H
2
OandCO
2
inmagma
sfro mtheMarian
aarc
andbackarcsystems
SallyNewma
nandEdwardStolper
Divisio
nofGeologica
landPlanetar
ySciences
,Californi
aInstitut
eofTechnolog
y,Pasadena
,Californi
a91125
(
sally@gps.caltech.edu)
RobertStern
Cente
rforLithospheri
cStudies
,Universit
yofTexa
satDallas
,Bo
x830688
,MSFA31
,Richardson
,Texa
s75083-0688
[1]
.
Abstract
:
Weexam
inedtheH
2
O
an dCO
2
content
sofglasse
sfro
m
lava
sandxenolith
sfro
m
the
Marian
aarcsystem
,anintraocean
icconvergen
tmargi
ninthewester
nPacific
,whic
hcontain
sanactive
volcani
carc
,anactivel
yspreadin
gbac
karcbasin
,andactiv
ebehind-the-ar
ccross-chai
nvolcanoes.
Sample
sinclud
e
(
1)glas
srim
sfro
m
Marian
aarc
,Marian
atrough
,an dcross-chai
nsubmarin
elavas
;
(
2)
glas
sinclusion
sinarcan dtroug
hphenocrysts
;an d
(
3)glas
sinclusion
sfro
m
agabbr
o+anorthosite
xenolit
hfro
m
Agriga
n
(
Marian
aarc
) .Glas
srim
sofsubmarin
ear clava
scontai
n0. 3±1.9wt%
H
2
O,and
CO
2
isbelo
w
detectio
nlimits .Wher
ethe
ycoul
dbecompared
,glas
sinclusion
sinarcphenocrysts
contai
nmor
eH
2
O
tha
nthei
rhos
tglasses
;mos
tarcglasse
sandphenocrys
tinclusion
sconta
inno
detec
tabl
eCO
2
,wit
htheexceptio
nofthos
efro
m
aNort
hHiyosh
ishoshonite
,whic
hcontain
s400±600
ppm .Th
eglas
sinclusion
sfro
m
theAgriga
nxenolit
hconta
in4±6%
H
2
O,andCO
2
isbelo
w
the
detec
tionlimit .Glasse
sfro
m
thecross-chai
nlava
saresimila
rtothos
efro
m
thearc
:H
2
O
content
sare
1.4±1.7wt%,andCO
2
isbelo
w
detec
tionlimits .
Volatil
econtent
sinMarian
atroug
hlav
aglas
srims
arevariable
:0.2±2.8wt%
H
2
O
and0±300ppm
CO
2
.Glas
sinclusion
sfro
m
troug
hphenocryst
shave
wate
rcontent
ssimila
rtothehos
tglass
,butthe
ycancontai
nupto875ppm
CO
2
.Volatil
econtent
sof
mel
tinclusion
s
fro
m
troug
h
and
arc
lava
s
and
fro
m
the
xenolit
h
impl
y
minimu
m
depth
s
of
crystallizatio
nof
1±
8km .H
2
OandCO
2
content
sofMarian
atroug
hglasse
sarenegat
ivel
ycorrelate
d,
indicatin
gsaturatio
noftheeruptin
gmagm
awit
haCO
2
±H
2
O
vapo
ratth epressur
eoferuptio
n
(
400
bar
sforthes
esamples
),wit
hthevapo
rrangin
gfro
m
nearl
ypur
eCO
2
attheCO
2
-ric
hendoftheglass
arra
ytonearl
ypur
eH
2
OattheH
2
O-ric
hend .Degassin
gofthes
emagma
sonascen
tanderuptio
nleads
tosignifican
tlos
sofCO
2
(
thereb
ymaskin
gpreeruptiv
eCO
2
contents
)butminima
ldisturban
ceof
preeruptiv
eH
2
Ocontents .Fo
rsubmarin
eMari
anaarcmagmas
,depth
swer
elowenoug
htha
tdegassing
onascen
tanderuptio
nledtolos
sofbot
hH
2
O
andCO
2
;asaresult
,H
2
O
content
sarepositively
corre
late
dwit
hwate
rdept
hforthes
esamples .Th
eH
2
O
content
sofpri
mitiv
eMarian
atroug
hmagmas
riches
tinth eslab-der
ive dcomponen
t
(
i.e.
,th emos
t``arc-lik
e''magmas
)ar e
2wt% .Although
evolve
dglasse
swit
hupto4±6wt%
H
2
O
arepresen
tamon
gMarian
aarcsamples
,weinterpre
tthe
glas
sinclusio
ndat
aasindicatin
gtha
tprimitiv
eMarian
aarcliquid
scontai
n1±3wt%
H
2
O .The
preeruptiv
eH
2
O
content
sofpri
mitiv
ecross-chai
nseamoun
tliquid
sare>1±2wt%.
Keywords
:
Subduction
;volatiles
;basalts
;Marianas
;bac
karcbasin
;degassing.
Indexterms:
Majo
relemen
tchemistry
;mino
relemen
tchemistry
;compositio
nofthecrust.
Receive
d
Octobe
r29,1999
;
Revised
Marc
h27,2000
;
Accepte
d
Marc
h31,2000;
Published
May 30, 2000.
Copyright 2000 by the American Geophysical Union
G
3
G
3
Geoche
mistry
Geophysics
Geosystems
Published by
AGU and the Geochemical Society
AN ELECT
RONIC JOUR
NAL OF
THE EA
RTH SCIENCES
Article
Volume 1
May 30, 2000
Paper
n
umber
1999GC00002
7
ISSN:
1525-2027
Geochemistry
Geophysics
Geosystems
New
man
,S.,E .Stolpe
r,andR .
Stern
,200
0 .
H
2
O
andCO
2
inmagma
sfro
m
theMarian
aarcandbac
karcsystems,
Geochem
.Geophys
.Geosyst.
,vol .1
,Pape
rnumbe
r1999GC00002
7
[15,062
words
,9figures
,6tables]
.
May 30,
2000.
1.Introduction
[2]
.
Althoug
hmuc
hhasbee
nwritte
nabou
tthe
importanc
eofvolatiles
,particularl
y
H
2
O,in
thegeneration
,fractionation
,anderuptio
nof
islan
darc
,bac
karc
,andcontinenta
larcmag-
mas
,unti
lrecentl
ylittl
ewa
sknow
noftheir
conc
entration
sin
un
degasse
d
ma
gma
sfrom
direc
tmeasurements .Th
eprincipa
ldifficulty
hasbee
ntha
tmos
tsample
darc-relate
dlavas
erup
tsubaeriall
yandconsequentl
ydega
ssub-
stant
iall y,so
preeruptiv
e
volatil
e
con
centra-
tion
s
canno
tusuall
y
be
directl
y
measured
(
althoug
h
indirec
tconstraint
s
ar e
available
fro
m
experi
mentall
y
determ
ined
phas
e
rela-
tion
s[e.g.
,
Sakuyama
,1983
;
Merzbache
rand
Eggler
,1984
;
SissonandGrove
,1993
;
Baker
etal
.,1994
;
Gardneretal
.,1995
;
Danyush-
evskyetal
.,1996]
) . Twothing
shav
echanged
in
recen
tyea
rs:Fir
st,the
deve
lop
men tof
microanalytica
ltool
s
has
enable
d
measure-
men
tof
the
concentration
s
of
volatile
s
in
glass
inclusio
ns
within
phenocrysts
,wh
ich
frequentl
y
remai
n
seale
d
durin
g
eruption
and
therefor
e
preserv
e
preeruptiv
e
volatile
content
s
[
Metrichetal
.,1993
;
Sissonand
Layne
,1993
;
SobolevandDanyushevsk
y
,
1994
;
SobolevandChaussidon
,1996;
Hauri
,
1998
;
SissonandBronto
,1998]
;andsecond,
submarin
earc-relate
d
lava
shav
ebecom
ein-
creasingl
y
available
,and
the
y
can
preserve
preeruptiv
evolatil
econtent
sbecaus
ethe
ydegas
incompletel
yoneruptio
nunde
rwate
ratpres-
sure
sofuptosevera
lhundre
dbar
s[
Sternand
Bibee
,1984
;
Devinean dSigurdsson
,1995;
Dixonetal
.,1995].
[3]
.
Inthi
spape
r,wetak
eadvantag
eofbot
hof
thes
edevelopment
stoprovid
eadditiona
lcon-
straint
sontheconcentration
sandrole
sofH
2
O
andCO
2
inthegeneratio
nofmagma
sfro
m
the
Marian
aarcanditsassociate
dbac
karcsystem.
Weusethemicroanalytica
lcapabilit
yofFou
r-
iertransfor
m
infrare
dspectroscop
y
(
FTIR
)to
measur
etheconcentration
softhes
evolatile
sin
theglass
yrim
sofsubmarin
eflow
sandinmelt
inclusion
strappe
dincrystallin
ephases .Impor-
tan
tfeature
sofou
rwor
karethemeasurement
ofCO
2
inthes
esample
sandthefac
ttha
tthey
com
efro
m
avariet
yofenvironment
s
(
i.e.
,the
magmati
cfron
tofthearc
;cross-ar
cseamount
chain
sextendin
gtotherea
rofthemagmatic
front
;andthebac
karcbasin) .Thes
efeatures
allo
w
characterizatio
nofthepreeruptiv
econ-
centration
s
of
bot
h
of
th e
majo
r
volatiles
(
i.e.
,H
2
O
and
CO
2
)in
magma
s
fro
m
this
oceani
carc/bac
karcregio
nandexamination
ofdifference
sbetwee
nsubenvironment
swith-
inthi
soveral
lsystem.
2.SampleDescription
2.1.Marian
aTrough
(
BackArc)
[4]
.
Sample
swer
eavailabl
efro
m
asubstantial
lengt
h
oftheMarian
atroug
h
axis
,fro
m
the
intersectio
nofthebac
karcspreadin
gcenter
wit
hthearcat
23
8
N,sout
hto
15
8
N
(
Figure
1) .Th
enatur
eofcrusta
lextensio
nandmag-
matis
m
varie
sove
rthelengt
hoftheextension
axis
,fro
m
riftin
gassociate
dwit
htheeruption
ofarc-lik
eevolve
dlava
sandporphyriti
cba-
salt
sinth eshallo
w
(
i.e.
,nea
rse alevel
)nort
hto
seafloo
rspreadin
ganderuptio
nofmid-ocean
ridg
ebasal
t
(
MORB)-lik
elava
sinth ecentral
and
souther
n
portion
softhetrough
,where
depth
stotheridg
eaxi
srang
efro
m
3.5km
at
13
8
N
to4.5±5km
at20
8
±21
8
N
[
Martine
zet
al
.,1995
;
Gribbleetal
.,1996
,1998
] .
Lavas
fro
m
theful
lrang
eoftectoni
csetting
sand
G
3
N
E
WMAN
ET AL: H
2
O
AND CO
2
IN THE MARIANA ARC
9
1999GC000027
Geochemist
r
y
Geophysics
Geosystems
G
3
s
p
a
n
ning the range of known magma types
from the trough are included in this study.
[5]
Samples were collected by dredge (WOK
series) and the
Alvin
submersible
(
ALV series)
from near 18
8
N by James Hawkins
(
Scripps
Institution of Oceanography) during a 1987
cruise of the R/V
Atlantis II
(
major and trace
elements reported by
Hawkins et al
.[1990];
volatile data initially published by
Stolper and
25
°
N
20
°
N
15
°
N15
°
N
20
°
N
25
°
N
150
°
E
145
°
E
140
°
E
140
°
E
145
°
E
150
°
E
300
0
2000
2000
200
0
200
0
100
0
N. Hiyoshi
grigan
A
15-17
°
N
Kasugas
Seamounts
Submarine trough locations
Submarine arc locations
Cross-chain volcanoes
Gabbroic xenolith
Inclusions
Contours in
1000 fathoms
2000
200
0
1000
M
a
r
i
a
n
a
T
r
e
n
c
h
3000
M
a
r
i
a
n
a
Arc
18
°
N
18-23
°
N
M
a
ri
a
n
a
T
r
o
u
g
h
Fukujin
x
Sample locations
F
i
g
ure 1.
Location map for samples included in this study .For the trough, individual sample locations are
too numerous to show separately .Therefore just the ranges are shown .
G
3
N
E
W
M
A
N
E
T
A
L
:
H
2
O
A
N
D
C
O
2
I
N
T
HE MARIANA ARC
1999GC000027
Geochemistry
Geophysics
Geosystems
G
3
N
ewman
[1994]); by dredge from the same
region during the 1988 GH88-1 cruise of the
Hakurei-Maru
(
samples D1009 and D1010;
major elements and rare gases are reported by
Ikeda et al
.[1998]); by dredge
(
DS and GTVA
series) from near 15
8
to 17
8
N by S .Bloomer
(
Oregon State University) and D .Steuben
(Karlsruhe University) during a 1990 cruise of
the R/V
Sonne of Germany
(
H
2
O contents and
major and trace element and isotopic composi-
tions are given by
Gribble et al
.[1996]); and by
dredge from 18
8
to 23
8
N during Leg 7 of the
1991 TUNES expedition of the R/V
Thomas
Washington
(
dredges 46 ± 82; H
2
O contents and
major and trace element and isotopic composi-
tions are given by
Gribble et al
.[1998]) .Sam-
pling locations, full sample names, and depths
of collection for the trough samples are listed in
Table 1 .Petrographic descriptions of samples
from most of these suites are given by
Hawkins
etal
.[1990] and
Gribbleetal
.[1996, 1998]; the
suite from which samples D1009 and D1010
come is described
(
although in somewhat less
detail) by
Ikeda et al
.[1998] .
[6]
Fresh basaltic glass was picked from the
glassy lavas, all of which were fragments of
pillow rinds, except for GTVA 71-1-7, which is
from a sheet flow .Most analyzed chips contain
at least one vesicle, generally >100
m
m across.
The glass fragments contain crystals in varying
proportions, primarily olivine and plagioclase
phenocrysts and plagioclase microlites; some
fragments show evidence of devitrification/
crystallization locally around these crystals,
and such regions were avoided during analysis.
Many of the olivine and plagioclase pheno-
crysts are >l mm across and contain inclusions
of brown glass .Most of these glass inclusions,
especially in plagioclase, are at least partially
devitrified and appear to be connected to a
crystal face by a capillary [
Anderson
, 1991].
We analyzed four glassy inclusions in three
olivine phenocrysts from WOK 28-3; none of
these have visible connections to grain bound-
aries .Two plagioclase-hosted glassy inclusions
were also analyzed: one, in plagioclase from
ALV 1833-1, is connected to the crystal ± host
glass interface by a capillary; the other, in
plagioclase from ALV 1832-2, appears to have
been isolated from the host liquid since entrap-
ment .Representative photomicrographs of
these inclusions are shown in Figure 2.
2.2. Mariana Arc
(
Magmatic Front)
[7]
Samples were available from a substantial
length of the Mariana arc: from North Hiyoshi
(
23.37
8
N), near the northern end of the arc just
north of where it is intersected by the Mariana
trough, to Agrigan
(
18.75
8
N), in the central
region of the arc
(
Figure 1) .With the exception
of a composite gabbro + anorthosite xenolith
from Agrigan island, the samples are all sub-
marine lavas dredged from seamounts by one
of us
(
R .Stern) and S .Bloomer during the
1985 TT-192 cruise of the R/V
T. G. Thomp-
son
.Geochemical data for these samples are
given by
Bloomer et al
.[1989a, 1989b],
Lin et
al
.[1989], and S .H .Bloomer and R .J .Stern
(
unpublished data, 1999) .No H
2
OorCO
2
measurements have been reported for these
samples, but
Garcia et al
.[1979] reported
measurements from samples dredged from Fu-
kujin seamount
(
21.92
8
N) .The samples stu-
died here include low-K and medium-K series
lavas with compositions ranging from basalt to
dacite or silicic andesite [
Bloomer et al
.,
1989a] .Lavas from the northernmost sea-
mounts
(
from dredges 53 and 54) are shosho-
nitic; these lavas may relate to the propagation
of the trough spreading center into the arc
[
Bloomer et al
., 1989a]. Sampling locations,
full sample names, and depths of collection
(
for the lavas) for the arc samples are listed in
Table 3.
[8]
The dredged lavas are generally porphyritic
and vesicular .The groundmass in most sam-
ples is partly crystalline, but glass fragments
G
3
NE
W
M
A
N
E
T
A
L: H
2
O AND CO
2
IN THE MARIANA ARC
1999GC000027
Geochemistry
Geophysics
Geosystems
G
3
Table 1.
Dissolved Volatile Contents and Depths of Collection and Calculated Vapor Saturation
Conditions
a
for Mariana Trough Matrix Glasses
Sample
Latitude,
8
N
Longitude,
8
EH
2
O, wt %
CO
2
, ppm
Depth of
Collection, m
Vapor-Saturated
Depth
b
,m
X
(
H
2
O)
vapor
c
15
8
±17
8
N
d
DS18-1-6
16.18
144.79
2.03 0.06
30 9
3838 ± 4052
4600
0.87
DS22-2-2
16.96
144.78
0.48 0.01
168 6
3055 ± 3287
3600
0.06
DS74-2-1
16.53
144.83
2.08 0.10
28 33
3880 ± 5000
4700
0.88
DS74-2-3
16.53
144.83
2.16 0.12
bdl
e
3880 ± 5000
4500
1.00
DS74-3-1
16.53
144.83
1.40 0.08
95 4
3880 ± 5000
3800
0.50
DS79-2-2
16.08
144.74
1.09 0.06
152 2
3625 ± 3747
4200
0.27
DS80-23-2
15.75
144.75
1.52 0.15
98 2
3730 ± 3922
4200
0.53
DS80-25-3
15.75
144.75
2.78 0.08
104 10
3730 ± 3922
9500
0.78
DS84-1-1
15.00
144.46
0.20 0.003
188 11
4046 ± 4272
3900
0.01
DS84-2-1
15.00
144.46
0.21 0.004
215 6
4046 ± 4272
4400
0.01
DS86-4-1
15.09
144.50
0.58 0.01
126 21
3386 ± 3518
2900
0.10
DS88-1-2
15.30
144.51
1.30 0.07
199 6
4279 ± 4671
5700
0.29
DS88-2-1
15.30
144.51
1.04 0.05
161 31
4279 ± 4671
4300
0.24
DS88-3-1
15.30
144.51
1.24 0.02
189 12
4279 ± 4671
5300
0.28
GTVA 71-1-7
16.99
144.83
1.34 0.14
111 19
3400 ± 3404
4000
0.44
GTVA 73-2-2
16.69
144.82
1.46 0.09
142 4
4124 ± 4236
5000
0.42
GTVA 75-1-1
16.41
144.85
2.21 0.07
bdl
3670 ± 3954
4900
1.00
18
8
N
f
WOK 5-4
18.3
144.7
1.60 0.05
64 6
3925
3800
0.66
WOK 10-1
18.4
144.65
1.14 0.02
120 1
3950 ± 4100
3700
0.34
WOK 16-2
18.1
144.75
0.64 0.01
184 7
3969 ± 3975
4100
0.09
WOK 28-3
17.6
144.9
0.50 0.01
183 10
4050 ± 4126
4000
0.05
ALV 1832-2
18.15
144.8
2.28 0.09
18 5
3694
5300
0.93
ALV 1833-1
18.1
144.75
2.43 0.05
24 15
3691
6100
0.92
ALV 1833-11
18.1
144.75
1.20 0.01
102 12
3693
3400
0.40
ALV 1839-21
18.2
144.7
1.21 0.03
94 6
4044
3300
0.42
ALV 1840-3
18.2
144.7
1.26 0.04
113 33
3283
3800
0.40
ALV 1846-9
18.3
144.7
1.89 0.05
bdl
3485
3400
1.00
ALV 1846-12
18.3
144.7
1.55 0.02
90 8
3751
4200
0.56
18
8
±23
8
N
g
46:1-6
20.82
143.55
1.41 0.09
111 9
3700
4200
0.46
47:1-5
20.97
143.44
1.76 0.08
66 6
3680 ± 5100
4300
0.69
48:1-3
21.31
143.36
1.56 0.10
bdl
2860 ± 2900
2400
1.00
54:1-1
22.79
142.42
1.69 0.02
61 6
3430 ± 3460
4000
0.69
55:1-1
22.87
142.32
1.79 0.06
bdl
3125 ± 3205
3600
1.00
68:1-2
21.35
143.28
1.84 0.08
69 6
3800 ± 3900
4700
0.70
71:1-14
20.31
143.93
1.82 0.36
128 17
4330 ± 4400
5800
0.56
72:2
19.83
144.3
2.23 0.09
74 13
4380 ± 4450
6300
0.77
73:2-1
19.73
144.4
1.15 0.05
160 19
3600 ± 4036
4500
0.28
74:1-1
19.67
144.39
1.00 0.06
112 4
3630 ± 3690
3200
0.29
75:1-2
19.44
144.48
0.72 0.02
227 1
4380 ± 4450
5100
0.09
76:1-1
19.45
145.54
0.73 0.01
180 16
3364 ± 4480
4100
0.12
80:1-3
19.12
144.67
0.57 0.003
295 7
4040 ± 4050
6300
0.05
82:1-1
18.75
144.66
1.69 0.20
109 18
4300 ± 4340
5000
0.56
D1009
18.15
144.73
1.20 0.03
138 2
3700 ± 3770
4200
0.33
D1010
18.15
144.73
1.13 0.07
139 1
3700 ± 3770
4000
0.30
Std JDF-D2
0.347 0.007 225 45
G
3
N
E
WMAN
E
T
A
L
:
H
2
O AND CO
2
IN THE MARIANA ARC
1999GC000027
Geochemistry
Geophysics
Geosystems
G
3
l
arge en
o
u
g
h for analysis were picked from all
samples except 38-2, in which the plagioclase-
rich groundmass was microcrystalline .These
samples contain phenocrysts of olivine, clin-
opyroxene, plagioclase, orthopyroxene, horn-
blende, and/or biotite (in the shoshonites),
many of which contain melt inclusions .Ten
glassy inclusions with no evidence of connec-
tion to a grain boundary were analyzed from
four different submarine arc samples .Each of
these inclusions was in a separate olivine,
plagioclase, or clinopyroxene host crystal.
Photomicrographs of two of these inclusions
are shown in Figure 2, one showing a large
vapor bubble.
[9]
The Agrigan xenolith was collected by R.
Stern in 1976 and described by
Stern
[1979].
It is a cumulate comprising distinct anortho-
sitic and gabbroic layers .We studied five
glass inclusions from five different cumulus
olivine grains from this sample .These inclu-
sions have no visible connections to the
surfaces of their host olivines .A representa-
tive photomicrograph of one of these inclu-
sions is shown in Figure 2, showing daughter
crystals growing in from the inclusion
boundary.
2.3. Cross-Chain Seamount Samples
[10]
Glassy samples from the Kasuga cross-
chain at
21.5
8
N[
Fryer et al
., 1997] were
studied .Sampling locations and depths of
collection for these samples are listed in
Table 3 .No glass inclusions within pheno-
crysts were analyzed from these samples.
There are no previous measurements of vo-
latile contents in these or related samples.
Notes to Table 1:
a
Pressures for vapor saturation and the composition of the coexisting vapor in the mixed volatile system were calculated using the
following relationship describing the thermodynamic condition for saturation:
a
m
H
2
O
=f
v
H
2
O
K
H
2
O
a
m
CO
2
=f
v
CO
2
K
CO
2
1,
where
a
is activity,
m
is melt, mol is molecular,
v
is vapor,
f
is fugacity of the pure volatile component at the pressure and
temperature of interest,
K
is solubility constant for the pure volatile component adjusted to the appropriate temperature and pressure .The
solubility of H
2
O is described in terms of the molecular H
2
O species, according to
Stolper
[1982] .Assuming Henrian behavior, the
activities can be taken to be equal to the mole fractions .Mixing between the components of the vapor phase is assumed to be ideal,
following
Dixon and Stolper
[1995] .The relationship given above was used to calculate saturation curves for mixed H
2
O±CO
2
volatile
phases for several pressures .Fugacities of pure H
2
O and CO
2
were calculated using a modified Redlich-Kwong model [
Holloway
, 1977]
and their solubilities are from
Dixon et al
.[1995] .The calculations were done at 1150
8
C, which is a reasonable temperature for Mariana
trough basalts, based on the results of
Hawkins et al
.[1990] and
Hawkins and Melchior
[1985], who calculated magmatic temperatures
of 1078
8
± 1199
8
C based on olivine/chromite and olivine/glass geothermometers .This relationship was also used to calculate the
saturation curves for mixed H
2
O±CO
2
volatile phases for several pressures shown in Figure 5.
b
Calculated depth of vapor saturation for reported H
2
OandCO
2
contents based on calculated pressure of vapor saturation
(
see
footnote a) and converting to water depth using 10-m water depth = 1 bar .Calculated depths have been rounded to the nearest 100 m .
c
Calculated mole fraction of H
2
O in vapor coexisting with melt with reported H
2
OandCO
2
contents
(
assumed to be 1 if CO
2
is below
detection limit) .See footnote a .
d
All these samples formally require the prefix ``Sonne 69''; i.e., DS18-1-6 is Sonne 69 DS18-1-6. The prefixes have been omitted for
simplicity .DS refers to dredge sample
,and
GTVA refers to grab television apparatus.
e
In all tables, bdl is below detection limit.
f
These same samples were previously analyzed by
Stolper and Newman
[1994] .The WOK samples are dredges; the ALV samples
were collected with the
Alvin
submersible .Results reported here are averages of these previous results and measurements on new sections
of these samples prepared for this study .CO
2
contents were determined as described in the text using a less subjective approach to
background subtraction .CO
2
contents presented here differ on average by +6%
(
relative) compared to the results presented by
Stolper
and Newman
[1994], although individual analyses can deviate by up to plus or minus several tens of percent
(
relative) .Excluding H
2
O-
rich samples ALV 1832-2 and ALV 1833-1, reported water contents are on average 0.3%
(
relative) higher than the results presented by
Stolper and Newman
[1994].
g
The D1009 and D1010 samples formally require the prefix ``GH88-1''; i.e., D1009 is GH88-1 D1009. The D indicates they are
dredge samples .All other samples in this section formally require the prefix ``TUN
ES7 D,"
where again the D indicates they are from
dredges .The prefixes have been omitted for simplicity .
G
3
N
E
W
MAN ET AL: H
2
O AND CO
2
IN THE MARIANA ARC
1999GC000027
Geochemistry
Geophysics
Geosystems
G
3
[
1
1
]
T
h
et
h
ree Kasuga seamounts form a ridge
that trends obliquely away from the northern
part of the arc to the SSW
(
Figure 1) .The
closest active volcano in the arc itself is Fuku-
jin, located
30 km NW of the Kasugas .The
three Kasuga seamount samples included in
this study were collected from the lower to
middle slopes of Kasuga 2 and Kasuga 3 by the
submersible
Alvin
in 1987 as described by
Fryer et al
.[1997] .These two seamounts are
active, on the basis of the occurrence of hydro-
thermal activity and fresh glass on both [
Fryer
et al
., 1997]. The two Kasuga 2 samples in this
study are high-K basalts, and the one Kasuga 3
sample is an absarokite .Petrographic descrip-
tions and geochemical data for the suite of
Kasuga samples, including those studied here,
are given by
Jackson
[1989],
Stern et al
.
[1993], and
Fryer et al
.[1997] .Lavas from
these seamounts have glassy margins from
which clean glass was picked; the samples we
studied are vesicular
(
5 ± 50%) and aphyric to
sparsely phyric
(
with phenocrysts of olivine,
plagioclase, and clinopyroxene) [
Stern et al
.,
1993;
Fryer et al
., 1997].
3. Analytical Techniques
3.1. H
2
O an dCO
2
[12]
For each fragment of glassy lava a dou-
bly polished section
0.5 ± 2 mm across and
50 ± 320
m
m thick was prepared .For glass
inclusions in phenocrysts the host phase was
doubly polished, exposing the inclusion on
both sides; after polishing, the inclusions were
15 ± 120
m
m across and
10 ± 100
m
m thick.
Dissolved H
2
O and CO
2
contents were deter-
mined using the FTIR techniques described by
a. ALV1833-1 melt inclusion in plagioclase
circled inclusion - 70
μ
m across
plane pola
rized light
crossed polars
neck to crack
b. ALV1832-2 melt inclusion in plagioclase
circled inclusion - 50 x 75
μ
m
c. WOK 28-3 OL1 melt inclusion i
n olivine
circled inclusi
on - 150 x 80
μ
m
d. AG 4-4A
inclusion in
olivine 5
(80
μ
m long)
from
Agrigan island
e. 54B inclusion in
olivine 3
(80
μ
m long) from
North Hiyoshi
seamount
f. 38-2
inclusion in
pyroxene 1
(100
μ
m long)
from
Fukuyama
seamount
Figure2.
Photomicrographs of representative melt
inclusions in phenocrysts from submarine Mariana
trough and arc lavas and in olivine from the Agrigan
xenolith.
G
3
NE
W
M
AN ET AL: H
2
O A
N
D
CO
2
IN THE MARIANA ARC
1999GC000027
Geochemistry
Geophysics
Geosystems
G
3
F
i
n
e
and Stolper
[1985/1986] and
Dixon et al
.
[1988], except most spectra were taken with a
Nicolet IR-Plan microscope attached to a Ni-
colet 60SX spectrometer .The double-aperture
system on the microscope [
Wopenka et al
.,
1990] was used, with adjustable rectangular
apertures ranging from
15 ± 110
m
m across:
apertures were positioned above and below the
sample to mask light not propagating through
the region of interest and to eliminate light
refracted significantly from normal while pas-
sing through the sample .In this configuration
the region of interest can be seen through the
Cassegrainian optics of the microscope, which
are also used to direct the infrared beam.
[13]
The absorption bands used to determine
dissolved H
2
O concentrations were the OH
stretching band at
3550 cm
1
(
primarily for
determining molecular H
2
O plus hydroxyl con-
tents of glasses with total H
2
O contents less
than
1 wt %, although it was also used for a
few thin samples with higher total H
2
O con-
tents), the
5200 cm
1
band resulting from the
combination stretching and bending mode of
H
2
O molecules, and the
4500 cm
1
band
resulting from the combination modes of X ±
OH groups
(
where X = Si, Al, etc .[
Stolper
,
1982]) .The antisymmetric stretching bands of
distorted CO
3
2
groups at 1515 and 1435 cm
1
were used to determine the amount of CO
2
dissolved as carbonate ions .Except for the
most silicic samples
(
see footnotes to Tables 3
and 4), a molar absorptivity of 63 3 L/mol cm
(
P .Dobson, S .Newman, S .Epstein, and E .
Stolper, unpublished data, 1987) was used to
calculate H
2
O concentrations from the intensity
of the 3550 cm
1
band; this is similar to the
value of
Pandya et al
.[1992]
(
61 1 L/mol
cm) and
Yamashita et al
.[1997]
(
64 1 L/mol
cm), but lower than that reported by
Jendrze-
jewski et al
.[1996b]
(
78 1 L/mol cm) .Molar
absorptivities of 0.67 0.03 and 0.62 0.07 L/
mol cm were used
(
except for the most silicic
samples; see footnotes to Tables 4 and 5) to
calculate the molecular H
2
O and hydroxyl
group concentrations from the 5200 and 4520
cm
1
bands [
Dixon et al
., 1995]. A molar
absorptivity of 375 20 L/mol cm [
Fine and
Stolper
, 1985/1986] was used to calculate the
concentration of CO
2
dissolved as CO
3
2
;this
value is similar to the 397 7 L/mol cm value
of
Jendrzejewski et al
.[1996a] .The density of
each glass was assumed to be 2.8 g/cm
3
.The
absence of absorption features at 2350 cm
1
indicates that dissolved molecular CO
2
is be-
low the detection limit in all samples
(
typically
less than
25 ppm for samples in the thickness
range used here), consistent with previous ob-
servations that CO
2
dissolves nearly entirely as
Table 2.
Volatile Concentrations in Mariana Trough Glass Inclusions
Sample
H
2
O, wt %
CO
2
,
a
ppm
Pressure for Vapor
Saturation,
b
bars
X
(
H
2
O)
vapor
c
WOK 28-3
(
in OL1)
0.47 0.02
691 28
1400
0.02
WOK 28-3
(
in OL2)
0.48 0.04
613 78
1200
0.02
WOK 28-3
(
large inclusion in OL3)
0.60 0.09
-
WOK 28-3
(
small inclusion in OL3)
0.55 0.01
-
ALV 1832-2
(
in PLAG)
2.23 0.07
875 141
2200
0.25
ALV 1833-1
(
in PLAG)
1.71 0.01
bdl
300
1.00
a
Dash indicates a sample in which the CO
2
content was not determined due to large interference fringes in the infrared spectrum .The
pressure of vapor saturation and composition of coexisting vapor were not calculated for these samples .Here bdl indicates the CO
2
content was below the detection limit.
b
Calculated pressure of vapor saturation for reported H
2
OandCO
2
contents
(
0% CO
2
assumed for samples in which CO
2
is bdl) .See
footnote a, Table 1 .Calculated pressures have been rounded to the nearest 100 bars .
c
Calculated mole fraction of H
2
O in vapor coexisting with melt with reported H
2
OandCO
2
contents
(
assumed to be 1 if CO
2
is bdl).
See footnote a, Table 1.
G
3
NE
WMAN ET AL: H
2
O AND CO
2
IN THE MARIANA ARC
1999GC000027
Geochemistry
Geophysics
Geosystems
G
3
Table 3.
Volatile Concentrations in Submarine Mariana-Arc-Lava Matrix Glasses and Cross-Chain ± Volcano Matrix Glasses
Sample
Volcano
Latitude,
8
N
Longitude,
8
EH
2
O, wt %
CO
2
,
ppm
Depth of Col-
lection, m
Vapor-Satu-
rated Depth
at 0%
CO
2
,
a
m
Vapor-Saturated
Depth at 50 ppm
CO
2
,
b
m
Model CO
2
,
c
ppm
Mariana Arc Seamounts
d
14-21
(
basaltic andesite)
supply reef
20.13
144.08
0.75 0.09
bdl
530 ± 1060
500
1500
15
14-24
(
basaltic andesite)
supply reef
20.13
144.08
0.28 0.08
bdl
530 ± 1060
60
1100
36
21-8
(
basaltic andesite)
NW Uracas
20.65
144.44
1.18 0.10
bdl
2000 ± 2200
1300
2300
39
25-14
(
andesite)
S. Daikoku
21.05
144.50
0.80 0.07
bdl
1430 ± 1450
600
1600
42
29:1-2
(
silicic andesite)
e
Daikoku
21.33
144.19
0.75 0.06
bdl
400 ± 920
700
1800
-
29:2-2
(
silicic andesite)
e
Daikoku
21.33
144.19
0.40 0.02
bdl
400 ± 920
200
1300
21
34-2-2
(
basaltic andesite)
Fukujin
21.92
143.46
0.44 0.08
bdl
1002 ± 1600
160
1200
55
34-3-3
(
basaltic andesite)
Fukujin
21.92
143.46
0.62 0.01
bdl
1002 ± 1600
330
1400
47
38-2
(b
asalt)
f
Fukuyama
22.37
143.12
1740 ± 1800
39-1
(
magnesian andesite)
Sakoyama
22.36
143.08
1.06 0.06
bdl
2100
1000
2100
52
53i
(
rhyolite)
e
N. Hiyoshi
23.37
141.80
1.91 0.18
bdl
1280 ± 1687
4000
5200
-
54B
(
basalt)
N. Hiyoshi
23.37
141.72
0.91 0.08
bdl
1250 ± 1640
700
1800
34
54G
(
basalt)
N. Hiyoshi
23.37
141.72
0.94 0.06
bdl
1250 ± 1640
800
1800
31
Kasuga Cross-Chain Seamounts
ALV 1880-3
(
high-K
basalt)
Kasuga 2
21.58
143.6
1.45 0.01
bdl
2200 ± 2600
2000
3000
20
ALV 1880-5
(
high-K
basalt)
Kasuga 2
21.58
143.6
1.55 0.22
bdl
2200 ± 2600
2300
3300
6
ALV 1884-10
(
absarokite)
Kasuga 3
21.37
143.5
1.69 0.13
bdl
2900 ± 3500
2700
3800
23
a
Calculated depth of vapor saturation for reported H
2
O content and 0% CO
2
based on calculated pressure of vapor saturation
(
see footnote a, Table 1) and converting to water depth
using 10-m water depth = 1 bar.
b
Calculated depth of vapor saturation for reported H
2
O content and 50 ppm CO
2
based on calculated pressure of vapor saturation
(
see footnote a, Table 1) and converting to water
depth using 10 m water depth = 1 bar .Calculated depths have been rounded to the nearest 100 m .
c
Calculated CO
2
content required for vapor saturation at the depth of collection
(
see footnote a, Table 1) .For dredged samples, this calculation was done for the midpoint of the depth
range of the dredge .No value is listed for 53i because it is oversaturated with respect to pure H
2
O vapor at the depth of collection (see Figure 4a).
d
All Mariana arc seamount samples formally require the prefix ``TT192 D''; i.e., 14-21 is TT192 D14-21. The prefixes have been omitted for simplicity.
e
For these silicic samples, values of molar absorptivities appropriate to rhyolitic glass [
Newman et al
., 1986;
Dobson et al
., 1989] were used to determine dissolved H
2
O contents.
Calculations of vapor saturation utilized the solubility model for rhyolitic melts from
Blank et al
.[1993] .
f
No glass available for this sample .Consequently, no measured volatile contents or calculated pressures or CO
2
contents at vapor saturation are given.
G
3
NE
WMAN ET AL: H
2
O AND CO
2
IN THE MARIANA ARC
1999GC000027
Geochemistry
Geophysics
Geosystems
G
3