Volatiles in Alkalic Basalts from the North Arch

JOURNAL OF PETROLOGY
VOLUME 38
NUMBER 7
PAGES 911–939
1997
Volatiles in Alkalic Basalts from the North
Arch Volcanic Field, Hawaii: Extensive
Degassing of Deep Submarine-erupted
Alkalic Series Lavas
JACQUELINE E. DIXON1∗, DAVID A. CLAGUE2, PAUL WALLACE3 AND
ROBERT POREDA4
1
DIVISION OF MARINE GEOLOGY AND GEOPHYSICS, ROSENSTIEL SCHOOL OF MARINE AND ATMOSPHERIC SCIENCE,
UNIVERSITY OF MIAMI, MIAMI, FL 33149, USA
2
MONTEREY BAY AQUARIUM RESEARCH INSTITUTE, MOSS LANDING, CA 95039-0628, USA
3
OCEAN DRILLING PROGRAM AND DEPARTMENT OF GEOLOGY AND GEOPHYSICS, TEXAS A&M UNIVERSITY, COLLEGE
STATION, TX 77845-9547, USA
4
DEPARTMENT OF GEOLOGICAL SCIENCE, UNIVERSITY OF ROCHESTER, ROCHESTER, NY 14627, USA
RECEIVED APRIL 30, 1996 REVISED TYPESCRIPT ACCEPTED FEBRUARY 26, 1997
The North Arch volcanic field is a submarine suite of alkali basaltic
to nephelinitic lavas on the seafloor north of Oahu at water depths
of 3900–4380 m. Glasses from these lavas were analyzed for
H2O, CO2, Cl, S, Fe3+/RFe, and noble gases to investigate the
role of volatiles in the generation, evolution, and degassing of these
alkalic series lavas. In contrast to the systematic negative correlation
between concentrations of SiO2 and nonvolatile incompatible elements
(e.g. P2O5), the behavior of the volatile components is much more
irregular. Concentrations of H2O in glasses vary by a factor of two
(~0·69–1·42 wt %) and show a poor correlation with melt
composition, whereas concentrations of dissolved CO2 in glasses
(260–800 p.p.m.) increase with increasing alkalinity of the glasses.
The H2O and CO2 concentrations in the glasses are in equilibrium
with an H2O–CO2 vapor at the depth of eruption (~400 bar
pressure). Samples collected directly from vent structures are highly
vesicular, suggesting that these samples were gas rich upon eruption.
Estimated bulk volatile contents of the two most vesicular vent
samples are high (1·9±0·1 wt % H2O and 5·4±0·4 CO2)
and are interpreted to have formed by closed system degassing.
Estimated bulk volatile contents in four other vesicular vent samples
are lower (1·3±0·2 wt % H2O and 2·0±0·4 wt % CO2),
and these samples are interpreted to have lost some gas during
eruption. Glass samples from inflated, flat lava flows are nonvesicular
and interpreted to have lost essentially all exsolved gas during
eruption and flow. Forward degassing models can predict the
observed range in dissolved H2O and CO2 contents, calculated
vapor compositions, and vesicularity as a function of SiO2. The
models involve open to closed system degassing of an H2O–CO2
vapor phase from melts initially having H2O/P2O5=3 and CO2/
H2O=1–4 by mass. Cl concentrations (400–1360 p.p.m.) in
glasses correlate with concentrations of nonvolatile, incompatible
elements. Concentrations of noble gases measured on bulk glass
samples are low compared with mid-oceanic ridge basalt (MORB).
The low concentrations result mainly from extensive vapor exsolution
from the magma. The helium isotopic ratios for gases released from
vesicles are similar to MORB values [6·8–8·5 times the air ratio
(RA)], whereas those released from glasses are lower than MORB
values as a result of in situ decay of U and Th. The S contents
(0·11–0·22 wt %) of most of the alkali olivine basaltic and
basanitic glasses are sufficient to saturate the silicate melt with
immiscible Fe–S–O liquid at the T and P of eruption and
quenching. However, two vesicular samples appear to have lost
some dissolved S owing to eruptive degassing. Magmatic oxygen
fugacities estimated from Fe3+/RFe range from DFMQ=–0·8
to +0·7, with the nephelinitic glasses being more oxidizing than
the less alkalic glasses. We infer that the mantle source region for
∗Corresponding author. Telephone (office): 305-361-4150. Fax: 305361-4632. e-mail: [email protected]
 Oxford University Press 1997
JOURNAL OF PETROLOGY
VOLUME 38
the North Arch magmas was homogeneous with respect to Fe3+/
RFe and that melting occurred in the absence of graphite or CH4rich fluid. The effect of variable partial melting on magmatic oxygen
fugacity may be a common feature of Hawaiian volcanism. These
complex data point to a simple result, namely that parental magma
compositions can be derived by variable extents of melting of a
homogeneous source followed by olivine crystallization and degassing
at 400 bar. If the parental liquids are produced by 1·6–9·0%
partial melting (±20% relative), then mantle volatile contents are
estimated to be 525±75 p.p.m. H2O, 1300±800 p.p.m. CO2
and 30±6 p.p.m. Cl.
KEY WORDS:
NUMBER 7
JULY 1997
alkalic; basalt; degassing; volatiles; mantle
(Sarda et al., 1985; Gerlach, 1986, 1989; Allègre et al.,
1986–1987; Bottinga & Javoy, 1990; Sarda & Graham,
1990; Blank et al., 1993; Pineau & Javoy, 1994). The
‘popping rocks’ from the Mid-Atlantic Ridge at 14°N
are a rare and important exception (Sarda & Graham,
1990; Gerlach, 1991; Graham & Sarda, 1991; Pineau &
Javoy, 1994). It is extremely difficult, therefore, to estimate mantle carbon contents based on measured values
in MORB glasses (e.g. Jambon, 1994; Dixon & Stolper,
1995). Even though the abundance and isotopic composition of carbon in basaltic glasses have been the subject
of many studies, there is still no consensus on the carbon
content of primary mantle-derived magmas or mantle
carbon variability (Pineau et al., 1976; Delaney et al.,
1978; Javoy et al., 1978; Muenow et al., 1979; Mathez &
Delaney, 1981; Pineau & Javoy, 1983, 1994; Des Marais
& Moore, 1984; Mattey et al., 1984, 1989; Sakai et al.,
1984; Fine & Stolper, 1986; Gerlach & Thomas, 1986;
Exley et al., 1986; Dixon et al., 1988, 1991; Gerlach &
Taylor, 1990; Javoy & Pineau, 1991; Blank et al., 1993;
Trull et al., 1993; Kingsley & Schilling, 1995).
Many alkalic basalts are thought to erupt quickly with
little or no residence time in a shallow crustal reservoir,
as inferred from their ability to carry mantle xenoliths
to the surface (e.g. Sparks et al., 1977; Wilson & Head,
1981; Spera, 1984; Clague, 1987). Any gas exsolved
during this more direct ascent may have less opportunity
to segregate and escape. If erupted on land, however,
these exsolved volatiles are completely lost during eruption and flow. For example, classic suites of oceanic
island alkalic lavas, such as the East Molokai Volcanic
Series on Molokai (Clague & Beeson, 1980), the Honolulu
Volcanic Series on Oahu (Clague & Frey, 1982) and the
Koloa Volcanic Series on Kauai (Clague & Dalrymple,
1988), are not suitable for volatile analysis because they
have lost most, if not all, of their pre-eruptive volatiles
owing to severe degassing. One solution to this problem
is to investigate pre-eruptive volatile contents in melt
inclusions trapped in phenocrysts. In subaerially erupted
lavas, however, even melt inclusions in phenocrysts often
record complex degassing of magmas within a summit
reservoir (e.g. Anderson & Brown, 1993). Another approach, adopted here, is to look at alkalic basalts erupted
deep on the seafloor, where much of the initial volatile
content in the form of exsolved (vesicles) and dissolved
species may be retained in quenched glassy rinds. These
submarine basalts may provide information on primary
magmatic and, by inference, mantle volatile contents.
The extent and style of degassing of tholeiitic basaltic
melts has been quantitatively modeled by comparing
accurate measurements of volatiles dissolved in quenched
glassy rinds with experimental determinations of water
and carbon dioxide solubilities (Greenland et al., 1985;
Gerlach, 1986; Dixon et al., 1988, 1991, 1995; Clague et
al., 1991; Stolper & Newman, 1994; Dixon & Stolper,
INTRODUCTION
Volatiles play an important role in the generation and
evolution of mantle-derived melts, affecting the extent of
mantle melting (Green & Ringwood, 1967; Kushiro et
al., 1968; Kushiro, 1970; Green, 1972, 1973; Green &
Wallace, 1988; Hirose & Kawamoto, 1995), liquidus
phase relationships (e.g. Yoder & Tilley, 1962; Yoder,
1965; Holloway & Burnham, 1972; Helz, 1973, 1976;
Michael & Chase, 1987; Gaetani et al., 1993; Sisson &
Grove, 1993), physical properties of melts (e.g. Lange,
1994; Watson, 1994), and eruptive style (Vergniolle &
Jaupart, 1986, 1990; Head & Wilson, 1987; Greenland
et al., 1988; Cashman & Mangan, 1994; Sparks et al.,
1994). Studies of oceanic island basalts have suggested
mantle source heterogeneities with respect to major and
trace elements and radiogenic isotopes, but because of
extensive degassing of these subaerial lavas, heterogeneities in mantle volatile contents are poorly constrained (e.g. Jambon, 1994). Determination of heterogeneities (or lack thereof ) in mantle volatiles is critical
to fully characterize mantle reservoirs, to compare the
role of volatiles in igneous processes in different tectonic
environments, and to constrain global cycling of these
elements.
It might seem that the best way to investigate mantle
volatiles would be to study its most voluminous product,
namely tholeiitic mid-ocean ridge basalt (MORB). This
logic is valid for water because the solubility of water in
basaltic liquids is high, hence degassing of water from
MORB erupted deeper than ~500 m is not significant
(Moore & Schilling, 1973; Moore et al., 1977; Moore,
1979; Jambon & Zimmermann, 1987; Dixon & Stolper,
1995). However, a different approach is required for
carbon because its solubility is much lower and most
MORB loses significant amounts (30–95%) of initial
carbon during residence in shallow crustal reservoirs
912
DIXON et al.
VOLATILES IN NORTH ARCH ALKALIC BASALTS
1995). These models have recently been extended to
include alkalic basalts (Dixon, 1997) by incorporation
of compositional parameterizations of CO2 and H2O
solubility into the degassing model. We define degassing
as the exsolution of volatiles from magma, but not necessarily subsequent loss from the system. Degassing can
occur as a closed system process, in which the vapor
remains in contact and in equilibrium with the melt
throughout its degassing history, or an open system
process, in which the vapor is instantaneously removed
from the melt.
In this study, we present dissolved H2O and CO2
concentrations in glasses from the North Arch volcanic
field, a submarine suite of alkalic to strongly alkalic lavas
erupted on the seafloor north of Oahu (Clague et al.,
1990). These alkalic lavas were all collected at water
depths of 3900–4380 m (eruption pressure ~400±40
bar) and provide a unique opportunity to investigate the
role of volatiles in the generation and evolution of alkalic
series magmas. Based on these data and degassing models
suitable for alkalic basalts, we model the effects of degassing and estimate (1) the concentration of volatiles in
plausible primary melts, and (2) the concentration of
volatiles in the mantle source.
We also present chlorine and noble gas concentrations,
and Fe3+/RFe, which can be used to estimate magmatic
oxygen fugacity. These data provide insight into sulfide
saturation of the melts, degassing of sulfur and noble
gases, and the dependence of oxidation state on extent
of partial melting.
Fig. 1. Simplified map of the lava flow field north of Oahu based on
GLORIA imagery showing the sample locations. Χ, Dredges; Ε,
cores. Map modified from Clague et al. (1990).
GEOLOGICAL BACKGROUND,
SAMPLE LOCATIONS AND
DESCRIPTIONS
The North Arch volcanic field is located north of Oahu
on the Hawaiian Arch, a 200 m high flexural arch formed
in response to loading of the Hawaiian Islands (Moore,
1970). Young lava flows cover an area of ~25 000 km2
at depths of 3900–4380 m (Fig. 1). Clague et al. (1990)
described the locations, petrography, and major and
minor element compositions of samples from 26 locations
within the volcanic field. The area consists of extensive
flat lava flows covered by >1 m of sediment within which
are isolated small (50–200 m high) hills interpreted as vent
structures. The lavas, with one exception, are estimated to
have erupted at 0·5–1·15 Ma, on the basis of stratigraphy,
sediment thicknesses [sedimentation rates determined by
paleomagnetic data on nearby cores (Rees, 1991)], and
palagonite thicknesses. One sample is estimated to have
erupted >1·6 Ma ago.
Several physical characteristics of the recovered
samples are specifically relevant to this study. Dredges
of the flat flows recovered only small, flat (2–3 mm thick)
chips of glass with palagonite alteration rinds on both
sides and up to 1 mm of Mn-oxide deposits usually on
one side only. All the flow samples are nonvesicular
(<0·1% vesicles) and are interpreted to be the outer
spalled glassy rims of sheetflows. In contrast, dredging of
the vent structures (dredges 21, 22, 23, 24, 26, 27, and
29) recovered mainly pillow joint blocks of vesicular
basalt and blocks of layered hydroclastite and volcanic
breccia. Even though these lavas were dredged at ~4000
m water depth, some vent lavas have vesicularities up to
57% (Table 1, Fig. 2a). Also, glass spheres <0·5 mm in
diameter are present in some of the hydroclastites found at
the vents (Fig. 2b). These glass spheres are compositionally
similar to lavas in the same dredge. Glass spheres from
other and shallower submarine locations (Vallier et al.,
1977; Melson et al., 1988; Smith & Batiza, 1989) have
been interpreted to have formed during submarine fountaining, resulting from rapid exsolution of large volumes
of gas during eruption.
913
JOURNAL OF PETROLOGY
VOLUME 38
NUMBER 7
JULY 1997
Table 1: Major and volatile element data and parental magma compositions
Sample:
29D
23D
21D-b
24D-a
24D-b
27D-a
27D-b
Type:
neph.
neph.
neph.
neph.
neph.
neph.
basanite
36D-a
basanite
Style:
vent?
vent
vent
vent
vent
vent
vent
flow
Glass analyses 1
SiO2
40·5
42·0
41·7
42·7
42·8
42·6
43·6
43·7
Al2O3
12·9
14·4
13·7
14·1
14·8
13·6
14·6
14·7
FeO
12·5
12·3
11·7
12·2
12·0
12·6
11·8
11·9
MnO
0·19
0·19
0·17
0·20
0·20
0·20
0·19
MgO
7·22
6·48
7·06
8·06
7·19
7·87
7·07
CaO
13·3
13·4
14·0
13·2
13·7
13·0
13·2
0·20
6·43
13·2
Na2O
4·84
4·62
4·40
4·30
4·43
4·13
3·97
K 2O
1·31
1·38
1·39
0·98
1·00
1·29
1·14
0·98
P2O5
0·82
0·71
0·71
0·65
0·66
0·59
0·55
0·55
TiO2
2·80
2·95
2·53
2·10
2·15
2·71
2·36
2·53
S
0·204
0·138
0·195
0·120
0·115
0·122
0·169
0·209
Cl
0·136
0·107
0·116
0·097
0·102
0·115
0·108
0·085
H2O (total)
0·96(5)
0·98(4)
1·31(3)
0·93(9)
1·00(6)
0·83(11)
1·02(5)
4·03
1·29(6)
CO2 (p.p.m.)
420(20)
800(40)
470(30)
600(40)
650(30)
630(90)
460(10)
440(30)
Total
97·7
99·7
99·0
99·5
100·2
99·7
99·8
99·5
Molecular H2O (wt %)
Fe3+/RFe
0·14(3)
n.a.
0·12(1)
0·20
0·28(4)
n.a.
DFMQ
n.c.
+0·2
% vesicles2
n.a.
35
2
n.c.
0·13(1)
0·16(1)
0·11(3)
0·17(2)
0·19
0·22
0·23
0·16
+0·1
+0·5
+0·7
−0·2
57
35
27
27
0·27(1)
n.a.
n.c.
<0·1
Parental magma compositions 3
SiO2
40·5
n.c.
41·5
42·3
42·3
42·1
42·9
Al2O3
10·3
n.c.
11·0
11·3
11·6
10·7
11·3
n.c.
FeO
12·3
n.c.
11·7
12·1
12·2
12·4
11·9
n.c.
MnO
0·20
n.c.
0·18
0·21
0·20
0·21
0·20
n.c.
n.c.
MgO
15·4
n.c.
15·1
16·0
15·8
16·4
16·1
n.c.
CaO
10·6
n.c.
11·2
10·6
10·9
10·2
10·2
n.c.
Na2O
3·86
n.c.
3·53
3·44
3·48
3·24
3·08
K 2O
1·05
n.c.
1·12
0·79
0·79
1·01
0·88
n.c.
P2O5
0·65
n.c.
0·57
0·52
0·52
0·46
0·43
n.c.
2·24
n.c.
TiO2
% ol added
25
n.c.
2·03
1·68
25
25
1·69
27
2·13
27
1·83
29
n.c.
n.c.
n.c.
Init. H2O4
1·95
n.c.
1·71
1·56
1·56
1·38
1·29
Init. CO24
4·9 (2·9)
n.c.
4·3 (2·5)
3·9 (2·3)
3·9 (2·3)
3·5 (2·0)
3·2 (2·0)
n.c.
Init. Cl4
0·11
n.c.
0·10
0·09
0·09
0·08
0·07
n.c.
ANALYTICAL TECHNIQUES
Infrared spectroscopy
Concentrations of dissolved water and carbon dioxide
were measured using IR spectroscopy. Glass chips were
doubly polished to a thickness of ~50–200 lm. The
position and size of the beam were controlled by placing
each glass chip over a 200 lm aperture. Transmission
IR spectra in the 4000–1200 cm–1 (2·5–8·3 lm) range
were collected using the microchamber on a Nicolet
60SX FTIR (Fourier transform infrared) spectrometer,
914
n.c.
a globar source, a KBr beamsplitter, an HgCdTe detector, and a mirror velocity of 1·57 cm/s. Typically, 4096
scans were collected for each spectrum. The spectrum of
a decarbonated basanite sample (1297D) or tholeiite
(TT152-21-35D) was subtracted from the sample spectra
as a background correction. Absorbance measurements
for the molecular water (1630 cm–1) and carbonate (1515
and 1430 cm–1) bands were made on reference subtracted
spectra. Determination of concentrations was done
through Beer–Lambert law calibration [see review by
Ihinger et al. (1994)]. The thickness, or path length, is
DIXON et al.
VOLATILES IN NORTH ARCH ALKALIC BASALTS
Table 1: continued
Sample:
26D-a
20G
9D
36D-b
17D-a
22D
Type:
basanite
basanite
basanite
AOB
AOB
AOB
RB02-a
AOB
Style:
vent
flow
flow
flow
flow
vent
vent?
Glass analyses 1
SiO2
43·6
43·4
44·9
46·0
45·0
47·6
47·8
Al2O3
14·5
13·9
14·6
14·6
14·2
15·1
15·0
FeO
11·7
12·0
11·8
11·3
11·4
11·1
11·1
MnO
0·19
0·19
0·18
0·19
0·16
0·18
MgO
7·01
7·89
7·55
7·33
8·71
8·29
CaO
13·2
12·2
12·3
12·4
11·6
11·7
0·17
7·02
11·9
Na2O
3·97
4·06
3·81
3·55
3·44
3·18
K 2O
1·12
0·97
0·85
0·74
0·78
0·38
0·42
P2O5
0·50
0·48
0·38
0·34
0·32
0·25
0·23
TiO2
2·36
2·24
2·16
1·88
1·89
1·36
1·62
S
0·163
0·117
0·137
0·146
0·133
0·139
0·135
Cl
0·108
0·076
0·085
0·063
0·055
0·045
0·040
H 2O
1·02(0)
1·42(3)
0·91(4)
0·97(5)
0·96(7)
0·69(3)
3·13
0·70(2)
CO2 (p.p.m.)
450(20)
380(50)
390(30)
330(20)
280(10)
260(20)
280(20)
Total
99·0
99·0
99·7
99·5
98·7
100·0
99·3
Molecular H2O (wt %)
0·17(2)
0·37(5)
0·12(1)
0·20(2)
0·13(1)
0·07(1)
0·12
0·14
0·13
0·13
0·10(1)
Fe3+/RFe
n.a.
DFMQ
n.c.
n.c.
−0·8
−0·5
−0·6
−0·4
55
n.a.
<0·1
<0·1
<0·1
<0·1
<0·1
% vesicles2
n.a.
n.a.
n.c.
Parental magma compositions 3
SiO2
42·9
42·8
43·9
44·7
44·1
46·1
45·9
Al2O3
11·2
11·0
11·3
11·2
11·4
11·9
11·2
FeO
11·8
11·9
11·8
11·5
11·4
11·2
11·4
MnO
0·20
MgO
16·1
CaO
10·2
Na2O
3·08
0·20
16·3
0·19
16·6
0·20
16·5
0·17
16·3
0·19
16·7
0·18
17·1
9·61
9·50
9·55
9·35
9·20
8·91
3·20
2·94
2·74
2·77
2·50
2·35
K 2O
0·87
0·76
0·66
0·57
0·63
0·30
0·32
P2O5
0·39
0·38
0·29
0·26
0·26
0·20
0·17
TiO2
% ol added
1·83
29
1·77
27
1·67
30
1·45
30
1·52
24
1·07
27
1·21
34
Init. H2O4
1·17
1·14
0·87
0·78
0·78
0·60
Init. CO24
2·9 (1·8)
2·8 (1·8)
2·1 (1·4)
2·0 (1·1)
2·0 (1·1)
1·5 (0·9)
1·3 (0·8)
Init. Cl4
0·07
0·06
0·05
0·04
0·04
0·03
0·03
0·51
n.a., not analyzed; n.c., not calculated; values in parentheses are 1r error in the last or last two decimal places.
1
Glass analyses from Clague et al. (1990).
2
Vesicle contents were measured using image analysis to determine area percent vesicles that were corrected to volume
percent vesicles by multiplying by a factor of 1·18 (Cashman & March, 1988; Mangan et al., 1993).
3
Compositions of parental magmas were calculated from glass compositions having >7 wt % MgO by addition of 0·1%
increments of equilibrium olivine using the average measured log fO2 of FMQ until the liquidus olivine is Fo91 (see Stolper
& Newman, 1994).
4
Initial H2O, CO2, and Cl contents in the parental magmas are calculated from the P2O5 contents using H2O=3×P2O5, CO2=
2·5±1·5×H2O, and Cl=0·17×P2O5 (see discussion of degassing models).
915
JOURNAL OF PETROLOGY
VOLUME 38
NUMBER 7
JULY 1997
Fig. 2. (a) Photomicrograph of vesicular pillow lava with 55 vol. % vesicles from dredge F1188HW-26D. Field of view is 5 mm wide. Four
nephelinites and two basanites collected from vent structures are highly vesicular with 27–55 vol. % vesicles; the remaining lavas have Ζ2 vol.
% vesicles. (b) Photomicrograph of a sphere of glass in hydroclastite from dredge 26D. These glass spheres may represent lava injected into the
water column during fountaining at 4000 m water depth. Sphere diameter is ~0·3 mm.
measured by a digital micrometer with a precision of
±1–2 lm. The glass density was calculated for each
sample using the Gladstone–Dale rule and the Church–
Johnson equation as described by Silver et al. (1990).
In this study we are examining a range of glass compositions, therefore the compositional dependence of the
molar absorptivities (proportionality constants between
the measured absorbances and the concentrations) must
be taken into account. The molar absorptivity for total
dissolved water using the fundamental OH stretching
band at 3535 cm–1 is not strongly compositionally dependent for basaltic compositions and we use a value of
63±5 L/mol cm (P. Dobson, S. Newman, S. Epstein &
E. Stolper, unpublished results). Dixon et al. (1995) showed
that the molar absorptivity for molecular water, however,
decreases as the proportion of tetrahedral cations decreases. Using their derived linear relation between the
value of molar absorptivity of the 5200 cm–1 band (e5200)
and the mole fraction of tetrahedral cations (Si4+, Al3+,
Fe3+, and Ti4+), and assuming a constant e1630/e5200 of
916
DIXON et al.
VOLATILES IN NORTH ARCH ALKALIC BASALTS
40·3, we predict a value of 19±3 L/mol cm for e1630 for
the North Arch glasses (24% lower than the value of
25±3 L/mol cm determined for tholeiitic basalt). The
molar absorptivity for carbon dissolved as carbonate in
silicate glasses also varies as a function of composition.
Dixon & Pan (1995) determined that the ratio of sodium
to calcium in the glass can be used to predict the molar
absorptivity of carbonate in basaltic glasses. Based on
their relation e1525=451·2 – 341·8[Na/(Na+Ca)], and
a range of Na/(Na+Ca) in the North Arch glasses of
0·32–0·40, we calculate an average molar absorptivity of
330±20 L/mol cm for CO2 dissolved as carbonate in the
North Arch glasses. Because concentrations are inversely
proportional to the molar absorptivity, our reported
concentrations are ~14% higher than if the tholeiitic
molar absorptivity (375±20) were used.
Precision of the analyses is about ±2% for total water
and ±15% for molecular water and carbonate. The
accuracy of the total water analyses is the same as
reported by Dixon et al. (1991) (about ±10%). Because
of the larger uncertainty in the compositional dependence
of the molar absorptivity for carbon dissolved as carbonate in silicate glasses, the accuracy of the CO2 analyses
is estimated to be ~20%.
Electron microprobe
Concentrations of sulfur and chlorine were determined
on a nine-spectrometer ARL electron microprobe using
natural and synthetic standards and instrumental parameters described by Clague et al. (1995). Mean-atomicnumber calculations, based on the backgrounds measured
on high and low mean-atomic-number standards, were
used to obtain the background counts. Sulfur analyses
determined by these procedures are consistent with those
measured by electron probe in MORB and seamount
glasses (Wallace & Carmichael, 1992), based on interlaboratory comparison of glasses from Loihi seamount
(D. Clague, unpublished data). Error in the S analyses
is estimated to be ±6%, based on analysis of a standard
with comparable S content (mean of 14 analyses on
standard VG-2 is 0·127±0·008 wt %, ±6% relative).
Error in the Cl analyses is estimated to be ±8% based
on analysis of standards A-99 (mean of 17 analyses is
0·024±0·002 wt %, ±8% relative) and VG-2 (mean of
12 analyses is 0·031±0·002 wt %, ±7% relative).
Determination of Fe2O3 and FeO
Determinations of Fe2O3 and FeO were made on nine
glass samples ranging in composition from nephelinitic
to alkali olivine basaltic using splits from the material
analyzed by electron probe and FTIR spectrometry.
Small chips of glass were handpicked so as to avoid any
oxidation, alteration, or vesicular portions. Wet chemical
measurements of ferrous iron were made by I. S. E.
Carmichael using techniques described by Christie et al.
(1986). The precision of this technique is generally better
than ±1% relative. Fe2O3 for each of the samples was
calculated by difference using the wet chemical ferrous
iron determination and the electron probe measurement
of total iron (FeOT). Uncertainties in the determination
of FeO and Fe2O3 combine to yield an error in Fe3+/
RFe of ~5% relative (Christie et al., 1986).
Noble gases
Because of the extreme vesicularity of many of the
samples and the difficulty in obtaining large quantities
of clean glass without palagonite or patches of devitrified
areas, only a small subset of the glasses were suitable for
noble gas analyses. All of the analyses were performed
at the University of Rochester on gases extracted by both
vacuum crushing and total fusion techniques (Poreda &
Farley, 1992). Data are listed in Table 2. Errors in the
3
He/4He ratio result from two effects: uncertainty in
counting statistics for the 3He peak and uncertainty in
the correction for blank and air helium. For samples that
contain >10 ncc/g He and have He/Ne ratios >50, the
error in the measured 3He/4He ratio is ±2%. For samples
that contain between 4 and 10 ncc/g, errors average
about ±5%. Samples that contain 1–4 ncc/g have errors
of ±10%. When the He/Ne ratio is <10, there is a
substantial (>3%) correction to the 3He/4He ratio for
the addition of atmospheric helium. The conservative
estimate of the amount of air helium uses the ratio of
the measured He/Ne ratio to the He/Ne ratio in air.
Errors in this simple model can add considerable uncertainty as the He/Ne ratio decreases below ten.
RESULTS
Review of petrography and major and
minor element compositions
Clague et al. (1990) presented major and minor element
compositions of the North Arch Lavas and showed that:
(1) the flows consist of alkalic series lavas ranging from
alkali olivine basalts to nephelinites; (2) olivine is the only
phenocryst mineral in most of the lavas, but clinopyroxene phenocrysts occur in several evolved samples;
and (3) the compositional range can be generated by
variable degrees of partial melting of sources similar to
or slightly more depleted than those inferred for the
Koloa Volcanics (Clague & Dalrymple, 1988), with alkali
olivine basalt representing the highest degree of melting
and nephelinite representing the lowest degree of melting.
We have included in Table 1 the previously published
917
JOURNAL OF PETROLOGY
VOLUME 38
NUMBER 7
JULY 1997
Table 2: Rare gas data
Sample
Temp.
(°C)
7D
17D-a
22D
26D
4
He/ He
4
22
36
84
132
(ncc/g)
(pcc/g)
(pcc/g)
(pcc/g)
(pcc/g)
6·5
3·3
440
48·8
1·14
363
He
Ne
800
3·1
7·9
5·9
1600
5·5
6·2
6·4
Total
4·15
14·1
Crush
7·6
800
4·4
1600
7·6
4·1
Total
4·68
46·1
4·5
42
Kr
Xe
40
Ar/36Ar
38
Ar/36Ar
0·1897
135
7·6
33
3·6
0·98
355
8·8
195
51·7
2·36
334
228
0·1873
338
6·6
800
7·7
224
13·4
361
3·5
0·5
301
0·1884
1600
6·6
54
6·9
136
2·1
1·53
397
0·1885
Total
7·48
278
Crush
8·1
58
85
800
8·5
1740
87
1600
5·5
Total
8·49
bl
20·9
Ar
Crush
Crush
35D
R/R A
3
1·9
35
497
3800
8·1
156
1742
327
22·5
3·6
793
0·1867
48
0·1
638
0·1908
3956
787
bl
800
2·3
11·2
14·9
252
5·2
0·51
335
0·1889
1600
4·7
2·9
5·8
148
5·8
0·29
408
0·1887
Total
2·79
14·1
400
Crush
6·2
6·1
95
13D
Crush
2·2
2·9
>50
21D
Crush
5·4
5·4
180
23D
Crush
1·3
1·4
120
34D
Crush
5·3
5·6
140
36D
Crush
7·9
6·4
48
362
Noble gas results for the North Arch lavas. Gases were released from the glass by either vacuum crushing or a two-step
heating technique that used previously reported procedures (Poreda & Farley, 1992). 7D and 35D are similar to alkali olivine
basalt sample 17D-a. 13D is similar to basanite sample 9D. 34D is similar to alkali olivine basalt sample 36D-b.
microprobe analyses (Clague et al., 1990) of those glasses
which we analyzed for volatiles in this study.
Systematic trends in nonvolatile minor elements provide a framework for the examination of volatile elements.
The negative correlation between P2O5 and SiO2 is
shown in Fig. 3a. Even without correction for fractional
crystallization (which causes a positive correlation between P2O5 and SiO2 for magmas of these compositions),
the data can be fitted by a line (P2O5=4·33 – 0·088SiO2)
allowing the P2O5 content to be predicted as a simple
function of SiO2. Both potassium and phosphorus display
highly incompatible element behavior in the North Arch
glasses and their ratio (K2O/P2O5) remains essentially
constant with a value of ~2 over the range of compositions
(Fig. 3b). As argued by Clague et al. (1990), the negative
correlations between SiO2 and incompatible trace element concentrations are consistent with generation by
variable extents of melting of a common homogeneous
source.
918
Water
Volatile contents are listed in Table 1. Concentrations
of total dissolved water range from 0·69 to 1·42 wt
%. In contrast to the well-correlated behavior of the
nonvolatile incompatible elements, concentrations of
water do not correlate with SiO2 or P2O5. This is surprising because previous work on mid-oceanic ridge and
Hawaiian tholeiitic basalts has shown that water usually
behaves incompatibly during melting and crystallization
processes with a distribution coefficient similar to that of
Ce or P2O5 (~0·01) (Michael, 1988, 1995; Dixon et al.,
1988). A plot of H2O vs P2O5 for the North Arch glasses
DIXON et al.
VOLATILES IN NORTH ARCH ALKALIC BASALTS
Fig. 3. (a) Negative correlation between P2O5 and SiO2 analyzed in
North Arch glasses (filled symbols) can be fitted by a line (P2O5=4·33
– 0·088SiO2) allowing the P2O5 content to be predicted as a simple
function of SiO2. Χ, alkali olivine basalts; Ε, basanites; Ο, nephelinites.
Parental magma compositions (open symbols) were calculated from
glass compositions by addition of 0·1% increments of equilibrium
olivine assuming a log f O2 of FMQ until the liquidus olivine is Fo91
(Stolper & Newman, 1994). Mass of olivine added was 28±4 wt %.
(b) Positive correlation between K2O and P2O5 in the North Arch
glasses. K2O/P2O5 is essentially constant with a value of ~2. These
trends in nonvolatile, incompatible trace element concentrations are
consistent with generation by variable extents of melting of a homogeneous source region.
shows considerable scatter (Fig. 4a), with the ratio of
H2O/P2O5 varying from three in the alkali olivine basalts
to 1·3 in the nephelinites. In general, samples with low
H2O/P2O5 ratios are also highly vesicular.
Water concentrations in many of the same samples
were determined previously on glassy whole-rock samples
by weight loss (Clague et al., 1990; see caption to Fig.
4b). In general, water concentrations based on analyses
of glassy whole-rock samples are higher than those based
on IR analyses (Fig. 4b). Given the age and condition of
these samples, it is likely that the higher and more variable
water concentrations measured on bulk samples reflect
inclusion of small amounts of altered material.
Most dissolved water occurs as hydroxyl groups, consistent with high-temperature speciation curves for basaltic glasses (Dixon et al., 1995). All samples contain
919
Fig. 4. (a) Concentrations of water plotted against P2O5. Water concentrations in the North Arch glasses show considerable scatter, with
the ratio of H2O/P2O5 varying from three in the alkali olivine basalts
to 1·3 in the nephelinites. Symbols are the same as in Fig. 3. (b) H2O+
determined on glassy, whole-rock samples plotted against H2O in glass
(this study). Continuous line is 1:1 line. H2O+ analyses of whole-rock
samples (Clague et al., 1990) were determined by weight loss after
heating 50 mg of sample with 150 mg of lead oxide and lead chromate
flux at 900–950°C and then subtracting the water lost by heating at
110°C ( Jackson et al., 1987). (c) Concentrations of molecular water
were determined using the height of the 1630 cm–1 band and a molar
absorptivity for molecular water of 19±3 L/mol cm (see text). These
data are consistent with the experimentally determined speciation
model for water in tholeiitic glass (Dixon et al., 1995).
JOURNAL OF PETROLOGY
VOLUME 38
small amounts of molecular water (0·07–0·37 wt %; Fig.
4c). Concentrations of molecular water in most glasses
(13 of 15 samples) plot within error (±15%) of the highT water speciation curve for tholeiitic basaltic glasses
(Dixon et al., 1995). The good agreement between observed and experimentally determined speciation model
for water in tholeiitic glass (Dixon et al., 1995) confirms
that this speciation model is valid over the range of
compositions studied here and that areas of glass analyzed
by IR spectroscopy were unaltered in most samples.
Two samples (D36-b and RBO2-a), however, have
molecular water concentrations slightly greater than that
predicted by the high-T speciation model. These excesses
in molecular water (0·06 and 0·03 wt % H2O, respectively,
greater than the predicted concentration) are small relative to the total water concentrations (0·97 and 0·70 wt
% H2O, respectively) and will not affect the conclusions
of this study. Theoretically, molecular water concentrations in glasses may be modified by equilibration
at lower temperatures or by low-temperature addition of
water (Pandya et al., 1992; Dixon et al., 1995). Though
determination of the T dependence (and consequently
quench rate dependence) of water speciation in silicate
glasses is currently an area of active research (e.g. McMillan, 1994), we do not consider the quench rate dependence of water speciation to be a significant source
of error in this study because (1) experimentally quenched
basaltic glasses produced by quench rates differing by
two orders of magnitude did not produce a detectable
difference in water speciation (Dixon et al., 1995) and (2)
water speciation in most of the North Arch glasses is
within error of the predicted speciation curve (Dixon et
al., 1995). Given the age and generally altered state of
the North Arch glasses, we consider the most likely
explanation for the excess of molecular water in these
two glasses to be minor low-temperature hydration.
Carbon dioxide
Concentrations of carbon dioxide in these glasses range
from 260 to 800 p.p.m. (Fig. 5a). No absorptions at 2350
cm–1 were observed, indicating that carbon is dissolved
only as carbonate and not as molecular CO2, consistent
with previous work on mafic silicate melts (Fine & Stolper,
1986; Blank & Brooker, 1994). The concentration of
dissolved CO2 increases with the degree of SiO2 undersaturation of the lavas and increasing P2O5 content (Fig.
5a), reflecting the strong compositional dependence of
CO2 solubility in basaltic melts (Blank & Brooker, 1994;
Holloway & Blank, 1994; Dixon, 1997).
Some of the same samples were analyzed previously
for CO2 on glassy, whole-rock samples (Clague et al.,
1990) using a coulometric technique ( Jackson et al.,
1987). Concentrations measured on glassy, whole-rock
920
NUMBER 7
JULY 1997
Fig. 5. (a) Concentration of CO2 (p.p.m.) dissolved as carbonate in
North Arch glasses plotted against the P2O5 content, showing increase
in CO2 with increasing SiO2 undersaturation. Symbols are the same
as in Fig. 3. (b) CO2 contents in whole-rock samples (Clague et al.,
1990) vs CO2 content in glasses (this study). Continuous line is the
1:1 line. CO2 contents in whole-rock samples were determined
coulometrically ( Jackson et al., 1987) and are lower than the IR analyses
of glasses in all but two samples.
samples ranged from 200 to 500 (±100) p.p.m. CO2,
lower than the concentrations in glasses presented here
in all but two samples (Fig. 5b). CO2 concentrations
from IR analyses of glasses are usually lower than those
obtained on bulk glass samples, because IR spectroscopy
measures only the carbon dissolved in the glass, whereas
bulk glass samples are likely to contain additional carbon
in vesicles or adsorbed onto surfaces. In the case of the
North Arch glasses, however, the bulk glass samples
probably contain patches of devitrified glass from which
CO2 may have been lost, thus yielding lower total CO2
contents than IR analyses of fresh glassy areas.
Chlorine
Chlorine contents range from 400 to 1360 p.p.m. and
correlate positively with P2O5 (Fig. 6) with a constant
Cl/P2O5 ratio of 0·17±0·02. Concentrations of Cl in
the North Arch lavas do not appear to be affected by
degassing. This is consistent with Cl being more soluble
DIXON et al.
VOLATILES IN NORTH ARCH ALKALIC BASALTS
buffer has DFMQ=+0·7. Values of DFMQ for the
North Arch volcanics range from –0·8 for the alkali
olivine basaltic glasses to DFMQ=+0·7 in the nephelinitic glasses (Fig. 7c). These values are similar to
those for submarine glasses from Loihi seamount
(DFMQ=–1·7 to +1·1).
Sulfur
Fig. 6. Cl vs P2O5 for North Arch glasses. Symbols are the same as in
Fig. 3. Cl correlates positively with P2O5 with a constant ratio of
0·17±0·02.
than H2O in basaltic liquids and is also consistent with
Cl solubility increasing with decreasing SiO2 (Iwasaki &
Katsura, 1967).
Fe3+/RFe and relative magmatic oxygen
fugacities
Values of Fe3+/RFe range from 0·12 in the alkali olivine
basaltic glasses to 0·23 in the nephelinitic glasses (Fig.
7a). These values are on average ~0·06 lower than those
reported by Clague et al. (1990) for glassy whole-rock
samples (Fig. 7b). Similar differences have been documented for MORB glasses and cogenetic pillow interiors
(Christie et al., 1986), and have been attributed to hydrogen loss from the pillow interiors during cooling and
crystallization. For the North Arch glasses, there is a
positive correlation (r 2=0·74) between Fe3+/RFe and
P2O5 content (Fig. 7a). This pattern is consistent with the
observation of Clague et al. (1990) that whole-rock Fe3+/
RFe increases with P2O5. These values are higher than
those typical of MORB glasses, which have average
Fe3+/RFe<0·1 (Christie et al., 1986), but are similar to
the values of 0·08–0·19 for submarine glasses from Loihi
seamount (Wallace & Carmichael, 1992).
Magmatic oxygen fugacities for these quenched glasses
can be estimated using the experimentally calibrated
relationship between oxygen fugacity, bulk composition,
temperature, and Fe3+/RFe (Sack et al., 1980; Kress &
Carmichael, 1991). Calculated values of f O2 are strongly
dependent on the assumed temperature of quenching.
However, by expressing the result relative to the oxygen
fugacity of the fayalite–magnetite–quartz (FMQ) buffer
curve at the same temperature, this strong temperature
dependence can be circumvented. Relative oxygen fugacity is then defined as DFMQ=log f O2 sample – log
f O2 FMQ (Carmichael & Ghiorso, 1986) and is arbitrarily
calculated at 1200°C. For reference, the Ni–NiO oxygen
The North Arch glasses have concentrations of sulfur
ranging from 0·11 to 0·22 wt %. As noted by Clague et
al. (1990), S concentrations in these glasses are greater
than those in MORB with comparable FeOT contents.
In this respect they are similar to submarine glasses from
Loihi seamount, which have S contents ranging from
0·10 to 0·19 wt %. The higher S contents relative to
MORB of Loihi and North Arch glasses are consistent
with their higher oxygen fugacities. For a basaltic melt
that is saturated with an immiscible Fe–S–O liquid phase,
an increase in oxygen fugacity increases the amount of
dissolved S that is necessary to saturate the melt (Fig. 7d;
Wallace & Carmichael, 1992). The dissolved S contents of
MORB glasses are controlled by saturation with Fe–S–O
liquid. The higher oxygen fugacities of the North Arch
melts relative to MORB would therefore result in increased S solubility for melts with comparable FeOT.
Noble gases
Concentrations and isotopic compositions
Concentrations of noble gases (Table 2) are extremely
low. For example, 4He concentrations in glasses range
from 10–8 to 10–6 cm3/g or 10–1000 times lower than
most MORB glass [(2–5)×10–5 cm3/g; Kurz & Jenkins,
1981; Sarda & Graham, 1990]. These low concentrations
imply that these glasses have undergone extensive magmatic degassing. However, because the noble gases were
measured on bulk glass samples (unlike H2O, CO2, Cl,
and S, which were measured using microbeam techniques), we cannot rule out the possibility that inclusion
of patches of devitrified glass contributed to the low He
values.
3
He/4He in the vesicle gases (crush samples) in samples
with >5 ncc/g of helium ranges from five to eight times
the air ratio (R A), typical of values obtained for posterosional Hawaiian lavas (e.g. Craig & Poreda, 1986)
and lower than values of ‘high 3He/4He’ plume helium
(R/R A=32) observed in Loihi and other Hawaiian volcanoes (Rison & Craig, 1983; Kurz et al., 1983). 3He/
4
He in the vesicle gases in samples with <5 ncc/g are
less reliable (see Rison & Craig, 1983). 3He/4He ratios
in the gas released by the two-step fusion procedure (melt
fraction) are typically lower than the values for the
corresponding crush analysis. Previous work has shown
921
JOURNAL OF PETROLOGY
VOLUME 38
NUMBER 7
JULY 1997
Fig. 7. (a) Fe3+/RFe vs P2O5 showing positive correlation (r 2=0·74). Symbols are the same as in Fig. 3. (b) Fe3+/RFe (whole rock, Clague et
al., 1990) vs Fe3+/RFe (this study). Continuous line is 1:1 line. Values determined on carefully hand-picked glass chips (this study) are ~0·06
lower than those reported by Clague et al. (1990) for glassy, whole-rock samples. (c) DFMQ vs P2O5. Relative oxygen fugacity is defined as
DFMQ=log f O2 sample – log f O2 FMQ and is calculated at 1200°C (Carmichael & Ghiorso, 1986). (d) S vs DFMQ. Continuous line is the calculated
sulfide liquid solubility at 1200°C and 1170°C, which represent the average temperatures of the alkali basalts and nephelinites, respectively, in
melts that are saturated with immiscible Fe–S–O liquid (Wallace & Carmichael, 1992). The dashed lines represent uncertainty of ±1 SD in
the calculated S saturation value. Error in the S analyses is estimated to be ±6% .
that as a sample ages, the 3He/4He dissolved in glass is
more affected by post eruptive 4He production by U and
Th decay compared with 3He/4He in vesicle gases, thus
resulting in isotopic disequilibrium between the melt and
crush fractions. Attempts to estimate the age for these
flows based on the helium isotope disequilibrium (Graham et al., 1987) were not successful because of incomplete
helium retention in the glass.
The isotopic compositions of the heavy noble gases
(Ne, Ar, Kr and Xe) in the melt fraction are almost
entirely atmospheric, except for a 20–30% excess of
radiogenic 40Ar ( 40Ar∗). The 36Ar contents of North Arch
glasses are comparable with or higher than the amount
seen in fresh MORB glass, even in samples that have
lost >99% of their 40Ar∗ (and mantle helium). The
ratios of heavy noble gases (Ne/Ar, Kr/Ar and Xe/Ar)
correspond to the solubility ratios of the gases in seawater
rather than the ratios in the atmosphere. It has been
shown that seawater contains the necessary concentration
of dissolved constituents to act as the source of atmospheric gases (Patterson et al., 1990; Honda et al.,
1991) and that this atmospheric component can be
incorporated during the devitrification process (e.g. Dymond & Hogan, 1973). The atmospheric noble gas
signature may reflect two processes: loss of mantle volatiles by magmatic degassing followed by incorporation
of seawater into the lava. The incorporation mechanism
may be complicated, but addition of as little as 0·1 wt
% seawater, probably added during minor devitrification,
can provide quantities of dissolved noble gases equal to
the concentrations observed in the lava.
922
EVALUATION OF VAPOR
SATURATION AND CALCULATION
OF VAPOR PHASE COMPOSITION
To model the degassing behavior of these magmas, we
need to first evaluate if they are vapor saturated with a
CO2–H2O vapor at the depth of eruption. In the case
of the vesicular samples, it is obvious that vapor saturation
DIXON et al.
VOLATILES IN NORTH ARCH ALKALIC BASALTS
was reached at some time during their ascent; however,
they could be supersaturated with respect to a CO2–H2O
vapor, as is observed for most MORB (Fine & Stolper,
1986; Stolper & Holloway, 1988; Dixon et al., 1988, 1995).
In the case of the nonvesicular samples, examination of
the dissolved H2O and CO2 contents provides a means
to check for vapor saturation.
CO2 and H2O concentrations in vapor-saturated melts
vary as a function of pressure, temperature and vapor
composition. Dixon (1997) presented compositional parameterizations of H2O and CO2 solubilities and used
these parameterizations to develop vapor saturation and
degassing models for alkalic basaltic liquids. Vapor saturation diagrams generated as a function of melt composition are used to determine the pressure at which
the melt was last in equilibrium with a vapor and the
composition of the vapor phase based on measured H2O
and CO2 contents in basaltic glasses. Specifically, from
tholeiitic through the nephelinitic compositions found in
the North Arch area, CO2 solubility is a strong function
of composition and increases by a factor of 5 (±15%),
whereas H2O solubility is a weak function of composition
and decreases by ~30% (±15%).
The H2O and CO2 data are plotted against the 400
bar vapor saturation curve (isobar) as a function of
composition in Fig. 8. Details of the calculation of equilibration pressure are given in the footnote to Table 3.
There is good agreement between the measured H2O
and CO2 contents of the glasses and the predicted vapor
saturation curves. One way to estimate the ‘goodness of
fit’ is to calculate the pressure of equilibration and compare it with the eruption pressure. All samples except
one (23D) have calculated equilibration pressures (mean
P equil=440±64 bar) within the estimated uncertainty of
the model (±30%; see footnote to Table 3) of P erupt=
400±40 bar. This agreement suggests that most of
the North Arch glasses were vapor saturated with an
H2O–CO2 vapor upon quenching on the seafloor at 400
bar.
The conclusion that the North Arch glasses are saturated with respect to an H2O–CO2 vapor is consistent
with a similar conclusion for other Hawaiian (Dixon et
al., 1991) and back-arc basin basalts (Newman, 1989,
1990), but contrasts with most MORB glasses, which are
supersaturated at their eruption depth with P equil>P erupt
by up to a factor of four (Fine & Stolper, 1986; Stolper
& Holloway, 1988; Dixon et al., 1988, 1995). Because
diffusion rates of H2O and CO2 in melts depend on the
water content, it may be that water-rich melts (e.g. ocean
island and back-arc basin basalts) are more likely to
maintain melt–vapor equilibrium on eruptive time scales
than dry melts (e.g. depleted MORB). Also, the higher
vesicularity of Hawaiian magmas relative to MORB
means larger vesicles and closer spacing between vesicles,
thus diffusing species would have less distance to travel
to reach a bubble.
The mean molar proportion of CO2 in the vapor
(X vapour
CO2 ) is 0·84 in the alkali basaltic glasses and 0·64–0·74
in the nephelinites and basanites depending on whether
m
0,m
it is calculated as X CO
/X CO
or (1 –X Hm2O/X H0,m2O) (Table
3). The small discrepancies between the two methods of
calculating X vapour
result from the larger uncertainty in
CO2
the CO2 analyses. Thus, the vapor phase in the nephelinites and basanites is slightly more H2O rich than
in the alkali olivine basalts, but the vapor phase remains
rich in CO2 throughout the entire compositional range.
2−
3
2−
3
CALCULATION OF BULK VOLATILE
CONTENTS IN VESICULAR SAMPLES
Bulk volatile contents (volatiles in glass plus vesicles) can
be used to infer volatile contents of undegassed magmas
if there has been no loss or gain of volatiles from the
system (e.g. Moore et al., 1977; Pineau & Javoy, 1983;
Des Marais & Moore, 1984; Gerlach, 1991; Graham &
Sarda, 1991). It is probable that some bubbles have
escaped, resulting in minimum estimates, though it is
possible for volatiles to be gained by accumulation of
bubbles. Vesicle contents for the North Arch lavas were
determined on the quenched glassy rims, therefore it is
especially important to evaluate how vesicle abundances
in glassy rims may be modified by eruption and flow.
We argue below that the vesicular vent samples have
bulk volatile contents very close to undegassed magmas,
whereas nonvesicular flow samples represent magmas
that have lost their exsolved gas.
In general, it is well documented that vesicularity in
tube-fed pahoehoe lava flows (probably the closest analog
to the North Arch sheet flows) decreases with transport
distance away from the vent (e.g. Swanson & Fabbi,
1973; Mangan et al., 1993; Cashman & Mangan, 1994;
Cashman et al., 1994). Constant H2O and S contents of
quenched glass in lava tube samples and in gas emitted
from the tube system indicate that open system, ‘passive’
rise and escape of larger bubbles to the lava surface
probably account for much of the decreasing vesicularity
with distance (Cashman et al., 1994).
The evolution of vesicle abundances and sizes, specifically in chilled margins on subaerial Hawaiian pahoehoe lava flows, has been described by Wilmoth &
Walker (1993). Spongy (S-type) pahoehoe is characterized
by >30% vesicles distributed roughly equally between
outer selvages and flow interiors. Vesicularity in glassy
margins of S-type pahoehoe flows is highest at the vent,
where it is interpreted that the original complement of
bubbles has been frozen in. Pipe vesicle-bearing (P-type)
pahoehoe lavas are characterized by lower vesicularity,
generally larger vesicles, and the occurrence of pipe
923
,
mol
2−
3
4d
Xm
CO ×10
924
0·81
Sat. indexh
0·078
0·149
6·55
N Hv Ok
v l
N CO
v m
M CO
59
99
81
%CO2 degassedo
%Total volatiles degassedp
2
%H2O degassedn
2
75
98
50
4·92
0·112
0·055
0·99
1·97
547
48
96
36
1·15
0·026
0·041
0·74
2·05
480
1·20
0·39
0·59
6·56
3·86
470
0·61
89·3
54·7
1·31
41·7
21D-b
76
99
48
5·29
0·120
0·048
0·86
1·79
450
1·13
0·72
0·84
5·87
4·94
600
0·28
94·2
26·8
0·93
42·7
24D-a
70
98
44
3·86
0·088
0·043
0·77
1·76
500
1·25
0·67
0·92
5·80
5·35
650
0·33
94·7
31·0
1·00
42·8
24D-b
84
99
55
8·43
0·192
0·055
0·99
1·82
439
1·10
0·78
0·88
5·94
5·20
630
0·22
93·7
20·9
0·83
42·6
27D-a
61
98
34
2·68
0·061
0·030
0·54
1·55
420
1·05
0·67
0·72
5·25
3·79
460
0·33
98·7
32·4
1·02
43·6
27D-b
28
93
17
0·57
0·013
0·014
0·25
1·53
486
1·16
0·71
0·64
5·18
3·33
440
0·52
99·1
51·5
1·29
43·7
36D-a
61
98
34
2·64
0·060
0·030
0·53
1·55
414
1·04
0·67
0·71
5·25
3·70
450
0·33
98·7
32·6
1·02
43·6
26D-a
16
86
12
0·23
0·005
0·010
0·19
1·61
495
1·24
0·34
0·58
5·39
3·11
380
0·66
97·7
64·4
1·42
43·4
20G
57
98
25
2·28
0·052
0·017
0·30
1·21
392
0·98
0·76
0·74
4·36
3·21
390
0·24
105
25·7
0·91
44·9
9D
—
—
0
—
—
—
0
0·92
499
1·25
0·74
0·99
3·60
3·55
430
0·26
110
29·1
0·97
46·0
36D-b
47
98
19
1·51
0·034
0·013
0·23
1·19
323
0·81
0·73
0·54
4·29
2·31
280
0·27
106
28·5
0·96
45·0
17D-a
39
97
7
0·97
0·022
0·003
0·06
0·75‡
391
0·98
0·88
0·86
2·51
2·15
260
0·12
118
14·5
0·69
47·6
22D
—
—
0
—
—
—
0
0·69‡
440
1·10
0·88
0·99
2·37
2·31
280
0·12
119
14·8
0·70
47·8
RB02-a
NUMBER 7
2
1·41
2
2·37
Calc init. H2Oi
h
1·37
0·67
1·0
6·35
6·59
800
0·33
90·8
30·0
0·98
42·0
23D
VOLUME 38
M Hv O (wt %)j
P equil
2
324
0·66
v
X CO
=1−XHv Og
2
0·47
2−
3
o, m f
Xm
CO / XCO
2−
3
7·38
3·46
420
0·34
X oCO, m ×104e
2−
3
c
mol
83·4
m
/Xo,
H O,
CO2 (p.p.m.)
X
2
m
H2O
2
×104b
mol
m
X o,
H O,
28·7
mol
Xm
H O,
2
0·96
40·5
29D
×104a
H2O (wt %)
SiO2
Vapor saturation calculation
Sample:
Table 3: Calculation of vapor composition, extent of degassing, and mantle CO2 contents
JOURNAL OF PETROLOGY
JULY 1997
C
D
fH2O(PO, TO)
(−Vo,H2mO ) (P-PO)
,
exp
fH2O(PO, TO)
RTO
925
2−
3
2−
3
m
m
XCO
(P, TO)=XCO
(PO, TO)
D
fCO2(P, TO)
(−DVo,r m)(P−PO)
,
exp
fCO2(PO, TO)
RTO
C
Theoretical mole fraction of carbonate in a melt in equilibrium with pure CO2 fluid at 400 bar is calculated using:
2−
2−
2−
3
m
o, m
3
3
CO2−
CO2−
o, m
m
3
H2O ,mol
CO2−
2−
3
2−
3
g
2−
3
2−
3
2−
3
2−
3
2−
3
2−
3
Assuming ideal mixing, X / X
represents the mole fraction of CO2 in the fluid phase.
m
m
At saturation, (XHm2O, mol / X
)+(X /XoCO, m )=1 (Dixon & Stolper, 1995), therefore (XCO
/ XoCO, m ) should be equivalent to (1−XHm2O, mol / Xo,
H2O , mol). Because analytical
errors for H2O are smaller than those for CO2, (1−XHm2O,mol/XoH,2Om,mol) is a more reliable estimate of the mole fraction of CO2 in the fluid.
h
m
m
o, m
The 400 bar saturation index (SI) is the sum (XHm2O,mol / Xo,
H2O , mol)+(XCO /XCO ), which equals unity if the melt is saturated, is >1 if the magma is supersaturated, and
m
m
o, m
<1 if the magma is undersaturated. Pequil=P at which (XHm2O, mol / Xo,
H2O , mol)+(XCO /XCO )=1. Uncertainty in calculated equilibrium pressure is estimated to be >±30%
based on the 1r uncertainties of the H2O analyses (>±5%), CO2 analyses (>±20%), and compositional dependence of H2O and CO2 solubilities (>±15%; Dixon,
1997).
i
Initial H2O contents for all samples except 22D and RBO2-a were calculated by first calculating P2O5 according to the linear fit to the P2O5 and SiO2 data (P2O5=
4·33–0·0875SiO2) and assuming H2O/P2O5=3. Initial H2O contents for 22D and RBO2-a were calculated by 3× the P2O5 content measured in the glass.
j
Mass of H2O in the vapor=initial H2O in the melt−measured H2O in the melt.
k
Number of moles of H2O in the vapor=MvH2O/18·015.
l
Number of moles of CO2 in the vapor=NvH2O×[(1−XvH2O)/XvH2O].
m
Mass of CO2 in the vapor=NvH2O×44·01.
n
Percent H2O degassed=100×(initial H2O−measured H2O)/initial H2O.
o
Percent CO2 degassed=100×(initial CO2−measured CO2)/initial CO2, where initial CO2=MvCO2+(p.p.m. CO2×10−4).
p
Percent total volatiles degassed=100×[(initial H2O – measured H2O)+(initial CO2−measured CO2)]/(initial H2O+initial CO2).
f
m
XCO
(PO, TO)=8·697×10−6−1·698×10−7SiO2.
are the molar volumes of the melt species in their standard states and have been taken to be independent of P, T, and melt composition (23 cm3/mol; Pan et al.,
1991). A 5× increase in CO2 solubility as SiO2 decreases from 49 to 40 wt % SiO2 is achieved by allowing the mole fraction of carbonate dissolved in the melt to
vary as a function of SiO2 (Dixon, 1997):
2−
3
,m
,m
− VoO, m and VoO, m and VoCO
DVo,r m= VoCO
A BA B
where variables are defined as in (b) with carbon dioxide replacing water and carbonate replacing molecular water.
e
2−
3
m
XCO
= {(wt % CO2/44)/[(100−wt % H2O−wt % CO2/36·6)+wt % H2O/18+wt % CO2/44]}.
whereX
(P, TO) is the mole fraction of molecular water in melt saturated with fluid with a fugacity of water of fH2O (P, TO) at pressure P and temperature TO
(1473·15 K); XHm2O, mol (PO, TO) is the mole fraction of molecular water in melt in equilibrium with vapor with a fugacity of water of fH2O (PO, TO) at pressure PO (1 bar)
and temperature T0; fH2O (PO, TO)=1 bar; Vo,H2mO, assumed constant over the range of compositions studied here, is the molar volume of water in the melt in its
standard state (12 cm3/mol; Dixon et al., 1995); and R is the gas content (83·15 cm3bar/mol K). A 30% decrease in the H2O solubility as SiO2 decreases from 49 to
40 wt % is achieved by allowing the mole fraction of molecular water dissolved in the melt at standard state to vary as a function of SiO2 (Dixon, 1997): XHm2O, mol
(PO, TO)=−3·0356×10−5 +1·2889×10−6 SiO2.
C
m
m
o, m
Assuming ideal mixing in the fluid, XHm2O,mol / Xo,
H2O ,mol represents the mole fraction of water in the fluid [in an H2O−CO2 fluid, XH2O,mol / XH2O , mol=1 for melt in
m
equilibrium with pure H2O fluid, XHm2O, mol / Xo,
H2 O, mol=0 for melt in equilibrium with pure CO2 fluid; Dixon & Stolper, 1995)].
d
Mole fraction of carbon dissolved as carbonate in the melt is calculated using:
m
H2O, mol
XHm2O, mol (P, TO)=XHm2O, mol(PO, TO)
a
Mole fraction of molecular water dissolved in the melt is calculated from measured total water contents using a regular solution model (Dixon et al., 1995) and
m
m
m
sum
equations to calculate mole fractions on a single oxygen basis: wt % H2Osum
tot =wt % OH+wt % H2Omol; XOH=2(XH−XH2O, mol); XB (total)={(wt % H2Otot /18)/
m
[(100−wt % H2O)/36·6+wt % H2Osum
tot /18+wt % CO2/44]}, XH2O, mol={ (wt % H2Omol/18)/[(100−wt % H2Otot)/36·6+wt % H2Otot/18+wt % CO2/44] }; where 36·6 is the molecular weight of anhydrous basalt on a single-oxygen basis.
b
Theoretical mole fraction of molecular water dissolved in the melt if the melt were saturated with a pure H2O fluid at 400 bar is calculated using the equation
for H2O solubility in tholeiite at 1200°C (Dixon et al., 1995):
DIXON et al.
VOLATILES IN NORTH ARCH ALKALIC BASALTS
JOURNAL OF PETROLOGY
VOLUME 38
Fig. 8. CO2 and H2O contents determined by FTIR spectroscopy in
North Arch glasses plotted against curves of vapor saturation at 400
bar pressure (isobars). Vapor compositions range from pure CO2 along
the y-axis to pure H2O along the x-axis. Position of the vapor saturation
surface varies as a function of silicate melt composition from nephelinitic
(SiO2=40 wt %) to alkali olivine basaltic (SiO2=48 wt %) and is
plotted at increments of 1 wt % SiO2. CO2 solubility increases by five
times and H2O solubility decreases by ~30% as SiO2 decreases from
49 to 40 wt % (Dixon, 1997). (a) Nephelinitic glasses, SiO2=40–42 wt
%. (b) Basanitic glasses, SiO2=43–45 wt %. (c) Alkali olivine basaltic
glasses, SiO2=46–48 wt %. The good agreement between the glass
data and the compositionally dependent vapor saturation curves is
consistent with these glasses being vapor saturated at the depth of
eruption.
NUMBER 7
JULY 1997
vesicles near the base of each flow. The outer selvage
(1–3 cm) is depleted in vesicles with respect to the rest
of the flow. P-type pahoehoe flows are most common on
the shallow (<4°) slopes of coastal terraces and are
interpreted to have resided in a lava tube for a day or
longer, to allow sufficient time for significant bubble loss
and coalescence, before emerging at the surface.
We interpret the nonvesicular glassy rims collected
from the North Arch lava flows to be analogous to the
vesicle depleted rims of P-type pahoehoe flows (Wilmoth
& Walker, 1993). Therefore, low vesicularities are not
representative of the original complement of bubbles.
Segregation of bubbles from the North Arch magmas
may have been facilitated by their low viscosities, resulting
from their high alkali and volatile contents. Using the
method of Shaw (1972), viscosities (in Pa s) at 1200°C
are calculated to be 1·2±0·3 for the nephelinitic melts,
2·6±0·6 for the basanitic liquids, and 7±3 for the
alkali olivine basaltic liquids, in contrast to 13±3 for a
Hawaiian tholeiite with 7 wt % MgO (Clague et al.,
1995). At these low viscosities, movement, and presumably escape of a vapor phase, seems likely during
flow away from the vent.
To estimate bulk volatile contents, we use only the
vesicular samples dredged from vent structures (23D,
24D-a, 24D-b, 27D-a, 27D-b, 26D-a). Bulk volatile contents are calculated assuming (a) ideal gas behavior for the
vesicle gases, (b) a pressure of 400 bar (the approximate
pressure for eruption beneath 4000 m of seawater), (c) a
‘rigid temperature’ of 1000°C (Moore et al., 1977), and
(d) a vapor composition for each sample as calculated
above.
The two most vesicular samples (24D-a and 26D-a)
have 56±1 vol. % vesicles and a vapor composition of
0·70±0·03 mol % CO2 (error is difference from the
average value) and are calculated to contain 1·0±0·1 wt
% H2O and 5·3±0·4 wt % CO2 in the vapor. The bulk
volatile contents of these samples (glass plus vesicles) are
1·9±0·1 wt % H2O and 5·4±0·4 wt % CO2 yielding a
bulk CO2/H2O (by mass) of 2·8±0·4. Thus, the vapor
represents 47% of the bulk H2O and 98% of the bulk
CO2.
The four other vesicular samples (23D, 24D-b, 27Da, and 27D-b) have 31±4 vol. % vesicles and a vapor
composition of 0·70±0·06 mol % CO2. These samples
are calculated to contain 0·4±0·1 wt % H2O and
1·9±0·4 wt % CO2 in the vapor. The bulk volatile
contents of these samples are 1·3±0·2 wt % H2O and
2·0±0·4 wt % CO2, yielding a lower bulk CO2/H2O
(by mass) of 1·5±0·3. In these samples, the vapor represents 31% of the bulk H2O and 95% of the bulk CO2.
The proportion of gas in a melt could possibly be
modified by either accumulation or loss of bubbles. If
the highly vesicular glasses are related to the moderately
vesicular ones by accumulation of bubbles instead of
926
DIXON et al.
VOLATILES IN NORTH ARCH ALKALIC BASALTS
closed system degassing, we would expect a bulk H2O/
P2O5 higher than that found in the alkali olivine basalts,
which are relatively undegassed with respect to H2O (this
also assumes that all the undegassed parental magmas
had initial H2O/P2O5 ratios similar to the alkali olivine
basalts). The two highly vesicular glasses have a bulk
H2O/P2O5=3·3±0·5, slightly higher than, but within
10% of, the value of three observed in the alkali olivine
basalts. If bubble accumulation has taken place, it probably accounts for <20% of the gas. Alternately, if the
moderately vesicular glasses lost gas with respect to the
highly vesicular group, we would expect their bulk H2O/
P2O5 to be <3. Also, loss of CO2-rich vapor would result
in a decrease in the CO2/H2O ratio of the bulk system,
as well as the bulk abundance of gas. In fact, this is what
we observe. The moderately vesicular samples have a
bulk H2O/P2O5 of 2·1 (30% lower than a value of three)
and lower bulk CO2/H2O than the highly vesicular
samples. Thus, we infer that the highly vesicular samples
contain the least modified information about initial volatile contents.
These estimated bulk volatile concentrations are large
(total volatile contents of 3·3–7·3 wt %) when compared
with most MORB glasses, which have only a few tenths
of a percent of H2O and CO2, or even compared with
the ‘popping rocks’, which are estimated to have 1·4 wt
% volatiles (Graham & Sarda, 1991). These high estimates
for volatiles in the North Arch glasses are consistent
with the high concentrations of nonvolatile incompatible
elements related to generation of these magmas by low
extents of melting.
Estimates of bulk CO2/H2O of ~1–3 may seem high,
but in fact they are consistent with the few existing
estimates for undegassed volatile contents in oceanic
basalts. For example, primary depleted MORB is thought
to have ~0·12 wt % CO2, with estimates ranging from
~0·07 to 0·26 wt % ( Jambon et al., 1985; Gerlach, 1989;
Jambon, 1994) and ~0·1 wt % H2O (Michael, 1988;
Dixon et al., 1988; Jambon, 1994) yielding a mean CO2/
H2O by mass of ~1. Estimates of bulk CO2 and H2O in
popping rocks from the Mid-Atlantic Ridge, perhaps
representative of undegassed MORB magma, are 0·8 wt
% CO2 and ~0·6 wt % H2O (Sarda & Graham, 1990;
Gerlach, 1991; Graham & Sarda, 1991) yielding a bulk
CO2/H2O of 1·3. Estimates for the bulk volatile contents
in Kilauea tholeiitic basalts range from 0·32 (Greenland
et al., 1985) to 0·65 wt % CO2 (Gerlach & Graeber,
1985) and 0·35 wt % H2O (Clague et al., 1991) yielding
CO2/H2O of 1–2. At this time we cite the similarities to
show that our estimates of bulk volatile contents are
reasonable. More work is required to understand what,
if any, fractionations occur between H2O and CO2 during
generation of basaltic magmas.
MODELING OF TRENDS IN H 2 O AND
CO 2 DATA
One goal of this study is to model the behavior of volatiles
during processes such as partial melting, fractional crystallization and degassing, to estimate mantle volatile
contents. An intriguing aspect of the North Arch data is
the poor correlation of H2O concentrations with nonvolatile incompatible elements. The decoupling of H2O
from other trace elements could have occurred by modification of the source region (metasomatic hypothesis) or
during eruption on the seafloor (degassing hypothesis).
In the metasomatic hypothesis, the source for the alkali
olivine basalts would have experienced addition of a
hydrous fluid with respect to the source for the nephelinites. At the same temperature, a hydrous mantle
will melt more than a relatively anhydrous mantle (Green
& Ringwood, 1967; Kushiro et al., 1968; Kushiro, 1970;
Green, 1972, 1973; Green & Wallace, 1988; Hirose &
Kawamoto, 1995), resulting in a positive correlation
between extent of melting and mantle water concentration. If the effects of addition of water to the source
and increase in the extent of melting essentially cancel,
water concentrations would remain low (<1 wt %) for
all parental magmas regardless of the extent of melting.
A CO2-rich vapor would exsolve during ascent and
eruption, but significant exsolution of water would not
be required to explain the relatively constant dissolved
H2O concentrations. There is increasing evidence that
metasomatic processes may affect the lithosphere under
ocean islands (Amundsen, 1987; Frey & Roden, 1987;
Sen, 1988). A similar metasomatic model successfully
explained trends in the H2O data in glasses from the
Mariana back-arc basin (Stolper & Newman, 1994) and
served as a model during initial evaluation of the trends
in the volatile data for the North Arch glasses (Clague
& Dixon, 1991, 1993).
Alternately, the mantle source region for the North
Arch lavas could be homogeneous with respect to volatiles. Initial volatile contents would increase proportionally with other incompatible trace elements (e.g.
P2O5) as the extent of melting decreases. Hence, low
degree partial melts (nephelinites) would have high initial
volatile concentrations. Consequently, both H2O and
CO2 would degas during eruption (lowering the H2O/
P2O5). In contrast, relatively large degree partial melts
(alkali olivine basalts) would have lower initial volatile
concentrations and would exsolve dominantly CO2 (because of the extremely low CO2 solubility) with only
minor amounts of H2O (little or no change to H2O/
P2O5). The extremely low noble gas concentrations and
the high vesicularity of the nephelinites suggest that
vapor exsolution has been significant. However, previous
models of degassing of tholeiitic basalts show that essentially all the CO2 in a magma will exsolve before
927
JOURNAL OF PETROLOGY
VOLUME 38
significant exsolution of H2O occurs (Khitarov & Kadik,
1973; Shilobreyeva et al., 1983; Gerlach, 1986; Newman,
1989, 1990; Bottinga & Javoy, 1990; Dixon & Stolper,
1995). Thus, at first glance, the relatively high CO2
contents in the nephelinites seem to argue against significant exsolution of H2O. Any model that explains the
relatively low H2O contents in the nephelinitic glasses
must also explain the relatively high CO2 contents. These
hypotheses will be tested below.
Forward degassing modeling
Forward degassing models predict the dissolved H2O
and CO2 concentrations and the volume and vapor
composition of the exsolved gas as a function of initial
volatile element contents, melt composition, style of degassing (open or closed), and final pressure of equilibration
(Dixon, 1997). Temperature is assumed to be 1200°C.
During discussion of the degassing models, the phrase
‘initial volatile contents’ refers to the undegassed values
in the observed magma compositions. Later in the discussion of parental magmas, we will show that most of
the erupted lavas have undergone olivine fractionation,
consequently the initial volatile contents in the parental
magmas used to infer mantle volatile concentrations will
be lower than those discussed here.
For each calculation, the final pressure (400 bar) and
style of degassing are specified. The SiO2 content is set to
a value equal to or between 49 and 40 wt % and the
concentrations of incompatible elements are calculated
as a function of SiO2. We choose P2O5 as our reference
incompatible element for both models. The initial P2O5
content as a function of SiO2 is calculated based on a linear
fit to the P2O5 vs SiO2 data (P2O5=4·33 – 0·088SiO2; Fig.
3a). Calculation of initial volatile element contents as a
function of P2O5 is discussed below within the discussion
of the metasomatic and degassing hypotheses. After calculation of the initial volatile contents, H2O and CO2 solubilities and the value of the melt–vapor fractionation
factor b (an expression for the equilibrium partitioning
of water and carbon dioxide into the vapor phase) are
calculated as a function of SiO2 and pressure. Degassing
of magma is accomplished by incrementally transferring
H2O and CO2 from the supersaturated melt into the vapor
phase until saturation is reached. Degassing was assumed
to occur at 400 bar pressure, but the results are independent of the P at which degassing begins. Each increment of volatiles partitioned into the vapor must satisfy
equations for mass balance, vapor–melt fractionation, and
speciation of water in melt [see equations and detailed
explanation given by Dixon & Stolper (1995) and Dixon
(1997)]. Calculations are repeated using different SiO2
contents stepping down from 49 to 40 wt %. Results of
each calculation for a given SiO2 content are then
NUMBER 7
JULY 1997
compared with the observed data. The program and
sample calculations are available from the first author.
Modeling results: metasomatic hypothesis
In the metasomatic hypothesis, the observed range in H2O
contents and H2O/P2O5 is assumed to originate in the
mantle source. This hypothesis was tested by allowing initial H2O content to vary, such that the initial H2O/P2O5
ratio of the melts was set to increase from one to three as
the SiO2 content of the melt increased from 40 to 49 wt %
consistent with the observed variations in the data. Initial
CO2 contents were modeled in several ways as discussed
below.
In the first set of calculations, the initial CO2 contents
were assumed to increase along with other nonvolatile
incompatible elements (i.e. CO2 not contained in the metasomatic fluid) and were calculated using constant initial
CO2/P2O5 ratios that varied from 0·5 to six. Exsolution of
H2O and CO2 from magmas having the first set of initial
conditions can produce the observed dissolved H2O and
CO2 concentrations only when the initial CO2 contents
are low (CO2/P2O5=0·5). Calculated vesicularities in this
case (<20 vol %) are lower than the observed vesicularities
in the vent samples.
In a second set of calculations, the metasomatic fluid
was assumed to contain both H2O and CO2. Under this
condition, metasomatism affects the concentrations of
both H2O and CO2, but not the CO2/H2O. This was
accomplished by holding the initial CO2/H2O ratio constant, while allowing the initial H2O/P2O5 to decrease with
SiO2 (e.g. larger extent melts have higher H2O and CO2
in source). As in the previous case, we could match the
concentrations of dissolved H2O and CO2 only when the
initial CO2 concentrations are low (CO2/H2O=0·5), resulting in calculated vesicularities (<20 vol %) that are,
again, lower that the observed values in the vent samples.
Also, the required initial CO2/H2O is lower than the bulk
CO2/H2O estimated for the vesicular samples. These discrepancies do not support a metasomatic hypothesis to
account for the trends in volatiles in the North Arch lavas.
We conclude that metasomatic addition of water
(Clague & Dixon, 1991, 1993) cannot explain the relatively
constant water contents and high vesicularities in the North
Arch lavas. If the mantle source region has experienced
metasomatism at some time in its history, then the metasomatism affected volumes of mantle large enough to be
homogeneous on the scale sampled by alkalic volcanism.
928
Degassing hypothesis
In the degassing hypothesis, the cause of the low H2O
contents in the basanites and nephelinites is assumed to
be exsolution of water into the vapor. We assume that
initial volatile contents in the erupted lavas correlate
negatively with SiO2 as do the concentrations of
DIXON et al.
VOLATILES IN NORTH ARCH ALKALIC BASALTS
nonvolatile incompatible elements. We assume that H2O
behaves similarly to P2O5 during melting and crystallization (Michael, 1988, 1995; Dixon et al., 1988) and
that the H2O/P2O5 of the mantle source region is three
(the upper limit of the observed ratios) and that this value
is not modified during melting and crystallization, thus
the initial H2O content is three times the P2O5 content.
For initial CO2 contents, we multiply the calculated initial
H2O content by a fixed CO2/H2O (by mass). We assume
a constant initial CO2/H2O for each calculation and
repeat the calculations using ratios that vary from one
to four to cover the range of values predicted by the bulk
volatile calculations.
Closed and open system degassing models calculated
using CO2/H2O ratios of 1–4 are shown in Fig. 9a–e.
H2O concentrations in the initial (undegassed) melts
increase as the SiO2 decreases and P2O5 increases, such
that alkali olivine basalts have ~0·9 wt % H2O, basanites
have ~1·4 wt % H2O, and nephelinites have ~2·7 wt %
H2O. In contrast, over this same compositional range
the water solubility at 400 bar decreases slightly from
2·0 to 1·6 wt % as SiO2 varies from 49 to 40 wt %
(Dixon, 1997). Thus the initial water concentrations in
nephelinitic magmas exceed the H2O solubility and H2O
will exsolve even in the absence of CO2. But to exsolve
enough water to diminish the dissolved H2O concentrations down to the ~1 wt % value observed in most
of the glasses, the melts have to be in equilibrium with
a vapor having P H2O<400 bar.
Most water contents measured in the North Arch
glasses (Fig. 9a) are bounded by the open system curve
and the closed system degassing curve using an initial
CO2/H2O ratio of four. The vesicular glasses (23D, 24Da, 24D-b, 27D-a, 27D-b, 26D-a) typically have lower
water contents at the same SiO2 than the nonvesicular
glasses and are bounded by the closed system degassing
curves using initial CO2/H2O ratios of one and four.
High initial CO2 contents (CO2/H2O>1) are required to
yield a vapor in which P CO2>P H2O, such that the vaporsaturated melts will have H2O contents of ~1·0 wt %.
The amount of water exsolved into the vapor is significant.
For example, after closed system degassing using an
intermediate ratio of initial CO2 to H2O of two, the alkali
olivine basaltic melts have exsolved ~15%, the basanitic
melts ~30%, and the nephelinitic melts ~60% of the
initial water present. The proportion of water exsolved
from the nephelinites is greater than that in the alkali
olivine basalts because of the higher initial water concentrations, lower H2O solubility, and smaller CO2–H2O
melt–vapor fractionation factors, which results in slightly
diminished preferential partitioning of CO2 into the vapor
relative to H2O (Dixon, 1997).
The nonvesicular samples (21D-b, 36D-a, 9D, 36D-b,
17D-a, 22D, RB02-a) are bounded below by the closed
system degassing curve using CO2/H2O of two and above
by the open system degassing curve. Either the melts
that quenched to form the nonvesicular glasses initially
had less CO2 than those that quenched as vesicular
glasses, or degassing of these melts occurred as a partially
open system. During open system degassing, fractionation
between CO2 and H2O is accentuated because CO2 is
preferentially partitioned into the vapor and removed,
thereby diminishing H2O exsolution until concentrations
close to the H2O solubility are reached. Thus, magmas
that have degassed as an open system will appear to have
lower initial CO2/H2O contents than those that have
degassed as a closed system. We interpret the low vesicularity as physical evidence for gas loss during flow on
the seafloor consistent with open (or partially open) system
degassing.
The larger proportion of initial H2O that exsolves from
SiO2-poor magmas at 400 bar results in melts that have
low H2O/P2O5 ratios (Fig. 9b). As is the case for dissolved
water concentrations, the H2O/P2O5 data for the vesicular samples (23D, 24D-a, 24D-b, 27D-a, 27D-b, 26Da) are bounded by closed system degassing curves with
CO2/H2O of 1–4, whereas the nonvesicular samples
(21D-b, 36D-a, 9D, 36D-b, 17D-a, 22D, RB02-a) have
higher H2O/P2O5 at the same SiO2 and are bounded by
the curves for open system to closed system with an
initial CO2/H2O of ~2.
The higher carbon dioxide contents measured in the
vesicular North Arch glasses are generally consistent with
the closed system degassing model using initial CO2/
H2O ratios of 1–4 (Fig. 9c), though the CO2 data have
more scatter, with sample 23D lying above the CO2/
H2O=4 curve. The nonvesicular samples have lower
dissolved CO2 contents bounded above by the closed
system curve using CO2/H2O of two and below by the
open system curve.
Using estimates of CO2/H2O of three, initial carbon
dioxide contents for the undegassed magmas are 0·41–2·8
wt % for the alkali olivine basalts, 2·8–4·7 wt % for the
basanites, and 4·7–7·5 wt % for the nephelinites. These
values are strikingly high (and are off scale in Fig. 9c)
compared with CO2 solubility at 400 bar ranging from
~200 p.p.m. for tholeiite to ~950 p.p.m. for nephelinite,
but are consistent with the bulk CO2 values calculated
for the most vesicular glasses. Because of the low solubility
of CO2 in these melts, almost all (>95%) of the carbon
dioxide is exsolved during decompression associated with
ascent and eruption.
Predicted vapor compositions remain CO2 rich over
the range of compositions modeled. In the closed system
degassing model using an initial CO2/H2O ratio of three,
X vapour
CO2 ranges from 0·99 in the alkali olivine basaltic melt
(SiO2=48 wt %) to 0·66 in the nephelinitic melt (SiO2=
40 wt %) (Fig. 9d). The predicted vapor compositions are
consistent with the vapor phase compositions calculated
based on measured H2O and CO2 contents in basaltic
929
JOURNAL OF PETROLOGY
VOLUME 38
NUMBER 7
JULY 1997
Fig. 9. Comparison of data with results of forward degassing models (see text for details of modeling). Curves connect saturation values calculated
at different SiO2 contents and do not represent evolution curves. Dashed curves represent results of forward degassing models for open and closed
system degassing of melts having a range of SiO2 contents, initial concentrations of P2O5 constrained by the relationship in Fig. 3a (P2O5=4·33 –
0·088 SiO2), and initial volatile concentrations constrained by H2O/P2O5=3 and CO2/H2O=1 (dotted), CO2/H2O=2 (short dashes), CO2/H2O=
3 (medium dashes), and CO2/H2O=4 (long dashes). Open system curve (dotted; closer spacing) calculated in the same way with only the CO2/
H2O=2 curve shown. Open system degassing is relatively insensitive to initial volatile contents. Continuous lines are initial undegassed values. Filled
symbols have vesicularities Ζ2 vol %. Open symbols have vesicularities [27 vol %. Circles are alkali olivine basalts. Squares are basanites. Diamonds
are nephelinites. Sample labeled V is a vent sample (29D) for which vesicularity was not measured. Sample labeled F is a flow sample (20G) for which
vesicularity was not measured. (a) H2O vs SiO2. The continuous line shows the negative correlation between SiO2 and H2O expected if H2O maintains
a constant proportionality with P2O5 during melt generation and olivine fractionation. The open system curve shows that little H2O is lost during
open system degassing until the H2O content of the melt has a value close to the H2O solubility, which occurs at the change in slope at ~43 wt %
SiO2. The flow samples roughly follow or are slightly below the open system trend. The vent samples require closed system degassing and an initial
CO2/H2O[1 to be saturated with ~1 wt % H2O at 400 bar. (b) H2O/P2O5 vs SiO2. As in (a), the decreasing H2O/P2O5 with decreasing SiO2 of the
vent samples is consistent with closed system degassing of melts having initial CO2/H2O[1, whereas the higher H2O/P2O5 of the flow samples is
consistent with open to mostly open system degassing. (c) CO2 vs SiO2. Though the CO2 analyses are less precise than those for H2O, the vent samples
typically have higher dissolved CO2 concentrations than the flow samples, consistent with the closed vs open trends described in (a) and (b). (d) Vapor
composition vs SiO2. The flow samples roughly follow or are slightly above the open system trend. The vent samples require closed system degassing
and an initial CO2/H2O[1 to be saturated with a vapor having ~70 mol % CO2 at 400 bar. (e) Volume percent vesicles vs SiO2. Highly vesicular
samples are bounded above by the closed system degassing curve using an initial value of CO2/H2O=3. The low abundance of vesicles (Ζ2 vol %)
in the flow samples is the most intuitively obvious consequence of open system degassing.
930
DIXON et al.
VOLATILES IN NORTH ARCH ALKALIC BASALTS
glasses. The vesicular glasses have predicted vapor compositions consistent with the closed system degassing
models with initial CO2/H2O of 1–4, whereas the nonvesicular samples have lower X vapour
bounded above by
CO2
the closed system curve using CO2/H2O=2 and bounded
below by the open system curve.
Predicted vesicularities are high, with the model curve
having initial CO2/H2O=3 forming an upper bound for
the highest values observed in the samples (Fig. 9e).
Calculated vesicularities using initial CO2/H2O=3 are
about 8–41% for alkali olivine basaltic magmas, 41–58%
for basanitic magmas, and 58–70% for nephelinitic
magmas. The highly vesicular samples lie along the closed
system curves, consistent with the idea that exsolved
vapor remained within the magma during ascent and
depressurization.
The variations of dissolved volatile contents, vapor
compositions, and vesicularities as a function of SiO2
content are best explained by extensive degassing at 400
bar pressure of magmas whose initial volatile contents
are proportional to concentrations of nonvolatile incompatible elements (e.g. P2O5). The exsolving vapor
escapes to varying degrees from the magma during
eruption and flow. The highly vesicular samples (24D-1
and 26D-a) have lost the least gas (closed system) as
manifested by their high gas content, high X vapour
CO2 , low
dissolved H2O, low H2O/P2O5, and high dissolved CO2
contents. That the system is closed with respect to exsolving volatiles suggests that these magmas did not reside
at shallow levels in the crust long enough for bubbles to
separate from the melt. The high bulk volatile contents
predicted for closed system degassing of the nephelinites
may provide the necessary force for producing submarine
fountaining, even at 4 km water depth, consistent with
the presence of glass spheres in hydroclastites in the
North Arch volcanic field. The other lavas lost some or
almost all exsolved gas during eruption and flow away
from the vent as manifested by their lower vesicularities,
higher dissolved H2O, higher H2O/P2O5, and low dissolved CO2 contents.
Though the models are complex, they result in a simple
conclusion; namely, that the mantle source for the North
Arch lavas is homogeneous with respect to ratios of
volatile to nonvolatile incompatible elements. Reduction
in the extent of melting produces enrichments in all
incompatible elements, including H2O and CO2. We do
not need to invoke ad hoc source heterogeneities to explain
the observed variations in volatiles.
shown in Fig. 7d. Also shown for reference are calculated
solubility curves for nephelinitic to alkali basaltic melts
that are saturated with immiscible Fe–S–O liquid. Because S solubility in melts is strongly temperature dependent, it is important to know the eruption
temperatures of the North Arch magmas to assess whether
they were saturated with Fe–S–O liquid. Eruption temperatures were calculated for the glasses at an assumed
pressure of 1 kbar using the MELTS program (Ghiorso
& Sack, 1995). Liquidus temperatures for melts with the
composition and dissolved H2O contents of the North
Arch glasses vary from 1175–1225°C for the alkali basalts
to 1145–1190°C for the nephelinites. The calculations
suggest that for melts with comparable MgO contents,
the alkali basalts have liquidus temperatures ~15°C
higher than the nephelinites. Based on these results,
sulfide liquid solubility curves are shown in Fig. 7d
at 1200°C and 1170°C, which represent the average
temperatures of the alkali basalts and nephelinites, respectively, that were analyzed for Fe3+/RFe. The alkali
olivine basalt, basanite, and two nephelinite glasses have
S contents that are within error of the saturation line
(Fig. 7d), suggesting that their S contents are controlled
by Fe–S–O liquid saturation. Two other nephelinitic
glasses (24D-a, 27D-a) have S contents slightly less than
that required for Fe–S–O saturated melts. The lower S
contents of these samples is probably the result of exsolution of S into the vapor during eruption on the
seafloor. It is not clear why only these samples should
have been affected by S loss, but we note that both
are highly vesicular vent samples. Using the observed
correlation between glass P2O5 contents and relative
oxygen fugacity (Fig. 7c) we have estimated DFMQ for
glasses that have not been analyzed for Fe3+/RFe. The
results, when combined with solubility calculations, suggest that the other samples were also saturated with
immiscible Fe–S–O liquid at the time of eruption, and
that they did not lose appreciable dissolved S to the
coexisting vapor phase during eruption and quenching.
Sulfide liquid saturation and degassing of
sulfur
The dissolved S contents of the North Arch glasses as a
function of their relative magmatic oxygen fugacities are
931
CALCULATION OF PARENTAL
MAGMA COMPOSITIONS AND
MANTLE VOLATILE
CONCENTRATIONS
To remove the effects of shallow-level fractionation, we
calculated the compositions of parental magmas from
actual glass compositions by addition of 0·1% increments
of equilibrium olivine until the liquidus olivine is Fo91 (see
Stolper & Newman, 1994). Parental magma compositions
were calculated for samples having MgO contents >7·0
wt % (Table 1). Two samples (23D and 36D-a) had MgO
contents <7·0 wt %, contained common clinopyroxene
JOURNAL OF PETROLOGY
VOLUME 38
phenocrysts, and were excluded from the parental magma
calculation. Calculations were performed using an initial
log f O2 equivalent to the average measured log f O2 of FMQ.
Amounts of olivine added to each glass composition
ranged from 22 to 34% with an average of 28±4%,
resulting in melts with MgO contents of 15·1–17·2 wt
%. The amounts of added olivine calculated here are
higher than values of 5–15% reported by Clague et al.
(1990) because they started with olivine-bearing, wholerock compositions. We do not think it is inconsistent that
the vesicular magmas are closed with respect to volatiles,
and yet appear to have lost some olivine, because olivine
probably crystallizes throughout the ascent path, whereas
exsolution of volatiles begins at a shallower depth. Calculated P2O5 contents in parental magmas range from
0·17 wt % in an alkali olivine basalt to 0·65 wt % in a
nephelinite (Table 1; Fig. 3a). Based on the argument
that concentrations of volatiles are proportional to P2O5,
parental magmas are calculated to contain 0·51–1·95 wt
% H2O (3×P2O5), and 1·3±0·8 to 4·9±2·9 wt % CO2
(2·5±1·5×H2O).
Estimates of extent of melting are indexed to concentrations of P2O5. Both P2O5 and K2O display highly
incompatible behavior in the North Arch glasses; we
selected P2O5 as an indicator of the extent of melting
(F m) for consistency with previous work on Hawaiian
alkalic lavas. Concentrations of P2O5 in the parental
magma compositions vary by a factor of ~3·7, ranging
from 0·18 wt % in the alkali olivine basalts to 0·66 wt
% in the nephelinites. It is not possible to simultaneously
define the extent of melting, the bulk distribution coefficient for P2O5 (D P0 2O5), and the initial concentration of
P2O5 in the source (c P0 2O5), but reasonable bounds can be
placed on these parameters based on previous studies of
Hawaiian lavas (Clague & Frey, 1982; Chen & Frey,
1983, 1985; Clague & Dalrymple, 1988; Chen et al.,
1991). Phosphorus has an incompatibility similar to that
of the light rare earth element Nd (e.g. Sun et al., 1979),
with estimates of D P0 2O5 in mantle assemblages dependent
on the amount of garnet in the source region. Using
partition coefficients from Ulmer (1989), D P0 2O5 is calculated to be 0·011 in a depleted mantle containing no
garnet (65% ol, 35% opx, 10% cpx) to 0·019 in a mantle
containing 7% garnet (53% ol, 25% opx, 15% cpx, 7%
gt). Given the similar incompatibility of phosphorus and
potassium in the North Arch lavas, we chose the lower
(and probably minimum) estimate and assert that there
is little residual garnet consistent with a relatively depleted
source region (Clague et al., 1990). Use of the lower value
for D P0 2O5 results in a conservative estimate of c P0 2O5 and
ultimately of c P0 2O5. Previous estimates of extents of melting
required to produce Hawaiian alkali olivine basalts range
from about 6·5–8·0% (Chen et al., 1991) to 9–17%
(Clague & Frey, 1982; Clague & Dalrymple, 1988; Clague
et al., 1990). Assuming D P0 2O5=0·011 and that the least
NUMBER 7
JULY 1997
SiO2-undersaturated alkali olivine basalts were derived
by ~9±2% batch melting, then we calculate that c P0 2O5
or D P0 2O5 is 175±25 p.p.m. and the most SiO2-undersaturated nephelinites were generated by ~1·6±0·3%
batch melting. Changes in c P0 2O5 or D P0 2O5 will shift the
entire data set, but will not affect the relative differences
between samples. For example, using a higher value of
0·019 for D P0 2O5 results in an increase in the value of c
P2O5
or D P0 2O5 to 198 p.p.m., which is the upper limit of
0
our estimated error.
Mantle volatile contents can be calculated based on
this estimate of 175±25 p.p.m. P2O5 in the mantle source.
The concentrations of mantle volatiles are calculated to
be 525±75 p.p.m. for H2O (3×P2O5), 1300±800 p.p.m.
(1 to 4×H2O=2·5±1·5×H2O) for CO2, and 30±6
p.p.m. (0·17×P2O5) for Cl. The implications of our
estimated mantle volatile contents on the origin of various
mantle end members required to model the temporal
variation of all Hawaiian lavas will be investigated within
the context of a more complete set of trace and isotopic
data (Frey et al., unpublished data, 1997). We emphasize,
however, that this is the first time that mantle volatile
contents have been estimated from measured concentrations in alkalic glasses of oceanic island affinity.
An estimate of 525 p.p.m. H2O in the mantle source
for the North Arch lavas is higher than estimates of
100–180 p.p.m. for depleted MORB and slightly higher
than the range estimates of 250–450 p.p.m. for enriched
MORB (Michael, 1988; Dixon et al., 1988). The estimated
parent magma for high-MgO, tholeiitic glasses from
Kilauea has 0·35 wt % H2O (Clague et al., 1991). Assuming H2O has a distribution coefficient of ~0·01 and
that the parent magma was generated by 15±5% melting, then water content in the mantle source region of
Kilauea tholeiites is calculated to be 555±170 p.p.m.,
similar to our estimate for the mantle source of the North
Arch lavas. For Cl, primary depleted MORB has ~20–50
p.p.m. Cl and enriched MORB has ~150–200 p.p.m.
Cl (Schilling et al., 1980; Michael & Schilling, 1989). If
these magmas were produced by ~15% melting, then
the source region for depleted MORB would contain
2–8 p.p.m. Cl and that for enriched MORB would
contain 23–30 p.p.m. Cl. Therefore our estimate of 30
p.p.m. Cl in the mantle source region for the North Arch
lavas is similar to or slightly higher than estimates for
the source of enriched MORB.
Our preferred interpretation of the constant initial
CO2/H2O required to produce the observed trends in
the volatile data is that both CO2 and H2O were undersaturated during magma genesis and behaved incompatibly during melting. This interpretation is
consistent with studies of the behavior of water (Michael,
1988, 1995; Dixon et al., 1988) and carbon (Kadik, 1995)
during melting, but it is not unique. One can imagine
another scenario in which the melts are saturated with
932
DIXON et al.
VOLATILES IN NORTH ARCH ALKALIC BASALTS
respect to carbon-bearing fluid during melt generation
and the increase in CO2 solubility by a factor of five as
SiO2 decreases coincidentally matches the increase in
incompatible elements by a factor of five as the extent
of melting decreases. If this were the case, then the
pressure at which the North Arch magmas are saturated
with the calculated initial volatile concentrations would
provide information on the depth of melt generation.
The pressure at which the parental melts would be
vapor saturated can be calculated using the estimated
H2O and CO2 contents in the parental magmas (Table
1) and the model of Dixon (1997). Parental alkali olivine
basalt with 0·5 wt % H2O and 1·3 wt % CO2 would
reach vapor saturation at 8·3 kbar (~30 km depth).
Parental nephelinite with 2·0 wt % H2O and 4·9 wt %
CO2 would be vapor saturated at 12·8 kbar (~40 km
depth). These estimates are about a factor of 2–3 lower
than current estimates of the pressure of generation of
alkalic basalts (e.g. Adam, 1988) and thus vapor saturation
during melt saturation is probably not the dominant
control of the initial CO2 content.
Dissolved S contents of parental magmas
Based on the saturation calculations discussed previously,
the dissolved S contents in the North Arch glasses appear
to have been controlled by saturation with immiscible
Fe–S–O liquid. Thus S would not behave as a highly
incompatible element during crystallization. However,
there is little constraint on whether the parental melts
were Fe–S–O liquid saturated, and it is therefore possible
that crystal fractionation of olivine from a sulfide liquidundersaturated parental melt caused an increase in dissolved S contents until saturation occurred. If the parental
melts were sulfide saturated during melt generation, their
estimated dissolved S contents would have ranged from
0·21 wt % for the most reduced magmas to 0·28 wt %
in the most oxidized. These S contents are estimated
from solubility calculations for sulfide-saturated melts at
~1325°C, the estimated average liquidus temperature
for the parental magmas. The calculated extents of
melting based on estimated parental melt P2O5 contents
vary from 1·6 to 9·0% (±20% relative), suggesting that
the upper-mantle source lithology would have [190
p.p.m. S for residual sulfide to be present throughout the
melting interval. This value is similar to an estimate of
200 p.p.m. for primitive upper mantle based on analyses
of mantle xenoliths (O’Neill, 1991). However, Clague et
al. (1990) have argued that the mantle source region for
the North Arch magmas is more depleted than that from
which magmas of the Honolulu series were derived, and
that minor phases such as phlogopite, amphibole, Tioxides, and apatite were not present in the residuum
during partial melting. Based on this evidence, it seems
plausible that the S content in the North Arch source
region was lower than that estimated for primitive upper
mantle (<200 p.p.m.), and that sulfide was therefore not
a residual phase during melting. In this case, the dissolved
S contents of the parental melts would have been lower
than the values calculated above assuming sulfide saturation during melt generation.
OXYGEN FUGACITY AND PARTIAL
MELTING
As discussed previously, negative correlations between
SiO2 and incompatible trace element concentrations (e.g.
P2O5) in the North Arch lavas are consistent with the
hypothesis that the magmas were generated by variable
extents of melting of a common homogeneous source
(Fig. 3). The good correlation of glass P2O5 contents with
Fe3+/RFe and relative oxygen fugacity further suggests
that magmatic oxygen fugacity may be related to the
degree of partial melting. To account for the effects
of olivine fractionation on glass composition we have
calculated Fe3+/RFe for the parental magmas in Table
1 by assuming that Fe3+ behaves as an incompatible
element during crystallization. As a result, the calculated
parental compositions all have lower Fe3+/RFe than the
analyzed glasses. Calculated relative oxygen fugacities of
the parental magmas are ~0·5 log units more reduced
than that of the corresponding analyzed glass. Owing to
the pressure dependence of ferric–ferrous equilibrium in
silicate melts, a crystal-poor melt ascending as a closed
system (i.e. fixed oxygen content) will undergo a decrease
in equilibrium oxygen fugacity with decreasing pressure
(Kress & Carmichael, 1991). This decrease roughly parallels that of the FMQ buffer so that DFMQ is relatively
independent of pressure (see Kress & Carmichael, 1991).
Calculated relative oxygen fugacities of the parental
magmas at an estimated pressure of melt generation of
30 kbar range from FMQ – 0·8 to FMQ+0·7. This range
of values is consistent with estimates for the equilibrium
oxygen fugacity of upper-mantle peridotites (FMQ – 4
to +1) determined by analysis of mantle xenoliths (e.g.
O’Neill & Wall, 1987; Wood & Virgo, 1989; Bryndzia
et al., 1989; Luth et al., 1990).
The positive correlation between parental magma P2O5
contents and estimated oxygen fugacities is consistent
with the interpretation that the lowest degree partial
melts (nephelinites) had the highest relative magmatic
oxygen fugacities and that relative magmatic oxygen
fugacity decreased with increasing extent of melting.
Several hypotheses that might account for this correlation
are discussed below. Because mantle phases containing
Fe and Mg are all solid solutions, and a fluid phase, if
present, will be a multicomponent C–O–H–S phase, the
equilibria that control oxygen fugacity during mantle
933
JOURNAL OF PETROLOGY
VOLUME 38
melting are all multivariant such that f O2 is not tightly
buffered (Ballhaus & Frost, 1994; Canil et al., 1994). In
the absence of a separate C–O–H–S fluid phase, the
oxygen fugacity of partial melts will be controlled by the
oxygen content of the system, as reflected in the bulk
Fe3+/RFe, and by the partitioning of Fe3+ between melt
and the relevant solid phases. Alternatively, if a fluid
phase is present during melting, then f O2 may be controlled by equilibria involving graphite and C–O–H–S
species.
Based on the analysis of Fe3+/RFe in minerals from
mantle xenoliths, O’Neill et al. (1993) suggested that
primitive upper mantle (both garnet and spinel lherzolite)
has whole-rock Fe3+/RFe in the range 0·015–0·04. Furthermore, these workers found that Fe3+ resides dominantly in clinopyroxene, orthopyroxene, spinel, and
garnet. During partial melting, whole-rock Fe3+/RFe in
residual peridotite decreases with extent of melting because of the large decrease in modal abundances of the
pyroxenes (Dick et al., 1984; O’Neill et al., 1993). This
causes Fe3+ to behave like an incompatible element,
in contrast to Fe2+, which behaves compatibly during
melting. Importantly, this behavior of Fe3+ should hold
regardless of whether f O2 is externally controlled by vapor
saturation, because it is the preferential melting of the
pyroxenes that determines Fe3+/RFe in both the melt
and the solid residuum (O’Neill et al., 1993). If we take
sol/liq
sol/liq
=1 and D Fe
~0·2 (O’Neill et al., 1993) and use
D FeO
2O3
the extent of melting values calculated for the glasses in
Table 1, then the calculated parental melt compositions
could have been generated from a mantle source lithology
with Fe3+/RFe=0·029±0·006. This value is in excellent
agreement with the estimates for primitive upper mantle
discussed above, demonstrating that the North Arch
magmas could have been generated in a mantle source
region that was homogeneous with respect to Fe3+/RFe.
An alternative possibility for the relationship between
degree of partial melting and relative f O2 is that melting
occurred in equilibrium with graphite and a C–O fluid
phase (Wood et al., 1990). In such a system, increasing
degrees of melting result in decreasing relative oxygen
fugacities because oxidation of graphite to CO2 reduces
Fe3+ in the silicate phases. However, at the estimated
pressure (~30 kbar) and temperature (~1325°C) of melting for the North Arch magmas, the relative oxygen
fugacities deduced for the parental melts (FMQ – 0·8 to
FMQ+0·7) are too high for graphite to be stable (e.g.
Ballhaus & Frost, 1994). Furthermore, our estimated
CO2 contents for the parental melts are too low to cause
saturation of the melt with CO2 in the melting region.
Another possibility is that melting in the North Arch
source region was triggered by redox reactions involving
methane (redox melting; e.g. Taylor & Green, 1988).
Oxidation of CH4-rich fluids to H2O and CO2 by reduction of Fe3+ could cause melting as water lowers the
NUMBER 7
JULY 1997
peridotite solidus temperature. This hypothesis predicts
that the relative f O2 of partial melts will be inversely
related to the amount of CH4 added to the source region.
As a result, partial melts with lower relative f O2 would
be generated in equilibrium with higher mantle water
contents because methane oxidation produces H2O. If
this process had occurred in the North Arch source
region, then the extent of melting should be controlled
by the amount of water present from oxidation of methane. Thus higher degree, lower f O2 partial melts should
show evidence of higher source concentrations of water,
but our modeling for the North Arch magmas does not
support the metasomatic hypothesis. In addition, as with
graphite, the oxygen fugacities for the North Arch parental melts, estimated based on analyzed Fe3+/RFe in
the glasses, are too high for methane to be present in
the source region (e.g. Wood et al., 1990; Ballhaus &
Frost, 1994).
Based on these lines of evidence, we conclude that the
North Arch magmas were generated in a mantle source
region that was homogeneous with respect to Fe3+/RFe
but did not contain either graphite or CH4-rich fluid.
Measurements of Fe3+/RFe for submarine glasses from
Loihi seamount (Wallace & Carmichael, 1992) show a
correlation with incompatible element abundances similar to that of the North Arch glasses, suggesting that the
effect of variable partial melting on magmatic oxygen
fugacity may be a common feature of Hawaiian volcanism. This pattern is not observed in MORB magmas
(Christie et al., 1986), although the average values of
Fe3+/RFe in MORB are consistent with estimates of
Fe3+/RFe in the upper mantle (O’Neill et al., 1993).
CONCLUSIONS
We have determined the volatile contents and Fe3+/RFe
in glasses from a suite of alkalic lavas from the North
Arch volcanic field, north of Oahu. These glasses were
quenched deep on the seafloor and have retained information as to their volatile-rich initial state. Based on
measured dissolved H2O and CO2 contents, these glasses
were vapor saturated with an H2O–CO2 vapor having
X vapour
of 0·6–0·88 upon quenching on the seafloor at
CO2
400 bar. The high vesicularity of vent samples suggests
that these magmas were extremely gas rich upon eruption.
Calculation of bulk volatile contents for the two most
vesicular glasses yields 1·9±0·1 wt % H2O and 5·4±0·4
wt % CO2.
Dissolved H2O and CO2 contents, vapor compositions,
and vesicularities are consistent with exsolution of >95%
of CO2 and up to 60% of H2O at 400 bar from magmas
whose initial volatile contents are proportional to concentrations of nonvolatile incompatible elements. Data
are bounded by calculated open system and closed system
934
DIXON et al.
VOLATILES IN NORTH ARCH ALKALIC BASALTS
degassing models with initial CO2/H2O=4, suggesting
that variable amounts of exsolving vapor escaped from
the magma during eruption and flow. The most vesicular
samples are least modified by gas loss from the magma
as manifested by their high gas contents, high X vapour
CO2 ,
estimated bulk H2O/P2O5 of ~3, low dissolved H2O, low
dissolved H2O/P2O5, and high dissolved CO2, and are
consistent with closed system degassing using an initial
CO2/H2O of 2·5±1·5. The other lavas lost gas to variable
extents during eruption and flow away from the vent as
manifested by their lower vesicularities, higher dissolved
H2O, higher H2O/P2O5, and low dissolved CO2 contents.
Predicted vesicularities are high for the low degree partial
melts and probably resulted in submarine fire fountaining
that produced the glassy spheres in hydroclastites.
Concentrations of noble gases in bulk glass samples
are extremely low, consistent with extensive magmatic
degassing. The 3He/4He in the crush samples with >5
ncc/g of helium ranges from 5×R A to 8×R A. The
isotopic composition of the heavy noble gases in the melt
fraction is almost entirely atmospheric, consistent with
loss of mantle volatiles by magmatic degassing and the
incorporation of seawater into the lava, possibly during
the minor amounts of devitrification of the glass.
The high dissolved S contents of most of the North
Arch glasses suggest that most samples were saturated
with immiscible Fe–S–O liquid at the time of eruption
and quenching. In contrast to other volatile trace elements
such as Cl, S does not behave simply as an incompatible
element during melting and crystallization because its
concentration is solubility controlled. If the original parental magmas were sulfide saturated during melt generation, then the mantle source must contain
[192 p.p.m. S.
We have presented a complex data set with a simple
conclusion: variations in major, trace, and volatile element concentrations and Fe3+/RFe in alkalic basalts
from the North Arch volcanic field can be produced by
variable extents of melting of a homogeneous source
followed by olivine crystallization and degassing that
reaches vapor saturation at the eruption depth. Assuming
the parental magmas were produced by 1·6–9·0% batch
melting and a D P0 2O5=0·011, the mantle source is estimated to contain 525±75 p.p.m. H2O, 1300±800
p.p.m. CO2, and 30±6 p.p.m. Cl, with a whole-rock
Fe3+/RFe=0·029±0·006.
ACKNOWLEDGEMENTS
J. Crisp graciously provided access to her image analysis
system at JPL for estimation of vesicle contents of lavas.
We thank I. Carmichael for the wet chemical analyses
of ferrous iron. Reviews by J. Moore, D. Graham, and
an anonymous reviewer greatly improved the manuscript.
Discussions with J. Natland concerning sulfur were enlightening. Support was provided by NSF OCE-9302574
to J.E.D.
REFERENCES
Adam, J., 1988. Dry, hydrous, and CO2-bearing liquidus phase relationships in the CMAS system at 28 kb, and their bearing on the
origin of alkali basalts. Journal of Geology 96, 709.
Allègre, C. J., Staudacher, Th. & Sarda, Ph., 1986–1987. Rare gas
systematics: formation of the atmosphere, evolution and structure
of the Earth’s mantle. Earth and Planetary Science Letters 81, 127–150.
Amundsen, H. E. F., 1987. Peridotite xenoliths from Gran Canaria,
Canary Islands: evidence for metasomatic process and partial melting
in the lower oceanic crust. Neues Jahrbuch für Mineralogie, Abhandlungen
156, 121–140.
Anderson, A. T., Jr & Brown, G. G., 1993. CO2 contents and formation
pressures of some Kilauean melt inclusions. American Mineralogist 78,
794–803.
Ballhaus, C. & Frost, B. R., 1994. The generation of oxidized CO2bearing basaltic melts from reduced CH4-bearing upper mantle
sources. Geochimica et Cosmochimica Acta 58, 4931–4940.
Blank, J. G. & Brooker, R. A., 1994. Experimental studies of carbon
dioxide in silicate melts: solubility, speciation, and stable carbon
isotope behavior. Mineralogical Society of America, Reviews in Mineralogy
30, 157–186.
Blank, J. G., Delaney, J. R. & Des Marais, D. J., 1993. The concentration
and isotopic composition of carbon in basaltic glasses from the Juan
de Fuca Ridge, Pacific Ocean. Geochimica et Cosmochimica Acta 57,
875–887.
Bottinga, Y. & Javoy, M., 1990. MORB degassing: bubble growth and
ascent. Chemical Geology 81, 255–270.
Bryndzia, L. T., Wood, B. J. & Dick, H. J. B., 1989. The oxidation state
of the Earth’s sub-oceanic mantle from oxygen thermobarometry of
abyssal spinel peridotites. Nature 341, 526–527.
Canil, D., O’Neill, H. St. C., Pearson, D. G., Rudnick, R. L., McDonough, W. F. & Carswell, D. A., 1994. Ferric iron in peridotites and
mantle oxidation states. Earth and Planetary Science Letters 123, 205–220.
Carmichael, I. S. E. & Ghiorso, M. S., 1986. Oxidation–reduction
relations in basic magma: a case for homogeneous equilibria. Earth
and Planetary Science Letters 78, 200–210.
Cashman, K. V. & Mangan, M. T., 1994. Physical aspects of magmatic
degassing II. Constraints on vesiculation processes from textural
studies of eruptive products. Mineralogical Society of America, Reviews in
Mineralogy 30, 447–478.
Cashman, K. V. & Marsh, B. D., 1988. Crystal size distribution
(CSD) in rocks and the kinetics and dynamics of crystallization. 2.
Makaopuhi Lava Lake. Contributions to Mineralogy and Petrology 99,
292–305.
Cashman, K. V., Mangan, M. T. & Newman, S., 1994. Surface
degassing and modifications to vesicle size distributions in Kilauea
basalt. Journal of Volcanology and Geothermal Research 61, 45–68.
Chen, C. Y. & Frey, F. A., 1983. Origin of Hawaiian tholeiites and
alkalic basalt. Nature 302, 785–789.
Chen, C. Y. & Frey, F. A., 1985. Trace element and isotopic geochemistry of lavas from Haleakala volcano, East Maui, Hawaii:
implications for the origin of Hawaiian basalts. Journal of Geophysical
Research 90, 8743–8768.
Chen, C.-Y., Frey, F. A., Garcia, M. O., Dalrymple, G. B. & Hart, S.
R., 1991. The tholeiite to alkalic basalt transition at Haleakala
935
JOURNAL OF PETROLOGY
VOLUME 38
Volcano, Maui, Hawaii. Contributions to Mineralogy and Petrology 106,
183–200.
Christie, D. M., Carmichael, I. S. E. & Langmuir, C. H., 1986.
Oxidation states of mid-ocean ridge basalt glasses. Earth and Planetary
Science Letters 79, 397–411.
Clague, C. A., 1987. Hawaiian xenolith populations, magma supply
rates, and development of magma chambers. Bulletin of Volcanology
49, 577–587.
Clague, D. A. & Beeson, M. H., 1980. Trace element geochemistry of
the East Molokai Volcanic Series, Hawaii. American Journal of Science
280-A, 820–844.
Clague, D. A. & Dalrymple, G. B., 1988. Age and petrology of
alkalic postshield and rejuvenated stage lava from Kauai, Hawaii.
Contributions to Mineralogy and Petrology 99, 202–218.
Clague, D. A. & Dixon, J. E., 1991. Volatiles in submarine nephelinitic
to basanitic glasses from the North Arch volcanic field. EOS, Transactions of the American Geophysical Union 72, Winter Meeting Supplement,
563.
Clague, D. A. & Dixon, J. E., 1993. Volatiles in Hawaiian magmas.
International Workshop on Interplate Volcanism—Polynesian Plume Province,
Papeete, Tahiti. Paris: Institut de Physique du Globe, p. 17.
Clague, D. A. & Frey, F. A., 1982. Petrology and trace element
geochemistry of the Honolulu Volcanics, implications for the oceanic
mantle beneath Hawaii. Journal of Petrology 23, 447–504.
Clague, D. A., Holcomb, R. T., Sinton, J. M., Detrick, R. S. &
Torresan, M. R., 1990. Pliocene and Pleistocene alkalic flood basalts
on the seafloor north of the Hawaiian islands. Earth and Planetary
Science Letters 98, 175–191.
Clague, D. A., Weber, W. S. & Dixon, J. E., 1991. Picritic glasses from
Hawaii. Nature 353, 553–556.
Clague, D. A., Moore, J. G., Dixon, J. E. & Friesen, W. B., 1995.
Petrology of submarine lavas from Kilauea’s Puna Ridge, Hawaii.
Journal of Petrology 36, 299–349.
Craig, H. & Poreda, R., 1986. Cosmogenic 3He in the summit lavas
of Maui. Proceedings of the National Academy of Science 83, 1970–1974.
Delaney, J. R., Muenow, D. W. & Graham, D. G., 1978. Abundance
and distribution of water, carbon and sulfur in the glassy rims of
submarine pillow basalts. Geochimica et Cosmochimica Acta 42, 581–594.
Des Marais, D. J. & Moore, J. G., 1984. Carbon and its isotopes in
mid-oceanic basaltic glasses. Earth and Planetary Science Letters 69,
43–57.
Dick, H. J. B., Fisher, R. L. & Bryan, W. B., 1984. Mineralogic
variability of the uppermost mantle along mid-ocean ridges. Earth
and Planetary Science Letters 69, 88–106.
Dixon, J. E., 1997. Degassing of alkalic basalts. American Mineralogist 82,
368–378.
Dixon, J. E. & Pan, V., 1995. Determination of the molar absorptivity
of dissolved carbonate in basanitic glass. American Mineralogist 80,
1339–1342.
Dixon, J. E. & Stolper, E. M., 1995. An experimental study of water
and carbon dioxide solubilities in mid-ocean ridge basaltic liquids.
Part II: Application to degassing of basaltic liquids. Journal of Petrology
36, 1633–1646.
Dixon, J. E., Stolper, E. M. & Delaney, J. R., 1988. Infrared spectroscopic measurements of CO2 and H2O glasses in the Juan de
Fuca Ridge basaltic glasses. Earth and Planetary Science Letters 90,
87–104.
Dixon, J. E., Clague, D. A. & Stolper, E. M., 1991. Degassing history
of water, sulfur, and carbon in submarine lavas from Kilauea
volcano, Hawaii. Journal of Geology 99, 371–394.
Dixon, J. E., Stolper, E. M. & Holloway, J. R., 1995. An experimental
study of water and carbon dioxide solubilities in mid-ocean ridge
NUMBER 7
JULY 1997
basaltic liquids. Part I: Calibration and solubility results. Journal of
Petrology 36, 1607–1631.
Dymond, J. & Hogan, L., 1973. Noble gas abundance patterns in
deep-sea basalts—primordial gases from the mantle. Earth and Planetary Science Letters 20, 131–139.
Exley, R. A., Mattey, D. P., Clague, D. A. & Pillinger, C. T., 1986.
Carbon isotope systematics of a mantle ‘hot spot’: a comparison of
Loihi Seamount and MORB glasses. Earth and Planetary Science Letters
78, 189–199.
Fine, G. & Stolper, E. M., 1986. Dissolved carbon dioxide in basaltic
glasses: concentrations and speciation. Earth and Planetary Science Letters
76, 263–278.
Frey, F. A. & Roden, M. F., 1987. The mantle source for the Hawaiian
Islands: constraints from the lavas and ultramafic inclusions. In:
Menzies, M. A. & Hawkesworth, C. J. (eds) Mantle Metasomatism.
London: Academic Press, pp. 423–463.
Gaetani, G. A., Grove, T. L. & Bryan, W. B., 1993. The influence of
water on the petrogenesis of subduction-related igneous rocks. Nature
365, 332–334.
Gerlach, T. M., 1986. Exsolution of H2O, CO2, and S during eruptive
episodes at Kilauea Volcano, Hawaii. Journal of Geophysical Research
91, 12177–12185.
Gerlach, T. M., 1989. Degassing of carbon dioxide from basaltic
magma at spreading centers: II. Mid-oceanic ridge basalts. Journal
of Volcanology and Geothermal Research 39, 221–232.
Gerlach, T. M., 1991. Comment on ‘Mid-ocean ridge popping rocks:
implications for degassing at ridge crest’ by P. Sarda and D. Graham.
Earth and Planetary Science Letters 105, 566–567.
Gerlach, T. M. & Graeber, E. J., 1985. Volatile budget of Kilauea
volcano. Nature 313, 273–277.
Gerlach, T. M. & Taylor, B. E., 1990. Carbon isotope constraints on
degassing of carbon dioxide from Kilauea Volcano. Geochimica et
Cosmochimica Acta 54, 2051–2058.
Gerlach, T. M. & Thomas, D. M., 1986. Carbon and sulfur isotopic
composition of Kilauea parental magma. Nature 319, 480–482.
Ghiorso, M. S. & Sack, R. O., 1995. Chemical mass transfer in
magmatic systems IV. A revised and internally consistent thermodynamic model for the interpolation and extrapolation of liquid–solid
equilibria in magmatic systems at elevated temperatures and pressures. Contributions to Mineralogy and Petrology 119, 197–212.
Graham, D. & Sarda, P., 1991. Reply to comment by T. M. Gerlach
on ‘Mid-ocean ridge popping rocks: implications for degassing at
ridge crests’. Earth and Planetary Science Letters 105, 568–573.
Graham, D. W., Jenkins, W. J., Kurz, M. D. & Batiza, R., 1987.
Helium isotope disequilibrium and geochronology of glassy submarine basalts. Nature 326, 384–386.
Green, D. H., 1972. Magmatic activity as the major process in the
chemical evolution of the Earth’s crust and mantle. Tectonophysics 13,
47–71.
Green, D. H., 1973. Contrasted melting relations in a pyrolite upper
mantle under mid-ocean ridge, stable crust and island arc environments. Tectonophysics 17, 285–297.
Green, D. H. & Ringwood, A. E., 1967. The genesis of basaltic
magmas. Contributions to Mineralogy and Petrology 15, 103–190.
Green, D. H. & Wallace, M. E., 1988. Mantle metasomatism by
ephemeral carbonatite melts. Nature 336, 459–462.
Greenland, L. P., Rose, W. I. & Stokes, J. B., 1985. An estimate of
gas emissions and magmatic gas content from Kilauea volcano.
Geochimica et Cosmochimica Acta 49, 125–129.
Greenland, L. P., Okamura, A. T. & Stokes, J. B., 1988. Constraints
on the mechanics of the eruption. US Geological Survey Professional
Paper 1463, 155–164.
936
DIXON et al.
VOLATILES IN NORTH ARCH ALKALIC BASALTS
Head, J. W. & Wilson, L., 1987. Lava fountain heights at Pu’u O’o,
Kilauea, Hawaii: indicators of amount and variations of exsolved
magma volatiles. Journal of Geophysical Research 92, 13715–13719.
Helz, R. T., 1973. Phase relations of basalt in their melting ranges at
PH2O=5 kb as a function of oxygen fugacity. Journal of Petrology 14,
249–302.
Helz, R. T., 1976. Phase relations of basalt in their melting ranges at
PH2O=5 kb. Part II: Melt compositions. Journal of Petrology 17,
139–193.
Hirose, K. & Kawamoto, T., 1995. Hydrous partial melting of lherzolite
at 1 GPa: the effect of H2O on the genesis of basaltic magmas. Earth
and Planetary Science Letters 133, 463–473.
Holloway, J. R. & Burnham, C. W., 1972. Melting relations of basalt
with equilibrium water pressure less than total pressure. Journal of
Petrology 13, 1–29.
Holloway, J. R. & Blank, J. G., 1994. Application of experimental
results to C–O–H species in natural melts. Mineralogical Society of
America, Reviews in Mineralogy 30, 187–230.
Honda, M., McDougall, I., Patterson, D. B., Doulgeris, A. & Clague,
D. A., 1991. Possible solar noble-gas component in Hawaiian basalts.
Nature 349, 149–151.
Ihinger, P. D., Hervig, R. L. & McMillan, P. F., 1994. Applications of
experimental results to C–O–H species in natural melts. Mineralogical
Society of America, Reviews in Mineralogy 30, 67–121.
Iwasaki, B. & Katsura, T., 1967. The solubility of hydrogen chloride
in volcanic rock melts at total pressure of one atmosphere and at
temperatures of 1200°C and 1290°C under anhydrous conditions.
Bulletin of the Chemical Society of Japan 40, 554.
Jackson, L. L., Brown, F. W. & Neil, S. T., 1987. Major and minor
elements requiring individual determination, classical whole-rock
analysis, and rapid rock analysis. In: Baedecker, P. A. (ed.) Methods
of Geochemical Analysis. US Geological Survey, Bulletin 1770, G1–G23.
Jambon, A., 1994. Earth degassing and large-scale geochemical cycling
of volatile elements. Mineralogical Society of America, Reviews in Mineralogy
30, 479–517.
Jambon, A. & Zimmermann, J. L., 1987. Major volatiles from a North
Atlantic MORB glass and calibration to He: a size fraction analysis.
Chemical Geology 62, 177–189.
Jambon, A., Weber, H. W. & Begemann, F., 1985. Helium and argon
from an Atlantic MORB glass: concentration, distribution and
isotopic composition. Earth and Planetary Science Letters 73, 255–267.
Javoy, M., Pineau, F. & Iiyama, I., 1978. Experimental determination
of the isotopic fractionation between gaseous CO2 and carbon
dissolved in a tholeiitic magma: preliminary results. Contributions to
Mineralogy and Petrology 67, 35–39.
Javoy, M. & Pineau, F., 1991. The volatiles record of a ‘popping’
rock from the Mid-Atlantic Ridge at 14°N: chemical and isotopic
composition of gas trapped in the vesicles. Earth and Planetary Science
Letters 107, 598–611.
Kadik, A. A., 1995. Formation of carbon species in terrestrial magmas.
In Farley, K. A. (ed.) Volatiles in the Earth and Solar System. American
Institute of Physics Conference Proceedings 341, 106–114.
Khitarov, N. I. & Kadik, A. A., 1973. Water and carbon dioxide in
magmatic melts and peculiarities of the melting process. Contributions
to Mineralogy and Petrology 41, 205–215.
Kingsley, R. H. & Schilling, J.-G., 1995. Carbon in Mid-Atlantic Ridge
basalt glasses from 28°N to 63°N: evidence for a carbon enriched
Azores mantle plume. Earth and Planetary Science Letters 129, 31–53.
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.
Contributions to Mineralogy and Petrology 108, 82–92.
Kurz, M. D. & Jenkins, W. J., 1981. The distribution of helium in
oceanic basalt glass. Earth and Planetary Science Letters 53, 41–54.
Kurz, M. D., Jenkins, W. J., Hart, S. R. & Clague, D. A., 1983.
Helium isotopic variations in volcanic rocks from Loihi Seamount
and the Island of Hawaii. Earth and Planetary Science Letters 66, 388–406.
Kushiro, I., 1970. Stability of amphibole and phlogopite in the upper
mantle. Carnegie Institute of Washington, Yearbook 68, 245–247.
Kushiro, I., Syono, Y. & Akimoto, S., 1968. Melting of a peridotite
nodule at high pressures and high water pressures. Journal of Geophysical
Research 73, 6023–6029.
Lange, R. A., 1994. The effect of H2O, CO2 and F on the density and
viscosity of silicate melts. Mineralogical Society of America, Reviews in
Mineralogy 30, 332–369.
Luth, R. W., Virgo, D., Boyd, F. R. & Wood, B. J., 1990. Ferric iron
in mantle-derived garnets, implications for thermobarometry and
for the oxidation state of the mantle. Contributions to Mineralogy and
Petrology 104, 56–72.
Mangan, M. R., Cashman, K. V. & Newman, S., 1993. Vesiculation
of basaltic magma during eruption. Geology 21, 157–160.
Mathez, E. A. & Delaney, J. R., 1981. The nature and distribution of
carbon in submarine basalts and peridotite nodules. Earth and Planetary
Science Letters 56, 217–232.
Mattey, D. P., Carr, R. H., Wright, I. P. & Pillinger, C. T., 1984.
Carbon isotopes in submarine basalts. Earth and Planetary Science Letters
70, 196–206.
Mattey, D. P., Exley, R. A. & Pillinger, C. T., 1989. Isotopic composition
of CO2 and dissolved carbon in glass. Geochimica et Cosmochimica Acta
53, 2377–2386.
McMillan, P., 1994. Water solubility and speciation models. Mineralogical
Society of America, Reviews in Mineralogy 30, 131–156.
Melson, W. G., O’Hearn, T. & Fredriksson, K., 1988. Composition
and origin of basaltic glass spherules in pelagic clay from the eastern
Pacific. Marine Geology 83, 253ff.
Michael, P. J., 1988. The concentration, behavior and storage of H2O in
the suboceanic upper mantle: implications for mantle metasomatism.
Geochimica et Cosmochimica Acta 52, 555–566.
Michael, P. J., 1995. Regionally distinctive sources of depleted MORB:
evidence from trace elements and H2O. Earth and Planetary Science
Letters 131, 301–320.
Michael, P. J. & Chase, R. L., 1987. The influence of primary
magma composition, H2O and pressure on mid-ocean ridge basalt
differentiation. Contributions to Mineralogy and Petrology 96, 245–263.
Michael, P. J. & Schilling, J.-G., 1989. Chlorine in mid-ocean ridge
magmas: evidence for assimilation of seawater-influenced components. Geochimica et Cosmochimica Acta 53, 3131–3143.
Moore, J. G., 1970. Relationship between subsidence and volcanic
load, Hawaii. Bulletin Volcanologique 34, 562–576.
Moore, J. G., 1979. Vesicularity and CO2 in mid-ocean ridge basalt.
Nature 282, 250–253.
Moore, J. G. & Schilling, J.-G., 1973. Vesicles, water, and sulfur in
Reykjanes Ridge basalts. Contributions to Mineralogy and Petrology 41,
105–118.
Moore, J. G., Batchelder, J. N. & Cunningham, C. G., 1977. CO2filled vesicles in mid-ocean basalt. Journal of Volcanology and Geothermal
Research 2, 309–327.
Muenow, D. W., Graham, D. G., Liu, N. W. K. & Delaney, J. R.,
1979. The abundance of volatiles in Hawaiian tholeiitic submarine
basalts. Earth and Planetary Science Letters 42, 71–76.
Newman, S., 1989. Water and carbon dioxide contents in basaltic
glasses from the Mariana Trough. EOS, Transactions of the American
Geophysical Union 70, Fall Meeting Supplemental, 1387.
Newman, S., 1990. Water and carbon dioxide contents of back arc
basin basalts. V. M. Goldschmidt Conference 1990 Programs and Abstracts,
69.
937
JOURNAL OF PETROLOGY
VOLUME 38
O’Neill, H. St. C., 1991. The origin of the Moon and the early history
of the Earth—a chemical model. Part 2: The Earth. Geochimica et
Cosmochimica Acta 55, 1159–1172.
O’Neill, H. St. C. & Wall, V. J., 1987. The olivine–
orthopyroxene–spinel oxygen geobarometer, the nickel precipitation
curve, and the oxygen fugacity of the Earth’s upper mantle. Journal
of Petrology 28, 1169–1191.
O’Neill, H. St. C., Rubie, D. C., Canil, D., Geiger, C. A., Ross, C.
R., II, Seifert, F. & Woodland, A. B., 1993. Ferric iron in the upper
mantle and in transition zone assemblages: implications for relative
oxygen fugacities in the mantle. In: Evolution of the Earth and Planets,
Geophysical Monograph, American Geophysical Union 74, IUGG Vol. 14,
73–88.
Pan, V., Holloway, J. R. & Hervig, R. L., 1991. The pressure and
temperature dependence of carbon dioxide solubility in tholeiitic
basalt melts. Geochimica et Cosmochimica Acta 55, 1587–1595.
Pandya, N., Sharma, S. K., Muenow, D. W., & Sherriff, B. L., 1992.
Hydration of alkali silicate glasses at ambient conditions. EOS,
Transactions American Geophysical Union 73, Spring Meeting Supplement,
361.
Patterson, D. B., Honda, M. & McDougall, I., 1990. Atmospheric
contamination: a possible source for heavy noble gases in basalts
from Loihi Seamount, Hawaii. Geophysical Research Letters 17, 705–708.
Pineau, F., Javoy, M. & Bottinga, Y., 1976. 13C/12C ratios of rocks and
inclusions in popping rocks of the mid-Atlantic ridge. Their bearing
on the problem of isotopic composition of deep seated carbon. Earth
and Planetary Science Letters 29, 413–421.
Pineau, F. & Javoy, M., 1983. Carbon isotopes and concentrations
in mid-oceanic ridge basalts. Earth and Planetary Science Letters 62,
239–257.
Pineau, F. & Javoy, M., 1994. Strong degassing at ridge crests: the
behaviour of dissolved carbon and water in basalt glasses at 14°N,
Mid-Atlantic Ridge. Earth and Planetary Science Letters 123, 179–198.
Poreda, R. J. & Farley, K. A., 1992. Rare gases in Samoan xenoliths.
Earth and Planetary Science Letters 113, 129–144.
Rees, B. A., 1991. Seismic stratigraphy and magnetostratigraphy of
the Hawaiian flexural moat. Master’s Thesis, University of Rhode
Island, 146 pp.
Rison, W. & Craig, H., 1983. Helium isotopes and mantle volatiles in
Loihi Seamount and Hawaiian Island basalts and xenoliths. Earth
and Planetary Science Letters 66, 407–426.
Sack, R. O., Carmichael, I. S. E., Rivers, M. & Ghiorso, M. S.,
1980. Ferric–ferrous equilibria in natural silicate liquids at 1 bar.
Contributions to Mineralogy and Petrology 75, 369–376.
Sakai, H., Des Marais, D. J., Ueda, A. & Moore, J. G., 1984.
Concentrations and isotope ratios of carbon, nitrogen and sulfur in
ocean-floor basalts. Geochimica et Cosmochimica Acta 48, 371–379.
Sarda, P. & Graham, D., 1990. Mid-ocean ridge popping rocks:
implications for degassing at ridge crests. Earth and Planetary Science
Letters 97, 268–289.
Sarda, Ph., Staudacher, Th. & Allègre, C. J., 1985. 40Ar/36Ar in MORB
glasses: constraints on atmosphere and mantle evolution. Earth and
Planetary Science Letters 72, 357–375.
Schilling, J.-G., Bergeron, M. B. & Evans, R., 1980. Halogens in the
mantle beneath the North Atlantic. Philosophical Transactions of the
Royal Society of London, Series A 297, 147–178.
Sen, G., 1988. Petrogenesis of spinel lherzolite and pyroxenite suite
xenoliths from the Koolau shield, Oahu, Hawaii: implications for
petrology for the post-eruptive lithosphere beneath Oahu. Contributions to Mineralogy and Petrology 100, 61–91.
Shaw, H. R., 1972. Viscosities of magmatic silicate liquids: an empirical
method of prediction. American Journal of Science 272, 870–893.
NUMBER 7
JULY 1997
Shilobreyeva, S. N., Kadik, A. A. & Lukanin, O. A., 1983. Outgassing
of ocean-floor magma as a reflection of volatile conditions in the
magma generation region. Geokhimiya 9, 1257–1274.
Silver, L. A., Ihinger, P. D. & Stolper, E. M., 1990. The influence of
bulk composition on the speciation of water in silicate glasses.
Contributions to Mineralogy and Petrology 104, 142–162.
Sisson, T. W. & Grove, T. L., 1993. Experimental investigations of
the role of H2O in calc-alkaline differentiation and subduction zone
magmatism. Contributions to Mineralogy and Petrology 113, 143–166.
Smith, T. L. & Batiza, R., 1989. New field and laboratory evidence
for the origin of hyaloclastite flows on seamount summits. Bulletin of
Volcanology 51, 96–114.
Sparks, R. S. J., Pinkerton, H. & MacDonald, R., 1977. The transport
of xenoliths in magma. Earth and Planetary Science Letters 35, 234–238.
Sparks, R. S. J., Barclay, J., Jaupart, C., Mader, H. M. & Phillips, J.
C., 1994. Physical aspects of magmatic degassing I. Experimental
and theoretical constraints on vesiculation. Mineralogical Society of
America, Reviews in Mineralogy 30, 413–445.
Spera, F. J., 1984. Carbon dioxide in igneous petrogenesis: III. Role
of volatiles in the ascent of alkaline magma with special reference
to xenolith-bearing mafic lavas. Contributions to Mineralogy and Petrology
88, 217–232.
Stolper, E. M. & Holloway, J. R., 1988. Experimental determination
of the solubility of carbon dioxide in molten basalt at low pressure.
Earth and Planetary Science Letters 87, 397–408.
Stolper, E. M. & Newman, S., 1994. The role of water in the
petrogenesis of Mariana trough magmas. Earth and Planetary Science
Letters 121, 293–325.
Sun, S.-S., Nesbitt, R. W. & Sharaskin, A. Ya., 1979. Geochemical
characteristics of mid-ocean ridge basalts. Earth and Planetary Science
Letters 44, 119–138.
Swanson, D. A. & Fabbi, B. P., 1973. Loss of volatiles during fountaining
and flowage of basaltic lava at Kilauea volcano, Hawaii. Journal of
Research, US Geological Survey 1, 649–658.
Taylor, W. R. & Green, D. H., 1988. Measurement of reduced
peridotite–C–O–H solidus and implications for redox melting of the
mantle. Nature 332, 349–352.
Trull, T., Nadeau, S., Pineau, F., Polvé, M. & Javoy, M., 1993. C–He
systematics in hotspot xenoliths: implications for mantle carbon
contents and carbon recycling. Earth and Planetary Science Letters 118,
43–64.
Ulmer, P., 1989. Partitioning of high-field strength elements among
olivine, pyroxene, garnet and calc-alkaline picrobasalt: experimental
results and an application. Carnegie Institute of Washington, Yearbook
1988–1989, 42–47.
Vallier, T. L., Bohrer, D., Moreland, G. & McKee, E. H., 1977. Origin
of basaltic microlapilli in lower Miocene pelagic sediment, north
eastern Pacific. Geological Society of America Bulletin 88, 787.
Vergniolle, S. & Jaupart, C., 1986. Separated two-phase flow and
basaltic eruptions. Journal of Geophysical Research 91, 12842–12860.
Vergniolle, S. & Jaupart, C., 1990. Dynamics of degassing at Kilauea
volcano, Hawaii. Journal of Geophysical Research 95, 2793–2809.
Wallace, P. & Carmichael, I. S. E., 1992. Sulfur in basaltic magmas.
Geochimica et Cosmochimica Acta 56, 1863–1874.
Watson, E. B., 1994. Diffusion in volatile-bearing magmas. Mineralogical
Society of America, Reviews in Mineralogy 30, 371–411.
Wilmoth, R. A. & Walker, G. P. L., 1993. P-type and S-type pahoehoe:
a study of vesicle distribution patterns in Hawaiian lava flows. Journal
of Volcanology and Geothermal Research 55, 129–142.
Wilson, L. & Head, J. W., III, 1981. Ascent and eruption of basaltic
magma on the earth and moon. Journal of Geophysical Research 86,
2971–3001.
938
DIXON et al.
VOLATILES IN NORTH ARCH ALKALIC BASALTS
Wood, B. J. & Virgo, D., 1989. Upper mantle oxidation state: ferric
iron contents of lherzolite spinels by 57Fe Mössbauer spectroscopy
and resultant oxygen fugacities. Geochimica et Cosmochimica Acta 53,
1277–1291.
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.
Yoder, H. S., 1965. Diopside–anorthite–water at five and ten kilobars
and its bearing on explosive volcanism. Carnegie Institute of Washington,
Yearbook 64, 82–89.
Yoder, H. S. & Tilley, C. E., 1962. Origin of basalt magmas; an
experimental study of natural and synthetic rock systems. Journal of
Petrology 3, 342–532.
939