Embed Size (px)
Citation preview
Hawkins, J., Parson, L., Allan, J., et al., 1994Proceedings of the Ocean Drilling Program, Scientific Results, Vol. 135
36. OXIDATION STATE AND SULFUR CONCENTRATIONS IN LAU BASIN BASALTS1
Kristen Nilsson Farley2
ABSTRACT
The backarc glasses recovered during Ocean Drilling Program Leg 135 are unique among submarine tholeiitic glasses withrespect to their oxygen fugacity and sulfur concentrations. Unlike mid-ocean-ridge basalt glasses,/O2 in these samples (inferredfrom ratios Fe3+/Fe2+) is high and variable, and S variations (90-1140 ppm) are not coupled with FeO concentration. Strongcorrelations occur between the alkali and alkaline-earth elements and both /O 2 (positive correlations) and S concentrations(negative correlations). Correlations between/O2 and various trace elements are strongest for those elements with a known affinityfor hydrous fluids (perhaps produced during slab dehydration), suggesting the presence of a hydrous fluid with high/O2 and highalkali and alkaline earth element concentrations in the Lau Basin mantle.
Concentrations of S and/O2 are strongly correlated; high/O2 samples are characterized by low S in addition to high alkali andalkaline earth element concentrations. The negative correlations between S and these trace elements are not consistent withincompatible behavior of S during crystallization. Mass balance considerations indicate that the S concentrations cannot result simplyfrom mixing between low-S and high-S sources. Furthermore, there is no relationship between S and other trace elements or isotoperatios that might indicate that the S variations reflect mixing processes. The S variations more likely reflect the fact that when silicatecoexists with an S-rich vapor phase the solubility of S in the silicate melt is a function of/O2 and is at a minimum at the/O2 conditionsrecorded by these glasses. The absence of Fe-sulfides and the high and variable vesicle contents are consistent with the idea that Sconcentrations reflect silicate-vapor equilibria rather than silicate-sulfide equilibria (as in MORB). The low S contents of somesamples, therefore, reflect the high/O2 of the supra-subduction zone environment rather than a low-S source component.
INTRODUCTION
Numerous studies have demonstrated that backarc basin basalts(BABB) span a compositional range between mid-ocean-ridge basalts(MORB) and island arc volcanics (IAV) (e.g., Hawkins and Melchior,1985; Perfit et al, 1987; Sinton and Fryer, 1987; Volpe et al., 1987,1988; Fryer et al., 1990; Hawkins et al., 1990). The oxygen fugacity(fθ2) of backarc magmas has never been systematically investigated,although it has been suggested to be higher than that of MORB(Hawkins and Melchior, 1985; Sinton and Fryer, 1987; Hochstaedter,Gill, Kusakabe, et al., 1990). This would have several importantimplications. The behavior of elements with multiple valence statesand the solubility of volatile species are strongly affected by/O2. Forexample, variations in the Fe3+/Fe2+ ratio influence the extent andtiming of oxide fractionation and the composition of residual magmas.Furthermore,/S2 and S solubility are intimately related to/O2 and thus/O 2 exerts a strong control on the behavior of S.
This chapter presents /O 2 and S concentrations of BABB glassrecovered in the Lau Basin during Ocean Drilling Program (ODP)Leg 135 (Fig. 1) so as to constrain the relationship between the twoparameters in a wide variety of BABB compositions. The Leg 135backarc samples are an ideal sample set because a large number ofchemically distinct units with adequate amounts of fresh glass wererecovered (Parson, Hawkins, Allan, et al., 1992). Fresh glass is criticalfor/O2 and S analyses because syn- and post-eruptive degassing andalteration reactions can render even fresh crystalline material unsuit-able (e.g., Sato, 1978; Mathez, 1984; Christie et al, 1986).
To address questions concerning the/O2 and S contents of the IAVsource properly, a comprehensive study of pristine, undegassed arcglasses would be desirable. Because such material is extremely rare,
Hawkins, J., Parson, L., Allan, J., et al, 1994. Proc. ODP, Sci. Results, 135: CollegeStation, TX (Ocean Drilling Program).
2 Scripps Institution of Oceanography, University of California, San Diego, La Jolla,CA 92093-0208, U.S.A. (present address: Department of Geology, Whittier College,Whittier, CA 90608).
BABB become a particularly valuable proxy. Numerous isotopic andtrace element studies have concluded that BABB magmas contain anarc-like component (e.g., Hawkins and Melchior, 1985; Sinton andFryer, 1987; Volpe et al., 1987, 1988; Hawkins et al., 1990; Hergt andHawkesworth, this volume); however, unlike IAV, BABB are not ex-tensively degassed and glassy material is abundant. In addition, BABBare typically less fractionated than arc lavas; the more Mg-rich samplesare more likely to reflect the source signature than the effects of crys-tallization. By studying BABB glasses, we can attempt to place someconstraints on the S and/O2 characteristics of the arc-like end-memberin BABB petrogenesis and perhaps on the IAV source as well.
The island arc volcanics, and some BABB, are enriched in certaintrace elements that are thought to be derived from a hydrous fluid (orpossibly melt) produced during dehydration of the downgoing slab(e.g., Perfit et al., 1980; Gill, 1981; Hawkins and Melchior, 1985;Hochstaedter, Gill, and Morris, 1990; McCullogh and Gamble, 1991;Hawkesworth et al., 1991). One goal of this work is to compare the/O 2 and S characteristics of the Leg 135 backarc samples to MORB(about which we know a great deal) and IAV (about which we knowvery little). Those samples that suggest significant deviations fromMORB-like compositions may reflect the incorporation of this hydrousfluid in their/O2 and S signatures.
METHODS
Major element and S concentrations were determined by electronmicroprobe on 73 glasses from the Leg 135 backarc sites. Many ofthese glasses were also analyzed for trace elements and radiogenicisotopes by Hergt and Hawkesworth (this volume). Major elementswere analyzed on the Cameca SX50 electron microprobe at the Uni-versity of California, Davis (UCD). Natural and synthetic standardswere used, operating voltage was 15 kV, sample current was 10 nA,counting time was 20 s per element, and beam diameter was 15 µm.Each analysis represents an average of at least five spots from a sin-gle glass fragment. Basalt glass JDFD2 was analyzed periodicallythroughout the run. Table 1 compares the values obtained for basaltglass JDFD2 with preferred values obtained by microprobe and wetchemical methods (C. Langmuir and J. Bender, unpubl. data, 1992).
603
K.N. FARLEY
15°S
New Zealand
179°W 177 175C
Figure 1. Map of the Lau Basin showing Sites 834, 835, 836, and 839 and the major geologic features of the Tonga Trench and the Lau Basinsystem. Islands shown include Zephyr Shoal (Z), Upolu (U), Tongatapu (T), 'Eua (E), Vava'u (V), 'Ata (A), and Niuafo'ou (NF). The CentralLau (CLSC) and Eastern Lau (ELSC) spreading centers, Valu Fa Ridge (VF), and Mangatolu Triple Junction (MTJ) are also shown. Contourinterval in kilometers. Map from Parson, Hawkins, Allan, et al. (1992).
LAU BASIN BASALTS
Table 1. Compositions of basalt glass JDFD2 ob-
tained by microprobe at UCD and the preferred
values for this glass.
SiO 2
TiO 2
A1 2O 3
FeO*
MnOMgOCaON a 2 O
K2OP 2 O 5
Total
UCDaverage
50.331.84
13.8012.120.216.77
10.692.870.210.21
99.05
Standarddeviation
0.310.060.170.130.040.120.170.070.030.03
Preferredvalue
50.801.93
13.8012.170.226.83
10.802.770.220.23
99.77
Notes: UCD average represents 12 analyses made throughoutthe microprobe run. Preferred value is the preferred com-position for this glass (C. Langmuir and J. Bender, un-publ. data, 1992). In Table 2 Siθ2, Tiθ2, and Na2θ werenormalized (1%, 5%, and 3%, respectively) to correct forthe small systematic differences.
Only for SiO2, TiO2, and Na2O were small normalizations necessaryto correct for consistent differences.
Sulfur concentrations were measured on the Camebax microprobeat the Lamont Doherty Earth Observatory (LDEO). Chalcopyrite wasused as a standard, spot size was 20 µm, operating current was 15 kV,and sample current was 100 nA. Each analysis represents an averageof four spots measured simultaneously on two spectrometers for 40 s(for a total of 320 s per glass). Replicate analyses at LDEO of basaltglass JDFD2 for S yielded an average of 1400 ± 40 ppm, in excellentagreement with the value of 1420 ± 40 ppm obtained by Wallace andCarmichael (1992).
Ferrous iron was determined on glass fragments chipped frompillow margins and hand-picked to avoid alteration products, coatingmaterial, or phenocrysts. Fe2+ was determined using titration methodssimilar to those described by Wilson (1960). All samples were run induplicate. Fe2O3 was determined by difference from FeO* as meas-ured by microprobe. To facilitate comparison with the large data setof Christie et al. (1986), four analyses of JDFD2 were made. The FeOvalues obtained (11.01 to 11.10) agree well with the FeO content of11.08 ± 0.09 reported by Christie et al. (1986).
Even dilute acid leaches of the glass fragments appear to producedisparate and spurious Fe3+/Fe2+ data (K. Nilsson, unpubl. data, 1990).Extreme care was used in picking the glass fragments to ensure that noleaching was necessary. In all but one case, samples from a singlegeochemical unit produced very similar/O2 results. Only in basalticglasses from Site 834 (Subunit 2a) were significant/O2 differencesobserved; other geochemical characteristics also suggest that this unitspans a large compositional range. In several cases, it was necessaryto combine glasses from the same compositional unit to obtain enoughclean material for the titrations. To determine if combining glasses froma single unit might introduce unacceptable errors, three separate glassesfrom Site 834 (Unit 10) and one bulk glass from this unit were analyzed.The aggregate glass (FeO = 8.58) and the individual samples (FeO =8.53 to 8.69) yielded similar results.
Oxygen fugacity was calculated from Fe3+/Fe2+ using the regres-sion equation of Sack et al. (1980) and the coefficients of Kilinc et al.(1983). Precision of/O2 calculated in this way has been estimated tobe 0.3 to 0.5 log units (Kilinc et al, 1983; Carmichael, 1991). Tocircumvent its strong temperature dependence,/O2 is expressed rela-tive to the curve of the nickel-nickel-oxide/O2 buffer (NNO; Huebnerand Sato, 1970). Thus, positive ΔNNO are relatively oxidizing con-ditions whereas negative ΔNNO reflect more reducing conditions.The FMQ (fayalite-magnetite-quartz) buffer lies at ΔNNO = -0.8(Carmichael, 1991).
RESULTS
Samples
Sites 834, 835, 836, and 839 yielded fresh glass suitable for in-clusion in this study (Fig. 1). Details concerning the geographic setting,water depth, and sample depth can be found in the relevant site reportsin Parson, Hawkins, Allan, et al. (1992). Water depth at these sitesranged from 2450 to 2905 m and the samples themselves were col-lected from between 25 and 500 m below seafloor (mbsf). Units 1,2, and 3 at Site 836 may not represent igneous basement (Parson,Hawkins, Allan, et al., 1992) as these glasses occurred in volcanicgravels (Units 1 and 2) and large sedimentary clasts (Unit 3) immedi-ately overlying the igneous basement.
No simple relationship exists between geochemical signature andeither depth or geographic position among the samples (Parson, Hawkins,Allan, et al., 1992; Hawkins and Allan, this volume; Hergt and Hawkes-worth, this volume; Hergt and Nilsson, this volume). Major elementvariations are discussed elsewhere (e.g., Ewart, this volume; Hawkinsand Allan, this volume; Hergt and Nilsson, this volume), and it willonly be noted that these glasses comprise a wide range of compositionsindicative of various degrees of fractionation (SiO2 = 49%-57%; Table2). The glasses vary in major and trace element composition fromprimitive basalts indistinguishable from MORB (Site 834, Unit 7) tobasalts and andesites with a strong resemblance to IAV lavas (Site 839,all units).
Detailed petrographic descriptions are presented in the relevant sitereports (Parson, Hawkins, Allan, et al., 1992), but those characteristicsthat pertain to this study will be briefly summarized. Oxide mineralsare common as skeletal grains interstitial to silicate minerals, but theyare extremely rare as phenocrysts. In ferroandesites from Site 834, Unit12 (hereafter referred to as Unit 834-12), euhedral titanomagnetitegrains occur in the glassy rim, suggesting that this was a liquidus phaseat the time of eruption. Only in the two relatively coarse-graineddiabasic units (834-5 and 836-4) do relatively large crystals of ilmeniteand magnetite coexist. Although some lamellae reflecting oxyexsolu-tion are present, their development is not extensive.
Only two basaltic units (834-7 and 834-6a) have Fe-sulfide blebscoexisting with basaltic glass. Sulfides are, however, fairly commonin the crystalline interiors of most rock types, where they are oftenobserved adjacent to or (rarely) included in Plagioclase phenocrysts.Sulfides are mostly pyrrhotite and chalcopyrite compositions (K.Nilsson, unpubl. data, 1992), similar in composition to sulfide glob-ules found in other seafloor lavas (Mathez and Yeats, 1976; Czaman-ske and Moore, 1977; Mathez, 1980; Perfit and Fornari, 1983).
Oxygen Fugacity
The Fe^/ZFe ratio ranges from 0.103 to 0.286 in these glasses,corresponding to a range in ΔNNO of-1.37 to +1.02 (Table 3). At agiven MgO content, these BABB glasses are 1 to 2 log units moreoxidized and the range in/O2 is greater than in MORB glasses (Fig.2). The Fe3+/Fe2+ ratio of some of the Leg 135 samples infers condi-tions more oxidized than the NNO buffer and is therefore higher thanhas been measured in any other deep-sea glasses. Some of the samplesfrom Site 834 may suggest a parallel trend to MORB at approximately1 log unit higher/O2. The scatter among the data set as a whole withrespect to MgO-/O2 variations indicates that the/O2 variations do notreflect rock types or degree of fractionation but must largely reflectvariations in the/O2 of the source.
Sulfur
Sulfur concentrations range from 90 to 1140 ppm (Table 2) andonly the basaltic glasses from Units 834-7, 834-6a, 835-1, 836-3, and836-4 have Fe-S relationships that place them near the empirically
605
K.N. FARLEY
Table 2. Major element (in wt%) and sulfur concentrations (in ppm) in Leg 135 backarc glasses.
Core, section,interval (cm)
135-834 A-13X-1, 18-02414X-1,25-31
135-834B-14R-1,28-3215R-1.0-815R-2, 35-3715R-2, 96-10418R-1.0-618R-1.7-1319R-1, 5-1022R-1, 41-4626R-1, 107-10930R-2, 98-10431R-1, 98-10431R-2, 37^231R-2,43^832R-1, 0-533R-l,0-634R-2, 60-6335R-1, 15-2335R-1, 31-3735R-2, 67-6935R-2, 70-7235R-2, 77-8336R-1, 0-736R-l,7-1037R-2, 42-4439R-1, 0-739R-l,7-1039R-1,9-1540R-1,45^741R-l,0-547R-1, 130-13550R-1,87-9251R-1, 10-1553R-1,7-2054R-1,0-554R-1, 5_io57R-1,0-556R-2, 111-11857R-1, 34-4059R-2, 30-3259R-2, 52-55
135-835B-3R-1, 123-1273R-2, 0-44R-1,67-734R-1,73-785R-1, 14-217R-3, 135-140
135-836A-3H-3, 23-243H-3, 90-953H4, 53-1103H-CC, 0-74H-CC0-137X-1, 15-22
135-836B-3R-1,47-529M-1, 97-102
135-839B-18R-l,0-619R-1, 15-1919R-1, 38^119R-1,64-6920R-1, 36-4121R-1, 53-5622R-1, 19-2322R-1, 24-2922R-1,30-3623R-1,28-3123R-1, 32-3523R-1, 40-4425R-1, 11-1625R-1, 48-55327R-1,29-3342R-1,0-6
Unit orsubunit
2a2a
6a6b6a6a6c6c6c77777777
888888
10a*10a10a10a10a10a1212121212121213131313
111111
112334
4a?
2333333333333469
SiO2
50.6251.57
52.3450.7952.2052.1651.5551.8351.7650.1249.9750.0950.2349.6250.1249.4450.3252.4952.6652.3652.2352.3552.3652.4052.3255.5751.9753.0452.3352.3753.2155.5456.0656.1555.0855.5955.1054.8051.2651.2051.3451.02
51.8651.8652.5552.3651.9152.29
55.6955.6057.9949.9550.0349.97
50.0853.13
55.6053.2652.8752.9853.2952.8453.0553.1153.1153.5852.9652.7652.7153.2053.3355.38
TiO2
1.711.60
1.961.292.001.951.391.441.411.371.441.351.431.451.411.371.431.271.351.261.251.291.281.251.282.011.691.711.681.791.762.152.212.252.182.312.252.251.381.331.301.38
1.071.041.101.671.121.02
1.271.311.150.890.840.89
0.901.13
1.100.690.640.680.740.700.710.730.720.700.700.730.710.820.681.05
A12O3
16.2416.01
15.7116.3815.6415.5916.5516.7316.7416.5316.5316.5316.4616.3516.4216.6416.4416.3616.2816.2916.4116.3116.3116.5016.3815.3615.3215.2315.3115.2815.3114.7014.5414.5514.5214.7514.6914.6216.1616.1116.0716.18
15.4315.5215.4715.3115.6115.70
15.1514.8214.7415.8116.0015.60
15.9114.93
14.5416.1516.9616.3515.9815.9016.0315.7615.7215.8615.6216.0815.5215.4716.3314.41
FeO
9.9010.20
10.429.19
10.2110.368.708.658.688.878.959.018.878.948.829.108.719.248.858.888.998.948.898.768.92
10.5611.3311.3811.4911.4711.5512.7512.5612.7912.8412.9212.9712.879.569.449.529.45
10.1810.2210.2511.4810.0410.15
11.6112.4011.349.299.249.00
9.2211.55
11.488.618.308.798.579.179.118.908.798.948.738.938.94
10.088.59
11.73
MnO
0.160.18
0.180.170.160.180.130.120.180.170.160.180.160.160.170.160.170.160.150.150.190.130.180.120.160.160.180.190.210.190.210.250.270.280.250.240.250.250.170.160.160.15
0.160.140.180.250.190.17
0.200.130.240.210.190.19
0.170.28
0.180.190.070.190.180.150.220.180.200.140.140.140.190.190.170.21
MgO
6.275.78
5.596.725.615.506.866.906.588.328.268.088.288.288.218.238.226.166.436.306.426.456.456.446.374.095.024.924.944.744.673.383.243.343.693.653.623.606.596.626.336.49
6.536.416.514.976.516.43
4.143.853.268.388.307.94
8.545.78
4.376.766.406.626.736.786.746.807.006.966.906.726.946.316.834.23
CaO
11.1210.57
9.9911.669.949.94
10.9110.7710.8511.7611.9211.9112.0312.0011.9111.8311.7510.9610.8810.9311.1011.1311.0110.9911.167.729.569.459.489.329.227.297.357.327.687.437.497.47
11.4411.3211.1411.32
11.1711.2511.269.51
11.2311.26
8.488.397.69
13.7213.5913.66
13.8510.17
8.8611.8612.4311.9611.8312.0612.1112.1812.4012.0212.5012.1012.3211.1811.959.05
Na2O
3.473.33
3.772.933.813.853.373.373.393.123.142.953.163.123.153.143.063.053.013.093.062.982.933.002.993.523.513.413.463.523.413.183.033.013.222.993.163.143.123.063.183.06
2.422.392.433.532.342.49
2.953.003.262.192.131.97
2.172.44
2.401.801.841.881.811.771.851.791.791.811.731.851.771.801.872.34
K2O
0.170.20
0.180.220.180.190.180.190.170.070.090.090.090.070.080.080.100.110.100.130.120.100.100.120.120.290.250.270.280.280.280.220.230.220.200.210.230.210.130.130.130.13
0.210.210.200.280.190.22
0.250.260.300.070.070.06
0.060.27
0.480.340.320.350.340.310.320.290.300.290.320.340.320.250.340.49
P 2 O 5
0.240.20
0.250.210.280.250.200.200.200.150.160.140.160.150.160.150.180.140.130.130.100.090.110.110.110.320.210.200.230.210.240.310.280.280.290.280.270.280.140.130.170.15
0.110.090.080.170.140.13
0.160.170.170.060.070.07
0.080.21
0.150.080.160.090.100.060.140.090.110.110.080.130.100.090.170.17
Total
99.9099.64
100.3999.56
100.0399.9799.84
100.2099.96
100.49100.62100.34100.87100.14100.44100.16100.3999.9499.8499.5299.8699.7799.6399.7099.8099.6199.0499.8099.4099.1699.8699.7799.77
100.1999.95
100.37100.0399.5099.9499.4999.3599.34
99.1499.12
100.0299.5299.2899.87
99.9099.92
100.13100.56100.4799.35
100.9899.90
99.1599.7399.9999.8999.5699.74
100.2899.84
100.13100.4299.6899.7999.5299.39
100.2799.03
S
1080740
1140540
11201140800830800
10201020101010101010101010401000620650580640680680660640420590580570570580740760780690770800780780780750740
107010801040450
10901080
260190240
mo970970
1020467
17011090
10090
110100100110110110110130210110260
Notes: Glass compositions generally correlate quite well with the units determined by shipboard studies (Parson, Hawkins, Allan, et al.,1992). The only petrographic units showing significant within-unit heterogeneities were Unit 6 and Subunit 10a at Site 834. Unit6 has been divided into Subunits 6a, 6b, and 6c and the single anomalous composition in Unit 834-10a is identified with an asterisk(*). Note also that the lowermost sample from Unit 834-12 and the uppermost sample from Unit 834-13 were mislabeled duringshipboard sampling and these have been transposed.
606
LAU BASIN BASALTS
Table 3. Fe^/XFLeg 135 glasses.
, MgO, FeO*, and S concentrations in
Site
834
836
839
Unit orsubunit
2a2a6a6c78
10a10a10a10a1212
1
33
Core,section
13X-114X-1AvgAvgAvgAvg39R-140R-141R-1Avg57R-1Avg3H-34H-CC19R-123R-1
FeO*
9.9010.2010.338.678.998.89
11.3311.4711.5511.4512.8712.8812.409.248.618.93
MgO
6.205.725.526.718.146.324.974.694.624.763.573.453.818.226.696.65
Fe 3 + lFe
0.1120.2230.1390.1540.1030.1710.2860.2780.2810.2820.1820.1770.2770.1310.2310.260
ΔNNO
-1.290.43
-0.77-0.41-1.37-0.12
1.020.961.071.040.01
-0.041.11
-0.750.761.04
S(ppm)
1080740
1140810
1010660570570580570780760190970110
no
Notes: Samples with a specific core designation can be identified bycomparison with Table 2. When glass from several samples had to becombined to provide enough material for titration, the result isindicated as an average (Avg). When glasses from different sampleswere combined, the results for S, FeO*, and MgO are given asaverage compositions.
1 .J
1.0
0.5
0.0
-0.5
-1.0
1 ^
-
MORE\
\-
1 1
1 1
A
A
— _
•
\
1 \ 1
• 834G 836A 839-
-
_
D
-
i
O
o
<
3 4 5 6 7 8 9
MgO (wt%)
Figure 2. ΔNNO vs. MgO in glass separates. Bold arrow labeled MORB is the
trend that MORB glasses appear to follow during fractionation (Juster et al.,
1989; Nilsson and Natland, 1991).
derived MORB sulfide-saturation line of Mathez (1980; Fig. 3).Basaltic glasses from Units 834-6a and 834-7 are the only glasseswith petrographic evidence for immiscible sulfide melts in the mag-mas. The glasses from Site 839 and Units 1 and 2 from Site 836 haveamong the lowest S ever observed in deep-sea glasses and approachthe very low values suggested for IAV magmas.
DISCUSSION
Because BABB have geochemical affinities with both MORB andIAV lavas, it is useful to consider our present understanding of/O2 andS behavior in both of these rock types. Oxygen fugacity of relativelyprimitive MORB (MgO > 7.5 wt%) is inferred from Fe3+/Fe2+ to beapproximately 2-3 log units below NNO (Christie et al., 1986). TheFe3+/Fe2+ ratio of basaltic melts increases with silicate fractionationuntil approximately 4.5 wt% MgO, at which point Fe-Ti oxide crystal-lization commences and consumes Fe3+, and both Fe3+/Fe2+ and /O 2
decrease (Juster et al., 1989; Nilsson and Natland, 1991). Becauseoxide precipitation appears to begin at/O 2 equal to that of the NNObuffer in MORB compositions, it has been suggested that MORBmagmas are restricted to/O2 less than NNO (Juster et al., 1989).
The /O 2 of IAV lavas is poorly constrained in part reflecting thepaucity of undegassed arc glass. Experimental data and two-oxideequilibrium studies suggest that/O2 in IAV lavas is two to five log unitshigher than that of MORB (Fudali, 1965; Gill, 1981, and referencestherein). It is unknown whether this relatively high/O2 reflects the IAVsource or processes associated with subaerial eruption.
Most of our understanding concerning S concentrations in oceanicrocks results from studies of MORB, again reflecting the scarcity ofundegassed glass in other environments. Sulfur concentrations inMORB glass are typically 800-1000 ppm and correlate positively withFe content (Mathez, 1976, 1980). Most MORB melts coexist withimmiscible sulfide liquids that occur as small blebs of Fe-sulfides in thebasaltic glass. Importantly, S concentration in basaltic melts is not anindependent parameter;/O2,/S2, Fe, and S content are linked (MacLean,1969; Connolly and Haughton, 1972; Haughton et al., 1974).
All IAV lavas that have been analyzed to date contain less than 200ppm S (Naldrett et al, 1978; Gill, 1981; Ueda and Sakai, 1984), how-ever, these analyses were made on subaerial crystalline material that islikely to have lost much of its S during eruptive degassing. A singledredge in the Mariana Island Arc recovered glassy IAV material at 1170m water depth (Garcia et al., 1979). These glasses also have low S
Q_
ZJGO
1
7 U D
-
-
Y
/ ^
A
1
Θ
A
i
• 834O 835"D 836A 839.
1»• -
-
D
i
1200
1000
800
600
400
200
06 7 8 9 10
Fe (wt%)Figure 3. S vs. Fe in the Leg 135 glasses. Also indicated is the MORB saturationline of Mathez (1980).
(30-100 ppm), but their vesicle content is not described and their highSiO2 (55%—61%) suggests that crystallization has been extensive. Theeffects of vesiculation and fractionation on S concentration in IAVlavas are unknown but are likely to be significant. Studies of meltinclusions in phenocrysts suggest a tremendous range in S concentra-tion from very low to nearly MORB-like (Anderson, 1974; Devine etal., 1984). Given the unconvincing available data, it is ambiguouswhether the low S measured in IAV lavas is a signature of the sourceor simply reflects processes associated with their petrogenesis.
Many of the BABB glasses in this study are significantly differentfrom MORB with respect to both/O2 and S; these differences mustreflect differences in their sources, their petrogenetic environments,or both. As S concentration is intimately related to/O2, the causes ofthe high and variable/O2 of these lavas will first be explored.
Oxygen Fugacity
Several authors have attempted to constrain fluxes of the alkaliand alkaline earth elements from the descending slab so as to quantifythe "subducted component" in IAV lavas (Tatsumi et al., 1986;
607
K.N. FARLEY
McCullogh and Gamble, 1991; Hawkesworth et al., 1991). Quantifi-cation of these models is extremely difficult because the compositionof the downgoing plate is variable and poorly constrained, the mantlewedge is variably depleted, and the effects of a volatile fluid onmelting and crystallization processes in the mantle wedge are onlypoorly known. Nonetheless, most authors agree that an alkali-rich,hydrous fluid is driven off of the descending slab and plays a role inarc and backarc petrogenesis (e.g., Perfit et al., 1980; Gill, 1981;Hawkins and Melchior, 1985; Sinton and Fryer, 1987; Hochstaedter,Gill, and Morris, 1990; McCullogh and Gamble, 1991; Hawkesworthet al., 1991). Although the oxidizing capability of such a fluid isunconstrained, it is likely to have the capacity to increase the/O2 ofthe depleted mantle. This is supported by the work Mattioli et al.(1980) who showed that lherzolite nodules with mineralogic andgeochemical evidence of metasomatism by a hydrous fluid havehigher/O2 than unmetasomatized nodules.
If the high-/O2 Lau Basin glasses reflect the incorporation of ahydrous fluid, there should be a relationship between/O2 and thosetrace elements present in such a fluid. Various trace element concen-trations and isotope ratios (data from Hergt and Hawkesworth, thisvolume) were plotted as a function of/O2 and least squares linearregressions were computed (examples in Fig. 4). If high/O2 reflectsthe incorporation of a hydrous fluid, the correlations should be strong-est for those elements with an affinity for such fluids.
Correlation coefficients (r2) range from 0 to 0.85 and are listed inorder of decreasing degree of correlation in Table 4. The correlationdata have been subdivided into three categories: strong correlations(r2 > 0.6), possible correlations (0.6 > r2 > 0.35), and no correlations(r2 < 0.35). As the r values incorporate the relative errors associatedwith analyses of the various elements and the magnitude of the totalvariation in concentration, small differences between r2 values are notas significant as the general order that is observed. The importantpoint is that for some elements,/O2 correlates well with concentration,whereas for other elements, no correlation is observed. In particular,concentrations of the large ion lithophile elements (LILE: e.g., Rb,Cs, Ba, and K) exhibit strong positive correlations with/O2 (Fig. 4A).
One should note that these calculations have not been corrected inany way for the effects of fractionation. This is not an ideal solutionbecause other studies have concluded that some of these lavas repre-sent moderately fractionated compositions (Parson, Hawkins, Allan,et al., 1992; Hergt and Nilsson, this volume; Ewart et al., this volume).Unfortunately, the effects of crystallization processes, on Fe^/XFecannot be accurately predicted (Fig. 3; Juster et al., 1989). Further-more, the variability in some trace element concentrations reflectsvariations beyond the effects of fractionation. Therefore, it is neces-sary to assume that the variability in both Fe^/LFe and some traceelement concentrations is dominantly a source signature and thesecondary effects of fractional crystallization have been relativelyminor. This assumption is supported by the observation that the mostmafic samples (MgO > 6 wt%), those that are least likely to have beenaffected by fractionation processes span nearly the entire range in/O2
space (circled points in Fig. 4).
When the elements and isotope ratios are listed by decreasing r2
values as determined above (Table 4), the order of the elements isremarkably similar to estimates of trace element affinity for a fluidphase as determined by Tatsumi et al. (1986) and McCullogh andGamble (1991). In an experimental study evaluating trace elementmobility during dehydration of serpentinite, Tatsumi et al. (1986)concluded that trace element affinity for such a fluid follows the orderRb > Ba > K > Sr > La > Y. Based on patterns observed on MORB-normalized trace element diagrams, McCullogh and Gamble (1991)suggested that trace element affinity for a slab-derived hydrous fluidfollows the order Rb > Ba > Pb > K > U > Th > Sr > La > Ce > Nb >Y. Those elements that correlate well with/O2 (particularly Rb, Ba,and K) are suggested to be the most fluid-mobile elements by bothmodels. The relatively poor r2 values for Pb, Th, and Sr are rather sur-prising because these elements have been suggested to be relatively
Table 4. Various trace elements and isotopic ratiosand the correlation coefficient (r ) between the con-centration of that element or the ratio and/02
Element/ratio
Strong correlations:
RbBaCsK2O87Sr/86Sr
Possible correlations:U207 p b /204 p b
WThPb
No correlations:TaNb208 p b / 204 p b
ZrCe206pb /204p b
LaHf1 4 3Nd/1 4 4NdSrY
Ba/LaCe/Pb
S
r2
0.850.820.740.650.65
0.580.480.470.450.37
0.210.150.150.070.060.050.030.030.020.000.00
0.740.52
0.80
Notes: Trace element and isotope data are from Hergt and
Hawkesworth (this volume). Note that the correlations of
fθ2 with S and Ce/Pb are in the opposite sense as the other
well-correlated elements.
mobile in a fluid phase (McCullogh and Gamble, 1991; Hawkesworthet al., 1991). The absence of a correlation between/O2 and Sr, despitethe good correlation between/O2 and 87Sr/86Sr, may reflect the second-ary effects of Plagioclase crystallization. According to one model(McCullogh and Gamble, 1991), Pb is a highly mobile element, yetonly a poor correlation with/O2 is observed (r2 = 0.37). Such discrep-ancies may reflect the model-dependent nature of fluid mobility stud-ies or differences in subduction/dehydration regimes in individualsubduction zones.
Because the effects of fractionation cannot be accounted for in ther2 calculations, it is desirable to observe correlations in elemental andisotopic ratios. Various trace element ratios, such as high Ba/La andlow Ce/Pb, are characteristic of IAV lavas and are thought to reflectthe incorporation of an alkali-rich (high Ba and Pb, low La and Ce)fluid in IAV petrogenesis. As these elements are highly incompatibleduring melting and crystallization processes, these ratios should re-main relatively constant and reflect their source even as their concen-trations change. Good correlations with/O2 for both Ba/La (positive;Fig. 4B) and Ce/Pb (negative) support the suggestion that the ele-mental correlations described above are not simply a fortuitous con-sequence of fractionation processes. Interestingly, the correlation ofCe/Pb with/O2 is stronger than that of either Ce or Pb alone.
Correlations with /O 2 also provide insight into the behavior ofradiogenic isotope ratios. The 87Sr/86Sr ratio correlates relatively wellwith/O2 (r2 = 0.65; Fig. 4C), whereas no correlation is observed forthe Nd isotopes (r2 < 0.1); this is consistent with the suggestion thatSr is carried by a fluid phase whereas Nd is not. Although there isgood coherency among the Pb isotope ratios themselves (Hergt andHawkesworth, this volume), the 207Pb/204Pb ratio shows a possiblecorrelation with /O2, whereas no correlation is observed for the208pb/204pb o r 206pb/204pb r a t i o s pe rhaps the "subducted component"has a stronger effect on 207Pb/204Pb than the other isotopes of Pb inthe Lau Basin mantle.
Although the suggestion of an oxidized, hydrous, LILE-rich fluidderived from a descending slab is not new, this is the first relationship
608
LAU BASIN BASALTS
0.7024-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
ΔNNO (log units)
834 π 836 839
Figure 4. Rb concentration (A), Ba/La (B), and 87Sr/86Sr (C) vs./O2 (ΔNNO)in Leg 135 glasses. Dashed lines are linear regressions through the data withcorrelation coefficients as indicated. Circled points indicate those glasses withMgO > 6 wt%. Note that these relatively high MgO glasses occur through-out and thus the correlations cannot result simply from the effects of frac-tional crystallization.
between/O2 and the elements likely to be carried by such a fluid thathas been observed in basaltic glass. If the order of r values is indica-tive of the composition of such a hydrous fluid in the Lau-Tonga sys-tem, then this component is characterized by high/O2, Rb, Ba, Cs, K,and 87Sr/86Sr, and by possible enrichments in U, Th, W, Pb, and207Pb/204Pb. Those elements that show no correlation with/O2 (r2 <0.35) are probably not introduced by this fluid phase; their variabilitymust reflect heterogeneities in the mantle and/or fractionation effects.
Sulfur
In this section, the low and variable S concentrations of these lavaswill be explored. In particular, (1) As these melts are below sulfidesaturation, does S behave as an incompatible element? (2) Doesmixing occur between high-S and low-S source components? and (3)Is the solubility of S decreased as a result of the elevated/O2? Doesthis lead to increased S-loss during submarine degassing?
/. Sulfur as an Incompatible Element
Most of the glasses do not coexist with a sulfide phase and thereforethe melts from which they formed are likely to have been undersatu-rated with respect to sulfide melts. If S is exhausted during partialmelting of the mantle source (e.g., Hamlyn et al., 1985; Hamlyn andKeays, 1986; Natland et al., 1991), then the resulting magmas will bebelow saturation. As long as sulfide-saturation is not reached duringfractional crystallization and S does not enter into any other crystal-lizing phases, S should behave as an incompatible element (i.e., itsconcentration in the melt should increase in proportion with other in-compatible elements). Using MgO as a measure of fractionation, thereis clearly no consistent increase in S concentration with decreasingMgO (Fig. 5). In fact, some of the least fractionated glasses have amongthe highest S concentrations (e.g., Units 834-7, 836-3, and 836-4) andvice versa (e.g., Units 834-10a, 836-1, 836-2, 839-1, and 839-9).
Furthermore, S concentrations are not correlated with those ofmost incompatible elements. In particular, there is no correlationbetween S and either the rare earth elements (REE) or the highfield-strength elements (HFSE; e.g., Nb, Ta, Zr, and Y). There is,however, a strong negative correlation between S content and theLILEs (e.g., Rb; Fig. 6A). Although a negative relationship betweentwo elements cannot indicate that they are both behaving incompat-ibly during crystallization, it does suggest that the concentration vari-ations are linked.
2. Mixing
Trace element and isotopic studies indicate that these lavas repre-sent mixing between at least three compositional end-members: twoasthenospheric components and at least one "arc-like" component(Hergt and Hawkesworth, this volume). It is therefore appealing totry to constrain the wide range of S contents by mixing of low-S(arc-like) and high-S (MORB-like) sources. Excellent negative cor-relations are observed between S and LILE concentrations (e.g., Rb;Fig. 6A) and many ratios involving these elements (e.g., Ba/La; Fig.6B). However, the particularly striking relationships are of the sortS/LILE vs. 1/LILE (Figs. 6C and 6D). A straight line on such adiagram is consistent with, although not conclusive evidence for,binary mixing between high-S/low-LILE and low-S/high-LILE com-ponents. The implied end-members from these diagrams are remark-ably similar to what have been suggested for MORB (-1000 ppm S,-0.05 wt% K2O, -0.03 ppm Rb, and -4 ppm Ba) and IAV (-50 ppmS, -0.35 wt% K2O, -7 ppm Rb, and -75 ppm Ba). Thus, it is possiblethat the wide range of both S and LILE concentrations represent
ECLQ_
m
1200
1000
800
600
400
200
0
• 834O 835π 836A 839
3 4 5 6 7 8 9
MgO (wt%)
Figure 5. S vs. MgO for Leg 135 glasses. Note that S does not increase withdecreasing MgO as would be expected if S behaves as an incompatible element.
609
K.N. FARLEY
7
6
? 5
CL 4Q.
X)QI 2
1
0
A'
a
~ A
-
-
I
A
I
•
1
1 1
-
A
a
200 400 600 800 1000 1200
Sulfur (ppm)
0.0005 0.0010 0.0015 0.0020
1/K
30
25
α 2 0
\ 15 -σ
m 10
5
0
I A
i
i
•
i
i
«
t
i
i
•f
i
i
α
200 400 600 800 1000 1200
Sulfur (ppm)
0.05 0.10 0.15 0.20 0.25
1/Bα
834 o 835 Ü 836 839
Figure 6. Various relationships between S and LILEs illustrating the general result that high LILE concentration iscoupled with low S and vice versa. A. S vs. Rb. B. S vs. Ba/La. C. S/K vs. I/K. D. S/Ba vs. 1/Ba.
mixing between depleted asthenosphere and a hydrous fluid intro-duced during subduction and dehydration of the slab.
To test this mixing hypothesis it is necessary to explore the massbalance constraints concerning the S and LILE budgets of the hypo-thetical components. It is generally agreed that in regions of IAV andBABB petrogenesis, the dominant mixing component is the depletedasthenosphere and that the subducted component acts as an LILE-richcontaminant. In the case of the LILEs, it is easy to envision how theaddition of a very small amount of trace element enriched fluid canmix with a large amount of trace element depleted mantle and have atremendous effect on the resulting composition. However, with re-spect to S, we must explain the opposite relationship; the arc compo-nent is apparently very low in S, yet it must mix with a large amountof a high-S component and generate low-S melts. To satisfy massbalance constraints, negative S concentrations are required in thesubducted component.
Another argument against the mixing hypothesis involves therelationships between S and other classes of trace elements whichhave distinctive abundances in MORB and IAV lavas. To explain thevery low S content of boninites, Hamlyn et al. (1985) and Hamlynand Keays (1986) argued that low S is a characteristic of second stagemelts in which the bulk of the S inventory was removed during aprevious partial melting event. This argument is often invoked toexplain the low HFSE concentrations in IAV and BABB lavas (e.g.,Gill, 1981; Hawkins et al., 1984; Hawkins and Melchior, 1985;McCullogh and Gamble, 1991). If previous depletion events controlboth S and HFSE, the relationship between S and HFSE should be atleast as strong as that between S and those elements introduced duringlater enrichment events. The lack of any correlation between S andthe HFSE, coupled with the excellent negative correlations betweenS and the LILEs, strongly suggest that S concentration is more closelyrelated to those processes that are responsible for the enrichment ofLILEs than for the depletion of HFSE. Although the excellent S-LILE
relationship could theoretically be consistent with binary mixing, boththe low S concentrations and the excellent S-LILE relationship musthave another origin.
3. Low Sulfur Solubility and Sulfur Loss During Degassing
The relationship between/O2 and S solubility will be reviewed sothat the significance of any relationship between S and/O2 for thesesamples can be determined. Katsura and Nagashima (1974) con-cluded that the solubility of S (with respect to silicate-vapor equilib-ria) in magmas of various compositions is strongly dependent on/O2
and reaches a minimum at/O2 of approximately NNO (their fig. 3).Furthermore, at/O2 lower than NNO, S is dissolved in the melt as S2"and behaves as observed in MORB. At/O2 greater than NNO, muchof the S dissolves as SO4" (Sakai et al., 1982; Carroll and Rutherford,1988; Nilsson and Peach, 1993) and the effects on basaltic magmaare essentially unexplored.
The term "solubility" in the petrologic literature refers to the S con-tent of the silicate melt in equilibrium with another S-rich phase (sulfideliquid, anhydrite, or a vapor phase). The Lau Basin magmas are, forthe most part, undersaturated with respect to a sulfide liquid, as evi-denced by the near absence of sulfide globules and the lack of a rela-tionship between S and Fe. In addition, there is no evidence that thesemagmas were saturated with anhydrite, which might result in a rela-tionship between S and Ca (Carroll and Rutherford, 1985,1987). If theLau Basin magmas were controlled by silicate-vapor systematics, theexperimental results of Katsura and Nagashima (1974) are applicableas these experiments were saturated only with an S-bearing vapor phase.
Significantly, a strong negative correlation (r2 = 0.80) is presentbetween S concentration and oxidation state in the Lau Basin glasses(Fig. 7), suggesting that the two parameters are intimately related. Sconcentration decreases steadily with increasing /O 2 until approxi-mately NNO + 1. These oxidized lavas are likely to be recording the
610
LAU BASIN BASALTS
minimum in S solubility, which occurs at approximately NNO (Kat-sura and Nagashima, 1974). In addition to the decreased solubility ofS in these oxidized melts, any degree of submarine degassing couldfurther decrease the magmatic S concentrations.
Previous studies of volatile behavior during submarine degassinghave concluded that at water depths greater than 200-500 m degas-sing is dominated by CO2; H2O and S are not lost in appreciablequantities at these pressures (e.g., Moore and Fabbi, 1971; Moore andSchilling, 1973; Dixon et al, 1991). However, the degassing behaviorof various volatile species from oxidized, volatile-rich lavas such asthose recovered on Leg 135 is essentially unknown. Insufficient dataare available on other volatile constituents in these lavas to make morethan intuitive inferences regarding degassing, but the high vesiclecontents (approaching 50 vol% in some glasses; Parson, Hawkins,Allan, et al., 1992) suggest that degassing processes might be quitedifferent in these backarc lavas than in MORB. Arelationship betweenS concentration and extent of degassing is supported by a crudenegative relationship between vesicles and average S concentrationfor a given geochemical unit (Fig. 8). Using vesicle contents of nearbywhole rocks as a proxy for degree of degassing is somewhat risky.Low vesicle contents can indicate both undegassed or highly degassedmagmas and vesicle content varies by as much as 20 vol% withinsome of the geochemical units (Parson, Hawkins, Allan, et al., 1992).The relatively low vesicle contents of the Site 839 lavas (given theirextremely low S) may suggest that they have already undergoneextensive degassing. More data on the behavior of other volatiles inthese lavas are crucial before quantitative conclusions concerningdegassing of these melts can be made.
In conclusion, the low S concentrations of the "arc-like" samples(particularly those from Site 839) probably reflect the effects of apetrogenetic process rather than a source characteristic. If LILEs areintroduced into the Lau Basin mantle by a hydrous fluid, the intro-duction of this fluid into the mantle may also increase the/O2. High/O 2 lowers the solubility of S, which could promote S loss duringmagmatic degassing. Thus, the low S concentrations in many of theseglasses reflect the higher/O2 of the supra-subduction zone environ-ment and not a low-S source.
Comparison to Other Backarc Glasses
Although a small amount of S data for various other BABB existsin the literature, they are not coupled with /O 2 data and thereforecannot be compared directly to the results of this study. Despite theuncertainties introduced by combining S data generated by variousanalytical methods (e.g., Dixon et al, 1991), the available data fromvarious backarc basins have been combined to ascertain whether thelow and variable S concentrations observed in the Leg 135 glassesare anomalous (Fig. 9). Although S concentrations from no single dataset suggest any correlation with Fe, the combined data indicate thatmost BABB magmas show a reasonable correlation between S andFe. Interestingly, the backarc samples always fall at lower S for agiven Fe content than do MORB glasses. This relationship is as wouldbe expected if additional Fe3+ acts simply as a dilutant with respect tothe Fe-S relationship of Mathez (1976, 1980). Simple calculationssuggest that the trend observed in Figure 9 is the trend that is predictedgiven a Fe^/XFe of approximately 0.2 (Nilsson, 1993). Althoughcareful n θ 2 studies of BABB glasses are rare, this Fe^/XFe isapproximately what has been observed and suggested (Hawkins andMelchior, 1985; Hochstaedter, Gill, Kusakabe, et al., 1990; Nilsson,1993). No explanation can be given for the very low S concentrationsfrom some of these localities, as the/O2 and degree of degassing atmost sites were not determined. It is possible that the low S in otherbackarc basins may be caused by low S solubility as a result of high/O 2 and that loss of S species from arc and backarc lavas duringdegassing may be a widespread phenomenon (even at considerablewater depths).
in
O
<
1.5
0.5 -
-0.5 -
-1.0
-1.5
. A Q
A
• 834D 836A 839
t
•
•0 200 400 600 800 1000 1200
Sulfur (ppm)Figure 7. Sulfur vs. ΔNNO.
Φ
rr
10
0 200 400 600 800 1000 1200
Average sulfur (ppm)
Figure 8. Average S concentration vs. range in vesicle content for each unit.Vesicle content (vol%) is the range determined by point counts of nearbysamples (Parson, Hawkins, Allan et al., 1992). Lines connect points at theextremes of vesicle contents for each unit and single points indicate that onlyone sample was counted from that unit. Fields enclose points from the samesite. Symbols as in previous figures.
CONCLUSIONS
The Leg 135 backarc glasses are characterized by relatively highand variable Fe3+/Fe2+ (and therefore /O2). A strong correlation be-tween/02 and the LILEs (and 87Sr/86Sr) suggests that a high/O2-highLILE component may be involved in the petrogenesis of these mag-mas. When the elements and ratios are ranked in order of decreasingdegree of correlation with /O2, the order is remarkably similar tosuggested trace element fluid mobility generated by "slab flux" models(e.g., Tatsumi et al., 1986; McCullogh and Gamble, 1991; Hawkes-worth et al., 1991). The high-/O2, high-LILE component is thereforesuggested to be a hydrous fluid produced during slab dehydration.
Sulfur concentration in the Leg 135 glasses is low and variableand is not coupled with FeO* as it is in MORB glasses. S does correlatenegatively with both /O 2 and the LILEs. The correlation is in the
611
K.N. FARLEY
EQ .
Ifur
1500-
1250-
1000-
750-
500-
250-
0
JA X
/ "
A °0
0Á
×
/
0
O
X
/
//
//0
°o 0
0
0
0×
0
A
•
0
ו
0
•
Scotia SeaLau BasinN Fiji BasinSumisu RiftMariana TroughMORB glasses
10 11 12 13
Fe (wt %)
Figure 9. S vs. Fe for other backarc basin data. MORB saturation line fromMathez (1980). Data sources are as follows: Scotia Sea = Muenow et al. (1980);Lau Basin = Davis et al. (1991); North Fiji Basin = Aggrey et al. (1988); SumisuRift = S from Hochstaedter, Gill, Kusakabe, et al. (1990) and FeO* fromHochstaedter, Gill, and Morris, (1990); Mariana Trough = S from Kusakabeet al. (1990) and FeO* from J. Hawkins (unpubl. data, 1988).
wrong sense to record incompatible behavior and does not extend toother trace elements and isotope ratios as it should if mixing wereinvolved. S solubility is strongly dependent on /O 2 and the low Sconcentrations reflect the high/O2 of these glasses.
The low-/O2 glasses (i.e., Unit 834-7) suggest very little influenceof an alkali-rich, high-/O2, hydrous fluid; their major and trace elementcharacteristics are much like MORB, S concentrations are high, andthey are saturated with a sulfide melt. Conversely, the high-/O2 glasses(i.e., those from Site 839) record the significant influence of sucha hydrous fluid. The LILEs are enriched, /O 2 is relatively high, Ssolubility is low and much of the S has been lost during degassing.
We must await more detailed studies of other BABB with coupledS and /O 2 data before determining if such a process is of globalsignificance in backarc basins and island arcs.
ACKNOWLEDGMENTS
Support for this work was provided by USSAC. The microprobework benefited from the assistance of Sarah Roeske and LinneaNorby at UCD and LDEO, respectively. Careful reviews by CherylPeach, Ken Farley, Ed Mathez, Mike Perfit, and an anonymousreviewer are gratefully acknowledged.
REFERENCES*
Aggrey, K.E., Muenow, D.W., and Sinton, J.M., 1988. Volatile abundances insubmarine glasses from the North Fiji and Lau back-arc basins. Geochim.Cosmochim. Acta, 52:2501-2506.
Anderson, A.T., 1974. Chlorine, sulfur, and water in magmas and oceans. Geol.Soc. Am. Bull, 85:1485-1492.
Carmichael, I.S.E., 1991. The redox states of basic and silicic magmas: areflection of their source region? Contrib. Mineral. Petrol, 106:129-141.
Carroll, M.R., and Rutherford, MJ., 1985. Sulfide and sulfate saturation inhydrous silicate melts. J. Geophys. Res., 90 (suppl.):C601-C612.
, 1987. The stability of igneous anhydrite: experimental results andimplications for sulfur behavior in the 1982 El Chichon trachyandesite andother evolved magmas. J. Petrol, 28:781-801.
-, 1988. Sulfur speciation in hydrous experimental glasses of varying
Abbreviations for names of organizations and publication titles in ODP reference listsfollow the style given in Chemical Abstracts Service Source Index (published byAmerican Chemical Society).
oxidation state: results from measured wavelength shifts of sulfur X-rays.Am. Mineral, 73:845-849.
Christie, D.M., Carmichael, I.S.E., andLangmuir, CH., 1986. Oxidation statesof mid-ocean ridge basalt glasses. Earth Planet. Sci. Lett., 79:397-411.
Connolly, J.W.D., and Haughton, D.R., 1972. The volume of sulfur in glass ofbasaltic composition formed under conditions of low oxidation potential.Am. Mineral, 57:1515-1517.
Czamanske, G.K., and Moore, J.G., 1977. Composition and phase chemistryof sulfide globules in basalt from the Mid-Atlantic Ridge rift valley near37°N lat. Geol Soc. Am. Bull, 88:587-599.
Davis, A.S., Clague, D.A., Schulz, M.S., and Hein, J.R., 1991. Low sulfurcontent in submarine lavas: an unreliable indicator of subaerial eruption.Geology, 19:750-753.
Devine, J.D., Sigurdsson, H., and Davis, A.N., 1984. Estimates of sulfur andchlorine yield to the atmosphere from volcanic eruptions and potentialclimatic effects. J. Geophys. Res., 89:6309-6325.
Dixon, J.E., Clague, D. A., and Stolper, E.M., 1991. Degassing history of water,sulfur, and carbon in submarine lavas from Kilauea volcano, Hawaii. J.Geol, 99:371-394.
Fryer, P., Taylor, B., Langmuir, CH., and Hochstaedter, A.G., 1990. Petrologyand geochemistry of lavas from the Sumisu and Torishima backarc rifts.Earth Planet. Sci. Lett, 100:161-178.
Fudali, R.F., 1965. Oxygen fugacities of basaltic and andesitic magmas.Geochim. Cosmochim. Acta, 29:1063-1075.
Garcia, M.O., Liu, N.W.K., and Muenow, D.W., 1979. Volatiles in submarinevolcanic rocks from the Mariana island arc and trough. Geochim. Cosmo-chim. Acta, 43:305-312.
Gill, J.B., 1981. Orogenic Andesites and Plate Tectonics: New York (Sprin-ger-Verlag).
Hamlyn, PR., and Keays, R.R., 1986. Sulfur saturation and second-stagemelts: application to the Bushveld platinum metal deposits. Econ. Geol,81:1431-1445.
Hamlyn, PR., Keays, R.R., Cameron, W.E., Crawford, A.J., and Waldron,H.M., 1985. Precious metals in magnesian low-Ti lavas: implications formetallogenesis and sulfur saturation in primary magmas. Geochim. Cos-mochim. Acta, 49:1797-1811.
Haughton, D.R., Roeder, PL., and Skinner, B.J., 1974. Solubility of sulfur inmafic magmas. Econ. Geol, 69:451^-67.
Hawkesworth, C.J., Hergt, J.M., Ellam, R.M., and McDermott, R, 1991.Element fluxes associated with subduction related magmatism. Philos.Trans. R. Soc. London A, 335:393^05.
Hawkins, J.W., Bloomer, S.H., Evans, CA., and Melchior, J.T., 1984. Evolu-tion of intra-oceanic arc-trench systems. Tectonophysics, 102:175-205.
Hawkins, J.W., Lonsdale, PR, Macdougall, J.D., and Volpe, A.M., 1990.Petrology of the axial ridge of the Mariana Trough back arc spreadingcenter. Earth Planet. Sci. Lett., 100:226-256.
Hawkins, J.W., and Melchior, J.T., 1985. Petrology of Mariana Trough andLau Basin basalts. J. Geophys. Res., 90:11431-11468.
Hochstaedter, A.G., Gill, J.B., Kusakabe, M., Newman, S., Pringle, M., Taylor,B., and Fryer, P., 1990. Volcanism in the Sumisu rift. I. Major element,volatile and stable isotope geochemistry. Earth Planet. Sci. Lett., 100:179—194.
Hochstaedter, A.G., Gill, J.B., and Morris, J.D., 1990. Volcanism in the SumisuRift. II. Subduction and non-subduction related components. Earth Planet.Sci. Lett., 100:195-209.
Huebner, J.S., and Sato, M., 1970. The oxygen fugacity-temperature relation-ships of manganese oxide and nickel oxide buffers. Am. Mineral, 55:934-952.
Juster, T C , Grove, T.L., and Perfit, M.R., 1989. Experimental constraints onthe generation of FeTi basalts, andesites, and rhyodacites at the GalapagosSpreading Center, 85°W and 95°W. /. Geophys. Res., 94:9251-9274.
Katsura, T, and Nagashima, S., 1974. Solubility of sulfur in some magmas at1 atmosphere. Geochim. Cosmochim. Acta, 38:517-531.
Kilinc, A., Carmichael, I.S.E., Rivers, M.L., and Sack, R.O., 1983. Theferric-ferrous ratio of natural silicate liquids equilibrated in air. Contrib.Mineral. Petrol, 83:136-140.
612
LAU BASIN BASALTS
Kusakabe, M., Mayeda, S., and Nakamura, E., 1990. S, O and Sr isotopesystematics of active vent materials from the Mariana backarc basinspreading axis at 18°N. Earth Planet. Sci. Lett., 100:275-282.
MacLean, W.H., 1969. Liquidus phase relations in the FeS-FeO-Fe3O4-SiO2
systems and their application in geology. Econ. Geol., 64:865-884.Mathez, E.A., 1976. Sulfur solubility and magmatic sulfides in submarine
basalt glass. J. Geophys. Res., 81:4269-4276., 1980. Sulfide relations in Hole 418A flows and sulfur contents of
glasses. In Donnelly, T., Francheteau, J., Bryan, W., Robinson, P., Flower,M., Salisbury, M., et al., Init. Repts. DSDP, 51, 52, 53 (Pt. 2): Washington(U.S. Govt. Printing Office), 1069-1085.
-, 1984. Influence of degassing on oxidation states of basaltic mag-mas. Nature, 310:371-375.
Mathez, E.A., and Yeats, R.S., 1976. Magmatic sulfides in basalt glass fromDSDP Hole 319A and Site 320, Nazca Plate. In Yeats, R.S., Hart, S.R., etal., Init. Repts. DSDP, 34: Washington (U.S. Govt. Printing Office),363-373.
Mattioli, G.S., Baker, M.B., Rutter, MJ., and Stolper, E.M., 1989. Uppermantle oxygen fugacity and its relationship to metasomatism. J. Geol.,97:521-536.
McCullogh, M.T., and Gamble, J.A., 1991. Geochemical and geodynamicalconstraints on subduction zone magmatism. Earth Planet. Sci. Lett.,102:358-374.
Moore, J.G., and Fabbi, B.P., 1971. An estimate of the juvenile sulfur contentof basalt. Contrib. Mineral. Petrol, 33:118-127.
Moore, J.G., and Schilling, J.-G., 1973. Vesicles, water, and sulfur inReykjanes Ridge basalts. Contrib. Mineral. Petrol, 41:105-118.
Muenow, D.W., Liu, N.W.K., Garcia, M.O., and Saunders, A.D., 1980. Vola-tiles in submarine volcanic rocks from the spreading axis of the East ScotiaSea backarc basin. Earth Planet. Sci. Lett., 47:272-278.
Naldrett, A.J., Goodwin, A.M., Fisher, T.L., and Ridler, R.G., 1978. The sulfurcontent of Archean volcanic rocks and a comparison with ocean floorbasalts. Can. J. Earth Sci., 15:715-728.
Natland, J.H., Meyer, P.S., Dick, H.J.B., and Bloomer, S.H., 1991. Magmaticoxides and sulfides in gabbroic rocks from Hole 735B and the laterdevelopment of the liquid line of descent. In Von Herzen, R.P., Robinson,RT., et al., Proc. ODP, Sci. Results, 118: College Station, TX (OceanDrilling Program), 75-112.
Nilsson, K., 1993. Oxidation state, sulfur speciation, and sulfur concentrationin basaltic magmas: examples from Hess Deep and the Lau Basin [Ph.D.dissert.]. Univ. of California, San Diego.
Nilsson, K., and Natland, J.H., 1991. Oxide crystallization and/O2 duringdifferentiation of abyssal tholeiites and gabbros at Hess Deep, E. Pacific.Eos, 72:526.
Nilsson, K., and Peach, C, 1993. Sulfur speciation, oxidation state, and sulfurconcentration in backarc magmas. Geochim. Cosmochim. Acta.
Parson, L., Hawkins, J., Allan, J., et al., 1992. Proc. ODP, Init. Repts., 135:College Station, TX (Ocean Drilling Program).
Perfit, M.R., and Fornari, D.J., 1983. Geochemical studies of abyssal lavasrecovered by DSRV Alvin from eastern Galapagos Rift, Inca Transform,and Ecuador Rift. 2. Phase chemistry and crystallization history. J. Geo-phys. Res., 88:10530-10550.
Perfit, M.R., Gust, D.A., Bence, A.E., Arculus, R.J., and Taylor, S.R., 1980.Chemical characteristics of island-arc basalts: implications for mantlesources. Chern. Geol, 30:227-256.
Perfit, M.R., Langmuir, CH., Balkisapa, M., Chappell, B., Johnson, R.W.,Staudigel, H., and Taylor, S.R., 1987. Geochemistry and petrology ofvolcanic rocks from the Woodlark Basin: addressing questions of ridgesubduction. In Taylor, B., and Exon, N.F. (Eds.), Marine Geology, Geo-physics, and Geochemistry of the Woodlark Basin—Solomon Islands.Circum-Pac. Counc. Energy Miner. Resour., Earth Sci. Ser., 113-154.
Sack, R.O., Carmichael, I.S.E., Rivers, M.L., and Ghiorso, M.S., 1980. Fer-ric-ferrous equilibria in natural silicate liquids at 1 bar. Contrib. Mineral.Petrol, 75:369-376.
Sakai, H., Casadevall, T.J., and Moore, J.G., 1982. Chemistry and isotoperatios of sulfur in basalts and volcanic gases at Kilauea volcano, Hawaii.Geochim. Cosmochim. Acta, 46:729—738.
Sato, M., 1978. Oxygen fugacity of basaltic magmas and the role of gasforming elements. Geophys. Res. Lett., 5:447-449.
Sinton, J.M., and Fryer, P., 1987. Mariana Trough lavas from 18°N: implica-tions for the origin of back arc basin basalts. J. Geophys. Res., 92:12782-12802.
Tatsumi, Y, Hamilton, D.L., andNesbitt, R.W., 1986. Chemical characteristicsof fluid phases from a subducted lithosphere and origin of arc magmas:evidence from high pressure experiments and natural rocks. J. VolcanolGeotherm. Res., 29:293-309.
Ueda, A., and Sakai, H., 1984. Sulfur isotope study of Quaternary volcanicrocks from the Japanese Islands Arc. Geochim. Cosmochim. Acta,48:1837-1848.
Volpe, A.M., Macdougall, J.D., and Hawkins, J.W., 1987. Mariana Troughbasalts (MTB): trace element and Sr-Nd isotopic evidence for mixing be-tween MORB-like and arc-like melts. Earth Planet. Sci. Lett., 82:241-254.
, 1988. Lau basin (LBB): trace element and Sr-Nd isotopic evidencefor heterogeneity in backarc basin mantle. Earth Planet. Sci. Lett., 90:174-186.
Wallace, P., and Carmichael, I.S.E., 1992. Sulfur in basaltic magmas. Geochim.Cosmochim. Acta, 56:1863-1874.
Wilson, A.D., 1960. The micro-determination of ferrous iron in silicate min-erals by a volumetric and a colorimetric method. Analyst, 85:823-827.
Date of initial receipt: 1 March 1993Date of acceptance: 29 June 1993Ms 135SR-145
613
