G3G3GeochemistryGeophysics

GeosystemsPublished by AGU and the Geochemical Society

AN ELECTRONIC JOURNAL OF THE EARTH SCIENCES

GeochemistryGeophysics

Geosystems

Article

Volume 11, Number 4

16 April 2010

Q04004, doi:10.1029/2009GC003009

ISSN: 1525!2027

ClickHere

for

FullArticle

Contamination of basaltic lava by seawater: Evidence foundin a lava pillar from Axial Seamount, Juan de Fuca Ridge

Peter Schiffman and Robert ZierenbergDepartment of Geology, University of California, Davis, California 95616, USA (pschiffman@ucdavis.edu)

William W. Chadwick Jr.Hatfield Marine Science Center, Oregon State University, NOAA, Newport, Oregon 97365, USA

David A. ClagueMonterey Bay Aquarium Research Institute, 7700 Sandholdt Road, Moss Landing, California 95039!0638, USA

Jacob LowensternU.S. Geological Survey, MS 910345, Middlefield Road, Menlo Park, California 94025, USA

[1] A lava pillar formed during the 1998 eruption at Axial Seamount exhibits compositional and texturalevidence for contamination by seawater under magmatic conditions. Glass immediately adjacent to anas-tomosing microfractures within 1 cm of the inner pillar wall is oxidized and significantly enriched in Naand Cl and depleted in Fe and K with respect to that in glassy selvages from the unaffected outer pillar wall.The affected glass contains up to 1 wt % Cl and is enriched by !2 wt % Na2O relative to unaffected glass,consistent with a nearly 1:1 (molar) incorporation of NaCl. Glass bordering the Cl!enriched glass in theinner pillar wall is depleted in Na but enriched in K. The presence of tiny (<10 mm) grains of Cu!Fe sulfidesand Fe sulfides as well as elemental Ni, Ag, and Au in the Na!depleted, K!enriched glass of the inner pillarwall implies significant reduction of this glass, presumably by hydrogen generated during seawater contam-ination and oxidation of lava adjacent to microfractures. We interpret the compositional anomalies we seein the glass of the interior pillar wall as caused by rapid incorporation of seawater into the still!molten lavaduring pillar growth, probably on the time scale of seconds to minutes. Only one of seven examined lavapillars shows this effect, and we interpret that seawater has to be trapped in contact with molten lava (insidethe lava pillar, in this case) to produce the effects we see. Thus, under the right conditions, seawater con-tamination of lavas during submarine eruptions is one means by which the oceanic crust can sequester Clduring its global flux cycle. However, since very few recent lava flows have been examined in similar detail,the global significance of this process in effecting Earth’s Cl budget remains uncertain.

Components: 6100 words, 7 figures, 1 table.

Keywords: lava pillar; seawater contamination; Axial Seamount.

Index Terms: 1034 Geochemistry: Hydrothermal systems (0450); 1039 Geochemistry: Alteration and weathering processes(3617); 8427 Volcanology: Subaqueous volcanism.

Received 5 January 2010; Revised 2 March 2010; Accepted 11 March 2010; Published 16 April 2010.

Schiffman, P., R. Zierenberg, W. W. Chadwick Jr., D. A. Clague, and J. Lowenstern (2010), Contamination of basaltic lava byseawater: Evidence found in a lava pillar from Axial Seamount, Juan de Fuca Ridge, Geochem. Geophys. Geosyst., 11,Q04004, doi:10.1029/2009GC003009.

Copyright 2010 by the American Geophysical Union 1 of 12

1. Introduction

[2] Seawater assimilation has been postulated as aprocess affecting the composition of MORBs andOIBs, especially the elevated Cl/K ratios found inanomalous MORBs [Michael and Schilling, 1989;Michael and Cornell, 1998]. A variety of potentialassimilants has long been suggested, includingCl!amphibole!bearing rock, a hydrous, Cl!rich vaporphase produced by extreme magmatic fractionation,an as yet undetected Cl!rich solid phase, and sea-water!derived brines [Michael and Schilling, 1989].Subsequent workers [e.g., le Roux et al., 2006;Coombs et al., 2004; Kent et al., 1999; Perfit et al.,1999; Jambon et al., 1995] have demonstrated thatMORB andOIBmagmas, as well as those from backarc basins [Kent et al., 2002], must have assimilatedsignificant amounts of a seawater!derived compo-nent on a regional scale. On the outcrop scale, va-porized seawater may be incorporated into submarinelava flows during their emplacement [Soule et al.,2006].

[3] The oceanic crust has long been recognized as amajor reservoir for Cl prior to its recycling throughthe subduction process [Anderson, 1974]. Cl isconcentrated in marine sedimentary pore fluids and,during hydrothermal alteration of basalt, seques-tered in some silicate minerals such as chlorite andespecially, amphibole [Ito et al., 1983]. Below wepresent petrographic and geochemical evidence thatCl may also be sequestered in the quenched glass ofsubmarine lava flows during their emplacement,under certain conditions.

[4] Lava pillars are hollow, basaltic pipes thatsupport the upper solidified surfaces of submarinelava ponds and serve as conduits for trapped sea-water to escape from beneath actively inflatingflows [Engels et al., 2003; Gregg and Chadwick,1996; Francheteau et al., 1979]. The outer wallsof lava pillars are covered with (the remnants of)lava shelves whose undersides are coated withlava drips or stalactites that imply vaporization ofinflowing seawater beneath the deflating, overly-ing crust [Soule et al., 2006; Perfit et al., 2003;Engels et al., 2003; Chadwick, 2003]. The presenceof pipe vesicles with associated high!temperaturesilicate mineral deposition on the walls of drips andadjacent septa imply significant high!temperature,vapor!magma interaction [Perfit et al., 2003].Mineralogical and geochemical evidence for local!scale contamination of the underside crusts of sub-marine sheet flows by a seawater!derived vaporunder magmatic conditions has previously beendocumented [Soule et al., 2006].

[5] Below, we present evidence for contamination,by a seawater!derived fluid, of an actively growinglava pillar formed during the 1998 eruption at AxialSeamount on the Juan de Fuca Ridge. The con-tamination apparently occurred while the inner wallof the lava pillar was brittle enough to micro-fracture, yet molten enough to sustain significantoxidation!reduction and compositional modifica-tion before quenching to a glass. This example oflava!seawater interaction has implications for theconditions under which such contamination occurs.

2. Sample Description and Petrography

[6] The January 1998 eruption at Axial Seamount[Embley et al., 1999] produced an extensive sheetflow with many lava pillars exposed in the collapsedinterior of the flow. Fortuitously, the expanding andinflating flow also uplifted a seafloor monitoringinstrument, deployed 4 months earlier, which sur-vived the eruption and precisely recorded the upliftof the lava flow surface as it inflated [Chadwick,2003]. Initially, the flow inflated nearly 1.7 m in5 min; following a 4 min pause, a further inflationof 1.2 m took place over the course of nearly anhour [Fox et al., 2001]. Drain back of the flow overthe next 1.5 h left behind numerous standing lavapillars that independently preserved a record offlow inflation. In September 2003, the ROVROPOScollected a fragment from a broken pillar (sampleR743!RK!0007), lying on the floor of a collapsearea within the 1998 flow. The fragment measured!30 cm in diameter, but the original height of thepillar and the fragment’s relative position fromwithin the pillar are both unknown.

[7] In cross section, the pillar sample has fourdistinct textural zones, which are radially sym-metric around its core (Figure 1). The outer pillarwall has a thin (<1 cm) black, glassy selvage,ornamented with closely spaced, cuspate lavashelves that record the continuous process of drainback following flow inflation [Chadwick, 2003;Gregg et al., 2000]. Moving inward, the next !3 cmof the pillar wall is composed of a gray, varioliticzone in which microcrystallinity coarsens inward,presumably in response to decreasing cooling rates,in a manner similar to that long!described frompillow lavas [e.g., Schiffman and Lofgren, 1982] andmore recently the crusts of submarine lava flows[Soule et al., 2006]. This zone grades into a !1 cmwide, variolitic zone containing slit!like pipevesicles, aligned radially relative to the long axis ofthe pillar. The innermost textural zone, also !1 cmin width, is bluish!gray glass, with a perlitic

GeochemistryGeophysicsGeosystems G3G3 SCHIFFMAN ET AL.: CONTAMINATION OF BASALTIC LAVA BY SEAWATER10.1029/2009GC003009

2 of 12

appearance, and ornamented with curvilinear micro-fractures. The surface of the inner pillar wall,directly beneath this last zone, has a tan!coloredskin (!1 mm). The features associated with theinner pillar wall’s contamination zone (i.e., perlitictexture and curvilinear microfractures) are observedalong the entire length of this pillar sample.

[8] Backscattered electron (BSE) imaging reveals amarked contrast between microtextures in the outer(Figure 2a) versus inner (Figure 2b) pillar walls.The latter is composed of aphyric, highly micro-vesiculated glass, unlike the selvage on the outerpillar wall, in which microphenocrysts of plagio-clase and clinopyroxene are present within sparselyvesiculated sideromelane. Within the inner pillarwall, irregularly shaped microvesicles (generally<50 mm) are concentrated near microfractures. Theglass and microvesicles are riddled with numerousmm!sized grains that appear extremely brightagainst the basaltic glass in BSE images (Figure 2b).Energy dispersive analyses of the largest of these

grains, which approach 5 mm in diameter, reveal thatthey include a wide variety of base and preciousmetals (i.e., Ni, Cu, Au, and Ag) primarily in theirelemental state (Figure 2c), as well as in Cu! andFe!bearing sulfides. However, the grains of ele-mental Ni, Ag, and Au are not intergrown withsulfides as would be expected if they were derivedfrom an immiscible sulfide liquid.

[9] The microfractures, which penetrate into thefirst cm of the inner pillar wall, have distinctive,!50 to 100 mm wide, subsymmetrical “haloes”surrounding them. In plane!polarized transmittedlight, the haloes are distinctly more reddened thanthe adjacent glass (Figure 3). Viewed in cross!polarized light, they are still isotropic. In BSEimages, the haloes have a mottled texture, appar-ently due to the presence of ubiquitous submicron!sized vesicles.

[10] X!ray dot mapping indicates that the haloes arestrongly enriched in Cl compared to the adjacent

Figure 1. (a) Photographic and (b) interpretive cross!sectional views through a portion of lava pillar sample R743!RK!007. Evidence for Cl enrichment occurs mainly within the first cm adjacent to the inner wall of the pillar. The“variolitic” zone becomes microcrystalline as it grades toward the inner pillow wall.

GeochemistryGeophysicsGeosystems G3G3 SCHIFFMAN ET AL.: CONTAMINATION OF BASALTIC LAVA BY SEAWATER10.1029/2009GC003009

3 of 12

Figure 2

GeochemistryGeophysicsGeosystems G3G3 SCHIFFMAN ET AL.: CONTAMINATION OF BASALTIC LAVA BY SEAWATER10.1029/2009GC003009

4 of 12

glass (Figure 4). X!ray diffraction analysis, per-formed on powders microdrilled from the regionsadjacent to these microfractures, does not reveal thepresence of any secondary minerals, consistent withthe isotropic nature of the affected glass and theapparent absence of hydrothermal alteration withinthe pillar.

3. Compositional Variations of LavaPillar Glasses

[11] Glasses from the lava pillar were analyzedwith a Cameca SX!100 electron microprobe oper-ated at 15 KeV, 10 nA beam current, and a 10 mmbeam diameter. The counting time on peak andbackground for Cl Ka was 30 s. The lower limitof detection for Cl under these conditions is!0.03 wt %. The calibration standard for Cl was

a well!characterized, natural biotite (R!2208) con-taining 1.2 wt % Cl. Using this standard, we repro-duce the reported Cl content of USNM R6600!1scapolite (1.43 wt %) to within 5%.

[12] We have recognized three distinct populationsof glass in the lava pillar sample (Table 1). Themost pristine of these is that found in the glassyselvage of the outer pillar wall that is composi-tionally similar to other MORB glasses reportedfrom Axial Seamount [e.g., Chadwick et al., 2005].Glass in the inner pillar walls is slightly enriched inK2O (by !0.05 wt %) and significantly depleted inNa2O (by !0.8 wt %) relative to glass in the outerpillar wall (Figure 5a). Glass within the haloesdirectly adjacent to microfractures of the innerpillar wall is significantly Na! and Cl!enrichedrelative to all other glasses. The maximum valuesrecorded in 32 individual spot analyses is 1.0 wt %

Figure 2. (a) Backscattered electron view of a portion of the outer wall of lava pillar sample R743!RK!007. Glasscontains abundant quench crystallites of plagioclase and clinopyroxene and relatively few rounded vesicles (!50–100 mm) and is generally devoid of fractures. (b) Backscattered electron view of a portion of the inner wall of lavapillar sample R743!RK!007. The outer portion of the zone of Cl enrichment is adjacent to the inner wall (on the rightside of image). Note anastomozing cracks propagating inward at right angles to the inner pillar wall. These crackshave thin (50–100 mm) Cl enrichment haloes adjacent to them. Also note the small (<50 mm), irregularly shapedvesicles which are mainly found within 1–2 mm of the inner pillar wall. The tiny (<10 mm) bright spots seen inmicrovesicles, as well as within glass, are grains of metallic Ni (with minor phosphorus), Cu, and rarer Au and Ag, aswell as base metal sulfides as determined by energy dispersive spectrometry. Unlike the outer pillar wall, the glass isdevoid of any quench crystallites. (c) Energy dispersive spectrum from a <5 mm diameter, Au!bearing grain such asseen in Figure 2b. The grain is essentially pure Au, and the Al, Si, Ca, and Fe peaks have been excited from the sur-rounding basaltic glass.

Figure 3. Plane polarized view of the glassy margin of the inner pillar wall of sample R743!RK!007. Note the dis-tinct reddening adjacent to microfractures. The width of the field of view is 2 mm.

GeochemistryGeophysicsGeosystems G3G3 SCHIFFMAN ET AL.: CONTAMINATION OF BASALTIC LAVA BY SEAWATER10.1029/2009GC003009

5 of 12

for Cl and 4.5 wt % for Na2O, and collectively, thedata conform closely to the trend expected if Naand Cl were added in a 1:1 molar basis (Figure 5b).The most Na2O!enriched of these glasses also ex-hibit significant Fe depletion (by up to !1.5 wt % asFeO) relative to the other pillar glasses (Figure 5c).Na, K, and Cl exhibit compositional gradients on theorder of 20–25 mm into inner pillar glass adjacentto the interface with the Na! and Cl!enriched haloes(Figure 6).

[13] Total dissolved water contents in doublypolished sections of pillar glasses were investigatedusing a Nicolet Magma 750 Fourier Transforminfrared spectrometer using techniques describedby Mastin et al. [2004]. Spot analyses 100 mm indiameter on both inner and outer pillar glassesaveraged 0.23 wt %. These results are in accordwith the oxide totals from the electron microprobeanalyses of the various pillar glasses (Table 1) thatimply that the total dissolved volatiles in the glassesare on the order of 1 wt % or less.

4. Discussion

[14] Several lines of textural and geochemical evi-dence lead to us to conclude that the lava adjacentto the inner wall of lava pillar sample R743!RK!0007 was contaminated by a seawater!derived fluidat magmatic temperatures. Within the inner pillarwall, the textural evidence includes the (1) densityand irregular shape of microfractures and micro-vesicles, (2) oxidation/reddening of glass withinNaCl!enriched haloes, (3) presence of mm!sizedgrains of elemental metals and sulfides, and (4) ab-

sence of quench crystallites. None of these featuresare observed in the outer pillar wall.

[15] The size, shape, and density of vesicles inglass adjacent to haloes of NaCl enrichment aredistinct from those in quenched glass on the outerpillar walls (Figure 2a versus Figure 2b), the latterwhich may be mainly derived from CO2 degassing[Chadwick et al., 2005]. In the inner pillar wall,vesicles are generally an order of magnitudesmaller (5–10 mm versus 50–100 mm) and lessspherical. The irregular shape of most vesicles mayreflect bubble coalescing that was in process as theglass quenched. Similar!appearing irregularlyshaped vesicles have also been described from thelower surfaces of lobate crusts on submarine sheetflows that have interacted with a seawater!derivedvapor phase [Perfit et al., 2003]. Collectively, thesize, shape, and distribution of these vesicles pre-sumably reflect the active and rapid interactionbetween the lava and the contaminating fluid. Thecurvilinear, anastomozing microfractures that strad-dle the haloes of NaCl enrichment appear to bespatially and texturally related to these micro-vesicles, and may have been initiated through rapidcoalescing of bubbles. Such microfractures are notfound along the inner walls of pillars that do notexhibit contamination chemistries.

[16] The observed reddening of NaCl!enrichedglass (Figure 3) presumably reflects the rapid oxi-dation of ferrous iron in basaltic melt in response toseawater contamination. This process would gen-erate H2 through the dissociation of H2O, via amechanism similar to that attending serpentiniza-tion of ferrous iron!bearing olivine in peridotites

Figure 4. Cl!K alpha X!ray dot map of a portion of the inner wall of lava pillar sample R743!RK!007. The edge ofthe inner pillar wall is approximately 1 mm from the bottom of the image, and the edge of the outer pillar wall islocated about 4 cm from the top of the image. The brightness denotes concentration of Cl: the haloes adjacent tothe fractures contain as much as 1 wt % Cl as indicated by wavelength dispersive electron microprobe spectrometry.The adjacent glass contains !400 ppm Cl on average.

GeochemistryGeophysicsGeosystems G3G3 SCHIFFMAN ET AL.: CONTAMINATION OF BASALTIC LAVA BY SEAWATER10.1029/2009GC003009

6 of 12

[Charlou et al., 2002]. We envision that the hydro-gen produced in the haloes of seawater contamina-tion would have rapidly diffused into the adjacentmelt, reducing metals like Au, Ag, and Ni, some to

an elemental state in which they precipitated asmm!sized grains.

[17] The striking absence of quench crystallitesin the glass of the inner pillar walls presumablyindicates that cooling rates were higher than thosein the outer pillar walls [Gregg et al., 2000], thatthe contamination process locally lowered themelting point of the basalt [Engels et al., 2003],and/or that the process of seawater contaminationsomehow suppressed nucleation and/or growth ofcrystallites along the inner pillar wall. The firstscenario seems unlikely since the contaminationprocess presumably entailed interaction of meltwith heated versus cold seawater. Nucleation andgrowth rates in basaltic melts are apparently affectedby sonic and ultrasonic frequencies [Bartels andFurman, 2002]. Shock waves must have beengenerated during the microfracturing that promotedseawater contamination within the inner pillarwalls, potentially suppressing growth of quenchcrystallites.

[18] Compositional variations of glasses in theinner wall of the lava pillar also are consistent withcontamination by NaCl!enriched fluid. The Na andCl contents of the haloes adjacent to microfracturesare elevated (with respect to “normal” glass in theouter pillar walls) on a nearly 1:1 molar basis(Figure 5b). Na enrichment in the haloes is stronglycorrelated with K depletion (Figure 5a). Potassiumand iron, the latter depleted by up to 1.5 wt % asFeO in the most Na! (and Cl!) enriched glasses(Figure 5c), were apparently leached from thehaloes, possibly by Cl!rich hydrothermal fluids.Although this “leaching” of K and Fe may havebeen mediated by hydrothermal fluids, there is nopetrographic or mineralogic evidence that thesefluids hydrated the glass or precipitated secondaryminerals within it.

Figure 5. (a) Variations of wt % K2O versus Na2O inglasses from lava pillar sample R743!RK!007 as deter-mined by wavelength dispersive electron microprobeanalyses. Note that in glasses from the inner pillar walls,Na enrichment strongly correlates with K depletion (andvice versa). (b) Variations of wt % Na versus Cl inglasses from lava pillar sample R743!RK!007. Note thatthe Na! and Cl!enriched glasses lie close to the calculatedline of 1:1 (molar) Na:Cl addition to the “normal” glassesfound within the outer pillar walls. (c) Variations of wt %FeO versus Na2O in glasses from lava pillar sampleR743!RK!007. Note that Na (and, but not shown, Cl)addition to glasses in the inner pillar wall appears to cor-relate well with Fe depletion.

GeochemistryGeophysicsGeosystems G3G3 SCHIFFMAN ET AL.: CONTAMINATION OF BASALTIC LAVA BY SEAWATER10.1029/2009GC003009

7 of 12

[19] Preserved concentration gradients for Na, K,and Cl directly adjacent to the enriched haloes areroughly consistent with the diffusion of these ionsunder magmatic conditions. Assuming diffusionrates for Na in basaltic melts of !10"6 cm2/s[Henderson et al., 1985], the observed, quenched!in concentration gradients, which are on the scaleof a few tens of mm’s (Figure 6), apparently reflectcooling through the solidus on a time scale of afew minutes or less.

[20] The maximum Cl content (i.e., 1 wt %) of theenriched haloes adjacent to microfractures is con-sistent with experimental data on Cl solubility inbasaltic melts [Webster et al., 1999] as well as withthe H2O contents in the contaminated glasses. At adepth of 1523 m where this lava pillar formed (i.e., afluid pressure of !15 MPa), the maximum solubilityof Cl is approximately 2.05 wt % [Malinin et al.,1989], but will be lower with increased H2O andCO2 content. At 15 MPa, H2O saturation in basalticmelts will be !1.2 wt % [Coombs et al., 2004], andwill also decrease with CO2 and Cl content. Thisvalue for maximumH2O solubility is consistent withthe high anhydrous oxide totals (!99.0–99.5 wt %)we measured by electron microprobe for all theglasses in the lava pillar (Table 1) as well with thewater contents of 0.23 wt % as determined by FTIR.Previously reported total H2O values for basalticglasses from Axial Seamount range from 0.14 to0.30 wt % [Chadwick et al., 2005; Dixon et al.,1988].

[21] The evidence for seawater contamination atmagmatic temperatures into the inner wall of this

lava pillar appears strong, but the mechanism forhow this occurred is less certain. The ubiquitouspresence of lava selvages and associated lava drips,observed on the outer walls of lava pillars, has beenused to argue for the presence of a seawater!derivedvapor beneath the upper crust of submarine sheetflows, either by ingress of cold seawater fromabove the flow [Chadwick, 2003] or by trapped,heated seawater from beneath it [Perfit et al., 2003;Engels et al., 2003]. Glass within the outer cm ofthe upper crust of lava pillars from the East PacificRise (EPR) is enriched in Cl up to 0.05 wt % locally,and the undersurfaces of the crusts of adjacent sheetflows are coated with minerals precipitated fromvaporized seawater [Soule et al., 2006]. However,to date, no one has reported Cl contamination inconcentrations comparable to this study.

[22] Previous models [e.g., Gregg and Chadwick,1996] for lava pillar formation have suggestedthat the hollow interior of these structures serve asconduits for the upflow of seawater trapped beneathan advancing flow, thereby advectively cooling thepillar from the inside out and prohibiting its thermalerosion from the surrounding and inflating flow.Estimated cooling times for quenching a !1 cmthick, glassy inner pillar wall are on the order of 102 s[Gregg et al., 2000], thus placing order ofmagnitudeconstraints on the time frame for seawater contam-ination into the Axial Seamount pillar.

[23] Six other lava pillars we have examined (fromthe 1998 flow at Axial Seamount and an older flowjust to the west, from the southern Cleft segment onthe Juan de Fuca Ridge, and from 31°S on the

Figure 6. Concentrations of Na2O, K2O, and Cl in glass within and directly adjacent to Na! and Cl!enriched glassfound in the inner pillar walls of sample R743!RK!007. The size of data symbols is roughly proportional to analyticalerrors.

GeochemistryGeophysicsGeosystems G3G3 SCHIFFMAN ET AL.: CONTAMINATION OF BASALTIC LAVA BY SEAWATER10.1029/2009GC003009

8 of 12

EPR) do not display similar enrichment features, sowe cannot speculate how widespread this previ-ously unrecognized process of seawater contami-nation might be. However, the evidence aboveseems to physically require that for this particularlava pillar at least, vaporized seawater had to be inextended contact with molten lava along the innerwalls of the pillar as it was still growing upward.Normally, seawater circulation through the pillarpipe quickly quenches exposed molten lava duringpillar growth, which suggests that in this case theseawater circulation had to be greatly restricted oreven completely blocked. There are two ways thiscould happen in an inflating pillar, either by lavabreaching the pillar walls and blocking the pipe orby a lobe of lava flowing over the top of the pillarand restricting or sealing the outlet.

[24] Both of these processes occur in inflating sheetflows. Gregg et al. [2000] reported the occurrenceof an 8 cm long lava drip on the inner wall of a lavapillar sample from a 1991 EPR flow, which impliedthat the previously solidified wall had been breachedby molten lava that oozed into the pillar pipe fromthe interior of the adjacent inflating sheet flow.Presumably, such a breach would occur near thevery top of a pillar, where the inner wall would bethinnest and most vulnerable to penetration. Lavaflowing over the top of a pillar would be the sub-marine equivalent of a breakout on an inflating,subaerial pahoehoe flow which are commonlyobserved to subsequently overrun adjacent flowlobes [Hon et al., 1994]. In either case, the pillartop and bottom would remain attached to the upperand lower crusts of the lava flow, and therefore thepillar would continue to grow upward, at least for atime. This would continue to expose molten lavaalong spreading cracks in the pillar walls andbecause the seawater is now trapped in the pillarconduit, it would quickly heat up and vaporize(Figure 7). The seawater would be driven into thenewly formed walls of the pillar along micro-fractures initiated by seawater vaporization. The ob-served, 100 mm maximum width of NaCl!enrichedhaloes adjacent to microfractures is consistent withexperimentally determined diffusivities (of !10"7to 10"6 cm2/s) for Cl in silicate melts at nearliquidus temperatures [Alletti et al., 2007], if thecontamination occurs over time scales of !102 s asimplied by the thickness of quenched glass on theinner wall of lava pillars.

[25] How long this process could continue woulddepend on whether the circulation of seawater insidethe pillar was only restricted or whether it wascompletely closed off. If it was merely restricted,T

able

1.Com

positio

nsof

Glasses

From

LavaPillarSampleR74

3!RK!007

a

Cl!EnrichedGlass

inInnerWall

Other

Glass

inInnerWall

Glass

inOuter

Wall

High!Cl

Axial

MORBc

Range

of32

Points

MeanTof

31Po

intsb

Representative

Analysis

Range

of13

Points

Meanof

13Po

ints

Representative

Analysis

Range

of9Po

ints

Meanof

9Po

ints

Representative

Analysis

SiO2

47.7–49.7

48.98(0.56)

49.71

49.1–5

0.5

49.86(0.39)

49.72

48.6–49.7

49.30(0.37)

49.48

49.96

Al 2O3

13.7–14.7

14.33(0.25)

14.36

14.4–1

5.5

14.77(0.36)

14.52

13.7–14.8

14.63(0.33)

14.49

14.51

CaO

11.5–13.6

12.25(0.53)

11.76

11.6–1

2.4

12.03(0.24)

12.28

11.6–12.4

11.92(0.19)

12.03

11.94

FeO

9.4–

10.9

10.19(0.43)

9.83

10.4–1

1.1

10.84(0.21)

10.66

10.8–12.0

10.90(0.36)

10.94

10.48

MgO

6.8–

7.7

7.16

(0.21)

6.89

7.1–7.5

7.27

(0.08)

7.26

6.9–

7.5

7.13

(0.19)

6.98

7.13

Na 2O

3.0–

4.5

3.84

(0.36)

4.48

1.6–2.8

2.06

(0.40)

2.77

2.7–

3.0

2.87

(0.06)

2.95

3.06

TiO

21.4–

1.7

1.53

(0.08)

1.51

1.4–1.6

1.55

(0.06)

1.58

1.4–

1.6

1.54

(0.07)

1.61

1.46

K2O

0.05

–0.14

0.09

(0.02)

0.08

0.17–0

.27

0.22

(0.03)

0.22

0.15

–0.20

0.17

(0.01)

0.17

0.20

MnO

0.11

–0.29

0.19

(0.04)

0.17

0.16–0

.27

0.20

(0.03)

0.20

0.14

–0.25

0.19

(0.03)

0.25

0.20

P 2O5

0.19

–0.26

0.22

(0.02)

0.21

0.21–0

.29

0.24

(0.02)

0.27

0.21

–0.31

0.24

(0.03)

0.22

0.16

Cl

0.28

–1.0

0.65

(0.02)

0.95

0.02–0

.06

0.04

(0.01)

0.04

0.02

–0.05

0.03

(0.01)

0.03

0.04

S0.03

–0.21

0.11

(0.10)

0.07

0.07–0

.13

0.10

(0.02)

0.12

0.08

–0.11

0.10

(0.02)

0.10

0.13

Cl/K

8.67

13.57

0.22

0.33

0.21

0.30

0.31

Total

99.54

100.02

99.18

99.64

99.02

99.25

99.27

a Unitiswt%.

b Stand

arddeviations

ofmeanin

parentheses.

c Axial

analysisissampleR49

7!20

from

Cha

dwicket

al.[200

5].

GeochemistryGeophysicsGeosystems G3G3 SCHIFFMAN ET AL.: CONTAMINATION OF BASALTIC LAVA BY SEAWATER10.1029/2009GC003009

9 of 12

Figure 7. Model for rapid seawater contamination within the inner pillar wall of sample R743!RK!007. In intervalT1, pillar grows as flow inflates; in interval T2, pillar top is overrun by adjacent flow lobe (or lava from breached pillarwall); in interval T3, trapped seawater within isolated pillar is vaporized and assimilated within microfractures; and ininterval T4, after flow deflation, pillar with assimilated seawater should be shorter than adjacent pillars. The timebetween intervals T2 and T3 is probably only a few seconds.

GeochemistryGeophysicsGeosystems G3G3 SCHIFFMAN ET AL.: CONTAMINATION OF BASALTIC LAVA BY SEAWATER10.1029/2009GC003009

10 of 12

one could imagine a balance in which brief sea-water vaporization was followed by enough slowcooling that the pillar would continue to growupward throughout the full inflation of the lavaflow. On the other hand, if the seawater circulationwere completely blocked, then the cooling effectinside the pillar would be greatly reduced and thepillar would be vulnerable to remelting and perhapseven becoming detached from the upper crust ofthe lava flow (Figure 7). Since pillar sample R743!RK!0007 is a fragment that was collected from amound of broken, fallen pillars, we cannot knowwhich of these scenarios might apply to this par-ticular pillar. If our model is correct, then lavapillars with this kind of seawater contaminationwould be difficult to recognize unless they werebroken apart and the interior of the pillar pipe werevisible.

[26] How important is this process of seawatercontamination as a means of sequestering Cl in aglobal context? Sequestration of Cl in lava flowsduring their emplacement is undoubtedly not nearlyas globally significant as that resulting from brineformation at deeper levels in the oceanic crust (e.g.,as described by Coombs et al. [2004]). But very fewrecent lava flows have been examined in similarpetrographic detail. Inflated submarine sheet flows,with attendant lava pillars, are a significant com-ponent of all mid!ocean ridges, especially at fastspreading centers. Our results show that whereseawater is trapped in contact with molten lava andcan vaporize, Cl assimilation can occur and thisprocess potentially contributes a nontrivial compo-nent to the Cl reservoir in the oceanic crust. Pre-sumably these assimilated glasses alter rapidly topalagonite or crystalline smectite during any sub-sequent diagenesis or low!temperature alteration.Whether these alteration products structurally retainthis assimilated Cl, or conversely, release it toaltering pore fluids would ultimately determinehow significant the above!described contaminationprocess is in contributing to the global Cl cycle.

5. Conclusions

[27] Textural and compositional evidence implythat the inner wall of a lava pillar from the 1998lava flow at Axial Seamount was contaminated bya seawater!derived fluid while it was either stillmolten or at temperatures within the glass transi-tion interval. Circulation of seawater inside thehollow interior of this lava pillar must have beengreatly restricted or completely blocked, either bylava entering the pillar pipe or overriding the pillar

top, thus allowing the seawater to vaporize. Up to1.0 wt % Cl has been incorporated into glass adja-cent to fractures along the inner pillar wall. Theresults of this study indicate that under someconditions, mafic lavas can be locally contami-nated by significant concentrations of Cl duringsubmarine eruptions. Admittedly, the contamina-tion processes described in this study do not havesignificant implications for global processes ofMORB petrogenesis. Rather, they provide a possi-ble mechanism for local Cl enrichment within theupper oceanic crust, and thus potentially a source forCl!enriched secondary minerals (e.g., amphiboles)which form during subsequent metamorphism ofthe oceanic crust.

Acknowledgments

[28] The NOAA Vents Program has supported the NeMO sea-floor observatory at Axial Seamount, including the 2003 expe-dition that collected the lava pillar sample described in thispaper. Thanks to the captain and crew of the R/V Thompsonand the ROPOS ROV team for their support at sea. This isNOAA/PMEL contribution 3326. Special thanks to Mike Perfitand the anonymous reviewers who critically read and madeexcellent suggestions for improving this manuscript.

References

Alletti, M., D. R. Baker, and C. Freda (2007), Halogen diffu-sion in basaltic melt, Geochim. Cosmochim. Acta, 71, 3570–3580, doi:10.1016/j.gca.2007.04.018.

Anderson, A. T. (1974), Chlorine, sulfur, and water in magmaand oceans, Geol. Soc. Am. Bull., 85, 1485–1492.

Bartels, K. S., and T. Furman (2002), Effect of sonic and ultra-sonic frequencies on the crystallization of basalt, Am. Mineral.,87, 217–226.

Chadwick, J., M. Perfit, W. I. Ridley, W. Kamenov,W. Chadwick Jr., P. le Roux, and M. Smith (2005), Effectsof the Cobb hot spot on the Juan de Fuca Ridge, J. Geophys.Res., 110, B03101, doi:10.1029/2003JB002767.

Chadwick, W. C., Jr. (2003), Quantitative constraints on thegrowth of submarine lava pillars from amonitoring instrumentthat was caught in a lava flow, J. Geophys. Res., 108(B11),2534, doi:10.1029/2003JB002422.

Charlou, J. L., J. P. Donval, Y. Fouquet, P. Jean!Baptiste, andN. Holm (2002), Geochemistry of high H2 and CH4 ventfluids issuing from ultramafic rocks at the Rainbow hydro-thermal field, Chem. Geol., 191, 345–359, doi:10.1016/S0009-2541(02)00134-1.

Coombs, M. L., T. W. Sisson, and J.!I. Kimura (2004), Ultra!high chlorine in submarine Kilauea glasses: Evidence fordirect assimilation of brine by magma, Earth Planet. Sci.Lett., 217, 297–313, doi:10.1016/S0012-821X(03)00631-9.

Dixon, J. E., E. Stolper, and J. R. Delaney (1988), Infraredspectroscopic measurements of CO2 and H2O in Juan deFuca basaltic glasses, Earth Planet. Sci. Lett., 90, 87–104,doi:10.1016/0012-821X(88)90114-8.

GeochemistryGeophysicsGeosystems G3G3 SCHIFFMAN ET AL.: CONTAMINATION OF BASALTIC LAVA BY SEAWATER10.1029/2009GC003009

11 of 12

Embley, R. W., W. W. Chadwick Jr., D. Clague, and D. Stakes(1999), The 1998 Eruption of Axial Volcano: Multibeamanomalies and seafloor observations, Geophys. Res. Lett.,26(23), 3425–3428, doi:10.1029/1999GL002328.

Engels, J. L., M. H. Edwards, D. J. Fornari, M. Perfit, and J. R.Cann (2003), A new model for submarine volcanic collapseformation, Geochem. Geophys. Geosyst., 4(9), 1077,doi:10.1029/2002GC000483.

Fox, C. G., W. W. Chadwick Jr., and R. W. Embley (2001),Direct observation of a submarine volcanic eruption from asea!floor instrument caught in a lava flow, Nature, 412,727–729, doi:10.1038/35089066.

Francheteau, J., T. Juteau, and C. Rangan (1979), Basalticpillars in collapsed lava!pools on the deep ocean!floor, Nature,281, 209–211, doi:10.1038/281209a0.

Gregg, T. K. P., and W. W. Chadwick Jr. (1996), Submarinelava!flow inflation: A model for the formation of lava pillars,Geology, 24, 981–984, doi:10.1130/0091-7613(1996)024<0981:SLFIAM>2.3.CO;2.

Gregg, T. K. P., D. J. Fornari, M. R. Perfit, W. I. Ridley, andM. D. Kurz (2000), Using submarine lava pillars to recordmid!ocean ridge eruption dynamics, Earth Planet. Sci. Lett.,178, 195–214, doi:10.1016/S0012-821X(00)00085-6.

Henderson, P., J. Nolan, G. C. Cunningham, and R. K. Lowry(1985), Structural controls and mechanism of diffusion innatural silicate melts, Contrib. Mineral. Petrol., 89, 263–272, doi:10.1007/BF00379459.

Hon, K., J. Kauahikaua, R. Denlinger, and K. Mackay (1994),Emplacement and inflation of pahoehoe sheet flows: Obser-vations and measurements of active lava flows on Kilaueavolcano, Hawaii, Geol. Soc. Am. Bull., 106, 351–370,doi:10.1130/0016-7606(1994)106<0351:EAIOPS>2.3.CO;2.

Ito, E., D. M. Harris, and A. T. Anderson Jr. (1983), Alter-ation of oceanic crust and geological cycling of chlorineand water, Geochim. Cosmochim. Acta, 47, 1613–1623,doi:10.1016/0016-7037(83)90188-6.

Jambon, A., B. Deruelle, G. Dreibus, and F. Pineau (1995),Chlorine and bromine abundance in MORB: The contrastingbehaviour of the Mid Atlantic Ridge and the East PacificRise and implications for the chlorine geodynamic cycle,Chem.Geol., 126, 101–117, doi:10.1016/0009-2541(95)00112-4.

Kent, A. J. R., D. A. Clague, M. Honda, E. M. Stolper, I. D.Hutcheon, and M. D. Norman (1999), Widespread assimila-tion of a seawater!derived component at Lohihi Seamount,Hawaii, Geochim. Cosmochim. Acta, 63, 2749–2761,doi:10.1016/S0016-7037(99)00215-X.

Kent, A. J. R., D. W. Peate, S. Newman, E. M. Stolper, andJ. A. Pearce (2002), Chlorine in submarine glasses from theLau Basin: Seawater contamination and constraints on thecomposition of slab!derived fluids, Earth Planet. Sci. Lett.,202, 361–377, doi:10.1016/S0012-821X(02)00786-0.

le Roux, P. J., S. B. Shirey, E. H. Hauri, M. R. Perfit, and J. F.Bender (2006), The effects of variable sources, processesand contaminants on the composition of northern EPRMORB (8–10°N and 12–14°N): Evidence from volatiles(H2O, CO2, S) and halogens (F, Cl), Earth Planet. Sci. Lett.,251, 209–231, doi:10.1016/j.epsl.2006.09.012.

Malinin, S. D., I. F. Kravchuk, and F. Delbove (1989), Chloridedistribution between phases in hydrated and dry chloride!aluminosilicate melt systems as a function of phase compo-sition, Geochem. Int., 26, 32–38.

Mastin, L. G., R. L. Christiansen, C. Thornber, J. Lowenstern,and M. Beeson (2004), What makes hydromagmatic erup-tions violent? Some insights from the Keanakako’i Ash,Kilauea Volcano, Hawai’i, J. Volcanol. Geotherm. Res.,137, 15–31, doi:10.1016/j.jvolgeores.2004.05.015.

Michael, P. J., andW. C. Cornell (1998), Influence of spreadingrate and magma supply on crystallization and assimilationbeneath mid!ocean ridges: Evidence from chlorine and majorelement chemistry of mid!ocean ridge basalts, J. Geophys.Res., 103(B8), 18,325–18,356, doi:10.1029/98JB00791.

Michael, P. J., and J.!G. Schilling (1989), Chlorine in mid!ocean ridge magmas: Evidence for assimilation of seawater!influenced components, Geochim. Cosmochim. Acta, 53,3131–3143, doi:10.1016/0016-7037(89)90094-X.

Perfit, M. R., W. I. Ridley, and I. Jonasson (1999), Geologic,petrologic and geochemical relationships between magma-tism and massive sulfide mineralization along the easternGalapagos Spreading Center, Rev. Econ. Geol., 8, 75–100.

Perfit, M. R., J. R. Cann, D. J. Fornari, J. Engels, D. K. Smith,W. I. Ridley, and M. H. Edwards (2003), Interaction of seawater and lava during submarine eruptions at mid!oceanridges, Nature, 426(6962), 62–65, doi:10.1038/nature02032.

Schiffman, P., and G. E. Lofgren (1982), Dynamic crystalliza-tion studies on the Grande Ronde pillow basalts, centralWashington, J. Geol., 90, 49–78, doi:10.1086/628651.

Soule, S. A., D. J. Fornari, M. Perfit, J. R. Cann, L. Montesi,and W. I. Ridley (2006), Incorporation of seawater into mid!ocean ridge lava flows during emplacement, Earth Planet.Sci. Lett., 252, 289–307, doi:10.1016/j.epsl.2006.09.043.

Webster, J. D., R. J. Kinzler, and E. A. Mathez (1999), Chlorideand water solubility in basalt and andesite melts and implica-tions for magmatic degassing, Geochim. Cosmochim. Acta,63, 729–738, doi:10.1016/S0016-7037(99)00043-5.

GeochemistryGeophysicsGeosystems G3G3 SCHIFFMAN ET AL.: CONTAMINATION OF BASALTIC LAVA BY SEAWATER10.1029/2009GC003009

12 of 12