crystallization and degassing in the basement sill ...geoscience.unlv.edu/files/boudreau and simon...

Crystallization and Degassing in the BasementSill, McMurdo DryValleys, Antarctica

ALAN BOUDREAU1* AND ADAM SIMON2

1DIVISION OF EARTH AND OCEAN SCIENCES, NICHOLAS SCHOOL OF THE ENVIRONMENT AND EARTH SCIENCES,

DUKE UNIVERSITY, DURHAM, NC 27708, USA2DEPARTMENT OF GEOSCIENCE, UNIVERSITY OF NEVADA, LAS VEGAS, NV 89154-4010, USA

RECEIVED AUGUST 8, 2006; ACCEPTED MARCH 30, 2007ADVANCE ACCESS PUBLICATION MAY 23, 2007

The Basement Sill is part of the Ferrar Large Igneous Province

exposed in the McMurdo Dry Valleys, Antarctica. The sill is

�330 m thick in the Bull Pass area and 450þm thick in the Dais

area, �12 km to the west, and is characterized by phenocryst-free

lower and upper margins and an orthopyroxene-rich central ‘tongue’

(opx 1^5 mm in size). Halogen variations in apatite from a suite of

samples collected along vertical transects through the sill were exam-

ined to evaluate the process of crystallization-induced degassing

(i.e. second boiling) and its effects on magma chemistry. Apatite

grains from any given sample are generally unzoned with respect to

Cl and F concentrations, but may vary by 20^30 mol% in the halo-

gen site between grains. Overall average Cl/F mass ratios increase

with height from the lower margin to the center of the sill, and then

decrease to near zero towards the top margin where the rocks are

relatively oxide-rich. The Cl/F trend parallels those of bulk MgO

and grain size.The upper margin contains abundant mafic pegma-

toids and the apatite in these segregations has lower Cl/F ratios

compared with that in the host-rocks, although REE show no

measurable difference. Numerical modeling illustrates that a cooling

and crystallizing sill initially develops two separate vapor-saturated

zones at the lower and upper margins owing to the irreversible heat

loss to the cooler country rock.Vapor separating from the lower zone

migrates upward into hotter silicate liquid, where it is resorbed, thus

increasing the Cl/F mass ratio of the liquid. This process leads to

saturation and precipitation of apatite from the liquid with a higher

Cl/F ratio than would otherwise occur.Volatile enrichment can also

aid compaction and grain growth in the central part of the sill.

In contrast, the relatively Fe-rich, Cl-poor nature of the upper zone

rocks suggests that these rocks may have crystallized from more

evolved, degassed silicate liquid, possibly compacted out of the under-

lying crystal mush. In addition, as vapor sourced from the lower and

central parts of the sill ascends into the cooler upper zone of the sill,

the vapor may be localized (along with late interstitial silicate

liquid) to form pegmatoids at temperatures at which Cl is less

favored in apatite and can be leached from existing apatite by the

ascending vapor, the latter causing the observed decrease in the Cl/F

mass ratio of apatite in the (evolved) pegmatoids.

KEY WORDS: Ferrar Igneous Province; halogens; fluid; apatite

I NTRODUCTIONMagmatic volatile phases can play a number of importantroles in the generation, crystallization, and eruption of maficmagmas (Roggensack et al., 1997; Huppert & Woods, 2002;Yang & Scott, 2002; Cervantes & Wallace, 2003; Wallace,2003; Cashman, 2004; Gonnerman & Manga, 2005;Hammer, 2006). However, in plutonic rocks the evidence ofdegassing is difficult to detect given the fugitive nature ofvolatile fluids (Candela & Blevin, 1995). Halogen variationsin apatite are one mechanism that has been found useful indeciphering the complex degassing history of layered intru-sions and associated mafic rocks (e.g. Brown & Peckett, 1977;Boudreau & McCallum, 1989;Willmore et al., 2000).This study details the halogen composition of apatite in

dolerite from the Basement Sill from the west Bull Pass andDais areas (where the sill is 330^350 and 450þ m thick,respectively; Marsh, 2004) of the Ferrar Igneous Provincein the McMurdo DryValleys, Antarctica. The crystallizedmagma bodies provide an excellent field laboratory for thestudy of vapor migration in a solidifying crystal pile asboth were emplaced as an orthopyroxene-phyric mush(Marsh, 2004) with a relatively simple emplacement and

*Corresponding author. Telephone: 919-684-5646. Fax: 919-684-5833.E-mail: [email protected]

� The Author 2007. Published by Oxford University Press. Allrights reserved. For Permissions, please e-mail: [email protected]

JOURNALOFPETROLOGY VOLUME 48 NUMBER 7 PAGES1369^1386 2007 doi:10.1093/petrology/egm022

crystallization history that can otherwise complicate theunderstanding of the compositional evolution of vaporfrom a basic igneous intrusion.

GENERAL GEOLOGY ANDPREV IOUS INTERPRETAT IONSThe Ferrar Large Igneous Province is located along strikeof the Trans-Antarctic Mountains, which run north^southand divide Antarctica into the Eastern and Western icesheets (Gunn & Warren, 1962; McKelvey & Webb, 1962;Gunn, 1966; Borg et al., 1990; Elliot & Fleming, 2004).The province was produced during the Jurassic-age(Heimann et al., 1994; Knight & Renne, 2005) fragmenta-tion of Gondwanaland. In the McMurdo Dry Valleys inSouthern Victoria Land the Ferrar consists of a series offour vertically interconnected dolerite sills (Fig. 1), inter-connecting dykes and extrusive rocks (i.e. the �500mthick Kirkpatrick flood basalt province). The high-aspectratio dolerite sills vary in thickness from 100 to 500m andcrop out over an area of the order of 10 000 km2. In orderfrom bottom to top they are called the Basement Sill,Peneplain Sill, Asgard Sill and Mt. Fleming Sill (Gunn &Warren, 1962; Hamilton, 1965). The sills are typicallyseparated from one another by several hundred meters,except in the area of Bull Pass, where the Basement Silland Peneplain Sill are in contact with one another andwere interpreted by Marsh (2004), on the basis of fieldevidence, to have formed from the top down.The compositions of the Ferrar sills are dominantly

quartz-normative dolerite. The lowermost exposed sill, theBasement Sill, is composed of a lower phenocryst-freegabbronorite chilled marginal zone, an orthopyroxenephenocryst-bearing central zone, identified as the‘Opx tongue’ by Marsh (2004), and a phenocryst-freegabbronorite upper margin (Gunn & Warren, 1962;Hamilton, 1965; Marsh, 2004). In the West Bull Pass area,the lower chilled margin is dark, aphanitic to microcrystal-line, contains plagioclase, orthopyroxene, augite andinverted pigeonite, and has an abrupt contact with theoverlying opx tongue. Moving up from the lower margin,the grain size and opx content of the sill both increasetoward the center of the Opx tongue, and then bothdecrease toward the top margin. The increase in orthopyr-oxene modal abundance in the center of the sill manifestsitself in a strong change in bulk-rock MgO concentrationfrom �7 wt % in both the upper and lower margins to�20 wt % in the tongue (Marsh, 2004). The dark-coloredupper margin superficially mimics the lower chilledmargin and is comprised dominantly of plagioclase, ortho-pyroxene, augite and inverted pigeonite; however, theupper margin has a more Fe-rich bulk composition and,for the most part, comparatively larger grain sizes (fine-to medium grained) relative to the aphanitic lower chilled

margin.The difference in grain size between the lower andupper chilled margins probably reflects higher countryrock temperatures above the sill at the time of emplace-ment owing to the high magma flux feeding theKirkpatrick flood basalts and the formation of earlier sills(i.e. Peneplain Sill) above the Basement Sill (Hersum et al.,submitted). Oxides (magnetite and ilmenite) are particu-larly abundant at the very top of the upper margin.Another difference between the lower and upper marginsis the common presence of mafic pegmatoids and grano-phyric segregations beginning in the upper part of theopx tongue and continuing to the top of the sill (Zavala &Marsh, 2001, 2002). The pegmatoid segregations arecomposed dominantly of plagioclase and pyroxene (thelatter to several centimeters in length) with minor modalamounts of quartz, alkali feldspar, amphibole, apatite, bio-tite and oxides. The pegmatoids are enriched in SiO2, FeOand incompatible elements (e.g. K, Ba, Rb, Nb and Ti)and have higher Ab:An (�25%) ratios in plagioclaserelative to plagioclase in the host-rock. The liquids thatcrystallized to yield the pegmatoids are inferred torepresent the end products of the crystallization of late-state interstitial liquids that were probably expelled(i.e. filter pressed) from the surrounding gabbronoriteduring compaction of the magma (Geist et al., 2005).The abrupt contact between the lower and upper

margins and the Opx tongue manifests itself in a distinctbreak in slope noted in a crystal size distribution study(Zieg & Forsha, 2005) as well as a dramatic rise intemperature (�2008C) inferred from two-pyroxenethermometry (Simon & Marsh, 2005). These results reflectdirectly the presence of the extensive Opx tongue andsuggest that the pyroxene phenocrysts in the tongue wereentrained in the ascending magma after having beentexturally equilibrated at a much deeper level of themagmatic plumbing system than the present level of sillemplacement. Notably higher concentrations of Cr and Alin the Opx phenocrysts in the tongue relative to the upperand lower margins lend additional support to this hypoth-esis (Simon, in preparation).The magma that formed mostof the Basement Sill probably intruded as a crystal-richmush (i.e. silicate liquidþphenocrysts; Marsh, 2004;Charrier & Marsh, 2005; Petford et al., 2005) and thespatial position of the tongue reflects the inability ofthe large orthopyroxene phenocrysts to settle throughthe viscous lower margin.Finally, the Dais Intrusion makes up the Dais, a

topographic feature about 12 km west of the Bull passarea, and is an extension of the Basement Sill. It differsfrom the sill in the Bull Pass area in being thicker(the exposed portion is �450m thick; the lower contactis not exposed) and in having more pronounced modallayering, particularly in the upper part of the Opx tongue(Marsh, 2004). The layering is defined by sharp changes

JOURNAL OF PETROLOGY VOLUME 48 NUMBER 7 JULY 2007

1370

Fig. 1. Top: geological map of part of the DryValleys, Antarctica, showing locations of the Bull Pass and Dais areas (after Marsh & Zieg, 2004).The Asgard and Mt. Fleming sills are spatially closely associated and are shown here as one unit. Bottom: simplified sections of the BasementSill in the Dais and west Bull Pass areas.

BOUDREAU & SIMON DEGASSING IN BASEMENT SILL, ANTARCTICA

1371

in the amounts of orthopyroxene and plagioclase andranges in thickness from centimeter to several meter scale.

Petrology of late-crystallizing mineralsAll halogen-bearing minerals are present in only minormodal abundance and none appear to have been earlyliquidus (i.e.cumulus) phases. Apatite is the principalmineral of interest as it is the most commonly occurring(and probably the first precipitating) halogen-bearingmineral. Neither mica nor amphibole shows any texturalevidence of crystallizing before apatite. This is consistentwith evidence from layered intrusions crystallized ina low-pressure environment (such as the SkaergaardIntrusion) where apatite is the first halogen-bearing‘cumulus’ mineral observed after �90% crystallization(e.g. Wager & Brown, 1967). The crystallization sequenceof a representative chilled margin composition was mod-eled using MELTS (Ghiorso & Sack, 1995), which predictsthat apatite is the sole halogen-bearing mineral duringcooling. Also, unlike the micas and amphiboles, whichhave a strong crystal-chemical control on halogen substitu-tion, the apatite halogen solid solution (i.e. Cl,F) is idealat high temperatures (e.g. Volfinger et al., 1985; Tacker &Stormer, 1989).The apatite grains analyzed in this study are interstitial

to early formed minerals (plagioclase and pyroxene),and are typically associated with other late-crystallizedphases such as quartz, Fe^Ti oxides and, relatively rarely,biotite. Apatite modal abundance is qualitatively corre-lated with (unpublished) whole-rock P2O5 concentrations,suggesting that apatite is the dominant phosphorus-bearing phase in the rocks. In most instances apatitedisplays a characteristic acicular habit, forming thin rods

usually no more than a few tens of microns in width butlocally up to about 300 mm in length. In the central partof the sill, however, grains can be considerably thickerand have a tabular habit (Fig. 2).Finally, apatite from a few samples of the exposed part

of the Dais was analyzed to see if lateral compositionalvariation is present in the Basement Sill between the BullPass and Dais Intrusion.

METHODSApatite grains in polished thin sections were analyzed bywavelength-dispersive spectrometry (WDS) using theCameca Camebax electron probe microanalyzer (EPMA)at Duke University, Durham, North Carolina, USA.Typical analytical conditions were: 15 keV accelerationvoltage, 15 nA beam current, and a focused beam(a requirement given the typically small size of individualgrains). Peak and background counting times for Cl and Fwere in the ranged of 20^30 s and 10^15 s, respectively.Other elements were counted on the peak for 30 s andbackground for 15 s. To minimize volatilization, F and Clwere counted first. Standards included natural chlorapatite(RM-1, Morton & Catanzaro, 1964) and fluorapatite(from Wilberforce, Ontario) for Ca, P and the halogens,and allanite (102522, #42; Frondel, 1964) for La and Cefor which the analyte lines are relatively free from interfer-ences (Roeder, 1985). Standard intensities for F and Clusing the apatite standards were also checked againstfluorite, topaz and halite. Fluorine was analyzed usinga synthetic Si^W layered diffracting crystal. The OH� ionand H2O were calculated by difference based on idealOH� site occupancy. Previous work with these analytical

Fig. 2. Backscattered electron image of apatite from the basal Ferrar sill, west Bull Pass, Antarctica. Sample 31U: acicular apatite from the lowerchilled margin, just below the orthopyroxene tongue. (SeeTable 1, analysis 1, for analysis.) Sample 14N: large, tabular apatite from the center ofthe orthopyroxene tongue. (SeeTable 1, analyses 5 and 6, for core and rim analysis.) Ap, apatite; Pl, plagioclase; Opx, orthopyroxene; Q, quartz.(Note different scales.)


1372

conditions (seeWillmore et al., 2000) has shown no signifi-cant apatite orientation effects on the analyses asreported by Stormer et al. (1993). However, to minimizeany potential orientation effects, preference was given tograins cut parallel to the c-axis.

RESULTSApatite compositionRepresentative apatite compositions are listed in Table 1(a complete set of analyses is available as an ElectronicAppendix at http://www.petrology.oxfordjournals.org/)and those from west Bull Pass are shown as a function ofheight in the sill in Fig. 3. The mole fraction of F rangesfrom 0�4 to end-member fluorapatite (and a few analysesfor which the mole fraction F exceeds 1�0), whereas Clranges from nears its limit of detection (LOD �0�01wt %) to a maximum of about 0�5mole fraction. As hasnow been observed in a number of studies (e.g. Boudreauet al., 1986, 1992, 1995; Willmore et al., 2000), individualapatite grains from any given sample tend to be homo-geneous and not significantly zoned in Cl and F (e.g.Table 1, analysis 5 and 6), whereas different grains in the

same sample can vary by as much as 30þ mol% in thehydroxyl/halogen site. Thus, detailed stratigraphic changescan be masked by this within-sample variation. However,overall there remains a broad increase in Cl and decreasein F from the lower margin towards the center of the sill,and Cl then decreases to below LOD in the upper margin.These stratigraphic changes in Cl broadly parallel trendsin the bulk-rock MgO and grain size (Fig. 3).The compositions of Ferrar apatite along with composi-

tional fields from a number of other mafic intrusions areplotted in an F^Cl^OH ternary diagram in Fig. 4.The few analyses from the Dais overlap those from theBull Pass area, suggesting that there are no significantregional variations in apatite chemistry. The Cl-poornature of most Ferrar apatite is typical of intrusions suchas the Skaergaard (Greenland), Great Dyke (Zimbabwe),Kiglapait (Canada), and Munni Munni (Australia),where the mole fraction of the chlorapatite component isgenerally less than 20mol% (see Boudreau, 1995). Exceptfor the few Cl-rich samples from the central part of theOpx tongue, none are as Cl-rich as is seen in the lower por-tions of the Stillwater and Bushveld intrusions below theeconomically important platinum-group element zones of

Table 1: Selected electron microprobe analyses of apatite in the basal Ferrar Sill, Antarctica

Analysis: 1 2 3 4 5 6 8 9 10 11

Sample: 31U 27N 24N 17U 14N-core 14N-rim 06N 01N DG-101 DG-103

Location: Bull Pass Bull Pass Bull Pass Bull Pass Bull Pass Bull Pass Bull Pass Bull Pass Dais Dais

Height� (m): 3 25 83 163 196 196 284 339 near top near top

Rock type: gabbro opx’nite opx’nite opx’nite opx’nite opx’nite gabbro gabbro pegmatoid host gabbro

wt %

CaO 55�3 55�1 54�4 55�5 55�2 54�4 54�7 54�7 54�5 55�4 (0�5)

P2O5 41�1 40�9 41�4 41�0 41�0 41�1 41�8 41�0 41�4 40�6 (0�4)

F 2�75 2�69 2�49 1�59 2�41 2�21 2�42 3�62 3�45 3�29 (0�24)

Cl 0�48 0�77 0�88 1�73 0�57 0�58 0�24 0�12 0�18 0�30 (0�03)

H2Oy 0�35 0�30 0�34 0�59 0�49 0�56 0�55 0�01 0�07 0�14

La3O3 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 0�10 0�14 (0�05)

Ce2O3 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 0�44 0�42 (0�05)

SiO2 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 0�15 0�29 (0�02)

Na2O n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 0�13 0�05 (0�02)

Total 99�9 100�6 99�5 100�4 98�9 99�8 99�7 99�4 100�4 100�6

O� F,Cl 1�26 1�31 1�25 1�06 1�05 1�14 1�07 1�55 1�49 1�45

Total 98�6 99�3 98�3 99�4 97�8 98�6 98�7 97�8 98�9 99�2

XCl 0�07 0�11 0�13 0�25 �08 �08 0�03 0�02 0�03 0�04

XF 0�73 0�72 0�68 0�42 �64 �60 0�65 0�98 0�93 0�88

XOHy 0�20 0�17 0�20 0�33 �28 �32 0�31 0�01 0�04 0�08

n.a., not analyzed; opx’nite, orthopyroxenite. Numbers in parentheses for analysis 11 are 1s counting statistic errors andare typical for all analyses.�Height relative to lower contact.yH2O and OH calculated on the bases of Clþ FþOH¼ 1 per formula unit.


1373

http://www.petrology.oxfordjournals.org/

these two intrusions. Apatite in the Stillwater and Bushveldintrusions is strongly enriched in the end-member chlora-patite component throughout rather sizeable stratigraphicthicknesses.

Variations in apatite compositions acrossa pegmatoid^host-rock contactApatite was analyzed in three samples collected acrossa pegmatoid^host-rock contact from the Dais layeredintrusion: one from within the pegmatoid segregation,one from the pegmatoid^gabbronorite host-rock contactand one 3m into the host-rock.The samples were collected

from the upper part of the Dais intrusion stratigraphicallyabove the Opx tongue. Data from these apatite analysesare shown in Fig. 5 and example analyses are provided inTable 1; analysis 10 is apatite from within the pegmatoidand 11 is apatite from the host gabbro. Apatite within thepegmatoid is Cl poor (all50�2 wt % Cl) and most grainsare nearly end-member fluorapatite. Furthermore, apatitein the pegmatoid contains even less Cl than apatite inthe surrounding upper zone host-rock. Bulk analysisof the pegmatoid demonstrates that the concentrations ofincompatible trace elements are elevated between twoand three times their concentrations in the host-rock

Fig. 3. Summary results from the Ferrar Sill, Bull Pass area. (a) Mole fraction Cl (filled symbols and F (open symbols) in apatite; (b) molarCl/F ratio of apatite; (c) bulk-rock MgO concentration (wt %); (d) average orthopyroxene grain size.


1374

(Geist et al., 2005); however, the concentrations of rareearth elements (REE) in apatite are essentially unchangedfrom host gabbronorite to the pegmatoid.

DISCUSSIONTo summarize, apatite in the Basement Sill shows signifi-cant Cl/F variations both within individual samples andas a function of stratigraphic position, with the highestCl/F ratios occurring in the middle of the sill. Assumingthat the apparent rapid quench typical of the lower chilledmargin preserves apatite halogen compositions in equili-brium with the initial (unfractionated) liquid, the initialapatite is characterized by X(Cl) �0�05 and X(F) �0�7.Despite the complexity of the magma dynamics indi-

cated by the characteristics of the Opx tongue, a notablefeature of the sill is that, once emplaced, it cooled uni-formly from the top and bottom such that the silicateliquid would have become vapor-saturated nearly simulta-neously at both margins. The sill took some 1000 yearsto solidify and evidence for post-solidification low-temperature hydrothermal alteration is absent (Simon &Marsh, 2005). As mentioned above, the finer grain size in

the lower margin relative to the upper margin suggeststhat the upper margin may have cooled slightly moreslowly; the proximity of the Peneplain Sill above theBasement Sill may have played a role in this (see Hersumet al., submitted). Thus, the rapid crystallization time andpreservation of magmatic halogen ratios in trace phasessuch as apatite provide the opportunity to study the lossof vapor evolved from the upper margin to the overlyingcountry rock and the migration of vapor evolved from thelower margin as this vapor ascended through the sill.Prior to the interstitial liquid becoming saturated in

a halogen-bearing phase (e.g. apatite, biotite, amphiboleor vapor), one would not expect there to be fractionationof halogens by crystallization of the observed anhydrousminerals. Ignoring vapor for the moment, and although itappears that apatite is the first halogen-bearing mineralphase to crystallize, the crystallization of apatite, biotiteor amphibole should all incorporate F in preference to Cl(e.g. Speer, 1984; Candela, 1986), particularly in rocks withhigh Mg/Fe ratio (e.g. Volfinger et al., 1985). Consideringonly apatite, Beswick & Carmichael (1978) noted thatfluorapatite is generally more stable (i.e. crystallizes ear-lier) than chlorapatite and hydroxyapatite at magmatic

Fig. 4. Comparison of Ferrar apatite from the Bull pass area (*) and the Dais Intrusion (þ) with those from the Stillwater and Bushveldcomplexes (only the range below the platinum reefs is shown), as well as the Munni Munni and Great Dyke. Inset shows the range of apatitefrom associated sill, dike and marginal rocks associated with these other intrusions. After Boudreau (1995).


1375

temperatures and the phase equilibrium interpretations ofTacker & Stormer (1993) are consistent with their observa-tions. This suggests that apatite behaves similarly to otherhalogen- and OH-bearing minerals in that F-bearingend-members have significantly higher thermal stabilities(Patin‹ o Douce & Roden, 2006). Hence, Cl/F variations inapatite are analogous to Mg/Fe variations in olivine ororthopyroxene (Cawthorn, 1994), such that an increasein the Cl/F ratio of apatite implies both an increase in theCl/F ratio of the interstitial silicate liquid and that Cl-rich

apatite formed from the primary Basement Sill silicateliquid should be stable only at lower temperatures. Thus,the Cl/F mass ratio in the silicate liquid should increaseafter apatite and other Cl- and F-sequestering mineralphases begin to crystallize from the interstitial liquid.An exsolving vapor will effectively decrease the Cl/Fmass ratio of the silicate liquid as degassing progresses(as discussed more fully below). As pockets of interstitialsilicate liquid that are saturated in apatite or other halo-gen-bearing phases become isolated during the final

Fig. 5. Comparison of apatite traverse from host-rock (upper left photomicrograph), through contact (center photomicrograph) and topegmatoid (upper right photomicrograph) from the upper margin of the Dais Intrusion. Plots are of (a) Cl, (b) F, and (c) Ce2O3 concentration(wt %) in apatite as a function of relative distance across the thin section shown in the photomicrographs at the top.


1376

crystallization of interstitial liquid, apatite of differentcompositions can be produced. This is considered to bethe main cause of the observed compositional variationin apatite within individual samples.The overall stratigraphic changes in the Cl/Fand OH/F

ratio of apatite can be caused by several crystallizationmechanisms, some of which have been discussed previously(Boudreau et al., 1986; Boudreau & McCallum, 1989;Boudreau & Kruger, 1990; Mathez & Webster, 2005).The effect of temperature and of variations in the activitiesof F and Cl on apatite compositions have been describedby Boudreau et al. (1992), who considered the equilibriumbetween apatite and an aqueous fluid, for which thermo-dynamic data are available, that are both in equilibriumwith a silicate liquid. Although the activities of the halo-gens in a silicate liquid are not expressed explicitly bysuch a treatment, an increase in the activity or chemicalpotential of either of the halogens in any phase mustoccur in all other coexisting halogen-bearing phasesas well.Let us consider the exchange reaction between apatite

(ap) and an aqueous fluid (af) in which HCl and HF arepresent as neutral species:

F� apatitesolid þHClaf ¼ Cl� apatitesolid þHFaf ð1Þ

and the accompanying apparent exchange constant:

logK ¼ logX

apCl a

afHF

XapF aafHCl

� �: ð2Þ

In equation (2), Xji denotes the mole fraction of component

i in phase j and aji denotes the activity of i in phase j.

Activity coefficients for these species are not known and,thus, are not included in the apparent exchange constant.The relationship between fluid composition and apatitecomposition is given by

XapCl

XapF

� �¼ K

aafHCl

aafHF

� �: ð3Þ

Analogous equations can be written to describe OH^Fexchange between apatite and aqueous fluid. Using theobservation that halogen substitution in apatite is idealabove 5008C (Tacker & Stormer, 1989), Boudreau et al.(1992) fitted expressions for log K to a simple tempera-ture-dependent function. Shown in Fig. 6 are calculatedapatite compositions that result from changing eitherthe Cl/For OH/F ratio of the silicate liquid at a fixed tem-perature of 11008C (dashed lines) or from changing thetemperature at fixed HCl/HF and H2O/HF activity ratiosof 3�0 and 10�0, respectively (continuous lines). The modeltrends in Fig. 6 suggest that the observed variations inapatite compositions may be caused by changing eitherthe HCl/HF and H2O/HF ratios by approximately anorder of magnitude, or by varying the temperatureat which apatite equilibrates by up to 5008C.

As noted above, there do not appear to be any halogen-sequestering minerals crystallizing in the sill prior toapatite saturation; hence, fractional crystallization of theobserved halogen-free minerals alone cannot lead to varia-tions in halogen ratios as a function of stratigraphicposition. That is, because P behaved as an incompatibletrace element during the crystallization of the BasementSill, and assuming as a first approximation that apatitecrystallization was controlled only by the P concentrationin the silicate liquid, then all interstitial silicate liquidshould have reached apatite saturation after the sameamount of crystallization. The system would have followeda single liquid line of descent with respect to apatitecrystallization and apatite would have begun to crystallizeat about the same temperature throughout the intrusion.Because this is inconsistent with a large temperaturerange in apatite equilibration, temperature effects alonecannot be the major cause of the stratigraphic variation inapatite composition. One would also expect the same forbiotite and amphibole crystallization.However, the separation of an aqueous fluid phase from

a crystallizing silicate liquid can fractionate the halogensowing to the strong affinity of Cl relative to F for anexsolved aqueous fluid phase relative to silicate liquid(Sourirajan & Kennedy, 1962; Helgeson, 1964; Burnham,1967; Roedder, 1984, 1992; Bodnar et al., 1985; Candela,1986; Chou, 1987; Shinohara et al., 1989; Chou et al., 1992;Metrich & Rutherford, 1992; Sterner et al., 1992; Anderko& Pitzer, 1993; Cline & Bodnar, 1994; Lowenstern, 1994;Piccoli & Candela, 1994; Webster et al., 1999, 2004;Webster, 2004; Webster & DeVivo, 2004; Mathez &Webster, 2005). Furthermore, as discussed below, migrationof exsolved vapor through a crystallizing magma canreadily redistribute halogens from one region to another,leading to variations in halogen ratios in both the silicateliquid and apatite.We suggest that vapor evolved from the lower half of the

sill and subsequently migrated upward through the crystalmush where it encountered hotter, fluid-undersaturatedsilicate liquid (i.e. silicate liquid that had not reachedcrystallization-induced fluid saturation). The free vaporphase would have had a high Cl/F mass ratio owing to thestrong mass transfer of Cl from silicate liquid to vaporrelative to F (Dingwell & Scarfe, 1983; Webster &Holloway, 1990; Candela & Piccoli, 1995). As the vaporascended into hotter silicate liquid, the vapor would havebeen resorbed by the silicate liquid; hence, resulting in ele-vated volatile concentrations and higher Cl/F ratios inthe interstitial silicate liquid in the central part of the sill.This process would ultimately yield higher Cl/F ratios inapatite crystallized from this interstitial silicate liquid.In contrast, early formed apatite crystallized from theupper margin of the Basement Sill and its associatedpegmatoidal and granophyric segregations grew from


1377

more evolved silicate liquid that had already degassed andpreferentially lost an unquantified amount of Cl to theexsolved volatile phase. In addition, aqueous fluids sepa-rating from the lower and eventually from the middle ofthe intrusion migrated into cooler rocks of the uppermargin. For aqueous fluids with a fixed HCl/HF ratio,the fluids would be in equilibrium with lower Cl/F apatiteand could leach Cl from pre-existing apatite.

Evidence from pegmatoidsThe strong enrichments in SiO2, FeO and incompatibletrace elements in the pegmatoid (Geist et al., 2005) areconsistent with these segregations having crystallizedfrom late-stage, evolved interstitial silicate liquid. The lackof variation in rare earth element (REE) apatite notedin this study suggests that the higher bulk-rock REEconcentrations in the pegmatoids simply reflect a higheramount of this evolved liquid than is seen in the host-rock,but that the evolved liquid in both the pegmatoid and theinterstitial liquid of the host otherwise have similar REEconcentrations. However, the lower Cl/F ratio of apatitein the pegmatoid as compared with host apatite impliesthat the apatite equilibrated with a liquid or vapor withlower Cl/Fratio or that this ratio was fixed but equilibratedat a lower temperature, or both.

It is suggested that the pegmatoidal texture developedas vapor, evolved from the lower parts of the sill,used these regions of higher amounts of silicate liquid(i.e. pegmatoids) as preferential conduits through whichto ascend. Further, these aqueous fluids were relativelyCl-poor and were not effective at moving significantquantities of REE. Data from saline aqueous inclusionsin quartz hosted by granophyre of the Capitan Pluton(New Mexico) suggest that the light REE (LREE) maybe fractionated from the heavy REE (HREE) duringvolatile phase exsolution, with the LREE having beenscavenged by the saline aqueous fluid (Banks et al., 1994).In another study, whole-rock analyses of potassicallyaltered and unaltered granodiorite also showed enrich-ment in LREE in the aqueous-fluid-altered rocks (Taylor& Fryer, 1980). The ability of aqueous fluids to scavengepreferentially the REE from crystallizing silicate liquidhas also been shown experimentally (Flynn & Burnham,1978; Sakagawa, 1989; Reed, 1995; Reed et al., 2000).Although the mass transfer of individual REE from silicateliquid to vapor is a complex function of the total salinity ofthe vapor phase (Candela, 1990) and the concentrations ofall other chloride-bound elements in the vapor (Reed,1995;Reed et al., 2000), in general, the exsolution of a Cl-bearingaqueous fluid from a silicate liquid will lead to a decreasein the LREE/HREE mass ratio of the silicate liquid.

Fig. 6. Plots of (a) the Cl/Fratio as a function of height and (b) OH/Fratio as a function of height, both from the Basal Sill from the Bull Passarea. In (a), the continuous lines show the calculated Cl/F ratio of apatite in equilibrium with an aqueous fluid with a fixed HCl/HFactivityratio of 3�0 and at the various temperatures indicated. The dashed lines are the calculated Cl/F ratio of apatite in equilibrium with an aqueousfluid at 11008C and with different HCl/HF activity ratios as noted. In (b) the continuous lines show the calculated OH/F ratio of apatite inequilibrium with an aqueous fluid with a fixed H2O/HFactivity ratio of 10�0 at the various temperatures indicated. The dashed lines are thecalculated OH/Fratio of apatite in equilibriumwith an aqueous fluid at 11008C and with different H2O/HFactivity ratios indicated. (See text foradditional discussion.)


1378

The partition coefficients describing the simple mass trans-fer of La and Lu between silicate liquid and vapor increasefrom approximately 0�01 and 0�001when the Cl concentra-tion of the vapor is 0�01molal to 2 and 0�7 when the Clconcentration of the vapor is 2molal. Thus, if ascendingvapor is resorbed into stratigraphically higher, hottersilicate liquids (as discussed below), this should increasethe LREE concentration of that interstitial silicate liquidand, even upon subsequent vapor saturation of that parcelof silicate liquid, the LREE/HREE mass ratio might beexpected to be higher than that of silicate liquid strati-graphically lower in the magma chamber. The fact thatREE concentrations in the pegmatoid and host-rockapatite are equivalent suggests that the late-state intersti-tial silicate liquid that crystallized to form the pegmatoidsdid not resorb a significant quantity of ascending vaporand, further, that the vapor exsolving from this segregatedsilicate liquid was of a low enough salinity or temperaturethat it did not effectively fractionate the REE.

Modeling degassing in a coolingand crystallizing intrusionVapor separation in the cooling and crystallizing BasementSill was modeled using a modified version of theIRIDIUM program (Boudreau, 2003). This program usesa simplified version of the MELTS free energy minimiza-tion program (Ghiorso et al., 1994; Ghiorso & Sack, 1995)coupled with one-dimensional heat and mass transfer tomodel a number of igneous processes such as compactionand silicate liquid percolation through a porous assem-blage. The program has been modified to include a morecomplete C^O^H^S fluid (see the Appendix). It also hasbeen modified to allow for an upper and lower thicknessof material to act as a thermal sink; originally the programsimply assumed a fixed heat loss from the top and bottomof the magma column. This modification allowsmore realistic modeling of the thermal profile of a sillthat is losing heat by thermal conduction into coolercountry rock.The initial conditions of the model assume a 350m

thick, pyroxene-phyric sill that is allowed to cool andcrystallize as heat is lost to the over- and underlying coun-try rocks. Because we are interested specifically in theeffects that volatile saturation and volatile migration mayhave on halogen ratios, the model initially assumes auniform mush column. (The phenocryst-free marginalrocks are ignored as they probably represent an initialphenocryst-free leading edge of intruding magma thatwas essentially chilled prior to the emplacement of theOpx tongue.) Paleodepth estimates suggest country rocktemperatures as low as 1508C (Simon & Marsh, 2005;Hersum et al., submitted), but as the rocks are likely tohave been preheated by intrusion of the overlyingPeneplain sill prior to emplacement of the Basement Sill,we used a uniform initial country rock temperature

of 3008C in the model. Pressure at the top of the sillis 100MPa, based on an intrusion depth of 1km andassuming a lithostatic pressure gradient of 100MPa/km.The bulk composition of the magma (i.e. silicate

liquidþcrystals) is composed of 60% chilled marginalcomposition taken from Gunn (1966) at the temperatureat which this liquid is saturated with orthopyroxene(11308C) plus 40% orthopyroxene of mg-number 85.To this is added 0�6 wt % H2O, 0�01 wt % CO2 and0�03 wt % S. The initial volatile contents of the Ferrarmagmas are not known, and we have used a value thatfalls within the high end of water contents observed inmafic magmas [see summary by Boudreau et al. (1997)].This value was used to prevent free energy minimizationproblems at low liquid fractions. Also, the model calcula-tion does not go to 100% crystallization so that the actualquantity of vapor generated is relatively small. After 90%crystallization the liquid has lost most of its CO2 andis degassing nearly pure H2O (as discussed below);this composition vapor would simply continue with addi-tional crystallization. Finally, because the chilled margincomposition of Gunn is not in equilibrium with the addedorthopyroxene and volatiles, the initial calculated solidassemblage is actually a mix of olivine (�10%) and ortho-pyroxene (�25%). Over the course of the run, olivine isreplaced with orthopyroxene by peritectic reactions withthe silicate liquid.Chlorine and F are treated as trace elements in the

calculation, and are assumed to have vapor^silicateliquid partition coefficients of 100 and 0�1, respectively.These partition coefficient values are within the rangeobserved in both natural and experimental systems acrossa wide range of silicate liquid compositions (Carroll &Webster, 1994; Saal et al., 2002). Chlorine and F are other-wise assumed to not partition into any solid phases.Minerals that crystallize over the course of the reactioninclude orthopyroxene, olivine, plagioclase, magnetite,alkali feldspar and an immiscible Fe^S liquid. In additionto crystallization, the mush column is allowed to compactfollowing the model of Shirley (1986). Any vapor that isgenerated is assumed to move upward at a fixed velocityequal to 1�56 cm/day; this vapor ascent velocity is one-halfthe characteristic compaction velocity at the start of thesimulation. Because each time step is calculated fromthe compaction velocity profile (which slows downover time as the sill solidifies), this value was selected sothat the vapor moves about one space step for each timestep over the course of crystallization. This number isotherwise arbitrary, as there are no models that expresshow bubbles move through a crystallizing and compactingmush column. Nor does vapor degassing or migrationaffect compaction rates in the model, except for the effectof dissolved H2O in the silicate liquid on viscosityof the latter.


1379

None of the results presented below are qualitativelyaffected by modest changes in any of the above initialmodel parameters, as long as the mush column is notinitially vapor-saturated. For example, changing thevolatile concentrations of the initial silicate liquid orvapor velocity can change the point at which the silicateliquid becomes vapor saturated or the time scale at whichvapor fronts move, but the relative change in CO2/H2Oexsolved over time will be similar to that discussed below.In this respect, the main conclusions are robust.Model results for times of 0, 31, 107, 210 and 432 years

are shown in Fig. 7. Heat loss and crystallization lead tocrystallization-induced vapor saturation (i.e. second boil-ing) initially occurring in the upper and lower margins asthe solidification fronts propagate in toward the center ofthe sill. The extents of these zones at the labeled timesare shown in the ‘Mass vapor’ plot in Fig. 7. Althoughthe initial silicate liquid is relatively H2O-rich, the initialvapor separating from both the upper and lower zonesis CO2-rich (i.e. high initial CO2/H2O ratio) owing tothe much lower solubility of CO2 (Eggler et al., 1974;Mattey, 1991; Lowenstern, 2000). As degassing progressesthe vapor separating at any stratigraphic level becomesmore H2O-rich and the CO2/H2O ratio continuallydecreases. The concentration of H2S in the vapor remainslow at all times. Spatially, this is expressed by the vaporin the upper and lower margins being more H2O-richthan the more interior gases at any given time, at leastduring the early stages of degassing. At some time between61 and 107 years the upper and lower vapor saturationzones merge and the entire sill becomes vapor-saturated.Over time, CO2 is effectively flushed from the sill suchthat by 432 years the fluid is essentially all H2O with

a few mol% H2S, except in the center of the sill wherethe concentration of CO2 remains high.In the upper zone, progressive devolatilization and the

separation and loss of vapor from the top of the intrusionto the overlying country rock results in the preferential lossof Cl from the interstitial silicate liquid. Also, although Clis continually being advected through the sill in vaporevolved and ascending from silicate melt in the lowermargin, the Cl content of interstitial liquid in theupper zone remains approximately constant. In contrast,the concentration of F in the interstitial liquid increasesas crystallization proceeds. The net result is a decreaseover time in the Cl/F mass ratio in the upper part ofthe sill.The generation and migration of vapor is more compli-

cated in the lower zone. Vapor exsolved near the lowermargin ascends into hotter interstitial liquids that are notyet vapor-saturated. This results in the resorption of vaporin the hotter, interstitial silicate liquid. This process hasfour effects. First, as Cl is preferentially partitioned intothe aqueous fluid relative to F, the resorption of vaporwith a high Cl/F mass ratio into the interstitial silicateliquid leads to the formation of a peak in the Cl concentra-tion in the interstitial silicate liquid that migrates upwardthrough the magma with time. Second, the enrichment ofvolatile components in stratigraphically higher interstitialsilicate liquid owing to resorption causes this interstitialsilicate liquid to achieve volatile saturation and exsolvea vapor phase earlier than would have occurred basedon the original volatile concentrations in this silicateliquid fraction. This results in the lower vapor-saturatedzone moving into the center of the sill faster thanthe upper zone. Third, this upward advancing ‘wet front’

Fig. 7. Results of IRIDIUM model of a solidifying and degassing 350m thick orthopyroxene-phyric sill at 11308C instantaneously intruded intocountry rock initially at 3008C. Profiles are shown at time of 0, 31, 107, 210 and 432 years. From left to right, plots are of: (a) temperature in thesill (white) and country rock (grey); (b) per cent solid; (c) mass of vapor exsolved during crystallization (wt %); (d) mole percentage of thevapor components H2O (continuous lines), CO2 (dashed lines) and H2S (dotted line, shown at 432 years only); (e) Cl (continuous lines) and F(dashed lines) concentrations in liquid, normalized to original concentrations. (See text for additional discussion.)


1380

produces a peak in the mass of (H2O-rich) vapor evolvedthat also moves upward a short distance behind the peakin Cl concentration in interstitial liquid. Fourth, progres-sive degassing from the lower margin produces moreextreme depletions in CO2 in the fluid than occurs in theupper zone. This difference occurs because, unlike theupper margin, the lower margin does not have a source ofCO2 from degassing interstitial liquids below it.The migration of both the ‘Mass vapor’ and Cl peaks

continues once the upper and lower vapor saturationzones merge. However, the peak Cl/F ratio of the intersti-tial silicate liquid declines over time. Although not shown,once the Cl peak passes above the center of the sill itencounters progressively colder magma and, as pointedout above, this will not favor increasing the Cl concentra-tion of apatite. Furthermore, this cooler magma containsmore evolved interstitial silicate liquid, which containsa higher concentration of F, and hence again Cl/F ratiosare not strongly elevated.

Additional compaction effectsShown in Fig. 8 are snapshots at 210 years of the phasemode profile (wt%), bulk MgO, mass quantity of vapor,solid and liquid velocities and the per cent deformation ofthe solid matrix. The profiles in most of these plots showthe effects of compaction. At 210 years, the model solidand liquid velocities in the margins are near zero.Thus, active compaction of the mush column at this timeis slowing considerably and is occurring mainly in thecentral part of the sill. The bulk MgO concentrationhas evolved from a straight profile to an ‘s’-shaped profileowing to compaction. It should be noted that, in the modelpresented here, the sill was assumed to be uniformin its initial phenocryst content. Although this is notentirely consistent with the conclusion of Marsh (2004),

we highlight that the new model results would be superim-posed on any more complicated initial distribution of phe-nocrysts. In the lower part of the intrusion, compactionleads to an increase in MgO as solids displace interstitialliquid. In contrast, for reasons discussed in some detail byBoudreau & Philpotts (2002), in the upper part of thesill the solid assemblage is undergoing extensional defor-mation. Briefly, this occurs because the increasing solidifi-cation of interstitial silicate liquid towards the uppercontact decreases the permeability and hence the solidvelocity decreases upward in the upper half of the sill.In effect, solids in the center of the sill can move downfaster than those near the upper margin. The effect isa zone of extension that is maximized at about two-thirdsup in the sill. Although it is beyond the scope of this study,we note that in the Dais intrusion the tensional regionis characterized by the appearance of pronouncedmafic-felsic layering.Finally, it is noted that the migration of exsolved vapor

affects the compaction-driven silicate liquid velocity.In general, the silicate liquid would have a positivevelocity value as the liquid moves upward in responseto solids moving downward (i.e. compacting). However,vapor ascends faster than silicate liquid, hence there isa local negative liquid velocity as a ‘backflow’ of silicateliquid displaces ‘lost’ vapor. This is most evident wherea negative liquid velocity peak is associated with thevapor mass peak in the lower part of the sill, but it isalso present in the upper part of the sill.

CONCLUSIONSApatite compositions and numerical modeling bothsuggest that degassing in a mafic sill can be relativelycomplicated. Like the magma itself, the vapor evolved

Fig. 8. Calculated profiles (from left to right) of: (a) phase mode (wt %); (b) bulk-rock MgO (wt %); (c) mass vapor (wt %); (d) the solid andliquid velocities (cm/day); (e) the deformation of the solid matrix. All are shown at 210 years of model calculation. (See text for additionaldiscussion.)


1381

from a cooling magma can undergo a protracted period ofcompositional evolution. This affects the estimationof volatile abundances of elements such as S, Cl and F.Degassing from the lower margin of the sill can influencestrongly the volatile-element budget of the overlyingmagma column as crystallization proceeds. Variation inhalogen ratios of apatite through the vertical extent ofa sill provide a convenient, if difficult, method for elucidat-ing its volatilization history. The model results agreebroadly with the fluid inclusion record preserved in theStillwater Complex (Hanley et al., 2005).

ACKNOWLEDGEMENTSThe senior author would like to thank Bruce Marsh forhis efforts to make the Field Laboratory Workshop in theMcMurdo Dry Valleys, Antarctica, a great success andfor the chance to participate in the study of these veryinteresting rocks. This paper was significantly improvedby comments from W. Bohrson, D. Geist, J. Hanley andan anonymous reviewer. The work was supported byNational Science Foundation Grant EAR 04-07928.

SUPPLEMENTARY DATASupplementary data for this paper are available atJournal of Petrology online.

REFERENCESAnderko, A. & Pitzer, K. S. (1993). Equation-of-state representation

of phase equilibria and volumetric properties of the systemNaCl^H2O above 573K. Geochimica et Cosmochimica Acta 57,1657^1680.

Banks, D. A.,Yardley, B.W. D., Campbell, A. R. & Jarvis, K. E. (1994).REE composition of an aqueous magmatic fluid: a fluid inclusionstudy from the Capitan Pluton, New Mexico, U.S.A. Chemical

Geology 113, 259^272.Beswick, A. E. & Carmichael, I. S. E. (1978). Constraints on mantle

source compositions imposed by phosphorus and the rare-earthelements. Contributions to Mineralogy and Petrology 67, 317^330.

Bodnar, R. J. Burnham, C.W. & Sterner, S. M. (1985). Synthetic fluidinclusions in natural quartz. III. Determination of phase equilib-rium properties in the system H2O^NaCl to 10008C and 1500 bars.Geochimica et Cosmochimica Acta 49, 1861^1873.

Borg, S. G., DePaolo, D. J. & Smith, B. M. (1990). Isotopic structureand tectonics of the central Transantarctic Mountains. Journal ofGeophysical Research 95, 6647^6667.

Boudreau, A. E. & Kroger, F. J. (1990).Variation in the composition ofapatite through the Merensky cyclic unit in the western BushveldComplex. Economic Geology 85, 737^745.

Boudreau, A. E. (1995). Fluid evolution in layered intrusions: evidencefrom the chemistry of the halogen-bearing minerals.In: Thompson, J. F. H. (ed.) Magmas, Fluids, and Ore Deposits.

Geological Society of Canada Short Course Series 23, 25^46.Boudreau, A. E. (1999). PELEça version of the MELTS software

program for the PC platform. Computers and Geosciences 25, 201^203.

Boudreau, A. E. (2003). IRIDIUNça program to model reaction ofsilicate liquid infiltrating a porous solid assemblage.. Computers andGeosciences 29, 423^429.

Boudreau, A. E. & McCallum, I. S. (1989). Investigations of theStillwater Complex: Part V. Apatites as indicators of evolving fluidcomposition. Contributions to Mineralogy and Petrology 102, 138^153.

Boudreau, A. E. & Philpotts, A. J. (2002). Quantitative modeling ofcompaction in the Holyoke flood basalt flow, Hartford Basin,Connecticut. Contributions to Mineralogy and Petrology 144, 176^184.

Boudreau, A. E., Mathez, E. A. & McCallum, I. S. (1986). Halogengeochemistry of the Stillwater and Bushveld Complexes: evidencefor transport of the platinum-group elements by Cl-rich fluids.Journal of Petrology 27, 967^986.

Boudreau, A. E., Love, C. & Hoatson, D. (1992). Halogen geochemis-try of the Munni Munni Complex and associated intrusions of thePilbara Block, W. Australia. Geochimica Cosmochimica Acta 57,4467^4477.

Boudreau, A. E., Love, C. & Prendergast, M. D. (1995). Halogengeochemistry of the Great Dyke, Zimbabwe. Contributions to

Mineralogy and Petrology 122, 289^300.Boudreau, A. E., Stewart, M. A. & Spivack, A. J. (1997). Stable

chlorine isotopes and the origin of high-Cl magmas of theStillwater Complex, Montana. Geology 25, 791^794.

Brown, G. M. & Peckett, A. (1977). Fluorapatites from the Skaergaardintrusion, East Greenland. Mineralogical Magazine 41, 227^232.

Burnham, C. W. (1967). Hydrothermal fluids at the magmatic stage.In: Barnes, H. L. (ed.) Geochemistry of Hydrothermal Ore Deposits.NewYork: JohnWiley, pp. 34^76.

Candela, P. A. (1986). Toward a thermodynamic model forthe halogens in magmatic systems: an application to melt^vapor^apatite equilibria. Chemical Geology 57, 289^301.

Candela, P. A. (1990) Theoretical constraints on the chemistry of themagmatic aqueous phase. In: Stein, H. J. & Hannah, J. L. (eds)Ore-bearing Granite Systems; Petrogenesis and mineralizing processes,Geological Society of America Special Paper 246, 11^20.

Candela, P. A. & Piccoli, P. M. (1995). Model ore-metal partitioningfrom melts into vapor and vapor-brine mixtures. In: Thompson,J.F.H. (ed.), Magmas, Fluids and Ore Deposits. Mineralogical Societyof Canada Short Course 23, 101^127.

Candela, P. A. & Blevin, P. L. (1995). Do some miarolitic granitespreserve evidence of magmatic volatile phase permeability?Economic Geology 90, 2310^2316.

Carroll, M. R. & Webster, J. D. (1994). Volatiles in magmas.In: Carroll, M. R. & Holloway, J. R. (eds) Volatiles in Magmas.

Mineralogical Society of America, Reviews in Mineralogy 30, 231^279.Cashman, K. V. (2004). Volatile controls on magma ascent and

degassing. In: The State of the Planet: Frontiers and Challenges in

Geophysics. Sparks, R. S. J & Hawkesworth, C. J. (eds) American

Geophysical Union Monograph 150, 109^124.Cawthorn, R. G. (1994). Formation of chlor- and fluor-apatite in

layered intrusions. Mineralogical Magazine 58, 299^306.Cervantes, P. & Wallace, P. (2003). Magma degassing and basaltic

eruption styles: a case study of the 2000 yr B.P. eruption of XitleVolcano, central Mexico. Journal of Volcanology and Geothermal

Research 120, 249^270.Charrier, A. D. & Marsh, B. D. (2005). Sill emplacement dynamics:

experimental textural modeling of a pulsing, cooling, particle-laden magma as applied to the Basement Sill, McMurdo DryValleys, Antarctica. EOS Transactions, American Geophysical Union

86(52),V23A^0686.Chou, I. C. (1987). Phase relations in the system NaCl^KCl^H2O: III,

Solubilities of halite in vapor saturated liquids above 4458C andredetermination of phase equilibrium properties in the system


1382

NaCl^H2O to 10008C and 1500 bars. Geochimica et Cosmochimica Acta51, 1965^1975.

Chou, I.-M., Sterner, S. M. & Pitzer, K. S. (1992). Phase relations inthe system NaCl^KCl^H2O: IV. Differential thermal analysis ofthe sylvite liquidus in the KCl^H2O binary, the liquidus in theNaCl^KCl^H2O ternary, and the solidus in the NaCl^KCl binaryto 2 kb pressure, and a summary of experimental data for thermo-dynamic-PTX analysis of solid^liquid equilibria at elevated P^Tconditions. Geochimica et Cosmochimica Acta 56, 2281^2293.

Cline, J. S. & Bodnar, R. J. (1994). Direct evolution of brine froma crystallizing silicic melt at the Questa, New Mexico, molybde-num deposit. Economic Geology 89, 1780^1802.

Dingwell, D. B. & Scarfe, C. M. (1983). Major element partitioning inthe system haplogranite-HF-H2O: implications for leucogranitesand high-silica rhyolites. EOS 64, 312.

Duan, Z., Moller, N. &Weare, J. H. (1996). A general equation of statefor supercritical fluid mixtures and molecular dynamics simulationof mixture PVTX properties. Geochimica et Cosmochimica Acta 60,1209^1216.

Eggler, D. H., Mysen, B. O. & Seitz, M. G. (1974). The solubility ofCO2 in silicate liquid and crystals. Carnegie Institute of Washington

Yearbook 73, 226^228.Elliot, D. H. & Fleming, T. H. (2004). Occurrence and dispersal of

magmas in the Jurassic Ferrar Large Igneous Province. GondwanaResearch 7, 223^237.

Flynn, R. T. & Burnham, C. W. (1978). An experimental determina-tion of rare earth partition coefficients between a chloridecontaining vapor phase and silicate melts. Geochimica

et Cosmochimica Acta 42, 685^701.Frondel, J.W. (1964).Variations of some rare earth elements in allanite.

American Mineralogist 49, 1159^1177.Geist, D., Boudreau, A., Garcia, M., Harpp, K., Mathez, E. &

Marsh, B. (2005). Origin of mafic pegmatoids in the DaisIntrusion, Wright Valley, Antarctica. EOS Transactions, American

Geophysical Union 86(52), Fall Meeting Supplement, AbstractV14C-07.

Ghiorso, M. S. & Sack, R. O. (1995). Chemical mass transfer inmagmatic processes. IV. A revised and internally consistentthermodynamic model for the interpretation and extrapolation ofliquid^solid equilibria in magmatic systems at elevatedtemperatures and pressures. Contributions to Mineralogy and Petrology

119, 197^212.Ghiorso, M. S., Hirschman, M. & Sack, R. O. (1994). MELTS: soft-

ware for thermodynamic modeling of magmatic systems. EOS

Transactions, American Geophysical Union 75, 571.Gonnerman, H. M. & Manga, M. (2005). Nonequilibrium magma

degassing: results from modeling of the ca. 1340A.D. eruption ofMono Craters, California. Earth and Planetary Science Letters 238,1^16.

Gunn, B. (1966). Modal and element variation in Antarctic tholeiites.Geochimica et Cosmochimica Acta 30, 881^920.

Gunn, B. & Warren, G. (1962). Geology of Victoria Land between the

Mawson and Mulock Glaciers, Antarctica. New Zealand Geological Survey

Bulletin 71, 157 pp.Hamilton,W. B. (1965). Diabase Sheets of theTaylor Glacier Region,Victoria

Land, Antarctica. US Geological Survey, Professional Papers 456-B, 71pp. þmap.

Hammer, J. E. (2006). Interpreting inclusive evidence. Nature 439,26^27.

Hanley, J., Mungall, J. E., Spooner, E. T. C. & Pettke,T. (2005). Fluidand melt inclusion evidence for platinum-group element transportby high salinity fluids and halide melts below the J-M reef,Stillwater Complex, Montana, U.S.A. In: T. O. To« ma« nen &

Alapieti, T. T. (eds) 10th International Platinum Sumposium, Extended

Abstracts. Geological Survey of Finland, Espoo, Finland, 94^97.Haughton, D. R., Roeder, P. L. & Skinner, B. J. (1974). Solubility of

sulfur in mafic magmas. Economic Geology 69, 451^467.Heimann, A., Fleming, T. H., Elliot, D. H. & Foland, K. A. (1994).

A short interval of Jurassic continental food basalt volcanism inAntarctica as demonstrated by 40Ar/39Ar geochronology. Earth andPlanetary Science Letters 121, 19^41.

Helgeson, H. C. (1964). Complexing and Hydrothermal Ore Deposits.NewYork: Macmillan.

Hersum, T., Marsh, B., Simon, A. (submitted). Dolerite partial melt-ing of granitic country rock, melt segregation, and re-injection asdikes, McMurdo DryValleys, Antarctica. Journal of Petrology.

Huppert, H. E. & Woods, A.W. (2002).The role of volatiles in magmachamber dynamics. Nature 420, 493^495.

Knight, K. B. & Renne, P. R. (2005). Evidence for extended (5^10Ma) emplacement of Ferrar Dolerite from 40Ar^39Ar geochronol-ogy. EOSTransactions, American Geophysical Union 86(52),V23A^0684.

Kress, V. C. & Carmichael, I. S. E. (1991). The compressibility ofsilicate liquids containing Fe2O3 and the effect of composition,temperature, oxygen fugacity and pressure on their redox states.Contributions to Mineralogy and Petrology 108, 82^92.

Lowenstern, J. B. (1994). Chlorine, fluid immiscibility, and degassingin peralkaline magmas from Pantelleria, Italy. American Mineralogist

79, 353^369.Lowenstern, J. B. (2000). The contrasting behavior of two magmatic

volatiles: Cl and CO2. Journal of Geochemical Exploration 69^70,287^290.

Marsh, B. D. (2004). A magmatic mush column Rosetta Stone: theMcMurdo Dry Valleys of Antarctica. EOS Transactions, American

Geophysical Union 85(47), 497^502.Marsh, B. D. & Zieg, M. J. (2004). Preliminary geologic map, Ferrar

Dolerites, DryValleys, Antarctica. In: Marsh, B. D. (ed.) Magmatic

Field LaboratoryWorkshop: McMurdo DryValleys, Antarctica. Baltimore,MD: Johns Hopkins University, unpublished workshop handbook,unpaginated.

Mathez, E. A. & Webster, J. D. (2005). Partitioning behavior of chlor-ine and fluorine in the system apatite^silicate melt^fluid. Geochimicaet Cosmochimica Acta 69, 1275^1286.

Mattey, D. P. (1991). Carbon dioxide solubility and carbon isotopefractionation in basaltic melt. Geochimica et Cosmochimica Acta 55(11),3467^3473.

Mavrogenes, J. A. & O’Neill, H. St. C. (1999). The relative effects ofpressure, temperature and oxygen fugacity on the solubility of sul-fide in mafic magmas. Geochimica et Cosmochimica Acta 63, 1173^1180.

McKelvey, B. C. & Webb, P. N. (1962). Geological investigations insouthernVictoria Land, Antarctica: Part 3, Geology of theWrightValley. New ZealandJournal of Geology and Geophysics 5, 143^162.

Metrich, N. & Rutherford, M. J. (1992). Experimental study of chlor-ine behavior in hydrous silicic melts. Geochimica et Cosmochimica Acta57, 607^616.

Morton, R. D. & Catanzaro, E. J. (1974). Stable chlorine isotope abun-dances in apatite from Odega‡rdens Verk, Norway. Norsk Geologisk

Tidsskrift 44, 307^313.O’Neill, H. St. C. & Mavrogenes, J. A. (2002). The sulfide capacity

and the sulfur content at sulfide saturation of silicate melts at14008C and 1bar. Journal of Petrology 43, 1049^1087.

Papale, P. (1999). Modeling the solubility of a two-componentH2OþCO2 fluid in silicate liquids. American Mineralogist 84,477^492.

Patin‹ o Douce, A. E. & Roden, M. (2006). Apatite as a probe of halo-gen and water fugacities in the terrestrial planets. Geochimica

et Cosmochimica Acta 70, 3173^3196.


1383

Petford, N., Jerram, D. & Davidson, J. (2005). Slurry flow and struc-tures formation in a magma mush: the Basement Sill, McMurdoDry Valleys, Antarctica. EOS Transactions, American Geophysical

Union 86(52),V13H^05.Piccoli, P. M. & Candela, P. A. (1994). Apatite in felsic rocks: a model

for the estimation of initial halogen concentrations in the BishopTuff and Tuolumne intrusive suite magmas. American Journal of

Science 294, 92^135.Reed M. J. (1995) Distribution of rare earth elements between aqu-

eous fluid and granitic melt. PhD dissertation, University ofMaryland College Park, p. 250.

Reed, M. J., Candela, P. A. & Piccoli, P. M. (2000).The distribution ofrare earth elements between monzogranitic melt and the aqueousvolatile phase in experimental investigations at 8008C and200MPa. Contributions to Mineralogy and Petrology 140, 251^262.

Roeder, P. L. (1985). Electron-microprobe analysis of minerals forrare-earth elements: use of calculated peak-overlap corrections.Canadian Mineralogist 23, 263^271.

Roedder, E. (ed.) (1984). Fluid Inclusions. Mineralogical Society of America,

Reviews in Mineralogy 12.Roedder, E. (1992). Fluid inclusion evidence for immiscibility in mag-

matic differentiation. Geochimica et Cosmochimica Acta 56, 5^20.Roggensack, K., Hervig, R. L., McKnight, S. B. & Williams, S. N.

(1997). Explosive basaltic volcanism from Cerro Negro Volcano:influence of volatiles on eruptive style. Science 277, 1639^1642.

Saal, A. E., Hauri, E. H., Langmuir, C. H. & Perfit, M. R. (2002).Vapour undersaturation in primitive mid-ocean-ridge basalt andthe volatile content of Earth’s upper mantle. Nature 419, 451^455.

Sakagawa, M. (1989). Experimental study of the partitioning orrare earth elements between granitic melts and aqueous fluids.MSc thesis, Tsukuba University, Institute of Geoscience, Ibaraki,Japan.

Shi, P. (1992). Fluid fugacities and phase equilibria in the Fe^Si^OH^Ssystem. American Mineralogist 77, 1050^1066.

Shinohara, H., Iiyama, J. T. & Matsuo, S. (1989). Partition of chlorinecompounds between silicate melt and hydrothermal solutions 1.Partition of NaCl^KCl. Geochimica et Cosmochimica Acta 53,2617^2630.

Shirley, D. N. (1986). Compaction of igneous cumulates. Journal ofGeology 94, 795^809.

Simon, A. C. & Marsh, B. (2005). Pyroxene thermometry in the OPXTongue of the Basement Sill, McMurdo Dry Valleys, Antarctica.EOS Transactions, American Geophysical Union 86(52), Fall MeetingSupplement, Abstract V13H-08.

Sourirajan, S. & Kennedy, G. C. (1962).The system H2O^NaCl at ele-vated temperatures and pressures. American Journal of Science 260,115^141.

Speer, J. A. (ed.) (1984) Micas in Igneous Rocks. Mineralogical Society of

American, Reviews in Mineralogy 13.Sterner, S. M., Chou, I.-M., Downs, R.T. & Pitzer, K. S. (1992). Phase

relations in the system NaCl^KCl^H2O: V. Thermodynamic-PTXanalysis of solid^liquid equilibria at high temperatures and pres-sures. Geochimica et Cosmochimica Acta 56, 2295^2309.

Stormer, J. C., Pierson, M. L. & Tacker, R. C. (1993). Variation of Fand Cl X-ray intensity due to anisotropic diffusion in apatiteduring electron microprobe analysis. American Mineralogist 78,641^648.

Tacker, R. C. & Stormer, J. C. (1989). A thermodynamic model forapatite solid solutions, applicable to high-temperature geologicproblems. American Mineralogist 74, 877^888.

Tacker, R. C. & Stormer, J. C. (1993). Thermodynamics of mixingof liquids in the system Ca3(PO4)2^CaF2^CaCl2^Ca(OH)2.Geochimica et Cosmochimica Acta 57, 4663^4676.

Taylor, R. P. & Fryer, B. J. (1980). Multiple-stage hydrothermal altera-tion in porphyry copper systems in northernTurkey: the temporalinterplay of potassic, propylitic, and phyllic fluids. CanadianJournalof Earth Sciences 17, 901^926.

Volfinger, M., Roberts, J.-L., Vielzeuf, D. & Neiva, A. M. R. (1985).Structural controls of the chlorine content of OH-bearing silicates(micas and amphiboles). Geochimica et Cosmochimica Acta 49, 37^48.

Wager, L. R. & Brown, G. M. (1967). Layered Igneous Rocks. Edinburgh:Oliver & Boyd, 588 pp.

Wallace, P. & Carmichael, I. S. E. (1992). Sulfur in basic magmas.Geochimica et Cosmochimica Acta 56, 1863^1874.

Wallace, P. J. (2003). From mantle to atmosphere: magma degassing,explosive eruptions, and volcanic volatile budgets. In: De Vivo, B.& Bodnar, R. J. (eds) Melt Inclusions in Volcanic Systems: Methods,

Applications and Problems. Developments inVolcanology 5, 105^127.Webster, J. D. & Holloway, J. R. (1990). Partitioning of F and Cl

between magmatic hydrothermal fluids and highly evolved grani-tic magmas. In: Stein, H. J. & Hannah, J. L. (eds) Ore-bearing

Granite Systems: Petrogenesis and Mineralizing Processes. GeologicalSociety of America Special Paper 246, 21^34.

Webster, J. D. (2004). The exsolution of magmatic hydrosaline melts.Chemical Geology 21, 33^48.

Webster, J. D. & DeVivo, B. (2002). Experimental and modeled solubi-lities of chlorine in aluminosilicate melts, consequences of magmaevolution, and implications for exsolution of hydrous chloridemelt at Mt. Somma^Vesuvius, Italy. American Mineralogist 87,1046^1061.

Webster, J. D., Kinzler, R. J. & Mathez, E. A. (1999). Chloride andwater solubility in basalt and andesite liquids and implications formagmatic degassing. Geochimica et Cosmochimica Acta 63, 729^738.

Webster, J. D.,Thomas, R., Fo« rster, H.-J., Seltmann, R. & Tappen, C.(2004). Geochemical evolution of halogen-enriched, granitemagmas and mineralizing fluids of the Zinnwald tin^tungstenmining district, Erzgebirge, Germany. Mineralium Deposita 39,452^472.

Wendlandt, R. F. (1982). Sulfide saturation of basalt and andesite meltsat high pressures and temperatures. American Mineralogist 67,877^885.

Willmore, C. C., Boudreau, A. E. & Kruger, F. J. (2000). The halogengeochemistry of the Bushveld Complex, Republic of South Africa:implications for chalcophile element distribution in the Lower andCritical zones. Journal of Petrology 41, 1517^1539.

Yang, K. & Scott, S. D. (2002). Magmatic degassing of volatiles andore metals into a hydrothermal system on the modern sea floor ofthe eastern Manus back-arc basin, western Pacific. Economic Geology97, 1079^1100.

Zavala, K. & Marsh, B. D. (2001). On the filling process formingsilicic segregations. EOS Transactions, American Geophysical Union

82(20), Fall Meeting Supplement, Abstract V41A-05.Zavala, K. & Marsh, B. D. (2002). Fluid dynamic experiments bear-

ing on the filling of silicic segregations in basaltic sills. EOS

Transactions, American Geophysical Union 83(19), Fall MeetingSupplement, Abstract V21B-04.

Zieg, M. J. & Forsha, C. J. (2005). Textural gradients in Antarctic andCanadian sills: evidence for episodic filling history. EOS

Transactions, American Geophysical Union 86(52), Fall MeetingSupplement, Abstract V14C-04.

APPENDIXThe modeling described in this paper uses an updatedversion of the IRIDIUM program (Boudreau, 2003).


1384

This program uses a simplified version of the MELTSprogram (Ghiorso et al., 1994; Ghiorso & Sack, 1995) forphase equilibrium determinations that are coupled withmass and heat transfer models. The main simplification

from MELTS is the use of simpler solid solution modelsfor some of the minerals (Boudreau, 1999). The updatesdescribed here include the addition of CO2 and FeS assystem components, which allows the model to exsolve

Fig. A1. Composition of vapor phase (in mole fractions) as a func-tion of log f(O2). Initial bulk composition is MORB liquid with 2�0wt % H2O, 0�15 wt % CO2 and 0�1 wt % S, equilibrated at 12258C,1000 bars. Log f(O2) varies between QFMþ 3 and QFM^ 3 (whereQFM is the quartz^fayalite^magnetite buffer).

Fig. A2. Volatile concentrations in (a) vapor and (b) silicate liquidfor fractional degassing. Initial MORB liquid with 2�0 wt % H2O,0�15 wt % CO2 and 0�1wt % S at 12258C. Liquid undergoes degassingwith loss of vapor while undergoing an isothermal pressure changefrom 2001 to 1bar.

Fig. A3. Calculated sulfur concentration at sulfide saturation inliquids of basaltic composition as a function of FeO total concentra-tion at 12008C, 1bar and f(O2)¼Ni^NiO. Samples taken fromHaughton et al. (1974).

Fig. A4. Sulfur concentration in silicate liquid at sulfide saturation asa function of (a) log f(O2) and (b) log f(S2) at 12008C and 1bar.Sample 8F1 from Haughton et al. (1974); FeOtotal is 26�2 wt %.


1385

both a CÔ^H^S fluid and to precipitate sulfides andC-bearing phases from the liquid.CO2 solubility in IRIDIUM essentially incorporates themodel of Papale (1999). It has been modified as follows: theCO2ôxide liquid interaction parameters have been recali-brated to the liquid components and interaction param-eters used in MELTS (the Papale model used those of theprogram SILMIN, an earlier version of MELTS).The ther-modynamic properties of pure CO2 and all other gasspecies are calculated by the general equation of state forsupercritical gases of Duan et al. (1996). Gases are otherwiseassumed to mix ideally. Possible gas species include H2O,CO2, CO, CH4, H2S, and SO2; however, for the modelingdonehere onlyH2O,CO2 andH2Sare considered.Graphite(C), calcite and dolomite are possible precipitating solids.Finally, because of its low solubility at low to moderatepressures, the model treats the CO2 liquid component asnot having any effect on the other liquid species other thanactingas a dilutant to themole fraction,X.The program uses liquid FeS (liquid troilite) for thesilicate liquid component for sulfur. Sulfur fugacity is cal-culated by the method of Wallace & Carmichael (1992).Immiscible sulfide liquid is assumed to be an ideal solutionof two components, FeS and FeO. Other possible sulfidephases that can precipitate are pyrrhotite, pyrite andtroilite.FeSôxide liquid interaction parameters were calibrated(with some arbitrary adjustment because of uncertainsulfide and silicate liquid compositions in the experi-mental data) from the data of Haughton et al. (1974),Wendlandt (1982), Mavrogenes & O’Neill (1999) andO’Neill & Mavrogenes (2002). Because these studies invari-ably tabulated total iron only, FeO and Fe2O3 are calcu-lated by the method of Kress & Carmichael (1991) based

on experimental f(O2), temperature and bulk composition.These experimental studies are mostly of basaltîntermediate composition, but the lower solubility calcu-lated for more silicic compositions appears reasonable.Above �1 GPa, sulfide saturation values are significantlylower than expected. As for CO2, the interaction para-meters do not affect the calculation of the chemicalpotential of other liquid components other than aminor effect on mole fractions.Examples of program output are shown in Figs A1Â4.Figure A1 shows an example of gas speciation as a functionof f(O2) for a gas-saturated mid-ocean ridge basalt(MORB) magma. It is necessary to note an importantcaveat regarding oxygen and sulfur fugacity as calculatedby MELTS/IRIDIUM; that is, the liquid interactionparameters used by MELTS were not explicitycalibrated against oxygen fugacity. Because of this,when f(O2) is calculated from reactions between liquidcomponents used by MELTS, it is not the same as thatcalculated by the empirical Kress & Carmichael (1991)model incorporated into MELTS. Nonetheless, gas specia-tion for S-bearing species is broadly consistent with themodel results of Shi (1992), for example.Figure A2 shows the changes in both gas and magma com-position as a vapor-saturated MORB magma undergoespolybaric decompressional fractional degassing from 2001to 1bar. This can be compared with similar plots of Papale(1999).Figure A3 shows the S concentration at sulfide saturation(SCSS) as a function of total iron concentration in theliquid for a selection of liquids taken from Haughton et al.(1974), and Fig. A4 shows the SCSS as a function of sulfurand oxygen fugacity, again for a sample from Haughtonet al. (1974).


1386

crystallization and degassing in the basement sill ...geoscience.unlv.edu/files/boudreau and simon...

Documents