Eur J Mineral2008 20 905ndash916Published online June 2008 This paper is dedicated to

the memory of Werner Schreyer

Lithium and its isotopes in tourmaline as indicators of the crystallizationprocess in the San Diego County pegmatites California USA

Jennifer S MALONEY1 4 Peter I NABELEK1 Mona-Liza C SIRBESCU2 and Ralf HALAMA3 5

1 Department of Geological Sciences University of Missouri Columbia MO 65211 USACorresponding author e-mail nabelekpmissouriedu

2 Department of Geology Central Michigan University Mt Pleasant MI 48859 USA3 Department of Geology University of Maryland College Park MD 20742 USA

4 Current address Newfield Exploration Company 363 N Sam Houston Parkway Houston TX 77060 USA5 Current address Institut fuumlr Geowissenschaften Christian-Albrechts-Universitaumlt Kiel 24098 Kiel Germany

Abstract In the lithium-cesium-tantalum-type pegmatite dikes of San Diego County California USA tourmaline is the mainreservoir for Li except in the cores and the pockets of the dikes where other Li-bearing minerals also occur Tourmaline fromthree subhorizontal dikes was analyzed for bulk Li concentrations and Li isotope ratios The bottom portion of each dike includesrhythmically layered aplite called line-rock Above the aplite is the lower pegmatite zone that crystallized upward whereas thehanging pegmatite zone crystallized downward The lower and hanging pegmatite zones are joined at the core zone Pockets thatwere once fluid-filled occur in the core zone

Tourmaline in the line-rocks and the upper border zones has 22ndash70 ppm Li and in the pegmatite zones 53ndash450 ppm Li Largetourmaline blades in the cores have 174ndash663 ppm Li Elbaite rims on prismatic tourmaline in the pockets have up to 5075 ppm LiThe progressive enrichment in Li from the wall-zones to the pockets is attributed to inward fractional crystallization of the dikesThe line-rock in each dike appears to have crystallized until the melt reached fluid saturation at which point the melt and the fluidbegan to unmix to form the pegmatite zones and the pockets The estimated initial Li concentration in the magma that producedthe dikes is sim 630 ppm At this low concentration Li has had much smaller effect on crystallization of the dikes than H2Oδ7Li in tourmaline in the line-rocks the cores and the pockets ranges from +112 to +161 with no systematic difference

between these textural zones However in radial tourmalines δ7Li is gt 19 The very elevated δ7Li may reflect Li isotopefractionation between the melt and the exsolving fluid at the time of crystallization of these tourmalines with 7Li preferringthe more strongly-bonded occupancy in the silicate melt over a hydrated ion occupancy in the fluid Alternatively the elevatedδ7Li may also have been caused by preferential accumulation of the slower-diffusing 7Li ahead of the rapidly-growing radialtourmalines The overall elevated δ7Li values of the dikes may have been acquired by Li isotope exchange with wall-rocks duringpassage of the pegmatite melts from their sources

Key-words tourmaline pegmatites lithium isotopes fractionation

Introduction

A hallmark of granitic pegmatites are their disequilibriumfeatures such as large variations in crystal size large radialcrystal splays and strong mineralogical textural and chem-ical zoning across dikes The zoning may including rhyth-mically crystallized aplitic border zones pegmatite zonescores and pockets The pockets which typically occur inthe cores of dikes represent the space that was once filledby an accumulated fluid The proposed processes for crys-tallization of granitic pegmatites range from closed sys-tem fractional crystallization of a hydrous melt (Jahns ampBurnham 1969) to rapid cooling to a glass with subsequentdevelopment of the pegmatitic texture by constitutionalzone refinement (Morgan amp London 1999) Although open

system processes have also been proposed to explain thecharacteristics of granitic pegmatites the prevalent viewis closed-system crystallization of fluxed melts (Cernyacute1991)

Water and other fluxing components such as Li and Bare central to most crystallization models of granitic peg-matites and their roles must be understood especially inview of the recognition that pegmatites can crystallize attemperatures lt 400 C (London 1986a Thomas et al1988 Morgan amp London 1999 Sirbescu amp Nabelek2003a b) The fluxing components may play a critical rolein permitting crystallization of pegmatitic melts at very lowtemperatures be it by unmixing involving production ofa fluxed F B and P-rich melt (Thomas amp Klemm 1997Thomas et al 2000) or by permitting rapid crystal growth

0935-1221080020-1823 $ 540DOI 1011270935-122120080020-1823 ccopy 2008 E Schweizerbartrsquosche Verlagsbuchhandlung D-70176 Stuttgart

906 JS Maloney et al

rates in highly undercooled melts (Webber et al 1999Nabelek 2007 Sirbescu et al 2008)

This study was inspired in part by the work of WernerSchreyer on the crystal-chemistry of tourmaline Hiswork has demonstrated the value of tourmaline as anindicator of the petrogenesis of igneous rocks includingpegmatites (eg Schreyer et al 2000 Kalt et al 2001)In the present study concentrations and isotopic ratiosof Li in tourmaline are used to constrain the role of Liin the crystallization of three pegmatite dikes in the SanDiego County California USA These dikes have beenstudied by many workers including as a foundation forthe classic pegmatite crystallization model of Jahns ampBurnham (1969) In the lithium-cesium-tantalum (LCT)family of pegmatites to which the San Diego Countydikes belong Li occurs at concentrations high enoughto cause crystallization of Li minerals including elbaitetourmaline [Na(Li15Al15)Al6B3Si6O27(OH)3(OH F)]spodumene [LiAlSi2O6] amblygonite-montebrasite[(Li Na)AlPO4(F OH)] and lepidolite along thepolylithionite [KLi2AlSi4O10(F OH)2] ndash trilithionite[KLi15Al15AlSi3O10(F OH)2] join (Cernyacute amp Ercit 2005)Because Li can substitute into tourmaline which is ubiq-uitous in all portions of the San Diego pegmatite dikestourmaline was used here as a proxy for determining thebehavior of Li and to assess its importance for fluxinggranite melts Isotope ratios of Li can be a mirror onmagma differentiation because the fractionation of Li iso-topes in a multiphase system is to a large extent controlledby its coordination in coexisting phases andor kinetics(Wenger amp Armbruster 1991 Richter et al 2003 Wunderet al 2006 2007)

Geologic context and samples

The pegmatites of San Diego County are located within thenorthwest trending subduction-related Peninsular RangesBatholith (PRB) that stretches from the San Jacinto Moun-tains of California into Baja California Mexico (Fig 1)The batholith is divided into the western and eastern zones(Walawender et al 1990 Fisher 2002 Todd et al 2003)The western zone is comprised of I-type gabbros quartzdiorites tonalites granodiorites and monzogranites rang-ing in age from 120 to 105 Ma in age The more fel-sic eastern zone lacks gabbros and is dominated by largeconcentrically-zoned I- and S-type tonalites and monzo-granites whose intrusion began at sim 105 Ma when mag-matism in the western zone stopped due to a change inplate dynamics Some of the intrusions include the 100and 89 Ma La Posta-type plutons that straddle to east-westboundary of the PRB (Walawender et al 1990 Symonset al 2003) La Posta-type plutons are zoned inwardfrom hornblende tonalite to muscovite-bearing monzogran-ite The pegmatite field is broadly contemporaneous withthe La Posta-type plutons and also straddles the east-westboundary The pegmatites occur as suites of parallel to sub-horizontal dikes mostly within the plutons and occasion-ally within intervening schists The dikes were emplaced at2ndash3 kbar based on densities of fluid inclusions trapped in

Fig 1 Regional sketch showing locations of the Cryo-Genie Lit-tle Three and Himalaya pegmatite dikes in the San Diego CountyCalifornia Inset shows location of the Peninsular Ranges Batholith(stippled) The black rectangular shape is San Diego County

pockets (London 1986b) The host plutons appear to havebeen brittle when the dikes were emplaced

Tourmaline was collected from three pegmatite dike sys-tems ndash the Cryo-Genie the Little Three and the Himalaya(Fig 1) Dikes in all three systems exhibit similar zoning inmineralogy texture and chemistry (Jahns amp Tuttle 1963Jahns 1979) A typical dike has a lower border zone that ischaracterized by rhythmically-layered aplite with oscillat-ing changes in mineralogy from Mg-Fe-poor layers domi-nated by quartz and albite to Mg-Fe-rich layers dominatedby tourmaline and sometimes garnet This layered apliteis colloquially called ldquoline-rockrdquo Above the line-rock isthe lower pegmatite zone with upward growth direction ofminerals The hanging part of the dike is characterized by athin border zone which occasionally is also layered Belowthis border zone is the hanging pegmatite with large euhe-dral K-feldspar crystals and then graphic quartz-feldsparintergrowths Between the hanging and lower pegmatites isthe coarse-grained core zone The core zone includes pock-ets that have variable size and occurrence along the dikersquosstrike Aside from gem-quality crystals the pockets are of-ten partially filled with clays that suggest low-temperaturealteration of core and pocket minerals by magmatic flu-ids (Foord et al 1986) When pockets are absent the de-marcation between the lower and hanging portions of thedike is shown by opposite growth directions of mineralsand occurrences of massive quartz large amount of micaand large tourmaline blades In this paper both a coarse-grained core and a wall of a pocket are referred to as theldquocore zonerdquo

The main Cryo-Genie dike intruded into sillimanite-grade metasedimentary rocks and a portion of the localLa Posta-type granite Field relationships with dated gran-ites suggest that the dike was emplaced between 98 and89 Ma (Kampf et al 2003) The dike is 2ndash4 m thick andcan be traced for more than 200 m The analyzed sampleswere collected at the Green Ledge surface excavations andaround the Payday pocket in underground excavations (Ta-ble 1 Fig 2a b) At Green Ledge only the upper portion of

Lithium and its isotopes in pegmatite tourmalines 907

Table 1 Tourmaline locations shapes and Li concentrations and isotope ratios

Sample Zone Tourmaline shape Li (ppm) δ7Li ()CG-1a middle hanging pegmatite radial 53 192CG-1c core zone large blade 174CG-1e core zone large blade 138CG-1i black pocket prism 273 159CG-1i green pocket prism 5075 147CG-1n pocket prism 416 112CG-3e core zone large blade 265 161CG-3i core zone large blade 663 112CG-3k middle hanging pegmatite radial 421CG-5b top border short blade 22CG-5d middle hanging pegmatite radial 93LT-1c core zone large blade 517 134LT-3f bottom line rock small prismatic 56 151LT-3g bottom line rock small prismatic 66LT-3i bottom line rock small prismatic 70 123LT-4c core zone large blade 268 148LT-5c bottom pegmatite radial 452 229LT-5d bottom pegmatite radial 113 191LT-5e bottom pegmatite small prismatic 135 142HM-2C black pocket prism 641 159HM-2c green pocket prism 1456 137HM-4C pocket prism 954 159

Fig 2 Sketches showing sample locations at the Cryo-Genie pegmatite dike (a) Green Ledge location on the surface and (b) Payday pocketunderground

908 JS Maloney et al

the dike from the upper contact to the core zone includinga pocket is exposed CG-5b is a short euhedral tourmalinefrom the border of the hanging pegmatite CG-1a and CG-5d are radially-grown tourmaline needles from the hangingpegmatite and CG-1c and CG-1e are large euhedral tour-malines from the core zone CG-1i and CG-1n are tourma-line prisms that grew into the pocket CG-1i has a schorlcore and green elbaite rim CG-3e and CG-3i are schorlblades from core zone around the Payday pocket whereasCG-3k is radial schorl from the pegmatite above the pocket

The Little Three dike system intrudes the Green Valleytonalite-gabbro of the western zone of the PRB (Stern et al1986) The system has five dikes including the Little Threemain dike and the Spaulding dike that were sampled forthis study The dikes vary between 1 and 2 m in thick-ness along strike The relative thickness of line-rock andpegmatite zones also varies Most of the analyzed samplescome from the Little Three main dike near the mine en-trance on Topaz Ledge (Fig 3) The bottom boundary ofthe dike with the country rock is a homogeneous aplite withline-rock above it The line rock terminates sharply againstthe lower pegmatite zone which contains small prismaticor bladed tourmaline within graphic intergrowths of quartzand feldspar

Samples LT-3f LT-3g and LT-3i come from the line-rock(Fig 3a) Samples LT-5c and 5d are radial tourmalines thatgrew in the lower pegmatite that cuts through the line rockalong strike from location 3 (Fig 3b) LT-5e is a prismatictourmaline in a smaller dike cutting through the line rockSample LT-4c is a large tourmaline blade from the corezone around a several-meter size pocket that occurs abovethe mine entrance Sample LT-1c is a similar tourmalinefrom the core zone of the Spaulding dike

The Himalaya dike system includes two subparallel dikeshosted by the San Marcos gabbro (Fisher et al 1999Webber et al 1999) The two dikes are separated by 3ndash10 m along the 915 m of their exposed length but convergeat the San Diego mine where the samples were collectedThe Himalaya dikes were emplaced sim 100 my ago asdated by fission track and K-Ar methods (Foord 1976)Samples HM-2c and HM-4c come from pockets in the un-derground workings of the San Diego mine TourmalineHM-2c includes a schorl core and a green elbaite rim whileHM-4c is exclusively schorl

The Cryo-Genie samples best represent a cross-sectionacross the hanging portion of a pegmatite dike while theLittle Three samples best represent the lower portion of adike The Himalaya samples were analyzed primarily to de-termine if there is a regional variability in Li isotopic com-positions

Methods

Bulk Li concentrations

Tourmaline is extremely difficult to dissolve in acids andtherefore we have developed a flux procedure that ac-complishes complete dissolution of ground-up tourmalineBlack cores and green rims of zoned pocket tourmalines

Fig 3 Sketches showing sample locations at the Little Three peg-matite dike Parts a) and b) are along strike on the main dike

were prepared separately When samples had only smallcrystals multiple grains were ground together 100 mgof each tourmaline powder were mixed with 400 mg ofground K2CO3 flux in a zirconia crucible and placed into afurnace at 500 C The temperature was then increased to900 C for 15 min after which the furnace was turned offThe crucibles were left in the furnace overnight to slowlycool to prevent cracking Each fluxed sample was first cen-trifuged in 15 ml of 10 HNO3 for four minutes Theliquid was then decanted into a Teflon beaker and stirredon a hot plate The remaining solid residue was again cen-trifuged in 15 ml of 10 HNO3 Any residue that was stillleft was dissolved in 3 ml of concentrated HNO3 and wasthen added to the already dissolved tourmaline and stirredfor additional 15 minutes on low heat The sample was thenbrought up to 50 ml in a volumetric flask with 10 HNO3In several samples a brown amorphous precipitate formedafter several days After the liquids were decanted the pre-cipitates were dissolved in a HF-HNO3 mixture and thendiluted to 50 ml with distilled water Analysis of these so-lutions revealed no Li

Lithium concentrations (Table 1) were analyzed using thePerkin-Elmer Optima 3300 Inductively Coupled Plasma

Lithium and its isotopes in pegmatite tourmalines 909

Optical Emission Spectrometer (ICP-OES) at the Univer-sity of Missouri Synthetic standards with 005 010 1and 5 ppm Li were prepared using blank K2CO3 flux so-lutions to create the calibration curves for Li Instrumentaldrift was accounted for by analyzing the prepared standardsat regular intervals during the analysis and aliquots of thestandard solutions were used as check standards Lithiumconcentrations were obtained from the strong 610362 nmemission line

Li isotope ratios

Analysis of Li isotope ratios (Table 1) was carried out ondried aliquots of solutions prepared for ICP-OES analy-sis The dried samples were prepared and analyzed at theGeochemistry Laboratory of the University of Maryland-College Park The preparation followed the procedure out-lined in Rudnick et al (2004) and Teng et al (2004) whichis based on the three-column procedure of Moriguti ampNakumura (1998) Because NaLi ratio in solution largerthan sim 5 may cause instability in the analysis the NaLiratio of each sample was determined semi-quantitativelyprior to analysis and excess Na was stripped-off by addi-tional column purification Measurements were done on aNu Plasma Multicollector Inductively Couple Plasma MassSpectrometer (MC-ICP-MS) Each analysis consisted oftwo blocks of twenty individual measurements Each anal-ysis was bracketed by the measurement of a 100 ppb L-SVEC standard The 7Li6Li ratio in L-SVEC for each2 times 20 measurements had an average 2σ of the mean 0003 The external precision better than 1 Two otherLi isotope standards IRMM-016 (Qi et al 1997) and thein-house standard UMD-1 were routinely analyzed dur-ing each analytical session The results for both (IRMM-016 +02 plusmn 04 UMD-1 +543 plusmn 02) agree well withpublished results (IRMM-016 ndash01 plusmn 02 to +02 plusmn 08UMD-1 +547 plusmn 10 Rudnick et al 2004 Teng et al2004 Halama et al 2007) Two USGS rock standardsBHVO-1 (+42 ) and QLO-1 (+66 ) were analyzedfor quality-control purposes The value for BHVO-1 waswithin the uncertainty of previously published results (+43to +58 Bouman et al 2004 Rudnick et al 2004) and thevalue for QLO-1 was within the range of previous analy-ses at the University of Maryland (+56 to +68 Halamaunpublished data)

Results

Li concentrations

Li concentrations in the tourmalines (Table 1) are shown inFig 4a They are plotted in terms of the zones in which thetourmalines grew There appear to be no systematic differ-ences in Li concentrations between corresponding zones ofthe different dikes based on the available data in spite ofthe spatial and probably some temporal separation of thedikes although with denser data sets for all three dikessome systematic differences could appear Nevertheless

Fig 4 (a) Li concentrations and (b) Li isotopic ratios in tourma-lines in three pegmatite dikes Data are grouped according to tex-tural zones from which samples came from Schorl tourmalines areshown by black symbols and elbaite rims on schorl by white sym-bols Two core-rim pairs are shown by connecting lines

the similar ranges in Li concentrations in the pegmatite andthe core zones of both the Cryo-Genie and the Little Threedikes for example point to a common petrogenetic processof the dikes that is underscored by their similar tectoniccontext the similar style of emplacement and the similartextural and mineral zoning across the dikes

There is a progressive increase of about two orders ofmagnitude in Li concentrations across the zones of individ-ual dikes For example in the Cryo-Genie dike concentra-tions range from 22 ppm to 5075 ppm The lowest Li con-centrations 70 ppm are in tourmalines in the line-rockof the Little Three main dike and the top border zone of the

910 JS Maloney et al

Cryo-Genie dike These tourmalines are at most few mil-limeters in length and are prismatic Li concentrations inradial tourmalines in the pegmatite zones of both the Cryo-Genie and Little Three dikes and in large tourmaline bladesin the core zones are on average higher by about an orderof magnitude The concentrations are highly variable how-ever In prismatic schorl crystals within pockets the con-centrations range from 273 to 954 ppm Two green over-growths on schorl cores have the highest Li concentrations1456 and 5075 ppm

Li isotope ratios

In contrast to the progressive increase in Li concentrationsin tourmalines across the zones of the dikes the variationin δ7Li is more complex (Fig 4b) There is no correlationwith Li concentrations but again there appears to be nosystematic difference in δ7Li between the three dikes Intourmalines from the wall-zoneline-rock core zone andpockets δ7Li has approximately the same range between112 and 161 In the pockets elbaite rims have lighterLi than schorl cores although the difference in the Cryo-Genie pair is within overlapping errors In three tourma-lines from the upper and the lower pegmatite zones ofboth Cryo-Genie and Little Three dikes δ7Li is very highgt 19 These three tourmalines came from radial splayswhereas others with lighter Li were prismatic or blade-shaped (Fig 3b) δ7Li in prismatic tourmaline from thebottom pegmatite zone of the Little Three dike is lower at142

The obtained δ7Li values and their range are among thehighest measured in rocks δ7Li in most unaltered crustalrocks typically does not exceed 10 (Tomascak 2004Teng et al 2004) but values up to 19 in quartz anda 10 difference between quartz and albite have beenreported for the Tin Mountain pegmatite in the Black HillsSouth Dakota USA (Teng et al 2006b)

Initial Li concentration in pegmatite melts

Lithium is often invoked as an element that may contributeto growth of large crystals that characterize pegmatites bydepolymerizing the silicate melt structure and thereforelowering its viscosity and increasing chemical diffusionrates In order to evaluate the fluxing effect of Li comparedto H2O in the San Diego County pegmatites a crude es-timate was made of its initial concentration in the meltsThe estimate comes from the Li concentrations in tourma-line the modal abundance of tourmaline and the Li con-centrations in fluid inclusions in pocket quartz assumingthat the pockets were the collection volumes for the sep-arated fluid Tourmaline is the largest reservoir for Li inthe line rock and pegmatite but in the cores and pocketswhere other Li-bearing minerals including lepidolite andamblygonite occur Li concentration in the fluid is best es-timated from primary fluid inclusions in quartz In the Lit-tle Three pegmatite these inclusions have a range of 7000 to12 000 ppm with average of 9200 ppm Li (unpublished data

determined by ion chromatography) The inclusion fluidscontain sim 3 wt NaCl based on microthermometric mea-surements

The initial concentration of Li in the pegmatite melts (Co)can be estimated using

Co = XLRCLR + XPEGCPEG + XPOCKCPOCK

where X is the mass proportion of each zone and C isthe concentration of Li in each zone Li concentrationsin the line-rock and the pegmatite zones were determinedfrom the average of Li concentrations in tourmaline in eachzone and the modal proportions of tourmaline Data for theCryo-Genie and Little Three dikes were combined Com-bining the data for the two dikes is justified because in thepegmatite zones which are volumetrically most abundantLi concentrations cover the same ranges The average Liconcentration in tourmaline in the Cryo-Genie pegmatitezones is 258 ppm and in the Little Three it is 298 ppmThe mass proportions of the three zones were determinedfrom their relative volumes and densities 2700 kgm3 forthe rocks and 500 kgm3 for the fluid The fluid density isappropriate for H2O with 3 wt NaCl at 2 kbar and 400 C(Anderko amp Pitzer 1993) The proportion of tourmaline inthe line-rock and pegmatite zones is lt 8 volume basedon image analysis of cut slabs The estimated bulk Li con-centration in the line-rock zones is only sim 5 ppm and in thepegmatite zones only sim 60 ppm It is readily evident thatmost Li from the original melt ended-up in the fluids

The pockets show that the exsolved fluid was collectedin discrete spaces instead of one continuous space betweenthe hanging and lower portions of the dikes Because thevolume and the distribution of the pockets are highly vari-able (Jahns 1979 Stern et al 1986) the volume of theexsolved fluid is difficult to estimate from the field occur-rences of the pockets A better estimate of the volume ofthe exsolved fluid comes from the maximum H2O solubil-ity in silicate melts which at 2 kbar is sim 6 wt (Holtzet al 1995) This amount of H2O given its molar volumeat 400 C and 2 kbar would occupy sim 28 of the chambervolume Using this proportion the calculated initial Li con-centration in the pegmatite dikes is only sim 630 ppm If theCryo-Genie data is left out of the calculation the differencein the result is only 1 ppm Although the ranges of the mea-sured Li concentrations and errors in the volume estimatesof tourmaline and the textural zones contribute to an erroron this estimate the estimate is dominated by the calculatedfluid volume and the Li concentration in the fluid as deter-mined from fluid inclusions Although 630 ppm Li couldappear to be a rather small concentration for a LCT-typemelt it is 18 times greater than the estimated average con-centration of 35 ppm in the upper continental crust (Tenget al 2004)

Discussion

Role of lithium in the pegmatite crystallization process

Since the classic model of Jahns amp Burnham (1969)for crystallization of pegmatite melts which involves the

Lithium and its isotopes in pegmatite tourmalines 911

separation of a fluid phase from a silicate melt it has beenrecognized that the formation of pegmatites also dependson kinetic controls during rapid cooling of dikes in par-ticular enrichments of components in melt boundary lay-ers ahead of rapidly crystallizing minerals (eg Rockholdet al 1987 London 1992 Webber et al 1999) In addi-tion to being responsible for crystallization of minerals inwhich it is an essential structural constituent Li is ofteninvoked as a possible component that along with H2O BF and other species can potentially flux granite melts Forexample fractional crystallization involving Li was partlyresponsible for mineral zoning seen in the large LCT-typeTin Mountain pegmatite in the Black Hills of South Dakotain which the outer zones are dominated by feldspars andthe inner zones by spodumene and quartz (Walker et al1986) Trapping temperatures of primary fluid inclusionsare lt 400 C and nearly invariable across the pegmatitesuggesting that it crystallized nearly isothermally as an un-dercooled liquid (Sirbescu amp Nabelek 2003a)

An extreme kinetic model is that of Morgan amp London(1999) for crystallization of the Little Three pegmatiteThey suggested that the low temperature and the fast cool-ing rate that must have occurred during solidification of thepegmatite did not allow for crystal nucleation until the melthas reached a glass state at sim 250 C below the equilibriumliquidus They proposed a constitutional zone-refining pro-cess in which a fluxed crystallization front swept a F Liand Mn-rich boundary layer through the solid or semi-soliddike eventually resulting in enrichment of these elementsin the pocket zone

Although we did not obtain electron microprobe data ontourmaline from the dikes the electron microprobe dataof Morgan and London (1999) on tourmaline in the Lit-tle Three dike suggests that the progressive increase in Liconcentrations from the line-rock and the upper wall-zoneto the pockets (Fig 4a) corresponds to changes in othertourmaline components Morgan amp London (1999) foundthat across the line-rock and the pegmatite zones Mg de-creases from sim 07 to near 0 per formula unit while Fe staysnearly constant The increase in Li and the decrease in Mgsuggests an increasing exchange of the elbaite componentfor the dravite component In the pockets tourmaline iszoned from schorl to elbaite but the zoning appears to becontinuous without an evidence for a miscibility gap (Mor-gan amp London 1999) consistent with evidence for com-plete solid solution between schorl and lithian olenite ina pegmatite from the eastern Alps (Kalt et al 2001) El-baite in the pockets has elevated F and Mn concentrations(Morgan amp London 1999) The occurrence of schorl coresin the pockets suggests that schorl grew while Fe-bearingmelt was still present in the dikes but the elbaite rims to-gether with other lithium minerals in pegmatite cores andpockets grew in equilibrium with Li-rich fluid collected inthe pockets The progressive increase in Li across the dikesis more consistent with progressive inward crystallizationof the dikes than a zone-refining process as Li appears tohave been progressively enriched in the residual liquid be-cause of its low solubility in early-crystallizing mineralsincluding schorl The change in tourmaline compositionacross the San Diego County pegmatites is analogous to

the tourmaline composition trend in the Bob Ingersol peg-matite in the Black Hills (Jolliff et al 1986)

The transition from the aplitic line-rock to pegmatiteprobably marks the point of fluid separation in the magmaFluid separation is suggested by the occurrences of elon-gated radial crystals that characterize the pegmatite zonesbecause crystals grow faster and become elongated whenH2O activity increases in the melt (Fenn 1977) HighH2O activity promotes an increase in the diffusion ratesof chemical components even when the melt is undergo-ing rapid cooling Crystallization of tourmaline (and gar-net) is controlled more by gradients in concentrations ofless mobile elements including Fe Mg and Mn than bygradients of rapidly-diffusing elements especially the al-kalis (Rockhold et al 1987 Webber et al 1999) Thelack of crystallization of Li minerals such as spodumeneand amblygonite until the core zones was probably pre-cluded by the initially low Li concentration and removalof Li by the exsolving fluid Although in Cl-absent water-peraluminous melt systems D(Li)fluidmelt is sim 04 (Londonet al 1988) D(Li)fluidmelt increases with the addition ofCl For example when an aqueous fluid at 800 C2 kbarhas sim 7 wt Cl D(Li)fluidmelt is sim 2 (Webster et al 1989)However even this D(Li)fluidmelt seems insufficient to ex-plain the very elevated Li in the pocket fluids by simplebatch partitioning between the fluid and the melt particu-larly because the fluid contained only sim 2 wt Cl Morelikely the large concentration of Li in the fluid is the re-sult of Rayleigh enrichment with crystallization dominatedby feldspars quartz and schorl tourmaline in which Li hassmall solubility compared to the melt and the fluid At 88 crystallization Li would have exceeded 05 wt (11 wtLi2O) in the melt and a correspondingly high concentrationin the accumulating fluid

Concentrations in excess of 1 wt Li2O that may haveexisted at later stages of fractional crystallization of thedikes are approached in some large spodumene-bearingpegmatite intrusions including the Tin Mountain peg-matite the Harding pegmatite in New Mexico and theTanco pegmatite in Manitoba (Norton 1994) These largepegmatites contain spodumene andor other Li-bearingminerals not only in their cores but also other zones (Nor-ton 1994) Li2O concentrations approaching 1 wt in asilicate melt may be required for crystallization of miner-als in which Li is an essential structural constituent In theSan Diego pegmatites such high concentrations apparentlyexisted only in the core zones and the pockets

The influence of the estimated initial 630 ppm Li on theviscosity of the dikes was likely far smaller in compari-son with the influence of the sim 6 wt H2O that wouldbe in the melt at the point of saturation Even in the corezones where Li2O may have reached 1 its effect wouldlikely have been much smaller 1 Li2O is equivalent to19 mole in a haplogranite melt whereas 6 wt H2Ois equivalent to sim 185 mole The addition of 1 wtof excess Li to a haplogranite melt lowers the viscosityby about one order of magnitude (Dingwell et al 1996)but in a peraluminous melt where Li may be complexedwith Al in a Si4+ = Al3+ + Li+ substitution the effect ofLi addition is probably smaller Even if Li exceeded its

912 JS Maloney et al

charge-balancing role in the melt upon removal of Al bycrystallization of the highly peraluminous tourmaline or ifAl was complexed with F in a non-network position in themelt (Giordano et al 2004 Mysen amp Richet 2005) Li isnot expected to have had a drastic effect on the viscosity ofthe dikes In comparison when H2O is added to a silicatemelt it initially dissolves by forming Al-OH and Si-OHcomplexes which depolymerize the tetrahedral networkViscosity drops by sim 45 orders of magnitude with the ad-dition of only 1 wt H2O to a peraluminous leucogranitemelt at 600 C and by an additional 6 orders of magnitudewith further addition of 5 wt H2O (Romano et al 2001Whittington et al 2004) The decreasing effect of H2O ad-dition on the viscosity comes from the increasingly greaterdissolution of H2O as a molecular species (Stolper 1982)Likewise diffusivities of ions in silicate melts dramaticallyincrease with the addition of only a small amount of H2Obut less so with further addition of H2O (Watson 1994)

Lithium isotope fractionation

A potentially large fractionation of Li-isotopes in any mul-tiphase system is due to the 17 mass difference be-tween 7Li and 6Li Presently available data show only verylimited Li isotope fractionation during crystallization ofhigh-temperature igneous systems (Tomascak et al 1999Magna et al 2006 Halama et al 2007) and during vapor-liquid separation in hydrothermal systems (Foustoukoset al 2004 Liebscher et al 2007) However significantfractionation may result from different coordination statesof Li in coexisting phases that include multiple miner-als melts and aqueous fluids in relatively low-temperaturegranitic systems (Wenger amp Armbruster 1991) In general6Li preferentially occupies sites with higher coordinationnumbers and therefore weaker bonds whereas 7Li prefer-entially enters sites with smaller coordination numbers andstronger bonds (Wunder et al 2007)

Fractionation of Li between minerals and fluids hasbeen experimentally determined only for a limited set ofminerals to date Wunder et al (2006 2007) found thatΔ7Listauroliteminusfluid = +13 and is essentially temperature-independent Δ7Lilepidoliteminusfluid is approximately ndash2 withsome temperature dependence in the 350ndash400 C rangeΔ7Lispodumeneminusfluid is also temperature-dependent but morenegative by about 3 For all three minerals the frac-tionation is insensitive to the Cl content of the fluid whichimplies that Li probably forms a tetrahedrally-coordinatedhydrated ion Li(H2O)+4 instead of a LiCl or LiOH com-plex Wunder et al (2007) concluded that equilibrium iso-topic fractionation is firstly controlled by Li coordina-tion with 7Li preferentially incorporated into the phasethat allows for a smaller coordination number and sec-ondly by the Li-O bond length giving the relationshipδ7Listaurolite gt δ

7Lilepidolite gt δ7Lispodumene In staurolite

Li substitutes for the divalent cations Fe2+ Mg and Znin the tetrahedral sites in lepidolite Li is octahedrally-coordinated between tetrahedral layers and in spodumeneit occupies the relatively large M2 octahedral site Thereare no experimental Li isotope fractionation data involv-

ing either tourmaline or melt but in tourmaline Li occupiesthe octahedrally-coordinated Y-site and in a peraluminousmelt Li is probably strongly bonded in its charge balanc-ing role with tetrahedrally-coordinated Al3+ in a LiAlSi3O8complex (Mysen amp Richet 2005) Therefore Li in such amelt should be isotopically relatively heavy

The results of Wunder et al (2007) involving muscoviteare quite different from the results of Lynton et al (2005)who found Δ7Limuscoviteminusfluid to range between +8 and +20in the 400ndash500 C interval Wunder et al (2007) attributedthe discrepancy to the diffusion mechanism that Lyntonet al (2005) used to introduce Li into muscovite IndeedTeng et al (2006a) ascribed very large variations in δ7Li ofcountry rocks in the aureole of the Tin Mountain pegmatiteto differential diffusion of the two Li isotopes which under-scores that in addition to the energy of bonds in lattices ki-netic effects may induce transient Li isotope fractionationwhich may be preserved in rapidly cooled systems

There have been no direct measurements of Li iso-tope fractionation between peraluminous silicate melts andaqueous fluids but it is expected that Li in a melt should beisotopically heavier because of strong bonds in associationwith charge balancing of Al that is in tetrahedral coordina-tion This inference is supported by isotopic compositionsof Li in fluid inclusions and host quartz in the Tin Mountainpegmatite in the Black Hills (Teng et al 2006b) Li in thefluid inclusions has much lower δ7Li values than Li in thequartz supporting the inference that 7Li prefers the strongbonds in quartz where it is possibly charge-balancing Althat is incorporated into the quartz structure and by anal-ogy bonds in high-silica melts over the weaker hydratedbonds in the fluid

The San Diego pegmatites were systems in which min-erals melts and aqueous fluids coexisted at various stagesof crystallization Assuming equilibrium the isotopic com-position of Li in tourmaline in the dikes is reflective of themedium from which the tourmaline crystallized HoweverLi isotopes may be strongly fractionated by kinetic effectsThere is a several permil heterogeneity in δ7Li even in indi-vidual zones of a single pegmatite Most interesting are thevery elevated gt 19 δ7Li values of radial tourmalines inboth the hanging and lower pegmatite zones A reason forthe elevated values may be that the tourmalines grew whenthe melt became saturated in the aqueous fluid at whichpoint Li isotopes were fractionated between the melt andthe aqueous fluid with most Li going into the fluid (Fig 5)If the isotopic composition of Li in tourmaline in each ofthe zones reflects the relative fractionation of Li isotopesbetween the melt and the fluid then the isotopically heavierLi in the pegmatite zone tourmaline is consistent with theexpected stronger Li bonds in the melt compared to bondsof hydrated Li in the fluid The similarity of δ7Li values intourmaline that crystallized in the line rock and the pock-ets is consistent with accumulation of the bulk of meltrsquosinitial Li into the pocket fluid as this would result in littlechange in the isotopic composition of Li from the initialfluid-undersatured melt to the eventually collected fluid inthe pockets Elbaite rims of pocket tourmalines have lowerδ7Li than their corresponding cores although one core-rimpair has overlapping analytical errors (Fig 4b) The lower

Lithium and its isotopes in pegmatite tourmalines 913

Fig 5 Schematic diagram illustrating a possible mechanism of pro-ducing high δ7Li tourmalines in pegmatite zones during fluid-meltseparation Because Li is used in charge-balancing with strongly-bonded Al in the silicate melt the δ7Li in the melt should be moreelevated than δ7Li in the fluid where Li probably occurs mostly asa hydrated ion (Wunder et al 2007) Tourmalines crystallizing inequilibrium with the melt and fluid respectively may reflect the iso-topic fractionation between the melt and the fluid

δ7Li values of the elbaite rims are consistent with crystal-lization of schorl cores while melt was still present andcrystallization of elbaite rims in the presence of the fluidonly

The lack of a systematic increase in δ7Li across the tex-tural zones of the dikes suggests that there was little in-fluence of tourmaline itself on the isotopic composition ofthe residual melt during crystallization Given that bothΔ7Lispodumeneminusfluid and Δ7Lilepidoliteminusfluid are both negativeΔ7Litourmalineminusmelt should be even more negative because Liin tourmaline is in octahedral coordination while in themelt it is associated with tetrahedral Al However becausethe proportion of tourmaline in the line-rock and pegmatitezones is lt 8 and the concentration of Li in the tourma-line is very small crystallization of tourmaline would havehad a negligible effect on the Li isotope ratio in the resid-ual melt Based on mass-balance calculations only sim 2 of the initial 630 ppm Li is contained in schorl in the line-rocks and the pegmatites zones

Kinetic effects

A kinetic cause for the elevated δ7Li values of the radialtourmalines in the pegmatite zones must also be consid-ered however The shape of the tourmalines suggests thatthey grew very rapidly in which case the Li isotope frac-tionation may have been kinetically controlled by differ-ential diffusion of Li isotopes at the crystal-liquid (andorcrystal-fluid) interface The relative diffusion rates of two

Fig 6 Schematic diagram illustrating a possible kinetic mechanismof producing high δ7Li radial tourmalines in pegmatite zones Be-cause the diffusion of 7Li is slower in the melt than the diffusion of6Li and when Li is an incompatible element relative to tourmaline7Li should become preferentially enriched in a boundary layer aheadof a tourmaline crystal that is growing faster than the rate at whichLi diffuses in the melt

isotopes of a given element in a silicate liquid are given by

D1

Dh=

(mh

m1

)β

where m is the mass of isotope ldquolrdquo stands for the light iso-tope and ldquohrdquo stands for the heavy isotope (Richter et al2003) Using basalt-rhyolite melt couples Richter et al(2003) experimentally determined that βLi asymp 0215 Thismeans that 6Li can diffuse substantially faster than 7Lithrough the melt away from a growing crystal so that apreferential enrichment of 7Li can potentially occur in thechemical boundary layer ahead of the crystal (Fig 6) re-sulting in elevated δ7Li The diffusion rate of Li in silicatemelts is orders of magnitude faster than the rates of othermajor and minor cations (Richter et al 2003) Under con-ditions of slow mineral growth homogeneous Li isotoperatios would be expected in the tourmalines Instead theobserved Li isotope heterogeneity suggests crystallizationconditions under which the diffusion of Li in the melt didnot keep-up with the rate of tourmaline growth

Elevated δ7Li values of pegmatites

The overall elevated δ7Li values seen in the San Diegopegmatites are similar to the values in the Tin Mountainpegmatite in the Black Hills δ7Li values in the associ-ated Harney Peak leucogranite and the host schists of theTin Mountain pegmatite have δ7Li values within a fewpermil of 0 which points to some process that leads tostrong Li isotope fractionation during generation of LCT-type pegmatite melts (Teng et al 2006b) The apparentlarge fractionation for these relatively low-temperature ig-neous systems contrasts with the minimal fractionation in

914 JS Maloney et al

high-temperature igneous systems (Tomascak 2004) Tenget al (2006b) suggested that elevated δ7Li values in peg-matite melts could potentially be acquired by crystal-liquidfractionation during crystallization of parental magmasHowever for this process to be effective a greater amountof Li would have to remain in the crystallized assemblage(parental granite) than in the residual liquid (pegmatite)This is inconsistent with for example the relative concen-trations of Li in the Harney Peak leucogranite and its poten-tially residual liquid now represented by the Tin Mountainpegmatite Li concentrations in the Harney Peak leucogran-ite range from 10ndash205 ppm (Teng et al 2006b)

A possible alternative explanation is that the structureof a hydrous pegmatite melt has more similarity to waterthan to minerals and therefore it incorporates 7Li pref-erentially over minerals in a fashion analogous to thelarge 7Li enrichment in crustal fluids compared to crustalrocks (Tomascak 2004) As hydrated pegmatite melts passthrough the crust from their sources they may acquire el-evated δ7Li values through rapid Li isotope exchange withthe surrounding rocks Matthews et al (2003) found thatthe oxygen and especially hydrogen isotopic compositionof pegmatite dikes on Naxos Greece reflects the isotopiccomposition of the host rocks which change in isotopiccomposition along strike of the dikes If hydrogen isotopescan be nearly fully exchanged between pegmatite melts andtheir host rocks then it is likely that Li isotopes can also bereadily exchanged given the fast diffusion rates of Li in sil-icate liquids (Richter et al 2003)

Acknowledgements We are grateful for access providedby Dana and Ken Gochenour to the Cryo-Genie propertyBill Calhoun to the San Diego mine property and LouisSpaulding Jr to the Little Three property Jeffrey Patter-son and Matt Taylor provided guidance around the peg-matite districts and Jim Student helped with sample col-lection Carol Nabelek oversaw the ICP-OES analysis BillMcDonough graciously gave access to JM to conduct anal-ysis in the isotope laboratory at the University of MarylandThe paper benefited from the constructive reviews of AxelLiebscher Jeffrey Ryan and Ed Grew and additional com-ments from Roberta Rudnick A Feodor-Lynen fellowshipto Halama by the Alexander von Humboldt Foundation isgratefully acknowledged The study was supported by Uni-versity of Missouri Research Board Grant D3508 and NSFGrant 408564 to Nabelek Additional funding came fromNSF Grant EAR 0606989 to Rudnick and McDonough

References

Anderko A amp Pitzer KS (1993) Equation-of-state representa-tion of phase equilibria and volumetric properties of the systemNaCl-H2O above 573 K Geochim Cosmochim Acta 57 1657-1680

Bouman C Elliott T Vroon PZ (2004) Lithium inputs to sub-duction zones Chem Geol 212 59-79

Cernyacute P (1991) Rare-element granitic pegmatites Part I Anatomyand internal evolution of pegmatite deposits GeoscienceCanada 18 49-67

Cernyacute P amp Ercit TS (2005) The classification of granitic peg-matites revisited Can Mineral 43 2005-2026

Dingwell DB Hess K-U Knoche R (1996) Granite andgranitic pegmatite melts volumes and viscosities Trans R SocEdinburgh Earth Sci 87 65-72

Fenn PM (1977) The nucleation and growth of alkali feldsparsfrom hydrous melts Can Mineral 15 135-161

Fisher J (2002) Gem and rare-element pegmatites of southernCalifornia Mineral Record 33 363-407

Fisher J Foord EE Bricker GA (1999) The geology miner-alogy and history of the Himalaya Mine Mesa Grande SanDiego County California California Geol 3-17

Foord EE (1976) Mineralogy and petrogenesis of layeredpegmatite-aplite dikes in the Mesa Grande District San DiegoCounty California PhD Dissertation Stanford University

Foord EE Starkey HC Taggard JE Jr (1986) Mineralogyand paragenesis of ldquopocketrdquo clays and associated minerals incomplex granitic pegmatites San Diego County California AmMineral 71 428-439

Foustoukos DI James RH Berndt ME Seyfried WE Jr(2004) Lithium isotopic systematics of hydrothermal vent flu-ids at the Main Endeavour Field Northern Juan de Fuca RidgeChem Geol 212 17-26

Giordano D Romano C Dingwell DB Poe B Behrens H(2004) The combined effects of water and fluorine on theviscosity of silicic magmas Geochim Cosmochim Acta 685159-5168

Halama R McDonough WF Rudnick RL Keller J KlaudiusJ (2007) The Li isotopic composition of Oldoinyo LengaiNature of the mantle sources and lack of isotopic fractionationduring carbonatite petrogenesis Earth Planet Sci Lett 25477-89

Holtz F Behrens H Dingwell DB Johannes W (1995) Watersolubility in haplogranite melts Compositional pressure andtemperature dependence Am Mineral 80 94-108

Jahns R H (1979) Gem-bearing pegmatites in San Diego CountyCalifornia The Stewart mine Pala district and the Himalayamine Mesa Grande district in ldquoMesozoic crystalline rocksPeninsular Ranges batholith and pegmatites Point Sol ophio-literdquo P L Abbott amp V R Todd eds San Diego State UniversitySan Diego California 3-38

Jahns RH amp Burnham CW (1969) Experimental studies of peg-matite genesis I A model for the derivation and crystallizationof granitic pegmatites Econ Geol 64 843-864

Jahns RH amp Tuttle OF (1963) Layered pegmatite-aplite intru-sives Mineral Soc Am Sp Pap 1 78-92

Jolliff BL Papike JJ Shearer CK Laul JC (1986)Tourmaline as a recorder of pegmatite evolution Bob Ingersolpegmatite Black Hills South Dakota Am Mineral 71 472-500

Kalt A Schreyer W Ludwig T Prowatke S Bernhardt H ErtlA (2001) Complete solid solution between magnesian schorland lithian excess-boron olenite in a pegmatite from the Koralpe(eastern Alps Austria) Eur J Mineral 13 1191-1205

Kampf AR Gochenour K Clanin J (2003) Tourmaline dis-covery at the Cryo-Genie mine San Diego County CaliforniaRocks and Minerals 78 156-163

Liebscher A Meixner A Romer RL Heinrich W (2007)Experimental calibration of the vapourndashliquid phase relationsand lithium isotope fractionation in the system H2OndashLiCl at400 Geofluids 7 1-7

906 JS Maloney et al

rates in highly undercooled melts (Webber et al 1999Nabelek 2007 Sirbescu et al 2008)

This study was inspired in part by the work of WernerSchreyer on the crystal-chemistry of tourmaline Hiswork has demonstrated the value of tourmaline as anindicator of the petrogenesis of igneous rocks includingpegmatites (eg Schreyer et al 2000 Kalt et al 2001)In the present study concentrations and isotopic ratiosof Li in tourmaline are used to constrain the role of Liin the crystallization of three pegmatite dikes in the SanDiego County California USA These dikes have beenstudied by many workers including as a foundation forthe classic pegmatite crystallization model of Jahns ampBurnham (1969) In the lithium-cesium-tantalum (LCT)family of pegmatites to which the San Diego Countydikes belong Li occurs at concentrations high enoughto cause crystallization of Li minerals including elbaitetourmaline [Na(Li15Al15)Al6B3Si6O27(OH)3(OH F)]spodumene [LiAlSi2O6] amblygonite-montebrasite[(Li Na)AlPO4(F OH)] and lepidolite along thepolylithionite [KLi2AlSi4O10(F OH)2] ndash trilithionite[KLi15Al15AlSi3O10(F OH)2] join (Cernyacute amp Ercit 2005)Because Li can substitute into tourmaline which is ubiq-uitous in all portions of the San Diego pegmatite dikestourmaline was used here as a proxy for determining thebehavior of Li and to assess its importance for fluxinggranite melts Isotope ratios of Li can be a mirror onmagma differentiation because the fractionation of Li iso-topes in a multiphase system is to a large extent controlledby its coordination in coexisting phases andor kinetics(Wenger amp Armbruster 1991 Richter et al 2003 Wunderet al 2006 2007)

Geologic context and samples

The pegmatites of San Diego County are located within thenorthwest trending subduction-related Peninsular RangesBatholith (PRB) that stretches from the San Jacinto Moun-tains of California into Baja California Mexico (Fig 1)The batholith is divided into the western and eastern zones(Walawender et al 1990 Fisher 2002 Todd et al 2003)The western zone is comprised of I-type gabbros quartzdiorites tonalites granodiorites and monzogranites rang-ing in age from 120 to 105 Ma in age The more fel-sic eastern zone lacks gabbros and is dominated by largeconcentrically-zoned I- and S-type tonalites and monzo-granites whose intrusion began at sim 105 Ma when mag-matism in the western zone stopped due to a change inplate dynamics Some of the intrusions include the 100and 89 Ma La Posta-type plutons that straddle to east-westboundary of the PRB (Walawender et al 1990 Symonset al 2003) La Posta-type plutons are zoned inwardfrom hornblende tonalite to muscovite-bearing monzogran-ite The pegmatite field is broadly contemporaneous withthe La Posta-type plutons and also straddles the east-westboundary The pegmatites occur as suites of parallel to sub-horizontal dikes mostly within the plutons and occasion-ally within intervening schists The dikes were emplaced at2ndash3 kbar based on densities of fluid inclusions trapped in

Fig 1 Regional sketch showing locations of the Cryo-Genie Lit-tle Three and Himalaya pegmatite dikes in the San Diego CountyCalifornia Inset shows location of the Peninsular Ranges Batholith(stippled) The black rectangular shape is San Diego County

pockets (London 1986b) The host plutons appear to havebeen brittle when the dikes were emplaced

Tourmaline was collected from three pegmatite dike sys-tems ndash the Cryo-Genie the Little Three and the Himalaya(Fig 1) Dikes in all three systems exhibit similar zoning inmineralogy texture and chemistry (Jahns amp Tuttle 1963Jahns 1979) A typical dike has a lower border zone that ischaracterized by rhythmically-layered aplite with oscillat-ing changes in mineralogy from Mg-Fe-poor layers domi-nated by quartz and albite to Mg-Fe-rich layers dominatedby tourmaline and sometimes garnet This layered apliteis colloquially called ldquoline-rockrdquo Above the line-rock isthe lower pegmatite zone with upward growth direction ofminerals The hanging part of the dike is characterized by athin border zone which occasionally is also layered Belowthis border zone is the hanging pegmatite with large euhe-dral K-feldspar crystals and then graphic quartz-feldsparintergrowths Between the hanging and lower pegmatites isthe coarse-grained core zone The core zone includes pock-ets that have variable size and occurrence along the dikersquosstrike Aside from gem-quality crystals the pockets are of-ten partially filled with clays that suggest low-temperaturealteration of core and pocket minerals by magmatic flu-ids (Foord et al 1986) When pockets are absent the de-marcation between the lower and hanging portions of thedike is shown by opposite growth directions of mineralsand occurrences of massive quartz large amount of micaand large tourmaline blades In this paper both a coarse-grained core and a wall of a pocket are referred to as theldquocore zonerdquo

The main Cryo-Genie dike intruded into sillimanite-grade metasedimentary rocks and a portion of the localLa Posta-type granite Field relationships with dated gran-ites suggest that the dike was emplaced between 98 and89 Ma (Kampf et al 2003) The dike is 2ndash4 m thick andcan be traced for more than 200 m The analyzed sampleswere collected at the Green Ledge surface excavations andaround the Payday pocket in underground excavations (Ta-ble 1 Fig 2a b) At Green Ledge only the upper portion of

Lithium and its isotopes in pegmatite tourmalines 907

Table 1 Tourmaline locations shapes and Li concentrations and isotope ratios

Sample Zone Tourmaline shape Li (ppm) δ7Li ()CG-1a middle hanging pegmatite radial 53 192CG-1c core zone large blade 174CG-1e core zone large blade 138CG-1i black pocket prism 273 159CG-1i green pocket prism 5075 147CG-1n pocket prism 416 112CG-3e core zone large blade 265 161CG-3i core zone large blade 663 112CG-3k middle hanging pegmatite radial 421CG-5b top border short blade 22CG-5d middle hanging pegmatite radial 93LT-1c core zone large blade 517 134LT-3f bottom line rock small prismatic 56 151LT-3g bottom line rock small prismatic 66LT-3i bottom line rock small prismatic 70 123LT-4c core zone large blade 268 148LT-5c bottom pegmatite radial 452 229LT-5d bottom pegmatite radial 113 191LT-5e bottom pegmatite small prismatic 135 142HM-2C black pocket prism 641 159HM-2c green pocket prism 1456 137HM-4C pocket prism 954 159

Fig 2 Sketches showing sample locations at the Cryo-Genie pegmatite dike (a) Green Ledge location on the surface and (b) Payday pocketunderground

908 JS Maloney et al

the dike from the upper contact to the core zone includinga pocket is exposed CG-5b is a short euhedral tourmalinefrom the border of the hanging pegmatite CG-1a and CG-5d are radially-grown tourmaline needles from the hangingpegmatite and CG-1c and CG-1e are large euhedral tour-malines from the core zone CG-1i and CG-1n are tourma-line prisms that grew into the pocket CG-1i has a schorlcore and green elbaite rim CG-3e and CG-3i are schorlblades from core zone around the Payday pocket whereasCG-3k is radial schorl from the pegmatite above the pocket

The Little Three dike system intrudes the Green Valleytonalite-gabbro of the western zone of the PRB (Stern et al1986) The system has five dikes including the Little Threemain dike and the Spaulding dike that were sampled forthis study The dikes vary between 1 and 2 m in thick-ness along strike The relative thickness of line-rock andpegmatite zones also varies Most of the analyzed samplescome from the Little Three main dike near the mine en-trance on Topaz Ledge (Fig 3) The bottom boundary ofthe dike with the country rock is a homogeneous aplite withline-rock above it The line rock terminates sharply againstthe lower pegmatite zone which contains small prismaticor bladed tourmaline within graphic intergrowths of quartzand feldspar

Samples LT-3f LT-3g and LT-3i come from the line-rock(Fig 3a) Samples LT-5c and 5d are radial tourmalines thatgrew in the lower pegmatite that cuts through the line rockalong strike from location 3 (Fig 3b) LT-5e is a prismatictourmaline in a smaller dike cutting through the line rockSample LT-4c is a large tourmaline blade from the corezone around a several-meter size pocket that occurs abovethe mine entrance Sample LT-1c is a similar tourmalinefrom the core zone of the Spaulding dike

The Himalaya dike system includes two subparallel dikeshosted by the San Marcos gabbro (Fisher et al 1999Webber et al 1999) The two dikes are separated by 3ndash10 m along the 915 m of their exposed length but convergeat the San Diego mine where the samples were collectedThe Himalaya dikes were emplaced sim 100 my ago asdated by fission track and K-Ar methods (Foord 1976)Samples HM-2c and HM-4c come from pockets in the un-derground workings of the San Diego mine TourmalineHM-2c includes a schorl core and a green elbaite rim whileHM-4c is exclusively schorl

The Cryo-Genie samples best represent a cross-sectionacross the hanging portion of a pegmatite dike while theLittle Three samples best represent the lower portion of adike The Himalaya samples were analyzed primarily to de-termine if there is a regional variability in Li isotopic com-positions

Methods

Bulk Li concentrations

Tourmaline is extremely difficult to dissolve in acids andtherefore we have developed a flux procedure that ac-complishes complete dissolution of ground-up tourmalineBlack cores and green rims of zoned pocket tourmalines

Fig 3 Sketches showing sample locations at the Little Three peg-matite dike Parts a) and b) are along strike on the main dike

were prepared separately When samples had only smallcrystals multiple grains were ground together 100 mgof each tourmaline powder were mixed with 400 mg ofground K2CO3 flux in a zirconia crucible and placed into afurnace at 500 C The temperature was then increased to900 C for 15 min after which the furnace was turned offThe crucibles were left in the furnace overnight to slowlycool to prevent cracking Each fluxed sample was first cen-trifuged in 15 ml of 10 HNO3 for four minutes Theliquid was then decanted into a Teflon beaker and stirredon a hot plate The remaining solid residue was again cen-trifuged in 15 ml of 10 HNO3 Any residue that was stillleft was dissolved in 3 ml of concentrated HNO3 and wasthen added to the already dissolved tourmaline and stirredfor additional 15 minutes on low heat The sample was thenbrought up to 50 ml in a volumetric flask with 10 HNO3In several samples a brown amorphous precipitate formedafter several days After the liquids were decanted the pre-cipitates were dissolved in a HF-HNO3 mixture and thendiluted to 50 ml with distilled water Analysis of these so-lutions revealed no Li

Lithium concentrations (Table 1) were analyzed using thePerkin-Elmer Optima 3300 Inductively Coupled Plasma

Lithium and its isotopes in pegmatite tourmalines 909

Optical Emission Spectrometer (ICP-OES) at the Univer-sity of Missouri Synthetic standards with 005 010 1and 5 ppm Li were prepared using blank K2CO3 flux so-lutions to create the calibration curves for Li Instrumentaldrift was accounted for by analyzing the prepared standardsat regular intervals during the analysis and aliquots of thestandard solutions were used as check standards Lithiumconcentrations were obtained from the strong 610362 nmemission line

Li isotope ratios

Analysis of Li isotope ratios (Table 1) was carried out ondried aliquots of solutions prepared for ICP-OES analy-sis The dried samples were prepared and analyzed at theGeochemistry Laboratory of the University of Maryland-College Park The preparation followed the procedure out-lined in Rudnick et al (2004) and Teng et al (2004) whichis based on the three-column procedure of Moriguti ampNakumura (1998) Because NaLi ratio in solution largerthan sim 5 may cause instability in the analysis the NaLiratio of each sample was determined semi-quantitativelyprior to analysis and excess Na was stripped-off by addi-tional column purification Measurements were done on aNu Plasma Multicollector Inductively Couple Plasma MassSpectrometer (MC-ICP-MS) Each analysis consisted oftwo blocks of twenty individual measurements Each anal-ysis was bracketed by the measurement of a 100 ppb L-SVEC standard The 7Li6Li ratio in L-SVEC for each2 times 20 measurements had an average 2σ of the mean 0003 The external precision better than 1 Two otherLi isotope standards IRMM-016 (Qi et al 1997) and thein-house standard UMD-1 were routinely analyzed dur-ing each analytical session The results for both (IRMM-016 +02 plusmn 04 UMD-1 +543 plusmn 02) agree well withpublished results (IRMM-016 ndash01 plusmn 02 to +02 plusmn 08UMD-1 +547 plusmn 10 Rudnick et al 2004 Teng et al2004 Halama et al 2007) Two USGS rock standardsBHVO-1 (+42 ) and QLO-1 (+66 ) were analyzedfor quality-control purposes The value for BHVO-1 waswithin the uncertainty of previously published results (+43to +58 Bouman et al 2004 Rudnick et al 2004) and thevalue for QLO-1 was within the range of previous analy-ses at the University of Maryland (+56 to +68 Halamaunpublished data)

Results

Li concentrations

Li concentrations in the tourmalines (Table 1) are shown inFig 4a They are plotted in terms of the zones in which thetourmalines grew There appear to be no systematic differ-ences in Li concentrations between corresponding zones ofthe different dikes based on the available data in spite ofthe spatial and probably some temporal separation of thedikes although with denser data sets for all three dikessome systematic differences could appear Nevertheless

Fig 4 (a) Li concentrations and (b) Li isotopic ratios in tourma-lines in three pegmatite dikes Data are grouped according to tex-tural zones from which samples came from Schorl tourmalines areshown by black symbols and elbaite rims on schorl by white sym-bols Two core-rim pairs are shown by connecting lines

the similar ranges in Li concentrations in the pegmatite andthe core zones of both the Cryo-Genie and the Little Threedikes for example point to a common petrogenetic processof the dikes that is underscored by their similar tectoniccontext the similar style of emplacement and the similartextural and mineral zoning across the dikes

There is a progressive increase of about two orders ofmagnitude in Li concentrations across the zones of individ-ual dikes For example in the Cryo-Genie dike concentra-tions range from 22 ppm to 5075 ppm The lowest Li con-centrations 70 ppm are in tourmalines in the line-rockof the Little Three main dike and the top border zone of the

910 JS Maloney et al

Cryo-Genie dike These tourmalines are at most few mil-limeters in length and are prismatic Li concentrations inradial tourmalines in the pegmatite zones of both the Cryo-Genie and Little Three dikes and in large tourmaline bladesin the core zones are on average higher by about an orderof magnitude The concentrations are highly variable how-ever In prismatic schorl crystals within pockets the con-centrations range from 273 to 954 ppm Two green over-growths on schorl cores have the highest Li concentrations1456 and 5075 ppm

Li isotope ratios

In contrast to the progressive increase in Li concentrationsin tourmalines across the zones of the dikes the variationin δ7Li is more complex (Fig 4b) There is no correlationwith Li concentrations but again there appears to be nosystematic difference in δ7Li between the three dikes Intourmalines from the wall-zoneline-rock core zone andpockets δ7Li has approximately the same range between112 and 161 In the pockets elbaite rims have lighterLi than schorl cores although the difference in the Cryo-Genie pair is within overlapping errors In three tourma-lines from the upper and the lower pegmatite zones ofboth Cryo-Genie and Little Three dikes δ7Li is very highgt 19 These three tourmalines came from radial splayswhereas others with lighter Li were prismatic or blade-shaped (Fig 3b) δ7Li in prismatic tourmaline from thebottom pegmatite zone of the Little Three dike is lower at142

The obtained δ7Li values and their range are among thehighest measured in rocks δ7Li in most unaltered crustalrocks typically does not exceed 10 (Tomascak 2004Teng et al 2004) but values up to 19 in quartz anda 10 difference between quartz and albite have beenreported for the Tin Mountain pegmatite in the Black HillsSouth Dakota USA (Teng et al 2006b)

Initial Li concentration in pegmatite melts

Lithium is often invoked as an element that may contributeto growth of large crystals that characterize pegmatites bydepolymerizing the silicate melt structure and thereforelowering its viscosity and increasing chemical diffusionrates In order to evaluate the fluxing effect of Li comparedto H2O in the San Diego County pegmatites a crude es-timate was made of its initial concentration in the meltsThe estimate comes from the Li concentrations in tourma-line the modal abundance of tourmaline and the Li con-centrations in fluid inclusions in pocket quartz assumingthat the pockets were the collection volumes for the sep-arated fluid Tourmaline is the largest reservoir for Li inthe line rock and pegmatite but in the cores and pocketswhere other Li-bearing minerals including lepidolite andamblygonite occur Li concentration in the fluid is best es-timated from primary fluid inclusions in quartz In the Lit-tle Three pegmatite these inclusions have a range of 7000 to12 000 ppm with average of 9200 ppm Li (unpublished data

determined by ion chromatography) The inclusion fluidscontain sim 3 wt NaCl based on microthermometric mea-surements

The initial concentration of Li in the pegmatite melts (Co)can be estimated using

Co = XLRCLR + XPEGCPEG + XPOCKCPOCK

where X is the mass proportion of each zone and C isthe concentration of Li in each zone Li concentrationsin the line-rock and the pegmatite zones were determinedfrom the average of Li concentrations in tourmaline in eachzone and the modal proportions of tourmaline Data for theCryo-Genie and Little Three dikes were combined Com-bining the data for the two dikes is justified because in thepegmatite zones which are volumetrically most abundantLi concentrations cover the same ranges The average Liconcentration in tourmaline in the Cryo-Genie pegmatitezones is 258 ppm and in the Little Three it is 298 ppmThe mass proportions of the three zones were determinedfrom their relative volumes and densities 2700 kgm3 forthe rocks and 500 kgm3 for the fluid The fluid density isappropriate for H2O with 3 wt NaCl at 2 kbar and 400 C(Anderko amp Pitzer 1993) The proportion of tourmaline inthe line-rock and pegmatite zones is lt 8 volume basedon image analysis of cut slabs The estimated bulk Li con-centration in the line-rock zones is only sim 5 ppm and in thepegmatite zones only sim 60 ppm It is readily evident thatmost Li from the original melt ended-up in the fluids

The pockets show that the exsolved fluid was collectedin discrete spaces instead of one continuous space betweenthe hanging and lower portions of the dikes Because thevolume and the distribution of the pockets are highly vari-able (Jahns 1979 Stern et al 1986) the volume of theexsolved fluid is difficult to estimate from the field occur-rences of the pockets A better estimate of the volume ofthe exsolved fluid comes from the maximum H2O solubil-ity in silicate melts which at 2 kbar is sim 6 wt (Holtzet al 1995) This amount of H2O given its molar volumeat 400 C and 2 kbar would occupy sim 28 of the chambervolume Using this proportion the calculated initial Li con-centration in the pegmatite dikes is only sim 630 ppm If theCryo-Genie data is left out of the calculation the differencein the result is only 1 ppm Although the ranges of the mea-sured Li concentrations and errors in the volume estimatesof tourmaline and the textural zones contribute to an erroron this estimate the estimate is dominated by the calculatedfluid volume and the Li concentration in the fluid as deter-mined from fluid inclusions Although 630 ppm Li couldappear to be a rather small concentration for a LCT-typemelt it is 18 times greater than the estimated average con-centration of 35 ppm in the upper continental crust (Tenget al 2004)

Discussion

Role of lithium in the pegmatite crystallization process

Since the classic model of Jahns amp Burnham (1969)for crystallization of pegmatite melts which involves the

Lithium and its isotopes in pegmatite tourmalines 911

separation of a fluid phase from a silicate melt it has beenrecognized that the formation of pegmatites also dependson kinetic controls during rapid cooling of dikes in par-ticular enrichments of components in melt boundary lay-ers ahead of rapidly crystallizing minerals (eg Rockholdet al 1987 London 1992 Webber et al 1999) In addi-tion to being responsible for crystallization of minerals inwhich it is an essential structural constituent Li is ofteninvoked as a possible component that along with H2O BF and other species can potentially flux granite melts Forexample fractional crystallization involving Li was partlyresponsible for mineral zoning seen in the large LCT-typeTin Mountain pegmatite in the Black Hills of South Dakotain which the outer zones are dominated by feldspars andthe inner zones by spodumene and quartz (Walker et al1986) Trapping temperatures of primary fluid inclusionsare lt 400 C and nearly invariable across the pegmatitesuggesting that it crystallized nearly isothermally as an un-dercooled liquid (Sirbescu amp Nabelek 2003a)

An extreme kinetic model is that of Morgan amp London(1999) for crystallization of the Little Three pegmatiteThey suggested that the low temperature and the fast cool-ing rate that must have occurred during solidification of thepegmatite did not allow for crystal nucleation until the melthas reached a glass state at sim 250 C below the equilibriumliquidus They proposed a constitutional zone-refining pro-cess in which a fluxed crystallization front swept a F Liand Mn-rich boundary layer through the solid or semi-soliddike eventually resulting in enrichment of these elementsin the pocket zone

Although we did not obtain electron microprobe data ontourmaline from the dikes the electron microprobe dataof Morgan and London (1999) on tourmaline in the Lit-tle Three dike suggests that the progressive increase in Liconcentrations from the line-rock and the upper wall-zoneto the pockets (Fig 4a) corresponds to changes in othertourmaline components Morgan amp London (1999) foundthat across the line-rock and the pegmatite zones Mg de-creases from sim 07 to near 0 per formula unit while Fe staysnearly constant The increase in Li and the decrease in Mgsuggests an increasing exchange of the elbaite componentfor the dravite component In the pockets tourmaline iszoned from schorl to elbaite but the zoning appears to becontinuous without an evidence for a miscibility gap (Mor-gan amp London 1999) consistent with evidence for com-plete solid solution between schorl and lithian olenite ina pegmatite from the eastern Alps (Kalt et al 2001) El-baite in the pockets has elevated F and Mn concentrations(Morgan amp London 1999) The occurrence of schorl coresin the pockets suggests that schorl grew while Fe-bearingmelt was still present in the dikes but the elbaite rims to-gether with other lithium minerals in pegmatite cores andpockets grew in equilibrium with Li-rich fluid collected inthe pockets The progressive increase in Li across the dikesis more consistent with progressive inward crystallizationof the dikes than a zone-refining process as Li appears tohave been progressively enriched in the residual liquid be-cause of its low solubility in early-crystallizing mineralsincluding schorl The change in tourmaline compositionacross the San Diego County pegmatites is analogous to

the tourmaline composition trend in the Bob Ingersol peg-matite in the Black Hills (Jolliff et al 1986)

The transition from the aplitic line-rock to pegmatiteprobably marks the point of fluid separation in the magmaFluid separation is suggested by the occurrences of elon-gated radial crystals that characterize the pegmatite zonesbecause crystals grow faster and become elongated whenH2O activity increases in the melt (Fenn 1977) HighH2O activity promotes an increase in the diffusion ratesof chemical components even when the melt is undergo-ing rapid cooling Crystallization of tourmaline (and gar-net) is controlled more by gradients in concentrations ofless mobile elements including Fe Mg and Mn than bygradients of rapidly-diffusing elements especially the al-kalis (Rockhold et al 1987 Webber et al 1999) Thelack of crystallization of Li minerals such as spodumeneand amblygonite until the core zones was probably pre-cluded by the initially low Li concentration and removalof Li by the exsolving fluid Although in Cl-absent water-peraluminous melt systems D(Li)fluidmelt is sim 04 (Londonet al 1988) D(Li)fluidmelt increases with the addition ofCl For example when an aqueous fluid at 800 C2 kbarhas sim 7 wt Cl D(Li)fluidmelt is sim 2 (Webster et al 1989)However even this D(Li)fluidmelt seems insufficient to ex-plain the very elevated Li in the pocket fluids by simplebatch partitioning between the fluid and the melt particu-larly because the fluid contained only sim 2 wt Cl Morelikely the large concentration of Li in the fluid is the re-sult of Rayleigh enrichment with crystallization dominatedby feldspars quartz and schorl tourmaline in which Li hassmall solubility compared to the melt and the fluid At 88 crystallization Li would have exceeded 05 wt (11 wtLi2O) in the melt and a correspondingly high concentrationin the accumulating fluid

Concentrations in excess of 1 wt Li2O that may haveexisted at later stages of fractional crystallization of thedikes are approached in some large spodumene-bearingpegmatite intrusions including the Tin Mountain peg-matite the Harding pegmatite in New Mexico and theTanco pegmatite in Manitoba (Norton 1994) These largepegmatites contain spodumene andor other Li-bearingminerals not only in their cores but also other zones (Nor-ton 1994) Li2O concentrations approaching 1 wt in asilicate melt may be required for crystallization of miner-als in which Li is an essential structural constituent In theSan Diego pegmatites such high concentrations apparentlyexisted only in the core zones and the pockets

The influence of the estimated initial 630 ppm Li on theviscosity of the dikes was likely far smaller in compari-son with the influence of the sim 6 wt H2O that wouldbe in the melt at the point of saturation Even in the corezones where Li2O may have reached 1 its effect wouldlikely have been much smaller 1 Li2O is equivalent to19 mole in a haplogranite melt whereas 6 wt H2Ois equivalent to sim 185 mole The addition of 1 wtof excess Li to a haplogranite melt lowers the viscosityby about one order of magnitude (Dingwell et al 1996)but in a peraluminous melt where Li may be complexedwith Al in a Si4+ = Al3+ + Li+ substitution the effect ofLi addition is probably smaller Even if Li exceeded its

912 JS Maloney et al

charge-balancing role in the melt upon removal of Al bycrystallization of the highly peraluminous tourmaline or ifAl was complexed with F in a non-network position in themelt (Giordano et al 2004 Mysen amp Richet 2005) Li isnot expected to have had a drastic effect on the viscosity ofthe dikes In comparison when H2O is added to a silicatemelt it initially dissolves by forming Al-OH and Si-OHcomplexes which depolymerize the tetrahedral networkViscosity drops by sim 45 orders of magnitude with the ad-dition of only 1 wt H2O to a peraluminous leucogranitemelt at 600 C and by an additional 6 orders of magnitudewith further addition of 5 wt H2O (Romano et al 2001Whittington et al 2004) The decreasing effect of H2O ad-dition on the viscosity comes from the increasingly greaterdissolution of H2O as a molecular species (Stolper 1982)Likewise diffusivities of ions in silicate melts dramaticallyincrease with the addition of only a small amount of H2Obut less so with further addition of H2O (Watson 1994)

Lithium isotope fractionation

A potentially large fractionation of Li-isotopes in any mul-tiphase system is due to the 17 mass difference be-tween 7Li and 6Li Presently available data show only verylimited Li isotope fractionation during crystallization ofhigh-temperature igneous systems (Tomascak et al 1999Magna et al 2006 Halama et al 2007) and during vapor-liquid separation in hydrothermal systems (Foustoukoset al 2004 Liebscher et al 2007) However significantfractionation may result from different coordination statesof Li in coexisting phases that include multiple miner-als melts and aqueous fluids in relatively low-temperaturegranitic systems (Wenger amp Armbruster 1991) In general6Li preferentially occupies sites with higher coordinationnumbers and therefore weaker bonds whereas 7Li prefer-entially enters sites with smaller coordination numbers andstronger bonds (Wunder et al 2007)

Fractionation of Li between minerals and fluids hasbeen experimentally determined only for a limited set ofminerals to date Wunder et al (2006 2007) found thatΔ7Listauroliteminusfluid = +13 and is essentially temperature-independent Δ7Lilepidoliteminusfluid is approximately ndash2 withsome temperature dependence in the 350ndash400 C rangeΔ7Lispodumeneminusfluid is also temperature-dependent but morenegative by about 3 For all three minerals the frac-tionation is insensitive to the Cl content of the fluid whichimplies that Li probably forms a tetrahedrally-coordinatedhydrated ion Li(H2O)+4 instead of a LiCl or LiOH com-plex Wunder et al (2007) concluded that equilibrium iso-topic fractionation is firstly controlled by Li coordina-tion with 7Li preferentially incorporated into the phasethat allows for a smaller coordination number and sec-ondly by the Li-O bond length giving the relationshipδ7Listaurolite gt δ

7Lilepidolite gt δ7Lispodumene In staurolite

Li substitutes for the divalent cations Fe2+ Mg and Znin the tetrahedral sites in lepidolite Li is octahedrally-coordinated between tetrahedral layers and in spodumeneit occupies the relatively large M2 octahedral site Thereare no experimental Li isotope fractionation data involv-

ing either tourmaline or melt but in tourmaline Li occupiesthe octahedrally-coordinated Y-site and in a peraluminousmelt Li is probably strongly bonded in its charge balanc-ing role with tetrahedrally-coordinated Al3+ in a LiAlSi3O8complex (Mysen amp Richet 2005) Therefore Li in such amelt should be isotopically relatively heavy

The results of Wunder et al (2007) involving muscoviteare quite different from the results of Lynton et al (2005)who found Δ7Limuscoviteminusfluid to range between +8 and +20in the 400ndash500 C interval Wunder et al (2007) attributedthe discrepancy to the diffusion mechanism that Lyntonet al (2005) used to introduce Li into muscovite IndeedTeng et al (2006a) ascribed very large variations in δ7Li ofcountry rocks in the aureole of the Tin Mountain pegmatiteto differential diffusion of the two Li isotopes which under-scores that in addition to the energy of bonds in lattices ki-netic effects may induce transient Li isotope fractionationwhich may be preserved in rapidly cooled systems

There have been no direct measurements of Li iso-tope fractionation between peraluminous silicate melts andaqueous fluids but it is expected that Li in a melt should beisotopically heavier because of strong bonds in associationwith charge balancing of Al that is in tetrahedral coordina-tion This inference is supported by isotopic compositionsof Li in fluid inclusions and host quartz in the Tin Mountainpegmatite in the Black Hills (Teng et al 2006b) Li in thefluid inclusions has much lower δ7Li values than Li in thequartz supporting the inference that 7Li prefers the strongbonds in quartz where it is possibly charge-balancing Althat is incorporated into the quartz structure and by anal-ogy bonds in high-silica melts over the weaker hydratedbonds in the fluid

The San Diego pegmatites were systems in which min-erals melts and aqueous fluids coexisted at various stagesof crystallization Assuming equilibrium the isotopic com-position of Li in tourmaline in the dikes is reflective of themedium from which the tourmaline crystallized HoweverLi isotopes may be strongly fractionated by kinetic effectsThere is a several permil heterogeneity in δ7Li even in indi-vidual zones of a single pegmatite Most interesting are thevery elevated gt 19 δ7Li values of radial tourmalines inboth the hanging and lower pegmatite zones A reason forthe elevated values may be that the tourmalines grew whenthe melt became saturated in the aqueous fluid at whichpoint Li isotopes were fractionated between the melt andthe aqueous fluid with most Li going into the fluid (Fig 5)If the isotopic composition of Li in tourmaline in each ofthe zones reflects the relative fractionation of Li isotopesbetween the melt and the fluid then the isotopically heavierLi in the pegmatite zone tourmaline is consistent with theexpected stronger Li bonds in the melt compared to bondsof hydrated Li in the fluid The similarity of δ7Li values intourmaline that crystallized in the line rock and the pock-ets is consistent with accumulation of the bulk of meltrsquosinitial Li into the pocket fluid as this would result in littlechange in the isotopic composition of Li from the initialfluid-undersatured melt to the eventually collected fluid inthe pockets Elbaite rims of pocket tourmalines have lowerδ7Li than their corresponding cores although one core-rimpair has overlapping analytical errors (Fig 4b) The lower

Lithium and its isotopes in pegmatite tourmalines 913

Fig 5 Schematic diagram illustrating a possible mechanism of pro-ducing high δ7Li tourmalines in pegmatite zones during fluid-meltseparation Because Li is used in charge-balancing with strongly-bonded Al in the silicate melt the δ7Li in the melt should be moreelevated than δ7Li in the fluid where Li probably occurs mostly asa hydrated ion (Wunder et al 2007) Tourmalines crystallizing inequilibrium with the melt and fluid respectively may reflect the iso-topic fractionation between the melt and the fluid

δ7Li values of the elbaite rims are consistent with crystal-lization of schorl cores while melt was still present andcrystallization of elbaite rims in the presence of the fluidonly

The lack of a systematic increase in δ7Li across the tex-tural zones of the dikes suggests that there was little in-fluence of tourmaline itself on the isotopic composition ofthe residual melt during crystallization Given that bothΔ7Lispodumeneminusfluid and Δ7Lilepidoliteminusfluid are both negativeΔ7Litourmalineminusmelt should be even more negative because Liin tourmaline is in octahedral coordination while in themelt it is associated with tetrahedral Al However becausethe proportion of tourmaline in the line-rock and pegmatitezones is lt 8 and the concentration of Li in the tourma-line is very small crystallization of tourmaline would havehad a negligible effect on the Li isotope ratio in the resid-ual melt Based on mass-balance calculations only sim 2 of the initial 630 ppm Li is contained in schorl in the line-rocks and the pegmatites zones

Kinetic effects

A kinetic cause for the elevated δ7Li values of the radialtourmalines in the pegmatite zones must also be consid-ered however The shape of the tourmalines suggests thatthey grew very rapidly in which case the Li isotope frac-tionation may have been kinetically controlled by differ-ential diffusion of Li isotopes at the crystal-liquid (andorcrystal-fluid) interface The relative diffusion rates of two

Fig 6 Schematic diagram illustrating a possible kinetic mechanismof producing high δ7Li radial tourmalines in pegmatite zones Be-cause the diffusion of 7Li is slower in the melt than the diffusion of6Li and when Li is an incompatible element relative to tourmaline7Li should become preferentially enriched in a boundary layer aheadof a tourmaline crystal that is growing faster than the rate at whichLi diffuses in the melt

isotopes of a given element in a silicate liquid are given by

D1

Dh=

(mh

m1

)β

where m is the mass of isotope ldquolrdquo stands for the light iso-tope and ldquohrdquo stands for the heavy isotope (Richter et al2003) Using basalt-rhyolite melt couples Richter et al(2003) experimentally determined that βLi asymp 0215 Thismeans that 6Li can diffuse substantially faster than 7Lithrough the melt away from a growing crystal so that apreferential enrichment of 7Li can potentially occur in thechemical boundary layer ahead of the crystal (Fig 6) re-sulting in elevated δ7Li The diffusion rate of Li in silicatemelts is orders of magnitude faster than the rates of othermajor and minor cations (Richter et al 2003) Under con-ditions of slow mineral growth homogeneous Li isotoperatios would be expected in the tourmalines Instead theobserved Li isotope heterogeneity suggests crystallizationconditions under which the diffusion of Li in the melt didnot keep-up with the rate of tourmaline growth

Elevated δ7Li values of pegmatites

The overall elevated δ7Li values seen in the San Diegopegmatites are similar to the values in the Tin Mountainpegmatite in the Black Hills δ7Li values in the associ-ated Harney Peak leucogranite and the host schists of theTin Mountain pegmatite have δ7Li values within a fewpermil of 0 which points to some process that leads tostrong Li isotope fractionation during generation of LCT-type pegmatite melts (Teng et al 2006b) The apparentlarge fractionation for these relatively low-temperature ig-neous systems contrasts with the minimal fractionation in

914 JS Maloney et al

high-temperature igneous systems (Tomascak 2004) Tenget al (2006b) suggested that elevated δ7Li values in peg-matite melts could potentially be acquired by crystal-liquidfractionation during crystallization of parental magmasHowever for this process to be effective a greater amountof Li would have to remain in the crystallized assemblage(parental granite) than in the residual liquid (pegmatite)This is inconsistent with for example the relative concen-trations of Li in the Harney Peak leucogranite and its poten-tially residual liquid now represented by the Tin Mountainpegmatite Li concentrations in the Harney Peak leucogran-ite range from 10ndash205 ppm (Teng et al 2006b)

A possible alternative explanation is that the structureof a hydrous pegmatite melt has more similarity to waterthan to minerals and therefore it incorporates 7Li pref-erentially over minerals in a fashion analogous to thelarge 7Li enrichment in crustal fluids compared to crustalrocks (Tomascak 2004) As hydrated pegmatite melts passthrough the crust from their sources they may acquire el-evated δ7Li values through rapid Li isotope exchange withthe surrounding rocks Matthews et al (2003) found thatthe oxygen and especially hydrogen isotopic compositionof pegmatite dikes on Naxos Greece reflects the isotopiccomposition of the host rocks which change in isotopiccomposition along strike of the dikes If hydrogen isotopescan be nearly fully exchanged between pegmatite melts andtheir host rocks then it is likely that Li isotopes can also bereadily exchanged given the fast diffusion rates of Li in sil-icate liquids (Richter et al 2003)

Acknowledgements We are grateful for access providedby Dana and Ken Gochenour to the Cryo-Genie propertyBill Calhoun to the San Diego mine property and LouisSpaulding Jr to the Little Three property Jeffrey Patter-son and Matt Taylor provided guidance around the peg-matite districts and Jim Student helped with sample col-lection Carol Nabelek oversaw the ICP-OES analysis BillMcDonough graciously gave access to JM to conduct anal-ysis in the isotope laboratory at the University of MarylandThe paper benefited from the constructive reviews of AxelLiebscher Jeffrey Ryan and Ed Grew and additional com-ments from Roberta Rudnick A Feodor-Lynen fellowshipto Halama by the Alexander von Humboldt Foundation isgratefully acknowledged The study was supported by Uni-versity of Missouri Research Board Grant D3508 and NSFGrant 408564 to Nabelek Additional funding came fromNSF Grant EAR 0606989 to Rudnick and McDonough

References

Anderko A amp Pitzer KS (1993) Equation-of-state representa-tion of phase equilibria and volumetric properties of the systemNaCl-H2O above 573 K Geochim Cosmochim Acta 57 1657-1680

Bouman C Elliott T Vroon PZ (2004) Lithium inputs to sub-duction zones Chem Geol 212 59-79

Cernyacute P (1991) Rare-element granitic pegmatites Part I Anatomyand internal evolution of pegmatite deposits GeoscienceCanada 18 49-67

Cernyacute P amp Ercit TS (2005) The classification of granitic peg-matites revisited Can Mineral 43 2005-2026

Dingwell DB Hess K-U Knoche R (1996) Granite andgranitic pegmatite melts volumes and viscosities Trans R SocEdinburgh Earth Sci 87 65-72

Fenn PM (1977) The nucleation and growth of alkali feldsparsfrom hydrous melts Can Mineral 15 135-161

Fisher J (2002) Gem and rare-element pegmatites of southernCalifornia Mineral Record 33 363-407

Fisher J Foord EE Bricker GA (1999) The geology miner-alogy and history of the Himalaya Mine Mesa Grande SanDiego County California California Geol 3-17

Foord EE (1976) Mineralogy and petrogenesis of layeredpegmatite-aplite dikes in the Mesa Grande District San DiegoCounty California PhD Dissertation Stanford University

Foord EE Starkey HC Taggard JE Jr (1986) Mineralogyand paragenesis of ldquopocketrdquo clays and associated minerals incomplex granitic pegmatites San Diego County California AmMineral 71 428-439

Foustoukos DI James RH Berndt ME Seyfried WE Jr(2004) Lithium isotopic systematics of hydrothermal vent flu-ids at the Main Endeavour Field Northern Juan de Fuca RidgeChem Geol 212 17-26

Giordano D Romano C Dingwell DB Poe B Behrens H(2004) The combined effects of water and fluorine on theviscosity of silicic magmas Geochim Cosmochim Acta 685159-5168

Halama R McDonough WF Rudnick RL Keller J KlaudiusJ (2007) The Li isotopic composition of Oldoinyo LengaiNature of the mantle sources and lack of isotopic fractionationduring carbonatite petrogenesis Earth Planet Sci Lett 25477-89

Holtz F Behrens H Dingwell DB Johannes W (1995) Watersolubility in haplogranite melts Compositional pressure andtemperature dependence Am Mineral 80 94-108

Jahns R H (1979) Gem-bearing pegmatites in San Diego CountyCalifornia The Stewart mine Pala district and the Himalayamine Mesa Grande district in ldquoMesozoic crystalline rocksPeninsular Ranges batholith and pegmatites Point Sol ophio-literdquo P L Abbott amp V R Todd eds San Diego State UniversitySan Diego California 3-38

Jahns RH amp Burnham CW (1969) Experimental studies of peg-matite genesis I A model for the derivation and crystallizationof granitic pegmatites Econ Geol 64 843-864

Jahns RH amp Tuttle OF (1963) Layered pegmatite-aplite intru-sives Mineral Soc Am Sp Pap 1 78-92

Jolliff BL Papike JJ Shearer CK Laul JC (1986)Tourmaline as a recorder of pegmatite evolution Bob Ingersolpegmatite Black Hills South Dakota Am Mineral 71 472-500

Kalt A Schreyer W Ludwig T Prowatke S Bernhardt H ErtlA (2001) Complete solid solution between magnesian schorland lithian excess-boron olenite in a pegmatite from the Koralpe(eastern Alps Austria) Eur J Mineral 13 1191-1205

Kampf AR Gochenour K Clanin J (2003) Tourmaline dis-covery at the Cryo-Genie mine San Diego County CaliforniaRocks and Minerals 78 156-163

Liebscher A Meixner A Romer RL Heinrich W (2007)Experimental calibration of the vapourndashliquid phase relationsand lithium isotope fractionation in the system H2OndashLiCl at400 Geofluids 7 1-7

Lithium and its isotopes in pegmatite tourmalines 907

Table 1 Tourmaline locations shapes and Li concentrations and isotope ratios

Sample Zone Tourmaline shape Li (ppm) δ7Li ()CG-1a middle hanging pegmatite radial 53 192CG-1c core zone large blade 174CG-1e core zone large blade 138CG-1i black pocket prism 273 159CG-1i green pocket prism 5075 147CG-1n pocket prism 416 112CG-3e core zone large blade 265 161CG-3i core zone large blade 663 112CG-3k middle hanging pegmatite radial 421CG-5b top border short blade 22CG-5d middle hanging pegmatite radial 93LT-1c core zone large blade 517 134LT-3f bottom line rock small prismatic 56 151LT-3g bottom line rock small prismatic 66LT-3i bottom line rock small prismatic 70 123LT-4c core zone large blade 268 148LT-5c bottom pegmatite radial 452 229LT-5d bottom pegmatite radial 113 191LT-5e bottom pegmatite small prismatic 135 142HM-2C black pocket prism 641 159HM-2c green pocket prism 1456 137HM-4C pocket prism 954 159

Fig 2 Sketches showing sample locations at the Cryo-Genie pegmatite dike (a) Green Ledge location on the surface and (b) Payday pocketunderground

908 JS Maloney et al

the dike from the upper contact to the core zone includinga pocket is exposed CG-5b is a short euhedral tourmalinefrom the border of the hanging pegmatite CG-1a and CG-5d are radially-grown tourmaline needles from the hangingpegmatite and CG-1c and CG-1e are large euhedral tour-malines from the core zone CG-1i and CG-1n are tourma-line prisms that grew into the pocket CG-1i has a schorlcore and green elbaite rim CG-3e and CG-3i are schorlblades from core zone around the Payday pocket whereasCG-3k is radial schorl from the pegmatite above the pocket

The Little Three dike system intrudes the Green Valleytonalite-gabbro of the western zone of the PRB (Stern et al1986) The system has five dikes including the Little Threemain dike and the Spaulding dike that were sampled forthis study The dikes vary between 1 and 2 m in thick-ness along strike The relative thickness of line-rock andpegmatite zones also varies Most of the analyzed samplescome from the Little Three main dike near the mine en-trance on Topaz Ledge (Fig 3) The bottom boundary ofthe dike with the country rock is a homogeneous aplite withline-rock above it The line rock terminates sharply againstthe lower pegmatite zone which contains small prismaticor bladed tourmaline within graphic intergrowths of quartzand feldspar

Samples LT-3f LT-3g and LT-3i come from the line-rock(Fig 3a) Samples LT-5c and 5d are radial tourmalines thatgrew in the lower pegmatite that cuts through the line rockalong strike from location 3 (Fig 3b) LT-5e is a prismatictourmaline in a smaller dike cutting through the line rockSample LT-4c is a large tourmaline blade from the corezone around a several-meter size pocket that occurs abovethe mine entrance Sample LT-1c is a similar tourmalinefrom the core zone of the Spaulding dike

The Himalaya dike system includes two subparallel dikeshosted by the San Marcos gabbro (Fisher et al 1999Webber et al 1999) The two dikes are separated by 3ndash10 m along the 915 m of their exposed length but convergeat the San Diego mine where the samples were collectedThe Himalaya dikes were emplaced sim 100 my ago asdated by fission track and K-Ar methods (Foord 1976)Samples HM-2c and HM-4c come from pockets in the un-derground workings of the San Diego mine TourmalineHM-2c includes a schorl core and a green elbaite rim whileHM-4c is exclusively schorl

The Cryo-Genie samples best represent a cross-sectionacross the hanging portion of a pegmatite dike while theLittle Three samples best represent the lower portion of adike The Himalaya samples were analyzed primarily to de-termine if there is a regional variability in Li isotopic com-positions

Methods

Bulk Li concentrations

Tourmaline is extremely difficult to dissolve in acids andtherefore we have developed a flux procedure that ac-complishes complete dissolution of ground-up tourmalineBlack cores and green rims of zoned pocket tourmalines

Fig 3 Sketches showing sample locations at the Little Three peg-matite dike Parts a) and b) are along strike on the main dike

were prepared separately When samples had only smallcrystals multiple grains were ground together 100 mgof each tourmaline powder were mixed with 400 mg ofground K2CO3 flux in a zirconia crucible and placed into afurnace at 500 C The temperature was then increased to900 C for 15 min after which the furnace was turned offThe crucibles were left in the furnace overnight to slowlycool to prevent cracking Each fluxed sample was first cen-trifuged in 15 ml of 10 HNO3 for four minutes Theliquid was then decanted into a Teflon beaker and stirredon a hot plate The remaining solid residue was again cen-trifuged in 15 ml of 10 HNO3 Any residue that was stillleft was dissolved in 3 ml of concentrated HNO3 and wasthen added to the already dissolved tourmaline and stirredfor additional 15 minutes on low heat The sample was thenbrought up to 50 ml in a volumetric flask with 10 HNO3In several samples a brown amorphous precipitate formedafter several days After the liquids were decanted the pre-cipitates were dissolved in a HF-HNO3 mixture and thendiluted to 50 ml with distilled water Analysis of these so-lutions revealed no Li

Lithium concentrations (Table 1) were analyzed using thePerkin-Elmer Optima 3300 Inductively Coupled Plasma

Lithium and its isotopes in pegmatite tourmalines 909

Optical Emission Spectrometer (ICP-OES) at the Univer-sity of Missouri Synthetic standards with 005 010 1and 5 ppm Li were prepared using blank K2CO3 flux so-lutions to create the calibration curves for Li Instrumentaldrift was accounted for by analyzing the prepared standardsat regular intervals during the analysis and aliquots of thestandard solutions were used as check standards Lithiumconcentrations were obtained from the strong 610362 nmemission line

Li isotope ratios

Analysis of Li isotope ratios (Table 1) was carried out ondried aliquots of solutions prepared for ICP-OES analy-sis The dried samples were prepared and analyzed at theGeochemistry Laboratory of the University of Maryland-College Park The preparation followed the procedure out-lined in Rudnick et al (2004) and Teng et al (2004) whichis based on the three-column procedure of Moriguti ampNakumura (1998) Because NaLi ratio in solution largerthan sim 5 may cause instability in the analysis the NaLiratio of each sample was determined semi-quantitativelyprior to analysis and excess Na was stripped-off by addi-tional column purification Measurements were done on aNu Plasma Multicollector Inductively Couple Plasma MassSpectrometer (MC-ICP-MS) Each analysis consisted oftwo blocks of twenty individual measurements Each anal-ysis was bracketed by the measurement of a 100 ppb L-SVEC standard The 7Li6Li ratio in L-SVEC for each2 times 20 measurements had an average 2σ of the mean 0003 The external precision better than 1 Two otherLi isotope standards IRMM-016 (Qi et al 1997) and thein-house standard UMD-1 were routinely analyzed dur-ing each analytical session The results for both (IRMM-016 +02 plusmn 04 UMD-1 +543 plusmn 02) agree well withpublished results (IRMM-016 ndash01 plusmn 02 to +02 plusmn 08UMD-1 +547 plusmn 10 Rudnick et al 2004 Teng et al2004 Halama et al 2007) Two USGS rock standardsBHVO-1 (+42 ) and QLO-1 (+66 ) were analyzedfor quality-control purposes The value for BHVO-1 waswithin the uncertainty of previously published results (+43to +58 Bouman et al 2004 Rudnick et al 2004) and thevalue for QLO-1 was within the range of previous analy-ses at the University of Maryland (+56 to +68 Halamaunpublished data)

Results

Li concentrations

Li concentrations in the tourmalines (Table 1) are shown inFig 4a They are plotted in terms of the zones in which thetourmalines grew There appear to be no systematic differ-ences in Li concentrations between corresponding zones ofthe different dikes based on the available data in spite ofthe spatial and probably some temporal separation of thedikes although with denser data sets for all three dikessome systematic differences could appear Nevertheless

Fig 4 (a) Li concentrations and (b) Li isotopic ratios in tourma-lines in three pegmatite dikes Data are grouped according to tex-tural zones from which samples came from Schorl tourmalines areshown by black symbols and elbaite rims on schorl by white sym-bols Two core-rim pairs are shown by connecting lines

the similar ranges in Li concentrations in the pegmatite andthe core zones of both the Cryo-Genie and the Little Threedikes for example point to a common petrogenetic processof the dikes that is underscored by their similar tectoniccontext the similar style of emplacement and the similartextural and mineral zoning across the dikes

There is a progressive increase of about two orders ofmagnitude in Li concentrations across the zones of individ-ual dikes For example in the Cryo-Genie dike concentra-tions range from 22 ppm to 5075 ppm The lowest Li con-centrations 70 ppm are in tourmalines in the line-rockof the Little Three main dike and the top border zone of the

910 JS Maloney et al

Cryo-Genie dike These tourmalines are at most few mil-limeters in length and are prismatic Li concentrations inradial tourmalines in the pegmatite zones of both the Cryo-Genie and Little Three dikes and in large tourmaline bladesin the core zones are on average higher by about an orderof magnitude The concentrations are highly variable how-ever In prismatic schorl crystals within pockets the con-centrations range from 273 to 954 ppm Two green over-growths on schorl cores have the highest Li concentrations1456 and 5075 ppm

Li isotope ratios

In contrast to the progressive increase in Li concentrationsin tourmalines across the zones of the dikes the variationin δ7Li is more complex (Fig 4b) There is no correlationwith Li concentrations but again there appears to be nosystematic difference in δ7Li between the three dikes Intourmalines from the wall-zoneline-rock core zone andpockets δ7Li has approximately the same range between112 and 161 In the pockets elbaite rims have lighterLi than schorl cores although the difference in the Cryo-Genie pair is within overlapping errors In three tourma-lines from the upper and the lower pegmatite zones ofboth Cryo-Genie and Little Three dikes δ7Li is very highgt 19 These three tourmalines came from radial splayswhereas others with lighter Li were prismatic or blade-shaped (Fig 3b) δ7Li in prismatic tourmaline from thebottom pegmatite zone of the Little Three dike is lower at142

The obtained δ7Li values and their range are among thehighest measured in rocks δ7Li in most unaltered crustalrocks typically does not exceed 10 (Tomascak 2004Teng et al 2004) but values up to 19 in quartz anda 10 difference between quartz and albite have beenreported for the Tin Mountain pegmatite in the Black HillsSouth Dakota USA (Teng et al 2006b)

Initial Li concentration in pegmatite melts

Lithium is often invoked as an element that may contributeto growth of large crystals that characterize pegmatites bydepolymerizing the silicate melt structure and thereforelowering its viscosity and increasing chemical diffusionrates In order to evaluate the fluxing effect of Li comparedto H2O in the San Diego County pegmatites a crude es-timate was made of its initial concentration in the meltsThe estimate comes from the Li concentrations in tourma-line the modal abundance of tourmaline and the Li con-centrations in fluid inclusions in pocket quartz assumingthat the pockets were the collection volumes for the sep-arated fluid Tourmaline is the largest reservoir for Li inthe line rock and pegmatite but in the cores and pocketswhere other Li-bearing minerals including lepidolite andamblygonite occur Li concentration in the fluid is best es-timated from primary fluid inclusions in quartz In the Lit-tle Three pegmatite these inclusions have a range of 7000 to12 000 ppm with average of 9200 ppm Li (unpublished data

determined by ion chromatography) The inclusion fluidscontain sim 3 wt NaCl based on microthermometric mea-surements

The initial concentration of Li in the pegmatite melts (Co)can be estimated using

Co = XLRCLR + XPEGCPEG + XPOCKCPOCK

where X is the mass proportion of each zone and C isthe concentration of Li in each zone Li concentrationsin the line-rock and the pegmatite zones were determinedfrom the average of Li concentrations in tourmaline in eachzone and the modal proportions of tourmaline Data for theCryo-Genie and Little Three dikes were combined Com-bining the data for the two dikes is justified because in thepegmatite zones which are volumetrically most abundantLi concentrations cover the same ranges The average Liconcentration in tourmaline in the Cryo-Genie pegmatitezones is 258 ppm and in the Little Three it is 298 ppmThe mass proportions of the three zones were determinedfrom their relative volumes and densities 2700 kgm3 forthe rocks and 500 kgm3 for the fluid The fluid density isappropriate for H2O with 3 wt NaCl at 2 kbar and 400 C(Anderko amp Pitzer 1993) The proportion of tourmaline inthe line-rock and pegmatite zones is lt 8 volume basedon image analysis of cut slabs The estimated bulk Li con-centration in the line-rock zones is only sim 5 ppm and in thepegmatite zones only sim 60 ppm It is readily evident thatmost Li from the original melt ended-up in the fluids

The pockets show that the exsolved fluid was collectedin discrete spaces instead of one continuous space betweenthe hanging and lower portions of the dikes Because thevolume and the distribution of the pockets are highly vari-able (Jahns 1979 Stern et al 1986) the volume of theexsolved fluid is difficult to estimate from the field occur-rences of the pockets A better estimate of the volume ofthe exsolved fluid comes from the maximum H2O solubil-ity in silicate melts which at 2 kbar is sim 6 wt (Holtzet al 1995) This amount of H2O given its molar volumeat 400 C and 2 kbar would occupy sim 28 of the chambervolume Using this proportion the calculated initial Li con-centration in the pegmatite dikes is only sim 630 ppm If theCryo-Genie data is left out of the calculation the differencein the result is only 1 ppm Although the ranges of the mea-sured Li concentrations and errors in the volume estimatesof tourmaline and the textural zones contribute to an erroron this estimate the estimate is dominated by the calculatedfluid volume and the Li concentration in the fluid as deter-mined from fluid inclusions Although 630 ppm Li couldappear to be a rather small concentration for a LCT-typemelt it is 18 times greater than the estimated average con-centration of 35 ppm in the upper continental crust (Tenget al 2004)

Discussion

Role of lithium in the pegmatite crystallization process

Since the classic model of Jahns amp Burnham (1969)for crystallization of pegmatite melts which involves the

Lithium and its isotopes in pegmatite tourmalines 911

separation of a fluid phase from a silicate melt it has beenrecognized that the formation of pegmatites also dependson kinetic controls during rapid cooling of dikes in par-ticular enrichments of components in melt boundary lay-ers ahead of rapidly crystallizing minerals (eg Rockholdet al 1987 London 1992 Webber et al 1999) In addi-tion to being responsible for crystallization of minerals inwhich it is an essential structural constituent Li is ofteninvoked as a possible component that along with H2O BF and other species can potentially flux granite melts Forexample fractional crystallization involving Li was partlyresponsible for mineral zoning seen in the large LCT-typeTin Mountain pegmatite in the Black Hills of South Dakotain which the outer zones are dominated by feldspars andthe inner zones by spodumene and quartz (Walker et al1986) Trapping temperatures of primary fluid inclusionsare lt 400 C and nearly invariable across the pegmatitesuggesting that it crystallized nearly isothermally as an un-dercooled liquid (Sirbescu amp Nabelek 2003a)

An extreme kinetic model is that of Morgan amp London(1999) for crystallization of the Little Three pegmatiteThey suggested that the low temperature and the fast cool-ing rate that must have occurred during solidification of thepegmatite did not allow for crystal nucleation until the melthas reached a glass state at sim 250 C below the equilibriumliquidus They proposed a constitutional zone-refining pro-cess in which a fluxed crystallization front swept a F Liand Mn-rich boundary layer through the solid or semi-soliddike eventually resulting in enrichment of these elementsin the pocket zone

Although we did not obtain electron microprobe data ontourmaline from the dikes the electron microprobe dataof Morgan and London (1999) on tourmaline in the Lit-tle Three dike suggests that the progressive increase in Liconcentrations from the line-rock and the upper wall-zoneto the pockets (Fig 4a) corresponds to changes in othertourmaline components Morgan amp London (1999) foundthat across the line-rock and the pegmatite zones Mg de-creases from sim 07 to near 0 per formula unit while Fe staysnearly constant The increase in Li and the decrease in Mgsuggests an increasing exchange of the elbaite componentfor the dravite component In the pockets tourmaline iszoned from schorl to elbaite but the zoning appears to becontinuous without an evidence for a miscibility gap (Mor-gan amp London 1999) consistent with evidence for com-plete solid solution between schorl and lithian olenite ina pegmatite from the eastern Alps (Kalt et al 2001) El-baite in the pockets has elevated F and Mn concentrations(Morgan amp London 1999) The occurrence of schorl coresin the pockets suggests that schorl grew while Fe-bearingmelt was still present in the dikes but the elbaite rims to-gether with other lithium minerals in pegmatite cores andpockets grew in equilibrium with Li-rich fluid collected inthe pockets The progressive increase in Li across the dikesis more consistent with progressive inward crystallizationof the dikes than a zone-refining process as Li appears tohave been progressively enriched in the residual liquid be-cause of its low solubility in early-crystallizing mineralsincluding schorl The change in tourmaline compositionacross the San Diego County pegmatites is analogous to

the tourmaline composition trend in the Bob Ingersol peg-matite in the Black Hills (Jolliff et al 1986)

The transition from the aplitic line-rock to pegmatiteprobably marks the point of fluid separation in the magmaFluid separation is suggested by the occurrences of elon-gated radial crystals that characterize the pegmatite zonesbecause crystals grow faster and become elongated whenH2O activity increases in the melt (Fenn 1977) HighH2O activity promotes an increase in the diffusion ratesof chemical components even when the melt is undergo-ing rapid cooling Crystallization of tourmaline (and gar-net) is controlled more by gradients in concentrations ofless mobile elements including Fe Mg and Mn than bygradients of rapidly-diffusing elements especially the al-kalis (Rockhold et al 1987 Webber et al 1999) Thelack of crystallization of Li minerals such as spodumeneand amblygonite until the core zones was probably pre-cluded by the initially low Li concentration and removalof Li by the exsolving fluid Although in Cl-absent water-peraluminous melt systems D(Li)fluidmelt is sim 04 (Londonet al 1988) D(Li)fluidmelt increases with the addition ofCl For example when an aqueous fluid at 800 C2 kbarhas sim 7 wt Cl D(Li)fluidmelt is sim 2 (Webster et al 1989)However even this D(Li)fluidmelt seems insufficient to ex-plain the very elevated Li in the pocket fluids by simplebatch partitioning between the fluid and the melt particu-larly because the fluid contained only sim 2 wt Cl Morelikely the large concentration of Li in the fluid is the re-sult of Rayleigh enrichment with crystallization dominatedby feldspars quartz and schorl tourmaline in which Li hassmall solubility compared to the melt and the fluid At 88 crystallization Li would have exceeded 05 wt (11 wtLi2O) in the melt and a correspondingly high concentrationin the accumulating fluid

Concentrations in excess of 1 wt Li2O that may haveexisted at later stages of fractional crystallization of thedikes are approached in some large spodumene-bearingpegmatite intrusions including the Tin Mountain peg-matite the Harding pegmatite in New Mexico and theTanco pegmatite in Manitoba (Norton 1994) These largepegmatites contain spodumene andor other Li-bearingminerals not only in their cores but also other zones (Nor-ton 1994) Li2O concentrations approaching 1 wt in asilicate melt may be required for crystallization of miner-als in which Li is an essential structural constituent In theSan Diego pegmatites such high concentrations apparentlyexisted only in the core zones and the pockets

The influence of the estimated initial 630 ppm Li on theviscosity of the dikes was likely far smaller in compari-son with the influence of the sim 6 wt H2O that wouldbe in the melt at the point of saturation Even in the corezones where Li2O may have reached 1 its effect wouldlikely have been much smaller 1 Li2O is equivalent to19 mole in a haplogranite melt whereas 6 wt H2Ois equivalent to sim 185 mole The addition of 1 wtof excess Li to a haplogranite melt lowers the viscosityby about one order of magnitude (Dingwell et al 1996)but in a peraluminous melt where Li may be complexedwith Al in a Si4+ = Al3+ + Li+ substitution the effect ofLi addition is probably smaller Even if Li exceeded its

912 JS Maloney et al

charge-balancing role in the melt upon removal of Al bycrystallization of the highly peraluminous tourmaline or ifAl was complexed with F in a non-network position in themelt (Giordano et al 2004 Mysen amp Richet 2005) Li isnot expected to have had a drastic effect on the viscosity ofthe dikes In comparison when H2O is added to a silicatemelt it initially dissolves by forming Al-OH and Si-OHcomplexes which depolymerize the tetrahedral networkViscosity drops by sim 45 orders of magnitude with the ad-dition of only 1 wt H2O to a peraluminous leucogranitemelt at 600 C and by an additional 6 orders of magnitudewith further addition of 5 wt H2O (Romano et al 2001Whittington et al 2004) The decreasing effect of H2O ad-dition on the viscosity comes from the increasingly greaterdissolution of H2O as a molecular species (Stolper 1982)Likewise diffusivities of ions in silicate melts dramaticallyincrease with the addition of only a small amount of H2Obut less so with further addition of H2O (Watson 1994)

Lithium isotope fractionation

A potentially large fractionation of Li-isotopes in any mul-tiphase system is due to the 17 mass difference be-tween 7Li and 6Li Presently available data show only verylimited Li isotope fractionation during crystallization ofhigh-temperature igneous systems (Tomascak et al 1999Magna et al 2006 Halama et al 2007) and during vapor-liquid separation in hydrothermal systems (Foustoukoset al 2004 Liebscher et al 2007) However significantfractionation may result from different coordination statesof Li in coexisting phases that include multiple miner-als melts and aqueous fluids in relatively low-temperaturegranitic systems (Wenger amp Armbruster 1991) In general6Li preferentially occupies sites with higher coordinationnumbers and therefore weaker bonds whereas 7Li prefer-entially enters sites with smaller coordination numbers andstronger bonds (Wunder et al 2007)

Fractionation of Li between minerals and fluids hasbeen experimentally determined only for a limited set ofminerals to date Wunder et al (2006 2007) found thatΔ7Listauroliteminusfluid = +13 and is essentially temperature-independent Δ7Lilepidoliteminusfluid is approximately ndash2 withsome temperature dependence in the 350ndash400 C rangeΔ7Lispodumeneminusfluid is also temperature-dependent but morenegative by about 3 For all three minerals the frac-tionation is insensitive to the Cl content of the fluid whichimplies that Li probably forms a tetrahedrally-coordinatedhydrated ion Li(H2O)+4 instead of a LiCl or LiOH com-plex Wunder et al (2007) concluded that equilibrium iso-topic fractionation is firstly controlled by Li coordina-tion with 7Li preferentially incorporated into the phasethat allows for a smaller coordination number and sec-ondly by the Li-O bond length giving the relationshipδ7Listaurolite gt δ

7Lilepidolite gt δ7Lispodumene In staurolite

Li substitutes for the divalent cations Fe2+ Mg and Znin the tetrahedral sites in lepidolite Li is octahedrally-coordinated between tetrahedral layers and in spodumeneit occupies the relatively large M2 octahedral site Thereare no experimental Li isotope fractionation data involv-

ing either tourmaline or melt but in tourmaline Li occupiesthe octahedrally-coordinated Y-site and in a peraluminousmelt Li is probably strongly bonded in its charge balanc-ing role with tetrahedrally-coordinated Al3+ in a LiAlSi3O8complex (Mysen amp Richet 2005) Therefore Li in such amelt should be isotopically relatively heavy

The results of Wunder et al (2007) involving muscoviteare quite different from the results of Lynton et al (2005)who found Δ7Limuscoviteminusfluid to range between +8 and +20in the 400ndash500 C interval Wunder et al (2007) attributedthe discrepancy to the diffusion mechanism that Lyntonet al (2005) used to introduce Li into muscovite IndeedTeng et al (2006a) ascribed very large variations in δ7Li ofcountry rocks in the aureole of the Tin Mountain pegmatiteto differential diffusion of the two Li isotopes which under-scores that in addition to the energy of bonds in lattices ki-netic effects may induce transient Li isotope fractionationwhich may be preserved in rapidly cooled systems

There have been no direct measurements of Li iso-tope fractionation between peraluminous silicate melts andaqueous fluids but it is expected that Li in a melt should beisotopically heavier because of strong bonds in associationwith charge balancing of Al that is in tetrahedral coordina-tion This inference is supported by isotopic compositionsof Li in fluid inclusions and host quartz in the Tin Mountainpegmatite in the Black Hills (Teng et al 2006b) Li in thefluid inclusions has much lower δ7Li values than Li in thequartz supporting the inference that 7Li prefers the strongbonds in quartz where it is possibly charge-balancing Althat is incorporated into the quartz structure and by anal-ogy bonds in high-silica melts over the weaker hydratedbonds in the fluid

The San Diego pegmatites were systems in which min-erals melts and aqueous fluids coexisted at various stagesof crystallization Assuming equilibrium the isotopic com-position of Li in tourmaline in the dikes is reflective of themedium from which the tourmaline crystallized HoweverLi isotopes may be strongly fractionated by kinetic effectsThere is a several permil heterogeneity in δ7Li even in indi-vidual zones of a single pegmatite Most interesting are thevery elevated gt 19 δ7Li values of radial tourmalines inboth the hanging and lower pegmatite zones A reason forthe elevated values may be that the tourmalines grew whenthe melt became saturated in the aqueous fluid at whichpoint Li isotopes were fractionated between the melt andthe aqueous fluid with most Li going into the fluid (Fig 5)If the isotopic composition of Li in tourmaline in each ofthe zones reflects the relative fractionation of Li isotopesbetween the melt and the fluid then the isotopically heavierLi in the pegmatite zone tourmaline is consistent with theexpected stronger Li bonds in the melt compared to bondsof hydrated Li in the fluid The similarity of δ7Li values intourmaline that crystallized in the line rock and the pock-ets is consistent with accumulation of the bulk of meltrsquosinitial Li into the pocket fluid as this would result in littlechange in the isotopic composition of Li from the initialfluid-undersatured melt to the eventually collected fluid inthe pockets Elbaite rims of pocket tourmalines have lowerδ7Li than their corresponding cores although one core-rimpair has overlapping analytical errors (Fig 4b) The lower

Lithium and its isotopes in pegmatite tourmalines 913

Fig 5 Schematic diagram illustrating a possible mechanism of pro-ducing high δ7Li tourmalines in pegmatite zones during fluid-meltseparation Because Li is used in charge-balancing with strongly-bonded Al in the silicate melt the δ7Li in the melt should be moreelevated than δ7Li in the fluid where Li probably occurs mostly asa hydrated ion (Wunder et al 2007) Tourmalines crystallizing inequilibrium with the melt and fluid respectively may reflect the iso-topic fractionation between the melt and the fluid

δ7Li values of the elbaite rims are consistent with crystal-lization of schorl cores while melt was still present andcrystallization of elbaite rims in the presence of the fluidonly

The lack of a systematic increase in δ7Li across the tex-tural zones of the dikes suggests that there was little in-fluence of tourmaline itself on the isotopic composition ofthe residual melt during crystallization Given that bothΔ7Lispodumeneminusfluid and Δ7Lilepidoliteminusfluid are both negativeΔ7Litourmalineminusmelt should be even more negative because Liin tourmaline is in octahedral coordination while in themelt it is associated with tetrahedral Al However becausethe proportion of tourmaline in the line-rock and pegmatitezones is lt 8 and the concentration of Li in the tourma-line is very small crystallization of tourmaline would havehad a negligible effect on the Li isotope ratio in the resid-ual melt Based on mass-balance calculations only sim 2 of the initial 630 ppm Li is contained in schorl in the line-rocks and the pegmatites zones

Kinetic effects

A kinetic cause for the elevated δ7Li values of the radialtourmalines in the pegmatite zones must also be consid-ered however The shape of the tourmalines suggests thatthey grew very rapidly in which case the Li isotope frac-tionation may have been kinetically controlled by differ-ential diffusion of Li isotopes at the crystal-liquid (andorcrystal-fluid) interface The relative diffusion rates of two

Fig 6 Schematic diagram illustrating a possible kinetic mechanismof producing high δ7Li radial tourmalines in pegmatite zones Be-cause the diffusion of 7Li is slower in the melt than the diffusion of6Li and when Li is an incompatible element relative to tourmaline7Li should become preferentially enriched in a boundary layer aheadof a tourmaline crystal that is growing faster than the rate at whichLi diffuses in the melt

isotopes of a given element in a silicate liquid are given by

D1

Dh=

(mh

m1

)β

where m is the mass of isotope ldquolrdquo stands for the light iso-tope and ldquohrdquo stands for the heavy isotope (Richter et al2003) Using basalt-rhyolite melt couples Richter et al(2003) experimentally determined that βLi asymp 0215 Thismeans that 6Li can diffuse substantially faster than 7Lithrough the melt away from a growing crystal so that apreferential enrichment of 7Li can potentially occur in thechemical boundary layer ahead of the crystal (Fig 6) re-sulting in elevated δ7Li The diffusion rate of Li in silicatemelts is orders of magnitude faster than the rates of othermajor and minor cations (Richter et al 2003) Under con-ditions of slow mineral growth homogeneous Li isotoperatios would be expected in the tourmalines Instead theobserved Li isotope heterogeneity suggests crystallizationconditions under which the diffusion of Li in the melt didnot keep-up with the rate of tourmaline growth

Elevated δ7Li values of pegmatites

The overall elevated δ7Li values seen in the San Diegopegmatites are similar to the values in the Tin Mountainpegmatite in the Black Hills δ7Li values in the associ-ated Harney Peak leucogranite and the host schists of theTin Mountain pegmatite have δ7Li values within a fewpermil of 0 which points to some process that leads tostrong Li isotope fractionation during generation of LCT-type pegmatite melts (Teng et al 2006b) The apparentlarge fractionation for these relatively low-temperature ig-neous systems contrasts with the minimal fractionation in

914 JS Maloney et al

high-temperature igneous systems (Tomascak 2004) Tenget al (2006b) suggested that elevated δ7Li values in peg-matite melts could potentially be acquired by crystal-liquidfractionation during crystallization of parental magmasHowever for this process to be effective a greater amountof Li would have to remain in the crystallized assemblage(parental granite) than in the residual liquid (pegmatite)This is inconsistent with for example the relative concen-trations of Li in the Harney Peak leucogranite and its poten-tially residual liquid now represented by the Tin Mountainpegmatite Li concentrations in the Harney Peak leucogran-ite range from 10ndash205 ppm (Teng et al 2006b)

A possible alternative explanation is that the structureof a hydrous pegmatite melt has more similarity to waterthan to minerals and therefore it incorporates 7Li pref-erentially over minerals in a fashion analogous to thelarge 7Li enrichment in crustal fluids compared to crustalrocks (Tomascak 2004) As hydrated pegmatite melts passthrough the crust from their sources they may acquire el-evated δ7Li values through rapid Li isotope exchange withthe surrounding rocks Matthews et al (2003) found thatthe oxygen and especially hydrogen isotopic compositionof pegmatite dikes on Naxos Greece reflects the isotopiccomposition of the host rocks which change in isotopiccomposition along strike of the dikes If hydrogen isotopescan be nearly fully exchanged between pegmatite melts andtheir host rocks then it is likely that Li isotopes can also bereadily exchanged given the fast diffusion rates of Li in sil-icate liquids (Richter et al 2003)

Acknowledgements We are grateful for access providedby Dana and Ken Gochenour to the Cryo-Genie propertyBill Calhoun to the San Diego mine property and LouisSpaulding Jr to the Little Three property Jeffrey Patter-son and Matt Taylor provided guidance around the peg-matite districts and Jim Student helped with sample col-lection Carol Nabelek oversaw the ICP-OES analysis BillMcDonough graciously gave access to JM to conduct anal-ysis in the isotope laboratory at the University of MarylandThe paper benefited from the constructive reviews of AxelLiebscher Jeffrey Ryan and Ed Grew and additional com-ments from Roberta Rudnick A Feodor-Lynen fellowshipto Halama by the Alexander von Humboldt Foundation isgratefully acknowledged The study was supported by Uni-versity of Missouri Research Board Grant D3508 and NSFGrant 408564 to Nabelek Additional funding came fromNSF Grant EAR 0606989 to Rudnick and McDonough

References

Anderko A amp Pitzer KS (1993) Equation-of-state representa-tion of phase equilibria and volumetric properties of the systemNaCl-H2O above 573 K Geochim Cosmochim Acta 57 1657-1680

Bouman C Elliott T Vroon PZ (2004) Lithium inputs to sub-duction zones Chem Geol 212 59-79

Cernyacute P (1991) Rare-element granitic pegmatites Part I Anatomyand internal evolution of pegmatite deposits GeoscienceCanada 18 49-67

Cernyacute P amp Ercit TS (2005) The classification of granitic peg-matites revisited Can Mineral 43 2005-2026

Dingwell DB Hess K-U Knoche R (1996) Granite andgranitic pegmatite melts volumes and viscosities Trans R SocEdinburgh Earth Sci 87 65-72

Fenn PM (1977) The nucleation and growth of alkali feldsparsfrom hydrous melts Can Mineral 15 135-161

Fisher J (2002) Gem and rare-element pegmatites of southernCalifornia Mineral Record 33 363-407

Fisher J Foord EE Bricker GA (1999) The geology miner-alogy and history of the Himalaya Mine Mesa Grande SanDiego County California California Geol 3-17

Foord EE (1976) Mineralogy and petrogenesis of layeredpegmatite-aplite dikes in the Mesa Grande District San DiegoCounty California PhD Dissertation Stanford University

Foord EE Starkey HC Taggard JE Jr (1986) Mineralogyand paragenesis of ldquopocketrdquo clays and associated minerals incomplex granitic pegmatites San Diego County California AmMineral 71 428-439

Foustoukos DI James RH Berndt ME Seyfried WE Jr(2004) Lithium isotopic systematics of hydrothermal vent flu-ids at the Main Endeavour Field Northern Juan de Fuca RidgeChem Geol 212 17-26

Giordano D Romano C Dingwell DB Poe B Behrens H(2004) The combined effects of water and fluorine on theviscosity of silicic magmas Geochim Cosmochim Acta 685159-5168

Halama R McDonough WF Rudnick RL Keller J KlaudiusJ (2007) The Li isotopic composition of Oldoinyo LengaiNature of the mantle sources and lack of isotopic fractionationduring carbonatite petrogenesis Earth Planet Sci Lett 25477-89

Holtz F Behrens H Dingwell DB Johannes W (1995) Watersolubility in haplogranite melts Compositional pressure andtemperature dependence Am Mineral 80 94-108

Jahns R H (1979) Gem-bearing pegmatites in San Diego CountyCalifornia The Stewart mine Pala district and the Himalayamine Mesa Grande district in ldquoMesozoic crystalline rocksPeninsular Ranges batholith and pegmatites Point Sol ophio-literdquo P L Abbott amp V R Todd eds San Diego State UniversitySan Diego California 3-38

Jahns RH amp Burnham CW (1969) Experimental studies of peg-matite genesis I A model for the derivation and crystallizationof granitic pegmatites Econ Geol 64 843-864

Jahns RH amp Tuttle OF (1963) Layered pegmatite-aplite intru-sives Mineral Soc Am Sp Pap 1 78-92

Jolliff BL Papike JJ Shearer CK Laul JC (1986)Tourmaline as a recorder of pegmatite evolution Bob Ingersolpegmatite Black Hills South Dakota Am Mineral 71 472-500

Kalt A Schreyer W Ludwig T Prowatke S Bernhardt H ErtlA (2001) Complete solid solution between magnesian schorland lithian excess-boron olenite in a pegmatite from the Koralpe(eastern Alps Austria) Eur J Mineral 13 1191-1205

Kampf AR Gochenour K Clanin J (2003) Tourmaline dis-covery at the Cryo-Genie mine San Diego County CaliforniaRocks and Minerals 78 156-163

Liebscher A Meixner A Romer RL Heinrich W (2007)Experimental calibration of the vapourndashliquid phase relationsand lithium isotope fractionation in the system H2OndashLiCl at400 Geofluids 7 1-7

908 JS Maloney et al

the dike from the upper contact to the core zone includinga pocket is exposed CG-5b is a short euhedral tourmalinefrom the border of the hanging pegmatite CG-1a and CG-5d are radially-grown tourmaline needles from the hangingpegmatite and CG-1c and CG-1e are large euhedral tour-malines from the core zone CG-1i and CG-1n are tourma-line prisms that grew into the pocket CG-1i has a schorlcore and green elbaite rim CG-3e and CG-3i are schorlblades from core zone around the Payday pocket whereasCG-3k is radial schorl from the pegmatite above the pocket

The Little Three dike system intrudes the Green Valleytonalite-gabbro of the western zone of the PRB (Stern et al1986) The system has five dikes including the Little Threemain dike and the Spaulding dike that were sampled forthis study The dikes vary between 1 and 2 m in thick-ness along strike The relative thickness of line-rock andpegmatite zones also varies Most of the analyzed samplescome from the Little Three main dike near the mine en-trance on Topaz Ledge (Fig 3) The bottom boundary ofthe dike with the country rock is a homogeneous aplite withline-rock above it The line rock terminates sharply againstthe lower pegmatite zone which contains small prismaticor bladed tourmaline within graphic intergrowths of quartzand feldspar

Samples LT-3f LT-3g and LT-3i come from the line-rock(Fig 3a) Samples LT-5c and 5d are radial tourmalines thatgrew in the lower pegmatite that cuts through the line rockalong strike from location 3 (Fig 3b) LT-5e is a prismatictourmaline in a smaller dike cutting through the line rockSample LT-4c is a large tourmaline blade from the corezone around a several-meter size pocket that occurs abovethe mine entrance Sample LT-1c is a similar tourmalinefrom the core zone of the Spaulding dike

The Himalaya dike system includes two subparallel dikeshosted by the San Marcos gabbro (Fisher et al 1999Webber et al 1999) The two dikes are separated by 3ndash10 m along the 915 m of their exposed length but convergeat the San Diego mine where the samples were collectedThe Himalaya dikes were emplaced sim 100 my ago asdated by fission track and K-Ar methods (Foord 1976)Samples HM-2c and HM-4c come from pockets in the un-derground workings of the San Diego mine TourmalineHM-2c includes a schorl core and a green elbaite rim whileHM-4c is exclusively schorl

The Cryo-Genie samples best represent a cross-sectionacross the hanging portion of a pegmatite dike while theLittle Three samples best represent the lower portion of adike The Himalaya samples were analyzed primarily to de-termine if there is a regional variability in Li isotopic com-positions

Methods

Bulk Li concentrations

Tourmaline is extremely difficult to dissolve in acids andtherefore we have developed a flux procedure that ac-complishes complete dissolution of ground-up tourmalineBlack cores and green rims of zoned pocket tourmalines

Fig 3 Sketches showing sample locations at the Little Three peg-matite dike Parts a) and b) are along strike on the main dike

were prepared separately When samples had only smallcrystals multiple grains were ground together 100 mgof each tourmaline powder were mixed with 400 mg ofground K2CO3 flux in a zirconia crucible and placed into afurnace at 500 C The temperature was then increased to900 C for 15 min after which the furnace was turned offThe crucibles were left in the furnace overnight to slowlycool to prevent cracking Each fluxed sample was first cen-trifuged in 15 ml of 10 HNO3 for four minutes Theliquid was then decanted into a Teflon beaker and stirredon a hot plate The remaining solid residue was again cen-trifuged in 15 ml of 10 HNO3 Any residue that was stillleft was dissolved in 3 ml of concentrated HNO3 and wasthen added to the already dissolved tourmaline and stirredfor additional 15 minutes on low heat The sample was thenbrought up to 50 ml in a volumetric flask with 10 HNO3In several samples a brown amorphous precipitate formedafter several days After the liquids were decanted the pre-cipitates were dissolved in a HF-HNO3 mixture and thendiluted to 50 ml with distilled water Analysis of these so-lutions revealed no Li

Lithium concentrations (Table 1) were analyzed using thePerkin-Elmer Optima 3300 Inductively Coupled Plasma

Lithium and its isotopes in pegmatite tourmalines 909

Optical Emission Spectrometer (ICP-OES) at the Univer-sity of Missouri Synthetic standards with 005 010 1and 5 ppm Li were prepared using blank K2CO3 flux so-lutions to create the calibration curves for Li Instrumentaldrift was accounted for by analyzing the prepared standardsat regular intervals during the analysis and aliquots of thestandard solutions were used as check standards Lithiumconcentrations were obtained from the strong 610362 nmemission line

Li isotope ratios

Analysis of Li isotope ratios (Table 1) was carried out ondried aliquots of solutions prepared for ICP-OES analy-sis The dried samples were prepared and analyzed at theGeochemistry Laboratory of the University of Maryland-College Park The preparation followed the procedure out-lined in Rudnick et al (2004) and Teng et al (2004) whichis based on the three-column procedure of Moriguti ampNakumura (1998) Because NaLi ratio in solution largerthan sim 5 may cause instability in the analysis the NaLiratio of each sample was determined semi-quantitativelyprior to analysis and excess Na was stripped-off by addi-tional column purification Measurements were done on aNu Plasma Multicollector Inductively Couple Plasma MassSpectrometer (MC-ICP-MS) Each analysis consisted oftwo blocks of twenty individual measurements Each anal-ysis was bracketed by the measurement of a 100 ppb L-SVEC standard The 7Li6Li ratio in L-SVEC for each2 times 20 measurements had an average 2σ of the mean 0003 The external precision better than 1 Two otherLi isotope standards IRMM-016 (Qi et al 1997) and thein-house standard UMD-1 were routinely analyzed dur-ing each analytical session The results for both (IRMM-016 +02 plusmn 04 UMD-1 +543 plusmn 02) agree well withpublished results (IRMM-016 ndash01 plusmn 02 to +02 plusmn 08UMD-1 +547 plusmn 10 Rudnick et al 2004 Teng et al2004 Halama et al 2007) Two USGS rock standardsBHVO-1 (+42 ) and QLO-1 (+66 ) were analyzedfor quality-control purposes The value for BHVO-1 waswithin the uncertainty of previously published results (+43to +58 Bouman et al 2004 Rudnick et al 2004) and thevalue for QLO-1 was within the range of previous analy-ses at the University of Maryland (+56 to +68 Halamaunpublished data)

Results

Li concentrations

Li concentrations in the tourmalines (Table 1) are shown inFig 4a They are plotted in terms of the zones in which thetourmalines grew There appear to be no systematic differ-ences in Li concentrations between corresponding zones ofthe different dikes based on the available data in spite ofthe spatial and probably some temporal separation of thedikes although with denser data sets for all three dikessome systematic differences could appear Nevertheless

Fig 4 (a) Li concentrations and (b) Li isotopic ratios in tourma-lines in three pegmatite dikes Data are grouped according to tex-tural zones from which samples came from Schorl tourmalines areshown by black symbols and elbaite rims on schorl by white sym-bols Two core-rim pairs are shown by connecting lines

the similar ranges in Li concentrations in the pegmatite andthe core zones of both the Cryo-Genie and the Little Threedikes for example point to a common petrogenetic processof the dikes that is underscored by their similar tectoniccontext the similar style of emplacement and the similartextural and mineral zoning across the dikes

There is a progressive increase of about two orders ofmagnitude in Li concentrations across the zones of individ-ual dikes For example in the Cryo-Genie dike concentra-tions range from 22 ppm to 5075 ppm The lowest Li con-centrations 70 ppm are in tourmalines in the line-rockof the Little Three main dike and the top border zone of the

910 JS Maloney et al

Cryo-Genie dike These tourmalines are at most few mil-limeters in length and are prismatic Li concentrations inradial tourmalines in the pegmatite zones of both the Cryo-Genie and Little Three dikes and in large tourmaline bladesin the core zones are on average higher by about an orderof magnitude The concentrations are highly variable how-ever In prismatic schorl crystals within pockets the con-centrations range from 273 to 954 ppm Two green over-growths on schorl cores have the highest Li concentrations1456 and 5075 ppm

Li isotope ratios

In contrast to the progressive increase in Li concentrationsin tourmalines across the zones of the dikes the variationin δ7Li is more complex (Fig 4b) There is no correlationwith Li concentrations but again there appears to be nosystematic difference in δ7Li between the three dikes Intourmalines from the wall-zoneline-rock core zone andpockets δ7Li has approximately the same range between112 and 161 In the pockets elbaite rims have lighterLi than schorl cores although the difference in the Cryo-Genie pair is within overlapping errors In three tourma-lines from the upper and the lower pegmatite zones ofboth Cryo-Genie and Little Three dikes δ7Li is very highgt 19 These three tourmalines came from radial splayswhereas others with lighter Li were prismatic or blade-shaped (Fig 3b) δ7Li in prismatic tourmaline from thebottom pegmatite zone of the Little Three dike is lower at142

The obtained δ7Li values and their range are among thehighest measured in rocks δ7Li in most unaltered crustalrocks typically does not exceed 10 (Tomascak 2004Teng et al 2004) but values up to 19 in quartz anda 10 difference between quartz and albite have beenreported for the Tin Mountain pegmatite in the Black HillsSouth Dakota USA (Teng et al 2006b)

Initial Li concentration in pegmatite melts

Lithium is often invoked as an element that may contributeto growth of large crystals that characterize pegmatites bydepolymerizing the silicate melt structure and thereforelowering its viscosity and increasing chemical diffusionrates In order to evaluate the fluxing effect of Li comparedto H2O in the San Diego County pegmatites a crude es-timate was made of its initial concentration in the meltsThe estimate comes from the Li concentrations in tourma-line the modal abundance of tourmaline and the Li con-centrations in fluid inclusions in pocket quartz assumingthat the pockets were the collection volumes for the sep-arated fluid Tourmaline is the largest reservoir for Li inthe line rock and pegmatite but in the cores and pocketswhere other Li-bearing minerals including lepidolite andamblygonite occur Li concentration in the fluid is best es-timated from primary fluid inclusions in quartz In the Lit-tle Three pegmatite these inclusions have a range of 7000 to12 000 ppm with average of 9200 ppm Li (unpublished data

determined by ion chromatography) The inclusion fluidscontain sim 3 wt NaCl based on microthermometric mea-surements

The initial concentration of Li in the pegmatite melts (Co)can be estimated using

Co = XLRCLR + XPEGCPEG + XPOCKCPOCK

where X is the mass proportion of each zone and C isthe concentration of Li in each zone Li concentrationsin the line-rock and the pegmatite zones were determinedfrom the average of Li concentrations in tourmaline in eachzone and the modal proportions of tourmaline Data for theCryo-Genie and Little Three dikes were combined Com-bining the data for the two dikes is justified because in thepegmatite zones which are volumetrically most abundantLi concentrations cover the same ranges The average Liconcentration in tourmaline in the Cryo-Genie pegmatitezones is 258 ppm and in the Little Three it is 298 ppmThe mass proportions of the three zones were determinedfrom their relative volumes and densities 2700 kgm3 forthe rocks and 500 kgm3 for the fluid The fluid density isappropriate for H2O with 3 wt NaCl at 2 kbar and 400 C(Anderko amp Pitzer 1993) The proportion of tourmaline inthe line-rock and pegmatite zones is lt 8 volume basedon image analysis of cut slabs The estimated bulk Li con-centration in the line-rock zones is only sim 5 ppm and in thepegmatite zones only sim 60 ppm It is readily evident thatmost Li from the original melt ended-up in the fluids

The pockets show that the exsolved fluid was collectedin discrete spaces instead of one continuous space betweenthe hanging and lower portions of the dikes Because thevolume and the distribution of the pockets are highly vari-able (Jahns 1979 Stern et al 1986) the volume of theexsolved fluid is difficult to estimate from the field occur-rences of the pockets A better estimate of the volume ofthe exsolved fluid comes from the maximum H2O solubil-ity in silicate melts which at 2 kbar is sim 6 wt (Holtzet al 1995) This amount of H2O given its molar volumeat 400 C and 2 kbar would occupy sim 28 of the chambervolume Using this proportion the calculated initial Li con-centration in the pegmatite dikes is only sim 630 ppm If theCryo-Genie data is left out of the calculation the differencein the result is only 1 ppm Although the ranges of the mea-sured Li concentrations and errors in the volume estimatesof tourmaline and the textural zones contribute to an erroron this estimate the estimate is dominated by the calculatedfluid volume and the Li concentration in the fluid as deter-mined from fluid inclusions Although 630 ppm Li couldappear to be a rather small concentration for a LCT-typemelt it is 18 times greater than the estimated average con-centration of 35 ppm in the upper continental crust (Tenget al 2004)

Discussion

Role of lithium in the pegmatite crystallization process

Since the classic model of Jahns amp Burnham (1969)for crystallization of pegmatite melts which involves the

Lithium and its isotopes in pegmatite tourmalines 911

separation of a fluid phase from a silicate melt it has beenrecognized that the formation of pegmatites also dependson kinetic controls during rapid cooling of dikes in par-ticular enrichments of components in melt boundary lay-ers ahead of rapidly crystallizing minerals (eg Rockholdet al 1987 London 1992 Webber et al 1999) In addi-tion to being responsible for crystallization of minerals inwhich it is an essential structural constituent Li is ofteninvoked as a possible component that along with H2O BF and other species can potentially flux granite melts Forexample fractional crystallization involving Li was partlyresponsible for mineral zoning seen in the large LCT-typeTin Mountain pegmatite in the Black Hills of South Dakotain which the outer zones are dominated by feldspars andthe inner zones by spodumene and quartz (Walker et al1986) Trapping temperatures of primary fluid inclusionsare lt 400 C and nearly invariable across the pegmatitesuggesting that it crystallized nearly isothermally as an un-dercooled liquid (Sirbescu amp Nabelek 2003a)

An extreme kinetic model is that of Morgan amp London(1999) for crystallization of the Little Three pegmatiteThey suggested that the low temperature and the fast cool-ing rate that must have occurred during solidification of thepegmatite did not allow for crystal nucleation until the melthas reached a glass state at sim 250 C below the equilibriumliquidus They proposed a constitutional zone-refining pro-cess in which a fluxed crystallization front swept a F Liand Mn-rich boundary layer through the solid or semi-soliddike eventually resulting in enrichment of these elementsin the pocket zone

Although we did not obtain electron microprobe data ontourmaline from the dikes the electron microprobe dataof Morgan and London (1999) on tourmaline in the Lit-tle Three dike suggests that the progressive increase in Liconcentrations from the line-rock and the upper wall-zoneto the pockets (Fig 4a) corresponds to changes in othertourmaline components Morgan amp London (1999) foundthat across the line-rock and the pegmatite zones Mg de-creases from sim 07 to near 0 per formula unit while Fe staysnearly constant The increase in Li and the decrease in Mgsuggests an increasing exchange of the elbaite componentfor the dravite component In the pockets tourmaline iszoned from schorl to elbaite but the zoning appears to becontinuous without an evidence for a miscibility gap (Mor-gan amp London 1999) consistent with evidence for com-plete solid solution between schorl and lithian olenite ina pegmatite from the eastern Alps (Kalt et al 2001) El-baite in the pockets has elevated F and Mn concentrations(Morgan amp London 1999) The occurrence of schorl coresin the pockets suggests that schorl grew while Fe-bearingmelt was still present in the dikes but the elbaite rims to-gether with other lithium minerals in pegmatite cores andpockets grew in equilibrium with Li-rich fluid collected inthe pockets The progressive increase in Li across the dikesis more consistent with progressive inward crystallizationof the dikes than a zone-refining process as Li appears tohave been progressively enriched in the residual liquid be-cause of its low solubility in early-crystallizing mineralsincluding schorl The change in tourmaline compositionacross the San Diego County pegmatites is analogous to

the tourmaline composition trend in the Bob Ingersol peg-matite in the Black Hills (Jolliff et al 1986)

The transition from the aplitic line-rock to pegmatiteprobably marks the point of fluid separation in the magmaFluid separation is suggested by the occurrences of elon-gated radial crystals that characterize the pegmatite zonesbecause crystals grow faster and become elongated whenH2O activity increases in the melt (Fenn 1977) HighH2O activity promotes an increase in the diffusion ratesof chemical components even when the melt is undergo-ing rapid cooling Crystallization of tourmaline (and gar-net) is controlled more by gradients in concentrations ofless mobile elements including Fe Mg and Mn than bygradients of rapidly-diffusing elements especially the al-kalis (Rockhold et al 1987 Webber et al 1999) Thelack of crystallization of Li minerals such as spodumeneand amblygonite until the core zones was probably pre-cluded by the initially low Li concentration and removalof Li by the exsolving fluid Although in Cl-absent water-peraluminous melt systems D(Li)fluidmelt is sim 04 (Londonet al 1988) D(Li)fluidmelt increases with the addition ofCl For example when an aqueous fluid at 800 C2 kbarhas sim 7 wt Cl D(Li)fluidmelt is sim 2 (Webster et al 1989)However even this D(Li)fluidmelt seems insufficient to ex-plain the very elevated Li in the pocket fluids by simplebatch partitioning between the fluid and the melt particu-larly because the fluid contained only sim 2 wt Cl Morelikely the large concentration of Li in the fluid is the re-sult of Rayleigh enrichment with crystallization dominatedby feldspars quartz and schorl tourmaline in which Li hassmall solubility compared to the melt and the fluid At 88 crystallization Li would have exceeded 05 wt (11 wtLi2O) in the melt and a correspondingly high concentrationin the accumulating fluid

Concentrations in excess of 1 wt Li2O that may haveexisted at later stages of fractional crystallization of thedikes are approached in some large spodumene-bearingpegmatite intrusions including the Tin Mountain peg-matite the Harding pegmatite in New Mexico and theTanco pegmatite in Manitoba (Norton 1994) These largepegmatites contain spodumene andor other Li-bearingminerals not only in their cores but also other zones (Nor-ton 1994) Li2O concentrations approaching 1 wt in asilicate melt may be required for crystallization of miner-als in which Li is an essential structural constituent In theSan Diego pegmatites such high concentrations apparentlyexisted only in the core zones and the pockets

The influence of the estimated initial 630 ppm Li on theviscosity of the dikes was likely far smaller in compari-son with the influence of the sim 6 wt H2O that wouldbe in the melt at the point of saturation Even in the corezones where Li2O may have reached 1 its effect wouldlikely have been much smaller 1 Li2O is equivalent to19 mole in a haplogranite melt whereas 6 wt H2Ois equivalent to sim 185 mole The addition of 1 wtof excess Li to a haplogranite melt lowers the viscosityby about one order of magnitude (Dingwell et al 1996)but in a peraluminous melt where Li may be complexedwith Al in a Si4+ = Al3+ + Li+ substitution the effect ofLi addition is probably smaller Even if Li exceeded its

912 JS Maloney et al

charge-balancing role in the melt upon removal of Al bycrystallization of the highly peraluminous tourmaline or ifAl was complexed with F in a non-network position in themelt (Giordano et al 2004 Mysen amp Richet 2005) Li isnot expected to have had a drastic effect on the viscosity ofthe dikes In comparison when H2O is added to a silicatemelt it initially dissolves by forming Al-OH and Si-OHcomplexes which depolymerize the tetrahedral networkViscosity drops by sim 45 orders of magnitude with the ad-dition of only 1 wt H2O to a peraluminous leucogranitemelt at 600 C and by an additional 6 orders of magnitudewith further addition of 5 wt H2O (Romano et al 2001Whittington et al 2004) The decreasing effect of H2O ad-dition on the viscosity comes from the increasingly greaterdissolution of H2O as a molecular species (Stolper 1982)Likewise diffusivities of ions in silicate melts dramaticallyincrease with the addition of only a small amount of H2Obut less so with further addition of H2O (Watson 1994)

Lithium isotope fractionation

A potentially large fractionation of Li-isotopes in any mul-tiphase system is due to the 17 mass difference be-tween 7Li and 6Li Presently available data show only verylimited Li isotope fractionation during crystallization ofhigh-temperature igneous systems (Tomascak et al 1999Magna et al 2006 Halama et al 2007) and during vapor-liquid separation in hydrothermal systems (Foustoukoset al 2004 Liebscher et al 2007) However significantfractionation may result from different coordination statesof Li in coexisting phases that include multiple miner-als melts and aqueous fluids in relatively low-temperaturegranitic systems (Wenger amp Armbruster 1991) In general6Li preferentially occupies sites with higher coordinationnumbers and therefore weaker bonds whereas 7Li prefer-entially enters sites with smaller coordination numbers andstronger bonds (Wunder et al 2007)

Fractionation of Li between minerals and fluids hasbeen experimentally determined only for a limited set ofminerals to date Wunder et al (2006 2007) found thatΔ7Listauroliteminusfluid = +13 and is essentially temperature-independent Δ7Lilepidoliteminusfluid is approximately ndash2 withsome temperature dependence in the 350ndash400 C rangeΔ7Lispodumeneminusfluid is also temperature-dependent but morenegative by about 3 For all three minerals the frac-tionation is insensitive to the Cl content of the fluid whichimplies that Li probably forms a tetrahedrally-coordinatedhydrated ion Li(H2O)+4 instead of a LiCl or LiOH com-plex Wunder et al (2007) concluded that equilibrium iso-topic fractionation is firstly controlled by Li coordina-tion with 7Li preferentially incorporated into the phasethat allows for a smaller coordination number and sec-ondly by the Li-O bond length giving the relationshipδ7Listaurolite gt δ

7Lilepidolite gt δ7Lispodumene In staurolite

Li substitutes for the divalent cations Fe2+ Mg and Znin the tetrahedral sites in lepidolite Li is octahedrally-coordinated between tetrahedral layers and in spodumeneit occupies the relatively large M2 octahedral site Thereare no experimental Li isotope fractionation data involv-

ing either tourmaline or melt but in tourmaline Li occupiesthe octahedrally-coordinated Y-site and in a peraluminousmelt Li is probably strongly bonded in its charge balanc-ing role with tetrahedrally-coordinated Al3+ in a LiAlSi3O8complex (Mysen amp Richet 2005) Therefore Li in such amelt should be isotopically relatively heavy

The results of Wunder et al (2007) involving muscoviteare quite different from the results of Lynton et al (2005)who found Δ7Limuscoviteminusfluid to range between +8 and +20in the 400ndash500 C interval Wunder et al (2007) attributedthe discrepancy to the diffusion mechanism that Lyntonet al (2005) used to introduce Li into muscovite IndeedTeng et al (2006a) ascribed very large variations in δ7Li ofcountry rocks in the aureole of the Tin Mountain pegmatiteto differential diffusion of the two Li isotopes which under-scores that in addition to the energy of bonds in lattices ki-netic effects may induce transient Li isotope fractionationwhich may be preserved in rapidly cooled systems

There have been no direct measurements of Li iso-tope fractionation between peraluminous silicate melts andaqueous fluids but it is expected that Li in a melt should beisotopically heavier because of strong bonds in associationwith charge balancing of Al that is in tetrahedral coordina-tion This inference is supported by isotopic compositionsof Li in fluid inclusions and host quartz in the Tin Mountainpegmatite in the Black Hills (Teng et al 2006b) Li in thefluid inclusions has much lower δ7Li values than Li in thequartz supporting the inference that 7Li prefers the strongbonds in quartz where it is possibly charge-balancing Althat is incorporated into the quartz structure and by anal-ogy bonds in high-silica melts over the weaker hydratedbonds in the fluid

The San Diego pegmatites were systems in which min-erals melts and aqueous fluids coexisted at various stagesof crystallization Assuming equilibrium the isotopic com-position of Li in tourmaline in the dikes is reflective of themedium from which the tourmaline crystallized HoweverLi isotopes may be strongly fractionated by kinetic effectsThere is a several permil heterogeneity in δ7Li even in indi-vidual zones of a single pegmatite Most interesting are thevery elevated gt 19 δ7Li values of radial tourmalines inboth the hanging and lower pegmatite zones A reason forthe elevated values may be that the tourmalines grew whenthe melt became saturated in the aqueous fluid at whichpoint Li isotopes were fractionated between the melt andthe aqueous fluid with most Li going into the fluid (Fig 5)If the isotopic composition of Li in tourmaline in each ofthe zones reflects the relative fractionation of Li isotopesbetween the melt and the fluid then the isotopically heavierLi in the pegmatite zone tourmaline is consistent with theexpected stronger Li bonds in the melt compared to bondsof hydrated Li in the fluid The similarity of δ7Li values intourmaline that crystallized in the line rock and the pock-ets is consistent with accumulation of the bulk of meltrsquosinitial Li into the pocket fluid as this would result in littlechange in the isotopic composition of Li from the initialfluid-undersatured melt to the eventually collected fluid inthe pockets Elbaite rims of pocket tourmalines have lowerδ7Li than their corresponding cores although one core-rimpair has overlapping analytical errors (Fig 4b) The lower

Lithium and its isotopes in pegmatite tourmalines 913

Fig 5 Schematic diagram illustrating a possible mechanism of pro-ducing high δ7Li tourmalines in pegmatite zones during fluid-meltseparation Because Li is used in charge-balancing with strongly-bonded Al in the silicate melt the δ7Li in the melt should be moreelevated than δ7Li in the fluid where Li probably occurs mostly asa hydrated ion (Wunder et al 2007) Tourmalines crystallizing inequilibrium with the melt and fluid respectively may reflect the iso-topic fractionation between the melt and the fluid

δ7Li values of the elbaite rims are consistent with crystal-lization of schorl cores while melt was still present andcrystallization of elbaite rims in the presence of the fluidonly

The lack of a systematic increase in δ7Li across the tex-tural zones of the dikes suggests that there was little in-fluence of tourmaline itself on the isotopic composition ofthe residual melt during crystallization Given that bothΔ7Lispodumeneminusfluid and Δ7Lilepidoliteminusfluid are both negativeΔ7Litourmalineminusmelt should be even more negative because Liin tourmaline is in octahedral coordination while in themelt it is associated with tetrahedral Al However becausethe proportion of tourmaline in the line-rock and pegmatitezones is lt 8 and the concentration of Li in the tourma-line is very small crystallization of tourmaline would havehad a negligible effect on the Li isotope ratio in the resid-ual melt Based on mass-balance calculations only sim 2 of the initial 630 ppm Li is contained in schorl in the line-rocks and the pegmatites zones

Kinetic effects

A kinetic cause for the elevated δ7Li values of the radialtourmalines in the pegmatite zones must also be consid-ered however The shape of the tourmalines suggests thatthey grew very rapidly in which case the Li isotope frac-tionation may have been kinetically controlled by differ-ential diffusion of Li isotopes at the crystal-liquid (andorcrystal-fluid) interface The relative diffusion rates of two

Fig 6 Schematic diagram illustrating a possible kinetic mechanismof producing high δ7Li radial tourmalines in pegmatite zones Be-cause the diffusion of 7Li is slower in the melt than the diffusion of6Li and when Li is an incompatible element relative to tourmaline7Li should become preferentially enriched in a boundary layer aheadof a tourmaline crystal that is growing faster than the rate at whichLi diffuses in the melt

isotopes of a given element in a silicate liquid are given by

D1

Dh=

(mh

m1

)β

where m is the mass of isotope ldquolrdquo stands for the light iso-tope and ldquohrdquo stands for the heavy isotope (Richter et al2003) Using basalt-rhyolite melt couples Richter et al(2003) experimentally determined that βLi asymp 0215 Thismeans that 6Li can diffuse substantially faster than 7Lithrough the melt away from a growing crystal so that apreferential enrichment of 7Li can potentially occur in thechemical boundary layer ahead of the crystal (Fig 6) re-sulting in elevated δ7Li The diffusion rate of Li in silicatemelts is orders of magnitude faster than the rates of othermajor and minor cations (Richter et al 2003) Under con-ditions of slow mineral growth homogeneous Li isotoperatios would be expected in the tourmalines Instead theobserved Li isotope heterogeneity suggests crystallizationconditions under which the diffusion of Li in the melt didnot keep-up with the rate of tourmaline growth

Elevated δ7Li values of pegmatites

The overall elevated δ7Li values seen in the San Diegopegmatites are similar to the values in the Tin Mountainpegmatite in the Black Hills δ7Li values in the associ-ated Harney Peak leucogranite and the host schists of theTin Mountain pegmatite have δ7Li values within a fewpermil of 0 which points to some process that leads tostrong Li isotope fractionation during generation of LCT-type pegmatite melts (Teng et al 2006b) The apparentlarge fractionation for these relatively low-temperature ig-neous systems contrasts with the minimal fractionation in

914 JS Maloney et al

high-temperature igneous systems (Tomascak 2004) Tenget al (2006b) suggested that elevated δ7Li values in peg-matite melts could potentially be acquired by crystal-liquidfractionation during crystallization of parental magmasHowever for this process to be effective a greater amountof Li would have to remain in the crystallized assemblage(parental granite) than in the residual liquid (pegmatite)This is inconsistent with for example the relative concen-trations of Li in the Harney Peak leucogranite and its poten-tially residual liquid now represented by the Tin Mountainpegmatite Li concentrations in the Harney Peak leucogran-ite range from 10ndash205 ppm (Teng et al 2006b)

A possible alternative explanation is that the structureof a hydrous pegmatite melt has more similarity to waterthan to minerals and therefore it incorporates 7Li pref-erentially over minerals in a fashion analogous to thelarge 7Li enrichment in crustal fluids compared to crustalrocks (Tomascak 2004) As hydrated pegmatite melts passthrough the crust from their sources they may acquire el-evated δ7Li values through rapid Li isotope exchange withthe surrounding rocks Matthews et al (2003) found thatthe oxygen and especially hydrogen isotopic compositionof pegmatite dikes on Naxos Greece reflects the isotopiccomposition of the host rocks which change in isotopiccomposition along strike of the dikes If hydrogen isotopescan be nearly fully exchanged between pegmatite melts andtheir host rocks then it is likely that Li isotopes can also bereadily exchanged given the fast diffusion rates of Li in sil-icate liquids (Richter et al 2003)

Acknowledgements We are grateful for access providedby Dana and Ken Gochenour to the Cryo-Genie propertyBill Calhoun to the San Diego mine property and LouisSpaulding Jr to the Little Three property Jeffrey Patter-son and Matt Taylor provided guidance around the peg-matite districts and Jim Student helped with sample col-lection Carol Nabelek oversaw the ICP-OES analysis BillMcDonough graciously gave access to JM to conduct anal-ysis in the isotope laboratory at the University of MarylandThe paper benefited from the constructive reviews of AxelLiebscher Jeffrey Ryan and Ed Grew and additional com-ments from Roberta Rudnick A Feodor-Lynen fellowshipto Halama by the Alexander von Humboldt Foundation isgratefully acknowledged The study was supported by Uni-versity of Missouri Research Board Grant D3508 and NSFGrant 408564 to Nabelek Additional funding came fromNSF Grant EAR 0606989 to Rudnick and McDonough

References

Anderko A amp Pitzer KS (1993) Equation-of-state representa-tion of phase equilibria and volumetric properties of the systemNaCl-H2O above 573 K Geochim Cosmochim Acta 57 1657-1680

Bouman C Elliott T Vroon PZ (2004) Lithium inputs to sub-duction zones Chem Geol 212 59-79

Cernyacute P (1991) Rare-element granitic pegmatites Part I Anatomyand internal evolution of pegmatite deposits GeoscienceCanada 18 49-67

Cernyacute P amp Ercit TS (2005) The classification of granitic peg-matites revisited Can Mineral 43 2005-2026

Dingwell DB Hess K-U Knoche R (1996) Granite andgranitic pegmatite melts volumes and viscosities Trans R SocEdinburgh Earth Sci 87 65-72

Fenn PM (1977) The nucleation and growth of alkali feldsparsfrom hydrous melts Can Mineral 15 135-161

Fisher J (2002) Gem and rare-element pegmatites of southernCalifornia Mineral Record 33 363-407

Fisher J Foord EE Bricker GA (1999) The geology miner-alogy and history of the Himalaya Mine Mesa Grande SanDiego County California California Geol 3-17

Foord EE (1976) Mineralogy and petrogenesis of layeredpegmatite-aplite dikes in the Mesa Grande District San DiegoCounty California PhD Dissertation Stanford University

Foord EE Starkey HC Taggard JE Jr (1986) Mineralogyand paragenesis of ldquopocketrdquo clays and associated minerals incomplex granitic pegmatites San Diego County California AmMineral 71 428-439

Foustoukos DI James RH Berndt ME Seyfried WE Jr(2004) Lithium isotopic systematics of hydrothermal vent flu-ids at the Main Endeavour Field Northern Juan de Fuca RidgeChem Geol 212 17-26

Giordano D Romano C Dingwell DB Poe B Behrens H(2004) The combined effects of water and fluorine on theviscosity of silicic magmas Geochim Cosmochim Acta 685159-5168

Halama R McDonough WF Rudnick RL Keller J KlaudiusJ (2007) The Li isotopic composition of Oldoinyo LengaiNature of the mantle sources and lack of isotopic fractionationduring carbonatite petrogenesis Earth Planet Sci Lett 25477-89

Holtz F Behrens H Dingwell DB Johannes W (1995) Watersolubility in haplogranite melts Compositional pressure andtemperature dependence Am Mineral 80 94-108

Jahns R H (1979) Gem-bearing pegmatites in San Diego CountyCalifornia The Stewart mine Pala district and the Himalayamine Mesa Grande district in ldquoMesozoic crystalline rocksPeninsular Ranges batholith and pegmatites Point Sol ophio-literdquo P L Abbott amp V R Todd eds San Diego State UniversitySan Diego California 3-38

Jahns RH amp Burnham CW (1969) Experimental studies of peg-matite genesis I A model for the derivation and crystallizationof granitic pegmatites Econ Geol 64 843-864

Jahns RH amp Tuttle OF (1963) Layered pegmatite-aplite intru-sives Mineral Soc Am Sp Pap 1 78-92

Jolliff BL Papike JJ Shearer CK Laul JC (1986)Tourmaline as a recorder of pegmatite evolution Bob Ingersolpegmatite Black Hills South Dakota Am Mineral 71 472-500

Kalt A Schreyer W Ludwig T Prowatke S Bernhardt H ErtlA (2001) Complete solid solution between magnesian schorland lithian excess-boron olenite in a pegmatite from the Koralpe(eastern Alps Austria) Eur J Mineral 13 1191-1205

Kampf AR Gochenour K Clanin J (2003) Tourmaline dis-covery at the Cryo-Genie mine San Diego County CaliforniaRocks and Minerals 78 156-163

Liebscher A Meixner A Romer RL Heinrich W (2007)Experimental calibration of the vapourndashliquid phase relationsand lithium isotope fractionation in the system H2OndashLiCl at400 Geofluids 7 1-7

Lithium and its isotopes in pegmatite tourmalines 909

Optical Emission Spectrometer (ICP-OES) at the Univer-sity of Missouri Synthetic standards with 005 010 1and 5 ppm Li were prepared using blank K2CO3 flux so-lutions to create the calibration curves for Li Instrumentaldrift was accounted for by analyzing the prepared standardsat regular intervals during the analysis and aliquots of thestandard solutions were used as check standards Lithiumconcentrations were obtained from the strong 610362 nmemission line

Li isotope ratios

Analysis of Li isotope ratios (Table 1) was carried out ondried aliquots of solutions prepared for ICP-OES analy-sis The dried samples were prepared and analyzed at theGeochemistry Laboratory of the University of Maryland-College Park The preparation followed the procedure out-lined in Rudnick et al (2004) and Teng et al (2004) whichis based on the three-column procedure of Moriguti ampNakumura (1998) Because NaLi ratio in solution largerthan sim 5 may cause instability in the analysis the NaLiratio of each sample was determined semi-quantitativelyprior to analysis and excess Na was stripped-off by addi-tional column purification Measurements were done on aNu Plasma Multicollector Inductively Couple Plasma MassSpectrometer (MC-ICP-MS) Each analysis consisted oftwo blocks of twenty individual measurements Each anal-ysis was bracketed by the measurement of a 100 ppb L-SVEC standard The 7Li6Li ratio in L-SVEC for each2 times 20 measurements had an average 2σ of the mean 0003 The external precision better than 1 Two otherLi isotope standards IRMM-016 (Qi et al 1997) and thein-house standard UMD-1 were routinely analyzed dur-ing each analytical session The results for both (IRMM-016 +02 plusmn 04 UMD-1 +543 plusmn 02) agree well withpublished results (IRMM-016 ndash01 plusmn 02 to +02 plusmn 08UMD-1 +547 plusmn 10 Rudnick et al 2004 Teng et al2004 Halama et al 2007) Two USGS rock standardsBHVO-1 (+42 ) and QLO-1 (+66 ) were analyzedfor quality-control purposes The value for BHVO-1 waswithin the uncertainty of previously published results (+43to +58 Bouman et al 2004 Rudnick et al 2004) and thevalue for QLO-1 was within the range of previous analy-ses at the University of Maryland (+56 to +68 Halamaunpublished data)

Results

Li concentrations

Li concentrations in the tourmalines (Table 1) are shown inFig 4a They are plotted in terms of the zones in which thetourmalines grew There appear to be no systematic differ-ences in Li concentrations between corresponding zones ofthe different dikes based on the available data in spite ofthe spatial and probably some temporal separation of thedikes although with denser data sets for all three dikessome systematic differences could appear Nevertheless

Fig 4 (a) Li concentrations and (b) Li isotopic ratios in tourma-lines in three pegmatite dikes Data are grouped according to tex-tural zones from which samples came from Schorl tourmalines areshown by black symbols and elbaite rims on schorl by white sym-bols Two core-rim pairs are shown by connecting lines

the similar ranges in Li concentrations in the pegmatite andthe core zones of both the Cryo-Genie and the Little Threedikes for example point to a common petrogenetic processof the dikes that is underscored by their similar tectoniccontext the similar style of emplacement and the similartextural and mineral zoning across the dikes

There is a progressive increase of about two orders ofmagnitude in Li concentrations across the zones of individ-ual dikes For example in the Cryo-Genie dike concentra-tions range from 22 ppm to 5075 ppm The lowest Li con-centrations 70 ppm are in tourmalines in the line-rockof the Little Three main dike and the top border zone of the

910 JS Maloney et al

Cryo-Genie dike These tourmalines are at most few mil-limeters in length and are prismatic Li concentrations inradial tourmalines in the pegmatite zones of both the Cryo-Genie and Little Three dikes and in large tourmaline bladesin the core zones are on average higher by about an orderof magnitude The concentrations are highly variable how-ever In prismatic schorl crystals within pockets the con-centrations range from 273 to 954 ppm Two green over-growths on schorl cores have the highest Li concentrations1456 and 5075 ppm

Li isotope ratios

In contrast to the progressive increase in Li concentrationsin tourmalines across the zones of the dikes the variationin δ7Li is more complex (Fig 4b) There is no correlationwith Li concentrations but again there appears to be nosystematic difference in δ7Li between the three dikes Intourmalines from the wall-zoneline-rock core zone andpockets δ7Li has approximately the same range between112 and 161 In the pockets elbaite rims have lighterLi than schorl cores although the difference in the Cryo-Genie pair is within overlapping errors In three tourma-lines from the upper and the lower pegmatite zones ofboth Cryo-Genie and Little Three dikes δ7Li is very highgt 19 These three tourmalines came from radial splayswhereas others with lighter Li were prismatic or blade-shaped (Fig 3b) δ7Li in prismatic tourmaline from thebottom pegmatite zone of the Little Three dike is lower at142

The obtained δ7Li values and their range are among thehighest measured in rocks δ7Li in most unaltered crustalrocks typically does not exceed 10 (Tomascak 2004Teng et al 2004) but values up to 19 in quartz anda 10 difference between quartz and albite have beenreported for the Tin Mountain pegmatite in the Black HillsSouth Dakota USA (Teng et al 2006b)

Initial Li concentration in pegmatite melts

Lithium is often invoked as an element that may contributeto growth of large crystals that characterize pegmatites bydepolymerizing the silicate melt structure and thereforelowering its viscosity and increasing chemical diffusionrates In order to evaluate the fluxing effect of Li comparedto H2O in the San Diego County pegmatites a crude es-timate was made of its initial concentration in the meltsThe estimate comes from the Li concentrations in tourma-line the modal abundance of tourmaline and the Li con-centrations in fluid inclusions in pocket quartz assumingthat the pockets were the collection volumes for the sep-arated fluid Tourmaline is the largest reservoir for Li inthe line rock and pegmatite but in the cores and pocketswhere other Li-bearing minerals including lepidolite andamblygonite occur Li concentration in the fluid is best es-timated from primary fluid inclusions in quartz In the Lit-tle Three pegmatite these inclusions have a range of 7000 to12 000 ppm with average of 9200 ppm Li (unpublished data

determined by ion chromatography) The inclusion fluidscontain sim 3 wt NaCl based on microthermometric mea-surements

The initial concentration of Li in the pegmatite melts (Co)can be estimated using

Co = XLRCLR + XPEGCPEG + XPOCKCPOCK

where X is the mass proportion of each zone and C isthe concentration of Li in each zone Li concentrationsin the line-rock and the pegmatite zones were determinedfrom the average of Li concentrations in tourmaline in eachzone and the modal proportions of tourmaline Data for theCryo-Genie and Little Three dikes were combined Com-bining the data for the two dikes is justified because in thepegmatite zones which are volumetrically most abundantLi concentrations cover the same ranges The average Liconcentration in tourmaline in the Cryo-Genie pegmatitezones is 258 ppm and in the Little Three it is 298 ppmThe mass proportions of the three zones were determinedfrom their relative volumes and densities 2700 kgm3 forthe rocks and 500 kgm3 for the fluid The fluid density isappropriate for H2O with 3 wt NaCl at 2 kbar and 400 C(Anderko amp Pitzer 1993) The proportion of tourmaline inthe line-rock and pegmatite zones is lt 8 volume basedon image analysis of cut slabs The estimated bulk Li con-centration in the line-rock zones is only sim 5 ppm and in thepegmatite zones only sim 60 ppm It is readily evident thatmost Li from the original melt ended-up in the fluids

The pockets show that the exsolved fluid was collectedin discrete spaces instead of one continuous space betweenthe hanging and lower portions of the dikes Because thevolume and the distribution of the pockets are highly vari-able (Jahns 1979 Stern et al 1986) the volume of theexsolved fluid is difficult to estimate from the field occur-rences of the pockets A better estimate of the volume ofthe exsolved fluid comes from the maximum H2O solubil-ity in silicate melts which at 2 kbar is sim 6 wt (Holtzet al 1995) This amount of H2O given its molar volumeat 400 C and 2 kbar would occupy sim 28 of the chambervolume Using this proportion the calculated initial Li con-centration in the pegmatite dikes is only sim 630 ppm If theCryo-Genie data is left out of the calculation the differencein the result is only 1 ppm Although the ranges of the mea-sured Li concentrations and errors in the volume estimatesof tourmaline and the textural zones contribute to an erroron this estimate the estimate is dominated by the calculatedfluid volume and the Li concentration in the fluid as deter-mined from fluid inclusions Although 630 ppm Li couldappear to be a rather small concentration for a LCT-typemelt it is 18 times greater than the estimated average con-centration of 35 ppm in the upper continental crust (Tenget al 2004)

Discussion

Role of lithium in the pegmatite crystallization process

Since the classic model of Jahns amp Burnham (1969)for crystallization of pegmatite melts which involves the

Lithium and its isotopes in pegmatite tourmalines 911

separation of a fluid phase from a silicate melt it has beenrecognized that the formation of pegmatites also dependson kinetic controls during rapid cooling of dikes in par-ticular enrichments of components in melt boundary lay-ers ahead of rapidly crystallizing minerals (eg Rockholdet al 1987 London 1992 Webber et al 1999) In addi-tion to being responsible for crystallization of minerals inwhich it is an essential structural constituent Li is ofteninvoked as a possible component that along with H2O BF and other species can potentially flux granite melts Forexample fractional crystallization involving Li was partlyresponsible for mineral zoning seen in the large LCT-typeTin Mountain pegmatite in the Black Hills of South Dakotain which the outer zones are dominated by feldspars andthe inner zones by spodumene and quartz (Walker et al1986) Trapping temperatures of primary fluid inclusionsare lt 400 C and nearly invariable across the pegmatitesuggesting that it crystallized nearly isothermally as an un-dercooled liquid (Sirbescu amp Nabelek 2003a)

An extreme kinetic model is that of Morgan amp London(1999) for crystallization of the Little Three pegmatiteThey suggested that the low temperature and the fast cool-ing rate that must have occurred during solidification of thepegmatite did not allow for crystal nucleation until the melthas reached a glass state at sim 250 C below the equilibriumliquidus They proposed a constitutional zone-refining pro-cess in which a fluxed crystallization front swept a F Liand Mn-rich boundary layer through the solid or semi-soliddike eventually resulting in enrichment of these elementsin the pocket zone

Although we did not obtain electron microprobe data ontourmaline from the dikes the electron microprobe dataof Morgan and London (1999) on tourmaline in the Lit-tle Three dike suggests that the progressive increase in Liconcentrations from the line-rock and the upper wall-zoneto the pockets (Fig 4a) corresponds to changes in othertourmaline components Morgan amp London (1999) foundthat across the line-rock and the pegmatite zones Mg de-creases from sim 07 to near 0 per formula unit while Fe staysnearly constant The increase in Li and the decrease in Mgsuggests an increasing exchange of the elbaite componentfor the dravite component In the pockets tourmaline iszoned from schorl to elbaite but the zoning appears to becontinuous without an evidence for a miscibility gap (Mor-gan amp London 1999) consistent with evidence for com-plete solid solution between schorl and lithian olenite ina pegmatite from the eastern Alps (Kalt et al 2001) El-baite in the pockets has elevated F and Mn concentrations(Morgan amp London 1999) The occurrence of schorl coresin the pockets suggests that schorl grew while Fe-bearingmelt was still present in the dikes but the elbaite rims to-gether with other lithium minerals in pegmatite cores andpockets grew in equilibrium with Li-rich fluid collected inthe pockets The progressive increase in Li across the dikesis more consistent with progressive inward crystallizationof the dikes than a zone-refining process as Li appears tohave been progressively enriched in the residual liquid be-cause of its low solubility in early-crystallizing mineralsincluding schorl The change in tourmaline compositionacross the San Diego County pegmatites is analogous to

the tourmaline composition trend in the Bob Ingersol peg-matite in the Black Hills (Jolliff et al 1986)

The transition from the aplitic line-rock to pegmatiteprobably marks the point of fluid separation in the magmaFluid separation is suggested by the occurrences of elon-gated radial crystals that characterize the pegmatite zonesbecause crystals grow faster and become elongated whenH2O activity increases in the melt (Fenn 1977) HighH2O activity promotes an increase in the diffusion ratesof chemical components even when the melt is undergo-ing rapid cooling Crystallization of tourmaline (and gar-net) is controlled more by gradients in concentrations ofless mobile elements including Fe Mg and Mn than bygradients of rapidly-diffusing elements especially the al-kalis (Rockhold et al 1987 Webber et al 1999) Thelack of crystallization of Li minerals such as spodumeneand amblygonite until the core zones was probably pre-cluded by the initially low Li concentration and removalof Li by the exsolving fluid Although in Cl-absent water-peraluminous melt systems D(Li)fluidmelt is sim 04 (Londonet al 1988) D(Li)fluidmelt increases with the addition ofCl For example when an aqueous fluid at 800 C2 kbarhas sim 7 wt Cl D(Li)fluidmelt is sim 2 (Webster et al 1989)However even this D(Li)fluidmelt seems insufficient to ex-plain the very elevated Li in the pocket fluids by simplebatch partitioning between the fluid and the melt particu-larly because the fluid contained only sim 2 wt Cl Morelikely the large concentration of Li in the fluid is the re-sult of Rayleigh enrichment with crystallization dominatedby feldspars quartz and schorl tourmaline in which Li hassmall solubility compared to the melt and the fluid At 88 crystallization Li would have exceeded 05 wt (11 wtLi2O) in the melt and a correspondingly high concentrationin the accumulating fluid

Concentrations in excess of 1 wt Li2O that may haveexisted at later stages of fractional crystallization of thedikes are approached in some large spodumene-bearingpegmatite intrusions including the Tin Mountain peg-matite the Harding pegmatite in New Mexico and theTanco pegmatite in Manitoba (Norton 1994) These largepegmatites contain spodumene andor other Li-bearingminerals not only in their cores but also other zones (Nor-ton 1994) Li2O concentrations approaching 1 wt in asilicate melt may be required for crystallization of miner-als in which Li is an essential structural constituent In theSan Diego pegmatites such high concentrations apparentlyexisted only in the core zones and the pockets

The influence of the estimated initial 630 ppm Li on theviscosity of the dikes was likely far smaller in compari-son with the influence of the sim 6 wt H2O that wouldbe in the melt at the point of saturation Even in the corezones where Li2O may have reached 1 its effect wouldlikely have been much smaller 1 Li2O is equivalent to19 mole in a haplogranite melt whereas 6 wt H2Ois equivalent to sim 185 mole The addition of 1 wtof excess Li to a haplogranite melt lowers the viscosityby about one order of magnitude (Dingwell et al 1996)but in a peraluminous melt where Li may be complexedwith Al in a Si4+ = Al3+ + Li+ substitution the effect ofLi addition is probably smaller Even if Li exceeded its

912 JS Maloney et al

charge-balancing role in the melt upon removal of Al bycrystallization of the highly peraluminous tourmaline or ifAl was complexed with F in a non-network position in themelt (Giordano et al 2004 Mysen amp Richet 2005) Li isnot expected to have had a drastic effect on the viscosity ofthe dikes In comparison when H2O is added to a silicatemelt it initially dissolves by forming Al-OH and Si-OHcomplexes which depolymerize the tetrahedral networkViscosity drops by sim 45 orders of magnitude with the ad-dition of only 1 wt H2O to a peraluminous leucogranitemelt at 600 C and by an additional 6 orders of magnitudewith further addition of 5 wt H2O (Romano et al 2001Whittington et al 2004) The decreasing effect of H2O ad-dition on the viscosity comes from the increasingly greaterdissolution of H2O as a molecular species (Stolper 1982)Likewise diffusivities of ions in silicate melts dramaticallyincrease with the addition of only a small amount of H2Obut less so with further addition of H2O (Watson 1994)

Lithium isotope fractionation

A potentially large fractionation of Li-isotopes in any mul-tiphase system is due to the 17 mass difference be-tween 7Li and 6Li Presently available data show only verylimited Li isotope fractionation during crystallization ofhigh-temperature igneous systems (Tomascak et al 1999Magna et al 2006 Halama et al 2007) and during vapor-liquid separation in hydrothermal systems (Foustoukoset al 2004 Liebscher et al 2007) However significantfractionation may result from different coordination statesof Li in coexisting phases that include multiple miner-als melts and aqueous fluids in relatively low-temperaturegranitic systems (Wenger amp Armbruster 1991) In general6Li preferentially occupies sites with higher coordinationnumbers and therefore weaker bonds whereas 7Li prefer-entially enters sites with smaller coordination numbers andstronger bonds (Wunder et al 2007)

Fractionation of Li between minerals and fluids hasbeen experimentally determined only for a limited set ofminerals to date Wunder et al (2006 2007) found thatΔ7Listauroliteminusfluid = +13 and is essentially temperature-independent Δ7Lilepidoliteminusfluid is approximately ndash2 withsome temperature dependence in the 350ndash400 C rangeΔ7Lispodumeneminusfluid is also temperature-dependent but morenegative by about 3 For all three minerals the frac-tionation is insensitive to the Cl content of the fluid whichimplies that Li probably forms a tetrahedrally-coordinatedhydrated ion Li(H2O)+4 instead of a LiCl or LiOH com-plex Wunder et al (2007) concluded that equilibrium iso-topic fractionation is firstly controlled by Li coordina-tion with 7Li preferentially incorporated into the phasethat allows for a smaller coordination number and sec-ondly by the Li-O bond length giving the relationshipδ7Listaurolite gt δ

7Lilepidolite gt δ7Lispodumene In staurolite

Li substitutes for the divalent cations Fe2+ Mg and Znin the tetrahedral sites in lepidolite Li is octahedrally-coordinated between tetrahedral layers and in spodumeneit occupies the relatively large M2 octahedral site Thereare no experimental Li isotope fractionation data involv-

ing either tourmaline or melt but in tourmaline Li occupiesthe octahedrally-coordinated Y-site and in a peraluminousmelt Li is probably strongly bonded in its charge balanc-ing role with tetrahedrally-coordinated Al3+ in a LiAlSi3O8complex (Mysen amp Richet 2005) Therefore Li in such amelt should be isotopically relatively heavy

The results of Wunder et al (2007) involving muscoviteare quite different from the results of Lynton et al (2005)who found Δ7Limuscoviteminusfluid to range between +8 and +20in the 400ndash500 C interval Wunder et al (2007) attributedthe discrepancy to the diffusion mechanism that Lyntonet al (2005) used to introduce Li into muscovite IndeedTeng et al (2006a) ascribed very large variations in δ7Li ofcountry rocks in the aureole of the Tin Mountain pegmatiteto differential diffusion of the two Li isotopes which under-scores that in addition to the energy of bonds in lattices ki-netic effects may induce transient Li isotope fractionationwhich may be preserved in rapidly cooled systems

There have been no direct measurements of Li iso-tope fractionation between peraluminous silicate melts andaqueous fluids but it is expected that Li in a melt should beisotopically heavier because of strong bonds in associationwith charge balancing of Al that is in tetrahedral coordina-tion This inference is supported by isotopic compositionsof Li in fluid inclusions and host quartz in the Tin Mountainpegmatite in the Black Hills (Teng et al 2006b) Li in thefluid inclusions has much lower δ7Li values than Li in thequartz supporting the inference that 7Li prefers the strongbonds in quartz where it is possibly charge-balancing Althat is incorporated into the quartz structure and by anal-ogy bonds in high-silica melts over the weaker hydratedbonds in the fluid

The San Diego pegmatites were systems in which min-erals melts and aqueous fluids coexisted at various stagesof crystallization Assuming equilibrium the isotopic com-position of Li in tourmaline in the dikes is reflective of themedium from which the tourmaline crystallized HoweverLi isotopes may be strongly fractionated by kinetic effectsThere is a several permil heterogeneity in δ7Li even in indi-vidual zones of a single pegmatite Most interesting are thevery elevated gt 19 δ7Li values of radial tourmalines inboth the hanging and lower pegmatite zones A reason forthe elevated values may be that the tourmalines grew whenthe melt became saturated in the aqueous fluid at whichpoint Li isotopes were fractionated between the melt andthe aqueous fluid with most Li going into the fluid (Fig 5)If the isotopic composition of Li in tourmaline in each ofthe zones reflects the relative fractionation of Li isotopesbetween the melt and the fluid then the isotopically heavierLi in the pegmatite zone tourmaline is consistent with theexpected stronger Li bonds in the melt compared to bondsof hydrated Li in the fluid The similarity of δ7Li values intourmaline that crystallized in the line rock and the pock-ets is consistent with accumulation of the bulk of meltrsquosinitial Li into the pocket fluid as this would result in littlechange in the isotopic composition of Li from the initialfluid-undersatured melt to the eventually collected fluid inthe pockets Elbaite rims of pocket tourmalines have lowerδ7Li than their corresponding cores although one core-rimpair has overlapping analytical errors (Fig 4b) The lower

Lithium and its isotopes in pegmatite tourmalines 913

Fig 5 Schematic diagram illustrating a possible mechanism of pro-ducing high δ7Li tourmalines in pegmatite zones during fluid-meltseparation Because Li is used in charge-balancing with strongly-bonded Al in the silicate melt the δ7Li in the melt should be moreelevated than δ7Li in the fluid where Li probably occurs mostly asa hydrated ion (Wunder et al 2007) Tourmalines crystallizing inequilibrium with the melt and fluid respectively may reflect the iso-topic fractionation between the melt and the fluid

δ7Li values of the elbaite rims are consistent with crystal-lization of schorl cores while melt was still present andcrystallization of elbaite rims in the presence of the fluidonly

The lack of a systematic increase in δ7Li across the tex-tural zones of the dikes suggests that there was little in-fluence of tourmaline itself on the isotopic composition ofthe residual melt during crystallization Given that bothΔ7Lispodumeneminusfluid and Δ7Lilepidoliteminusfluid are both negativeΔ7Litourmalineminusmelt should be even more negative because Liin tourmaline is in octahedral coordination while in themelt it is associated with tetrahedral Al However becausethe proportion of tourmaline in the line-rock and pegmatitezones is lt 8 and the concentration of Li in the tourma-line is very small crystallization of tourmaline would havehad a negligible effect on the Li isotope ratio in the resid-ual melt Based on mass-balance calculations only sim 2 of the initial 630 ppm Li is contained in schorl in the line-rocks and the pegmatites zones

Kinetic effects

A kinetic cause for the elevated δ7Li values of the radialtourmalines in the pegmatite zones must also be consid-ered however The shape of the tourmalines suggests thatthey grew very rapidly in which case the Li isotope frac-tionation may have been kinetically controlled by differ-ential diffusion of Li isotopes at the crystal-liquid (andorcrystal-fluid) interface The relative diffusion rates of two

Fig 6 Schematic diagram illustrating a possible kinetic mechanismof producing high δ7Li radial tourmalines in pegmatite zones Be-cause the diffusion of 7Li is slower in the melt than the diffusion of6Li and when Li is an incompatible element relative to tourmaline7Li should become preferentially enriched in a boundary layer aheadof a tourmaline crystal that is growing faster than the rate at whichLi diffuses in the melt

isotopes of a given element in a silicate liquid are given by

D1

Dh=

(mh

m1

)β

where m is the mass of isotope ldquolrdquo stands for the light iso-tope and ldquohrdquo stands for the heavy isotope (Richter et al2003) Using basalt-rhyolite melt couples Richter et al(2003) experimentally determined that βLi asymp 0215 Thismeans that 6Li can diffuse substantially faster than 7Lithrough the melt away from a growing crystal so that apreferential enrichment of 7Li can potentially occur in thechemical boundary layer ahead of the crystal (Fig 6) re-sulting in elevated δ7Li The diffusion rate of Li in silicatemelts is orders of magnitude faster than the rates of othermajor and minor cations (Richter et al 2003) Under con-ditions of slow mineral growth homogeneous Li isotoperatios would be expected in the tourmalines Instead theobserved Li isotope heterogeneity suggests crystallizationconditions under which the diffusion of Li in the melt didnot keep-up with the rate of tourmaline growth

Elevated δ7Li values of pegmatites

The overall elevated δ7Li values seen in the San Diegopegmatites are similar to the values in the Tin Mountainpegmatite in the Black Hills δ7Li values in the associ-ated Harney Peak leucogranite and the host schists of theTin Mountain pegmatite have δ7Li values within a fewpermil of 0 which points to some process that leads tostrong Li isotope fractionation during generation of LCT-type pegmatite melts (Teng et al 2006b) The apparentlarge fractionation for these relatively low-temperature ig-neous systems contrasts with the minimal fractionation in

914 JS Maloney et al

high-temperature igneous systems (Tomascak 2004) Tenget al (2006b) suggested that elevated δ7Li values in peg-matite melts could potentially be acquired by crystal-liquidfractionation during crystallization of parental magmasHowever for this process to be effective a greater amountof Li would have to remain in the crystallized assemblage(parental granite) than in the residual liquid (pegmatite)This is inconsistent with for example the relative concen-trations of Li in the Harney Peak leucogranite and its poten-tially residual liquid now represented by the Tin Mountainpegmatite Li concentrations in the Harney Peak leucogran-ite range from 10ndash205 ppm (Teng et al 2006b)

A possible alternative explanation is that the structureof a hydrous pegmatite melt has more similarity to waterthan to minerals and therefore it incorporates 7Li pref-erentially over minerals in a fashion analogous to thelarge 7Li enrichment in crustal fluids compared to crustalrocks (Tomascak 2004) As hydrated pegmatite melts passthrough the crust from their sources they may acquire el-evated δ7Li values through rapid Li isotope exchange withthe surrounding rocks Matthews et al (2003) found thatthe oxygen and especially hydrogen isotopic compositionof pegmatite dikes on Naxos Greece reflects the isotopiccomposition of the host rocks which change in isotopiccomposition along strike of the dikes If hydrogen isotopescan be nearly fully exchanged between pegmatite melts andtheir host rocks then it is likely that Li isotopes can also bereadily exchanged given the fast diffusion rates of Li in sil-icate liquids (Richter et al 2003)

Acknowledgements We are grateful for access providedby Dana and Ken Gochenour to the Cryo-Genie propertyBill Calhoun to the San Diego mine property and LouisSpaulding Jr to the Little Three property Jeffrey Patter-son and Matt Taylor provided guidance around the peg-matite districts and Jim Student helped with sample col-lection Carol Nabelek oversaw the ICP-OES analysis BillMcDonough graciously gave access to JM to conduct anal-ysis in the isotope laboratory at the University of MarylandThe paper benefited from the constructive reviews of AxelLiebscher Jeffrey Ryan and Ed Grew and additional com-ments from Roberta Rudnick A Feodor-Lynen fellowshipto Halama by the Alexander von Humboldt Foundation isgratefully acknowledged The study was supported by Uni-versity of Missouri Research Board Grant D3508 and NSFGrant 408564 to Nabelek Additional funding came fromNSF Grant EAR 0606989 to Rudnick and McDonough

References

Anderko A amp Pitzer KS (1993) Equation-of-state representa-tion of phase equilibria and volumetric properties of the systemNaCl-H2O above 573 K Geochim Cosmochim Acta 57 1657-1680

Bouman C Elliott T Vroon PZ (2004) Lithium inputs to sub-duction zones Chem Geol 212 59-79

Cernyacute P (1991) Rare-element granitic pegmatites Part I Anatomyand internal evolution of pegmatite deposits GeoscienceCanada 18 49-67

Cernyacute P amp Ercit TS (2005) The classification of granitic peg-matites revisited Can Mineral 43 2005-2026

Dingwell DB Hess K-U Knoche R (1996) Granite andgranitic pegmatite melts volumes and viscosities Trans R SocEdinburgh Earth Sci 87 65-72

Fenn PM (1977) The nucleation and growth of alkali feldsparsfrom hydrous melts Can Mineral 15 135-161

Fisher J (2002) Gem and rare-element pegmatites of southernCalifornia Mineral Record 33 363-407

Fisher J Foord EE Bricker GA (1999) The geology miner-alogy and history of the Himalaya Mine Mesa Grande SanDiego County California California Geol 3-17

Foord EE (1976) Mineralogy and petrogenesis of layeredpegmatite-aplite dikes in the Mesa Grande District San DiegoCounty California PhD Dissertation Stanford University

Foord EE Starkey HC Taggard JE Jr (1986) Mineralogyand paragenesis of ldquopocketrdquo clays and associated minerals incomplex granitic pegmatites San Diego County California AmMineral 71 428-439

Foustoukos DI James RH Berndt ME Seyfried WE Jr(2004) Lithium isotopic systematics of hydrothermal vent flu-ids at the Main Endeavour Field Northern Juan de Fuca RidgeChem Geol 212 17-26

Giordano D Romano C Dingwell DB Poe B Behrens H(2004) The combined effects of water and fluorine on theviscosity of silicic magmas Geochim Cosmochim Acta 685159-5168

Halama R McDonough WF Rudnick RL Keller J KlaudiusJ (2007) The Li isotopic composition of Oldoinyo LengaiNature of the mantle sources and lack of isotopic fractionationduring carbonatite petrogenesis Earth Planet Sci Lett 25477-89

Holtz F Behrens H Dingwell DB Johannes W (1995) Watersolubility in haplogranite melts Compositional pressure andtemperature dependence Am Mineral 80 94-108

Jahns R H (1979) Gem-bearing pegmatites in San Diego CountyCalifornia The Stewart mine Pala district and the Himalayamine Mesa Grande district in ldquoMesozoic crystalline rocksPeninsular Ranges batholith and pegmatites Point Sol ophio-literdquo P L Abbott amp V R Todd eds San Diego State UniversitySan Diego California 3-38

Jahns RH amp Burnham CW (1969) Experimental studies of peg-matite genesis I A model for the derivation and crystallizationof granitic pegmatites Econ Geol 64 843-864

Jahns RH amp Tuttle OF (1963) Layered pegmatite-aplite intru-sives Mineral Soc Am Sp Pap 1 78-92

Jolliff BL Papike JJ Shearer CK Laul JC (1986)Tourmaline as a recorder of pegmatite evolution Bob Ingersolpegmatite Black Hills South Dakota Am Mineral 71 472-500

Kalt A Schreyer W Ludwig T Prowatke S Bernhardt H ErtlA (2001) Complete solid solution between magnesian schorland lithian excess-boron olenite in a pegmatite from the Koralpe(eastern Alps Austria) Eur J Mineral 13 1191-1205

Kampf AR Gochenour K Clanin J (2003) Tourmaline dis-covery at the Cryo-Genie mine San Diego County CaliforniaRocks and Minerals 78 156-163

Liebscher A Meixner A Romer RL Heinrich W (2007)Experimental calibration of the vapourndashliquid phase relationsand lithium isotope fractionation in the system H2OndashLiCl at400 Geofluids 7 1-7

910 JS Maloney et al

Cryo-Genie dike These tourmalines are at most few mil-limeters in length and are prismatic Li concentrations inradial tourmalines in the pegmatite zones of both the Cryo-Genie and Little Three dikes and in large tourmaline bladesin the core zones are on average higher by about an orderof magnitude The concentrations are highly variable how-ever In prismatic schorl crystals within pockets the con-centrations range from 273 to 954 ppm Two green over-growths on schorl cores have the highest Li concentrations1456 and 5075 ppm

Li isotope ratios

In contrast to the progressive increase in Li concentrationsin tourmalines across the zones of the dikes the variationin δ7Li is more complex (Fig 4b) There is no correlationwith Li concentrations but again there appears to be nosystematic difference in δ7Li between the three dikes Intourmalines from the wall-zoneline-rock core zone andpockets δ7Li has approximately the same range between112 and 161 In the pockets elbaite rims have lighterLi than schorl cores although the difference in the Cryo-Genie pair is within overlapping errors In three tourma-lines from the upper and the lower pegmatite zones ofboth Cryo-Genie and Little Three dikes δ7Li is very highgt 19 These three tourmalines came from radial splayswhereas others with lighter Li were prismatic or blade-shaped (Fig 3b) δ7Li in prismatic tourmaline from thebottom pegmatite zone of the Little Three dike is lower at142

The obtained δ7Li values and their range are among thehighest measured in rocks δ7Li in most unaltered crustalrocks typically does not exceed 10 (Tomascak 2004Teng et al 2004) but values up to 19 in quartz anda 10 difference between quartz and albite have beenreported for the Tin Mountain pegmatite in the Black HillsSouth Dakota USA (Teng et al 2006b)

Initial Li concentration in pegmatite melts

Lithium is often invoked as an element that may contributeto growth of large crystals that characterize pegmatites bydepolymerizing the silicate melt structure and thereforelowering its viscosity and increasing chemical diffusionrates In order to evaluate the fluxing effect of Li comparedto H2O in the San Diego County pegmatites a crude es-timate was made of its initial concentration in the meltsThe estimate comes from the Li concentrations in tourma-line the modal abundance of tourmaline and the Li con-centrations in fluid inclusions in pocket quartz assumingthat the pockets were the collection volumes for the sep-arated fluid Tourmaline is the largest reservoir for Li inthe line rock and pegmatite but in the cores and pocketswhere other Li-bearing minerals including lepidolite andamblygonite occur Li concentration in the fluid is best es-timated from primary fluid inclusions in quartz In the Lit-tle Three pegmatite these inclusions have a range of 7000 to12 000 ppm with average of 9200 ppm Li (unpublished data

determined by ion chromatography) The inclusion fluidscontain sim 3 wt NaCl based on microthermometric mea-surements

The initial concentration of Li in the pegmatite melts (Co)can be estimated using

Co = XLRCLR + XPEGCPEG + XPOCKCPOCK

where X is the mass proportion of each zone and C isthe concentration of Li in each zone Li concentrationsin the line-rock and the pegmatite zones were determinedfrom the average of Li concentrations in tourmaline in eachzone and the modal proportions of tourmaline Data for theCryo-Genie and Little Three dikes were combined Com-bining the data for the two dikes is justified because in thepegmatite zones which are volumetrically most abundantLi concentrations cover the same ranges The average Liconcentration in tourmaline in the Cryo-Genie pegmatitezones is 258 ppm and in the Little Three it is 298 ppmThe mass proportions of the three zones were determinedfrom their relative volumes and densities 2700 kgm3 forthe rocks and 500 kgm3 for the fluid The fluid density isappropriate for H2O with 3 wt NaCl at 2 kbar and 400 C(Anderko amp Pitzer 1993) The proportion of tourmaline inthe line-rock and pegmatite zones is lt 8 volume basedon image analysis of cut slabs The estimated bulk Li con-centration in the line-rock zones is only sim 5 ppm and in thepegmatite zones only sim 60 ppm It is readily evident thatmost Li from the original melt ended-up in the fluids

The pockets show that the exsolved fluid was collectedin discrete spaces instead of one continuous space betweenthe hanging and lower portions of the dikes Because thevolume and the distribution of the pockets are highly vari-able (Jahns 1979 Stern et al 1986) the volume of theexsolved fluid is difficult to estimate from the field occur-rences of the pockets A better estimate of the volume ofthe exsolved fluid comes from the maximum H2O solubil-ity in silicate melts which at 2 kbar is sim 6 wt (Holtzet al 1995) This amount of H2O given its molar volumeat 400 C and 2 kbar would occupy sim 28 of the chambervolume Using this proportion the calculated initial Li con-centration in the pegmatite dikes is only sim 630 ppm If theCryo-Genie data is left out of the calculation the differencein the result is only 1 ppm Although the ranges of the mea-sured Li concentrations and errors in the volume estimatesof tourmaline and the textural zones contribute to an erroron this estimate the estimate is dominated by the calculatedfluid volume and the Li concentration in the fluid as deter-mined from fluid inclusions Although 630 ppm Li couldappear to be a rather small concentration for a LCT-typemelt it is 18 times greater than the estimated average con-centration of 35 ppm in the upper continental crust (Tenget al 2004)

Discussion

Role of lithium in the pegmatite crystallization process

Since the classic model of Jahns amp Burnham (1969)for crystallization of pegmatite melts which involves the

Lithium and its isotopes in pegmatite tourmalines 911

separation of a fluid phase from a silicate melt it has beenrecognized that the formation of pegmatites also dependson kinetic controls during rapid cooling of dikes in par-ticular enrichments of components in melt boundary lay-ers ahead of rapidly crystallizing minerals (eg Rockholdet al 1987 London 1992 Webber et al 1999) In addi-tion to being responsible for crystallization of minerals inwhich it is an essential structural constituent Li is ofteninvoked as a possible component that along with H2O BF and other species can potentially flux granite melts Forexample fractional crystallization involving Li was partlyresponsible for mineral zoning seen in the large LCT-typeTin Mountain pegmatite in the Black Hills of South Dakotain which the outer zones are dominated by feldspars andthe inner zones by spodumene and quartz (Walker et al1986) Trapping temperatures of primary fluid inclusionsare lt 400 C and nearly invariable across the pegmatitesuggesting that it crystallized nearly isothermally as an un-dercooled liquid (Sirbescu amp Nabelek 2003a)

An extreme kinetic model is that of Morgan amp London(1999) for crystallization of the Little Three pegmatiteThey suggested that the low temperature and the fast cool-ing rate that must have occurred during solidification of thepegmatite did not allow for crystal nucleation until the melthas reached a glass state at sim 250 C below the equilibriumliquidus They proposed a constitutional zone-refining pro-cess in which a fluxed crystallization front swept a F Liand Mn-rich boundary layer through the solid or semi-soliddike eventually resulting in enrichment of these elementsin the pocket zone

Although we did not obtain electron microprobe data ontourmaline from the dikes the electron microprobe dataof Morgan and London (1999) on tourmaline in the Lit-tle Three dike suggests that the progressive increase in Liconcentrations from the line-rock and the upper wall-zoneto the pockets (Fig 4a) corresponds to changes in othertourmaline components Morgan amp London (1999) foundthat across the line-rock and the pegmatite zones Mg de-creases from sim 07 to near 0 per formula unit while Fe staysnearly constant The increase in Li and the decrease in Mgsuggests an increasing exchange of the elbaite componentfor the dravite component In the pockets tourmaline iszoned from schorl to elbaite but the zoning appears to becontinuous without an evidence for a miscibility gap (Mor-gan amp London 1999) consistent with evidence for com-plete solid solution between schorl and lithian olenite ina pegmatite from the eastern Alps (Kalt et al 2001) El-baite in the pockets has elevated F and Mn concentrations(Morgan amp London 1999) The occurrence of schorl coresin the pockets suggests that schorl grew while Fe-bearingmelt was still present in the dikes but the elbaite rims to-gether with other lithium minerals in pegmatite cores andpockets grew in equilibrium with Li-rich fluid collected inthe pockets The progressive increase in Li across the dikesis more consistent with progressive inward crystallizationof the dikes than a zone-refining process as Li appears tohave been progressively enriched in the residual liquid be-cause of its low solubility in early-crystallizing mineralsincluding schorl The change in tourmaline compositionacross the San Diego County pegmatites is analogous to

the tourmaline composition trend in the Bob Ingersol peg-matite in the Black Hills (Jolliff et al 1986)

The transition from the aplitic line-rock to pegmatiteprobably marks the point of fluid separation in the magmaFluid separation is suggested by the occurrences of elon-gated radial crystals that characterize the pegmatite zonesbecause crystals grow faster and become elongated whenH2O activity increases in the melt (Fenn 1977) HighH2O activity promotes an increase in the diffusion ratesof chemical components even when the melt is undergo-ing rapid cooling Crystallization of tourmaline (and gar-net) is controlled more by gradients in concentrations ofless mobile elements including Fe Mg and Mn than bygradients of rapidly-diffusing elements especially the al-kalis (Rockhold et al 1987 Webber et al 1999) Thelack of crystallization of Li minerals such as spodumeneand amblygonite until the core zones was probably pre-cluded by the initially low Li concentration and removalof Li by the exsolving fluid Although in Cl-absent water-peraluminous melt systems D(Li)fluidmelt is sim 04 (Londonet al 1988) D(Li)fluidmelt increases with the addition ofCl For example when an aqueous fluid at 800 C2 kbarhas sim 7 wt Cl D(Li)fluidmelt is sim 2 (Webster et al 1989)However even this D(Li)fluidmelt seems insufficient to ex-plain the very elevated Li in the pocket fluids by simplebatch partitioning between the fluid and the melt particu-larly because the fluid contained only sim 2 wt Cl Morelikely the large concentration of Li in the fluid is the re-sult of Rayleigh enrichment with crystallization dominatedby feldspars quartz and schorl tourmaline in which Li hassmall solubility compared to the melt and the fluid At 88 crystallization Li would have exceeded 05 wt (11 wtLi2O) in the melt and a correspondingly high concentrationin the accumulating fluid

Concentrations in excess of 1 wt Li2O that may haveexisted at later stages of fractional crystallization of thedikes are approached in some large spodumene-bearingpegmatite intrusions including the Tin Mountain peg-matite the Harding pegmatite in New Mexico and theTanco pegmatite in Manitoba (Norton 1994) These largepegmatites contain spodumene andor other Li-bearingminerals not only in their cores but also other zones (Nor-ton 1994) Li2O concentrations approaching 1 wt in asilicate melt may be required for crystallization of miner-als in which Li is an essential structural constituent In theSan Diego pegmatites such high concentrations apparentlyexisted only in the core zones and the pockets

The influence of the estimated initial 630 ppm Li on theviscosity of the dikes was likely far smaller in compari-son with the influence of the sim 6 wt H2O that wouldbe in the melt at the point of saturation Even in the corezones where Li2O may have reached 1 its effect wouldlikely have been much smaller 1 Li2O is equivalent to19 mole in a haplogranite melt whereas 6 wt H2Ois equivalent to sim 185 mole The addition of 1 wtof excess Li to a haplogranite melt lowers the viscosityby about one order of magnitude (Dingwell et al 1996)but in a peraluminous melt where Li may be complexedwith Al in a Si4+ = Al3+ + Li+ substitution the effect ofLi addition is probably smaller Even if Li exceeded its

912 JS Maloney et al

charge-balancing role in the melt upon removal of Al bycrystallization of the highly peraluminous tourmaline or ifAl was complexed with F in a non-network position in themelt (Giordano et al 2004 Mysen amp Richet 2005) Li isnot expected to have had a drastic effect on the viscosity ofthe dikes In comparison when H2O is added to a silicatemelt it initially dissolves by forming Al-OH and Si-OHcomplexes which depolymerize the tetrahedral networkViscosity drops by sim 45 orders of magnitude with the ad-dition of only 1 wt H2O to a peraluminous leucogranitemelt at 600 C and by an additional 6 orders of magnitudewith further addition of 5 wt H2O (Romano et al 2001Whittington et al 2004) The decreasing effect of H2O ad-dition on the viscosity comes from the increasingly greaterdissolution of H2O as a molecular species (Stolper 1982)Likewise diffusivities of ions in silicate melts dramaticallyincrease with the addition of only a small amount of H2Obut less so with further addition of H2O (Watson 1994)

Lithium isotope fractionation

A potentially large fractionation of Li-isotopes in any mul-tiphase system is due to the 17 mass difference be-tween 7Li and 6Li Presently available data show only verylimited Li isotope fractionation during crystallization ofhigh-temperature igneous systems (Tomascak et al 1999Magna et al 2006 Halama et al 2007) and during vapor-liquid separation in hydrothermal systems (Foustoukoset al 2004 Liebscher et al 2007) However significantfractionation may result from different coordination statesof Li in coexisting phases that include multiple miner-als melts and aqueous fluids in relatively low-temperaturegranitic systems (Wenger amp Armbruster 1991) In general6Li preferentially occupies sites with higher coordinationnumbers and therefore weaker bonds whereas 7Li prefer-entially enters sites with smaller coordination numbers andstronger bonds (Wunder et al 2007)

Fractionation of Li between minerals and fluids hasbeen experimentally determined only for a limited set ofminerals to date Wunder et al (2006 2007) found thatΔ7Listauroliteminusfluid = +13 and is essentially temperature-independent Δ7Lilepidoliteminusfluid is approximately ndash2 withsome temperature dependence in the 350ndash400 C rangeΔ7Lispodumeneminusfluid is also temperature-dependent but morenegative by about 3 For all three minerals the frac-tionation is insensitive to the Cl content of the fluid whichimplies that Li probably forms a tetrahedrally-coordinatedhydrated ion Li(H2O)+4 instead of a LiCl or LiOH com-plex Wunder et al (2007) concluded that equilibrium iso-topic fractionation is firstly controlled by Li coordina-tion with 7Li preferentially incorporated into the phasethat allows for a smaller coordination number and sec-ondly by the Li-O bond length giving the relationshipδ7Listaurolite gt δ

7Lilepidolite gt δ7Lispodumene In staurolite

Li substitutes for the divalent cations Fe2+ Mg and Znin the tetrahedral sites in lepidolite Li is octahedrally-coordinated between tetrahedral layers and in spodumeneit occupies the relatively large M2 octahedral site Thereare no experimental Li isotope fractionation data involv-

ing either tourmaline or melt but in tourmaline Li occupiesthe octahedrally-coordinated Y-site and in a peraluminousmelt Li is probably strongly bonded in its charge balanc-ing role with tetrahedrally-coordinated Al3+ in a LiAlSi3O8complex (Mysen amp Richet 2005) Therefore Li in such amelt should be isotopically relatively heavy

The results of Wunder et al (2007) involving muscoviteare quite different from the results of Lynton et al (2005)who found Δ7Limuscoviteminusfluid to range between +8 and +20in the 400ndash500 C interval Wunder et al (2007) attributedthe discrepancy to the diffusion mechanism that Lyntonet al (2005) used to introduce Li into muscovite IndeedTeng et al (2006a) ascribed very large variations in δ7Li ofcountry rocks in the aureole of the Tin Mountain pegmatiteto differential diffusion of the two Li isotopes which under-scores that in addition to the energy of bonds in lattices ki-netic effects may induce transient Li isotope fractionationwhich may be preserved in rapidly cooled systems

There have been no direct measurements of Li iso-tope fractionation between peraluminous silicate melts andaqueous fluids but it is expected that Li in a melt should beisotopically heavier because of strong bonds in associationwith charge balancing of Al that is in tetrahedral coordina-tion This inference is supported by isotopic compositionsof Li in fluid inclusions and host quartz in the Tin Mountainpegmatite in the Black Hills (Teng et al 2006b) Li in thefluid inclusions has much lower δ7Li values than Li in thequartz supporting the inference that 7Li prefers the strongbonds in quartz where it is possibly charge-balancing Althat is incorporated into the quartz structure and by anal-ogy bonds in high-silica melts over the weaker hydratedbonds in the fluid

The San Diego pegmatites were systems in which min-erals melts and aqueous fluids coexisted at various stagesof crystallization Assuming equilibrium the isotopic com-position of Li in tourmaline in the dikes is reflective of themedium from which the tourmaline crystallized HoweverLi isotopes may be strongly fractionated by kinetic effectsThere is a several permil heterogeneity in δ7Li even in indi-vidual zones of a single pegmatite Most interesting are thevery elevated gt 19 δ7Li values of radial tourmalines inboth the hanging and lower pegmatite zones A reason forthe elevated values may be that the tourmalines grew whenthe melt became saturated in the aqueous fluid at whichpoint Li isotopes were fractionated between the melt andthe aqueous fluid with most Li going into the fluid (Fig 5)If the isotopic composition of Li in tourmaline in each ofthe zones reflects the relative fractionation of Li isotopesbetween the melt and the fluid then the isotopically heavierLi in the pegmatite zone tourmaline is consistent with theexpected stronger Li bonds in the melt compared to bondsof hydrated Li in the fluid The similarity of δ7Li values intourmaline that crystallized in the line rock and the pock-ets is consistent with accumulation of the bulk of meltrsquosinitial Li into the pocket fluid as this would result in littlechange in the isotopic composition of Li from the initialfluid-undersatured melt to the eventually collected fluid inthe pockets Elbaite rims of pocket tourmalines have lowerδ7Li than their corresponding cores although one core-rimpair has overlapping analytical errors (Fig 4b) The lower

Lithium and its isotopes in pegmatite tourmalines 913

Fig 5 Schematic diagram illustrating a possible mechanism of pro-ducing high δ7Li tourmalines in pegmatite zones during fluid-meltseparation Because Li is used in charge-balancing with strongly-bonded Al in the silicate melt the δ7Li in the melt should be moreelevated than δ7Li in the fluid where Li probably occurs mostly asa hydrated ion (Wunder et al 2007) Tourmalines crystallizing inequilibrium with the melt and fluid respectively may reflect the iso-topic fractionation between the melt and the fluid

δ7Li values of the elbaite rims are consistent with crystal-lization of schorl cores while melt was still present andcrystallization of elbaite rims in the presence of the fluidonly

The lack of a systematic increase in δ7Li across the tex-tural zones of the dikes suggests that there was little in-fluence of tourmaline itself on the isotopic composition ofthe residual melt during crystallization Given that bothΔ7Lispodumeneminusfluid and Δ7Lilepidoliteminusfluid are both negativeΔ7Litourmalineminusmelt should be even more negative because Liin tourmaline is in octahedral coordination while in themelt it is associated with tetrahedral Al However becausethe proportion of tourmaline in the line-rock and pegmatitezones is lt 8 and the concentration of Li in the tourma-line is very small crystallization of tourmaline would havehad a negligible effect on the Li isotope ratio in the resid-ual melt Based on mass-balance calculations only sim 2 of the initial 630 ppm Li is contained in schorl in the line-rocks and the pegmatites zones

Kinetic effects

A kinetic cause for the elevated δ7Li values of the radialtourmalines in the pegmatite zones must also be consid-ered however The shape of the tourmalines suggests thatthey grew very rapidly in which case the Li isotope frac-tionation may have been kinetically controlled by differ-ential diffusion of Li isotopes at the crystal-liquid (andorcrystal-fluid) interface The relative diffusion rates of two

Fig 6 Schematic diagram illustrating a possible kinetic mechanismof producing high δ7Li radial tourmalines in pegmatite zones Be-cause the diffusion of 7Li is slower in the melt than the diffusion of6Li and when Li is an incompatible element relative to tourmaline7Li should become preferentially enriched in a boundary layer aheadof a tourmaline crystal that is growing faster than the rate at whichLi diffuses in the melt

isotopes of a given element in a silicate liquid are given by

D1

Dh=

(mh

m1

)β

where m is the mass of isotope ldquolrdquo stands for the light iso-tope and ldquohrdquo stands for the heavy isotope (Richter et al2003) Using basalt-rhyolite melt couples Richter et al(2003) experimentally determined that βLi asymp 0215 Thismeans that 6Li can diffuse substantially faster than 7Lithrough the melt away from a growing crystal so that apreferential enrichment of 7Li can potentially occur in thechemical boundary layer ahead of the crystal (Fig 6) re-sulting in elevated δ7Li The diffusion rate of Li in silicatemelts is orders of magnitude faster than the rates of othermajor and minor cations (Richter et al 2003) Under con-ditions of slow mineral growth homogeneous Li isotoperatios would be expected in the tourmalines Instead theobserved Li isotope heterogeneity suggests crystallizationconditions under which the diffusion of Li in the melt didnot keep-up with the rate of tourmaline growth

Elevated δ7Li values of pegmatites

The overall elevated δ7Li values seen in the San Diegopegmatites are similar to the values in the Tin Mountainpegmatite in the Black Hills δ7Li values in the associ-ated Harney Peak leucogranite and the host schists of theTin Mountain pegmatite have δ7Li values within a fewpermil of 0 which points to some process that leads tostrong Li isotope fractionation during generation of LCT-type pegmatite melts (Teng et al 2006b) The apparentlarge fractionation for these relatively low-temperature ig-neous systems contrasts with the minimal fractionation in

914 JS Maloney et al

high-temperature igneous systems (Tomascak 2004) Tenget al (2006b) suggested that elevated δ7Li values in peg-matite melts could potentially be acquired by crystal-liquidfractionation during crystallization of parental magmasHowever for this process to be effective a greater amountof Li would have to remain in the crystallized assemblage(parental granite) than in the residual liquid (pegmatite)This is inconsistent with for example the relative concen-trations of Li in the Harney Peak leucogranite and its poten-tially residual liquid now represented by the Tin Mountainpegmatite Li concentrations in the Harney Peak leucogran-ite range from 10ndash205 ppm (Teng et al 2006b)

A possible alternative explanation is that the structureof a hydrous pegmatite melt has more similarity to waterthan to minerals and therefore it incorporates 7Li pref-erentially over minerals in a fashion analogous to thelarge 7Li enrichment in crustal fluids compared to crustalrocks (Tomascak 2004) As hydrated pegmatite melts passthrough the crust from their sources they may acquire el-evated δ7Li values through rapid Li isotope exchange withthe surrounding rocks Matthews et al (2003) found thatthe oxygen and especially hydrogen isotopic compositionof pegmatite dikes on Naxos Greece reflects the isotopiccomposition of the host rocks which change in isotopiccomposition along strike of the dikes If hydrogen isotopescan be nearly fully exchanged between pegmatite melts andtheir host rocks then it is likely that Li isotopes can also bereadily exchanged given the fast diffusion rates of Li in sil-icate liquids (Richter et al 2003)

Acknowledgements We are grateful for access providedby Dana and Ken Gochenour to the Cryo-Genie propertyBill Calhoun to the San Diego mine property and LouisSpaulding Jr to the Little Three property Jeffrey Patter-son and Matt Taylor provided guidance around the peg-matite districts and Jim Student helped with sample col-lection Carol Nabelek oversaw the ICP-OES analysis BillMcDonough graciously gave access to JM to conduct anal-ysis in the isotope laboratory at the University of MarylandThe paper benefited from the constructive reviews of AxelLiebscher Jeffrey Ryan and Ed Grew and additional com-ments from Roberta Rudnick A Feodor-Lynen fellowshipto Halama by the Alexander von Humboldt Foundation isgratefully acknowledged The study was supported by Uni-versity of Missouri Research Board Grant D3508 and NSFGrant 408564 to Nabelek Additional funding came fromNSF Grant EAR 0606989 to Rudnick and McDonough

References

Anderko A amp Pitzer KS (1993) Equation-of-state representa-tion of phase equilibria and volumetric properties of the systemNaCl-H2O above 573 K Geochim Cosmochim Acta 57 1657-1680

Bouman C Elliott T Vroon PZ (2004) Lithium inputs to sub-duction zones Chem Geol 212 59-79

Cernyacute P (1991) Rare-element granitic pegmatites Part I Anatomyand internal evolution of pegmatite deposits GeoscienceCanada 18 49-67

Cernyacute P amp Ercit TS (2005) The classification of granitic peg-matites revisited Can Mineral 43 2005-2026

Dingwell DB Hess K-U Knoche R (1996) Granite andgranitic pegmatite melts volumes and viscosities Trans R SocEdinburgh Earth Sci 87 65-72

Fenn PM (1977) The nucleation and growth of alkali feldsparsfrom hydrous melts Can Mineral 15 135-161

Fisher J (2002) Gem and rare-element pegmatites of southernCalifornia Mineral Record 33 363-407

Fisher J Foord EE Bricker GA (1999) The geology miner-alogy and history of the Himalaya Mine Mesa Grande SanDiego County California California Geol 3-17

Foord EE (1976) Mineralogy and petrogenesis of layeredpegmatite-aplite dikes in the Mesa Grande District San DiegoCounty California PhD Dissertation Stanford University

Foord EE Starkey HC Taggard JE Jr (1986) Mineralogyand paragenesis of ldquopocketrdquo clays and associated minerals incomplex granitic pegmatites San Diego County California AmMineral 71 428-439

Foustoukos DI James RH Berndt ME Seyfried WE Jr(2004) Lithium isotopic systematics of hydrothermal vent flu-ids at the Main Endeavour Field Northern Juan de Fuca RidgeChem Geol 212 17-26

Giordano D Romano C Dingwell DB Poe B Behrens H(2004) The combined effects of water and fluorine on theviscosity of silicic magmas Geochim Cosmochim Acta 685159-5168

Halama R McDonough WF Rudnick RL Keller J KlaudiusJ (2007) The Li isotopic composition of Oldoinyo LengaiNature of the mantle sources and lack of isotopic fractionationduring carbonatite petrogenesis Earth Planet Sci Lett 25477-89

Holtz F Behrens H Dingwell DB Johannes W (1995) Watersolubility in haplogranite melts Compositional pressure andtemperature dependence Am Mineral 80 94-108

Jahns R H (1979) Gem-bearing pegmatites in San Diego CountyCalifornia The Stewart mine Pala district and the Himalayamine Mesa Grande district in ldquoMesozoic crystalline rocksPeninsular Ranges batholith and pegmatites Point Sol ophio-literdquo P L Abbott amp V R Todd eds San Diego State UniversitySan Diego California 3-38

Jahns RH amp Burnham CW (1969) Experimental studies of peg-matite genesis I A model for the derivation and crystallizationof granitic pegmatites Econ Geol 64 843-864

Jahns RH amp Tuttle OF (1963) Layered pegmatite-aplite intru-sives Mineral Soc Am Sp Pap 1 78-92

Jolliff BL Papike JJ Shearer CK Laul JC (1986)Tourmaline as a recorder of pegmatite evolution Bob Ingersolpegmatite Black Hills South Dakota Am Mineral 71 472-500

Kalt A Schreyer W Ludwig T Prowatke S Bernhardt H ErtlA (2001) Complete solid solution between magnesian schorland lithian excess-boron olenite in a pegmatite from the Koralpe(eastern Alps Austria) Eur J Mineral 13 1191-1205

Kampf AR Gochenour K Clanin J (2003) Tourmaline dis-covery at the Cryo-Genie mine San Diego County CaliforniaRocks and Minerals 78 156-163

Liebscher A Meixner A Romer RL Heinrich W (2007)Experimental calibration of the vapourndashliquid phase relationsand lithium isotope fractionation in the system H2OndashLiCl at400 Geofluids 7 1-7

Lithium and its isotopes in pegmatite tourmalines 911

separation of a fluid phase from a silicate melt it has beenrecognized that the formation of pegmatites also dependson kinetic controls during rapid cooling of dikes in par-ticular enrichments of components in melt boundary lay-ers ahead of rapidly crystallizing minerals (eg Rockholdet al 1987 London 1992 Webber et al 1999) In addi-tion to being responsible for crystallization of minerals inwhich it is an essential structural constituent Li is ofteninvoked as a possible component that along with H2O BF and other species can potentially flux granite melts Forexample fractional crystallization involving Li was partlyresponsible for mineral zoning seen in the large LCT-typeTin Mountain pegmatite in the Black Hills of South Dakotain which the outer zones are dominated by feldspars andthe inner zones by spodumene and quartz (Walker et al1986) Trapping temperatures of primary fluid inclusionsare lt 400 C and nearly invariable across the pegmatitesuggesting that it crystallized nearly isothermally as an un-dercooled liquid (Sirbescu amp Nabelek 2003a)

An extreme kinetic model is that of Morgan amp London(1999) for crystallization of the Little Three pegmatiteThey suggested that the low temperature and the fast cool-ing rate that must have occurred during solidification of thepegmatite did not allow for crystal nucleation until the melthas reached a glass state at sim 250 C below the equilibriumliquidus They proposed a constitutional zone-refining pro-cess in which a fluxed crystallization front swept a F Liand Mn-rich boundary layer through the solid or semi-soliddike eventually resulting in enrichment of these elementsin the pocket zone

Although we did not obtain electron microprobe data ontourmaline from the dikes the electron microprobe dataof Morgan and London (1999) on tourmaline in the Lit-tle Three dike suggests that the progressive increase in Liconcentrations from the line-rock and the upper wall-zoneto the pockets (Fig 4a) corresponds to changes in othertourmaline components Morgan amp London (1999) foundthat across the line-rock and the pegmatite zones Mg de-creases from sim 07 to near 0 per formula unit while Fe staysnearly constant The increase in Li and the decrease in Mgsuggests an increasing exchange of the elbaite componentfor the dravite component In the pockets tourmaline iszoned from schorl to elbaite but the zoning appears to becontinuous without an evidence for a miscibility gap (Mor-gan amp London 1999) consistent with evidence for com-plete solid solution between schorl and lithian olenite ina pegmatite from the eastern Alps (Kalt et al 2001) El-baite in the pockets has elevated F and Mn concentrations(Morgan amp London 1999) The occurrence of schorl coresin the pockets suggests that schorl grew while Fe-bearingmelt was still present in the dikes but the elbaite rims to-gether with other lithium minerals in pegmatite cores andpockets grew in equilibrium with Li-rich fluid collected inthe pockets The progressive increase in Li across the dikesis more consistent with progressive inward crystallizationof the dikes than a zone-refining process as Li appears tohave been progressively enriched in the residual liquid be-cause of its low solubility in early-crystallizing mineralsincluding schorl The change in tourmaline compositionacross the San Diego County pegmatites is analogous to

the tourmaline composition trend in the Bob Ingersol peg-matite in the Black Hills (Jolliff et al 1986)

The transition from the aplitic line-rock to pegmatiteprobably marks the point of fluid separation in the magmaFluid separation is suggested by the occurrences of elon-gated radial crystals that characterize the pegmatite zonesbecause crystals grow faster and become elongated whenH2O activity increases in the melt (Fenn 1977) HighH2O activity promotes an increase in the diffusion ratesof chemical components even when the melt is undergo-ing rapid cooling Crystallization of tourmaline (and gar-net) is controlled more by gradients in concentrations ofless mobile elements including Fe Mg and Mn than bygradients of rapidly-diffusing elements especially the al-kalis (Rockhold et al 1987 Webber et al 1999) Thelack of crystallization of Li minerals such as spodumeneand amblygonite until the core zones was probably pre-cluded by the initially low Li concentration and removalof Li by the exsolving fluid Although in Cl-absent water-peraluminous melt systems D(Li)fluidmelt is sim 04 (Londonet al 1988) D(Li)fluidmelt increases with the addition ofCl For example when an aqueous fluid at 800 C2 kbarhas sim 7 wt Cl D(Li)fluidmelt is sim 2 (Webster et al 1989)However even this D(Li)fluidmelt seems insufficient to ex-plain the very elevated Li in the pocket fluids by simplebatch partitioning between the fluid and the melt particu-larly because the fluid contained only sim 2 wt Cl Morelikely the large concentration of Li in the fluid is the re-sult of Rayleigh enrichment with crystallization dominatedby feldspars quartz and schorl tourmaline in which Li hassmall solubility compared to the melt and the fluid At 88 crystallization Li would have exceeded 05 wt (11 wtLi2O) in the melt and a correspondingly high concentrationin the accumulating fluid

Concentrations in excess of 1 wt Li2O that may haveexisted at later stages of fractional crystallization of thedikes are approached in some large spodumene-bearingpegmatite intrusions including the Tin Mountain peg-matite the Harding pegmatite in New Mexico and theTanco pegmatite in Manitoba (Norton 1994) These largepegmatites contain spodumene andor other Li-bearingminerals not only in their cores but also other zones (Nor-ton 1994) Li2O concentrations approaching 1 wt in asilicate melt may be required for crystallization of miner-als in which Li is an essential structural constituent In theSan Diego pegmatites such high concentrations apparentlyexisted only in the core zones and the pockets

The influence of the estimated initial 630 ppm Li on theviscosity of the dikes was likely far smaller in compari-son with the influence of the sim 6 wt H2O that wouldbe in the melt at the point of saturation Even in the corezones where Li2O may have reached 1 its effect wouldlikely have been much smaller 1 Li2O is equivalent to19 mole in a haplogranite melt whereas 6 wt H2Ois equivalent to sim 185 mole The addition of 1 wtof excess Li to a haplogranite melt lowers the viscosityby about one order of magnitude (Dingwell et al 1996)but in a peraluminous melt where Li may be complexedwith Al in a Si4+ = Al3+ + Li+ substitution the effect ofLi addition is probably smaller Even if Li exceeded its

912 JS Maloney et al

charge-balancing role in the melt upon removal of Al bycrystallization of the highly peraluminous tourmaline or ifAl was complexed with F in a non-network position in themelt (Giordano et al 2004 Mysen amp Richet 2005) Li isnot expected to have had a drastic effect on the viscosity ofthe dikes In comparison when H2O is added to a silicatemelt it initially dissolves by forming Al-OH and Si-OHcomplexes which depolymerize the tetrahedral networkViscosity drops by sim 45 orders of magnitude with the ad-dition of only 1 wt H2O to a peraluminous leucogranitemelt at 600 C and by an additional 6 orders of magnitudewith further addition of 5 wt H2O (Romano et al 2001Whittington et al 2004) The decreasing effect of H2O ad-dition on the viscosity comes from the increasingly greaterdissolution of H2O as a molecular species (Stolper 1982)Likewise diffusivities of ions in silicate melts dramaticallyincrease with the addition of only a small amount of H2Obut less so with further addition of H2O (Watson 1994)

Lithium isotope fractionation

A potentially large fractionation of Li-isotopes in any mul-tiphase system is due to the 17 mass difference be-tween 7Li and 6Li Presently available data show only verylimited Li isotope fractionation during crystallization ofhigh-temperature igneous systems (Tomascak et al 1999Magna et al 2006 Halama et al 2007) and during vapor-liquid separation in hydrothermal systems (Foustoukoset al 2004 Liebscher et al 2007) However significantfractionation may result from different coordination statesof Li in coexisting phases that include multiple miner-als melts and aqueous fluids in relatively low-temperaturegranitic systems (Wenger amp Armbruster 1991) In general6Li preferentially occupies sites with higher coordinationnumbers and therefore weaker bonds whereas 7Li prefer-entially enters sites with smaller coordination numbers andstronger bonds (Wunder et al 2007)

Fractionation of Li between minerals and fluids hasbeen experimentally determined only for a limited set ofminerals to date Wunder et al (2006 2007) found thatΔ7Listauroliteminusfluid = +13 and is essentially temperature-independent Δ7Lilepidoliteminusfluid is approximately ndash2 withsome temperature dependence in the 350ndash400 C rangeΔ7Lispodumeneminusfluid is also temperature-dependent but morenegative by about 3 For all three minerals the frac-tionation is insensitive to the Cl content of the fluid whichimplies that Li probably forms a tetrahedrally-coordinatedhydrated ion Li(H2O)+4 instead of a LiCl or LiOH com-plex Wunder et al (2007) concluded that equilibrium iso-topic fractionation is firstly controlled by Li coordina-tion with 7Li preferentially incorporated into the phasethat allows for a smaller coordination number and sec-ondly by the Li-O bond length giving the relationshipδ7Listaurolite gt δ

7Lilepidolite gt δ7Lispodumene In staurolite

Li substitutes for the divalent cations Fe2+ Mg and Znin the tetrahedral sites in lepidolite Li is octahedrally-coordinated between tetrahedral layers and in spodumeneit occupies the relatively large M2 octahedral site Thereare no experimental Li isotope fractionation data involv-

ing either tourmaline or melt but in tourmaline Li occupiesthe octahedrally-coordinated Y-site and in a peraluminousmelt Li is probably strongly bonded in its charge balanc-ing role with tetrahedrally-coordinated Al3+ in a LiAlSi3O8complex (Mysen amp Richet 2005) Therefore Li in such amelt should be isotopically relatively heavy

The results of Wunder et al (2007) involving muscoviteare quite different from the results of Lynton et al (2005)who found Δ7Limuscoviteminusfluid to range between +8 and +20in the 400ndash500 C interval Wunder et al (2007) attributedthe discrepancy to the diffusion mechanism that Lyntonet al (2005) used to introduce Li into muscovite IndeedTeng et al (2006a) ascribed very large variations in δ7Li ofcountry rocks in the aureole of the Tin Mountain pegmatiteto differential diffusion of the two Li isotopes which under-scores that in addition to the energy of bonds in lattices ki-netic effects may induce transient Li isotope fractionationwhich may be preserved in rapidly cooled systems

There have been no direct measurements of Li iso-tope fractionation between peraluminous silicate melts andaqueous fluids but it is expected that Li in a melt should beisotopically heavier because of strong bonds in associationwith charge balancing of Al that is in tetrahedral coordina-tion This inference is supported by isotopic compositionsof Li in fluid inclusions and host quartz in the Tin Mountainpegmatite in the Black Hills (Teng et al 2006b) Li in thefluid inclusions has much lower δ7Li values than Li in thequartz supporting the inference that 7Li prefers the strongbonds in quartz where it is possibly charge-balancing Althat is incorporated into the quartz structure and by anal-ogy bonds in high-silica melts over the weaker hydratedbonds in the fluid

The San Diego pegmatites were systems in which min-erals melts and aqueous fluids coexisted at various stagesof crystallization Assuming equilibrium the isotopic com-position of Li in tourmaline in the dikes is reflective of themedium from which the tourmaline crystallized HoweverLi isotopes may be strongly fractionated by kinetic effectsThere is a several permil heterogeneity in δ7Li even in indi-vidual zones of a single pegmatite Most interesting are thevery elevated gt 19 δ7Li values of radial tourmalines inboth the hanging and lower pegmatite zones A reason forthe elevated values may be that the tourmalines grew whenthe melt became saturated in the aqueous fluid at whichpoint Li isotopes were fractionated between the melt andthe aqueous fluid with most Li going into the fluid (Fig 5)If the isotopic composition of Li in tourmaline in each ofthe zones reflects the relative fractionation of Li isotopesbetween the melt and the fluid then the isotopically heavierLi in the pegmatite zone tourmaline is consistent with theexpected stronger Li bonds in the melt compared to bondsof hydrated Li in the fluid The similarity of δ7Li values intourmaline that crystallized in the line rock and the pock-ets is consistent with accumulation of the bulk of meltrsquosinitial Li into the pocket fluid as this would result in littlechange in the isotopic composition of Li from the initialfluid-undersatured melt to the eventually collected fluid inthe pockets Elbaite rims of pocket tourmalines have lowerδ7Li than their corresponding cores although one core-rimpair has overlapping analytical errors (Fig 4b) The lower

Lithium and its isotopes in pegmatite tourmalines 913

Fig 5 Schematic diagram illustrating a possible mechanism of pro-ducing high δ7Li tourmalines in pegmatite zones during fluid-meltseparation Because Li is used in charge-balancing with strongly-bonded Al in the silicate melt the δ7Li in the melt should be moreelevated than δ7Li in the fluid where Li probably occurs mostly asa hydrated ion (Wunder et al 2007) Tourmalines crystallizing inequilibrium with the melt and fluid respectively may reflect the iso-topic fractionation between the melt and the fluid

δ7Li values of the elbaite rims are consistent with crystal-lization of schorl cores while melt was still present andcrystallization of elbaite rims in the presence of the fluidonly

The lack of a systematic increase in δ7Li across the tex-tural zones of the dikes suggests that there was little in-fluence of tourmaline itself on the isotopic composition ofthe residual melt during crystallization Given that bothΔ7Lispodumeneminusfluid and Δ7Lilepidoliteminusfluid are both negativeΔ7Litourmalineminusmelt should be even more negative because Liin tourmaline is in octahedral coordination while in themelt it is associated with tetrahedral Al However becausethe proportion of tourmaline in the line-rock and pegmatitezones is lt 8 and the concentration of Li in the tourma-line is very small crystallization of tourmaline would havehad a negligible effect on the Li isotope ratio in the resid-ual melt Based on mass-balance calculations only sim 2 of the initial 630 ppm Li is contained in schorl in the line-rocks and the pegmatites zones

Kinetic effects

A kinetic cause for the elevated δ7Li values of the radialtourmalines in the pegmatite zones must also be consid-ered however The shape of the tourmalines suggests thatthey grew very rapidly in which case the Li isotope frac-tionation may have been kinetically controlled by differ-ential diffusion of Li isotopes at the crystal-liquid (andorcrystal-fluid) interface The relative diffusion rates of two

Fig 6 Schematic diagram illustrating a possible kinetic mechanismof producing high δ7Li radial tourmalines in pegmatite zones Be-cause the diffusion of 7Li is slower in the melt than the diffusion of6Li and when Li is an incompatible element relative to tourmaline7Li should become preferentially enriched in a boundary layer aheadof a tourmaline crystal that is growing faster than the rate at whichLi diffuses in the melt

isotopes of a given element in a silicate liquid are given by

D1

Dh=

(mh

m1

)β

where m is the mass of isotope ldquolrdquo stands for the light iso-tope and ldquohrdquo stands for the heavy isotope (Richter et al2003) Using basalt-rhyolite melt couples Richter et al(2003) experimentally determined that βLi asymp 0215 Thismeans that 6Li can diffuse substantially faster than 7Lithrough the melt away from a growing crystal so that apreferential enrichment of 7Li can potentially occur in thechemical boundary layer ahead of the crystal (Fig 6) re-sulting in elevated δ7Li The diffusion rate of Li in silicatemelts is orders of magnitude faster than the rates of othermajor and minor cations (Richter et al 2003) Under con-ditions of slow mineral growth homogeneous Li isotoperatios would be expected in the tourmalines Instead theobserved Li isotope heterogeneity suggests crystallizationconditions under which the diffusion of Li in the melt didnot keep-up with the rate of tourmaline growth

Elevated δ7Li values of pegmatites

The overall elevated δ7Li values seen in the San Diegopegmatites are similar to the values in the Tin Mountainpegmatite in the Black Hills δ7Li values in the associ-ated Harney Peak leucogranite and the host schists of theTin Mountain pegmatite have δ7Li values within a fewpermil of 0 which points to some process that leads tostrong Li isotope fractionation during generation of LCT-type pegmatite melts (Teng et al 2006b) The apparentlarge fractionation for these relatively low-temperature ig-neous systems contrasts with the minimal fractionation in

914 JS Maloney et al

high-temperature igneous systems (Tomascak 2004) Tenget al (2006b) suggested that elevated δ7Li values in peg-matite melts could potentially be acquired by crystal-liquidfractionation during crystallization of parental magmasHowever for this process to be effective a greater amountof Li would have to remain in the crystallized assemblage(parental granite) than in the residual liquid (pegmatite)This is inconsistent with for example the relative concen-trations of Li in the Harney Peak leucogranite and its poten-tially residual liquid now represented by the Tin Mountainpegmatite Li concentrations in the Harney Peak leucogran-ite range from 10ndash205 ppm (Teng et al 2006b)

A possible alternative explanation is that the structureof a hydrous pegmatite melt has more similarity to waterthan to minerals and therefore it incorporates 7Li pref-erentially over minerals in a fashion analogous to thelarge 7Li enrichment in crustal fluids compared to crustalrocks (Tomascak 2004) As hydrated pegmatite melts passthrough the crust from their sources they may acquire el-evated δ7Li values through rapid Li isotope exchange withthe surrounding rocks Matthews et al (2003) found thatthe oxygen and especially hydrogen isotopic compositionof pegmatite dikes on Naxos Greece reflects the isotopiccomposition of the host rocks which change in isotopiccomposition along strike of the dikes If hydrogen isotopescan be nearly fully exchanged between pegmatite melts andtheir host rocks then it is likely that Li isotopes can also bereadily exchanged given the fast diffusion rates of Li in sil-icate liquids (Richter et al 2003)

Acknowledgements We are grateful for access providedby Dana and Ken Gochenour to the Cryo-Genie propertyBill Calhoun to the San Diego mine property and LouisSpaulding Jr to the Little Three property Jeffrey Patter-son and Matt Taylor provided guidance around the peg-matite districts and Jim Student helped with sample col-lection Carol Nabelek oversaw the ICP-OES analysis BillMcDonough graciously gave access to JM to conduct anal-ysis in the isotope laboratory at the University of MarylandThe paper benefited from the constructive reviews of AxelLiebscher Jeffrey Ryan and Ed Grew and additional com-ments from Roberta Rudnick A Feodor-Lynen fellowshipto Halama by the Alexander von Humboldt Foundation isgratefully acknowledged The study was supported by Uni-versity of Missouri Research Board Grant D3508 and NSFGrant 408564 to Nabelek Additional funding came fromNSF Grant EAR 0606989 to Rudnick and McDonough

References

Anderko A amp Pitzer KS (1993) Equation-of-state representa-tion of phase equilibria and volumetric properties of the systemNaCl-H2O above 573 K Geochim Cosmochim Acta 57 1657-1680

Bouman C Elliott T Vroon PZ (2004) Lithium inputs to sub-duction zones Chem Geol 212 59-79

Cernyacute P (1991) Rare-element granitic pegmatites Part I Anatomyand internal evolution of pegmatite deposits GeoscienceCanada 18 49-67

Cernyacute P amp Ercit TS (2005) The classification of granitic peg-matites revisited Can Mineral 43 2005-2026

Dingwell DB Hess K-U Knoche R (1996) Granite andgranitic pegmatite melts volumes and viscosities Trans R SocEdinburgh Earth Sci 87 65-72

Fenn PM (1977) The nucleation and growth of alkali feldsparsfrom hydrous melts Can Mineral 15 135-161

Fisher J (2002) Gem and rare-element pegmatites of southernCalifornia Mineral Record 33 363-407

Fisher J Foord EE Bricker GA (1999) The geology miner-alogy and history of the Himalaya Mine Mesa Grande SanDiego County California California Geol 3-17

Foord EE (1976) Mineralogy and petrogenesis of layeredpegmatite-aplite dikes in the Mesa Grande District San DiegoCounty California PhD Dissertation Stanford University

Foord EE Starkey HC Taggard JE Jr (1986) Mineralogyand paragenesis of ldquopocketrdquo clays and associated minerals incomplex granitic pegmatites San Diego County California AmMineral 71 428-439

Foustoukos DI James RH Berndt ME Seyfried WE Jr(2004) Lithium isotopic systematics of hydrothermal vent flu-ids at the Main Endeavour Field Northern Juan de Fuca RidgeChem Geol 212 17-26

Giordano D Romano C Dingwell DB Poe B Behrens H(2004) The combined effects of water and fluorine on theviscosity of silicic magmas Geochim Cosmochim Acta 685159-5168

Halama R McDonough WF Rudnick RL Keller J KlaudiusJ (2007) The Li isotopic composition of Oldoinyo LengaiNature of the mantle sources and lack of isotopic fractionationduring carbonatite petrogenesis Earth Planet Sci Lett 25477-89

Holtz F Behrens H Dingwell DB Johannes W (1995) Watersolubility in haplogranite melts Compositional pressure andtemperature dependence Am Mineral 80 94-108

Jahns R H (1979) Gem-bearing pegmatites in San Diego CountyCalifornia The Stewart mine Pala district and the Himalayamine Mesa Grande district in ldquoMesozoic crystalline rocksPeninsular Ranges batholith and pegmatites Point Sol ophio-literdquo P L Abbott amp V R Todd eds San Diego State UniversitySan Diego California 3-38

Jahns RH amp Burnham CW (1969) Experimental studies of peg-matite genesis I A model for the derivation and crystallizationof granitic pegmatites Econ Geol 64 843-864

Jahns RH amp Tuttle OF (1963) Layered pegmatite-aplite intru-sives Mineral Soc Am Sp Pap 1 78-92

Jolliff BL Papike JJ Shearer CK Laul JC (1986)Tourmaline as a recorder of pegmatite evolution Bob Ingersolpegmatite Black Hills South Dakota Am Mineral 71 472-500

Kalt A Schreyer W Ludwig T Prowatke S Bernhardt H ErtlA (2001) Complete solid solution between magnesian schorland lithian excess-boron olenite in a pegmatite from the Koralpe(eastern Alps Austria) Eur J Mineral 13 1191-1205

Kampf AR Gochenour K Clanin J (2003) Tourmaline dis-covery at the Cryo-Genie mine San Diego County CaliforniaRocks and Minerals 78 156-163

Liebscher A Meixner A Romer RL Heinrich W (2007)Experimental calibration of the vapourndashliquid phase relationsand lithium isotope fractionation in the system H2OndashLiCl at400 Geofluids 7 1-7

912 JS Maloney et al

charge-balancing role in the melt upon removal of Al bycrystallization of the highly peraluminous tourmaline or ifAl was complexed with F in a non-network position in themelt (Giordano et al 2004 Mysen amp Richet 2005) Li isnot expected to have had a drastic effect on the viscosity ofthe dikes In comparison when H2O is added to a silicatemelt it initially dissolves by forming Al-OH and Si-OHcomplexes which depolymerize the tetrahedral networkViscosity drops by sim 45 orders of magnitude with the ad-dition of only 1 wt H2O to a peraluminous leucogranitemelt at 600 C and by an additional 6 orders of magnitudewith further addition of 5 wt H2O (Romano et al 2001Whittington et al 2004) The decreasing effect of H2O ad-dition on the viscosity comes from the increasingly greaterdissolution of H2O as a molecular species (Stolper 1982)Likewise diffusivities of ions in silicate melts dramaticallyincrease with the addition of only a small amount of H2Obut less so with further addition of H2O (Watson 1994)

Lithium isotope fractionation

A potentially large fractionation of Li-isotopes in any mul-tiphase system is due to the 17 mass difference be-tween 7Li and 6Li Presently available data show only verylimited Li isotope fractionation during crystallization ofhigh-temperature igneous systems (Tomascak et al 1999Magna et al 2006 Halama et al 2007) and during vapor-liquid separation in hydrothermal systems (Foustoukoset al 2004 Liebscher et al 2007) However significantfractionation may result from different coordination statesof Li in coexisting phases that include multiple miner-als melts and aqueous fluids in relatively low-temperaturegranitic systems (Wenger amp Armbruster 1991) In general6Li preferentially occupies sites with higher coordinationnumbers and therefore weaker bonds whereas 7Li prefer-entially enters sites with smaller coordination numbers andstronger bonds (Wunder et al 2007)

Fractionation of Li between minerals and fluids hasbeen experimentally determined only for a limited set ofminerals to date Wunder et al (2006 2007) found thatΔ7Listauroliteminusfluid = +13 and is essentially temperature-independent Δ7Lilepidoliteminusfluid is approximately ndash2 withsome temperature dependence in the 350ndash400 C rangeΔ7Lispodumeneminusfluid is also temperature-dependent but morenegative by about 3 For all three minerals the frac-tionation is insensitive to the Cl content of the fluid whichimplies that Li probably forms a tetrahedrally-coordinatedhydrated ion Li(H2O)+4 instead of a LiCl or LiOH com-plex Wunder et al (2007) concluded that equilibrium iso-topic fractionation is firstly controlled by Li coordina-tion with 7Li preferentially incorporated into the phasethat allows for a smaller coordination number and sec-ondly by the Li-O bond length giving the relationshipδ7Listaurolite gt δ

7Lilepidolite gt δ7Lispodumene In staurolite

Li substitutes for the divalent cations Fe2+ Mg and Znin the tetrahedral sites in lepidolite Li is octahedrally-coordinated between tetrahedral layers and in spodumeneit occupies the relatively large M2 octahedral site Thereare no experimental Li isotope fractionation data involv-

ing either tourmaline or melt but in tourmaline Li occupiesthe octahedrally-coordinated Y-site and in a peraluminousmelt Li is probably strongly bonded in its charge balanc-ing role with tetrahedrally-coordinated Al3+ in a LiAlSi3O8complex (Mysen amp Richet 2005) Therefore Li in such amelt should be isotopically relatively heavy

The results of Wunder et al (2007) involving muscoviteare quite different from the results of Lynton et al (2005)who found Δ7Limuscoviteminusfluid to range between +8 and +20in the 400ndash500 C interval Wunder et al (2007) attributedthe discrepancy to the diffusion mechanism that Lyntonet al (2005) used to introduce Li into muscovite IndeedTeng et al (2006a) ascribed very large variations in δ7Li ofcountry rocks in the aureole of the Tin Mountain pegmatiteto differential diffusion of the two Li isotopes which under-scores that in addition to the energy of bonds in lattices ki-netic effects may induce transient Li isotope fractionationwhich may be preserved in rapidly cooled systems

There have been no direct measurements of Li iso-tope fractionation between peraluminous silicate melts andaqueous fluids but it is expected that Li in a melt should beisotopically heavier because of strong bonds in associationwith charge balancing of Al that is in tetrahedral coordina-tion This inference is supported by isotopic compositionsof Li in fluid inclusions and host quartz in the Tin Mountainpegmatite in the Black Hills (Teng et al 2006b) Li in thefluid inclusions has much lower δ7Li values than Li in thequartz supporting the inference that 7Li prefers the strongbonds in quartz where it is possibly charge-balancing Althat is incorporated into the quartz structure and by anal-ogy bonds in high-silica melts over the weaker hydratedbonds in the fluid

The San Diego pegmatites were systems in which min-erals melts and aqueous fluids coexisted at various stagesof crystallization Assuming equilibrium the isotopic com-position of Li in tourmaline in the dikes is reflective of themedium from which the tourmaline crystallized HoweverLi isotopes may be strongly fractionated by kinetic effectsThere is a several permil heterogeneity in δ7Li even in indi-vidual zones of a single pegmatite Most interesting are thevery elevated gt 19 δ7Li values of radial tourmalines inboth the hanging and lower pegmatite zones A reason forthe elevated values may be that the tourmalines grew whenthe melt became saturated in the aqueous fluid at whichpoint Li isotopes were fractionated between the melt andthe aqueous fluid with most Li going into the fluid (Fig 5)If the isotopic composition of Li in tourmaline in each ofthe zones reflects the relative fractionation of Li isotopesbetween the melt and the fluid then the isotopically heavierLi in the pegmatite zone tourmaline is consistent with theexpected stronger Li bonds in the melt compared to bondsof hydrated Li in the fluid The similarity of δ7Li values intourmaline that crystallized in the line rock and the pock-ets is consistent with accumulation of the bulk of meltrsquosinitial Li into the pocket fluid as this would result in littlechange in the isotopic composition of Li from the initialfluid-undersatured melt to the eventually collected fluid inthe pockets Elbaite rims of pocket tourmalines have lowerδ7Li than their corresponding cores although one core-rimpair has overlapping analytical errors (Fig 4b) The lower

Lithium and its isotopes in pegmatite tourmalines 913

Fig 5 Schematic diagram illustrating a possible mechanism of pro-ducing high δ7Li tourmalines in pegmatite zones during fluid-meltseparation Because Li is used in charge-balancing with strongly-bonded Al in the silicate melt the δ7Li in the melt should be moreelevated than δ7Li in the fluid where Li probably occurs mostly asa hydrated ion (Wunder et al 2007) Tourmalines crystallizing inequilibrium with the melt and fluid respectively may reflect the iso-topic fractionation between the melt and the fluid

δ7Li values of the elbaite rims are consistent with crystal-lization of schorl cores while melt was still present andcrystallization of elbaite rims in the presence of the fluidonly

The lack of a systematic increase in δ7Li across the tex-tural zones of the dikes suggests that there was little in-fluence of tourmaline itself on the isotopic composition ofthe residual melt during crystallization Given that bothΔ7Lispodumeneminusfluid and Δ7Lilepidoliteminusfluid are both negativeΔ7Litourmalineminusmelt should be even more negative because Liin tourmaline is in octahedral coordination while in themelt it is associated with tetrahedral Al However becausethe proportion of tourmaline in the line-rock and pegmatitezones is lt 8 and the concentration of Li in the tourma-line is very small crystallization of tourmaline would havehad a negligible effect on the Li isotope ratio in the resid-ual melt Based on mass-balance calculations only sim 2 of the initial 630 ppm Li is contained in schorl in the line-rocks and the pegmatites zones

Kinetic effects

A kinetic cause for the elevated δ7Li values of the radialtourmalines in the pegmatite zones must also be consid-ered however The shape of the tourmalines suggests thatthey grew very rapidly in which case the Li isotope frac-tionation may have been kinetically controlled by differ-ential diffusion of Li isotopes at the crystal-liquid (andorcrystal-fluid) interface The relative diffusion rates of two

Fig 6 Schematic diagram illustrating a possible kinetic mechanismof producing high δ7Li radial tourmalines in pegmatite zones Be-cause the diffusion of 7Li is slower in the melt than the diffusion of6Li and when Li is an incompatible element relative to tourmaline7Li should become preferentially enriched in a boundary layer aheadof a tourmaline crystal that is growing faster than the rate at whichLi diffuses in the melt

isotopes of a given element in a silicate liquid are given by

D1

Dh=

(mh

m1

)β

where m is the mass of isotope ldquolrdquo stands for the light iso-tope and ldquohrdquo stands for the heavy isotope (Richter et al2003) Using basalt-rhyolite melt couples Richter et al(2003) experimentally determined that βLi asymp 0215 Thismeans that 6Li can diffuse substantially faster than 7Lithrough the melt away from a growing crystal so that apreferential enrichment of 7Li can potentially occur in thechemical boundary layer ahead of the crystal (Fig 6) re-sulting in elevated δ7Li The diffusion rate of Li in silicatemelts is orders of magnitude faster than the rates of othermajor and minor cations (Richter et al 2003) Under con-ditions of slow mineral growth homogeneous Li isotoperatios would be expected in the tourmalines Instead theobserved Li isotope heterogeneity suggests crystallizationconditions under which the diffusion of Li in the melt didnot keep-up with the rate of tourmaline growth

Elevated δ7Li values of pegmatites

The overall elevated δ7Li values seen in the San Diegopegmatites are similar to the values in the Tin Mountainpegmatite in the Black Hills δ7Li values in the associ-ated Harney Peak leucogranite and the host schists of theTin Mountain pegmatite have δ7Li values within a fewpermil of 0 which points to some process that leads tostrong Li isotope fractionation during generation of LCT-type pegmatite melts (Teng et al 2006b) The apparentlarge fractionation for these relatively low-temperature ig-neous systems contrasts with the minimal fractionation in

914 JS Maloney et al

high-temperature igneous systems (Tomascak 2004) Tenget al (2006b) suggested that elevated δ7Li values in peg-matite melts could potentially be acquired by crystal-liquidfractionation during crystallization of parental magmasHowever for this process to be effective a greater amountof Li would have to remain in the crystallized assemblage(parental granite) than in the residual liquid (pegmatite)This is inconsistent with for example the relative concen-trations of Li in the Harney Peak leucogranite and its poten-tially residual liquid now represented by the Tin Mountainpegmatite Li concentrations in the Harney Peak leucogran-ite range from 10ndash205 ppm (Teng et al 2006b)

A possible alternative explanation is that the structureof a hydrous pegmatite melt has more similarity to waterthan to minerals and therefore it incorporates 7Li pref-erentially over minerals in a fashion analogous to thelarge 7Li enrichment in crustal fluids compared to crustalrocks (Tomascak 2004) As hydrated pegmatite melts passthrough the crust from their sources they may acquire el-evated δ7Li values through rapid Li isotope exchange withthe surrounding rocks Matthews et al (2003) found thatthe oxygen and especially hydrogen isotopic compositionof pegmatite dikes on Naxos Greece reflects the isotopiccomposition of the host rocks which change in isotopiccomposition along strike of the dikes If hydrogen isotopescan be nearly fully exchanged between pegmatite melts andtheir host rocks then it is likely that Li isotopes can also bereadily exchanged given the fast diffusion rates of Li in sil-icate liquids (Richter et al 2003)

Acknowledgements We are grateful for access providedby Dana and Ken Gochenour to the Cryo-Genie propertyBill Calhoun to the San Diego mine property and LouisSpaulding Jr to the Little Three property Jeffrey Patter-son and Matt Taylor provided guidance around the peg-matite districts and Jim Student helped with sample col-lection Carol Nabelek oversaw the ICP-OES analysis BillMcDonough graciously gave access to JM to conduct anal-ysis in the isotope laboratory at the University of MarylandThe paper benefited from the constructive reviews of AxelLiebscher Jeffrey Ryan and Ed Grew and additional com-ments from Roberta Rudnick A Feodor-Lynen fellowshipto Halama by the Alexander von Humboldt Foundation isgratefully acknowledged The study was supported by Uni-versity of Missouri Research Board Grant D3508 and NSFGrant 408564 to Nabelek Additional funding came fromNSF Grant EAR 0606989 to Rudnick and McDonough

References

Anderko A amp Pitzer KS (1993) Equation-of-state representa-tion of phase equilibria and volumetric properties of the systemNaCl-H2O above 573 K Geochim Cosmochim Acta 57 1657-1680

Bouman C Elliott T Vroon PZ (2004) Lithium inputs to sub-duction zones Chem Geol 212 59-79

Cernyacute P (1991) Rare-element granitic pegmatites Part I Anatomyand internal evolution of pegmatite deposits GeoscienceCanada 18 49-67

Cernyacute P amp Ercit TS (2005) The classification of granitic peg-matites revisited Can Mineral 43 2005-2026

Dingwell DB Hess K-U Knoche R (1996) Granite andgranitic pegmatite melts volumes and viscosities Trans R SocEdinburgh Earth Sci 87 65-72

Fenn PM (1977) The nucleation and growth of alkali feldsparsfrom hydrous melts Can Mineral 15 135-161

Fisher J (2002) Gem and rare-element pegmatites of southernCalifornia Mineral Record 33 363-407

Fisher J Foord EE Bricker GA (1999) The geology miner-alogy and history of the Himalaya Mine Mesa Grande SanDiego County California California Geol 3-17

Foord EE (1976) Mineralogy and petrogenesis of layeredpegmatite-aplite dikes in the Mesa Grande District San DiegoCounty California PhD Dissertation Stanford University

Foord EE Starkey HC Taggard JE Jr (1986) Mineralogyand paragenesis of ldquopocketrdquo clays and associated minerals incomplex granitic pegmatites San Diego County California AmMineral 71 428-439

Foustoukos DI James RH Berndt ME Seyfried WE Jr(2004) Lithium isotopic systematics of hydrothermal vent flu-ids at the Main Endeavour Field Northern Juan de Fuca RidgeChem Geol 212 17-26

Giordano D Romano C Dingwell DB Poe B Behrens H(2004) The combined effects of water and fluorine on theviscosity of silicic magmas Geochim Cosmochim Acta 685159-5168

Halama R McDonough WF Rudnick RL Keller J KlaudiusJ (2007) The Li isotopic composition of Oldoinyo LengaiNature of the mantle sources and lack of isotopic fractionationduring carbonatite petrogenesis Earth Planet Sci Lett 25477-89

Holtz F Behrens H Dingwell DB Johannes W (1995) Watersolubility in haplogranite melts Compositional pressure andtemperature dependence Am Mineral 80 94-108

Jahns R H (1979) Gem-bearing pegmatites in San Diego CountyCalifornia The Stewart mine Pala district and the Himalayamine Mesa Grande district in ldquoMesozoic crystalline rocksPeninsular Ranges batholith and pegmatites Point Sol ophio-literdquo P L Abbott amp V R Todd eds San Diego State UniversitySan Diego California 3-38

Jahns RH amp Burnham CW (1969) Experimental studies of peg-matite genesis I A model for the derivation and crystallizationof granitic pegmatites Econ Geol 64 843-864

Jahns RH amp Tuttle OF (1963) Layered pegmatite-aplite intru-sives Mineral Soc Am Sp Pap 1 78-92

Jolliff BL Papike JJ Shearer CK Laul JC (1986)Tourmaline as a recorder of pegmatite evolution Bob Ingersolpegmatite Black Hills South Dakota Am Mineral 71 472-500

Kalt A Schreyer W Ludwig T Prowatke S Bernhardt H ErtlA (2001) Complete solid solution between magnesian schorland lithian excess-boron olenite in a pegmatite from the Koralpe(eastern Alps Austria) Eur J Mineral 13 1191-1205

Kampf AR Gochenour K Clanin J (2003) Tourmaline dis-covery at the Cryo-Genie mine San Diego County CaliforniaRocks and Minerals 78 156-163

Liebscher A Meixner A Romer RL Heinrich W (2007)Experimental calibration of the vapourndashliquid phase relationsand lithium isotope fractionation in the system H2OndashLiCl at400 Geofluids 7 1-7

Lithium and its isotopes in pegmatite tourmalines 913

Fig 5 Schematic diagram illustrating a possible mechanism of pro-ducing high δ7Li tourmalines in pegmatite zones during fluid-meltseparation Because Li is used in charge-balancing with strongly-bonded Al in the silicate melt the δ7Li in the melt should be moreelevated than δ7Li in the fluid where Li probably occurs mostly asa hydrated ion (Wunder et al 2007) Tourmalines crystallizing inequilibrium with the melt and fluid respectively may reflect the iso-topic fractionation between the melt and the fluid

δ7Li values of the elbaite rims are consistent with crystal-lization of schorl cores while melt was still present andcrystallization of elbaite rims in the presence of the fluidonly

The lack of a systematic increase in δ7Li across the tex-tural zones of the dikes suggests that there was little in-fluence of tourmaline itself on the isotopic composition ofthe residual melt during crystallization Given that bothΔ7Lispodumeneminusfluid and Δ7Lilepidoliteminusfluid are both negativeΔ7Litourmalineminusmelt should be even more negative because Liin tourmaline is in octahedral coordination while in themelt it is associated with tetrahedral Al However becausethe proportion of tourmaline in the line-rock and pegmatitezones is lt 8 and the concentration of Li in the tourma-line is very small crystallization of tourmaline would havehad a negligible effect on the Li isotope ratio in the resid-ual melt Based on mass-balance calculations only sim 2 of the initial 630 ppm Li is contained in schorl in the line-rocks and the pegmatites zones

Kinetic effects

A kinetic cause for the elevated δ7Li values of the radialtourmalines in the pegmatite zones must also be consid-ered however The shape of the tourmalines suggests thatthey grew very rapidly in which case the Li isotope frac-tionation may have been kinetically controlled by differ-ential diffusion of Li isotopes at the crystal-liquid (andorcrystal-fluid) interface The relative diffusion rates of two

Fig 6 Schematic diagram illustrating a possible kinetic mechanismof producing high δ7Li radial tourmalines in pegmatite zones Be-cause the diffusion of 7Li is slower in the melt than the diffusion of6Li and when Li is an incompatible element relative to tourmaline7Li should become preferentially enriched in a boundary layer aheadof a tourmaline crystal that is growing faster than the rate at whichLi diffuses in the melt

isotopes of a given element in a silicate liquid are given by

D1

Dh=

(mh

m1

)β

where m is the mass of isotope ldquolrdquo stands for the light iso-tope and ldquohrdquo stands for the heavy isotope (Richter et al2003) Using basalt-rhyolite melt couples Richter et al(2003) experimentally determined that βLi asymp 0215 Thismeans that 6Li can diffuse substantially faster than 7Lithrough the melt away from a growing crystal so that apreferential enrichment of 7Li can potentially occur in thechemical boundary layer ahead of the crystal (Fig 6) re-sulting in elevated δ7Li The diffusion rate of Li in silicatemelts is orders of magnitude faster than the rates of othermajor and minor cations (Richter et al 2003) Under con-ditions of slow mineral growth homogeneous Li isotoperatios would be expected in the tourmalines Instead theobserved Li isotope heterogeneity suggests crystallizationconditions under which the diffusion of Li in the melt didnot keep-up with the rate of tourmaline growth

Elevated δ7Li values of pegmatites

The overall elevated δ7Li values seen in the San Diegopegmatites are similar to the values in the Tin Mountainpegmatite in the Black Hills δ7Li values in the associ-ated Harney Peak leucogranite and the host schists of theTin Mountain pegmatite have δ7Li values within a fewpermil of 0 which points to some process that leads tostrong Li isotope fractionation during generation of LCT-type pegmatite melts (Teng et al 2006b) The apparentlarge fractionation for these relatively low-temperature ig-neous systems contrasts with the minimal fractionation in

914 JS Maloney et al

high-temperature igneous systems (Tomascak 2004) Tenget al (2006b) suggested that elevated δ7Li values in peg-matite melts could potentially be acquired by crystal-liquidfractionation during crystallization of parental magmasHowever for this process to be effective a greater amountof Li would have to remain in the crystallized assemblage(parental granite) than in the residual liquid (pegmatite)This is inconsistent with for example the relative concen-trations of Li in the Harney Peak leucogranite and its poten-tially residual liquid now represented by the Tin Mountainpegmatite Li concentrations in the Harney Peak leucogran-ite range from 10ndash205 ppm (Teng et al 2006b)

A possible alternative explanation is that the structureof a hydrous pegmatite melt has more similarity to waterthan to minerals and therefore it incorporates 7Li pref-erentially over minerals in a fashion analogous to thelarge 7Li enrichment in crustal fluids compared to crustalrocks (Tomascak 2004) As hydrated pegmatite melts passthrough the crust from their sources they may acquire el-evated δ7Li values through rapid Li isotope exchange withthe surrounding rocks Matthews et al (2003) found thatthe oxygen and especially hydrogen isotopic compositionof pegmatite dikes on Naxos Greece reflects the isotopiccomposition of the host rocks which change in isotopiccomposition along strike of the dikes If hydrogen isotopescan be nearly fully exchanged between pegmatite melts andtheir host rocks then it is likely that Li isotopes can also bereadily exchanged given the fast diffusion rates of Li in sil-icate liquids (Richter et al 2003)

Acknowledgements We are grateful for access providedby Dana and Ken Gochenour to the Cryo-Genie propertyBill Calhoun to the San Diego mine property and LouisSpaulding Jr to the Little Three property Jeffrey Patter-son and Matt Taylor provided guidance around the peg-matite districts and Jim Student helped with sample col-lection Carol Nabelek oversaw the ICP-OES analysis BillMcDonough graciously gave access to JM to conduct anal-ysis in the isotope laboratory at the University of MarylandThe paper benefited from the constructive reviews of AxelLiebscher Jeffrey Ryan and Ed Grew and additional com-ments from Roberta Rudnick A Feodor-Lynen fellowshipto Halama by the Alexander von Humboldt Foundation isgratefully acknowledged The study was supported by Uni-versity of Missouri Research Board Grant D3508 and NSFGrant 408564 to Nabelek Additional funding came fromNSF Grant EAR 0606989 to Rudnick and McDonough

References

Anderko A amp Pitzer KS (1993) Equation-of-state representa-tion of phase equilibria and volumetric properties of the systemNaCl-H2O above 573 K Geochim Cosmochim Acta 57 1657-1680

Bouman C Elliott T Vroon PZ (2004) Lithium inputs to sub-duction zones Chem Geol 212 59-79

Cernyacute P (1991) Rare-element granitic pegmatites Part I Anatomyand internal evolution of pegmatite deposits GeoscienceCanada 18 49-67

Cernyacute P amp Ercit TS (2005) The classification of granitic peg-matites revisited Can Mineral 43 2005-2026

Dingwell DB Hess K-U Knoche R (1996) Granite andgranitic pegmatite melts volumes and viscosities Trans R SocEdinburgh Earth Sci 87 65-72

Fenn PM (1977) The nucleation and growth of alkali feldsparsfrom hydrous melts Can Mineral 15 135-161

Fisher J (2002) Gem and rare-element pegmatites of southernCalifornia Mineral Record 33 363-407

Fisher J Foord EE Bricker GA (1999) The geology miner-alogy and history of the Himalaya Mine Mesa Grande SanDiego County California California Geol 3-17

Foord EE (1976) Mineralogy and petrogenesis of layeredpegmatite-aplite dikes in the Mesa Grande District San DiegoCounty California PhD Dissertation Stanford University

Foord EE Starkey HC Taggard JE Jr (1986) Mineralogyand paragenesis of ldquopocketrdquo clays and associated minerals incomplex granitic pegmatites San Diego County California AmMineral 71 428-439

Foustoukos DI James RH Berndt ME Seyfried WE Jr(2004) Lithium isotopic systematics of hydrothermal vent flu-ids at the Main Endeavour Field Northern Juan de Fuca RidgeChem Geol 212 17-26

Giordano D Romano C Dingwell DB Poe B Behrens H(2004) The combined effects of water and fluorine on theviscosity of silicic magmas Geochim Cosmochim Acta 685159-5168

Halama R McDonough WF Rudnick RL Keller J KlaudiusJ (2007) The Li isotopic composition of Oldoinyo LengaiNature of the mantle sources and lack of isotopic fractionationduring carbonatite petrogenesis Earth Planet Sci Lett 25477-89

Holtz F Behrens H Dingwell DB Johannes W (1995) Watersolubility in haplogranite melts Compositional pressure andtemperature dependence Am Mineral 80 94-108

Jahns R H (1979) Gem-bearing pegmatites in San Diego CountyCalifornia The Stewart mine Pala district and the Himalayamine Mesa Grande district in ldquoMesozoic crystalline rocksPeninsular Ranges batholith and pegmatites Point Sol ophio-literdquo P L Abbott amp V R Todd eds San Diego State UniversitySan Diego California 3-38

Jahns RH amp Burnham CW (1969) Experimental studies of peg-matite genesis I A model for the derivation and crystallizationof granitic pegmatites Econ Geol 64 843-864

Jahns RH amp Tuttle OF (1963) Layered pegmatite-aplite intru-sives Mineral Soc Am Sp Pap 1 78-92

Jolliff BL Papike JJ Shearer CK Laul JC (1986)Tourmaline as a recorder of pegmatite evolution Bob Ingersolpegmatite Black Hills South Dakota Am Mineral 71 472-500

Kalt A Schreyer W Ludwig T Prowatke S Bernhardt H ErtlA (2001) Complete solid solution between magnesian schorland lithian excess-boron olenite in a pegmatite from the Koralpe(eastern Alps Austria) Eur J Mineral 13 1191-1205

Kampf AR Gochenour K Clanin J (2003) Tourmaline dis-covery at the Cryo-Genie mine San Diego County CaliforniaRocks and Minerals 78 156-163

Liebscher A Meixner A Romer RL Heinrich W (2007)Experimental calibration of the vapourndashliquid phase relationsand lithium isotope fractionation in the system H2OndashLiCl at400 Geofluids 7 1-7

914 JS Maloney et al

high-temperature igneous systems (Tomascak 2004) Tenget al (2006b) suggested that elevated δ7Li values in peg-matite melts could potentially be acquired by crystal-liquidfractionation during crystallization of parental magmasHowever for this process to be effective a greater amountof Li would have to remain in the crystallized assemblage(parental granite) than in the residual liquid (pegmatite)This is inconsistent with for example the relative concen-trations of Li in the Harney Peak leucogranite and its poten-tially residual liquid now represented by the Tin Mountainpegmatite Li concentrations in the Harney Peak leucogran-ite range from 10ndash205 ppm (Teng et al 2006b)

A possible alternative explanation is that the structureof a hydrous pegmatite melt has more similarity to waterthan to minerals and therefore it incorporates 7Li pref-erentially over minerals in a fashion analogous to thelarge 7Li enrichment in crustal fluids compared to crustalrocks (Tomascak 2004) As hydrated pegmatite melts passthrough the crust from their sources they may acquire el-evated δ7Li values through rapid Li isotope exchange withthe surrounding rocks Matthews et al (2003) found thatthe oxygen and especially hydrogen isotopic compositionof pegmatite dikes on Naxos Greece reflects the isotopiccomposition of the host rocks which change in isotopiccomposition along strike of the dikes If hydrogen isotopescan be nearly fully exchanged between pegmatite melts andtheir host rocks then it is likely that Li isotopes can also bereadily exchanged given the fast diffusion rates of Li in sil-icate liquids (Richter et al 2003)

Acknowledgements We are grateful for access providedby Dana and Ken Gochenour to the Cryo-Genie propertyBill Calhoun to the San Diego mine property and LouisSpaulding Jr to the Little Three property Jeffrey Patter-son and Matt Taylor provided guidance around the peg-matite districts and Jim Student helped with sample col-lection Carol Nabelek oversaw the ICP-OES analysis BillMcDonough graciously gave access to JM to conduct anal-ysis in the isotope laboratory at the University of MarylandThe paper benefited from the constructive reviews of AxelLiebscher Jeffrey Ryan and Ed Grew and additional com-ments from Roberta Rudnick A Feodor-Lynen fellowshipto Halama by the Alexander von Humboldt Foundation isgratefully acknowledged The study was supported by Uni-versity of Missouri Research Board Grant D3508 and NSFGrant 408564 to Nabelek Additional funding came fromNSF Grant EAR 0606989 to Rudnick and McDonough

References

Anderko A amp Pitzer KS (1993) Equation-of-state representa-tion of phase equilibria and volumetric properties of the systemNaCl-H2O above 573 K Geochim Cosmochim Acta 57 1657-1680

Bouman C Elliott T Vroon PZ (2004) Lithium inputs to sub-duction zones Chem Geol 212 59-79

Cernyacute P (1991) Rare-element granitic pegmatites Part I Anatomyand internal evolution of pegmatite deposits GeoscienceCanada 18 49-67

Cernyacute P amp Ercit TS (2005) The classification of granitic peg-matites revisited Can Mineral 43 2005-2026

Dingwell DB Hess K-U Knoche R (1996) Granite andgranitic pegmatite melts volumes and viscosities Trans R SocEdinburgh Earth Sci 87 65-72

Fenn PM (1977) The nucleation and growth of alkali feldsparsfrom hydrous melts Can Mineral 15 135-161

Fisher J (2002) Gem and rare-element pegmatites of southernCalifornia Mineral Record 33 363-407

Fisher J Foord EE Bricker GA (1999) The geology miner-alogy and history of the Himalaya Mine Mesa Grande SanDiego County California California Geol 3-17

Foord EE (1976) Mineralogy and petrogenesis of layeredpegmatite-aplite dikes in the Mesa Grande District San DiegoCounty California PhD Dissertation Stanford University

Foord EE Starkey HC Taggard JE Jr (1986) Mineralogyand paragenesis of ldquopocketrdquo clays and associated minerals incomplex granitic pegmatites San Diego County California AmMineral 71 428-439

Foustoukos DI James RH Berndt ME Seyfried WE Jr(2004) Lithium isotopic systematics of hydrothermal vent flu-ids at the Main Endeavour Field Northern Juan de Fuca RidgeChem Geol 212 17-26

Giordano D Romano C Dingwell DB Poe B Behrens H(2004) The combined effects of water and fluorine on theviscosity of silicic magmas Geochim Cosmochim Acta 685159-5168

Halama R McDonough WF Rudnick RL Keller J KlaudiusJ (2007) The Li isotopic composition of Oldoinyo LengaiNature of the mantle sources and lack of isotopic fractionationduring carbonatite petrogenesis Earth Planet Sci Lett 25477-89

Holtz F Behrens H Dingwell DB Johannes W (1995) Watersolubility in haplogranite melts Compositional pressure andtemperature dependence Am Mineral 80 94-108

Jahns R H (1979) Gem-bearing pegmatites in San Diego CountyCalifornia The Stewart mine Pala district and the Himalayamine Mesa Grande district in ldquoMesozoic crystalline rocksPeninsular Ranges batholith and pegmatites Point Sol ophio-literdquo P L Abbott amp V R Todd eds San Diego State UniversitySan Diego California 3-38

Jahns RH amp Burnham CW (1969) Experimental studies of peg-matite genesis I A model for the derivation and crystallizationof granitic pegmatites Econ Geol 64 843-864

Jahns RH amp Tuttle OF (1963) Layered pegmatite-aplite intru-sives Mineral Soc Am Sp Pap 1 78-92

Jolliff BL Papike JJ Shearer CK Laul JC (1986)Tourmaline as a recorder of pegmatite evolution Bob Ingersolpegmatite Black Hills South Dakota Am Mineral 71 472-500

Kalt A Schreyer W Ludwig T Prowatke S Bernhardt H ErtlA (2001) Complete solid solution between magnesian schorland lithian excess-boron olenite in a pegmatite from the Koralpe(eastern Alps Austria) Eur J Mineral 13 1191-1205

Kampf AR Gochenour K Clanin J (2003) Tourmaline dis-covery at the Cryo-Genie mine San Diego County CaliforniaRocks and Minerals 78 156-163

Liebscher A Meixner A Romer RL Heinrich W (2007)Experimental calibration of the vapourndashliquid phase relationsand lithium isotope fractionation in the system H2OndashLiCl at400 Geofluids 7 1-7

Lithium and its isotopes in pegmatite tourmalines 915

London D (1986a) Magmatic-hydrothermal transition in the Tancorare-element pegmatite Evidence from fluid inclusions andphase-equilibrium experiments Am Mineral 71 376-395

ndash (1986b) Formation of tourmaline-rich gem pockets in miaroliticpegmatites Am Mineral 71 396-405

ndash (1992) The application of experimental petrology to the gene-sis and crystallization of granitic pegmatites Can Mineral 30499-540

London D Hervig RL Morgan GB VI (1988) Melt-vaporsolubilities and elemental partitioning in peraluminous granite-pegmatite systems experimental results with Macusani glass at200 MPa Contrib Mineral Petrol 99 360-373

Lynton SJ Walker RJ Candela PA (2005) Lithium isotopesin the system Qz-Ms-fluid An experimental study GeochimCosmochim Acta 69 3337-3347

Magna T Wiechert U Grove TL Halliday AN (2006)Lithium isotope fractionation in the southern Cascadia subduc-tion zone Earth Planet Sci Lett 250 428-443

Matthews A Putlitz B Hamiel Y Hervig RL (2003) Volatiletransport during the crystallization of anatectic melts oxygenboron and hydrogen stable isotope study on the metamorphiccomplex of Naxos Greece Geochim Cosmochim Acta 673145-3163

Morgan GB VI amp London D (1999) Crystallization of theLittle Three layered pegmatite-aplite dike Ramona DistrictCalifornia Contrib Mineral Petrol 136 310-330

Moriguti T Nakamura E (1998) High-yield lithium separationand the precise isotopic analysis for natural rock and aqueoussamples Chem Geol 145 91-104

Mysen BO and Richet P (2005) Silicate glasses and melts prop-erties and structure Elsevier Amsterdam 544 p

Nabelek PI (2007) A kinetic model for crystallization of graniticpegmatites at very low temperatures 6th Hutton Symposium150-151

Norton JJ (1994) Structure and bulk composition of the TinMountain Pegmatite Black Hills South Dakota Econ Geol89 1167-1175

Qi HP Taylor PDP Berglund M De Bievre P (1997)Calibrated measurements of the isotopic composition andatomic weight of the natural Li isotopic reference materialIRMM-016 Int J Mass Spectrom Ion Process 171 263-268

Richter FM Davis AM Depaolo DJ Watson EB (2003)Isotope fractionation by chemical diffusion between moltenbasalt and rhyolite Geochim Cosmochim Acta 67 3905-3923

Rockhold JR Nabelek PI Glascock MD (1987) Origin ofrhythmic layering in the Calamity Peak satellite pluton ofthe Harney Peak Granite South Dakota The role of boronGeochim Cosmochim Acta 51 487-496

Romano C Poe B Mincione V Hess KU Dingwell DB(2001) The viscosities of dry and hydrous XAlSi3O8 (X = LiNa K Ca05 Mg05) melts Chem Geol 174 115-132

Rudnick RL Tomascak PB Njo HB Gardner LR (2004)Extreme lithium isotopic fractionation during continentalweathering revealed in saprolites from South Carolina ChemGeol 212 45-57

Schreyer W Wodara U Marler B van Aken PA Seifert FRobert J-L (2000) Synthetic tourmaline (olenite) with excess

boron replacing silicon in the tetrahedral site I Synthesis con-ditions chemical and spectroscopic evidence Eur J Mineral12 529-541

Sirbescu MC amp Nabelek PI (2003a) Crystallization condi-tions and evolution of magmatic fluids in the Harney PeakGranite and associated pegmatites Black Hills South Dakotandash Evidence from fluid inclusions Geochim Cosmochim Acta67 2443-2465

ndashndash (2003b) Crustal melts below 400 C Geology 31 685-688

Sirbescu MC Hartwick EE Student JJ (2008) Rapid crys-tallization of the Animikie Red Ace Pegmatite FlorenceCounty Northeastern Wisconsin Inclusion microthermometryand conductive-cooling modeling Contrib Mineral Petrol inpress

Stern LA Brown GE Bird DK Jahns RH Foord EEShigley JE Spaulding LB Jr (1986) Mineralogy and geo-chemical evolution of the Little Three pegmatite-aplite layeredintrusive Ramona California Am Mineral 71 406-427

Stolper E (1982) The speciation of water in silicate meltsGeochim Cosmochim Acta 46 2609-2620

Symons DTA Walawender MJ Smith TE Molnar SEHarris MJ Blackburn WH (2003) Palomagnetism and geo-barometry of the La Posta pluton California in Geol SocAm Spec Pap 374 Tectonic Evolution of Northwestern Mexicoand the Southwestern USA SE Johnson SR Paterson JMFletcher DL Kimbrough A Martin-Barajas eds 93-116

Teng F McDonough WF Rudnick RL Dalpeacute C TomascakPB Chappell BW Gao S (2004) Lithium isotopic composi-tion and concentration of the upper continental crust GeochimCosmochim Acta 68 4167-4178

Teng F McDonough WF Rudnick RL Walker RJ (2006a)Diffusion-driven extreme lithium isotopic fractionation in coun-try rocks of the Tin Mountain pegmatite Earth Planet Sci Lett243 701-710

Teng F McDonough WF Rudnick RL Walker RJ SirbescuMC (2006b) Lithium isotopic systematics of granites and peg-matites from the Black Hills South Dakota Am Mineral 911488-1498

Thomas AV Bray CJ Spooner ETC (1988) A discussion ofthe Jahns-Burnham proposal for the formation of zoned graniticpegmatites using solid-liquid-vapour inclusions from the TancoPegmatite SE Manitoba Canada Trans R Soc EdinburghEarth Sci 7 299-315

Thomas R amp Klemm W (1997) Microthermometric study of sil-icate melt inclusions in Variscan granites from SE GermanyVolatile contents and entrapment conditions J Petrol 381753-1765

Thomas R Webster JD Heinrich W (2000) Melt inclusions inpegmatite quartz complete miscibility between silicate meltsand hydrous fluids at low pressure Contrib Mineral Petrol139 394-401

Todd VR Shaw SE Hammarstrom JM (2003) Cretaceous plu-tons of the Peninsular Ranges batholith San Diego and west-ernmost Imperial Counties California Intrusion across a LateJurassic continental margin Geol Soc Am Spec Pap 374185-235

Tomascak PB (2004) Developments in the understanding and ap-plication of lithium isotopes in the Earth and planetary sciencesRev Mineral Geochem 55 153-195

916 JS Maloney et al

Tomascak PB Tera F Helz RT Walker RJ (1999) The ab-sence of lithium isotope fractionation during basalt differen-tiation new measurements by multicollector sector ICP-MSGeochim Cosmochim Acta 63 907-910

Walawender MJ Gastil RG Clinkenbeard JP McCormickWV Eastman BG Wernicke RS Wardlaw MS GunnSH Smith BM (1990) Origin and evolution of the zoned LaPosta-type plutons eastern Peninsular Ranges batholith south-ern and Baja California in ldquoThe nature and origin of Cordilleranmagmatismrdquo J L Anderson ed Boulder Colorado 1-18

Walker RJ Hanson GN Papike JJ Orsquoneil JR Laul JC(1986) Internal evolution of the Tin Mountain pegmatite BlackHills South Dakota Am Mineral 71 440-459

Watson EB (1994) Diffusion in volatile-bearing magmas RevMineral 30 371-411

Webber KL Simmons WB Falster AU Foord EE (1999)Cooling rates and crystallization dynamics of shallow levelpegmatite-aplite dikes San Diego County Califronia AmMineral 84 708-717

Webster JD Holloway JR Hervig RL (1989) Partitioning oflithophile trace elements between H2O and H2O + CO2 fluidsand topaz rhyolites Econ Geol 84 116-134

Wenger M amp Armbruster T (1991) Crystal-chemistry of lithiumndashoxygen coordination and bonding Eur J Mineral 3 387-399

Whittington A Richet P Behrens H Holtz F Scaillet B(2004) Experimental temperature-X(H2O)-viscosity relation-ship for leucogranites and comparison with synthetic silicic liq-uids Trans R Soc Edinburgh Earth Sci 95 59-71

Wunder B Meixner A Romer RL Heinrich W (2006)Temperature-dependent isotopic fractionation of lithium be-tween clinopyroxene and high-pressure hydrous fluids ContribMineral Petrol 51 112-120

Wunder B Meixner A Romer RL Feenstra A Shettler GHeinrich W (2007) Lithium isotope fractionation between Li-bearing staurolite Li-mica and aqueous fluids An experimentalstudy Chem Geol 238 277-290

Received 12 November 2007Modified version received 28 February 2008Accepted 29 February 2008