A geochemical and Sr-Nd-O isotopic study of theProterozoic Eriksfjord Basalts Gardar Province SouthGreenland Reconstruction of an OIB signature in crustallycontaminated rift-related basalts

R HALAMA1 T WENZEL

1 B G J UPTON2 W SIEBEL

1AND G MARKL

11 Universita t Tubingen Institut fur Geowissenschaften Wilhelmstr 56 D-72074 Tubingen Germany2 Department of Geology and Geophysics The University of Edinburgh West Mains Road Edinburgh EH9 3JW UK

ABSTRACT

Basalts from the volcano-sedimentary Eriksfjord Formation (Gardar Province South Greenland) wereerupted at around 12 Ga into rift-related graben structures The basalts have compositions transitionalbetween tholeiite and alkaline basalt with MgO contents lt7 wt and they display LREE-enrichmentrelative to a chondritic source Most of the trace element and REE characteristics are similar to those ofbasalts derived from OIB-like mantle sources Initial 87Sr86Sr ratios of clinopyroxene separates rangefrom 070278 to 070383 and initial eNd values vary from ndash32 to +21 The most unradiogenic samplesoverlap with the field defined by carbonatites of similar age and can be explained by mixing ofisotopically depleted and enriched mantle components Using AFC modelling equations the Sr-Ndisotope data of the more radiogenic basalts can successfully be modelled by addition of lt5 lowercrustal granulite-facies gneisses as contaminants d18Ov-smow values of separated clinopyroxene rangefrom +52 to +60 and fall within the range of typical mantle-derived rocks However up to 10mixing with an average lower crustal component are permitted by the data

KEY WORDS geochemistry Sr-Nd-O isotopes Gardar Province basalt Greenland

Introduction

STUDY of rift-related basaltic magmatism isessential for an improvement in our understandingof the chemical composition of the sub-conti-nental mantle and also the mechanisms respon-sible for the generation of ood basalts alkalinerocks and carbonatites It has been shown thatseveral distinct mantle components can contributeto magmatism in rift-related environments (egPerry et al 1987 Paces and Bell 1989 Heamanand Machado 1992) but that the relativeimportance of these sources is highly variable inboth space and time (Shirey et al 1994

Nicholson et al 1997) For example manyoccurrences of recent rift-related magmatismhave a mantle-plume association (eg Baker etal 1997 Bell and Tilton 2001) Studies ofProterozoic continental volcanic rocks provideinsight into the composition of the mantle and theinterplay between mantle and crustal magmasources at that time (eg Lightfoot et al 1991Shirey et al 1994) In most models three mainmantle source components have been invoked toexplain the geochemical signatures of the eruptedrocks (1) Depleted MORB mantle (DMM) iemantle sources of mid-ocean ridge basalts(MORBs) that are depleted in all incompatibleelements (2) OIB-like mantle ie mantle knownfrom ocean island basalts (OIBs) that is inoceanic regions restricted to islands that are notrelated to subduction (Hofmann 1997) This typeof mantle is less depleted or even enriched in

E-mail markluni-tuebingendeDOI 1011800026461036750147

Mineralogical Magazine October 2003 Vol 67(5) pp 831ndash853

2003 The Mineralogical Society

incompatible elements compared to the primitivemantle There are several types of OIB compo-nents well de ned by Sr-Nd-Pb isotopic studies(Zindler and Hart 1986 Hart et al 1992) Theseare HIMU (lsquohigh mrsquo m 238U206Pb) enrichedmantle 1 (EM-1) and enriched mantle 2 (EM-2)(3) Subcontinental lithospheric mantle (SCLM)representing the mantle below the continentalcrust The SCLM is usually enriched inincompatible elements but it can be veryheterogeneous

Interaction of an OIB-like source with litho-spheric mantle (Ormerod et al 1988 Paslick etal 1995 Molzahn et al 1996) or DMM (egNicholson et al 1997) may have played a majorrole in magma genesis of continental volcanicrocks Lavas chemically similar to OIBs havebeen observed in many continental ood basaltprovinces eg Deccan Traps (Mahoney 1988)British Tertiary Volcanic Province (Fitton et al1997) and Parana-Etendeka (Gibson et al 1997)These are believed to represent mantle-derivedmelts from plumes not contaminated by litho-spheric material but they represent only a smallproportion of the total magma volume erupted(Gibson et al 1999) Therefore the SCLM hasbeen invoked in various models either as a sourceof large-scale melts (eg Turner et al 1996) or asassimilant in plume-derived melts (eg Gibson etal 1995) The SCLM appears to be especiallyimportant in the genesis of small-degree melts(Gibson et al 1999) Post-Archaean lithosphericmantle is considered to have been suf cientlyfertile to have contributed to the genesis of oodbasalts (Hawkesworth et al 1990)

The Gardar Province in South Greenland is arift-related igneous province with magmatismlasting from about 135 to 114 Ga (Emeleusand Upton 1976 Upton and Emeleus 1987Upton et al 2003) It comprises about 12 mainigneous complexes numerous dykes of variablechemical composition and a sequence of rift-related interlayered lavas and sediments(Eriksfjord Formation) The basaltic lavas of theEriksfjord Formation are among the geochemi-cally most primitive rocks of the ProvinceHowever except for detailed palaeomagneticinvestigations (Thomas and Piper 1992 Piper etal 1999) they have been studied relatively little(Larsen 1977 Paslick et al 1993) Due to theircomparatively primitive nature the basaltic rocksof the Eriksfjord Formation provide a goodopportunity to discern the geochemical signaturesof the mantle sources of the magmatic activity of

the Province Similar sources may have beeninvolved in the petrogenesis of the Gardaranorthosites alkali gabbros and nephelinesyenites In this study we present new geochem-ical data (XRF analyses of major and traceelements Sr Nd and O isotopic analyses REEanalyses of whole-rock samples and in situ REEanalyses of clinopyroxene) to constrain mantlesources and to determine the role of crustalcontamination in the petrogenesis of theEriksfjord Formation basalts In particular high-precision laser- uorination oxygen isotope datahave only recently become an important tool forevaluating the contribution of mantle andcontinental crust in suites of continental oodbasalts (Baker et al 2000) Since continentalintraplate basalts are the only major class ofterrestrial basalts that have not yet been exploredsystematically using these methods (Eiler 2001)the laser uorination oxygen isotope data ofclinopyroxene separates presented herein are animportant contribution to the knowledge ofoxygen isotopic compositions of continentalbasalts in general

Geological setting

The Eriksfjord Formation (EF) represents an~3500 m thick sequence of Precambrian supra-crustal rocks comprising continental sandstonesand conglomerates and eruptive rocks and sillsthat range from basalt through trachyte to alkalinephonolite and carbonatite (Larsen 1977) Therocks lie unconformably on the KetilidianJulianehaEcirc b batholith dated at 185 to 179 Ga(van Breemen et al 1974 Garde et al 2002) andare cut by the 116 Ga Ilotildeacutemaussaq intrusion(Waight et al 2002) The type-section of theEriksfjord Formation occurs on the Ilotildeacutemaussaqpeninsula and has been subdivided into sixstratigraphic groups (Thomas and Piper 1992)originally de ned as members (Poulsen 1964)(Fig 1) The three volcanic groups are known asfrom oldest to youngest Mussartut Ulukasik andIlotildeacutemaussaq The Mussartut Group is subdividedinto nine sub-units and the Il otilde maussaq Group issubdivided into three sub-units called SermilikTunugdliar k and Narssaq Fjeld (Fig 1) Asandstone unit precedes each volcanic group(Thomas and Piper 1992) The Mussartut Groupcontains the Qassiarsuk carbonatite complexdated at ~12 Ga by Rb-Sr and Pb-Pb isochrons(Andersen 1997) Paslick et al (1993) obtainedtwo Sm-Nd isochron ages of 117plusmn003 and

832

R HALAMA ETAL

120plusmn003 Ga for two samples of the UlukasikGroup However palaeomagnetic data indicate anage of 135 ndash131 Ga for the entire succession(Piper et al 1999) and the exact age of the EF isas yet unresolved (Upton et al 2003) Parts of theEriksfjord lavas have been subjected to extensivemetasomatism related to the intrusion of alkalinemagmas (Ranlov and Dymek 1991 Coulson andChambers 1996) The local presence of epidoteimplies thermal metamorphism at temperatures ofup to 300 ndash350ordmC (Piper et al 1999) Higheruption rates are inferred for the volcanic systemand the whole lifespan is considered to belt~5 Ma (Piper et al 1999)

Analytical procedures

Electron microprobe analysisMineral compositions were determined using aJEOL 8900 electron microprobe (EMP) at theInstitut fur Geowissenschaften UniversitatTubingen Germany An internal jrz correctionof the raw data was applied (Armstrong 1991)Both natural and synthetic standards were usedfor major and minor elements Measuring times

were 16 s and 30 s on the peak positions formajor and minor elements respectively Theemission current was 15 nA and the accelerationvoltage 15 kV For feldspar analyses a beamdiameter of 5 mm was used to avoid errorsresulting from diffusion of Na

XRF analysis

Whole-rock analyses were performed by standardX-ray uorescence (XRF) techniques at theInstitut fur Geowissenschaften at the JohannesGutenberg-Universitat Mainz Germany using aPhilips PW 1404 spectrometer and at theUniversitat Freiburg using a Philips PW 2404spectrometer Prior to analysis whole-rocksamples were crushed and milled in tungstencarbide mills The powders obtained were dried at105ordmC Loss on ignition (LOI) determinationswere carried out on 1 g of pulverized samplematerial which was heated for 2 h at 1000ordmCPressed powder pellets and fused glass discs wereprepared to measure contents of trace and majorelements respectively Natural standards wereused for calibration Detection limits vary

FIG 1 Map of the Eriksfjord Formation on the Narssaq peninsula with sample locations (after Poulsen 1964 andLarsen 1977) Sample 101472 (not shown) is from a succession ~2 km south of the settlement of Narsarsuaq further

to the NE

PETROGENESIS OF THE ERIKSFJORD BASALTS

833

between 1 and 9 ppm depending on the speci ctrace element and on the instrument used

Whole-rock REE analyses

Whole-rock REE analyses were carried out by theNatural Environment Research Council (NERC)inductively coupled plasma-atomic emissionspectrometry (ICP-AES) facility at RoyalHolloway University of London Solutions wereprepared using a combined HF dissolution andalkali fusion as described in Walsh et al (1981)Resultant solutions were analysed for 12 REEsand selected potential interferences using a PerkinElmer Optima 3300RL ICP-AES

Laser ICP-MS in situ REE measurement of clinopyroxene

In situ laser ablation inductively coupled plasma-mass spectrometer (LA-ICP-MS) REE analyses on~150 mm thick polished thin sections wereperformed at the EU Large-Scale GeochemicalFacility (University of Bristol) using a VGElemental PlasmaQuad 3 + S-Option ICP-MSequipped with a 266 nm Nd-YAG laser (VGMicroProbe II) The laser beam diameter at thesample surface was ~20 mm All measurementswere made using Thermo Elemental PlasmaLablsquotime-resolved analysisrsquo (TRA) data acquisitionsoftware with a total acquisition time of 100 s peranalysis allowing ~40 s for background followedby 50 s for laser ablation NIST 610 glass was usedfor instrument calibration and NIST 612 was usedas a secondary standard Ca was used as an internalstandard to correct the ablation yield differencesbetween and during individual analyses on bothstandards and samples To avoid analyticaluncertaintiesdue to variations in the concentrationsof the internal standard Ca concentrations werequantitatively measured within 20 mm of the laserablation pits using the EMP at the Institut furGeowissenschaften Universitat Tubingen Theprecision of trace-element concentrations basedon repeated analyses of standards is ~plusmn5 forelement concentrations gt10 ppm and plusmn10 forconcentrations lt10 ppm Data processing wascarried out of ine using the same PlasmaLabsoftware used for data collection and variouscustom-designed Excel spreadsheets The limitsof detection are de ned as 328 standard deviationsabove background level which is equal to a 95con dence that the measured signal is signi cantlyabove background Typical detection limits for theREE in this study were 004 ndash06 ppm

Sr and Nd isotope analysisFor Sr and Nd isotope analyses 15 ndash20 mgsamples of hand-picked clinopyroxene werespiked with mixed 84Sr ndash 87Rb and 150Nd ndash 149Smtracers before dissolution under high pressure inHF at 180ordmC in poly-tetrauor-ethylene (PTFE)reaction bombs The Rb and Sr were separated inquartz columns containing a 5 ml resin bed ofAG50W-X12 200ndash400 mesh equilibrated with25 N HCl The Sm and Nd separation wasperformed in quartz columns using 17 mlTe on powder coated with HDEHP (Di-EthylHexyl Phosphate) as the cation exchange mediumequilibrated with 018 N HCl All analyses weremade with a multicollector Finnigan MAT 262thermal ionization mass spectrometer (TIMS) instatic collection mode at the Institut furGeowissenschaften Universitat Tubingen TheSr was loaded with a Ta-HF activator andmeasured on a single W lament The Rb Smand Nd were measured in a double Re- lamentcon guration mode 87Sr86Sr and 143Nd144Ndratios were normalized for mass fractionation to86Sr88Sr = 01194 and 146Nd144Nd = 07219respectively The average 143Nd144Nd ratioobtained for the Ames Nd-standard (GeologicalSurvey of Canada Roddick et al 1992) was0512119plusmn10 (n = 42) and the average 87Sr86Srratio for the NBS 987 Sr-standard was0710261plusmn16 (n = 30) during the course of thisstudy 143Nd144Nd ratios have been cross-checked with the La Jolla Nd-standard whichgave 0511831plusmn30 (n = 12) Total proceduralblanks (chemistry and loading) were lt200 pg forSr and lt100 pg for Nd

Oxygen isotope analysis

The oxygen isotope composition of hand-pickedclinopyroxene separates was measured using amethod similar to that described by Sharp (1990)and Rumble and Hoering (1994) 1 ndash2 mg ofhand-picked clinopyroxenewere loaded onto a Pt-sample holder The sample chamber was pumpedout to a vacuum of ~10 ndash6 mbar and pre- uorinated overnight Samples were then heatedwith a CO2-laser in an atmosphere of 50 mbar ofpure F2 Excess F2 is separated from O2 byconversion to Cl2 using KCl held at 150ordmC Theextracted O2 is collected on a molecular sieve(13X) Oxygen isotopic compositions weremeasured on a Finnigan MAT 252 mass spectro-meter at the Institut fur GeowissenschaftenUniversitat Tubingen The results are reported as

834

R HALAMA ETAL

the per mil deviation from Vienna Standard MeanOcean Water (V-SMOW) in the standard d-notation Replicate oxygen isotope analyses ofthe NBS-28 quartz standard (Valley et al 1995)had an average precision of plusmn01 for d18O Ineach run standards were analysed at thebeginning and the end of the sample set Acorrection was applied to the data equal to theaverage of the difference between the meanmeasured value and the accepted value for thestandard (964)

Oxygen isotope compositions of powderedwhole-rock samples were determined by aconventional method modi ed after Clayton andMayeda (1963) using BrF5 as reagent andconverting the liberated oxygen to CO2 beforemass spectrometric analysis

Petrography

175 samples were collected from all threevolcanic units of the Eriksfjord Formation butmost of them were extensively altered Theycontained secondary chlorite Fe-talc Ca-Fesilicates and carbonate as alteration productsSince the focus of this study was in situ analysesof minerals or analyses of mineral separates onlynine samples with a signi cant proportion of freshclinopyroxene and relatively few alterationproducts were selected for analysis Due to thelimited number of suitable samples we did notattempt to investigate the lavas in terms of theirstratigraphic position In addition to the ninesamples from which pyroxene separates wereobtained REE whole-rock data of six othersamples were added to the data set

In general the Mussartut Group lavas areolivine-free but contain up to 50 chloriticmatrix that probably results from devitri cationof former glass In contrast the majority ofUlukasik and Il otilde maussaq Group lavas containolivine-pseudomorphs and tend to be coarsergrained and holocrystalline Many of themcontain lsquosnow- akersquo glomerocrysts of plagioclaseindicating plagioclase-saturation or super-satura-tion at near-surface levels (Upton 1996) Thesamples from which clinopyroxenes have beenseparated are ne-grained basalts with plagioclaselaths embedded in interstitial clinopyroxeneas themajor and apatite and Fe-Ti oxides as subordinateprimary magmatic mineral phases The texturecan be described as ophitic and is believed toresult from the cotectic crystallization of plagio-clase and clinopyroxene(Philpotts 1990) Olivine

has been present in a few samples but it is nowcompletely altered Pyrite pyrrhotite and chalco-pyrite occur as accessory phases Some samplescontain two oxides (ilmenite + titanomagnetite)whereas others contain a single oxide exsolvedinto ilmenite and titanomagnetite or ilmenite andhematite Texturally they are mostly subhedral toanhedral although few euhedral grains occurHematite is also present around both discrete andexsolved ilmenite and titanomagnetite grains andthe exsolved hematite might be entirely ofsecondary origin formed by oxidation of titano-magnetite Late-stage interstitial quartz occursrarely Sample EF 108 (Ulukasik Group) isremarkable in having an intense reddish colourunlike the other samples that are almost black

Results

Mineral chemistryTypical analyses of clinopyroxenes are given inTable 1 Compositions range from En41Wo44Fs16

to En28Wo45Fs27 and broadly overlap for allgroups (Fig 2) According to the classi cation ofMorimoto et al (1988) clinopyroxenes of theMussartut and Ulukasik Groups are mainlyaugites with few diopsides present whereas inthe Il otilde maussaq Group samples of diopsidiccompositions dominate over augitic onesClinopyroxene analyses from the varioussamples form tight clusters on the pyroxenequadrilateral except for those from sample EF108 that are more scattered With only a singlepyroxene phase present in all samples clino-pyroxene compositions indicate minimum crystal-lization temperatures (Lindsley 1983) of750 ndash1000ordmC (Fig 2) Plagioclase is normallyzoned and its composition typically variesbetween An70Ab29Or1 and An16Ab72Or12 Lateinterstitial albite can reach compositions ofAn0Ab99Or1 There are no signi cant composi-

1200 Ecirc C

800 Ecirc C

25 50 75

25 25

En Fs

HdDi

Mussartucirct Group

Iliacutemaussaq Group

Ulukasik Group

FIG 2 Clinopyroxene analyses of the Eriksfjord basaltsplotted in the pyroxene quadrilateral with isotherms

after Lindsley (1983)

PETROGENESIS OF THE ERIKSFJORD BASALTS

835

tional changes among the various samplesIlmenite from two samples (EF 024 and EF 039)which contain coexisting ilmenite and titanomag-netite includes a signi cant component ofpyrophanite (MnTiO3) with up to 87 wtMnO It s com pos i t i ons va r y be t weenI l m 8 6 H e m 5 P y r 8 a nd I l m 7 6 H e m 5 P yr 1 9 Titanomagnetite analyses were unsatisfactory asall grains show strong alteration They areconsidered unreliable and are not presented here

Whole-rock geochemistry and classif|cation

Concentrations of major and trace elements aresummarized in Table 2 The basalts can beclassi ed according to the TAS diagram (LeMaitre et al 1989) as basalts and trachybasaltswith both alkaline and subalkaline compositions(not shown) Due to possible mobility of the Naand K the samples were also plotted into NbYvs ZrP2O5 and P2O5 vs Zr diagrams (Floyd andWinchester 1975) which can discriminate

between tholeiitic and alkaline series Thesediagrams do not yield consistent discriminationsfor the EF basalts but indicate that the characterof the sample suite is transitional betweentholeiitic and alkaline (not shown)

Mg [100Mg(Mg+Fe2+)] calculated assumingFe2O3FeO = 02 as recommended for basalts byMiddlemost (1989) varies between 54 and 40(Table 2) Ni concentrations range from 46 to115 ppm and Cr contents from 29 to 126 ppmTaking Mg as fractionation index a positivecorrelation (r2 = 061) with Ni contents indicatesfractionation of olivine (Fig 3) The relativelypoor positive correlations of Mg with Cr (r2 =044) and Sc (r2 = 023) argue against a signi cantfractionation of Cr-spinel or clinopyroxene Thisis consistent with the lack of clinopyroxenephenocrysts and the absence of any Cr-spinelThe negative correlations of Mg with Sr TiO2

and P2O5 suggest that fractionation of plagioclaseFe-Ti oxides and apatite was not important in thesample set investigated which is again consistent

TABLE 1 Typical clinopyroxene analyses of Eriksfjord Formation basalts determined by EMP

Sample EF 024 EF 024 EF 039 EF 059 EF 072 EF 108 EF 168 EF 174Cpx no cpx 2 cpx 3 cpx 5 cpx 3 cpx 5 cpx 7 cpx 4 cpx 1

SiO2 5060 5041 5128 4988 5145 4950 5175 5069TiO2 173 176 121 169 140 162 096 121Al2O3 297 300 179 223 171 264 155 213FeO 1160 1068 1218 1405 993 1234 979 1129MnO 028 021 030 040 019 028 025 031MgO 1254 1324 1263 1182 1325 1163 1307 1214CaO 1959 2029 2002 1907 2106 2099 2174 2143Na2O 034 031 029 049 033 055 052 054

Total 9965 9989 9971 9963 9932 9955 9962 9974

Formulae based on 4 cations and 6 oxygensSi 191 189 194 190 194 188 194 191Al 013 013 008 010 008 012 007 009Ti 005 005 003 005 004 005 003 003Fe3+ - - - 003 - 007 003 005Mg 071 074 071 067 075 066 073 068Fe2+ 037 033 039 041 031 032 028 031Mn 001 001 001 001 001 001 001 001Ca 079 082 081 078 085 085 087 087Na 002 002 002 004 002 004 004 004

Total 400 400 400 400 400 400 400 400

Pyroxene projection after Lindsley (1983)Wo 039 041 041 041 043 043 045 044En 040 041 039 037 040 038 040 038Fs 021 018 021 023 017 019 015 017

836

R HALAMA ETAL

TABLE 2 XRF whole-rock analyses of Eriksfjord Formation basalts

Sample EF 024 EF 063 EF 061 EF 039 EF 059 EF 108 EF 072 EF 174 EF 168Group Mussartut Mussartut Mussartut Mussartut Mussartut Ulukasik Ulukasik Ilotildeacutemaussaq IlotildeacutemaussaqSubunit 4 4 5 6 7 Lower Unit Upper Unit Sermilik Narssaq

Fjeld

Major elements (wt)SiO2 4573 4585 4688 4660 4875 4698 4656 4601 4545TiO2 230 200 180 175 191 265 203 299 277Al2O3 1578 1630 1600 1612 1645 1682 1668 1645 1719Fe2O3 1510 1356 1331 1383 1225 1394 1321 1438 1392MnO 028 018 018 018 017 014 017 018 017MgO 525 683 653 673 568 396 647 440 480CaO 842 849 872 857 920 641 808 746 737Na2O 349 274 289 276 236 448 303 323 376K2O 087 083 068 032 027 159 070 205 132P2O5 041 035 023 023 023 048 035 111 083LOI 199 210 240 238 242 204 210 052 156

Total 9962 9923 9962 9947 9969 9949 9938 9878 9914

Mg 453 545 538 536 524 403 538 421 450

Trace elements (ppm)Sc 24 22 25 22 26 19 20 21 16V 209 175 215 195 211 142 170 164 203Cr 81 108 66 65 126 48 82 33 29Co 63 78 70 62 71 44 68 35 46Ni 99 115 85 85 101 58 106 46 51Cu 65 49 46 60 97 41 59 38 41Zn 103 102 110 97 110 93 82 122 105Ga 20 18 21 19 20 19 16 21 19Rb 24 26 16 lt 10 lt 10 38 12 39 45Sr 504 503 512 392 373 712 507 850 992Y 31 26 28 23 28 31 17 31 25Zr 188 165 121 116 125 125 88 164 139Nb 12 11 10 lt 10 10 9 lt 10 31 27Ba 445 375 415 302 178 582 405 1293 900Pb 6 7 3 4 3 4 4 3 5La ndash ndash 15 ndash 12 11 ndash 37 28Ce ndash ndash 40 ndash 38 39 ndash 92 71Pr ndash ndash 1 ndash 1 2 ndash 9 4Nd ndash ndash 16 ndash 15 18 ndash 43 34Sm ndash ndash 6 ndash 7 6 ndash 5 5

CIPW normq ndash ndash ndash ndash 313 ndash ndash ndash ndashor 514 491 402 189 160 940 414 1212 780ab 2807 2319 2445 2335 1997 3207 2564 2733 2981an 2482 2973 2868 3065 3350 2109 2985 2433 2613ne 079 ndash ndash ndash ndash 316 ndash ndash 109di 1187 830 1076 853 877 633 654 452 416hy ndash 890 1059 1650 2217 ndash 1097 315 ndashol 1670 1309 1044 780 ndash 1472 1120 1388 1689mt 365 328 322 335 296 336 319 348 336il 437 380 342 332 363 503 386 568 526ap 095 081 053 053 053 111 081 257 192

Mg was calculated as Mg = 100Mg2+(Mg2++Fe2+) using Fe2O3FeO = 02 Normative mineral compositionswere determined according to the CIPW calculation scheme

PETROGENESIS OF THE ERIKSFJORD BASALTS

837

1 5

2 0

2 5

3 0

3 5

40

60

80

100

120

0

0 5

1

1 5

200

400

600

800

1000

0

50

100

150

15

20

25

30

4 0 45 50 5 5

r2 = 023

r2 = 044r2 = 061

r2 = 053

r2 = 082 r2 = 052

Ni (ppm)

Sc (ppm)

P2O5 (wt)

Sr (ppm)

TiO2 (wt)

Cr (ppm)

Mg Mg

Mg Mg

Mg

40 4 5 50 55

Mg

40 45 50 55

40 45 50 55

40 45 50 5540 4 5 50 55

FIG 3 Whole-rock variation diagrams for EF basalts with Mg as fractionation index Mg is calculated asMg(Mg+Fe2+) with Fe2O3FeO = 02 (Middlemost 1989)

838

R HALAMA ETAL

with the mostly subhedral Fe-Ti oxides and theophitic texture Ratios of incompatible traceelements such as ZrY (range 40 ndash63) andTiY (range 385 ndash716) are relatively uniformthroughout all samples but some scatter isevident Notable geochemical features are highSr contents up to 1000 ppm and very high Bacontents up to 1300 ppm

Normative mineral compositions were deter-mined according to the CIPW calculation scheme(Cross et al 1903 Cox et al 1979) Mostsamples are hypersthene-normative some arenepheline-normative and one (EF 059) is quartz-normative (Table 2) Despite the uncertaintiesinvolved in the CIPW norm calculation the dataindicate that some of the basalts crystallized fromsilica-undersaturated magmas and others fromdistinct silica-saturatedor -oversaturatedmagmas

Multi-element diagrams normalized to primi-tive mantle values show that the EF basalts areenriched in all incompatible elements relative toprimitive mantle values (Fig 4) The patterns ofthe EF basalts are characterized by a prominentBa peak a subordinate P peak and a decreasingelement enrichment with increasing compatibilityIn comparison with normal MORB (N-MORB)enriched MORB (E-MORB) and average OIB(Sun and McDonough 1989) the patterns of theEF basalts are closest to those of OIB both inshape and normalized concentrations

Whole-rock REE data

Whole-rock REE data are presented in Table 3and REE concentrations normalized to chondriticvalues of Boynton (1984) are shown in Fig 5aAll samples are characterized by LREE-enrichedpatterns with no or slightly positive Euanomalies (EuEu = 097 to 113) La concentra-tions are roughly 20 ndash60 times and Lu concentra-tions are 10 times chondritic values NormalizedLREE abundances from La to Nd are fairlyconstant but the patterns show a steady andsmooth decline from Nd towards heavier REEThe REE patterns of the EF basalts fall in betweenthose of typical OIB and E-MORB (Fig 5b)They have similar patterns and absolute abun-dances compared to basaltic dykes from theIvittuut area South Greenland (Goodenough etal 2002) LaNSmN as a measure of LREEfractionation has a restricted range from 130 to165 The degree of HREE fractionationexpressed as GdNYbN is slightly more variableand ranges from 157 to 257 Both these ratios of

the EF basalts fall generally between typical OIBand E-MORB compositions

Clinopyroxene REE patterns

Typical analyses of REE concentrations inclinopyroxenes are given in Table 3 They showthat REE concentrations in individual minerals ofone particular sample can be quite variableChondrite-normalized REE patterns of clinopyr-oxenes are very similar in the samples from theMussartut and Ulukasik Groups (Fig 6ab) Theyare enriched in all REE relative to chondriticvalues and the patterns show a fairly steepincrease from La to Sm followed by a gradualsmooth decrease from Sm to Lu Clinopyroxenesfrom sample EF 168 (Il otildeacutemaussaq Group) aredistinct as they have generally higher REEcontents especially in the LREE and a weak Euanomaly (Fig 6c) The 3 ndash106 enrichment in Nd

c (sa

mpl

e)c

(prim

itive

man

tle)

c (sa

mpl

e)c

(prim

itive

man

tle)

1

10

100

1000

Mussartucirct GroupUlukasik GroupIliacutemaussaq Group

Rb Ba K Nb La Ce Sr Nd P Zr Ti Y

Rb Ba K Nb La Ce Sr Nd P Zr Ti Y01

1

10

100

1000N-MORBE-MORBOIB

FIG 4 Multi-element plots normalized to primitivemantle values (McDonough and Sun 1995) Data forN-MORB E-MORB and average OIB (Sun and

McDonough 1989) are shown for comparison

PETROGENESIS OF THE ERIKSFJORD BASALTS

839

TA

BL

E3

RE

Eda

taof

who

lero

cks

and

clin

opyr

oxen

es

the

latt

erde

term

ined

byL

aser

ICP

-MS

Who

le-r

ock

anal

yses

R

epre

sent

ativ

ecl

inop

yrox

ene

anal

yses

S

ampl

eno

10

1231

1014

4610

1457

1014

7218

6389

1863

95E

F02

4-2

EF

039-

1E

F03

9-2

EF

039-

4E

F07

2-2

EF

072-

3E

F16

8-3

EF

168-

5

RE

Eco

ncen

trat

ions

(ppm

)L

a11

65

180

610

29

661

155

412

28

440

266

199

150

273

304

165

527

59

Ce

290

346

05

249

617

13

378

130

08

180

411

46

868

717

105

012

60

572

399

67

Pr

434

67

388

268

527

433

331

216

181

128

249

262

102

917

43

Nd

200

631

71

176

412

49

250

920

16

190

213

82

122

28

9815

64

157

258

51

960

6S

m4

756

974

043

215

914

826

664

194

283

276

077

0615

82

265

9E

u1

742

51

571

071

971

721

831

491

111

212

001

943

877

07G

d4

826

584

043

555

84

996

965

615

244

326

156

0514

75

252

7T

b ndash

ndash ndash

ndash ndash

ndash1

351

010

770

711

111

212

313

61D

y4

475

223

613

565

354

816

865

934

784

255

996

8613

10

205

2H

o0

921

040

730

731

11

011

481

401

001

001

211

292

394

24E

r2

432

531

951

972

872

663

893

342

642

112

993

646

539

60T

m ndash

ndash ndash

ndash ndash

ndash0

450

390

360

290

350

440

861

27Y

b2

182

071

741

832

592

363

782

812

492

452

703

016

788

89L

u0

340

310

260

270

390

350

460

410

300

310

520

411

251

60

(Eu

Eu

) N1

111

131

190

971

031

070

820

940

720

991

000

910

770

83

Eu

anom

aly

isca

lcul

ated

as(E

uE

u) N

=E

u N(

SmN6

Gd N

)05

norm

aliz

edto

chon

drit

eva

lues

from

Boy

nton

(198

4)

840

R HALAMA ETAL

in clinopyroxenes of EF 168 compared to theother samples (EF 024 039 and 072) seems onlypartly due to a higher degree of fractionationsincethe Mg value of EF 168 is only slightly lowerThus the LREE enrichment appears to underlinethe more alkaline character of this sample Acomparison with published REE patterns ofclinopyroxenes (Fig 6d) shows that the EFpatterns are unlike patterns from olivine gabbroswith MORB-like signatures (Benoit et al 1996)but fairly similar to clinopyroxene megacrystpatterns from an OIB-like alkaline basalt (Nimisand Vannucci 1995)

Sr Nd and O isotopic composition

Sr Nd and O isotopic compositions of all samplesare presented in Table 4 Sr and Nd isotopiccompositions of the clinopyroxene separates havebeen recalculated to an approximate eruption age

of 12 Ga (Paslick et al 1993) using decayconstants of 142610ndash11 andash1 for 87Rb (Steigerand Jager 1977) and 654610ndash12 a ndash1 for 147Sm(Lugmair and Marti 1978) Present day CHURvalues of 01967 for 147Sm144Nd (Jacobsen andWasserburg 1980) and 0512638 for 143Nd144Nd(Goldstein et al 1984) were used to calculateinitial eNd values 87Sr86Sr initial ratios are fairlyconstant and range from 070278 to 070383(Table 4) eNd(i) values vary from ndash32 to +21(Table 4) On the Sr-Nd isotope diagram the EFbasalts de ne a relatively steep trend (Fig 7) withinitial 87Sr86Sr ratios similar to those of the BulkSilicate Earth (BSE) Only one sample is slightlydisplaced towards a more radiogenic initial87Sr86Sr ratio None of the samples has an eNd

value close to depleted mantle values estimatedto be between eNd = +53 (calculated afterDePaolo 1981) and eNd = +73 (calculated afterGoldstein et al 1984) at 12 Ga The two alkalinebasalts of the Il otilde maussaq Group are within therange of the other data but they de ne a restricted eld around CHUR and BSE respectively Errorsinduced by the uncertainty in the age of the EFbasalts are largely within the error of the isotopeanalyses and do not change the overall picture Incomparison with other published Sr-Nd isotopedata from the Gardar Province the initial Srisotope ratios of the EF basalts are very similar torelatively primitive basaltic lamprophyric andcarbonatitic dyke rocks which have (87Sr86Sr)i of~0703 (Pearce and Leng 1996 Andersen 1997Goodenough et al 2002) So far no evidence hasbeen presented for an isotopicallyenriched mantlesource in the Gardar Province as the existingeNd(i) data of ndash05 to +52 show an isotopicallydepleted signature (Pearce and Leng 1996Andersen 1997 Goodenough et al 2002Upton et al 2003) Clinopyroxene from the EFsamples of Paslick et al (1993) yield eNd(i) of+22 The Nd isotope data of this study are withinthe lower range of published values from theGardar Province and extend the spread to morenegative eNd(i) values

Several recent oxygen isotope studies haveshown that mineral separates are less sensitive tolow-T alteration and secondary water take-up thanwhole-rock samples (eg Vroon et al 2001)However only few oxygen isotope data ofmineral separates from continental volcanics areavailable (eg Dobosi et al 1998 Baker et al2000) Extensive studies of mineral separates ofocean island basalts (Eiler et al 1997) andoceanic-arc lavas (Eiler et al 2000a) demon-

c (sa

mpl

e)c

(cho

ndri

te)

c (sa

mp

le)c

(cho

ndrit

e)

1

10

100

200

101446

101472

101231

101457

186389

186395

La Ce Pr Nd Sm Eu Gd Dy Ho Er Yb Lu

1

10

100

200

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

N-MORBE-MORBOIBBasaltic dykes from Ivittuut

a

b

FIG 5 (a) Whole-rock REE data of EF basaltsnormalized to chondritic values from Boynton (1984)(b) Typical compositions of N-MORB E-MORB andOIB (Sun and McDonough 1989) are shown for

comparison

PETROGENESIS OF THE ERIKSFJORD BASALTS

841

strated that analyses of mineral separates consis-tently span a narrower range than comparablewhole-rock data In our study we used clino-pyroxene because it is the only fresh mineralphase present in the samples Oxygen isotopemeasurements of clinopyroxene separates fromthe EF basalts show a range in d18Ocpx from +52to +60 with an average value of +566plusmn023(1s) All samples overlap the range of d18Ocpx

values from mantle peridotites (+525 to +590Mattey et al 1994) from continental oodbasalts from Yemen (+55 to +69 Baker etal 2000) and from alkaline basalts of Hungary(+507 to +534 Dobosi et al 1998) There is asmall difference between the averages ofMussartut Group (d18Ocpx-avg = +575plusmn015(1s)) and Ulukasik Group (d18Ocpx-avg =+575plusmn009 (1s)) samples compared toIl otildeacutemaussaq Group (d18Ocpx-avg = +533plusmn021(1s)) samples

d18O values of melts coexisting with clino-pyroxene were calculated assuming that oxygen

in pyroxene and melt were in isotopic equilibriumand T was 1100ordmC Dmelt-cpx can then be calculatedusing the following equation of Kalamarides(1986)

Dmelt-cpx = ndash00008T + 143 (T in Kelvin)

Although fractionation varies with magmatictemperatures we consider the temperature of1100ordmC and the resulting Dmelt-cpx of +03 assuf ciently representative considering the analy-tical errors involved It is similar to Dmelt-cpx

values used in previous studies (+039 Vroonet al 2001 +02 Dobosi et al 1998)Resulting d18Omelt values of EF basalts rangefrom +55 to +63 These values can becompared to whole-rock data from continentalbasalts given by Harmon and Hoefs (1995)Intraplate basalts from continental rift zonesshow a range from +46 to +83 with a meanof +61 and continental ood basalts which arerather poorly represented in the database havevalues between +43 and +65 Calculated EF

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Cpx from MORB-like

olivine gabbro

Cpx megacrysts from OIB-like

alkaline basalts

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 024EF 039

Mussartucirct Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 168

Iliacutemaussaq Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

1

10

100

1000

EF 072

Ulukasik Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb LuLa Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

c

ba

d

1

10

100

01

FIG 6 REE patterns of clinopyroxenes (cpx) from Mussartut (a) Ulukasik (b) and Il otilde maussaq (c) Group samplesnormalized to chondritic values from Boynton (1984) (d) REE patterns of cpx from MORB-type olivine gabbro(Benoit et al 1996) and OIB-type alkaline basalt (Nimis and Vannucci 1995) for comparison (note the change of

scale)

842

R HALAMA ETAL

TA

BL

E4

Sr

Nd

and

Ois

otop

icco

mpo

siti

ons

ofcl

inop

yrox

ene

sepa

rate

sfr

omE

riks

fjor

dF

orm

atio

nba

salt

sW

hole

-roc

kan

alys

esfr

omth

eK

etil

idia

nba

sem

ent

and

from

sedi

men

tary

rock

sof

the

Eri

ksfj

ord

For

mat

ion

are

give

nfo

rco

mpa

riso

n

Sam

ple

Mem

ber

Sr

Rb

87R

b86S

r87S

r86Sr

87Sr

86Sr

init

ial

SmN

d147Sm

144N

d143N

d144N

de N

d(i

)d1

8O

(ppm

)(p

pm)

(ppm

)(p

pm)

Bas

alts

E

F02

4M

ussa

rtut

646

50

486

002

180

7042

01plusmn

100

7038

267

3822

80

019

560

5124

67plusmn

10ndash

32

567

EF

063

Mus

sart

ut5

60plusmn

011

EF

061

Mus

sart

ut89

95

099

90

0321

070

3818

plusmn10

070

3266

574

163

90

2115

051

2629

plusmn10

ndash2

55

79plusmn

028

EF

039

Mus

sart

ut73

74

058

30

0229

070

3492

plusmn10

070

3098

662

191

30

2092

051

2626

plusmn10

ndash2

25

71plusmn

001

EF

059

Mus

sart

ut54

47

013

70

0073

070

3375

plusmn12

070

3250

606

168

90

2171

051

2727

plusmn09

ndash1

45

99plusmn

001

EF

108

Ulu

kasi

k66

92

111

00

0480

070

3741

plusmn09

070

2916

109

936

82

018

050

5126

17plusmn

10+

21

581

EF

072

Ulu

kasi

k69

41

032

20

0134

070

3014

plusmn10

070

2784

818

251

90

1962

051

2574

plusmn10

ndash1

25

68E

F17

4Il

otildemau

ssaq

125

42

324

005

360

7039

35plusmn

100

7030

1418

76

796

10

1424

051

2230

plusmn10

+0

45

47E

F16

8Il

otildemau

ssaq

866

71

422

004

750

7037

99plusmn

070

7029

8316

45

641

60

1550

051

2304

plusmn09

ndash0

15

18

Bas

emen

tan

dse

dim

enta

ryro

cks

JG02

Juli

aneh

aEcirc b30

01

169

71

6375

074

4789

plusmn10

071

6647

105

362

28

010

220

5114

70plusmn

09ndash

83

78

Gra

nite

EF

021

Qua

rtzi

te11

5E

F02

8A

rkos

ic6

8ar

enit

e

Sam

ples

are

list

edin

stra

tigra

phic

orde

rfr

omol

dest

toyo

unge

st87

Sr86

Sr

init

ialr

atio

san

de N

d(i)

valu

esw

ere

calc

ulat

edfo

ran

age

of1

2G

aS

tand

ard

devi

atio

nsfo

rd18

Oar

egi

ven

for

sam

ples

that

wer

ean

alys

edtw

ice

PETROGENESIS OF THE ERIKSFJORD BASALTS

843

model melt oxygen isotope data are within theranges of both groups Compared to Igaliko dykerocks from further SE in the Gardar Province(Pearce and Leng 1996) the EF basalts cover amuch more restricted eld possibly because d18Owhole-rock analyses are more prone to secondaryalteration than analyses of mineral separates(Fig 8) Whole-rock d18O measurements of theunderlying JulianehaEcirc b Granite and of the sedi-mentary sandstone members of the EriksfjordFormation yield values of 78 and 68 ndash115respectively (Table 4) and are shown forcomparison

Discussion

Mantle source regions of the EF basaltsTo reconstruct the mantle sources involved in theformation of the EF basalts we use Sr-Nd-Oisotope and incompatible trace element data On aSr-Nd isotope diagram the isotopic compositionof DMM was modelled for an age of 12 Ga(DePaolo 1981 Goldstein et al 1984 Taylor andMcLennan 1985) The fairly constant 87Sr86Srinitial isotope ratios of the EF basalts are ~0703which falls within the range typical for mantle-derived rocks but slightly above calculated valuesfor DMM (Fig 7) eNd(i) values are more

-4

-2

0

2

4

6

8

0701 0702 0703 0704 0705 0706

DMM Typical 2 s error

CHUR

BS

E

(87Sr86Sr)i

e Nd

(i)

EM-1

Qassiarsuk carbonatites and lamprophyres

EF basalts

Depleted MORB mantle

Ivittuut basaltic and lamprophyric dikes

Igaliko carbonatites and lamprophyres

Enriched mantle 1

Carbonatites from other regions of similar age

FIG 7 eNd(i) vs (87Sr86Sr)i diagram calculated for t = 12 Ga which is considered as the eruption age of the EFbasalts (Paslick et al 1993) The reference line of Bulk Silicate Earth (BSE) for the Sr isotopic composition wascalculated after DePaolo (1988) assuming present-day values of 00827 for 87Rb86Sr and 07045 for 87Sr86SrIsotopic composition of the depleted MORB mantle (DMM) was calculated after DePaolo (1981) Goldstein et al(1984) and Taylor and McLennan (1985) EM-1 was calculated after Schleicher et al (1998) (see text for details)Data sources are Goodenough et al (2002) Ivittuut (n = 6) Pearce and Leng (1996) Igaliko (n = 5) Andersen(1997) Qassiarsuk (n = 25) Bell and Blenkinsop (1989) other carbonatites of similar age (Goudini Schryburt Lake

and Prairie Lake n = 6)

844

R HALAMA ETAL

scattered and indicate that the rocks contain amixture of isotopic components from both mantleand crustal reservoirs or a heterogeneous mantlesource Only one sample (EF 108) has an eNd(i)value signi cantly above 0 This sample exhibitsthe largest spread of clinopyroxene compositionswhich may re ect strong alteration Goodenoughet al (2002) also note that the highest eNd occursin the most altered sample of their sample setHowever Paslick et al (1993) report clino-pyroxene data from two Ulukasik Groupsamples with eNd(i) gt +2 and those sampleswere apparently not disturbed with regard to theirSm-Nd age information On the other hand Piperet al (1999) suggest that the Sm-Nd systematicsof those samples may be erroneous due to laterdisturbance by uids derived from the Il otildeacutemaussaqintrusion Therefore although we are presentlynot able to decide whether positive eNd(i) values

for the EF basalts are real it is clear that none ofthose values is close to values of a 12 Ga oldDMM source (ie eNd = +5 to +7)

The Sr-Nd isotopic composition for OIB-typesources in the Proterozoic is not as wellconstrained as DMM Today OIB-type mantlesources have a wide range of isotopic composi-tions that partly overlap with DMM but eNd

values are generally lower (Hart et al 1992Hofmann 1997) The EM-1 mantle component ischaracterized by extremely low 143Nd144Ndratios and eNd values of between 0 and ndash5(Hofmann 1997) A possible 12 Ga position forEM-1 is plotted in Fig 7 calculated according toSchleicher et al (1998) with 147Sm144Nd = 018and 87Rb86Sr = 010 The present-day values usedin the calculation are 143Nd144Nd = 051234 and87Sr86Sr = 07057 taken from ocean islandbasalts closest to an EM-1 end-member(Hofmann 1997) Other Sr-Nd isotope datafrom the Gardar Province are plotted to putfurther constraints on the mantle sources of the EFbasalts (Fig 7) We also added comparative datafrom African and Canadian carbonatitesof similarage because there is a well known similaritybetween Sr-Nd isotopic data of OIBs andcarbonatites (Bell and Blenkinsop 1989) Thelinear array of the carbonatite data is consideredto be the result of mixing between the two mantlecomponents HIMU and EM-1 (Bell andBlenkinsop 1989 Bell and Tilton 2001) Thecarbonatites from the Gardar Province havesimilar Nd but slightly more radiogenic Srisotopic compositions than the reference carbo-natite data (Fig 7) The least contaminatedsamples overlap with the linear array and themore radiogenic Sr isotopic composition can beexplained by crustal contamination (Andersen1997) We therefore consider it possible that theGardar carbonatites also represent mixing ofhypothetical HIMU and EM-1 components of anOIB-like source Since the isotopically mostprimitive EF basalt samples overlap with theQassiarsuk carbonatites it could be possible thatthe basalts also contain such mantle componentsalthough Pb isotope data for the EF basalts thatcould clarify the importance of a HIMU sourceare lacking Based on the Sr-Nd isotope data theEF basalts appear to be derived from aheterogeneous OIB-like mantle source thatcomprises isotopically depleted and enrichedcomponents OIB-like and carbonatitic magma-tism are often attributed to mantle plume activity(Bell and Simonetti 1996 Hofmann 1997

2

4

6

8

10

12

-10 -8 -6 -4 -2 0 2 4 6 8

DMM

d18 O

[permil]

e Nd (i)

Typical2s error

EM-1

EF basalts - cpx

EF basalts - model melt

Ketilidian basement

Igaliko dykes

Depleted MORB mantle

Enriched mantle 1

EM-1 lavas - cpx

FIG 8 Oxygen isotopi c compositio ns (d18OV-SMOW) of clinopyroxene from EF basalts calculatedmelt (see text for details) and whole-rock Ketilidianbasement vs eNd(i) Whole-rock data for Igaliko dykerocks (n = 7) are from Pearce and Leng (1996) DMMdata derived from fresh MORB glasses are from Eiler etal (2000b) The eld of EM-1 lavas is taken fromclinopyroxene data of Gough and Tristan da CunhaIsland lavas (Harris et al 2000) and the EM-1 end-member composition is based on the calculations afterSchleicher et al (1998) for eNd(i) and an average value

of 578 (n = 18) from the EM-1 lavas for d18O

PETROGENESIS OF THE ERIKSFJORD BASALTS

845

Simonetti et al 1998) A plume input to theformation of the EF basalts is supported by therelatively large volume of magmas and the shortduration (lt5 Ma) of magmatism (Piper et al1999) It is also compatible with the geologicalsetting of a continental breakup Additionally aplume component appears to be necessary toexplain the thermal input as it is dif cult to melt acold lithosphere section (Arndt and Christensen1992)

Oxygen isotope data of the EF basalts aregenerally consistent with a mantle derivationincluding an OIB-like mantle source On the d18Ovs eNd(i) diagram (Fig 8) the overlap betweenthe EF data clinopyroxene data from EM-1-likeoceanic basalts and the hypothetical EM-1 end-member suggests an involvement of an EM-1-type mantle source Olivines from HIMU-likesources are known to be slightly lower in d18Othan typical upper mantle values from MORBs(Eiler et al 1997) but the EF data are notsuf cient to assess if the slightly lower d18Ovalues in the Il otildeacutemaussaq Group basalts are causedby a relatively higher input from a HIMU source

The contribution of the various mantle sourcesto the EF basalts can also be evaluated on theprimitive mantle-normalized multi-elementdiagram (Fig 4) and on the REE plots (Figs 5and 6) Both the whole-rock REE and theclinopyroxene REE patterns of the EF basaltsare unlike those representative of N-MORBsources We conclude again that DMM did notplay a major role in the petrogenesis of the EFbasalts On the other hand similarities with OIB-like mantle sources are more prominent Theshape of the clinopyroxene REE pattern closelyresembles those from clinopyroxene of OIB-likealkaline basalts despite differences in absoluteconcentrations The whole-rock REE patterns ofthe EF basalts lie between E-MORB and averageOIB patterns as do the multi-element patterns Onthe latter the EF basalts have marked positive Baand P peaks that are not present in typical OIBsHowever Ba contents should be interpreted withcaution since highly variable ZrBa ratios from014 to 070 are an indication that Ba was re-mobilized after emplacement of the basalts (egPrice et al 1991 Oliveira and Tarney 1995)Likewise P mobility was demonstrated in alteredbasaltic rocks (Winchester and Floyd 1976) Insummary the REE and trace element data areevidence that the mantle source(s) of the EFbasalts were enriched in incompatible traceelements relative to primitive mantle

For relatively primitive hy-normative basalticrocks of the Ivittuut area further north in theGardar Province it has been argued that some ofthem were derived from a lithospheric mantlesource previously enriched by subduction-related processes This interpretation is basedon certain geochemical characteristics such ashigh LaNb ratios and negative Nb anomalies(Goodenough et al 2002) For the EF basaltsnegative Nb anomalies are not clearly discern-ible but LaNb ratios are gt1 which seems to be acharacteristic feature of a lithospheric mantlesource (Fitton 1995) LaBa LaRb and LaKratios were also used to argue for a lithosphericorigin of the Gardar basic magmas although amantle plume involvement was assumed toprovide the required energy (Upton andEmeleus 1987 Upton et al 2003) Sincefeldspar is the only primary magmatic mineralin the EF basalts in which Ba Rb and K areincorporated in signi cant amounts and thefeldspars are generally altered these elementsmight have been easily mobilized and wetherefore do not dare to base the interpretationof the EF basalts sources on these data Inaddition the lack of precise geochemical data onthe composition of the lithosphere below theGardar Province complicates an evaluation ofthe lithospheric contribution to the EF basalticmagmatism

Certain ratios of incompatible trace elements ofthe EF basalts are compared to several occur-rences of ocean island basalts representative ofEM-1 and HIMU sources in Table 5 Ratiosbetween highly incompatible trace elements donot change signi cantly during partial melting orfractional crystallization in basaltic systems andmay therefore constrain the ratios in their sources(Weaver et al 1987) The EF basalts aresubdivided into the more alkaline isotopicallyprimitive Il otildeacutemaussaq Group samples and thosesamples with eNd(i) lt ndash1 The Il otilde maussaq Groupsamples have higher SrNd LaNb and ZrNbratios than HIMU and overlap with the upperrange of EM-1 ratios The EF basalts with eNd(i)lt ndash1 have generally higher ratios than the oceanicisland basalts and partly overlap with typicalvalues of the lower crust It therefore seems likelythat assimilation of lower crustal material mayhave changed the primary trace element ratios inthe basalts Therefore we will evaluate thepossible in uence of assimilation of crustalmaterial on the geochemistry of the EF basaltsin the following section

846

R HALAMA ETAL

The role of crustal contamination

Large amounts of bulk assimilation of crustalmaterial which is relatively SiO2-rich comparedto a basaltic melt would drive the compositions tohigher SiO2 contents and quartz-normativecompositions Partial melts derived from crustalrocks would also be relatively SiO2-rich (Beard etal 1993 Carroll and Wyllie 1990) and wouldhave the same effect The ne-normative characterof the EF basalts (Table 1) suggests thatassimilation of crustal material especially ofsilica-rich upper crustal rocks was not signi cantwith the possible exception of sample EF 059

On the primitive mantle-normalized multi-element diagrams (Fig 4) Ti troughs are typicalfor crustal rocks especially for the Upper Crust(Rudnick and Fountain 1995) and they maytherefore be present in rocks extensively contami-nated by continental crust Ti troughs aregenerally lacking in the patterns of the EFbasalts and a major upper crustal contribution tothe magmas can therefore be excluded However

small amounts of assimilation of lower crustalmaterial cannot be ruled out because the averagelower crust has higher Ti contents

Steep trends on Sr-Nd isotope plots have beenattributed to assimilation of granulite-faciesgneisses (eg Bernstein et al 1998) AFCmodelling based on an equation developed byDePaolo (1981) was performed to test thehypothesis of small amounts of contaminationwith lower crustal material (Fig 9) The basalticmagma which represents one end-member for thecalculations has the isotopic signature of theisotopically most primitive samples from thepresent data set and from the literature(Table 6) Two possible contaminants repre-senting upper and lower crustal compositionswere selected for modelling (Table 6) Therepresentative samples are a granite from theJulianehaEcirc b batholith for the upper crust (sampleJG 02) and an average of ve granulite-faciesgneisses with low RbSr ratios from the Archaeancraton of West Greenland (Taylor et al 1984) for

TABLE 5 Trace element ratios of EF basalts compared to ocean island basalts(OIBs) representing EM-1 and HIMU mantle components and the lowercrust

SrNd LaNb ZrNb

EF basalts with eNd lt ndash 1 249 ndash396 120 ndash150 121 ndash157Il otildeacutemaussaq Group EF basalts 198 ndash292 104 ndash119 51 ndash53EM-1 90 ndash207 067 ndash111 39 ndash109HIMU 45 -163 063 ndash080 27 ndash 36Lower continental crust 30 16 136

Data sources EM-1 Humphris and Thompson (1983 Walvis Ridge) and Cliff et al(1991 Inaccessible Islands) HIMU Chauvel et al (1992 Tubuai Island) Lowercontinental crust Rudnick and Fountain (1995)

TABLE 6 End-member compositions used for AFC modelling

End-member Sr (87Sr86Sr)i Nd eNd(i) Data source(ppm) (T = 12 Ga) (ppm) (T = 12 Ga)

Primitive EF basaltic 507 070278 18 + 22 Sample EF 072magma Paslick et al (1993)Upper crust 300 071665 62 ndash83 Sample JG 02(JulianehaEcirc b Granite)Lower crust 747 070668 63 ndash 234 Taylor et al (1984)(Granulite facies gneiss)

The composition of the granulite-facies gneiss is an average of 5 samples with low RbSr ratios lt02

PETROGENESIS OF THE ERIKSFJORD BASALTS

847

the lower crust An in uence of Archaean rockson Gardar magmas has been demonstrated by Pbisotope studies (Taylor and Upton 1993) andseismic data were also interpreted to re ect theexistence of a wedge of Archaean crust thatextends southwards to Lindenow Fjord (Dahl-Jensen et al 1998) We have shown in theprevious sections on the basis of Ni trends thatfractionation of olivine was most relevant to themajor element evolution of the lavas whereasfractionation of other phases was apparently ofminor importance We therefore modelled theAFC processes with olivine fractionation usingmineral-melt partition coef cients of 0014 for Srand 00066 for Nd (Rollinson 1993)

The modelling of AFC processes is stronglydependent on the parameter r de ned as the ratioof mass assimilation rate and fractional crystal-lization rate (DePaolo 1981) The change inisotopic signature becomes enhanced for highervalues of r Reiners et al (1995) have shown thatr values might be very high and even gt1 in theearly stages of magmatic evolution at hightemperatures since crystallization can besuppressed by assimilation Therefore we calcu-

lated AFC processes with two r values of 05 and08 that represent a relatively high and a middle tolow assimilation rate to demonstrate the effects ofvariable r (Fig 9) The results indicate that AFCprocesses involving upper crust similar to theJulianehaEcirc b Granite cannot explain the trend of theEF basalts because Sr isotopic compositionswould shift to considerably more radiogenicvalues This result is consistent with theconclusions derived from trace element dataThe AFC curves for assimilation of lower crusthowever agree well with the bulk of the data Themost negative eNd values can be explained byAFC processes with intermediate assimilationrates (r = 05) involving assimilation of lt5lower crustal material With a greater degree ofassimilation and an r value of 08 lt4assimilation is indicated

Lower crustal rocks of igneous origin have anaverage d18O value of +75plusmn14 with an overallrange from +54 to +125 (Fowler and Harmon1990) 5 to 10 bulk assimilation of lower crustalmaterial with d18O = +75 into a typical mantle-derived magma with d18O = +57 results ind18O values of the contaminated magma from

-24

-20

-16

-12

-8

-4

0

4

0700 0703 0706 0709 0712 0715 0718

r = 08

r = 05

r = 05

r = 08

EF Basalts

Endmembers of AFC calculations

Upper Crust

Lower Crust

Primitive EF basaltic magma

(87Sr86Sr)i

e Nd

(i)

09

08

06

09

08

06

09

0806

09

08

07

07

04

Typical errorCHUR

BS

E

FIG 9 eNd(i) vs (87Sr86Sr)i diagram showing the results of AFC modelling End-member compositions are listed inTable 6 Olivine is the only fractionating mineral Partition coef cients for olivinemelt are from the compilation ofRollinson (1993) Curves were calculated for two values of r (r = rate of assimilation rate of fractionalcrystallization) Dashed lines and numbers in italics are for r = 05 solid lines and regular numbers are for r = 08

Values at the curves are for F values (fraction of melt remaining) with 10 tick marks

848

R HALAMA ETAL

+579 to +588 Therefore the d18O data cannotbe used to rule out small amounts of assimilationof lower crust To evaluate how much contamina-tion would be permissible from the presentoxygen isotope data we calculated how muchcontamination of a mantle-derived magma with18O = +57 with a contaminant within the lowerrange of lower crustal xenoliths (d18O = +67)would drive the oxygen isotopic composition tothe highest value of the EF basalts (d18O~+63) Although simple mixing shows that60 assimilation is permissible this amountwould drive the whole-rock geochemistry tomore silicic compositions than is observed in theEF basalts In Fig 8 there is no obvious trendfrom the basalts towards the Ketilidian basementsuggesting again that upper crust was notassimilated in signi cant amounts

Ratios of trace elements incompatible inbasaltic systems can be used to distinguishbetween crust and mantle reservoirs (egWeaver 1991) Based on the AFC modellingEF basalts with eNd(i) lt ndash1 are those which aremost likely in uenced by crustal contaminationThe relatively high LaNb and ZrNb ratios ofthese samples are consistent with assimilation oflower crustal material SrNd ratios are evenhigher than the average lower crustal composi-tion but SrNd ratios in those samples which weretaken as representative for the lower crust in theAFC modelling can reach values of 80 ie wellabove the highest EF basalt ratios Again theSrNd ratios support the hypothesis of some lowercrustal contamination

The absence of more primitive basaltic rockswith MgO gt8 wt might also indicate lowercrustal assimilation It suggests that magmas mayhave been retained at depth where they acquiredmore evolved compositions by fractional crystal-lization and had the opportunity to assimilatelower crustal material during crustal underplating(Upton et al 2003)

Summary and conclusions

The new geochemical data of the EriksfjordFormation basalts in the Gardar Provinceprovided in this study are used to placepetrogenetic constraints on the sources of theirparental magmas Based on trace-element patternsand Sr-Nd isotopic constraints there seems to beno indication that a DMM-like source played animportant role in magma genesis of the EFbasalts Clinopyroxene and whole-rock REE

patterns are more compatible with an OIB-likesource but we are not able to conclude whetherthe observed enrichment in REE re ects thesourcersquos signature or magmatic processes Theisotopically most primitive EF basalts have Sr-Ndisotopic compositions that are similar to Gardarcarbonatites Due to the isotopic similaritybetween carbonatites and OIBs many arethought to be related to mantle plumes (Bell2001) We therefore speculate that OIB-likemantle components from a plume source werealso involved in the parental EF magmas but wecannot exclude interaction with the SCLM Twodistinct mantle components one isotopicallydepleted and another one enriched could havecontributed to the EF magmas AFC modelling ofSr-Nd isotope data indicates that assimilation oflt5 lower crust could explain the steep vector onthe Sr-Nd isotope plot but assimilation of uppercrustal material is unlikely Oxygen isotope dataof clinopyroxene separates are consistent with amantle origin but permit some assimilation oflower crustal rocks In summary the data areconsistent with the hypothesis that the EFmagmas could have originated from a mantleplume source which is supported by geologicalarguments and that they underwent someassimilation of lower crustal material

Acknowledgements

We would like to thank Bruce Paterson whoprovided invaluable help during Laser ICP-MSmeasurements at the Large-Scale GeochemicalFacility supported by the European Community-Access to Research Infrastructure action of theImproving Human Potential Programme contractnumber HPRI-CT-1999-00008 awarded to ProfBJ Wood (University of Bristol) Gabi Stoschekand Torsten Vennemann are thanked for their helpwith oxygen isotope measurements Elmar Reitterexpertly assisted with preparation and measure-ments of radiogenic isotopes and MathiasWestphal helped with microprobe measurementsREE whole-rock analyses were kindly providedby the NERC facility at Royal HollowayUniversity of London Michael Marks is thankedfor his pleasant company during eldwork Veryconstructive reviews by Tom Andersen andAlexei S Rukhlov greatly helped to improve themanuscript We also thank Ian Coulson for nalediting Funding of this work by the DeutscheForschungsgemeinschaft (grant Ma-21351-2) isgratefully acknowledged

PETROGENESIS OF THE ERIKSFJORD BASALTS

849

References

Andersen T (1997) Age and petrogenesis of theQassiarsuk carbonatite-alkaline silicate volcaniccomplex in the Gardar rift South GreenlandMineralogical Magazine 61 499 ndash513

Armstrong JT (1991) Quantitative elemental analysisof individual microparticles with electron beaminstruments Pp 261 ndash315 in Electron ProbeQuantitation (KFJ Heinrich and DE Newburyeditors) Plenum Press New York and London

Arndt NT and Christensen U (1992) The role oflithospheric mantle in continental ood volcanismThermal and geochemical constraints Journal ofGeophysical Research 97 10967 ndash 10981

Baker JA Menzies MA Thirlwall MF andMacpherson CG (199 7) Pet rogenes is ofQuaternary Intraplate Vocanism Sanarsquoa YemenImplications for Plume-Lithosphere Interaction andPolybaric Melt Hybridization Journal of Petrology 38 1359 ndash 1390

Baker JA MacPherson CG Menzies MAThirlwall MF Al-Kadasi M and Mattey DP(2000) Resolving crustal and mantle contributions tocontinental ood volcanism Yemen constraintsfrom mineral oxygen isotope data Journal ofPetrology 41 1805 ndash1820

Beard JS Abitz RJ and Lofgren GE (1993)Experimental melting of crustal xenoliths fromKilbourne Hole New Mexico and implications forthe contamination and genesi s of magmasContributions to Mineralogy and Petrology 11588 ndash102

Bell K (2001) Carbonatites relationships to mantle-plume activity Geological Society of AmericaSpecial Paper 352 267 ndash290

Bell K and Blenkinsop J (1989) Neodymium andStrontium Isotope Geochemistry of Carbonatites Pp278 ndash300 in Carbonatites ndash Genesis and Evolution(K Bell editor) Unwin Hyman London

Bell K and Simonetti A (1996) Carbonat iteMagmatism and Plume Activity Implications fromthe Nd Pb and Sr Isotope Systematics of OldoinyoLengai Journal of Petrology 37 1321 ndash1339

Bell K and Tilton GR (2001) Nd Pb and Sr IsotopicCompositions of East African Carbonat itesEvide nce for Mant le Mixing and PlumeInhomogene ity Journa l of Petrology 42 1927 ndash1945

Benoit M Polve M and Ceuleneer G (1996) Traceelement and isotopic characterization of ma ccumulates in a fossil mantle diapir (Oman ophiolite)Chemical Geology 134 199 ndash 214

Bernstein S Kelemen PB Tegner C Kurz MDBlusztajn J and Kent Brooks C (1998) Post-breakup basaltic magmatism along the EastGreenland Tertiary rifted margin Earth and

Planetary Science Letters 160 845 ndash862Boynton WV (1984) Geochemistry of the rare earth

elements meteorite studies Pp 63 ndash114 in RareEarth Element Geochemistry (P Henderson editor)Elsevier Amsterdam

Carroll MR and Wyllie PJ (1990) The systemtonalite-H2O at 15 kbar and the genesis of calc-alkaline magmas American Mineralogist 75345 ndash 357

Chauvel C Hofmann AW and Vidal P (1992)HIMU-EM The French Polynesian connectionEarth and Planetary Science Letters 110 99 ndash119

Clayton RN and Mayeda TK (1963) The use ofbromine penta uoride in the extraction of oxygenfrom oxides and silicates for isotope analysisGeochimica et Cosmochimica Acta 27 43 ndash52

Coulson IM and Chambers AD (1996) Patterns ofzonation in rare-earth-bearing minerals in nephelinesyenites of the North Qoroq center South GreenlandThe Canadian Mineralogist 34 1163 ndash1178

Cox KG Bell JD and Pankhurst RJ (1979) TheInterpretation of Igneous Rocks George Allen andUnwin London

Cross W Iddings JP Pirsson LV and WashingtonHS (1903) Quantitative Classi cation of IgneousRocks University of Chicago Press

Dahl-Jensen T Thybo T Hopper H and Rosing M(1998) Crustal structure at the SE Greenland marginfrom wide-angle and normal incidence seismic dataTectonophysics 288 191 ndash 198

DePaolo DJ (1981) Trace element and isotopic effectsof combined wallrock assimilation and fractionalcrystallisation Earth and Planetary Science Letters53 189 ndash 202

De P a o lo D J (1 9 8 8 ) Neodymium Iso topeGeochemistry An Introduction Springer-VerlagNew York

Dobosi G Downes H Mattey D and Embey-IsztinA (1998) Oxygen isotope ratios of phenocrysts fromalkali basalts of the Pannonian basin Evidence foran O-isotopically homogeneous upper mantle be-neath a subduction-in uenced area Lithos 42213 ndash 223

Eiler JM (2001) Oxygen isotope variations of basalticlavas and upper mantle rocks Pp 319 ndash364 inStable Isotope Geochemistry (JW Valley and DRCole editor s) Reviews in Minera logy andGeochemistry 43 The Mineralogical Society ofAmerica Washington DC

Eiler JM Farley KA Valley JW Hauri E CraigH Hart SR and Stolper EM (1997) Oxygenisotope variations in ocean island basalt phenocrystsGeochimica et Cosmochimica Acta 61 2281 ndash2293

Eiler JM Crawford A Elliot T Farley KAValley JW and Stolper EM (2000a) Oxygenisotope geochemistry of oceanic-arc lavas Journal

850

R HALAMA ETAL

incompatible elements compared to the primitivemantle There are several types of OIB compo-nents well de ned by Sr-Nd-Pb isotopic studies(Zindler and Hart 1986 Hart et al 1992) Theseare HIMU (lsquohigh mrsquo m 238U206Pb) enrichedmantle 1 (EM-1) and enriched mantle 2 (EM-2)(3) Subcontinental lithospheric mantle (SCLM)representing the mantle below the continentalcrust The SCLM is usually enriched inincompatible elements but it can be veryheterogeneous

Interaction of an OIB-like source with litho-spheric mantle (Ormerod et al 1988 Paslick etal 1995 Molzahn et al 1996) or DMM (egNicholson et al 1997) may have played a majorrole in magma genesis of continental volcanicrocks Lavas chemically similar to OIBs havebeen observed in many continental ood basaltprovinces eg Deccan Traps (Mahoney 1988)British Tertiary Volcanic Province (Fitton et al1997) and Parana-Etendeka (Gibson et al 1997)These are believed to represent mantle-derivedmelts from plumes not contaminated by litho-spheric material but they represent only a smallproportion of the total magma volume erupted(Gibson et al 1999) Therefore the SCLM hasbeen invoked in various models either as a sourceof large-scale melts (eg Turner et al 1996) or asassimilant in plume-derived melts (eg Gibson etal 1995) The SCLM appears to be especiallyimportant in the genesis of small-degree melts(Gibson et al 1999) Post-Archaean lithosphericmantle is considered to have been suf cientlyfertile to have contributed to the genesis of oodbasalts (Hawkesworth et al 1990)

The Gardar Province in South Greenland is arift-related igneous province with magmatismlasting from about 135 to 114 Ga (Emeleusand Upton 1976 Upton and Emeleus 1987Upton et al 2003) It comprises about 12 mainigneous complexes numerous dykes of variablechemical composition and a sequence of rift-related interlayered lavas and sediments(Eriksfjord Formation) The basaltic lavas of theEriksfjord Formation are among the geochemi-cally most primitive rocks of the ProvinceHowever except for detailed palaeomagneticinvestigations (Thomas and Piper 1992 Piper etal 1999) they have been studied relatively little(Larsen 1977 Paslick et al 1993) Due to theircomparatively primitive nature the basaltic rocksof the Eriksfjord Formation provide a goodopportunity to discern the geochemical signaturesof the mantle sources of the magmatic activity of

the Province Similar sources may have beeninvolved in the petrogenesis of the Gardaranorthosites alkali gabbros and nephelinesyenites In this study we present new geochem-ical data (XRF analyses of major and traceelements Sr Nd and O isotopic analyses REEanalyses of whole-rock samples and in situ REEanalyses of clinopyroxene) to constrain mantlesources and to determine the role of crustalcontamination in the petrogenesis of theEriksfjord Formation basalts In particular high-precision laser- uorination oxygen isotope datahave only recently become an important tool forevaluating the contribution of mantle andcontinental crust in suites of continental oodbasalts (Baker et al 2000) Since continentalintraplate basalts are the only major class ofterrestrial basalts that have not yet been exploredsystematically using these methods (Eiler 2001)the laser uorination oxygen isotope data ofclinopyroxene separates presented herein are animportant contribution to the knowledge ofoxygen isotopic compositions of continentalbasalts in general

Geological setting

The Eriksfjord Formation (EF) represents an~3500 m thick sequence of Precambrian supra-crustal rocks comprising continental sandstonesand conglomerates and eruptive rocks and sillsthat range from basalt through trachyte to alkalinephonolite and carbonatite (Larsen 1977) Therocks lie unconformably on the KetilidianJulianehaEcirc b batholith dated at 185 to 179 Ga(van Breemen et al 1974 Garde et al 2002) andare cut by the 116 Ga Ilotildeacutemaussaq intrusion(Waight et al 2002) The type-section of theEriksfjord Formation occurs on the Ilotildeacutemaussaqpeninsula and has been subdivided into sixstratigraphic groups (Thomas and Piper 1992)originally de ned as members (Poulsen 1964)(Fig 1) The three volcanic groups are known asfrom oldest to youngest Mussartut Ulukasik andIlotildeacutemaussaq The Mussartut Group is subdividedinto nine sub-units and the Il otilde maussaq Group issubdivided into three sub-units called SermilikTunugdliar k and Narssaq Fjeld (Fig 1) Asandstone unit precedes each volcanic group(Thomas and Piper 1992) The Mussartut Groupcontains the Qassiarsuk carbonatite complexdated at ~12 Ga by Rb-Sr and Pb-Pb isochrons(Andersen 1997) Paslick et al (1993) obtainedtwo Sm-Nd isochron ages of 117plusmn003 and

832

R HALAMA ETAL

120plusmn003 Ga for two samples of the UlukasikGroup However palaeomagnetic data indicate anage of 135 ndash131 Ga for the entire succession(Piper et al 1999) and the exact age of the EF isas yet unresolved (Upton et al 2003) Parts of theEriksfjord lavas have been subjected to extensivemetasomatism related to the intrusion of alkalinemagmas (Ranlov and Dymek 1991 Coulson andChambers 1996) The local presence of epidoteimplies thermal metamorphism at temperatures ofup to 300 ndash350ordmC (Piper et al 1999) Higheruption rates are inferred for the volcanic systemand the whole lifespan is considered to belt~5 Ma (Piper et al 1999)

Analytical procedures

Electron microprobe analysisMineral compositions were determined using aJEOL 8900 electron microprobe (EMP) at theInstitut fur Geowissenschaften UniversitatTubingen Germany An internal jrz correctionof the raw data was applied (Armstrong 1991)Both natural and synthetic standards were usedfor major and minor elements Measuring times

were 16 s and 30 s on the peak positions formajor and minor elements respectively Theemission current was 15 nA and the accelerationvoltage 15 kV For feldspar analyses a beamdiameter of 5 mm was used to avoid errorsresulting from diffusion of Na

XRF analysis

Whole-rock analyses were performed by standardX-ray uorescence (XRF) techniques at theInstitut fur Geowissenschaften at the JohannesGutenberg-Universitat Mainz Germany using aPhilips PW 1404 spectrometer and at theUniversitat Freiburg using a Philips PW 2404spectrometer Prior to analysis whole-rocksamples were crushed and milled in tungstencarbide mills The powders obtained were dried at105ordmC Loss on ignition (LOI) determinationswere carried out on 1 g of pulverized samplematerial which was heated for 2 h at 1000ordmCPressed powder pellets and fused glass discs wereprepared to measure contents of trace and majorelements respectively Natural standards wereused for calibration Detection limits vary

FIG 1 Map of the Eriksfjord Formation on the Narssaq peninsula with sample locations (after Poulsen 1964 andLarsen 1977) Sample 101472 (not shown) is from a succession ~2 km south of the settlement of Narsarsuaq further

to the NE

PETROGENESIS OF THE ERIKSFJORD BASALTS

833

between 1 and 9 ppm depending on the speci ctrace element and on the instrument used

Whole-rock REE analyses

Whole-rock REE analyses were carried out by theNatural Environment Research Council (NERC)inductively coupled plasma-atomic emissionspectrometry (ICP-AES) facility at RoyalHolloway University of London Solutions wereprepared using a combined HF dissolution andalkali fusion as described in Walsh et al (1981)Resultant solutions were analysed for 12 REEsand selected potential interferences using a PerkinElmer Optima 3300RL ICP-AES

Laser ICP-MS in situ REE measurement of clinopyroxene

In situ laser ablation inductively coupled plasma-mass spectrometer (LA-ICP-MS) REE analyses on~150 mm thick polished thin sections wereperformed at the EU Large-Scale GeochemicalFacility (University of Bristol) using a VGElemental PlasmaQuad 3 + S-Option ICP-MSequipped with a 266 nm Nd-YAG laser (VGMicroProbe II) The laser beam diameter at thesample surface was ~20 mm All measurementswere made using Thermo Elemental PlasmaLablsquotime-resolved analysisrsquo (TRA) data acquisitionsoftware with a total acquisition time of 100 s peranalysis allowing ~40 s for background followedby 50 s for laser ablation NIST 610 glass was usedfor instrument calibration and NIST 612 was usedas a secondary standard Ca was used as an internalstandard to correct the ablation yield differencesbetween and during individual analyses on bothstandards and samples To avoid analyticaluncertaintiesdue to variations in the concentrationsof the internal standard Ca concentrations werequantitatively measured within 20 mm of the laserablation pits using the EMP at the Institut furGeowissenschaften Universitat Tubingen Theprecision of trace-element concentrations basedon repeated analyses of standards is ~plusmn5 forelement concentrations gt10 ppm and plusmn10 forconcentrations lt10 ppm Data processing wascarried out of ine using the same PlasmaLabsoftware used for data collection and variouscustom-designed Excel spreadsheets The limitsof detection are de ned as 328 standard deviationsabove background level which is equal to a 95con dence that the measured signal is signi cantlyabove background Typical detection limits for theREE in this study were 004 ndash06 ppm

Sr and Nd isotope analysisFor Sr and Nd isotope analyses 15 ndash20 mgsamples of hand-picked clinopyroxene werespiked with mixed 84Sr ndash 87Rb and 150Nd ndash 149Smtracers before dissolution under high pressure inHF at 180ordmC in poly-tetrauor-ethylene (PTFE)reaction bombs The Rb and Sr were separated inquartz columns containing a 5 ml resin bed ofAG50W-X12 200ndash400 mesh equilibrated with25 N HCl The Sm and Nd separation wasperformed in quartz columns using 17 mlTe on powder coated with HDEHP (Di-EthylHexyl Phosphate) as the cation exchange mediumequilibrated with 018 N HCl All analyses weremade with a multicollector Finnigan MAT 262thermal ionization mass spectrometer (TIMS) instatic collection mode at the Institut furGeowissenschaften Universitat Tubingen TheSr was loaded with a Ta-HF activator andmeasured on a single W lament The Rb Smand Nd were measured in a double Re- lamentcon guration mode 87Sr86Sr and 143Nd144Ndratios were normalized for mass fractionation to86Sr88Sr = 01194 and 146Nd144Nd = 07219respectively The average 143Nd144Nd ratioobtained for the Ames Nd-standard (GeologicalSurvey of Canada Roddick et al 1992) was0512119plusmn10 (n = 42) and the average 87Sr86Srratio for the NBS 987 Sr-standard was0710261plusmn16 (n = 30) during the course of thisstudy 143Nd144Nd ratios have been cross-checked with the La Jolla Nd-standard whichgave 0511831plusmn30 (n = 12) Total proceduralblanks (chemistry and loading) were lt200 pg forSr and lt100 pg for Nd

Oxygen isotope analysis

The oxygen isotope composition of hand-pickedclinopyroxene separates was measured using amethod similar to that described by Sharp (1990)and Rumble and Hoering (1994) 1 ndash2 mg ofhand-picked clinopyroxenewere loaded onto a Pt-sample holder The sample chamber was pumpedout to a vacuum of ~10 ndash6 mbar and pre- uorinated overnight Samples were then heatedwith a CO2-laser in an atmosphere of 50 mbar ofpure F2 Excess F2 is separated from O2 byconversion to Cl2 using KCl held at 150ordmC Theextracted O2 is collected on a molecular sieve(13X) Oxygen isotopic compositions weremeasured on a Finnigan MAT 252 mass spectro-meter at the Institut fur GeowissenschaftenUniversitat Tubingen The results are reported as

834

R HALAMA ETAL

the per mil deviation from Vienna Standard MeanOcean Water (V-SMOW) in the standard d-notation Replicate oxygen isotope analyses ofthe NBS-28 quartz standard (Valley et al 1995)had an average precision of plusmn01 for d18O Ineach run standards were analysed at thebeginning and the end of the sample set Acorrection was applied to the data equal to theaverage of the difference between the meanmeasured value and the accepted value for thestandard (964)

Oxygen isotope compositions of powderedwhole-rock samples were determined by aconventional method modi ed after Clayton andMayeda (1963) using BrF5 as reagent andconverting the liberated oxygen to CO2 beforemass spectrometric analysis

Petrography

175 samples were collected from all threevolcanic units of the Eriksfjord Formation butmost of them were extensively altered Theycontained secondary chlorite Fe-talc Ca-Fesilicates and carbonate as alteration productsSince the focus of this study was in situ analysesof minerals or analyses of mineral separates onlynine samples with a signi cant proportion of freshclinopyroxene and relatively few alterationproducts were selected for analysis Due to thelimited number of suitable samples we did notattempt to investigate the lavas in terms of theirstratigraphic position In addition to the ninesamples from which pyroxene separates wereobtained REE whole-rock data of six othersamples were added to the data set

In general the Mussartut Group lavas areolivine-free but contain up to 50 chloriticmatrix that probably results from devitri cationof former glass In contrast the majority ofUlukasik and Il otilde maussaq Group lavas containolivine-pseudomorphs and tend to be coarsergrained and holocrystalline Many of themcontain lsquosnow- akersquo glomerocrysts of plagioclaseindicating plagioclase-saturation or super-satura-tion at near-surface levels (Upton 1996) Thesamples from which clinopyroxenes have beenseparated are ne-grained basalts with plagioclaselaths embedded in interstitial clinopyroxeneas themajor and apatite and Fe-Ti oxides as subordinateprimary magmatic mineral phases The texturecan be described as ophitic and is believed toresult from the cotectic crystallization of plagio-clase and clinopyroxene(Philpotts 1990) Olivine

has been present in a few samples but it is nowcompletely altered Pyrite pyrrhotite and chalco-pyrite occur as accessory phases Some samplescontain two oxides (ilmenite + titanomagnetite)whereas others contain a single oxide exsolvedinto ilmenite and titanomagnetite or ilmenite andhematite Texturally they are mostly subhedral toanhedral although few euhedral grains occurHematite is also present around both discrete andexsolved ilmenite and titanomagnetite grains andthe exsolved hematite might be entirely ofsecondary origin formed by oxidation of titano-magnetite Late-stage interstitial quartz occursrarely Sample EF 108 (Ulukasik Group) isremarkable in having an intense reddish colourunlike the other samples that are almost black

Results

Mineral chemistryTypical analyses of clinopyroxenes are given inTable 1 Compositions range from En41Wo44Fs16

to En28Wo45Fs27 and broadly overlap for allgroups (Fig 2) According to the classi cation ofMorimoto et al (1988) clinopyroxenes of theMussartut and Ulukasik Groups are mainlyaugites with few diopsides present whereas inthe Il otilde maussaq Group samples of diopsidiccompositions dominate over augitic onesClinopyroxene analyses from the varioussamples form tight clusters on the pyroxenequadrilateral except for those from sample EF108 that are more scattered With only a singlepyroxene phase present in all samples clino-pyroxene compositions indicate minimum crystal-lization temperatures (Lindsley 1983) of750 ndash1000ordmC (Fig 2) Plagioclase is normallyzoned and its composition typically variesbetween An70Ab29Or1 and An16Ab72Or12 Lateinterstitial albite can reach compositions ofAn0Ab99Or1 There are no signi cant composi-

1200 Ecirc C

800 Ecirc C

25 50 75

25 25

En Fs

HdDi

Mussartucirct Group

Iliacutemaussaq Group

Ulukasik Group

FIG 2 Clinopyroxene analyses of the Eriksfjord basaltsplotted in the pyroxene quadrilateral with isotherms

after Lindsley (1983)

PETROGENESIS OF THE ERIKSFJORD BASALTS

835

tional changes among the various samplesIlmenite from two samples (EF 024 and EF 039)which contain coexisting ilmenite and titanomag-netite includes a signi cant component ofpyrophanite (MnTiO3) with up to 87 wtMnO It s com pos i t i ons va r y be t weenI l m 8 6 H e m 5 P y r 8 a nd I l m 7 6 H e m 5 P yr 1 9 Titanomagnetite analyses were unsatisfactory asall grains show strong alteration They areconsidered unreliable and are not presented here

Whole-rock geochemistry and classif|cation

Concentrations of major and trace elements aresummarized in Table 2 The basalts can beclassi ed according to the TAS diagram (LeMaitre et al 1989) as basalts and trachybasaltswith both alkaline and subalkaline compositions(not shown) Due to possible mobility of the Naand K the samples were also plotted into NbYvs ZrP2O5 and P2O5 vs Zr diagrams (Floyd andWinchester 1975) which can discriminate

between tholeiitic and alkaline series Thesediagrams do not yield consistent discriminationsfor the EF basalts but indicate that the characterof the sample suite is transitional betweentholeiitic and alkaline (not shown)

Mg [100Mg(Mg+Fe2+)] calculated assumingFe2O3FeO = 02 as recommended for basalts byMiddlemost (1989) varies between 54 and 40(Table 2) Ni concentrations range from 46 to115 ppm and Cr contents from 29 to 126 ppmTaking Mg as fractionation index a positivecorrelation (r2 = 061) with Ni contents indicatesfractionation of olivine (Fig 3) The relativelypoor positive correlations of Mg with Cr (r2 =044) and Sc (r2 = 023) argue against a signi cantfractionation of Cr-spinel or clinopyroxene Thisis consistent with the lack of clinopyroxenephenocrysts and the absence of any Cr-spinelThe negative correlations of Mg with Sr TiO2

and P2O5 suggest that fractionation of plagioclaseFe-Ti oxides and apatite was not important in thesample set investigated which is again consistent

TABLE 1 Typical clinopyroxene analyses of Eriksfjord Formation basalts determined by EMP

Sample EF 024 EF 024 EF 039 EF 059 EF 072 EF 108 EF 168 EF 174Cpx no cpx 2 cpx 3 cpx 5 cpx 3 cpx 5 cpx 7 cpx 4 cpx 1

SiO2 5060 5041 5128 4988 5145 4950 5175 5069TiO2 173 176 121 169 140 162 096 121Al2O3 297 300 179 223 171 264 155 213FeO 1160 1068 1218 1405 993 1234 979 1129MnO 028 021 030 040 019 028 025 031MgO 1254 1324 1263 1182 1325 1163 1307 1214CaO 1959 2029 2002 1907 2106 2099 2174 2143Na2O 034 031 029 049 033 055 052 054

Total 9965 9989 9971 9963 9932 9955 9962 9974

Formulae based on 4 cations and 6 oxygensSi 191 189 194 190 194 188 194 191Al 013 013 008 010 008 012 007 009Ti 005 005 003 005 004 005 003 003Fe3+ - - - 003 - 007 003 005Mg 071 074 071 067 075 066 073 068Fe2+ 037 033 039 041 031 032 028 031Mn 001 001 001 001 001 001 001 001Ca 079 082 081 078 085 085 087 087Na 002 002 002 004 002 004 004 004

Total 400 400 400 400 400 400 400 400

Pyroxene projection after Lindsley (1983)Wo 039 041 041 041 043 043 045 044En 040 041 039 037 040 038 040 038Fs 021 018 021 023 017 019 015 017

836

R HALAMA ETAL

TABLE 2 XRF whole-rock analyses of Eriksfjord Formation basalts

Sample EF 024 EF 063 EF 061 EF 039 EF 059 EF 108 EF 072 EF 174 EF 168Group Mussartut Mussartut Mussartut Mussartut Mussartut Ulukasik Ulukasik Ilotildeacutemaussaq IlotildeacutemaussaqSubunit 4 4 5 6 7 Lower Unit Upper Unit Sermilik Narssaq

Fjeld

Major elements (wt)SiO2 4573 4585 4688 4660 4875 4698 4656 4601 4545TiO2 230 200 180 175 191 265 203 299 277Al2O3 1578 1630 1600 1612 1645 1682 1668 1645 1719Fe2O3 1510 1356 1331 1383 1225 1394 1321 1438 1392MnO 028 018 018 018 017 014 017 018 017MgO 525 683 653 673 568 396 647 440 480CaO 842 849 872 857 920 641 808 746 737Na2O 349 274 289 276 236 448 303 323 376K2O 087 083 068 032 027 159 070 205 132P2O5 041 035 023 023 023 048 035 111 083LOI 199 210 240 238 242 204 210 052 156

Total 9962 9923 9962 9947 9969 9949 9938 9878 9914

Mg 453 545 538 536 524 403 538 421 450

Trace elements (ppm)Sc 24 22 25 22 26 19 20 21 16V 209 175 215 195 211 142 170 164 203Cr 81 108 66 65 126 48 82 33 29Co 63 78 70 62 71 44 68 35 46Ni 99 115 85 85 101 58 106 46 51Cu 65 49 46 60 97 41 59 38 41Zn 103 102 110 97 110 93 82 122 105Ga 20 18 21 19 20 19 16 21 19Rb 24 26 16 lt 10 lt 10 38 12 39 45Sr 504 503 512 392 373 712 507 850 992Y 31 26 28 23 28 31 17 31 25Zr 188 165 121 116 125 125 88 164 139Nb 12 11 10 lt 10 10 9 lt 10 31 27Ba 445 375 415 302 178 582 405 1293 900Pb 6 7 3 4 3 4 4 3 5La ndash ndash 15 ndash 12 11 ndash 37 28Ce ndash ndash 40 ndash 38 39 ndash 92 71Pr ndash ndash 1 ndash 1 2 ndash 9 4Nd ndash ndash 16 ndash 15 18 ndash 43 34Sm ndash ndash 6 ndash 7 6 ndash 5 5

CIPW normq ndash ndash ndash ndash 313 ndash ndash ndash ndashor 514 491 402 189 160 940 414 1212 780ab 2807 2319 2445 2335 1997 3207 2564 2733 2981an 2482 2973 2868 3065 3350 2109 2985 2433 2613ne 079 ndash ndash ndash ndash 316 ndash ndash 109di 1187 830 1076 853 877 633 654 452 416hy ndash 890 1059 1650 2217 ndash 1097 315 ndashol 1670 1309 1044 780 ndash 1472 1120 1388 1689mt 365 328 322 335 296 336 319 348 336il 437 380 342 332 363 503 386 568 526ap 095 081 053 053 053 111 081 257 192

Mg was calculated as Mg = 100Mg2+(Mg2++Fe2+) using Fe2O3FeO = 02 Normative mineral compositionswere determined according to the CIPW calculation scheme

PETROGENESIS OF THE ERIKSFJORD BASALTS

837

1 5

2 0

2 5

3 0

3 5

40

60

80

100

120

0

0 5

1

1 5

200

400

600

800

1000

0

50

100

150

15

20

25

30

4 0 45 50 5 5

r2 = 023

r2 = 044r2 = 061

r2 = 053

r2 = 082 r2 = 052

Ni (ppm)

Sc (ppm)

P2O5 (wt)

Sr (ppm)

TiO2 (wt)

Cr (ppm)

Mg Mg

Mg Mg

Mg

40 4 5 50 55

Mg

40 45 50 55

40 45 50 55

40 45 50 5540 4 5 50 55

FIG 3 Whole-rock variation diagrams for EF basalts with Mg as fractionation index Mg is calculated asMg(Mg+Fe2+) with Fe2O3FeO = 02 (Middlemost 1989)

838

R HALAMA ETAL

with the mostly subhedral Fe-Ti oxides and theophitic texture Ratios of incompatible traceelements such as ZrY (range 40 ndash63) andTiY (range 385 ndash716) are relatively uniformthroughout all samples but some scatter isevident Notable geochemical features are highSr contents up to 1000 ppm and very high Bacontents up to 1300 ppm

Normative mineral compositions were deter-mined according to the CIPW calculation scheme(Cross et al 1903 Cox et al 1979) Mostsamples are hypersthene-normative some arenepheline-normative and one (EF 059) is quartz-normative (Table 2) Despite the uncertaintiesinvolved in the CIPW norm calculation the dataindicate that some of the basalts crystallized fromsilica-undersaturated magmas and others fromdistinct silica-saturatedor -oversaturatedmagmas

Multi-element diagrams normalized to primi-tive mantle values show that the EF basalts areenriched in all incompatible elements relative toprimitive mantle values (Fig 4) The patterns ofthe EF basalts are characterized by a prominentBa peak a subordinate P peak and a decreasingelement enrichment with increasing compatibilityIn comparison with normal MORB (N-MORB)enriched MORB (E-MORB) and average OIB(Sun and McDonough 1989) the patterns of theEF basalts are closest to those of OIB both inshape and normalized concentrations

Whole-rock REE data

Whole-rock REE data are presented in Table 3and REE concentrations normalized to chondriticvalues of Boynton (1984) are shown in Fig 5aAll samples are characterized by LREE-enrichedpatterns with no or slightly positive Euanomalies (EuEu = 097 to 113) La concentra-tions are roughly 20 ndash60 times and Lu concentra-tions are 10 times chondritic values NormalizedLREE abundances from La to Nd are fairlyconstant but the patterns show a steady andsmooth decline from Nd towards heavier REEThe REE patterns of the EF basalts fall in betweenthose of typical OIB and E-MORB (Fig 5b)They have similar patterns and absolute abun-dances compared to basaltic dykes from theIvittuut area South Greenland (Goodenough etal 2002) LaNSmN as a measure of LREEfractionation has a restricted range from 130 to165 The degree of HREE fractionationexpressed as GdNYbN is slightly more variableand ranges from 157 to 257 Both these ratios of

the EF basalts fall generally between typical OIBand E-MORB compositions

Clinopyroxene REE patterns

Typical analyses of REE concentrations inclinopyroxenes are given in Table 3 They showthat REE concentrations in individual minerals ofone particular sample can be quite variableChondrite-normalized REE patterns of clinopyr-oxenes are very similar in the samples from theMussartut and Ulukasik Groups (Fig 6ab) Theyare enriched in all REE relative to chondriticvalues and the patterns show a fairly steepincrease from La to Sm followed by a gradualsmooth decrease from Sm to Lu Clinopyroxenesfrom sample EF 168 (Il otildeacutemaussaq Group) aredistinct as they have generally higher REEcontents especially in the LREE and a weak Euanomaly (Fig 6c) The 3 ndash106 enrichment in Nd

c (sa

mpl

e)c

(prim

itive

man

tle)

c (sa

mpl

e)c

(prim

itive

man

tle)

1

10

100

1000

Mussartucirct GroupUlukasik GroupIliacutemaussaq Group

Rb Ba K Nb La Ce Sr Nd P Zr Ti Y

Rb Ba K Nb La Ce Sr Nd P Zr Ti Y01

1

10

100

1000N-MORBE-MORBOIB

FIG 4 Multi-element plots normalized to primitivemantle values (McDonough and Sun 1995) Data forN-MORB E-MORB and average OIB (Sun and

McDonough 1989) are shown for comparison

PETROGENESIS OF THE ERIKSFJORD BASALTS

839

TA

BL

E3

RE

Eda

taof

who

lero

cks

and

clin

opyr

oxen

es

the

latt

erde

term

ined

byL

aser

ICP

-MS

Who

le-r

ock

anal

yses

R

epre

sent

ativ

ecl

inop

yrox

ene

anal

yses

S

ampl

eno

10

1231

1014

4610

1457

1014

7218

6389

1863

95E

F02

4-2

EF

039-

1E

F03

9-2

EF

039-

4E

F07

2-2

EF

072-

3E

F16

8-3

EF

168-

5

RE

Eco

ncen

trat

ions

(ppm

)L

a11

65

180

610

29

661

155

412

28

440

266

199

150

273

304

165

527

59

Ce

290

346

05

249

617

13

378

130

08

180

411

46

868

717

105

012

60

572

399

67

Pr

434

67

388

268

527

433

331

216

181

128

249

262

102

917

43

Nd

200

631

71

176

412

49

250

920

16

190

213

82

122

28

9815

64

157

258

51

960

6S

m4

756

974

043

215

914

826

664

194

283

276

077

0615

82

265

9E

u1

742

51

571

071

971

721

831

491

111

212

001

943

877

07G

d4

826

584

043

555

84

996

965

615

244

326

156

0514

75

252

7T

b ndash

ndash ndash

ndash ndash

ndash1

351

010

770

711

111

212

313

61D

y4

475

223

613

565

354

816

865

934

784

255

996

8613

10

205

2H

o0

921

040

730

731

11

011

481

401

001

001

211

292

394

24E

r2

432

531

951

972

872

663

893

342

642

112

993

646

539

60T

m ndash

ndash ndash

ndash ndash

ndash0

450

390

360

290

350

440

861

27Y

b2

182

071

741

832

592

363

782

812

492

452

703

016

788

89L

u0

340

310

260

270

390

350

460

410

300

310

520

411

251

60

(Eu

Eu

) N1

111

131

190

971

031

070

820

940

720

991

000

910

770

83

Eu

anom

aly

isca

lcul

ated

as(E

uE

u) N

=E

u N(

SmN6

Gd N

)05

norm

aliz

edto

chon

drit

eva

lues

from

Boy

nton

(198

4)

840

R HALAMA ETAL

in clinopyroxenes of EF 168 compared to theother samples (EF 024 039 and 072) seems onlypartly due to a higher degree of fractionationsincethe Mg value of EF 168 is only slightly lowerThus the LREE enrichment appears to underlinethe more alkaline character of this sample Acomparison with published REE patterns ofclinopyroxenes (Fig 6d) shows that the EFpatterns are unlike patterns from olivine gabbroswith MORB-like signatures (Benoit et al 1996)but fairly similar to clinopyroxene megacrystpatterns from an OIB-like alkaline basalt (Nimisand Vannucci 1995)

Sr Nd and O isotopic composition

Sr Nd and O isotopic compositions of all samplesare presented in Table 4 Sr and Nd isotopiccompositions of the clinopyroxene separates havebeen recalculated to an approximate eruption age

of 12 Ga (Paslick et al 1993) using decayconstants of 142610ndash11 andash1 for 87Rb (Steigerand Jager 1977) and 654610ndash12 a ndash1 for 147Sm(Lugmair and Marti 1978) Present day CHURvalues of 01967 for 147Sm144Nd (Jacobsen andWasserburg 1980) and 0512638 for 143Nd144Nd(Goldstein et al 1984) were used to calculateinitial eNd values 87Sr86Sr initial ratios are fairlyconstant and range from 070278 to 070383(Table 4) eNd(i) values vary from ndash32 to +21(Table 4) On the Sr-Nd isotope diagram the EFbasalts de ne a relatively steep trend (Fig 7) withinitial 87Sr86Sr ratios similar to those of the BulkSilicate Earth (BSE) Only one sample is slightlydisplaced towards a more radiogenic initial87Sr86Sr ratio None of the samples has an eNd

value close to depleted mantle values estimatedto be between eNd = +53 (calculated afterDePaolo 1981) and eNd = +73 (calculated afterGoldstein et al 1984) at 12 Ga The two alkalinebasalts of the Il otilde maussaq Group are within therange of the other data but they de ne a restricted eld around CHUR and BSE respectively Errorsinduced by the uncertainty in the age of the EFbasalts are largely within the error of the isotopeanalyses and do not change the overall picture Incomparison with other published Sr-Nd isotopedata from the Gardar Province the initial Srisotope ratios of the EF basalts are very similar torelatively primitive basaltic lamprophyric andcarbonatitic dyke rocks which have (87Sr86Sr)i of~0703 (Pearce and Leng 1996 Andersen 1997Goodenough et al 2002) So far no evidence hasbeen presented for an isotopicallyenriched mantlesource in the Gardar Province as the existingeNd(i) data of ndash05 to +52 show an isotopicallydepleted signature (Pearce and Leng 1996Andersen 1997 Goodenough et al 2002Upton et al 2003) Clinopyroxene from the EFsamples of Paslick et al (1993) yield eNd(i) of+22 The Nd isotope data of this study are withinthe lower range of published values from theGardar Province and extend the spread to morenegative eNd(i) values

Several recent oxygen isotope studies haveshown that mineral separates are less sensitive tolow-T alteration and secondary water take-up thanwhole-rock samples (eg Vroon et al 2001)However only few oxygen isotope data ofmineral separates from continental volcanics areavailable (eg Dobosi et al 1998 Baker et al2000) Extensive studies of mineral separates ofocean island basalts (Eiler et al 1997) andoceanic-arc lavas (Eiler et al 2000a) demon-

c (sa

mpl

e)c

(cho

ndri

te)

c (sa

mp

le)c

(cho

ndrit

e)

1

10

100

200

101446

101472

101231

101457

186389

186395

La Ce Pr Nd Sm Eu Gd Dy Ho Er Yb Lu

1

10

100

200

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

N-MORBE-MORBOIBBasaltic dykes from Ivittuut

a

b

FIG 5 (a) Whole-rock REE data of EF basaltsnormalized to chondritic values from Boynton (1984)(b) Typical compositions of N-MORB E-MORB andOIB (Sun and McDonough 1989) are shown for

comparison

PETROGENESIS OF THE ERIKSFJORD BASALTS

841

strated that analyses of mineral separates consis-tently span a narrower range than comparablewhole-rock data In our study we used clino-pyroxene because it is the only fresh mineralphase present in the samples Oxygen isotopemeasurements of clinopyroxene separates fromthe EF basalts show a range in d18Ocpx from +52to +60 with an average value of +566plusmn023(1s) All samples overlap the range of d18Ocpx

values from mantle peridotites (+525 to +590Mattey et al 1994) from continental oodbasalts from Yemen (+55 to +69 Baker etal 2000) and from alkaline basalts of Hungary(+507 to +534 Dobosi et al 1998) There is asmall difference between the averages ofMussartut Group (d18Ocpx-avg = +575plusmn015(1s)) and Ulukasik Group (d18Ocpx-avg =+575plusmn009 (1s)) samples compared toIl otildeacutemaussaq Group (d18Ocpx-avg = +533plusmn021(1s)) samples

d18O values of melts coexisting with clino-pyroxene were calculated assuming that oxygen

in pyroxene and melt were in isotopic equilibriumand T was 1100ordmC Dmelt-cpx can then be calculatedusing the following equation of Kalamarides(1986)

Dmelt-cpx = ndash00008T + 143 (T in Kelvin)

Although fractionation varies with magmatictemperatures we consider the temperature of1100ordmC and the resulting Dmelt-cpx of +03 assuf ciently representative considering the analy-tical errors involved It is similar to Dmelt-cpx

values used in previous studies (+039 Vroonet al 2001 +02 Dobosi et al 1998)Resulting d18Omelt values of EF basalts rangefrom +55 to +63 These values can becompared to whole-rock data from continentalbasalts given by Harmon and Hoefs (1995)Intraplate basalts from continental rift zonesshow a range from +46 to +83 with a meanof +61 and continental ood basalts which arerather poorly represented in the database havevalues between +43 and +65 Calculated EF

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Cpx from MORB-like

olivine gabbro

Cpx megacrysts from OIB-like

alkaline basalts

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 024EF 039

Mussartucirct Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 168

Iliacutemaussaq Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

1

10

100

1000

EF 072

Ulukasik Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb LuLa Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

c

ba

d

1

10

100

01

FIG 6 REE patterns of clinopyroxenes (cpx) from Mussartut (a) Ulukasik (b) and Il otilde maussaq (c) Group samplesnormalized to chondritic values from Boynton (1984) (d) REE patterns of cpx from MORB-type olivine gabbro(Benoit et al 1996) and OIB-type alkaline basalt (Nimis and Vannucci 1995) for comparison (note the change of

scale)

842

R HALAMA ETAL

TA

BL

E4

Sr

Nd

and

Ois

otop

icco

mpo

siti

ons

ofcl

inop

yrox

ene

sepa

rate

sfr

omE

riks

fjor

dF

orm

atio

nba

salt

sW

hole

-roc

kan

alys

esfr

omth

eK

etil

idia

nba

sem

ent

and

from

sedi

men

tary

rock

sof

the

Eri

ksfj

ord

For

mat

ion

are

give

nfo

rco

mpa

riso

n

Sam

ple

Mem

ber

Sr

Rb

87R

b86S

r87S

r86Sr

87Sr

86Sr

init

ial

SmN

d147Sm

144N

d143N

d144N

de N

d(i

)d1

8O

(ppm

)(p

pm)

(ppm

)(p

pm)

Bas

alts

E

F02

4M

ussa

rtut

646

50

486

002

180

7042

01plusmn

100

7038

267

3822

80

019

560

5124

67plusmn

10ndash

32

567

EF

063

Mus

sart

ut5

60plusmn

011

EF

061

Mus

sart

ut89

95

099

90

0321

070

3818

plusmn10

070

3266

574

163

90

2115

051

2629

plusmn10

ndash2

55

79plusmn

028

EF

039

Mus

sart

ut73

74

058

30

0229

070

3492

plusmn10

070

3098

662

191

30

2092

051

2626

plusmn10

ndash2

25

71plusmn

001

EF

059

Mus

sart

ut54

47

013

70

0073

070

3375

plusmn12

070

3250

606

168

90

2171

051

2727

plusmn09

ndash1

45

99plusmn

001

EF

108

Ulu

kasi

k66

92

111

00

0480

070

3741

plusmn09

070

2916

109

936

82

018

050

5126

17plusmn

10+

21

581

EF

072

Ulu

kasi

k69

41

032

20

0134

070

3014

plusmn10

070

2784

818

251

90

1962

051

2574

plusmn10

ndash1

25

68E

F17

4Il

otildemau

ssaq

125

42

324

005

360

7039

35plusmn

100

7030

1418

76

796

10

1424

051

2230

plusmn10

+0

45

47E

F16

8Il

otildemau

ssaq

866

71

422

004

750

7037

99plusmn

070

7029

8316

45

641

60

1550

051

2304

plusmn09

ndash0

15

18

Bas

emen

tan

dse

dim

enta

ryro

cks

JG02

Juli

aneh

aEcirc b30

01

169

71

6375

074

4789

plusmn10

071

6647

105

362

28

010

220

5114

70plusmn

09ndash

83

78

Gra

nite

EF

021

Qua

rtzi

te11

5E

F02

8A

rkos

ic6

8ar

enit

e

Sam

ples

are

list

edin

stra

tigra

phic

orde

rfr

omol

dest

toyo

unge

st87

Sr86

Sr

init

ialr

atio

san

de N

d(i)

valu

esw

ere

calc

ulat

edfo

ran

age

of1

2G

aS

tand

ard

devi

atio

nsfo

rd18

Oar

egi

ven

for

sam

ples

that

wer

ean

alys

edtw

ice

PETROGENESIS OF THE ERIKSFJORD BASALTS

843

model melt oxygen isotope data are within theranges of both groups Compared to Igaliko dykerocks from further SE in the Gardar Province(Pearce and Leng 1996) the EF basalts cover amuch more restricted eld possibly because d18Owhole-rock analyses are more prone to secondaryalteration than analyses of mineral separates(Fig 8) Whole-rock d18O measurements of theunderlying JulianehaEcirc b Granite and of the sedi-mentary sandstone members of the EriksfjordFormation yield values of 78 and 68 ndash115respectively (Table 4) and are shown forcomparison

Discussion

Mantle source regions of the EF basaltsTo reconstruct the mantle sources involved in theformation of the EF basalts we use Sr-Nd-Oisotope and incompatible trace element data On aSr-Nd isotope diagram the isotopic compositionof DMM was modelled for an age of 12 Ga(DePaolo 1981 Goldstein et al 1984 Taylor andMcLennan 1985) The fairly constant 87Sr86Srinitial isotope ratios of the EF basalts are ~0703which falls within the range typical for mantle-derived rocks but slightly above calculated valuesfor DMM (Fig 7) eNd(i) values are more

-4

-2

0

2

4

6

8

0701 0702 0703 0704 0705 0706

DMM Typical 2 s error

CHUR

BS

E

(87Sr86Sr)i

e Nd

(i)

EM-1

Qassiarsuk carbonatites and lamprophyres

EF basalts

Depleted MORB mantle

Ivittuut basaltic and lamprophyric dikes

Igaliko carbonatites and lamprophyres

Enriched mantle 1

Carbonatites from other regions of similar age

FIG 7 eNd(i) vs (87Sr86Sr)i diagram calculated for t = 12 Ga which is considered as the eruption age of the EFbasalts (Paslick et al 1993) The reference line of Bulk Silicate Earth (BSE) for the Sr isotopic composition wascalculated after DePaolo (1988) assuming present-day values of 00827 for 87Rb86Sr and 07045 for 87Sr86SrIsotopic composition of the depleted MORB mantle (DMM) was calculated after DePaolo (1981) Goldstein et al(1984) and Taylor and McLennan (1985) EM-1 was calculated after Schleicher et al (1998) (see text for details)Data sources are Goodenough et al (2002) Ivittuut (n = 6) Pearce and Leng (1996) Igaliko (n = 5) Andersen(1997) Qassiarsuk (n = 25) Bell and Blenkinsop (1989) other carbonatites of similar age (Goudini Schryburt Lake

and Prairie Lake n = 6)

844

R HALAMA ETAL

scattered and indicate that the rocks contain amixture of isotopic components from both mantleand crustal reservoirs or a heterogeneous mantlesource Only one sample (EF 108) has an eNd(i)value signi cantly above 0 This sample exhibitsthe largest spread of clinopyroxene compositionswhich may re ect strong alteration Goodenoughet al (2002) also note that the highest eNd occursin the most altered sample of their sample setHowever Paslick et al (1993) report clino-pyroxene data from two Ulukasik Groupsamples with eNd(i) gt +2 and those sampleswere apparently not disturbed with regard to theirSm-Nd age information On the other hand Piperet al (1999) suggest that the Sm-Nd systematicsof those samples may be erroneous due to laterdisturbance by uids derived from the Il otildeacutemaussaqintrusion Therefore although we are presentlynot able to decide whether positive eNd(i) values

for the EF basalts are real it is clear that none ofthose values is close to values of a 12 Ga oldDMM source (ie eNd = +5 to +7)

The Sr-Nd isotopic composition for OIB-typesources in the Proterozoic is not as wellconstrained as DMM Today OIB-type mantlesources have a wide range of isotopic composi-tions that partly overlap with DMM but eNd

values are generally lower (Hart et al 1992Hofmann 1997) The EM-1 mantle component ischaracterized by extremely low 143Nd144Ndratios and eNd values of between 0 and ndash5(Hofmann 1997) A possible 12 Ga position forEM-1 is plotted in Fig 7 calculated according toSchleicher et al (1998) with 147Sm144Nd = 018and 87Rb86Sr = 010 The present-day values usedin the calculation are 143Nd144Nd = 051234 and87Sr86Sr = 07057 taken from ocean islandbasalts closest to an EM-1 end-member(Hofmann 1997) Other Sr-Nd isotope datafrom the Gardar Province are plotted to putfurther constraints on the mantle sources of the EFbasalts (Fig 7) We also added comparative datafrom African and Canadian carbonatitesof similarage because there is a well known similaritybetween Sr-Nd isotopic data of OIBs andcarbonatites (Bell and Blenkinsop 1989) Thelinear array of the carbonatite data is consideredto be the result of mixing between the two mantlecomponents HIMU and EM-1 (Bell andBlenkinsop 1989 Bell and Tilton 2001) Thecarbonatites from the Gardar Province havesimilar Nd but slightly more radiogenic Srisotopic compositions than the reference carbo-natite data (Fig 7) The least contaminatedsamples overlap with the linear array and themore radiogenic Sr isotopic composition can beexplained by crustal contamination (Andersen1997) We therefore consider it possible that theGardar carbonatites also represent mixing ofhypothetical HIMU and EM-1 components of anOIB-like source Since the isotopically mostprimitive EF basalt samples overlap with theQassiarsuk carbonatites it could be possible thatthe basalts also contain such mantle componentsalthough Pb isotope data for the EF basalts thatcould clarify the importance of a HIMU sourceare lacking Based on the Sr-Nd isotope data theEF basalts appear to be derived from aheterogeneous OIB-like mantle source thatcomprises isotopically depleted and enrichedcomponents OIB-like and carbonatitic magma-tism are often attributed to mantle plume activity(Bell and Simonetti 1996 Hofmann 1997

2

4

6

8

10

12

-10 -8 -6 -4 -2 0 2 4 6 8

DMM

d18 O

[permil]

e Nd (i)

Typical2s error

EM-1

EF basalts - cpx

EF basalts - model melt

Ketilidian basement

Igaliko dykes

Depleted MORB mantle

Enriched mantle 1

EM-1 lavas - cpx

FIG 8 Oxygen isotopi c compositio ns (d18OV-SMOW) of clinopyroxene from EF basalts calculatedmelt (see text for details) and whole-rock Ketilidianbasement vs eNd(i) Whole-rock data for Igaliko dykerocks (n = 7) are from Pearce and Leng (1996) DMMdata derived from fresh MORB glasses are from Eiler etal (2000b) The eld of EM-1 lavas is taken fromclinopyroxene data of Gough and Tristan da CunhaIsland lavas (Harris et al 2000) and the EM-1 end-member composition is based on the calculations afterSchleicher et al (1998) for eNd(i) and an average value

of 578 (n = 18) from the EM-1 lavas for d18O

PETROGENESIS OF THE ERIKSFJORD BASALTS

845

Simonetti et al 1998) A plume input to theformation of the EF basalts is supported by therelatively large volume of magmas and the shortduration (lt5 Ma) of magmatism (Piper et al1999) It is also compatible with the geologicalsetting of a continental breakup Additionally aplume component appears to be necessary toexplain the thermal input as it is dif cult to melt acold lithosphere section (Arndt and Christensen1992)

Oxygen isotope data of the EF basalts aregenerally consistent with a mantle derivationincluding an OIB-like mantle source On the d18Ovs eNd(i) diagram (Fig 8) the overlap betweenthe EF data clinopyroxene data from EM-1-likeoceanic basalts and the hypothetical EM-1 end-member suggests an involvement of an EM-1-type mantle source Olivines from HIMU-likesources are known to be slightly lower in d18Othan typical upper mantle values from MORBs(Eiler et al 1997) but the EF data are notsuf cient to assess if the slightly lower d18Ovalues in the Il otildeacutemaussaq Group basalts are causedby a relatively higher input from a HIMU source

The contribution of the various mantle sourcesto the EF basalts can also be evaluated on theprimitive mantle-normalized multi-elementdiagram (Fig 4) and on the REE plots (Figs 5and 6) Both the whole-rock REE and theclinopyroxene REE patterns of the EF basaltsare unlike those representative of N-MORBsources We conclude again that DMM did notplay a major role in the petrogenesis of the EFbasalts On the other hand similarities with OIB-like mantle sources are more prominent Theshape of the clinopyroxene REE pattern closelyresembles those from clinopyroxene of OIB-likealkaline basalts despite differences in absoluteconcentrations The whole-rock REE patterns ofthe EF basalts lie between E-MORB and averageOIB patterns as do the multi-element patterns Onthe latter the EF basalts have marked positive Baand P peaks that are not present in typical OIBsHowever Ba contents should be interpreted withcaution since highly variable ZrBa ratios from014 to 070 are an indication that Ba was re-mobilized after emplacement of the basalts (egPrice et al 1991 Oliveira and Tarney 1995)Likewise P mobility was demonstrated in alteredbasaltic rocks (Winchester and Floyd 1976) Insummary the REE and trace element data areevidence that the mantle source(s) of the EFbasalts were enriched in incompatible traceelements relative to primitive mantle

For relatively primitive hy-normative basalticrocks of the Ivittuut area further north in theGardar Province it has been argued that some ofthem were derived from a lithospheric mantlesource previously enriched by subduction-related processes This interpretation is basedon certain geochemical characteristics such ashigh LaNb ratios and negative Nb anomalies(Goodenough et al 2002) For the EF basaltsnegative Nb anomalies are not clearly discern-ible but LaNb ratios are gt1 which seems to be acharacteristic feature of a lithospheric mantlesource (Fitton 1995) LaBa LaRb and LaKratios were also used to argue for a lithosphericorigin of the Gardar basic magmas although amantle plume involvement was assumed toprovide the required energy (Upton andEmeleus 1987 Upton et al 2003) Sincefeldspar is the only primary magmatic mineralin the EF basalts in which Ba Rb and K areincorporated in signi cant amounts and thefeldspars are generally altered these elementsmight have been easily mobilized and wetherefore do not dare to base the interpretationof the EF basalts sources on these data Inaddition the lack of precise geochemical data onthe composition of the lithosphere below theGardar Province complicates an evaluation ofthe lithospheric contribution to the EF basalticmagmatism

Certain ratios of incompatible trace elements ofthe EF basalts are compared to several occur-rences of ocean island basalts representative ofEM-1 and HIMU sources in Table 5 Ratiosbetween highly incompatible trace elements donot change signi cantly during partial melting orfractional crystallization in basaltic systems andmay therefore constrain the ratios in their sources(Weaver et al 1987) The EF basalts aresubdivided into the more alkaline isotopicallyprimitive Il otildeacutemaussaq Group samples and thosesamples with eNd(i) lt ndash1 The Il otilde maussaq Groupsamples have higher SrNd LaNb and ZrNbratios than HIMU and overlap with the upperrange of EM-1 ratios The EF basalts with eNd(i)lt ndash1 have generally higher ratios than the oceanicisland basalts and partly overlap with typicalvalues of the lower crust It therefore seems likelythat assimilation of lower crustal material mayhave changed the primary trace element ratios inthe basalts Therefore we will evaluate thepossible in uence of assimilation of crustalmaterial on the geochemistry of the EF basaltsin the following section

846

R HALAMA ETAL

The role of crustal contamination

Large amounts of bulk assimilation of crustalmaterial which is relatively SiO2-rich comparedto a basaltic melt would drive the compositions tohigher SiO2 contents and quartz-normativecompositions Partial melts derived from crustalrocks would also be relatively SiO2-rich (Beard etal 1993 Carroll and Wyllie 1990) and wouldhave the same effect The ne-normative characterof the EF basalts (Table 1) suggests thatassimilation of crustal material especially ofsilica-rich upper crustal rocks was not signi cantwith the possible exception of sample EF 059

On the primitive mantle-normalized multi-element diagrams (Fig 4) Ti troughs are typicalfor crustal rocks especially for the Upper Crust(Rudnick and Fountain 1995) and they maytherefore be present in rocks extensively contami-nated by continental crust Ti troughs aregenerally lacking in the patterns of the EFbasalts and a major upper crustal contribution tothe magmas can therefore be excluded However

small amounts of assimilation of lower crustalmaterial cannot be ruled out because the averagelower crust has higher Ti contents

Steep trends on Sr-Nd isotope plots have beenattributed to assimilation of granulite-faciesgneisses (eg Bernstein et al 1998) AFCmodelling based on an equation developed byDePaolo (1981) was performed to test thehypothesis of small amounts of contaminationwith lower crustal material (Fig 9) The basalticmagma which represents one end-member for thecalculations has the isotopic signature of theisotopically most primitive samples from thepresent data set and from the literature(Table 6) Two possible contaminants repre-senting upper and lower crustal compositionswere selected for modelling (Table 6) Therepresentative samples are a granite from theJulianehaEcirc b batholith for the upper crust (sampleJG 02) and an average of ve granulite-faciesgneisses with low RbSr ratios from the Archaeancraton of West Greenland (Taylor et al 1984) for

TABLE 5 Trace element ratios of EF basalts compared to ocean island basalts(OIBs) representing EM-1 and HIMU mantle components and the lowercrust

SrNd LaNb ZrNb

EF basalts with eNd lt ndash 1 249 ndash396 120 ndash150 121 ndash157Il otildeacutemaussaq Group EF basalts 198 ndash292 104 ndash119 51 ndash53EM-1 90 ndash207 067 ndash111 39 ndash109HIMU 45 -163 063 ndash080 27 ndash 36Lower continental crust 30 16 136

Data sources EM-1 Humphris and Thompson (1983 Walvis Ridge) and Cliff et al(1991 Inaccessible Islands) HIMU Chauvel et al (1992 Tubuai Island) Lowercontinental crust Rudnick and Fountain (1995)

TABLE 6 End-member compositions used for AFC modelling

End-member Sr (87Sr86Sr)i Nd eNd(i) Data source(ppm) (T = 12 Ga) (ppm) (T = 12 Ga)

Primitive EF basaltic 507 070278 18 + 22 Sample EF 072magma Paslick et al (1993)Upper crust 300 071665 62 ndash83 Sample JG 02(JulianehaEcirc b Granite)Lower crust 747 070668 63 ndash 234 Taylor et al (1984)(Granulite facies gneiss)

The composition of the granulite-facies gneiss is an average of 5 samples with low RbSr ratios lt02

PETROGENESIS OF THE ERIKSFJORD BASALTS

847

the lower crust An in uence of Archaean rockson Gardar magmas has been demonstrated by Pbisotope studies (Taylor and Upton 1993) andseismic data were also interpreted to re ect theexistence of a wedge of Archaean crust thatextends southwards to Lindenow Fjord (Dahl-Jensen et al 1998) We have shown in theprevious sections on the basis of Ni trends thatfractionation of olivine was most relevant to themajor element evolution of the lavas whereasfractionation of other phases was apparently ofminor importance We therefore modelled theAFC processes with olivine fractionation usingmineral-melt partition coef cients of 0014 for Srand 00066 for Nd (Rollinson 1993)

The modelling of AFC processes is stronglydependent on the parameter r de ned as the ratioof mass assimilation rate and fractional crystal-lization rate (DePaolo 1981) The change inisotopic signature becomes enhanced for highervalues of r Reiners et al (1995) have shown thatr values might be very high and even gt1 in theearly stages of magmatic evolution at hightemperatures since crystallization can besuppressed by assimilation Therefore we calcu-

lated AFC processes with two r values of 05 and08 that represent a relatively high and a middle tolow assimilation rate to demonstrate the effects ofvariable r (Fig 9) The results indicate that AFCprocesses involving upper crust similar to theJulianehaEcirc b Granite cannot explain the trend of theEF basalts because Sr isotopic compositionswould shift to considerably more radiogenicvalues This result is consistent with theconclusions derived from trace element dataThe AFC curves for assimilation of lower crusthowever agree well with the bulk of the data Themost negative eNd values can be explained byAFC processes with intermediate assimilationrates (r = 05) involving assimilation of lt5lower crustal material With a greater degree ofassimilation and an r value of 08 lt4assimilation is indicated

Lower crustal rocks of igneous origin have anaverage d18O value of +75plusmn14 with an overallrange from +54 to +125 (Fowler and Harmon1990) 5 to 10 bulk assimilation of lower crustalmaterial with d18O = +75 into a typical mantle-derived magma with d18O = +57 results ind18O values of the contaminated magma from

-24

-20

-16

-12

-8

-4

0

4

0700 0703 0706 0709 0712 0715 0718

r = 08

r = 05

r = 05

r = 08

EF Basalts

Endmembers of AFC calculations

Upper Crust

Lower Crust

Primitive EF basaltic magma

(87Sr86Sr)i

e Nd

(i)

09

08

06

09

08

06

09

0806

09

08

07

07

04

Typical errorCHUR

BS

E

FIG 9 eNd(i) vs (87Sr86Sr)i diagram showing the results of AFC modelling End-member compositions are listed inTable 6 Olivine is the only fractionating mineral Partition coef cients for olivinemelt are from the compilation ofRollinson (1993) Curves were calculated for two values of r (r = rate of assimilation rate of fractionalcrystallization) Dashed lines and numbers in italics are for r = 05 solid lines and regular numbers are for r = 08

Values at the curves are for F values (fraction of melt remaining) with 10 tick marks

848

R HALAMA ETAL

+579 to +588 Therefore the d18O data cannotbe used to rule out small amounts of assimilationof lower crust To evaluate how much contamina-tion would be permissible from the presentoxygen isotope data we calculated how muchcontamination of a mantle-derived magma with18O = +57 with a contaminant within the lowerrange of lower crustal xenoliths (d18O = +67)would drive the oxygen isotopic composition tothe highest value of the EF basalts (d18O~+63) Although simple mixing shows that60 assimilation is permissible this amountwould drive the whole-rock geochemistry tomore silicic compositions than is observed in theEF basalts In Fig 8 there is no obvious trendfrom the basalts towards the Ketilidian basementsuggesting again that upper crust was notassimilated in signi cant amounts

Ratios of trace elements incompatible inbasaltic systems can be used to distinguishbetween crust and mantle reservoirs (egWeaver 1991) Based on the AFC modellingEF basalts with eNd(i) lt ndash1 are those which aremost likely in uenced by crustal contaminationThe relatively high LaNb and ZrNb ratios ofthese samples are consistent with assimilation oflower crustal material SrNd ratios are evenhigher than the average lower crustal composi-tion but SrNd ratios in those samples which weretaken as representative for the lower crust in theAFC modelling can reach values of 80 ie wellabove the highest EF basalt ratios Again theSrNd ratios support the hypothesis of some lowercrustal contamination

The absence of more primitive basaltic rockswith MgO gt8 wt might also indicate lowercrustal assimilation It suggests that magmas mayhave been retained at depth where they acquiredmore evolved compositions by fractional crystal-lization and had the opportunity to assimilatelower crustal material during crustal underplating(Upton et al 2003)

Summary and conclusions

The new geochemical data of the EriksfjordFormation basalts in the Gardar Provinceprovided in this study are used to placepetrogenetic constraints on the sources of theirparental magmas Based on trace-element patternsand Sr-Nd isotopic constraints there seems to beno indication that a DMM-like source played animportant role in magma genesis of the EFbasalts Clinopyroxene and whole-rock REE

patterns are more compatible with an OIB-likesource but we are not able to conclude whetherthe observed enrichment in REE re ects thesourcersquos signature or magmatic processes Theisotopically most primitive EF basalts have Sr-Ndisotopic compositions that are similar to Gardarcarbonatites Due to the isotopic similaritybetween carbonatites and OIBs many arethought to be related to mantle plumes (Bell2001) We therefore speculate that OIB-likemantle components from a plume source werealso involved in the parental EF magmas but wecannot exclude interaction with the SCLM Twodistinct mantle components one isotopicallydepleted and another one enriched could havecontributed to the EF magmas AFC modelling ofSr-Nd isotope data indicates that assimilation oflt5 lower crust could explain the steep vector onthe Sr-Nd isotope plot but assimilation of uppercrustal material is unlikely Oxygen isotope dataof clinopyroxene separates are consistent with amantle origin but permit some assimilation oflower crustal rocks In summary the data areconsistent with the hypothesis that the EFmagmas could have originated from a mantleplume source which is supported by geologicalarguments and that they underwent someassimilation of lower crustal material

Acknowledgements

We would like to thank Bruce Paterson whoprovided invaluable help during Laser ICP-MSmeasurements at the Large-Scale GeochemicalFacility supported by the European Community-Access to Research Infrastructure action of theImproving Human Potential Programme contractnumber HPRI-CT-1999-00008 awarded to ProfBJ Wood (University of Bristol) Gabi Stoschekand Torsten Vennemann are thanked for their helpwith oxygen isotope measurements Elmar Reitterexpertly assisted with preparation and measure-ments of radiogenic isotopes and MathiasWestphal helped with microprobe measurementsREE whole-rock analyses were kindly providedby the NERC facility at Royal HollowayUniversity of London Michael Marks is thankedfor his pleasant company during eldwork Veryconstructive reviews by Tom Andersen andAlexei S Rukhlov greatly helped to improve themanuscript We also thank Ian Coulson for nalediting Funding of this work by the DeutscheForschungsgemeinschaft (grant Ma-21351-2) isgratefully acknowledged

PETROGENESIS OF THE ERIKSFJORD BASALTS

849

References

Andersen T (1997) Age and petrogenesis of theQassiarsuk carbonatite-alkaline silicate volcaniccomplex in the Gardar rift South GreenlandMineralogical Magazine 61 499 ndash513

Armstrong JT (1991) Quantitative elemental analysisof individual microparticles with electron beaminstruments Pp 261 ndash315 in Electron ProbeQuantitation (KFJ Heinrich and DE Newburyeditors) Plenum Press New York and London

Arndt NT and Christensen U (1992) The role oflithospheric mantle in continental ood volcanismThermal and geochemical constraints Journal ofGeophysical Research 97 10967 ndash 10981

Baker JA Menzies MA Thirlwall MF andMacpherson CG (199 7) Pet rogenes is ofQuaternary Intraplate Vocanism Sanarsquoa YemenImplications for Plume-Lithosphere Interaction andPolybaric Melt Hybridization Journal of Petrology 38 1359 ndash 1390

Baker JA MacPherson CG Menzies MAThirlwall MF Al-Kadasi M and Mattey DP(2000) Resolving crustal and mantle contributions tocontinental ood volcanism Yemen constraintsfrom mineral oxygen isotope data Journal ofPetrology 41 1805 ndash1820

Beard JS Abitz RJ and Lofgren GE (1993)Experimental melting of crustal xenoliths fromKilbourne Hole New Mexico and implications forthe contamination and genesi s of magmasContributions to Mineralogy and Petrology 11588 ndash102

Bell K (2001) Carbonatites relationships to mantle-plume activity Geological Society of AmericaSpecial Paper 352 267 ndash290

Bell K and Blenkinsop J (1989) Neodymium andStrontium Isotope Geochemistry of Carbonatites Pp278 ndash300 in Carbonatites ndash Genesis and Evolution(K Bell editor) Unwin Hyman London

Bell K and Simonetti A (1996) Carbonat iteMagmatism and Plume Activity Implications fromthe Nd Pb and Sr Isotope Systematics of OldoinyoLengai Journal of Petrology 37 1321 ndash1339

Bell K and Tilton GR (2001) Nd Pb and Sr IsotopicCompositions of East African Carbonat itesEvide nce for Mant le Mixing and PlumeInhomogene ity Journa l of Petrology 42 1927 ndash1945

Benoit M Polve M and Ceuleneer G (1996) Traceelement and isotopic characterization of ma ccumulates in a fossil mantle diapir (Oman ophiolite)Chemical Geology 134 199 ndash 214

Bernstein S Kelemen PB Tegner C Kurz MDBlusztajn J and Kent Brooks C (1998) Post-breakup basaltic magmatism along the EastGreenland Tertiary rifted margin Earth and

Planetary Science Letters 160 845 ndash862Boynton WV (1984) Geochemistry of the rare earth

elements meteorite studies Pp 63 ndash114 in RareEarth Element Geochemistry (P Henderson editor)Elsevier Amsterdam

Carroll MR and Wyllie PJ (1990) The systemtonalite-H2O at 15 kbar and the genesis of calc-alkaline magmas American Mineralogist 75345 ndash 357

Chauvel C Hofmann AW and Vidal P (1992)HIMU-EM The French Polynesian connectionEarth and Planetary Science Letters 110 99 ndash119

Clayton RN and Mayeda TK (1963) The use ofbromine penta uoride in the extraction of oxygenfrom oxides and silicates for isotope analysisGeochimica et Cosmochimica Acta 27 43 ndash52

Coulson IM and Chambers AD (1996) Patterns ofzonation in rare-earth-bearing minerals in nephelinesyenites of the North Qoroq center South GreenlandThe Canadian Mineralogist 34 1163 ndash1178

Cox KG Bell JD and Pankhurst RJ (1979) TheInterpretation of Igneous Rocks George Allen andUnwin London

Cross W Iddings JP Pirsson LV and WashingtonHS (1903) Quantitative Classi cation of IgneousRocks University of Chicago Press

Dahl-Jensen T Thybo T Hopper H and Rosing M(1998) Crustal structure at the SE Greenland marginfrom wide-angle and normal incidence seismic dataTectonophysics 288 191 ndash 198

DePaolo DJ (1981) Trace element and isotopic effectsof combined wallrock assimilation and fractionalcrystallisation Earth and Planetary Science Letters53 189 ndash 202

De P a o lo D J (1 9 8 8 ) Neodymium Iso topeGeochemistry An Introduction Springer-VerlagNew York

Dobosi G Downes H Mattey D and Embey-IsztinA (1998) Oxygen isotope ratios of phenocrysts fromalkali basalts of the Pannonian basin Evidence foran O-isotopically homogeneous upper mantle be-neath a subduction-in uenced area Lithos 42213 ndash 223

Eiler JM (2001) Oxygen isotope variations of basalticlavas and upper mantle rocks Pp 319 ndash364 inStable Isotope Geochemistry (JW Valley and DRCole editor s) Reviews in Minera logy andGeochemistry 43 The Mineralogical Society ofAmerica Washington DC

Eiler JM Farley KA Valley JW Hauri E CraigH Hart SR and Stolper EM (1997) Oxygenisotope variations in ocean island basalt phenocrystsGeochimica et Cosmochimica Acta 61 2281 ndash2293

Eiler JM Crawford A Elliot T Farley KAValley JW and Stolper EM (2000a) Oxygenisotope geochemistry of oceanic-arc lavas Journal

850

R HALAMA ETAL

120plusmn003 Ga for two samples of the UlukasikGroup However palaeomagnetic data indicate anage of 135 ndash131 Ga for the entire succession(Piper et al 1999) and the exact age of the EF isas yet unresolved (Upton et al 2003) Parts of theEriksfjord lavas have been subjected to extensivemetasomatism related to the intrusion of alkalinemagmas (Ranlov and Dymek 1991 Coulson andChambers 1996) The local presence of epidoteimplies thermal metamorphism at temperatures ofup to 300 ndash350ordmC (Piper et al 1999) Higheruption rates are inferred for the volcanic systemand the whole lifespan is considered to belt~5 Ma (Piper et al 1999)

Analytical procedures

Electron microprobe analysisMineral compositions were determined using aJEOL 8900 electron microprobe (EMP) at theInstitut fur Geowissenschaften UniversitatTubingen Germany An internal jrz correctionof the raw data was applied (Armstrong 1991)Both natural and synthetic standards were usedfor major and minor elements Measuring times

were 16 s and 30 s on the peak positions formajor and minor elements respectively Theemission current was 15 nA and the accelerationvoltage 15 kV For feldspar analyses a beamdiameter of 5 mm was used to avoid errorsresulting from diffusion of Na

XRF analysis

Whole-rock analyses were performed by standardX-ray uorescence (XRF) techniques at theInstitut fur Geowissenschaften at the JohannesGutenberg-Universitat Mainz Germany using aPhilips PW 1404 spectrometer and at theUniversitat Freiburg using a Philips PW 2404spectrometer Prior to analysis whole-rocksamples were crushed and milled in tungstencarbide mills The powders obtained were dried at105ordmC Loss on ignition (LOI) determinationswere carried out on 1 g of pulverized samplematerial which was heated for 2 h at 1000ordmCPressed powder pellets and fused glass discs wereprepared to measure contents of trace and majorelements respectively Natural standards wereused for calibration Detection limits vary

FIG 1 Map of the Eriksfjord Formation on the Narssaq peninsula with sample locations (after Poulsen 1964 andLarsen 1977) Sample 101472 (not shown) is from a succession ~2 km south of the settlement of Narsarsuaq further

to the NE

PETROGENESIS OF THE ERIKSFJORD BASALTS

833

between 1 and 9 ppm depending on the speci ctrace element and on the instrument used

Whole-rock REE analyses

Whole-rock REE analyses were carried out by theNatural Environment Research Council (NERC)inductively coupled plasma-atomic emissionspectrometry (ICP-AES) facility at RoyalHolloway University of London Solutions wereprepared using a combined HF dissolution andalkali fusion as described in Walsh et al (1981)Resultant solutions were analysed for 12 REEsand selected potential interferences using a PerkinElmer Optima 3300RL ICP-AES

Laser ICP-MS in situ REE measurement of clinopyroxene

In situ laser ablation inductively coupled plasma-mass spectrometer (LA-ICP-MS) REE analyses on~150 mm thick polished thin sections wereperformed at the EU Large-Scale GeochemicalFacility (University of Bristol) using a VGElemental PlasmaQuad 3 + S-Option ICP-MSequipped with a 266 nm Nd-YAG laser (VGMicroProbe II) The laser beam diameter at thesample surface was ~20 mm All measurementswere made using Thermo Elemental PlasmaLablsquotime-resolved analysisrsquo (TRA) data acquisitionsoftware with a total acquisition time of 100 s peranalysis allowing ~40 s for background followedby 50 s for laser ablation NIST 610 glass was usedfor instrument calibration and NIST 612 was usedas a secondary standard Ca was used as an internalstandard to correct the ablation yield differencesbetween and during individual analyses on bothstandards and samples To avoid analyticaluncertaintiesdue to variations in the concentrationsof the internal standard Ca concentrations werequantitatively measured within 20 mm of the laserablation pits using the EMP at the Institut furGeowissenschaften Universitat Tubingen Theprecision of trace-element concentrations basedon repeated analyses of standards is ~plusmn5 forelement concentrations gt10 ppm and plusmn10 forconcentrations lt10 ppm Data processing wascarried out of ine using the same PlasmaLabsoftware used for data collection and variouscustom-designed Excel spreadsheets The limitsof detection are de ned as 328 standard deviationsabove background level which is equal to a 95con dence that the measured signal is signi cantlyabove background Typical detection limits for theREE in this study were 004 ndash06 ppm

Sr and Nd isotope analysisFor Sr and Nd isotope analyses 15 ndash20 mgsamples of hand-picked clinopyroxene werespiked with mixed 84Sr ndash 87Rb and 150Nd ndash 149Smtracers before dissolution under high pressure inHF at 180ordmC in poly-tetrauor-ethylene (PTFE)reaction bombs The Rb and Sr were separated inquartz columns containing a 5 ml resin bed ofAG50W-X12 200ndash400 mesh equilibrated with25 N HCl The Sm and Nd separation wasperformed in quartz columns using 17 mlTe on powder coated with HDEHP (Di-EthylHexyl Phosphate) as the cation exchange mediumequilibrated with 018 N HCl All analyses weremade with a multicollector Finnigan MAT 262thermal ionization mass spectrometer (TIMS) instatic collection mode at the Institut furGeowissenschaften Universitat Tubingen TheSr was loaded with a Ta-HF activator andmeasured on a single W lament The Rb Smand Nd were measured in a double Re- lamentcon guration mode 87Sr86Sr and 143Nd144Ndratios were normalized for mass fractionation to86Sr88Sr = 01194 and 146Nd144Nd = 07219respectively The average 143Nd144Nd ratioobtained for the Ames Nd-standard (GeologicalSurvey of Canada Roddick et al 1992) was0512119plusmn10 (n = 42) and the average 87Sr86Srratio for the NBS 987 Sr-standard was0710261plusmn16 (n = 30) during the course of thisstudy 143Nd144Nd ratios have been cross-checked with the La Jolla Nd-standard whichgave 0511831plusmn30 (n = 12) Total proceduralblanks (chemistry and loading) were lt200 pg forSr and lt100 pg for Nd

Oxygen isotope analysis

The oxygen isotope composition of hand-pickedclinopyroxene separates was measured using amethod similar to that described by Sharp (1990)and Rumble and Hoering (1994) 1 ndash2 mg ofhand-picked clinopyroxenewere loaded onto a Pt-sample holder The sample chamber was pumpedout to a vacuum of ~10 ndash6 mbar and pre- uorinated overnight Samples were then heatedwith a CO2-laser in an atmosphere of 50 mbar ofpure F2 Excess F2 is separated from O2 byconversion to Cl2 using KCl held at 150ordmC Theextracted O2 is collected on a molecular sieve(13X) Oxygen isotopic compositions weremeasured on a Finnigan MAT 252 mass spectro-meter at the Institut fur GeowissenschaftenUniversitat Tubingen The results are reported as

834

R HALAMA ETAL

the per mil deviation from Vienna Standard MeanOcean Water (V-SMOW) in the standard d-notation Replicate oxygen isotope analyses ofthe NBS-28 quartz standard (Valley et al 1995)had an average precision of plusmn01 for d18O Ineach run standards were analysed at thebeginning and the end of the sample set Acorrection was applied to the data equal to theaverage of the difference between the meanmeasured value and the accepted value for thestandard (964)

Oxygen isotope compositions of powderedwhole-rock samples were determined by aconventional method modi ed after Clayton andMayeda (1963) using BrF5 as reagent andconverting the liberated oxygen to CO2 beforemass spectrometric analysis

Petrography

175 samples were collected from all threevolcanic units of the Eriksfjord Formation butmost of them were extensively altered Theycontained secondary chlorite Fe-talc Ca-Fesilicates and carbonate as alteration productsSince the focus of this study was in situ analysesof minerals or analyses of mineral separates onlynine samples with a signi cant proportion of freshclinopyroxene and relatively few alterationproducts were selected for analysis Due to thelimited number of suitable samples we did notattempt to investigate the lavas in terms of theirstratigraphic position In addition to the ninesamples from which pyroxene separates wereobtained REE whole-rock data of six othersamples were added to the data set

In general the Mussartut Group lavas areolivine-free but contain up to 50 chloriticmatrix that probably results from devitri cationof former glass In contrast the majority ofUlukasik and Il otilde maussaq Group lavas containolivine-pseudomorphs and tend to be coarsergrained and holocrystalline Many of themcontain lsquosnow- akersquo glomerocrysts of plagioclaseindicating plagioclase-saturation or super-satura-tion at near-surface levels (Upton 1996) Thesamples from which clinopyroxenes have beenseparated are ne-grained basalts with plagioclaselaths embedded in interstitial clinopyroxeneas themajor and apatite and Fe-Ti oxides as subordinateprimary magmatic mineral phases The texturecan be described as ophitic and is believed toresult from the cotectic crystallization of plagio-clase and clinopyroxene(Philpotts 1990) Olivine

has been present in a few samples but it is nowcompletely altered Pyrite pyrrhotite and chalco-pyrite occur as accessory phases Some samplescontain two oxides (ilmenite + titanomagnetite)whereas others contain a single oxide exsolvedinto ilmenite and titanomagnetite or ilmenite andhematite Texturally they are mostly subhedral toanhedral although few euhedral grains occurHematite is also present around both discrete andexsolved ilmenite and titanomagnetite grains andthe exsolved hematite might be entirely ofsecondary origin formed by oxidation of titano-magnetite Late-stage interstitial quartz occursrarely Sample EF 108 (Ulukasik Group) isremarkable in having an intense reddish colourunlike the other samples that are almost black

Results

Mineral chemistryTypical analyses of clinopyroxenes are given inTable 1 Compositions range from En41Wo44Fs16

to En28Wo45Fs27 and broadly overlap for allgroups (Fig 2) According to the classi cation ofMorimoto et al (1988) clinopyroxenes of theMussartut and Ulukasik Groups are mainlyaugites with few diopsides present whereas inthe Il otilde maussaq Group samples of diopsidiccompositions dominate over augitic onesClinopyroxene analyses from the varioussamples form tight clusters on the pyroxenequadrilateral except for those from sample EF108 that are more scattered With only a singlepyroxene phase present in all samples clino-pyroxene compositions indicate minimum crystal-lization temperatures (Lindsley 1983) of750 ndash1000ordmC (Fig 2) Plagioclase is normallyzoned and its composition typically variesbetween An70Ab29Or1 and An16Ab72Or12 Lateinterstitial albite can reach compositions ofAn0Ab99Or1 There are no signi cant composi-

1200 Ecirc C

800 Ecirc C

25 50 75

25 25

En Fs

HdDi

Mussartucirct Group

Iliacutemaussaq Group

Ulukasik Group

FIG 2 Clinopyroxene analyses of the Eriksfjord basaltsplotted in the pyroxene quadrilateral with isotherms

after Lindsley (1983)

PETROGENESIS OF THE ERIKSFJORD BASALTS

835

tional changes among the various samplesIlmenite from two samples (EF 024 and EF 039)which contain coexisting ilmenite and titanomag-netite includes a signi cant component ofpyrophanite (MnTiO3) with up to 87 wtMnO It s com pos i t i ons va r y be t weenI l m 8 6 H e m 5 P y r 8 a nd I l m 7 6 H e m 5 P yr 1 9 Titanomagnetite analyses were unsatisfactory asall grains show strong alteration They areconsidered unreliable and are not presented here

Whole-rock geochemistry and classif|cation

Concentrations of major and trace elements aresummarized in Table 2 The basalts can beclassi ed according to the TAS diagram (LeMaitre et al 1989) as basalts and trachybasaltswith both alkaline and subalkaline compositions(not shown) Due to possible mobility of the Naand K the samples were also plotted into NbYvs ZrP2O5 and P2O5 vs Zr diagrams (Floyd andWinchester 1975) which can discriminate

between tholeiitic and alkaline series Thesediagrams do not yield consistent discriminationsfor the EF basalts but indicate that the characterof the sample suite is transitional betweentholeiitic and alkaline (not shown)

Mg [100Mg(Mg+Fe2+)] calculated assumingFe2O3FeO = 02 as recommended for basalts byMiddlemost (1989) varies between 54 and 40(Table 2) Ni concentrations range from 46 to115 ppm and Cr contents from 29 to 126 ppmTaking Mg as fractionation index a positivecorrelation (r2 = 061) with Ni contents indicatesfractionation of olivine (Fig 3) The relativelypoor positive correlations of Mg with Cr (r2 =044) and Sc (r2 = 023) argue against a signi cantfractionation of Cr-spinel or clinopyroxene Thisis consistent with the lack of clinopyroxenephenocrysts and the absence of any Cr-spinelThe negative correlations of Mg with Sr TiO2

and P2O5 suggest that fractionation of plagioclaseFe-Ti oxides and apatite was not important in thesample set investigated which is again consistent

TABLE 1 Typical clinopyroxene analyses of Eriksfjord Formation basalts determined by EMP

Sample EF 024 EF 024 EF 039 EF 059 EF 072 EF 108 EF 168 EF 174Cpx no cpx 2 cpx 3 cpx 5 cpx 3 cpx 5 cpx 7 cpx 4 cpx 1

SiO2 5060 5041 5128 4988 5145 4950 5175 5069TiO2 173 176 121 169 140 162 096 121Al2O3 297 300 179 223 171 264 155 213FeO 1160 1068 1218 1405 993 1234 979 1129MnO 028 021 030 040 019 028 025 031MgO 1254 1324 1263 1182 1325 1163 1307 1214CaO 1959 2029 2002 1907 2106 2099 2174 2143Na2O 034 031 029 049 033 055 052 054

Total 9965 9989 9971 9963 9932 9955 9962 9974

Formulae based on 4 cations and 6 oxygensSi 191 189 194 190 194 188 194 191Al 013 013 008 010 008 012 007 009Ti 005 005 003 005 004 005 003 003Fe3+ - - - 003 - 007 003 005Mg 071 074 071 067 075 066 073 068Fe2+ 037 033 039 041 031 032 028 031Mn 001 001 001 001 001 001 001 001Ca 079 082 081 078 085 085 087 087Na 002 002 002 004 002 004 004 004

Total 400 400 400 400 400 400 400 400

Pyroxene projection after Lindsley (1983)Wo 039 041 041 041 043 043 045 044En 040 041 039 037 040 038 040 038Fs 021 018 021 023 017 019 015 017

836

R HALAMA ETAL

TABLE 2 XRF whole-rock analyses of Eriksfjord Formation basalts

Sample EF 024 EF 063 EF 061 EF 039 EF 059 EF 108 EF 072 EF 174 EF 168Group Mussartut Mussartut Mussartut Mussartut Mussartut Ulukasik Ulukasik Ilotildeacutemaussaq IlotildeacutemaussaqSubunit 4 4 5 6 7 Lower Unit Upper Unit Sermilik Narssaq

Fjeld

Major elements (wt)SiO2 4573 4585 4688 4660 4875 4698 4656 4601 4545TiO2 230 200 180 175 191 265 203 299 277Al2O3 1578 1630 1600 1612 1645 1682 1668 1645 1719Fe2O3 1510 1356 1331 1383 1225 1394 1321 1438 1392MnO 028 018 018 018 017 014 017 018 017MgO 525 683 653 673 568 396 647 440 480CaO 842 849 872 857 920 641 808 746 737Na2O 349 274 289 276 236 448 303 323 376K2O 087 083 068 032 027 159 070 205 132P2O5 041 035 023 023 023 048 035 111 083LOI 199 210 240 238 242 204 210 052 156

Total 9962 9923 9962 9947 9969 9949 9938 9878 9914

Mg 453 545 538 536 524 403 538 421 450

Trace elements (ppm)Sc 24 22 25 22 26 19 20 21 16V 209 175 215 195 211 142 170 164 203Cr 81 108 66 65 126 48 82 33 29Co 63 78 70 62 71 44 68 35 46Ni 99 115 85 85 101 58 106 46 51Cu 65 49 46 60 97 41 59 38 41Zn 103 102 110 97 110 93 82 122 105Ga 20 18 21 19 20 19 16 21 19Rb 24 26 16 lt 10 lt 10 38 12 39 45Sr 504 503 512 392 373 712 507 850 992Y 31 26 28 23 28 31 17 31 25Zr 188 165 121 116 125 125 88 164 139Nb 12 11 10 lt 10 10 9 lt 10 31 27Ba 445 375 415 302 178 582 405 1293 900Pb 6 7 3 4 3 4 4 3 5La ndash ndash 15 ndash 12 11 ndash 37 28Ce ndash ndash 40 ndash 38 39 ndash 92 71Pr ndash ndash 1 ndash 1 2 ndash 9 4Nd ndash ndash 16 ndash 15 18 ndash 43 34Sm ndash ndash 6 ndash 7 6 ndash 5 5

CIPW normq ndash ndash ndash ndash 313 ndash ndash ndash ndashor 514 491 402 189 160 940 414 1212 780ab 2807 2319 2445 2335 1997 3207 2564 2733 2981an 2482 2973 2868 3065 3350 2109 2985 2433 2613ne 079 ndash ndash ndash ndash 316 ndash ndash 109di 1187 830 1076 853 877 633 654 452 416hy ndash 890 1059 1650 2217 ndash 1097 315 ndashol 1670 1309 1044 780 ndash 1472 1120 1388 1689mt 365 328 322 335 296 336 319 348 336il 437 380 342 332 363 503 386 568 526ap 095 081 053 053 053 111 081 257 192

Mg was calculated as Mg = 100Mg2+(Mg2++Fe2+) using Fe2O3FeO = 02 Normative mineral compositionswere determined according to the CIPW calculation scheme

PETROGENESIS OF THE ERIKSFJORD BASALTS

837

1 5

2 0

2 5

3 0

3 5

40

60

80

100

120

0

0 5

1

1 5

200

400

600

800

1000

0

50

100

150

15

20

25

30

4 0 45 50 5 5

r2 = 023

r2 = 044r2 = 061

r2 = 053

r2 = 082 r2 = 052

Ni (ppm)

Sc (ppm)

P2O5 (wt)

Sr (ppm)

TiO2 (wt)

Cr (ppm)

Mg Mg

Mg Mg

Mg

40 4 5 50 55

Mg

40 45 50 55

40 45 50 55

40 45 50 5540 4 5 50 55

FIG 3 Whole-rock variation diagrams for EF basalts with Mg as fractionation index Mg is calculated asMg(Mg+Fe2+) with Fe2O3FeO = 02 (Middlemost 1989)

838

R HALAMA ETAL

with the mostly subhedral Fe-Ti oxides and theophitic texture Ratios of incompatible traceelements such as ZrY (range 40 ndash63) andTiY (range 385 ndash716) are relatively uniformthroughout all samples but some scatter isevident Notable geochemical features are highSr contents up to 1000 ppm and very high Bacontents up to 1300 ppm

Normative mineral compositions were deter-mined according to the CIPW calculation scheme(Cross et al 1903 Cox et al 1979) Mostsamples are hypersthene-normative some arenepheline-normative and one (EF 059) is quartz-normative (Table 2) Despite the uncertaintiesinvolved in the CIPW norm calculation the dataindicate that some of the basalts crystallized fromsilica-undersaturated magmas and others fromdistinct silica-saturatedor -oversaturatedmagmas

Multi-element diagrams normalized to primi-tive mantle values show that the EF basalts areenriched in all incompatible elements relative toprimitive mantle values (Fig 4) The patterns ofthe EF basalts are characterized by a prominentBa peak a subordinate P peak and a decreasingelement enrichment with increasing compatibilityIn comparison with normal MORB (N-MORB)enriched MORB (E-MORB) and average OIB(Sun and McDonough 1989) the patterns of theEF basalts are closest to those of OIB both inshape and normalized concentrations

Whole-rock REE data

Whole-rock REE data are presented in Table 3and REE concentrations normalized to chondriticvalues of Boynton (1984) are shown in Fig 5aAll samples are characterized by LREE-enrichedpatterns with no or slightly positive Euanomalies (EuEu = 097 to 113) La concentra-tions are roughly 20 ndash60 times and Lu concentra-tions are 10 times chondritic values NormalizedLREE abundances from La to Nd are fairlyconstant but the patterns show a steady andsmooth decline from Nd towards heavier REEThe REE patterns of the EF basalts fall in betweenthose of typical OIB and E-MORB (Fig 5b)They have similar patterns and absolute abun-dances compared to basaltic dykes from theIvittuut area South Greenland (Goodenough etal 2002) LaNSmN as a measure of LREEfractionation has a restricted range from 130 to165 The degree of HREE fractionationexpressed as GdNYbN is slightly more variableand ranges from 157 to 257 Both these ratios of

the EF basalts fall generally between typical OIBand E-MORB compositions

Clinopyroxene REE patterns

Typical analyses of REE concentrations inclinopyroxenes are given in Table 3 They showthat REE concentrations in individual minerals ofone particular sample can be quite variableChondrite-normalized REE patterns of clinopyr-oxenes are very similar in the samples from theMussartut and Ulukasik Groups (Fig 6ab) Theyare enriched in all REE relative to chondriticvalues and the patterns show a fairly steepincrease from La to Sm followed by a gradualsmooth decrease from Sm to Lu Clinopyroxenesfrom sample EF 168 (Il otildeacutemaussaq Group) aredistinct as they have generally higher REEcontents especially in the LREE and a weak Euanomaly (Fig 6c) The 3 ndash106 enrichment in Nd

c (sa

mpl

e)c

(prim

itive

man

tle)

c (sa

mpl

e)c

(prim

itive

man

tle)

1

10

100

1000

Mussartucirct GroupUlukasik GroupIliacutemaussaq Group

Rb Ba K Nb La Ce Sr Nd P Zr Ti Y

Rb Ba K Nb La Ce Sr Nd P Zr Ti Y01

1

10

100

1000N-MORBE-MORBOIB

FIG 4 Multi-element plots normalized to primitivemantle values (McDonough and Sun 1995) Data forN-MORB E-MORB and average OIB (Sun and

McDonough 1989) are shown for comparison

PETROGENESIS OF THE ERIKSFJORD BASALTS

839

TA

BL

E3

RE

Eda

taof

who

lero

cks

and

clin

opyr

oxen

es

the

latt

erde

term

ined

byL

aser

ICP

-MS

Who

le-r

ock

anal

yses

R

epre

sent

ativ

ecl

inop

yrox

ene

anal

yses

S

ampl

eno

10

1231

1014

4610

1457

1014

7218

6389

1863

95E

F02

4-2

EF

039-

1E

F03

9-2

EF

039-

4E

F07

2-2

EF

072-

3E

F16

8-3

EF

168-

5

RE

Eco

ncen

trat

ions

(ppm

)L

a11

65

180

610

29

661

155

412

28

440

266

199

150

273

304

165

527

59

Ce

290

346

05

249

617

13

378

130

08

180

411

46

868

717

105

012

60

572

399

67

Pr

434

67

388

268

527

433

331

216

181

128

249

262

102

917

43

Nd

200

631

71

176

412

49

250

920

16

190

213

82

122

28

9815

64

157

258

51

960

6S

m4

756

974

043

215

914

826

664

194

283

276

077

0615

82

265

9E

u1

742

51

571

071

971

721

831

491

111

212

001

943

877

07G

d4

826

584

043

555

84

996

965

615

244

326

156

0514

75

252

7T

b ndash

ndash ndash

ndash ndash

ndash1

351

010

770

711

111

212

313

61D

y4

475

223

613

565

354

816

865

934

784

255

996

8613

10

205

2H

o0

921

040

730

731

11

011

481

401

001

001

211

292

394

24E

r2

432

531

951

972

872

663

893

342

642

112

993

646

539

60T

m ndash

ndash ndash

ndash ndash

ndash0

450

390

360

290

350

440

861

27Y

b2

182

071

741

832

592

363

782

812

492

452

703

016

788

89L

u0

340

310

260

270

390

350

460

410

300

310

520

411

251

60

(Eu

Eu

) N1

111

131

190

971

031

070

820

940

720

991

000

910

770

83

Eu

anom

aly

isca

lcul

ated

as(E

uE

u) N

=E

u N(

SmN6

Gd N

)05

norm

aliz

edto

chon

drit

eva

lues

from

Boy

nton

(198

4)

840

R HALAMA ETAL

in clinopyroxenes of EF 168 compared to theother samples (EF 024 039 and 072) seems onlypartly due to a higher degree of fractionationsincethe Mg value of EF 168 is only slightly lowerThus the LREE enrichment appears to underlinethe more alkaline character of this sample Acomparison with published REE patterns ofclinopyroxenes (Fig 6d) shows that the EFpatterns are unlike patterns from olivine gabbroswith MORB-like signatures (Benoit et al 1996)but fairly similar to clinopyroxene megacrystpatterns from an OIB-like alkaline basalt (Nimisand Vannucci 1995)

Sr Nd and O isotopic composition

Sr Nd and O isotopic compositions of all samplesare presented in Table 4 Sr and Nd isotopiccompositions of the clinopyroxene separates havebeen recalculated to an approximate eruption age

of 12 Ga (Paslick et al 1993) using decayconstants of 142610ndash11 andash1 for 87Rb (Steigerand Jager 1977) and 654610ndash12 a ndash1 for 147Sm(Lugmair and Marti 1978) Present day CHURvalues of 01967 for 147Sm144Nd (Jacobsen andWasserburg 1980) and 0512638 for 143Nd144Nd(Goldstein et al 1984) were used to calculateinitial eNd values 87Sr86Sr initial ratios are fairlyconstant and range from 070278 to 070383(Table 4) eNd(i) values vary from ndash32 to +21(Table 4) On the Sr-Nd isotope diagram the EFbasalts de ne a relatively steep trend (Fig 7) withinitial 87Sr86Sr ratios similar to those of the BulkSilicate Earth (BSE) Only one sample is slightlydisplaced towards a more radiogenic initial87Sr86Sr ratio None of the samples has an eNd

value close to depleted mantle values estimatedto be between eNd = +53 (calculated afterDePaolo 1981) and eNd = +73 (calculated afterGoldstein et al 1984) at 12 Ga The two alkalinebasalts of the Il otilde maussaq Group are within therange of the other data but they de ne a restricted eld around CHUR and BSE respectively Errorsinduced by the uncertainty in the age of the EFbasalts are largely within the error of the isotopeanalyses and do not change the overall picture Incomparison with other published Sr-Nd isotopedata from the Gardar Province the initial Srisotope ratios of the EF basalts are very similar torelatively primitive basaltic lamprophyric andcarbonatitic dyke rocks which have (87Sr86Sr)i of~0703 (Pearce and Leng 1996 Andersen 1997Goodenough et al 2002) So far no evidence hasbeen presented for an isotopicallyenriched mantlesource in the Gardar Province as the existingeNd(i) data of ndash05 to +52 show an isotopicallydepleted signature (Pearce and Leng 1996Andersen 1997 Goodenough et al 2002Upton et al 2003) Clinopyroxene from the EFsamples of Paslick et al (1993) yield eNd(i) of+22 The Nd isotope data of this study are withinthe lower range of published values from theGardar Province and extend the spread to morenegative eNd(i) values

Several recent oxygen isotope studies haveshown that mineral separates are less sensitive tolow-T alteration and secondary water take-up thanwhole-rock samples (eg Vroon et al 2001)However only few oxygen isotope data ofmineral separates from continental volcanics areavailable (eg Dobosi et al 1998 Baker et al2000) Extensive studies of mineral separates ofocean island basalts (Eiler et al 1997) andoceanic-arc lavas (Eiler et al 2000a) demon-

c (sa

mpl

e)c

(cho

ndri

te)

c (sa

mp

le)c

(cho

ndrit

e)

1

10

100

200

101446

101472

101231

101457

186389

186395

La Ce Pr Nd Sm Eu Gd Dy Ho Er Yb Lu

1

10

100

200

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

N-MORBE-MORBOIBBasaltic dykes from Ivittuut

a

b

FIG 5 (a) Whole-rock REE data of EF basaltsnormalized to chondritic values from Boynton (1984)(b) Typical compositions of N-MORB E-MORB andOIB (Sun and McDonough 1989) are shown for

comparison

PETROGENESIS OF THE ERIKSFJORD BASALTS

841

strated that analyses of mineral separates consis-tently span a narrower range than comparablewhole-rock data In our study we used clino-pyroxene because it is the only fresh mineralphase present in the samples Oxygen isotopemeasurements of clinopyroxene separates fromthe EF basalts show a range in d18Ocpx from +52to +60 with an average value of +566plusmn023(1s) All samples overlap the range of d18Ocpx

values from mantle peridotites (+525 to +590Mattey et al 1994) from continental oodbasalts from Yemen (+55 to +69 Baker etal 2000) and from alkaline basalts of Hungary(+507 to +534 Dobosi et al 1998) There is asmall difference between the averages ofMussartut Group (d18Ocpx-avg = +575plusmn015(1s)) and Ulukasik Group (d18Ocpx-avg =+575plusmn009 (1s)) samples compared toIl otildeacutemaussaq Group (d18Ocpx-avg = +533plusmn021(1s)) samples

d18O values of melts coexisting with clino-pyroxene were calculated assuming that oxygen

in pyroxene and melt were in isotopic equilibriumand T was 1100ordmC Dmelt-cpx can then be calculatedusing the following equation of Kalamarides(1986)

Dmelt-cpx = ndash00008T + 143 (T in Kelvin)

Although fractionation varies with magmatictemperatures we consider the temperature of1100ordmC and the resulting Dmelt-cpx of +03 assuf ciently representative considering the analy-tical errors involved It is similar to Dmelt-cpx

values used in previous studies (+039 Vroonet al 2001 +02 Dobosi et al 1998)Resulting d18Omelt values of EF basalts rangefrom +55 to +63 These values can becompared to whole-rock data from continentalbasalts given by Harmon and Hoefs (1995)Intraplate basalts from continental rift zonesshow a range from +46 to +83 with a meanof +61 and continental ood basalts which arerather poorly represented in the database havevalues between +43 and +65 Calculated EF

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Cpx from MORB-like

olivine gabbro

Cpx megacrysts from OIB-like

alkaline basalts

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 024EF 039

Mussartucirct Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 168

Iliacutemaussaq Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

1

10

100

1000

EF 072

Ulukasik Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb LuLa Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

c

ba

d

1

10

100

01

FIG 6 REE patterns of clinopyroxenes (cpx) from Mussartut (a) Ulukasik (b) and Il otilde maussaq (c) Group samplesnormalized to chondritic values from Boynton (1984) (d) REE patterns of cpx from MORB-type olivine gabbro(Benoit et al 1996) and OIB-type alkaline basalt (Nimis and Vannucci 1995) for comparison (note the change of

scale)

842

R HALAMA ETAL

TA

BL

E4

Sr

Nd

and

Ois

otop

icco

mpo

siti

ons

ofcl

inop

yrox

ene

sepa

rate

sfr

omE

riks

fjor

dF

orm

atio

nba

salt

sW

hole

-roc

kan

alys

esfr

omth

eK

etil

idia

nba

sem

ent

and

from

sedi

men

tary

rock

sof

the

Eri

ksfj

ord

For

mat

ion

are

give

nfo

rco

mpa

riso

n

Sam

ple

Mem

ber

Sr

Rb

87R

b86S

r87S

r86Sr

87Sr

86Sr

init

ial

SmN

d147Sm

144N

d143N

d144N

de N

d(i

)d1

8O

(ppm

)(p

pm)

(ppm

)(p

pm)

Bas

alts

E

F02

4M

ussa

rtut

646

50

486

002

180

7042

01plusmn

100

7038

267

3822

80

019

560

5124

67plusmn

10ndash

32

567

EF

063

Mus

sart

ut5

60plusmn

011

EF

061

Mus

sart

ut89

95

099

90

0321

070

3818

plusmn10

070

3266

574

163

90

2115

051

2629

plusmn10

ndash2

55

79plusmn

028

EF

039

Mus

sart

ut73

74

058

30

0229

070

3492

plusmn10

070

3098

662

191

30

2092

051

2626

plusmn10

ndash2

25

71plusmn

001

EF

059

Mus

sart

ut54

47

013

70

0073

070

3375

plusmn12

070

3250

606

168

90

2171

051

2727

plusmn09

ndash1

45

99plusmn

001

EF

108

Ulu

kasi

k66

92

111

00

0480

070

3741

plusmn09

070

2916

109

936

82

018

050

5126

17plusmn

10+

21

581

EF

072

Ulu

kasi

k69

41

032

20

0134

070

3014

plusmn10

070

2784

818

251

90

1962

051

2574

plusmn10

ndash1

25

68E

F17

4Il

otildemau

ssaq

125

42

324

005

360

7039

35plusmn

100

7030

1418

76

796

10

1424

051

2230

plusmn10

+0

45

47E

F16

8Il

otildemau

ssaq

866

71

422

004

750

7037

99plusmn

070

7029

8316

45

641

60

1550

051

2304

plusmn09

ndash0

15

18

Bas

emen

tan

dse

dim

enta

ryro

cks

JG02

Juli

aneh

aEcirc b30

01

169

71

6375

074

4789

plusmn10

071

6647

105

362

28

010

220

5114

70plusmn

09ndash

83

78

Gra

nite

EF

021

Qua

rtzi

te11

5E

F02

8A

rkos

ic6

8ar

enit

e

Sam

ples

are

list

edin

stra

tigra

phic

orde

rfr

omol

dest

toyo

unge

st87

Sr86

Sr

init

ialr

atio

san

de N

d(i)

valu

esw

ere

calc

ulat

edfo

ran

age

of1

2G

aS

tand

ard

devi

atio

nsfo

rd18

Oar

egi

ven

for

sam

ples

that

wer

ean

alys

edtw

ice

PETROGENESIS OF THE ERIKSFJORD BASALTS

843

model melt oxygen isotope data are within theranges of both groups Compared to Igaliko dykerocks from further SE in the Gardar Province(Pearce and Leng 1996) the EF basalts cover amuch more restricted eld possibly because d18Owhole-rock analyses are more prone to secondaryalteration than analyses of mineral separates(Fig 8) Whole-rock d18O measurements of theunderlying JulianehaEcirc b Granite and of the sedi-mentary sandstone members of the EriksfjordFormation yield values of 78 and 68 ndash115respectively (Table 4) and are shown forcomparison

Discussion

Mantle source regions of the EF basaltsTo reconstruct the mantle sources involved in theformation of the EF basalts we use Sr-Nd-Oisotope and incompatible trace element data On aSr-Nd isotope diagram the isotopic compositionof DMM was modelled for an age of 12 Ga(DePaolo 1981 Goldstein et al 1984 Taylor andMcLennan 1985) The fairly constant 87Sr86Srinitial isotope ratios of the EF basalts are ~0703which falls within the range typical for mantle-derived rocks but slightly above calculated valuesfor DMM (Fig 7) eNd(i) values are more

-4

-2

0

2

4

6

8

0701 0702 0703 0704 0705 0706

DMM Typical 2 s error

CHUR

BS

E

(87Sr86Sr)i

e Nd

(i)

EM-1

Qassiarsuk carbonatites and lamprophyres

EF basalts

Depleted MORB mantle

Ivittuut basaltic and lamprophyric dikes

Igaliko carbonatites and lamprophyres

Enriched mantle 1

Carbonatites from other regions of similar age

FIG 7 eNd(i) vs (87Sr86Sr)i diagram calculated for t = 12 Ga which is considered as the eruption age of the EFbasalts (Paslick et al 1993) The reference line of Bulk Silicate Earth (BSE) for the Sr isotopic composition wascalculated after DePaolo (1988) assuming present-day values of 00827 for 87Rb86Sr and 07045 for 87Sr86SrIsotopic composition of the depleted MORB mantle (DMM) was calculated after DePaolo (1981) Goldstein et al(1984) and Taylor and McLennan (1985) EM-1 was calculated after Schleicher et al (1998) (see text for details)Data sources are Goodenough et al (2002) Ivittuut (n = 6) Pearce and Leng (1996) Igaliko (n = 5) Andersen(1997) Qassiarsuk (n = 25) Bell and Blenkinsop (1989) other carbonatites of similar age (Goudini Schryburt Lake

and Prairie Lake n = 6)

844

R HALAMA ETAL

scattered and indicate that the rocks contain amixture of isotopic components from both mantleand crustal reservoirs or a heterogeneous mantlesource Only one sample (EF 108) has an eNd(i)value signi cantly above 0 This sample exhibitsthe largest spread of clinopyroxene compositionswhich may re ect strong alteration Goodenoughet al (2002) also note that the highest eNd occursin the most altered sample of their sample setHowever Paslick et al (1993) report clino-pyroxene data from two Ulukasik Groupsamples with eNd(i) gt +2 and those sampleswere apparently not disturbed with regard to theirSm-Nd age information On the other hand Piperet al (1999) suggest that the Sm-Nd systematicsof those samples may be erroneous due to laterdisturbance by uids derived from the Il otildeacutemaussaqintrusion Therefore although we are presentlynot able to decide whether positive eNd(i) values

for the EF basalts are real it is clear that none ofthose values is close to values of a 12 Ga oldDMM source (ie eNd = +5 to +7)

The Sr-Nd isotopic composition for OIB-typesources in the Proterozoic is not as wellconstrained as DMM Today OIB-type mantlesources have a wide range of isotopic composi-tions that partly overlap with DMM but eNd

values are generally lower (Hart et al 1992Hofmann 1997) The EM-1 mantle component ischaracterized by extremely low 143Nd144Ndratios and eNd values of between 0 and ndash5(Hofmann 1997) A possible 12 Ga position forEM-1 is plotted in Fig 7 calculated according toSchleicher et al (1998) with 147Sm144Nd = 018and 87Rb86Sr = 010 The present-day values usedin the calculation are 143Nd144Nd = 051234 and87Sr86Sr = 07057 taken from ocean islandbasalts closest to an EM-1 end-member(Hofmann 1997) Other Sr-Nd isotope datafrom the Gardar Province are plotted to putfurther constraints on the mantle sources of the EFbasalts (Fig 7) We also added comparative datafrom African and Canadian carbonatitesof similarage because there is a well known similaritybetween Sr-Nd isotopic data of OIBs andcarbonatites (Bell and Blenkinsop 1989) Thelinear array of the carbonatite data is consideredto be the result of mixing between the two mantlecomponents HIMU and EM-1 (Bell andBlenkinsop 1989 Bell and Tilton 2001) Thecarbonatites from the Gardar Province havesimilar Nd but slightly more radiogenic Srisotopic compositions than the reference carbo-natite data (Fig 7) The least contaminatedsamples overlap with the linear array and themore radiogenic Sr isotopic composition can beexplained by crustal contamination (Andersen1997) We therefore consider it possible that theGardar carbonatites also represent mixing ofhypothetical HIMU and EM-1 components of anOIB-like source Since the isotopically mostprimitive EF basalt samples overlap with theQassiarsuk carbonatites it could be possible thatthe basalts also contain such mantle componentsalthough Pb isotope data for the EF basalts thatcould clarify the importance of a HIMU sourceare lacking Based on the Sr-Nd isotope data theEF basalts appear to be derived from aheterogeneous OIB-like mantle source thatcomprises isotopically depleted and enrichedcomponents OIB-like and carbonatitic magma-tism are often attributed to mantle plume activity(Bell and Simonetti 1996 Hofmann 1997

2

4

6

8

10

12

-10 -8 -6 -4 -2 0 2 4 6 8

DMM

d18 O

[permil]

e Nd (i)

Typical2s error

EM-1

EF basalts - cpx

EF basalts - model melt

Ketilidian basement

Igaliko dykes

Depleted MORB mantle

Enriched mantle 1

EM-1 lavas - cpx

FIG 8 Oxygen isotopi c compositio ns (d18OV-SMOW) of clinopyroxene from EF basalts calculatedmelt (see text for details) and whole-rock Ketilidianbasement vs eNd(i) Whole-rock data for Igaliko dykerocks (n = 7) are from Pearce and Leng (1996) DMMdata derived from fresh MORB glasses are from Eiler etal (2000b) The eld of EM-1 lavas is taken fromclinopyroxene data of Gough and Tristan da CunhaIsland lavas (Harris et al 2000) and the EM-1 end-member composition is based on the calculations afterSchleicher et al (1998) for eNd(i) and an average value

of 578 (n = 18) from the EM-1 lavas for d18O

PETROGENESIS OF THE ERIKSFJORD BASALTS

845

Simonetti et al 1998) A plume input to theformation of the EF basalts is supported by therelatively large volume of magmas and the shortduration (lt5 Ma) of magmatism (Piper et al1999) It is also compatible with the geologicalsetting of a continental breakup Additionally aplume component appears to be necessary toexplain the thermal input as it is dif cult to melt acold lithosphere section (Arndt and Christensen1992)

Oxygen isotope data of the EF basalts aregenerally consistent with a mantle derivationincluding an OIB-like mantle source On the d18Ovs eNd(i) diagram (Fig 8) the overlap betweenthe EF data clinopyroxene data from EM-1-likeoceanic basalts and the hypothetical EM-1 end-member suggests an involvement of an EM-1-type mantle source Olivines from HIMU-likesources are known to be slightly lower in d18Othan typical upper mantle values from MORBs(Eiler et al 1997) but the EF data are notsuf cient to assess if the slightly lower d18Ovalues in the Il otildeacutemaussaq Group basalts are causedby a relatively higher input from a HIMU source

The contribution of the various mantle sourcesto the EF basalts can also be evaluated on theprimitive mantle-normalized multi-elementdiagram (Fig 4) and on the REE plots (Figs 5and 6) Both the whole-rock REE and theclinopyroxene REE patterns of the EF basaltsare unlike those representative of N-MORBsources We conclude again that DMM did notplay a major role in the petrogenesis of the EFbasalts On the other hand similarities with OIB-like mantle sources are more prominent Theshape of the clinopyroxene REE pattern closelyresembles those from clinopyroxene of OIB-likealkaline basalts despite differences in absoluteconcentrations The whole-rock REE patterns ofthe EF basalts lie between E-MORB and averageOIB patterns as do the multi-element patterns Onthe latter the EF basalts have marked positive Baand P peaks that are not present in typical OIBsHowever Ba contents should be interpreted withcaution since highly variable ZrBa ratios from014 to 070 are an indication that Ba was re-mobilized after emplacement of the basalts (egPrice et al 1991 Oliveira and Tarney 1995)Likewise P mobility was demonstrated in alteredbasaltic rocks (Winchester and Floyd 1976) Insummary the REE and trace element data areevidence that the mantle source(s) of the EFbasalts were enriched in incompatible traceelements relative to primitive mantle

For relatively primitive hy-normative basalticrocks of the Ivittuut area further north in theGardar Province it has been argued that some ofthem were derived from a lithospheric mantlesource previously enriched by subduction-related processes This interpretation is basedon certain geochemical characteristics such ashigh LaNb ratios and negative Nb anomalies(Goodenough et al 2002) For the EF basaltsnegative Nb anomalies are not clearly discern-ible but LaNb ratios are gt1 which seems to be acharacteristic feature of a lithospheric mantlesource (Fitton 1995) LaBa LaRb and LaKratios were also used to argue for a lithosphericorigin of the Gardar basic magmas although amantle plume involvement was assumed toprovide the required energy (Upton andEmeleus 1987 Upton et al 2003) Sincefeldspar is the only primary magmatic mineralin the EF basalts in which Ba Rb and K areincorporated in signi cant amounts and thefeldspars are generally altered these elementsmight have been easily mobilized and wetherefore do not dare to base the interpretationof the EF basalts sources on these data Inaddition the lack of precise geochemical data onthe composition of the lithosphere below theGardar Province complicates an evaluation ofthe lithospheric contribution to the EF basalticmagmatism

Certain ratios of incompatible trace elements ofthe EF basalts are compared to several occur-rences of ocean island basalts representative ofEM-1 and HIMU sources in Table 5 Ratiosbetween highly incompatible trace elements donot change signi cantly during partial melting orfractional crystallization in basaltic systems andmay therefore constrain the ratios in their sources(Weaver et al 1987) The EF basalts aresubdivided into the more alkaline isotopicallyprimitive Il otildeacutemaussaq Group samples and thosesamples with eNd(i) lt ndash1 The Il otilde maussaq Groupsamples have higher SrNd LaNb and ZrNbratios than HIMU and overlap with the upperrange of EM-1 ratios The EF basalts with eNd(i)lt ndash1 have generally higher ratios than the oceanicisland basalts and partly overlap with typicalvalues of the lower crust It therefore seems likelythat assimilation of lower crustal material mayhave changed the primary trace element ratios inthe basalts Therefore we will evaluate thepossible in uence of assimilation of crustalmaterial on the geochemistry of the EF basaltsin the following section

846

R HALAMA ETAL

The role of crustal contamination

Large amounts of bulk assimilation of crustalmaterial which is relatively SiO2-rich comparedto a basaltic melt would drive the compositions tohigher SiO2 contents and quartz-normativecompositions Partial melts derived from crustalrocks would also be relatively SiO2-rich (Beard etal 1993 Carroll and Wyllie 1990) and wouldhave the same effect The ne-normative characterof the EF basalts (Table 1) suggests thatassimilation of crustal material especially ofsilica-rich upper crustal rocks was not signi cantwith the possible exception of sample EF 059

On the primitive mantle-normalized multi-element diagrams (Fig 4) Ti troughs are typicalfor crustal rocks especially for the Upper Crust(Rudnick and Fountain 1995) and they maytherefore be present in rocks extensively contami-nated by continental crust Ti troughs aregenerally lacking in the patterns of the EFbasalts and a major upper crustal contribution tothe magmas can therefore be excluded However

small amounts of assimilation of lower crustalmaterial cannot be ruled out because the averagelower crust has higher Ti contents

Steep trends on Sr-Nd isotope plots have beenattributed to assimilation of granulite-faciesgneisses (eg Bernstein et al 1998) AFCmodelling based on an equation developed byDePaolo (1981) was performed to test thehypothesis of small amounts of contaminationwith lower crustal material (Fig 9) The basalticmagma which represents one end-member for thecalculations has the isotopic signature of theisotopically most primitive samples from thepresent data set and from the literature(Table 6) Two possible contaminants repre-senting upper and lower crustal compositionswere selected for modelling (Table 6) Therepresentative samples are a granite from theJulianehaEcirc b batholith for the upper crust (sampleJG 02) and an average of ve granulite-faciesgneisses with low RbSr ratios from the Archaeancraton of West Greenland (Taylor et al 1984) for

TABLE 5 Trace element ratios of EF basalts compared to ocean island basalts(OIBs) representing EM-1 and HIMU mantle components and the lowercrust

SrNd LaNb ZrNb

EF basalts with eNd lt ndash 1 249 ndash396 120 ndash150 121 ndash157Il otildeacutemaussaq Group EF basalts 198 ndash292 104 ndash119 51 ndash53EM-1 90 ndash207 067 ndash111 39 ndash109HIMU 45 -163 063 ndash080 27 ndash 36Lower continental crust 30 16 136

Data sources EM-1 Humphris and Thompson (1983 Walvis Ridge) and Cliff et al(1991 Inaccessible Islands) HIMU Chauvel et al (1992 Tubuai Island) Lowercontinental crust Rudnick and Fountain (1995)

TABLE 6 End-member compositions used for AFC modelling

End-member Sr (87Sr86Sr)i Nd eNd(i) Data source(ppm) (T = 12 Ga) (ppm) (T = 12 Ga)

Primitive EF basaltic 507 070278 18 + 22 Sample EF 072magma Paslick et al (1993)Upper crust 300 071665 62 ndash83 Sample JG 02(JulianehaEcirc b Granite)Lower crust 747 070668 63 ndash 234 Taylor et al (1984)(Granulite facies gneiss)

The composition of the granulite-facies gneiss is an average of 5 samples with low RbSr ratios lt02

PETROGENESIS OF THE ERIKSFJORD BASALTS

847

the lower crust An in uence of Archaean rockson Gardar magmas has been demonstrated by Pbisotope studies (Taylor and Upton 1993) andseismic data were also interpreted to re ect theexistence of a wedge of Archaean crust thatextends southwards to Lindenow Fjord (Dahl-Jensen et al 1998) We have shown in theprevious sections on the basis of Ni trends thatfractionation of olivine was most relevant to themajor element evolution of the lavas whereasfractionation of other phases was apparently ofminor importance We therefore modelled theAFC processes with olivine fractionation usingmineral-melt partition coef cients of 0014 for Srand 00066 for Nd (Rollinson 1993)

The modelling of AFC processes is stronglydependent on the parameter r de ned as the ratioof mass assimilation rate and fractional crystal-lization rate (DePaolo 1981) The change inisotopic signature becomes enhanced for highervalues of r Reiners et al (1995) have shown thatr values might be very high and even gt1 in theearly stages of magmatic evolution at hightemperatures since crystallization can besuppressed by assimilation Therefore we calcu-

lated AFC processes with two r values of 05 and08 that represent a relatively high and a middle tolow assimilation rate to demonstrate the effects ofvariable r (Fig 9) The results indicate that AFCprocesses involving upper crust similar to theJulianehaEcirc b Granite cannot explain the trend of theEF basalts because Sr isotopic compositionswould shift to considerably more radiogenicvalues This result is consistent with theconclusions derived from trace element dataThe AFC curves for assimilation of lower crusthowever agree well with the bulk of the data Themost negative eNd values can be explained byAFC processes with intermediate assimilationrates (r = 05) involving assimilation of lt5lower crustal material With a greater degree ofassimilation and an r value of 08 lt4assimilation is indicated

Lower crustal rocks of igneous origin have anaverage d18O value of +75plusmn14 with an overallrange from +54 to +125 (Fowler and Harmon1990) 5 to 10 bulk assimilation of lower crustalmaterial with d18O = +75 into a typical mantle-derived magma with d18O = +57 results ind18O values of the contaminated magma from

-24

-20

-16

-12

-8

-4

0

4

0700 0703 0706 0709 0712 0715 0718

r = 08

r = 05

r = 05

r = 08

EF Basalts

Endmembers of AFC calculations

Upper Crust

Lower Crust

Primitive EF basaltic magma

(87Sr86Sr)i

e Nd

(i)

09

08

06

09

08

06

09

0806

09

08

07

07

04

Typical errorCHUR

BS

E

FIG 9 eNd(i) vs (87Sr86Sr)i diagram showing the results of AFC modelling End-member compositions are listed inTable 6 Olivine is the only fractionating mineral Partition coef cients for olivinemelt are from the compilation ofRollinson (1993) Curves were calculated for two values of r (r = rate of assimilation rate of fractionalcrystallization) Dashed lines and numbers in italics are for r = 05 solid lines and regular numbers are for r = 08

Values at the curves are for F values (fraction of melt remaining) with 10 tick marks

848

R HALAMA ETAL

+579 to +588 Therefore the d18O data cannotbe used to rule out small amounts of assimilationof lower crust To evaluate how much contamina-tion would be permissible from the presentoxygen isotope data we calculated how muchcontamination of a mantle-derived magma with18O = +57 with a contaminant within the lowerrange of lower crustal xenoliths (d18O = +67)would drive the oxygen isotopic composition tothe highest value of the EF basalts (d18O~+63) Although simple mixing shows that60 assimilation is permissible this amountwould drive the whole-rock geochemistry tomore silicic compositions than is observed in theEF basalts In Fig 8 there is no obvious trendfrom the basalts towards the Ketilidian basementsuggesting again that upper crust was notassimilated in signi cant amounts

Ratios of trace elements incompatible inbasaltic systems can be used to distinguishbetween crust and mantle reservoirs (egWeaver 1991) Based on the AFC modellingEF basalts with eNd(i) lt ndash1 are those which aremost likely in uenced by crustal contaminationThe relatively high LaNb and ZrNb ratios ofthese samples are consistent with assimilation oflower crustal material SrNd ratios are evenhigher than the average lower crustal composi-tion but SrNd ratios in those samples which weretaken as representative for the lower crust in theAFC modelling can reach values of 80 ie wellabove the highest EF basalt ratios Again theSrNd ratios support the hypothesis of some lowercrustal contamination

The absence of more primitive basaltic rockswith MgO gt8 wt might also indicate lowercrustal assimilation It suggests that magmas mayhave been retained at depth where they acquiredmore evolved compositions by fractional crystal-lization and had the opportunity to assimilatelower crustal material during crustal underplating(Upton et al 2003)

Summary and conclusions

The new geochemical data of the EriksfjordFormation basalts in the Gardar Provinceprovided in this study are used to placepetrogenetic constraints on the sources of theirparental magmas Based on trace-element patternsand Sr-Nd isotopic constraints there seems to beno indication that a DMM-like source played animportant role in magma genesis of the EFbasalts Clinopyroxene and whole-rock REE

patterns are more compatible with an OIB-likesource but we are not able to conclude whetherthe observed enrichment in REE re ects thesourcersquos signature or magmatic processes Theisotopically most primitive EF basalts have Sr-Ndisotopic compositions that are similar to Gardarcarbonatites Due to the isotopic similaritybetween carbonatites and OIBs many arethought to be related to mantle plumes (Bell2001) We therefore speculate that OIB-likemantle components from a plume source werealso involved in the parental EF magmas but wecannot exclude interaction with the SCLM Twodistinct mantle components one isotopicallydepleted and another one enriched could havecontributed to the EF magmas AFC modelling ofSr-Nd isotope data indicates that assimilation oflt5 lower crust could explain the steep vector onthe Sr-Nd isotope plot but assimilation of uppercrustal material is unlikely Oxygen isotope dataof clinopyroxene separates are consistent with amantle origin but permit some assimilation oflower crustal rocks In summary the data areconsistent with the hypothesis that the EFmagmas could have originated from a mantleplume source which is supported by geologicalarguments and that they underwent someassimilation of lower crustal material

Acknowledgements

We would like to thank Bruce Paterson whoprovided invaluable help during Laser ICP-MSmeasurements at the Large-Scale GeochemicalFacility supported by the European Community-Access to Research Infrastructure action of theImproving Human Potential Programme contractnumber HPRI-CT-1999-00008 awarded to ProfBJ Wood (University of Bristol) Gabi Stoschekand Torsten Vennemann are thanked for their helpwith oxygen isotope measurements Elmar Reitterexpertly assisted with preparation and measure-ments of radiogenic isotopes and MathiasWestphal helped with microprobe measurementsREE whole-rock analyses were kindly providedby the NERC facility at Royal HollowayUniversity of London Michael Marks is thankedfor his pleasant company during eldwork Veryconstructive reviews by Tom Andersen andAlexei S Rukhlov greatly helped to improve themanuscript We also thank Ian Coulson for nalediting Funding of this work by the DeutscheForschungsgemeinschaft (grant Ma-21351-2) isgratefully acknowledged

PETROGENESIS OF THE ERIKSFJORD BASALTS

849

References

Andersen T (1997) Age and petrogenesis of theQassiarsuk carbonatite-alkaline silicate volcaniccomplex in the Gardar rift South GreenlandMineralogical Magazine 61 499 ndash513

Armstrong JT (1991) Quantitative elemental analysisof individual microparticles with electron beaminstruments Pp 261 ndash315 in Electron ProbeQuantitation (KFJ Heinrich and DE Newburyeditors) Plenum Press New York and London

Arndt NT and Christensen U (1992) The role oflithospheric mantle in continental ood volcanismThermal and geochemical constraints Journal ofGeophysical Research 97 10967 ndash 10981

Baker JA Menzies MA Thirlwall MF andMacpherson CG (199 7) Pet rogenes is ofQuaternary Intraplate Vocanism Sanarsquoa YemenImplications for Plume-Lithosphere Interaction andPolybaric Melt Hybridization Journal of Petrology 38 1359 ndash 1390

Baker JA MacPherson CG Menzies MAThirlwall MF Al-Kadasi M and Mattey DP(2000) Resolving crustal and mantle contributions tocontinental ood volcanism Yemen constraintsfrom mineral oxygen isotope data Journal ofPetrology 41 1805 ndash1820

Beard JS Abitz RJ and Lofgren GE (1993)Experimental melting of crustal xenoliths fromKilbourne Hole New Mexico and implications forthe contamination and genesi s of magmasContributions to Mineralogy and Petrology 11588 ndash102

Bell K (2001) Carbonatites relationships to mantle-plume activity Geological Society of AmericaSpecial Paper 352 267 ndash290

Bell K and Blenkinsop J (1989) Neodymium andStrontium Isotope Geochemistry of Carbonatites Pp278 ndash300 in Carbonatites ndash Genesis and Evolution(K Bell editor) Unwin Hyman London

Bell K and Simonetti A (1996) Carbonat iteMagmatism and Plume Activity Implications fromthe Nd Pb and Sr Isotope Systematics of OldoinyoLengai Journal of Petrology 37 1321 ndash1339

Bell K and Tilton GR (2001) Nd Pb and Sr IsotopicCompositions of East African Carbonat itesEvide nce for Mant le Mixing and PlumeInhomogene ity Journa l of Petrology 42 1927 ndash1945

Benoit M Polve M and Ceuleneer G (1996) Traceelement and isotopic characterization of ma ccumulates in a fossil mantle diapir (Oman ophiolite)Chemical Geology 134 199 ndash 214

Bernstein S Kelemen PB Tegner C Kurz MDBlusztajn J and Kent Brooks C (1998) Post-breakup basaltic magmatism along the EastGreenland Tertiary rifted margin Earth and

Planetary Science Letters 160 845 ndash862Boynton WV (1984) Geochemistry of the rare earth

elements meteorite studies Pp 63 ndash114 in RareEarth Element Geochemistry (P Henderson editor)Elsevier Amsterdam

Carroll MR and Wyllie PJ (1990) The systemtonalite-H2O at 15 kbar and the genesis of calc-alkaline magmas American Mineralogist 75345 ndash 357

Chauvel C Hofmann AW and Vidal P (1992)HIMU-EM The French Polynesian connectionEarth and Planetary Science Letters 110 99 ndash119

Clayton RN and Mayeda TK (1963) The use ofbromine penta uoride in the extraction of oxygenfrom oxides and silicates for isotope analysisGeochimica et Cosmochimica Acta 27 43 ndash52

Coulson IM and Chambers AD (1996) Patterns ofzonation in rare-earth-bearing minerals in nephelinesyenites of the North Qoroq center South GreenlandThe Canadian Mineralogist 34 1163 ndash1178

Cox KG Bell JD and Pankhurst RJ (1979) TheInterpretation of Igneous Rocks George Allen andUnwin London

Cross W Iddings JP Pirsson LV and WashingtonHS (1903) Quantitative Classi cation of IgneousRocks University of Chicago Press

Dahl-Jensen T Thybo T Hopper H and Rosing M(1998) Crustal structure at the SE Greenland marginfrom wide-angle and normal incidence seismic dataTectonophysics 288 191 ndash 198

DePaolo DJ (1981) Trace element and isotopic effectsof combined wallrock assimilation and fractionalcrystallisation Earth and Planetary Science Letters53 189 ndash 202

De P a o lo D J (1 9 8 8 ) Neodymium Iso topeGeochemistry An Introduction Springer-VerlagNew York

Dobosi G Downes H Mattey D and Embey-IsztinA (1998) Oxygen isotope ratios of phenocrysts fromalkali basalts of the Pannonian basin Evidence foran O-isotopically homogeneous upper mantle be-neath a subduction-in uenced area Lithos 42213 ndash 223

Eiler JM (2001) Oxygen isotope variations of basalticlavas and upper mantle rocks Pp 319 ndash364 inStable Isotope Geochemistry (JW Valley and DRCole editor s) Reviews in Minera logy andGeochemistry 43 The Mineralogical Society ofAmerica Washington DC

Eiler JM Farley KA Valley JW Hauri E CraigH Hart SR and Stolper EM (1997) Oxygenisotope variations in ocean island basalt phenocrystsGeochimica et Cosmochimica Acta 61 2281 ndash2293

Eiler JM Crawford A Elliot T Farley KAValley JW and Stolper EM (2000a) Oxygenisotope geochemistry of oceanic-arc lavas Journal

850

R HALAMA ETAL

between 1 and 9 ppm depending on the speci ctrace element and on the instrument used

Whole-rock REE analyses

Whole-rock REE analyses were carried out by theNatural Environment Research Council (NERC)inductively coupled plasma-atomic emissionspectrometry (ICP-AES) facility at RoyalHolloway University of London Solutions wereprepared using a combined HF dissolution andalkali fusion as described in Walsh et al (1981)Resultant solutions were analysed for 12 REEsand selected potential interferences using a PerkinElmer Optima 3300RL ICP-AES

Laser ICP-MS in situ REE measurement of clinopyroxene

In situ laser ablation inductively coupled plasma-mass spectrometer (LA-ICP-MS) REE analyses on~150 mm thick polished thin sections wereperformed at the EU Large-Scale GeochemicalFacility (University of Bristol) using a VGElemental PlasmaQuad 3 + S-Option ICP-MSequipped with a 266 nm Nd-YAG laser (VGMicroProbe II) The laser beam diameter at thesample surface was ~20 mm All measurementswere made using Thermo Elemental PlasmaLablsquotime-resolved analysisrsquo (TRA) data acquisitionsoftware with a total acquisition time of 100 s peranalysis allowing ~40 s for background followedby 50 s for laser ablation NIST 610 glass was usedfor instrument calibration and NIST 612 was usedas a secondary standard Ca was used as an internalstandard to correct the ablation yield differencesbetween and during individual analyses on bothstandards and samples To avoid analyticaluncertaintiesdue to variations in the concentrationsof the internal standard Ca concentrations werequantitatively measured within 20 mm of the laserablation pits using the EMP at the Institut furGeowissenschaften Universitat Tubingen Theprecision of trace-element concentrations basedon repeated analyses of standards is ~plusmn5 forelement concentrations gt10 ppm and plusmn10 forconcentrations lt10 ppm Data processing wascarried out of ine using the same PlasmaLabsoftware used for data collection and variouscustom-designed Excel spreadsheets The limitsof detection are de ned as 328 standard deviationsabove background level which is equal to a 95con dence that the measured signal is signi cantlyabove background Typical detection limits for theREE in this study were 004 ndash06 ppm

Sr and Nd isotope analysisFor Sr and Nd isotope analyses 15 ndash20 mgsamples of hand-picked clinopyroxene werespiked with mixed 84Sr ndash 87Rb and 150Nd ndash 149Smtracers before dissolution under high pressure inHF at 180ordmC in poly-tetrauor-ethylene (PTFE)reaction bombs The Rb and Sr were separated inquartz columns containing a 5 ml resin bed ofAG50W-X12 200ndash400 mesh equilibrated with25 N HCl The Sm and Nd separation wasperformed in quartz columns using 17 mlTe on powder coated with HDEHP (Di-EthylHexyl Phosphate) as the cation exchange mediumequilibrated with 018 N HCl All analyses weremade with a multicollector Finnigan MAT 262thermal ionization mass spectrometer (TIMS) instatic collection mode at the Institut furGeowissenschaften Universitat Tubingen TheSr was loaded with a Ta-HF activator andmeasured on a single W lament The Rb Smand Nd were measured in a double Re- lamentcon guration mode 87Sr86Sr and 143Nd144Ndratios were normalized for mass fractionation to86Sr88Sr = 01194 and 146Nd144Nd = 07219respectively The average 143Nd144Nd ratioobtained for the Ames Nd-standard (GeologicalSurvey of Canada Roddick et al 1992) was0512119plusmn10 (n = 42) and the average 87Sr86Srratio for the NBS 987 Sr-standard was0710261plusmn16 (n = 30) during the course of thisstudy 143Nd144Nd ratios have been cross-checked with the La Jolla Nd-standard whichgave 0511831plusmn30 (n = 12) Total proceduralblanks (chemistry and loading) were lt200 pg forSr and lt100 pg for Nd

Oxygen isotope analysis

The oxygen isotope composition of hand-pickedclinopyroxene separates was measured using amethod similar to that described by Sharp (1990)and Rumble and Hoering (1994) 1 ndash2 mg ofhand-picked clinopyroxenewere loaded onto a Pt-sample holder The sample chamber was pumpedout to a vacuum of ~10 ndash6 mbar and pre- uorinated overnight Samples were then heatedwith a CO2-laser in an atmosphere of 50 mbar ofpure F2 Excess F2 is separated from O2 byconversion to Cl2 using KCl held at 150ordmC Theextracted O2 is collected on a molecular sieve(13X) Oxygen isotopic compositions weremeasured on a Finnigan MAT 252 mass spectro-meter at the Institut fur GeowissenschaftenUniversitat Tubingen The results are reported as

834

R HALAMA ETAL

the per mil deviation from Vienna Standard MeanOcean Water (V-SMOW) in the standard d-notation Replicate oxygen isotope analyses ofthe NBS-28 quartz standard (Valley et al 1995)had an average precision of plusmn01 for d18O Ineach run standards were analysed at thebeginning and the end of the sample set Acorrection was applied to the data equal to theaverage of the difference between the meanmeasured value and the accepted value for thestandard (964)

Oxygen isotope compositions of powderedwhole-rock samples were determined by aconventional method modi ed after Clayton andMayeda (1963) using BrF5 as reagent andconverting the liberated oxygen to CO2 beforemass spectrometric analysis

Petrography

175 samples were collected from all threevolcanic units of the Eriksfjord Formation butmost of them were extensively altered Theycontained secondary chlorite Fe-talc Ca-Fesilicates and carbonate as alteration productsSince the focus of this study was in situ analysesof minerals or analyses of mineral separates onlynine samples with a signi cant proportion of freshclinopyroxene and relatively few alterationproducts were selected for analysis Due to thelimited number of suitable samples we did notattempt to investigate the lavas in terms of theirstratigraphic position In addition to the ninesamples from which pyroxene separates wereobtained REE whole-rock data of six othersamples were added to the data set

In general the Mussartut Group lavas areolivine-free but contain up to 50 chloriticmatrix that probably results from devitri cationof former glass In contrast the majority ofUlukasik and Il otilde maussaq Group lavas containolivine-pseudomorphs and tend to be coarsergrained and holocrystalline Many of themcontain lsquosnow- akersquo glomerocrysts of plagioclaseindicating plagioclase-saturation or super-satura-tion at near-surface levels (Upton 1996) Thesamples from which clinopyroxenes have beenseparated are ne-grained basalts with plagioclaselaths embedded in interstitial clinopyroxeneas themajor and apatite and Fe-Ti oxides as subordinateprimary magmatic mineral phases The texturecan be described as ophitic and is believed toresult from the cotectic crystallization of plagio-clase and clinopyroxene(Philpotts 1990) Olivine

has been present in a few samples but it is nowcompletely altered Pyrite pyrrhotite and chalco-pyrite occur as accessory phases Some samplescontain two oxides (ilmenite + titanomagnetite)whereas others contain a single oxide exsolvedinto ilmenite and titanomagnetite or ilmenite andhematite Texturally they are mostly subhedral toanhedral although few euhedral grains occurHematite is also present around both discrete andexsolved ilmenite and titanomagnetite grains andthe exsolved hematite might be entirely ofsecondary origin formed by oxidation of titano-magnetite Late-stage interstitial quartz occursrarely Sample EF 108 (Ulukasik Group) isremarkable in having an intense reddish colourunlike the other samples that are almost black

Results

Mineral chemistryTypical analyses of clinopyroxenes are given inTable 1 Compositions range from En41Wo44Fs16

to En28Wo45Fs27 and broadly overlap for allgroups (Fig 2) According to the classi cation ofMorimoto et al (1988) clinopyroxenes of theMussartut and Ulukasik Groups are mainlyaugites with few diopsides present whereas inthe Il otilde maussaq Group samples of diopsidiccompositions dominate over augitic onesClinopyroxene analyses from the varioussamples form tight clusters on the pyroxenequadrilateral except for those from sample EF108 that are more scattered With only a singlepyroxene phase present in all samples clino-pyroxene compositions indicate minimum crystal-lization temperatures (Lindsley 1983) of750 ndash1000ordmC (Fig 2) Plagioclase is normallyzoned and its composition typically variesbetween An70Ab29Or1 and An16Ab72Or12 Lateinterstitial albite can reach compositions ofAn0Ab99Or1 There are no signi cant composi-

1200 Ecirc C

800 Ecirc C

25 50 75

25 25

En Fs

HdDi

Mussartucirct Group

Iliacutemaussaq Group

Ulukasik Group

FIG 2 Clinopyroxene analyses of the Eriksfjord basaltsplotted in the pyroxene quadrilateral with isotherms

after Lindsley (1983)

PETROGENESIS OF THE ERIKSFJORD BASALTS

835

tional changes among the various samplesIlmenite from two samples (EF 024 and EF 039)which contain coexisting ilmenite and titanomag-netite includes a signi cant component ofpyrophanite (MnTiO3) with up to 87 wtMnO It s com pos i t i ons va r y be t weenI l m 8 6 H e m 5 P y r 8 a nd I l m 7 6 H e m 5 P yr 1 9 Titanomagnetite analyses were unsatisfactory asall grains show strong alteration They areconsidered unreliable and are not presented here

Whole-rock geochemistry and classif|cation

Concentrations of major and trace elements aresummarized in Table 2 The basalts can beclassi ed according to the TAS diagram (LeMaitre et al 1989) as basalts and trachybasaltswith both alkaline and subalkaline compositions(not shown) Due to possible mobility of the Naand K the samples were also plotted into NbYvs ZrP2O5 and P2O5 vs Zr diagrams (Floyd andWinchester 1975) which can discriminate

between tholeiitic and alkaline series Thesediagrams do not yield consistent discriminationsfor the EF basalts but indicate that the characterof the sample suite is transitional betweentholeiitic and alkaline (not shown)

Mg [100Mg(Mg+Fe2+)] calculated assumingFe2O3FeO = 02 as recommended for basalts byMiddlemost (1989) varies between 54 and 40(Table 2) Ni concentrations range from 46 to115 ppm and Cr contents from 29 to 126 ppmTaking Mg as fractionation index a positivecorrelation (r2 = 061) with Ni contents indicatesfractionation of olivine (Fig 3) The relativelypoor positive correlations of Mg with Cr (r2 =044) and Sc (r2 = 023) argue against a signi cantfractionation of Cr-spinel or clinopyroxene Thisis consistent with the lack of clinopyroxenephenocrysts and the absence of any Cr-spinelThe negative correlations of Mg with Sr TiO2

and P2O5 suggest that fractionation of plagioclaseFe-Ti oxides and apatite was not important in thesample set investigated which is again consistent

TABLE 1 Typical clinopyroxene analyses of Eriksfjord Formation basalts determined by EMP

Sample EF 024 EF 024 EF 039 EF 059 EF 072 EF 108 EF 168 EF 174Cpx no cpx 2 cpx 3 cpx 5 cpx 3 cpx 5 cpx 7 cpx 4 cpx 1

SiO2 5060 5041 5128 4988 5145 4950 5175 5069TiO2 173 176 121 169 140 162 096 121Al2O3 297 300 179 223 171 264 155 213FeO 1160 1068 1218 1405 993 1234 979 1129MnO 028 021 030 040 019 028 025 031MgO 1254 1324 1263 1182 1325 1163 1307 1214CaO 1959 2029 2002 1907 2106 2099 2174 2143Na2O 034 031 029 049 033 055 052 054

Total 9965 9989 9971 9963 9932 9955 9962 9974

Formulae based on 4 cations and 6 oxygensSi 191 189 194 190 194 188 194 191Al 013 013 008 010 008 012 007 009Ti 005 005 003 005 004 005 003 003Fe3+ - - - 003 - 007 003 005Mg 071 074 071 067 075 066 073 068Fe2+ 037 033 039 041 031 032 028 031Mn 001 001 001 001 001 001 001 001Ca 079 082 081 078 085 085 087 087Na 002 002 002 004 002 004 004 004

Total 400 400 400 400 400 400 400 400

Pyroxene projection after Lindsley (1983)Wo 039 041 041 041 043 043 045 044En 040 041 039 037 040 038 040 038Fs 021 018 021 023 017 019 015 017

836

R HALAMA ETAL

TABLE 2 XRF whole-rock analyses of Eriksfjord Formation basalts

Sample EF 024 EF 063 EF 061 EF 039 EF 059 EF 108 EF 072 EF 174 EF 168Group Mussartut Mussartut Mussartut Mussartut Mussartut Ulukasik Ulukasik Ilotildeacutemaussaq IlotildeacutemaussaqSubunit 4 4 5 6 7 Lower Unit Upper Unit Sermilik Narssaq

Fjeld

Major elements (wt)SiO2 4573 4585 4688 4660 4875 4698 4656 4601 4545TiO2 230 200 180 175 191 265 203 299 277Al2O3 1578 1630 1600 1612 1645 1682 1668 1645 1719Fe2O3 1510 1356 1331 1383 1225 1394 1321 1438 1392MnO 028 018 018 018 017 014 017 018 017MgO 525 683 653 673 568 396 647 440 480CaO 842 849 872 857 920 641 808 746 737Na2O 349 274 289 276 236 448 303 323 376K2O 087 083 068 032 027 159 070 205 132P2O5 041 035 023 023 023 048 035 111 083LOI 199 210 240 238 242 204 210 052 156

Total 9962 9923 9962 9947 9969 9949 9938 9878 9914

Mg 453 545 538 536 524 403 538 421 450

Trace elements (ppm)Sc 24 22 25 22 26 19 20 21 16V 209 175 215 195 211 142 170 164 203Cr 81 108 66 65 126 48 82 33 29Co 63 78 70 62 71 44 68 35 46Ni 99 115 85 85 101 58 106 46 51Cu 65 49 46 60 97 41 59 38 41Zn 103 102 110 97 110 93 82 122 105Ga 20 18 21 19 20 19 16 21 19Rb 24 26 16 lt 10 lt 10 38 12 39 45Sr 504 503 512 392 373 712 507 850 992Y 31 26 28 23 28 31 17 31 25Zr 188 165 121 116 125 125 88 164 139Nb 12 11 10 lt 10 10 9 lt 10 31 27Ba 445 375 415 302 178 582 405 1293 900Pb 6 7 3 4 3 4 4 3 5La ndash ndash 15 ndash 12 11 ndash 37 28Ce ndash ndash 40 ndash 38 39 ndash 92 71Pr ndash ndash 1 ndash 1 2 ndash 9 4Nd ndash ndash 16 ndash 15 18 ndash 43 34Sm ndash ndash 6 ndash 7 6 ndash 5 5

CIPW normq ndash ndash ndash ndash 313 ndash ndash ndash ndashor 514 491 402 189 160 940 414 1212 780ab 2807 2319 2445 2335 1997 3207 2564 2733 2981an 2482 2973 2868 3065 3350 2109 2985 2433 2613ne 079 ndash ndash ndash ndash 316 ndash ndash 109di 1187 830 1076 853 877 633 654 452 416hy ndash 890 1059 1650 2217 ndash 1097 315 ndashol 1670 1309 1044 780 ndash 1472 1120 1388 1689mt 365 328 322 335 296 336 319 348 336il 437 380 342 332 363 503 386 568 526ap 095 081 053 053 053 111 081 257 192

Mg was calculated as Mg = 100Mg2+(Mg2++Fe2+) using Fe2O3FeO = 02 Normative mineral compositionswere determined according to the CIPW calculation scheme

PETROGENESIS OF THE ERIKSFJORD BASALTS

837

1 5

2 0

2 5

3 0

3 5

40

60

80

100

120

0

0 5

1

1 5

200

400

600

800

1000

0

50

100

150

15

20

25

30

4 0 45 50 5 5

r2 = 023

r2 = 044r2 = 061

r2 = 053

r2 = 082 r2 = 052

Ni (ppm)

Sc (ppm)

P2O5 (wt)

Sr (ppm)

TiO2 (wt)

Cr (ppm)

Mg Mg

Mg Mg

Mg

40 4 5 50 55

Mg

40 45 50 55

40 45 50 55

40 45 50 5540 4 5 50 55

FIG 3 Whole-rock variation diagrams for EF basalts with Mg as fractionation index Mg is calculated asMg(Mg+Fe2+) with Fe2O3FeO = 02 (Middlemost 1989)

838

R HALAMA ETAL

with the mostly subhedral Fe-Ti oxides and theophitic texture Ratios of incompatible traceelements such as ZrY (range 40 ndash63) andTiY (range 385 ndash716) are relatively uniformthroughout all samples but some scatter isevident Notable geochemical features are highSr contents up to 1000 ppm and very high Bacontents up to 1300 ppm

Normative mineral compositions were deter-mined according to the CIPW calculation scheme(Cross et al 1903 Cox et al 1979) Mostsamples are hypersthene-normative some arenepheline-normative and one (EF 059) is quartz-normative (Table 2) Despite the uncertaintiesinvolved in the CIPW norm calculation the dataindicate that some of the basalts crystallized fromsilica-undersaturated magmas and others fromdistinct silica-saturatedor -oversaturatedmagmas

Multi-element diagrams normalized to primi-tive mantle values show that the EF basalts areenriched in all incompatible elements relative toprimitive mantle values (Fig 4) The patterns ofthe EF basalts are characterized by a prominentBa peak a subordinate P peak and a decreasingelement enrichment with increasing compatibilityIn comparison with normal MORB (N-MORB)enriched MORB (E-MORB) and average OIB(Sun and McDonough 1989) the patterns of theEF basalts are closest to those of OIB both inshape and normalized concentrations

Whole-rock REE data

Whole-rock REE data are presented in Table 3and REE concentrations normalized to chondriticvalues of Boynton (1984) are shown in Fig 5aAll samples are characterized by LREE-enrichedpatterns with no or slightly positive Euanomalies (EuEu = 097 to 113) La concentra-tions are roughly 20 ndash60 times and Lu concentra-tions are 10 times chondritic values NormalizedLREE abundances from La to Nd are fairlyconstant but the patterns show a steady andsmooth decline from Nd towards heavier REEThe REE patterns of the EF basalts fall in betweenthose of typical OIB and E-MORB (Fig 5b)They have similar patterns and absolute abun-dances compared to basaltic dykes from theIvittuut area South Greenland (Goodenough etal 2002) LaNSmN as a measure of LREEfractionation has a restricted range from 130 to165 The degree of HREE fractionationexpressed as GdNYbN is slightly more variableand ranges from 157 to 257 Both these ratios of

the EF basalts fall generally between typical OIBand E-MORB compositions

Clinopyroxene REE patterns

Typical analyses of REE concentrations inclinopyroxenes are given in Table 3 They showthat REE concentrations in individual minerals ofone particular sample can be quite variableChondrite-normalized REE patterns of clinopyr-oxenes are very similar in the samples from theMussartut and Ulukasik Groups (Fig 6ab) Theyare enriched in all REE relative to chondriticvalues and the patterns show a fairly steepincrease from La to Sm followed by a gradualsmooth decrease from Sm to Lu Clinopyroxenesfrom sample EF 168 (Il otildeacutemaussaq Group) aredistinct as they have generally higher REEcontents especially in the LREE and a weak Euanomaly (Fig 6c) The 3 ndash106 enrichment in Nd

c (sa

mpl

e)c

(prim

itive

man

tle)

c (sa

mpl

e)c

(prim

itive

man

tle)

1

10

100

1000

Mussartucirct GroupUlukasik GroupIliacutemaussaq Group

Rb Ba K Nb La Ce Sr Nd P Zr Ti Y

Rb Ba K Nb La Ce Sr Nd P Zr Ti Y01

1

10

100

1000N-MORBE-MORBOIB

FIG 4 Multi-element plots normalized to primitivemantle values (McDonough and Sun 1995) Data forN-MORB E-MORB and average OIB (Sun and

McDonough 1989) are shown for comparison

PETROGENESIS OF THE ERIKSFJORD BASALTS

839

TA

BL

E3

RE

Eda

taof

who

lero

cks

and

clin

opyr

oxen

es

the

latt

erde

term

ined

byL

aser

ICP

-MS

Who

le-r

ock

anal

yses

R

epre

sent

ativ

ecl

inop

yrox

ene

anal

yses

S

ampl

eno

10

1231

1014

4610

1457

1014

7218

6389

1863

95E

F02

4-2

EF

039-

1E

F03

9-2

EF

039-

4E

F07

2-2

EF

072-

3E

F16

8-3

EF

168-

5

RE

Eco

ncen

trat

ions

(ppm

)L

a11

65

180

610

29

661

155

412

28

440

266

199

150

273

304

165

527

59

Ce

290

346

05

249

617

13

378

130

08

180

411

46

868

717

105

012

60

572

399

67

Pr

434

67

388

268

527

433

331

216

181

128

249

262

102

917

43

Nd

200

631

71

176

412

49

250

920

16

190

213

82

122

28

9815

64

157

258

51

960

6S

m4

756

974

043

215

914

826

664

194

283

276

077

0615

82

265

9E

u1

742

51

571

071

971

721

831

491

111

212

001

943

877

07G

d4

826

584

043

555

84

996

965

615

244

326

156

0514

75

252

7T

b ndash

ndash ndash

ndash ndash

ndash1

351

010

770

711

111

212

313

61D

y4

475

223

613

565

354

816

865

934

784

255

996

8613

10

205

2H

o0

921

040

730

731

11

011

481

401

001

001

211

292

394

24E

r2

432

531

951

972

872

663

893

342

642

112

993

646

539

60T

m ndash

ndash ndash

ndash ndash

ndash0

450

390

360

290

350

440

861

27Y

b2

182

071

741

832

592

363

782

812

492

452

703

016

788

89L

u0

340

310

260

270

390

350

460

410

300

310

520

411

251

60

(Eu

Eu

) N1

111

131

190

971

031

070

820

940

720

991

000

910

770

83

Eu

anom

aly

isca

lcul

ated

as(E

uE

u) N

=E

u N(

SmN6

Gd N

)05

norm

aliz

edto

chon

drit

eva

lues

from

Boy

nton

(198

4)

840

R HALAMA ETAL

in clinopyroxenes of EF 168 compared to theother samples (EF 024 039 and 072) seems onlypartly due to a higher degree of fractionationsincethe Mg value of EF 168 is only slightly lowerThus the LREE enrichment appears to underlinethe more alkaline character of this sample Acomparison with published REE patterns ofclinopyroxenes (Fig 6d) shows that the EFpatterns are unlike patterns from olivine gabbroswith MORB-like signatures (Benoit et al 1996)but fairly similar to clinopyroxene megacrystpatterns from an OIB-like alkaline basalt (Nimisand Vannucci 1995)

Sr Nd and O isotopic composition

Sr Nd and O isotopic compositions of all samplesare presented in Table 4 Sr and Nd isotopiccompositions of the clinopyroxene separates havebeen recalculated to an approximate eruption age

of 12 Ga (Paslick et al 1993) using decayconstants of 142610ndash11 andash1 for 87Rb (Steigerand Jager 1977) and 654610ndash12 a ndash1 for 147Sm(Lugmair and Marti 1978) Present day CHURvalues of 01967 for 147Sm144Nd (Jacobsen andWasserburg 1980) and 0512638 for 143Nd144Nd(Goldstein et al 1984) were used to calculateinitial eNd values 87Sr86Sr initial ratios are fairlyconstant and range from 070278 to 070383(Table 4) eNd(i) values vary from ndash32 to +21(Table 4) On the Sr-Nd isotope diagram the EFbasalts de ne a relatively steep trend (Fig 7) withinitial 87Sr86Sr ratios similar to those of the BulkSilicate Earth (BSE) Only one sample is slightlydisplaced towards a more radiogenic initial87Sr86Sr ratio None of the samples has an eNd

value close to depleted mantle values estimatedto be between eNd = +53 (calculated afterDePaolo 1981) and eNd = +73 (calculated afterGoldstein et al 1984) at 12 Ga The two alkalinebasalts of the Il otilde maussaq Group are within therange of the other data but they de ne a restricted eld around CHUR and BSE respectively Errorsinduced by the uncertainty in the age of the EFbasalts are largely within the error of the isotopeanalyses and do not change the overall picture Incomparison with other published Sr-Nd isotopedata from the Gardar Province the initial Srisotope ratios of the EF basalts are very similar torelatively primitive basaltic lamprophyric andcarbonatitic dyke rocks which have (87Sr86Sr)i of~0703 (Pearce and Leng 1996 Andersen 1997Goodenough et al 2002) So far no evidence hasbeen presented for an isotopicallyenriched mantlesource in the Gardar Province as the existingeNd(i) data of ndash05 to +52 show an isotopicallydepleted signature (Pearce and Leng 1996Andersen 1997 Goodenough et al 2002Upton et al 2003) Clinopyroxene from the EFsamples of Paslick et al (1993) yield eNd(i) of+22 The Nd isotope data of this study are withinthe lower range of published values from theGardar Province and extend the spread to morenegative eNd(i) values

Several recent oxygen isotope studies haveshown that mineral separates are less sensitive tolow-T alteration and secondary water take-up thanwhole-rock samples (eg Vroon et al 2001)However only few oxygen isotope data ofmineral separates from continental volcanics areavailable (eg Dobosi et al 1998 Baker et al2000) Extensive studies of mineral separates ofocean island basalts (Eiler et al 1997) andoceanic-arc lavas (Eiler et al 2000a) demon-

c (sa

mpl

e)c

(cho

ndri

te)

c (sa

mp

le)c

(cho

ndrit

e)

1

10

100

200

101446

101472

101231

101457

186389

186395

La Ce Pr Nd Sm Eu Gd Dy Ho Er Yb Lu

1

10

100

200

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

N-MORBE-MORBOIBBasaltic dykes from Ivittuut

a

b

FIG 5 (a) Whole-rock REE data of EF basaltsnormalized to chondritic values from Boynton (1984)(b) Typical compositions of N-MORB E-MORB andOIB (Sun and McDonough 1989) are shown for

comparison

PETROGENESIS OF THE ERIKSFJORD BASALTS

841

strated that analyses of mineral separates consis-tently span a narrower range than comparablewhole-rock data In our study we used clino-pyroxene because it is the only fresh mineralphase present in the samples Oxygen isotopemeasurements of clinopyroxene separates fromthe EF basalts show a range in d18Ocpx from +52to +60 with an average value of +566plusmn023(1s) All samples overlap the range of d18Ocpx

values from mantle peridotites (+525 to +590Mattey et al 1994) from continental oodbasalts from Yemen (+55 to +69 Baker etal 2000) and from alkaline basalts of Hungary(+507 to +534 Dobosi et al 1998) There is asmall difference between the averages ofMussartut Group (d18Ocpx-avg = +575plusmn015(1s)) and Ulukasik Group (d18Ocpx-avg =+575plusmn009 (1s)) samples compared toIl otildeacutemaussaq Group (d18Ocpx-avg = +533plusmn021(1s)) samples

d18O values of melts coexisting with clino-pyroxene were calculated assuming that oxygen

in pyroxene and melt were in isotopic equilibriumand T was 1100ordmC Dmelt-cpx can then be calculatedusing the following equation of Kalamarides(1986)

Dmelt-cpx = ndash00008T + 143 (T in Kelvin)

Although fractionation varies with magmatictemperatures we consider the temperature of1100ordmC and the resulting Dmelt-cpx of +03 assuf ciently representative considering the analy-tical errors involved It is similar to Dmelt-cpx

values used in previous studies (+039 Vroonet al 2001 +02 Dobosi et al 1998)Resulting d18Omelt values of EF basalts rangefrom +55 to +63 These values can becompared to whole-rock data from continentalbasalts given by Harmon and Hoefs (1995)Intraplate basalts from continental rift zonesshow a range from +46 to +83 with a meanof +61 and continental ood basalts which arerather poorly represented in the database havevalues between +43 and +65 Calculated EF

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Cpx from MORB-like

olivine gabbro

Cpx megacrysts from OIB-like

alkaline basalts

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 024EF 039

Mussartucirct Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 168

Iliacutemaussaq Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

1

10

100

1000

EF 072

Ulukasik Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb LuLa Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

c

ba

d

1

10

100

01

FIG 6 REE patterns of clinopyroxenes (cpx) from Mussartut (a) Ulukasik (b) and Il otilde maussaq (c) Group samplesnormalized to chondritic values from Boynton (1984) (d) REE patterns of cpx from MORB-type olivine gabbro(Benoit et al 1996) and OIB-type alkaline basalt (Nimis and Vannucci 1995) for comparison (note the change of

scale)

842

R HALAMA ETAL

TA

BL

E4

Sr

Nd

and

Ois

otop

icco

mpo

siti

ons

ofcl

inop

yrox

ene

sepa

rate

sfr

omE

riks

fjor

dF

orm

atio

nba

salt

sW

hole

-roc

kan

alys

esfr

omth

eK

etil

idia

nba

sem

ent

and

from

sedi

men

tary

rock

sof

the

Eri

ksfj

ord

For

mat

ion

are

give

nfo

rco

mpa

riso

n

Sam

ple

Mem

ber

Sr

Rb

87R

b86S

r87S

r86Sr

87Sr

86Sr

init

ial

SmN

d147Sm

144N

d143N

d144N

de N

d(i

)d1

8O

(ppm

)(p

pm)

(ppm

)(p

pm)

Bas

alts

E

F02

4M

ussa

rtut

646

50

486

002

180

7042

01plusmn

100

7038

267

3822

80

019

560

5124

67plusmn

10ndash

32

567

EF

063

Mus

sart

ut5

60plusmn

011

EF

061

Mus

sart

ut89

95

099

90

0321

070

3818

plusmn10

070

3266

574

163

90

2115

051

2629

plusmn10

ndash2

55

79plusmn

028

EF

039

Mus

sart

ut73

74

058

30

0229

070

3492

plusmn10

070

3098

662

191

30

2092

051

2626

plusmn10

ndash2

25

71plusmn

001

EF

059

Mus

sart

ut54

47

013

70

0073

070

3375

plusmn12

070

3250

606

168

90

2171

051

2727

plusmn09

ndash1

45

99plusmn

001

EF

108

Ulu

kasi

k66

92

111

00

0480

070

3741

plusmn09

070

2916

109

936

82

018

050

5126

17plusmn

10+

21

581

EF

072

Ulu

kasi

k69

41

032

20

0134

070

3014

plusmn10

070

2784

818

251

90

1962

051

2574

plusmn10

ndash1

25

68E

F17

4Il

otildemau

ssaq

125

42

324

005

360

7039

35plusmn

100

7030

1418

76

796

10

1424

051

2230

plusmn10

+0

45

47E

F16

8Il

otildemau

ssaq

866

71

422

004

750

7037

99plusmn

070

7029

8316

45

641

60

1550

051

2304

plusmn09

ndash0

15

18

Bas

emen

tan

dse

dim

enta

ryro

cks

JG02

Juli

aneh

aEcirc b30

01

169

71

6375

074

4789

plusmn10

071

6647

105

362

28

010

220

5114

70plusmn

09ndash

83

78

Gra

nite

EF

021

Qua

rtzi

te11

5E

F02

8A

rkos

ic6

8ar

enit

e

Sam

ples

are

list

edin

stra

tigra

phic

orde

rfr

omol

dest

toyo

unge

st87

Sr86

Sr

init

ialr

atio

san

de N

d(i)

valu

esw

ere

calc

ulat

edfo

ran

age

of1

2G

aS

tand

ard

devi

atio

nsfo

rd18

Oar

egi

ven

for

sam

ples

that

wer

ean

alys

edtw

ice

PETROGENESIS OF THE ERIKSFJORD BASALTS

843

model melt oxygen isotope data are within theranges of both groups Compared to Igaliko dykerocks from further SE in the Gardar Province(Pearce and Leng 1996) the EF basalts cover amuch more restricted eld possibly because d18Owhole-rock analyses are more prone to secondaryalteration than analyses of mineral separates(Fig 8) Whole-rock d18O measurements of theunderlying JulianehaEcirc b Granite and of the sedi-mentary sandstone members of the EriksfjordFormation yield values of 78 and 68 ndash115respectively (Table 4) and are shown forcomparison

Discussion

Mantle source regions of the EF basaltsTo reconstruct the mantle sources involved in theformation of the EF basalts we use Sr-Nd-Oisotope and incompatible trace element data On aSr-Nd isotope diagram the isotopic compositionof DMM was modelled for an age of 12 Ga(DePaolo 1981 Goldstein et al 1984 Taylor andMcLennan 1985) The fairly constant 87Sr86Srinitial isotope ratios of the EF basalts are ~0703which falls within the range typical for mantle-derived rocks but slightly above calculated valuesfor DMM (Fig 7) eNd(i) values are more

-4

-2

0

2

4

6

8

0701 0702 0703 0704 0705 0706

DMM Typical 2 s error

CHUR

BS

E

(87Sr86Sr)i

e Nd

(i)

EM-1

Qassiarsuk carbonatites and lamprophyres

EF basalts

Depleted MORB mantle

Ivittuut basaltic and lamprophyric dikes

Igaliko carbonatites and lamprophyres

Enriched mantle 1

Carbonatites from other regions of similar age

FIG 7 eNd(i) vs (87Sr86Sr)i diagram calculated for t = 12 Ga which is considered as the eruption age of the EFbasalts (Paslick et al 1993) The reference line of Bulk Silicate Earth (BSE) for the Sr isotopic composition wascalculated after DePaolo (1988) assuming present-day values of 00827 for 87Rb86Sr and 07045 for 87Sr86SrIsotopic composition of the depleted MORB mantle (DMM) was calculated after DePaolo (1981) Goldstein et al(1984) and Taylor and McLennan (1985) EM-1 was calculated after Schleicher et al (1998) (see text for details)Data sources are Goodenough et al (2002) Ivittuut (n = 6) Pearce and Leng (1996) Igaliko (n = 5) Andersen(1997) Qassiarsuk (n = 25) Bell and Blenkinsop (1989) other carbonatites of similar age (Goudini Schryburt Lake

and Prairie Lake n = 6)

844

R HALAMA ETAL

scattered and indicate that the rocks contain amixture of isotopic components from both mantleand crustal reservoirs or a heterogeneous mantlesource Only one sample (EF 108) has an eNd(i)value signi cantly above 0 This sample exhibitsthe largest spread of clinopyroxene compositionswhich may re ect strong alteration Goodenoughet al (2002) also note that the highest eNd occursin the most altered sample of their sample setHowever Paslick et al (1993) report clino-pyroxene data from two Ulukasik Groupsamples with eNd(i) gt +2 and those sampleswere apparently not disturbed with regard to theirSm-Nd age information On the other hand Piperet al (1999) suggest that the Sm-Nd systematicsof those samples may be erroneous due to laterdisturbance by uids derived from the Il otildeacutemaussaqintrusion Therefore although we are presentlynot able to decide whether positive eNd(i) values

for the EF basalts are real it is clear that none ofthose values is close to values of a 12 Ga oldDMM source (ie eNd = +5 to +7)

The Sr-Nd isotopic composition for OIB-typesources in the Proterozoic is not as wellconstrained as DMM Today OIB-type mantlesources have a wide range of isotopic composi-tions that partly overlap with DMM but eNd

values are generally lower (Hart et al 1992Hofmann 1997) The EM-1 mantle component ischaracterized by extremely low 143Nd144Ndratios and eNd values of between 0 and ndash5(Hofmann 1997) A possible 12 Ga position forEM-1 is plotted in Fig 7 calculated according toSchleicher et al (1998) with 147Sm144Nd = 018and 87Rb86Sr = 010 The present-day values usedin the calculation are 143Nd144Nd = 051234 and87Sr86Sr = 07057 taken from ocean islandbasalts closest to an EM-1 end-member(Hofmann 1997) Other Sr-Nd isotope datafrom the Gardar Province are plotted to putfurther constraints on the mantle sources of the EFbasalts (Fig 7) We also added comparative datafrom African and Canadian carbonatitesof similarage because there is a well known similaritybetween Sr-Nd isotopic data of OIBs andcarbonatites (Bell and Blenkinsop 1989) Thelinear array of the carbonatite data is consideredto be the result of mixing between the two mantlecomponents HIMU and EM-1 (Bell andBlenkinsop 1989 Bell and Tilton 2001) Thecarbonatites from the Gardar Province havesimilar Nd but slightly more radiogenic Srisotopic compositions than the reference carbo-natite data (Fig 7) The least contaminatedsamples overlap with the linear array and themore radiogenic Sr isotopic composition can beexplained by crustal contamination (Andersen1997) We therefore consider it possible that theGardar carbonatites also represent mixing ofhypothetical HIMU and EM-1 components of anOIB-like source Since the isotopically mostprimitive EF basalt samples overlap with theQassiarsuk carbonatites it could be possible thatthe basalts also contain such mantle componentsalthough Pb isotope data for the EF basalts thatcould clarify the importance of a HIMU sourceare lacking Based on the Sr-Nd isotope data theEF basalts appear to be derived from aheterogeneous OIB-like mantle source thatcomprises isotopically depleted and enrichedcomponents OIB-like and carbonatitic magma-tism are often attributed to mantle plume activity(Bell and Simonetti 1996 Hofmann 1997

2

4

6

8

10

12

-10 -8 -6 -4 -2 0 2 4 6 8

DMM

d18 O

[permil]

e Nd (i)

Typical2s error

EM-1

EF basalts - cpx

EF basalts - model melt

Ketilidian basement

Igaliko dykes

Depleted MORB mantle

Enriched mantle 1

EM-1 lavas - cpx

FIG 8 Oxygen isotopi c compositio ns (d18OV-SMOW) of clinopyroxene from EF basalts calculatedmelt (see text for details) and whole-rock Ketilidianbasement vs eNd(i) Whole-rock data for Igaliko dykerocks (n = 7) are from Pearce and Leng (1996) DMMdata derived from fresh MORB glasses are from Eiler etal (2000b) The eld of EM-1 lavas is taken fromclinopyroxene data of Gough and Tristan da CunhaIsland lavas (Harris et al 2000) and the EM-1 end-member composition is based on the calculations afterSchleicher et al (1998) for eNd(i) and an average value

of 578 (n = 18) from the EM-1 lavas for d18O

PETROGENESIS OF THE ERIKSFJORD BASALTS

845

Simonetti et al 1998) A plume input to theformation of the EF basalts is supported by therelatively large volume of magmas and the shortduration (lt5 Ma) of magmatism (Piper et al1999) It is also compatible with the geologicalsetting of a continental breakup Additionally aplume component appears to be necessary toexplain the thermal input as it is dif cult to melt acold lithosphere section (Arndt and Christensen1992)

Oxygen isotope data of the EF basalts aregenerally consistent with a mantle derivationincluding an OIB-like mantle source On the d18Ovs eNd(i) diagram (Fig 8) the overlap betweenthe EF data clinopyroxene data from EM-1-likeoceanic basalts and the hypothetical EM-1 end-member suggests an involvement of an EM-1-type mantle source Olivines from HIMU-likesources are known to be slightly lower in d18Othan typical upper mantle values from MORBs(Eiler et al 1997) but the EF data are notsuf cient to assess if the slightly lower d18Ovalues in the Il otildeacutemaussaq Group basalts are causedby a relatively higher input from a HIMU source

The contribution of the various mantle sourcesto the EF basalts can also be evaluated on theprimitive mantle-normalized multi-elementdiagram (Fig 4) and on the REE plots (Figs 5and 6) Both the whole-rock REE and theclinopyroxene REE patterns of the EF basaltsare unlike those representative of N-MORBsources We conclude again that DMM did notplay a major role in the petrogenesis of the EFbasalts On the other hand similarities with OIB-like mantle sources are more prominent Theshape of the clinopyroxene REE pattern closelyresembles those from clinopyroxene of OIB-likealkaline basalts despite differences in absoluteconcentrations The whole-rock REE patterns ofthe EF basalts lie between E-MORB and averageOIB patterns as do the multi-element patterns Onthe latter the EF basalts have marked positive Baand P peaks that are not present in typical OIBsHowever Ba contents should be interpreted withcaution since highly variable ZrBa ratios from014 to 070 are an indication that Ba was re-mobilized after emplacement of the basalts (egPrice et al 1991 Oliveira and Tarney 1995)Likewise P mobility was demonstrated in alteredbasaltic rocks (Winchester and Floyd 1976) Insummary the REE and trace element data areevidence that the mantle source(s) of the EFbasalts were enriched in incompatible traceelements relative to primitive mantle

For relatively primitive hy-normative basalticrocks of the Ivittuut area further north in theGardar Province it has been argued that some ofthem were derived from a lithospheric mantlesource previously enriched by subduction-related processes This interpretation is basedon certain geochemical characteristics such ashigh LaNb ratios and negative Nb anomalies(Goodenough et al 2002) For the EF basaltsnegative Nb anomalies are not clearly discern-ible but LaNb ratios are gt1 which seems to be acharacteristic feature of a lithospheric mantlesource (Fitton 1995) LaBa LaRb and LaKratios were also used to argue for a lithosphericorigin of the Gardar basic magmas although amantle plume involvement was assumed toprovide the required energy (Upton andEmeleus 1987 Upton et al 2003) Sincefeldspar is the only primary magmatic mineralin the EF basalts in which Ba Rb and K areincorporated in signi cant amounts and thefeldspars are generally altered these elementsmight have been easily mobilized and wetherefore do not dare to base the interpretationof the EF basalts sources on these data Inaddition the lack of precise geochemical data onthe composition of the lithosphere below theGardar Province complicates an evaluation ofthe lithospheric contribution to the EF basalticmagmatism

Certain ratios of incompatible trace elements ofthe EF basalts are compared to several occur-rences of ocean island basalts representative ofEM-1 and HIMU sources in Table 5 Ratiosbetween highly incompatible trace elements donot change signi cantly during partial melting orfractional crystallization in basaltic systems andmay therefore constrain the ratios in their sources(Weaver et al 1987) The EF basalts aresubdivided into the more alkaline isotopicallyprimitive Il otildeacutemaussaq Group samples and thosesamples with eNd(i) lt ndash1 The Il otilde maussaq Groupsamples have higher SrNd LaNb and ZrNbratios than HIMU and overlap with the upperrange of EM-1 ratios The EF basalts with eNd(i)lt ndash1 have generally higher ratios than the oceanicisland basalts and partly overlap with typicalvalues of the lower crust It therefore seems likelythat assimilation of lower crustal material mayhave changed the primary trace element ratios inthe basalts Therefore we will evaluate thepossible in uence of assimilation of crustalmaterial on the geochemistry of the EF basaltsin the following section

846

R HALAMA ETAL

The role of crustal contamination

Large amounts of bulk assimilation of crustalmaterial which is relatively SiO2-rich comparedto a basaltic melt would drive the compositions tohigher SiO2 contents and quartz-normativecompositions Partial melts derived from crustalrocks would also be relatively SiO2-rich (Beard etal 1993 Carroll and Wyllie 1990) and wouldhave the same effect The ne-normative characterof the EF basalts (Table 1) suggests thatassimilation of crustal material especially ofsilica-rich upper crustal rocks was not signi cantwith the possible exception of sample EF 059

On the primitive mantle-normalized multi-element diagrams (Fig 4) Ti troughs are typicalfor crustal rocks especially for the Upper Crust(Rudnick and Fountain 1995) and they maytherefore be present in rocks extensively contami-nated by continental crust Ti troughs aregenerally lacking in the patterns of the EFbasalts and a major upper crustal contribution tothe magmas can therefore be excluded However

small amounts of assimilation of lower crustalmaterial cannot be ruled out because the averagelower crust has higher Ti contents

Steep trends on Sr-Nd isotope plots have beenattributed to assimilation of granulite-faciesgneisses (eg Bernstein et al 1998) AFCmodelling based on an equation developed byDePaolo (1981) was performed to test thehypothesis of small amounts of contaminationwith lower crustal material (Fig 9) The basalticmagma which represents one end-member for thecalculations has the isotopic signature of theisotopically most primitive samples from thepresent data set and from the literature(Table 6) Two possible contaminants repre-senting upper and lower crustal compositionswere selected for modelling (Table 6) Therepresentative samples are a granite from theJulianehaEcirc b batholith for the upper crust (sampleJG 02) and an average of ve granulite-faciesgneisses with low RbSr ratios from the Archaeancraton of West Greenland (Taylor et al 1984) for

TABLE 5 Trace element ratios of EF basalts compared to ocean island basalts(OIBs) representing EM-1 and HIMU mantle components and the lowercrust

SrNd LaNb ZrNb

EF basalts with eNd lt ndash 1 249 ndash396 120 ndash150 121 ndash157Il otildeacutemaussaq Group EF basalts 198 ndash292 104 ndash119 51 ndash53EM-1 90 ndash207 067 ndash111 39 ndash109HIMU 45 -163 063 ndash080 27 ndash 36Lower continental crust 30 16 136

Data sources EM-1 Humphris and Thompson (1983 Walvis Ridge) and Cliff et al(1991 Inaccessible Islands) HIMU Chauvel et al (1992 Tubuai Island) Lowercontinental crust Rudnick and Fountain (1995)

TABLE 6 End-member compositions used for AFC modelling

End-member Sr (87Sr86Sr)i Nd eNd(i) Data source(ppm) (T = 12 Ga) (ppm) (T = 12 Ga)

Primitive EF basaltic 507 070278 18 + 22 Sample EF 072magma Paslick et al (1993)Upper crust 300 071665 62 ndash83 Sample JG 02(JulianehaEcirc b Granite)Lower crust 747 070668 63 ndash 234 Taylor et al (1984)(Granulite facies gneiss)

The composition of the granulite-facies gneiss is an average of 5 samples with low RbSr ratios lt02

PETROGENESIS OF THE ERIKSFJORD BASALTS

847

the lower crust An in uence of Archaean rockson Gardar magmas has been demonstrated by Pbisotope studies (Taylor and Upton 1993) andseismic data were also interpreted to re ect theexistence of a wedge of Archaean crust thatextends southwards to Lindenow Fjord (Dahl-Jensen et al 1998) We have shown in theprevious sections on the basis of Ni trends thatfractionation of olivine was most relevant to themajor element evolution of the lavas whereasfractionation of other phases was apparently ofminor importance We therefore modelled theAFC processes with olivine fractionation usingmineral-melt partition coef cients of 0014 for Srand 00066 for Nd (Rollinson 1993)

The modelling of AFC processes is stronglydependent on the parameter r de ned as the ratioof mass assimilation rate and fractional crystal-lization rate (DePaolo 1981) The change inisotopic signature becomes enhanced for highervalues of r Reiners et al (1995) have shown thatr values might be very high and even gt1 in theearly stages of magmatic evolution at hightemperatures since crystallization can besuppressed by assimilation Therefore we calcu-

lated AFC processes with two r values of 05 and08 that represent a relatively high and a middle tolow assimilation rate to demonstrate the effects ofvariable r (Fig 9) The results indicate that AFCprocesses involving upper crust similar to theJulianehaEcirc b Granite cannot explain the trend of theEF basalts because Sr isotopic compositionswould shift to considerably more radiogenicvalues This result is consistent with theconclusions derived from trace element dataThe AFC curves for assimilation of lower crusthowever agree well with the bulk of the data Themost negative eNd values can be explained byAFC processes with intermediate assimilationrates (r = 05) involving assimilation of lt5lower crustal material With a greater degree ofassimilation and an r value of 08 lt4assimilation is indicated

Lower crustal rocks of igneous origin have anaverage d18O value of +75plusmn14 with an overallrange from +54 to +125 (Fowler and Harmon1990) 5 to 10 bulk assimilation of lower crustalmaterial with d18O = +75 into a typical mantle-derived magma with d18O = +57 results ind18O values of the contaminated magma from

-24

-20

-16

-12

-8

-4

0

4

0700 0703 0706 0709 0712 0715 0718

r = 08

r = 05

r = 05

r = 08

EF Basalts

Endmembers of AFC calculations

Upper Crust

Lower Crust

Primitive EF basaltic magma

(87Sr86Sr)i

e Nd

(i)

09

08

06

09

08

06

09

0806

09

08

07

07

04

Typical errorCHUR

BS

E

FIG 9 eNd(i) vs (87Sr86Sr)i diagram showing the results of AFC modelling End-member compositions are listed inTable 6 Olivine is the only fractionating mineral Partition coef cients for olivinemelt are from the compilation ofRollinson (1993) Curves were calculated for two values of r (r = rate of assimilation rate of fractionalcrystallization) Dashed lines and numbers in italics are for r = 05 solid lines and regular numbers are for r = 08

Values at the curves are for F values (fraction of melt remaining) with 10 tick marks

848

R HALAMA ETAL

+579 to +588 Therefore the d18O data cannotbe used to rule out small amounts of assimilationof lower crust To evaluate how much contamina-tion would be permissible from the presentoxygen isotope data we calculated how muchcontamination of a mantle-derived magma with18O = +57 with a contaminant within the lowerrange of lower crustal xenoliths (d18O = +67)would drive the oxygen isotopic composition tothe highest value of the EF basalts (d18O~+63) Although simple mixing shows that60 assimilation is permissible this amountwould drive the whole-rock geochemistry tomore silicic compositions than is observed in theEF basalts In Fig 8 there is no obvious trendfrom the basalts towards the Ketilidian basementsuggesting again that upper crust was notassimilated in signi cant amounts

Ratios of trace elements incompatible inbasaltic systems can be used to distinguishbetween crust and mantle reservoirs (egWeaver 1991) Based on the AFC modellingEF basalts with eNd(i) lt ndash1 are those which aremost likely in uenced by crustal contaminationThe relatively high LaNb and ZrNb ratios ofthese samples are consistent with assimilation oflower crustal material SrNd ratios are evenhigher than the average lower crustal composi-tion but SrNd ratios in those samples which weretaken as representative for the lower crust in theAFC modelling can reach values of 80 ie wellabove the highest EF basalt ratios Again theSrNd ratios support the hypothesis of some lowercrustal contamination

The absence of more primitive basaltic rockswith MgO gt8 wt might also indicate lowercrustal assimilation It suggests that magmas mayhave been retained at depth where they acquiredmore evolved compositions by fractional crystal-lization and had the opportunity to assimilatelower crustal material during crustal underplating(Upton et al 2003)

Summary and conclusions

The new geochemical data of the EriksfjordFormation basalts in the Gardar Provinceprovided in this study are used to placepetrogenetic constraints on the sources of theirparental magmas Based on trace-element patternsand Sr-Nd isotopic constraints there seems to beno indication that a DMM-like source played animportant role in magma genesis of the EFbasalts Clinopyroxene and whole-rock REE

patterns are more compatible with an OIB-likesource but we are not able to conclude whetherthe observed enrichment in REE re ects thesourcersquos signature or magmatic processes Theisotopically most primitive EF basalts have Sr-Ndisotopic compositions that are similar to Gardarcarbonatites Due to the isotopic similaritybetween carbonatites and OIBs many arethought to be related to mantle plumes (Bell2001) We therefore speculate that OIB-likemantle components from a plume source werealso involved in the parental EF magmas but wecannot exclude interaction with the SCLM Twodistinct mantle components one isotopicallydepleted and another one enriched could havecontributed to the EF magmas AFC modelling ofSr-Nd isotope data indicates that assimilation oflt5 lower crust could explain the steep vector onthe Sr-Nd isotope plot but assimilation of uppercrustal material is unlikely Oxygen isotope dataof clinopyroxene separates are consistent with amantle origin but permit some assimilation oflower crustal rocks In summary the data areconsistent with the hypothesis that the EFmagmas could have originated from a mantleplume source which is supported by geologicalarguments and that they underwent someassimilation of lower crustal material

Acknowledgements

We would like to thank Bruce Paterson whoprovided invaluable help during Laser ICP-MSmeasurements at the Large-Scale GeochemicalFacility supported by the European Community-Access to Research Infrastructure action of theImproving Human Potential Programme contractnumber HPRI-CT-1999-00008 awarded to ProfBJ Wood (University of Bristol) Gabi Stoschekand Torsten Vennemann are thanked for their helpwith oxygen isotope measurements Elmar Reitterexpertly assisted with preparation and measure-ments of radiogenic isotopes and MathiasWestphal helped with microprobe measurementsREE whole-rock analyses were kindly providedby the NERC facility at Royal HollowayUniversity of London Michael Marks is thankedfor his pleasant company during eldwork Veryconstructive reviews by Tom Andersen andAlexei S Rukhlov greatly helped to improve themanuscript We also thank Ian Coulson for nalediting Funding of this work by the DeutscheForschungsgemeinschaft (grant Ma-21351-2) isgratefully acknowledged

PETROGENESIS OF THE ERIKSFJORD BASALTS

849

References

Andersen T (1997) Age and petrogenesis of theQassiarsuk carbonatite-alkaline silicate volcaniccomplex in the Gardar rift South GreenlandMineralogical Magazine 61 499 ndash513

Armstrong JT (1991) Quantitative elemental analysisof individual microparticles with electron beaminstruments Pp 261 ndash315 in Electron ProbeQuantitation (KFJ Heinrich and DE Newburyeditors) Plenum Press New York and London

Arndt NT and Christensen U (1992) The role oflithospheric mantle in continental ood volcanismThermal and geochemical constraints Journal ofGeophysical Research 97 10967 ndash 10981

Baker JA Menzies MA Thirlwall MF andMacpherson CG (199 7) Pet rogenes is ofQuaternary Intraplate Vocanism Sanarsquoa YemenImplications for Plume-Lithosphere Interaction andPolybaric Melt Hybridization Journal of Petrology 38 1359 ndash 1390

Baker JA MacPherson CG Menzies MAThirlwall MF Al-Kadasi M and Mattey DP(2000) Resolving crustal and mantle contributions tocontinental ood volcanism Yemen constraintsfrom mineral oxygen isotope data Journal ofPetrology 41 1805 ndash1820

Beard JS Abitz RJ and Lofgren GE (1993)Experimental melting of crustal xenoliths fromKilbourne Hole New Mexico and implications forthe contamination and genesi s of magmasContributions to Mineralogy and Petrology 11588 ndash102

Bell K (2001) Carbonatites relationships to mantle-plume activity Geological Society of AmericaSpecial Paper 352 267 ndash290

Bell K and Blenkinsop J (1989) Neodymium andStrontium Isotope Geochemistry of Carbonatites Pp278 ndash300 in Carbonatites ndash Genesis and Evolution(K Bell editor) Unwin Hyman London

Bell K and Simonetti A (1996) Carbonat iteMagmatism and Plume Activity Implications fromthe Nd Pb and Sr Isotope Systematics of OldoinyoLengai Journal of Petrology 37 1321 ndash1339

Bell K and Tilton GR (2001) Nd Pb and Sr IsotopicCompositions of East African Carbonat itesEvide nce for Mant le Mixing and PlumeInhomogene ity Journa l of Petrology 42 1927 ndash1945

Benoit M Polve M and Ceuleneer G (1996) Traceelement and isotopic characterization of ma ccumulates in a fossil mantle diapir (Oman ophiolite)Chemical Geology 134 199 ndash 214

Bernstein S Kelemen PB Tegner C Kurz MDBlusztajn J and Kent Brooks C (1998) Post-breakup basaltic magmatism along the EastGreenland Tertiary rifted margin Earth and

Planetary Science Letters 160 845 ndash862Boynton WV (1984) Geochemistry of the rare earth

elements meteorite studies Pp 63 ndash114 in RareEarth Element Geochemistry (P Henderson editor)Elsevier Amsterdam

Carroll MR and Wyllie PJ (1990) The systemtonalite-H2O at 15 kbar and the genesis of calc-alkaline magmas American Mineralogist 75345 ndash 357

Chauvel C Hofmann AW and Vidal P (1992)HIMU-EM The French Polynesian connectionEarth and Planetary Science Letters 110 99 ndash119

Clayton RN and Mayeda TK (1963) The use ofbromine penta uoride in the extraction of oxygenfrom oxides and silicates for isotope analysisGeochimica et Cosmochimica Acta 27 43 ndash52

Coulson IM and Chambers AD (1996) Patterns ofzonation in rare-earth-bearing minerals in nephelinesyenites of the North Qoroq center South GreenlandThe Canadian Mineralogist 34 1163 ndash1178

Cox KG Bell JD and Pankhurst RJ (1979) TheInterpretation of Igneous Rocks George Allen andUnwin London

Cross W Iddings JP Pirsson LV and WashingtonHS (1903) Quantitative Classi cation of IgneousRocks University of Chicago Press

Dahl-Jensen T Thybo T Hopper H and Rosing M(1998) Crustal structure at the SE Greenland marginfrom wide-angle and normal incidence seismic dataTectonophysics 288 191 ndash 198

DePaolo DJ (1981) Trace element and isotopic effectsof combined wallrock assimilation and fractionalcrystallisation Earth and Planetary Science Letters53 189 ndash 202

De P a o lo D J (1 9 8 8 ) Neodymium Iso topeGeochemistry An Introduction Springer-VerlagNew York

Dobosi G Downes H Mattey D and Embey-IsztinA (1998) Oxygen isotope ratios of phenocrysts fromalkali basalts of the Pannonian basin Evidence foran O-isotopically homogeneous upper mantle be-neath a subduction-in uenced area Lithos 42213 ndash 223

Eiler JM (2001) Oxygen isotope variations of basalticlavas and upper mantle rocks Pp 319 ndash364 inStable Isotope Geochemistry (JW Valley and DRCole editor s) Reviews in Minera logy andGeochemistry 43 The Mineralogical Society ofAmerica Washington DC

Eiler JM Farley KA Valley JW Hauri E CraigH Hart SR and Stolper EM (1997) Oxygenisotope variations in ocean island basalt phenocrystsGeochimica et Cosmochimica Acta 61 2281 ndash2293

Eiler JM Crawford A Elliot T Farley KAValley JW and Stolper EM (2000a) Oxygenisotope geochemistry of oceanic-arc lavas Journal

850

R HALAMA ETAL

the per mil deviation from Vienna Standard MeanOcean Water (V-SMOW) in the standard d-notation Replicate oxygen isotope analyses ofthe NBS-28 quartz standard (Valley et al 1995)had an average precision of plusmn01 for d18O Ineach run standards were analysed at thebeginning and the end of the sample set Acorrection was applied to the data equal to theaverage of the difference between the meanmeasured value and the accepted value for thestandard (964)

Oxygen isotope compositions of powderedwhole-rock samples were determined by aconventional method modi ed after Clayton andMayeda (1963) using BrF5 as reagent andconverting the liberated oxygen to CO2 beforemass spectrometric analysis

Petrography

175 samples were collected from all threevolcanic units of the Eriksfjord Formation butmost of them were extensively altered Theycontained secondary chlorite Fe-talc Ca-Fesilicates and carbonate as alteration productsSince the focus of this study was in situ analysesof minerals or analyses of mineral separates onlynine samples with a signi cant proportion of freshclinopyroxene and relatively few alterationproducts were selected for analysis Due to thelimited number of suitable samples we did notattempt to investigate the lavas in terms of theirstratigraphic position In addition to the ninesamples from which pyroxene separates wereobtained REE whole-rock data of six othersamples were added to the data set

In general the Mussartut Group lavas areolivine-free but contain up to 50 chloriticmatrix that probably results from devitri cationof former glass In contrast the majority ofUlukasik and Il otilde maussaq Group lavas containolivine-pseudomorphs and tend to be coarsergrained and holocrystalline Many of themcontain lsquosnow- akersquo glomerocrysts of plagioclaseindicating plagioclase-saturation or super-satura-tion at near-surface levels (Upton 1996) Thesamples from which clinopyroxenes have beenseparated are ne-grained basalts with plagioclaselaths embedded in interstitial clinopyroxeneas themajor and apatite and Fe-Ti oxides as subordinateprimary magmatic mineral phases The texturecan be described as ophitic and is believed toresult from the cotectic crystallization of plagio-clase and clinopyroxene(Philpotts 1990) Olivine

has been present in a few samples but it is nowcompletely altered Pyrite pyrrhotite and chalco-pyrite occur as accessory phases Some samplescontain two oxides (ilmenite + titanomagnetite)whereas others contain a single oxide exsolvedinto ilmenite and titanomagnetite or ilmenite andhematite Texturally they are mostly subhedral toanhedral although few euhedral grains occurHematite is also present around both discrete andexsolved ilmenite and titanomagnetite grains andthe exsolved hematite might be entirely ofsecondary origin formed by oxidation of titano-magnetite Late-stage interstitial quartz occursrarely Sample EF 108 (Ulukasik Group) isremarkable in having an intense reddish colourunlike the other samples that are almost black

Results

Mineral chemistryTypical analyses of clinopyroxenes are given inTable 1 Compositions range from En41Wo44Fs16

to En28Wo45Fs27 and broadly overlap for allgroups (Fig 2) According to the classi cation ofMorimoto et al (1988) clinopyroxenes of theMussartut and Ulukasik Groups are mainlyaugites with few diopsides present whereas inthe Il otilde maussaq Group samples of diopsidiccompositions dominate over augitic onesClinopyroxene analyses from the varioussamples form tight clusters on the pyroxenequadrilateral except for those from sample EF108 that are more scattered With only a singlepyroxene phase present in all samples clino-pyroxene compositions indicate minimum crystal-lization temperatures (Lindsley 1983) of750 ndash1000ordmC (Fig 2) Plagioclase is normallyzoned and its composition typically variesbetween An70Ab29Or1 and An16Ab72Or12 Lateinterstitial albite can reach compositions ofAn0Ab99Or1 There are no signi cant composi-

1200 Ecirc C

800 Ecirc C

25 50 75

25 25

En Fs

HdDi

Mussartucirct Group

Iliacutemaussaq Group

Ulukasik Group

FIG 2 Clinopyroxene analyses of the Eriksfjord basaltsplotted in the pyroxene quadrilateral with isotherms

after Lindsley (1983)

PETROGENESIS OF THE ERIKSFJORD BASALTS

835

tional changes among the various samplesIlmenite from two samples (EF 024 and EF 039)which contain coexisting ilmenite and titanomag-netite includes a signi cant component ofpyrophanite (MnTiO3) with up to 87 wtMnO It s com pos i t i ons va r y be t weenI l m 8 6 H e m 5 P y r 8 a nd I l m 7 6 H e m 5 P yr 1 9 Titanomagnetite analyses were unsatisfactory asall grains show strong alteration They areconsidered unreliable and are not presented here

Whole-rock geochemistry and classif|cation

Concentrations of major and trace elements aresummarized in Table 2 The basalts can beclassi ed according to the TAS diagram (LeMaitre et al 1989) as basalts and trachybasaltswith both alkaline and subalkaline compositions(not shown) Due to possible mobility of the Naand K the samples were also plotted into NbYvs ZrP2O5 and P2O5 vs Zr diagrams (Floyd andWinchester 1975) which can discriminate

between tholeiitic and alkaline series Thesediagrams do not yield consistent discriminationsfor the EF basalts but indicate that the characterof the sample suite is transitional betweentholeiitic and alkaline (not shown)

Mg [100Mg(Mg+Fe2+)] calculated assumingFe2O3FeO = 02 as recommended for basalts byMiddlemost (1989) varies between 54 and 40(Table 2) Ni concentrations range from 46 to115 ppm and Cr contents from 29 to 126 ppmTaking Mg as fractionation index a positivecorrelation (r2 = 061) with Ni contents indicatesfractionation of olivine (Fig 3) The relativelypoor positive correlations of Mg with Cr (r2 =044) and Sc (r2 = 023) argue against a signi cantfractionation of Cr-spinel or clinopyroxene Thisis consistent with the lack of clinopyroxenephenocrysts and the absence of any Cr-spinelThe negative correlations of Mg with Sr TiO2

and P2O5 suggest that fractionation of plagioclaseFe-Ti oxides and apatite was not important in thesample set investigated which is again consistent

TABLE 1 Typical clinopyroxene analyses of Eriksfjord Formation basalts determined by EMP

Sample EF 024 EF 024 EF 039 EF 059 EF 072 EF 108 EF 168 EF 174Cpx no cpx 2 cpx 3 cpx 5 cpx 3 cpx 5 cpx 7 cpx 4 cpx 1

SiO2 5060 5041 5128 4988 5145 4950 5175 5069TiO2 173 176 121 169 140 162 096 121Al2O3 297 300 179 223 171 264 155 213FeO 1160 1068 1218 1405 993 1234 979 1129MnO 028 021 030 040 019 028 025 031MgO 1254 1324 1263 1182 1325 1163 1307 1214CaO 1959 2029 2002 1907 2106 2099 2174 2143Na2O 034 031 029 049 033 055 052 054

Total 9965 9989 9971 9963 9932 9955 9962 9974

Formulae based on 4 cations and 6 oxygensSi 191 189 194 190 194 188 194 191Al 013 013 008 010 008 012 007 009Ti 005 005 003 005 004 005 003 003Fe3+ - - - 003 - 007 003 005Mg 071 074 071 067 075 066 073 068Fe2+ 037 033 039 041 031 032 028 031Mn 001 001 001 001 001 001 001 001Ca 079 082 081 078 085 085 087 087Na 002 002 002 004 002 004 004 004

Total 400 400 400 400 400 400 400 400

Pyroxene projection after Lindsley (1983)Wo 039 041 041 041 043 043 045 044En 040 041 039 037 040 038 040 038Fs 021 018 021 023 017 019 015 017

836

R HALAMA ETAL

TABLE 2 XRF whole-rock analyses of Eriksfjord Formation basalts

Sample EF 024 EF 063 EF 061 EF 039 EF 059 EF 108 EF 072 EF 174 EF 168Group Mussartut Mussartut Mussartut Mussartut Mussartut Ulukasik Ulukasik Ilotildeacutemaussaq IlotildeacutemaussaqSubunit 4 4 5 6 7 Lower Unit Upper Unit Sermilik Narssaq

Fjeld

Major elements (wt)SiO2 4573 4585 4688 4660 4875 4698 4656 4601 4545TiO2 230 200 180 175 191 265 203 299 277Al2O3 1578 1630 1600 1612 1645 1682 1668 1645 1719Fe2O3 1510 1356 1331 1383 1225 1394 1321 1438 1392MnO 028 018 018 018 017 014 017 018 017MgO 525 683 653 673 568 396 647 440 480CaO 842 849 872 857 920 641 808 746 737Na2O 349 274 289 276 236 448 303 323 376K2O 087 083 068 032 027 159 070 205 132P2O5 041 035 023 023 023 048 035 111 083LOI 199 210 240 238 242 204 210 052 156

Total 9962 9923 9962 9947 9969 9949 9938 9878 9914

Mg 453 545 538 536 524 403 538 421 450

Trace elements (ppm)Sc 24 22 25 22 26 19 20 21 16V 209 175 215 195 211 142 170 164 203Cr 81 108 66 65 126 48 82 33 29Co 63 78 70 62 71 44 68 35 46Ni 99 115 85 85 101 58 106 46 51Cu 65 49 46 60 97 41 59 38 41Zn 103 102 110 97 110 93 82 122 105Ga 20 18 21 19 20 19 16 21 19Rb 24 26 16 lt 10 lt 10 38 12 39 45Sr 504 503 512 392 373 712 507 850 992Y 31 26 28 23 28 31 17 31 25Zr 188 165 121 116 125 125 88 164 139Nb 12 11 10 lt 10 10 9 lt 10 31 27Ba 445 375 415 302 178 582 405 1293 900Pb 6 7 3 4 3 4 4 3 5La ndash ndash 15 ndash 12 11 ndash 37 28Ce ndash ndash 40 ndash 38 39 ndash 92 71Pr ndash ndash 1 ndash 1 2 ndash 9 4Nd ndash ndash 16 ndash 15 18 ndash 43 34Sm ndash ndash 6 ndash 7 6 ndash 5 5

CIPW normq ndash ndash ndash ndash 313 ndash ndash ndash ndashor 514 491 402 189 160 940 414 1212 780ab 2807 2319 2445 2335 1997 3207 2564 2733 2981an 2482 2973 2868 3065 3350 2109 2985 2433 2613ne 079 ndash ndash ndash ndash 316 ndash ndash 109di 1187 830 1076 853 877 633 654 452 416hy ndash 890 1059 1650 2217 ndash 1097 315 ndashol 1670 1309 1044 780 ndash 1472 1120 1388 1689mt 365 328 322 335 296 336 319 348 336il 437 380 342 332 363 503 386 568 526ap 095 081 053 053 053 111 081 257 192

Mg was calculated as Mg = 100Mg2+(Mg2++Fe2+) using Fe2O3FeO = 02 Normative mineral compositionswere determined according to the CIPW calculation scheme

PETROGENESIS OF THE ERIKSFJORD BASALTS

837

1 5

2 0

2 5

3 0

3 5

40

60

80

100

120

0

0 5

1

1 5

200

400

600

800

1000

0

50

100

150

15

20

25

30

4 0 45 50 5 5

r2 = 023

r2 = 044r2 = 061

r2 = 053

r2 = 082 r2 = 052

Ni (ppm)

Sc (ppm)

P2O5 (wt)

Sr (ppm)

TiO2 (wt)

Cr (ppm)

Mg Mg

Mg Mg

Mg

40 4 5 50 55

Mg

40 45 50 55

40 45 50 55

40 45 50 5540 4 5 50 55

FIG 3 Whole-rock variation diagrams for EF basalts with Mg as fractionation index Mg is calculated asMg(Mg+Fe2+) with Fe2O3FeO = 02 (Middlemost 1989)

838

R HALAMA ETAL

with the mostly subhedral Fe-Ti oxides and theophitic texture Ratios of incompatible traceelements such as ZrY (range 40 ndash63) andTiY (range 385 ndash716) are relatively uniformthroughout all samples but some scatter isevident Notable geochemical features are highSr contents up to 1000 ppm and very high Bacontents up to 1300 ppm

Normative mineral compositions were deter-mined according to the CIPW calculation scheme(Cross et al 1903 Cox et al 1979) Mostsamples are hypersthene-normative some arenepheline-normative and one (EF 059) is quartz-normative (Table 2) Despite the uncertaintiesinvolved in the CIPW norm calculation the dataindicate that some of the basalts crystallized fromsilica-undersaturated magmas and others fromdistinct silica-saturatedor -oversaturatedmagmas

Multi-element diagrams normalized to primi-tive mantle values show that the EF basalts areenriched in all incompatible elements relative toprimitive mantle values (Fig 4) The patterns ofthe EF basalts are characterized by a prominentBa peak a subordinate P peak and a decreasingelement enrichment with increasing compatibilityIn comparison with normal MORB (N-MORB)enriched MORB (E-MORB) and average OIB(Sun and McDonough 1989) the patterns of theEF basalts are closest to those of OIB both inshape and normalized concentrations

Whole-rock REE data

Whole-rock REE data are presented in Table 3and REE concentrations normalized to chondriticvalues of Boynton (1984) are shown in Fig 5aAll samples are characterized by LREE-enrichedpatterns with no or slightly positive Euanomalies (EuEu = 097 to 113) La concentra-tions are roughly 20 ndash60 times and Lu concentra-tions are 10 times chondritic values NormalizedLREE abundances from La to Nd are fairlyconstant but the patterns show a steady andsmooth decline from Nd towards heavier REEThe REE patterns of the EF basalts fall in betweenthose of typical OIB and E-MORB (Fig 5b)They have similar patterns and absolute abun-dances compared to basaltic dykes from theIvittuut area South Greenland (Goodenough etal 2002) LaNSmN as a measure of LREEfractionation has a restricted range from 130 to165 The degree of HREE fractionationexpressed as GdNYbN is slightly more variableand ranges from 157 to 257 Both these ratios of

the EF basalts fall generally between typical OIBand E-MORB compositions

Clinopyroxene REE patterns

Typical analyses of REE concentrations inclinopyroxenes are given in Table 3 They showthat REE concentrations in individual minerals ofone particular sample can be quite variableChondrite-normalized REE patterns of clinopyr-oxenes are very similar in the samples from theMussartut and Ulukasik Groups (Fig 6ab) Theyare enriched in all REE relative to chondriticvalues and the patterns show a fairly steepincrease from La to Sm followed by a gradualsmooth decrease from Sm to Lu Clinopyroxenesfrom sample EF 168 (Il otildeacutemaussaq Group) aredistinct as they have generally higher REEcontents especially in the LREE and a weak Euanomaly (Fig 6c) The 3 ndash106 enrichment in Nd

c (sa

mpl

e)c

(prim

itive

man

tle)

c (sa

mpl

e)c

(prim

itive

man

tle)

1

10

100

1000

Mussartucirct GroupUlukasik GroupIliacutemaussaq Group

Rb Ba K Nb La Ce Sr Nd P Zr Ti Y

Rb Ba K Nb La Ce Sr Nd P Zr Ti Y01

1

10

100

1000N-MORBE-MORBOIB

FIG 4 Multi-element plots normalized to primitivemantle values (McDonough and Sun 1995) Data forN-MORB E-MORB and average OIB (Sun and

McDonough 1989) are shown for comparison

PETROGENESIS OF THE ERIKSFJORD BASALTS

839

TA

BL

E3

RE

Eda

taof

who

lero

cks

and

clin

opyr

oxen

es

the

latt

erde

term

ined

byL

aser

ICP

-MS

Who

le-r

ock

anal

yses

R

epre

sent

ativ

ecl

inop

yrox

ene

anal

yses

S

ampl

eno

10

1231

1014

4610

1457

1014

7218

6389

1863

95E

F02

4-2

EF

039-

1E

F03

9-2

EF

039-

4E

F07

2-2

EF

072-

3E

F16

8-3

EF

168-

5

RE

Eco

ncen

trat

ions

(ppm

)L

a11

65

180

610

29

661

155

412

28

440

266

199

150

273

304

165

527

59

Ce

290

346

05

249

617

13

378

130

08

180

411

46

868

717

105

012

60

572

399

67

Pr

434

67

388

268

527

433

331

216

181

128

249

262

102

917

43

Nd

200

631

71

176

412

49

250

920

16

190

213

82

122

28

9815

64

157

258

51

960

6S

m4

756

974

043

215

914

826

664

194

283

276

077

0615

82

265

9E

u1

742

51

571

071

971

721

831

491

111

212

001

943

877

07G

d4

826

584

043

555

84

996

965

615

244

326

156

0514

75

252

7T

b ndash

ndash ndash

ndash ndash

ndash1

351

010

770

711

111

212

313

61D

y4

475

223

613

565

354

816

865

934

784

255

996

8613

10

205

2H

o0

921

040

730

731

11

011

481

401

001

001

211

292

394

24E

r2

432

531

951

972

872

663

893

342

642

112

993

646

539

60T

m ndash

ndash ndash

ndash ndash

ndash0

450

390

360

290

350

440

861

27Y

b2

182

071

741

832

592

363

782

812

492

452

703

016

788

89L

u0

340

310

260

270

390

350

460

410

300

310

520

411

251

60

(Eu

Eu

) N1

111

131

190

971

031

070

820

940

720

991

000

910

770

83

Eu

anom

aly

isca

lcul

ated

as(E

uE

u) N

=E

u N(

SmN6

Gd N

)05

norm

aliz

edto

chon

drit

eva

lues

from

Boy

nton

(198

4)

840

R HALAMA ETAL

in clinopyroxenes of EF 168 compared to theother samples (EF 024 039 and 072) seems onlypartly due to a higher degree of fractionationsincethe Mg value of EF 168 is only slightly lowerThus the LREE enrichment appears to underlinethe more alkaline character of this sample Acomparison with published REE patterns ofclinopyroxenes (Fig 6d) shows that the EFpatterns are unlike patterns from olivine gabbroswith MORB-like signatures (Benoit et al 1996)but fairly similar to clinopyroxene megacrystpatterns from an OIB-like alkaline basalt (Nimisand Vannucci 1995)

Sr Nd and O isotopic composition

Sr Nd and O isotopic compositions of all samplesare presented in Table 4 Sr and Nd isotopiccompositions of the clinopyroxene separates havebeen recalculated to an approximate eruption age

of 12 Ga (Paslick et al 1993) using decayconstants of 142610ndash11 andash1 for 87Rb (Steigerand Jager 1977) and 654610ndash12 a ndash1 for 147Sm(Lugmair and Marti 1978) Present day CHURvalues of 01967 for 147Sm144Nd (Jacobsen andWasserburg 1980) and 0512638 for 143Nd144Nd(Goldstein et al 1984) were used to calculateinitial eNd values 87Sr86Sr initial ratios are fairlyconstant and range from 070278 to 070383(Table 4) eNd(i) values vary from ndash32 to +21(Table 4) On the Sr-Nd isotope diagram the EFbasalts de ne a relatively steep trend (Fig 7) withinitial 87Sr86Sr ratios similar to those of the BulkSilicate Earth (BSE) Only one sample is slightlydisplaced towards a more radiogenic initial87Sr86Sr ratio None of the samples has an eNd

value close to depleted mantle values estimatedto be between eNd = +53 (calculated afterDePaolo 1981) and eNd = +73 (calculated afterGoldstein et al 1984) at 12 Ga The two alkalinebasalts of the Il otilde maussaq Group are within therange of the other data but they de ne a restricted eld around CHUR and BSE respectively Errorsinduced by the uncertainty in the age of the EFbasalts are largely within the error of the isotopeanalyses and do not change the overall picture Incomparison with other published Sr-Nd isotopedata from the Gardar Province the initial Srisotope ratios of the EF basalts are very similar torelatively primitive basaltic lamprophyric andcarbonatitic dyke rocks which have (87Sr86Sr)i of~0703 (Pearce and Leng 1996 Andersen 1997Goodenough et al 2002) So far no evidence hasbeen presented for an isotopicallyenriched mantlesource in the Gardar Province as the existingeNd(i) data of ndash05 to +52 show an isotopicallydepleted signature (Pearce and Leng 1996Andersen 1997 Goodenough et al 2002Upton et al 2003) Clinopyroxene from the EFsamples of Paslick et al (1993) yield eNd(i) of+22 The Nd isotope data of this study are withinthe lower range of published values from theGardar Province and extend the spread to morenegative eNd(i) values

Several recent oxygen isotope studies haveshown that mineral separates are less sensitive tolow-T alteration and secondary water take-up thanwhole-rock samples (eg Vroon et al 2001)However only few oxygen isotope data ofmineral separates from continental volcanics areavailable (eg Dobosi et al 1998 Baker et al2000) Extensive studies of mineral separates ofocean island basalts (Eiler et al 1997) andoceanic-arc lavas (Eiler et al 2000a) demon-

c (sa

mpl

e)c

(cho

ndri

te)

c (sa

mp

le)c

(cho

ndrit

e)

1

10

100

200

101446

101472

101231

101457

186389

186395

La Ce Pr Nd Sm Eu Gd Dy Ho Er Yb Lu

1

10

100

200

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

N-MORBE-MORBOIBBasaltic dykes from Ivittuut

a

b

FIG 5 (a) Whole-rock REE data of EF basaltsnormalized to chondritic values from Boynton (1984)(b) Typical compositions of N-MORB E-MORB andOIB (Sun and McDonough 1989) are shown for

comparison

PETROGENESIS OF THE ERIKSFJORD BASALTS

841

strated that analyses of mineral separates consis-tently span a narrower range than comparablewhole-rock data In our study we used clino-pyroxene because it is the only fresh mineralphase present in the samples Oxygen isotopemeasurements of clinopyroxene separates fromthe EF basalts show a range in d18Ocpx from +52to +60 with an average value of +566plusmn023(1s) All samples overlap the range of d18Ocpx

values from mantle peridotites (+525 to +590Mattey et al 1994) from continental oodbasalts from Yemen (+55 to +69 Baker etal 2000) and from alkaline basalts of Hungary(+507 to +534 Dobosi et al 1998) There is asmall difference between the averages ofMussartut Group (d18Ocpx-avg = +575plusmn015(1s)) and Ulukasik Group (d18Ocpx-avg =+575plusmn009 (1s)) samples compared toIl otildeacutemaussaq Group (d18Ocpx-avg = +533plusmn021(1s)) samples

d18O values of melts coexisting with clino-pyroxene were calculated assuming that oxygen

in pyroxene and melt were in isotopic equilibriumand T was 1100ordmC Dmelt-cpx can then be calculatedusing the following equation of Kalamarides(1986)

Dmelt-cpx = ndash00008T + 143 (T in Kelvin)

Although fractionation varies with magmatictemperatures we consider the temperature of1100ordmC and the resulting Dmelt-cpx of +03 assuf ciently representative considering the analy-tical errors involved It is similar to Dmelt-cpx

values used in previous studies (+039 Vroonet al 2001 +02 Dobosi et al 1998)Resulting d18Omelt values of EF basalts rangefrom +55 to +63 These values can becompared to whole-rock data from continentalbasalts given by Harmon and Hoefs (1995)Intraplate basalts from continental rift zonesshow a range from +46 to +83 with a meanof +61 and continental ood basalts which arerather poorly represented in the database havevalues between +43 and +65 Calculated EF

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Cpx from MORB-like

olivine gabbro

Cpx megacrysts from OIB-like

alkaline basalts

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 024EF 039

Mussartucirct Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 168

Iliacutemaussaq Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

1

10

100

1000

EF 072

Ulukasik Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb LuLa Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

c

ba

d

1

10

100

01

FIG 6 REE patterns of clinopyroxenes (cpx) from Mussartut (a) Ulukasik (b) and Il otilde maussaq (c) Group samplesnormalized to chondritic values from Boynton (1984) (d) REE patterns of cpx from MORB-type olivine gabbro(Benoit et al 1996) and OIB-type alkaline basalt (Nimis and Vannucci 1995) for comparison (note the change of

scale)

842

R HALAMA ETAL

TA

BL

E4

Sr

Nd

and

Ois

otop

icco

mpo

siti

ons

ofcl

inop

yrox

ene

sepa

rate

sfr

omE

riks

fjor

dF

orm

atio

nba

salt

sW

hole

-roc

kan

alys

esfr

omth

eK

etil

idia

nba

sem

ent

and

from

sedi

men

tary

rock

sof

the

Eri

ksfj

ord

For

mat

ion

are

give

nfo

rco

mpa

riso

n

Sam

ple

Mem

ber

Sr

Rb

87R

b86S

r87S

r86Sr

87Sr

86Sr

init

ial

SmN

d147Sm

144N

d143N

d144N

de N

d(i

)d1

8O

(ppm

)(p

pm)

(ppm

)(p

pm)

Bas

alts

E

F02

4M

ussa

rtut

646

50

486

002

180

7042

01plusmn

100

7038

267

3822

80

019

560

5124

67plusmn

10ndash

32

567

EF

063

Mus

sart

ut5

60plusmn

011

EF

061

Mus

sart

ut89

95

099

90

0321

070

3818

plusmn10

070

3266

574

163

90

2115

051

2629

plusmn10

ndash2

55

79plusmn

028

EF

039

Mus

sart

ut73

74

058

30

0229

070

3492

plusmn10

070

3098

662

191

30

2092

051

2626

plusmn10

ndash2

25

71plusmn

001

EF

059

Mus

sart

ut54

47

013

70

0073

070

3375

plusmn12

070

3250

606

168

90

2171

051

2727

plusmn09

ndash1

45

99plusmn

001

EF

108

Ulu

kasi

k66

92

111

00

0480

070

3741

plusmn09

070

2916

109

936

82

018

050

5126

17plusmn

10+

21

581

EF

072

Ulu

kasi

k69

41

032

20

0134

070

3014

plusmn10

070

2784

818

251

90

1962

051

2574

plusmn10

ndash1

25

68E

F17

4Il

otildemau

ssaq

125

42

324

005

360

7039

35plusmn

100

7030

1418

76

796

10

1424

051

2230

plusmn10

+0

45

47E

F16

8Il

otildemau

ssaq

866

71

422

004

750

7037

99plusmn

070

7029

8316

45

641

60

1550

051

2304

plusmn09

ndash0

15

18

Bas

emen

tan

dse

dim

enta

ryro

cks

JG02

Juli

aneh

aEcirc b30

01

169

71

6375

074

4789

plusmn10

071

6647

105

362

28

010

220

5114

70plusmn

09ndash

83

78

Gra

nite

EF

021

Qua

rtzi

te11

5E

F02

8A

rkos

ic6

8ar

enit

e

Sam

ples

are

list

edin

stra

tigra

phic

orde

rfr

omol

dest

toyo

unge

st87

Sr86

Sr

init

ialr

atio

san

de N

d(i)

valu

esw

ere

calc

ulat

edfo

ran

age

of1

2G

aS

tand

ard

devi

atio

nsfo

rd18

Oar

egi

ven

for

sam

ples

that

wer

ean

alys

edtw

ice

PETROGENESIS OF THE ERIKSFJORD BASALTS

843

model melt oxygen isotope data are within theranges of both groups Compared to Igaliko dykerocks from further SE in the Gardar Province(Pearce and Leng 1996) the EF basalts cover amuch more restricted eld possibly because d18Owhole-rock analyses are more prone to secondaryalteration than analyses of mineral separates(Fig 8) Whole-rock d18O measurements of theunderlying JulianehaEcirc b Granite and of the sedi-mentary sandstone members of the EriksfjordFormation yield values of 78 and 68 ndash115respectively (Table 4) and are shown forcomparison

Discussion

Mantle source regions of the EF basaltsTo reconstruct the mantle sources involved in theformation of the EF basalts we use Sr-Nd-Oisotope and incompatible trace element data On aSr-Nd isotope diagram the isotopic compositionof DMM was modelled for an age of 12 Ga(DePaolo 1981 Goldstein et al 1984 Taylor andMcLennan 1985) The fairly constant 87Sr86Srinitial isotope ratios of the EF basalts are ~0703which falls within the range typical for mantle-derived rocks but slightly above calculated valuesfor DMM (Fig 7) eNd(i) values are more

-4

-2

0

2

4

6

8

0701 0702 0703 0704 0705 0706

DMM Typical 2 s error

CHUR

BS

E

(87Sr86Sr)i

e Nd

(i)

EM-1

Qassiarsuk carbonatites and lamprophyres

EF basalts

Depleted MORB mantle

Ivittuut basaltic and lamprophyric dikes

Igaliko carbonatites and lamprophyres

Enriched mantle 1

Carbonatites from other regions of similar age

FIG 7 eNd(i) vs (87Sr86Sr)i diagram calculated for t = 12 Ga which is considered as the eruption age of the EFbasalts (Paslick et al 1993) The reference line of Bulk Silicate Earth (BSE) for the Sr isotopic composition wascalculated after DePaolo (1988) assuming present-day values of 00827 for 87Rb86Sr and 07045 for 87Sr86SrIsotopic composition of the depleted MORB mantle (DMM) was calculated after DePaolo (1981) Goldstein et al(1984) and Taylor and McLennan (1985) EM-1 was calculated after Schleicher et al (1998) (see text for details)Data sources are Goodenough et al (2002) Ivittuut (n = 6) Pearce and Leng (1996) Igaliko (n = 5) Andersen(1997) Qassiarsuk (n = 25) Bell and Blenkinsop (1989) other carbonatites of similar age (Goudini Schryburt Lake

and Prairie Lake n = 6)

844

R HALAMA ETAL

scattered and indicate that the rocks contain amixture of isotopic components from both mantleand crustal reservoirs or a heterogeneous mantlesource Only one sample (EF 108) has an eNd(i)value signi cantly above 0 This sample exhibitsthe largest spread of clinopyroxene compositionswhich may re ect strong alteration Goodenoughet al (2002) also note that the highest eNd occursin the most altered sample of their sample setHowever Paslick et al (1993) report clino-pyroxene data from two Ulukasik Groupsamples with eNd(i) gt +2 and those sampleswere apparently not disturbed with regard to theirSm-Nd age information On the other hand Piperet al (1999) suggest that the Sm-Nd systematicsof those samples may be erroneous due to laterdisturbance by uids derived from the Il otildeacutemaussaqintrusion Therefore although we are presentlynot able to decide whether positive eNd(i) values

for the EF basalts are real it is clear that none ofthose values is close to values of a 12 Ga oldDMM source (ie eNd = +5 to +7)

The Sr-Nd isotopic composition for OIB-typesources in the Proterozoic is not as wellconstrained as DMM Today OIB-type mantlesources have a wide range of isotopic composi-tions that partly overlap with DMM but eNd

values are generally lower (Hart et al 1992Hofmann 1997) The EM-1 mantle component ischaracterized by extremely low 143Nd144Ndratios and eNd values of between 0 and ndash5(Hofmann 1997) A possible 12 Ga position forEM-1 is plotted in Fig 7 calculated according toSchleicher et al (1998) with 147Sm144Nd = 018and 87Rb86Sr = 010 The present-day values usedin the calculation are 143Nd144Nd = 051234 and87Sr86Sr = 07057 taken from ocean islandbasalts closest to an EM-1 end-member(Hofmann 1997) Other Sr-Nd isotope datafrom the Gardar Province are plotted to putfurther constraints on the mantle sources of the EFbasalts (Fig 7) We also added comparative datafrom African and Canadian carbonatitesof similarage because there is a well known similaritybetween Sr-Nd isotopic data of OIBs andcarbonatites (Bell and Blenkinsop 1989) Thelinear array of the carbonatite data is consideredto be the result of mixing between the two mantlecomponents HIMU and EM-1 (Bell andBlenkinsop 1989 Bell and Tilton 2001) Thecarbonatites from the Gardar Province havesimilar Nd but slightly more radiogenic Srisotopic compositions than the reference carbo-natite data (Fig 7) The least contaminatedsamples overlap with the linear array and themore radiogenic Sr isotopic composition can beexplained by crustal contamination (Andersen1997) We therefore consider it possible that theGardar carbonatites also represent mixing ofhypothetical HIMU and EM-1 components of anOIB-like source Since the isotopically mostprimitive EF basalt samples overlap with theQassiarsuk carbonatites it could be possible thatthe basalts also contain such mantle componentsalthough Pb isotope data for the EF basalts thatcould clarify the importance of a HIMU sourceare lacking Based on the Sr-Nd isotope data theEF basalts appear to be derived from aheterogeneous OIB-like mantle source thatcomprises isotopically depleted and enrichedcomponents OIB-like and carbonatitic magma-tism are often attributed to mantle plume activity(Bell and Simonetti 1996 Hofmann 1997

2

4

6

8

10

12

-10 -8 -6 -4 -2 0 2 4 6 8

DMM

d18 O

[permil]

e Nd (i)

Typical2s error

EM-1

EF basalts - cpx

EF basalts - model melt

Ketilidian basement

Igaliko dykes

Depleted MORB mantle

Enriched mantle 1

EM-1 lavas - cpx

FIG 8 Oxygen isotopi c compositio ns (d18OV-SMOW) of clinopyroxene from EF basalts calculatedmelt (see text for details) and whole-rock Ketilidianbasement vs eNd(i) Whole-rock data for Igaliko dykerocks (n = 7) are from Pearce and Leng (1996) DMMdata derived from fresh MORB glasses are from Eiler etal (2000b) The eld of EM-1 lavas is taken fromclinopyroxene data of Gough and Tristan da CunhaIsland lavas (Harris et al 2000) and the EM-1 end-member composition is based on the calculations afterSchleicher et al (1998) for eNd(i) and an average value

of 578 (n = 18) from the EM-1 lavas for d18O

PETROGENESIS OF THE ERIKSFJORD BASALTS

845

Simonetti et al 1998) A plume input to theformation of the EF basalts is supported by therelatively large volume of magmas and the shortduration (lt5 Ma) of magmatism (Piper et al1999) It is also compatible with the geologicalsetting of a continental breakup Additionally aplume component appears to be necessary toexplain the thermal input as it is dif cult to melt acold lithosphere section (Arndt and Christensen1992)

Oxygen isotope data of the EF basalts aregenerally consistent with a mantle derivationincluding an OIB-like mantle source On the d18Ovs eNd(i) diagram (Fig 8) the overlap betweenthe EF data clinopyroxene data from EM-1-likeoceanic basalts and the hypothetical EM-1 end-member suggests an involvement of an EM-1-type mantle source Olivines from HIMU-likesources are known to be slightly lower in d18Othan typical upper mantle values from MORBs(Eiler et al 1997) but the EF data are notsuf cient to assess if the slightly lower d18Ovalues in the Il otildeacutemaussaq Group basalts are causedby a relatively higher input from a HIMU source

The contribution of the various mantle sourcesto the EF basalts can also be evaluated on theprimitive mantle-normalized multi-elementdiagram (Fig 4) and on the REE plots (Figs 5and 6) Both the whole-rock REE and theclinopyroxene REE patterns of the EF basaltsare unlike those representative of N-MORBsources We conclude again that DMM did notplay a major role in the petrogenesis of the EFbasalts On the other hand similarities with OIB-like mantle sources are more prominent Theshape of the clinopyroxene REE pattern closelyresembles those from clinopyroxene of OIB-likealkaline basalts despite differences in absoluteconcentrations The whole-rock REE patterns ofthe EF basalts lie between E-MORB and averageOIB patterns as do the multi-element patterns Onthe latter the EF basalts have marked positive Baand P peaks that are not present in typical OIBsHowever Ba contents should be interpreted withcaution since highly variable ZrBa ratios from014 to 070 are an indication that Ba was re-mobilized after emplacement of the basalts (egPrice et al 1991 Oliveira and Tarney 1995)Likewise P mobility was demonstrated in alteredbasaltic rocks (Winchester and Floyd 1976) Insummary the REE and trace element data areevidence that the mantle source(s) of the EFbasalts were enriched in incompatible traceelements relative to primitive mantle

For relatively primitive hy-normative basalticrocks of the Ivittuut area further north in theGardar Province it has been argued that some ofthem were derived from a lithospheric mantlesource previously enriched by subduction-related processes This interpretation is basedon certain geochemical characteristics such ashigh LaNb ratios and negative Nb anomalies(Goodenough et al 2002) For the EF basaltsnegative Nb anomalies are not clearly discern-ible but LaNb ratios are gt1 which seems to be acharacteristic feature of a lithospheric mantlesource (Fitton 1995) LaBa LaRb and LaKratios were also used to argue for a lithosphericorigin of the Gardar basic magmas although amantle plume involvement was assumed toprovide the required energy (Upton andEmeleus 1987 Upton et al 2003) Sincefeldspar is the only primary magmatic mineralin the EF basalts in which Ba Rb and K areincorporated in signi cant amounts and thefeldspars are generally altered these elementsmight have been easily mobilized and wetherefore do not dare to base the interpretationof the EF basalts sources on these data Inaddition the lack of precise geochemical data onthe composition of the lithosphere below theGardar Province complicates an evaluation ofthe lithospheric contribution to the EF basalticmagmatism

Certain ratios of incompatible trace elements ofthe EF basalts are compared to several occur-rences of ocean island basalts representative ofEM-1 and HIMU sources in Table 5 Ratiosbetween highly incompatible trace elements donot change signi cantly during partial melting orfractional crystallization in basaltic systems andmay therefore constrain the ratios in their sources(Weaver et al 1987) The EF basalts aresubdivided into the more alkaline isotopicallyprimitive Il otildeacutemaussaq Group samples and thosesamples with eNd(i) lt ndash1 The Il otilde maussaq Groupsamples have higher SrNd LaNb and ZrNbratios than HIMU and overlap with the upperrange of EM-1 ratios The EF basalts with eNd(i)lt ndash1 have generally higher ratios than the oceanicisland basalts and partly overlap with typicalvalues of the lower crust It therefore seems likelythat assimilation of lower crustal material mayhave changed the primary trace element ratios inthe basalts Therefore we will evaluate thepossible in uence of assimilation of crustalmaterial on the geochemistry of the EF basaltsin the following section

846

R HALAMA ETAL

The role of crustal contamination

Large amounts of bulk assimilation of crustalmaterial which is relatively SiO2-rich comparedto a basaltic melt would drive the compositions tohigher SiO2 contents and quartz-normativecompositions Partial melts derived from crustalrocks would also be relatively SiO2-rich (Beard etal 1993 Carroll and Wyllie 1990) and wouldhave the same effect The ne-normative characterof the EF basalts (Table 1) suggests thatassimilation of crustal material especially ofsilica-rich upper crustal rocks was not signi cantwith the possible exception of sample EF 059

On the primitive mantle-normalized multi-element diagrams (Fig 4) Ti troughs are typicalfor crustal rocks especially for the Upper Crust(Rudnick and Fountain 1995) and they maytherefore be present in rocks extensively contami-nated by continental crust Ti troughs aregenerally lacking in the patterns of the EFbasalts and a major upper crustal contribution tothe magmas can therefore be excluded However

small amounts of assimilation of lower crustalmaterial cannot be ruled out because the averagelower crust has higher Ti contents

Steep trends on Sr-Nd isotope plots have beenattributed to assimilation of granulite-faciesgneisses (eg Bernstein et al 1998) AFCmodelling based on an equation developed byDePaolo (1981) was performed to test thehypothesis of small amounts of contaminationwith lower crustal material (Fig 9) The basalticmagma which represents one end-member for thecalculations has the isotopic signature of theisotopically most primitive samples from thepresent data set and from the literature(Table 6) Two possible contaminants repre-senting upper and lower crustal compositionswere selected for modelling (Table 6) Therepresentative samples are a granite from theJulianehaEcirc b batholith for the upper crust (sampleJG 02) and an average of ve granulite-faciesgneisses with low RbSr ratios from the Archaeancraton of West Greenland (Taylor et al 1984) for

TABLE 5 Trace element ratios of EF basalts compared to ocean island basalts(OIBs) representing EM-1 and HIMU mantle components and the lowercrust

SrNd LaNb ZrNb

EF basalts with eNd lt ndash 1 249 ndash396 120 ndash150 121 ndash157Il otildeacutemaussaq Group EF basalts 198 ndash292 104 ndash119 51 ndash53EM-1 90 ndash207 067 ndash111 39 ndash109HIMU 45 -163 063 ndash080 27 ndash 36Lower continental crust 30 16 136

Data sources EM-1 Humphris and Thompson (1983 Walvis Ridge) and Cliff et al(1991 Inaccessible Islands) HIMU Chauvel et al (1992 Tubuai Island) Lowercontinental crust Rudnick and Fountain (1995)

TABLE 6 End-member compositions used for AFC modelling

End-member Sr (87Sr86Sr)i Nd eNd(i) Data source(ppm) (T = 12 Ga) (ppm) (T = 12 Ga)

Primitive EF basaltic 507 070278 18 + 22 Sample EF 072magma Paslick et al (1993)Upper crust 300 071665 62 ndash83 Sample JG 02(JulianehaEcirc b Granite)Lower crust 747 070668 63 ndash 234 Taylor et al (1984)(Granulite facies gneiss)

The composition of the granulite-facies gneiss is an average of 5 samples with low RbSr ratios lt02

PETROGENESIS OF THE ERIKSFJORD BASALTS

847

the lower crust An in uence of Archaean rockson Gardar magmas has been demonstrated by Pbisotope studies (Taylor and Upton 1993) andseismic data were also interpreted to re ect theexistence of a wedge of Archaean crust thatextends southwards to Lindenow Fjord (Dahl-Jensen et al 1998) We have shown in theprevious sections on the basis of Ni trends thatfractionation of olivine was most relevant to themajor element evolution of the lavas whereasfractionation of other phases was apparently ofminor importance We therefore modelled theAFC processes with olivine fractionation usingmineral-melt partition coef cients of 0014 for Srand 00066 for Nd (Rollinson 1993)

The modelling of AFC processes is stronglydependent on the parameter r de ned as the ratioof mass assimilation rate and fractional crystal-lization rate (DePaolo 1981) The change inisotopic signature becomes enhanced for highervalues of r Reiners et al (1995) have shown thatr values might be very high and even gt1 in theearly stages of magmatic evolution at hightemperatures since crystallization can besuppressed by assimilation Therefore we calcu-

lated AFC processes with two r values of 05 and08 that represent a relatively high and a middle tolow assimilation rate to demonstrate the effects ofvariable r (Fig 9) The results indicate that AFCprocesses involving upper crust similar to theJulianehaEcirc b Granite cannot explain the trend of theEF basalts because Sr isotopic compositionswould shift to considerably more radiogenicvalues This result is consistent with theconclusions derived from trace element dataThe AFC curves for assimilation of lower crusthowever agree well with the bulk of the data Themost negative eNd values can be explained byAFC processes with intermediate assimilationrates (r = 05) involving assimilation of lt5lower crustal material With a greater degree ofassimilation and an r value of 08 lt4assimilation is indicated

Lower crustal rocks of igneous origin have anaverage d18O value of +75plusmn14 with an overallrange from +54 to +125 (Fowler and Harmon1990) 5 to 10 bulk assimilation of lower crustalmaterial with d18O = +75 into a typical mantle-derived magma with d18O = +57 results ind18O values of the contaminated magma from

-24

-20

-16

-12

-8

-4

0

4

0700 0703 0706 0709 0712 0715 0718

r = 08

r = 05

r = 05

r = 08

EF Basalts

Endmembers of AFC calculations

Upper Crust

Lower Crust

Primitive EF basaltic magma

(87Sr86Sr)i

e Nd

(i)

09

08

06

09

08

06

09

0806

09

08

07

07

04

Typical errorCHUR

BS

E

FIG 9 eNd(i) vs (87Sr86Sr)i diagram showing the results of AFC modelling End-member compositions are listed inTable 6 Olivine is the only fractionating mineral Partition coef cients for olivinemelt are from the compilation ofRollinson (1993) Curves were calculated for two values of r (r = rate of assimilation rate of fractionalcrystallization) Dashed lines and numbers in italics are for r = 05 solid lines and regular numbers are for r = 08

Values at the curves are for F values (fraction of melt remaining) with 10 tick marks

848

R HALAMA ETAL

+579 to +588 Therefore the d18O data cannotbe used to rule out small amounts of assimilationof lower crust To evaluate how much contamina-tion would be permissible from the presentoxygen isotope data we calculated how muchcontamination of a mantle-derived magma with18O = +57 with a contaminant within the lowerrange of lower crustal xenoliths (d18O = +67)would drive the oxygen isotopic composition tothe highest value of the EF basalts (d18O~+63) Although simple mixing shows that60 assimilation is permissible this amountwould drive the whole-rock geochemistry tomore silicic compositions than is observed in theEF basalts In Fig 8 there is no obvious trendfrom the basalts towards the Ketilidian basementsuggesting again that upper crust was notassimilated in signi cant amounts

Ratios of trace elements incompatible inbasaltic systems can be used to distinguishbetween crust and mantle reservoirs (egWeaver 1991) Based on the AFC modellingEF basalts with eNd(i) lt ndash1 are those which aremost likely in uenced by crustal contaminationThe relatively high LaNb and ZrNb ratios ofthese samples are consistent with assimilation oflower crustal material SrNd ratios are evenhigher than the average lower crustal composi-tion but SrNd ratios in those samples which weretaken as representative for the lower crust in theAFC modelling can reach values of 80 ie wellabove the highest EF basalt ratios Again theSrNd ratios support the hypothesis of some lowercrustal contamination

The absence of more primitive basaltic rockswith MgO gt8 wt might also indicate lowercrustal assimilation It suggests that magmas mayhave been retained at depth where they acquiredmore evolved compositions by fractional crystal-lization and had the opportunity to assimilatelower crustal material during crustal underplating(Upton et al 2003)

Summary and conclusions

The new geochemical data of the EriksfjordFormation basalts in the Gardar Provinceprovided in this study are used to placepetrogenetic constraints on the sources of theirparental magmas Based on trace-element patternsand Sr-Nd isotopic constraints there seems to beno indication that a DMM-like source played animportant role in magma genesis of the EFbasalts Clinopyroxene and whole-rock REE

patterns are more compatible with an OIB-likesource but we are not able to conclude whetherthe observed enrichment in REE re ects thesourcersquos signature or magmatic processes Theisotopically most primitive EF basalts have Sr-Ndisotopic compositions that are similar to Gardarcarbonatites Due to the isotopic similaritybetween carbonatites and OIBs many arethought to be related to mantle plumes (Bell2001) We therefore speculate that OIB-likemantle components from a plume source werealso involved in the parental EF magmas but wecannot exclude interaction with the SCLM Twodistinct mantle components one isotopicallydepleted and another one enriched could havecontributed to the EF magmas AFC modelling ofSr-Nd isotope data indicates that assimilation oflt5 lower crust could explain the steep vector onthe Sr-Nd isotope plot but assimilation of uppercrustal material is unlikely Oxygen isotope dataof clinopyroxene separates are consistent with amantle origin but permit some assimilation oflower crustal rocks In summary the data areconsistent with the hypothesis that the EFmagmas could have originated from a mantleplume source which is supported by geologicalarguments and that they underwent someassimilation of lower crustal material

Acknowledgements

We would like to thank Bruce Paterson whoprovided invaluable help during Laser ICP-MSmeasurements at the Large-Scale GeochemicalFacility supported by the European Community-Access to Research Infrastructure action of theImproving Human Potential Programme contractnumber HPRI-CT-1999-00008 awarded to ProfBJ Wood (University of Bristol) Gabi Stoschekand Torsten Vennemann are thanked for their helpwith oxygen isotope measurements Elmar Reitterexpertly assisted with preparation and measure-ments of radiogenic isotopes and MathiasWestphal helped with microprobe measurementsREE whole-rock analyses were kindly providedby the NERC facility at Royal HollowayUniversity of London Michael Marks is thankedfor his pleasant company during eldwork Veryconstructive reviews by Tom Andersen andAlexei S Rukhlov greatly helped to improve themanuscript We also thank Ian Coulson for nalediting Funding of this work by the DeutscheForschungsgemeinschaft (grant Ma-21351-2) isgratefully acknowledged

PETROGENESIS OF THE ERIKSFJORD BASALTS

849

References

Andersen T (1997) Age and petrogenesis of theQassiarsuk carbonatite-alkaline silicate volcaniccomplex in the Gardar rift South GreenlandMineralogical Magazine 61 499 ndash513

Armstrong JT (1991) Quantitative elemental analysisof individual microparticles with electron beaminstruments Pp 261 ndash315 in Electron ProbeQuantitation (KFJ Heinrich and DE Newburyeditors) Plenum Press New York and London

Arndt NT and Christensen U (1992) The role oflithospheric mantle in continental ood volcanismThermal and geochemical constraints Journal ofGeophysical Research 97 10967 ndash 10981

Baker JA Menzies MA Thirlwall MF andMacpherson CG (199 7) Pet rogenes is ofQuaternary Intraplate Vocanism Sanarsquoa YemenImplications for Plume-Lithosphere Interaction andPolybaric Melt Hybridization Journal of Petrology 38 1359 ndash 1390

Baker JA MacPherson CG Menzies MAThirlwall MF Al-Kadasi M and Mattey DP(2000) Resolving crustal and mantle contributions tocontinental ood volcanism Yemen constraintsfrom mineral oxygen isotope data Journal ofPetrology 41 1805 ndash1820

Beard JS Abitz RJ and Lofgren GE (1993)Experimental melting of crustal xenoliths fromKilbourne Hole New Mexico and implications forthe contamination and genesi s of magmasContributions to Mineralogy and Petrology 11588 ndash102

Bell K (2001) Carbonatites relationships to mantle-plume activity Geological Society of AmericaSpecial Paper 352 267 ndash290

Bell K and Blenkinsop J (1989) Neodymium andStrontium Isotope Geochemistry of Carbonatites Pp278 ndash300 in Carbonatites ndash Genesis and Evolution(K Bell editor) Unwin Hyman London

Bell K and Simonetti A (1996) Carbonat iteMagmatism and Plume Activity Implications fromthe Nd Pb and Sr Isotope Systematics of OldoinyoLengai Journal of Petrology 37 1321 ndash1339

Bell K and Tilton GR (2001) Nd Pb and Sr IsotopicCompositions of East African Carbonat itesEvide nce for Mant le Mixing and PlumeInhomogene ity Journa l of Petrology 42 1927 ndash1945

Benoit M Polve M and Ceuleneer G (1996) Traceelement and isotopic characterization of ma ccumulates in a fossil mantle diapir (Oman ophiolite)Chemical Geology 134 199 ndash 214

Bernstein S Kelemen PB Tegner C Kurz MDBlusztajn J and Kent Brooks C (1998) Post-breakup basaltic magmatism along the EastGreenland Tertiary rifted margin Earth and

Planetary Science Letters 160 845 ndash862Boynton WV (1984) Geochemistry of the rare earth

elements meteorite studies Pp 63 ndash114 in RareEarth Element Geochemistry (P Henderson editor)Elsevier Amsterdam

Carroll MR and Wyllie PJ (1990) The systemtonalite-H2O at 15 kbar and the genesis of calc-alkaline magmas American Mineralogist 75345 ndash 357

Chauvel C Hofmann AW and Vidal P (1992)HIMU-EM The French Polynesian connectionEarth and Planetary Science Letters 110 99 ndash119

Clayton RN and Mayeda TK (1963) The use ofbromine penta uoride in the extraction of oxygenfrom oxides and silicates for isotope analysisGeochimica et Cosmochimica Acta 27 43 ndash52

Coulson IM and Chambers AD (1996) Patterns ofzonation in rare-earth-bearing minerals in nephelinesyenites of the North Qoroq center South GreenlandThe Canadian Mineralogist 34 1163 ndash1178

Cox KG Bell JD and Pankhurst RJ (1979) TheInterpretation of Igneous Rocks George Allen andUnwin London

Cross W Iddings JP Pirsson LV and WashingtonHS (1903) Quantitative Classi cation of IgneousRocks University of Chicago Press

Dahl-Jensen T Thybo T Hopper H and Rosing M(1998) Crustal structure at the SE Greenland marginfrom wide-angle and normal incidence seismic dataTectonophysics 288 191 ndash 198

DePaolo DJ (1981) Trace element and isotopic effectsof combined wallrock assimilation and fractionalcrystallisation Earth and Planetary Science Letters53 189 ndash 202

De P a o lo D J (1 9 8 8 ) Neodymium Iso topeGeochemistry An Introduction Springer-VerlagNew York

Dobosi G Downes H Mattey D and Embey-IsztinA (1998) Oxygen isotope ratios of phenocrysts fromalkali basalts of the Pannonian basin Evidence foran O-isotopically homogeneous upper mantle be-neath a subduction-in uenced area Lithos 42213 ndash 223

Eiler JM (2001) Oxygen isotope variations of basalticlavas and upper mantle rocks Pp 319 ndash364 inStable Isotope Geochemistry (JW Valley and DRCole editor s) Reviews in Minera logy andGeochemistry 43 The Mineralogical Society ofAmerica Washington DC

Eiler JM Farley KA Valley JW Hauri E CraigH Hart SR and Stolper EM (1997) Oxygenisotope variations in ocean island basalt phenocrystsGeochimica et Cosmochimica Acta 61 2281 ndash2293

Eiler JM Crawford A Elliot T Farley KAValley JW and Stolper EM (2000a) Oxygenisotope geochemistry of oceanic-arc lavas Journal

850

R HALAMA ETAL

tional changes among the various samplesIlmenite from two samples (EF 024 and EF 039)which contain coexisting ilmenite and titanomag-netite includes a signi cant component ofpyrophanite (MnTiO3) with up to 87 wtMnO It s com pos i t i ons va r y be t weenI l m 8 6 H e m 5 P y r 8 a nd I l m 7 6 H e m 5 P yr 1 9 Titanomagnetite analyses were unsatisfactory asall grains show strong alteration They areconsidered unreliable and are not presented here

Whole-rock geochemistry and classif|cation

Concentrations of major and trace elements aresummarized in Table 2 The basalts can beclassi ed according to the TAS diagram (LeMaitre et al 1989) as basalts and trachybasaltswith both alkaline and subalkaline compositions(not shown) Due to possible mobility of the Naand K the samples were also plotted into NbYvs ZrP2O5 and P2O5 vs Zr diagrams (Floyd andWinchester 1975) which can discriminate

between tholeiitic and alkaline series Thesediagrams do not yield consistent discriminationsfor the EF basalts but indicate that the characterof the sample suite is transitional betweentholeiitic and alkaline (not shown)

Mg [100Mg(Mg+Fe2+)] calculated assumingFe2O3FeO = 02 as recommended for basalts byMiddlemost (1989) varies between 54 and 40(Table 2) Ni concentrations range from 46 to115 ppm and Cr contents from 29 to 126 ppmTaking Mg as fractionation index a positivecorrelation (r2 = 061) with Ni contents indicatesfractionation of olivine (Fig 3) The relativelypoor positive correlations of Mg with Cr (r2 =044) and Sc (r2 = 023) argue against a signi cantfractionation of Cr-spinel or clinopyroxene Thisis consistent with the lack of clinopyroxenephenocrysts and the absence of any Cr-spinelThe negative correlations of Mg with Sr TiO2

and P2O5 suggest that fractionation of plagioclaseFe-Ti oxides and apatite was not important in thesample set investigated which is again consistent

TABLE 1 Typical clinopyroxene analyses of Eriksfjord Formation basalts determined by EMP

Sample EF 024 EF 024 EF 039 EF 059 EF 072 EF 108 EF 168 EF 174Cpx no cpx 2 cpx 3 cpx 5 cpx 3 cpx 5 cpx 7 cpx 4 cpx 1

SiO2 5060 5041 5128 4988 5145 4950 5175 5069TiO2 173 176 121 169 140 162 096 121Al2O3 297 300 179 223 171 264 155 213FeO 1160 1068 1218 1405 993 1234 979 1129MnO 028 021 030 040 019 028 025 031MgO 1254 1324 1263 1182 1325 1163 1307 1214CaO 1959 2029 2002 1907 2106 2099 2174 2143Na2O 034 031 029 049 033 055 052 054

Total 9965 9989 9971 9963 9932 9955 9962 9974

Formulae based on 4 cations and 6 oxygensSi 191 189 194 190 194 188 194 191Al 013 013 008 010 008 012 007 009Ti 005 005 003 005 004 005 003 003Fe3+ - - - 003 - 007 003 005Mg 071 074 071 067 075 066 073 068Fe2+ 037 033 039 041 031 032 028 031Mn 001 001 001 001 001 001 001 001Ca 079 082 081 078 085 085 087 087Na 002 002 002 004 002 004 004 004

Total 400 400 400 400 400 400 400 400

Pyroxene projection after Lindsley (1983)Wo 039 041 041 041 043 043 045 044En 040 041 039 037 040 038 040 038Fs 021 018 021 023 017 019 015 017

836

R HALAMA ETAL

TABLE 2 XRF whole-rock analyses of Eriksfjord Formation basalts

Sample EF 024 EF 063 EF 061 EF 039 EF 059 EF 108 EF 072 EF 174 EF 168Group Mussartut Mussartut Mussartut Mussartut Mussartut Ulukasik Ulukasik Ilotildeacutemaussaq IlotildeacutemaussaqSubunit 4 4 5 6 7 Lower Unit Upper Unit Sermilik Narssaq

Fjeld

Major elements (wt)SiO2 4573 4585 4688 4660 4875 4698 4656 4601 4545TiO2 230 200 180 175 191 265 203 299 277Al2O3 1578 1630 1600 1612 1645 1682 1668 1645 1719Fe2O3 1510 1356 1331 1383 1225 1394 1321 1438 1392MnO 028 018 018 018 017 014 017 018 017MgO 525 683 653 673 568 396 647 440 480CaO 842 849 872 857 920 641 808 746 737Na2O 349 274 289 276 236 448 303 323 376K2O 087 083 068 032 027 159 070 205 132P2O5 041 035 023 023 023 048 035 111 083LOI 199 210 240 238 242 204 210 052 156

Total 9962 9923 9962 9947 9969 9949 9938 9878 9914

Mg 453 545 538 536 524 403 538 421 450

Trace elements (ppm)Sc 24 22 25 22 26 19 20 21 16V 209 175 215 195 211 142 170 164 203Cr 81 108 66 65 126 48 82 33 29Co 63 78 70 62 71 44 68 35 46Ni 99 115 85 85 101 58 106 46 51Cu 65 49 46 60 97 41 59 38 41Zn 103 102 110 97 110 93 82 122 105Ga 20 18 21 19 20 19 16 21 19Rb 24 26 16 lt 10 lt 10 38 12 39 45Sr 504 503 512 392 373 712 507 850 992Y 31 26 28 23 28 31 17 31 25Zr 188 165 121 116 125 125 88 164 139Nb 12 11 10 lt 10 10 9 lt 10 31 27Ba 445 375 415 302 178 582 405 1293 900Pb 6 7 3 4 3 4 4 3 5La ndash ndash 15 ndash 12 11 ndash 37 28Ce ndash ndash 40 ndash 38 39 ndash 92 71Pr ndash ndash 1 ndash 1 2 ndash 9 4Nd ndash ndash 16 ndash 15 18 ndash 43 34Sm ndash ndash 6 ndash 7 6 ndash 5 5

CIPW normq ndash ndash ndash ndash 313 ndash ndash ndash ndashor 514 491 402 189 160 940 414 1212 780ab 2807 2319 2445 2335 1997 3207 2564 2733 2981an 2482 2973 2868 3065 3350 2109 2985 2433 2613ne 079 ndash ndash ndash ndash 316 ndash ndash 109di 1187 830 1076 853 877 633 654 452 416hy ndash 890 1059 1650 2217 ndash 1097 315 ndashol 1670 1309 1044 780 ndash 1472 1120 1388 1689mt 365 328 322 335 296 336 319 348 336il 437 380 342 332 363 503 386 568 526ap 095 081 053 053 053 111 081 257 192

Mg was calculated as Mg = 100Mg2+(Mg2++Fe2+) using Fe2O3FeO = 02 Normative mineral compositionswere determined according to the CIPW calculation scheme

PETROGENESIS OF THE ERIKSFJORD BASALTS

837

1 5

2 0

2 5

3 0

3 5

40

60

80

100

120

0

0 5

1

1 5

200

400

600

800

1000

0

50

100

150

15

20

25

30

4 0 45 50 5 5

r2 = 023

r2 = 044r2 = 061

r2 = 053

r2 = 082 r2 = 052

Ni (ppm)

Sc (ppm)

P2O5 (wt)

Sr (ppm)

TiO2 (wt)

Cr (ppm)

Mg Mg

Mg Mg

Mg

40 4 5 50 55

Mg

40 45 50 55

40 45 50 55

40 45 50 5540 4 5 50 55

FIG 3 Whole-rock variation diagrams for EF basalts with Mg as fractionation index Mg is calculated asMg(Mg+Fe2+) with Fe2O3FeO = 02 (Middlemost 1989)

838

R HALAMA ETAL

with the mostly subhedral Fe-Ti oxides and theophitic texture Ratios of incompatible traceelements such as ZrY (range 40 ndash63) andTiY (range 385 ndash716) are relatively uniformthroughout all samples but some scatter isevident Notable geochemical features are highSr contents up to 1000 ppm and very high Bacontents up to 1300 ppm

Normative mineral compositions were deter-mined according to the CIPW calculation scheme(Cross et al 1903 Cox et al 1979) Mostsamples are hypersthene-normative some arenepheline-normative and one (EF 059) is quartz-normative (Table 2) Despite the uncertaintiesinvolved in the CIPW norm calculation the dataindicate that some of the basalts crystallized fromsilica-undersaturated magmas and others fromdistinct silica-saturatedor -oversaturatedmagmas

Multi-element diagrams normalized to primi-tive mantle values show that the EF basalts areenriched in all incompatible elements relative toprimitive mantle values (Fig 4) The patterns ofthe EF basalts are characterized by a prominentBa peak a subordinate P peak and a decreasingelement enrichment with increasing compatibilityIn comparison with normal MORB (N-MORB)enriched MORB (E-MORB) and average OIB(Sun and McDonough 1989) the patterns of theEF basalts are closest to those of OIB both inshape and normalized concentrations

Whole-rock REE data

Whole-rock REE data are presented in Table 3and REE concentrations normalized to chondriticvalues of Boynton (1984) are shown in Fig 5aAll samples are characterized by LREE-enrichedpatterns with no or slightly positive Euanomalies (EuEu = 097 to 113) La concentra-tions are roughly 20 ndash60 times and Lu concentra-tions are 10 times chondritic values NormalizedLREE abundances from La to Nd are fairlyconstant but the patterns show a steady andsmooth decline from Nd towards heavier REEThe REE patterns of the EF basalts fall in betweenthose of typical OIB and E-MORB (Fig 5b)They have similar patterns and absolute abun-dances compared to basaltic dykes from theIvittuut area South Greenland (Goodenough etal 2002) LaNSmN as a measure of LREEfractionation has a restricted range from 130 to165 The degree of HREE fractionationexpressed as GdNYbN is slightly more variableand ranges from 157 to 257 Both these ratios of

the EF basalts fall generally between typical OIBand E-MORB compositions

Clinopyroxene REE patterns

Typical analyses of REE concentrations inclinopyroxenes are given in Table 3 They showthat REE concentrations in individual minerals ofone particular sample can be quite variableChondrite-normalized REE patterns of clinopyr-oxenes are very similar in the samples from theMussartut and Ulukasik Groups (Fig 6ab) Theyare enriched in all REE relative to chondriticvalues and the patterns show a fairly steepincrease from La to Sm followed by a gradualsmooth decrease from Sm to Lu Clinopyroxenesfrom sample EF 168 (Il otildeacutemaussaq Group) aredistinct as they have generally higher REEcontents especially in the LREE and a weak Euanomaly (Fig 6c) The 3 ndash106 enrichment in Nd

c (sa

mpl

e)c

(prim

itive

man

tle)

c (sa

mpl

e)c

(prim

itive

man

tle)

1

10

100

1000

Mussartucirct GroupUlukasik GroupIliacutemaussaq Group

Rb Ba K Nb La Ce Sr Nd P Zr Ti Y

Rb Ba K Nb La Ce Sr Nd P Zr Ti Y01

1

10

100

1000N-MORBE-MORBOIB

FIG 4 Multi-element plots normalized to primitivemantle values (McDonough and Sun 1995) Data forN-MORB E-MORB and average OIB (Sun and

McDonough 1989) are shown for comparison

PETROGENESIS OF THE ERIKSFJORD BASALTS

839

TA

BL

E3

RE

Eda

taof

who

lero

cks

and

clin

opyr

oxen

es

the

latt

erde

term

ined

byL

aser

ICP

-MS

Who

le-r

ock

anal

yses

R

epre

sent

ativ

ecl

inop

yrox

ene

anal

yses

S

ampl

eno

10

1231

1014

4610

1457

1014

7218

6389

1863

95E

F02

4-2

EF

039-

1E

F03

9-2

EF

039-

4E

F07

2-2

EF

072-

3E

F16

8-3

EF

168-

5

RE

Eco

ncen

trat

ions

(ppm

)L

a11

65

180

610

29

661

155

412

28

440

266

199

150

273

304

165

527

59

Ce

290

346

05

249

617

13

378

130

08

180

411

46

868

717

105

012

60

572

399

67

Pr

434

67

388

268

527

433

331

216

181

128

249

262

102

917

43

Nd

200

631

71

176

412

49

250

920

16

190

213

82

122

28

9815

64

157

258

51

960

6S

m4

756

974

043

215

914

826

664

194

283

276

077

0615

82

265

9E

u1

742

51

571

071

971

721

831

491

111

212

001

943

877

07G

d4

826

584

043

555

84

996

965

615

244

326

156

0514

75

252

7T

b ndash

ndash ndash

ndash ndash

ndash1

351

010

770

711

111

212

313

61D

y4

475

223

613

565

354

816

865

934

784

255

996

8613

10

205

2H

o0

921

040

730

731

11

011

481

401

001

001

211

292

394

24E

r2

432

531

951

972

872

663

893

342

642

112

993

646

539

60T

m ndash

ndash ndash

ndash ndash

ndash0

450

390

360

290

350

440

861

27Y

b2

182

071

741

832

592

363

782

812

492

452

703

016

788

89L

u0

340

310

260

270

390

350

460

410

300

310

520

411

251

60

(Eu

Eu

) N1

111

131

190

971

031

070

820

940

720

991

000

910

770

83

Eu

anom

aly

isca

lcul

ated

as(E

uE

u) N

=E

u N(

SmN6

Gd N

)05

norm

aliz

edto

chon

drit

eva

lues

from

Boy

nton

(198

4)

840

R HALAMA ETAL

in clinopyroxenes of EF 168 compared to theother samples (EF 024 039 and 072) seems onlypartly due to a higher degree of fractionationsincethe Mg value of EF 168 is only slightly lowerThus the LREE enrichment appears to underlinethe more alkaline character of this sample Acomparison with published REE patterns ofclinopyroxenes (Fig 6d) shows that the EFpatterns are unlike patterns from olivine gabbroswith MORB-like signatures (Benoit et al 1996)but fairly similar to clinopyroxene megacrystpatterns from an OIB-like alkaline basalt (Nimisand Vannucci 1995)

Sr Nd and O isotopic composition

Sr Nd and O isotopic compositions of all samplesare presented in Table 4 Sr and Nd isotopiccompositions of the clinopyroxene separates havebeen recalculated to an approximate eruption age

of 12 Ga (Paslick et al 1993) using decayconstants of 142610ndash11 andash1 for 87Rb (Steigerand Jager 1977) and 654610ndash12 a ndash1 for 147Sm(Lugmair and Marti 1978) Present day CHURvalues of 01967 for 147Sm144Nd (Jacobsen andWasserburg 1980) and 0512638 for 143Nd144Nd(Goldstein et al 1984) were used to calculateinitial eNd values 87Sr86Sr initial ratios are fairlyconstant and range from 070278 to 070383(Table 4) eNd(i) values vary from ndash32 to +21(Table 4) On the Sr-Nd isotope diagram the EFbasalts de ne a relatively steep trend (Fig 7) withinitial 87Sr86Sr ratios similar to those of the BulkSilicate Earth (BSE) Only one sample is slightlydisplaced towards a more radiogenic initial87Sr86Sr ratio None of the samples has an eNd

value close to depleted mantle values estimatedto be between eNd = +53 (calculated afterDePaolo 1981) and eNd = +73 (calculated afterGoldstein et al 1984) at 12 Ga The two alkalinebasalts of the Il otilde maussaq Group are within therange of the other data but they de ne a restricted eld around CHUR and BSE respectively Errorsinduced by the uncertainty in the age of the EFbasalts are largely within the error of the isotopeanalyses and do not change the overall picture Incomparison with other published Sr-Nd isotopedata from the Gardar Province the initial Srisotope ratios of the EF basalts are very similar torelatively primitive basaltic lamprophyric andcarbonatitic dyke rocks which have (87Sr86Sr)i of~0703 (Pearce and Leng 1996 Andersen 1997Goodenough et al 2002) So far no evidence hasbeen presented for an isotopicallyenriched mantlesource in the Gardar Province as the existingeNd(i) data of ndash05 to +52 show an isotopicallydepleted signature (Pearce and Leng 1996Andersen 1997 Goodenough et al 2002Upton et al 2003) Clinopyroxene from the EFsamples of Paslick et al (1993) yield eNd(i) of+22 The Nd isotope data of this study are withinthe lower range of published values from theGardar Province and extend the spread to morenegative eNd(i) values

Several recent oxygen isotope studies haveshown that mineral separates are less sensitive tolow-T alteration and secondary water take-up thanwhole-rock samples (eg Vroon et al 2001)However only few oxygen isotope data ofmineral separates from continental volcanics areavailable (eg Dobosi et al 1998 Baker et al2000) Extensive studies of mineral separates ofocean island basalts (Eiler et al 1997) andoceanic-arc lavas (Eiler et al 2000a) demon-

c (sa

mpl

e)c

(cho

ndri

te)

c (sa

mp

le)c

(cho

ndrit

e)

1

10

100

200

101446

101472

101231

101457

186389

186395

La Ce Pr Nd Sm Eu Gd Dy Ho Er Yb Lu

1

10

100

200

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

N-MORBE-MORBOIBBasaltic dykes from Ivittuut

a

b

FIG 5 (a) Whole-rock REE data of EF basaltsnormalized to chondritic values from Boynton (1984)(b) Typical compositions of N-MORB E-MORB andOIB (Sun and McDonough 1989) are shown for

comparison

PETROGENESIS OF THE ERIKSFJORD BASALTS

841

strated that analyses of mineral separates consis-tently span a narrower range than comparablewhole-rock data In our study we used clino-pyroxene because it is the only fresh mineralphase present in the samples Oxygen isotopemeasurements of clinopyroxene separates fromthe EF basalts show a range in d18Ocpx from +52to +60 with an average value of +566plusmn023(1s) All samples overlap the range of d18Ocpx

values from mantle peridotites (+525 to +590Mattey et al 1994) from continental oodbasalts from Yemen (+55 to +69 Baker etal 2000) and from alkaline basalts of Hungary(+507 to +534 Dobosi et al 1998) There is asmall difference between the averages ofMussartut Group (d18Ocpx-avg = +575plusmn015(1s)) and Ulukasik Group (d18Ocpx-avg =+575plusmn009 (1s)) samples compared toIl otildeacutemaussaq Group (d18Ocpx-avg = +533plusmn021(1s)) samples

d18O values of melts coexisting with clino-pyroxene were calculated assuming that oxygen

in pyroxene and melt were in isotopic equilibriumand T was 1100ordmC Dmelt-cpx can then be calculatedusing the following equation of Kalamarides(1986)

Dmelt-cpx = ndash00008T + 143 (T in Kelvin)

Although fractionation varies with magmatictemperatures we consider the temperature of1100ordmC and the resulting Dmelt-cpx of +03 assuf ciently representative considering the analy-tical errors involved It is similar to Dmelt-cpx

values used in previous studies (+039 Vroonet al 2001 +02 Dobosi et al 1998)Resulting d18Omelt values of EF basalts rangefrom +55 to +63 These values can becompared to whole-rock data from continentalbasalts given by Harmon and Hoefs (1995)Intraplate basalts from continental rift zonesshow a range from +46 to +83 with a meanof +61 and continental ood basalts which arerather poorly represented in the database havevalues between +43 and +65 Calculated EF

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Cpx from MORB-like

olivine gabbro

Cpx megacrysts from OIB-like

alkaline basalts

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 024EF 039

Mussartucirct Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 168

Iliacutemaussaq Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

1

10

100

1000

EF 072

Ulukasik Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb LuLa Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

c

ba

d

1

10

100

01

FIG 6 REE patterns of clinopyroxenes (cpx) from Mussartut (a) Ulukasik (b) and Il otilde maussaq (c) Group samplesnormalized to chondritic values from Boynton (1984) (d) REE patterns of cpx from MORB-type olivine gabbro(Benoit et al 1996) and OIB-type alkaline basalt (Nimis and Vannucci 1995) for comparison (note the change of

scale)

842

R HALAMA ETAL

TA

BL

E4

Sr

Nd

and

Ois

otop

icco

mpo

siti

ons

ofcl

inop

yrox

ene

sepa

rate

sfr

omE

riks

fjor

dF

orm

atio

nba

salt

sW

hole

-roc

kan

alys

esfr

omth

eK

etil

idia

nba

sem

ent

and

from

sedi

men

tary

rock

sof

the

Eri

ksfj

ord

For

mat

ion

are

give

nfo

rco

mpa

riso

n

Sam

ple

Mem

ber

Sr

Rb

87R

b86S

r87S

r86Sr

87Sr

86Sr

init

ial

SmN

d147Sm

144N

d143N

d144N

de N

d(i

)d1

8O

(ppm

)(p

pm)

(ppm

)(p

pm)

Bas

alts

E

F02

4M

ussa

rtut

646

50

486

002

180

7042

01plusmn

100

7038

267

3822

80

019

560

5124

67plusmn

10ndash

32

567

EF

063

Mus

sart

ut5

60plusmn

011

EF

061

Mus

sart

ut89

95

099

90

0321

070

3818

plusmn10

070

3266

574

163

90

2115

051

2629

plusmn10

ndash2

55

79plusmn

028

EF

039

Mus

sart

ut73

74

058

30

0229

070

3492

plusmn10

070

3098

662

191

30

2092

051

2626

plusmn10

ndash2

25

71plusmn

001

EF

059

Mus

sart

ut54

47

013

70

0073

070

3375

plusmn12

070

3250

606

168

90

2171

051

2727

plusmn09

ndash1

45

99plusmn

001

EF

108

Ulu

kasi

k66

92

111

00

0480

070

3741

plusmn09

070

2916

109

936

82

018

050

5126

17plusmn

10+

21

581

EF

072

Ulu

kasi

k69

41

032

20

0134

070

3014

plusmn10

070

2784

818

251

90

1962

051

2574

plusmn10

ndash1

25

68E

F17

4Il

otildemau

ssaq

125

42

324

005

360

7039

35plusmn

100

7030

1418

76

796

10

1424

051

2230

plusmn10

+0

45

47E

F16

8Il

otildemau

ssaq

866

71

422

004

750

7037

99plusmn

070

7029

8316

45

641

60

1550

051

2304

plusmn09

ndash0

15

18

Bas

emen

tan

dse

dim

enta

ryro

cks

JG02

Juli

aneh

aEcirc b30

01

169

71

6375

074

4789

plusmn10

071

6647

105

362

28

010

220

5114

70plusmn

09ndash

83

78

Gra

nite

EF

021

Qua

rtzi

te11

5E

F02

8A

rkos

ic6

8ar

enit

e

Sam

ples

are

list

edin

stra

tigra

phic

orde

rfr

omol

dest

toyo

unge

st87

Sr86

Sr

init

ialr

atio

san

de N

d(i)

valu

esw

ere

calc

ulat

edfo

ran

age

of1

2G

aS

tand

ard

devi

atio

nsfo

rd18

Oar

egi

ven

for

sam

ples

that

wer

ean

alys

edtw

ice

PETROGENESIS OF THE ERIKSFJORD BASALTS

843

model melt oxygen isotope data are within theranges of both groups Compared to Igaliko dykerocks from further SE in the Gardar Province(Pearce and Leng 1996) the EF basalts cover amuch more restricted eld possibly because d18Owhole-rock analyses are more prone to secondaryalteration than analyses of mineral separates(Fig 8) Whole-rock d18O measurements of theunderlying JulianehaEcirc b Granite and of the sedi-mentary sandstone members of the EriksfjordFormation yield values of 78 and 68 ndash115respectively (Table 4) and are shown forcomparison

Discussion

Mantle source regions of the EF basaltsTo reconstruct the mantle sources involved in theformation of the EF basalts we use Sr-Nd-Oisotope and incompatible trace element data On aSr-Nd isotope diagram the isotopic compositionof DMM was modelled for an age of 12 Ga(DePaolo 1981 Goldstein et al 1984 Taylor andMcLennan 1985) The fairly constant 87Sr86Srinitial isotope ratios of the EF basalts are ~0703which falls within the range typical for mantle-derived rocks but slightly above calculated valuesfor DMM (Fig 7) eNd(i) values are more

-4

-2

0

2

4

6

8

0701 0702 0703 0704 0705 0706

DMM Typical 2 s error

CHUR

BS

E

(87Sr86Sr)i

e Nd

(i)

EM-1

Qassiarsuk carbonatites and lamprophyres

EF basalts

Depleted MORB mantle

Ivittuut basaltic and lamprophyric dikes

Igaliko carbonatites and lamprophyres

Enriched mantle 1

Carbonatites from other regions of similar age

FIG 7 eNd(i) vs (87Sr86Sr)i diagram calculated for t = 12 Ga which is considered as the eruption age of the EFbasalts (Paslick et al 1993) The reference line of Bulk Silicate Earth (BSE) for the Sr isotopic composition wascalculated after DePaolo (1988) assuming present-day values of 00827 for 87Rb86Sr and 07045 for 87Sr86SrIsotopic composition of the depleted MORB mantle (DMM) was calculated after DePaolo (1981) Goldstein et al(1984) and Taylor and McLennan (1985) EM-1 was calculated after Schleicher et al (1998) (see text for details)Data sources are Goodenough et al (2002) Ivittuut (n = 6) Pearce and Leng (1996) Igaliko (n = 5) Andersen(1997) Qassiarsuk (n = 25) Bell and Blenkinsop (1989) other carbonatites of similar age (Goudini Schryburt Lake

and Prairie Lake n = 6)

844

R HALAMA ETAL

scattered and indicate that the rocks contain amixture of isotopic components from both mantleand crustal reservoirs or a heterogeneous mantlesource Only one sample (EF 108) has an eNd(i)value signi cantly above 0 This sample exhibitsthe largest spread of clinopyroxene compositionswhich may re ect strong alteration Goodenoughet al (2002) also note that the highest eNd occursin the most altered sample of their sample setHowever Paslick et al (1993) report clino-pyroxene data from two Ulukasik Groupsamples with eNd(i) gt +2 and those sampleswere apparently not disturbed with regard to theirSm-Nd age information On the other hand Piperet al (1999) suggest that the Sm-Nd systematicsof those samples may be erroneous due to laterdisturbance by uids derived from the Il otildeacutemaussaqintrusion Therefore although we are presentlynot able to decide whether positive eNd(i) values

for the EF basalts are real it is clear that none ofthose values is close to values of a 12 Ga oldDMM source (ie eNd = +5 to +7)

The Sr-Nd isotopic composition for OIB-typesources in the Proterozoic is not as wellconstrained as DMM Today OIB-type mantlesources have a wide range of isotopic composi-tions that partly overlap with DMM but eNd

values are generally lower (Hart et al 1992Hofmann 1997) The EM-1 mantle component ischaracterized by extremely low 143Nd144Ndratios and eNd values of between 0 and ndash5(Hofmann 1997) A possible 12 Ga position forEM-1 is plotted in Fig 7 calculated according toSchleicher et al (1998) with 147Sm144Nd = 018and 87Rb86Sr = 010 The present-day values usedin the calculation are 143Nd144Nd = 051234 and87Sr86Sr = 07057 taken from ocean islandbasalts closest to an EM-1 end-member(Hofmann 1997) Other Sr-Nd isotope datafrom the Gardar Province are plotted to putfurther constraints on the mantle sources of the EFbasalts (Fig 7) We also added comparative datafrom African and Canadian carbonatitesof similarage because there is a well known similaritybetween Sr-Nd isotopic data of OIBs andcarbonatites (Bell and Blenkinsop 1989) Thelinear array of the carbonatite data is consideredto be the result of mixing between the two mantlecomponents HIMU and EM-1 (Bell andBlenkinsop 1989 Bell and Tilton 2001) Thecarbonatites from the Gardar Province havesimilar Nd but slightly more radiogenic Srisotopic compositions than the reference carbo-natite data (Fig 7) The least contaminatedsamples overlap with the linear array and themore radiogenic Sr isotopic composition can beexplained by crustal contamination (Andersen1997) We therefore consider it possible that theGardar carbonatites also represent mixing ofhypothetical HIMU and EM-1 components of anOIB-like source Since the isotopically mostprimitive EF basalt samples overlap with theQassiarsuk carbonatites it could be possible thatthe basalts also contain such mantle componentsalthough Pb isotope data for the EF basalts thatcould clarify the importance of a HIMU sourceare lacking Based on the Sr-Nd isotope data theEF basalts appear to be derived from aheterogeneous OIB-like mantle source thatcomprises isotopically depleted and enrichedcomponents OIB-like and carbonatitic magma-tism are often attributed to mantle plume activity(Bell and Simonetti 1996 Hofmann 1997

2

4

6

8

10

12

-10 -8 -6 -4 -2 0 2 4 6 8

DMM

d18 O

[permil]

e Nd (i)

Typical2s error

EM-1

EF basalts - cpx

EF basalts - model melt

Ketilidian basement

Igaliko dykes

Depleted MORB mantle

Enriched mantle 1

EM-1 lavas - cpx

FIG 8 Oxygen isotopi c compositio ns (d18OV-SMOW) of clinopyroxene from EF basalts calculatedmelt (see text for details) and whole-rock Ketilidianbasement vs eNd(i) Whole-rock data for Igaliko dykerocks (n = 7) are from Pearce and Leng (1996) DMMdata derived from fresh MORB glasses are from Eiler etal (2000b) The eld of EM-1 lavas is taken fromclinopyroxene data of Gough and Tristan da CunhaIsland lavas (Harris et al 2000) and the EM-1 end-member composition is based on the calculations afterSchleicher et al (1998) for eNd(i) and an average value

of 578 (n = 18) from the EM-1 lavas for d18O

PETROGENESIS OF THE ERIKSFJORD BASALTS

845

Simonetti et al 1998) A plume input to theformation of the EF basalts is supported by therelatively large volume of magmas and the shortduration (lt5 Ma) of magmatism (Piper et al1999) It is also compatible with the geologicalsetting of a continental breakup Additionally aplume component appears to be necessary toexplain the thermal input as it is dif cult to melt acold lithosphere section (Arndt and Christensen1992)

Oxygen isotope data of the EF basalts aregenerally consistent with a mantle derivationincluding an OIB-like mantle source On the d18Ovs eNd(i) diagram (Fig 8) the overlap betweenthe EF data clinopyroxene data from EM-1-likeoceanic basalts and the hypothetical EM-1 end-member suggests an involvement of an EM-1-type mantle source Olivines from HIMU-likesources are known to be slightly lower in d18Othan typical upper mantle values from MORBs(Eiler et al 1997) but the EF data are notsuf cient to assess if the slightly lower d18Ovalues in the Il otildeacutemaussaq Group basalts are causedby a relatively higher input from a HIMU source

The contribution of the various mantle sourcesto the EF basalts can also be evaluated on theprimitive mantle-normalized multi-elementdiagram (Fig 4) and on the REE plots (Figs 5and 6) Both the whole-rock REE and theclinopyroxene REE patterns of the EF basaltsare unlike those representative of N-MORBsources We conclude again that DMM did notplay a major role in the petrogenesis of the EFbasalts On the other hand similarities with OIB-like mantle sources are more prominent Theshape of the clinopyroxene REE pattern closelyresembles those from clinopyroxene of OIB-likealkaline basalts despite differences in absoluteconcentrations The whole-rock REE patterns ofthe EF basalts lie between E-MORB and averageOIB patterns as do the multi-element patterns Onthe latter the EF basalts have marked positive Baand P peaks that are not present in typical OIBsHowever Ba contents should be interpreted withcaution since highly variable ZrBa ratios from014 to 070 are an indication that Ba was re-mobilized after emplacement of the basalts (egPrice et al 1991 Oliveira and Tarney 1995)Likewise P mobility was demonstrated in alteredbasaltic rocks (Winchester and Floyd 1976) Insummary the REE and trace element data areevidence that the mantle source(s) of the EFbasalts were enriched in incompatible traceelements relative to primitive mantle

For relatively primitive hy-normative basalticrocks of the Ivittuut area further north in theGardar Province it has been argued that some ofthem were derived from a lithospheric mantlesource previously enriched by subduction-related processes This interpretation is basedon certain geochemical characteristics such ashigh LaNb ratios and negative Nb anomalies(Goodenough et al 2002) For the EF basaltsnegative Nb anomalies are not clearly discern-ible but LaNb ratios are gt1 which seems to be acharacteristic feature of a lithospheric mantlesource (Fitton 1995) LaBa LaRb and LaKratios were also used to argue for a lithosphericorigin of the Gardar basic magmas although amantle plume involvement was assumed toprovide the required energy (Upton andEmeleus 1987 Upton et al 2003) Sincefeldspar is the only primary magmatic mineralin the EF basalts in which Ba Rb and K areincorporated in signi cant amounts and thefeldspars are generally altered these elementsmight have been easily mobilized and wetherefore do not dare to base the interpretationof the EF basalts sources on these data Inaddition the lack of precise geochemical data onthe composition of the lithosphere below theGardar Province complicates an evaluation ofthe lithospheric contribution to the EF basalticmagmatism

Certain ratios of incompatible trace elements ofthe EF basalts are compared to several occur-rences of ocean island basalts representative ofEM-1 and HIMU sources in Table 5 Ratiosbetween highly incompatible trace elements donot change signi cantly during partial melting orfractional crystallization in basaltic systems andmay therefore constrain the ratios in their sources(Weaver et al 1987) The EF basalts aresubdivided into the more alkaline isotopicallyprimitive Il otildeacutemaussaq Group samples and thosesamples with eNd(i) lt ndash1 The Il otilde maussaq Groupsamples have higher SrNd LaNb and ZrNbratios than HIMU and overlap with the upperrange of EM-1 ratios The EF basalts with eNd(i)lt ndash1 have generally higher ratios than the oceanicisland basalts and partly overlap with typicalvalues of the lower crust It therefore seems likelythat assimilation of lower crustal material mayhave changed the primary trace element ratios inthe basalts Therefore we will evaluate thepossible in uence of assimilation of crustalmaterial on the geochemistry of the EF basaltsin the following section

846

R HALAMA ETAL

The role of crustal contamination

Large amounts of bulk assimilation of crustalmaterial which is relatively SiO2-rich comparedto a basaltic melt would drive the compositions tohigher SiO2 contents and quartz-normativecompositions Partial melts derived from crustalrocks would also be relatively SiO2-rich (Beard etal 1993 Carroll and Wyllie 1990) and wouldhave the same effect The ne-normative characterof the EF basalts (Table 1) suggests thatassimilation of crustal material especially ofsilica-rich upper crustal rocks was not signi cantwith the possible exception of sample EF 059

On the primitive mantle-normalized multi-element diagrams (Fig 4) Ti troughs are typicalfor crustal rocks especially for the Upper Crust(Rudnick and Fountain 1995) and they maytherefore be present in rocks extensively contami-nated by continental crust Ti troughs aregenerally lacking in the patterns of the EFbasalts and a major upper crustal contribution tothe magmas can therefore be excluded However

small amounts of assimilation of lower crustalmaterial cannot be ruled out because the averagelower crust has higher Ti contents

Steep trends on Sr-Nd isotope plots have beenattributed to assimilation of granulite-faciesgneisses (eg Bernstein et al 1998) AFCmodelling based on an equation developed byDePaolo (1981) was performed to test thehypothesis of small amounts of contaminationwith lower crustal material (Fig 9) The basalticmagma which represents one end-member for thecalculations has the isotopic signature of theisotopically most primitive samples from thepresent data set and from the literature(Table 6) Two possible contaminants repre-senting upper and lower crustal compositionswere selected for modelling (Table 6) Therepresentative samples are a granite from theJulianehaEcirc b batholith for the upper crust (sampleJG 02) and an average of ve granulite-faciesgneisses with low RbSr ratios from the Archaeancraton of West Greenland (Taylor et al 1984) for

TABLE 5 Trace element ratios of EF basalts compared to ocean island basalts(OIBs) representing EM-1 and HIMU mantle components and the lowercrust

SrNd LaNb ZrNb

EF basalts with eNd lt ndash 1 249 ndash396 120 ndash150 121 ndash157Il otildeacutemaussaq Group EF basalts 198 ndash292 104 ndash119 51 ndash53EM-1 90 ndash207 067 ndash111 39 ndash109HIMU 45 -163 063 ndash080 27 ndash 36Lower continental crust 30 16 136

Data sources EM-1 Humphris and Thompson (1983 Walvis Ridge) and Cliff et al(1991 Inaccessible Islands) HIMU Chauvel et al (1992 Tubuai Island) Lowercontinental crust Rudnick and Fountain (1995)

TABLE 6 End-member compositions used for AFC modelling

End-member Sr (87Sr86Sr)i Nd eNd(i) Data source(ppm) (T = 12 Ga) (ppm) (T = 12 Ga)

Primitive EF basaltic 507 070278 18 + 22 Sample EF 072magma Paslick et al (1993)Upper crust 300 071665 62 ndash83 Sample JG 02(JulianehaEcirc b Granite)Lower crust 747 070668 63 ndash 234 Taylor et al (1984)(Granulite facies gneiss)

The composition of the granulite-facies gneiss is an average of 5 samples with low RbSr ratios lt02

PETROGENESIS OF THE ERIKSFJORD BASALTS

847

the lower crust An in uence of Archaean rockson Gardar magmas has been demonstrated by Pbisotope studies (Taylor and Upton 1993) andseismic data were also interpreted to re ect theexistence of a wedge of Archaean crust thatextends southwards to Lindenow Fjord (Dahl-Jensen et al 1998) We have shown in theprevious sections on the basis of Ni trends thatfractionation of olivine was most relevant to themajor element evolution of the lavas whereasfractionation of other phases was apparently ofminor importance We therefore modelled theAFC processes with olivine fractionation usingmineral-melt partition coef cients of 0014 for Srand 00066 for Nd (Rollinson 1993)

The modelling of AFC processes is stronglydependent on the parameter r de ned as the ratioof mass assimilation rate and fractional crystal-lization rate (DePaolo 1981) The change inisotopic signature becomes enhanced for highervalues of r Reiners et al (1995) have shown thatr values might be very high and even gt1 in theearly stages of magmatic evolution at hightemperatures since crystallization can besuppressed by assimilation Therefore we calcu-

lated AFC processes with two r values of 05 and08 that represent a relatively high and a middle tolow assimilation rate to demonstrate the effects ofvariable r (Fig 9) The results indicate that AFCprocesses involving upper crust similar to theJulianehaEcirc b Granite cannot explain the trend of theEF basalts because Sr isotopic compositionswould shift to considerably more radiogenicvalues This result is consistent with theconclusions derived from trace element dataThe AFC curves for assimilation of lower crusthowever agree well with the bulk of the data Themost negative eNd values can be explained byAFC processes with intermediate assimilationrates (r = 05) involving assimilation of lt5lower crustal material With a greater degree ofassimilation and an r value of 08 lt4assimilation is indicated

Lower crustal rocks of igneous origin have anaverage d18O value of +75plusmn14 with an overallrange from +54 to +125 (Fowler and Harmon1990) 5 to 10 bulk assimilation of lower crustalmaterial with d18O = +75 into a typical mantle-derived magma with d18O = +57 results ind18O values of the contaminated magma from

-24

-20

-16

-12

-8

-4

0

4

0700 0703 0706 0709 0712 0715 0718

r = 08

r = 05

r = 05

r = 08

EF Basalts

Endmembers of AFC calculations

Upper Crust

Lower Crust

Primitive EF basaltic magma

(87Sr86Sr)i

e Nd

(i)

09

08

06

09

08

06

09

0806

09

08

07

07

04

Typical errorCHUR

BS

E

FIG 9 eNd(i) vs (87Sr86Sr)i diagram showing the results of AFC modelling End-member compositions are listed inTable 6 Olivine is the only fractionating mineral Partition coef cients for olivinemelt are from the compilation ofRollinson (1993) Curves were calculated for two values of r (r = rate of assimilation rate of fractionalcrystallization) Dashed lines and numbers in italics are for r = 05 solid lines and regular numbers are for r = 08

Values at the curves are for F values (fraction of melt remaining) with 10 tick marks

848

R HALAMA ETAL

+579 to +588 Therefore the d18O data cannotbe used to rule out small amounts of assimilationof lower crust To evaluate how much contamina-tion would be permissible from the presentoxygen isotope data we calculated how muchcontamination of a mantle-derived magma with18O = +57 with a contaminant within the lowerrange of lower crustal xenoliths (d18O = +67)would drive the oxygen isotopic composition tothe highest value of the EF basalts (d18O~+63) Although simple mixing shows that60 assimilation is permissible this amountwould drive the whole-rock geochemistry tomore silicic compositions than is observed in theEF basalts In Fig 8 there is no obvious trendfrom the basalts towards the Ketilidian basementsuggesting again that upper crust was notassimilated in signi cant amounts

Ratios of trace elements incompatible inbasaltic systems can be used to distinguishbetween crust and mantle reservoirs (egWeaver 1991) Based on the AFC modellingEF basalts with eNd(i) lt ndash1 are those which aremost likely in uenced by crustal contaminationThe relatively high LaNb and ZrNb ratios ofthese samples are consistent with assimilation oflower crustal material SrNd ratios are evenhigher than the average lower crustal composi-tion but SrNd ratios in those samples which weretaken as representative for the lower crust in theAFC modelling can reach values of 80 ie wellabove the highest EF basalt ratios Again theSrNd ratios support the hypothesis of some lowercrustal contamination

The absence of more primitive basaltic rockswith MgO gt8 wt might also indicate lowercrustal assimilation It suggests that magmas mayhave been retained at depth where they acquiredmore evolved compositions by fractional crystal-lization and had the opportunity to assimilatelower crustal material during crustal underplating(Upton et al 2003)

Summary and conclusions

The new geochemical data of the EriksfjordFormation basalts in the Gardar Provinceprovided in this study are used to placepetrogenetic constraints on the sources of theirparental magmas Based on trace-element patternsand Sr-Nd isotopic constraints there seems to beno indication that a DMM-like source played animportant role in magma genesis of the EFbasalts Clinopyroxene and whole-rock REE

patterns are more compatible with an OIB-likesource but we are not able to conclude whetherthe observed enrichment in REE re ects thesourcersquos signature or magmatic processes Theisotopically most primitive EF basalts have Sr-Ndisotopic compositions that are similar to Gardarcarbonatites Due to the isotopic similaritybetween carbonatites and OIBs many arethought to be related to mantle plumes (Bell2001) We therefore speculate that OIB-likemantle components from a plume source werealso involved in the parental EF magmas but wecannot exclude interaction with the SCLM Twodistinct mantle components one isotopicallydepleted and another one enriched could havecontributed to the EF magmas AFC modelling ofSr-Nd isotope data indicates that assimilation oflt5 lower crust could explain the steep vector onthe Sr-Nd isotope plot but assimilation of uppercrustal material is unlikely Oxygen isotope dataof clinopyroxene separates are consistent with amantle origin but permit some assimilation oflower crustal rocks In summary the data areconsistent with the hypothesis that the EFmagmas could have originated from a mantleplume source which is supported by geologicalarguments and that they underwent someassimilation of lower crustal material

Acknowledgements

We would like to thank Bruce Paterson whoprovided invaluable help during Laser ICP-MSmeasurements at the Large-Scale GeochemicalFacility supported by the European Community-Access to Research Infrastructure action of theImproving Human Potential Programme contractnumber HPRI-CT-1999-00008 awarded to ProfBJ Wood (University of Bristol) Gabi Stoschekand Torsten Vennemann are thanked for their helpwith oxygen isotope measurements Elmar Reitterexpertly assisted with preparation and measure-ments of radiogenic isotopes and MathiasWestphal helped with microprobe measurementsREE whole-rock analyses were kindly providedby the NERC facility at Royal HollowayUniversity of London Michael Marks is thankedfor his pleasant company during eldwork Veryconstructive reviews by Tom Andersen andAlexei S Rukhlov greatly helped to improve themanuscript We also thank Ian Coulson for nalediting Funding of this work by the DeutscheForschungsgemeinschaft (grant Ma-21351-2) isgratefully acknowledged

PETROGENESIS OF THE ERIKSFJORD BASALTS

849

References

Andersen T (1997) Age and petrogenesis of theQassiarsuk carbonatite-alkaline silicate volcaniccomplex in the Gardar rift South GreenlandMineralogical Magazine 61 499 ndash513

Armstrong JT (1991) Quantitative elemental analysisof individual microparticles with electron beaminstruments Pp 261 ndash315 in Electron ProbeQuantitation (KFJ Heinrich and DE Newburyeditors) Plenum Press New York and London

Arndt NT and Christensen U (1992) The role oflithospheric mantle in continental ood volcanismThermal and geochemical constraints Journal ofGeophysical Research 97 10967 ndash 10981

Baker JA Menzies MA Thirlwall MF andMacpherson CG (199 7) Pet rogenes is ofQuaternary Intraplate Vocanism Sanarsquoa YemenImplications for Plume-Lithosphere Interaction andPolybaric Melt Hybridization Journal of Petrology 38 1359 ndash 1390

Baker JA MacPherson CG Menzies MAThirlwall MF Al-Kadasi M and Mattey DP(2000) Resolving crustal and mantle contributions tocontinental ood volcanism Yemen constraintsfrom mineral oxygen isotope data Journal ofPetrology 41 1805 ndash1820

Beard JS Abitz RJ and Lofgren GE (1993)Experimental melting of crustal xenoliths fromKilbourne Hole New Mexico and implications forthe contamination and genesi s of magmasContributions to Mineralogy and Petrology 11588 ndash102

Bell K (2001) Carbonatites relationships to mantle-plume activity Geological Society of AmericaSpecial Paper 352 267 ndash290

Bell K and Blenkinsop J (1989) Neodymium andStrontium Isotope Geochemistry of Carbonatites Pp278 ndash300 in Carbonatites ndash Genesis and Evolution(K Bell editor) Unwin Hyman London

Bell K and Simonetti A (1996) Carbonat iteMagmatism and Plume Activity Implications fromthe Nd Pb and Sr Isotope Systematics of OldoinyoLengai Journal of Petrology 37 1321 ndash1339

Bell K and Tilton GR (2001) Nd Pb and Sr IsotopicCompositions of East African Carbonat itesEvide nce for Mant le Mixing and PlumeInhomogene ity Journa l of Petrology 42 1927 ndash1945

Benoit M Polve M and Ceuleneer G (1996) Traceelement and isotopic characterization of ma ccumulates in a fossil mantle diapir (Oman ophiolite)Chemical Geology 134 199 ndash 214

Bernstein S Kelemen PB Tegner C Kurz MDBlusztajn J and Kent Brooks C (1998) Post-breakup basaltic magmatism along the EastGreenland Tertiary rifted margin Earth and

Planetary Science Letters 160 845 ndash862Boynton WV (1984) Geochemistry of the rare earth

elements meteorite studies Pp 63 ndash114 in RareEarth Element Geochemistry (P Henderson editor)Elsevier Amsterdam

Carroll MR and Wyllie PJ (1990) The systemtonalite-H2O at 15 kbar and the genesis of calc-alkaline magmas American Mineralogist 75345 ndash 357

Chauvel C Hofmann AW and Vidal P (1992)HIMU-EM The French Polynesian connectionEarth and Planetary Science Letters 110 99 ndash119

Clayton RN and Mayeda TK (1963) The use ofbromine penta uoride in the extraction of oxygenfrom oxides and silicates for isotope analysisGeochimica et Cosmochimica Acta 27 43 ndash52

Coulson IM and Chambers AD (1996) Patterns ofzonation in rare-earth-bearing minerals in nephelinesyenites of the North Qoroq center South GreenlandThe Canadian Mineralogist 34 1163 ndash1178

Cox KG Bell JD and Pankhurst RJ (1979) TheInterpretation of Igneous Rocks George Allen andUnwin London

Cross W Iddings JP Pirsson LV and WashingtonHS (1903) Quantitative Classi cation of IgneousRocks University of Chicago Press

Dahl-Jensen T Thybo T Hopper H and Rosing M(1998) Crustal structure at the SE Greenland marginfrom wide-angle and normal incidence seismic dataTectonophysics 288 191 ndash 198

DePaolo DJ (1981) Trace element and isotopic effectsof combined wallrock assimilation and fractionalcrystallisation Earth and Planetary Science Letters53 189 ndash 202

De P a o lo D J (1 9 8 8 ) Neodymium Iso topeGeochemistry An Introduction Springer-VerlagNew York

Dobosi G Downes H Mattey D and Embey-IsztinA (1998) Oxygen isotope ratios of phenocrysts fromalkali basalts of the Pannonian basin Evidence foran O-isotopically homogeneous upper mantle be-neath a subduction-in uenced area Lithos 42213 ndash 223

Eiler JM (2001) Oxygen isotope variations of basalticlavas and upper mantle rocks Pp 319 ndash364 inStable Isotope Geochemistry (JW Valley and DRCole editor s) Reviews in Minera logy andGeochemistry 43 The Mineralogical Society ofAmerica Washington DC

Eiler JM Farley KA Valley JW Hauri E CraigH Hart SR and Stolper EM (1997) Oxygenisotope variations in ocean island basalt phenocrystsGeochimica et Cosmochimica Acta 61 2281 ndash2293

Eiler JM Crawford A Elliot T Farley KAValley JW and Stolper EM (2000a) Oxygenisotope geochemistry of oceanic-arc lavas Journal

850

R HALAMA ETAL

TABLE 2 XRF whole-rock analyses of Eriksfjord Formation basalts

Sample EF 024 EF 063 EF 061 EF 039 EF 059 EF 108 EF 072 EF 174 EF 168Group Mussartut Mussartut Mussartut Mussartut Mussartut Ulukasik Ulukasik Ilotildeacutemaussaq IlotildeacutemaussaqSubunit 4 4 5 6 7 Lower Unit Upper Unit Sermilik Narssaq

Fjeld

Major elements (wt)SiO2 4573 4585 4688 4660 4875 4698 4656 4601 4545TiO2 230 200 180 175 191 265 203 299 277Al2O3 1578 1630 1600 1612 1645 1682 1668 1645 1719Fe2O3 1510 1356 1331 1383 1225 1394 1321 1438 1392MnO 028 018 018 018 017 014 017 018 017MgO 525 683 653 673 568 396 647 440 480CaO 842 849 872 857 920 641 808 746 737Na2O 349 274 289 276 236 448 303 323 376K2O 087 083 068 032 027 159 070 205 132P2O5 041 035 023 023 023 048 035 111 083LOI 199 210 240 238 242 204 210 052 156

Total 9962 9923 9962 9947 9969 9949 9938 9878 9914

Mg 453 545 538 536 524 403 538 421 450

Trace elements (ppm)Sc 24 22 25 22 26 19 20 21 16V 209 175 215 195 211 142 170 164 203Cr 81 108 66 65 126 48 82 33 29Co 63 78 70 62 71 44 68 35 46Ni 99 115 85 85 101 58 106 46 51Cu 65 49 46 60 97 41 59 38 41Zn 103 102 110 97 110 93 82 122 105Ga 20 18 21 19 20 19 16 21 19Rb 24 26 16 lt 10 lt 10 38 12 39 45Sr 504 503 512 392 373 712 507 850 992Y 31 26 28 23 28 31 17 31 25Zr 188 165 121 116 125 125 88 164 139Nb 12 11 10 lt 10 10 9 lt 10 31 27Ba 445 375 415 302 178 582 405 1293 900Pb 6 7 3 4 3 4 4 3 5La ndash ndash 15 ndash 12 11 ndash 37 28Ce ndash ndash 40 ndash 38 39 ndash 92 71Pr ndash ndash 1 ndash 1 2 ndash 9 4Nd ndash ndash 16 ndash 15 18 ndash 43 34Sm ndash ndash 6 ndash 7 6 ndash 5 5

CIPW normq ndash ndash ndash ndash 313 ndash ndash ndash ndashor 514 491 402 189 160 940 414 1212 780ab 2807 2319 2445 2335 1997 3207 2564 2733 2981an 2482 2973 2868 3065 3350 2109 2985 2433 2613ne 079 ndash ndash ndash ndash 316 ndash ndash 109di 1187 830 1076 853 877 633 654 452 416hy ndash 890 1059 1650 2217 ndash 1097 315 ndashol 1670 1309 1044 780 ndash 1472 1120 1388 1689mt 365 328 322 335 296 336 319 348 336il 437 380 342 332 363 503 386 568 526ap 095 081 053 053 053 111 081 257 192

Mg was calculated as Mg = 100Mg2+(Mg2++Fe2+) using Fe2O3FeO = 02 Normative mineral compositionswere determined according to the CIPW calculation scheme

PETROGENESIS OF THE ERIKSFJORD BASALTS

837

1 5

2 0

2 5

3 0

3 5

40

60

80

100

120

0

0 5

1

1 5

200

400

600

800

1000

0

50

100

150

15

20

25

30

4 0 45 50 5 5

r2 = 023

r2 = 044r2 = 061

r2 = 053

r2 = 082 r2 = 052

Ni (ppm)

Sc (ppm)

P2O5 (wt)

Sr (ppm)

TiO2 (wt)

Cr (ppm)

Mg Mg

Mg Mg

Mg

40 4 5 50 55

Mg

40 45 50 55

40 45 50 55

40 45 50 5540 4 5 50 55

FIG 3 Whole-rock variation diagrams for EF basalts with Mg as fractionation index Mg is calculated asMg(Mg+Fe2+) with Fe2O3FeO = 02 (Middlemost 1989)

838

R HALAMA ETAL

with the mostly subhedral Fe-Ti oxides and theophitic texture Ratios of incompatible traceelements such as ZrY (range 40 ndash63) andTiY (range 385 ndash716) are relatively uniformthroughout all samples but some scatter isevident Notable geochemical features are highSr contents up to 1000 ppm and very high Bacontents up to 1300 ppm

Normative mineral compositions were deter-mined according to the CIPW calculation scheme(Cross et al 1903 Cox et al 1979) Mostsamples are hypersthene-normative some arenepheline-normative and one (EF 059) is quartz-normative (Table 2) Despite the uncertaintiesinvolved in the CIPW norm calculation the dataindicate that some of the basalts crystallized fromsilica-undersaturated magmas and others fromdistinct silica-saturatedor -oversaturatedmagmas

Multi-element diagrams normalized to primi-tive mantle values show that the EF basalts areenriched in all incompatible elements relative toprimitive mantle values (Fig 4) The patterns ofthe EF basalts are characterized by a prominentBa peak a subordinate P peak and a decreasingelement enrichment with increasing compatibilityIn comparison with normal MORB (N-MORB)enriched MORB (E-MORB) and average OIB(Sun and McDonough 1989) the patterns of theEF basalts are closest to those of OIB both inshape and normalized concentrations

Whole-rock REE data

Whole-rock REE data are presented in Table 3and REE concentrations normalized to chondriticvalues of Boynton (1984) are shown in Fig 5aAll samples are characterized by LREE-enrichedpatterns with no or slightly positive Euanomalies (EuEu = 097 to 113) La concentra-tions are roughly 20 ndash60 times and Lu concentra-tions are 10 times chondritic values NormalizedLREE abundances from La to Nd are fairlyconstant but the patterns show a steady andsmooth decline from Nd towards heavier REEThe REE patterns of the EF basalts fall in betweenthose of typical OIB and E-MORB (Fig 5b)They have similar patterns and absolute abun-dances compared to basaltic dykes from theIvittuut area South Greenland (Goodenough etal 2002) LaNSmN as a measure of LREEfractionation has a restricted range from 130 to165 The degree of HREE fractionationexpressed as GdNYbN is slightly more variableand ranges from 157 to 257 Both these ratios of

the EF basalts fall generally between typical OIBand E-MORB compositions

Clinopyroxene REE patterns

Typical analyses of REE concentrations inclinopyroxenes are given in Table 3 They showthat REE concentrations in individual minerals ofone particular sample can be quite variableChondrite-normalized REE patterns of clinopyr-oxenes are very similar in the samples from theMussartut and Ulukasik Groups (Fig 6ab) Theyare enriched in all REE relative to chondriticvalues and the patterns show a fairly steepincrease from La to Sm followed by a gradualsmooth decrease from Sm to Lu Clinopyroxenesfrom sample EF 168 (Il otildeacutemaussaq Group) aredistinct as they have generally higher REEcontents especially in the LREE and a weak Euanomaly (Fig 6c) The 3 ndash106 enrichment in Nd

c (sa

mpl

e)c

(prim

itive

man

tle)

c (sa

mpl

e)c

(prim

itive

man

tle)

1

10

100

1000

Mussartucirct GroupUlukasik GroupIliacutemaussaq Group

Rb Ba K Nb La Ce Sr Nd P Zr Ti Y

Rb Ba K Nb La Ce Sr Nd P Zr Ti Y01

1

10

100

1000N-MORBE-MORBOIB

FIG 4 Multi-element plots normalized to primitivemantle values (McDonough and Sun 1995) Data forN-MORB E-MORB and average OIB (Sun and

McDonough 1989) are shown for comparison

PETROGENESIS OF THE ERIKSFJORD BASALTS

839

TA

BL

E3

RE

Eda

taof

who

lero

cks

and

clin

opyr

oxen

es

the

latt

erde

term

ined

byL

aser

ICP

-MS

Who

le-r

ock

anal

yses

R

epre

sent

ativ

ecl

inop

yrox

ene

anal

yses

S

ampl

eno

10

1231

1014

4610

1457

1014

7218

6389

1863

95E

F02

4-2

EF

039-

1E

F03

9-2

EF

039-

4E

F07

2-2

EF

072-

3E

F16

8-3

EF

168-

5

RE

Eco

ncen

trat

ions

(ppm

)L

a11

65

180

610

29

661

155

412

28

440

266

199

150

273

304

165

527

59

Ce

290

346

05

249

617

13

378

130

08

180

411

46

868

717

105

012

60

572

399

67

Pr

434

67

388

268

527

433

331

216

181

128

249

262

102

917

43

Nd

200

631

71

176

412

49

250

920

16

190

213

82

122

28

9815

64

157

258

51

960

6S

m4

756

974

043

215

914

826

664

194

283

276

077

0615

82

265

9E

u1

742

51

571

071

971

721

831

491

111

212

001

943

877

07G

d4

826

584

043

555

84

996

965

615

244

326

156

0514

75

252

7T

b ndash

ndash ndash

ndash ndash

ndash1

351

010

770

711

111

212

313

61D

y4

475

223

613

565

354

816

865

934

784

255

996

8613

10

205

2H

o0

921

040

730

731

11

011

481

401

001

001

211

292

394

24E

r2

432

531

951

972

872

663

893

342

642

112

993

646

539

60T

m ndash

ndash ndash

ndash ndash

ndash0

450

390

360

290

350

440

861

27Y

b2

182

071

741

832

592

363

782

812

492

452

703

016

788

89L

u0

340

310

260

270

390

350

460

410

300

310

520

411

251

60

(Eu

Eu

) N1

111

131

190

971

031

070

820

940

720

991

000

910

770

83

Eu

anom

aly

isca

lcul

ated

as(E

uE

u) N

=E

u N(

SmN6

Gd N

)05

norm

aliz

edto

chon

drit

eva

lues

from

Boy

nton

(198

4)

840

R HALAMA ETAL

in clinopyroxenes of EF 168 compared to theother samples (EF 024 039 and 072) seems onlypartly due to a higher degree of fractionationsincethe Mg value of EF 168 is only slightly lowerThus the LREE enrichment appears to underlinethe more alkaline character of this sample Acomparison with published REE patterns ofclinopyroxenes (Fig 6d) shows that the EFpatterns are unlike patterns from olivine gabbroswith MORB-like signatures (Benoit et al 1996)but fairly similar to clinopyroxene megacrystpatterns from an OIB-like alkaline basalt (Nimisand Vannucci 1995)

Sr Nd and O isotopic composition

Sr Nd and O isotopic compositions of all samplesare presented in Table 4 Sr and Nd isotopiccompositions of the clinopyroxene separates havebeen recalculated to an approximate eruption age

of 12 Ga (Paslick et al 1993) using decayconstants of 142610ndash11 andash1 for 87Rb (Steigerand Jager 1977) and 654610ndash12 a ndash1 for 147Sm(Lugmair and Marti 1978) Present day CHURvalues of 01967 for 147Sm144Nd (Jacobsen andWasserburg 1980) and 0512638 for 143Nd144Nd(Goldstein et al 1984) were used to calculateinitial eNd values 87Sr86Sr initial ratios are fairlyconstant and range from 070278 to 070383(Table 4) eNd(i) values vary from ndash32 to +21(Table 4) On the Sr-Nd isotope diagram the EFbasalts de ne a relatively steep trend (Fig 7) withinitial 87Sr86Sr ratios similar to those of the BulkSilicate Earth (BSE) Only one sample is slightlydisplaced towards a more radiogenic initial87Sr86Sr ratio None of the samples has an eNd

value close to depleted mantle values estimatedto be between eNd = +53 (calculated afterDePaolo 1981) and eNd = +73 (calculated afterGoldstein et al 1984) at 12 Ga The two alkalinebasalts of the Il otilde maussaq Group are within therange of the other data but they de ne a restricted eld around CHUR and BSE respectively Errorsinduced by the uncertainty in the age of the EFbasalts are largely within the error of the isotopeanalyses and do not change the overall picture Incomparison with other published Sr-Nd isotopedata from the Gardar Province the initial Srisotope ratios of the EF basalts are very similar torelatively primitive basaltic lamprophyric andcarbonatitic dyke rocks which have (87Sr86Sr)i of~0703 (Pearce and Leng 1996 Andersen 1997Goodenough et al 2002) So far no evidence hasbeen presented for an isotopicallyenriched mantlesource in the Gardar Province as the existingeNd(i) data of ndash05 to +52 show an isotopicallydepleted signature (Pearce and Leng 1996Andersen 1997 Goodenough et al 2002Upton et al 2003) Clinopyroxene from the EFsamples of Paslick et al (1993) yield eNd(i) of+22 The Nd isotope data of this study are withinthe lower range of published values from theGardar Province and extend the spread to morenegative eNd(i) values

Several recent oxygen isotope studies haveshown that mineral separates are less sensitive tolow-T alteration and secondary water take-up thanwhole-rock samples (eg Vroon et al 2001)However only few oxygen isotope data ofmineral separates from continental volcanics areavailable (eg Dobosi et al 1998 Baker et al2000) Extensive studies of mineral separates ofocean island basalts (Eiler et al 1997) andoceanic-arc lavas (Eiler et al 2000a) demon-

c (sa

mpl

e)c

(cho

ndri

te)

c (sa

mp

le)c

(cho

ndrit

e)

1

10

100

200

101446

101472

101231

101457

186389

186395

La Ce Pr Nd Sm Eu Gd Dy Ho Er Yb Lu

1

10

100

200

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

N-MORBE-MORBOIBBasaltic dykes from Ivittuut

a

b

FIG 5 (a) Whole-rock REE data of EF basaltsnormalized to chondritic values from Boynton (1984)(b) Typical compositions of N-MORB E-MORB andOIB (Sun and McDonough 1989) are shown for

comparison

PETROGENESIS OF THE ERIKSFJORD BASALTS

841

strated that analyses of mineral separates consis-tently span a narrower range than comparablewhole-rock data In our study we used clino-pyroxene because it is the only fresh mineralphase present in the samples Oxygen isotopemeasurements of clinopyroxene separates fromthe EF basalts show a range in d18Ocpx from +52to +60 with an average value of +566plusmn023(1s) All samples overlap the range of d18Ocpx

values from mantle peridotites (+525 to +590Mattey et al 1994) from continental oodbasalts from Yemen (+55 to +69 Baker etal 2000) and from alkaline basalts of Hungary(+507 to +534 Dobosi et al 1998) There is asmall difference between the averages ofMussartut Group (d18Ocpx-avg = +575plusmn015(1s)) and Ulukasik Group (d18Ocpx-avg =+575plusmn009 (1s)) samples compared toIl otildeacutemaussaq Group (d18Ocpx-avg = +533plusmn021(1s)) samples

d18O values of melts coexisting with clino-pyroxene were calculated assuming that oxygen

in pyroxene and melt were in isotopic equilibriumand T was 1100ordmC Dmelt-cpx can then be calculatedusing the following equation of Kalamarides(1986)

Dmelt-cpx = ndash00008T + 143 (T in Kelvin)

Although fractionation varies with magmatictemperatures we consider the temperature of1100ordmC and the resulting Dmelt-cpx of +03 assuf ciently representative considering the analy-tical errors involved It is similar to Dmelt-cpx

values used in previous studies (+039 Vroonet al 2001 +02 Dobosi et al 1998)Resulting d18Omelt values of EF basalts rangefrom +55 to +63 These values can becompared to whole-rock data from continentalbasalts given by Harmon and Hoefs (1995)Intraplate basalts from continental rift zonesshow a range from +46 to +83 with a meanof +61 and continental ood basalts which arerather poorly represented in the database havevalues between +43 and +65 Calculated EF

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Cpx from MORB-like

olivine gabbro

Cpx megacrysts from OIB-like

alkaline basalts

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 024EF 039

Mussartucirct Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 168

Iliacutemaussaq Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

1

10

100

1000

EF 072

Ulukasik Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb LuLa Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

c

ba

d

1

10

100

01

FIG 6 REE patterns of clinopyroxenes (cpx) from Mussartut (a) Ulukasik (b) and Il otilde maussaq (c) Group samplesnormalized to chondritic values from Boynton (1984) (d) REE patterns of cpx from MORB-type olivine gabbro(Benoit et al 1996) and OIB-type alkaline basalt (Nimis and Vannucci 1995) for comparison (note the change of

scale)

842

R HALAMA ETAL

TA

BL

E4

Sr

Nd

and

Ois

otop

icco

mpo

siti

ons

ofcl

inop

yrox

ene

sepa

rate

sfr

omE

riks

fjor

dF

orm

atio

nba

salt

sW

hole

-roc

kan

alys

esfr

omth

eK

etil

idia

nba

sem

ent

and

from

sedi

men

tary

rock

sof

the

Eri

ksfj

ord

For

mat

ion

are

give

nfo

rco

mpa

riso

n

Sam

ple

Mem

ber

Sr

Rb

87R

b86S

r87S

r86Sr

87Sr

86Sr

init

ial

SmN

d147Sm

144N

d143N

d144N

de N

d(i

)d1

8O

(ppm

)(p

pm)

(ppm

)(p

pm)

Bas

alts

E

F02

4M

ussa

rtut

646

50

486

002

180

7042

01plusmn

100

7038

267

3822

80

019

560

5124

67plusmn

10ndash

32

567

EF

063

Mus

sart

ut5

60plusmn

011

EF

061

Mus

sart

ut89

95

099

90

0321

070

3818

plusmn10

070

3266

574

163

90

2115

051

2629

plusmn10

ndash2

55

79plusmn

028

EF

039

Mus

sart

ut73

74

058

30

0229

070

3492

plusmn10

070

3098

662

191

30

2092

051

2626

plusmn10

ndash2

25

71plusmn

001

EF

059

Mus

sart

ut54

47

013

70

0073

070

3375

plusmn12

070

3250

606

168

90

2171

051

2727

plusmn09

ndash1

45

99plusmn

001

EF

108

Ulu

kasi

k66

92

111

00

0480

070

3741

plusmn09

070

2916

109

936

82

018

050

5126

17plusmn

10+

21

581

EF

072

Ulu

kasi

k69

41

032

20

0134

070

3014

plusmn10

070

2784

818

251

90

1962

051

2574

plusmn10

ndash1

25

68E

F17

4Il

otildemau

ssaq

125

42

324

005

360

7039

35plusmn

100

7030

1418

76

796

10

1424

051

2230

plusmn10

+0

45

47E

F16

8Il

otildemau

ssaq

866

71

422

004

750

7037

99plusmn

070

7029

8316

45

641

60

1550

051

2304

plusmn09

ndash0

15

18

Bas

emen

tan

dse

dim

enta

ryro

cks

JG02

Juli

aneh

aEcirc b30

01

169

71

6375

074

4789

plusmn10

071

6647

105

362

28

010

220

5114

70plusmn

09ndash

83

78

Gra

nite

EF

021

Qua

rtzi

te11

5E

F02

8A

rkos

ic6

8ar

enit

e

Sam

ples

are

list

edin

stra

tigra

phic

orde

rfr

omol

dest

toyo

unge

st87

Sr86

Sr

init

ialr

atio

san

de N

d(i)

valu

esw

ere

calc

ulat

edfo

ran

age

of1

2G

aS

tand

ard

devi

atio

nsfo

rd18

Oar

egi

ven

for

sam

ples

that

wer

ean

alys

edtw

ice

PETROGENESIS OF THE ERIKSFJORD BASALTS

843

model melt oxygen isotope data are within theranges of both groups Compared to Igaliko dykerocks from further SE in the Gardar Province(Pearce and Leng 1996) the EF basalts cover amuch more restricted eld possibly because d18Owhole-rock analyses are more prone to secondaryalteration than analyses of mineral separates(Fig 8) Whole-rock d18O measurements of theunderlying JulianehaEcirc b Granite and of the sedi-mentary sandstone members of the EriksfjordFormation yield values of 78 and 68 ndash115respectively (Table 4) and are shown forcomparison

Discussion

Mantle source regions of the EF basaltsTo reconstruct the mantle sources involved in theformation of the EF basalts we use Sr-Nd-Oisotope and incompatible trace element data On aSr-Nd isotope diagram the isotopic compositionof DMM was modelled for an age of 12 Ga(DePaolo 1981 Goldstein et al 1984 Taylor andMcLennan 1985) The fairly constant 87Sr86Srinitial isotope ratios of the EF basalts are ~0703which falls within the range typical for mantle-derived rocks but slightly above calculated valuesfor DMM (Fig 7) eNd(i) values are more

-4

-2

0

2

4

6

8

0701 0702 0703 0704 0705 0706

DMM Typical 2 s error

CHUR

BS

E

(87Sr86Sr)i

e Nd

(i)

EM-1

Qassiarsuk carbonatites and lamprophyres

EF basalts

Depleted MORB mantle

Ivittuut basaltic and lamprophyric dikes

Igaliko carbonatites and lamprophyres

Enriched mantle 1

Carbonatites from other regions of similar age

FIG 7 eNd(i) vs (87Sr86Sr)i diagram calculated for t = 12 Ga which is considered as the eruption age of the EFbasalts (Paslick et al 1993) The reference line of Bulk Silicate Earth (BSE) for the Sr isotopic composition wascalculated after DePaolo (1988) assuming present-day values of 00827 for 87Rb86Sr and 07045 for 87Sr86SrIsotopic composition of the depleted MORB mantle (DMM) was calculated after DePaolo (1981) Goldstein et al(1984) and Taylor and McLennan (1985) EM-1 was calculated after Schleicher et al (1998) (see text for details)Data sources are Goodenough et al (2002) Ivittuut (n = 6) Pearce and Leng (1996) Igaliko (n = 5) Andersen(1997) Qassiarsuk (n = 25) Bell and Blenkinsop (1989) other carbonatites of similar age (Goudini Schryburt Lake

and Prairie Lake n = 6)

844

R HALAMA ETAL

scattered and indicate that the rocks contain amixture of isotopic components from both mantleand crustal reservoirs or a heterogeneous mantlesource Only one sample (EF 108) has an eNd(i)value signi cantly above 0 This sample exhibitsthe largest spread of clinopyroxene compositionswhich may re ect strong alteration Goodenoughet al (2002) also note that the highest eNd occursin the most altered sample of their sample setHowever Paslick et al (1993) report clino-pyroxene data from two Ulukasik Groupsamples with eNd(i) gt +2 and those sampleswere apparently not disturbed with regard to theirSm-Nd age information On the other hand Piperet al (1999) suggest that the Sm-Nd systematicsof those samples may be erroneous due to laterdisturbance by uids derived from the Il otildeacutemaussaqintrusion Therefore although we are presentlynot able to decide whether positive eNd(i) values

for the EF basalts are real it is clear that none ofthose values is close to values of a 12 Ga oldDMM source (ie eNd = +5 to +7)

The Sr-Nd isotopic composition for OIB-typesources in the Proterozoic is not as wellconstrained as DMM Today OIB-type mantlesources have a wide range of isotopic composi-tions that partly overlap with DMM but eNd

values are generally lower (Hart et al 1992Hofmann 1997) The EM-1 mantle component ischaracterized by extremely low 143Nd144Ndratios and eNd values of between 0 and ndash5(Hofmann 1997) A possible 12 Ga position forEM-1 is plotted in Fig 7 calculated according toSchleicher et al (1998) with 147Sm144Nd = 018and 87Rb86Sr = 010 The present-day values usedin the calculation are 143Nd144Nd = 051234 and87Sr86Sr = 07057 taken from ocean islandbasalts closest to an EM-1 end-member(Hofmann 1997) Other Sr-Nd isotope datafrom the Gardar Province are plotted to putfurther constraints on the mantle sources of the EFbasalts (Fig 7) We also added comparative datafrom African and Canadian carbonatitesof similarage because there is a well known similaritybetween Sr-Nd isotopic data of OIBs andcarbonatites (Bell and Blenkinsop 1989) Thelinear array of the carbonatite data is consideredto be the result of mixing between the two mantlecomponents HIMU and EM-1 (Bell andBlenkinsop 1989 Bell and Tilton 2001) Thecarbonatites from the Gardar Province havesimilar Nd but slightly more radiogenic Srisotopic compositions than the reference carbo-natite data (Fig 7) The least contaminatedsamples overlap with the linear array and themore radiogenic Sr isotopic composition can beexplained by crustal contamination (Andersen1997) We therefore consider it possible that theGardar carbonatites also represent mixing ofhypothetical HIMU and EM-1 components of anOIB-like source Since the isotopically mostprimitive EF basalt samples overlap with theQassiarsuk carbonatites it could be possible thatthe basalts also contain such mantle componentsalthough Pb isotope data for the EF basalts thatcould clarify the importance of a HIMU sourceare lacking Based on the Sr-Nd isotope data theEF basalts appear to be derived from aheterogeneous OIB-like mantle source thatcomprises isotopically depleted and enrichedcomponents OIB-like and carbonatitic magma-tism are often attributed to mantle plume activity(Bell and Simonetti 1996 Hofmann 1997

2

4

6

8

10

12

-10 -8 -6 -4 -2 0 2 4 6 8

DMM

d18 O

[permil]

e Nd (i)

Typical2s error

EM-1

EF basalts - cpx

EF basalts - model melt

Ketilidian basement

Igaliko dykes

Depleted MORB mantle

Enriched mantle 1

EM-1 lavas - cpx

FIG 8 Oxygen isotopi c compositio ns (d18OV-SMOW) of clinopyroxene from EF basalts calculatedmelt (see text for details) and whole-rock Ketilidianbasement vs eNd(i) Whole-rock data for Igaliko dykerocks (n = 7) are from Pearce and Leng (1996) DMMdata derived from fresh MORB glasses are from Eiler etal (2000b) The eld of EM-1 lavas is taken fromclinopyroxene data of Gough and Tristan da CunhaIsland lavas (Harris et al 2000) and the EM-1 end-member composition is based on the calculations afterSchleicher et al (1998) for eNd(i) and an average value

of 578 (n = 18) from the EM-1 lavas for d18O

PETROGENESIS OF THE ERIKSFJORD BASALTS

845

Simonetti et al 1998) A plume input to theformation of the EF basalts is supported by therelatively large volume of magmas and the shortduration (lt5 Ma) of magmatism (Piper et al1999) It is also compatible with the geologicalsetting of a continental breakup Additionally aplume component appears to be necessary toexplain the thermal input as it is dif cult to melt acold lithosphere section (Arndt and Christensen1992)

Oxygen isotope data of the EF basalts aregenerally consistent with a mantle derivationincluding an OIB-like mantle source On the d18Ovs eNd(i) diagram (Fig 8) the overlap betweenthe EF data clinopyroxene data from EM-1-likeoceanic basalts and the hypothetical EM-1 end-member suggests an involvement of an EM-1-type mantle source Olivines from HIMU-likesources are known to be slightly lower in d18Othan typical upper mantle values from MORBs(Eiler et al 1997) but the EF data are notsuf cient to assess if the slightly lower d18Ovalues in the Il otildeacutemaussaq Group basalts are causedby a relatively higher input from a HIMU source

The contribution of the various mantle sourcesto the EF basalts can also be evaluated on theprimitive mantle-normalized multi-elementdiagram (Fig 4) and on the REE plots (Figs 5and 6) Both the whole-rock REE and theclinopyroxene REE patterns of the EF basaltsare unlike those representative of N-MORBsources We conclude again that DMM did notplay a major role in the petrogenesis of the EFbasalts On the other hand similarities with OIB-like mantle sources are more prominent Theshape of the clinopyroxene REE pattern closelyresembles those from clinopyroxene of OIB-likealkaline basalts despite differences in absoluteconcentrations The whole-rock REE patterns ofthe EF basalts lie between E-MORB and averageOIB patterns as do the multi-element patterns Onthe latter the EF basalts have marked positive Baand P peaks that are not present in typical OIBsHowever Ba contents should be interpreted withcaution since highly variable ZrBa ratios from014 to 070 are an indication that Ba was re-mobilized after emplacement of the basalts (egPrice et al 1991 Oliveira and Tarney 1995)Likewise P mobility was demonstrated in alteredbasaltic rocks (Winchester and Floyd 1976) Insummary the REE and trace element data areevidence that the mantle source(s) of the EFbasalts were enriched in incompatible traceelements relative to primitive mantle

For relatively primitive hy-normative basalticrocks of the Ivittuut area further north in theGardar Province it has been argued that some ofthem were derived from a lithospheric mantlesource previously enriched by subduction-related processes This interpretation is basedon certain geochemical characteristics such ashigh LaNb ratios and negative Nb anomalies(Goodenough et al 2002) For the EF basaltsnegative Nb anomalies are not clearly discern-ible but LaNb ratios are gt1 which seems to be acharacteristic feature of a lithospheric mantlesource (Fitton 1995) LaBa LaRb and LaKratios were also used to argue for a lithosphericorigin of the Gardar basic magmas although amantle plume involvement was assumed toprovide the required energy (Upton andEmeleus 1987 Upton et al 2003) Sincefeldspar is the only primary magmatic mineralin the EF basalts in which Ba Rb and K areincorporated in signi cant amounts and thefeldspars are generally altered these elementsmight have been easily mobilized and wetherefore do not dare to base the interpretationof the EF basalts sources on these data Inaddition the lack of precise geochemical data onthe composition of the lithosphere below theGardar Province complicates an evaluation ofthe lithospheric contribution to the EF basalticmagmatism

Certain ratios of incompatible trace elements ofthe EF basalts are compared to several occur-rences of ocean island basalts representative ofEM-1 and HIMU sources in Table 5 Ratiosbetween highly incompatible trace elements donot change signi cantly during partial melting orfractional crystallization in basaltic systems andmay therefore constrain the ratios in their sources(Weaver et al 1987) The EF basalts aresubdivided into the more alkaline isotopicallyprimitive Il otildeacutemaussaq Group samples and thosesamples with eNd(i) lt ndash1 The Il otilde maussaq Groupsamples have higher SrNd LaNb and ZrNbratios than HIMU and overlap with the upperrange of EM-1 ratios The EF basalts with eNd(i)lt ndash1 have generally higher ratios than the oceanicisland basalts and partly overlap with typicalvalues of the lower crust It therefore seems likelythat assimilation of lower crustal material mayhave changed the primary trace element ratios inthe basalts Therefore we will evaluate thepossible in uence of assimilation of crustalmaterial on the geochemistry of the EF basaltsin the following section

846

R HALAMA ETAL

The role of crustal contamination

Large amounts of bulk assimilation of crustalmaterial which is relatively SiO2-rich comparedto a basaltic melt would drive the compositions tohigher SiO2 contents and quartz-normativecompositions Partial melts derived from crustalrocks would also be relatively SiO2-rich (Beard etal 1993 Carroll and Wyllie 1990) and wouldhave the same effect The ne-normative characterof the EF basalts (Table 1) suggests thatassimilation of crustal material especially ofsilica-rich upper crustal rocks was not signi cantwith the possible exception of sample EF 059

On the primitive mantle-normalized multi-element diagrams (Fig 4) Ti troughs are typicalfor crustal rocks especially for the Upper Crust(Rudnick and Fountain 1995) and they maytherefore be present in rocks extensively contami-nated by continental crust Ti troughs aregenerally lacking in the patterns of the EFbasalts and a major upper crustal contribution tothe magmas can therefore be excluded However

small amounts of assimilation of lower crustalmaterial cannot be ruled out because the averagelower crust has higher Ti contents

Steep trends on Sr-Nd isotope plots have beenattributed to assimilation of granulite-faciesgneisses (eg Bernstein et al 1998) AFCmodelling based on an equation developed byDePaolo (1981) was performed to test thehypothesis of small amounts of contaminationwith lower crustal material (Fig 9) The basalticmagma which represents one end-member for thecalculations has the isotopic signature of theisotopically most primitive samples from thepresent data set and from the literature(Table 6) Two possible contaminants repre-senting upper and lower crustal compositionswere selected for modelling (Table 6) Therepresentative samples are a granite from theJulianehaEcirc b batholith for the upper crust (sampleJG 02) and an average of ve granulite-faciesgneisses with low RbSr ratios from the Archaeancraton of West Greenland (Taylor et al 1984) for

TABLE 5 Trace element ratios of EF basalts compared to ocean island basalts(OIBs) representing EM-1 and HIMU mantle components and the lowercrust

SrNd LaNb ZrNb

EF basalts with eNd lt ndash 1 249 ndash396 120 ndash150 121 ndash157Il otildeacutemaussaq Group EF basalts 198 ndash292 104 ndash119 51 ndash53EM-1 90 ndash207 067 ndash111 39 ndash109HIMU 45 -163 063 ndash080 27 ndash 36Lower continental crust 30 16 136

Data sources EM-1 Humphris and Thompson (1983 Walvis Ridge) and Cliff et al(1991 Inaccessible Islands) HIMU Chauvel et al (1992 Tubuai Island) Lowercontinental crust Rudnick and Fountain (1995)

TABLE 6 End-member compositions used for AFC modelling

End-member Sr (87Sr86Sr)i Nd eNd(i) Data source(ppm) (T = 12 Ga) (ppm) (T = 12 Ga)

Primitive EF basaltic 507 070278 18 + 22 Sample EF 072magma Paslick et al (1993)Upper crust 300 071665 62 ndash83 Sample JG 02(JulianehaEcirc b Granite)Lower crust 747 070668 63 ndash 234 Taylor et al (1984)(Granulite facies gneiss)

The composition of the granulite-facies gneiss is an average of 5 samples with low RbSr ratios lt02

PETROGENESIS OF THE ERIKSFJORD BASALTS

847

the lower crust An in uence of Archaean rockson Gardar magmas has been demonstrated by Pbisotope studies (Taylor and Upton 1993) andseismic data were also interpreted to re ect theexistence of a wedge of Archaean crust thatextends southwards to Lindenow Fjord (Dahl-Jensen et al 1998) We have shown in theprevious sections on the basis of Ni trends thatfractionation of olivine was most relevant to themajor element evolution of the lavas whereasfractionation of other phases was apparently ofminor importance We therefore modelled theAFC processes with olivine fractionation usingmineral-melt partition coef cients of 0014 for Srand 00066 for Nd (Rollinson 1993)

The modelling of AFC processes is stronglydependent on the parameter r de ned as the ratioof mass assimilation rate and fractional crystal-lization rate (DePaolo 1981) The change inisotopic signature becomes enhanced for highervalues of r Reiners et al (1995) have shown thatr values might be very high and even gt1 in theearly stages of magmatic evolution at hightemperatures since crystallization can besuppressed by assimilation Therefore we calcu-

lated AFC processes with two r values of 05 and08 that represent a relatively high and a middle tolow assimilation rate to demonstrate the effects ofvariable r (Fig 9) The results indicate that AFCprocesses involving upper crust similar to theJulianehaEcirc b Granite cannot explain the trend of theEF basalts because Sr isotopic compositionswould shift to considerably more radiogenicvalues This result is consistent with theconclusions derived from trace element dataThe AFC curves for assimilation of lower crusthowever agree well with the bulk of the data Themost negative eNd values can be explained byAFC processes with intermediate assimilationrates (r = 05) involving assimilation of lt5lower crustal material With a greater degree ofassimilation and an r value of 08 lt4assimilation is indicated

Lower crustal rocks of igneous origin have anaverage d18O value of +75plusmn14 with an overallrange from +54 to +125 (Fowler and Harmon1990) 5 to 10 bulk assimilation of lower crustalmaterial with d18O = +75 into a typical mantle-derived magma with d18O = +57 results ind18O values of the contaminated magma from

-24

-20

-16

-12

-8

-4

0

4

0700 0703 0706 0709 0712 0715 0718

r = 08

r = 05

r = 05

r = 08

EF Basalts

Endmembers of AFC calculations

Upper Crust

Lower Crust

Primitive EF basaltic magma

(87Sr86Sr)i

e Nd

(i)

09

08

06

09

08

06

09

0806

09

08

07

07

04

Typical errorCHUR

BS

E

FIG 9 eNd(i) vs (87Sr86Sr)i diagram showing the results of AFC modelling End-member compositions are listed inTable 6 Olivine is the only fractionating mineral Partition coef cients for olivinemelt are from the compilation ofRollinson (1993) Curves were calculated for two values of r (r = rate of assimilation rate of fractionalcrystallization) Dashed lines and numbers in italics are for r = 05 solid lines and regular numbers are for r = 08

Values at the curves are for F values (fraction of melt remaining) with 10 tick marks

848

R HALAMA ETAL

+579 to +588 Therefore the d18O data cannotbe used to rule out small amounts of assimilationof lower crust To evaluate how much contamina-tion would be permissible from the presentoxygen isotope data we calculated how muchcontamination of a mantle-derived magma with18O = +57 with a contaminant within the lowerrange of lower crustal xenoliths (d18O = +67)would drive the oxygen isotopic composition tothe highest value of the EF basalts (d18O~+63) Although simple mixing shows that60 assimilation is permissible this amountwould drive the whole-rock geochemistry tomore silicic compositions than is observed in theEF basalts In Fig 8 there is no obvious trendfrom the basalts towards the Ketilidian basementsuggesting again that upper crust was notassimilated in signi cant amounts

Ratios of trace elements incompatible inbasaltic systems can be used to distinguishbetween crust and mantle reservoirs (egWeaver 1991) Based on the AFC modellingEF basalts with eNd(i) lt ndash1 are those which aremost likely in uenced by crustal contaminationThe relatively high LaNb and ZrNb ratios ofthese samples are consistent with assimilation oflower crustal material SrNd ratios are evenhigher than the average lower crustal composi-tion but SrNd ratios in those samples which weretaken as representative for the lower crust in theAFC modelling can reach values of 80 ie wellabove the highest EF basalt ratios Again theSrNd ratios support the hypothesis of some lowercrustal contamination

The absence of more primitive basaltic rockswith MgO gt8 wt might also indicate lowercrustal assimilation It suggests that magmas mayhave been retained at depth where they acquiredmore evolved compositions by fractional crystal-lization and had the opportunity to assimilatelower crustal material during crustal underplating(Upton et al 2003)

Summary and conclusions

The new geochemical data of the EriksfjordFormation basalts in the Gardar Provinceprovided in this study are used to placepetrogenetic constraints on the sources of theirparental magmas Based on trace-element patternsand Sr-Nd isotopic constraints there seems to beno indication that a DMM-like source played animportant role in magma genesis of the EFbasalts Clinopyroxene and whole-rock REE

patterns are more compatible with an OIB-likesource but we are not able to conclude whetherthe observed enrichment in REE re ects thesourcersquos signature or magmatic processes Theisotopically most primitive EF basalts have Sr-Ndisotopic compositions that are similar to Gardarcarbonatites Due to the isotopic similaritybetween carbonatites and OIBs many arethought to be related to mantle plumes (Bell2001) We therefore speculate that OIB-likemantle components from a plume source werealso involved in the parental EF magmas but wecannot exclude interaction with the SCLM Twodistinct mantle components one isotopicallydepleted and another one enriched could havecontributed to the EF magmas AFC modelling ofSr-Nd isotope data indicates that assimilation oflt5 lower crust could explain the steep vector onthe Sr-Nd isotope plot but assimilation of uppercrustal material is unlikely Oxygen isotope dataof clinopyroxene separates are consistent with amantle origin but permit some assimilation oflower crustal rocks In summary the data areconsistent with the hypothesis that the EFmagmas could have originated from a mantleplume source which is supported by geologicalarguments and that they underwent someassimilation of lower crustal material

Acknowledgements

We would like to thank Bruce Paterson whoprovided invaluable help during Laser ICP-MSmeasurements at the Large-Scale GeochemicalFacility supported by the European Community-Access to Research Infrastructure action of theImproving Human Potential Programme contractnumber HPRI-CT-1999-00008 awarded to ProfBJ Wood (University of Bristol) Gabi Stoschekand Torsten Vennemann are thanked for their helpwith oxygen isotope measurements Elmar Reitterexpertly assisted with preparation and measure-ments of radiogenic isotopes and MathiasWestphal helped with microprobe measurementsREE whole-rock analyses were kindly providedby the NERC facility at Royal HollowayUniversity of London Michael Marks is thankedfor his pleasant company during eldwork Veryconstructive reviews by Tom Andersen andAlexei S Rukhlov greatly helped to improve themanuscript We also thank Ian Coulson for nalediting Funding of this work by the DeutscheForschungsgemeinschaft (grant Ma-21351-2) isgratefully acknowledged

PETROGENESIS OF THE ERIKSFJORD BASALTS

849

References

Andersen T (1997) Age and petrogenesis of theQassiarsuk carbonatite-alkaline silicate volcaniccomplex in the Gardar rift South GreenlandMineralogical Magazine 61 499 ndash513

Armstrong JT (1991) Quantitative elemental analysisof individual microparticles with electron beaminstruments Pp 261 ndash315 in Electron ProbeQuantitation (KFJ Heinrich and DE Newburyeditors) Plenum Press New York and London

Arndt NT and Christensen U (1992) The role oflithospheric mantle in continental ood volcanismThermal and geochemical constraints Journal ofGeophysical Research 97 10967 ndash 10981

Baker JA Menzies MA Thirlwall MF andMacpherson CG (199 7) Pet rogenes is ofQuaternary Intraplate Vocanism Sanarsquoa YemenImplications for Plume-Lithosphere Interaction andPolybaric Melt Hybridization Journal of Petrology 38 1359 ndash 1390

Baker JA MacPherson CG Menzies MAThirlwall MF Al-Kadasi M and Mattey DP(2000) Resolving crustal and mantle contributions tocontinental ood volcanism Yemen constraintsfrom mineral oxygen isotope data Journal ofPetrology 41 1805 ndash1820

Beard JS Abitz RJ and Lofgren GE (1993)Experimental melting of crustal xenoliths fromKilbourne Hole New Mexico and implications forthe contamination and genesi s of magmasContributions to Mineralogy and Petrology 11588 ndash102

Bell K (2001) Carbonatites relationships to mantle-plume activity Geological Society of AmericaSpecial Paper 352 267 ndash290

Bell K and Blenkinsop J (1989) Neodymium andStrontium Isotope Geochemistry of Carbonatites Pp278 ndash300 in Carbonatites ndash Genesis and Evolution(K Bell editor) Unwin Hyman London

Bell K and Simonetti A (1996) Carbonat iteMagmatism and Plume Activity Implications fromthe Nd Pb and Sr Isotope Systematics of OldoinyoLengai Journal of Petrology 37 1321 ndash1339

Bell K and Tilton GR (2001) Nd Pb and Sr IsotopicCompositions of East African Carbonat itesEvide nce for Mant le Mixing and PlumeInhomogene ity Journa l of Petrology 42 1927 ndash1945

Benoit M Polve M and Ceuleneer G (1996) Traceelement and isotopic characterization of ma ccumulates in a fossil mantle diapir (Oman ophiolite)Chemical Geology 134 199 ndash 214

Bernstein S Kelemen PB Tegner C Kurz MDBlusztajn J and Kent Brooks C (1998) Post-breakup basaltic magmatism along the EastGreenland Tertiary rifted margin Earth and

Planetary Science Letters 160 845 ndash862Boynton WV (1984) Geochemistry of the rare earth

elements meteorite studies Pp 63 ndash114 in RareEarth Element Geochemistry (P Henderson editor)Elsevier Amsterdam

Carroll MR and Wyllie PJ (1990) The systemtonalite-H2O at 15 kbar and the genesis of calc-alkaline magmas American Mineralogist 75345 ndash 357

Chauvel C Hofmann AW and Vidal P (1992)HIMU-EM The French Polynesian connectionEarth and Planetary Science Letters 110 99 ndash119

Clayton RN and Mayeda TK (1963) The use ofbromine penta uoride in the extraction of oxygenfrom oxides and silicates for isotope analysisGeochimica et Cosmochimica Acta 27 43 ndash52

Coulson IM and Chambers AD (1996) Patterns ofzonation in rare-earth-bearing minerals in nephelinesyenites of the North Qoroq center South GreenlandThe Canadian Mineralogist 34 1163 ndash1178

Cox KG Bell JD and Pankhurst RJ (1979) TheInterpretation of Igneous Rocks George Allen andUnwin London

Cross W Iddings JP Pirsson LV and WashingtonHS (1903) Quantitative Classi cation of IgneousRocks University of Chicago Press

Dahl-Jensen T Thybo T Hopper H and Rosing M(1998) Crustal structure at the SE Greenland marginfrom wide-angle and normal incidence seismic dataTectonophysics 288 191 ndash 198

DePaolo DJ (1981) Trace element and isotopic effectsof combined wallrock assimilation and fractionalcrystallisation Earth and Planetary Science Letters53 189 ndash 202

De P a o lo D J (1 9 8 8 ) Neodymium Iso topeGeochemistry An Introduction Springer-VerlagNew York

Dobosi G Downes H Mattey D and Embey-IsztinA (1998) Oxygen isotope ratios of phenocrysts fromalkali basalts of the Pannonian basin Evidence foran O-isotopically homogeneous upper mantle be-neath a subduction-in uenced area Lithos 42213 ndash 223

Eiler JM (2001) Oxygen isotope variations of basalticlavas and upper mantle rocks Pp 319 ndash364 inStable Isotope Geochemistry (JW Valley and DRCole editor s) Reviews in Minera logy andGeochemistry 43 The Mineralogical Society ofAmerica Washington DC

Eiler JM Farley KA Valley JW Hauri E CraigH Hart SR and Stolper EM (1997) Oxygenisotope variations in ocean island basalt phenocrystsGeochimica et Cosmochimica Acta 61 2281 ndash2293

Eiler JM Crawford A Elliot T Farley KAValley JW and Stolper EM (2000a) Oxygenisotope geochemistry of oceanic-arc lavas Journal

850

R HALAMA ETAL

1 5

2 0

2 5

3 0

3 5

40

60

80

100

120

0

0 5

1

1 5

200

400

600

800

1000

0

50

100

150

15

20

25

30

4 0 45 50 5 5

r2 = 023

r2 = 044r2 = 061

r2 = 053

r2 = 082 r2 = 052

Ni (ppm)

Sc (ppm)

P2O5 (wt)

Sr (ppm)

TiO2 (wt)

Cr (ppm)

Mg Mg

Mg Mg

Mg

40 4 5 50 55

Mg

40 45 50 55

40 45 50 55

40 45 50 5540 4 5 50 55

FIG 3 Whole-rock variation diagrams for EF basalts with Mg as fractionation index Mg is calculated asMg(Mg+Fe2+) with Fe2O3FeO = 02 (Middlemost 1989)

838

R HALAMA ETAL

with the mostly subhedral Fe-Ti oxides and theophitic texture Ratios of incompatible traceelements such as ZrY (range 40 ndash63) andTiY (range 385 ndash716) are relatively uniformthroughout all samples but some scatter isevident Notable geochemical features are highSr contents up to 1000 ppm and very high Bacontents up to 1300 ppm

Normative mineral compositions were deter-mined according to the CIPW calculation scheme(Cross et al 1903 Cox et al 1979) Mostsamples are hypersthene-normative some arenepheline-normative and one (EF 059) is quartz-normative (Table 2) Despite the uncertaintiesinvolved in the CIPW norm calculation the dataindicate that some of the basalts crystallized fromsilica-undersaturated magmas and others fromdistinct silica-saturatedor -oversaturatedmagmas

Multi-element diagrams normalized to primi-tive mantle values show that the EF basalts areenriched in all incompatible elements relative toprimitive mantle values (Fig 4) The patterns ofthe EF basalts are characterized by a prominentBa peak a subordinate P peak and a decreasingelement enrichment with increasing compatibilityIn comparison with normal MORB (N-MORB)enriched MORB (E-MORB) and average OIB(Sun and McDonough 1989) the patterns of theEF basalts are closest to those of OIB both inshape and normalized concentrations

Whole-rock REE data

Whole-rock REE data are presented in Table 3and REE concentrations normalized to chondriticvalues of Boynton (1984) are shown in Fig 5aAll samples are characterized by LREE-enrichedpatterns with no or slightly positive Euanomalies (EuEu = 097 to 113) La concentra-tions are roughly 20 ndash60 times and Lu concentra-tions are 10 times chondritic values NormalizedLREE abundances from La to Nd are fairlyconstant but the patterns show a steady andsmooth decline from Nd towards heavier REEThe REE patterns of the EF basalts fall in betweenthose of typical OIB and E-MORB (Fig 5b)They have similar patterns and absolute abun-dances compared to basaltic dykes from theIvittuut area South Greenland (Goodenough etal 2002) LaNSmN as a measure of LREEfractionation has a restricted range from 130 to165 The degree of HREE fractionationexpressed as GdNYbN is slightly more variableand ranges from 157 to 257 Both these ratios of

the EF basalts fall generally between typical OIBand E-MORB compositions

Clinopyroxene REE patterns

Typical analyses of REE concentrations inclinopyroxenes are given in Table 3 They showthat REE concentrations in individual minerals ofone particular sample can be quite variableChondrite-normalized REE patterns of clinopyr-oxenes are very similar in the samples from theMussartut and Ulukasik Groups (Fig 6ab) Theyare enriched in all REE relative to chondriticvalues and the patterns show a fairly steepincrease from La to Sm followed by a gradualsmooth decrease from Sm to Lu Clinopyroxenesfrom sample EF 168 (Il otildeacutemaussaq Group) aredistinct as they have generally higher REEcontents especially in the LREE and a weak Euanomaly (Fig 6c) The 3 ndash106 enrichment in Nd

c (sa

mpl

e)c

(prim

itive

man

tle)

c (sa

mpl

e)c

(prim

itive

man

tle)

1

10

100

1000

Mussartucirct GroupUlukasik GroupIliacutemaussaq Group

Rb Ba K Nb La Ce Sr Nd P Zr Ti Y

Rb Ba K Nb La Ce Sr Nd P Zr Ti Y01

1

10

100

1000N-MORBE-MORBOIB

FIG 4 Multi-element plots normalized to primitivemantle values (McDonough and Sun 1995) Data forN-MORB E-MORB and average OIB (Sun and

McDonough 1989) are shown for comparison

PETROGENESIS OF THE ERIKSFJORD BASALTS

839

TA

BL

E3

RE

Eda

taof

who

lero

cks

and

clin

opyr

oxen

es

the

latt

erde

term

ined

byL

aser

ICP

-MS

Who

le-r

ock

anal

yses

R

epre

sent

ativ

ecl

inop

yrox

ene

anal

yses

S

ampl

eno

10

1231

1014

4610

1457

1014

7218

6389

1863

95E

F02

4-2

EF

039-

1E

F03

9-2

EF

039-

4E

F07

2-2

EF

072-

3E

F16

8-3

EF

168-

5

RE

Eco

ncen

trat

ions

(ppm

)L

a11

65

180

610

29

661

155

412

28

440

266

199

150

273

304

165

527

59

Ce

290

346

05

249

617

13

378

130

08

180

411

46

868

717

105

012

60

572

399

67

Pr

434

67

388

268

527

433

331

216

181

128

249

262

102

917

43

Nd

200

631

71

176

412

49

250

920

16

190

213

82

122

28

9815

64

157

258

51

960

6S

m4

756

974

043

215

914

826

664

194

283

276

077

0615

82

265

9E

u1

742

51

571

071

971

721

831

491

111

212

001

943

877

07G

d4

826

584

043

555

84

996

965

615

244

326

156

0514

75

252

7T

b ndash

ndash ndash

ndash ndash

ndash1

351

010

770

711

111

212

313

61D

y4

475

223

613

565

354

816

865

934

784

255

996

8613

10

205

2H

o0

921

040

730

731

11

011

481

401

001

001

211

292

394

24E

r2

432

531

951

972

872

663

893

342

642

112

993

646

539

60T

m ndash

ndash ndash

ndash ndash

ndash0

450

390

360

290

350

440

861

27Y

b2

182

071

741

832

592

363

782

812

492

452

703

016

788

89L

u0

340

310

260

270

390

350

460

410

300

310

520

411

251

60

(Eu

Eu

) N1

111

131

190

971

031

070

820

940

720

991

000

910

770

83

Eu

anom

aly

isca

lcul

ated

as(E

uE

u) N

=E

u N(

SmN6

Gd N

)05

norm

aliz

edto

chon

drit

eva

lues

from

Boy

nton

(198

4)

840

R HALAMA ETAL

in clinopyroxenes of EF 168 compared to theother samples (EF 024 039 and 072) seems onlypartly due to a higher degree of fractionationsincethe Mg value of EF 168 is only slightly lowerThus the LREE enrichment appears to underlinethe more alkaline character of this sample Acomparison with published REE patterns ofclinopyroxenes (Fig 6d) shows that the EFpatterns are unlike patterns from olivine gabbroswith MORB-like signatures (Benoit et al 1996)but fairly similar to clinopyroxene megacrystpatterns from an OIB-like alkaline basalt (Nimisand Vannucci 1995)

Sr Nd and O isotopic composition

Sr Nd and O isotopic compositions of all samplesare presented in Table 4 Sr and Nd isotopiccompositions of the clinopyroxene separates havebeen recalculated to an approximate eruption age

of 12 Ga (Paslick et al 1993) using decayconstants of 142610ndash11 andash1 for 87Rb (Steigerand Jager 1977) and 654610ndash12 a ndash1 for 147Sm(Lugmair and Marti 1978) Present day CHURvalues of 01967 for 147Sm144Nd (Jacobsen andWasserburg 1980) and 0512638 for 143Nd144Nd(Goldstein et al 1984) were used to calculateinitial eNd values 87Sr86Sr initial ratios are fairlyconstant and range from 070278 to 070383(Table 4) eNd(i) values vary from ndash32 to +21(Table 4) On the Sr-Nd isotope diagram the EFbasalts de ne a relatively steep trend (Fig 7) withinitial 87Sr86Sr ratios similar to those of the BulkSilicate Earth (BSE) Only one sample is slightlydisplaced towards a more radiogenic initial87Sr86Sr ratio None of the samples has an eNd

value close to depleted mantle values estimatedto be between eNd = +53 (calculated afterDePaolo 1981) and eNd = +73 (calculated afterGoldstein et al 1984) at 12 Ga The two alkalinebasalts of the Il otilde maussaq Group are within therange of the other data but they de ne a restricted eld around CHUR and BSE respectively Errorsinduced by the uncertainty in the age of the EFbasalts are largely within the error of the isotopeanalyses and do not change the overall picture Incomparison with other published Sr-Nd isotopedata from the Gardar Province the initial Srisotope ratios of the EF basalts are very similar torelatively primitive basaltic lamprophyric andcarbonatitic dyke rocks which have (87Sr86Sr)i of~0703 (Pearce and Leng 1996 Andersen 1997Goodenough et al 2002) So far no evidence hasbeen presented for an isotopicallyenriched mantlesource in the Gardar Province as the existingeNd(i) data of ndash05 to +52 show an isotopicallydepleted signature (Pearce and Leng 1996Andersen 1997 Goodenough et al 2002Upton et al 2003) Clinopyroxene from the EFsamples of Paslick et al (1993) yield eNd(i) of+22 The Nd isotope data of this study are withinthe lower range of published values from theGardar Province and extend the spread to morenegative eNd(i) values

Several recent oxygen isotope studies haveshown that mineral separates are less sensitive tolow-T alteration and secondary water take-up thanwhole-rock samples (eg Vroon et al 2001)However only few oxygen isotope data ofmineral separates from continental volcanics areavailable (eg Dobosi et al 1998 Baker et al2000) Extensive studies of mineral separates ofocean island basalts (Eiler et al 1997) andoceanic-arc lavas (Eiler et al 2000a) demon-

c (sa

mpl

e)c

(cho

ndri

te)

c (sa

mp

le)c

(cho

ndrit

e)

1

10

100

200

101446

101472

101231

101457

186389

186395

La Ce Pr Nd Sm Eu Gd Dy Ho Er Yb Lu

1

10

100

200

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

N-MORBE-MORBOIBBasaltic dykes from Ivittuut

a

b

FIG 5 (a) Whole-rock REE data of EF basaltsnormalized to chondritic values from Boynton (1984)(b) Typical compositions of N-MORB E-MORB andOIB (Sun and McDonough 1989) are shown for

comparison

PETROGENESIS OF THE ERIKSFJORD BASALTS

841

strated that analyses of mineral separates consis-tently span a narrower range than comparablewhole-rock data In our study we used clino-pyroxene because it is the only fresh mineralphase present in the samples Oxygen isotopemeasurements of clinopyroxene separates fromthe EF basalts show a range in d18Ocpx from +52to +60 with an average value of +566plusmn023(1s) All samples overlap the range of d18Ocpx

values from mantle peridotites (+525 to +590Mattey et al 1994) from continental oodbasalts from Yemen (+55 to +69 Baker etal 2000) and from alkaline basalts of Hungary(+507 to +534 Dobosi et al 1998) There is asmall difference between the averages ofMussartut Group (d18Ocpx-avg = +575plusmn015(1s)) and Ulukasik Group (d18Ocpx-avg =+575plusmn009 (1s)) samples compared toIl otildeacutemaussaq Group (d18Ocpx-avg = +533plusmn021(1s)) samples

d18O values of melts coexisting with clino-pyroxene were calculated assuming that oxygen

in pyroxene and melt were in isotopic equilibriumand T was 1100ordmC Dmelt-cpx can then be calculatedusing the following equation of Kalamarides(1986)

Dmelt-cpx = ndash00008T + 143 (T in Kelvin)

Although fractionation varies with magmatictemperatures we consider the temperature of1100ordmC and the resulting Dmelt-cpx of +03 assuf ciently representative considering the analy-tical errors involved It is similar to Dmelt-cpx

values used in previous studies (+039 Vroonet al 2001 +02 Dobosi et al 1998)Resulting d18Omelt values of EF basalts rangefrom +55 to +63 These values can becompared to whole-rock data from continentalbasalts given by Harmon and Hoefs (1995)Intraplate basalts from continental rift zonesshow a range from +46 to +83 with a meanof +61 and continental ood basalts which arerather poorly represented in the database havevalues between +43 and +65 Calculated EF

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Cpx from MORB-like

olivine gabbro

Cpx megacrysts from OIB-like

alkaline basalts

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 024EF 039

Mussartucirct Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 168

Iliacutemaussaq Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

1

10

100

1000

EF 072

Ulukasik Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb LuLa Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

c

ba

d

1

10

100

01

FIG 6 REE patterns of clinopyroxenes (cpx) from Mussartut (a) Ulukasik (b) and Il otilde maussaq (c) Group samplesnormalized to chondritic values from Boynton (1984) (d) REE patterns of cpx from MORB-type olivine gabbro(Benoit et al 1996) and OIB-type alkaline basalt (Nimis and Vannucci 1995) for comparison (note the change of

scale)

842

R HALAMA ETAL

TA

BL

E4

Sr

Nd

and

Ois

otop

icco

mpo

siti

ons

ofcl

inop

yrox

ene

sepa

rate

sfr

omE

riks

fjor

dF

orm

atio

nba

salt

sW

hole

-roc

kan

alys

esfr

omth

eK

etil

idia

nba

sem

ent

and

from

sedi

men

tary

rock

sof

the

Eri

ksfj

ord

For

mat

ion

are

give

nfo

rco

mpa

riso

n

Sam

ple

Mem

ber

Sr

Rb

87R

b86S

r87S

r86Sr

87Sr

86Sr

init

ial

SmN

d147Sm

144N

d143N

d144N

de N

d(i

)d1

8O

(ppm

)(p

pm)

(ppm

)(p

pm)

Bas

alts

E

F02

4M

ussa

rtut

646

50

486

002

180

7042

01plusmn

100

7038

267

3822

80

019

560

5124

67plusmn

10ndash

32

567

EF

063

Mus

sart

ut5

60plusmn

011

EF

061

Mus

sart

ut89

95

099

90

0321

070

3818

plusmn10

070

3266

574

163

90

2115

051

2629

plusmn10

ndash2

55

79plusmn

028

EF

039

Mus

sart

ut73

74

058

30

0229

070

3492

plusmn10

070

3098

662

191

30

2092

051

2626

plusmn10

ndash2

25

71plusmn

001

EF

059

Mus

sart

ut54

47

013

70

0073

070

3375

plusmn12

070

3250

606

168

90

2171

051

2727

plusmn09

ndash1

45

99plusmn

001

EF

108

Ulu

kasi

k66

92

111

00

0480

070

3741

plusmn09

070

2916

109

936

82

018

050

5126

17plusmn

10+

21

581

EF

072

Ulu

kasi

k69

41

032

20

0134

070

3014

plusmn10

070

2784

818

251

90

1962

051

2574

plusmn10

ndash1

25

68E

F17

4Il

otildemau

ssaq

125

42

324

005

360

7039

35plusmn

100

7030

1418

76

796

10

1424

051

2230

plusmn10

+0

45

47E

F16

8Il

otildemau

ssaq

866

71

422

004

750

7037

99plusmn

070

7029

8316

45

641

60

1550

051

2304

plusmn09

ndash0

15

18

Bas

emen

tan

dse

dim

enta

ryro

cks

JG02

Juli

aneh

aEcirc b30

01

169

71

6375

074

4789

plusmn10

071

6647

105

362

28

010

220

5114

70plusmn

09ndash

83

78

Gra

nite

EF

021

Qua

rtzi

te11

5E

F02

8A

rkos

ic6

8ar

enit

e

Sam

ples

are

list

edin

stra

tigra

phic

orde

rfr

omol

dest

toyo

unge

st87

Sr86

Sr

init

ialr

atio

san

de N

d(i)

valu

esw

ere

calc

ulat

edfo

ran

age

of1

2G

aS

tand

ard

devi

atio

nsfo

rd18

Oar

egi

ven

for

sam

ples

that

wer

ean

alys

edtw

ice

PETROGENESIS OF THE ERIKSFJORD BASALTS

843

model melt oxygen isotope data are within theranges of both groups Compared to Igaliko dykerocks from further SE in the Gardar Province(Pearce and Leng 1996) the EF basalts cover amuch more restricted eld possibly because d18Owhole-rock analyses are more prone to secondaryalteration than analyses of mineral separates(Fig 8) Whole-rock d18O measurements of theunderlying JulianehaEcirc b Granite and of the sedi-mentary sandstone members of the EriksfjordFormation yield values of 78 and 68 ndash115respectively (Table 4) and are shown forcomparison

Discussion

Mantle source regions of the EF basaltsTo reconstruct the mantle sources involved in theformation of the EF basalts we use Sr-Nd-Oisotope and incompatible trace element data On aSr-Nd isotope diagram the isotopic compositionof DMM was modelled for an age of 12 Ga(DePaolo 1981 Goldstein et al 1984 Taylor andMcLennan 1985) The fairly constant 87Sr86Srinitial isotope ratios of the EF basalts are ~0703which falls within the range typical for mantle-derived rocks but slightly above calculated valuesfor DMM (Fig 7) eNd(i) values are more

-4

-2

0

2

4

6

8

0701 0702 0703 0704 0705 0706

DMM Typical 2 s error

CHUR

BS

E

(87Sr86Sr)i

e Nd

(i)

EM-1

Qassiarsuk carbonatites and lamprophyres

EF basalts

Depleted MORB mantle

Ivittuut basaltic and lamprophyric dikes

Igaliko carbonatites and lamprophyres

Enriched mantle 1

Carbonatites from other regions of similar age

FIG 7 eNd(i) vs (87Sr86Sr)i diagram calculated for t = 12 Ga which is considered as the eruption age of the EFbasalts (Paslick et al 1993) The reference line of Bulk Silicate Earth (BSE) for the Sr isotopic composition wascalculated after DePaolo (1988) assuming present-day values of 00827 for 87Rb86Sr and 07045 for 87Sr86SrIsotopic composition of the depleted MORB mantle (DMM) was calculated after DePaolo (1981) Goldstein et al(1984) and Taylor and McLennan (1985) EM-1 was calculated after Schleicher et al (1998) (see text for details)Data sources are Goodenough et al (2002) Ivittuut (n = 6) Pearce and Leng (1996) Igaliko (n = 5) Andersen(1997) Qassiarsuk (n = 25) Bell and Blenkinsop (1989) other carbonatites of similar age (Goudini Schryburt Lake

and Prairie Lake n = 6)

844

R HALAMA ETAL

scattered and indicate that the rocks contain amixture of isotopic components from both mantleand crustal reservoirs or a heterogeneous mantlesource Only one sample (EF 108) has an eNd(i)value signi cantly above 0 This sample exhibitsthe largest spread of clinopyroxene compositionswhich may re ect strong alteration Goodenoughet al (2002) also note that the highest eNd occursin the most altered sample of their sample setHowever Paslick et al (1993) report clino-pyroxene data from two Ulukasik Groupsamples with eNd(i) gt +2 and those sampleswere apparently not disturbed with regard to theirSm-Nd age information On the other hand Piperet al (1999) suggest that the Sm-Nd systematicsof those samples may be erroneous due to laterdisturbance by uids derived from the Il otildeacutemaussaqintrusion Therefore although we are presentlynot able to decide whether positive eNd(i) values

for the EF basalts are real it is clear that none ofthose values is close to values of a 12 Ga oldDMM source (ie eNd = +5 to +7)

The Sr-Nd isotopic composition for OIB-typesources in the Proterozoic is not as wellconstrained as DMM Today OIB-type mantlesources have a wide range of isotopic composi-tions that partly overlap with DMM but eNd

values are generally lower (Hart et al 1992Hofmann 1997) The EM-1 mantle component ischaracterized by extremely low 143Nd144Ndratios and eNd values of between 0 and ndash5(Hofmann 1997) A possible 12 Ga position forEM-1 is plotted in Fig 7 calculated according toSchleicher et al (1998) with 147Sm144Nd = 018and 87Rb86Sr = 010 The present-day values usedin the calculation are 143Nd144Nd = 051234 and87Sr86Sr = 07057 taken from ocean islandbasalts closest to an EM-1 end-member(Hofmann 1997) Other Sr-Nd isotope datafrom the Gardar Province are plotted to putfurther constraints on the mantle sources of the EFbasalts (Fig 7) We also added comparative datafrom African and Canadian carbonatitesof similarage because there is a well known similaritybetween Sr-Nd isotopic data of OIBs andcarbonatites (Bell and Blenkinsop 1989) Thelinear array of the carbonatite data is consideredto be the result of mixing between the two mantlecomponents HIMU and EM-1 (Bell andBlenkinsop 1989 Bell and Tilton 2001) Thecarbonatites from the Gardar Province havesimilar Nd but slightly more radiogenic Srisotopic compositions than the reference carbo-natite data (Fig 7) The least contaminatedsamples overlap with the linear array and themore radiogenic Sr isotopic composition can beexplained by crustal contamination (Andersen1997) We therefore consider it possible that theGardar carbonatites also represent mixing ofhypothetical HIMU and EM-1 components of anOIB-like source Since the isotopically mostprimitive EF basalt samples overlap with theQassiarsuk carbonatites it could be possible thatthe basalts also contain such mantle componentsalthough Pb isotope data for the EF basalts thatcould clarify the importance of a HIMU sourceare lacking Based on the Sr-Nd isotope data theEF basalts appear to be derived from aheterogeneous OIB-like mantle source thatcomprises isotopically depleted and enrichedcomponents OIB-like and carbonatitic magma-tism are often attributed to mantle plume activity(Bell and Simonetti 1996 Hofmann 1997

2

4

6

8

10

12

-10 -8 -6 -4 -2 0 2 4 6 8

DMM

d18 O

[permil]

e Nd (i)

Typical2s error

EM-1

EF basalts - cpx

EF basalts - model melt

Ketilidian basement

Igaliko dykes

Depleted MORB mantle

Enriched mantle 1

EM-1 lavas - cpx

FIG 8 Oxygen isotopi c compositio ns (d18OV-SMOW) of clinopyroxene from EF basalts calculatedmelt (see text for details) and whole-rock Ketilidianbasement vs eNd(i) Whole-rock data for Igaliko dykerocks (n = 7) are from Pearce and Leng (1996) DMMdata derived from fresh MORB glasses are from Eiler etal (2000b) The eld of EM-1 lavas is taken fromclinopyroxene data of Gough and Tristan da CunhaIsland lavas (Harris et al 2000) and the EM-1 end-member composition is based on the calculations afterSchleicher et al (1998) for eNd(i) and an average value

of 578 (n = 18) from the EM-1 lavas for d18O

PETROGENESIS OF THE ERIKSFJORD BASALTS

845

Simonetti et al 1998) A plume input to theformation of the EF basalts is supported by therelatively large volume of magmas and the shortduration (lt5 Ma) of magmatism (Piper et al1999) It is also compatible with the geologicalsetting of a continental breakup Additionally aplume component appears to be necessary toexplain the thermal input as it is dif cult to melt acold lithosphere section (Arndt and Christensen1992)

Oxygen isotope data of the EF basalts aregenerally consistent with a mantle derivationincluding an OIB-like mantle source On the d18Ovs eNd(i) diagram (Fig 8) the overlap betweenthe EF data clinopyroxene data from EM-1-likeoceanic basalts and the hypothetical EM-1 end-member suggests an involvement of an EM-1-type mantle source Olivines from HIMU-likesources are known to be slightly lower in d18Othan typical upper mantle values from MORBs(Eiler et al 1997) but the EF data are notsuf cient to assess if the slightly lower d18Ovalues in the Il otildeacutemaussaq Group basalts are causedby a relatively higher input from a HIMU source

The contribution of the various mantle sourcesto the EF basalts can also be evaluated on theprimitive mantle-normalized multi-elementdiagram (Fig 4) and on the REE plots (Figs 5and 6) Both the whole-rock REE and theclinopyroxene REE patterns of the EF basaltsare unlike those representative of N-MORBsources We conclude again that DMM did notplay a major role in the petrogenesis of the EFbasalts On the other hand similarities with OIB-like mantle sources are more prominent Theshape of the clinopyroxene REE pattern closelyresembles those from clinopyroxene of OIB-likealkaline basalts despite differences in absoluteconcentrations The whole-rock REE patterns ofthe EF basalts lie between E-MORB and averageOIB patterns as do the multi-element patterns Onthe latter the EF basalts have marked positive Baand P peaks that are not present in typical OIBsHowever Ba contents should be interpreted withcaution since highly variable ZrBa ratios from014 to 070 are an indication that Ba was re-mobilized after emplacement of the basalts (egPrice et al 1991 Oliveira and Tarney 1995)Likewise P mobility was demonstrated in alteredbasaltic rocks (Winchester and Floyd 1976) Insummary the REE and trace element data areevidence that the mantle source(s) of the EFbasalts were enriched in incompatible traceelements relative to primitive mantle

For relatively primitive hy-normative basalticrocks of the Ivittuut area further north in theGardar Province it has been argued that some ofthem were derived from a lithospheric mantlesource previously enriched by subduction-related processes This interpretation is basedon certain geochemical characteristics such ashigh LaNb ratios and negative Nb anomalies(Goodenough et al 2002) For the EF basaltsnegative Nb anomalies are not clearly discern-ible but LaNb ratios are gt1 which seems to be acharacteristic feature of a lithospheric mantlesource (Fitton 1995) LaBa LaRb and LaKratios were also used to argue for a lithosphericorigin of the Gardar basic magmas although amantle plume involvement was assumed toprovide the required energy (Upton andEmeleus 1987 Upton et al 2003) Sincefeldspar is the only primary magmatic mineralin the EF basalts in which Ba Rb and K areincorporated in signi cant amounts and thefeldspars are generally altered these elementsmight have been easily mobilized and wetherefore do not dare to base the interpretationof the EF basalts sources on these data Inaddition the lack of precise geochemical data onthe composition of the lithosphere below theGardar Province complicates an evaluation ofthe lithospheric contribution to the EF basalticmagmatism

Certain ratios of incompatible trace elements ofthe EF basalts are compared to several occur-rences of ocean island basalts representative ofEM-1 and HIMU sources in Table 5 Ratiosbetween highly incompatible trace elements donot change signi cantly during partial melting orfractional crystallization in basaltic systems andmay therefore constrain the ratios in their sources(Weaver et al 1987) The EF basalts aresubdivided into the more alkaline isotopicallyprimitive Il otildeacutemaussaq Group samples and thosesamples with eNd(i) lt ndash1 The Il otilde maussaq Groupsamples have higher SrNd LaNb and ZrNbratios than HIMU and overlap with the upperrange of EM-1 ratios The EF basalts with eNd(i)lt ndash1 have generally higher ratios than the oceanicisland basalts and partly overlap with typicalvalues of the lower crust It therefore seems likelythat assimilation of lower crustal material mayhave changed the primary trace element ratios inthe basalts Therefore we will evaluate thepossible in uence of assimilation of crustalmaterial on the geochemistry of the EF basaltsin the following section

846

R HALAMA ETAL

The role of crustal contamination

Large amounts of bulk assimilation of crustalmaterial which is relatively SiO2-rich comparedto a basaltic melt would drive the compositions tohigher SiO2 contents and quartz-normativecompositions Partial melts derived from crustalrocks would also be relatively SiO2-rich (Beard etal 1993 Carroll and Wyllie 1990) and wouldhave the same effect The ne-normative characterof the EF basalts (Table 1) suggests thatassimilation of crustal material especially ofsilica-rich upper crustal rocks was not signi cantwith the possible exception of sample EF 059

On the primitive mantle-normalized multi-element diagrams (Fig 4) Ti troughs are typicalfor crustal rocks especially for the Upper Crust(Rudnick and Fountain 1995) and they maytherefore be present in rocks extensively contami-nated by continental crust Ti troughs aregenerally lacking in the patterns of the EFbasalts and a major upper crustal contribution tothe magmas can therefore be excluded However

small amounts of assimilation of lower crustalmaterial cannot be ruled out because the averagelower crust has higher Ti contents

Steep trends on Sr-Nd isotope plots have beenattributed to assimilation of granulite-faciesgneisses (eg Bernstein et al 1998) AFCmodelling based on an equation developed byDePaolo (1981) was performed to test thehypothesis of small amounts of contaminationwith lower crustal material (Fig 9) The basalticmagma which represents one end-member for thecalculations has the isotopic signature of theisotopically most primitive samples from thepresent data set and from the literature(Table 6) Two possible contaminants repre-senting upper and lower crustal compositionswere selected for modelling (Table 6) Therepresentative samples are a granite from theJulianehaEcirc b batholith for the upper crust (sampleJG 02) and an average of ve granulite-faciesgneisses with low RbSr ratios from the Archaeancraton of West Greenland (Taylor et al 1984) for

TABLE 5 Trace element ratios of EF basalts compared to ocean island basalts(OIBs) representing EM-1 and HIMU mantle components and the lowercrust

SrNd LaNb ZrNb

EF basalts with eNd lt ndash 1 249 ndash396 120 ndash150 121 ndash157Il otildeacutemaussaq Group EF basalts 198 ndash292 104 ndash119 51 ndash53EM-1 90 ndash207 067 ndash111 39 ndash109HIMU 45 -163 063 ndash080 27 ndash 36Lower continental crust 30 16 136

Data sources EM-1 Humphris and Thompson (1983 Walvis Ridge) and Cliff et al(1991 Inaccessible Islands) HIMU Chauvel et al (1992 Tubuai Island) Lowercontinental crust Rudnick and Fountain (1995)

TABLE 6 End-member compositions used for AFC modelling

End-member Sr (87Sr86Sr)i Nd eNd(i) Data source(ppm) (T = 12 Ga) (ppm) (T = 12 Ga)

Primitive EF basaltic 507 070278 18 + 22 Sample EF 072magma Paslick et al (1993)Upper crust 300 071665 62 ndash83 Sample JG 02(JulianehaEcirc b Granite)Lower crust 747 070668 63 ndash 234 Taylor et al (1984)(Granulite facies gneiss)

The composition of the granulite-facies gneiss is an average of 5 samples with low RbSr ratios lt02

PETROGENESIS OF THE ERIKSFJORD BASALTS

847

the lower crust An in uence of Archaean rockson Gardar magmas has been demonstrated by Pbisotope studies (Taylor and Upton 1993) andseismic data were also interpreted to re ect theexistence of a wedge of Archaean crust thatextends southwards to Lindenow Fjord (Dahl-Jensen et al 1998) We have shown in theprevious sections on the basis of Ni trends thatfractionation of olivine was most relevant to themajor element evolution of the lavas whereasfractionation of other phases was apparently ofminor importance We therefore modelled theAFC processes with olivine fractionation usingmineral-melt partition coef cients of 0014 for Srand 00066 for Nd (Rollinson 1993)

The modelling of AFC processes is stronglydependent on the parameter r de ned as the ratioof mass assimilation rate and fractional crystal-lization rate (DePaolo 1981) The change inisotopic signature becomes enhanced for highervalues of r Reiners et al (1995) have shown thatr values might be very high and even gt1 in theearly stages of magmatic evolution at hightemperatures since crystallization can besuppressed by assimilation Therefore we calcu-

lated AFC processes with two r values of 05 and08 that represent a relatively high and a middle tolow assimilation rate to demonstrate the effects ofvariable r (Fig 9) The results indicate that AFCprocesses involving upper crust similar to theJulianehaEcirc b Granite cannot explain the trend of theEF basalts because Sr isotopic compositionswould shift to considerably more radiogenicvalues This result is consistent with theconclusions derived from trace element dataThe AFC curves for assimilation of lower crusthowever agree well with the bulk of the data Themost negative eNd values can be explained byAFC processes with intermediate assimilationrates (r = 05) involving assimilation of lt5lower crustal material With a greater degree ofassimilation and an r value of 08 lt4assimilation is indicated

Lower crustal rocks of igneous origin have anaverage d18O value of +75plusmn14 with an overallrange from +54 to +125 (Fowler and Harmon1990) 5 to 10 bulk assimilation of lower crustalmaterial with d18O = +75 into a typical mantle-derived magma with d18O = +57 results ind18O values of the contaminated magma from

-24

-20

-16

-12

-8

-4

0

4

0700 0703 0706 0709 0712 0715 0718

r = 08

r = 05

r = 05

r = 08

EF Basalts

Endmembers of AFC calculations

Upper Crust

Lower Crust

Primitive EF basaltic magma

(87Sr86Sr)i

e Nd

(i)

09

08

06

09

08

06

09

0806

09

08

07

07

04

Typical errorCHUR

BS

E

FIG 9 eNd(i) vs (87Sr86Sr)i diagram showing the results of AFC modelling End-member compositions are listed inTable 6 Olivine is the only fractionating mineral Partition coef cients for olivinemelt are from the compilation ofRollinson (1993) Curves were calculated for two values of r (r = rate of assimilation rate of fractionalcrystallization) Dashed lines and numbers in italics are for r = 05 solid lines and regular numbers are for r = 08

Values at the curves are for F values (fraction of melt remaining) with 10 tick marks

848

R HALAMA ETAL

+579 to +588 Therefore the d18O data cannotbe used to rule out small amounts of assimilationof lower crust To evaluate how much contamina-tion would be permissible from the presentoxygen isotope data we calculated how muchcontamination of a mantle-derived magma with18O = +57 with a contaminant within the lowerrange of lower crustal xenoliths (d18O = +67)would drive the oxygen isotopic composition tothe highest value of the EF basalts (d18O~+63) Although simple mixing shows that60 assimilation is permissible this amountwould drive the whole-rock geochemistry tomore silicic compositions than is observed in theEF basalts In Fig 8 there is no obvious trendfrom the basalts towards the Ketilidian basementsuggesting again that upper crust was notassimilated in signi cant amounts

Ratios of trace elements incompatible inbasaltic systems can be used to distinguishbetween crust and mantle reservoirs (egWeaver 1991) Based on the AFC modellingEF basalts with eNd(i) lt ndash1 are those which aremost likely in uenced by crustal contaminationThe relatively high LaNb and ZrNb ratios ofthese samples are consistent with assimilation oflower crustal material SrNd ratios are evenhigher than the average lower crustal composi-tion but SrNd ratios in those samples which weretaken as representative for the lower crust in theAFC modelling can reach values of 80 ie wellabove the highest EF basalt ratios Again theSrNd ratios support the hypothesis of some lowercrustal contamination

The absence of more primitive basaltic rockswith MgO gt8 wt might also indicate lowercrustal assimilation It suggests that magmas mayhave been retained at depth where they acquiredmore evolved compositions by fractional crystal-lization and had the opportunity to assimilatelower crustal material during crustal underplating(Upton et al 2003)

Summary and conclusions

The new geochemical data of the EriksfjordFormation basalts in the Gardar Provinceprovided in this study are used to placepetrogenetic constraints on the sources of theirparental magmas Based on trace-element patternsand Sr-Nd isotopic constraints there seems to beno indication that a DMM-like source played animportant role in magma genesis of the EFbasalts Clinopyroxene and whole-rock REE

patterns are more compatible with an OIB-likesource but we are not able to conclude whetherthe observed enrichment in REE re ects thesourcersquos signature or magmatic processes Theisotopically most primitive EF basalts have Sr-Ndisotopic compositions that are similar to Gardarcarbonatites Due to the isotopic similaritybetween carbonatites and OIBs many arethought to be related to mantle plumes (Bell2001) We therefore speculate that OIB-likemantle components from a plume source werealso involved in the parental EF magmas but wecannot exclude interaction with the SCLM Twodistinct mantle components one isotopicallydepleted and another one enriched could havecontributed to the EF magmas AFC modelling ofSr-Nd isotope data indicates that assimilation oflt5 lower crust could explain the steep vector onthe Sr-Nd isotope plot but assimilation of uppercrustal material is unlikely Oxygen isotope dataof clinopyroxene separates are consistent with amantle origin but permit some assimilation oflower crustal rocks In summary the data areconsistent with the hypothesis that the EFmagmas could have originated from a mantleplume source which is supported by geologicalarguments and that they underwent someassimilation of lower crustal material

Acknowledgements

We would like to thank Bruce Paterson whoprovided invaluable help during Laser ICP-MSmeasurements at the Large-Scale GeochemicalFacility supported by the European Community-Access to Research Infrastructure action of theImproving Human Potential Programme contractnumber HPRI-CT-1999-00008 awarded to ProfBJ Wood (University of Bristol) Gabi Stoschekand Torsten Vennemann are thanked for their helpwith oxygen isotope measurements Elmar Reitterexpertly assisted with preparation and measure-ments of radiogenic isotopes and MathiasWestphal helped with microprobe measurementsREE whole-rock analyses were kindly providedby the NERC facility at Royal HollowayUniversity of London Michael Marks is thankedfor his pleasant company during eldwork Veryconstructive reviews by Tom Andersen andAlexei S Rukhlov greatly helped to improve themanuscript We also thank Ian Coulson for nalediting Funding of this work by the DeutscheForschungsgemeinschaft (grant Ma-21351-2) isgratefully acknowledged

PETROGENESIS OF THE ERIKSFJORD BASALTS

849

References

Andersen T (1997) Age and petrogenesis of theQassiarsuk carbonatite-alkaline silicate volcaniccomplex in the Gardar rift South GreenlandMineralogical Magazine 61 499 ndash513

Armstrong JT (1991) Quantitative elemental analysisof individual microparticles with electron beaminstruments Pp 261 ndash315 in Electron ProbeQuantitation (KFJ Heinrich and DE Newburyeditors) Plenum Press New York and London

Arndt NT and Christensen U (1992) The role oflithospheric mantle in continental ood volcanismThermal and geochemical constraints Journal ofGeophysical Research 97 10967 ndash 10981

Baker JA Menzies MA Thirlwall MF andMacpherson CG (199 7) Pet rogenes is ofQuaternary Intraplate Vocanism Sanarsquoa YemenImplications for Plume-Lithosphere Interaction andPolybaric Melt Hybridization Journal of Petrology 38 1359 ndash 1390

Baker JA MacPherson CG Menzies MAThirlwall MF Al-Kadasi M and Mattey DP(2000) Resolving crustal and mantle contributions tocontinental ood volcanism Yemen constraintsfrom mineral oxygen isotope data Journal ofPetrology 41 1805 ndash1820

Beard JS Abitz RJ and Lofgren GE (1993)Experimental melting of crustal xenoliths fromKilbourne Hole New Mexico and implications forthe contamination and genesi s of magmasContributions to Mineralogy and Petrology 11588 ndash102

Bell K (2001) Carbonatites relationships to mantle-plume activity Geological Society of AmericaSpecial Paper 352 267 ndash290

Bell K and Blenkinsop J (1989) Neodymium andStrontium Isotope Geochemistry of Carbonatites Pp278 ndash300 in Carbonatites ndash Genesis and Evolution(K Bell editor) Unwin Hyman London

Bell K and Simonetti A (1996) Carbonat iteMagmatism and Plume Activity Implications fromthe Nd Pb and Sr Isotope Systematics of OldoinyoLengai Journal of Petrology 37 1321 ndash1339

Bell K and Tilton GR (2001) Nd Pb and Sr IsotopicCompositions of East African Carbonat itesEvide nce for Mant le Mixing and PlumeInhomogene ity Journa l of Petrology 42 1927 ndash1945

Benoit M Polve M and Ceuleneer G (1996) Traceelement and isotopic characterization of ma ccumulates in a fossil mantle diapir (Oman ophiolite)Chemical Geology 134 199 ndash 214

Bernstein S Kelemen PB Tegner C Kurz MDBlusztajn J and Kent Brooks C (1998) Post-breakup basaltic magmatism along the EastGreenland Tertiary rifted margin Earth and

Planetary Science Letters 160 845 ndash862Boynton WV (1984) Geochemistry of the rare earth

elements meteorite studies Pp 63 ndash114 in RareEarth Element Geochemistry (P Henderson editor)Elsevier Amsterdam

Carroll MR and Wyllie PJ (1990) The systemtonalite-H2O at 15 kbar and the genesis of calc-alkaline magmas American Mineralogist 75345 ndash 357

Chauvel C Hofmann AW and Vidal P (1992)HIMU-EM The French Polynesian connectionEarth and Planetary Science Letters 110 99 ndash119

Clayton RN and Mayeda TK (1963) The use ofbromine penta uoride in the extraction of oxygenfrom oxides and silicates for isotope analysisGeochimica et Cosmochimica Acta 27 43 ndash52

Coulson IM and Chambers AD (1996) Patterns ofzonation in rare-earth-bearing minerals in nephelinesyenites of the North Qoroq center South GreenlandThe Canadian Mineralogist 34 1163 ndash1178

Cox KG Bell JD and Pankhurst RJ (1979) TheInterpretation of Igneous Rocks George Allen andUnwin London

Cross W Iddings JP Pirsson LV and WashingtonHS (1903) Quantitative Classi cation of IgneousRocks University of Chicago Press

Dahl-Jensen T Thybo T Hopper H and Rosing M(1998) Crustal structure at the SE Greenland marginfrom wide-angle and normal incidence seismic dataTectonophysics 288 191 ndash 198

DePaolo DJ (1981) Trace element and isotopic effectsof combined wallrock assimilation and fractionalcrystallisation Earth and Planetary Science Letters53 189 ndash 202

De P a o lo D J (1 9 8 8 ) Neodymium Iso topeGeochemistry An Introduction Springer-VerlagNew York

Dobosi G Downes H Mattey D and Embey-IsztinA (1998) Oxygen isotope ratios of phenocrysts fromalkali basalts of the Pannonian basin Evidence foran O-isotopically homogeneous upper mantle be-neath a subduction-in uenced area Lithos 42213 ndash 223

Eiler JM (2001) Oxygen isotope variations of basalticlavas and upper mantle rocks Pp 319 ndash364 inStable Isotope Geochemistry (JW Valley and DRCole editor s) Reviews in Minera logy andGeochemistry 43 The Mineralogical Society ofAmerica Washington DC

Eiler JM Farley KA Valley JW Hauri E CraigH Hart SR and Stolper EM (1997) Oxygenisotope variations in ocean island basalt phenocrystsGeochimica et Cosmochimica Acta 61 2281 ndash2293

Eiler JM Crawford A Elliot T Farley KAValley JW and Stolper EM (2000a) Oxygenisotope geochemistry of oceanic-arc lavas Journal

850

R HALAMA ETAL

with the mostly subhedral Fe-Ti oxides and theophitic texture Ratios of incompatible traceelements such as ZrY (range 40 ndash63) andTiY (range 385 ndash716) are relatively uniformthroughout all samples but some scatter isevident Notable geochemical features are highSr contents up to 1000 ppm and very high Bacontents up to 1300 ppm

Normative mineral compositions were deter-mined according to the CIPW calculation scheme(Cross et al 1903 Cox et al 1979) Mostsamples are hypersthene-normative some arenepheline-normative and one (EF 059) is quartz-normative (Table 2) Despite the uncertaintiesinvolved in the CIPW norm calculation the dataindicate that some of the basalts crystallized fromsilica-undersaturated magmas and others fromdistinct silica-saturatedor -oversaturatedmagmas

Multi-element diagrams normalized to primi-tive mantle values show that the EF basalts areenriched in all incompatible elements relative toprimitive mantle values (Fig 4) The patterns ofthe EF basalts are characterized by a prominentBa peak a subordinate P peak and a decreasingelement enrichment with increasing compatibilityIn comparison with normal MORB (N-MORB)enriched MORB (E-MORB) and average OIB(Sun and McDonough 1989) the patterns of theEF basalts are closest to those of OIB both inshape and normalized concentrations

Whole-rock REE data

Whole-rock REE data are presented in Table 3and REE concentrations normalized to chondriticvalues of Boynton (1984) are shown in Fig 5aAll samples are characterized by LREE-enrichedpatterns with no or slightly positive Euanomalies (EuEu = 097 to 113) La concentra-tions are roughly 20 ndash60 times and Lu concentra-tions are 10 times chondritic values NormalizedLREE abundances from La to Nd are fairlyconstant but the patterns show a steady andsmooth decline from Nd towards heavier REEThe REE patterns of the EF basalts fall in betweenthose of typical OIB and E-MORB (Fig 5b)They have similar patterns and absolute abun-dances compared to basaltic dykes from theIvittuut area South Greenland (Goodenough etal 2002) LaNSmN as a measure of LREEfractionation has a restricted range from 130 to165 The degree of HREE fractionationexpressed as GdNYbN is slightly more variableand ranges from 157 to 257 Both these ratios of

the EF basalts fall generally between typical OIBand E-MORB compositions

Clinopyroxene REE patterns

Typical analyses of REE concentrations inclinopyroxenes are given in Table 3 They showthat REE concentrations in individual minerals ofone particular sample can be quite variableChondrite-normalized REE patterns of clinopyr-oxenes are very similar in the samples from theMussartut and Ulukasik Groups (Fig 6ab) Theyare enriched in all REE relative to chondriticvalues and the patterns show a fairly steepincrease from La to Sm followed by a gradualsmooth decrease from Sm to Lu Clinopyroxenesfrom sample EF 168 (Il otildeacutemaussaq Group) aredistinct as they have generally higher REEcontents especially in the LREE and a weak Euanomaly (Fig 6c) The 3 ndash106 enrichment in Nd

c (sa

mpl

e)c

(prim

itive

man

tle)

c (sa

mpl

e)c

(prim

itive

man

tle)

1

10

100

1000

Mussartucirct GroupUlukasik GroupIliacutemaussaq Group

Rb Ba K Nb La Ce Sr Nd P Zr Ti Y

Rb Ba K Nb La Ce Sr Nd P Zr Ti Y01

1

10

100

1000N-MORBE-MORBOIB

FIG 4 Multi-element plots normalized to primitivemantle values (McDonough and Sun 1995) Data forN-MORB E-MORB and average OIB (Sun and

McDonough 1989) are shown for comparison

PETROGENESIS OF THE ERIKSFJORD BASALTS

839

TA

BL

E3

RE

Eda

taof

who

lero

cks

and

clin

opyr

oxen

es

the

latt

erde

term

ined

byL

aser

ICP

-MS

Who

le-r

ock

anal

yses

R

epre

sent

ativ

ecl

inop

yrox

ene

anal

yses

S

ampl

eno

10

1231

1014

4610

1457

1014

7218

6389

1863

95E

F02

4-2

EF

039-

1E

F03

9-2

EF

039-

4E

F07

2-2

EF

072-

3E

F16

8-3

EF

168-

5

RE

Eco

ncen

trat

ions

(ppm

)L

a11

65

180

610

29

661

155

412

28

440

266

199

150

273

304

165

527

59

Ce

290

346

05

249

617

13

378

130

08

180

411

46

868

717

105

012

60

572

399

67

Pr

434

67

388

268

527

433

331

216

181

128

249

262

102

917

43

Nd

200

631

71

176

412

49

250

920

16

190

213

82

122

28

9815

64

157

258

51

960

6S

m4

756

974

043

215

914

826

664

194

283

276

077

0615

82

265

9E

u1

742

51

571

071

971

721

831

491

111

212

001

943

877

07G

d4

826

584

043

555

84

996

965

615

244

326

156

0514

75

252

7T

b ndash

ndash ndash

ndash ndash

ndash1

351

010

770

711

111

212

313

61D

y4

475

223

613

565

354

816

865

934

784

255

996

8613

10

205

2H

o0

921

040

730

731

11

011

481

401

001

001

211

292

394

24E

r2

432

531

951

972

872

663

893

342

642

112

993

646

539

60T

m ndash

ndash ndash

ndash ndash

ndash0

450

390

360

290

350

440

861

27Y

b2

182

071

741

832

592

363

782

812

492

452

703

016

788

89L

u0

340

310

260

270

390

350

460

410

300

310

520

411

251

60

(Eu

Eu

) N1

111

131

190

971

031

070

820

940

720

991

000

910

770

83

Eu

anom

aly

isca

lcul

ated

as(E

uE

u) N

=E

u N(

SmN6

Gd N

)05

norm

aliz

edto

chon

drit

eva

lues

from

Boy

nton

(198

4)

840

R HALAMA ETAL

in clinopyroxenes of EF 168 compared to theother samples (EF 024 039 and 072) seems onlypartly due to a higher degree of fractionationsincethe Mg value of EF 168 is only slightly lowerThus the LREE enrichment appears to underlinethe more alkaline character of this sample Acomparison with published REE patterns ofclinopyroxenes (Fig 6d) shows that the EFpatterns are unlike patterns from olivine gabbroswith MORB-like signatures (Benoit et al 1996)but fairly similar to clinopyroxene megacrystpatterns from an OIB-like alkaline basalt (Nimisand Vannucci 1995)

Sr Nd and O isotopic composition

Sr Nd and O isotopic compositions of all samplesare presented in Table 4 Sr and Nd isotopiccompositions of the clinopyroxene separates havebeen recalculated to an approximate eruption age

of 12 Ga (Paslick et al 1993) using decayconstants of 142610ndash11 andash1 for 87Rb (Steigerand Jager 1977) and 654610ndash12 a ndash1 for 147Sm(Lugmair and Marti 1978) Present day CHURvalues of 01967 for 147Sm144Nd (Jacobsen andWasserburg 1980) and 0512638 for 143Nd144Nd(Goldstein et al 1984) were used to calculateinitial eNd values 87Sr86Sr initial ratios are fairlyconstant and range from 070278 to 070383(Table 4) eNd(i) values vary from ndash32 to +21(Table 4) On the Sr-Nd isotope diagram the EFbasalts de ne a relatively steep trend (Fig 7) withinitial 87Sr86Sr ratios similar to those of the BulkSilicate Earth (BSE) Only one sample is slightlydisplaced towards a more radiogenic initial87Sr86Sr ratio None of the samples has an eNd

value close to depleted mantle values estimatedto be between eNd = +53 (calculated afterDePaolo 1981) and eNd = +73 (calculated afterGoldstein et al 1984) at 12 Ga The two alkalinebasalts of the Il otilde maussaq Group are within therange of the other data but they de ne a restricted eld around CHUR and BSE respectively Errorsinduced by the uncertainty in the age of the EFbasalts are largely within the error of the isotopeanalyses and do not change the overall picture Incomparison with other published Sr-Nd isotopedata from the Gardar Province the initial Srisotope ratios of the EF basalts are very similar torelatively primitive basaltic lamprophyric andcarbonatitic dyke rocks which have (87Sr86Sr)i of~0703 (Pearce and Leng 1996 Andersen 1997Goodenough et al 2002) So far no evidence hasbeen presented for an isotopicallyenriched mantlesource in the Gardar Province as the existingeNd(i) data of ndash05 to +52 show an isotopicallydepleted signature (Pearce and Leng 1996Andersen 1997 Goodenough et al 2002Upton et al 2003) Clinopyroxene from the EFsamples of Paslick et al (1993) yield eNd(i) of+22 The Nd isotope data of this study are withinthe lower range of published values from theGardar Province and extend the spread to morenegative eNd(i) values

Several recent oxygen isotope studies haveshown that mineral separates are less sensitive tolow-T alteration and secondary water take-up thanwhole-rock samples (eg Vroon et al 2001)However only few oxygen isotope data ofmineral separates from continental volcanics areavailable (eg Dobosi et al 1998 Baker et al2000) Extensive studies of mineral separates ofocean island basalts (Eiler et al 1997) andoceanic-arc lavas (Eiler et al 2000a) demon-

c (sa

mpl

e)c

(cho

ndri

te)

c (sa

mp

le)c

(cho

ndrit

e)

1

10

100

200

101446

101472

101231

101457

186389

186395

La Ce Pr Nd Sm Eu Gd Dy Ho Er Yb Lu

1

10

100

200

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

N-MORBE-MORBOIBBasaltic dykes from Ivittuut

a

b

FIG 5 (a) Whole-rock REE data of EF basaltsnormalized to chondritic values from Boynton (1984)(b) Typical compositions of N-MORB E-MORB andOIB (Sun and McDonough 1989) are shown for

comparison

PETROGENESIS OF THE ERIKSFJORD BASALTS

841

strated that analyses of mineral separates consis-tently span a narrower range than comparablewhole-rock data In our study we used clino-pyroxene because it is the only fresh mineralphase present in the samples Oxygen isotopemeasurements of clinopyroxene separates fromthe EF basalts show a range in d18Ocpx from +52to +60 with an average value of +566plusmn023(1s) All samples overlap the range of d18Ocpx

values from mantle peridotites (+525 to +590Mattey et al 1994) from continental oodbasalts from Yemen (+55 to +69 Baker etal 2000) and from alkaline basalts of Hungary(+507 to +534 Dobosi et al 1998) There is asmall difference between the averages ofMussartut Group (d18Ocpx-avg = +575plusmn015(1s)) and Ulukasik Group (d18Ocpx-avg =+575plusmn009 (1s)) samples compared toIl otildeacutemaussaq Group (d18Ocpx-avg = +533plusmn021(1s)) samples

d18O values of melts coexisting with clino-pyroxene were calculated assuming that oxygen

in pyroxene and melt were in isotopic equilibriumand T was 1100ordmC Dmelt-cpx can then be calculatedusing the following equation of Kalamarides(1986)

Dmelt-cpx = ndash00008T + 143 (T in Kelvin)

Although fractionation varies with magmatictemperatures we consider the temperature of1100ordmC and the resulting Dmelt-cpx of +03 assuf ciently representative considering the analy-tical errors involved It is similar to Dmelt-cpx

values used in previous studies (+039 Vroonet al 2001 +02 Dobosi et al 1998)Resulting d18Omelt values of EF basalts rangefrom +55 to +63 These values can becompared to whole-rock data from continentalbasalts given by Harmon and Hoefs (1995)Intraplate basalts from continental rift zonesshow a range from +46 to +83 with a meanof +61 and continental ood basalts which arerather poorly represented in the database havevalues between +43 and +65 Calculated EF

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Cpx from MORB-like

olivine gabbro

Cpx megacrysts from OIB-like

alkaline basalts

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 024EF 039

Mussartucirct Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 168

Iliacutemaussaq Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

1

10

100

1000

EF 072

Ulukasik Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb LuLa Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

c

ba

d

1

10

100

01

FIG 6 REE patterns of clinopyroxenes (cpx) from Mussartut (a) Ulukasik (b) and Il otilde maussaq (c) Group samplesnormalized to chondritic values from Boynton (1984) (d) REE patterns of cpx from MORB-type olivine gabbro(Benoit et al 1996) and OIB-type alkaline basalt (Nimis and Vannucci 1995) for comparison (note the change of

scale)

842

R HALAMA ETAL

TA

BL

E4

Sr

Nd

and

Ois

otop

icco

mpo

siti

ons

ofcl

inop

yrox

ene

sepa

rate

sfr

omE

riks

fjor

dF

orm

atio

nba

salt

sW

hole

-roc

kan

alys

esfr

omth

eK

etil

idia

nba

sem

ent

and

from

sedi

men

tary

rock

sof

the

Eri

ksfj

ord

For

mat

ion

are

give

nfo

rco

mpa

riso

n

Sam

ple

Mem

ber

Sr

Rb

87R

b86S

r87S

r86Sr

87Sr

86Sr

init

ial

SmN

d147Sm

144N

d143N

d144N

de N

d(i

)d1

8O

(ppm

)(p

pm)

(ppm

)(p

pm)

Bas

alts

E

F02

4M

ussa

rtut

646

50

486

002

180

7042

01plusmn

100

7038

267

3822

80

019

560

5124

67plusmn

10ndash

32

567

EF

063

Mus

sart

ut5

60plusmn

011

EF

061

Mus

sart

ut89

95

099

90

0321

070

3818

plusmn10

070

3266

574

163

90

2115

051

2629

plusmn10

ndash2

55

79plusmn

028

EF

039

Mus

sart

ut73

74

058

30

0229

070

3492

plusmn10

070

3098

662

191

30

2092

051

2626

plusmn10

ndash2

25

71plusmn

001

EF

059

Mus

sart

ut54

47

013

70

0073

070

3375

plusmn12

070

3250

606

168

90

2171

051

2727

plusmn09

ndash1

45

99plusmn

001

EF

108

Ulu

kasi

k66

92

111

00

0480

070

3741

plusmn09

070

2916

109

936

82

018

050

5126

17plusmn

10+

21

581

EF

072

Ulu

kasi

k69

41

032

20

0134

070

3014

plusmn10

070

2784

818

251

90

1962

051

2574

plusmn10

ndash1

25

68E

F17

4Il

otildemau

ssaq

125

42

324

005

360

7039

35plusmn

100

7030

1418

76

796

10

1424

051

2230

plusmn10

+0

45

47E

F16

8Il

otildemau

ssaq

866

71

422

004

750

7037

99plusmn

070

7029

8316

45

641

60

1550

051

2304

plusmn09

ndash0

15

18

Bas

emen

tan

dse

dim

enta

ryro

cks

JG02

Juli

aneh

aEcirc b30

01

169

71

6375

074

4789

plusmn10

071

6647

105

362

28

010

220

5114

70plusmn

09ndash

83

78

Gra

nite

EF

021

Qua

rtzi

te11

5E

F02

8A

rkos

ic6

8ar

enit

e

Sam

ples

are

list

edin

stra

tigra

phic

orde

rfr

omol

dest

toyo

unge

st87

Sr86

Sr

init

ialr

atio

san

de N

d(i)

valu

esw

ere

calc

ulat

edfo

ran

age

of1

2G

aS

tand

ard

devi

atio

nsfo

rd18

Oar

egi

ven

for

sam

ples

that

wer

ean

alys

edtw

ice

PETROGENESIS OF THE ERIKSFJORD BASALTS

843

model melt oxygen isotope data are within theranges of both groups Compared to Igaliko dykerocks from further SE in the Gardar Province(Pearce and Leng 1996) the EF basalts cover amuch more restricted eld possibly because d18Owhole-rock analyses are more prone to secondaryalteration than analyses of mineral separates(Fig 8) Whole-rock d18O measurements of theunderlying JulianehaEcirc b Granite and of the sedi-mentary sandstone members of the EriksfjordFormation yield values of 78 and 68 ndash115respectively (Table 4) and are shown forcomparison

Discussion

Mantle source regions of the EF basaltsTo reconstruct the mantle sources involved in theformation of the EF basalts we use Sr-Nd-Oisotope and incompatible trace element data On aSr-Nd isotope diagram the isotopic compositionof DMM was modelled for an age of 12 Ga(DePaolo 1981 Goldstein et al 1984 Taylor andMcLennan 1985) The fairly constant 87Sr86Srinitial isotope ratios of the EF basalts are ~0703which falls within the range typical for mantle-derived rocks but slightly above calculated valuesfor DMM (Fig 7) eNd(i) values are more

-4

-2

0

2

4

6

8

0701 0702 0703 0704 0705 0706

DMM Typical 2 s error

CHUR

BS

E

(87Sr86Sr)i

e Nd

(i)

EM-1

Qassiarsuk carbonatites and lamprophyres

EF basalts

Depleted MORB mantle

Ivittuut basaltic and lamprophyric dikes

Igaliko carbonatites and lamprophyres

Enriched mantle 1

Carbonatites from other regions of similar age

FIG 7 eNd(i) vs (87Sr86Sr)i diagram calculated for t = 12 Ga which is considered as the eruption age of the EFbasalts (Paslick et al 1993) The reference line of Bulk Silicate Earth (BSE) for the Sr isotopic composition wascalculated after DePaolo (1988) assuming present-day values of 00827 for 87Rb86Sr and 07045 for 87Sr86SrIsotopic composition of the depleted MORB mantle (DMM) was calculated after DePaolo (1981) Goldstein et al(1984) and Taylor and McLennan (1985) EM-1 was calculated after Schleicher et al (1998) (see text for details)Data sources are Goodenough et al (2002) Ivittuut (n = 6) Pearce and Leng (1996) Igaliko (n = 5) Andersen(1997) Qassiarsuk (n = 25) Bell and Blenkinsop (1989) other carbonatites of similar age (Goudini Schryburt Lake

and Prairie Lake n = 6)

844

R HALAMA ETAL

scattered and indicate that the rocks contain amixture of isotopic components from both mantleand crustal reservoirs or a heterogeneous mantlesource Only one sample (EF 108) has an eNd(i)value signi cantly above 0 This sample exhibitsthe largest spread of clinopyroxene compositionswhich may re ect strong alteration Goodenoughet al (2002) also note that the highest eNd occursin the most altered sample of their sample setHowever Paslick et al (1993) report clino-pyroxene data from two Ulukasik Groupsamples with eNd(i) gt +2 and those sampleswere apparently not disturbed with regard to theirSm-Nd age information On the other hand Piperet al (1999) suggest that the Sm-Nd systematicsof those samples may be erroneous due to laterdisturbance by uids derived from the Il otildeacutemaussaqintrusion Therefore although we are presentlynot able to decide whether positive eNd(i) values

for the EF basalts are real it is clear that none ofthose values is close to values of a 12 Ga oldDMM source (ie eNd = +5 to +7)

The Sr-Nd isotopic composition for OIB-typesources in the Proterozoic is not as wellconstrained as DMM Today OIB-type mantlesources have a wide range of isotopic composi-tions that partly overlap with DMM but eNd

values are generally lower (Hart et al 1992Hofmann 1997) The EM-1 mantle component ischaracterized by extremely low 143Nd144Ndratios and eNd values of between 0 and ndash5(Hofmann 1997) A possible 12 Ga position forEM-1 is plotted in Fig 7 calculated according toSchleicher et al (1998) with 147Sm144Nd = 018and 87Rb86Sr = 010 The present-day values usedin the calculation are 143Nd144Nd = 051234 and87Sr86Sr = 07057 taken from ocean islandbasalts closest to an EM-1 end-member(Hofmann 1997) Other Sr-Nd isotope datafrom the Gardar Province are plotted to putfurther constraints on the mantle sources of the EFbasalts (Fig 7) We also added comparative datafrom African and Canadian carbonatitesof similarage because there is a well known similaritybetween Sr-Nd isotopic data of OIBs andcarbonatites (Bell and Blenkinsop 1989) Thelinear array of the carbonatite data is consideredto be the result of mixing between the two mantlecomponents HIMU and EM-1 (Bell andBlenkinsop 1989 Bell and Tilton 2001) Thecarbonatites from the Gardar Province havesimilar Nd but slightly more radiogenic Srisotopic compositions than the reference carbo-natite data (Fig 7) The least contaminatedsamples overlap with the linear array and themore radiogenic Sr isotopic composition can beexplained by crustal contamination (Andersen1997) We therefore consider it possible that theGardar carbonatites also represent mixing ofhypothetical HIMU and EM-1 components of anOIB-like source Since the isotopically mostprimitive EF basalt samples overlap with theQassiarsuk carbonatites it could be possible thatthe basalts also contain such mantle componentsalthough Pb isotope data for the EF basalts thatcould clarify the importance of a HIMU sourceare lacking Based on the Sr-Nd isotope data theEF basalts appear to be derived from aheterogeneous OIB-like mantle source thatcomprises isotopically depleted and enrichedcomponents OIB-like and carbonatitic magma-tism are often attributed to mantle plume activity(Bell and Simonetti 1996 Hofmann 1997

2

4

6

8

10

12

-10 -8 -6 -4 -2 0 2 4 6 8

DMM

d18 O

[permil]

e Nd (i)

Typical2s error

EM-1

EF basalts - cpx

EF basalts - model melt

Ketilidian basement

Igaliko dykes

Depleted MORB mantle

Enriched mantle 1

EM-1 lavas - cpx

FIG 8 Oxygen isotopi c compositio ns (d18OV-SMOW) of clinopyroxene from EF basalts calculatedmelt (see text for details) and whole-rock Ketilidianbasement vs eNd(i) Whole-rock data for Igaliko dykerocks (n = 7) are from Pearce and Leng (1996) DMMdata derived from fresh MORB glasses are from Eiler etal (2000b) The eld of EM-1 lavas is taken fromclinopyroxene data of Gough and Tristan da CunhaIsland lavas (Harris et al 2000) and the EM-1 end-member composition is based on the calculations afterSchleicher et al (1998) for eNd(i) and an average value

of 578 (n = 18) from the EM-1 lavas for d18O

PETROGENESIS OF THE ERIKSFJORD BASALTS

845

Simonetti et al 1998) A plume input to theformation of the EF basalts is supported by therelatively large volume of magmas and the shortduration (lt5 Ma) of magmatism (Piper et al1999) It is also compatible with the geologicalsetting of a continental breakup Additionally aplume component appears to be necessary toexplain the thermal input as it is dif cult to melt acold lithosphere section (Arndt and Christensen1992)

Oxygen isotope data of the EF basalts aregenerally consistent with a mantle derivationincluding an OIB-like mantle source On the d18Ovs eNd(i) diagram (Fig 8) the overlap betweenthe EF data clinopyroxene data from EM-1-likeoceanic basalts and the hypothetical EM-1 end-member suggests an involvement of an EM-1-type mantle source Olivines from HIMU-likesources are known to be slightly lower in d18Othan typical upper mantle values from MORBs(Eiler et al 1997) but the EF data are notsuf cient to assess if the slightly lower d18Ovalues in the Il otildeacutemaussaq Group basalts are causedby a relatively higher input from a HIMU source

The contribution of the various mantle sourcesto the EF basalts can also be evaluated on theprimitive mantle-normalized multi-elementdiagram (Fig 4) and on the REE plots (Figs 5and 6) Both the whole-rock REE and theclinopyroxene REE patterns of the EF basaltsare unlike those representative of N-MORBsources We conclude again that DMM did notplay a major role in the petrogenesis of the EFbasalts On the other hand similarities with OIB-like mantle sources are more prominent Theshape of the clinopyroxene REE pattern closelyresembles those from clinopyroxene of OIB-likealkaline basalts despite differences in absoluteconcentrations The whole-rock REE patterns ofthe EF basalts lie between E-MORB and averageOIB patterns as do the multi-element patterns Onthe latter the EF basalts have marked positive Baand P peaks that are not present in typical OIBsHowever Ba contents should be interpreted withcaution since highly variable ZrBa ratios from014 to 070 are an indication that Ba was re-mobilized after emplacement of the basalts (egPrice et al 1991 Oliveira and Tarney 1995)Likewise P mobility was demonstrated in alteredbasaltic rocks (Winchester and Floyd 1976) Insummary the REE and trace element data areevidence that the mantle source(s) of the EFbasalts were enriched in incompatible traceelements relative to primitive mantle

For relatively primitive hy-normative basalticrocks of the Ivittuut area further north in theGardar Province it has been argued that some ofthem were derived from a lithospheric mantlesource previously enriched by subduction-related processes This interpretation is basedon certain geochemical characteristics such ashigh LaNb ratios and negative Nb anomalies(Goodenough et al 2002) For the EF basaltsnegative Nb anomalies are not clearly discern-ible but LaNb ratios are gt1 which seems to be acharacteristic feature of a lithospheric mantlesource (Fitton 1995) LaBa LaRb and LaKratios were also used to argue for a lithosphericorigin of the Gardar basic magmas although amantle plume involvement was assumed toprovide the required energy (Upton andEmeleus 1987 Upton et al 2003) Sincefeldspar is the only primary magmatic mineralin the EF basalts in which Ba Rb and K areincorporated in signi cant amounts and thefeldspars are generally altered these elementsmight have been easily mobilized and wetherefore do not dare to base the interpretationof the EF basalts sources on these data Inaddition the lack of precise geochemical data onthe composition of the lithosphere below theGardar Province complicates an evaluation ofthe lithospheric contribution to the EF basalticmagmatism

Certain ratios of incompatible trace elements ofthe EF basalts are compared to several occur-rences of ocean island basalts representative ofEM-1 and HIMU sources in Table 5 Ratiosbetween highly incompatible trace elements donot change signi cantly during partial melting orfractional crystallization in basaltic systems andmay therefore constrain the ratios in their sources(Weaver et al 1987) The EF basalts aresubdivided into the more alkaline isotopicallyprimitive Il otildeacutemaussaq Group samples and thosesamples with eNd(i) lt ndash1 The Il otilde maussaq Groupsamples have higher SrNd LaNb and ZrNbratios than HIMU and overlap with the upperrange of EM-1 ratios The EF basalts with eNd(i)lt ndash1 have generally higher ratios than the oceanicisland basalts and partly overlap with typicalvalues of the lower crust It therefore seems likelythat assimilation of lower crustal material mayhave changed the primary trace element ratios inthe basalts Therefore we will evaluate thepossible in uence of assimilation of crustalmaterial on the geochemistry of the EF basaltsin the following section

846

R HALAMA ETAL

The role of crustal contamination

Large amounts of bulk assimilation of crustalmaterial which is relatively SiO2-rich comparedto a basaltic melt would drive the compositions tohigher SiO2 contents and quartz-normativecompositions Partial melts derived from crustalrocks would also be relatively SiO2-rich (Beard etal 1993 Carroll and Wyllie 1990) and wouldhave the same effect The ne-normative characterof the EF basalts (Table 1) suggests thatassimilation of crustal material especially ofsilica-rich upper crustal rocks was not signi cantwith the possible exception of sample EF 059

On the primitive mantle-normalized multi-element diagrams (Fig 4) Ti troughs are typicalfor crustal rocks especially for the Upper Crust(Rudnick and Fountain 1995) and they maytherefore be present in rocks extensively contami-nated by continental crust Ti troughs aregenerally lacking in the patterns of the EFbasalts and a major upper crustal contribution tothe magmas can therefore be excluded However

small amounts of assimilation of lower crustalmaterial cannot be ruled out because the averagelower crust has higher Ti contents

Steep trends on Sr-Nd isotope plots have beenattributed to assimilation of granulite-faciesgneisses (eg Bernstein et al 1998) AFCmodelling based on an equation developed byDePaolo (1981) was performed to test thehypothesis of small amounts of contaminationwith lower crustal material (Fig 9) The basalticmagma which represents one end-member for thecalculations has the isotopic signature of theisotopically most primitive samples from thepresent data set and from the literature(Table 6) Two possible contaminants repre-senting upper and lower crustal compositionswere selected for modelling (Table 6) Therepresentative samples are a granite from theJulianehaEcirc b batholith for the upper crust (sampleJG 02) and an average of ve granulite-faciesgneisses with low RbSr ratios from the Archaeancraton of West Greenland (Taylor et al 1984) for

TABLE 5 Trace element ratios of EF basalts compared to ocean island basalts(OIBs) representing EM-1 and HIMU mantle components and the lowercrust

SrNd LaNb ZrNb

EF basalts with eNd lt ndash 1 249 ndash396 120 ndash150 121 ndash157Il otildeacutemaussaq Group EF basalts 198 ndash292 104 ndash119 51 ndash53EM-1 90 ndash207 067 ndash111 39 ndash109HIMU 45 -163 063 ndash080 27 ndash 36Lower continental crust 30 16 136

Data sources EM-1 Humphris and Thompson (1983 Walvis Ridge) and Cliff et al(1991 Inaccessible Islands) HIMU Chauvel et al (1992 Tubuai Island) Lowercontinental crust Rudnick and Fountain (1995)

TABLE 6 End-member compositions used for AFC modelling

End-member Sr (87Sr86Sr)i Nd eNd(i) Data source(ppm) (T = 12 Ga) (ppm) (T = 12 Ga)

Primitive EF basaltic 507 070278 18 + 22 Sample EF 072magma Paslick et al (1993)Upper crust 300 071665 62 ndash83 Sample JG 02(JulianehaEcirc b Granite)Lower crust 747 070668 63 ndash 234 Taylor et al (1984)(Granulite facies gneiss)

The composition of the granulite-facies gneiss is an average of 5 samples with low RbSr ratios lt02

PETROGENESIS OF THE ERIKSFJORD BASALTS

847

the lower crust An in uence of Archaean rockson Gardar magmas has been demonstrated by Pbisotope studies (Taylor and Upton 1993) andseismic data were also interpreted to re ect theexistence of a wedge of Archaean crust thatextends southwards to Lindenow Fjord (Dahl-Jensen et al 1998) We have shown in theprevious sections on the basis of Ni trends thatfractionation of olivine was most relevant to themajor element evolution of the lavas whereasfractionation of other phases was apparently ofminor importance We therefore modelled theAFC processes with olivine fractionation usingmineral-melt partition coef cients of 0014 for Srand 00066 for Nd (Rollinson 1993)

The modelling of AFC processes is stronglydependent on the parameter r de ned as the ratioof mass assimilation rate and fractional crystal-lization rate (DePaolo 1981) The change inisotopic signature becomes enhanced for highervalues of r Reiners et al (1995) have shown thatr values might be very high and even gt1 in theearly stages of magmatic evolution at hightemperatures since crystallization can besuppressed by assimilation Therefore we calcu-

lated AFC processes with two r values of 05 and08 that represent a relatively high and a middle tolow assimilation rate to demonstrate the effects ofvariable r (Fig 9) The results indicate that AFCprocesses involving upper crust similar to theJulianehaEcirc b Granite cannot explain the trend of theEF basalts because Sr isotopic compositionswould shift to considerably more radiogenicvalues This result is consistent with theconclusions derived from trace element dataThe AFC curves for assimilation of lower crusthowever agree well with the bulk of the data Themost negative eNd values can be explained byAFC processes with intermediate assimilationrates (r = 05) involving assimilation of lt5lower crustal material With a greater degree ofassimilation and an r value of 08 lt4assimilation is indicated

Lower crustal rocks of igneous origin have anaverage d18O value of +75plusmn14 with an overallrange from +54 to +125 (Fowler and Harmon1990) 5 to 10 bulk assimilation of lower crustalmaterial with d18O = +75 into a typical mantle-derived magma with d18O = +57 results ind18O values of the contaminated magma from

-24

-20

-16

-12

-8

-4

0

4

0700 0703 0706 0709 0712 0715 0718

r = 08

r = 05

r = 05

r = 08

EF Basalts

Endmembers of AFC calculations

Upper Crust

Lower Crust

Primitive EF basaltic magma

(87Sr86Sr)i

e Nd

(i)

09

08

06

09

08

06

09

0806

09

08

07

07

04

Typical errorCHUR

BS

E

FIG 9 eNd(i) vs (87Sr86Sr)i diagram showing the results of AFC modelling End-member compositions are listed inTable 6 Olivine is the only fractionating mineral Partition coef cients for olivinemelt are from the compilation ofRollinson (1993) Curves were calculated for two values of r (r = rate of assimilation rate of fractionalcrystallization) Dashed lines and numbers in italics are for r = 05 solid lines and regular numbers are for r = 08

Values at the curves are for F values (fraction of melt remaining) with 10 tick marks

848

R HALAMA ETAL

+579 to +588 Therefore the d18O data cannotbe used to rule out small amounts of assimilationof lower crust To evaluate how much contamina-tion would be permissible from the presentoxygen isotope data we calculated how muchcontamination of a mantle-derived magma with18O = +57 with a contaminant within the lowerrange of lower crustal xenoliths (d18O = +67)would drive the oxygen isotopic composition tothe highest value of the EF basalts (d18O~+63) Although simple mixing shows that60 assimilation is permissible this amountwould drive the whole-rock geochemistry tomore silicic compositions than is observed in theEF basalts In Fig 8 there is no obvious trendfrom the basalts towards the Ketilidian basementsuggesting again that upper crust was notassimilated in signi cant amounts

Ratios of trace elements incompatible inbasaltic systems can be used to distinguishbetween crust and mantle reservoirs (egWeaver 1991) Based on the AFC modellingEF basalts with eNd(i) lt ndash1 are those which aremost likely in uenced by crustal contaminationThe relatively high LaNb and ZrNb ratios ofthese samples are consistent with assimilation oflower crustal material SrNd ratios are evenhigher than the average lower crustal composi-tion but SrNd ratios in those samples which weretaken as representative for the lower crust in theAFC modelling can reach values of 80 ie wellabove the highest EF basalt ratios Again theSrNd ratios support the hypothesis of some lowercrustal contamination

The absence of more primitive basaltic rockswith MgO gt8 wt might also indicate lowercrustal assimilation It suggests that magmas mayhave been retained at depth where they acquiredmore evolved compositions by fractional crystal-lization and had the opportunity to assimilatelower crustal material during crustal underplating(Upton et al 2003)

Summary and conclusions

The new geochemical data of the EriksfjordFormation basalts in the Gardar Provinceprovided in this study are used to placepetrogenetic constraints on the sources of theirparental magmas Based on trace-element patternsand Sr-Nd isotopic constraints there seems to beno indication that a DMM-like source played animportant role in magma genesis of the EFbasalts Clinopyroxene and whole-rock REE

patterns are more compatible with an OIB-likesource but we are not able to conclude whetherthe observed enrichment in REE re ects thesourcersquos signature or magmatic processes Theisotopically most primitive EF basalts have Sr-Ndisotopic compositions that are similar to Gardarcarbonatites Due to the isotopic similaritybetween carbonatites and OIBs many arethought to be related to mantle plumes (Bell2001) We therefore speculate that OIB-likemantle components from a plume source werealso involved in the parental EF magmas but wecannot exclude interaction with the SCLM Twodistinct mantle components one isotopicallydepleted and another one enriched could havecontributed to the EF magmas AFC modelling ofSr-Nd isotope data indicates that assimilation oflt5 lower crust could explain the steep vector onthe Sr-Nd isotope plot but assimilation of uppercrustal material is unlikely Oxygen isotope dataof clinopyroxene separates are consistent with amantle origin but permit some assimilation oflower crustal rocks In summary the data areconsistent with the hypothesis that the EFmagmas could have originated from a mantleplume source which is supported by geologicalarguments and that they underwent someassimilation of lower crustal material

Acknowledgements

We would like to thank Bruce Paterson whoprovided invaluable help during Laser ICP-MSmeasurements at the Large-Scale GeochemicalFacility supported by the European Community-Access to Research Infrastructure action of theImproving Human Potential Programme contractnumber HPRI-CT-1999-00008 awarded to ProfBJ Wood (University of Bristol) Gabi Stoschekand Torsten Vennemann are thanked for their helpwith oxygen isotope measurements Elmar Reitterexpertly assisted with preparation and measure-ments of radiogenic isotopes and MathiasWestphal helped with microprobe measurementsREE whole-rock analyses were kindly providedby the NERC facility at Royal HollowayUniversity of London Michael Marks is thankedfor his pleasant company during eldwork Veryconstructive reviews by Tom Andersen andAlexei S Rukhlov greatly helped to improve themanuscript We also thank Ian Coulson for nalediting Funding of this work by the DeutscheForschungsgemeinschaft (grant Ma-21351-2) isgratefully acknowledged

PETROGENESIS OF THE ERIKSFJORD BASALTS

849

References

Andersen T (1997) Age and petrogenesis of theQassiarsuk carbonatite-alkaline silicate volcaniccomplex in the Gardar rift South GreenlandMineralogical Magazine 61 499 ndash513

Armstrong JT (1991) Quantitative elemental analysisof individual microparticles with electron beaminstruments Pp 261 ndash315 in Electron ProbeQuantitation (KFJ Heinrich and DE Newburyeditors) Plenum Press New York and London

Arndt NT and Christensen U (1992) The role oflithospheric mantle in continental ood volcanismThermal and geochemical constraints Journal ofGeophysical Research 97 10967 ndash 10981

Baker JA Menzies MA Thirlwall MF andMacpherson CG (199 7) Pet rogenes is ofQuaternary Intraplate Vocanism Sanarsquoa YemenImplications for Plume-Lithosphere Interaction andPolybaric Melt Hybridization Journal of Petrology 38 1359 ndash 1390

Baker JA MacPherson CG Menzies MAThirlwall MF Al-Kadasi M and Mattey DP(2000) Resolving crustal and mantle contributions tocontinental ood volcanism Yemen constraintsfrom mineral oxygen isotope data Journal ofPetrology 41 1805 ndash1820

Beard JS Abitz RJ and Lofgren GE (1993)Experimental melting of crustal xenoliths fromKilbourne Hole New Mexico and implications forthe contamination and genesi s of magmasContributions to Mineralogy and Petrology 11588 ndash102

Bell K (2001) Carbonatites relationships to mantle-plume activity Geological Society of AmericaSpecial Paper 352 267 ndash290

Bell K and Blenkinsop J (1989) Neodymium andStrontium Isotope Geochemistry of Carbonatites Pp278 ndash300 in Carbonatites ndash Genesis and Evolution(K Bell editor) Unwin Hyman London

Bell K and Simonetti A (1996) Carbonat iteMagmatism and Plume Activity Implications fromthe Nd Pb and Sr Isotope Systematics of OldoinyoLengai Journal of Petrology 37 1321 ndash1339

Bell K and Tilton GR (2001) Nd Pb and Sr IsotopicCompositions of East African Carbonat itesEvide nce for Mant le Mixing and PlumeInhomogene ity Journa l of Petrology 42 1927 ndash1945

Benoit M Polve M and Ceuleneer G (1996) Traceelement and isotopic characterization of ma ccumulates in a fossil mantle diapir (Oman ophiolite)Chemical Geology 134 199 ndash 214

Bernstein S Kelemen PB Tegner C Kurz MDBlusztajn J and Kent Brooks C (1998) Post-breakup basaltic magmatism along the EastGreenland Tertiary rifted margin Earth and

Planetary Science Letters 160 845 ndash862Boynton WV (1984) Geochemistry of the rare earth

elements meteorite studies Pp 63 ndash114 in RareEarth Element Geochemistry (P Henderson editor)Elsevier Amsterdam

Carroll MR and Wyllie PJ (1990) The systemtonalite-H2O at 15 kbar and the genesis of calc-alkaline magmas American Mineralogist 75345 ndash 357

Chauvel C Hofmann AW and Vidal P (1992)HIMU-EM The French Polynesian connectionEarth and Planetary Science Letters 110 99 ndash119

Clayton RN and Mayeda TK (1963) The use ofbromine penta uoride in the extraction of oxygenfrom oxides and silicates for isotope analysisGeochimica et Cosmochimica Acta 27 43 ndash52

Coulson IM and Chambers AD (1996) Patterns ofzonation in rare-earth-bearing minerals in nephelinesyenites of the North Qoroq center South GreenlandThe Canadian Mineralogist 34 1163 ndash1178

Cox KG Bell JD and Pankhurst RJ (1979) TheInterpretation of Igneous Rocks George Allen andUnwin London

Cross W Iddings JP Pirsson LV and WashingtonHS (1903) Quantitative Classi cation of IgneousRocks University of Chicago Press

Dahl-Jensen T Thybo T Hopper H and Rosing M(1998) Crustal structure at the SE Greenland marginfrom wide-angle and normal incidence seismic dataTectonophysics 288 191 ndash 198

DePaolo DJ (1981) Trace element and isotopic effectsof combined wallrock assimilation and fractionalcrystallisation Earth and Planetary Science Letters53 189 ndash 202

De P a o lo D J (1 9 8 8 ) Neodymium Iso topeGeochemistry An Introduction Springer-VerlagNew York

Dobosi G Downes H Mattey D and Embey-IsztinA (1998) Oxygen isotope ratios of phenocrysts fromalkali basalts of the Pannonian basin Evidence foran O-isotopically homogeneous upper mantle be-neath a subduction-in uenced area Lithos 42213 ndash 223

Eiler JM (2001) Oxygen isotope variations of basalticlavas and upper mantle rocks Pp 319 ndash364 inStable Isotope Geochemistry (JW Valley and DRCole editor s) Reviews in Minera logy andGeochemistry 43 The Mineralogical Society ofAmerica Washington DC

Eiler JM Farley KA Valley JW Hauri E CraigH Hart SR and Stolper EM (1997) Oxygenisotope variations in ocean island basalt phenocrystsGeochimica et Cosmochimica Acta 61 2281 ndash2293

Eiler JM Crawford A Elliot T Farley KAValley JW and Stolper EM (2000a) Oxygenisotope geochemistry of oceanic-arc lavas Journal

850

R HALAMA ETAL

TA

BL

E3

RE

Eda

taof

who

lero

cks

and

clin

opyr

oxen

es

the

latt

erde

term

ined

byL

aser

ICP

-MS

Who

le-r

ock

anal

yses

R

epre

sent

ativ

ecl

inop

yrox

ene

anal

yses

S

ampl

eno

10

1231

1014

4610

1457

1014

7218

6389

1863

95E

F02

4-2

EF

039-

1E

F03

9-2

EF

039-

4E

F07

2-2

EF

072-

3E

F16

8-3

EF

168-

5

RE

Eco

ncen

trat

ions

(ppm

)L

a11

65

180

610

29

661

155

412

28

440

266

199

150

273

304

165

527

59

Ce

290

346

05

249

617

13

378

130

08

180

411

46

868

717

105

012

60

572

399

67

Pr

434

67

388

268

527

433

331

216

181

128

249

262

102

917

43

Nd

200

631

71

176

412

49

250

920

16

190

213

82

122

28

9815

64

157

258

51

960

6S

m4

756

974

043

215

914

826

664

194

283

276

077

0615

82

265

9E

u1

742

51

571

071

971

721

831

491

111

212

001

943

877

07G

d4

826

584

043

555

84

996

965

615

244

326

156

0514

75

252

7T

b ndash

ndash ndash

ndash ndash

ndash1

351

010

770

711

111

212

313

61D

y4

475

223

613

565

354

816

865

934

784

255

996

8613

10

205

2H

o0

921

040

730

731

11

011

481

401

001

001

211

292

394

24E

r2

432

531

951

972

872

663

893

342

642

112

993

646

539

60T

m ndash

ndash ndash

ndash ndash

ndash0

450

390

360

290

350

440

861

27Y

b2

182

071

741

832

592

363

782

812

492

452

703

016

788

89L

u0

340

310

260

270

390

350

460

410

300

310

520

411

251

60

(Eu

Eu

) N1

111

131

190

971

031

070

820

940

720

991

000

910

770

83

Eu

anom

aly

isca

lcul

ated

as(E

uE

u) N

=E

u N(

SmN6

Gd N

)05

norm

aliz

edto

chon

drit

eva

lues

from

Boy

nton

(198

4)

840

R HALAMA ETAL

in clinopyroxenes of EF 168 compared to theother samples (EF 024 039 and 072) seems onlypartly due to a higher degree of fractionationsincethe Mg value of EF 168 is only slightly lowerThus the LREE enrichment appears to underlinethe more alkaline character of this sample Acomparison with published REE patterns ofclinopyroxenes (Fig 6d) shows that the EFpatterns are unlike patterns from olivine gabbroswith MORB-like signatures (Benoit et al 1996)but fairly similar to clinopyroxene megacrystpatterns from an OIB-like alkaline basalt (Nimisand Vannucci 1995)

Sr Nd and O isotopic composition

Sr Nd and O isotopic compositions of all samplesare presented in Table 4 Sr and Nd isotopiccompositions of the clinopyroxene separates havebeen recalculated to an approximate eruption age

of 12 Ga (Paslick et al 1993) using decayconstants of 142610ndash11 andash1 for 87Rb (Steigerand Jager 1977) and 654610ndash12 a ndash1 for 147Sm(Lugmair and Marti 1978) Present day CHURvalues of 01967 for 147Sm144Nd (Jacobsen andWasserburg 1980) and 0512638 for 143Nd144Nd(Goldstein et al 1984) were used to calculateinitial eNd values 87Sr86Sr initial ratios are fairlyconstant and range from 070278 to 070383(Table 4) eNd(i) values vary from ndash32 to +21(Table 4) On the Sr-Nd isotope diagram the EFbasalts de ne a relatively steep trend (Fig 7) withinitial 87Sr86Sr ratios similar to those of the BulkSilicate Earth (BSE) Only one sample is slightlydisplaced towards a more radiogenic initial87Sr86Sr ratio None of the samples has an eNd

value close to depleted mantle values estimatedto be between eNd = +53 (calculated afterDePaolo 1981) and eNd = +73 (calculated afterGoldstein et al 1984) at 12 Ga The two alkalinebasalts of the Il otilde maussaq Group are within therange of the other data but they de ne a restricted eld around CHUR and BSE respectively Errorsinduced by the uncertainty in the age of the EFbasalts are largely within the error of the isotopeanalyses and do not change the overall picture Incomparison with other published Sr-Nd isotopedata from the Gardar Province the initial Srisotope ratios of the EF basalts are very similar torelatively primitive basaltic lamprophyric andcarbonatitic dyke rocks which have (87Sr86Sr)i of~0703 (Pearce and Leng 1996 Andersen 1997Goodenough et al 2002) So far no evidence hasbeen presented for an isotopicallyenriched mantlesource in the Gardar Province as the existingeNd(i) data of ndash05 to +52 show an isotopicallydepleted signature (Pearce and Leng 1996Andersen 1997 Goodenough et al 2002Upton et al 2003) Clinopyroxene from the EFsamples of Paslick et al (1993) yield eNd(i) of+22 The Nd isotope data of this study are withinthe lower range of published values from theGardar Province and extend the spread to morenegative eNd(i) values

Several recent oxygen isotope studies haveshown that mineral separates are less sensitive tolow-T alteration and secondary water take-up thanwhole-rock samples (eg Vroon et al 2001)However only few oxygen isotope data ofmineral separates from continental volcanics areavailable (eg Dobosi et al 1998 Baker et al2000) Extensive studies of mineral separates ofocean island basalts (Eiler et al 1997) andoceanic-arc lavas (Eiler et al 2000a) demon-

c (sa

mpl

e)c

(cho

ndri

te)

c (sa

mp

le)c

(cho

ndrit

e)

1

10

100

200

101446

101472

101231

101457

186389

186395

La Ce Pr Nd Sm Eu Gd Dy Ho Er Yb Lu

1

10

100

200

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

N-MORBE-MORBOIBBasaltic dykes from Ivittuut

a

b

FIG 5 (a) Whole-rock REE data of EF basaltsnormalized to chondritic values from Boynton (1984)(b) Typical compositions of N-MORB E-MORB andOIB (Sun and McDonough 1989) are shown for

comparison

PETROGENESIS OF THE ERIKSFJORD BASALTS

841

strated that analyses of mineral separates consis-tently span a narrower range than comparablewhole-rock data In our study we used clino-pyroxene because it is the only fresh mineralphase present in the samples Oxygen isotopemeasurements of clinopyroxene separates fromthe EF basalts show a range in d18Ocpx from +52to +60 with an average value of +566plusmn023(1s) All samples overlap the range of d18Ocpx

values from mantle peridotites (+525 to +590Mattey et al 1994) from continental oodbasalts from Yemen (+55 to +69 Baker etal 2000) and from alkaline basalts of Hungary(+507 to +534 Dobosi et al 1998) There is asmall difference between the averages ofMussartut Group (d18Ocpx-avg = +575plusmn015(1s)) and Ulukasik Group (d18Ocpx-avg =+575plusmn009 (1s)) samples compared toIl otildeacutemaussaq Group (d18Ocpx-avg = +533plusmn021(1s)) samples

d18O values of melts coexisting with clino-pyroxene were calculated assuming that oxygen

in pyroxene and melt were in isotopic equilibriumand T was 1100ordmC Dmelt-cpx can then be calculatedusing the following equation of Kalamarides(1986)

Dmelt-cpx = ndash00008T + 143 (T in Kelvin)

Although fractionation varies with magmatictemperatures we consider the temperature of1100ordmC and the resulting Dmelt-cpx of +03 assuf ciently representative considering the analy-tical errors involved It is similar to Dmelt-cpx

values used in previous studies (+039 Vroonet al 2001 +02 Dobosi et al 1998)Resulting d18Omelt values of EF basalts rangefrom +55 to +63 These values can becompared to whole-rock data from continentalbasalts given by Harmon and Hoefs (1995)Intraplate basalts from continental rift zonesshow a range from +46 to +83 with a meanof +61 and continental ood basalts which arerather poorly represented in the database havevalues between +43 and +65 Calculated EF

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Cpx from MORB-like

olivine gabbro

Cpx megacrysts from OIB-like

alkaline basalts

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 024EF 039

Mussartucirct Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 168

Iliacutemaussaq Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

1

10

100

1000

EF 072

Ulukasik Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb LuLa Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

c

ba

d

1

10

100

01

FIG 6 REE patterns of clinopyroxenes (cpx) from Mussartut (a) Ulukasik (b) and Il otilde maussaq (c) Group samplesnormalized to chondritic values from Boynton (1984) (d) REE patterns of cpx from MORB-type olivine gabbro(Benoit et al 1996) and OIB-type alkaline basalt (Nimis and Vannucci 1995) for comparison (note the change of

scale)

842

R HALAMA ETAL

TA

BL

E4

Sr

Nd

and

Ois

otop

icco

mpo

siti

ons

ofcl

inop

yrox

ene

sepa

rate

sfr

omE

riks

fjor

dF

orm

atio

nba

salt

sW

hole

-roc

kan

alys

esfr

omth

eK

etil

idia

nba

sem

ent

and

from

sedi

men

tary

rock

sof

the

Eri

ksfj

ord

For

mat

ion

are

give

nfo

rco

mpa

riso

n

Sam

ple

Mem

ber

Sr

Rb

87R

b86S

r87S

r86Sr

87Sr

86Sr

init

ial

SmN

d147Sm

144N

d143N

d144N

de N

d(i

)d1

8O

(ppm

)(p

pm)

(ppm

)(p

pm)

Bas

alts

E

F02

4M

ussa

rtut

646

50

486

002

180

7042

01plusmn

100

7038

267

3822

80

019

560

5124

67plusmn

10ndash

32

567

EF

063

Mus

sart

ut5

60plusmn

011

EF

061

Mus

sart

ut89

95

099

90

0321

070

3818

plusmn10

070

3266

574

163

90

2115

051

2629

plusmn10

ndash2

55

79plusmn

028

EF

039

Mus

sart

ut73

74

058

30

0229

070

3492

plusmn10

070

3098

662

191

30

2092

051

2626

plusmn10

ndash2

25

71plusmn

001

EF

059

Mus

sart

ut54

47

013

70

0073

070

3375

plusmn12

070

3250

606

168

90

2171

051

2727

plusmn09

ndash1

45

99plusmn

001

EF

108

Ulu

kasi

k66

92

111

00

0480

070

3741

plusmn09

070

2916

109

936

82

018

050

5126

17plusmn

10+

21

581

EF

072

Ulu

kasi

k69

41

032

20

0134

070

3014

plusmn10

070

2784

818

251

90

1962

051

2574

plusmn10

ndash1

25

68E

F17

4Il

otildemau

ssaq

125

42

324

005

360

7039

35plusmn

100

7030

1418

76

796

10

1424

051

2230

plusmn10

+0

45

47E

F16

8Il

otildemau

ssaq

866

71

422

004

750

7037

99plusmn

070

7029

8316

45

641

60

1550

051

2304

plusmn09

ndash0

15

18

Bas

emen

tan

dse

dim

enta

ryro

cks

JG02

Juli

aneh

aEcirc b30

01

169

71

6375

074

4789

plusmn10

071

6647

105

362

28

010

220

5114

70plusmn

09ndash

83

78

Gra

nite

EF

021

Qua

rtzi

te11

5E

F02

8A

rkos

ic6

8ar

enit

e

Sam

ples

are

list

edin

stra

tigra

phic

orde

rfr

omol

dest

toyo

unge

st87

Sr86

Sr

init

ialr

atio

san

de N

d(i)

valu

esw

ere

calc

ulat

edfo

ran

age

of1

2G

aS

tand

ard

devi

atio

nsfo

rd18

Oar

egi

ven

for

sam

ples

that

wer

ean

alys

edtw

ice

PETROGENESIS OF THE ERIKSFJORD BASALTS

843

model melt oxygen isotope data are within theranges of both groups Compared to Igaliko dykerocks from further SE in the Gardar Province(Pearce and Leng 1996) the EF basalts cover amuch more restricted eld possibly because d18Owhole-rock analyses are more prone to secondaryalteration than analyses of mineral separates(Fig 8) Whole-rock d18O measurements of theunderlying JulianehaEcirc b Granite and of the sedi-mentary sandstone members of the EriksfjordFormation yield values of 78 and 68 ndash115respectively (Table 4) and are shown forcomparison

Discussion

Mantle source regions of the EF basaltsTo reconstruct the mantle sources involved in theformation of the EF basalts we use Sr-Nd-Oisotope and incompatible trace element data On aSr-Nd isotope diagram the isotopic compositionof DMM was modelled for an age of 12 Ga(DePaolo 1981 Goldstein et al 1984 Taylor andMcLennan 1985) The fairly constant 87Sr86Srinitial isotope ratios of the EF basalts are ~0703which falls within the range typical for mantle-derived rocks but slightly above calculated valuesfor DMM (Fig 7) eNd(i) values are more

-4

-2

0

2

4

6

8

0701 0702 0703 0704 0705 0706

DMM Typical 2 s error

CHUR

BS

E

(87Sr86Sr)i

e Nd

(i)

EM-1

Qassiarsuk carbonatites and lamprophyres

EF basalts

Depleted MORB mantle

Ivittuut basaltic and lamprophyric dikes

Igaliko carbonatites and lamprophyres

Enriched mantle 1

Carbonatites from other regions of similar age

FIG 7 eNd(i) vs (87Sr86Sr)i diagram calculated for t = 12 Ga which is considered as the eruption age of the EFbasalts (Paslick et al 1993) The reference line of Bulk Silicate Earth (BSE) for the Sr isotopic composition wascalculated after DePaolo (1988) assuming present-day values of 00827 for 87Rb86Sr and 07045 for 87Sr86SrIsotopic composition of the depleted MORB mantle (DMM) was calculated after DePaolo (1981) Goldstein et al(1984) and Taylor and McLennan (1985) EM-1 was calculated after Schleicher et al (1998) (see text for details)Data sources are Goodenough et al (2002) Ivittuut (n = 6) Pearce and Leng (1996) Igaliko (n = 5) Andersen(1997) Qassiarsuk (n = 25) Bell and Blenkinsop (1989) other carbonatites of similar age (Goudini Schryburt Lake

and Prairie Lake n = 6)

844

R HALAMA ETAL

scattered and indicate that the rocks contain amixture of isotopic components from both mantleand crustal reservoirs or a heterogeneous mantlesource Only one sample (EF 108) has an eNd(i)value signi cantly above 0 This sample exhibitsthe largest spread of clinopyroxene compositionswhich may re ect strong alteration Goodenoughet al (2002) also note that the highest eNd occursin the most altered sample of their sample setHowever Paslick et al (1993) report clino-pyroxene data from two Ulukasik Groupsamples with eNd(i) gt +2 and those sampleswere apparently not disturbed with regard to theirSm-Nd age information On the other hand Piperet al (1999) suggest that the Sm-Nd systematicsof those samples may be erroneous due to laterdisturbance by uids derived from the Il otildeacutemaussaqintrusion Therefore although we are presentlynot able to decide whether positive eNd(i) values

for the EF basalts are real it is clear that none ofthose values is close to values of a 12 Ga oldDMM source (ie eNd = +5 to +7)

The Sr-Nd isotopic composition for OIB-typesources in the Proterozoic is not as wellconstrained as DMM Today OIB-type mantlesources have a wide range of isotopic composi-tions that partly overlap with DMM but eNd

values are generally lower (Hart et al 1992Hofmann 1997) The EM-1 mantle component ischaracterized by extremely low 143Nd144Ndratios and eNd values of between 0 and ndash5(Hofmann 1997) A possible 12 Ga position forEM-1 is plotted in Fig 7 calculated according toSchleicher et al (1998) with 147Sm144Nd = 018and 87Rb86Sr = 010 The present-day values usedin the calculation are 143Nd144Nd = 051234 and87Sr86Sr = 07057 taken from ocean islandbasalts closest to an EM-1 end-member(Hofmann 1997) Other Sr-Nd isotope datafrom the Gardar Province are plotted to putfurther constraints on the mantle sources of the EFbasalts (Fig 7) We also added comparative datafrom African and Canadian carbonatitesof similarage because there is a well known similaritybetween Sr-Nd isotopic data of OIBs andcarbonatites (Bell and Blenkinsop 1989) Thelinear array of the carbonatite data is consideredto be the result of mixing between the two mantlecomponents HIMU and EM-1 (Bell andBlenkinsop 1989 Bell and Tilton 2001) Thecarbonatites from the Gardar Province havesimilar Nd but slightly more radiogenic Srisotopic compositions than the reference carbo-natite data (Fig 7) The least contaminatedsamples overlap with the linear array and themore radiogenic Sr isotopic composition can beexplained by crustal contamination (Andersen1997) We therefore consider it possible that theGardar carbonatites also represent mixing ofhypothetical HIMU and EM-1 components of anOIB-like source Since the isotopically mostprimitive EF basalt samples overlap with theQassiarsuk carbonatites it could be possible thatthe basalts also contain such mantle componentsalthough Pb isotope data for the EF basalts thatcould clarify the importance of a HIMU sourceare lacking Based on the Sr-Nd isotope data theEF basalts appear to be derived from aheterogeneous OIB-like mantle source thatcomprises isotopically depleted and enrichedcomponents OIB-like and carbonatitic magma-tism are often attributed to mantle plume activity(Bell and Simonetti 1996 Hofmann 1997

2

4

6

8

10

12

-10 -8 -6 -4 -2 0 2 4 6 8

DMM

d18 O

[permil]

e Nd (i)

Typical2s error

EM-1

EF basalts - cpx

EF basalts - model melt

Ketilidian basement

Igaliko dykes

Depleted MORB mantle

Enriched mantle 1

EM-1 lavas - cpx

FIG 8 Oxygen isotopi c compositio ns (d18OV-SMOW) of clinopyroxene from EF basalts calculatedmelt (see text for details) and whole-rock Ketilidianbasement vs eNd(i) Whole-rock data for Igaliko dykerocks (n = 7) are from Pearce and Leng (1996) DMMdata derived from fresh MORB glasses are from Eiler etal (2000b) The eld of EM-1 lavas is taken fromclinopyroxene data of Gough and Tristan da CunhaIsland lavas (Harris et al 2000) and the EM-1 end-member composition is based on the calculations afterSchleicher et al (1998) for eNd(i) and an average value

of 578 (n = 18) from the EM-1 lavas for d18O

PETROGENESIS OF THE ERIKSFJORD BASALTS

845

Simonetti et al 1998) A plume input to theformation of the EF basalts is supported by therelatively large volume of magmas and the shortduration (lt5 Ma) of magmatism (Piper et al1999) It is also compatible with the geologicalsetting of a continental breakup Additionally aplume component appears to be necessary toexplain the thermal input as it is dif cult to melt acold lithosphere section (Arndt and Christensen1992)

Oxygen isotope data of the EF basalts aregenerally consistent with a mantle derivationincluding an OIB-like mantle source On the d18Ovs eNd(i) diagram (Fig 8) the overlap betweenthe EF data clinopyroxene data from EM-1-likeoceanic basalts and the hypothetical EM-1 end-member suggests an involvement of an EM-1-type mantle source Olivines from HIMU-likesources are known to be slightly lower in d18Othan typical upper mantle values from MORBs(Eiler et al 1997) but the EF data are notsuf cient to assess if the slightly lower d18Ovalues in the Il otildeacutemaussaq Group basalts are causedby a relatively higher input from a HIMU source

The contribution of the various mantle sourcesto the EF basalts can also be evaluated on theprimitive mantle-normalized multi-elementdiagram (Fig 4) and on the REE plots (Figs 5and 6) Both the whole-rock REE and theclinopyroxene REE patterns of the EF basaltsare unlike those representative of N-MORBsources We conclude again that DMM did notplay a major role in the petrogenesis of the EFbasalts On the other hand similarities with OIB-like mantle sources are more prominent Theshape of the clinopyroxene REE pattern closelyresembles those from clinopyroxene of OIB-likealkaline basalts despite differences in absoluteconcentrations The whole-rock REE patterns ofthe EF basalts lie between E-MORB and averageOIB patterns as do the multi-element patterns Onthe latter the EF basalts have marked positive Baand P peaks that are not present in typical OIBsHowever Ba contents should be interpreted withcaution since highly variable ZrBa ratios from014 to 070 are an indication that Ba was re-mobilized after emplacement of the basalts (egPrice et al 1991 Oliveira and Tarney 1995)Likewise P mobility was demonstrated in alteredbasaltic rocks (Winchester and Floyd 1976) Insummary the REE and trace element data areevidence that the mantle source(s) of the EFbasalts were enriched in incompatible traceelements relative to primitive mantle

For relatively primitive hy-normative basalticrocks of the Ivittuut area further north in theGardar Province it has been argued that some ofthem were derived from a lithospheric mantlesource previously enriched by subduction-related processes This interpretation is basedon certain geochemical characteristics such ashigh LaNb ratios and negative Nb anomalies(Goodenough et al 2002) For the EF basaltsnegative Nb anomalies are not clearly discern-ible but LaNb ratios are gt1 which seems to be acharacteristic feature of a lithospheric mantlesource (Fitton 1995) LaBa LaRb and LaKratios were also used to argue for a lithosphericorigin of the Gardar basic magmas although amantle plume involvement was assumed toprovide the required energy (Upton andEmeleus 1987 Upton et al 2003) Sincefeldspar is the only primary magmatic mineralin the EF basalts in which Ba Rb and K areincorporated in signi cant amounts and thefeldspars are generally altered these elementsmight have been easily mobilized and wetherefore do not dare to base the interpretationof the EF basalts sources on these data Inaddition the lack of precise geochemical data onthe composition of the lithosphere below theGardar Province complicates an evaluation ofthe lithospheric contribution to the EF basalticmagmatism

Certain ratios of incompatible trace elements ofthe EF basalts are compared to several occur-rences of ocean island basalts representative ofEM-1 and HIMU sources in Table 5 Ratiosbetween highly incompatible trace elements donot change signi cantly during partial melting orfractional crystallization in basaltic systems andmay therefore constrain the ratios in their sources(Weaver et al 1987) The EF basalts aresubdivided into the more alkaline isotopicallyprimitive Il otildeacutemaussaq Group samples and thosesamples with eNd(i) lt ndash1 The Il otilde maussaq Groupsamples have higher SrNd LaNb and ZrNbratios than HIMU and overlap with the upperrange of EM-1 ratios The EF basalts with eNd(i)lt ndash1 have generally higher ratios than the oceanicisland basalts and partly overlap with typicalvalues of the lower crust It therefore seems likelythat assimilation of lower crustal material mayhave changed the primary trace element ratios inthe basalts Therefore we will evaluate thepossible in uence of assimilation of crustalmaterial on the geochemistry of the EF basaltsin the following section

846

R HALAMA ETAL

The role of crustal contamination

Large amounts of bulk assimilation of crustalmaterial which is relatively SiO2-rich comparedto a basaltic melt would drive the compositions tohigher SiO2 contents and quartz-normativecompositions Partial melts derived from crustalrocks would also be relatively SiO2-rich (Beard etal 1993 Carroll and Wyllie 1990) and wouldhave the same effect The ne-normative characterof the EF basalts (Table 1) suggests thatassimilation of crustal material especially ofsilica-rich upper crustal rocks was not signi cantwith the possible exception of sample EF 059

On the primitive mantle-normalized multi-element diagrams (Fig 4) Ti troughs are typicalfor crustal rocks especially for the Upper Crust(Rudnick and Fountain 1995) and they maytherefore be present in rocks extensively contami-nated by continental crust Ti troughs aregenerally lacking in the patterns of the EFbasalts and a major upper crustal contribution tothe magmas can therefore be excluded However

small amounts of assimilation of lower crustalmaterial cannot be ruled out because the averagelower crust has higher Ti contents

Steep trends on Sr-Nd isotope plots have beenattributed to assimilation of granulite-faciesgneisses (eg Bernstein et al 1998) AFCmodelling based on an equation developed byDePaolo (1981) was performed to test thehypothesis of small amounts of contaminationwith lower crustal material (Fig 9) The basalticmagma which represents one end-member for thecalculations has the isotopic signature of theisotopically most primitive samples from thepresent data set and from the literature(Table 6) Two possible contaminants repre-senting upper and lower crustal compositionswere selected for modelling (Table 6) Therepresentative samples are a granite from theJulianehaEcirc b batholith for the upper crust (sampleJG 02) and an average of ve granulite-faciesgneisses with low RbSr ratios from the Archaeancraton of West Greenland (Taylor et al 1984) for

TABLE 5 Trace element ratios of EF basalts compared to ocean island basalts(OIBs) representing EM-1 and HIMU mantle components and the lowercrust

SrNd LaNb ZrNb

EF basalts with eNd lt ndash 1 249 ndash396 120 ndash150 121 ndash157Il otildeacutemaussaq Group EF basalts 198 ndash292 104 ndash119 51 ndash53EM-1 90 ndash207 067 ndash111 39 ndash109HIMU 45 -163 063 ndash080 27 ndash 36Lower continental crust 30 16 136

Data sources EM-1 Humphris and Thompson (1983 Walvis Ridge) and Cliff et al(1991 Inaccessible Islands) HIMU Chauvel et al (1992 Tubuai Island) Lowercontinental crust Rudnick and Fountain (1995)

TABLE 6 End-member compositions used for AFC modelling

End-member Sr (87Sr86Sr)i Nd eNd(i) Data source(ppm) (T = 12 Ga) (ppm) (T = 12 Ga)

Primitive EF basaltic 507 070278 18 + 22 Sample EF 072magma Paslick et al (1993)Upper crust 300 071665 62 ndash83 Sample JG 02(JulianehaEcirc b Granite)Lower crust 747 070668 63 ndash 234 Taylor et al (1984)(Granulite facies gneiss)

The composition of the granulite-facies gneiss is an average of 5 samples with low RbSr ratios lt02

PETROGENESIS OF THE ERIKSFJORD BASALTS

847

the lower crust An in uence of Archaean rockson Gardar magmas has been demonstrated by Pbisotope studies (Taylor and Upton 1993) andseismic data were also interpreted to re ect theexistence of a wedge of Archaean crust thatextends southwards to Lindenow Fjord (Dahl-Jensen et al 1998) We have shown in theprevious sections on the basis of Ni trends thatfractionation of olivine was most relevant to themajor element evolution of the lavas whereasfractionation of other phases was apparently ofminor importance We therefore modelled theAFC processes with olivine fractionation usingmineral-melt partition coef cients of 0014 for Srand 00066 for Nd (Rollinson 1993)

The modelling of AFC processes is stronglydependent on the parameter r de ned as the ratioof mass assimilation rate and fractional crystal-lization rate (DePaolo 1981) The change inisotopic signature becomes enhanced for highervalues of r Reiners et al (1995) have shown thatr values might be very high and even gt1 in theearly stages of magmatic evolution at hightemperatures since crystallization can besuppressed by assimilation Therefore we calcu-

lated AFC processes with two r values of 05 and08 that represent a relatively high and a middle tolow assimilation rate to demonstrate the effects ofvariable r (Fig 9) The results indicate that AFCprocesses involving upper crust similar to theJulianehaEcirc b Granite cannot explain the trend of theEF basalts because Sr isotopic compositionswould shift to considerably more radiogenicvalues This result is consistent with theconclusions derived from trace element dataThe AFC curves for assimilation of lower crusthowever agree well with the bulk of the data Themost negative eNd values can be explained byAFC processes with intermediate assimilationrates (r = 05) involving assimilation of lt5lower crustal material With a greater degree ofassimilation and an r value of 08 lt4assimilation is indicated

Lower crustal rocks of igneous origin have anaverage d18O value of +75plusmn14 with an overallrange from +54 to +125 (Fowler and Harmon1990) 5 to 10 bulk assimilation of lower crustalmaterial with d18O = +75 into a typical mantle-derived magma with d18O = +57 results ind18O values of the contaminated magma from

-24

-20

-16

-12

-8

-4

0

4

0700 0703 0706 0709 0712 0715 0718

r = 08

r = 05

r = 05

r = 08

EF Basalts

Endmembers of AFC calculations

Upper Crust

Lower Crust

Primitive EF basaltic magma

(87Sr86Sr)i

e Nd

(i)

09

08

06

09

08

06

09

0806

09

08

07

07

04

Typical errorCHUR

BS

E

FIG 9 eNd(i) vs (87Sr86Sr)i diagram showing the results of AFC modelling End-member compositions are listed inTable 6 Olivine is the only fractionating mineral Partition coef cients for olivinemelt are from the compilation ofRollinson (1993) Curves were calculated for two values of r (r = rate of assimilation rate of fractionalcrystallization) Dashed lines and numbers in italics are for r = 05 solid lines and regular numbers are for r = 08

Values at the curves are for F values (fraction of melt remaining) with 10 tick marks

848

R HALAMA ETAL

+579 to +588 Therefore the d18O data cannotbe used to rule out small amounts of assimilationof lower crust To evaluate how much contamina-tion would be permissible from the presentoxygen isotope data we calculated how muchcontamination of a mantle-derived magma with18O = +57 with a contaminant within the lowerrange of lower crustal xenoliths (d18O = +67)would drive the oxygen isotopic composition tothe highest value of the EF basalts (d18O~+63) Although simple mixing shows that60 assimilation is permissible this amountwould drive the whole-rock geochemistry tomore silicic compositions than is observed in theEF basalts In Fig 8 there is no obvious trendfrom the basalts towards the Ketilidian basementsuggesting again that upper crust was notassimilated in signi cant amounts

Ratios of trace elements incompatible inbasaltic systems can be used to distinguishbetween crust and mantle reservoirs (egWeaver 1991) Based on the AFC modellingEF basalts with eNd(i) lt ndash1 are those which aremost likely in uenced by crustal contaminationThe relatively high LaNb and ZrNb ratios ofthese samples are consistent with assimilation oflower crustal material SrNd ratios are evenhigher than the average lower crustal composi-tion but SrNd ratios in those samples which weretaken as representative for the lower crust in theAFC modelling can reach values of 80 ie wellabove the highest EF basalt ratios Again theSrNd ratios support the hypothesis of some lowercrustal contamination

The absence of more primitive basaltic rockswith MgO gt8 wt might also indicate lowercrustal assimilation It suggests that magmas mayhave been retained at depth where they acquiredmore evolved compositions by fractional crystal-lization and had the opportunity to assimilatelower crustal material during crustal underplating(Upton et al 2003)

Summary and conclusions

The new geochemical data of the EriksfjordFormation basalts in the Gardar Provinceprovided in this study are used to placepetrogenetic constraints on the sources of theirparental magmas Based on trace-element patternsand Sr-Nd isotopic constraints there seems to beno indication that a DMM-like source played animportant role in magma genesis of the EFbasalts Clinopyroxene and whole-rock REE

patterns are more compatible with an OIB-likesource but we are not able to conclude whetherthe observed enrichment in REE re ects thesourcersquos signature or magmatic processes Theisotopically most primitive EF basalts have Sr-Ndisotopic compositions that are similar to Gardarcarbonatites Due to the isotopic similaritybetween carbonatites and OIBs many arethought to be related to mantle plumes (Bell2001) We therefore speculate that OIB-likemantle components from a plume source werealso involved in the parental EF magmas but wecannot exclude interaction with the SCLM Twodistinct mantle components one isotopicallydepleted and another one enriched could havecontributed to the EF magmas AFC modelling ofSr-Nd isotope data indicates that assimilation oflt5 lower crust could explain the steep vector onthe Sr-Nd isotope plot but assimilation of uppercrustal material is unlikely Oxygen isotope dataof clinopyroxene separates are consistent with amantle origin but permit some assimilation oflower crustal rocks In summary the data areconsistent with the hypothesis that the EFmagmas could have originated from a mantleplume source which is supported by geologicalarguments and that they underwent someassimilation of lower crustal material

Acknowledgements

We would like to thank Bruce Paterson whoprovided invaluable help during Laser ICP-MSmeasurements at the Large-Scale GeochemicalFacility supported by the European Community-Access to Research Infrastructure action of theImproving Human Potential Programme contractnumber HPRI-CT-1999-00008 awarded to ProfBJ Wood (University of Bristol) Gabi Stoschekand Torsten Vennemann are thanked for their helpwith oxygen isotope measurements Elmar Reitterexpertly assisted with preparation and measure-ments of radiogenic isotopes and MathiasWestphal helped with microprobe measurementsREE whole-rock analyses were kindly providedby the NERC facility at Royal HollowayUniversity of London Michael Marks is thankedfor his pleasant company during eldwork Veryconstructive reviews by Tom Andersen andAlexei S Rukhlov greatly helped to improve themanuscript We also thank Ian Coulson for nalediting Funding of this work by the DeutscheForschungsgemeinschaft (grant Ma-21351-2) isgratefully acknowledged

PETROGENESIS OF THE ERIKSFJORD BASALTS

849

References

Andersen T (1997) Age and petrogenesis of theQassiarsuk carbonatite-alkaline silicate volcaniccomplex in the Gardar rift South GreenlandMineralogical Magazine 61 499 ndash513

Armstrong JT (1991) Quantitative elemental analysisof individual microparticles with electron beaminstruments Pp 261 ndash315 in Electron ProbeQuantitation (KFJ Heinrich and DE Newburyeditors) Plenum Press New York and London

Arndt NT and Christensen U (1992) The role oflithospheric mantle in continental ood volcanismThermal and geochemical constraints Journal ofGeophysical Research 97 10967 ndash 10981

Baker JA Menzies MA Thirlwall MF andMacpherson CG (199 7) Pet rogenes is ofQuaternary Intraplate Vocanism Sanarsquoa YemenImplications for Plume-Lithosphere Interaction andPolybaric Melt Hybridization Journal of Petrology 38 1359 ndash 1390

Baker JA MacPherson CG Menzies MAThirlwall MF Al-Kadasi M and Mattey DP(2000) Resolving crustal and mantle contributions tocontinental ood volcanism Yemen constraintsfrom mineral oxygen isotope data Journal ofPetrology 41 1805 ndash1820

Beard JS Abitz RJ and Lofgren GE (1993)Experimental melting of crustal xenoliths fromKilbourne Hole New Mexico and implications forthe contamination and genesi s of magmasContributions to Mineralogy and Petrology 11588 ndash102

Bell K (2001) Carbonatites relationships to mantle-plume activity Geological Society of AmericaSpecial Paper 352 267 ndash290

Bell K and Blenkinsop J (1989) Neodymium andStrontium Isotope Geochemistry of Carbonatites Pp278 ndash300 in Carbonatites ndash Genesis and Evolution(K Bell editor) Unwin Hyman London

Bell K and Simonetti A (1996) Carbonat iteMagmatism and Plume Activity Implications fromthe Nd Pb and Sr Isotope Systematics of OldoinyoLengai Journal of Petrology 37 1321 ndash1339

Bell K and Tilton GR (2001) Nd Pb and Sr IsotopicCompositions of East African Carbonat itesEvide nce for Mant le Mixing and PlumeInhomogene ity Journa l of Petrology 42 1927 ndash1945

Benoit M Polve M and Ceuleneer G (1996) Traceelement and isotopic characterization of ma ccumulates in a fossil mantle diapir (Oman ophiolite)Chemical Geology 134 199 ndash 214

Bernstein S Kelemen PB Tegner C Kurz MDBlusztajn J and Kent Brooks C (1998) Post-breakup basaltic magmatism along the EastGreenland Tertiary rifted margin Earth and

Planetary Science Letters 160 845 ndash862Boynton WV (1984) Geochemistry of the rare earth

elements meteorite studies Pp 63 ndash114 in RareEarth Element Geochemistry (P Henderson editor)Elsevier Amsterdam

Carroll MR and Wyllie PJ (1990) The systemtonalite-H2O at 15 kbar and the genesis of calc-alkaline magmas American Mineralogist 75345 ndash 357

Chauvel C Hofmann AW and Vidal P (1992)HIMU-EM The French Polynesian connectionEarth and Planetary Science Letters 110 99 ndash119

Clayton RN and Mayeda TK (1963) The use ofbromine penta uoride in the extraction of oxygenfrom oxides and silicates for isotope analysisGeochimica et Cosmochimica Acta 27 43 ndash52

Coulson IM and Chambers AD (1996) Patterns ofzonation in rare-earth-bearing minerals in nephelinesyenites of the North Qoroq center South GreenlandThe Canadian Mineralogist 34 1163 ndash1178

Cox KG Bell JD and Pankhurst RJ (1979) TheInterpretation of Igneous Rocks George Allen andUnwin London

Cross W Iddings JP Pirsson LV and WashingtonHS (1903) Quantitative Classi cation of IgneousRocks University of Chicago Press

Dahl-Jensen T Thybo T Hopper H and Rosing M(1998) Crustal structure at the SE Greenland marginfrom wide-angle and normal incidence seismic dataTectonophysics 288 191 ndash 198

DePaolo DJ (1981) Trace element and isotopic effectsof combined wallrock assimilation and fractionalcrystallisation Earth and Planetary Science Letters53 189 ndash 202

De P a o lo D J (1 9 8 8 ) Neodymium Iso topeGeochemistry An Introduction Springer-VerlagNew York

Dobosi G Downes H Mattey D and Embey-IsztinA (1998) Oxygen isotope ratios of phenocrysts fromalkali basalts of the Pannonian basin Evidence foran O-isotopically homogeneous upper mantle be-neath a subduction-in uenced area Lithos 42213 ndash 223

Eiler JM (2001) Oxygen isotope variations of basalticlavas and upper mantle rocks Pp 319 ndash364 inStable Isotope Geochemistry (JW Valley and DRCole editor s) Reviews in Minera logy andGeochemistry 43 The Mineralogical Society ofAmerica Washington DC

Eiler JM Farley KA Valley JW Hauri E CraigH Hart SR and Stolper EM (1997) Oxygenisotope variations in ocean island basalt phenocrystsGeochimica et Cosmochimica Acta 61 2281 ndash2293

Eiler JM Crawford A Elliot T Farley KAValley JW and Stolper EM (2000a) Oxygenisotope geochemistry of oceanic-arc lavas Journal

850

R HALAMA ETAL

in clinopyroxenes of EF 168 compared to theother samples (EF 024 039 and 072) seems onlypartly due to a higher degree of fractionationsincethe Mg value of EF 168 is only slightly lowerThus the LREE enrichment appears to underlinethe more alkaline character of this sample Acomparison with published REE patterns ofclinopyroxenes (Fig 6d) shows that the EFpatterns are unlike patterns from olivine gabbroswith MORB-like signatures (Benoit et al 1996)but fairly similar to clinopyroxene megacrystpatterns from an OIB-like alkaline basalt (Nimisand Vannucci 1995)

Sr Nd and O isotopic composition

Sr Nd and O isotopic compositions of all samplesare presented in Table 4 Sr and Nd isotopiccompositions of the clinopyroxene separates havebeen recalculated to an approximate eruption age

of 12 Ga (Paslick et al 1993) using decayconstants of 142610ndash11 andash1 for 87Rb (Steigerand Jager 1977) and 654610ndash12 a ndash1 for 147Sm(Lugmair and Marti 1978) Present day CHURvalues of 01967 for 147Sm144Nd (Jacobsen andWasserburg 1980) and 0512638 for 143Nd144Nd(Goldstein et al 1984) were used to calculateinitial eNd values 87Sr86Sr initial ratios are fairlyconstant and range from 070278 to 070383(Table 4) eNd(i) values vary from ndash32 to +21(Table 4) On the Sr-Nd isotope diagram the EFbasalts de ne a relatively steep trend (Fig 7) withinitial 87Sr86Sr ratios similar to those of the BulkSilicate Earth (BSE) Only one sample is slightlydisplaced towards a more radiogenic initial87Sr86Sr ratio None of the samples has an eNd

value close to depleted mantle values estimatedto be between eNd = +53 (calculated afterDePaolo 1981) and eNd = +73 (calculated afterGoldstein et al 1984) at 12 Ga The two alkalinebasalts of the Il otilde maussaq Group are within therange of the other data but they de ne a restricted eld around CHUR and BSE respectively Errorsinduced by the uncertainty in the age of the EFbasalts are largely within the error of the isotopeanalyses and do not change the overall picture Incomparison with other published Sr-Nd isotopedata from the Gardar Province the initial Srisotope ratios of the EF basalts are very similar torelatively primitive basaltic lamprophyric andcarbonatitic dyke rocks which have (87Sr86Sr)i of~0703 (Pearce and Leng 1996 Andersen 1997Goodenough et al 2002) So far no evidence hasbeen presented for an isotopicallyenriched mantlesource in the Gardar Province as the existingeNd(i) data of ndash05 to +52 show an isotopicallydepleted signature (Pearce and Leng 1996Andersen 1997 Goodenough et al 2002Upton et al 2003) Clinopyroxene from the EFsamples of Paslick et al (1993) yield eNd(i) of+22 The Nd isotope data of this study are withinthe lower range of published values from theGardar Province and extend the spread to morenegative eNd(i) values

Several recent oxygen isotope studies haveshown that mineral separates are less sensitive tolow-T alteration and secondary water take-up thanwhole-rock samples (eg Vroon et al 2001)However only few oxygen isotope data ofmineral separates from continental volcanics areavailable (eg Dobosi et al 1998 Baker et al2000) Extensive studies of mineral separates ofocean island basalts (Eiler et al 1997) andoceanic-arc lavas (Eiler et al 2000a) demon-

c (sa

mpl

e)c

(cho

ndri

te)

c (sa

mp

le)c

(cho

ndrit

e)

1

10

100

200

101446

101472

101231

101457

186389

186395

La Ce Pr Nd Sm Eu Gd Dy Ho Er Yb Lu

1

10

100

200

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

N-MORBE-MORBOIBBasaltic dykes from Ivittuut

a

b

FIG 5 (a) Whole-rock REE data of EF basaltsnormalized to chondritic values from Boynton (1984)(b) Typical compositions of N-MORB E-MORB andOIB (Sun and McDonough 1989) are shown for

comparison

PETROGENESIS OF THE ERIKSFJORD BASALTS

841

strated that analyses of mineral separates consis-tently span a narrower range than comparablewhole-rock data In our study we used clino-pyroxene because it is the only fresh mineralphase present in the samples Oxygen isotopemeasurements of clinopyroxene separates fromthe EF basalts show a range in d18Ocpx from +52to +60 with an average value of +566plusmn023(1s) All samples overlap the range of d18Ocpx

values from mantle peridotites (+525 to +590Mattey et al 1994) from continental oodbasalts from Yemen (+55 to +69 Baker etal 2000) and from alkaline basalts of Hungary(+507 to +534 Dobosi et al 1998) There is asmall difference between the averages ofMussartut Group (d18Ocpx-avg = +575plusmn015(1s)) and Ulukasik Group (d18Ocpx-avg =+575plusmn009 (1s)) samples compared toIl otildeacutemaussaq Group (d18Ocpx-avg = +533plusmn021(1s)) samples

d18O values of melts coexisting with clino-pyroxene were calculated assuming that oxygen

in pyroxene and melt were in isotopic equilibriumand T was 1100ordmC Dmelt-cpx can then be calculatedusing the following equation of Kalamarides(1986)

Dmelt-cpx = ndash00008T + 143 (T in Kelvin)

Although fractionation varies with magmatictemperatures we consider the temperature of1100ordmC and the resulting Dmelt-cpx of +03 assuf ciently representative considering the analy-tical errors involved It is similar to Dmelt-cpx

values used in previous studies (+039 Vroonet al 2001 +02 Dobosi et al 1998)Resulting d18Omelt values of EF basalts rangefrom +55 to +63 These values can becompared to whole-rock data from continentalbasalts given by Harmon and Hoefs (1995)Intraplate basalts from continental rift zonesshow a range from +46 to +83 with a meanof +61 and continental ood basalts which arerather poorly represented in the database havevalues between +43 and +65 Calculated EF

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Cpx from MORB-like

olivine gabbro

Cpx megacrysts from OIB-like

alkaline basalts

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 024EF 039

Mussartucirct Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 168

Iliacutemaussaq Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

1

10

100

1000

EF 072

Ulukasik Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb LuLa Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

c

ba

d

1

10

100

01

FIG 6 REE patterns of clinopyroxenes (cpx) from Mussartut (a) Ulukasik (b) and Il otilde maussaq (c) Group samplesnormalized to chondritic values from Boynton (1984) (d) REE patterns of cpx from MORB-type olivine gabbro(Benoit et al 1996) and OIB-type alkaline basalt (Nimis and Vannucci 1995) for comparison (note the change of

scale)

842

R HALAMA ETAL

TA

BL

E4

Sr

Nd

and

Ois

otop

icco

mpo

siti

ons

ofcl

inop

yrox

ene

sepa

rate

sfr

omE

riks

fjor

dF

orm

atio

nba

salt

sW

hole

-roc

kan

alys

esfr

omth

eK

etil

idia

nba

sem

ent

and

from

sedi

men

tary

rock

sof

the

Eri

ksfj

ord

For

mat

ion

are

give

nfo

rco

mpa

riso

n

Sam

ple

Mem

ber

Sr

Rb

87R

b86S

r87S

r86Sr

87Sr

86Sr

init

ial

SmN

d147Sm

144N

d143N

d144N

de N

d(i

)d1

8O

(ppm

)(p

pm)

(ppm

)(p

pm)

Bas

alts

E

F02

4M

ussa

rtut

646

50

486

002

180

7042

01plusmn

100

7038

267

3822

80

019

560

5124

67plusmn

10ndash

32

567

EF

063

Mus

sart

ut5

60plusmn

011

EF

061

Mus

sart

ut89

95

099

90

0321

070

3818

plusmn10

070

3266

574

163

90

2115

051

2629

plusmn10

ndash2

55

79plusmn

028

EF

039

Mus

sart

ut73

74

058

30

0229

070

3492

plusmn10

070

3098

662

191

30

2092

051

2626

plusmn10

ndash2

25

71plusmn

001

EF

059

Mus

sart

ut54

47

013

70

0073

070

3375

plusmn12

070

3250

606

168

90

2171

051

2727

plusmn09

ndash1

45

99plusmn

001

EF

108

Ulu

kasi

k66

92

111

00

0480

070

3741

plusmn09

070

2916

109

936

82

018

050

5126

17plusmn

10+

21

581

EF

072

Ulu

kasi

k69

41

032

20

0134

070

3014

plusmn10

070

2784

818

251

90

1962

051

2574

plusmn10

ndash1

25

68E

F17

4Il

otildemau

ssaq

125

42

324

005

360

7039

35plusmn

100

7030

1418

76

796

10

1424

051

2230

plusmn10

+0

45

47E

F16

8Il

otildemau

ssaq

866

71

422

004

750

7037

99plusmn

070

7029

8316

45

641

60

1550

051

2304

plusmn09

ndash0

15

18

Bas

emen

tan

dse

dim

enta

ryro

cks

JG02

Juli

aneh

aEcirc b30

01

169

71

6375

074

4789

plusmn10

071

6647

105

362

28

010

220

5114

70plusmn

09ndash

83

78

Gra

nite

EF

021

Qua

rtzi

te11

5E

F02

8A

rkos

ic6

8ar

enit

e

Sam

ples

are

list

edin

stra

tigra

phic

orde

rfr

omol

dest

toyo

unge

st87

Sr86

Sr

init

ialr

atio

san

de N

d(i)

valu

esw

ere

calc

ulat

edfo

ran

age

of1

2G

aS

tand

ard

devi

atio

nsfo

rd18

Oar

egi

ven

for

sam

ples

that

wer

ean

alys

edtw

ice

PETROGENESIS OF THE ERIKSFJORD BASALTS

843

model melt oxygen isotope data are within theranges of both groups Compared to Igaliko dykerocks from further SE in the Gardar Province(Pearce and Leng 1996) the EF basalts cover amuch more restricted eld possibly because d18Owhole-rock analyses are more prone to secondaryalteration than analyses of mineral separates(Fig 8) Whole-rock d18O measurements of theunderlying JulianehaEcirc b Granite and of the sedi-mentary sandstone members of the EriksfjordFormation yield values of 78 and 68 ndash115respectively (Table 4) and are shown forcomparison

Discussion

Mantle source regions of the EF basaltsTo reconstruct the mantle sources involved in theformation of the EF basalts we use Sr-Nd-Oisotope and incompatible trace element data On aSr-Nd isotope diagram the isotopic compositionof DMM was modelled for an age of 12 Ga(DePaolo 1981 Goldstein et al 1984 Taylor andMcLennan 1985) The fairly constant 87Sr86Srinitial isotope ratios of the EF basalts are ~0703which falls within the range typical for mantle-derived rocks but slightly above calculated valuesfor DMM (Fig 7) eNd(i) values are more

-4

-2

0

2

4

6

8

0701 0702 0703 0704 0705 0706

DMM Typical 2 s error

CHUR

BS

E

(87Sr86Sr)i

e Nd

(i)

EM-1

Qassiarsuk carbonatites and lamprophyres

EF basalts

Depleted MORB mantle

Ivittuut basaltic and lamprophyric dikes

Igaliko carbonatites and lamprophyres

Enriched mantle 1

Carbonatites from other regions of similar age

FIG 7 eNd(i) vs (87Sr86Sr)i diagram calculated for t = 12 Ga which is considered as the eruption age of the EFbasalts (Paslick et al 1993) The reference line of Bulk Silicate Earth (BSE) for the Sr isotopic composition wascalculated after DePaolo (1988) assuming present-day values of 00827 for 87Rb86Sr and 07045 for 87Sr86SrIsotopic composition of the depleted MORB mantle (DMM) was calculated after DePaolo (1981) Goldstein et al(1984) and Taylor and McLennan (1985) EM-1 was calculated after Schleicher et al (1998) (see text for details)Data sources are Goodenough et al (2002) Ivittuut (n = 6) Pearce and Leng (1996) Igaliko (n = 5) Andersen(1997) Qassiarsuk (n = 25) Bell and Blenkinsop (1989) other carbonatites of similar age (Goudini Schryburt Lake

and Prairie Lake n = 6)

844

R HALAMA ETAL

scattered and indicate that the rocks contain amixture of isotopic components from both mantleand crustal reservoirs or a heterogeneous mantlesource Only one sample (EF 108) has an eNd(i)value signi cantly above 0 This sample exhibitsthe largest spread of clinopyroxene compositionswhich may re ect strong alteration Goodenoughet al (2002) also note that the highest eNd occursin the most altered sample of their sample setHowever Paslick et al (1993) report clino-pyroxene data from two Ulukasik Groupsamples with eNd(i) gt +2 and those sampleswere apparently not disturbed with regard to theirSm-Nd age information On the other hand Piperet al (1999) suggest that the Sm-Nd systematicsof those samples may be erroneous due to laterdisturbance by uids derived from the Il otildeacutemaussaqintrusion Therefore although we are presentlynot able to decide whether positive eNd(i) values

for the EF basalts are real it is clear that none ofthose values is close to values of a 12 Ga oldDMM source (ie eNd = +5 to +7)

The Sr-Nd isotopic composition for OIB-typesources in the Proterozoic is not as wellconstrained as DMM Today OIB-type mantlesources have a wide range of isotopic composi-tions that partly overlap with DMM but eNd

values are generally lower (Hart et al 1992Hofmann 1997) The EM-1 mantle component ischaracterized by extremely low 143Nd144Ndratios and eNd values of between 0 and ndash5(Hofmann 1997) A possible 12 Ga position forEM-1 is plotted in Fig 7 calculated according toSchleicher et al (1998) with 147Sm144Nd = 018and 87Rb86Sr = 010 The present-day values usedin the calculation are 143Nd144Nd = 051234 and87Sr86Sr = 07057 taken from ocean islandbasalts closest to an EM-1 end-member(Hofmann 1997) Other Sr-Nd isotope datafrom the Gardar Province are plotted to putfurther constraints on the mantle sources of the EFbasalts (Fig 7) We also added comparative datafrom African and Canadian carbonatitesof similarage because there is a well known similaritybetween Sr-Nd isotopic data of OIBs andcarbonatites (Bell and Blenkinsop 1989) Thelinear array of the carbonatite data is consideredto be the result of mixing between the two mantlecomponents HIMU and EM-1 (Bell andBlenkinsop 1989 Bell and Tilton 2001) Thecarbonatites from the Gardar Province havesimilar Nd but slightly more radiogenic Srisotopic compositions than the reference carbo-natite data (Fig 7) The least contaminatedsamples overlap with the linear array and themore radiogenic Sr isotopic composition can beexplained by crustal contamination (Andersen1997) We therefore consider it possible that theGardar carbonatites also represent mixing ofhypothetical HIMU and EM-1 components of anOIB-like source Since the isotopically mostprimitive EF basalt samples overlap with theQassiarsuk carbonatites it could be possible thatthe basalts also contain such mantle componentsalthough Pb isotope data for the EF basalts thatcould clarify the importance of a HIMU sourceare lacking Based on the Sr-Nd isotope data theEF basalts appear to be derived from aheterogeneous OIB-like mantle source thatcomprises isotopically depleted and enrichedcomponents OIB-like and carbonatitic magma-tism are often attributed to mantle plume activity(Bell and Simonetti 1996 Hofmann 1997

2

4

6

8

10

12

-10 -8 -6 -4 -2 0 2 4 6 8

DMM

d18 O

[permil]

e Nd (i)

Typical2s error

EM-1

EF basalts - cpx

EF basalts - model melt

Ketilidian basement

Igaliko dykes

Depleted MORB mantle

Enriched mantle 1

EM-1 lavas - cpx

FIG 8 Oxygen isotopi c compositio ns (d18OV-SMOW) of clinopyroxene from EF basalts calculatedmelt (see text for details) and whole-rock Ketilidianbasement vs eNd(i) Whole-rock data for Igaliko dykerocks (n = 7) are from Pearce and Leng (1996) DMMdata derived from fresh MORB glasses are from Eiler etal (2000b) The eld of EM-1 lavas is taken fromclinopyroxene data of Gough and Tristan da CunhaIsland lavas (Harris et al 2000) and the EM-1 end-member composition is based on the calculations afterSchleicher et al (1998) for eNd(i) and an average value

of 578 (n = 18) from the EM-1 lavas for d18O

PETROGENESIS OF THE ERIKSFJORD BASALTS

845

Simonetti et al 1998) A plume input to theformation of the EF basalts is supported by therelatively large volume of magmas and the shortduration (lt5 Ma) of magmatism (Piper et al1999) It is also compatible with the geologicalsetting of a continental breakup Additionally aplume component appears to be necessary toexplain the thermal input as it is dif cult to melt acold lithosphere section (Arndt and Christensen1992)

Oxygen isotope data of the EF basalts aregenerally consistent with a mantle derivationincluding an OIB-like mantle source On the d18Ovs eNd(i) diagram (Fig 8) the overlap betweenthe EF data clinopyroxene data from EM-1-likeoceanic basalts and the hypothetical EM-1 end-member suggests an involvement of an EM-1-type mantle source Olivines from HIMU-likesources are known to be slightly lower in d18Othan typical upper mantle values from MORBs(Eiler et al 1997) but the EF data are notsuf cient to assess if the slightly lower d18Ovalues in the Il otildeacutemaussaq Group basalts are causedby a relatively higher input from a HIMU source

The contribution of the various mantle sourcesto the EF basalts can also be evaluated on theprimitive mantle-normalized multi-elementdiagram (Fig 4) and on the REE plots (Figs 5and 6) Both the whole-rock REE and theclinopyroxene REE patterns of the EF basaltsare unlike those representative of N-MORBsources We conclude again that DMM did notplay a major role in the petrogenesis of the EFbasalts On the other hand similarities with OIB-like mantle sources are more prominent Theshape of the clinopyroxene REE pattern closelyresembles those from clinopyroxene of OIB-likealkaline basalts despite differences in absoluteconcentrations The whole-rock REE patterns ofthe EF basalts lie between E-MORB and averageOIB patterns as do the multi-element patterns Onthe latter the EF basalts have marked positive Baand P peaks that are not present in typical OIBsHowever Ba contents should be interpreted withcaution since highly variable ZrBa ratios from014 to 070 are an indication that Ba was re-mobilized after emplacement of the basalts (egPrice et al 1991 Oliveira and Tarney 1995)Likewise P mobility was demonstrated in alteredbasaltic rocks (Winchester and Floyd 1976) Insummary the REE and trace element data areevidence that the mantle source(s) of the EFbasalts were enriched in incompatible traceelements relative to primitive mantle

For relatively primitive hy-normative basalticrocks of the Ivittuut area further north in theGardar Province it has been argued that some ofthem were derived from a lithospheric mantlesource previously enriched by subduction-related processes This interpretation is basedon certain geochemical characteristics such ashigh LaNb ratios and negative Nb anomalies(Goodenough et al 2002) For the EF basaltsnegative Nb anomalies are not clearly discern-ible but LaNb ratios are gt1 which seems to be acharacteristic feature of a lithospheric mantlesource (Fitton 1995) LaBa LaRb and LaKratios were also used to argue for a lithosphericorigin of the Gardar basic magmas although amantle plume involvement was assumed toprovide the required energy (Upton andEmeleus 1987 Upton et al 2003) Sincefeldspar is the only primary magmatic mineralin the EF basalts in which Ba Rb and K areincorporated in signi cant amounts and thefeldspars are generally altered these elementsmight have been easily mobilized and wetherefore do not dare to base the interpretationof the EF basalts sources on these data Inaddition the lack of precise geochemical data onthe composition of the lithosphere below theGardar Province complicates an evaluation ofthe lithospheric contribution to the EF basalticmagmatism

Certain ratios of incompatible trace elements ofthe EF basalts are compared to several occur-rences of ocean island basalts representative ofEM-1 and HIMU sources in Table 5 Ratiosbetween highly incompatible trace elements donot change signi cantly during partial melting orfractional crystallization in basaltic systems andmay therefore constrain the ratios in their sources(Weaver et al 1987) The EF basalts aresubdivided into the more alkaline isotopicallyprimitive Il otildeacutemaussaq Group samples and thosesamples with eNd(i) lt ndash1 The Il otilde maussaq Groupsamples have higher SrNd LaNb and ZrNbratios than HIMU and overlap with the upperrange of EM-1 ratios The EF basalts with eNd(i)lt ndash1 have generally higher ratios than the oceanicisland basalts and partly overlap with typicalvalues of the lower crust It therefore seems likelythat assimilation of lower crustal material mayhave changed the primary trace element ratios inthe basalts Therefore we will evaluate thepossible in uence of assimilation of crustalmaterial on the geochemistry of the EF basaltsin the following section

846

R HALAMA ETAL

The role of crustal contamination

Large amounts of bulk assimilation of crustalmaterial which is relatively SiO2-rich comparedto a basaltic melt would drive the compositions tohigher SiO2 contents and quartz-normativecompositions Partial melts derived from crustalrocks would also be relatively SiO2-rich (Beard etal 1993 Carroll and Wyllie 1990) and wouldhave the same effect The ne-normative characterof the EF basalts (Table 1) suggests thatassimilation of crustal material especially ofsilica-rich upper crustal rocks was not signi cantwith the possible exception of sample EF 059

On the primitive mantle-normalized multi-element diagrams (Fig 4) Ti troughs are typicalfor crustal rocks especially for the Upper Crust(Rudnick and Fountain 1995) and they maytherefore be present in rocks extensively contami-nated by continental crust Ti troughs aregenerally lacking in the patterns of the EFbasalts and a major upper crustal contribution tothe magmas can therefore be excluded However

small amounts of assimilation of lower crustalmaterial cannot be ruled out because the averagelower crust has higher Ti contents

Steep trends on Sr-Nd isotope plots have beenattributed to assimilation of granulite-faciesgneisses (eg Bernstein et al 1998) AFCmodelling based on an equation developed byDePaolo (1981) was performed to test thehypothesis of small amounts of contaminationwith lower crustal material (Fig 9) The basalticmagma which represents one end-member for thecalculations has the isotopic signature of theisotopically most primitive samples from thepresent data set and from the literature(Table 6) Two possible contaminants repre-senting upper and lower crustal compositionswere selected for modelling (Table 6) Therepresentative samples are a granite from theJulianehaEcirc b batholith for the upper crust (sampleJG 02) and an average of ve granulite-faciesgneisses with low RbSr ratios from the Archaeancraton of West Greenland (Taylor et al 1984) for

TABLE 5 Trace element ratios of EF basalts compared to ocean island basalts(OIBs) representing EM-1 and HIMU mantle components and the lowercrust

SrNd LaNb ZrNb

EF basalts with eNd lt ndash 1 249 ndash396 120 ndash150 121 ndash157Il otildeacutemaussaq Group EF basalts 198 ndash292 104 ndash119 51 ndash53EM-1 90 ndash207 067 ndash111 39 ndash109HIMU 45 -163 063 ndash080 27 ndash 36Lower continental crust 30 16 136

Data sources EM-1 Humphris and Thompson (1983 Walvis Ridge) and Cliff et al(1991 Inaccessible Islands) HIMU Chauvel et al (1992 Tubuai Island) Lowercontinental crust Rudnick and Fountain (1995)

TABLE 6 End-member compositions used for AFC modelling

End-member Sr (87Sr86Sr)i Nd eNd(i) Data source(ppm) (T = 12 Ga) (ppm) (T = 12 Ga)

Primitive EF basaltic 507 070278 18 + 22 Sample EF 072magma Paslick et al (1993)Upper crust 300 071665 62 ndash83 Sample JG 02(JulianehaEcirc b Granite)Lower crust 747 070668 63 ndash 234 Taylor et al (1984)(Granulite facies gneiss)

The composition of the granulite-facies gneiss is an average of 5 samples with low RbSr ratios lt02

PETROGENESIS OF THE ERIKSFJORD BASALTS

847

the lower crust An in uence of Archaean rockson Gardar magmas has been demonstrated by Pbisotope studies (Taylor and Upton 1993) andseismic data were also interpreted to re ect theexistence of a wedge of Archaean crust thatextends southwards to Lindenow Fjord (Dahl-Jensen et al 1998) We have shown in theprevious sections on the basis of Ni trends thatfractionation of olivine was most relevant to themajor element evolution of the lavas whereasfractionation of other phases was apparently ofminor importance We therefore modelled theAFC processes with olivine fractionation usingmineral-melt partition coef cients of 0014 for Srand 00066 for Nd (Rollinson 1993)

The modelling of AFC processes is stronglydependent on the parameter r de ned as the ratioof mass assimilation rate and fractional crystal-lization rate (DePaolo 1981) The change inisotopic signature becomes enhanced for highervalues of r Reiners et al (1995) have shown thatr values might be very high and even gt1 in theearly stages of magmatic evolution at hightemperatures since crystallization can besuppressed by assimilation Therefore we calcu-

lated AFC processes with two r values of 05 and08 that represent a relatively high and a middle tolow assimilation rate to demonstrate the effects ofvariable r (Fig 9) The results indicate that AFCprocesses involving upper crust similar to theJulianehaEcirc b Granite cannot explain the trend of theEF basalts because Sr isotopic compositionswould shift to considerably more radiogenicvalues This result is consistent with theconclusions derived from trace element dataThe AFC curves for assimilation of lower crusthowever agree well with the bulk of the data Themost negative eNd values can be explained byAFC processes with intermediate assimilationrates (r = 05) involving assimilation of lt5lower crustal material With a greater degree ofassimilation and an r value of 08 lt4assimilation is indicated

Lower crustal rocks of igneous origin have anaverage d18O value of +75plusmn14 with an overallrange from +54 to +125 (Fowler and Harmon1990) 5 to 10 bulk assimilation of lower crustalmaterial with d18O = +75 into a typical mantle-derived magma with d18O = +57 results ind18O values of the contaminated magma from

-24

-20

-16

-12

-8

-4

0

4

0700 0703 0706 0709 0712 0715 0718

r = 08

r = 05

r = 05

r = 08

EF Basalts

Endmembers of AFC calculations

Upper Crust

Lower Crust

Primitive EF basaltic magma

(87Sr86Sr)i

e Nd

(i)

09

08

06

09

08

06

09

0806

09

08

07

07

04

Typical errorCHUR

BS

E

FIG 9 eNd(i) vs (87Sr86Sr)i diagram showing the results of AFC modelling End-member compositions are listed inTable 6 Olivine is the only fractionating mineral Partition coef cients for olivinemelt are from the compilation ofRollinson (1993) Curves were calculated for two values of r (r = rate of assimilation rate of fractionalcrystallization) Dashed lines and numbers in italics are for r = 05 solid lines and regular numbers are for r = 08

Values at the curves are for F values (fraction of melt remaining) with 10 tick marks

848

R HALAMA ETAL

+579 to +588 Therefore the d18O data cannotbe used to rule out small amounts of assimilationof lower crust To evaluate how much contamina-tion would be permissible from the presentoxygen isotope data we calculated how muchcontamination of a mantle-derived magma with18O = +57 with a contaminant within the lowerrange of lower crustal xenoliths (d18O = +67)would drive the oxygen isotopic composition tothe highest value of the EF basalts (d18O~+63) Although simple mixing shows that60 assimilation is permissible this amountwould drive the whole-rock geochemistry tomore silicic compositions than is observed in theEF basalts In Fig 8 there is no obvious trendfrom the basalts towards the Ketilidian basementsuggesting again that upper crust was notassimilated in signi cant amounts

Ratios of trace elements incompatible inbasaltic systems can be used to distinguishbetween crust and mantle reservoirs (egWeaver 1991) Based on the AFC modellingEF basalts with eNd(i) lt ndash1 are those which aremost likely in uenced by crustal contaminationThe relatively high LaNb and ZrNb ratios ofthese samples are consistent with assimilation oflower crustal material SrNd ratios are evenhigher than the average lower crustal composi-tion but SrNd ratios in those samples which weretaken as representative for the lower crust in theAFC modelling can reach values of 80 ie wellabove the highest EF basalt ratios Again theSrNd ratios support the hypothesis of some lowercrustal contamination

The absence of more primitive basaltic rockswith MgO gt8 wt might also indicate lowercrustal assimilation It suggests that magmas mayhave been retained at depth where they acquiredmore evolved compositions by fractional crystal-lization and had the opportunity to assimilatelower crustal material during crustal underplating(Upton et al 2003)

Summary and conclusions

The new geochemical data of the EriksfjordFormation basalts in the Gardar Provinceprovided in this study are used to placepetrogenetic constraints on the sources of theirparental magmas Based on trace-element patternsand Sr-Nd isotopic constraints there seems to beno indication that a DMM-like source played animportant role in magma genesis of the EFbasalts Clinopyroxene and whole-rock REE

patterns are more compatible with an OIB-likesource but we are not able to conclude whetherthe observed enrichment in REE re ects thesourcersquos signature or magmatic processes Theisotopically most primitive EF basalts have Sr-Ndisotopic compositions that are similar to Gardarcarbonatites Due to the isotopic similaritybetween carbonatites and OIBs many arethought to be related to mantle plumes (Bell2001) We therefore speculate that OIB-likemantle components from a plume source werealso involved in the parental EF magmas but wecannot exclude interaction with the SCLM Twodistinct mantle components one isotopicallydepleted and another one enriched could havecontributed to the EF magmas AFC modelling ofSr-Nd isotope data indicates that assimilation oflt5 lower crust could explain the steep vector onthe Sr-Nd isotope plot but assimilation of uppercrustal material is unlikely Oxygen isotope dataof clinopyroxene separates are consistent with amantle origin but permit some assimilation oflower crustal rocks In summary the data areconsistent with the hypothesis that the EFmagmas could have originated from a mantleplume source which is supported by geologicalarguments and that they underwent someassimilation of lower crustal material

Acknowledgements

We would like to thank Bruce Paterson whoprovided invaluable help during Laser ICP-MSmeasurements at the Large-Scale GeochemicalFacility supported by the European Community-Access to Research Infrastructure action of theImproving Human Potential Programme contractnumber HPRI-CT-1999-00008 awarded to ProfBJ Wood (University of Bristol) Gabi Stoschekand Torsten Vennemann are thanked for their helpwith oxygen isotope measurements Elmar Reitterexpertly assisted with preparation and measure-ments of radiogenic isotopes and MathiasWestphal helped with microprobe measurementsREE whole-rock analyses were kindly providedby the NERC facility at Royal HollowayUniversity of London Michael Marks is thankedfor his pleasant company during eldwork Veryconstructive reviews by Tom Andersen andAlexei S Rukhlov greatly helped to improve themanuscript We also thank Ian Coulson for nalediting Funding of this work by the DeutscheForschungsgemeinschaft (grant Ma-21351-2) isgratefully acknowledged

PETROGENESIS OF THE ERIKSFJORD BASALTS

849

References

Andersen T (1997) Age and petrogenesis of theQassiarsuk carbonatite-alkaline silicate volcaniccomplex in the Gardar rift South GreenlandMineralogical Magazine 61 499 ndash513

Armstrong JT (1991) Quantitative elemental analysisof individual microparticles with electron beaminstruments Pp 261 ndash315 in Electron ProbeQuantitation (KFJ Heinrich and DE Newburyeditors) Plenum Press New York and London

Arndt NT and Christensen U (1992) The role oflithospheric mantle in continental ood volcanismThermal and geochemical constraints Journal ofGeophysical Research 97 10967 ndash 10981

Baker JA Menzies MA Thirlwall MF andMacpherson CG (199 7) Pet rogenes is ofQuaternary Intraplate Vocanism Sanarsquoa YemenImplications for Plume-Lithosphere Interaction andPolybaric Melt Hybridization Journal of Petrology 38 1359 ndash 1390

Baker JA MacPherson CG Menzies MAThirlwall MF Al-Kadasi M and Mattey DP(2000) Resolving crustal and mantle contributions tocontinental ood volcanism Yemen constraintsfrom mineral oxygen isotope data Journal ofPetrology 41 1805 ndash1820

Beard JS Abitz RJ and Lofgren GE (1993)Experimental melting of crustal xenoliths fromKilbourne Hole New Mexico and implications forthe contamination and genesi s of magmasContributions to Mineralogy and Petrology 11588 ndash102

Bell K (2001) Carbonatites relationships to mantle-plume activity Geological Society of AmericaSpecial Paper 352 267 ndash290

Bell K and Blenkinsop J (1989) Neodymium andStrontium Isotope Geochemistry of Carbonatites Pp278 ndash300 in Carbonatites ndash Genesis and Evolution(K Bell editor) Unwin Hyman London

Bell K and Simonetti A (1996) Carbonat iteMagmatism and Plume Activity Implications fromthe Nd Pb and Sr Isotope Systematics of OldoinyoLengai Journal of Petrology 37 1321 ndash1339

Bell K and Tilton GR (2001) Nd Pb and Sr IsotopicCompositions of East African Carbonat itesEvide nce for Mant le Mixing and PlumeInhomogene ity Journa l of Petrology 42 1927 ndash1945

Benoit M Polve M and Ceuleneer G (1996) Traceelement and isotopic characterization of ma ccumulates in a fossil mantle diapir (Oman ophiolite)Chemical Geology 134 199 ndash 214

Bernstein S Kelemen PB Tegner C Kurz MDBlusztajn J and Kent Brooks C (1998) Post-breakup basaltic magmatism along the EastGreenland Tertiary rifted margin Earth and

Planetary Science Letters 160 845 ndash862Boynton WV (1984) Geochemistry of the rare earth

elements meteorite studies Pp 63 ndash114 in RareEarth Element Geochemistry (P Henderson editor)Elsevier Amsterdam

Carroll MR and Wyllie PJ (1990) The systemtonalite-H2O at 15 kbar and the genesis of calc-alkaline magmas American Mineralogist 75345 ndash 357

Chauvel C Hofmann AW and Vidal P (1992)HIMU-EM The French Polynesian connectionEarth and Planetary Science Letters 110 99 ndash119

Clayton RN and Mayeda TK (1963) The use ofbromine penta uoride in the extraction of oxygenfrom oxides and silicates for isotope analysisGeochimica et Cosmochimica Acta 27 43 ndash52

Coulson IM and Chambers AD (1996) Patterns ofzonation in rare-earth-bearing minerals in nephelinesyenites of the North Qoroq center South GreenlandThe Canadian Mineralogist 34 1163 ndash1178

Cox KG Bell JD and Pankhurst RJ (1979) TheInterpretation of Igneous Rocks George Allen andUnwin London

Cross W Iddings JP Pirsson LV and WashingtonHS (1903) Quantitative Classi cation of IgneousRocks University of Chicago Press

Dahl-Jensen T Thybo T Hopper H and Rosing M(1998) Crustal structure at the SE Greenland marginfrom wide-angle and normal incidence seismic dataTectonophysics 288 191 ndash 198

DePaolo DJ (1981) Trace element and isotopic effectsof combined wallrock assimilation and fractionalcrystallisation Earth and Planetary Science Letters53 189 ndash 202

De P a o lo D J (1 9 8 8 ) Neodymium Iso topeGeochemistry An Introduction Springer-VerlagNew York

Dobosi G Downes H Mattey D and Embey-IsztinA (1998) Oxygen isotope ratios of phenocrysts fromalkali basalts of the Pannonian basin Evidence foran O-isotopically homogeneous upper mantle be-neath a subduction-in uenced area Lithos 42213 ndash 223

Eiler JM (2001) Oxygen isotope variations of basalticlavas and upper mantle rocks Pp 319 ndash364 inStable Isotope Geochemistry (JW Valley and DRCole editor s) Reviews in Minera logy andGeochemistry 43 The Mineralogical Society ofAmerica Washington DC

Eiler JM Farley KA Valley JW Hauri E CraigH Hart SR and Stolper EM (1997) Oxygenisotope variations in ocean island basalt phenocrystsGeochimica et Cosmochimica Acta 61 2281 ndash2293

Eiler JM Crawford A Elliot T Farley KAValley JW and Stolper EM (2000a) Oxygenisotope geochemistry of oceanic-arc lavas Journal

850

R HALAMA ETAL

strated that analyses of mineral separates consis-tently span a narrower range than comparablewhole-rock data In our study we used clino-pyroxene because it is the only fresh mineralphase present in the samples Oxygen isotopemeasurements of clinopyroxene separates fromthe EF basalts show a range in d18Ocpx from +52to +60 with an average value of +566plusmn023(1s) All samples overlap the range of d18Ocpx

values from mantle peridotites (+525 to +590Mattey et al 1994) from continental oodbasalts from Yemen (+55 to +69 Baker etal 2000) and from alkaline basalts of Hungary(+507 to +534 Dobosi et al 1998) There is asmall difference between the averages ofMussartut Group (d18Ocpx-avg = +575plusmn015(1s)) and Ulukasik Group (d18Ocpx-avg =+575plusmn009 (1s)) samples compared toIl otildeacutemaussaq Group (d18Ocpx-avg = +533plusmn021(1s)) samples

d18O values of melts coexisting with clino-pyroxene were calculated assuming that oxygen

in pyroxene and melt were in isotopic equilibriumand T was 1100ordmC Dmelt-cpx can then be calculatedusing the following equation of Kalamarides(1986)

Dmelt-cpx = ndash00008T + 143 (T in Kelvin)

Although fractionation varies with magmatictemperatures we consider the temperature of1100ordmC and the resulting Dmelt-cpx of +03 assuf ciently representative considering the analy-tical errors involved It is similar to Dmelt-cpx

values used in previous studies (+039 Vroonet al 2001 +02 Dobosi et al 1998)Resulting d18Omelt values of EF basalts rangefrom +55 to +63 These values can becompared to whole-rock data from continentalbasalts given by Harmon and Hoefs (1995)Intraplate basalts from continental rift zonesshow a range from +46 to +83 with a meanof +61 and continental ood basalts which arerather poorly represented in the database havevalues between +43 and +65 Calculated EF

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Cpx from MORB-like

olivine gabbro

Cpx megacrysts from OIB-like

alkaline basalts

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 024EF 039

Mussartucirct Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

1

10

100

1000

EF 168

Iliacutemaussaq Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

1

10

100

1000

EF 072

Ulukasik Group - clinopyroxene data

c (sa

mpl

e)c

(cho

ndrit

e)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb LuLa Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

c

ba

d

1

10

100

01

FIG 6 REE patterns of clinopyroxenes (cpx) from Mussartut (a) Ulukasik (b) and Il otilde maussaq (c) Group samplesnormalized to chondritic values from Boynton (1984) (d) REE patterns of cpx from MORB-type olivine gabbro(Benoit et al 1996) and OIB-type alkaline basalt (Nimis and Vannucci 1995) for comparison (note the change of

scale)

842

R HALAMA ETAL

TA

BL

E4

Sr

Nd

and

Ois

otop

icco

mpo

siti

ons

ofcl

inop

yrox

ene

sepa

rate

sfr

omE

riks

fjor

dF

orm

atio

nba

salt

sW

hole

-roc

kan

alys

esfr

omth

eK

etil

idia

nba

sem

ent

and

from

sedi

men

tary

rock

sof

the

Eri

ksfj

ord

For

mat

ion

are

give

nfo

rco

mpa

riso

n

Sam

ple

Mem

ber

Sr

Rb

87R

b86S

r87S

r86Sr

87Sr

86Sr

init

ial

SmN

d147Sm

144N

d143N

d144N

de N

d(i

)d1

8O

(ppm

)(p

pm)

(ppm

)(p

pm)

Bas

alts

E

F02

4M

ussa

rtut

646

50

486

002

180

7042

01plusmn

100

7038

267

3822

80

019

560

5124

67plusmn

10ndash

32

567

EF

063

Mus

sart

ut5

60plusmn

011

EF

061

Mus

sart

ut89

95

099

90

0321

070

3818

plusmn10

070

3266

574

163

90

2115

051

2629

plusmn10

ndash2

55

79plusmn

028

EF

039

Mus

sart

ut73

74

058

30

0229

070

3492

plusmn10

070

3098

662

191

30

2092

051

2626

plusmn10

ndash2

25

71plusmn

001

EF

059

Mus

sart

ut54

47

013

70

0073

070

3375

plusmn12

070

3250

606

168

90

2171

051

2727

plusmn09

ndash1

45

99plusmn

001

EF

108

Ulu

kasi

k66

92

111

00

0480

070

3741

plusmn09

070

2916

109

936

82

018

050

5126

17plusmn

10+

21

581

EF

072

Ulu

kasi

k69

41

032

20

0134

070

3014

plusmn10

070

2784

818

251

90

1962

051

2574

plusmn10

ndash1

25

68E

F17

4Il

otildemau

ssaq

125

42

324

005

360

7039

35plusmn

100

7030

1418

76

796

10

1424

051

2230

plusmn10

+0

45

47E

F16

8Il

otildemau

ssaq

866

71

422

004

750

7037

99plusmn

070

7029

8316

45

641

60

1550

051

2304

plusmn09

ndash0

15

18

Bas

emen

tan

dse

dim

enta

ryro

cks

JG02

Juli

aneh

aEcirc b30

01

169

71

6375

074

4789

plusmn10

071

6647

105

362

28

010

220

5114

70plusmn

09ndash

83

78

Gra

nite

EF

021

Qua

rtzi

te11

5E

F02

8A

rkos

ic6

8ar

enit

e

Sam

ples

are

list

edin

stra

tigra

phic

orde

rfr

omol

dest

toyo

unge

st87

Sr86

Sr

init

ialr

atio

san

de N

d(i)

valu

esw

ere

calc

ulat

edfo

ran

age

of1

2G

aS

tand

ard

devi

atio

nsfo

rd18

Oar

egi

ven

for

sam

ples

that

wer

ean

alys

edtw

ice

PETROGENESIS OF THE ERIKSFJORD BASALTS

843

model melt oxygen isotope data are within theranges of both groups Compared to Igaliko dykerocks from further SE in the Gardar Province(Pearce and Leng 1996) the EF basalts cover amuch more restricted eld possibly because d18Owhole-rock analyses are more prone to secondaryalteration than analyses of mineral separates(Fig 8) Whole-rock d18O measurements of theunderlying JulianehaEcirc b Granite and of the sedi-mentary sandstone members of the EriksfjordFormation yield values of 78 and 68 ndash115respectively (Table 4) and are shown forcomparison

Discussion

Mantle source regions of the EF basaltsTo reconstruct the mantle sources involved in theformation of the EF basalts we use Sr-Nd-Oisotope and incompatible trace element data On aSr-Nd isotope diagram the isotopic compositionof DMM was modelled for an age of 12 Ga(DePaolo 1981 Goldstein et al 1984 Taylor andMcLennan 1985) The fairly constant 87Sr86Srinitial isotope ratios of the EF basalts are ~0703which falls within the range typical for mantle-derived rocks but slightly above calculated valuesfor DMM (Fig 7) eNd(i) values are more

-4

-2

0

2

4

6

8

0701 0702 0703 0704 0705 0706

DMM Typical 2 s error

CHUR

BS

E

(87Sr86Sr)i

e Nd

(i)

EM-1

Qassiarsuk carbonatites and lamprophyres

EF basalts

Depleted MORB mantle

Ivittuut basaltic and lamprophyric dikes

Igaliko carbonatites and lamprophyres

Enriched mantle 1

Carbonatites from other regions of similar age

FIG 7 eNd(i) vs (87Sr86Sr)i diagram calculated for t = 12 Ga which is considered as the eruption age of the EFbasalts (Paslick et al 1993) The reference line of Bulk Silicate Earth (BSE) for the Sr isotopic composition wascalculated after DePaolo (1988) assuming present-day values of 00827 for 87Rb86Sr and 07045 for 87Sr86SrIsotopic composition of the depleted MORB mantle (DMM) was calculated after DePaolo (1981) Goldstein et al(1984) and Taylor and McLennan (1985) EM-1 was calculated after Schleicher et al (1998) (see text for details)Data sources are Goodenough et al (2002) Ivittuut (n = 6) Pearce and Leng (1996) Igaliko (n = 5) Andersen(1997) Qassiarsuk (n = 25) Bell and Blenkinsop (1989) other carbonatites of similar age (Goudini Schryburt Lake

and Prairie Lake n = 6)

844

R HALAMA ETAL

scattered and indicate that the rocks contain amixture of isotopic components from both mantleand crustal reservoirs or a heterogeneous mantlesource Only one sample (EF 108) has an eNd(i)value signi cantly above 0 This sample exhibitsthe largest spread of clinopyroxene compositionswhich may re ect strong alteration Goodenoughet al (2002) also note that the highest eNd occursin the most altered sample of their sample setHowever Paslick et al (1993) report clino-pyroxene data from two Ulukasik Groupsamples with eNd(i) gt +2 and those sampleswere apparently not disturbed with regard to theirSm-Nd age information On the other hand Piperet al (1999) suggest that the Sm-Nd systematicsof those samples may be erroneous due to laterdisturbance by uids derived from the Il otildeacutemaussaqintrusion Therefore although we are presentlynot able to decide whether positive eNd(i) values

for the EF basalts are real it is clear that none ofthose values is close to values of a 12 Ga oldDMM source (ie eNd = +5 to +7)

The Sr-Nd isotopic composition for OIB-typesources in the Proterozoic is not as wellconstrained as DMM Today OIB-type mantlesources have a wide range of isotopic composi-tions that partly overlap with DMM but eNd

values are generally lower (Hart et al 1992Hofmann 1997) The EM-1 mantle component ischaracterized by extremely low 143Nd144Ndratios and eNd values of between 0 and ndash5(Hofmann 1997) A possible 12 Ga position forEM-1 is plotted in Fig 7 calculated according toSchleicher et al (1998) with 147Sm144Nd = 018and 87Rb86Sr = 010 The present-day values usedin the calculation are 143Nd144Nd = 051234 and87Sr86Sr = 07057 taken from ocean islandbasalts closest to an EM-1 end-member(Hofmann 1997) Other Sr-Nd isotope datafrom the Gardar Province are plotted to putfurther constraints on the mantle sources of the EFbasalts (Fig 7) We also added comparative datafrom African and Canadian carbonatitesof similarage because there is a well known similaritybetween Sr-Nd isotopic data of OIBs andcarbonatites (Bell and Blenkinsop 1989) Thelinear array of the carbonatite data is consideredto be the result of mixing between the two mantlecomponents HIMU and EM-1 (Bell andBlenkinsop 1989 Bell and Tilton 2001) Thecarbonatites from the Gardar Province havesimilar Nd but slightly more radiogenic Srisotopic compositions than the reference carbo-natite data (Fig 7) The least contaminatedsamples overlap with the linear array and themore radiogenic Sr isotopic composition can beexplained by crustal contamination (Andersen1997) We therefore consider it possible that theGardar carbonatites also represent mixing ofhypothetical HIMU and EM-1 components of anOIB-like source Since the isotopically mostprimitive EF basalt samples overlap with theQassiarsuk carbonatites it could be possible thatthe basalts also contain such mantle componentsalthough Pb isotope data for the EF basalts thatcould clarify the importance of a HIMU sourceare lacking Based on the Sr-Nd isotope data theEF basalts appear to be derived from aheterogeneous OIB-like mantle source thatcomprises isotopically depleted and enrichedcomponents OIB-like and carbonatitic magma-tism are often attributed to mantle plume activity(Bell and Simonetti 1996 Hofmann 1997

2

4

6

8

10

12

-10 -8 -6 -4 -2 0 2 4 6 8

DMM

d18 O

[permil]

e Nd (i)

Typical2s error

EM-1

EF basalts - cpx

EF basalts - model melt

Ketilidian basement

Igaliko dykes

Depleted MORB mantle

Enriched mantle 1

EM-1 lavas - cpx

FIG 8 Oxygen isotopi c compositio ns (d18OV-SMOW) of clinopyroxene from EF basalts calculatedmelt (see text for details) and whole-rock Ketilidianbasement vs eNd(i) Whole-rock data for Igaliko dykerocks (n = 7) are from Pearce and Leng (1996) DMMdata derived from fresh MORB glasses are from Eiler etal (2000b) The eld of EM-1 lavas is taken fromclinopyroxene data of Gough and Tristan da CunhaIsland lavas (Harris et al 2000) and the EM-1 end-member composition is based on the calculations afterSchleicher et al (1998) for eNd(i) and an average value

of 578 (n = 18) from the EM-1 lavas for d18O

PETROGENESIS OF THE ERIKSFJORD BASALTS

845

Simonetti et al 1998) A plume input to theformation of the EF basalts is supported by therelatively large volume of magmas and the shortduration (lt5 Ma) of magmatism (Piper et al1999) It is also compatible with the geologicalsetting of a continental breakup Additionally aplume component appears to be necessary toexplain the thermal input as it is dif cult to melt acold lithosphere section (Arndt and Christensen1992)

Oxygen isotope data of the EF basalts aregenerally consistent with a mantle derivationincluding an OIB-like mantle source On the d18Ovs eNd(i) diagram (Fig 8) the overlap betweenthe EF data clinopyroxene data from EM-1-likeoceanic basalts and the hypothetical EM-1 end-member suggests an involvement of an EM-1-type mantle source Olivines from HIMU-likesources are known to be slightly lower in d18Othan typical upper mantle values from MORBs(Eiler et al 1997) but the EF data are notsuf cient to assess if the slightly lower d18Ovalues in the Il otildeacutemaussaq Group basalts are causedby a relatively higher input from a HIMU source

The contribution of the various mantle sourcesto the EF basalts can also be evaluated on theprimitive mantle-normalized multi-elementdiagram (Fig 4) and on the REE plots (Figs 5and 6) Both the whole-rock REE and theclinopyroxene REE patterns of the EF basaltsare unlike those representative of N-MORBsources We conclude again that DMM did notplay a major role in the petrogenesis of the EFbasalts On the other hand similarities with OIB-like mantle sources are more prominent Theshape of the clinopyroxene REE pattern closelyresembles those from clinopyroxene of OIB-likealkaline basalts despite differences in absoluteconcentrations The whole-rock REE patterns ofthe EF basalts lie between E-MORB and averageOIB patterns as do the multi-element patterns Onthe latter the EF basalts have marked positive Baand P peaks that are not present in typical OIBsHowever Ba contents should be interpreted withcaution since highly variable ZrBa ratios from014 to 070 are an indication that Ba was re-mobilized after emplacement of the basalts (egPrice et al 1991 Oliveira and Tarney 1995)Likewise P mobility was demonstrated in alteredbasaltic rocks (Winchester and Floyd 1976) Insummary the REE and trace element data areevidence that the mantle source(s) of the EFbasalts were enriched in incompatible traceelements relative to primitive mantle

For relatively primitive hy-normative basalticrocks of the Ivittuut area further north in theGardar Province it has been argued that some ofthem were derived from a lithospheric mantlesource previously enriched by subduction-related processes This interpretation is basedon certain geochemical characteristics such ashigh LaNb ratios and negative Nb anomalies(Goodenough et al 2002) For the EF basaltsnegative Nb anomalies are not clearly discern-ible but LaNb ratios are gt1 which seems to be acharacteristic feature of a lithospheric mantlesource (Fitton 1995) LaBa LaRb and LaKratios were also used to argue for a lithosphericorigin of the Gardar basic magmas although amantle plume involvement was assumed toprovide the required energy (Upton andEmeleus 1987 Upton et al 2003) Sincefeldspar is the only primary magmatic mineralin the EF basalts in which Ba Rb and K areincorporated in signi cant amounts and thefeldspars are generally altered these elementsmight have been easily mobilized and wetherefore do not dare to base the interpretationof the EF basalts sources on these data Inaddition the lack of precise geochemical data onthe composition of the lithosphere below theGardar Province complicates an evaluation ofthe lithospheric contribution to the EF basalticmagmatism

Certain ratios of incompatible trace elements ofthe EF basalts are compared to several occur-rences of ocean island basalts representative ofEM-1 and HIMU sources in Table 5 Ratiosbetween highly incompatible trace elements donot change signi cantly during partial melting orfractional crystallization in basaltic systems andmay therefore constrain the ratios in their sources(Weaver et al 1987) The EF basalts aresubdivided into the more alkaline isotopicallyprimitive Il otildeacutemaussaq Group samples and thosesamples with eNd(i) lt ndash1 The Il otilde maussaq Groupsamples have higher SrNd LaNb and ZrNbratios than HIMU and overlap with the upperrange of EM-1 ratios The EF basalts with eNd(i)lt ndash1 have generally higher ratios than the oceanicisland basalts and partly overlap with typicalvalues of the lower crust It therefore seems likelythat assimilation of lower crustal material mayhave changed the primary trace element ratios inthe basalts Therefore we will evaluate thepossible in uence of assimilation of crustalmaterial on the geochemistry of the EF basaltsin the following section

846

R HALAMA ETAL

The role of crustal contamination

Large amounts of bulk assimilation of crustalmaterial which is relatively SiO2-rich comparedto a basaltic melt would drive the compositions tohigher SiO2 contents and quartz-normativecompositions Partial melts derived from crustalrocks would also be relatively SiO2-rich (Beard etal 1993 Carroll and Wyllie 1990) and wouldhave the same effect The ne-normative characterof the EF basalts (Table 1) suggests thatassimilation of crustal material especially ofsilica-rich upper crustal rocks was not signi cantwith the possible exception of sample EF 059

On the primitive mantle-normalized multi-element diagrams (Fig 4) Ti troughs are typicalfor crustal rocks especially for the Upper Crust(Rudnick and Fountain 1995) and they maytherefore be present in rocks extensively contami-nated by continental crust Ti troughs aregenerally lacking in the patterns of the EFbasalts and a major upper crustal contribution tothe magmas can therefore be excluded However

small amounts of assimilation of lower crustalmaterial cannot be ruled out because the averagelower crust has higher Ti contents

Steep trends on Sr-Nd isotope plots have beenattributed to assimilation of granulite-faciesgneisses (eg Bernstein et al 1998) AFCmodelling based on an equation developed byDePaolo (1981) was performed to test thehypothesis of small amounts of contaminationwith lower crustal material (Fig 9) The basalticmagma which represents one end-member for thecalculations has the isotopic signature of theisotopically most primitive samples from thepresent data set and from the literature(Table 6) Two possible contaminants repre-senting upper and lower crustal compositionswere selected for modelling (Table 6) Therepresentative samples are a granite from theJulianehaEcirc b batholith for the upper crust (sampleJG 02) and an average of ve granulite-faciesgneisses with low RbSr ratios from the Archaeancraton of West Greenland (Taylor et al 1984) for

TABLE 5 Trace element ratios of EF basalts compared to ocean island basalts(OIBs) representing EM-1 and HIMU mantle components and the lowercrust

SrNd LaNb ZrNb

EF basalts with eNd lt ndash 1 249 ndash396 120 ndash150 121 ndash157Il otildeacutemaussaq Group EF basalts 198 ndash292 104 ndash119 51 ndash53EM-1 90 ndash207 067 ndash111 39 ndash109HIMU 45 -163 063 ndash080 27 ndash 36Lower continental crust 30 16 136

Data sources EM-1 Humphris and Thompson (1983 Walvis Ridge) and Cliff et al(1991 Inaccessible Islands) HIMU Chauvel et al (1992 Tubuai Island) Lowercontinental crust Rudnick and Fountain (1995)

TABLE 6 End-member compositions used for AFC modelling

End-member Sr (87Sr86Sr)i Nd eNd(i) Data source(ppm) (T = 12 Ga) (ppm) (T = 12 Ga)

Primitive EF basaltic 507 070278 18 + 22 Sample EF 072magma Paslick et al (1993)Upper crust 300 071665 62 ndash83 Sample JG 02(JulianehaEcirc b Granite)Lower crust 747 070668 63 ndash 234 Taylor et al (1984)(Granulite facies gneiss)

The composition of the granulite-facies gneiss is an average of 5 samples with low RbSr ratios lt02

PETROGENESIS OF THE ERIKSFJORD BASALTS

847

the lower crust An in uence of Archaean rockson Gardar magmas has been demonstrated by Pbisotope studies (Taylor and Upton 1993) andseismic data were also interpreted to re ect theexistence of a wedge of Archaean crust thatextends southwards to Lindenow Fjord (Dahl-Jensen et al 1998) We have shown in theprevious sections on the basis of Ni trends thatfractionation of olivine was most relevant to themajor element evolution of the lavas whereasfractionation of other phases was apparently ofminor importance We therefore modelled theAFC processes with olivine fractionation usingmineral-melt partition coef cients of 0014 for Srand 00066 for Nd (Rollinson 1993)

The modelling of AFC processes is stronglydependent on the parameter r de ned as the ratioof mass assimilation rate and fractional crystal-lization rate (DePaolo 1981) The change inisotopic signature becomes enhanced for highervalues of r Reiners et al (1995) have shown thatr values might be very high and even gt1 in theearly stages of magmatic evolution at hightemperatures since crystallization can besuppressed by assimilation Therefore we calcu-

lated AFC processes with two r values of 05 and08 that represent a relatively high and a middle tolow assimilation rate to demonstrate the effects ofvariable r (Fig 9) The results indicate that AFCprocesses involving upper crust similar to theJulianehaEcirc b Granite cannot explain the trend of theEF basalts because Sr isotopic compositionswould shift to considerably more radiogenicvalues This result is consistent with theconclusions derived from trace element dataThe AFC curves for assimilation of lower crusthowever agree well with the bulk of the data Themost negative eNd values can be explained byAFC processes with intermediate assimilationrates (r = 05) involving assimilation of lt5lower crustal material With a greater degree ofassimilation and an r value of 08 lt4assimilation is indicated

Lower crustal rocks of igneous origin have anaverage d18O value of +75plusmn14 with an overallrange from +54 to +125 (Fowler and Harmon1990) 5 to 10 bulk assimilation of lower crustalmaterial with d18O = +75 into a typical mantle-derived magma with d18O = +57 results ind18O values of the contaminated magma from

-24

-20

-16

-12

-8

-4

0

4

0700 0703 0706 0709 0712 0715 0718

r = 08

r = 05

r = 05

r = 08

EF Basalts

Endmembers of AFC calculations

Upper Crust

Lower Crust

Primitive EF basaltic magma

(87Sr86Sr)i

e Nd

(i)

09

08

06

09

08

06

09

0806

09

08

07

07

04

Typical errorCHUR

BS

E

FIG 9 eNd(i) vs (87Sr86Sr)i diagram showing the results of AFC modelling End-member compositions are listed inTable 6 Olivine is the only fractionating mineral Partition coef cients for olivinemelt are from the compilation ofRollinson (1993) Curves were calculated for two values of r (r = rate of assimilation rate of fractionalcrystallization) Dashed lines and numbers in italics are for r = 05 solid lines and regular numbers are for r = 08

Values at the curves are for F values (fraction of melt remaining) with 10 tick marks

848

R HALAMA ETAL

+579 to +588 Therefore the d18O data cannotbe used to rule out small amounts of assimilationof lower crust To evaluate how much contamina-tion would be permissible from the presentoxygen isotope data we calculated how muchcontamination of a mantle-derived magma with18O = +57 with a contaminant within the lowerrange of lower crustal xenoliths (d18O = +67)would drive the oxygen isotopic composition tothe highest value of the EF basalts (d18O~+63) Although simple mixing shows that60 assimilation is permissible this amountwould drive the whole-rock geochemistry tomore silicic compositions than is observed in theEF basalts In Fig 8 there is no obvious trendfrom the basalts towards the Ketilidian basementsuggesting again that upper crust was notassimilated in signi cant amounts

Ratios of trace elements incompatible inbasaltic systems can be used to distinguishbetween crust and mantle reservoirs (egWeaver 1991) Based on the AFC modellingEF basalts with eNd(i) lt ndash1 are those which aremost likely in uenced by crustal contaminationThe relatively high LaNb and ZrNb ratios ofthese samples are consistent with assimilation oflower crustal material SrNd ratios are evenhigher than the average lower crustal composi-tion but SrNd ratios in those samples which weretaken as representative for the lower crust in theAFC modelling can reach values of 80 ie wellabove the highest EF basalt ratios Again theSrNd ratios support the hypothesis of some lowercrustal contamination

The absence of more primitive basaltic rockswith MgO gt8 wt might also indicate lowercrustal assimilation It suggests that magmas mayhave been retained at depth where they acquiredmore evolved compositions by fractional crystal-lization and had the opportunity to assimilatelower crustal material during crustal underplating(Upton et al 2003)

Summary and conclusions

The new geochemical data of the EriksfjordFormation basalts in the Gardar Provinceprovided in this study are used to placepetrogenetic constraints on the sources of theirparental magmas Based on trace-element patternsand Sr-Nd isotopic constraints there seems to beno indication that a DMM-like source played animportant role in magma genesis of the EFbasalts Clinopyroxene and whole-rock REE

patterns are more compatible with an OIB-likesource but we are not able to conclude whetherthe observed enrichment in REE re ects thesourcersquos signature or magmatic processes Theisotopically most primitive EF basalts have Sr-Ndisotopic compositions that are similar to Gardarcarbonatites Due to the isotopic similaritybetween carbonatites and OIBs many arethought to be related to mantle plumes (Bell2001) We therefore speculate that OIB-likemantle components from a plume source werealso involved in the parental EF magmas but wecannot exclude interaction with the SCLM Twodistinct mantle components one isotopicallydepleted and another one enriched could havecontributed to the EF magmas AFC modelling ofSr-Nd isotope data indicates that assimilation oflt5 lower crust could explain the steep vector onthe Sr-Nd isotope plot but assimilation of uppercrustal material is unlikely Oxygen isotope dataof clinopyroxene separates are consistent with amantle origin but permit some assimilation oflower crustal rocks In summary the data areconsistent with the hypothesis that the EFmagmas could have originated from a mantleplume source which is supported by geologicalarguments and that they underwent someassimilation of lower crustal material

Acknowledgements

We would like to thank Bruce Paterson whoprovided invaluable help during Laser ICP-MSmeasurements at the Large-Scale GeochemicalFacility supported by the European Community-Access to Research Infrastructure action of theImproving Human Potential Programme contractnumber HPRI-CT-1999-00008 awarded to ProfBJ Wood (University of Bristol) Gabi Stoschekand Torsten Vennemann are thanked for their helpwith oxygen isotope measurements Elmar Reitterexpertly assisted with preparation and measure-ments of radiogenic isotopes and MathiasWestphal helped with microprobe measurementsREE whole-rock analyses were kindly providedby the NERC facility at Royal HollowayUniversity of London Michael Marks is thankedfor his pleasant company during eldwork Veryconstructive reviews by Tom Andersen andAlexei S Rukhlov greatly helped to improve themanuscript We also thank Ian Coulson for nalediting Funding of this work by the DeutscheForschungsgemeinschaft (grant Ma-21351-2) isgratefully acknowledged

PETROGENESIS OF THE ERIKSFJORD BASALTS

849

References

Andersen T (1997) Age and petrogenesis of theQassiarsuk carbonatite-alkaline silicate volcaniccomplex in the Gardar rift South GreenlandMineralogical Magazine 61 499 ndash513

Armstrong JT (1991) Quantitative elemental analysisof individual microparticles with electron beaminstruments Pp 261 ndash315 in Electron ProbeQuantitation (KFJ Heinrich and DE Newburyeditors) Plenum Press New York and London

Arndt NT and Christensen U (1992) The role oflithospheric mantle in continental ood volcanismThermal and geochemical constraints Journal ofGeophysical Research 97 10967 ndash 10981

Baker JA Menzies MA Thirlwall MF andMacpherson CG (199 7) Pet rogenes is ofQuaternary Intraplate Vocanism Sanarsquoa YemenImplications for Plume-Lithosphere Interaction andPolybaric Melt Hybridization Journal of Petrology 38 1359 ndash 1390

Baker JA MacPherson CG Menzies MAThirlwall MF Al-Kadasi M and Mattey DP(2000) Resolving crustal and mantle contributions tocontinental ood volcanism Yemen constraintsfrom mineral oxygen isotope data Journal ofPetrology 41 1805 ndash1820

Beard JS Abitz RJ and Lofgren GE (1993)Experimental melting of crustal xenoliths fromKilbourne Hole New Mexico and implications forthe contamination and genesi s of magmasContributions to Mineralogy and Petrology 11588 ndash102

Bell K (2001) Carbonatites relationships to mantle-plume activity Geological Society of AmericaSpecial Paper 352 267 ndash290

Bell K and Blenkinsop J (1989) Neodymium andStrontium Isotope Geochemistry of Carbonatites Pp278 ndash300 in Carbonatites ndash Genesis and Evolution(K Bell editor) Unwin Hyman London

Bell K and Simonetti A (1996) Carbonat iteMagmatism and Plume Activity Implications fromthe Nd Pb and Sr Isotope Systematics of OldoinyoLengai Journal of Petrology 37 1321 ndash1339

Bell K and Tilton GR (2001) Nd Pb and Sr IsotopicCompositions of East African Carbonat itesEvide nce for Mant le Mixing and PlumeInhomogene ity Journa l of Petrology 42 1927 ndash1945

Benoit M Polve M and Ceuleneer G (1996) Traceelement and isotopic characterization of ma ccumulates in a fossil mantle diapir (Oman ophiolite)Chemical Geology 134 199 ndash 214

Bernstein S Kelemen PB Tegner C Kurz MDBlusztajn J and Kent Brooks C (1998) Post-breakup basaltic magmatism along the EastGreenland Tertiary rifted margin Earth and

Planetary Science Letters 160 845 ndash862Boynton WV (1984) Geochemistry of the rare earth

elements meteorite studies Pp 63 ndash114 in RareEarth Element Geochemistry (P Henderson editor)Elsevier Amsterdam

Carroll MR and Wyllie PJ (1990) The systemtonalite-H2O at 15 kbar and the genesis of calc-alkaline magmas American Mineralogist 75345 ndash 357

Chauvel C Hofmann AW and Vidal P (1992)HIMU-EM The French Polynesian connectionEarth and Planetary Science Letters 110 99 ndash119

Clayton RN and Mayeda TK (1963) The use ofbromine penta uoride in the extraction of oxygenfrom oxides and silicates for isotope analysisGeochimica et Cosmochimica Acta 27 43 ndash52

Coulson IM and Chambers AD (1996) Patterns ofzonation in rare-earth-bearing minerals in nephelinesyenites of the North Qoroq center South GreenlandThe Canadian Mineralogist 34 1163 ndash1178

Cox KG Bell JD and Pankhurst RJ (1979) TheInterpretation of Igneous Rocks George Allen andUnwin London

Cross W Iddings JP Pirsson LV and WashingtonHS (1903) Quantitative Classi cation of IgneousRocks University of Chicago Press

Dahl-Jensen T Thybo T Hopper H and Rosing M(1998) Crustal structure at the SE Greenland marginfrom wide-angle and normal incidence seismic dataTectonophysics 288 191 ndash 198

DePaolo DJ (1981) Trace element and isotopic effectsof combined wallrock assimilation and fractionalcrystallisation Earth and Planetary Science Letters53 189 ndash 202

De P a o lo D J (1 9 8 8 ) Neodymium Iso topeGeochemistry An Introduction Springer-VerlagNew York

Dobosi G Downes H Mattey D and Embey-IsztinA (1998) Oxygen isotope ratios of phenocrysts fromalkali basalts of the Pannonian basin Evidence foran O-isotopically homogeneous upper mantle be-neath a subduction-in uenced area Lithos 42213 ndash 223

Eiler JM (2001) Oxygen isotope variations of basalticlavas and upper mantle rocks Pp 319 ndash364 inStable Isotope Geochemistry (JW Valley and DRCole editor s) Reviews in Minera logy andGeochemistry 43 The Mineralogical Society ofAmerica Washington DC

Eiler JM Farley KA Valley JW Hauri E CraigH Hart SR and Stolper EM (1997) Oxygenisotope variations in ocean island basalt phenocrystsGeochimica et Cosmochimica Acta 61 2281 ndash2293

Eiler JM Crawford A Elliot T Farley KAValley JW and Stolper EM (2000a) Oxygenisotope geochemistry of oceanic-arc lavas Journal

850

R HALAMA ETAL

TA

BL

E4

Sr

Nd

and

Ois

otop

icco

mpo

siti

ons

ofcl

inop

yrox

ene

sepa

rate

sfr

omE

riks

fjor

dF

orm

atio

nba

salt

sW

hole

-roc

kan

alys

esfr

omth

eK

etil

idia

nba

sem

ent

and

from

sedi

men

tary

rock

sof

the

Eri

ksfj

ord

For

mat

ion

are

give

nfo

rco

mpa

riso

n

Sam

ple

Mem

ber

Sr

Rb

87R

b86S

r87S

r86Sr

87Sr

86Sr

init

ial

SmN

d147Sm

144N

d143N

d144N

de N

d(i

)d1

8O

(ppm

)(p

pm)

(ppm

)(p

pm)

Bas

alts

E

F02

4M

ussa

rtut

646

50

486

002

180

7042

01plusmn

100

7038

267

3822

80

019

560

5124

67plusmn

10ndash

32

567

EF

063

Mus

sart

ut5

60plusmn

011

EF

061

Mus

sart

ut89

95

099

90

0321

070

3818

plusmn10

070

3266

574

163

90

2115

051

2629

plusmn10

ndash2

55

79plusmn

028

EF

039

Mus

sart

ut73

74

058

30

0229

070

3492

plusmn10

070

3098

662

191

30

2092

051

2626

plusmn10

ndash2

25

71plusmn

001

EF

059

Mus

sart

ut54

47

013

70

0073

070

3375

plusmn12

070

3250

606

168

90

2171

051

2727

plusmn09

ndash1

45

99plusmn

001

EF

108

Ulu

kasi

k66

92

111

00

0480

070

3741

plusmn09

070

2916

109

936

82

018

050

5126

17plusmn

10+

21

581

EF

072

Ulu

kasi

k69

41

032

20

0134

070

3014

plusmn10

070

2784

818

251

90

1962

051

2574

plusmn10

ndash1

25

68E

F17

4Il

otildemau

ssaq

125

42

324

005

360

7039

35plusmn

100

7030

1418

76

796

10

1424

051

2230

plusmn10

+0

45

47E

F16

8Il

otildemau

ssaq

866

71

422

004

750

7037

99plusmn

070

7029

8316

45

641

60

1550

051

2304

plusmn09

ndash0

15

18

Bas

emen

tan

dse

dim

enta

ryro

cks

JG02

Juli

aneh

aEcirc b30

01

169

71

6375

074

4789

plusmn10

071

6647

105

362

28

010

220

5114

70plusmn

09ndash

83

78

Gra

nite

EF

021

Qua

rtzi

te11

5E

F02

8A

rkos

ic6

8ar

enit

e

Sam

ples

are

list

edin

stra

tigra

phic

orde

rfr

omol

dest

toyo

unge

st87

Sr86

Sr

init

ialr

atio

san

de N

d(i)

valu

esw

ere

calc

ulat

edfo

ran

age

of1

2G

aS

tand

ard

devi

atio

nsfo

rd18

Oar

egi

ven

for

sam

ples

that

wer

ean

alys

edtw

ice

PETROGENESIS OF THE ERIKSFJORD BASALTS

843

model melt oxygen isotope data are within theranges of both groups Compared to Igaliko dykerocks from further SE in the Gardar Province(Pearce and Leng 1996) the EF basalts cover amuch more restricted eld possibly because d18Owhole-rock analyses are more prone to secondaryalteration than analyses of mineral separates(Fig 8) Whole-rock d18O measurements of theunderlying JulianehaEcirc b Granite and of the sedi-mentary sandstone members of the EriksfjordFormation yield values of 78 and 68 ndash115respectively (Table 4) and are shown forcomparison

Discussion

Mantle source regions of the EF basaltsTo reconstruct the mantle sources involved in theformation of the EF basalts we use Sr-Nd-Oisotope and incompatible trace element data On aSr-Nd isotope diagram the isotopic compositionof DMM was modelled for an age of 12 Ga(DePaolo 1981 Goldstein et al 1984 Taylor andMcLennan 1985) The fairly constant 87Sr86Srinitial isotope ratios of the EF basalts are ~0703which falls within the range typical for mantle-derived rocks but slightly above calculated valuesfor DMM (Fig 7) eNd(i) values are more

-4

-2

0

2

4

6

8

0701 0702 0703 0704 0705 0706

DMM Typical 2 s error

CHUR

BS

E

(87Sr86Sr)i

e Nd

(i)

EM-1

Qassiarsuk carbonatites and lamprophyres

EF basalts

Depleted MORB mantle

Ivittuut basaltic and lamprophyric dikes

Igaliko carbonatites and lamprophyres

Enriched mantle 1

Carbonatites from other regions of similar age

FIG 7 eNd(i) vs (87Sr86Sr)i diagram calculated for t = 12 Ga which is considered as the eruption age of the EFbasalts (Paslick et al 1993) The reference line of Bulk Silicate Earth (BSE) for the Sr isotopic composition wascalculated after DePaolo (1988) assuming present-day values of 00827 for 87Rb86Sr and 07045 for 87Sr86SrIsotopic composition of the depleted MORB mantle (DMM) was calculated after DePaolo (1981) Goldstein et al(1984) and Taylor and McLennan (1985) EM-1 was calculated after Schleicher et al (1998) (see text for details)Data sources are Goodenough et al (2002) Ivittuut (n = 6) Pearce and Leng (1996) Igaliko (n = 5) Andersen(1997) Qassiarsuk (n = 25) Bell and Blenkinsop (1989) other carbonatites of similar age (Goudini Schryburt Lake

and Prairie Lake n = 6)

844

R HALAMA ETAL

scattered and indicate that the rocks contain amixture of isotopic components from both mantleand crustal reservoirs or a heterogeneous mantlesource Only one sample (EF 108) has an eNd(i)value signi cantly above 0 This sample exhibitsthe largest spread of clinopyroxene compositionswhich may re ect strong alteration Goodenoughet al (2002) also note that the highest eNd occursin the most altered sample of their sample setHowever Paslick et al (1993) report clino-pyroxene data from two Ulukasik Groupsamples with eNd(i) gt +2 and those sampleswere apparently not disturbed with regard to theirSm-Nd age information On the other hand Piperet al (1999) suggest that the Sm-Nd systematicsof those samples may be erroneous due to laterdisturbance by uids derived from the Il otildeacutemaussaqintrusion Therefore although we are presentlynot able to decide whether positive eNd(i) values

for the EF basalts are real it is clear that none ofthose values is close to values of a 12 Ga oldDMM source (ie eNd = +5 to +7)

The Sr-Nd isotopic composition for OIB-typesources in the Proterozoic is not as wellconstrained as DMM Today OIB-type mantlesources have a wide range of isotopic composi-tions that partly overlap with DMM but eNd

values are generally lower (Hart et al 1992Hofmann 1997) The EM-1 mantle component ischaracterized by extremely low 143Nd144Ndratios and eNd values of between 0 and ndash5(Hofmann 1997) A possible 12 Ga position forEM-1 is plotted in Fig 7 calculated according toSchleicher et al (1998) with 147Sm144Nd = 018and 87Rb86Sr = 010 The present-day values usedin the calculation are 143Nd144Nd = 051234 and87Sr86Sr = 07057 taken from ocean islandbasalts closest to an EM-1 end-member(Hofmann 1997) Other Sr-Nd isotope datafrom the Gardar Province are plotted to putfurther constraints on the mantle sources of the EFbasalts (Fig 7) We also added comparative datafrom African and Canadian carbonatitesof similarage because there is a well known similaritybetween Sr-Nd isotopic data of OIBs andcarbonatites (Bell and Blenkinsop 1989) Thelinear array of the carbonatite data is consideredto be the result of mixing between the two mantlecomponents HIMU and EM-1 (Bell andBlenkinsop 1989 Bell and Tilton 2001) Thecarbonatites from the Gardar Province havesimilar Nd but slightly more radiogenic Srisotopic compositions than the reference carbo-natite data (Fig 7) The least contaminatedsamples overlap with the linear array and themore radiogenic Sr isotopic composition can beexplained by crustal contamination (Andersen1997) We therefore consider it possible that theGardar carbonatites also represent mixing ofhypothetical HIMU and EM-1 components of anOIB-like source Since the isotopically mostprimitive EF basalt samples overlap with theQassiarsuk carbonatites it could be possible thatthe basalts also contain such mantle componentsalthough Pb isotope data for the EF basalts thatcould clarify the importance of a HIMU sourceare lacking Based on the Sr-Nd isotope data theEF basalts appear to be derived from aheterogeneous OIB-like mantle source thatcomprises isotopically depleted and enrichedcomponents OIB-like and carbonatitic magma-tism are often attributed to mantle plume activity(Bell and Simonetti 1996 Hofmann 1997

2

4

6

8

10

12

-10 -8 -6 -4 -2 0 2 4 6 8

DMM

d18 O

[permil]

e Nd (i)

Typical2s error

EM-1

EF basalts - cpx

EF basalts - model melt

Ketilidian basement

Igaliko dykes

Depleted MORB mantle

Enriched mantle 1

EM-1 lavas - cpx

FIG 8 Oxygen isotopi c compositio ns (d18OV-SMOW) of clinopyroxene from EF basalts calculatedmelt (see text for details) and whole-rock Ketilidianbasement vs eNd(i) Whole-rock data for Igaliko dykerocks (n = 7) are from Pearce and Leng (1996) DMMdata derived from fresh MORB glasses are from Eiler etal (2000b) The eld of EM-1 lavas is taken fromclinopyroxene data of Gough and Tristan da CunhaIsland lavas (Harris et al 2000) and the EM-1 end-member composition is based on the calculations afterSchleicher et al (1998) for eNd(i) and an average value

of 578 (n = 18) from the EM-1 lavas for d18O

PETROGENESIS OF THE ERIKSFJORD BASALTS

845

Simonetti et al 1998) A plume input to theformation of the EF basalts is supported by therelatively large volume of magmas and the shortduration (lt5 Ma) of magmatism (Piper et al1999) It is also compatible with the geologicalsetting of a continental breakup Additionally aplume component appears to be necessary toexplain the thermal input as it is dif cult to melt acold lithosphere section (Arndt and Christensen1992)

Oxygen isotope data of the EF basalts aregenerally consistent with a mantle derivationincluding an OIB-like mantle source On the d18Ovs eNd(i) diagram (Fig 8) the overlap betweenthe EF data clinopyroxene data from EM-1-likeoceanic basalts and the hypothetical EM-1 end-member suggests an involvement of an EM-1-type mantle source Olivines from HIMU-likesources are known to be slightly lower in d18Othan typical upper mantle values from MORBs(Eiler et al 1997) but the EF data are notsuf cient to assess if the slightly lower d18Ovalues in the Il otildeacutemaussaq Group basalts are causedby a relatively higher input from a HIMU source

The contribution of the various mantle sourcesto the EF basalts can also be evaluated on theprimitive mantle-normalized multi-elementdiagram (Fig 4) and on the REE plots (Figs 5and 6) Both the whole-rock REE and theclinopyroxene REE patterns of the EF basaltsare unlike those representative of N-MORBsources We conclude again that DMM did notplay a major role in the petrogenesis of the EFbasalts On the other hand similarities with OIB-like mantle sources are more prominent Theshape of the clinopyroxene REE pattern closelyresembles those from clinopyroxene of OIB-likealkaline basalts despite differences in absoluteconcentrations The whole-rock REE patterns ofthe EF basalts lie between E-MORB and averageOIB patterns as do the multi-element patterns Onthe latter the EF basalts have marked positive Baand P peaks that are not present in typical OIBsHowever Ba contents should be interpreted withcaution since highly variable ZrBa ratios from014 to 070 are an indication that Ba was re-mobilized after emplacement of the basalts (egPrice et al 1991 Oliveira and Tarney 1995)Likewise P mobility was demonstrated in alteredbasaltic rocks (Winchester and Floyd 1976) Insummary the REE and trace element data areevidence that the mantle source(s) of the EFbasalts were enriched in incompatible traceelements relative to primitive mantle

For relatively primitive hy-normative basalticrocks of the Ivittuut area further north in theGardar Province it has been argued that some ofthem were derived from a lithospheric mantlesource previously enriched by subduction-related processes This interpretation is basedon certain geochemical characteristics such ashigh LaNb ratios and negative Nb anomalies(Goodenough et al 2002) For the EF basaltsnegative Nb anomalies are not clearly discern-ible but LaNb ratios are gt1 which seems to be acharacteristic feature of a lithospheric mantlesource (Fitton 1995) LaBa LaRb and LaKratios were also used to argue for a lithosphericorigin of the Gardar basic magmas although amantle plume involvement was assumed toprovide the required energy (Upton andEmeleus 1987 Upton et al 2003) Sincefeldspar is the only primary magmatic mineralin the EF basalts in which Ba Rb and K areincorporated in signi cant amounts and thefeldspars are generally altered these elementsmight have been easily mobilized and wetherefore do not dare to base the interpretationof the EF basalts sources on these data Inaddition the lack of precise geochemical data onthe composition of the lithosphere below theGardar Province complicates an evaluation ofthe lithospheric contribution to the EF basalticmagmatism

Certain ratios of incompatible trace elements ofthe EF basalts are compared to several occur-rences of ocean island basalts representative ofEM-1 and HIMU sources in Table 5 Ratiosbetween highly incompatible trace elements donot change signi cantly during partial melting orfractional crystallization in basaltic systems andmay therefore constrain the ratios in their sources(Weaver et al 1987) The EF basalts aresubdivided into the more alkaline isotopicallyprimitive Il otildeacutemaussaq Group samples and thosesamples with eNd(i) lt ndash1 The Il otilde maussaq Groupsamples have higher SrNd LaNb and ZrNbratios than HIMU and overlap with the upperrange of EM-1 ratios The EF basalts with eNd(i)lt ndash1 have generally higher ratios than the oceanicisland basalts and partly overlap with typicalvalues of the lower crust It therefore seems likelythat assimilation of lower crustal material mayhave changed the primary trace element ratios inthe basalts Therefore we will evaluate thepossible in uence of assimilation of crustalmaterial on the geochemistry of the EF basaltsin the following section

846

R HALAMA ETAL

The role of crustal contamination

Large amounts of bulk assimilation of crustalmaterial which is relatively SiO2-rich comparedto a basaltic melt would drive the compositions tohigher SiO2 contents and quartz-normativecompositions Partial melts derived from crustalrocks would also be relatively SiO2-rich (Beard etal 1993 Carroll and Wyllie 1990) and wouldhave the same effect The ne-normative characterof the EF basalts (Table 1) suggests thatassimilation of crustal material especially ofsilica-rich upper crustal rocks was not signi cantwith the possible exception of sample EF 059

On the primitive mantle-normalized multi-element diagrams (Fig 4) Ti troughs are typicalfor crustal rocks especially for the Upper Crust(Rudnick and Fountain 1995) and they maytherefore be present in rocks extensively contami-nated by continental crust Ti troughs aregenerally lacking in the patterns of the EFbasalts and a major upper crustal contribution tothe magmas can therefore be excluded However

small amounts of assimilation of lower crustalmaterial cannot be ruled out because the averagelower crust has higher Ti contents

Steep trends on Sr-Nd isotope plots have beenattributed to assimilation of granulite-faciesgneisses (eg Bernstein et al 1998) AFCmodelling based on an equation developed byDePaolo (1981) was performed to test thehypothesis of small amounts of contaminationwith lower crustal material (Fig 9) The basalticmagma which represents one end-member for thecalculations has the isotopic signature of theisotopically most primitive samples from thepresent data set and from the literature(Table 6) Two possible contaminants repre-senting upper and lower crustal compositionswere selected for modelling (Table 6) Therepresentative samples are a granite from theJulianehaEcirc b batholith for the upper crust (sampleJG 02) and an average of ve granulite-faciesgneisses with low RbSr ratios from the Archaeancraton of West Greenland (Taylor et al 1984) for

TABLE 5 Trace element ratios of EF basalts compared to ocean island basalts(OIBs) representing EM-1 and HIMU mantle components and the lowercrust

SrNd LaNb ZrNb

EF basalts with eNd lt ndash 1 249 ndash396 120 ndash150 121 ndash157Il otildeacutemaussaq Group EF basalts 198 ndash292 104 ndash119 51 ndash53EM-1 90 ndash207 067 ndash111 39 ndash109HIMU 45 -163 063 ndash080 27 ndash 36Lower continental crust 30 16 136

Data sources EM-1 Humphris and Thompson (1983 Walvis Ridge) and Cliff et al(1991 Inaccessible Islands) HIMU Chauvel et al (1992 Tubuai Island) Lowercontinental crust Rudnick and Fountain (1995)

TABLE 6 End-member compositions used for AFC modelling

End-member Sr (87Sr86Sr)i Nd eNd(i) Data source(ppm) (T = 12 Ga) (ppm) (T = 12 Ga)

Primitive EF basaltic 507 070278 18 + 22 Sample EF 072magma Paslick et al (1993)Upper crust 300 071665 62 ndash83 Sample JG 02(JulianehaEcirc b Granite)Lower crust 747 070668 63 ndash 234 Taylor et al (1984)(Granulite facies gneiss)

The composition of the granulite-facies gneiss is an average of 5 samples with low RbSr ratios lt02

PETROGENESIS OF THE ERIKSFJORD BASALTS

847

the lower crust An in uence of Archaean rockson Gardar magmas has been demonstrated by Pbisotope studies (Taylor and Upton 1993) andseismic data were also interpreted to re ect theexistence of a wedge of Archaean crust thatextends southwards to Lindenow Fjord (Dahl-Jensen et al 1998) We have shown in theprevious sections on the basis of Ni trends thatfractionation of olivine was most relevant to themajor element evolution of the lavas whereasfractionation of other phases was apparently ofminor importance We therefore modelled theAFC processes with olivine fractionation usingmineral-melt partition coef cients of 0014 for Srand 00066 for Nd (Rollinson 1993)

The modelling of AFC processes is stronglydependent on the parameter r de ned as the ratioof mass assimilation rate and fractional crystal-lization rate (DePaolo 1981) The change inisotopic signature becomes enhanced for highervalues of r Reiners et al (1995) have shown thatr values might be very high and even gt1 in theearly stages of magmatic evolution at hightemperatures since crystallization can besuppressed by assimilation Therefore we calcu-

lated AFC processes with two r values of 05 and08 that represent a relatively high and a middle tolow assimilation rate to demonstrate the effects ofvariable r (Fig 9) The results indicate that AFCprocesses involving upper crust similar to theJulianehaEcirc b Granite cannot explain the trend of theEF basalts because Sr isotopic compositionswould shift to considerably more radiogenicvalues This result is consistent with theconclusions derived from trace element dataThe AFC curves for assimilation of lower crusthowever agree well with the bulk of the data Themost negative eNd values can be explained byAFC processes with intermediate assimilationrates (r = 05) involving assimilation of lt5lower crustal material With a greater degree ofassimilation and an r value of 08 lt4assimilation is indicated

Lower crustal rocks of igneous origin have anaverage d18O value of +75plusmn14 with an overallrange from +54 to +125 (Fowler and Harmon1990) 5 to 10 bulk assimilation of lower crustalmaterial with d18O = +75 into a typical mantle-derived magma with d18O = +57 results ind18O values of the contaminated magma from

-24

-20

-16

-12

-8

-4

0

4

0700 0703 0706 0709 0712 0715 0718

r = 08

r = 05

r = 05

r = 08

EF Basalts

Endmembers of AFC calculations

Upper Crust

Lower Crust

Primitive EF basaltic magma

(87Sr86Sr)i

e Nd

(i)

09

08

06

09

08

06

09

0806

09

08

07

07

04

Typical errorCHUR

BS

E

FIG 9 eNd(i) vs (87Sr86Sr)i diagram showing the results of AFC modelling End-member compositions are listed inTable 6 Olivine is the only fractionating mineral Partition coef cients for olivinemelt are from the compilation ofRollinson (1993) Curves were calculated for two values of r (r = rate of assimilation rate of fractionalcrystallization) Dashed lines and numbers in italics are for r = 05 solid lines and regular numbers are for r = 08

Values at the curves are for F values (fraction of melt remaining) with 10 tick marks

848

R HALAMA ETAL

+579 to +588 Therefore the d18O data cannotbe used to rule out small amounts of assimilationof lower crust To evaluate how much contamina-tion would be permissible from the presentoxygen isotope data we calculated how muchcontamination of a mantle-derived magma with18O = +57 with a contaminant within the lowerrange of lower crustal xenoliths (d18O = +67)would drive the oxygen isotopic composition tothe highest value of the EF basalts (d18O~+63) Although simple mixing shows that60 assimilation is permissible this amountwould drive the whole-rock geochemistry tomore silicic compositions than is observed in theEF basalts In Fig 8 there is no obvious trendfrom the basalts towards the Ketilidian basementsuggesting again that upper crust was notassimilated in signi cant amounts

Ratios of trace elements incompatible inbasaltic systems can be used to distinguishbetween crust and mantle reservoirs (egWeaver 1991) Based on the AFC modellingEF basalts with eNd(i) lt ndash1 are those which aremost likely in uenced by crustal contaminationThe relatively high LaNb and ZrNb ratios ofthese samples are consistent with assimilation oflower crustal material SrNd ratios are evenhigher than the average lower crustal composi-tion but SrNd ratios in those samples which weretaken as representative for the lower crust in theAFC modelling can reach values of 80 ie wellabove the highest EF basalt ratios Again theSrNd ratios support the hypothesis of some lowercrustal contamination

The absence of more primitive basaltic rockswith MgO gt8 wt might also indicate lowercrustal assimilation It suggests that magmas mayhave been retained at depth where they acquiredmore evolved compositions by fractional crystal-lization and had the opportunity to assimilatelower crustal material during crustal underplating(Upton et al 2003)

Summary and conclusions

The new geochemical data of the EriksfjordFormation basalts in the Gardar Provinceprovided in this study are used to placepetrogenetic constraints on the sources of theirparental magmas Based on trace-element patternsand Sr-Nd isotopic constraints there seems to beno indication that a DMM-like source played animportant role in magma genesis of the EFbasalts Clinopyroxene and whole-rock REE

patterns are more compatible with an OIB-likesource but we are not able to conclude whetherthe observed enrichment in REE re ects thesourcersquos signature or magmatic processes Theisotopically most primitive EF basalts have Sr-Ndisotopic compositions that are similar to Gardarcarbonatites Due to the isotopic similaritybetween carbonatites and OIBs many arethought to be related to mantle plumes (Bell2001) We therefore speculate that OIB-likemantle components from a plume source werealso involved in the parental EF magmas but wecannot exclude interaction with the SCLM Twodistinct mantle components one isotopicallydepleted and another one enriched could havecontributed to the EF magmas AFC modelling ofSr-Nd isotope data indicates that assimilation oflt5 lower crust could explain the steep vector onthe Sr-Nd isotope plot but assimilation of uppercrustal material is unlikely Oxygen isotope dataof clinopyroxene separates are consistent with amantle origin but permit some assimilation oflower crustal rocks In summary the data areconsistent with the hypothesis that the EFmagmas could have originated from a mantleplume source which is supported by geologicalarguments and that they underwent someassimilation of lower crustal material

Acknowledgements

We would like to thank Bruce Paterson whoprovided invaluable help during Laser ICP-MSmeasurements at the Large-Scale GeochemicalFacility supported by the European Community-Access to Research Infrastructure action of theImproving Human Potential Programme contractnumber HPRI-CT-1999-00008 awarded to ProfBJ Wood (University of Bristol) Gabi Stoschekand Torsten Vennemann are thanked for their helpwith oxygen isotope measurements Elmar Reitterexpertly assisted with preparation and measure-ments of radiogenic isotopes and MathiasWestphal helped with microprobe measurementsREE whole-rock analyses were kindly providedby the NERC facility at Royal HollowayUniversity of London Michael Marks is thankedfor his pleasant company during eldwork Veryconstructive reviews by Tom Andersen andAlexei S Rukhlov greatly helped to improve themanuscript We also thank Ian Coulson for nalediting Funding of this work by the DeutscheForschungsgemeinschaft (grant Ma-21351-2) isgratefully acknowledged

PETROGENESIS OF THE ERIKSFJORD BASALTS

849

References

Andersen T (1997) Age and petrogenesis of theQassiarsuk carbonatite-alkaline silicate volcaniccomplex in the Gardar rift South GreenlandMineralogical Magazine 61 499 ndash513

Armstrong JT (1991) Quantitative elemental analysisof individual microparticles with electron beaminstruments Pp 261 ndash315 in Electron ProbeQuantitation (KFJ Heinrich and DE Newburyeditors) Plenum Press New York and London

Arndt NT and Christensen U (1992) The role oflithospheric mantle in continental ood volcanismThermal and geochemical constraints Journal ofGeophysical Research 97 10967 ndash 10981

Baker JA Menzies MA Thirlwall MF andMacpherson CG (199 7) Pet rogenes is ofQuaternary Intraplate Vocanism Sanarsquoa YemenImplications for Plume-Lithosphere Interaction andPolybaric Melt Hybridization Journal of Petrology 38 1359 ndash 1390

Baker JA MacPherson CG Menzies MAThirlwall MF Al-Kadasi M and Mattey DP(2000) Resolving crustal and mantle contributions tocontinental ood volcanism Yemen constraintsfrom mineral oxygen isotope data Journal ofPetrology 41 1805 ndash1820

Beard JS Abitz RJ and Lofgren GE (1993)Experimental melting of crustal xenoliths fromKilbourne Hole New Mexico and implications forthe contamination and genesi s of magmasContributions to Mineralogy and Petrology 11588 ndash102

Bell K (2001) Carbonatites relationships to mantle-plume activity Geological Society of AmericaSpecial Paper 352 267 ndash290

Bell K and Blenkinsop J (1989) Neodymium andStrontium Isotope Geochemistry of Carbonatites Pp278 ndash300 in Carbonatites ndash Genesis and Evolution(K Bell editor) Unwin Hyman London

Bell K and Simonetti A (1996) Carbonat iteMagmatism and Plume Activity Implications fromthe Nd Pb and Sr Isotope Systematics of OldoinyoLengai Journal of Petrology 37 1321 ndash1339

Bell K and Tilton GR (2001) Nd Pb and Sr IsotopicCompositions of East African Carbonat itesEvide nce for Mant le Mixing and PlumeInhomogene ity Journa l of Petrology 42 1927 ndash1945

Benoit M Polve M and Ceuleneer G (1996) Traceelement and isotopic characterization of ma ccumulates in a fossil mantle diapir (Oman ophiolite)Chemical Geology 134 199 ndash 214

Bernstein S Kelemen PB Tegner C Kurz MDBlusztajn J and Kent Brooks C (1998) Post-breakup basaltic magmatism along the EastGreenland Tertiary rifted margin Earth and

Planetary Science Letters 160 845 ndash862Boynton WV (1984) Geochemistry of the rare earth

elements meteorite studies Pp 63 ndash114 in RareEarth Element Geochemistry (P Henderson editor)Elsevier Amsterdam

Carroll MR and Wyllie PJ (1990) The systemtonalite-H2O at 15 kbar and the genesis of calc-alkaline magmas American Mineralogist 75345 ndash 357

Chauvel C Hofmann AW and Vidal P (1992)HIMU-EM The French Polynesian connectionEarth and Planetary Science Letters 110 99 ndash119

Clayton RN and Mayeda TK (1963) The use ofbromine penta uoride in the extraction of oxygenfrom oxides and silicates for isotope analysisGeochimica et Cosmochimica Acta 27 43 ndash52

Coulson IM and Chambers AD (1996) Patterns ofzonation in rare-earth-bearing minerals in nephelinesyenites of the North Qoroq center South GreenlandThe Canadian Mineralogist 34 1163 ndash1178

Cox KG Bell JD and Pankhurst RJ (1979) TheInterpretation of Igneous Rocks George Allen andUnwin London

Cross W Iddings JP Pirsson LV and WashingtonHS (1903) Quantitative Classi cation of IgneousRocks University of Chicago Press

Dahl-Jensen T Thybo T Hopper H and Rosing M(1998) Crustal structure at the SE Greenland marginfrom wide-angle and normal incidence seismic dataTectonophysics 288 191 ndash 198

DePaolo DJ (1981) Trace element and isotopic effectsof combined wallrock assimilation and fractionalcrystallisation Earth and Planetary Science Letters53 189 ndash 202

De P a o lo D J (1 9 8 8 ) Neodymium Iso topeGeochemistry An Introduction Springer-VerlagNew York

Dobosi G Downes H Mattey D and Embey-IsztinA (1998) Oxygen isotope ratios of phenocrysts fromalkali basalts of the Pannonian basin Evidence foran O-isotopically homogeneous upper mantle be-neath a subduction-in uenced area Lithos 42213 ndash 223

Eiler JM (2001) Oxygen isotope variations of basalticlavas and upper mantle rocks Pp 319 ndash364 inStable Isotope Geochemistry (JW Valley and DRCole editor s) Reviews in Minera logy andGeochemistry 43 The Mineralogical Society ofAmerica Washington DC

Eiler JM Farley KA Valley JW Hauri E CraigH Hart SR and Stolper EM (1997) Oxygenisotope variations in ocean island basalt phenocrystsGeochimica et Cosmochimica Acta 61 2281 ndash2293

Eiler JM Crawford A Elliot T Farley KAValley JW and Stolper EM (2000a) Oxygenisotope geochemistry of oceanic-arc lavas Journal

850

R HALAMA ETAL

model melt oxygen isotope data are within theranges of both groups Compared to Igaliko dykerocks from further SE in the Gardar Province(Pearce and Leng 1996) the EF basalts cover amuch more restricted eld possibly because d18Owhole-rock analyses are more prone to secondaryalteration than analyses of mineral separates(Fig 8) Whole-rock d18O measurements of theunderlying JulianehaEcirc b Granite and of the sedi-mentary sandstone members of the EriksfjordFormation yield values of 78 and 68 ndash115respectively (Table 4) and are shown forcomparison

Discussion

Mantle source regions of the EF basaltsTo reconstruct the mantle sources involved in theformation of the EF basalts we use Sr-Nd-Oisotope and incompatible trace element data On aSr-Nd isotope diagram the isotopic compositionof DMM was modelled for an age of 12 Ga(DePaolo 1981 Goldstein et al 1984 Taylor andMcLennan 1985) The fairly constant 87Sr86Srinitial isotope ratios of the EF basalts are ~0703which falls within the range typical for mantle-derived rocks but slightly above calculated valuesfor DMM (Fig 7) eNd(i) values are more

-4

-2

0

2

4

6

8

0701 0702 0703 0704 0705 0706

DMM Typical 2 s error

CHUR

BS

E

(87Sr86Sr)i

e Nd

(i)

EM-1

Qassiarsuk carbonatites and lamprophyres

EF basalts

Depleted MORB mantle

Ivittuut basaltic and lamprophyric dikes

Igaliko carbonatites and lamprophyres

Enriched mantle 1

Carbonatites from other regions of similar age

FIG 7 eNd(i) vs (87Sr86Sr)i diagram calculated for t = 12 Ga which is considered as the eruption age of the EFbasalts (Paslick et al 1993) The reference line of Bulk Silicate Earth (BSE) for the Sr isotopic composition wascalculated after DePaolo (1988) assuming present-day values of 00827 for 87Rb86Sr and 07045 for 87Sr86SrIsotopic composition of the depleted MORB mantle (DMM) was calculated after DePaolo (1981) Goldstein et al(1984) and Taylor and McLennan (1985) EM-1 was calculated after Schleicher et al (1998) (see text for details)Data sources are Goodenough et al (2002) Ivittuut (n = 6) Pearce and Leng (1996) Igaliko (n = 5) Andersen(1997) Qassiarsuk (n = 25) Bell and Blenkinsop (1989) other carbonatites of similar age (Goudini Schryburt Lake

and Prairie Lake n = 6)

844

R HALAMA ETAL

scattered and indicate that the rocks contain amixture of isotopic components from both mantleand crustal reservoirs or a heterogeneous mantlesource Only one sample (EF 108) has an eNd(i)value signi cantly above 0 This sample exhibitsthe largest spread of clinopyroxene compositionswhich may re ect strong alteration Goodenoughet al (2002) also note that the highest eNd occursin the most altered sample of their sample setHowever Paslick et al (1993) report clino-pyroxene data from two Ulukasik Groupsamples with eNd(i) gt +2 and those sampleswere apparently not disturbed with regard to theirSm-Nd age information On the other hand Piperet al (1999) suggest that the Sm-Nd systematicsof those samples may be erroneous due to laterdisturbance by uids derived from the Il otildeacutemaussaqintrusion Therefore although we are presentlynot able to decide whether positive eNd(i) values

for the EF basalts are real it is clear that none ofthose values is close to values of a 12 Ga oldDMM source (ie eNd = +5 to +7)

The Sr-Nd isotopic composition for OIB-typesources in the Proterozoic is not as wellconstrained as DMM Today OIB-type mantlesources have a wide range of isotopic composi-tions that partly overlap with DMM but eNd

values are generally lower (Hart et al 1992Hofmann 1997) The EM-1 mantle component ischaracterized by extremely low 143Nd144Ndratios and eNd values of between 0 and ndash5(Hofmann 1997) A possible 12 Ga position forEM-1 is plotted in Fig 7 calculated according toSchleicher et al (1998) with 147Sm144Nd = 018and 87Rb86Sr = 010 The present-day values usedin the calculation are 143Nd144Nd = 051234 and87Sr86Sr = 07057 taken from ocean islandbasalts closest to an EM-1 end-member(Hofmann 1997) Other Sr-Nd isotope datafrom the Gardar Province are plotted to putfurther constraints on the mantle sources of the EFbasalts (Fig 7) We also added comparative datafrom African and Canadian carbonatitesof similarage because there is a well known similaritybetween Sr-Nd isotopic data of OIBs andcarbonatites (Bell and Blenkinsop 1989) Thelinear array of the carbonatite data is consideredto be the result of mixing between the two mantlecomponents HIMU and EM-1 (Bell andBlenkinsop 1989 Bell and Tilton 2001) Thecarbonatites from the Gardar Province havesimilar Nd but slightly more radiogenic Srisotopic compositions than the reference carbo-natite data (Fig 7) The least contaminatedsamples overlap with the linear array and themore radiogenic Sr isotopic composition can beexplained by crustal contamination (Andersen1997) We therefore consider it possible that theGardar carbonatites also represent mixing ofhypothetical HIMU and EM-1 components of anOIB-like source Since the isotopically mostprimitive EF basalt samples overlap with theQassiarsuk carbonatites it could be possible thatthe basalts also contain such mantle componentsalthough Pb isotope data for the EF basalts thatcould clarify the importance of a HIMU sourceare lacking Based on the Sr-Nd isotope data theEF basalts appear to be derived from aheterogeneous OIB-like mantle source thatcomprises isotopically depleted and enrichedcomponents OIB-like and carbonatitic magma-tism are often attributed to mantle plume activity(Bell and Simonetti 1996 Hofmann 1997

2

4

6

8

10

12

-10 -8 -6 -4 -2 0 2 4 6 8

DMM

d18 O

[permil]

e Nd (i)

Typical2s error

EM-1

EF basalts - cpx

EF basalts - model melt

Ketilidian basement

Igaliko dykes

Depleted MORB mantle

Enriched mantle 1

EM-1 lavas - cpx

FIG 8 Oxygen isotopi c compositio ns (d18OV-SMOW) of clinopyroxene from EF basalts calculatedmelt (see text for details) and whole-rock Ketilidianbasement vs eNd(i) Whole-rock data for Igaliko dykerocks (n = 7) are from Pearce and Leng (1996) DMMdata derived from fresh MORB glasses are from Eiler etal (2000b) The eld of EM-1 lavas is taken fromclinopyroxene data of Gough and Tristan da CunhaIsland lavas (Harris et al 2000) and the EM-1 end-member composition is based on the calculations afterSchleicher et al (1998) for eNd(i) and an average value

of 578 (n = 18) from the EM-1 lavas for d18O

PETROGENESIS OF THE ERIKSFJORD BASALTS

845

Simonetti et al 1998) A plume input to theformation of the EF basalts is supported by therelatively large volume of magmas and the shortduration (lt5 Ma) of magmatism (Piper et al1999) It is also compatible with the geologicalsetting of a continental breakup Additionally aplume component appears to be necessary toexplain the thermal input as it is dif cult to melt acold lithosphere section (Arndt and Christensen1992)

Oxygen isotope data of the EF basalts aregenerally consistent with a mantle derivationincluding an OIB-like mantle source On the d18Ovs eNd(i) diagram (Fig 8) the overlap betweenthe EF data clinopyroxene data from EM-1-likeoceanic basalts and the hypothetical EM-1 end-member suggests an involvement of an EM-1-type mantle source Olivines from HIMU-likesources are known to be slightly lower in d18Othan typical upper mantle values from MORBs(Eiler et al 1997) but the EF data are notsuf cient to assess if the slightly lower d18Ovalues in the Il otildeacutemaussaq Group basalts are causedby a relatively higher input from a HIMU source

The contribution of the various mantle sourcesto the EF basalts can also be evaluated on theprimitive mantle-normalized multi-elementdiagram (Fig 4) and on the REE plots (Figs 5and 6) Both the whole-rock REE and theclinopyroxene REE patterns of the EF basaltsare unlike those representative of N-MORBsources We conclude again that DMM did notplay a major role in the petrogenesis of the EFbasalts On the other hand similarities with OIB-like mantle sources are more prominent Theshape of the clinopyroxene REE pattern closelyresembles those from clinopyroxene of OIB-likealkaline basalts despite differences in absoluteconcentrations The whole-rock REE patterns ofthe EF basalts lie between E-MORB and averageOIB patterns as do the multi-element patterns Onthe latter the EF basalts have marked positive Baand P peaks that are not present in typical OIBsHowever Ba contents should be interpreted withcaution since highly variable ZrBa ratios from014 to 070 are an indication that Ba was re-mobilized after emplacement of the basalts (egPrice et al 1991 Oliveira and Tarney 1995)Likewise P mobility was demonstrated in alteredbasaltic rocks (Winchester and Floyd 1976) Insummary the REE and trace element data areevidence that the mantle source(s) of the EFbasalts were enriched in incompatible traceelements relative to primitive mantle

For relatively primitive hy-normative basalticrocks of the Ivittuut area further north in theGardar Province it has been argued that some ofthem were derived from a lithospheric mantlesource previously enriched by subduction-related processes This interpretation is basedon certain geochemical characteristics such ashigh LaNb ratios and negative Nb anomalies(Goodenough et al 2002) For the EF basaltsnegative Nb anomalies are not clearly discern-ible but LaNb ratios are gt1 which seems to be acharacteristic feature of a lithospheric mantlesource (Fitton 1995) LaBa LaRb and LaKratios were also used to argue for a lithosphericorigin of the Gardar basic magmas although amantle plume involvement was assumed toprovide the required energy (Upton andEmeleus 1987 Upton et al 2003) Sincefeldspar is the only primary magmatic mineralin the EF basalts in which Ba Rb and K areincorporated in signi cant amounts and thefeldspars are generally altered these elementsmight have been easily mobilized and wetherefore do not dare to base the interpretationof the EF basalts sources on these data Inaddition the lack of precise geochemical data onthe composition of the lithosphere below theGardar Province complicates an evaluation ofthe lithospheric contribution to the EF basalticmagmatism

Certain ratios of incompatible trace elements ofthe EF basalts are compared to several occur-rences of ocean island basalts representative ofEM-1 and HIMU sources in Table 5 Ratiosbetween highly incompatible trace elements donot change signi cantly during partial melting orfractional crystallization in basaltic systems andmay therefore constrain the ratios in their sources(Weaver et al 1987) The EF basalts aresubdivided into the more alkaline isotopicallyprimitive Il otildeacutemaussaq Group samples and thosesamples with eNd(i) lt ndash1 The Il otilde maussaq Groupsamples have higher SrNd LaNb and ZrNbratios than HIMU and overlap with the upperrange of EM-1 ratios The EF basalts with eNd(i)lt ndash1 have generally higher ratios than the oceanicisland basalts and partly overlap with typicalvalues of the lower crust It therefore seems likelythat assimilation of lower crustal material mayhave changed the primary trace element ratios inthe basalts Therefore we will evaluate thepossible in uence of assimilation of crustalmaterial on the geochemistry of the EF basaltsin the following section

846

R HALAMA ETAL

The role of crustal contamination

Large amounts of bulk assimilation of crustalmaterial which is relatively SiO2-rich comparedto a basaltic melt would drive the compositions tohigher SiO2 contents and quartz-normativecompositions Partial melts derived from crustalrocks would also be relatively SiO2-rich (Beard etal 1993 Carroll and Wyllie 1990) and wouldhave the same effect The ne-normative characterof the EF basalts (Table 1) suggests thatassimilation of crustal material especially ofsilica-rich upper crustal rocks was not signi cantwith the possible exception of sample EF 059

On the primitive mantle-normalized multi-element diagrams (Fig 4) Ti troughs are typicalfor crustal rocks especially for the Upper Crust(Rudnick and Fountain 1995) and they maytherefore be present in rocks extensively contami-nated by continental crust Ti troughs aregenerally lacking in the patterns of the EFbasalts and a major upper crustal contribution tothe magmas can therefore be excluded However

small amounts of assimilation of lower crustalmaterial cannot be ruled out because the averagelower crust has higher Ti contents

Steep trends on Sr-Nd isotope plots have beenattributed to assimilation of granulite-faciesgneisses (eg Bernstein et al 1998) AFCmodelling based on an equation developed byDePaolo (1981) was performed to test thehypothesis of small amounts of contaminationwith lower crustal material (Fig 9) The basalticmagma which represents one end-member for thecalculations has the isotopic signature of theisotopically most primitive samples from thepresent data set and from the literature(Table 6) Two possible contaminants repre-senting upper and lower crustal compositionswere selected for modelling (Table 6) Therepresentative samples are a granite from theJulianehaEcirc b batholith for the upper crust (sampleJG 02) and an average of ve granulite-faciesgneisses with low RbSr ratios from the Archaeancraton of West Greenland (Taylor et al 1984) for

TABLE 5 Trace element ratios of EF basalts compared to ocean island basalts(OIBs) representing EM-1 and HIMU mantle components and the lowercrust

SrNd LaNb ZrNb

EF basalts with eNd lt ndash 1 249 ndash396 120 ndash150 121 ndash157Il otildeacutemaussaq Group EF basalts 198 ndash292 104 ndash119 51 ndash53EM-1 90 ndash207 067 ndash111 39 ndash109HIMU 45 -163 063 ndash080 27 ndash 36Lower continental crust 30 16 136

Data sources EM-1 Humphris and Thompson (1983 Walvis Ridge) and Cliff et al(1991 Inaccessible Islands) HIMU Chauvel et al (1992 Tubuai Island) Lowercontinental crust Rudnick and Fountain (1995)

TABLE 6 End-member compositions used for AFC modelling

End-member Sr (87Sr86Sr)i Nd eNd(i) Data source(ppm) (T = 12 Ga) (ppm) (T = 12 Ga)

Primitive EF basaltic 507 070278 18 + 22 Sample EF 072magma Paslick et al (1993)Upper crust 300 071665 62 ndash83 Sample JG 02(JulianehaEcirc b Granite)Lower crust 747 070668 63 ndash 234 Taylor et al (1984)(Granulite facies gneiss)

The composition of the granulite-facies gneiss is an average of 5 samples with low RbSr ratios lt02

PETROGENESIS OF THE ERIKSFJORD BASALTS

847

the lower crust An in uence of Archaean rockson Gardar magmas has been demonstrated by Pbisotope studies (Taylor and Upton 1993) andseismic data were also interpreted to re ect theexistence of a wedge of Archaean crust thatextends southwards to Lindenow Fjord (Dahl-Jensen et al 1998) We have shown in theprevious sections on the basis of Ni trends thatfractionation of olivine was most relevant to themajor element evolution of the lavas whereasfractionation of other phases was apparently ofminor importance We therefore modelled theAFC processes with olivine fractionation usingmineral-melt partition coef cients of 0014 for Srand 00066 for Nd (Rollinson 1993)

The modelling of AFC processes is stronglydependent on the parameter r de ned as the ratioof mass assimilation rate and fractional crystal-lization rate (DePaolo 1981) The change inisotopic signature becomes enhanced for highervalues of r Reiners et al (1995) have shown thatr values might be very high and even gt1 in theearly stages of magmatic evolution at hightemperatures since crystallization can besuppressed by assimilation Therefore we calcu-

lated AFC processes with two r values of 05 and08 that represent a relatively high and a middle tolow assimilation rate to demonstrate the effects ofvariable r (Fig 9) The results indicate that AFCprocesses involving upper crust similar to theJulianehaEcirc b Granite cannot explain the trend of theEF basalts because Sr isotopic compositionswould shift to considerably more radiogenicvalues This result is consistent with theconclusions derived from trace element dataThe AFC curves for assimilation of lower crusthowever agree well with the bulk of the data Themost negative eNd values can be explained byAFC processes with intermediate assimilationrates (r = 05) involving assimilation of lt5lower crustal material With a greater degree ofassimilation and an r value of 08 lt4assimilation is indicated

Lower crustal rocks of igneous origin have anaverage d18O value of +75plusmn14 with an overallrange from +54 to +125 (Fowler and Harmon1990) 5 to 10 bulk assimilation of lower crustalmaterial with d18O = +75 into a typical mantle-derived magma with d18O = +57 results ind18O values of the contaminated magma from

-24

-20

-16

-12

-8

-4

0

4

0700 0703 0706 0709 0712 0715 0718

r = 08

r = 05

r = 05

r = 08

EF Basalts

Endmembers of AFC calculations

Upper Crust

Lower Crust

Primitive EF basaltic magma

(87Sr86Sr)i

e Nd

(i)

09

08

06

09

08

06

09

0806

09

08

07

07

04

Typical errorCHUR

BS

E

FIG 9 eNd(i) vs (87Sr86Sr)i diagram showing the results of AFC modelling End-member compositions are listed inTable 6 Olivine is the only fractionating mineral Partition coef cients for olivinemelt are from the compilation ofRollinson (1993) Curves were calculated for two values of r (r = rate of assimilation rate of fractionalcrystallization) Dashed lines and numbers in italics are for r = 05 solid lines and regular numbers are for r = 08

Values at the curves are for F values (fraction of melt remaining) with 10 tick marks

848

R HALAMA ETAL

+579 to +588 Therefore the d18O data cannotbe used to rule out small amounts of assimilationof lower crust To evaluate how much contamina-tion would be permissible from the presentoxygen isotope data we calculated how muchcontamination of a mantle-derived magma with18O = +57 with a contaminant within the lowerrange of lower crustal xenoliths (d18O = +67)would drive the oxygen isotopic composition tothe highest value of the EF basalts (d18O~+63) Although simple mixing shows that60 assimilation is permissible this amountwould drive the whole-rock geochemistry tomore silicic compositions than is observed in theEF basalts In Fig 8 there is no obvious trendfrom the basalts towards the Ketilidian basementsuggesting again that upper crust was notassimilated in signi cant amounts

Ratios of trace elements incompatible inbasaltic systems can be used to distinguishbetween crust and mantle reservoirs (egWeaver 1991) Based on the AFC modellingEF basalts with eNd(i) lt ndash1 are those which aremost likely in uenced by crustal contaminationThe relatively high LaNb and ZrNb ratios ofthese samples are consistent with assimilation oflower crustal material SrNd ratios are evenhigher than the average lower crustal composi-tion but SrNd ratios in those samples which weretaken as representative for the lower crust in theAFC modelling can reach values of 80 ie wellabove the highest EF basalt ratios Again theSrNd ratios support the hypothesis of some lowercrustal contamination

The absence of more primitive basaltic rockswith MgO gt8 wt might also indicate lowercrustal assimilation It suggests that magmas mayhave been retained at depth where they acquiredmore evolved compositions by fractional crystal-lization and had the opportunity to assimilatelower crustal material during crustal underplating(Upton et al 2003)

Summary and conclusions

The new geochemical data of the EriksfjordFormation basalts in the Gardar Provinceprovided in this study are used to placepetrogenetic constraints on the sources of theirparental magmas Based on trace-element patternsand Sr-Nd isotopic constraints there seems to beno indication that a DMM-like source played animportant role in magma genesis of the EFbasalts Clinopyroxene and whole-rock REE

patterns are more compatible with an OIB-likesource but we are not able to conclude whetherthe observed enrichment in REE re ects thesourcersquos signature or magmatic processes Theisotopically most primitive EF basalts have Sr-Ndisotopic compositions that are similar to Gardarcarbonatites Due to the isotopic similaritybetween carbonatites and OIBs many arethought to be related to mantle plumes (Bell2001) We therefore speculate that OIB-likemantle components from a plume source werealso involved in the parental EF magmas but wecannot exclude interaction with the SCLM Twodistinct mantle components one isotopicallydepleted and another one enriched could havecontributed to the EF magmas AFC modelling ofSr-Nd isotope data indicates that assimilation oflt5 lower crust could explain the steep vector onthe Sr-Nd isotope plot but assimilation of uppercrustal material is unlikely Oxygen isotope dataof clinopyroxene separates are consistent with amantle origin but permit some assimilation oflower crustal rocks In summary the data areconsistent with the hypothesis that the EFmagmas could have originated from a mantleplume source which is supported by geologicalarguments and that they underwent someassimilation of lower crustal material

Acknowledgements

We would like to thank Bruce Paterson whoprovided invaluable help during Laser ICP-MSmeasurements at the Large-Scale GeochemicalFacility supported by the European Community-Access to Research Infrastructure action of theImproving Human Potential Programme contractnumber HPRI-CT-1999-00008 awarded to ProfBJ Wood (University of Bristol) Gabi Stoschekand Torsten Vennemann are thanked for their helpwith oxygen isotope measurements Elmar Reitterexpertly assisted with preparation and measure-ments of radiogenic isotopes and MathiasWestphal helped with microprobe measurementsREE whole-rock analyses were kindly providedby the NERC facility at Royal HollowayUniversity of London Michael Marks is thankedfor his pleasant company during eldwork Veryconstructive reviews by Tom Andersen andAlexei S Rukhlov greatly helped to improve themanuscript We also thank Ian Coulson for nalediting Funding of this work by the DeutscheForschungsgemeinschaft (grant Ma-21351-2) isgratefully acknowledged

PETROGENESIS OF THE ERIKSFJORD BASALTS

849

References

Andersen T (1997) Age and petrogenesis of theQassiarsuk carbonatite-alkaline silicate volcaniccomplex in the Gardar rift South GreenlandMineralogical Magazine 61 499 ndash513

Armstrong JT (1991) Quantitative elemental analysisof individual microparticles with electron beaminstruments Pp 261 ndash315 in Electron ProbeQuantitation (KFJ Heinrich and DE Newburyeditors) Plenum Press New York and London

Arndt NT and Christensen U (1992) The role oflithospheric mantle in continental ood volcanismThermal and geochemical constraints Journal ofGeophysical Research 97 10967 ndash 10981

Baker JA Menzies MA Thirlwall MF andMacpherson CG (199 7) Pet rogenes is ofQuaternary Intraplate Vocanism Sanarsquoa YemenImplications for Plume-Lithosphere Interaction andPolybaric Melt Hybridization Journal of Petrology 38 1359 ndash 1390

Baker JA MacPherson CG Menzies MAThirlwall MF Al-Kadasi M and Mattey DP(2000) Resolving crustal and mantle contributions tocontinental ood volcanism Yemen constraintsfrom mineral oxygen isotope data Journal ofPetrology 41 1805 ndash1820

Beard JS Abitz RJ and Lofgren GE (1993)Experimental melting of crustal xenoliths fromKilbourne Hole New Mexico and implications forthe contamination and genesi s of magmasContributions to Mineralogy and Petrology 11588 ndash102

Bell K (2001) Carbonatites relationships to mantle-plume activity Geological Society of AmericaSpecial Paper 352 267 ndash290

Bell K and Blenkinsop J (1989) Neodymium andStrontium Isotope Geochemistry of Carbonatites Pp278 ndash300 in Carbonatites ndash Genesis and Evolution(K Bell editor) Unwin Hyman London

Bell K and Simonetti A (1996) Carbonat iteMagmatism and Plume Activity Implications fromthe Nd Pb and Sr Isotope Systematics of OldoinyoLengai Journal of Petrology 37 1321 ndash1339

Bell K and Tilton GR (2001) Nd Pb and Sr IsotopicCompositions of East African Carbonat itesEvide nce for Mant le Mixing and PlumeInhomogene ity Journa l of Petrology 42 1927 ndash1945

Benoit M Polve M and Ceuleneer G (1996) Traceelement and isotopic characterization of ma ccumulates in a fossil mantle diapir (Oman ophiolite)Chemical Geology 134 199 ndash 214

Bernstein S Kelemen PB Tegner C Kurz MDBlusztajn J and Kent Brooks C (1998) Post-breakup basaltic magmatism along the EastGreenland Tertiary rifted margin Earth and

Planetary Science Letters 160 845 ndash862Boynton WV (1984) Geochemistry of the rare earth

elements meteorite studies Pp 63 ndash114 in RareEarth Element Geochemistry (P Henderson editor)Elsevier Amsterdam

Carroll MR and Wyllie PJ (1990) The systemtonalite-H2O at 15 kbar and the genesis of calc-alkaline magmas American Mineralogist 75345 ndash 357

Chauvel C Hofmann AW and Vidal P (1992)HIMU-EM The French Polynesian connectionEarth and Planetary Science Letters 110 99 ndash119

Clayton RN and Mayeda TK (1963) The use ofbromine penta uoride in the extraction of oxygenfrom oxides and silicates for isotope analysisGeochimica et Cosmochimica Acta 27 43 ndash52

Coulson IM and Chambers AD (1996) Patterns ofzonation in rare-earth-bearing minerals in nephelinesyenites of the North Qoroq center South GreenlandThe Canadian Mineralogist 34 1163 ndash1178

Cox KG Bell JD and Pankhurst RJ (1979) TheInterpretation of Igneous Rocks George Allen andUnwin London

Cross W Iddings JP Pirsson LV and WashingtonHS (1903) Quantitative Classi cation of IgneousRocks University of Chicago Press

Dahl-Jensen T Thybo T Hopper H and Rosing M(1998) Crustal structure at the SE Greenland marginfrom wide-angle and normal incidence seismic dataTectonophysics 288 191 ndash 198

DePaolo DJ (1981) Trace element and isotopic effectsof combined wallrock assimilation and fractionalcrystallisation Earth and Planetary Science Letters53 189 ndash 202

De P a o lo D J (1 9 8 8 ) Neodymium Iso topeGeochemistry An Introduction Springer-VerlagNew York

Dobosi G Downes H Mattey D and Embey-IsztinA (1998) Oxygen isotope ratios of phenocrysts fromalkali basalts of the Pannonian basin Evidence foran O-isotopically homogeneous upper mantle be-neath a subduction-in uenced area Lithos 42213 ndash 223

Eiler JM (2001) Oxygen isotope variations of basalticlavas and upper mantle rocks Pp 319 ndash364 inStable Isotope Geochemistry (JW Valley and DRCole editor s) Reviews in Minera logy andGeochemistry 43 The Mineralogical Society ofAmerica Washington DC

Eiler JM Farley KA Valley JW Hauri E CraigH Hart SR and Stolper EM (1997) Oxygenisotope variations in ocean island basalt phenocrystsGeochimica et Cosmochimica Acta 61 2281 ndash2293

Eiler JM Crawford A Elliot T Farley KAValley JW and Stolper EM (2000a) Oxygenisotope geochemistry of oceanic-arc lavas Journal

850

R HALAMA ETAL

scattered and indicate that the rocks contain amixture of isotopic components from both mantleand crustal reservoirs or a heterogeneous mantlesource Only one sample (EF 108) has an eNd(i)value signi cantly above 0 This sample exhibitsthe largest spread of clinopyroxene compositionswhich may re ect strong alteration Goodenoughet al (2002) also note that the highest eNd occursin the most altered sample of their sample setHowever Paslick et al (1993) report clino-pyroxene data from two Ulukasik Groupsamples with eNd(i) gt +2 and those sampleswere apparently not disturbed with regard to theirSm-Nd age information On the other hand Piperet al (1999) suggest that the Sm-Nd systematicsof those samples may be erroneous due to laterdisturbance by uids derived from the Il otildeacutemaussaqintrusion Therefore although we are presentlynot able to decide whether positive eNd(i) values

for the EF basalts are real it is clear that none ofthose values is close to values of a 12 Ga oldDMM source (ie eNd = +5 to +7)

The Sr-Nd isotopic composition for OIB-typesources in the Proterozoic is not as wellconstrained as DMM Today OIB-type mantlesources have a wide range of isotopic composi-tions that partly overlap with DMM but eNd

values are generally lower (Hart et al 1992Hofmann 1997) The EM-1 mantle component ischaracterized by extremely low 143Nd144Ndratios and eNd values of between 0 and ndash5(Hofmann 1997) A possible 12 Ga position forEM-1 is plotted in Fig 7 calculated according toSchleicher et al (1998) with 147Sm144Nd = 018and 87Rb86Sr = 010 The present-day values usedin the calculation are 143Nd144Nd = 051234 and87Sr86Sr = 07057 taken from ocean islandbasalts closest to an EM-1 end-member(Hofmann 1997) Other Sr-Nd isotope datafrom the Gardar Province are plotted to putfurther constraints on the mantle sources of the EFbasalts (Fig 7) We also added comparative datafrom African and Canadian carbonatitesof similarage because there is a well known similaritybetween Sr-Nd isotopic data of OIBs andcarbonatites (Bell and Blenkinsop 1989) Thelinear array of the carbonatite data is consideredto be the result of mixing between the two mantlecomponents HIMU and EM-1 (Bell andBlenkinsop 1989 Bell and Tilton 2001) Thecarbonatites from the Gardar Province havesimilar Nd but slightly more radiogenic Srisotopic compositions than the reference carbo-natite data (Fig 7) The least contaminatedsamples overlap with the linear array and themore radiogenic Sr isotopic composition can beexplained by crustal contamination (Andersen1997) We therefore consider it possible that theGardar carbonatites also represent mixing ofhypothetical HIMU and EM-1 components of anOIB-like source Since the isotopically mostprimitive EF basalt samples overlap with theQassiarsuk carbonatites it could be possible thatthe basalts also contain such mantle componentsalthough Pb isotope data for the EF basalts thatcould clarify the importance of a HIMU sourceare lacking Based on the Sr-Nd isotope data theEF basalts appear to be derived from aheterogeneous OIB-like mantle source thatcomprises isotopically depleted and enrichedcomponents OIB-like and carbonatitic magma-tism are often attributed to mantle plume activity(Bell and Simonetti 1996 Hofmann 1997

2

4

6

8

10

12

-10 -8 -6 -4 -2 0 2 4 6 8

DMM

d18 O

[permil]

e Nd (i)

Typical2s error

EM-1

EF basalts - cpx

EF basalts - model melt

Ketilidian basement

Igaliko dykes

Depleted MORB mantle

Enriched mantle 1

EM-1 lavas - cpx

FIG 8 Oxygen isotopi c compositio ns (d18OV-SMOW) of clinopyroxene from EF basalts calculatedmelt (see text for details) and whole-rock Ketilidianbasement vs eNd(i) Whole-rock data for Igaliko dykerocks (n = 7) are from Pearce and Leng (1996) DMMdata derived from fresh MORB glasses are from Eiler etal (2000b) The eld of EM-1 lavas is taken fromclinopyroxene data of Gough and Tristan da CunhaIsland lavas (Harris et al 2000) and the EM-1 end-member composition is based on the calculations afterSchleicher et al (1998) for eNd(i) and an average value

of 578 (n = 18) from the EM-1 lavas for d18O

PETROGENESIS OF THE ERIKSFJORD BASALTS

845

Simonetti et al 1998) A plume input to theformation of the EF basalts is supported by therelatively large volume of magmas and the shortduration (lt5 Ma) of magmatism (Piper et al1999) It is also compatible with the geologicalsetting of a continental breakup Additionally aplume component appears to be necessary toexplain the thermal input as it is dif cult to melt acold lithosphere section (Arndt and Christensen1992)

Oxygen isotope data of the EF basalts aregenerally consistent with a mantle derivationincluding an OIB-like mantle source On the d18Ovs eNd(i) diagram (Fig 8) the overlap betweenthe EF data clinopyroxene data from EM-1-likeoceanic basalts and the hypothetical EM-1 end-member suggests an involvement of an EM-1-type mantle source Olivines from HIMU-likesources are known to be slightly lower in d18Othan typical upper mantle values from MORBs(Eiler et al 1997) but the EF data are notsuf cient to assess if the slightly lower d18Ovalues in the Il otildeacutemaussaq Group basalts are causedby a relatively higher input from a HIMU source

The contribution of the various mantle sourcesto the EF basalts can also be evaluated on theprimitive mantle-normalized multi-elementdiagram (Fig 4) and on the REE plots (Figs 5and 6) Both the whole-rock REE and theclinopyroxene REE patterns of the EF basaltsare unlike those representative of N-MORBsources We conclude again that DMM did notplay a major role in the petrogenesis of the EFbasalts On the other hand similarities with OIB-like mantle sources are more prominent Theshape of the clinopyroxene REE pattern closelyresembles those from clinopyroxene of OIB-likealkaline basalts despite differences in absoluteconcentrations The whole-rock REE patterns ofthe EF basalts lie between E-MORB and averageOIB patterns as do the multi-element patterns Onthe latter the EF basalts have marked positive Baand P peaks that are not present in typical OIBsHowever Ba contents should be interpreted withcaution since highly variable ZrBa ratios from014 to 070 are an indication that Ba was re-mobilized after emplacement of the basalts (egPrice et al 1991 Oliveira and Tarney 1995)Likewise P mobility was demonstrated in alteredbasaltic rocks (Winchester and Floyd 1976) Insummary the REE and trace element data areevidence that the mantle source(s) of the EFbasalts were enriched in incompatible traceelements relative to primitive mantle

For relatively primitive hy-normative basalticrocks of the Ivittuut area further north in theGardar Province it has been argued that some ofthem were derived from a lithospheric mantlesource previously enriched by subduction-related processes This interpretation is basedon certain geochemical characteristics such ashigh LaNb ratios and negative Nb anomalies(Goodenough et al 2002) For the EF basaltsnegative Nb anomalies are not clearly discern-ible but LaNb ratios are gt1 which seems to be acharacteristic feature of a lithospheric mantlesource (Fitton 1995) LaBa LaRb and LaKratios were also used to argue for a lithosphericorigin of the Gardar basic magmas although amantle plume involvement was assumed toprovide the required energy (Upton andEmeleus 1987 Upton et al 2003) Sincefeldspar is the only primary magmatic mineralin the EF basalts in which Ba Rb and K areincorporated in signi cant amounts and thefeldspars are generally altered these elementsmight have been easily mobilized and wetherefore do not dare to base the interpretationof the EF basalts sources on these data Inaddition the lack of precise geochemical data onthe composition of the lithosphere below theGardar Province complicates an evaluation ofthe lithospheric contribution to the EF basalticmagmatism

Certain ratios of incompatible trace elements ofthe EF basalts are compared to several occur-rences of ocean island basalts representative ofEM-1 and HIMU sources in Table 5 Ratiosbetween highly incompatible trace elements donot change signi cantly during partial melting orfractional crystallization in basaltic systems andmay therefore constrain the ratios in their sources(Weaver et al 1987) The EF basalts aresubdivided into the more alkaline isotopicallyprimitive Il otildeacutemaussaq Group samples and thosesamples with eNd(i) lt ndash1 The Il otilde maussaq Groupsamples have higher SrNd LaNb and ZrNbratios than HIMU and overlap with the upperrange of EM-1 ratios The EF basalts with eNd(i)lt ndash1 have generally higher ratios than the oceanicisland basalts and partly overlap with typicalvalues of the lower crust It therefore seems likelythat assimilation of lower crustal material mayhave changed the primary trace element ratios inthe basalts Therefore we will evaluate thepossible in uence of assimilation of crustalmaterial on the geochemistry of the EF basaltsin the following section

846

R HALAMA ETAL

The role of crustal contamination

Large amounts of bulk assimilation of crustalmaterial which is relatively SiO2-rich comparedto a basaltic melt would drive the compositions tohigher SiO2 contents and quartz-normativecompositions Partial melts derived from crustalrocks would also be relatively SiO2-rich (Beard etal 1993 Carroll and Wyllie 1990) and wouldhave the same effect The ne-normative characterof the EF basalts (Table 1) suggests thatassimilation of crustal material especially ofsilica-rich upper crustal rocks was not signi cantwith the possible exception of sample EF 059

On the primitive mantle-normalized multi-element diagrams (Fig 4) Ti troughs are typicalfor crustal rocks especially for the Upper Crust(Rudnick and Fountain 1995) and they maytherefore be present in rocks extensively contami-nated by continental crust Ti troughs aregenerally lacking in the patterns of the EFbasalts and a major upper crustal contribution tothe magmas can therefore be excluded However

small amounts of assimilation of lower crustalmaterial cannot be ruled out because the averagelower crust has higher Ti contents

Steep trends on Sr-Nd isotope plots have beenattributed to assimilation of granulite-faciesgneisses (eg Bernstein et al 1998) AFCmodelling based on an equation developed byDePaolo (1981) was performed to test thehypothesis of small amounts of contaminationwith lower crustal material (Fig 9) The basalticmagma which represents one end-member for thecalculations has the isotopic signature of theisotopically most primitive samples from thepresent data set and from the literature(Table 6) Two possible contaminants repre-senting upper and lower crustal compositionswere selected for modelling (Table 6) Therepresentative samples are a granite from theJulianehaEcirc b batholith for the upper crust (sampleJG 02) and an average of ve granulite-faciesgneisses with low RbSr ratios from the Archaeancraton of West Greenland (Taylor et al 1984) for

TABLE 5 Trace element ratios of EF basalts compared to ocean island basalts(OIBs) representing EM-1 and HIMU mantle components and the lowercrust

SrNd LaNb ZrNb

EF basalts with eNd lt ndash 1 249 ndash396 120 ndash150 121 ndash157Il otildeacutemaussaq Group EF basalts 198 ndash292 104 ndash119 51 ndash53EM-1 90 ndash207 067 ndash111 39 ndash109HIMU 45 -163 063 ndash080 27 ndash 36Lower continental crust 30 16 136

Data sources EM-1 Humphris and Thompson (1983 Walvis Ridge) and Cliff et al(1991 Inaccessible Islands) HIMU Chauvel et al (1992 Tubuai Island) Lowercontinental crust Rudnick and Fountain (1995)

TABLE 6 End-member compositions used for AFC modelling

End-member Sr (87Sr86Sr)i Nd eNd(i) Data source(ppm) (T = 12 Ga) (ppm) (T = 12 Ga)

Primitive EF basaltic 507 070278 18 + 22 Sample EF 072magma Paslick et al (1993)Upper crust 300 071665 62 ndash83 Sample JG 02(JulianehaEcirc b Granite)Lower crust 747 070668 63 ndash 234 Taylor et al (1984)(Granulite facies gneiss)

The composition of the granulite-facies gneiss is an average of 5 samples with low RbSr ratios lt02

PETROGENESIS OF THE ERIKSFJORD BASALTS

847

the lower crust An in uence of Archaean rockson Gardar magmas has been demonstrated by Pbisotope studies (Taylor and Upton 1993) andseismic data were also interpreted to re ect theexistence of a wedge of Archaean crust thatextends southwards to Lindenow Fjord (Dahl-Jensen et al 1998) We have shown in theprevious sections on the basis of Ni trends thatfractionation of olivine was most relevant to themajor element evolution of the lavas whereasfractionation of other phases was apparently ofminor importance We therefore modelled theAFC processes with olivine fractionation usingmineral-melt partition coef cients of 0014 for Srand 00066 for Nd (Rollinson 1993)

The modelling of AFC processes is stronglydependent on the parameter r de ned as the ratioof mass assimilation rate and fractional crystal-lization rate (DePaolo 1981) The change inisotopic signature becomes enhanced for highervalues of r Reiners et al (1995) have shown thatr values might be very high and even gt1 in theearly stages of magmatic evolution at hightemperatures since crystallization can besuppressed by assimilation Therefore we calcu-

lated AFC processes with two r values of 05 and08 that represent a relatively high and a middle tolow assimilation rate to demonstrate the effects ofvariable r (Fig 9) The results indicate that AFCprocesses involving upper crust similar to theJulianehaEcirc b Granite cannot explain the trend of theEF basalts because Sr isotopic compositionswould shift to considerably more radiogenicvalues This result is consistent with theconclusions derived from trace element dataThe AFC curves for assimilation of lower crusthowever agree well with the bulk of the data Themost negative eNd values can be explained byAFC processes with intermediate assimilationrates (r = 05) involving assimilation of lt5lower crustal material With a greater degree ofassimilation and an r value of 08 lt4assimilation is indicated

Lower crustal rocks of igneous origin have anaverage d18O value of +75plusmn14 with an overallrange from +54 to +125 (Fowler and Harmon1990) 5 to 10 bulk assimilation of lower crustalmaterial with d18O = +75 into a typical mantle-derived magma with d18O = +57 results ind18O values of the contaminated magma from

-24

-20

-16

-12

-8

-4

0

4

0700 0703 0706 0709 0712 0715 0718

r = 08

r = 05

r = 05

r = 08

EF Basalts

Endmembers of AFC calculations

Upper Crust

Lower Crust

Primitive EF basaltic magma

(87Sr86Sr)i

e Nd

(i)

09

08

06

09

08

06

09

0806

09

08

07

07

04

Typical errorCHUR

BS

E

FIG 9 eNd(i) vs (87Sr86Sr)i diagram showing the results of AFC modelling End-member compositions are listed inTable 6 Olivine is the only fractionating mineral Partition coef cients for olivinemelt are from the compilation ofRollinson (1993) Curves were calculated for two values of r (r = rate of assimilation rate of fractionalcrystallization) Dashed lines and numbers in italics are for r = 05 solid lines and regular numbers are for r = 08

Values at the curves are for F values (fraction of melt remaining) with 10 tick marks

848

R HALAMA ETAL

+579 to +588 Therefore the d18O data cannotbe used to rule out small amounts of assimilationof lower crust To evaluate how much contamina-tion would be permissible from the presentoxygen isotope data we calculated how muchcontamination of a mantle-derived magma with18O = +57 with a contaminant within the lowerrange of lower crustal xenoliths (d18O = +67)would drive the oxygen isotopic composition tothe highest value of the EF basalts (d18O~+63) Although simple mixing shows that60 assimilation is permissible this amountwould drive the whole-rock geochemistry tomore silicic compositions than is observed in theEF basalts In Fig 8 there is no obvious trendfrom the basalts towards the Ketilidian basementsuggesting again that upper crust was notassimilated in signi cant amounts

Ratios of trace elements incompatible inbasaltic systems can be used to distinguishbetween crust and mantle reservoirs (egWeaver 1991) Based on the AFC modellingEF basalts with eNd(i) lt ndash1 are those which aremost likely in uenced by crustal contaminationThe relatively high LaNb and ZrNb ratios ofthese samples are consistent with assimilation oflower crustal material SrNd ratios are evenhigher than the average lower crustal composi-tion but SrNd ratios in those samples which weretaken as representative for the lower crust in theAFC modelling can reach values of 80 ie wellabove the highest EF basalt ratios Again theSrNd ratios support the hypothesis of some lowercrustal contamination

The absence of more primitive basaltic rockswith MgO gt8 wt might also indicate lowercrustal assimilation It suggests that magmas mayhave been retained at depth where they acquiredmore evolved compositions by fractional crystal-lization and had the opportunity to assimilatelower crustal material during crustal underplating(Upton et al 2003)

Summary and conclusions

The new geochemical data of the EriksfjordFormation basalts in the Gardar Provinceprovided in this study are used to placepetrogenetic constraints on the sources of theirparental magmas Based on trace-element patternsand Sr-Nd isotopic constraints there seems to beno indication that a DMM-like source played animportant role in magma genesis of the EFbasalts Clinopyroxene and whole-rock REE

patterns are more compatible with an OIB-likesource but we are not able to conclude whetherthe observed enrichment in REE re ects thesourcersquos signature or magmatic processes Theisotopically most primitive EF basalts have Sr-Ndisotopic compositions that are similar to Gardarcarbonatites Due to the isotopic similaritybetween carbonatites and OIBs many arethought to be related to mantle plumes (Bell2001) We therefore speculate that OIB-likemantle components from a plume source werealso involved in the parental EF magmas but wecannot exclude interaction with the SCLM Twodistinct mantle components one isotopicallydepleted and another one enriched could havecontributed to the EF magmas AFC modelling ofSr-Nd isotope data indicates that assimilation oflt5 lower crust could explain the steep vector onthe Sr-Nd isotope plot but assimilation of uppercrustal material is unlikely Oxygen isotope dataof clinopyroxene separates are consistent with amantle origin but permit some assimilation oflower crustal rocks In summary the data areconsistent with the hypothesis that the EFmagmas could have originated from a mantleplume source which is supported by geologicalarguments and that they underwent someassimilation of lower crustal material

Acknowledgements

We would like to thank Bruce Paterson whoprovided invaluable help during Laser ICP-MSmeasurements at the Large-Scale GeochemicalFacility supported by the European Community-Access to Research Infrastructure action of theImproving Human Potential Programme contractnumber HPRI-CT-1999-00008 awarded to ProfBJ Wood (University of Bristol) Gabi Stoschekand Torsten Vennemann are thanked for their helpwith oxygen isotope measurements Elmar Reitterexpertly assisted with preparation and measure-ments of radiogenic isotopes and MathiasWestphal helped with microprobe measurementsREE whole-rock analyses were kindly providedby the NERC facility at Royal HollowayUniversity of London Michael Marks is thankedfor his pleasant company during eldwork Veryconstructive reviews by Tom Andersen andAlexei S Rukhlov greatly helped to improve themanuscript We also thank Ian Coulson for nalediting Funding of this work by the DeutscheForschungsgemeinschaft (grant Ma-21351-2) isgratefully acknowledged

PETROGENESIS OF THE ERIKSFJORD BASALTS

849

References

Andersen T (1997) Age and petrogenesis of theQassiarsuk carbonatite-alkaline silicate volcaniccomplex in the Gardar rift South GreenlandMineralogical Magazine 61 499 ndash513

Armstrong JT (1991) Quantitative elemental analysisof individual microparticles with electron beaminstruments Pp 261 ndash315 in Electron ProbeQuantitation (KFJ Heinrich and DE Newburyeditors) Plenum Press New York and London

Arndt NT and Christensen U (1992) The role oflithospheric mantle in continental ood volcanismThermal and geochemical constraints Journal ofGeophysical Research 97 10967 ndash 10981

Baker JA Menzies MA Thirlwall MF andMacpherson CG (199 7) Pet rogenes is ofQuaternary Intraplate Vocanism Sanarsquoa YemenImplications for Plume-Lithosphere Interaction andPolybaric Melt Hybridization Journal of Petrology 38 1359 ndash 1390

Baker JA MacPherson CG Menzies MAThirlwall MF Al-Kadasi M and Mattey DP(2000) Resolving crustal and mantle contributions tocontinental ood volcanism Yemen constraintsfrom mineral oxygen isotope data Journal ofPetrology 41 1805 ndash1820

Beard JS Abitz RJ and Lofgren GE (1993)Experimental melting of crustal xenoliths fromKilbourne Hole New Mexico and implications forthe contamination and genesi s of magmasContributions to Mineralogy and Petrology 11588 ndash102

Bell K (2001) Carbonatites relationships to mantle-plume activity Geological Society of AmericaSpecial Paper 352 267 ndash290

Bell K and Blenkinsop J (1989) Neodymium andStrontium Isotope Geochemistry of Carbonatites Pp278 ndash300 in Carbonatites ndash Genesis and Evolution(K Bell editor) Unwin Hyman London

Bell K and Simonetti A (1996) Carbonat iteMagmatism and Plume Activity Implications fromthe Nd Pb and Sr Isotope Systematics of OldoinyoLengai Journal of Petrology 37 1321 ndash1339

Bell K and Tilton GR (2001) Nd Pb and Sr IsotopicCompositions of East African Carbonat itesEvide nce for Mant le Mixing and PlumeInhomogene ity Journa l of Petrology 42 1927 ndash1945

Benoit M Polve M and Ceuleneer G (1996) Traceelement and isotopic characterization of ma ccumulates in a fossil mantle diapir (Oman ophiolite)Chemical Geology 134 199 ndash 214

Bernstein S Kelemen PB Tegner C Kurz MDBlusztajn J and Kent Brooks C (1998) Post-breakup basaltic magmatism along the EastGreenland Tertiary rifted margin Earth and

Planetary Science Letters 160 845 ndash862Boynton WV (1984) Geochemistry of the rare earth

elements meteorite studies Pp 63 ndash114 in RareEarth Element Geochemistry (P Henderson editor)Elsevier Amsterdam

Carroll MR and Wyllie PJ (1990) The systemtonalite-H2O at 15 kbar and the genesis of calc-alkaline magmas American Mineralogist 75345 ndash 357

Chauvel C Hofmann AW and Vidal P (1992)HIMU-EM The French Polynesian connectionEarth and Planetary Science Letters 110 99 ndash119

Clayton RN and Mayeda TK (1963) The use ofbromine penta uoride in the extraction of oxygenfrom oxides and silicates for isotope analysisGeochimica et Cosmochimica Acta 27 43 ndash52

Coulson IM and Chambers AD (1996) Patterns ofzonation in rare-earth-bearing minerals in nephelinesyenites of the North Qoroq center South GreenlandThe Canadian Mineralogist 34 1163 ndash1178

Cox KG Bell JD and Pankhurst RJ (1979) TheInterpretation of Igneous Rocks George Allen andUnwin London

Cross W Iddings JP Pirsson LV and WashingtonHS (1903) Quantitative Classi cation of IgneousRocks University of Chicago Press

Dahl-Jensen T Thybo T Hopper H and Rosing M(1998) Crustal structure at the SE Greenland marginfrom wide-angle and normal incidence seismic dataTectonophysics 288 191 ndash 198

DePaolo DJ (1981) Trace element and isotopic effectsof combined wallrock assimilation and fractionalcrystallisation Earth and Planetary Science Letters53 189 ndash 202

De P a o lo D J (1 9 8 8 ) Neodymium Iso topeGeochemistry An Introduction Springer-VerlagNew York

Dobosi G Downes H Mattey D and Embey-IsztinA (1998) Oxygen isotope ratios of phenocrysts fromalkali basalts of the Pannonian basin Evidence foran O-isotopically homogeneous upper mantle be-neath a subduction-in uenced area Lithos 42213 ndash 223

Eiler JM (2001) Oxygen isotope variations of basalticlavas and upper mantle rocks Pp 319 ndash364 inStable Isotope Geochemistry (JW Valley and DRCole editor s) Reviews in Minera logy andGeochemistry 43 The Mineralogical Society ofAmerica Washington DC

Eiler JM Farley KA Valley JW Hauri E CraigH Hart SR and Stolper EM (1997) Oxygenisotope variations in ocean island basalt phenocrystsGeochimica et Cosmochimica Acta 61 2281 ndash2293

Eiler JM Crawford A Elliot T Farley KAValley JW and Stolper EM (2000a) Oxygenisotope geochemistry of oceanic-arc lavas Journal

850

R HALAMA ETAL

Simonetti et al 1998) A plume input to theformation of the EF basalts is supported by therelatively large volume of magmas and the shortduration (lt5 Ma) of magmatism (Piper et al1999) It is also compatible with the geologicalsetting of a continental breakup Additionally aplume component appears to be necessary toexplain the thermal input as it is dif cult to melt acold lithosphere section (Arndt and Christensen1992)

Oxygen isotope data of the EF basalts aregenerally consistent with a mantle derivationincluding an OIB-like mantle source On the d18Ovs eNd(i) diagram (Fig 8) the overlap betweenthe EF data clinopyroxene data from EM-1-likeoceanic basalts and the hypothetical EM-1 end-member suggests an involvement of an EM-1-type mantle source Olivines from HIMU-likesources are known to be slightly lower in d18Othan typical upper mantle values from MORBs(Eiler et al 1997) but the EF data are notsuf cient to assess if the slightly lower d18Ovalues in the Il otildeacutemaussaq Group basalts are causedby a relatively higher input from a HIMU source

The contribution of the various mantle sourcesto the EF basalts can also be evaluated on theprimitive mantle-normalized multi-elementdiagram (Fig 4) and on the REE plots (Figs 5and 6) Both the whole-rock REE and theclinopyroxene REE patterns of the EF basaltsare unlike those representative of N-MORBsources We conclude again that DMM did notplay a major role in the petrogenesis of the EFbasalts On the other hand similarities with OIB-like mantle sources are more prominent Theshape of the clinopyroxene REE pattern closelyresembles those from clinopyroxene of OIB-likealkaline basalts despite differences in absoluteconcentrations The whole-rock REE patterns ofthe EF basalts lie between E-MORB and averageOIB patterns as do the multi-element patterns Onthe latter the EF basalts have marked positive Baand P peaks that are not present in typical OIBsHowever Ba contents should be interpreted withcaution since highly variable ZrBa ratios from014 to 070 are an indication that Ba was re-mobilized after emplacement of the basalts (egPrice et al 1991 Oliveira and Tarney 1995)Likewise P mobility was demonstrated in alteredbasaltic rocks (Winchester and Floyd 1976) Insummary the REE and trace element data areevidence that the mantle source(s) of the EFbasalts were enriched in incompatible traceelements relative to primitive mantle

For relatively primitive hy-normative basalticrocks of the Ivittuut area further north in theGardar Province it has been argued that some ofthem were derived from a lithospheric mantlesource previously enriched by subduction-related processes This interpretation is basedon certain geochemical characteristics such ashigh LaNb ratios and negative Nb anomalies(Goodenough et al 2002) For the EF basaltsnegative Nb anomalies are not clearly discern-ible but LaNb ratios are gt1 which seems to be acharacteristic feature of a lithospheric mantlesource (Fitton 1995) LaBa LaRb and LaKratios were also used to argue for a lithosphericorigin of the Gardar basic magmas although amantle plume involvement was assumed toprovide the required energy (Upton andEmeleus 1987 Upton et al 2003) Sincefeldspar is the only primary magmatic mineralin the EF basalts in which Ba Rb and K areincorporated in signi cant amounts and thefeldspars are generally altered these elementsmight have been easily mobilized and wetherefore do not dare to base the interpretationof the EF basalts sources on these data Inaddition the lack of precise geochemical data onthe composition of the lithosphere below theGardar Province complicates an evaluation ofthe lithospheric contribution to the EF basalticmagmatism

Certain ratios of incompatible trace elements ofthe EF basalts are compared to several occur-rences of ocean island basalts representative ofEM-1 and HIMU sources in Table 5 Ratiosbetween highly incompatible trace elements donot change signi cantly during partial melting orfractional crystallization in basaltic systems andmay therefore constrain the ratios in their sources(Weaver et al 1987) The EF basalts aresubdivided into the more alkaline isotopicallyprimitive Il otildeacutemaussaq Group samples and thosesamples with eNd(i) lt ndash1 The Il otilde maussaq Groupsamples have higher SrNd LaNb and ZrNbratios than HIMU and overlap with the upperrange of EM-1 ratios The EF basalts with eNd(i)lt ndash1 have generally higher ratios than the oceanicisland basalts and partly overlap with typicalvalues of the lower crust It therefore seems likelythat assimilation of lower crustal material mayhave changed the primary trace element ratios inthe basalts Therefore we will evaluate thepossible in uence of assimilation of crustalmaterial on the geochemistry of the EF basaltsin the following section

846

R HALAMA ETAL

The role of crustal contamination

Large amounts of bulk assimilation of crustalmaterial which is relatively SiO2-rich comparedto a basaltic melt would drive the compositions tohigher SiO2 contents and quartz-normativecompositions Partial melts derived from crustalrocks would also be relatively SiO2-rich (Beard etal 1993 Carroll and Wyllie 1990) and wouldhave the same effect The ne-normative characterof the EF basalts (Table 1) suggests thatassimilation of crustal material especially ofsilica-rich upper crustal rocks was not signi cantwith the possible exception of sample EF 059

On the primitive mantle-normalized multi-element diagrams (Fig 4) Ti troughs are typicalfor crustal rocks especially for the Upper Crust(Rudnick and Fountain 1995) and they maytherefore be present in rocks extensively contami-nated by continental crust Ti troughs aregenerally lacking in the patterns of the EFbasalts and a major upper crustal contribution tothe magmas can therefore be excluded However

small amounts of assimilation of lower crustalmaterial cannot be ruled out because the averagelower crust has higher Ti contents

Steep trends on Sr-Nd isotope plots have beenattributed to assimilation of granulite-faciesgneisses (eg Bernstein et al 1998) AFCmodelling based on an equation developed byDePaolo (1981) was performed to test thehypothesis of small amounts of contaminationwith lower crustal material (Fig 9) The basalticmagma which represents one end-member for thecalculations has the isotopic signature of theisotopically most primitive samples from thepresent data set and from the literature(Table 6) Two possible contaminants repre-senting upper and lower crustal compositionswere selected for modelling (Table 6) Therepresentative samples are a granite from theJulianehaEcirc b batholith for the upper crust (sampleJG 02) and an average of ve granulite-faciesgneisses with low RbSr ratios from the Archaeancraton of West Greenland (Taylor et al 1984) for

TABLE 5 Trace element ratios of EF basalts compared to ocean island basalts(OIBs) representing EM-1 and HIMU mantle components and the lowercrust

SrNd LaNb ZrNb

EF basalts with eNd lt ndash 1 249 ndash396 120 ndash150 121 ndash157Il otildeacutemaussaq Group EF basalts 198 ndash292 104 ndash119 51 ndash53EM-1 90 ndash207 067 ndash111 39 ndash109HIMU 45 -163 063 ndash080 27 ndash 36Lower continental crust 30 16 136

Data sources EM-1 Humphris and Thompson (1983 Walvis Ridge) and Cliff et al(1991 Inaccessible Islands) HIMU Chauvel et al (1992 Tubuai Island) Lowercontinental crust Rudnick and Fountain (1995)

TABLE 6 End-member compositions used for AFC modelling

End-member Sr (87Sr86Sr)i Nd eNd(i) Data source(ppm) (T = 12 Ga) (ppm) (T = 12 Ga)

Primitive EF basaltic 507 070278 18 + 22 Sample EF 072magma Paslick et al (1993)Upper crust 300 071665 62 ndash83 Sample JG 02(JulianehaEcirc b Granite)Lower crust 747 070668 63 ndash 234 Taylor et al (1984)(Granulite facies gneiss)

The composition of the granulite-facies gneiss is an average of 5 samples with low RbSr ratios lt02

PETROGENESIS OF THE ERIKSFJORD BASALTS

847

the lower crust An in uence of Archaean rockson Gardar magmas has been demonstrated by Pbisotope studies (Taylor and Upton 1993) andseismic data were also interpreted to re ect theexistence of a wedge of Archaean crust thatextends southwards to Lindenow Fjord (Dahl-Jensen et al 1998) We have shown in theprevious sections on the basis of Ni trends thatfractionation of olivine was most relevant to themajor element evolution of the lavas whereasfractionation of other phases was apparently ofminor importance We therefore modelled theAFC processes with olivine fractionation usingmineral-melt partition coef cients of 0014 for Srand 00066 for Nd (Rollinson 1993)

The modelling of AFC processes is stronglydependent on the parameter r de ned as the ratioof mass assimilation rate and fractional crystal-lization rate (DePaolo 1981) The change inisotopic signature becomes enhanced for highervalues of r Reiners et al (1995) have shown thatr values might be very high and even gt1 in theearly stages of magmatic evolution at hightemperatures since crystallization can besuppressed by assimilation Therefore we calcu-

lated AFC processes with two r values of 05 and08 that represent a relatively high and a middle tolow assimilation rate to demonstrate the effects ofvariable r (Fig 9) The results indicate that AFCprocesses involving upper crust similar to theJulianehaEcirc b Granite cannot explain the trend of theEF basalts because Sr isotopic compositionswould shift to considerably more radiogenicvalues This result is consistent with theconclusions derived from trace element dataThe AFC curves for assimilation of lower crusthowever agree well with the bulk of the data Themost negative eNd values can be explained byAFC processes with intermediate assimilationrates (r = 05) involving assimilation of lt5lower crustal material With a greater degree ofassimilation and an r value of 08 lt4assimilation is indicated

Lower crustal rocks of igneous origin have anaverage d18O value of +75plusmn14 with an overallrange from +54 to +125 (Fowler and Harmon1990) 5 to 10 bulk assimilation of lower crustalmaterial with d18O = +75 into a typical mantle-derived magma with d18O = +57 results ind18O values of the contaminated magma from

-24

-20

-16

-12

-8

-4

0

4

0700 0703 0706 0709 0712 0715 0718

r = 08

r = 05

r = 05

r = 08

EF Basalts

Endmembers of AFC calculations

Upper Crust

Lower Crust

Primitive EF basaltic magma

(87Sr86Sr)i

e Nd

(i)

09

08

06

09

08

06

09

0806

09

08

07

07

04

Typical errorCHUR

BS

E

FIG 9 eNd(i) vs (87Sr86Sr)i diagram showing the results of AFC modelling End-member compositions are listed inTable 6 Olivine is the only fractionating mineral Partition coef cients for olivinemelt are from the compilation ofRollinson (1993) Curves were calculated for two values of r (r = rate of assimilation rate of fractionalcrystallization) Dashed lines and numbers in italics are for r = 05 solid lines and regular numbers are for r = 08

Values at the curves are for F values (fraction of melt remaining) with 10 tick marks

848

R HALAMA ETAL

+579 to +588 Therefore the d18O data cannotbe used to rule out small amounts of assimilationof lower crust To evaluate how much contamina-tion would be permissible from the presentoxygen isotope data we calculated how muchcontamination of a mantle-derived magma with18O = +57 with a contaminant within the lowerrange of lower crustal xenoliths (d18O = +67)would drive the oxygen isotopic composition tothe highest value of the EF basalts (d18O~+63) Although simple mixing shows that60 assimilation is permissible this amountwould drive the whole-rock geochemistry tomore silicic compositions than is observed in theEF basalts In Fig 8 there is no obvious trendfrom the basalts towards the Ketilidian basementsuggesting again that upper crust was notassimilated in signi cant amounts

Ratios of trace elements incompatible inbasaltic systems can be used to distinguishbetween crust and mantle reservoirs (egWeaver 1991) Based on the AFC modellingEF basalts with eNd(i) lt ndash1 are those which aremost likely in uenced by crustal contaminationThe relatively high LaNb and ZrNb ratios ofthese samples are consistent with assimilation oflower crustal material SrNd ratios are evenhigher than the average lower crustal composi-tion but SrNd ratios in those samples which weretaken as representative for the lower crust in theAFC modelling can reach values of 80 ie wellabove the highest EF basalt ratios Again theSrNd ratios support the hypothesis of some lowercrustal contamination

The absence of more primitive basaltic rockswith MgO gt8 wt might also indicate lowercrustal assimilation It suggests that magmas mayhave been retained at depth where they acquiredmore evolved compositions by fractional crystal-lization and had the opportunity to assimilatelower crustal material during crustal underplating(Upton et al 2003)

Summary and conclusions

The new geochemical data of the EriksfjordFormation basalts in the Gardar Provinceprovided in this study are used to placepetrogenetic constraints on the sources of theirparental magmas Based on trace-element patternsand Sr-Nd isotopic constraints there seems to beno indication that a DMM-like source played animportant role in magma genesis of the EFbasalts Clinopyroxene and whole-rock REE

patterns are more compatible with an OIB-likesource but we are not able to conclude whetherthe observed enrichment in REE re ects thesourcersquos signature or magmatic processes Theisotopically most primitive EF basalts have Sr-Ndisotopic compositions that are similar to Gardarcarbonatites Due to the isotopic similaritybetween carbonatites and OIBs many arethought to be related to mantle plumes (Bell2001) We therefore speculate that OIB-likemantle components from a plume source werealso involved in the parental EF magmas but wecannot exclude interaction with the SCLM Twodistinct mantle components one isotopicallydepleted and another one enriched could havecontributed to the EF magmas AFC modelling ofSr-Nd isotope data indicates that assimilation oflt5 lower crust could explain the steep vector onthe Sr-Nd isotope plot but assimilation of uppercrustal material is unlikely Oxygen isotope dataof clinopyroxene separates are consistent with amantle origin but permit some assimilation oflower crustal rocks In summary the data areconsistent with the hypothesis that the EFmagmas could have originated from a mantleplume source which is supported by geologicalarguments and that they underwent someassimilation of lower crustal material

Acknowledgements

We would like to thank Bruce Paterson whoprovided invaluable help during Laser ICP-MSmeasurements at the Large-Scale GeochemicalFacility supported by the European Community-Access to Research Infrastructure action of theImproving Human Potential Programme contractnumber HPRI-CT-1999-00008 awarded to ProfBJ Wood (University of Bristol) Gabi Stoschekand Torsten Vennemann are thanked for their helpwith oxygen isotope measurements Elmar Reitterexpertly assisted with preparation and measure-ments of radiogenic isotopes and MathiasWestphal helped with microprobe measurementsREE whole-rock analyses were kindly providedby the NERC facility at Royal HollowayUniversity of London Michael Marks is thankedfor his pleasant company during eldwork Veryconstructive reviews by Tom Andersen andAlexei S Rukhlov greatly helped to improve themanuscript We also thank Ian Coulson for nalediting Funding of this work by the DeutscheForschungsgemeinschaft (grant Ma-21351-2) isgratefully acknowledged

PETROGENESIS OF THE ERIKSFJORD BASALTS

849

References

Andersen T (1997) Age and petrogenesis of theQassiarsuk carbonatite-alkaline silicate volcaniccomplex in the Gardar rift South GreenlandMineralogical Magazine 61 499 ndash513

Armstrong JT (1991) Quantitative elemental analysisof individual microparticles with electron beaminstruments Pp 261 ndash315 in Electron ProbeQuantitation (KFJ Heinrich and DE Newburyeditors) Plenum Press New York and London

Arndt NT and Christensen U (1992) The role oflithospheric mantle in continental ood volcanismThermal and geochemical constraints Journal ofGeophysical Research 97 10967 ndash 10981

Baker JA Menzies MA Thirlwall MF andMacpherson CG (199 7) Pet rogenes is ofQuaternary Intraplate Vocanism Sanarsquoa YemenImplications for Plume-Lithosphere Interaction andPolybaric Melt Hybridization Journal of Petrology 38 1359 ndash 1390

Baker JA MacPherson CG Menzies MAThirlwall MF Al-Kadasi M and Mattey DP(2000) Resolving crustal and mantle contributions tocontinental ood volcanism Yemen constraintsfrom mineral oxygen isotope data Journal ofPetrology 41 1805 ndash1820

Beard JS Abitz RJ and Lofgren GE (1993)Experimental melting of crustal xenoliths fromKilbourne Hole New Mexico and implications forthe contamination and genesi s of magmasContributions to Mineralogy and Petrology 11588 ndash102

Bell K (2001) Carbonatites relationships to mantle-plume activity Geological Society of AmericaSpecial Paper 352 267 ndash290

Bell K and Blenkinsop J (1989) Neodymium andStrontium Isotope Geochemistry of Carbonatites Pp278 ndash300 in Carbonatites ndash Genesis and Evolution(K Bell editor) Unwin Hyman London

Bell K and Simonetti A (1996) Carbonat iteMagmatism and Plume Activity Implications fromthe Nd Pb and Sr Isotope Systematics of OldoinyoLengai Journal of Petrology 37 1321 ndash1339

Bell K and Tilton GR (2001) Nd Pb and Sr IsotopicCompositions of East African Carbonat itesEvide nce for Mant le Mixing and PlumeInhomogene ity Journa l of Petrology 42 1927 ndash1945

Benoit M Polve M and Ceuleneer G (1996) Traceelement and isotopic characterization of ma ccumulates in a fossil mantle diapir (Oman ophiolite)Chemical Geology 134 199 ndash 214

Bernstein S Kelemen PB Tegner C Kurz MDBlusztajn J and Kent Brooks C (1998) Post-breakup basaltic magmatism along the EastGreenland Tertiary rifted margin Earth and

Planetary Science Letters 160 845 ndash862Boynton WV (1984) Geochemistry of the rare earth

elements meteorite studies Pp 63 ndash114 in RareEarth Element Geochemistry (P Henderson editor)Elsevier Amsterdam

Carroll MR and Wyllie PJ (1990) The systemtonalite-H2O at 15 kbar and the genesis of calc-alkaline magmas American Mineralogist 75345 ndash 357

Chauvel C Hofmann AW and Vidal P (1992)HIMU-EM The French Polynesian connectionEarth and Planetary Science Letters 110 99 ndash119

Clayton RN and Mayeda TK (1963) The use ofbromine penta uoride in the extraction of oxygenfrom oxides and silicates for isotope analysisGeochimica et Cosmochimica Acta 27 43 ndash52

Coulson IM and Chambers AD (1996) Patterns ofzonation in rare-earth-bearing minerals in nephelinesyenites of the North Qoroq center South GreenlandThe Canadian Mineralogist 34 1163 ndash1178

Cox KG Bell JD and Pankhurst RJ (1979) TheInterpretation of Igneous Rocks George Allen andUnwin London

Cross W Iddings JP Pirsson LV and WashingtonHS (1903) Quantitative Classi cation of IgneousRocks University of Chicago Press

Dahl-Jensen T Thybo T Hopper H and Rosing M(1998) Crustal structure at the SE Greenland marginfrom wide-angle and normal incidence seismic dataTectonophysics 288 191 ndash 198

DePaolo DJ (1981) Trace element and isotopic effectsof combined wallrock assimilation and fractionalcrystallisation Earth and Planetary Science Letters53 189 ndash 202

De P a o lo D J (1 9 8 8 ) Neodymium Iso topeGeochemistry An Introduction Springer-VerlagNew York

Dobosi G Downes H Mattey D and Embey-IsztinA (1998) Oxygen isotope ratios of phenocrysts fromalkali basalts of the Pannonian basin Evidence foran O-isotopically homogeneous upper mantle be-neath a subduction-in uenced area Lithos 42213 ndash 223

Eiler JM (2001) Oxygen isotope variations of basalticlavas and upper mantle rocks Pp 319 ndash364 inStable Isotope Geochemistry (JW Valley and DRCole editor s) Reviews in Minera logy andGeochemistry 43 The Mineralogical Society ofAmerica Washington DC

Eiler JM Farley KA Valley JW Hauri E CraigH Hart SR and Stolper EM (1997) Oxygenisotope variations in ocean island basalt phenocrystsGeochimica et Cosmochimica Acta 61 2281 ndash2293

Eiler JM Crawford A Elliot T Farley KAValley JW and Stolper EM (2000a) Oxygenisotope geochemistry of oceanic-arc lavas Journal

850

R HALAMA ETAL

The role of crustal contamination

Large amounts of bulk assimilation of crustalmaterial which is relatively SiO2-rich comparedto a basaltic melt would drive the compositions tohigher SiO2 contents and quartz-normativecompositions Partial melts derived from crustalrocks would also be relatively SiO2-rich (Beard etal 1993 Carroll and Wyllie 1990) and wouldhave the same effect The ne-normative characterof the EF basalts (Table 1) suggests thatassimilation of crustal material especially ofsilica-rich upper crustal rocks was not signi cantwith the possible exception of sample EF 059

On the primitive mantle-normalized multi-element diagrams (Fig 4) Ti troughs are typicalfor crustal rocks especially for the Upper Crust(Rudnick and Fountain 1995) and they maytherefore be present in rocks extensively contami-nated by continental crust Ti troughs aregenerally lacking in the patterns of the EFbasalts and a major upper crustal contribution tothe magmas can therefore be excluded However

small amounts of assimilation of lower crustalmaterial cannot be ruled out because the averagelower crust has higher Ti contents

Steep trends on Sr-Nd isotope plots have beenattributed to assimilation of granulite-faciesgneisses (eg Bernstein et al 1998) AFCmodelling based on an equation developed byDePaolo (1981) was performed to test thehypothesis of small amounts of contaminationwith lower crustal material (Fig 9) The basalticmagma which represents one end-member for thecalculations has the isotopic signature of theisotopically most primitive samples from thepresent data set and from the literature(Table 6) Two possible contaminants repre-senting upper and lower crustal compositionswere selected for modelling (Table 6) Therepresentative samples are a granite from theJulianehaEcirc b batholith for the upper crust (sampleJG 02) and an average of ve granulite-faciesgneisses with low RbSr ratios from the Archaeancraton of West Greenland (Taylor et al 1984) for

TABLE 5 Trace element ratios of EF basalts compared to ocean island basalts(OIBs) representing EM-1 and HIMU mantle components and the lowercrust

SrNd LaNb ZrNb

EF basalts with eNd lt ndash 1 249 ndash396 120 ndash150 121 ndash157Il otildeacutemaussaq Group EF basalts 198 ndash292 104 ndash119 51 ndash53EM-1 90 ndash207 067 ndash111 39 ndash109HIMU 45 -163 063 ndash080 27 ndash 36Lower continental crust 30 16 136

Data sources EM-1 Humphris and Thompson (1983 Walvis Ridge) and Cliff et al(1991 Inaccessible Islands) HIMU Chauvel et al (1992 Tubuai Island) Lowercontinental crust Rudnick and Fountain (1995)

TABLE 6 End-member compositions used for AFC modelling

End-member Sr (87Sr86Sr)i Nd eNd(i) Data source(ppm) (T = 12 Ga) (ppm) (T = 12 Ga)

Primitive EF basaltic 507 070278 18 + 22 Sample EF 072magma Paslick et al (1993)Upper crust 300 071665 62 ndash83 Sample JG 02(JulianehaEcirc b Granite)Lower crust 747 070668 63 ndash 234 Taylor et al (1984)(Granulite facies gneiss)

The composition of the granulite-facies gneiss is an average of 5 samples with low RbSr ratios lt02

PETROGENESIS OF THE ERIKSFJORD BASALTS

847

the lower crust An in uence of Archaean rockson Gardar magmas has been demonstrated by Pbisotope studies (Taylor and Upton 1993) andseismic data were also interpreted to re ect theexistence of a wedge of Archaean crust thatextends southwards to Lindenow Fjord (Dahl-Jensen et al 1998) We have shown in theprevious sections on the basis of Ni trends thatfractionation of olivine was most relevant to themajor element evolution of the lavas whereasfractionation of other phases was apparently ofminor importance We therefore modelled theAFC processes with olivine fractionation usingmineral-melt partition coef cients of 0014 for Srand 00066 for Nd (Rollinson 1993)

The modelling of AFC processes is stronglydependent on the parameter r de ned as the ratioof mass assimilation rate and fractional crystal-lization rate (DePaolo 1981) The change inisotopic signature becomes enhanced for highervalues of r Reiners et al (1995) have shown thatr values might be very high and even gt1 in theearly stages of magmatic evolution at hightemperatures since crystallization can besuppressed by assimilation Therefore we calcu-

lated AFC processes with two r values of 05 and08 that represent a relatively high and a middle tolow assimilation rate to demonstrate the effects ofvariable r (Fig 9) The results indicate that AFCprocesses involving upper crust similar to theJulianehaEcirc b Granite cannot explain the trend of theEF basalts because Sr isotopic compositionswould shift to considerably more radiogenicvalues This result is consistent with theconclusions derived from trace element dataThe AFC curves for assimilation of lower crusthowever agree well with the bulk of the data Themost negative eNd values can be explained byAFC processes with intermediate assimilationrates (r = 05) involving assimilation of lt5lower crustal material With a greater degree ofassimilation and an r value of 08 lt4assimilation is indicated

Lower crustal rocks of igneous origin have anaverage d18O value of +75plusmn14 with an overallrange from +54 to +125 (Fowler and Harmon1990) 5 to 10 bulk assimilation of lower crustalmaterial with d18O = +75 into a typical mantle-derived magma with d18O = +57 results ind18O values of the contaminated magma from

-24

-20

-16

-12

-8

-4

0

4

0700 0703 0706 0709 0712 0715 0718

r = 08

r = 05

r = 05

r = 08

EF Basalts

Endmembers of AFC calculations

Upper Crust

Lower Crust

Primitive EF basaltic magma

(87Sr86Sr)i

e Nd

(i)

09

08

06

09

08

06

09

0806

09

08

07

07

04

Typical errorCHUR

BS

E

FIG 9 eNd(i) vs (87Sr86Sr)i diagram showing the results of AFC modelling End-member compositions are listed inTable 6 Olivine is the only fractionating mineral Partition coef cients for olivinemelt are from the compilation ofRollinson (1993) Curves were calculated for two values of r (r = rate of assimilation rate of fractionalcrystallization) Dashed lines and numbers in italics are for r = 05 solid lines and regular numbers are for r = 08

Values at the curves are for F values (fraction of melt remaining) with 10 tick marks

848

R HALAMA ETAL

+579 to +588 Therefore the d18O data cannotbe used to rule out small amounts of assimilationof lower crust To evaluate how much contamina-tion would be permissible from the presentoxygen isotope data we calculated how muchcontamination of a mantle-derived magma with18O = +57 with a contaminant within the lowerrange of lower crustal xenoliths (d18O = +67)would drive the oxygen isotopic composition tothe highest value of the EF basalts (d18O~+63) Although simple mixing shows that60 assimilation is permissible this amountwould drive the whole-rock geochemistry tomore silicic compositions than is observed in theEF basalts In Fig 8 there is no obvious trendfrom the basalts towards the Ketilidian basementsuggesting again that upper crust was notassimilated in signi cant amounts

Ratios of trace elements incompatible inbasaltic systems can be used to distinguishbetween crust and mantle reservoirs (egWeaver 1991) Based on the AFC modellingEF basalts with eNd(i) lt ndash1 are those which aremost likely in uenced by crustal contaminationThe relatively high LaNb and ZrNb ratios ofthese samples are consistent with assimilation oflower crustal material SrNd ratios are evenhigher than the average lower crustal composi-tion but SrNd ratios in those samples which weretaken as representative for the lower crust in theAFC modelling can reach values of 80 ie wellabove the highest EF basalt ratios Again theSrNd ratios support the hypothesis of some lowercrustal contamination

The absence of more primitive basaltic rockswith MgO gt8 wt might also indicate lowercrustal assimilation It suggests that magmas mayhave been retained at depth where they acquiredmore evolved compositions by fractional crystal-lization and had the opportunity to assimilatelower crustal material during crustal underplating(Upton et al 2003)

Summary and conclusions

The new geochemical data of the EriksfjordFormation basalts in the Gardar Provinceprovided in this study are used to placepetrogenetic constraints on the sources of theirparental magmas Based on trace-element patternsand Sr-Nd isotopic constraints there seems to beno indication that a DMM-like source played animportant role in magma genesis of the EFbasalts Clinopyroxene and whole-rock REE

patterns are more compatible with an OIB-likesource but we are not able to conclude whetherthe observed enrichment in REE re ects thesourcersquos signature or magmatic processes Theisotopically most primitive EF basalts have Sr-Ndisotopic compositions that are similar to Gardarcarbonatites Due to the isotopic similaritybetween carbonatites and OIBs many arethought to be related to mantle plumes (Bell2001) We therefore speculate that OIB-likemantle components from a plume source werealso involved in the parental EF magmas but wecannot exclude interaction with the SCLM Twodistinct mantle components one isotopicallydepleted and another one enriched could havecontributed to the EF magmas AFC modelling ofSr-Nd isotope data indicates that assimilation oflt5 lower crust could explain the steep vector onthe Sr-Nd isotope plot but assimilation of uppercrustal material is unlikely Oxygen isotope dataof clinopyroxene separates are consistent with amantle origin but permit some assimilation oflower crustal rocks In summary the data areconsistent with the hypothesis that the EFmagmas could have originated from a mantleplume source which is supported by geologicalarguments and that they underwent someassimilation of lower crustal material

Acknowledgements

We would like to thank Bruce Paterson whoprovided invaluable help during Laser ICP-MSmeasurements at the Large-Scale GeochemicalFacility supported by the European Community-Access to Research Infrastructure action of theImproving Human Potential Programme contractnumber HPRI-CT-1999-00008 awarded to ProfBJ Wood (University of Bristol) Gabi Stoschekand Torsten Vennemann are thanked for their helpwith oxygen isotope measurements Elmar Reitterexpertly assisted with preparation and measure-ments of radiogenic isotopes and MathiasWestphal helped with microprobe measurementsREE whole-rock analyses were kindly providedby the NERC facility at Royal HollowayUniversity of London Michael Marks is thankedfor his pleasant company during eldwork Veryconstructive reviews by Tom Andersen andAlexei S Rukhlov greatly helped to improve themanuscript We also thank Ian Coulson for nalediting Funding of this work by the DeutscheForschungsgemeinschaft (grant Ma-21351-2) isgratefully acknowledged

PETROGENESIS OF THE ERIKSFJORD BASALTS

849

References

Andersen T (1997) Age and petrogenesis of theQassiarsuk carbonatite-alkaline silicate volcaniccomplex in the Gardar rift South GreenlandMineralogical Magazine 61 499 ndash513

Armstrong JT (1991) Quantitative elemental analysisof individual microparticles with electron beaminstruments Pp 261 ndash315 in Electron ProbeQuantitation (KFJ Heinrich and DE Newburyeditors) Plenum Press New York and London

Arndt NT and Christensen U (1992) The role oflithospheric mantle in continental ood volcanismThermal and geochemical constraints Journal ofGeophysical Research 97 10967 ndash 10981

Baker JA Menzies MA Thirlwall MF andMacpherson CG (199 7) Pet rogenes is ofQuaternary Intraplate Vocanism Sanarsquoa YemenImplications for Plume-Lithosphere Interaction andPolybaric Melt Hybridization Journal of Petrology 38 1359 ndash 1390

Baker JA MacPherson CG Menzies MAThirlwall MF Al-Kadasi M and Mattey DP(2000) Resolving crustal and mantle contributions tocontinental ood volcanism Yemen constraintsfrom mineral oxygen isotope data Journal ofPetrology 41 1805 ndash1820

Beard JS Abitz RJ and Lofgren GE (1993)Experimental melting of crustal xenoliths fromKilbourne Hole New Mexico and implications forthe contamination and genesi s of magmasContributions to Mineralogy and Petrology 11588 ndash102

Bell K (2001) Carbonatites relationships to mantle-plume activity Geological Society of AmericaSpecial Paper 352 267 ndash290

Bell K and Blenkinsop J (1989) Neodymium andStrontium Isotope Geochemistry of Carbonatites Pp278 ndash300 in Carbonatites ndash Genesis and Evolution(K Bell editor) Unwin Hyman London

Bell K and Simonetti A (1996) Carbonat iteMagmatism and Plume Activity Implications fromthe Nd Pb and Sr Isotope Systematics of OldoinyoLengai Journal of Petrology 37 1321 ndash1339

Bell K and Tilton GR (2001) Nd Pb and Sr IsotopicCompositions of East African Carbonat itesEvide nce for Mant le Mixing and PlumeInhomogene ity Journa l of Petrology 42 1927 ndash1945

Benoit M Polve M and Ceuleneer G (1996) Traceelement and isotopic characterization of ma ccumulates in a fossil mantle diapir (Oman ophiolite)Chemical Geology 134 199 ndash 214

Bernstein S Kelemen PB Tegner C Kurz MDBlusztajn J and Kent Brooks C (1998) Post-breakup basaltic magmatism along the EastGreenland Tertiary rifted margin Earth and

Planetary Science Letters 160 845 ndash862Boynton WV (1984) Geochemistry of the rare earth

elements meteorite studies Pp 63 ndash114 in RareEarth Element Geochemistry (P Henderson editor)Elsevier Amsterdam

Carroll MR and Wyllie PJ (1990) The systemtonalite-H2O at 15 kbar and the genesis of calc-alkaline magmas American Mineralogist 75345 ndash 357

Chauvel C Hofmann AW and Vidal P (1992)HIMU-EM The French Polynesian connectionEarth and Planetary Science Letters 110 99 ndash119

Clayton RN and Mayeda TK (1963) The use ofbromine penta uoride in the extraction of oxygenfrom oxides and silicates for isotope analysisGeochimica et Cosmochimica Acta 27 43 ndash52

Coulson IM and Chambers AD (1996) Patterns ofzonation in rare-earth-bearing minerals in nephelinesyenites of the North Qoroq center South GreenlandThe Canadian Mineralogist 34 1163 ndash1178

Cox KG Bell JD and Pankhurst RJ (1979) TheInterpretation of Igneous Rocks George Allen andUnwin London

Cross W Iddings JP Pirsson LV and WashingtonHS (1903) Quantitative Classi cation of IgneousRocks University of Chicago Press

Dahl-Jensen T Thybo T Hopper H and Rosing M(1998) Crustal structure at the SE Greenland marginfrom wide-angle and normal incidence seismic dataTectonophysics 288 191 ndash 198

DePaolo DJ (1981) Trace element and isotopic effectsof combined wallrock assimilation and fractionalcrystallisation Earth and Planetary Science Letters53 189 ndash 202

De P a o lo D J (1 9 8 8 ) Neodymium Iso topeGeochemistry An Introduction Springer-VerlagNew York

Dobosi G Downes H Mattey D and Embey-IsztinA (1998) Oxygen isotope ratios of phenocrysts fromalkali basalts of the Pannonian basin Evidence foran O-isotopically homogeneous upper mantle be-neath a subduction-in uenced area Lithos 42213 ndash 223

Eiler JM (2001) Oxygen isotope variations of basalticlavas and upper mantle rocks Pp 319 ndash364 inStable Isotope Geochemistry (JW Valley and DRCole editor s) Reviews in Minera logy andGeochemistry 43 The Mineralogical Society ofAmerica Washington DC

Eiler JM Farley KA Valley JW Hauri E CraigH Hart SR and Stolper EM (1997) Oxygenisotope variations in ocean island basalt phenocrystsGeochimica et Cosmochimica Acta 61 2281 ndash2293

Eiler JM Crawford A Elliot T Farley KAValley JW and Stolper EM (2000a) Oxygenisotope geochemistry of oceanic-arc lavas Journal

850

R HALAMA ETAL

the lower crust An in uence of Archaean rockson Gardar magmas has been demonstrated by Pbisotope studies (Taylor and Upton 1993) andseismic data were also interpreted to re ect theexistence of a wedge of Archaean crust thatextends southwards to Lindenow Fjord (Dahl-Jensen et al 1998) We have shown in theprevious sections on the basis of Ni trends thatfractionation of olivine was most relevant to themajor element evolution of the lavas whereasfractionation of other phases was apparently ofminor importance We therefore modelled theAFC processes with olivine fractionation usingmineral-melt partition coef cients of 0014 for Srand 00066 for Nd (Rollinson 1993)

The modelling of AFC processes is stronglydependent on the parameter r de ned as the ratioof mass assimilation rate and fractional crystal-lization rate (DePaolo 1981) The change inisotopic signature becomes enhanced for highervalues of r Reiners et al (1995) have shown thatr values might be very high and even gt1 in theearly stages of magmatic evolution at hightemperatures since crystallization can besuppressed by assimilation Therefore we calcu-

lated AFC processes with two r values of 05 and08 that represent a relatively high and a middle tolow assimilation rate to demonstrate the effects ofvariable r (Fig 9) The results indicate that AFCprocesses involving upper crust similar to theJulianehaEcirc b Granite cannot explain the trend of theEF basalts because Sr isotopic compositionswould shift to considerably more radiogenicvalues This result is consistent with theconclusions derived from trace element dataThe AFC curves for assimilation of lower crusthowever agree well with the bulk of the data Themost negative eNd values can be explained byAFC processes with intermediate assimilationrates (r = 05) involving assimilation of lt5lower crustal material With a greater degree ofassimilation and an r value of 08 lt4assimilation is indicated

Lower crustal rocks of igneous origin have anaverage d18O value of +75plusmn14 with an overallrange from +54 to +125 (Fowler and Harmon1990) 5 to 10 bulk assimilation of lower crustalmaterial with d18O = +75 into a typical mantle-derived magma with d18O = +57 results ind18O values of the contaminated magma from

-24

-20

-16

-12

-8

-4

0

4

0700 0703 0706 0709 0712 0715 0718

r = 08

r = 05

r = 05

r = 08

EF Basalts

Endmembers of AFC calculations

Upper Crust

Lower Crust

Primitive EF basaltic magma

(87Sr86Sr)i

e Nd

(i)

09

08

06

09

08

06

09

0806

09

08

07

07

04

Typical errorCHUR

BS

E

FIG 9 eNd(i) vs (87Sr86Sr)i diagram showing the results of AFC modelling End-member compositions are listed inTable 6 Olivine is the only fractionating mineral Partition coef cients for olivinemelt are from the compilation ofRollinson (1993) Curves were calculated for two values of r (r = rate of assimilation rate of fractionalcrystallization) Dashed lines and numbers in italics are for r = 05 solid lines and regular numbers are for r = 08

Values at the curves are for F values (fraction of melt remaining) with 10 tick marks

848

R HALAMA ETAL

+579 to +588 Therefore the d18O data cannotbe used to rule out small amounts of assimilationof lower crust To evaluate how much contamina-tion would be permissible from the presentoxygen isotope data we calculated how muchcontamination of a mantle-derived magma with18O = +57 with a contaminant within the lowerrange of lower crustal xenoliths (d18O = +67)would drive the oxygen isotopic composition tothe highest value of the EF basalts (d18O~+63) Although simple mixing shows that60 assimilation is permissible this amountwould drive the whole-rock geochemistry tomore silicic compositions than is observed in theEF basalts In Fig 8 there is no obvious trendfrom the basalts towards the Ketilidian basementsuggesting again that upper crust was notassimilated in signi cant amounts

Ratios of trace elements incompatible inbasaltic systems can be used to distinguishbetween crust and mantle reservoirs (egWeaver 1991) Based on the AFC modellingEF basalts with eNd(i) lt ndash1 are those which aremost likely in uenced by crustal contaminationThe relatively high LaNb and ZrNb ratios ofthese samples are consistent with assimilation oflower crustal material SrNd ratios are evenhigher than the average lower crustal composi-tion but SrNd ratios in those samples which weretaken as representative for the lower crust in theAFC modelling can reach values of 80 ie wellabove the highest EF basalt ratios Again theSrNd ratios support the hypothesis of some lowercrustal contamination

The absence of more primitive basaltic rockswith MgO gt8 wt might also indicate lowercrustal assimilation It suggests that magmas mayhave been retained at depth where they acquiredmore evolved compositions by fractional crystal-lization and had the opportunity to assimilatelower crustal material during crustal underplating(Upton et al 2003)

Summary and conclusions

The new geochemical data of the EriksfjordFormation basalts in the Gardar Provinceprovided in this study are used to placepetrogenetic constraints on the sources of theirparental magmas Based on trace-element patternsand Sr-Nd isotopic constraints there seems to beno indication that a DMM-like source played animportant role in magma genesis of the EFbasalts Clinopyroxene and whole-rock REE

patterns are more compatible with an OIB-likesource but we are not able to conclude whetherthe observed enrichment in REE re ects thesourcersquos signature or magmatic processes Theisotopically most primitive EF basalts have Sr-Ndisotopic compositions that are similar to Gardarcarbonatites Due to the isotopic similaritybetween carbonatites and OIBs many arethought to be related to mantle plumes (Bell2001) We therefore speculate that OIB-likemantle components from a plume source werealso involved in the parental EF magmas but wecannot exclude interaction with the SCLM Twodistinct mantle components one isotopicallydepleted and another one enriched could havecontributed to the EF magmas AFC modelling ofSr-Nd isotope data indicates that assimilation oflt5 lower crust could explain the steep vector onthe Sr-Nd isotope plot but assimilation of uppercrustal material is unlikely Oxygen isotope dataof clinopyroxene separates are consistent with amantle origin but permit some assimilation oflower crustal rocks In summary the data areconsistent with the hypothesis that the EFmagmas could have originated from a mantleplume source which is supported by geologicalarguments and that they underwent someassimilation of lower crustal material

Acknowledgements

We would like to thank Bruce Paterson whoprovided invaluable help during Laser ICP-MSmeasurements at the Large-Scale GeochemicalFacility supported by the European Community-Access to Research Infrastructure action of theImproving Human Potential Programme contractnumber HPRI-CT-1999-00008 awarded to ProfBJ Wood (University of Bristol) Gabi Stoschekand Torsten Vennemann are thanked for their helpwith oxygen isotope measurements Elmar Reitterexpertly assisted with preparation and measure-ments of radiogenic isotopes and MathiasWestphal helped with microprobe measurementsREE whole-rock analyses were kindly providedby the NERC facility at Royal HollowayUniversity of London Michael Marks is thankedfor his pleasant company during eldwork Veryconstructive reviews by Tom Andersen andAlexei S Rukhlov greatly helped to improve themanuscript We also thank Ian Coulson for nalediting Funding of this work by the DeutscheForschungsgemeinschaft (grant Ma-21351-2) isgratefully acknowledged

PETROGENESIS OF THE ERIKSFJORD BASALTS

849

References

Andersen T (1997) Age and petrogenesis of theQassiarsuk carbonatite-alkaline silicate volcaniccomplex in the Gardar rift South GreenlandMineralogical Magazine 61 499 ndash513

Armstrong JT (1991) Quantitative elemental analysisof individual microparticles with electron beaminstruments Pp 261 ndash315 in Electron ProbeQuantitation (KFJ Heinrich and DE Newburyeditors) Plenum Press New York and London

Arndt NT and Christensen U (1992) The role oflithospheric mantle in continental ood volcanismThermal and geochemical constraints Journal ofGeophysical Research 97 10967 ndash 10981

Baker JA Menzies MA Thirlwall MF andMacpherson CG (199 7) Pet rogenes is ofQuaternary Intraplate Vocanism Sanarsquoa YemenImplications for Plume-Lithosphere Interaction andPolybaric Melt Hybridization Journal of Petrology 38 1359 ndash 1390

Baker JA MacPherson CG Menzies MAThirlwall MF Al-Kadasi M and Mattey DP(2000) Resolving crustal and mantle contributions tocontinental ood volcanism Yemen constraintsfrom mineral oxygen isotope data Journal ofPetrology 41 1805 ndash1820

Beard JS Abitz RJ and Lofgren GE (1993)Experimental melting of crustal xenoliths fromKilbourne Hole New Mexico and implications forthe contamination and genesi s of magmasContributions to Mineralogy and Petrology 11588 ndash102

Bell K (2001) Carbonatites relationships to mantle-plume activity Geological Society of AmericaSpecial Paper 352 267 ndash290

Bell K and Blenkinsop J (1989) Neodymium andStrontium Isotope Geochemistry of Carbonatites Pp278 ndash300 in Carbonatites ndash Genesis and Evolution(K Bell editor) Unwin Hyman London

Bell K and Simonetti A (1996) Carbonat iteMagmatism and Plume Activity Implications fromthe Nd Pb and Sr Isotope Systematics of OldoinyoLengai Journal of Petrology 37 1321 ndash1339

Bell K and Tilton GR (2001) Nd Pb and Sr IsotopicCompositions of East African Carbonat itesEvide nce for Mant le Mixing and PlumeInhomogene ity Journa l of Petrology 42 1927 ndash1945

Benoit M Polve M and Ceuleneer G (1996) Traceelement and isotopic characterization of ma ccumulates in a fossil mantle diapir (Oman ophiolite)Chemical Geology 134 199 ndash 214

Bernstein S Kelemen PB Tegner C Kurz MDBlusztajn J and Kent Brooks C (1998) Post-breakup basaltic magmatism along the EastGreenland Tertiary rifted margin Earth and

Planetary Science Letters 160 845 ndash862Boynton WV (1984) Geochemistry of the rare earth

elements meteorite studies Pp 63 ndash114 in RareEarth Element Geochemistry (P Henderson editor)Elsevier Amsterdam

Carroll MR and Wyllie PJ (1990) The systemtonalite-H2O at 15 kbar and the genesis of calc-alkaline magmas American Mineralogist 75345 ndash 357

Chauvel C Hofmann AW and Vidal P (1992)HIMU-EM The French Polynesian connectionEarth and Planetary Science Letters 110 99 ndash119

Clayton RN and Mayeda TK (1963) The use ofbromine penta uoride in the extraction of oxygenfrom oxides and silicates for isotope analysisGeochimica et Cosmochimica Acta 27 43 ndash52

Coulson IM and Chambers AD (1996) Patterns ofzonation in rare-earth-bearing minerals in nephelinesyenites of the North Qoroq center South GreenlandThe Canadian Mineralogist 34 1163 ndash1178

Cox KG Bell JD and Pankhurst RJ (1979) TheInterpretation of Igneous Rocks George Allen andUnwin London

Cross W Iddings JP Pirsson LV and WashingtonHS (1903) Quantitative Classi cation of IgneousRocks University of Chicago Press

Dahl-Jensen T Thybo T Hopper H and Rosing M(1998) Crustal structure at the SE Greenland marginfrom wide-angle and normal incidence seismic dataTectonophysics 288 191 ndash 198

DePaolo DJ (1981) Trace element and isotopic effectsof combined wallrock assimilation and fractionalcrystallisation Earth and Planetary Science Letters53 189 ndash 202

De P a o lo D J (1 9 8 8 ) Neodymium Iso topeGeochemistry An Introduction Springer-VerlagNew York

Dobosi G Downes H Mattey D and Embey-IsztinA (1998) Oxygen isotope ratios of phenocrysts fromalkali basalts of the Pannonian basin Evidence foran O-isotopically homogeneous upper mantle be-neath a subduction-in uenced area Lithos 42213 ndash 223

Eiler JM (2001) Oxygen isotope variations of basalticlavas and upper mantle rocks Pp 319 ndash364 inStable Isotope Geochemistry (JW Valley and DRCole editor s) Reviews in Minera logy andGeochemistry 43 The Mineralogical Society ofAmerica Washington DC

Eiler JM Farley KA Valley JW Hauri E CraigH Hart SR and Stolper EM (1997) Oxygenisotope variations in ocean island basalt phenocrystsGeochimica et Cosmochimica Acta 61 2281 ndash2293

Eiler JM Crawford A Elliot T Farley KAValley JW and Stolper EM (2000a) Oxygenisotope geochemistry of oceanic-arc lavas Journal

850

R HALAMA ETAL

+579 to +588 Therefore the d18O data cannotbe used to rule out small amounts of assimilationof lower crust To evaluate how much contamina-tion would be permissible from the presentoxygen isotope data we calculated how muchcontamination of a mantle-derived magma with18O = +57 with a contaminant within the lowerrange of lower crustal xenoliths (d18O = +67)would drive the oxygen isotopic composition tothe highest value of the EF basalts (d18O~+63) Although simple mixing shows that60 assimilation is permissible this amountwould drive the whole-rock geochemistry tomore silicic compositions than is observed in theEF basalts In Fig 8 there is no obvious trendfrom the basalts towards the Ketilidian basementsuggesting again that upper crust was notassimilated in signi cant amounts

Ratios of trace elements incompatible inbasaltic systems can be used to distinguishbetween crust and mantle reservoirs (egWeaver 1991) Based on the AFC modellingEF basalts with eNd(i) lt ndash1 are those which aremost likely in uenced by crustal contaminationThe relatively high LaNb and ZrNb ratios ofthese samples are consistent with assimilation oflower crustal material SrNd ratios are evenhigher than the average lower crustal composi-tion but SrNd ratios in those samples which weretaken as representative for the lower crust in theAFC modelling can reach values of 80 ie wellabove the highest EF basalt ratios Again theSrNd ratios support the hypothesis of some lowercrustal contamination

The absence of more primitive basaltic rockswith MgO gt8 wt might also indicate lowercrustal assimilation It suggests that magmas mayhave been retained at depth where they acquiredmore evolved compositions by fractional crystal-lization and had the opportunity to assimilatelower crustal material during crustal underplating(Upton et al 2003)

Summary and conclusions

The new geochemical data of the EriksfjordFormation basalts in the Gardar Provinceprovided in this study are used to placepetrogenetic constraints on the sources of theirparental magmas Based on trace-element patternsand Sr-Nd isotopic constraints there seems to beno indication that a DMM-like source played animportant role in magma genesis of the EFbasalts Clinopyroxene and whole-rock REE

patterns are more compatible with an OIB-likesource but we are not able to conclude whetherthe observed enrichment in REE re ects thesourcersquos signature or magmatic processes Theisotopically most primitive EF basalts have Sr-Ndisotopic compositions that are similar to Gardarcarbonatites Due to the isotopic similaritybetween carbonatites and OIBs many arethought to be related to mantle plumes (Bell2001) We therefore speculate that OIB-likemantle components from a plume source werealso involved in the parental EF magmas but wecannot exclude interaction with the SCLM Twodistinct mantle components one isotopicallydepleted and another one enriched could havecontributed to the EF magmas AFC modelling ofSr-Nd isotope data indicates that assimilation oflt5 lower crust could explain the steep vector onthe Sr-Nd isotope plot but assimilation of uppercrustal material is unlikely Oxygen isotope dataof clinopyroxene separates are consistent with amantle origin but permit some assimilation oflower crustal rocks In summary the data areconsistent with the hypothesis that the EFmagmas could have originated from a mantleplume source which is supported by geologicalarguments and that they underwent someassimilation of lower crustal material

Acknowledgements

We would like to thank Bruce Paterson whoprovided invaluable help during Laser ICP-MSmeasurements at the Large-Scale GeochemicalFacility supported by the European Community-Access to Research Infrastructure action of theImproving Human Potential Programme contractnumber HPRI-CT-1999-00008 awarded to ProfBJ Wood (University of Bristol) Gabi Stoschekand Torsten Vennemann are thanked for their helpwith oxygen isotope measurements Elmar Reitterexpertly assisted with preparation and measure-ments of radiogenic isotopes and MathiasWestphal helped with microprobe measurementsREE whole-rock analyses were kindly providedby the NERC facility at Royal HollowayUniversity of London Michael Marks is thankedfor his pleasant company during eldwork Veryconstructive reviews by Tom Andersen andAlexei S Rukhlov greatly helped to improve themanuscript We also thank Ian Coulson for nalediting Funding of this work by the DeutscheForschungsgemeinschaft (grant Ma-21351-2) isgratefully acknowledged

PETROGENESIS OF THE ERIKSFJORD BASALTS

849

References

Andersen T (1997) Age and petrogenesis of theQassiarsuk carbonatite-alkaline silicate volcaniccomplex in the Gardar rift South GreenlandMineralogical Magazine 61 499 ndash513

Armstrong JT (1991) Quantitative elemental analysisof individual microparticles with electron beaminstruments Pp 261 ndash315 in Electron ProbeQuantitation (KFJ Heinrich and DE Newburyeditors) Plenum Press New York and London

Arndt NT and Christensen U (1992) The role oflithospheric mantle in continental ood volcanismThermal and geochemical constraints Journal ofGeophysical Research 97 10967 ndash 10981

Baker JA Menzies MA Thirlwall MF andMacpherson CG (199 7) Pet rogenes is ofQuaternary Intraplate Vocanism Sanarsquoa YemenImplications for Plume-Lithosphere Interaction andPolybaric Melt Hybridization Journal of Petrology 38 1359 ndash 1390

Baker JA MacPherson CG Menzies MAThirlwall MF Al-Kadasi M and Mattey DP(2000) Resolving crustal and mantle contributions tocontinental ood volcanism Yemen constraintsfrom mineral oxygen isotope data Journal ofPetrology 41 1805 ndash1820

Beard JS Abitz RJ and Lofgren GE (1993)Experimental melting of crustal xenoliths fromKilbourne Hole New Mexico and implications forthe contamination and genesi s of magmasContributions to Mineralogy and Petrology 11588 ndash102

Bell K (2001) Carbonatites relationships to mantle-plume activity Geological Society of AmericaSpecial Paper 352 267 ndash290

Bell K and Blenkinsop J (1989) Neodymium andStrontium Isotope Geochemistry of Carbonatites Pp278 ndash300 in Carbonatites ndash Genesis and Evolution(K Bell editor) Unwin Hyman London

Bell K and Simonetti A (1996) Carbonat iteMagmatism and Plume Activity Implications fromthe Nd Pb and Sr Isotope Systematics of OldoinyoLengai Journal of Petrology 37 1321 ndash1339

Bell K and Tilton GR (2001) Nd Pb and Sr IsotopicCompositions of East African Carbonat itesEvide nce for Mant le Mixing and PlumeInhomogene ity Journa l of Petrology 42 1927 ndash1945

Benoit M Polve M and Ceuleneer G (1996) Traceelement and isotopic characterization of ma ccumulates in a fossil mantle diapir (Oman ophiolite)Chemical Geology 134 199 ndash 214

Bernstein S Kelemen PB Tegner C Kurz MDBlusztajn J and Kent Brooks C (1998) Post-breakup basaltic magmatism along the EastGreenland Tertiary rifted margin Earth and

Planetary Science Letters 160 845 ndash862Boynton WV (1984) Geochemistry of the rare earth

elements meteorite studies Pp 63 ndash114 in RareEarth Element Geochemistry (P Henderson editor)Elsevier Amsterdam

Carroll MR and Wyllie PJ (1990) The systemtonalite-H2O at 15 kbar and the genesis of calc-alkaline magmas American Mineralogist 75345 ndash 357

Chauvel C Hofmann AW and Vidal P (1992)HIMU-EM The French Polynesian connectionEarth and Planetary Science Letters 110 99 ndash119

Clayton RN and Mayeda TK (1963) The use ofbromine penta uoride in the extraction of oxygenfrom oxides and silicates for isotope analysisGeochimica et Cosmochimica Acta 27 43 ndash52

Coulson IM and Chambers AD (1996) Patterns ofzonation in rare-earth-bearing minerals in nephelinesyenites of the North Qoroq center South GreenlandThe Canadian Mineralogist 34 1163 ndash1178

Cox KG Bell JD and Pankhurst RJ (1979) TheInterpretation of Igneous Rocks George Allen andUnwin London

Cross W Iddings JP Pirsson LV and WashingtonHS (1903) Quantitative Classi cation of IgneousRocks University of Chicago Press

Dahl-Jensen T Thybo T Hopper H and Rosing M(1998) Crustal structure at the SE Greenland marginfrom wide-angle and normal incidence seismic dataTectonophysics 288 191 ndash 198

DePaolo DJ (1981) Trace element and isotopic effectsof combined wallrock assimilation and fractionalcrystallisation Earth and Planetary Science Letters53 189 ndash 202

De P a o lo D J (1 9 8 8 ) Neodymium Iso topeGeochemistry An Introduction Springer-VerlagNew York

Dobosi G Downes H Mattey D and Embey-IsztinA (1998) Oxygen isotope ratios of phenocrysts fromalkali basalts of the Pannonian basin Evidence foran O-isotopically homogeneous upper mantle be-neath a subduction-in uenced area Lithos 42213 ndash 223

Eiler JM (2001) Oxygen isotope variations of basalticlavas and upper mantle rocks Pp 319 ndash364 inStable Isotope Geochemistry (JW Valley and DRCole editor s) Reviews in Minera logy andGeochemistry 43 The Mineralogical Society ofAmerica Washington DC

Eiler JM Farley KA Valley JW Hauri E CraigH Hart SR and Stolper EM (1997) Oxygenisotope variations in ocean island basalt phenocrystsGeochimica et Cosmochimica Acta 61 2281 ndash2293

Eiler JM Crawford A Elliot T Farley KAValley JW and Stolper EM (2000a) Oxygenisotope geochemistry of oceanic-arc lavas Journal

850

R HALAMA ETAL

References

Andersen T (1997) Age and petrogenesis of theQassiarsuk carbonatite-alkaline silicate volcaniccomplex in the Gardar rift South GreenlandMineralogical Magazine 61 499 ndash513

Armstrong JT (1991) Quantitative elemental analysisof individual microparticles with electron beaminstruments Pp 261 ndash315 in Electron ProbeQuantitation (KFJ Heinrich and DE Newburyeditors) Plenum Press New York and London

Arndt NT and Christensen U (1992) The role oflithospheric mantle in continental ood volcanismThermal and geochemical constraints Journal ofGeophysical Research 97 10967 ndash 10981

Baker JA Menzies MA Thirlwall MF andMacpherson CG (199 7) Pet rogenes is ofQuaternary Intraplate Vocanism Sanarsquoa YemenImplications for Plume-Lithosphere Interaction andPolybaric Melt Hybridization Journal of Petrology 38 1359 ndash 1390

Baker JA MacPherson CG Menzies MAThirlwall MF Al-Kadasi M and Mattey DP(2000) Resolving crustal and mantle contributions tocontinental ood volcanism Yemen constraintsfrom mineral oxygen isotope data Journal ofPetrology 41 1805 ndash1820

Beard JS Abitz RJ and Lofgren GE (1993)Experimental melting of crustal xenoliths fromKilbourne Hole New Mexico and implications forthe contamination and genesi s of magmasContributions to Mineralogy and Petrology 11588 ndash102

Bell K (2001) Carbonatites relationships to mantle-plume activity Geological Society of AmericaSpecial Paper 352 267 ndash290

Bell K and Blenkinsop J (1989) Neodymium andStrontium Isotope Geochemistry of Carbonatites Pp278 ndash300 in Carbonatites ndash Genesis and Evolution(K Bell editor) Unwin Hyman London

Bell K and Simonetti A (1996) Carbonat iteMagmatism and Plume Activity Implications fromthe Nd Pb and Sr Isotope Systematics of OldoinyoLengai Journal of Petrology 37 1321 ndash1339

Bell K and Tilton GR (2001) Nd Pb and Sr IsotopicCompositions of East African Carbonat itesEvide nce for Mant le Mixing and PlumeInhomogene ity Journa l of Petrology 42 1927 ndash1945

Benoit M Polve M and Ceuleneer G (1996) Traceelement and isotopic characterization of ma ccumulates in a fossil mantle diapir (Oman ophiolite)Chemical Geology 134 199 ndash 214

Bernstein S Kelemen PB Tegner C Kurz MDBlusztajn J and Kent Brooks C (1998) Post-breakup basaltic magmatism along the EastGreenland Tertiary rifted margin Earth and

Planetary Science Letters 160 845 ndash862Boynton WV (1984) Geochemistry of the rare earth

elements meteorite studies Pp 63 ndash114 in RareEarth Element Geochemistry (P Henderson editor)Elsevier Amsterdam

Carroll MR and Wyllie PJ (1990) The systemtonalite-H2O at 15 kbar and the genesis of calc-alkaline magmas American Mineralogist 75345 ndash 357

Chauvel C Hofmann AW and Vidal P (1992)HIMU-EM The French Polynesian connectionEarth and Planetary Science Letters 110 99 ndash119

Clayton RN and Mayeda TK (1963) The use ofbromine penta uoride in the extraction of oxygenfrom oxides and silicates for isotope analysisGeochimica et Cosmochimica Acta 27 43 ndash52

Coulson IM and Chambers AD (1996) Patterns ofzonation in rare-earth-bearing minerals in nephelinesyenites of the North Qoroq center South GreenlandThe Canadian Mineralogist 34 1163 ndash1178

Cox KG Bell JD and Pankhurst RJ (1979) TheInterpretation of Igneous Rocks George Allen andUnwin London

Cross W Iddings JP Pirsson LV and WashingtonHS (1903) Quantitative Classi cation of IgneousRocks University of Chicago Press

Dahl-Jensen T Thybo T Hopper H and Rosing M(1998) Crustal structure at the SE Greenland marginfrom wide-angle and normal incidence seismic dataTectonophysics 288 191 ndash 198

DePaolo DJ (1981) Trace element and isotopic effectsof combined wallrock assimilation and fractionalcrystallisation Earth and Planetary Science Letters53 189 ndash 202

De P a o lo D J (1 9 8 8 ) Neodymium Iso topeGeochemistry An Introduction Springer-VerlagNew York

Dobosi G Downes H Mattey D and Embey-IsztinA (1998) Oxygen isotope ratios of phenocrysts fromalkali basalts of the Pannonian basin Evidence foran O-isotopically homogeneous upper mantle be-neath a subduction-in uenced area Lithos 42213 ndash 223

Eiler JM (2001) Oxygen isotope variations of basalticlavas and upper mantle rocks Pp 319 ndash364 inStable Isotope Geochemistry (JW Valley and DRCole editor s) Reviews in Minera logy andGeochemistry 43 The Mineralogical Society ofAmerica Washington DC

Eiler JM Farley KA Valley JW Hauri E CraigH Hart SR and Stolper EM (1997) Oxygenisotope variations in ocean island basalt phenocrystsGeochimica et Cosmochimica Acta 61 2281 ndash2293

Eiler JM Crawford A Elliot T Farley KAValley JW and Stolper EM (2000a) Oxygenisotope geochemistry of oceanic-arc lavas Journal

850

R HALAMA ETAL

of Petrology 41 229 ndash256Eiler JM Schiano P Kitchen N and Stolper EM

(2000b) Oxygen-isotope evidence for recycled crustin the sources of mid-ocean-ridge basalts Nature403 530 ndash 534

Emeleus CH and Upton BGJ (1976) The Gardarperiod in southern Greenland Pp 152 ndash181 inGeology of Greenland (A Escher and WS Watted ito rs) Geolo gical Survey of Greenl and Copenhagen

Fitton JG (1995) Coupled molybdenum and niobiumdepletion in continental basalts Earth and PlanetaryScience Letters 136 715 ndash 721

Fitton JG Saunders AD Norry MJ HardarsonBS and Taylor RN (1997) Thermal and chemicalstructure of the Iceland plume Earth and PlanetaryScience Letters 153 197 ndash 208

Floyd PA and Winchester JA (1975) Magma-typeand tectonic setting discrimination using immobileelements Earth and Planetary Science Letters 27211 ndash218

Fowler MB and Harmon RS (1990) The oxygenisotope composition of lower crustal granulitexenoliths Pp 493 ndash506 in Granulites and CrustalEvolution (DVielzeuf and P Vidal editors) KluwerAcademic Publishers Dordrecht The Netherlands

Garde AA Hamilton MA Chadwick B Grocott Jand McCaffrey KJW (2002) The Ketilidian orogenof South Greenland geochronology tectonicsmagmat ism and fore -arc accretion duringPalaeoproterozoic oblique convergence CanadianJournal of Earth Sciences 39 765 ndash793

Gibson SA Thompson RN Dickin AP andLeonardos OH (1995) High-Ti and Low-Ti ma cpotassic magmas key to plume-lithosphere interac-tions and continental ood-basalt genesis Earth andPlanetary Science Letters 136 149 ndash165

Gibson SA Thompson RN Weska RK and DickinAP (1997) Late-Cretaceous rift-related upwellingand melting of the Trinidade starting mantle-plumehead beneath western Brazil Contributions toMineralogy and Petrology 126 303 ndash314

Gibson SA Thompson RN Leonardos OHDickin AP and Mitchell JG (1999) The limitedextent of plume-lithosphere interactions duringcontinental ood-basalt genesis geochemical evi-dence from Cretaceous magmatism in southernBrazil Contributions to Mineralogy and Petrology 137 147 ndash 169

Goldstein SL OrsquoNions RK and Hamilton PJ(1984) A Sm-Nd isotopic study of the atmosphericdust and particulates from major river systems Earthand Planetary Science Letters 70 221 ndash236

Goodenough KM Upton BGJ and Ellam RM(2002) Long-term memory of subduction processesin the lithospheric mantle evidence from the

geochemistry of basic dykes in the Gardar Provinceof south Greenland Journal of the GeologicalSociety of London 159 705 ndash714

Harmon RS and Hoefs J (1995) Oxygen isotopeheterogeneity of the mantle deduced from global 18Osystematics of basalts from different geotectonicsettings Contributions to Mineralogy and Petrology 120 95 ndash114

Harris C Smith HS and le Roex AP (2000) Oxygenisotope composition of phenocrysts from Tristan daCunha and Gough Island lavas variation withfractional crystallization and evidence for assimila-tion Contributions to Mineralogy and Petrology 138 164 ndash175

Hart SR Hauri EH Oschmann LA and WhiteheadJA (1992) Mantle plumes and entrainment Isotopicevidence Science 256 517 ndash 520

Hawkesworth CJ Kempton PD Rogers NWEllam RM and van Calsteren PW (1990)Continental mantle lithosphere and shallow levelenrichment processes in the Earthrsquos mantle Earthand Planetary Science Letters 96 256 ndash268

Heaman LM and Machado N (1992) Timing andorigin of midcontinent rift alkaline magmatismNorth America evidence from the ColdwellComplex Contribution s to Minera logy andPetrology 110 289 ndash303

Hofmann A (1997) Mantle geochemistry the messagefrom oceanic volcanism Nature 385 219 ndash 229

Humphris SE and Thompson G (1983) Geochemistryof rare earth elements in basalts from the WalvisRidge implications for its origin and evolutionEarth and Planetary Science Letters 66 223 ndash242

Jacobsen SB and Wasserburg GJ (1980) Sm-Ndisotopic evolutio n of chondrit es Earth andPlanetary Science Letters 50 139 ndash155

Kalamarides RI (1986) High-temperature oxygenisotope fractionation among the phases of Kiglapaitintrusion Labrador Canada Chemical Geology 58303 ndash310

Larsen JG (1977) Petrology of the late lavas of theEriksfjord Formation Gardar province SouthGreenland Bul le tin Groslashnlands Geolog iskeUndersoslashgelse 125 31 pp

Le Maitre RW Bateman P Dudek A Keller J LeBas MJ Sabine PA Schmid R Soslashrensen HStreckeisen A Woolley AR and Zanettin B(1989) A Classi cation of Igneous Rocks andGlossary of Terms Blackwell Oxford UK

Lightfoot PC Sutcliffe RH and Doherty W (1991)Crustal contamination identi ed in KeweenawanOsler Group Tholeiites Ontario A trace elementperspective Journal of Geology 99 739 ndash760

Lindsley DH (1983) Pyroxene thermometry AmericanMineralogist 68 477 ndash 493

Lugmair GW and Marti K (1978) Lunar initial

PETROGENESIS OF THE ERIKSFJORD BASALTS

851

143Nd144Nd differential evolution of the lunar crustand mantle Earth and Planetary Science Letters 39349 ndash357

Mahoney JJ (1988) Deccan traps Pp 151 ndash 194 inContinental Flood Basalts (JD MacDougall edi-tor) Kluwer Academic Publishers Dordrecht TheNetherlands

Mattey D Lowry D and Macpherson C (1994)Oxygen isotope composition of mantle peridotiteEarth and Planetary Science Letters 128 231 ndash 241

McDonough WF and Sun SS (1995) The composi-tion of the Earth Chemical Geology 120 223 ndash 253

Middlemost EAK (1989) Iron oxidation ratios normsand the classi cation of volcanic rocks ChemicalGeology 77 19 ndash26

Molzahn M Reisberg L and Worner G (1996) OsSr Nd Pb O isotope and trace element data from theFerrar ood basalts Antarctica evidence for anenriched subcontinental lithospheric source Earthand Planetary Science Letters 144 529 ndash546

Morimoto N Fabrie J Ferguson AK GinzburgIV Ross M Seifert FA Zussman J Aoki Kand Gottardi G (1988) Nomenclature of pyroxenesMineralogical Magazine 52 535 ndash550

Nicholson SW Shirey SB Schulz KJ and GreenJC (1997) Rift-wide correlati on of 11 GaMidcontinent rift system basalts implications formultiple mantle sources during rift developmentCanadian Journal of Earth Sciences 34 504 ndash520

Nimis P and Vannucci R (1995) An ion microprobestudy of clinopyroxenes in websteritic and mega-crystic xenoliths from Hyblean Plateau (SE SicilyItaly) constraints on HFSEREESr fractionation atmantle depth Chemical Geology 124 185 ndash197

Oliveira EP and Tarney J (1995) Petrogenesis of theLate Proterozoic Curaca ma c dyke swarm Brazilasthenospheric magmatism associated with conti-nental collision Mineralogy and Petrology 5327 ndash48

Ormerod DS Hawkesworth CJ Rogers NWLeeman WP and Menzies MA (1988) Tectonicand magmatic transitions in the Western Great BasinUSA Nature 333 349 ndash353

Paces JB and Bell K (1989) Non-depleted sub-continental mantle beneath the Superior Province ofthe Canadian Shield Nd-Sr isotopic and traceelement evidence from Midcontinent Rift basaltsGeochimica et Cosmochimica Acta 53 2023 ndash2035

Paslick CR Halliday AN Davies GR Mezger Kand Upton BGJ (1993) Timing of proterozoicmagmatism in the Gardar Province southernGreenland Bulletin of the Geological Society ofAmerica 105 272 ndash278

Paslick CR Halliday AN James D and DawsonJB (1995) Enrichment of the continental lithosphereby OIB melts Isotopic evidence from the volcanic

province of northern Tanzania Earth and PlanetaryScience Letters 130 109 ndash126

Pearce NJG and Leng MJ (1996) The origin ofcarbonatites and related rocks from the Igaliko DykeSwarm Gardar Province South Greenland eldgeochemical and C-O-Sr-Nd isotope evidenceLithos 39 21 ndash 40

Perry FV Baldridge WS and DePaolo DJ (1987)Role of asthenosphere and lithosphere in the genesisof late cenozoic basaltic rocks from the Rio Granderift and adjacent regions of the Southwestern UnitedStates Journal of Geophysical Research 929193 ndash9213

Philpotts AR (1990) Principles of Igneous andMetamorphic Petrology Prentice Hall New JerseyUSA

Piper JDA Thomas DN Share S and Zhang Qi Rui(1999) The palaeomagnetism of (Mesoproterozoic)Eriksfjord Group red beds South Greenland multi-phase remagnetizat ion during the Gardar andGre nvil le episode s Geophysica l JournalInternational 136 739 ndash756

Poulsen V (1964) The sandstones of the PrecambrianEriksfjord Formation in South Greenland RapportGroslashnlands Geologiske Undersoslashgelse 2 16 pp

Price RC Gray CM Wilson RE Frey FA andTaylor SR (1991) The effects of weathering onrare-earth element Y and Ba abundances in Tertiarybasalts from southeastern Australia ChemicalGeology 93 245 ndash265

Ranloslashv J and Dymek RF (1991) Compositionalzoning in hydrothermal aegirine from fenites in theProterozoic Gardar Province South GreenlandEuropean Journal of Mineralogy 3 837 ndash 853

Reiners PW Nelson BK and Ghiorso MS (1995)Assimilation of felsic crust by basaltic magmathermal limits and extents of crustal contaminationof mantle-derived magmas Geology 23 563 ndash566

Roddick JC Sullivan RW and Dudas FO (1992)Precise calibration of Nd tracer isotopic compositionfor Sm-Nd studies Chemical Geology 97 1 ndash8

Rollinson H (1993) Using Geochemical DataEvaluation Presentation Interpretation LongmanGroup UK Limited London

Rudnick R and Fountain DM (1995) Nature andcomposition of the continental crust A lower crustalperspective Reviews of Geophysics 33 267 ndash 309

Rumble D and Hoering TC (1994) Analysis ofoxygen and sulfur isotope ratios in oxide and sul deminerals by spot heating with a carbon dioxide laserin a uorine atmosphrere Accounts of ChemicalResearch 27 237 ndash241

Schleicher H Kramm U Pernicka E SchidlowskiM Schmidt F Subramanian V Todt W andViladkar SG (1998) Enriched subcontinental uppermantle beneath Southern India Evidence from Pb

852

R HALAMA ETAL

Nd Sr and C-O isotopic studies on Tamil Naducarbonatites Journal of Petrology 39 1765 ndash1785

Sharp ZD (1990) A laser-based microanalyticalmethod for the in-situ determination of oxygenisotope ratios of silicates and oxides Geochimicaet Cosmochimica Acta 54 1353 ndash1357

Shirey SB Klewin KW Berg JH and CarlsonRW (1994) Temporal changes in the sources of ood basalts Isotopic and trace element evidencefrom the 1100 Ma old Keweenawan Mamainse PointFormation Ontario Canada Geochimica etCosmochimica Acta 58 4475 ndash4490

Simonetti A Goldstein SL Schmidberger SS andViladkar SG (1998) Geochemical and Nd Pb andSr isotope data from Deccan alkaline complexes ndashinferences for mantle sources and plume-lithosphereinteraction Journal of Petrology 39 1847 ndash1864

Steiger RH and Jager E (1977) Subcommision ongeochronology conventions of the use of decayconstants in geo- and cosmochronology Earth andPlanetary Science Letters 36 359 ndash362

Sun SS and McDonough WF (1989) Chemical andisotopic systematics of oceanic basalts implicationsfor mantle composition and processes Pp 313 ndash345in Magmatism in Ocean Basins (AD Saunders andMJ Norry editors) Special Publication 42Geological Society of London

Taylor PN and Upton BGJ (1993) Contrasting Pbisotopic compositions in two intrusive complexes ofthe Gardar Magmatic Province of South GreenlandChemical Geology 104 261 ndash 268

Taylor PN Jones NW and Moorbath S (1984)Isotopic assessment of relative contributions fromcrust and mantle sources to the magma genesis ofPrecam brian granit oid rocks PhilosophicalTransactions of the Royal Society of LondonA310 605 ndash625

Taylor SR and McLennan SM (1985) TheContinental Crust its Composition and Evolution Blackwell Scienti c Publications Oxford UK

Thomas DN and Piper JDA (1992) A revisedmagnetostratigraphy for the Mid-ProterozoicGarda r lava succ ess ion South Green land Tectonophysics 201 1 ndash16

Turner SP Hawkesworth CJ Gallagher KGStewart K Peate D and Mantovani M (1996)Mantle plumes ood basalts and thermal models formelt generation beneath continents assessment of aconductive heating model Journal of GeophysicalResearch 101 11503 ndash 11518

Upton BGJ (1996) Anorthosites and troctolites of theGardar Magmatic Province Pp 19 ndash34 in Petrologyand Geochemistry of Magmatic Suites of Rocks inthe Continental and Oceanic Crusts (D Demaiffeeditor) A volume dedicated to Professor JeanMichot Universite Libre de Bruxelles Royal

Museum for Central Africa (Tervuren)Upton BGJ and Emeleus CH (1987) Mid-

Proterozoic alkalin e magmatism in southernGreenland the Gardar province Pp 449 ndash471 inThe Alkaline Rocks (JG Fitton and BGJ Uptoneditors) Special Publication 30 Geological Societyof London

Upton BGJ Emeleus CH Heaman LMGoodenou gh KM and Finch AA (2003)Magmati sm of the mid-Protero zoic GardarProvince South Greenland chronology petro-genesis and geological setting Lithos 68 43 ndash65

Valley JW Kitchen N Kohn MJ Niendorf CRand Spicuzza MJ (1995) UWG-2 a garnet standardfor oxygen isotope ratios strategies for highprecis ion and accura cy with laser heatin gGeochimica et Cosmochimica Acta 59 5223 ndash5231

van Breemen O Aftalion M and Allaart JH (1974)Isotopic and geochronologic studies on granites fromthe Ketilidian Mobile Belt of South GreenlandBulletin of the Geological Society of America 85403 ndash412

Vroon PZ Lowry D Van Bergen MJ Boyce AJand Mattey DP (2001) Oxygen isotope systematicsof the Banda Arc Low d18O despite involvement ofsubducted continental material in magma genesisGeochimica et Cosmochimica Acta 65 589 ndash609

Waight T Baker J and Willigers B (2002) Rbisotope dilution analyses by MC-ICPMS using Zr tocorrect for mass fractionation towards improved Rb-Sr geochronology Chemical Geology 186 99 ndash116

Walsh JN Buckley F and Barker J (1981) Thesimultaneous determination of the rare-earth ele-ments in rocks using Inductively Coupled PlasmaSource Spectrometry Chemical Geology 33141 ndash153

Weaver BL (1991) The origin of ocean island basaltend-member compositions trace element and iso-topic constraints Earth and Planetary ScienceLetters 104 381 ndash397

Weaver BL Wood DA Tarney J and Joron J-L(1987) Geochemistry of ocean island basalts fromthe South Atlantic Ascension Bouvet St HelenaGough and Tristan da Cunha Pp 253 ndash267 in TheAlkaline Rocks (JG Fitton and BGJ Uptoneditors) Special Publicat ion 30 GeologicalSociety of London

Winchester JA and Floyd PA (1976) Geochemicalmagma type discrimination application to alteredand metamorphosed basic igneous rocks Earth andPlanetary Science Letters 28 459 ndash469

Zindler A and Hart SR (1986) Chemical geody-namics Annual Reviews of Earth and PlanetarySciences 14 493 ndash571

[Manuscript received 27 November 2002revised 12 June 2003]

PETROGENESIS OF THE ERIKSFJORD BASALTS

853