ORIGINAL PAPER

Gregor Markl Æ Lukas Baumgartner

pH changes in peralkaline late-magmatic fluids

Received: 24 April 2002 /Accepted: 25 July 2002 / Published online: 11 September 2002� Springer-Verlag 2002

Abstract The 1.15-Ga-old Ilimaussaq intrusive complexin South Greenland shows an extensive fractionationtrend from alkaline augite syenite to various varieties ofstrongly peralkaline, agpaitic nepheline syenites. Theperalkaline nepheline-bearing syenites crystallizedbetween ca. 900 and 450 �C at 1 kbar and they are cut bylate-magmatic hydrothermal veins with nepheline-absentassemblages of albite + aegirine + analcime ± sodalite± Na–Be-silicates (tugtupite, chkalovite, sørensenite) ±ussingite (NaAlSi3O8*NaOH). Based on fluid inclusionsand phase equilibria, these veins crystallized between 300and 500 �C at 1 kbar. Textures indicate that the hydro-thermal veins at least partly replaced earlier Ilimaussaqrocks. The occurrence of ussingite and tugtupite suggeststhat the late-magmatic fluids had strongly basic pH val-ues. Speciation calculations show that the pH in fluids ofthe system Na–Al–Si–O–H–Cl mainly depends on theNa/Cl ratio and, to a lesser degree, on salinity and tem-perature. If the Na/Cl ratio is greater than 1, pH (at400 �C and 1 kbar, where neutrality is about at pH 5) liesbetween 7 and 12. Because Na/Cl tends to decrease in thefinal stages of magmatism and during crystallization ofthe vein assemblage, pH of late-magmatic fluids gener-ally should become more acidic, and only two processescan increase Na/Cl and, thus, pH: dissolution of aCl-poor or Cl-free Na silicate or unmixing of anHCl-enriched vapour phase. Field observations andmicrotextures suggest that replacement reactions areresponsible for the change to basic pH at least in somealteration assemblages. While replacement of 1 molnepheline by 1 mol analcime would not alter the pH, the

volume-conserving reaction 1:85 Neþ 2:3 H2Oþ 0:19H4SiO4 ¼ 1:02 Anlþ 0:83 Naþ þ 0:83 AlðOHÞ�4 can beused to model the replacement process quantitatively,provided it occurred in a more or less closed system.Progress of this reaction leads to successively increasingpH of the fluid during fluid–rock interaction andstabilizes minerals such as ussingite and tugtupite.Transferring the two processes to a larger scale, it isproposed that the extreme ‘hyper-agpaitic’ assemblagesat Ilimaussaq or at the Kola peninsula, which includecopious amounts of very basic, water-soluble mineralssuch as trona, villiaumite or thermonatrite, are formedeither in this way by autometasomatic reactions oflate-magmatic fluids or melts (or supercritical fluid-melt-mixtures) with earlier crystallized rocks of the sameplutonic complex or by large-scale vapour unmixing inthe very final stages of magmatism.

Introduction

Agpaitic rocks, i.e. peralkaline rocks with complex Na–Zr silicates such as eudialyte or rinkite, are still enigmaticmembers of the peralkaline rock family. Their conditionsof formation, their relation to miaskitic rocks and theextreme enrichment of some incompatible elements arestill poorly understood (Sørensen 1997; Bailey et al.2001). They deserve, however, interest due to their po-tential as Zr, P, Nb and REE deposits, due to their wealthof rare or even unique minerals and due to their extremefractionation trend, which leads – possibly in a transitionthrough a supercritical fluid-rich melt as described bySowerby and Keppler (2001), to some extreme pegmatitecompositions, or as suggested by Bureau and Keppler(1999), to high-pressure basaltic melts – to the so-calledhyperagpaitic rocks, which could be regarded as the mostfractionated magmatic rocks known (Khomyakov 1995,2001; Sørensen and Larsen 2001). In addition to a gen-erally extremely Na-rich mineralogy, these hyperagpaiticrocks consist to a certain extent of very water-soluble

Contrib Mineral Petrol (2002) 144: 331–346DOI 10.1007/s00410-002-0401-6

G. Markl (&)Institut fur Mineralogie, Petrologie und Geochemie,Eberhard-Karls-Universitat, Wilhelmstrasse 56,Tubingen, GermanyE-mail: markl@uni-tuebingen.de

L. BaumgartnerInstitut de Mineralogie et Petrographie,BFSH-2, Lausanne, Switzerland

Editorial responsibility: J. Hoefs

minerals such as trona, thermonatrite or villiaumite,some of which indicate a very basic pH of the fluids orliquids from which were they formed. Also, ussingite(NaAlSi3O8*NaOH) is common in these most differen-tiated rocks (Sørensen and Larsen 2001), the composi-tion of which points to a basic pH during formation, too.The detailed mechanisms, however, by which these fluidsbecome so basic, are still unknown.

Late-magmatic veins consisting of analcime, albite,aegirine, in places ussingite and other minerals arecommonly associated with highly differentiated agpaiticcomplexes (e.g. Engell et al. 1971; Khomyakov 1995;Sørensen 1997; Horvath and Horvath-Pfenninger 2000;Pekov 2000; Sørensen and Larsen 2001). In the Ili-maussaq complex of South Greenland, the most differ-entiated lujavritic rocks crystallized at temperatures aslow as 450 �C (Sood and Edgar 1970; Larsen 1976;Larsen and Sørensen 1987), while the analcime + albite+ acmite ± ussingite ± tugtupite veins formed between300 and 400 �C based on fluid inclusion homogenizationtemperatures and phase equilibria (Sobolev et al. 1970;Konnerup-Madsen and Rose-Hansen 1982; Markl2001a). Hence, it appears likely that these hydrothermalveins formed from fluids related to the hyper-agpaiticrocks, e.g. from fluids expelled from the hyperagpaitesduring crystallization. The present contribution dealswith the question of what parameters affect the pH of aperalkaline solution and how these fluids became asbasic as they obviously were.

Regional geology

The Ilimaussaq intrusion is part of the 1.1–1.3-Ga-oldGardar failed rift province of South Greenland (Uptonand Emeleus 1987). During this period, ten major plu-tonic complexes of gabbroic and nepheline syenitic toquartz-saturated granitic rocks intruded into a basementconsisting of early Proterozoic granites and gneisses(Emeleus and Upton 1976), which are unconformablyoverlain by early Gardar basalts and sandstones of theEriksfjord formation (Poulsen 1964). The association ofbasalts, gabbroic, and granitoid intrusions with rela-tively abundant anorthosite xenoliths and rare carbon-atites has been interpreted to reflect large-scale meltingprocesses of asthenospheric mantle. During fractiona-tion in a deep-seated magma chamber, probably at thecrust–mantle boundary, these alkali basaltic melts arebelieved to have formed massive-type anorthosites asflotation cumulates, and later alkaline to peralkalinemelts as residual liquids (Larsen and Sørensen 1987).

According to Ussing (1912) and Larsen and Sørensen(1987), the Ilimaussaq igneous complex was formed bysuccessive intrusion of three melt batches released fromthe top of such a deep-seatedmagma chamber. Thesemeltbatches intruded to approximately the same depth ofabout 3–4 km (1 kbar, Konnerup-Madsen and Rose-Hansen 1984). The earliest intrusive phase (stage I, Larsenand Sørensen 1987), the augite syenite of Ussing (1912)

and Ferguson (1964), is found both as a few-hundred-meters-thick shell along the southern and south-easternmargins of the intrusion and as an about 150-m-thick lidon the stratigraphic top of the intrusion (Fig. 1). Theaugite syenite lid was later intruded by an alkali granite(stage II; Nielsen and Steenfelt 1979; Larsen and Sørensen1987). In the peralkaline stage III, the so-called ‘sodalitefoyaite’ crystallized from the roof downwards, while the‘naujaite’ represents a flotation cumulate of sodalite(Ussing 1912; Ferguson 1964). Both rock types arenepheline- and sodalite-bearing syenites.Mafic cumulatescorresponding to this flotation cumulate are not exposed,but their existence is indicated by a strong positive gravityanomaly beneath Ilimaussaq (Blundell 1978; Forsbergand Rasmussen 1978). The layered, agpaitic kakortokitesaccumulated at the lowermost stratigraphic position ofthe present outcrops of the complex and are separatedfrom the overlying naujaites and foyaites by large massesof residual liquids: the lujavrites (Ussing 1912; Ferguson1964). Again, kakortokites and lujavrites are texturalvarieties of nepheline- and sodalite-bearing syenites. Thelujavrites, additionally, are found as crosscutting veinsthroughout the whole complex. Again, kakortokite andlujavrite are chemically and texturally different varietiesof agpaitic nepheline syenite. Typical monitors of frac-tionation, such as Fe/Mg, Ca/(Na+K) or contents ofincompatible elements, indicate that the lujavrites are themost evolved rock types of the complex. K/Rb ratioscontinuously decrease during fractionation from around500 in the augite syenite to 35 in the most evolvedlujavriteswhileZr/Hf andCl/Br continuously rise from45to 97 and from 170 to 15,000, respectively (Bailey et al.2001). Total rare earth element contents as well as LREE/HREE ratios show a general increase with fractionation,reaching extreme values of almost 10 in the most evolvedlujavrites (Bailey et al. 2001). The petrology and mineralchemistry of the major rock-forming minerals in thevarious rock types have been discussed by Marks andMarkl (2001), Markl et al. (2001), Larsen (1976, 1977,1981).

Late-stage phenomena in Ilimaussaq include veins ofimmiscible peralkaline liquids formed by liquid immis-cibility of most fractionated lujavrite (Markl 2001a),hydrothermal veins consisting of analcime + albite +acmite ± ussingite ± tugtupite ± chkalovite cuttingacross all other rock types of the complex (Engell et al.1971; Markl 2001b) and various special types of ‘lujav-rites’ such as a murmanite [Na3(Ti,Nb)4(Si2O7)2O4

4H2O]- and a villiaumite (NaF)-bearing variety, whichare interpreted as metasomatized country rocks(Sørensen and Larsen 2001).

Petrography and mineral chemistry

Petrography

There are several types of hydrothermal veins in theIlimaussaq complex, which differ in their mineralogy

332

and in their mode of emplacement (Engell et al. 1971):some cut across their host rocks with not much macro-scopically visible alteration (although microscopically,alteration in the host rocks may be abundant!), andother veins and lenses show irregular contacts towardsthe surrounding rocks, suggesting an origin by replace-ment (‘replacement veins’). All of these structures have amajor modal content of analcime, albite and aegirine incommon, but they may additionally contain sodalite,ussingite or Be-bearing minerals such as chkalovite,sørensenite or tugtupite (Engell et al. 1971; Markl2001b). Chkalovite and tugtupite occur both in thestraight and in the irregular veins, whereas ussingitepredominantly occurs in lenses or short replacementveins. Newly formed nepheline is notably absent from allof these veins.

Because this will be important in the following dis-cussion, we state that in addition to the microtexturessuch as the one from Fig. 2, at least some of the mac-roscopic veins and lenses indicate an approximatelyvolume-conserving replacement of the former magmaticmineralogy by later hydrothermal assemblages. For themicrotextures, this notion is based on the replacementtexture itself (the replacement textures closely follow theformer outline of grain boundaries) and based on thetexture of the surrounding silicates, which show noindication of any volume change such as deformation-induced subgrain formation or dissolution features. In

some of the macroscopic veins – namely those with theirregular contacts towards the host rocks – the mostimportant observation supporting constant volume re-placement is the continuous gradation from unreplacedrocks through progressively more replaced rocks tocompletely replaced rocks in the centre of the veins. Inthe outer zones of this structure, the original texture maystill be visible through the ‘network’ of replacementtextures. Growth in an open fracture may be an addi-tional process, but in the about two to five times morevoluminous marginal zones it is certainly not. The veinswith straight contacts towards the host rocks may betypical fracture fillings and these may have originated inanother way and not by constant-volume replacement.

Whereas analcime, albite and sodalite typically occuras anhedral grains of a few micrometres to some centi-metres size, aegirine invariably forms euhedral,prismatic crystals of a bright green colour. The needle-shaped crystals may reach a length of 1 cm and arecommonly radially oriented. In rare vugs, also analcimeand sodalite may form crystals of up to some centimetresin size. All of these minerals are probably co-precipi-tated and in apparent textural equilibrium.

In rocks adjacent to the veins, magmatic nepheline istypically partially altered to analcime, sodalite or amixture of both. Figure 2 shows a zoned reaction tex-ture, where at the grain boundary between magmaticcrystals of nepheline and sodalite in a nepheline-bearing

Fig. 1 Simplified geologicalmap of the Ilimaussaq intrusivecomplex, South Greenland,after Ussing (1912) andFerguson (1964)

333

host rock (sodalite foyaite) a symplectite of analcimewith secondary sodalite and a pure analcime layerdeveloped. This mode of occurrence is typical in thewhole Ilimaussaq complex, but it is the more extensivethe closer a hydrothermal vein or lens is, and it is bestdeveloped around the ‘replacement veins’. This indicatesthat it is related to the infiltration of an aqueous fluid. Itis important to note that this fluid must be late-mag-matic because there exists no sign of any post-magmatictectonic, metamorphic or hydrothermal activity in theIlimaussaq area.

Mineral chemistry

Analytical techniques

Wavelength-dispersive microprobe analyses were performed on aCameca SX100 at the University of Freiburg, Germany, and on aJEOL 8900 at the University of Tubingen, Germany. Naturalstandards were used for calibration. Analytical conditions were10 nA and 15 kV, with counting times of 20 s for all elements on thepeak and 10 s on the background. Raw data were processed byprocedures described in Pouchor and Pichoir (1984, Cameca) andArmstrong (1991, JEOL). Representative electron microprobeanalyses are reported in Tables 1 and 2.

Albite

The feldspar in the veins and secondary feldspar in rocks adjacentto the veins is invariably albite with more than 95% albite com-ponent and with no measurable An component.

Analcime

The chemical composition of analcime does not show significantdifferences based on its mode of occurrence. Analcime in hydro-thermal veins and in the country rocks (replacing nepheline) showsanalytical totals of around 91–93 wt% and is close to ideal endmember composition. Na + Al against Si exchange is insignificant,but the exchanges of (Si+3 vacancies) against 4Na (i.e. a quartzcomponent) and of (Al+2 vacancies against 3Na) are responsiblefor the low Na contents, which may be as low as 0.8 atoms performula unit because it was ensured that no Na evaporation duringanalysis occurred. No other elements such as K or Fe were detectedin amounts >0.1 wt% in analcime.

Aegirine

Hydrothermally formed clinopyroxene in Ilimaussaq has compo-sitions close to the aegirine end member composition, but contains– in some samples – a significant component of jadeite as well. The

Fig. 2 Typical texture from the host rocks (here a sodalite foyaite)close to a hydrothermal vein: magmatic nepheline and sodalite 1crystals are replaced along grain boundaries by newly grownanalcime (two different textures, analcime 1 and 2, but of same age)and sodalite 2. Sample GM 1346 from Schwinn (1999), scale bar is200 lm

Table 1 Representativemicroprobe analyses of miner-als from Ilimaussaq hydrother-mal veins. Clinopyroxeneformula calculation is based onsix oxygens and four cations inorder to evaluate the content ofFe3+

Analysis no. 1374-376 1374-378 1374-380 1257-391 1257-397 1257-400 1401-2-267 1401-2-268Mineral Anl Anl Anl Anl Anl Ab Cpx Cpx

SiO2 50.84 50.33 51.38 49.76 56.34 68.86 53.15 52.97TiO2 0.00 0.00 0.00 0.00 0.00 0.00 1.14 0.38Al2O3 30.54 29.60 30.69 29.97 24.90 19.94 0.71 0.88FeO 0.31 0.27 0.00 0.03 0.06 0.04 30.06 31.17MnO 0.00 0.00 0.01 0.03 0.04 0.00 0.54 0.43MgO 0.00 0.00 0.00 0.00 0.02 0.00 0.05 0.12CaO 0.01 0.02 0.03 0.00 0.01 0.00 0.05 0.15Na2O 12.06 12.95 11.74 13.06 12.49 11.78 13.70 14.06K2O 0.05 0.06 0.04 0.03 0.07 0.13 0.02 0.00Cl 0.00 0.00 0.00 0.00 0.01 0.00 0.00 0.00Total 93.83 93.22 93.91 92.87 93.94 100.75 99.42 100.17

No. of Oformula isbased on

6 6 6 6 6 8 6 6

Si 1.82 1.82 1.83 1.81 2.00 2.99 1.98 1.99Al 1.29 1.26 1.29 1.28 1.04 1.02 0.03 0.04Ti 0.00 0.00 0.00 0.00 0.00 0.00 0.03 0.01Fe3+ 0.00 0.00 0.00 0.00 0.00 0.00 0.93 0.94Mg 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01Fe 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00Mn 0.00 0.00 0.00 0.00 0.00 0.00 0.02 0.01Ca 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01Na 0.84 0.91 0.81 0.92 0.86 0.99 0.99 0.99K 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.00Cl 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00Sum 3.95 4.00 3.93 4.01 3.91 5.00 4.00 4.00

334

maximum measured values are about 5% jadeite component inclinopyroxene from ussingite-free hydrothermal veins and up to19% in some ussingite-bearing, hydrothermally altered lujavritesfrom country rocks (see Markl et al. 2001).

Sodalite

Primary magmatic sodalite in Ilimaussaq intrusive rocks (sodalite 1in Fig. 2) has typically very low contents of K, Ca or SO3 and isregarded as an almost pure end member. In sodalite from hydro-thermal veins, Cl content varies between 1.85 and 2 atoms performula unit (afu) and K is below 0.1 per formula unit, secondarysodalite in the reaction textures, such as sodalite 2 in Fig. 2, mayhave up to 0.2 afu K, up to 0.25 afu SO3 and about 1.75 afu Cl(Table 2). Still, this is very close to end member composition.

Nepheline

The composition of magmatic nepheline is highly variable in therocks of the Ilimaussaq complex. Accordingly, also the relics ofthese magmatic nephelines in the alteration and replacement tex-ture discussed here (such as the one on Fig. 2) have widely variablecompositions: between Ne67Ks17Qtz16 to Ne74Ks11Qtz15 in theaugite syenite, a wider range in SiO2 contents and higher Ks con-tents up to Ne74Ks25Qtz3 in sodalite foyaite, naujaite and kakor-tokite and very SiO2-poor compositions in lujavrite (see also Marklet al. 2001). The CaO content in nepheline of all rock types isalways below 1 wt%. The spread in excess silica is related to patchycompositional differences within single grains and between grains,but not to a continuous growth zonation.

Fluid–solid phase equilibria

Principles of the calculations

The hydrothermal veins in the Ilimaussaq complexand the fluids from which they formed are investigated

in the simplified model system Na2O–Al2O3–SiO2–H2O–NaCl in which a H2O–NaCl fluid and the endmember phases analcime, nepheline, albite, sodaliteand jadeite (and, if necessary, halite) are discussed.More components such as Be or Fe would, in theabsence of more solid phase constraints, render thecalculations non-unique because the lack of solidphase equilibria would lead to a lack of constraintsnecessary to solve the linear system of equations. Forsolid phases principally present in at least some of theveins – such as tugtupite or acmite – the thermody-namic data are missing and, hence, they also do notprovide the necessary constraints. At a given pressure,temperature and total salinity, the three solid phasesobserved in the veins (albite, analcime, jadeite-bearingaegirine) buffer three component activities or, inthe fluid, concentrations of dissolved species accordingto the formalism of Eugster and Baumgartner(1987) using the program SOLUBLE (Roselle andBaumgartner 1995). It should be noted that theDebye–Huckel equation used for calculation of acti-vity coefficients for aqueous charged species is limitedto an ionic strength of less than about 1 in the system(Hauzenberger et al. 2001). For ionic strength largerthan 1, the uncertainty of the calculations increases.Nevertheless, it is the only model available at this timefor these types of solutions.

A combination of three of the following equilibriainvolving solid phases, together with a total mass con-straint such as total Cl concentration, constrains theactivities of the basis dissolved species Cl–, Na+,Al(OH)4

– and H4SiO4 (=SiO2, aq.), which are thedominant species for the components and are at theconditions of interest:

Table 2 Representative mi-croprobe analyses of mineralsfrom the texture displayed inFig. 2

Analysis no. 1346S1-88 1346rea-79 N1346M1-2 1346M1-1 1346N1-3 1346N1-7Mineral Sod I Sod II Anl I Anl II Ne Ne

SiO2 37.18 38.52 55.69 57.19 46.50 46.66Al2O3 35.68 35.16 24.26 25.16 32.40 31.83FeO 0.05 0.00 0.00 0.00 1.03 0.95MnO 0.00 0.00 0.05 0.00 0.00 0.02MgO 0.01 0.01 0.01 0.00 0.00 0.01CaO 0.00 0.04 0.02 0.00 0.01 0.00Na2O 24.51 23.00 13.66 11.84 16.07 16.59K2O 0.03 0.04 0.09 0.00 5.12 4.61SO3 0.04 1.17 0.00 0.00 0.00 0.00Cl 7.51 6.54 0.00 0.00 0.00 0.00Total 105.01 104.48 93.82 94.19 101.13 100.67

No. of Oformula isbased on

25 25 6 6 32 32

Si 5.76 5.96 1.99 2.01 8.74 8.80Al 6.52 6.41 1.02 1.04 7.17 7.07Mg 0.00 0.00 0.00 0.00 0.15 0.14Fe 0.01 0.00 0.00 0.00 0.00 0.00Mn 0.00 0.00 0.00 0.00 0.00 0.00Ca 0.00 0.01 0.00 0.00 0.00 0.00Na 7.37 6.90 0.95 0.81 5.85 6.07K 0.00 0.01 0.00 0.00 1.23 1.11SO3 0.00 0.14 0.00 0.00 0.00 0.00Cl 1.97 1.72 0.00 0.00 0.00 0.00Sum 19.66 19.29 3.97 3.87 23.14 23.19

335

NaAlSi3O8 þ 8H2O ¼ Naþ þAlðOHÞ�4 þ 3H4SiO4 ð1Þ

NaAlSi2O6 þ 6H2O ¼ Naþ þAlðOHÞ�4 þ 2H4SiO4 ð2Þ

NaAlSi2O6 �H2Oþ 5H2O ¼ Naþ þAlðOHÞ�4 þ 2H4SiO4 ð3Þ

Na8Al6Si6O24Cl2þ 24H2O¼ 8Naþ þ 6AlðOHÞ�4 þ 6H4SiO4þ 2Cl�

ð4Þ

NaAlSiO4þ 4H2O¼Naþ þAlðOHÞ�4 þH4SiO4 ð5Þ

Of these reactions, only three are linearly independentbecause of the co-linearity of albite, jadeite, nephelineand analcime in composition space. However, nephelineis strongly K-bearing and clinopyroxene contains highamounts of Fe3+ and, hence, their co-linearity is ap-parent, but not real. If for the idealized calculations onlytwo solid phase constraints are used (which is typically

necessary because of the compositional co-linearity ofthe idealized system), then one has to add another totalmass constraint, such as the total molality of Na inthe system, for example. Another eight fluid species arerelated to the above basis species by the followingequilibria among dissolved species:

NaCl ¼ Naþ þ Cl� ð6Þ

NaOH ¼ Naþ þOH� ð7Þ

AlðOHÞþ2 þ 2OH� ¼ AlðOHÞ�4 ð8Þ

NaAlðOHÞ4 ¼ Naþ þAlðOHÞ�4 ð9Þ

HSiO�3 þ 2H2O ¼ H4SiO4 þOH� ð10Þ

NaH5SiO5 ¼ Naþ þOH� þH4SiO4 ð11Þ

Hþ þOH� ¼ H2O ð12Þ

Table 3 Speciation of ahydrothermal fluid coexistingwith albite and a clinopyroxenewith 5 mol% jadeite compo-nent; see the text for details ofthe calculations . Pressure ofcalculations: 1 kbar

T Na/Cl pH Cltot Natot mH4SiO4 mHSiO3– mNa+ mNaOH mNaCl

300 0.002 1.46 5 0.01 0.006 4.1E–09 0.005 1.4E–11 0.005300 0.003 1.54 3 0.01 0.006 4.0E–09 0.006 1.9E–11 0.004300 0.010 1.73 1 0.01 0.007 3.8E–09 0.007 3.8E–11 0.003300 0.020 1.87 0.5 0.01 0.007 3.9E–09 0.008 5.6E–11 0.002300 0.020 0.97 5 0.1 0.006 1.3E–09 0.050 4.5E–11 0.050300 0.033 1.06 3 0.1 0.006 1.3E–09 0.057 6.3E–11 0.042300 0.100 2.21 0.1 0.01 0.007 4.5E–09 0.009 1.4E–10 0.001300 0.100 1.27 1 0.1 0.007 1.3E–09 0.072 1.3E–10 0.028300 0.200 1.43 0.5 0.1 0.007 1.4E–09 0.079 2.1E–10 0.021300 0.200 0.53 5 1 0.006 4.5E–10 0.525 1.7E–10 0.475300 0.333 0.67 3 1 0.006 4.8E–10 0.600 2.7E–10 0.400300 1.000 4.12 0.1 0.1 0.007 3.6E–07 0.091 1.2E–07 0.009300 1.000 4.47 1 1 0.007 1.8E–06 0.743 2.1E–06 0.257300 2.000 9.17 0.5 1 0.007 8.9E–02 0.670 9.7E–02 0.123300 10.000 9.45 0.1 1 0.007 1.6E–01 0.615 1.7E–01 0.024400 0.002 1.84 5 0.01 0.007 4.9E–09 0.003 1.6E–10 0.006400 0.003 1.92 3 0.01 0.007 4.4E–09 0.003 2.2E–10 0.006400 0.010 2.10 1 0.01 0.008 3.6E–09 0.004 4.4E–10 0.005400 0.020 2.22 0.5 0.01 0.008 3.2E–09 0.005 6.8E–10 0.004400 0.020 1.34 5 0.1 0.007 1.5E–09 0.031 5.5E–10 0.068400 0.033 1.42 3 0.1 0.007 1.3E–09 0.037 7.8E–10 0.062400 0.100 2.52 0.1 0.01 0.008 2.8E–09 0.007 1.8E–09 0.002400 0.100 1.62 1 0.1 0.008 1.1E–09 0.049 1.7E–09 0.050400 0.200 1.78 0.5 0.1 0.008 1.0E–09 0.057 2.7E–09 0.042400 0.200 0.92 5 1 0.007 4.5E–10 0.346 2.2E–09 0.653400 0.333 1.04 3 1 0.007 4.5E–10 0.410 3.6E–09 0.590400 1.000 3.30 0.1 0.1 0.008 1.5E–08 0.075 1.2E–07 0.024400 1.000 3.63 1 1 0.008 8.8E–08 0.542 1.9E–06 0.457400 2.000 8.99 0.5 1 0.008 1.8E–02 0.428 3.3E–01 0.216400 10.000 9.33 0.1 1 0.008 3.4E–02 0.338 5.8E–01 0.041500 0.020 1.88 5 0.1 0.007 1.4E–09 0.015 1.3E–08 0.072500 0.033 1.99 3 0.1 0.008 1.1E–09 0.017 1.8E–08 0.069500 0.100 2.27 1 0.1 0.009 6.7E–10 0.021 4.3E–08 0.064500 0.200 2.47 0.5 0.1 0.009 5.1E–10 0.024 7.8E–08 0.061500 0.200 1.60 5 1 0.007 3.0E–10 0.205 8.6E–08 0.781500 0.333 1.81 3 1 0.008 2.9E–10 0.227 1.6E–07 0.759500 1.000 3.51 0.1 0.1 0.009 2.1E–09 0.033 1.2E–06 0.051500 1.000 3.80 1 1 0.009 1.6E–08 0.251 1.8E–05 0.734500 2.000 8.45 0.5 1 0.009 4.8E–04 0.150 4.7E–01 0.362500 10.000 9.11 0.1 1 0.009 1.2E–03 0.060 8.6E–01 0.067600 0.200 3.58 5 1 0.007 1.5E–10 0.039 4.5E–05 0.852600 0.333 3.87 3 1 0.008 2.9E–10 0.039 8.9E–05 0.845600 1.000 5.09 1 1 0.008 4.9E–09 0.038 1.5E–03 0.835600 2.000 7.68 0.5 1 0.008 1.3E–06 0.024 3.7E–01 0.477600 10.000 8.47 0.1 1 0.008 3.8E–06 0.008 7.7E–01 0.094

336

HClþOH� ¼ H2Oþ Cl� ð13Þ

This total of 12 fluid species plus OH– and H2O wasconsidered a reasonable proxy for the late-magmatic,hydrothermal fluids. The equilibrium constants of theabove 12 equations at constant pressure and tempera-ture in the range from 300 to 600 �C at a lithostaticpressure of 1 kbar were calculated using the programSUPCRT92 (Johnson et al. 1992) with the revisedthermodynamic database of Oelkers et al. (1995). Theequilibrium constant of reaction (2) was adjusted tomeasured clinopyroxene compositions, whereas anal-cime, albite and sodalite were treated as pure endmembers due to their close-to-end member composition.Jadeite activities calculated with the model of Holland(1990) were in the range from 0.02 to 0.25. Whennepheline was involved in the calculations, its activitywas calculated using an ideal solution model and aver-age compositions in the respective textures such as the

one shown in Fig. 2. Resulting nepheline activities variedbetween 0.4 and 0.8.

Fluid composition and pH in equilibriumwith the hydrothermal assemblage

pH of the solution

Selected results of the above calculations in the range300–600 �C at 1 kbar are presented in Tables 3 4, and onFigs. 3, 4 and 5. It is obvious that the results do notdepend strongly on the presence of either nepheline orjadeite of various activities (Fig. 3). Calculations withanalcime (not shown) are almost indistinguishable fromthose using jadeite as solid phase. In other words, thefluid has basically the same composition in equilibriumwith albite plus jadeite or nepheline or analcime. Also,

Table 4 Speciation of ahydrothermal fluid coexistingwith albite and a clinopyroxenewith 5 mol% jadeite compo-nent; for details of the calcula-tions see the text. Pressure ofcalculations: 1 kbar

T mNaH5SiO5 mNaAl(OH)4 mAl(OH)2+ mAl(OH)4- mHCl mCl– mOH– mH+ mH2O

300 1.2E–11 2.3E–05 1.2E–11 0.005 0.06 4.94 9.5E–09 0.035 0.84794300 1.9E–11 2.4E–05 1.2E–11 0.004 0.04 2.96 9.2E–09 0.029 0.90283300 4.2E–11 2.6E–05 9.6E–01 0.002 0.01 0.99 8.9E–09 0.018 0.96529300 6.6E–11 2.6E–05 4.7E–01 0.001 0.01 0.49 8.9E–09 0.014 0.98228300 4.0E–11 2.3E–05 1.2E–11 0.000 0.17 4.77 3.0E–09 0.108 0.85012300 6.3E–11 2.4E–05 1.2E–11 0.000 0.11 2.85 3.0E–09 0.088 0.90452300 1.7E–10 2.7E–05 8.4E–02 0.001 0.00 0.10 1.0E–08 0.006 0.9963300 1.5E–10 2.6E–05 8.1E–01 0.000 0.03 0.94 3.0E–09 0.054 0.96615300 2.4E–10 2.6E–05 3.5E–01 0.000 0.02 0.46 3.2E–09 0.037 0.98282300 1.5E–10 2.3E–05 1.2E–11 0.000 0.43 4.09 1.0E–09 0.293 0.8591300 2.7E–10 2.4E–05 1.2E–11 0.000 0.23 2.36 1.1E–09 0.216 0.91172300 1.4E–07 2.7E–05 1.3E–06 0.000 0.00 0.09 8.2E–07 0.000 0.99644300 2.4E–06 2.6E–05 3.1E–08 0.000 0.00 0.74 4.2E–06 0.000 0.96944300 1.1E–01 2.6E–05 1.4E–14 0.000 0.00 0.38 2.0E–01 0.000 0.97068300 1.9E–01 2.6E–05 4.1E–14 0.000 0.00 0.08 3.8E–01 0.000 0.97162400 8.1E–12 8.2E–04 1.2E–11 0.101 0.14 4.84 3.5E–08 0.015 0.84659400 1.3E–11 8.8E–04 1.2E–11 0.070 0.10 2.89 3.2E–08 0.012 0.90176400 3.0E–11 9.4E–04 9.8E–01 0.031 0.04 0.96 2.6E–08 0.008 0.96476400 4.7E–11 9.6E–04 4.8E–01 0.018 0.02 0.47 2.3E–08 0.006 0.98195400 2.9E–11 8.3E–04 1.2E–11 0.009 0.44 4.48 1.1E–08 0.046 0.85368400 4.6E–11 8.8E–04 1.2E–11 0.006 0.29 2.64 9.6E–09 0.038 0.90732400 1.3E–10 9.7E–04 8.9E–02 0.006 0.00 0.09 2.0E–08 0.003 0.99616400 1.1E–10 9.4E–04 7.7E–01 0.003 0.11 0.84 8.1E–09 0.024 0.96766400 1.9E–10 9.6E–04 3.3E–01 0.001 0.06 0.40 7.6E–09 0.017 0.9838400 1.2E–10 8.3E–04 1.2E–11 0.001 0.00 3.31 3.3E–09 0.122 0.86957400 2.2E–10 8.8E–04 1.2E–11 0.000 0.58 1.82 3.2E–09 0.090 0.91985400 8.6E–09 9.7E–04 2.2E–04 0.000 0.00 0.08 1.1E–07 0.001 0.99668400 1.3E–07 9.4E–04 6.7E–06 0.000 0.00 0.54 6.4E–07 0.000 0.97281400 2.3E–02 9.4E–04 1.6E–12 0.000 0.00 0.28 1.3E–01 0.000 0.97478400 4.0E–02 9.4E–04 4.3E–13 0.000 0.00 0.06 2.4E–01 0.000 0.97632500 2.2E–11 1.3E–02 1.2E–11 0.073 0.00 2.85 2.9E–08 0.013 0.87373500 3.6E–11 1.4E–02 1.2E–11 0.044 0.00 1.54 2.4E–08 0.010 0.92262500 9.3E–11 1.5E–02 3.6E–01 0.012 0.56 0.38 1.4E–08 0.005 0.97499500 1.7E–10 1.5E–02 1.3E–01 0.005 0.29 0.15 1.1E–08 0.003 0.98786500 1.6E–10 1.3E–02 7.9E–01 0.002 0.00 1.02 6.4E–09 0.025 0.90183500 3.2E–10 1.4E–02 2.7E–01 0.001 0.00 0.51 6.1E–09 0.015 0.94012500 2.8E–09 1.6E–02 7.4E–04 0.001 0.02 0.03 4.5E–08 0.000 0.99712500 3.8E–08 1.5E–02 2.6E–05 0.001 0.02 0.25 3.4E–07 0.000 0.97754500 1.0E–03 1.5E–02 2.1E–14 0.001 0.00 0.14 1.0E–02 0.000 0.97956500 1.9E–03 1.5E–02 2.6E–11 0.001 0.00 0.03 2.5E–02 0.000 0.98111600 1.3E–09 1.1E–01 2.6E–03 0.001 0.00 0.04 1.1E–08 0.000 0.91497600 2.8E–09 1.2E–01 6.9E–04 0.001 0.00 0.04 2.2E–08 0.000 0.94606600 5.1E–08 1.2E–01 2.5E–06 0.001 0.13 0.04 3.6E–07 0.000 0.97929600 1.3E–05 1.3E–01 2.6E–11 0.001 0.00 0.02 9.5E–05 0.000 0.98174600 2.6E–05 1.3E–01 2.0E–12 0.001 0.00 0.01 2.8E–04 0.000 0.98202

337

the three-phase assemblage albite + nepheline +sodalite does not alter the results significantly (Fig. 4),although the activity of NaCl is fixed by this assemblageat fixed P and T.

However, the calculations clearly indicate that twoparameters determine the speciation and the pH of thehydrothermal solutions: the ratio of Na/Cl and thetemperature (Figs. 3, 4 and 5). It is important to stressagain, that all these calculations are charge balanced.Especially the Na/Cl ratio is the most importantparameter controlling the pH of the solution (Fig. 3). At350 �C and at Na/Cl ratios less than 1, fluids have astrongly acid pH between 0.5 and 3, whereas at Na/Clgreater than 1, pH is on the order of 8 to 10 and hence,strongly basic. At 350 �C and 1 kbar, neutrality liesapproximately at pH=5.3. In the region aroundNa=Cl, pH is extremely sensitive to very small changesin the Na/Cl ratio: addition of a small amount of Na toan NaCl solution in equilibrium with two Na-Al-sili-

cates produces a strongly basic pH, whereas Cl additionrenders the pH acidic. Temperature has a smaller effecton pH than the Na/Cl ratio (Fig. 5). Decreasing tem-perature from 600 to 300 �C at constant bulk composi-tion (i.e. both constant Natot and constant Na/Cl ratio)changes the pH less than 1 log unit. If the pH is basic athigh temperature, it becomes more basic during cooling,but if it is acidic at high temperature, it becomes moreacidic. This means that cooling a hydrothermal solutionin equilibrium with two Na-Al-silicates does not result ina fundamental change of pH if no reaction with a solidphase takes place which changes the Na/Cl ratio of thesolution.

Speciation of dissolved ions in the solution

Speciation calculations reveal that H4SiO4 is always themost important Si-bearing and Cl– the most importantCl-species, whereas the most important Al-bearing andNa-bearing species change depending on pH (Tables 3and 4, Fig. 6). At low pH, Al(OH)2

+, NaCl and Cl– arethe most important species, at intermediate pH, Na+

may become important, and at basic pH, NaOH is thedominant Na-bearing and Al(OH)4

– the dominant Al-bearing species. HCl is present in some extremely acidsolutions and may be important at low Natot concen-trations, whereas species such as HSiO3

–, NaH5SiO5 orNaAl(OH)4 have exceedingly low concentrations in allof our calculations. Figure 7 illustrates, how changingNa/Cl, i.e. changing pH, changes the molality of NaCl inthe hydrothermal fluid at various total concentrations ofNa.

The temperature effect on fluid speciation at constantNa/Cl and constant Natot is shown in Fig. 8. Decreasingtemperature from 600 to 300 �C decreases – depending

Fig. 3 The pH of hydrothermal fluids in equilibrium with theassemblage a albite + clinopyroxene (of variable jadeite contentbetween 5 and 25 mol%) and b albite + nepheline (variable Neendmember content between 40 and 80 mol%) at 350 �C, 1 kbarversus Na/Cl ratios for three different Na-concentrations

Fig. 4 The pH of hydrothermal fluids in equilibrium with theassemblage albite + nepheline (Ne80) + sodalite at 1 kbar andvarious temperatures for various total molalities of Na. Theassemblage fixes the NaCl activity at a value of 0.0036

338

on Na/Cl (equivalent to pH) – the NaCl, NaOH andHCl molalities drastically, the NaCl molality slightly,but increases the Na+ and Cl– molality. Hence, single-ion species are preferred at low temperatures (Fig. 8).

Quantitative model of the fluid–rock interaction process

Which process does the Na/Cl ratio of the hydrothermalfluid change?

The above calculations have shown that a fluid inequilibrium with two Na-Al-silicates cannot change itspH drastically just by decreasing temperature, but thatits composition and, specifically, its Na/Cl ratio has tochange in order to change its pH. Because we observethe NaOH-bearing mineral ussingite in some of theveins, a basic pH of the hydrothermal solutions from

which this mineral formed may be a reasonable as-sumption. Because ussingite is typically one of the latestminerals and because it does not appear in all of thehydrothermal veins, we take this as an indication ofincreasing pH during formation of the vein assemblage.In order to increase the pH, however, Na/Cl has to beincreased (see Fig. 3) and, thus, either Cl has to beremoved or Na has to be added to the fluid. Removal ofCl from a hydrothermal solution only happens when aCl-rich vapour exsolves or when a Cl-bearing mineralprecipitates. Removal of Cl alone by precipitation of aCl-rich mineral without contemporaneous removal ofNa is unlikely because the Cl-bearing minerals in ourassemblages, such as sodalite or tugtupite, would re-move much more Na than Cl in the same instance. If novapour-phase separation occurs, addition of Na bydissolution of Na-bearing and Cl-poor or Cl-free min-erals thus appears to be the only way to change the pHof the fluid towards a more basic pH. The differenttextures observed along grain boundaries, in andaround veins and hydrothermal lenses indicate thatboth processes could have worked in the Ilimaussaqintrusive complex.

Liquid–vapour separation

At low pressures, a miscibility gap opens in the systemNaCl–H2O, and a highly saline brine coexists with alow-density vapour (‘boiling’; e.g. Shmulovich et al.1995). If this phase separation occurs, the hydrolysis ofNaCl

NaClþH2O ¼ NaOHþHCl ð14Þ

leads to enrichment of HCl in the low-density vapourphase and of NaOH in the coexisting liquid (Bischoffand Pitzer 1989; Shmulovich et al. 1995). This meansthat a highly Cl-bearing, acidic vapour separates from aresidual, Na-enriched and accordingly basic liquid. Thelow-density vapour phase typically escapes easier fromthe system than the saline brine. The difference inquench pH between the two phases may reach up to 6log units at 600 �C and about 2 log units at about 400 �C(Shmulovich et al. 1995). The pressures at whichimmiscibility occurs lie between 200 and 500 bar in thepure NaCl–H2O system (Bischoff and Pitzer 1989;Shmulovich et al. 1995). These pressures appear too highfor the Ilimaussaq intrusion for which a pressure of in-trusion of about 1 kbar has been estimated (Konnerup-Madsen and Rose-Hansen 1982; Markl et al. 2001).However, at these high crustal levels, pressure maychange from hydrostatic to lithostatic by hydrofractur-ing of the country rocks, and at least for short periods oftime, this may lead to a significant pressure decreasedown to about 400 bar (hydrostatic), which may drive aliquid into the immiscibility field. Hence, the fluid insuch newly opened fractures may unmix and developbasic fluids to react with the country rocks and to pre-cipitate minerals stable at moderately basic pH. We

Fig. 5 The pH of hydrothermal fluids in equilibrium with albite +clinopyroxene (Jd0.05) at 1 kbar and at various temperatures andvarious Na/Cl ratios

339

suggest that the straight veins with no large surroundingreaction zones in Ilimaussaq may have formed this way.These veins, for example, bear the Na–Be silicate tug-tupite (Engell et al. 1971; Markl 2001b). However, ac-cording to the experimental data of Bischoff and Pitzer(1989) and Shmulovich et al. (1995), pressures wouldhave to be below 400 bar to allow for phase separationat 300–400 �C and the resulting pH of the NaOH-en-riched liquid would only be between 7 and 8 (quenchpH). Therefore, additional fluid–rock interaction had tooccur to drive either the NaOH-pre-enriched liquidtowards higher pH and towards ussingite stability(although we cannot quantify at which pH ussingitebecomes stable, unfortunately), or to basify fluids thatdid not unmix at all due to the slightly too high pressurein the Ilimaussaq late-magmatic system. The latterprocess probably lead to the veins with irregular outlinesand large replacement zones around them, and to theussingite-bearing lenses, as well as to the microtexturessuch as the one of Fig. 2. Hence, different types of veinsprobably formed from different physico-chemicalprocesses.

Fig. 6 Speciation of hydrothermal fluids in equilibrium with albite+ jadeite0.05 at 400 �C and 1 kbar depending on pH and on totalmolality of Na

Fig. 7 The molality of NaCl in hydrothermal fluids at 400 �C and1 kbar, which have variable Na/Cl ratios, variable total Namolalities and which are in equilibrium with albite + clinopyrox-ene (Jd0.05). As shown on the two lowermost curves, the assemblagealbite + nepheline (Ne0.4–0.8) does not alter the results significantly

340

Quantitative model for fluid–rock interaction

Some of the Ilimaussaq late-stage veins replace formerlycrystallized Ilimaussaq rocks around fractures and lensesinterpreted as late-magmatic fluid ‘pods’ (the ussingite-bearing ones definitely belong to this category). Texturesin and around the ‘replacement veins’ provide evidencefor replacement of nepheline (e.g. Fig. 2) amongstother minerals such as K-feldspar or arfvedsonite. Thenepheline-consuming reaction was taken as a proxy forhost rock replacement because nepheline is in manyIlimaussaq rocks a modally important constituent. It canbe nicely modelled in the Na–Al–Si–O–H–Cl system.

The easiest reaction to replace nepheline by, e.g.analcime (or albite, or jadeite), is the following:

Neþ SiO2 þH2O ¼ Anl ð15Þ

Here, 1 mol of nepheline is dissolved and 1 mol ofanalcime is formed, but Na is neither set free nor con-sumed and, hence, there is no change in the pH of thefluid.

However, thin section observations, like the one inFig. 2 and field observations, suggest that the replace-

ment is rather volume- than mass-conserving and, hence,the replacement reaction should be formulated asfollows:

1:85Neþ 2:3H2Oþ 0:19H4SiO4 ¼ 1:02Anlþ 0:83Naþ

þ 0:83AlðOHÞ�4 ð16Þ

In this reaction, 100 cm2 of nepheline was dissolvedand replaced by 100 cm2 of analcime. Here, almost halfof the nepheline’s Na is set free and, accordingly, bothNa/Cl and pH will increase. Replacement of nephelineby albite would have almost the same effect due to thevery similar molar volumes of albite and analcime,replacement by zeolites would have even greater effectsbecause they have very large molar volumes. However,replacement by jadeite (if there were pure jadeite at all)and sodalite would have much smaller effects andreplacement of primary sodalite by secondary analcimewould have no effect on the pH at all, again due to themolar volumes of the minerals taking part in thereplacement reactions.

Apart from pH, salinity also changes during thesereplacement reactions. Replacement of an anhydrousmineral by analcime or zeolites increases the salinity due

Fig. 8 Speciation of hydro-thermal fluids in equilibriumwith albite + clinopyroxene(Jd0.05) at various temperaturesand Na/Cl ratios and at 1 mol/ltotal Na molality

341

to consumption of the solute water, while replacement ofnepheline by albite leaves salinity virtually constant, andreplacement of nepheline by sodalite decreases thesalinity because NaCl is incorporated into sodalite.Figure 9 shows quantitative calculations of how the pHand salinity change when fluid reacts with a nepheline-or sodalite-bearing rock at hydrothermal conditions.The extent of pH and salinity changes depends on theratio between the volume of fluid entering the rock andthe volume of mineral being replaced.

As a quantitative – although admittedly crude andsimplified – example, we show how salinity and pH ofa fluid changes during the reaction depicted in Fig. 2.The details of this calculation are reported in Table 5.For these calculations, the molar volume data from thesame updated SUPCRT database of Oelkers et al.(1995) as above have been used. It was assumed thatthe pure analcime zone replaced the primary sodalite,whereas the symplectite of secondary sodalite and

analcime formed at the expense of nepheline. Theformation of the symplectite, instead of pure analcime,at the expense of nepheline was probably triggered bya local increase in fluid salinity, when primary Cl-bearing sodalite 1 was replaced by Cl-free analcime 1.The measured volumetric proportions of analcime 1 toanalcime 2 + sodalite 2-symplectite (see Fig. 2 forexplanation) and of analcime 2 to sodalite 2 within thissymplectite were both about 1:1. The calculations havebeen performed for a fluid volume of 100 cm2, and thevolumetric mineral/fluid ratio was varied between0 and 1, where ‘mineral’ means the volume of themineral(s) replaced. The starting fluid was chosen as anaqueous fluid with about 3 wt% NaCl and a Na/Clratio by weight of about 0.8. These figures were chosenbecause fluid inclusions in the most evolved lujavritesof the Ilimaussaq complex are aqueous with 3–5 wt%salinity (Markl et al. 2001) and because it is likely thatthe fluid starts with a slight Cl dominance due toprecipitation of Na-bearing, Cl-free minerals before.With increasing reaction, the pH changes to more basicvalues of up to 9.5 (Fig. 10), because Na/Cl changessystematically from 0.79 to 3.97 when the mineral/fluid

Fig. 9 The effect of various late-magmatic, hydrothermal reactionson the salinity and the pH of a coexisting fluid at 400 �C and 1 kbarin dependence on the volumetric mineral fluid ratio

342

ratio changes from 0 to 1. Salinity increases to valuesof about 17 wt% at a mineral/fluid ratio of 1. The Clmolality increased from 0.066 to 0.34 mol/100 g H2O.Hence, the observed replacement (Fig. 2) of magmaticsodalite and nepheline by an assemblage of analcimeand secondary sodalite results in a drastic increase inboth the salinity and the pH of the fluid. Even atmoderate mineral/fluid ratios such as 0.2, the pH isalready as basic as 8.5. At mineral/fluid ratios below0.2 (which means that five times more fluid entered therock than minerals were transformed) the effect ofthese reactions is still measurable, although progres-sively smaller.

Similar calculations for reactions in which nephelineis replaced by analcime or by albite alone show similarresults. Both salinity and Na/Cl increase dramatically atmoderate fluid/rock ratios despite the fact that duringthis replacement process, new Na-Al-silicates arecontinuously precipitated.

Summary and conclusions

Hydrothermal veins and hydrothermal replacementtextures of microscopic scale in peralkaline rocks of theIlimaussaq complex are believed to be related to theaction of late-magmatic fluids of the Ilimaussaq complexitself. These late-stage fluids were released at tempera-tures between 300 and 500 �C from the last crystallizingperalkaline liquids, the lujavrites, and they penetratedthe whole complex along large fractures – which wereprobably formed by hydrofracturing under litho- tohydrostatic pressure conditions – and along grainboundaries. Some of the fluids may have experiencedliquid–vapour separation, which may, for example, beresponsible for the highly variable salinity of fluid

inclusions in the hydrothermal assemblages (e.g. Sobolevet al. 1970; Markl 2001b). This process may have pro-duced moderately basic brines in these fractures. Thelate-stage fluids – either unmixed or not – reacted withthe earlier crystallized rocks, and especially with thenepheline which is not stable in equilibrium with thesefluids. Various types of precipitation and replacementreactions produced albite + analcime + aegirine ±sodalite ± ussingite ± tugtupite assemblages mainly inveins and lenses and to a lesser extent along grainboundaries (here no ussingite or tugtupite). Thereplacement reactions were volume-, but not mass-con-serving and they strongly changed the salinity and thepH of the fluids, in some cases from a low-salinity,aqueous fluid of low pH to a highly saline brine withstrongly basic pH.

Various types of hydrothermal veins at Ilimaussaq

It was shown above that volumetric 1:1 replacement ofnepheline- and sodalite-bearing rocks by albite- andanalcime-dominated assemblages results in strongincrease of both pH and salinity of the coexistinghydrothermal fluid. Although the model is certainlyoversimplified, as it does not take into account othercomponents and phases in the rock as it deals with avery simplistic fluid composition, it nevertheless pro-vides evidence of how hydrothermal fluid compositionmay change during liquid–vapour exsolution or duringlate-magmatic autometasomatic fluid–rock interactionprocesses. This model explains why primary fluidcompositions in the Ilimaussaq lujavrites typicallyindicate very water-rich fluids of low salinities and whyin the late-stage hydrothermal veins a large variation influid salinity has been reported (e.g. Sobolev et al.

Table 5 Fluid speciation and mass calculations during formation of the reaction texture in Fig. 2. Calculations are for 100 cm2 of fluidwith a starting salinity of 4 wt% (3 wt% NaCl and 1 wt% KCl)

Ne Sod Anl Na+ Cl– H2O H4SiO2 Al(OH)4– Mol H2O

finalMol Clfinal

Mol Nafinal

FinalNa/Cl ratio

Final salinityin wt%

1. Partial reaction: NEfiSod2+Anl2 (when volume of Sod2=volume of Anl2)Reactionstoichiometry

–1.85 0.12 0.515 0.38 –0.24 –1.95 0.10 0.62

Mineral/fluid ratio0.01 0.00 0.00 –0.02 0.00 0.01 3.98 0.05 0.04 0.91 3.840.05 0.02 –0.01 –0.10 0.01 0.03 3.90 0.04 0.06 1.56 3.180.2 0.08 –0.05 –0.41 0.02 0.13 3.59 0.00 0.12 –79.87 0.121 0.38 –0.24 –1.95 0.10 0.62 2.06 –0.19 0.41 –2.17 –41.892. Partial reaction: Sod1fiAnl1

0.00 –0.24 1.03

0.89 0.48 –0.61 –0.62 0.410.01 0.01 0.00 –0.01 –0.01 0.00 3.99 0.05 0.05 0.89 4.360.05 0.04 0.02 –0.03 –0.03 0.02 3.97 0.07 0.08 1.14 5.790.2 0.19 0.10 –0.13 –0.13 0.09 3.87 0.15 0.23 1.51 11.261 0.89 0.48 –0.61 –0.62 0.41 3.39 0.53 0.93 1.76 33.57Complete reaction (when volumes of Anl1:Anl2:Sod2=2:1:1) 8.00 0.10 0.08 0.80 4.000.01 7.97 0.10 0.09 0.90 4.000.05 7.87 0.11 0.14 1.28 4.410.2 7.46 0.15 0.34 2.32 6.121 5.45 0.34 1.34 3.97 16.78

343

1970; Konnerup-Madsen 1982). Depending on thespecific process, the fluid/mineral ratio and on thereactions going on around a specific vein, fluid salini-ties and the pH of the fluids may vary significantly (e.g.Figs. 9 and 10). In the replacement process, the verysame fluid may evolve towards high salinity and high

pH (mainly nepheline is replaced by analcime oralbite), high salinity and low pH (mainly sodalite isreplaced by analcime) or low salinity and high pH(mainly nepheline is replaced by secondary sodalite orby albite) depending on the rock type the fluid isreacting with and the specific assemblage that precipi-tates. Fluids developing high pH will ultimately pre-cipitate ussingite and those with high pH and highsalinity will precipitate minerals such as tugtupite,which are stabilized by high Cl content and high pH ofthe fluid (see Markl 2001b). The various physico-chemical processes and the various possibilities ofevolution probably explain why there are differenttypes of late-magmatic hydrothermal veins atIlimaussaq (e.g. Engell et al. 1971), although it iscertainly reasonable to assume only one late-magmatic‘parental’ fluid, which is the fluid expelled from thecrystallizing lujavrite.

Implications for hyper-agpaites

Both the unmixing and the autometasomatic processmay also work on a larger scale and produce the largeroccurrences of hyper-agpaites, which are known fromthe Kola peninsula (Khomyakov 1995, 2001; Pekov2000). These rocks would accordingly represent largeportions of alkali syenitic rocks replaced by late-mag-matic minerals due to interaction with fluids becoming(when reacting) or already being (when unmixed) basic.The wealth of water-soluble minerals in these assem-blages is not surprising because the water is successivelyremoved by hydration reactions and the dissolved min-erals will successively precipitate while the fluid ‘driesout’.

The key why this process only appears to work inagpaitic rocks may be the specific relatively simple fluidcomposition (based on the fluid inclusion work pub-lished) expelled from the last crystallizing melt. It cantypically be described in the system Na2O–Al2O3–SiO2–H2O–NaCl with or without K2O. This fluid is likely ofnear neutral pH because strongly acidic solutionswould reach alumina and silica saturation upon onlylittle dissolution of the rock. Saturation with silicawould buffer the fluid at low pH values anyway(Hauzenberger et al. 2001; Baumgartner, unpublisheddata). As the fluid progresses towards a basic pH,aluminium and silica solubility increase. In fact, An-derson and Burnham (1983) found that feldspar dis-solved near congruently in basic solutions where NaOHor KOH was added. Hence, acidic solutions wouldremain buffered at low pH, whereas near neutralsolutions can, by dissolution of nepheline, evolvetowards basic solutions.

Another important feature of the proposed process isthat replacement must involve minerals such as nephe-line or sodalite in order to change the fluid compositionsignificantly because the decrease in Na per formula unitmust be large. This is not the case for replacement of

Fig. 10 The effect of the reaction depicted in Fig. 2 on the totalmolality of Cl, the Na/Cl ratio (=the pH) and on the salinity of acoexisting fluid depending on the volumetric mineral/fluid ratio at1 kbar and 400 �C. See text for detailed discussion

344

feldspar, for example, because feldspar itself has a rel-atively large molar volume. Hence, the replacement typeof autometasomatic process will only lead to the de-scribed features in peralkaline, silica-undersaturatedmagmatic complexes.

Acknowledgements Nathalie Kranz helped with the microprobeanalyses. Important improvements of this paper were initiated bythe critical and constructive remarks of Wilhelm Heinrich and ananonymous reviewer. We express our gratefulness for their re-marks, especially for Wilhelm Heinrich’s suggestions about liquid–vapour phase separation and its possible importance at Ilimaussaq.This work was funded by the Deutsche Forschungsgemeinschaft bygrants Ma2135/1-1 and 1-2.

References

Anderson GM, Burnham CW (1983) Feldspar solubility and thetransport of aluminum under metamorphic conditions. AmJ Sci 283A:283–297

Armstrong JT (1991) Quantitative elemental analysis of individualmicroparticles with electron beam instruments. In: HeinrichKFJ, Newbury DE (eds) Electron probe quantification. PlenumPress, New York, pp 261–315

Bailey JC, Gwozdz R, Rose-Hansen J, Sørensen H (2001)Geochemical overview of the Ilimaussaq alkaline complex,South Greenland. Geol Survey Denmark Greenland Bull190:35–54

Bischoff JL, Pitzer KS (1989) Liquid–vapour relations for thesystem NaCl–H2O: summary of the P–T–x surface from 300 �Cto 500 �C. Am J Sci 289:217–248

Blundell DJ (1978) A gravity survey across the Gardar igneousprovince, SW Greenland. J Geol Soc Lond 135:645–654

Bureau H, Keppler H (1999) Complete miscibility between silicatemelts and hydrous fluids in the upper mantle; experimentalevidence and geochemical implications. Earth Planet Sci Lett165:187–196

Emeleus CH, Upton BGJ (1976) The Gardar Period in SouthGreenland. In: Escher A, Watt WS (eds) Geology of Greenland.Grønlands Geologiske Undersøgelse, pp 152–180

Engell J, Hansen J, Jensen M, Kunzendorf H, Løvborg L (1971)Beryllium mineralization in the Ilımaussaq intrusion, SouthGreenland. Rapport Grønlands Geologiske Undersøgelse33:1–40

Eugster H, Baumgartner L (1987) Mineral solubilities and specia-tion in supercritical metamorphic fluids. Mineral Soc Am RevMineral 17:367–404

Ferguson J (1964) Geology of the Ilimaussaq alkaline intrusion,south Greenland. Bull Grønlands Geologiske Undersøgelsevol 39

Forsberg G Rasmussen E (1978) Gravity and rock densities in theIlimaussaq area, South Greenland. Rapport GrønlandsGeologiske Undersøgelse 90:81–84

Johnson JW, Oelkers EH, Helgeson H (1992) SUPCRT92: a soft-ware package for calculating the standard molal thermody-namic properties of minerals, gases, aqueous species, andreactions from 1 to 5,000 bars and 0 to 1,000 �C. ComputGeosci 18:899–947

Hauzenberger CA, Baumgartner LP, Pak TM (2001) Experimentalstudy on the solubility of the ‘model’-pelite assemblage albite +K-feldspar + andalusite + quartz in supercritical chloride-richsupercritical aqueous solutions at 0.2 GPa and 600 �C.Geochim Cosmochim Acta 65:4493–4507

Holland TJB (1990) Activities of components in omphacitic solidsolutions. Contrib Mineral Petrol 105:446–453

Horvath L, Horvath-Pfenninger E (2000) Die Mineralien des MontSaint-Hilaire. LAPIS 25:23–61

Khomyakov AP (1995) Mineralogy of hyperagpaitic alkalinerocks. Clarendon Press, Oxford

Khomyakov AP (2001) The distribution of minerals in hyper-ag-paitic rocks in terms of symmetry: evolution of views on thenumber and symmetry of minerals. Geol Greenland Surv Bull190:73–82

Konnerup-Madsen J, Rose-Hansen J (1982) Volatiles associatedwith alkaline igneous rift activity: fluid inclusions in the Ili-maussaq intrusion and the Gardar granitic complexes (SouthGreenland). Chem Geol 37:79–93

Konnerup-Madsen J, Rose-Hansen J (1984) Composition andsignificance of fluid inclusions in the Ilimaussaq peralkalinegranite, South Greenland. Bull Mineral 107:317–326

Larsen LM (1976) Clinopyroxenes and coexisting mafic mineralsfrom the alkaline Ilimaussaq Intrusion, South Greenland.J Petrol 17:258–290

Larsen LM (1977) Aenigmatites from the Ilimaussaq intrusion,south Greenland: chemistry and petrological implications. Li-thos 10:257–270

Larsen LM (1981) Chemistry of feldspars in the Ilimaussaq augitesyenite with additional data on some other minerals. RapportGrønlands Geologiske Undersøgelse 103:31–37

Larsen LM, Sørensen H (1987) The Ilimaussaq Intrusion – pro-gressive crystallization and formation of layering in an agpaiticmagma. In: Fitton JG, Upton BGJ (eds) Alkaline igneousrocks. Geol Soc Spec Publ 30:473–488

Markl G (2001a) A new type of silicate liquid immiscibilitymacroscopically visible in Proterozoic peralkaline rocks fromIlimaussaq, South Greenland. Contrib Mineral Petrol141:458–472

Markl G (2001b) Stability of Na–Be minerals in late-magmaticfluids of the Ilımaussaq alkaline complex, South Greenland.Geol Surv Denmark Greenland Bull 190:145–158

Markl G, Marks M, Schwinn G, Sommer H (2001) Phase equi-librium constraints on intensive crystallization parameters ofthe Ilımaussaq complex, South Greenland. J Petrol42:2231–2258

Marks M, Markl G (2001) Fractionation and assimilation pro-cesses deduced from phase equilibria in the alkaline augite sy-enite unit of the Ilimaussaq Intrusion, South Greenland.J Petrol 42:1947–1969

Nielsen BL, Steenfelt A (1979) Intrusive events at Kvanefjeld in theIlimaussaq igneous complex. Bull Geol Soc Denmark 27:143–155

Oelkers EH, Helgeson HC, Shock EI, Sverjensky DA, Johnson JW,Porkrovskii VA (1995) Summary of apparent partial molalGibbs free energy of formation of aqueous species and gases atpressures 1 to 5,000 bars and temperatures 25 to 1,000 degreesC. J Chem Ref Data 24:1401–1560

Pekov IV (2000) Lovozero Massif. Ocean Pictures, MoscowPouchor JL, Pichoir F (1984) A new model for quantitative X-ray

microanalysis. Part1: applications to the analysis of homogenesamples. Rech Aerespatiale 3:13–38

Poulsen V (1964) The sandstones of the Precambrian EriksfjordFormation in South Greenland. Rapport Grønlands Geolog-iske Undersøgelse, vol 2

Roselle GT, Baumgartner LP (1995) Experimental determinationof anorthite solubility and calcium speciation in supercriticalchloride solutions at 2 kbar from 400 to 600 �C. GeochimCosmochim Acta 59:1539–1549

Schwinn G (1999) Die Petrologie der Sodalith-Foyaite, Naujaiteund Kakortokite von Ilimaussaq, Sud-Gronland. MSc Thesis,University of Freiburg

Shmulovich KI, Tkachenko SI, Plyasunova NV (1995) Phaseequilibria in fluid systems at high pressures and temperatures.In: Shmulovich KI, Yardley BWD, Gonchar G (eds) Fluid inthe crust. Chapman and Hall, London, pp 193–214

Sobolev VS, Bazarova TY, Shugurova NA, Bazarov LS, DolgovYA, Sørensen H (1970) A preliminary examination of fluidinclusions in nepheline, sørensenite, tugtupite and chkalovitefrom the Ilimaussaq alkaline intrusion, South Greenland. BullGeol Surv Greenland 81:1–32

Sørensen H (1997) The agpaitic rocks; an overview. Mineral Mag61:485–498

345

Sørensen H, Larsen LM (2001) The hyper-agpaitic stage in theevolution of the Ilimaussaq alkaline complex, South Greenland.Geol Greenland Surv Bull 190:83–94

Sood MK, Edgar AD (1970) Melting relations of undersaturatedalkaline rocks. Meddelser Grønland, vol 181

Sowerby JR, Keppler H (2001) The effect of fluorine, boron andexcess sodium on the critical curve in the albite–H2O system.Contrib Mineral Petrol 143:32–37

Upton BGJ, Emeleus CH (1987) Mid-Proterozoic alkaline mag-matism in southern Greenland: the Gardar province. In: FittonJG, Upton BGJ (eds) Alkaline igneous rocks. Geol Soc SpecPubl 30:449–471

Ussing NV (1912) Geology of the country around Julianehab,Greenland. Meddelser Grønland vol 38

346