Chemical Geology 288 (2011) 39ndash52

Contents lists available at ScienceDirect

Chemical Geology

j ourna l homepage wwwe lsev ie rcom locate chemgeo

Research papers

Cooling history of a dike as revealed by mineral chemistry A case study fromMt Etna volcano

Silvio Mollo a Gabriele Lanzafame b Matteo Masotta c Gianluca Iezzi dCarmelo Ferlito b Piergiorgio Scarlato a

a Istituto Nazionale di Geofisica e Vulcanologia Via di Vigna Murata 605 I-00143 Roma Italyb Universitagrave di Catania Dipartimento di Scienze Geologiche Corso Italia 57 I-95129 Catania Italyc Dipartimento di Scienze della Terra Sapienza Universitagrave di Roma Ple Aldo Moro 5 I-00176 Roma Italyd Dipartimento DIGAT Universitagrave G dAnnunzio Via Dei Vestini 30 I-66013 Chieti Italy

Corresponding author Tel +39 0651860674 faxE-mail address molloingvit (S Mollo)

0009-2541$ ndash see front matter copy 2011 Elsevier BV Aldoi101016jchemgeo201106016

a b s t r a c t

a r t i c l e i n f o

Article historyReceived 15 March 2011Received in revised form 16 June 2011Accepted 28 June 2011Available online 7 July 2011

Editor DB Dingwell

KeywordsDikePlagioclaseClinopyroxeneCooling ratePartition coefficientThermometer

Seven rock samples were systematically collected from innermost to the outermost portion of a dikeoutcropping at Mt Etna volcano Results show that from dike core-to-rim plagioclase clinopyroxene andtitanomagnetite show compositional variations due to increasing cooling rate Plagioclase is progressivelyenriched in An from innermost to the outermost part of the dike Similarly clinopyroxene components En+CaTs+CaFeTs increase whereas Di+Hd decrease The Usp content in titanomagnetite also systematicallydecrease from dike core-to-rim Partition coefficients and thermometers based on the crystal-liquid exchangereaction indicate that due to rapid cooling rates at the dike outer portions early-formed crystal nuclei do notre-equilibrate with the melt The chemistry of minerals progressively deviates from that of equilibriumconsequently from dike core-to-rim mineral compositions resemble those of high-temperature formationThe chemical variations of clinopyroxene and plagioclase in dike samples mirror those obtained from coolingexperiments carried out on alkaline basalts Accordingly we used an experimental equation based onclinopyroxene compositional variation as a function of cooling rate to determine the cooling conditionsexperienced by the crystals during dike emplacement The estimated cooling rates are comparable to thosepredicted by thermal modeling based on an explicit finite-difference scheme

+39 0651860507

l rights reserved

copy 2011 Elsevier BV All rights reserved

1 Introduction

Rock texture crystallization sequence and chemical compositionof minerals are strongly affected by kinetic effects especially at highcooling and decompression rates (eg Hammer 2008 Iezzi et al2008 Brugger andHammer 2010 Del Gaudio et al 2010Mollo et al2010 in press-a

Recently experimental studies have related the chemical variationof crystallizing phases to the cooling history of magmas (Hammer2006 Lofgren et al 2006 Mollo et al 2010 in press-a Iezzi et al2011) They showed that in addition to textural variations themineral chemistry also reflects different cooling regimes

Dikes especially those with meter-size thickness are very usefulto highlight the effect of cooling rate on mineral compositions Therate at which temperature decreases is determined by the distancefrom the host rock for lower distances the heat is released frommagma to the host rock more efficiently thus causing rapid coolingrates As a consequence crystal chemical variations induced by

cooling regimes in dikes can be used to reconstruct the solidificationpath of magmas (Coish and Taylor 1979 Ujike 1982 Pietranik et al2006 Chistyakova and Latypov 2009 Latypov et al 2011)

In this study we investigate the compositional variations ofclinopyroxene plagioclase and titanomagnetite from a dike outcrop-ping at Mt Etna volcano Our results indicate that these crystallinephases grew at increasing cooling rates from innermost to theoutermost portion of the dike The marked and systematic mineralchemical variations highlight that care must be taken to assesswhether equilibrium was attained Moreover kinetic effects have tobe taken into consideration when estimating dike crystallizationconditions by means of thermometers based on clinopyroxene- andplagioclase-liquid pairs (Putirka et al 1996 2003 Putirka 20052008 Lange et al 2009)

2 Sampling and analytical methods

For the purpose of this study we have selected a dike from MtEtna volcano (Sicily Italy) the largest sub-aerial volcano in Europeand one of themost active andmost intenselymonitored on Earth MtEtna is a 1200 km2 wide and 33 km high stratovolcano (Fig 1) Thegreater part of Mt Etna was constructed by the overlapping products

Fig 2 Compositions of products erupted at Mt Etna volcano In the Na2O+K2O versusSiO2 diagram the bulk rock analysis of the dike (closed star) and deposits of Trifogletto II(solid line) Vavalaci (dotted line) and Ellittico (dashed line) eruptive centers areplotted

40 S Mollo et al Chemical Geology 288 (2011) 39ndash52

associated to distinct centers of activity Lavas emitted in the last300 ka are of alkalic affinity and range from hawaiite to benmoreite incomposition (Fig 2) Dikes crop out over the entire Valle del Bove areaand they represent the uppermost segment of magmatic intrusionsthat feed many eruptive centers also associated to regional tectonics(Ferlito and Nicotra 2010 and references therein)

We investigated the texture and mineral chemistry of a dikelocated at an altitude of 1550 m asl at the head of a deep valleynamed Canalone dei Faggi from the southern border of the Valle delBove depression (Fig 1) This dike intrudes the hawaiitic tobenmoreitic lava flows and pyroclastic deposits of Trifogletto IIVavalaci and Ellittico eruptive centers (Fig 2) Stratigraphically themagmatic intrusion is subsequent to the activity of the Trifoglietto IIand Vavalaci centers and can be ascribed to the Ellittico magmaticactivity (Fig 2) We sampled the shallowest part of the dike that atthe time of the intrusion was at a depth of no more than 100 mbeneath the original surface This depthwas inferred by the altitude ofuneroded lava flows cropping out at the border of the Valle del BoveConsidering a rock density of 2750 kgm3 a depth of 100 mcorresponds to a lithostatic pressure of about 27 MPa The dike inthe sampling section has a total width of 43 m (Fig 1) and appears tobe a simple body showing sharp parallel contacts with the host rockneither field nor laboratory petrographic studies revealed in situcontamination by host rock or internal boundaries such as chilledzones or comb layers (eg Chistyakova and Latypov 2009 andreferences therein)

Seven samples hereafter named DK1 DK2 DK3 DK4 DK5 DK6and DK7 were collected from the innermost to the outermost portionof the dike using a sampling distance of 36 cm from sample to sample(Fig 1) The bulk analysis of the samples was performed by XRF onpowder pellets corrected for matrix effects (Franzini and Leoni1972) at the Dipartimento di Scienze Geologiche University ofCatania For major elements no significant variations were observedamong dike samples having an averaged composition of SiO2=4859(plusmn032) TiO2=186 (plusmn002) Al2O3=1688 (plusmn019) FeO=1129

Fig 1 Schematic map of Mt Etna volcano (insert panel) Seven samples ie DK1 DK2DK3 DK4 DK5 DK6 and DK7 were collected from innermost to the outermost part ofthe dike using a sampling distance of 36 cm from sample to sample

(plusmn016) MnO=020 (plusmn001) MgO=419 (plusmn010) CaO=1037(plusmn009) Na2O=380 (plusmn015) K2O=161 (plusmn01) P2O5=121(plusmn001) (all wt) The rock can be classified as trachybasalt(Fig 2) Moreover its chemical homogeneity allows us to excludean internal chemical zonation due to (i) changes in composition of aninflowing magma (ii) flow differentiation and (iii) accumulation orconcentration of crystals (eg Chistyakova and Latypov 2009) Thedike results from the intrusion of a single pulse of magma andconsequently textural and chemical variations of mineral phases areattributable to the cooling process operating during dike emplace-ment (Coish and Taylor 1979 Ujike 1982)

Textural and chemical analyses were carried out at the HP-HTLaboratory of Experimental Volcanology and Geophysics of theIstituto Nazionale di Geofisica e Vulcanologia in Roma (Italy) Sevendouble-polished thin sections (one for each sample) were investigat-ed using a Jeol-JXA8200 EDS-WDS combined electron microprobeequipped with five wavelength-dispersive spectrometers 15 kVaccelerating voltage and 10 nA beam current Core and rim of crystalswere analyzed with a beam size of 2 μm and counting time of 20 and10 s on peaks and background respectively The following standardshave been adopted for the various chemical elements jadeite (Si andNa) corundum (Al) forsterite (Mg) andradite (Fe) rutile (Ti)orthoclase (K) barite (Ba) apatite (P) and spessartine (Mn) Sodiumand potassium were analyzed first to reduce possible volatilizationeffects Precision was better than 5 for all cations Textural imageswere obtained with a Jeol FE-SEM 6500 F equipped with an energydispersion microanalysis system

3 Results

31 Textures

The dike samples are poorly vesicular (b5 vol) and microcrys-talline Their texture is holocrystalline characterized by fine-grainedgroundmass (Fig 3) Because of the low crystal size variable from 200to 2 μm a quantification of the textural variation from dike core-to-rim was not feasible This is also beyond the scope of this study

Significant changes in crystal content were not observed from dikecore-to-rim On average the phase assemblage consists of plagioclase(68 vol) clinopyroxene (20 vol) olivine (6 vol) and titanomag-netite (6 vol)

Plagioclase clinopyroxene and titanomagnetite are characterizedby variable textures from innermost to the outermost part of thedike (Fig 3) In the internal part crystals have tabular (ie plagioclase)and prismatic (ie clinopyroxene) shapes with well defined edges

Fig 3 Textural features of plagioclase (plg) clinopyroxene (cpx) titanomagnetite (timt) and olivine (ol) from innermost (a) to the outermost part of the dike (b) Black scale bar is100 μm White scale bar is 10 μm

41S Mollo et al Chemical Geology 288 (2011) 39ndash52

(Fig 3a) In contrast at the dike margin dendritic or acicular crystalsexhibit sieve textured and resorbed rims (Fig 3b)

Regardless for the position of dike samples olivine is alwayssubrounded in shape frequently showing resorbed and irregularboundaries (Fig 3a) Crystals have a maximum size of 30 μm and insome cases are poikilitic from dike core-to-rim (Fig 3a)

32 Mineral compositions

Chemical analyses of both plagioclase core and rim are reported inTable 1 for each dike sample Plagioclase is labradoritic in compositionwith a pronounced normal zoning Fig 4 shows that the chemistry ofboth plagioclase core and rim systematically changes as a function ofthe sample position Al Ca Fe and Mg increase and Si Na and Kdecrease from innermost to the outermost part of dike ie fromsample DK1 to DK7 Notably the core of plagioclase shows an overallenrichment in Al Ca Fe and Mg relative to rim In Fig 4 the crystalvariability is also reported in terms of anorthite (An) albite (Ab) andorthoclase (Or) components for both core and rim of plagioclase theAn content rapidly increases from dike core-to-rim in contrast Aband Or contents progressively decrease

Representative analyses of clinopyroxene are reported in Table 2Site occupancies molecular components and Fe3+Fe2+ ratios werecalculated on the basis of six oxygen atoms following Putirka (1999)Crystals are zoned with more magnesian cores (Mg=83plusmn1Mg=atomic Mg(Mg+Fe2+)) relative to rims (Mg=79plusmn1)Clinopyroxenes are diopsidic in compositions (Morimoto 1988)However each major cation incorporated in clinopyroxene variessystematically as a function of the sample position Fig 5 shows thatfrom innermost (sample DK1) to the outermost (sample DK7) part ofthe dike crystals are progressively depleted in Ca+Mg+Fe2++Si

counter balanced by enrichments in Na+Fe3++Al (mainly IVAl)+Ti(Table 2 and Fig 5) Changes of clinopyroxene components are alsoreported in Fig 5 from sample DK1 to DK7 diopside (Di) andhedenbergite (Hd) decrease whereas enstatite (En) Ca-Tschermak(CaTs) and CaFe-Tschermak (CaFeTs) increase The Tschermak-richcomponents result from the coupled substitution IVAl with Si and M2

(Mg Fe2+) with M1(Al Fe3+) notably the progressive CaFeTsmolecular increment compensates the decrement of IVAl with Fe3+

(Table 2)Titanomagnetite is the only opaque mineral present in the dike

samples The composition of magnetite (Mt IVFe3+ VI(Fe2+Fe3+)O4)-ulvospinel (Usp IVFe2+ VI(Fe2+Ti)O4) solid solution has beenrecalculated following Stormer (1983) on the basis of oxidestoichiometry we determined the Fe2O3 and FeO from total ironand cations per formula unit of minor costituents ie Al2O3 MnO andMgO (Table 3) Results are plotted in Fig 6 on the TiO2ndashFeOndash12Fe2O3

ternary diagram It is worth noting that the ulvospinel (Usp) contentof titanomagnetite decreases from Usp63 to Usp39 being the crystalsprogressively depleted in Ti from dike core-to-rim (Table 3)

Differently to other mineral phases olivine crystals do not showsignificant chemical variations as a function of the sample position(Table 4) However crystals are zoned with more magnesian cores(Fo66-67) relative to rims (Fo59-60)

4 Discussion

Plagioclase clinopyroxene olivine and titanomagnetite found indike samples are common minerals for the majority of Etneanhistorical and pre-historical products typical of low pressurecrystallization (Tanguy et al 1997 and references therein)

Table 1Representative electron microprobe analyses of plagioclases sd represents the standard deviation (in parenthesis number of averaged analyses)

DK1 DK2 DK3 DK4 DK5 DK6 DK7

wt Core sd(10)

Rim sd(10)

Core sd(10)

Rim sd(10)

Core sd(10)

Rim sd(10)

Core sd(10)

Rim sd(10)

Core sd(10)

Rim sd(10)

Core sd(10)

Rim sd(10)

Core sd(10)

Rim sd(10)

SiO2 5125 021 5321 022 5093 011 5297 011 5062 014 5274 015 5046 009 5262 009 5030 047 5250 049 4999 026 5227 027 4967 017 5203 018TiO2 005 000 008 000 006 000 009 000 007 000 010 000 008 000 011 000 009 000 012 000 010 000 013 000 011 000 014 000Al2O3 3017 023 2929 023 3031 013 2940 013 3045 009 2950 009 3052 009 2956 009 3059 038 2961 036 3073 034 2971 033 3087 021 2982 020FeO 080 001 068 000 089 001 076 000 098 001 084 001 102 001 088 001 106 001 091 001 115 001 099 001 124 001 107 001MnO 003 000 001 000 003 000 001 000 003 000 001 000 003 000 001 000 002 000 002 000 002 000 002 000 002 000 002 000MgO 015 000 009 000 015 000 010 000 016 000 011 000 016 000 011 000 016 000 011 000 017 001 012 000 017 000 013 000CaO 1393 016 1121 013 1415 010 1148 008 1438 014 1175 011 1449 010 1189 008 1460 022 1203 018 1483 022 1230 018 1505 015 1257 013Na2O 335 004 494 006 323 004 473 006 311 003 452 005 305 002 441 003 299 008 430 011 287 004 409 006 275 003 388 004K2O 023 000 037 000 020 000 033 000 016 000 030 000 015 000 028 000 013 000 026 001 009 000 023 000 006 000 019 000Total 9996 9988 9996 9987 9995 9987 9995 9986 9995 9986 9994 9985 9994 9985

Formula on the basis of 8 oxygensSi 2338 2416 2326 2406 2314 2397 2308 2393 2301 2388 2289 2379 2277 2369Ti 0001 0001 0001 0002 0001 0002 0001 0002 0001 0002 0002 0002 0002 0002Al 1623 1568 1632 1574 1641 1581 1645 1584 1650 1588 1659 1594 1668 1601Fe 0027 0023 0031 0026 0034 0029 0035 0030 0037 0031 0040 0034 0043 0037Mn 0001 0001 0001 0001 0001 0001 0001 0001 0001 0001 0001 0001 0001 0001Mg 0010 0006 0010 0007 0011 0007 0011 0007 0011 0008 0011 0008 0012 0009Ca 0681 0545 0693 0559 0704 0572 0710 0579 0716 0586 0728 0600 0739 0613Na 0296 0435 0286 0416 0276 0398 0270 0389 0265 0380 0255 0361 0244 0343K 0013 0021 0011 0019 0009 0017 0008 0016 0007 0015 0005 0013 0004 0011An 6873 5446 6996 5620 7118 5796 7180 5886 7241 5976 7365 6158 7488 6343Ab 2991 4343 2889 4188 2786 4030 2735 3950 2683 3870 2580 3708 2476 3543Or 135 212 115 193 095 174 086 164 076 154 056 134 036 114

42SM

olloet

alChem

icalGeology

288(2011)

39ndash52

Fig 4 Plagioclase composition from sample DK1 to DK7 Recalculation of feldspar analyses is done on an 8-oxygen basis Major element concentrations are reported in atoms performula unit (apfu) Anorthite (An) Albite (Ab) and Orthoclase (Or) components are expressed in mole percent Error bars are within symbols

43S Mollo et al Chemical Geology 288 (2011) 39ndash52

It is worth noting that Mt Etna is an open-system degassingvolcano generating magmas that contain relatively high watercontents (up to 35 wt) (cf Ferlito and Lanzafame 2010 andreferences therein) Therefore the crystallization of some Etneanmagmas can be driven by volatile exsolution due to decompression(Metrich and Rutherford 1998) However in our dike samples thelow amount of bubbles (b5 vol) and the absence of hydrousminerals (ie amphibole and phlogopite) characteristic of water-bearing Etnean magmas (cf Tanguy et al 1997) indicate that acrystallization induced by decompression minimally affected thesolidification of magma and that according to previous studiesvolatile phases migrate passively through cracks and vugs ofoverlapping lava flows (Heap et al 2009 2011)

41 Plagioclase clinopyroxene and titanomagnetite compositionalvariations

Compositions of plagioclase clinopyroxene and titanomagnetitefrom our samples resemble those observed for crystals found in manyEtnean lavas However minerals show marked and continuouscompositional variations from sample DK1 to DK7 (Figs 4 5 and 6)Such variations testify to an increasing disequilibrium growth of thecrystals from the innermost and slowly cooled part of the dike to theoutermost and rapidly cooled portion (Smith and Lindsley 1971Mevel and Velde 1976 Coish and Taylor 1979 Loomis and Welber1982 Ujike 1982 Pietranik et al 2006 Baginski et al 2009Chistyakova and Latypov 2009)

According to previous studies dealing with dikes (Loomis andWelber 1982 Ujike 1982 Pietranik et al 2006 Chistyakova andLatypov 2009) plagioclase exhibits steep compositional gradients interms of An contents (Fig 4) coupled with Mg+Fe increase and Na+K decrease from dike core-to-rim (Fig 4) Notably Fe increases withincreasing An (Smith 1983 Smith and Brown 1988) in agreementwith plagioclase stoichiometry decreasing amounts of Si4+ leavelarger available T- and A-sites for Fe substitution as the An content isincreased (Tegner 1997) Recently Mollo et al (in press-a)experimentally demonstrated that as the cooling rate is increasedthe plagioclase composition is characterized by decreasing Si+Na+K

and increasing Al+Ca+Fe+Mg contents Moreover Iezzi et al(2011) showed that early formed microlites of plagioclase do not re-equilibrate with the melt under dynamic conditions thus resulting inmuch more An-rich plagioclase than equilibrium crystals Accordingto crystal nucleation theory in silicate crystals and liquids theenergetic barrier of nucleation roughly scales with the number ofIVSiO bonds and to a less extent of IVAlO bonds ie degree ofpolymerization and average bond strength (Iezzi et al 2008 andreferences therein) consequently the nucleation of more basic (ieAn-rich) plagioclases prevails over that of more acid (ie Ab-rich)crystals (Iezzi et al 2011)

Similarly to plagioclase the composition of clinopyroxene alsochanges under variable kinetic conditions (Smith and Lindsley 1971Mevel and Velde 1976 Grove and Bence 1977 1979 Grove andRaudsepp 1978 Coish and Taylor 1979 Gamble and Taylor 1980Lofgren et al 2006 Mollo et al 2010) Fig 5 shows thatclinopyroxenes from dike core-to-rim are enriched in Al Ti and Fe3+This occurs in response to a disequilibrium growth of the crystalsleading to increasing concentration of incompatible elements withincreasing cooling rate (Lofgren et al 2006) Mollo et al (2010) haveexperimentally observed that as the cooling rate is increased Al ismuchmore compatible in clinopyroxene and that such occurrence isresponsible for the Tschermak-rich crystals found at the dike chilledmargins (Ujike 1982 Baginski et al 2009) Moreover our clinopyrox-enes are characterized by high IVAl contents (N0173 apfu) whereas theamount of VIAl is consistently low (b005 apfu) as a consequence of lowpressure conditions (Muntildeoz and Sagredo 1974) and in agreementwiththe dike stratigraphic position

Several studies evidenced that titanomagnetites found in rapidlycooled rocks have compositions indicative of departure fromequilibrium (Hammond and Taylor 1982 Nakamura 1995 Venezkyand Rutherford 1997) Generally the ulvospinel component oftitanomagnetite drastically decreases with increasing cooling rate(Nakada and Motomura 1999 Zhou et al 2000) Consequently Usp-poor titanomagnetites occur at the dike chilled margins where rapidcooling rates occur (Smith and Prevot 1977) Accordingly Fig 6shows that the kinetic control on titanomagnetite crystallizationcauses the ulvospinel content to progressively decrease from dike

Fig 4 Plagioclase composition from sample DK1 to DK7 Recalculation of feldspar analyses is done on an 8-oxygen basis Major element concentrations are reported in atoms performula unit (apfu) Anorthite (An) Albite (Ab) and Orthoclase (Or) components are expressed in mole percent Error bars are within symbols

43S Mollo et al Chemical Geology 288 (2011) 39ndash52

It is worth noting that Mt Etna is an open-system degassingvolcano generating magmas that contain relatively high watercontents (up to 35 wt) (cf Ferlito and Lanzafame 2010 andreferences therein) Therefore the crystallization of some Etneanmagmas can be driven by volatile exsolution due to decompression(Metrich and Rutherford 1998) However in our dike samples thelow amount of bubbles (b5 vol) and the absence of hydrousminerals (ie amphibole and phlogopite) characteristic of water-bearing Etnean magmas (cf Tanguy et al 1997) indicate that acrystallization induced by decompression minimally affected thesolidification of magma and that according to previous studiesvolatile phases migrate passively through cracks and vugs ofoverlapping lava flows (Heap et al 2009 2011)

41 Plagioclase clinopyroxene and titanomagnetite compositionalvariations

Compositions of plagioclase clinopyroxene and titanomagnetitefrom our samples resemble those observed for crystals found in manyEtnean lavas However minerals show marked and continuouscompositional variations from sample DK1 to DK7 (Figs 4 5 and 6)Such variations testify to an increasing disequilibrium growth of thecrystals from the innermost and slowly cooled part of the dike to theoutermost and rapidly cooled portion (Smith and Lindsley 1971Mevel and Velde 1976 Coish and Taylor 1979 Loomis and Welber1982 Ujike 1982 Pietranik et al 2006 Baginski et al 2009Chistyakova and Latypov 2009)

According to previous studies dealing with dikes (Loomis andWelber 1982 Ujike 1982 Pietranik et al 2006 Chistyakova andLatypov 2009) plagioclase exhibits steep compositional gradients interms of An contents (Fig 4) coupled with Mg+Fe increase and Na+K decrease from dike core-to-rim (Fig 4) Notably Fe increases withincreasing An (Smith 1983 Smith and Brown 1988) in agreementwith plagioclase stoichiometry decreasing amounts of Si4+ leavelarger available T- and A-sites for Fe substitution as the An content isincreased (Tegner 1997) Recently Mollo et al (in press-a)experimentally demonstrated that as the cooling rate is increasedthe plagioclase composition is characterized by decreasing Si+Na+K

and increasing Al+Ca+Fe+Mg contents Moreover Iezzi et al(2011) showed that early formed microlites of plagioclase do not re-equilibrate with the melt under dynamic conditions thus resulting inmuch more An-rich plagioclase than equilibrium crystals Accordingto crystal nucleation theory in silicate crystals and liquids theenergetic barrier of nucleation roughly scales with the number ofIVSiO bonds and to a less extent of IVAlO bonds ie degree ofpolymerization and average bond strength (Iezzi et al 2008 andreferences therein) consequently the nucleation of more basic (ieAn-rich) plagioclases prevails over that of more acid (ie Ab-rich)crystals (Iezzi et al 2011)

Similarly to plagioclase the composition of clinopyroxene alsochanges under variable kinetic conditions (Smith and Lindsley 1971Mevel and Velde 1976 Grove and Bence 1977 1979 Grove andRaudsepp 1978 Coish and Taylor 1979 Gamble and Taylor 1980Lofgren et al 2006 Mollo et al 2010) Fig 5 shows thatclinopyroxenes from dike core-to-rim are enriched in Al Ti and Fe3+This occurs in response to a disequilibrium growth of the crystalsleading to increasing concentration of incompatible elements withincreasing cooling rate (Lofgren et al 2006) Mollo et al (2010) haveexperimentally observed that as the cooling rate is increased Al ismuchmore compatible in clinopyroxene and that such occurrence isresponsible for the Tschermak-rich crystals found at the dike chilledmargins (Ujike 1982 Baginski et al 2009) Moreover our clinopyrox-enes are characterized by high IVAl contents (N0173 apfu) whereas theamount of VIAl is consistently low (b005 apfu) as a consequence of lowpressure conditions (Muntildeoz and Sagredo 1974) and in agreementwiththe dike stratigraphic position

Several studies evidenced that titanomagnetites found in rapidlycooled rocks have compositions indicative of departure fromequilibrium (Hammond and Taylor 1982 Nakamura 1995 Venezkyand Rutherford 1997) Generally the ulvospinel component oftitanomagnetite drastically decreases with increasing cooling rate(Nakada and Motomura 1999 Zhou et al 2000) Consequently Usp-poor titanomagnetites occur at the dike chilled margins where rapidcooling rates occur (Smith and Prevot 1977) Accordingly Fig 6shows that the kinetic control on titanomagnetite crystallizationcauses the ulvospinel content to progressively decrease from dike

Table 2Representative electron microprobe analyses of clinopyroxenes sd represents the standard deviation (in parenthesis number of averaged analyses)

DK1 DK2 DK3 DK4 DK5 DK6 DK7

wt Core sd(10)

Rim sd(10)

Core sd(10)

Rim sd(10)

Core sd(10)

Rim sd(10)

Core sd(10)

Rim sd(10)

Core sd(10)

Rim sd(10)

Core sd(10)

Rim sd(10)

Core sd(10)

Rim sd(10)

SiO2 4516 018 4884 020 4484 010 4835 010 4452 013 4787 014 4436 015 4763 016 4419 036 4738 039 4387 037 4690 039 4355 015 4641 016TiO2 247 004 140 002 260 002 154 001 273 003 168 002 279 001 175 001 286 004 182 003 298 005 196 003 311 004 210 003Al2O3 813 006 487 004 839 004 529 002 865 003 572 002 878 003 593 002 891 011 614 008 917 010 657 007 943 006 699 005FeO 916 006 864 006 931 005 879 005 945 006 894 006 952 006 902 005 960 010 909 009 974 009 924 008 989 006 939 006MnO 017 002 021 003 017 002 022 003 018 002 022 003 018 002 023 003 018 003 023 003 019 003 024 004 019 002 024 003MgO 1290 031 1261 031 1276 017 1239 017 1262 021 1217 020 1255 017 1206 017 1248 036 1194 035 1234 040 1172 038 1220 026 1150 024CaO 2104 025 2254 027 2089 014 2245 015 2074 020 2237 021 2067 014 2233 015 2059 031 2228 033 2044 032 2220 035 2029 021 2211 023Na2O 040 000 048 001 043 001 050 001 046 000 053 001 048 000 054 000 049 001 055 001 052 001 058 001 055 001 060 001K2O 001 000 003 000 001 000 003 000 002 000 004 000 002 000 004 000 002 000 004 000 003 000 004 000 003 000 004 000Total 9944 9962 9940 9957 9936 9953 9934 9951 9932 9948 9928 9944 9924 9939

Formula on the basis of 6 oxygensSi 1689 1827 1678 1810 1668 1794 1663 1786 1657 1778 1647 1762 1636 1746Ti 0311 0173 0322 0190 0332 0206 0337 0214 0343 0222 0353 0238 0364 0254AlIV 0047 0041 0049 0044 0050 0047 0051 0049 0051 0050 0053 0053 0054 0056AlVI 0069 0039 0073 0043 0077 0047 0079 0049 0081 0051 0084 0055 0088 0059Fe3+ 0154 0090 0159 0097 0163 0104 0165 0108 0167 0111 0171 0118 0175 0125Fe2+ 0132 0181 0133 0178 0133 0176 0134 0175 0134 0174 0135 0172 0135 0170Mg 0719 0703 0712 0691 0705 0680 0701 0674 0697 0668 0690 0656 0683 0645Mn 0005 0007 0006 0007 0006 0007 0006 0007 0006 0007 0006 0007 0006 0008Ca 0843 0903 0838 0901 0833 0898 0830 0897 0827 0896 0822 0894 0817 0891Na 0029 0035 0031 0037 0033 0038 0035 0039 0036 0040 0038 0042 0040 0044K 0000 0001 0001 0002 0001 0002 0001 0002 0001 0002 0001 0002 0001 0002Fe2+Fe3+ 1167 0497 1194 0543 1220 0589 1233 0613 1246 0637 1272 0686 1298 0737Mg 8448 7957 8429 7949 8409 7940 8399 7936 8389 7931 8369 7922 8349 7913Di 0508 0612 0497 0600 0485 0588 0480 0581 0474 0575 0463 0563 0452 0551Hd 0093 0157 0093 0155 0092 0152 0091 0151 0091 0150 0090 0148 0089 0145En 0106 0045 0108 0046 0110 0046 0111 0046 0112 0046 0114 0047 0116 0047Fs 0019 0012 0020 0012 0021 0012 0021 0012 0021 0012 0022 0012 0023 0012CaTs 0077 0045 0079 0048 0081 0052 0082 0054 0083 0056 0086 0059 0088 0063CaFeTs 0069 0039 0073 0043 0077 0047 0079 0049 0081 0051 0084 0055 0088 0059CaTiTs 0095 0050 0096 0054 0097 0059 0098 0061 0098 0064 0099 0068 0100 0073Jd 0029 0036 0032 0038 0034 0040 0035 0041 0037 0042 0039 0044 0042 0046

44SM

olloet

alChem

icalGeology

288(2011)

39ndash52

Fig 5 Clinopyroxene composition from sample DK1 to DK7 Recalculation of mineral analyses is done on a 6-oxygen basis Major element concentrations are reported in atoms performula unit (apfu) Diopside (Di) hedenbergite (Hd) enstatite (En) calcium Tschermak (CaTs) and iron-bearing calcium Tschermak (CaFeTs) components are expressed in molepercent Error bars are within symbols

45S Mollo et al Chemical Geology 288 (2011) 39ndash52

sample DK1 to DK7 Moreover the disequilibrium growth oftitanomagnetite also favors higher concentrations of incompatibleelements ie Al and Mg (Table 3) that are incorporated in the crystallattice

10To summarize from dike core-to-rim crystals of plagioclaseclinopyroxene and titanomagnetite are characterized by increasing Al

Table 3Representative electron microprobe analyses of titanomagnetites sd represents the standa

DK1 DK2 DK3 DK4

wt sd(10) sd(10) sd(10)

TiO2 2187 216 1986 253 1785 253 168Al2O3 102 001 193 001 285 002 33FeO 7235 043 7304 088 7374 111 740MnO 096 013 083 011 070 010 06MgO 114 007 138 008 162 009 17Total

Formula the basis of 3 cationsTi 0612 0553 0494 04Al 0045 0084 0124 01Mn 0030 0026 0022 00Mg 0063 0076 0089 00Fe3+ 0732 0810 0888 09Fe2+ 1518 1451 1384 13Usp 6255 5846 5403 516FeO mol 60828 60228 59592 592Fe2O3 mol 14664 16820 19114 203TiO2 mol 24507 22952 21294 204

contents (Figs 1 and2 andTable 3) At the same time as the cooling rateis increased both silicate and oxide crystals show lower amounts of Si4+

and Ti4+ cations (Tables 1 2 and 3) respectively This feature reflectsthe nucleation process that requires breaking and formation ofmolecular units the most energetic bonds of Si4+ and Ti4+ tetravalentcations require a higher energy for molecular rearrangement and

rd deviation (in parenthesis number of averaged analyses)

DK5 DK6 DK7

sd(10) sd(10) sd(10) sd(10)

4 219 1583 207 1382 185 1181 1411 001 376 001 468 002 559 0029 168 7443 090 7513 046 7582 0403 009 057 008 044 006 031 0044 015 185 009 209 010 233 009

65 0436 0379 032343 0162 0201 023920 0018 0014 001095 0101 0114 012626 0965 1040 111650 1318 1252 11877 4924 4412 386560 58918 58193 5741221 21565 24181 2698619 19517 17626 15602

Fig 5 Clinopyroxene composition from sample DK1 to DK7 Recalculation of mineral analyses is done on a 6-oxygen basis Major element concentrations are reported in atoms performula unit (apfu) Diopside (Di) hedenbergite (Hd) enstatite (En) calcium Tschermak (CaTs) and iron-bearing calcium Tschermak (CaFeTs) components are expressed in molepercent Error bars are within symbols

45S Mollo et al Chemical Geology 288 (2011) 39ndash52

sample DK1 to DK7 Moreover the disequilibrium growth oftitanomagnetite also favors higher concentrations of incompatibleelements ie Al and Mg (Table 3) that are incorporated in the crystallattice

10To summarize from dike core-to-rim crystals of plagioclaseclinopyroxene and titanomagnetite are characterized by increasing Al

Table 3Representative electron microprobe analyses of titanomagnetites sd represents the standa

DK1 DK2 DK3 DK4

wt sd(10) sd(10) sd(10)

TiO2 2187 216 1986 253 1785 253 168Al2O3 102 001 193 001 285 002 33FeO 7235 043 7304 088 7374 111 740MnO 096 013 083 011 070 010 06MgO 114 007 138 008 162 009 17Total

Formula the basis of 3 cationsTi 0612 0553 0494 04Al 0045 0084 0124 01Mn 0030 0026 0022 00Mg 0063 0076 0089 00Fe3+ 0732 0810 0888 09Fe2+ 1518 1451 1384 13Usp 6255 5846 5403 516FeO mol 60828 60228 59592 592Fe2O3 mol 14664 16820 19114 203TiO2 mol 24507 22952 21294 204

contents (Figs 1 and2 andTable 3) At the same time as the cooling rateis increased both silicate and oxide crystals show lower amounts of Si4+

and Ti4+ cations (Tables 1 2 and 3) respectively This feature reflectsthe nucleation process that requires breaking and formation ofmolecular units the most energetic bonds of Si4+ and Ti4+ tetravalentcations require a higher energy for molecular rearrangement and

rd deviation (in parenthesis number of averaged analyses)

DK5 DK6 DK7

sd(10) sd(10) sd(10) sd(10)

4 219 1583 207 1382 185 1181 1411 001 376 001 468 002 559 0029 168 7443 090 7513 046 7582 0403 009 057 008 044 006 031 0044 015 185 009 209 010 233 009

65 0436 0379 032343 0162 0201 023920 0018 0014 001095 0101 0114 012626 0965 1040 111650 1318 1252 11877 4924 4412 386560 58918 58193 5741221 21565 24181 2698619 19517 17626 15602

Fig 6 Titanomagnetite compositions plotted in the system FeOndashTiO2ndash12Fe2O3 fromsample DK1 to DK7 Oxides are aligned to the ulvospinel (usp)ndashmagnetite (mt) tie line

Table4

Represen

tative

electron

microprob

ean

alyses

ofolivinessdrepresen

tsthestan

dard

deviation(inpa

renthe

sisnu

mbe

rof

averag

edan

alyses)

DK1

DK2

DK3

DK4

DK5

DK6

DK7

wt

Core

sd (10)

Rim

sd (10)

Core

sd (10)

Rim

sd (10)

Core

sd (10)

Rim

sd (10)

Core

sd (10)

Rim

sd (10)

Core

sd (10)

Rim

sd (10)

Core

sd (10)

Rim

sd (10)

Core

sd (10)

Rim

sd (10)

SiO2

3677

006

3570

006

3703

008

3595

008

3679

012

3572

012

3702

012

3594

012

3680

009

3573

009

3701

005

3593

005

3691

007

3583

007

TiO2

006

001

007

001

006

001

007

001

006

001

007

001

006

001

007

001

006

001

007

001

006

001

007

001

006

001

007

001

FeO

2933

040

3430

047

2886

035

3376

041

2930

045

3428

052

2889

066

3378

077

2928

036

3425

042

2891

015

3381

018

2912

031

3406

036

MnO

089

012

122

017

062

008

085

012

088

012

120

016

063

009

086

012

086

012

118

016

065

008

088

011

077

009

105

012

MgO

3329

203

2857

174

3211

147

2755

126

3323

152

2851

130

3217

245

2761

210

3317

151

2846

130

3223

128

2766

110

3276

130

2811

112

CaO

045

000

085

001

044

000

084

001

045

000

085

001

044

000

084

001

045

000

085

000

044

000

084

000

045

000

084

001

Total

10079

10071

9912

9902

10071

10063

9921

9911

10062

10055

9929

9919

10005

9997

Form

ulaon

theba

sisof

4ox

ygen

sSi

098

8098

7100

7100

7098

9098

8100

6100

6099

0098

9100

5100

5099

7099

6Fe

065

9079

3065

6079

1065

9079

3065

7079

1065

8079

3065

7079

1065

8079

2Mn

002

0002

9001

4002

0002

0002

8001

5002

0002

0002

8001

5002

1001

8002

5Mg

133

3117

8130

2115

1133

1117

7130

4115

2133

0117

5130

5115

3131

9116

5Ca

001

3002

5001

3002

5001

3002

5001

3002

5001

3002

5001

3002

5001

3002

5Fo

6693

5975

6648

5927

6690

5973

6650

5930

6688

5971

6653

5932

6673

5954

Fa33

07

4025

3352

4073

3310

4027

3350

4070

3312

4029

3347

4068

3327

4046

46 S Mollo et al Chemical Geology 288 (2011) 39ndash52

consequently these cations are less incorporated in nucleating phaseswith respect to Al3+ and Fe3+ more mobile cations (Kirkpatrick 1975Mysen and Richet 2005 Iezzi et al 2011)

42 Olivine disequilibrium composition

The forsterite content in olivine is strictly related to the melt Mgover a wide range of bulk compositions temperatures and oxygenfugacities (Berndt et al 2005 Toplis 2005)

Olivines from dike samples have a forsterite content (Fo66-67)lower than that (Fo85) would be predicted to crystallize from thebasalt (Mg=55) at equilibrium by means of olivine-melt pairs fromnatural samples (Tanguy et al 1997) laboratory experiments(Metrich and Rutherford 1998) and computations performed withthe PETROLOG program (Danyushevsky 2001) based on the parti-tioning of Fe and Mg between olivine and melt Generally trachyba-salts (56bMgb68) from Mt Etna volcano equilibrate with moreforsteritic (Fo75-88) olivines (Tanguy et al 1997) Metrich andRutherford (1998) experimentally demonstrated that these olivinescrystallize at low pressures (27ndash80 MPa) and under water-presentconditions (27ndash16 wt) thus suggesting shallow reservoirs fortrachybasaltic melts from Mt Etna volcano (see also Mollo et al inpress-b and references therein)

Moreover differently to that observed for clinopyroxene plagio-clase and titanomagnetite there is no evidence of textural maturationand compositional variation for olivine crystals from dike core-to-rimHowever kinetically-controlled experiments conducted on a Fe-richbasaltic melt by Hammer (2008) have evidenced that the forsteritecontent in olivine increases with increasing cooling rate This impliesthat under disequilibrium growth conditions the olivine compositionapproaches to that of high-temperature formation Such a featurematches with the experimental observation that early-formedcrystals of clinopyroxene (Hammer 2008 Mollo et al 2010) andplagioclase (Iezzi et al 2011 Mollo et al in press-a) do not re-equilibrate with the melt at fast cooling rates consequentlythermometers based on crystal-liquid exchange reactions faithfullyestimate higher crystallization temperatures Therefore if olivinesfound in dike samples formed under dynamic conditions a higherforsterite content would have been expected in the crystals relative tothat predicted at equilibrium Additionally cooling experimentsconducted on a trachybasalt from Mt Etna reproducing dikeemplacements under hydrous and anhydrous conditions havedemonstrated that clinopyroxene crystallization prevails over thatof olivine (Del Gaudio et al 2010) consequently olivine is not astable in Etnean trachybasalts experiencing rapid cooling rates

Fig 7 cpx-meltKdFe-Mg values are plotted as a function of sample position Error bars arewithin symbols cpx-meltKdFe-Mg is calculated according to both Roeder and Emslie(1970) (closed square) and Putirka (1999) models (closed diamond)

Fig 8 pl-liqKdFe-Mg pl-liqKdCa-Na and pl-liqKdAb-An are plotted as a function of sampleposition Error bars are within symbols

47S Mollo et al Chemical Geology 288 (2011) 39ndash52

We therefore conclude that Fe-rich olivines found in our dikesamples must have been introduced to the original melt asxenocrystals Magma mixing has been recently evidenced at MtEtna volcano immediately before andor during the start of magmaascent from the crustal reservoir (Viccaro et al 2006 Ferlito et al2008 2009 Corsaro et al 2009) Moreover Tanguy et al (1997) havesuggested that during the Ellittico magmatic activity low-pressurefractionation processes have occasionally shifted the original trachy-basaltic composition toward more evolved trachyandesitic andtrachytic melts in equilibrium with Fe-rich olivines

43 Mineral-melt partition coefficients

Clinopyroxene and plagioclase are the most important magmaticminerals as they frequently occur as a liquidus phase or during cotecticcrystallization (Phinney 1992) Their crystal-melt partition coefficientsare broadly used in geochemical and petrological applications such asfor the modeling of melt evolution through fractionation and partialmelting as well as the assessment of equilibrium crystallizationconditions The most used partition coefficients are those based onFendashMg (xls-liqKdFe-Mg=(xlsXFeO xlsXMgO) (liqXFeO liqXMgO)) and Ca-Na(xls-liqKdCa-Na=(xlsXCaO xlsXNa2O) (liqXCaO liqXNa2O)) exchangereactions between clinopyroxene plagioclase and melt The Ab-Anexchange (xls-liqKdAb-An=(xlsXAbtimes liqXAlO15times liqXCaO) (xlsXAntimesliqXNaO05times liqXSiO2)) partition coefficient between plagioclase andmelt was also proposed by Putirka (2008) as a new tool to assessthe equilibrium crystallization condition for plagioclase-bearingrocks

With the aim to constrain the initial cooling conditions ofintruding magma we calculated values of xls-liqKdFe-Mg xls-liqKdCa-Naand xls-liqKdAb-An for each dike sample using the chemistry of early-formed clinopyroxene and plagioclase crystals (ie the crystal cores)and the initial melt composition (ie the dike bulk composition)Notably at the time of cooling the initial dike melt becameprogressively more differentiated as crystallization proceeded there-fore crystal rims have equilibrated with a more evolved melt(possibly trachyandesite or trachyte) from which at the final stageof solidification the fine-grained groundmass formed Consequentlycrystal rims result to be in strong disequilibriumwith the original dikebulk composition eg cpx-liqKdFe-Mg=037ndash040 and plg-liqKdAb-An=035ndash049

Theoretical and experimental studies demonstrated that FendashMgexchange partition coefficient between clinopyroxene and liquid maybe extremely useful for the assessment of equilibrium conditions(Putirka et al 2003 and references therein) These studies evidencedthat cpx-liqKdFe-Mg has a constant equilibrium value of 027plusmn003(Putirka 1999) Fig 7 shows a fairly close correspondence betweenthe equilibrium range for cpx-liqKdFe-Mg and the values measuredfor each dike sample A further test for equilibrium was alsoperformed using the T-sensitive model reported in Putirka (2008)This model is based on the deviations in observed and calculatedvalues for cpx-liqKdFe-Mg derived from experimental observationsthese values are not perfectly invariant and consequently they show

a slight temperature dependency Results for cpx-liqKdFe-Mg are strictlyclose to the equilibrium value of 026 (Fig 7) Using this valuecalculations with PETROLOG program (Danyushevsky 2001) high-light that differently from olivine clinopyroxene (Mg=83plusmn1) is inequilibrium with the dike melt (Mg=55)

pl-liqKdFe-Mg pl-liqKdCa-Na and pl-liqKdAb-An for plagioclase-meltpairs have been the topic of many studies investigating their variationas a function of pressure temperature and melt composition (egLonghi et al 1976 Grove et al 1982 Mahood and Baker 1986 Usslerand Glazner 1989 Sugawara 2001) Recently it has been alsodemonstrated that as the cooling rate is increased plagioclasecrystals are progressively enriched in Fe and Ca leading to highervalues for partition coefficients (Conte et al 2006 Brugger andHammer 2010 Mollo et al in press-a) Accordingly from theinnermost and slowly cooled part of the dike toward the outermostand rapidly cooled portion pl-liqKdFe-Mg and pl-liqKdCa-Na progressivelyincrease from 198 to 271 and from 152 to 2 respectively (Fig 8)Notably values for pl-liqKdFe-Mg are close to those experimentally deter-mined under intermediate oxidizing conditions ie from NNO+1 toNNO+2 buffer (Mollo et al in press-a) that are common for theEtnean magmas (Metrich and Rutherford 1998) Fig 8 also showsthat pl-liqKdAb-An value progressively decreases from 027 to 020from innermost to the outermost part of the dike Such variation isdue to the higher An content in plagioclase (Fig 4) over the coolingtimescale of the dike (Iezzi et al 2011 Mollo et al in press-a)Notably regression analysis on plagioclase-liquid pairs reported inPutirka (2008) have evidenced that pl-liqKdAb-An yields a constantvalue over two temperature intervals pl-liqKdAb-An=010plusmn005at Tb1050 degC and 027plusmn011 for Tge1050 degC According to theseanalyses pl-liqKdAb-An values determined in this study suggest thatplagioclases should have equilibrated at Tge1050 degC (see below)

To conclude it is worth nothing that clinopyroxene and plagio-clase partition coefficients recalculated by subtracting the contribu-tion of inherited olivines from the dike bulk composition do not varysignificantly this is due to the lower content of olivine (6 vol)relative to that of clinopyroxene (20 vol) and plagioclase (68 vol)

Fig 10 Melt-water concentration predicted for each dike sample using Model H ofPutirka et al (2005) and the hygrometer of Lange et al (2009) Error bars are withinsymbols

Fig 9 Temperature estimates for each dike sample using thermometers modeled on the exchange reactions DiHd-Jd and Hd-CaTs for clinopyroxene (Putirka et al 1996 2003Putirka 2008) and AbAn for plagioclase (Putirka 2005 2008) error bars are within symbols (a) Differences (Δ) between the clinopyroxene components predicted using themodelof Putirka (1999) and those measured for crystals from dike samples ΔDiHd (Di+Hd) ΔsumTs (CaTs+CaTiTs) and ΔJd values are plotted for each dike sample

48 S Mollo et al Chemical Geology 288 (2011) 39ndash52

44 Dike crystallization conditions

Since partition coefficients determined using the compositions ofthe crystal cores and the dike bulk composition haveminor deviationsfrom the equilibrium range determined for crystal-liquid pairstherefore thermometers based on clinopyroxene-liquid and plagio-clase-liquid equilibria will allow us to constrain the dike crystalliza-tion temperature Thermometers modeled on the exchange reactionsDiHd-Jd and Hd-CaTs for clinopyroxene (Putirka et al 1996 2003Putirka 2008) and AbAn for plagioclase (Putirka 2005 2008) wereused Results are displayed in Fig 9a Thermometers predict acrystallization temperature of 1126plusmn26 degC and 1119plusmn19 degC forclinopyroxene and plagioclase respectively These two temperaturesand their associated uncertainties were determined considering thewhole compositional variations of clinopyroxene and plagioclasecrystals from dike core-to-rim as well as the temperature intervalspredicted by means of both clinopyroxene-liquid and plagioclase-liquid thermometers Notably two differentmineral-melt equilibriumyield similar temperatures suggesting a few degrees differencebetween the clinopyroxene and plagioclase crystallization Moreoverthe predicted temperatures were never less than 1050 degC inagreement with the temperature interval proposed by Putirka(2008) on the basis of pl-liqKdAb-An Since the plagioclase thermometeris sensitive to melt-water content (Putirka 2008) a constant watervalue of 16 wt was used for the calculation (see below) NotablyFig 9a shows that from innermost to the outermost part of the dikethermometers yield an increasing temperature formation for thecrystals the predicted temperature slightly increases for clinopyrox-ene and plagioclase (not appreciable for plagioclase because of thelarge scale of Fig 9a) Such a feature is due to the kinetic control ofcooling rate on the composition of early-formed crystals Authorshave experimentally demonstrated that nuclei of clinopyroxene andplagioclase do not re-equilibrate with the melt over a rapid coolingrate (Mollo et al 2010 in press-a) Therefore mineral compositionsare shifted toward those of high-temperature formation ie tscher-mak- (Fig 5) and anorthite-rich (Fig 4) for clinopyroxene andplagioclase respectively In contrast a slow cooling rate causesmineral compositions to migrate towards those corresponding tothe equilibrium crystallization temperature Since thermometers arebased on the assumption of equilibrium thus any form of disequilib-rium is evidenced by monotonic trends toward higher temperatureformations (Fig 9a)

Moreover Mollo et al 2010 experimentally demonstrated that anequilibrium value for cpx-liqKdFe-Mg (Fig 7) does not imply equilibriumcrystallization conditions This occurs because during disequilibriumgrowth elements incompatible with clinopyroxene are much more

incorporated in the crystal lattice compared to the compatible onesie Fe and Mg that do not significantly fractionate (Lofgren et al2006 Mollo et al 2010) Putirka (1999) proposed an extensivemodelto test clinopyroxenendashliquid equilibria based on the comparison of thepredicted components for clinopyroxene (calculated via regressionanalysis of clinopyroxenendashliquid pairs in equilibrium conditions) withthose analyzed in the investigated crystals The author suggested thatdifferences (Δ) between the predicted and observed components willprovide a more robust test for equilibrium than cpx-liqKdFe-Mg InFig 9b we compared DiHd (Di+Hd) sumTs (CaTs+CaTiTs) and Jdcomponents predicted using the model of Putirka (1999) with thosecalculated for our clinopyroxenes (Table 2) Results indicate thatΔDiHd ΔsumTs and ΔJd progressively increase from dike core-to-rimyielding higher temperature predictions Therefore tests proposed byPutirka (1999) should be mostly considered in order to determineequilibrium crystallization conditions

The water content of the dike melt was calculated using bothhygrometers of Putirka (2005) and Lange et al (2009) based on theplagioclase-liquid exchange reaction between the anorthite and albitecomponents Since melt-water content is highly sensitive to temper-ature we used for the calculation the average temperature of 1126 degCsuggested by clinopyroxene thermometers Results are reported inFig 10 The water concentration predicted for the melt is 16plusmn03This value agrees with that determined for trachybasaltic lavas fromMt Etna volcano (Del Gaudio et al 2010 and references therein)Notably Mollo et al (in press-a) evidenced that the different behaviorof the two hygrometers (Fig 10) is due to the effect of coolingdynamics the model of Putirka (2005) reflects the effect of coolingrate on plagioclase composition (ie higher An contents) in contrastthe model of Lange et al (2009) acts as a proxy for temperature (ie1126 degC) yielding an almost constant melt-water concentration(Fig 10)

49S Mollo et al Chemical Geology 288 (2011) 39ndash52

Alkaline rocks are usually characterized by a high oxygen fugacitythat varies from QFM to NNO+2 buffer (Ryabchikov and Kogarko2006 and references therein) Oxygen fugacity controls the speciationof chemical components in volatile and melt and their impact onmagma genesis and differentiation Importantly the fO2 influences thevalence state of iron in the melt and consequently the ironpartitioning between melt and crystals We determined the redoxstate of magma using the Kress and Carmichael (1991) equation andthe Fe3+Fe2+ ratio recorded by clinopyroxene cores Results arereported in Fig 11 for each dike sample and the averaged value ofNNO+17 (plusmn01) agrees with oxidizing conditions estimated forEtnean magmas (Metrich and Rutherford 1998) One could infer thatthe strong magma degassing at Mt Etna may influence the oxygenfugacity of the system A commonly invoked mechanism for theoxidation of magmatic liquids is related to the preferential loss of H2H2S or CO component during degassing (Mathez 1984 Sisson andGrove 1993) However it has been demonstrated that in activevolcanoes (i) the oxygen fugacity of the exsolved gasses is controlledby that of the host magma (Gerlach 1993) (ii) the fO2 value of thehost magma does not change even after substantial degassing coolingand crystallization (Lange and Carmichael 1996) and (iii) coolingdikes are characterized by crystallization at almost constant oxygenfugacity (Gibb 1971 Ujike 1982) Notably in a recent study Franceet al (2010) reported a new method to estimate fO2 from electronprobe microanalyses of plagioclase and clinopyroxene This method isnot based on stoichiometric calculations but on the differentpartitioning behavior of ferrous and ferric iron between bothmineralsand the melt plagioclase can incorporate more Fe3+ than Fe2+ whileclinopyroxene can incorporate more Fe2+ than Fe3+ We used themodel of France et al (2010) to estimate the redox state of the meltResults show that fO2 increases of about 1 log unit from dike core-to-rim (Fig 11) Consequently the model of France et al (2010) isaffected by a larger error on estimate than the almost constant valuedetermined by the equation of Kress and Carmichael (1991) Such afeature is due to the progressive incorporation of iron in plagioclasecrystal lattice with increasing cooling rate (Fig 4) Since iron isincompatible with plagioclase it results much more susceptible todynamic crystallization conditions relative to clinopyroxene (Molloet al in press-a) Therefore early-formed microlites of plagioclasebecomemuch more Fe-rich than equilibrium crystals over the coolingtimescale of the dike This highlights the importance of decipheringthe cooling history of dikes mostly when their crystals grow underdynamic conditions

To conclude melt crystallization conditions estimated above wererun with the MELTS program (Ghiorso and Sack 1995) Resultssuggest a liquidus temperature of about 1200 degC very close to that(1211 degC) experimentally determined by Del Gaudio et al (2010) for atrachybasaltic composition fromMt Etna volcano Moreover the meltis cosaturated by clinopyroxene and plagioclase at an equilibriumtemperature of 1100 degC that is comparable to clinopyroxene (1126plusmn

Fig 11 Oxygen fugacity of melt predicted for each dike sample using themodel of Kressand Carmichael (1991) and of France et al (2010) Error bars are within symbols

26 degC) and plagioclase (1119plusmn19 degC) crystallization temperaturepredicted by thermometers Notably the compositions of plagioclase(An73Ab25Or1) and clinopyroxene (Di88Hd2En8Fs2) calculated usingMELTS code match with those of the crystals found in the dikesamples importantly olivine was never formed in thermodynamicequilibrium with the dike melt under any plausible conditions

45 Dike cooling history

According to what is discussed above the chemical variationrecorded by plagioclase clinopyroxene and titanomagnetite is due tomore rapid cooling conditions from dike core-to-rim Mollo et al(2010) proposed an empirical equation relating the clinopyroxenecompositional variation (DHTE) to the cooling rate (CR) experiencedby trachybasaltic magmas from Mt Etna volcano As it is displayed inFig 12a themodel is based on the assumption that as the cooling rateis increased the clinopyroxene components Di+Hd (DH) decreasewhereas En+CaTs+CaFeTs (TE) increase due to a continuoussubstitution of M2(Mg Fe2+) with M1(Al Fe3+) and of TSi4+

with TAl3+ (Deer et al 2001 Etzel et al 2007) We used thismodel to calculate the cooling rates responsible to clinopyroxenecompositional variations from sample DK1 to DK7 Results are shownin Fig 12b where experimental data from Mollo et al (2010) are alsoreported for comparison From innermost to the outermost part of thedike the cooling rate increases from 002plusmn0002 (ie 12plusmn01 degCh)to 113plusmn011 degCmin (ie 678plusmn54 degCh) This cooling interval is veryclose to that (002ndash17 degCmin) determined by Ujike (1982) for a set ofsamples collected at the Shirotori-Hikeda dike swarm (ShikokuJapan) over a distance of 18 m Moreover cooling rates determinedfrom this study are comparable to those estimated from inner to the

Fig 12 DH (Di+Hd) versus TE (CaTs+CaFeTs+En) plot of clinopyroxenes from sampleDK1 to DK7 the linear fit is reported as gray line and error bars are within symbols(a) Cooling rates experienced by clinopyroxene from each dike sample determined by theequation of Mollo et al (2010) the fit is reported as gray dashed line (b)

Fig 13 Dike cooling history determined by thermal modeling based on an explicitfinite-difference scheme

50 S Mollo et al Chemical Geology 288 (2011) 39ndash52

outer part of lava flows (eg Wilding et al 1995 Zhou et al 2000Gottsmann et al 2004)

However it must be noted that intruding magmas experiencevariable cooling rates over solidification time The cooling rateoccurring in any portion of a dike does not decrease linearly andmonotonically as in experiments actually in natural environmentsthe cooling of magma is also determined by the variation of thermalgradient with time and the variation of thermal diffusivity withtemperature (Blundy et al 2006 Whittington et al 2009) Thereforethe experimental equation of Mollo et al (2010) allows us todetermine the cooling rate recorded by clinopyroxene compositionduring the early stage of crystallization from the dike melt With theaim to compare the clinopyroxene cooling dynamics with the thermalregime of the intruding magma we calculated the conductive coolingof the dike through a set of numerical simulations based on an explicitfinite-difference scheme (eg Wohletz et al 1999 and referencestherein) The following conditions were assumed in the calculations(i) an initial temperature of 1150 degC for the dike melt ie themaximum crystallization temperature of the dike melt as inferred bythermometers based on crystalndashliquid pairs and 30 degC for the wallrock (Corsaro and Pompilio 2004 and references therein) (ii) aspecific heat and thermal conductivity of 1200 JkgK and 18 WmKand 1150 JkgK and 3 WmK for the intruding magma and host rockrespectively (Wohletz et al 1999) and (iii) a density of 2750 kgm3

for the trachybasaltic products from Mt Etna volcano (Lange andCarmichael 1987) Latent heat of crystallization and heat convectivetransfer were assumed negligible Results are shown in Fig 13 wheretemperature profiles at 1 10 50 and 100 h are reported for a cross-section at about 1 m from the top of the dike ie the presumedstratigraphic position of the dike samples As expected over a timeinterval of 1ndash100 h the dike outermost part is rapidly cooled to430 degC in contrast the innermost portion keeps a higher temperatureof 1010 degC (Fig 13) Thermal gradient profiles evidence that crystals atthe dike rim experienced a rapid cooling rate of 80 degCh whereasthose at the dike core cooled over a lower rate of 1 degCh These coolingconditions are comparable with those determined through the modelof Mollo et al (2010) based on clinopyroxene compositionalvariations However it is worth stressing that Potuzak and Dingwell(2001) determined the glass transition temperature of Etna trachy-

basalt to be 680 degC Below this temperature the mineral-meltequilibrium is frozen in the crystal growth is halted and no furtherequilibration takes place According to this at the dike core early-formed nuclei re-equilibrate with the residual melt over a coolingtime longer than 100 h by contrast at the dike rim no more changesin mineral compositions occur after a cooling time of 9 h

5 Conclusions

Results from this study show that from dike core-to-rim crystalsshow compositional variations due to more rapid cooling ratesPlagioclase is progressively enriched in An from innermost to theoutermost part of the dike Similarly clinopyroxene components En+CaTs+CaFeTs increase whereas Di+Hd decrease The Usp content intitanomagnetite also systematically decreases from dike core-to-rimSuch compositional variations are due to the nucleation and growth ofthe crystals under variable dynamic conditions Partition coefficientsand thermometers based on the crystalndashliquid exchange reactionindicate that at the dike outer portions early-formed nuclei do not re-equilibrate with themelt due to faster cooling rates as a consequencemineral compositions are shifted towards those of high-temperatureformation We used the clinopyroxene compositional variationsobserved from dike core-to-rim to determine the cooling ratesexperienced by the crystals These cooling rates are comparable tothose estimated by thermal modeling based on an explicit finite-difference scheme

Acknowledgments

Authors are grateful to Professor K Putirka one anonymousreviewer and the editor Professor D B Dindwell for their useful andconstructive suggestions

A Cavallo is acknowledged for assistance during electronmicroprobe analysis This work was supported by Project FIRB MIURldquoDevelopment of innovative technologies for the environmentalprotection from natural eventsrdquo

References

Baginski B Dzierzanowski P Macdonald R Upton BGJ 2009 Complex relation-ships among coexisting pyroxenes the Paleogene Eskdalemuir dyke ScotlandMineral Mag 73 929ndash942

Berndt J Koepke J Holtz F 2005 An experimental investigation of the influence ofwater and oxygen fugacity on differentiation of MORB at 200 MPa J Petrol 46135ndash167

Blundy J Cashman K Humphreys M 2006 Magma heating by decompression-driven crystallization beneath andesite volcanoes Nature 443 76ndash80

Brugger CR Hammer JE 2010 Crystallization kinetics in continuous decompressionexperiments implications for interpreting natural magma ascent processesJ Petrol 51 1941ndash1965

Chistyakova S Latypov R 2009 On the development of internal chemical zonation insmall mafic dykes Geol Mag 147 1ndash12

Coish RA Taylor LA 1979 The effect of cooling rate on texture and pyroxenechemistry in DSDP Leg 34 basalt a microprobe study Earth Planet Sci Lett 42389ndash398

Conte A Perinelli C Trigila R 2006 Cooling kinetics experiments on differentStromboli lavas effects on crystal morphologies and phase compositionJ Volcanol Geoth Res 155 179ndash200

Corsaro RA Pompilio M 2004 Buoyancy-controlled eruption of magmas at Mt EtnaTerra Nova 16 (1) 16ndash22

Corsaro RA Civetta L Di Renzo V Miraglia L 2009 Petrology of lavas from the2004ndash2005 flank eruption of Mt Etna Italy inferences on the dynamics of magmain the shallow plumbing system Bull Volcanol 71 781ndash793

Danyushevsky LV 2001 The effect of small amounts of H2O on crystallisation of mid-ocean ridge and backarc basin magmas J Volcanol Geotherm Res 110 265ndash280

Deer WA Howie RA Zussman J 2001 Framework Silicates Feldspars secondedition The Geological Society London

Del Gaudio P Mollo S Ventura G Iezzi G Taddeucci J Cavallo A 2010 Coolingrate-induced differentiation in anhydrous and hydrous basalts at 500 MPaimplications for the storage and transport of magmas in dikes Chem Geol 270164ndash178

Etzel K Benisek A Dachs E Cemic L 2007 Thermodynamic mixing behavior ofsynthetic Ca-Tschermakndashdiopside pyroxene solid solutions I Volume and heatcapacity of mixing Phys Chem Miner 34 733ndash746

51S Mollo et al Chemical Geology 288 (2011) 39ndash52

Ferlito C Lanzafame G 2010 The role of supercritical fluids in the potassiumenrichment of magmas at Mount Etna volcano (Italy) Lithos 119 642ndash650

Ferlito C Nicotra E 2010 The dyke swarm ofMount Calanna (Etna Italy) an exampleof the uppermost portion of a volcanic plumbing system Bull Volcanol 72 (10)1191ndash1207

Ferlito C Viccaro M Cristofolini R 2008 Volatile-induced magma differentiation inthe plumbing system of Mt Etna volcano (Italy) evidence from glass in tephra ofthe 2001 eruption Bull Volcanol 70 455ndash473

Ferlito C Coltorti M Cristofolini R Giacomoni PP 2009 The contemporaneousemission of low-K and high-K trachybasalts and the role of the NE Rift during the2002 eruptive event Mt Etna Italy Bull Volcanol 71 575ndash587

France L Ildefonse B Koepke J Bech F 2010 A new method to estimate theoxidation state of basaltic series from microprobe analyses J Volcanol GeothermRes 189 340ndash346

Franzini M Leoni L 1972 A full matrix correction in X-ray fluorescence analysis ofrock samples Atti Soc Tosc Sci Nat Mem Ser A 79 7ndash22

Gamble RP Taylor LA 1980 Crystalliquid partitioning in augite effects of coolingrate Earth Plan Sci Lett 47 21ndash33

Gerlach TM 1993 Oxygen buffering of Kilauea volcanic gases and the oxygen fugacityof Kilauea basalt Geochim Cosmochim Acta 57 795ndash814

Ghiorso MS Sack RO 1995 Chemical mass-transfer in magmatic processes 4 Arevised and internally consistent thermodynamic model for the interpolation andextrapolation of liquidusndashsolid equilbria in magmatic systems at elevatedtemperatures and pressures Contrib Mineral Petrol 119 197ndash212

Gibb FGF 1971 Crystalndashliquid relationships in some ultrabasic dykes and theirpetrological significance Contrib Mineral Petrol 30 103ndash118

Gottsmann J Harris AJL Dingwell DB 2004 Thermal history of Hawaiian pahoehoelava crusts at the glass transition implications for flow rheology and emplacementEarth Planet Sci Lett 228 343ndash353

Grove TL Bence AE 1977 Experimental study of pyroxenendashliquid interactionin quartz-normative basalt 15597 Proceedings of Lunar and Planetary ScienceConference 8th pp 1549ndash1579

Grove TL Bence AE 1979 Crystallization kinetics in a multiply saturated basaltmagma an experimental study of Luna 24 ferrobasalt Proceedings of Lunar andPlanetary Science Conference 10th pp 439ndash478

Grove TL Raudsepp M 1978 Effects of kinetics on the crystallization ofquartznormative basalt 15597 an experimental study Proceedings of Lunar andPlanetary Science Conference 9th pp 585ndash599

Grove TL Gerlach DC Sando TW 1982 Origin of calc-alkaline series lavas atMedicine Lake Volcano by fractionation assimilation and mixing Contrib MineralPetrol 80 160ndash182

Hammer JE 2006 Influence of fO2 and cooling rate on the kinetics and energetics ofFe-rich basalt crystallization Earth Planet Sci Lett 248 618ndash637

Hammer JE 2008 Experimental studies of the kinetics and energetics of magmacrystallization In Putirka KD Tepley FJ (Eds) Minerals Inclusions and VolcanicProcesses Rev Mineral Geochem 69 pp 9ndash59

Hammond PA Taylor LA 1982 The ilmenite-titanomagnetite assemblage kineticsof re-equilibration Earth Planet Sci Lett 61 143ndash150

Heap MJ Vinciguerra S Meredith PG 2009 The evolution of elastic moduli withincreasing crack damage during cyclic stressing of a basalt from Mt Etna volcanoTectonophysics 471 153ndash160

Heap MJ Baud P Meredith G Vinciguerra S Bell AF Main IG 2011 Brittle creepin basalt and its application to time-dependent volcano deformation Earth PlanSci Lett 307 71ndash82 doi101016jepsl201104035

Iezzi G Mollo S Ventura G Cavallo A Romano C 2008 Experimental solidificationof anhydrous latitic and trachytic melts at different cooling rates the role ofnucleation kinetics Chem Geol 253 91ndash101

Iezzi G Mollo S Torresi G Ventura G Cavallo A Scarlato P 2011 Experimentalsolidification of an andesitic melt by cooling Chem Geol doi101016jchemgeo201101024

Kirkpatrick RJ 1975 Crystal growth from the melt a review Am Mineral 60 798ndash814Kress VC Carmichael ISE 1991 The compressibility of silicate liquids containing

Fe2O3 and the effect of composition temperature oxygen fugacity and pressure ontheir redox states Contrib Mineral Petrol 108 82ndash92

Lange RA Carmichael ISE 1987 Densities of Na2O-K2O-CaO-MgO-FeO-Fe2O3-Al2O3-TiO2-SiO2 liquids new measurements and derived partial molar propertiesGeochim Cosmochim Acta 53 2195ndash2204

Lange RA Carmichael ISE 1996 The Aurora volcanic field California-Nevadaoxygen fugacity constraints on the development of andesitic magma ContribMineral Petrol 125 167ndash185

Lange RA Frey HM Hector J 2009 A thermodynamic model for the plagioclase-liquid hygrometerthermometer Am Mineral 94 494ndash506

Latypov R Hanski E Lavrenchuk A Huhma H Havela T 2011 Alsquothree-increasemodelrsquo for the origin of the marginal reversal of the Koitelainen layered intrusionFinland J Petrol 52 733ndash764

Lofgren GE Huss GR Wasserburg GJ 2006 An experimental study of traceelementpartitioning between TindashAlndashclinopyroxene and melt equilibrium and kineticeffects including sector zoning Am Mineral 91 1596ndash1606

Longhi J Walker D Hays JF 1976 Fe andMg in plagioclase Proceedings of the LunarScience Conference 7th pp 1281ndash1300

Loomis TP Welber PW 1982 Crystallization processes in the Rocky Hill Sanodioritepluton California an interpretation based on compositional zoning of plagioclaseContrib Mineral Petrol 81 230ndash239

Mahood GA Baker DR 1986 Experimental constraints on depths of fractionation ofmildly alkalic basalts and associated felsic rocks Pantelleria Strait of Sicily ContribMineral Petrol 93 251ndash264

Mathez EA 1984 Influence of degassing on oxidation states of basaltic magmasNature 310 371ndash375

Meacutetrich N Rutherford MJ 1998 Low pressure crystallization paths of H2O-saturatedbasaltic-hawaiitic melts from Mt Etna implications for open-system degassing ofbasaltic volcanoes Geochim Cosmochim Acta 62 1195ndash1205

Mevel C Velde D 1976 Clinopyroxenes in Mesozoic pillow lavas from the FrenchAlps influence of cooling rate on compositional trends Earth Planet Sci Lett 32158ndash164

Mollo S Del Gaudio P Ventura G Iezzi G Scarlato P 2010 Dependence ofclinopyroxene composition on cooling rate in basaltic magmas implications forthermobarometry Lithos 118 302ndash312

Mollo S Putirka K Del Gaudio P Scarlato P in press-a Plagioclase-melt (dis)equilibrium due to cooling dynamics implications for thermometry barometryand hygrometry Lithos doi101016jlithos201102008

Mollo S Vinciguerra S Iezzi G Iarocci A Scarlato P Heap MJ Dingwell DB inpress-b Volcanic edifice weakening via devolatilisation reactions Geophys J Intdoi101111j1365-246X201105097x

Morimoto N 1988 Nomenclature of pyroxenes Mineral Mag 52 535ndash550Muntildeoz M Sagredo J 1974 Clinopyroxenes as geobarometric indicators in mafic and

ultramafic rocks from Canary Islands Contrib Mineral Petrol 44 139ndash147Mysen BO Richet P 2005 Silicate Glasses and Melts Properties and Structure

Elsevier AmsterdamNakada S Motomura Y 1999 Petrology of the 1991ndash1995 eruption at Unzen effusion

pulsation and groundmass crystallization J Volcanol Geotherm Res 89 173ndash196Nakamura M 1995 Continuous mixing of crystal mush and replenished magma in the

ongoing Unzen eruption Geology 23 807ndash810Phinney WC 1992 Partitioning coefficients for iron between plagioclase and basalt as

a function of oxygen fugacity implications for Archean and lunar anorthositesGeochim Cosmochim Acta 56 1885ndash1895

Pietranik A Koepke J Puziewicz J 2006 The study of crystallization and resorptionin plutonic plagioclase implications on evolution of granodiorite magma (GęsiniecGranodiorite Strzelin Crystalline Massif SW Poland) Lithos 86 260ndash280

Potuzak M Dingwell DB 2001 Temperature-dependent thermal expansivitiesof multicomponent natural melts between 993 and 1803 K Chem Geol 22910ndash27

Putirka K 1999 Clinopyroxene+liquid equilibria ContribMineral Petrol 135 151ndash163Putirka K 2005 Igneous thermometers and barometers based on plagioclase+liquid

equilibria test of some existing models and new calibrations Am Mineral 90336ndash346

Putirka KD 2008 Thermometers and barometers for volcanic systems In PutirkaKD Tepley F (Eds) Minerals Inclusions and Volcanic Processes Rev in MineralGeochem 69 pp 61ndash120

Putirka K Johnson M Kinzler R Walker D 1996 Thermobarometry of maficigneous rocks based on clinopyroxenendashliquid equilibria 0ndash30 kbar ContribMineral Petrol 123 92ndash108

Putirka K Ryerson FJ Mikaelian H 2003 New igneous thermobarometers for maficand evolved lava compositions based on clinopyroxene+liquid equilibria AmMineral 88 1542ndash1554

Roeder PL Emslie RF 1970 Olivinendashliquid equilibrium Contrib Mineral Petrol 29275ndash289

Ryabchikov ID Kogarko LN 2006 Magnetite compositions and oxygen fugacities ofthe Khibina magmatic system Lithos 91 35ndash45

Sisson TW Grove TL 1993 Experimental investigations of the role of H2O in calc-alkaline differentiation and subduction zone magmatism Contrib Mineral Petrol113 143ndash166

Smith CB 1983 Pb Sr and Nd isotopic evidence for sources of Southern AfricanCretaceous kimberlites Nature 30 51ndash54

Smith JV BrownWL 1988 Feldspar Minerals 1 Crystal Structures Physical Chemicaland Microtextural Properties 2nd edition Springer-Verlag New York 828 pp

Smith D Lindsley DH 1971 Stable and metastable augite crystallization trends in asingle basalt flow Am Mineral 56 225ndash233

Smith BM Preacutevot M 1977 Variation of the magnetic properties in a basaltic dykewith concentric cooling zones Phys Earth Planet Inter 14 120ndash136

Stormer JC 1983 The effects of recalculation on estimates of temperature and oxygenfugacity from analyses of multicomponent ironndashtitanium oxides Am Mineral 68586ndash594

Sugawara T 2001 Ferric iron partitioning between plagioclase and silicate liquidthermodynamics and petrological applications Contrib Mineral Petrol 141659ndash686

Tanguy JC Condomines M Kieffer G 1997 Evolution of the Mount Etna magmaconstraints on the present feeding system and eruptive mechanism J VolcanolGeotherm Res 75 221ndash250

Tegner C 1997 Iron in plagioclase as a monitor of the differentiation of the Skaergaardintrusion Contrib Mineral Petrol 128 45ndash51

Toplis MJ 2005 The thermodynamics of iron and magnesium partitioning betweenolivine and liquid criteria for assessing and predicting equilibrium in natural andexperimental systems Contrib Mineral Petrol 149 22ndash39

Ujike O 1982 Microprobe mineralogy of plagioclase clinopyroxene and amphibole asrecords of cooling rate in the Shirotori-Hiketa dike swarm northeastern ShikokuJapan Lithos 15 281ndash293

Ussler III W Glazner AF 1989 Phase equilibria along a basalt-rhyolite mixing lineimplications for the origin of calc-alkaline intermediate rocks Contrib MineralPetrol 101 232ndash244

Venezky DY Rutherford MJ 1997 Preeruption conditions and timing of dacitendashandesitemagma mixing in the 22 ka eruption at Mount Rainier J Geophys Res 10220069ndash20086

52 S Mollo et al Chemical Geology 288 (2011) 39ndash52

Viccaro M Ferlito C Cortesogno L Cristofolini R Gaggero L 2006 Magma mixingduring the 2001 event at Mount Etna (Italy) effect on the eruptive dynamicsJ Volcanol Geotherm Res 149 139ndash159

Whittington AG Hofmeister AM Nabelek PI 2009 Temperature-dependentThermal Diffusivity of the Earths Crust and Implications for Magmatism

Wilding MC Webb SL Dingwell DB 1995 Evaluation of a relaxation geospeed-ometer for volcanic glasses Chem Geol 125 137ndash148

Wohletz KH Orsi G Civetta L 1999 Thermal evolution of the Phlegraean magmaticsystem J Volcanol Geotherm Res 91 381ndash414

Zhou W Van der Voo R Peacor DR Zhang Y 2000 Variable Ti-content and grainsize of titanomagnetite as a function of cooling rate in very young MORB EarthPlanet Sci Lett 179 9ndash20