Deciphering temperature pressure andoxygen-activity conditions of chlorite

formation

OLIVIER VIDAL1 dagger P IERRE LANARI2 MANUEL MUNOZ1 FRANCK BOURDELLE3 AND VINCENT DE ANDRADE4

1 Isterre CNRS Universiteacute Grenoble Alpes 1381 Rue de la Piscine BP 53 Grenoble Cedex 09 38041 France2Department of Geological Sciences University of Bern Baltzerstrasse 1 + 3 Bern CH3012 Switzerland

3 LGCgE Universiteacute Lille Bacirct SN5 Villeneuve drsquoAscq 59655 France4 Argonne National Laboratory 9700 South Cass Ave Bldg 438-B007 Lemont IL 60439 USA

(Received 1 November 2015 revised 1 May 2016 Guest editor Jeff Walker)

ABSTRACT The advantages and limits of empirical semi-empirical and thermodynamic methodsdevoted to the estimation of chlorite-formation temperature are discussed briefly The results of semi-empirical and thermodynamic approaches with different assumptions regarding the redox state of iron inchlorite are compared for a large set of natural data covering a range of pressure conditions from a fewhundred bars to 18 kbar and temperature from 100 to 500degC The T-XFe3+ evolution estimated using thethermodynamic approach of Vidal et al (2005) shows a systematic increase in XFe3+ with decreasingtemperature which is compatible with the decrease in aO2 buffered by magnetite- or hematite-chloriteequilibrium This trend as well as the observed increase in vacancies in chlorite with decreasingtemperature is interpreted as the incorporation of Fe3+-sudoite The standard-state properties of this end-member have been derived to reproduce the observed T-aO2-XFe

3+evolutions It can be used to estimateT-aO2-XFe

3 values with a Chl-Qtz-H2O multi-equilibrium approach When combining our results withthose of other studies published recently it appears that thermodynamic approaches and mappingtechniques developed for metamorphic rocks can be used to discuss the conditions of formation of verylow-grade rocks where kinetics is much more sluggish than in metamorphic rocks This requires use ofappropriate analytical tools and techniques with a spatial resolution of a few hundred nanometres

KEYWORDS chlorite oxidation state thermodynamics

In the 1980s and 1990s several empirical calibrationswere proposed to link the tetrahedral Al or vacancycontent of chlorite with temperature (Cathelineau ampNieva 1985KranidiotisampMacLean 1987Cathelineau1988 Hillier amp Velde 1991 Jowett 1991 Zang amp Fyfe1995) Although these approaches are still used widely

for low-grade rocks the lack of theoretical backgroundsand the reliability of the temperatures estimated have beencriticized (eg de Caritat et al 1993 Essene amp Peacor1995 Essene 2009 Bourdelle et al 2013a) and theirapplication to metamorphic rocks remains questionable(Vidal et al 2006) The thermodynamic models ofchlorite proposed by several authors (eg Vidal et al2001 2005 2006 Walshe 1986 Lanari et al 2014)offer an alternative to empirical approaches Vidal et al(2006) and Lanari et al (2014) proposed thermodynamicmodels constrained from high-temperature experimentsand natural data from low- to medium-grade meta-morphic rocks They showed that these models could beused to constrain the temperature of chlorite formation in

E-mail oliviervidalujf-grenoblefrdaggerThis work was originally presented during the sessionlsquoThe many faces of chloritersquo part of the Euroclay 2015conference held in July 2015 in Edinburgh UK

DOI 101180claymin2016051406

copy 2016 The Mineralogical Society

Clay Minerals (2016) 51 615ndash633

quartz-bearing rocks over a large range of temperatureconditions Vidal et al (2006) claimed that the proportionof Fe3+ in chlorite could be estimated from a criterionbased on the convergence of two Chl-Qtz-H2O equilibriathat are compositionally dependant but assumed to bethermodynamically independent The same authors andalso de Andrade et al (2011) Ganne et al (2012) Lanariet al (2012 2013 2014) Yamato et al (2007) Groschet al (2014) andTrincal et al (2015) showed thatmaps ofchlorite crystallization temperature can be computed atthe thin-section scale and their superposition onmicrostructures allows one to discuss the T-deformationhistory of metapelitic samples Application to low-grademetamorphic rocks indicates that temperatures as low as100degC could be calculated which suggests that thethermodynamic approach could potentially be used atvery low-T conditions Finally Inoue et al (2009)Bourdelle et al (2013a) and Lanari et al (2014) useddifferent compilations of natural samples to derive semi-empirical equations that also link the composition ofchlorite with its temperature of formation The maindifference compared to the thermodynamic approach is asimplification of the Gibbsrsquo free energy equation activitymodels and assumptions regarding chlorite Fe3+ contentThese semi-empirical thermometers have the advantageof being easier to use than full thermodynamicmodellingand they have been proved to reproduce successfullytemperatures estimated independently in low-temperatureseries However only Lanari et al (2014) verified theconsistency of the semi-empirical equations with theexisting experimental data and with natural compositionsof chlorite in metamorphic rocks

The purpose of the present study was to give anoverview of the advantages and pitfalls of the differentapproaches mentioned above and to discuss thepotential of thermodynamic approaches developed byVidal et al (2001 2005 2006) and Lanari et al (2014)for metamorphic rocks to very low-grade conditionsParticular emphasis is placed on the role of Fe3+ and itsevolution with temperature pressure and oxygenactivity To this end preliminary thermodynamic dataof a new chlorite Fe3+-bearing end-member (ferri-sudoite) have been constrained

RANGE OF CHLOR ITECOMPOS IT ION RELEVANT END -MEMBER COMPOS IT IONS AND

EQUIL IBR IA

Many studies dealing with the variation of chloritecomposition with intensive parameters and bulk-rockcomposition have already been published (see eg

Curtis et al 1985 Walshe 1986 Wiewioacutera amp Weiss1990 Hillier amp Velde 1991 Vidal et al 2001 Inoueet al 2009 Bourdelle et al 2013a b Lanari et al2014 Bourdelle amp Cathelineau 2015 and referencestherein) and the results will not be repeated extensivelyhere Vidal et al (2001) and Lanari et al (2014)restricted their analysis to chlorites with Si lt 3 apfuwhich are the most common compositions in metape-lites and they compiled a reference set ofsim300 analysesfrom samples covering a wide range of temperature (T )and pressure (P) (Fig 1) In the present study thiscompilation was complemented by 475 analyses fromthe literature and 700 analyses used by Grosch et al(2012) From their original set of analyses Vidal et al(2001) and Lanari et al (2014) showed that thecompositional variability of chlorites result from threemain substitutions Fe-Mg (FM) Tschermak (TK) Si(FeMg) = AlIVAlVI and ditrioctahedral 3(MgFe2+) =2AlVI substitutions between the tritri-octahedral Si-rich end-members clinochlore (Clin) and daphnite(Daph) the tritri-octahedral Al-rich end-members Fe-amesite (Fe-Am) andMg-amesite (Mg-Am) and the ditri-octahedral end-members Fe-sudoite (Fe-Sud) andMg-sudoite (Mg-Sud) (Fig 2) In addition to these threemain substitutions several studies have shown that theincorporation of ferric iron in chlorite can be significantin a low-grade context (see eg Inoue et al 2009 andreferences therein Munoz et al 2013) This incorpor-ation results from the Al3+ = Fe3+ andor di-trioctahedralsubstitution 3(Fe2+Mg2+) =2Fe3+ (Trincal andLanari 2016 this issue) Walshe (1986) suggested that

FIG 1 Pressure and temperature conditions of the samplesused in the present study

616 O Vidal et al

the incorporation of ferric ironmight also result from theexchange Fe2+ + OHminus = Fe3+ + O2minus but to our knowl-edge this exchange has not been confirmed thus far byanalytical data

In the absence of any experimental data on themagnitude of such substitutions with P T and fO2

nothermodynamic properties of an Fe3+-bearing chloriteend-member are available yet in the existing internallyconsistent databases from Holland amp Powell (1998) orBerman (1988) Nevertheless Vidal et al (2005 2006)proposed an estimate of the magnitude of Fe3+ ofnatural chlorites using a criterion based on theconvergence of two independent equilibria that can bewritten with the end-members used to compute Fig 2This point is discussed in more detail in the next section

For any chlorite formed in quartz-bearing rocks andoccurring within the grey area in Fig 2 at least twoequilibria can be written which correspond to theintersection of the tie linesClin orDaph-Sud andAm-Qtz

2 Clinthorn 3 Mg-Sud frac14 4 Mg-Amthorn 7 Qtz

thorn 4 H2O (1)

2 Daphthorn 3 Fe-Sud frac14 4 Fe-Amthorn 7 Qtz

thorn 4 H2O (2)

Similar equilibria can be written for chlorite indiaspore- (Dsp) or corundum- (Co) bearing rocks

Clinthorn Dsp=Co frac14 Mg-AmthornMg-Sudthorn H2O (3)

Daph thorn Dsp=Co frac14 Fe-Am thorn Fe-Sud thorn H2O (4)

For any chlorite composition equilibria 1 and 2 or 3and 4 are achieved simultaneously at the P and Tconditions of equilibrium

In the presence of quartz and pure water thecondition of equilibrium 1 and 2 reads

DGWPT (1)

frac14RT ln (a(Mg-Am)4a(Clin)2a(Mg-Sud)3)(5)

DGWPT (2)

frac14RT ln (a(Fe-Am)4a(Daph)2a(Fe-Sud)3)(6)

where lsquoarsquo and ΔGdegPT stand for the activity of end-members in chlorite and for the variation in the Gibbsfree energy of reaction for pure end-members respect-ively Different combinations of activities can lead tothe same activity ratio (equilibrium constant) inequations 5 and 6 Consequently the same equilibriumconstant and equilibrium temperature can be obtainedwith chlorite of different compositions and differentAlIV or vacancy content crystallizing at the same P-Tconditions in rocks of different bulk compositions

Conversely chlorite with the same AlIV or vacancycontent can eventually form at different temperaturesFor these reasons empirical thermometers based on thesole AlIVor vacancy content cannot be used for wide-ranging rock compositions Thermometers involving aratio of end-member activities or proportions are theonly valid equations when dealing with chlorites fromrocks of contrasted compositions With rising tempera-ture and for a fixed chlorite composition equilibria 1 to4 are shifted from the left-hand side (LHS) containingthe hydrated assemblage to the right-hand side (RHS)containing the dehydrated assemblage and freewater Inorder to maintain equilibrium with increasing tempera-ture the activity of the RHS end-members increases atthe expense of that of the LHS end-members Thealuminous end-members Mg-Am and Fe-Am arelocated in the RHS of equilibria 1 to 4 so that theiractivity ndash and therefore the amount of AlIVand AlVI inchlorite equilibrated with quartz or DspCo ndash isexpected to increase with temperature and pressureSimilarly the amount of vacancy (Mg- and Fe-Sud inthe LHS of equilibria 1 and 2) and Si (Sud Clin andDaph in the LHS) in chlorite equilibrated with quartzdecreases with temperature The same conclusion isdrawn from inspection of the following equilibriawritten with the thermodynamic components used byWalshe (1986)

3 Si4Al4A2O10(OH)8 thorn 5 Si4Mg6O10(OH)8frac14 6 Clinthorn 14 Qtzthorn 8 H2O

(7)

The vacancy-bearing and Si-rich end-members are alsolocated in the LHS of equilibrium 7 which suggests

FIG 2 Composition of the end-members and range ofchlorite composition (grey area) considered in the presentstudy plotted in a SiO2-FeO+MgO-Al2O3 ternarydiagram The intersection of the continuous and dashedlines corresponds to reactions 1 and 3 listed in the text

617Conditions of formation of chlorite

that the amount of vacancies and Si in chloritedecreases with increasing temperature In contrastequilibria 3 and 4 indicate that the amount of vacanciesin chlorite equilibrated with DspCo increases withtemperature

In low-pressure rocks such as those used to calibratethe various AlIV empirical thermometers published todate the effect of pressure is negligible compared tothat of temperature and the variation of AlIV is a goodproxy for temperature This is not true for high-pressure low-temperature chlorite crystallizing atconditions typical of subduction zones In such asetting the effect of pressure on the equilibriumtemperature of equilibria 1 and 2 cannot be furtherignored and high-pressure chlorites can have vacan-cies and Al contents similar to chlorite crystallizing atmuch lower temperature and pressure conditions Thispoint is illustrated in Fig 3 which shows the T and Pdependence of chlorite composition inferred from thenatural data of Lanari et al (2014) At P lt 4 kbar and Tlt 300degC the amount of Al on M1 (amesite content)increases almost linearly with temperature (Fig 3a)which is consistent with the empirical AlIV-Tempiricalequations published to date At higher P and T theamount of amesite in chlorite does not follow the trendobserved at low temperature because it is alsocontrolled by pressure Thus the large scatter ofamesite content vs T in Fig 3a at P gt 6 kbar reflectsthe variability of the TP conditions among thedifferent samples A similar observation is made forthe vacancy-T evolution (Fig 3b) which definesdifferent trends for the low-pressure and medium- tohigh-pressure samples Finally Fig 3c shows that theFeMg ratio of chlorite (considering FeT = Fe2+) doesnot show any consistent relationship with T and P This

ratio is controlled primarily by the bulk-rock compos-ition and the FeMg correction introduced in empiricalAlIV thermometers does not seem to be supportedeither for theoretical reasons or by observations whenconsidering large sets of natural data

MODELL ING THE CONDIT IONS OFCHLOR ITE FORMATION AND FE 3 +

CONTENT

The discussion above offers a theoretical backgroundfor the evolution of AlIV and vacancy contents withtemperature that has been documented by manyauthors in the past It suggests that empirical thermo-meters based on the sole variation of AlIV or vacancycontent cannot be extrapolated outside the range of Pconditions and bulk-rock composition at which theywere calibrated This explains why a myriad ofdifferent equations have been published in the pastthey were derived from different sets of naturalsamples

As an alternative approach to the empirical thermo-meters the temperature condition of chlorite formationcan be estimated from the thermodynamic calculationof equilibria similar to those listed above (1ndash4) Thisrequires knowledge of the standard-state thermo-dynamic properties of the relevant end-members andactivity-composition models Walshe (1986) proposedthe first chlorite model which was able to estimate theformation temperature ferric iron and water contents ofchlorite coexisting with quartz and water at known orassumed pressure The composition of chlorite wasrepresented by six independent thermodynamic com-ponents Si4Mg6O10(OH)8 Clin Daph Si3Al2(Fe

2+)5O10(OH)8 Si4Al4O10(OH)8 and Si3Al2(Fe

2+)4Fe3+

FIG 3 Al on M1 (a) vacancy on M1 (b) and XFe of chlorite (c) as a function of temperature and pressure (after Lanariet al 2014) assuming that FeTotal = Fe2+ AlM1 and ⧠M1 are proportional to the amount of amesite and sudoite end-members respectively (AlM1 = 1 for pure amesite ⧠M1 = 1 for pure sudoite AlM1 =⧠M1 = 0 for amesite-free and

sudoite-free chlorite) The grey area in part b encompasses the vacancy contents reported by Vidal et al (2001)

618 O Vidal et al

O11(OH)7 which can be used to write four chlorite-quartz-water equilibria (two independent) This modelwas not used in the present contribution because thestandard-state and solid-solution thermodynamic proper-ties were not constrained by experimental data and theyare not compatible with the existing internally consistentdatabases Two kinds of thermodynamic models com-patiblewith the existing internally consistent databases ofHolland amp Powell (1998) or Berman (1988) have beenproposed so far they have different philosophies andassumptions

(1) Vidal et al (2001 2006) considered a cationrepartition on two tetrahedral sites T1 (SiSi)2 and T2(SiAl)2 and three octahedral sitesM1 (AlFe2+Mg2+)1M2ndash3 (Mg2+Fe2+Al)4 and M4 (AlFe3+)1 and assumedequipartition of Fe and Mg on M2ndashM3 and M1 Theconditions of chlorite formation in the presence of quartzandH2Oare calculated using four equilibria involving fivechlorite end-members (Clin Daph Fe-Am Mg-Am andMg-Sud)

2 Clinthorn 3 Mg-Sud frac14 4 Mg-Amthorn 7 Qtz

thorn 4 H2O (7)

4 Clinthorn 5 Fe-Am frac14 4 Daphthorn 5 Mg-Am (8)

16 Daphthorn 15 Mg-Sud frac14 20 Fe-Amthorn 6 Clin

thorn 35 Qtzthorn 20 H2O (9)

4 Daphthorn 6 Mg-Sud frac14 5 Fe-Amthorn 3 Mg-Am

thorn 14 Qtzthorn 8 H2O (10)

The thermodynamic properties and non-ideal mixingproperties (Table 1) are believed to be compatible withthe internally consistent thermodynamic database ofBerman (1988) The properties of Clin Mg-Am Mg-Sud and the WAl-Mg Margules parameter wereconstrained by experimental data gathered in theMgO-Al2O3-SiO2-H2O chemical system and thoseof Daph and Fe-Ames and WAl-Fe were constrainedindependently by experimental data obtained in theFeO-Al2O3-SiO2-H2O (FASH) system (Fig 4) Thisapproach leads to contrasted values of WAl-Fe(minus106 kJ at 300degC 5 kbar) and WAl-Mg (+68 kJ at300degC 5 kbar) which seems unlikely in view of thesimilar ionic potential of Fe2+ and Mg2+ Howeverattempts to resolve this problem while maintaining agood fit of the experimental constraints in MASH andin FASH proved unsuccessful For the cation reparti-tion used by Vidal et al (2005 2006) the end-members Clin Daph Mg-Am and Fe-Am areindependent because the site M1 is filled either byMg in the case of Clin or Fe in the case of Daph

Equilibrium 8 corresponds thus to the intracrystallineexchange reaction MgM1(Fe5Mg4)M4 = FeM1(Mg5Fe4)M4 This internal equilibrium is achieved atdefined P and T conditions along a univariant line atfixed Fe3+ values Two of the four equilibria (1 8 9and 10) are independent and the P T XFe3+ conditionsof chlorite + quartz + water equilibrium are defined bytheir intersection in a P-T field This approach allowsone to estimate aminimum value ofXFe3+ = Fe3+FeTotalrequired to obtain convergence of equilibria at assumedpressure In view of the analytical uncertainties andunknown uncertainties of the thermodynamic propertiesof the chlorite end-members and non-ideal solid-solution parameters a certain scatter between equilibria(1 8 9 and 10) is accepted In the present studyconvergence was assumed to be achieved when thetemperature difference between equilibrium 8 ndash themost sensitive to small variations of compositions ndash andthe other three equilibria was lt30degC Equilibriumtemperatures and (XFe3+)Chl were calculated for allchlorite compositions at the pressure conditionsreported in the original publications or at 1 to 2 kbarfor the low-temperature samples Calculation wasconducted starting from XFe3+ = 0 and increasing thevalue by 1 until convergence was achieved It isemphasized that convergence at lower temperature canbe maintained with a further increase of XFe3+ Theestimated values of XFe3+ discussed in the followingare therefore minimum values at maximum tempera-tures The effect of a further increase of XFe3+ onequilibrium temperature is weak at low temperature butit can be very significant at high temperature (seeGrosch et al 2012) Note also that the quantity of Feconsidered as Fe3+ influences the activity of all chloriteend-members Increasing XFe3+ leads to an increase invacancies in structural formulae calculated on a 14anhydrous oxygen basis so that even the activity ofMg-Sud and the location of equilibrium 1 involving Fe-freeend-members are affected

(2) Lanari et al (2014) considered a cation-siterepartition identical to that adopted by Vidal et al(2001 2006) but they followed the formalism ofPowell amp Holland (1999) which assumes that onlythree of the four end-members Mg-Ames Fe-Am Clinand Daph are independent It follows that only one ofequilibria (1 8 9 and 10) is independent Thetemperature of chlorite formation is calculated fromthe location of equilibrium 1 with an assumption thatXFe3+ and the Gibbs free energy of the dependent end-member Fe-Am can be estimated with relationshipsbetween the Mg- and Fe-bearing end-members (seeLanari et al 2014) Unfortunately this approach does

619Conditions of formation of chlorite

not allow us to fit the experimental data of chloritecomposition reported by Parra et al (2005) in theFASH system (Fig 4)

Whatever the thermodynamic model at least onetemperature of chlorite-quartz-water equilibrium can becalculated at fixed pressure with assumptions regardingthe proportion of ferric iron in chlorite This calculationinvolves evaluation of equilibrium constants expressedas a ratio of chlorite end-member activities andtherefore cation proportions It follows that chloritesof different compositions (and different AlIV contents)crystallizing at identical P-T conditions but in rocks of

different compositions can have identical equilibriumconstants of (1) and (8) to (10) Inoue et al (2009)Bourdelle et al (2013a) and Lanari et al (2014)proposed several semi-empirical chlorite thermometersalso involving equilibrium constants calculated as theratio of chlorite end-member activities The maindifference compared to the thermodynamic approachis a simplification of the Gibbs free energy equationactivity models and assumptions on the ferric ironcontent in chlorite These semi-empirical thermometersare easy to use and they are believed to reproducesuccessfully temperatures estimated independently in

TABLE 1 Standard-state properties of chlorite end-members and Margules parameters of the chlorite solid solution (seeBerman 1988 Vidal et al 2006 for the format)

Mg-sudoite (Vidal et al 1992 2006) [Si2]T1[Si Al]T2)IV([A]M1[Mg2 Al2]M2[Al]M4)

VIO10(OH)8ST minus86552730 39050 20510C1 111584 minus85075 minus19777440 2959147000V1 264515406 00 minus018194707 00Ferri-sudoite (present study) [Si2]T1[Si Al]T2)

IV([A]M1[Mg2 Al2]M2plusmn3[Fe3thorn]M4)

VIO10(OH)8ST minus8221273a 40550 20400

minus8230400b

C1 112225 minus85695 minus17431790 2552588000V1 264515406 00 minus018194707 00Daphnite (Vidal et al 2001 2006) [Si2]T1[Si Al]T2)

IV([Fe2thorn]M1[Fe2thorn4 ]M2plusmn3[Al]M4)

VIO10(OH)8ST minus7120845 55940 21600C1 1229233 minus102565 minus12276900 2121510000V1 264515406 00 minus018194707 00Fe-amesite (Vidal et al 2005 2006) [Si2]T1[Al2]T2)

IV([Al]M1[Fe2thorn4 ]M2plusmn3[Al]M4)

VIO10(OH)8ST minus76074600 51480 20900C1 121981 minus100215 minus14439720 2392326000V1 264515406 00 minus018194707 00Clinochlore (Berman 1988) [Si2]T1[Si Al]T2)

IV([Mg]M1[Mg4]M2plusmn3[Al]M4)VIO10(OH)8

ST minus890958984 43515 21147C1 1214284 minus1121713 0000 minus1256253184V1 264515406 00 minus018194707 00Mg-amesite (Vidal et al 2001 2006) [Si2]T1[Al2]T2)

IV([Al]M1[Mg4]M2plusmn3[Al]M4)VIO10(OH)8

ST minus90359005 40320 20520C1 114445 minus83272 minus20058040 2820786000V1 264515406 00 minus018194707 00Margules parameters on site M1 (Vidal et al 2006)WMg-Al minus9400 minus30 minus02WMg- 10000 minus25 09WFe-Al 1200 31 07WFe- 2000 minus15 04WAl- minus10000 minus30 09

ST ΔHdegf (J) Sdeg(JK) Vdeg(Jb) Cp K0 K1 K2 K3 (Cp = K0 + K1Tminus05 + K2Tminus2 + K3Tminus3) V(PT) V1 V2 V3 V4 (V(PT)V(1298) = 1 + V1(Tminus 298) + V2(Tminus 298)2 + V3(Pminus1) + V4(Pminus 1)2)a estimated using equilibria 1 and 21ndash24 b estimated as (ΔHdegf (Mg-sudoite)minus ΔHdegf (corundum)2 + ΔHdegf (hematite)2)The Margules parameters list the WH WS and Wv values (W =WH + T(K)WS + (Pminus 1)Wv)

620 O Vidal et al

low-temperature series (Inoue et al 2009 Bourdelleet al 2013a Lanari et al 2014) and metamorphicrocks (Lanari et al 2014) The weakness of the semi-empirical equations is that their compatibility withexperimental data obtained in MASH and FASHsystems has not always been checked and they arerestricted to quartz-bearing rocks

Finally the main difference between the thermo-dynamic approach of Vidal et al (2005 2006) andthose of Vidal et al (2001) Lanari et al (2014) or thesemi-empirical thermometers of Inoue et al (2009)

Bourdelle et al (2013a) and Lanari et al (2014) is thatthe former leads to the estimation of T-(XFe3+)Chl pairsat given pressure whereas the latter estimate T atassumed or measured XFe3+ and pressure or assuminga negligible role of pressure This last assumption isacceptable at low-pressure conditions but becomesproblematic for metamorphic samples because theamount of Mg-Sud end-member at fixed temperaturehas been shown to increase with pressure (Fig 3b)

A comparison of the approaches of Vidal et al(2006) Lanari et al (2014) and Inoue et al (2009) is

FIG 4 Comparison of measured and calculated Si content of chlorite in the FASH systemwith almandine and fayalite (a)or chloritoid and hercynite (b) Boxes experimental data of Parra et al (2005) continuous lines Si content calculatedwith the thermodynamic properties of Fe-amesite and daphnite and the solid solution model of Vidal et al (2005 2006)dashed lines Si content calculated with the Holland et al (1998) model updated by Coggon amp Holland (2002) which

assumes that the thermodynamic properties of Fe-Am depend on those of Clin Mg-Am and Daph

621Conditions of formation of chlorite

shown in Fig 5 which reports temperatures estimatedfor the set of 1541 analyses used in the present studyThe temperatures calculated with the assumption Fe3+

= 0 (equation 38 of Lanari et al 2014) or Fe3+ gt 0 (allother approaches) are labelled TFe2+ and TFe3+respectively in Fig 5 and in the following Allcalculations were conducted with a MATLAB scriptwith a GUI interface available on request from the firstauthor

The temperature conditions estimated for blueschist-facies metamorphic samples with the approach ofVidal et al (2006) (TFe3+ Vidal) cover a wide range oftemperature from 275 to 550degC (Fig 5a) About onethird of the samples show temperatures of lt350degCwhich is 50ndash100degC lower than the published peaktemperatures derived for the samples studied Usingthe XFe3+ estimated with the Vidal et al (2006)approach the temperatures estimated with the semi-empirical equations of Inoue et al (2009) (TFe3+

Inoue) or Lanari et al (2014) (TFe3+ Lanari) show asimilar evolution but the estimated temperatures are50ndash100degC lower in the case of Inoue et al (2009) andhigher in the case of Lanari et al (2014) The mostimportant deviation is observed between TFe2+ Lanariand the other temperature estimates A significantproportion of chlorites calculated to have crystallized atTFe3+asymp 300degC give much higher TFe2+Lanari (gt500degCopen symbols in Fig 5a) and plot on a different trendThe same observation is made for the Barbertonsamples from Grosch et al (2012) (Fig 5c) In bothcases the maximum TFe2+ Lanari can be as high as900degC but the highest temperatures are obtained forsamples containing very small amounts of vacancies(lt005 pfu open circles below the dashed line in thetemperature-vacancy plot shown in Fig 5a) The Mg-Sud activity and temperatures derived from theseanalyses are very sensitive to small variations incomposition and Fe3+-content and they should not beused for reliable thermometric estimates In contrastFig 5b shows that similar temperatures are estimatedwith all approaches for low-pressure and low-gradesamples

This comparison indicates that similar results areobtained with all methods at T lt 300degC and similartrends are observed at higher temperature for the threemethods allowing the incorporation of Fe3+ althoughwith a significant scatter at T gt 400degC With the XFe3+

values derived using the Vidal et al (2006) approachTFe3+ Vidal and TFe3+ Inoue are quite consistentHowever the highest TFe3+ Inoue estimated forblueschist metamorphic samples are lower whereasTFe3+ Lanari are higher In all approaches the

temperature calculated depends heavily on thesudoite activity which is proportional to the amountof vacancy in chlorite (a(Sud) = 64 (XSi XAl)T2 XM1 (XMg)2M2ndash3 (XAl)2M2ndash3 (XAl)M4))This is illustrated in Fig 5 which shows that theamount of vacancy in chlorite increases with a drop intemperature As noted by Laird (1988) the amount ofvacancy in chlorite structural formulae calculated on acharge-normalized 14 anhydrous-oxygen basisincreases with increasing XFe3+ It is therefore notsurprising that like vacancies XFe3+ increases withreducing temperature This trend is illustrated in Fig 6where the TFe3+Vidal vs XFe3+ values of samples usedto compute Fig 5 have been plotted Figure 6 alsoindicates that larger XFe3+ values are calculated for thehigh-pressure samples than for the low-pressuresamples Nevertheless vacancies are also noted toincreasewith reducing Tunder the assumption that Fe2+

= FeTotal (see Fig 3b and the open circles in Fig 5ab)Therefore the increase in vacancies with reducingtemperature is not just a mathematical artefact resultingfrom the calculation of Fe3+-bearing structural formulaeon a 14-oxygen basis Interestingly even at lowtemperature where XFe3+ calculated with the Vidalet al (2006) approach is at a maximum TFe2+ Lanari issimilar to the three TFe3+ (Fig 5b) This indicates thatthe TFe2+ Lanari equation includes the systematicvariation of XFe3+ with temperature observed inFig 6 which explains why similar TFe2+ and TFe3+

are obtained at low T and P (Fig 5b) for contrastingamounts of vacancies and ferric iron contentsNevertheless it is evident from Fig 5ac that the XFe3+ vs temperature relationship that was used to derive theTFe2+ Lanari equation is not fully consistent with the T-XFe3+ relationship obtained with the thermodynamicapproach of Vidal et al (2006) This is illustrated by thedivergence between TFe2+ Lanari and the three TFe3+

calculated for the low-pressure Barberton samples(Fig 5c) and several HP samples Fig 5a

COMPARISON OF EST IMATED ANDMEASURED (XFE 3 + )CHL

At this stage we can only conclude from thecomparison of TFe2+ and TFe3+ in Fig 5ac thateither some XFe3+ values derived with the Vidal et al(2006) approach are overestimated and the low TFe3+

obtained for some HP-HT samples are underestimatedor that TFe2+Lanari only pertains for fO2

conditionsbuffered by specific assemblages An overestimationof XFe3+ due to errors in the standard-state thermo-dynamic properties and the non-ideal Margules

622 O Vidal et al

FIG 5 Left Plot of temperatures estimated with equations 38 (TFe3+ Lanari) and 40 (TFe2+ Lanari) of Lanari et al(2014) and with the thermometer of Inoue et al (2009) (TFe3+ Inoue) vs temperatures estimated using the approach ofVidal et al (2006) TFe3+ Lanari and TFe3+ Inoue were calculated using the XFe3+ values obtained with the approach ofVidal et al (2006) and TFe2+ Lanari assuming XFe3+ = 0 Right Plot of vacancy content vs estimated temperatures(a) and (b) are high-pressure (gt6 kbar) and low- to medium-pressure (lt6 kbar) respectively (c) shows the resultsobtained for the low-pressure samples of Grosch et al (2012) The unrealistically high TFe2+ Lanari are obtained forchlorites with lt005 vacancy pfu (dashed line in the vacancy-temperature plots) The histograms show the repartition

of temperatures estimated with the different approaches

623Conditions of formation of chlorite

parameters of Vidal et al (2006) is possible butthis should be demonstrated by a systematic compari-son of measured and estimated XFe3+ from naturalsamples covering a wide range of P T and fO2

conditions Such a study is still lacking but a first-order comparison of measured and estimated XFe3+

contents for samples of known temperature ofcrystallization is shown in Fig 7 which was obtainedfor the Rouez Niger and Saint-Martin low-T samplescompiled by Inoue et al (2009) the Glarus sample ofBourdelle et al (2013c) and a blueschist-facies sampleretrogressed into the greeschist facies from Sambagawa(de Andrade et al 2006 Munoz et al 2006 Vidalet al 2006) T-XFe3+ values were calculated at 2 kbarexcept for the Sambagawa sample for which calcula-tions were made at 2 5 and 10 kbar for the low-medium- and high-temperature chlorite groups identi-fied by de Andrade et al (2006) and Vidal et al (2006)respectively

FIG 6 Estimated TFe3+ Vidal and XFe3+ pairs for low- andmedium-pressure samples (open circles) high-pressuresamples (black circles) and the samples from Grosch et al

(2012) (grey squares)

FIG 7 Boxes measured XFe3+ and reference temperatures Symbols Tmaximum-XFe3+minimum pairs calculated to

obtain convergence (plusmn30degC) of equilibria 1 8 9 and 10 with the thermodynamic model of Vidal et al (2006)

624 O Vidal et al

For each series of Inoue et al (2009) between 50and 60 of the chlorite analyses contain Si lt3 apfuand led to a convergence of equations 1 8 9 and 10within 30degC The calculated T-XFe3+ pairs are plottedin Fig 7 and compared with the range of referencetemperature (TRef) estimated independently and XFe3+

measures reported by Inoue et al (2009) A fairagreement of the measured and calculated TFe3+ Vidal-XFe3+ values is observed for the Niger and St-Martinanalyses and a good agreement of the TRef and TFe3+ Vidal was observed for the Rouez samples but thecalculated valueswere smaller than themeasured valuesfor XFe3+ for this samples However the XFe3+ valuesreported in Fig 7 are minimum values required toobtain convergence of equilibria 1 and 8ndash10 within30degC Convergence is maintained at higher valuesof XFe3+ and TFe3+ Vidal for the Rouez samples arestill located in the range of TRef when XFe3+ is setat gt15 (lower limit of the reference values) TheTFe3+ Vidal values calculated for the Glarus chloriteanalyses listed in Bourdelle et al (2013c) are 139degC 203degC and 224degC in fair agreement with thoseestimated independently by Bourdelle et al (2013c)(135degC 170degC and 182degC respectively) or with thesemi-empirical thermometer of Inoue et al (2009)(145degC 196degC and 210degC respectively) For thethree samples studied the XFe3+ values measuredusing XANES (X-ray absorption near edge struc-ture) spectra in the transmission mode of scanningtransmission X-ray microscopy (STXM) were foundto be 223 238 and 279 These values comparefavourably with the estimated ones (32 27 and 28respectively) although the measured values suggestan increase in XFe3+ with temperature which iscontrary the trend inferred from the thermodynamicestimates and evidenced in Figs 6 and 7 Finally therange of XFe3+ estimated for the different generationsof chlorite in the Sambagawa sample (ellipses inFig 7) are slightly higher but still in fair agreementwith those inferred from the XANES measurementsIn conclusion Fig 7 shows that the values estimatedon the basis of the convergence of equilibria (1 8 9and 10) are consistent with the TRef and measuredXFe3+ for various samples covering a range oftemperature from 100degC to 500degC pressure from sim2to 10 kbar and XFe3+ values from 5 to 35 Boththe measured and estimated XFe3+ values plotted inFig 7 seem to confirm the systematic decrease ofXFe3+ with increasing temperature and increase withincreasing pressure that was observed in Fig 6 Thistrend is similar to the decrease in vacancy observedin Fig 5

aO2 -TEMPERATURE EST IMATESAND ADDIT ION OF A FERR I -

SUDOITE END -MEMBER

The general trends inferred from our compilation ofnatural samples indicate that XFe3+ increases withdecreasing temperature and increasing pressureAlthough the model of Vidal et al (2006) involvesno Fe3+-bearing end-member the oxygen activity(aO2) prevailing during the formation of chloritesused to compute Fig 7 can be estimated assuming thatit was buffered by the presence of pure magnetite (Mt)in equilibrium with chlorite and quartz Following thisassumption aO2 for each sample has been calculatedusing equilibria (1 8 9 10) and the followingadditional equilibria which intersect at the same P T(plusmn30degC) (XFe3+)Chl and Log(aO2) (plusmn2) (Fig 8)

15 Mg-Sudthorn 40 Mtthorn 45 Qtzthorn 60 H2O

frac14 6 Clinthorn 24 Daphthorn 20 O2

(11)

6 Mg-Sudthorn 10 Mtthorn 6 Qtzthorn 12 H2O

frac14 3 Mg-Amthorn 6 Daphthorn 5 O2

(12)

15 Mg-Sudthorn 6 Clinthorn 10 Mt

frac14 15 Qtzthorn 15 Mg-Amthorn 6 Daphthorn 5 O2

(13)

6 Daphthorn 4 Clinthorn 5 O2

frac14 20 Qtzthorn 10 Mtthorn 5 Mg-Amthorn 20 H2O(14)

60 Mg-Sudthorn 12 Clinthorn 70 Mtthorn 60 H2O

frac14 45 Mg-Amthorn 42 Daphthorn 35 O2

(15)

4 Qtzthorn 2 Mtthorn Fe-Amthorn 4 H2O

frac14 2 Daphthorn O2

(16)

15 Mg-Sudthorn 6 Daphthorn 10 Mt

frac14 15 Qtzthorn 6 Clinthorn 15 Fe-Amthorn 5 O2

(17)

60 Mg-Sudthorn 70 Mtthorn 60 H2O

frac14 24 Clinthorn 45 Fe-Amthorn 6 Daphthorn 35 O2

(18)

15 Mg-Sudthorn 16 Mtthorn 12 H2O

frac14 3 Qtzthorn 6 Clinthorn 12 Fe-Amthorn 8 O2

(19)

6 Mg-Sudthorn 4 Mt

frac14 6 Qtzthorn 3 Mg-Amthorn 3 Fe-Amthorn 2 O2

(20)

12 Mg-Sudthorn 14 Mtthorn 12 H2O

frac14 6 Mg-Amthorn 3 Fe-Amthorn 6 Daphthorn 7 O2

(21)

5 Mg-Amthorn 4 Mtthorn 8 Qtzthorn 8 H2O

frac14 4 Clinthorn 3 Fe-Amthorn 2 O2

(22)

24 Mg-Sudthorn 3 Mg-Amthorn 28 Mtthorn 24 H2O

frac14 12 Clinthorn 21 Fe-Amthorn 14 O2

(23)

All calculated Log(aO2) except those of the Saint-

625Conditions of formation of chlorite

Martin samples follow the same trend and decreasewith decreasing temperature from simminus27 at 450ndash500degC10 kbar (high-temperature chlorites of Sambagawa) tominus50degC at 125degC and 1 kbar (Niger samples) (Fig 8)The Log(aO2) values of the high-T and -P samples plotin the stability field of magnetite whereas those of thelow-T samples plot in the stability field of hematitewhich is not consistent with the assumption of ChlndashMtndashQtzndashH2O equilibrium The presence of hematite hasbeen reported in the Saint-Martin samples only(Beaufort et al 1992) which plot above the trenddefined by the other samples (upper grey ellipse inFig 8) The relative position of this hematite-bearingsample is consistent with expected higher oxygenactivities required for its formation For the other low-T and apparently hematite-free samples it is possiblethat the calculated Log(aO2) values are too high or thatoxygen was not buffered by an equilibrium ofmagnetitewith chlorite In any case Fig 8 indicates that thesystematic increase in (XFe3+)Chl with decreasingtemperature observed in Figs 6 and 7 is consistentwith a lower aO2 and amount of free oxygen at low-Tthan high-T in systems where aO2 is buffered by thepresence of chlorite and iron oxides To maintain this

equilibrium with lowering temperature in metamorphicrocks where oxygen is immobile oxygen has to beconsumed and one way for that is to produce chloritethat is richer in Fe3+ This is a possible interpretationfor the formation of low-temperature Fe3+-rich chloritecoexisting with high-temperature Fe3+-poor chloritein the same thin section of the Sambagwa sample(Fig 9)

In order to discuss the Log(aO2)-T-XFe3+ evolution

in more detail for rocks where aO2 is not buffered bythe presence of magnetite it is necessary to define anFe3+-bearing chlorite end-member The simultaneousincrease of vacancy and Fe3+ content with decreasingtemperature observed in Figs 5ndash7 is explained by the3(Mg Fe2+) = 2(Fe3+) ditrioctahedral substitution(see also Trincal amp Lanari 2016 this issue) Thisexchange can be modelled with a ferri-sudoite end-member that materializes the relationship betweenXFe3+ and vacancy The standard-state thermo-dynamic properties of ferri-sudoite were constrainedas follows

(1) Because the maximum amount of (Fe3+)Chlcalculated in the previous section never exceeded1 apfu the cation distribution of Vidal et al (2006)

FIG 8 Left Ranges of Log(aO2) conditions calculated for the samples used to compute Fig 7 at the temperature andXFe3+ contents estimated from the Chl-Qtz-H2O equilibrium conditions Dashed lines hematite-magnetite-O2

equilibrium calculated using the thermodynamic properties of Berman (1988) and unit activities of Ht and Mt RightSample calculation at 2 kbar for a Rouez sample using reactions 8ndash20 (upper) or 26ndash40 (lower) The vertical linescorrespond to oxygen- and magnetite-free reactions 1 5 6 and 7 used to compute Figs 5ndash7 which converge at +30degCThe circles in the lower figure represent values calculated with (ΔHdegf ) ferri-Sud constrained with the chlorite-magnetite-quartz-water equilibria 8ndash20 (open circle) or estimated as ΔHdegf (Mg-sudoite) ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2

(filled circle) (see Table 1 and text)

626 O Vidal et al

was kept unchanged and the Fe3+-Al3+ exchange wasassumed to be restricted to the M4 site of chlorite Thechosen formula of ferri-Sud was thus ([Si2]T1[SiAl]T2)

IV([]M1[Mg2Al2]M2-3[Fe3+]M4)

VIO10(OH)8The standard-state thermodynamic properties of allother chlorite end-members the activity model and theMargules parameters were kept unchanged

(2) The third-law entropy of ferri-sudoite (Sdeg =415 JmolK) was estimated according to Holland(1989) with a configurational entropy of 2305 JmolKand a ferromagnetic contribution of 1304 JmolK itsvolume (Vdeg = 204 Jb) was estimated to be 01 Jblower than that of Mg-sudoite (2051 Jb) and the Cp(T ) function was estimated according to Berman amp

FIG 9 Rock-sample mapping of a Sambagawa sample (a) maps of estimated XFe3+ (Vidal et al 2006 Munoz et al2006) and Log(aO2) (estimated assuming that it is buffered by magnetite-chlorite equilibria see text and Fig 8)(b) measured Fe XANES edge-position proportional to Fe redox (Munoz et al 2006) (c) estimated temperature (Vidal

et al 2006 Munoz et al 2006)

627Conditions of formation of chlorite

Brown (1985) The formation enthalpy (ΔHdegf = 82254 plusmn 5 kJmol) was first estimated as the average valuebetween that calculated according to Chermak ampRimstidt (1989) (minus82204 kJmol) and that(minus82304 kJmol) calculted as ΔHdegf (Mg-sudoite) ndashΔHdegf (corundum)2 + ΔHdegf (hematite)2 using thevalues of Vidal et al (1992) for sudoite and Berman(1988) for hematite

(3) In a first step ΔHdegf and Sdeg of ferri-sudoite wereadjusted in order to reproduce the temperaturesreported in Fig 7 with oxygen-free reactions andassuming that magnetite is present in all samples(Chl-Qtz-Mt assemblage) ΔHdegf was allowed to varybetweenminus8220 andminus8230 kJmol and Sdeg within 10 JmolK of the estimated value The addition ofmagnetite and ferri-sudoite to quartz and water leadsto 17 additional equilibria and the following wereused to constrain ΔHdegf and Sdeg of ferri-sudoite

ClinthornMg-Sud frac14 Mg-Amthorn Qtzthorn H2O (1)

4 Clinthorn Daphthorn 10 ferri-Sud

frac14 10 Mg-Amthorn 25 Qtzthorn 30 Mtthorn 20 H2O(24)

7 Daphthorn 70 ferri-Sudthorn 30 Mg-Am

frac14 22 Clinthorn 75 Mg-Sudthorn 35 Mtthorn 40 H2O(25)

28 ferri-Sudthorn 11 Fe-AmthornMg-Am

frac14 6 Daphthorn 30 Mg-Sudthorn 14 Mtthorn 16 H2O(26)

Daphthorn 10 ferri-Sudthorn 10 Mg-Amthorn 10 Qtz

frac14 6 Clinthorn 15 Mg-Sudthorn 5 Mtthorn 16 H2O(27)

The ferri-Sud properties derived with the assumptionof ChlndashQtzndashMt equilibrium are listed in Table 1(referenced a)With these values the oxygen activity atthe P T and (XFe3+)Chl of all equilibria used tocompute Fig 7 can be calculated with the followingoxygen-bearing and magnetite-free equilibria whichconverge with equilibria 1 8 9 and 10 at the same PTand aO2 conditions

16 Qtzthorn 7 Mg-Amthorn 4 Daphthorn 6 Mg-Sud

thorn 5 O2 thorn 12 H2O frac14 20 ferri-Sud(28)

Fe-Amthorn 2 Mg-Amthorn 6 Qtzthorn 4 H2Othorn O2

frac14 4 ferri-Sud(29)

Mg-Amthorn 4 Daphthorn 6 Mg-Sudthorn 2 O2

frac14 3 Fe-Amthorn 8 ferri-Sudthorn 2 Qtz(30)

5 Mg-Amthorn 12 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 8 Fe-Amthorn 28 ferri-Sud

(31)

8 Daphthorn 12 Mg-Sudthorn 3 O2 frac14 7 Fe-Am

thorn 12 ferri-Sudthorn 10 Qtzthorn 4 H2O(32)

6 Clinthorn 4 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Mg-Amthorn 20 ferri-Sudthorn 5 Qtz(33)

4 Clinthorn 20 ferri-Sud frac14 30 Qtz

thorn 15 Mg-Amthorn 4 Daphthorn 5 O2 thorn 20 H2O(34)

90 Mg-Sudthorn 28 Daphthorn 32 Clinthorn 20 H2Othorn 35 O2

frac14 140 ferri-Sudthorn 15 Mg-Am

(35)

15 Qtzthorn 14 Clinthorn 16 Daphthorn 45 Mg-Sud

thorn 20 O2 thorn 20 H2O frac14 80 ferri-Sud(36)

2 Clinthorn FeAmthorn 3 Mg-Sudthorn O2

frac14 2 Mg-Amthorn 4 ferri-Sudthorn Qtz(37)

12 Clinthorn 7 Fe-Amthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 10 Mg-Amthorn 28 ferri-Sud

(38)

5 Qtzthorn 2 Clinthorn 2 Fe-Amthorn 3 Mg-Sud

thorn 2 O2 thorn 4 H2O frac14 8 ferri-Sud(39)

2 Clinthorn 8 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Fe-Amthorn 20 ferri-Sudthorn 5 Qtz(40)

30 Qtzthorn 8 Clinthorn 15 Fe-Amthorn 5 O2 thorn 20 H2O

frac14 8 Daphthorn 20 ferri-Sud

(41)

4 Clinthorn 8 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 3 Fe-Amthorn 28 ferri-Sud

(42)

Because the ferri-sudoite standard-state propertieswere constrained by the chlorite-quartz-magnetite-O2-H2O equilibrium the Log(aO2) calculated withthe equilibria above follow naturally the trend shown inFig 8 and the Log(aO2) values calculated withreactions 28ndash42 of low-temperature samples plot inthe hematite stability field This suggests that theΔGdegPT values of ferri-sudoite derived above with theassumption that aO2 is buffered by the presence ofmagnetite are possibly too high A reduction of ΔHdegf(ferri-sudoite) to 82304 kJmol (=ΔHdegf (Mg-sudoite)ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2 see Table 1)leads to Log(aO2) reduced by 45 Log units (see Fig 8)

628 O Vidal et al

and most calculated values except those of thehematite-bearing St Martin samples plot in themagnetite stability field

A more detailed study is required to refine the ferri-sudoite thermodynamic data However the resultsobtained using the preliminary values listed in Table 1show that this end-member can be used to reproduceconsistently the T-XFe3+ and T- trends obtained fromclassical Chl-Qtz-H2O calculations (Figs 5ndash7) Theincrease in XFe3+ with reducing temperature is consist-ent with the expected reduction in aO2 and the amountof free oxygen that is consumed to produce Fe3+-bearingchlorite Ferric iron is therefore a major component thathas to be taken into account when dealing with low-temperature samples Empirical or semi-empiricalchlorite thermometers which assume that all Fe is inthe ferrous state may include one of the XFe3+-T andaO2-T trends illustrated in Figs 7 and 8 but they cannotaccount for the different XFe3+-T trends that areexpected in the contexts of different aO2 conditions

FUTURE CHALLENGES LOOK INGAT LOW-GRADE ROCKS SUCH ASMETAMORPH IC ROCKS AND THE

DYNAMICS OF CHLOR ITECRYSTALL IZAT ION AT THE

MICROMETR IC TO NANOMETR ICSCALE

T-P-XFe3+ mapping at the thin-section scaleevidence of chlorite composition re-equilibration in metamorphic and low-gradesamples

Although natural samples offer valuable constraintson thermodynamic models there are two maindifficulties in using them to evaluate the reliability ofexisting chlorite models for thermometric purposes

(1) In the case of low-grade samples referencepressures and temperatures are generally known withina reasonable absolute uncertainty but the size ofchlorite grains is often small and good microprobeanalyses are sometimes difficult to obtain Moreoverthe kinetics of reaction are so sluggish that inheritedchlorite grains might remain unaffected by latervariations in temperature which makes thetemperature-composition relationships difficult toconstrain Lastly extrapolation at HT and HP ofempirical or semi-empirical models constrained byLT samples only may be erroneous in particular theeffect of pressure or eventual systematic variations ofXFe3+ cannot be appraised

(2) In the case of metamorphic samplesmicroprobe analyses are easier to obtain from largechlorite grains but chlorites of contrasted compositionthat crystallized at different temperatures and times ofthe metamorphic history are often observed to coexistmetastably in different microstructural sites of the samethin section In such cases the definition of a singlepair of reference temperature and pressure isinappropriate and cannot be restricted to the peak-temperature conditions

In order to overcome these difficulties and to bettervisualize the link between mineral compositiondeformation and P-T conditions at the thin-sectionscale de Andrade et al (2006 2011) Vidal et al(2006) and Lanari et al (2013) developed an approachwhich combines elementary maps of composition witheither a thermodynamic or a semi-empirical estimate ofchlorite crystallization temperature Maps of rawmicroprobe analyses were standardized using good-quality point analyses of minerals present in themapped area The maps of standardized analyseswere transformed into maps of structural formulaewhere each pixel analysed is converted into individualmineral structural formulae Depending on the thermo-metric approach it was then possible to compute mapsof temperature or T-XFe3+ conditions assuming thateach chlorite pixel was in equilibrium with quartz andwater The results reported by Vidal et al (2006)Yamato et al (2007) Ganne et al (2012) Lanari et al(2013 2014) and Scheffer et al (2015) showed a closelink between deformation and mineral reaction at themicroscopic scale and maps of chlorite temperaturecould be used as a basis for interpreting the micro-structures observed in thin sections of low- to middle-grade metapelites In all these studies the metastablepersistence of different generations of chlorite crystal-lizing at different temperatures during the exhumationof metamorphic samples could be used to constrain theT-deformation history Moreover the map of (XFe3+)Chl calculated by Vidal et al (2006) was in goodqualitative agreement with the in situ mapping of thisratio using the X-ray absorption near-edge structure(XANES) techniques (Fig 9) (see also Munoz et al2006) The calculated temperature and XFe3+ mapsobtained on a single sample showed the same trend ofreducing (XFe3+)Chl with increasing temperature asobserved in Figs 6 and 7

Bourdelle et al (2013c) applied a similar approachfor low-temperature samples with chlorite of smallgrain size using transmission electron microscopy-energy dispersive X-ray (TEM-EDX) spectroscopicanalyses of focused-ion beam (FIB) foils instead of

629Conditions of formation of chlorite

microprobe analyses of thin sections (Fig 10) Thoseauthors were able to observe heterogeneities in XFe3+

at the nm scale using STXM-XANES measurementsThis last study was particularly interesting as it showedthat approaches similar to those used for studyingmetamorphic rocks could be applied to low-grade anddiagenetic samples

Looking at diagenetic rocks like metamorphicrocks

The use of equilibrium thermodynamics to studylow-grade rocks has been often criticized becausesluggish kinetics might preclude achievement of globalequilibrium at low temperature and is responsible forthe crystallization of metastable phases with a lowerenergy of nucleation than stable ones Essene amp Peacor(1995) claimed that the coexistence of chemicallyinhomogeneous phases at a very fine scale in low-grade rocks is indicative of disequilibrium and thatmineral reaction is better described by a kinetic ratherthan thermodynamic approach However the coexist-ence of chemically inhomogeneous phases is also acommon observation in metamorphic rocks wherezoned grains and the coexistence for example ofchlorite grains of different compositions indicate thatthermodynamic equilibrium is approached only locallyand at a given time in the metamorphic history In factthe main difference between diagenetic and higher-temperature rocks is the size of the domains in localequilibrium The more sluggish kinetics of reactionand diffusivity at low temperature lead to small grainsand domains in equilibrium but these domains can beidentified when using appropriate analytical tools andprocedures developed recently (Bourdelle et al2013bc)

FIG 10 Quantitative Fe redox nanomapping on a FIB foilof Glarus sample (T lt 300degC Switzerland) calculatedfrom the 7087ndash71025 eV images ratio (STXM-XANES)

(from Bourdelle et al 2013c)

FIG 11 Analysis of an illite-chlorite assemblage (Gulf Coast sample Alamo 1 well 5825 m) (a) SEM image of thepetrographic thin section showing the emplacement of the FIB foil to be cut across the illitendashchlorite interface (b)Bright-field TEM image of the FIB foil extracted from the section with the platinum strap on top of it (cd) Enlargementof part b showing the Al and Mg contents of chlorite (apfu) and their variation between crystal core and crystal rim

630 O Vidal et al

Figure 10 shows that it is possible to demonstrateheterogeneities in XFe3+ contents at the nanometerscale Consistent variations of other major elements arealso observed with TEM technics on FIB foils Anexample is shown in Fig 11 which depicts thevariations of Al and Mg contents in a chlorite grainfrom the Gulf Coast (Bourdelle et al 2013b) In thiscase the FIB foil was cut parallel to the interfacebetween illite and chlorite and a nm-scale investiga-tion of composition was conducted by TEM-EDXanalysis The rim and core compositions showed azoning of chlorite with Mg- and Al-richer rims thancore compositions This variation in composition on ascale of lt1 μm was related to various crystallizationstages during burial Chlorite sampled at differentdepths and various locations of the Gulf Coast areashowed that zoned compositions similar to thatobserved in Fig 11 are frequent (Bourdelle et al2013b) The AlIV in chlorite rims display a Tdependence that is consistent with the km-scale andregional evolution of temperature ie with thediagenetic history of Gulf Coast (burial P-T condi-tions detritalauthigenic compositions etchellip)(Fig 12) This study suggests again that approachesused for metamorphic rocks can be applied to low-grade rocks when using adapted analytical tools able toidentify sub-μm variations of composition

CONCLUS ION

The variability of chlorite composition reflects thevariability of its conditions of formation Numerousefforts have been made over the last 30 years to

translate the composition of chlorite in terms of itstemperature of formation The first empirical equationsbased on the AlIVor vacancy contents are restricted tobulk-rock compositions and pressure conditions atwhich they were calibrated Thermodynamicapproaches and semi-empirical equations based onthe ratio of end-member activities do not present thislimit Among the different thermodynamic models ofchlorite published so far Vidal et al (2006) proposed amodel that can be used to estimate T and (XFe3+)Chlsimultaneously The estimated (XFe3+)Chl values areconsistent with those measured by Moumlssbauer orXANES spectroscopy and for the same (XFe3+)Chlthe temperatures estimated using the method of Vidalet al (2006) are similar to those obtained with thesemi-empirical equations of Inoue et al (2009) andLanari et al (2014) Different T-XFe3+ trends areobserved for different pressure gradients and oxygen-buffering assemblages (Lanari et al 2014) butoverall the systematic increase in XFe3+ with areduction in temperature is compatible with thedecrease in aO2 in a closed system The increase inXFe3+ and chlorite vacancy with decreasing tempera-ture is interpreted as reflecting the incorporation of theFe3+-sudoite component which can be used toestimate T-aO2-XFe

3+ values with a Chl-Qtz-H2Omulti-equilibrium approach

Similar temperatures are calculated for low-tem-perature chlorites (T lt 300degC and measured XFe3+ upto 30) with all approaches assuming that iron inchlorite is either completely ferrous or with estimatedvalues of XFe3+ This indicates that the equations ofsemi-empirical thermometers assuming that all iron inchlorite is ferrous include some inherent Fe2+Fe3+ vstemperature correction However very different tem-peratures can be obtained for higher-temperaturesamples (gt300degC) with approaches involving ferriciron or assuming that all of the iron is ferrous Thissuggests that the Fe2+Fe3+ vs temperature correctionmight be restricted to particular aO2 conditions andpressure gradients More work is necessary to betterunderstand this point and to improve the thermo-dynamic models and semi-empirical thermometersThis requires high-quality and well constrained naturaldata To this end mapping techniques that highlight thelink between the variations in temperature and pressureand small-scale chlorite composition including XFe3+

variations offer much promise When combining theresults of the present study with those of other studiespublished recently it appears that thermodynamicapproaches and mapping techniques developed formetamorphic rocks could be used to study low-grade

FIG 12 Range of crystal-rim AlIV content (white boxes)of Gulf Coast chlorites vs measured temperatures Thegrey area underlines the range of maximum AlIV contentwhich is assumed to represent the closest approach toequilibrium composition (after Bourdelle et al 2013b)

631Conditions of formation of chlorite

rocks where kinetics are much more sluggish Thisrequires use of appropriate analytical tools and techni-ques with a spatial resolution of just a few hundrednanometres which are now becoming available

ACKNOWLEDGMENTS

Constructive reviews by S Hillier F Nieto and ananonymous reviewer are gratefully acknowledged

REFERENCES

Beaufort D Patrier P Meunier A amp Ottaviani MM(1992) Chemical variations in assemblages includingepidote andor chlorite in the fossil hydrothermalsystem of Saint Martin (Lesser Antilles) Journal ofVolcanology and Geothermal Research 51 95ndash114

Berman R (1988) Internally-consistent thermodynamicdata for minerals in the system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2Journal of Petrology 29 445ndash522

Berman R amp Brown T (1985) Heat capacity of minerals inthe system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2 representation estima-tion and high temperature extrapolation Contributionsto Mineralogy and Petrology 89 168ndash183

Bourdelle F amp Cathelineau M (2015) Low-temperaturechlorite geothermometry a graphical representationbased on a TndashR2+ndashSi diagram European Journal ofMineralogy 27 617ndash626

Bourdelle F Parra T Chopin C amp Beyssac O (2013a) Anew chlorite geothermometer for diagenetic to low-grade metamorphic conditions Contributions toMineralogy and Petrology 165 723ndash735

Bourdelle F Parra T Beyssac O Chopin C amp Vidal O(2013b) Clay minerals thermometry a comparativestudy based on high-resolution chemical analyses ofillite and chlorite in sandstones from the Gulf Coast(Texas USA) American Mineralogist 98 914ndash926

Bourdelle F Benzerara K Beyssac O Cosmidis JNeuville DR Brown GE amp Paineau E (2013c)Quantification of the ferric ferrous iron ratio insilicates by scanning transmission X-ray microscopyat the Fe L23 edges Contribution to Mineralogy andPetrology 166 423ndash434

Cathelineau M (1988) Cation site occupancy in chloritesand illites as function of temperature Clay Minerals23 471ndash485

Cathelineau M amp Nieva D (1985) A chlorite solidsolution geothermometer the Los Azufres (Mexico)geothermal system Contributions to Mineralogy andPetrology 91 235ndash244

Chermak JA amp Rimstidt JD (1989) Estimating thethermodynamic properties (ΔGdegf and ΔHdegf) of silicate

minerals at 298K from the sum of polyhedral contribu-tions American Mineralogist 74 1023ndash1031

Coggon R amp Holland TJB (2002) Mixing properties ofphengitic micas and revised garnent-phengite thermo-barometers Journal of Metamorphic Geology 20683ndash696

Curtis CD Hughes CR Whiteman JA ampWhittle CK(1985) Compositional variation within some sedi-mentary chlorites and some comments on their originMineralogical Magazine 49 375ndash386

de Andrade V Vidal O Lewin E OrsquoBrien P amp Agard P(2006) Quantification of electron microprobe com-positional maps of rock thin sections an optimizedmethod and examples Journal of MetamorphicGeology 24 655ndash668

de Andrade V Susini J Salomeacute M Beraldin O RigaultC Heymes T Lewin E amp Vidal O (2011)Submicrometer hyperspectral X-ray imaging of het-erogeneous rocks and geomaterials applications at theFe k-edge Analytical Chemistry 83 4220ndash4227

de Caritat P Hutcheon I amp Walshe J (1993) Chloritegeothermometry a review Clays and Clay Minerals41 219ndash239

Essene EJ (2009) Thermobarometry gone astray Pp101ndash129 in Physics and Chemistry of Earthrsquos Interior(AK Gupta amp S Dasgupta editors) PlatinumJubilee Indian National Science Academy Springer-Verlag Berlin

Essene EJ amp Peacor DR (1995) Clay mineralthermometry ndash a critical perspective Clays and ClayMinerals 43 540ndash553

Ganne J de Andrade V Weinberg RF Vidal O DubacqB Kagambega N Naba S Baratoux L Jessell M ampAllibon J (2012) Modern-style plate subductionpreserved in the Palaeoproterozoic West Africancraton Nature Geoscience 5 60ndash65

Grosch EG Vidal O Abu-Alam T amp McLoughlin N(2012) PTconstraints on the metamorphic evolution ofthe Paleoarchean Kromberg type-section BarbertonGreenstone Belt South-Africa Journal of Petrology53 513ndash545

Grosch E McLoughlin N Lanari P Erambert M ampVidal O (2014) Microscale mapping of alterationconditions and potential biosignatures in basaltic-ultramafic rocks on early Earth and beyondAstrobiology 14 216ndash228

Hillier S amp Velde B (1991) Octahedral occupancy andchemical composition of diagenetic (low-temperature)chlorites Clay Minerals 26 149ndash168

Holland TJB (1989) Dependance of entropy on volumefor silicates and oxide minerals A review and apredictive model American Mineralogist 74 5ndash13

Holland T amp Powell R (1998) An internally consistentthermodynamic data set for phases of petrologicalinterest Journal ofMetamorphic Geology 16 309ndash343

Holland T Baker J amp Powell R (1998)Mixing propertiesand activity composition and relationships of chlorites

632 O Vidal et al

the incorporation of ferric ironmight also result from theexchange Fe2+ + OHminus = Fe3+ + O2minus but to our knowl-edge this exchange has not been confirmed thus far byanalytical data

In the absence of any experimental data on themagnitude of such substitutions with P T and fO2

nothermodynamic properties of an Fe3+-bearing chloriteend-member are available yet in the existing internallyconsistent databases from Holland amp Powell (1998) orBerman (1988) Nevertheless Vidal et al (2005 2006)proposed an estimate of the magnitude of Fe3+ ofnatural chlorites using a criterion based on theconvergence of two independent equilibria that can bewritten with the end-members used to compute Fig 2This point is discussed in more detail in the next section

For any chlorite formed in quartz-bearing rocks andoccurring within the grey area in Fig 2 at least twoequilibria can be written which correspond to theintersection of the tie linesClin orDaph-Sud andAm-Qtz

2 Clinthorn 3 Mg-Sud frac14 4 Mg-Amthorn 7 Qtz

thorn 4 H2O (1)

2 Daphthorn 3 Fe-Sud frac14 4 Fe-Amthorn 7 Qtz

thorn 4 H2O (2)

Similar equilibria can be written for chlorite indiaspore- (Dsp) or corundum- (Co) bearing rocks

Clinthorn Dsp=Co frac14 Mg-AmthornMg-Sudthorn H2O (3)

Daph thorn Dsp=Co frac14 Fe-Am thorn Fe-Sud thorn H2O (4)

For any chlorite composition equilibria 1 and 2 or 3and 4 are achieved simultaneously at the P and Tconditions of equilibrium

In the presence of quartz and pure water thecondition of equilibrium 1 and 2 reads

DGWPT (1)

frac14RT ln (a(Mg-Am)4a(Clin)2a(Mg-Sud)3)(5)

DGWPT (2)

frac14RT ln (a(Fe-Am)4a(Daph)2a(Fe-Sud)3)(6)

where lsquoarsquo and ΔGdegPT stand for the activity of end-members in chlorite and for the variation in the Gibbsfree energy of reaction for pure end-members respect-ively Different combinations of activities can lead tothe same activity ratio (equilibrium constant) inequations 5 and 6 Consequently the same equilibriumconstant and equilibrium temperature can be obtainedwith chlorite of different compositions and differentAlIV or vacancy content crystallizing at the same P-Tconditions in rocks of different bulk compositions

Conversely chlorite with the same AlIV or vacancycontent can eventually form at different temperaturesFor these reasons empirical thermometers based on thesole AlIVor vacancy content cannot be used for wide-ranging rock compositions Thermometers involving aratio of end-member activities or proportions are theonly valid equations when dealing with chlorites fromrocks of contrasted compositions With rising tempera-ture and for a fixed chlorite composition equilibria 1 to4 are shifted from the left-hand side (LHS) containingthe hydrated assemblage to the right-hand side (RHS)containing the dehydrated assemblage and freewater Inorder to maintain equilibrium with increasing tempera-ture the activity of the RHS end-members increases atthe expense of that of the LHS end-members Thealuminous end-members Mg-Am and Fe-Am arelocated in the RHS of equilibria 1 to 4 so that theiractivity ndash and therefore the amount of AlIVand AlVI inchlorite equilibrated with quartz or DspCo ndash isexpected to increase with temperature and pressureSimilarly the amount of vacancy (Mg- and Fe-Sud inthe LHS of equilibria 1 and 2) and Si (Sud Clin andDaph in the LHS) in chlorite equilibrated with quartzdecreases with temperature The same conclusion isdrawn from inspection of the following equilibriawritten with the thermodynamic components used byWalshe (1986)

3 Si4Al4A2O10(OH)8 thorn 5 Si4Mg6O10(OH)8frac14 6 Clinthorn 14 Qtzthorn 8 H2O

(7)

The vacancy-bearing and Si-rich end-members are alsolocated in the LHS of equilibrium 7 which suggests

FIG 2 Composition of the end-members and range ofchlorite composition (grey area) considered in the presentstudy plotted in a SiO2-FeO+MgO-Al2O3 ternarydiagram The intersection of the continuous and dashedlines corresponds to reactions 1 and 3 listed in the text

617Conditions of formation of chlorite

that the amount of vacancies and Si in chloritedecreases with increasing temperature In contrastequilibria 3 and 4 indicate that the amount of vacanciesin chlorite equilibrated with DspCo increases withtemperature

In low-pressure rocks such as those used to calibratethe various AlIV empirical thermometers published todate the effect of pressure is negligible compared tothat of temperature and the variation of AlIV is a goodproxy for temperature This is not true for high-pressure low-temperature chlorite crystallizing atconditions typical of subduction zones In such asetting the effect of pressure on the equilibriumtemperature of equilibria 1 and 2 cannot be furtherignored and high-pressure chlorites can have vacan-cies and Al contents similar to chlorite crystallizing atmuch lower temperature and pressure conditions Thispoint is illustrated in Fig 3 which shows the T and Pdependence of chlorite composition inferred from thenatural data of Lanari et al (2014) At P lt 4 kbar and Tlt 300degC the amount of Al on M1 (amesite content)increases almost linearly with temperature (Fig 3a)which is consistent with the empirical AlIV-Tempiricalequations published to date At higher P and T theamount of amesite in chlorite does not follow the trendobserved at low temperature because it is alsocontrolled by pressure Thus the large scatter ofamesite content vs T in Fig 3a at P gt 6 kbar reflectsthe variability of the TP conditions among thedifferent samples A similar observation is made forthe vacancy-T evolution (Fig 3b) which definesdifferent trends for the low-pressure and medium- tohigh-pressure samples Finally Fig 3c shows that theFeMg ratio of chlorite (considering FeT = Fe2+) doesnot show any consistent relationship with T and P This

ratio is controlled primarily by the bulk-rock compos-ition and the FeMg correction introduced in empiricalAlIV thermometers does not seem to be supportedeither for theoretical reasons or by observations whenconsidering large sets of natural data

MODELL ING THE CONDIT IONS OFCHLOR ITE FORMATION AND FE 3 +

CONTENT

The discussion above offers a theoretical backgroundfor the evolution of AlIV and vacancy contents withtemperature that has been documented by manyauthors in the past It suggests that empirical thermo-meters based on the sole variation of AlIV or vacancycontent cannot be extrapolated outside the range of Pconditions and bulk-rock composition at which theywere calibrated This explains why a myriad ofdifferent equations have been published in the pastthey were derived from different sets of naturalsamples

As an alternative approach to the empirical thermo-meters the temperature condition of chlorite formationcan be estimated from the thermodynamic calculationof equilibria similar to those listed above (1ndash4) Thisrequires knowledge of the standard-state thermo-dynamic properties of the relevant end-members andactivity-composition models Walshe (1986) proposedthe first chlorite model which was able to estimate theformation temperature ferric iron and water contents ofchlorite coexisting with quartz and water at known orassumed pressure The composition of chlorite wasrepresented by six independent thermodynamic com-ponents Si4Mg6O10(OH)8 Clin Daph Si3Al2(Fe

2+)5O10(OH)8 Si4Al4O10(OH)8 and Si3Al2(Fe

2+)4Fe3+

FIG 3 Al on M1 (a) vacancy on M1 (b) and XFe of chlorite (c) as a function of temperature and pressure (after Lanariet al 2014) assuming that FeTotal = Fe2+ AlM1 and ⧠M1 are proportional to the amount of amesite and sudoite end-members respectively (AlM1 = 1 for pure amesite ⧠M1 = 1 for pure sudoite AlM1 =⧠M1 = 0 for amesite-free and

sudoite-free chlorite) The grey area in part b encompasses the vacancy contents reported by Vidal et al (2001)

618 O Vidal et al

O11(OH)7 which can be used to write four chlorite-quartz-water equilibria (two independent) This modelwas not used in the present contribution because thestandard-state and solid-solution thermodynamic proper-ties were not constrained by experimental data and theyare not compatible with the existing internally consistentdatabases Two kinds of thermodynamic models com-patiblewith the existing internally consistent databases ofHolland amp Powell (1998) or Berman (1988) have beenproposed so far they have different philosophies andassumptions

(1) Vidal et al (2001 2006) considered a cationrepartition on two tetrahedral sites T1 (SiSi)2 and T2(SiAl)2 and three octahedral sitesM1 (AlFe2+Mg2+)1M2ndash3 (Mg2+Fe2+Al)4 and M4 (AlFe3+)1 and assumedequipartition of Fe and Mg on M2ndashM3 and M1 Theconditions of chlorite formation in the presence of quartzandH2Oare calculated using four equilibria involving fivechlorite end-members (Clin Daph Fe-Am Mg-Am andMg-Sud)

2 Clinthorn 3 Mg-Sud frac14 4 Mg-Amthorn 7 Qtz

thorn 4 H2O (7)

4 Clinthorn 5 Fe-Am frac14 4 Daphthorn 5 Mg-Am (8)

16 Daphthorn 15 Mg-Sud frac14 20 Fe-Amthorn 6 Clin

thorn 35 Qtzthorn 20 H2O (9)

4 Daphthorn 6 Mg-Sud frac14 5 Fe-Amthorn 3 Mg-Am

thorn 14 Qtzthorn 8 H2O (10)

The thermodynamic properties and non-ideal mixingproperties (Table 1) are believed to be compatible withthe internally consistent thermodynamic database ofBerman (1988) The properties of Clin Mg-Am Mg-Sud and the WAl-Mg Margules parameter wereconstrained by experimental data gathered in theMgO-Al2O3-SiO2-H2O chemical system and thoseof Daph and Fe-Ames and WAl-Fe were constrainedindependently by experimental data obtained in theFeO-Al2O3-SiO2-H2O (FASH) system (Fig 4) Thisapproach leads to contrasted values of WAl-Fe(minus106 kJ at 300degC 5 kbar) and WAl-Mg (+68 kJ at300degC 5 kbar) which seems unlikely in view of thesimilar ionic potential of Fe2+ and Mg2+ Howeverattempts to resolve this problem while maintaining agood fit of the experimental constraints in MASH andin FASH proved unsuccessful For the cation reparti-tion used by Vidal et al (2005 2006) the end-members Clin Daph Mg-Am and Fe-Am areindependent because the site M1 is filled either byMg in the case of Clin or Fe in the case of Daph

Equilibrium 8 corresponds thus to the intracrystallineexchange reaction MgM1(Fe5Mg4)M4 = FeM1(Mg5Fe4)M4 This internal equilibrium is achieved atdefined P and T conditions along a univariant line atfixed Fe3+ values Two of the four equilibria (1 8 9and 10) are independent and the P T XFe3+ conditionsof chlorite + quartz + water equilibrium are defined bytheir intersection in a P-T field This approach allowsone to estimate aminimum value ofXFe3+ = Fe3+FeTotalrequired to obtain convergence of equilibria at assumedpressure In view of the analytical uncertainties andunknown uncertainties of the thermodynamic propertiesof the chlorite end-members and non-ideal solid-solution parameters a certain scatter between equilibria(1 8 9 and 10) is accepted In the present studyconvergence was assumed to be achieved when thetemperature difference between equilibrium 8 ndash themost sensitive to small variations of compositions ndash andthe other three equilibria was lt30degC Equilibriumtemperatures and (XFe3+)Chl were calculated for allchlorite compositions at the pressure conditionsreported in the original publications or at 1 to 2 kbarfor the low-temperature samples Calculation wasconducted starting from XFe3+ = 0 and increasing thevalue by 1 until convergence was achieved It isemphasized that convergence at lower temperature canbe maintained with a further increase of XFe3+ Theestimated values of XFe3+ discussed in the followingare therefore minimum values at maximum tempera-tures The effect of a further increase of XFe3+ onequilibrium temperature is weak at low temperature butit can be very significant at high temperature (seeGrosch et al 2012) Note also that the quantity of Feconsidered as Fe3+ influences the activity of all chloriteend-members Increasing XFe3+ leads to an increase invacancies in structural formulae calculated on a 14anhydrous oxygen basis so that even the activity ofMg-Sud and the location of equilibrium 1 involving Fe-freeend-members are affected

(2) Lanari et al (2014) considered a cation-siterepartition identical to that adopted by Vidal et al(2001 2006) but they followed the formalism ofPowell amp Holland (1999) which assumes that onlythree of the four end-members Mg-Ames Fe-Am Clinand Daph are independent It follows that only one ofequilibria (1 8 9 and 10) is independent Thetemperature of chlorite formation is calculated fromthe location of equilibrium 1 with an assumption thatXFe3+ and the Gibbs free energy of the dependent end-member Fe-Am can be estimated with relationshipsbetween the Mg- and Fe-bearing end-members (seeLanari et al 2014) Unfortunately this approach does

619Conditions of formation of chlorite

not allow us to fit the experimental data of chloritecomposition reported by Parra et al (2005) in theFASH system (Fig 4)

Whatever the thermodynamic model at least onetemperature of chlorite-quartz-water equilibrium can becalculated at fixed pressure with assumptions regardingthe proportion of ferric iron in chlorite This calculationinvolves evaluation of equilibrium constants expressedas a ratio of chlorite end-member activities andtherefore cation proportions It follows that chloritesof different compositions (and different AlIV contents)crystallizing at identical P-T conditions but in rocks of

different compositions can have identical equilibriumconstants of (1) and (8) to (10) Inoue et al (2009)Bourdelle et al (2013a) and Lanari et al (2014)proposed several semi-empirical chlorite thermometersalso involving equilibrium constants calculated as theratio of chlorite end-member activities The maindifference compared to the thermodynamic approachis a simplification of the Gibbs free energy equationactivity models and assumptions on the ferric ironcontent in chlorite These semi-empirical thermometersare easy to use and they are believed to reproducesuccessfully temperatures estimated independently in

TABLE 1 Standard-state properties of chlorite end-members and Margules parameters of the chlorite solid solution (seeBerman 1988 Vidal et al 2006 for the format)

Mg-sudoite (Vidal et al 1992 2006) [Si2]T1[Si Al]T2)IV([A]M1[Mg2 Al2]M2[Al]M4)

VIO10(OH)8ST minus86552730 39050 20510C1 111584 minus85075 minus19777440 2959147000V1 264515406 00 minus018194707 00Ferri-sudoite (present study) [Si2]T1[Si Al]T2)

IV([A]M1[Mg2 Al2]M2plusmn3[Fe3thorn]M4)

VIO10(OH)8ST minus8221273a 40550 20400

minus8230400b

C1 112225 minus85695 minus17431790 2552588000V1 264515406 00 minus018194707 00Daphnite (Vidal et al 2001 2006) [Si2]T1[Si Al]T2)

IV([Fe2thorn]M1[Fe2thorn4 ]M2plusmn3[Al]M4)

VIO10(OH)8ST minus7120845 55940 21600C1 1229233 minus102565 minus12276900 2121510000V1 264515406 00 minus018194707 00Fe-amesite (Vidal et al 2005 2006) [Si2]T1[Al2]T2)

IV([Al]M1[Fe2thorn4 ]M2plusmn3[Al]M4)

VIO10(OH)8ST minus76074600 51480 20900C1 121981 minus100215 minus14439720 2392326000V1 264515406 00 minus018194707 00Clinochlore (Berman 1988) [Si2]T1[Si Al]T2)

IV([Mg]M1[Mg4]M2plusmn3[Al]M4)VIO10(OH)8

ST minus890958984 43515 21147C1 1214284 minus1121713 0000 minus1256253184V1 264515406 00 minus018194707 00Mg-amesite (Vidal et al 2001 2006) [Si2]T1[Al2]T2)

IV([Al]M1[Mg4]M2plusmn3[Al]M4)VIO10(OH)8

ST minus90359005 40320 20520C1 114445 minus83272 minus20058040 2820786000V1 264515406 00 minus018194707 00Margules parameters on site M1 (Vidal et al 2006)WMg-Al minus9400 minus30 minus02WMg- 10000 minus25 09WFe-Al 1200 31 07WFe- 2000 minus15 04WAl- minus10000 minus30 09

ST ΔHdegf (J) Sdeg(JK) Vdeg(Jb) Cp K0 K1 K2 K3 (Cp = K0 + K1Tminus05 + K2Tminus2 + K3Tminus3) V(PT) V1 V2 V3 V4 (V(PT)V(1298) = 1 + V1(Tminus 298) + V2(Tminus 298)2 + V3(Pminus1) + V4(Pminus 1)2)a estimated using equilibria 1 and 21ndash24 b estimated as (ΔHdegf (Mg-sudoite)minus ΔHdegf (corundum)2 + ΔHdegf (hematite)2)The Margules parameters list the WH WS and Wv values (W =WH + T(K)WS + (Pminus 1)Wv)

620 O Vidal et al

low-temperature series (Inoue et al 2009 Bourdelleet al 2013a Lanari et al 2014) and metamorphicrocks (Lanari et al 2014) The weakness of the semi-empirical equations is that their compatibility withexperimental data obtained in MASH and FASHsystems has not always been checked and they arerestricted to quartz-bearing rocks

Finally the main difference between the thermo-dynamic approach of Vidal et al (2005 2006) andthose of Vidal et al (2001) Lanari et al (2014) or thesemi-empirical thermometers of Inoue et al (2009)

Bourdelle et al (2013a) and Lanari et al (2014) is thatthe former leads to the estimation of T-(XFe3+)Chl pairsat given pressure whereas the latter estimate T atassumed or measured XFe3+ and pressure or assuminga negligible role of pressure This last assumption isacceptable at low-pressure conditions but becomesproblematic for metamorphic samples because theamount of Mg-Sud end-member at fixed temperaturehas been shown to increase with pressure (Fig 3b)

A comparison of the approaches of Vidal et al(2006) Lanari et al (2014) and Inoue et al (2009) is

FIG 4 Comparison of measured and calculated Si content of chlorite in the FASH systemwith almandine and fayalite (a)or chloritoid and hercynite (b) Boxes experimental data of Parra et al (2005) continuous lines Si content calculatedwith the thermodynamic properties of Fe-amesite and daphnite and the solid solution model of Vidal et al (2005 2006)dashed lines Si content calculated with the Holland et al (1998) model updated by Coggon amp Holland (2002) which

assumes that the thermodynamic properties of Fe-Am depend on those of Clin Mg-Am and Daph

621Conditions of formation of chlorite

shown in Fig 5 which reports temperatures estimatedfor the set of 1541 analyses used in the present studyThe temperatures calculated with the assumption Fe3+

= 0 (equation 38 of Lanari et al 2014) or Fe3+ gt 0 (allother approaches) are labelled TFe2+ and TFe3+respectively in Fig 5 and in the following Allcalculations were conducted with a MATLAB scriptwith a GUI interface available on request from the firstauthor

The temperature conditions estimated for blueschist-facies metamorphic samples with the approach ofVidal et al (2006) (TFe3+ Vidal) cover a wide range oftemperature from 275 to 550degC (Fig 5a) About onethird of the samples show temperatures of lt350degCwhich is 50ndash100degC lower than the published peaktemperatures derived for the samples studied Usingthe XFe3+ estimated with the Vidal et al (2006)approach the temperatures estimated with the semi-empirical equations of Inoue et al (2009) (TFe3+

Inoue) or Lanari et al (2014) (TFe3+ Lanari) show asimilar evolution but the estimated temperatures are50ndash100degC lower in the case of Inoue et al (2009) andhigher in the case of Lanari et al (2014) The mostimportant deviation is observed between TFe2+ Lanariand the other temperature estimates A significantproportion of chlorites calculated to have crystallized atTFe3+asymp 300degC give much higher TFe2+Lanari (gt500degCopen symbols in Fig 5a) and plot on a different trendThe same observation is made for the Barbertonsamples from Grosch et al (2012) (Fig 5c) In bothcases the maximum TFe2+ Lanari can be as high as900degC but the highest temperatures are obtained forsamples containing very small amounts of vacancies(lt005 pfu open circles below the dashed line in thetemperature-vacancy plot shown in Fig 5a) The Mg-Sud activity and temperatures derived from theseanalyses are very sensitive to small variations incomposition and Fe3+-content and they should not beused for reliable thermometric estimates In contrastFig 5b shows that similar temperatures are estimatedwith all approaches for low-pressure and low-gradesamples

This comparison indicates that similar results areobtained with all methods at T lt 300degC and similartrends are observed at higher temperature for the threemethods allowing the incorporation of Fe3+ althoughwith a significant scatter at T gt 400degC With the XFe3+

values derived using the Vidal et al (2006) approachTFe3+ Vidal and TFe3+ Inoue are quite consistentHowever the highest TFe3+ Inoue estimated forblueschist metamorphic samples are lower whereasTFe3+ Lanari are higher In all approaches the

temperature calculated depends heavily on thesudoite activity which is proportional to the amountof vacancy in chlorite (a(Sud) = 64 (XSi XAl)T2 XM1 (XMg)2M2ndash3 (XAl)2M2ndash3 (XAl)M4))This is illustrated in Fig 5 which shows that theamount of vacancy in chlorite increases with a drop intemperature As noted by Laird (1988) the amount ofvacancy in chlorite structural formulae calculated on acharge-normalized 14 anhydrous-oxygen basisincreases with increasing XFe3+ It is therefore notsurprising that like vacancies XFe3+ increases withreducing temperature This trend is illustrated in Fig 6where the TFe3+Vidal vs XFe3+ values of samples usedto compute Fig 5 have been plotted Figure 6 alsoindicates that larger XFe3+ values are calculated for thehigh-pressure samples than for the low-pressuresamples Nevertheless vacancies are also noted toincreasewith reducing Tunder the assumption that Fe2+

= FeTotal (see Fig 3b and the open circles in Fig 5ab)Therefore the increase in vacancies with reducingtemperature is not just a mathematical artefact resultingfrom the calculation of Fe3+-bearing structural formulaeon a 14-oxygen basis Interestingly even at lowtemperature where XFe3+ calculated with the Vidalet al (2006) approach is at a maximum TFe2+ Lanari issimilar to the three TFe3+ (Fig 5b) This indicates thatthe TFe2+ Lanari equation includes the systematicvariation of XFe3+ with temperature observed inFig 6 which explains why similar TFe2+ and TFe3+

are obtained at low T and P (Fig 5b) for contrastingamounts of vacancies and ferric iron contentsNevertheless it is evident from Fig 5ac that the XFe3+ vs temperature relationship that was used to derive theTFe2+ Lanari equation is not fully consistent with the T-XFe3+ relationship obtained with the thermodynamicapproach of Vidal et al (2006) This is illustrated by thedivergence between TFe2+ Lanari and the three TFe3+

calculated for the low-pressure Barberton samples(Fig 5c) and several HP samples Fig 5a

COMPARISON OF EST IMATED ANDMEASURED (XFE 3 + )CHL

At this stage we can only conclude from thecomparison of TFe2+ and TFe3+ in Fig 5ac thateither some XFe3+ values derived with the Vidal et al(2006) approach are overestimated and the low TFe3+

obtained for some HP-HT samples are underestimatedor that TFe2+Lanari only pertains for fO2

conditionsbuffered by specific assemblages An overestimationof XFe3+ due to errors in the standard-state thermo-dynamic properties and the non-ideal Margules

622 O Vidal et al

FIG 5 Left Plot of temperatures estimated with equations 38 (TFe3+ Lanari) and 40 (TFe2+ Lanari) of Lanari et al(2014) and with the thermometer of Inoue et al (2009) (TFe3+ Inoue) vs temperatures estimated using the approach ofVidal et al (2006) TFe3+ Lanari and TFe3+ Inoue were calculated using the XFe3+ values obtained with the approach ofVidal et al (2006) and TFe2+ Lanari assuming XFe3+ = 0 Right Plot of vacancy content vs estimated temperatures(a) and (b) are high-pressure (gt6 kbar) and low- to medium-pressure (lt6 kbar) respectively (c) shows the resultsobtained for the low-pressure samples of Grosch et al (2012) The unrealistically high TFe2+ Lanari are obtained forchlorites with lt005 vacancy pfu (dashed line in the vacancy-temperature plots) The histograms show the repartition

of temperatures estimated with the different approaches

623Conditions of formation of chlorite

parameters of Vidal et al (2006) is possible butthis should be demonstrated by a systematic compari-son of measured and estimated XFe3+ from naturalsamples covering a wide range of P T and fO2

conditions Such a study is still lacking but a first-order comparison of measured and estimated XFe3+

contents for samples of known temperature ofcrystallization is shown in Fig 7 which was obtainedfor the Rouez Niger and Saint-Martin low-T samplescompiled by Inoue et al (2009) the Glarus sample ofBourdelle et al (2013c) and a blueschist-facies sampleretrogressed into the greeschist facies from Sambagawa(de Andrade et al 2006 Munoz et al 2006 Vidalet al 2006) T-XFe3+ values were calculated at 2 kbarexcept for the Sambagawa sample for which calcula-tions were made at 2 5 and 10 kbar for the low-medium- and high-temperature chlorite groups identi-fied by de Andrade et al (2006) and Vidal et al (2006)respectively

FIG 6 Estimated TFe3+ Vidal and XFe3+ pairs for low- andmedium-pressure samples (open circles) high-pressuresamples (black circles) and the samples from Grosch et al

(2012) (grey squares)

FIG 7 Boxes measured XFe3+ and reference temperatures Symbols Tmaximum-XFe3+minimum pairs calculated to

obtain convergence (plusmn30degC) of equilibria 1 8 9 and 10 with the thermodynamic model of Vidal et al (2006)

624 O Vidal et al

For each series of Inoue et al (2009) between 50and 60 of the chlorite analyses contain Si lt3 apfuand led to a convergence of equations 1 8 9 and 10within 30degC The calculated T-XFe3+ pairs are plottedin Fig 7 and compared with the range of referencetemperature (TRef) estimated independently and XFe3+

measures reported by Inoue et al (2009) A fairagreement of the measured and calculated TFe3+ Vidal-XFe3+ values is observed for the Niger and St-Martinanalyses and a good agreement of the TRef and TFe3+ Vidal was observed for the Rouez samples but thecalculated valueswere smaller than themeasured valuesfor XFe3+ for this samples However the XFe3+ valuesreported in Fig 7 are minimum values required toobtain convergence of equilibria 1 and 8ndash10 within30degC Convergence is maintained at higher valuesof XFe3+ and TFe3+ Vidal for the Rouez samples arestill located in the range of TRef when XFe3+ is setat gt15 (lower limit of the reference values) TheTFe3+ Vidal values calculated for the Glarus chloriteanalyses listed in Bourdelle et al (2013c) are 139degC 203degC and 224degC in fair agreement with thoseestimated independently by Bourdelle et al (2013c)(135degC 170degC and 182degC respectively) or with thesemi-empirical thermometer of Inoue et al (2009)(145degC 196degC and 210degC respectively) For thethree samples studied the XFe3+ values measuredusing XANES (X-ray absorption near edge struc-ture) spectra in the transmission mode of scanningtransmission X-ray microscopy (STXM) were foundto be 223 238 and 279 These values comparefavourably with the estimated ones (32 27 and 28respectively) although the measured values suggestan increase in XFe3+ with temperature which iscontrary the trend inferred from the thermodynamicestimates and evidenced in Figs 6 and 7 Finally therange of XFe3+ estimated for the different generationsof chlorite in the Sambagawa sample (ellipses inFig 7) are slightly higher but still in fair agreementwith those inferred from the XANES measurementsIn conclusion Fig 7 shows that the values estimatedon the basis of the convergence of equilibria (1 8 9and 10) are consistent with the TRef and measuredXFe3+ for various samples covering a range oftemperature from 100degC to 500degC pressure from sim2to 10 kbar and XFe3+ values from 5 to 35 Boththe measured and estimated XFe3+ values plotted inFig 7 seem to confirm the systematic decrease ofXFe3+ with increasing temperature and increase withincreasing pressure that was observed in Fig 6 Thistrend is similar to the decrease in vacancy observedin Fig 5

aO2 -TEMPERATURE EST IMATESAND ADDIT ION OF A FERR I -

SUDOITE END -MEMBER

The general trends inferred from our compilation ofnatural samples indicate that XFe3+ increases withdecreasing temperature and increasing pressureAlthough the model of Vidal et al (2006) involvesno Fe3+-bearing end-member the oxygen activity(aO2) prevailing during the formation of chloritesused to compute Fig 7 can be estimated assuming thatit was buffered by the presence of pure magnetite (Mt)in equilibrium with chlorite and quartz Following thisassumption aO2 for each sample has been calculatedusing equilibria (1 8 9 10) and the followingadditional equilibria which intersect at the same P T(plusmn30degC) (XFe3+)Chl and Log(aO2) (plusmn2) (Fig 8)

15 Mg-Sudthorn 40 Mtthorn 45 Qtzthorn 60 H2O

frac14 6 Clinthorn 24 Daphthorn 20 O2

(11)

6 Mg-Sudthorn 10 Mtthorn 6 Qtzthorn 12 H2O

frac14 3 Mg-Amthorn 6 Daphthorn 5 O2

(12)

15 Mg-Sudthorn 6 Clinthorn 10 Mt

frac14 15 Qtzthorn 15 Mg-Amthorn 6 Daphthorn 5 O2

(13)

6 Daphthorn 4 Clinthorn 5 O2

frac14 20 Qtzthorn 10 Mtthorn 5 Mg-Amthorn 20 H2O(14)

60 Mg-Sudthorn 12 Clinthorn 70 Mtthorn 60 H2O

frac14 45 Mg-Amthorn 42 Daphthorn 35 O2

(15)

4 Qtzthorn 2 Mtthorn Fe-Amthorn 4 H2O

frac14 2 Daphthorn O2

(16)

15 Mg-Sudthorn 6 Daphthorn 10 Mt

frac14 15 Qtzthorn 6 Clinthorn 15 Fe-Amthorn 5 O2

(17)

60 Mg-Sudthorn 70 Mtthorn 60 H2O

frac14 24 Clinthorn 45 Fe-Amthorn 6 Daphthorn 35 O2

(18)

15 Mg-Sudthorn 16 Mtthorn 12 H2O

frac14 3 Qtzthorn 6 Clinthorn 12 Fe-Amthorn 8 O2

(19)

6 Mg-Sudthorn 4 Mt

frac14 6 Qtzthorn 3 Mg-Amthorn 3 Fe-Amthorn 2 O2

(20)

12 Mg-Sudthorn 14 Mtthorn 12 H2O

frac14 6 Mg-Amthorn 3 Fe-Amthorn 6 Daphthorn 7 O2

(21)

5 Mg-Amthorn 4 Mtthorn 8 Qtzthorn 8 H2O

frac14 4 Clinthorn 3 Fe-Amthorn 2 O2

(22)

24 Mg-Sudthorn 3 Mg-Amthorn 28 Mtthorn 24 H2O

frac14 12 Clinthorn 21 Fe-Amthorn 14 O2

(23)

All calculated Log(aO2) except those of the Saint-

625Conditions of formation of chlorite

Martin samples follow the same trend and decreasewith decreasing temperature from simminus27 at 450ndash500degC10 kbar (high-temperature chlorites of Sambagawa) tominus50degC at 125degC and 1 kbar (Niger samples) (Fig 8)The Log(aO2) values of the high-T and -P samples plotin the stability field of magnetite whereas those of thelow-T samples plot in the stability field of hematitewhich is not consistent with the assumption of ChlndashMtndashQtzndashH2O equilibrium The presence of hematite hasbeen reported in the Saint-Martin samples only(Beaufort et al 1992) which plot above the trenddefined by the other samples (upper grey ellipse inFig 8) The relative position of this hematite-bearingsample is consistent with expected higher oxygenactivities required for its formation For the other low-T and apparently hematite-free samples it is possiblethat the calculated Log(aO2) values are too high or thatoxygen was not buffered by an equilibrium ofmagnetitewith chlorite In any case Fig 8 indicates that thesystematic increase in (XFe3+)Chl with decreasingtemperature observed in Figs 6 and 7 is consistentwith a lower aO2 and amount of free oxygen at low-Tthan high-T in systems where aO2 is buffered by thepresence of chlorite and iron oxides To maintain this

equilibrium with lowering temperature in metamorphicrocks where oxygen is immobile oxygen has to beconsumed and one way for that is to produce chloritethat is richer in Fe3+ This is a possible interpretationfor the formation of low-temperature Fe3+-rich chloritecoexisting with high-temperature Fe3+-poor chloritein the same thin section of the Sambagwa sample(Fig 9)

In order to discuss the Log(aO2)-T-XFe3+ evolution

in more detail for rocks where aO2 is not buffered bythe presence of magnetite it is necessary to define anFe3+-bearing chlorite end-member The simultaneousincrease of vacancy and Fe3+ content with decreasingtemperature observed in Figs 5ndash7 is explained by the3(Mg Fe2+) = 2(Fe3+) ditrioctahedral substitution(see also Trincal amp Lanari 2016 this issue) Thisexchange can be modelled with a ferri-sudoite end-member that materializes the relationship betweenXFe3+ and vacancy The standard-state thermo-dynamic properties of ferri-sudoite were constrainedas follows

(1) Because the maximum amount of (Fe3+)Chlcalculated in the previous section never exceeded1 apfu the cation distribution of Vidal et al (2006)

FIG 8 Left Ranges of Log(aO2) conditions calculated for the samples used to compute Fig 7 at the temperature andXFe3+ contents estimated from the Chl-Qtz-H2O equilibrium conditions Dashed lines hematite-magnetite-O2

equilibrium calculated using the thermodynamic properties of Berman (1988) and unit activities of Ht and Mt RightSample calculation at 2 kbar for a Rouez sample using reactions 8ndash20 (upper) or 26ndash40 (lower) The vertical linescorrespond to oxygen- and magnetite-free reactions 1 5 6 and 7 used to compute Figs 5ndash7 which converge at +30degCThe circles in the lower figure represent values calculated with (ΔHdegf ) ferri-Sud constrained with the chlorite-magnetite-quartz-water equilibria 8ndash20 (open circle) or estimated as ΔHdegf (Mg-sudoite) ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2

(filled circle) (see Table 1 and text)

626 O Vidal et al

was kept unchanged and the Fe3+-Al3+ exchange wasassumed to be restricted to the M4 site of chlorite Thechosen formula of ferri-Sud was thus ([Si2]T1[SiAl]T2)

IV([]M1[Mg2Al2]M2-3[Fe3+]M4)

VIO10(OH)8The standard-state thermodynamic properties of allother chlorite end-members the activity model and theMargules parameters were kept unchanged

(2) The third-law entropy of ferri-sudoite (Sdeg =415 JmolK) was estimated according to Holland(1989) with a configurational entropy of 2305 JmolKand a ferromagnetic contribution of 1304 JmolK itsvolume (Vdeg = 204 Jb) was estimated to be 01 Jblower than that of Mg-sudoite (2051 Jb) and the Cp(T ) function was estimated according to Berman amp

FIG 9 Rock-sample mapping of a Sambagawa sample (a) maps of estimated XFe3+ (Vidal et al 2006 Munoz et al2006) and Log(aO2) (estimated assuming that it is buffered by magnetite-chlorite equilibria see text and Fig 8)(b) measured Fe XANES edge-position proportional to Fe redox (Munoz et al 2006) (c) estimated temperature (Vidal

et al 2006 Munoz et al 2006)

627Conditions of formation of chlorite

Brown (1985) The formation enthalpy (ΔHdegf = 82254 plusmn 5 kJmol) was first estimated as the average valuebetween that calculated according to Chermak ampRimstidt (1989) (minus82204 kJmol) and that(minus82304 kJmol) calculted as ΔHdegf (Mg-sudoite) ndashΔHdegf (corundum)2 + ΔHdegf (hematite)2 using thevalues of Vidal et al (1992) for sudoite and Berman(1988) for hematite

(3) In a first step ΔHdegf and Sdeg of ferri-sudoite wereadjusted in order to reproduce the temperaturesreported in Fig 7 with oxygen-free reactions andassuming that magnetite is present in all samples(Chl-Qtz-Mt assemblage) ΔHdegf was allowed to varybetweenminus8220 andminus8230 kJmol and Sdeg within 10 JmolK of the estimated value The addition ofmagnetite and ferri-sudoite to quartz and water leadsto 17 additional equilibria and the following wereused to constrain ΔHdegf and Sdeg of ferri-sudoite

ClinthornMg-Sud frac14 Mg-Amthorn Qtzthorn H2O (1)

4 Clinthorn Daphthorn 10 ferri-Sud

frac14 10 Mg-Amthorn 25 Qtzthorn 30 Mtthorn 20 H2O(24)

7 Daphthorn 70 ferri-Sudthorn 30 Mg-Am

frac14 22 Clinthorn 75 Mg-Sudthorn 35 Mtthorn 40 H2O(25)

28 ferri-Sudthorn 11 Fe-AmthornMg-Am

frac14 6 Daphthorn 30 Mg-Sudthorn 14 Mtthorn 16 H2O(26)

Daphthorn 10 ferri-Sudthorn 10 Mg-Amthorn 10 Qtz

frac14 6 Clinthorn 15 Mg-Sudthorn 5 Mtthorn 16 H2O(27)

The ferri-Sud properties derived with the assumptionof ChlndashQtzndashMt equilibrium are listed in Table 1(referenced a)With these values the oxygen activity atthe P T and (XFe3+)Chl of all equilibria used tocompute Fig 7 can be calculated with the followingoxygen-bearing and magnetite-free equilibria whichconverge with equilibria 1 8 9 and 10 at the same PTand aO2 conditions

16 Qtzthorn 7 Mg-Amthorn 4 Daphthorn 6 Mg-Sud

thorn 5 O2 thorn 12 H2O frac14 20 ferri-Sud(28)

Fe-Amthorn 2 Mg-Amthorn 6 Qtzthorn 4 H2Othorn O2

frac14 4 ferri-Sud(29)

Mg-Amthorn 4 Daphthorn 6 Mg-Sudthorn 2 O2

frac14 3 Fe-Amthorn 8 ferri-Sudthorn 2 Qtz(30)

5 Mg-Amthorn 12 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 8 Fe-Amthorn 28 ferri-Sud

(31)

8 Daphthorn 12 Mg-Sudthorn 3 O2 frac14 7 Fe-Am

thorn 12 ferri-Sudthorn 10 Qtzthorn 4 H2O(32)

6 Clinthorn 4 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Mg-Amthorn 20 ferri-Sudthorn 5 Qtz(33)

4 Clinthorn 20 ferri-Sud frac14 30 Qtz

thorn 15 Mg-Amthorn 4 Daphthorn 5 O2 thorn 20 H2O(34)

90 Mg-Sudthorn 28 Daphthorn 32 Clinthorn 20 H2Othorn 35 O2

frac14 140 ferri-Sudthorn 15 Mg-Am

(35)

15 Qtzthorn 14 Clinthorn 16 Daphthorn 45 Mg-Sud

thorn 20 O2 thorn 20 H2O frac14 80 ferri-Sud(36)

2 Clinthorn FeAmthorn 3 Mg-Sudthorn O2

frac14 2 Mg-Amthorn 4 ferri-Sudthorn Qtz(37)

12 Clinthorn 7 Fe-Amthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 10 Mg-Amthorn 28 ferri-Sud

(38)

5 Qtzthorn 2 Clinthorn 2 Fe-Amthorn 3 Mg-Sud

thorn 2 O2 thorn 4 H2O frac14 8 ferri-Sud(39)

2 Clinthorn 8 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Fe-Amthorn 20 ferri-Sudthorn 5 Qtz(40)

30 Qtzthorn 8 Clinthorn 15 Fe-Amthorn 5 O2 thorn 20 H2O

frac14 8 Daphthorn 20 ferri-Sud

(41)

4 Clinthorn 8 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 3 Fe-Amthorn 28 ferri-Sud

(42)

Because the ferri-sudoite standard-state propertieswere constrained by the chlorite-quartz-magnetite-O2-H2O equilibrium the Log(aO2) calculated withthe equilibria above follow naturally the trend shown inFig 8 and the Log(aO2) values calculated withreactions 28ndash42 of low-temperature samples plot inthe hematite stability field This suggests that theΔGdegPT values of ferri-sudoite derived above with theassumption that aO2 is buffered by the presence ofmagnetite are possibly too high A reduction of ΔHdegf(ferri-sudoite) to 82304 kJmol (=ΔHdegf (Mg-sudoite)ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2 see Table 1)leads to Log(aO2) reduced by 45 Log units (see Fig 8)

628 O Vidal et al

and most calculated values except those of thehematite-bearing St Martin samples plot in themagnetite stability field

A more detailed study is required to refine the ferri-sudoite thermodynamic data However the resultsobtained using the preliminary values listed in Table 1show that this end-member can be used to reproduceconsistently the T-XFe3+ and T- trends obtained fromclassical Chl-Qtz-H2O calculations (Figs 5ndash7) Theincrease in XFe3+ with reducing temperature is consist-ent with the expected reduction in aO2 and the amountof free oxygen that is consumed to produce Fe3+-bearingchlorite Ferric iron is therefore a major component thathas to be taken into account when dealing with low-temperature samples Empirical or semi-empiricalchlorite thermometers which assume that all Fe is inthe ferrous state may include one of the XFe3+-T andaO2-T trends illustrated in Figs 7 and 8 but they cannotaccount for the different XFe3+-T trends that areexpected in the contexts of different aO2 conditions

FUTURE CHALLENGES LOOK INGAT LOW-GRADE ROCKS SUCH ASMETAMORPH IC ROCKS AND THE

DYNAMICS OF CHLOR ITECRYSTALL IZAT ION AT THE

MICROMETR IC TO NANOMETR ICSCALE

T-P-XFe3+ mapping at the thin-section scaleevidence of chlorite composition re-equilibration in metamorphic and low-gradesamples

Although natural samples offer valuable constraintson thermodynamic models there are two maindifficulties in using them to evaluate the reliability ofexisting chlorite models for thermometric purposes

(1) In the case of low-grade samples referencepressures and temperatures are generally known withina reasonable absolute uncertainty but the size ofchlorite grains is often small and good microprobeanalyses are sometimes difficult to obtain Moreoverthe kinetics of reaction are so sluggish that inheritedchlorite grains might remain unaffected by latervariations in temperature which makes thetemperature-composition relationships difficult toconstrain Lastly extrapolation at HT and HP ofempirical or semi-empirical models constrained byLT samples only may be erroneous in particular theeffect of pressure or eventual systematic variations ofXFe3+ cannot be appraised

(2) In the case of metamorphic samplesmicroprobe analyses are easier to obtain from largechlorite grains but chlorites of contrasted compositionthat crystallized at different temperatures and times ofthe metamorphic history are often observed to coexistmetastably in different microstructural sites of the samethin section In such cases the definition of a singlepair of reference temperature and pressure isinappropriate and cannot be restricted to the peak-temperature conditions

In order to overcome these difficulties and to bettervisualize the link between mineral compositiondeformation and P-T conditions at the thin-sectionscale de Andrade et al (2006 2011) Vidal et al(2006) and Lanari et al (2013) developed an approachwhich combines elementary maps of composition witheither a thermodynamic or a semi-empirical estimate ofchlorite crystallization temperature Maps of rawmicroprobe analyses were standardized using good-quality point analyses of minerals present in themapped area The maps of standardized analyseswere transformed into maps of structural formulaewhere each pixel analysed is converted into individualmineral structural formulae Depending on the thermo-metric approach it was then possible to compute mapsof temperature or T-XFe3+ conditions assuming thateach chlorite pixel was in equilibrium with quartz andwater The results reported by Vidal et al (2006)Yamato et al (2007) Ganne et al (2012) Lanari et al(2013 2014) and Scheffer et al (2015) showed a closelink between deformation and mineral reaction at themicroscopic scale and maps of chlorite temperaturecould be used as a basis for interpreting the micro-structures observed in thin sections of low- to middle-grade metapelites In all these studies the metastablepersistence of different generations of chlorite crystal-lizing at different temperatures during the exhumationof metamorphic samples could be used to constrain theT-deformation history Moreover the map of (XFe3+)Chl calculated by Vidal et al (2006) was in goodqualitative agreement with the in situ mapping of thisratio using the X-ray absorption near-edge structure(XANES) techniques (Fig 9) (see also Munoz et al2006) The calculated temperature and XFe3+ mapsobtained on a single sample showed the same trend ofreducing (XFe3+)Chl with increasing temperature asobserved in Figs 6 and 7

Bourdelle et al (2013c) applied a similar approachfor low-temperature samples with chlorite of smallgrain size using transmission electron microscopy-energy dispersive X-ray (TEM-EDX) spectroscopicanalyses of focused-ion beam (FIB) foils instead of

629Conditions of formation of chlorite

microprobe analyses of thin sections (Fig 10) Thoseauthors were able to observe heterogeneities in XFe3+

at the nm scale using STXM-XANES measurementsThis last study was particularly interesting as it showedthat approaches similar to those used for studyingmetamorphic rocks could be applied to low-grade anddiagenetic samples

Looking at diagenetic rocks like metamorphicrocks

The use of equilibrium thermodynamics to studylow-grade rocks has been often criticized becausesluggish kinetics might preclude achievement of globalequilibrium at low temperature and is responsible forthe crystallization of metastable phases with a lowerenergy of nucleation than stable ones Essene amp Peacor(1995) claimed that the coexistence of chemicallyinhomogeneous phases at a very fine scale in low-grade rocks is indicative of disequilibrium and thatmineral reaction is better described by a kinetic ratherthan thermodynamic approach However the coexist-ence of chemically inhomogeneous phases is also acommon observation in metamorphic rocks wherezoned grains and the coexistence for example ofchlorite grains of different compositions indicate thatthermodynamic equilibrium is approached only locallyand at a given time in the metamorphic history In factthe main difference between diagenetic and higher-temperature rocks is the size of the domains in localequilibrium The more sluggish kinetics of reactionand diffusivity at low temperature lead to small grainsand domains in equilibrium but these domains can beidentified when using appropriate analytical tools andprocedures developed recently (Bourdelle et al2013bc)

FIG 10 Quantitative Fe redox nanomapping on a FIB foilof Glarus sample (T lt 300degC Switzerland) calculatedfrom the 7087ndash71025 eV images ratio (STXM-XANES)

(from Bourdelle et al 2013c)

FIG 11 Analysis of an illite-chlorite assemblage (Gulf Coast sample Alamo 1 well 5825 m) (a) SEM image of thepetrographic thin section showing the emplacement of the FIB foil to be cut across the illitendashchlorite interface (b)Bright-field TEM image of the FIB foil extracted from the section with the platinum strap on top of it (cd) Enlargementof part b showing the Al and Mg contents of chlorite (apfu) and their variation between crystal core and crystal rim

630 O Vidal et al

Figure 10 shows that it is possible to demonstrateheterogeneities in XFe3+ contents at the nanometerscale Consistent variations of other major elements arealso observed with TEM technics on FIB foils Anexample is shown in Fig 11 which depicts thevariations of Al and Mg contents in a chlorite grainfrom the Gulf Coast (Bourdelle et al 2013b) In thiscase the FIB foil was cut parallel to the interfacebetween illite and chlorite and a nm-scale investiga-tion of composition was conducted by TEM-EDXanalysis The rim and core compositions showed azoning of chlorite with Mg- and Al-richer rims thancore compositions This variation in composition on ascale of lt1 μm was related to various crystallizationstages during burial Chlorite sampled at differentdepths and various locations of the Gulf Coast areashowed that zoned compositions similar to thatobserved in Fig 11 are frequent (Bourdelle et al2013b) The AlIV in chlorite rims display a Tdependence that is consistent with the km-scale andregional evolution of temperature ie with thediagenetic history of Gulf Coast (burial P-T condi-tions detritalauthigenic compositions etchellip)(Fig 12) This study suggests again that approachesused for metamorphic rocks can be applied to low-grade rocks when using adapted analytical tools able toidentify sub-μm variations of composition

CONCLUS ION

The variability of chlorite composition reflects thevariability of its conditions of formation Numerousefforts have been made over the last 30 years to

translate the composition of chlorite in terms of itstemperature of formation The first empirical equationsbased on the AlIVor vacancy contents are restricted tobulk-rock compositions and pressure conditions atwhich they were calibrated Thermodynamicapproaches and semi-empirical equations based onthe ratio of end-member activities do not present thislimit Among the different thermodynamic models ofchlorite published so far Vidal et al (2006) proposed amodel that can be used to estimate T and (XFe3+)Chlsimultaneously The estimated (XFe3+)Chl values areconsistent with those measured by Moumlssbauer orXANES spectroscopy and for the same (XFe3+)Chlthe temperatures estimated using the method of Vidalet al (2006) are similar to those obtained with thesemi-empirical equations of Inoue et al (2009) andLanari et al (2014) Different T-XFe3+ trends areobserved for different pressure gradients and oxygen-buffering assemblages (Lanari et al 2014) butoverall the systematic increase in XFe3+ with areduction in temperature is compatible with thedecrease in aO2 in a closed system The increase inXFe3+ and chlorite vacancy with decreasing tempera-ture is interpreted as reflecting the incorporation of theFe3+-sudoite component which can be used toestimate T-aO2-XFe

3+ values with a Chl-Qtz-H2Omulti-equilibrium approach

Similar temperatures are calculated for low-tem-perature chlorites (T lt 300degC and measured XFe3+ upto 30) with all approaches assuming that iron inchlorite is either completely ferrous or with estimatedvalues of XFe3+ This indicates that the equations ofsemi-empirical thermometers assuming that all iron inchlorite is ferrous include some inherent Fe2+Fe3+ vstemperature correction However very different tem-peratures can be obtained for higher-temperaturesamples (gt300degC) with approaches involving ferriciron or assuming that all of the iron is ferrous Thissuggests that the Fe2+Fe3+ vs temperature correctionmight be restricted to particular aO2 conditions andpressure gradients More work is necessary to betterunderstand this point and to improve the thermo-dynamic models and semi-empirical thermometersThis requires high-quality and well constrained naturaldata To this end mapping techniques that highlight thelink between the variations in temperature and pressureand small-scale chlorite composition including XFe3+

variations offer much promise When combining theresults of the present study with those of other studiespublished recently it appears that thermodynamicapproaches and mapping techniques developed formetamorphic rocks could be used to study low-grade

FIG 12 Range of crystal-rim AlIV content (white boxes)of Gulf Coast chlorites vs measured temperatures Thegrey area underlines the range of maximum AlIV contentwhich is assumed to represent the closest approach toequilibrium composition (after Bourdelle et al 2013b)

631Conditions of formation of chlorite

rocks where kinetics are much more sluggish Thisrequires use of appropriate analytical tools and techni-ques with a spatial resolution of just a few hundrednanometres which are now becoming available

ACKNOWLEDGMENTS

Constructive reviews by S Hillier F Nieto and ananonymous reviewer are gratefully acknowledged

REFERENCES

Beaufort D Patrier P Meunier A amp Ottaviani MM(1992) Chemical variations in assemblages includingepidote andor chlorite in the fossil hydrothermalsystem of Saint Martin (Lesser Antilles) Journal ofVolcanology and Geothermal Research 51 95ndash114

Berman R (1988) Internally-consistent thermodynamicdata for minerals in the system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2Journal of Petrology 29 445ndash522

Berman R amp Brown T (1985) Heat capacity of minerals inthe system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2 representation estima-tion and high temperature extrapolation Contributionsto Mineralogy and Petrology 89 168ndash183

Bourdelle F amp Cathelineau M (2015) Low-temperaturechlorite geothermometry a graphical representationbased on a TndashR2+ndashSi diagram European Journal ofMineralogy 27 617ndash626

Bourdelle F Parra T Chopin C amp Beyssac O (2013a) Anew chlorite geothermometer for diagenetic to low-grade metamorphic conditions Contributions toMineralogy and Petrology 165 723ndash735

Bourdelle F Parra T Beyssac O Chopin C amp Vidal O(2013b) Clay minerals thermometry a comparativestudy based on high-resolution chemical analyses ofillite and chlorite in sandstones from the Gulf Coast(Texas USA) American Mineralogist 98 914ndash926

Bourdelle F Benzerara K Beyssac O Cosmidis JNeuville DR Brown GE amp Paineau E (2013c)Quantification of the ferric ferrous iron ratio insilicates by scanning transmission X-ray microscopyat the Fe L23 edges Contribution to Mineralogy andPetrology 166 423ndash434

Cathelineau M (1988) Cation site occupancy in chloritesand illites as function of temperature Clay Minerals23 471ndash485

Cathelineau M amp Nieva D (1985) A chlorite solidsolution geothermometer the Los Azufres (Mexico)geothermal system Contributions to Mineralogy andPetrology 91 235ndash244

Chermak JA amp Rimstidt JD (1989) Estimating thethermodynamic properties (ΔGdegf and ΔHdegf) of silicate

minerals at 298K from the sum of polyhedral contribu-tions American Mineralogist 74 1023ndash1031

Coggon R amp Holland TJB (2002) Mixing properties ofphengitic micas and revised garnent-phengite thermo-barometers Journal of Metamorphic Geology 20683ndash696

Curtis CD Hughes CR Whiteman JA ampWhittle CK(1985) Compositional variation within some sedi-mentary chlorites and some comments on their originMineralogical Magazine 49 375ndash386

de Andrade V Vidal O Lewin E OrsquoBrien P amp Agard P(2006) Quantification of electron microprobe com-positional maps of rock thin sections an optimizedmethod and examples Journal of MetamorphicGeology 24 655ndash668

de Andrade V Susini J Salomeacute M Beraldin O RigaultC Heymes T Lewin E amp Vidal O (2011)Submicrometer hyperspectral X-ray imaging of het-erogeneous rocks and geomaterials applications at theFe k-edge Analytical Chemistry 83 4220ndash4227

de Caritat P Hutcheon I amp Walshe J (1993) Chloritegeothermometry a review Clays and Clay Minerals41 219ndash239

Essene EJ (2009) Thermobarometry gone astray Pp101ndash129 in Physics and Chemistry of Earthrsquos Interior(AK Gupta amp S Dasgupta editors) PlatinumJubilee Indian National Science Academy Springer-Verlag Berlin

Essene EJ amp Peacor DR (1995) Clay mineralthermometry ndash a critical perspective Clays and ClayMinerals 43 540ndash553

Ganne J de Andrade V Weinberg RF Vidal O DubacqB Kagambega N Naba S Baratoux L Jessell M ampAllibon J (2012) Modern-style plate subductionpreserved in the Palaeoproterozoic West Africancraton Nature Geoscience 5 60ndash65

Grosch EG Vidal O Abu-Alam T amp McLoughlin N(2012) PTconstraints on the metamorphic evolution ofthe Paleoarchean Kromberg type-section BarbertonGreenstone Belt South-Africa Journal of Petrology53 513ndash545

Grosch E McLoughlin N Lanari P Erambert M ampVidal O (2014) Microscale mapping of alterationconditions and potential biosignatures in basaltic-ultramafic rocks on early Earth and beyondAstrobiology 14 216ndash228

Hillier S amp Velde B (1991) Octahedral occupancy andchemical composition of diagenetic (low-temperature)chlorites Clay Minerals 26 149ndash168

Holland TJB (1989) Dependance of entropy on volumefor silicates and oxide minerals A review and apredictive model American Mineralogist 74 5ndash13

Holland T amp Powell R (1998) An internally consistentthermodynamic data set for phases of petrologicalinterest Journal ofMetamorphic Geology 16 309ndash343

Holland T Baker J amp Powell R (1998)Mixing propertiesand activity composition and relationships of chlorites

632 O Vidal et al

that the amount of vacancies and Si in chloritedecreases with increasing temperature In contrastequilibria 3 and 4 indicate that the amount of vacanciesin chlorite equilibrated with DspCo increases withtemperature

In low-pressure rocks such as those used to calibratethe various AlIV empirical thermometers published todate the effect of pressure is negligible compared tothat of temperature and the variation of AlIV is a goodproxy for temperature This is not true for high-pressure low-temperature chlorite crystallizing atconditions typical of subduction zones In such asetting the effect of pressure on the equilibriumtemperature of equilibria 1 and 2 cannot be furtherignored and high-pressure chlorites can have vacan-cies and Al contents similar to chlorite crystallizing atmuch lower temperature and pressure conditions Thispoint is illustrated in Fig 3 which shows the T and Pdependence of chlorite composition inferred from thenatural data of Lanari et al (2014) At P lt 4 kbar and Tlt 300degC the amount of Al on M1 (amesite content)increases almost linearly with temperature (Fig 3a)which is consistent with the empirical AlIV-Tempiricalequations published to date At higher P and T theamount of amesite in chlorite does not follow the trendobserved at low temperature because it is alsocontrolled by pressure Thus the large scatter ofamesite content vs T in Fig 3a at P gt 6 kbar reflectsthe variability of the TP conditions among thedifferent samples A similar observation is made forthe vacancy-T evolution (Fig 3b) which definesdifferent trends for the low-pressure and medium- tohigh-pressure samples Finally Fig 3c shows that theFeMg ratio of chlorite (considering FeT = Fe2+) doesnot show any consistent relationship with T and P This

ratio is controlled primarily by the bulk-rock compos-ition and the FeMg correction introduced in empiricalAlIV thermometers does not seem to be supportedeither for theoretical reasons or by observations whenconsidering large sets of natural data

MODELL ING THE CONDIT IONS OFCHLOR ITE FORMATION AND FE 3 +

CONTENT

The discussion above offers a theoretical backgroundfor the evolution of AlIV and vacancy contents withtemperature that has been documented by manyauthors in the past It suggests that empirical thermo-meters based on the sole variation of AlIV or vacancycontent cannot be extrapolated outside the range of Pconditions and bulk-rock composition at which theywere calibrated This explains why a myriad ofdifferent equations have been published in the pastthey were derived from different sets of naturalsamples

As an alternative approach to the empirical thermo-meters the temperature condition of chlorite formationcan be estimated from the thermodynamic calculationof equilibria similar to those listed above (1ndash4) Thisrequires knowledge of the standard-state thermo-dynamic properties of the relevant end-members andactivity-composition models Walshe (1986) proposedthe first chlorite model which was able to estimate theformation temperature ferric iron and water contents ofchlorite coexisting with quartz and water at known orassumed pressure The composition of chlorite wasrepresented by six independent thermodynamic com-ponents Si4Mg6O10(OH)8 Clin Daph Si3Al2(Fe

2+)5O10(OH)8 Si4Al4O10(OH)8 and Si3Al2(Fe

2+)4Fe3+

FIG 3 Al on M1 (a) vacancy on M1 (b) and XFe of chlorite (c) as a function of temperature and pressure (after Lanariet al 2014) assuming that FeTotal = Fe2+ AlM1 and ⧠M1 are proportional to the amount of amesite and sudoite end-members respectively (AlM1 = 1 for pure amesite ⧠M1 = 1 for pure sudoite AlM1 =⧠M1 = 0 for amesite-free and

sudoite-free chlorite) The grey area in part b encompasses the vacancy contents reported by Vidal et al (2001)

618 O Vidal et al

O11(OH)7 which can be used to write four chlorite-quartz-water equilibria (two independent) This modelwas not used in the present contribution because thestandard-state and solid-solution thermodynamic proper-ties were not constrained by experimental data and theyare not compatible with the existing internally consistentdatabases Two kinds of thermodynamic models com-patiblewith the existing internally consistent databases ofHolland amp Powell (1998) or Berman (1988) have beenproposed so far they have different philosophies andassumptions

(1) Vidal et al (2001 2006) considered a cationrepartition on two tetrahedral sites T1 (SiSi)2 and T2(SiAl)2 and three octahedral sitesM1 (AlFe2+Mg2+)1M2ndash3 (Mg2+Fe2+Al)4 and M4 (AlFe3+)1 and assumedequipartition of Fe and Mg on M2ndashM3 and M1 Theconditions of chlorite formation in the presence of quartzandH2Oare calculated using four equilibria involving fivechlorite end-members (Clin Daph Fe-Am Mg-Am andMg-Sud)

2 Clinthorn 3 Mg-Sud frac14 4 Mg-Amthorn 7 Qtz

thorn 4 H2O (7)

4 Clinthorn 5 Fe-Am frac14 4 Daphthorn 5 Mg-Am (8)

16 Daphthorn 15 Mg-Sud frac14 20 Fe-Amthorn 6 Clin

thorn 35 Qtzthorn 20 H2O (9)

4 Daphthorn 6 Mg-Sud frac14 5 Fe-Amthorn 3 Mg-Am

thorn 14 Qtzthorn 8 H2O (10)

The thermodynamic properties and non-ideal mixingproperties (Table 1) are believed to be compatible withthe internally consistent thermodynamic database ofBerman (1988) The properties of Clin Mg-Am Mg-Sud and the WAl-Mg Margules parameter wereconstrained by experimental data gathered in theMgO-Al2O3-SiO2-H2O chemical system and thoseof Daph and Fe-Ames and WAl-Fe were constrainedindependently by experimental data obtained in theFeO-Al2O3-SiO2-H2O (FASH) system (Fig 4) Thisapproach leads to contrasted values of WAl-Fe(minus106 kJ at 300degC 5 kbar) and WAl-Mg (+68 kJ at300degC 5 kbar) which seems unlikely in view of thesimilar ionic potential of Fe2+ and Mg2+ Howeverattempts to resolve this problem while maintaining agood fit of the experimental constraints in MASH andin FASH proved unsuccessful For the cation reparti-tion used by Vidal et al (2005 2006) the end-members Clin Daph Mg-Am and Fe-Am areindependent because the site M1 is filled either byMg in the case of Clin or Fe in the case of Daph

Equilibrium 8 corresponds thus to the intracrystallineexchange reaction MgM1(Fe5Mg4)M4 = FeM1(Mg5Fe4)M4 This internal equilibrium is achieved atdefined P and T conditions along a univariant line atfixed Fe3+ values Two of the four equilibria (1 8 9and 10) are independent and the P T XFe3+ conditionsof chlorite + quartz + water equilibrium are defined bytheir intersection in a P-T field This approach allowsone to estimate aminimum value ofXFe3+ = Fe3+FeTotalrequired to obtain convergence of equilibria at assumedpressure In view of the analytical uncertainties andunknown uncertainties of the thermodynamic propertiesof the chlorite end-members and non-ideal solid-solution parameters a certain scatter between equilibria(1 8 9 and 10) is accepted In the present studyconvergence was assumed to be achieved when thetemperature difference between equilibrium 8 ndash themost sensitive to small variations of compositions ndash andthe other three equilibria was lt30degC Equilibriumtemperatures and (XFe3+)Chl were calculated for allchlorite compositions at the pressure conditionsreported in the original publications or at 1 to 2 kbarfor the low-temperature samples Calculation wasconducted starting from XFe3+ = 0 and increasing thevalue by 1 until convergence was achieved It isemphasized that convergence at lower temperature canbe maintained with a further increase of XFe3+ Theestimated values of XFe3+ discussed in the followingare therefore minimum values at maximum tempera-tures The effect of a further increase of XFe3+ onequilibrium temperature is weak at low temperature butit can be very significant at high temperature (seeGrosch et al 2012) Note also that the quantity of Feconsidered as Fe3+ influences the activity of all chloriteend-members Increasing XFe3+ leads to an increase invacancies in structural formulae calculated on a 14anhydrous oxygen basis so that even the activity ofMg-Sud and the location of equilibrium 1 involving Fe-freeend-members are affected

(2) Lanari et al (2014) considered a cation-siterepartition identical to that adopted by Vidal et al(2001 2006) but they followed the formalism ofPowell amp Holland (1999) which assumes that onlythree of the four end-members Mg-Ames Fe-Am Clinand Daph are independent It follows that only one ofequilibria (1 8 9 and 10) is independent Thetemperature of chlorite formation is calculated fromthe location of equilibrium 1 with an assumption thatXFe3+ and the Gibbs free energy of the dependent end-member Fe-Am can be estimated with relationshipsbetween the Mg- and Fe-bearing end-members (seeLanari et al 2014) Unfortunately this approach does

619Conditions of formation of chlorite

not allow us to fit the experimental data of chloritecomposition reported by Parra et al (2005) in theFASH system (Fig 4)

Whatever the thermodynamic model at least onetemperature of chlorite-quartz-water equilibrium can becalculated at fixed pressure with assumptions regardingthe proportion of ferric iron in chlorite This calculationinvolves evaluation of equilibrium constants expressedas a ratio of chlorite end-member activities andtherefore cation proportions It follows that chloritesof different compositions (and different AlIV contents)crystallizing at identical P-T conditions but in rocks of

different compositions can have identical equilibriumconstants of (1) and (8) to (10) Inoue et al (2009)Bourdelle et al (2013a) and Lanari et al (2014)proposed several semi-empirical chlorite thermometersalso involving equilibrium constants calculated as theratio of chlorite end-member activities The maindifference compared to the thermodynamic approachis a simplification of the Gibbs free energy equationactivity models and assumptions on the ferric ironcontent in chlorite These semi-empirical thermometersare easy to use and they are believed to reproducesuccessfully temperatures estimated independently in

TABLE 1 Standard-state properties of chlorite end-members and Margules parameters of the chlorite solid solution (seeBerman 1988 Vidal et al 2006 for the format)

Mg-sudoite (Vidal et al 1992 2006) [Si2]T1[Si Al]T2)IV([A]M1[Mg2 Al2]M2[Al]M4)

VIO10(OH)8ST minus86552730 39050 20510C1 111584 minus85075 minus19777440 2959147000V1 264515406 00 minus018194707 00Ferri-sudoite (present study) [Si2]T1[Si Al]T2)

IV([A]M1[Mg2 Al2]M2plusmn3[Fe3thorn]M4)

VIO10(OH)8ST minus8221273a 40550 20400

minus8230400b

C1 112225 minus85695 minus17431790 2552588000V1 264515406 00 minus018194707 00Daphnite (Vidal et al 2001 2006) [Si2]T1[Si Al]T2)

IV([Fe2thorn]M1[Fe2thorn4 ]M2plusmn3[Al]M4)

VIO10(OH)8ST minus7120845 55940 21600C1 1229233 minus102565 minus12276900 2121510000V1 264515406 00 minus018194707 00Fe-amesite (Vidal et al 2005 2006) [Si2]T1[Al2]T2)

IV([Al]M1[Fe2thorn4 ]M2plusmn3[Al]M4)

VIO10(OH)8ST minus76074600 51480 20900C1 121981 minus100215 minus14439720 2392326000V1 264515406 00 minus018194707 00Clinochlore (Berman 1988) [Si2]T1[Si Al]T2)

IV([Mg]M1[Mg4]M2plusmn3[Al]M4)VIO10(OH)8

ST minus890958984 43515 21147C1 1214284 minus1121713 0000 minus1256253184V1 264515406 00 minus018194707 00Mg-amesite (Vidal et al 2001 2006) [Si2]T1[Al2]T2)

IV([Al]M1[Mg4]M2plusmn3[Al]M4)VIO10(OH)8

ST minus90359005 40320 20520C1 114445 minus83272 minus20058040 2820786000V1 264515406 00 minus018194707 00Margules parameters on site M1 (Vidal et al 2006)WMg-Al minus9400 minus30 minus02WMg- 10000 minus25 09WFe-Al 1200 31 07WFe- 2000 minus15 04WAl- minus10000 minus30 09

ST ΔHdegf (J) Sdeg(JK) Vdeg(Jb) Cp K0 K1 K2 K3 (Cp = K0 + K1Tminus05 + K2Tminus2 + K3Tminus3) V(PT) V1 V2 V3 V4 (V(PT)V(1298) = 1 + V1(Tminus 298) + V2(Tminus 298)2 + V3(Pminus1) + V4(Pminus 1)2)a estimated using equilibria 1 and 21ndash24 b estimated as (ΔHdegf (Mg-sudoite)minus ΔHdegf (corundum)2 + ΔHdegf (hematite)2)The Margules parameters list the WH WS and Wv values (W =WH + T(K)WS + (Pminus 1)Wv)

620 O Vidal et al

low-temperature series (Inoue et al 2009 Bourdelleet al 2013a Lanari et al 2014) and metamorphicrocks (Lanari et al 2014) The weakness of the semi-empirical equations is that their compatibility withexperimental data obtained in MASH and FASHsystems has not always been checked and they arerestricted to quartz-bearing rocks

Finally the main difference between the thermo-dynamic approach of Vidal et al (2005 2006) andthose of Vidal et al (2001) Lanari et al (2014) or thesemi-empirical thermometers of Inoue et al (2009)

Bourdelle et al (2013a) and Lanari et al (2014) is thatthe former leads to the estimation of T-(XFe3+)Chl pairsat given pressure whereas the latter estimate T atassumed or measured XFe3+ and pressure or assuminga negligible role of pressure This last assumption isacceptable at low-pressure conditions but becomesproblematic for metamorphic samples because theamount of Mg-Sud end-member at fixed temperaturehas been shown to increase with pressure (Fig 3b)

A comparison of the approaches of Vidal et al(2006) Lanari et al (2014) and Inoue et al (2009) is

FIG 4 Comparison of measured and calculated Si content of chlorite in the FASH systemwith almandine and fayalite (a)or chloritoid and hercynite (b) Boxes experimental data of Parra et al (2005) continuous lines Si content calculatedwith the thermodynamic properties of Fe-amesite and daphnite and the solid solution model of Vidal et al (2005 2006)dashed lines Si content calculated with the Holland et al (1998) model updated by Coggon amp Holland (2002) which

assumes that the thermodynamic properties of Fe-Am depend on those of Clin Mg-Am and Daph

621Conditions of formation of chlorite

shown in Fig 5 which reports temperatures estimatedfor the set of 1541 analyses used in the present studyThe temperatures calculated with the assumption Fe3+

= 0 (equation 38 of Lanari et al 2014) or Fe3+ gt 0 (allother approaches) are labelled TFe2+ and TFe3+respectively in Fig 5 and in the following Allcalculations were conducted with a MATLAB scriptwith a GUI interface available on request from the firstauthor

The temperature conditions estimated for blueschist-facies metamorphic samples with the approach ofVidal et al (2006) (TFe3+ Vidal) cover a wide range oftemperature from 275 to 550degC (Fig 5a) About onethird of the samples show temperatures of lt350degCwhich is 50ndash100degC lower than the published peaktemperatures derived for the samples studied Usingthe XFe3+ estimated with the Vidal et al (2006)approach the temperatures estimated with the semi-empirical equations of Inoue et al (2009) (TFe3+

Inoue) or Lanari et al (2014) (TFe3+ Lanari) show asimilar evolution but the estimated temperatures are50ndash100degC lower in the case of Inoue et al (2009) andhigher in the case of Lanari et al (2014) The mostimportant deviation is observed between TFe2+ Lanariand the other temperature estimates A significantproportion of chlorites calculated to have crystallized atTFe3+asymp 300degC give much higher TFe2+Lanari (gt500degCopen symbols in Fig 5a) and plot on a different trendThe same observation is made for the Barbertonsamples from Grosch et al (2012) (Fig 5c) In bothcases the maximum TFe2+ Lanari can be as high as900degC but the highest temperatures are obtained forsamples containing very small amounts of vacancies(lt005 pfu open circles below the dashed line in thetemperature-vacancy plot shown in Fig 5a) The Mg-Sud activity and temperatures derived from theseanalyses are very sensitive to small variations incomposition and Fe3+-content and they should not beused for reliable thermometric estimates In contrastFig 5b shows that similar temperatures are estimatedwith all approaches for low-pressure and low-gradesamples

This comparison indicates that similar results areobtained with all methods at T lt 300degC and similartrends are observed at higher temperature for the threemethods allowing the incorporation of Fe3+ althoughwith a significant scatter at T gt 400degC With the XFe3+

values derived using the Vidal et al (2006) approachTFe3+ Vidal and TFe3+ Inoue are quite consistentHowever the highest TFe3+ Inoue estimated forblueschist metamorphic samples are lower whereasTFe3+ Lanari are higher In all approaches the

temperature calculated depends heavily on thesudoite activity which is proportional to the amountof vacancy in chlorite (a(Sud) = 64 (XSi XAl)T2 XM1 (XMg)2M2ndash3 (XAl)2M2ndash3 (XAl)M4))This is illustrated in Fig 5 which shows that theamount of vacancy in chlorite increases with a drop intemperature As noted by Laird (1988) the amount ofvacancy in chlorite structural formulae calculated on acharge-normalized 14 anhydrous-oxygen basisincreases with increasing XFe3+ It is therefore notsurprising that like vacancies XFe3+ increases withreducing temperature This trend is illustrated in Fig 6where the TFe3+Vidal vs XFe3+ values of samples usedto compute Fig 5 have been plotted Figure 6 alsoindicates that larger XFe3+ values are calculated for thehigh-pressure samples than for the low-pressuresamples Nevertheless vacancies are also noted toincreasewith reducing Tunder the assumption that Fe2+

= FeTotal (see Fig 3b and the open circles in Fig 5ab)Therefore the increase in vacancies with reducingtemperature is not just a mathematical artefact resultingfrom the calculation of Fe3+-bearing structural formulaeon a 14-oxygen basis Interestingly even at lowtemperature where XFe3+ calculated with the Vidalet al (2006) approach is at a maximum TFe2+ Lanari issimilar to the three TFe3+ (Fig 5b) This indicates thatthe TFe2+ Lanari equation includes the systematicvariation of XFe3+ with temperature observed inFig 6 which explains why similar TFe2+ and TFe3+

are obtained at low T and P (Fig 5b) for contrastingamounts of vacancies and ferric iron contentsNevertheless it is evident from Fig 5ac that the XFe3+ vs temperature relationship that was used to derive theTFe2+ Lanari equation is not fully consistent with the T-XFe3+ relationship obtained with the thermodynamicapproach of Vidal et al (2006) This is illustrated by thedivergence between TFe2+ Lanari and the three TFe3+

calculated for the low-pressure Barberton samples(Fig 5c) and several HP samples Fig 5a

COMPARISON OF EST IMATED ANDMEASURED (XFE 3 + )CHL

At this stage we can only conclude from thecomparison of TFe2+ and TFe3+ in Fig 5ac thateither some XFe3+ values derived with the Vidal et al(2006) approach are overestimated and the low TFe3+

obtained for some HP-HT samples are underestimatedor that TFe2+Lanari only pertains for fO2

conditionsbuffered by specific assemblages An overestimationof XFe3+ due to errors in the standard-state thermo-dynamic properties and the non-ideal Margules

622 O Vidal et al

FIG 5 Left Plot of temperatures estimated with equations 38 (TFe3+ Lanari) and 40 (TFe2+ Lanari) of Lanari et al(2014) and with the thermometer of Inoue et al (2009) (TFe3+ Inoue) vs temperatures estimated using the approach ofVidal et al (2006) TFe3+ Lanari and TFe3+ Inoue were calculated using the XFe3+ values obtained with the approach ofVidal et al (2006) and TFe2+ Lanari assuming XFe3+ = 0 Right Plot of vacancy content vs estimated temperatures(a) and (b) are high-pressure (gt6 kbar) and low- to medium-pressure (lt6 kbar) respectively (c) shows the resultsobtained for the low-pressure samples of Grosch et al (2012) The unrealistically high TFe2+ Lanari are obtained forchlorites with lt005 vacancy pfu (dashed line in the vacancy-temperature plots) The histograms show the repartition

of temperatures estimated with the different approaches

623Conditions of formation of chlorite

parameters of Vidal et al (2006) is possible butthis should be demonstrated by a systematic compari-son of measured and estimated XFe3+ from naturalsamples covering a wide range of P T and fO2

conditions Such a study is still lacking but a first-order comparison of measured and estimated XFe3+

contents for samples of known temperature ofcrystallization is shown in Fig 7 which was obtainedfor the Rouez Niger and Saint-Martin low-T samplescompiled by Inoue et al (2009) the Glarus sample ofBourdelle et al (2013c) and a blueschist-facies sampleretrogressed into the greeschist facies from Sambagawa(de Andrade et al 2006 Munoz et al 2006 Vidalet al 2006) T-XFe3+ values were calculated at 2 kbarexcept for the Sambagawa sample for which calcula-tions were made at 2 5 and 10 kbar for the low-medium- and high-temperature chlorite groups identi-fied by de Andrade et al (2006) and Vidal et al (2006)respectively

FIG 6 Estimated TFe3+ Vidal and XFe3+ pairs for low- andmedium-pressure samples (open circles) high-pressuresamples (black circles) and the samples from Grosch et al

(2012) (grey squares)

FIG 7 Boxes measured XFe3+ and reference temperatures Symbols Tmaximum-XFe3+minimum pairs calculated to

obtain convergence (plusmn30degC) of equilibria 1 8 9 and 10 with the thermodynamic model of Vidal et al (2006)

624 O Vidal et al

For each series of Inoue et al (2009) between 50and 60 of the chlorite analyses contain Si lt3 apfuand led to a convergence of equations 1 8 9 and 10within 30degC The calculated T-XFe3+ pairs are plottedin Fig 7 and compared with the range of referencetemperature (TRef) estimated independently and XFe3+

measures reported by Inoue et al (2009) A fairagreement of the measured and calculated TFe3+ Vidal-XFe3+ values is observed for the Niger and St-Martinanalyses and a good agreement of the TRef and TFe3+ Vidal was observed for the Rouez samples but thecalculated valueswere smaller than themeasured valuesfor XFe3+ for this samples However the XFe3+ valuesreported in Fig 7 are minimum values required toobtain convergence of equilibria 1 and 8ndash10 within30degC Convergence is maintained at higher valuesof XFe3+ and TFe3+ Vidal for the Rouez samples arestill located in the range of TRef when XFe3+ is setat gt15 (lower limit of the reference values) TheTFe3+ Vidal values calculated for the Glarus chloriteanalyses listed in Bourdelle et al (2013c) are 139degC 203degC and 224degC in fair agreement with thoseestimated independently by Bourdelle et al (2013c)(135degC 170degC and 182degC respectively) or with thesemi-empirical thermometer of Inoue et al (2009)(145degC 196degC and 210degC respectively) For thethree samples studied the XFe3+ values measuredusing XANES (X-ray absorption near edge struc-ture) spectra in the transmission mode of scanningtransmission X-ray microscopy (STXM) were foundto be 223 238 and 279 These values comparefavourably with the estimated ones (32 27 and 28respectively) although the measured values suggestan increase in XFe3+ with temperature which iscontrary the trend inferred from the thermodynamicestimates and evidenced in Figs 6 and 7 Finally therange of XFe3+ estimated for the different generationsof chlorite in the Sambagawa sample (ellipses inFig 7) are slightly higher but still in fair agreementwith those inferred from the XANES measurementsIn conclusion Fig 7 shows that the values estimatedon the basis of the convergence of equilibria (1 8 9and 10) are consistent with the TRef and measuredXFe3+ for various samples covering a range oftemperature from 100degC to 500degC pressure from sim2to 10 kbar and XFe3+ values from 5 to 35 Boththe measured and estimated XFe3+ values plotted inFig 7 seem to confirm the systematic decrease ofXFe3+ with increasing temperature and increase withincreasing pressure that was observed in Fig 6 Thistrend is similar to the decrease in vacancy observedin Fig 5

aO2 -TEMPERATURE EST IMATESAND ADDIT ION OF A FERR I -

SUDOITE END -MEMBER

The general trends inferred from our compilation ofnatural samples indicate that XFe3+ increases withdecreasing temperature and increasing pressureAlthough the model of Vidal et al (2006) involvesno Fe3+-bearing end-member the oxygen activity(aO2) prevailing during the formation of chloritesused to compute Fig 7 can be estimated assuming thatit was buffered by the presence of pure magnetite (Mt)in equilibrium with chlorite and quartz Following thisassumption aO2 for each sample has been calculatedusing equilibria (1 8 9 10) and the followingadditional equilibria which intersect at the same P T(plusmn30degC) (XFe3+)Chl and Log(aO2) (plusmn2) (Fig 8)

15 Mg-Sudthorn 40 Mtthorn 45 Qtzthorn 60 H2O

frac14 6 Clinthorn 24 Daphthorn 20 O2

(11)

6 Mg-Sudthorn 10 Mtthorn 6 Qtzthorn 12 H2O

frac14 3 Mg-Amthorn 6 Daphthorn 5 O2

(12)

15 Mg-Sudthorn 6 Clinthorn 10 Mt

frac14 15 Qtzthorn 15 Mg-Amthorn 6 Daphthorn 5 O2

(13)

6 Daphthorn 4 Clinthorn 5 O2

frac14 20 Qtzthorn 10 Mtthorn 5 Mg-Amthorn 20 H2O(14)

60 Mg-Sudthorn 12 Clinthorn 70 Mtthorn 60 H2O

frac14 45 Mg-Amthorn 42 Daphthorn 35 O2

(15)

4 Qtzthorn 2 Mtthorn Fe-Amthorn 4 H2O

frac14 2 Daphthorn O2

(16)

15 Mg-Sudthorn 6 Daphthorn 10 Mt

frac14 15 Qtzthorn 6 Clinthorn 15 Fe-Amthorn 5 O2

(17)

60 Mg-Sudthorn 70 Mtthorn 60 H2O

frac14 24 Clinthorn 45 Fe-Amthorn 6 Daphthorn 35 O2

(18)

15 Mg-Sudthorn 16 Mtthorn 12 H2O

frac14 3 Qtzthorn 6 Clinthorn 12 Fe-Amthorn 8 O2

(19)

6 Mg-Sudthorn 4 Mt

frac14 6 Qtzthorn 3 Mg-Amthorn 3 Fe-Amthorn 2 O2

(20)

12 Mg-Sudthorn 14 Mtthorn 12 H2O

frac14 6 Mg-Amthorn 3 Fe-Amthorn 6 Daphthorn 7 O2

(21)

5 Mg-Amthorn 4 Mtthorn 8 Qtzthorn 8 H2O

frac14 4 Clinthorn 3 Fe-Amthorn 2 O2

(22)

24 Mg-Sudthorn 3 Mg-Amthorn 28 Mtthorn 24 H2O

frac14 12 Clinthorn 21 Fe-Amthorn 14 O2

(23)

All calculated Log(aO2) except those of the Saint-

625Conditions of formation of chlorite

Martin samples follow the same trend and decreasewith decreasing temperature from simminus27 at 450ndash500degC10 kbar (high-temperature chlorites of Sambagawa) tominus50degC at 125degC and 1 kbar (Niger samples) (Fig 8)The Log(aO2) values of the high-T and -P samples plotin the stability field of magnetite whereas those of thelow-T samples plot in the stability field of hematitewhich is not consistent with the assumption of ChlndashMtndashQtzndashH2O equilibrium The presence of hematite hasbeen reported in the Saint-Martin samples only(Beaufort et al 1992) which plot above the trenddefined by the other samples (upper grey ellipse inFig 8) The relative position of this hematite-bearingsample is consistent with expected higher oxygenactivities required for its formation For the other low-T and apparently hematite-free samples it is possiblethat the calculated Log(aO2) values are too high or thatoxygen was not buffered by an equilibrium ofmagnetitewith chlorite In any case Fig 8 indicates that thesystematic increase in (XFe3+)Chl with decreasingtemperature observed in Figs 6 and 7 is consistentwith a lower aO2 and amount of free oxygen at low-Tthan high-T in systems where aO2 is buffered by thepresence of chlorite and iron oxides To maintain this

equilibrium with lowering temperature in metamorphicrocks where oxygen is immobile oxygen has to beconsumed and one way for that is to produce chloritethat is richer in Fe3+ This is a possible interpretationfor the formation of low-temperature Fe3+-rich chloritecoexisting with high-temperature Fe3+-poor chloritein the same thin section of the Sambagwa sample(Fig 9)

In order to discuss the Log(aO2)-T-XFe3+ evolution

in more detail for rocks where aO2 is not buffered bythe presence of magnetite it is necessary to define anFe3+-bearing chlorite end-member The simultaneousincrease of vacancy and Fe3+ content with decreasingtemperature observed in Figs 5ndash7 is explained by the3(Mg Fe2+) = 2(Fe3+) ditrioctahedral substitution(see also Trincal amp Lanari 2016 this issue) Thisexchange can be modelled with a ferri-sudoite end-member that materializes the relationship betweenXFe3+ and vacancy The standard-state thermo-dynamic properties of ferri-sudoite were constrainedas follows

(1) Because the maximum amount of (Fe3+)Chlcalculated in the previous section never exceeded1 apfu the cation distribution of Vidal et al (2006)

FIG 8 Left Ranges of Log(aO2) conditions calculated for the samples used to compute Fig 7 at the temperature andXFe3+ contents estimated from the Chl-Qtz-H2O equilibrium conditions Dashed lines hematite-magnetite-O2

equilibrium calculated using the thermodynamic properties of Berman (1988) and unit activities of Ht and Mt RightSample calculation at 2 kbar for a Rouez sample using reactions 8ndash20 (upper) or 26ndash40 (lower) The vertical linescorrespond to oxygen- and magnetite-free reactions 1 5 6 and 7 used to compute Figs 5ndash7 which converge at +30degCThe circles in the lower figure represent values calculated with (ΔHdegf ) ferri-Sud constrained with the chlorite-magnetite-quartz-water equilibria 8ndash20 (open circle) or estimated as ΔHdegf (Mg-sudoite) ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2

(filled circle) (see Table 1 and text)

626 O Vidal et al

was kept unchanged and the Fe3+-Al3+ exchange wasassumed to be restricted to the M4 site of chlorite Thechosen formula of ferri-Sud was thus ([Si2]T1[SiAl]T2)

IV([]M1[Mg2Al2]M2-3[Fe3+]M4)

VIO10(OH)8The standard-state thermodynamic properties of allother chlorite end-members the activity model and theMargules parameters were kept unchanged

(2) The third-law entropy of ferri-sudoite (Sdeg =415 JmolK) was estimated according to Holland(1989) with a configurational entropy of 2305 JmolKand a ferromagnetic contribution of 1304 JmolK itsvolume (Vdeg = 204 Jb) was estimated to be 01 Jblower than that of Mg-sudoite (2051 Jb) and the Cp(T ) function was estimated according to Berman amp

FIG 9 Rock-sample mapping of a Sambagawa sample (a) maps of estimated XFe3+ (Vidal et al 2006 Munoz et al2006) and Log(aO2) (estimated assuming that it is buffered by magnetite-chlorite equilibria see text and Fig 8)(b) measured Fe XANES edge-position proportional to Fe redox (Munoz et al 2006) (c) estimated temperature (Vidal

et al 2006 Munoz et al 2006)

627Conditions of formation of chlorite

Brown (1985) The formation enthalpy (ΔHdegf = 82254 plusmn 5 kJmol) was first estimated as the average valuebetween that calculated according to Chermak ampRimstidt (1989) (minus82204 kJmol) and that(minus82304 kJmol) calculted as ΔHdegf (Mg-sudoite) ndashΔHdegf (corundum)2 + ΔHdegf (hematite)2 using thevalues of Vidal et al (1992) for sudoite and Berman(1988) for hematite

(3) In a first step ΔHdegf and Sdeg of ferri-sudoite wereadjusted in order to reproduce the temperaturesreported in Fig 7 with oxygen-free reactions andassuming that magnetite is present in all samples(Chl-Qtz-Mt assemblage) ΔHdegf was allowed to varybetweenminus8220 andminus8230 kJmol and Sdeg within 10 JmolK of the estimated value The addition ofmagnetite and ferri-sudoite to quartz and water leadsto 17 additional equilibria and the following wereused to constrain ΔHdegf and Sdeg of ferri-sudoite

ClinthornMg-Sud frac14 Mg-Amthorn Qtzthorn H2O (1)

4 Clinthorn Daphthorn 10 ferri-Sud

frac14 10 Mg-Amthorn 25 Qtzthorn 30 Mtthorn 20 H2O(24)

7 Daphthorn 70 ferri-Sudthorn 30 Mg-Am

frac14 22 Clinthorn 75 Mg-Sudthorn 35 Mtthorn 40 H2O(25)

28 ferri-Sudthorn 11 Fe-AmthornMg-Am

frac14 6 Daphthorn 30 Mg-Sudthorn 14 Mtthorn 16 H2O(26)

Daphthorn 10 ferri-Sudthorn 10 Mg-Amthorn 10 Qtz

frac14 6 Clinthorn 15 Mg-Sudthorn 5 Mtthorn 16 H2O(27)

The ferri-Sud properties derived with the assumptionof ChlndashQtzndashMt equilibrium are listed in Table 1(referenced a)With these values the oxygen activity atthe P T and (XFe3+)Chl of all equilibria used tocompute Fig 7 can be calculated with the followingoxygen-bearing and magnetite-free equilibria whichconverge with equilibria 1 8 9 and 10 at the same PTand aO2 conditions

16 Qtzthorn 7 Mg-Amthorn 4 Daphthorn 6 Mg-Sud

thorn 5 O2 thorn 12 H2O frac14 20 ferri-Sud(28)

Fe-Amthorn 2 Mg-Amthorn 6 Qtzthorn 4 H2Othorn O2

frac14 4 ferri-Sud(29)

Mg-Amthorn 4 Daphthorn 6 Mg-Sudthorn 2 O2

frac14 3 Fe-Amthorn 8 ferri-Sudthorn 2 Qtz(30)

5 Mg-Amthorn 12 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 8 Fe-Amthorn 28 ferri-Sud

(31)

8 Daphthorn 12 Mg-Sudthorn 3 O2 frac14 7 Fe-Am

thorn 12 ferri-Sudthorn 10 Qtzthorn 4 H2O(32)

6 Clinthorn 4 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Mg-Amthorn 20 ferri-Sudthorn 5 Qtz(33)

4 Clinthorn 20 ferri-Sud frac14 30 Qtz

thorn 15 Mg-Amthorn 4 Daphthorn 5 O2 thorn 20 H2O(34)

90 Mg-Sudthorn 28 Daphthorn 32 Clinthorn 20 H2Othorn 35 O2

frac14 140 ferri-Sudthorn 15 Mg-Am

(35)

15 Qtzthorn 14 Clinthorn 16 Daphthorn 45 Mg-Sud

thorn 20 O2 thorn 20 H2O frac14 80 ferri-Sud(36)

2 Clinthorn FeAmthorn 3 Mg-Sudthorn O2

frac14 2 Mg-Amthorn 4 ferri-Sudthorn Qtz(37)

12 Clinthorn 7 Fe-Amthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 10 Mg-Amthorn 28 ferri-Sud

(38)

5 Qtzthorn 2 Clinthorn 2 Fe-Amthorn 3 Mg-Sud

thorn 2 O2 thorn 4 H2O frac14 8 ferri-Sud(39)

2 Clinthorn 8 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Fe-Amthorn 20 ferri-Sudthorn 5 Qtz(40)

30 Qtzthorn 8 Clinthorn 15 Fe-Amthorn 5 O2 thorn 20 H2O

frac14 8 Daphthorn 20 ferri-Sud

(41)

4 Clinthorn 8 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 3 Fe-Amthorn 28 ferri-Sud

(42)

Because the ferri-sudoite standard-state propertieswere constrained by the chlorite-quartz-magnetite-O2-H2O equilibrium the Log(aO2) calculated withthe equilibria above follow naturally the trend shown inFig 8 and the Log(aO2) values calculated withreactions 28ndash42 of low-temperature samples plot inthe hematite stability field This suggests that theΔGdegPT values of ferri-sudoite derived above with theassumption that aO2 is buffered by the presence ofmagnetite are possibly too high A reduction of ΔHdegf(ferri-sudoite) to 82304 kJmol (=ΔHdegf (Mg-sudoite)ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2 see Table 1)leads to Log(aO2) reduced by 45 Log units (see Fig 8)

628 O Vidal et al

and most calculated values except those of thehematite-bearing St Martin samples plot in themagnetite stability field

A more detailed study is required to refine the ferri-sudoite thermodynamic data However the resultsobtained using the preliminary values listed in Table 1show that this end-member can be used to reproduceconsistently the T-XFe3+ and T- trends obtained fromclassical Chl-Qtz-H2O calculations (Figs 5ndash7) Theincrease in XFe3+ with reducing temperature is consist-ent with the expected reduction in aO2 and the amountof free oxygen that is consumed to produce Fe3+-bearingchlorite Ferric iron is therefore a major component thathas to be taken into account when dealing with low-temperature samples Empirical or semi-empiricalchlorite thermometers which assume that all Fe is inthe ferrous state may include one of the XFe3+-T andaO2-T trends illustrated in Figs 7 and 8 but they cannotaccount for the different XFe3+-T trends that areexpected in the contexts of different aO2 conditions

FUTURE CHALLENGES LOOK INGAT LOW-GRADE ROCKS SUCH ASMETAMORPH IC ROCKS AND THE

DYNAMICS OF CHLOR ITECRYSTALL IZAT ION AT THE

MICROMETR IC TO NANOMETR ICSCALE

T-P-XFe3+ mapping at the thin-section scaleevidence of chlorite composition re-equilibration in metamorphic and low-gradesamples

Although natural samples offer valuable constraintson thermodynamic models there are two maindifficulties in using them to evaluate the reliability ofexisting chlorite models for thermometric purposes

(1) In the case of low-grade samples referencepressures and temperatures are generally known withina reasonable absolute uncertainty but the size ofchlorite grains is often small and good microprobeanalyses are sometimes difficult to obtain Moreoverthe kinetics of reaction are so sluggish that inheritedchlorite grains might remain unaffected by latervariations in temperature which makes thetemperature-composition relationships difficult toconstrain Lastly extrapolation at HT and HP ofempirical or semi-empirical models constrained byLT samples only may be erroneous in particular theeffect of pressure or eventual systematic variations ofXFe3+ cannot be appraised

(2) In the case of metamorphic samplesmicroprobe analyses are easier to obtain from largechlorite grains but chlorites of contrasted compositionthat crystallized at different temperatures and times ofthe metamorphic history are often observed to coexistmetastably in different microstructural sites of the samethin section In such cases the definition of a singlepair of reference temperature and pressure isinappropriate and cannot be restricted to the peak-temperature conditions

In order to overcome these difficulties and to bettervisualize the link between mineral compositiondeformation and P-T conditions at the thin-sectionscale de Andrade et al (2006 2011) Vidal et al(2006) and Lanari et al (2013) developed an approachwhich combines elementary maps of composition witheither a thermodynamic or a semi-empirical estimate ofchlorite crystallization temperature Maps of rawmicroprobe analyses were standardized using good-quality point analyses of minerals present in themapped area The maps of standardized analyseswere transformed into maps of structural formulaewhere each pixel analysed is converted into individualmineral structural formulae Depending on the thermo-metric approach it was then possible to compute mapsof temperature or T-XFe3+ conditions assuming thateach chlorite pixel was in equilibrium with quartz andwater The results reported by Vidal et al (2006)Yamato et al (2007) Ganne et al (2012) Lanari et al(2013 2014) and Scheffer et al (2015) showed a closelink between deformation and mineral reaction at themicroscopic scale and maps of chlorite temperaturecould be used as a basis for interpreting the micro-structures observed in thin sections of low- to middle-grade metapelites In all these studies the metastablepersistence of different generations of chlorite crystal-lizing at different temperatures during the exhumationof metamorphic samples could be used to constrain theT-deformation history Moreover the map of (XFe3+)Chl calculated by Vidal et al (2006) was in goodqualitative agreement with the in situ mapping of thisratio using the X-ray absorption near-edge structure(XANES) techniques (Fig 9) (see also Munoz et al2006) The calculated temperature and XFe3+ mapsobtained on a single sample showed the same trend ofreducing (XFe3+)Chl with increasing temperature asobserved in Figs 6 and 7

Bourdelle et al (2013c) applied a similar approachfor low-temperature samples with chlorite of smallgrain size using transmission electron microscopy-energy dispersive X-ray (TEM-EDX) spectroscopicanalyses of focused-ion beam (FIB) foils instead of

629Conditions of formation of chlorite

microprobe analyses of thin sections (Fig 10) Thoseauthors were able to observe heterogeneities in XFe3+

at the nm scale using STXM-XANES measurementsThis last study was particularly interesting as it showedthat approaches similar to those used for studyingmetamorphic rocks could be applied to low-grade anddiagenetic samples

Looking at diagenetic rocks like metamorphicrocks

The use of equilibrium thermodynamics to studylow-grade rocks has been often criticized becausesluggish kinetics might preclude achievement of globalequilibrium at low temperature and is responsible forthe crystallization of metastable phases with a lowerenergy of nucleation than stable ones Essene amp Peacor(1995) claimed that the coexistence of chemicallyinhomogeneous phases at a very fine scale in low-grade rocks is indicative of disequilibrium and thatmineral reaction is better described by a kinetic ratherthan thermodynamic approach However the coexist-ence of chemically inhomogeneous phases is also acommon observation in metamorphic rocks wherezoned grains and the coexistence for example ofchlorite grains of different compositions indicate thatthermodynamic equilibrium is approached only locallyand at a given time in the metamorphic history In factthe main difference between diagenetic and higher-temperature rocks is the size of the domains in localequilibrium The more sluggish kinetics of reactionand diffusivity at low temperature lead to small grainsand domains in equilibrium but these domains can beidentified when using appropriate analytical tools andprocedures developed recently (Bourdelle et al2013bc)

FIG 10 Quantitative Fe redox nanomapping on a FIB foilof Glarus sample (T lt 300degC Switzerland) calculatedfrom the 7087ndash71025 eV images ratio (STXM-XANES)

(from Bourdelle et al 2013c)

FIG 11 Analysis of an illite-chlorite assemblage (Gulf Coast sample Alamo 1 well 5825 m) (a) SEM image of thepetrographic thin section showing the emplacement of the FIB foil to be cut across the illitendashchlorite interface (b)Bright-field TEM image of the FIB foil extracted from the section with the platinum strap on top of it (cd) Enlargementof part b showing the Al and Mg contents of chlorite (apfu) and their variation between crystal core and crystal rim

630 O Vidal et al

Figure 10 shows that it is possible to demonstrateheterogeneities in XFe3+ contents at the nanometerscale Consistent variations of other major elements arealso observed with TEM technics on FIB foils Anexample is shown in Fig 11 which depicts thevariations of Al and Mg contents in a chlorite grainfrom the Gulf Coast (Bourdelle et al 2013b) In thiscase the FIB foil was cut parallel to the interfacebetween illite and chlorite and a nm-scale investiga-tion of composition was conducted by TEM-EDXanalysis The rim and core compositions showed azoning of chlorite with Mg- and Al-richer rims thancore compositions This variation in composition on ascale of lt1 μm was related to various crystallizationstages during burial Chlorite sampled at differentdepths and various locations of the Gulf Coast areashowed that zoned compositions similar to thatobserved in Fig 11 are frequent (Bourdelle et al2013b) The AlIV in chlorite rims display a Tdependence that is consistent with the km-scale andregional evolution of temperature ie with thediagenetic history of Gulf Coast (burial P-T condi-tions detritalauthigenic compositions etchellip)(Fig 12) This study suggests again that approachesused for metamorphic rocks can be applied to low-grade rocks when using adapted analytical tools able toidentify sub-μm variations of composition

CONCLUS ION

The variability of chlorite composition reflects thevariability of its conditions of formation Numerousefforts have been made over the last 30 years to

translate the composition of chlorite in terms of itstemperature of formation The first empirical equationsbased on the AlIVor vacancy contents are restricted tobulk-rock compositions and pressure conditions atwhich they were calibrated Thermodynamicapproaches and semi-empirical equations based onthe ratio of end-member activities do not present thislimit Among the different thermodynamic models ofchlorite published so far Vidal et al (2006) proposed amodel that can be used to estimate T and (XFe3+)Chlsimultaneously The estimated (XFe3+)Chl values areconsistent with those measured by Moumlssbauer orXANES spectroscopy and for the same (XFe3+)Chlthe temperatures estimated using the method of Vidalet al (2006) are similar to those obtained with thesemi-empirical equations of Inoue et al (2009) andLanari et al (2014) Different T-XFe3+ trends areobserved for different pressure gradients and oxygen-buffering assemblages (Lanari et al 2014) butoverall the systematic increase in XFe3+ with areduction in temperature is compatible with thedecrease in aO2 in a closed system The increase inXFe3+ and chlorite vacancy with decreasing tempera-ture is interpreted as reflecting the incorporation of theFe3+-sudoite component which can be used toestimate T-aO2-XFe

3+ values with a Chl-Qtz-H2Omulti-equilibrium approach

Similar temperatures are calculated for low-tem-perature chlorites (T lt 300degC and measured XFe3+ upto 30) with all approaches assuming that iron inchlorite is either completely ferrous or with estimatedvalues of XFe3+ This indicates that the equations ofsemi-empirical thermometers assuming that all iron inchlorite is ferrous include some inherent Fe2+Fe3+ vstemperature correction However very different tem-peratures can be obtained for higher-temperaturesamples (gt300degC) with approaches involving ferriciron or assuming that all of the iron is ferrous Thissuggests that the Fe2+Fe3+ vs temperature correctionmight be restricted to particular aO2 conditions andpressure gradients More work is necessary to betterunderstand this point and to improve the thermo-dynamic models and semi-empirical thermometersThis requires high-quality and well constrained naturaldata To this end mapping techniques that highlight thelink between the variations in temperature and pressureand small-scale chlorite composition including XFe3+

variations offer much promise When combining theresults of the present study with those of other studiespublished recently it appears that thermodynamicapproaches and mapping techniques developed formetamorphic rocks could be used to study low-grade

FIG 12 Range of crystal-rim AlIV content (white boxes)of Gulf Coast chlorites vs measured temperatures Thegrey area underlines the range of maximum AlIV contentwhich is assumed to represent the closest approach toequilibrium composition (after Bourdelle et al 2013b)

631Conditions of formation of chlorite

rocks where kinetics are much more sluggish Thisrequires use of appropriate analytical tools and techni-ques with a spatial resolution of just a few hundrednanometres which are now becoming available

ACKNOWLEDGMENTS

Constructive reviews by S Hillier F Nieto and ananonymous reviewer are gratefully acknowledged

REFERENCES

Beaufort D Patrier P Meunier A amp Ottaviani MM(1992) Chemical variations in assemblages includingepidote andor chlorite in the fossil hydrothermalsystem of Saint Martin (Lesser Antilles) Journal ofVolcanology and Geothermal Research 51 95ndash114

Berman R (1988) Internally-consistent thermodynamicdata for minerals in the system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2Journal of Petrology 29 445ndash522

Berman R amp Brown T (1985) Heat capacity of minerals inthe system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2 representation estima-tion and high temperature extrapolation Contributionsto Mineralogy and Petrology 89 168ndash183

Bourdelle F amp Cathelineau M (2015) Low-temperaturechlorite geothermometry a graphical representationbased on a TndashR2+ndashSi diagram European Journal ofMineralogy 27 617ndash626

Bourdelle F Parra T Chopin C amp Beyssac O (2013a) Anew chlorite geothermometer for diagenetic to low-grade metamorphic conditions Contributions toMineralogy and Petrology 165 723ndash735

Bourdelle F Parra T Beyssac O Chopin C amp Vidal O(2013b) Clay minerals thermometry a comparativestudy based on high-resolution chemical analyses ofillite and chlorite in sandstones from the Gulf Coast(Texas USA) American Mineralogist 98 914ndash926

Bourdelle F Benzerara K Beyssac O Cosmidis JNeuville DR Brown GE amp Paineau E (2013c)Quantification of the ferric ferrous iron ratio insilicates by scanning transmission X-ray microscopyat the Fe L23 edges Contribution to Mineralogy andPetrology 166 423ndash434

Cathelineau M (1988) Cation site occupancy in chloritesand illites as function of temperature Clay Minerals23 471ndash485

Cathelineau M amp Nieva D (1985) A chlorite solidsolution geothermometer the Los Azufres (Mexico)geothermal system Contributions to Mineralogy andPetrology 91 235ndash244

Chermak JA amp Rimstidt JD (1989) Estimating thethermodynamic properties (ΔGdegf and ΔHdegf) of silicate

minerals at 298K from the sum of polyhedral contribu-tions American Mineralogist 74 1023ndash1031

Coggon R amp Holland TJB (2002) Mixing properties ofphengitic micas and revised garnent-phengite thermo-barometers Journal of Metamorphic Geology 20683ndash696

Curtis CD Hughes CR Whiteman JA ampWhittle CK(1985) Compositional variation within some sedi-mentary chlorites and some comments on their originMineralogical Magazine 49 375ndash386

de Andrade V Vidal O Lewin E OrsquoBrien P amp Agard P(2006) Quantification of electron microprobe com-positional maps of rock thin sections an optimizedmethod and examples Journal of MetamorphicGeology 24 655ndash668

de Andrade V Susini J Salomeacute M Beraldin O RigaultC Heymes T Lewin E amp Vidal O (2011)Submicrometer hyperspectral X-ray imaging of het-erogeneous rocks and geomaterials applications at theFe k-edge Analytical Chemistry 83 4220ndash4227

de Caritat P Hutcheon I amp Walshe J (1993) Chloritegeothermometry a review Clays and Clay Minerals41 219ndash239

Essene EJ (2009) Thermobarometry gone astray Pp101ndash129 in Physics and Chemistry of Earthrsquos Interior(AK Gupta amp S Dasgupta editors) PlatinumJubilee Indian National Science Academy Springer-Verlag Berlin

Essene EJ amp Peacor DR (1995) Clay mineralthermometry ndash a critical perspective Clays and ClayMinerals 43 540ndash553

Ganne J de Andrade V Weinberg RF Vidal O DubacqB Kagambega N Naba S Baratoux L Jessell M ampAllibon J (2012) Modern-style plate subductionpreserved in the Palaeoproterozoic West Africancraton Nature Geoscience 5 60ndash65

Grosch EG Vidal O Abu-Alam T amp McLoughlin N(2012) PTconstraints on the metamorphic evolution ofthe Paleoarchean Kromberg type-section BarbertonGreenstone Belt South-Africa Journal of Petrology53 513ndash545

Grosch E McLoughlin N Lanari P Erambert M ampVidal O (2014) Microscale mapping of alterationconditions and potential biosignatures in basaltic-ultramafic rocks on early Earth and beyondAstrobiology 14 216ndash228

Hillier S amp Velde B (1991) Octahedral occupancy andchemical composition of diagenetic (low-temperature)chlorites Clay Minerals 26 149ndash168

Holland TJB (1989) Dependance of entropy on volumefor silicates and oxide minerals A review and apredictive model American Mineralogist 74 5ndash13

Holland T amp Powell R (1998) An internally consistentthermodynamic data set for phases of petrologicalinterest Journal ofMetamorphic Geology 16 309ndash343

Holland T Baker J amp Powell R (1998)Mixing propertiesand activity composition and relationships of chlorites

632 O Vidal et al

O11(OH)7 which can be used to write four chlorite-quartz-water equilibria (two independent) This modelwas not used in the present contribution because thestandard-state and solid-solution thermodynamic proper-ties were not constrained by experimental data and theyare not compatible with the existing internally consistentdatabases Two kinds of thermodynamic models com-patiblewith the existing internally consistent databases ofHolland amp Powell (1998) or Berman (1988) have beenproposed so far they have different philosophies andassumptions

(1) Vidal et al (2001 2006) considered a cationrepartition on two tetrahedral sites T1 (SiSi)2 and T2(SiAl)2 and three octahedral sitesM1 (AlFe2+Mg2+)1M2ndash3 (Mg2+Fe2+Al)4 and M4 (AlFe3+)1 and assumedequipartition of Fe and Mg on M2ndashM3 and M1 Theconditions of chlorite formation in the presence of quartzandH2Oare calculated using four equilibria involving fivechlorite end-members (Clin Daph Fe-Am Mg-Am andMg-Sud)

2 Clinthorn 3 Mg-Sud frac14 4 Mg-Amthorn 7 Qtz

thorn 4 H2O (7)

4 Clinthorn 5 Fe-Am frac14 4 Daphthorn 5 Mg-Am (8)

16 Daphthorn 15 Mg-Sud frac14 20 Fe-Amthorn 6 Clin

thorn 35 Qtzthorn 20 H2O (9)

4 Daphthorn 6 Mg-Sud frac14 5 Fe-Amthorn 3 Mg-Am

thorn 14 Qtzthorn 8 H2O (10)

The thermodynamic properties and non-ideal mixingproperties (Table 1) are believed to be compatible withthe internally consistent thermodynamic database ofBerman (1988) The properties of Clin Mg-Am Mg-Sud and the WAl-Mg Margules parameter wereconstrained by experimental data gathered in theMgO-Al2O3-SiO2-H2O chemical system and thoseof Daph and Fe-Ames and WAl-Fe were constrainedindependently by experimental data obtained in theFeO-Al2O3-SiO2-H2O (FASH) system (Fig 4) Thisapproach leads to contrasted values of WAl-Fe(minus106 kJ at 300degC 5 kbar) and WAl-Mg (+68 kJ at300degC 5 kbar) which seems unlikely in view of thesimilar ionic potential of Fe2+ and Mg2+ Howeverattempts to resolve this problem while maintaining agood fit of the experimental constraints in MASH andin FASH proved unsuccessful For the cation reparti-tion used by Vidal et al (2005 2006) the end-members Clin Daph Mg-Am and Fe-Am areindependent because the site M1 is filled either byMg in the case of Clin or Fe in the case of Daph

Equilibrium 8 corresponds thus to the intracrystallineexchange reaction MgM1(Fe5Mg4)M4 = FeM1(Mg5Fe4)M4 This internal equilibrium is achieved atdefined P and T conditions along a univariant line atfixed Fe3+ values Two of the four equilibria (1 8 9and 10) are independent and the P T XFe3+ conditionsof chlorite + quartz + water equilibrium are defined bytheir intersection in a P-T field This approach allowsone to estimate aminimum value ofXFe3+ = Fe3+FeTotalrequired to obtain convergence of equilibria at assumedpressure In view of the analytical uncertainties andunknown uncertainties of the thermodynamic propertiesof the chlorite end-members and non-ideal solid-solution parameters a certain scatter between equilibria(1 8 9 and 10) is accepted In the present studyconvergence was assumed to be achieved when thetemperature difference between equilibrium 8 ndash themost sensitive to small variations of compositions ndash andthe other three equilibria was lt30degC Equilibriumtemperatures and (XFe3+)Chl were calculated for allchlorite compositions at the pressure conditionsreported in the original publications or at 1 to 2 kbarfor the low-temperature samples Calculation wasconducted starting from XFe3+ = 0 and increasing thevalue by 1 until convergence was achieved It isemphasized that convergence at lower temperature canbe maintained with a further increase of XFe3+ Theestimated values of XFe3+ discussed in the followingare therefore minimum values at maximum tempera-tures The effect of a further increase of XFe3+ onequilibrium temperature is weak at low temperature butit can be very significant at high temperature (seeGrosch et al 2012) Note also that the quantity of Feconsidered as Fe3+ influences the activity of all chloriteend-members Increasing XFe3+ leads to an increase invacancies in structural formulae calculated on a 14anhydrous oxygen basis so that even the activity ofMg-Sud and the location of equilibrium 1 involving Fe-freeend-members are affected

(2) Lanari et al (2014) considered a cation-siterepartition identical to that adopted by Vidal et al(2001 2006) but they followed the formalism ofPowell amp Holland (1999) which assumes that onlythree of the four end-members Mg-Ames Fe-Am Clinand Daph are independent It follows that only one ofequilibria (1 8 9 and 10) is independent Thetemperature of chlorite formation is calculated fromthe location of equilibrium 1 with an assumption thatXFe3+ and the Gibbs free energy of the dependent end-member Fe-Am can be estimated with relationshipsbetween the Mg- and Fe-bearing end-members (seeLanari et al 2014) Unfortunately this approach does

619Conditions of formation of chlorite

not allow us to fit the experimental data of chloritecomposition reported by Parra et al (2005) in theFASH system (Fig 4)

Whatever the thermodynamic model at least onetemperature of chlorite-quartz-water equilibrium can becalculated at fixed pressure with assumptions regardingthe proportion of ferric iron in chlorite This calculationinvolves evaluation of equilibrium constants expressedas a ratio of chlorite end-member activities andtherefore cation proportions It follows that chloritesof different compositions (and different AlIV contents)crystallizing at identical P-T conditions but in rocks of

different compositions can have identical equilibriumconstants of (1) and (8) to (10) Inoue et al (2009)Bourdelle et al (2013a) and Lanari et al (2014)proposed several semi-empirical chlorite thermometersalso involving equilibrium constants calculated as theratio of chlorite end-member activities The maindifference compared to the thermodynamic approachis a simplification of the Gibbs free energy equationactivity models and assumptions on the ferric ironcontent in chlorite These semi-empirical thermometersare easy to use and they are believed to reproducesuccessfully temperatures estimated independently in

TABLE 1 Standard-state properties of chlorite end-members and Margules parameters of the chlorite solid solution (seeBerman 1988 Vidal et al 2006 for the format)

Mg-sudoite (Vidal et al 1992 2006) [Si2]T1[Si Al]T2)IV([A]M1[Mg2 Al2]M2[Al]M4)

VIO10(OH)8ST minus86552730 39050 20510C1 111584 minus85075 minus19777440 2959147000V1 264515406 00 minus018194707 00Ferri-sudoite (present study) [Si2]T1[Si Al]T2)

IV([A]M1[Mg2 Al2]M2plusmn3[Fe3thorn]M4)

VIO10(OH)8ST minus8221273a 40550 20400

minus8230400b

C1 112225 minus85695 minus17431790 2552588000V1 264515406 00 minus018194707 00Daphnite (Vidal et al 2001 2006) [Si2]T1[Si Al]T2)

IV([Fe2thorn]M1[Fe2thorn4 ]M2plusmn3[Al]M4)

VIO10(OH)8ST minus7120845 55940 21600C1 1229233 minus102565 minus12276900 2121510000V1 264515406 00 minus018194707 00Fe-amesite (Vidal et al 2005 2006) [Si2]T1[Al2]T2)

IV([Al]M1[Fe2thorn4 ]M2plusmn3[Al]M4)

VIO10(OH)8ST minus76074600 51480 20900C1 121981 minus100215 minus14439720 2392326000V1 264515406 00 minus018194707 00Clinochlore (Berman 1988) [Si2]T1[Si Al]T2)

IV([Mg]M1[Mg4]M2plusmn3[Al]M4)VIO10(OH)8

ST minus890958984 43515 21147C1 1214284 minus1121713 0000 minus1256253184V1 264515406 00 minus018194707 00Mg-amesite (Vidal et al 2001 2006) [Si2]T1[Al2]T2)

IV([Al]M1[Mg4]M2plusmn3[Al]M4)VIO10(OH)8

ST minus90359005 40320 20520C1 114445 minus83272 minus20058040 2820786000V1 264515406 00 minus018194707 00Margules parameters on site M1 (Vidal et al 2006)WMg-Al minus9400 minus30 minus02WMg- 10000 minus25 09WFe-Al 1200 31 07WFe- 2000 minus15 04WAl- minus10000 minus30 09

ST ΔHdegf (J) Sdeg(JK) Vdeg(Jb) Cp K0 K1 K2 K3 (Cp = K0 + K1Tminus05 + K2Tminus2 + K3Tminus3) V(PT) V1 V2 V3 V4 (V(PT)V(1298) = 1 + V1(Tminus 298) + V2(Tminus 298)2 + V3(Pminus1) + V4(Pminus 1)2)a estimated using equilibria 1 and 21ndash24 b estimated as (ΔHdegf (Mg-sudoite)minus ΔHdegf (corundum)2 + ΔHdegf (hematite)2)The Margules parameters list the WH WS and Wv values (W =WH + T(K)WS + (Pminus 1)Wv)

620 O Vidal et al

low-temperature series (Inoue et al 2009 Bourdelleet al 2013a Lanari et al 2014) and metamorphicrocks (Lanari et al 2014) The weakness of the semi-empirical equations is that their compatibility withexperimental data obtained in MASH and FASHsystems has not always been checked and they arerestricted to quartz-bearing rocks

Finally the main difference between the thermo-dynamic approach of Vidal et al (2005 2006) andthose of Vidal et al (2001) Lanari et al (2014) or thesemi-empirical thermometers of Inoue et al (2009)

Bourdelle et al (2013a) and Lanari et al (2014) is thatthe former leads to the estimation of T-(XFe3+)Chl pairsat given pressure whereas the latter estimate T atassumed or measured XFe3+ and pressure or assuminga negligible role of pressure This last assumption isacceptable at low-pressure conditions but becomesproblematic for metamorphic samples because theamount of Mg-Sud end-member at fixed temperaturehas been shown to increase with pressure (Fig 3b)

A comparison of the approaches of Vidal et al(2006) Lanari et al (2014) and Inoue et al (2009) is

FIG 4 Comparison of measured and calculated Si content of chlorite in the FASH systemwith almandine and fayalite (a)or chloritoid and hercynite (b) Boxes experimental data of Parra et al (2005) continuous lines Si content calculatedwith the thermodynamic properties of Fe-amesite and daphnite and the solid solution model of Vidal et al (2005 2006)dashed lines Si content calculated with the Holland et al (1998) model updated by Coggon amp Holland (2002) which

assumes that the thermodynamic properties of Fe-Am depend on those of Clin Mg-Am and Daph

621Conditions of formation of chlorite

shown in Fig 5 which reports temperatures estimatedfor the set of 1541 analyses used in the present studyThe temperatures calculated with the assumption Fe3+

= 0 (equation 38 of Lanari et al 2014) or Fe3+ gt 0 (allother approaches) are labelled TFe2+ and TFe3+respectively in Fig 5 and in the following Allcalculations were conducted with a MATLAB scriptwith a GUI interface available on request from the firstauthor

The temperature conditions estimated for blueschist-facies metamorphic samples with the approach ofVidal et al (2006) (TFe3+ Vidal) cover a wide range oftemperature from 275 to 550degC (Fig 5a) About onethird of the samples show temperatures of lt350degCwhich is 50ndash100degC lower than the published peaktemperatures derived for the samples studied Usingthe XFe3+ estimated with the Vidal et al (2006)approach the temperatures estimated with the semi-empirical equations of Inoue et al (2009) (TFe3+

Inoue) or Lanari et al (2014) (TFe3+ Lanari) show asimilar evolution but the estimated temperatures are50ndash100degC lower in the case of Inoue et al (2009) andhigher in the case of Lanari et al (2014) The mostimportant deviation is observed between TFe2+ Lanariand the other temperature estimates A significantproportion of chlorites calculated to have crystallized atTFe3+asymp 300degC give much higher TFe2+Lanari (gt500degCopen symbols in Fig 5a) and plot on a different trendThe same observation is made for the Barbertonsamples from Grosch et al (2012) (Fig 5c) In bothcases the maximum TFe2+ Lanari can be as high as900degC but the highest temperatures are obtained forsamples containing very small amounts of vacancies(lt005 pfu open circles below the dashed line in thetemperature-vacancy plot shown in Fig 5a) The Mg-Sud activity and temperatures derived from theseanalyses are very sensitive to small variations incomposition and Fe3+-content and they should not beused for reliable thermometric estimates In contrastFig 5b shows that similar temperatures are estimatedwith all approaches for low-pressure and low-gradesamples

This comparison indicates that similar results areobtained with all methods at T lt 300degC and similartrends are observed at higher temperature for the threemethods allowing the incorporation of Fe3+ althoughwith a significant scatter at T gt 400degC With the XFe3+

values derived using the Vidal et al (2006) approachTFe3+ Vidal and TFe3+ Inoue are quite consistentHowever the highest TFe3+ Inoue estimated forblueschist metamorphic samples are lower whereasTFe3+ Lanari are higher In all approaches the

temperature calculated depends heavily on thesudoite activity which is proportional to the amountof vacancy in chlorite (a(Sud) = 64 (XSi XAl)T2 XM1 (XMg)2M2ndash3 (XAl)2M2ndash3 (XAl)M4))This is illustrated in Fig 5 which shows that theamount of vacancy in chlorite increases with a drop intemperature As noted by Laird (1988) the amount ofvacancy in chlorite structural formulae calculated on acharge-normalized 14 anhydrous-oxygen basisincreases with increasing XFe3+ It is therefore notsurprising that like vacancies XFe3+ increases withreducing temperature This trend is illustrated in Fig 6where the TFe3+Vidal vs XFe3+ values of samples usedto compute Fig 5 have been plotted Figure 6 alsoindicates that larger XFe3+ values are calculated for thehigh-pressure samples than for the low-pressuresamples Nevertheless vacancies are also noted toincreasewith reducing Tunder the assumption that Fe2+

= FeTotal (see Fig 3b and the open circles in Fig 5ab)Therefore the increase in vacancies with reducingtemperature is not just a mathematical artefact resultingfrom the calculation of Fe3+-bearing structural formulaeon a 14-oxygen basis Interestingly even at lowtemperature where XFe3+ calculated with the Vidalet al (2006) approach is at a maximum TFe2+ Lanari issimilar to the three TFe3+ (Fig 5b) This indicates thatthe TFe2+ Lanari equation includes the systematicvariation of XFe3+ with temperature observed inFig 6 which explains why similar TFe2+ and TFe3+

are obtained at low T and P (Fig 5b) for contrastingamounts of vacancies and ferric iron contentsNevertheless it is evident from Fig 5ac that the XFe3+ vs temperature relationship that was used to derive theTFe2+ Lanari equation is not fully consistent with the T-XFe3+ relationship obtained with the thermodynamicapproach of Vidal et al (2006) This is illustrated by thedivergence between TFe2+ Lanari and the three TFe3+

calculated for the low-pressure Barberton samples(Fig 5c) and several HP samples Fig 5a

COMPARISON OF EST IMATED ANDMEASURED (XFE 3 + )CHL

At this stage we can only conclude from thecomparison of TFe2+ and TFe3+ in Fig 5ac thateither some XFe3+ values derived with the Vidal et al(2006) approach are overestimated and the low TFe3+

obtained for some HP-HT samples are underestimatedor that TFe2+Lanari only pertains for fO2

conditionsbuffered by specific assemblages An overestimationof XFe3+ due to errors in the standard-state thermo-dynamic properties and the non-ideal Margules

622 O Vidal et al

FIG 5 Left Plot of temperatures estimated with equations 38 (TFe3+ Lanari) and 40 (TFe2+ Lanari) of Lanari et al(2014) and with the thermometer of Inoue et al (2009) (TFe3+ Inoue) vs temperatures estimated using the approach ofVidal et al (2006) TFe3+ Lanari and TFe3+ Inoue were calculated using the XFe3+ values obtained with the approach ofVidal et al (2006) and TFe2+ Lanari assuming XFe3+ = 0 Right Plot of vacancy content vs estimated temperatures(a) and (b) are high-pressure (gt6 kbar) and low- to medium-pressure (lt6 kbar) respectively (c) shows the resultsobtained for the low-pressure samples of Grosch et al (2012) The unrealistically high TFe2+ Lanari are obtained forchlorites with lt005 vacancy pfu (dashed line in the vacancy-temperature plots) The histograms show the repartition

of temperatures estimated with the different approaches

623Conditions of formation of chlorite

parameters of Vidal et al (2006) is possible butthis should be demonstrated by a systematic compari-son of measured and estimated XFe3+ from naturalsamples covering a wide range of P T and fO2

conditions Such a study is still lacking but a first-order comparison of measured and estimated XFe3+

contents for samples of known temperature ofcrystallization is shown in Fig 7 which was obtainedfor the Rouez Niger and Saint-Martin low-T samplescompiled by Inoue et al (2009) the Glarus sample ofBourdelle et al (2013c) and a blueschist-facies sampleretrogressed into the greeschist facies from Sambagawa(de Andrade et al 2006 Munoz et al 2006 Vidalet al 2006) T-XFe3+ values were calculated at 2 kbarexcept for the Sambagawa sample for which calcula-tions were made at 2 5 and 10 kbar for the low-medium- and high-temperature chlorite groups identi-fied by de Andrade et al (2006) and Vidal et al (2006)respectively

FIG 6 Estimated TFe3+ Vidal and XFe3+ pairs for low- andmedium-pressure samples (open circles) high-pressuresamples (black circles) and the samples from Grosch et al

(2012) (grey squares)

FIG 7 Boxes measured XFe3+ and reference temperatures Symbols Tmaximum-XFe3+minimum pairs calculated to

obtain convergence (plusmn30degC) of equilibria 1 8 9 and 10 with the thermodynamic model of Vidal et al (2006)

624 O Vidal et al

For each series of Inoue et al (2009) between 50and 60 of the chlorite analyses contain Si lt3 apfuand led to a convergence of equations 1 8 9 and 10within 30degC The calculated T-XFe3+ pairs are plottedin Fig 7 and compared with the range of referencetemperature (TRef) estimated independently and XFe3+

measures reported by Inoue et al (2009) A fairagreement of the measured and calculated TFe3+ Vidal-XFe3+ values is observed for the Niger and St-Martinanalyses and a good agreement of the TRef and TFe3+ Vidal was observed for the Rouez samples but thecalculated valueswere smaller than themeasured valuesfor XFe3+ for this samples However the XFe3+ valuesreported in Fig 7 are minimum values required toobtain convergence of equilibria 1 and 8ndash10 within30degC Convergence is maintained at higher valuesof XFe3+ and TFe3+ Vidal for the Rouez samples arestill located in the range of TRef when XFe3+ is setat gt15 (lower limit of the reference values) TheTFe3+ Vidal values calculated for the Glarus chloriteanalyses listed in Bourdelle et al (2013c) are 139degC 203degC and 224degC in fair agreement with thoseestimated independently by Bourdelle et al (2013c)(135degC 170degC and 182degC respectively) or with thesemi-empirical thermometer of Inoue et al (2009)(145degC 196degC and 210degC respectively) For thethree samples studied the XFe3+ values measuredusing XANES (X-ray absorption near edge struc-ture) spectra in the transmission mode of scanningtransmission X-ray microscopy (STXM) were foundto be 223 238 and 279 These values comparefavourably with the estimated ones (32 27 and 28respectively) although the measured values suggestan increase in XFe3+ with temperature which iscontrary the trend inferred from the thermodynamicestimates and evidenced in Figs 6 and 7 Finally therange of XFe3+ estimated for the different generationsof chlorite in the Sambagawa sample (ellipses inFig 7) are slightly higher but still in fair agreementwith those inferred from the XANES measurementsIn conclusion Fig 7 shows that the values estimatedon the basis of the convergence of equilibria (1 8 9and 10) are consistent with the TRef and measuredXFe3+ for various samples covering a range oftemperature from 100degC to 500degC pressure from sim2to 10 kbar and XFe3+ values from 5 to 35 Boththe measured and estimated XFe3+ values plotted inFig 7 seem to confirm the systematic decrease ofXFe3+ with increasing temperature and increase withincreasing pressure that was observed in Fig 6 Thistrend is similar to the decrease in vacancy observedin Fig 5

aO2 -TEMPERATURE EST IMATESAND ADDIT ION OF A FERR I -

SUDOITE END -MEMBER

The general trends inferred from our compilation ofnatural samples indicate that XFe3+ increases withdecreasing temperature and increasing pressureAlthough the model of Vidal et al (2006) involvesno Fe3+-bearing end-member the oxygen activity(aO2) prevailing during the formation of chloritesused to compute Fig 7 can be estimated assuming thatit was buffered by the presence of pure magnetite (Mt)in equilibrium with chlorite and quartz Following thisassumption aO2 for each sample has been calculatedusing equilibria (1 8 9 10) and the followingadditional equilibria which intersect at the same P T(plusmn30degC) (XFe3+)Chl and Log(aO2) (plusmn2) (Fig 8)

15 Mg-Sudthorn 40 Mtthorn 45 Qtzthorn 60 H2O

frac14 6 Clinthorn 24 Daphthorn 20 O2

(11)

6 Mg-Sudthorn 10 Mtthorn 6 Qtzthorn 12 H2O

frac14 3 Mg-Amthorn 6 Daphthorn 5 O2

(12)

15 Mg-Sudthorn 6 Clinthorn 10 Mt

frac14 15 Qtzthorn 15 Mg-Amthorn 6 Daphthorn 5 O2

(13)

6 Daphthorn 4 Clinthorn 5 O2

frac14 20 Qtzthorn 10 Mtthorn 5 Mg-Amthorn 20 H2O(14)

60 Mg-Sudthorn 12 Clinthorn 70 Mtthorn 60 H2O

frac14 45 Mg-Amthorn 42 Daphthorn 35 O2

(15)

4 Qtzthorn 2 Mtthorn Fe-Amthorn 4 H2O

frac14 2 Daphthorn O2

(16)

15 Mg-Sudthorn 6 Daphthorn 10 Mt

frac14 15 Qtzthorn 6 Clinthorn 15 Fe-Amthorn 5 O2

(17)

60 Mg-Sudthorn 70 Mtthorn 60 H2O

frac14 24 Clinthorn 45 Fe-Amthorn 6 Daphthorn 35 O2

(18)

15 Mg-Sudthorn 16 Mtthorn 12 H2O

frac14 3 Qtzthorn 6 Clinthorn 12 Fe-Amthorn 8 O2

(19)

6 Mg-Sudthorn 4 Mt

frac14 6 Qtzthorn 3 Mg-Amthorn 3 Fe-Amthorn 2 O2

(20)

12 Mg-Sudthorn 14 Mtthorn 12 H2O

frac14 6 Mg-Amthorn 3 Fe-Amthorn 6 Daphthorn 7 O2

(21)

5 Mg-Amthorn 4 Mtthorn 8 Qtzthorn 8 H2O

frac14 4 Clinthorn 3 Fe-Amthorn 2 O2

(22)

24 Mg-Sudthorn 3 Mg-Amthorn 28 Mtthorn 24 H2O

frac14 12 Clinthorn 21 Fe-Amthorn 14 O2

(23)

All calculated Log(aO2) except those of the Saint-

625Conditions of formation of chlorite

Martin samples follow the same trend and decreasewith decreasing temperature from simminus27 at 450ndash500degC10 kbar (high-temperature chlorites of Sambagawa) tominus50degC at 125degC and 1 kbar (Niger samples) (Fig 8)The Log(aO2) values of the high-T and -P samples plotin the stability field of magnetite whereas those of thelow-T samples plot in the stability field of hematitewhich is not consistent with the assumption of ChlndashMtndashQtzndashH2O equilibrium The presence of hematite hasbeen reported in the Saint-Martin samples only(Beaufort et al 1992) which plot above the trenddefined by the other samples (upper grey ellipse inFig 8) The relative position of this hematite-bearingsample is consistent with expected higher oxygenactivities required for its formation For the other low-T and apparently hematite-free samples it is possiblethat the calculated Log(aO2) values are too high or thatoxygen was not buffered by an equilibrium ofmagnetitewith chlorite In any case Fig 8 indicates that thesystematic increase in (XFe3+)Chl with decreasingtemperature observed in Figs 6 and 7 is consistentwith a lower aO2 and amount of free oxygen at low-Tthan high-T in systems where aO2 is buffered by thepresence of chlorite and iron oxides To maintain this

equilibrium with lowering temperature in metamorphicrocks where oxygen is immobile oxygen has to beconsumed and one way for that is to produce chloritethat is richer in Fe3+ This is a possible interpretationfor the formation of low-temperature Fe3+-rich chloritecoexisting with high-temperature Fe3+-poor chloritein the same thin section of the Sambagwa sample(Fig 9)

In order to discuss the Log(aO2)-T-XFe3+ evolution

in more detail for rocks where aO2 is not buffered bythe presence of magnetite it is necessary to define anFe3+-bearing chlorite end-member The simultaneousincrease of vacancy and Fe3+ content with decreasingtemperature observed in Figs 5ndash7 is explained by the3(Mg Fe2+) = 2(Fe3+) ditrioctahedral substitution(see also Trincal amp Lanari 2016 this issue) Thisexchange can be modelled with a ferri-sudoite end-member that materializes the relationship betweenXFe3+ and vacancy The standard-state thermo-dynamic properties of ferri-sudoite were constrainedas follows

(1) Because the maximum amount of (Fe3+)Chlcalculated in the previous section never exceeded1 apfu the cation distribution of Vidal et al (2006)

FIG 8 Left Ranges of Log(aO2) conditions calculated for the samples used to compute Fig 7 at the temperature andXFe3+ contents estimated from the Chl-Qtz-H2O equilibrium conditions Dashed lines hematite-magnetite-O2

equilibrium calculated using the thermodynamic properties of Berman (1988) and unit activities of Ht and Mt RightSample calculation at 2 kbar for a Rouez sample using reactions 8ndash20 (upper) or 26ndash40 (lower) The vertical linescorrespond to oxygen- and magnetite-free reactions 1 5 6 and 7 used to compute Figs 5ndash7 which converge at +30degCThe circles in the lower figure represent values calculated with (ΔHdegf ) ferri-Sud constrained with the chlorite-magnetite-quartz-water equilibria 8ndash20 (open circle) or estimated as ΔHdegf (Mg-sudoite) ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2

(filled circle) (see Table 1 and text)

626 O Vidal et al

was kept unchanged and the Fe3+-Al3+ exchange wasassumed to be restricted to the M4 site of chlorite Thechosen formula of ferri-Sud was thus ([Si2]T1[SiAl]T2)

IV([]M1[Mg2Al2]M2-3[Fe3+]M4)

VIO10(OH)8The standard-state thermodynamic properties of allother chlorite end-members the activity model and theMargules parameters were kept unchanged

(2) The third-law entropy of ferri-sudoite (Sdeg =415 JmolK) was estimated according to Holland(1989) with a configurational entropy of 2305 JmolKand a ferromagnetic contribution of 1304 JmolK itsvolume (Vdeg = 204 Jb) was estimated to be 01 Jblower than that of Mg-sudoite (2051 Jb) and the Cp(T ) function was estimated according to Berman amp

FIG 9 Rock-sample mapping of a Sambagawa sample (a) maps of estimated XFe3+ (Vidal et al 2006 Munoz et al2006) and Log(aO2) (estimated assuming that it is buffered by magnetite-chlorite equilibria see text and Fig 8)(b) measured Fe XANES edge-position proportional to Fe redox (Munoz et al 2006) (c) estimated temperature (Vidal

et al 2006 Munoz et al 2006)

627Conditions of formation of chlorite

Brown (1985) The formation enthalpy (ΔHdegf = 82254 plusmn 5 kJmol) was first estimated as the average valuebetween that calculated according to Chermak ampRimstidt (1989) (minus82204 kJmol) and that(minus82304 kJmol) calculted as ΔHdegf (Mg-sudoite) ndashΔHdegf (corundum)2 + ΔHdegf (hematite)2 using thevalues of Vidal et al (1992) for sudoite and Berman(1988) for hematite

(3) In a first step ΔHdegf and Sdeg of ferri-sudoite wereadjusted in order to reproduce the temperaturesreported in Fig 7 with oxygen-free reactions andassuming that magnetite is present in all samples(Chl-Qtz-Mt assemblage) ΔHdegf was allowed to varybetweenminus8220 andminus8230 kJmol and Sdeg within 10 JmolK of the estimated value The addition ofmagnetite and ferri-sudoite to quartz and water leadsto 17 additional equilibria and the following wereused to constrain ΔHdegf and Sdeg of ferri-sudoite

ClinthornMg-Sud frac14 Mg-Amthorn Qtzthorn H2O (1)

4 Clinthorn Daphthorn 10 ferri-Sud

frac14 10 Mg-Amthorn 25 Qtzthorn 30 Mtthorn 20 H2O(24)

7 Daphthorn 70 ferri-Sudthorn 30 Mg-Am

frac14 22 Clinthorn 75 Mg-Sudthorn 35 Mtthorn 40 H2O(25)

28 ferri-Sudthorn 11 Fe-AmthornMg-Am

frac14 6 Daphthorn 30 Mg-Sudthorn 14 Mtthorn 16 H2O(26)

Daphthorn 10 ferri-Sudthorn 10 Mg-Amthorn 10 Qtz

frac14 6 Clinthorn 15 Mg-Sudthorn 5 Mtthorn 16 H2O(27)

The ferri-Sud properties derived with the assumptionof ChlndashQtzndashMt equilibrium are listed in Table 1(referenced a)With these values the oxygen activity atthe P T and (XFe3+)Chl of all equilibria used tocompute Fig 7 can be calculated with the followingoxygen-bearing and magnetite-free equilibria whichconverge with equilibria 1 8 9 and 10 at the same PTand aO2 conditions

16 Qtzthorn 7 Mg-Amthorn 4 Daphthorn 6 Mg-Sud

thorn 5 O2 thorn 12 H2O frac14 20 ferri-Sud(28)

Fe-Amthorn 2 Mg-Amthorn 6 Qtzthorn 4 H2Othorn O2

frac14 4 ferri-Sud(29)

Mg-Amthorn 4 Daphthorn 6 Mg-Sudthorn 2 O2

frac14 3 Fe-Amthorn 8 ferri-Sudthorn 2 Qtz(30)

5 Mg-Amthorn 12 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 8 Fe-Amthorn 28 ferri-Sud

(31)

8 Daphthorn 12 Mg-Sudthorn 3 O2 frac14 7 Fe-Am

thorn 12 ferri-Sudthorn 10 Qtzthorn 4 H2O(32)

6 Clinthorn 4 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Mg-Amthorn 20 ferri-Sudthorn 5 Qtz(33)

4 Clinthorn 20 ferri-Sud frac14 30 Qtz

thorn 15 Mg-Amthorn 4 Daphthorn 5 O2 thorn 20 H2O(34)

90 Mg-Sudthorn 28 Daphthorn 32 Clinthorn 20 H2Othorn 35 O2

frac14 140 ferri-Sudthorn 15 Mg-Am

(35)

15 Qtzthorn 14 Clinthorn 16 Daphthorn 45 Mg-Sud

thorn 20 O2 thorn 20 H2O frac14 80 ferri-Sud(36)

2 Clinthorn FeAmthorn 3 Mg-Sudthorn O2

frac14 2 Mg-Amthorn 4 ferri-Sudthorn Qtz(37)

12 Clinthorn 7 Fe-Amthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 10 Mg-Amthorn 28 ferri-Sud

(38)

5 Qtzthorn 2 Clinthorn 2 Fe-Amthorn 3 Mg-Sud

thorn 2 O2 thorn 4 H2O frac14 8 ferri-Sud(39)

2 Clinthorn 8 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Fe-Amthorn 20 ferri-Sudthorn 5 Qtz(40)

30 Qtzthorn 8 Clinthorn 15 Fe-Amthorn 5 O2 thorn 20 H2O

frac14 8 Daphthorn 20 ferri-Sud

(41)

4 Clinthorn 8 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 3 Fe-Amthorn 28 ferri-Sud

(42)

Because the ferri-sudoite standard-state propertieswere constrained by the chlorite-quartz-magnetite-O2-H2O equilibrium the Log(aO2) calculated withthe equilibria above follow naturally the trend shown inFig 8 and the Log(aO2) values calculated withreactions 28ndash42 of low-temperature samples plot inthe hematite stability field This suggests that theΔGdegPT values of ferri-sudoite derived above with theassumption that aO2 is buffered by the presence ofmagnetite are possibly too high A reduction of ΔHdegf(ferri-sudoite) to 82304 kJmol (=ΔHdegf (Mg-sudoite)ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2 see Table 1)leads to Log(aO2) reduced by 45 Log units (see Fig 8)

628 O Vidal et al

and most calculated values except those of thehematite-bearing St Martin samples plot in themagnetite stability field

A more detailed study is required to refine the ferri-sudoite thermodynamic data However the resultsobtained using the preliminary values listed in Table 1show that this end-member can be used to reproduceconsistently the T-XFe3+ and T- trends obtained fromclassical Chl-Qtz-H2O calculations (Figs 5ndash7) Theincrease in XFe3+ with reducing temperature is consist-ent with the expected reduction in aO2 and the amountof free oxygen that is consumed to produce Fe3+-bearingchlorite Ferric iron is therefore a major component thathas to be taken into account when dealing with low-temperature samples Empirical or semi-empiricalchlorite thermometers which assume that all Fe is inthe ferrous state may include one of the XFe3+-T andaO2-T trends illustrated in Figs 7 and 8 but they cannotaccount for the different XFe3+-T trends that areexpected in the contexts of different aO2 conditions

FUTURE CHALLENGES LOOK INGAT LOW-GRADE ROCKS SUCH ASMETAMORPH IC ROCKS AND THE

DYNAMICS OF CHLOR ITECRYSTALL IZAT ION AT THE

MICROMETR IC TO NANOMETR ICSCALE

T-P-XFe3+ mapping at the thin-section scaleevidence of chlorite composition re-equilibration in metamorphic and low-gradesamples

Although natural samples offer valuable constraintson thermodynamic models there are two maindifficulties in using them to evaluate the reliability ofexisting chlorite models for thermometric purposes

(1) In the case of low-grade samples referencepressures and temperatures are generally known withina reasonable absolute uncertainty but the size ofchlorite grains is often small and good microprobeanalyses are sometimes difficult to obtain Moreoverthe kinetics of reaction are so sluggish that inheritedchlorite grains might remain unaffected by latervariations in temperature which makes thetemperature-composition relationships difficult toconstrain Lastly extrapolation at HT and HP ofempirical or semi-empirical models constrained byLT samples only may be erroneous in particular theeffect of pressure or eventual systematic variations ofXFe3+ cannot be appraised

(2) In the case of metamorphic samplesmicroprobe analyses are easier to obtain from largechlorite grains but chlorites of contrasted compositionthat crystallized at different temperatures and times ofthe metamorphic history are often observed to coexistmetastably in different microstructural sites of the samethin section In such cases the definition of a singlepair of reference temperature and pressure isinappropriate and cannot be restricted to the peak-temperature conditions

In order to overcome these difficulties and to bettervisualize the link between mineral compositiondeformation and P-T conditions at the thin-sectionscale de Andrade et al (2006 2011) Vidal et al(2006) and Lanari et al (2013) developed an approachwhich combines elementary maps of composition witheither a thermodynamic or a semi-empirical estimate ofchlorite crystallization temperature Maps of rawmicroprobe analyses were standardized using good-quality point analyses of minerals present in themapped area The maps of standardized analyseswere transformed into maps of structural formulaewhere each pixel analysed is converted into individualmineral structural formulae Depending on the thermo-metric approach it was then possible to compute mapsof temperature or T-XFe3+ conditions assuming thateach chlorite pixel was in equilibrium with quartz andwater The results reported by Vidal et al (2006)Yamato et al (2007) Ganne et al (2012) Lanari et al(2013 2014) and Scheffer et al (2015) showed a closelink between deformation and mineral reaction at themicroscopic scale and maps of chlorite temperaturecould be used as a basis for interpreting the micro-structures observed in thin sections of low- to middle-grade metapelites In all these studies the metastablepersistence of different generations of chlorite crystal-lizing at different temperatures during the exhumationof metamorphic samples could be used to constrain theT-deformation history Moreover the map of (XFe3+)Chl calculated by Vidal et al (2006) was in goodqualitative agreement with the in situ mapping of thisratio using the X-ray absorption near-edge structure(XANES) techniques (Fig 9) (see also Munoz et al2006) The calculated temperature and XFe3+ mapsobtained on a single sample showed the same trend ofreducing (XFe3+)Chl with increasing temperature asobserved in Figs 6 and 7

Bourdelle et al (2013c) applied a similar approachfor low-temperature samples with chlorite of smallgrain size using transmission electron microscopy-energy dispersive X-ray (TEM-EDX) spectroscopicanalyses of focused-ion beam (FIB) foils instead of

629Conditions of formation of chlorite

microprobe analyses of thin sections (Fig 10) Thoseauthors were able to observe heterogeneities in XFe3+

at the nm scale using STXM-XANES measurementsThis last study was particularly interesting as it showedthat approaches similar to those used for studyingmetamorphic rocks could be applied to low-grade anddiagenetic samples

Looking at diagenetic rocks like metamorphicrocks

The use of equilibrium thermodynamics to studylow-grade rocks has been often criticized becausesluggish kinetics might preclude achievement of globalequilibrium at low temperature and is responsible forthe crystallization of metastable phases with a lowerenergy of nucleation than stable ones Essene amp Peacor(1995) claimed that the coexistence of chemicallyinhomogeneous phases at a very fine scale in low-grade rocks is indicative of disequilibrium and thatmineral reaction is better described by a kinetic ratherthan thermodynamic approach However the coexist-ence of chemically inhomogeneous phases is also acommon observation in metamorphic rocks wherezoned grains and the coexistence for example ofchlorite grains of different compositions indicate thatthermodynamic equilibrium is approached only locallyand at a given time in the metamorphic history In factthe main difference between diagenetic and higher-temperature rocks is the size of the domains in localequilibrium The more sluggish kinetics of reactionand diffusivity at low temperature lead to small grainsand domains in equilibrium but these domains can beidentified when using appropriate analytical tools andprocedures developed recently (Bourdelle et al2013bc)

FIG 10 Quantitative Fe redox nanomapping on a FIB foilof Glarus sample (T lt 300degC Switzerland) calculatedfrom the 7087ndash71025 eV images ratio (STXM-XANES)

(from Bourdelle et al 2013c)

FIG 11 Analysis of an illite-chlorite assemblage (Gulf Coast sample Alamo 1 well 5825 m) (a) SEM image of thepetrographic thin section showing the emplacement of the FIB foil to be cut across the illitendashchlorite interface (b)Bright-field TEM image of the FIB foil extracted from the section with the platinum strap on top of it (cd) Enlargementof part b showing the Al and Mg contents of chlorite (apfu) and their variation between crystal core and crystal rim

630 O Vidal et al

Figure 10 shows that it is possible to demonstrateheterogeneities in XFe3+ contents at the nanometerscale Consistent variations of other major elements arealso observed with TEM technics on FIB foils Anexample is shown in Fig 11 which depicts thevariations of Al and Mg contents in a chlorite grainfrom the Gulf Coast (Bourdelle et al 2013b) In thiscase the FIB foil was cut parallel to the interfacebetween illite and chlorite and a nm-scale investiga-tion of composition was conducted by TEM-EDXanalysis The rim and core compositions showed azoning of chlorite with Mg- and Al-richer rims thancore compositions This variation in composition on ascale of lt1 μm was related to various crystallizationstages during burial Chlorite sampled at differentdepths and various locations of the Gulf Coast areashowed that zoned compositions similar to thatobserved in Fig 11 are frequent (Bourdelle et al2013b) The AlIV in chlorite rims display a Tdependence that is consistent with the km-scale andregional evolution of temperature ie with thediagenetic history of Gulf Coast (burial P-T condi-tions detritalauthigenic compositions etchellip)(Fig 12) This study suggests again that approachesused for metamorphic rocks can be applied to low-grade rocks when using adapted analytical tools able toidentify sub-μm variations of composition

CONCLUS ION

The variability of chlorite composition reflects thevariability of its conditions of formation Numerousefforts have been made over the last 30 years to

translate the composition of chlorite in terms of itstemperature of formation The first empirical equationsbased on the AlIVor vacancy contents are restricted tobulk-rock compositions and pressure conditions atwhich they were calibrated Thermodynamicapproaches and semi-empirical equations based onthe ratio of end-member activities do not present thislimit Among the different thermodynamic models ofchlorite published so far Vidal et al (2006) proposed amodel that can be used to estimate T and (XFe3+)Chlsimultaneously The estimated (XFe3+)Chl values areconsistent with those measured by Moumlssbauer orXANES spectroscopy and for the same (XFe3+)Chlthe temperatures estimated using the method of Vidalet al (2006) are similar to those obtained with thesemi-empirical equations of Inoue et al (2009) andLanari et al (2014) Different T-XFe3+ trends areobserved for different pressure gradients and oxygen-buffering assemblages (Lanari et al 2014) butoverall the systematic increase in XFe3+ with areduction in temperature is compatible with thedecrease in aO2 in a closed system The increase inXFe3+ and chlorite vacancy with decreasing tempera-ture is interpreted as reflecting the incorporation of theFe3+-sudoite component which can be used toestimate T-aO2-XFe

3+ values with a Chl-Qtz-H2Omulti-equilibrium approach

Similar temperatures are calculated for low-tem-perature chlorites (T lt 300degC and measured XFe3+ upto 30) with all approaches assuming that iron inchlorite is either completely ferrous or with estimatedvalues of XFe3+ This indicates that the equations ofsemi-empirical thermometers assuming that all iron inchlorite is ferrous include some inherent Fe2+Fe3+ vstemperature correction However very different tem-peratures can be obtained for higher-temperaturesamples (gt300degC) with approaches involving ferriciron or assuming that all of the iron is ferrous Thissuggests that the Fe2+Fe3+ vs temperature correctionmight be restricted to particular aO2 conditions andpressure gradients More work is necessary to betterunderstand this point and to improve the thermo-dynamic models and semi-empirical thermometersThis requires high-quality and well constrained naturaldata To this end mapping techniques that highlight thelink between the variations in temperature and pressureand small-scale chlorite composition including XFe3+

variations offer much promise When combining theresults of the present study with those of other studiespublished recently it appears that thermodynamicapproaches and mapping techniques developed formetamorphic rocks could be used to study low-grade

FIG 12 Range of crystal-rim AlIV content (white boxes)of Gulf Coast chlorites vs measured temperatures Thegrey area underlines the range of maximum AlIV contentwhich is assumed to represent the closest approach toequilibrium composition (after Bourdelle et al 2013b)

631Conditions of formation of chlorite

rocks where kinetics are much more sluggish Thisrequires use of appropriate analytical tools and techni-ques with a spatial resolution of just a few hundrednanometres which are now becoming available

ACKNOWLEDGMENTS

Constructive reviews by S Hillier F Nieto and ananonymous reviewer are gratefully acknowledged

REFERENCES

Beaufort D Patrier P Meunier A amp Ottaviani MM(1992) Chemical variations in assemblages includingepidote andor chlorite in the fossil hydrothermalsystem of Saint Martin (Lesser Antilles) Journal ofVolcanology and Geothermal Research 51 95ndash114

Berman R (1988) Internally-consistent thermodynamicdata for minerals in the system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2Journal of Petrology 29 445ndash522

Berman R amp Brown T (1985) Heat capacity of minerals inthe system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2 representation estima-tion and high temperature extrapolation Contributionsto Mineralogy and Petrology 89 168ndash183

Bourdelle F amp Cathelineau M (2015) Low-temperaturechlorite geothermometry a graphical representationbased on a TndashR2+ndashSi diagram European Journal ofMineralogy 27 617ndash626

Bourdelle F Parra T Chopin C amp Beyssac O (2013a) Anew chlorite geothermometer for diagenetic to low-grade metamorphic conditions Contributions toMineralogy and Petrology 165 723ndash735

Bourdelle F Parra T Beyssac O Chopin C amp Vidal O(2013b) Clay minerals thermometry a comparativestudy based on high-resolution chemical analyses ofillite and chlorite in sandstones from the Gulf Coast(Texas USA) American Mineralogist 98 914ndash926

Bourdelle F Benzerara K Beyssac O Cosmidis JNeuville DR Brown GE amp Paineau E (2013c)Quantification of the ferric ferrous iron ratio insilicates by scanning transmission X-ray microscopyat the Fe L23 edges Contribution to Mineralogy andPetrology 166 423ndash434

Cathelineau M (1988) Cation site occupancy in chloritesand illites as function of temperature Clay Minerals23 471ndash485

Cathelineau M amp Nieva D (1985) A chlorite solidsolution geothermometer the Los Azufres (Mexico)geothermal system Contributions to Mineralogy andPetrology 91 235ndash244

Chermak JA amp Rimstidt JD (1989) Estimating thethermodynamic properties (ΔGdegf and ΔHdegf) of silicate

minerals at 298K from the sum of polyhedral contribu-tions American Mineralogist 74 1023ndash1031

Coggon R amp Holland TJB (2002) Mixing properties ofphengitic micas and revised garnent-phengite thermo-barometers Journal of Metamorphic Geology 20683ndash696

Curtis CD Hughes CR Whiteman JA ampWhittle CK(1985) Compositional variation within some sedi-mentary chlorites and some comments on their originMineralogical Magazine 49 375ndash386

de Andrade V Vidal O Lewin E OrsquoBrien P amp Agard P(2006) Quantification of electron microprobe com-positional maps of rock thin sections an optimizedmethod and examples Journal of MetamorphicGeology 24 655ndash668

de Andrade V Susini J Salomeacute M Beraldin O RigaultC Heymes T Lewin E amp Vidal O (2011)Submicrometer hyperspectral X-ray imaging of het-erogeneous rocks and geomaterials applications at theFe k-edge Analytical Chemistry 83 4220ndash4227

de Caritat P Hutcheon I amp Walshe J (1993) Chloritegeothermometry a review Clays and Clay Minerals41 219ndash239

Essene EJ (2009) Thermobarometry gone astray Pp101ndash129 in Physics and Chemistry of Earthrsquos Interior(AK Gupta amp S Dasgupta editors) PlatinumJubilee Indian National Science Academy Springer-Verlag Berlin

Essene EJ amp Peacor DR (1995) Clay mineralthermometry ndash a critical perspective Clays and ClayMinerals 43 540ndash553

Ganne J de Andrade V Weinberg RF Vidal O DubacqB Kagambega N Naba S Baratoux L Jessell M ampAllibon J (2012) Modern-style plate subductionpreserved in the Palaeoproterozoic West Africancraton Nature Geoscience 5 60ndash65

Grosch EG Vidal O Abu-Alam T amp McLoughlin N(2012) PTconstraints on the metamorphic evolution ofthe Paleoarchean Kromberg type-section BarbertonGreenstone Belt South-Africa Journal of Petrology53 513ndash545

Grosch E McLoughlin N Lanari P Erambert M ampVidal O (2014) Microscale mapping of alterationconditions and potential biosignatures in basaltic-ultramafic rocks on early Earth and beyondAstrobiology 14 216ndash228

Hillier S amp Velde B (1991) Octahedral occupancy andchemical composition of diagenetic (low-temperature)chlorites Clay Minerals 26 149ndash168

Holland TJB (1989) Dependance of entropy on volumefor silicates and oxide minerals A review and apredictive model American Mineralogist 74 5ndash13

Holland T amp Powell R (1998) An internally consistentthermodynamic data set for phases of petrologicalinterest Journal ofMetamorphic Geology 16 309ndash343

Holland T Baker J amp Powell R (1998)Mixing propertiesand activity composition and relationships of chlorites

632 O Vidal et al

not allow us to fit the experimental data of chloritecomposition reported by Parra et al (2005) in theFASH system (Fig 4)

Whatever the thermodynamic model at least onetemperature of chlorite-quartz-water equilibrium can becalculated at fixed pressure with assumptions regardingthe proportion of ferric iron in chlorite This calculationinvolves evaluation of equilibrium constants expressedas a ratio of chlorite end-member activities andtherefore cation proportions It follows that chloritesof different compositions (and different AlIV contents)crystallizing at identical P-T conditions but in rocks of

different compositions can have identical equilibriumconstants of (1) and (8) to (10) Inoue et al (2009)Bourdelle et al (2013a) and Lanari et al (2014)proposed several semi-empirical chlorite thermometersalso involving equilibrium constants calculated as theratio of chlorite end-member activities The maindifference compared to the thermodynamic approachis a simplification of the Gibbs free energy equationactivity models and assumptions on the ferric ironcontent in chlorite These semi-empirical thermometersare easy to use and they are believed to reproducesuccessfully temperatures estimated independently in

TABLE 1 Standard-state properties of chlorite end-members and Margules parameters of the chlorite solid solution (seeBerman 1988 Vidal et al 2006 for the format)

Mg-sudoite (Vidal et al 1992 2006) [Si2]T1[Si Al]T2)IV([A]M1[Mg2 Al2]M2[Al]M4)

VIO10(OH)8ST minus86552730 39050 20510C1 111584 minus85075 minus19777440 2959147000V1 264515406 00 minus018194707 00Ferri-sudoite (present study) [Si2]T1[Si Al]T2)

IV([A]M1[Mg2 Al2]M2plusmn3[Fe3thorn]M4)

VIO10(OH)8ST minus8221273a 40550 20400

minus8230400b

C1 112225 minus85695 minus17431790 2552588000V1 264515406 00 minus018194707 00Daphnite (Vidal et al 2001 2006) [Si2]T1[Si Al]T2)

IV([Fe2thorn]M1[Fe2thorn4 ]M2plusmn3[Al]M4)

VIO10(OH)8ST minus7120845 55940 21600C1 1229233 minus102565 minus12276900 2121510000V1 264515406 00 minus018194707 00Fe-amesite (Vidal et al 2005 2006) [Si2]T1[Al2]T2)

IV([Al]M1[Fe2thorn4 ]M2plusmn3[Al]M4)

VIO10(OH)8ST minus76074600 51480 20900C1 121981 minus100215 minus14439720 2392326000V1 264515406 00 minus018194707 00Clinochlore (Berman 1988) [Si2]T1[Si Al]T2)

IV([Mg]M1[Mg4]M2plusmn3[Al]M4)VIO10(OH)8

ST minus890958984 43515 21147C1 1214284 minus1121713 0000 minus1256253184V1 264515406 00 minus018194707 00Mg-amesite (Vidal et al 2001 2006) [Si2]T1[Al2]T2)

IV([Al]M1[Mg4]M2plusmn3[Al]M4)VIO10(OH)8

ST minus90359005 40320 20520C1 114445 minus83272 minus20058040 2820786000V1 264515406 00 minus018194707 00Margules parameters on site M1 (Vidal et al 2006)WMg-Al minus9400 minus30 minus02WMg- 10000 minus25 09WFe-Al 1200 31 07WFe- 2000 minus15 04WAl- minus10000 minus30 09

ST ΔHdegf (J) Sdeg(JK) Vdeg(Jb) Cp K0 K1 K2 K3 (Cp = K0 + K1Tminus05 + K2Tminus2 + K3Tminus3) V(PT) V1 V2 V3 V4 (V(PT)V(1298) = 1 + V1(Tminus 298) + V2(Tminus 298)2 + V3(Pminus1) + V4(Pminus 1)2)a estimated using equilibria 1 and 21ndash24 b estimated as (ΔHdegf (Mg-sudoite)minus ΔHdegf (corundum)2 + ΔHdegf (hematite)2)The Margules parameters list the WH WS and Wv values (W =WH + T(K)WS + (Pminus 1)Wv)

620 O Vidal et al

low-temperature series (Inoue et al 2009 Bourdelleet al 2013a Lanari et al 2014) and metamorphicrocks (Lanari et al 2014) The weakness of the semi-empirical equations is that their compatibility withexperimental data obtained in MASH and FASHsystems has not always been checked and they arerestricted to quartz-bearing rocks

Finally the main difference between the thermo-dynamic approach of Vidal et al (2005 2006) andthose of Vidal et al (2001) Lanari et al (2014) or thesemi-empirical thermometers of Inoue et al (2009)

Bourdelle et al (2013a) and Lanari et al (2014) is thatthe former leads to the estimation of T-(XFe3+)Chl pairsat given pressure whereas the latter estimate T atassumed or measured XFe3+ and pressure or assuminga negligible role of pressure This last assumption isacceptable at low-pressure conditions but becomesproblematic for metamorphic samples because theamount of Mg-Sud end-member at fixed temperaturehas been shown to increase with pressure (Fig 3b)

A comparison of the approaches of Vidal et al(2006) Lanari et al (2014) and Inoue et al (2009) is

FIG 4 Comparison of measured and calculated Si content of chlorite in the FASH systemwith almandine and fayalite (a)or chloritoid and hercynite (b) Boxes experimental data of Parra et al (2005) continuous lines Si content calculatedwith the thermodynamic properties of Fe-amesite and daphnite and the solid solution model of Vidal et al (2005 2006)dashed lines Si content calculated with the Holland et al (1998) model updated by Coggon amp Holland (2002) which

assumes that the thermodynamic properties of Fe-Am depend on those of Clin Mg-Am and Daph

621Conditions of formation of chlorite

shown in Fig 5 which reports temperatures estimatedfor the set of 1541 analyses used in the present studyThe temperatures calculated with the assumption Fe3+

= 0 (equation 38 of Lanari et al 2014) or Fe3+ gt 0 (allother approaches) are labelled TFe2+ and TFe3+respectively in Fig 5 and in the following Allcalculations were conducted with a MATLAB scriptwith a GUI interface available on request from the firstauthor

The temperature conditions estimated for blueschist-facies metamorphic samples with the approach ofVidal et al (2006) (TFe3+ Vidal) cover a wide range oftemperature from 275 to 550degC (Fig 5a) About onethird of the samples show temperatures of lt350degCwhich is 50ndash100degC lower than the published peaktemperatures derived for the samples studied Usingthe XFe3+ estimated with the Vidal et al (2006)approach the temperatures estimated with the semi-empirical equations of Inoue et al (2009) (TFe3+

Inoue) or Lanari et al (2014) (TFe3+ Lanari) show asimilar evolution but the estimated temperatures are50ndash100degC lower in the case of Inoue et al (2009) andhigher in the case of Lanari et al (2014) The mostimportant deviation is observed between TFe2+ Lanariand the other temperature estimates A significantproportion of chlorites calculated to have crystallized atTFe3+asymp 300degC give much higher TFe2+Lanari (gt500degCopen symbols in Fig 5a) and plot on a different trendThe same observation is made for the Barbertonsamples from Grosch et al (2012) (Fig 5c) In bothcases the maximum TFe2+ Lanari can be as high as900degC but the highest temperatures are obtained forsamples containing very small amounts of vacancies(lt005 pfu open circles below the dashed line in thetemperature-vacancy plot shown in Fig 5a) The Mg-Sud activity and temperatures derived from theseanalyses are very sensitive to small variations incomposition and Fe3+-content and they should not beused for reliable thermometric estimates In contrastFig 5b shows that similar temperatures are estimatedwith all approaches for low-pressure and low-gradesamples

This comparison indicates that similar results areobtained with all methods at T lt 300degC and similartrends are observed at higher temperature for the threemethods allowing the incorporation of Fe3+ althoughwith a significant scatter at T gt 400degC With the XFe3+

values derived using the Vidal et al (2006) approachTFe3+ Vidal and TFe3+ Inoue are quite consistentHowever the highest TFe3+ Inoue estimated forblueschist metamorphic samples are lower whereasTFe3+ Lanari are higher In all approaches the

temperature calculated depends heavily on thesudoite activity which is proportional to the amountof vacancy in chlorite (a(Sud) = 64 (XSi XAl)T2 XM1 (XMg)2M2ndash3 (XAl)2M2ndash3 (XAl)M4))This is illustrated in Fig 5 which shows that theamount of vacancy in chlorite increases with a drop intemperature As noted by Laird (1988) the amount ofvacancy in chlorite structural formulae calculated on acharge-normalized 14 anhydrous-oxygen basisincreases with increasing XFe3+ It is therefore notsurprising that like vacancies XFe3+ increases withreducing temperature This trend is illustrated in Fig 6where the TFe3+Vidal vs XFe3+ values of samples usedto compute Fig 5 have been plotted Figure 6 alsoindicates that larger XFe3+ values are calculated for thehigh-pressure samples than for the low-pressuresamples Nevertheless vacancies are also noted toincreasewith reducing Tunder the assumption that Fe2+

= FeTotal (see Fig 3b and the open circles in Fig 5ab)Therefore the increase in vacancies with reducingtemperature is not just a mathematical artefact resultingfrom the calculation of Fe3+-bearing structural formulaeon a 14-oxygen basis Interestingly even at lowtemperature where XFe3+ calculated with the Vidalet al (2006) approach is at a maximum TFe2+ Lanari issimilar to the three TFe3+ (Fig 5b) This indicates thatthe TFe2+ Lanari equation includes the systematicvariation of XFe3+ with temperature observed inFig 6 which explains why similar TFe2+ and TFe3+

are obtained at low T and P (Fig 5b) for contrastingamounts of vacancies and ferric iron contentsNevertheless it is evident from Fig 5ac that the XFe3+ vs temperature relationship that was used to derive theTFe2+ Lanari equation is not fully consistent with the T-XFe3+ relationship obtained with the thermodynamicapproach of Vidal et al (2006) This is illustrated by thedivergence between TFe2+ Lanari and the three TFe3+

calculated for the low-pressure Barberton samples(Fig 5c) and several HP samples Fig 5a

COMPARISON OF EST IMATED ANDMEASURED (XFE 3 + )CHL

At this stage we can only conclude from thecomparison of TFe2+ and TFe3+ in Fig 5ac thateither some XFe3+ values derived with the Vidal et al(2006) approach are overestimated and the low TFe3+

obtained for some HP-HT samples are underestimatedor that TFe2+Lanari only pertains for fO2

conditionsbuffered by specific assemblages An overestimationof XFe3+ due to errors in the standard-state thermo-dynamic properties and the non-ideal Margules

622 O Vidal et al

FIG 5 Left Plot of temperatures estimated with equations 38 (TFe3+ Lanari) and 40 (TFe2+ Lanari) of Lanari et al(2014) and with the thermometer of Inoue et al (2009) (TFe3+ Inoue) vs temperatures estimated using the approach ofVidal et al (2006) TFe3+ Lanari and TFe3+ Inoue were calculated using the XFe3+ values obtained with the approach ofVidal et al (2006) and TFe2+ Lanari assuming XFe3+ = 0 Right Plot of vacancy content vs estimated temperatures(a) and (b) are high-pressure (gt6 kbar) and low- to medium-pressure (lt6 kbar) respectively (c) shows the resultsobtained for the low-pressure samples of Grosch et al (2012) The unrealistically high TFe2+ Lanari are obtained forchlorites with lt005 vacancy pfu (dashed line in the vacancy-temperature plots) The histograms show the repartition

of temperatures estimated with the different approaches

623Conditions of formation of chlorite

parameters of Vidal et al (2006) is possible butthis should be demonstrated by a systematic compari-son of measured and estimated XFe3+ from naturalsamples covering a wide range of P T and fO2

conditions Such a study is still lacking but a first-order comparison of measured and estimated XFe3+

contents for samples of known temperature ofcrystallization is shown in Fig 7 which was obtainedfor the Rouez Niger and Saint-Martin low-T samplescompiled by Inoue et al (2009) the Glarus sample ofBourdelle et al (2013c) and a blueschist-facies sampleretrogressed into the greeschist facies from Sambagawa(de Andrade et al 2006 Munoz et al 2006 Vidalet al 2006) T-XFe3+ values were calculated at 2 kbarexcept for the Sambagawa sample for which calcula-tions were made at 2 5 and 10 kbar for the low-medium- and high-temperature chlorite groups identi-fied by de Andrade et al (2006) and Vidal et al (2006)respectively

FIG 6 Estimated TFe3+ Vidal and XFe3+ pairs for low- andmedium-pressure samples (open circles) high-pressuresamples (black circles) and the samples from Grosch et al

(2012) (grey squares)

FIG 7 Boxes measured XFe3+ and reference temperatures Symbols Tmaximum-XFe3+minimum pairs calculated to

obtain convergence (plusmn30degC) of equilibria 1 8 9 and 10 with the thermodynamic model of Vidal et al (2006)

624 O Vidal et al

For each series of Inoue et al (2009) between 50and 60 of the chlorite analyses contain Si lt3 apfuand led to a convergence of equations 1 8 9 and 10within 30degC The calculated T-XFe3+ pairs are plottedin Fig 7 and compared with the range of referencetemperature (TRef) estimated independently and XFe3+

measures reported by Inoue et al (2009) A fairagreement of the measured and calculated TFe3+ Vidal-XFe3+ values is observed for the Niger and St-Martinanalyses and a good agreement of the TRef and TFe3+ Vidal was observed for the Rouez samples but thecalculated valueswere smaller than themeasured valuesfor XFe3+ for this samples However the XFe3+ valuesreported in Fig 7 are minimum values required toobtain convergence of equilibria 1 and 8ndash10 within30degC Convergence is maintained at higher valuesof XFe3+ and TFe3+ Vidal for the Rouez samples arestill located in the range of TRef when XFe3+ is setat gt15 (lower limit of the reference values) TheTFe3+ Vidal values calculated for the Glarus chloriteanalyses listed in Bourdelle et al (2013c) are 139degC 203degC and 224degC in fair agreement with thoseestimated independently by Bourdelle et al (2013c)(135degC 170degC and 182degC respectively) or with thesemi-empirical thermometer of Inoue et al (2009)(145degC 196degC and 210degC respectively) For thethree samples studied the XFe3+ values measuredusing XANES (X-ray absorption near edge struc-ture) spectra in the transmission mode of scanningtransmission X-ray microscopy (STXM) were foundto be 223 238 and 279 These values comparefavourably with the estimated ones (32 27 and 28respectively) although the measured values suggestan increase in XFe3+ with temperature which iscontrary the trend inferred from the thermodynamicestimates and evidenced in Figs 6 and 7 Finally therange of XFe3+ estimated for the different generationsof chlorite in the Sambagawa sample (ellipses inFig 7) are slightly higher but still in fair agreementwith those inferred from the XANES measurementsIn conclusion Fig 7 shows that the values estimatedon the basis of the convergence of equilibria (1 8 9and 10) are consistent with the TRef and measuredXFe3+ for various samples covering a range oftemperature from 100degC to 500degC pressure from sim2to 10 kbar and XFe3+ values from 5 to 35 Boththe measured and estimated XFe3+ values plotted inFig 7 seem to confirm the systematic decrease ofXFe3+ with increasing temperature and increase withincreasing pressure that was observed in Fig 6 Thistrend is similar to the decrease in vacancy observedin Fig 5

aO2 -TEMPERATURE EST IMATESAND ADDIT ION OF A FERR I -

SUDOITE END -MEMBER

The general trends inferred from our compilation ofnatural samples indicate that XFe3+ increases withdecreasing temperature and increasing pressureAlthough the model of Vidal et al (2006) involvesno Fe3+-bearing end-member the oxygen activity(aO2) prevailing during the formation of chloritesused to compute Fig 7 can be estimated assuming thatit was buffered by the presence of pure magnetite (Mt)in equilibrium with chlorite and quartz Following thisassumption aO2 for each sample has been calculatedusing equilibria (1 8 9 10) and the followingadditional equilibria which intersect at the same P T(plusmn30degC) (XFe3+)Chl and Log(aO2) (plusmn2) (Fig 8)

15 Mg-Sudthorn 40 Mtthorn 45 Qtzthorn 60 H2O

frac14 6 Clinthorn 24 Daphthorn 20 O2

(11)

6 Mg-Sudthorn 10 Mtthorn 6 Qtzthorn 12 H2O

frac14 3 Mg-Amthorn 6 Daphthorn 5 O2

(12)

15 Mg-Sudthorn 6 Clinthorn 10 Mt

frac14 15 Qtzthorn 15 Mg-Amthorn 6 Daphthorn 5 O2

(13)

6 Daphthorn 4 Clinthorn 5 O2

frac14 20 Qtzthorn 10 Mtthorn 5 Mg-Amthorn 20 H2O(14)

60 Mg-Sudthorn 12 Clinthorn 70 Mtthorn 60 H2O

frac14 45 Mg-Amthorn 42 Daphthorn 35 O2

(15)

4 Qtzthorn 2 Mtthorn Fe-Amthorn 4 H2O

frac14 2 Daphthorn O2

(16)

15 Mg-Sudthorn 6 Daphthorn 10 Mt

frac14 15 Qtzthorn 6 Clinthorn 15 Fe-Amthorn 5 O2

(17)

60 Mg-Sudthorn 70 Mtthorn 60 H2O

frac14 24 Clinthorn 45 Fe-Amthorn 6 Daphthorn 35 O2

(18)

15 Mg-Sudthorn 16 Mtthorn 12 H2O

frac14 3 Qtzthorn 6 Clinthorn 12 Fe-Amthorn 8 O2

(19)

6 Mg-Sudthorn 4 Mt

frac14 6 Qtzthorn 3 Mg-Amthorn 3 Fe-Amthorn 2 O2

(20)

12 Mg-Sudthorn 14 Mtthorn 12 H2O

frac14 6 Mg-Amthorn 3 Fe-Amthorn 6 Daphthorn 7 O2

(21)

5 Mg-Amthorn 4 Mtthorn 8 Qtzthorn 8 H2O

frac14 4 Clinthorn 3 Fe-Amthorn 2 O2

(22)

24 Mg-Sudthorn 3 Mg-Amthorn 28 Mtthorn 24 H2O

frac14 12 Clinthorn 21 Fe-Amthorn 14 O2

(23)

All calculated Log(aO2) except those of the Saint-

625Conditions of formation of chlorite

Martin samples follow the same trend and decreasewith decreasing temperature from simminus27 at 450ndash500degC10 kbar (high-temperature chlorites of Sambagawa) tominus50degC at 125degC and 1 kbar (Niger samples) (Fig 8)The Log(aO2) values of the high-T and -P samples plotin the stability field of magnetite whereas those of thelow-T samples plot in the stability field of hematitewhich is not consistent with the assumption of ChlndashMtndashQtzndashH2O equilibrium The presence of hematite hasbeen reported in the Saint-Martin samples only(Beaufort et al 1992) which plot above the trenddefined by the other samples (upper grey ellipse inFig 8) The relative position of this hematite-bearingsample is consistent with expected higher oxygenactivities required for its formation For the other low-T and apparently hematite-free samples it is possiblethat the calculated Log(aO2) values are too high or thatoxygen was not buffered by an equilibrium ofmagnetitewith chlorite In any case Fig 8 indicates that thesystematic increase in (XFe3+)Chl with decreasingtemperature observed in Figs 6 and 7 is consistentwith a lower aO2 and amount of free oxygen at low-Tthan high-T in systems where aO2 is buffered by thepresence of chlorite and iron oxides To maintain this

equilibrium with lowering temperature in metamorphicrocks where oxygen is immobile oxygen has to beconsumed and one way for that is to produce chloritethat is richer in Fe3+ This is a possible interpretationfor the formation of low-temperature Fe3+-rich chloritecoexisting with high-temperature Fe3+-poor chloritein the same thin section of the Sambagwa sample(Fig 9)

In order to discuss the Log(aO2)-T-XFe3+ evolution

in more detail for rocks where aO2 is not buffered bythe presence of magnetite it is necessary to define anFe3+-bearing chlorite end-member The simultaneousincrease of vacancy and Fe3+ content with decreasingtemperature observed in Figs 5ndash7 is explained by the3(Mg Fe2+) = 2(Fe3+) ditrioctahedral substitution(see also Trincal amp Lanari 2016 this issue) Thisexchange can be modelled with a ferri-sudoite end-member that materializes the relationship betweenXFe3+ and vacancy The standard-state thermo-dynamic properties of ferri-sudoite were constrainedas follows

(1) Because the maximum amount of (Fe3+)Chlcalculated in the previous section never exceeded1 apfu the cation distribution of Vidal et al (2006)

FIG 8 Left Ranges of Log(aO2) conditions calculated for the samples used to compute Fig 7 at the temperature andXFe3+ contents estimated from the Chl-Qtz-H2O equilibrium conditions Dashed lines hematite-magnetite-O2

equilibrium calculated using the thermodynamic properties of Berman (1988) and unit activities of Ht and Mt RightSample calculation at 2 kbar for a Rouez sample using reactions 8ndash20 (upper) or 26ndash40 (lower) The vertical linescorrespond to oxygen- and magnetite-free reactions 1 5 6 and 7 used to compute Figs 5ndash7 which converge at +30degCThe circles in the lower figure represent values calculated with (ΔHdegf ) ferri-Sud constrained with the chlorite-magnetite-quartz-water equilibria 8ndash20 (open circle) or estimated as ΔHdegf (Mg-sudoite) ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2

(filled circle) (see Table 1 and text)

626 O Vidal et al

was kept unchanged and the Fe3+-Al3+ exchange wasassumed to be restricted to the M4 site of chlorite Thechosen formula of ferri-Sud was thus ([Si2]T1[SiAl]T2)

IV([]M1[Mg2Al2]M2-3[Fe3+]M4)

VIO10(OH)8The standard-state thermodynamic properties of allother chlorite end-members the activity model and theMargules parameters were kept unchanged

(2) The third-law entropy of ferri-sudoite (Sdeg =415 JmolK) was estimated according to Holland(1989) with a configurational entropy of 2305 JmolKand a ferromagnetic contribution of 1304 JmolK itsvolume (Vdeg = 204 Jb) was estimated to be 01 Jblower than that of Mg-sudoite (2051 Jb) and the Cp(T ) function was estimated according to Berman amp

FIG 9 Rock-sample mapping of a Sambagawa sample (a) maps of estimated XFe3+ (Vidal et al 2006 Munoz et al2006) and Log(aO2) (estimated assuming that it is buffered by magnetite-chlorite equilibria see text and Fig 8)(b) measured Fe XANES edge-position proportional to Fe redox (Munoz et al 2006) (c) estimated temperature (Vidal

et al 2006 Munoz et al 2006)

627Conditions of formation of chlorite

Brown (1985) The formation enthalpy (ΔHdegf = 82254 plusmn 5 kJmol) was first estimated as the average valuebetween that calculated according to Chermak ampRimstidt (1989) (minus82204 kJmol) and that(minus82304 kJmol) calculted as ΔHdegf (Mg-sudoite) ndashΔHdegf (corundum)2 + ΔHdegf (hematite)2 using thevalues of Vidal et al (1992) for sudoite and Berman(1988) for hematite

(3) In a first step ΔHdegf and Sdeg of ferri-sudoite wereadjusted in order to reproduce the temperaturesreported in Fig 7 with oxygen-free reactions andassuming that magnetite is present in all samples(Chl-Qtz-Mt assemblage) ΔHdegf was allowed to varybetweenminus8220 andminus8230 kJmol and Sdeg within 10 JmolK of the estimated value The addition ofmagnetite and ferri-sudoite to quartz and water leadsto 17 additional equilibria and the following wereused to constrain ΔHdegf and Sdeg of ferri-sudoite

ClinthornMg-Sud frac14 Mg-Amthorn Qtzthorn H2O (1)

4 Clinthorn Daphthorn 10 ferri-Sud

frac14 10 Mg-Amthorn 25 Qtzthorn 30 Mtthorn 20 H2O(24)

7 Daphthorn 70 ferri-Sudthorn 30 Mg-Am

frac14 22 Clinthorn 75 Mg-Sudthorn 35 Mtthorn 40 H2O(25)

28 ferri-Sudthorn 11 Fe-AmthornMg-Am

frac14 6 Daphthorn 30 Mg-Sudthorn 14 Mtthorn 16 H2O(26)

Daphthorn 10 ferri-Sudthorn 10 Mg-Amthorn 10 Qtz

frac14 6 Clinthorn 15 Mg-Sudthorn 5 Mtthorn 16 H2O(27)

The ferri-Sud properties derived with the assumptionof ChlndashQtzndashMt equilibrium are listed in Table 1(referenced a)With these values the oxygen activity atthe P T and (XFe3+)Chl of all equilibria used tocompute Fig 7 can be calculated with the followingoxygen-bearing and magnetite-free equilibria whichconverge with equilibria 1 8 9 and 10 at the same PTand aO2 conditions

16 Qtzthorn 7 Mg-Amthorn 4 Daphthorn 6 Mg-Sud

thorn 5 O2 thorn 12 H2O frac14 20 ferri-Sud(28)

Fe-Amthorn 2 Mg-Amthorn 6 Qtzthorn 4 H2Othorn O2

frac14 4 ferri-Sud(29)

Mg-Amthorn 4 Daphthorn 6 Mg-Sudthorn 2 O2

frac14 3 Fe-Amthorn 8 ferri-Sudthorn 2 Qtz(30)

5 Mg-Amthorn 12 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 8 Fe-Amthorn 28 ferri-Sud

(31)

8 Daphthorn 12 Mg-Sudthorn 3 O2 frac14 7 Fe-Am

thorn 12 ferri-Sudthorn 10 Qtzthorn 4 H2O(32)

6 Clinthorn 4 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Mg-Amthorn 20 ferri-Sudthorn 5 Qtz(33)

4 Clinthorn 20 ferri-Sud frac14 30 Qtz

thorn 15 Mg-Amthorn 4 Daphthorn 5 O2 thorn 20 H2O(34)

90 Mg-Sudthorn 28 Daphthorn 32 Clinthorn 20 H2Othorn 35 O2

frac14 140 ferri-Sudthorn 15 Mg-Am

(35)

15 Qtzthorn 14 Clinthorn 16 Daphthorn 45 Mg-Sud

thorn 20 O2 thorn 20 H2O frac14 80 ferri-Sud(36)

2 Clinthorn FeAmthorn 3 Mg-Sudthorn O2

frac14 2 Mg-Amthorn 4 ferri-Sudthorn Qtz(37)

12 Clinthorn 7 Fe-Amthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 10 Mg-Amthorn 28 ferri-Sud

(38)

5 Qtzthorn 2 Clinthorn 2 Fe-Amthorn 3 Mg-Sud

thorn 2 O2 thorn 4 H2O frac14 8 ferri-Sud(39)

2 Clinthorn 8 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Fe-Amthorn 20 ferri-Sudthorn 5 Qtz(40)

30 Qtzthorn 8 Clinthorn 15 Fe-Amthorn 5 O2 thorn 20 H2O

frac14 8 Daphthorn 20 ferri-Sud

(41)

4 Clinthorn 8 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 3 Fe-Amthorn 28 ferri-Sud

(42)

Because the ferri-sudoite standard-state propertieswere constrained by the chlorite-quartz-magnetite-O2-H2O equilibrium the Log(aO2) calculated withthe equilibria above follow naturally the trend shown inFig 8 and the Log(aO2) values calculated withreactions 28ndash42 of low-temperature samples plot inthe hematite stability field This suggests that theΔGdegPT values of ferri-sudoite derived above with theassumption that aO2 is buffered by the presence ofmagnetite are possibly too high A reduction of ΔHdegf(ferri-sudoite) to 82304 kJmol (=ΔHdegf (Mg-sudoite)ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2 see Table 1)leads to Log(aO2) reduced by 45 Log units (see Fig 8)

628 O Vidal et al

and most calculated values except those of thehematite-bearing St Martin samples plot in themagnetite stability field

A more detailed study is required to refine the ferri-sudoite thermodynamic data However the resultsobtained using the preliminary values listed in Table 1show that this end-member can be used to reproduceconsistently the T-XFe3+ and T- trends obtained fromclassical Chl-Qtz-H2O calculations (Figs 5ndash7) Theincrease in XFe3+ with reducing temperature is consist-ent with the expected reduction in aO2 and the amountof free oxygen that is consumed to produce Fe3+-bearingchlorite Ferric iron is therefore a major component thathas to be taken into account when dealing with low-temperature samples Empirical or semi-empiricalchlorite thermometers which assume that all Fe is inthe ferrous state may include one of the XFe3+-T andaO2-T trends illustrated in Figs 7 and 8 but they cannotaccount for the different XFe3+-T trends that areexpected in the contexts of different aO2 conditions

FUTURE CHALLENGES LOOK INGAT LOW-GRADE ROCKS SUCH ASMETAMORPH IC ROCKS AND THE

DYNAMICS OF CHLOR ITECRYSTALL IZAT ION AT THE

MICROMETR IC TO NANOMETR ICSCALE

T-P-XFe3+ mapping at the thin-section scaleevidence of chlorite composition re-equilibration in metamorphic and low-gradesamples

Although natural samples offer valuable constraintson thermodynamic models there are two maindifficulties in using them to evaluate the reliability ofexisting chlorite models for thermometric purposes

(1) In the case of low-grade samples referencepressures and temperatures are generally known withina reasonable absolute uncertainty but the size ofchlorite grains is often small and good microprobeanalyses are sometimes difficult to obtain Moreoverthe kinetics of reaction are so sluggish that inheritedchlorite grains might remain unaffected by latervariations in temperature which makes thetemperature-composition relationships difficult toconstrain Lastly extrapolation at HT and HP ofempirical or semi-empirical models constrained byLT samples only may be erroneous in particular theeffect of pressure or eventual systematic variations ofXFe3+ cannot be appraised

(2) In the case of metamorphic samplesmicroprobe analyses are easier to obtain from largechlorite grains but chlorites of contrasted compositionthat crystallized at different temperatures and times ofthe metamorphic history are often observed to coexistmetastably in different microstructural sites of the samethin section In such cases the definition of a singlepair of reference temperature and pressure isinappropriate and cannot be restricted to the peak-temperature conditions

In order to overcome these difficulties and to bettervisualize the link between mineral compositiondeformation and P-T conditions at the thin-sectionscale de Andrade et al (2006 2011) Vidal et al(2006) and Lanari et al (2013) developed an approachwhich combines elementary maps of composition witheither a thermodynamic or a semi-empirical estimate ofchlorite crystallization temperature Maps of rawmicroprobe analyses were standardized using good-quality point analyses of minerals present in themapped area The maps of standardized analyseswere transformed into maps of structural formulaewhere each pixel analysed is converted into individualmineral structural formulae Depending on the thermo-metric approach it was then possible to compute mapsof temperature or T-XFe3+ conditions assuming thateach chlorite pixel was in equilibrium with quartz andwater The results reported by Vidal et al (2006)Yamato et al (2007) Ganne et al (2012) Lanari et al(2013 2014) and Scheffer et al (2015) showed a closelink between deformation and mineral reaction at themicroscopic scale and maps of chlorite temperaturecould be used as a basis for interpreting the micro-structures observed in thin sections of low- to middle-grade metapelites In all these studies the metastablepersistence of different generations of chlorite crystal-lizing at different temperatures during the exhumationof metamorphic samples could be used to constrain theT-deformation history Moreover the map of (XFe3+)Chl calculated by Vidal et al (2006) was in goodqualitative agreement with the in situ mapping of thisratio using the X-ray absorption near-edge structure(XANES) techniques (Fig 9) (see also Munoz et al2006) The calculated temperature and XFe3+ mapsobtained on a single sample showed the same trend ofreducing (XFe3+)Chl with increasing temperature asobserved in Figs 6 and 7

Bourdelle et al (2013c) applied a similar approachfor low-temperature samples with chlorite of smallgrain size using transmission electron microscopy-energy dispersive X-ray (TEM-EDX) spectroscopicanalyses of focused-ion beam (FIB) foils instead of

629Conditions of formation of chlorite

microprobe analyses of thin sections (Fig 10) Thoseauthors were able to observe heterogeneities in XFe3+

at the nm scale using STXM-XANES measurementsThis last study was particularly interesting as it showedthat approaches similar to those used for studyingmetamorphic rocks could be applied to low-grade anddiagenetic samples

Looking at diagenetic rocks like metamorphicrocks

The use of equilibrium thermodynamics to studylow-grade rocks has been often criticized becausesluggish kinetics might preclude achievement of globalequilibrium at low temperature and is responsible forthe crystallization of metastable phases with a lowerenergy of nucleation than stable ones Essene amp Peacor(1995) claimed that the coexistence of chemicallyinhomogeneous phases at a very fine scale in low-grade rocks is indicative of disequilibrium and thatmineral reaction is better described by a kinetic ratherthan thermodynamic approach However the coexist-ence of chemically inhomogeneous phases is also acommon observation in metamorphic rocks wherezoned grains and the coexistence for example ofchlorite grains of different compositions indicate thatthermodynamic equilibrium is approached only locallyand at a given time in the metamorphic history In factthe main difference between diagenetic and higher-temperature rocks is the size of the domains in localequilibrium The more sluggish kinetics of reactionand diffusivity at low temperature lead to small grainsand domains in equilibrium but these domains can beidentified when using appropriate analytical tools andprocedures developed recently (Bourdelle et al2013bc)

FIG 10 Quantitative Fe redox nanomapping on a FIB foilof Glarus sample (T lt 300degC Switzerland) calculatedfrom the 7087ndash71025 eV images ratio (STXM-XANES)

(from Bourdelle et al 2013c)

FIG 11 Analysis of an illite-chlorite assemblage (Gulf Coast sample Alamo 1 well 5825 m) (a) SEM image of thepetrographic thin section showing the emplacement of the FIB foil to be cut across the illitendashchlorite interface (b)Bright-field TEM image of the FIB foil extracted from the section with the platinum strap on top of it (cd) Enlargementof part b showing the Al and Mg contents of chlorite (apfu) and their variation between crystal core and crystal rim

630 O Vidal et al

Figure 10 shows that it is possible to demonstrateheterogeneities in XFe3+ contents at the nanometerscale Consistent variations of other major elements arealso observed with TEM technics on FIB foils Anexample is shown in Fig 11 which depicts thevariations of Al and Mg contents in a chlorite grainfrom the Gulf Coast (Bourdelle et al 2013b) In thiscase the FIB foil was cut parallel to the interfacebetween illite and chlorite and a nm-scale investiga-tion of composition was conducted by TEM-EDXanalysis The rim and core compositions showed azoning of chlorite with Mg- and Al-richer rims thancore compositions This variation in composition on ascale of lt1 μm was related to various crystallizationstages during burial Chlorite sampled at differentdepths and various locations of the Gulf Coast areashowed that zoned compositions similar to thatobserved in Fig 11 are frequent (Bourdelle et al2013b) The AlIV in chlorite rims display a Tdependence that is consistent with the km-scale andregional evolution of temperature ie with thediagenetic history of Gulf Coast (burial P-T condi-tions detritalauthigenic compositions etchellip)(Fig 12) This study suggests again that approachesused for metamorphic rocks can be applied to low-grade rocks when using adapted analytical tools able toidentify sub-μm variations of composition

CONCLUS ION

The variability of chlorite composition reflects thevariability of its conditions of formation Numerousefforts have been made over the last 30 years to

translate the composition of chlorite in terms of itstemperature of formation The first empirical equationsbased on the AlIVor vacancy contents are restricted tobulk-rock compositions and pressure conditions atwhich they were calibrated Thermodynamicapproaches and semi-empirical equations based onthe ratio of end-member activities do not present thislimit Among the different thermodynamic models ofchlorite published so far Vidal et al (2006) proposed amodel that can be used to estimate T and (XFe3+)Chlsimultaneously The estimated (XFe3+)Chl values areconsistent with those measured by Moumlssbauer orXANES spectroscopy and for the same (XFe3+)Chlthe temperatures estimated using the method of Vidalet al (2006) are similar to those obtained with thesemi-empirical equations of Inoue et al (2009) andLanari et al (2014) Different T-XFe3+ trends areobserved for different pressure gradients and oxygen-buffering assemblages (Lanari et al 2014) butoverall the systematic increase in XFe3+ with areduction in temperature is compatible with thedecrease in aO2 in a closed system The increase inXFe3+ and chlorite vacancy with decreasing tempera-ture is interpreted as reflecting the incorporation of theFe3+-sudoite component which can be used toestimate T-aO2-XFe

3+ values with a Chl-Qtz-H2Omulti-equilibrium approach

Similar temperatures are calculated for low-tem-perature chlorites (T lt 300degC and measured XFe3+ upto 30) with all approaches assuming that iron inchlorite is either completely ferrous or with estimatedvalues of XFe3+ This indicates that the equations ofsemi-empirical thermometers assuming that all iron inchlorite is ferrous include some inherent Fe2+Fe3+ vstemperature correction However very different tem-peratures can be obtained for higher-temperaturesamples (gt300degC) with approaches involving ferriciron or assuming that all of the iron is ferrous Thissuggests that the Fe2+Fe3+ vs temperature correctionmight be restricted to particular aO2 conditions andpressure gradients More work is necessary to betterunderstand this point and to improve the thermo-dynamic models and semi-empirical thermometersThis requires high-quality and well constrained naturaldata To this end mapping techniques that highlight thelink between the variations in temperature and pressureand small-scale chlorite composition including XFe3+

variations offer much promise When combining theresults of the present study with those of other studiespublished recently it appears that thermodynamicapproaches and mapping techniques developed formetamorphic rocks could be used to study low-grade

FIG 12 Range of crystal-rim AlIV content (white boxes)of Gulf Coast chlorites vs measured temperatures Thegrey area underlines the range of maximum AlIV contentwhich is assumed to represent the closest approach toequilibrium composition (after Bourdelle et al 2013b)

631Conditions of formation of chlorite

rocks where kinetics are much more sluggish Thisrequires use of appropriate analytical tools and techni-ques with a spatial resolution of just a few hundrednanometres which are now becoming available

ACKNOWLEDGMENTS

Constructive reviews by S Hillier F Nieto and ananonymous reviewer are gratefully acknowledged

REFERENCES

Beaufort D Patrier P Meunier A amp Ottaviani MM(1992) Chemical variations in assemblages includingepidote andor chlorite in the fossil hydrothermalsystem of Saint Martin (Lesser Antilles) Journal ofVolcanology and Geothermal Research 51 95ndash114

Berman R (1988) Internally-consistent thermodynamicdata for minerals in the system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2Journal of Petrology 29 445ndash522

Berman R amp Brown T (1985) Heat capacity of minerals inthe system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2 representation estima-tion and high temperature extrapolation Contributionsto Mineralogy and Petrology 89 168ndash183

Bourdelle F amp Cathelineau M (2015) Low-temperaturechlorite geothermometry a graphical representationbased on a TndashR2+ndashSi diagram European Journal ofMineralogy 27 617ndash626

Bourdelle F Parra T Chopin C amp Beyssac O (2013a) Anew chlorite geothermometer for diagenetic to low-grade metamorphic conditions Contributions toMineralogy and Petrology 165 723ndash735

Bourdelle F Parra T Beyssac O Chopin C amp Vidal O(2013b) Clay minerals thermometry a comparativestudy based on high-resolution chemical analyses ofillite and chlorite in sandstones from the Gulf Coast(Texas USA) American Mineralogist 98 914ndash926

Bourdelle F Benzerara K Beyssac O Cosmidis JNeuville DR Brown GE amp Paineau E (2013c)Quantification of the ferric ferrous iron ratio insilicates by scanning transmission X-ray microscopyat the Fe L23 edges Contribution to Mineralogy andPetrology 166 423ndash434

Cathelineau M (1988) Cation site occupancy in chloritesand illites as function of temperature Clay Minerals23 471ndash485

Cathelineau M amp Nieva D (1985) A chlorite solidsolution geothermometer the Los Azufres (Mexico)geothermal system Contributions to Mineralogy andPetrology 91 235ndash244

Chermak JA amp Rimstidt JD (1989) Estimating thethermodynamic properties (ΔGdegf and ΔHdegf) of silicate

minerals at 298K from the sum of polyhedral contribu-tions American Mineralogist 74 1023ndash1031

Coggon R amp Holland TJB (2002) Mixing properties ofphengitic micas and revised garnent-phengite thermo-barometers Journal of Metamorphic Geology 20683ndash696

Curtis CD Hughes CR Whiteman JA ampWhittle CK(1985) Compositional variation within some sedi-mentary chlorites and some comments on their originMineralogical Magazine 49 375ndash386

de Andrade V Vidal O Lewin E OrsquoBrien P amp Agard P(2006) Quantification of electron microprobe com-positional maps of rock thin sections an optimizedmethod and examples Journal of MetamorphicGeology 24 655ndash668

de Andrade V Susini J Salomeacute M Beraldin O RigaultC Heymes T Lewin E amp Vidal O (2011)Submicrometer hyperspectral X-ray imaging of het-erogeneous rocks and geomaterials applications at theFe k-edge Analytical Chemistry 83 4220ndash4227

de Caritat P Hutcheon I amp Walshe J (1993) Chloritegeothermometry a review Clays and Clay Minerals41 219ndash239

Essene EJ (2009) Thermobarometry gone astray Pp101ndash129 in Physics and Chemistry of Earthrsquos Interior(AK Gupta amp S Dasgupta editors) PlatinumJubilee Indian National Science Academy Springer-Verlag Berlin

Essene EJ amp Peacor DR (1995) Clay mineralthermometry ndash a critical perspective Clays and ClayMinerals 43 540ndash553

Ganne J de Andrade V Weinberg RF Vidal O DubacqB Kagambega N Naba S Baratoux L Jessell M ampAllibon J (2012) Modern-style plate subductionpreserved in the Palaeoproterozoic West Africancraton Nature Geoscience 5 60ndash65

Grosch EG Vidal O Abu-Alam T amp McLoughlin N(2012) PTconstraints on the metamorphic evolution ofthe Paleoarchean Kromberg type-section BarbertonGreenstone Belt South-Africa Journal of Petrology53 513ndash545

Grosch E McLoughlin N Lanari P Erambert M ampVidal O (2014) Microscale mapping of alterationconditions and potential biosignatures in basaltic-ultramafic rocks on early Earth and beyondAstrobiology 14 216ndash228

Hillier S amp Velde B (1991) Octahedral occupancy andchemical composition of diagenetic (low-temperature)chlorites Clay Minerals 26 149ndash168

Holland TJB (1989) Dependance of entropy on volumefor silicates and oxide minerals A review and apredictive model American Mineralogist 74 5ndash13

Holland T amp Powell R (1998) An internally consistentthermodynamic data set for phases of petrologicalinterest Journal ofMetamorphic Geology 16 309ndash343

Holland T Baker J amp Powell R (1998)Mixing propertiesand activity composition and relationships of chlorites

632 O Vidal et al

low-temperature series (Inoue et al 2009 Bourdelleet al 2013a Lanari et al 2014) and metamorphicrocks (Lanari et al 2014) The weakness of the semi-empirical equations is that their compatibility withexperimental data obtained in MASH and FASHsystems has not always been checked and they arerestricted to quartz-bearing rocks

Finally the main difference between the thermo-dynamic approach of Vidal et al (2005 2006) andthose of Vidal et al (2001) Lanari et al (2014) or thesemi-empirical thermometers of Inoue et al (2009)

Bourdelle et al (2013a) and Lanari et al (2014) is thatthe former leads to the estimation of T-(XFe3+)Chl pairsat given pressure whereas the latter estimate T atassumed or measured XFe3+ and pressure or assuminga negligible role of pressure This last assumption isacceptable at low-pressure conditions but becomesproblematic for metamorphic samples because theamount of Mg-Sud end-member at fixed temperaturehas been shown to increase with pressure (Fig 3b)

A comparison of the approaches of Vidal et al(2006) Lanari et al (2014) and Inoue et al (2009) is

FIG 4 Comparison of measured and calculated Si content of chlorite in the FASH systemwith almandine and fayalite (a)or chloritoid and hercynite (b) Boxes experimental data of Parra et al (2005) continuous lines Si content calculatedwith the thermodynamic properties of Fe-amesite and daphnite and the solid solution model of Vidal et al (2005 2006)dashed lines Si content calculated with the Holland et al (1998) model updated by Coggon amp Holland (2002) which

assumes that the thermodynamic properties of Fe-Am depend on those of Clin Mg-Am and Daph

621Conditions of formation of chlorite

shown in Fig 5 which reports temperatures estimatedfor the set of 1541 analyses used in the present studyThe temperatures calculated with the assumption Fe3+

= 0 (equation 38 of Lanari et al 2014) or Fe3+ gt 0 (allother approaches) are labelled TFe2+ and TFe3+respectively in Fig 5 and in the following Allcalculations were conducted with a MATLAB scriptwith a GUI interface available on request from the firstauthor

The temperature conditions estimated for blueschist-facies metamorphic samples with the approach ofVidal et al (2006) (TFe3+ Vidal) cover a wide range oftemperature from 275 to 550degC (Fig 5a) About onethird of the samples show temperatures of lt350degCwhich is 50ndash100degC lower than the published peaktemperatures derived for the samples studied Usingthe XFe3+ estimated with the Vidal et al (2006)approach the temperatures estimated with the semi-empirical equations of Inoue et al (2009) (TFe3+

Inoue) or Lanari et al (2014) (TFe3+ Lanari) show asimilar evolution but the estimated temperatures are50ndash100degC lower in the case of Inoue et al (2009) andhigher in the case of Lanari et al (2014) The mostimportant deviation is observed between TFe2+ Lanariand the other temperature estimates A significantproportion of chlorites calculated to have crystallized atTFe3+asymp 300degC give much higher TFe2+Lanari (gt500degCopen symbols in Fig 5a) and plot on a different trendThe same observation is made for the Barbertonsamples from Grosch et al (2012) (Fig 5c) In bothcases the maximum TFe2+ Lanari can be as high as900degC but the highest temperatures are obtained forsamples containing very small amounts of vacancies(lt005 pfu open circles below the dashed line in thetemperature-vacancy plot shown in Fig 5a) The Mg-Sud activity and temperatures derived from theseanalyses are very sensitive to small variations incomposition and Fe3+-content and they should not beused for reliable thermometric estimates In contrastFig 5b shows that similar temperatures are estimatedwith all approaches for low-pressure and low-gradesamples

This comparison indicates that similar results areobtained with all methods at T lt 300degC and similartrends are observed at higher temperature for the threemethods allowing the incorporation of Fe3+ althoughwith a significant scatter at T gt 400degC With the XFe3+

values derived using the Vidal et al (2006) approachTFe3+ Vidal and TFe3+ Inoue are quite consistentHowever the highest TFe3+ Inoue estimated forblueschist metamorphic samples are lower whereasTFe3+ Lanari are higher In all approaches the

temperature calculated depends heavily on thesudoite activity which is proportional to the amountof vacancy in chlorite (a(Sud) = 64 (XSi XAl)T2 XM1 (XMg)2M2ndash3 (XAl)2M2ndash3 (XAl)M4))This is illustrated in Fig 5 which shows that theamount of vacancy in chlorite increases with a drop intemperature As noted by Laird (1988) the amount ofvacancy in chlorite structural formulae calculated on acharge-normalized 14 anhydrous-oxygen basisincreases with increasing XFe3+ It is therefore notsurprising that like vacancies XFe3+ increases withreducing temperature This trend is illustrated in Fig 6where the TFe3+Vidal vs XFe3+ values of samples usedto compute Fig 5 have been plotted Figure 6 alsoindicates that larger XFe3+ values are calculated for thehigh-pressure samples than for the low-pressuresamples Nevertheless vacancies are also noted toincreasewith reducing Tunder the assumption that Fe2+

= FeTotal (see Fig 3b and the open circles in Fig 5ab)Therefore the increase in vacancies with reducingtemperature is not just a mathematical artefact resultingfrom the calculation of Fe3+-bearing structural formulaeon a 14-oxygen basis Interestingly even at lowtemperature where XFe3+ calculated with the Vidalet al (2006) approach is at a maximum TFe2+ Lanari issimilar to the three TFe3+ (Fig 5b) This indicates thatthe TFe2+ Lanari equation includes the systematicvariation of XFe3+ with temperature observed inFig 6 which explains why similar TFe2+ and TFe3+

are obtained at low T and P (Fig 5b) for contrastingamounts of vacancies and ferric iron contentsNevertheless it is evident from Fig 5ac that the XFe3+ vs temperature relationship that was used to derive theTFe2+ Lanari equation is not fully consistent with the T-XFe3+ relationship obtained with the thermodynamicapproach of Vidal et al (2006) This is illustrated by thedivergence between TFe2+ Lanari and the three TFe3+

calculated for the low-pressure Barberton samples(Fig 5c) and several HP samples Fig 5a

COMPARISON OF EST IMATED ANDMEASURED (XFE 3 + )CHL

At this stage we can only conclude from thecomparison of TFe2+ and TFe3+ in Fig 5ac thateither some XFe3+ values derived with the Vidal et al(2006) approach are overestimated and the low TFe3+

obtained for some HP-HT samples are underestimatedor that TFe2+Lanari only pertains for fO2

conditionsbuffered by specific assemblages An overestimationof XFe3+ due to errors in the standard-state thermo-dynamic properties and the non-ideal Margules

622 O Vidal et al

FIG 5 Left Plot of temperatures estimated with equations 38 (TFe3+ Lanari) and 40 (TFe2+ Lanari) of Lanari et al(2014) and with the thermometer of Inoue et al (2009) (TFe3+ Inoue) vs temperatures estimated using the approach ofVidal et al (2006) TFe3+ Lanari and TFe3+ Inoue were calculated using the XFe3+ values obtained with the approach ofVidal et al (2006) and TFe2+ Lanari assuming XFe3+ = 0 Right Plot of vacancy content vs estimated temperatures(a) and (b) are high-pressure (gt6 kbar) and low- to medium-pressure (lt6 kbar) respectively (c) shows the resultsobtained for the low-pressure samples of Grosch et al (2012) The unrealistically high TFe2+ Lanari are obtained forchlorites with lt005 vacancy pfu (dashed line in the vacancy-temperature plots) The histograms show the repartition

of temperatures estimated with the different approaches

623Conditions of formation of chlorite

parameters of Vidal et al (2006) is possible butthis should be demonstrated by a systematic compari-son of measured and estimated XFe3+ from naturalsamples covering a wide range of P T and fO2

conditions Such a study is still lacking but a first-order comparison of measured and estimated XFe3+

contents for samples of known temperature ofcrystallization is shown in Fig 7 which was obtainedfor the Rouez Niger and Saint-Martin low-T samplescompiled by Inoue et al (2009) the Glarus sample ofBourdelle et al (2013c) and a blueschist-facies sampleretrogressed into the greeschist facies from Sambagawa(de Andrade et al 2006 Munoz et al 2006 Vidalet al 2006) T-XFe3+ values were calculated at 2 kbarexcept for the Sambagawa sample for which calcula-tions were made at 2 5 and 10 kbar for the low-medium- and high-temperature chlorite groups identi-fied by de Andrade et al (2006) and Vidal et al (2006)respectively

FIG 6 Estimated TFe3+ Vidal and XFe3+ pairs for low- andmedium-pressure samples (open circles) high-pressuresamples (black circles) and the samples from Grosch et al

(2012) (grey squares)

FIG 7 Boxes measured XFe3+ and reference temperatures Symbols Tmaximum-XFe3+minimum pairs calculated to

obtain convergence (plusmn30degC) of equilibria 1 8 9 and 10 with the thermodynamic model of Vidal et al (2006)

624 O Vidal et al

For each series of Inoue et al (2009) between 50and 60 of the chlorite analyses contain Si lt3 apfuand led to a convergence of equations 1 8 9 and 10within 30degC The calculated T-XFe3+ pairs are plottedin Fig 7 and compared with the range of referencetemperature (TRef) estimated independently and XFe3+

measures reported by Inoue et al (2009) A fairagreement of the measured and calculated TFe3+ Vidal-XFe3+ values is observed for the Niger and St-Martinanalyses and a good agreement of the TRef and TFe3+ Vidal was observed for the Rouez samples but thecalculated valueswere smaller than themeasured valuesfor XFe3+ for this samples However the XFe3+ valuesreported in Fig 7 are minimum values required toobtain convergence of equilibria 1 and 8ndash10 within30degC Convergence is maintained at higher valuesof XFe3+ and TFe3+ Vidal for the Rouez samples arestill located in the range of TRef when XFe3+ is setat gt15 (lower limit of the reference values) TheTFe3+ Vidal values calculated for the Glarus chloriteanalyses listed in Bourdelle et al (2013c) are 139degC 203degC and 224degC in fair agreement with thoseestimated independently by Bourdelle et al (2013c)(135degC 170degC and 182degC respectively) or with thesemi-empirical thermometer of Inoue et al (2009)(145degC 196degC and 210degC respectively) For thethree samples studied the XFe3+ values measuredusing XANES (X-ray absorption near edge struc-ture) spectra in the transmission mode of scanningtransmission X-ray microscopy (STXM) were foundto be 223 238 and 279 These values comparefavourably with the estimated ones (32 27 and 28respectively) although the measured values suggestan increase in XFe3+ with temperature which iscontrary the trend inferred from the thermodynamicestimates and evidenced in Figs 6 and 7 Finally therange of XFe3+ estimated for the different generationsof chlorite in the Sambagawa sample (ellipses inFig 7) are slightly higher but still in fair agreementwith those inferred from the XANES measurementsIn conclusion Fig 7 shows that the values estimatedon the basis of the convergence of equilibria (1 8 9and 10) are consistent with the TRef and measuredXFe3+ for various samples covering a range oftemperature from 100degC to 500degC pressure from sim2to 10 kbar and XFe3+ values from 5 to 35 Boththe measured and estimated XFe3+ values plotted inFig 7 seem to confirm the systematic decrease ofXFe3+ with increasing temperature and increase withincreasing pressure that was observed in Fig 6 Thistrend is similar to the decrease in vacancy observedin Fig 5

aO2 -TEMPERATURE EST IMATESAND ADDIT ION OF A FERR I -

SUDOITE END -MEMBER

The general trends inferred from our compilation ofnatural samples indicate that XFe3+ increases withdecreasing temperature and increasing pressureAlthough the model of Vidal et al (2006) involvesno Fe3+-bearing end-member the oxygen activity(aO2) prevailing during the formation of chloritesused to compute Fig 7 can be estimated assuming thatit was buffered by the presence of pure magnetite (Mt)in equilibrium with chlorite and quartz Following thisassumption aO2 for each sample has been calculatedusing equilibria (1 8 9 10) and the followingadditional equilibria which intersect at the same P T(plusmn30degC) (XFe3+)Chl and Log(aO2) (plusmn2) (Fig 8)

15 Mg-Sudthorn 40 Mtthorn 45 Qtzthorn 60 H2O

frac14 6 Clinthorn 24 Daphthorn 20 O2

(11)

6 Mg-Sudthorn 10 Mtthorn 6 Qtzthorn 12 H2O

frac14 3 Mg-Amthorn 6 Daphthorn 5 O2

(12)

15 Mg-Sudthorn 6 Clinthorn 10 Mt

frac14 15 Qtzthorn 15 Mg-Amthorn 6 Daphthorn 5 O2

(13)

6 Daphthorn 4 Clinthorn 5 O2

frac14 20 Qtzthorn 10 Mtthorn 5 Mg-Amthorn 20 H2O(14)

60 Mg-Sudthorn 12 Clinthorn 70 Mtthorn 60 H2O

frac14 45 Mg-Amthorn 42 Daphthorn 35 O2

(15)

4 Qtzthorn 2 Mtthorn Fe-Amthorn 4 H2O

frac14 2 Daphthorn O2

(16)

15 Mg-Sudthorn 6 Daphthorn 10 Mt

frac14 15 Qtzthorn 6 Clinthorn 15 Fe-Amthorn 5 O2

(17)

60 Mg-Sudthorn 70 Mtthorn 60 H2O

frac14 24 Clinthorn 45 Fe-Amthorn 6 Daphthorn 35 O2

(18)

15 Mg-Sudthorn 16 Mtthorn 12 H2O

frac14 3 Qtzthorn 6 Clinthorn 12 Fe-Amthorn 8 O2

(19)

6 Mg-Sudthorn 4 Mt

frac14 6 Qtzthorn 3 Mg-Amthorn 3 Fe-Amthorn 2 O2

(20)

12 Mg-Sudthorn 14 Mtthorn 12 H2O

frac14 6 Mg-Amthorn 3 Fe-Amthorn 6 Daphthorn 7 O2

(21)

5 Mg-Amthorn 4 Mtthorn 8 Qtzthorn 8 H2O

frac14 4 Clinthorn 3 Fe-Amthorn 2 O2

(22)

24 Mg-Sudthorn 3 Mg-Amthorn 28 Mtthorn 24 H2O

frac14 12 Clinthorn 21 Fe-Amthorn 14 O2

(23)

All calculated Log(aO2) except those of the Saint-

625Conditions of formation of chlorite

Martin samples follow the same trend and decreasewith decreasing temperature from simminus27 at 450ndash500degC10 kbar (high-temperature chlorites of Sambagawa) tominus50degC at 125degC and 1 kbar (Niger samples) (Fig 8)The Log(aO2) values of the high-T and -P samples plotin the stability field of magnetite whereas those of thelow-T samples plot in the stability field of hematitewhich is not consistent with the assumption of ChlndashMtndashQtzndashH2O equilibrium The presence of hematite hasbeen reported in the Saint-Martin samples only(Beaufort et al 1992) which plot above the trenddefined by the other samples (upper grey ellipse inFig 8) The relative position of this hematite-bearingsample is consistent with expected higher oxygenactivities required for its formation For the other low-T and apparently hematite-free samples it is possiblethat the calculated Log(aO2) values are too high or thatoxygen was not buffered by an equilibrium ofmagnetitewith chlorite In any case Fig 8 indicates that thesystematic increase in (XFe3+)Chl with decreasingtemperature observed in Figs 6 and 7 is consistentwith a lower aO2 and amount of free oxygen at low-Tthan high-T in systems where aO2 is buffered by thepresence of chlorite and iron oxides To maintain this

equilibrium with lowering temperature in metamorphicrocks where oxygen is immobile oxygen has to beconsumed and one way for that is to produce chloritethat is richer in Fe3+ This is a possible interpretationfor the formation of low-temperature Fe3+-rich chloritecoexisting with high-temperature Fe3+-poor chloritein the same thin section of the Sambagwa sample(Fig 9)

In order to discuss the Log(aO2)-T-XFe3+ evolution

in more detail for rocks where aO2 is not buffered bythe presence of magnetite it is necessary to define anFe3+-bearing chlorite end-member The simultaneousincrease of vacancy and Fe3+ content with decreasingtemperature observed in Figs 5ndash7 is explained by the3(Mg Fe2+) = 2(Fe3+) ditrioctahedral substitution(see also Trincal amp Lanari 2016 this issue) Thisexchange can be modelled with a ferri-sudoite end-member that materializes the relationship betweenXFe3+ and vacancy The standard-state thermo-dynamic properties of ferri-sudoite were constrainedas follows

(1) Because the maximum amount of (Fe3+)Chlcalculated in the previous section never exceeded1 apfu the cation distribution of Vidal et al (2006)

FIG 8 Left Ranges of Log(aO2) conditions calculated for the samples used to compute Fig 7 at the temperature andXFe3+ contents estimated from the Chl-Qtz-H2O equilibrium conditions Dashed lines hematite-magnetite-O2

equilibrium calculated using the thermodynamic properties of Berman (1988) and unit activities of Ht and Mt RightSample calculation at 2 kbar for a Rouez sample using reactions 8ndash20 (upper) or 26ndash40 (lower) The vertical linescorrespond to oxygen- and magnetite-free reactions 1 5 6 and 7 used to compute Figs 5ndash7 which converge at +30degCThe circles in the lower figure represent values calculated with (ΔHdegf ) ferri-Sud constrained with the chlorite-magnetite-quartz-water equilibria 8ndash20 (open circle) or estimated as ΔHdegf (Mg-sudoite) ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2

(filled circle) (see Table 1 and text)

626 O Vidal et al

was kept unchanged and the Fe3+-Al3+ exchange wasassumed to be restricted to the M4 site of chlorite Thechosen formula of ferri-Sud was thus ([Si2]T1[SiAl]T2)

IV([]M1[Mg2Al2]M2-3[Fe3+]M4)

VIO10(OH)8The standard-state thermodynamic properties of allother chlorite end-members the activity model and theMargules parameters were kept unchanged

(2) The third-law entropy of ferri-sudoite (Sdeg =415 JmolK) was estimated according to Holland(1989) with a configurational entropy of 2305 JmolKand a ferromagnetic contribution of 1304 JmolK itsvolume (Vdeg = 204 Jb) was estimated to be 01 Jblower than that of Mg-sudoite (2051 Jb) and the Cp(T ) function was estimated according to Berman amp

FIG 9 Rock-sample mapping of a Sambagawa sample (a) maps of estimated XFe3+ (Vidal et al 2006 Munoz et al2006) and Log(aO2) (estimated assuming that it is buffered by magnetite-chlorite equilibria see text and Fig 8)(b) measured Fe XANES edge-position proportional to Fe redox (Munoz et al 2006) (c) estimated temperature (Vidal

et al 2006 Munoz et al 2006)

627Conditions of formation of chlorite

Brown (1985) The formation enthalpy (ΔHdegf = 82254 plusmn 5 kJmol) was first estimated as the average valuebetween that calculated according to Chermak ampRimstidt (1989) (minus82204 kJmol) and that(minus82304 kJmol) calculted as ΔHdegf (Mg-sudoite) ndashΔHdegf (corundum)2 + ΔHdegf (hematite)2 using thevalues of Vidal et al (1992) for sudoite and Berman(1988) for hematite

(3) In a first step ΔHdegf and Sdeg of ferri-sudoite wereadjusted in order to reproduce the temperaturesreported in Fig 7 with oxygen-free reactions andassuming that magnetite is present in all samples(Chl-Qtz-Mt assemblage) ΔHdegf was allowed to varybetweenminus8220 andminus8230 kJmol and Sdeg within 10 JmolK of the estimated value The addition ofmagnetite and ferri-sudoite to quartz and water leadsto 17 additional equilibria and the following wereused to constrain ΔHdegf and Sdeg of ferri-sudoite

ClinthornMg-Sud frac14 Mg-Amthorn Qtzthorn H2O (1)

4 Clinthorn Daphthorn 10 ferri-Sud

frac14 10 Mg-Amthorn 25 Qtzthorn 30 Mtthorn 20 H2O(24)

7 Daphthorn 70 ferri-Sudthorn 30 Mg-Am

frac14 22 Clinthorn 75 Mg-Sudthorn 35 Mtthorn 40 H2O(25)

28 ferri-Sudthorn 11 Fe-AmthornMg-Am

frac14 6 Daphthorn 30 Mg-Sudthorn 14 Mtthorn 16 H2O(26)

Daphthorn 10 ferri-Sudthorn 10 Mg-Amthorn 10 Qtz

frac14 6 Clinthorn 15 Mg-Sudthorn 5 Mtthorn 16 H2O(27)

The ferri-Sud properties derived with the assumptionof ChlndashQtzndashMt equilibrium are listed in Table 1(referenced a)With these values the oxygen activity atthe P T and (XFe3+)Chl of all equilibria used tocompute Fig 7 can be calculated with the followingoxygen-bearing and magnetite-free equilibria whichconverge with equilibria 1 8 9 and 10 at the same PTand aO2 conditions

16 Qtzthorn 7 Mg-Amthorn 4 Daphthorn 6 Mg-Sud

thorn 5 O2 thorn 12 H2O frac14 20 ferri-Sud(28)

Fe-Amthorn 2 Mg-Amthorn 6 Qtzthorn 4 H2Othorn O2

frac14 4 ferri-Sud(29)

Mg-Amthorn 4 Daphthorn 6 Mg-Sudthorn 2 O2

frac14 3 Fe-Amthorn 8 ferri-Sudthorn 2 Qtz(30)

5 Mg-Amthorn 12 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 8 Fe-Amthorn 28 ferri-Sud

(31)

8 Daphthorn 12 Mg-Sudthorn 3 O2 frac14 7 Fe-Am

thorn 12 ferri-Sudthorn 10 Qtzthorn 4 H2O(32)

6 Clinthorn 4 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Mg-Amthorn 20 ferri-Sudthorn 5 Qtz(33)

4 Clinthorn 20 ferri-Sud frac14 30 Qtz

thorn 15 Mg-Amthorn 4 Daphthorn 5 O2 thorn 20 H2O(34)

90 Mg-Sudthorn 28 Daphthorn 32 Clinthorn 20 H2Othorn 35 O2

frac14 140 ferri-Sudthorn 15 Mg-Am

(35)

15 Qtzthorn 14 Clinthorn 16 Daphthorn 45 Mg-Sud

thorn 20 O2 thorn 20 H2O frac14 80 ferri-Sud(36)

2 Clinthorn FeAmthorn 3 Mg-Sudthorn O2

frac14 2 Mg-Amthorn 4 ferri-Sudthorn Qtz(37)

12 Clinthorn 7 Fe-Amthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 10 Mg-Amthorn 28 ferri-Sud

(38)

5 Qtzthorn 2 Clinthorn 2 Fe-Amthorn 3 Mg-Sud

thorn 2 O2 thorn 4 H2O frac14 8 ferri-Sud(39)

2 Clinthorn 8 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Fe-Amthorn 20 ferri-Sudthorn 5 Qtz(40)

30 Qtzthorn 8 Clinthorn 15 Fe-Amthorn 5 O2 thorn 20 H2O

frac14 8 Daphthorn 20 ferri-Sud

(41)

4 Clinthorn 8 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 3 Fe-Amthorn 28 ferri-Sud

(42)

Because the ferri-sudoite standard-state propertieswere constrained by the chlorite-quartz-magnetite-O2-H2O equilibrium the Log(aO2) calculated withthe equilibria above follow naturally the trend shown inFig 8 and the Log(aO2) values calculated withreactions 28ndash42 of low-temperature samples plot inthe hematite stability field This suggests that theΔGdegPT values of ferri-sudoite derived above with theassumption that aO2 is buffered by the presence ofmagnetite are possibly too high A reduction of ΔHdegf(ferri-sudoite) to 82304 kJmol (=ΔHdegf (Mg-sudoite)ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2 see Table 1)leads to Log(aO2) reduced by 45 Log units (see Fig 8)

628 O Vidal et al

and most calculated values except those of thehematite-bearing St Martin samples plot in themagnetite stability field

A more detailed study is required to refine the ferri-sudoite thermodynamic data However the resultsobtained using the preliminary values listed in Table 1show that this end-member can be used to reproduceconsistently the T-XFe3+ and T- trends obtained fromclassical Chl-Qtz-H2O calculations (Figs 5ndash7) Theincrease in XFe3+ with reducing temperature is consist-ent with the expected reduction in aO2 and the amountof free oxygen that is consumed to produce Fe3+-bearingchlorite Ferric iron is therefore a major component thathas to be taken into account when dealing with low-temperature samples Empirical or semi-empiricalchlorite thermometers which assume that all Fe is inthe ferrous state may include one of the XFe3+-T andaO2-T trends illustrated in Figs 7 and 8 but they cannotaccount for the different XFe3+-T trends that areexpected in the contexts of different aO2 conditions

FUTURE CHALLENGES LOOK INGAT LOW-GRADE ROCKS SUCH ASMETAMORPH IC ROCKS AND THE

DYNAMICS OF CHLOR ITECRYSTALL IZAT ION AT THE

MICROMETR IC TO NANOMETR ICSCALE

T-P-XFe3+ mapping at the thin-section scaleevidence of chlorite composition re-equilibration in metamorphic and low-gradesamples

Although natural samples offer valuable constraintson thermodynamic models there are two maindifficulties in using them to evaluate the reliability ofexisting chlorite models for thermometric purposes

(1) In the case of low-grade samples referencepressures and temperatures are generally known withina reasonable absolute uncertainty but the size ofchlorite grains is often small and good microprobeanalyses are sometimes difficult to obtain Moreoverthe kinetics of reaction are so sluggish that inheritedchlorite grains might remain unaffected by latervariations in temperature which makes thetemperature-composition relationships difficult toconstrain Lastly extrapolation at HT and HP ofempirical or semi-empirical models constrained byLT samples only may be erroneous in particular theeffect of pressure or eventual systematic variations ofXFe3+ cannot be appraised

(2) In the case of metamorphic samplesmicroprobe analyses are easier to obtain from largechlorite grains but chlorites of contrasted compositionthat crystallized at different temperatures and times ofthe metamorphic history are often observed to coexistmetastably in different microstructural sites of the samethin section In such cases the definition of a singlepair of reference temperature and pressure isinappropriate and cannot be restricted to the peak-temperature conditions

In order to overcome these difficulties and to bettervisualize the link between mineral compositiondeformation and P-T conditions at the thin-sectionscale de Andrade et al (2006 2011) Vidal et al(2006) and Lanari et al (2013) developed an approachwhich combines elementary maps of composition witheither a thermodynamic or a semi-empirical estimate ofchlorite crystallization temperature Maps of rawmicroprobe analyses were standardized using good-quality point analyses of minerals present in themapped area The maps of standardized analyseswere transformed into maps of structural formulaewhere each pixel analysed is converted into individualmineral structural formulae Depending on the thermo-metric approach it was then possible to compute mapsof temperature or T-XFe3+ conditions assuming thateach chlorite pixel was in equilibrium with quartz andwater The results reported by Vidal et al (2006)Yamato et al (2007) Ganne et al (2012) Lanari et al(2013 2014) and Scheffer et al (2015) showed a closelink between deformation and mineral reaction at themicroscopic scale and maps of chlorite temperaturecould be used as a basis for interpreting the micro-structures observed in thin sections of low- to middle-grade metapelites In all these studies the metastablepersistence of different generations of chlorite crystal-lizing at different temperatures during the exhumationof metamorphic samples could be used to constrain theT-deformation history Moreover the map of (XFe3+)Chl calculated by Vidal et al (2006) was in goodqualitative agreement with the in situ mapping of thisratio using the X-ray absorption near-edge structure(XANES) techniques (Fig 9) (see also Munoz et al2006) The calculated temperature and XFe3+ mapsobtained on a single sample showed the same trend ofreducing (XFe3+)Chl with increasing temperature asobserved in Figs 6 and 7

Bourdelle et al (2013c) applied a similar approachfor low-temperature samples with chlorite of smallgrain size using transmission electron microscopy-energy dispersive X-ray (TEM-EDX) spectroscopicanalyses of focused-ion beam (FIB) foils instead of

629Conditions of formation of chlorite

microprobe analyses of thin sections (Fig 10) Thoseauthors were able to observe heterogeneities in XFe3+

at the nm scale using STXM-XANES measurementsThis last study was particularly interesting as it showedthat approaches similar to those used for studyingmetamorphic rocks could be applied to low-grade anddiagenetic samples

Looking at diagenetic rocks like metamorphicrocks

The use of equilibrium thermodynamics to studylow-grade rocks has been often criticized becausesluggish kinetics might preclude achievement of globalequilibrium at low temperature and is responsible forthe crystallization of metastable phases with a lowerenergy of nucleation than stable ones Essene amp Peacor(1995) claimed that the coexistence of chemicallyinhomogeneous phases at a very fine scale in low-grade rocks is indicative of disequilibrium and thatmineral reaction is better described by a kinetic ratherthan thermodynamic approach However the coexist-ence of chemically inhomogeneous phases is also acommon observation in metamorphic rocks wherezoned grains and the coexistence for example ofchlorite grains of different compositions indicate thatthermodynamic equilibrium is approached only locallyand at a given time in the metamorphic history In factthe main difference between diagenetic and higher-temperature rocks is the size of the domains in localequilibrium The more sluggish kinetics of reactionand diffusivity at low temperature lead to small grainsand domains in equilibrium but these domains can beidentified when using appropriate analytical tools andprocedures developed recently (Bourdelle et al2013bc)

FIG 10 Quantitative Fe redox nanomapping on a FIB foilof Glarus sample (T lt 300degC Switzerland) calculatedfrom the 7087ndash71025 eV images ratio (STXM-XANES)

(from Bourdelle et al 2013c)

FIG 11 Analysis of an illite-chlorite assemblage (Gulf Coast sample Alamo 1 well 5825 m) (a) SEM image of thepetrographic thin section showing the emplacement of the FIB foil to be cut across the illitendashchlorite interface (b)Bright-field TEM image of the FIB foil extracted from the section with the platinum strap on top of it (cd) Enlargementof part b showing the Al and Mg contents of chlorite (apfu) and their variation between crystal core and crystal rim

630 O Vidal et al

Figure 10 shows that it is possible to demonstrateheterogeneities in XFe3+ contents at the nanometerscale Consistent variations of other major elements arealso observed with TEM technics on FIB foils Anexample is shown in Fig 11 which depicts thevariations of Al and Mg contents in a chlorite grainfrom the Gulf Coast (Bourdelle et al 2013b) In thiscase the FIB foil was cut parallel to the interfacebetween illite and chlorite and a nm-scale investiga-tion of composition was conducted by TEM-EDXanalysis The rim and core compositions showed azoning of chlorite with Mg- and Al-richer rims thancore compositions This variation in composition on ascale of lt1 μm was related to various crystallizationstages during burial Chlorite sampled at differentdepths and various locations of the Gulf Coast areashowed that zoned compositions similar to thatobserved in Fig 11 are frequent (Bourdelle et al2013b) The AlIV in chlorite rims display a Tdependence that is consistent with the km-scale andregional evolution of temperature ie with thediagenetic history of Gulf Coast (burial P-T condi-tions detritalauthigenic compositions etchellip)(Fig 12) This study suggests again that approachesused for metamorphic rocks can be applied to low-grade rocks when using adapted analytical tools able toidentify sub-μm variations of composition

CONCLUS ION

The variability of chlorite composition reflects thevariability of its conditions of formation Numerousefforts have been made over the last 30 years to

translate the composition of chlorite in terms of itstemperature of formation The first empirical equationsbased on the AlIVor vacancy contents are restricted tobulk-rock compositions and pressure conditions atwhich they were calibrated Thermodynamicapproaches and semi-empirical equations based onthe ratio of end-member activities do not present thislimit Among the different thermodynamic models ofchlorite published so far Vidal et al (2006) proposed amodel that can be used to estimate T and (XFe3+)Chlsimultaneously The estimated (XFe3+)Chl values areconsistent with those measured by Moumlssbauer orXANES spectroscopy and for the same (XFe3+)Chlthe temperatures estimated using the method of Vidalet al (2006) are similar to those obtained with thesemi-empirical equations of Inoue et al (2009) andLanari et al (2014) Different T-XFe3+ trends areobserved for different pressure gradients and oxygen-buffering assemblages (Lanari et al 2014) butoverall the systematic increase in XFe3+ with areduction in temperature is compatible with thedecrease in aO2 in a closed system The increase inXFe3+ and chlorite vacancy with decreasing tempera-ture is interpreted as reflecting the incorporation of theFe3+-sudoite component which can be used toestimate T-aO2-XFe

3+ values with a Chl-Qtz-H2Omulti-equilibrium approach

Similar temperatures are calculated for low-tem-perature chlorites (T lt 300degC and measured XFe3+ upto 30) with all approaches assuming that iron inchlorite is either completely ferrous or with estimatedvalues of XFe3+ This indicates that the equations ofsemi-empirical thermometers assuming that all iron inchlorite is ferrous include some inherent Fe2+Fe3+ vstemperature correction However very different tem-peratures can be obtained for higher-temperaturesamples (gt300degC) with approaches involving ferriciron or assuming that all of the iron is ferrous Thissuggests that the Fe2+Fe3+ vs temperature correctionmight be restricted to particular aO2 conditions andpressure gradients More work is necessary to betterunderstand this point and to improve the thermo-dynamic models and semi-empirical thermometersThis requires high-quality and well constrained naturaldata To this end mapping techniques that highlight thelink between the variations in temperature and pressureand small-scale chlorite composition including XFe3+

variations offer much promise When combining theresults of the present study with those of other studiespublished recently it appears that thermodynamicapproaches and mapping techniques developed formetamorphic rocks could be used to study low-grade

FIG 12 Range of crystal-rim AlIV content (white boxes)of Gulf Coast chlorites vs measured temperatures Thegrey area underlines the range of maximum AlIV contentwhich is assumed to represent the closest approach toequilibrium composition (after Bourdelle et al 2013b)

631Conditions of formation of chlorite

rocks where kinetics are much more sluggish Thisrequires use of appropriate analytical tools and techni-ques with a spatial resolution of just a few hundrednanometres which are now becoming available

ACKNOWLEDGMENTS

Constructive reviews by S Hillier F Nieto and ananonymous reviewer are gratefully acknowledged

REFERENCES

Beaufort D Patrier P Meunier A amp Ottaviani MM(1992) Chemical variations in assemblages includingepidote andor chlorite in the fossil hydrothermalsystem of Saint Martin (Lesser Antilles) Journal ofVolcanology and Geothermal Research 51 95ndash114

Berman R (1988) Internally-consistent thermodynamicdata for minerals in the system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2Journal of Petrology 29 445ndash522

Berman R amp Brown T (1985) Heat capacity of minerals inthe system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2 representation estima-tion and high temperature extrapolation Contributionsto Mineralogy and Petrology 89 168ndash183

Bourdelle F amp Cathelineau M (2015) Low-temperaturechlorite geothermometry a graphical representationbased on a TndashR2+ndashSi diagram European Journal ofMineralogy 27 617ndash626

Bourdelle F Parra T Chopin C amp Beyssac O (2013a) Anew chlorite geothermometer for diagenetic to low-grade metamorphic conditions Contributions toMineralogy and Petrology 165 723ndash735

Bourdelle F Parra T Beyssac O Chopin C amp Vidal O(2013b) Clay minerals thermometry a comparativestudy based on high-resolution chemical analyses ofillite and chlorite in sandstones from the Gulf Coast(Texas USA) American Mineralogist 98 914ndash926

Bourdelle F Benzerara K Beyssac O Cosmidis JNeuville DR Brown GE amp Paineau E (2013c)Quantification of the ferric ferrous iron ratio insilicates by scanning transmission X-ray microscopyat the Fe L23 edges Contribution to Mineralogy andPetrology 166 423ndash434

Cathelineau M (1988) Cation site occupancy in chloritesand illites as function of temperature Clay Minerals23 471ndash485

Cathelineau M amp Nieva D (1985) A chlorite solidsolution geothermometer the Los Azufres (Mexico)geothermal system Contributions to Mineralogy andPetrology 91 235ndash244

Chermak JA amp Rimstidt JD (1989) Estimating thethermodynamic properties (ΔGdegf and ΔHdegf) of silicate

minerals at 298K from the sum of polyhedral contribu-tions American Mineralogist 74 1023ndash1031

Coggon R amp Holland TJB (2002) Mixing properties ofphengitic micas and revised garnent-phengite thermo-barometers Journal of Metamorphic Geology 20683ndash696

Curtis CD Hughes CR Whiteman JA ampWhittle CK(1985) Compositional variation within some sedi-mentary chlorites and some comments on their originMineralogical Magazine 49 375ndash386

de Andrade V Vidal O Lewin E OrsquoBrien P amp Agard P(2006) Quantification of electron microprobe com-positional maps of rock thin sections an optimizedmethod and examples Journal of MetamorphicGeology 24 655ndash668

de Andrade V Susini J Salomeacute M Beraldin O RigaultC Heymes T Lewin E amp Vidal O (2011)Submicrometer hyperspectral X-ray imaging of het-erogeneous rocks and geomaterials applications at theFe k-edge Analytical Chemistry 83 4220ndash4227

de Caritat P Hutcheon I amp Walshe J (1993) Chloritegeothermometry a review Clays and Clay Minerals41 219ndash239

Essene EJ (2009) Thermobarometry gone astray Pp101ndash129 in Physics and Chemistry of Earthrsquos Interior(AK Gupta amp S Dasgupta editors) PlatinumJubilee Indian National Science Academy Springer-Verlag Berlin

Essene EJ amp Peacor DR (1995) Clay mineralthermometry ndash a critical perspective Clays and ClayMinerals 43 540ndash553

Ganne J de Andrade V Weinberg RF Vidal O DubacqB Kagambega N Naba S Baratoux L Jessell M ampAllibon J (2012) Modern-style plate subductionpreserved in the Palaeoproterozoic West Africancraton Nature Geoscience 5 60ndash65

Grosch EG Vidal O Abu-Alam T amp McLoughlin N(2012) PTconstraints on the metamorphic evolution ofthe Paleoarchean Kromberg type-section BarbertonGreenstone Belt South-Africa Journal of Petrology53 513ndash545

Grosch E McLoughlin N Lanari P Erambert M ampVidal O (2014) Microscale mapping of alterationconditions and potential biosignatures in basaltic-ultramafic rocks on early Earth and beyondAstrobiology 14 216ndash228

Hillier S amp Velde B (1991) Octahedral occupancy andchemical composition of diagenetic (low-temperature)chlorites Clay Minerals 26 149ndash168

Holland TJB (1989) Dependance of entropy on volumefor silicates and oxide minerals A review and apredictive model American Mineralogist 74 5ndash13

Holland T amp Powell R (1998) An internally consistentthermodynamic data set for phases of petrologicalinterest Journal ofMetamorphic Geology 16 309ndash343

Holland T Baker J amp Powell R (1998)Mixing propertiesand activity composition and relationships of chlorites

632 O Vidal et al

shown in Fig 5 which reports temperatures estimatedfor the set of 1541 analyses used in the present studyThe temperatures calculated with the assumption Fe3+

= 0 (equation 38 of Lanari et al 2014) or Fe3+ gt 0 (allother approaches) are labelled TFe2+ and TFe3+respectively in Fig 5 and in the following Allcalculations were conducted with a MATLAB scriptwith a GUI interface available on request from the firstauthor

The temperature conditions estimated for blueschist-facies metamorphic samples with the approach ofVidal et al (2006) (TFe3+ Vidal) cover a wide range oftemperature from 275 to 550degC (Fig 5a) About onethird of the samples show temperatures of lt350degCwhich is 50ndash100degC lower than the published peaktemperatures derived for the samples studied Usingthe XFe3+ estimated with the Vidal et al (2006)approach the temperatures estimated with the semi-empirical equations of Inoue et al (2009) (TFe3+

Inoue) or Lanari et al (2014) (TFe3+ Lanari) show asimilar evolution but the estimated temperatures are50ndash100degC lower in the case of Inoue et al (2009) andhigher in the case of Lanari et al (2014) The mostimportant deviation is observed between TFe2+ Lanariand the other temperature estimates A significantproportion of chlorites calculated to have crystallized atTFe3+asymp 300degC give much higher TFe2+Lanari (gt500degCopen symbols in Fig 5a) and plot on a different trendThe same observation is made for the Barbertonsamples from Grosch et al (2012) (Fig 5c) In bothcases the maximum TFe2+ Lanari can be as high as900degC but the highest temperatures are obtained forsamples containing very small amounts of vacancies(lt005 pfu open circles below the dashed line in thetemperature-vacancy plot shown in Fig 5a) The Mg-Sud activity and temperatures derived from theseanalyses are very sensitive to small variations incomposition and Fe3+-content and they should not beused for reliable thermometric estimates In contrastFig 5b shows that similar temperatures are estimatedwith all approaches for low-pressure and low-gradesamples

This comparison indicates that similar results areobtained with all methods at T lt 300degC and similartrends are observed at higher temperature for the threemethods allowing the incorporation of Fe3+ althoughwith a significant scatter at T gt 400degC With the XFe3+

values derived using the Vidal et al (2006) approachTFe3+ Vidal and TFe3+ Inoue are quite consistentHowever the highest TFe3+ Inoue estimated forblueschist metamorphic samples are lower whereasTFe3+ Lanari are higher In all approaches the

temperature calculated depends heavily on thesudoite activity which is proportional to the amountof vacancy in chlorite (a(Sud) = 64 (XSi XAl)T2 XM1 (XMg)2M2ndash3 (XAl)2M2ndash3 (XAl)M4))This is illustrated in Fig 5 which shows that theamount of vacancy in chlorite increases with a drop intemperature As noted by Laird (1988) the amount ofvacancy in chlorite structural formulae calculated on acharge-normalized 14 anhydrous-oxygen basisincreases with increasing XFe3+ It is therefore notsurprising that like vacancies XFe3+ increases withreducing temperature This trend is illustrated in Fig 6where the TFe3+Vidal vs XFe3+ values of samples usedto compute Fig 5 have been plotted Figure 6 alsoindicates that larger XFe3+ values are calculated for thehigh-pressure samples than for the low-pressuresamples Nevertheless vacancies are also noted toincreasewith reducing Tunder the assumption that Fe2+

= FeTotal (see Fig 3b and the open circles in Fig 5ab)Therefore the increase in vacancies with reducingtemperature is not just a mathematical artefact resultingfrom the calculation of Fe3+-bearing structural formulaeon a 14-oxygen basis Interestingly even at lowtemperature where XFe3+ calculated with the Vidalet al (2006) approach is at a maximum TFe2+ Lanari issimilar to the three TFe3+ (Fig 5b) This indicates thatthe TFe2+ Lanari equation includes the systematicvariation of XFe3+ with temperature observed inFig 6 which explains why similar TFe2+ and TFe3+

are obtained at low T and P (Fig 5b) for contrastingamounts of vacancies and ferric iron contentsNevertheless it is evident from Fig 5ac that the XFe3+ vs temperature relationship that was used to derive theTFe2+ Lanari equation is not fully consistent with the T-XFe3+ relationship obtained with the thermodynamicapproach of Vidal et al (2006) This is illustrated by thedivergence between TFe2+ Lanari and the three TFe3+

calculated for the low-pressure Barberton samples(Fig 5c) and several HP samples Fig 5a

COMPARISON OF EST IMATED ANDMEASURED (XFE 3 + )CHL

At this stage we can only conclude from thecomparison of TFe2+ and TFe3+ in Fig 5ac thateither some XFe3+ values derived with the Vidal et al(2006) approach are overestimated and the low TFe3+

obtained for some HP-HT samples are underestimatedor that TFe2+Lanari only pertains for fO2

conditionsbuffered by specific assemblages An overestimationof XFe3+ due to errors in the standard-state thermo-dynamic properties and the non-ideal Margules

622 O Vidal et al

FIG 5 Left Plot of temperatures estimated with equations 38 (TFe3+ Lanari) and 40 (TFe2+ Lanari) of Lanari et al(2014) and with the thermometer of Inoue et al (2009) (TFe3+ Inoue) vs temperatures estimated using the approach ofVidal et al (2006) TFe3+ Lanari and TFe3+ Inoue were calculated using the XFe3+ values obtained with the approach ofVidal et al (2006) and TFe2+ Lanari assuming XFe3+ = 0 Right Plot of vacancy content vs estimated temperatures(a) and (b) are high-pressure (gt6 kbar) and low- to medium-pressure (lt6 kbar) respectively (c) shows the resultsobtained for the low-pressure samples of Grosch et al (2012) The unrealistically high TFe2+ Lanari are obtained forchlorites with lt005 vacancy pfu (dashed line in the vacancy-temperature plots) The histograms show the repartition

of temperatures estimated with the different approaches

623Conditions of formation of chlorite

parameters of Vidal et al (2006) is possible butthis should be demonstrated by a systematic compari-son of measured and estimated XFe3+ from naturalsamples covering a wide range of P T and fO2

conditions Such a study is still lacking but a first-order comparison of measured and estimated XFe3+

contents for samples of known temperature ofcrystallization is shown in Fig 7 which was obtainedfor the Rouez Niger and Saint-Martin low-T samplescompiled by Inoue et al (2009) the Glarus sample ofBourdelle et al (2013c) and a blueschist-facies sampleretrogressed into the greeschist facies from Sambagawa(de Andrade et al 2006 Munoz et al 2006 Vidalet al 2006) T-XFe3+ values were calculated at 2 kbarexcept for the Sambagawa sample for which calcula-tions were made at 2 5 and 10 kbar for the low-medium- and high-temperature chlorite groups identi-fied by de Andrade et al (2006) and Vidal et al (2006)respectively

FIG 6 Estimated TFe3+ Vidal and XFe3+ pairs for low- andmedium-pressure samples (open circles) high-pressuresamples (black circles) and the samples from Grosch et al

(2012) (grey squares)

FIG 7 Boxes measured XFe3+ and reference temperatures Symbols Tmaximum-XFe3+minimum pairs calculated to

obtain convergence (plusmn30degC) of equilibria 1 8 9 and 10 with the thermodynamic model of Vidal et al (2006)

624 O Vidal et al

For each series of Inoue et al (2009) between 50and 60 of the chlorite analyses contain Si lt3 apfuand led to a convergence of equations 1 8 9 and 10within 30degC The calculated T-XFe3+ pairs are plottedin Fig 7 and compared with the range of referencetemperature (TRef) estimated independently and XFe3+

measures reported by Inoue et al (2009) A fairagreement of the measured and calculated TFe3+ Vidal-XFe3+ values is observed for the Niger and St-Martinanalyses and a good agreement of the TRef and TFe3+ Vidal was observed for the Rouez samples but thecalculated valueswere smaller than themeasured valuesfor XFe3+ for this samples However the XFe3+ valuesreported in Fig 7 are minimum values required toobtain convergence of equilibria 1 and 8ndash10 within30degC Convergence is maintained at higher valuesof XFe3+ and TFe3+ Vidal for the Rouez samples arestill located in the range of TRef when XFe3+ is setat gt15 (lower limit of the reference values) TheTFe3+ Vidal values calculated for the Glarus chloriteanalyses listed in Bourdelle et al (2013c) are 139degC 203degC and 224degC in fair agreement with thoseestimated independently by Bourdelle et al (2013c)(135degC 170degC and 182degC respectively) or with thesemi-empirical thermometer of Inoue et al (2009)(145degC 196degC and 210degC respectively) For thethree samples studied the XFe3+ values measuredusing XANES (X-ray absorption near edge struc-ture) spectra in the transmission mode of scanningtransmission X-ray microscopy (STXM) were foundto be 223 238 and 279 These values comparefavourably with the estimated ones (32 27 and 28respectively) although the measured values suggestan increase in XFe3+ with temperature which iscontrary the trend inferred from the thermodynamicestimates and evidenced in Figs 6 and 7 Finally therange of XFe3+ estimated for the different generationsof chlorite in the Sambagawa sample (ellipses inFig 7) are slightly higher but still in fair agreementwith those inferred from the XANES measurementsIn conclusion Fig 7 shows that the values estimatedon the basis of the convergence of equilibria (1 8 9and 10) are consistent with the TRef and measuredXFe3+ for various samples covering a range oftemperature from 100degC to 500degC pressure from sim2to 10 kbar and XFe3+ values from 5 to 35 Boththe measured and estimated XFe3+ values plotted inFig 7 seem to confirm the systematic decrease ofXFe3+ with increasing temperature and increase withincreasing pressure that was observed in Fig 6 Thistrend is similar to the decrease in vacancy observedin Fig 5

aO2 -TEMPERATURE EST IMATESAND ADDIT ION OF A FERR I -

SUDOITE END -MEMBER

The general trends inferred from our compilation ofnatural samples indicate that XFe3+ increases withdecreasing temperature and increasing pressureAlthough the model of Vidal et al (2006) involvesno Fe3+-bearing end-member the oxygen activity(aO2) prevailing during the formation of chloritesused to compute Fig 7 can be estimated assuming thatit was buffered by the presence of pure magnetite (Mt)in equilibrium with chlorite and quartz Following thisassumption aO2 for each sample has been calculatedusing equilibria (1 8 9 10) and the followingadditional equilibria which intersect at the same P T(plusmn30degC) (XFe3+)Chl and Log(aO2) (plusmn2) (Fig 8)

15 Mg-Sudthorn 40 Mtthorn 45 Qtzthorn 60 H2O

frac14 6 Clinthorn 24 Daphthorn 20 O2

(11)

6 Mg-Sudthorn 10 Mtthorn 6 Qtzthorn 12 H2O

frac14 3 Mg-Amthorn 6 Daphthorn 5 O2

(12)

15 Mg-Sudthorn 6 Clinthorn 10 Mt

frac14 15 Qtzthorn 15 Mg-Amthorn 6 Daphthorn 5 O2

(13)

6 Daphthorn 4 Clinthorn 5 O2

frac14 20 Qtzthorn 10 Mtthorn 5 Mg-Amthorn 20 H2O(14)

60 Mg-Sudthorn 12 Clinthorn 70 Mtthorn 60 H2O

frac14 45 Mg-Amthorn 42 Daphthorn 35 O2

(15)

4 Qtzthorn 2 Mtthorn Fe-Amthorn 4 H2O

frac14 2 Daphthorn O2

(16)

15 Mg-Sudthorn 6 Daphthorn 10 Mt

frac14 15 Qtzthorn 6 Clinthorn 15 Fe-Amthorn 5 O2

(17)

60 Mg-Sudthorn 70 Mtthorn 60 H2O

frac14 24 Clinthorn 45 Fe-Amthorn 6 Daphthorn 35 O2

(18)

15 Mg-Sudthorn 16 Mtthorn 12 H2O

frac14 3 Qtzthorn 6 Clinthorn 12 Fe-Amthorn 8 O2

(19)

6 Mg-Sudthorn 4 Mt

frac14 6 Qtzthorn 3 Mg-Amthorn 3 Fe-Amthorn 2 O2

(20)

12 Mg-Sudthorn 14 Mtthorn 12 H2O

frac14 6 Mg-Amthorn 3 Fe-Amthorn 6 Daphthorn 7 O2

(21)

5 Mg-Amthorn 4 Mtthorn 8 Qtzthorn 8 H2O

frac14 4 Clinthorn 3 Fe-Amthorn 2 O2

(22)

24 Mg-Sudthorn 3 Mg-Amthorn 28 Mtthorn 24 H2O

frac14 12 Clinthorn 21 Fe-Amthorn 14 O2

(23)

All calculated Log(aO2) except those of the Saint-

625Conditions of formation of chlorite

Martin samples follow the same trend and decreasewith decreasing temperature from simminus27 at 450ndash500degC10 kbar (high-temperature chlorites of Sambagawa) tominus50degC at 125degC and 1 kbar (Niger samples) (Fig 8)The Log(aO2) values of the high-T and -P samples plotin the stability field of magnetite whereas those of thelow-T samples plot in the stability field of hematitewhich is not consistent with the assumption of ChlndashMtndashQtzndashH2O equilibrium The presence of hematite hasbeen reported in the Saint-Martin samples only(Beaufort et al 1992) which plot above the trenddefined by the other samples (upper grey ellipse inFig 8) The relative position of this hematite-bearingsample is consistent with expected higher oxygenactivities required for its formation For the other low-T and apparently hematite-free samples it is possiblethat the calculated Log(aO2) values are too high or thatoxygen was not buffered by an equilibrium ofmagnetitewith chlorite In any case Fig 8 indicates that thesystematic increase in (XFe3+)Chl with decreasingtemperature observed in Figs 6 and 7 is consistentwith a lower aO2 and amount of free oxygen at low-Tthan high-T in systems where aO2 is buffered by thepresence of chlorite and iron oxides To maintain this

equilibrium with lowering temperature in metamorphicrocks where oxygen is immobile oxygen has to beconsumed and one way for that is to produce chloritethat is richer in Fe3+ This is a possible interpretationfor the formation of low-temperature Fe3+-rich chloritecoexisting with high-temperature Fe3+-poor chloritein the same thin section of the Sambagwa sample(Fig 9)

In order to discuss the Log(aO2)-T-XFe3+ evolution

in more detail for rocks where aO2 is not buffered bythe presence of magnetite it is necessary to define anFe3+-bearing chlorite end-member The simultaneousincrease of vacancy and Fe3+ content with decreasingtemperature observed in Figs 5ndash7 is explained by the3(Mg Fe2+) = 2(Fe3+) ditrioctahedral substitution(see also Trincal amp Lanari 2016 this issue) Thisexchange can be modelled with a ferri-sudoite end-member that materializes the relationship betweenXFe3+ and vacancy The standard-state thermo-dynamic properties of ferri-sudoite were constrainedas follows

(1) Because the maximum amount of (Fe3+)Chlcalculated in the previous section never exceeded1 apfu the cation distribution of Vidal et al (2006)

FIG 8 Left Ranges of Log(aO2) conditions calculated for the samples used to compute Fig 7 at the temperature andXFe3+ contents estimated from the Chl-Qtz-H2O equilibrium conditions Dashed lines hematite-magnetite-O2

equilibrium calculated using the thermodynamic properties of Berman (1988) and unit activities of Ht and Mt RightSample calculation at 2 kbar for a Rouez sample using reactions 8ndash20 (upper) or 26ndash40 (lower) The vertical linescorrespond to oxygen- and magnetite-free reactions 1 5 6 and 7 used to compute Figs 5ndash7 which converge at +30degCThe circles in the lower figure represent values calculated with (ΔHdegf ) ferri-Sud constrained with the chlorite-magnetite-quartz-water equilibria 8ndash20 (open circle) or estimated as ΔHdegf (Mg-sudoite) ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2

(filled circle) (see Table 1 and text)

626 O Vidal et al

was kept unchanged and the Fe3+-Al3+ exchange wasassumed to be restricted to the M4 site of chlorite Thechosen formula of ferri-Sud was thus ([Si2]T1[SiAl]T2)

IV([]M1[Mg2Al2]M2-3[Fe3+]M4)

VIO10(OH)8The standard-state thermodynamic properties of allother chlorite end-members the activity model and theMargules parameters were kept unchanged

(2) The third-law entropy of ferri-sudoite (Sdeg =415 JmolK) was estimated according to Holland(1989) with a configurational entropy of 2305 JmolKand a ferromagnetic contribution of 1304 JmolK itsvolume (Vdeg = 204 Jb) was estimated to be 01 Jblower than that of Mg-sudoite (2051 Jb) and the Cp(T ) function was estimated according to Berman amp

FIG 9 Rock-sample mapping of a Sambagawa sample (a) maps of estimated XFe3+ (Vidal et al 2006 Munoz et al2006) and Log(aO2) (estimated assuming that it is buffered by magnetite-chlorite equilibria see text and Fig 8)(b) measured Fe XANES edge-position proportional to Fe redox (Munoz et al 2006) (c) estimated temperature (Vidal

et al 2006 Munoz et al 2006)

627Conditions of formation of chlorite

Brown (1985) The formation enthalpy (ΔHdegf = 82254 plusmn 5 kJmol) was first estimated as the average valuebetween that calculated according to Chermak ampRimstidt (1989) (minus82204 kJmol) and that(minus82304 kJmol) calculted as ΔHdegf (Mg-sudoite) ndashΔHdegf (corundum)2 + ΔHdegf (hematite)2 using thevalues of Vidal et al (1992) for sudoite and Berman(1988) for hematite

(3) In a first step ΔHdegf and Sdeg of ferri-sudoite wereadjusted in order to reproduce the temperaturesreported in Fig 7 with oxygen-free reactions andassuming that magnetite is present in all samples(Chl-Qtz-Mt assemblage) ΔHdegf was allowed to varybetweenminus8220 andminus8230 kJmol and Sdeg within 10 JmolK of the estimated value The addition ofmagnetite and ferri-sudoite to quartz and water leadsto 17 additional equilibria and the following wereused to constrain ΔHdegf and Sdeg of ferri-sudoite

ClinthornMg-Sud frac14 Mg-Amthorn Qtzthorn H2O (1)

4 Clinthorn Daphthorn 10 ferri-Sud

frac14 10 Mg-Amthorn 25 Qtzthorn 30 Mtthorn 20 H2O(24)

7 Daphthorn 70 ferri-Sudthorn 30 Mg-Am

frac14 22 Clinthorn 75 Mg-Sudthorn 35 Mtthorn 40 H2O(25)

28 ferri-Sudthorn 11 Fe-AmthornMg-Am

frac14 6 Daphthorn 30 Mg-Sudthorn 14 Mtthorn 16 H2O(26)

Daphthorn 10 ferri-Sudthorn 10 Mg-Amthorn 10 Qtz

frac14 6 Clinthorn 15 Mg-Sudthorn 5 Mtthorn 16 H2O(27)

The ferri-Sud properties derived with the assumptionof ChlndashQtzndashMt equilibrium are listed in Table 1(referenced a)With these values the oxygen activity atthe P T and (XFe3+)Chl of all equilibria used tocompute Fig 7 can be calculated with the followingoxygen-bearing and magnetite-free equilibria whichconverge with equilibria 1 8 9 and 10 at the same PTand aO2 conditions

16 Qtzthorn 7 Mg-Amthorn 4 Daphthorn 6 Mg-Sud

thorn 5 O2 thorn 12 H2O frac14 20 ferri-Sud(28)

Fe-Amthorn 2 Mg-Amthorn 6 Qtzthorn 4 H2Othorn O2

frac14 4 ferri-Sud(29)

Mg-Amthorn 4 Daphthorn 6 Mg-Sudthorn 2 O2

frac14 3 Fe-Amthorn 8 ferri-Sudthorn 2 Qtz(30)

5 Mg-Amthorn 12 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 8 Fe-Amthorn 28 ferri-Sud

(31)

8 Daphthorn 12 Mg-Sudthorn 3 O2 frac14 7 Fe-Am

thorn 12 ferri-Sudthorn 10 Qtzthorn 4 H2O(32)

6 Clinthorn 4 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Mg-Amthorn 20 ferri-Sudthorn 5 Qtz(33)

4 Clinthorn 20 ferri-Sud frac14 30 Qtz

thorn 15 Mg-Amthorn 4 Daphthorn 5 O2 thorn 20 H2O(34)

90 Mg-Sudthorn 28 Daphthorn 32 Clinthorn 20 H2Othorn 35 O2

frac14 140 ferri-Sudthorn 15 Mg-Am

(35)

15 Qtzthorn 14 Clinthorn 16 Daphthorn 45 Mg-Sud

thorn 20 O2 thorn 20 H2O frac14 80 ferri-Sud(36)

2 Clinthorn FeAmthorn 3 Mg-Sudthorn O2

frac14 2 Mg-Amthorn 4 ferri-Sudthorn Qtz(37)

12 Clinthorn 7 Fe-Amthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 10 Mg-Amthorn 28 ferri-Sud

(38)

5 Qtzthorn 2 Clinthorn 2 Fe-Amthorn 3 Mg-Sud

thorn 2 O2 thorn 4 H2O frac14 8 ferri-Sud(39)

2 Clinthorn 8 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Fe-Amthorn 20 ferri-Sudthorn 5 Qtz(40)

30 Qtzthorn 8 Clinthorn 15 Fe-Amthorn 5 O2 thorn 20 H2O

frac14 8 Daphthorn 20 ferri-Sud

(41)

4 Clinthorn 8 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 3 Fe-Amthorn 28 ferri-Sud

(42)

Because the ferri-sudoite standard-state propertieswere constrained by the chlorite-quartz-magnetite-O2-H2O equilibrium the Log(aO2) calculated withthe equilibria above follow naturally the trend shown inFig 8 and the Log(aO2) values calculated withreactions 28ndash42 of low-temperature samples plot inthe hematite stability field This suggests that theΔGdegPT values of ferri-sudoite derived above with theassumption that aO2 is buffered by the presence ofmagnetite are possibly too high A reduction of ΔHdegf(ferri-sudoite) to 82304 kJmol (=ΔHdegf (Mg-sudoite)ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2 see Table 1)leads to Log(aO2) reduced by 45 Log units (see Fig 8)

628 O Vidal et al

and most calculated values except those of thehematite-bearing St Martin samples plot in themagnetite stability field

A more detailed study is required to refine the ferri-sudoite thermodynamic data However the resultsobtained using the preliminary values listed in Table 1show that this end-member can be used to reproduceconsistently the T-XFe3+ and T- trends obtained fromclassical Chl-Qtz-H2O calculations (Figs 5ndash7) Theincrease in XFe3+ with reducing temperature is consist-ent with the expected reduction in aO2 and the amountof free oxygen that is consumed to produce Fe3+-bearingchlorite Ferric iron is therefore a major component thathas to be taken into account when dealing with low-temperature samples Empirical or semi-empiricalchlorite thermometers which assume that all Fe is inthe ferrous state may include one of the XFe3+-T andaO2-T trends illustrated in Figs 7 and 8 but they cannotaccount for the different XFe3+-T trends that areexpected in the contexts of different aO2 conditions

FUTURE CHALLENGES LOOK INGAT LOW-GRADE ROCKS SUCH ASMETAMORPH IC ROCKS AND THE

DYNAMICS OF CHLOR ITECRYSTALL IZAT ION AT THE

MICROMETR IC TO NANOMETR ICSCALE

T-P-XFe3+ mapping at the thin-section scaleevidence of chlorite composition re-equilibration in metamorphic and low-gradesamples

Although natural samples offer valuable constraintson thermodynamic models there are two maindifficulties in using them to evaluate the reliability ofexisting chlorite models for thermometric purposes

(1) In the case of low-grade samples referencepressures and temperatures are generally known withina reasonable absolute uncertainty but the size ofchlorite grains is often small and good microprobeanalyses are sometimes difficult to obtain Moreoverthe kinetics of reaction are so sluggish that inheritedchlorite grains might remain unaffected by latervariations in temperature which makes thetemperature-composition relationships difficult toconstrain Lastly extrapolation at HT and HP ofempirical or semi-empirical models constrained byLT samples only may be erroneous in particular theeffect of pressure or eventual systematic variations ofXFe3+ cannot be appraised

(2) In the case of metamorphic samplesmicroprobe analyses are easier to obtain from largechlorite grains but chlorites of contrasted compositionthat crystallized at different temperatures and times ofthe metamorphic history are often observed to coexistmetastably in different microstructural sites of the samethin section In such cases the definition of a singlepair of reference temperature and pressure isinappropriate and cannot be restricted to the peak-temperature conditions

In order to overcome these difficulties and to bettervisualize the link between mineral compositiondeformation and P-T conditions at the thin-sectionscale de Andrade et al (2006 2011) Vidal et al(2006) and Lanari et al (2013) developed an approachwhich combines elementary maps of composition witheither a thermodynamic or a semi-empirical estimate ofchlorite crystallization temperature Maps of rawmicroprobe analyses were standardized using good-quality point analyses of minerals present in themapped area The maps of standardized analyseswere transformed into maps of structural formulaewhere each pixel analysed is converted into individualmineral structural formulae Depending on the thermo-metric approach it was then possible to compute mapsof temperature or T-XFe3+ conditions assuming thateach chlorite pixel was in equilibrium with quartz andwater The results reported by Vidal et al (2006)Yamato et al (2007) Ganne et al (2012) Lanari et al(2013 2014) and Scheffer et al (2015) showed a closelink between deformation and mineral reaction at themicroscopic scale and maps of chlorite temperaturecould be used as a basis for interpreting the micro-structures observed in thin sections of low- to middle-grade metapelites In all these studies the metastablepersistence of different generations of chlorite crystal-lizing at different temperatures during the exhumationof metamorphic samples could be used to constrain theT-deformation history Moreover the map of (XFe3+)Chl calculated by Vidal et al (2006) was in goodqualitative agreement with the in situ mapping of thisratio using the X-ray absorption near-edge structure(XANES) techniques (Fig 9) (see also Munoz et al2006) The calculated temperature and XFe3+ mapsobtained on a single sample showed the same trend ofreducing (XFe3+)Chl with increasing temperature asobserved in Figs 6 and 7

Bourdelle et al (2013c) applied a similar approachfor low-temperature samples with chlorite of smallgrain size using transmission electron microscopy-energy dispersive X-ray (TEM-EDX) spectroscopicanalyses of focused-ion beam (FIB) foils instead of

629Conditions of formation of chlorite

microprobe analyses of thin sections (Fig 10) Thoseauthors were able to observe heterogeneities in XFe3+

at the nm scale using STXM-XANES measurementsThis last study was particularly interesting as it showedthat approaches similar to those used for studyingmetamorphic rocks could be applied to low-grade anddiagenetic samples

Looking at diagenetic rocks like metamorphicrocks

The use of equilibrium thermodynamics to studylow-grade rocks has been often criticized becausesluggish kinetics might preclude achievement of globalequilibrium at low temperature and is responsible forthe crystallization of metastable phases with a lowerenergy of nucleation than stable ones Essene amp Peacor(1995) claimed that the coexistence of chemicallyinhomogeneous phases at a very fine scale in low-grade rocks is indicative of disequilibrium and thatmineral reaction is better described by a kinetic ratherthan thermodynamic approach However the coexist-ence of chemically inhomogeneous phases is also acommon observation in metamorphic rocks wherezoned grains and the coexistence for example ofchlorite grains of different compositions indicate thatthermodynamic equilibrium is approached only locallyand at a given time in the metamorphic history In factthe main difference between diagenetic and higher-temperature rocks is the size of the domains in localequilibrium The more sluggish kinetics of reactionand diffusivity at low temperature lead to small grainsand domains in equilibrium but these domains can beidentified when using appropriate analytical tools andprocedures developed recently (Bourdelle et al2013bc)

FIG 10 Quantitative Fe redox nanomapping on a FIB foilof Glarus sample (T lt 300degC Switzerland) calculatedfrom the 7087ndash71025 eV images ratio (STXM-XANES)

(from Bourdelle et al 2013c)

FIG 11 Analysis of an illite-chlorite assemblage (Gulf Coast sample Alamo 1 well 5825 m) (a) SEM image of thepetrographic thin section showing the emplacement of the FIB foil to be cut across the illitendashchlorite interface (b)Bright-field TEM image of the FIB foil extracted from the section with the platinum strap on top of it (cd) Enlargementof part b showing the Al and Mg contents of chlorite (apfu) and their variation between crystal core and crystal rim

630 O Vidal et al

Figure 10 shows that it is possible to demonstrateheterogeneities in XFe3+ contents at the nanometerscale Consistent variations of other major elements arealso observed with TEM technics on FIB foils Anexample is shown in Fig 11 which depicts thevariations of Al and Mg contents in a chlorite grainfrom the Gulf Coast (Bourdelle et al 2013b) In thiscase the FIB foil was cut parallel to the interfacebetween illite and chlorite and a nm-scale investiga-tion of composition was conducted by TEM-EDXanalysis The rim and core compositions showed azoning of chlorite with Mg- and Al-richer rims thancore compositions This variation in composition on ascale of lt1 μm was related to various crystallizationstages during burial Chlorite sampled at differentdepths and various locations of the Gulf Coast areashowed that zoned compositions similar to thatobserved in Fig 11 are frequent (Bourdelle et al2013b) The AlIV in chlorite rims display a Tdependence that is consistent with the km-scale andregional evolution of temperature ie with thediagenetic history of Gulf Coast (burial P-T condi-tions detritalauthigenic compositions etchellip)(Fig 12) This study suggests again that approachesused for metamorphic rocks can be applied to low-grade rocks when using adapted analytical tools able toidentify sub-μm variations of composition

CONCLUS ION

The variability of chlorite composition reflects thevariability of its conditions of formation Numerousefforts have been made over the last 30 years to

translate the composition of chlorite in terms of itstemperature of formation The first empirical equationsbased on the AlIVor vacancy contents are restricted tobulk-rock compositions and pressure conditions atwhich they were calibrated Thermodynamicapproaches and semi-empirical equations based onthe ratio of end-member activities do not present thislimit Among the different thermodynamic models ofchlorite published so far Vidal et al (2006) proposed amodel that can be used to estimate T and (XFe3+)Chlsimultaneously The estimated (XFe3+)Chl values areconsistent with those measured by Moumlssbauer orXANES spectroscopy and for the same (XFe3+)Chlthe temperatures estimated using the method of Vidalet al (2006) are similar to those obtained with thesemi-empirical equations of Inoue et al (2009) andLanari et al (2014) Different T-XFe3+ trends areobserved for different pressure gradients and oxygen-buffering assemblages (Lanari et al 2014) butoverall the systematic increase in XFe3+ with areduction in temperature is compatible with thedecrease in aO2 in a closed system The increase inXFe3+ and chlorite vacancy with decreasing tempera-ture is interpreted as reflecting the incorporation of theFe3+-sudoite component which can be used toestimate T-aO2-XFe

3+ values with a Chl-Qtz-H2Omulti-equilibrium approach

Similar temperatures are calculated for low-tem-perature chlorites (T lt 300degC and measured XFe3+ upto 30) with all approaches assuming that iron inchlorite is either completely ferrous or with estimatedvalues of XFe3+ This indicates that the equations ofsemi-empirical thermometers assuming that all iron inchlorite is ferrous include some inherent Fe2+Fe3+ vstemperature correction However very different tem-peratures can be obtained for higher-temperaturesamples (gt300degC) with approaches involving ferriciron or assuming that all of the iron is ferrous Thissuggests that the Fe2+Fe3+ vs temperature correctionmight be restricted to particular aO2 conditions andpressure gradients More work is necessary to betterunderstand this point and to improve the thermo-dynamic models and semi-empirical thermometersThis requires high-quality and well constrained naturaldata To this end mapping techniques that highlight thelink between the variations in temperature and pressureand small-scale chlorite composition including XFe3+

variations offer much promise When combining theresults of the present study with those of other studiespublished recently it appears that thermodynamicapproaches and mapping techniques developed formetamorphic rocks could be used to study low-grade

FIG 12 Range of crystal-rim AlIV content (white boxes)of Gulf Coast chlorites vs measured temperatures Thegrey area underlines the range of maximum AlIV contentwhich is assumed to represent the closest approach toequilibrium composition (after Bourdelle et al 2013b)

631Conditions of formation of chlorite

rocks where kinetics are much more sluggish Thisrequires use of appropriate analytical tools and techni-ques with a spatial resolution of just a few hundrednanometres which are now becoming available

ACKNOWLEDGMENTS

Constructive reviews by S Hillier F Nieto and ananonymous reviewer are gratefully acknowledged

REFERENCES

Beaufort D Patrier P Meunier A amp Ottaviani MM(1992) Chemical variations in assemblages includingepidote andor chlorite in the fossil hydrothermalsystem of Saint Martin (Lesser Antilles) Journal ofVolcanology and Geothermal Research 51 95ndash114

Berman R (1988) Internally-consistent thermodynamicdata for minerals in the system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2Journal of Petrology 29 445ndash522

Berman R amp Brown T (1985) Heat capacity of minerals inthe system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2 representation estima-tion and high temperature extrapolation Contributionsto Mineralogy and Petrology 89 168ndash183

Bourdelle F amp Cathelineau M (2015) Low-temperaturechlorite geothermometry a graphical representationbased on a TndashR2+ndashSi diagram European Journal ofMineralogy 27 617ndash626

Bourdelle F Parra T Chopin C amp Beyssac O (2013a) Anew chlorite geothermometer for diagenetic to low-grade metamorphic conditions Contributions toMineralogy and Petrology 165 723ndash735

Bourdelle F Parra T Beyssac O Chopin C amp Vidal O(2013b) Clay minerals thermometry a comparativestudy based on high-resolution chemical analyses ofillite and chlorite in sandstones from the Gulf Coast(Texas USA) American Mineralogist 98 914ndash926

Bourdelle F Benzerara K Beyssac O Cosmidis JNeuville DR Brown GE amp Paineau E (2013c)Quantification of the ferric ferrous iron ratio insilicates by scanning transmission X-ray microscopyat the Fe L23 edges Contribution to Mineralogy andPetrology 166 423ndash434

Cathelineau M (1988) Cation site occupancy in chloritesand illites as function of temperature Clay Minerals23 471ndash485

Cathelineau M amp Nieva D (1985) A chlorite solidsolution geothermometer the Los Azufres (Mexico)geothermal system Contributions to Mineralogy andPetrology 91 235ndash244

Chermak JA amp Rimstidt JD (1989) Estimating thethermodynamic properties (ΔGdegf and ΔHdegf) of silicate

minerals at 298K from the sum of polyhedral contribu-tions American Mineralogist 74 1023ndash1031

Coggon R amp Holland TJB (2002) Mixing properties ofphengitic micas and revised garnent-phengite thermo-barometers Journal of Metamorphic Geology 20683ndash696

Curtis CD Hughes CR Whiteman JA ampWhittle CK(1985) Compositional variation within some sedi-mentary chlorites and some comments on their originMineralogical Magazine 49 375ndash386

de Andrade V Vidal O Lewin E OrsquoBrien P amp Agard P(2006) Quantification of electron microprobe com-positional maps of rock thin sections an optimizedmethod and examples Journal of MetamorphicGeology 24 655ndash668

de Andrade V Susini J Salomeacute M Beraldin O RigaultC Heymes T Lewin E amp Vidal O (2011)Submicrometer hyperspectral X-ray imaging of het-erogeneous rocks and geomaterials applications at theFe k-edge Analytical Chemistry 83 4220ndash4227

de Caritat P Hutcheon I amp Walshe J (1993) Chloritegeothermometry a review Clays and Clay Minerals41 219ndash239

Essene EJ (2009) Thermobarometry gone astray Pp101ndash129 in Physics and Chemistry of Earthrsquos Interior(AK Gupta amp S Dasgupta editors) PlatinumJubilee Indian National Science Academy Springer-Verlag Berlin

Essene EJ amp Peacor DR (1995) Clay mineralthermometry ndash a critical perspective Clays and ClayMinerals 43 540ndash553

Ganne J de Andrade V Weinberg RF Vidal O DubacqB Kagambega N Naba S Baratoux L Jessell M ampAllibon J (2012) Modern-style plate subductionpreserved in the Palaeoproterozoic West Africancraton Nature Geoscience 5 60ndash65

Grosch EG Vidal O Abu-Alam T amp McLoughlin N(2012) PTconstraints on the metamorphic evolution ofthe Paleoarchean Kromberg type-section BarbertonGreenstone Belt South-Africa Journal of Petrology53 513ndash545

Grosch E McLoughlin N Lanari P Erambert M ampVidal O (2014) Microscale mapping of alterationconditions and potential biosignatures in basaltic-ultramafic rocks on early Earth and beyondAstrobiology 14 216ndash228

Hillier S amp Velde B (1991) Octahedral occupancy andchemical composition of diagenetic (low-temperature)chlorites Clay Minerals 26 149ndash168

Holland TJB (1989) Dependance of entropy on volumefor silicates and oxide minerals A review and apredictive model American Mineralogist 74 5ndash13

Holland T amp Powell R (1998) An internally consistentthermodynamic data set for phases of petrologicalinterest Journal ofMetamorphic Geology 16 309ndash343

Holland T Baker J amp Powell R (1998)Mixing propertiesand activity composition and relationships of chlorites

632 O Vidal et al

FIG 5 Left Plot of temperatures estimated with equations 38 (TFe3+ Lanari) and 40 (TFe2+ Lanari) of Lanari et al(2014) and with the thermometer of Inoue et al (2009) (TFe3+ Inoue) vs temperatures estimated using the approach ofVidal et al (2006) TFe3+ Lanari and TFe3+ Inoue were calculated using the XFe3+ values obtained with the approach ofVidal et al (2006) and TFe2+ Lanari assuming XFe3+ = 0 Right Plot of vacancy content vs estimated temperatures(a) and (b) are high-pressure (gt6 kbar) and low- to medium-pressure (lt6 kbar) respectively (c) shows the resultsobtained for the low-pressure samples of Grosch et al (2012) The unrealistically high TFe2+ Lanari are obtained forchlorites with lt005 vacancy pfu (dashed line in the vacancy-temperature plots) The histograms show the repartition

of temperatures estimated with the different approaches

623Conditions of formation of chlorite

parameters of Vidal et al (2006) is possible butthis should be demonstrated by a systematic compari-son of measured and estimated XFe3+ from naturalsamples covering a wide range of P T and fO2

conditions Such a study is still lacking but a first-order comparison of measured and estimated XFe3+

contents for samples of known temperature ofcrystallization is shown in Fig 7 which was obtainedfor the Rouez Niger and Saint-Martin low-T samplescompiled by Inoue et al (2009) the Glarus sample ofBourdelle et al (2013c) and a blueschist-facies sampleretrogressed into the greeschist facies from Sambagawa(de Andrade et al 2006 Munoz et al 2006 Vidalet al 2006) T-XFe3+ values were calculated at 2 kbarexcept for the Sambagawa sample for which calcula-tions were made at 2 5 and 10 kbar for the low-medium- and high-temperature chlorite groups identi-fied by de Andrade et al (2006) and Vidal et al (2006)respectively

FIG 6 Estimated TFe3+ Vidal and XFe3+ pairs for low- andmedium-pressure samples (open circles) high-pressuresamples (black circles) and the samples from Grosch et al

(2012) (grey squares)

FIG 7 Boxes measured XFe3+ and reference temperatures Symbols Tmaximum-XFe3+minimum pairs calculated to

obtain convergence (plusmn30degC) of equilibria 1 8 9 and 10 with the thermodynamic model of Vidal et al (2006)

624 O Vidal et al

For each series of Inoue et al (2009) between 50and 60 of the chlorite analyses contain Si lt3 apfuand led to a convergence of equations 1 8 9 and 10within 30degC The calculated T-XFe3+ pairs are plottedin Fig 7 and compared with the range of referencetemperature (TRef) estimated independently and XFe3+

measures reported by Inoue et al (2009) A fairagreement of the measured and calculated TFe3+ Vidal-XFe3+ values is observed for the Niger and St-Martinanalyses and a good agreement of the TRef and TFe3+ Vidal was observed for the Rouez samples but thecalculated valueswere smaller than themeasured valuesfor XFe3+ for this samples However the XFe3+ valuesreported in Fig 7 are minimum values required toobtain convergence of equilibria 1 and 8ndash10 within30degC Convergence is maintained at higher valuesof XFe3+ and TFe3+ Vidal for the Rouez samples arestill located in the range of TRef when XFe3+ is setat gt15 (lower limit of the reference values) TheTFe3+ Vidal values calculated for the Glarus chloriteanalyses listed in Bourdelle et al (2013c) are 139degC 203degC and 224degC in fair agreement with thoseestimated independently by Bourdelle et al (2013c)(135degC 170degC and 182degC respectively) or with thesemi-empirical thermometer of Inoue et al (2009)(145degC 196degC and 210degC respectively) For thethree samples studied the XFe3+ values measuredusing XANES (X-ray absorption near edge struc-ture) spectra in the transmission mode of scanningtransmission X-ray microscopy (STXM) were foundto be 223 238 and 279 These values comparefavourably with the estimated ones (32 27 and 28respectively) although the measured values suggestan increase in XFe3+ with temperature which iscontrary the trend inferred from the thermodynamicestimates and evidenced in Figs 6 and 7 Finally therange of XFe3+ estimated for the different generationsof chlorite in the Sambagawa sample (ellipses inFig 7) are slightly higher but still in fair agreementwith those inferred from the XANES measurementsIn conclusion Fig 7 shows that the values estimatedon the basis of the convergence of equilibria (1 8 9and 10) are consistent with the TRef and measuredXFe3+ for various samples covering a range oftemperature from 100degC to 500degC pressure from sim2to 10 kbar and XFe3+ values from 5 to 35 Boththe measured and estimated XFe3+ values plotted inFig 7 seem to confirm the systematic decrease ofXFe3+ with increasing temperature and increase withincreasing pressure that was observed in Fig 6 Thistrend is similar to the decrease in vacancy observedin Fig 5

aO2 -TEMPERATURE EST IMATESAND ADDIT ION OF A FERR I -

SUDOITE END -MEMBER

The general trends inferred from our compilation ofnatural samples indicate that XFe3+ increases withdecreasing temperature and increasing pressureAlthough the model of Vidal et al (2006) involvesno Fe3+-bearing end-member the oxygen activity(aO2) prevailing during the formation of chloritesused to compute Fig 7 can be estimated assuming thatit was buffered by the presence of pure magnetite (Mt)in equilibrium with chlorite and quartz Following thisassumption aO2 for each sample has been calculatedusing equilibria (1 8 9 10) and the followingadditional equilibria which intersect at the same P T(plusmn30degC) (XFe3+)Chl and Log(aO2) (plusmn2) (Fig 8)

15 Mg-Sudthorn 40 Mtthorn 45 Qtzthorn 60 H2O

frac14 6 Clinthorn 24 Daphthorn 20 O2

(11)

6 Mg-Sudthorn 10 Mtthorn 6 Qtzthorn 12 H2O

frac14 3 Mg-Amthorn 6 Daphthorn 5 O2

(12)

15 Mg-Sudthorn 6 Clinthorn 10 Mt

frac14 15 Qtzthorn 15 Mg-Amthorn 6 Daphthorn 5 O2

(13)

6 Daphthorn 4 Clinthorn 5 O2

frac14 20 Qtzthorn 10 Mtthorn 5 Mg-Amthorn 20 H2O(14)

60 Mg-Sudthorn 12 Clinthorn 70 Mtthorn 60 H2O

frac14 45 Mg-Amthorn 42 Daphthorn 35 O2

(15)

4 Qtzthorn 2 Mtthorn Fe-Amthorn 4 H2O

frac14 2 Daphthorn O2

(16)

15 Mg-Sudthorn 6 Daphthorn 10 Mt

frac14 15 Qtzthorn 6 Clinthorn 15 Fe-Amthorn 5 O2

(17)

60 Mg-Sudthorn 70 Mtthorn 60 H2O

frac14 24 Clinthorn 45 Fe-Amthorn 6 Daphthorn 35 O2

(18)

15 Mg-Sudthorn 16 Mtthorn 12 H2O

frac14 3 Qtzthorn 6 Clinthorn 12 Fe-Amthorn 8 O2

(19)

6 Mg-Sudthorn 4 Mt

frac14 6 Qtzthorn 3 Mg-Amthorn 3 Fe-Amthorn 2 O2

(20)

12 Mg-Sudthorn 14 Mtthorn 12 H2O

frac14 6 Mg-Amthorn 3 Fe-Amthorn 6 Daphthorn 7 O2

(21)

5 Mg-Amthorn 4 Mtthorn 8 Qtzthorn 8 H2O

frac14 4 Clinthorn 3 Fe-Amthorn 2 O2

(22)

24 Mg-Sudthorn 3 Mg-Amthorn 28 Mtthorn 24 H2O

frac14 12 Clinthorn 21 Fe-Amthorn 14 O2

(23)

All calculated Log(aO2) except those of the Saint-

625Conditions of formation of chlorite

Martin samples follow the same trend and decreasewith decreasing temperature from simminus27 at 450ndash500degC10 kbar (high-temperature chlorites of Sambagawa) tominus50degC at 125degC and 1 kbar (Niger samples) (Fig 8)The Log(aO2) values of the high-T and -P samples plotin the stability field of magnetite whereas those of thelow-T samples plot in the stability field of hematitewhich is not consistent with the assumption of ChlndashMtndashQtzndashH2O equilibrium The presence of hematite hasbeen reported in the Saint-Martin samples only(Beaufort et al 1992) which plot above the trenddefined by the other samples (upper grey ellipse inFig 8) The relative position of this hematite-bearingsample is consistent with expected higher oxygenactivities required for its formation For the other low-T and apparently hematite-free samples it is possiblethat the calculated Log(aO2) values are too high or thatoxygen was not buffered by an equilibrium ofmagnetitewith chlorite In any case Fig 8 indicates that thesystematic increase in (XFe3+)Chl with decreasingtemperature observed in Figs 6 and 7 is consistentwith a lower aO2 and amount of free oxygen at low-Tthan high-T in systems where aO2 is buffered by thepresence of chlorite and iron oxides To maintain this

equilibrium with lowering temperature in metamorphicrocks where oxygen is immobile oxygen has to beconsumed and one way for that is to produce chloritethat is richer in Fe3+ This is a possible interpretationfor the formation of low-temperature Fe3+-rich chloritecoexisting with high-temperature Fe3+-poor chloritein the same thin section of the Sambagwa sample(Fig 9)

In order to discuss the Log(aO2)-T-XFe3+ evolution

in more detail for rocks where aO2 is not buffered bythe presence of magnetite it is necessary to define anFe3+-bearing chlorite end-member The simultaneousincrease of vacancy and Fe3+ content with decreasingtemperature observed in Figs 5ndash7 is explained by the3(Mg Fe2+) = 2(Fe3+) ditrioctahedral substitution(see also Trincal amp Lanari 2016 this issue) Thisexchange can be modelled with a ferri-sudoite end-member that materializes the relationship betweenXFe3+ and vacancy The standard-state thermo-dynamic properties of ferri-sudoite were constrainedas follows

(1) Because the maximum amount of (Fe3+)Chlcalculated in the previous section never exceeded1 apfu the cation distribution of Vidal et al (2006)

FIG 8 Left Ranges of Log(aO2) conditions calculated for the samples used to compute Fig 7 at the temperature andXFe3+ contents estimated from the Chl-Qtz-H2O equilibrium conditions Dashed lines hematite-magnetite-O2

equilibrium calculated using the thermodynamic properties of Berman (1988) and unit activities of Ht and Mt RightSample calculation at 2 kbar for a Rouez sample using reactions 8ndash20 (upper) or 26ndash40 (lower) The vertical linescorrespond to oxygen- and magnetite-free reactions 1 5 6 and 7 used to compute Figs 5ndash7 which converge at +30degCThe circles in the lower figure represent values calculated with (ΔHdegf ) ferri-Sud constrained with the chlorite-magnetite-quartz-water equilibria 8ndash20 (open circle) or estimated as ΔHdegf (Mg-sudoite) ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2

(filled circle) (see Table 1 and text)

626 O Vidal et al

was kept unchanged and the Fe3+-Al3+ exchange wasassumed to be restricted to the M4 site of chlorite Thechosen formula of ferri-Sud was thus ([Si2]T1[SiAl]T2)

IV([]M1[Mg2Al2]M2-3[Fe3+]M4)

VIO10(OH)8The standard-state thermodynamic properties of allother chlorite end-members the activity model and theMargules parameters were kept unchanged

(2) The third-law entropy of ferri-sudoite (Sdeg =415 JmolK) was estimated according to Holland(1989) with a configurational entropy of 2305 JmolKand a ferromagnetic contribution of 1304 JmolK itsvolume (Vdeg = 204 Jb) was estimated to be 01 Jblower than that of Mg-sudoite (2051 Jb) and the Cp(T ) function was estimated according to Berman amp

FIG 9 Rock-sample mapping of a Sambagawa sample (a) maps of estimated XFe3+ (Vidal et al 2006 Munoz et al2006) and Log(aO2) (estimated assuming that it is buffered by magnetite-chlorite equilibria see text and Fig 8)(b) measured Fe XANES edge-position proportional to Fe redox (Munoz et al 2006) (c) estimated temperature (Vidal

et al 2006 Munoz et al 2006)

627Conditions of formation of chlorite

Brown (1985) The formation enthalpy (ΔHdegf = 82254 plusmn 5 kJmol) was first estimated as the average valuebetween that calculated according to Chermak ampRimstidt (1989) (minus82204 kJmol) and that(minus82304 kJmol) calculted as ΔHdegf (Mg-sudoite) ndashΔHdegf (corundum)2 + ΔHdegf (hematite)2 using thevalues of Vidal et al (1992) for sudoite and Berman(1988) for hematite

(3) In a first step ΔHdegf and Sdeg of ferri-sudoite wereadjusted in order to reproduce the temperaturesreported in Fig 7 with oxygen-free reactions andassuming that magnetite is present in all samples(Chl-Qtz-Mt assemblage) ΔHdegf was allowed to varybetweenminus8220 andminus8230 kJmol and Sdeg within 10 JmolK of the estimated value The addition ofmagnetite and ferri-sudoite to quartz and water leadsto 17 additional equilibria and the following wereused to constrain ΔHdegf and Sdeg of ferri-sudoite

ClinthornMg-Sud frac14 Mg-Amthorn Qtzthorn H2O (1)

4 Clinthorn Daphthorn 10 ferri-Sud

frac14 10 Mg-Amthorn 25 Qtzthorn 30 Mtthorn 20 H2O(24)

7 Daphthorn 70 ferri-Sudthorn 30 Mg-Am

frac14 22 Clinthorn 75 Mg-Sudthorn 35 Mtthorn 40 H2O(25)

28 ferri-Sudthorn 11 Fe-AmthornMg-Am

frac14 6 Daphthorn 30 Mg-Sudthorn 14 Mtthorn 16 H2O(26)

Daphthorn 10 ferri-Sudthorn 10 Mg-Amthorn 10 Qtz

frac14 6 Clinthorn 15 Mg-Sudthorn 5 Mtthorn 16 H2O(27)

The ferri-Sud properties derived with the assumptionof ChlndashQtzndashMt equilibrium are listed in Table 1(referenced a)With these values the oxygen activity atthe P T and (XFe3+)Chl of all equilibria used tocompute Fig 7 can be calculated with the followingoxygen-bearing and magnetite-free equilibria whichconverge with equilibria 1 8 9 and 10 at the same PTand aO2 conditions

16 Qtzthorn 7 Mg-Amthorn 4 Daphthorn 6 Mg-Sud

thorn 5 O2 thorn 12 H2O frac14 20 ferri-Sud(28)

Fe-Amthorn 2 Mg-Amthorn 6 Qtzthorn 4 H2Othorn O2

frac14 4 ferri-Sud(29)

Mg-Amthorn 4 Daphthorn 6 Mg-Sudthorn 2 O2

frac14 3 Fe-Amthorn 8 ferri-Sudthorn 2 Qtz(30)

5 Mg-Amthorn 12 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 8 Fe-Amthorn 28 ferri-Sud

(31)

8 Daphthorn 12 Mg-Sudthorn 3 O2 frac14 7 Fe-Am

thorn 12 ferri-Sudthorn 10 Qtzthorn 4 H2O(32)

6 Clinthorn 4 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Mg-Amthorn 20 ferri-Sudthorn 5 Qtz(33)

4 Clinthorn 20 ferri-Sud frac14 30 Qtz

thorn 15 Mg-Amthorn 4 Daphthorn 5 O2 thorn 20 H2O(34)

90 Mg-Sudthorn 28 Daphthorn 32 Clinthorn 20 H2Othorn 35 O2

frac14 140 ferri-Sudthorn 15 Mg-Am

(35)

15 Qtzthorn 14 Clinthorn 16 Daphthorn 45 Mg-Sud

thorn 20 O2 thorn 20 H2O frac14 80 ferri-Sud(36)

2 Clinthorn FeAmthorn 3 Mg-Sudthorn O2

frac14 2 Mg-Amthorn 4 ferri-Sudthorn Qtz(37)

12 Clinthorn 7 Fe-Amthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 10 Mg-Amthorn 28 ferri-Sud

(38)

5 Qtzthorn 2 Clinthorn 2 Fe-Amthorn 3 Mg-Sud

thorn 2 O2 thorn 4 H2O frac14 8 ferri-Sud(39)

2 Clinthorn 8 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Fe-Amthorn 20 ferri-Sudthorn 5 Qtz(40)

30 Qtzthorn 8 Clinthorn 15 Fe-Amthorn 5 O2 thorn 20 H2O

frac14 8 Daphthorn 20 ferri-Sud

(41)

4 Clinthorn 8 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 3 Fe-Amthorn 28 ferri-Sud

(42)

Because the ferri-sudoite standard-state propertieswere constrained by the chlorite-quartz-magnetite-O2-H2O equilibrium the Log(aO2) calculated withthe equilibria above follow naturally the trend shown inFig 8 and the Log(aO2) values calculated withreactions 28ndash42 of low-temperature samples plot inthe hematite stability field This suggests that theΔGdegPT values of ferri-sudoite derived above with theassumption that aO2 is buffered by the presence ofmagnetite are possibly too high A reduction of ΔHdegf(ferri-sudoite) to 82304 kJmol (=ΔHdegf (Mg-sudoite)ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2 see Table 1)leads to Log(aO2) reduced by 45 Log units (see Fig 8)

628 O Vidal et al

and most calculated values except those of thehematite-bearing St Martin samples plot in themagnetite stability field

A more detailed study is required to refine the ferri-sudoite thermodynamic data However the resultsobtained using the preliminary values listed in Table 1show that this end-member can be used to reproduceconsistently the T-XFe3+ and T- trends obtained fromclassical Chl-Qtz-H2O calculations (Figs 5ndash7) Theincrease in XFe3+ with reducing temperature is consist-ent with the expected reduction in aO2 and the amountof free oxygen that is consumed to produce Fe3+-bearingchlorite Ferric iron is therefore a major component thathas to be taken into account when dealing with low-temperature samples Empirical or semi-empiricalchlorite thermometers which assume that all Fe is inthe ferrous state may include one of the XFe3+-T andaO2-T trends illustrated in Figs 7 and 8 but they cannotaccount for the different XFe3+-T trends that areexpected in the contexts of different aO2 conditions

FUTURE CHALLENGES LOOK INGAT LOW-GRADE ROCKS SUCH ASMETAMORPH IC ROCKS AND THE

DYNAMICS OF CHLOR ITECRYSTALL IZAT ION AT THE

MICROMETR IC TO NANOMETR ICSCALE

T-P-XFe3+ mapping at the thin-section scaleevidence of chlorite composition re-equilibration in metamorphic and low-gradesamples

Although natural samples offer valuable constraintson thermodynamic models there are two maindifficulties in using them to evaluate the reliability ofexisting chlorite models for thermometric purposes

(1) In the case of low-grade samples referencepressures and temperatures are generally known withina reasonable absolute uncertainty but the size ofchlorite grains is often small and good microprobeanalyses are sometimes difficult to obtain Moreoverthe kinetics of reaction are so sluggish that inheritedchlorite grains might remain unaffected by latervariations in temperature which makes thetemperature-composition relationships difficult toconstrain Lastly extrapolation at HT and HP ofempirical or semi-empirical models constrained byLT samples only may be erroneous in particular theeffect of pressure or eventual systematic variations ofXFe3+ cannot be appraised

(2) In the case of metamorphic samplesmicroprobe analyses are easier to obtain from largechlorite grains but chlorites of contrasted compositionthat crystallized at different temperatures and times ofthe metamorphic history are often observed to coexistmetastably in different microstructural sites of the samethin section In such cases the definition of a singlepair of reference temperature and pressure isinappropriate and cannot be restricted to the peak-temperature conditions

In order to overcome these difficulties and to bettervisualize the link between mineral compositiondeformation and P-T conditions at the thin-sectionscale de Andrade et al (2006 2011) Vidal et al(2006) and Lanari et al (2013) developed an approachwhich combines elementary maps of composition witheither a thermodynamic or a semi-empirical estimate ofchlorite crystallization temperature Maps of rawmicroprobe analyses were standardized using good-quality point analyses of minerals present in themapped area The maps of standardized analyseswere transformed into maps of structural formulaewhere each pixel analysed is converted into individualmineral structural formulae Depending on the thermo-metric approach it was then possible to compute mapsof temperature or T-XFe3+ conditions assuming thateach chlorite pixel was in equilibrium with quartz andwater The results reported by Vidal et al (2006)Yamato et al (2007) Ganne et al (2012) Lanari et al(2013 2014) and Scheffer et al (2015) showed a closelink between deformation and mineral reaction at themicroscopic scale and maps of chlorite temperaturecould be used as a basis for interpreting the micro-structures observed in thin sections of low- to middle-grade metapelites In all these studies the metastablepersistence of different generations of chlorite crystal-lizing at different temperatures during the exhumationof metamorphic samples could be used to constrain theT-deformation history Moreover the map of (XFe3+)Chl calculated by Vidal et al (2006) was in goodqualitative agreement with the in situ mapping of thisratio using the X-ray absorption near-edge structure(XANES) techniques (Fig 9) (see also Munoz et al2006) The calculated temperature and XFe3+ mapsobtained on a single sample showed the same trend ofreducing (XFe3+)Chl with increasing temperature asobserved in Figs 6 and 7

Bourdelle et al (2013c) applied a similar approachfor low-temperature samples with chlorite of smallgrain size using transmission electron microscopy-energy dispersive X-ray (TEM-EDX) spectroscopicanalyses of focused-ion beam (FIB) foils instead of

629Conditions of formation of chlorite

microprobe analyses of thin sections (Fig 10) Thoseauthors were able to observe heterogeneities in XFe3+

at the nm scale using STXM-XANES measurementsThis last study was particularly interesting as it showedthat approaches similar to those used for studyingmetamorphic rocks could be applied to low-grade anddiagenetic samples

Looking at diagenetic rocks like metamorphicrocks

The use of equilibrium thermodynamics to studylow-grade rocks has been often criticized becausesluggish kinetics might preclude achievement of globalequilibrium at low temperature and is responsible forthe crystallization of metastable phases with a lowerenergy of nucleation than stable ones Essene amp Peacor(1995) claimed that the coexistence of chemicallyinhomogeneous phases at a very fine scale in low-grade rocks is indicative of disequilibrium and thatmineral reaction is better described by a kinetic ratherthan thermodynamic approach However the coexist-ence of chemically inhomogeneous phases is also acommon observation in metamorphic rocks wherezoned grains and the coexistence for example ofchlorite grains of different compositions indicate thatthermodynamic equilibrium is approached only locallyand at a given time in the metamorphic history In factthe main difference between diagenetic and higher-temperature rocks is the size of the domains in localequilibrium The more sluggish kinetics of reactionand diffusivity at low temperature lead to small grainsand domains in equilibrium but these domains can beidentified when using appropriate analytical tools andprocedures developed recently (Bourdelle et al2013bc)

FIG 10 Quantitative Fe redox nanomapping on a FIB foilof Glarus sample (T lt 300degC Switzerland) calculatedfrom the 7087ndash71025 eV images ratio (STXM-XANES)

(from Bourdelle et al 2013c)

FIG 11 Analysis of an illite-chlorite assemblage (Gulf Coast sample Alamo 1 well 5825 m) (a) SEM image of thepetrographic thin section showing the emplacement of the FIB foil to be cut across the illitendashchlorite interface (b)Bright-field TEM image of the FIB foil extracted from the section with the platinum strap on top of it (cd) Enlargementof part b showing the Al and Mg contents of chlorite (apfu) and their variation between crystal core and crystal rim

630 O Vidal et al

Figure 10 shows that it is possible to demonstrateheterogeneities in XFe3+ contents at the nanometerscale Consistent variations of other major elements arealso observed with TEM technics on FIB foils Anexample is shown in Fig 11 which depicts thevariations of Al and Mg contents in a chlorite grainfrom the Gulf Coast (Bourdelle et al 2013b) In thiscase the FIB foil was cut parallel to the interfacebetween illite and chlorite and a nm-scale investiga-tion of composition was conducted by TEM-EDXanalysis The rim and core compositions showed azoning of chlorite with Mg- and Al-richer rims thancore compositions This variation in composition on ascale of lt1 μm was related to various crystallizationstages during burial Chlorite sampled at differentdepths and various locations of the Gulf Coast areashowed that zoned compositions similar to thatobserved in Fig 11 are frequent (Bourdelle et al2013b) The AlIV in chlorite rims display a Tdependence that is consistent with the km-scale andregional evolution of temperature ie with thediagenetic history of Gulf Coast (burial P-T condi-tions detritalauthigenic compositions etchellip)(Fig 12) This study suggests again that approachesused for metamorphic rocks can be applied to low-grade rocks when using adapted analytical tools able toidentify sub-μm variations of composition

CONCLUS ION

The variability of chlorite composition reflects thevariability of its conditions of formation Numerousefforts have been made over the last 30 years to

translate the composition of chlorite in terms of itstemperature of formation The first empirical equationsbased on the AlIVor vacancy contents are restricted tobulk-rock compositions and pressure conditions atwhich they were calibrated Thermodynamicapproaches and semi-empirical equations based onthe ratio of end-member activities do not present thislimit Among the different thermodynamic models ofchlorite published so far Vidal et al (2006) proposed amodel that can be used to estimate T and (XFe3+)Chlsimultaneously The estimated (XFe3+)Chl values areconsistent with those measured by Moumlssbauer orXANES spectroscopy and for the same (XFe3+)Chlthe temperatures estimated using the method of Vidalet al (2006) are similar to those obtained with thesemi-empirical equations of Inoue et al (2009) andLanari et al (2014) Different T-XFe3+ trends areobserved for different pressure gradients and oxygen-buffering assemblages (Lanari et al 2014) butoverall the systematic increase in XFe3+ with areduction in temperature is compatible with thedecrease in aO2 in a closed system The increase inXFe3+ and chlorite vacancy with decreasing tempera-ture is interpreted as reflecting the incorporation of theFe3+-sudoite component which can be used toestimate T-aO2-XFe

3+ values with a Chl-Qtz-H2Omulti-equilibrium approach

Similar temperatures are calculated for low-tem-perature chlorites (T lt 300degC and measured XFe3+ upto 30) with all approaches assuming that iron inchlorite is either completely ferrous or with estimatedvalues of XFe3+ This indicates that the equations ofsemi-empirical thermometers assuming that all iron inchlorite is ferrous include some inherent Fe2+Fe3+ vstemperature correction However very different tem-peratures can be obtained for higher-temperaturesamples (gt300degC) with approaches involving ferriciron or assuming that all of the iron is ferrous Thissuggests that the Fe2+Fe3+ vs temperature correctionmight be restricted to particular aO2 conditions andpressure gradients More work is necessary to betterunderstand this point and to improve the thermo-dynamic models and semi-empirical thermometersThis requires high-quality and well constrained naturaldata To this end mapping techniques that highlight thelink between the variations in temperature and pressureand small-scale chlorite composition including XFe3+

variations offer much promise When combining theresults of the present study with those of other studiespublished recently it appears that thermodynamicapproaches and mapping techniques developed formetamorphic rocks could be used to study low-grade

FIG 12 Range of crystal-rim AlIV content (white boxes)of Gulf Coast chlorites vs measured temperatures Thegrey area underlines the range of maximum AlIV contentwhich is assumed to represent the closest approach toequilibrium composition (after Bourdelle et al 2013b)

631Conditions of formation of chlorite

rocks where kinetics are much more sluggish Thisrequires use of appropriate analytical tools and techni-ques with a spatial resolution of just a few hundrednanometres which are now becoming available

ACKNOWLEDGMENTS

Constructive reviews by S Hillier F Nieto and ananonymous reviewer are gratefully acknowledged

REFERENCES

Beaufort D Patrier P Meunier A amp Ottaviani MM(1992) Chemical variations in assemblages includingepidote andor chlorite in the fossil hydrothermalsystem of Saint Martin (Lesser Antilles) Journal ofVolcanology and Geothermal Research 51 95ndash114

Berman R (1988) Internally-consistent thermodynamicdata for minerals in the system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2Journal of Petrology 29 445ndash522

Berman R amp Brown T (1985) Heat capacity of minerals inthe system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2 representation estima-tion and high temperature extrapolation Contributionsto Mineralogy and Petrology 89 168ndash183

Bourdelle F amp Cathelineau M (2015) Low-temperaturechlorite geothermometry a graphical representationbased on a TndashR2+ndashSi diagram European Journal ofMineralogy 27 617ndash626

Bourdelle F Parra T Chopin C amp Beyssac O (2013a) Anew chlorite geothermometer for diagenetic to low-grade metamorphic conditions Contributions toMineralogy and Petrology 165 723ndash735

Bourdelle F Parra T Beyssac O Chopin C amp Vidal O(2013b) Clay minerals thermometry a comparativestudy based on high-resolution chemical analyses ofillite and chlorite in sandstones from the Gulf Coast(Texas USA) American Mineralogist 98 914ndash926

Bourdelle F Benzerara K Beyssac O Cosmidis JNeuville DR Brown GE amp Paineau E (2013c)Quantification of the ferric ferrous iron ratio insilicates by scanning transmission X-ray microscopyat the Fe L23 edges Contribution to Mineralogy andPetrology 166 423ndash434

Cathelineau M (1988) Cation site occupancy in chloritesand illites as function of temperature Clay Minerals23 471ndash485

Cathelineau M amp Nieva D (1985) A chlorite solidsolution geothermometer the Los Azufres (Mexico)geothermal system Contributions to Mineralogy andPetrology 91 235ndash244

Chermak JA amp Rimstidt JD (1989) Estimating thethermodynamic properties (ΔGdegf and ΔHdegf) of silicate

minerals at 298K from the sum of polyhedral contribu-tions American Mineralogist 74 1023ndash1031

Coggon R amp Holland TJB (2002) Mixing properties ofphengitic micas and revised garnent-phengite thermo-barometers Journal of Metamorphic Geology 20683ndash696

Curtis CD Hughes CR Whiteman JA ampWhittle CK(1985) Compositional variation within some sedi-mentary chlorites and some comments on their originMineralogical Magazine 49 375ndash386

de Andrade V Vidal O Lewin E OrsquoBrien P amp Agard P(2006) Quantification of electron microprobe com-positional maps of rock thin sections an optimizedmethod and examples Journal of MetamorphicGeology 24 655ndash668

de Andrade V Susini J Salomeacute M Beraldin O RigaultC Heymes T Lewin E amp Vidal O (2011)Submicrometer hyperspectral X-ray imaging of het-erogeneous rocks and geomaterials applications at theFe k-edge Analytical Chemistry 83 4220ndash4227

de Caritat P Hutcheon I amp Walshe J (1993) Chloritegeothermometry a review Clays and Clay Minerals41 219ndash239

Essene EJ (2009) Thermobarometry gone astray Pp101ndash129 in Physics and Chemistry of Earthrsquos Interior(AK Gupta amp S Dasgupta editors) PlatinumJubilee Indian National Science Academy Springer-Verlag Berlin

Essene EJ amp Peacor DR (1995) Clay mineralthermometry ndash a critical perspective Clays and ClayMinerals 43 540ndash553

Ganne J de Andrade V Weinberg RF Vidal O DubacqB Kagambega N Naba S Baratoux L Jessell M ampAllibon J (2012) Modern-style plate subductionpreserved in the Palaeoproterozoic West Africancraton Nature Geoscience 5 60ndash65

Grosch EG Vidal O Abu-Alam T amp McLoughlin N(2012) PTconstraints on the metamorphic evolution ofthe Paleoarchean Kromberg type-section BarbertonGreenstone Belt South-Africa Journal of Petrology53 513ndash545

Grosch E McLoughlin N Lanari P Erambert M ampVidal O (2014) Microscale mapping of alterationconditions and potential biosignatures in basaltic-ultramafic rocks on early Earth and beyondAstrobiology 14 216ndash228

Hillier S amp Velde B (1991) Octahedral occupancy andchemical composition of diagenetic (low-temperature)chlorites Clay Minerals 26 149ndash168

Holland TJB (1989) Dependance of entropy on volumefor silicates and oxide minerals A review and apredictive model American Mineralogist 74 5ndash13

Holland T amp Powell R (1998) An internally consistentthermodynamic data set for phases of petrologicalinterest Journal ofMetamorphic Geology 16 309ndash343

Holland T Baker J amp Powell R (1998)Mixing propertiesand activity composition and relationships of chlorites

632 O Vidal et al

parameters of Vidal et al (2006) is possible butthis should be demonstrated by a systematic compari-son of measured and estimated XFe3+ from naturalsamples covering a wide range of P T and fO2

conditions Such a study is still lacking but a first-order comparison of measured and estimated XFe3+

contents for samples of known temperature ofcrystallization is shown in Fig 7 which was obtainedfor the Rouez Niger and Saint-Martin low-T samplescompiled by Inoue et al (2009) the Glarus sample ofBourdelle et al (2013c) and a blueschist-facies sampleretrogressed into the greeschist facies from Sambagawa(de Andrade et al 2006 Munoz et al 2006 Vidalet al 2006) T-XFe3+ values were calculated at 2 kbarexcept for the Sambagawa sample for which calcula-tions were made at 2 5 and 10 kbar for the low-medium- and high-temperature chlorite groups identi-fied by de Andrade et al (2006) and Vidal et al (2006)respectively

FIG 6 Estimated TFe3+ Vidal and XFe3+ pairs for low- andmedium-pressure samples (open circles) high-pressuresamples (black circles) and the samples from Grosch et al

(2012) (grey squares)

FIG 7 Boxes measured XFe3+ and reference temperatures Symbols Tmaximum-XFe3+minimum pairs calculated to

obtain convergence (plusmn30degC) of equilibria 1 8 9 and 10 with the thermodynamic model of Vidal et al (2006)

624 O Vidal et al

For each series of Inoue et al (2009) between 50and 60 of the chlorite analyses contain Si lt3 apfuand led to a convergence of equations 1 8 9 and 10within 30degC The calculated T-XFe3+ pairs are plottedin Fig 7 and compared with the range of referencetemperature (TRef) estimated independently and XFe3+

measures reported by Inoue et al (2009) A fairagreement of the measured and calculated TFe3+ Vidal-XFe3+ values is observed for the Niger and St-Martinanalyses and a good agreement of the TRef and TFe3+ Vidal was observed for the Rouez samples but thecalculated valueswere smaller than themeasured valuesfor XFe3+ for this samples However the XFe3+ valuesreported in Fig 7 are minimum values required toobtain convergence of equilibria 1 and 8ndash10 within30degC Convergence is maintained at higher valuesof XFe3+ and TFe3+ Vidal for the Rouez samples arestill located in the range of TRef when XFe3+ is setat gt15 (lower limit of the reference values) TheTFe3+ Vidal values calculated for the Glarus chloriteanalyses listed in Bourdelle et al (2013c) are 139degC 203degC and 224degC in fair agreement with thoseestimated independently by Bourdelle et al (2013c)(135degC 170degC and 182degC respectively) or with thesemi-empirical thermometer of Inoue et al (2009)(145degC 196degC and 210degC respectively) For thethree samples studied the XFe3+ values measuredusing XANES (X-ray absorption near edge struc-ture) spectra in the transmission mode of scanningtransmission X-ray microscopy (STXM) were foundto be 223 238 and 279 These values comparefavourably with the estimated ones (32 27 and 28respectively) although the measured values suggestan increase in XFe3+ with temperature which iscontrary the trend inferred from the thermodynamicestimates and evidenced in Figs 6 and 7 Finally therange of XFe3+ estimated for the different generationsof chlorite in the Sambagawa sample (ellipses inFig 7) are slightly higher but still in fair agreementwith those inferred from the XANES measurementsIn conclusion Fig 7 shows that the values estimatedon the basis of the convergence of equilibria (1 8 9and 10) are consistent with the TRef and measuredXFe3+ for various samples covering a range oftemperature from 100degC to 500degC pressure from sim2to 10 kbar and XFe3+ values from 5 to 35 Boththe measured and estimated XFe3+ values plotted inFig 7 seem to confirm the systematic decrease ofXFe3+ with increasing temperature and increase withincreasing pressure that was observed in Fig 6 Thistrend is similar to the decrease in vacancy observedin Fig 5

aO2 -TEMPERATURE EST IMATESAND ADDIT ION OF A FERR I -

SUDOITE END -MEMBER

The general trends inferred from our compilation ofnatural samples indicate that XFe3+ increases withdecreasing temperature and increasing pressureAlthough the model of Vidal et al (2006) involvesno Fe3+-bearing end-member the oxygen activity(aO2) prevailing during the formation of chloritesused to compute Fig 7 can be estimated assuming thatit was buffered by the presence of pure magnetite (Mt)in equilibrium with chlorite and quartz Following thisassumption aO2 for each sample has been calculatedusing equilibria (1 8 9 10) and the followingadditional equilibria which intersect at the same P T(plusmn30degC) (XFe3+)Chl and Log(aO2) (plusmn2) (Fig 8)

15 Mg-Sudthorn 40 Mtthorn 45 Qtzthorn 60 H2O

frac14 6 Clinthorn 24 Daphthorn 20 O2

(11)

6 Mg-Sudthorn 10 Mtthorn 6 Qtzthorn 12 H2O

frac14 3 Mg-Amthorn 6 Daphthorn 5 O2

(12)

15 Mg-Sudthorn 6 Clinthorn 10 Mt

frac14 15 Qtzthorn 15 Mg-Amthorn 6 Daphthorn 5 O2

(13)

6 Daphthorn 4 Clinthorn 5 O2

frac14 20 Qtzthorn 10 Mtthorn 5 Mg-Amthorn 20 H2O(14)

60 Mg-Sudthorn 12 Clinthorn 70 Mtthorn 60 H2O

frac14 45 Mg-Amthorn 42 Daphthorn 35 O2

(15)

4 Qtzthorn 2 Mtthorn Fe-Amthorn 4 H2O

frac14 2 Daphthorn O2

(16)

15 Mg-Sudthorn 6 Daphthorn 10 Mt

frac14 15 Qtzthorn 6 Clinthorn 15 Fe-Amthorn 5 O2

(17)

60 Mg-Sudthorn 70 Mtthorn 60 H2O

frac14 24 Clinthorn 45 Fe-Amthorn 6 Daphthorn 35 O2

(18)

15 Mg-Sudthorn 16 Mtthorn 12 H2O

frac14 3 Qtzthorn 6 Clinthorn 12 Fe-Amthorn 8 O2

(19)

6 Mg-Sudthorn 4 Mt

frac14 6 Qtzthorn 3 Mg-Amthorn 3 Fe-Amthorn 2 O2

(20)

12 Mg-Sudthorn 14 Mtthorn 12 H2O

frac14 6 Mg-Amthorn 3 Fe-Amthorn 6 Daphthorn 7 O2

(21)

5 Mg-Amthorn 4 Mtthorn 8 Qtzthorn 8 H2O

frac14 4 Clinthorn 3 Fe-Amthorn 2 O2

(22)

24 Mg-Sudthorn 3 Mg-Amthorn 28 Mtthorn 24 H2O

frac14 12 Clinthorn 21 Fe-Amthorn 14 O2

(23)

All calculated Log(aO2) except those of the Saint-

625Conditions of formation of chlorite

Martin samples follow the same trend and decreasewith decreasing temperature from simminus27 at 450ndash500degC10 kbar (high-temperature chlorites of Sambagawa) tominus50degC at 125degC and 1 kbar (Niger samples) (Fig 8)The Log(aO2) values of the high-T and -P samples plotin the stability field of magnetite whereas those of thelow-T samples plot in the stability field of hematitewhich is not consistent with the assumption of ChlndashMtndashQtzndashH2O equilibrium The presence of hematite hasbeen reported in the Saint-Martin samples only(Beaufort et al 1992) which plot above the trenddefined by the other samples (upper grey ellipse inFig 8) The relative position of this hematite-bearingsample is consistent with expected higher oxygenactivities required for its formation For the other low-T and apparently hematite-free samples it is possiblethat the calculated Log(aO2) values are too high or thatoxygen was not buffered by an equilibrium ofmagnetitewith chlorite In any case Fig 8 indicates that thesystematic increase in (XFe3+)Chl with decreasingtemperature observed in Figs 6 and 7 is consistentwith a lower aO2 and amount of free oxygen at low-Tthan high-T in systems where aO2 is buffered by thepresence of chlorite and iron oxides To maintain this

equilibrium with lowering temperature in metamorphicrocks where oxygen is immobile oxygen has to beconsumed and one way for that is to produce chloritethat is richer in Fe3+ This is a possible interpretationfor the formation of low-temperature Fe3+-rich chloritecoexisting with high-temperature Fe3+-poor chloritein the same thin section of the Sambagwa sample(Fig 9)

In order to discuss the Log(aO2)-T-XFe3+ evolution

in more detail for rocks where aO2 is not buffered bythe presence of magnetite it is necessary to define anFe3+-bearing chlorite end-member The simultaneousincrease of vacancy and Fe3+ content with decreasingtemperature observed in Figs 5ndash7 is explained by the3(Mg Fe2+) = 2(Fe3+) ditrioctahedral substitution(see also Trincal amp Lanari 2016 this issue) Thisexchange can be modelled with a ferri-sudoite end-member that materializes the relationship betweenXFe3+ and vacancy The standard-state thermo-dynamic properties of ferri-sudoite were constrainedas follows

(1) Because the maximum amount of (Fe3+)Chlcalculated in the previous section never exceeded1 apfu the cation distribution of Vidal et al (2006)

FIG 8 Left Ranges of Log(aO2) conditions calculated for the samples used to compute Fig 7 at the temperature andXFe3+ contents estimated from the Chl-Qtz-H2O equilibrium conditions Dashed lines hematite-magnetite-O2

equilibrium calculated using the thermodynamic properties of Berman (1988) and unit activities of Ht and Mt RightSample calculation at 2 kbar for a Rouez sample using reactions 8ndash20 (upper) or 26ndash40 (lower) The vertical linescorrespond to oxygen- and magnetite-free reactions 1 5 6 and 7 used to compute Figs 5ndash7 which converge at +30degCThe circles in the lower figure represent values calculated with (ΔHdegf ) ferri-Sud constrained with the chlorite-magnetite-quartz-water equilibria 8ndash20 (open circle) or estimated as ΔHdegf (Mg-sudoite) ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2

(filled circle) (see Table 1 and text)

626 O Vidal et al

was kept unchanged and the Fe3+-Al3+ exchange wasassumed to be restricted to the M4 site of chlorite Thechosen formula of ferri-Sud was thus ([Si2]T1[SiAl]T2)

IV([]M1[Mg2Al2]M2-3[Fe3+]M4)

VIO10(OH)8The standard-state thermodynamic properties of allother chlorite end-members the activity model and theMargules parameters were kept unchanged

(2) The third-law entropy of ferri-sudoite (Sdeg =415 JmolK) was estimated according to Holland(1989) with a configurational entropy of 2305 JmolKand a ferromagnetic contribution of 1304 JmolK itsvolume (Vdeg = 204 Jb) was estimated to be 01 Jblower than that of Mg-sudoite (2051 Jb) and the Cp(T ) function was estimated according to Berman amp

FIG 9 Rock-sample mapping of a Sambagawa sample (a) maps of estimated XFe3+ (Vidal et al 2006 Munoz et al2006) and Log(aO2) (estimated assuming that it is buffered by magnetite-chlorite equilibria see text and Fig 8)(b) measured Fe XANES edge-position proportional to Fe redox (Munoz et al 2006) (c) estimated temperature (Vidal

et al 2006 Munoz et al 2006)

627Conditions of formation of chlorite

Brown (1985) The formation enthalpy (ΔHdegf = 82254 plusmn 5 kJmol) was first estimated as the average valuebetween that calculated according to Chermak ampRimstidt (1989) (minus82204 kJmol) and that(minus82304 kJmol) calculted as ΔHdegf (Mg-sudoite) ndashΔHdegf (corundum)2 + ΔHdegf (hematite)2 using thevalues of Vidal et al (1992) for sudoite and Berman(1988) for hematite

(3) In a first step ΔHdegf and Sdeg of ferri-sudoite wereadjusted in order to reproduce the temperaturesreported in Fig 7 with oxygen-free reactions andassuming that magnetite is present in all samples(Chl-Qtz-Mt assemblage) ΔHdegf was allowed to varybetweenminus8220 andminus8230 kJmol and Sdeg within 10 JmolK of the estimated value The addition ofmagnetite and ferri-sudoite to quartz and water leadsto 17 additional equilibria and the following wereused to constrain ΔHdegf and Sdeg of ferri-sudoite

ClinthornMg-Sud frac14 Mg-Amthorn Qtzthorn H2O (1)

4 Clinthorn Daphthorn 10 ferri-Sud

frac14 10 Mg-Amthorn 25 Qtzthorn 30 Mtthorn 20 H2O(24)

7 Daphthorn 70 ferri-Sudthorn 30 Mg-Am

frac14 22 Clinthorn 75 Mg-Sudthorn 35 Mtthorn 40 H2O(25)

28 ferri-Sudthorn 11 Fe-AmthornMg-Am

frac14 6 Daphthorn 30 Mg-Sudthorn 14 Mtthorn 16 H2O(26)

Daphthorn 10 ferri-Sudthorn 10 Mg-Amthorn 10 Qtz

frac14 6 Clinthorn 15 Mg-Sudthorn 5 Mtthorn 16 H2O(27)

The ferri-Sud properties derived with the assumptionof ChlndashQtzndashMt equilibrium are listed in Table 1(referenced a)With these values the oxygen activity atthe P T and (XFe3+)Chl of all equilibria used tocompute Fig 7 can be calculated with the followingoxygen-bearing and magnetite-free equilibria whichconverge with equilibria 1 8 9 and 10 at the same PTand aO2 conditions

16 Qtzthorn 7 Mg-Amthorn 4 Daphthorn 6 Mg-Sud

thorn 5 O2 thorn 12 H2O frac14 20 ferri-Sud(28)

Fe-Amthorn 2 Mg-Amthorn 6 Qtzthorn 4 H2Othorn O2

frac14 4 ferri-Sud(29)

Mg-Amthorn 4 Daphthorn 6 Mg-Sudthorn 2 O2

frac14 3 Fe-Amthorn 8 ferri-Sudthorn 2 Qtz(30)

5 Mg-Amthorn 12 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 8 Fe-Amthorn 28 ferri-Sud

(31)

8 Daphthorn 12 Mg-Sudthorn 3 O2 frac14 7 Fe-Am

thorn 12 ferri-Sudthorn 10 Qtzthorn 4 H2O(32)

6 Clinthorn 4 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Mg-Amthorn 20 ferri-Sudthorn 5 Qtz(33)

4 Clinthorn 20 ferri-Sud frac14 30 Qtz

thorn 15 Mg-Amthorn 4 Daphthorn 5 O2 thorn 20 H2O(34)

90 Mg-Sudthorn 28 Daphthorn 32 Clinthorn 20 H2Othorn 35 O2

frac14 140 ferri-Sudthorn 15 Mg-Am

(35)

15 Qtzthorn 14 Clinthorn 16 Daphthorn 45 Mg-Sud

thorn 20 O2 thorn 20 H2O frac14 80 ferri-Sud(36)

2 Clinthorn FeAmthorn 3 Mg-Sudthorn O2

frac14 2 Mg-Amthorn 4 ferri-Sudthorn Qtz(37)

12 Clinthorn 7 Fe-Amthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 10 Mg-Amthorn 28 ferri-Sud

(38)

5 Qtzthorn 2 Clinthorn 2 Fe-Amthorn 3 Mg-Sud

thorn 2 O2 thorn 4 H2O frac14 8 ferri-Sud(39)

2 Clinthorn 8 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Fe-Amthorn 20 ferri-Sudthorn 5 Qtz(40)

30 Qtzthorn 8 Clinthorn 15 Fe-Amthorn 5 O2 thorn 20 H2O

frac14 8 Daphthorn 20 ferri-Sud

(41)

4 Clinthorn 8 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 3 Fe-Amthorn 28 ferri-Sud

(42)

Because the ferri-sudoite standard-state propertieswere constrained by the chlorite-quartz-magnetite-O2-H2O equilibrium the Log(aO2) calculated withthe equilibria above follow naturally the trend shown inFig 8 and the Log(aO2) values calculated withreactions 28ndash42 of low-temperature samples plot inthe hematite stability field This suggests that theΔGdegPT values of ferri-sudoite derived above with theassumption that aO2 is buffered by the presence ofmagnetite are possibly too high A reduction of ΔHdegf(ferri-sudoite) to 82304 kJmol (=ΔHdegf (Mg-sudoite)ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2 see Table 1)leads to Log(aO2) reduced by 45 Log units (see Fig 8)

628 O Vidal et al

and most calculated values except those of thehematite-bearing St Martin samples plot in themagnetite stability field

A more detailed study is required to refine the ferri-sudoite thermodynamic data However the resultsobtained using the preliminary values listed in Table 1show that this end-member can be used to reproduceconsistently the T-XFe3+ and T- trends obtained fromclassical Chl-Qtz-H2O calculations (Figs 5ndash7) Theincrease in XFe3+ with reducing temperature is consist-ent with the expected reduction in aO2 and the amountof free oxygen that is consumed to produce Fe3+-bearingchlorite Ferric iron is therefore a major component thathas to be taken into account when dealing with low-temperature samples Empirical or semi-empiricalchlorite thermometers which assume that all Fe is inthe ferrous state may include one of the XFe3+-T andaO2-T trends illustrated in Figs 7 and 8 but they cannotaccount for the different XFe3+-T trends that areexpected in the contexts of different aO2 conditions

FUTURE CHALLENGES LOOK INGAT LOW-GRADE ROCKS SUCH ASMETAMORPH IC ROCKS AND THE

DYNAMICS OF CHLOR ITECRYSTALL IZAT ION AT THE

MICROMETR IC TO NANOMETR ICSCALE

T-P-XFe3+ mapping at the thin-section scaleevidence of chlorite composition re-equilibration in metamorphic and low-gradesamples

Although natural samples offer valuable constraintson thermodynamic models there are two maindifficulties in using them to evaluate the reliability ofexisting chlorite models for thermometric purposes

(1) In the case of low-grade samples referencepressures and temperatures are generally known withina reasonable absolute uncertainty but the size ofchlorite grains is often small and good microprobeanalyses are sometimes difficult to obtain Moreoverthe kinetics of reaction are so sluggish that inheritedchlorite grains might remain unaffected by latervariations in temperature which makes thetemperature-composition relationships difficult toconstrain Lastly extrapolation at HT and HP ofempirical or semi-empirical models constrained byLT samples only may be erroneous in particular theeffect of pressure or eventual systematic variations ofXFe3+ cannot be appraised

(2) In the case of metamorphic samplesmicroprobe analyses are easier to obtain from largechlorite grains but chlorites of contrasted compositionthat crystallized at different temperatures and times ofthe metamorphic history are often observed to coexistmetastably in different microstructural sites of the samethin section In such cases the definition of a singlepair of reference temperature and pressure isinappropriate and cannot be restricted to the peak-temperature conditions

In order to overcome these difficulties and to bettervisualize the link between mineral compositiondeformation and P-T conditions at the thin-sectionscale de Andrade et al (2006 2011) Vidal et al(2006) and Lanari et al (2013) developed an approachwhich combines elementary maps of composition witheither a thermodynamic or a semi-empirical estimate ofchlorite crystallization temperature Maps of rawmicroprobe analyses were standardized using good-quality point analyses of minerals present in themapped area The maps of standardized analyseswere transformed into maps of structural formulaewhere each pixel analysed is converted into individualmineral structural formulae Depending on the thermo-metric approach it was then possible to compute mapsof temperature or T-XFe3+ conditions assuming thateach chlorite pixel was in equilibrium with quartz andwater The results reported by Vidal et al (2006)Yamato et al (2007) Ganne et al (2012) Lanari et al(2013 2014) and Scheffer et al (2015) showed a closelink between deformation and mineral reaction at themicroscopic scale and maps of chlorite temperaturecould be used as a basis for interpreting the micro-structures observed in thin sections of low- to middle-grade metapelites In all these studies the metastablepersistence of different generations of chlorite crystal-lizing at different temperatures during the exhumationof metamorphic samples could be used to constrain theT-deformation history Moreover the map of (XFe3+)Chl calculated by Vidal et al (2006) was in goodqualitative agreement with the in situ mapping of thisratio using the X-ray absorption near-edge structure(XANES) techniques (Fig 9) (see also Munoz et al2006) The calculated temperature and XFe3+ mapsobtained on a single sample showed the same trend ofreducing (XFe3+)Chl with increasing temperature asobserved in Figs 6 and 7

Bourdelle et al (2013c) applied a similar approachfor low-temperature samples with chlorite of smallgrain size using transmission electron microscopy-energy dispersive X-ray (TEM-EDX) spectroscopicanalyses of focused-ion beam (FIB) foils instead of

629Conditions of formation of chlorite

microprobe analyses of thin sections (Fig 10) Thoseauthors were able to observe heterogeneities in XFe3+

at the nm scale using STXM-XANES measurementsThis last study was particularly interesting as it showedthat approaches similar to those used for studyingmetamorphic rocks could be applied to low-grade anddiagenetic samples

Looking at diagenetic rocks like metamorphicrocks

The use of equilibrium thermodynamics to studylow-grade rocks has been often criticized becausesluggish kinetics might preclude achievement of globalequilibrium at low temperature and is responsible forthe crystallization of metastable phases with a lowerenergy of nucleation than stable ones Essene amp Peacor(1995) claimed that the coexistence of chemicallyinhomogeneous phases at a very fine scale in low-grade rocks is indicative of disequilibrium and thatmineral reaction is better described by a kinetic ratherthan thermodynamic approach However the coexist-ence of chemically inhomogeneous phases is also acommon observation in metamorphic rocks wherezoned grains and the coexistence for example ofchlorite grains of different compositions indicate thatthermodynamic equilibrium is approached only locallyand at a given time in the metamorphic history In factthe main difference between diagenetic and higher-temperature rocks is the size of the domains in localequilibrium The more sluggish kinetics of reactionand diffusivity at low temperature lead to small grainsand domains in equilibrium but these domains can beidentified when using appropriate analytical tools andprocedures developed recently (Bourdelle et al2013bc)

FIG 10 Quantitative Fe redox nanomapping on a FIB foilof Glarus sample (T lt 300degC Switzerland) calculatedfrom the 7087ndash71025 eV images ratio (STXM-XANES)

(from Bourdelle et al 2013c)

FIG 11 Analysis of an illite-chlorite assemblage (Gulf Coast sample Alamo 1 well 5825 m) (a) SEM image of thepetrographic thin section showing the emplacement of the FIB foil to be cut across the illitendashchlorite interface (b)Bright-field TEM image of the FIB foil extracted from the section with the platinum strap on top of it (cd) Enlargementof part b showing the Al and Mg contents of chlorite (apfu) and their variation between crystal core and crystal rim

630 O Vidal et al

Figure 10 shows that it is possible to demonstrateheterogeneities in XFe3+ contents at the nanometerscale Consistent variations of other major elements arealso observed with TEM technics on FIB foils Anexample is shown in Fig 11 which depicts thevariations of Al and Mg contents in a chlorite grainfrom the Gulf Coast (Bourdelle et al 2013b) In thiscase the FIB foil was cut parallel to the interfacebetween illite and chlorite and a nm-scale investiga-tion of composition was conducted by TEM-EDXanalysis The rim and core compositions showed azoning of chlorite with Mg- and Al-richer rims thancore compositions This variation in composition on ascale of lt1 μm was related to various crystallizationstages during burial Chlorite sampled at differentdepths and various locations of the Gulf Coast areashowed that zoned compositions similar to thatobserved in Fig 11 are frequent (Bourdelle et al2013b) The AlIV in chlorite rims display a Tdependence that is consistent with the km-scale andregional evolution of temperature ie with thediagenetic history of Gulf Coast (burial P-T condi-tions detritalauthigenic compositions etchellip)(Fig 12) This study suggests again that approachesused for metamorphic rocks can be applied to low-grade rocks when using adapted analytical tools able toidentify sub-μm variations of composition

CONCLUS ION

The variability of chlorite composition reflects thevariability of its conditions of formation Numerousefforts have been made over the last 30 years to

translate the composition of chlorite in terms of itstemperature of formation The first empirical equationsbased on the AlIVor vacancy contents are restricted tobulk-rock compositions and pressure conditions atwhich they were calibrated Thermodynamicapproaches and semi-empirical equations based onthe ratio of end-member activities do not present thislimit Among the different thermodynamic models ofchlorite published so far Vidal et al (2006) proposed amodel that can be used to estimate T and (XFe3+)Chlsimultaneously The estimated (XFe3+)Chl values areconsistent with those measured by Moumlssbauer orXANES spectroscopy and for the same (XFe3+)Chlthe temperatures estimated using the method of Vidalet al (2006) are similar to those obtained with thesemi-empirical equations of Inoue et al (2009) andLanari et al (2014) Different T-XFe3+ trends areobserved for different pressure gradients and oxygen-buffering assemblages (Lanari et al 2014) butoverall the systematic increase in XFe3+ with areduction in temperature is compatible with thedecrease in aO2 in a closed system The increase inXFe3+ and chlorite vacancy with decreasing tempera-ture is interpreted as reflecting the incorporation of theFe3+-sudoite component which can be used toestimate T-aO2-XFe

3+ values with a Chl-Qtz-H2Omulti-equilibrium approach

Similar temperatures are calculated for low-tem-perature chlorites (T lt 300degC and measured XFe3+ upto 30) with all approaches assuming that iron inchlorite is either completely ferrous or with estimatedvalues of XFe3+ This indicates that the equations ofsemi-empirical thermometers assuming that all iron inchlorite is ferrous include some inherent Fe2+Fe3+ vstemperature correction However very different tem-peratures can be obtained for higher-temperaturesamples (gt300degC) with approaches involving ferriciron or assuming that all of the iron is ferrous Thissuggests that the Fe2+Fe3+ vs temperature correctionmight be restricted to particular aO2 conditions andpressure gradients More work is necessary to betterunderstand this point and to improve the thermo-dynamic models and semi-empirical thermometersThis requires high-quality and well constrained naturaldata To this end mapping techniques that highlight thelink between the variations in temperature and pressureand small-scale chlorite composition including XFe3+

variations offer much promise When combining theresults of the present study with those of other studiespublished recently it appears that thermodynamicapproaches and mapping techniques developed formetamorphic rocks could be used to study low-grade

FIG 12 Range of crystal-rim AlIV content (white boxes)of Gulf Coast chlorites vs measured temperatures Thegrey area underlines the range of maximum AlIV contentwhich is assumed to represent the closest approach toequilibrium composition (after Bourdelle et al 2013b)

631Conditions of formation of chlorite

rocks where kinetics are much more sluggish Thisrequires use of appropriate analytical tools and techni-ques with a spatial resolution of just a few hundrednanometres which are now becoming available

ACKNOWLEDGMENTS

Constructive reviews by S Hillier F Nieto and ananonymous reviewer are gratefully acknowledged

REFERENCES

Beaufort D Patrier P Meunier A amp Ottaviani MM(1992) Chemical variations in assemblages includingepidote andor chlorite in the fossil hydrothermalsystem of Saint Martin (Lesser Antilles) Journal ofVolcanology and Geothermal Research 51 95ndash114

Berman R (1988) Internally-consistent thermodynamicdata for minerals in the system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2Journal of Petrology 29 445ndash522

Berman R amp Brown T (1985) Heat capacity of minerals inthe system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2 representation estima-tion and high temperature extrapolation Contributionsto Mineralogy and Petrology 89 168ndash183

Bourdelle F amp Cathelineau M (2015) Low-temperaturechlorite geothermometry a graphical representationbased on a TndashR2+ndashSi diagram European Journal ofMineralogy 27 617ndash626

Bourdelle F Parra T Chopin C amp Beyssac O (2013a) Anew chlorite geothermometer for diagenetic to low-grade metamorphic conditions Contributions toMineralogy and Petrology 165 723ndash735

Bourdelle F Parra T Beyssac O Chopin C amp Vidal O(2013b) Clay minerals thermometry a comparativestudy based on high-resolution chemical analyses ofillite and chlorite in sandstones from the Gulf Coast(Texas USA) American Mineralogist 98 914ndash926

Bourdelle F Benzerara K Beyssac O Cosmidis JNeuville DR Brown GE amp Paineau E (2013c)Quantification of the ferric ferrous iron ratio insilicates by scanning transmission X-ray microscopyat the Fe L23 edges Contribution to Mineralogy andPetrology 166 423ndash434

Cathelineau M (1988) Cation site occupancy in chloritesand illites as function of temperature Clay Minerals23 471ndash485

Cathelineau M amp Nieva D (1985) A chlorite solidsolution geothermometer the Los Azufres (Mexico)geothermal system Contributions to Mineralogy andPetrology 91 235ndash244

Chermak JA amp Rimstidt JD (1989) Estimating thethermodynamic properties (ΔGdegf and ΔHdegf) of silicate

minerals at 298K from the sum of polyhedral contribu-tions American Mineralogist 74 1023ndash1031

Coggon R amp Holland TJB (2002) Mixing properties ofphengitic micas and revised garnent-phengite thermo-barometers Journal of Metamorphic Geology 20683ndash696

Curtis CD Hughes CR Whiteman JA ampWhittle CK(1985) Compositional variation within some sedi-mentary chlorites and some comments on their originMineralogical Magazine 49 375ndash386

de Andrade V Vidal O Lewin E OrsquoBrien P amp Agard P(2006) Quantification of electron microprobe com-positional maps of rock thin sections an optimizedmethod and examples Journal of MetamorphicGeology 24 655ndash668

de Andrade V Susini J Salomeacute M Beraldin O RigaultC Heymes T Lewin E amp Vidal O (2011)Submicrometer hyperspectral X-ray imaging of het-erogeneous rocks and geomaterials applications at theFe k-edge Analytical Chemistry 83 4220ndash4227

de Caritat P Hutcheon I amp Walshe J (1993) Chloritegeothermometry a review Clays and Clay Minerals41 219ndash239

Essene EJ (2009) Thermobarometry gone astray Pp101ndash129 in Physics and Chemistry of Earthrsquos Interior(AK Gupta amp S Dasgupta editors) PlatinumJubilee Indian National Science Academy Springer-Verlag Berlin

Essene EJ amp Peacor DR (1995) Clay mineralthermometry ndash a critical perspective Clays and ClayMinerals 43 540ndash553

Ganne J de Andrade V Weinberg RF Vidal O DubacqB Kagambega N Naba S Baratoux L Jessell M ampAllibon J (2012) Modern-style plate subductionpreserved in the Palaeoproterozoic West Africancraton Nature Geoscience 5 60ndash65

Grosch EG Vidal O Abu-Alam T amp McLoughlin N(2012) PTconstraints on the metamorphic evolution ofthe Paleoarchean Kromberg type-section BarbertonGreenstone Belt South-Africa Journal of Petrology53 513ndash545

Grosch E McLoughlin N Lanari P Erambert M ampVidal O (2014) Microscale mapping of alterationconditions and potential biosignatures in basaltic-ultramafic rocks on early Earth and beyondAstrobiology 14 216ndash228

Hillier S amp Velde B (1991) Octahedral occupancy andchemical composition of diagenetic (low-temperature)chlorites Clay Minerals 26 149ndash168

Holland TJB (1989) Dependance of entropy on volumefor silicates and oxide minerals A review and apredictive model American Mineralogist 74 5ndash13

Holland T amp Powell R (1998) An internally consistentthermodynamic data set for phases of petrologicalinterest Journal ofMetamorphic Geology 16 309ndash343

Holland T Baker J amp Powell R (1998)Mixing propertiesand activity composition and relationships of chlorites

632 O Vidal et al

For each series of Inoue et al (2009) between 50and 60 of the chlorite analyses contain Si lt3 apfuand led to a convergence of equations 1 8 9 and 10within 30degC The calculated T-XFe3+ pairs are plottedin Fig 7 and compared with the range of referencetemperature (TRef) estimated independently and XFe3+

measures reported by Inoue et al (2009) A fairagreement of the measured and calculated TFe3+ Vidal-XFe3+ values is observed for the Niger and St-Martinanalyses and a good agreement of the TRef and TFe3+ Vidal was observed for the Rouez samples but thecalculated valueswere smaller than themeasured valuesfor XFe3+ for this samples However the XFe3+ valuesreported in Fig 7 are minimum values required toobtain convergence of equilibria 1 and 8ndash10 within30degC Convergence is maintained at higher valuesof XFe3+ and TFe3+ Vidal for the Rouez samples arestill located in the range of TRef when XFe3+ is setat gt15 (lower limit of the reference values) TheTFe3+ Vidal values calculated for the Glarus chloriteanalyses listed in Bourdelle et al (2013c) are 139degC 203degC and 224degC in fair agreement with thoseestimated independently by Bourdelle et al (2013c)(135degC 170degC and 182degC respectively) or with thesemi-empirical thermometer of Inoue et al (2009)(145degC 196degC and 210degC respectively) For thethree samples studied the XFe3+ values measuredusing XANES (X-ray absorption near edge struc-ture) spectra in the transmission mode of scanningtransmission X-ray microscopy (STXM) were foundto be 223 238 and 279 These values comparefavourably with the estimated ones (32 27 and 28respectively) although the measured values suggestan increase in XFe3+ with temperature which iscontrary the trend inferred from the thermodynamicestimates and evidenced in Figs 6 and 7 Finally therange of XFe3+ estimated for the different generationsof chlorite in the Sambagawa sample (ellipses inFig 7) are slightly higher but still in fair agreementwith those inferred from the XANES measurementsIn conclusion Fig 7 shows that the values estimatedon the basis of the convergence of equilibria (1 8 9and 10) are consistent with the TRef and measuredXFe3+ for various samples covering a range oftemperature from 100degC to 500degC pressure from sim2to 10 kbar and XFe3+ values from 5 to 35 Boththe measured and estimated XFe3+ values plotted inFig 7 seem to confirm the systematic decrease ofXFe3+ with increasing temperature and increase withincreasing pressure that was observed in Fig 6 Thistrend is similar to the decrease in vacancy observedin Fig 5

aO2 -TEMPERATURE EST IMATESAND ADDIT ION OF A FERR I -

SUDOITE END -MEMBER

The general trends inferred from our compilation ofnatural samples indicate that XFe3+ increases withdecreasing temperature and increasing pressureAlthough the model of Vidal et al (2006) involvesno Fe3+-bearing end-member the oxygen activity(aO2) prevailing during the formation of chloritesused to compute Fig 7 can be estimated assuming thatit was buffered by the presence of pure magnetite (Mt)in equilibrium with chlorite and quartz Following thisassumption aO2 for each sample has been calculatedusing equilibria (1 8 9 10) and the followingadditional equilibria which intersect at the same P T(plusmn30degC) (XFe3+)Chl and Log(aO2) (plusmn2) (Fig 8)

15 Mg-Sudthorn 40 Mtthorn 45 Qtzthorn 60 H2O

frac14 6 Clinthorn 24 Daphthorn 20 O2

(11)

6 Mg-Sudthorn 10 Mtthorn 6 Qtzthorn 12 H2O

frac14 3 Mg-Amthorn 6 Daphthorn 5 O2

(12)

15 Mg-Sudthorn 6 Clinthorn 10 Mt

frac14 15 Qtzthorn 15 Mg-Amthorn 6 Daphthorn 5 O2

(13)

6 Daphthorn 4 Clinthorn 5 O2

frac14 20 Qtzthorn 10 Mtthorn 5 Mg-Amthorn 20 H2O(14)

60 Mg-Sudthorn 12 Clinthorn 70 Mtthorn 60 H2O

frac14 45 Mg-Amthorn 42 Daphthorn 35 O2

(15)

4 Qtzthorn 2 Mtthorn Fe-Amthorn 4 H2O

frac14 2 Daphthorn O2

(16)

15 Mg-Sudthorn 6 Daphthorn 10 Mt

frac14 15 Qtzthorn 6 Clinthorn 15 Fe-Amthorn 5 O2

(17)

60 Mg-Sudthorn 70 Mtthorn 60 H2O

frac14 24 Clinthorn 45 Fe-Amthorn 6 Daphthorn 35 O2

(18)

15 Mg-Sudthorn 16 Mtthorn 12 H2O

frac14 3 Qtzthorn 6 Clinthorn 12 Fe-Amthorn 8 O2

(19)

6 Mg-Sudthorn 4 Mt

frac14 6 Qtzthorn 3 Mg-Amthorn 3 Fe-Amthorn 2 O2

(20)

12 Mg-Sudthorn 14 Mtthorn 12 H2O

frac14 6 Mg-Amthorn 3 Fe-Amthorn 6 Daphthorn 7 O2

(21)

5 Mg-Amthorn 4 Mtthorn 8 Qtzthorn 8 H2O

frac14 4 Clinthorn 3 Fe-Amthorn 2 O2

(22)

24 Mg-Sudthorn 3 Mg-Amthorn 28 Mtthorn 24 H2O

frac14 12 Clinthorn 21 Fe-Amthorn 14 O2

(23)

All calculated Log(aO2) except those of the Saint-

625Conditions of formation of chlorite

Martin samples follow the same trend and decreasewith decreasing temperature from simminus27 at 450ndash500degC10 kbar (high-temperature chlorites of Sambagawa) tominus50degC at 125degC and 1 kbar (Niger samples) (Fig 8)The Log(aO2) values of the high-T and -P samples plotin the stability field of magnetite whereas those of thelow-T samples plot in the stability field of hematitewhich is not consistent with the assumption of ChlndashMtndashQtzndashH2O equilibrium The presence of hematite hasbeen reported in the Saint-Martin samples only(Beaufort et al 1992) which plot above the trenddefined by the other samples (upper grey ellipse inFig 8) The relative position of this hematite-bearingsample is consistent with expected higher oxygenactivities required for its formation For the other low-T and apparently hematite-free samples it is possiblethat the calculated Log(aO2) values are too high or thatoxygen was not buffered by an equilibrium ofmagnetitewith chlorite In any case Fig 8 indicates that thesystematic increase in (XFe3+)Chl with decreasingtemperature observed in Figs 6 and 7 is consistentwith a lower aO2 and amount of free oxygen at low-Tthan high-T in systems where aO2 is buffered by thepresence of chlorite and iron oxides To maintain this

equilibrium with lowering temperature in metamorphicrocks where oxygen is immobile oxygen has to beconsumed and one way for that is to produce chloritethat is richer in Fe3+ This is a possible interpretationfor the formation of low-temperature Fe3+-rich chloritecoexisting with high-temperature Fe3+-poor chloritein the same thin section of the Sambagwa sample(Fig 9)

In order to discuss the Log(aO2)-T-XFe3+ evolution

in more detail for rocks where aO2 is not buffered bythe presence of magnetite it is necessary to define anFe3+-bearing chlorite end-member The simultaneousincrease of vacancy and Fe3+ content with decreasingtemperature observed in Figs 5ndash7 is explained by the3(Mg Fe2+) = 2(Fe3+) ditrioctahedral substitution(see also Trincal amp Lanari 2016 this issue) Thisexchange can be modelled with a ferri-sudoite end-member that materializes the relationship betweenXFe3+ and vacancy The standard-state thermo-dynamic properties of ferri-sudoite were constrainedas follows

(1) Because the maximum amount of (Fe3+)Chlcalculated in the previous section never exceeded1 apfu the cation distribution of Vidal et al (2006)

FIG 8 Left Ranges of Log(aO2) conditions calculated for the samples used to compute Fig 7 at the temperature andXFe3+ contents estimated from the Chl-Qtz-H2O equilibrium conditions Dashed lines hematite-magnetite-O2

equilibrium calculated using the thermodynamic properties of Berman (1988) and unit activities of Ht and Mt RightSample calculation at 2 kbar for a Rouez sample using reactions 8ndash20 (upper) or 26ndash40 (lower) The vertical linescorrespond to oxygen- and magnetite-free reactions 1 5 6 and 7 used to compute Figs 5ndash7 which converge at +30degCThe circles in the lower figure represent values calculated with (ΔHdegf ) ferri-Sud constrained with the chlorite-magnetite-quartz-water equilibria 8ndash20 (open circle) or estimated as ΔHdegf (Mg-sudoite) ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2

(filled circle) (see Table 1 and text)

626 O Vidal et al

was kept unchanged and the Fe3+-Al3+ exchange wasassumed to be restricted to the M4 site of chlorite Thechosen formula of ferri-Sud was thus ([Si2]T1[SiAl]T2)

IV([]M1[Mg2Al2]M2-3[Fe3+]M4)

VIO10(OH)8The standard-state thermodynamic properties of allother chlorite end-members the activity model and theMargules parameters were kept unchanged

(2) The third-law entropy of ferri-sudoite (Sdeg =415 JmolK) was estimated according to Holland(1989) with a configurational entropy of 2305 JmolKand a ferromagnetic contribution of 1304 JmolK itsvolume (Vdeg = 204 Jb) was estimated to be 01 Jblower than that of Mg-sudoite (2051 Jb) and the Cp(T ) function was estimated according to Berman amp

FIG 9 Rock-sample mapping of a Sambagawa sample (a) maps of estimated XFe3+ (Vidal et al 2006 Munoz et al2006) and Log(aO2) (estimated assuming that it is buffered by magnetite-chlorite equilibria see text and Fig 8)(b) measured Fe XANES edge-position proportional to Fe redox (Munoz et al 2006) (c) estimated temperature (Vidal

et al 2006 Munoz et al 2006)

627Conditions of formation of chlorite

Brown (1985) The formation enthalpy (ΔHdegf = 82254 plusmn 5 kJmol) was first estimated as the average valuebetween that calculated according to Chermak ampRimstidt (1989) (minus82204 kJmol) and that(minus82304 kJmol) calculted as ΔHdegf (Mg-sudoite) ndashΔHdegf (corundum)2 + ΔHdegf (hematite)2 using thevalues of Vidal et al (1992) for sudoite and Berman(1988) for hematite

(3) In a first step ΔHdegf and Sdeg of ferri-sudoite wereadjusted in order to reproduce the temperaturesreported in Fig 7 with oxygen-free reactions andassuming that magnetite is present in all samples(Chl-Qtz-Mt assemblage) ΔHdegf was allowed to varybetweenminus8220 andminus8230 kJmol and Sdeg within 10 JmolK of the estimated value The addition ofmagnetite and ferri-sudoite to quartz and water leadsto 17 additional equilibria and the following wereused to constrain ΔHdegf and Sdeg of ferri-sudoite

ClinthornMg-Sud frac14 Mg-Amthorn Qtzthorn H2O (1)

4 Clinthorn Daphthorn 10 ferri-Sud

frac14 10 Mg-Amthorn 25 Qtzthorn 30 Mtthorn 20 H2O(24)

7 Daphthorn 70 ferri-Sudthorn 30 Mg-Am

frac14 22 Clinthorn 75 Mg-Sudthorn 35 Mtthorn 40 H2O(25)

28 ferri-Sudthorn 11 Fe-AmthornMg-Am

frac14 6 Daphthorn 30 Mg-Sudthorn 14 Mtthorn 16 H2O(26)

Daphthorn 10 ferri-Sudthorn 10 Mg-Amthorn 10 Qtz

frac14 6 Clinthorn 15 Mg-Sudthorn 5 Mtthorn 16 H2O(27)

The ferri-Sud properties derived with the assumptionof ChlndashQtzndashMt equilibrium are listed in Table 1(referenced a)With these values the oxygen activity atthe P T and (XFe3+)Chl of all equilibria used tocompute Fig 7 can be calculated with the followingoxygen-bearing and magnetite-free equilibria whichconverge with equilibria 1 8 9 and 10 at the same PTand aO2 conditions

16 Qtzthorn 7 Mg-Amthorn 4 Daphthorn 6 Mg-Sud

thorn 5 O2 thorn 12 H2O frac14 20 ferri-Sud(28)

Fe-Amthorn 2 Mg-Amthorn 6 Qtzthorn 4 H2Othorn O2

frac14 4 ferri-Sud(29)

Mg-Amthorn 4 Daphthorn 6 Mg-Sudthorn 2 O2

frac14 3 Fe-Amthorn 8 ferri-Sudthorn 2 Qtz(30)

5 Mg-Amthorn 12 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 8 Fe-Amthorn 28 ferri-Sud

(31)

8 Daphthorn 12 Mg-Sudthorn 3 O2 frac14 7 Fe-Am

thorn 12 ferri-Sudthorn 10 Qtzthorn 4 H2O(32)

6 Clinthorn 4 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Mg-Amthorn 20 ferri-Sudthorn 5 Qtz(33)

4 Clinthorn 20 ferri-Sud frac14 30 Qtz

thorn 15 Mg-Amthorn 4 Daphthorn 5 O2 thorn 20 H2O(34)

90 Mg-Sudthorn 28 Daphthorn 32 Clinthorn 20 H2Othorn 35 O2

frac14 140 ferri-Sudthorn 15 Mg-Am

(35)

15 Qtzthorn 14 Clinthorn 16 Daphthorn 45 Mg-Sud

thorn 20 O2 thorn 20 H2O frac14 80 ferri-Sud(36)

2 Clinthorn FeAmthorn 3 Mg-Sudthorn O2

frac14 2 Mg-Amthorn 4 ferri-Sudthorn Qtz(37)

12 Clinthorn 7 Fe-Amthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 10 Mg-Amthorn 28 ferri-Sud

(38)

5 Qtzthorn 2 Clinthorn 2 Fe-Amthorn 3 Mg-Sud

thorn 2 O2 thorn 4 H2O frac14 8 ferri-Sud(39)

2 Clinthorn 8 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Fe-Amthorn 20 ferri-Sudthorn 5 Qtz(40)

30 Qtzthorn 8 Clinthorn 15 Fe-Amthorn 5 O2 thorn 20 H2O

frac14 8 Daphthorn 20 ferri-Sud

(41)

4 Clinthorn 8 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 3 Fe-Amthorn 28 ferri-Sud

(42)

Because the ferri-sudoite standard-state propertieswere constrained by the chlorite-quartz-magnetite-O2-H2O equilibrium the Log(aO2) calculated withthe equilibria above follow naturally the trend shown inFig 8 and the Log(aO2) values calculated withreactions 28ndash42 of low-temperature samples plot inthe hematite stability field This suggests that theΔGdegPT values of ferri-sudoite derived above with theassumption that aO2 is buffered by the presence ofmagnetite are possibly too high A reduction of ΔHdegf(ferri-sudoite) to 82304 kJmol (=ΔHdegf (Mg-sudoite)ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2 see Table 1)leads to Log(aO2) reduced by 45 Log units (see Fig 8)

628 O Vidal et al

and most calculated values except those of thehematite-bearing St Martin samples plot in themagnetite stability field

A more detailed study is required to refine the ferri-sudoite thermodynamic data However the resultsobtained using the preliminary values listed in Table 1show that this end-member can be used to reproduceconsistently the T-XFe3+ and T- trends obtained fromclassical Chl-Qtz-H2O calculations (Figs 5ndash7) Theincrease in XFe3+ with reducing temperature is consist-ent with the expected reduction in aO2 and the amountof free oxygen that is consumed to produce Fe3+-bearingchlorite Ferric iron is therefore a major component thathas to be taken into account when dealing with low-temperature samples Empirical or semi-empiricalchlorite thermometers which assume that all Fe is inthe ferrous state may include one of the XFe3+-T andaO2-T trends illustrated in Figs 7 and 8 but they cannotaccount for the different XFe3+-T trends that areexpected in the contexts of different aO2 conditions

FUTURE CHALLENGES LOOK INGAT LOW-GRADE ROCKS SUCH ASMETAMORPH IC ROCKS AND THE

DYNAMICS OF CHLOR ITECRYSTALL IZAT ION AT THE

MICROMETR IC TO NANOMETR ICSCALE

T-P-XFe3+ mapping at the thin-section scaleevidence of chlorite composition re-equilibration in metamorphic and low-gradesamples

Although natural samples offer valuable constraintson thermodynamic models there are two maindifficulties in using them to evaluate the reliability ofexisting chlorite models for thermometric purposes

(1) In the case of low-grade samples referencepressures and temperatures are generally known withina reasonable absolute uncertainty but the size ofchlorite grains is often small and good microprobeanalyses are sometimes difficult to obtain Moreoverthe kinetics of reaction are so sluggish that inheritedchlorite grains might remain unaffected by latervariations in temperature which makes thetemperature-composition relationships difficult toconstrain Lastly extrapolation at HT and HP ofempirical or semi-empirical models constrained byLT samples only may be erroneous in particular theeffect of pressure or eventual systematic variations ofXFe3+ cannot be appraised

(2) In the case of metamorphic samplesmicroprobe analyses are easier to obtain from largechlorite grains but chlorites of contrasted compositionthat crystallized at different temperatures and times ofthe metamorphic history are often observed to coexistmetastably in different microstructural sites of the samethin section In such cases the definition of a singlepair of reference temperature and pressure isinappropriate and cannot be restricted to the peak-temperature conditions

In order to overcome these difficulties and to bettervisualize the link between mineral compositiondeformation and P-T conditions at the thin-sectionscale de Andrade et al (2006 2011) Vidal et al(2006) and Lanari et al (2013) developed an approachwhich combines elementary maps of composition witheither a thermodynamic or a semi-empirical estimate ofchlorite crystallization temperature Maps of rawmicroprobe analyses were standardized using good-quality point analyses of minerals present in themapped area The maps of standardized analyseswere transformed into maps of structural formulaewhere each pixel analysed is converted into individualmineral structural formulae Depending on the thermo-metric approach it was then possible to compute mapsof temperature or T-XFe3+ conditions assuming thateach chlorite pixel was in equilibrium with quartz andwater The results reported by Vidal et al (2006)Yamato et al (2007) Ganne et al (2012) Lanari et al(2013 2014) and Scheffer et al (2015) showed a closelink between deformation and mineral reaction at themicroscopic scale and maps of chlorite temperaturecould be used as a basis for interpreting the micro-structures observed in thin sections of low- to middle-grade metapelites In all these studies the metastablepersistence of different generations of chlorite crystal-lizing at different temperatures during the exhumationof metamorphic samples could be used to constrain theT-deformation history Moreover the map of (XFe3+)Chl calculated by Vidal et al (2006) was in goodqualitative agreement with the in situ mapping of thisratio using the X-ray absorption near-edge structure(XANES) techniques (Fig 9) (see also Munoz et al2006) The calculated temperature and XFe3+ mapsobtained on a single sample showed the same trend ofreducing (XFe3+)Chl with increasing temperature asobserved in Figs 6 and 7

Bourdelle et al (2013c) applied a similar approachfor low-temperature samples with chlorite of smallgrain size using transmission electron microscopy-energy dispersive X-ray (TEM-EDX) spectroscopicanalyses of focused-ion beam (FIB) foils instead of

629Conditions of formation of chlorite

microprobe analyses of thin sections (Fig 10) Thoseauthors were able to observe heterogeneities in XFe3+

at the nm scale using STXM-XANES measurementsThis last study was particularly interesting as it showedthat approaches similar to those used for studyingmetamorphic rocks could be applied to low-grade anddiagenetic samples

Looking at diagenetic rocks like metamorphicrocks

The use of equilibrium thermodynamics to studylow-grade rocks has been often criticized becausesluggish kinetics might preclude achievement of globalequilibrium at low temperature and is responsible forthe crystallization of metastable phases with a lowerenergy of nucleation than stable ones Essene amp Peacor(1995) claimed that the coexistence of chemicallyinhomogeneous phases at a very fine scale in low-grade rocks is indicative of disequilibrium and thatmineral reaction is better described by a kinetic ratherthan thermodynamic approach However the coexist-ence of chemically inhomogeneous phases is also acommon observation in metamorphic rocks wherezoned grains and the coexistence for example ofchlorite grains of different compositions indicate thatthermodynamic equilibrium is approached only locallyand at a given time in the metamorphic history In factthe main difference between diagenetic and higher-temperature rocks is the size of the domains in localequilibrium The more sluggish kinetics of reactionand diffusivity at low temperature lead to small grainsand domains in equilibrium but these domains can beidentified when using appropriate analytical tools andprocedures developed recently (Bourdelle et al2013bc)

FIG 10 Quantitative Fe redox nanomapping on a FIB foilof Glarus sample (T lt 300degC Switzerland) calculatedfrom the 7087ndash71025 eV images ratio (STXM-XANES)

(from Bourdelle et al 2013c)

FIG 11 Analysis of an illite-chlorite assemblage (Gulf Coast sample Alamo 1 well 5825 m) (a) SEM image of thepetrographic thin section showing the emplacement of the FIB foil to be cut across the illitendashchlorite interface (b)Bright-field TEM image of the FIB foil extracted from the section with the platinum strap on top of it (cd) Enlargementof part b showing the Al and Mg contents of chlorite (apfu) and their variation between crystal core and crystal rim

630 O Vidal et al

Figure 10 shows that it is possible to demonstrateheterogeneities in XFe3+ contents at the nanometerscale Consistent variations of other major elements arealso observed with TEM technics on FIB foils Anexample is shown in Fig 11 which depicts thevariations of Al and Mg contents in a chlorite grainfrom the Gulf Coast (Bourdelle et al 2013b) In thiscase the FIB foil was cut parallel to the interfacebetween illite and chlorite and a nm-scale investiga-tion of composition was conducted by TEM-EDXanalysis The rim and core compositions showed azoning of chlorite with Mg- and Al-richer rims thancore compositions This variation in composition on ascale of lt1 μm was related to various crystallizationstages during burial Chlorite sampled at differentdepths and various locations of the Gulf Coast areashowed that zoned compositions similar to thatobserved in Fig 11 are frequent (Bourdelle et al2013b) The AlIV in chlorite rims display a Tdependence that is consistent with the km-scale andregional evolution of temperature ie with thediagenetic history of Gulf Coast (burial P-T condi-tions detritalauthigenic compositions etchellip)(Fig 12) This study suggests again that approachesused for metamorphic rocks can be applied to low-grade rocks when using adapted analytical tools able toidentify sub-μm variations of composition

CONCLUS ION

The variability of chlorite composition reflects thevariability of its conditions of formation Numerousefforts have been made over the last 30 years to

translate the composition of chlorite in terms of itstemperature of formation The first empirical equationsbased on the AlIVor vacancy contents are restricted tobulk-rock compositions and pressure conditions atwhich they were calibrated Thermodynamicapproaches and semi-empirical equations based onthe ratio of end-member activities do not present thislimit Among the different thermodynamic models ofchlorite published so far Vidal et al (2006) proposed amodel that can be used to estimate T and (XFe3+)Chlsimultaneously The estimated (XFe3+)Chl values areconsistent with those measured by Moumlssbauer orXANES spectroscopy and for the same (XFe3+)Chlthe temperatures estimated using the method of Vidalet al (2006) are similar to those obtained with thesemi-empirical equations of Inoue et al (2009) andLanari et al (2014) Different T-XFe3+ trends areobserved for different pressure gradients and oxygen-buffering assemblages (Lanari et al 2014) butoverall the systematic increase in XFe3+ with areduction in temperature is compatible with thedecrease in aO2 in a closed system The increase inXFe3+ and chlorite vacancy with decreasing tempera-ture is interpreted as reflecting the incorporation of theFe3+-sudoite component which can be used toestimate T-aO2-XFe

3+ values with a Chl-Qtz-H2Omulti-equilibrium approach

Similar temperatures are calculated for low-tem-perature chlorites (T lt 300degC and measured XFe3+ upto 30) with all approaches assuming that iron inchlorite is either completely ferrous or with estimatedvalues of XFe3+ This indicates that the equations ofsemi-empirical thermometers assuming that all iron inchlorite is ferrous include some inherent Fe2+Fe3+ vstemperature correction However very different tem-peratures can be obtained for higher-temperaturesamples (gt300degC) with approaches involving ferriciron or assuming that all of the iron is ferrous Thissuggests that the Fe2+Fe3+ vs temperature correctionmight be restricted to particular aO2 conditions andpressure gradients More work is necessary to betterunderstand this point and to improve the thermo-dynamic models and semi-empirical thermometersThis requires high-quality and well constrained naturaldata To this end mapping techniques that highlight thelink between the variations in temperature and pressureand small-scale chlorite composition including XFe3+

variations offer much promise When combining theresults of the present study with those of other studiespublished recently it appears that thermodynamicapproaches and mapping techniques developed formetamorphic rocks could be used to study low-grade

FIG 12 Range of crystal-rim AlIV content (white boxes)of Gulf Coast chlorites vs measured temperatures Thegrey area underlines the range of maximum AlIV contentwhich is assumed to represent the closest approach toequilibrium composition (after Bourdelle et al 2013b)

631Conditions of formation of chlorite

rocks where kinetics are much more sluggish Thisrequires use of appropriate analytical tools and techni-ques with a spatial resolution of just a few hundrednanometres which are now becoming available

ACKNOWLEDGMENTS

Constructive reviews by S Hillier F Nieto and ananonymous reviewer are gratefully acknowledged

REFERENCES

Beaufort D Patrier P Meunier A amp Ottaviani MM(1992) Chemical variations in assemblages includingepidote andor chlorite in the fossil hydrothermalsystem of Saint Martin (Lesser Antilles) Journal ofVolcanology and Geothermal Research 51 95ndash114

Berman R (1988) Internally-consistent thermodynamicdata for minerals in the system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2Journal of Petrology 29 445ndash522

Berman R amp Brown T (1985) Heat capacity of minerals inthe system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2 representation estima-tion and high temperature extrapolation Contributionsto Mineralogy and Petrology 89 168ndash183

Bourdelle F amp Cathelineau M (2015) Low-temperaturechlorite geothermometry a graphical representationbased on a TndashR2+ndashSi diagram European Journal ofMineralogy 27 617ndash626

Bourdelle F Parra T Chopin C amp Beyssac O (2013a) Anew chlorite geothermometer for diagenetic to low-grade metamorphic conditions Contributions toMineralogy and Petrology 165 723ndash735

Bourdelle F Parra T Beyssac O Chopin C amp Vidal O(2013b) Clay minerals thermometry a comparativestudy based on high-resolution chemical analyses ofillite and chlorite in sandstones from the Gulf Coast(Texas USA) American Mineralogist 98 914ndash926

Bourdelle F Benzerara K Beyssac O Cosmidis JNeuville DR Brown GE amp Paineau E (2013c)Quantification of the ferric ferrous iron ratio insilicates by scanning transmission X-ray microscopyat the Fe L23 edges Contribution to Mineralogy andPetrology 166 423ndash434

Cathelineau M (1988) Cation site occupancy in chloritesand illites as function of temperature Clay Minerals23 471ndash485

Cathelineau M amp Nieva D (1985) A chlorite solidsolution geothermometer the Los Azufres (Mexico)geothermal system Contributions to Mineralogy andPetrology 91 235ndash244

Chermak JA amp Rimstidt JD (1989) Estimating thethermodynamic properties (ΔGdegf and ΔHdegf) of silicate

minerals at 298K from the sum of polyhedral contribu-tions American Mineralogist 74 1023ndash1031

Coggon R amp Holland TJB (2002) Mixing properties ofphengitic micas and revised garnent-phengite thermo-barometers Journal of Metamorphic Geology 20683ndash696

Curtis CD Hughes CR Whiteman JA ampWhittle CK(1985) Compositional variation within some sedi-mentary chlorites and some comments on their originMineralogical Magazine 49 375ndash386

de Andrade V Vidal O Lewin E OrsquoBrien P amp Agard P(2006) Quantification of electron microprobe com-positional maps of rock thin sections an optimizedmethod and examples Journal of MetamorphicGeology 24 655ndash668

de Andrade V Susini J Salomeacute M Beraldin O RigaultC Heymes T Lewin E amp Vidal O (2011)Submicrometer hyperspectral X-ray imaging of het-erogeneous rocks and geomaterials applications at theFe k-edge Analytical Chemistry 83 4220ndash4227

de Caritat P Hutcheon I amp Walshe J (1993) Chloritegeothermometry a review Clays and Clay Minerals41 219ndash239

Essene EJ (2009) Thermobarometry gone astray Pp101ndash129 in Physics and Chemistry of Earthrsquos Interior(AK Gupta amp S Dasgupta editors) PlatinumJubilee Indian National Science Academy Springer-Verlag Berlin

Essene EJ amp Peacor DR (1995) Clay mineralthermometry ndash a critical perspective Clays and ClayMinerals 43 540ndash553

Ganne J de Andrade V Weinberg RF Vidal O DubacqB Kagambega N Naba S Baratoux L Jessell M ampAllibon J (2012) Modern-style plate subductionpreserved in the Palaeoproterozoic West Africancraton Nature Geoscience 5 60ndash65

Grosch EG Vidal O Abu-Alam T amp McLoughlin N(2012) PTconstraints on the metamorphic evolution ofthe Paleoarchean Kromberg type-section BarbertonGreenstone Belt South-Africa Journal of Petrology53 513ndash545

Grosch E McLoughlin N Lanari P Erambert M ampVidal O (2014) Microscale mapping of alterationconditions and potential biosignatures in basaltic-ultramafic rocks on early Earth and beyondAstrobiology 14 216ndash228

Hillier S amp Velde B (1991) Octahedral occupancy andchemical composition of diagenetic (low-temperature)chlorites Clay Minerals 26 149ndash168

Holland TJB (1989) Dependance of entropy on volumefor silicates and oxide minerals A review and apredictive model American Mineralogist 74 5ndash13

Holland T amp Powell R (1998) An internally consistentthermodynamic data set for phases of petrologicalinterest Journal ofMetamorphic Geology 16 309ndash343

Holland T Baker J amp Powell R (1998)Mixing propertiesand activity composition and relationships of chlorites

632 O Vidal et al

Martin samples follow the same trend and decreasewith decreasing temperature from simminus27 at 450ndash500degC10 kbar (high-temperature chlorites of Sambagawa) tominus50degC at 125degC and 1 kbar (Niger samples) (Fig 8)The Log(aO2) values of the high-T and -P samples plotin the stability field of magnetite whereas those of thelow-T samples plot in the stability field of hematitewhich is not consistent with the assumption of ChlndashMtndashQtzndashH2O equilibrium The presence of hematite hasbeen reported in the Saint-Martin samples only(Beaufort et al 1992) which plot above the trenddefined by the other samples (upper grey ellipse inFig 8) The relative position of this hematite-bearingsample is consistent with expected higher oxygenactivities required for its formation For the other low-T and apparently hematite-free samples it is possiblethat the calculated Log(aO2) values are too high or thatoxygen was not buffered by an equilibrium ofmagnetitewith chlorite In any case Fig 8 indicates that thesystematic increase in (XFe3+)Chl with decreasingtemperature observed in Figs 6 and 7 is consistentwith a lower aO2 and amount of free oxygen at low-Tthan high-T in systems where aO2 is buffered by thepresence of chlorite and iron oxides To maintain this

equilibrium with lowering temperature in metamorphicrocks where oxygen is immobile oxygen has to beconsumed and one way for that is to produce chloritethat is richer in Fe3+ This is a possible interpretationfor the formation of low-temperature Fe3+-rich chloritecoexisting with high-temperature Fe3+-poor chloritein the same thin section of the Sambagwa sample(Fig 9)

In order to discuss the Log(aO2)-T-XFe3+ evolution

in more detail for rocks where aO2 is not buffered bythe presence of magnetite it is necessary to define anFe3+-bearing chlorite end-member The simultaneousincrease of vacancy and Fe3+ content with decreasingtemperature observed in Figs 5ndash7 is explained by the3(Mg Fe2+) = 2(Fe3+) ditrioctahedral substitution(see also Trincal amp Lanari 2016 this issue) Thisexchange can be modelled with a ferri-sudoite end-member that materializes the relationship betweenXFe3+ and vacancy The standard-state thermo-dynamic properties of ferri-sudoite were constrainedas follows

(1) Because the maximum amount of (Fe3+)Chlcalculated in the previous section never exceeded1 apfu the cation distribution of Vidal et al (2006)

FIG 8 Left Ranges of Log(aO2) conditions calculated for the samples used to compute Fig 7 at the temperature andXFe3+ contents estimated from the Chl-Qtz-H2O equilibrium conditions Dashed lines hematite-magnetite-O2

equilibrium calculated using the thermodynamic properties of Berman (1988) and unit activities of Ht and Mt RightSample calculation at 2 kbar for a Rouez sample using reactions 8ndash20 (upper) or 26ndash40 (lower) The vertical linescorrespond to oxygen- and magnetite-free reactions 1 5 6 and 7 used to compute Figs 5ndash7 which converge at +30degCThe circles in the lower figure represent values calculated with (ΔHdegf ) ferri-Sud constrained with the chlorite-magnetite-quartz-water equilibria 8ndash20 (open circle) or estimated as ΔHdegf (Mg-sudoite) ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2

(filled circle) (see Table 1 and text)

626 O Vidal et al

was kept unchanged and the Fe3+-Al3+ exchange wasassumed to be restricted to the M4 site of chlorite Thechosen formula of ferri-Sud was thus ([Si2]T1[SiAl]T2)

IV([]M1[Mg2Al2]M2-3[Fe3+]M4)

VIO10(OH)8The standard-state thermodynamic properties of allother chlorite end-members the activity model and theMargules parameters were kept unchanged

(2) The third-law entropy of ferri-sudoite (Sdeg =415 JmolK) was estimated according to Holland(1989) with a configurational entropy of 2305 JmolKand a ferromagnetic contribution of 1304 JmolK itsvolume (Vdeg = 204 Jb) was estimated to be 01 Jblower than that of Mg-sudoite (2051 Jb) and the Cp(T ) function was estimated according to Berman amp

FIG 9 Rock-sample mapping of a Sambagawa sample (a) maps of estimated XFe3+ (Vidal et al 2006 Munoz et al2006) and Log(aO2) (estimated assuming that it is buffered by magnetite-chlorite equilibria see text and Fig 8)(b) measured Fe XANES edge-position proportional to Fe redox (Munoz et al 2006) (c) estimated temperature (Vidal

et al 2006 Munoz et al 2006)

627Conditions of formation of chlorite

Brown (1985) The formation enthalpy (ΔHdegf = 82254 plusmn 5 kJmol) was first estimated as the average valuebetween that calculated according to Chermak ampRimstidt (1989) (minus82204 kJmol) and that(minus82304 kJmol) calculted as ΔHdegf (Mg-sudoite) ndashΔHdegf (corundum)2 + ΔHdegf (hematite)2 using thevalues of Vidal et al (1992) for sudoite and Berman(1988) for hematite

(3) In a first step ΔHdegf and Sdeg of ferri-sudoite wereadjusted in order to reproduce the temperaturesreported in Fig 7 with oxygen-free reactions andassuming that magnetite is present in all samples(Chl-Qtz-Mt assemblage) ΔHdegf was allowed to varybetweenminus8220 andminus8230 kJmol and Sdeg within 10 JmolK of the estimated value The addition ofmagnetite and ferri-sudoite to quartz and water leadsto 17 additional equilibria and the following wereused to constrain ΔHdegf and Sdeg of ferri-sudoite

ClinthornMg-Sud frac14 Mg-Amthorn Qtzthorn H2O (1)

4 Clinthorn Daphthorn 10 ferri-Sud

frac14 10 Mg-Amthorn 25 Qtzthorn 30 Mtthorn 20 H2O(24)

7 Daphthorn 70 ferri-Sudthorn 30 Mg-Am

frac14 22 Clinthorn 75 Mg-Sudthorn 35 Mtthorn 40 H2O(25)

28 ferri-Sudthorn 11 Fe-AmthornMg-Am

frac14 6 Daphthorn 30 Mg-Sudthorn 14 Mtthorn 16 H2O(26)

Daphthorn 10 ferri-Sudthorn 10 Mg-Amthorn 10 Qtz

frac14 6 Clinthorn 15 Mg-Sudthorn 5 Mtthorn 16 H2O(27)

The ferri-Sud properties derived with the assumptionof ChlndashQtzndashMt equilibrium are listed in Table 1(referenced a)With these values the oxygen activity atthe P T and (XFe3+)Chl of all equilibria used tocompute Fig 7 can be calculated with the followingoxygen-bearing and magnetite-free equilibria whichconverge with equilibria 1 8 9 and 10 at the same PTand aO2 conditions

16 Qtzthorn 7 Mg-Amthorn 4 Daphthorn 6 Mg-Sud

thorn 5 O2 thorn 12 H2O frac14 20 ferri-Sud(28)

Fe-Amthorn 2 Mg-Amthorn 6 Qtzthorn 4 H2Othorn O2

frac14 4 ferri-Sud(29)

Mg-Amthorn 4 Daphthorn 6 Mg-Sudthorn 2 O2

frac14 3 Fe-Amthorn 8 ferri-Sudthorn 2 Qtz(30)

5 Mg-Amthorn 12 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 8 Fe-Amthorn 28 ferri-Sud

(31)

8 Daphthorn 12 Mg-Sudthorn 3 O2 frac14 7 Fe-Am

thorn 12 ferri-Sudthorn 10 Qtzthorn 4 H2O(32)

6 Clinthorn 4 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Mg-Amthorn 20 ferri-Sudthorn 5 Qtz(33)

4 Clinthorn 20 ferri-Sud frac14 30 Qtz

thorn 15 Mg-Amthorn 4 Daphthorn 5 O2 thorn 20 H2O(34)

90 Mg-Sudthorn 28 Daphthorn 32 Clinthorn 20 H2Othorn 35 O2

frac14 140 ferri-Sudthorn 15 Mg-Am

(35)

15 Qtzthorn 14 Clinthorn 16 Daphthorn 45 Mg-Sud

thorn 20 O2 thorn 20 H2O frac14 80 ferri-Sud(36)

2 Clinthorn FeAmthorn 3 Mg-Sudthorn O2

frac14 2 Mg-Amthorn 4 ferri-Sudthorn Qtz(37)

12 Clinthorn 7 Fe-Amthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 10 Mg-Amthorn 28 ferri-Sud

(38)

5 Qtzthorn 2 Clinthorn 2 Fe-Amthorn 3 Mg-Sud

thorn 2 O2 thorn 4 H2O frac14 8 ferri-Sud(39)

2 Clinthorn 8 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Fe-Amthorn 20 ferri-Sudthorn 5 Qtz(40)

30 Qtzthorn 8 Clinthorn 15 Fe-Amthorn 5 O2 thorn 20 H2O

frac14 8 Daphthorn 20 ferri-Sud

(41)

4 Clinthorn 8 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 3 Fe-Amthorn 28 ferri-Sud

(42)

Because the ferri-sudoite standard-state propertieswere constrained by the chlorite-quartz-magnetite-O2-H2O equilibrium the Log(aO2) calculated withthe equilibria above follow naturally the trend shown inFig 8 and the Log(aO2) values calculated withreactions 28ndash42 of low-temperature samples plot inthe hematite stability field This suggests that theΔGdegPT values of ferri-sudoite derived above with theassumption that aO2 is buffered by the presence ofmagnetite are possibly too high A reduction of ΔHdegf(ferri-sudoite) to 82304 kJmol (=ΔHdegf (Mg-sudoite)ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2 see Table 1)leads to Log(aO2) reduced by 45 Log units (see Fig 8)

628 O Vidal et al

and most calculated values except those of thehematite-bearing St Martin samples plot in themagnetite stability field

A more detailed study is required to refine the ferri-sudoite thermodynamic data However the resultsobtained using the preliminary values listed in Table 1show that this end-member can be used to reproduceconsistently the T-XFe3+ and T- trends obtained fromclassical Chl-Qtz-H2O calculations (Figs 5ndash7) Theincrease in XFe3+ with reducing temperature is consist-ent with the expected reduction in aO2 and the amountof free oxygen that is consumed to produce Fe3+-bearingchlorite Ferric iron is therefore a major component thathas to be taken into account when dealing with low-temperature samples Empirical or semi-empiricalchlorite thermometers which assume that all Fe is inthe ferrous state may include one of the XFe3+-T andaO2-T trends illustrated in Figs 7 and 8 but they cannotaccount for the different XFe3+-T trends that areexpected in the contexts of different aO2 conditions

FUTURE CHALLENGES LOOK INGAT LOW-GRADE ROCKS SUCH ASMETAMORPH IC ROCKS AND THE

DYNAMICS OF CHLOR ITECRYSTALL IZAT ION AT THE

MICROMETR IC TO NANOMETR ICSCALE

T-P-XFe3+ mapping at the thin-section scaleevidence of chlorite composition re-equilibration in metamorphic and low-gradesamples

Although natural samples offer valuable constraintson thermodynamic models there are two maindifficulties in using them to evaluate the reliability ofexisting chlorite models for thermometric purposes

(1) In the case of low-grade samples referencepressures and temperatures are generally known withina reasonable absolute uncertainty but the size ofchlorite grains is often small and good microprobeanalyses are sometimes difficult to obtain Moreoverthe kinetics of reaction are so sluggish that inheritedchlorite grains might remain unaffected by latervariations in temperature which makes thetemperature-composition relationships difficult toconstrain Lastly extrapolation at HT and HP ofempirical or semi-empirical models constrained byLT samples only may be erroneous in particular theeffect of pressure or eventual systematic variations ofXFe3+ cannot be appraised

(2) In the case of metamorphic samplesmicroprobe analyses are easier to obtain from largechlorite grains but chlorites of contrasted compositionthat crystallized at different temperatures and times ofthe metamorphic history are often observed to coexistmetastably in different microstructural sites of the samethin section In such cases the definition of a singlepair of reference temperature and pressure isinappropriate and cannot be restricted to the peak-temperature conditions

In order to overcome these difficulties and to bettervisualize the link between mineral compositiondeformation and P-T conditions at the thin-sectionscale de Andrade et al (2006 2011) Vidal et al(2006) and Lanari et al (2013) developed an approachwhich combines elementary maps of composition witheither a thermodynamic or a semi-empirical estimate ofchlorite crystallization temperature Maps of rawmicroprobe analyses were standardized using good-quality point analyses of minerals present in themapped area The maps of standardized analyseswere transformed into maps of structural formulaewhere each pixel analysed is converted into individualmineral structural formulae Depending on the thermo-metric approach it was then possible to compute mapsof temperature or T-XFe3+ conditions assuming thateach chlorite pixel was in equilibrium with quartz andwater The results reported by Vidal et al (2006)Yamato et al (2007) Ganne et al (2012) Lanari et al(2013 2014) and Scheffer et al (2015) showed a closelink between deformation and mineral reaction at themicroscopic scale and maps of chlorite temperaturecould be used as a basis for interpreting the micro-structures observed in thin sections of low- to middle-grade metapelites In all these studies the metastablepersistence of different generations of chlorite crystal-lizing at different temperatures during the exhumationof metamorphic samples could be used to constrain theT-deformation history Moreover the map of (XFe3+)Chl calculated by Vidal et al (2006) was in goodqualitative agreement with the in situ mapping of thisratio using the X-ray absorption near-edge structure(XANES) techniques (Fig 9) (see also Munoz et al2006) The calculated temperature and XFe3+ mapsobtained on a single sample showed the same trend ofreducing (XFe3+)Chl with increasing temperature asobserved in Figs 6 and 7

Bourdelle et al (2013c) applied a similar approachfor low-temperature samples with chlorite of smallgrain size using transmission electron microscopy-energy dispersive X-ray (TEM-EDX) spectroscopicanalyses of focused-ion beam (FIB) foils instead of

629Conditions of formation of chlorite

microprobe analyses of thin sections (Fig 10) Thoseauthors were able to observe heterogeneities in XFe3+

at the nm scale using STXM-XANES measurementsThis last study was particularly interesting as it showedthat approaches similar to those used for studyingmetamorphic rocks could be applied to low-grade anddiagenetic samples

Looking at diagenetic rocks like metamorphicrocks

The use of equilibrium thermodynamics to studylow-grade rocks has been often criticized becausesluggish kinetics might preclude achievement of globalequilibrium at low temperature and is responsible forthe crystallization of metastable phases with a lowerenergy of nucleation than stable ones Essene amp Peacor(1995) claimed that the coexistence of chemicallyinhomogeneous phases at a very fine scale in low-grade rocks is indicative of disequilibrium and thatmineral reaction is better described by a kinetic ratherthan thermodynamic approach However the coexist-ence of chemically inhomogeneous phases is also acommon observation in metamorphic rocks wherezoned grains and the coexistence for example ofchlorite grains of different compositions indicate thatthermodynamic equilibrium is approached only locallyand at a given time in the metamorphic history In factthe main difference between diagenetic and higher-temperature rocks is the size of the domains in localequilibrium The more sluggish kinetics of reactionand diffusivity at low temperature lead to small grainsand domains in equilibrium but these domains can beidentified when using appropriate analytical tools andprocedures developed recently (Bourdelle et al2013bc)

FIG 10 Quantitative Fe redox nanomapping on a FIB foilof Glarus sample (T lt 300degC Switzerland) calculatedfrom the 7087ndash71025 eV images ratio (STXM-XANES)

(from Bourdelle et al 2013c)

FIG 11 Analysis of an illite-chlorite assemblage (Gulf Coast sample Alamo 1 well 5825 m) (a) SEM image of thepetrographic thin section showing the emplacement of the FIB foil to be cut across the illitendashchlorite interface (b)Bright-field TEM image of the FIB foil extracted from the section with the platinum strap on top of it (cd) Enlargementof part b showing the Al and Mg contents of chlorite (apfu) and their variation between crystal core and crystal rim

630 O Vidal et al

Figure 10 shows that it is possible to demonstrateheterogeneities in XFe3+ contents at the nanometerscale Consistent variations of other major elements arealso observed with TEM technics on FIB foils Anexample is shown in Fig 11 which depicts thevariations of Al and Mg contents in a chlorite grainfrom the Gulf Coast (Bourdelle et al 2013b) In thiscase the FIB foil was cut parallel to the interfacebetween illite and chlorite and a nm-scale investiga-tion of composition was conducted by TEM-EDXanalysis The rim and core compositions showed azoning of chlorite with Mg- and Al-richer rims thancore compositions This variation in composition on ascale of lt1 μm was related to various crystallizationstages during burial Chlorite sampled at differentdepths and various locations of the Gulf Coast areashowed that zoned compositions similar to thatobserved in Fig 11 are frequent (Bourdelle et al2013b) The AlIV in chlorite rims display a Tdependence that is consistent with the km-scale andregional evolution of temperature ie with thediagenetic history of Gulf Coast (burial P-T condi-tions detritalauthigenic compositions etchellip)(Fig 12) This study suggests again that approachesused for metamorphic rocks can be applied to low-grade rocks when using adapted analytical tools able toidentify sub-μm variations of composition

CONCLUS ION

The variability of chlorite composition reflects thevariability of its conditions of formation Numerousefforts have been made over the last 30 years to

translate the composition of chlorite in terms of itstemperature of formation The first empirical equationsbased on the AlIVor vacancy contents are restricted tobulk-rock compositions and pressure conditions atwhich they were calibrated Thermodynamicapproaches and semi-empirical equations based onthe ratio of end-member activities do not present thislimit Among the different thermodynamic models ofchlorite published so far Vidal et al (2006) proposed amodel that can be used to estimate T and (XFe3+)Chlsimultaneously The estimated (XFe3+)Chl values areconsistent with those measured by Moumlssbauer orXANES spectroscopy and for the same (XFe3+)Chlthe temperatures estimated using the method of Vidalet al (2006) are similar to those obtained with thesemi-empirical equations of Inoue et al (2009) andLanari et al (2014) Different T-XFe3+ trends areobserved for different pressure gradients and oxygen-buffering assemblages (Lanari et al 2014) butoverall the systematic increase in XFe3+ with areduction in temperature is compatible with thedecrease in aO2 in a closed system The increase inXFe3+ and chlorite vacancy with decreasing tempera-ture is interpreted as reflecting the incorporation of theFe3+-sudoite component which can be used toestimate T-aO2-XFe

3+ values with a Chl-Qtz-H2Omulti-equilibrium approach

Similar temperatures are calculated for low-tem-perature chlorites (T lt 300degC and measured XFe3+ upto 30) with all approaches assuming that iron inchlorite is either completely ferrous or with estimatedvalues of XFe3+ This indicates that the equations ofsemi-empirical thermometers assuming that all iron inchlorite is ferrous include some inherent Fe2+Fe3+ vstemperature correction However very different tem-peratures can be obtained for higher-temperaturesamples (gt300degC) with approaches involving ferriciron or assuming that all of the iron is ferrous Thissuggests that the Fe2+Fe3+ vs temperature correctionmight be restricted to particular aO2 conditions andpressure gradients More work is necessary to betterunderstand this point and to improve the thermo-dynamic models and semi-empirical thermometersThis requires high-quality and well constrained naturaldata To this end mapping techniques that highlight thelink between the variations in temperature and pressureand small-scale chlorite composition including XFe3+

variations offer much promise When combining theresults of the present study with those of other studiespublished recently it appears that thermodynamicapproaches and mapping techniques developed formetamorphic rocks could be used to study low-grade

FIG 12 Range of crystal-rim AlIV content (white boxes)of Gulf Coast chlorites vs measured temperatures Thegrey area underlines the range of maximum AlIV contentwhich is assumed to represent the closest approach toequilibrium composition (after Bourdelle et al 2013b)

631Conditions of formation of chlorite

rocks where kinetics are much more sluggish Thisrequires use of appropriate analytical tools and techni-ques with a spatial resolution of just a few hundrednanometres which are now becoming available

ACKNOWLEDGMENTS

Constructive reviews by S Hillier F Nieto and ananonymous reviewer are gratefully acknowledged

REFERENCES

Beaufort D Patrier P Meunier A amp Ottaviani MM(1992) Chemical variations in assemblages includingepidote andor chlorite in the fossil hydrothermalsystem of Saint Martin (Lesser Antilles) Journal ofVolcanology and Geothermal Research 51 95ndash114

Berman R (1988) Internally-consistent thermodynamicdata for minerals in the system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2Journal of Petrology 29 445ndash522

Berman R amp Brown T (1985) Heat capacity of minerals inthe system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2 representation estima-tion and high temperature extrapolation Contributionsto Mineralogy and Petrology 89 168ndash183

Bourdelle F amp Cathelineau M (2015) Low-temperaturechlorite geothermometry a graphical representationbased on a TndashR2+ndashSi diagram European Journal ofMineralogy 27 617ndash626

Bourdelle F Parra T Chopin C amp Beyssac O (2013a) Anew chlorite geothermometer for diagenetic to low-grade metamorphic conditions Contributions toMineralogy and Petrology 165 723ndash735

Bourdelle F Parra T Beyssac O Chopin C amp Vidal O(2013b) Clay minerals thermometry a comparativestudy based on high-resolution chemical analyses ofillite and chlorite in sandstones from the Gulf Coast(Texas USA) American Mineralogist 98 914ndash926

Bourdelle F Benzerara K Beyssac O Cosmidis JNeuville DR Brown GE amp Paineau E (2013c)Quantification of the ferric ferrous iron ratio insilicates by scanning transmission X-ray microscopyat the Fe L23 edges Contribution to Mineralogy andPetrology 166 423ndash434

Cathelineau M (1988) Cation site occupancy in chloritesand illites as function of temperature Clay Minerals23 471ndash485

Cathelineau M amp Nieva D (1985) A chlorite solidsolution geothermometer the Los Azufres (Mexico)geothermal system Contributions to Mineralogy andPetrology 91 235ndash244

Chermak JA amp Rimstidt JD (1989) Estimating thethermodynamic properties (ΔGdegf and ΔHdegf) of silicate

minerals at 298K from the sum of polyhedral contribu-tions American Mineralogist 74 1023ndash1031

Coggon R amp Holland TJB (2002) Mixing properties ofphengitic micas and revised garnent-phengite thermo-barometers Journal of Metamorphic Geology 20683ndash696

Curtis CD Hughes CR Whiteman JA ampWhittle CK(1985) Compositional variation within some sedi-mentary chlorites and some comments on their originMineralogical Magazine 49 375ndash386

de Andrade V Vidal O Lewin E OrsquoBrien P amp Agard P(2006) Quantification of electron microprobe com-positional maps of rock thin sections an optimizedmethod and examples Journal of MetamorphicGeology 24 655ndash668

de Andrade V Susini J Salomeacute M Beraldin O RigaultC Heymes T Lewin E amp Vidal O (2011)Submicrometer hyperspectral X-ray imaging of het-erogeneous rocks and geomaterials applications at theFe k-edge Analytical Chemistry 83 4220ndash4227

de Caritat P Hutcheon I amp Walshe J (1993) Chloritegeothermometry a review Clays and Clay Minerals41 219ndash239

Essene EJ (2009) Thermobarometry gone astray Pp101ndash129 in Physics and Chemistry of Earthrsquos Interior(AK Gupta amp S Dasgupta editors) PlatinumJubilee Indian National Science Academy Springer-Verlag Berlin

Essene EJ amp Peacor DR (1995) Clay mineralthermometry ndash a critical perspective Clays and ClayMinerals 43 540ndash553

Ganne J de Andrade V Weinberg RF Vidal O DubacqB Kagambega N Naba S Baratoux L Jessell M ampAllibon J (2012) Modern-style plate subductionpreserved in the Palaeoproterozoic West Africancraton Nature Geoscience 5 60ndash65

Grosch EG Vidal O Abu-Alam T amp McLoughlin N(2012) PTconstraints on the metamorphic evolution ofthe Paleoarchean Kromberg type-section BarbertonGreenstone Belt South-Africa Journal of Petrology53 513ndash545

Grosch E McLoughlin N Lanari P Erambert M ampVidal O (2014) Microscale mapping of alterationconditions and potential biosignatures in basaltic-ultramafic rocks on early Earth and beyondAstrobiology 14 216ndash228

Hillier S amp Velde B (1991) Octahedral occupancy andchemical composition of diagenetic (low-temperature)chlorites Clay Minerals 26 149ndash168

Holland TJB (1989) Dependance of entropy on volumefor silicates and oxide minerals A review and apredictive model American Mineralogist 74 5ndash13

Holland T amp Powell R (1998) An internally consistentthermodynamic data set for phases of petrologicalinterest Journal ofMetamorphic Geology 16 309ndash343

Holland T Baker J amp Powell R (1998)Mixing propertiesand activity composition and relationships of chlorites

632 O Vidal et al

was kept unchanged and the Fe3+-Al3+ exchange wasassumed to be restricted to the M4 site of chlorite Thechosen formula of ferri-Sud was thus ([Si2]T1[SiAl]T2)

IV([]M1[Mg2Al2]M2-3[Fe3+]M4)

VIO10(OH)8The standard-state thermodynamic properties of allother chlorite end-members the activity model and theMargules parameters were kept unchanged

(2) The third-law entropy of ferri-sudoite (Sdeg =415 JmolK) was estimated according to Holland(1989) with a configurational entropy of 2305 JmolKand a ferromagnetic contribution of 1304 JmolK itsvolume (Vdeg = 204 Jb) was estimated to be 01 Jblower than that of Mg-sudoite (2051 Jb) and the Cp(T ) function was estimated according to Berman amp

FIG 9 Rock-sample mapping of a Sambagawa sample (a) maps of estimated XFe3+ (Vidal et al 2006 Munoz et al2006) and Log(aO2) (estimated assuming that it is buffered by magnetite-chlorite equilibria see text and Fig 8)(b) measured Fe XANES edge-position proportional to Fe redox (Munoz et al 2006) (c) estimated temperature (Vidal

et al 2006 Munoz et al 2006)

627Conditions of formation of chlorite

Brown (1985) The formation enthalpy (ΔHdegf = 82254 plusmn 5 kJmol) was first estimated as the average valuebetween that calculated according to Chermak ampRimstidt (1989) (minus82204 kJmol) and that(minus82304 kJmol) calculted as ΔHdegf (Mg-sudoite) ndashΔHdegf (corundum)2 + ΔHdegf (hematite)2 using thevalues of Vidal et al (1992) for sudoite and Berman(1988) for hematite

(3) In a first step ΔHdegf and Sdeg of ferri-sudoite wereadjusted in order to reproduce the temperaturesreported in Fig 7 with oxygen-free reactions andassuming that magnetite is present in all samples(Chl-Qtz-Mt assemblage) ΔHdegf was allowed to varybetweenminus8220 andminus8230 kJmol and Sdeg within 10 JmolK of the estimated value The addition ofmagnetite and ferri-sudoite to quartz and water leadsto 17 additional equilibria and the following wereused to constrain ΔHdegf and Sdeg of ferri-sudoite

ClinthornMg-Sud frac14 Mg-Amthorn Qtzthorn H2O (1)

4 Clinthorn Daphthorn 10 ferri-Sud

frac14 10 Mg-Amthorn 25 Qtzthorn 30 Mtthorn 20 H2O(24)

7 Daphthorn 70 ferri-Sudthorn 30 Mg-Am

frac14 22 Clinthorn 75 Mg-Sudthorn 35 Mtthorn 40 H2O(25)

28 ferri-Sudthorn 11 Fe-AmthornMg-Am

frac14 6 Daphthorn 30 Mg-Sudthorn 14 Mtthorn 16 H2O(26)

Daphthorn 10 ferri-Sudthorn 10 Mg-Amthorn 10 Qtz

frac14 6 Clinthorn 15 Mg-Sudthorn 5 Mtthorn 16 H2O(27)

The ferri-Sud properties derived with the assumptionof ChlndashQtzndashMt equilibrium are listed in Table 1(referenced a)With these values the oxygen activity atthe P T and (XFe3+)Chl of all equilibria used tocompute Fig 7 can be calculated with the followingoxygen-bearing and magnetite-free equilibria whichconverge with equilibria 1 8 9 and 10 at the same PTand aO2 conditions

16 Qtzthorn 7 Mg-Amthorn 4 Daphthorn 6 Mg-Sud

thorn 5 O2 thorn 12 H2O frac14 20 ferri-Sud(28)

Fe-Amthorn 2 Mg-Amthorn 6 Qtzthorn 4 H2Othorn O2

frac14 4 ferri-Sud(29)

Mg-Amthorn 4 Daphthorn 6 Mg-Sudthorn 2 O2

frac14 3 Fe-Amthorn 8 ferri-Sudthorn 2 Qtz(30)

5 Mg-Amthorn 12 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 8 Fe-Amthorn 28 ferri-Sud

(31)

8 Daphthorn 12 Mg-Sudthorn 3 O2 frac14 7 Fe-Am

thorn 12 ferri-Sudthorn 10 Qtzthorn 4 H2O(32)

6 Clinthorn 4 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Mg-Amthorn 20 ferri-Sudthorn 5 Qtz(33)

4 Clinthorn 20 ferri-Sud frac14 30 Qtz

thorn 15 Mg-Amthorn 4 Daphthorn 5 O2 thorn 20 H2O(34)

90 Mg-Sudthorn 28 Daphthorn 32 Clinthorn 20 H2Othorn 35 O2

frac14 140 ferri-Sudthorn 15 Mg-Am

(35)

15 Qtzthorn 14 Clinthorn 16 Daphthorn 45 Mg-Sud

thorn 20 O2 thorn 20 H2O frac14 80 ferri-Sud(36)

2 Clinthorn FeAmthorn 3 Mg-Sudthorn O2

frac14 2 Mg-Amthorn 4 ferri-Sudthorn Qtz(37)

12 Clinthorn 7 Fe-Amthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 10 Mg-Amthorn 28 ferri-Sud

(38)

5 Qtzthorn 2 Clinthorn 2 Fe-Amthorn 3 Mg-Sud

thorn 2 O2 thorn 4 H2O frac14 8 ferri-Sud(39)

2 Clinthorn 8 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Fe-Amthorn 20 ferri-Sudthorn 5 Qtz(40)

30 Qtzthorn 8 Clinthorn 15 Fe-Amthorn 5 O2 thorn 20 H2O

frac14 8 Daphthorn 20 ferri-Sud

(41)

4 Clinthorn 8 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 3 Fe-Amthorn 28 ferri-Sud

(42)

Because the ferri-sudoite standard-state propertieswere constrained by the chlorite-quartz-magnetite-O2-H2O equilibrium the Log(aO2) calculated withthe equilibria above follow naturally the trend shown inFig 8 and the Log(aO2) values calculated withreactions 28ndash42 of low-temperature samples plot inthe hematite stability field This suggests that theΔGdegPT values of ferri-sudoite derived above with theassumption that aO2 is buffered by the presence ofmagnetite are possibly too high A reduction of ΔHdegf(ferri-sudoite) to 82304 kJmol (=ΔHdegf (Mg-sudoite)ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2 see Table 1)leads to Log(aO2) reduced by 45 Log units (see Fig 8)

628 O Vidal et al

and most calculated values except those of thehematite-bearing St Martin samples plot in themagnetite stability field

A more detailed study is required to refine the ferri-sudoite thermodynamic data However the resultsobtained using the preliminary values listed in Table 1show that this end-member can be used to reproduceconsistently the T-XFe3+ and T- trends obtained fromclassical Chl-Qtz-H2O calculations (Figs 5ndash7) Theincrease in XFe3+ with reducing temperature is consist-ent with the expected reduction in aO2 and the amountof free oxygen that is consumed to produce Fe3+-bearingchlorite Ferric iron is therefore a major component thathas to be taken into account when dealing with low-temperature samples Empirical or semi-empiricalchlorite thermometers which assume that all Fe is inthe ferrous state may include one of the XFe3+-T andaO2-T trends illustrated in Figs 7 and 8 but they cannotaccount for the different XFe3+-T trends that areexpected in the contexts of different aO2 conditions

FUTURE CHALLENGES LOOK INGAT LOW-GRADE ROCKS SUCH ASMETAMORPH IC ROCKS AND THE

DYNAMICS OF CHLOR ITECRYSTALL IZAT ION AT THE

MICROMETR IC TO NANOMETR ICSCALE

T-P-XFe3+ mapping at the thin-section scaleevidence of chlorite composition re-equilibration in metamorphic and low-gradesamples

Although natural samples offer valuable constraintson thermodynamic models there are two maindifficulties in using them to evaluate the reliability ofexisting chlorite models for thermometric purposes

(1) In the case of low-grade samples referencepressures and temperatures are generally known withina reasonable absolute uncertainty but the size ofchlorite grains is often small and good microprobeanalyses are sometimes difficult to obtain Moreoverthe kinetics of reaction are so sluggish that inheritedchlorite grains might remain unaffected by latervariations in temperature which makes thetemperature-composition relationships difficult toconstrain Lastly extrapolation at HT and HP ofempirical or semi-empirical models constrained byLT samples only may be erroneous in particular theeffect of pressure or eventual systematic variations ofXFe3+ cannot be appraised

(2) In the case of metamorphic samplesmicroprobe analyses are easier to obtain from largechlorite grains but chlorites of contrasted compositionthat crystallized at different temperatures and times ofthe metamorphic history are often observed to coexistmetastably in different microstructural sites of the samethin section In such cases the definition of a singlepair of reference temperature and pressure isinappropriate and cannot be restricted to the peak-temperature conditions

In order to overcome these difficulties and to bettervisualize the link between mineral compositiondeformation and P-T conditions at the thin-sectionscale de Andrade et al (2006 2011) Vidal et al(2006) and Lanari et al (2013) developed an approachwhich combines elementary maps of composition witheither a thermodynamic or a semi-empirical estimate ofchlorite crystallization temperature Maps of rawmicroprobe analyses were standardized using good-quality point analyses of minerals present in themapped area The maps of standardized analyseswere transformed into maps of structural formulaewhere each pixel analysed is converted into individualmineral structural formulae Depending on the thermo-metric approach it was then possible to compute mapsof temperature or T-XFe3+ conditions assuming thateach chlorite pixel was in equilibrium with quartz andwater The results reported by Vidal et al (2006)Yamato et al (2007) Ganne et al (2012) Lanari et al(2013 2014) and Scheffer et al (2015) showed a closelink between deformation and mineral reaction at themicroscopic scale and maps of chlorite temperaturecould be used as a basis for interpreting the micro-structures observed in thin sections of low- to middle-grade metapelites In all these studies the metastablepersistence of different generations of chlorite crystal-lizing at different temperatures during the exhumationof metamorphic samples could be used to constrain theT-deformation history Moreover the map of (XFe3+)Chl calculated by Vidal et al (2006) was in goodqualitative agreement with the in situ mapping of thisratio using the X-ray absorption near-edge structure(XANES) techniques (Fig 9) (see also Munoz et al2006) The calculated temperature and XFe3+ mapsobtained on a single sample showed the same trend ofreducing (XFe3+)Chl with increasing temperature asobserved in Figs 6 and 7

Bourdelle et al (2013c) applied a similar approachfor low-temperature samples with chlorite of smallgrain size using transmission electron microscopy-energy dispersive X-ray (TEM-EDX) spectroscopicanalyses of focused-ion beam (FIB) foils instead of

629Conditions of formation of chlorite

microprobe analyses of thin sections (Fig 10) Thoseauthors were able to observe heterogeneities in XFe3+

at the nm scale using STXM-XANES measurementsThis last study was particularly interesting as it showedthat approaches similar to those used for studyingmetamorphic rocks could be applied to low-grade anddiagenetic samples

Looking at diagenetic rocks like metamorphicrocks

The use of equilibrium thermodynamics to studylow-grade rocks has been often criticized becausesluggish kinetics might preclude achievement of globalequilibrium at low temperature and is responsible forthe crystallization of metastable phases with a lowerenergy of nucleation than stable ones Essene amp Peacor(1995) claimed that the coexistence of chemicallyinhomogeneous phases at a very fine scale in low-grade rocks is indicative of disequilibrium and thatmineral reaction is better described by a kinetic ratherthan thermodynamic approach However the coexist-ence of chemically inhomogeneous phases is also acommon observation in metamorphic rocks wherezoned grains and the coexistence for example ofchlorite grains of different compositions indicate thatthermodynamic equilibrium is approached only locallyand at a given time in the metamorphic history In factthe main difference between diagenetic and higher-temperature rocks is the size of the domains in localequilibrium The more sluggish kinetics of reactionand diffusivity at low temperature lead to small grainsand domains in equilibrium but these domains can beidentified when using appropriate analytical tools andprocedures developed recently (Bourdelle et al2013bc)

FIG 10 Quantitative Fe redox nanomapping on a FIB foilof Glarus sample (T lt 300degC Switzerland) calculatedfrom the 7087ndash71025 eV images ratio (STXM-XANES)

(from Bourdelle et al 2013c)

FIG 11 Analysis of an illite-chlorite assemblage (Gulf Coast sample Alamo 1 well 5825 m) (a) SEM image of thepetrographic thin section showing the emplacement of the FIB foil to be cut across the illitendashchlorite interface (b)Bright-field TEM image of the FIB foil extracted from the section with the platinum strap on top of it (cd) Enlargementof part b showing the Al and Mg contents of chlorite (apfu) and their variation between crystal core and crystal rim

630 O Vidal et al

Figure 10 shows that it is possible to demonstrateheterogeneities in XFe3+ contents at the nanometerscale Consistent variations of other major elements arealso observed with TEM technics on FIB foils Anexample is shown in Fig 11 which depicts thevariations of Al and Mg contents in a chlorite grainfrom the Gulf Coast (Bourdelle et al 2013b) In thiscase the FIB foil was cut parallel to the interfacebetween illite and chlorite and a nm-scale investiga-tion of composition was conducted by TEM-EDXanalysis The rim and core compositions showed azoning of chlorite with Mg- and Al-richer rims thancore compositions This variation in composition on ascale of lt1 μm was related to various crystallizationstages during burial Chlorite sampled at differentdepths and various locations of the Gulf Coast areashowed that zoned compositions similar to thatobserved in Fig 11 are frequent (Bourdelle et al2013b) The AlIV in chlorite rims display a Tdependence that is consistent with the km-scale andregional evolution of temperature ie with thediagenetic history of Gulf Coast (burial P-T condi-tions detritalauthigenic compositions etchellip)(Fig 12) This study suggests again that approachesused for metamorphic rocks can be applied to low-grade rocks when using adapted analytical tools able toidentify sub-μm variations of composition

CONCLUS ION

The variability of chlorite composition reflects thevariability of its conditions of formation Numerousefforts have been made over the last 30 years to

translate the composition of chlorite in terms of itstemperature of formation The first empirical equationsbased on the AlIVor vacancy contents are restricted tobulk-rock compositions and pressure conditions atwhich they were calibrated Thermodynamicapproaches and semi-empirical equations based onthe ratio of end-member activities do not present thislimit Among the different thermodynamic models ofchlorite published so far Vidal et al (2006) proposed amodel that can be used to estimate T and (XFe3+)Chlsimultaneously The estimated (XFe3+)Chl values areconsistent with those measured by Moumlssbauer orXANES spectroscopy and for the same (XFe3+)Chlthe temperatures estimated using the method of Vidalet al (2006) are similar to those obtained with thesemi-empirical equations of Inoue et al (2009) andLanari et al (2014) Different T-XFe3+ trends areobserved for different pressure gradients and oxygen-buffering assemblages (Lanari et al 2014) butoverall the systematic increase in XFe3+ with areduction in temperature is compatible with thedecrease in aO2 in a closed system The increase inXFe3+ and chlorite vacancy with decreasing tempera-ture is interpreted as reflecting the incorporation of theFe3+-sudoite component which can be used toestimate T-aO2-XFe

3+ values with a Chl-Qtz-H2Omulti-equilibrium approach

Similar temperatures are calculated for low-tem-perature chlorites (T lt 300degC and measured XFe3+ upto 30) with all approaches assuming that iron inchlorite is either completely ferrous or with estimatedvalues of XFe3+ This indicates that the equations ofsemi-empirical thermometers assuming that all iron inchlorite is ferrous include some inherent Fe2+Fe3+ vstemperature correction However very different tem-peratures can be obtained for higher-temperaturesamples (gt300degC) with approaches involving ferriciron or assuming that all of the iron is ferrous Thissuggests that the Fe2+Fe3+ vs temperature correctionmight be restricted to particular aO2 conditions andpressure gradients More work is necessary to betterunderstand this point and to improve the thermo-dynamic models and semi-empirical thermometersThis requires high-quality and well constrained naturaldata To this end mapping techniques that highlight thelink between the variations in temperature and pressureand small-scale chlorite composition including XFe3+

variations offer much promise When combining theresults of the present study with those of other studiespublished recently it appears that thermodynamicapproaches and mapping techniques developed formetamorphic rocks could be used to study low-grade

FIG 12 Range of crystal-rim AlIV content (white boxes)of Gulf Coast chlorites vs measured temperatures Thegrey area underlines the range of maximum AlIV contentwhich is assumed to represent the closest approach toequilibrium composition (after Bourdelle et al 2013b)

631Conditions of formation of chlorite

rocks where kinetics are much more sluggish Thisrequires use of appropriate analytical tools and techni-ques with a spatial resolution of just a few hundrednanometres which are now becoming available

ACKNOWLEDGMENTS

Constructive reviews by S Hillier F Nieto and ananonymous reviewer are gratefully acknowledged

REFERENCES

Beaufort D Patrier P Meunier A amp Ottaviani MM(1992) Chemical variations in assemblages includingepidote andor chlorite in the fossil hydrothermalsystem of Saint Martin (Lesser Antilles) Journal ofVolcanology and Geothermal Research 51 95ndash114

Berman R (1988) Internally-consistent thermodynamicdata for minerals in the system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2Journal of Petrology 29 445ndash522

Berman R amp Brown T (1985) Heat capacity of minerals inthe system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2 representation estima-tion and high temperature extrapolation Contributionsto Mineralogy and Petrology 89 168ndash183

Bourdelle F amp Cathelineau M (2015) Low-temperaturechlorite geothermometry a graphical representationbased on a TndashR2+ndashSi diagram European Journal ofMineralogy 27 617ndash626

Bourdelle F Parra T Chopin C amp Beyssac O (2013a) Anew chlorite geothermometer for diagenetic to low-grade metamorphic conditions Contributions toMineralogy and Petrology 165 723ndash735

Bourdelle F Parra T Beyssac O Chopin C amp Vidal O(2013b) Clay minerals thermometry a comparativestudy based on high-resolution chemical analyses ofillite and chlorite in sandstones from the Gulf Coast(Texas USA) American Mineralogist 98 914ndash926

Bourdelle F Benzerara K Beyssac O Cosmidis JNeuville DR Brown GE amp Paineau E (2013c)Quantification of the ferric ferrous iron ratio insilicates by scanning transmission X-ray microscopyat the Fe L23 edges Contribution to Mineralogy andPetrology 166 423ndash434

Cathelineau M (1988) Cation site occupancy in chloritesand illites as function of temperature Clay Minerals23 471ndash485

Cathelineau M amp Nieva D (1985) A chlorite solidsolution geothermometer the Los Azufres (Mexico)geothermal system Contributions to Mineralogy andPetrology 91 235ndash244

Chermak JA amp Rimstidt JD (1989) Estimating thethermodynamic properties (ΔGdegf and ΔHdegf) of silicate

minerals at 298K from the sum of polyhedral contribu-tions American Mineralogist 74 1023ndash1031

Coggon R amp Holland TJB (2002) Mixing properties ofphengitic micas and revised garnent-phengite thermo-barometers Journal of Metamorphic Geology 20683ndash696

Curtis CD Hughes CR Whiteman JA ampWhittle CK(1985) Compositional variation within some sedi-mentary chlorites and some comments on their originMineralogical Magazine 49 375ndash386

de Andrade V Vidal O Lewin E OrsquoBrien P amp Agard P(2006) Quantification of electron microprobe com-positional maps of rock thin sections an optimizedmethod and examples Journal of MetamorphicGeology 24 655ndash668

de Andrade V Susini J Salomeacute M Beraldin O RigaultC Heymes T Lewin E amp Vidal O (2011)Submicrometer hyperspectral X-ray imaging of het-erogeneous rocks and geomaterials applications at theFe k-edge Analytical Chemistry 83 4220ndash4227

de Caritat P Hutcheon I amp Walshe J (1993) Chloritegeothermometry a review Clays and Clay Minerals41 219ndash239

Essene EJ (2009) Thermobarometry gone astray Pp101ndash129 in Physics and Chemistry of Earthrsquos Interior(AK Gupta amp S Dasgupta editors) PlatinumJubilee Indian National Science Academy Springer-Verlag Berlin

Essene EJ amp Peacor DR (1995) Clay mineralthermometry ndash a critical perspective Clays and ClayMinerals 43 540ndash553

Ganne J de Andrade V Weinberg RF Vidal O DubacqB Kagambega N Naba S Baratoux L Jessell M ampAllibon J (2012) Modern-style plate subductionpreserved in the Palaeoproterozoic West Africancraton Nature Geoscience 5 60ndash65

Grosch EG Vidal O Abu-Alam T amp McLoughlin N(2012) PTconstraints on the metamorphic evolution ofthe Paleoarchean Kromberg type-section BarbertonGreenstone Belt South-Africa Journal of Petrology53 513ndash545

Grosch E McLoughlin N Lanari P Erambert M ampVidal O (2014) Microscale mapping of alterationconditions and potential biosignatures in basaltic-ultramafic rocks on early Earth and beyondAstrobiology 14 216ndash228

Hillier S amp Velde B (1991) Octahedral occupancy andchemical composition of diagenetic (low-temperature)chlorites Clay Minerals 26 149ndash168

Holland TJB (1989) Dependance of entropy on volumefor silicates and oxide minerals A review and apredictive model American Mineralogist 74 5ndash13

Holland T amp Powell R (1998) An internally consistentthermodynamic data set for phases of petrologicalinterest Journal ofMetamorphic Geology 16 309ndash343

Holland T Baker J amp Powell R (1998)Mixing propertiesand activity composition and relationships of chlorites

632 O Vidal et al

Brown (1985) The formation enthalpy (ΔHdegf = 82254 plusmn 5 kJmol) was first estimated as the average valuebetween that calculated according to Chermak ampRimstidt (1989) (minus82204 kJmol) and that(minus82304 kJmol) calculted as ΔHdegf (Mg-sudoite) ndashΔHdegf (corundum)2 + ΔHdegf (hematite)2 using thevalues of Vidal et al (1992) for sudoite and Berman(1988) for hematite

(3) In a first step ΔHdegf and Sdeg of ferri-sudoite wereadjusted in order to reproduce the temperaturesreported in Fig 7 with oxygen-free reactions andassuming that magnetite is present in all samples(Chl-Qtz-Mt assemblage) ΔHdegf was allowed to varybetweenminus8220 andminus8230 kJmol and Sdeg within 10 JmolK of the estimated value The addition ofmagnetite and ferri-sudoite to quartz and water leadsto 17 additional equilibria and the following wereused to constrain ΔHdegf and Sdeg of ferri-sudoite

ClinthornMg-Sud frac14 Mg-Amthorn Qtzthorn H2O (1)

4 Clinthorn Daphthorn 10 ferri-Sud

frac14 10 Mg-Amthorn 25 Qtzthorn 30 Mtthorn 20 H2O(24)

7 Daphthorn 70 ferri-Sudthorn 30 Mg-Am

frac14 22 Clinthorn 75 Mg-Sudthorn 35 Mtthorn 40 H2O(25)

28 ferri-Sudthorn 11 Fe-AmthornMg-Am

frac14 6 Daphthorn 30 Mg-Sudthorn 14 Mtthorn 16 H2O(26)

Daphthorn 10 ferri-Sudthorn 10 Mg-Amthorn 10 Qtz

frac14 6 Clinthorn 15 Mg-Sudthorn 5 Mtthorn 16 H2O(27)

The ferri-Sud properties derived with the assumptionof ChlndashQtzndashMt equilibrium are listed in Table 1(referenced a)With these values the oxygen activity atthe P T and (XFe3+)Chl of all equilibria used tocompute Fig 7 can be calculated with the followingoxygen-bearing and magnetite-free equilibria whichconverge with equilibria 1 8 9 and 10 at the same PTand aO2 conditions

16 Qtzthorn 7 Mg-Amthorn 4 Daphthorn 6 Mg-Sud

thorn 5 O2 thorn 12 H2O frac14 20 ferri-Sud(28)

Fe-Amthorn 2 Mg-Amthorn 6 Qtzthorn 4 H2Othorn O2

frac14 4 ferri-Sud(29)

Mg-Amthorn 4 Daphthorn 6 Mg-Sudthorn 2 O2

frac14 3 Fe-Amthorn 8 ferri-Sudthorn 2 Qtz(30)

5 Mg-Amthorn 12 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 8 Fe-Amthorn 28 ferri-Sud

(31)

8 Daphthorn 12 Mg-Sudthorn 3 O2 frac14 7 Fe-Am

thorn 12 ferri-Sudthorn 10 Qtzthorn 4 H2O(32)

6 Clinthorn 4 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Mg-Amthorn 20 ferri-Sudthorn 5 Qtz(33)

4 Clinthorn 20 ferri-Sud frac14 30 Qtz

thorn 15 Mg-Amthorn 4 Daphthorn 5 O2 thorn 20 H2O(34)

90 Mg-Sudthorn 28 Daphthorn 32 Clinthorn 20 H2Othorn 35 O2

frac14 140 ferri-Sudthorn 15 Mg-Am

(35)

15 Qtzthorn 14 Clinthorn 16 Daphthorn 45 Mg-Sud

thorn 20 O2 thorn 20 H2O frac14 80 ferri-Sud(36)

2 Clinthorn FeAmthorn 3 Mg-Sudthorn O2

frac14 2 Mg-Amthorn 4 ferri-Sudthorn Qtz(37)

12 Clinthorn 7 Fe-Amthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 10 Mg-Amthorn 28 ferri-Sud

(38)

5 Qtzthorn 2 Clinthorn 2 Fe-Amthorn 3 Mg-Sud

thorn 2 O2 thorn 4 H2O frac14 8 ferri-Sud(39)

2 Clinthorn 8 Daphthorn 15 Mg-Sudthorn 5 O2

frac14 5 Fe-Amthorn 20 ferri-Sudthorn 5 Qtz(40)

30 Qtzthorn 8 Clinthorn 15 Fe-Amthorn 5 O2 thorn 20 H2O

frac14 8 Daphthorn 20 ferri-Sud

(41)

4 Clinthorn 8 Daphthorn 18 Mg-Sudthorn 7 O2 thorn 4 H2O

frac14 3 Fe-Amthorn 28 ferri-Sud

(42)

Because the ferri-sudoite standard-state propertieswere constrained by the chlorite-quartz-magnetite-O2-H2O equilibrium the Log(aO2) calculated withthe equilibria above follow naturally the trend shown inFig 8 and the Log(aO2) values calculated withreactions 28ndash42 of low-temperature samples plot inthe hematite stability field This suggests that theΔGdegPT values of ferri-sudoite derived above with theassumption that aO2 is buffered by the presence ofmagnetite are possibly too high A reduction of ΔHdegf(ferri-sudoite) to 82304 kJmol (=ΔHdegf (Mg-sudoite)ndash ΔHdegf (corundum)2 + ΔHdegf (hematite)2 see Table 1)leads to Log(aO2) reduced by 45 Log units (see Fig 8)

628 O Vidal et al

and most calculated values except those of thehematite-bearing St Martin samples plot in themagnetite stability field

A more detailed study is required to refine the ferri-sudoite thermodynamic data However the resultsobtained using the preliminary values listed in Table 1show that this end-member can be used to reproduceconsistently the T-XFe3+ and T- trends obtained fromclassical Chl-Qtz-H2O calculations (Figs 5ndash7) Theincrease in XFe3+ with reducing temperature is consist-ent with the expected reduction in aO2 and the amountof free oxygen that is consumed to produce Fe3+-bearingchlorite Ferric iron is therefore a major component thathas to be taken into account when dealing with low-temperature samples Empirical or semi-empiricalchlorite thermometers which assume that all Fe is inthe ferrous state may include one of the XFe3+-T andaO2-T trends illustrated in Figs 7 and 8 but they cannotaccount for the different XFe3+-T trends that areexpected in the contexts of different aO2 conditions

FUTURE CHALLENGES LOOK INGAT LOW-GRADE ROCKS SUCH ASMETAMORPH IC ROCKS AND THE

DYNAMICS OF CHLOR ITECRYSTALL IZAT ION AT THE

MICROMETR IC TO NANOMETR ICSCALE

T-P-XFe3+ mapping at the thin-section scaleevidence of chlorite composition re-equilibration in metamorphic and low-gradesamples

Although natural samples offer valuable constraintson thermodynamic models there are two maindifficulties in using them to evaluate the reliability ofexisting chlorite models for thermometric purposes

(1) In the case of low-grade samples referencepressures and temperatures are generally known withina reasonable absolute uncertainty but the size ofchlorite grains is often small and good microprobeanalyses are sometimes difficult to obtain Moreoverthe kinetics of reaction are so sluggish that inheritedchlorite grains might remain unaffected by latervariations in temperature which makes thetemperature-composition relationships difficult toconstrain Lastly extrapolation at HT and HP ofempirical or semi-empirical models constrained byLT samples only may be erroneous in particular theeffect of pressure or eventual systematic variations ofXFe3+ cannot be appraised

(2) In the case of metamorphic samplesmicroprobe analyses are easier to obtain from largechlorite grains but chlorites of contrasted compositionthat crystallized at different temperatures and times ofthe metamorphic history are often observed to coexistmetastably in different microstructural sites of the samethin section In such cases the definition of a singlepair of reference temperature and pressure isinappropriate and cannot be restricted to the peak-temperature conditions

In order to overcome these difficulties and to bettervisualize the link between mineral compositiondeformation and P-T conditions at the thin-sectionscale de Andrade et al (2006 2011) Vidal et al(2006) and Lanari et al (2013) developed an approachwhich combines elementary maps of composition witheither a thermodynamic or a semi-empirical estimate ofchlorite crystallization temperature Maps of rawmicroprobe analyses were standardized using good-quality point analyses of minerals present in themapped area The maps of standardized analyseswere transformed into maps of structural formulaewhere each pixel analysed is converted into individualmineral structural formulae Depending on the thermo-metric approach it was then possible to compute mapsof temperature or T-XFe3+ conditions assuming thateach chlorite pixel was in equilibrium with quartz andwater The results reported by Vidal et al (2006)Yamato et al (2007) Ganne et al (2012) Lanari et al(2013 2014) and Scheffer et al (2015) showed a closelink between deformation and mineral reaction at themicroscopic scale and maps of chlorite temperaturecould be used as a basis for interpreting the micro-structures observed in thin sections of low- to middle-grade metapelites In all these studies the metastablepersistence of different generations of chlorite crystal-lizing at different temperatures during the exhumationof metamorphic samples could be used to constrain theT-deformation history Moreover the map of (XFe3+)Chl calculated by Vidal et al (2006) was in goodqualitative agreement with the in situ mapping of thisratio using the X-ray absorption near-edge structure(XANES) techniques (Fig 9) (see also Munoz et al2006) The calculated temperature and XFe3+ mapsobtained on a single sample showed the same trend ofreducing (XFe3+)Chl with increasing temperature asobserved in Figs 6 and 7

Bourdelle et al (2013c) applied a similar approachfor low-temperature samples with chlorite of smallgrain size using transmission electron microscopy-energy dispersive X-ray (TEM-EDX) spectroscopicanalyses of focused-ion beam (FIB) foils instead of

629Conditions of formation of chlorite

microprobe analyses of thin sections (Fig 10) Thoseauthors were able to observe heterogeneities in XFe3+

at the nm scale using STXM-XANES measurementsThis last study was particularly interesting as it showedthat approaches similar to those used for studyingmetamorphic rocks could be applied to low-grade anddiagenetic samples

Looking at diagenetic rocks like metamorphicrocks

The use of equilibrium thermodynamics to studylow-grade rocks has been often criticized becausesluggish kinetics might preclude achievement of globalequilibrium at low temperature and is responsible forthe crystallization of metastable phases with a lowerenergy of nucleation than stable ones Essene amp Peacor(1995) claimed that the coexistence of chemicallyinhomogeneous phases at a very fine scale in low-grade rocks is indicative of disequilibrium and thatmineral reaction is better described by a kinetic ratherthan thermodynamic approach However the coexist-ence of chemically inhomogeneous phases is also acommon observation in metamorphic rocks wherezoned grains and the coexistence for example ofchlorite grains of different compositions indicate thatthermodynamic equilibrium is approached only locallyand at a given time in the metamorphic history In factthe main difference between diagenetic and higher-temperature rocks is the size of the domains in localequilibrium The more sluggish kinetics of reactionand diffusivity at low temperature lead to small grainsand domains in equilibrium but these domains can beidentified when using appropriate analytical tools andprocedures developed recently (Bourdelle et al2013bc)

FIG 10 Quantitative Fe redox nanomapping on a FIB foilof Glarus sample (T lt 300degC Switzerland) calculatedfrom the 7087ndash71025 eV images ratio (STXM-XANES)

(from Bourdelle et al 2013c)

FIG 11 Analysis of an illite-chlorite assemblage (Gulf Coast sample Alamo 1 well 5825 m) (a) SEM image of thepetrographic thin section showing the emplacement of the FIB foil to be cut across the illitendashchlorite interface (b)Bright-field TEM image of the FIB foil extracted from the section with the platinum strap on top of it (cd) Enlargementof part b showing the Al and Mg contents of chlorite (apfu) and their variation between crystal core and crystal rim

630 O Vidal et al

Figure 10 shows that it is possible to demonstrateheterogeneities in XFe3+ contents at the nanometerscale Consistent variations of other major elements arealso observed with TEM technics on FIB foils Anexample is shown in Fig 11 which depicts thevariations of Al and Mg contents in a chlorite grainfrom the Gulf Coast (Bourdelle et al 2013b) In thiscase the FIB foil was cut parallel to the interfacebetween illite and chlorite and a nm-scale investiga-tion of composition was conducted by TEM-EDXanalysis The rim and core compositions showed azoning of chlorite with Mg- and Al-richer rims thancore compositions This variation in composition on ascale of lt1 μm was related to various crystallizationstages during burial Chlorite sampled at differentdepths and various locations of the Gulf Coast areashowed that zoned compositions similar to thatobserved in Fig 11 are frequent (Bourdelle et al2013b) The AlIV in chlorite rims display a Tdependence that is consistent with the km-scale andregional evolution of temperature ie with thediagenetic history of Gulf Coast (burial P-T condi-tions detritalauthigenic compositions etchellip)(Fig 12) This study suggests again that approachesused for metamorphic rocks can be applied to low-grade rocks when using adapted analytical tools able toidentify sub-μm variations of composition

CONCLUS ION

The variability of chlorite composition reflects thevariability of its conditions of formation Numerousefforts have been made over the last 30 years to

translate the composition of chlorite in terms of itstemperature of formation The first empirical equationsbased on the AlIVor vacancy contents are restricted tobulk-rock compositions and pressure conditions atwhich they were calibrated Thermodynamicapproaches and semi-empirical equations based onthe ratio of end-member activities do not present thislimit Among the different thermodynamic models ofchlorite published so far Vidal et al (2006) proposed amodel that can be used to estimate T and (XFe3+)Chlsimultaneously The estimated (XFe3+)Chl values areconsistent with those measured by Moumlssbauer orXANES spectroscopy and for the same (XFe3+)Chlthe temperatures estimated using the method of Vidalet al (2006) are similar to those obtained with thesemi-empirical equations of Inoue et al (2009) andLanari et al (2014) Different T-XFe3+ trends areobserved for different pressure gradients and oxygen-buffering assemblages (Lanari et al 2014) butoverall the systematic increase in XFe3+ with areduction in temperature is compatible with thedecrease in aO2 in a closed system The increase inXFe3+ and chlorite vacancy with decreasing tempera-ture is interpreted as reflecting the incorporation of theFe3+-sudoite component which can be used toestimate T-aO2-XFe

3+ values with a Chl-Qtz-H2Omulti-equilibrium approach

Similar temperatures are calculated for low-tem-perature chlorites (T lt 300degC and measured XFe3+ upto 30) with all approaches assuming that iron inchlorite is either completely ferrous or with estimatedvalues of XFe3+ This indicates that the equations ofsemi-empirical thermometers assuming that all iron inchlorite is ferrous include some inherent Fe2+Fe3+ vstemperature correction However very different tem-peratures can be obtained for higher-temperaturesamples (gt300degC) with approaches involving ferriciron or assuming that all of the iron is ferrous Thissuggests that the Fe2+Fe3+ vs temperature correctionmight be restricted to particular aO2 conditions andpressure gradients More work is necessary to betterunderstand this point and to improve the thermo-dynamic models and semi-empirical thermometersThis requires high-quality and well constrained naturaldata To this end mapping techniques that highlight thelink between the variations in temperature and pressureand small-scale chlorite composition including XFe3+

variations offer much promise When combining theresults of the present study with those of other studiespublished recently it appears that thermodynamicapproaches and mapping techniques developed formetamorphic rocks could be used to study low-grade

FIG 12 Range of crystal-rim AlIV content (white boxes)of Gulf Coast chlorites vs measured temperatures Thegrey area underlines the range of maximum AlIV contentwhich is assumed to represent the closest approach toequilibrium composition (after Bourdelle et al 2013b)

631Conditions of formation of chlorite

rocks where kinetics are much more sluggish Thisrequires use of appropriate analytical tools and techni-ques with a spatial resolution of just a few hundrednanometres which are now becoming available

ACKNOWLEDGMENTS

Constructive reviews by S Hillier F Nieto and ananonymous reviewer are gratefully acknowledged

REFERENCES

Beaufort D Patrier P Meunier A amp Ottaviani MM(1992) Chemical variations in assemblages includingepidote andor chlorite in the fossil hydrothermalsystem of Saint Martin (Lesser Antilles) Journal ofVolcanology and Geothermal Research 51 95ndash114

Berman R (1988) Internally-consistent thermodynamicdata for minerals in the system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2Journal of Petrology 29 445ndash522

Berman R amp Brown T (1985) Heat capacity of minerals inthe system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2 representation estima-tion and high temperature extrapolation Contributionsto Mineralogy and Petrology 89 168ndash183

Bourdelle F amp Cathelineau M (2015) Low-temperaturechlorite geothermometry a graphical representationbased on a TndashR2+ndashSi diagram European Journal ofMineralogy 27 617ndash626

Bourdelle F Parra T Chopin C amp Beyssac O (2013a) Anew chlorite geothermometer for diagenetic to low-grade metamorphic conditions Contributions toMineralogy and Petrology 165 723ndash735

Bourdelle F Parra T Beyssac O Chopin C amp Vidal O(2013b) Clay minerals thermometry a comparativestudy based on high-resolution chemical analyses ofillite and chlorite in sandstones from the Gulf Coast(Texas USA) American Mineralogist 98 914ndash926

Bourdelle F Benzerara K Beyssac O Cosmidis JNeuville DR Brown GE amp Paineau E (2013c)Quantification of the ferric ferrous iron ratio insilicates by scanning transmission X-ray microscopyat the Fe L23 edges Contribution to Mineralogy andPetrology 166 423ndash434

Cathelineau M (1988) Cation site occupancy in chloritesand illites as function of temperature Clay Minerals23 471ndash485

Cathelineau M amp Nieva D (1985) A chlorite solidsolution geothermometer the Los Azufres (Mexico)geothermal system Contributions to Mineralogy andPetrology 91 235ndash244

Chermak JA amp Rimstidt JD (1989) Estimating thethermodynamic properties (ΔGdegf and ΔHdegf) of silicate

minerals at 298K from the sum of polyhedral contribu-tions American Mineralogist 74 1023ndash1031

Coggon R amp Holland TJB (2002) Mixing properties ofphengitic micas and revised garnent-phengite thermo-barometers Journal of Metamorphic Geology 20683ndash696

Curtis CD Hughes CR Whiteman JA ampWhittle CK(1985) Compositional variation within some sedi-mentary chlorites and some comments on their originMineralogical Magazine 49 375ndash386

de Andrade V Vidal O Lewin E OrsquoBrien P amp Agard P(2006) Quantification of electron microprobe com-positional maps of rock thin sections an optimizedmethod and examples Journal of MetamorphicGeology 24 655ndash668

de Andrade V Susini J Salomeacute M Beraldin O RigaultC Heymes T Lewin E amp Vidal O (2011)Submicrometer hyperspectral X-ray imaging of het-erogeneous rocks and geomaterials applications at theFe k-edge Analytical Chemistry 83 4220ndash4227

de Caritat P Hutcheon I amp Walshe J (1993) Chloritegeothermometry a review Clays and Clay Minerals41 219ndash239

Essene EJ (2009) Thermobarometry gone astray Pp101ndash129 in Physics and Chemistry of Earthrsquos Interior(AK Gupta amp S Dasgupta editors) PlatinumJubilee Indian National Science Academy Springer-Verlag Berlin

Essene EJ amp Peacor DR (1995) Clay mineralthermometry ndash a critical perspective Clays and ClayMinerals 43 540ndash553

Ganne J de Andrade V Weinberg RF Vidal O DubacqB Kagambega N Naba S Baratoux L Jessell M ampAllibon J (2012) Modern-style plate subductionpreserved in the Palaeoproterozoic West Africancraton Nature Geoscience 5 60ndash65

Grosch EG Vidal O Abu-Alam T amp McLoughlin N(2012) PTconstraints on the metamorphic evolution ofthe Paleoarchean Kromberg type-section BarbertonGreenstone Belt South-Africa Journal of Petrology53 513ndash545

Grosch E McLoughlin N Lanari P Erambert M ampVidal O (2014) Microscale mapping of alterationconditions and potential biosignatures in basaltic-ultramafic rocks on early Earth and beyondAstrobiology 14 216ndash228

Hillier S amp Velde B (1991) Octahedral occupancy andchemical composition of diagenetic (low-temperature)chlorites Clay Minerals 26 149ndash168

Holland TJB (1989) Dependance of entropy on volumefor silicates and oxide minerals A review and apredictive model American Mineralogist 74 5ndash13

Holland T amp Powell R (1998) An internally consistentthermodynamic data set for phases of petrologicalinterest Journal ofMetamorphic Geology 16 309ndash343

Holland T Baker J amp Powell R (1998)Mixing propertiesand activity composition and relationships of chlorites

632 O Vidal et al

and most calculated values except those of thehematite-bearing St Martin samples plot in themagnetite stability field

A more detailed study is required to refine the ferri-sudoite thermodynamic data However the resultsobtained using the preliminary values listed in Table 1show that this end-member can be used to reproduceconsistently the T-XFe3+ and T- trends obtained fromclassical Chl-Qtz-H2O calculations (Figs 5ndash7) Theincrease in XFe3+ with reducing temperature is consist-ent with the expected reduction in aO2 and the amountof free oxygen that is consumed to produce Fe3+-bearingchlorite Ferric iron is therefore a major component thathas to be taken into account when dealing with low-temperature samples Empirical or semi-empiricalchlorite thermometers which assume that all Fe is inthe ferrous state may include one of the XFe3+-T andaO2-T trends illustrated in Figs 7 and 8 but they cannotaccount for the different XFe3+-T trends that areexpected in the contexts of different aO2 conditions

FUTURE CHALLENGES LOOK INGAT LOW-GRADE ROCKS SUCH ASMETAMORPH IC ROCKS AND THE

DYNAMICS OF CHLOR ITECRYSTALL IZAT ION AT THE

MICROMETR IC TO NANOMETR ICSCALE

T-P-XFe3+ mapping at the thin-section scaleevidence of chlorite composition re-equilibration in metamorphic and low-gradesamples

Although natural samples offer valuable constraintson thermodynamic models there are two maindifficulties in using them to evaluate the reliability ofexisting chlorite models for thermometric purposes

(1) In the case of low-grade samples referencepressures and temperatures are generally known withina reasonable absolute uncertainty but the size ofchlorite grains is often small and good microprobeanalyses are sometimes difficult to obtain Moreoverthe kinetics of reaction are so sluggish that inheritedchlorite grains might remain unaffected by latervariations in temperature which makes thetemperature-composition relationships difficult toconstrain Lastly extrapolation at HT and HP ofempirical or semi-empirical models constrained byLT samples only may be erroneous in particular theeffect of pressure or eventual systematic variations ofXFe3+ cannot be appraised

(2) In the case of metamorphic samplesmicroprobe analyses are easier to obtain from largechlorite grains but chlorites of contrasted compositionthat crystallized at different temperatures and times ofthe metamorphic history are often observed to coexistmetastably in different microstructural sites of the samethin section In such cases the definition of a singlepair of reference temperature and pressure isinappropriate and cannot be restricted to the peak-temperature conditions

In order to overcome these difficulties and to bettervisualize the link between mineral compositiondeformation and P-T conditions at the thin-sectionscale de Andrade et al (2006 2011) Vidal et al(2006) and Lanari et al (2013) developed an approachwhich combines elementary maps of composition witheither a thermodynamic or a semi-empirical estimate ofchlorite crystallization temperature Maps of rawmicroprobe analyses were standardized using good-quality point analyses of minerals present in themapped area The maps of standardized analyseswere transformed into maps of structural formulaewhere each pixel analysed is converted into individualmineral structural formulae Depending on the thermo-metric approach it was then possible to compute mapsof temperature or T-XFe3+ conditions assuming thateach chlorite pixel was in equilibrium with quartz andwater The results reported by Vidal et al (2006)Yamato et al (2007) Ganne et al (2012) Lanari et al(2013 2014) and Scheffer et al (2015) showed a closelink between deformation and mineral reaction at themicroscopic scale and maps of chlorite temperaturecould be used as a basis for interpreting the micro-structures observed in thin sections of low- to middle-grade metapelites In all these studies the metastablepersistence of different generations of chlorite crystal-lizing at different temperatures during the exhumationof metamorphic samples could be used to constrain theT-deformation history Moreover the map of (XFe3+)Chl calculated by Vidal et al (2006) was in goodqualitative agreement with the in situ mapping of thisratio using the X-ray absorption near-edge structure(XANES) techniques (Fig 9) (see also Munoz et al2006) The calculated temperature and XFe3+ mapsobtained on a single sample showed the same trend ofreducing (XFe3+)Chl with increasing temperature asobserved in Figs 6 and 7

Bourdelle et al (2013c) applied a similar approachfor low-temperature samples with chlorite of smallgrain size using transmission electron microscopy-energy dispersive X-ray (TEM-EDX) spectroscopicanalyses of focused-ion beam (FIB) foils instead of

629Conditions of formation of chlorite

microprobe analyses of thin sections (Fig 10) Thoseauthors were able to observe heterogeneities in XFe3+

at the nm scale using STXM-XANES measurementsThis last study was particularly interesting as it showedthat approaches similar to those used for studyingmetamorphic rocks could be applied to low-grade anddiagenetic samples

Looking at diagenetic rocks like metamorphicrocks

The use of equilibrium thermodynamics to studylow-grade rocks has been often criticized becausesluggish kinetics might preclude achievement of globalequilibrium at low temperature and is responsible forthe crystallization of metastable phases with a lowerenergy of nucleation than stable ones Essene amp Peacor(1995) claimed that the coexistence of chemicallyinhomogeneous phases at a very fine scale in low-grade rocks is indicative of disequilibrium and thatmineral reaction is better described by a kinetic ratherthan thermodynamic approach However the coexist-ence of chemically inhomogeneous phases is also acommon observation in metamorphic rocks wherezoned grains and the coexistence for example ofchlorite grains of different compositions indicate thatthermodynamic equilibrium is approached only locallyand at a given time in the metamorphic history In factthe main difference between diagenetic and higher-temperature rocks is the size of the domains in localequilibrium The more sluggish kinetics of reactionand diffusivity at low temperature lead to small grainsand domains in equilibrium but these domains can beidentified when using appropriate analytical tools andprocedures developed recently (Bourdelle et al2013bc)

FIG 10 Quantitative Fe redox nanomapping on a FIB foilof Glarus sample (T lt 300degC Switzerland) calculatedfrom the 7087ndash71025 eV images ratio (STXM-XANES)

(from Bourdelle et al 2013c)

FIG 11 Analysis of an illite-chlorite assemblage (Gulf Coast sample Alamo 1 well 5825 m) (a) SEM image of thepetrographic thin section showing the emplacement of the FIB foil to be cut across the illitendashchlorite interface (b)Bright-field TEM image of the FIB foil extracted from the section with the platinum strap on top of it (cd) Enlargementof part b showing the Al and Mg contents of chlorite (apfu) and their variation between crystal core and crystal rim

630 O Vidal et al

Figure 10 shows that it is possible to demonstrateheterogeneities in XFe3+ contents at the nanometerscale Consistent variations of other major elements arealso observed with TEM technics on FIB foils Anexample is shown in Fig 11 which depicts thevariations of Al and Mg contents in a chlorite grainfrom the Gulf Coast (Bourdelle et al 2013b) In thiscase the FIB foil was cut parallel to the interfacebetween illite and chlorite and a nm-scale investiga-tion of composition was conducted by TEM-EDXanalysis The rim and core compositions showed azoning of chlorite with Mg- and Al-richer rims thancore compositions This variation in composition on ascale of lt1 μm was related to various crystallizationstages during burial Chlorite sampled at differentdepths and various locations of the Gulf Coast areashowed that zoned compositions similar to thatobserved in Fig 11 are frequent (Bourdelle et al2013b) The AlIV in chlorite rims display a Tdependence that is consistent with the km-scale andregional evolution of temperature ie with thediagenetic history of Gulf Coast (burial P-T condi-tions detritalauthigenic compositions etchellip)(Fig 12) This study suggests again that approachesused for metamorphic rocks can be applied to low-grade rocks when using adapted analytical tools able toidentify sub-μm variations of composition

CONCLUS ION

The variability of chlorite composition reflects thevariability of its conditions of formation Numerousefforts have been made over the last 30 years to

translate the composition of chlorite in terms of itstemperature of formation The first empirical equationsbased on the AlIVor vacancy contents are restricted tobulk-rock compositions and pressure conditions atwhich they were calibrated Thermodynamicapproaches and semi-empirical equations based onthe ratio of end-member activities do not present thislimit Among the different thermodynamic models ofchlorite published so far Vidal et al (2006) proposed amodel that can be used to estimate T and (XFe3+)Chlsimultaneously The estimated (XFe3+)Chl values areconsistent with those measured by Moumlssbauer orXANES spectroscopy and for the same (XFe3+)Chlthe temperatures estimated using the method of Vidalet al (2006) are similar to those obtained with thesemi-empirical equations of Inoue et al (2009) andLanari et al (2014) Different T-XFe3+ trends areobserved for different pressure gradients and oxygen-buffering assemblages (Lanari et al 2014) butoverall the systematic increase in XFe3+ with areduction in temperature is compatible with thedecrease in aO2 in a closed system The increase inXFe3+ and chlorite vacancy with decreasing tempera-ture is interpreted as reflecting the incorporation of theFe3+-sudoite component which can be used toestimate T-aO2-XFe

3+ values with a Chl-Qtz-H2Omulti-equilibrium approach

Similar temperatures are calculated for low-tem-perature chlorites (T lt 300degC and measured XFe3+ upto 30) with all approaches assuming that iron inchlorite is either completely ferrous or with estimatedvalues of XFe3+ This indicates that the equations ofsemi-empirical thermometers assuming that all iron inchlorite is ferrous include some inherent Fe2+Fe3+ vstemperature correction However very different tem-peratures can be obtained for higher-temperaturesamples (gt300degC) with approaches involving ferriciron or assuming that all of the iron is ferrous Thissuggests that the Fe2+Fe3+ vs temperature correctionmight be restricted to particular aO2 conditions andpressure gradients More work is necessary to betterunderstand this point and to improve the thermo-dynamic models and semi-empirical thermometersThis requires high-quality and well constrained naturaldata To this end mapping techniques that highlight thelink between the variations in temperature and pressureand small-scale chlorite composition including XFe3+

variations offer much promise When combining theresults of the present study with those of other studiespublished recently it appears that thermodynamicapproaches and mapping techniques developed formetamorphic rocks could be used to study low-grade

FIG 12 Range of crystal-rim AlIV content (white boxes)of Gulf Coast chlorites vs measured temperatures Thegrey area underlines the range of maximum AlIV contentwhich is assumed to represent the closest approach toequilibrium composition (after Bourdelle et al 2013b)

631Conditions of formation of chlorite

rocks where kinetics are much more sluggish Thisrequires use of appropriate analytical tools and techni-ques with a spatial resolution of just a few hundrednanometres which are now becoming available

ACKNOWLEDGMENTS

Constructive reviews by S Hillier F Nieto and ananonymous reviewer are gratefully acknowledged

REFERENCES

Beaufort D Patrier P Meunier A amp Ottaviani MM(1992) Chemical variations in assemblages includingepidote andor chlorite in the fossil hydrothermalsystem of Saint Martin (Lesser Antilles) Journal ofVolcanology and Geothermal Research 51 95ndash114

Berman R (1988) Internally-consistent thermodynamicdata for minerals in the system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2Journal of Petrology 29 445ndash522

Berman R amp Brown T (1985) Heat capacity of minerals inthe system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2 representation estima-tion and high temperature extrapolation Contributionsto Mineralogy and Petrology 89 168ndash183

Bourdelle F amp Cathelineau M (2015) Low-temperaturechlorite geothermometry a graphical representationbased on a TndashR2+ndashSi diagram European Journal ofMineralogy 27 617ndash626

Bourdelle F Parra T Chopin C amp Beyssac O (2013a) Anew chlorite geothermometer for diagenetic to low-grade metamorphic conditions Contributions toMineralogy and Petrology 165 723ndash735

Bourdelle F Parra T Beyssac O Chopin C amp Vidal O(2013b) Clay minerals thermometry a comparativestudy based on high-resolution chemical analyses ofillite and chlorite in sandstones from the Gulf Coast(Texas USA) American Mineralogist 98 914ndash926

Bourdelle F Benzerara K Beyssac O Cosmidis JNeuville DR Brown GE amp Paineau E (2013c)Quantification of the ferric ferrous iron ratio insilicates by scanning transmission X-ray microscopyat the Fe L23 edges Contribution to Mineralogy andPetrology 166 423ndash434

Cathelineau M (1988) Cation site occupancy in chloritesand illites as function of temperature Clay Minerals23 471ndash485

Cathelineau M amp Nieva D (1985) A chlorite solidsolution geothermometer the Los Azufres (Mexico)geothermal system Contributions to Mineralogy andPetrology 91 235ndash244

Chermak JA amp Rimstidt JD (1989) Estimating thethermodynamic properties (ΔGdegf and ΔHdegf) of silicate

minerals at 298K from the sum of polyhedral contribu-tions American Mineralogist 74 1023ndash1031

Coggon R amp Holland TJB (2002) Mixing properties ofphengitic micas and revised garnent-phengite thermo-barometers Journal of Metamorphic Geology 20683ndash696

Curtis CD Hughes CR Whiteman JA ampWhittle CK(1985) Compositional variation within some sedi-mentary chlorites and some comments on their originMineralogical Magazine 49 375ndash386

de Andrade V Vidal O Lewin E OrsquoBrien P amp Agard P(2006) Quantification of electron microprobe com-positional maps of rock thin sections an optimizedmethod and examples Journal of MetamorphicGeology 24 655ndash668

de Andrade V Susini J Salomeacute M Beraldin O RigaultC Heymes T Lewin E amp Vidal O (2011)Submicrometer hyperspectral X-ray imaging of het-erogeneous rocks and geomaterials applications at theFe k-edge Analytical Chemistry 83 4220ndash4227

de Caritat P Hutcheon I amp Walshe J (1993) Chloritegeothermometry a review Clays and Clay Minerals41 219ndash239

Essene EJ (2009) Thermobarometry gone astray Pp101ndash129 in Physics and Chemistry of Earthrsquos Interior(AK Gupta amp S Dasgupta editors) PlatinumJubilee Indian National Science Academy Springer-Verlag Berlin

Essene EJ amp Peacor DR (1995) Clay mineralthermometry ndash a critical perspective Clays and ClayMinerals 43 540ndash553

Ganne J de Andrade V Weinberg RF Vidal O DubacqB Kagambega N Naba S Baratoux L Jessell M ampAllibon J (2012) Modern-style plate subductionpreserved in the Palaeoproterozoic West Africancraton Nature Geoscience 5 60ndash65

Grosch EG Vidal O Abu-Alam T amp McLoughlin N(2012) PTconstraints on the metamorphic evolution ofthe Paleoarchean Kromberg type-section BarbertonGreenstone Belt South-Africa Journal of Petrology53 513ndash545

Grosch E McLoughlin N Lanari P Erambert M ampVidal O (2014) Microscale mapping of alterationconditions and potential biosignatures in basaltic-ultramafic rocks on early Earth and beyondAstrobiology 14 216ndash228

Hillier S amp Velde B (1991) Octahedral occupancy andchemical composition of diagenetic (low-temperature)chlorites Clay Minerals 26 149ndash168

Holland TJB (1989) Dependance of entropy on volumefor silicates and oxide minerals A review and apredictive model American Mineralogist 74 5ndash13

Holland T amp Powell R (1998) An internally consistentthermodynamic data set for phases of petrologicalinterest Journal ofMetamorphic Geology 16 309ndash343

Holland T Baker J amp Powell R (1998)Mixing propertiesand activity composition and relationships of chlorites

632 O Vidal et al

microprobe analyses of thin sections (Fig 10) Thoseauthors were able to observe heterogeneities in XFe3+

at the nm scale using STXM-XANES measurementsThis last study was particularly interesting as it showedthat approaches similar to those used for studyingmetamorphic rocks could be applied to low-grade anddiagenetic samples

Looking at diagenetic rocks like metamorphicrocks

The use of equilibrium thermodynamics to studylow-grade rocks has been often criticized becausesluggish kinetics might preclude achievement of globalequilibrium at low temperature and is responsible forthe crystallization of metastable phases with a lowerenergy of nucleation than stable ones Essene amp Peacor(1995) claimed that the coexistence of chemicallyinhomogeneous phases at a very fine scale in low-grade rocks is indicative of disequilibrium and thatmineral reaction is better described by a kinetic ratherthan thermodynamic approach However the coexist-ence of chemically inhomogeneous phases is also acommon observation in metamorphic rocks wherezoned grains and the coexistence for example ofchlorite grains of different compositions indicate thatthermodynamic equilibrium is approached only locallyand at a given time in the metamorphic history In factthe main difference between diagenetic and higher-temperature rocks is the size of the domains in localequilibrium The more sluggish kinetics of reactionand diffusivity at low temperature lead to small grainsand domains in equilibrium but these domains can beidentified when using appropriate analytical tools andprocedures developed recently (Bourdelle et al2013bc)

FIG 10 Quantitative Fe redox nanomapping on a FIB foilof Glarus sample (T lt 300degC Switzerland) calculatedfrom the 7087ndash71025 eV images ratio (STXM-XANES)

(from Bourdelle et al 2013c)

FIG 11 Analysis of an illite-chlorite assemblage (Gulf Coast sample Alamo 1 well 5825 m) (a) SEM image of thepetrographic thin section showing the emplacement of the FIB foil to be cut across the illitendashchlorite interface (b)Bright-field TEM image of the FIB foil extracted from the section with the platinum strap on top of it (cd) Enlargementof part b showing the Al and Mg contents of chlorite (apfu) and their variation between crystal core and crystal rim

630 O Vidal et al

Figure 10 shows that it is possible to demonstrateheterogeneities in XFe3+ contents at the nanometerscale Consistent variations of other major elements arealso observed with TEM technics on FIB foils Anexample is shown in Fig 11 which depicts thevariations of Al and Mg contents in a chlorite grainfrom the Gulf Coast (Bourdelle et al 2013b) In thiscase the FIB foil was cut parallel to the interfacebetween illite and chlorite and a nm-scale investiga-tion of composition was conducted by TEM-EDXanalysis The rim and core compositions showed azoning of chlorite with Mg- and Al-richer rims thancore compositions This variation in composition on ascale of lt1 μm was related to various crystallizationstages during burial Chlorite sampled at differentdepths and various locations of the Gulf Coast areashowed that zoned compositions similar to thatobserved in Fig 11 are frequent (Bourdelle et al2013b) The AlIV in chlorite rims display a Tdependence that is consistent with the km-scale andregional evolution of temperature ie with thediagenetic history of Gulf Coast (burial P-T condi-tions detritalauthigenic compositions etchellip)(Fig 12) This study suggests again that approachesused for metamorphic rocks can be applied to low-grade rocks when using adapted analytical tools able toidentify sub-μm variations of composition

CONCLUS ION

The variability of chlorite composition reflects thevariability of its conditions of formation Numerousefforts have been made over the last 30 years to

translate the composition of chlorite in terms of itstemperature of formation The first empirical equationsbased on the AlIVor vacancy contents are restricted tobulk-rock compositions and pressure conditions atwhich they were calibrated Thermodynamicapproaches and semi-empirical equations based onthe ratio of end-member activities do not present thislimit Among the different thermodynamic models ofchlorite published so far Vidal et al (2006) proposed amodel that can be used to estimate T and (XFe3+)Chlsimultaneously The estimated (XFe3+)Chl values areconsistent with those measured by Moumlssbauer orXANES spectroscopy and for the same (XFe3+)Chlthe temperatures estimated using the method of Vidalet al (2006) are similar to those obtained with thesemi-empirical equations of Inoue et al (2009) andLanari et al (2014) Different T-XFe3+ trends areobserved for different pressure gradients and oxygen-buffering assemblages (Lanari et al 2014) butoverall the systematic increase in XFe3+ with areduction in temperature is compatible with thedecrease in aO2 in a closed system The increase inXFe3+ and chlorite vacancy with decreasing tempera-ture is interpreted as reflecting the incorporation of theFe3+-sudoite component which can be used toestimate T-aO2-XFe

3+ values with a Chl-Qtz-H2Omulti-equilibrium approach

Similar temperatures are calculated for low-tem-perature chlorites (T lt 300degC and measured XFe3+ upto 30) with all approaches assuming that iron inchlorite is either completely ferrous or with estimatedvalues of XFe3+ This indicates that the equations ofsemi-empirical thermometers assuming that all iron inchlorite is ferrous include some inherent Fe2+Fe3+ vstemperature correction However very different tem-peratures can be obtained for higher-temperaturesamples (gt300degC) with approaches involving ferriciron or assuming that all of the iron is ferrous Thissuggests that the Fe2+Fe3+ vs temperature correctionmight be restricted to particular aO2 conditions andpressure gradients More work is necessary to betterunderstand this point and to improve the thermo-dynamic models and semi-empirical thermometersThis requires high-quality and well constrained naturaldata To this end mapping techniques that highlight thelink between the variations in temperature and pressureand small-scale chlorite composition including XFe3+

variations offer much promise When combining theresults of the present study with those of other studiespublished recently it appears that thermodynamicapproaches and mapping techniques developed formetamorphic rocks could be used to study low-grade

FIG 12 Range of crystal-rim AlIV content (white boxes)of Gulf Coast chlorites vs measured temperatures Thegrey area underlines the range of maximum AlIV contentwhich is assumed to represent the closest approach toequilibrium composition (after Bourdelle et al 2013b)

631Conditions of formation of chlorite

rocks where kinetics are much more sluggish Thisrequires use of appropriate analytical tools and techni-ques with a spatial resolution of just a few hundrednanometres which are now becoming available

ACKNOWLEDGMENTS

Constructive reviews by S Hillier F Nieto and ananonymous reviewer are gratefully acknowledged

REFERENCES

Beaufort D Patrier P Meunier A amp Ottaviani MM(1992) Chemical variations in assemblages includingepidote andor chlorite in the fossil hydrothermalsystem of Saint Martin (Lesser Antilles) Journal ofVolcanology and Geothermal Research 51 95ndash114

Berman R (1988) Internally-consistent thermodynamicdata for minerals in the system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2Journal of Petrology 29 445ndash522

Berman R amp Brown T (1985) Heat capacity of minerals inthe system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2 representation estima-tion and high temperature extrapolation Contributionsto Mineralogy and Petrology 89 168ndash183

Bourdelle F amp Cathelineau M (2015) Low-temperaturechlorite geothermometry a graphical representationbased on a TndashR2+ndashSi diagram European Journal ofMineralogy 27 617ndash626

Bourdelle F Parra T Chopin C amp Beyssac O (2013a) Anew chlorite geothermometer for diagenetic to low-grade metamorphic conditions Contributions toMineralogy and Petrology 165 723ndash735

Bourdelle F Parra T Beyssac O Chopin C amp Vidal O(2013b) Clay minerals thermometry a comparativestudy based on high-resolution chemical analyses ofillite and chlorite in sandstones from the Gulf Coast(Texas USA) American Mineralogist 98 914ndash926

Bourdelle F Benzerara K Beyssac O Cosmidis JNeuville DR Brown GE amp Paineau E (2013c)Quantification of the ferric ferrous iron ratio insilicates by scanning transmission X-ray microscopyat the Fe L23 edges Contribution to Mineralogy andPetrology 166 423ndash434

Cathelineau M (1988) Cation site occupancy in chloritesand illites as function of temperature Clay Minerals23 471ndash485

Cathelineau M amp Nieva D (1985) A chlorite solidsolution geothermometer the Los Azufres (Mexico)geothermal system Contributions to Mineralogy andPetrology 91 235ndash244

Chermak JA amp Rimstidt JD (1989) Estimating thethermodynamic properties (ΔGdegf and ΔHdegf) of silicate

minerals at 298K from the sum of polyhedral contribu-tions American Mineralogist 74 1023ndash1031

Coggon R amp Holland TJB (2002) Mixing properties ofphengitic micas and revised garnent-phengite thermo-barometers Journal of Metamorphic Geology 20683ndash696

Curtis CD Hughes CR Whiteman JA ampWhittle CK(1985) Compositional variation within some sedi-mentary chlorites and some comments on their originMineralogical Magazine 49 375ndash386

de Andrade V Vidal O Lewin E OrsquoBrien P amp Agard P(2006) Quantification of electron microprobe com-positional maps of rock thin sections an optimizedmethod and examples Journal of MetamorphicGeology 24 655ndash668

de Andrade V Susini J Salomeacute M Beraldin O RigaultC Heymes T Lewin E amp Vidal O (2011)Submicrometer hyperspectral X-ray imaging of het-erogeneous rocks and geomaterials applications at theFe k-edge Analytical Chemistry 83 4220ndash4227

de Caritat P Hutcheon I amp Walshe J (1993) Chloritegeothermometry a review Clays and Clay Minerals41 219ndash239

Essene EJ (2009) Thermobarometry gone astray Pp101ndash129 in Physics and Chemistry of Earthrsquos Interior(AK Gupta amp S Dasgupta editors) PlatinumJubilee Indian National Science Academy Springer-Verlag Berlin

Essene EJ amp Peacor DR (1995) Clay mineralthermometry ndash a critical perspective Clays and ClayMinerals 43 540ndash553

Ganne J de Andrade V Weinberg RF Vidal O DubacqB Kagambega N Naba S Baratoux L Jessell M ampAllibon J (2012) Modern-style plate subductionpreserved in the Palaeoproterozoic West Africancraton Nature Geoscience 5 60ndash65

Grosch EG Vidal O Abu-Alam T amp McLoughlin N(2012) PTconstraints on the metamorphic evolution ofthe Paleoarchean Kromberg type-section BarbertonGreenstone Belt South-Africa Journal of Petrology53 513ndash545

Grosch E McLoughlin N Lanari P Erambert M ampVidal O (2014) Microscale mapping of alterationconditions and potential biosignatures in basaltic-ultramafic rocks on early Earth and beyondAstrobiology 14 216ndash228

Hillier S amp Velde B (1991) Octahedral occupancy andchemical composition of diagenetic (low-temperature)chlorites Clay Minerals 26 149ndash168

Holland TJB (1989) Dependance of entropy on volumefor silicates and oxide minerals A review and apredictive model American Mineralogist 74 5ndash13

Holland T amp Powell R (1998) An internally consistentthermodynamic data set for phases of petrologicalinterest Journal ofMetamorphic Geology 16 309ndash343

Holland T Baker J amp Powell R (1998)Mixing propertiesand activity composition and relationships of chlorites

632 O Vidal et al

Figure 10 shows that it is possible to demonstrateheterogeneities in XFe3+ contents at the nanometerscale Consistent variations of other major elements arealso observed with TEM technics on FIB foils Anexample is shown in Fig 11 which depicts thevariations of Al and Mg contents in a chlorite grainfrom the Gulf Coast (Bourdelle et al 2013b) In thiscase the FIB foil was cut parallel to the interfacebetween illite and chlorite and a nm-scale investiga-tion of composition was conducted by TEM-EDXanalysis The rim and core compositions showed azoning of chlorite with Mg- and Al-richer rims thancore compositions This variation in composition on ascale of lt1 μm was related to various crystallizationstages during burial Chlorite sampled at differentdepths and various locations of the Gulf Coast areashowed that zoned compositions similar to thatobserved in Fig 11 are frequent (Bourdelle et al2013b) The AlIV in chlorite rims display a Tdependence that is consistent with the km-scale andregional evolution of temperature ie with thediagenetic history of Gulf Coast (burial P-T condi-tions detritalauthigenic compositions etchellip)(Fig 12) This study suggests again that approachesused for metamorphic rocks can be applied to low-grade rocks when using adapted analytical tools able toidentify sub-μm variations of composition

CONCLUS ION

The variability of chlorite composition reflects thevariability of its conditions of formation Numerousefforts have been made over the last 30 years to

translate the composition of chlorite in terms of itstemperature of formation The first empirical equationsbased on the AlIVor vacancy contents are restricted tobulk-rock compositions and pressure conditions atwhich they were calibrated Thermodynamicapproaches and semi-empirical equations based onthe ratio of end-member activities do not present thislimit Among the different thermodynamic models ofchlorite published so far Vidal et al (2006) proposed amodel that can be used to estimate T and (XFe3+)Chlsimultaneously The estimated (XFe3+)Chl values areconsistent with those measured by Moumlssbauer orXANES spectroscopy and for the same (XFe3+)Chlthe temperatures estimated using the method of Vidalet al (2006) are similar to those obtained with thesemi-empirical equations of Inoue et al (2009) andLanari et al (2014) Different T-XFe3+ trends areobserved for different pressure gradients and oxygen-buffering assemblages (Lanari et al 2014) butoverall the systematic increase in XFe3+ with areduction in temperature is compatible with thedecrease in aO2 in a closed system The increase inXFe3+ and chlorite vacancy with decreasing tempera-ture is interpreted as reflecting the incorporation of theFe3+-sudoite component which can be used toestimate T-aO2-XFe

3+ values with a Chl-Qtz-H2Omulti-equilibrium approach

Similar temperatures are calculated for low-tem-perature chlorites (T lt 300degC and measured XFe3+ upto 30) with all approaches assuming that iron inchlorite is either completely ferrous or with estimatedvalues of XFe3+ This indicates that the equations ofsemi-empirical thermometers assuming that all iron inchlorite is ferrous include some inherent Fe2+Fe3+ vstemperature correction However very different tem-peratures can be obtained for higher-temperaturesamples (gt300degC) with approaches involving ferriciron or assuming that all of the iron is ferrous Thissuggests that the Fe2+Fe3+ vs temperature correctionmight be restricted to particular aO2 conditions andpressure gradients More work is necessary to betterunderstand this point and to improve the thermo-dynamic models and semi-empirical thermometersThis requires high-quality and well constrained naturaldata To this end mapping techniques that highlight thelink between the variations in temperature and pressureand small-scale chlorite composition including XFe3+

variations offer much promise When combining theresults of the present study with those of other studiespublished recently it appears that thermodynamicapproaches and mapping techniques developed formetamorphic rocks could be used to study low-grade

FIG 12 Range of crystal-rim AlIV content (white boxes)of Gulf Coast chlorites vs measured temperatures Thegrey area underlines the range of maximum AlIV contentwhich is assumed to represent the closest approach toequilibrium composition (after Bourdelle et al 2013b)

631Conditions of formation of chlorite

rocks where kinetics are much more sluggish Thisrequires use of appropriate analytical tools and techni-ques with a spatial resolution of just a few hundrednanometres which are now becoming available

ACKNOWLEDGMENTS

Constructive reviews by S Hillier F Nieto and ananonymous reviewer are gratefully acknowledged

REFERENCES

Beaufort D Patrier P Meunier A amp Ottaviani MM(1992) Chemical variations in assemblages includingepidote andor chlorite in the fossil hydrothermalsystem of Saint Martin (Lesser Antilles) Journal ofVolcanology and Geothermal Research 51 95ndash114

Berman R (1988) Internally-consistent thermodynamicdata for minerals in the system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2Journal of Petrology 29 445ndash522

Berman R amp Brown T (1985) Heat capacity of minerals inthe system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2 representation estima-tion and high temperature extrapolation Contributionsto Mineralogy and Petrology 89 168ndash183

Bourdelle F amp Cathelineau M (2015) Low-temperaturechlorite geothermometry a graphical representationbased on a TndashR2+ndashSi diagram European Journal ofMineralogy 27 617ndash626

Bourdelle F Parra T Chopin C amp Beyssac O (2013a) Anew chlorite geothermometer for diagenetic to low-grade metamorphic conditions Contributions toMineralogy and Petrology 165 723ndash735

Bourdelle F Parra T Beyssac O Chopin C amp Vidal O(2013b) Clay minerals thermometry a comparativestudy based on high-resolution chemical analyses ofillite and chlorite in sandstones from the Gulf Coast(Texas USA) American Mineralogist 98 914ndash926

Bourdelle F Benzerara K Beyssac O Cosmidis JNeuville DR Brown GE amp Paineau E (2013c)Quantification of the ferric ferrous iron ratio insilicates by scanning transmission X-ray microscopyat the Fe L23 edges Contribution to Mineralogy andPetrology 166 423ndash434

Cathelineau M (1988) Cation site occupancy in chloritesand illites as function of temperature Clay Minerals23 471ndash485

Cathelineau M amp Nieva D (1985) A chlorite solidsolution geothermometer the Los Azufres (Mexico)geothermal system Contributions to Mineralogy andPetrology 91 235ndash244

Chermak JA amp Rimstidt JD (1989) Estimating thethermodynamic properties (ΔGdegf and ΔHdegf) of silicate

minerals at 298K from the sum of polyhedral contribu-tions American Mineralogist 74 1023ndash1031

Coggon R amp Holland TJB (2002) Mixing properties ofphengitic micas and revised garnent-phengite thermo-barometers Journal of Metamorphic Geology 20683ndash696

Curtis CD Hughes CR Whiteman JA ampWhittle CK(1985) Compositional variation within some sedi-mentary chlorites and some comments on their originMineralogical Magazine 49 375ndash386

de Andrade V Vidal O Lewin E OrsquoBrien P amp Agard P(2006) Quantification of electron microprobe com-positional maps of rock thin sections an optimizedmethod and examples Journal of MetamorphicGeology 24 655ndash668

de Andrade V Susini J Salomeacute M Beraldin O RigaultC Heymes T Lewin E amp Vidal O (2011)Submicrometer hyperspectral X-ray imaging of het-erogeneous rocks and geomaterials applications at theFe k-edge Analytical Chemistry 83 4220ndash4227

de Caritat P Hutcheon I amp Walshe J (1993) Chloritegeothermometry a review Clays and Clay Minerals41 219ndash239

Essene EJ (2009) Thermobarometry gone astray Pp101ndash129 in Physics and Chemistry of Earthrsquos Interior(AK Gupta amp S Dasgupta editors) PlatinumJubilee Indian National Science Academy Springer-Verlag Berlin

Essene EJ amp Peacor DR (1995) Clay mineralthermometry ndash a critical perspective Clays and ClayMinerals 43 540ndash553

Ganne J de Andrade V Weinberg RF Vidal O DubacqB Kagambega N Naba S Baratoux L Jessell M ampAllibon J (2012) Modern-style plate subductionpreserved in the Palaeoproterozoic West Africancraton Nature Geoscience 5 60ndash65

Grosch EG Vidal O Abu-Alam T amp McLoughlin N(2012) PTconstraints on the metamorphic evolution ofthe Paleoarchean Kromberg type-section BarbertonGreenstone Belt South-Africa Journal of Petrology53 513ndash545

Grosch E McLoughlin N Lanari P Erambert M ampVidal O (2014) Microscale mapping of alterationconditions and potential biosignatures in basaltic-ultramafic rocks on early Earth and beyondAstrobiology 14 216ndash228

Hillier S amp Velde B (1991) Octahedral occupancy andchemical composition of diagenetic (low-temperature)chlorites Clay Minerals 26 149ndash168

Holland TJB (1989) Dependance of entropy on volumefor silicates and oxide minerals A review and apredictive model American Mineralogist 74 5ndash13

Holland T amp Powell R (1998) An internally consistentthermodynamic data set for phases of petrologicalinterest Journal ofMetamorphic Geology 16 309ndash343

Holland T Baker J amp Powell R (1998)Mixing propertiesand activity composition and relationships of chlorites

632 O Vidal et al

rocks where kinetics are much more sluggish Thisrequires use of appropriate analytical tools and techni-ques with a spatial resolution of just a few hundrednanometres which are now becoming available

ACKNOWLEDGMENTS

Constructive reviews by S Hillier F Nieto and ananonymous reviewer are gratefully acknowledged

REFERENCES

Beaufort D Patrier P Meunier A amp Ottaviani MM(1992) Chemical variations in assemblages includingepidote andor chlorite in the fossil hydrothermalsystem of Saint Martin (Lesser Antilles) Journal ofVolcanology and Geothermal Research 51 95ndash114

Berman R (1988) Internally-consistent thermodynamicdata for minerals in the system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2Journal of Petrology 29 445ndash522

Berman R amp Brown T (1985) Heat capacity of minerals inthe system Na2OndashK2OndashCaOndashMgOndashFeOndashFe2O3ndashAl2O3ndashSiO2ndashTiO2ndashH2OndashCO2 representation estima-tion and high temperature extrapolation Contributionsto Mineralogy and Petrology 89 168ndash183

Bourdelle F amp Cathelineau M (2015) Low-temperaturechlorite geothermometry a graphical representationbased on a TndashR2+ndashSi diagram European Journal ofMineralogy 27 617ndash626

Bourdelle F Parra T Chopin C amp Beyssac O (2013a) Anew chlorite geothermometer for diagenetic to low-grade metamorphic conditions Contributions toMineralogy and Petrology 165 723ndash735

Bourdelle F Parra T Beyssac O Chopin C amp Vidal O(2013b) Clay minerals thermometry a comparativestudy based on high-resolution chemical analyses ofillite and chlorite in sandstones from the Gulf Coast(Texas USA) American Mineralogist 98 914ndash926

Bourdelle F Benzerara K Beyssac O Cosmidis JNeuville DR Brown GE amp Paineau E (2013c)Quantification of the ferric ferrous iron ratio insilicates by scanning transmission X-ray microscopyat the Fe L23 edges Contribution to Mineralogy andPetrology 166 423ndash434

Cathelineau M (1988) Cation site occupancy in chloritesand illites as function of temperature Clay Minerals23 471ndash485

Cathelineau M amp Nieva D (1985) A chlorite solidsolution geothermometer the Los Azufres (Mexico)geothermal system Contributions to Mineralogy andPetrology 91 235ndash244

Chermak JA amp Rimstidt JD (1989) Estimating thethermodynamic properties (ΔGdegf and ΔHdegf) of silicate

minerals at 298K from the sum of polyhedral contribu-tions American Mineralogist 74 1023ndash1031

Coggon R amp Holland TJB (2002) Mixing properties ofphengitic micas and revised garnent-phengite thermo-barometers Journal of Metamorphic Geology 20683ndash696

Curtis CD Hughes CR Whiteman JA ampWhittle CK(1985) Compositional variation within some sedi-mentary chlorites and some comments on their originMineralogical Magazine 49 375ndash386

de Andrade V Vidal O Lewin E OrsquoBrien P amp Agard P(2006) Quantification of electron microprobe com-positional maps of rock thin sections an optimizedmethod and examples Journal of MetamorphicGeology 24 655ndash668

de Andrade V Susini J Salomeacute M Beraldin O RigaultC Heymes T Lewin E amp Vidal O (2011)Submicrometer hyperspectral X-ray imaging of het-erogeneous rocks and geomaterials applications at theFe k-edge Analytical Chemistry 83 4220ndash4227

de Caritat P Hutcheon I amp Walshe J (1993) Chloritegeothermometry a review Clays and Clay Minerals41 219ndash239

Essene EJ (2009) Thermobarometry gone astray Pp101ndash129 in Physics and Chemistry of Earthrsquos Interior(AK Gupta amp S Dasgupta editors) PlatinumJubilee Indian National Science Academy Springer-Verlag Berlin

Essene EJ amp Peacor DR (1995) Clay mineralthermometry ndash a critical perspective Clays and ClayMinerals 43 540ndash553

Ganne J de Andrade V Weinberg RF Vidal O DubacqB Kagambega N Naba S Baratoux L Jessell M ampAllibon J (2012) Modern-style plate subductionpreserved in the Palaeoproterozoic West Africancraton Nature Geoscience 5 60ndash65

Grosch EG Vidal O Abu-Alam T amp McLoughlin N(2012) PTconstraints on the metamorphic evolution ofthe Paleoarchean Kromberg type-section BarbertonGreenstone Belt South-Africa Journal of Petrology53 513ndash545

Grosch E McLoughlin N Lanari P Erambert M ampVidal O (2014) Microscale mapping of alterationconditions and potential biosignatures in basaltic-ultramafic rocks on early Earth and beyondAstrobiology 14 216ndash228

Hillier S amp Velde B (1991) Octahedral occupancy andchemical composition of diagenetic (low-temperature)chlorites Clay Minerals 26 149ndash168

Holland TJB (1989) Dependance of entropy on volumefor silicates and oxide minerals A review and apredictive model American Mineralogist 74 5ndash13

Holland T amp Powell R (1998) An internally consistentthermodynamic data set for phases of petrologicalinterest Journal ofMetamorphic Geology 16 309ndash343

Holland T Baker J amp Powell R (1998)Mixing propertiesand activity composition and relationships of chlorites

632 O Vidal et al

in the system MgOndashFeOndashAl2O3ndashSiO2ndashH2OEuropean Journal of Mineralogy 10 395ndash406

Inoue A Meunier A Patrier-Mas P Rigault C BeaufortD amp Vieillard P (2009) Application of chemicalgeothermometry to low-temperature trioctahedralchlorites Clays and Clay Minerals 57 371ndash382

Jowett E (1991) Fitting iron and magnesium into thehydrothermal chlorite geothermometer ProgramAbstract 16 A62

Kranidiotis P amp MacLean W (1987) Systematics ofchlorite alteration at the Phelps Dodge massive sulfidedeposit Matagami Quebec Economic Geology 821898ndash1911

Laird J (1988) Chlorites Metamorphic petrology Pp405ndash453 in Hydrous Phyllosilicates Exclusive ofMicas (SW Bailey editor) Reviews in Mineralogy19 Mineralogical Society of America Washington DC

Lanari P Guillot S Schwartz S Vidal O Tricart P RielN amp Beyssac O (2012) Diachronous evolution of thealpine continental subduction wedge evidence fromP-T estimates in the Brianccedilonnais Zone houillere(FrancemdashWestern Alps) Journal of Geodynamics 5639ndash54

Lanari P Rolland Y Schwartz S Vidal O Guillot STricart P amp Dumont T (2013) P-T-t estimation of syn-kinematic strain in low-grade quartz-feldspar-bearingrocks using thermodynamic modeling and 40Ar39Ardating techniques example of the Plan de Phasy shearzone unit (Brianccedilonnais Zone Western Alps) TerraNova 26 130ndash138

Lanari P Wagner T amp Vidal O (2014) A thermodynamicmodel for di-trioctahedral chlorite from experimentaland natural data in the system MgOndashFeOndashAl2O3ndashSiO2ndashH2O applications to PndashT sections and geother-mometry Contributions to Mineralogy and Petrology167 1ndash19

Munoz M De Andrade V Vidal O Lewin E PascarelliS amp Susini J (2006) Redox and speciation micro-mapping using dispersive X-ray absorption spectros-copy application to iron in chlorite mineral of ametamorphic rock thin section GeochemistryGeophysics Geosystems 7 Q11020

MunozM Vidal O Marcaillou C Pascarelli O MathonO amp Farges F (2013) Iron oxidation state inphyllosilicate single crystals using Fe-K pre-edgeand XANES spectroscopy effects of the linearpolarization of the synchrotron X-ray beamAmerican Mineralogy 98 1187ndash1197

Parra T Vidal O amp Theye T (2005) Experimental data onthe Tschermak substitution in Fe-chlorite AmericanMineralogist 90 359ndash370

Powell R amp Holland T (1999) Relating formulations ofthe thermodynamics of mineral solid solutionsactivity modeling of pyroxenes amphiboles andmicas American Mineralogist 84 1ndash14

Scheffer C Vanderhaeghe O Lanari P Tarantola APonthus L Photiades A amp France L (2015) Syn topost-orogenic exhumation of high-grade nappesstructure and thermobarometry of the western Attic-Cycladic metamorphic complex (Lavrion Greece)Journal of Geodynamics 51 doi101016jjog201508005

Trincal V amp Lanari P (2016) Al-free di-trioctahedralsubstitution in chlorite and a ferri-sudoite end-memberClay Minerals 51 675ndash689

Trincal V Lanari P Buatier M Lacroix B CharpentierD Labaume P amp Munoz M (2015) Temperaturemicro-mapping in oscillatory zoned chloriteApplication to the study of a green-schist facies faultzone near Gavarnie (Pyrenees Spain) AmericanMineralogist 100 2468ndash2483

Vidal O Goffeacute B amp Theye T (1992) Experimental studyof the stability of sudoite and magnesiocarpholite andcalculation of a new petrogenetic grid for the systemFeOndashMgOndashAl2O3ndashSiO2ndashH2O Journal ofMetamorphic Geology 10 603ndash614

Vidal O Parra T amp Trotet F (2001) A thermodynamicmodel for FendashMg aluminous chlorite using datafrom phase equilibrium experiments and naturalpelitic assemblages in the 100 to 600 C 1 to25 kbar range American Journal of Science 301557ndash592

Vidal O Parra T amp Vieillard P (2005) Thermodynamicproperties of theTschermak solid solution in Fe-chloriteapplication to natural examples and possible role ofoxidation American Mineralogist 90 347ndash358

Vidal O de Andrade V Lewin E Muntildeoz M Parra T ampPascarelli S (2006) P-T-deformation-Fe3+Fe2+ mappingat the thin section scale and comparison with XANESmapping application to a garnet-bearing metapelitefrom the Sambagawametamorphic belt (Japan) Journalof Metamorphic Geology 24 669ndash683

Walshe JL (1986) A six-component chlorite solidsolution model and the conditions of chlorite forma-tion in hydrothermal and geothermal systemsEconomic Geology 81 681ndash703

Wiewioacutera A amp Weiss Z (1990) Crystallochemicalclassifications of phyllosilicates based on the unifiedsystem of projection of chemical composition II Thechlorite group Clay Minerals 25 83ndash92

Yamato P Agard P Burov E Le Pourhiet L Jolivet L ampTiberi C (2007) Burial and exhumation in asubduction wedge mutual constraints from thermo-mechanical modeling and natural P-T-t data (SchistesLustreacutes western Alps) Journal of GeophysicalResearch 112 B07410

Zang W amp Fyfe W (1995) Chloritization of thehydrothermally altered bedrock at the Igarape Bahiagold deposit Carajas Brazil Mineralium Deposita30 30ndash38

633Conditions of formation of chlorite