ultrahigh-pressuremetamorphism ... · frontiers ultrahigh-pressuremetamorphism:...

Frontiers

Ultrahigh-pressure metamorphism:tracing continental crust into the mantle

Christian ChopinLaboratoire de Ge¤ologie, UMR 8538 du CNRS, Ecole normale supe¤rieure, 24 rue Lhomond, 75005 Paris, France

Received 16 December 2002; received in revised form 6 May 2003; accepted 6 May 2003

Abstract

More and more evidence is being discovered in Phanerozoic collision belts of the burial of crustal rocks topreviously unsuspected (and ever increasing) depths, presently on the order of 150^200 km, and of exhumation fromsuch depths. This extends by almost one order of magnitude the depth classically ascribed to the metamorphic cyclingof continental crust, and demonstrates its possible subduction. The pieces of evidence for this new, ultrahigh-pressure(UHP) metamorphism exclusively occur in the form of relics of high-pressure minerals that escaped back-transformation during decompression. The main UHP mineral indicators are the high-pressure polymorphs of silicaand carbon, coesite and microdiamond, respectively; the latter often demonstrably precipitated from a metamorphicfluid and is completely unrelated to kimberlitic diamond or any shock event. Recent discoveries of pyroxeneexsolutions in garnet and of coesite exsolutions in titanite suggest a precursor garnet or titanite containing six-foldcoordinated silicon, therefore still higher pressures than implied by diamond stability, on the order of 6 GPa. TheUHP rocks raise a formidable geological problem: that of the mechanisms responsible for their burial and, morepressingly, for their exhumation from the relevant depths. The petrological record indicates that large tracts of UHProcks were buried to conditions of low T/P ratio, consistent with a subduction-zone context. Decompression occurredin most instances under continuous cooling, implying continuous heat loss to the footwall and hangingwall of therising body. This rise along the subduction channel ^ an obvious mechanical discontinuity and weak zone ^ may bedriven by buoyancy up to mid-crustal levels as a result of the lesser density of the acidic crustal rocks (even ifcompletely re-equilibrated at depth) after delamination from the lower crust, in a convergent setting. Chronologicalstudies suggest that the rates involved are typical plate velocities (1^2 cm/yr), especially during early stages ofexhumation, and bear no relation to normal erosion rates. Important observations are that: (i) as a result of strainpartitioning and fluid channelling, significant volumes of subducted crust may remain unreacted (i.e. metastable) evenat conditions as high as 700‡C and 3 GPa ^ with implications as to geophysical modeling; (ii) subducted continentalcrust shows no isotopic or geochemical evidence of interaction with mantle material. An unknown proportion ofsubducted continental crust must have escaped exhumation and effectively recycled into the mantle, with geochemicalimplications still to be explored, bearing in mind the above inefficiency of mixing. The repeated occurrence of UHPmetamorphism, hence of continental subduction, through time and space since at least the late Proterozoic shows thatit must be considered a common process, inherent to continental collision. Evidence of older, Precambrian UHPmetamorphism is to be sought in high-pressure granulite-facies terranes.8 2003 Elsevier Science B.V. All rights reserved.

0012-821X / 03 / $ ^ see front matter 8 2003 Elsevier Science B.V. All rights reserved.doi:10.1016/S0012-821X(03)00261-9

E-mail address: [email protected] (C. Chopin).

EPSL 6683 19-6-03 Cyaan Magenta Geel Zwart

Earth and Planetary Science Letters 212 (2003) 1^14

R

Available online at www.sciencedirect.com

www.elsevier.com/locate/epsl

mailto:[email protected]

http://www.elsevier.com/locate/epsl

Keywords: metamorphism; ultrahigh pressure; continental subduction; exhumation; mantle^crust interaction; collision; buoyancy;eclogite; coesite; diamond; majorite; kinetics; £uids

1. Introduction

One of the key features in the theory of platetectonics is the opposition of a young, continu-ously renewed and subducted oceanic lithosphere,to an old and everlasting continental lithosphere.In this paradigm, continental crust was assigned,because of its low density, the role of a £oatingobject which could be accreted, deformed, thick-ened and eroded in collision zones, but the fate ofwhich was restricted to an erosion^sedimentation^metamorphism/melting cycle encompassing atmost the 30 or 40 outer kilometers of the solidEarth. The existence of crustal mountain roots asdeep as 70 km was evident from gravity and seis-mic data, but it was unclear whether samplescould ever be brought from such depth to thesurface by tectonic processes.A ¢rst hint was given by the recognition that

rocks of oceanic origin in the central Alps, afterthey were hydrothermally transformed undernear-surface conditions, were metamorphosed toconditions of ca. 3 GPa (i.e. near 100 km depth)[1] and that the enclosing continental basementmay have shared the same burial [2]. In the early80s the successive discoveries of coesite, a high-pressure polymorph of quartz, in the western Alps[3] and the Caledonian chain of Norway [4] led tothe inescapable conclusion that continental crustcan be buried to 100 km depths and be broughtback to outcrop during the course of orogenicprocesses. The geological problems at stake be-came even more formidable after the discoveryof demonstrably metamorphic diamond in theCambrian orogen of the Kokchetav massif, Ka-zakhstan [5], implying minimum formation pres-sures of about 4 GPa, i.e. burial to depths exceed-ing 120 km.These ¢ndings, which rely entirely on careful

microscopic observation, have been the incentivefor a blooming ¢eld-oriented high-pressure re-search activity. A wealth of new ¢ndings over adecade con¢rms worldwide the generality of theearly discoveries, brings answers to the more

pressing questions but also pushes ever deeperthe limits of continental metamorphism, throughthe recognition of ever more tenuous indicators ofever higher pressures ^ and so raises new ques-tions. We examine here these recent developments^ made in parallel with those of mantle mineral-ogy [6] as revealed by high-pressure experimentalwork and by mineral inclusions in ‘super-deep’kimberlitic diamonds ^ as well as their limitationsand implications.

2. The force of mineralogical evidence

2.1. The classics: simple polymorphic indicators

The recognition of ultrahigh-pressure metamor-phic rocks has essentially relied on the identi¢ca-tion of the high-pressure polymorphs of silica(coesite) or of carbon (diamond). The simplicityof the ¢rst-order phase transitions involved indeedmakes the very presence of one such phase, what-ever its grain size or abundance, an immediate in-dicator of minimum pressures attained, on theorder of 3 and 4 GPa, respectively (Fig. 1). Theunderlying assumption that these phases formedin their stability ¢eld is borne out by independentlines of evidence, as shown below.The rock mineralogy tending to adapt to

changing pressure (P) and temperature (T) condi-tions, such UHP phases normally disappear dur-ing decompression, i.e. during the movement ofthe rock toward the surface ( = exhumation), thisthe more so if the rock enters the stability ¢eld ofthe respective low-pressure phase at high temper-ature. As a matter of fact, the metastable persis-tence of these indicators up to surface conditionsis commonly the result of their inclusion in me-chanically strong host minerals like garnet or zir-con (Fig. 2A,B,D). These act as pressure vesselsaround each inclusion, insulating it from meta-morphic £uid and preventing the volume increaseof the back-transformation, i.e. maintaining theinclusion pressure on the relevant transition curve


C. Chopin / Earth and Planetary Science Letters 212 (2003) 1^142

as long as the tensile strength of the host mineralcan sustain the di¡erence between the internal (in-clusion) and external pressure. Although early an-ticipated [7], the e¡ectiveness of this mechanismwas only recently demonstrated for small (10^20Wm) coesite inclusions in garnet and zircon, inwhich Raman microspectroscopy revealed resid-ual internal pressures that could exceed 2 GPa[8,9] ! Such residual pressures are also evidenceagainst metastable coesite formation. Note that,once the host mineral is fractured (Fig. 2A) ^ orin the absence of a ‘container’ ^ there is no other

obstacle than kinetics to the complete transforma-tion of the high-P phase into the lower-P poly-morph, i.e., for coesite/quartz, temperatures lowerthan about 375^400‡C [10] or a completely drysystem [11].

2.2. Relics and pseudomorphs: the limits ofevidence

The tiniest relic of a UHP indicator in a meta-morphic rock is in itself compelling evidence thatthe rock passed at some stage through UHP con-ditions, and this may imply entire reconsiderationof the geological or structural record for the area.Given these far-reaching implications, a prerequi-site is that the mineralogical evidence is unambig-uously characterized ^ and demonstrably in situ.Techniques with high spatial resolution, such asRaman microspectroscopy, are invaluable com-plements of optical observation in this respect.Yet, the limit of what can be done convincinglyhas probably been reached with a weak coesiteRaman signal in what is optically a polycrystallinequartz inclusion in Antarctic eclogite [12], or withthe report of diamond from the Rhodope, Greece[13], in which the Raman spectrum of ‘diamond’,usually an intense sharp band, can be interpretedas well as that of a defective graphite [14]. Be-sides, diamond is so prone to artifact, to beplucked o¡ or introduced into a sample duringprocessing (sawing, grinding) that utmost careand special techniques are required to ensure itsidenti¢cation in situ (e.g. [15]), the safest waybeing to work exclusively with crystals that areentirely included in another mineral, small sizeand confocal techniques permitting [5,16]. Thismay be the reason why the original reports oflarge UHP diamond from Dabie Shan, easternChina [17], have not yet been independently con-¢rmed by micro-¢ndings.However, absence of evidence is not evidence of

absence, and this may be a golden rule for UHPstudies. The original coesite report from the Ca-ledonides [4] could not be duplicated for a decadein spite of the e¡orts of several expert groups ^until a Ph.D. student was able to demonstrate theregional occurrence of coesite there [18,19]. Neweyes are sometimes required to open frontiersT

00 200200 400400 600600 800800 100010000

1

2

3

4

5

Fig. 1. P^T diagram showing the stability ¢eld of the mainUHP index minerals, coesite and diamond, and the decom-pression trajectory common to most UHP terranes (broadarrow). By de¢nition, UHP metamorphic conditions arereached above the coesite^quartz transition. For comparison,one of the coldest T/P ratios achievable in subduction zonesis shown (5‡C/km), as well as the 20‡C/km line, which maybe seen as a lower boundary value for common, low- to in-termediate-pressure metamorphism. Incipient partial meltingof granitic to granodioritic compositions in the presence andabsence of a hydrous £uid is also shown. The thin solid line(arrowed) is the P^T path computed by Roselle and Engi[82] for the top of a 5 km thick continental crustal fragmentsubducted to 100 km depth with a plate velocity of 3 cm/yrand a 30‡ dip, detached from the subducting plate, held atconstant depth for 3 Ma, then exhumed along the subduc-tion shear zone at a constant exhumation rate of 1 cm/yr(frictional shear heating is considered).


C. Chopin / Earth and Planetary Science Letters 212 (2003) 1^14 3

Fig. 2. Photomicrographs of characteristic or potential mineral indicators of UHP conditions (plane-polarized light unless other-wise speci¢ed). (A) Coesite (coes) inclusion in pyrope garnet (gt); note the incipient transformation into quartz (qz) and, as a re-sult of the related volume increase, the radial cracks in the host garnet [3] ; Parigi, Dora-Maira massif. (B) Ellenbergerite (Elb) in-cluded in pyrope garnet along with rutile (ru); Parigi, Dora-Maira massif. This Mg^Al^(Ti,Zr)-silicate contains 8 wt% H2O andis stable at pressures exceeding 2.7 GPa (and T6 725‡C). (C) Oriented quartz needles precipitated within omphacitic clinopyrox-ene (cpx); Blumenau eclogite, Erzgebirge [86]. This feature is often encountered in UHP eclogite. (D) Microdiamond (diam) in-clusions in zircon (zr); garnet gneiss, Seidenbach, Erzgebirge [16]. (E) Oriented orthopyroxene (opx) precipitates in garnet, evi-dence of a former super-silicic, majoritic garnet; Ugelvik peridotite, OtrTy island, western Norway [28]. (F) Oriented K-feldspar(Kfs) precipitates in clinopyroxene; crossed nicols; pyroxenite in marble, Kumdi Kol, Kokchetav massif.



The evidence may also become more elusive,when the indisputable relics of UHP mineralshave disappeared and only pseudomorphs thereof,i.e. breakdown products retaining the originalshape of the parent mineral, are present. Graphiteoctahedra [20] are a straightforward example ofpseudomorphs after diamond, but what is the di-agnostic value of graphite cuboids in schist [21]?Likewise, the texture of quartz pseudomorphingcoesite (Fig. 2A) is characteristic [7] but tends todisappear through recrystallization upon pro-longed thermal annealing. What then is the diag-nostic value of a well-annealed polycrystallinequartz inclusion in fractured garnet? Clearly atsome stage the force of the evidence declines,and only an array of converging hints may thenconvince the sceptical petrologist.

2.3. Beyond the polymorphic record:deeper and deeper with other UHP indicators

The intensive study of ascertained or potentialUHP rocks has revealed a number of uncommonmineralogical features and assemblages, of moreor less de¢nite UHP value. Some of these miner-alogical hints for UHP are illustrated in Fig. 2;many have recently been reviewed [22,23], so onlythe most salient or new features are addressedhere. Importantly, none of them involves a poly-morphic transition, thus avoiding the suspicionbearing on such phase transitions, namely thatthey may be a¡ected by high deviatoric stresses(e.g. [3]) ^ which is de¢nitely not the case formicrodiamond in the Erzgebirge UHP gneisses,which precipitated as a daughter mineral in super-critical COH £uid/melt inclusions in garnet [24,25].

2.3.1. K-bearing clinopyroxeneThe occurrence, in calc^silicate rocks of the di-

amond-bearing Kokchetav massif, of clinopyrox-ene with extremely high potassium contents (up to1.5 wt% K2O) [5] was a con¢rmation of stablediamond occurrence. So far solely known froma few eclogitic xenoliths in kimberlite, this featurewas experimentally shown to be stable at pres-sures of 4^10 GPa. Upon decompression this po-tassic pyroxene shows characteristic textures, with

oriented precipitates of K-feldspar (Fig. 2F) [5,26], sometimes of a later K-bearing mica [27].

2.3.2. Majoritic garnetAnother most signi¢cant discovery is, in the

peridotite body of OtrTy Island, western Norway,that of microtextural evidence for exsolution oforthopyroxene in coarse garnet, implying a‘super-silicic’ precursor garnet, i.e. the presenceof several mol% majorite component [28,29]. Itwas known from experiment [30] that with in-creasing pressure pyropic garnet can incorporatemore and more MgSiO3 as majorite component(to ¢t the garnet formula, better written Mg3 MgVISi IVSi3O12, in which one-fourth of the siliconatoms is six-fold coordinated). Modal estimates ofthe majorite component originally present in theNorwegian garnet, i.e. before exsolution, lead toformation pressures exceeding 6 GPa (s 200 km).Such sensational exsolution had been discoveredin garnet from kimberlite xenoliths a few yearsearlier [31] but it was completely unexpectedthat the delicate textures (Fig. 2E) could be pre-served during the long ascent and slow coolingof an orogenic peridotite body, as opposed toquenching by explosive kimberlite sampling. Ad-mittedly, mantle rather than crustal material wasinvolved in either case, but the demonstration wasmade that, even with the rates attending meta-morphic processes, such evidence could be pre-served ^ opening the prospect that it may befound in UHP crustal rocks too.Since then, indeed, three reports of related tex-

tures were made in UHP terranes. In garnet^peri-dotite of the Su-Lu terrane, eastern China [32],the textural evidence of pyroxene exsolutionfrom garnet is convincing but also involves ori-ented micrometer-size rods of rutile and apatite;however, it is unclear whether the extreme condi-tions recorded by the ultrama¢c body were sharedby the associated UHP (coesite-grade) crustalrocks. As in Norway ([28], cf. [33,34]), the extremeconditions may record an earlier, deeper mantlestage before the body was amalgamated withcrustal material during ‘normal’ UHP metamor-phism. In two other reports from the Greek Rho-dope [13,35], the crustal nature of the rocks isclear but the evidence of exsolution from a former



majoritic garnet is less compelling [14] : in garnetfrom kyanite^biotite gneiss, no pyroxene occursbut oriented rods of quartz and rutile [13] ; andin garnet from metabasite, with oriented rutilerods, clinopyroxene occurs as blebs [35], whichmay or may not represent recrystallized exsolu-tion products (alternatively they could be inclu-sions trapped during garnet growth).

2.3.3. ‘Super-silicic’ titaniteOther spectacular evidence for the metamor-

phism of crustal material at unsuspected depthsis the discovery of coesite precipitates in titanite,CaTiSiO5, in marble from the Kokchetav massif[27]. This suggests the existence of a super-silicicprecursor titanite, as experimentally anticipated ina high-pressure study of the series CaTiSiO5^CaSi2O5, in which the Ca-silicate end-member isstable above 8 GPa and has the titanite structurewith 50% six-fold coordinated silicon [36]. Theprecursor compositions reconstructed by integra-tion of the exsolved phases (also minor calcite and

apatite) point again to pressures that may haveexceeded 6 GPa [27], i.e. near 200 km burial. Par-adoxically, coesite in such a case plays the role ofa decompressional, lower-pressure phase!A general feature of these UHP conditions,

whether they approach the 3 GPa/650^800‡Crange in ‘standard’ coesite-bearing terranes orthe 4^6 GPa/900^1000‡C range in the diamond-bearing terranes, is that they represent a low T/Pratio at mantle depths. The most likely if not thesole geodynamic context in which such a low ratiois realized is subduction zones.

3. UHP through space and time

After an incubation period of 5 years after the¢rst coesite reports, more and more evidence forUHP metamorphism is being discovered world-wide (e.g. [23]), even in orogens where high-pres-sure metamorphic rocks were so far virtually ab-sent or terribly underestimated, like the Himalaya

Fig. 3. The UHP metamorphic terranes presently known, on the basis of coesite or metamorphic diamond occurrence. Adaptedfrom [57]; references in text.



[37,38], Indonesia [8], Antarctica [12] or SouthAmerica [39]. In short, de¢nite evidence of UHPmetamorphism is now missing only from Austra-lia and North America (Fig. 3).

3.1. Coherent UHP terranes

A major result concerns the actual extent ofUHP rocks within a given metamorphic terrane.Pieces of de¢nite evidence for UHP remain rareand occur in rock types (eclogitic metabasite,magnesian schist, marble or pelitic metasediment)that are subordinate with respect to the over-whelming country rock of felsic gneiss (metagran-ite), in which evidence for UHP has long re-mained elusive. Yet the areal extent of UHPoccurrences is over kilometers in the Alps [40],tens of kilometers in Norway [19] and the Kok-chetav massif [9], and hundreds of kilometersalong the Hong’an-Dabie-Su-Lu Triassic belt ofeast central China (e.g. [22]).The key point is whether the UHP rocks and

their country-rock gneiss, which bears only low-grade assemblages, share the same UHP historyor were tectonically amalgamated during a laterstage, i.e. whether the metamorphic terrane as awhole forms a coherent UHP terrane or a mixtureof UHP and low-pressure rocks. New problem,new methods: beside extensive zircon U^Pbpoint-dating [41] and stable-isotope data [42],the systematic study by Raman spectroscopy ofthe minute mineral inclusions in zircon concen-trates extracted from the country-rock gneisseshas proved to be most e¡ective in unraveling theformer extent of UHP conditions in rocks thatbear now only low-grade mineral assemblages, es-pecially in the huge Dabie-Su-Lu belt [43^45].There the resulting evidence is that the gneisses,whether of sedimentary derivation (paragneiss as-sociated with eclogite and marble), or of mag-matic derivation (orthogneiss, former graniteforming large tracts of country rock), both com-monly bear coesite relics in zircon, thereby dem-onstrating the coherency of the UHP terrane. Asimilar conclusion was reached in the small UHPterrane (15U5U1 km) of the Dora-Maira massif,western Alps [40] and might hold in Norway[18,19,46], as well as in the Erzgebirge [47] and

the Kokchetav massif [9], in both of which micro-diamonds are also gneiss-hosted. However, theimplications are particularly formidable in China,given the scale of the UHP terrane: several hun-dred kilometers in length, a few tens of kilometerslateral exposure, and some kilometers thickness!Obviously large tracts of continental crust wereinvolved in UHP metamorphism.

3.2. Shaping the continents

One of the most fascinating aspects of recent¢ndings is the repeated occurrence through timeof UHP metamorphism of continental crust. Thecomposite Eurasian continent bears the scars ofthe successive orogenies that make it a collage ofmicrocontinental blocks. In each of these scarsUHP rocks do occur, marking the steps of con-tinental accretion (Fig. 3) : from the CambrianKokchetav massif in Kazakhstan (530 Ma[48,49]), to the Early Paleozoic central Chinabelt (Altun-Qaidam and Qinling; ca. 500 Ma[50], Jingsui Yang et al., article in revision), tothe Triassic Hong’an-Dabie-Su-Lu belt (220^240Ma, e.g. [41]), to the Cretaceous merging of theLhassa block prolongation in Indonesia (110^120Ma [51]), to the Tertiary Himalayan belt (45 Ma?[37]). On the western side of the Eurasian block,UHP crustal metamorphism is likewise well rep-resented in the Caledonian chain (Laurentia^Bal-tica collision) from eastern Greenland [52] to Nor-way, in the Variscan belt (Laurentia^Gondwanacollision) from Central to Western Europe (Bohe-mian massif [16] and Monts du Lyonnais, FrenchMassif Central [53], respectively), but also in theMeso-Cenozoic Alpine chain (Europe^Apulia col-lision), from the Rhodope massif (70^75 Ma [35])to the western Alps (38^35 Ma [54^56]). By con-trast, the processes active along the circum-Paci¢cmargins seem to be unsuitable to produce or,more likely, to exhume UHP rocks. In the caseof Eurasia, the UHP material involved may be ofoceanic origin in a few instances (Sulawesi, Montsdu Lyonnais, Cignana) but continental in mostcases, attesting to the general rule that continentalsubduction has operated at least since the lateProterozoic.This apparent age limit as well as the paucity



of Precambrian high-pressure, low-temperaturerocks (blueschists and eclogites) have led to thesuggestion of a rapid decrease of the subduction-zone thermal gradient during the late Proterozoic,after a massive heat release made possible by thedislocation of the Rodinia supercontinent [57].However, the existence of Proterozoic UHP meta-morphism in the Pan-African belt of Mali (ca. 600Ma [58]) and Brazil [39], of 2 Ga old eclogites inthe Usagaran belt of Tanzania [59] and of 2.6 Gahigh-pressure granulites in the western Canadianshield [60] indicates that zones of low T/P ratiosdid exist early in the Earth’s history. The preser-vation of the HP/UHP rocks formed was admit-tedly more di⁄cult because of the Earth’s higherheat production, and the likely higher averagegeothermal gradient and lesser plate thickness.This may be su⁄cient to account for the paucityof preserved Precambrian blueschists and eclo-gites.The important point is that in collision belts ^

as opposed to Paci¢c-type margins ^ subductionof continental crust to depths reaching 100^200km appears to be the rule since at least the Pro-terozoic. The implications in terms of tectonicsand geodynamics are far-reaching. Were it simplyfor the Himalaya, most of what has previouslybeen considered relevant for the archetype of thecollision belts ^ in terms of thermal regime, meta-morphism, partial melting, uplift and cooling ^actually pertains to a late stage of the chain evo-lution: metamorphism of continental crust startedabout 20 Ma earlier with the subduction of theleading edge of the Indian plate to coesite-formingdepths (90^100 km); the present-day shallow dipof India below southern Tibet bears no relation toan initially much steeper dip, and the average ex-humation rates are twice as high as previouslythought (ca. 10 mm/yr) [37].

4. A window into subduction-zone processes

Another important aspect of these ¢ndings isthat the exposure of large tracts of HP/UHProcks represents windows opening into subduc-tion-zone processes. Aspects thus accessible to ob-servation and measurement include the extent of

mineral transformation and the role of £uids anddeformation under such conditions, the degree ofinteraction with mantle material ^ in the footwallof the subduction zone or in the overlying mantlewedge ^ and its possible geochemical signature.Admittedly, the message is in essence a palimpsestthat has to be deciphered, since the features ac-quired under UHP conditions may have been al-tered, overprinted or erased during the later de-compression history. Yet it is a unique record.

4.1. Extent of rock transformation at UHP

A repeated observation in HP/UHP terranes isthat signi¢cant rock volumes may escape defor-mation and mineral equilibration at the prevailingP^T conditions, despite temperatures as high as650^750‡C for UHP rocks. Famous examples arethe metagranite body of Flatraket in the WesternGneiss Region, Norway [61], or the Brossascometagranite in the Dora-Maira massif [62] (Fig.4). Pure volume di¡usion is clearly an ine⁄cientprocess; the partitioning of deformation and therelated £uid access, both triggering nucleation andmass transfer [63,64], are actually much more ef-¢cient than temperature.This has led thermomechanical modelers to

consider not only the changing mineralogy ac-cording to P^T conditions, in order to accountfor changing physical properties like density or

Fig. 4. Undeformed UHP metagranite sample, Brossasco,Dora-Maira massif. Would you believe that this Hercyniangranite underwent conditions of 700^750‡C at 100 km depthduring Alpine times? Yet it did, according to de¢nite miner-alogical evidence [62].



rheology, but also the e¡ectiveness of this trans-formation (e.g. [65^67]), i.e. metastable persis-tence or absence thereof. The consequences interms of density changes and buoyancy are signif-icant (see below).

4.2. Isotopic behavior at UHP

This contrast between reactive and metastablesystems, regardless of the high temperatures at-tained, is also entirely re£ected in their geochem-ical and isotopic behavior, thereby questioningthe meaning and the applicability of closure tem-peratures. Monazite U^Pb ages in undeformedfacies of the Brossasco metagranite are intermedi-ate between the Hercynian intrusion age and theAlpine age of the UHP event, whereas they arede¢nitely Alpine in nearby UHP rocks like mag-nesian schists [68] which were completely re-worked during Alpine times. The same holds forU^Pb in titanite, for which formation ages allowone to date the successive steps of UHP metamor-phism, incipient overprint and low-grade over-print, depending on the rock type addressed[56]. Evidently the rocks showing the largest min-eral metastability under UHP conditions showRb^Sr disequilibrium among their phases [69]and are the worst target for ArÂr dating (e.g.[70]).Independent of the problem of the equilibration

scale, there is an inherent problem to KÂr sys-tematics in UHP terranes. Under UHP conditionsthe partial pressure of argon is expected to be oneorder of magnitude higher than in low-pressureterranes, leading to signi¢cant argon incorpora-tion during the growth of some UHP phases, inparticular white mica. The result is commonlythat the KÂr ages of UHP micas (even ArÂrplateau ages) are paradoxically higher than thosederived from the more retentive U^Pb or Rb^Srsystems, and geologically meaningless [70,71].

4.3. Fluid behavior: open vs. closed system,mantle^crust interaction, rheology

In contrast to reports of low-pressure terranes£ushed by ‘oceans’ of hydrothermal £uids, theexperience gained in HP and UHP terranes is

that of very limited £uid £ow and at best centi-meter-scale isotopic equilibration, in spite of theuncommon P^T conditions. As summarized byRumble [42], ‘stable isotope evidence denies theexistence of a pervasive £uid free to in¢ltrateacross lithologic contacts during UHP metamor-phism. Furthermore, the preservation of high-temperature oxygen isotope fractionations amongminerals argues against the presence of free £uidafter peak metamorphism, during exhumationand cooling. Residence in the upper mantle hadno discernible metasomatic e¡ect on the stableisotope composition of crustal rocks subductedduring continental collision’. The best evidencefor this is the preservation, throughout the UHPevent in Dabie-Su-Lu, of a huge oxygen and hy-drogen isotope negative anomaly that dates backto a premetamorphic, Neoproterozoic hydrother-mal system involving meteoric water from a coldclimate [72]. This does not mean the rocks weredry: most UHP assemblages bear hydrous miner-als like micas or epidote that are stable underthese conditions, and therefore evolve little £uid.The presence or absence of a £uid also directly

a¡ects the temperature of partial melting (Fig. 1).Therefore, depending on both £uid availabilityand the P^T path followed, in particular duringdecompression, melting may be expected to beminimal or quite general. Interestingly, underUHP conditions there may be complete miscibilitybetween silicate melts and hydrous £uids for arange of compositions, with the consequencethat the supercritical £uids generated at UHP bydehydration reactions during subduction can dis-solve considerable amounts of silicate materialwith increasing T, and therefore be of restrictedmobility (as compared to the more hydrous £uidsproduced at lower pressure, in the subcriticalrange) [73,74]. Besides, the presence of such £uids,whenever water is available, has a decisive weak-ening e¡ect on the rheological strength of rocksduring deep subduction, precluding notable shearheating [75].

5. Exhumation: rates and processes

The realization that crustal segments may re-



turn to the surface from depths exceeding 120 kmraises a di⁄cult geological problem: that of themechanisms responsible for their burial and, morepressingly, for their exhumation from suchdepths. The petrological record indicates thatlarge tracts of crustal rocks were buried to UHPconditions of low T/P ratios, which is consistentwith and best accounted for by a subduction-zonecontext. It has indeed long been recognized bymodeling that several hundred kilometers of theleading edge of a continental plate can be en-trained into a subduction zone by the sole e¡ectof body forces, all the more so as the continentalmargin is thinner and the cohesion between con-tinental and sinking oceanic lithosphere is higher.Continental subduction during incipient collisionis therefore an easy way to account for the rele-vant burial.As to exhumation, the petrological record in

most instances constrains decompression to haveoccurred under continuous cooling [3,9,23,40,64],implying continuous heat loss to the footwall and/or hangingwall of the rising body. In addition,isotopic dating shows that formation ages andcooling ages obtained on UHP rocks span a nar-row range of a few million years in most instances[35,41,48,49,54^56,76], implying high coolingrates and, through the petrological constraints,uncommonly high decompression rates, especiallyduring early stages of exhumation. The exhuma-tion rates obtained (up to 1^2 cm/yr) are indeedmore akin to plate velocities and bear no relationto normal erosion rates. It was soon recognizedthat erosion and tectonic processes like extension-al faulting or ‘corner £ow’ are unable to accountfor the critical, early stage of extrusion, even ifthey may become e¡ective during later stages, athigher structural levels. Besides, it is certainly notfortuitous that most UHP terranes presently ex-posed are essentially made of light, upper-crustalmaterial, thus pointing to the role of buoyancy asa main (?) factor of exhumation (e.g. [77]). Theconstraint of continuous cooling precludes anymechanism like diapiric rise across the overlyingmantle wedge and makes the return back alongthe subduction channel ^ a ‘two-way street’ andobvious mechanical discontinuity ^ the mostlikely exhumation path. The process is essentially

driven by buoyancy up to mid-crustal levels, as aresult of the smaller density of the acidic crustalrocks (even if completely re-equilibrated atdepth), after delamination from the lower crust.Field data (e.g. [76,79,80]) as well as physical [81],thermal [82] or thermo-mechanical [66,67,87]modeling support the idea that most of the exhu-mation of UHP rocks may be achieved by such‘extrusion’ of light crustal material along the sub-duction channel, in a continuously convergent set-ting, this without the need to resort to slab break-o¡ as a ‘deus ex machina’.

6. Consequences and prospects

A new picture emerges for the fate of continen-tal margins and microcontinents. Continental sub-duction must now be considered a normal pro-cess, inherent to continental-plates convergence,collision and orogeny ^ at least since the late Pro-terozoic. Part of (most of?) the subducted uppercrust may be delaminated and exhumed, bringingtestimony of subduction-zone processes back tothe surface, with a clear geochemical message:the ine⁄ciency of mixing. Such a message couldhelp us shape our view of the deeper mantle interms of its heterogeneity and inheritance fromsubducted slabs (e.g. [83]). An open question isstill the extent to which continental lithosphere,upper and lower crust possibly included, may ac-tually disappear into subduction zones and con-tribute to magma generation [78], mantle geo-chemistry, geophysical signature, etc. Anotherconcern is that our present view of UHP meta-morphism might be intrinsically biased by thenecessary preservation of a mineralogical record.It may well be that subducted continental frag-ments were exhumed and decompressed undernear-isothermal or even increasing temperatureconditions, leading to dehydration reactions andpartial melting (Fig. 1). This would likely erasemost records of the early UHP stage. High-grade,granulite-facies metamorphic terranes are there-fore also open to UHP detective work [84].Given the exploding number of UHP ¢ndings

during the past few years, the author is con¢dentthat UHP metamorphism will sooner or later be



recognized on all continents and back into thePrecambrian as well. An important issue will beincreasing chronological and spatial resolution inthe dating of the highest-pressure and early-exhu-mation stages, for instance in the dating of suc-cessive growth zones of crystals. Conversely, theever-increasing spatial resolution of characteriza-tion techniques also hints at one of the limits tobe encountered. More and more polymorphs hav-ing a UHP thermodynamic stability ¢eld will beidenti¢ed at the submicrometer scale, down to‘crystals’ having only a few unit-cells extension;thereby the limit is reached under which surfaceenergies may become preponderant over enthalpicproperties, casting some doubt as to the actualP^T formation conditions. The recent ¢ndingsof nanometer-size TiO2 with the K-PbO2 structurein Erzgebirge [85] show that this concern is al-ready no longer a purely theoretical one.

Acknowledgements

Sharing the excitement of ‘high-pressure hunt-ing’ with Bruno Go¡e¤ and Werner Schreyer hasbeen a life experience. The help of B. Go¡e¤ andA. Michard through comments on an early ver-sion, of H.-J. Massonne, H.-P. Schertl and E.Schma«dicke for photographs or rock samples,and of M. Engi, W. Schreyer and an anonymousreferee through helpful reviews are gratefully ac-knowledged, as well as the editorial patience ofA.N. Halliday.[AH]

References

[1] B.W. Evans, V. Trommsdor¡, W. Richter, Petrology ofan eclogite-metarodingite suite at Cima di Gagnone, Tici-no, Switzerland, Am. Mineral. 64 (1979) 15^31.

[2] C.A. Heinrich, Eclogite facies regional metamorphism ofhydrous ma¢c rocks in the central alpine Adula nappe,J. Petrol. 27 (1986) 123^154.

[3] C. Chopin, Coesite and pure pyrope in high-grade blue-schists of the western Alps: a ¢rst record and some con-sequences, Contrib. Mineral. Petrol. 86 (1984) 107^118.

[4] D.C. Smith, Coesite in clinopyroxene in the Caledonidesand its implications for geodynamics, Nature 310 (1984)641^644.

[5] N.V. Sobolev, V.S. Shatsky, Diamond inclusions in garnet

from metamorphic rocks: a new environment for dia-mond formation, Nature 343 (1990) 742^746.

[6] R.J. Hemley (Ed.), Ultrahigh-pressure Mineralogy: Phys-ics and Chemistry of the Earth’s Deep Interior, MSARev. Mineral. 37 (1998).

[7] Ph. Gillet, J. Ingrin, C. Chopin, Coesite in subductedcontinental crust: P-T history deduced from an elasticmodel, Earth Planet. Sci. Lett. 70 (1984) 426^436.

[8] C.D. Parkinson, I. Katayama, Present-day ultrahigh-pres-sure conditions of coesite inclusions in zircon and garnet:Evidence from laser Raman microspectroscopy, Geology27 (1999) 979^982.

[9] C.D. Parkinson, Coesite inclusions and prograde compo-sitional zonation of garnet in whiteschist of the HP-UHPM Kokchetav massif, Kazakhstan: a record of pro-gressive UHP metamorphism, Lithos 52 (2000) 215^233.

[10] J.L. Mosenfelder, S.R. Bohlen, Kinetics of coesite toquartz transformation, Earth Planet. Sci. Lett. 153(1997) 133^147.

[11] J.G. Liou, R.Y. Zhang, Occurrences of intergranular coe-site in ultrahigh-P rocks from the Su-Lu region, easternChina: implications for lack of £uid during exhumation,Am. Mineral. 81 (1996) 1217^1221.

[12] B. Ghiribelli, M.L. Frezzotti, R. Palmeri, Coesite in eclo-gites of the Lanterman Range (Antarctica): Evidencefrom textural and Raman studies, Eur. J. Mineral. 14(2002) 355^360.

[13] E.D. Mposkos, D.K. Kostopoulos, Diamond, former coe-site and supersilicic garnet in metasedimentary rocks fromthe Greek Rhodope: a new ultrahigh-pressure metamor-phic province established, Earth Planet. Sci. Lett. 192(2001) 497^506.

[14] O. Beyssac, C. Chopin, Comment on ‘Diamond, formercoesite and supersilicic garnet in metasedimentary rocksfrom the Greek Rhodope: a new ultrahigh-pressure meta-morphic province established’ by Mposkos and Kosto-poulos, Earth Planet. Sci. Lett., accepted for publication.

[15] H.J. Massonne, H.J. Bernhardt, D. Dettmar, E. Kessler,O. Medenbach, T. Westphal, Simple identi¢cation andquanti¢cation of microdiamonds in rock thin-sections,Eur. J. Mineral. 10 (1998) 497^504.

[16] L. Nasdala, H.J. Massonne, Microdiamonds from theSaxonian Erzgebirge, Germany: in situ micro-Ramancharacterization, Eur. J. Mineral. 12 (2000) 495^498.

[17] S. Xu, A.I. Okay, S. Ji, A.M.C. Sengo«r, W. Su, Y. Liu, L.Jiang, Diamond from the Dabie Shan metamorphic rocksand its implications for tectonic setting, Science 256(1992) 80^82.

[18] A. Wain, New evidence for coesite in eclogites andgneisses: De¢ning an ultrahigh-pressure province in theWestern Gneiss Region of Norway, Geology 25 (1997)927^930.

[19] A. Wain, D. Waters, A. Jephcoat, H. Olijynk, The high-pressure to ultrahigh-pressure eclogite transition in theWestern Gneiss Region, Norway, Eur. J. Mineral. 12(2000) 667^687.

[20] D.G. Pearson, G.R. Davies, P.H. Nixon, H.J. Milledge,



Graphitized diamonds from a peridotite massif in Moroc-co and its implication for anomalous diamond occur-rences, Nature 338 (1989) 60^62.

[21] M.L. Leech, W.G. Ernst, Graphite pseudomorphs afterdiamond? A carbon isotope and spectroscopic study ofgraphite cuboids from the Maksyutov Complex, southUral Mountains, Russia, Geochim. Cosmochim. Acta 62(1998) 2143^2154.

[22] J.G. Liou, R.Y. Zhang, W.G. Ernst, D. Rumble III, S.Maruyama, High-pressure minerals from deeply sub-ducted metamorphic rocks, Rev. Mineral. 37 (1998) 33^96.

[23] D.A. Carswell, R.Y. Zhang, Petrographic characteristicsand metamorphic evolution of ultrahigh-pressure eclogitesin plate-collision belts, Int. Geol. Rev. 41 (1999) 781^798.

[24] B. Sto«ckhert, J. Duyster, C. Trepmann, H.-J. Massonne,Microdiamond daughter crystals precipitated from super-critical COH plus silicate £uids included in garnet, Erzge-birge, Germany, Geology 29 (2001) 391^394.

[25] S.L. Hwang, P. Shen, H.T. Chu, T.F. Yui, C.C. Lin,Genesis of microdiamonds from melt and associated mul-tiphase inclusions in garnet of ultrahigh-pressure gneissfrom Erzgebirge, Germany, Earth Planet. Sci. Lett. 188(2001) 9^15.

[26] R.Y. Zhang, J.G. Liou, W.G. Ernst, R.G. Coleman, N.V.Sobolev, V.S. Shatsky, Metamorphic evolution of dia-mond-bearing and associated rocks from the KokchetavMassif, northern Kazakhstan, J. Metamorph. Geol. 15(1997) 479^496.

[27] Y. Ogasawara, K. Fukasawa, S. Maruyama, Coesite ex-solution from supersilicic titanite in UHP marble from theKokchetav Massif, northern Kazakhstan, Am. Mineral.87 (2002) 454^461.

[28] H.L.M. Van Roermund, M.R. Drury, Ultra-high pressure(Ps 6 GPa) garnet peridotites in Western Norway: exhu-mation of mantle rocks from s 185 km depth, TerraNova 10 (1998) 295^301.

[29] H.L.M. Van Roermund, M.R. Drury, A. Barnhoorn, A.De Ronde, Relict majoritic garnet microstructures fromultra-deep orogenic peridotites in Western Norway, J. Pe-trol. 42 (2001) 117^130.

[30] M. Akaogi, S. Akimoto, Pyroxene-garnet solid-solutionequilibria in the systems Mg4Si4O12-Mg3Al2Si3O12 andFe4Si4O12-Fe3Al2Si3O12 at high pressures and tempera-tures, Phys. Earth Planet. Int. 15 (1977) 90^106.

[31] S.E. Haggerty, V. Sautter, Ultradeep (greater than 300kilometers), ultrama¢c upper mantle xenoliths, Science248 (1990) 993^996.

[32] K. Ye, B. Cong, D. Ye, The possible subduction of con-tinental material to depths greater than 200 km, Nature407 (2000) 734^736.

[33] H.L.M. Van Roermund, D.A. Carswell, M.R. Drury,T.C. Heijboer, Micro-diamonds in a megacrystic garnet-websterite pod from Bardane on the island of FjTrtoft,western Norway, Geology 30 (2002) 959^962.

[34] H.K. Brueckner, D.A. Carswell, W.L. Gri⁄n, Paleozoicdiamonds within a Precambrian peridotite lens in UHP

gneisses of the Norwegian Caledonides, Earth Planet. Sci.Lett. 203 (2002) 805^816.

[35] A. Liati, D. Gebauer, R. Wysoczanski, U-Pb SHRIMP-dating of zircon domains from UHP garnet-rich ma¢crocks and late pegmatoids in the Rhodope zone (NGreece); evidence for Early Cretaceous crystallizationand Late Cretaceous metamorphism, Chem. Geol. 184(2002) 281^299.

[36] R. Knoche, R. Angel, F. Seifert, T. Fliervoet, Completesubstitution of Si for Ti in titanite Ca(Ti1�xSix)SiO5, Am.Mineral. 83 (1998) 1168^1175.

[37] P.J. O’Brien, N. Zotov, R. Law, M.A. Khan, M.Q. Jan,Coesite in Himalayan eclogite and implications for modelsof India-Asia collision, Geology 29 (2001) 435^438.

[38] B.K. Mukherjee, H.K. Sachan, Discovery of coesite fromIndian Himalaya: A record of ultra-high pressure meta-morphism in Indian Continental Crust, Curr. Sci. (India)81 (2001) 1358^1361.

[39] C.D. Parkinson, A. Motoki, C.T. Onishi, S. Maruyama,Ultrahigh-pressure pyrope-kyanite granulites and associ-ated eclogites in Neoproterozoic nappes of southeast Bra-zil, Abstr. UHPM Workshop Waseda University, 2001,pp. 87^90.

[40] C. Chopin, H.P. Schertl, The UHP unit in the Dora-Maira massif, Western Alps, Int. Geol. Rev. 41 (1999)765^780.

[41] B.R. Hacker, L. Ratschbacher, L. Webb, T. Ireland, D.Walker, D. Shuwen, U/Pb zircon ages constrain the archi-tecture of the ultrahigh-pressure Qinling-Dabie Orogen,China, Earth Planet. Sci. Lett. 161 (1998) 215^230.

[42] D. Rumble, Stable isotope geochemistry of ultrahigh-pres-sure rocks, in: B.R. Hacker, J.G. Liou (Eds.), When Con-tinents Collide: Geodynamics and Geochemistry of Ultra-high-Pressure Rocks, Kluwer, 1998, pp. 241^259.

[43] H. Tabata, K. Yamauchi, S. Maruyama, Tracing the ex-tent of a UHP terrane: Mineral-inclusion study of zirconsin gneisses from the Dabie Shan, in: B.R. Hacker, J.G.Liou (Eds.), When Continents Collide: Geodynamics andGeochemistry of Ultrahigh-Pressure Rocks, Kluwer, 1998,pp. 261^273.

[44] K. Ye, Y.P. Yao, I. Katayama, B. Cong, Q. Wang, S.Maruyama, Large areal extent of ultrahigh-pressure meta-morphism in the Sulu ultrahigh-pressure terrane of EastChina: new implications from coesite and omphacite in-clusions in zircon of granitic gneiss, Lithos 52 (2000) 157^164.

[45] F.L. Liu, Z. Xu, J.G. Liou, I. Katayama, H. Masago, S.Maruyama, J. Yang, Ultrahigh-pressure mineral inclu-sions in zircons from gneissic core samples of the ChineseContinental Scienti¢c Drilling Site in eastern China, Eur.J. Mineral. 14 (2002) 499^512.

[46] S. Cuthbert, D.A. Carswell, E.J. Krogh-Ravna, A. Wain,Eclogites and eclogites in the Western Gneiss Region,Norwegian Caledonides, Lithos 52 (2000) 165^195.

[47] H.J. Massonne, First ¢nd of coesite in the ultrahigh-pres-sure metamorphic area of the central Erzgebirge, Ger-many, Eur. J. Mineral. 13 (2001) 565^570.



[48] J. Hermann, D. Rubatto, A. Korsakov, V.S. Shatsky,Multiple zircon growth during fast exhumation of dia-mondiferous, deeply subducted continental crust (Kok-chetav Massif, Kazakhstan), Contrib. Mineral. Petrol.141 (2001) 66^82.

[49] I. Katayama, S. Maruyama, C.D. Parkinson, K. Terada,Y. Sano, Ion micro-probe U-Pb zircon geochronology ofpeak and retrograde stages of ultrahigh-pressure meta-morphic rocks from the Kokchetav massif, northern Ka-zakhstan, Earth Planet. Sci. Lett. 188 (2001) 185^198.

[50] J. Yang, Z. Xu, S. Song, J. Zhang, C. Wu, R. Shi, H. Li,M. Brunel, Discovery of coesite in the North QaidamEarly Palaeozoic ultrahigh pressure (UHP) metamorphicbelt, NW China, C. R. Acad. Sci. Paris 333 (2001) 719^724.

[51] C.D. Parkinson, K. Miyazaki, K. Wakita, A.J. Barber,D.A. Carswell, An overview and tectonic synthesis ofthe pre-Tertiary very-high-pressure metamorphic and as-sociated rocks of Java, Sulawesi and Kalimantan, Indo-nesia, Island Arc 7 (1998) 184^200.

[52] J.A. Giletti, E.J. Krogh-Ravna, First evidence for ultra-high-pressure metamorphism in the North-East Green-land Caledonides, Geology 30 (2002) 551^554.

[53] J.M. Lardeaux, P. Ledru, I. Daniel, S. Duche“ne, The Va-riscan French Massif Central - a new addition to theultrahigh pressure metamorphic ‘club’: exhumation pro-cesses and geodynamic consequences, Tectonophysics 332(2001) 143^167.

[54] G.R. Tilton, W. Schreyer, H.P. Schertl, Pb-Sr-Nd isotopicbehavior of deeply subducted crustal rocks from theDora-Maira-Massif, Western Alps-II: what is the age ofthe ultrahigh-pressure metamorphism?, Contrib. Mineral.Petrol. 108 (1991) 22^33.

[55] D. Gebauer, H.P. Schertl, M. Brix, W. Schreyer, 35 Maold ultrahigh-pressure metamorphism and evidence forvery rapid exhumation in the Dora-Maira-Massif, West-ern Alps, Lithos 41 (1997) 5^24.

[56] D. Rubatto, J. Hermann, Exhumation as fast as subduc-tion?, Geology 29 (2001) 3^6.

[57] S. Maruyama, J.G. Liou, Initiation of ultrahigh-pressuremetamorphism and its signi¢cance on the Proterozoic-Phanerozoic boundary, Island Arc 7 (1998) 6^35.

[58] B. Jahn, R. Caby, P. Monie¤, The oldest UHP eclogites ofthe world: age of UHP metamorphism, nature of proto-liths and tectonic implications, Chem. Geol. 178 (2001)143^158.

[59] A. Mo«ller, P. Appel, K. Metzger, V. Schenk, Evidence fora 2 Ga subduction zone Eclogites in the Usagaran belt ofTanzania, Geology 23 (1995) 1067^1070.

[60] D.R. Snoeyenbos, M.L. Williams, S. Hanmer, Archeanhigh-pressure metamorphism in the western CanadianShield, Eur. J. Mineral. 7 (1995) 1251^1272.

[61] A.L. Wain, D.J. Waters, H. Austrheim, Metastability ofgranulites and process of eclogitisation in the UHP regionof western Norway, J. Metamorph. Geol. 19 (2001) 609^626.

[62] M. Bruno, R. Compagnoni, M. Rubbo, The ultra-high

pressure coronitic and pseudomorphous reactions in ametagranodiorite from the Brossasco-Isasca Unit, Dora-Maira Massif, western Italian Alps: a petrographic studyand equilibrium thermodynamic modelling, J. Meta-morph. Geol. 19 (2001) 33^43.

[63] A.K. Straume, H. Austrheim, Importance of fracturingduring retro-metamorphism of eclogites, J. Metamorph.Geol. 17 (1999) 637^652.

[64] G. Simon, C. Chopin, Enstatite-sapphirine crack-relatedassemblages in ultrahigh-pressure pyrope megablasts,Dora-Maira massif, western Alps, Contrib. Mineral. Pe-trol. 140 (2001) 422^440.

[65] R. Bousquet, B. Go¡e¤, P. Henry, X. Le Pichon, C. Cho-pin, Kinematic, thermal and petrological model of theCentral Alps: Lepontine metamorphism in the uppercrust and eclogitisation of the lower crust, Tectonophysics273 (1997) 105^127.

[66] M.P. Doin, P. Henry, Subduction initiation and continen-tal crust recycling the roles of rheology and eclogitization,Tectonophysics 342 (2001) 163^191.

[67] E. Burov, L. Jolivet, L. Le Pourhiet, A. Poliakov, A ther-momechanical model of exhumation of high pressure(HP) and ultra-high pressure (UHP) metamorphic rocksin Alpine-type collision belts, Tectonophysics 342 (2001)113^136.

[68] J.L. Paquette, J.M. Montel, C. Chopin, U-Pb-Th datingof the Brossasco ultrahigh-pressure metagranite, Dora-Maira massif, western Alps, Eur. J. Mineral. 11 (1999)69^77.

[69] G.R. Tilton, L. Ames, H.P. Schertl, W. Schreyer, Recon-naissance isotopic investigations on rocks of an unde-formed granite contact within the coesite-bearing unit ofthe Dora-Maira-Massif, Lithos 41 (1997) 25^36.

[70] S. Scaillet, K-Ar (40Ar/39Ar) geochronology of ultrahighpressure rocks, in: B.R. Hacker, J.G. Liou (Eds.), WhenContinents Collide: Geodynamics and Geochemistry ofUltrahigh-Pressure Rocks, Kluwer, 1998, pp. 161^201.

[71] D. Giorgis, M. Cosca, S. Li, Distribution and signi¢canceof extraneous argon in UHP eclogite (Sulu terrain, Chi-na): insight from in situ 40Ar/39Ar UV-laser ablation anal-ysis, Earth Planet. Sci. Lett. 181 (2000) 605^615.

[72] D. Rumble, D. Giorgis, T. Ireland, Z. Zhang, H. Xu, T.F.Yui, J. Yang, Z. Xu, Low N

18O zircons, U-Pb dating, andthe age of the Qinglongshan oxygen and hydrogen isotopeanomaly near Donghai in Jiangsu Province, China, Geo-chim. Cosmochim. Acta 66 (2002) 2299^2306.

[73] H. Bureau, H. Keppler, Complete miscibility between sil-icate melts and hydrous £uids in the upper mantle: Ex-perimental evidence and geochemical implications, EarthPlanet. Sci. Lett. 165 (1999) 187^196.

[74] W. Schreyer, Experimental aspects of UHP metamor-phism: granitic systems, Int. Geol. Rev. 41 (1999) 701^710.

[75] B. Sto«ckhert, J. Renner, Rheology of crustal rocks atultrahigh pressure, in: B.R. Hacker, J.G. Liou (Eds.),When Continents Collide: Geodynamics and Geochemis-try of Ultrahigh-Pressure Rocks, Kluwer, 1998, pp. 57^95.



[76] B.R. Hacker, L. Ratschbacher, L. Webb et al., Exhuma-tion of ultrahigh-pressure continental crust in east centralChina: Late Triassic-Early Jurassic tectonic unroo¢ng,J. Geophys. Res. 105 (2000) 13,339^13,364.

[77] W.G. Ernst, S. Maruyama, S. Wallis, Buoyancy-driven,rapid exhumation of ultrahigh-pressure metamorphosedcontinental crust, Proc. Natl. Acad. Sci. USA 94 (1997)9532^9537.

[78] W.G. Ernst, Subduction, ultrahigh-pressure metamor-phism, and regurgitation of buoyant crustal slices - impli-cations for arcs and continental growth, Phys. Earth Plan-et. Int. 127 (2001) 253^275.

[79] A. Michard, C. Henry, C. Chopin, Structures in UHPMrocks: a case study from the Alps, in: R.G. Coleman, X.Wang (Eds.), Ultrahigh Pressure Metamorphism, Cam-bridge University Press, New York, 1995, pp. 132^158.

[80] D. Avigad, C. Chopin, R. Le Bayon, Thrusting and ex-tension in the southern Dora-Maira ultra-high-pressuremassif (Western Alps): view from below the coesite-bear-ing unit, J. Geol. 111 (2003) 57^70.

[81] A.I. Chemenda, M. Mattauer, A.N. Bokun, Continentalsubduction and a mechanism for exhumation of high-pressure metamorphic rocks: new modelling, ¢eld datafrom Oman, Earth Planet. Sci. Lett. 143 (1996) 173^182.

[82] G.T. Roselle, M. Engi, Ultra high pressure (UHP) ter-rains: lessons from thermal modelling, Am. J. Sci. 302(2002) 410^441.

[83] N. Merveilleux du Vignaux, L. Fleitout, Stretching andmixing of viscous blobs in Earth’s mantle, J. Geophys.Res. 106 (2001) 30893^30908.

[84] P.J. O’Brien, J. Ro«tzler, High-pressure granulites: forma-

tion, recovery of peak conditions and implications fortectonics, J. Metamorph. Geol. 21 (2003) 3^20.

[85] S.L. Hwang, P.Y. Shen, H.T. Chu, T.F. Yui, Nanometer-size K-PbO2-type TiO2 in garnet: A thermobarometer forultrahigh-pressure metamorphism, Science 288 (2000)321^324.

[86] E. Schma«dicke, W.F. Mu«ller, Unusual exsolution phe-nomena in omphacite and partial replacement of phengiteby phlogopite+kyanite in an eclogite from the Erzgebirge,Contrib. Mineral. Petrol. 139 (2000) 629^642.

[87] T.V. Gerya, B. Sto«ckhert, A.L. Perchuk, Exhumation ofhigh-pressure metamorphic rocks in a subduction chan-nel: A numerical simulation, Tectonics 21 (2002) 10.1029/2002TC001406

Christian Chopin, 48, has a CNRSresearcher position at the Ecolenormale supe¤rieure, Paris. Hislong-standing interests are in thephase equilibria, mineral chemistryand transformations of rocks underhigh-pressure conditions, and theimplications they have for moun-tain-building and geological pro-cesses in general. This led him tothe ¢rst discovery of metamorphic

coesite. Trying to maintain a balance between experimentaland ¢eld-oriented petrological work, he is a ¢rm believer inthe unsurpassable value of observational data: Nature as thelargest laboratory.



ultrahigh-pressuremetamorphism ... · frontiers ultrahigh-pressuremetamorphism:...

Documents