UHT_metamorphism

7/29/2019 UHT_metamorphism

1/6

ELEMENTS, VOL . 7, PP. 235240 AUGUST 2011235

1811-5209/11/0007-0235$2.50 DOI: 10.2113/gselements.7.4.235

How Does the ContinentalCrust Get Really Hot?

INTRODUCTION

Evidence for the pressuretemperature (PT) conditionsunder which Earths crust has generated large volumes ofmagma is provided by metamorphic rocks that representthe solid residue of partial melting. Many of these rockspreserve minerals formed at moderate pressures and veryhigh temperatures, conditions consistent with substantialpartial melting of continental crust. Although originallyregarded as isolated anomalies, there is increasing evidencethat these ultrahigh-temperature (UHT) conditions were

attained repeatedly in time and space. Our ability to quan-tify this record has increased dramatically in recent yearswith improved thermodynamic constraints on mineralPTstability that allow us to derive robust PTdata for meta-morphic rocks. These data can then be compared withgeothermal gradients predicted using mathematical modelsthat describe the thermal behaviour of continental crustin simple tectonic settings. While standard numericalmodels for mountain building can reproduce the condi-tions recorded by most metamorphic rocks, UHT metamor-phism is difficult to replicate. In this article we examine anumber of heat sources that might account for theseextreme temperatures.

RECOGNITION OF UHT METAMORPHISM

Metamorphic conditions are classified using metamorphicfacies, which arePTfields defined by distinctive mineralassemblages (FIG. 1A). UHT conditions lie at the high-temperature extreme of the granulite facies and weredefined by Harley (1998) as temperatures in excess of 900 C

and pressures of 0.7 to 1.3 GPa.Brown (2006) proposed a revisedupper pressure limit equivalent toa P/T gradient of 750 C GPa1,close to the kyanitesillimanitereaction boundary (FIG. 1A). Thelower temperature limit of 900 Cis somewhat arbitrary, but it placesthe onset of UHT metamorphismbeyond the conditions at which

many crustal rocks start to melt, aprocess that represents a signifi-cant barrier to the attainment ofhigher temperatures.

Recognition of UHT metamor-phism is problematic because few rocks develop diagnosticminerals at these conditions and widespread chemicalequilibration during cooling makes temperature estimatesbased on mineral composition unreliable. Although rarein metamorphic belts, Mg-rich mudstone does developdiagnostic mineral assemblages at UHT conditions, mostnotably sapphirine + quartz (FIG. 1B), but also orthopy-roxene + sillimanite + quartz, spinel + quartz, and osumilite+ garnet (Harley 2008). However, the stability of theseassemblages is highly sensitive to minor chemical compo-

nents and the redox state of the rock, making them unreli-able indicators of UHT conditions. Thus sapphirine +quartz is stable down to 850 C in highly oxidised systems(Taylor-Jones and Powell 2010), and components such asFe3+, Cr, Zn and Ti can extend spinel + quartz stability tobelow 900 C (Harley 2008). Other evidence for UHT condi-tions includes high Zr contents in rutile; aluminous ortho-pyroxene coexisting with garnet, although the presence ofFe3+ can again lead to temperature overestimates; andextensive solid solution in feldspar and pyroxene resultingin mesoperthite and pigeonite exsolution, although caremust be taken to ensure these are not igneous relics (Harley2008).

Temperature estimates based on the distribution of Fe and

Mg between different minerals are reset on cooling fromUHT conditions, but calculations that correct for this effectreveal a continuum in estimated peak temperature fromthe lower granulite facies into the UHT field (F IG. 1A;Pattison et al. 2003). This suggests that UHT metamor-phism occurs in similar tectonic settings to lower-temper-ature granulite metamorphism and is not a result ofanomalous processes, a conclusion supported by thediscovery of UHT metamorphism at more than 40 localitiesworldwide (Kelsey 2008) with ages spanning the last 3000million years (Brown 2006). Inferred geothermal gradientsbeneath the Himalaya are consistent with UHT conditionsat depth (Hacker et al. 2000), implying that the apparentscarcity of Phanerozoic UHT metamorphism reflects thetime taken for deep crustal rocks to reach the surface and

There is widespread evidence that ultrahigh temperatures of 9001000 C

have been generated in the Earths crust repeatedly in time and space.

These temperatures were associated with thickened crust in collisional

mountain belts and the production of large volumes of magma. Numerical

modelling indicates that a long-lived mountain plateau with high internal

concentrations of heat-producing elements and low erosion rates is the most

likely setting for such extreme conditions. Preferential thickening of already-

hot back-arc basins and mechanical heating by deformation in ductile shear

zones might also contribute to elevated temperatures.

KEYWORDS: metamorphism, ultrahigh temperature, heat production, mountain

belt, thermal modelling

Chris Clark1, Ian C. W. Fitzsimons1, David Healy2and Simon L. Harley3

1 The Institute for Geoscience Research (TIGeR), Department ofApplied Geology, Curtin Univers ity, GPO Box U1987Perth WA 6845, AustraliaE-mail: [email protected]

2 School of Geosciences, Kings College, University of AberdeenAberdeen, AB24 3UE, UK

3 Grant Institute of Earth Science, The University of EdinburghEdinburgh, EH9 3JW, UK


2/6

ELEMENTS AUGUST 2011236

not an absence of UHT conditions. The ages of UHT meta-morphism show some correlation with periods of super-

continent assembly (Brown 2006), suggesting that UHTmetamorphism occurs during continental collision or thatUHT terranes associated with collision are more likely tobe preserved than those formed in other settings.

WHAT DRIVES UHT METAMORPHISM?

While there is widespread agreement that UHT rocks occurin many metamorphic belts, there is no consensus on theheat source for such extreme temperatures. Pervasive defor-mation and widespread chemical and textural re-equili-bration at high temperature have destroyed much of thefield and petrological evidence for how UHT conditionsare achieved (FIG. 1C). Some constraints are provided bythe exposure of ancient UHT terranes at the surface of crustthat is now of normal thickness and by mineral reactions

in UHT rocks indicating that peak conditions are typicallyfollowed by decompression. These relationships suggestthat UHT rocks form in the mid levels of thickened crust,consistent with metamorphism during continental colli-sion. However, some terranes, including the NapierComplex of Antarctica, record prolonged cooling from UHTconditions at near constant pressure, implying that theseareas were in isostatic equilibrium during and after meta-morphism. Another important observation is that UHTmetamorphism is typically not associated with the intru-sion of substantial mafic or ultramafic rock, ruling outmantle-derived magma as a major heat source.

Given the limited geological evidence, the best quantitativeconstraints on the cause of UHT metamorphism come fromnumerical predictions of temperature var iations in simple

tectonic settings. Two-dimensional numerical models areincreasingly used to reproduce the evolution of mountainbelts (Jamieson and Beaumont 2010), but these require anunderstanding of regional-scale structure and rock distribu-tion that is lacking for deeply eroded UHT terranes. Forthis reason we investigate the factors that promote, or limit,UHT metamorphism in simple models of crustal thickeningusing the one-dimensional heat flow equation:

, (1)

where Tis temperature, tis time, z is depth, is thermaldiffusivity, is density, cp is specific heat capacity, u is

vertical transport velocity relative to Earths surface (rateof burial, or exhumation if negative), andArad,Amech andAchem are rates of heat production per unit volume by radio-active decay, mechanical deformation, and chemical reac-tion, respectively. The first term on the right-hand side ofequation 1 describes conductive heat flow between rocksof different temperature, the second quantifies vertical heattransport by advection (heat carried with rocks movingrelative to Earths surface), and the third describes threemechanisms that create heat (and also consume heat inthe case ofAchem). One-dimensional models cannot accountfor lateral movement of heat or rock, which will be signifi-cant close to plate boundaries and other large-scale dippingstructures in mountain belts, but they do provide a first-order assessment of potential heat sources for UHT meta-morphism, particularly for rocks located some distancefrom plate boundaries. The use of one-dimensional modelsalso maximizes the likelihood of replicating UHT condi-tions, given that lateral heat flow will move heat away fromhigh-temperature rocks.

Important parameters in our models include the thick-nesses of the crust and lithosphere before and after thick-ening, the temperaturedepth profile before thickening,the geometry of thickening (e.g. homogenous deformationor thrust stacking), the erosion rate, the values of, andcp, the magnitude of heat flow from the mantle into thelithosphere, the magnitudes and spatial and/or temporaldistributions ofArad,Amech and Achem, and the magnitudeof heat advection by magma into or within the crust. Theseare constrained to varying degrees by geologic and experi-

mental data, and there has been considerable uniformityin values used over the last 30 years (England andThompson 1984); however, recent studies have questionedsome of these assumptions. In particular, new experimentsshow that has a much stronger temperature dependencethan thought previously, with values at lower crustaltemperatures being about 50% of those used in mostmodels (Whittington et al. 2009). This reduces the ratesof conductive heat flow, allowing regions of high radioac-tive, mechanical or chemical heat production to attainhigher temperatures, and we adopt temperature-dependentvalues of in our models. Unlike many studies that assumeAchem to be negligible, we incorporate a term for the heat

FIGURE 1(A) PTconditions of UHT and other styles of meta-morphism, from Brown (2007). Red circles are PT

estimates for granulite facies rocks (Harley 1998; Pattison et al.2003); their distribution shows that UHT metamorphism is contin-uous with the granulite facies. Field abbreviations: A, amphibolitefacie s; BS, blueschist facies ; E-HPG, medium-T ec logite high- Pgranulite facies; G, granulite facies; GS, greenschist facies; UHP,ultrahigh-pressure metamorphism. Mineral abbreviations: Ab,

albite; Coe, coesite; Jd, jadeite; Ky, kyanite; Qtz, quartz; Sil, silli-manite. (B AND C) Mineralogical indicators and field relationshipsof UHT metamorphism in the Napier Complex, Antarctica.(B) Sapphirine (Spr) + orthopyroxene (Opx) + quartz (Qtz) assem-blage. Opx contains up to 10 wt% Al2O3. (C) Interlayered sequenceof UHT metamorphic rocks including quartzofeldspathic gneiss (a),garnet-sillimanite metapelite with Spr-bearing layers (b),metatonalite (c) and granodioritic gneiss (d)

A B C


3/6


consumed by melting reactions, which could be significantunder UHT conditions given the potential for extensivemelt generation.

We investigate three heat sources that have been proposedto account for UHT conditions during continentalcollision:

Elevated radioactive heat production in thickened crust

Increased mantle heat input to back-arc basins

Mechanical heating in ductile shear zones

Another possible heat source is the addition of mantle-derived magma to the crust, but we ignore this given thelack of evidence for significant mafic magmatism in UHTterranes. We also ignore the effects of magma movementwithin the crust because this does not add extra heat tothe system and cannot, on its own, drive UHT metamor-phism. Partial melting could, however, play an importantrole in limiting crustal heat production and enabling thecrust to attain higher temperatures in later metamorphicevents, and we discuss this at the end of the article.

Radioactive Heat ProductionRadioactive decay of U, Th and K has long been recognisedas an important heat source in continental crust, withtypical heat-production values of 0.13.0 W m-3 (Vil etal. 2010). The influence of radioactive heating duringmountain building depends on the initial distribution ofheat-producing elements (generally assumed to be greaterin the upper c rust due to its more felsic composition) andhow this distribution is modified during collision, includingthe addition of radioactive material by thickening and its

loss by erosion (Jamieson et al. 1998; Sandiford andMcLaren 2002). We investigate these parameters using aone-dimensional model in which crust, comprising a20 km thick upper radioactive layer (Arad = 2.0 W m3)and a 15 km thick lower non-radioactive layer, is instan-taneously doubled in thickness by homogenous deforma-tion (FIG. 2). There is no thickening of mantle lithosphere,consistent with its partial detachment or subductionduring collision, and, following modelling studies of theEuropean Alps (England and Thompson 1984), erosion

pre-thickening post-thickening

Arad=0W m-3

upper crust

lower crust

20

km

40km

15km

30km

upper crust

lower crust

mantle

mantle

Base of lithosphere:

T = 1300 C

Top of lithosphere: T = 0 C Top of lithosphere: T = 0 C

lithosphere=150km

lithosphere=185km

MOHO at 35 km

MOHO at 70 km

Arad=0W m-3

Arad=0W m-3

Arad=2W m-3

Arad=2W m-3

post-erosion

5km

30km

upper crust

lower crust

mantle

Top of lithosphere: T = 0 C

lithosphere=150kmMOHO at 35 km

Arad=0W m-3

Arad=2W m-3

Arad=0W m-3

Arad=0W m-3

Erosion rate = 0.7 mm y-1

0 300 600 900 1200

10

20

30

40

50

60

70

Temperature (C)

D

epth(km)

0 My (before thickening)0 My (after thickening) 0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

Pressure(GPa)

PTtfor initial depth 30 km



0 300 600 900 1200Temperature (C)

UHT UHT

20 My

40 My

60 My

80 My

100 My

120 My

A

B C


T = 1300 C


T = 1300 C

FIGURE 21-D thermal model for instantaneous doubling ofcrustal thickness by homogenous deformation, with

erosion at 0.7 mm y-1 starting 20 My after thickening. (A) Modelgeometry and Arad immediately before thickening, immediatelyafter thickening and 120 My after thickening when erosion hasreturned crust to its original thickness. (B) Evolution of thegeothermal gradient with time, showing gradients immediatelybefore and after thickening and then at 20 My intervals. (C) PTt

particle paths for rocks buried to 30, 50 and 70 km depths onthickening. All models solve equation 1 by finite difference withfixed Tat the surface (0 C) and the base of the lithosphere(1300 C). The latent heat of melting is 320 kJ kg-1 (see FIG. 4C), andthe T-dependent expressions for and cp are from McKenzie et al.(2005) for the mantle and from Mottaghy et al. (2008) for thecrust. Boxes in (B) and (C) mark UHT conditions.

A

B C


4/6


commences 20 million years (My) after thickening at a rateof 0.7 mm y1. These last aspects of the model set-up maxi-mise the likelihood of UHT metamorphism. The resultsare illustrated as a series of geotherms showing how thetemperaturedepth profile changes with time (FIG. 2B) andas particle paths depictingPThistories of rocks originatingat different depths in the mountain belt (FIG. 2C). Thegeothermal profile cools instantaneously as thickeningtransports rock to greater depths, then heats up as the extraradioactivity in thickened crust takes effect, and finally

cools as erosion removes heat-producing crust at thesurface. Although a thermal anomaly develops near thebase of the radioactive upper crust 4060 My after thick-ening, this temperature falls 300 C short of UHT condi-tions. Particle paths show an initial 20 My period ofheating at constant pressure and then decompression asthe onset of erosion moves rocks towards the surface, buttemperatures at mid-crusta l levels never exceed 600 C.

Several authors have suggested that elevated mid-crustaltemperatures result from higher-than-normal radioactiveheating (e.g. Chamberlain and Sonder 1990). We investi-gate this effect in FIGURE 3A, which compares thermal histo-ries for the Alpine model of FIGURE 2 using different valuesofArad. For each value we plot the temperature history ofa rock at the base of the heat-producing upper crust (40 km

depth immediately after thickening). Peak temperaturesoccur 5560 My after thickening, and an Aradvalue of atleast 3.5 W m3 is needed to achieve UHT conditions.Although such values are inferred at least locally for UHTterranes (e.g. Andreoli et al. 2006), our model underesti-mates the Arad needed for UHT metamorphism if heat isremoved by lateral flow, as might be expected in narrowAlpine mountain belts, or if erosion starts immediatelyafter collision.

The effect of erosion is illustrated in F IGURE 3B, whichcompares the thermal history of rocks buried initially to40 km for various erosion rates starting immediately afterthickening and for an upper-crustalArad value of 3 W m3.Despite above-average heat production, only the lowesterosion rate of 0.05 mm y1 allows UHT conditions to beattained. This result is consistent with metamorphism inthe middle of a wide mountain plateau in a Himalayan-style mountain belt (Lal et al. 2004). Our model suggeststhe plateau must be long-lived (120 My) to attain UHTconditions, unless the upper-crustalArad value is substan-tially higher than 3 W m3. Other studies have reproducedUHT conditions 90 My after plateau formation with anupper-crustalArad value of only 2 W m3, but these studiesignored latent heat of melting and used either a muchthicker radioactive upper crust (60 km; McKenzie andPriestley 2008) or a heat-producing lower crust (Arad =0.75 W m3; Jamieson and Beaumont 2010).

Mantle Heat in Back-Arc BasinsBack-arc basins are regions of thinned continental litho-sphere with high mantle heat flow and Moho temperaturesas high as 800 C (Currie and Hyndman 2006). If thickenedin response to continental collision, temperatures in thisalready hotter-than-normal crust are augmented byincreased radioactive heat production (Brown 2006). Weinvestigate this with a one-dimensional model that startswith 20 km thick heat-producing upper crust and 15 kmthick non-radioactive lower crust, but the total lithosphericthickness is only 5070 km, typical of back-arc basins, andthe upper-crustalArad value is relatively low (1.5 W m3).The thickness of both crust and mantle lithosphere is

instantaneously doubled, and erosion commences imme-diately at 0.7 mm y1. FIGURE 3C compares the thermalhistory of rocks buried initially to 40 km for initial litho-spheric thicknesses of 50, 60 and 70 km. Peak temperaturesare attained 3035 My after thickening, as for normal litho-spheric thicknesses and the same erosion rate (FIG. 2B), butas in other studies of back-arc thickening (Thompson etal. 2001) none of our models reach UHT conditions.Increasing Arad, decreasing erosion rates, or adding heatfrom magmas would raise peak temperature, but we againemphasise that our models overestimate temperaturebecause they assume no lateral heat flow.

Mechanical Heating in Shear ZonesThere has been much discussion of whether mechanical

heating is a negligible or significant contributor to meta-morphic temperatures (e.g. Nabelek et al. 2010). Themagnitude ofAmech is given by the product of strain rate() and shear stress (), where the latter is a measure of rockstrength. While there is reasonable agreement on likelystrain rates during mountain building, maximum shearstress is strongly dependent on rock type and decreasesmarkedly with increasing temperature. We investigateAmech using a one-dimensional model with the same initialconditions as in FIGURE 2, but we double crustal thicknessinstantaneously by stacking one 35 km thick crustal blockon top of another. The shear zone between them is 3 kmwide and is active for 50 My with a displacement velocityof 30 mm y1. Erosion commences 20 My after thickening

200

400

600

800

1000

020 40 60 80 100 120

Time (My)

Tem

peratu

re(C)

1.0

1.5

2.0

2.5

3.0

3.5

4.01200

Varying heat production (W m-3)

A

200

400

600

800

1000

020 40 60 80 100 120

Time (My)

Tem

peratu

re(C)

0.05

0.35

0.70

1.00

Varying erosion rate (mm y-1)

B

200

400

600

800

1000

020 40 60 80 100 120

Time (My)

Tempe

rature(

C)

50

60

70

Varying initial lithospherethickness (km)

C

FIGURE 3(A) Ttevolution of rocks at 40 km depth immediatelyafter thickening for different values of upper crustal

Arad. All other parameters are identical to those in FIGURE 2. (B) Ttevolution of rocks at 40 km depth immediately after thickening fordifferent erosion rates starting immediately after thickening. Theupper crustal Arad value is 3 W m3 and other parameters are iden-tical to those in FIGURE 2. (C) Ttevolution of rocks buried to 40 kmdepth by thickening of a back-arc basin with varying initial litho-spheric thickness. The crustal geometry is as in F IGURE 2, but theupper crustal Aradvalue is 1.5 W m

3 and the total lithosphericthickness is 50, 60 or 70 km. Instantaneous homogenous deforma-tion doubles the thickness of both crust and lithosphere, and isfollowed immediately by erosion at 0.7 mm y1.

A B C


5/6


at 0.7 mm y1, but we keep the shear zone at a fixed depth.FIGURE 4A compares the thermal history of rocks at 40 kmdepth immediately after thrusting for four values of.Although varies with temperature, strain rate and rocktype (FIG. 4B), we use temperature-insensitive values hereto highlight the effect of different rock strengths, but isset to zero once the temperature reaches 750 C to reflectthe substantial drop in strength as rocks begin to melt.

Peak temperatures of 700 C are attained 50 My afterthrusting for zeroAmech(= 0), compared to 600 C in thecomparable homogenous deformation model (FIG. 2),reflecting the deep burial of radioactive upper crust bythrust stacking; also, higher temperatures are achieved as increases. Each rock in FIGURE 4A moves up into the centreof the shear zone 27 My after thrusting because of erosion,and shear zone temperature at this time is 700 C for =10 MPa, 800 C for = 30 MPa, and 950 C for = 100 MPa.Felsic rocks can have shear strengths of 100 MPa at300650 C (FIG. 4B), but these values decrease to 30 MPaat 700750 C. Strengths drop much more with as little as10% partial melting or a switch of deformation mechanismfrom dislocation creep to diffusion creep, which ispromoted by a decrease in grain size (Franek et al. 2011).This makes it unlikely that deformation of felsic rocksproduces significant heat as UHT conditions are approached.

Rocks such as clinopyroxenite retain sufficient strength at700900 C to generate UHT conditions by mechanicalheating in less than 20 My, as shown by Nabelek et al.(2010), but deformation is likely to be focussed in weakerfelsic rocks that dominate the upper and mid crust. Thusalthough mechanical heating could rapidly increase mid-crustal temperature in the early, cooler stages of collision,its contribution diminishes greatly as temperatures rise.

THE ROLE OF CRUSTAL MELTING

UHT metamorphism occurs at temperatures above theonset of partial melting in most crustal rocks, and meltingis an endothermic process that consumes heat and bufferstemperature (Stwe 1995). FIGURE 4C illustrates the effectof latent heat of melting (L) on rocks buried to 40 km after

thickening in the model of FIGURE 2, but with a higherAradvalue (3.5 W m3) to ensure that the rocks attain UHTconditions. Peak temperatures forL = 100, 320 and 500 kJkg1 are respectively 10, 35 and 50 C lower than those for

zero latent heat. We use 320 kJ kg 1 in FIGURES 24B, a widelyacceptedvalue for felsic rocks. The heating curve forArad =3.5 W m3 in FIGURE 3A would have reached a peak of980 C for L = 0, rather than 950 C, showing that UHTconditions are attained more readily if melting is suppressed.Melt loss is one way to limit partia l melting during a latermetamorphism, and this wi ll also strengthen the crust andincreaseAmech, suggesting that UHT conditions are easierto achieve in terranes that have experienced multiplethermal events. This effect should be offset by depletion

of heat-producing elements, given that U, Th and K parti-tion into melt, but the behaviour of heat-producingelements during partial melting is not fully understood.Metamorphic rocks that have lost melt can be enriched inU and Th, while many granites derived by high-tempera-ture crustal melting have lower-than-expected U and Thcontents (Villaseca et al. 2007).

SUMMARY AND FUTURE DIRECTIONS

UHT metamorphism is characteristic of the middle to lowercrust in many collisional orogens, but the heat sourceresponsible for generating temperatures in excess of 900 Cis controversial. It is likely that elevated concentrations ofheat-producing elements are a critical component, coupledwith crustal thickening to form a wide plateau that must

survive long enough for enhanced radioactive heating tosubstantially raise crustal temperatures. In such cases, theretrograde decompression typical of UHT terranes reflectsplateau collapse once convergent tectonic forces can nolonger sustain its gravitational potential. Mechanicalheating in shear zones and preferential thickening of back-arc basins with al ready-elevated geothermal gradients cancontribute to higher temperatures, part icularly early in themetamorphic history, but will not usually lead to UHTconditions without above-average radioactive heat produc-tion. Irrespective of heat source, UHT conditions areattained more easily in terranes that have already under-gone at least one episode of partial melting, unless thisalso results in reduced radioactive heat production.

Future studies must involve basic fieldwork to constrain

the regional geometry of UHT terranes and allow construc-tion of two-dimensional thermo-mechanical models thatprovide more robust constraints than the models used here.This work should include the systematic collection of

200

400

600

800

020 40 60 80 100 120

Time (My)

em

peratur e

(C)

1000

0

10

30100

Varying rock strength (MPa)

A

101

102

103

104

105

0200 400 600 800 1000

Temperature (C)

Shear

Streng

th(MPa

)

Strain rate

= 3x10-13 s-1

Granite (Carter

et al. 1981)

Quartz (Hirth

et al. 2001)

Quartz (Rutter

& Brodie 2004)

Clinopyroxenite (Kirby

& Kronenberg 1984)

B

-40

-30

-20

-10

0

-5020 40

Time (My)

Tempera

ture(C

)

100

320

500

Varying latent heat

of melting (kJ kg-1)

60

C

FIGURE 4(A) Ttevolution of rocks buried to 40 km by instan-taneous stacking of one 35 km thick crustal block on

top of another along a 3 km wide shear zone that generates heatfor four dif ferent rock st rengths (). The initial geometry and Aradvalues are from FIGURE 2A. The shear zone is active for 50 My, with avelocity of 3 cm yr1(shear strain rate = 3 10 13 s-1). Values of aretaken as constant at temperatures below 750C, but are set to zeroat higherTto simulate melt weakening. Erosion at 0.7 mm y-1 starts20 My after thickening. (B) Plot of shear strength against Tfordifferent rock types undergoing dislocation creep at a strain rate of3 10 -13 s-1 (after Nabelek et al. 2010), showing that rock strengthdecreases markedly on heating. (C) Ttevolution of rocks buried to

40 km by homogeneous thickening for three values of latent heatof melting (L). Tis the difference between the actual Tfor aselected value ofL and Tobtained ifL = 0. The model set-up isfrom F IGURE 2, apart from the upper-crustal Arad value (3.5 W m

3).We assume that the consumpt ion of L increases linearly with meltfract ion over the melt ing interval , and melt f raction increaseslinearly with Taccording to the best-fit line through the experi-mental data for natural metapelite. Latent heat is also released oncooling in our model, reducing retrograde cooling rates, but this isless realistic given that melt crystallization is not a simple reversalof melting and that some partial melt will escape to highercrustal levels.

A B C


6/6


REFERENCESAndreoli MAG, Hart RJ, Ashwal LD,

Coetzee H (2006) Correlations betweenU, Th content and metamorphic gradein the western Namaqualand belt,South Africa, with implications forradioactive heating of the crust. Journalof Petrology 47: 1095-1118

Brown M (2006) Duality of thermalregimes is the distinctive characteristicof plate tectonics since the Neoarchean.Geology 34: 961-964

Brown M (2007) Metamorphic conditionsin orogenic belts: A record of secularchange. International Geology Review49: 193-234

Carter NL, Anderson DA, Hansen FD,Krans RL (1981) Creep and c reeprupture of granitic rocks. In: Carter NL,Friedman M, Logan JM, Sterns DW (eds)Mechanical Behavior of Crustal Rocks.American Geophysical UnionGeophysical Monograph 24, pp 61-82

Chamberlain CP, Sonder LJ (1990) Heat-producing elements and the thermaland baric patterns of metamorphicbelts. Science 250: 763-769

Currie CA, Hyndman RD (2006) Thethermal structure of subduction zoneback arcs. Journal of GeophysicalResearch 111: B08404,doi:10.1029/2005JB004024

England PC, Thompson AB (1984)Pressuretemperaturetime paths ofregional metamorphism I. Heat transferduring the evolution of thickeningcontinental crust. Journal of Petrology25: 894-928

Franek J, Schulmann K, Lexa O, Ulrich S,tpsk P, Haloda J, Tcov P (2011)Origin of felsic granulite microstructureby heterogeneous decomposition ofalkali feldspar and extreme weakening

of orogenic lower crust during theVariscan orogeny. Journal ofMetamorphic Geology 29: 103-130

Hacker BR, Gnos E, Ratschbacher L, GroveM, McWilliams M, Sobolev SV, Wan J,Zhenhan W (2000) Hot and dry deepcrustal xenoliths from Tibet. Science287: 2463-2466

Harley SL (1998) On the occurrence andcharacterization of ultrahigh-tempera-ture metamorphism. In: Treloar PJ,OBrien PJ (eds) What DrivesMetamorphism and MetamorphicReactions? Geological Society ofLondon Special Publication 138,pp 81-107

Harley SL (2008) Refining thePTrecordsof UHT crustal metamorphism. Journalof Metamorphic Geology 26: 125-154

Hirth G, Teyssier C, Dunlap JW (2001) Anevaluation of quartzite flow laws basedon comparisons between experimen-tally and naturally deformed rocks.International Journal of Earth Sciences90: 77-87

Jamieson RA, Beaumont C (2010) Coevalthrusting and extension during lowercrustal ductile flow implications forexhumation of high-grade metamorphicrocks. Journal of Metamorphic Geology29: 33-51

Jamieson RA, Beaumont C, Ful lsack P, LeeB (1998) Barrovian regional metamor-phism: wheres the heat? In: Treloar PJ,OBrien PJ (eds) What DrivesMetamorphism and MetamorphicReactions? Geological Society ofLondon Special Publication 138,pp 23-51

Kelsey DE (2008) On ultrahigh-tempera-ture crustal metamorphism. GondwanaResearch 13: 1-29

Kirby SH, Kronenberg AK (1984)Deformation of clinopyroxenite:Evidence for a transition in flow mecha-nisms and semibrittle behavior. Journalof Geophysical Research 89(B5):3177-3192

Lal D, Harris NBW, Sharma KK, Gu Z ,Ding L, Liu T, Dong W, Caffee MW, JullAJT (2004) Erosion history of theTibetan Plateau since the last intergla-cial: constraints from the first studies ofcosmogenic 10Be from Tibetan bedrock.Earth and Planetary Science Letters 217:33-42

McKenzie D, Priestley K (2008) The influ-ence of lithospheric thickness variationson continental evolution. Lithos 102:1-11

McKenzie D, Jackson J, Priestley K (2005)Thermal structure of oceanic and conti-nental lithosphere. Earth and PlanetaryScience Letters 233: 337-349

Mottaghy D, Vosteen H-D, SchellschmidtR (2008) Temperature dependence ofthe relationship of thermal diffusivityversus thermal conductivity for crystal-line rocks. International Journal ofEarth Sciences 97: 435-442

Nabelek PI, Whittington AG, HofmeisterAM (2010) Strain-heating as a mecha-nism for partial melting and ultrahightemperature metamorphism in conver-gent orogens: Implications of tempera-

ture-dependent thermal diffusivity andrheology. Journal of GeophysicalResearch 115: B12417,doi:10.1029/2010JB007727

Pattison DRM, Chacko T, Farquhar J,McFarlane CRM (2003) Temperatures ofgranulite-facies metamorphism:Constraints from experimental phaseequilibria and thermobarometry

corrected for retrograde exchange.Journal of Pet rology 44: 867-900

Rutter EH, Brodie KH (2004)Experimental intracrystalline plasticflow in hot-pressed synthetic quartziteprepared from Brazilian quartz crystals.

Journal of St ructural Geology 26:259-270

Sandiford M, McLaren S (2002) Tectonicfeedback and the ordering of heatproducing elements within the conti-nental lithosphere. Earth and PlanetaryScience Letters 204: 133-150

Stwe K (1995) Thermal buffering effectsat the solidus. Implications for theequilibration of partially melted meta-morphic rocks. Tectonophysics 248:

39-51Taylor-Jones K, Powell R (2010) The

stability of sapphirine + quartz: calcu-lated phase equilibria in FeOMgOAl2O3SiO2TiO2O. Journal ofMetamorphic Geology 28: 615-633

Thompson AB, Schulmann K, Jezek J,Tolar V (2001) Thermally softenedcontinental extensional zones (arcs andrifts) as precursors to thickenedorogenic belts. Tectonophysics 332:115-141

Vil M, Fernndez M, Jimnez-Munt I(2010) Radiogenic heat production vari-ability of some common lithologicalgroups and its significance to litho-spheric thermal modeling.Tectonophysics 490: 152-164

Villaseca C, Orejana D, Paterson BA(2007) ZrLREE rich minerals inresidual peraluminous granulites,another factor in the origin of low ZrLREE granitic melts? Lithos 96: 375-386

Whittington AG, Hofmeister AM, NabelekPI (2009) Temperature-dependentthermal diffusivity of the Earths crustand implications for magmatism.Nature 458: 319-321

radioactive heat-production data, coupled with petrologicaland geochemical studies to establish the extent and impli-cations of melt production and melt loss. Geochronologicalwork should constrain the onset of collision as well as peakUHT metamorphism, and it should determine whether thetime taken to reach UHT conditions is consistent withradioactive heating in a long-lived plateau or with morerapid heating mechanisms. It is also important to establishwhether UHT conditions develop preferentially in terranesthat have undergone prior metamorphism and melt loss.

UHT metamorphism must be linked to the generation oflarge volumes of magma and chemical differentiation ofcontinental crust. An important question is whether therepeated occurrence of UHT metamorphism through Earth

history reflects multiple episodes of voluminous magmageneration, or whether most melt production in UHTterranes occurred in unrelated earlier events that primedrocks for later UHT metamorphism.

ACKNOWLEDGMENTS

We acknowledge fruitful discussions of high-temperaturemetamorphism with many colleagues, including D. Kelsey,M. Brown, M. Hand, R. White, N. Kelly, M. Santosh and A.Collins. We thank R. Jamieson, O. Lexa and E. Sawyer for

perceptive comments on an earlier version of this manu-script. Our work is supported by the Australian ResearchCouncil (Discovery Program Grant DP0664679).

Documents

UHT_metamorphism