sign in sign up
<<
Home , Blueschist, Eclogite, Metamorphic facies

The importance of —» dehydration reactions in subducting oceanic

SIMON M. PEACOCK Department of , Arizona State University, Tempe, Arizona 85287-1404 and Institut für Mineralogie und Pétrographie, ETH Zentrum, CH-8092 Zürich, Switzerland

ABSTRACT that may be stable in subducting to depths of—100 km, including chlorite (Delany and Helgeson, 1978; Bebout, 1991), phlog- The metamorphic evolution and dehydration of subducting oceanic opite (Wyllie, 1973; Fyfe and McBirney, 1975), phengite (Nicolls and crust may be predicted by combining calculated pressure-temperature others, 1980), and (Tatsumi, 1989; Peacock, 1990b). (P-T) paths with a model of metabasalt phase equilibria. In steady-state The research presented in this paper differs from most previous zones with high rates of shear heating, the upper parts of the investigations in two important respects. (1) Rather than using end- subducting oceanic crust progress through the —» amphib- member dehydration reactions, I use the metamorphic- concept olite —» —»eclogite facies, whereas lower parts of the subducting to determine the important dehydration reactions that occur in a oceanic crust progress through the blueschist —» eclogite facies. In of oceanic- composition. In P-T space, metamorphic-facies steady-state subduction zones with moderate rates of shear heating, most boundaries are broader than end-member univariant reaction curves, of the subducting oceanic crust passes through the blueschist —» eclogite but I believe that the metamorphic-facies concept provides a more transition. In steady-state subduction -zones with low rates of shear realistic picture of the continuous dehydration reactions that occur in heating, the entire subducting oceanic crust lies within the blueschist subducting oceanic crust. (2) In addition to presenting cross sections facies to depths greater than 70 km. For oceanic crust containing 1-2 through subducting oceanic crust, I superpose calculated P-T paths on wt% H20, dehydration will not begin until the onset of eclogite- or a diagram to illustrate and dehy- -facies metamorphism, depending on the P-T path. For dration more clearly in subducting oceanic crust as a function of time, many subduction zones, the most important dehydration reactions in the position in the slab, and subduction-zone parameters. subducting oceanic crust occur at the blueschist —> eclogite facies tran- I begin by presenting a metamorphic-facies diagram and an es- sition associated with the breakdown of (or clinozoisite), glau- timate derived from the literature of the bulk composition of oceanic cophane, and chlorite. Large amounts of H20 released by blueschist -» crust. For each metamorphic facies, I calculate a modal mineralogy eclogite dehydration reactions could trigger partial melting in the over- from which the maximum amount of bound H20 may be estimated. lying wedge and may play a crucial role in the generation of arc Using Molnar and England's (1990) analytical expressions, I calculate . a range of possible P-T paths for subducting oceanic crust for different rates of shear heating. The overlaying of calculated P-T paths on the INTRODUCTION metamorphic facies diagram limits the maximum H20 content of sub- ducting oceanic crust, and important dehydration reactions that occur The geochemistry of most arc magmas indicate that partial melt- in metabasaltic compositions may be identified. As is shown below, ing occurs in the mantle wedge above the subducting slab as a result most P-T paths calculated for subducting oceanic crust pass through

of the infiltration of slab-derived, H20-rich fluids (Gill, 1981). How is the blueschist -» eclogite facies in contrast to higher-temperature H20 transported in the subducted slab to depths of 100-150 km? What greenschist —» amphibolite —» eclogite paths proposed by previous hydrous are stable in subduction zones? The answers to these workers (for example, Wyllie and Sekine, 1982; Wyllie, 1988). questions lie in determining the position of dehydration reactions in subducting oceanic crust. Determining the amount and position of DEVELOPMENT OF THE THERMAL-PETROLOGIC MODEL HzO released from subducting oceanic crust as a function of the P-T path and other variables is an important step toward understanding Bulk Composition of the Oceanic Crust subduction-zone processes, arc genesis, and the recycling of H20 into the mantle. In this contribution, calculated P-T paths are Average oceanic crust consists of —7 km of basaltic and gabbroic combined with metabasalt phase equilibria in order to depict the met- rocks. The composition of the oceanic crust is dominated by tholeiitic amorphic evolution and dehydration of subducting oceanic crust. basalt emplaced at mid-ocean ridges, with lesser amounts of alkali and Global estimates of subduction-zone volatile budgets suggest that tholeiitic basalt emplaced at hot spots (oceanic-island ). The

as much as —90% of the bound H20 subducted past the accretionary circulation of hydrothermal fluids through the oceanic crust may prism is contained in the oceanic crust (Ito and others, 1983; change the bulk composition of the oceanic crust, primarily through

Peacock, 1990a); the remaining subducted HzO is contained in oceanic the addition of volatiles (H20 and C02) and alkali elements (Na and sediments and hydrated uppermost mantle. During subduction, in- K). Basalt consists of eleven major oxides plus H20 (Table 1). creases in pressure (P) and temperature (T) tend to liberate H20 from In order to calculate metamorphic modes, the bulk com- subducting oceanic crust as hydrous silicates become progressively positions in Table 1 were reduced to the five component Na20-Ca0- unstable. Previous research has suggested a variety of hydrous phases Mg0-Al203-Si02 (NCMAS) system, based on the following crystal

Geological Society of America Bulletin, v. 105, p. 684-694, 7 figs., 4 tables, May 1993.

684

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/105/5/684/3381769/i0016-7606-105-5-684.pdf by guest on 29 September 2021 DEHYDRATION REACTIONS IN SUBDUCTING OCEANIC CRUST

TABLE 1. ESTIMATES OF THE BULK COMPOSITION OF OCEANIC CRUST BASED ON A DRILL CORES, DREDGE HAULS, AND

Oceanic Oceanic Oceanic Oceanic Oceanic Alkaline Spilite basalt tholeiite basalt basalt basalt basalt

Reference 1 2 3 4 5 6 7 No. of analyses (460) (161) (101) n.r. n.r. (199) (124) Si02 48.5 49.3 50.53 49.5 49.6 47.1 48.8 TiOj 2.6 1.8 1.56 1.5 1.5 2.7 1.3 A1A) 15.0 15.2 15.27 16.0 16.8 15.3 15.7 Fe203 3.1 2.4 n.r. n.r. n.r. 4.3 3.8 FeO 8.5 8.0 10.46 10.5 8.8 8.3 6.6 MnO 0.17 0.17 n.r. n.r. 0.2 0.17 0.15 MgO 7.2 8.3 7.47 7.7 7.2 7.0 6.1 CaO 10.5 10.8 11.49 11.3 11.8 9.0 7.1 NazO 2.5 2.6 2.62 2.8 2.7 3.4 4.4 K,0 0.8 0.24 0.16 0.15 0.2 1.2 1.0 P2O5 0.31 0.21 0.13 n.r. 0.2 0.41 0.34 Total 99.18 99.02 99.69 99.45 99.00 99.88 95.29

References: 1. Manson (1967); chemical analyses compiled from literature; Pacific, Atlantic, and Indian Oceans. 2. Hyndman (Tables 4-8, 1972); oceanic tholeiite analyses compiled from literature. 3. Melson and others (1976); oceanic basalt glasses from Mid-Atlantic Ridge, East Pacific Rise, Figure 1. Ternary ACF diagram with composition of mineral and Indian Ocean. 4. Taylor and McLennan (1985); estimated composition of basaltic layer 3. phases in the NCMASH system. Oceanic basalt compositions (Table 1), 5. Condie (Tabic 4.5, 1989); miscellaneous sources, including dredge samples, core samples, and selected ophiolites. labeled 1 through 7, plot in a narrow field shown in trapezoidal inset. 6. Hyndman (Tables 4-8, 1972); alkali olivine basalt, analyses compiled from literature. A = AI0l s + FeOj 5-(NaO0 s + KO0 S), C = CaO, F = FeO + MgO 7. Hyndman (Tables 4-1, 1972); spilite analyses compiled from literature, n.r., not reported + MnO. Projected through Si02, H20, and either (NaAlSi3Os) or (NaAlSi206). Grt, solid solution; Amph, amphibole solid solution. See Table 2 for mineral abbreviations.

chemical similarities: (1) Na20 = Na20 + K20; (2) CaO = CaO; (3) sent a subdivision of metamorphic conditions based on mineral as- MgO = MgO + FeO + MnO; (4) A1203 = A1203 + Fe203; and (5) semblages (or mineral reactions) (Fig. 2). For a given bulk composi- Si02 = Si02. The oxides Ti02 and P205 were ignored, and the cal- tion, the metamorphic mineral assemblage depends only on the culations were performed on a molar basis. For those chemical anal- metamorphic conditions (for example, P, T, oxygen fugacity, fluid

yses that did not report Fe203, 20% of the reported as FeO was composition). Solid solutions, particularly Fe-Mg solutions in am- assumed to be ferric iron. The different oceanic-basalt compositions phibole, chlorite, and , cause the boundaries between met- plot in a narrow field on the ACF diagram (Fig. 1); the small differences amorphic facies to be transitional in nature and to be generally much

result in part from the uncertainty in Fe203 content of bulk compo- broader than shown in Figure 2. I consider the metamorphic-facies sitions 3, 4, and 5. For this paper, I used Hyndman's (1972) average boundaries depicted in Figure 2 to have widths of at least 50-100 °C oceanic tholeiite as a model for oceanic crust, because he differenti- and 0.1-0.2 GPa. In addition, oxygen fugacity and fluid composition

ated Fe203 from FeO, and because he also presented average bulk strongly influence the position of metamorphic facies; for example, the compositions for olivine alkali basalt and spilite. size of the albite--amphibolite facies is significantly reduced at

The focus of this paper is on HzO which increases the number of low oxygen fugacities (Apted and Liou, 1983). Keeping the transi- chemical components to six (NCMAS + H20). Fresh oceanic basalt tional nature of the metamorphic-facies boundaries in mind, the met- contains between 0.1 and 0.5 wt% HzO (Dixon and others, 1988). amorphic-facies diagram provides a useful framework in which to Ocean-floor metamorphism, caused by large-scale hydrothermal cir- discuss the metamorphic evolution and dehydration of subducting culation, and submarine weathering substantially increase the amount oceanic crust.

of bound H20 in the oceanic crust. Hydration of the oceanic crust In order to calculate the mineral assemblage and maximum H20 occurs preferentially along permeable zones, such as fractures, with content of each of the metamorphic facies, a metamorphic mineral the result that the oceanic crust has a highly variable HzO content. The norm was calculated (Tables 2 and 3) that is similar to the CIPW norms average H20 content of the oceanic crust is difficult to determine; used in igneous . I could have used published chemical

estimates range from 1-2 wt% H20 (Ito and others, 1983; Peacock, analyses to estimate the H20 content of different metamorphic facies, 1990a) up to 6 wt% H20 (Anderson and others, 1976). In this paper, as was done by Fyfe and others (1978), but published analyses are rare,

I treat H20 as an excess component and calculate the maximum and it is not clear that metabasalts are fully hydrated. I used Hynd- amount of bound H20 that may be present in a given metamorphic man's (1972) oceanic tholeiite for the bulk composition of the meta- facies. This approach provides a useful reference frame in which to basalt (Table 1, composition 2), simplified to the NCMAS system. In discuss dehydration reactions that occur in subducting oceanic crust the six-component NCMASH system, a divariant field in P-T space

containing variable amounts of H20. (such as a metamorphic facies) will have six phases. In these calcu- lations, H20 fluid is assumed to be present in excess, leaving five solid Metamorphic Facies, Calculated Mineral Modes, and H20 Contents phases. For each metamorphic facies, I chose a mineral assemblage that is characteristic of the metamorphic facies and encloses the ba- The mineralogical changes in metabasalt as a function of P and saltic bulk composition in the five-component chemical space. T are best illustrated using the concept of metamorphic facies (Eskola, The presence of a relatively small amount of a very hydrous

1920), because most hydrous minerals in metabasalt break down by mineral can have a relatively large effect on the calculated H20 con- a complex series of continuous reactions. Metamorphic facies repre- tent (Table 3). For example, the 4.6 modal percent chlorite present in

Geological Society of America Bulletin, May 1993 685

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/105/5/684/3381769/i0016-7606-105-5-684.pdf by guest on 29 September 2021 TABLE 3. CALCULATED MINERAL MODES AND H20 CONTENT

Metamorphic Mineral Mode* HjO Comments facies assemblage content*

Albite 58.1 2.3 Epidote Clinozoisite 11.9 Amphibolite Chlorite 6.0 Albite 17.1 7.0 Amphibolite^ Hornblende 60.0 2.1 Note that hornblende + (low T) Clinozoisite 7.9 make up >80% of the rock. Chlorite 4.6 AnaiAbgo 20.8 Quartz 6.6 Amphibolite5 Hornblende 58.9 . 1.3 Note that hornblende + plagioclase (high T) PrpaoGrs^ 5.2 make up >85% of the rock. 2.1 AnwAböo 28.6 Quartz 5.2 Eclogite® Hornblende 25.6 0.8 (low P-T) Clinozoisite 14.8 •M50DÌ50 33.0 17.7 Quartz 8.8 Eclogite® Diopside 18.5 0.0 This is a three-phase assemblage: garnet (high P-T) Jadeite 19.6 (Prp68Grs32) + clinopyroxene (Jd54DÌ46) 17.7 + quartz. Pyrope 33.5 Quartz 10.8

Epidote- Tr50Gln50 48.3 2.6 blueschist Clinozoisite 29.1 Figure 2. Pressure-temperature diagram of metamorphic fades Chlorite 11.8 Albite 10.0 based on experimental and theoretical facies diagrams presented by Quartz 0.7 Turner (1981), Liou and others (1985), Oh and others (1988), and Evans Granulite5 Hornblende 51.8 1.2 The presence of orthopyroxene makes this a (low T) Enstatite 6.2 granulite-facies assemblage. Note that (1990). All facies boundaries are transitional in nature and generally Diopside 2.5 hornblende + plagioclase make up >85% of the rock. broader than depicted here (see text for discussion). For each metamor- An50Ab50 36.5 Quartz 3.1 phic facies, the calculated maximum H20 content is shown for a basaltic Granulite5 10.8 0.0 This is a four-phase assemblage: bulk composition. Metamorphic facies abbreviations: AM, amphibolite (highT) Enstatite 8.7 plagioclase (An57Ab43) + Di + Fo + En. Diopside 15.1 This assemblage is identical to the facies; EA, epidote-amphibolite facies; EB, epidote-blueschist facies; Albite 27.8 igneous CIPW norm and shows that the EC, eclogite facies; GN, granulite facies; GS, greenschist facies; LB, 37.6 bulk composition is olivine-normative. lawsonite-blueschist facies; LC, lawsonite-chlorite facies; PP, - Greenschist 29.6 3.4 Clinozoisite 25.2 facies; PrA, prehnite- facies; PA, pumpellyite- Chlorite 17.9 Albite 27.2 actinolite facies; ZE, facies. Quartz 0.1 Lawsonite- 18.1 5.9 Increasing the jadeite content of the Cpx blueschist Lawsonite 28.1 occurs at the expense of glaucophane. Chlorite 19.1 J1I30DÌ70 29.3 Quartz 5.4 Lawsonite- Tremolite 26.9 5.4 chlorite Pumpellyite 19.2 Lawsonite 11.5 Chlorite 16.1 Albite 26.4 TABLE 2. MINERAL FORMULAE AND H20 CONTENT Prehnite- Tremolite 15.9 5.0 actinolite Prehnite 30.6 Chlorite 25.9 Mineral Mineral Minerai Formula H20 1 Albite 25.6 abbrev. group* content Quartz 2.1 Prehnite- Pumpellyite 19.3 5.9 Lmt CaAl2SÌ40i2 • 4H20 15.3 pumpellyite** 9.6 Chi Chlorite (Mg5Al)(AlSi3)O,0(OH)8 13.0 Chlorite 33.0 Lws Lawsonite CaAl2Si2OT(OH)2 • H20 11.5 Albite 24.0 Pmp Pumpellyite CaiMgAlsSisO^fOH), 6.7 Quartz 14.1 Won Wonesite NaMg3(AlSi3)Ö10(OH)2 4.5 Prh Prehnite Ca2Al2Si3O|0(OH)2 4.4 Pumpellyite- Tremolite 20.7 5.1 Gin Glaucophane Amphibole •NazMgjAlzSigOafOH^ 2.3 actinolite Pumpellyite 30.9 Hbl Hornblende Amphibole Nao.sCa^MgiAlKAluSis.sJO^OHJj 2.2 Chlorite 19.2 Tr Tremolite Amphibole •Ca2Mg5Si8022(0H)2 2.2 Albite 26.4 Czo Clinozoisite Ca2Al3Si30,2(0H) 2.0 Quartz 2.9 Ab Albite Feldspar NaAlSi 0 0 3 8 Zeolite** Laumontite 27.6 8.5 An Anorthite Feldspar CaAl2Si208 0 Cal Calcite CaC0 0 Calcite 12.1 3 Chlorite 31.6 Di Diopside Clinopyroxene CaMgSi206 0 Jd Jadeite Clinopyroxene NaAlSi 0 0 Albite 21.3 2 6 Quartz 7.5 Omp Clinopyroxene (Na0.5Ca0.5)(Mg0.5Al„.5)Si2O6 0 En Enstatite Mg2Si206 0 Fo Forsterite Mg2Si04 0 *Mineral modes in percent by volume. Grs Grossular Garnet CajAl2Si3012 0 *H20 content in percent by weight. ftp Pyrope Garnet Mg3Al2Si3012 0 5These facies occur over a wide range in P-T space. Two possible mineral modes are presented Qtz Quartz Si02 0 representing H20-rich and H20-poor end members. **A Ca-rich, Al-poor mineral is required to encompass the basaltic bulk composition for these *Solid solutions considered in this paper. facies. Calcite, which lies outside the NCMASH system, was used in these calculations because it is *H20 content in percent by weight. a common mineral in low-grade metabasalt.

686 Geological Society of America Bulletin, May 1993

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/105/5/684/3381769/i0016-7606-105-5-684.pdf by guest on 29 September 2021 DEHYDRATION REACTIONS IN SUBDUCTING OCEANIC CRUST

TABLE 4. POSSIBLE MODES AND H20 CONTENTS OF GREENSCHIST-FACIES In order to focus on the most important controls on P-7" paths, all METABASALT of the P-T paths depicted in Figure 3 were calculated (Appendix 1) 2 Chemical Mineral Oceanic Oceanic Spilite using the following parameters: basal heat flux, Q0 = 0.05 W/m system assemblage tholeiite alkali (corresponding to the mean observed heat flow out of > 80 Ma oceanic basalt lithosphere; Sclater and others, 1980); thermal conductivity, k - 2.5 6 2 NCMASH Tremolile* 29.6 23.3 16.1 W/(m-K); thermal diffusivity, K = 1 x 10 m /s; subduction angle, Clinozoisite 25.2 22.0 18.5 Chlorite 17.9 19.5 19.0 8 = 20° (corresponding to the average dip of Benioff zones from 0 to Aibite 27.2 41.7 50.2 60-km depth; Jarrard, 1986); and density, p = 3,000 kg/m3. The effect Quartz 0.1 (6.5) (3.7) J l20 contend 3.38 3.41 3.15 of varying these parameters on subduction-zone thermal structures

NCMASH Tremolite 29.7 15.4 11.5 and subduction-zone P-T paths has been explored by Molnar and Clinozoisite 25.2 26.6 21.1 England (1990) and Peacock (1992), respectively. Chlorite 18.1 10.4 13.8 Aibite 27.4 29.7 43.5 The most important variable influencing steady-state P-Tpaths is Wonesite (0.3) 17.9 10.2 11 ¿0 content 3.39 2.98 2.90 the rate of shear heating [= shear stress (T) X slip rate (V)] in the

KNCMASH Tremolile 28.6 18.2 11.8 subduction (Fig. 3A). The magnitude of shear stresses in Clinozoisite 25.8 24.7 20.7 subduction zones is considerably uncertain; estimates of T range from Chlorite 16.9 13.7 14.2 Aibite 25.6 34.0 43.8 < 10 MPa (Shreve and Cloos, 1986) to 10-50 MPa (Bird, 1978; van den Phlogopite 2.3 11.8 9.8 Quartz 0.9 (2.4) (0.3) Beukel and Wortel, 1988; Dumitru, 1991; Peacock, 1992) to >50-100 H2O content 3.33 3.13 2.92 MPa (Graham and England, 1976; Davies, 1980; Honda, 1985; Molnar KNCMASH Tremolite 33.6 5.3 10.0 and England, 1990). For T = 0 and V = 10 cm/yr, steady-state P-T Clinozoisite 26.6 22.6 20.4 Chlorite 15.0 18.3 14.8 paths calculated for the top of the subducting oceanic crust intersect Aibite 25.6 33.9 43.8 1 GPa (10 kb) at T = 95 °C. For T = 100 MPa and V = 10 cm/yr, Phiogopite 2.3 11.8 9.8 Diopsidc (3.2) 8.2 1.2 steady-state P-T paths calculated for the top of the subducting oceanic H-J> content 3.25 3.34 2.95 crust intersect 1 GPa at T = 705 °C, more than 600 °C higher than in •Mineral modes in percent by volume. the absence of shear heating. In order to discuss the dehydration of ' HjO content in percent by weight. subducting oceanic crust, I chose four representative P-T paths (V = 10 cm/tyr; T = 100, 67, 33, and 0 MPa) that cover the plausible range of P-T paths followed by subducting oceanic crust (Fig. 3B). the low-temperature amphibolite fades accounts for —25% of the total The thermal structure of a subduction zone cools and approaches bound H20 calculated for the rock. Because the amount of hydrous a steady state over a period of several to tens of million years. The time minerals, such as chlorite, in a calculated norm is sensitive to the bulk it takes for a subduction zone to reach steady state depends on the composition of the system, the calculated HzO contents may be sen- convergence rate, angle of subduction (5), and the depth of interest (z) sitive to the assumed bulk composition of the oceanic crust. In order (see Molnar and England, 1990). For V = 10 cm/yr and 8 = 20°, to test the sensitivity of calculated H20 contents to the bulk compo- steady-state conditions are approached after —5 Ma for z = 34 km sition, a series of norm calculations were performed for three different (P = 1 GPa) and after -17 Ma for z = 68 km (P = 2 GPa). For V = basaltic compositions taken from Hyndman (1972): oceanic tholeiite, 3 cm/yr, steady-state conditions are approached after —7 Ma for z = oceanic alkali basalt, and spilite (Table 4). For each composition, four 34 km and after —22 Ma forz = 68 km. Oceanic crust subducted during possible greenschist-facies mineral assemblages were considered, two this early transient period will follow warmer P-T paths than will in the NCMASH system and two in the KNCMASH system. The oceanic crust subducted after steady state has been achieved. Figure calculated H20 contents for the 12 different norms are remarkably 3C illustrates the difference between initial and steady-state P-T paths similar, averaging 3.18 ± 0.20 wt% H20, despite substantially differ- for the case of V = 10 cm/yr and t = 33 MPa. The top of the first ent mineral modes. Six of the norms have negative amounts of one oceanic crust to be subducted reaches a temperature of 475 °C at P = mineral; the average H20 content of the six acceptable norms is 1 GPa. The subduction zone cools with time, and oceanic crust sub- 3.15 ± 0.22 wt% H,0. Using the acceptable norms, a greenschist- ducted under steady-state conditions reaches temperatures of only facies oceanic tholeiite could contain up to —3.4 wt% H20, a green- 300 °C at P = 1 GPa. -facies oceanic alkali basalt could contain up to —3.2 wt% HzO, A third important variable is the position of the rock within the and a greenschist-facies spilite could contain up to —2.9 wt% H20. subducting oceanic crust. If shear heating is a major heat source in Based on this analysis, I estimate that the calculated maximum H20 subduction zones, then the top of the subducting oceanic crust follows contents of metabasalt have uncertainties of —10 relative percent a much warmer P-T path than does the base of the subducting oceanic because of variations in bulk composition. crust. For the case of V = 10 cm/yr and t = 67 MPa, the top of the oceanic crust is 300 °C hotter than the base of the 7-km-thick oceanic Calculated P-T Paths for Subducting Oceanic Crust crust atP = 1 GPa (Fig. 3D). If shear heating is insignificant, then the top of the subducting oceanic crust follows a slightly cooler P-T path Subduction zone P-T paths depend on the magnitude of shear than does the base. heating, the geometry, and rate of subduction, the thermal structure (age) of the subducting lithosphere, the vigor of induced mantle con- DISCUSSION vection, and the thermal properties of the rocks (Peacock, 1991). P-T paths followed by subducting oceanic crust may be calculated using P-T Paths and Metamorphic Facies analytical expressions (Molnar and England, 1990) or numerical meth- ods (Peacock, 1991). In this paper, I used analytical expressions based The metamorphic evolution of subducting oceanic crust depends on Molnar and England's work, because these expressions are rela- on several important variables, of which the rate of shear heating in tively easy to evaluate and closely approximate the actual solutions. the subduction shear zone is perhaps the most important (Fig. 3A). For

Geological Society of America Bulletin, May 1993 687

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/105/5/684/3381769/i0016-7606-105-5-684.pdf by guest on 29 September 2021 Temperature (°C) Temperature (°C)

Temperature (°C) Temperature (°C) Figure 3. Calculated pressure-temperature (P-T) paths followed by subducting oceanic crust overlaid on a metamorphic-facies diagram. See 2 3 6 2 Appendix 1 for equations used to calculate P-T paths. Q0 = 0.05 W/m , 8 = 20°, p = 3,000 kg/m , k = 2.5 W/(m-K), K = 1 x 10" m /s. (A) Steady-state P-T paths followed by the top of the subducting oceanic crust for different shear stresses and convergence rates. Curves are labeled with shear stress in MPa. Higher shear stresses result in warmer steady-state P-T paths. (B) Representative steady-state P-T paths (V = 10 cm/yr) used to discuss different evolutions of subducted oceanic crust. (C) Initial and steady-state P-T paths followed by the top of the subducting oceanic crust (V = 10 cm/yr, T = 33 MPa). The initial P-T path represents conditions encountered by the first rock to be subducted. (D) Steady-state P-T paths followed by the top and the base of 7-km-thick oceanic crust (V = 10 cm/yr, T = 67 MPa). Prior to subduction, the top of the oceanic crust is cooler than the base, but shear heating along the subduction shear zone causes inverted thermal gradients to develop in subducting oceanic crust.

688 Geological Society of America Bulletin, May 1993

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/105/5/684/3381769/i0016-7606-105-5-684.pdf by guest on 29 September 2021 DEHYDRATION REACTIONS IN SUBDUCTING OCEANIC CRUST

high rates of shear heating (for example, V = 10 cm/yr and r ~ 100 pends strongly on the rate of shear heating in the subduction shear MPa), calculated steady-state P-T paths for the top of the subducting zone. As discussed above, for high rates of shear heating (V = 10 oceanic crust pass through the high-temperature greenschist, epidote- cm/yr, T = 100 MPa), the top of the slab passes through the greenschist amphibolite, amphibolite, and granulite facies. For low to moderate epidote-amphibolite —» amphibolite —» granulite -* eclogite rates of shear heating (T ~ 20-60 MPa), calculated steady-state P-T facies (Fig. 4A); the lower parts of the subducting oceanic crust un- paths pass through the blueschist facies. For the case of no shear dergo a strikingly different metamorphic evolution, passing through heating (T = 0), calculated steady-state P-T paths are very cold, in- the lawsonite-blueschist —> epidote-blueschist —> eclogite facies tersecting P = 1 GPa at T < 200 °C. Given these very low tempera- (Fig. 4A). The lowermost subducting oceanic crust remains in the tures, metamorphic reactions may not even occur in the subducting lawsonite-blueschist field at depths >70 km, even for high rates of oceanic crust until much greater depths are reached. Is it possible that shear heating (Fig. 4A). For moderate rates of shear heating (Figs. 4B some of the apparently unmetamorphosed rocks present in ancient and 4C), the subducting oceanic crust does not pass through the subduction zones have actually undergone high pressures, but very higher-temperature amphibolite and granulite facies, more of the sub- low temperatures? ducting oceanic crust passes through the blueschist facies, and the The thermal structure of a subduction zone cools over time, and blueschist -> eclogite transition occurs at greater depths. subducting oceanic crust follows progressively cooler P-T paths until Because shear heating is proportional to the convergence rate, steady-state conditions are achieved. Subducting oceanic crust will the rate of shear heating in Figure 5 is 30% of the rate in Figure 4 for encounter different metamorphic facies as the subduction zone the same value of shear stress. For the case of V = 3 cm/yr and T = evolves. In the example depicted in Figure 3C (V = 10 cm/yr, T = 33 100 MPa (Fig. 5A), the top of the subducting oceanic crust passes MPa), the earliest subducted oceanic crust passes through the fol- through the greenschist epidote-blueschist —» eclogite facies; the lowing sequence of metamorphic facies: greenschist -> epidote-am- higher-temperature epidote-amphibolite, amphibolite, and granulite phibolite eclogite. For the same parameters, oceanic crust sub- facies are not encountered by the subducting oceanic crust. For V = ducted in a subduction zone that has reached thermal steady state will 3 cm/yr, all of the oceanic crust passes through the blueschist facies undergo lawsonite-blueschist —> eclogite facies metamorphism. The for the shear stresses considered here. As was the case for V = 10 variation in metamorphism with time depends on the rate of shear cm/yr, the blueschist —> eclogite transition occurs at greater depths for heating and, to a lesser extent, the convergence rate. Lower rates of lower rates of shear heating. shear heating and higher convergence rates result in greater differ- ences between the initial and steady-state P-Tpaths. In general, as the Dehydration Reactions in Subducting Oceanic Crust subduction zone cools with time, subducting oceanic crust follows progressively cooler P-T paths, and the dehydration of the oceanic Almost all subducting oceanic crust passes through the blueschist crust migrates to deeper levels. facies (Figs. 4 and 5). Based on the norm calculations presented above,

The preceding discussion focused on P-T paths followed by the metabasalt may contain up to 2.6 wt% H20 in the epidote-blueschist top of the subducting oceanic crust that is in contact with the over- facies and up to 5.9 wt% HzO in the lawsonite-blueschist facies. In riding plate. Rocks within the oceanic crust are located some distance contrast, metabasalt in the eclogite facies contains between 0.8 and 0.0 from the subduction shear zone and will follow different P-Tpaths than wt% H20. Therefore, large amounts of H20 should be released from will the top of the oceanic crust (Fig. 3D). Moderate rates of shear subducting oceanic crust as rocks pass from the blueschist facies into heating result in the top of the oceanic crust following warmer P-T the eclogite facies, owing to the breakdown of lawsonite, clinozoisite, paths than does the base of the oceanic crust. The lower parts of the chlorite, and amphibole. oceanic crust follow cooler P-T paths and would be expected to de- In order to illustrate different dehydration histories for subducting

hydrate at greater depth than would the upper parts of the subducting oceanic crust, curves showing the maximum amount of H20 as a oceanic crust. In the example in Figure 3D, the top of the oceanic crust function of temperature and pressure were constructed (Fig. 6) for will enter the eclogite facies at P ~ 1.2 GPa and T ~ 560 °C while the three of the representative P-T paths shown in Figure 3B. Oceanic base of the oceanic crust still lies in the lawsonite-blueschist facies. If crust following cooler P-T paths has the potential to subduct larger

the lower part of the oceanic crust contains substantial amounts of amounts of H20 to greater depths than does oceanic crust following H20, then the relatively cold P-T paths followed by the lower crust warmer .P-T paths (Fig. 6). Although the phase equilibria of metabasalt would allow this water to be subducted to depths of 70 km or more. at high pressures and low temperatures are not well known, oceanic In thermal steady state, the thermal structure of the subducting crust following cool P-T paths could subduct very large quantities of oceanic crust does not change with time, and the locations of the H20 to depths greater than 70 km.

metamorphic-facies boundaries in the subducting oceanic crust re- Prior to subduction, the H20 content of the oceanic crust is highly main fixed relative to the surface. For a steady-state subduction zone, variable on the scale of millimeters to kilometers. During subduction,

therefore, it is possible to construct cross sections through the sub- parts of the oceanic crust containing greater amounts of H20 will ducting oceanic crust, showing the distribution of metamorphic facies dehydrate before rocks containing less H20 will. The released water for different convergence rates and rates of shear heating (Figs. 4 and may leave the system or may be reabsorbed by adjacent dry rocks. 5). The metamorphic evolution of a rock within the oceanic crust may The net effect of local dehydration and hydration reactions will be to

be traced by following a line parallel to the top of the subducting homogenize the H20 content of the subducting oceanic crust. Such a oceanic crust. These cross sections represent a different way of vis- process could account for the more uniform H20 contents of green- ualizing the same information presented in the P-T path diagrams schist and blueschist in paleosubduction as compared to the (Fig. 3), but such cross sections make the geometry of metamorphic variably hydrated nature of basalt in and oceanic drill cores.

facies and dehydration reactions within the subducting oceanic crust Most estimates of the amount of HzO contained in the oceanic easier to visualize. crust are substantially less than the maximum amount of H20 that can The metamorphic structure of the subducting oceanic crust de- be accommodated in basalt in the and other low-grade

Geological Society of America Bulletin, May 1993 689

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/105/5/684/3381769/i0016-7606-105-5-684.pdf by guest on 29 September 2021 S. M. PEACOCK

(A) Distance (km)

Figure 4. Cross sections of subducting oce- anic crust showing the distribution of meta- morphic facies under steady-state conditions for three different amounts of shear heating and V = 10 cm/yr. In thermal steady state, temperature structure and facies boundaries are constant over time. Rocks within the sub- ducting oceanic crust follow gray lines parallel to the top of the slab. Most subducting oceanic crust passes through the blueschist —* eclogite facies transition. The geometry of the top of the subducting oceanic crust is to scale, but the 7-km-thick oceanic crust has been stretched vertically by a factor of five to depict the facies boundaries more clearly; dashed lines on the right-hand side illustrate this vertical stretch. Diagonal gray lines with positive slope are iso- bars (labeled in MPa) which would be hori- zontal if the oceanic crust were not stretched vertically. See text for discussion and Figure 2 for metamorphic-facies abbreviations. Note that metamorphic-facies boundaries within subducting oceanic crust will be broader than depicted here. (A) T = 100 MPa. (B) T = 67 MPa. (C) T = 33 MPa.

metamorphic facies. Oceanic crust that contains less than the maxi- within the amphibolite facies associated with the breakdown of chlo-

mum amount of HzO in a given metamorphic facies has the potential rite and clinozoisite. For P-T path 2, oceanic crust containing 2 wt% to undergo hydration reactions if the rock is infiltrated by H20-rich H20 will begin dehydrating at the epidote-amphibolite -» eclogite fluids. A potential source of such fluid is dehydration reactions oc- facies boundary associated with the breakdown of hornblende + chlo- curring at deeper levels in the subduction zones. Hydration, rather rite + clinozoisite to form garnet + omphacite (Figs. 6C and 6D). For than dehydration, of subducting oceanic crust may be an important P-T paths lying between paths 2 and 3, dehydration will begin at the process at shallow levels in subduction zones where very hydrous epidote-blueschist -» eclogite facies boundary associated with the minerals such as zeolite, smectite, and chlorite are stable (Fyfe and reaction glaucophane + clinozoisite —» omphacite + garnet + H20 McBirney, 1975; Bebout, 1991). (Evans, 1990) or at the lawsonite-blueschist —» eclogite facies bound-

The release of H20 from subducting oceanic crust will not occur ary associated with the reaction glaucophane + lawsonite —> ompha- until the rock reaches P- T conditions for which the maximum amount cite + garnet + HzO (Evans, 1990). For P-T path 3, given our current

of H20 that can be accommodated in hydrous minerals is less than the knowledge of phase equilibria, oceanic crust containing 2 wt% HzO H20 content of the rock. Consider the subduction of oceanic crust may not dehydrate at P < 2 GPa (Figs. 6E and 6F). containing 2 wt% H20 (Fig. 6). For P-T path 1, net dehydration Oceanic crust containing 1 wt% HzO will begin dehydrating reactions will not occur in the zeolite, prehnite-pumpellyite, prehnite- within the granulite facies (P-T path 1), at the onset of the eclogite actinolite, greenschist, and epidote-amphibolite facies, because the facies (P-T path 2), or at P > 2 GPa (P-T paths 3 and 4). Net dehy- mineralogy of each of these facies can accommodate more than 2 wt% dration will not occur at low metamorphic grades.

H20 (Figs. 6A and 6B). Dehydration of the oceanic crust will begin Oceanic crust containing 4 wt% H20 may begin dehydrating in

690 Geological Society of America Bulletin, May 1993

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/105/5/684/3381769/i0016-7606-105-5-684.pdf by guest on 29 September 2021 DEHYDRATION REACTIONS IN SUBDUCTING OCEANIC CRUST

(A) Distance (km)

Figure 5. Same as Figure 4 except V = 3 cm/yr. (A) T = 100 MPa. (B) T = 67 MPa. (C) T = 33 MPa.

-

(km)

the greenschist or epidote-blueschist fades, depending on the P-T upward, then the H20 will pass through oceanic crust lying in the path. Cool P-T paths are subparallel to the lawsonite-blueschist (5.9 low-temperature eclogite field.

wt% H20 maximum) epidote-blueschist (2.6 wt% H20 maximum) Dehydration reactions associated with the blueschist —' eclogite facies boundary (Fig. 3A); the calculated depth at which lawsonite transition will occur at depths >70 km for subduction zones with breaks down to form clinozoisite is very sensitive to theP-Tpath. This moderate to low rates of shear heating (Figs. 4C and 5C). At depths reaction could release substantial quantities of H20 if the subducting >50-100 km, the thermal structure of the subducting oceanic crust oceanic crust contains more than 2.6 wt% bound H20. will be influenced by the induced in the overlying asthen- For many P-T paths, subducting oceanic crust containing 1-2 ospheric wedge. The analytical expressions used to calculate the P-T

wt% H20 will lose a substantial amount of its HzO at the blue- paths presented in this paper (Appendix 1) do not incorporate induced schist eclogite transition. This transition is a strong function of asthenospheric convection; therefore the P-T paths in Figure 3 cannot temperature and occurs at T —500 °C (Fig. 2). For high rates of shear be extrapolated to depths >70 km (P > 2 GPa). Two-dimensional heating, the 500 °C isotherm and the blueschist —> eclogite transition numerical models that incorporate induced asthenospheric convec- cut down through the oceanic crust and dip in the same direction as tion suggest that P-T paths followed by the subducting oceanic crust the subduction zone (Figs. 4A, 4B, 5A, and 5B). Lower parts of the are substantially warmer than in the absence of induced asthen- oceanic crust undergo dehydration associated with the blueschist ospheric convection (Peacock, 1991), but the difference in P-T paths

eclogite transition at greater depths than do the upper parts. If H20 begins only after the oceanic crust passes beneath the convecting released from blueschist -* eclogite dehydration reactions migrates asthenospheric wedge. Calculated P-T paths at shallower depths, such

Geological Society of America Bulletin, May 1993 691

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/105/5/684/3381769/i0016-7606-105-5-684.pdf by guest on 29 September 2021 S. M. PEACOCK

Figure 6. Maximum H 0 content of subducting oceanic 0 200 400 600 800 1000 0.4 0.8 1.2 1.6 2 Temperature (°C) Pressure (GPa) crust as a function of temperature and pressure for P- T paths 10 1 1 I" 1, 2, and 3 (see Fig. 3B). Rocks following cooler P-T paths D j have the potential to subduct more H20 to greater depths. ? S— ( Path 2) Ì8 Oceanic crust containing 2 wt% HzO (horizontal dashed C line) will not dehydrate until the maximum H20 curve drops

•

" ZE -LC LB

1 1 1 200 400 600 800 1000 0.4 0.8 1.2 1.6 2.0 Temperature (°C) Pressure (GPa)

as those presented in this paper, are unaffected by induced astheno- For mostP-Tpaths, subducted oceanic crust containing 1-2 wt% spheric convection. Induced asthenospheric convection will cause H20 will begin to dehydrate at the eclogite-facies boundary. Eclogite heating of the subducting slab and may drive the blueschist —» eclogite is commonly thought of as dry rock consisting of omphacite + garnet, transition within the subducting oceanic crust at depths >70 km. but low-temperature contain a wide variety of hydrous min- erals, including (glaucophane, tremolite, nyboite), Hydrous Minerals within the Eclogite Facies (phengite, , , phlogopite), lawsonite, -epidote, chlorite, and chlorotoid (Newton, 1986; Carswell, 1990). In many Many P-T paths calculated for subducting oceanic crust pass cases, these hydrous phases are in apparent textural equilibrium with through the eclogite facies. The low-pressure and low-tempera- omphacite and garnet and are interpreted as stable eclogite-facies ture boundaries of the eclogite facies are distinctly different. Cool minerals. The onset of eclogite-facies metamorphism, therefore, does P-T paths intersect the low-temperature eclogite boundary, which not represent the final dehydration of subducting oceanic crust. has a steep, negative dP/dT associated with the dehydration of Thermodynamic calculations using the THERMOCALC pro- glaucophane + lawsonite/clinozoisite to form garnet + omphacite gram (Powell and Holland, 1988; Holland and Powell, 1990) indicate

+ HzO (Evans, 1990). Warm P-T paths intersect the low-pressure that the amphibole end-member tremolite should break down to form eclogite boundary, which has a shallow, positive dP/dT and is talc + diopside at P —2.5 GPa. H20 is not released in this solid-solid primarily associated with the breakdown of plagioclase compo- reaction; the H20 bound in the tremolite structure goes into talc, nents (for example, albite —» jadeite + quartz). Compared to the which remains stable to much higher pressures. Other amphibole end low-temperature eclogite boundary, relatively little dehydration members, such as glaucophane, may undergo similar reactions to form occurs across the low-pressure eclogite boundary; only minor + pyroxene, but chemically complex amphibole could remain

amounts of HzO are released by the partial breakdown of horn- stable to pressures significantly greater than the tremolite-breakdown blende ± clinozoisite ± chlorite. reaction. Recent subsolidus phase equilibria experiments by Pawley

692 Geological Society of America Bulletin, May 1993

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/105/5/684/3381769/i0016-7606-105-5-684.pdf by guest on 29 September 2021 DEHYDRATION REACTIONS IN SUBDUCTING OCEANIC CRUST

and Holloway (1992) demonstrated that amphibole breaks down be- km and is probably driven by heat derived from the overlying con- tween 2 and 2.5 GPa, but other hydrous minerals remain stable to vecting asthenospheric wedge. The results of this investigation sug- higher pressures, including epidote and lawsonite. gest that oceanic crust can subduct large amounts of H20 to depths Given our current knowledge of metabasaltic phase equilibria, it of—100 km, and that arc magmatism may result from infiltration of the appears that basalt which was metamorphosed under low-temperature mantle wedge by H20-rich fluids derived from blueschist —> eclogite

eclogite-facies conditions could contain up to 0.5 wt% H20 and pos- dehydration reactions in subducting oceanic crust. sibly more. If this view is correct, then substantial amounts of H20 could be subducted to depths > 100 km or more. ACKNOWLEDGMENTS Subducted oceanic crust following cold P-rpaths would be even

more effective at subducting H20 to great depths. Experiments have I thank my many colleagues at the ETH, particularly J.A.D. not been conducted at the high pressures and low temperatures in- Connolly, G. L. Friih-Green, T. A. Rushmer, R. J. Sweeney, A. B. tersected by P-rpath 4 (Fig. 3B), and thermodynamic calculations at Thompson, and P. Ulmer, for numerous spirited discussions of sub- these conditions are very uncertain. If subduction-zone shear stresses duction-zone processes. Constructive reviews by J. G. Liou, R. J. are low (T < 20 MPa), and cold P-T paths are characteristic of sub- Tracy, A. G. Sylvester, and an anonymous reviewer substantially

duction zones, then almost all of the H20 in the subducting oceanic improved the manuscript. This work was supported by National Sci- crust could be subducted to depths >100 km. ence Foundation Grant EAR-9105741 and by a Guest Professorship provided by the Swiss Federal Institute of Technology (ETH, Partial Melting of Subducting Oceanic Crust Zurich).

Subducting oceanic crust following hot P-T paths, caused by high APPENDIX 1. DERIVATION OF ANALYTICAL EXPRESSIONS FOR P-T rates of shear heating, will intersect partial melting reactions at P < PATHS FOR SUBDUCTING OCEANIC CRUST

2 GPa (Peacock, 1990a, 1991). If H20 is present as a distinct fluid phase, then partial melting will occur where P-rpaths intersect the wet The following derivations are based to a large extent on derivations pre- sented in Molnar and England (1990), and I use their coordinate system in this basalt solidus. The amount of partial melting that will occur at the wet discussion (Fig. Al). solidus is proportional to the amount of HzO available. For example, if the porosity of subducting oceanic crust is 0.1% and the pores A. Steady-State P-T Paths for the Top of the Subducting contain pure H20 (pH2o ~ 0.3 prock), then a melt containing 3 wt% H20 Oceanic Crust would form from 1% partial melting. Changing the porosity of the subducting oceanic crust by an order of magnitude would change the In thermal steady state, the P-7" path followed by the top of the subducting amount of partial melt formed at the wet solidus by an order of oceanic crust coincides with the P-T conditions along the subduction shear magnitude. zone. In the absence of significant radiogenic heating, steady-state tempera- tures along a subduction shear zone are accurately approximated by (Molnar More substantial partial melting of the oceanic crust would occur and England, 1990, equations 16 and 23) where P-T paths intersect the fluid-absent partial melting reactions associated with the breakdown of hornblende, which occurs at tem- peratures of 800-1100 °C at pressures of 1-2 GPa (Green, 1982). In steady-state subduction zones, such conditions are achieved only 2 where T = temperature (K), Q0 = basal heat flux (W/m ), T = shear stress (Pa), if the rate of shear heating is high (for example, V —10 cm/yr and V = convergence rate (m/s), zf = depth to the fault (m), k = thermal conduc- T —100 MPa). Such conditions may also be achieved in very young, tivity (W/m-K), and S = a divisor that accounts for advection given by non-steady-state subduction zones where young oceanic lithosphere V zf sin 5 is being subducted (Peacock, 1990a, 1991). The steady-state P-T paths S = \+b-J—J— (A2) calculated for subduction zones with low to moderate rates of shear heating indicate that the subducting oceanic crust will not undergo where b = a constant (=1 based on numerical experiments), 5 = angle of partial melting. subduction, and K = thermal diffusivity (m2/s). In this model, the magnitude of the shear stress is a constant value along the subduction shear zone. CONCLUSIONS

The metamorphism and dehydration of subducting oceanic crust maybe predicted by integratingP-rpaths derived from heat-transfer theory with metabasalt phase equilibria. For a wide range of param- eters, subducting oceanic crust passes through the blueschist -» eclo- gite facies. Higher-temperature greenschist, epidote-amphibolite, am- phibolite, and granulite facies conditions are encountered by the upper part of the subducting oceanic crust only in subduction zones char- acterized by high rates of shear heating (for example, V—10 cm/yr and

T —100 MPa). Oceanic crust containing 1-2 wt% H20 will release substantia] amounts of H20 at the blueschist —» eclogite boundary associated with dehydration reactions involving lawsonite (or clino- zoisite), glaucophane, and chlorite. The depth of the blueschist —» eclogite transition depends to a large extent on the rate of shear Figure Al. Coordinate system and geometry used to model heat heating. If shear stresses in subduction zones are moderate (T = 20-40 transfer in subducting oceanic crust. See Appendix 1 for discussion of MPa), then the blueschist —• eclogite transition occurs at depths >70 parameters.

Geological Society of America Bulletin, May 1993 693

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/105/5/684/3381769/i0016-7606-105-5-684.pdf by guest on 29 September 2021 S. M. PEACOCK

Depth is related to pressure by REFERENCES CITED Anderson, R. N., Uyeda, S., and Miyashiro, A., 1976, Geophysical and geochemical constraints at P = pgz (A3) converging plate boundaries—Part I: Dehydration in the downgoing slab: Royal Astronomical Society Geophysical Journal, v. 44, p. 333-357. where p = density (kg/m3) and g = gravitational constant (9.8 m/s2). Apted, M. J., and Liou, J. G., 1983, Phase relations among greenschistj epidote amphibolite, and amphibolite in a basaltic system: American Journal of Science, v. 283-A, p. 32&-354. Bebout, G. E;, 1991, Field-based evidence for devolatilization in subductioni zones: Implications for arc B. Steady-State P-T Paths for Rocks within the Subducting magmatism: Science, v. 251, p. 413-416. Bird, P., 1978, Stress and temperature in subduction shear zones: Tonga and Mariana: Royal Astro- Oceanic Crust nomical Society Geophysical Journal, v. 55, p. 411-434. Carslaw, H. S., and Jaeger, J. C., 1959, Conduction of heat in solids: New York, Oxford University Press, 510 p. In a thrust system, the top part of the lower plate warms as it passes Carswell, D. A., ed., 1990, Eclogite facies rocks: London, England, Blackie, 396 p. beneath the upper plate. If we choose a moving reference frame that follows Condie, K. C., 1989, and crustal evolution: Oxford, England, Pergamon, 476 p. Delany, J. M., and Helgeson, H. C., 1978, Calculation of the thermodynamic consequences of dehy- the descending lower plate, then the thermal structure of the lower plate may dration in subducting oceanic crust to 100 kb and >800°C: American Journal of Science, v. 278, be accurately calculated using the solution for conduction of heat into a semi- p. 638-686. Dixon, J. E., Stolper, E., and Delaney, J. R., 1988, Infrared spectroscopic measurements of CO2 and infinite half space where the temperature at the upper surface increases with H2O in Juan de Fuca Ridge basaltic glasses: and Planetary Science Letters, v. 90, p. 87-104. time (Carslaw and Jaeger, 1959). In the steady state, the boundary condition Dumitru, T. A., 1991, Effects of subduction parameters on geothermal gradients in forearcs, with an at the top of the descending slab is given by equation Al, which may be application to Franciscan subduction in California: Journal of Geophysical Research, v. 96, p. 621-641. rewritten as a function of time (t) by substituting Eskola, P., 1920, The mineral fades of rocks: Norsk Geologisk Tidsskrift, v. 6, p. 143-194. Evans, B. W., 1990, Phase relations of epidote : Lithos, v. 25, p. 3-23. : Vt sin 5 (Bl) Fyfe, W. S., and McBimey, A. R., 1975, Subduction and the structure of andesitic volcanic belts: American Journal of Science, v. 275-A, p. 285-297. which yields Fyfe, W. S., Price, N. J., and Thompson, A. B., 1978, Fluids in the Earth's crust: Amsterdam, the Netherlands, Elsevier, 365 p. Gill, J., 1981, Orogenic andesites and plate tectonics: New York, Springer-Verlag, 390 p. Graham, C. M., and England, P. C., 1976, Thermal regimes and regional metamorphism in the vicinity (B2) of overthrust faults: An example of shear heating and inverted metamorphic zonation from south- ern California: Earth and Planetaiy Science Letters, v. 31, p. 142-152. Green, T.H., 1982, Anatexis of mafic crust and high pressure crystallization of andesite, in Thorpe, R. S., Note that time in this equation represents the time it takes for a rock to descend ed., Andesites: New York, Wiley, p. 465-487. Holland, T.J.B., and Powell, R., 1990, An enlarged and updated internally consistent thermodynamic to depth zf and not the time since the beginning of subduction. This derivation dataset with uncertainties and correlations: The system K20-Na20-Ca0-Mg0-Mn0-Fe0-Fe203- applies to the thermal steady state which is achieved only after several to tens AI2O3-TÌO2-SÌO2-C-H2-O2: Journal of Metamorphic Geology, v. 8, p. 89-124. of million years of subduction. Honda, S., 1985, Thermal structure beneath Tohoku, northeast Japan—A case study for understanding the detailed thermal structure of the subduction zone: Tectonophysics, v. 112, p. 69-102. The initial thermal structure of the lower plate before it descends beneath Hyndman, D. W., 1972, Petrology of igneous and metamorphic rocks: New York, McGraw-Hill, 533 p. the upper plate is approximated using a linear : Ito, E., Harris, D. M., and Anderson, A. T., Jr., 1983, Alteration of oceanic crust and geologic cycling of chlorine and water: Geochimica et Cosmochimica Acta, v. 47, p. 1613-1624. Jarrard, R. D., 1986, Relations among subduction parameters: Reviews of Geophysics, v. 24, p. 217-284. T (w, t = 0) = ^ w (B3) Liou, J. G., Maruyama, S., and Cho, M., 1985, Phase equilibria and mineral parageneses of metabasites k in low-grade metamorphism: Mineralogical Magazine, v. 49, p. 321-333. Manson, V., 1967, Geochemistry of basaltic rocks: Major elements, in Hess, H. H., and Poldervaart, À., eds., Basalts: New York, Wiley, p. 215-269. where w = distance measured perpendicular to the . Melson, W. G., Vallier, T. L., Wright, T. L., Byerly, G., and Nelen, J., 1976, Chemical diversity of Using the analytical solution of Carslaw and Jaeger (1959, p. 63, equa- abyssal volcanic glass erupted along Pacific, Atlantic, and Indian Ocean sea-floor spreading systems,/« Sutton, G. H., Manghnani, M. H., and Moberly, R., eds., The geophysics of the Pacific tion 7) for the conduction of heat into a semi-infinite solid with a surface Ocean basin and its margin: Washington, D.C., American Geophysical Union Geophysical Mono- temperature that varies as a function of time, to which must be added the initial graph 19, p. 351-367. temperature of the lower plate, we calculate Molnar, P., and England, P. C., 1990, Temperatures, heat flux, and frictional stress near major thrust faults: Journal of Geophysical Research, v. 95, p. 4833-4856. Newton, R. C., 1986, Metamorphic temperatures and pressures of Group B and C eclogites: Geological Qo Ksin 5 I Qu + T Society of America Memoir 164, p. 17-30. (B4) Nicholls, I. A., Whitford, D. J., Harris, K. L., and Taylor, S. R., 1980, Variation in the geochemistry "I of mantle sources for tholeiitic and calc-alkaline mafic magmas, Western Sunda , Indonesia: Chemical Geology, v. 30, p. 177-199. where i2erfc (x) is the second integral of the complementary error function. Oh, C. W., Liou, J. G., and Krogh, E. J., 1988, A for the eclogite and related facies at high pressure metamorphism: Geological Society of America Abstracts with Programs, v. 20, Rewriting equation (B4) in terms of depth: p. A344. Pawley, A. R., and Holloway, J. R., 1992, Stability of hydrous minerals in subduction zones [abs.]: Eos (American Geophysical Union Transactions), v. 73, p. 141. Qo Zf Qo + TV 2 T(w, z) = — w + — 4 i erfc (B5) Peacock, S. M., 1990a, Fluid processes in subduction zones: Science, v. 248, p. 329-337. k S 1 k 2Vi

MANUSCRIPT RECEIVED BY THE SOCIETY FEBRUARY 25, 1992 0.5 Qo Zf TK / Kzf (C3) REVISED MANUSCRIPT RECEIVED OCTOBER 22,1992 V TT V sin 8 MANUSCRIPT ACCEPTED OCTOBER 23,1992

Printed in U.S.A.

694 Geological Society of America Bulletin, May 1993

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/105/5/684/3381769/i0016-7606-105-5-684.pdf by guest on 29 September 2021