Invited review doi: 10.1111/j.1365-3121.2008.00843.x Plate tectonics, flood basalts and the evolution of Earth’s oceans

Jun Korenaga Department of Geology and Geophysics, Yale University, PO Box 208109, New Haven, CT 06520-8109, USA

ABSTRACT The chemical composition of the bulk silicate Earth (BSE) thicker, buoyant basaltic crust, and the subductability of indicates that the present-day thermal budget of Earth is likely oceanic lithosphere is a critical factor regarding the emer- to be characterized by a significant excess of surface heat loss gence of plate tectonics before the Proterozoic. Moreover, over internal heat generation, indicating an important role of sluggish convection in the past is equivalent to reduced secular cooling in Earth’s history. When combined with secular cooling, thus suggesting a more minor role of mantle petrological constraints on the degree of secular cooling, this plumes in the early Earth. Finally, deeper ocean basins are thermal budget places a tight constraint on permissible heat- possible with slower plate motion in the past, and Earth’s flow scaling for mantle convection, along with implications for oceans in the Archean is suggested to have had about twice as the operation of plate tectonics on Earth, the history of mantle much water as today, and the mantle may have started as dry plumes and flood basalt magmatism, and the origin and and have been gradually hydrated by subduction. The global evolution of Earth’s oceans. In the presence of plate tectonics, water cycle may thus be dominated by regassing, rather than hotter mantle may have convected more slowly because it degassing, pointing towards the impact origin of Earth’s generates thicker dehydrated lithosphere, which could slow oceans, which is shown to be supported by the revised down subduction. The intervals of globally synchronous composition of the BSE. orogenies are consistent with the predicted variation of plate velocity for the last 3.6 Gyr. Hotter mantle also produces Terra Nova, 20, 419–439, 2008

2005; Labrosse and Jaupart, 2007), example, are often thought to be more Introduction and a review on so far published active in the past, but such temporal The thermal history of Earth, after the models on Earth’s thermal evolution trend is in direct conflict with what initial period of global magma ocean, has recently been published (Korena- thermal evolution indicates. Also, the is governed by the balance of surface ga, 2008b). Estimating a thermal his- volume of oceans is unlikely to be heat loss and internal heat generation. tory involves both geophysics (the constant, and it is probably decreasing An initially hot Earth has been grad- physics of convective heat loss) and with time. Whereas these topics tend ually cooled down with time because, geochemistry (the abundance of heat- to be considered independently, each for most of Earth’s history, surface producing elements), and many pub- of them represents a different aspect heat loss has been greater than internal lished models do not account for both on the evolution of the integrated heat supply by radiogenic isotopes. aspects simultaneously. Available geo- Earth system, so a proper understand- This simple energy balance places first- chemical and geological data place a ing of Earth’s thermal history is order constraints on various aspects of strict constraint on the permissible essential to make self-consistent pre- physical and chemical processes oper- range of thermal history, and the most dictions about them. ating on the surface of Earth and likely scenario is briefly reviewed in the within its deep interior. There have next section. Thermal history of earth been a number of theoretical studies The purpose of this review article is on Earth’s thermal history (e.g. to discuss some of major geological Global heat balance equations McKenzie and Weiss, 1975; Davies, processes that are directly related to 1980; Schubert et al., 1980; Christen- the thermal evolution of Earth, on the The thermal history of Earth may be sen, 1985; Richter, 1985; Solomatov, basis of this latest understanding. modelled by integrating the following 2001; Korenaga, 2003; Grigne et al., Topics covered include the onset of global energy balance (e.g. Stevenson plate tectonics, the history of mantle et al., 1983): plumes, continental growth, and the Correspondence: Jun Korenaga, Depart- origin and evolution of Earth’s dTmðtÞ ment of Geology and Geophysics, Yale Cm ¼ HmðtÞQðtÞþQCMBðtÞ; University, PO Box 208109, New Haven, oceans. They are difficult problems, dt CT 06520-8109, USA. Tel.: +1 (203) 432 and little can be said with certainty at ð1Þ present. Because they are intimately 7381; fax: +1 (203) 432 3134; e-mail: and [email protected] coupled with thermal evolution, how- ever, we can still derive a few impor- dTcðtÞ dmicðtÞ This article was commissioned by the tant constraints. These constraints C ¼ðL þ E Þ þ H ðtÞ c dt g dt c Editors of Terra Nova to mark the Inter- may be counter-intuitive or not obvi- national Year of Planet Earth. ous at first sight. Mantle plumes, for QCMBðtÞ; ð2Þ

2008 Blackwell Publishing Ltd 419 Plate tectonics, flood basalts and Earth’s oceans • J. Korenaga Terra Nova, Vol 20, No. 6, 419–439 ...... where the subscripts m and c denote Stevenson et al., 1983). Thus, the first As the physical state of the early the mantle and core components, two terms on the right-hand side of Earth is uncertain, the above differen- respectively, Ci is heat capacity, Ti(t) Eq. (2) may be neglected for simplic- tial equation is integrated backward in is average temperature, and Hi(t)is ity, and Eqs (1) and (2) may be time, starting with the present-day internal heat production. Core heat combined to yield: condition. flux, Q (t), represents heat CMB dT ðtÞ exchange between the mantle and the m C HmðtÞQðtÞ Present-day thermal budget core, and Q(t) is the surface heat flux. dt The mass of the inner core is denoted dDTCMBðtÞ The present-day thermal budget, i.e. þ Cc ; ð3Þ by mic(t), and the first term on the dt Q(0) and Hm(0), may be estimated as right-hand side of Eq. (2) describes the where C is the heat capacity of the follows. The global heat flux is esti- release of latent heat and gravitational mated to be 46 ± 3 TW (Jaupart entire Earth [=Cm+Cc, 7 · energy associated with the growth of 1027 JK)1 (Stacey, 1981)] and et al., 2007), which is the sum of heat the inner core. As mantle evolution is flux due to mantle convection, Q(0), DTCMB(t) is the temperature contrast tracked by a single average tempera- at the core–mantle boundary. The and radiogenic heat production in ture, this formulation corresponds to present-day contrast is estimated to continental crust, Hcc(0). The latter is whole-mantle convection. Though be on the order of 1000 K (e.g. Boeh- estimated to be 7.5 ± 2.5 TW (Rud- layered-mantle convection has been a ler, 1996; Williams, 1998), but how it nick and Gao, 2003), so we may write popular concept (e.g. Jacobsen and should change with time is not under- Hccð0Þ¼7:5 þ 2:5e1 ½TW; ð5Þ Wasserburg, 1979; Richter, 1985; All- stood well. To estimate the temporal egre et al., 1996; Kellogg et al., 1999), variability of the temperature con- and whole-mantle convection is probably trast, we need to calculate the core sufficient to explain available geophys- Qð0Þ¼46 Hccð0Þþ3e2 ½TW; ð6Þ heat flux QCMB(t), which depends ical and geochemical data if the uncer- critically on the poorly known rheo- where e1 and e2 are random variables tainties of these data and the logy of the lowermost mantle (e.g. following the Gaussian distribution of limitation of theoretical predictions Solomatov, 1996; Karato, 1998; zero mean and unit standard devia- are taken into account (Lyubetskaya Korenaga, 2005a) and also on other tion. The use of two independent and Korenaga, 2007b; Korenaga, complications such as the presence of random variables signifies that the 2008b). Note that layered mantle con- chemical heterogeneities (e.g. Farne- uncertainty of the global heat flux vection is still possible. Because it is tani, 1997; Jellinek and Manga, 2002), and that of continental heat produc- not well defined (Korenaga, 2008b), the post-perovskite phase transition tion are not correlated. Assuming however, exploring its implication for (e.g. Murakami et al., 2004; Oganov whole-mantle convection, heat pro- Earth’s evolution would require us to and Ono, 2004), and drastic changes duction in the bulk silicate Earth deal with more degrees of freedom. in thermal properties (e.g. Badro (BSE) (i.e. crust and mantle) is esti- For simplicity, therefore, we restrict et al., 2004; Lin et al., 2005). We do mated to be 16 ± 3 TW (Lyubetskaya ourselves to whole-mantle convection not even know why the present-day and Korenaga, 2007b), so heat pro- in this review, which should serve as contrast happens to be 1000 K. duction in the convecting mantle may a reference when considering more Thus, it may be reasonable to treat be expressed as: complex convection models (Table 1 dDT (t) ⁄dt as a free parameter and CMB H ð0Þ¼16 H ð0Þþ3e ½TW; ð7Þ presents a summary of major assump- see how thermal evolution would be m cc 3 tions made in this article). affected by this term. As the simplest where e3 is another random variable Internal heat production in the core option, it is set to zero here, and other following the same Gaussian distribu- is controversial (e.g. Gessmann and possibilities will be explored later (see tion, and the Ôconvective Urey ratio’ Wood, 2002; McDonough, 2003; Core heat flux and the possibility of (Korenaga, 2008b) is defined as: Rama Murthy et al., 2003; Lee et al., superheated core). With these simpli- 2004; Lassiter, 2006), but even if the fying assumptions, the global heat Ur(0) ¼ Hmð0Þ=Qð0Þ: ð8Þ core contains potassium at 100 p.p.m. balance can be expressed as: level, Hc would be only 0.7 TW at These uncertainties are relatively present. The energy release due to the dTmðtÞ large, allowing unrealistic values [e.g. C ¼ HmðtÞQðtÞ: ð4Þ inner core growth is similarly small, dt negative values for Hm(0)], so the being on the order of 2 TW (e.g. following additional constraints are imposed: 5 TW £ Hcc(0) £ 14 TW and Table 1 List of major assumptions employed in this article. Hm(0) ‡ 3 TW. These a priori bounds Assumption Relevant section are based on published composition models for the continental crust and Earth’s mantle convects as a single layer Thermal History of Earth the convecting mantle (e.g. Jochum Internal heat production in the core is negligible Thermal History of Earth et al., 1983; Taylor and McLennan, Mantle dehydration by melting controls global mantle dynamics Thermal History of Earth 1985; Wedepohl, 1995; Rudnick and Mean sea level has always been close to mean continental level History of Ocean Volume Gao, 2003; Salters and Stracke, 2004; Seafloor age–area relationship follows a triangular distribution History of Ocean Volume Workman and Hart, 2005). The Zero-age depth of seafloor has been approximately constant History of Ocean Volume resulting probability distribution Average thickness of continental crust has been approximately constant History of Ocean Volume functions are shown in Fig. 1. It can

420 2008 Blackwell Publishing Ltd Terra Nova, Vol 20, No. 6, 419–439 J. Korenaga • Plate tectonics, flood basalts and Earth’s oceans ......

14 tional to the temperature difference 14 44 42 between the surface and the interior. It 12 12 40 has been repeatedly shown that the 10 10 38 positive temperature dependency does (TW) (TW) (TW) m cc 8 36 not produce a sensible thermal history Q H H 8 6 34 consistent with the present-day ther- 32 mal budget (e.g. Christensen, 1985; 6 4 30 Korenaga, 2003). One could resolve 0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5 Ur Ur Ur this problem by postulating a high Urey ratio at present (e.g. 0.7) Fig. 1 Joint probability distribution functions for continental crust heat production (Davies, 1980, 1993, 2007; Schubert Hcc, mantle heat production Hm, convective heat flux Q, and the convective Urey et al., 1980; Schubert and Reymer, ratio Ur, based on Eqs (5)–(8) with the a priori bounds described in the text. Brighter 1985; Williams and Pan, 1992; Mc- colour denotes higher probability. Stars correspond to mean values, and ellipses to Namara and van Keken, 2000), but 68% confidence regions assuming the normal distribution. such high Urey ratio appears to be in conflict with geochemical data (Mc- Donough and Sun, 1995; Lyubetskaya be seen that Hcc(0) 8 TW, Hm(0) (e.g. Schubert et al., 1980; Christen- and Korenaga, 2007a). Future anti- 8.5 TW, Q(0) 38 TW, and Ur(0) sen, 1985; Gurnis, 1989; Solomatov, neutrino data should give direct con- 0.22. For the sake of simplicity, 1995; Conrad and Hager, 1999b; Ko- straints on mantle radioactivity (Araki thermal evolution models in this paper renaga, 2003). Figure 2 shows three et al., 2005; McDonough, 2005). are all calculated with these mean representative scaling laws (see Kore- Recently, there were some attempts values, but it is important to keep in naga (2006) for their derivations). The to rescue the conventional scaling by mind that the present-day thermal Ôconventional’ scaling predicts higher invoking temporal fluctuations in budget has non-trivial uncertainties heat flux for hotter mantle, and the plate tectonics (e.g. Grigne et al., and that some uncertainties are temperature dependency is dictated by 2005; Silver and Behn, 2008), but strongly correlated [e.g. Hm(0) and the activation energy of mantle vis- these studies have been suggested to Ur(0)]. See Korenaga (2008b) for the cosity [a few hundred kJ mol)1, (Ka- be inconsistent either with known sea effects of such uncertainties on the rato and Wu, 1993)]. Even in the limit level changes (Korenaga, 2007a) or prediction of thermal histories. of zero activation energy (Ôisoviscous with the physics of heat transfer (Ko- convection’ in Fig. 2), the temperature renaga, 2008a). It is also important to dependency of heat flow is still posi- understand what underlies the con- Internal heating and surface heat flux tive because heat flow is also propor- ventional scaling (e.g. Howard, 1966); At present, the convecting mantle contains only 50% of heat-produc- ing elements in the BSE, but it must have had a larger fraction when the mass of continental crust was smaller 200 in the past. This may be expressed as: Conventional 100 HmðtÞ¼Hm0ðtÞþð1 FCðtÞÞHccðtÞ; scaling ð9Þ Isoviscous convection where Hm0(t) and Hcc(t) denote the 50

(hypothetical) evolution of internal (TW) heating in the mantle and the conti- Q nental crust, respectively, without "Plate tectonics" mass transfer between them, and 20 FC(t) is the mass fraction of continen- tal crust with respect to the present- day value [assuming FC(t) £ 1 for 10 all t]. The calculation of these internal heating functions is straightforward 1200 1300 1400 1500 1600 1700 1800 once the present-day values are spec- Tm(°C) ified (e.g. Spohn and Breuer, 1993; Grigne and Labrosse, 2001). Fig. 2 Three classes of heat-flow scaling for mantle convection. ÔConventional’ and The evolution of surface heat flux Ôisoviscous’ parametrizations are both based on the heat-flow scaling of simple Q(t) has to be estimated through heat- thermal convection. Mantle viscosity is temperature-dependent for the former, with flow scaling for mantle convection as the activation energy of 300 kJ mol)1, and it is constant for the latter. ÔPlate tectonics’ there are no direct observations on scaling incorporates the effects of mantle melting beneath mid-ocean ridges such as global heat flow in the past. This dehydration stiffening and compositional buoyancy. See Korenaga (2006) for scaling issue has been controversial derivations.

2008 Blackwell Publishing Ltd 421 Plate tectonics, flood basalts and Earth’s oceans • J. Korenaga Terra Nova, Vol 20, No. 6, 419–439 ...... hotter mantle is predicted to yield importance of grain growth kinetics in subject to further investigation, higher heat flux because the boundary mantle convection (particularly in the including fully self-consistent numeri- layer (i.e. plates) becomes thinner lower mantle) is not well understood cal modelling. Important points are owing to greater convective instabil- yet, heat-flow scaling with mantle that there is no a priori reason for ity. Conventional scaling and substan- melting depends mainly on upper Earth’s mantle to follow the conven- tial boundary-layer growth are thus mantle properties, which are reason- tional scaling, and that we now have mutually exclusive, but mixing these ably well known. Thus, the scaling of some physically plausible mechanisms opposing concepts seems to be re- Korenaga (2006) (Ôplate tectonics’ in that may modify drastically the appar- quired if intermittent plate tectonics Fig. 3) is adopted for thermal history ent temperature sensitivity of convec- were to moderate surface heat flux. calculations here. Note that the con- tive heat flux. It appears, therefore, that some cept of hotter and stiffer mantle (So- The calculated thermal history is kind of negative temperature depen- lomatov, 1996) could enhance the presented with three different models dency is essential to prevent the so- negative temperature dependency fur- of continental growth (Fig. 3). called thermal catastrophe (Fig. 3a), ther. As will be suggested later, the Table 2 lists key model parameters and such scaling may be possible if the mantle may have been drier on aver- and their (present-day) values adopted effects of grain growth kinetics (Solo- age, which would also help suppress- here. The present-day mantle poten- matov, 1996, 2001) or mantle melting ing convective heat flux from a hotter tial temperature is assumed to be associated with plate tectonics (Kore- mantle in the past. These suggestions 1350 C (e.g. Kinzler, 1997; Herzberg naga, 2003, 2006) are important for for unconventional heat-flow scaling et al., 2007). It can be observed that global mantle dynamics. Whereas the are far from being established and the detail of continental growth is not important; radically different growth models give rise to only 100 K (a) (b) difference in the early Earth (Fig. 3a). 1800 Perhaps the most important feature is 1.0 Thermal Armstrong [1981] the robustness of model predictions catastrophe Campbell [2003] 1700 0.8 owing to a negative feedback imple- mented by the adopted heat-flow scal- 1600 0.6 ing. Lower heat flux for hotter mantle implies that internal heating may have 1500 0.4 McLennan & Talyor [1982] been higher than surface heat flux 1400 0.2 sometime in the past (Fig. 3c), and Ur(0) = 0.22 also that plate tectonics was more 1300 Fraction of cont. crust 0.0 sluggish in the past (Fig. 3d). Mantle potential temp. (°C) 0 1 2 3 4 0 1 2 3 4 Time B.P. (Ga) Time B.P. (Ga) Before interpreting these predic- tions any further, at least two issues (c) (d) need to be discussed. The first one is 60 10 (Gondwanaland) Rodinia Pangea Nuna Kenorland First GCO? the validity of the assumed heat-flow scaling for plate tectonics. The physics 8 Q of plate tectonics is not fully under- 40 stood yet (e.g. Bercovici et al., 2000;

(TW) 6

m Bercovici, 2003). We still do not H 4

& know, for example, why Earth exhib- 20 its plate tectonics and other terrestrial

conv H m 2 planets do not. There are various Q suggestions (e.g. Tozer, 1985; Regen- 0 Plate velocity (cm/year) 0 0 1 2 3 4 0 1 2 3 4 auer-Lieb et al., 2001; Korenaga, Time B.P. (Ga) Time B.P. (Ga) 2007b), many of which call for the existence of surface water, but we do Fig. 3 Thermal evolution modelling with continental growth, starting with the not have a quantitative understanding present-day mantle temperature of 1350 C and the present-day convective Urey ratio for under what conditions plate tec- of 0.22. The global heat balance of Eq. (4) is assumed, and the details of integration tonics can take place. Without such procedure can be found in Korenaga (2006). The heat-flow scaling of plate tectonics fundamental understanding, then, (Fig. 2) is adopted. (a) The history of mantle potential temperature T . The result m why should we trust the proposed with the conventional heat-flow scaling is also shown by dashed line. (b) Three models of continental growth: Ôinstantaneous’ growth (Armstrong, 1981), Ôgradual’ growth scaling for plate tectonics? The answer (McLennan and Taylor, 1982), and somewhere in-between (Campbell, 2003). lies in the nature of scaling laws. Heat- (c) Predicted history of convective heat flux (solid) and mantle internal heating flow scaling prescribes the sensitivity (dashed). Different colours correspond different models of continental growth. It may of surface heat flow to variations in mantle temperature, and all of scaling be seen that the Urey ratio, Ur(t)=Hm(t) ⁄ Q(t), gradually increases from the present-day value of 0.22 to higher values in the past. (d) Predicted history of average laws in Fig. 2 are normalized with plate velocity. Dashed lines denote the range of geological estimate based on the respect to the present-day convective intervals between supercontinent assemblies or globally synchronous orogenies (Kore- heat flow. We know plate tectonics is naga,2006),thetimingofwhichareshownbygreenbars(Hoffman,1997;Nutman,2006). operating today, and we can estimate

422 2008 Blackwell Publishing Ltd Terra Nova, Vol 20, No. 6, 419–439 J. Korenaga • Plate tectonics, flood basalts and Earth’s oceans ......

Table 2 Key parameters used in mantle and ocean evolution modelling. prediction based on thermal evolution modelling (Fig. 3d). One could ques- Parameter Definition Value* Note tion this connection between the his- ) C Heat capacity of the entire Earth 7 · 1027 JK 1 Eqs (3) and (4) tory of supercontinents and the vigour Tm(t) Mantle potential temperature 1350 C Eqs (3) and (4) of plate tectonics by suggesting that Hm(t) Heat production in the mantle 8.5 TW Eqs (3), (4) and (7); Fig. 1 plate tectonics could have been pro- Q(t) Convective heat fluxà 38 TW Eqs (3), (4) and (6); Fig. 1 ceeding vigorously for long periods 27 )1 Cc Heat capacity of the core 1.3 · 10 JK Eq. (3) with no significant change in conti- DTCMB(t) Temperature contrast at CMB§ 1000 K Eq. (3) nental configuration, but as noted Hcc(t) Heat production in continental crust– 8 TW Eq. (9); Fig. 1 earlier (see Internal heating and sur- smax(t) Maximum age of seafloor** 180 Ma Eq. (23) ) face heat flux), more vigorous convec- v(t) Average plate velocity 4cmyr 1 Eq. (23) 14 3 tion (i.e. higher heat flow) in the past Ao(t) Total area of ocean basins 3.1 · 10 m Eqs (16) and (23) ) G(t) Plate creation rate 3.45 km2 yr 1 Eqs (22), (15) and (23) may not allow us to construct a V(t) Total ocean volume 1.51 · 1018 m3 Eq. (18) reasonable thermal history. Nonethe- less, the speed of continental motion d0(t) Zero-age depth of seafloor 2654 màà Eq. (19) ) b(t) Seafloor subsidence rate 323 m Myr 1 ⁄ 2 Eqs (19) and (20) may be better regarded as a minimum estimate on the global average, so the *For parameters that depend on time t, present-day values are listed. agreement with model prediction Past heat production is calculated based on the present-day value and U : Th : K of 1 : 4 : 1.27 · 104 [see should not be taken at face value. Eq. (3) and Table 1 of Korenaga (2006)]. The effect of continental growth is taken into account by Eq. (9). The interpretation of geological àPast convective heat flux is calculated using the Ôplate tectonics’ scaling law shown in Fig. 2. records prior to c. 3 Ga has been §This is assumed to be constant in Figs 3, 4 and 9. The effect of its possible temporal variation on mantle controversial. It is important to keep evolution is minor (see Section on Core heat flux and the possibility of superheated core), though it is in mind that, even if plate tectonics important for the thermal history of the core (Fig. 5). was operating in the early Archean, it –Past heat production in the continental crust is calculated in a similar way to that in the mantle but with does not have to resemble modern- U : Th : K of 1 : 5 : 104. style plate tectonics. The most impor- **The triangular distribution of seafloor age-depth is assumed [Eq. (15)]. 2 tant ingredient is the subduction of Past plate velocity is calculated using its relationship to convective heat flux, v / (Q ⁄ Tm) [Eq. (18) of Korenaga (2006)]. oceanic lithosphere, and other aspects ààThis is assumed to be constant in Fig. 9. of plate tectonics could have been different. It follows that a mere change in the style of tectonics [e.g. at how potential energy release is bal- require the closure of ocean basins c. 3.2 Ga as suggested by the geology anced by viscous dissipation for the that existed between different conti- of the Pilbara Craton in Australia present-day mantle (e.g. Conrad and nents. The subduction of oceanic lith- (Van Kranendonk et al., 2007)] does Hager, 1999a). Heat-flow scaling is osphere should be involved in closing not necessarily coincide with the emer- constructed by considering how this ocean basins, so the timing of ancient gence of plate tectonics. Plates can energy balance should change for orogenies provides important con- coexist with vertical tectonics, as cur- different temperatures. As far as plate straints on the first emergence of plate rently observed at the Basin and tectonics is operating, therefore, the tectonics on Earth. The oldest Ôsuper- Range (Sleep, 2007). In fact, the assumed heat flow scaling is unlikely continent’ Kenorland was assembled operation of plate tectonics as early to be grossly in error, though the at c. 2.7 Ga (Hoffman, 1997), so it as 3.7–3.8 Ga was suggested by Ko- details of viscous dissipation in the would be reasonable to assume the miya et al. (1999) based on the Isua mantle could be debatable. This operation of plate tectonics in the late supracrustal belt in Greenland, and brings up the second issue, which is Archean. Whether Kenorland was a the age of episodic granite production the onset of plate tectonics on Earth. true supercontinent or not (Bleeker, recorded in the Istaq Gneiss Complex Figure 3 shows the backward inte- 2003) is not so important in the of Greenland (Nutman et al., 2002) as gration of global heat balance to the current context; there is field evidence well as the Narryer Gneiss Complex of beginning of Earth’s history (4.6 Ga), for globally synchronous orogeny at Western Australia (Kinny and Nut- and this is equivalent to assuming that c. 2.7 Ga, which is sufficient for discus- man, 1996) may point to an early plate tectonics started as soon as sion here. Also, recycling of oceanic global collisional orogeny at c. 3.6 Ga Earth was created. However, if plate crust by subduction back to c. 3Gais (Nutman, 2006). Interestingly, this tectonics started at 3 Ga, for example, supported by the age of eclogite xeno- timing is consistent with what thermal model predictions for older times bear liths from subcontinental lithospheric evolution indicates (Fig. 3d). little significance. This issue is mantle (e.g. Jacob et al., 1994; Shirey The possibility of plate tectonics in discussed in detail next. et al., 2001). As noted by Korenaga even earlier times has also been widely (2006), the time intervals of so far debated, and it is centred on the known five supercontinents (Pangea, interpretation of the Hadean Onset of plate tectonics Gondwanaland, Rodinia, Nuna and (>4 Ga) detrital zircons (e.g. Mojzsis Kenorland) may be used to estimate et al., 2001; Wilde et al., 2001; Harri- Geological and geochemical evidence the spatially and temporally averaged son et al., 2005; Valley et al., 2005; The assembly of continental masses rate of plate motion, and such esti- Coogan and Hinton, 2006; Grimes and associated collisional processes mate appears to be consistent with the et al., 2007). Some authors have

2008 Blackwell Publishing Ltd 423 Plate tectonics, flood basalts and Earth’s oceans • J. Korenaga Terra Nova, Vol 20, No. 6, 419–439 ...... suggested the existence and recycling (6.4 · 106 m). For a 300-km-diame- negatively buoyant, at least on aver- of continental crust as early as 4.5 Ga ter impactor, DT is only 15 K. Of age, with respect to the hot interior. (Harrison et al., 2005), which implies course, impact heating is a highly This buoyancy constraint is not an the operation of plate tectonics (or localized phenomenon, and a much issue for simple thermal convection just recycling of oceanic lithosphere) greater temperature rise is expected in with a homogeneous material because shortly after the formation of Earth. the vicinity of an impact site. The late the top thermal boundary layer is The interpretation of out-of-context heavy bombardment is certainly always denser than the hot interior. detrital zircons is, however, not un- important for surface conditions as it Convection in Earth’s mantle, how- ique especially for the Hadean era could vaporize at least a part of ever, induces pressure-release melting (e.g. Sleep, 2007), unless we invoke oceans. The above order-of-magni- and differentiates the shallow mantle Uniformitarianism for relevant geo- tude estimate is probably sufficient, into basaltic crust and depleted mantle logical processes. What the Hadean however, to show that the effect of lithosphere, both of which are less zircons actually constrain seems to be even the biggest impactor during the dense than the mantle before melting an open question. A related issue is late bombardment would not be if compared at the same temperature. that, if continental crust existed in the significant for global mantle dynam- For the top boundary layer to become Hadean, how voluminous it could ics such as the operation of plate denser than the interior, therefore, it have been then, and it will be dis- tectonics. has to be cooled for a sufficiently long cussed later (see Continental growth: time so that negative buoyancy due to instantaneous, gradual, or discontinu- thermal contraction overcomes the Geodynamical considerations ous?). intrinsic chemical buoyancy [note that Estimating the onset of plate tectonics basalt is transformed to denser eclo- on a purely theoretical ground is gite at 60 km depth (Ringwood and A note on late heavy bombardment difficult because the very physics of Green, 1964), so this chemical buo- On the basis of the lunar cratering plate-tectonic convection is not well yancy issue is important only for the record, it has been suggested that understood (e.g. Bercovici et al., initiation of subduction, not for its Earth experienced a spike of meteorite 2000). There still exist, however, cer- continuous operation]. For the pres- bombardment at 4.0–3.8 Ga (Hart- tain physical constraints that have to ent-day condition, this time-scale is mann et al., 2000; Ryder, 2002). This be considered when discussing the estimated to be c. 20 Myr, but when event is called as the late heavy style and vigour of mantle convection the mantle is hotter than present by bombardment, and it is sometimes in the early Earth. In particular, the 200–300 K, the time-scale could be on discussed as if it could have had a interpretation of geochemical data is the order of 100 Myr (Davies, 1992). strong influence on the stability of often assisted by or compared to That is, when the mantle was hotter in lithosphere (and thus the operation of geodynamical considerations (some- the Hadean, it must have had thicker plate tectonics) (e.g. Shirey et al., times subconsciously), and this is buoyant crust as well, which could 2008). The total mass of impactors where a preconception could poten- return to the interior only after c. 100- that hit the Moon during this period is tially lead to biased interpretations. Myr long surface cooling. This time- estimated to be 3 · 1018 kg (Ryder, For example, vigorous convection is scale is comparable to the present-day 2002), and by extrapolation, Earth is often assumed for the Hadean dynam- time-scale to renew ocean basins. expected to have been hit by a few big ics (e.g. Sleep and Zahnle, 2001), but it Vigorously convecting hotter mantle (>300 km diameter) impactors and is not clear if such vigour can really be could be at odds with more copious numerous smaller ones (about a dozen attained in the early Earth. Though melting expected for such mantle. of 200-km-diameter impactors, a convection in a magma ocean should The subductability argument may hundred of 100-km-diameter impac- have been extremely vigorous (Solo- be countered because thick oceanic tors, and so on) (e.g. Sleep, 2007). The matov, 2007), the transition from such crust such as oceanic plateaus does mass of a 300-km-diameter impactor intense liquid convection to solid-state subduct at present (e.g. Sleep and would be 4 · 1019 kg, and its influ- mantle convection is currently poorly Windley, 1982), but such thick crust ence of the thermal state of Earth’s understood [see Sleep (2007) for the is localized at present as opposed to its lithosphere may be estimated by possibility of Ômush ocean’]. If mantle likely global occurrence in the Arche- equating its potential energy and the convection was in the mode of plate an. The argument may also be ques- energy required to heat up the litho- tectonics, for example, the scaling of tioned because the seafloor is sphere by DT as: Korenaga (2006) might be applicable, currently subducting irrespective of which predicts slower, not faster, plate its age (Parsons, 1982). Indeed, near GMEm MLcpDT ; ð10Þ motion. Even without calling for this zero-age crust is subducting under the r E particular scaling, however, a simple western margin of the North Ameri- where ML is the mass of lithosphere physical argument centred on the can plate, whereas the oldest crust in (1.7 · 1023 kg for the thickness of subductability of oceanic lithosphere the Atlantic (c. 180 Ma) is not sub- 3 100 km), cp is its specific heat (10 (e.g. Bickle, 1986; Davies, 1992) seems ducting at all, so the simple buoyancy Jkg)1 K)1), G is the gravitational to question rapid convection when argument may seem to be irrelevant to ) ) ) constant (6.67 · 10 11 m3 kg 1 s 2), Earth was much hotter than today. actual plate dynamics. Young sub- ME is the mass of Earth (6 · For the top boundary layer to be ducting seafloor is, however, attached 1024 kg), m is the mass of the impac- recycled into the interior by free con- to an older, already subducted plate, tor, and rE is the radius of Earth vection, the boundary layer must be and the presence of old, non-subduct-

424 2008 Blackwell Publishing Ltd Terra Nova, Vol 20, No. 6, 419–439 J. Korenaga • Plate tectonics, flood basalts and Earth’s oceans ...... ing seafloor tells us that negative negative buoyancy would be too long Subductability thus remains to be buoyancy alone is not sufficient to (i.e. c. 100 Myr) to be achieved in an important factor to be considered, initiate subduction. Probably a more vigorously convecting mantle, but this and the backward integration of adequate measure is the global aver- argument depends on the conventional Earth’s thermal history seems to pro- age of the age of subducting seafloor, heat-flow scaling (Fig. 2). Also, van vide an important perspective related which is c. 49 Ma (rate average) or Thienen et al. (2004) suggested that to subductability, namely, the role of c. 61 Ma (area average) [based on plate tectonics may not be possible internal heating on the initiation of Table 1 of Parsons (1982)]. In either when the mantle potential temperature plate tectonics. Figure 4 shows two case, it is old enough to achieve is higher than 1500 C based on sub- kinds of lithospheric thickness. One is sufficient negative buoyancy under ductability, but their calculation as- the minimum thickness of subductable the present-day condition. sumes the so-called plate model for the oceanic lithosphere [which is equiva- The concept of subductability evolution of oceanic lithosphere (Par- lent to the Ôcritical thickness’ defined depends on two factors: thermal dif- sons and Sclater, 1977; Stein and Stein, by Korenaga (2006)]; lithosphere must fusivity and intrinsic chemical buoy- 1992), in which the growth of litho- exceed this thickness to become neg- ancy. Whereas the former is virtually sphere is inhibited after c. 80 Myr. atively buoyant. This thickness is a constant under plausible mantle con- Though convective instability can limit function of mantle temperature, so its ditions, the chemical buoyancy is a the growth of lithosphere (Parsons and temporal variation is determined by function of mantle temperature, and McKenzie, 1978), this instability is a an assumed thermal history (Fig. 3a). the functionality is less certain. To function of mantle viscosity, and it is The other is the equilibrium thickness predict the density of oceanic crust, difficult to justify the use of a constant of a hypothetical stagnant lid, which is one has to first calculate the compo- maximum thickness over a range of a function of internal heat generation sition of primary mantle melt for a mantle temperature (cf. Korenaga, as (Korenaga, 2006): range of mantle temperature, and then 2003). Note that the plate model con- kAT calculate the density of mineral aggre- tains an artificial bottom boundary h m ; ð11Þ SL H gates expected to be solidified from a condition to suppress cooling, and a m given melt composition. The current recent global analysis of seafloor where k is thermal conductivity and A understanding of mid-ocean ridge topography casts a doubt on the is the total surface area of Earth. magmatism is probably sufficient to observational basis for this model At a thermal equilibrium, lithosphere conduct the first step with moderate (Korenaga and Korenaga, 2008). should become thinner for higher accuracy (e.g. Langmuir et al., 1992; Recently, Davies (2006) proposed internal heating, and Fig. 4 indicates Kinzler, 1997; Walter, 1998), but the that the subductability issue would that the equilibrium thickness may second step is more uncertain because not present a major obstacle for have been less than the minimum the details of crystallizing phases the operation of plate tectonics if the thickness before c. 2.5–3 Ga. In the depend on the temperature and pres- mantle was already depleted in the Archean and the Hadean, the amount sure conditions within newly forming early Earth by earlier melting events. of internal heat generation in the crust and thus on how exactly crust is In his numerical model, subduction is mantle was greater than present by a constructed. Even for the present-day forced by a surface velocity condition, factor of >2 (Fig. 3c), and this high oceanic crust, the physical mechanism and subducted oceanic crust segre- internal heating may have suppressed of crustal accretion is still under gates from depleted mantle litho- the growth of the top thermal bound- debates (e.g. Phipps Morgan and sphere and sinks to the lower mantle, ary layer and thus prevented litho- Chen, 1993; Korenaga and Kelemen, leading to a gradual depletion of the sphere to become negatively buoyant. 1998; Wilson et al., 2006), and we do upper mantle, the melting of which This argument is, however, probably not know, from first principles, how to does not yield thick oceanic crust. too simplistic because the equilibrium construct much thicker crust corre- This scenario, however, requires some thickness is unlikely to be achieved sponding to hotter mantle. The sub- kind of tectonics (other than plate instantaneously. The thermal adjust- ductability calculation by Korenaga tectonics) that can subduct oceanic ment time-scale can be long, on the (2006), for example, depends on the crust, and a time-scale to achieve the order of 1 Gyr (Daly, 1980), and this crustal density parametrization of required mantle depletion is uncertain long time-scale in a sense justifies Korenaga et al. (2002), which assumes [in the model of Davies (2006), sub- Eq. (1), in which surface heat flux low-pressure crystallization. Future duction was achieved by assuming the Q(t) is usually parameterized as a studies on the crustal structure of continuous operation of plate tecton- function of mantle temperature large igneous provinces may provide ics]. The once highly depleted upper (Fig. 2) and can vary independently important field constraints on this mantle would also need to be refertil- from internal heat production Hm(t). issue. ized later to explain the present-day The physics of thermal adjustment is, Provided that intrinsic chemical upper mantle. In other words, the however, not fully explored for mantle buoyancy remains significant for hot- upper mantle needs to change its convection with realistic rheology, ter mantle, subductability is a robust composition so that melting always and the significance of the thickness physical constraint, but its use needs yields relatively thin oceanic crust crossover in Fig. 4 is an open ques- some care. Davies (1992), for example, regardless of its temperature. The tion. As noted by Sleep (2000) and argued that plate tectonics was un- plausibility of this mechanism seems Stevenson (2003), a change in the likely when the mantle was much to hinge on a delicate balance between mode of mantle convection may lead hotter because the time-scale for mantle mixing and secular cooling. to non-monotonic thermal histories.

2008 Blackwell Publishing Ltd 425 Plate tectonics, flood basalts and Earth’s oceans • J. Korenaga Terra Nova, Vol 20, No. 6, 419–439 ......

Phanerozoic Proterozoic Archean Hadean of hotspot magmatism has been con- 350 troversial (e.g. Anderson, 1998; Foul- ger et al., 2005). Thermal anomalies 300 such as mantle plumes are certainly one way to generate hotspot islands "Equilibrium" thickness and flood basalts, but not the only of stagnant lid 250 way because chemical and ⁄or dynam- ical anomalies may also result in similar magmatic activities (e.g. Sleep, 200 1984; Tackley and Stevenson, 1993; Korenaga and Jordan, 2002; Ander- 150 son, 2005; Ito and Mahoney, 2005; Korenaga, 2005b). For the sake of discussion, however, 100 Plate thickness [km] Plate thickness let us assume that most of hotspots Minimum thickness of and flood basalts are formed by the 50 "subductable" oceanic lithosphere melting of mantle plumes that origi- nate in the core–mantle boundary region. In this case, the thermal evo- 0 0 1 2 3 4 lution of Earth suggests that hotspot magmatism should have been more Time B.P. [Ga] reduced in the past. As will be explained shortly, this is a straightfor- Fig. 4 The predicted variation of the minimum thickness of subductable oceanic ward consequence of a geologically lithosphere and the equilibrium thickness of stagnant lid, based on the thermal history plausible thermal history (Sleep et al., shown in Fig. 3. The equilibrium lid thickness is thinner in the past due to higher mantle heat production, and is thinner than the minimum subductable thickness 1988), though this fact does not seem before the Proterozoic, suggesting that the subductability of oceanic lithosphere is a to be widely recognized. For example, key factor for the emergence of plate tectonics in the early Earth. a plume-dominated regime is often suggested as an alternative mode of mantle convection in the Archean (i.e. Though it could be a coincidence, the instead of the plate-tectonic regime) Secular cooling and flood basalt timing of crossover is close to the (e.g. Fyfe, 1978; Van Kranendonk volcanism Archean-Proterozoic boundary, at et al., 2007), and some models for which surviving continental crust be- continental growth call for a promi- Did mantle plumes exist in the gan to emerge in abundance. A quan- nent role of flood basalts or mantle Archean? titative relationship between mantle plumes in the early Earth (e.g. Abbott convection and continental recycling Apart from metal-silicate segregation and Mooney, 1995; Albare` de, 1998). throughout Earth’s history is one of that took place within the first hun- Along with the notion of more vigor- the subjects that should be explored dred million years of Earth’s history ous convection, plume activities in the by future geodynamical studies. (e.g. Halliday, 2003), chemical differ- Archean are commonly assumed to To go beyond this simple buoyancy entiation in Earth’s interior refers to have been similar to or higher than argument, it is imperative to advance the melting of silicate rocks. The today. Indeed, the vigor of mantle our understanding of the physics of melting of shallow upper mantle usu- convection and the intensity of plumes plate tectonics. Though plate-tectonic- ally takes one of the following three may be related through the thermal like convection can be successfully types of surface manifestation: mid- budget of Earth, but their temporal simulated in numerical models (e.g. ocean ridge magmatism, arc magma- variations do not have to be positively Moresi and Solomatov, 1998; Tackley, tism and hotspot magmatism (e.g. correlated. It is possible, for example, 2000; Richards et al., 2001; Ogawa, Wilson, 1989; McBirney, 1993). The to have a reduced plume flux while 2003; Gurnis et al., 2004; Stein et al., first two are associated with plate maintaining the vigor of convection, 2004), currently available models treat tectonics. Here, the term hotspot and the thermal history shown in the strength of oceanic lithosphere as a magmatism is used in a broad sense Fig. 3 indicates that such possibility free parameter, which must be ad- to cover not only hotspots such as is likely. To discuss this further, we justed to achieve plate tectonics. It is Hawaii and Iceland but also continen- need to relax one of the assumptions difficult to discuss the onset of plate tal and oceanic flood basalt provinces behind Eq. (4) and consider the pos- tectonics on the basis of those models such as the Deccan Traps and the sibility of differential core cooling because the free parameter may not be Ontong Java Plateau (Coffin and Eld- using Eq. (3). constant over the geological time. It holm, 1994). This type of magmatism would desirable to predict the strength is commonly explained by the upwell- Core heat flux and the possibility of of lithosphere from first principles ing of mantle plumes (e.g. Morgan, superheated core based on tangible physical processes; 1971; Richards et al., 1989; Campbell such research effort is still in its and Griffiths, 1990; White and The reconstructed thermal history of infancy (e.g. Korenaga, 2007b). McKenzie, 1995) though the origin Fig. 3 is based on Eq. (4), in which the

426 2008 Blackwell Publishing Ltd Terra Nova, Vol 20, No. 6, 419–439 J. Korenaga • Plate tectonics, flood basalts and Earth’s oceans ...... mantle and the core are assumed to present-day value and is predicted to dynamics of the core–mantle bound- have cooled at the same rate. That is, be negative before the early Archean ary region; the rheology of the lower- the temperature contrast at the core– (i.e. core heating instead of cooling most mantle is currently unknown. To mantle boundary is assumed to have should take place then). With the explore the significance of time-depen- been constant. In this case, the core nearly constant surface heat flux but dent DTCMB, two different variations heat flux is directly related to the with gradually decaying radiogenic are considered (cases 2 and 3 in secular cooling of the mantle as: heat source (Fig. 3c), the secular cool- Fig. 5b). In case 2, the contrast ing of Earth and thus the core heat decreases linearly by 500 K over the dTmðtÞ QCMBðtÞ¼Cc ð12Þ flux are likely to have been lower in entire Earth history, whereas in case 3 dt the past. it decreases quadratically by a similar [from Eq. (2)]. The history of core The details of the predicted core amount. Thermal evolution was heat flux with this assumption is evolution are, however, subject to solved again, but using Eq. (3), and shown as case 1 in Fig. 5. The pres- large uncertainties. The temperature the corresponding core heat flux was ent-day core heat flux is estimated to contrast DTCMB, which is assumed to calculated as: be 6 TW, and more important, the be constant, is almost a free parameter dT ðtÞ dDT ðtÞ past core heat flux is lower than the given our limited knowledge of the Q ðtÞ¼C m C CMB : CMB c dt c dt ð13Þ (a) 12 It may be seen that differential core cooling could modify substantially 10 Case 2 predicted core heat flux (Fig. 5a). On the other hand, the thermal history of 8 Case 3 the mantle (not shown) is not affected much by this modification. This is expected because the differential core 6 cooling of case 2, for example, pro- vides additional core heat flux of 4 TW, which is significant for the core 4 Case 1

Core heat flux [TW] thermal budget, but not for the mantle thermal budget. Cases 1 through 3 are 2 all consistent with the observational constraints on the present-day core heat flux [6)12 TW, (Buffett, 2003)]. 0 0 1 2 3 4 At the core–mantle boundary, the core side is considered to be hotter Time B.P. (Ga) than the mantle side by at least 1000 K at present (Williams, 1998). (b) 600 The present-day contrast is uncertain because the estimate of the core-side 500 temperature is based on the phase diagram of the core, which depends (0) (K) 400 on its chemical composition and in

CMB particular on the (uncertain) light ele- T 300 Δ ment composition (e.g. McDonough, )–

t 2003). Nevertheless, a temperature ( Case 2 200 contrast more than a few hundred K CMB

T is probably robust, and this suggests Case 3 Δ 100 either (1) that there was no contrast at Case 1 the beginning of the Earth history, but 0 the mantle cooled faster than the core, 0 1 2 3 4 or (2) that the core was initially hotter than the mantle, and the temperature Time B.P. (Ga) contrast has not been entirely removed. The first scenario seems Fig. 5 (a) Predicted history of core heat flux QCMB incorporating the effect of differential core cooling. Different colours correspond to different models of unlikely because it would predict core continental growth (Fig. 3b). (b) Three cases of differential cooling are shown: 1, heat flux lower than case 1 and thus no differential cooling (the core cools at the same rate as the mantle), 2, constant rate becomes incompatible with the pres- of differential cooling (temperature contrast at the core–mantle boundary linearly ent-day core heat flux estimate and decrease by 500 K over 4.6 Gyr), and 3, decreasing rate of differential cooling (heat also with the history of the geo- flux due to differential cooling decreases linearly from 10 TW at 4.6 Ga to zero at magnetic field (e.g. McElhinny and present). Senanayake, 1980; Buffett, 2002;

2008 Blackwell Publishing Ltd 427 Plate tectonics, flood basalts and Earth’s oceans • J. Korenaga Terra Nova, Vol 20, No. 6, 419–439 ......

Tarduno et al., 2007). The second 2002). If the average plume size Another mechanism that may pro- possibility is physically plausible be- remains similar through time, then, duce large igneous provinces is the cause either the gravitational segrega- the number of plumes should have upwelling of chemically anomalous tion of the core or the Moon-forming been lower, and if we instead assume mantle that has been fertilized by giant impact is expected to deposit a the constant frequency of plume for- recycled oceanic or continental crust considerable amount of heat into the mation, the average plume size should (e.g. Korenaga and Kelemen, 2000; core (Solomatov, 2007). The initial have been smaller. In either case, we Yaxley, 2000; Anderson, 2005). More temperature of this superheated core expect a reduced volume of flood fertile mantle is usually intrinsically and its later evolution are uncertain, basalt magmatism in the past as a denser (O’Hara, 1975), so its upwell- but it would be fortuitous if the initial corollary of the mantle plume hypo- ing probably requires special tectonic temperature contrast has been main- thesis. This does not necessarily mean environments (Korenaga, 2004, tained to the present day; it is possible, that we cannot expect greater flood 2005b). Compensating chemical den- but it would require a specific dynam- basalt magmatism in the past. First of sity anomalies by thermal buoyancy is ics of the core-boundary boundary all, Ônormal’ mantle in the Archean is possible, but it would require unreal- region. Figure 5(a) suggests that the likely to be 300 K hotter than pres- istically hotter mantle [e.g. DT possibility of differential core cooling ent (Fig. 3a), so flood basalt magma- 500)600 K (Lin and van Keken, (or the current uncertainty of the core– tism in the Archean may not 2005)], which may not be consistent mantle boundary dynamics) provides necessarily require an unusual source with available petrological constraints important degrees of freedom for the mantle. High temperature alone, how- (DT 100)300 K) (e.g. White and coupled core–mantle evolution. ever, is probably insufficient to explain McKenzie, 1995; Herzberg, 2004). Explaining both the present-day core focused magmatic events such as flood Setting aside this dynamical difficulty, heat flux and the history of the geo- basalts, and there are a few non-plume it is also unclear how abundant such magnetic field has been a conundrum mechanisms as discussed below. fertile mantle would have been in the for the thermal history of Earth’s core One popular concept is the episodic past. For one thing, the recycling rate (e.g. Labrosse et al., 2001; Buffett, overturn of layered-mantle convection of oceanic crust is controlled by plate 2003; Nimmo et al., 2004; Butler et al., (e.g. Stein and Hofmann, 1994; Con- motion, which may not have been 2005), and these extra degrees of free- die, 1998; Rino et al., 2004). In the different from present as discussed dom may help to resolve it. layered-mantle convection mode, the earlier (Internal heating and surface A linear decrease in the tempera- lower mantle cools less efficiently and heat flux). On the other hand, the ture contrast (case 2) shifts core heat thus becomes hotter than the upper recycling of continental crust is free flux almost uniformly, thus unaffect- mantle. Numerical modelling in the from this constraint, and primordial ing the trend of lower heat flux in the early 1990s suggested that, whereas it heterogeneities created during the past. To reverse this trend, we need is difficult to maintain a purely layered magma ocean may have been abun- to invoke a greater temporal varia- state, episodically layered convection dant in the early Earth. The possibility tion in the past (e.g. case 3), but there may take place with the endothermic of fertile mantle is a wild card, as our is a negative feedback mechanism to phase change at the base of the mantle understanding of the dynamics of suppress core heat flux in the early transition zone (i.e. at the 660-km chemically heterogeneous mantle is Earth. Higher differential core cool- discontinuity) (e.g. Machetel and still immature (Korenaga, 2008b). ing is equivalent to higher internal Weber, 1991; Honda et al., 1993; Geological indicators for flood bas- heating in the mantle [Eq. (3)], which Tackley et al., 1993; Solheim and alts are common in Archean terranes would then reduce the secular cooling Peltier, 1994). When a temporally (e.g. Campbell et al., 1989; Ernst and of the entire Earth (i.e. including the layered state is broken, a large volume Buchan, 2003; Sandiford et al., 2004), core). This is why even case 3 pre- of the hot lower mantle material can which is probably the basis for the dicts vanishing core heat flux in the be brought to the surface, potentially notion of more active plume activities. early Earth (Fig. 5a). It appears that resulting in massive melting events. One way to reconcile the apparent core heat flux was probably lower Note that the plausibility of episodic discrepancy between the theoretical than (cases 1 and 2) or similar to overturns depends critically on the expectation and the field observation (case 3) the present-day value, and magnitude of the (negative) Clapeyron is to call for preservation bias; the substantially higher core heat flux in slope for the endothermic phase continental crust with flood basalts the early Earth probably requires an change. To insulate the lower mantle may have better survived for the unrealistic degree of differential core from cooling due to subducted slabs following reason. The genesis of flood cooling. and make it substantially hotter than basalts, produced by either thermal or the upper mantle, subducted slabs chemical anomalies, involves the melt- must be supported globally by the ing of a large volume of the mantle. Origins of flood basalts and phase change for several hundred Because mantle melting also deh- preservation bias million years. Recent experimental ydrates and stiffens the residual man- With reduced core heat flux, it would studies suggest that the Clapeyron tle (Karato, 1986; Hirth and be difficult to expect more vigorous slope is not as strongly negative as Kohlstedt, 1996), this large-scale melt- plume activities in the early Earth. previously thought (Katsura et al., ing would produce a voluminous, stiff Furthermore, plume heat flux is likely 2003; Fei et al., 2004), potentially mantle root, which could protect the to be smaller than the total core heat undermining the physical basis for overlying continental crust from flux (e.g. Davies, 1993; Labrosse, episodically layered convection. tectonic disturbances. It is unclear

428 2008 Blackwell Publishing Ltd Terra Nova, Vol 20, No. 6, 419–439 J. Korenaga • Plate tectonics, flood basalts and Earth’s oceans ...... how much of such dehydrated residual a few times in the past, and the use of maximum yield stress), which may mantle has contributed to what we single heat-flow scaling in thermal not be constant over Earth’s history. call today as continental tectosphere evolution modelling, as done in most Equally important, mantle viscosity is (Jordan, 1988; Pearson, 1999), but the of previous studies, may become a function of not only temperature, role of dehydrated mantle in ancient overly simplistic in this case. Thus, but also other parameters such as continental dynamics is an important the debate over continental growth grain size and water content (e.g. dynamical problem to consider (e.g. has a critical relevance to the theoret- Karato and Wu, 1993; Hirth and Doin et al., 1997; Lenardic and Moresi, ical formulation of thermal evolution Kohlstedt, 2003; Korenaga and Ka- 1999). Oceanic crust older than modelling. rato, 2008). Solomatov (1996), for c. 180 Myr old is all subducted, and The episodic overturn of layered- example, suggested that hotter mantle we have only continental crust to dis- mantle convection has been a popular could become more viscous if grain- cuss anything older. It is natural to concept as a plausible mechanism that size-dependent viscosity is considered. hope for a minimal preservation bias, may explain the episodic growth of Assuming weaker convective stress but when interpreting billion-years-old continental growth (e.g. Stein and from a hotter mantle is equivalent to continental crust, it would be hard to Hofmann, 1994), but as discussed holding these other variables constant overestimate preservation bias. earlier, this idea is based on early through time, which may not be numerical convection models with a warranted. In fact, as shown in the strong endothermic phase transition, next section, the present-day thermal Continental growth and the history the assumption of which does not budget indicates that the mantle was of ocean volume seem to be valid in light of recent probably drier in the past, which may experimental studies. Recently, O’Ne- compensate a decrease in viscosity due Continental growth: instantaneous, ill et al. (2007) proposed that plate to temperature dependency. gradual, or discontinuous? tectonics itself might have been inter- The diversity of continental growth When the continental crust emerged in mittent in the Precambrian. Whereas models is partly because different the Earth history and how it has their argument using palaeomagnetic geochemists place different weights evolved to its preset figure have been data is weak given the likelihood of on relevant geochemical data such as debated over several decades (e.g. true power wander (Evans, 2003), 143Nd ⁄ 144Nd and Nb ⁄Th. For exam- Hurley and Rand, 1969; O’Nions they offer a plausible dynamical rea- ple, a significant volume of continen- et al., 1979; DePaolo, 1980; Arm- soning. Mantle convection models tal crust in the Hadean, as implied by strong, 1981; Taylor and McLennan, with pseudo-plastic rheology are the hafnium isotope data of ancient 1985; Jacobsen, 1988; Collerson and known to exhibit three modes of zircons (Harrison et al., 2005), seems Kamber, 1999; Campbell, 2003; Har- convection (stagnant-lid, intermittent to be incompatible with the Th–U–Nb rison et al., 2005), and continental plate tectonics, and continuous plate systematics of depleted-mantle-de- growth is still a highly controversial tectonics), depending on the assumed rived rocks (Collerson and Kamber, topic. Being highly enriched in heat- strength of lithosphere (or its maxi- 1999). The interpretation of geochem- producing elements, its growth history mum yield strength) (e.g. Moresi and ical data in terms of geological pro- can influence the thermal evolution of Solomatov, 1998; Stein et al., 2004). cesses, however, often depends on Earth by depleting the convecting When the mantle was hotter in the simple box models (e.g. DePaolo, mantle [Eq. (9)], but this type of past, its internal viscosity may be 1980; Jacobsen, 1988), which in turn influence is of relatively minor impor- lower due to temperature dependency, assumes rapid mantle mixing (Coltice tance (Fig. 3). Probably a more so convective stress could become too et al., 2000). Also, geochemical inter- important aspect is whether the pro- low to sustain the continuous opera- pretations usually require the chemi- duction of continental crust has been tion of plate tectonics. In this case, cal or isotopic composition of the BSE continuous or discontinuous. Gradual intermittent plate tectonics is possible as a reference baseline, but the uncer- growth models (e.g. McLennan and for a certain range of maximum yield tainty of such reference value is not Taylor, 1982; Jacobsen, 1988; Camp- stress. Unlike the model proposed by trivial. Continental growth models bell, 2003) are obviously continuous, Silver and Behn (2008), however, this based on the Nd isotope evolution and instantaneous growth models (e.g. original version of intermittent plate (e.g. Bennett, 2003), for example, Armstrong, 1981; Harrison et al., tectonics proposed by Moresi and assumes that the Sm ⁄Nd ratio of the 2005) are also mostly continuous in Solomatov (1998) hardly modifies BSE is identical to that of chondrites this sense, because the constant conti- conventional heat-flow scaling, be- within 1% uncertainty [the Ôstrong’ nental mass is assumed to have been cause low heat flux during the stag- version of chondrite assumption (Ko- maintained by balancing continuous nant-lid mode is compensated by high renaga, 2008b)]. Available isotopic production and destruction. These heat flux during the plate tectonics data from terrestrial samples, on the Ôcontinuous’ growth models are com- mode [see, for example, Figure 3 of other hand, require only the Ôweak’ patible with the continuous operation Moresi and Solomatov (1998)]. So it version of chondrite assumption, i.e. of plate tectonics. ÔDiscontinuous’ may be able to explain continental the ratios of refractory lithophile ele- growth models (e.g. Rino et al., growth but not thermal evolution. ments such as Sm and Nd should not 2004; Hawkesworth and Kemp, Also, as noted earlier, the pseudo- be different from the chondritic aver- 2006; Parman, 2007), on the other plastic model of Moresi and Soloma- age more than a few per cents (Lyu- hand, suggest that the mode of mantle tov (1998) needs to assume the betskaya and Korenaga, 2007b). This convection may have changed at least strength of lithosphere (i.e. its type of uncertainty may be essential

2008 Blackwell Publishing Ltd 429 Plate tectonics, flood basalts and Earth’s oceans • J. Korenaga Terra Nova, Vol 20, No. 6, 419–439 ...... when constructing a geochemical (a) Constant ocean volume (b) Variable ocean volume model that can reconcile different [e.g., Reymer & Schubert, 1984] geochemical data simultaneously. Present Present Global water cycle and net water influx Continental Continental Frequent inundations throughout the crust crust Phanerozoic suggest that the mean sea level has always been close to the mean continental level, which is known as the constant freeboard (Wise, 1974). This constancy is at Archean Archean least geologically reasonable; conti- nental crust is subject to erosion when Faster plate motion Slower plate it is above the sea level, and when a motion part of continents subsides below the sea level, it would likely be the locus of Younger, shallower Deeper deposition. The abundant occurrence ocean basin ocean basin of submarine flood basalt magmatism in the Archean and Proterozoic (Arndt, 1999), however, implies that Fig. 6 Cartoon illustrating possible relations among heat-flow scaling, continental the mean sea level had been high growth, and ocean volume, under the assumption of the constant freeboard. (a) When enough to inundate a substantial frac- the ocean volume is assumed to be constant, plate motion must have been faster to create younger and shallower seafloor when there was less continental crust. Faster tion of continents through the Pre- plate motion (i.e. higher heat flux) in the past, however, results in thermal cambrian. The constant freeboard is catastrophe. (b) Slower plate motion, which is more consistent with the present-day thus probably too simplified an thermal budget, predicts a greater ocean volume in the past. assumption, but the following point seems to be robust: there has always Galer, 1991; Kasting and Holm, 1992; where dAo ⁄ dt(s) is the area–age dis- been a sufficient volume of water to fill Galer and Mezger, 1998; Harrison, tribution of seafloor, and q(s) is heat up the ocean basins, at least to the 1999; Hynes, 2001; Ru¨ pke et al., 2004; flow from a seafloor of age s . Both of mean continental level. In other Kasting et al., 2006). them can also be a function of a words, the constant freeboard is still The present-day thermal budget geological time t. The present-day useful to quantify the lower bound on characterized by a low Urey ratio area–age distribution is best approxi- the ocean volume. instead suggests less vigorous convec- mated by a triangular distribution (i.e. Reymer and Schubert (1984) cou- tion thus slower plate motion, the the area of younger seafloor is greater pled the constant freeboard with the corollary of which is older and deeper than that of older seafloor), and we thermal evolution of Earth and esti- ocean basins in the past (Fig. 6b). In assume this distribution for older mated the history of continental this case, the volume of Earth’s oceans times because it arises naturally from growth assuming that the ocean vol- may have been greater to maintain the subduction irrespective of seafloor age ume has been constant (Fig. 6a). Their constant freeboard, as many geochem- (Parsons, 1982). Thus we have model is based on the conventional ical studies suggest smaller continental dA s heat-flow scaling (Fig. 2), which pre- mass in the Archean. This interesting o ðs; tÞ¼GðtÞ 1 ; ð15Þ dicts faster plate motion (i.e. higher possibility has not been seriously con- ds smaxðtÞ heat flow) for hotter mantle in the sidered probably because it is difficult where s (t) is the age of the oldest past. Faster plate motion means youn- max to test. It can be demonstrated, how- seafloor. The coefficient G(t) is con- ger and thus shallower seafloor, and ever, that the constant ocean volume strained by the total area of ocean to maintain the constant freeboard is inconsistent with the present-day basins, Ao(t), as: without changing the ocean volume, thermal budget, without using any continental mass needs to be reduced. heat-flow scaling law. GðtÞ A ðtÞ¼A A ðtÞ¼ s ðtÞ; ð16Þ Note that the conventional scaling A key concept is that, for a given o c 2 max cannot reconstruct a reasonable ther- continental growth history, the where A is the total area of Earth’s mal history unless internal heat pro- assumptions of constant freeboard surface and A (t) is the total area of duction in the convecting mantle is and constant ocean volume are suffi- c continents. Seafloor heat flow is as- much higher than geochemical con- cient to determine oceanic heat flux, sumed to follow half-space cooling as: straints (see Internal heating and which represents a major fraction of surface heat flux). This important fact convective heat flux. In general, oce- Bffiffiffi TmðtÞ tends to be overlooked, and the anic heat flux at a time t, Q (t), may qðs; tÞ¼p ; ð17Þ o s Tmð0Þ notion of more vigorous convection be expressed as: in the past has been entrenched in the Z where B is 550 mW m)2 Myr)1 ⁄ 2 smax literature on the continental freeboard dAo QoðtÞ¼ ðs; tÞqðs; tÞds; ð14Þ (Korenaga and Korenaga, 2008). At or the history of ocean volume (e.g. 0 ds present (i.e. t = 0), G(0)=3.45

430 2008 Blackwell Publishing Ltd Terra Nova, Vol 20, No. 6, 419–439 J. Korenaga • Plate tectonics, flood basalts and Earth’s oceans ......

2 )1 km yr and smax(0)=180 Myr (Par- total convective heat flux (i.e. subcon- oceanic heat flux may be summarized sons, 1982), and Eq. (14) gives 34 tinental heat flux is neglected), and as Ôempirical’ heat-flow scaling TW for the total oceanic heat flux, solved the heat balance of Eq. (4) (Fig. 7d). Alternatively, we may fix which is in good agreement with the starting with the present-day Urey the maximum seafloor age as 180- actual estimate [32 ± 2 TW (Pollack ratio of 0.22. Here the zero-age sea- Myr-old, and solve Eq. (18) for the et al., 1993; Jaupart et al., 2007)]. floor depth d0 is assumed to be con- zero-age depth. Results for this case are Similarly, assuming the constant free- stant. This assumption may be shown in Fig. 8. The temperature of board, the total ocean volume may be justified because the existence of vol- the Archean mantle is still too high, and expressed as: canogenic massive sulphide deposits the predicted zero-age depth in the Z since the Archean (e.g. Nisbet et al., Archean is probably too shallow smax dA V ðtÞ¼ odðs; tÞds; ð18Þ 1987) indicates that the depth of mid- (Ohmoto, 1996; Kitajima et al., 2001). 0 ds ocean ridges should always be greater The constant ocean volume, there- where d(s) is the seafloor depth as a than 2 km below sea level (Ohmoto, fore, seems to be incompatible with function of seafloor age, 1996). As shown in Fig. 7a, this cal- the thermal budget of Earth, and the culation leads to thermal catastrophe possibility of time-dependent ocean pffiffiffi dðs; tÞ¼d0ðtÞþbðtÞ s; ð19Þ even for the instantaneous growth volume deserves some attention. As a model, in which no net continental preliminary attempt, the thermal his- Here d0 is the zero-age depth (average growth takes place during the last tory of Fig. 3 may be used to calculate depth to mid-ocean ridge axis), and b 4 Gyr. This is because hotter mantle by the history of ocean volume (Fig. 9). is the subsidence rate due to half-space itself results in greater thermal sub- First, noting that the coefficient G(t) cooling. At the present, we have sidence [Eq. (20)], so to keep the same in the area–age distribution can be d0(0) = 2654 m and b(0) = 323 m ocean volume, the maximum seafloor expressed as: ) Myr 1 ⁄ 2 (Korenaga and Korenaga, age should decrease (Fig. 7c), i.e. sea- GðtÞ¼LvðtÞ; ð22Þ 2008). The subsidence rate is linearly floor should become younger on aver- proportional to a temperature con- age, resulting in an increase in oceanic where L is the total length of divergent trast between the surface and the heat flux (Fig. 7b). This positive corre- plate boundaries and v(t) is the aver- interior (e.g. Turcotte and Schubert, lation between mantle temperature and age spreading rate, the maximum age 1982), so we have T ðtÞ bðtÞ¼bð0Þ m : ð20Þ Tmð0Þ (a)

By assuming the constant ocean vol- 1800 (b) 200 ume, therefore, we can solve Eq. (18) 1700 for the maximum seafloor age as: 150

 (TW) 2 15 V ð0Þ 1600 m smaxðtÞ¼ d0ðtÞ ; H 100 8b t A t Convective urey ratio ð Þ oð Þ & 1500 = 0.22 @ present Qconv

ð21Þ conv 50 1400 Q and Q(t) can then be calculated by Hm 1300 0 combining Eqs (15)–(17). The total (°C) Mantle potential temp. 0 1 2 3 4 0 1 2 3 4 area of continents in Eq. (16) is Time B.P. (Ga) Time B.P. (Ga) calculated from a given history of (d) continental growth, assuming that the (c) 200 average thickness of continental crust 200 150 has been approximately constant at 100 35–45 km (Durrheim and Mooney, (TW) 1991; Galer and Mezger, 1998). Note (Ma) 100 50 conv that Galer and Mezger (1998) sug- max τ gested that continental crust could 50 Q 20 have been thicker in the Archean than 10 at present by 5 km, based on origi- 0 0 1 2 3 4 1200 1400 1600 1800 nal burial pressures estimated for Time B.P. (Ga) Mantle potential temp. (°C) exposed Archean granite-greenstone segments. Thicker crust in the past Fig. 7 A possible consequence of the constant ocean volume, with the assumption of would only substantiate the following the constant zero-age depth. The modelling procedure is identical to that used for argument because it means more Fig. 3, expect that convective heat flux is based on Eq. (14). (a) Mantle potential ocean basins [Eq. (16)] and flatter temperature as a function of time. (b) Convective heat flux and internal heating. seafloor topography to satisfy the (c) Maximum age of seafloor. (d) A posteriori relationship between convective heat constant freeboard. flux and mantle temperature. For comparison, three heat-flow scaling laws of Fig. 2 We used the oceanic heat flux of are shown in grey. Legend is the same as Fig. 3; note that the cases of McLennan and Eq. (14) as the lower bound on the Taylor (1982) (blue) and Campbell (2003) (black) are identical in (a) and (d).

2008 Blackwell Publishing Ltd 431 Plate tectonics, flood basalts and Earth’s oceans • J. Korenaga Terra Nova, Vol 20, No. 6, 419–439 ......

(a) (b) to what is estimated to have been lost 1800 100 from Earth’s oceans (Fig. 9b). This coincidence may suggest that the man- 1700 80 tle was dry in the Hadean and has

(TW) been progressively hydrated by sub- 1600 m 60 duction. As mentioned earlier, such H Qconv Convective urey ratio 1500 & 40 gradual hydration may be important = 0.22 @ present for the continuous operation of plate 1400 conv 20 Q tectonics. Note that the dry Archean Hm mantle does not preclude the Ôwet’ 1300 0 Mantle potential temp. (°C) 0 1 2 3 4 0 1 2 3 4 origin of komatiite (Allegre, 1982; Time B.P. (Ga) Time B.P. (Ga) Parman et al., 1997; Grove and Par- man, 2004), because arc environments (c)3000 (d) can be locally hydrated by subduction. 200 The regassing-dominated global water 2000 100 cycle suggests that the most of terres- trial water may have originated at (TW)

(m) 50

o Earth’s surface, pointing towards the d 1000 conv impact origin of Earth’s oceans. As Q 20 discussed next, this issue has an 10 intriguing connection to the thermal 0 0 1 2 3 4 1200 1400 1600 1800 budget of Earth. Time B.P. (Ga) Mantle potential temp. (°C) The origin of terrestrial water Fig. 8 Same as Fig. 7, but with the assumption of the constant maximum age of seafloor. The history of zero-age depth is shown in (c). The origin of Earth’s oceans, or how this planet acquired its present amount of water in the early solar system, depends on a number of of seafloor may be calculated from the magmatism, so the net water influx factors that are currently not known predicted plate velocity as: may close to be zero. It is difficult, precisely enough, such as the relative however, to completely dehydrate timing of Earth accretion, core segre- 2AoðtÞ vð0Þ smaxðtÞ¼ : ð23Þ subducting slabs because nominally gation, the dissipation of the solar Gð0Þ vðtÞ anhydrous minerals can hold a nebula, and the disappearance of the Here, L is assumed to be constant non-trivial amount of water (e.g. Hir- massive asteroid belt (e.g. Abe et al., because the average size of plates is schmann et al., 2005). Continuous 2000; Marty and Yokochi, 2006). unlikely to change substantially with subduction at the current rate could The hydrogen isotope ratio (D ⁄H) ) time (Korenaga, 2006). The total easily bring one-ocean-worth water is 150 · 10 6 for Earth’s oceans, ) ocean volume can then be calculated into the deep mantle over a few billion 25 · 10 6 for the solar nebula, ) from Eq. (18), assuming the constant years (Smyth and Jacobsen, 2006). The 310 · 10 6 for comets and 130– ) freeboard as well as the constant zero- average rate of the predicted influx for 180 · 10 6 for chondrites, and with age depth (Fig. 9b). Results indicate the last 2 Gyr is also compatible with this information alone, Earth’s oceans that Archean ocean basins may have the estimated range of present-day could have been derived simply from been deep enough to hold twice as influx (Jarrard, 2003) (Fig. 9c). the accretion of chondritic materials, much water as today. The continental crust contains 1 or from the solar nebula with isotopic Obviously, the predicted history of wt% water on average (e.g. Wed- fractionation, or from the mixing of ocean volume should not be taken at epohl, 1995), equivalent to 0.18 multiple sources (e.g. Dauphas et al., face value, as it is based on a number ocean. The MORB-source mantle is 2000; Drake, 2005; Genda and Ikoma, of assumptions with varying credibil- estimated to have 142 ± 85 p.p.m. 2008). When combined with carbon ity (Table 1). At the same time, how- water (Saal et al., 2002), and the and nitrogen isotope constraints, how- ever, losing one-ocean-worth water global mass balance of silicate reser- ever, the late accretion (i.e. accretion since the Archean is not surprising, voirs, on the basis of the new compo- after the core formation) of chondritic or may even be expected in light of the sition model of Earth’s primitive materials appear to be the most likely global water cycle. It has long been mantle (Lyubetskaya and Korenaga, source of terrestrial water (Marty and known that water flux into the mantle 2007a), indicates that the MORB- Yokochi, 2006), and indeed, such late from the hydrosphere by subduction source mantle is representative for accretion with 0.5–1% Earth’s mass is exceeds water flux out of the mantle the bulk of the mantle (Lyubetskaya suggested by the abundance of highly by mid-ocean magmatism by an order and Korenaga, 2007b), so the mantle siderophile elements (HSE) in Earth’s of magnitude (e.g. Ito et al., 1983; may hold 0.42 ± 0.25 ocean of mantle (e.g. Morgan, 1986) and is also Jarrard, 2003). Most of such water water. Thus, the amount of water physically plausible according to influx is usually believed to quickly currently stored in the BSE (crust the dynamical simulations of plane- return to the surface through arc and mantle), 0.5–1 ocean, is similar tary accretion (e.g. Morbidelli et al.,

432 2008 Blackwell Publishing Ltd Terra Nova, Vol 20, No. 6, 419–439 J. Korenaga • Plate tectonics, flood basalts and Earth’s oceans ......

(a) the isotopic ratio for the BSE may be 600 estimated based on the elemental com- position model of BSE. With the 500 conventional estimate for the BSE Al2O3 content (4.2 wt%), the ob- 400 served correlation suggests the BSE

(Ma) isotope ratio of 0.1289–0.1304 (95% confidence limit) (Meisel et al., 2001). max 300 τ On the other hand, the so far observed range of the 187Os ⁄ 188Os ratio is 200 0.1255–0.1270 for carbonaceous chon- drites, 0.1270–0.1305 for ordinary 100 chondrites and 0.1270–0.1290 for 0 1 2 3 4 enstatite chondrites (Meisel et al., Time B.P. (Ga) 1996). Thus, Earth’s 187Os ⁄ 188Os is consistent with ordinary or enstatite (b) 3.0 chondrites, but not with carbonaceous chondrites. Ordinary and enstatite 2.5 chondrites are, however, much drier (<1% H2O) than carbonaceous chon- drites (10% H2O) (Robert, 2003), so (0) 2.0 o the addition of 0.5–1% Earth’s mass V could account for only a small fraction (t)/

o 1.5 of Earth’s water budget. The osmium V constraint has motivated a variety of 1.0 more elaborate ways to deliver water to Earth and satisfy geochemical con- 0.5 straints at the same time (e.g. Dauphas 0 1 2 3 4 and Marty, 2002; Drake and Righter, Time B.P. (Ga) 2002; Marty and Yokochi, 2006). This argument may not be so robust (c) because, if the late accretion of HSE 30 were made by the Moon-forming impactor (or similarly large impac- tors) and partial core addition, there 20 g/year) would be no simple relationship be-

14 tween the added masses of HSE and water. Even if we limit ourselves to 10 Present-day simple end-member mixing, however, influx the osmium constraint has one weak- ness that has been overlooked. The 0 composition model of BSE is based

Water influx (10 primarily on noisy geochemical trends exhibited by mantle rocks, but the 0 1 2 3 4 model uncertainty has not been well Time B.P. (Ga) quantified. A new statistical method was recently built to address this issue, Fig. 9 (a) The maximum age of seafloor predicted from the thermal history of Fig. 3 resulting in not only quantifying the [Eq. (23)]. (b) The predicted history of ocean volume normalized by the present-day uncertainty but also revising the mod- volume. (c) Net water influx from the hydrosphere to the mantle, assuming that the el itself in a non-trivial manner (Lyu- predicted change of ocean volume shown in (b) is due to the hydration of the mantle by subduction. Legend is the same as in Fig. 3. betskaya and Korenaga, 2007a). The new BSE model suggests the Al2O3 content of 3.52 ± 0.60 wt%. Revisit- ing the correlation, we would obtain 2000). Also, the Moon-forming giant known to present a serious impedi- the new 95% confidence limit of impact could have been this late ment to this late accretion hypothesis 0.1267–0.1277 for BSE 187Os ⁄ 188Os, veneer if a fraction of the impactor’s for Earth’s oceans (Meisel et al., 1996, which turns out to be consistent with core were mechanically mixed with the 2001). Mantle samples such as mantle any kind of chondrites (Fig. 10). Note Earth’s mantle. xenoliths and massif peridotites exhi- that this confidence limit is derived Osmium is one of those HSEs, bit a linear correlation between this from the uncertainty of the linear however, and its isotopic ratio, isotopic ratio and the index of melt trend and does not reflect the uncer- 187 188 Os ⁄ Os, of Earth’s mantle is depletion such as Al2O3 (Fig. 10), and tainty of the BSE model. With the

2008 Blackwell Publishing Ltd 433 Plate tectonics, flood basalts and Earth’s oceans • J. Korenaga Terra Nova, Vol 20, No. 6, 419–439 ......

0.135 Allegre, C.J., 1982. Genesis of Archaean komatiites in a wet ultramafic subducted

Ordinary plate. In: Komatiites (N.T. Arndt and chondrites E.G. Nisbet, eds), pp. 495–500. George 0.130 Allen & Unwin, London. Allegre, C.J., Hofmann, A. and O’Nions, K., 1996. The argon constraints on

Enstatite mantle structure. Geophys. Res. Lett., 23, 0.125 chondrites 3555–3557.

Os Carbonaceous chondrites Anderson, D.L., 1998. The EDGES of the 188 Mantle, in the Core-Mantle Boundary Region, pp. 255–271. American Geo- Os/ 0.120 physical Union, Washington, DC. 187 Anderson, D.L., 2005. Large igneous 3.52 ± 0.60 wt% New BSE estimate 4.2 wt% Old BSE estimate provinces, delamination, and fertile mantle. Elements, 1, 271–275. 0.115 Araki, T., Enomoto, S., Furuno, K. et al.,

Europe Southwest US 2005. Experimental investigation of Mexico Canada Pacific Australia geologically produced antineutrinos with Asia massif KamLAND. Nature, 436, 499–503. 0.110 Armstrong, R.L., 1981. Radiogenic iso- 0 1 2 3 4 5 topes: the case for crustal recycling on a Al2O3 (wt%) near-steady-state no-continental-growth Earth. Phil. Trans. R. Soc. Lond. A, 301, 187 188 443–472. Fig. 10 Covariation of Al2O3 and Os ⁄ Os for mantle xenoliths and massif peridotites [data are from Meisel et al. (2001) and references therein]. Different Arndt, N., 1999. Why was flood volcanism symbols denote different localities as indicated by the legend. Linear regression by on submerged continental platforms so Meisel et al. (2001) is shown by solid dash line (y = 0.1160 + 0.003253x). Linear common in the Precambrian? Precambrian Res., 97, 155–164. regression by the bootstrap resampling method is shown by red dashed line Badro, J., Rueff, J.-P., Vanko, G., Mon- (y=0.1161 + 0.003170x) with pink shade for the 95% confidence limit. The small aco, G., Figuet, G. and Guyot, F., 2004. difference from the original regression probably reflects that unpublished Mexico Electronic transition in perovskite: samples used by Meisel et al. (2001) are not used here. The 95% confidence limit on possible non-convecting layers in the the osmium isotope rate is shown for the case with the old BSE Al2O3 content (4.2 lower mantle. Science, 305, 383–386. wt%) and the original regression (grey arrow) and for that with the new BSE Al2O3 Bennett, V.C., 2003. Compositional evolu- content (3.52 wt%) and the bootstrap regression (blue arrow). The influence of the tion of the mantle. In: Treatise on uncertainty of the new BSE model (±0.60 wt%) is indicated by light blue arrow. The Geochemistry (H.D. Holland and osmium isotope ranges observed for three major types of chondrites are also indicated K.K. Turekian, eds), Vol. 2, pp. 493– (blue, carbonaceous; red, ordinary; green, enstatite). 519. Elsevier, Amsterdam. Bercovici, D., 2003. The generation of plate tectonics from mantle convection. Earth uncertainty of ±0.60 wt%, the likely Acknowledgements Planet. Sci. Lett., 205, 107–121. range of the BSE 187Os ⁄ 188Os would Bercovici, D., Ricard, Y. and Richards, The author thanks Alfred Kro¨ ner for M.A., 2000. The relation between mantle be widened as 0.1253–0.1291. The invitation to write a review paper on the dynamics and plate tectonics: a primer. other potential water sources can still thermal evolution of Earth, and Kent In: The History and Dynamics of Global contribute to some degree, but the Condie, Norm Sleep, Geoff Davies, and Plate Motions (M.A. Richards, R.G. most dominant source may well be an anonymous reviewer for constructive Gordon, R.D. van der Hilst, ed.), pp. 5– reviews. This work was sponsored by the carbonaceous chondrites. 46. American Geophysical Union, U.S. National Science Foundation under It is noted that efforts to quantify Washington, DC. grant EAR-0449517. the reliability of the BSE model were Bickle, M.J., 1986. Implications of melting motivated solely by a need to build a for stabilisation of the lithosphere and self-consistent geochemical model, heat loss in the Archean. Earth Planet. References which is important for long-standing Sci. Lett., 80, 314–324. debates over the structure and evolu- Abbott, D. and Mooney, W., 1995. The Bleeker,W.,2003.ThelateArcheanrecord:a tion of Earth’s mantle (Lyubetskaya structure and geochemical evolution of puzzle in ca. 35 pieces. Lithos, 71, 99–134. and Korenaga, 2007b; Korenaga, the continental crust: Support for the Boehler, R., 1996. Melting temperature of oceanic plateau model of continental the Earth’s mantle and core: Earth’s 2008b). The relevance to the late growth. Rev. Geophys. Suppl., 33, 231– thermal structure. Ann. Rev. Earth accretion hypothesis is entirely seren- 242. Planet. Sci., 24, 15–40. dipitous. Though the new BSE model Abe, Y., Ohtani, E., Okuchi, T., Righter, Buffett, B.A., 2002. Estimates of heat flow is not a final answer and should be K. and Drake, M., 2000. Water in the in the deep mantle based on the power revised in future when additional data early earth. In: The Origin of Earth and requirements for the geodynamo. become available, it does seem to Moon (K. Righter and R.M. Canup, Geophys. Res. Lett., 29, Doi: 10.1029/ bring us towards simpler, less arbi- eds), pp. 413–433. University of Arizona, 2001GL014649. trary hypotheses for the structure of Tucson. Buffett, B.A., 2003. On the thermal state mantle convection as well as the origin Albare` de, F., 1998. The growth of conti- of Earth’s core. Science, 299, 1675– of Earth’s oceans. nental crust. Tectonophysics, 296, 1–14. 1676.

434 2008 Blackwell Publishing Ltd Terra Nova, Vol 20, No. 6, 419–439 J. Korenaga • Plate tectonics, flood basalts and Earth’s oceans ......

Butler, S.L., Peltier, W.R. and Costin, Davies, G.F., 1992. On the emergence of level in the Archean and cooling of the S.O., 2005. Numerical models of the plate tectonics. Geology, 20, 963–966. Earth. Precambrian Res., 92, 389–412. Earth’s thermal history: effects of inner- Davies, G.F., 1993. Cooling the core and Genda, H. and Ikoma, M., 2008. Origin of core solidification and core potassim. mantle by plume and plate flows. the ocean on the Earth: early evolution Phys. Earth Planet. Int., 152, 22–42. Geophys. J. Int., 115, 132–146. of water D ⁄ H in a hydrogen-rich Campbell, I.H., 2003. Constraints on Davies, G.F., 2006. Gravitational depltion atmosphere. Icarus, 194, 42–52. continental growth models from Nb ⁄ U of the early Earth’s upper mantle and the Gessmann, C.K. and Wood, B.J., 2002. ratios in the 3.5 Ga Barberton and viability of early plate tectonics. Earth Potassium in the Earth’s core? Earth other Archaean basalt-komatiite suites. Planet. Sci. Lett., 243, 376–382. Planet. Sci. Lett., 200, 63–78. Am. J. Sci., 303, 319–351. Davies, G.F., 2007. Mantle regulation of Grigne, C. and Labrosse, S., 2001. Effects Campbell, I.H. and Griffiths, R.W., 1990. core cooling: a geodynamo without core of continents on Earth cooling: thermal Implications of mantle plume structure radioactivity? Phys. Earth Planet. Int., blanketing and depletion in radioactive for the evolution of flood basalts. Earth 160, 215–229. elements. Geophys. Res. Lett., 28, 2707– Planet. Sci. Lett., 99, 79–93. DePaolo, D.J., 1980. Crustal growth and 2710. Campbell, I.H., Griffiths, R.W. and Hill, mantle evolution: inferences from mod- Grigne, C., Labrosse, S. and Tackley, P.J., R.I., 1989. Melting in an Archean mantle els of element transport and Nd and 2005. Convective heat transfer as a plume: heads it’s basalts, tails it’s Sr isotopes. Geochim. Cosmochim. Acta, function of wavelength: Implications for komatiits. Nature, 339, 697–699. 44, 1185–1196. the cooling of the Earth. J. Geophys. Christensen, U.R., 1985. Thermal evolu- Doin, M.-P., Fleitout, L. and Christensen, Res., 110, B03409. Doi: 10.1029/ tion models for the Earth. J. Geophys. U., 1997. Mantle convection and stabil- 2004JB003376. Res., 90, 2995–3007. ity of depleted and undepleted conti- Grimes, C.B., John, B.E., Kelemen, P.B., Coffin, M.F. and Eldholm, O., 1994. Large nental lithosphere. J. Geophys. Res., Mazdab, F.K., Wooden, J.L., Cheadle, igneous provinces: crustal structure, 102, 2772–2787. M.J., Kanghøj, K. and Schwartz, J.J., dimensions, and external consequences. Drake, M.J., 2005. Origin of water in the 2007. Trace element chemistry of zircons Rev. Geophys., 32, 1–36. terrestrial planets. Meteorit. Planet. Sci., from oceanic crust: a method for distin- Collerson, K.D. and Kamber, B.S., 1999. 40, 519–527. guishing detrital zircon provenance. Evolution of the continents and the Drake, M.J. and Righter, K., 2002. Geology, 35, 643–646. atmosphere inferred from Th-U-Nb Determining the composition of the Grove, T.L. and Parman, S.W., 2004. systematics of the depleted mantle. Earth. Nature, 416, 39–44. Thermal evolution of the Earth as Science, 283, 1519–1522. Durrheim, R.J. and Mooney, W.D., 1991. recorded by komatiites. Earth Planet. Coltice, N., Albarede, F. and Gillet, P., Archean and Proterozoic crustal evolu- Sci. Lett., 219, 173–187. 2000. 40K-40Ar constraints on recycling tion: evidence from crustal seismology. Gurnis, M., 1989. A reassessment of the continental crust into the mantle. Geology, 19, 606–609. heat transport by variable viscosity Science, 288, 845–847. Ernst, R.E. and Buchan, K.L., 2003. convection with plates and lids. Geophys. Condie, K.C., 1998. Episodic continental Recognizing mantle plumes in the geo- Res. Lett., 16, 179–182. growth and supercontinents: a mantle logical record. Annu. Rev. Earth Planet. Gurnis, M., Hall, C. and Lavier, L., 2004. avalanche connection? Earth Planet. Sci., 31, 469–523. Evolving force balance during incipient Sci. Lett., 163, 97–108. Evans, D.A.D., 2003. True polar wander subduction. Geochem. Geophys. Geosys., Conrad, C.P. and Hager, B.H., 1999a. and supercontinents. Tectonophysics, 5, Q07. Doi: 10.1029/2003GC000681. Effects of plate bending and fault strength 362, 303–320. Halliday, A.N., 2003. The origin and at subduction zones on plate dynamics. Farnetani, C.G., 1997. Exess temperature earliest history of the earth. In: Treatise J. Geophys. Res., 104, 17,551–17,571. of mantle plumes: the role of chemical on Geochemistry (H.D. Holland and Conrad, C.P. and Hager, B.H., 1999b. The stratification across D’’. Geophys. Res. K.K. Turekian, eds), Vol. 1, pp. 509– thermal evolution of an earth with Lett., 24, 1583–1586. 557, Elsevier, Amsterdam. strong subduction zones. Geophys. Res. Fei, Y., Van Orman, J., Li, J., van Harrison, C.G.A., 1999. Constraints on Lett., 26, 3041–3044. Westrennen, W., Sanloup, C., Minarik, ocean volume change since the Archean. Coogan, L.A. and Hinton, R.W., 2006. Do W., Hirose, K., Komabayashi, T., Geophys. Res. Lett., 26, 1913–1916. the trace element compositions of detri- Walter, M. and Funakoshi, K., 2004. Harrison, T.M., Blichert-Toft, J., M¨ uller, tal zircons require Hadean continental Experimentally determined postspinel W., Albarede, F., Holden, P. and crust? Geology, 34, 633–636. transformation boundary in Mg2SiO4 Mojzsis, S.J., 2005. Heterogeneous Daly, S.F., 1980. Convection with decaying using MgO as an internal pressure Hadean hafnium: evidence of continen- heat sources: constant viscosity. Geo- standard and its geophysical implica- tal crust at 4.4 to 4.5 Ga. Science, phys. J. R. Astron. Soc., 61, 519–547. tions. J. Geophys. Res., 109, B02305. 310, 1947–1950. Dauphas, N. and Marty, B., 2002. Infer- Doi: 10.1029/2003JB002652. Hartmann, W.K., Ryder, G., Dones, L. ence on the nature and the mass of Foulger, G.R., Natland, J.H., Presnell, and Grinspoon, D., 2000. The time- Earth’s late veneer from noble metals D.C. and Anderson, D.L., 2005. Plates, dependent intense bombardment of and gases. J. Geophys. Res., 107, 5129. Plumes, and Paradigms. Geological the primordial Earth ⁄ Moon system. Doi: 10.1029/2001JE001617. Society of America, Boulder, CO. In: Orign of the Earth and Moon Dauphas, N., Robert, F. and Marty, B., Fyfe, W.S., 1978. The evolution of the (R.M. Canup and K. Righter, eds), 2000. The late asteroidal and cometary earth’s crust: Modern plate tectonics to pp. 493–512. University of Arizona bombardment of Earth as recorded in ancient hot spot tectonics? Chem. Geol., Press, Tucson, AZ. water deuterium to protium ratio. Icarus, 23, 89–114. Hawkesworth, C.J. and Kemp, A.I.S., 148, 508–512. Galer, S.J.G., 1991. Interrelationships 2006. Evolution of the continental crust. Davies, G.F., 1980. Thermal histories between continental freeboard, tectonics Nature, 443, 811–817. of convective Earth models and con- and mantle temperature. Earth Planet. Herzberg, C., 2004. Partial melting below straints on radiogenic heat production Sci. Lett., 105, 214–228. the Ontong Java Plateau. In: Origin and in the earth. J. Geophys. Res., 85, Galer, S.J.G. and Mezger, K., 1998. Evolution of the Ontong Java Plateau 2517–2530. Metamorphism, denudation and sea (G. Fitton, J. Mahoney, P. Wallace and

2008 Blackwell Publishing Ltd 435 Plate tectonics, flood basalts and Earth’s oceans • J. Korenaga Terra Nova, Vol 20, No. 6, 419–439 ......

A. Saunders, eds), pp. 179–184. Geo- Jacobsen, S.B. and Wasserburg, G.J., 1979. Kitajima, K., Maruyama, S., Utsunomiya, logical Society of London, London. The mean age of mantle and crustal S. and Liou, J.G., 2001. Seafloor Herzberg, C., Asimow, P.D., Arndt, N., reservoirs. J. Geophys. Res., 84, 7411– hydrothermal alteration at an Archaean Niu, Y., C.Lesher, M., Fitton, J.G., 7427. mid-ocean ridge. J. Metamor. Geol., 19, Cheadle, M.J. and Saunders, A.D., 2007. Jarrard, R.D., 2003. Subduction fluxes of 583–599. Temperatures in ambient mantle and water, carbon dioxide, chlorine, and Komiya, T., Maruyama, S., Masuda, T., plumes: constraints from basalts, potassium. Geochem. Geophys. Geosys., Nohda, S., Hayashi, M. and Okamotmo, picrites, and komatiites. Geochem. 4, 8905. Doi: 10.1029/2002GC000392.. K., 1999. Plate tectonics at 3.8–3.7 Ga: Geophys. Geosys., 8, Q02. Doi: 10.1029/ Jaupart, C., Labrosse, S. and Mareschal, field evidence from the Isua accretionary 2006GC001390. J.-C., 2007. Temperatures, heat and complex, southern west Greenland. Hirschmann, M.M., Aubaud, C. and energy in the mantle of the Earth. In: J. Geol., 107, 515–554. Withers, A.C., 2005. Storage capacity of Treatise on Geophysics (G. Schubert, Korenaga, J., 2003. Energetics of mantle H2O in nominally anhydrous minerals in ed.), Vol. 7, pp. 253–303. Elsevier, convection and the fate of fossil heat. the upper mantle. Earth Planet. Sci. Amsterdam. Geophys. Res. Lett., 30, 1437. Doi: Lett., 236, 167–181. Jellinek, A.M. and Manga, M., 2002. The 10.1029/2003GL016982. Hirth, G. and Kohlstedt, D.L., 1996. influence of a chemical boundary layer Korenaga, J., 2004. Mantle mixing and Water in the oceanic mantle: Implica- on the fixity, spacing and lifetime of continental breakup magmatism. Earth tions for rheology, melt extraction, and mantle plumes. Nature, 418, 760–763. Planet. Sci. Lett., 218, 463–473. the evolution of the lithosphere. Earth Jochum, K.P., Hofmann, A.W., Ito, E., Korenaga, J., 2005a. Firm mantle plumes Planet. Sci. Lett., 144, 93–108. Seufert, H.M. and White, W.M., 1983. and the nature of the core-mantle Hirth, G. and Kohlstedt, D., 2003. Rheol- K, U and Th in mid-ocean ridge basalt boundary region. Earth Planet. Sci. ogy of the upper mantle and the mantle glasses and heat production, K ⁄ U and Lett., 232, 29–37. wedge: a view from the experimentalists. K ⁄ Rb in the mantle. Nature, 306, Korenaga, J., 2005b. Why did not the In: Inside the Subduction Factory 431–436. Ontong Java Plateau form subaerially? (J. Eiler, ed.), pp. 83–105. American Jordan, T.H., 1988. Structure and forma- Earth Planet. Sci. Lett., 234, 385–399. Geophysical Union, Washington, DC. tion of the continental tectosphere. Korenaga, J., 2006. Archean geodynamics Hoffman, P.F., 1997. Tectonic genealogy J. Petrol. Spec., 11–37. and the thermal evolution of Earth. In: of North America. In: Earth Structure: Karato, S., 1986. Does partial melting Archean Geodynamics and Environments An Introduction to Structural Geology reduce the creep strength of the upper (K. Benn, J.-C. Mareschal and and Tectonics (B.A. van der Pluijm and mantle? Nature, 319, 309–310. K. Condie, eds), pp. 7–32. American S. Marshak, eds), pp. 459–464. Karato, S., 1998. Seismic anisotropy in the Geophysical Union, Washington, DC. McGraw-Hill, New York. deep mantle, boundary layers and the Korenaga, J., 2007a. Eustasy, superconti- Honda, S., Yuen, D.A., Balachandar, S. geometry of mantle convection. Pure nental insulation, and the temporal and Reuteler, D., 1993. Three-dimen- Appl. Geophys., 151, 565–587. variability of terrestrial heat flux. Earth sional instabilities of mantle convection Karato, S. and Wu, P., 1993. Rheology of Planet. Sci. Lett., 257, 350–358. with multiple phase transitions. Science, the upper mantle: a synthesis. Science, Korenaga, J., 2007b. Thermal cracking and 259, 1308–1311. 260, 771–778. the deep hydration of oceanic litho- Howard, L.N., 1966. Convection at high Kasting, J.F. and Holm, N.G., 1992. What sphere: a key to the generation of plate Rayleigh number. In: Proceedings of the determines the volume of the oceans? tectonics? J. Geophys. Res., 112, B05408. Eleventh International Congress of Earth Planet. Sci. Lett., 109, 507–515. Doi: 10.1029/2006JB004502. Applied Mechanics (H. Gortler, ed.), pp. Kasting, J.F., Tazewell Howard, M., Korenaga, J., 2008a. Comment on ‘‘Inter- 1109–1115. Springer-Verlag, New York. Wallmann, K., Veizer, J., Shields, G. and mittent plate tectonics?’’. Science, 320, Hurley, P.M. and Rand, J.R., 1969. Jaffre´ s, J., 2006. Paleoclimates, ocean 1291a. Pre-drift continental nuclei. Science, depth, and the oxygen isotopic compo- Korenaga, J., 2008b. Urey ratio and the 164, 1229–1242. sition of seawater. Earth Planet. structure and evolution of Earth’s man- Hynes, A., 2001. Freeboard revisited: con- Sci. Lett., 252, 82–93. tle. Rev. Geophys., 46, RG2007. tinental growth, crustal thickness change Katsura, T., Yamada, H., Shinmei, T., Doi: 10.1029/2007RG000241. and Earth’s thermal efficiency. Earth Kubo, A., Ono, S., Kanzaki, M., Yone- Korenaga, J. and Jordan, T.H., 2002. On Planet. Sci. Lett., 185, 161–172. da, A., Walter, M.J., Ito, E., Urakawa, the state of sublithospheric upper mantle Ito, G. and Mahoney, J.J., 2005. Flow and S., Funakoshi, K. and Utsumi, W., 2003. beneath a supercontient. Geophys. melting of a heterogeneous mantle: 2. Post-spinel transition in Mg2SiO4 deter- J. Int., 149, 179–189. implications for a chemically nonlayered mined by high P)T in situ X-ray Korenaga, J. and Karato, S., 2008. A new mantle. Earth Planet. Sci. Lett., 230, deffractometry. Phys. Earth Planet. analysis of experimental data on olivine 47–63. Int., 136, 11–24. rheology. J. Geophys. Res., 113, B02403. Ito, E., Harris, D.M. and Anderson, A.T., Kellogg, L.H., Hager, B.H. and van der Doi: 10.1029/2007JB005100. 1983. Alteration of oceanic crust and Hilst, R.D., 1999. Compositional strati- Korenaga, J. and Kelemen, P.B., 1998. geologic cycling of chlorine and water. fication in the deep mantle. Science, 283, Melt migration through the oceanic Geochim. Cosmochim. Acta, 47, 1613– 1881–1884. lower crust: a constraint from melt 1624. Kinny, P.D. and Nutman, A.P., 1996. percolation modeling with finite solid Jacob, D., Jagoutz, E., Lowry, D., Mattey, Zirconology of the Meeberrie gneiss, diffusion. Earth Planet. Sci. Lett., D. and Kudrjavtseva, G., 1994. Dia- Yilgarn Craton, Western Australia: an 156, 1–11. mondiferous eclogites from Siberia: early Archaean migmatite. Precambrian Korenaga, J. and Kelemen, P.B., 2000. remnants of Archean oceanic crust. Res., 78, 165–178. Major element heterogeneity of the Geochim. Cosmochim. Acta, 58, 5191– Kinzler, R.J., 1997. Melting of mantle mantle source in the North Atlantic 5207. peridotite at pressures approaching the igneous province. Earth Planet. Sci. Jacobsen, S.B., 1988. Isotopic constraints spinel to garnet transition: application to Lett., 184, 251–268. on crustal growth and recyling. Earth mid-ocean ridge basalt petrogenesis. Korenaga, T. and Korenaga, J., 2008. Planet. Sci. Lett., 90, 315–329. J. Geophys. Res., 102, 852–874. Subsidence of normal oceanic

436 2008 Blackwell Publishing Ltd Terra Nova, Vol 20, No. 6, 419–439 J. Korenaga • Plate tectonics, flood basalts and Earth’s oceans ......

lithosphere, apparent thermal expansiv- Machetel, P. and Weber, P., 1991. Inter- Murakami, M., Hirose, K., Kawamura, ity, and seafloor flattening. Earth Planet. mittent layered convection in a model K., Sata, N. and Ohishi, Y., 2004. Post- Sci. Lett., 268, 44–51. with an endothermic phase change at perovskite phase transition in MgSiO3. Korenaga, J., Kelemen, P.B. and Hol- 670 km. Nature, 350, 55–57. Science, 304, 855–858. brook, W.S., 2002. Methods for resolv- Marty, B. and Yokochi, R., 2006. Water in Nimmo, F., Price, G.D., Brodholt, J. and ing the origin of large igneous provinces the early Earth. Rev. Mineral. Geochem., Gubbins, D., 2004. The influence of from crustal seismology. J. Geophys. 62, 421–450. potassium on core and geodynamo Res., 107, 2178. Doi: 10.1029/ McBirney, A.R., 1993. Igneous Petrology, evolution. Geophys. J. Int., 156, 363–376. 2001B001030. 2nd edn. Jones and Bartlett Publishers, Nisbet, E.G., Arndt, N.T., Bickle, M.J., Labrosse, S., 2002. Hotspots, mantle Boston, MA. Cameron, W.E., Chauvel, C., Cheadle, plumes and core heat loss. Earth Planet. McDonough, W.F., 2003. Compositional M., Hegner, E., Kyser, T.K., Martin, A., Sci. Lett., 199, 147–156. model for the Earth’s core. In: Treatise Renner, R. and Roedder, E., 1987. Un- Labrosse, S. and Jaupart, C., 2007. Ther- on Geochemistry (H.D. Holland and iquely fresh 2.7 Ga komatiites from the mal evolution of the Earth: secular K.K. Turekian, eds), Vol. 2, pp. 547– Belingwe greenstone belt, Zimbabwe. changes and fluctuations of plate char- 568. Elsevier, Amsterdam. Geology, 15, 1147–1150. acteristics. Earth Planet. Sci. Lett., 260, McDonough, W.F., 2005. Ghosts from Nutman, A.P., 2006. Antiquity of the 465–481. within. Nature, 436, 467–468. oceans and continents. Elements, 2, Labrosse, S., Poirier, J.-P. and Mouel, McDonough, W.F. and Sun, S.-s., 1995. 223–227. J.-L.L., 2001. The age of the inner core. The composition of the Earth. Chem. Nutman, A.P., Friend, C.R.L. and Earth Planet. Sci. Lett., 190, 111–123. Geol., 120, 223–253. Bennett, V.C., 2002. Evidence for 3650– Langmuir, C.H., Klein, E.M. and Plank, McElhinny, M.W. and Senanayake, W.E., 3600 Ma assembly of the northern end T., 1992. Petrological systematics of 1980. Paleomagnetic evidence for the of the Itsaq Gneiss Complex, Greenland: mid-ocean ridge basalts: constraints on existence of the geomagnetic field implication for early Archean tectonics. melt generation beneath ocean ridges. 3.5 Ga ago. J. Geophys. Res., 85, Tectonics, 21, 1005. Doi: 10.1029/ In: Mantle Flow and Melt Generation at 3523–3528. 2000TC001203. Mid-Ocean Ridges, Geophys. Monogr. McKenzie, D. and Weiss, N., 1975. Spec- O’Hara, M.J., 1975. Is there an icelandic Ser. (J. Phipps Morgan, D.K. ulations on the thermal and tectonic mantle plume? Nature, 253, 708–710. Blackman and J.M. Sinton, eds), Vol. history of the Earth. Geophys. J. R. O’Neill, C., Lenardic, A., Moresi, L., 71, pp. 183–280. AGU, Washington, Astron. Soc., 42, 131–174. Torsvik, T.H. and Lee, C.-T., 2007. DC. McLennan, S.M. and Taylor, R.S., 1982. Episodic Precambrian subduction. Earth Lassiter, J.C., 2006. Constraints on the Geochemical constraints on the growth Planet. Sci. Lett., 262, 552–562. coupled thermal evolution of the of the continental crust. J. Geol., 90, O’Nions, R.K., Evensen, N.M. and Ham- Earth’s core and mantle, the age of the 347–361. ilton, P.J., 1979. Geochemical modeling inner core, and the origin of the McNamara, A.K. and van Keken, P.E., of mantle differentiation and crustal 186Os ⁄ 188Os ‘‘core signal’’ in plume- 2000. Cooling of the Earth: a para- growth. J. Geophys. Res., 84, 6091–6101. derived lavas. Earth Planet. Sci. Lett., meterized convection study of whole Oganov, A.R. and Ono, S., 2004. Theo- 250, 306–317. versus layered models. Geochem. Geo- retical and experimental evidence for a Lee, K.K.M., Steinle-Neumann, G. and phys. Geosys., 1, 2000GC000045. post-perovskite phase of MgSiO3 in Jeanloz, R., 2004. Ab-initio high-pres- Meisel, T., Walker, R.J. and Morgan, J.W., Earth’s D’’ layer. Nature, 430, 445–448. sure alloying of iron and potassium: 1996. The osmium isotopic composition Ogawa, M., 2003. Plate-like regime of a implications for the Earth’s core. of the Earth’s primitive upper mantle. numerically modeled thermal convection Geophys. Res. Lett., 31, L11603. Doi: Nature, 383, 517–520. in a fluid with temperature-, pressure-, 10.1029/2004GL019839. Meisel, T., Walker, R.J., Irving, A.J. and and stress-history-dependent viscosity. Lenardic, A. and Moresi, L.-N., 1999. Lorand, J.-P., 2001. Osmium isotopic J. Geophys. Res., 108, 2067. Some thoughts on the stability of cra- compositions of mantle xenoliths: a Doi: 10.1029/2000JB000069. tonic lithosphere: effects of buoyancy global perspective. Geochim. Cosmochim. Ohmoto, H., 1996. Formation of volcano- and viscosity. J. Geophys. Res., 104, Acta, 65, 1311–1323. genic massive sulfide deposits: the 12,747–12,758. Mojzsis, S.J., Harrison, T.M. and Pidgeon, Kuroko perspective. Ore Geol. Rev., Lin, S.-C. and van Keken, P.E., 2005. R.T., 2001. Oxygen-isotope evidence 10, 135–177. Multiple volcanic episodes of flood from ancient zircons for liquid water Parman, S.W., 2007. Helium isotopic evi- basalts caused by thermochemical man- at the Earth’s surface 4,300 Myr ago. dence for episodic mantle melting and tle plumes. Nature, 436, 250–252. Nature, 409, 178–181. crustal growth. Nature, 446, 900–903. Lin, J., Struzhkin, V., Jacobsen, S., Hu, Morbidelli, A., Chambers, J., Lunine, J.I., Parman, S.W., Dann, J.C., Grove, T.L. M., Chow, P., Kung, J., Liu, H., Mao, Petit, J.M., Robert, F., Valsecchi, G.B. and de Wit, M.J., 1997. Emplacement H. and Hemley, R., 2005. Spin transition and Cyr, K.E., 2000. Source regions and conditions of komatiite magmas from of iron in magnesiowu¨ ste in the Earth’s timescales for the delivery of water to the the 3.49 Ga Komati Formation, Bar- lower mantle. Nature, 436, 377–380. Earth. Meteorit. Planet. Sci., 35, 1309– berton Greenstone Belt, South Africa. Lyubetskaya, T. and Korenaga, J., 2007a. 1320. Earth Planet. Sci. Lett., 150, 303–323. Chemical composition of Earth’s primi- Moresi, L. and Solomatov, V., 1998. Parsons, B., 1982. Causes and conse- tive mantle and its variance, 1, methods Mantle convection with a brittle litho- quences of the relation between area and and results. J. Geophys. Res., 112, sphere: thoughts on the global tectonic age of the ocean floor. J. Geophys. Res., B03211. Doi: 10.1029/2005JB004223. styles of the Earth and Venus. Geophys. 87, 289–302. Lyubetskaya, T. and Korenaga, J., 2007b. J. Int., 133, 669–682. Parsons, B. and McKenzie, D., 1978. Chemical composition of Earth’s primi- Morgan, W.J., 1971. Convection plumes in Mantle convection and the thermal tive mantle and its variance, 2, implica- the lower mantle. Nature, 230, 42–43. structure of the plates. J. Geophys. Res., tions for global geodynamics. Morgan, J.W., 1986. Ultramafic xenoliths – 83, 4485–4496. J. Geophys. Res., 112, B03212. Doi: clues to Earth’s late accretional history. Parsons, B. and Sclater, J.G., 1977. An 10.1029/2005JB004224. J. Geophys. Res., 91, 2375–2387. analysis of the variation of ocean floor

2008 Blackwell Publishing Ltd 437 Plate tectonics, flood basalts and Earth’s oceans • J. Korenaga Terra Nova, Vol 20, No. 6, 419–439 ......

bathymetry and heat flow with age. Saal, A.E., Hauri, E.H., Langmuir, C.H. the presence of preexisting convection. J. Geophys. Res., 82, 803–827. and Perfit, M.R., 2002. Vapour J. Geophys. Res., 93, 7672–7689. Pearson, D.G., 1999. The age of continen- undersaturation in primitive mid-ocean- Smyth, J.R. and Jacobsen, S.D., 2006. tal roots. Lithos, 48, 171–194. ridge basalt and the volatile content of Nominally anhydrous minerals and Phipps Morgan, J. and Chen, Y., 1993. The Earth’s upper mantle. Nature, 419, Earth’s deep water cycle. In: Earth’s genesis of oceanic crust: Magma injec- 451–455. Deep Water Cycle (S. Jacobsen and tion, hydrothermal circulation, and Salters, V.J.M. and Stracke, A., 2004. S. van der Lee), pp. 1–11. American crustal flow. J. Geophys. Res., 98, Composition of the depleted mantle. Geophysical Union, Washington, DC. 6283–6297. Geochem. Geophys. Geosys., 5, Q05004. Solheim, L.P. and Peltier, W.R., 1994. Pollack, H.N., Hurter, S.J. and Johnson, Doi: 10.1029/2003GC000597. Avalanche effects in phase transition J.R., 1993. Heat flow from the Earth’s Sandiford, M., Van Kranendonk, M.J. and modulated thermal convection. interior: analysis of the global data set. Bodorkos, S., 2004. Conductive incuba- J. Geophys. Res., 99, 6997–7018. Rev. Geophys., 31, 267–280. tion and the origin of dome-and-keel Solomatov, V.S., 1995. Scaling of Rama Murthy, V., van Westrenen, W. and structure in Archean granite-greenstone temperature- and stress-dependent Fei, Y., 2003. Experimental evidence terrains: a model based on the eastern viscosity convection. Phys. Fluids, 7, that potassium is a substantial radio- Pilbara Craton, Western Australia. 266–274. active heat source in planetary cores. Tectonics, 23, TC1009. Doi: 10.1029/ Solomatov, V.S., 1996. Can hotter mantle Nature, 423, 163–165. 2002TC001452. have a larger viscosity? Geophys. Res. Regenauer-Lieb, K., Yuen, D.A. and Schubert, G. and Reymer, A.P.S., 1985. Lett., 23, 937–940. Branlund, J., 2001. The initiation of Continental volume and freeboard Solomatov, V.S., 2001. Grain size-depen- subduction: criticality by addition of through geological time. Nature, 316, dent viscosity convection and the ther- water? Science, 294, 578–580. 336–339. mal evolution of the earth. Earth Planet. Reymer, A. and Schubert, G., 1984. Schubert, G., Stevenson, D. and Cassen, Sci. Lett., 191, 203–212. Phanerozoic addition rates to the conti- P., 1980. Whole planet cooling and the Solomatov, V., 2007. Magma oceans and nental crust and crustal growth. Tecton- radiogenic heat source contents of the primordial mantle differentiation. In: ics, 3, 63–77. earth and moon. J. Geophys. Res., 85, Treatise on Geophysics (G. Schubert, Richards, M.A., Duncan, R.A. and 2531–2538. ed.), Vol. 9, pp. 91–119. Elsevier, Courtillot, V.E., 1989. Flood basalts and Shirey, S.B., Carlson, R.W., Richardson, Amsterdam. hot-spot tracks: plume heads and tails. S.H., Menzies, A., Gurney, J., Pearson, Spohn, T. and Breuer, D., 1993. Mantle Science, 246, 103–107. D.G., Harris, J.W. and Wiechert, U., differentiation through continental Richards, M.A., Yang, W.-S., 2001. Archean emplacement of eclogite growth and recycling and the thermal Baumgardner, J.R. and Bunge, H.-P., components into the lithospheric man- evolution of the Earth. In: Evolution of 2001. Role of a low-viscosity zone in tle during formation of the Kaapvaal the Earth and Planets, Geophys. Monogr. stabilizing plate tectonics: implications Craton. Geophys. Res. Lett., 28, 2509– Ser. (E. Takahashi, R. Jeanloz and R. for comparative terrestrial planetology. 2512. Rudie, eds), pp. 55–71. AGU, Washing- Geochem. Geophys. Geosys., 2, Shirey, S.B., Kamber, B.S., Whitehouse, ton, DC. 2000GC000115. M.J., Mueller, P.A. and Basu, A.R., Stacey, F.D., 1981. Cooling of the Earth – Richter, F.M., 1985. Models for the 2008. A review of the isotopic evidence a constraint on paleotectonic hypo- archean thermal regime. Earth Planet. for mantle and crustal processes in the theses. In: Evolution of the Earth (R.J. Sci. Lett., 73, 350–360. Hadean and Archian: implications for O’Connell and W.S. Fyfe, eds), pp. 272– Ringwood, A.E. and Green, D.H., 1964. the onset plate tectonic subduction. In: 276. American Geophysical Union, Experimental investigations bearing on When Did Plate Tectonics Start on Washington, DC. the nature of the Mohorovicic disconti- Earth? (K.C. Condie and V. Pease, eds), Stein, M. and Hofmann, A.W., 1994. nuity. Nature, 201, 566–567. pp. 1–29. Geological Society of America, Mantle plumes and episodic crustal Rino, S., Komiya, T., Windley, B.F., Boulder, CO. growth. Nature, 372, 63–68. Katayama, I., Motoki, A. and Hirata, Silver, P.G. and Behn, M.D., 2008. Inter- Stein, C.A. and Stein, S., 1992. A model for T., 2004. Major episodic increases of mittent plate tectonics? Science, 319, the global variation in oceanic depth and continental crustal growth determined 85–88. heat flow with lithospheric age. Nature, from zircon ages of river sands; impli- Sleep, N.H., 1984. Tapping of magmas 359, 123–129. cations for mantle overturns in the Early from ubiquitous mantle heterogeneities: Stein, C., Schmalzl, J. and Hansen, U., Precambrian. Phys. Earth Planet. Int., an alternative to mantle plumes? 2004. The effect of rheological para- 146, 369–394. J. Geophys. Res., 89, 10,029–10,041. meters on plate behavior in a self con- Robert, F., 2003. The D ⁄ H ratio in chon- Sleep, N.H., 2000. Evolution of the mode sistent model of mantle convection. drites. Space Sci. Rev., 106, 87–101. of convection within terrestrial planets. Phys. Earth Planet. Int., 142, 225–255. Rudnick, R.L. and Gao, S., 2003. Com- J. Geophys. Res., 105, 17,563–17,578. Stevenson, D.J., 2003. Styles of mantle position of the continental crust. In: Sleep, N.H., 2007. Plate tectonics through convection and their influence on plane- Treatise on Geochemistry (H.D. Holland time. In: Treatise on Geophysics (G. tary evolution. C. R. Geosci., 335, 99– and K.K. Turekian, eds), Vol. 3, Schubert, ed.), Vol. 9, pp. 101–117. 111. pp. 1–64. Elsevier, Amsterdam. Elsevier, Amsterdam. Stevenson, D.J., Spohn, T. and Schubert, Ru¨ pke, L.H., Phipps Morgan, J., Hort, M. Sleep, N.H. and Windley, B.F., 1982. G., 1983. Magnetism and thermal evo- and Connolly, J.A.D., 2004. Serpentine Archean plate tectonics: constraints and lution of the terrestrial planets. Icarus, and the subduction zone water cycle. inferences. J. Geol., 90, 363–379. 54, 466–489. Earth Planet. Sci. Lett., 223, 17–34. Sleep, N.H. and Zahnle, K., 2001. Carbon Tackley, P.J., 2000. Mantle convection and Ryder, G., 2002. Mass flux in the ancient dioxide cycling and implications for plate tectonics: toward an integrated Earth-Moon system and benign impli- climate on ancient Earth. J. Geophys. physical and chemical theory. Science, cations for the origin of life on Earth. Res., 106, 1373–1399. 288, 2002–2007. J. Geophys. Res., 107, 5022. Doi: Sleep, N.H., Richards, M.A. and Hager, Tackley, P.J. and Stevenson, D., 1993. A 10.1029/2001JE001583. B.H., 1988. Onset of mantle plumes in mechanism for spontaneous self-perpet-

438 2008 Blackwell Publishing Ltd Terra Nova, Vol 20, No. 6, 419–439 J. Korenaga • Plate tectonics, flood basalts and Earth’s oceans ......

uating volcanism on the terrestorial crustal maturation: oxygen isotope eds), pp. 73–81. American Geophysical planets. In: Flow and Creep in the Solar ratios of magmatic zircon. Contrib. Union, Washington, DC. System: Observations, Modeling and Mineral. Petrol., 150, 561–580. Williams, D.R. and Pan, V., 1992. Inter- Theory (D.B. Stone and S.K. Runcorn, Van Kranendonk, M.J., Smithies, R.H., nally heated mantle convection and the eds), pp. 307–321. Kluwer Academic, Hickman, A.H. and Champion, D., thermal and degassing history of the Dordrecht. 2007. Review: secular tectonic evolution Earth. J. Geophys. Res., 97, 8937–8950. Tarduno, J.A., Cottrel, R.D., Watkeys, of Archean continental crust: interplay Wilson, M., 1989. Igneous Petrogenesis. M.K. and Bauch, D., 2007. Geomagnetic between horizontal and vertical pro- Springer, Netherlands. field strength 3.2 billion years ago cesses in the formation of the Pilbara Wilson, D.S., Teagle, D.A.H., Alt, J.C. recorded by single silicate crystals. Craton, Australia. Terra Nova, 19, et al., 2006. Drilling to gabbro in intact Nature, 446, 657–660. 1–38. ocean crust. Science, 312, 1016–1020. Taylor, S.R. and McLennan, S.M., 1985. Walter, M.J., 1998. Melting of garnet Wise, D.U., 1974. Continental margins, The Continental Crust: its Composition peridotite and the origin of komatiite freeboard and the volumes of continents and Evolution. Blackwell, Boston. and depleted lithosphere. J. Petrol., 39, and oceans through time. In: Geology van Thienen, P., Vlaar, N.J. and van den 29–60. of Continental Margins (C.A. Burk and Berg, A.P., 2004. Plate tectonics on the Wedepohl, K.H., 1995. The composition of C.L. Drake, eds), pp. 45–58. Springer, terrestrial planets. Phys. Earth Planet. the continental crust. Geochim. Cosmo- New York. Int., 142, 61–74. chim. Acta, 59, 1217–1232. Workman, R.K. and Hart, S.R., 2005. Tozer, D.C., 1985. Heat transfer and White, R.S. and McKenzie, D., 1995. Major and trace element composition of planetary evolution. Geophys. Surv., 7, Mantle plumes and flood basalts. the depleted MORB mantle (DMM). 213–246. J. Geophys. Res., 100, 17,543–17,585. Earth Planet. Sci. Lett., 231, 53–72. Turcotte, D.L. and Schubert, G., 1982. Wilde, S.A., Valley, J.W., Peck, W.H. and Yaxley, G.M., 2000. Experimental study of Geodynamics: Applications of Continuum Graham, C.M., 2001. Evidence from the phase and melting relations of Physics to Geological Problems. John detrial zircons for the existence of con- homogeneous basalt + peridotite Wiley, New York. tinental crust and oceans on the Earth mixtures and implications for the Valley, J.W., Lackey, J.S., Clechenko, 4.4 Gyr ago. Nature, 409, 175–178. petrogenesis of flood basalts. Contrib. A.J.C.C.C., Spicuzza, M.J., Bindeman, Williams, Q., 1998. The temperature con- Mineral. Petrol., 139, 326–338. M.A.S.B.I.N., Ferreira, V.P., Sial, A.N., trast across D’’. In: The Core-Mantle Peck, E.M.K.W.H., Sinha, A.K. and Boundary Region (M. Gurnis, M.E. Received 2 April 2008; revised version Wei, C.S., 2005. 4.4 billion years of Wysession, E. Knittle and B.A. Buffett, accepted 5 September 2008

2008 Blackwell Publishing Ltd 439