Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
2 Earthâs Interior A s planets age they slowly evolve as the heat tion 6 deals with Earth materials properties, which trapped and generated in the interior is trans- control many of the internal processes discussed in ported to the surface. The internal planetary this chapter. processes that move this heatâincluding volcanism and convectionâhave a huge influence on the nature of QUESTION 4: HOW DOES EARTHâS planetary surfaces. Yet the vast interior is inaccessible to INTERIOR WORK, AND HOW DOES IT direct study and must be understood with geophysical AFFECT THE SURFACE? observations, experimental studies of materials under deep-Earth conditions, and theoretical models. For The previous chapter discussed evidence that Earth over a century, seismic wave, geomagnetic, and gravity and the Moon, and by extension the other terrestrial measurements made at the surface have been improv- planets, started out with high internal temperatures ing our characterization of Earthâs internal structure. about 4.5 billion years ago. Once the planetary accre- Experimental and theoretical determinations of mate- tion process tails off, the planets cool, first through a rial properties at high temperatures and pressures and period of active geological processes and ultimately to numerical modeling of mantle and core heat transport a state of geological quiescence. When the planet is and convection over very long timescales also play key geologically active, evidence of that activity is reflected roles in studies of internal dynamics. However, despite in the nature of its surface and atmosphere and perhaps continuing advances, we still cannot uniquely describe the existence of a magnetic field. After the interior Earthâs mantle structure or explain in any detail how cools and its viscosity increases sufficiently, geological the core and mantle work, why Earth differs from other activity grinds to a halt, and the planetâs surface stops planets, or how it may change in the future. regenerating. Thereafter, only external processes, such The three questions included in this chapter de- as bombardment with asteroids, further modify the scribe scientific challenges for understanding Earthâs surface. evolution and internal dynamics. Question 4 addresses Some planetary bodies, like the Moon, cooled deep-Earth dynamical processes, from the inner metal- quickly and have been geologically inactive for bil- lic core at the center of Earth to the convecting mantle lions of years. Despite rapid cooling after the Moon- to the volcanoes at the surface. Question 5 focuses on forming impact (Questions 1 and 2), Earth produced the near-surface features of Earthâold continents, and retained enough heat to power geological activity young ocean basins, and plate tectonicsâthat make until the present, and it is likely to do so for several Earth unique among Solar System planets and that also billion more years. However, both the amount of seem inextricably linked to the presence of water and Earthâs cooling and resulting changes in the internal the preservation of life-sustaining conditions. Ques- dynamics and surface environment are still poorly 35
36 ORIGIN AND EVOLUTION OF EARTH known. Although we know that heat is transported by mantle and then sinks again in a turbulent pattern that mantle convection, we do not yet have the capability is affected by rotation and the magnetic field the flow to exactly describe these convective patterns, calculate generates. By contrast, mantle motions are ponderous. with confidence how different they were in the past, or Typical velocities are about 5 cm/yr (based on geodetic, predict how they will change in the future. Resolving magnetic, seismic, and geological measurements), and the critical questions about planetary evolution will at this rate the nominal âround-tripâ journey of a mantle require much more advanced knowledge of planetary wide convection cellâacross the surface for 5,000 km, materials and how they affect convection (Question down 2,900 km to the bottom of the mantle, and back 6), better constraints from seismology on the present to the surface againâwould take about 300 million configuration of mantle flow at both large and small years. This rate of travel is consistent with simple ther- scales, and significant advances in mathematical mod- mal convection models that treat the mantle as if it were eling of convection that is driven by both temperature a liquid with a viscosity (estimated from postglacial and chemical variations. rebound rates) of about 1021 Pa-s. The configuration of convection in Earthâs mantle provides the primary Convection and Heat Flow control on how Earth cools, mainly because the mantle makes up roughly two-thirds of Earthâs mass and 85 About 43 TW (1012 J/s) of heat flows from Earthâs percent of its volume (Figure 2.1). interior through its surface at present, based on global Mantle motions carry hot material from deep heat flow measurements and thermal models for cool- inside Earth toward the surface, where heat is lost to ing oceanic lithosphere. Sources of this surface heat the atmosphere and ultimately to space, and also carry flow include the slow cooling of the mantle and core cold surface rocks down to great depths. Unresolved over the history of the planet; heating produced by issues concerning mantle convection arise from un- radioactive decay of U, Th, and K; and minor sources certainties about material properties at high pressures such as tidal heating. The exact contribution of each and temperatures. Experiments and field evidence to the planetâs heat flow is uncertain. For example, we show that mantle rock becomes soft enough to flow do not know how much U, Th, and K are contained over geological time periods at depths of just 30 to 60 inside Earth and how these elements are distributed km, where the temperature surpasses 700Â°C and pres- (McDonough, 2007). These elements are more effec- sure reaches several thousand atmospheres. At higher tive at keeping Earth hot if they are located deep within temperatureâabove 1200Â°Câthe viscosity of mantle the mantle, or even to some degree in the core, rather rock is low enough that it behaves much like a thick than near the surface. As a result of these uncertainties, liquid; almost all of the mantle is hotter than 1200Â°C. we cannot yet answer the simple question: How fast is Mantle viscosity exerts the primary control on the form Earth cooling? of convection and the efficiency at which heat is moved The primary mechanism for transporting heat toward Earthâs surface. However, other factors also are within Earthâs interior is convection. It was once be- important. For example, viscous dissipation associated lieved that mantle convection was impossible because with deformation of stiff lithospheric plates at subduc- the mantle was demonstrably solid. But much like a tion zones strongly affects the form of convection and glacier, the mantle can behave like both a brittle solid the relationship between convective vigor and surface and a liquid: it fractures when deformed rapidly but heat flow. The largest uncertainties are for the lower flows on long timescales. We now know that both mantle. Seismological data suggest that the flow pat- the mantle and the outer core circulate in a complex tern there is complex. Other observations suggest that pattern of large- and small-scale flows. In the molten viscosity increases in the lower mantle, and numerical outer core, which has very low viscosity (some estimates models indicate that flow velocities in the lower mantle suggest a value similar to that of liquid mercury), con- may be much slower than plate velocities such that the vection is rapid. Hot liquid metal circulates up to the overturn time is a billion years or more (Kellogg et al., top of the core where it loses heat to the base of the 1999; Ren et al., 2007).
EARTHâS INTERIOR 37 FIGURE 2.1â Cutaway view of Earthâs interior showing major layers (oceanic and continental crust, upper mantle, lower mantle, outer core, inner core) and features (mantle plumes, subduction zones, midocean ridges, convection currents, magnetic field). SOURCE: Lamb and Sington (1998). Copyright 1998 Princeton University Press. Reprinted with permission.
38 ORIGIN AND EVOLUTION OF EARTH How Are Mantle Convection and Earthâs Thermal understanding is insufficient to explain many of the Evolution Related? most important geological and geochemical features of our planet. We know that mantle convection is driven by the heat We are even further from understanding the in- of Earthâs interior, but what controls Earthâs tem- ternal evolution of other rocky bodies of our Solar perature? The current understanding is that the mantle System, where we have fewer data, and interactions itself acts as Earthâs primary âtemperature regulator,â between thermal evolution and orbital evolution pro- and its actions depend on the atomic-scale properties vide additional complications (see Box 2.1). Earth of mantle minerals that determine viscosity. The ef- (and possibly Venus) has apparently maintained a high fective viscosity of the mantle depends on the rate at enough internal temperature to ensure continued geo- which the mineral grains can deform in response to an logical activity. However, on smaller planetary bodies, applied stress, which in turn is strongly dependent on geological surface activity has either long since stopped temperature. Laboratory data indicate that for a given (Moon) or slowed greatly (Mars). It is believed that stress a 100Â°C temperature increase lowers the viscosity the mantles of other terrestrial planets should function by about a factor of 10. Consequently, if Earth were to in the same way as Earthâs, unless there are different heat up, it would convect more vigorously and lose heat amounts of radioactive elements or different amounts faster. As heat is lost, temperature drops and convection of water dissolved in the mantle minerals. The addi- slows, decreasing the rate of heat loss. This tempera- tion of tiny amounts of water to mantle minerals would ture-viscosity feedback should keep Earthâs internal lower both the viscosity of the mantle and the melting temperature well regulated. The temperature at which temperature (Question 6) and may prolong a planetâs the thermostat is most likely to be set is just below the geologically active life. melting point of mantle rock because there is an even faster decrease in viscosity with temperature once the What Do Mantle Plumes Tell Us About mantle begins to melt. Convection and Heat Transport? The temperature-viscosity feedback model is useful but it implies a steady system that undergoes only slow The viscosity of Earthâs mantle is sufficiently low changes over long periods of time. This implication and sensitive to temperature that convection can in- is at odds with much of what we know and suspect clude complex small-scale currents. Evidence of this about mantle materials and geological history. For small-scale convection is provided by hot spotsâlarge example, the continents, which are an end product of clusters of volcanoes, the most active of which are in Earthâs evolution, show evidence of rapid growth spurts Hawaii, Iceland, the Galapagos Islands, Yellowstone, (Question 5), which may or may not be associated and RÃ©union (Indian Ocean). Hot spots are usually with accelerated plate tectonics (Hoffman and Bow- explained as the surface outpourings of magma formed ring, 1984). The seafloor of the western Pacific Ocean in mantle plumes, which are cylindrical upwellings of contains enormous volcanic mountain ranges, which hot (and hence low viscosity) rock that are thought to suggests that the Cretaceous Period (65 to 150 Ma) form near the base of the mantle and rise to the surface was a time of exceptionally intense volcanic activity at rates much faster than plate velocities (Figure 2.2). and possibly also fast seafloor spreading (Engebretson Mantle plumes should form as a consequence of heat et al., 1992). We also know that the Cretaceous was a entering the bottom of the mantle from the much hot- period of exceptional global warmth and high sea level ter outer core. (Question 7) and stability of Earthâs magnetic field. Mantle plumes may also be responsible for large These observations as well as theoretical considerations igneous provinces, which are vast basalt lava plateaus on raise the question of whether Earthâs thermal evolution continents and the ocean floor. The best current expla- and internal processes are adequately described by our nation is that they form when the bulbous top of a new (quasi-) steady state models or whether the evolution plume approaches Earthâs surface (Figure 2.2), then has been unsteady and punctuated by catastrophic re- spreads out and causes widespread melting (Ernst et configurations. Thus, even though we understand the al., 2005). These large, rapid lava outpourings may have most basic features of mantle convection, our level of caused major perturbations to Earthâs climate (Ques-
EARTHâS INTERIOR 39 BOX 2.1 Planetary Comparisons Our understanding of our planetary neighbors has advanced substantially over the last several decades through spacecraft exploration and analysis of lunar samples and meteorites from Mars and the Moon. The other terrestrial planetary bodies (Venus, Mars, Mercury, and the Moon) formed by the same processes as Earth (see Question 1) and are governed by the same physical and chemical laws and processes. Nevertheless, each has taken a distinct evolutionary track, deepening the questions we pose for how Earth works the way it does. Venus, at 0.8 Earth masses, is sometimes called Earthâs âsister planet,â but its massive carbon dioxide atmosphere (90-bar surface pressure) and global cloud cover have led to a runaway greenhouse, a surface temperature of 470Â°C, and the loss of most water from the atmosphere. Venus also lacks Earth-like plate tectonics, but the planet has been subjected to resurfacingâprobably by some form of lithospheric recycling not understoodâat least once and perhaps multiple times. The density of impact craters indicates that the surface has an average age between several hundred million years and 1 billion years. There are mountain belts and pervasively deformed plateaus, both of which are stratigraphically older than the widespread volcanic plains, known to be basaltic from spacecraft lander measurements. Unlike Earth, Venus has no detectable internal magnetic field. A strong correlation of long-wavelength gravity and topography in the plains is the signature of ongoing mantle convection. Rifting and volcanism have occurred more recently than the average surface age, and the planet is likely to be volcanically and tectonically active at present. Mars, at 0.1 Earth masses, evolved more rapidly than Earth or Venus. Isotopic evidence from Martian meteorites indicates that Mars formed its core, mantle, and most of its crust within a few tens of millions of years after the beginning of Solar System formation, probably without any plate tectonics era. Large segments of the most ancient crust on Mars are strongly magnetized, relics of a core dynamo that began early in Martian history but probably died out after several hundred million years. The Martian surface has seen a mix of plains volcanism and more focused magmatism in regional centers, dominated by the Tharsis volcanic province, largely constructed before 4 Ga (billion years ago). Fluvial landforms, widespread chemical alteration, and sedimentary deposits visited by surface rovers all indicate that water was an important agent of geological change early in Martian history. At about 4 Ga, Mars lost its global magnetic field, its carbon dioxide atmosphere was substantially thinned by solar wind stripping, the climate cooled, and water lost its dominant role in surface change. Martian volcanism continued at generally declining rates, and the planet may still be active at low levels today. The Moon and Mercury, at 0.01 and 0.05 Earth masses, respectively, have heavily cratered surfaces and only extremely tenuous atmospheres, but their similarities end there. The Moon began largely molten, presumably the result of the accumulation of hot ejecta from a giant impact on the early Earth. Cooling and solidification of the resulting magma ocean led to formation of the crust and the mantle source regions of later volcanic lavas. Those lavas erupted to partially fill the lunar maria, mostly on the lunar nearside at 3 to 4 Ga, but there are also isolated younger volcanic deposits. The Moon may have a small iron-rich core, but if so it is no more than a few percent by mass. Lunar rocks from 3 to 4 Ga are magnetized, but whether the magnetizing field was a central core dynamo or transient field generated during surface impacts is an open question. The Moon is seismically active at low levels today. Shallow moonquakes are probably the signature of interior cooling, whereas deep moonquakes occur in clusters and appear to be triggered by tidal stresses. Mercury, in contrast, has such a high bulk density that its iron-rich core comprises at least 60 percent of the planetâs mass. Mercury has a global magnetic field, dipolar like that of Earth, and the outer core is known to be molten on the basis of the amplitude of the planetâs libration forced by solar torques as Mercury progresses along its elliptical orbit. The planet has an ancient, heavily cratered crust, as well as somewhat younger plains units that may be volcanic in origin. The surface composition is poorly known, but Earth-based measurements indicate that surface silicate materials have little or no ferrous iron. The dominant tectonic landforms on Mercury are high-relief lobate scarps, the surface expressions of large-offset thrust faults. Because of the extensive distribution and apparently random orientation of these features, the lobate scarps have been interpreted to record an extended period of global contraction, the result of some combination of interior cooling and solidification of an inner core. tion 7) and perhaps even major extinctions (Question 200 to 600 km, while others seem to extend almost to 8). Other indications of plumes include broad bulges the core-mantle boundary. However, there is abun- in the ocean floor, such as those around Hawaii, and dant evidence for much larger, domical or irregularly the tremendously excessive amount of lava produced at shaped low-velocity features in the lower mantle that Iceland in comparison to other places along the Mid- are sometimes called superplumes (Figure 2.3). Does Atlantic ridge. this mean that thermal plumes do not exist in the Although there is good geological evidence that lower mantle or that the seismic resolution is still too mantle plumes exist, seismological evidence for the low to make them out? Seismic data suggest that the existence of narrow, hot, cylindrical upwellings in large low-seismic-velocity regions near the base of the the lower mantle is only equivocal. Some cylindrical mantle are anomalously dense, which is contrary to regions of low velocity appear to extend downward to expectations for buoyant thermal upwellings (Ishii and
40 ORIGIN AND EVOLUTION OF EARTH FIGURE 2.2â Sketch of mantle convec- tion and structure based on inferences from fluid mechanics and seismologi- cal data. SOURCE: Courtesy of Geoff Davies, Australian National University. After Davies (1999). Copyright 1999 by Elsevier Science and Technology Jour- nals. Reproduced with permission. Figure 2.2.eps Tromp, 1999). However, it is becoming better appreci- or in layers. Models, geochemical analyses of mantle ated that temperature variations may not be the only magmas erupted on the surface, and interpretations of source of buoyant upwellings in the mantle. Chemical seismic waves that have passed through Earth have all variations may be large enough to affect large-scale yielded different answers. In general, mantle models mantle flow, and mantle plumes can have both thermal based on geochemistry suggest that mantle convection and chemical components to their buoyancy (Davaille, occurs in two layers, whereas most geophysical evidence 1999; Farnetani and Samuel, 2003). and numerical models strongly support whole-mantle convection. Reconciling these differences is impor- Does Convection Occur Through the Whole tant for understanding Earthâs volcanic and thermal Mantle or in Layers? evolution. A key question about the modern form of mantle flow Geochemical analysis. Interactions of the mantle with is whether convection occurs through the whole mantle the core and surface chemically alter the upper and lower boundary regions of the mantle (discussed be- low). Convection then stirs this altered material back into the main volume of the mantle. The chemical com- position of lavas derived from the mantle provides clues about the extent to which these heterogeneities persist in time and hence about the nature of mantle convec- tion (Van Keken et al., 2002). Lavas (and most other rocks) contain every one of the 90 naturally occurring elements in the periodic table, although about 75 are present in small abundances. With new techniques the concentration of each of the 90 elements and the rela- tive amounts of isotopes of about half of the elements can be measured precisely. The isotopes formed by ra- dioactive decay (206Pb, 207Pb, 208Pb, 87Sr, 143Nd, 230Th, 226Ra, and others) provide detailed information about mantle evolution as well as the processes that produce FIGURE 2.3â Representation of large-scale seismic velocity and transport magma. structure of the mantle. Red zones have relatively slow P-wave velocity and blue zones are relatively fast. Slow velocities Low-abundance trace elements and isotopes of are thought to represent hotter parts of the mantle. SOURCE: Pb, Sr, Nd, Hf, He, and Os show large, nonrandom <http://www.seismology.harvard.edu/Projects.html>. See also variations among volcanic rocks. Basalt lavas erupted Su et al. (1994). Used with permission. at midocean ridges differ systematically from those
EARTHâS INTERIOR 41 erupted at hot spots. Midocean ridge lavas also vary bulk composition derived from meteorites (Ques- from ridge to ridge and along individual ridges (Figure tion 1) and with the distribution of heterogeneities 2.4). The chemical differences between hot spots and at depth. midocean ridges have long been considered evidence that the lower mantle (whence mantle plumes presum- Seismic interpretation. The most direct observations ably come) is different, and convects separately, from available for inferring the present-day configuration the upper mantle (Hofmann, 1997). of mantle convection are provided by seismicity in Nevertheless, there are complications in the isoto- subduction zones and three-dimensional seismic pic data. For example, 3He data suggest that parts of the tomography models of the interior. Seismic velocity mantle are relatively unaltered, or at least less degassed, variations are caused by changes in pressure, tem- but other isotopes (Nd, Sr, Pb, Hf ) tell a different story perature, composition, and mineral alignment, so (Moreira et al., 2001). The mantle overall does not interpretation of the models requires information seem to have an Nd or Hf isotopic composition that from mineral physics (Question 6) and geodynamics. properly complements that of the continental crust. High-seismic-velocity features corresponding to cold Many such chemical clues must be sorted out before sinking oceanic lithosphere are clearly observed in re- we can develop a model for mantle convection and the gional and global seismic tomography models (Figure mantle-crust system that agrees with models for Earthâs 2.5). Low-velocity features (presumably signifying FIGURE 2.4 (Left) Bathymetry of the Mid-Atlantic ridge and topography of adjacent continents. SOURCE: <http://www.ngdc.noaa. gov/mgg/image/2minrelief.html>. (Right) Variations of neodymium isotopic composition in basalt lavas from along the Mid-Atlantic ridge plotted against latitude. Zero on the epsilon scale corresponds to the bulk Earth value, which assumes Earth has the same Sm/ Nd ratio as chondritic meteorites. The high degree of heterogeneity indicates that diverse materials are generated in the mantle by melting and subduction and that these heterogeneities are not homogenized by convection. SOURCE: Data from the online database PetDB, averaged by ridge segment by Su (2002).
42 ORIGIN AND EVOLUTION OF EARTH FIGURE 2.5â Seismic tomography data indicating that in some areas subduct- ed slabs extend through the 660-km discontinuity and well down into the lower mantle. Blue shading indicates higher seismic body wave (P) and shear wave (S) velocity, both of which should correlate with lower temperature. The thickness of the cold slab, however, is only about 50 to 100 km, whereas the thickness of the high-velocity (blue) zone is close to 500 km in the lower mantle. The greater thickness in the lower mantle could be due to deformation of the slab or to decreased spatial resolution of the image at greater depth. SOURCE: After Trampert and van der Hilst (2005). Copyright 2005 American Geophysical Union. Reproduced with permission. Figure 2.5.eps relatively high temperature) underlie ocean ridges, back bitmap image clearly penetrating the 660-km boundary. The variable arc basins, and tectonically active areas of continents. depth of lithospheric slab subduction is not easily un- Continental cratonic areas are underlain by high-seis- derstood in the context of simple thermal convection mic-velocity regions extending 250 to 350 km deep, in- and is the primary observation driving consideration dicating fundamental differences between oceanic and of more complex convection models involving both continental plates (Question 5). Deeper seismic-veloc- thermal and chemical effects. ity structures are less easily related to surface tectonics, with very large scale structures tending to dominate in Models. Mantle convection models have progressed the transition zone from 410 to 660 km deep, and in the from simplified two-dimensional models to complex lowermost mantle above the core-mantle boundary. For three-dimensional simulations, in concert with increas- several decades the resolution of seismic tomography ing computing power and improving knowledge of models has been improving, and this is guiding numeri- mantle material properties (Figure 2.6; see also Cohen, cal modeling of mantle flow processes. 2005). Comparison with seismological models allows Seismic evidence shows a large velocity disconti- some parameters in convection models to be tested, but nuity 660 km below the surface, which is thought to many issues are still unresolved. Among the challenges involve mineral phase transformations that tend to of simulating mantle convection are the strong depen- impede flow through the transition depth. However, dence of viscosity on temperature and composition, seismological data also show some subducted slabs ex- mineralogical heterogeneity in the mantle on both large tending to depths greater than 1,000 km (Figure 2.5), and small scales, departures from simple fluid behavior,
EARTHâS INTERIOR 43 FIGURE 2.6â Computer simulation of mantle convection in two dimensions. Red-green-blue color scale depicts temperatures from 4000Â°C to 0Â°C. Fine-scale features, which arise from extreme variations in material properties at small length scales, are not well Figure 2.6.eps represented in this simulation, but hot upwellings from the core-mantle boundary region, and cold downwellings (analogous to subduc- tion) from the cold surface boundary layer, are prominent features. SOURCE: Butler et al. (2005). Copyright 2005 by Elsevier Science and Technology Journals. Reproduced with permission. and the effects of melting and phase changes on mate- seismological and geodynamic results tend to favor rial properties. Although the simulations are guided by an intermediate model of mantle convection that observations and experimental measurements, these are is neither strictly layered nor simple whole-mantle often indirect or subject to varying interpretations as a convection. result of the difficulty in specifying material properties at conditions of high pressure and temperature. For When Did Earthâs Inner Core Form? example, the uppermost mantle is mostly made of three minerals: olivine, orthopyroxene, and clinopyroxene. Earthâs thermal evolution is reflected in and strongly In the lower mantle these minerals are transformed by influenced by the temperature of the liquid outer core. pressure into higher density forms, and the size and The fact that Earthâs outer core is liquid rather than even the composition of the mineral grains are poorly solid is evidence for the hot origin of Earth, and the fact known. It is the deformation characteristics of these that the core has not completely solidified over Earthâs mineral aggregates that determine the nature of mantle 4.5-billion-year history means that it has been pre- convection. Because the grain size and other proper- vented from losing heat too quickly. Laboratory experi- ties of the deep mantle have yet to be determined, our ments suggest that the top of the core is about 1500Â°C ideas about convection in the lower mantle involve large hotter than the deep mantle (Figure 2.7). Therefore, extrapolations of the properties we can determine for heat must be flowing from the outer core into the lower Earth materials at lower pressure and temperature (see mantle, and the core must be cooling. The core must Question 6). also be close to its solidification temperature because Numerical simulations of mantle convection show the inner core is solid. As the core cools, it solidifies that even with phase transitions inhibiting flow and from the bottom up, so we deduce that the solid inner a viscosity increase in the lower mantle, it is plausible core is growing and the liquid outer core is shrinking. that large-scale transport of material between the The inner coreâouter core boundary must have a upper and lower mantle does occur. All in all, current temperature exactly equal to the melting temperature of
44 ORIGIN AND EVOLUTION OF EARTH FIGURE 2.7â Schematic representation of average temperature in Earthâs interior versus depth. Viscosity estimates are also shown. Temperature is highly uncertain below about 500-km depth. The average mantle temperature (red line) is based on an adiabatic gradient and a temperature of 1350Â°C at a pressure of 1 bar. Higher and lower temperatures for plumes and subducted slabs are approximate but close to estimates. Temperature in the core and the large temperature drop at the base of the mantle are poorly constrained. See van der Hilst et al. (2007) for a recent estimate of temperature at the core-mantle boundary and Bunge (2005) for a discussion of nonadiabatic temperature structure in the mantle. the core at the corresponding pressure. The core melting started, it would slow cooling of the core because temperature is uncertain because the core contains mi- crystallization releases heat. It has recently been nor elements other than Fe, and it is not known exactly inferred from convection models that the inner core which elements and how much of them. Hence, the may be relatively young; it may have begun forming melting temperature of the core is likely to be a complex about 1.5 billion to 2 billion years ago (Labrosse et function of both composition and material properties al., 2001). This idea, however, is inconsistent with at high temperature and pressure. Ongoing research is theoretical models that suggest the presence of a examining the possibility that heat-producing elements solid inner core may be important for the strength (e.g., potassium) may be present in the core and may of the magnetic dipole field and for the occurrence contribute to a slowing of core cooling. of reversals. Moreover, there is evidence that Earthâs How long the inner core has existed, its rate of magnetic field is older than 2 billion years (Tarduno growth, and why the core has not fully solidified are et al., 2007). This apparent conundrum may be partly fundamental unresolved issues (Butler et al., 2005). a consequence of our still poor understanding of the Part of the answer seems to be that the core has been characteristics and processes near the core-mantle kept in a molten state by the mantle, which because boundary, including the values of the temperature of its much higher viscosity does not remove heat fast contrast and the amount of heat flowing across the enough. Also, once crystallization of the inner core boundary (e.g., Bunge, 2005).
EARTHâS INTERIOR 45 How Has Earthâs Magnetic Field Evolved Through Time? It has long been recognized that the main part of the geomagnetic field is sustained by fluid motions in Earthâs electrically conducting outer core. These mo- tions cause the magnetic field to change over many timescales, from diurnal to annual to geological time- scales. However, a unified picture of how the geody- namo and the core fit into the Earth system has not yet emerged. Important questions about the internal op- eration of the geodynamo and the relationship between the geodynamo and other Earth processes remain unanswered. These include: How do the geodynamo, mantle convection, and plate tectonics interact? What role did the geodynamo play in Earthâs early history? The age of the magnetic field is of interest because the magnetosphere may help keep Earth habitable. For example, the magnetosphere may have been necessary to help Earth retain its atmosphere against the eroding powers of the solar wind, and it partly shields Earthâs FIGURE 2.8â A snapshot of a three-dimensional computer surface against radiation from space. How important simulation of the geodynamo. The magnetic field is illustrated using lines of force; blue lines represent the inward directed field the latter is in preserving life or in modulating the and yellow lines represent the outward directed field. The field is rates of evolution is not agreed upon. New insights on intense and complicated in the modelâs fluid iron core, where it is these questions will come from continued satellite and generated by fluid motions. Like Earthâs field, the simulated field has a dominantly dipolar structure outside the core. SOURCE: ground-based observations of the geomagnetic field Courtesy of Gary Glatzmaier, University of California, Santa and the paleomagnetic field, dynamical interpreta- Cruz; and Paul Roberts, University of California, Los Angeles. tions of the coreâs seismic structure, and sophisticated numerical dynamos (e.g., Figure 2.8) and models of core evolution. carbon dioxide, and other gases, continually renew- What Are the Chemical Consequences of Mantle ing the oceans and atmosphere. Mountain building, Convection? erosion, and subduction, which also reflect the effects of mantle convection, remove these same materials The mantle interacts with Earthâs surface environ- and tend to recycle them into the deepest parts of the ment through volcanism, heat and mass exchange at mantle. At the core-mantle boundary we infer there is midocean ridges, and subduction. The mantle may also mainly heat exchange, but there is tantalizing evidence exchange material with Earthâs outer core. Overall, the of chemical interaction as well (Brandon et al., 1999). mantle mediates a grand-scale circulation of materials Still unknown are whether the processes that mediate that may extend from the core-mantle boundary to the these exchanges were different in the past. An inter- surface and back again. The nature of this mantle cir- esting possibility is that the nature of continents and culation and the processes that produce the interactions oceans that support a habitable surface environment at the mantle boundaries are critical to understanding today reflect only a particular phase of Earthâs cooling how Earthâs chemistry is continually modified. For ex- and hence might have been absent or much different ample, volcanism builds oceanic and continental crust in the past and might also be much different in the (Question 5) and releases to the atmosphere water, future.
46 ORIGIN AND EVOLUTION OF EARTH BOX 2.2 Midocean Ridge Hydrothermal Systems During seafloor production at midocean ridges, magma rises under the ridge to within a few kilometers of the ocean floor. This shallow heat source, combined with the faulting and fracturing that accompany seafloor spreading, causes large volumes of ocean water to circulate down into the oceanic crust, be heated, and then rise rapidly and return to the ocean at the ridge. On its way through the oceanic crust, the water reacts chemically with the rocks, changing their mineralogy and exchanging chemical elements. The hot water that returns to the ocean holds a much different suite of dissolved constituents than the seawater that enters the rocks on the ridge flanks. About 25 percent of Earthâs internal heat that is being lost to the surface today is carried by the water flowing through the oceanic crust at and near ocean ridges. The chemical exchange is large and has a global impact on the chemistry of the oceans. The best-known expressions of the high- temperature part of ridge hydrothermal systems are the âblack smokers,â which are mounds of sulfide minerals that form where hot (>350Â°C) fluids enter the ocean. However, most chemical exchange occurs farther away from the ridges, where the fluids are much cooler and more difficult to detect as they reenter the ocean. The chemical exchange at midocean ridges tends to add CO2 to the atmosphere-ocean system and make the ocean more acidic (Question 7). These effects are generally balanced over the long term by weathering of continental surface rocks, which tends to remove CO2 and make the ocean more alkaline, with rivers carrying the alkalinity to the oceans. The balance between hydrothermal acidity and river alkalinity is affected strongly by the mantle convection system of which the ocean ridge system is a part. The chemically altered oceanic crust is returned to the deep mantle in subduction zones, which allows the hydrothermal systems to affect the chemical composition of the entire mantle. Idealized sketch of the geological structure under a midocean ridge, showing the path that circulating water takes into the oceanic crust on the ridge flanks Box 2.2 figure.eps and back to the ocean near the ridge. The ocean sediment is pealed back in this picture to show the underlying layer of basalt lava. Gabbro is the crystal- line equivalent of basalt, and peridotite is the typical rock of the upper mantle. The partly crystallized magma under the ridge (the âcrystal mushâ), which has a temperature close to 1000Â°C, is the heat engine that drives the water circulation. SOURCE: Press and Siever (2001). Used with permission. Exchange at the surface: Volcanoes. Volcanoes and that keep Earthâs surface habitable. Volcanoes also their associated hydrothermal systems (Box 2.2) produce oceanic and continental crust (Question 5). provide the primary means by which the mantle On some other planets and moons, such as Venus passes material to the oceans, atmosphere, and crust. and Io, volcanoes are almost exclusively responsible Volcanoes probably created Earthâs early atmosphere for the surface morphology. The vast majority of and oceans, and they continue to resupply these Earthâs volcanic activity and crust production takes regions with water, CO 2, and other constituents place at midocean ridges. The current model of ocean
EARTHâS INTERIOR 47 BOX 2.3 Volcanic Origin of Oceanic Crust Earthâs mantle melts not because the temperature is raised but because the pressure is lowered as convection carries already hot rock material upward. Melting occurs during slow adiabatic cooling of this upward-moving rock and comes about because the temperature of melting decreases by about 3Â°C for every kilometer of upward movement, whereas the temperature of the rock decreases by only about 0.3Â°C/km due to expansion. The temperature and composition of lava erupted at the surface can be used to estimate the depth and temperature at which melting occurs: 150 to 90 km and 1600Â°C to 1450Â°C for especially hot mantle plumes like Hawaii (Ribe and Christensen, 1999) and 70 to 10 km and 1400Â°C to 1250Â°C under midocean ridges (Asimow and Langmuir, 2003). As a result of cooling during ascent, lava erupts with a lower temperature, typically between 1100Â°C and 1200Â°C. An essential aspect of melting in most planetary interiors is that mantle must be moving upward to melt. On Earth this happens at midocean ridges, mantle plumes, and subduction zones. Other planetsâlike Mars and Venusâdo not have plate tectonics and hence have neither midocean ridges nor subduction zones. Magmatism on these planets may result entirely from mantle plumes (Ernst et al., 2005). The rate of generation of new oceanic crust is proportional to the area of the triangular melting region multiplied by the velocity of upward movement and the melting rate per unit of upward movement (typically about 0.25 percent melting of solid peridotite per kilometer of upward motion). The depth of the onset of melting, shown here as 70 km, is determined by the mantle temperature and water content. Increasing either will increase the depth where melting begins and hence increase the amount of magma generated per unit of time and thereby increase the thickness of the oceanic crust that is formed. SOURCE: Adapted from Langmuir et al. (1992) and Richter and McKenzie (1984). floor generation by magmatic processes at midocean a conceptual leap in our understanding of melting ridges is a major success of Earth science, explaining in the mantle (see Box 2.3) that, in turn, was made the origin and characteristics of 60 percent of Earthâs possible by decades of research on the melting be- surface and relating both the thickness of the oceanic havior of mantle rocks and the percolation of magma crust and the depth of ridges below sea level to the through partially molten rock (McKenzie, 1985). temperature of the mantle upwelling under the ridge But while magma formation under ridges explains (Langmuir et al., 1992). The model emerged from the origin and thickness of the oceanic crust, many
48 ORIGIN AND EVOLUTION OF EARTH other aspects of the process must be resolved before we melting above the slab, such as small-scale convection can fully understand how volcanoes work and extend driven by frictional drag on the slab. the ocean ridge model to magma generation in other In general, our knowledge of volcanic processes is environments (see Question 9). For example, current much better for near-surface regions than for deeper models do not explain how magma produced in a regions where magma initially forms. A major objective broad, 150-km-wide zone under midocean ridges is is to understand volcanism from the bottom upâthat focused to erupt mainly within a narrow 10-km-wide is, to learn to predict the volume, composition, and zone at the ridges. The chemistry and Th-isotope ra- eruptive behavior of volcanoes from models of convec- tios of midocean ridge lavas also do not match model tion and heat transfer processes in the upper mantle predictions of the depth of the melting region or the and lower lithosphere. The bottom-up approach con- way magmas with different compositions and viscosi- trasts with traditional volcanology, which is motivated ties move and mix under the ridges (Sims et al., 2002). by hazard assessment to study volcanoes from the top More comprehensive numerical models are begin- down (Question 9). Bottom-up volcanology may also ning to incorporate chemical reactions accompanying benefit from studies of other planets, such as Mars and magma flow but still suffer from our limited knowledge Io, where boundary conditions are different enough of the mechanical properties of partially molten rock from those of Earth to allow models to be tested. and our inability to represent the chemical reactions Better models for the deep structure of volcanoes and accurately. long-term degassing of planetary interiors will require A less well understood type of magmatism occurs major leaps in our knowledge of partially molten rock in association with subduction zones. Although mod- and magma, the role of water in melting, the effect of est in number, subduction zone volcanoes represent melting on the viscosity of partially molten rock, and nearly all of the explosive volcanoes (Question 9) and the distribution of volatile elements between solids the mechanism by which much of the continental crust and liquids. is produced (Question 5). That volcanoes are located above relatively cold parts of the mantle is evidence Exchange in the interior: Subduction and mantle plumes. that a fundamentally different mechanism(s), perhaps Subduction occurs when old oceanic seafloor moves unique to Earth, is responsible for producing magma. slowly away from an oceanic ridge and across the ocean Although small-scale convection driven by frictional bottom, cools, and sinks into the mantle (Question 5). drag on the slab may cause melting above the slab, Cold subducted slabs contain rock that has reacted water is the melting mechanism invoked most often. chemically with ocean water (Box 2.2) and sediment Water (in the form of OHâ groups in minerals like derived from continents and shell-forming organisms amphibole) is carried into the mantle by subducting in the oceans. Although much of the sediment may slabs and then is lost as the slabs are metamorphosed be scraped off in the shallow part of the subduction (Tatsumi and Eggins, 1995). This water lowers the zones, the slabs carry some of it, plus chemical and melting temperature of the mantle by 200Â°C or more. isotopic traces of reaction with the ocean, down into If the released water moves upward from the cool slab the mantle. In this way subduction changes both mantle into hotter mantle above, it can produce the magma geochemistry and the volume and composition of the needed to generate volcanoes. The supply of water by oceans (Question 7). subduction to the magma-producing regions located The extent to which subducted slabs are assimilated 100 km or more below the volcanoes is confirmed by into the mantle is an open question. Some seismologi- the presence of the short-lived isotope 10Be, derived cal images have high-velocity tabular features in the from the atmosphere, in some island arc lavas. The midmantle and even at the base of the mantle that are mechanisms by which water- and CO2-rich fluids move suspected to be former oceanic lithosphere. Numerical in the mantle are poorly understood but central to this models indicate that it is plausible that sinking slabs puzzle; these mechanisms also influence the chemical remain cool and coherent all the way to the base of the and isotopic tracers that subducting slabs carry back mantle, where they pile up in a âslab graveyardâ (Figure into the deep mantle. Other processes may also cause 2.2; Christensen and Hofmann, 1994). If this happens,
EARTHâS INTERIOR 49 they could be reheated by heat flow from the core (and munication between the core and the lower mantle and their own radioactive elements) and return to the near- what processes could allow this communication to be surface environment as mantle plumes. There is geo- significant are topics of intense debate (e.g., ScherstÃ©n chemical support for this notion, which offers a direct et al., 2004). mechanism for chemical exchange between the surface and the deep mantle (e.g., Hofmann, 1997; Bizimis et Summary al., 2007). Some models suggest that subducted slabs do not sink that far before being thermally reassimilated Earthâs internal evolution governs much of the planetâs by the mantle and that the basal layers of the mantle evolution as a whole, but because the interior is mostly may be very old and relatively pristine. There is also inaccessible to direct sampling, its study requires a com- geochemical evidence for this latter model in the form bination of approaches. The seismic waves of earth- of high primordial 3He contents of large mantle plumes quakes can be used to determine the elastic properties (Courtillot et al., 2003). of Earthâs interior, and three-dimensional images of the Both models are mute on whether there is chemi- mantle and core from these waves are being produced cal exchange between the mantle and core analogous at systematically higher resolutions. The structures re- to that between the mantle and the oceans. However, vealed by seismology are interpreted using new knowl- Os isotope data suggest that some mantle plumes edge about Earth materials at high pressure, and great contain components that may have come from the advances have been made in experimental and theoreti- core (Brandon et al., 1999). This observation, if con- cal mineral physics. We now have sophisticated models firmed, would be consistent with a deep origin of for convection in the mantle and core and more precise mantle plumes, although uncertainty remains about geochemical and isotopic measurements of mantle whether this core signal could be transmitted through rocks. But there are still first-order inconsistencies in a basal mantle layer, which is both denser and more the interpretations of available observations, especially heterogeneous than the rest of the mantle (Figure 2.9; for the style of convection and the number and origin Garnero, 2000). Whether there is any chemical com- of mantle plumes. Recent discoveries of structure and evidence for an unanticipated phase at the base of the mantle have added a new dimension to mantle studies, as knowledge from seismology, fluid dynamics, geo- chemistry, and cosmochemistry comes together. Earthâs deep interior and surface are connected by volcanism and subduction. Volcanism modifies the internal chemical structure of planets, and great strides have been made in understanding the formation of magma and its transport from the mantle to the sur- face. But there is still no consensus on many aspects of Earthâs magmatic and geochemical history and their relation to surface conditions. For example, we do not know how much of Earthâs past volcanism was produced by mantle plumes and how much by plate tectonics, or why there were short periods of intense volcanic activity that could have changed the ocean basins, continents, and even global climate. In addition, we still have only FIGURE 2.9â Inferred features at the core-mantle boundary hints about how subduction zones work and how the (CMB). The notation Dâ³ is the seismological designation of the very existence of plates feeds back on the energetics of heterogeneous zone at the base of the mantle. ULVZ is ultra- low-velocity zone. SOURCE: Garnero (2000). Reprinted with the mantle convection system. Finally, we are only now permission from Annual Review of Earth and Planetary Sciences. exploring the most fundamental connections between Copyright 2000 by Annual Reviews. Earthâs core, magnetic field, mantle, and surface.
50 ORIGIN AND EVOLUTION OF EARTH QUESTION 5: WHY DOES EARTH HAVE What Is Plate Tectonics? PLATE TECTONICS AND CONTINENTS? Plate tectonics is the representation of Earthâs outermost Plate tectonics became a central organizing paradigm rock layers in terms of a small number of rigid spherical for geology over 30 years ago. The tenets of plate caps or plates. These plates are in relative motion, and tectonics theory have been so thoroughly assimilated their boundaries form the seismic (earthquake-produc- by the scientific population, and their implications ing) and tectonic (volcanic and mountain-building) so extensively pursued, that in some ways this report belts of the world. The plates interact at three types of could be considered a description of Earth science in boundariesâdivergent, convergent, and transformâall the âpost-plate tectonics era.â The questions regarding marked by the occurrence of earthquakes (Figure 2.10). plate tectonics that have now come to the fore have At divergent boundaries, plates move away from one less to do with the soundness of the theory than with another as new crust forms between them. The most the even more basic questions of why Earth has plate common type of divergent boundary occurs at the tectonics in the first place and how closely it is related midocean ridge system, which takes the expression of to other unique aspects of Earthâthe abundant water, a 40,000-km-long submarine mountain range that rises the continent-ocean elevation dichotomy, the existence about 2.5 km above the average ocean floor (Figure of life. We do not know whether it is possible to have 2.11a). At convergent boundaries of oceanic plates, one aspect without the others or how exactly they are one oceanic plate bends and subducts into the mantle. interdependent. Can these questions help us under- Convergent boundaries are the loci of the major deep stand why Earth is different from the other terrestrial earthquakes (>100 km below the surface); the principal planets? volcanic belts, notably the âring of fireâ around the Pa- 0Â° 60Â° 120Â° 180Â° 240Â° 300Â° 360Â° 90Â° N 90Â° N 60Â° N 60Â° N 30Â° N 30Â° N 0Â° 0Â° 30Â° S 30Â° S 60Â° S 60Â° S Depth (km) 90Â° S 90Â° S 0Â° 60Â° 120Â° 180Â° 240Â° 300Â° 360Â° 0 70 300 800 Figure 2.10.eps FIGURE 2.10â Locations of earthquakes of Richter magnitude 5 or greater for the period 1991 to 1997. These earthquakes mark the edges of Earthâs tectonic plates. SOURCE: Romanowicz (2008). Reprinted by permission from Macmillan Publishers Ltd.: Nature, copyright 2008.
EARTHâS INTERIOR 51 FIGURE 2.11 (a) Global topography contoured at 500-m intervals. The pre- ponderance of the continental area lies between 0 and 500 m above sea level. The ocean depth varies with the age of the seafloor. Young seafloor near ocean ridges is only about 2,000 m below sea level. Seafloor that is older than about 60 million years (see Figure 2.12) lies (a) at depths of 5,000 m or more. The elevations shown for Antarctica and Greenland represent the top of the ice sheets; the rocky surface of both areas is below sea level, where the ice is thick- est. (b) Contour map of the thickness of the crust. Continents are about 40 Â± 5 km thick except in areas of active moun- tain uplift, where they are thicker, and at their edges, where they are thinner. Oceanic crust is between 5 and 10 km thick, except in areas where there are thick volcanic plateaus. SOURCE: Data from the 2-degree resolution database CRUST 2.0; <http://mahi.ucsd.edu/ (b) Gabi/rem.dir/crust/crust2.html>. Figure 2.11.eps cific rim; and mountain building, as in the Himalayas, ferences in the types of crust that comprise the plates. the Caucasus, and the Alps. At transform boundaries, The two types of rocky crust, oceanic and continental, plates slide past one another, as along the San Andreas are distinguished by thickness, composition, and age fault in California, commonly producing large earth- (see also Question 2). Oceanic crust is thin (5 to 9 quakes but little volcanic activity. km), young (less than 200 million years old), and for A key component of the plate tectonics model is the most part fairly uniform in chemical composition, the nearly rigid moving plate. The plates are nearly consisting of basalt, which is a volcanic rock with silica rigid because the rocks near Earthâs surface are cool (SiO2) content of about 50 percent by weight. In con- and therefore strong and difficult to deform, even on trast, continental crust is thick (30 to 70 km), varies in geological timescales. At greater depths, temperatures age from young to very old (4 billion years), and also rise and the rocks become soft and deformable (Ques- varies greatly in composition. The average composition tions 4 and 6). As a consequence, most plates extend to is andesitic, which is a volcanic rock with about 58 per- a depth of only about 50 to 200 km below the surface. cent SiO2, but locally the composition varies from less The relative strength of the plates allows them to move than 40 percent to greater than 70 percent SiO2, with without significant internal deformation. The motion the upper crust being much more silica rich than the of all points on any rigid plate can be fully described by lower crust. In general, rock with higher SiO2 content only two pieces of information: the location of a âpoleâ is less dense, melts at a lower temperature, and is more about which the plate rotates and the rate of rotation deformable than rock with lower SiO2 content. Thirty about the pole; this property of rigid plates gives plate to forty percent of the radioactive heat-producing ele- tectonics its simplicity and mathematical elegance. ments are concentrated in the continental crust, and as Complexities in the plate model arise from dif- a consequence the deep parts of the continental crust
52 ORIGIN AND EVOLUTION OF EARTH FIGURE 2.12â Map of the age of the ocean floor, with age in million years before pres- ent (Ma). Solid black lines are midocean ridges. SOURCE: MÃ¼ller et al. (1997). Copy- right 1997 American Geophysical Union. Reproduced with permission. are significantly hotter and more deformable than rocks of the oceanic crust, such as why the oceanic crust be- at a comparable depth under the oceans. comes older, and the ocean deeper, with distance from The greater thickness, lower density, and more midocean ridges (Figures 2.11a and 2.12). This age- deformable character of continental crust cause it to depth correlation is almost entirely explained by the behave differently than oceanic crust. The different aging and cooling of the plate as it moves away from behaviors of oceanic and continental crust influence the the ridge. The oceanic crust is relatively young because nature of plate boundaries. Boundaries that are within it sinks back into the mantle via subduction zones. oceanic crust tend to be narrow, except in cases where Plate tectonics also accounts for âcontinental driftâ and the relative motion between plates is very slow (Royer allows us to reconstruct where continents were in the and Gordon, 1997; Zatman et al., 2001). Boundaries past and where they will go in the future. But it still that are between continents tend to be broad because leaves us with significant puzzles about fundamental the continental crust is more deformable and much large-scale features of Earthâs crust: the occurrence of more difficult to subduct, although there is a large hot spots (Question 4), the existence and durability of variation of deformation styles within the continental continental crust, and the complex structure of large crust. Similarly, boundaries that juxtapose oceanic and mountain ranges where continents collide. It also leaves continental crust exhibit a wide range of deformation open the question of why some areas have suffered styles, from wide to narrow, sometimes changing from broadly distributed deformation (e.g., in the Basin and one to the other over time. Approximately 10 to 15 Range Province the distance between what is now Salt percent of Earthâs surface is made of broadly deforming Lake City and the West Coast has doubled in about 30 regions, while the rest is comprised of the rigid plates million years), rather than behaving rigidly as is com- characteristic of the plate tectonics model. mon of plate interiors. The plate tectonics model provides a coherent and simple explanation for many important features of Why Plate Tectonics? Earthâs surface that are not predicted simply by mantle convection. Perhaps its most elegant achievement is Plate tectonics is a kinematic notionâa description of to explain the relative youth and other major features how things move. Although thermal convection in the
EARTHâS INTERIOR 53 mantle (Question 4) has long been recognized as the offer a clue about why Earth is the only terrestrial ultimate driving mechanism (Turcotte and Oxburgh, planet with anything that even remotely resembles plate 1967; Richter, 1973), an outstanding question is why tectonics. It is well documented that trace amounts of the plate tectonics style of thermal convection extends water within minerals greatly reduce rock strength and to the surface and generates the observed plate move- also seem to reduce the ability of faults to resist slippage ment. It was once thought that the plates are pushed (Questions 6 and 9). This helps explain how the fric- apart by convective stresses at the ridge crest. How- tional resistance between plates is overcome in Earthâs ever, the notable feature of the plate tectonics style of outer 10 to 30 km. In the deeper Earth, where rocks convection is that the rocks at the surface take part by deform ductilely, relatively low mantle viscosities have plunging back into the mantle at subduction zones. The been inferred from seismic waves just below the base of descent of cold, dense oceanic crust into the mantle the plates (Richards et al., 2001). The low viscosity, or at subduction zones is responsible for most plate mo- tendency to flow, probably results from waterâs ability to tions; the subducting rock drags the rest of the plate both weaken minerals and lower the melting tempera- along with it. Additional small contributions to plate ture of mantle rocks. Computer models can produce movement come from topography on the base of the a convection pattern that looks more or less like plate plates, but very little comes directly from upwelling of tectonics if this low-viscosity zone exists in the upper mantle beneath the midocean ridges. The involvement mantle and if the plate (the near-surface rock layer) of a buoyant surface boundary layer distinguishes plate behaves as if it were perfectly plastic. But this model tectonics from other forms of planetary convection. still requires that the frictional strength along the Planetary convection would be easier to understand and boundaries of the plates be several orders of magnitude model if it took place beneath a rigid unbroken surface smaller than that measured on rocks in laboratory ex- layer, as seems to be the case for Venus and Mars. periments, a difference that could be partly explained by It is easier to understand how plate tectonics can the presence of water. Ultimately, uncertainties about persist once it has started than to explain why it ex- the origin of plate tectonics may boil down to better ists, why it exists on Earth but not on other terrestrial understanding of rock strength on large scales, which planets, how and when it started on Earth, whether it is still poorly known for plate-sized rock bodies and for has always been operating or has stopped and restarted the ultraslow rates at which plates deform. at times, and whether it might eventually come to a halt. Plate tectonics would likely be difficult to start, When Did Plate Tectonics Begin? because in order to subduct surface rocks, the cold and strong surface rock layer needs to be broken, the plate It might help to understand the origin of plate tecton- needs to bend downward, and once moving downward ics if we could use the geological record to determine it must overcome the friction along the boundary with when it became the dominant style of convection. This the neighboring plate. All three processes require huge information would be useful because we know that amounts of energy. conditions in the early Earth were different from mod- The early Earth was much hotter than todayâs ern conditions (Questions 2 and 4), and hence plate Earth, providing more thermal energy to drive mantle tectonics may have started only after Earth had cooled convection. However, the higher temperatures within to a certain degree from its initial state. Most models the mantle would have led to thinner tectonic plates, indicate that the heat flux from Earth was higher by a and perhaps also thicker oceanic crust, making the factor of 3 or more 4 billion years ago and even higher plates of the early Earth more buoyant than they are 4.4 billion years ago. In the present plate tectonics re- today. The net effect on plate tectonics is poorly under- gime, the heat flux is approximately proportional to the stood, but it is possible that the Hadean Earth, while square root of the rate of generation of new seafloor. almost certainly convecting vigorously, had a style of If a similar relationship existed before 4 billion years convection quite different from that of today. ago, then either the speed of plate motions was at least Earth is distinguished from other planets by the 10 times faster than today or there must have been a presence of significant amounts of water, which may significantly larger number of smaller plates. However,
54 ORIGIN AND EVOLUTION OF EARTH increased heat loss through faster plate creation is po- one side of the terrane to the other. Spreading ridges tentially problematic in that the plates might not have can be destroyed through subduction, as exemplified had sufficient time to cool at the surface until they were by the subduction of the eastern extension of the dense enough to subduct. The processes responsible for Galapagos rift under South America. Along the east- the higher heat loss in the past remain an unresolved ern margin of the Pacific Ocean, at least one plate (the issue. Kula plate) and most of another (the Farallon plate), One clue about the beginning of plate tectonics along with their plate boundaries, have disappeared is that granitic rocks, whose formation is presumed to beneath North America during the past 100 million depend on the presence of water at depths of 100 km years. New subduction zones can be seen to form (see discussion below), had already formed by 4 billion where transform faults already provide deep cracks in years ago (Question 2). This is indirect evidence that the lithosphere. This appears to be happening today subduction operated in the very early Earth. However, beneath New Zealand, where a transform boundary it is also true that granitic rocks can form by means has been converted to nascent subduction beginning that do not involve subduction. For example, mantle- only about 5 million years ago. New spreading ridges plume-type volcanism would continually raise new lava apparently form and split continents, as has happened to Earthâs surface, gradually pushing the older lavas along the Red Sea, where a new region of oceanic crust down into the mantle where they would heat up, melt, has been forming between Africa and Arabia over the and produce granitic magma (Richter, 1985). past 10 million years. The same process appears to be Other evidence for the early establishment of plate occurring today along the East African Rift, where the tectonics on Earth comes from geological structures, northeastern corner of Africa is beginning to move especially folded and metamorphosed 3.5- to 3.2- eastward away from the greater African continent. billion-year-old rocks. This evidence suggests that There are many other examples of the formation of lateral compressive stresses and probably large-scale new plate boundaries and the termination of old ones. horizontal motions existed during that time, given the What we know of these examples can be deduced from old rocksâ similarity to those produced along modern geological observations, but the causal mechanisms plate boundaries. While other forces are also capable of remain elusive. producing horizontal motions and compression without the features of modern plate tectonics, the sum of avail- How Did the Continents Form? able data suggests that something like plate tectonics has been the dominant mechanism for shaping Earthâs The existence and persistence of continental crust surface since around 3.5 billion years ago and perhaps present their own set of questions that are perhaps as earlier. The scant rock record from before 3.5 billion fundamental as those of plate tectonics. Continental years makes it difficult to prove the existence of plate crust is crucial to Earth as we know it, both because it tectonics, but establishing when it started would help makes the land surface habitable and because erosion us understand the conditions needed for its existence and weathering of the continental surface play a role in and remains an important objective for both theoretical regulating Earthâs climate (Question 7). But how has and field studies of ancient terrains. the continental crust been preserved at Earthâs surface for billions of years, allowing land life to evolve as it What Causes New Plate Boundaries to Form? has? How were the continents created, and how are they likely to evolve in the future? Plate boundaries are transient features, so there must Just as water plays a central role in plate tectonics, be mechanisms to continuously create and destroy it also seems essential in âseedingâ continent formation. them. For example, subduction zones commonly be- The role of water in continent formation begins at sub- come extinct when two continents collide at a subduc- duction zones, where vigorous volcanic activity tends tion boundary. When microcontinents (or terranes) to occur. These zones produce thick lava accumulations collide with a continent, subduction can jump from called island arcs that stick up above sea level and are
EARTHâS INTERIOR 55 difficult to subduct. The location of subduction zones Other processes might also contribute to the par- may well be determined by the presence of sufficient ticular composition of the continents. In general, if water in rocks to weaken them. It seems inescapable, the raw materials for continents have basalt-like SiO2 based on our knowledge of the melting behavior of concentrations, but the preserved crust has higher SiO2, rocks, that the formation of island arc volcanoes requires then some low-SiO2 material must be returned to the water to be carried down within the minerals in the mantle. This low-SiO2 material cannot be sediment subducting slab. Furthermore, transporting especially because sediment is high in SiO2. There does exist a large amounts of water into the mantle via subduction subduction-like process, called lower crustal founder- zones can help produce the high-SiO2 magma needed ing, that can return low-SiO2 crustal rock to the mantle. to initiate continent formation. So there is a strong Seismic studies have begun to document what appear to suspicion that water begets subduction, which begets be large âdripsâ of high-density material that are slowly plate tectonics and continents. The planets that do not falling off the bottoms of continents in some areas have plate tectonicsâMercury, Mars, and Venusâhave (Calvert et al., 2000; Zandt et al., 2004), and studies of virtually no surface water and probably very little water xenoliths and lavas also provide evidence that the pro- dissolved in mantle minerals. cess has occurred (Ducea, 2002; Gao et al., 2004). The If it is true that subduction in the presence of questions that remain are how common lower crustal oceans (or water?) inevitably leads to the formation of foundering is, what stimulates it, and how long it has strips of thick volcanic crust like the modern Aleutian operated. The existence of this continental destruction Island chain and the Marianas, and if these strips of mechanism is consistent with specific variations in the crust cannot be subducted, they may be moved around strength and density of rocks in the lower continental by plates until they collide and combine to build larger crust and uppermost mantle, but the conditions leading land masses, or proto-continents. The bigger these land to its initiation must await more detailed knowledge of masses get, the more difficult they would be to subduct. rock properties (Question 6). In this way, subduction would act as a kind of filter, al- Another poorly understood contributor to conti- lowing thick and buoyant crust to remain at the surface nent formation and modification is mantle plumes. In and destroying crust that is thin and dense. oceanic regions, mantle plumes produce large patches However, this island-arc origin of continents can- of thick crust with the composition of basalt. But this not be the whole story because the continents are more dense, plume-produced crust could still be sufficiently silica rich than island arcs. We know that continents buoyant to resist subduction and therefore accrete to undergo geological processing and reprocessing that the continents. Iceland, Hawaii, and the large plateaus is too complex to be described in a one-stage model. of the western Pacific may all be examples of potential For example, any land above sea level is subject to new continental crust produced by mantle plumes. erosion and weathering. Weathering tends to leach Mantle plumes can also deposit their volcanic prod- away âmantle-typeâ elements like Mg and Ca, as well ucts directly on (or within) continental crust, thereby as others, and generally leaves Si and Al behind. The adding to the continental mass. And the heat provided sediment is transported to the ocean margins, where by mantle plume magmas entering the crust can cause some is subducted and some is plastered onto the crustal melting, uplift, and erosion and could even margin of the continent or squeezed between colliding contribute to the breakup of continents. The role of continents. Volcanism also occurs within continents, mantle plumes in the evolution of continental crust is and the lower part of the crust can itself melt and feed a fundamental unresolved issue, one that becomes more volcanoes in continental margin subduction zones like urgent and less tractable when considering the oldest the Andes and in continental collision zones like the continental crust. Whether all continental crust has Alps. So continents tend to be repeatedly modified been produced by island arc volcanism, or whether an after their initial formation, and they are also broken alternative mechanism involving wet mantle melting apart and reassembled by processes related to plate existed early in Earthâs history, remains hotly debated tectonics. but essentially unknown.
56 ORIGIN AND EVOLUTION OF EARTH How Does the Underlying Mantle Influence How Have the Continents Evolved Through Time? Continental Formation? The processes by which oceanic crust is created (at Continental plates consist not only of crust but also midocean ridges) and destroyed (at subduction zones) of relatively cold underlying mantle. The peculiarities are well established. The processes by which conti- of this subcontinental mantle provide additional clues nental crust is made are still hazy, and the processes about how continents form and persist, but we have by which continents are destroyed are even less well not yet been able to fully understand the message. In documented. Yet there is good reason to believe that many places the oldest continental plates reach about continental crust is not permanentâsimply long lived. 250 km or more in thickness, much greater than the The general questions of interest are how the volume thickest oceanic plates. Evidence suggests that the at- of the continents has changed through Earthâs history tached mantle under continents is a melt residue (it has and how the continental volume, ocean volume, thick- had magma extracted from it in the past) and that the ness of continents, and sea level are related. Corollary processes that formed it changed fundamentally about questions concern the mechanisms of production and 2.5 billion years ago. removal of continental crust and whether they have The most ancient continental cores, which formed changed with time. more than 2.5 billion years ago, are areas of prolonged Wholesale subduction of large tracts of continental stability. These regions, called cratons, are apparently crust is generally considered unlikely because of its low stronger than surrounding younger regions of the con- density and great thickness. However, the thin fringes tinents that experience periodic disruption. The mantle of continental crust (Figure 2.11b) that surround most portion of the cratons is cold and especially thick but continents are not buoyant enough to resist subduction has low density due to iron depletion ( Jordan, 1988). if they are attached to the underlying dense mantle. The iron depletion is difficult to explain unless this Calculations, observations in young mountain systems, mantle was melted extensively, implying melt initiation and the scarcity of deep continental margin rocks in at high pressures, probably due to the higher tempera- older mountain belts all suggest that subduction of ture of the Archean Earth. The fact that these thick, thinned continental crust may be common. Geolo- low-density continental âkeelsâ are found only beneath gists have also discovered rare exposures of continental crust that is older than 2.5 billion years means that they crustal rocks that were subducted to depths of at least must have stopped forming at that time (Sleep, 2005). 200 km, recrystallized, and then returned to the surface Their presence may also account for the longevity of (Figure 2.13; see Rumble et al., 2003). These so-called these old patches of continental crust. For example, if ultrahigh-pressure metamorphic rocks bear witness they were extensively melted, they would lack water, to the subduction of continental crust, but we do not making them stronger than other parts of the mantle know whether any of the crust remains in the mantle (Pollack, 1986). Alternatively, because earthquakes are (i.e., that the process effectively recycles continental confined to the continental crust and do not occur in crust). Continent-derived sediment deposited in oce- the continental mantle, the crust may be responsible anic trenches is another source of subducted continental for the strength of the continental lithosphere. In this crust. This mechanism of continent removal depends scenario the mantle keel limits heat flow into the base on erosion, and erosion is most effective in areas of high of the crust and thus strengthens it (McKenzie et al., elevation that are produced by continental collisions. 2005). In addition, despite their similar ages, the much As a result, over Earthâs lifetime the total volume more silicic composition of the overlying crust com- of recycled continental crust may be equal to or even pared to the melts extracted from the keels shows that exceed the current volume of the continents. Some they are not simple melt-residue pairs. The origin of studies suggest that the volume of continental crust is thick mantle lithosphere under the oldest continental steady, with the amount of subduction approximately regions and its role in continent preservation remain equaling the amount of new crust formed by upwell- intriguing fundamental questions. ing magma (von Huene and Scholl, 1991). However,
EARTHâS INTERIOR 57 ary, or is simply stirred into the mantle by convection. This question has profound implications for chemical cycling and layering within Earth (Question 4). How Do Climate, Tectonics, and Erosion Shape Landscapes? A recent advance in geology is the discovery that ero- sion, precipitation, and mountain building are inter- linked in unexpected ways, causing us to rethink one of the most familiar of geological processes. It has long been known that erosion modifies continents, preparing them to be subducted as sediment and enabling mass to be redistributed across Earthâs surface. The chemical FIGURE 2.13â Ultrahigh-pressure metamorphic rocks, Dabie weathering that accompanies erosion plays a major role Shan, China. SOURCE: Gray Bebout, Lehigh University. Used with permission. in regulating climate (Question 7) and affects the com- position of the continents, the oceans, the atmosphere, and the mantle. Erosion affects mainly the rock masses that protrude above sea level and is effective at reduc- such estimates are imprecise, and there is little evidence ing their elevation down to a value close to sea level. that they can be extended to the early Earth. Thus Most of the continental area has an elevation just a few the volume of the continental crust through time, the hundred meters above sea level (Figure 2.11a). volume of continental material reworked at subduc- The introduction of numerical modeling to moun- tion boundaries, and the total volume of continental tain-building studies, however, shows that mountain crust subducted remain highly uncertain. Each of these building and climatic processes are coupled. Uplift of represents a first-order issue for understanding Earthâs mountainous areas is driven by a combination of crustal chemical differentiation. A corollary question regards thickening and erosion and therefore is affected by cli- the extent to which continental material preserved mate and climate changes. In most mountain ranges, from Archean time (more than 2.5 billion years ago) rainfall is higher on one side than the other (Figure 2.14), is typical of Earthâs early continents. These remaining and hence erosion rates are not strictly correlated with bits of ancient continents could have been preserved either elevation or average slope. And because mountains by chance, but is it also possible that they have unusual influence rainfall patterns, they aid in their own destruc- properties, such as especially low radioactive heat pro- tion by focusing rainfall onto themselves. duction (Perry et al., 2006) and the factors discussed Because the region of maximum erosion in moun- above, that made them difficult to destroy? tain belts may be offset from the region of highest The fate of continental crust recycled back into the elevation, a complex pattern of mass redistribution can mantle is almost entirely unknown. At depths of about develop (Figure 2.15). In effect, erosion lowers surface 250 to 300 km, extremely dense minerals like stisho- elevation but draws the rock upward toward the surface. vite and hollandite can form, potentially rendering the By this process, precipitation patterns across a moun- metamorphosed continental rock more dense than the tain range can affect the height, width, and symmetry surrounding mantle, thereby contributing to subduc- of mountains, as well as the distribution of fault activ- tion. Little is known about the phase transitions and ity, and can even affect the lateral flow of rock deep in metamorphic reactions that might occur at these depths the crust. In other words, deformation and movement as needed laboratory work has not been done. Hence of Earthâs crust in mountain belts, long thought to be we have little insight into whether subducted conti- caused entirely by plate tectonic forces, can be heavily nental crust is returned near the surface, remains in influenced by surface processes. This understanding the upper mantle, descends to the core-mantle bound- has prompted an intense effort to correlate spatial and
58 ORIGIN AND EVOLUTION OF EARTH presents challenges, especially for threshold-dependent processes such as landsliding. New tools, especially cos- mogenic radionuclide dating and thermochronology, are now enabling us to determine the rates of processes through space and time, but others will be needed, for example to incorporate biotic effects (Question 8). The other critical component of models for moun- tain building, as well as for plate tectonics, is the rheology (deformation behavior) of rocks deep in the continental crust and in the upper mantle beneath the mountains. As Figure 2.15 implies, deep crustal rock flows laterally when pressure is decreased by erosion. The rate of flow depends on the rock properties, which in turn depend on mineralogy, temperature, pressure, FIGURE 2.14â Satellite photo of the central Himalaya and stress, and the flow rate itself. Although it is possible Tibetan plateau. The strongly asymmetric distribution of rainfall to determine the deformation behavior by laboratory is reflected in the vegetation pattern and distribution of glaciers. measurements, these measurements do not appear to The regions with dark green color, on the south side of the mountains, have the highest rainfall and also have the young- replicate deformation of most rocks under natural con- est metamorphic rocks exposed at the surface due to the rapid ditions. On average, the strength of rocks determined erosion. SOURCE: National Aeronautics and Space Administra- from laboratory measurements is much greater than the tion, <http://rapidfire.sci.gsfc.nasa.gov/gallery/?2002348-1214/ strength inferred from the study of regional geological Tibet2.A2002348.0505.1km.jpg>. systems (Question 6). This discrepancy is probably a matter of scaling, since natural systems are many orders of magnitude larger, and deform many orders of magni- temporal patterns of erosion and uplift with varying tude more slowly, than laboratory samples. Some large- patterns of precipitation (e.g., Burbank et al., 2003). scale mechanisms of deformation, like faulting, are not These studies, in turn, require the latest techniques reproducible in small-scale experimental samples. Also, of measuring erosion rates and crustal movement and fault systems within the crust may self-organize to cre- imaging the deeper parts of the continental crust and ate high fluid pressure along zones of active deforma- upper mantle. The large scale of tectonic systems has tion, further lowering the stresses needed for continued made them challenging to study, a task recently made large-scale deformation (Sleep, 2002). In such cases the easier by satellite sensors and systems like Interfero- strength of deforming rock masses is inversely related metric Synthetic Aperture Radar and the Global Po- to their spatial dimension. Although hypotheses like sitioning System. this one can qualitatively account for field observations, Critical to understanding the coupling of climate a fundamental theory for the rheology of rocks under and tectonics are empirical models that relate rainfall planetary conditions and scales awaits development and topography to the production and transport of sed- (Questions 4 and 6). iment and the erosion of bedrock. These geomorphic transport laws are still in their infancy. Experimentally Summary tested, field parameterized, and theoretically sound ex- pressions for most surface processes, especially as they Although plate tectonics theory explains many of apply to geological temporal and spatial scales, do not Earthâs surface features, fundamental questions re- yet exist. This gap in erosion process theory presents a main. There is increasing evidence that the existence great opportunity for scientific advanceâand a chal- of plate tectonics on Earth is related to the presence of lenging one because most relevant processes cannot abundant water, both at the surface and within Earthâs be easily simulated in controlled laboratory settings. interior, and that water plays a major role in the creation Furthermore, the heterogeneity of Earth materials and destruction of continents. However, there is still no
EARTHâS INTERIOR 59 FIGURE 2.15â Links among tectonics, climate, erosion, and topography at convergent plate boundaries illustrated with a hypothetical cross section of con- vergent plates. The resulting mountain range in (a) is located near the bound- ary between the plates. A finite-element numerical model (b) assumes stronger rainfall on the left (windward) side of the mountain, which leads to faster ero- sion there, and general flow of crustal rock toward the region of rapid erosion (curved black arrow). Warmer colors correspond to higher strain rates, the magenta line is the topographic surface, and the gray portion of the mesh shows the eroded mass. (c) Simplified plot of exhumation, elevation, and precipita- tion for the model. Figure modified from Willett (1999). SOURCE: Dietrich and Perron (2006). Reprinted by permission from Macmillan Publishers Ltd: Nature, copyright 2006. Figure 2.15.eps comprehensive theory that explains how plate tectonics from magma. Neither the mechanism of producing arises naturally from thermal convection. Establishing continental crust nor the process of destroying it and the criteria for plate tectonic convection is a fundamen- returning it to the mantle is well understood. Nor do we tal research goal for geologists and doing so will require know whether the continents were smaller or larger in better models for rock deformation properties and im- Earthâs past or whether the processes that produce and proved approaches to representing those properties in shape them were the same. The contribution of mantle numerical models of planetary convection. Other clues plumes to continental formation has gained particular will almost certainly come from the history of plate attention, as has the origin of the mantle roots under tectonics on Earth, studies of modern plate boundaries, the oldest parts of the continents. and comparisons with other planets. The past decade has seen a new understanding of The origin of continents can be plausibly attrib- the roles of erosion and climate in controlling the struc- uted to the existence of plate tectonics, in particular ture and shape of mountain ranges. This knowledge to the existence of subduction zones. However, the has become central to understanding the processes that apparent silica-rich composition of the continental affect continents and the changes that must be made crust indicates that the continents are not made in a to plate tectonics paradigms as applied to continental simple process like that which produces oceanic crust collisions. This search has intensified the desire to
60 ORIGIN AND EVOLUTION OF EARTH quantitatively predict the dependence of erosion rate and chemistry at the extreme conditions of planetary on other variables and the strength and deformation interiors and at the smallest scales of mineral surfaces properties of rocks in the lower continental crust and and nanoparticles. Advances at the other end of the the upper mantle. spectrum, when the scale is extremely large and/or the processes are extremely slow, will require advances QUESTION 6: HOW ARE EARTH in experiment, theory, computation, and observation. PROCESSES CONTROLLED BY MATERIAL Only the combination of all four is likely to bring PROPERTIES? progress. Geology is founded on the central insight that rocks What Minerals Comprise Planetary Interiors? can be read as a record of Earthâs history. Rocks and minerals are produced and altered by geological As noted in Questions 4 and 5, the nature of the con- processesâmelting, eruption, weathering, erosion, vection and deformation that affect Earthâs mantle and deformation, and metamorphism. Therefore, deci- crust, and hence models for plate tectonics and Earthâs phering the secrets of the rock record begins with an temperature history, depends directly on the material understanding of large-scale geological processes. The properties of rocks and minerals at the high tempera- keys to understanding these processes are the basic tures and pressures of planetary interiors. The pressure physics and chemistry of the materials that make up is 136 GPa (1.36 million atmospheres) at the base of the planet. Scientists now recognize that macroscale the mantle and 364 GPa at Earthâs center, while the behaviorsâplate tectonics, volcanism, and so onâarise temperature reaches 4000 K at the base of the mantle from the microscale composition of Earth materials and 6000 K at Earthâs center (similar to the temperature and indeed from the smallest details of their atomic at the surface of the Sun; Figure 2.17). structures. Understanding materials at this microscale is essential for comprehending Earthâs history (NRC, Phase transformations. The pressure in Earthâs interior 1987) and making reasonable predictions about how is so enormous that it alters the fundamental properties things may change in the future. of elements; for example, it can convert insulators to The high pressures and temperatures of Earthâs metals and cause magnetism to collapse (Figure 2.18). interior, the enormous size of Earth and its structures, Such changes occur because pressure compresses and the long expanse of geological time, and the vast di- distorts the electron orbitals, thereby changing the most versity of materials and properties present challenges basic properties of the materials. Changing pressures to investigation. Moreover, minerals are complicated bring about many kinds of phase transformations. The solids that generally contain not only their essential most familiar of these are melting and freezing, but chemical constituents but also trace amounts of almost many more complex phase transformations have been every element known in nature. Although we can learn identified. Structural phase transitions are also com- much about Earth from the study of pure compounds mon. The transition from graphite to diamond is well that approximate real minerals, we also know that even known, but more important for Earth processes is how minute amounts of other chemical elements can radi- mantle olivine and pyroxenes change at high pressure. cally change a mineralâs behavior. High-pressure mineral transformations, and their Fortunately, the surge of interest in understanding dependence on temperature, allow us to estimate the Earth materials at the atomic level has been accom- temperature of the deep Earth and provide constraints panied by rapid development of new tools, including on how mantle convection works. Temperatures inside new synchrotron sources that bring the ability to probe Earth can be estimated by comparing the pressure and the atomic structure of minerals and liquids (Figure temperature conditions at which mineral transforma- 2.16); high-pressure devices to simulate the distortion tions occur in the laboratory to the depths at which of atomic arrangements under huge pressures; and sudden changes in the physical properties of the mantle advanced quantum mechanical theory, which prom- and core occur (Figure 2.19). We know, for example, ises major advances in our understanding of physics that the boundary between the liquid outer core and the
EARTHâS INTERIOR 61 FIGURE 2.16â (Top) Aerial view of the storage ring at the Advanced Photon source. Such third-generation synchrotron sources have revolutionized the study of Earth materials by dramatically increasing spatial and temporal resolution of experimental measurements and allowing for the study of much smaller samples than had been possible. A similar qualitative advance is expected when the first fourth-generation synchrotron sources (X-ray-free electron lasers) come online in 2009. ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ SOURCE: <www.aps.anl.gov/About/APS_ Overview/index.html>. ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ Courtesy of Argonne National Laboratory, managed and operated by the University of Chicago, Argonne, LLC, for the U.S. Department of Energy. (Bottom) Results of a quantum mechanical computation based on density function theory, showing the predicted structure and distribution of electrons in SiO2 at high pressure. Such computational methods can provide estimates of material properties over the vast range of pressures and temperatures encountered in planetary interiors. SOURCE: Oganov et al. (2005). Reprinted with permission. Copyright 2005 by the American Physical Society. solid inner core must be at the melting temperature of or enhance the sinking of subducted slabs or change the core (Question 4), although the temperature is not the size and shape of mantle plumes as they rise. A known precisely due to uncertainty in the composition previously unknown phase transformation was recently of the core and the difficulties of exploring these high discovered at pressures well beyond those previously temperatures in the laboratory. The temperature of the probed (Murakami et al., 2004). The new transforma- most important changes of seismic wave velocity in tion, from perovskite, the main mineral structure of the the mantle, which happen at depths of about 400 and deep mantle, to a higher pressure postperovskite form, 660 km, is well constrained by laboratory studies of the occurs at the top of the Dâ³ region, an anomalous zone conversion of olivine and pyroxene to higher density above the core-mantle boundary (corresponding to minerals. These phase transformations are so drastic some 100-GPa pressure) that exhibits intriguing and that they can influence mantle convection; a phase highly variable seismological features (see Question 4), transformation that causes a large change in density some of which may be caused by the transformations. can work either for or against the thermal buoyancy that drives convection. What is the melting temperature of rocks under pressure? Although the effects of phase transitions on mantle Much of what we know about how Earthâs interior convection are generally appreciated, we still do not works is based on knowledge of the melting temperature know how the natural system actually worksâfor ex- of rock and metal, and how this temperature changes ample, the extent to which the phase transitions impede with pressure (Question 4). To expand this knowledge,
62 ORIGIN AND EVOLUTION OF EARTH FIGURE 2.17â Diamond-anvil apparatus (top). The sample is placed between two opposed diamond anvils, the tips of which range from 0.01 to 1 mm across, depending on the pressure range of interest. The vertically oriented strip is a metal gasket that prevents the sample from extruding. Diamond is ideal for high-pressure studies because is it strong, chemically inert, and transparent to most light. SOURCE: <http://www.physics. missouristate.edu/Faculty/Mayanovic/mayanovic_research_ webpage.htm>. Used with permission. A shock wave experiment can be carried out using a gun (right), magnetic drive, or laser. The projectile can produce pressures and temperatures that exceed those at Earthâs center (like a diamond anvil cell) but for very short periods of time (in contrast to static anvil experiments). New methods combine both static and dynamic approaches to reach pressure-temperature domains (Jeanloz et al., 2007). SOURCE: <www.gps.caltech.edu/~sue/TJA_LindhurstLab Website/index.html>. Used with permission. (a) HS Maj A. HS Maj O (c) LS Maj C. LS Maj O O O O O Fe Fe O O O O O O FIGURE 2.18â Influence of pressure on the iron atom. Shown is the predicted charge density of the doubly charged iron cation (Fe) Figure 2.18.eps in the mineral ferropericlase (Mg,Fe)O, in which it is surrounded by six oxygens (O). (Left) At low pressure the spins of the d electrons are maximally aligned, producing a net magnetic moment on each iron atom (called the high-spin or HS state) and the magnetic properties that we are familiar with, such as the tendency of magnetic minerals to align with the magnetic north pole. (Right) At high pressures characteristic of Earthâs deep mantle the spins pair (called the low-spin or LS state), the atomic magnetic moments vanish, and iron-bearing minerals are nonmagnetic. The figures show that the size and shape of the iron cation also change across the high-spin to low-spin transition: iron is smaller (by about 10 percent in volume) and less spherical in the low-spin state, which should produce a change in density and other physical properties of iron-bearing minerals. SOURCE: Tsuchiya et al. (2006). Reprinted with permission. Copyright 2006 by the American Physical Society.
EARTHâS INTERIOR 63 also increases with pressure but more slowly. Although there is so far only scant experimental and theoretical evidence, it suggests that magma can be denser than mantle rock deep inside Earth (Figure 2.20; Miller et al., 1991). The consequences of this for Earthâs evolu- tion would be profound. If silicate melt sinks instead of rising toward the surface, it could be stored at depth for long periods, where it would be kept hot. The geo- chemical consequences of this inverted gravitational separation could also be important, but little is known about the distribution of trace elements between solids and liquids at high pressures. Iron-rich liquid would likely exist as a separate, denser phase than Earthâs silicate fraction and sink to the center, forming the core FIGURE 2.19 Photograph looking into a diamond cell at a (Question 2). The timescale of this descent and the par- 100-micron blue single crystal of hydrous ringwoodite (ideally Mg2SiO4 composition) held in situ at 30 GPa, corresponding titioning of elements between the iron-rich and silicate to a depth of 800 km in Earth. The brown spots indicate where portions during core formation are still uncertain and the sample has been heated with a laser to a few thousand have profound implications for the chemical composi- degrees, causing a phase transformation to the assemblage MgSiO3 perovskite + MgO periclase that is thought to comprise tion of the core and the origin of the geomagnetic field most of Earthâs mantle below a depth of 660 km. SOURCE: (Question 4). Courtesy of Steven Jacobsen, Northwestern University. Used There is confirming evidence that liquid may be with permission. present in the deep mantle, especially near the core- mantle boundary. Seismologists have identified thin layers of extremely low shear wave velocity at the base we need to understand the processes that control the of the mantle, a characteristic of liquid. It has been sug- melting and freezing of rocks and minerals in the gested that this region could be made of dense, partially planetary interior. Melting of rocks involves complex solidified magma and that it could even be a remnant chemistry, because rocks are typically composed of four of the Hadean planetary magma ocean (Williams and or more mineral phases, none of which are pure. As Garnero, 1996; see Question 2). If U, Th, and K are rock melts, the composition and density of the liquid concentrated in this deep liquid, it could mean that the portion are different from those of the solid, and thus, base of the mantle produces extra heat from radioactiv- with the help of gravity, one can segregate from the ity, which would affect how we think about the core other. For example, the lava that erupts from volcanoes dynamo and about the overall chemical composition of is both less dense and compositionally different from the mantle. If mantle liquid is in contact with the liquid the parent mantle rock. Over Earthâs long history the outer core, it would also mean that chemical exchange repeated processes of melting, melt ascent due to buoy- across the boundary would be much more effective than ancy, and eruption onto the surface have completely if the mantle is solid; this would change the way we rearranged many of its chemical elements. This process think about the origin of chemical heterogeneity in the of planetary differentiation, making chemically distinct mantle (Question 4). To resolve these issues we need to domains out of a homogeneous starting material, is one know much more about the properties of silicate liquids of the most fundamental features of planetary evolution and solids at very high pressures and temperatures. Re- (Questions 2, 4, and 5). cent experimental advances, including measurements One of the more intriguing questions about melt- of liquid structure in situ at high pressure (Shen et al., ing is whether, under some conditions, magma may 2004), will work hand in hand with theoretical and be denser than the surrounding solid mantle. Magma computer modeling. Modeling of high-pressure prop- is highly compressible, so its density must increase erties (Figure 2.20), using the principles of quantum rapidly with increasing pressure. The density of solids mechanics, shows promise, although at present only a
64 ORIGIN AND EVOLUTION OF EARTH be seen in the inner core, where longitudinal seismic waves travel 3 percent faster along the rotational axis than in the equatorial plane. This difference may be due to alignment of iron crystals in the core, although the mechanism for producing the alignment is still uncer- tain (Stixrude and Brown, 1998). Understanding the origin of this alignment is likely to tell us a great deal about the dynamics at Earthâs center, the history of the core, and the origin of the geomagnetic field. FIGURE 2.20â Predicted atomic-scale structure of a model magma (MgSiO3 composition) showing that the large compress- How Much Water Is in the Solid Earth? ibility of liquids is caused by rearrangement of the structure from an open configuration near zero pressure (left) to a much more Earth is unique in the Solar System for its abundant compact and highly coordinated structure at the pressure of surface water, and most models for the early Earth the core-mantle boundary (right). Silicon-oxygen coordination polyhedra are shown in blue and magnesium ions in yellow. suggest that the source of this water was the mantle via SOURCE: Stixrude and Karki (2005). Reprinted with permission volcanic eruptions. Based on recent research, it seems from the American Association for the Advancement of Science likely that the interior continues to be a major reser- (AAAS). voir of both water and carbon dioxide (Williams and Hemley, 2001). Earth is so massive that if the mantle is only 0.03 percent water, it would hold the equivalent of small number of atoms can be modeled, which means all the water in the modern oceans. Upwelling mantle that it is not yet possible to use this approach to explore material at midocean ridges appears to contain about how trace elements behave. this much water, so at present Earthâs interior has at least one oceanâs worth of water. How much more it Can seismic waves be used to uniquely determine mantle might have and how this amount has changed over mineralogy? Material properties and seismology are in- Earthâs history are outstanding questions. terdependent in a fundamental way. Seismologists can We do not know whether Earth has always had measure the speed at which seismic waves traverse the the present amount of water at its surface, but the mantle and use this information to construct pictures of answer has implications for a variety of processes. To the deep Earth in a process analogous to a medical CAT reach the answer, we need a deeper understanding scan. At the same time, pictures of the deep mantle of where water and carbon dioxide are stored in the cannot be interpreted without information about mantle. We know of two potential reservoirs of water: mantle minerals and rocks, just as radiologists need hydrous phases, such as clays that contain predictable to know how bone and other types of tissue transmit amounts of water within their crystal structures, and X-rays. The changes in seismic wave velocity through nominally anhydrous phases, such as olivine (the most different structures in the deep Earth are smallâabout abundant mineral in the upper mantle), which include 1 percentâso the elastic properties of the minerals hydrogen as defects (Figure 2.21). Knowing more about need to be known precisely to interpret the changes. these reservoirs may frame our view of the long-term As these properties become better known, geologists evolution of the hydrosphere, including formation of hope to use seismic images to map the temperature and the oceans (Question 2). Understanding the evolution composition variations in the mantle and perhaps even of the deep hydrosphere is also central to our view of the pattern of convection. The latter is possible because mantle dynamics, since even small amounts of hydro- seismic wave velocity is dependent on direction, or gen can change the viscosity of the mantle by orders anisotropy, and can be related to flow patterns if there of magnitude and the melting temperature of rocks is sufficient knowledge of the elasticity of minerals and by hundreds of degrees (Question 4). For example, if the mechanisms by which they deform (Karato, 1998). the mantle has more water, it might convect faster and A striking example of anisotropy inside Earth may produce more volcanism, by which it loses water to the
EARTHâS INTERIOR 65 about how minerals and fluids react. Recent studies have shown that reactivity is exquisitely sensitive to the finest details of surface structure. For example, the rate of exchange with water for oxygen atoms on distinct but structurally similar sites on an aluminum hydroxide surface may vary by seven orders of magnitude (Phillips et al., 2000). In addition, a major new realization is that most of the mineral surface area in the environment may be in the form of nanophases: extremely small mineral particles, 1 to 100 nm in size, orders of magnitude too small to see with the naked eye. These very small mineral grains have dramatically different physical and chemical properties than larger ones (Banfield and Zhang, 2001). The surface energy of nanophases is so important that it can stabilize structures that do not ex- ist in bulk material (Navrotsky, 2004). These structures may have unique reactive sites, adsorptive properties, FIGURE 2.21 An example of how water might be stored in and reaction kinetics. The structures of nanophases Earthâs interior. Shown is the predicted structure of a nominally also vary depending on whether they are surrounded by anhydrous mantle mineral (stishovite, ideally SiO2) with trace amounts of hydrogen incorporated via the replacement: Si4+ = water, air, or organic ligands. Nanophases are important Al3+ + H+. Dark- and light-blue polyhedra are SiO6 and AlO6 for their role as a unique reactive surface area, and they coordination environments, respectively; red spheres are oxygen also help us understand how minerals form, since all atoms; and the green sphere is a hydrogen atom. The solubil- minerals start out as nanophases in the form of small ity of water in this mineral reaches a few percent at conditions typical of the shallow lower mantle. SOURCE: Courtesy of Lars nucleation centers (Figure 2.22). Stixrude, University of Michigan. At and near Earthâs surface, the formation and dissolution of minerals take place in the presence of microorganisms, and there is a growing awareness that surface. If the mantle loses too much water, volcanism biology plays a significant role in mediating chemical might slow down until enough water is returned to the reactions at mineral surfaces (Question 8). In addition, mantle by subduction. This type of feedback may help many minerals are formed entirely by living organisms, regulate Earthâs surface environment and the water both large and small. Limestone, for example, is almost content of the mantle (see also Question 7). entirely formed as calcium carbonate shell material by small marine organisms. Much of the modern study of mineral formation lies at the interface of biology, How Do Minerals and Fluids React? chemistry, and geology. With new analytical techniques Chemical reactions between minerals and water enable it is becoming possible to study how minerals are made the oceans and atmosphere to exchange chemicals with by organisms and to compare biological and inorganic the rocks of the crust and mantle. These chemical reac- processes. For example, it is possible that an organism tions control the mineral weathering that accompanies can produce a microenvironment that causes calcite erosion and ultimately affect the composition of sea- to be precipitated essentially by inorganic processes. water, the bioavailability of nutrients and toxins in the By altering the microenvironment, the organism can environment, and the amount of carbon dioxide in the control the particular form, and hence trace element atmosphere (Question 7). All of this chemistry occurs composition, of the mineral that is precipitated (Bentov in the microenvironment at the surfaces of minerals. and Erez, 2006). We may have much to learn about how New data about natural materials, especially about the minerals form by carefully watching how organisms microstructure of mineral surfaces, are changing ideas make them (Figures 2.22 and 2.23).
66 ORIGIN AND EVOLUTION OF EARTH FIGURE 2.22â Necklace of titania nanocrystals that have aggregated spontaneously by oriented attachment. In this mineral growth pathway, crystals no more than a few nanometers in diameter aggregate and rotate so that adjacent surfaces share the same crys- tallographic orientation. The pair of adjacent interfaces is eliminated and the pair of nanoparticles is converted to a larger single crystal. Individual atoms are visible in the lower view. SOURCE: (Top) Penn and Banfield (1999). Copyright 1999 by Elsevier Science and Technology Journals. Reproduced with permission. (Bottom) Courtesy of Lee Penn, University of Minnesota, and Jillian Banfield, University of California, Berkeley. Used with permission. Can Large-Domain, Multiscale, and Extremely to be concentrated in narrow zones rather than being Slow Earth Processes Be Predicted? widely distributed. Other feedbacks of this sort include thermal weakening and damage weakening (Bercovici Many properties and processes depend on length scale and Karato, 2002). In the latter, deformation either and timescale in ways that are difficult to predict. The reduces grain size or increases crack density, making general idea of scaling, or inferring the behavior of ma- the material easier to deform. There are many ways that terials at one scale from knowledge of those materials rocks can behave when stressed; these different defor- at another scale, underlies much of our thinking about mation processes affect one another; and the larger the Earth. For example, our understanding of mantle con- rock body under consideration, the more processes that vection is founded on our ability to relate planet-scale can come into play. Hence, predicting what will happen (large) and laboratory-scale (small) flows that have the at a large scale from information about what happens same ratio of buoyancy forces to viscous resisting forces at a small scale is a major challenge. (the Rayleigh number). Laboratory analogs are likely The behavior of faults raises many scale-related to be accurate for some aspects of mantle convection, fundamental questions: How are earthquakes (large but they have limits. For example, we know that the scale) generated and can we predict them using small- crust and uppermost mantle exhibit nonfluid behavior, scale models (Question 9)? What localized (small- or there would be no plate tectonics (Question 5). We scale) process and set of conditions trigger a (large) also know that most of the surface deformation caused fault to rupture a particular distance on a particular day? by plate tectonics takes place in narrow zones at the How much of continental deformation (large) is caused edges of the plates. The localization of deformation by slip on faults (small)? Some of the most influential probably has an origin in complex failure processes predictors of fault movement have been laboratory that are dependent on both size and timescale. Rocks measurements of rock strength: squeeze a rock in one and even magmas can exhibit a behavior called strain direction and eventually it will break or slide along softening, which means that as the amount or rate of preexisting faults, once friction is overcome. However, deformation increases, the resistance to deformation rock at the scale of a great earthquake rupture is much decreases, which increases the amount and rate of de- weaker than rock in the laboratory. One possible ex- formation further. Consequently, deformation is most planation is that water is pervasive in the crust and likely to continue wherever it has already started and weakens fault planes by acting as an easily sheared but
EARTHâS INTERIOR 67 FIGURE 2.23â Orange, polymer-laden ferric iron oxyhydroxides from a submerged mine. The slime consists of colloidal aggregates of nanoparticles, mineralized cell products, and cells (left) of two bacteria. The twisted stalks are characteristic of iron-oxidizing bacteria belonging to the Gallionella genus, while sheathed elongate cells are typical of bacteria belonging to the iron-oxidizing Lepthothrix genus. The contrast is due to iron oxyhydroxide nanoparticles. (Right) A closeup of the nanoparticle aggregates reveals that while the individual particles are separated (white regions), they have been bio-assembled so that they are crystallographically oriented in the same direction. SOURCE: Banfield et al. (2000). Reprinted with permission from AAAS. incompressible lubricant that dramatically reduces the magma migration in the mantle and formation of the friction between the two rock surfaces (Figure 2.24). core (Questions 2 and 4; Holtzman et al., 2003). The But as noted above, there are many other possible ways mantle is made of solid minerals with varying strengths. to cause Earthâs crust to appear weak in comparison to Just as in the case of magma channelization, mantle rocks in the laboratory. convection may organize weaker and stronger miner- Another reason scaling is challenging is that als into layers (foliation), dramatically influencing the Earth is heterogeneous: material properties, including viscosity as well as the seismic signal and our interpre- viscosity, electrical and thermal conductivity, chemical tation of it in terms of composition, temperature, and diffusivity, and elasticity, may vary spatially by orders flow pattern. Chemical reaction of fluids and melts with of magnitude on scales ranging from nanometers to surrounding solids can also produce channels, which kilometers. Heterogeneity may dramatically influence can significantly influence the composition of the dynamics. Cappuccino drinkers are familiar with the magma and our inferences about its origin (Spiegelman fluid dynamical oddities of composites, seen in the and Kelemen, 2003; Figure 2.25). relative stiffness of milk foam as compared with its constituents, air and milk. Analogous phenomena The importance of time. The solid-like or fluid-like are common in nature. For example, as magma forms behavior of the mantle illustrates the importance of by melting inside Earth, it juxtaposes relatively fluid time in the material properties of large domains. The magma with mineral crystals that are essentially rigid. boundary between fluid-like and solid-like behavior The viscosity of crystal mush, which largely determines is set by the Maxwell relaxation timeâthe ratio of how fast it rises (or sinks), depends strongly and non- viscosity to shear modulusâwhich is on the order of linearly on the amount of suspended solid crystals it 1,000 years for the mantle. This means that we can contains. Deformation and/or dissolution of the solid only determine the viscosity of mantle materials in the matrix through which magma moves can also organize laboratory at extremely slow rates of deformation or solid and liquid fractions so that the liquid becomes at unrealistically high temperatures to bring the Max- channelized, dramatically increasing the rate of liq- well relaxation time within the window achievable by uid-solid segregation, with important implications for experiment. Just as solids behave like fluids on long
68 ORIGIN AND EVOLUTION OF EARTH Figure 2.24 top.eps bitmap image FIGURE 2.24â Photograph (upper left) and thin section (upper right) of the Punch Bowl fault in southern California. The principal slip surface (pss) is thought to have accommodated several kilometers of slip. The slip is localized to a 1-mm (white) region, including a Figure 2.24 bottom.eps microshear zone with more intense shearing (dark) occurring within a few hundred microns. SOURCE: (Upper left) Chester and Chester type has been converted to paths (1998). Copyright 1998 Elsevier, reprinted with permission. (Upper right) Courtesy of Judith Chester, Texas A&M University. (Bottom) Results of experiments on fault slip in natural rocks showing that the friction coefficient depends on slip velocity and nearly vanishes for slip velocities similar to those of earthquakes (1 m/s). SOURCE: Di Toro et al. (2004). Reprinted by permission from Macmillan Publishers Ltd.: Nature, copyright 2004. timescales, fluids behave like solids and rupture on Summary short timescales. When magma is deformed very rap- idlyâfor example, during an eruptionâit may fracture. Understanding how Earth works depends on knowl- Understanding this behavior is helping us sort out the edge of the properties of rocks and minerals. After dynamics of volcanic eruptions (Question 9) and how a period of steady progress, breakthroughs are now these depend on features such as magma composition at hand because of new analytical tools provided by (e.g., Gonnermann and Manga, 2003). advanced radiation sources (e.g., synchrotron, neutron, and laser facilities) and advanced computing. Much of
EARTHâS INTERIOR 69 the essential physics and chemistry of Earth materials arises from structures and processes that occur at the atomic level. The new tools allow these small scales to be studied directly as well as simulated, bridging the gap between quantum mechanics and microscopy and paving the way for a new level of understanding of planetary processes at longer length scales. Earth materials present a challenge to understand- ing because of their complex chemical composition and the high pressures and temperatures of planetary inte- riors. The long timescales of geological processes also create difficulties because some of the critical processes that affect planetary evolution take place so slowly that they cannot be simulated in the laboratory and because they may be caused by mechanisms that are not impor- FIGURE 2.25 Simulation of the distribution of melt (as mea- tant or even perceptible at laboratory timescales. The sured by porosity) in a deforming, reacting matrix. The melt organizes itself into channels that vary in width, position, and physics of large domains, long timescales, and multiple melt content with time. SOURCE: Courtesy of Marc Spiegelman, interacting scales remains a major challenge in Earth Columbia University. Used with permission. See also Spiegelman science and one that will advance only with interdis- et al. (2001). ciplinary effort.