26.1. EARTH'S OCEAN AND THE FORMATION OF THE SOLAR SYSTEM
 The problems of the origin and evolution of the ocean and the atmosphere cannot be resolved realistically without referring to the processes by which the Earth itself formed. The observational data from lunar and planetary exploration do not support the previously common but vague notion that the Earth had somehow already formed when differentiation took place and the ocean and atmosphere began to develop. On the contrary, the processes leading to the formation of the Earth must themselves play a decisive role in producing differentiation (secs. 12.12-12.13) and in giving rise to the precursors for the present ocean and atmosphere. The present properties of the ocean-atmosphere system furthermore place boundary conditions on the accretion history of the planet. They contribute to the implausibility of the planetary evolution, particularly the instantaneous formation of the planets, that follows from the Laplacian type of concept of solar-system formation. The major objections against such concepts, however, come from the modern knowledge of the behavior of matter in space (ch. 1).
In accordance with the models developed in the preceding chapters and with modern knowledge of plasma physics and hydromagnetics, we conclude that when the formation of our solar system began, neutral gas in the circumsolar region fell in toward the Sun and was ionized upon reaching the critical velocity for ionization. The same processes occurred around the magnetized protoplanets (Jupiter, Saturn, Uranus, and probably also Neptune and Earth) in the later stages of their formation. The plasma revolving around the Sun provided the source or the capturing medium (ch. 19) for the material that, in the form of small particles, aggregated to larger bodies which ultimately gave rise to the planets (chs. 12,17, and 18).
 26.2. THE REMOTE PRECURSOR STAGES
26.2.1. Occlusion of Volatiles in Solid Condensates
Vapor-grown crystals are abundant components of certain types of meteorites which presumably form by the processes discussed in chs. 6 and 22 (also see fig. 7.7.1). This meteoritic material has chemical features indicative of the conditions of growth. Among these is the occurrence in some types of crystals of volatile components such as noble-gas atoms and halogen and hydroxyl ions. Because the inert gas atoms do not develop strong chemical bonds with the host structure, they are particularly useful for studying modes of incorporation.
The noble gas fraction which is of particular interest to the problems of the Earth is observed to be strongly bound in the interior of the crystals and to require high activation energies for release when the meteorite material is heated for analysis. This indicates that the gas was incorporated in the crystals during growth from the vapor phase. In most crystal structures in meteorites the packing density is high and hence solid solubilities of inert gas atoms are virtually nil. The comparatively high concentrations of occluded noble gases must therefore be achieved by their incorporation in dislocations and other growth imperfections.
Besides the presumably growth-occluded component, meteorites also contain surface implanted and radiogenic noble-gas components which have distinct, characteristic signatures (Signer and Suess, 1963); these need not be further discussed here.
The fact that the occluded noble gases are strongly bound internally in the crystals shows that incorporation took place as a part of the crystallization process and not as a surface adsorption or other low-energy processes occurring after formation of the grains as is sometimes suggested. Furthermore, it is well known from experiments that for noble-gas occlusion to be significant at crystal growth the temperature of the crystals has to be below the range 400-600K. The vapor phase temperature, however, must have been considerably higher. This follows from fundamental considerations of radiation from grain-gas systems in space (see, e.g., secs. 1.4 and 22.1; Lindblad, 1935; Lehnert, 1970a; Arrhenius and De, 1973).
Furthermore, as emphasized in secs. 1.4 and 15.6, any gas cloud in space with the dimensions visualized for a solar nebula must even at low temperatures be controlled by magnetohydrodynamic processes, and hence generate strong fields and electric currents and display a substantial degree of ionization. Therefore when considering the condensation and growth of solids in a primordial nebula we are concerned with a thermal state that must be common in gas-solid systems in space and where crystallizing grains at comparatively low temperature are immersed in, and exchange  matter with, a hot, optically thin, partially ionized gas. This state is manifest in a wide variety of phenomena active in the solar system today or recorded during the early stage of formation. These phenomena are discussed in context throughout this work.
26.2.2. Primordial Grains As Carriers of Atmospheric and Oceanic Components
The composition of the occluded noble-gas component in primordial condensates should be compared to the composition of the atmosphere of the Earth and the formation of its ocean. Measurements on meteorites show that this component characteristically has a relative abundance distribution of primordial noble-gas species which is rather similar to that of the Earth's atmosphere (Signer and Suess, 1963). In contrast, the noble gas isotopic abundances derived by interpolation between isotopic abundances of neighboring elements in the periodic table (Suess and Urey, 1956) give an entirely different distribution with a much higher abundance of light noble gases.
These facts suggest that the special noble-gas composition as found in meteorites and in the terrestrial atmosphere was established in the plasma from which the primordial condensates grew; in the former case, in the region of space where the parent materials of meteorites formed, and, in the latter case, in the region where the parent materials of the Earth condensed. Several mechanisms may have contributed to the observed noble-gas fractionation in the circumsolar region (see review in Arrhenius, 1972).
The Earth would then have acquired its atmosphere and ocean as it grew from primordial grains and aggregates similar to, but not necessarily identifiable with, those found in meteorites. The release of the volatiles, mainly during the accretion process, would form a primordial atmosphere from which the present one has gradually developed.
Although the important discovery of the "planetary" component of noble gases in meteorites was made over a decade ago, the full implications of a genetic relationship were not realized until recently (Wasson, 1969; Fanale,1971). Fanale aptly ascribes this delay to a climate of opinion which for a long time fostered a belief that the primordial atmosphere of the Earth must have been entirely removed by some ad hoc process. The present atmosphere would under these circumstances have evolved entirely by degassing of the interior of the planet, which somehow would have retained a sufficient mass of volatile components.
As demonstrated by Fanale, this is unlikely to have been the case; the primordial noble gases, with the possible exception of xenon, must at accretion by Earth have been largely transferred to the atmosphere where they  still reside. They are not even noticeable as a group in the present gas flux from the Earth's interior, where the noble-gas component is dominated by radiogenic species; nor has a noble-gas group with these element proportions yet been found occluded in igneous rocks. Other chemically reactive volatiles show a complex partition between the atmosphere and the solid Earth as discussed below.
26.2.3. Extraterrestrial Sources of Water
In view of the small mass of the hydrosphere compared to the mantle (1: 3000), concentrations as small as 300 parts per million of available hydroxyl in the accreting silicates that formed the Earth are sufficient to generate the total mass of the hydrosphere. Thus the material in meteorites fallen on the Earth and on the Moon (Gibson and Moore, 1973; Apollo 16 PET, 1973) would provide ample sources for both the ocean and the atmosphere; they have a content of hydroxyl and water ranging from a few hundred ppm to several percent.
The component of primordial solids of major importance as a source for terrestrial water is hydroxyl ion. This ion forms a regular structural component in magnesium and iron hydroxysilicates, which form the major mass of carbonaceous chondrites of Type I (Wiik, 1956). (Crystal hydrates of magnesium and sodium sulfates found in carbonaceous chondrites are probably not generated in space where they are unstable; they are likely to be forming by reaction with water vapor in terrestrial museums.)
It was previously believed (solely on the basis of geological intuition) that the hydroxysilicates in meteorites must be understood as a secondary reaction product between anhydrous silicates and water in vapor form or even as liquid water in rivers and swamps on a planet from which the sediments would subsequently have been removed as meteorites when the planet exploded. Apart from the prohibitive physical difficulties that meet such exploded-planet theories (sec. 22.2), it is now known from experiment (Meyer, 1969, 1971) that magnesium hydroxysilicates, analogous to those in meteorites, can crystallize directly at grain temperatures below about 500K from plasmas containing magnesium, silicon, hydrogen, and oxygen species. Furthermore, minor substitution with hydroxyl also occurs in terrestrial silicates common in space, such as olivine and pyroxene (Martin and Donnay, 1972). Such partial hydroxylation is also likely to occur during the growth of these silicates in free space, particularly in vapor crystallization at high relative pressure of atomic and ionic species of oxygen and hydrogen.
The fact that meteorite materials carry sufficient hydroxyl to account for the entire hydrosphere on Earth should not be taken to mean that the Earth formed from any of these specific materials, which probably represent different condensation events and regions in space. But the observations  imply that primordial condensates in different parts of the solar system, although varying markedly in chemical composition (ch. 20), have incorporated substantial amounts of volatiles, which were subsequently released in the accretional hot-spot front during the formation of the planets (sec. 26.3.2).
26.2.4. Reservoir of Inert and Reactive Volatiles
An important related question concerns the chemical composition of the Earth's total store of primordial volatiles, determined by the average composition of the planetesimals from which the Earth was built and modified by the loss processes discussed below. In the case of the primordial noble gases thus accreted, the observations mentioned above indicate relative elemental and isotopic proportions similar to those found in the occluded noble-gas component in meteorites.
In contrast, the content and proportions of reactive volatiles in the Earth's source material (primarily species of H, C, N, O, S, and the halogens) are obscured by the fact that it is totally unknown how much of these elements is hidden in the Earth's interior. Analyses of crustal rocks and extrusions from the upper mantle are not informative on this point since they are likely to be contaminated by the oceanic and atmospheric reservoirs. Extraterrestrial materials do not, at the present state of knowledge, provide much quantitative guidance on this point either, since their absolute and relative contents of reactive volatiles are extremely variable (Bogard et al., 1973; Gibson and Johnson, 1971, 1972; Collins et al., 1974).
26.3. THE IMMEDIATE PRECURSOR STAGES
26.3.1. Evolution of the Earth's Precursor Planetesimals
As shown above, we can, with some assurance, trace the Earth's ocean and atmosphere back in time to the plasma phase which preceded the formation of solid grains in circumsolar and transplanetary space. The evolutionary stages of grain formation in eccentric Kepler orbits around the magnetized gravitating central body have been discussed in chs. 16 -18, 21, and 23. Once jet streams have formed (ch. 6), accretional processes can become active (chs. 7 and 12).
In the case of the Earth the runaway accretion of the protoplanet and the exhaustion of the parent jet stream at time tc occurred very early during formation of the solar system; according to sec. 12.8, 3.5 X 107 yr after the onset of condensation in the terrestrial region of space. The mass present at that time sufficed to give rise to a protoplanet with about  half Earth's present radius (fig. 26.3.1). During the remaining part of the time period of infall of gas, assumed to last approximately 3 X 108 yr, growth was maintained at a low and steady rate, determined by the rate of injection of newly condensed material into the jet stream and hence by the rate of inflow of gas into the B cloud (sec. 21.11.1). At the end of the infall time tinf the jet stream was rapidly exhausted and the accretion of the planet terminated, as shown in fig. 26.3.1.
26.3.2. Temperature Distribution in the Growing Protoplanet
When an impacting planetesimal is brought to rest on the surface of the embryo its kinetic energy is almost entirely converted to heat energy, part of it locally and part of it in other regions of the embryo. The discussion in secs. 12.10 12.11 established that the temperature profile of a growing protoplanet is a function of the number and mass of impacting planetesimals, which reach a maximum during runaway accretion. We concluded, therefore, that the inner core of the Earth accreted cold, the accretion temperature rose to a maximum when the outer core formed, and the accretion temperature then fell abruptly and remained low (averaged over the entire surface of the Earth) during the accretion of the mantle, as depicted in fig. 26.3.1.
It is tempting to see in this primeval heat distribution of the Earth an explanation of the fact that, in its present state, our planet is known to have a solid inner core and mantle and a liquid outer core. Acceptance of this explanation requires that since the formative era the heat distribution has not changed very much due to thermal conduction. Further, radioactive heating would add another component to the heat profile in a manner depending on the largely unknown distribution of uranium, thorium, and potassium.
26.3.3. The Core of the Earth
It should be noted that the above interpretation of the Earth's internal structure presupposes that the core of the Earth is a primary feature. Still, 10 years ago there was no compelling evidence against the ingenious and widely accepted suggestion by Elsasser (1963) that the Earth's core formed at a relatively late time in geological history when radioactive heating of an originally homogeneous Earth had proceeded far enough to cause melting of iron (or iron sulfide) in an outer zone of the planet. Gravitational settling of the molten metal toward the center of the planet would release large amounts of gravitational energy and lead to a thermal runaway process, completely melting the Earth.
The following observations place such a development in doubt:
(1) Preserved crustal segments have been found to extend as far back in time as 3.7 Gyr (Black et al., 1971). This is difficult to reconcile with the necessary rate of cooling, particularly if the total store of volatiles, with the exception of a small fraction in solution in the melt, was transferred into a thick insulating atmosphere. The example from Venus further suggests that such a situation may be irreversible.
(2) Rocks derived from the upper mantle characteristically have high concentrations of nickel and platinum; nickel concentrations are mostly of the order of 10-3. It has been pointed out by Ringwood (1966) that the concentrations of these noble metals in the silicates would be brought to  much lower levels if they had been in contact with and approached equilibrium with molten iron or iron sulfide. That such extraction of nickel and platinum into the metallic phase actually takes place under similar kinetic conditions is indicated by the composition of lunar rocks, where most metallic iron and iron sulfides from the source planetesimals have been drained away from the surface layer in the accretional front of hot spots. As a result lunar silicates have nickel and platinum contents which are an order of magnitude lower thin their counterparts in terrestrial mantle rocks.
To satisfy the need for a core formed concurrently with, rather than subsequent to, the formation of the Earth, we need to assume either that the material accumulating in the region of the terrestrial planets during the first approximately 4 X 107 yr (0 < t tinf; fig. 26.3.1) was particularly rich in iron or that the core, as suggested by Ramsey (1948, 1949), consists of a compressed metallic material with chemical composition similar to that of the mantle. These alternatives are discussed in sec. 20.5.
26.3.4. Heat Release and Volatilization of Water During Accretion
Sections 12.6-12.9 have treated the mass and time relationships for accretion of planets in detail. The heating of the accreted material, carrying in it the volatile sources of the ocean and the atmosphere, is of crucial importance for fractionation of the volatiles and their ultimate disposition. The major amount of heat in the accretion process derived from the conversion into thermal energy of the kinetic energy of the impacting bodies (planetesimals).
When a planetesimal hits an embryo (protoplanet), its impact velocity is
where ves is the escape velocity for the embryo and u is the original velocity of the planetesimal relative to the embryo. In the later stage of accretion, u becomes small compared to ves. Hence, the amount of kinetic energy released at each impact is slightly above 1/2mves2, where m is the mass of the impacting planetesimal. A fraction of this energy will be converted to thermal energy of fusion within the planetesimal, melting the mass-fraction ,, given by
 L being the latent heat of fusion for the projectile material. If we take iron-magnesium silicates to be representative of the solid material in the planetesimals, the latent heat of fusion (Fe2SiO4: 295 J/g, MgSiO3: 616 J/g, Mg2SiO4: 455 J/g) may be taken to be of the order of 500 J/g.
As an example, when the embryo has grown to half the present size of the Earth, we find on putting , and L = 500 J/g that
The factor depends on the structure of the planetesimals. If these are hard solids some of the energy will be transmitted as shock waves which are dissipated at depth in the embryo (Levin, 1972). If they are fluffy aggregates a large fraction will be dissipated locally. Even if were as low as 4 percent, there is energy enough for the whole planetesimal to be melted. It is likely that the target material will be heated at the same time. Hence it is possible that a considerable fraction, if not all of the planetesimal, will be melted and/or heated to sufficiently high temperatures for the major part of its volatile components to be released in the form of gas.
The extent to which water vapor and other volatile compounds will be retained as an atmosphere around the protoplanet is determined by the balance between thermal escape of the molecules and the increasing gravitational retention by the protoplanet as its mass grows. Thus there will be a gradual accumulation of water vapor with time, and, under suitable conditions, this may condense to form liquid water.
These conditions are largely determined by the temperature of the surface of the growing protoplanet, which in turn depends on the planetesimal impact rate and the heat release at each impact. Before we proceed in sec. 26.4.3 to outline the process of the accumulation of water, we shall therefore briefly review the characteristics of the accretional heat distribution.
26.4. ACCUMULATION OF WATER DURING THE ACCRETION OF THE EARTH
26.4.1. Simple Model
The rate of increase of mass with radius for an embryo of uniform density is (see sec. 7.3)
 Let us suppose that each mass unit of impacting matter releases mass units of water. Then the rate of increase of water content in the environment of the embryo is
where MH2O is the mass of the water released.
The water vapor thus accumulated will form a part of the atmosphere around the embryo. At the top of this atmosphere the water molecules will approach a Maxwellian velocity distribution and a corresponding equilibrium temperature. The molecules which have thermal velocity in excess of the escape velocity for the embryo can escape eventually from the neighborhood of the embryo. As shown by Jeans, if the root mean square velocity of a gas is only of the order of 20 percent of the escape velocity, the gas can escape entirely in the course of a billion years or so. Hence we can make a crude model by assuming that prior to the Earth's being large enough to have an escape velocity greater than five times the thermal velocity no vapor is gravitationally retained by the embryo. Once the escape velocity equals or exceeds five times the thermal velocity, all the vapor is retained. The relevant temperature of the water vapor that will determine its rate of gravitational escape is the temperature characteristic of the atmosphere that is formed by the release of the occluded gases. This temperature is not related to the accretionally heated surface temperature of the embryo but is determined by the radiation fields of the Sun and of the plasma in the primordial magnetosphere surrounding the Earth (De, 1973).
The thermal conditions at the top of this proto-atmosphere may be comparable to those in the Earth's exosphere today, possibly having a characteristic temperature of about 1000K corresponding to a thermal velocity of about 1 km/sec for the main constituents of the atmosphere. If the escape velocity must be 5 times this we find that the Earth must have reached a size of about half its present value in order to retain the atmospheric gases and water vapor. This is about the present size of Mars and is reconcilable with the fact that Mars seems to be close to the limit where its gravitation is large enough to keep an atmosphere.
Figure 26.3.1 shows the primeval heat structure of the Earth resulting from accretion as discussed in sec. 26.3.2. The ordinate (left) for this curve is proportional to the temperature. We note that, after the low temperature accretion of the inner core, the temperature of the surface layer of the embryonic Earth continues to rise and culminates at Hence water vapor cannot condense during this period and must remain in the  atmosphere. However, the gravitational retention of water vapor at this stage is negligible. As accretion proceeds, now at a low rate determined by the injection of source material into the terrestrial region, the surface temperature of the protoplanet falls to a low average value which is probably close to the present surface temperature of the Earth. This would allow the water vapor to condense and begin the formation of a proto-ocean.
Figure 26.3.1 also shows the accumulation of water with increasing radius of the protoplanet calculated under the assumption that all the atmosphere is lost if but retained if . The total accumulation when the radius reaches the present value has been matched to equal the present ocean mass.
Meteorite materials of the type discussed in sec. 26.2.3 have sufficient hydroxyl contents to account for the present hydrosphere. Hence if the primordial grains had the same water content they would be an ample source for the present ocean.
26.4.2. Accretional Hot-Spot Front and State of Water
As was shown in secs. 12.10-12.11 above and in fig. 26.3.1 for the case of the Earth, heat delivery to the surface layer of the protoplanet first reached a maximum and then declined to a low mean value when the size of the present outer core was reached. After this culmination, the accretion of the outer regions of the Earth proceeded at a low rate, controlled by the continued injection rate of matter (assumed here to be constant) into the terrestrial region of space and terminating at the time tinf when this injection ceased. During the era between tc and tinf the average rate of heating of the surface of the protoplanet hence must have been low. At the same time, however, local heating at each individual impact site continued to be high and actually increased due to the increase of ves. The transformation of kinetic energy of the infalling bodies to thermal energy has been discussed in sec. 26.3.4. Since the major fraction of mass, and hence potential thermal energy, is concentrated in the largest embryos impacting on the growing Earth (Safronov, 1969; Ip, 1974a), it is these large projectiles that control the thermal evolution.
Assuming that the size distribution of accreting planetesimals was such as to place the major fraction of mass in bodies sufficiently large to penetrate the atmosphere and the ocean, the major fraction of heat was delivered in large impacts repeated relatively rarely at any given location (once every ten to a few hundred years in any impact area) during the era of mantle and crust formation. As pointed out in secs. 12.12-12.13, each major impact is likely to have created a deep subsurface region of molten rock which, in contrast to secondary ejecta and a thin surface crust, would cool slowly. In such melt reservoirs differentiation of magma could take place with the  heavy components sinking to the bottom and the light materials accumulating at the top. Although the average surface temperature of the Earth during this era would have remained low, each individual impact region would, in the course of time, be remelted and differentiated several times over. Radial progression of this accretional front of hot spots, discontinuous in space and time, resulted in the selective removal toward the surface of light differentiates forming the Earth's crust and of volatiles forming the atmosphere and the ocean.
The water vapor released at individual impacts after time tc, would condense and contribute to the growing proto-ocean due to the low average surface temperature during this era.
26.4.3. Details of the Model
The development discussed in secs. 26.3-26.4 above has purposely been made simplistic to reiterate in principle the energetics of growth of the planet and to illustrate the course of retention of oceanic and atmospheric components with time. There are several complicating factors, some of which can be discussed qualitatively with some certainty at the present time; for others observational basis is still lacking. Some of the resulting modifications and uncertainties are discussed in the following sections.
184.108.40.206 Atmospheric loss mechanism. In sec. 26.4, it was assumed that water vapor was lost from the exosphere by molecular evaporation during the embryonic growth stage of the planet. After achieving such a size that water molecules cannot escape the gravitational field, other mechanisms of water loss must predominate. If one assumes solar energy flux of at least the present magnitude during the major fraction of Earth's history (see sec. 25.5), water vapor in the upper atmosphere will be dissociated and form a number of species including atomic and molecular hydrogen, hydroxyl, and oxygen ions; of these the hydrogen species have a high escape rate and are preferentially lost to space. The escape rate is probably controlled by the water-vapor transfer rate from the troposphere across the stratospheric cold trap (Harteck and Jensen, 1948; Urey, 1952, 1959).
It is thus generally believed that a part of the terrestrial oxygen is the residue of water from which the hydrogen component has escaped. An estimate of the relative importance of this selective loss can be obtained from the budget shown in table 26.4.1.
The table shows that, if we make the extreme assumption that the oxygen now present in limestone derives entirely from dissociated water by reaction of such oxygen with primoridial carbon compounds, then limestone would be a major store of such oxygen. However, the limestone may partly or entirely have formed by other reactions instead; carbon dioxide may have been one of the primordial gas components of planetesimals (Gibson and....
Hydrosphere (including sediment pore water)
Excess in oxidized iron compounds
....Moore, 1973), it may have been produced by reaction of planetesimal carbon with oxygen in iron silicate in the accretional heat front (Ringwood, 1959), or carbonates could have formed by reaction of methane and water with silicates (Urey, 1952). Hence the largest conceivable loss of water by escape of hydrogen would amount to about 25 percent of the present mass of water; the actual amount is probably much smaller.
The amount of atmospheric oxygen used up by oxidation of transition element compounds, primarily those of iron, has been estimated on the basis of the extreme assumptions: (1) of an original oxygen-iron average oxidation state corresponding to FeO; and (2) that all iron in present-day sediments occurs as Fe2O3 and forms on the average 3.5 percent of shale and deep-sea sediments. The total thus obtained is only a small fraction of the oxygen in the present ocean. However, this calculation ignores the unknown amount of water-derived oxygen bound to divalent or trivalent iron in the mantle and in crustal igenous rocks (see Holland, 1964). Particularly the amount in the mantle constitutes a substantial uncertainty.
The rate of removal of gas from bodies in space is also affected by interaction with corpuscular radiation from the Sun. It is sometimes assumed that a "solar gale" arose after the planets had formed, removing all planetary atmospheres in the inner part of the solar system.
The need for such an ad hoc mechanism was rooted in the belief that the primordial components were missing from the Earth's atmosphere. As discussed in sec. 26.2, it is now realized that on the contrary our present atmosphere can only be understood as a product of the primordial accretion modified by loss of hydrogen and helium, by photochemical and biological processes, by reaction with the solid Earth, and by the radiogenic gas flux from the Earth's interior. The records from the Moon and from meteorites also have failed to give evidence of any major enhancement of solar corpuscular radiation during or after the formative era. For a discussion of the corpuscular radiation effects during this era, see secs. 16.8 and 25.5.
220.127.116.11 Effect of atmosphere and ocean on accretional heating. In principle the developing hydrosphere and atmosphere could alter the  distribution of accretional heat. The atmosphere and ocean would dissipate projectile energy by frictional heating and would decrease the radiative cooling efficiency of the collision-heated spots on the surface. The latter effect would become important if a large fraction of accumulated water were evaporated into a hot atmosphere. This is, however, not likely to have taken place since such a runaway greenhouse effect (Rasool and De Bergh, 1970) might be irreversible, whereas the geological record shows existence of sediments and organic life of Earth already at the - 3 Gyr level (Engel et al., 1968). The lack of development of a hot atmosphere can be understood since the calculated size distribution of accumulating planetesimals places the major amount of mass in large projectiles (sec. 26.4.2). This would concentrate the accretional heat in limited regions, and, with sufficient intervening time available between major impact events, efficient reradiation of surficial heat into space would take place.
At a large projectile mass/surface ratio, energy dissipation in the atmosphere and the ocean would also become small compared to the energy release after penetration to the solid surface, even in the case of objects with the assumed properties of comets (Lin, 1966).
18.104.22.168 Effect of planetesimal impact. Terrestrial experience gives little guidance concerning the nature of impact processes of the magnitude involved in planetary accretion. In the projectile mass range studied in controlled experiments on Earth with massive projectiles, the mass of ejecta exceeds that of the projectile for hypersonic impacts (sec. 7.4).
At projectile masses far beyond this range, however, the fraction of projectile material retained in the target would be expected to increase particularly at impact speeds several times the velocity of sound in the projectile material. This is indicated by the effects of the largest impacts on the lunar surface. Hence local implantation of kinetic energy converted to heat is likely to have been an important process during the accretion of the Earth.
26.5. INTRODUCTION OF WATER IN THE LITHOSPHERE
26.5.1. The Assumption of Primordial Impregnation
Crustal igneous rocks on Earth have a low but persistent content of water and occasionally very high contents of carbon dioxide (von Eckermann, 1948, 1958; Tuttle and Gittens, 1966). Because of the unknown extent of these components at greater depth in the Earth, the total store of volatiles in the solid Earth is highly uncertain. The questions of how and when these volatiles became buried are important to the problem of the formation of the ocean and atmosphere. One commonly made, intuitively based suggestion is that an excess over the present amount was somehow introduced into the  interior of the Earth during its early history. This situation would be or would become metastable, causing a net transport of water from the lithosphere to the ocean during a substantial fraction of geological time and possibly still today. No observational basis has been found for this assumption, which was originally made to secure a storage place for the present ocean and atmosphere while the original atmosphere was supposed to be destroyed. As discussed above, such a catastrophe is counterindicated by the noble-gas distribution in the atmosphere; hence the need for any such temporary ocean storage has disappeared.
To explain the present content of reactive volatiles (primarily water and carbon dioxide) in igneous rocks, Fanale (1971), on the basis of a proposal that the Earth became completely melted (Hanks and Anderson, 1969), suggested that the volatiles were partitioned in equilibrium between the melted Earth and a hot, high-pressure atmosphere in contact with it. This would seem excluded on the basis of quantitative considerations of the accretion process (ch. 12). These indicate early exhaustion of the Earth's jet stream and slow subsequent growth during the major part of the approximately 108-yr accretion period (fig. 26.3.1). Under these circumstances, the average temperature of the Earth's surface must have been low during accumulation of the mantle and the crust. The thorough outgassing of the noble gases recognized by Fanale is, as demonstrated by the late bombardment effects on the Moon, the natural consequence of local heating at each individual impact and does not in itself require or suggest simultaneous heating of the whole surface layer of the Earth.
It is furthermore doubtful that a thoroughly melted Earth would have had time to cool enough to yield a still preserved crust 0.7 Gyr after forma tion, particularly with a hot atmosphere containing a major part of the present ocean and of the carbon dioxide reservoir. Finally, the spotty occurrence of deep-seated igneous rocks rich in carbon dioxide suggests that this was introduced locally by a mechanism such as described below, rather than by equilibration of a molten Earth with a hot, massive atmosphere
26.5.2. Steady-State Impregnation and Release
There is indeed a straightforward and observationally supported way in which the igneous rocks of the crust and upper mantle would be continuous!, impregnated with reactive volatiles from the atmosphere and the ocean. The evidence for convection-driven lateral movement of large plates of the Earth's crust suggests strongly that water and carbonate containing sediments and hydrated submarine eruptives are sinking and assimilating into, the upper mantle in subduction zones, compensating for the rise of magma; and generation of new crust in the seafloor spreading zones. This vertical mixing is sufficiently fast (approximately 5 cm/year) to have drowned all  ocean sediments appreciably older than a few percent of the Earth's estimated age; so all reactive volatiles now found in igneous rocks can be understood as contamination mainly from the ocean, introduced into the solid Earth much later than the time of formation of the Earth's primordial crust.
Thus an efficient mechanism for circulation of volatiles between the ocean atmosphere system and the upper mantle has been operating through the geological eras recorded on the ocean floor and presumably during the entire history of the Earth after its formation. This does not exclude the possibility that some (probably small) fraction of the primordial volatiles was left behind in the growing lithosphere as a result of incomplete outgassing during accretion of the Earth.
26.5.3. Possible Remains of Planetesimal Volatiles in Earth's Crust and Mantle
At atmospheric pressure most gases are practically insoluble in silicate melts. However, considerable excess amounts of gas can be incorporated during shock melting of porous materials and can, at solidification, be retained in disequilibrium in such melts when they solidify due to the inefficiency of diffusion-limited removal processes (Fredriksson and De Carli, 1964). On the other hand, convection in such melts, and stripping by boiling of components such as hydrocarbons and monoxides of carbon, silicon, and potassium, contribute toward relieving such disequilibria. These retention and removal phenomena are exemplified in the lunar igneous rocks where frothing due to gas escaping from the melts is common.
Conditions in the lunar crust also indicate that, in the culminating stage of accretional heating (which on Earth probably occurred at the outer core and on the Moon near or at the present surface; see fig. 12.11.1), the removal of any water vapor possibly associated with the molten and vaporized projectile material was highly efficient, resulting in oxygen partial pressures less than 10-14 b. The sporadic occurrences of volatiles in lunar materials are considered to derive from postformative impact of volatile rich projectiles on the cold lunar surface and in some instances perhaps to be due to vapor transport through crustal fractures from the coldly accreted inner core (which could be considerably warmer today due to radioactive heating).
During the accretion of the Earth's mantle and crust, large impacts could well have implanted hydroxyl-containing material sufficiently deep so that the (pressure dependent) solubility in the melt remained comparatively high, and removal was not complete before solidification, in spite of repeated remelting by new impacts and the gravitational upward removal of light components which produced the crust. Because of the complexity of these processes and our lack of knowledge of large-scale impact effects, it is  difficult now to estimate the ultimate efficiency of material separation by the accretional hot-spot front.
A continued systematic search for primordial gas components such as radiogenic Xe129 from the Earth's crust and mantle could narrow the limits of uncertainty (Boulos and Manuel, 1971). Primordial ratios of appropriate neon and argon isotopes associated with He3 found in terrestrial materials (Clarke et al., 1969) would also serve as indicators of the possible importance of residual primordial gases.
Improved knowledge of the temperature distribution in the mantle would also contribute to the vertical transport efficiency problem since at least at moderate pressures the large cations of the elements contributing to radioactive heating are concentrated in the light component migrating toward the surface in the accretional heat front.
26.6. THE OCEAN AND THE EARTH-MOON SYSTEM
The evolution of the ocean must have been markedly affected by the fact that an abnormally massive body causing significant tidal effects exists in the vicinity of the Earth. A similar case is that of Neptune and its captured satellite Triton which has an orbit which decidedly is tidally modified (sec. 24.4) (McCord, 1966).
Tidal forces in the early evolution of the Earth-Moon system should be of considerable importance, and the question arises of the relative role of the ocean in tidal dissipation. Since dissipation in the solid Earth is considered insignificant (Munk, 1968), the ocean would provide the most important medium for tidal energy exchange.
It was believed earlier that capture of the Moon (ch. 24) must have had catastrophic tidal effects on Earth leading to complete evaporation of the ocean to form a hot atmosphere. However, the long duration of the high magnetic field immersion, indicated by the magnetization of lunar rocks in the time interval - 4 to - 3 Gyr (Strangway et al., 1972; Alfvén and Lindberg, 1974) suggests that the capture and the subsequent approach and recession of the Moon to its present orbit were associated with resonance effects (fig. 24.5.1). Such resonance effects could limit the closest approach of the Moon to distances much larger than the Roche limit.
All these questions concerning the history of the Moon need to be answered before we can have a detailed picture of the evolution of the Earth's ocean-atmosphere system.
26.7. SUMMARY AND CONCLUSIONS
(1) Physically acceptable models for accretion of planets and their source planetesimals are limited by the dynamic laws for motion of the primordial  solid condensate grains and by the boundary conditions for kinetic evolution of assemblages of particles in Kepler orbits.
(2) Analysis of the preplanetary conditions indicates a slow and cold accretion of the inner core of the Earth which temporarily changed into a rapid and hot accretion when Earth had reached approximately half of its present radius and about 10 percent of its present mass.
(3) In the subsequent phase, during which 90 percent of the Earth formed, accretion was slow and controlled by the influx of source material in the terrestrial region of space. During this period, extending over the order of 108 yr, each impacting planetesimal must have produced intense local heating, so that every part of the Earth became melted several times, but this heating was discontinuous in time and place so that the average temperature of the surface of the growing protoplanet remained low. During this period most of the gas, with the exception of hydrogen and helium, was retained.
(4) Due to the low average temperature of the Earth's surface, water vapor released in individual local impacts would during the slow, major phase of accretion condense to form a growing hydrosphere.
(5) The noble-gas composition of the present atmosphere indicates that it is directly inherited from the source planetesimals. The present atmosphere must consequently be considered as original. It differs from its primordial state only by escape of H and He, change in molecular composition due to photosynthesis, and removal of carbon into the crust, mainly as calcium carbonate.
(6) There is no need for the assumption of a solar gale removing the primitive atmosphere. Such an assumption also lacks support in the meteorite irradiation record.
(7) The observed present-day flux of volatiles from the crust into the ocean-atmosphere system must largely represent the return of volatiles which have been recycled from the ocean and atmosphere through the crust and upper mantle several times during geological history. The removal branch in this cycle is the dragging down of water and carbonate-containing sediments into the crust and upper mantle in the subduction zones, resulting from or driving the observed lateral motion of crustal plates.
(8) There is, consequently, no longer any basis for the earlier notion that the ocean and atmosphere have gradually emerged at the surface of the Earth during geological history. Instead, available evidence indicates that the ocean and the atmosphere have essentially been in place not only during the entire history of the Earth as an adult planet, but also during the major phase of accretion beginning at the stage that the proto-Earth was roughly of the size of Mars.