H2-rich fluids from serpentinization:
Geochemical and biotic implications
N. H. Sleep*†, A. Meibom‡, Th. Fridriksson†‡§, R. G. Coleman‡, and D. K. Bird‡
Departments of *Geophysics and‡Geological and Environmental Sciences, Stanford University, Stanford, CA 94305
Contributed by N. H. Sleep, July 21, 2004
serpentinites, composed of serpentine group minerals and varying
amounts of brucite, magnetite, and?or FeNi alloys. These minerals
buffer metamorphic fluids to extremely reducing conditions that
are capable of producing hydrogen gas. Awaruite, FeNi3, forms
early in this process when the serpentinite minerals are Fe-rich.
Olivine with the current mantle Fe?Mg ratio was oxidized during
serpentinization after the Moon-forming impact. This process
formed some of the ferric iron in the Earth’s mantle. For the rest of
Earth’s history, serpentinites covered only a small fraction of the
Earth’s surface but were an important prebiotic and biotic envi-
ronment. Extant methanogens react H2with CO2to form methane.
This is a likely habitable environment on large silicate planets. The
catalytic properties of FeNi3allow complex organic compounds to
form within serpentinite and, when mixed with atmospherically
produced complex organic matter and waters that circulated
through basalts, constitutes an attractive prebiotic substrate. Con-
versely, inorganic catalysis of methane by FeNi3 competes with
nascent and extant life.
lithospheres of early terrestrial planets. On early Earth these
rocks formed by eruption as near total melts from the mantle, by
solidification of cumulates from partial crystallization of mafic
melts at shallow depths, and by exhumation from the mantle
through impacts and tectonics. Tectonic exposure and cumulate
formation of ultramafic rocks continue on modern Earth. The
most common exposure today is along the axis of slow and very
slow spreading ridges where the mantle is too cool to form
significant basalt; ?10% of oceanic crust forms at such ridges (1).
Here, we consider thermodynamic constraints on the forma-
tion of H2-rich fluids by reaction of H2O with ultramafic rocks
at moderate temperatures and pressures. At temperatures
?300°C and low concentrations of aqueous silica, these rocks
react with H2O to form hydrous Mg and Fe2?silicates and
hydroxides (2, 3). The fluid oxidizes some of the Fe2?to
magnetite liberating gaseous and aqueous H2 to the environ-
occur on land, sometimes producing natural flames (4). Geo-
chemists report hydrogen-bearing gases trapped in spring waters
issuing from serpentinizing peridotites and ancient serpentinites
undergoing modern weathering from Oman (5, 6), the Zambales
ophiolite in the Philippines (4–8), Kansas oil wells (9, 10), the
Sakalin and Koryak Plateau in Russia (11, 12), and Milford
Sound in New Zealand (13). More recently, marine geologists
discovered low-temperature fluids from vents rich in methane
and hydrogen in midocean ridge systems (3, 14). The Lost City
vent field along the mid-Atlantic ridge system supports microbial
colonies with anaerobic thermophiles (3). Robert Rye (personal
communication) has found similar microbial colonies within
highly alkaline springs at the Cedars ultramafic complex in
California, within the Franciscan trench melange (15).
Buffering of solutions by serpentinite mineral assemblages
and production of H2, considered in this study, are of interest to
early biotic, and perhaps prebiotic, evolution on the Earth and
other rocky planets. In addition, we suggest a global geochemical
ltramafic rocks, composed mainly of olivine and pyroxene
(Mg and Fe2?silicates), are a common feature in the
implication of H2production in the formation of ferric oxide in
the Earth’s mantle.
An important class of habitable environments supported by H2
involves mixing of serpentinite-derived water with CO2-rich
water. The CO2reacts with H2to form methane by the reaction
4H2? CO23 CH4? 2H2O.
This reaction occurs abiotically in hydrothermal systems in
serpentine (3, 16) and lower-temperature environments beneath
oceans (17) and continents (18). Over geological time, meta-
morphism of carbonates and the degassing of lavas continually
recharge ocean and atmosphere with CO2.
Reaction 1 supports extant biota and abiotic processes. Meth-
anogens in rocks can thrive at H2concentrations of ?13 nM (19,
20), orders of magnitude below the concentration in equilibrium
with serpentinite presented below. Magnetite and awaruite
(FeNi3, a common trace mineral in serpentinite) both catalyze
methane production in the laboratory at hydrothermal condi-
tions (21, 22). Complex organic matter forms abiotically from
this process both in the laboratory and nature (3, 16, 18, 21, 22).
The formation of abiotic methane and more complex organic
not exist before life. Rather, abiotic methanogenesis was a mixed
blessing to nascent life forms. Abiotic formation of methane
competes with life by removing the energy source in reaction 1.
This is particularly true if awaruite or magnetite efficiently
catalyzes the reaction (22). The reaction (along with the forma-
tion of MgCO3) is also likely to occur when CO2-rich water
penetrates hot serpentine. This competition is evident in modern
subsurface environments with a slow supply of the limiting
reactant (either CO2or H2) (18). Methanogens do exist there,
but the abiotic consumption of their food reduces their produc-
outcompete life. At present, this limit is poorly constrained, but
it could be fairly high for inept prebiotic autocatalysis and early
We continue with the positive aspects of serpentinization for
life. Once life originated, methanogenesis was a continual sub-
surface niche on the Earth. This niche likely exists on Mars and
Europa (16). The use of Ni, a common element in serpentinites
and a rare element in other rocks, in the key enzyme of
methanogens points to the great antiquity of this process on
Earth and the probable origin of methanogenesis within fluids
that reacted with serpentinite (23, 24).
The H2-rich waters formed by low-CO2hydrothermal fluids
mixing with other fluids and organic substrates. Mixing pro-
duced disequilibria and gathered the full repertoire of biological
Abbreviations: QFM, quartz-fayalite-magnetite; SBM, serpentine-brucite-magnetite.
†To whom correspondence should be addressed. E-mail: email@example.com.
§Present address: Iceland GeoSurvey, Grensa ´svegur 9, 108 Reykjavik, Iceland.
© 2004 by The National Academy of Sciences of the USA
August 31, 2004 ?
vol. 101 ?
no. 35 www.pnas.org?cgi?doi?10.1073?pnas.0405289101
elements. First, the H2and the methane formed a very reduced
atmosphere. Lightning is known to produce complex organic
compounds in this situation (25). More transient reducing
atmospheres from the oxidation of metallic Fe occurred after
large asteroid impacts (26). In both cases, the presence of H2
increases the yield of complex organic compounds (27). Simi-
larly, complex organic compounds formed within the highly
reducing subsurface environments in serpentinites (16).
Below, we provide examples of gathering of elements by flow
through various serpentinite environments. Mixing may yield a
solution enriched in several biologically important elements. For
example, surface waters may contain amino acids formed from
nitrogen, whereas ground waters in basalt may contain signifi-
cant phosphorus. Finally, the strong catalytic properties of FeNi3
grains in serpentinites make them a potential site for prebiotic
chemistry. We are aware of no extant or fossil organisms that put
these surfaces to a useful purpose or that use FeNi3within their
cells. We feel that a diligent search is warranted.
To quantify our biological motivations, we compute concentra-
tions of aqueous H2in fluids in equilibrium with serpentinites
and the conditions where awaruite is stable. We use equilibrium
thermodynamic relations to illustrate implications of a well
known geologic process. We represent rocks as a mixture of
minerals of limited compositional variation and the redox state
in terms of the molality of aqueous H2. In the case of ultramafic
rocks, we consider the thermodynamic components H2O, H2,
FeO, MgO, SiO2, and NiO. We ignore sulfur and refer the reader
to the work of Frost (28). Fru ¨h-Green et al. (3) discuss similar
systems with significant carbon.
The mineralogy of ultramafic rocks is mostly olivine
[(Mg,Fe)2SiO4] and pyroxene [both orthopyroxenes (Mg,Fe)-
SiO3and clinopyroxenes Ca(Mg,Fe)Si2O6], with atomic Mg?Fe
ratios of ?9:1. Traditionally, the redox state of these geologic
systems is controlled by the quartz-fayalite-magnetite (QFM)
buffer, represented by equilibrium in the aqueous fluid for the
3Fe2SiO4? 2H2O 7 3SiO2? 2Fe3O4? 2H2(aq);
fayalite ? water 7 quartz ? magnetite ? hydrogen.
This buffer is relevant to redox conditions of ultramafic rocks
under igneous conditions and the more silicic mafic rocks (30)
even to temperatures of hydrothermal processes (31). Redox
conditions defined by reaction 2 are inappropriate at tempera-
tures ??300°C for ultramafic rocks, as the stable phases are
serpentine [(Mg,Fe)3Si2O5(OH)4], brucite [(Mg,Fe)(OH)2], and
magnetite, where the activity of aqueous silica is far below that
required to form quartz. Such hydrated metamorphic ultramafic
rocks are called serpentinites (32).
During serpentinization of ultramafic rocks Mg and Fe2?
partition between serpentine and brucite so that the chemical
potentials of MgO and FeO are the same in both phases (33, 34).
We represent appropriate redox reaction for serpentinite by
2Mg3Si2O5(OH)4? 2Fe3O4? 2H2(aq)? 4H2O,
Fe-chrysotile ? Mg-brucite 7
chrysotile ? magnetite ? hydrogen ? water,
where the stoichiometry corresponding to the Fe-chrysotile
[Fe3Si2O5(OH)4], chrysotile [Mg3Si2O5(OH)4] and Mg-brucite
[Mg(OH)2], and Fe-brucite [Fe(OH)2] denotes components of
brucite, respectively. We base our thermodynamic evaluation of
Fe2?substitution in serpentine and brucite on the mineral
compositions of geologic serpentinites (see Appendix). This
analysis demonstrates the importance of modest Fe2?substitu-
tion in these minerals on H2generation in planetary crusts.
Reaction 3 applies when there is enough Mg plus Fe(II) to
form brucite; that is, (Mg?Fe(II)) ?1.5 Si, as occurs in dunites
[nearly monomineralic olivine with (Mg?Fe(II)):Si ? 2] and
some harzburgites (rocks with olivine and minor orthopyrox-
ene). These rock types occurred on the ancient Earth (35). With
a lower ratio Mg?Fe(II)?Si than 1.5, the assemblage serpentine
plus talc is more oxidized. We consider serpentine plus brucite
as we are interested in the more reduced assemblage that
equilibrates with a higher concentration of H2. We recognize
that the assemblage serpentine-brucite-magnetite (SBM) does
form directly from olivine, and we restrict our analysis to
equilibrium conditions near open fractures.
Metallic iron was present on the early Earth during accretion,
just after the Moon-forming impact, and locally after major
asteroid impacts. Awaruite, a Ni-rich alloy with compositions
close to that of the ordered mineral Ni3Fe, occurs within modern
serpentinites (35, 36). The formation of awaruite in the presence
of magnetite is of interest here because H2is produced during
formation of magnetite from ferrous iron. We represent an
additional set of redox reactions involving metallic Ni-Fe
FeO ? H27 Fe ? H2O,
NiO ? H27 Ni ? H2O,
where Fe and Ni oxides denote thermodynamic components in
SBM (reaction 3), and Fe and Ni denote components in Ni-Fe
alloy awaruite. Reaction 4a allows us to represent equilibria
between SBM (reaction 3) and awaruite in terms of the activity
of Fe in awaruite as:
4Mg3Si2O5(OH)4? 3Fe3O4? 3Fe ? 12H2O,
Fe-chrysotile ? Mg-brucite 7
chrysotile ? magnetite ? Fe in awaruite ? water,
3Fe(OH)2? Fe3Si2O5(OH)4? 3 Mg(OH)27
2 Fe3O4? Mg3Si2O5(OH)4? 4H2O ? 2H2,aq
Fe-brucite ? Fe-chrysotile ? Mg-brucite 7
magnetite ? chrysotile ? water ? hydrogen.
These reactions provide compositional constraints among ser-
pentine and brucite in equilibrium with magnetite as a function
of the activity of Fe in awaruite and the molality of H2 (see
We compute phase equilibria by using thermodynamic data,
standard state conventions, and solid solution approximations
defined in Appendix. Our intent is to illustrate basic features of
aqueous H2concentrations and fluid saturation with H2consis-
tent with the common serpentinite mineral assemblages: (Mg-
Fe)-chrysotile, (Mg-Fe)-brucite, and magnetite, for reactions in
Sleep et al.PNAS ?
August 31, 2004 ?
vol. 101 ?
no. 35 ?
open pore space where water can potentially ascend toward the
surface of biological interest. We assume that abundant solid
phases control the equilibrium of locally small amounts of fluid,
composed mostly of H2O. At temperatures ?200°C, olivine and
orthopyroxene hydrate to serpentine and brucite at laboratory
time scales (37, 38). In rocks, the serpentine polymorph chry-
sotile grows within pore space, whereas the more common
olivine is commonly in contact with lizardite.
At Earth surface temperatures, olivine and orthopyroxene
rapidly dissolve in ground waters, making the water supersatu-
rated in serpentine and brucite that precipitate slowly (40). We
cannot simply apply reaction 3 because kinetically controlled
supersaturated reactants occur on both sides. It has not escaped
us that disequilibrium between dissolved Fe(II), water, and the
products magnetite and H2is a potential biotic reaction in this
We present diagrams that help constrain mass balance of a
biological or an industrial process based on aqueous H2 in
serpentinite. The concentration of H2per mass of fluid stays
constant as a fluid-dominated region heats or cools, unless the
fluid becomes saturated with respect to H2 gas. This lets us
represent a water-dominated part of fluid circulation, like a vein.
We assume a fluid pressure of 500 bars, which is appropriate for
high-level crustal rocks during the early Earth when much of its
water had not yet entered the mantle and was either in the ocean
or the atmosphere, and for modern reactions occurring ?2 km
deep in modern oceanic crust near ridge axes. The solid-phase
equilibria and the computed partial pressure of hydrogen are not
sensitive to pressure at temperatures below the critical point of
water, ?374°C and 220 bars.
Equilibria for the buffer reactions QFM (reaction 2) and SBM
(reaction 3) equilibria are shown in Fig. 1 as a function of
temperature and the molality of aqueous H2. The isopleth
labeled S0.1BM denotes equilibrium for reaction 3, where the
mole fraction XFe' Fe?(Mg?Fe) of Fe(II) in serpentine is 0.1
and XFein the brucite is in equilibrium with the serpentine (see
Appendix). The S0.1BM isopleth represents a ‘‘high-Fe(II)’’ sys-
tem. For comparison, the isopleth-labeled S0.025BM represents
equilibrium with ‘‘low-Fe(II)’’ substitution in serpentine and
brucite. Note the differences of aqueous hydrogen concentra-
tions required by the mineral buffers QFM and SBM and the
sensitivity of H2(aq) concentrations of SBM equilibria to Fe(II)
substitution. The curve labeled H2Saturation in Fig. 1 denotes
saturation of the fluid with H2(gas)at 500 bars total fluid pressure
(see Appendix); mineral reactions to the right of this saturation
curve are metastable. If a more reduced phase is initially present
(for example, metallic Fe), it is oxidized, producing hydrogen
gas. Gas bubbles may build up in the fluid because the equilib-
rium does not depend on the amount of bubbles present.
Reactions to the left of the solubility curve cannot saturate the
fluid with H2(gas)at a total fluid pressure of 500 bars. Finally,
note the curve S0.1BM in Fig. 1 represents the initial stage of
The atomic fraction XFeis 0.1 in serpentine and ?0.18 in brucite
(S0.1B0.18M), in total XFe, being approximately the same as in
primary ultramafic rocks, since the mode of serpentine is much
larger than that of brucite.
S0.1BM is an effective buffer in serpentinites. The addition of
hydrogen to a previously closed system equilibrium will result in
reduction of magnetite, which consumes some of the added
hydrogen and some of the magnetite. The new equilibrium has
Fluids buffered by the assemblage S0.1BM (Fig. 1) are super-
saturated with respect to H2(gas) at temperatures more than
?160°C. The buffer assemblage S0.025BM (Fig. 1) represents an
extensively oxidized serpentinite, where most of the Fe is in
magnetite and XFe? 0.025 in serpentine and XFe? ?0.05 in
brucite (S0.025B0.05M). Fluids in equilibrium with serpentine and
brucite of these compositions are undersaturated with respect to
still may proceed if the H2concentration is undersaturated with
H2(gas). We expect this situation when fluid vigorously circulates
through the rock, removing dissolved hydrogen. The hydrogen-
enriched fluid then vents to the surface and is replaced at depth
with hydrogen-poor fluid from the ocean.
Fig. 2 illustrates the SBM buffer equilibrium as a function of
temperature and H2(aq)concentrations for serpentine composi-
tions ranging from XFe0.025 to 0.1. The curve labeled 500 bars
denotes fluid saturation with respect to H2(gas)and the curves
ature at 500 bars total fluid pressure. Isopleth S0.1BM represents high-Fe
serpentinite, with XFe? 0.1 in crysotile and XFe? 0.07 in brucite, and S0.025BM
represents low-Fe serpentinite, with XFe? 0.025 in crysotile and 0.0175 in
brucite. QFM buffer and hydrogen solubility bars are shown.
Molality of H2(aq)for SBM buffer (reaction 3) as a function of temper-
for serpentinites with XFein cryostile ranging from 0.1 to 0.025. H2partial
pressure contours are thick gray lines.
Phase diagram as in Fig. 1 showing isopleths of SBM buffer equilibria
www.pnas.org?cgi?doi?10.1073?pnas.0405289101 Sleep et al.
labeled 5 and 50 bars represent isopleths of H2(gas) partial
pressures. Note in Fig. 2 that equilibrium of SBM with H2(gas)
saturated fluid requires an increase in the Fe(II) content of
serpentine and brucite with decreasing temperature. Similar
features are apparent for the cooling of fluids at constant partial
pressures of H2(gas). As an example, consider reaction at 260°C
serpentine, brucite, and water are in equilibrium for XFein ser-
pentine of 0.0625 and XFein brucite is ?0.12 (S0.065B0.12M). The
quick yield of hydrogen is that needed to oxidize the rock to this
composition. Any further oxidation would lower Fe?Mg in
serpentine and brucite, leaving the fluid undersaturated with
respect to hydrogen. Conversely, XFe in serpentine forms in
equilibrium only on the left side of the H2(gas)saturation curve.
For example, at 50-bar partial pressure of H2(gas) and 120°C
would have XFein serpentine ?0.0625.
The stability field of the Ni-Fe alloy awaruite in the presence of
serpentine, brucite, and magnetite is shown in Fig. 3, which illus-
trates the two-phase field of Ni-rich solid solutions plus awarurite
(shaded area), and the region of awaruite stability (see Appendix).
Also shown are isopleths (solid curves) of XFein serpentine for the
assemblage serpentine, brucite, magnetite, and awaruite as repre-
of partial pressures of H2(gas)consistent with reactions 5b and S3 in
Appendix. The curve labeled 500 bars denotes H2(gas)saturation in
the fluid, and the vertical dashed line denotes the composition of
stoichiometric awaruite (FeNi3). Fig. 3 is compatible with previous
magnetite-bearing assemblages replace earlier awaruite-bearing
ones as the rock becomes increasing oxidized (3). Note that there
is a minimum XFein serpentine for awaruite to be stable, which
decreases with increasing temperature, which is a weak function of
increases with temperature. For example, at 200°C, awaruite of
composition FeNi3cannot form unless the H2partial pressure is
more than ?320 bars, which would preclude awaruite formation in
excess of 50 bars H2is needed to form awaruite.
Ascent and Venting of Hydrogen-Rich Water
and off-axis locations (17), together with mineralogic and iso-
topic measurements of fossil magma-hydrothermal and regional
metamorphic terrains, demonstrate that aqueous fluids locally
flow in large quantities in crustal rocks (41–43). Metasomatic
reaction zones around veins in these rocks indicate that fluids
rapidly react with their lithologic environment.
Fig. 4 illustrates hypothetical paths of fluid-dominated phase
equilibrium with the local composition of serpentine, brucite,
and magnetite is represented by pressure-temperature point A in
with hydrogen. It continues to cool to point C, losing hydrogen
to bubbles. Finally, it ascends upward to lower pressures and
temperatures, losing more hydrogen into bubbles along the path
originally in solution at point A escapes as bubbles. Between
points A and E, the fluid is more reducing than the serpentine
from which it formed. The fluid may locally react with the rock,
consuming magnetite as represented by reaction 3. Boiling may
produce locally hydrogen-rich fluids. In general, a boiled fluid
once quenched is likely to be saturated with hydrogen. This
process is relevant to producing hydrogen seeps on land where
the total pressure is low.
Moon-Forming Impact and Ferric Iron in the Earth’s Mantle
The Earth’s Moon formed when a Mars-sized object collided
with the Venus-sized Proto-Earth. As summarized by Sleep et al.
(44), the impact left the Earth surrounded by a rock-vapor
atmosphere, which condensed in a few thousand years. The
Earth’s water and CO2formed a dense atmosphere heated from
below. Internal heating became an insignificant effect on climate
after ?2 million years. By that time, liquid oceans condensed
beneath a dense CO2atmosphere, and the surface temperature
was ?200°C. The Earth’s gravity was able to hold CO2, water,
brucite, magnetite, and awaruite as a function of XFein awaruite and tem-
perature. The two-phase region (awaruite plus Ni solid solution) is shaded.
Hydrogen partial pressures and XFein serpentine in equilibrium with
Fig. 2. The path A–F represents venting of fluid in a deep system on land that
is initially in equilibrium with serpentinite at point A. The second path (R–S–
Examples of two paths of fluid composition on phase diagram as in
Sleep et al. PNAS ?
August 31, 2004 ?
vol. 101 ?
no. 35 ?
and significant H2. In contrast, the Moon did not retain these
gases and lacks potentially habitable ground water.
Thereafter, the capacity of the Earth’s surface rocks to hold
CO2 in carbonates controlled the CO2 pressure and hence
(along with methane) climate (44–46). Rocks of the lithocap
of this magma ocean were, however, insufficient to contain the
available CO2. Rather, the Earth’s interior sequestered CO2
when the surface layer foundered, a process analogous to
modern subduction of carbonates in altered oceanic basalts.
This storage required that the foundered crust did not imme-
diately degas its CO2. Instead much of the CO2mixed into the
deep interior (47). This situation prevails now in that only a
modest fraction of subducted CO2immediately returns to the
surface at arc volcanoes.
Our concern here is the effect of the aftermath of the
Moon-forming impact on oxidation state of the Earth’s mantle.
The Earth’s mantle contains a significant amount of ferric iron
that is out of equilibrium with metallic iron in accreting bodies,
which formed the Earth’s core (e.g., ref. 48). Ferric iron probably
formed after the bulk of Earth’s metallic iron was sequestered in
the core, at least partially after the Moon-forming impact. In
addition, a ‘‘veneer’’ component of platinum group elements
exists in the Earth’s mantle. This material formed along with the
oxidation of metallic iron, much from the core of the impacting
body. Water and preexisting ferric iron in the Proto-Earth’s
mantle are the probable oxidants.
Mass balance indicates that that these processes produced
massive amounts of hydrogen gas (e.g., ref. 48). The ferric iron
requires reduction of 4.5 ? 1022mol of water (60% of the mass
of the current hydrosphere) to H2. The platinum-group elements
require 7.5 ? 1022mol of water (about the mass of the current
hydrosphere). The hydrogen gas from each source (if all in the
air at the same time) would have produced equivalent pressures
of 20 and 30 bars, respectively for a total of 50 bars. That is, an
equivalent pressure of 50 bars is a reasonable upper limit for the
amount of H2in the aftermath of the Moon-forming impact.
The relative timing of the escape of atmospheric H2to space
and the demise of CO2by forming carbonates is unknown. In
addition, much of the hydrogen production (partially that from
the oxidation of metallic iron) occurred at magmatic tempera-
tures inside the Earth or when the Earth’s surface was still too
a massive CO2atmosphere as carbon species are not included.
However, we can represent the behavior of a clement massive
hydrogen atmosphere above the liquid-water ocean.
We consider the behavior of serpentine with the Fe?2?Mg
ratio of the present mantle, by using Fig. 2. If no redox reactions
which produces hydrogen at 160°C at all reasonable H2atmo-
spheric partial pressures. At 20°C, this serpentinite is in equi-
Even in the extreme case of 50 bars of surface H2, oxidation of
serpentinite proceeds during hydrothermal circulation. For ex-
ample, fluid saturated with 50 bars of H2descends from recharge
point R (Fig. 4) into the subsurface as a closed system to point
S. It then heats up and reaches equilibrium with moderately
oxidized serpentinite at point A. The water then ascends back to
the surface, forming bubbles along the way on the path B–C–R.
More than 90% of the H2present at point A eventually forms
bubbles and escapes to the surface.
Thus significant oxidation of the mantle could have occurred
at hydrothermal conditions on the early Earth or on other bodies
where ultramafic rocks covered the surface. Conversely, the
iron is still ferrous, not 2?3 ferric, as in magnetite. Also, we still
have an ocean. There is no obvious hydrothermal buffer that
would yield the current mantle composition from a long-lived
Conclusions About the Later History of the Earth
Ultramafic rocks are uncommon on the surface of the modern
Earth. This situation prevailed soon after the Moon-forming
impact when the Earth’s interior cooled to the point where the
most common igneous rock was basalt, formed by partial melting
of the mostly solid mantle. The dominant buffer since that time
has been approximately QFM (reaction 2). Although uncom-
mon, serpentinized ultramafic rocks are important biotic envi-
ronments on the modern Earth. They are also likely to have been
a prebiotic environment on any silicate planet. These situations
are likely to resemble the modern Earth, where both low- and
high-temperature hydrothermal circulation occurs.
Figs. 2 and 3 suffice to delineate the environments where H2
may be generated, and where the minerals magnetite and FeNi3
may form. H2 forms in molar to millimolar quantities at all
temperatures even after much of the ferric iron in serpentine is
oxidized. This finding is compatible with the observation of
hydrogen in warm (40–70°C) oceanic hydrothermal vents (17),
hot hydrothermal vents (16), 2-km deep waters within the
Canadian Shield (18), and even natural flames above seeps on
land (4). Magnetite is a stable phase in these environments. In
contrast, FeNi3forms only at high XFein serpentine, and hence
high H2pressure. It is not expected to form at shallow depths or
mineral in surface samples and placers, however, indicates that
it oxidizes quite slowly at modern surface temperatures.
Appendix: Thermodynamic Calculations
We compute the phase relations depicted in Figs. 1–4 for
equilibrium of reactions 1, 3, and 5 by using equilibrium con-
stants derived from SUPCRT92 (49). Standard state conventions
are unit activity at any temperature and pressure for stoichio-
metric minerals and H2O, unit activity of a hypothetical one
molal solution at infinite dilution for aqueous species, and unit
activity of pure gases at any temperature and one bar.
We assume unit activities of quartz, magnetite, and water and
compute activities of other solid phases, assuming ideal site-
mixing solid solutions. The activity of fayalite is equal to 0.0064,
which corresponds to the mole fraction of fayalite in olivine of
mantle composition (0.08). We considered several compositions
of serpentine (Fe-Mg chrysotile) and Fe-Mg brucite; activities of
mineral components chrysotile, Fe-chrysotile, Mg-brucite, and
Fe-brucite (i.e., aFe-chrysotile ? X3Fe,chrysotile and aFe-brucite ?
XFe,brucite, and vice versa for the Mg-endmembers) are computed
consistent with the exchange reaction
3 Fe(OH)2? Mg3Si2O5(OH)47
Fe3SiO5(OH)4? 3 Mg(OH)2
Trommsdorff (50), which is based on coexisting compositions of
lizardite and brucite from serpentinites reported in the litera-
ture. The ratio of the mole fraction of Fe(II) in brucite to
serpentine solid solutions is ?2, which we assume to be inde-
pendent of temperature and pressure at ?300°C and 500 bars.
We compute the activity of Fe in FeNi3(awaruite) considering
the stability of completely disordered FeNi3phase that corre-
sponds to Fe metal with the activity of Fe equal to 0.25. We then
compute the activity of ordered awaruite by noting that the
entropy difference between the two ordering states is equal to
the configurational entropy Scon (Scon is zero in the ordered
phase and equal to 1.12 cal?mol?1?K?1per atom in the formula
unit for the disordered phase) and that the order?disorder
coefficient of H2,aqas a function of temperature was calculated
by using EQ3 (51) and the b-dot extended Debye-Huckel equa-
tion for a model sea water solution.
www.pnas.org?cgi?doi?10.1073?pnas.0405289101Sleep et al.
Thermodynamic data for H2O, H2(aq), H2(g), and minerals Download full-text
(except metallic iron and Fe-brucite) are from the slop98.dat
database (ref. 49 and http:??geopig.asu.edu?supcrt92?data?
slop98.dat). As a first-order approximation we use the thermo-
dynamic data for greenalite reported in the slop98 database for
Fe-chrysotile. We used thermodynamic data for metallic iron
enthalpy of formation for Fe-brucite from Wagman et al. (53),
which is consistent with Weast (54). Heat capacity of Fe-brucite
is estimated by assuming zero heat capacity for the reaction
2 Fe(OH)2? Mg2SiO47 2 Mg(OH)2? Fe2SiO4
using heat capacity data from Helgeson et al. (55). Molar volume
of Fe-brucite is based on density measurements of Fe(OH)2by
Kozlov and Levshov (56).
Solubility curves of H2,gin water in Figs. 1–4 denote equilib-
as a function of temperature at 30, 500 bars, etc. by using
SUPCRT92 (49). We assume that H2behaves as an ideal gas under
the pressure and temperature conditions considered. H2solu-
bility curves represent the solubility of a pure H2gas.
W. B. Evans and D. Rumble reviewed this article. This work was
supported by National Science Foundation Grants EAR-0000743 (to
N.H.S.) and EAR-0001113 (to D.K.B. and Th.F.). This work was
performed as part of a collaboration with the National Aeronautics and
Space Administration Astrobiology Institute Virtual Planetary Labora-
tory Lead Team.
1. Dick, H. J. B., Lin, J. & Schouten, H. (2003) Nature 426, 405–412.
2. Hemley, J. J., Montoya, J. W., Shaw, D. R. & Luce, R. W. (1977) Am. J. Sci.
3. Fru ¨h-Green, G. L., Connolly, J. A. D., Plas, A., Kelley, D. S. & Grobe ´ty, B.
(2004) AGU Mongr., in press.
4. Abrajano, Y. A., Sturchio, N. C., Bohlke, J. K., Lyon, G. L., Pordea, R. J. &
Stevens, C. M. (1988) Chem. Geol. 71, 211–222.
5. Barnes, I., O’Neil, J. R. & Trescases, J. J. (1978) Geochim. Cosmochim. Acta
6. Neal, C. & Stanger, G. (1983) Earth Planetary Sci. Lett. 66, 315–320.
7. Abrajano, T. A., Sturchio, N. C., Kennedy, B. M., Lyon, G. L., Muehlenbachs,
K. & Bohlke, J. K. (1990) Appl. Geochem. 5, 625–630.
8. Sturchio, N. C., Abrajano, T. A., Murdwchick, J. & Muelenbachs, K. (1989)
Tectonophysics 168, 101–107.
9. Coveney, R. M. (1971) Econ. Geol. 66, 1265–1266.
10. Coveney, R. M., Jr., Goebel, E. D., Zeller, E. J., Dreschhoff, G. A. M. &
Angino, E. E. (1987) Am. Assoc. Petroleum Geol. Bull. 71, 39–48.
11. Yurkova, R. M., Slonimskyaya, M. V., Dainiyak, B. A. & Drits, V. A. (1982)
Dokl. Akad. Nauk SSSR 263, 420–425.
12. Velinski, V. V. (1978) Geol. Geofiz. 3, 52–62.
13. Wood, B. L. (1970) New Zealand J. Geol. Geophys. 15, 88–128.
14. Kelley, D. S. & Fru ¨h-Green, G. L. (1999) J. Geophys. Res. 104, 10439–10460.
15. Barnes, I., LaMarche, V. C. & Himmelberg, G. (1967) Science 156, 830–832.
16. Holm, N. G. & Charlou, J. T. L. (2001) Earth Planet. Sci. Lett. 191, 1–8.
17. Kelley, D. S., Karson, J. A., Blackman, D. K., Fru ¨h-Green, G. L., Butterfield,
D. A., Lilley, M. D., Olson, E. J., Schrenk, M. O., Roe, K. K., Lebon, G. T., et
al. (2001) Nature 412, 145–149.
18. Sherwood Lollar, B., Westgate, T. D., Ward, J. A., Slater, G. F. & Lacrampe-
Couloume, G. (2002) Nature 416, 522–524.
19. Kral, T. A., Brink, K. M., Miller, S. L. & McCay, C. P. (1998) Origins Life Evol.
Biosphere 28, 311–319.
20. Chapelle, F. H., O’Neill, K., Bradley, P. M., Methe ´, B. A., Ciufo, S. A., Knobel,
L. L. & Lovley, D. R. (1997) Nature 1202, 312–315.
21. Berndt, M. E., Allen, D. E. & Seyfried, W. E., Jr. (1996) Geology 24, 351–354.
22. Horita, J. & Berndt, M. E. (1999) Sciences 285, 1055–1057.
23. Nisbet, E. G. & Fowler, C. M. R. (1982) in Tectonic and Biological Segmentation
of Mid-Ocean Ridges, eds. MacLeod, C. J., Tyler, P. A. & Walker, C. L.
(Geological Society, London), Spec. Pub. 188, pp. 239–251.
24. Frau ´sto da Silva, J. J. R. & Williams, R. J. P. (2001) The Biological Chemistry
of the Elements: The Inorganic Chemistry of Life (Oxford Univ. Press, New
York), 2nd Ed.
25. Miller, S. L. & Urey, H. (1959) Science 130, 245–251.
26. Kasting, J. F. (1990) Origin Life Evol. Biosphere 20, 199–231.
27. Miyakawa, S., Yamanashi, H., Kobayashi, K., Cleaves, H. J. & Miller, S. L.
(2002) Proc. Natl. Acad. Sci. USA 99, 14628–14631.
28. Frost, B. R. (1985) J. Petrol. 26, 31–63.
29. Basaltic Volcanism Study Project (1981) Basaltic Volcanism on the Terrestrial
Planets (Pergamon, New York).
30. Morse, S. A., Lindsley, D. H. & Williams, R. J. (1980) Am. J. Sci. 280, 159–170.
31. Bird, D. K., Rogers, R. D. & Manning, C. E. (1986) Meddelelser Grønland 16,
32. Coleman, R. G. (1971) Geol. Soc. Am. Bull. 82, 897–917.
33. Coleman, R. G. & Jove, C. (1993) in The Vegetation of Ultramafic (Serpentine)
Soils, ed. Reeves, R. D. (Intercept, Andover, U.K.), pp. 1–17.
34. O’Hanley, D. S. (1996) Serpentinites: Records of Tectonic and Petrological
History (Oxford Univ. Press, New York).
35. Eckstrand, O. R. (1975) Econ. Geol. 70, 183–201.
36. Railton, G. L. & Watters, W. A. (1990) New Zealand Geol. Survey Bull. 104,
37. Martin, B. & Fyfe, W. S. (1970) Chem. Geol. 6, 185–202.
38. Wegner, W. W. & Ernst, W. G. (1983) Am. J. Sci. 283, 151–180.
39. Evans, B. W. (2004). Int. Geol. Rev. 46, 479–506.
40. Nesbitt, H. W. & Bricker, O. P. (1978) Geochim. Cosmochim. Acta 42, 403–409.
41. U. S. National Research Council, Commission on Geosciences, Environment,
and Resources, Geophysics Study Committee (1990) The Role of Fluids in
Crustal Processes (Natl. Acad. Press, Washington, DC).
42. Lasaga, A. C., Rye, D. M., Lu ¨ttge, A. & Bolton, E. W. (2001) Geochim.
Cosmochim. Acta 65, 1161–1185.
43. Ingebritsen, S. E. & Manning, C. E. (2002) Proc. Natl. Acad. Sci. USA 99,
44. Sleep, N. H., Zahnle, K. & Neuhoff, P. S. (2001) Proc. Natl. Acad. Sci. USA 98,
45. Sleep, N. H. & Zahnle, K. (2001) J. Geophys. Res. 106, 1373–1399.
46. Pavlov, A. A., Kasting, J. F., Brown, L. L., Rages, K. A. & Freedman, R. (2000)
J. Geophys. Res. 105, 11981–11900.
47. Keppler, H., Wiedenbeck, M. & Shcheka, S. S. (2003) Nature 424, 414–416.
48. Kuramoto, K. & Matsui, T. (1996) J. Geophys. Res. 101, 14909–14932.
49. Johnson, J. W., Oelkers, E. H. & Helgeson, H. C. (1992) Comput. Geosci. 18,
50. Evans, B. W. & Trommsdorff, V. (1972) Schweiz. Mineral. Petrogr. Mitte. 52,
51. Wolery, T. J. (1991) EQ3NR, A Computer Program for Geochemical Aqueous
Documentation (Lawrence Livermore National Laboratory, Livermore, CA),
52. Robie, R. A. & Hemingway, B. S. (1995) Thermodynamic Properties of Minerals
and Related Substances at 298.15 K and 1 Bar (105 Pascals) Pressure and at
Higher Temperatures (U.S. Geological Survey, Washington, DC), Bulletin
53. Wagman, D. D., Evans, W. H., Parker, V. B., Schumm, R. H., Halow, I., Bailey,
S. M., Churney, K. L. & Nuttall, R. L. (1982) J. Phys. Chem. Ref. Data 11, Suppl.
54. Weast, R. C., ed. (1984) CRC Handbook of Chemistry and Physics (CRC, Boca
55. Helgeson, H. C., Delany, J. M., Nesbitt, H. W. & Bird, D. K. (1978) Am. J. Sci.
56. Kozlov, T. & Levshov, P. P. (1962) Am. Mineral. 47, 1218–1218 (Russian
Sleep et al.PNAS ?
August 31, 2004 ?
vol. 101 ?
no. 35 ?