Hypothesis Paper Did Earthquakes Keep the Early Crust Habitable?
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
ASTROBIOLOGY
Volume 7, Number 6, 2007
© Mary Ann Liebert, Inc.
DOI: 10.1089/ast.2006.0091
Hypothesis Paper
Did Earthquakes Keep the Early Crust Habitable?
NORMAN H. SLEEP and MARK D. ZOBACK
ABSTRACT
The shallow habitable region of cratonal crust deforms with a strain rate on the order of
1019 s1. This is rapid enough that small seismic events are expected on one-kilometer spa-
tial scales and one-million-year timescales. Rock faulting has the potential to release batches
of biological substrate, such as dissolved H2, permitting transient blooms. In addition, the
steady-state deformation of the brittle crust causes numerous small faults to be permeable
enough (on the order of 1015 m2) for water to flow on a kilometer scale over relatively short
geological times (105 yr). Hence, active faults act as concentrated niches capable of episod-
ically tapping resources in the bulk volume of the rock. Radiolysis and ferrous iron are po-
tentially bases of sustainable hard-rock niches. Key Words: Methanogens—Extreme environ-
ments—Cratons—Continental tectonics. Astrobiology 7, 1023–1032.
INTRODUCTION namic requirements for life. We then examine
stress and faulting within the crust and its effect
T ERRESTRIAL CRATONS ARE REGIONS of relative
geological inactivity and provide an ana-
logue to one-plate planets such as Mars. The crust
on rationing and gathering resources. Finally, we
address flow and sequestration of hydrous flu-
ids, which is a major part of the process.
of these regions is inhabited down to the depths
of the deepest mines, currently 4.1 km. Microbes
at these depths obtain their energy from the rock. THERMODYNAMICS AND LIFE
They do not depend on the products of photo-
synthesis at the surface. Life needs to gather chemicals and energy from
The purpose of this paper is to follow a sug- its environment. It can assimilate complex organic
gestion by Sherwood Lollar et al. (2006) that compounds already present, as we do when we
episodic fracturing in the crust leads to the obtain vitamin C from lemons. It can capture pho-
episodic release of H2- and CH4-rich abiogenic tons, produce complex organic compounds, and
gas. This process parcels out an energy resource store energy for later use, as does the lemon tree.
to biota. We use geodynamics to quantify Life can also use chemical disequilibria present in
processes that maintain habitability over long ge- its environment as an energy source. Only the last
ological periods. We begin with the thermody- option is available to microbes in deep hard rocks.
Department of Geophysics, Stanford University, Stanford, California.
10231024 SLEEP AND ZOBACK
Gibbs energy requirements clusions do not depend on the precise value of
this amount.
Following Hoehler et al. (1998, 2001) and
Consortia of organisms, where the product of
Hoehler (2005), we use the methanogen reaction
one is the substrate for the next, lower the con-
as a generic example of a deep niche. Hydrogen
centration of final substrate and remaining Gibbs
and carbon dioxide react to form methane and
energy (Jackson and McInerney, 2002). This op-
water
tion is unavailable to methanogens in rocks, as
their product methane is the minimum Gibbs en-
CO2 4H2 ⇒ CH4 2H2O. (1) ergy compound that cannot serve as a substrate
in the absence of an oxidant.
This reaction provides Gibbs energy, G. To benefit,
the methanogens must couple with a Gibbs-energy
consuming reaction, such as ADP (Adenosine Rationing and sustainability
diphosphate) ⇒ ATP (Adenosine triphosphate). The total amount of energy available from a
The need to couple an energy-consuming re- block of crustal rock is limited. For the crust to
action limits habitability. To illustrate this effect, be habitable over geological time, the supply of
we normalize Reaction 1 to one mole of H2 and energy to microbes cannot be too fast or too slow.
assume that the reactants are in aqueous solution We begin with the former issue. Consider that a
(i.e., 0.25CO2 H2 0.25CH4 0.5H2O). We homogeneous region has water with abundant
vary only the activity of H2 in Reaction 1 and ex- H2 and CO2. If there are also plenty of nutrients,
press the activity of hydrogen in equilibrium with the methanogen population blooms and con-
the biological reaction: sumes the reactants in (1) until the limiting reac-
tant is depleted to the thermodynamic limit in (2).
AH2 (AH2)A exp(GB/RT) (2) If no more reactants arrive, the methanogens
eventually die off. An organism might live on
where (AH2)A is the partial pressure in equilib- their remains, but the region would become ster-
rium with the uncoupled reaction when all other ile over geological time.
reactants are held constant, GB is the extra Gibbs Neither is a very slow supply of the reactants
energy needed for the coupled reaction, R is the good for microbes. Abiotic reactions may deplete
gas constant, and T is absolute temperature. a reactant below the thermodynamic limit in (2).
The reaction of ADP ⇒ ATP requires 36 kJ Even if this does not occur, the flux of reactants
mol1 (e.g., Hoehler et al., 2001). The Gibbs en- may be too low to sustain the energy requirement
ergy associated with methanogenesis appears to of the microbes. Finally, the microbes occasion-
be small enough compared with this so that (2) ally need to have a surfeit of energy to reproduce.
poses real limits to habitability. Microbes draw Otherwise, they cannot evolve. They will die off
down H2 to the thermodynamic limit if the reac- individually when local conditions become un-
tant CO2 is present in excess. The concentration tenable. Chemical, spatial, and temporal mecha-
of the remaining substrate and the amount of re- nisms that naturally ration and episodically dole
maining Gibbs energy are variable and depend out reactants keep microbes from this fate.
on species and substrate (e.g., Jackson and Mc- Biota in ancient deep-sea sediments illustrate
Inerney, 2002). For methanogens, Hoehler et al. natural rationing mechanisms (e.g., Parkes et al.,
(2001) report 10.6 kJ mol1 per methane in Reac- 2000). Slow ingress of oxidants limits consump-
tion 1 for wild archaea, which is 2.65 kJ mol1 tion in organic-rich sediments. Microbes very
of hydrogen. In contrast, methanogens investi- slowly consume refractory organic matter where
gated in the laboratory by Kral et al. (1998) oxidants are available.
require 3 times as much energy (33 kJ mol1 Interfaces provide local lush environments for
per methane). The coupled reactions use 12 subsurface biota that do not rapidly deplete large
(Hoehler et al., 2001) and 4 (Kral et al., 1998) H2 reservoirs (e.g., Parkes et al., 2000). In the case of
per ATP, respectively. The final H2 concentration fractures in rocks, the microbes may live with a
is small but not zero. For example, hot spring significant number of cells per volume of water
methanogens investigated by Chapelle et al. but consume resources at a slow rate per volume
(2002) deplete hydrogen to 13 nmol. Our con- of bulk rock. A zone of mixing between two flu-EARLY HABITABLE CRUST 1025
ids in a fracture system is a small, but potentially The finite rate of water flow limits the rate of
bountiful, niche. consumption of ferrous iron in basalt. For exam-
ple, the H2 concentration in freshwater Icelandic
systems extrapolates to 10 M at 100°C (Stefáns-
Subsurface energy sources son and Arnórsson, 2002). As 1 km3 is 1012 L, this
In the case of deep biota discussed by Sher- concentration results in a volume water/rock ra-
wood Lollar et al. (2006), sealed cracks serve to tio of 1.4 105 to remove the potentially avail-
sequester H2-rich water, and their rupture leads able H2 from the basalt. The end-member H2 con-
to occasional bounty. Some of the hydrogen centration from serpentinite in the marine Lost
comes from radiolysis from natural radioactivity City vent is 15 mM (Kelley et al., 2005), which im-
in the rock (Lin et al., 2005). The resource is re- plies a water/rock ratio of 102.
newable over geological time.
All water-rock reactions are potentially ex-
haustible. For example, hydrogen is a product of DYNAMICS OF STABLE CRUST
the reaction of ferrous iron with water to make
magnetite We now consider geodynamics to quantify
processes that form fractures and new surface area,
3FeO (in silicate or hydroxide) H2O and the rate that water flows through them. We
begin with the long-term deformation rate. We
⇒ Fe3O4 (magnetite) H2 (aqueous). (3) then obtain the rates of seismicity and fracturing,
and the rate of groundwater flow through them.
Iron sulfide is another hydrogen source
Renewable and nonrenewable stresses and
FeS (in rock) H2S (aqueous)
geological evidence of deformation
⇒ FeS2 (pyrite) H2 (aqueous) (4) We can constrain the current rate of deforma-
tion of plate interiors from a variety of lines of
(Wachtershauser, 1988; Drobner et al., 1990; evidence. From the perspective of geodesy, pre-
Rickard, 1997; Rickard and Luther, 1997; Hoehler, cise technology has been available for only a few
2005). These reactions are both potential energy years (108 s), and the breadth of plate interiors
sources for microbes and sources of hydrogen for limits base lines to 104 km. As modern geodetic
methanogens. We do not consider more oxidized techniques allow for the detection of baseline
rocks, such as red granite, where the mineral buffer changes of 1 cm, we can resolve strain rates of
involves oxidation of magnetite to hematite. We 1017 s1. Such platewide deformation is not ob-
also consider only iron-rich mafic and ultramafic served in intraplate areas around the world (e.g.,
rocks, which in aggregate are a significant fraction Calais et al., 2006). Transient deformation associ-
of cratonal crust and the crust of Mars. ated with glacial rebound is of this order and lim-
We quantify exhaustibility by comparing the its what might be learned about long-term de-
productivity from radiolysis to that from ferrous formation from more precise studies.
iron in rocks. The flux of H2 per surface area is The geological stability of plate interiors pro-
8 106 mol m2 yr1 in the Witwatersrand vides one upper-bound constraint on intraplate
Basin of South Africa (Lin et al., 2005). There is deformation rates. From a geological perspective,
bountiful energy in hard rock if water circulation if a strain rate of 1017 s1 were to persist for
brings in reactants and removes products. For ex- 1 billion years (3 1016 s), it would thin (or
ample, there are 1.4 106 mol m2 of H2 poten- thicken) 40 km–thick crust by 12 km, which is un-
tially available from Reaction 3 from a 1 km col- acceptable in a stable region, as such variations
umn of mafic (or ultramafic) rock with 10% FeO are not observed (Mooney et al., 1998). It would
by mass and density of 3000 kg m3. This quan- also deform continents sufficiently over geologi-
tity is large compared to current average-crust ra- cal time so that the jigsaw of continents of plate
diolysis flux. For example, it would provide a flux tectonics would no longer fit together. A strain
of 300 106 mol m2 yr1 if evenly distributed rate of 1018 s1 would cause a 1.2 km thickness
over the age of Earth. change, which would still cause observable struc-1026 SLEEP AND ZOBACK
tures in some places. A strain rate less than this associated with nonrenewable stress. These
would not have obvious geological effects, given processes include deformation from the change
the stability of cratons. Hence, an upper limit on of the Earth’s radius as the interior cools, changes
intraplate strain rates is 1018 (Fig. 1). in the ellipticity from tidal despinning, and north-
We can also obtain a lower limit on strain rate south plate movement over the elliptic surface of
by considering renewable and nonrenewable Earth (Solomon, 1987). Combined, they lead to
sources of intraplate stress. Body forces cause re- strain rates of a few 1021 s1.
newable stress. In analogy, consider a weight Another way to view this is in terms of the
hanging from a viscoelastic spring. The spring changes in the local geotherm over time, which
creeps slowly over time but the weight maintains generate thermoelastic stresses. Xenolith geot-
a constant force and constant stress on the spring. herm data provide some constraint on the rate
Now, consider the same spring stretched between that temperature changes (Bell et al., 2003). Part
2 fixed attachments. The spring relaxes, leaving of the change comes from the gradual cooling of
it unstressed. Earth’s sublithospheric mantle at 50 K per bil-
Body forces, mainly from the buoyancy of mid- lion years (Abbott et al., 1994; Galer and Mezger,
oceanic ridges (ridge push), and intraplate varia- 1998). Intracontinental basin formation (e.g.,
tions of lithosphere thickness and density (lithos- Kaminski and Jaupart, 2000) and mantle plumes
pheric buoyancy) maintain renewable stresses on may cause more rapid nonmonotonic changes.
the plates. The stress orientation from borehole The strain rate scales to the strain at the depth
measurements and earthquake mechanisms indi- where the material behaves rigidly, crudely
cates that these forces dominate within the stable 500°C, that is,
parts of the plate interiors (Zoback et al., 1989).
TRT
The strain rate from this stress must therefore ex- , (5)
ceed the strain rates for nonrenewable stress. As TLt
with a spring stretched between 2 attachments, where is the linear thermal expansion coeffi-
kinematics constrains the strains and strain rate cient [0.8 105 K1 for granite and 0.5
FIG. 1. Schematic diagram shows observational constraints on permeability and strain rate. The strain rate in cra-
tons needs to be greater than 1020 s1; otherwise nonrenewal stresses would dominate. It needs to be less than 1018
because deformation of the crust is not evident. The permeability is greater than 1017 m2 because the fluid pressure
is near hydrostatic. It needs to be less than 1016 m2 so that convection does not obviously perturb the geotherm from
a conductive gradient in a deep well. These ranges imply sluggish but finite fluid flow favorable to sustained habit-
ability.EARLY HABITABLE CRUST 1027
105 K1 for mafic rocks in the range 20–100°C [SN PF], (6)
(Skinner, 1966, Table 6-10)], TR is the temperature
where the lithosphere behaves rigidly, TL is the where is the coefficient of friction, 0.6–1.0. The
temperature at the base of the lithosphere, and term in the brackets is the effective stress, where
the temperature at the base of the lithosphere SN is the normal traction resolved on a fault sur-
changes T over time t. A change in the tem- face and PF is the fluid pressure. The shear trac-
perature at the base of the lithosphere of 50 K in tion is close to failure on many pre-existing faults
a billion years causes a strain rate of 0.6 1020 in the brittle crust, even in stable continental in-
s1. Mantle plumes (e.g., Sleep, 2006) would cause teriors (Zoback et al., 2002).
episodic strain rates a few times this amount. The Coulomb formula (6) suffices to show that
Reiterating, the strain rate in cratons needs to stresses have relaxed in the upper crust. The nor-
be greater than strain rates that cause nonrenew- mal traction in (6) scales with lithostatic pressure
able stress, that is, a few times, 1020 s1. It needs
to be less than 1018 s1, which would produce SL rgZ, (7)
evident deformation over geological time (Fig. 1).
As we deal only with processes that scale linearly where r is the density of the rock, g is the accel-
with strain rate, we use a single value 1019 s1 eration of gravity, and Z is depth. The hydrosta-
in examples with the caveat that the actual strain tic pressure gradient is
rate is likely to vary in space and time. Zoback
and Townend (2001) obtained 1020 s1 as an up- PH wgZ, (8)
per bound strain rate for stable regions, using rhe-
ological models for creep within the lithosphere where w is the density of water. Measurements
and estimates of the magnitude of buoyancy-re- indicate that the fluid pressure within the crys-
lated forces. talline basement rocks of stable crust is very close
The observed strains on Mars are of the same to hydrostatic (see review in Townend and
order as those in stable cratons on Earth, again Zoback, 2000). Letting the acceleration of gravity
with the caveat that observed strain rates vary in equal 9.8 m s2, the coefficient of friction 0.7,
space and time. For example, the strain in Lunae the water density equal 1000 kg m3, and the rock
Planum is 0.003 (Plescia, 1991), and the strain in density equal 2800 kg m3 yields a shear stress
Arcadia Planitia is 0.0006 (Plescia, 1993). Averag- in (6) of 62 MPa at 5 km depth. Deep brines in
ing these strains over 3 billion years gives rates cratonal rocks are often saline (e.g., Bottomley et
of 0.6 to 3 1020 s1. This is not surprising, as al., 2002, 2003; Fehn and Snyder, 2005; Négrel and
the processes are similar to terrestrial ones: global Casanova, 2005). For example, the water density
cooling of the interior, local cooling of the deep of 1100 kg m3 implies a shear stress in (6) of 58
lithosphere, and body forces from lateral hetero- MPa at 5 km depth.
geneities. The ridge push buoyancy force imposes a
stress resultant on the plate that determines the
integral of the membrane stress over depth
Stresses and strains within the habitable crust
Our nominal strain rate of 1018 s1 would pro- Y (Sxx SL)dz, (9)
duce significant stresses over geological time if it
were not relieved. We consider the stress balance the indices xx indicate a component of the hori-
within the lithosphere to show what happens zontal stress. The difference between horizontal
within the upper few habitable kilometers where and vertical stress causes creep in ductile materi-
temperatures are less than 100°C. als and faulting if the Coulomb criterion (6) is ex-
Numerous boreholes have penetrated the hab- ceeded. Faulting relaxes the stresses within the
itable region of Earth’s crust. Stress measure- shallow part of the crust so that the stress differ-
ments, including induced seismicity, show that ence in (9) satisfies (6). Creep relaxes stresses in
the crust is critically stressed so that it is near fric- deeper and hotter parts of the plate, typically 15
tional failure at depths down to at least 8 kilo- km depth. The stress in (9) thus concentrates in
meters (see summary in Townend and Zoback, the middle part of the plate, where it is too deep
2000). That is, the resolved shear tractions on all for faulting and too cold for rapid creep. Ductile
the fractures satisfy the Coulomb inequality creep within this highly viscous region is the rate-1028 SLEEP AND ZOBACK
limiting step. At steady state, the plate deforms quakes, the surface area of fault slip is a lower
over its full depth at a constant horizontal strain bound on the surface area created by a slip event.
rate. This deformation generates little stress in the We are now ready to use the globally observed
deep hot part of the plate and maintains the stress relationship that the number of events in a mag-
near frictional failure in the upper crust (Zoback nitude bin decreases by a factor of 10 for each
et al., 2002). magnitude, m. The total moment release is
mmax
Earthquake recurrence MT Cm 100.5m dm, (14)
As noted above, earthquakes on faults main- where Cm is a constant, mmax is the magnitude of
tain permeability in the crust and, in fact, episod- the largest event, and the integral converges
ically release trapped water by opening new rapidly enough that the lower limit is taken as
pathways. The statistics of large earthquakes . The moment released by events with mag-
must be considered because earthquakes release nitudes between mmax and mmax m is
much of the strain within the brittle zone, though
mmax
we are mainly interested in small frequent events. M(m) Cm 100.5m dm
For purposes of illustration, we bin earthquakes mmaxm
by 1 unit of magnitude as this leads to simple sta- MT [1 100.5m]. (15)
tistics. The Guttenberg and Richter relationship
(e.g., Kanamori, 1977; Marsan, 2005) implies that, (e.g., Marsan, 2005). This implies that the largest
for every 1 magnitude-7 event, there will be about events in a region release most of the moment.
10 magnitude-6 events, 100 magnitude-5 events, Conversely, the total moment release is statisti-
and so on. cally the moments of the earthquakes in the max-
We use seismic moment to relate seismicity to imum moment bin (m 1) divided by 1 10 .
the average cratonal strain rate of 1019 s1. By In addition, the Guttenberg and Richter relation-
definition, the moment is ship between strain and moment release (12), and
the relationship between moment and magnitude
M SA, (10) (13) imply that total surface area (the area A of
an individual event times the number of events
where is the shear modulus (40 GPa for
with that magnitude) is the same in each magni-
crustal rocks), S is the displacement during the
tude bin.
earthquake, and A is the surface area of fault rup-
The number of small events, however, is not
ture (Kanamori, 1997). From Hooke’s law, the slip
strongly dependent on the magnitude of the
on the fault is proportional to its dimension x
largest events. We let the largest magnitude of in-
A and the stress drop . This gives the mo-
traplate earthquakes in most regions be either 6
ment in terms of stress drop, which is empirically
or 7. The number of magnitude-2 events in a cu-
independent of earthquake size
bic kilometer per million years is 6 and 2, re-
spectively, for a strain rate of 1018 s1. There are
M x3. (11)
between 600–200 magnitude-0 events, each with
The strain averaged over some crustal volume V a surface area of 100 m2 and 60,000–20,000 mag-
is nitude-2 events each with a surface area of 1 m2.
The total ruptured area depends linearly on the
M SA number of magnitude bins that actually occur.
. (12)
V V We assume 5 bins with magnitude 2 being the
The moment magnitude m is defined by largest that can be reasonably expected on a kilo-
meter scale over a reasonably short geological
M (1.25 109 nt-m)m1.5. (13) time and magnitude 2 as a cut off for small events.
This gives a lower bound of surface-area pro-
(Kanamori, 1977). For example, (11) and (13) im- duction rate of 600,000–200,000 m2 per million
ply that a magnitude-2 event with a typical stress years.
drop of 1.25 MPa has a surface area of 104 m2. Be- Biota directly benefiting from earthquake rup-
cause there is potentially a great deal of rock frac- ture (releasing, for example, H2) thus have to per-
turing and comminution accompanying earth- sist over times between events. This conclusionEARLY HABITABLE CRUST 1029
holds even though our recurrence times are volume thermal expansion coefficient (6 104
somewhat pessimistic. The actual distribution of K1) of water at hydrostatic pressure at 5 km
events in space is patchy. Faults favorably ori- depth and 100°C. The permeability is between
ented for slip in a given stress field would be ex- 1016 and 1017 m2 (Townend and Zoback, 2000).
pected to rupture repeatedly in otherwise intact This implies a macroscopic velocity of 1011 and
rock. 1012 m s1 or 3 104 to 3 105 m yr1, re-
spectively (Fig. 1). The fluid velocity within a
Permeability and fluid flow given fracture is a factor of the inverse of the open
crack porosity times the macroscopic velocity, 100
We now consider what happens during the in- to 1000 times greater. This gives a particle veloc-
terseismic period in continental crust. Direct mea- ity range of 3–300 m per millennium. This implies
surements indicate that the crust is permeable that the water is not stagnant and that substrates,
enough that pore water is essentially at hydro- such as H2, which are produced at depth and are
static pressure and fractures aligned favorably to soluble in water, readily move around on a short
the stress field carry most of the fluid flow (Tow- geological timescale once they are in major
nend and Zoback, 2000; Zoback and Townend, cracks. Hence faulting opens new pathways,
2001). which allow circulating fluids to “mine” the nu-
Circulating flow is relevant to biota, as it brings trients. The bounty is likely to cause microbial
in reactants such as CO2 and sweeps out reaction blooms. Unfaulted domains in the crust have per-
products such as CH4. As explained in the Ap- meability on the order of 1019 to 1020 m2 (Tow-
pendix, it would be inappropriate to consider the nend and Zoback, 2000). These rocks sequester
stability of convection in a permeable laterally ho- nutrients but at such low permeabilities that fluid
mogeneous region. We show here that hetero- movement would be so slow as not to provide
geneity implies that a small amount of hy- sufficient nutrients to sustain life. Conversely, if
drothermal circulation always occurs. permeability were much higher than 1016 m2,
We begin with D’Arcy’s law; the macroscopic fluid movement would be so rapid as to exhaust
velocity of flow is dimensionally nutrients from a given volume of rock in short
kwg geological periods of time. We show in the Ap-
VD , (16) pendix that such high permeabilities can be ruled
w
out by observations of conductive heat flow to
where k is the permeability, w is the local lat- great depths in the crust.
eral variation of water density from lateral vari- Studies of deep brines provide evidence of the
ations in temperature, and w is the viscosity of properties of deeply trapped water. Iodine-129
the water. The particle velocity of the water is studies indicate that brines have remained
VD trapped (that is, not mixed with surface water)
VP , (17) for tens of millions of years (Bottomley et al., 2002;
Fehn and Snyder, 2005). Although the origin of
where is the through-going porosity. the brines is not clear, it is evident that grain-scale
We give an example for the 5 km–deep base of permeability was high enough that extensive wa-
the habitable zone where the temperature is ter-rock reaction occurred (Bottomley et al., 2002,
100°C and the fluid pressure is 50 MPa. The vis- 2003; Négrel and Casanova, 2005; Fehn and Sny-
cosity is 0.3 103 Pa s. Our treatment also ap- der, 2005).
plies to Mars if water is present, as we obtain or-
der-of-magnitude answers. Lateral variations in
temperature result from lateral variations in ther- Resource sustainability
mal conductivity, lateral variations in radioactive We now use these computed flow rates to ap-
element distribution, and lateral variations in the praise whether hard rocks have enough resources
heat flow coming from below. Lateral variation to sustain life for geologically long times. We start
of a few K over the distance of a kilometer is rea- with the water to rock ratio. Dimensionally, the
sonable at this depth. To provide a quantitative volume ratio is
example, we assume lateral temperature varia-
tions of 5 K (Clauser et al., 1997; Popov et al., 1999), V Dt
R , (18)
which implies density variations 3 kg m3 for the Z1030 SLEEP AND ZOBACK
where the circulation persists for time t, and an The water moves on the scale of kilometers over
equivalent thickness Z is available for reaction. millennium to million-year timescales. The cu-
For 1 billion years, the computed flow rates, and mulative water/rock ratio over geological time is
a thickness of 5 km, the volume ratio is 100–1000. large, 100–1000 by volume. There is enough FeO
The ratio for stagnant brines is lower. For exam- in basalt to provide hydrogen to biota over bil-
ple, 1% porosity and a residence time of 10 mil- lions of years of geological time.
lion years (see Bottomley et al., 2002; Fehn and The situation for microbes in the crust would
Snyder, 2005) imply a ratio of 1 over a billion differ significantly if the permeability and flow
years. rate were much larger or much smaller than we
We illustrate the implications of this ratio with assumed. Lower flow rates imply a much smaller
an example for basalt. The reactant (10% by mass sustainable population. Higher flow rates imply
FeO) is present on the order of 1 mole per liter of thermal convection with hot upwelling regions
rock. The product H2 is released at 10 M of and cool downwellings. The available resources,
water using the Icelandic data (Stefánsson and especially within upwellings, could be exhausted
Arnórsson, 2002). Our higher estimated water/ over relatively short periods of geological time.
rock ratio, 1000, depletes only 1% of this resource.
Water flow, however, is able to deplete rocks
that are more reactive than basalt, such as ser-
APPENDIX: FREE CONVECTION
pentinite. For example, the H2 concentration of
end-member Lost City marine vent fluid is 15 mM
Rapid fluid circulation would significantly per-
(Kelley et al., 2005). As already noted, a
turb the temperature field in the crust if it were
water/rock ratio of 100 would suffice to remove
significantly more permeable than we have as-
the H2 produced by the 10% FeO in the ser-
sumed. We summarize the formalism with sim-
pentinite rock.
ple equations.
The crack heterogeneity of the rock tends to ra-
In the absence of hydrothermal circulation,
tion such potentially exhaustible resources. Crack
conduction would carry heat from Earth’s inte-
margins react, but intact masses of rock stay fresh
rior to its surface. To a first order, the vertical heat
until tapped by new fractures. Trapped water
flow,
equilibrates with nearby grains, as indicated by
old brines (Bottomley et al., 2002; Fehn and Sny- T
q K , (19)
der, 2005), but does not transport reactants. Z
where K is the thermal conductivity and Z is
depth, would not vary much with depth. In fact,
CONCLUSION
deep boreholes in hard rock show little variation
of heat flow with depth. Deviations of the ob-
We began with the suggestion by Sherwood
served geothermal gradient in bores from the ge-
Lollar et al. (2006) that episodic faulting releases
othermal gradient implied by constant heat flow
batches of H2-rich or CH4-rich water into hard-
are modest, that is, several K (Clauser et al., 1997;
rock environments. This transiently puts the re-
Popov et al., 1999). This observation provides an
actant well above the level that abiotic processes
upper limit on the rate of fluid flow and an up-
can consume it. The microbial population blooms
per limit on permeability.
during the temporary period of bounty. The
The convective heat flow is dimensionally
probable strain rate in stable crust is rapid
enough that numerous cracks fail in a cubic kilo-
qv wCwVDT (20)
meter volume on a million year timescale. The
mechanism is attractive for maintaining habit-
where Cw is the specific heat of the water 4
ability in the upper 5 km or so of the cratonic
103 J kg1 K1, w is the density of water, and T
crust.
is the temperature contrast between the water
The water and the organisms within the deep
and the surface. The Darcy velocity in terms of
crust should not be regarded as spatially stag-
water properties is
nant. The stress in the habitable zone also main-
tains a population of permeable cracks, which are k wwgT
VD , (21)
local niches that tap large volumes of bulk rock. wEARLY HABITABLE CRUST 1031
where w is the volume thermal compaction co- differences, tens of Kelvin, cannot cause dense
efficient of water. Combining (20) and (21) yields brines to ascend into less dense fresh waters.
However, deep dense brines can thermally con-
w
2C
w wkgT
2
qv (22) vect internally if they are not locally stratified. In
w this case, the brine may have circulated consid-
It is illustrative to rewrite (22) in terms of the di- erably in the subsurface yet be old in the sense
mensionless Raleigh number Ra that it has not mixed with surface reservoirs.
Thermohaline convection is complicated and
kT 2wgCw wTkZ
qv qdT (23) beyond the scope of this paper. We note that
Z wK episodes of rapid overturn can occur when the
where Z is the distance between 2 horizontal density change from temperature gradients over-
isothermal boundaries and T is the temperature whelms the chemical stratification. Conversely,
contrast between the boundaries. When the local unstable differences in salinity may drive
Raleigh number is below a critical value, convec- flow.
tion does not occur within a homogeneous re-
gion. In Earth, local lateral temperature gradients
associated with heterogeneities drive sluggish
convection. Convection occurs within a homoge- REFERENCES
neous region when the Raleigh number exceeds
Abbott, D.L., Burgess, L., Longhi, J., and Smith, W.H.F.
the critical value. If the Raleigh number modestly (1994) An empirical thermal history of the Earth’s up-
exceeds the critical value, the heat flux by vigor- per mantle. J. Geophys. Res. 99, 13835–13850.
ous convection is proportional to the Raleigh Bell, D.R., Schmitz, M.D., and Janney, P.E. (2003) Meso-
number. The critical Raleigh number is 4 2 zoic thermal evolution of the southern African mantle
39.5 for 2 permeable boundaries and 27.1 for a lithosphere. Lithos 71, 273–287.
lower impermeable boundary and a constant Bottomley, D.J., Renaud, R., Kotzer, T., and Clark, I.D.
(2002) Iodine-129 constraints on residence times of deep
pressure upper boundary (Turcotte and Schubert,
marine brines in the Canadian Shield. Geology 30,
2002). 587–590.
In the case of the upper continental crust, the Bottomley, D.J., Chan, L.H., Katz, A., Starinsky, A., and
thermal state of Earth’s interior determines the Clark, I.D. (2003) Lithium isotope geochemistry and ori-
heat flow from below; it is 40 mW m2 within gin of Canadian Shield brines. Ground Water 41,
stable shields (e.g., Kaminski and Jaupart, 2000). 847–856.
The thermal gradient is then 20 K km1 as we Calais, E., Han, J.Y., DeMets, C., and Nocquet, J.M. (2006)
assumed above. The permeability at the critical Deformation of the North American plate interior from
a decade of continuous GPS measurements. J. Geophys.
Raleigh number in terms of a given conductive
Res. 111, B06402.
thermal gradient from (23) is Chapelle, F.H., O’Neill, K., Bradley, P.M., Methé, B.A.,
4.2 1010 T 1 Ciufo, S.A., Knobel, L.L., and Lovley, D.R. (2002) A hy-
kcrit (24) drogen-based subsurface microbial community domi-
Z2 Z nated by methanogens. Nature 415, 312–315.
where all the terms are in SI units and the term Clauser, C., Giese, P., Huenges, E., Kohl, T., Lehmann, H.,
in brackets is the conductive thermal gradient Rybach, L., Šafanda, J., Wilhelm, H., Windloff, K., and
(Turcotte and Schubert, 2002, Equation 9-133). Zoth, G. (1997) The thermal regime of the crystalline
continental crust: implications from the KTB. J. Geophys.
The critical permeability depends strongly on the
Res. 102, 18417–18441.
thickness of the convective region: 8.3 1016 for Drobner, E., Huber, H.J., Wachterhauser, G., Rose, D., and
a 5 km–thick layer and 0.93 1016 for a 15 km Stetter, K.O. (1990) Pyrite formation linked with hy-
layer. We can exclude a through-going perme- drogen evolution under anaerobic conditions. Nature
ability of more than a few 1016 m2 as all deep 346, 742–744.
borehole data available confirm a nearly con- Galer, S.J.G. and Mezger, K. (1998) Metamorphism de-
ducting geothermal gradient (Clauser et al., 1997; nudation and sea level in the Archean and cooling of
the Earth. Precambrian Res. 92, 387–412.
Popov et al., 1999).
Fehn, U. and Snyder, G.T. (2005) Residence times and
Finally, the existence of dense brine in the deep source ages of deep crustal fluids: interpretation of 129I
subsurface (Bottomley et al., 2002, 2003; Négrel and 36Cl results from the KTB-VB drill site, Germany.
and Casanova, 2005; Fehn and Snyder, 2005) pro- Geofluids 5, 42–51.
vides stable stratification. Modest temperature Hoehler, T.M. (2005) Biochemistry of dihydrogen (H2). In1032 SLEEP AND ZOBACK
Metal Ions in Biological Systems, Volume 43: Biogeochemi- between 25 and 125°C: the rate equation. Geochim. Cos-
cal Cycles of Elements, edited by A. Sigel, H. Sigel, and mochim. Acta 61, 115–134.
R.K.O. Sigel, Taylor and Francis, Boca Raton, pp. 9–48. Rickard, D. and Luther, G.W., III (1997) Kinetics of pyrite
Hoehler, T.M., Alperin, M.J., Albert, D.B., and Martens, formation by the H2S oxidation of iron (II) monosulfide
C.S. (1998) Thermodynamic control on H2 concentra- in aqueous solutions between 25 and 125°C: the mech-
tions in an anoxic marine sediment. Geochim. Cos- anism. Geochim. Cosmochim. Acta 61, 135–147.
mochim. Acta 62, 1745–1756. Sherwood Lollar, B., Lacrampe-Couloume, G., Slater,
Hoehler, T.M., Alperin, M.J., Albert, D.B., and Martens, G.F., Ward, J., Moser, D.P., Gihring, T.M., Lin, L.H., and
C.S. (2001) Apparent minimum free energy require- Onstott, T.C. (2006) Unravelling abiogenic and biogenic
ments for methanogenic Archaea and sulfate-reducing sources of methane in the Earth’s deep subsurface.
bacteria in an anoxic marine sediment. FEMS Microbi- Chem. Geol. 226, 328–339.
ology Ecology 38, 33–41. Skinner, B.J. (1966) Thermal expansion. In Handbook of
Jackson, B.E. and McInerney, M.J. (2002) Anaerobic mi- Physical Constants, Geological Society of America, Memoir
crobial metabolism can proceed close to thermody- 97, Geological Society of America, Boulder, CO, pp.
namic limits. Nature 415, 454–456. 75–96.
Kaminski, E. and Jaupart, C. (2000) Lithospheric structure Sleep, N.H. (2006) Mantle plumes from top to bottom.
beneath the Phanerozoic intracratonal basins of North Earth-Science Reviews 77, 231–271.
America. Earth Planet. Sci. Lett. 178, 139–149. Solomon, S.C. (1987) Secular cooling of the Earth as a
Kanamori, H. (1977) The energy release in great earth- source of intraplate stress. Earth Planet. Sci. Lett. 83,
quakes. J. Geophys. Res. 82, 2981–2987. 153–158.
Kelley, D.S., et al. (2005) A serpentine-hosted ecosystem: Stefánsson, A. and Arnórsson, S. (2002) Gas pressures and
the Lost City hydrothermal field. Science 307, 1428–1434. redox reactions in geothermal fluids in Iceland. Chem.
Kral, T.A., Brink, K.M., Miller S.L., and McCay, C.P. (1998) Geol. 190, 251–271.
Hydrogen consumptions by methanogens on the early Townend, J. and Zoback, M.D. (2000) How faulting keeps
Earth. Orig. Life Evol. Biosph. 28, 311–319. the crust strong. Geology 28, 399–402.
Lin, L.-H., Hall, J., Lippmann-Pipke, J., Ward, J.A., Sher- Turcotte, D.L. and Schubert, G. (2002) Geodynamics, 2nd
wood Lollar, B., DeFlaun, M., Rothmel, R., Moser, D., ed., Cambridge University Press, Cambridge.
Gihring, T.M., Mislowack, B., and Onstott, T.C. (2005) Wachtershauser, G. (1988) Pyrite formation, the first en-
Radiolytic H2 in continental crust: nuclear power for ergy source for life: a hypothesis. System Appl. Micro-
deep subsurface microbial communities. Geochem. Geo- biol. 10, 207–210.
phys. Geosyst. 6, Q07003. Zoback, M.D. and Townend, J. (2001) Implications of hy-
Marsan, D. (2005) The role of small earthquakes in redis- drostatic pore pressures and high crustal strength for
tributing crustal elastic stress. Geophys. J. Int. 163, the deformation of intraplate lithosphere. Tectono-
141–151. physics 336, 19–30.
Mooney, W.D., Laske, G., and Masters, T.G. (1998) CRUST Zoback, M.L., Zoback, M.D., Adams, J., Assumpcao, M.,
5.1: a global crustal model at 5° 5°. J. Geophys. Res. and Bell, S. (1989) Global patterns of tectonic stress. Na-
103, 727–747. ture 341, 291–298.
Négrel, P. and Casanova, J. (2005) Comparison of the Sr Zoback, M.D., Townend, J., and Grollimund, B. (2002)
isotopic signatures in brines of the Canadian and Steady-state failure equilibrium and deformation of in-
Fennoscandian shields. Appl. Geochem. 20, 749–766. traplate lithosphere. Int. Geol. Rev. 44, 383–401.
Parkes, R.J., Cragg, B.A., and Wellsbury, P. (2000) Recent
studies on bacterial populations and processes in sub-
seafloor sediments: a review. Hydrogeology J. 8, 11–28.
Plescia, J.B. (1991) Wrinkle ridges in Lunae Planum Mars: Address reprint requests to:
implications for shortening and strain. Geophys. Res. Norman H. Sleep
Lett. 18, 913–916. Department of Geophysics
Plescia, J.B. (1993) Wrinkle ridges of Arcadia Planitia, Mitchell Building, Room 366
Mars. J. Geophys. Res. 98, 15049–15059.
397 Panama Mall
Popov, Y.A., Pevzner, S.L., Pimenov, V.P., and Ro-
mushkevich, R.A. (1999) New geothermal data from the
Stanford University
Kola superdeep well SG-3. Tectonophysics 306, 345–366. Stanford, California 94305 USA
Rickard, D. (1997) Kinetics of pyrite formation by the H2S
oxidation of iron (II) monosulfide in aqueous solutions E-mail: norm@stanford.eduYou can also read
