You are on page 1of 15

ICARUS 90, 222-236 (1991)

Mantle Differentiation and Thermal Evolution of Mars, Mercury,


and Venus
TILMAN SPOHN

Westf61ische Wilhehns-Universitdt, lnstitut,ft~r Planetologie, W. Klemmstrasse 10, 1)-4400 Mfinster, Germany

Received January 22, 199(): revised July 24, 1990

1981]. Like the E a r t h ' s crust, the possible basaltic crusts


Thermal evolution models for the terrestrial planets Mars, Mer- of these terrestrial planets and that of Mercury are likely
cury, and Venus with core and mantle chemical differentiation, to be produced by the differentiation of their mantles.
lithosphere growth, and volcanic heat transfer have been calcu- Partial melting of mantle rock, most probably due to de-
lated. The mantle differentiates by forming a crust and the core compression in convective upwellings, produces bouyant
differentiates by inner core solidification. Continued volcanic activ- magmatic fluids which intrude near surface layers of rock
ity for billions-of-years is found to be possible even on small terres- or extrude on the surface where they crystallize as crustal
trial planets if crust growth is limited by lithosphere growth during rock. On Earth, the crust forms a thin layer of a few tens-
the early evolution. Later, crust formation may be limited by the of-kilometer thickness on top of the lithosphere. It has
declining vigor of mantle convection. The thicknesses of the crust
been speculated that the crusts of Venus and Mars may
and lithosphere are found to depend mainly on planet size, on the
be significantly thicker than that of Earth and may consti-
bulk concentration of radiogenic elements in the planet, and on
the ratio between volcanic and conductive heat transfer through tute a larger part of their lithospheres (e.g., Anderson
the lithosphere. Two end-member models have been calculated 1980, BVSP 1981). On one-plate planets the rheological
and the concentration of radiogenics in the planet has been varied. lithosphere forms a stagnant lid that overlies the convect-
In the first model, heat transfer from the mantle to the surface ing mantle. The theological lithosphere is part of the
occurs via heat conduction through the lithosphere, while in the thicker thermal lithosphere. The latter is defined as the
second model, mantle heat is advected via volcanic vents. Geologic conductive outer layer of a planet and comprises the rheo-
evidence for volcanism on Mars and Mercury for at least 3.5 Ga logical lithosphere and the underlying thermal boundary
and up to 1 Ga, respectively, the absence of a magnetic field on layer of the mantle convection. For Earth, the term litho-
Mars, and the presence of such a field on Mercury suggest that sphere is often associated with the thermal lithosphere.
heat transfer in these planets was dominated by heat conduction
To avoid confusion, I emphasize that the term lithosphere
through the lithosphere for most of their thermal history. The
in this paper refers to the theological lithosphere. The
present crust of Mercury is estimated to be a few tens of kilometers
radioactive isotopes 4°K, L4°Th, and 256U, the decay of
thick and about 10% of the mantle initial inventory of heat sources
is fractionated into the crust. The Martian crust may be 50-100 which is thought to be the primary mode of heat generation
km thick, possibly constituting more than a third of the lithosphere. in the terrestrial planets, are presumably enriched in the
Volcanic heat piping may have been an important heat transfer magmas and thus in the crusts because of their large ionic
mechanism on Venus and volcanic activity may continue to the radii and their valence states (Philpotts and Schnetzler,
present day. Venus may have a crust that may constitute almost 1970). Accordingly, planetary differentiation and crustal
the entire lithosphere but crustal thickness may be limited by the growth deplete the mantles of heat sources and potentially
basalt-eclogite phase transformation to 60 to 80 km. It is estimated are processes of primary importance to the thermal evolu-
that the present mantles of Mars and Venus are similarly depleted tions of the terrestrial planets. As is particularly evident
of about 20 to 40% of their initial heat source inventory. , ~99~
for the Earth, however, mantle differentiation is not an
Academic Press, Inc.
irreversible process because crust may be remixed with
the mantle. For one-plate planets, for which plate tecton-
1. INTRODUCTION ics is not available to remix crust, remixing with the man-
tle may be promoted by crustal delamination (e.g., Head
Basaltic rock has been observed on the surfaces of 1986), for instance.
Mars, Venus, and the Moon which show evidence for past Phillips and Malin (1983) and, recently, Turcotte (1989a)
volcanism [e.g., Basaltic Vokanism Study Project (BVSP) and Schubert et al. (1990) have presented models of

0019-1035/91 $3.00
Copyright © 1991 by Academic Press, Inc.
All rights of reproduction in any form reserved.
DIFFERENTIATION AND THERMAL EVOLUTION 223

crustal growth and planetary differentiation for Venus and convective heat flow from the sublithospheric mantle into
Mars. Phillips and Malin (1983) assume a reasonable but the lithosphere base (Schubert et al. 1979, Spohn and
otherwise unconstrained exponential growth law for the Schubert 1982a, 1983). On one-plate planets, conductive
crust and a final thickness of the Venusian crust of 10 km. heat transfer across the stagnant lithosphere is a bottle-
Turcotte (1989a) and Schubert et al. (1990) assume that neck for heat flow from the interior. That bottleneck can
the crustal growth and planetary differentiation rates of be circumvented to some extent by heat piping through
both planets are regulated by the mantle convection volcanism (O'Reilly and Davies 1981, Stevenson and
speed. Mantle convection speed, in turn, is regulated by McNamara 1988, Turcotte 1988, 1989b). Schubert et al.
the temperature-dependent mantle rheology. These au- (1979) were the first to consider lithosphere growth with
thors make the crucial assumption that mantle differentia- thermal evolution models of the terrestrial planets. They
tion is irreversible and obtain very rapid crustal growth in obtained hundreds-of-kilometers-thick present-day litho-
the first few 100 Ma of the thermal evolution of the planets spheres because they neglected heat generation by the
and very little growth thereafter. This result is at variance decay of radioactive elements in the mantles of their mod-
with evidence for continued volcanic activity on Mars for els. Present-day lithosphere thicknesses between 150 and
probably 3.5 Ga or more of its evolution (Neukum and 200 km have been calculated by Schubert et al. (1988)
Hiller 1981, BVSP 1981, McGill 1989) and possibly pres- for Mercury. For Mars, Schubert and Spohn (1990) have
ent volcanic activity on Venus (Schaber 1982, Scarf and calculated a present-day lithosphere thickness of about
Russel 1983, Esposito 1984, Head and Wilson 1986). Also, 100 km. These lithosphere thicknesses were calculated
the volcanic plains on Mercury may be as young as the by neglecting the effects of mantle differentiation and
lunar maria or even somewhat younger (e.g., BVSP 1981), volcanic heat transfer.
which suggests a period of at least some volcanic activity In addition to lithosphere thickness, mantle convection
of about I Ga since the accretion of that planet. Moreover, speed, and remixing rate, the crustal differentiation rate
for efficient differentiation, the early crustal thickness of should depend on the efficiency of magma generation in
Mars given by Schubert et al. (1990) is much larger than the mantle and on the magma transport rate to the crust.
the early lithosphere thickness calculated recently by The mantle may also become chemically layered during
Schubert and Spohn (1990) for that planet. differentiation. However, mantle chemical layering is be-
It can be argued that the crusts of the terrestrial planets yond the scope of the present paper which assumes that
cannot become thicker than their lithospheres and that the mantle is differentiated homogeneously. The effi-
crustal growth rate is limited by lithosphere growth rate. ciency of magma generation is essentially the ratio be-
It is conceivable, if not likely, that sublithosphere mantle tween the mantle convection turnover time and the char-
convection would remix any extra crustal material under- acteristic time for crustal fractionation. These quantities
neath the lithosphere with the mantle. This process may are difficult to estimate. In this paper, I therefore first
be helped by the basalt-eclogite phase transformation present models of the thermal evolutions of the terrestrial
(e.g., Turcotte 1989b). A similar mechanism of mass re- planets Mars, Mercury, and Venus with crustal growth
turn to the planetary interior has been proposed for Io simply limited by the growth of the lithosphere. These
by O'Reilly and Davies (1981). Recent calculations have models inherently assume that the mantle is always well
shown that mantle convection is quite effective at mixing mixed and that differentiation is so efficient that the entire
chemical heterogeneities (e.g., Hoffman and McKenzie lithosphere is composed of basalt and that the crust is
1985, Christensen 1989). Of course, mantle convection identical with the lithosphere. If, as I have argued above,
may become too sluggish to effectively remix the extra lithosphere growth limits crustal growth and planetary
crust with the mantle as the planet cools and the mantle differentiation then these models provide crusts of maxi-
viscosity increases. However, mantle differentiation mum thickness and can be used to discuss the effect of
should also decrease with decreasing convective vigor as mantle differentiation on the thermal evolution of the
the crustal material removed from the upper mantle must planet in a maximum limiting case. The effects of heat
be replenished from the lower mantle to keep differentia- piping by volcanic heat transfer on lithosphere thickness,
tion going. mantle differentiation, and thermal evolution of the larger
Lithosphere growth may be influenced to some extent planets Mars and Venus are also considered. I then pro-
by crustal growth through the effect of the latter on the ceed to discuss additional limits placed on the differentia-
heat production rate in the mantle. Lithosphere grows tion rate by the declining mantle convective vigor and the
as the temperature of upper mantle rock falls below the basalt-to-eclogite phase transformation and estimate more
temperature at which rock begins to flow on geologic time realistic present-day crustal thicknesses for the planets. I
scales as a consequence of the secular cooling of a planet. argue that lithosphere growth limited crustal growth in the
The growth rate is proportional to the difference between early evolutions of Mercury and Mars and convective
the conductive heat flow across the lithosphere and the vigor should have limited crustal growth thereafter. For
224 TILMAN SPOHN

Venus, lithosphere growth and the basalt-to-eclogite glected. ]Fm replaces F~ of Stevenson et al. (1983). 1 use
phase transformation may presently be the limiting mech- the notation F m instead of F~ for mantle heat flow to
anisms. The results together with the evidence from plane- avoid confusion with surface heat flow.] Equation (!) is an
tary geology suggest crustal thicknesses of a few tens of energy balance equation for the lithosphere base. If the
kilometers for Mercury, 100 km for Mars, and up to 80 lithosphere thickens by dl in time dt, then a quantity of
km for Venus. Accordingly, the mantles of the planets are heat PmCm(Tu - TOdl per unit area must be removed from
depleted of up to 40% of their initial inventory of heat the sublithospheric mantle. This cooling is provided by
sources. the difference between the upward conductive heat flux
out of the base of the lithosphere k(OT/Oz)]z=t and the
2. MODEL upward heat flux from the convecting mantle into the
base of the lithosphere Fml. Equation (1) has been used
I adopt and extend the thermal history model of Steven-
previously by Schubert et al. (1979), Spohn and Schubert
son et al. (1983) recently extended by Schubert et al.
(1982a, 1983), Schubert et al. (1988), and Schubert and
(1988) and Schubert and Spohn (1990). In this model, the
Spohn (1990) to calculate the evolution of lithosphere
planets are initially hot and differentiated into a silicate
thickness. Equation (1) mathematically resembles the Ste-
mantle and an iron-rich core. The initial mantle tempera-
fan boundary condition for melting and solidification (e.g.,
ture is subsolidus and the initial core temperature is super-
Carslaw and Jaeger 1954).
liquidus. The subsequent evolutions of the planets in this
Heat is transferred by heat conduction in the litho-
model consist of simple cooling with increasing mantle
sphere and the top surface is at constant temperature T~.
viscosity. The model also calculates a detailed core ther-
Volcanic heat transfer may bypass the lithosphere and
mal evolution with inner core solidification and estimates
transfer mantle heat directly to the surface. Depending on
of planetary magnetic moment. However, I pay relatively
the ratio between mantle heat transferred through the
little attention to core chemistry, core evolution, and mag-
lithosphere and that transferred through volcanic heat
netic field generation in this paper.
pipes, mantle heat flow Fm must be divided between Fm~
Heat transfer across the mantle is a main process that
and the volcanic heat transfer rate. That ratio is difficult
regulates the thermal evolution. The heat transfer rate is
to estimate and is likely to vary with time during the
calculated by using a simple Nusseit number-Rayleigh
evolutions of the planets. Therefore, I consider two end-
number relation. The mantle viscosity is assumed homo-
member models: (1) a conductive lid model, for which I
geneous and temperature dependent. The temperature de-
assume that volcanic heat transfer is negligible and FmL
pendence of the mantle viscosity is an important feature
e q u a l s F m ~, and (2) a heat-pipe model, for which I assume
of the present and similar models and regulates mantle
that mantle heat is entirely transferred by heat piping and
heat transfer via a thermostat effect. Although the exact
that Fret is zero. The heat conduction equation in the
form of the heat transport parameterization differs among
lithosphere is solved with an implicit finite difference
various authors, the approach is well established as a way
scheme.
of calculating thermal evolution models (e.g., Sharpe and
I assume for the present model that the lithosphere
Peltier 1979, Schubert et al. 1979, 1986, Stevenson and
is composed entirely of basalt so that the lithosphere is
Turner 1979, Cook and Turcotte 1981, Spohn and Schu-
identical with the crust. For simplicity, I assume that the
bert 1982b, Spohn 1984, Peltier 1989). The reader is re-
thermal diffusivity of the mantle and that of the basaltic
ferred to Stevenson et al. (1983) for a detailed description
lithosphere are the same. However, the lithosphere is
of the model.
enriched in heat sources as compared with the mantle.
I first modified the model to allow a calculation of litho-
The formation of basaltic crust typically requires partial
sphere thickness versus time. The equation for litho-
melting of mantle rock by 10 to 20% which suggests an
sphere thickening, first introduced by Schubert et al.
enrichment of the magma over the mantle by a factor of
(1979), is
4 to 5 (BVSP 1981). This is also approximately the ratio
dl OT between the heat generation rate in a typical fresh terres-
pmCm( Tu - - Tl)d-t = k-2"-lzoz=/ - - Fml (1) trial tholeiitic basalt of 2.1 × 10 ~ W kg -~ (BVSP 1981)
and an estimate of the Earth mantle heat generation rate
where Pm is mantle density, C m is mantle heat capacity, that takes into account the time lag between heat produc-
Tu is the temperature in the upper convecting mantle, TI tion and heat flow predicted by most thermal evolution
is the temperature defining the base of the lithosphere, / models (e.g., Spohn 1984) and by geochemical data (e.g.,
is lithosphere thickness, t is time, k is thermal conductiv- DePaolo 1981). Assuming that the ratio between magma
ity, and z is depth. Fmj is the heat flow from the convecting and mantle heat source densities is constant with time one
mantle into the lithosphere base. Fm~ is equal to the heat can calculate a potential thickness of crust which is the
flow from the mantle F m if volcanic heat transfer is ne- thickness of the crust for an entirely depleted mantle
DIFFERENTIATION AND THERMAL EVOLUTION 225

Rp with a moment-of-inertia factor of 0.365 of Schubert and


Op°t ~ 3-A (2) Spohn (1990). These models are compatible with the pres-
ently available observations of planetary magnetic fields.
where Rp is the surface radius of the planet and A is the 1 vary the initial mantle heat source densities and the ratio
magma enrichment factor. Assuming a magma enrichment between mantle heat conducted through the lithosphere
factor of 5, I calculate potential thicknesses of 165 km for and that removed via volcanic heat piping. Aside from the
Mercury, 225 km for Mars, and 405 km for Venus. For rheology, these parameters are most likely to affect the
Mars and Mercury, these potential thicknesses are com- current estimates of crustal thicknesses, growth rates, and
parable to previous estimates of lithosphere thicknesses mantle differentiation rates. Mantle heat source density is
(e.g., Schubert et al. 1988, 1990). For Venus, the potential an important parameter for any planetary thermal history
crust thickness is ample in comparison with most previous calculation, more important for the mantle energy balance
estimates of crust and lithosphere thickness (e.g., Phillips than, for instance, the rheology parameters according to
and Malin 1983). Mass balance gives an estimate of the the results of Earth thermal history calculations of Spohn
initial concentration of crustal material in the mantle of and Schubert (1982b). In the present calculations, mantle
heat source density determines lithosphere growth rate
1 Pc
through the dependence of the latter on k(aT/Oz)lz=t and
CO ~ - - - - (3) Fml. While Fmj depends directly on mantle heat source
A,om
density, k(OT/az)lz=l depends on the mantle heat sources
that were fractionated into the crust. Moreover, as will
where Pc and Pm are the densities of crust and mantle.
be shown in the discussion section, mantle convective
With a ratio of crust to mantle density of approximately
vigor also depends on mantle heat source density. If the
0.75, I find an initial concentration of crustal material of
crustal growth rate depends on lithosphere growth rate
15%. This value is somewhat larger than the value of 10%
and/or on convective vigor, then it will also depend on
assumed by Schubert et al. (1990) for Mars.
mantle heat source density. For the present models, I use
The mantle heat production rate Qm decreases with time
initial (time t = 0) mantle heat source densities between
in response to crustal growth. I assume that
0.85 × 10 -7 and 1.7 x 10 -7 W m -3. The value of 1.7 x
10 -7 W m -3 is a representative value for the Earth's man-
(4) tie and, together with an average decay constant h of 1.38
x 10-13 s-l, gives a present-day value of Earth mantle
heat generation rate in equilibrium with about 60% of the
where Q is the rate of mantle heat generation without average present-day heat flow of 80 mW m -2. This value
differentiation. Due to the decay of radioactive elements, was also used by Stevenson et al. (1983), Schubert et al.
Q decreases with time according to (1988, 1990), and Schubert and Spohn (1990) for their
thermal evolution models. I derived the value of 0.85 x
Q = Q0exp(-M) (5) 10 -7 W m -3 by using the radiogenic element abundances
of both Treiman et al. (1986) and Laul et al. (1986) for
where h is the decay constant and Q0 is the initial mantle Mars from the study of SNC meteorites and a ratio be-
heat production rate. The rate of change with time of Qm tween Earth and Mars mantle densities of 1.3. The heat
is, therefore, source density calculated from the element abundances
of Venus surface rock (Surkov 1983) is up to 100 times
1 dQm= X I dl (6) larger than the preferred Earth value and is mostly likely
Om dt (Dpot - /) dt" not characteristic of the Venusian mantle. For easy refer-
ence, standard parameter values are summarized in Ta-
In my model, Qm replaces Q in the mantle energy bal- bles I and II.
ance equation [Eq. (8)] of Stevenson et al. (1983). Steven-
son et al. (1983) have considered a wide range of models 3. RESULTS OF THERMAL HISTORY CALCULATIONS
for the terrestrial planets. Schubert and Spohn (1990) have WITH CRUST GROWTH LIMITED BY
recently significantly extended the range of parameter LITHOSPHERE GROWTH
values investigated for Mars. In the present paper, I do
3.1. Mars
not vary parameters that mostly determine core evolution
as, for instance, core chemistry. I also do not vary mantle Figure I and Table III summarize the results of model
rheology parameters. Rather, I adopt models Me4 and V1 calculations for Mars. Figure 1 shows the evolutions for
from Stevenson et al. (1983) for Mercury and Venus. For the conductive lid model and the heat-pipe model of (a)
Mars, I adopt the model with 14% core sulfur content and the basaltic lithosphere thickness and the ratio between
226 TILMAN SPOHN

TABLE I to about 300 km are possible for Mars according to the


Parameter Values for Thermal Evolution Models model calculations. These values depend on the initial
k~ (W m -t K i) 40.00
heat source density and the degree of heat piping.
Core thermal conductivity
Mantle thermal conductivity k m (W m t K - t ) 4.00 Additional calculations have shown that the evolution
Mantle thermal diffusivity Km (m 2 s t) 10 ~ of the lithosphere thickness is quite insensitive to the
Viscosity constant v0 (m 2 s i) 4000. assumed magma heat source enrichment factor. For the
Activation parameter A (K -I) 5.2 x 104 conductive lid model, present-day basaltic lithosphere
Thermal expansivity o~ (K -I) 2. × 10 s
,k(s I) 1.38 × I0 17 thicknesses are between 110 and 170 km and the mantle
Radioactive decay constant
Critical Rayleigh number Rac, 50(I. is depleted of 45 to 75% of its heat source inventory
Heat transfer exponent B 0.3 (Fig. lc). A comparison between the lithosphere thick-
Lithosphere boundary temperature I] (K) 1073. nesses of the conductive lid model and the model of
Note. For further explanation of parameters see Stevenson et al.
(19831.
T A B L E II
Parameter Values for Thermal Evolution M o d e l s

Mars Mercury Venus


basaltic lithosphere thickness and potential crustal thick-
ness, (b) the basaltic lithosphere growth rate, and (c) the Planet radius Rp (km) 3389. 2440. 6051.
ratio between the mantle heat source density Qm and the Gravity g ( m s 2) 3.7 3.8 9.0
undifferentiated mantle heat source density Q. The latter Surface temperature 7", (K) 220. 440. 730.
Core radius Re (km) 1726. 1900. 3110.
ratio measures the degree of depletion of the mantle due
Core density Pc (kg m 3) 7000. 8200. 12500.
to differentiation. Three different values of initial mantle Initial sulfur content x0 0.14 0.05 0.10
heat source density of 0.85, 1.28, and 1.70 x 10 7 W m 3 Pressure at CMB Pcm (GPa) 19. 10. 130.
have been considered. For the heat-pipe model and initial Central pressure Pc (GPa) 411. 40. 290.
heat source densities of 0.85 and 1.28 × 10-7 W m 3 the Core liquidus Tin,, (K) 1880. 1880. 1960.
parameters Tmj (K T P a t) 13.6 13.6 6.14
basaltic lithosphere becomes thicker than my standard
T m 2 ( K T P a 2) -62.2 62.2 -4.5
value of potential crustal thickness for Mars. For these Core adiabat T~,~ tK T P a t) 8.00 8.00 3.96
models, I assumed magma heat source enrichment factors parameters T,,, (K T P A -21 39. -39. 3.3
of 3.8 and 4.5, respectively, thereby allowing for basaltic Core latent heat L (kJ kg ~) 250. 250. 250.
lithosphere thicknesses of 250 and 300 kin. Rate of gravitational Eg (kJ kg ~) 250. 250. 750.
energy release
Table III gives present-day values of conductive sur-
upon inner core
face heat flow, mantle heat flow, and core heat flow, freeze out
upper mantle temperature and core mantle boundary Ratio between ~c 1.1 1. I 1.2
temperature, mantle viscosity and Rayleigh number, representative
lithosphere thickness and upper thermal boundary layer core temperature
and C M B
thickness, and inner core radius for the present models
temperature
and for the comparable model of Schubert and Spohn Mantle density Pm (kg m ~) 3469. 3300. 4500.
(1990). The latter model assumes an initial heat source Ratio between ~m 1. I. 1.3
density of 1.70 x 10 -7 W m 3. The important difference representative
between the present model and the model of Schubert mantle
temperature and
and Spohn (1990) is the differentiation of the mantle. A
upper mantle
comparison of these results shows clearly how differenti- temperature
ation and volcanic heat transfer affect the present ther- Ratio between ~: 1. I. 1.6
mal state of the planet. Although the mantle heat flow upper mantle
values differ by almost a third between the present temperature and
CMB temperature
conductive lid model and the model of Schubert and
Initial upper mantle Tmo (Kt 2573. 2573. 2573.
Spohn (1990), which also features a conductive lid, temperature
differences in the values of upper mantle and core Initial C M B Tomo (K) 3000. 3000. 4700.
temperatures, Rayleigh number, and viscosity are of temperature
much lesser significance. There are significant differ-
N o t e . C M B refers to the core-mantle boundary. For further explana-
ences, however, between the model of Schubert and
tion of the listed parameters see Stevenson et al. (19831. Parameter
Spohn (1990) and the heat-pipe model which features a values for Mercury and Venus are from models Me4 and V 1 of Stevenson
much cooler and more evolved planetary mantle. et al., respectively. Parameter values for Mars are from Schubert and
Present-day crust and lithosphere thicknesses of up Spohn (19901.
DIFFERENTIATION AND THERMAL EVOLUTION 227

a
300.0
~b
"~ 20.0
?
"~ 250.0
1.0 ~.
200.0
0.0 ~

"~ 180.0 O
0.6 ~ *'~ lO.O
ca
\ o
I- ~.", heat pipe
I00.0
0.4 ~ ~)
~ ~.o
0 50.0
,I.,=

0.0 , , i , I i i f i [ i i i i [ i i i i [ i i o.o ~
1000 2000 3000 4000 ~ o.o t i ~ L I t t i t I i t t t I i i i i I t t

0 1000 2000 3000 4000


~me (~)
hi,m e (Ma)
C
1.0

0.9 "~':'".7.~ conducive lid


0.8

0.7

0.0

~ 0.8

0.4

0.3

0.2.

0. I

i i i I i i i i t i ~ i i I i L , i I J
0.0
! 000 2000 3000 4000
~me (~)
F I G . 1. Results of thermal evolution calculations for Mars. Mantle heat is entirely removed by volcanic heat transfer for heat pipe models. For
conductive lid models, mantle heat is removed by conduction through the lithosphere. The solid lines refer to initial heat source densities of 0.85
× 10 -7 W m -3. the broken lines refer to initial heat source densities of 1.28 × 10 -7 W m -3, and the dash-dotted lines refer to initial heat source
densities of 1.7 z 10 -7 W m -3. (a) Basaltic lithosphere thickness and ratio between basaltic lithosphere thickness and potential crustal thickness
as a function of time. (b) Basaltic lithosphere growth rate as a function of time. (c) Evolution of the ratio between mantle heat source density Qm
and mantle heat source density without differentiation Q. This latter ratio measures the degree of depletion of the mantle of heat sources due to
differentiation.

Schubert and Spohn (1990) in Table III shows that fundamentally differently from the crustal growth rates
mantle differentiation and crust formation cause an of Turcotte (1989a) and Schubert et al. (1990). Their
about 10% thicker lithosphere after 4.5 Ga. This differ- growth rates decrease very rapidly during the first few
ence does not depend sensitively on the initial heat hundred million years and become essentially zero after
source density as additional calculations have shown. 500 Ma. The basaltic lithosphere growth rate calculated
For the heat-pipe model, present-day basaltic lithosphere here also decreases rapidly during the first billion years.
thicknesses between 200 and 300 km are calculated and However, it attains a value between 2 and 6 km 3 year
the mantle is almost completely depleted of heat sources thereafter and increases by another km 3 year -I during
in 4.5 Ga (Fig. lc). Obviously, the results in Table III the remainder of the evolution. The increase in growth
suggest that the fractionation of mantle heat sources rate is a consequence of the decrease with time of
into the basaltic lithosphere is not as important for the mantle radioactive heating rate. If the mantle heat source
thermal states of the mantle and core as the overall density were constant, the growth rate would decrease
value of mantle heat production rate and the mode of monotonically with time. The present-day growth rates
lithosphere heat transfer. for both the heat-pipe and the conductive lid models
The basaltic lithosphere growth rate (Fig. lb) evolves are about the same. The evolution of the growth rate
228 T I L M A N SPOHN

TABLE III
Results for Mars Thermal History Models

Conductive lid Heat pipe

1.70 1.28 0.85 1.70 1.28 0.85 S&S

Fc~ (mW m-2) 39.7 33.5 27. I 30.8 25.9 20.3 39.7
F m (roW m 2) 20.5 15.3 10.0 10.3 8.2 6.5 29.9
F c (roW m -2) 3.6 3.4 3.2 3.4 3.5 3.5 3.5
Tu (K) 1696. 1655. 1602. 1605. 1577. 1549. 1750.
Tcm(K) 1806. 1785. 1767. 1771. 1763. 1754. 1829.
p (1016 m 2 s -~) 8. 2(I. 50. 500. 800. 1600. 3.
Ra ( 106) 3. I. 0.3 0.3 0.2 0.08 6.5
/ (km) 108. 131. 169. 207. 241. 293. 96.2
6 (km/ 121. 154. 21(I. 206. 246. 298. 90.7
R i (km) . . . . 270. 41 I. --
Onset (Ga) . . . . 4.3 4.0 --

N o t e . Present-day values of the conductive surface heat flow Fc,, the mantle heat flow f m .
the core heat flow F c , the upper mantle temperature To, the core-mantle boundary temperature
Tc,,,, the mantle viscosity u, the mantle Rayleigh number Ra, the lithosphere thickness l, the
upper mantle thermal boundary layer thickness 6, and the inner core radius R i are listed. " O n s e t "
gives the time of onset of inner core growth. S&S is a comparable model of Schubert and Spohn
(1989).

for both the heat-pipe and the conductive lid models with those of the conductive lid model for Q0 equal to 1.7
certainly allows for an extended period of Martian x l0 7 W m 3. Stevenson et al. have not calculated
volcanism provided that differentition is efficient lithosphere or crustal thicknesses, however. As for Mars,
enough. I discuss differentiation efficiency in the discus- I find that thermal variables are not much affected by
sion section. the differentiation of the mantle, except for mantle heat
Differentiation of the mantle and heat piping may flow.
have effects on the thermal evolution of the core. For For the same parameter values as for Mars, I find the
no heat piping, the effect is quite small. The core lithosphere thickness to become approximately equal to
remains liquid as is expected from the calculations of the potential crustal thickness of 165 km after 4.5 Ga if
Schubert and Spohn (1990). However, for very efficient the initial heat source density is 0.85 x 10 7 W m 3.
heat piping and for initial heat source densities less than Mercury's present mantle is then almost completely de-
1.28 x 10 - 7 W m -3, an inner core begins to grow after pleted of heat sources. The present-day lithosphere thick-
4. to 4.4 Ga of evolution. The energy made available ness is 122 km or 0.75 potential thickness for Q0 equal to
by the growth of the inner core (e.g., Stevenson et al. 1.7 x 10 -.7 W m -3. Mercury's mantle is more readily
1983, Schubert and Spohn 1990) may help to drive a depleted of heat sources as compared with Mars's mantle
magnetic dynamo with a magnetic moment about two if lithosphere growth entirely controls crustal growth be-
orders of magnitude smaller than the Earth's present cause Mercury's thinner mantle cools more rapidly. Onset
magnetic moment. The presently available magnetic field of inner core growth affects lithosphere growth as Fig. 2b
measurements suggest a dipole moment of not more shows. There is a kink in the lithosphere growth rate
than 5 x 10 - 4 times Earth's present magnetic moment, curve at the beginning of inner core freezeout. The kink is
however. caused by the latent heat and gravitational energy release
upon inner core growth which increases the heat flow
3.2. Mercury from the core.
Plots of basaltic lithosphere thickness, basaltic litho- 3.3. Venus
sphere growth rate, and Qm/Q versus time for a conduc-
tive lid model of Mercury are shown in Fig. 2. Other Plots of basaltic lithosphere thickness, basaltic litho-
results and a comparison with model Me4 of Stevenson sphere growth rate, and Qm/Q versus time for Venus con-
et al. (1983) corrected for misprints of present-day mantle ductive lid and heat-pipe model are shown in Fig. 3. Other
viscosity and Rayleigh number are given in Table IV. results and a comparison with model V1 of Stevenson et
Because Me4 used an upper boundary temperature for al. (1983) are given in Table V. Because Stevenson et al.
mantle convection of 1073 K, its results are comparable did not incorporate a lithosphere into their model and
D I F F E R E N T I A T I O N AND T H E R M A L E V O L U T I O N 229

" ~ fSO.O
,O4o
,O4o
0.8 ~j, '~ 5.0
on
¢n
o
,,~ 100.0 0.6
! N
0

0.4 \ "x ", ....... ~-?'2. . -


,#. 50.0 2.0
x

0.2 o
0 co.duet'/re l~d

i l i i I i . i i I t , i J I i . , r I I I
0.0 0.0 ~ ~ o.o [ I I I I i i i i I i i i i I i i i i I i i
t 000 2000 8000 4000 .o 0 1000 2000 3000 4000
~me (l~a)
c
1.0

0.9 ,,x

0.8

0.7

0.6

0.5

0.4:
• . ". %.
0.3

0.2

O.f

0.0 i i i , I , j ~ i I , , , i I i t i i I ,

1000 2000 3000 4000


t~me (~a)
FIG. 2. Results of thermal evolution calculations for a conductive lid model o f Mercury. For further explanation see Fig. 1.

because they used the Venusian surface temperature of and 5 km 3 year-J depending on heat source density. For
730 K as the upper boundary temperature for mantle con- efficient heat piping, the lithosphere may become as thick
vection, their results for Venus should be compared with as 170 km for an initial heat source density of 0.85 × 10 -7
the heat-pipe model with Q0 equal to 1.7 × 10 -7 W m -3. W m -3 and 135 km for an initial heat source density of 1.7
As for Mars and Mercury, I find that thermal variables × 10 -7 W m -3. These thicknesses are between about 0.3
are not much affected by the differentiation of the mantle, and 0.45 of the potential crustal thickness. Accordingly,
except for the mantle heat flow. Remaining differences present-day values of Qm/Q are between 0.55 and 0.7 and
are due mostly to their choice of upper boundary tempera- the present-day growth rates are between 10 and 12 km 3
ture for mantle convection of 730 K rather than the litho- year-~. Clearly, the relatively thin basaltic lithospheres
sphere boundary temperature of 1073 K that is used in the as compared with the potential crust thicknesses and the
present work. relatively small degrees of depletion, even for efficient
For the conductive lid model, I find maximum litho- heat piping, suggest that there is potential for present-day
sphere thicknesses of only up to approximately 40 km. volcanism on Venus.
This is about 10% of the potential crust thickness assum-
ing a magma enrichment factor A of 5. Accordingly, the 4. DISCUSSION AND CONCLUSIONS
ratio Qm/Q does not become smaller than 0.9, suggesting
a small degree of differentiation and depletion for the The model calculations suggest that continued volcanic
present Venus. Lithosphere growth rate for Venus varies activity for periods extending over billions-of-years is pos-
between 1 and 5 km 3 year-J and presently is between 3 sible if recycling of crust in excess of lithosphere helps to
230 TILMAN SPOHN

TABLE IV the surface, and AHf the latent heat. For Venus, I/in km 3
Results for Mercury Thermal History Models year-~ is approximately 2.25 times Fro, with F m in mW
m 2. To balance the present-day value of F m of approxi-
1.70 1.28 0.85 Me4
mately 40 mW m 2 (Table V) a surface extrusion rate of
F~ (mW m --~) 29.5 26.0 22. I -- 90 km 3 year ~ is required. This is approximately seven
F m (roW m 2) 10.0 8.4 7.1 17.3 times the maximum present-day volumetric basaltic litho-
F~ (mW m 2) 8.2 8.0 7.7 7.7 sphere growth rate calculated for my model (Fig. 3b) and
Tu (K) 1580. 1559. 1540. [637. would require substantial crustal recycling. The above
1"~,1 (K) 1860. 185 I. 1844. 1854.
~, (10 t6 m 2 s -I) 78. 121. 187. 25.
estimate of the required surface extrusion rate is a factor
Ra ( 104) 0.5 11.3 11.2 (1.6 of 3 smaller than a more pessimistic estimate of Turcotte
/ (kin) 121.7 138.5 161.3 -- (1989b) but still a factor of 45 larger than an estimate of
6 (km) 202.7 230.9 265.7 -- the volcanic flux of 2 km 3 year-~ by Grimm and Solomon
R i (kin) 140 I. 1413. 1424. 1408. (1987) from impact crater density. Similar values of volca-
Onset (]02 Ma) 7.0 7.0 6.5 6.11
nic flux between 0.4 and 11 km 3 year- i have recently been
N o t e . Present-day values of the conductive surface heat flow F~,. the estimated by Fegley and Prinn 11989) to be required
mantle heat flow Fm, the core heat flow F~, the upper mantle temperature for the maintenance of global H2SO 4 clouds. Equation
T,, the c o r e - m a n t l e boundary temperature 7"~m, the mantle viscosity t,, (7) neglects the possible contribution to the heat flux
the mantle Rayleigh n u m b e r Ra, the lithosphere thickness I. the upper of heat conduction through thin lithosphere in the
mantle thermal boundary layer thickness ~, and the inner core radius R i
immediate neighborhood of a volcano, however. The
are listed. " ' O n s e t " gives the time of onset of inner core growth. Me4 is
a comparable model of Stevenson et al. (1983). latter contribution may surmount the actual heat transfer
rate of volcanic effusion by an order of magnitude
(Phillips and Malin 1983) and may make the heat-pipe
model more acceptable for the present Venus. For Mars,
keep the planet alive. The model thus offers a simple
the present-day mantle heat flow of about 10 mW m 2
explanation for the continued volcanic activity apparently
for the heat-pipe model (Table II1) could be balanced
documented in the surface record of the planets. It is
by a surface extrusion rate equal to the volumetric
certainly true, however, that the detailed physics of man-
lithosphere growth rate of 5 km 3 year ~ that is
tle differentiation including magma generation, magma
representative of the calculated Martian history after
transport, and mantle mixing is much more complicated
1 Ga.
than my simple model suggests. And there may be other,
Because I assumed that differentiation efficiency is not
additional processes that may limit mantle differentiation
restricted by the declining mantle convective vigor, the
as, for instance, effects of chemical bouyancy in a mantle
calculated present day basaltic lithosphere thicknesses
that becomes layered during melting and melt extrac-
give upper bounds for crustal thicknesses. Lithosphere
tion.
thickness may also be thought to be overestimated from
4.1. Limits of the Model the model as one might argue that less efficient differentia-
tion would result in a hotter mantle with a thinner litho-
The models presented in the preceding sections are end- sphere (Schubert et al. 1990). This lithosphere would be
member models in several respects. The conductive lid layered with a basalt crustal layer on top and a mantle
model neglects the contribution of volcanic heat transfer layer below. The comparison with a model of Mars with-
and essentially assumes that the crust grows by basalt out differentiation in Table III, however, suggests that
addition to the lower crust. The heat-pipe model neglects lithosphere thicknesses are overestimated by less than
the effects of declining volcanic activity through crust and about 10%. (The reader should be reminded at this point
lithosphere thickening that tends to diminish the impor- that the lithosphere thicknesses calculated in this paper
tance of volcanic heat transfer with time. A simple esti- refer to the theological lithosphere. The thermal litho-
mate shows how the volcanic mass flow is related to the sphere which is often referred to as the lithosphere for the
heat transfer rate. To balance the mantle heat flow F m by Earth would actually comprise the theological lithosphere
volcanic heat transfer, a volumetric surface extrusion rate and the underlying thermal boundary layer of mantle con-
V of vection. As can be seen from Tables III to V the thermal
lithosphere is approximately two to three times thicker
1/= 47rRpFm (7) than the theological lithosphere.) The effect of mantle
pc(CcAT + AHf) differentiation on lithosphere thickness can be further elu-
cidated with the following simple calculation. Consider a
is required, with Pc the crust density, Cc the crust specific lithosphere of thickness I in which heat is generated at the
heat, AT the temperature difference between magma and rate Qc, the bottom is at temperature T~, and the top is at
DIFFERENTIATION AND THERMAL EVOLUTION 231

a ~b
,M 20,0
0,4 ~
° ) i',
~__~ t S O . O
oo
oo ~_ ~ 15.0
0.3
ca 100.0
~ a~ pipe
0 ~ lO.O
0.2

,~ 6o.o
o.,~ ~ ~.o conauoeive

0.0 o.o ~ 0.0 , , i i I i r , i I i i , i I ~ i i i I J i


tO00 2000 3000 4000 t 000 2000 8000 4000
~me 01") ~e (~)
C
1.0

0.9 - " ~ " ~

0.8

0.?

0.8

0.6 i i , , ~ i J i i i i i , , i , i i i i i i
0 1000 2000 3000 4000

FIG. 3. Results of thermal evolution calculations for Venus. For further explanation see Fig. 1.

temperature T~. Heat is transferred by heat conduction inventory and the present-day lithosphere thickness is
and the surface heat flow is Fc~. Heat flow at the bottom about 170 km (Table III). A breakdown of the contribu-
is assumed to be given by tions to the surface heat flow of about 27 mW m- 2 shows
that 10 mW m 2 (Table III) is convected from the mantle,
OT dl 2.5 mW m -2 is due to lithosphere growth, the equivalent
k-~zlZ= I = p m C m ( T u - TI)~-~ + fml (8) of 10 mW m - 2 is generated in the basaltic lithosphere, and
5 mW m 2 is due to lithosphere cooling, k(OT/Oz)l~=l is
[compare Eq. (1)]. Energy balance assuming stationary 12.5 mW m -2 in this case. If the mantle would not differen-
heat conduction gives an estimate of lithosphere thickness tiate at all during lithosphere growth, then the heat flux
of due to heat generation in the lithosphere would be reduced
by a factor of 5 to 2 mW m -2 and the convective mantle
heat flow would be 18 mW m - 2 . And k(OT/OZ)Iz=l would
2k(Tl- Ts)
then be equal to 20.5 mW m -z. The resulting difference
1~ F~s + k(OT/Oz)]z=t" (9) in lithosphere thickness, however, would be less than
20%. This difference could be further reduced by the
For the conductive lid model, Fcs is independent of effects of upward differentiation of heat sources within
mantle differentiation, and lithosphere thickness changes the lithosphere. As a matter of fact, it is possible that the
with k(OT/Oz)lz=l. Consider as an illustrative example the lithosphere thicknesses given in Tables III to V are even
conductive lid model of Mars with an initial heat source underestimates. Assume again that the mantle is depleted
density of 0.85 × 10 - 7 W m 3. The present day-mantle of 75% of its initial inventory of heat-producing elements
in this model is depleted of 75% of its initial heat source but that these heat sources are concentrated in a 40-km-
232 TILMAN SPOHN

TABLE V
R e s u l t s for V e n u s T h e r m a l H i s t o r y Models

Conductive lid Heat pipe

1.7(I 1.28 0.85 1.70 1.28 0.85 VI

Fc~ (mW m 2) 62.2 50.6 38.5 18.6 16.2 13.5 --


F m (mW m 2) 58.2 46.7 34.7 45.7 36.2 27.0 59.9
F c (mW m 2) 11.5 10.8 10.3 11.0 10.6 12.0 11.0
Tu (K) 1815. 1781. 1736. 1759. 1724. 1682. 1722.
Tcrn (K) 3051. 3013. 2971. 298 I. 2947. 2924. 2933.
u (1016 m 2 s 11 I. 2. 4. 3. 5. I1. 5.
Ra (108) 4. 2. I. 2. I. 0.5 2.
/ (kin) 22.6 27.9 37.0 133. 149. 174. --
(km) 51.0 60.6 76.4 60.0 73.1 90.3 --
R i (km) . . . . . 681.
Onset (Ga) . . . . . 4.0

Note. Present-day values of the conductive surface heat flow Fc~, the mantle heat flow F m,
the core heat flow F c , the upper mantle temperature T~, the core-mantle boundary temperature
T~m, the mantle viscosity ~, the mantle Rayleigh number Ra, the lithosphere thickness I, the
upper mantle thermal boundary layer thickness 8, and the inner core radius R i are listed. " O n s e t "
gives the time of onset of inner core growth. V I is a comparable model of Stevenson e t a l .
( 19831.

thick crust. An estimate similar to the derivation of Eq. et al. (1990) give lithosphere thicknesses that in some
(9) shows that the calculated present-day lithosphere cases are much larger than the thicknesses calculated
thickness is then underestimated by 20%. The magma here. In estimating their lithosphere thicknesses they use
enrichment factor would have to be larger than 20 in this a version of Eq. (9) in which Fc~ is absent and in which
case, however. k(OT/Oz)l~_l equals F m, thereby neglecting the effects of
For the heat-pipe model, Fml is zero and present-day crustal heat sources, lithosphere growth, volcanic heat
lithosphere thickness depends mainly on lithosphere heat piping, and lithosphere cooling altogether.
source density. Fc~ in this case is not independent of differ- Crust thicknesses are much less reliably predicted by
entiation but increases with lithosphere heat source con- the calculated basaltic lithosphere thicknesses because
tent. Additional model calculations have shown that k(OT/ crust formation depends on the ratio between mantle tem-
Oz)l~_l changes comparatively little with degree of differ- perature and solidus temperature and, most likely, de-
entiation since dl/dt is almost independent of differentia- pends on mantle convection speed. The convection speed
tion. Figures lb, 2b, and 3b also show lithosphere growth matters if crustal fractionation occurs mostly in the upper
rate to depend rather little on heat source density. Conse- mantle like in the present Earth. Any crustal material
quently, and contrary to the conductive lid case, present- removed from the upper mantle must then be efficiently
day lithosphere thickness decreases with degree of differ- replenished to keep differentiation going. It is possible
entiation for the heat-pipe model. According to the addi- that magma is efficiently transferred to the surface from
tional calculations, the difference in lithosphere thickness the lower mantle, for instance, by magmons (e.g., McKen-
between a fully differentiated Mars and an undifferenti- zie 1984, Scott and Stevenson 1984, 1986). But magma
ated planet could be about 30% of the lithosphere thick- transport from the lower mantle may be limited by the
ness of the former. Stolper effect (e.g., Stolper et al. 1981), the possible more
Lithosphere thickness also depends on the chosen value rapid increase with pressure of magma density as com-
of the boundary temperature. A 100 K difference in the pared with solid mantle density at pressures above 6 to 10
assumed value of T~translates into a difference in present- GPa caused by the larger compressibility of the magma.
day lithosphere thickness of about 10%. However, I con-
4.2. Convective Vigor and Crustal Thicknesses
clude that lithosphere thickness does not depend strongly
on differentiation. If heat is transferred through the litho- To relate the mantle differentiation rate to the mantle
sphere mostly by conduction then lithosphere thickness convection speed one may write following Schubert et al.
at any given time increases with increasing differentiation. (1990)
This increase may be counteracted by the effects of heat
piping on lithosphere thickness and by the effects of up- 1 dQm u
ward differentiation of heat sources in the crust. Schubert Qm d~ - )t - X~Te. (10)
D I F F E R E N T I A T I O N AND T H E R M A L E V O L U T I O N 233

In (10), X is a parameter measuring the efficiency of dD/dt as a function of time for Mars, Mercury, and Venus.
magma generation and u is the representative mantle con- Note that dD/dt is actually independent of the convection
vection speed. X is essentially the ratio between the mantle speed scale. For negative values of the logarithm of the
convection turnover time and the characteristic time for ratio between dl/dt and dD/dt, the lithosphere is com-
crustal fractionation. X can also be taken as a measure of posed entirely of basalt and the crust is equal to the litho-
the percentage of melt in the mantle. A value of X of 1, sphere. Extra crustal material is recycled with the mantle
for example, suggests that a l/e fraction of the potential in this regime. For positive values of the logarithm of
thickness of the crust is fractionated in one convection the ratio between dl/dt and dD/dt, a layered lithosphere
overturn time. This essentially requires that the equiva- evolves with a crustal layer on top and a mantle layer
lent of the potential crust in the mantle is molten, approxi- below. A value of the logarithm of the ratio between dl/
mately 15% of the mantle in my model. X is assumed dt and dD/dt of 3, for instance, indicates that the crust
constant although it should actually be a function of time grows I m for every kilometer of lithosphere growth.
through its dependence on temperature and melting tem- There may, of course, be volcanism in this regime. How-
perature. Note also that the melting temperature of the ever, volcanic activity is expected to decrease with de-
mantle should increase with time as the mantle becomes creasing relative values of dD/dt. The importance of vol-
more depleted during differentiation. Differentiating (4) canic heat transfer should decrease accordingly. Crust
with respect to time after replacing I with crustal thickness thicknesses with crustal growth limited by lithosphere
D, one finds using (10) growth and depending on the declining mantle convective
vigor can be calculated by integrating dl/dt over time for
dl/dt <- dD/dt, integrating dD/dt over time for dD/dt <-
dD (Dpo t -- D)XU dl/dt, and adding the two integrals. This simple procedure
(11)
_

dt Rp. is possible because the model calculations discussed in


the previous section have shown that mantle Rayleigh
number and lithosphere thickness do not depend strongly
Equation (l l) gives a potential crustal growth rate for
on mantle differentiation even in the extreme case where
crust growing unimpeded by lithosphere thickness and
the crust was assumed to be always as thick as the litho-
growth rate. Boundary layer theory (e.g., Turcotte and
sphere.
Oxburgh 1967) suggests that u is related to the ratio be-
The results in Fig. 4 show that, in general, dD/dt is
tween the Rayleigh number Ra and the critical Rayleigh
larger than dl/dt in the early evolution of a terrestrial
number Rac~ according to
planet. Thereafter, the evolution of the ratio between dl/
dt and dD/dt depends on planet size, the initial concentra-
[ Ra ]2~ tion of heat sources, the importance of heat piping, and,
u = u0 - - (12) of course, the chosen value of magma generation effi-
~,Racr)
ciency X. For Mercury and Mars, dl/dt is larger than dD/
dt during most of the evolution if the magma generation
where u0 is a convection speed scale. With (1) and (11) efficiency of the present Earth is representative of the
one can calculate the ratio between lithosphere growth entire evolutions of these planets. For Venus, on the con-
rate dl/dt and potential crustal growth rate dD/dt assum- trary, dD/dt is larger than dl/dt in these circumstances
ing appropriate values for X. A representative value for X even for the present day.
is difficult to pick, however. Schubert et al. (1990) use For Mars (Fig. 4a), for very efficient heat piping and for
values between 10 -3 and 10 -2. Using the present Earth an initial heat source density of 0.85 × l 0 - 7 W m - 3 , the
as a guide, I find a value o f x of 4 x 10 -3. The latter value heat source density in accord with the data from the SNC
was derived using a present-day value of magma flux meteorites, the ratio between dl/dt and dD/dt increases
of 17 km 3 year ~, representative of the magma flux at rapidly. After the first few 100 Ma, the declining convec-
midoceanic ridges (Turcotte 1989b), a convection speed tive vigor has reduced the crustal growth rate to less than
scale of 10 cm year -], and a Rayleigh number of 10~, 10% of the lithosphere growth rate. The heat-pipe model
appropriate for Earth whole mantle convection. Continen- may then no longer be strictly valid. From a ratio of
tal flood basaits suggest a smaller magma flux of only 10 between dl/dt and dD/dt and the basaltic lithosphere
about 5 km 3 year-J (McLean 1985). Of course, the above growth rate of Fig. lb, it is calculated by using Eq. (7)
value of X may have been larger for the early mantles of that only 10% of the mantle heat flow given in Table III
the terrestrial planets as their degree of partial melting can then be balanced by volcanic heat transfer. If early
may have been considerably larger than that of the present Mars was 100 times more efficient at crustal fractionation
Earth's mantle. than the present Earth, then dD/dt may have been larger
Figure 4 shows the logarithm of the ratio between litho- than dl/dt for almost 2 Ga and volcanic activity with an
sphere growth rate dl/dt and potential crustal growth rate extrusion rate decreasing from the value representative
234 TILMAN SPOHN

t-
a ~'b
5.0 ~.~ 6.0

o 0
4.0 5.0

3.0 4.0

2.0
...-_.2._2:::7"57::2''57~2-" " ""
0 0 3.0

1.0
2•0

t"
0.0
1.0

-I.0 0
I I I I I I I I i I I I I I I i i I I i 0.0 I I L I l l
, I , , , , I , I , '40100 ' '
0 1000 2000 3000 4000 0 1000 2000 3000
~,~ (~)
L~

0 0.0

-1.0

0
~ -2.0

"~ -3.0
0
N I I I I ~ I I I I I I I [ L I I I I I I
1000 2000 3000 4000
~e (~)

FIG. 4. Lithosphere growth rate dl/dt divided by the potential growth rate dD / dt of a crust growing ummpeded by lithosphere growth rate as
a function of time for (a) Mars, (b) Mercury, and (c) Venus. For further explanation see Fig. 1.

of continental flood basalts on Earth to one-tenth of that effusion rate calculated from these data is about 6 x 10 z
value may have continued for another billion years• (Sim- km 3 year ~, about two orders of magnitude less than the
ply multiply the results in Fig. 4a by 100 to obtain the ratio volumetric crust production rate for the Mars models in
between dl/dt and dD/dt for a 100 times larger value of Fig. lb. The results for the conductive lid model with an
X.) This would require an about 15% partially molten initial heat source density of 0.85 x 10 -7 W m 3 shown
mantle of Mars for an unlikely 2 Ga, however. in Figs. lb and 4a combined suggest a crustal growth rate
If strong heat piping was restricted to the very early decreasing from 40 x 10 -2 km 3 year -1 at 1 Ga (4 km 3
evolution of Mars, however, then the results of the con- year ~from Fig. lb multiplied by 10 -2 from Fig. 4a) to 2
ductive lid model should be more representative of the x 10 2 km 3 year J a t 4 G a ( 6 k m 3 y e a r -j from Fig. lb
later evolution of the planet• The results for this model multiplied by 3 x 10 -3 from Fig. 4a), in nice agreement
suggest that crust growth and volcanic activity for almost with the above data. The evidence for continued volca-
the entire thermal history of Mars are possible with the nism in the surface record of Mars for at least 3.5 Ga
fractionation efficiency of the present Earth. Greeley (BVSP 1981, N e u k u m and Hiller 1981, Greeley 1987,
(1987) has recently estimated volumes of volcanic extru- McGill 1989) therefore argues against an extended period
sion o v e r the Martian history from geological maps based of very efficient heat piping for that planet. In any case,
on the Viking orbiter data. Estimated volumes of volcanic Mars may have a substantial crust as a comparison of Fig.
material decrease from about 72 x l 0 6 km 3 in the early 4a with Fig. la suggests• For efficient heat piping, the
Hesperian to about 21 x 10 6 km 3 in the Middle Amazonian crust may grow to 50-100 km of thickness in the first few
to about 5 x 10 6 km 3 in the Late Amazonian. The average hundred million years while the same thickness may be
DIFFERENTIATION AND THERMAL EVOLUTION 235

reached for a conductive lid model in less than 3 Ga. which may consist largely of basalt. However, as Turcotte
In both cases, Mars mantle would be depleted of about (1989b) has noted, the thickness of the Venusian crust may
20-40% of its initial heat source inventory. The present- be limited by the basalt-to-eclogite phase transformation:
day lithosphere thickness will be around 200 km as is The dense eclogite phase that is likely to form in the crust
required from tectonic modeling reviewed in Schubert et below 60 to 80 km depth is gravitationally unstable with
al. (1990) if the initial heat source density is 0.85 × 10 7 respect to the mantle below. When the crust is sufficiently
W m -3 as calculated from the SNC data. thick, crustal delamination can occur. A thick basaltic
For Mercury, the ratio between d l / d t and d D / d t in Fig. crust has been suggested by Anderson (1980) to explain
4b increases rapidly and differentiation of a crust compris- the intrinsically smaller density of Venus as compared
ing a substantial fraction of the lithosphere would require with Earth. A thick lithosphere is required by the model-
magma generation efficiencies much larger than the pres- ing of Kiefer et al. (1986) of the Venusian gravity field. A
ent efficiency of the Earth's mantle. The rapid increase in thick bouyant lithosphere on Venus helps to explain the
the ratio between d l / d t and d D / d t even for the conductive high topography and the associated large gravity anoma-
lid model argues against heat piping as an important mech- lies on Venus that are difficult to reconcile with a thin
anism of heat removal from the interior. The absence of lithosphere (Phillips et al. 1981, Phillips and Malin 1983,
prominent volcanoes on Mercury also suggests that heat Bowin 1983, Bowin et al. 1985, Turcotte 1989b). The
piping could not have been very important. Moreover, the calculated basaltic lithosphere thicknesses for the heat-
existence of a magnetic field argues against efficient heat pipe models are close to the required crustal thicknesses
piping because heat piping would cool the planet too rap- discussed in Turcotte (1989b) to explain the topography
idly. Efficient cooling of the mantle would cause solidifi- and gravity anomalies observed for Ishtar Terra and Beta
cation of the core and would not allow an active dynamo Regio. A thick crust does not require the mantle to be
in the core. Additional calculations have shown that heat almost completely depleted of heat sources, however, as
piping would cause the core to freeze even for an initial Phillips et al. (1981) and Phillips and Malin (1983) have
sulfur content of 5%. However, it has been suggested that argued, if the mantle heat can be focused into heat pipes.
the volcanic plains on Mercury may be as young as the
lunar maria or even somewhat younger (e.g., BVSP 1981), ACKNOWLEDGMENTS
indicating a period of some volcanism of about 1 Ga since
the accretion of the planet. According to the present calcu- I thankG. Schubertfor helpfuland stimulatingdiscussions. This paper
lations and assuming a fractionation efficiency representa- was completed during a sabbatical visit to UCLA. This research was
supported by DFG and by the Fullbright Foundation.
tive of the present Earth the potential crustal growth rate
was about 1/100th of the lithosphere growth rate after I
REFERENCES
Ga (Fig. 4b). According to Fig. 2b, this is equivalent to a
magma production rate of about 10 -2 km 3 year -I, still ANDERSON, D. 1980. Tectonics and composition of Venus. Geophys.
approximately equal to the present magma generation rate Res. Lett. 7, 101-102.
of the Hawaiian swell (Turcotte 1989b). A present-day Basaltic VolcanismStudy Project 1981.Basaltic Volcanism on the Ter-
crust thickness of a few tens of kilometers from an integra- restrial Planets. Pergamon, New York.
tion of d D / d t over time is calculated for Mercury and the BOWIN,C. 1983. Gravity, topography,and crustal evolutionof Venus.
Icarus 56, 345-371.
interior may be depleted of about 10% of its initial heat
source inventory. BOWIN,C., G. ABERS,ANDL. SHURE1985.Gravityof Venus at constant
altitude and comparison with Earth. J. Geophys. Res. 90, C757-C770.
For Venus, the results shown in Fig. 4c suggest that
CARSLAW, H. S., AND J. C. JAEGER 1954. Conduction o f Heat in Solids.
crust fractionation may today still be limited by litho- Oxford Univ. Press (Clarendon), London.
sphere growth and, together with the small degree of man- CHRJSTENSEN,U. 1989. Mixing by time-dependent convection. Earth
tie depletion shown in Fig. 3c, the results suggest that Planet. Sci. Lett. 95, 382-394.
there still is a high potential for a volcanically active COOK,F. A., ANDD. L. TURCOTTE1981.Parameterizedconvectionand
planet. The rate of magma generation of the present Venus the thermal history of the Earth. Tectonophysics 75, 1-17.
may be a quarter to half the value of that of the present DEPAOLO, D. J. 1981. Nd isotopic studies: Some new perspectives on
Earth midoceanic ridges. This result is incompatible with earth structure and evolution. EOS Trans. A G U 62, 137-140.
the estimated values of the Venusian volcanic flux from ESpOSITO, L. W. 1984. Sulfur dioxide: Episodic injection shows evi-
crater statistics and maintenance of n 2 s o 4 clouds quoted dence for active Venus volcanism. Science 223, 1072-1074.
above, however, unless crust grows mostly by volcanic FEGLEY, B., AND R. G. PRINN 1989. Estimation of the rate of volcanism
on Venus from reaction rate measurements. Nature 337, 55-58.
intrusion and surface volcanism is restricted topologically
GREELEY, R. 1987. Release of juvenile water on Mars: Estimated
to a few volcanic sites. The thermal characteristics of such amounts and timing associated with volcanism. Science 236,
a model may still be close to that of the heat-pipe model. 1653-1654.
Venus may have a 100-km-thick or thicker lithosphere GRIMM, R. E., AND S. C. SOLOMON 1987. Limits on modes oflithosphe-
236 TILMAN SPOHN

ric heat transport on Venus from impact crater density. Geophys. Res. SCHUBERT, G., S. C. SOLOMON, D. L. TURCOTTE, M. J. DRAKE, AND
Lett. 13, 14-17. N. H. SLEEP 1991. Origin and Thermal Evolution of Marx. In Mars
HEAD, J. W. 1986. lshtar Terra Venus: A simple model for large-scale (H. Kieffer, B. Jakosky, C. Snyder, and M. S. Matthews, Eds.), in
tectonic convergence, crustal thickness, and possible delamination. press, Univ. of Arizona Press, Tucson.
LPSC X V l l Abstracts, 323-324. SCHUBERT, G., AND T. SPOHN 1990. Thermal history of Mars and the
HEAD, J. W., AND L. WILSON 1986. Volcanic provinces and landforms sulfur content of its core. J. Geophys. Res., 95, 14095-141(14.
on Venus: Theory, predictions, and observations. J. Geophys. Res. SCHUBERT, G., T. SPOHN, AND R. T. REYNOLDS 1986. Thermal histor-
91, 9407-9446. ies, compositions, and internal structures of the moons of the solar
HOFFMAN N. R. A.. AND D. P. MCKENzIE 1985. The destruction of system. In Satellites (J. A. Burns and M. Matthews, Eds.), pp.
geochemical heterogeneities by differential fluid motions during man- 224-292. Univ. of Arizona Press, Tucson.
tle convection. Geophys. J. R. Astron. Soe. 82, 163-206. SCOTT, D. R., AND D. J. STEVENSON 1984. Magma solitons. Geophys.
KIEFER, W. S., M. A. R1CHARDS, B. H. HAGER, AND B. G. BILLS 1986. Res. Lett. 11, 1161-1164.
A dynamic model of Venus's gravity field. Geophys. Rex. Lett. 13,
SCOTT, D. R,, AND D. J. STEVENSON 1986. Magma ascent by porous
14-17.
flow. J. Geophys. Res. 91, 9283-9296.
LAUL J. C., M. R. SMITH, H. W,~NKE, E. JAGOUTZ, G. DREIBUS, H.
SHARPE, H. N., AND W. R. PELTIER 1979. A thermal history for the
PALME, B. SPETTEL, A. BURGHELE, M. E. LIPSCHUTZ, AND R. M.
Earth with parameterized convection. Geophys. J. R. Astrong. Soc.
VERKOUTEREN 1986. Chemical systematics of the Shergotty meteorite
59, 171-203.
and the composition of its parent body (Mars). Geochim. Coxmochim.
Acta 50, 909-926. SPOHN, T. 1984. Die thermische Evolution der Erde. J. Geophys. 54,
77-96.
McGILL, D. L. 1989. Geologic evidence supporting an endogenic origin
for the Martian crustal dichotomy. LPSC X X Abstracts, 667-668. SPOHN, Y., AND G. SCHUBERT 1982a. Convective thinning of the litho-
MCKENZlE D. P. 1985. The generation and compaction of partial melts. sphere: A mechanism for continental rifting. J. Geophys. Res. 87,
J. Petrol. 25, 713-765. 4669-4681.
MCLEAN, D. M. 1985. Mantle degassing induced dead ocean in the SPOHN, T., AND G. SCHUBERT 1982b. Modes of mantle convection and
Cretaceous-Tertiary transition. In The Carbon Cycle and Atmo- the removal of heat from the Earth interior. J. Geophys. Res. 87,
spheric C02: Natural Variationx Archaean to Present (E. T. Sund- 4682-4696.
quist and W. S. Broecker, Eds.), Geophys. Monogr. Ser., Vol 32, pp. SPOHN, T., AND G. SCHUBERT 1983. Convective thinning of the litho-
493-503. AGU, Washington, D.C. sphere: A mechanism for continental rifting and mid-plate volcanism
NEUKUM, G., AND K. HILLER 1981. Martian ages. J. Geophy~s. Res. 86, on Earth, Venus, and Mars. Tectonophysivs 94, 67-90.
3097-3121. STEVENSON. D. J., AND S. C. MCNAMARA 1988. Background heatflow
O'RHLLY, T. C., AND G. F. DAWES 1981. Magma transport of heat on on hotspot planets: 1o and Venus. Geophys. Res. Lett. 15, 1455-1458.
1o: A mechanism allowing a thick lithosphere. Geophys. Rex. Lett. 8, STEVENSON, D. J., T. SPOHN, AND G. SCHUBERT 1983. Magnetism and
313-316.
thermal evolution of the terrestrial planets. Icarus 54, 466-489.
PELTIER, W. R. 1989. Models of the thermal history of the Earth. EOS
STEVENSON, D. J., AND J. S. TURNER 1979. Fluid models of mantle
Trans. AGU 70, 1000.
convection. In The Earth, Its Origin, Structure, and Evolution (M.
PHILLIPS, R. G., W. M. KAULA, G. E. McGILL, AND M. C. MALIN McElhinny, Ed.), pp. 227-263. Academic Press, New York.
1981. Tectonics and evolution of Venus. Science 212, 879-887.
STOLPER, E. 1)., WALKER. B. H. HAGER. AND J. F. HAYS 1981. Melt
PHILHPS, R. G., AND M. C. MAHN 1983. The interior of Venus and
segregation from partially molten source regions: The importance of
tectonic implications. In Venus (D. M. Hunten, L. Colin, T. M. Do-
melt density and source region size. J. Geophys. Res. 91,6261-6271.
nahue, and V. I. Moroz, Eds.), pp. 159-214. Univ. of Arizona Press.
Tucson. SURKOV, Yu. A. 1983. Studies of Venus rocks by Veneras 8, 9, and 10.
In Venus (D. M. Hunten, L. Colin, T. M. Donahue, and V. I. Moroz,
PHILPOTTS, J. A., AND C. C. SCHNETZLER 1970. Phenocryst-matrix
Eds.), pp. 154-158. Univ. of Arizona Press, Tucson.
partition coefficients for K, Rb, St, and Ba, with applications to
anorthosite and basalt genesis. Cosmochim. Geochim. Acta 34, TRFIMAN, A. H.. M. J. DRAKE, M.-J. JANSSENS, R. WOLD, AND M.
307-322. EBIHARA 1986. Core formation in the Earth and shergottite parent
SCARF, F. L., AND C. T. RUSSEL 1983. Lightning measurements from body (SPB): Chemical evidence from basalts. Geochim. Cosmochim.
the Pioneer Venus Orbiter. Geophys. Res. Lett. 10, 1192-1195. Acta 50, 11171-1091.
SCHABER, G. G. 1982. Venus: Limited extension and volcanism along TURCOTTE D. L. 1988. A heat pipe mechanism for volcanism and tecton-
zones of lithospheric weakness. Geophys. Res. Lett. 9, 499-502. ics on Venus. LPSC X1X Abstracts. 1207-1208.
SCHUBERT, G., P. CASSEN, AND R. E. YOUNG 1979. Subsolidus convec- TURCOTTE, D. L. 1989a. Thermal evolution of Mars and Venus including
tive cooling histories of the terrestrial planets. Icarus 38, 192-211. irreversible fractionation. LPSC X X Abstracts, 1138-1139.
SCHUBERT, G., M. N. ROSS, D. J. STEVENSON, AND T. SPOHN 1988. TURCOTTE, D. L. 1989b. A heat pipe mechanism for volcanism and
Mercury's thermal history and the generation of its magnetic field. In tectonics on Venus. J. Geophys. Res. 94, 2779-2785.
Mercury (F. Vilas, C. R. Chapman, and M. S. Matthews, Eds.), pp. TURCOTTE D. L., AND E. R. OXBURGH 1967. Finite amplitude convec-
429-460. Univ. of Arizona Press, Tucson. tive cells and continental drift. J. Fluid Mech. 28, 29-42.

You might also like