Geochimica et Cosmochimica Acta, Vol. 65, No. 2, pp. 311–321, 2001
Copyright © 2001 Elsevier Science Ltd
Printed in the USA. All rights reserved
0016-7037/01 $20.00 ⫹ .00
Pergamon
PII S0016-7037(00)00528-7
The temperature of formation of carbonate in Martian meteorite ALH84001:
Constraints from cation diffusion
A. J. R. KENT,1,*,† I. D. HUTCHEON,1 F. J. RYERSON,2 and D. L. PHINNEY1
1
Analytical and Nuclear Chemistry Division, Lawrence Livermore National Laboratory, Livermore CA 94550, USA
Institute of Geophysics and Planetary Physics, Lawrence Livermore National Laboratory, Livermore CA 94550, USA
2
(Received December 8, 1999; accepted in revised form July 18, 2000)
Abstract—We have measured the rates of chemical diffusion of Mg in calcite and Ca in magnesite and used
these new data to constrain the formation temperature and thermal history of carbonates in the Martian
meteorite ALH84001. Our data have been collected at lower temperatures than in previous studies and provide
improved constraints on carbonate formation during relatively low-temperature processes (ⱕ400°C). Measured log D0 values for chemical diffusion of Mg in calcite and Ca in magnesite are ⫺16.0 ⫾ 1.1 and ⫺7.8 ⫾
4.3 m2/s and the activation energies (EA) are 76 ⫾ 16 and 214 ⫾ 60 kJ/mol, respectively. Measured diffusion
rates of Mg in calcite at temperatures between 400 and 550°C are substantially faster than expected from
extrapolation of existing higher-temperature data, suggesting that different mechanisms may govern diffusion
of Mg at temperatures above and below ⬃550°C.
We have used these data to constrain thermal histories which will allow the ⬃1 ␮m variations in Ca-Mg
composition in ALH84001 carbonates to survive homogenization by diffusion. Our results are generally
consistent with models for formation of carbonates in ALH84001 at low temperatures. For initial cooling rates
of between ⬃10⫺1 and 103°/Ma our results demonstrate that carbonates formed at temperatures no higher than
400°C and most probably less than 200°C. This range of cooling rates is similar to those observed within the
terrestrial crust, and suggests that formation of the carbonates by igneous, metamorphic or hydrothermal (or
other) processes in the Martian crust most plausibly occurred at temperatures below 200 to 400°C. Models that
suggest ALH84001 carbonates formed during a Martian impact event are also constrained by our data. The
thermal histories of terrestrial impact structures suggest that cooling rates sufficiently rapid to allow preservation of the observed carbonate zoning at formation temperatures in excess of 600°C (⬎⬃107°C/Ma) occur
only within the uppermost, melt-rich portions of an impact structure. Material deeper within the impact
structure (where cooling is dominated by uplifted crustal material and where much of the formation of
hydrothermal minerals is concentrated) cools much slower, typically at rates of ⬃102 to 103°/Ma. Carbonates
formed within this region would also only preserve ⬃1 ␮m compositional zoning at formation temperatures
of less than ⬃200 to 400°C. Copyright © 2001 Elsevier Science Ltd
from ⬃0 to ⬎650°C (see summary in Table 1). In addition,
many temperature estimates are model-dependent and rely on
assumptions about chemical and isotopic equilibrium and
closed versus open-system behaviour.
In this study we have measured the rates of Mg and Ca
diffusion in carbonate minerals and used these new data to
constrain the formation temperature and thermal history of
ALH84001carbonates. The carbonate “rosettes” in ALH84001
that host the putative biomarkers are chemically zoned and
exhibit compositional heterogeneity on the submicron scale
(e.g., Cooney et al., 1999; Schwandt et al., 1999). Zonation is
evident as Mg-rich rims grading to Ca- and Fe-rich cores, and
also by more marked changes in Mg, Fe and Ca concentrations
over shorter length scales (e.g., Harvey and McSween Jr., 1996;
McKay and Lofgren, 1997; Schwandt et al., 1999; Scott et al.,
1998; Treiman, 1995; Treiman and Romanek, 1998). Short
length-scale compositional variations occur adjacent to the
Mg-rich rims of the rosettes as well as in intermediate locations
and represent dramatic compositional gradients—with changes
exceeding 10 mol cation % over distances of ⬃1 ␮m (Scott et
al., 1997; Scott et al., 1998). Recent high-resolution field emission SEM images and micro-Raman studies confirm the indications from electron probe analyses that carbonates are heterogeneous over very short length scales (Cooney et al., 1999;
1. INTRODUCTION
An important test of the hypothesis that Martian meteorite
ALH84001 contains the fossil remnants of an ancient Martian
biota (McKay et al., 1996) is posed by the formation temperature and thermal history of the carbonate minerals associated
with the proposed biomarkers. Terrestrial microbial life is
known to occur at temperatures between ⫺10 to 113°C (Golden et al., 2000 and references therein); if it can be demonstrated
that carbonates in ALH84001 formed at temperatures outside
this range, then it is unlikely that these minerals formed in
association with a terrestrial-like biota. However, before the
temperature of carbonate formation can to be used to assess the
possible presence of ancient Martian life in ALH84001, there is
a need for robust and unequivocal estimates of that temperature. Unfortunately, there is little present agreement on the
temperature of formation of carbonate minerals in ALH84001.
Current estimates, based on a variety of techniques including
textural interpretation, phase equilibrium, stable isotope geothermometry, paleomagnetism, and diffusion modeling range
*Author to whom correspondence should be addressed ([email protected].
dk).
†
Present address: Danish Lithosphere Centre, Øster Voldgade 10, 1350
Copenhagen K, Denmark.
311
312
A. J. R. Kent et al.
Table 1. Estimates of the formation temperatures of carbonates in ALH84001.
Study and technique
Estimated formation temperature
Romanek et al. (1994)
O and C isotope measurement of bulk
carbonate sample
Mittlefehldt (1994)
Chemical composition of carbonates and
equilibrium thermodynamic arguments
Harvey and McSween (1996)
Chemical composition of carbonates and
equilibrium thermodynamic arguments
Hutchins and Jakosky (1997)
Re-evaulation of data from
Romaneck et al. (1994)
Kirschvink et al. (1997)
paleomagnetic studies
Scott et al. (1997); Scott et al. (1998)
Petrographic observations
Valley et al. (1997)
Ion-probe O and C isotope measurements
and mineralogical arguments
Saxton et al. (1998)
Ion-probe O isotope measurements
0–80°C
Leshin et al. (1998)
Ion-probe O isotope measurements
1. ⬃125°C with fluctuations up to
250°C (open system)
2. ⬎500°C (closed system)
Fisler and Cygan (1998)
Modelling of cation diffusion
Golden et al. (2000)
Experimental preparation of analogous
carbonate globules
This Study
Modelling of cation diffusion
⬍300°C
Mechanism of carbonate formation
Open system precipitation of carbonate
from aqueous fluids during
hydrothermal circulation
Carbonates formed during hydrothermal
circulation
700°C
⬃650°C
Impact-related metasomatism of ultramafic
rocks by CO2-rich fluids
40–250°C
⬍325°C
High temperature (⬍⬃600°C?)
⬍300°C
Initially ⬍⬃400°C, final stages
of precipitation ⬍ 150°C
⬃150°C
Crystallization of carbonate from
shock-melted material
Low-temperature non-equilibrium
crystallization
Precipitation from hydrothermal fluids at
temperatures ⬍⬃400°C, the final stages
of carbonate deposition may have
occurred at ⬍⬃150°C
1. Low-temperature precipitation from
open system with variable temperatures
2. High-temperature closed system
precipitation from limited amount of
CO2-rich fluid
Precipitation from Mg-rich supersaturated
solutions
⬍200°C over the range of
terrestrial crustal cooling rates
Schwandt et al., 1999) and demonstrate that the variations in
composition measured by the electron probe are not due to
small inclusions of magnetite or other minerals.
Given an appropriate thermal history (i.e., if temperatures are
hot enough for long enough), cation diffusion will act to homogenize chemical variations (Fisler and Cygan, 1998; Sheng
et al., 1992). Thus the observation that compositional variations
in ALH84001 carbonates occur at the ⬃1 ␮m scale can be used
in combination with the rates of cation diffusion to constrain
the formation temperature and thermal history of ALH84001
carbonates.
We have experimentally measured the rates of chemical
diffusion of Ca in magnesite and Mg in calcite and the rate of
self-diffusion of Mg in magnesite. Although our approach is
similar to that used in a recent study by Fisler and Cygan
(1998), one important difference is that our experimental design enables us to measure much slower diffusion rates, as low
as 10⫺25 m2/s, and to perform experiments at temperatures as
low as 400°C (compared to 550°C in previous studies; Fisler
and Cygan, 1998; 1999). This approach reduces errors associated with the down-temperature extrapolation required for calculations in the 100 to 400°C temperature interval and is
particularly important for calculations involving Mg diffusion
in carbonates, as this is unexpectedly fast at temperatures below
⬃550°C. Overall, our results provide a substantial refinement
of the temperature constraints that can be placed on the formation of carbonates in ALH84001.
We also note that although ALH84001 carbonates also contain a substantial amount of Fe, experimental difficulties (predominantly the instability of the Fe-rich carbonates siderite and
ankerite under our experimental conditions) prevented us from
obtaining data on cation diffusion rates in Fe-rich carbonates.
These data are a future goal for our experimental program.
To avoid confusion, we have applied the following usage in
our discussion of diffusion in carbonates (cf. Watson, 1994):
Chemical diffusion refers to that which occurs in the presence
of a chemical gradient, whereas self-diffusion is that which
occurs without a chemical gradient.
2. EXPERIMENTAL METHOD
The chemical diffusion rates of Ca and Mg in carbonates were
measured using a simple powder source technique similar to that
described by Cherniak (1997; 1998). Selected cleavage fragments of
natural calcite (CH-1 from Chihuahua, Mexico) and magnesite (from
Brumado, Brazil) were placed in platinum capsules together with dried
analytical-grade powder of the appropriate composition (CaCO3 for
magnesite, MgO and MgCO3 for calcite). The chemical compositions
of magnesite and calcite used for our experiments are given in Table 2.
All diffusion measurements were made using natural (101̄1) cleavage
surfaces of calcite and magnetite. Carbonate cleavage fragments were
not preannealed before experiments, and although no specific effort
was made to determine if crystallographic orientation has an observable
Formation temperature of ALH84001 carbonate
Table 2. Compositions of carbonate minerals used for diffusion
experiments.
Chihuahua
Calcite
CH-1
MgO
CaO
MnO
FeO
CO2
Number of
analyses
0.20
57.52
0.19
0.10
42.00
21
⫾1␴
0.01
0.46
0.02
0.03
0.49
Brumado
Magnesite
BR-1
41.55
0.30
0.19
0.19
57.77
27
⫾1␴
0.42
0.02
0.02
0.02
0.43
Compositions given in oxide wt.%. Carbonate compositions measured with a JEOL-733 electron microprobe, using a 10 nA, 20 ␮m
diameter electron beam and an accelerating voltage of 15 kV. Count
times for all elements were 60 s. Natural carbonates were used for
elemental standards. CO2 contents were calculated by difference assuming an oxide total of 100%.
effect on Mg and Ca diffusion rates, other studies suggest that there are
no detectable crystallographic orientation effects in calcite for C and O
self-diffusion between 500 to 800°C (Kronenberg et al., 1984) and Ca
self-diffusion between 700 to 900°C (Farver and Yund, 1996). For
calcite ⫹ MgO experiments a small amount of CaCO3 powder was also
added to the capsule to enhance stability of the calcite surface (Cherniak, 1997; Cherniak, 1998). Packed capsules were dried in a vacuum
oven at 60°C overnight and sealed in evacuated silica glass tubes before
being annealed in ThermolyneTM muffle furnaces for periods ranging
from 24 h to 150 d. Type-K chrome/alumel thermocouples were used
to monitor temperatures throughout the experiments, and the temperature uncertainty is estimated at less than ⫾3°C.
After annealing, samples were quenched by removal of the silica
tubes from the furnace. After cooling, carbonate fragments were removed from the silica tubes and the platinum capsules and rinsed
separately in both distilled H2O and in ethanol in an ultrasonic bath for
5 to 10 min to remove any adhering powdered source material. Annealed minerals were mounted with double-sided carbon tape on glass
discs and gold coated. Mineral surfaces were first examined for any
evidence of breakdown or residual source material using the SEM and
optical microscope and then analyzed with the ion microprobe. Within
a given experimental system the upper temperature bounds mark the
limit of stability of the carbonate cleavage faces (see below). The lower
temperature bounds in CaCO3 ⫹ magnesite and MgO ⫹ calcite experiments represent the minimum temperatures at which measurable diffusion gradients could be generated under convenient laboratory time
scales.
We report data from a total of 39 diffusion experiments. Experiments
were annealed at temperatures of 400, 450 and 500°C for magnesite ⫹
CaCO3 powder and 400, 450, 500, 550, and 600°C for calcite ⫹ MgO
powder. To examine the possible effects of differences in diffusant
composition, experiments were also run at 400, 450 and 500°C using
MgCO3 powder as the diffusant source for Mg chemical diffusion in
calcite and at 450°C using 26MgO powder to measure the rate of Mg
self-diffusion in magnesite. In addition, as a test of the veracity of our
experimental technique, both time series, where experiments are run
over a range of times for a given temperature, and zero time experiments for MgO ⫹ calcite and CaCO3 ⫹ magnesite were also performed. For the zero time experiments diffusion couples were brought
up to the experimental temperature for ⬃5 min and then quenched.
Diffusion profiles were analyzed with a modified Cameca ims-3f ion
microprobe, using a 16O⫺ primary ion beam with 17 keV impact energy
and ⬃1 to 10 nA current. To provide adequate depth resolution and
ensure uniform sputter rates, the primary ion beam was rastered over a
100 ⫻ 100 ␮m area. To avoid crater-edge effects an aperture, inserted
in the sample image plane, allowed only secondary ions originating
from an ⬃30 ␮m diameter region in the center of the rastered area to
enter the mass spectrometer. The mass spectrometer was tuned to
accept only high energy secondary ions by offsetting the accelerating
voltage by ⫺80 V relative to the voltage at which the intensity of 16O⫹
313
dropped to 10% of its maximum value on the low energy side of the
energy distribution. The width of the energy slit was adjusted to accept
secondary ions over a ⬃40 eV bandpass. The energy distribution of
16 ⫹
O was re-measured every 5 to 10 cycles to compensate for sample
charging during an analysis; the magnitude of charging usually ranged
between ⬃2 to 5 V. A typical analysis consisted of 200 to 1000
individual scans over the masses 25Mg⫹, 26Mg⫹, 40Ca⫹, 42Ca⫹, 44Ca⫹
(with count times of 1–2 s per mass) using low mass resolving power
(m/⌬m of ⬃500). The measured ion intensities were corrected for
counting system deadtime, taking into account the instantaneous count
rate.
Analysis times were converted to progressive depth from the mineral
surface via measurement of the depth of the final rastered crater with a
DektakTM stylus profilometer and assuming a constant sputter rate. The
overall accuracy of this measurement varied with surface roughness
and crater depth but is estimated to be better than ⫾5 to 10% relative;
the uncertainty in crater depth is not a significant contributor to the
uncertainty in the calculated diffusion coefficients.
A slightly modified analytical protocol to that described above was
used for the analysis of the 26Mg isotopic tracer experiments. To
resolve the isobaric interference of 24MgH⫹ on 25Mg⫹ and to obtain
better precision of Mg isotopic ratios, samples were analysed using a
mass-resolving power of ⬃2600 and no energy filtering, i.e., secondary
ions energies of 15 ⫾ 40V.
Diffusion coefficients for each experiment were calculated by performing a least-squares regression using the normalized 25Mg, 26Mg or
42
Ca depth profiles to the solution of Fick’s diffusion equation for
diffusion from a semi-infinite medium (1)
冉
x
共C x ⫺ C 1兲
⫽ erf
共C 0 ⫺ C 1兲
2
冑Dt
冊
,
(1)
where Cx is the concentration at depth x, Co is the initial concentration
in the calcite, C1 is the concentration at the crystal surface, D is the
diffusion coefficient and t is the duration of the anneal. For the
Mg-in-calcite experiments 25Mg⫹/42Ca⫹ and 26Mg⫹/42Ca⫹profiles
were inverted through Eqn. 1 and then fitted, while for the Ca-inmagnesite experiments the 40Ca⫹/25Mg⫹ profile was used. For the Mg
self-diffusion experiments in magnesite the 26Mg⫹/24Mg⫹ profile was
used. The formal solution to Eqn. 1 requires the sample surface to be
well defined in terms of both the concentration of the diffusant and
position relative to the measured depth profile. In practice the surface
concentration of the diffusant is approached asymptotically, making it
difficult to determine accurately. Small amounts of adhering diffusant
source material can also give erroneously high surface concentrations.
To minimize these problems, the surface concentration was estimated
iteratively by adjusting the value to minimize the reduced chi-squared
statistic for the least-squares fit; typically, the chi-squared values
ranged from 1 to 4. A more extensive discussion of this protocol is
given by Ryerson and McKeegan (1994).
The surfaces of all mineral fragments used for diffusion experiments
were examined after annealing and ultrasonic cleaning by optical
and/or scanning electron microscopy for evidence of dissolution or
degradation. Alteration of the mineral surface was apparent as small,
⬍1 to 5 ␮m diameter, regularly spaced pits in the cleavage surface.
Based on these observations, we established upper limits to the temperatures at which the host mineral remained stable in the presence of
the diffusant source. The limit of surface stability for calcite ⫹ MgO
experiments was ⬃600°C, for magnesite ⫹ CaCO3 experiments
⬃550°C, and for magnesite ⫹ 26MgO experiments ⬃500°C. At the
temperatures for which surface alteration was apparent in a given
experimental system, the degree of alteration appeared to be an approximate function of anneal time, with short-duration experiments
showing little or no sign of alteration, whereas longer duration experiments produced significantly altered mineral surfaces. For one calcite ⫹ MgO experiment annealed for 26.4 h at 600°C, several parts of
the host crystal appeared to have experienced no detectable alteration,
although elsewhere on the same crystal fragment alteration was readily
visible. Data from the analysis of one of these “unaltered” regions is
given in Table 3 and used herein under the proviso that this experiment
may have been affected by incipient surface alteration.
314
A. J. R. Kent et al.
Table 3. Data for the diffusion of Mg and Ca in calcite and magnesite.
Temperature
(°C)
Experimenta
Mg diffusion in calcite
A17
A2
A15
E4b
26
Time
(hours)
D
(m2/sec)
log D
(Ave. ⫾ 1␴)c
400
400
400
400
1440
2256
3648
1608
6.80 ⫻ 10⫺22
1.43 ⫻ 10⫺22
1.00 ⫻ 10⫺22
1.04 ⫻ 10⫺22
⫺21.75 ⫾ 0.39
A11
M47A
M47B
A12
A18
A16
E1b
450
450
450
450
450
450
450
744
1080
1080
1440
2256
3648
1608
1.10 ⫻ 10⫺21
8.06 ⫻ 10⫺22
3.26 ⫻ 10⫺22
1.10 ⫻ 10⫺22
8.57 ⫻ 10⫺23
1.04 ⫻ 10⫺22
8.81 ⫻ 10⫺22
⫺21.51 ⫾ 0.45
A5
A13 (i)
A13 (ii)
A4 (i)
A4 (ii)
M13 (i)
M13 (ii)
A8
M22b
500
500
500
500
500
500
500
500
500
240
240
240
576
576
600
600
1632
600
8.14 ⫻ 10⫺22
2.07 ⫻ 10⫺21
1.04 ⫻ 10⫺21
3.10 ⫻ 10⫺22
5.90 ⫻ 10⫺22
8.97 ⫻ 10⫺22
8.62 ⫻ 10⫺22
4.50 ⫻ 10⫺22
1.48 ⫻ 10⫺21
⫺21.01 ⫾ 0.21
A19
M48
M48
M49
M49
555
555
555
555
555
362
384
384
504
504
2.07 ⫻ 10⫺21
1.17 ⫻ 10⫺21
1.05 ⫻ 10⫺21
1.00 ⫻ 10⫺21
7.70 ⫻ 10⫺22
⫺20.92 ⫾ 0.03
(i)
(ii)
(i)
(ii)
600
Mg diffusion in magnesite
C2 (i)
C2 (ii)
450
450
1104
1104
7.30 ⫻ 10⫺23
1.67 ⫻ 10⫺22
⫺21.96 ⫾ 0.25
400
400
2256
3648
1.49 ⫻ 10⫺25
4.40 ⫻ 10⫺25
⫺24.59 ⫾ 0.33
B8
B3
B4 (i)
B4 (ii)
B1
450
450
450
450
450
720
1440
2256
2256
3648
7.54 ⫻ 10⫺24
5.00 ⫻ 10⫺25
3.50 ⫻ 10⫺23
6.60 ⫻ 10⫺23
5.90 ⫻ 10⫺24
⫺23.06 ⫾ 0.82
B6
B7
B9
500
500
500
240
816
1632
1.58 ⫻ 10⫺23
1.09 ⫻ 10⫺22
3.83 ⫻ 10⫺23
⫺22.43 ⫾ 0.43
Ca diffusion in magnesite
B16
B5
a
b
c
26.4
1.97 ⫻ 10⫺20
M14
Small Roman numerals denote repeat analyses of the same experiment.
Experiment performed with MgCO3 powder and calcite.
Averages for Mg-diffusion include data from MgCO3 and MgO-source experiments.
3. RESULTS
Data from two typical SIMS depth profiles are shown in
Figure 1. The results of the diffusion experiments are given in
Table 3 together with the corresponding experimental conditions and are shown graphically in Figures 2 and 3. The depth
of the analysis craters varied between ⬃100 and 5000 nm.
Replicate analyses show that the reproducibility of a measured
diffusion coefficient for a given experiment is ⫾ ⬃ 0.2 log
units (Table 3). However, the overall reproducibility for all
experiments conducted at a given temperature is generally ⫾
⬃ 0.4 log units (1␴; Table 3). This uncertainty is taken to be
representative of all the individual diffusion measurements, and
is similar to the variability reported in other experimental
studies of diffusion in calcite (e.g., (Cherniak, 1997; 1998).
One exception is the measured diffusion rate of Ca-in-magnesite at 450°C, which has a higher standard deviation of 0.82 log
units (Table 3). The reason for the higher standard deviation for
these particular experiments is uncertain but may be partly due
to shorter (and thus more difficult to measure) diffusion profiles
in Ca diffusion experiments.
The results of the zero time experiments are also shown in
Figure 1. In this figure the measured zero time profiles for the
25
Mg⫹/42Ca⫹ and 40Ca⫹/25Mg⫹ ratios are compared to the
diffusion gradients from the shortest duration and lowest temperature Mg and Ca diffusion experiments for which we report
data (i.e., the experiments with the shortest diffusion profiles).
Formation temperature of ALH84001 carbonate
315
Fig. 1. Results from Mg and Ca-diffusion experiments. A. Measured
diffusion gradient for chemical diffusion of Mg in calcite for experiment A17, annealed at 400°C for 1440 h (60 d). Triangles represent
individual ion-probe measurements of the 25Mg⫹/42Ca⫹ ratio as a
function of depth. The thick solid line shows the error function fitted to
these data and corresponds to a diffusion coefficient of 6.80 ⫻ 10⫺22
m2/s. Conversion of analysis time to depth was done by measuring the
depth of the ion probe crater. To emphasize the quality of the error
function fitting procedure, the dashed lines show diffusion profiles
calculated for a factor of two variation in the diffusion coefficient (D ⫽
3.40 ⫻ 10⫺22 and D ⫽ 13.60 ⫻ 10⫺22 m2/s). Diamonds represent the
25
Mg⫹/42Ca⫹ ratios measured in a zero time experiment conducted at
400°C (see text for details). B. Measured diffusion gradient for chemical diffusion of Ca in magnesite for experiment B16, annealed at
400°C for 2256 h (94 d). Triangles represent individual ion-probe
measurements of the 40Ca⫹/25Mg⫹ ratio as a function of depth. The
thick solid line shows the error function fitted to these data and
corresponds to a diffusion coefficient of 1.49 ⫻ 10⫺25 m2/s. The dashed
lines show diffusion profiles calculated for a factor of two variation in
the diffusion coefficient (D ⫽ 2.98 ⫻ 10⫺25 and D ⫽ 0.75 ⫻ 10⫺22
m2/s). Diamonds represent the 40Ca⫹/25Mg⫹ ratios measured in a zero
time experiment conducted at 400°C.
Although these results show that detectable amounts of residual
diffusant material do remain on the sample surfaces after cleaning (as shown by the slightly elevated 25Mg⫹/42Ca⫹ and
40
Ca⫹/25Mg⫹ ratios at the shallowest depths in zero time
profiles), the degree to which the 25Mg⫹/42Ca⫹ and 40Ca⫹/
25
Mg⫹ ratios are elevated is small compared to the size of the
measured diffusion profiles (Fig. 1). We are thus confident our
diffusion measurements are not significantly effected by residual diffusant material present on the sample surface during
analysis. The zero time experiments also allow us to estimate
the smallest diffusion gradients that we can measure with our
experimental protocol: for MgO ⫹ calcite experiments the
smallest measurable profiles are on the order of ⬃10 to 15 nm,
whereas for CaCO3 ⫹ magnesite profiles as small as 5 nm can
be measured (Fig. 1).
The results of time series experiments are shown in Figure 2.
Fig. 2. Time series plots for Mg and Ca diffusion experiments in
calcite and magnesite. Hollow and gray squares represent data for Mg
diffusion in calcite measured using MgO and MgCO3 respectively. The
star represent the rate of self diffusion of 26Mg in magnesite. Circles
represent the rate of Ca diffusion in magnesite. Solid lines show the
average log Do value for Mg diffusion in calcite and dashed lines show
the average value for Ca diffusion in magnesite. Averages do not
include data for MgCO3-source diffusion experiments. Error bars on
the individual symbols represent estimated ⫾1␴ uncertainty. A. 400; B.
450; C. 500; and D. 550°C.
For all temperatures examined, and for anneal times that vary
by up to a factor of ⬃5 for the 400, 450 and 500°C experiments
and ⬃1.5 for the 550°C experiments, there is no consistent
316
A. J. R. Kent et al.
mined by a weighted least-squares fit to the individual measured diffusivities (as opposed to the average values) at each
temperature. The arrays for Mg-in-calcite and Ca-in-magnesite
diffusion correspond to respective activation energies (EA) of
76 ⫾ 16 and 214 ⫾ 60 kJ/mol and log Do values of ⫺16.0 ⫾
1.1 and ⫺7.8 ⫾ 4.3 m2/s. The quality of the fit of the data to
linear arrays on the Arrhenius plot again suggests that a single
transport mechanism dominates the chemical diffusion of Ca
and Mg over the examined temperature range. The Arrhenius
parameters obtained for Mg-in-calcite diffusion remain essentially unchanged if the single data point from the diffusion
experiment at 600°C is removed (log Do ⫽ ⫺16.9 ⫾ 2.6 and
EA⫽ 63 ⫾ 14 kJ/mol). As discussed above, this experiment
may have been subject to incipient surface alteration.
The measured Mg diffusion rates appear independent of the
composition of the diffusant source, as Mg diffusion rates
determined using MgO and MgCO3 as diffusant sources are
within analytical uncertainty of each other at 500, 450 and
400°C (Fig. 2, 3). In addition, the array defined by the three
MgCO3-source experiments corresponds to EA and Do values
(117 ⫾ 36 and ⫺12.8 ⫾ 2.6) that are within error (1␴) of the
EA and Do values defined by the regression of data from
MgO-source experiments (76 ⫾ 16 and ⫺16.0 ⫾ 1.1). The rate
of 26Mg self-diffusion in magnesite at 450°C (with an average
value of log DMg ⫽ ⫺21.96 ⫾ 0.25 cm2/s) is also within the
estimated 1␴ analytical uncertainty of chemical diffusion rates
of Mg in calcite measured at the same temperature (Table 3,
Fig. 3).
4. DISCUSSION
4.1. Diffusion Rates of Ca and Mg in Calcite and
Magnesite
Fig. 3. A. Arrhenius plot showing our data for diffusion of Mg in
calcite and magnesite and Ca in magnesite. Symbols are the same as for
Figure 2 and are also shown in the accompanying legend. Representative ⫾1␴ error bars are shown at the bottom left. Data for Ca and Mg
self-diffusion in calcite from Farver and Yund (1996) (“F&Y 96”) and
Fisler and Cygan (1998), Fisler and Cygan (1999) (“F&C 98”) are also
shown by the labeled thin solid lines. Thin dashed line shows the least
squares best fit for MgCO3-source Mg diffusion data, solid dashed line
shows best fit for all our Mg diffusion data collected in this study and
solid thick line shows the best fit for our Ca diffusion data. B. Comparison of our data for Mg and Ca diffusion with data for diffusion of
REE, Sr, Pb, and O (additional data from Cherniak, 1997; 1998; Farver,
1994; Kronenberg et al., 1984).
change in diffusion coefficient with time for either Ca or Mg
diffusion (Fig. 2). This absence of variation suggests that the
transport of Ca and Mg is dominated by a single mechanism,
lattice diffusion, over the range of time scales investigated.
Our data for Mg and Ca diffusion are plotted on an Arrhenius
plot in Figure 3. Collectively, data for Ca diffusion in magnesite and Mg diffusion in calcite form separate, negativelysloped arrays on this plot, suggesting that the rates of diffusion
of Mg and Ca in carbonate minerals are substantially different
over this temperature interval. The Arrhenius parameters, activation energy (E A ) and preexponential factor (D 0 ), were deter-
Our measured rates of chemical diffusion of Mg in calcite
agree well with the data reported by Fisler and Cygan (1998;
1999) for self-diffusion of Mg in calcite between 600 and
550°C (Fig. 3A). However, at lower temperatures our diffusion
rates are substantially faster (between 2–3 orders of magnitude
at 400°C) than predicted from simple down-temperature extrapolation of their results. Our calculated activation energy for
Mg diffusion in calcite between 600 to 400°C (76 ⫾ 14 kJ/mol)
is also substantially different from the 284 ⫾ 74 kJ/mol value
reported by Fisler and Cygan (1998; 1999) for Mg diffusion in
calcite between 550 and 800°C.
The distinct change in the slope of the Arrhenius array for
Mg diffusion in calcite between our data and those of Fisler and
Cygan (1998; 1999) indicates that a substantial difference in
the measured activation energy exists between the two data sets
(as well as significantly different rates of Mg diffusion when
the data of Fisler and Cygan (1998; 1999) are extrapolated to
400°C). This difference could be related to differences in
experimental technique (i.e., measurement of the rates of selfdiffusion using isotopic tracers sputtered onto the mineral surface, compared to measurement of chemical diffusion from a
powder source). However, given that the diffusion rates for Mg
measured at 600 and 550°C by both studies are within analytical error of each other (Fig. 3), the observed difference in
measured activation energies may also indicate that different
transport mechanisms dominate Mg diffusion in calcite at temperatures above and below ⬃550°C. Fisler and Cygan (1998)
Formation temperature of ALH84001 carbonate
speculated that higher diffusivity of Mg in calcite at temperatures
below 550°C (their lowest experimental temperature) would be
consistent with a change from an intrinsic diffusion process to an
extrinsic diffusion process (or processes) at lower temperatures.
While our results are consistent with this suggestion, we note that,
similar changes in activation energy at ⬃550°C are not observed
for diffusion of Ca in magnesite (Fig. 3) or for several other
cations (Sr, Pb, REE) or O in calcite (Cherniak, 1997; 1998;
Farver, 1994).
One potential explanation for the differences in diffusivities
and activation energies between our Mg diffusion data and
those of Fisler and Cygan (1998; 1999) is that they reflect
differences in the composition of the diffusant source material
(in this case oxide for the majority of our Mg diffusion experiments and for the experiments of Fisler and Cygan (1998;
1999) and carbonate for our Ca diffusion measurements). The
composition of the source material used in diffusion experiments can potentially influence surface defect characteristics
and thus may affect the measured rates of diffusion. Although
the majority of our data for Mg diffusion in calcite is from
MgO-source experiments, we have tested the possible effect of
variations in diffusant source composition by also conducting
experiments using MgCO3 powder as the diffusant for Mg
diffusion in calcite. Our results suggest that no significant
difference exists between Mg diffusion data derived from
MgO-source and MgCO3-source experiments (Table 2; Figs. 2,
3). Experiments using MgCO3 powder as a source for Mg
diffusion have measured Mg diffusivities that are within analytical uncertainty of measured MgO-source Mg diffusivities at
400, 450 and 500°C (Table 2; Figs. 2, 3) and the regression of
data from the three MgCO3 experiments produces EA and Do
values that are within analytical error of the values calculated from
the regression of MgO-source diffusion data. In addition, the
diffusivities for Mg measured at 450 and 400°C using MgCO3 as
a diffusant source are substantially faster than both the measured
rates of Ca diffusion (using CaCO3 as a source) from our study
and the extrapolated diffusivities of Mg from the study of Fisler
and Cygan (1998; 1999). These data suggest that the rate of Mg
diffusion is independent of the choice of oxide or carbonate
powder for source material and that the differences in activation
energy and diffusivity that we observe between Ca and Mg may
be a fundamental feature of diffusive transport in carbonate
minerals.
An alternate explanation for the differences between our data
and those of Fisler and Cygan (1998; 1999) is the possibility
that these could be due to different concentrations of minor
elements in the calcites used for experiments. This possibility
requires further investigation. Variations in the concentrations
of Mn in calcite have been observed to have a marked effect on
the rate of self diffusion of O (Kronenberg et al., 1984),
although the same study found that the diffusion rate of C
appeared to be unaffected by the Mn concentration. We note
that although we have used calcite from the same locality
(Chihuahua, Mexico) as Fisler and Cygan (1998; 1999) for our
experiments, the MnO contents measured by us (0.19 ⫾ 0.02
wt.%) are much greater than the ⬃0.02 wt.% MnO reported by
Fisler and Cygan (1999). In addition, the concentration of FeO in
the calcite used for our experiment is also greater (0.10 compared
to 0.02 wt.%) than that reported by Fisler and Cygan (1999). This
could also have an effect on rates of cation diffusion. If Mg
317
diffusion below 550°C is an extrinsic process governed by the
density of Fe-based defects, diffusion would be expected to be
faster at higher FeO concentrations.
The rates of chemical diffusion of Ca in magnesite determined here are essentially indistinguishable from the rates of
self-diffusion of Ca in calcite indicated by the down-temperature extrapolation of the diffusivities measured by Fisler and
Cygan (1998; 1999) (Fig. 3). The activation energy for chemical iffusion of Ca in magnesite between 400 to 500°C determined here, 202 ⫾ 51 kJ/mol, is within analytical uncertainty
of the 271 ⫾ 80 kJ/mol activation energy for self-diffusion of
Ca in calcite reported by Fisler and Cygan (1998; 1999). Both
the activation energy and the diffusivity of Ca measured by us
and by Fisler and Cygan (1998; 1999) distinct from the corresponding values for self-diffusion of Ca in calcite measured at
temperatures of 700 to 900°C by Farver and Yund (1996)
[EA ⫽ 382 ⫾ 37kJ/mol]. Fisler and Cygan (1998) suggested
that this discrepancy may be the result of Ca diffusion occurring via a different mechanism at the relatively high temperatures (700 –900°C) investigated by Farver and Yund (1996).
4.2. Constraints on the Thermal History of ALH84001
Carbonates
Application of our data to investigate the thermal history of
ALH84001 requires that the chemical diffusion rates that we
have measured in calcite and magnesite are representative of
the diffusion rates of Ca and Mg in the carbonate minerals in
ALH84001. The removal of zoning and heterogeneity in the
Ca-Mg composition of carbonates by diffusive transport requires that both Ca and Mg diffuse simultaneously within the
carbonate lattice. Assuming that this is a binary diffusion
process and that diffusion occurs with thermodynamically ideal
mixing behaviour, the rate at which binary diffusion occurs is
related to the self-diffusion rates of Mg and Ca by the following
equation (Chakraborty, 1995):
D MgCa ⫽
D *MgD *Ca
,
X MgD *Mg ⫹ X CaD *Ca
(2)
where DMgCa is the binary diffusion coefficient, XMg, XCa are
the respective mole fractions of Ca and Mg and D*Mg, D*Ca are
the respective self-diffusion coefficients for Mg and Ca.
Inspection of Eqn. 2 shows that for calcite, where XCa
approaches 1, the Ca-Mg binary diffusion coefficient will be
⬇ D*Mg. This approximation is confirmed by the similarity
between the measured rates for the self-diffusion of Mg in
magnesite and chemical diffusion of Mg in calcite at 450°C
(Table 3, Fig. 3). Similarly, for magnesite the Ca-Mg binary
diffusion coefficient (which for a Mg-rich carbonate is the same
as the rate of chemical diffusion of Ca) will be ⬇D*Ca. Thus, the
rates of chemical diffusion of Mg in near-pure calcite and Ca in
near-pure magnesite (which are measures of the rates of Mg-Ca
interdiffusion in each mineral) are approximately the same as
the rates of self-diffusion of Ca and Mg. For carbonates containing both Ca and Mg, the binary diffusion coefficient will
have an intermediate value, and the rate at which cation diffusion will remove Mg-Ca compositional differences in carbonates may reasonably be expected to fall within the bounds
calculated from the measured rates of chemical diffusion of Ca
in magnesite and Mg in calcite (Fisler et al., 1998).
318
A. J. R. Kent et al.
Fig. 4. Measured diffusion profile for chemical diffusion of Mg in
calcite, annealed at 400°C for 94 d (experiment A2). Squares mark the
individual ion-probe measurements of 25Mg⫹/42Ca⫹ for this experiment and the solid line shows the error function fitted to these data.
Dashed lines show the calculated diffusion profiles that would result
from heating durations of 10, 100, 1000 yr.
For ALH84001, we believe that the rate of chemical diffusion of Mg in calcite provides the most stringent constraint on
the formation temperature of carbonates. Mg diffusion is much
more rapid than that of Ca and thus in Ca-rich carbonates,
where the binary diffusion coefficient is dominated by the Mg
diffusivity, the most stringent temperature constraints on the
formation temperature of ALH84001 carbonates will be provided by Mg diffusivity. Carbonate compositions in
ALH84001 vary considerably. Ca-rich carbonates are known to
occur (Harvey and McSween, 1996; Scott et al., 1998) and,
importantly, are associated with fine-scale compositional zoning and are juxtaposed with other composition carbonates on
the ⬃1 ␮m scale (e.g., Fig. 2 in Scott et al., 1997).
A simple treatment of our data shows that the diffusion rate
of Mg in calcite is sufficiently fast at temperatures of 400°C
that it is unlikely that short-length scale differences in Mg or Ca
concentration could survive in Ca-rich carbonates that formed
at or above this temperature. This is shown in Figure 4, where
we plot the measured diffusion profile for Mg in calcite annealed at 400°C for 94 d (experiment A2). Using the diffusion
coefficient calculated from this experiment (Do ⫽ 1.43 ⫻
10⫺22 m2/s) and Eqn. 1, we have calculated and plotted the
diffusion profiles that would result from continued annealing of
this experiment at 400°C for 10, 100 and 1000 yr. The initial
step function in Mg composition (i.e., the t ⫽ 0 Mg concentration profile) is quickly flattened and removed over the 1 ␮m
distance (similar to the observed length scale of zoning observed in ALH84001) used for the x-axis, even over the geologically short durations (10 –1000 yr) used for the calculation.
The only assumption required in this treatment is that the
chemical diffusion rate of Mg in calcite determined experimentally reflects the rate at which Mg-Ca compositional differences
would be removed by diffusion at 400°C in Ca-rich carbonate
in ALH84001. This simple analysis suggests that carbonates in
ALH84001 have not experienced temperatures of 400°C or
greater for periods of time exceeding 100 yr. We conclude that
if the ALH84001 carbonates formed at temperatures greater
Fig. 5. Plot of cooling rate versus the length scale of compositional
zoning (that would survive diffusional homogenization) for a range of
carbonate formation temperatures, calculated following the approach of
Sheng et al. (1992). Thick solid lines represent formation temperatures
of 600, 300 and 200°C, calculated using our data for Mg diffusion in
calcite. The thin solid line represents an initial formation temperature of
400°C, calculated using our data for Ca diffusion in magnesite. Dashed
lines represent formation temperatures of 450 and 600°C, calculated
from the Mg self-diffusion data of Fisler and Cygan (1998; 1999). The
area below each line represents the conditions under which carbonate
compositional heterogeneity would survive homogenization from diffusion during cooling from the given formation temperature.
than 400°C, they must have cooled extremely rapidly to preserve the observed variations in Mg concentration.
To investigate further the relationship between formation
temperature, cooling rate and survival of short length-scale
chemical zonation in ALH84001 carbonates over a more realistic range of thermal histories, we consider the case of monotonic cooling from a maximum initial temperature (Ganguly et
al., 1994; Kaiser and Wasserburg, 1983; Sheng et al., 1992).
From the formulation of (Sheng et al., 1992):
x2 ⬵
RT 20 D共T 0兲
,
r 0E A
(3)
where x is the minimum distance over which carbonate will
preserve Mg-Ca compositional differences, To is the initial
formation temperature, D(To) is the diffusivity at To, ro is the
initial cooling rate at To, EA is the activation energy and R is the
gas constant. Application of Eqn. 3 requires that the term E/RTo
is ⬎⬎ 1 (for Mg diffusion in calcite with T0 ⫽ 600°C this term
is ⬃10) and that the cooling history following formation at
temperature To is linear in 1/T. For conductive cooling following an initial formation temperature or decay of a transient
thermal spike, the latter assumption will be approximately true
(Dodson, 1973).
The results of our calculations are shown in Figure 5, where
we plot the length scale over which Ca-Mg heterogeneity
would survive as a function of the cooling rate and temperature
at the time of formation of the host mineral. For calculations
based on the diffusion of Mg in calcite we have plotted three
curves, representing formation temperatures of 200, 300 and
600°C. For comparison, we have also plotted two curves using
the Mg diffusion data of Fisler and Cygan, (1998; 1999) and
formation temperatures of 450 and 600°C. For calculations
based on the rate of Ca diffusion in magnesite we have plotted
a curve representing a formation temperature of 400°C. As
Formation temperature of ALH84001 carbonate
discussed above, the rate of Mg diffusion in calcite controls the
rate at which Mg-Ca compositional variations in Ca-rich carbonates will be removed by diffusion and, as Mg diffusion is
considerably faster than Ca, provides the most stringent constraint on the formation temperature. In Mg-rich carbonate the
rate at which Mg-Ca compositional variations can be homogenized by diffusion is determined by the slower rate of Ca
diffusion, and calculations based on Ca diffusion in magnesite
will provide an upper limit for the formation temperature. The
curves in Figure 5 illustrate the relationship between the initial
cooling rate and the characteristic diffusion distance for a large
range of potential cooling rates (10 orders of magnitude) and
form a basis for evaluating the role of different processes in the
formation of ALH84001 carbonates.
In Figure 5 we have highlighted the range of cooling rates
known from rocks that cool within the terrestrial crust (⬃0.1–
500°C/Ma), with the slowest rates equivalent to those experienced by stable Archaean crustal regions (e.g., Kent, 1994) and
the fastest rates from rapidly uplifted terranes such as metamorphic core complexes (Baldwin et al., 1993). We believe that
this range approximates the range of cooling rates that would
expected to be associated with igneous, metamorphic, tectonic,
hydrothermal or other processes within the Martian crust. The
special case of carbonate formation during and after an impact
event is dealt with separately below.
From Figure 5 it is apparent that the rate of Mg-Ca diffusion
in Ca-rich carbonates is sufficiently rapid that preservation of
⬃1 ␮m scale Mg-Ca compositional heterogeneity in carbonates
formed at temperatures of 600°C or greater would require an
initial cooling rate of at least ⬃107°C/Ma. Lowering the formation temperature to ⬃300°C still requires cooling rates considerably faster than those known from within the terrestrial
crust. Formation temperatures of ⬃200°C or less are required
to preserve the 1 ␮m scale Mg-Ca zoning in Ca-rich carbonates
in ALH 84001 for the range of cooling rates known from within
the terrestrial crust. For Mg-rich carbonates, calculations based
on the rate of Ca diffusion indicate that formation temperatures
of ⬃400°C or less are required to preserve 1 ␮m Ca-Mg
zoning. We thus conclude that if ALH84001 carbonates formed
within the Martian crust, the carbonate formation temperatures
did not exceed 400°C and were most plausibly below 200°C.
These conclusions strongly favor models that involve lowtemperature processes for carbonate formation and place tighter
constraints on the formation temperature of ALH84001 carbonates than previous studies (Fisler and Cygan, 1998; 1999). We
note that the Mg diffusion data of Fisler and Cygan (1998;
1999), if used for a similar analysis, only require formation of
ALH84001 carbonates at temperatures below ⬃450°C (Fig. 5).
4.2.1. Carbonates formed during an impact event
Several authors have suggested that carbonate rosettes in
ALH84001 formed at relatively high temperatures (⬃600 –
700°C) during heating and/or high temperature fluid circulation
associated with a Martian impact event (Harvey and McSween,
1996; Scott et al., 1997; Scott et al., 1998). Although the
calculations discussed in the preceding section indicate that the
formation of carbonates at temperatures of ⬎600°C requires
initial cooling rates in excess of 107°C/Ma (Fig. 5), cooling
319
rates following impact events may be substantially higher than
those related to terrestrial tectonic processes (Crossey et al.,
1994), and it is possible that fine-scale compositional zonation
in carbonates formed at temperatures of ⱖ600°C during an
impact event could survive. In addition, petrographic observations of ALH84001 indicate that fine-scale chemical zoning in
carbonates has survived at least one impact event, as finelyzoned carbonate grains are disrupted and offset in ALH84001
by impact-produced, plagioclase-composition melt glass
(Shearer et al., 1999; Treiman, 1995; 1998). Cooling rates
following any impact or thermal event must have been fast
enough to preserve the fine-scale carbonate zonation.
Our data on cation diffusivities can still be used within the
context of a monotonic cooling history to place some constraints on models that argue for formation of ALH84001
carbonates during impact events. Cooling rates (and thus the
potential survival of fine-scale chemical zonation in carbonates) within impact zones during and after impact vary significantly depending upon the geometry of the structure and
location within the impact structure (Crossey et al., 1994;
McCarville and Crossey, 1996). Studies of terrestrial impact
sites have shown that rocks within impact structures, including
those that like ALH84001 contain diaplectic and melt glasses
(and thus reached peak temperatures of up to 800°C or more
during impact; Grieve, 1987), experienced relatively slow cooling after impact (Crossey et al., 1994). This slow cooling is a
result of postimpact rebound and uplift of basement rocks
within the central region of an impact site. This causes lower
crustal isotherms to move closer to the surface, and the postimpact decay of the resulting thermal anomaly then promotes
relatively slow cooling compared to the uppermost melt-rich
portions of the impact structure (Crossey et al., 1994). For
example, cooling rates within the lower portions of the Manson
impact structure (peak temperatures of 600 –1000°C) were
⬃500 to 800°C/Ma (Crossey et al., 1994)—similar to the
fastest known terrestrial cooling rates associated with rapid
tectonic uplift shown in Figure 5. Only the uppermost, meltrich parts of the Manson structure experienced substantially
faster cooling (⬃ 106–107°C/Ma). Carbonate minerals formed
within an impact structure, with the exception of those that may
have formed within the relatively thin and rapidly cooled upper
melt sheet, would be subject to the same thermal constraints
shown in Figure 5 and are thus are unlikely to have formed at
temperatures greater than 200 to 400°C.
In addition, carbonate minerals that form during an impact
event are more likely to occur within the lower, more slowlycooled portions of an impact structure because it is this region
that experiences the most intense hydrothermal activity (Allen
et al., 1982; McCarville and Crossey, 1996). In terrestrial
environments the formation of carbonate minerals is common
in hydrothermal circulation zones and many workers have
suggested that carbonate rosettes in ALH84001 are also the
result of hydrothermal processes (Table 1). Studies of hydrothermal minerals and fluid inclusions from the Manson impact
structure show that hydrothermal circulation within the uplifted
central region occurred continuously from temperatures of
⬎300°C, down to ambient temperatures (Boer et al., 1996;
McCarville and Crossey, 1996).
320
A. J. R. Kent et al.
5. CONCLUSIONS
We have measured the rates of chemical diffusion of Ca in
magnesite and Mg in calcite and applied these new data to a
study of the formation temperature and thermal history of
carbonates in Martian meteorite ALH84001. Measured log D0
values for chemical diffusion of Mg in calcite and Ca in
magnesite are ⫺16.0 ⫾ 1.1 and ⫺7.8 ⫾ 4.3 m2/s and the
activation energies (EA) are 76 ⫾ 16 and 214 ⫾ 60 kJ/mol,
respectively. Diffusion rates of Mg in magnesite at temperatures between 400 to 550°C are substantially faster than expected from extrapolation of existing data, suggesting that
different mechanisms may govern diffusion of Mg at temperatures above and below ⬃550°C.
Considerations of isothermal and slow cooling thermal histories for carbonates suggest that, for a range of cooling rates
similar to those observed within the terrestrial crust,
ALH84001 carbonates are likely to have formed at temperatures at least ⬍400°C and most likely ⬍200°C. This supports
models involving low-temperature processes for carbonate formation.
In addition, evaluation of the range of thermal histories
evident within impact structures suggests that carbonates
formed within the lower parts of an impact structure would also
have cooled at rates slow enough to restrict carbonate formation to temperatures below ⬃200 to 400°C.
Acknowledgments—This work was funded by Lawrence Livermore
National Laboratory under the Laboratory Directed Research and Development (LDRD) program, and by NASA through work order
W-18845. Reviews by A. Treiman and two anonymous reviewers
significantly improved the quality of this manuscript. We are also
grateful to Haley Ryerson for helping to choose the prettiest carbonate
samples. Work performed under the auspices of the U.S. Department of
Energy by Lawrence Livermore National Laboratory under Contract
W-7405-ENG-48.
Associate editor: H. E. Newsom
REFERENCES
Allen C. C., Gooding J. L., and Keil K. (1982) Hydrothermally altered
impact melt rock and breccia: Contributions to the soil of Mars. J.
Geophys. Res. 87, 10083–10101.
Baldwin S. L., Lister G. S., Hill E. J., Foster D. A., and McDougall I.
(1993) Thermochronological constrains on the tectonic evolution of
the active metamorphic core complexes, De’Entrecasteaux Islands,
Papua New Guinea. Tectonics. 12, 611– 628.
Boer R. H., Reimold W. U., Koeberl C., and Kesler S. E. (1996) Fluid
inclusion studies on drill core samples from the Manson impact
crater: Evidence for post-impact hydrothermal activity. In The Manson Impact Structure, Iowa: Anatomy of an Impact Crater (ed. C.
Koeberl and R. A. Anderson), Vol. 302, pp. 377–382. Geological
Society of America Special Paper.
Chakraborty S. (1995) Diffusion in silicate melts. In Structure, dynamics and properties of silicate melts (ed. J. F. Stebbins, P. F. McMillan, and D. B. Dingwell), Vol. 32, pp. 411–503. Mineralogical
Society of America.
Cherniak D. J. (1997) An experimental study of strontium and lead
diffusion in calcite and implications for carbonate diagenesis and
metamorphism. Geochim. Cosmochim. Acta 61, 4173– 4179.
Cherniak D. J. (1998) REE diffusion in calcite. Earth Planet. Sci. Lett.
160, 273–287.
Cooney T. F., Scott E. R. D., Krot, A. N., Sharma, S. K., and Yamaguchi, A. (1999) Vibrational spectroscopic study of minerals in the
Martian meteorite ALH84001. Am. Mineral. 84, 1569 –1576.
Crossey L. J., Kudo A. M., and McCarville P. (1994) Post-impact
hydrothermal systems: Manson impact structure, Manson, Iowa.
Lunar Planet. Sci. XXIV. 299 –300.
Dodson M. H. (1973) Closure temperature in cooling geochronological
and petrological systems. Contrib. Mineral. Petrol. 40, 259 –274.
Farver J. R. and Yund R. A. (1996) Volume and grain boundary
diffusion of calcium in natural and hot-pressed calcite aggregates.
Contrib. Mineral. Petrol. 123, 77–91.
Farver R. J. (1994) Oxygen self-diffusion in calcite: Dependence on
temperature and water fugacity. Earth Planet. Sci. Lett. 121, 575–
587.
Fisler D. K. and Cygan R. T. (1998) Cation diffusion in calcite:
Determining closure temperatures and the thermal history for the
Allan Hills 84001 meteorite. Meteor. Planet. Sci. 33, 785–789.
Fisler D. K. and Cygan R. T. (1999) Diffusion of Ca and Mg in calcite.
Am. Mineral. 84, 1392–1399.
Fisler D. K., Cygan R. T., and Westrich H. R. (1998) Cation diffusion
in carbonate minerals: determining closure temperatures and the
thermal history for the ALH 84001 meteorite. Lunar Planet. Sci.
XXVIV, 221–222.
Ganguly J., Yang H., and Ghose S. (1994) Thermal histories of mesosiderites: Quantitative constrains from compositional zoning and
Fe-Mg ordering in orthopyroxenes. Geochim. Cosmochim. Acta 58,
2711–2723.
Golden, D. C., Ming D. W., Schwandt C. S., Morris R. V., Yang S. V.,
and Lofgren G. E. (2000). An experimental study on kineticallydriven precipitation of calcium-magnesium-iron carbonates from
solution: Implications for the low-temperature formation of carbonates in martian meteorite ALH84001. Meteor. Planet. Sci. 35, 457–
465.
Grieve R. A. F. (1987) Terrestrial impact structures. Ann. Rev. Earth
Planet. Sci. 15, 245–270.
Harvey R. P. and McSween Jr. H. Y. (1996) A possible high-temperature origin for the carbonates in the Martian meteorite ALH84001.
Nature 382, 49 –51.
Hutchins K. S. and Jakosky B. M. (1997) Carbonates in Martian
meteorite ALH84001: A planetary perspective. Geophys. Res. Lett.
24, 819 – 822.
Kaiser T. and Wasserburg G. J. (1983) The isotopic composition and
concentration of Ag in iron-meteorites and the origin of exotic silver.
Geochim. Cosmochim. Acta 47, 43–58.
Kent A. J. R. (1994) Geochronological constraints on the timing of
Archaean gold mineralisation in the Yilgarn Craton, Western Australia. Ph.D., Australian National University.
Kirschvink J. L., Maine A. T., and Vali H. (1997) paleomagnetic
evidence of low-temperature origin of carbonate in the Martian
meteorite ALH 84001. Science 275, 1629 –1633.
Kronenberg A. K., Yund A. R., and Gilletti B. J. (1984) Carbon and
oxygen diffusion in calcite: Effects of Mn content and PH2O. Phys.
Chem. Min. 11, 101–112.
Leshin L. A., McKeegan K. D., Carpenter P. K., and Harvey R. P.
(1998) Oxygen isotope constraints on the genesis of carbonates from
Martian meteorite ALH84001. Geochim. Cosmochim. Acta 62, 3–13.
McCarville P. and Crossey L. J. (1996) Post-impact hydrothermal
alteration of the Manson impact structure. In The Manson Impact
Structure, Iowa: Anatomy of an Impact Crater (ed. C. Koeberl and
R. A. Anderson), Vol. 302, pp. 347–376. Geological Society of
America Special Paper.
McKay D. S., Gibson Jr. E. K., Thomas-Keptra K. L., V. H., Romanek
C. S., Clemett S. J., Chillier X. D. F., Maechling C. R., and Zare
R. N. (1996) Search for past life on Mars: Possible relic biogenic
activity in Martian meteorite ALH84001. Science 273, 924 –930.
McKay G. A. and Lofgren G. E. (1997) Carbonates in ALH84001:
Evidence for kinetically controlled growth. Lunar and Planetary
Science XXVIII, 921–922.
Mittlefehldt D. W. (1994) ALH84001, a cumulate orthopyroxenite of
the Martian meteorite clan. Meteoritics 29, 214 –221.
Romanek C. S., Grady M. M., Wright I. P., Mittlefehldt D. W., Socki
R. A., Pillinger C. T., and Gibson Jr. E. K. (1994) Record of
fluid-rock interactions on Mars from the meteorite ALH84001. Nature 372, 655– 657.
Ryerson F. J. and McKeegan K. D. (1994) Determination of oxygen
self-diffusion in akermanite, anorthite, diopside, and spinel: Impli-
Formation temperature of ALH84001 carbonate
cations for oxygen isotopic anomalies and the thermal histories of
Ca-Al-rich inclusions. Geochim. Cosmochim. Acta 58, 3713–3734.
Saxton J. M., Lyon I. C., and Turner G. (1998) Correlated chemical and
isotopic zoning in carbonates in the Martian meteorite ALH84001.
Earth Planet. Sci. Lett. 160, 811– 822.
Schwandt C. S., McKay G. A., and Lofgren G. E. (1999) FESEM
imaging reveals previously unseen detail and enhances interpretations of ALH84001 carbonate petrogenesis. Lunar Planet. Sci. XXX,
1346 –1347.
Scott E. D., Yamaguchi A., and Krot A. N. (1997) Petrologic evidence
for shock melting of carbonates in the Martian meteorite ALH84001.
Nature 387, 377–379.
Scott E. R. D., Krot A. N., and Yamaguchi A. (1998) Carbonates in
fractures of Martian meteorite ALH84001: Petrologic evidence for
impact origin. Meteor. Planet. Sci. 33, 709 –719.
Shearer C. K., Leshin L. A., and Adcock C. T. (1999) Olivine in
Martian meteorite Allan Hills 84001: Evidence for a high-temperature origin and implications for signs of life. Meteor. Planet. Sci. 34,
331–339.
321
Sheng Y. J., Wasserburg G. J., and Hutcheon I. D. (1992) Selfdiffusion in spinel and in equilibrium melts: Constrains on flash
heating of silicates. Geochim. Cosmochim. Acta 56, 2535–2546.
Treiman A. H. (1995) A petrographic history of Martian meteorite
ALH84001: Two shocks and an ancient age. Meteoritics 30, 294 –
302.
Treiman A. H. (1998) The history of Allan Hills 84001 revised:
Multiple shock events. Meteor. Planet. Sci. 33, 753–764.
Treiman A. H. and Romanek C. S. (1998) Chemical and stable isotopic
disequilibrium in carbonate minerals of martian meteorite ALH
84001: Inconsistent with high formation temperature. Meteor.
Planet. Sci. 33, 737–742.
Valley J. V., Eiler J. M., Graham C. M., Gibson Jr. E. K., Romanek
C. S., and Stolper E. M. (1997) Low-temperature carbonate concretions in the Martian meteorite ALH84001: Evidence from stable
isotopes and mineralogy. Science 275, 1633–1638.
Watson, E. B. (1994) Diffusion in volatile-bearing magmas. In Volatiles in Magmas (ed. M. R. Carroll and J. R. Holloway), Vol. 30, pp.
371– 411. Mineralogical Society of America.