Hydrothermal processes at Gusev Crater: An evaluation of Paso
Robles class soils
A. S. Yen,1 R. V. Morris,2 B. C. Clark,3 R. Gellert,4 A. T. Knudson,5 S. Squyres,6
D. W. Mittlefehldt,2 D. W. Ming,2 R. Arvidson,7 T. McCoy,8 M. Schmidt,8 J. Hurowitz,1
R. Li,9 and J. R. Johnson10
Received 2 August 2007; revised 16 November 2007; accepted 18 December 2007; published 22 April 2008.
[1] The Mars Exploration Rover Spirit analyzed multiple occurrences of sulfur-rich, light-
toned soils along its traverse within Gusev Crater. These hydrated deposits are not readily
apparent in images of undisturbed soil but are present at shallow depths and were exposed
by the actions of the rover wheels. Referred to as ??Paso Robles?? class soils, they are
dominated by ferric iron sulfates, silica, and Mg-sulfates. Ca-sulfates, Ca-phosphates, and
other minor phases are also indicated in certain specific samples. The chemical
compositions are highly variable over both centimeter-scale distances and between the
widely separated exposures, but they clearly reflect the elemental signatures of nearby
rocks. The quantity of typical basaltic soil mixed into the light-toned materials prior to
excavation by the rover wheels is minimal, suggesting negligible reworking of the
deposits after their initial formation. The mineralogy, geochemistry, variability, association
with local compositions, and geologic setting of the deposits suggest that Paso Robles
class soils likely formed as hydrothermal and fumarolic condensates derived from magma
degassing and/or oxidative alteration of crustal iron sulfide deposits. Their occurrence as
unconsolidated, near-surface soils permits, though does not require, an age that is
significantly younger than that of the surrounding rocks.
Citation: Yen, A. S., et al. (2008), Hydrothermal processes at Gusev Crater: An evaluation of Paso Robles class soils, J. Geophys.
Res., 113, E06S10, doi:10.1029/2007JE002978.
1. Introduction
[2] The selection of Gusev Crater as a landing site for the
Mars Exploration Rover Spirit was based on the goal of
characterizing the nature and extent of aqueous processes at
the Martian surface [Golombek et al., 2003; Cabrol et al.,
2003]. The presence of Ma?adim Vallis, a 900 km valley
debouching into the crater from the south, suggested that
Gusev Crater was a location that might permit study of
fluvial, lacustrine, and other aqueous processes. In addition,
the proximity of Apollinaris Patera, an ancient volcano
250 km north of Gusev, suggested the possibility of mag-
ma-water interactions.
[3] After more than 150 sols on the Gusev plains where
the geology was dominated by relatively unweathered
olivine-rich basaltic soils and rocks, the first compelling
evidence for significant water-related weathering at Gusev
Crater was the discovery of goethite [Morris et al., 2006],
an iron oxyhydroxide, which requires the presence of water
for its formation. In contrast to this finding, however, are
numerous examples indicating that interactions with liquid
water have been relatively minimal. For example, the
discovery of an exposed ultramafic sequence of rocks
[Mittlefehldt et al., 2006] where over 70% of the iron is
in unaltered olivine and the ferric to total iron ratio is only

0.1, comparable to unaltered shergottites [Burns and
Martinez, 1991], indicates that certain materials within
Gusev Crater have been minimally altered by aqueous
processes. Even some of the most altered rocks in the
Columbia Hills, such as the Watchtower Class rocks that
dominate much of the northern side of Husband Hill,
apparently underwent nearly isochemical alteration, sug-
gesting very low water-to-rock ratios [Ming et al., 2006;
Squyres et al., 2006].
[4] Analyses of several isolated occurrences of light-
toned soil deposits excavated by the rover wheels provide
additional compelling evidence of significant water-related
activity at the Spirit landing site. These samples occur in
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E06S10, doi:10.1029/2007JE002978, 2008
1Jet Propulsion Laboratory, California Institute of Technology,
Pasadena, California, USA.
2NASA Johnson Space Center, Houston, Texas, USA.
3Lockheed Martin Corporation, Littleton, Colorado, USA.
4Department of Physics, University of Guelph, Guelph, Ontario,
Canada.
5Department of Earth and Planetary Sciences, Arizona State University,
Tempe, Arizona, USA.
6Department of Astronomy, Cornell University, Ithaca, New York,
USA.
7Department of Earth and Planetary Sciences, Washington University,
St. Louis, Missouri, USA.
8National Museum of Natural History, Smithsonian Institution,
Washington, D.C., USA.
9Center for Mapping, Ohio State University, Columbus, Ohio, USA.
10U.S. Geological Survey, Flagstaff, Arizona, USA.
Copyright 2008 by the American Geophysical Union.
0148-0227/08/2007JE002978
E06S10 1 of 19
two groupings: (1) The sulfur-rich Paso Robles class depos-
its, named after the first sample of such material that was
analyzed by all of the rover instruments [Gellert et al.,
2006; Morris et al., 2006; Ming et al., 2006], and (2) the
distinct, but related, silica-dominated materials encountered
after sol 1150 (S. W. Squyres et al., Discovery of silica-rich
deposits on Mars by the Spirit Rover, submitted to Science,
2008). There are a significant number of light-toned soils
which were not analyzed by the in situ instruments, pre-
venting classification into these two or possibly alternative
groups. The focus of this work is the Paso Robles class
soils, and a summary of the previously published and new
sets of instrument data, reconstructions of possible miner-
alogies, candidate formation processes, and implications of
these findings are discussed in sections 2?9.
2. Geologic Setting
[5] The Paso Robles class of light-toned soils encoun-
tered by Spirit is found on and around Husband Hill and
within the Inner Basin (Figure 1). There are three confirmed
locations with exposures of this soil class: The original
??Paso Robles?? site, ??Arad?? (also known as ??Dead Sea??),
and ??Tyrone.?? (See R. E. Arvidson et al. (Overview of the
Spirit Mars Exploration Rover mission to Gusev Crater,
Columbia Hills: Independence outcrop to the ascent onto
Home Plate, manuscript in preparation, 2008) for a descrip-
tion of the rover traverse.) Several additional occurrences of
light-toned soil which may also be Paso Robles class
deposits are indicated in Figure 1. The deposits at the
Tyrone location are found in a shallow depression
(Figures 2 and 3a) bounded primarily by the extensions of
??Low?? and ??Mitcheltree?? ridges to the west and the slopes
of McCool Hill to the east. The Arad exposures occur at the
base of a hill at the western margin of a 
150 m wide
valley. There is an abrupt change in slope at this location,
but unlike the Tyrone deposit, the Arad samples are not
found within a defined topographic minimum (Figure 3b).
Similarly, one of the possible exposures (not confirmed by
in situ analyses) of Paso Robles class materials, referred to
as ??Shredded,?? also occurs near the base of a hillslope
(Figure 3c). The location of the original Paso Robles
measurements, however, does not appear to be influenced
by topography and is on a slope of approximately 12

(Figure 3d). The suspected Paso Robles class materials near
??Wishstone?? (again, in situ analyses were not conducted)
are also on comparable slopes (Figure 3e).
[6] The elevations of the Paso Robles class soils span a
significant portion of the topographic range encountered by
Spirit and do not exhibit a consistent pattern. For reference,
the summit of Husband Hill is approximately 106 m above
the elevation of the lander on the Gusev Plains. Exposures
of Paso Robles class soils occur at approximately 33 m
(Arad), 45 m (Tyrone), and 66 m (Paso Robles) above the
lander elevation. The suspected, but not confirmed, expo-
Figure 1. Locations of Paso Robles class soil exposures analyzed in detail by the rover instruments are
shown in green superimposed on a High Resolution Imaging Science Experiment (HiRISE) image of the
Husband Hill complex. The locations indicated in yellow represent the major suspected Paso Robles class
soil exposures (not confirmed by the entire instrument complement). The location of the red circle
indicates distinct, but likely related, deposits of amorphous silica. The red line illustrates the traverse of
the rover.
E06S10 YEN ET AL.: HYDROTHERMAL PROCESSES AT GUSEV CRATER
2 of 19
E06S10
sures of Paso Robles class soils at Shredded and Wishstone
are found 
24 and 
45 m above the lander, respectively.
3. Instrument Analyses
[7] Paso Robles class deposits tend to be fine-grained and
poorly compacted, which are soil properties that present
challenges to rover mobility. In fact, several occurrences of
light-toned soil, including Paso Robles itself, were first
discovered when wheel slip impeded rover motion and
excavated soil deposits around rover wheels. Microscopic
Imager (MI) [Herkenhoff et al., 2003] images (e.g., Figure 4)
show weakly consolidated samples with clumps that may
have been induced by mixing by the rover wheels. Pancam
[Bell et al., 2003] false color images of the major occur-
rences are shown in Figure 5. This visible/near-infrared data
set exhibits relatively uncommon spectral features which are
consistent with ferric sulfates with various levels of hydra-
tion [Johnson et al., 2007]. Evidence that these deposits are
hydrated is provided by a strong 6 mm molecular water
feature observed in Mini-Thermal Emission Spectrometer
(Mini-TES) [Christensen et al., 2003] spectra (Figure 6).
The focus of the work described here is on data collected
from the Mo?ssbauer [Klingelho?fer et al., 2003] and Alpha
Particle X-ray Spectrometers (APXS) [Rieder et al., 2003].
Figure 2. Pancam false color view from the ??McMurdo??
panorama of the Tyrone deposit of Paso Robles class soils.
The view is toward the east (facing McCool Hill). The
exposed material (center) resides in a localized topographic
depression. Material dragged from Tyrone by the rover
wheels was deposited at the edge of the tracks.
Figure 3. Paso Robles class soil deposits are found in a variety of situations relative to local topography
(1 m topographic contours shown in red; white indicates areas not covered in the original images):
(a) Tyrone in a localized depression, (b) Arad and (c) Shredded at the base of slopes, and (d) Paso
Robles and (e) Wishstone on slopes of 
12
. Note that the Shredded and Wishstone exposures are
not confirmed exposures of sulfur-rich soils.
E06S10 YEN ET AL.: HYDROTHERMAL PROCESSES AT GUSEV CRATER
3 of 19
E06S10
Table 1 lists the analyses conducted to date on this class of
soil samples by these two instruments.
[8] The first in situ measurement of this class of soils was a
short (
30 min) APXS analysis on a target called Pasadena
Marengo at what is now referred to as the Paso Robles site. In
this daytime measurement, elevated instrument temperatures
significantly compromised the data quality but did not
prevent the extraction of a signal that indicated a level of
sulfur far greater than in any previous measurement at Gusev
Crater. This exploratory measurement was followed by a
longer, nighttime APXS analysis (Pasadena Paso Robles) at
appropriate detector temperatures. An associated Mo?ssbauer
measurement of this sample was also obtained. After driving
away from this location before the results could be fully
assessed, the science team decided to return to the location
for a more detailed evaluation, resulting in APXS and
Mo?ssbauer data from the Paso Robles2 PasoLight1 target.
Measurements were also made in an attempt to extract the
composition of the light-toned material from mixtures with
typical Gusev (basaltic) soils. These two additional measure-
ments (Table 1), however, were too close in composition to
typical Gusev soils to be useful in mixing models. Thus, the
Pasadena Paso Robles and the Paso Robles2 PasoLight1
analyses are the most useful of the data collected from this
site for understanding the composition and origin of this class
of soils. Interpretations of these data are given by Morris et
al. [2006] and Ming et al. [2006]. In the text that follows,
analyses at the Paso Robles ??location?? or ??site?? refers to
these two measurements, and Paso Robles ??class?? soils refer
to all exposures of light-toned soils at the Gusev site which
have related compositions.
[9] The APXS and Mo?ssbauer analyses of the light-toned
soils at the Arad site (Samra and Hula) have not been
previously published and provide additional insight into the
formation of Paso Robles class soils. These samples are
south of Husband Hill in the Inner Basin (Figure 1), and
provide the measurements with the least contamination from
surrounding basaltic soils. Each target was analyzed by the
APXS, and Mo?ssbauer data were collected from Samra.
[10] During the rush to find a north facing slope that
would allow Spirit to survive its second winter on Mars, the
rover became mired in the Tyrone deposit of light-toned
soils, just southeast of Home Plate (Figure 1) (R. E.
Arvidson et al., manuscript in preparation, 2008). A quick
escape was necessary, and APXS and Mo?ssbauer analyses
were not conducted at that time. Studies of Tyrone were
resumed the following spring. However, because of the
possible risk of becoming permanently stuck at Tyrone,
measurements were instead made on the material that had
been fortuitously dragged out of the Tyrone area by the
rover wheels and deposited on the edge of the wheel tracks
(Berkner Island and Mount Darwin). The Berkner Island
sample is approximately 40 m from the original Tyrone
location while the Mount Darwin target is much closer at a
distance of approximately 10 m.
3.1. Mo?ssbauer Data
[11] Mo?ssbauer spectra of Paso Robles class soils are
dominated by narrow doublets indicative of ferric iron
(Figure 7). The isomer shifts (d), quadrupole splittings
(DEQ), and full widths at half maxima (G) of fits to the ferric
doublets are listed in Table 2 (see Morris et al. [2006] for a
description of these Mo?ssbauer parameters). The dominant
ferric doublet is assigned to Fe3+ sulfate [Morris et al., 2006,
2007]. Variations in DEQ and G, which are significantly
greater than the 0.02 mm/sec uncertainties, may reflect ferric
sulfates with different hydration states, different degrees of
crystallinity, and/or different levels of substitutional impuri-
ties (e.g., Al3+ for Fe3+). Because of the range of values of G
and especially DEQ, we attempted fits with two ferric
doublets, but the quality of the fits did not improve signifi-
cantly. Without a compelling reason to pursue more complex
models, the results of the single doublet fits to the ferric
signatures of Arad Samra and the two Paso Robles analyses
are reported in Table 2. Other methods for fitting the data
included the use of the ferric doublet parameters for Arad
Samra as a constraint on the fits to the Paso Robles samples,
but this approach did not produce reasonable results.
[12] The Tyrone Berkner Island and Mount Darwin sam-
ples had major contributions from typical basaltic soil, and
the fidelity of the fits was enhanced by constraining the
parameters of this soil component. The results of using the
Mo?ssbauer spectrum of the nearby Bear Island basaltic soil as
a constraint on the fits to these two light-toned soils, as well as
single doublet fits to the ferric signatures, are listed in Table 2.
The use of other basaltic soil spectra in the constraints were
also investigated, and the differences were within the uncer-
tainty of the fits. The measurement at Tyrone Mount Darwin
was further compromised by a larger than desired distance
between the instrument sensor head and sample, resulting in
poor counting statistics. In order to obtain area fractions of
the iron minerals in Mount Darwin, it was necessary to
constrain parameters of the ferric doublet using results from
Tyrone Berkner Island measurement.
[13] Table 3 lists the percentage of total Fe associated
with specific Fe-bearing phases as determined by the
Figure 4. Mosaic of four Microscopic Imager images of
light-toned soil at Paso Robles (sol 400, approximately 5 
5 cm). The imprint of the Mo?ssbauer contact sensor (37 mm
outer diameter) is visible in the mosaic.
E06S10 YEN ET AL.: HYDROTHERMAL PROCESSES AT GUSEV CRATER
4 of 19
E06S10
Mo?ssbauer spectrometer. The uncertainty is ?2% (absolute)
for the Paso Robles and Arad measurements. The uncer-
tainties in the Tyrone samples are higher because of larger
quantities of basaltic soil and poor instrument positioning
for the Mount Darwin analysis. The two measurements at
Paso Robles, which are separated by a distance of only 
1
m, exhibit significant differences (over a factor of two) in
the quantity of hematite. The measurement at the Arad
location has the least amount of contamination from typical
basaltic soils and one of the highest ferric to total iron ratios
(0.91) measured to date by either rover.
3.2. Alpha Particle X-Ray Spectrometer Data
[14] The elemental compositions of Paso Robles class soil
samples were determined by the APXS (Tables 4a and 4b). A
defining attribute of these light-toned deposits is a sulfur
content in excess of 30 wt % SO3, unless the sample is
significantly contaminated by typical Gusev basaltic soil
Figure 5. False color Pancam images of suspected (Shredded and Wishstone) and confirmed (all others
listed) Paso Robles class soil exposures: (a?c) Shredded region (sol 180, p2387, p2544, p2545), (d)
Wishstone area (sol 351, p2588), (e?f) PasoRobles1 (sol 400, p2551; sol 404, p2559), (g?h)
PasoRobles2 (sol 431, p2599, p2530), (i) PasoRobles region (sol 445, p2548), (j? l) Arad area (sol 721,
p2538, sol 723, p2542, sol 725, p2547), (m) Tyrone (sol 788, p2396), (n) Berkner Island (from
??McMurdo Panorama??), (o) Mount Darwin (sol 1102, p2553).
E06S10 YEN ET AL.: HYDROTHERMAL PROCESSES AT GUSEV CRATER
5 of 19
E06S10
(Tables 4a and 4b). This elevated concentration of sulfur is the
highest of all measurements made by either rover, including
Meridiani outcrop rocks. The amounts of Na, Al, and K in
Paso Robles class deposits are consistently lower than in
typical Gusev soils and may represent minor levels of con-
tamination from basaltic material. Calculations constraining
the extent of basaltic soil mixing are described in section 4.
3.2.1. Compositional Variability
[15] Aside from elevated sulfur and variable contamina-
tion from basaltic soils, the collection of Paso Robles class
materials does not display any clear or consistent chemical
signatures. The two measurements of light-toned material at
the Paso Robles site, which are separated horizontally by
approximately 1 m, have differences in Al, P, Cl, K, Ti, Cr,
Fe and Ni of 20% or greater. Similarly, the Samra and Hula
samples at the Arad location are approximately 30 cm from
each other and have greater than 30% differences in Na,
Mg, Al, Si, Cl, Ca, Ti, Mn, and Zn. When comparing
average measurements at Paso Robles with those at Arad,
all measured elements with the exception of Fe and S are
different in concentration by over 30%, with many differ-
ences exceeding 100%. Figures 8 and 9 illustrate the
compositional variability of these analyses.
[16] The two measurements of material dragged from the
Tyrone site are not substantially different from one another
because they represent samples that were mixed within the
rover wheel. The primary difference between the Berkner
Island and the Mount Darwin analyses is a larger amount of
basaltic soil in the sample (Berkner Island) that was dragged
further from its origin. In comparison to the samples from
the Paso Robles and Arad sites, the Tyrone material ana-
lyzed by the APXS contains approximately double the
concentrations of Zn and Ca. It is clear that the chemical
compositions of this class of soils are highly variable over
both short and long distance scales. The variability in
elemental composition precludes the possibility of extensive
redistribution of materials from distant sources, which
would tend to homogenize samples, and may be related to
the observed color variations in these materials observed in
Pancam data [Johnson et al., 2007].
3.2.2. Localized Processes
[17] Chemical signatures of nearby rocks are evident in
Paso Robles class soil deposits. For many elements, the
compositions of the Paso Robles class soils are more similar
to surrounding rocks than to other samples within this soil
class (Figures 10 and 11). The most obvious indicator is the
significantly elevated phosphorous at Paso Robles itself (4.7
and 5.6 wt % P2O5), where the nearby rocks are of Wish-
stone and Watchtower class. Members of this rock class
have measured phosphate levels of up to 5.2% P2O5, the
highest of all rocks found by either rover [Ming et al., 2006;
Gellert et al., 2006]. A related indicator comes from a few
tens of ppm yttrium, a trace element identified in Wish-
stone/Watchtower class rocks [Clark et al., 2007], that is
also present in the soil at the Paso Robles site (Figure 12).
Figure 6. Mini-TES observations of Paso Robles class
soils show enhanced hydration in a peak at 6 mm (dashed
line in all plots). The light-toned soils are compared with
dark disturbed soils in rover tracks at (a) Paso Robles (sols
400?404), (b) Arad (sols 721?724), and (c) Tyrone (sols
1095?1100). The exposures at Tyrone are the largest and
include bright materials that are tinted yellow in addition to
white [Johnson et al., 2007]. Averages of the Mini-TES
observations from each color unit are shown. In all areas the
disturbed dark soil in rover tracks shows a small hydration
peak due to the presence of dust [Christensen et al., 2004]
mixed with the basaltic sand in the soil. All spectra collected
after sol 420 spectra have been corrected for dust
accumulations on the mirror after the method described by
A. T. Knudson et al. (manuscript in preparation, 2008).
E06S10 YEN ET AL.: HYDROTHERMAL PROCESSES AT GUSEV CRATER
6 of 19
E06S10
Rare earth elements such as Y are largely held in phosphatic
minerals such as whitlockite and apatite in Martian meteor-
ites, and is therefore consistent with the elevated P. Another
characteristic aspect of the rocks near Paso Robles reflected
in the chemical analyses of nearby light-toned soils is
anomalously low Cr (at detection limit) and Ni (
100 ppm).
[18] In contrast to the Paso Robles site, the soil measure-
ments at Arad, which are on the south side of Husband Hill
and not surrounded by phosphate-rich rocks, have only 0.5
wt % P2O5. Instead of showing similarities to Wishstone/
Watchtower rocks, the Arad measurements appear to reflect
the high Mg, Cr, and Ni and the low Al, K, Ca, and Zn of
the Algonquin class rocks [Squyres et al., 2007] exposed
less than 150 m away. Yttrium is not observed in Arad soils.
[19] The Mount Darwin analysis of Tyrone materials con-
tains the largest amount of CaO measured for Gusev samples
(>9.0 wt %). This sample is diluted by 
40% contamination
from basaltic soil (section 4.5), which averages 
6.3 wt %
CaO, so the true Ca content of the Tyrone deposits could be in
excess of 11 wt %. Within 40 m of the Tyrone exposure is the
Halley Brunt sample, also with one of the highest CaO
concentrations measured at Gusev (8.8 wt %). Unlike the
previous examples of Paso Robles class deposits containing
signatures of nearby rocks, the Halley Brunt target is not a
primary igneous rock, and the possibility that these two
materials formed via the same process cannot be ruled out.
Nonetheless, taken collectively, it is clear that these light-
toned soils are diverse in composition and chemically related
to local rocks.
[20] Ratios of the measured elements in the Paso Robles
class soils indicate that the chemical relationships to nearby
rocks are not a result of physical admixtures of comminuted
fines. The ratio of P to Ti in light-toned soils at the Paso
Robles location, for example, is 4 to 5 times greater than in
the surrounding rocks. In addition, the Samra measurement
has the elevated Ni signature of local rocks, but no indica-
tions of the olivine which may be the host of the Ni. The
chemical enrichments in the Paso Robles class soils which
establish the connection to local rocks are in elements that
are readily mobilized in aqueous solutions.
Table 1. Summary of APXS and Mo?ssbauer Spectrometer Data Collected from Paso Robles Class Soil Deposits
Sol(s) Target APXS MB Comments
399 Pasadena_Marengo 30 min 0 Warm detector. Indication of significant S
400?403 Pasadena_PasoRobles 3 h 47 h First quality measurement of this material
426?428 PasoRobles2_PasoDark 6.5 h 6 h Typical dark soil at Gusev
427?431 PasoRobles2_PasoLight1 6 h 27 h
430 PasoRobles2_PasoDarkLight 9 h 0 Minor enhancements in P, S, Ti, Fe
723?726 Arad_Samra 13 h 33 h Least amount of basaltic soil contamination
724 Arad_Hula 12 h 0
1013?1016 Tyrone_BerknerIsland1 3 h 44 h Deposit dragged from Tyrone by rover wheels
1098?1101 Tyrone_MountDarwin 3 h 22 h Deposit dragged from Tyrone by rover wheels
Figure 7. Mo?ssbauer spectra for the S-rich, Paso Robles Class soils at Gusev crater. (a)
Pasadena_PasoRobles; (b) PasoRobles2_PasoLight1; (c) Arad_Samra; and (d) Tyrone_BerknerIsland1
(model B, see Table 3). The spectrum for another S-rich soil (Tyrone MountDarwin) is similar to Figure
7d. Component subspectra for Fe in different Fe-bearing phases are shown by different colors, and the
computed fit is shown by the solid red line. Values corresponding to TC/BC  1.0 = 0 and its maximum
value are shown on each y axis. TC, total counts; BC, baseline counts.
E06S10 YEN ET AL.: HYDROTHERMAL PROCESSES AT GUSEV CRATER
7 of 19
E06S10
3.2.3. Hydration
[21] Scattering of the Pu X-ray lines emitted by the 244Cm
source of the APXS depends on the concentration of light
elements that are not directly detected by the instrument,
and this effect has been used to determine the extent of
hydration in the Paso Robles class soils [Campbell et al.,
2008]. This approach is completely independent of the
Mini-TES results shown in Figure 6, and produces results
indicating 16 ? 8, 15 ? 6, 18 ? 5, 65
+8 wt % water for the
analyses of Pasadena Paso Robles, Paso Robles2 Paso
Light1, Arad Samra, and Arad Hula, respectively [Campbell
et al., 2008]. The specific mineralogical composition of the
hydrated/hydroxylated phases, however, is not constrained
by this analysis.
4. Mineral Mixtures
[22] The combination of the Mo?ssbauer parameters for
the narrow ferric doublet, the high Fe3+/FeT ratio in the
Mo?ssbauer data, and the observation that Fe and S are the
two most abundant elements after Si indicate that a primary
constituent of Paso Robles class soils is ferric sulfate.
Basaltic soil, other sulfates, and various nonsulfates are
also present in these samples. Using a combination of
APXS and Mo?ssbauer data, sections 4.1?4.7 attempt to
constrain the minerals phases.
4.1. Arad Samra
[23] The Samra sample at the Arad site is the most
chemically pure example of Paso Robles class soils. The
four most abundant of the measured elements expressed as
oxides (SiO2, SO3, FeO, and MgO) constitute over 95 wt %
of the material. Because of this relatively simple chemistry,
calculating a candidate composition for Samra is more
straightforward than for the other samples.
4.1.1. Basaltic Soil Contamination
[24] The Samra sample was originally covered with
basaltic soil prior to excavation by the rover wheels. An
Table 2. Mo?ssbauer Parameters d, DEQ, and G for the Fe3D2 (Fe3+-Sulfate) Doublet Subspectruma
Targetb
Generic Name (Assignment) Fe3D2 (Fe3+-Sulfate)
T, Kd, mm/s DEQ, mm/s G, mm/s
A401SD0 (Pasadena_PasoRobles)c 0.42d 0.62 0.60 190?280
A429SD0 (PasoRobles2_PasoLight1)c 0.43 0.55 0.68 200?280
A723SD0 (Arad_Samra)e 0.41 0.51 0.54 190?270
A1015SD0 (Tyrone_BerknerIsland1) model Ae,f 0.42 0.81 0.63 190?250
A1015SD0 (Tyrone_BerknerIsland1) model Be,g 0.44 0.79 0.63 190?250
A1101SD0 (Tyrone_MountDarwin) model Ae,f [0.42]h 0.78 0.57 190?270
A1101SD0 (Tyrone_MountDarwin) model Be,g [0.44] 0.75 0.55 190?270
aMB parameters were calculated from spectra summed over the temperature interval given in the last column. Values of d are with respect to metallic iron
foil at the same temperature as the sample; d is isomer shift; DEQ is quadrupole splitting; G is full width.
bTarget naming convention: AwwwwSD0 (Feature-name_Target-name); A = MER-A (Gusev Crater); wwww is Gusev Crater sol number that data
product was returned to Earth; SD is disturbed soil. Alphanumeric strings before parentheses are unique target identifiers.
cData from Morris et al. [2006].
dMB parameter uncertainty is ?0.02 mm/s.
eData from Morris et al. [2007].
fFrom fit using one Fe3+ doublet.
gFrom fit using A1018SD0 (BearIsland_BearIsland) as the model for nonsulfate soil subspectrum.
hMB parameters in brackets are constraints used in the fitting procedure.
Table 3. Percentage of Total Fe Associated With Specific Fe-bearing Phases (f Factor Corrected) and Fe3+/FeT (190?270 K) for Paso
Robles Class Soils at Gusev Crater
Generic Namea
Sum Fe3+/FeTFe2D1 Fe2D2 Fe3D1 Fe3D2 Fe3D2 Fe3S1 Fe2.5S1 Fe3S2
Phase Assignmentb Target Ol Px npOx Fe3Sulfate Mt Mt(3) Mt(2.5) Hm
A401SD0 (Pasadena_PasoRobles)c 3d 10 0 62 5 2 3 20 100e 0.86f
A429SD0 (PasoRobles2_PasoLight1)c 10 8 0 69 6 1 5 7 100 0.79
A723SD0 (Arad_Samra)g 7 3 0 86 0 0 0 4 100 0.90
A1015SD0 (Tyrone_BerknerIsland1) model Ag,h 22 23 0 43 0 0 0 12 100 0.56
A1015SD0 (Tyrone_BerknerIsland1) model Bg,i 22 23 [9]j 34 0 0 0 12 100 0.56
A1101SD0 (Tyrone_MountDarwin) model Ag,h 23 30 0 42 0 0 0 5 100 0.47
A1101SD0 (Tyrone_MountDarwin) model Bg,i 22 31 [12]j 30 0 0 0 5 100 0.47
aNames have the format FeXYZ: X is oxidation state of Fe, Y is D or S (doublet or sextet), Z is sequential number for subspectra having the same FeXY.
bOl, olivine; Px, pyroxene; npOx, nanophase ferric oxide; Fe3Sulfate = Fe3+-sulfate; Mt = magnetite; and Hm, hematite.
cData are from Morris et al. [2006].
dUncertainty in subspectral area is ?2% absolute unless stated otherwise.
eBecause Mt = Mt(3) + Mt(2.5), Sum = Ol + Px + npOx + Fe3Sulfate + Mt + Hm. For all samples, ilmenite = 0.
fFe3+/FeT = (npOx + Fe3D2 + Mt(3) + 0.5(Mt(2.5)) + Hm)/Sum. Uncertainty in Fe
3+/FeT is ?0.03.
gData are from Morris et al. [2007].
hFrom fit using one Fe3+ doublet. Uncertainty in subspectral area is ?4% absolute.
iFrom fit using A1018SD0 (BearIsland_BearIsland) as the non-sulfate soil spectrum with AnpOx 
 0.25(AOl +APx) used as a constraint for the fitting
procedure. Uncertainty in subspectral area is ?4% absolute.
jMB parameters in brackets are constraints used in the fitting procedure.
E06S10 YEN ET AL.: HYDROTHERMAL PROCESSES AT GUSEV CRATER
8 of 19
E06S10
upper limit on the amount of basaltic material mixed into
the sample can be estimated by calculating how much of an
average Gusev soil composition (Tables 4a and 4b) can fit
within the Samra APXS measurement before any elemental
concentrations are exceeded. Even without direct measure-
ments of the chemical composition of the basaltic soil
adjacent to Samra, this approach is appropriate given the
compositional uniformity of typical Gusev soil [Yen et al.,
2005]. The amounts of Na and K in Samra are essentially at
the detection limits. Accounting for the amount of these
elements allowed by the measurement uncertainty results in
a maximum of 5 wt % contamination from basaltic soil.
[25] Another approach for estimating the quantity of
basaltic soil in Samra is using the Mo?ssbauer results.
Assuming that the olivine and pyroxene components of
the Samra measurement originate from basaltic contamina-
tion, an average soil composition can be used to calculate
the contribution from all measured elements. In this case,
10% of the iron is determined to be from the ferrous phases
(Table 3). Scaling an average basaltic soil composition to
match this iron content produces an estimate of 13 wt % for
the amount of contamination from typical soils.
[26] A third technique involves extrapolation from a plot of
Na versus Al for a variety of samples analyzed at Gusev
Crater (Figure 13). Assuming that the roughly linear rela-
tionship represents a plagioclase feldspar component and that
the only source of plagioclase in the measured material is
from the basaltic soil, the upper limit for the basaltic soil
component is approximately 3 wt %. This result is not
independent of the limits based on Na and K calculated
above, because Na is also used in this calculation. However,
this approach does show that constraints using Al, in addition
to Na and K, yield approximately consistent results.
[27] The difference between the 3% to 5% basaltic
contamination calculated from APXS data and the 13%
inferred from Mo?ssbauer data may result from analysis area
and depth differences. The APXS samples a 
30 mm
diameter circular area to a depth of 2 to 100 mm, depending
upon the energy of the X-ray [Rieder et al., 2003]. The
Mo?ssbauer analyzes a 15 mm diameter circular region to a
depth, depending on density, of 0.1 to 3 mm [Klingelho?fer et
al., 2003], more than an order of magnitude deeper than the
APXS analysis. Lateral inhomogeneities may have resulted
in more basaltic soil within the larger APXS field of view.
Alternatively, the mixing of the dust and sand mantle by the
rover wheels may have buried basaltic material under the
immediate surface to a depth sampled by the Mo?ssbauer
spectrometer, but not by the APXS. The relative amounts of
the two ferrous phases in Samra is not the 
1:1 ratio
observed for typical basaltic soils, and the possibility that
a small amount of an unidentified ferrous mineral phase
contributes to the peaks assigned to olivine and/or pyroxene
cannot be excluded. Given that the calculations described in
this section for the Samra sample are based primarily on
elemental chemistry, the 5% basaltic contamination derived
from the APXS results is used.
[28] A significant implication of the relatively small
amounts of basaltic contamination in the Samra deposit is
that these light-toned materials have not been extensive
redistributed from the location of initial formation. Fine
grained basaltic sand and dust are pervasive in the Martian
surface environment, and it is unlikely that the Paso Robles
class materials could have been transported very far without
developing a significant admixture of basaltic grains. The
amount of typical Gusev soil that is observed in this deposit
likely results from mixing of the surficial basaltic dust and
sand layer by the rover wheels during the excavation
process. These light-toned soils likely formed in the places
they have been found.
4.1.2. Sulfates and Other Phases
[29] The Mo?ssbauer analysis of Samra indicates that 4%
of the iron (equivalent to 1% of the sample) is present as
Table 4a. APXS Elemental Compositions Expressed as Weight Percent of Common Oxidesa
Sol Sample Na2O MgO Al2O3 SiO2 P2O5 SO3 Cl K2O CaO TiO2 Cr2O3 MnO FeO Ni Zn Br
401 Pasadena_PasoRobles 1.64 5.53 4.13 21.8 5.61 31.7 0.55 0.19 6.84 0.62 0.04 0.25 21.0 109 98 494
427 PasoRobles2_PasoLight1 1.42 5.19 6.27 24.9 4.69 31.6 0.73 0.40 6.94 0.88 0.35 0.30 16.1 561 116 478
428 PasoRobles2_PasoDark 3.24 8.74 10.4 46.1 0.96 5.68 0.55 0.39 6.25 0.89 0.38 0.33 15.9 414 209 52
430 PasoRobles2_PasoDarkLight 3.02 8.64 9.72 43.2 1.38 7.73 0.58 0.38 6.39 0.99 0.37 0.34 17.2 395 235 69
723 Arad_Samra 0.14 4.04 1.21 37.7 0.50 35.1 0.31 0.01 0.81 0.53 0.48 0.05 19.1 748 67 92
724 Arad_Hula 0.39 11.5 2.46 28.4 0.51 32.8 0.66 0.04 5.15 0.25 0.46 0.13 17.2 887 91 72
1013 Tyrone_BerknerIsland1 1.98 7.09 6.21 36.9 0.88 22.1 0.39 0.28 7.62 0.50 0.43 0.25 15.2 747 191 82
1098 Tyrone_MountDarwin 1.06 6.54 5.06 32.9 0.95 27.6 0.42 0.27 9.02 0.51 0.51 0.18 14.8 781 246 86
- Average Basaltic Soila 3.07 8.62 10.3 46.5 0.87 5.76 0.68 0.43 6.31 0.88 0.34 0.33 15.9 464 271 43
aNi, Zn, and Br are in ppm. Based on 20 high-quality measurements of fine-grained samples with no known irregularities.
Table 4b. Uncertainties in the APXS Values Listed in Table 4a
Sol Target Na2O MgO Al2O3 SiO2 P2O5 SO3 Cl K2O CaO TiO2 Cr2O3 MnO FeO Ni Zn Br
401 Pasadena_PasoRobles 0.21 0.09 0.07 0.24 0.14 0.30 0.02 0.06 0.06 0.07 0.03 0.01 0.12 49 18 26
427 PasoRobles2_PasoLight1 0.18 0.08 0.08 0.22 0.11 0.25 0.02 0.06 0.05 0.07 0.03 0.01 0.09 46 12 20
428 PasoRobles2_PasoDark 0.21 0.11 0.13 0.43 0.07 0.07 0.01 0.06 0.05 0.07 0.03 0.01 0.11 43 13 15
430 PasoRobles2_PasoDarkLight 0.26 0.12 0.11 0.40 0.08 0.08 0.01 0.06 0.05 0.07 0.03 0.01 0.11 40 12 15
723 Arad_Samra 0.14 0.06 0.03 0.26 0.07 0.24 0.01 0.05 0.01 0.06 0.03 0.00 0.07 41 8 14
724 Arad_Hula 0.16 0.12 0.04 0.26 0.07 0.27 0.01 0.05 0.04 0.06 0.03 0.01 0.10 44 9 16
1013 Tyrone_BerknerIsland1 0.28 0.13 0.11 0.40 0.09 0.24 0.01 0.06 0.07 0.07 0.04 0.01 0.12 55 16 17
1098 Tyrone_MountDarwin 0.28 0.15 0.11 0.33 0.10 0.30 0.02 0.06 0.09 0.07 0.05 0.02 0.10 78 28 21
- Average Basaltic Soila 0.20 0.10 0.11 0.40 0.07 0.08 0.01 0.06 0.04 0.06 0.03 0.01 0.10 39 12 15
aValues for average basaltic soil are for a typical individual measurement.
E06S10 YEN ET AL.: HYDROTHERMAL PROCESSES AT GUSEV CRATER
9 of 19
E06S10
hematite. After removing the basaltic component and the
hematite from the Samra analysis, the phases in which the
remaining elements reside are not fully constrained. Candi-
date assemblages can be derived, but the results are not
unique. The primary challenges include uncertainties in the
total oxygen content, the absence of constraints on the
abundance of (OH) and H2O groups, and not knowing if
other undetected elements are abundant in the sample.
Nonetheless, potential compositions can be reconstructed
and examined for reasonableness.
[30] Calcium and phosphorous are minor components
of the Samra deposit, and if they were combined into a
Ca-phosphate, this phase would represent 1 wt % of the
measured elements. The calculated Ca:P ratio is 1.3:1, but
the uncertainty is sufficiently large that it could be as
high as 1.7:1 as in the apatite family or as low as 1:1 as
in brushite. At these low concentrations, Ca and P could
also be substitutional impurities in other constituents
without developing into a distinct mineral phase.
[31] Mg-sulfates have been detected in many locations at
the Martian surface by the Mars Express OMEGA instru-
Figure 8. Elemental ratios of Paso Robles class soils to
average Gusev basaltic soil (Table 4) illustrating significant
differences from typical soils.
Figure 9. Elemental ratio plot of Paso Robles class soils to
the first Paso Robles measurement indicating significant
variability between the members of this soil class.
Figure 10. Correspondence analysis diagram [e.g., Larsen
et al., 2000] of samples analyzed by the Spirit APXS. The
data points (colors) are grouped by geographic locations,
and the brown lines represent the measured elements. Paso
Robles class soils clearly plot within the field of nearby
rocks.
Figure 11. A molar Ca versus P plot indicating that
signatures of nearby rocks are contained in the Paso Robles
class soil measurements. The points representing the Paso
Robles, Arad, and Tyrone deposits are widely separated
indicating distinct compositions. However, soils at the Paso
Robles site plot near Wishstone and Watchtower class
rocks; Arad soils plot adjacent to Algonquin class rocks;
and Tyrone soils are similar to Halley. Each of these rock
types are found in reasonably close proximity to the
associated Paso Robles class soil exposure indicating that
for certain elements, these soils are more similar to nearby
rocks than they are to each other.
E06S10 YEN ET AL.: HYDROTHERMAL PROCESSES AT GUSEV CRATER
10 of 19
E06S10
ment [Bibring et al., 2005]. They are inferred to be present
at Meridiani Planum on the basis of APXS data [Clark et
al., 2005], which is supported by Mini-TES [Glotch et al.,
2006] and orbital measurements [Griffes et al., 2007]. It is
therefore reasonable to consider the presence of an MgSO4 
nH2O phase in the Samra analysis. Assuming that all of the
Mg is consumed in this phase with a Mg:S mole ratio of 1:1,
11 wt % of the sample would be present as a Mg-sulfate.
[32] The remaining Fe and S (46 wt %) can be combined at
a Fe:S ratio of 2:2.8, which with uncertainties, fits well within
the family of ferric sulfates of the form (Fe3+)2(SO4)3  nH2O
(see section 5.2 for additional discussion). After accounting
for the phases listed above, the following remains: 34 wt %
silica, 1 wt % Al (possibly substituting for the Fe3+ in the
ferric sulfate), and 2 wt % Cl, Ti, and Cr. The possibility of
excess silica in this analysis is inescapable, as there are few
remaining elements for other compositions. In this recon-
struction of candidate mineral phases, 90 wt % of Samra
consists of silica, ferric sulfate, and Mg-sulfate. These
compositions are summarized in Table 5. It is worth noting
that such calculations assume end-member Fe-, Ca-, and
Mg-sulfate mineralogy. It is possible that there could be
some degree of solid solution among the various sulfate
mineralogies [Tosca and McLennan, 2007].
4.2. Arad Hula
[33] The Hula sample at the Arad location appears similar
in composition to Samra, but with nearly 3 times the Mg
and over 6 times the Ca. Unfortunately, Mo?ssbauer data
were not collected from this sample because of the pressing
need to move the rover to a safe location for the impending
Martian winter, so direct information on the iron mineralogy
is not available. The amount of basaltic soil contamination
in this sample can be constrained to less than 15 wt % using
Na and K as the limiting elements. This value is consistent
with the 12% inferred from Figure 13, which is based on Na
and Al in a plagioclase component.
[34] In order of abundance, the major remaining elements
after the maximum amount of basaltic soil is subtracted are
S, Si, Mg, Fe, and Ca. All other elements are present at
approximately 3% combined. An attempt at charge balance
using (SO4)
2, Mg2+, Fe3+, Ca2+, and minor amounts of
Al3+, Cl, and (PO4)
3 indicates a significant excess (10 to
15 mole%) of cations. Si is assumed to be present as silica
and not included in this attempted balance. This result
suggests that oxides, hydroxides, oxyhydroxides, fluorides,
borates and/or other silicates not present in the Samra
sample are found in the Hula measurement just a few tens
of centimeters away. Carbonates below approximately 10 wt
% cannot be excluded on the basis of the alpha channel data
[Rieder et al., 2003] from the APXS, though it is unlikely
that carbonates would form in combination with ferric
sulfates. After removal of the basaltic soil and 21% silica,
up to 57% of the remaining measured elements can be
present as sulfates, with a maximum of 38%, 30%, and 9%
of ferric sulfate (Fe:S = 2:3), Mg-sulfate (Mg:S = 1:1), and
Ca-sulfate (Ca:S = 1:1), respectively.
[35] An alternative method for considering the composi-
tion of Hula is to calculate the limiting quantity of the
Samra composition that can be accommodated by Hula.
This maximum value is 56% with Si as the limiting element.
In either approach to reconstructing the composition of
Hula, it is clear that this sample is significantly different
from the Samra deposit, that the difference does not result
simply from admixture of basaltic soil, and that additional
anions not directly measured by the APXS are necessary to
maintain charge balance. The possible composition of Hula
is summarized in Table 5.
4.3. Paso Robles I
[36] On the basis of the limited Cr abundance in APXS
data, there can be a maximum of 15% basaltic soil in the
first Paso Robles sample (Pasadena Paso Robles) that was
analyzed in detail. The Mo?ssbauer data, on the other hand,
indicate that 3% of the iron is in olivine and 10% of the iron
is in pyroxene, which corresponds to 28% basaltic soil when
Figure 12. The trace element yttrium (peak at 14.9 keV) is
found in the light-toned soil exposures at the Paso Robles
location and in surrounding rocks. This element is not
present in the Arad or Tyrone samples.
Figure 13. Molar plot of Na versus Al for rocks that are
presumed to be plagioclase-rich and basaltic soils at the
Gusev landing site. The fit to the plotted data approximately
represents an An35 composition (andesine).
E06S10 YEN ET AL.: HYDROTHERMAL PROCESSES AT GUSEV CRATER
11 of 19
E06S10
the contributions of the other elements are included. Calcu-
lations based on the possible inclusion of a plagioclase
component (Figure 13) suggest up to 35% basaltic soil.
Which of these values is likely to be the most representa-
tive? The estimate based on Na and Al is perhaps suspect in
the Paso Robles samples given that the surrounding Wish-
stone and Watchtower class rocks have significant enhance-
ments in these elements and because it is clear that these
light-toned soil deposits reflect, to a significant degree, the
chemistry of the local materials (section 3.2.2). That is, the
Na and Al enhancements could, in this case, be unique
attributes of the nonbasaltic portion of the Paso Robles
sample, rather than a direct indicator of a basaltic soil
contribution. This 35% value, however, is still valid as an
upper limit for the basaltic contamination. The larger
sampling depth and/or smaller field of view of the Mo?ss-
bauer spectrometer relative to the APXS may be responsible
for the larger amount of basaltic soil inferred from the
Mo?ssbauer measurement. There is also a possibility that an
unidentified ferrous phase results in an overestimate of the
amount of olivine and pyroxene, ultimately producing a
larger estimate for the amount of basaltic soil. Accounting
for possible uncertainties in the Cr measurement and the
standard deviation in Cr values in over 35 Gusev soil
measurements, both of which expand the limits of possible
mixing, a value of 20% basaltic soil contamination was
adopted for this reconstruction, but this value likely has an
uncertainty of at least 5% absolute. Note that the low Cr
signature of the nearby rocks [Gellert et al., 2006] is not
reflected in local basaltic soils, so establishing this con-
straint on the basis of Cr remains valid.
[37] After removal of 20% basaltic soil, 12% silica, and
5% hematite based on Mo?ssbauer results, the primary
remaining elements are S, Fe, Ca, Mg, and P in roughly a
15:8:4:4:3 molar ratio. Assuming these elements are present
as (SO4)
2, Fe3+, Ca2+, Mg2+, and (PO4)
3, overall charge
balance is maintained to within a few percent. One possible
composition of the sulfates and phosphates consistent with
the APXS results is 37% (Fe3+)2(SO4)3  nH2O, 11%
MgSO4  nH2O, and 11% Ca5(PO4)3(OH). See section 5.2
for additional discussion of the ferric sulfate phases. Evi-
dence for significant amounts of Y in these samples is
consistent with the presence of a distinct phosphate mineral.
In shergottites, for example, the rare earth elements are
dominantly contained in phosphates such as whitlockite and
apatite [McSween and Treiman, 1998]. Other phases present
in the first detailed Paso Robles analysis may include 
2%
NaCl and 
3% Al as allophane [Ming et al., 2006], alunite,
or a substitutional impurity. This candidate composition for
the first Paso Robles sample is summarized in Table 5.
Other combinations of these elements are possible, but in all
cases, ferric sulfates are still indicated. For example, the
presence of ferric iron phosphates [Lane et al., 2007] of up
to 11% (e.g., (Fe3+)(PO4)  nH2O or (Fe3+)3(PO4)2(OH)3 
nH2O), cannot be excluded on the basis of elemental
chemistry. If iron phosphates were present, a contribution
from Ca-sulfates may be inferred. However, modeling of
multispectral Pancam data of Paso Robles class soils did not
detect any phosphates [Johnson et al., 2007]. Mini-TES
data of Paso Robles can be reasonably fit without ferric
phosphates and the deconvolutions do not improve signif-
icantly when they are added (A. T. Knudson et al., Mini-
TES derived mineralogy of soils at the Mars Exploration
Rover ??Spirit?? landing site, manuscript in preparation,
2008). In addition, phosphates are a negligible component
in two of the three major occurrences of this soil class
(Table 4a) and are therefore not a defining characteristic.
4.4. Paso Robles II
[38] In the second set of analyses of the light-toned soils
at the Paso Robles location (Paso Robles2 PasoLight1), the
Mo?ssbauer spectrometer places the tightest constraint on the
amount of mixing from basaltic soil. Constraints based on
Na with Al (Figure 13) and on Na alone both establish an
upper limit of approximately 50% basaltic soil. With an
estimated 10%, 8%, and 6% of the Fe reported by the
Mo?ssbauer analysis as olivine, pyroxene, and magnetite,
respectively (Table 3), and perhaps a few percent of
unreported nanophase ferric oxides obscured by the ferric
sulfate doublet, up to 25% of the Fe could be present as a
basaltic contaminant. Using analyses of average Gusev
soils, this Mo?ssbauer measurement corresponds to 30% of
the observed Paso Robles II elemental composition. In
previous examples (Samra and Paso Robles I), the Mo?ss-
bauer results overestimate the quantity of basaltic material
for reconstructions of the elemental chemistry. Therefore, a
value of 25% basaltic soil is adopted for this analysis.
[39] After removal of 13% silica and 1% hematite, the
remaining composition is roughly similar to the Paso Robles
I analysis. The primary differences include 
20% less Mg
and P, 
30% less Fe, and nearly double the amount of Al
resulting in an Al abundance in Paso Robles II that is
comparable to the quantity of Mg. While it is not necessary
that all the Si be represented as silica, there is nearly as
much Si as Al, Mg, and Ca combined, and the majority of
these other cations are likely involved in sulfates and
phosphates. Therefore, there is a significant likelihood that
excess silica is present in this sample as well. Attempts at
charge balance based on (SO4)
2, Fe3+, Ca2+, Mg2+, and
(PO4)
3 result in a 13% anion excess, but this can be
remedied with the addition of Al3+. One possible composi-
Table 5. Reconstructed Compositions of the Paso Robles Class Soilsa
Sample Basaltic Soil Silica Mg-Sulfate Ca-Sulfate Fe-Sulfate Ca-P Hematite Otherb
Arad_Samra 5 34 11 0 46 1 1 2
Arad_Hula 15 21 30 9 38 1 unknown 2
Pasadena_PasoRobles 20 12 11 0 37 11 5 4
PasoRobles2_PasoLight1 25 13 9 0 36 10 1 6
Tyrone_BerknerIsland1 65 6 4 8 14 1 1 1
Tyrone_MountDarwin 40 14 9 15 18 1 1 2
aReconstructed compositions are in wt %.
bRepresents percentages of measured elements that do not fit within the other columns of the table.
E06S10 YEN ET AL.: HYDROTHERMAL PROCESSES AT GUSEV CRATER
12 of 19
E06S10
tion of the sulfates and phosphates consistent with the
APXS results is 36% (Al3+, Fe3+)2(SO4)3  nH2O, 9%
MgSO4  nH2O, and 10% Ca5(PO4)3(OH). See section 5.2
for additional discussion of the ferric sulfate phases. Halite,
at an abundance of up to 2%, may also be present. This
candidate composition is summarized in Table 5.
4.5. Mount Darwin
[40] Light-toned soil dragged from Tyrone by the rover
wheels created a thin trail that was mixed with and resting
on top of basaltic soil (Figure 2). This material was
accessible to the rover for analysis, and likely consists of
a mix of the yellowish and the whiter components of the
excavated soils (Figure 5m and Johnson et al. [2007]). The
Mo?ssbauer fits using an assumed abundance of npOx
consistent with typical soil measurements suggest that
approximately 60% of the iron is in components of basaltic
soil. It should, however, be noted that the quality of the
Mo?ssbauer data from this target is extremely poor given a
placement anomaly which resulted in a larger than desired
standoff distance between the sensor head and the sample.
When the Mo?ssbauer results are scaled to account for all
elements that are in average Gusev soils, the estimate for
basaltic soil contamination is approximately 65%. However,
given that the compositional analyses are based primarily on
APXS data and because of the relatively poor quality of the
Mo?ssbauer data, we use instead the 40% upper limit for
basaltic soil contribution established by APXS measure-
ments of Na.
[41] A major difference between the Mount Darwin
measurement and the analyses at Arad and Paso Robles is
a significant abundance of Ca. Given that essentially all the
P in this sample is from the basaltic soil contamination, the
Ca is likely in the form of sulfates. Assuming Mg-sulfates
(Mg:S = 1:1) and silica are in this sample, the major
remaining component is the ferric sulfate with an apparent
Fe:S of 1:1.1, which is well within the range expected for a
plausible suite of phases (section 5.2). The candidate
composition of Mount Darwin is summarized in Table 5.
4.6. Berkner Island
[42] The Tyrone Berkner Island1 sample is more distant
from the original Tyrone deposit than Mount Darwin, and it
is reasonable to expect it to be more diluted by basaltic soil.
Five elements in the APXS analysis (Na, Al, Cl, K, and Ti)
establish the allowable contribution of basaltic soil at
approximately 65%, which is tighter than the constraints
obtained from the Mo?ssbauer results. After removing the
basaltic contamination, candidate abundances of silica and
Mg-, Ca-, and Fe-sulfates can be calculated as described in
previous examples. The results are summarized in Table 5,
and possible Fe-sulfate phases are discussed in section 5.2.
4.7. Mineralogical Summary
[43] The basaltic soil components estimated in the pre-
ceding paragraphs represent contaminants mixed in by the
rover wheels and are not original constituents of the light-
toned soil. It is therefore reasonable to remove the basaltic
soil contribution and renormalize the remaining phases to
determine the true composition of the Paso Robles class
materials. The result (Figure 14) indicates that these deposits
are dominated by silica, ferric sulfate, and Mg-sulfate. The
measurements at the original Paso Robles location also
contains phosphates, represented here as a Ca phase. The
deposits excavated at Tyrone also contain significant Ca,
likely present as a sulfate.
[44] In these renormalizations after basaltic soil removal,
it is apparent that the Mount Darwin and Berkner Island
samples are very similar in bulk chemistry, and hence in
Figure 14. Derived compositions of the Paso Robles class samples after removal of basaltic soil
contamination (based on Table 5). ??Other?? represents the remaining constituents after depicted phases
and basaltic soil are removed. The available constraints on the Arad Hula sample prevent an accurate
subdivision of the ??total sulfate?? wedge.
E06S10 YEN ET AL.: HYDROTHERMAL PROCESSES AT GUSEV CRATER
13 of 19
E06S10
inferred bulk mineralogy as well. This is as expected given
that these analyses are from samples of the same material
falling out of the rover wheels at different points along the
traverse. The two Paso Robles samples are also similar to
each other in bulk chemistry and inferred mineralogy but
different from the Tyrone materials. The Samra deposit is
unique because it does not contain a significant Ca phase.
The Hula mineralogy remains the most poorly constrained
of the six APXS analyses, because Mo?ssbauer data were not
acquired from this target and elements analyzed by APXS to
not provide a satisfactory charge balance.
5. Acidic Environment
[45] Ferric sulfates form under oxidizing, low pH con-
ditions [Bigham and Nordstrom, 2000], and it is therefore
clear that these soils formed in a highly acidic environment.
Furthermore, as discussed below, the possibility of elemen-
tal sulfur and/or unreacted sulfuric acid in Paso Robles class
deposits cannot be excluded.
5.1. Excess Sulfur?
[46] A Fe:S ratio of approximately 2:3 is obtained directly
from the reconstruction of the Arad Samra sample once the
basaltic soil, measured hematite, silica, and Mg-sulfate
(Mg:S = 1:1) are removed from elemental chemistry
(section 4.1). This is a reasonably satisfying result given that
(Fe3+)2(SO4)3  nH2O is a common family of ferric sulfates
(Table 6). This Fe:S ratio is also consistent with the
chemical composition obtained from the other analyses of
Paso Robles class soils. However, reconstructions of possi-
ble mineral phases based on elemental chemistry are not
unique, even with good constraints on the iron oxidation
states and certain identified iron-bearing minerals.
[47] Ferric sulfates in other forms cannot be excluded on
the basis of the available data. Hydrogen in the deposit
could be present as H2O, (OH)
, or both, and a dominant
ferric sulfate in the Paso Robles samples of the form
(Fe3+)(SO4)(OH)  nH2O is allowed by the APXS results.
In this example, the Fe:S ratio is 1:1, implying approxi-
mately 6.5 mole% excess sulfur in the Samra analysis.
Ferric sulfates with more Fe than S would result in even
higher amounts of excess sulfur, which could be in the form
of native S or possibly unreacted H2SO4. The available data
do not require that excess sulfur be present, but they cannot
exclude this possibility. In addition, it is also possible that
native sulfur was a component of the original deposit and
was subsequently oxidized to sulfate phases.
5.2. Ferric Sulfate Phase(s)
[48] The specific mineral form(s) of the iron sulfate(s)
present in Paso Robles class soils can be constrained using
the available instrument data. Possible phases are listed in
Table 6. Only ferric phases are listed because minerals
dominated by ferrous iron are not consistent with the
Mo?ssbauer results and cannot be significant contributors
to the overall composition. Models using Pancam visible/
near-infrared data indicate that rhomboclase, fibroferrite,
hydronium jarosite, and/or ferricopiapite could be signifi-
cant contributors to Paso Robles class soils [Johnson et al.,
2007]. The likelihood of the various candidates in Table 6 as
dominant ferric sulfates in Paso Robles class soils can also
be evaluated on the basis of iron-sulfur ratios. Phases with
Fe:S of 1:1.5 fit well with APXS measurements of all Paso
Robles class deposits. Lower ratios (e.g., Fe:S of 1:1) would
imply excess sulfur (section 5.1) or a large contribution
from unmeasured cations such as H+, H3O
+, or NH4
+, though
NH4
+ is unlikely given the oxidizing conditions implied by
the presence of ferric phases. Higher ratios (e.g., Fe:S of
1:2) can be present in low concentrations, but cannot be
supported as the primary phase in the Samra analysis. Even
if all of the Mg were in nonsulfates, the maximum Fe:S
would be 1:1.8 in Samra. The percentage of Fe in the sulfate
is well constrained by the Mo?ssbauer spectrometer, and the
measured abundance of S is insufficient to achieve a ratio of
1:2. This result excludes phases such as rhomboclase as the
only ferric sulfate in Samra, though mixtures with other
phases to reduce the average Fe:S ratio are allowed.
Table 6. Candidate Ferric Sulfates in Paso Robles Class Soilsa
Mineral Phase Formula Fe:S 1b 2c 3d 4e
Mikasaite (Fe3+)2(SO4)3 2:3 X
Lausenite (Fe3+)2 (SO4)3 . 6H2O 2:3
Kornelite (Fe3+)2 (SO4)3 . 7?8H2O 2:3
Coquimbite (Fe3+)2 (SO4)3 . 9H2O 2:3 X X
Paracoquimbite (Fe3+)2 (SO4)3 . 9H2O 2:3 X X
Quenstedtite (Fe3+)2 (SO4)3 . 10?11H2O 2:3
Butlerite Fe3+(SO4)(OH) . 2H2O 1:1 X
Parabutlerite Fe3+(SO4)(OH) . 2H2O 1:1 X
Fibroferrite Fe3+(SO4)(OH) . 5H2O 1:1 X X
Amarantite (Fe3+)2 O(SO4)2 . 7H2O 1:1 X
Hohmannite (Fe3+)2 O(SO4)2 . 8H2O 1:1 X
Metahohmannite (Fe3+)2 (SO4)2(OH)2 . 3H2O 1:1 X
Ferricopiapite (Fe3+)2/3(Fe
3+)4 (SO4)6(OH)2 . 20H2O 7:9 X
Magnesiocopiapite Mg(Fe3+)4 (SO4)6(OH)2 . 20H2O 2:3
Botryogen MgFe3+(SO4)2(OH) . 7H2O 1:2 X
Hydronium jarosite (H3O)(Fe
3+)3 (SO4)2(OH)6 3:2 X X
Rhomboclase H5Fe
3+O2(SO4)2 . 2H2O 1:2 X X
Schwertmannite (Fe3+)16 O16(SO4)2(OH)12 . nH2O 8:1 X
aFerrous phases are not listed as Mo?ssbauer data constrain Fe2+ phases to less than 1% of the total.
bNot consistent with Mo?ssbauer parameters.
cFe:S of 1:1 implies 
5% excess sulfur at Arad Samra.
dFe:S of 1:2 not compatible as the only ferric sulfate at Arad Samra.
eIdentified as a possible constituent based on visible/near-infrared data [Johnson et al., 2007].
E06S10 YEN ET AL.: HYDROTHERMAL PROCESSES AT GUSEV CRATER
14 of 19
E06S10
Multiple phases present as solid solutions may also be
possible.
[49] The Mo?ssbauer parameters of the ferric sulfate listed
in Table 2 can be used to specifically exclude certain
phases. Mikasaite [Nomura et al., 2005], coquimbite and
paracoquimbite (R. V. Morris, unpublished laboratory data,
2007), for example, exhibit quadrupole splittings (DEQ)
which are too narrow to be consistent with the Paso Robles
class soils. Jarosite [Klingelho?fer et al., 2004; Morris et al.,
2007], the nanophase ferric oxide (npOx) [Morris et al.,
2004, 2006] identified in other Martian samples, and
schwertmannite, which is similar to np-Ox in Mo?ssbauer
analyses, all have distinctly different parameters from the
Paso Robles class measurements. Identifying or excluding
other phases listed in Table 6 on the basis of laboratory
measurements are complicated by the metastability of many
of these hydrated phases and the corresponding challenges
associated with verifying the true mineralogical composi-
tion of the material analyzed.
[50] Establishing the mineral phases present in the Paso
Robles class soil deposits is clearly important for determin-
ing the nature of the formation environment. However,
many of the phases listed in Table 6 which are not
specifically excluded by the Mo?ssbauer data form in close
association with each other in a well-defined environment.
For example, on Earth, lausenite, kornelite, quenstedtite,
butlerite, parabutlerite, fibroferrite, amanterite, hohmannite,
metahohmannite, ferricopiapite, magnesiocopiapite, botry-
ogen, and rhomboclase generally form together as precip-
itates of oxidative weathering of iron sulfide deposits
[Jambor et al., 2000; Bigham and Nordstrom, 2000]. Thus,
identifying the presence of ferric sulfates in general, may
provide nearly as much information about the formation
environment of Paso Robles class soils as establishing the
relative proportions of the specific phases themselves.
6. High-Temperature Environment
[51] The Home Plate feature within the Inner Basin is
likely an accumulation of pyroclastic materials resulting
from explosive interactions between basaltic magma and
water [Squyres et al., 2007]. All of the confirmed exposures
of Paso Robles class soils are within 1 km of Home Plate,
and Tyrone is only 50 m from the margin of the feature. Few
constraints for establishing the relative ages of Home Plate
and Paso Robles class soils are available, but it is clear that
both water and high temperatures were available in the
vicinity of Paso Robles class soil exposures at some point in
Martian history.
[52] Geochemical evidence is also consistent with forma-
tion of Paso Robles class soils at elevated temperatures. The
reconstructions of elemental chemistry described in section
4 indicate the presence of silica in Paso Robles class soils.
After removal of the basaltic soil component, the concen-
trations of Na, Mg, Al, K, and Ca are relatively low. In the
Samra analysis, for example, the molar abundance of Si is
five times greater than the amount of these other five
elements combined. With at least some of the Mg and Ca
in sulfates and/or phosphates, there are not enough cations
for typical silicate mineral phases. Thus, the occurrence of
excess silica in these soils is indicated. There are at least two
possible mechanisms for developing these silica deposits:
(1) precipitation from solutions rich in dissolved silica and
(2) acid bleaching of preexisting silicate phases. Both of
these processes are suggestive of interactions with fluids
and/or gases at elevated temperatures.
[53] The solubility of silica is relatively independent of
pH under acid to neutral conditions [Alexander et al., 1954;
Krauskopf, 1956] but strongly dependent on the temperature
of the fluid [White et al., 1956]. This relationship is
commonly used as a geothermometer in hydrothermal
systems [Fournier and Potter, 1982]. Water in aquifers
heated geothermally interacts with subsurface rocks liberat-
ing silica and other species in both acid sulfate and neutral
chloride environments. As silica-saturated solutions rise to
the surface and cool, precipitates of excess silica are formed.
This process occurs at Yellowstone National Park where the
interactions between thermal waters and volcanic rocks main-
tain up to 750 ppm silica in solution prior to cooling,
evaporation, and precipitation of amorphous silica [Channing
and Butler, 2007]. Elevated temperatures are consistent with
the development of silica in Paso Robles class soils, but are
not required. In the absence of hydrothermal and biological
activity, 120 ppm silica could still be present in aqueous
solution. Low-temperature, evaporitic sediments at Meri-
diani [McLennan et al., 2005], for example, contain approx-
imately 25% silica [Glotch and Bandfield, 2006].
[54] An alternative mechanism for concentrating silica is
through acid bleaching, which, in contrast, likely requires
elevated temperatures. In this process, dissolution of rocks
and soils by acidic fluids and vapors (e.g., sulfuric and/or
hydrochloric) is followed by immediate reprecipitation of
amorphous SiO2  nH2O and typically anatase (TiO2
polymorph) [Morris et al., 2000]. Other elements are
leached resulting in high concentrations of silica with
elevated Ti. A likely example of this process is found in
the Kenosha Comets sample at the eastern margin of Home
Plate. This light-toned soil contains over 90 wt % silica and
50% more Ti than average basaltic soil. A number of high-
silica, light-toned soils distinct from the high-sulfur Paso
Robles class materials have been analyzed near Home Plate.
The Si to Ti ratios of the high-silica and the Paso Robles
class materials are comparable, suggesting that these soils
are related. The proximity of the Kenosha Comets sample to
Tyrone (
50 m away) also supports a coupled relationship.
Formation at elevated temperatures is likely for both classes
of light-toned soils.
7. Formation Mechanism
[55] What processes resulted in the deposition of Paso
Robles class soils? How did the material form initially, and
what, if any, mechanisms may have redistributed or other-
wise altered the deposits?
[56] The attributes of these deposits discussed in sections
2?6 will help address these questions. Paso Robles class
soils (1) contain chemical signatures of local rocks, (2) dis-
play significant compositional heterogeneity over distances
of tens of centimeters, (3) exhibit major differences in
composition across the analyzed exposures, (4) contain
mineral phases indicative of formation in an acidic envi-
ronment, (5) are hydrated, (6) contain indicators consistent
with high-temperature processes, (7) are found in a variety
of geologic settings relative to topography and elevation,
E06S10 YEN ET AL.: HYDROTHERMAL PROCESSES AT GUSEV CRATER
15 of 19
E06S10
and (8) have not been extensively redistributed and likely
formed in the places where they were discovered.
[57] It is clear from this evidence that water played a key
role in the formation of these deposits. A number of
possible scenarios involving aqueous processes are pre-
sented in the following paragraphs.
7.1. Scenario 1: Lacustrine Sediments
[58] Paso Robles class soils precipitated from low-tem-
perature aqueous solutions in local lacustrine settings. The
occurrence of the Tyrone deposit in a localized topographic
depression (section 2) led to the suggestion that it might
have formed as layered aqueous precipitates from standing
bodies of water [Wang et al., 2007]. Other light-toned soils
rich in ferric sulfates, however, are found on slopes of
approximately 12
 (section 2), indicating that topographic
control in the formation is not important. Assuming a
consistent process for the development of all Paso Robles
class materials, the observed distribution argues strongly
against the idea that these soils are precipitates from
standing water. In addition, highly variable compositions
over lateral distances of less than 1 m contradict the
possibility of a quasi-equilibrium aqueous precipitate result-
ing from the evaporation of water. We therefore conclude
that the origin of Paso Robles class soils as a lacustrine
sediment is unlikely.
7.2. Scenario 2: Seeps and Efflorescent Salts
[59] Water interacting with subsurface iron sulfides mi-
grated toward the surface (e.g., by capillary action) and
rapidly evaporated resulting in a nonequilibrium accumula-
tion of mineral precipitates. Water interacting with iron
sulfides in the terrestrial environment can produce low pH
solutions that precipitate minerals such as gypsum, copia-
pite, alunogen, metahohmannite, coquimbite, and paraco-
quimbite [Joeckel et al., 2005]. Rhomboclase, ferricopiapite,
bilinite, gypsum, hexahydrite, and Al-sulfates are com-
monly produced in acid mine drainages [Keith et al., 2001;
Nordstrom and Alpers, 1999]. Many of these phases are
consistent with the possible mineralogies of the Paso Robles
class soils and are stable in an arid environment. Thus, the
observed mineral assemblages in Paso Robles class soils may
be reasonable for an evaporitic deposit resulting from fluid
interactions with local rocks. The substantial heterogeneities
over short length scales imply low water-to-rock ratios, and
the occurrence on slopes implies rapid evaporation upon
exposure to the atmosphere.
[60] Several observations, however, present challenges to
this formation scenario. The Paso Robles exposure of light-
toned soil occurs approximately 66 m above the Gusev
plains and 40 m below the summit of Husband Hill. Low-
temperature seeps and fluid drainage would not be expected
at this elevation; thermally driven processes may be more
likely. In addition, efflorescent salts are typically found as
cements in preexisting sediments [Joeckel et al., 2005]
rather than as widespread deposits of high purity. Further-
more, evidence that Home Plate is a hydrovolcanic deposit
and the formation of, likely related, silica-rich deposits by
acid sulfate bleaching processes indicate that thermal pro-
cesses have been important in at least the Inner Basin. With
the available observations, the development of Paso Robles
class soil deposits through low-temperature seeps and
evaporative processes cannot be ruled out, but a hydrother-
mal scenario seems more plausible.
7.3. Scenario 3: Phreatic Eruption
[61] Volcanic eruptions into near-surface water produced
steam explosions which distributed ash layers to the ob-
served locations of the Paso Robles class soils. The Home
Plate feature has been interpreted as a construct of explosive
volcanism [Squyres et al., 2007], so it is reasonable to
consider the possibility that Paso Robles class soils formed
hydrothermally during magma interactions with water. Al-
though mineral phases such as cristobalite, tridymite, and
gypsum can form in such events, magmatic mineral phases
generally dominate the deposits [Ohba and Nakagawa,
2002]. In addition, the observations of significant compo-
sitional diversity over both large and small length scales are
not consistent with an air fall deposit. Settling from the
atmosphere would tend to produce deposits that were more
uniform in composition over a large area. Instead, the
signatures of local rocks in these soils are strong indicators
that the formation process operates on local rather than
regional scales. It is also not clear how concentrations of
ferric sulfates and silica could form in abundance under
these conditions. A phreatic eruption scenario is therefore
unlikely to be responsible for the formation of Paso Robles
class soils.
7.4. Scenario 4: Fumarolic and Hydrothermal Deposits
[62] Paso Robles class soils formed as condensates of gas
emissions and precipitates from hydrothermal fluids asso-
ciated with nearby volcanic processes. The Home Plate
structure is likely a remnant of an explosive volcanic
deposit [Squyres et al., 2007], indicating that heated volca-
nic fluids and gases were once common in this vicinity.
Significant involvement of volatiles in the Home Plate
region is also evidenced by the abundance of alkali-rich
vesicular basalts [Crumpler et al., 2007]. Abundant volatiles
in a volcanic setting are clearly consistent with the likeli-
hood of fumarolic emissions and hydrothermal precipitates.
[63] Of the likely mineral phases identified in Paso Robles
class soils, silica, gypsum, anhydrite, and hematite are
commonly found in fumarolic deposits on Earth [Africano
and Bernard, 2000; Keith, 1991]. Ferric sulfates in terrestrial
occurrences are typically associated with acid mine drain-
ages, but there is clear evidence for copiapite, quenstedtite,
fibroferrite, coquimbite, and other Fe hydroxysulfates as
fumarolic condensates [Zimbelman et al., 2005; Stoiber and
Rose, 1974]. The occurrence of ferric sulfates is supported
by thermochemical modeling of products formed by rapid
cooling of volcanic gases which predicts the formation of
Fe2(SO4)3  H2O [Getahun et al., 1996]. In addition,
kieserite, hexahydrite, and epsomite have been reported
[Stoiber and Rose, 1974]. The relative simplicity of the
likely mineral assemblages in the Paso Robles class deposits
is not unprecedented. Examples of fumarolic sublimates
consisting of sulfur, gypsum, halotrichite, and little else
have been reported [Stoiber and Rose, 1974]. Furthermore,
it is possible that an initially complex assortment of meta-
stable phases equilibrated with the ambient environment
over geological timescales producing the observed compo-
sitions. Thus, the likely mineral phases in Paso Robles class
deposits are consistent with fumarolic deposits.
E06S10 YEN ET AL.: HYDROTHERMAL PROCESSES AT GUSEV CRATER
16 of 19
E06S10
[64] Zinc and potassium are elements characteristically
enhanced in high-temperature regions of terrestrial volcanic
sublimates [Symonds et al., 1987] which are also detectable
by the Mars Exploration Rover (MER) APXS. The highest
concentrations of Zn and K thus far detected on Mars are
within a few tens of meters of the Tyrone exposure of Paso
Robles class soils. The sample with the highest Zn levels
(nearly 2300 ppm) contains hematite and exhibits compelling
evidence for Ca-sulfate [Yen, 2006], both of which are
common fumarolic minerals. Neither Zn nor K concentra-
tions are significantly enhanced in the Paso Robles class soils
themselves, which may indicate that they are more distal
deposits or formed from fluid runoff of vapor condensates.
[65] The chemical signatures of the local rocks in analy-
ses of the Paso Robles class soils is indicative of fluids and
vapor migrating through subsurface materials prior to form-
ing precipitates at the surface. Heterogeneities in the com-
position over short length scales are characteristic of
fumarolic deposits, which are nonequilibrium condensates.
The occurrence of hematite and, particularly, silica are
consistent with formation in a high-temperature environ-
ment. The apparent fractionation of S-rich vapors (produc-
ing the Paso Robles class soils) and Cl-rich brines (indicated
in analyses of rocks at Home Plate [Squyres et al., 2007])
may be related to boiling of a hydrothermal brine at depth
(M. Schmidt et al., The hydrothermal origin of excess
halogens at Home Plate, Gusev Crater, submitted to Journal
of Geophysical Research, 2007).
[66] Bleaching by acidic volatiles (e.g., fumarolic gases
and vapor condensates) is indicated by several high-silica
analyses near Home Plate (section 6 and Squyres et al.
(submitted manuscript, 2008)). In addition, a rock sample
(??Ben?s Clod??) found at the Paso Robles location may
represent a precursor to the smectite montmorillonite or its
chemical equivalent [Clark et al., 2007]. Figure 15 shows
that the composition of Ben?s Clod may be consistent with
an altered Wishstone or Watchtower class rock.
[67] Taken collectively, these observations indicate that
the origin of the Paso Robles class soils is most consistent
with condensation from volcanic vapors and precipitation
from high-temperature fluids mobilized by volcanic activity
in an acid sulfate system. In the terrestrial fumarolic
environment, deposits can form as sublimates directly from
the gas phase [Symonds et al., 1987]. Liquid condensates,
consisting of water with abundant anions and cations can
collect and form solid precipitates as the water cools and
evaporates [Stoiber and Rose, 1970]. A combination of
these processes associated with fumarolic emissions are
likely responsible for the formation of the Paso Robles
class deposits.
[68] This interpretation implies a sulfur-rich magmatic
source, which has been previously hypothesized [Clark
and Baird, 1979], or the presence of nearby iron sulfide
deposits. The possible detection of pyrite and/or marcasite
in one rock sample on Home Plate [Morris et al., 2007]
supports the idea of crustal sulfides. However, this identi-
fication is not certain [Squyres et al., 2007], and the
continued search for sulfides with the Mo?ssbauer spectrom-
eter is warranted.
8. Biological Implications and Timescales
[69] A sulfurous hydrothermal environment, as indicated
by the Paso Robles class soils, can provide the necessary
access to water and an energy source to support potential
Martian microorganisms. Chemoautotrophic bacteria can
utilize H2, H2S, and elemental sulfur found in hydrothermal
environments in a redox reaction to produce sufficient
energy to support metabolic functions [McKay, 1997]. This
environment was likely highly acidic, as evidenced by the
occurrence of ferric sulfates. Acidophilic bacteria on Earth
are known to thrive at pH levels of less than 1 [Nealson,
1999], although these acidophiles apparently evolved from
organisms that initially arose under more neutral pH con-
ditions. Highly acidic conditions may present a significant
obstacle to the origin of life [Knoll et al., 2005], and future
astrobiological investigations of Mars should be focused on
regions with mineral deposits formed from more neutral
solutions.
[70] Few observational constraints are available to esti-
mate the duration of hydrothermal activity in the Columbia
Hills where the Paso Robles class soils are found. A sustained
presence would clearly be more conducive to biological
processes than a brief one. The age of the deposits in relation
to other geological units (e.g., the Gusev plains or Home
Plate) is similarly difficult to constrain. The Columbia Hills
may have formed as a result of uplift associated with the
formation of ancient craters [McCoy et al., 2008]. These
impacts may have resulted in hydrothermal activity [Newsom
et al., 1999] similar to the volcanic processes suggested for
the development of the Paso Robles class soils. However, it is
perhaps significant in terms of their age that these extraordi-
narily sulfate-rich deposits at Gusev are all found in shallow
soil exposures. Soil on Mars undergoes mixing by processes
that include redistribution by wind and stirring by impacts.
These processes clearly were important at the time of
emplacement of the local bedrock, but they have also
continued, albeit at lower rates, up to the present. Given the
apparently undisturbed nature of the Paso Robles class soils,
Figure 15. Ratio of Wishstone average (blue) and Watch-
tower average (red) compositions to the composition of
Ben?s Clod, a small Ti- and P-rich rock found at the Paso
Robles location. The precursor to Ben?s Clod may have
been of the Wishstone/Watchtower class.
E06S10 YEN ET AL.: HYDROTHERMAL PROCESSES AT GUSEV CRATER
17 of 19
E06S10
it would seem unlikely that they would have remained intact
from a time when impact-related hydrothermal processes
were relevant. In addition, we cannot rule out the possibility
that the local hydrothermal activity that produced Paso
Robles class soils took place significantly more recently than
major volcanic events in Gusev Crater such as the emplace-
ment of Home Plate and the Adirondack class basalts of the
plains. Regardless of the timing, however, the existence of
the Paso Robles class soils indicates that transient warm and
wet regions consistent with localized habitable environments
have been available on Mars.
9. Conclusions
[71] Light-toned soil exposures referred to as Paso Robles
class deposits analyzed by the Spirit rover within Gusev
crater are hydrated and dominated by ferric sulfates, silica,
and magnesium sulfates. Calcium sulfates and phosphates
may also be present in certain samples. The presence of
ferric sulfates indicates that these soils formed under oxi-
dizing and acidic conditions.
[72] These soils contain the elemental signatures of mo-
bile elements in nearby rocks providing compelling evi-
dence for fluid interactions with these rocks prior to
formation of the soil deposits. Compositional differences
across the multiple exposures of Paso Robles class soils
indicate that these fluids were localized. Significant chem-
ical variability over meter scales is evidence that the
quantity of water was limited and that the precipitates did
not form under large-scale equilibrium conditions.
[73] Paso Robles class soils are found immediately be-
neath thin layers of basaltic soil in a variety of topographic
settings including within local depressions and on slopes.
The formation of this class of soils is therefore not con-
trolled by topography. Minimal contamination from basaltic
soil, with the exception of the material mixed in by the rover
wheels, indicates that these soils were discovered at or very
near the locations where they formed.
[74] The likely formation mechanism for Paso Robles
class soils involves hydrothermal fluids and volcanic vapors
rich in sulfur, either from magma degassing or perhaps
through interactions with crustal sulfide deposits, rising to
the surface and producing condensates. The mineral
assemblages, chemical signatures, and geologic setting are
all consistent with this formation scenario.
[75] Localized warm and wet conditions with sulfur-rich
gases provided a potentially habitable environment, al-
though very low pH would have presented a significant
environmental challenge to organisms not adapted to such
conditions.
[76] Acknowledgments. We thank the members of the MER project
who enable daily science observations at the Spirit and Opportunity landing
sites. We thank K. Herkenhoff for Microscopic Imager images, W. Chen for
traverse maps, K. Di for contour plots, and S. McLennan for helpful
discussions. J. Bishop and H. Newsom provided constructive reviews of the
manuscript. A portion of the work described in this paper was conducted at
the Jet Propulsion Laboratory, California Institute of Technology, under a
contract with the National Aeronautics and Space Administration.
References
Africano, F., and A. Bernard (2000), Acid alteration in the fumarolic en-
vironment of Usu volcano, Hokkaido, Japan, J. Volcanol. Geotherm.
Res., 97, 475?495.
Alexander, G. B., W. M. Heston, and R. K. Iler (1954), The solubility of
amorphous silica in water, J. Phys. Chem., 58, 453?455.
Bell, J. F., III, et al. (2003), Mars Exploration Rover Athena Panoramic
Camera (Pancam) investigation, J. Geophys. Res., 108(E12), 8063,
doi:10.1029/2003JE002070.
Bibring, J. P., et al. (2005), Mars surface diversity as revealed by the
OMEGA/Mars Express observations, Science, 307, 1576 ? 1581,
doi:10.1126/science.1108806.
Bigham, J. M., and D. K. Nordstrom (2000), Iron and aluminum hydro-
xysulfates from acid sulfate waters, in Sulfate Minerals: Crystallography,
Geochemistry, and Environmental Significance, Rev. Mineral., vol. 40,
edited by C. N. Alpers, J. L. Jambor, and D. K. Nordstrom, pp. 351?403,
Mineral. Soc. of Am., Washington, D. C.
Burns, R. G., and S. L. Martinez (1991), Mo?ssbauer spectra of olivine-
rich achondrites: Evidence for preterrestrial redox reactions, Proc.
Lunar Planet. Sci., 21, 331?340.
Cabrol, N. A., et al. (2003), Exploring Gusev Crater with Spirit: Review of
science objectives and testable hypotheses, J. Geophys. Res., 108(E12),
8076, doi:10.1029/2002JE002026.
Campbell, J. L., R. Gellert, M. Lee, C. L. Mallet, J. A. Maxwell, and J. M.
O?Meara (2008), Quantitative in situ determination of hydration of bright
high-sulfate Martian soils, J. Geophys. Res., 113, E06S11, doi:10.1029/
2007JE002959.
Channing, A., and I. B. Butler (2007), Cryogenic opal-A deposition from
Yellowstone hot springs, Earth Planet. Sci. Lett., 257, 121?131.
Christensen, P. R., et al. (2003), Miniature Thermal Emission Spectrometer
for the Mars Exploration Rovers, J. Geophys. Res., 108(E12), 8064,
doi:10.1029/2003JE002117.
Christensen, P. R., et al. (2004), Initial results from the Mini-TES experi-
ment in Gusev Crater from the Spirit Rover, Science, 305, 837?842,
doi:10.1126/science.1100564.
Clark, B. C., and A. K. Baird (1979), Is the Martian lithosphere sulfur rich?,
J. Geophys. Res., 84, 8395?8403.
Clark, B. C., et al. (2005), Chemistry and mineralogy of outcrops at Mer-
idiani Planum, Earth Planet. Sci. Lett., 240, 73?94.
Clark, B. C., et al. (2007), Evidence for Montmorillonite or its composi-
tional precursors in Columbia Hills, Mars, J. Geophys. Res., 112,
E06S01, doi:10.1029/2006JE002756.
Crumpler, L. S., T. McCoy, M. Schmidt, and N. Cabrol (2007), Physical
volcanology at Gusev Crater, Spirit Rover, in Seventh International Con-
ference on Mars, Abstract 3385, Jet Propul. Lab., Pasadena, Calif.
Fournier, R. O., and R. W. Potter II (1982), An equation correlating the
solubility of quartz in water from 20
 to 900
C at pressures up to 10,000
bars, Geochim. Cosmochim. Acta, 46, 1969?1973.
Gellert, R., et al. (2006), Alpha Particle X-Ray Spectrometer (APXS):
Results from Gusev Crater and calibration report, J. Geophys. Res.,
111, E02S05, doi:10.1029/2005JE002555.
Getahun, A., M. H. Reed, and R. Symonds (1996), Mount St. Augustine
volcano fumarole wall rock alteration: Mineralogy, zoning, composition
and numerical models of its formation process, J. Volcanol. Geotherm.
Res., 71, 73?107.
Glotch, T. D., and J. L. Bandfield (2006), Determination and interpretation
of surface and atmospheric Miniature Thermal Emission Spectrometer
spectral end-members at the Meridiani Planum landing site, J. Geophys.
Res., 111, E12S06, doi:10.1029/2005JE002671.
Glotch, T. D., J. L. Bandfield, P. R. Christensen, W. M. Calvin, S. M.
McLennan, B. C. Clark, A. D. Rogers, and S. W. Squyres (2006), Miner-
alogy of the light-toned outcrop at Meridiani Planum as seen by the
Miniature Thermal Emission Spectrometer and implications for its for-
mation, J. Geophys. Res., 111, E12S03, doi:10.1029/2005JE002672.
Golombek, M. P., et al. (2003), Selection of the Mars Exploration Rover land-
ing sites, J. Geophys. Res., 108(E12), 8072, doi:10.1029/2003JE002074.
Griffes, J. L., R. E. Arvidson, F. Poulet, and A. Gendrin (2007), Geologic
and spectral mapping of etched terrain deposits in northern Meridiani
Planum, J. Geophys. Res., 112, E08S09, doi:10.1029/2006JE002811.
Herkenhoff, K. E., et al. (2003), Athena Microscopic Imager investigation,
J. Geophys. Res., 108(E12), 8065, doi:10.1029/2003JE002076.
Jambor, J. L., D. K. Nordstrom, and C. N. Alpers (2000), Metal-sulfate salts
from sulfide mineral oxidation, in Sulfate Minerals: Crystallography,
Geochemistry, and Environmental Significance, Rev. Mineral., vol. 40,
edited by C. N. Alpers, J. L. Jambor, and D. K. Nordstrom, pp. 303?350,
Mineral. Soc. of Am., Washington, D. C.
Joeckel, R. M., B. J. Ang Clement, and L. R. VanFleet Bates (2005),
Sulfate-mineral crusts from pyrite weathering and acid rock drainage in
the Dakota Formation and Graneros Shale, Jefferson County, Nebraska,
Chem. Geol., 215, 433?452, doi:10.1016/j.chemgeo.2004.06.044.
Johnson, J. R., J. F. Bell III, E. Cloutis, M. Staid, W. H. Farrand, T. McCoy,
M. Rice, A. Wang, and A. Yen (2007), Mineralogic constraints on sulfur-
rich soils from Pancam spectra at Gusev crater, Mars, Geophys. Res. Lett.,
34, L13202, doi:10.1029/2007GL029894.
E06S10 YEN ET AL.: HYDROTHERMAL PROCESSES AT GUSEV CRATER
18 of 19
E06S10
Keith, D. C., D. D. Runnells, K. J. Esposito, J. A. Chermak, D. B. Levy,
S. R. Hannula, M. Watts, and L. Hall (2001), Geochemical models of
the impact of acidic groundwater and evaporative sulfate salts on
Boulder Creek at Iron Mountain, California, Appl. Geochem, 16,
947?961.
Keith, T. E. C. (1991), Fossil and active fumaroles in the 1912 eruptive
deposits, Valley of Ten Thousand Smokes, Alaska, J. Volcanol.
Geotherm. Res., 45, 227?254.
Klingelho?fer, G., et al. (2003), Athena MIMOS II Mo?ssbauer spectrometer
investigation, J. Geophys. Res., 108(E12), 8067, doi:10.1029/
2003JE002138.
Klingelho?fer, G., et al. (2004), Jarosite and hematite at Meridiani Planum
from Opportunity?s Mo?ssbauer Spectrometer, Science, 306, 1740?1745,
doi:10.1126/science.1104653.
Knoll, A. H., et al. (2005), An astrobiological perspective on Meridiani
Planum, Earth Planet. Sci. Lett., 240, 179 ? 189, doi:10.1016/
j.epsl.2005.09.045.
Krauskopf, K. B. (1956), Dissolution and precipitation of silica at low
temperatures, Geochim. Cosmochim. Acta, 10, 1?26.
Lane, M. D., J. L. Bishop, M. D. Dyar, M. Parente, P. L. King, and B. C.
Hyde (2007), The ferric sulfate and ferric phosphate minerals in the light-
toned Paso Robles rover track soils: A multi-instrument analysis, in
Seventh International Conference on Mars, Abstract 3331, Jet Propul.
Lab., Pasadena, Calif.
Larsen, K. W., R. E. Arvidson, B. L. Jolliff, and B. C. Clark (2000),
Correspondence and least squares analyses of soil and rock compositions
for the Viking Lander 1 and Pathfinder landing sites, J. Geophys. Res.,
105, 29,207?29,221.
McCoy, T. J., et al. (2008), Structure, stratigraphy, and origin of Husband
Hill, Columbia Hills, Gusev Crater, Mars, J. Geophys. Res., doi:10.1029/
2007JE003041, in press.
McKay, C. P. (1997), The search for life on Mars, Origins Life Evol.
Biosphere, 27, 263?289.
McLennan, S. M., et al. (2005), Provenance and diagenesis of the evapor-
ite-bearing Burns formation, Meridiani Planum, Mars, Earth Planet. Sci.
Lett., 240, 95?121, doi:10.1016/j.epsl.2005.09.041.
McSween, H. Y., and A. H. Treiman (1998), Martian meteorites, in
Planetary Materials, Rev. Mineral., vol. 36, edited by J. J. Papike,
pp. 6-1 to 6-53, Mineral. Soc. of Am., Washington, D. C.
Ming, D. W., et al. (2006), Geochemical and mineralogical indicators for
aqueous processes in the Columbia Hills of Gusev crater, Mars, J. Geo-
phys. Res., 111, E02S12, doi:10.1029/2005JE002560.
Mittlefehldt, D. W., R. Gellert, T. McCoy, H. Y. McSween Jr., R. Li, and the
Athena Science Team (2006), Possible Ni-rich mafic-ultramafic mag-
matic sequence in the Columbia Hills: Evidence from the Spirit rover,
Lunar Planet. Sci., XXXVII, 1505.
Morris, R. V., et al. (2000), Acid sulfate alteration products of a tholeiitic
basalt: Implications for interpretation of Martian thermal emission spec-
tra, Lunar Planet. Sci. Conf., XXXI, 2014.
Morris, R. V., et al. (2004), Mineralogy at Gusev Crater from the Mo?ss-
bauer spectrometer on the Spirit Rover, Science, 305, 833 ? 836,
doi:10.1126/science.1100020.
Morris, R. V., et al. (2006), Mo?ssbauer mineralogy of rock, soil, and dust at
Gusev crater, Mars: Spirit?s journey through weakly altered olivine basalt
on the plains and pervasively altered basalt in the Columbia Hills,
J. Geophys. Res., 111, E02S13, doi:10.1029/2005JE002584.
Morris, R. V., et al. (2007), Possible evidence for iron sulfate, iron sulfides,
and elemental sulfur at Gusev Crater, Mars, from MER, CRISM, and
analog data, in Seventh International Conference on Mars, Abstract 3393,
Jet Propul. Lab., Pasadena, Calif.
Nealson, K. H. (1999), Post-Viking microbiology: New approaches, new
data, new insights, Origins Life Evol. Biosphere, 29, 73?93.
Newsom, H. E., J. J. Hagerty, and F. Goff (1999), Mixed hydrothermal fluids
and the origin of the Martian soil, J. Geophys. Res., 104, 8717?8728.
Nomura, K., M. Takeda, T. Iiyama, and H. Sakai (2005), Mo?ssbauer studies of
jarosite, mikasaite and yavapaiite, and implication to their Martian counter-
parts,Hyperfine Interact., 166, 657?664, doi:10.1007/s10751-006-9337-y.
Nordstrom, D. K., and C. N. Alpers (1999), Negative pH, efflorescent
mineralogy, and consequences for environmental restoration at the Iron
Mountain Superfund site, California, Proc. Natl. Acad. Sci. U. S. A., 96,
3455?3462.
Ohba, T., and M. Nakagawa (2002), Minerals in volcanic ash 2: Non-
magmatic minerals, Global Environ. Res., 6, 53?59.
Rieder, R., R. Gellert, J. Bru?ckner, G. Klingelho?fer, G. Dreibus, A. Yen, and
S. W. Squyres (2003), The new Athena alpha particle X-ray spectrometer
for the Mars Exploration Rovers, J. Geophys. Res., 108(E12), 8066,
doi:10.1029/2003JE002150.
Squyres, S. W., et al. (2006), Rocks of the Columbia Hills, J. Geophys.
Res., 111, E02S11, doi:10.1029/2005JE002562.
Squyres, S. W., et al. (2007), Pyroclastic activity at Home Plate in Gusev
Crater, Mars, Science, 316, 738?742, doi:10.1126/science.1139045.
Stoiber, R. E., and W. I. Rose (1970), The geochemistry of Central Amer-
ican volcanic gas condensates, Geol. Soc. Am. Bull., 81, 2891?2912.
Stoiber, R. E., and W. I. Rose (1974), Fumarole incrustations at active
Central American volcanoes, Geochim. Cosmochim. Acta, 38, 495?516.
Symonds, R. B., W. I. Rose, M. H. Reed, F. E. Lichte, and D. L. Finnegan
(1987), Volatilization, transport and sublimation of metallic and non-me-
tallic elements in high temperature gases at Merapi Volcano, Indonesia,
Geochim. Cosmochim. Acta, 51, 2083?2101.
Tosca, N. J., and S. M. McLennan (2007), An experimental approach to
evaporation processes at the Martian surface, Lunar Planet. Sci.,
XXXVIII, Abstract 1893.
Wang, A., et al. (2007), Sulfate-rich soils exposed by Spirit rover at multi-
ple locations in Gusev Crater on Mars, in Seventh International Confer-
ence on Mars, Abstract 3348, Jet Propul. Lab., Pasadena, Calif.
White, D. E., W. W. Brannock, and K. J. Murata (1956), Silica in hot-spring
waters, Geochim. Cosmochim. Acta., 10, 27?59.
Yen, A. S. (2006), Martian sulfates as observed by the Mars Exploration
Rover Mo?ssbauer and alpha particle X-ray spectrometers, Eos Trans.
AGU, 87(52), Fall Meet. Suppl., Abstract P22A-02.
Yen, A. S., et al. (2005), An integrated view of the chemistry and miner-
alogy of Martian soils, Nature, 436, 49?54, doi:10.1038/nature03637.
Zimbelman, D. R., R. O. Rye, and G. N. Breit (2005), Origin of secondary
sulfate minerals on active andesitic stratovolcanoes, Chem. Geol., 215,
37?60, doi:10.1016/j.chemgeo.2004.06.056.

R. Arvidson, Department of Earth and Planetary Sciences, Washington
University Campus Box 1169, One Brookings Drive, St. Louis, MO 63130,
USA.
B. C. Clark, Lockheed Martin Corporation, MS S8000, P.O. Box 179,
12257 State Highway 121, Littleton, CO 80127, USA.
R. Gellert, Department of Physics, University of Guelph, Guelph, ON,
Canada N1G 2W1.
J. Hurowitz and A. S. Yen, Jet Propulsion Laboratory, Mail Stop 183-501,
4800 Oak Grove Drive, Pasadena, CA 91109, USA.
J. R. Johnson, U.S. Geological Survey, Astrogeology Team, 2255 N.
Gemini Drive, Flagstaff, AZ 86001, USA.
A. T. Knudson, Department of Earth and Planetary Sciences, Arizona
State University, Mail Code 6305, Tempe, AZ 85287, USA.
R. Li, CEEGS/Center for Mapping, Ohio State University, 470 Hitchcock
Hall, 2070 Neil Avenue, Columbus, OH 43210, USA.
T. McCoy and M. Schmidt, Department of Mineral Sciences, National
Museum of Natural History, Smithsonian Institution, 10th and Constitution
Avenues, N.W., Washington, DC 20560, USA.
D. W. Ming, NASA Johnson Space Center, Code KX, 2101 NASA
Parkway, Houston, TX 77058, USA.
D. W. Mittlefehldt and R. V. Morris, NASA Johnson Space Center, Code
KR, 2101 NASA Parkway, Houston, TX 77058, USA.
S. Squyres, Department of Astronomy, Cornell University, 428 Space
Sciences, Ithaca, NY 14853, USA.
E06S10 YEN ET AL.: HYDROTHERMAL PROCESSES AT GUSEV CRATER
19 of 19
E06S10