66
Please print only if necessary Please use double-sided printing Please recycle paper Thanks! Please print only if necessary Please use double-sided printing Please recycle paper Thanks! Weathering of olivine under CO2 atmosphere: A martian perspective 1 2 E. Dehouck a,* , A. Gaudin a , N. Mangold a , L. Lajaunie b , A. Dauzères c , O. Grauby d , E. Le 3 Menn a 4 5 a Laboratoire de Planétologie et Géodynamique de Nantes (LPGN), CNRS/Université de 6 Nantes, 44322 Nantes, France 7 b Institut des Matériaux Jean Rouxel (IMN), CNRS/Université de Nantes, 44322 Nantes, 8 France 9 c Institut de Radioprotection et de Sûreté Nucléaire (IRSN), Fontenay-aux-Roses, France 10 d Centre Interdisciplinaire de Nanoscience de Marseille (CINaM), CNRS/Aix-Marseille 11 Université, Campus de Luminy, 13288 Marseille, France 12 13 14 *Corresponding author. Now at: Department of Geosciences, State University of New York at 15 Stony Brook, Stony Brook, NY 11794-2100, USA. 16 E-mail address: [email protected] 17 18 19

Weathering of olivine under CO2 atmosphere: a martian perspective

  • Upload
    uca-es

  • View
    4

  • Download
    0

Embed Size (px)

Citation preview

Please print only if necessary – Please use double-sided printing – Please recycle paper – Thanks!

Please print only if necessary – Please use double-sided printing – Please recycle paper – Thanks!

Weathering of olivine under CO2 atmosphere: A martian perspective 1

2

E. Dehouck a,*, A. Gaudin a, N. Mangold a, L. Lajaunie b, A. Dauzères c, O. Grauby d, E. Le 3

Menn a 4

5

aLaboratoire de Planétologie et Géodynamique de Nantes (LPGN), CNRS/Université de 6

Nantes, 44322 Nantes, France 7

bInstitut des Matériaux Jean Rouxel (IMN), CNRS/Université de Nantes, 44322 Nantes, 8

France 9

cInstitut de Radioprotection et de Sûreté Nucléaire (IRSN), Fontenay-aux-Roses, France 10

dCentre Interdisciplinaire de Nanoscience de Marseille (CINaM), CNRS/Aix-Marseille 11

Université, Campus de Luminy, 13288 Marseille, France 12

13

14

*Corresponding author. Now at: Department of Geosciences, State University of New York at 15

Stony Brook, Stony Brook, NY 11794-2100, USA. 16

E-mail address: [email protected] 17

18

19

2

ABSTRACT 20

21

Recent analyses from the Curiosity rover at Yellowknife Bay (Gale crater, Mars) show 22

sedimentary rocks deposited in a lacustrine environment and containing smectite clays 23

thought to derive from the alteration of olivine. However, little is known about the weathering 24

processes of olivine under early martian conditions, and about the stability of smectite clays in 25

particular. Here, we present a 3-month experiment investigating the weathering of forsteritic 26

olivine powders (Fo90) under a dense CO2 atmosphere, and under present-day terrestrial 27

conditions for comparison. The experiment also evaluates the potential effects of hydrogen 28

peroxide (H2O2), as a representation of the highly oxidizing compounds produced by 29

photochemical reactions throughout martian history. The weathered samples were 30

characterized by means of near-infrared spectroscopy (NIR), X-ray diffraction (XRD), 31

transmission electron microscopy with energy dispersive X-ray spectrometry (TEM-EDX), 32

Mössbauer spectroscopy and thermogravimetry. The results show that a Mg-rich smectite 33

phase formed from the weathering of olivine under CO2 conditions, although in lower 34

abundance than under terrestrial conditions. The main secondary phase formed under CO2 35

turns out to be a silica-rich phase (possibly acting as a “passivating” layer) with a non-36

diagnostic near-infrared spectral signature. The use of H2O2 highlights the critical importance 37

of both the redox conditions and Fe content of the initial olivine on the nature of the 38

secondary phases. 39

40

Keywords: Mars, alteration, olivine, smectites, phyllosilicates, silica. 41

42

43

3

1. INTRODUCTION 44

45

The martian crust is known to have undergone aqueous alteration in its early history: 46

various secondary minerals – including clays (mostly Fe-Mg smectites), hydrated sulfates, Fe-47

(oxy)hydroxides and carbonates – have been identified during the last decade by 48

hyperspectral imagery, almost exclusively in terrains of Noachian or Hesperian age (Bibring 49

et al., 2005; 2006; Poulet et al., 2005; Carter et al., 2013). While the detections of these 50

secondary minerals and the involvement of water in their formation are widely accepted, the 51

climatic and geological settings in which such processes occurred remain controversial. The 52

correlation between hydrous minerals and valley networks, alluvial/delta fans and paleolakes 53

in ancient terrains (e.g., Ehlmann et al., 2008a; Dehouck et al., 2010; Ansan et al., 2011) 54

suggest that at least some of the secondary minerals were formed by rock-water-atmosphere 55

interactions, such as weathering. Numerous observations of vertical sections of Al-rich clays 56

over Fe/Mg-smectites, typical of pedogenesis processes, also favors such a scenario (e.g., 57

Loizeau et al., 2007; 2010; Ehmann et al., 2009; Murchie et al., 2009; Wray et al., 2009; Noe 58

Dobrea et al., 2010; Gaudin et al., 2011; Le Deit et al. 2012; Carter et al., 2013). However, 59

this hypothesis has been challenged based on the “missing” widespread carbonates – which 60

are expected to form in a thick CO2 atmosphere – (e.g., Bullock and Moore, 2007; Fernandez-61

Remolar et al., 2011) and the difficulty to model an early martian atmosphere having the 62

adequate properties for liquid water to be available (e.g., Forget et al., 2012; Wordsworth et 63

al., 2012). This led some authors to propose that most of the martian clays have been formed 64

at depth (i.e., isolated from the atmosphere) by hydrothermal groundwater circulation, and 65

later exposed at the surface by erosion (e.g., Ehlmann et al., 2011). 66

Therefore, it is important to better understand how primary rocks and minerals 67

weather under Mars-relevant surface conditions in order to evaluate if the secondary 68

4

mineralogy currently observed on Mars may derive from rock-water-atmosphere interactions 69

(i.e., weathering), or if other processes (deep, hydrothermal alteration) should be favored 70

instead. Specifically, it is important to clarify if Fe/Mg-smectites – which are the most 71

common clay minerals found on Mars (Carter et al., 2013) – can be produced by weathering 72

under a dense CO2 atmosphere and associated low pH, because it is known that low pH 73

conditions tend to destabilize Fe/Mg smectites (e.g., Galan et al., 1999; Bauer et al., 2001; 74

Murakami et al., 2004; Amram and Ganor, 2005; Golubev et al., 2006) and to favor Al-rich 75

clays instead (e.g., Komadel and Madejová, 2006; and references therein). Yet, X-ray 76

diffraction analyses from the Curiosity rover of fine-grained lacustrine sedimentary rocks 77

display patterns consistent with a proportion of ~20% of a smectite clay, most probably a 78

ferroan saponite, suggesting that significant alteration occurred at Gale crater (Vaniman et al., 79

2014). The lack of high temperature phases (e.g., serpentine, chlorite) and the overall 80

mineralogy suggest a formation by alteration of forsteritic olivine through low-grade 81

diagenesis (McLennan et al., 2014), thus highlighting the role of olivine in alteration 82

processes at this place. 83

In this context, only few experimental studies so far have compared the weathering of 84

primary silicates under terrestrial and simulated martian conditions. Accordingly, the main 85

goal of this study was to determine the effects of early martian conditions on the weathering 86

of forsteritic olivine (Fo90), with a particular focus on the secondary products. Owing to its 87

nesosilicate structure (unpolymerized silica tetrahedra), olivine is one of the most reactive 88

silicates in near-surface conditions and, for this reason, is particularly appropriate for such 89

study. Using four batch experiments, we evaluated the effects of a CO2 atmosphere, but also 90

those of highly oxidizing compounds (represented here by hydrogen peroxide, H2O2) that 91

have been detected on present-day Mars and may have been more abundant in the past. A 92

detailed chemical, petrographic and mineralogical characterization of the final samples was 93

5

achieved using an appropriate set of analytical techniques, including near-infrared 94

spectroscopy (NIR), X-ray diffraction (XRD), transmission electron microscopy with energy 95

dispersive X-ray spectrometry (TEM-EDX), Mössbauer spectroscopy and thermogravimetry. 96

Implications for the alteration processes on early Mars are discussed. 97

98

2. APPROACH AND METHODS 99

100

2.1. Background: previous work on olivine alteration under CO2 101

102

Because of its high chemical reactivity, olivine has been frequently used in laboratory 103

experiments examining the dissolution rates of silicates in various conditions (e.g., Wogelius 104

and Walther, 1992; Pokrovsky and Schott, 2000; Golubev et al., 2005; Hänchen et al., 2006). 105

Olsen and Rimstidt (2007) have summarized the results of numerous of these studies and used 106

them to evaluate the lifetime of an olivine grain exposed to weathering on the martian surface. 107

Taking several parameters into account (temperature, composition, grain size, pH), they 108

estimated values ranging from a few thousand and several million years (up to 30 Ma for a 109

forsterite grain of 1 mm at 273 K and a pH of 7.5). 110

A number of experiments have also examined the potential of forsteritic olivine for 111

CO2 sequestration, in the context of Earth’s climate change mitigation (e.g., Giammar et al., 112

2005; Gerdemann et al., 2007; Bearat et al., 2006; Garcia et al., 2010; King et al., 2010; Daval 113

et al., 2011). These experiments have led to a good understanding of the olivine carbonation 114

process. The most common secondary phases reported are magnesite and amorphous silica. 115

Although the results of these studies are not directly applicable to weathering processes on 116

early Mars because of their high temperatures (up to several hundred degrees C) and CO2 117

pressures (up to several hundred bars), they can provide relevant information about the 118

6

mechanisms and environmental parameters that control the alteration of olivine under CO2 119

conditions. Nevertheless, the discrepancy between the results of such studies and the lack of 120

widespread carbonates on Mars justifies studying the alteration of olivine in conditions that 121

are more relevant to surface processes in terms of temperature and pressure. 122

Lastly, the experiments conducted by Dehouck et al. (2012) included the weathering 123

of forsteritic olivine, among other silicates, under simulated early martian conditions – 0.8 bar 124

of CO2, with or without H2O2 – and at low liquid-to-rock (L/R) ratio. The authors reported the 125

formation of nesquehonite, an hydrated Mg-carbonate. 126

127

2.2. Selection of the starting material 128

129

As a starting material, we selected a natural sample of forsteritic (Mg-rich) olivine 130

from San Carlos, Arizona (USA). Olivine is a common component of ultramafic and mafic 131

rocks, from which the martian crust is mainly made. It has been unambiguously detected in 132

various regions of Mars based on orbital data (Hoefen et al., 2003; Mustard et al., 2005; 133

Poulet et al., 2007; Ody et al., 2013) and, despite the global enrichment in Fe of the martian 134

crust (e.g., Longhi et al., 1992), most of the detections correspond to forsteritic and/or fine-135

grained olivine (both Fe content and grain size have a similar influence on the NIR signature 136

of olivine, so that it is difficult to retrieve definitive compositions from orbit; Ody et al., 137

2013). Moreover, among the secondary minerals discovered so far in olivine-rich regions of 138

Mars (Mangold et al., 2007; Ehlmann et al., 2009; Gaudin et al., 2011; Bishop et al., 2013), 139

several species of phyllosilicates (smectites, serpentines) and carbonates (magnesite, 140

hydromagnesite) are known to derive from olivine or olivine-bearing rocks in terrestrial 141

environments (e.g., Delvigne et al., 1979; Velbel et al., 1991; Wilson, 2004). Finally, 142

forsteritic olivine is weakly resistant to weathering, but it is also sufficiently common on the 143

7

surface of the Earth – as opposed to fayalitic olivine, in particular – to be found in “fresh” 144

state and in sufficient amount (>150 g) to meet the requirements of this study. 145

146

2.3. Weathering conditions on early Mars 147

148

Designing a laboratory experiment under “simulated early martian conditions” 149

requires making some assumptions. Based on the present-day martian atmosphere, composed 150

of >95% CO2 (Owen, 1982), the easiest way to obtain higher surface temperatures is by 151

increasing the total pressure. Recent results from general circulation models suggest that 152

additional greenhouse gases or other factors (impacts, volcanism) may have been necessary to 153

reach temperatures above the freezing point (e.g., Forget et al., 2012). However, since these 154

factors are still poorly constrained, we have decided to consider only the effect of a dense CO2 155

atmosphere in the present study. Other experiments are devoted to test the effect of SO2 on the 156

weathering processes (e.g., Chevrier et al., 2012). 157

Another hypothesis of our study is that the highly oxidizing compounds found in the 158

regolith (Bullock et al., 1994; Zent, 1998; Yen et al., 2000; Hurowitz et al., 2007) and in the 159

atmosphere (Clancy et al., 2004; Encrenaz et al., 2004) of present-day Mars may have been 160

more abundant at the time of formation of the alteration minerals. Indeed, these compounds 161

derive from the UV-induced photolysis of water; thus, if water was more abundant on early 162

Mars, highly oxidizing compounds are expected to be more abundant, too. In our experiment, 163

these oxidizing compounds have been represented by hydrogen peroxide (H2O2). Although 164

limited to the first meters or first tens of meters of the soil (Bullock et al., 1994; Zent, 1998), 165

hydrogen peroxide combined with a dense CO2 atmosphere may have influenced alteration 166

processes over long time scales by allowing the existence of both (moderately) acid and 167

(highly) oxidizing solutions (e.g., Chevrier et al., 2006; Hurowitz et al., 2010; Fernandez-168

8

Remolar et al., 2011). Yet, only few experimental studies to date have explored its effect on 169

the weathering of silicates (Dehouck et al., 2012). Because our experimental device is 170

exposed to sunlight and heat, H2O2 is expected to rapidly undergo disproportionation into 171

H2O and O2 during the course of the experiment. However, since disproportionation does not 172

involve transfer of an electron, the redox state of the system is not changed and the effect of 173

highly oxidizing conditions can still be tested. 174

175

2.4. Protocols 176

177

2.4.1. Starting material: preparation and characterization 178

179

The forsteritic olivine used in this study originated from San Carlos, Arizona (USA) 180

and consisted of olive-green, centimeter-sized monocrystals. Preliminary SEM-EDX 181

(scanning electron microscopy coupled with energy dispersive X-ray spectrometry) analyses 182

(N=24) performed on polished grains confirmed the mean formula to be ~Fo90Fa10 (Fo: 183

forsterite; Fa: fayalite). They also revealed the presence in minor quantity of small (a few tens 184

of micrometers for the longest ones; Fig. 1), elongated crystals containing P, Ca as well as a 185

little F, and interpreted to be apatite. 186

The cleanest grains were selected from the original lot and minor bright impurities 187

macroscopically visible on the surface of some grains were removed using a diamond-covered 188

abrasion tool. The grains were then cleaned using an ultrasonic bath in ethyl alcohol and 189

finely crushed using an automatic crusher with agate balls. The resulting, white-colored 190

powder was homogenized. SEM observations showed that >98% of the powder grains were 191

<60 µm in size. 192

9

To monitor the purity of the olivine powder before the experiment, we performed a 193

precise mineralogical and chemical characterization using near-infrared spectroscopy (NIR), 194

powder X-ray diffraction (XRD), and inductively coupled plasma optical emission 195

spectroscopy (ICP-OES). The NIR spectrum showed a highly pure olivine with very limited 196

hydration (Fig. 1). Shallow absorption bands at 2.31-µm (Hunt et al., 1972) and 4.02 µm are 197

attributable to minor apatite (Fig. 1). XRD pattern showed no crystallized phase other than 198

olivine. Finally, the chemical analysis by ICP-OES (Table 1) again confirmed the 199

composition of the olivine to be Fo90Fa10. Apart from Si, Mg and Fe, other elements like Mn, 200

Ca, Al and Na are also detected in small amounts. In contrast, P is below the detection limit, 201

which allows determining an upper limit of 0.12 wt% for the quantity of apatite contained in 202

the powder. 203

204

2.4.2. Experimental device 205

206

For the purposes of this study, we designed and assembled the apparatus shown in Fig. 207

2. Its primary function is to provide controlled weathering conditions in terms of atmospheric 208

composition and temperature. It consists of four borosilicate glass (Schott® Duran) flasks – or 209

“reactors” hereafter – of 1 liter, each placed on a hot plate. The first two reactors are 210

hermetically closed and linked to a stainless-steel and PTFE circuit allowing introduction of 211

gaseous CO2 (Air Liquide® Alphagaz CO2 N45). Independent gas lines and valves ensure 212

that each reactor remains fully isolated from the second one. A pair of manometers provides a 213

direct visual control of the internal pressure of the two reactors. The apparatus also allows 214

sampling both the solution and the gas without exposing the samples to ambient air. Lastly, a 215

stainless-steel thermometer is immersed inside the solution to continuously control its 216

temperature. 217

10

The two other reactors are filled with ambient air, but are also hermetically closed to 218

prevent escape of water vapor or other gases originating from the solution. They are not 219

equipped for pressure monitoring or gas sampling. 220

221

2.4.3. Initial protocol 222

223

For each reactor, we precisely weighted 30 g of sample and poured it into 300 mL of 224

liquid, thus giving a L/R ratio of 10. In two reactors, the liquid consisted only of ultrapure 225

water with a resistivity of 18 MΩ.cm. In the two other reactors, it consisted of hydrogen 226

peroxide (H2O2; VWR® Prolabo GPR Rectapur) at ~1 vol% in ultrapure water. 227

The four reactors were then mounted on the experimental device. The two reactors 228

used for the terrestrial conditions (one with H2O2 – hereafter abbreviated as “Air-H2O2” – and 229

the other without H2O2 – hereafter “Air”; Fig. 2) were simply closed with their original 230

stopper and put on the hot plates. The two reactors used for simulated early martian 231

conditions were connected to the circuit described above, which includes its own stoppers. 232

Then, the two reactors were purged by a continued flow of CO2 (the solutions “bubbled”), 233

while the valve for gas sampling stayed opened (Fig. 2). The purge was maintained during 234

one hour for the reactor without H2O2 (hereafter abbreviated as “CO2”) in order to completely 235

remove dissolved nitrogen and oxygen. In contrast, this operation was limited to a few 236

minutes for the other reactor containing H2O2 (hereafter “CO2-H2O2”) since hydrogen 237

peroxide was expected to produce O2 anyway. The apparatus was then closed and the initial 238

CO2 pressure in the “Mars reactors” was set at ~1.5 bar. This value slightly above the ambient 239

atmospheric pressure was chosen to ensure that any tiny leak would cause the escape of CO2 240

and not the entrance of ambient air. This value is also well within the range of atmospheric 241

11

pressure usually explored by global circulation models of early Mars (e.g., Forget et al., 242

2012). 243

Lastly, the temperature of the hot plates (outside the reactor) was set at ~70 °C in 244

order to obtain a solution temperature of ~45 °C (± 5 °C, depending on the room 245

temperature). Aluminum sheets were wrapped around the reactors to help maintaining a 246

homogeneous temperature. The value of 45 °C was chosen as a compromise between two 247

opposite requirements: (1) increasing reaction rates to maintain the experiment in a reasonable 248

timeframe and (2) keeping realistic conditions for surface aqueous processes. To avoid a non-249

homogeneous distribution of the secondary products within the olivine powders due to the 250

slight temperature gradient within the solution, the reactors were manually shaken daily 251

during the course of the experiment. 252

253

2.4.4. Monitoring, solution sampling and final samples retrieval 254

255

Throughout the experiment, a daily reading of temperature and pressure was achieved, 256

so that adjustments could be made if necessary. In particular, our apparatus would allow the 257

reintroduction of gaseous CO2 in the “Mars reactors” in the case of a pressure decrease under 258

a security level that we placed at 1.2 bar. However, this operation never became necessary, 259

demonstrating the efficient airtightness of the apparatus (our estimated leak rate is <1 mbar 260

per day). 261

As expected, hydrogen peroxide rapidly underwent disproportionation into H2O and 262

O2 under the combined action of sunlight and heat. As a result, the pressure inside the “CO2-263

H2O2” reactor reached 2.8 bar after 3 days (1.5 bar of CO2 and 1.3 bar of O2). The same 264

process occurred inside the “Air-H2O2” reactor, with an expected – but not verified – total 265

pressure of 2.3 bar (1 bar of air and 1.3 bar of additional O2). 266

12

The solutions of the four reactors were sampled after 3, 15 and 31 days. A final 267

sampling was made after 95 days, just before the end of the experiment. At each sampling, 268

~10 mL of solution was pumped using a syringe, so that the decrease of the L/R ratio after 30 269

days was relatively limited (from 10 to 9). Each aliquot was filtered at 0.2 µm and its pH was 270

immediately measured. Note here that for both “Earth reactors”, the pH values at day 3 were 271

dismissed because of unstable measures (probably due to the low concentrations of 272

electrolytes in the solutions; see section 3.1.2). Then, the solution was acidified at 2% with 273

nitric acid (HNO3) and analyzed for major elements – Na, Mg, Al, Si, P, K, Ca, Ti, Mn, Fe – 274

by ICP-OES at the SARM-CRPG laboratory (Vandoeuvre-lès-Nancy, France). A blank 275

sample (obtained from a 3-month blank experiment using ultrapure water but no solid) was 276

also analyzed in order to check any possible contamination originating from the experimental 277

device or from the equipments used to manipulate the sampled solutions. 278

Concerning the “Air-H2O2” reactor, the opening of the stopper caused the escape of 279

excess O2 originating from the disproportionation of hydrogen peroxide. As a consequence, 280

the appropriate quantity of liquid H2O2 was poured inside the reactor after each sampling in 281

order to replace the lost O2. 282

The solution sampling at 95 days marked the end of the experiment. The hot plates 283

were turned off and a gas sampling was achieved for the two “Mars reactors” (see next 284

section). Two distinct methods were used to collect the weathered olivine powders: the “CO2” 285

and “CO2-H2O2” reactors were vacuum-dried in order to avoid contact with the room 286

atmosphere, whereas the “Air” and “Air-H2O2” flasks were dried in an oven at 50°C. In both 287

cases, a portion of the solution was first extracted with a syringe to facilitate the drying 288

process. The drying procedure lasted ~2-3 days for each reactor. 289

290

2.4.5. Gas sampling and analysis 291

13

292

Our experimental device is equipped with a gas sampling outlet for each of the “Mars 293

reactors” (Fig. 2). This sampling was achieved only at the end of the experiment (95 days) to 294

avoid any disturbance of the system. The internal air of the stainless-steel bulbs used for the 295

sampling was pumped for one hour before allowing the gas of the reactors to enter. Then, the 296

samples were analyzed by gas chromatography (GC) at the Subatech laboratory (École des 297

Mines, University of Nantes, France). To calibrate the instrument, we used the CO2 bottle of 298

the experimental device (Air Liquide® Alphagaz CO2 N45) as well as two gas mixtures 299

optimized for low O2 concentrations (0.5% O2–99.5% CO2 and 5% O2–95% CO2; Air 300

Liquide® Crystal). Outside this range and for any other gases, measured values must be 301

considered as semi-quantitative. 302

As expected, the analysis of the gas from the “CO2-H2O2” reactor shows a high 303

quantity of O2 produced by the disproportionation of hydrogen peroxide (~2/3 of the mixture, 304

but because of the calibration procedure, the uncertainty is large). In addition, a very small 305

quantity of dihydrogen (H2) is detected (<0.1%). 306

The measured O2 concentration in the “CO2” reactor is minimal: 0.17%, i.e. in the 307

same order of magnitude as in the present-day atmosphere of Mars (Owen, 1982). 308

Furthermore, this value is an upper limit, since some small contamination may have occurred 309

during the preparation and analysis of the gas sample. H2 is detected at a higher level than for 310

the “CO2-H2O2” reactor (~0.5%, so higher than O2), indicating that an oxidation process 311

occurred in the absence of free O2, by dissociation of H2O molecules (e.g., Hurowitz et al., 312

2010). In contrast, no traces of methane have been found (Neubeck et al., 2011). 313

314

2.5. Analytical methods for the solid samples 315

316

14

All analytical methods described below were applied to the bulk powder samples, and 317

some of them were also applied to the <2-µm fraction of these powders, which was separated 318

by settling according to Stokes’ law and retrieved by centrifugation. This was done in an 319

attempt to facilitate the detection and identification of potential clay minerals, which are 320

concentrated in the finer fraction when present as individual particles. Also, smaller olivine 321

grains have higher surface-to-volume ratio, which makes secondary phases easier to detect 322

relatively to coarser fractions. 323

NIR spectra of our initial and weathered powders were acquired at the LPGN 324

laboratory (Nantes, France) using a Nicolet® 5700 Fourier transform infrared spectrometer 325

(FTIR) equipped with a tungsten-halogen white-light source, a CaF2 beam splitter and a 326

DTGS detector. For each sample, the FTIR chamber was first purged for four minutes with 327

dry and CO2-free air. Spectra presented in this paper are the average of 200 measurements in 328

the wavelength range of 1-5 µm (10000-2000 cm-1) with a resolution of 4 cm-1. Background 329

spectra were acquired using a Labsphere® Infragold reference, with a rough surface 330

optimized for powder analyses. Data acquisition and background correction were done using 331

the OMNIC® software. NIR spectra were acquired for both bulk samples and <2-µm 332

fractions. Unfortunately, the <2-µm fraction of the initial olivine was polluted during 333

preparation due to the failure of an ultrasonic agitator, making it unusable for NIR 334

spectroscopy. 335

Powder X-ray diffraction patterns were acquired at the IMN laboratory (Nantes, 336

France) using a Siemens® D5000 diffractometer equipped with a copper source (operated at a 337

voltage of 40 kV and a current of 40 mA), a monochromator and a Moxtek® PF2400 Si PIN 338

detector. For bulk powders, measurements were done between 2θ = 3.5° and 70° with steps of 339

0.016° and a counting time per step of 150 s. Oriented mounts were prepared from the <2-µm 340

fractions by wet smearing on glass slides in order to facilitate the detection of (001) 341

15

diffraction peaks of potential clay minerals. Measurements of the oriented mounts were done 342

between 2θ = 3.5° and 33° with steps of 0.016° and a counting time per step of 320 s. 343

Transmission electron microscopy (TEM) observations were performed at the IMN 344

laboratory using a Hitachi® HF 2000 equipped with a cold-field emission gun operated at an 345

acceleration voltage of 100 kV. TEM-EDX analyses were performed at the CINaM laboratory 346

(Marseille, France) using a JEOL® JEM 2010 equipped with a LaB6 electron gun (operated at 347

200 kV) and a Bruker® EDX detector. 348

57Fe Mössbauer spectra were acquired at the Mössbauer facility of the IMN laboratory, 349

which is equipped with a room-temperature 57Co(Rh) source. The spectrometer is operated in 350

transmission geometry and in constant acceleration mode, with a symmetric velocity ranging 351

between ±4.5 mm/s. Calibration of the velocity scale was done using absorption lines of iron 352

foil. After folding, the spectra (256 channels) were computed with a least-squares routine 353

using Lorentzian lines. 354

Thermogravimetric analyses (TGA) were performed at the CEA laboratory (Saclay, 355

France) using a Netzsch® STA 409 PC LUXX thermobalance operated at a heating rate of 356

10 °C/min from 25 to 980 °C, under a dinitrogen flow of 60 mL/min. Data were then 357

converted into differential thermogravimetric (DTG) curves. 358

Lastly, measures of total inorganic carbon (TIC) were performed at the IRSN 359

laboratory (Fontenay-aux-Roses, France) using an Elementar® Vario TOC Cube. This 360

instrument has a limit of detection of the order of ppb. 361

362

3. RESULTS 363

364

3.1. Chemistry of the solutions 365

366

16

3.1.1. pH evolution 367

368

Measurements show that the pH is more acidic in the two reactors containing a CO2 369

atmosphere than in those containing terrestrial air. In the “Mars reactors”, the pH reaches a 370

value of ~6 after 3 days and remains nearly unchanged afterwards, whereas in both “Earth 371

reactors” it reaches 8.4 at the end of the experiment (Fig. 3). The comparison of these 372

measures with theoretical initial values of 3.9 for a CO2 atmosphere (PCO2=1) and of 5.6 for a 373

terrestrial atmosphere (PCO2=4.10-4) indicates a strong buffering effect of olivine crystals, 374

especially fast in the “Mars reactors”. This suggests a rapid consumption of protons due to the 375

hydrolysis of olivine. 376

The pH values for the “CO2-H2O2” and “CO2” reactors are always very close to each 377

other (respectively 6.1 and 6.2 at the end of the experiment). The situation is a bit more 378

complex concerning the “Earth reactors”: the pH values at 14 and 31 days are slightly lower 379

in the “Air-H2O2” reactor than in the “Air” reactor. However, final values are found to be 380

identical (pH=8.4; Fig. 3). Thus, the pH appears unaffected by the presence of H2O2. 381

382

3.1.2. Dissolution of major elements 383

384

Mass concentrations over time of the three most abundant cations of the system (Si, 385

Mg, Fe) are shown in Fig. 3. The complete dataset (10 elements) with associated uncertainties 386

is available in the electronic annex (Table S1). 387

The evolutions of pH and major elements concentrations as a function of time and 388

atmospheric composition are well correlated: (1) as for pH, there is a rapid increase in 389

concentrations in the first days of the experiment followed by a relative stabilization 390

afterwards; (2) concentrations of dissolved elements are clearly much more important in the 391

17

“Mars reactors” than in the “Earth reactors” – acidic conditions under CO2 favor the 392

hydrolysis of olivine and thus the release of elements in the solution; (3) no systematic 393

influence of H2O2 is noticed (although some differences do exist for Mg, see below). 394

At the end of the experiment, Si is about 14 times more abundant in the “Mars 395

reactors” than in the “Earth reactors”. Beyond the values by themselves, the evolution over 396

time is also different between the two types of conditions: under CO2, the concentration is 397

nearly equal at 14 and 31 days, and has slightly decreased at 95 days; in contrast, under air, a 398

slow but continuous increase is observed. The effect of hydrogen peroxide on the dissolution 399

of Si appears negligible, since the two curves (with and without H2O2) are very close to each 400

other for the two types of conditions. 401

Mg is by far the most abundant element in the solutions. At the end of the experiment, 402

its mass concentration is 2.8, 4.1, 3.7 and 1.9 times higher than for Si in the “CO2-H2O2”, 403

“CO2”, “Air-H2O2” and “Air” reactors, respectively. Hence, the latter is the only one which 404

could tend toward a congruent dissolution (corresponding to a ratio of 1.6). However, ratios 405

for the sampling at 3, 14 and 31 days are all above 3. Also, contrary to Si, the curves appear 406

different with or without hydrogen peroxide: the Mg concentration decreases after 31 days in 407

the “CO2-H2O2” reactor, whereas a nearly linear increase is observed after 14 days in the 408

“CO2” reactor. Under terrestrial atmosphere, Mg concentrations increase continuously all 409

along the experiment (Fig. 3). Lastly, while the Mg concentration is higher with H2O2 for the 410

“Mars reactors”, the contrary is observed for the “Earth reactors”. 411

Iron solubility depends on its oxidation state: ferrous iron (Fe2+) is soluble whereas 412

ferric iron (Fe3+) is insoluble. Therefore, it is not surprising that dissolved Fe is solely 413

detected in the “CO2” reactor, which is the only one to have reducing conditions (no O2 or 414

H2O2). The decrease of concentration after 14 days suggests that some Fe2+ initially dissolved 415

has been subsequently incorporated in a Fe2+- or Fe3+-bearing solid phase. Indeed, even 416

18

without free O2 available, oxidation is possible through the dissociation of H2O molecules, as 417

shown by the detection of H2 in the GC analyses. 418

After Si, Mg and Fe, the most abundant elements are K, Ca and Na. The latter is the 419

only one to have a higher concentration under terrestrial atmosphere than under CO2 (Table 420

S1). Finally, the other analyzed elements (Al, P, Ti and Mn) present very low to non-421

measurable concentrations. 422

423

3.2. Weathered olivine powders 424

425

3.2.1. Near-infrared spectroscopy 426

427

The overall shape of the NIR spectra obtained from our weathered samples (Fig. 4) is 428

still dominated by the spectral signature of olivine (especially the broad, Fe2+-related band at 429

1.04 µm), which is not surprising given the relatively limited timeframe and low temperature 430

of the experiment. However, some significant differences can be observed between 1.8 and 431

2.5 µm (Fig. 4B). In this wavelength range, the four weathered samples show an obvious 432

absorption band at 1.91 µm, which was nearly-absent in the spectrum of the initial olivine. 433

This band is due to the formation of at least one alteration phase containing H2O molecules in 434

its structure (e.g., Hunt, 1977). Furthermore, another absorption band is visible at 2.31 µm in 435

the spectra of the “Air” and “Air-H2O2” samples, but also in the one of the “CO2” sample 436

(although more subtle here). This band indicates the formation of Mg-OH bonds in the altered 437

material (Clark et al., 1990). 438

These visual observations can be confirmed and refined by calculating the band depths 439

for several series of spectra (Fig. 5). Firstly, the band at 1.91 µm turns out to be deeper for the 440

powders weathered under CO2 than for those weathered under terrestrial air. Secondly, the 441

19

band depth at 2.31 µm is indeed greater for the “Air”, “Air-H2O2” and “CO2” samples 442

compared to the initial olivine. Thirdly, for a given type of atmosphere (CO2 or air), the band 443

depth – both at 1.91 and 2.31 µm – tends to be slightly lower for the samples weathered in the 444

presence of H2O2. 445

The combination of a H2O-related band at 1.9 µm with a metal-OH band around 2.2-446

2.3 µm is a typical spectral feature of hydrated phyllosilicates (Fig. 4D). The best match here 447

is obtained for a Mg-rich trioctahedral smectite, such as saponite, which is consistent with the 448

composition of our initial olivine. The small absorption band at 2.39 µm visible in the 449

reference spectrum of saponite is also found in the spectra of the “Air” and “Air-H2O2” 450

samples, confirming the good match. The combination of the 1.9-µm and 2.31-µm bands is 451

also observed in the “CO2” sample, but this time the correlation is not as good as previously 452

in terms of intensity, because the 1.9-µm band is deeper than for the “Air” and “Air-H2O2” 453

samples, whereas the 2.31-µm one is shallower. Therefore, it is probable that this portion of 454

the spectrum of the “CO2” sample corresponds to the superimposition of the signature of a 455

Mg-rich smectite with the one of a distinct Mg-poor hydrated phase. 456

Another wavelength range of interest is the one located between 3.8 and 4.2 µm, 457

because this is where the main absorption bands related to carbonate minerals should appear 458

(Fig. 4E and F). However, such carbonate-related bands are not obvious in the spectra of our 459

weathered samples (Fig. 4C). Indeed, the small 4.02-µm band found in the spectrum of the 460

initial olivine is still present without significant change, except perhaps a subtle enlargement 461

toward the short wavelengths for the “CO2” sample. This indicates that no (or nearly no) 462

carbonates were formed during the experiment, whatever the weathering conditions. 463

NIR spectra were also obtained from the <2-µm fraction of the weathered powders and 464

are reported in Fig. 6. These spectra are in good agreement with those of the bulk samples 465

(Fig. 4): the spectral signature of olivine appears preserved, while absorption bands related to 466

20

secondary phases are obvious at 1.91 and 2.31 µm. Moreover, the latter is the shallowest for 467

the “CO2-H2O2” sample, where it is probably inherited from the initial olivine. The intensity 468

of the 2.31-µm band then increases for the “CO2”, “Air-H2O2” and “Air” samples, 469

respectively. This confirms the trend shown in Fig. 5. Finally, a small absorption band located 470

at 1.39 µm is visible in the four spectra, while it was absent from Fig. 4. However, its 471

presence is not surprising, since it is classically associated to the one at 1.91 µm in hydrated 472

samples (Hunt, 1977). 473

474

3.2.2. X-ray diffraction 475

476

No clay signal appeared in the diffraction patterns of the weathered bulk samples (data 477

not shown), nor in those of the air-dried oriented <2-µm fractions (Fig. 7). This suggests that 478

the smectite phase detected in three samples by NIR spectroscopy has a low abundance and/or 479

a low crystallinity, both possibilities being consistent with the timeframe of the experiment. 480

481

3.2.3. Transmission electron microscopy 482

483

Grains from the initial olivine observed by TEM show clean and smooth surfaces, 484

which are the result of grinding (Fig. 8A). In contrast, and although some unmodified surfaces 485

persisted, the formation of two new phases directly on the olivine grains is clearly observed in 486

the weathered samples. The first type is characterized by its “cotton-like” texture (Fig. 8B) 487

and the second type by its “filamentous” texture (Fig. 8C). 488

The newly-formed phase with a cotton-like texture is observed only in the two 489

samples weathered under CO2. At low magnification, this phase appears as a semi-transparent 490

layer that surrounds a whole olivine grain, or even a group of several grains. Its thickness is 491

21

usually comprised between 50 and 100 nm, but can reach 400 nm in some cases. Its edge is 492

often darker, i.e., more opaque. At higher magnification, it turns out to be composed of small 493

rounded shapes difficult to separate visually, which gives the so-called cotton-like texture. 494

The newly-formed phase with a filamentous texture is abundant in the two samples 495

weathered under terrestrial air, where it appears on most olivine grains. It is also observed in 496

the “CO2” sample – although less frequently (<10% of the grains) –, but not in “CO2-H2O2”. 497

At low magnification, this phase gives a “hairy” aspect to the olivine grains. Its thickness is 498

never above a few tens of nanometers. At higher magnification, it becomes possible to 499

distinguish individual filaments, which are pointing more or less outward, depending on the 500

observed grain. 501

The good correlation between the band depth at 2.31 µm in the NIR spectra and the 502

abundance of the filamentous phase in TEM imagery suggests that this phase corresponds to 503

the smectite phase. As highlighted in Fig. 8D, this is further confirmed by a comparison with 504

phyllosilicates described in the literature, which have very similar morphologies (e.g., Smith 505

et al., 1987; Giorgetti et al., 2001; Jones and Brearley, 2006; Bishop et al., 2007; Lantenois et 506

al., 2008). Unfortunately, it was not possible to directly measure the distance between 507

smectite layers by high-resolution TEM because this phase was highly unstable under the 508

electron beam. 509

In order to precisely determine the chemistry of the newly-formed phases, EDX 510

measurements were achieved on the “CO2” and “Air” samples. The initial olivine was also 511

analyzed to obtain a reference composition. The resulting dataset is presented here in the form 512

of a ternary diagram Si-Mg-Fe (Fig. 9). The mean compositions calculated for each phase are 513

available in the electronic annex (Table S2). 514

The compositions measured by EDX for grains of the initial olivine are 515

homogeneously distributed around the bulk composition determined by ICP-OES, proving a 516

22

good agreement between the two methods (Fig. 9A and B). In contrast, the data points 517

corresponding to the newly-formed phases are not concentrated around a precise composition, 518

but rather distributed along mixing lines between the initial olivine and the secondary phase 519

itself (Fig. 9C, D and E). For the filamentous phase, this chemical mixing is explained by the 520

irregularity and low thickness of the weathered layer and by its direct proximity with the 521

unaltered core of the olivine grain: in these conditions, it is difficult to retrieve signal from the 522

secondary phase only. For the cotton-like phase, which is easier to analyze due to its greater 523

thickness, the mixing line may either reflect different stages in the alteration process or be due 524

to the presence of remaining olivine fragments within the weathered layer. 525

The filamentous phase is chemically closer to the initial olivine compared to the 526

cotton-like phase. Indeed, although both are characterized by a loss of Mg relative to Si, this 527

loss is much more pronounced for the cotton-like phase. 528

Compositions measured for the filamentous phase are compatible with a Mg/Fe-529

bearing smectite, intermediate between the saponite and nontronite endmembers. However, 530

there are some differences between the filamentous phase of the “Air” sample and the one of 531

the “CO2” sample. In the first case, the loss of Fe is important and so, the data points tend 532

toward a Mg-rich smectite close to the saponite endmember (Fig. 9A and C). In the second 533

case, the loss of Fe is less pronounced and some points even indicate an enrichment in Fe 534

(Fig. 9A and D). Hence, the corresponding smectite would be more Fe-rich than in the “Air” 535

sample, although still closer to saponite than nontronite. These observations are confirmed by 536

the mean compositions (Table S2): the filamentous phase of “Air” sample lost 21% of Mg and 537

24% of Fe relative to the initial olivine, while its counterpart in the “CO2” sample lost 42% 538

and 7%, respectively. 539

The data points measured for the cotton-like phase clearly tend toward the Si apex of 540

the ternary diagram (Fig. 9A and E). This reflects a major loss of both Mg and Fe relative to 541

23

the initial olivine (76% and 60%, respectively; Table S2). The most likely single phase with 542

such Si-rich composition – Si/(Si+Mg) > 0.8 – is silica (SiO2·nH2O) because most 543

phyllosilicate minerals have Si/(Si+Mg) < 0.7. Furthermore, the cotton-like phase is 544

composed of rounded shapes similar to the spheres observed in natural opals, although the 545

latter are usually slightly bigger (~150-350 nm; e.g., Jones and Segnit, 1969; Rondeau et al., 546

2004). 547

Table S2 also shows the behavior of some minor elements. For example, Na, Cl and K 548

are always less abundant in the secondary phases compared to the initial olivine, which is 549

consistent with their high solubility. On the contrary, the Al content is always higher in the 550

secondary phases, which reflects its low solubility. 551

552

3.2.4. Mössbauer spectroscopy 553

554

The four types of atmospheres used in our experiment cover a large span of redox 555

states, from reducing conditions (CO2 atmosphere) to moderately oxidizing (terrestrial 556

atmosphere) to highly oxidizing (H2O2). Analyses of the solutions have shown that, as 557

expected, only the reducing conditions allowed ferrous iron to stay dissolved. In contrast, in 558

the three other conditions, any Fe2+ extracted from olivine must have been converted to Fe3+ 559

and incorporated to (oxy)hydroxides or other Fe3+-bearing phase. We acquired Mössbauer 560

spectra of our samples (Fig. S3; included in the electronic annex) in an attempt to identify and 561

quantify such ferric phases (e.g., Schröder et al., 2004; and references therein). However, in 562

all the samples, we obtained only the signature of olivine, through the absorptions of Fe2+. 563

The absence of Fe3+ detection gives an upper limit of the order of a few percent. This shows 564

that the addition of hydrogen peroxide has not increase the oxidation of Fe2+ in measurable 565

proportions compared to the other environments tested here. 566

24

567

3.2.5. Thermogravimetry and total inorganic carbon 568

569

A primary mineral like olivine contains negligible volatiles: as a consequence, no 570

mass loss is expected in a thermogravimetric analysis, and this is verified here (Fig. 10). In 571

contrast, during weathering, secondary phases incorporate volatiles, including water (in the 572

form of H2O molecules and/or hydroxyl groups OH) in the case of smectites or silica and CO2 573

in the case of carbonates. 574

Differential thermogravimetric (DTG) curves obtained for the “CO2-H2O2” and “CO2” 575

samples are similar (Fig. 10): they show mass losses centered at ~100 °C and ~410 °C. In 576

addition, the “CO2” sample presents two minor mass losses at ~550 °C and ~630 °C. In 577

comparison, the curve of the “Air” sample is closer to the one of the initial olivine: its shows a 578

unique mass loss centered at ~80 °C. Note here that the “Air-H2O2” sample has not been 579

analyzed due to its high similarity with the “Air” sample, as revealed by previous results. 580

Mass losses located at 100 °C or less are attributable to the release of weakly-bound 581

water (e.g., Ek et al., 2001). Having in mind the NIR and TEM results, this water probably 582

comes essentially from hydrated silica for the samples weathered under CO2 and from the 583

smectite phase for the “Air” sample. A small contribution from the smectite phase is also 584

expected for the “CO2” sample. 585

Mass losses located at 410 °C for the “CO2-H2O2” and “CO2” samples are attributable 586

to the release of hydroxyl groups. Several alteration phases could have produced them. For 587

example, brucite Mg(OH)2 undergoes dehydroxylation at this temperature (Kotra et al., 1982), 588

but this hypothesis can be ruled out because it is not consistent with the NIR spectra (Fig. 4 589

and 6): if present, brucite would produce a strong absorption band at 1.4 µm. The thermal 590

decomposition of Fe-hydroxides is more plausible. Indeed, although goethite and 591

25

lepidocrocite undergo dehydroxylation at less than 300 °C, it has been shown that the 592

decomposition temperature is above 400 °C for hydrated forms close to ferrihydrite (Prasad 593

and Sitakara Rao, 1984; Mitov et al., 2002). However, no Fe-(oxy)hydroxides has been 594

formally identified by NIR, XRD or TEM. A third possibility is the dehydroxylation of 595

hydrated silica, which occurs at higher temperature than its dehydration (Ek et al., 2001). This 596

last hypothesis would be consistent with the detection of an abundant Si-rich phase by TEM 597

and with the fact that the mass losses at 100 °C and 410 °C appear to be correlated in the 598

curves of the two samples containing this phase. 599

Finally, minor mass losses at 550 °C and 630 °C observed only for the “CO2” sample 600

can be attributed to the release of CO2, which implies the presence of minor amounts of 601

carbonates in this powder (e.g., Kotra et al., 1982). This detection is coherent with the very 602

small amount of TIC measured in this sample (~0.09 mg/g), in contrast with null values for 603

the initial olivine and the other weathered samples. Since two distinct gas releases are 604

observed, they must correspond to two distinct mineral phases, for example siderite and 605

magnesite, respectively (Kotra et al., 1982). Because the direct precipitation of magnesite in 606

our experimental conditions is unlikely (Hänchen et al., 2008), hydrated Mg-carbonates such 607

as nesquehonite or hydromagnesite may have formed instead and then been converted to 608

magnesite during the thermogravimetric analysis. 609

Combining the above qualitative interpretations with the quantitative mass losses 610

observed – in addition to results from other analytical methods – allows to estimate directly or 611

indirectly the abundances of the secondary phases, as summarized in Table 2. 612

613

4. DISCUSSION 614

615

4.1. Nature and abundance of secondary phases 616

26

617

4.1.1. Smectite phase 618

619

The formation of a Mg-rich clay in the samples weathered under terrestrial air and in 620

lower amount in the “CO2” sample is attested by NIR spectra, TEM imagery and EDX 621

analyses (Table 2). Although definitive identification of the exact species by XRD was not 622

possible, the NIR signature is best matched by a Mg-rich member of the smectite group, 623

probably close to saponite (Fig. 4 and 6). On the contrary, the NIR spectra of our samples lack 624

features typical of some other Mg-bearing clays, such as talc (sharp and deep band at 1.39 625

µm), montmorillonite (2.21- or 2.29-µm band), palygorskite (2.2-µm band) and serpentine 626

(2.1-µm band). Serpentine can be further discounted because it does not match the chemical 627

compositions retrieved by EDX (Fig. 9). 628

We proposed in section 3.2.2 that the non-detection of the smectite phase by XRD – 629

even using oriented <2-µm mounts – could be due a low abundance and/or a low crystallinity 630

of the smectite. Consistently, TEM observations have shown that the filamentous phase was 631

really thin (several tens of nanometers at most). In addition, filaments were never seen 632

isolated from the olivine grains; therefore, they may have been unable to deposit flat, which 633

would have in turn limited the intensity of the (001) reflections of clay particles in the XRD 634

analyses. Finally, the hypothesis of a low crystallinity remains plausible, given the limited 635

timeframe of the experiment. As a consequence, the filamentous phase may not be a mature 636

smectite, but rather an early stage of crystallization or “precursor” (e.g., Santiago Buey et al., 637

2000). 638

Finally, a possibly important result of our study is that the smectite phase has been 639

observed in the “CO2” sample but not in the “CO2-H2O2” sample. This suggests that hydrogen 640

27

peroxide had an inhibiting effect on this secondary phase. This point will be further discussed 641

in section 4.2.2. 642

643

4.1.2. Si-rich phase 644

645

TEM observations of the “CO2” and “CO2-H2O2” samples have revealed a newly-646

formed phase with a so-called cotton-like texture, which is absent from the powders 647

weathered under terrestrial atmosphere. EDX analyses have shown that this phase clearly 648

tends toward a composition of silica (SiO2·nH2O). The formation of such Si-rich phase is 649

consistent with the observed evolution of Si concentrations in solution. Indeed, under CO2, Si 650

concentrations stabilized after 14 or 31 days, what may reflect equilibrium with a solid phase 651

(Fig. 3). Because of the low abundance of the smectite phase in “CO2” and its absence in 652

“CO2-H2O2”, it cannot explain the intensity of the hydration band at 1.9 µm in the NIR 653

spectra of these two samples (Fig. 5), nor the H2O-related mass losses in their 654

thermogravimetric analyses (Fig. 10). Therefore, hydration is most probably borne by the Si-655

rich phase, consistently with the formula of SiO2·nH2O. 656

From the Si-Mg-Fe ternary diagram (Fig. 9), the filamentous (smectite) and cotton-657

like (Si-rich) phases of the “CO2” sample seem to follow a same trend between the initial 658

olivine and the Si apex. This could suggest that they represent two different stages of a unique 659

alteration process of olivine. Interestingly, King et al. (2010) have reported the transformation 660

of a Si-rich phase into lizardite during the alteration of olivine under CO2 at 200 °C. A similar 661

mechanism may have occurred in our experiment, but the trend in Fig. 9 may also be 662

consistent with the reverse, i.e. the transformation of the smectite phase into the Si-rich phase. 663

In this case, the coexistence of the two phases would be the result of differential weathering 664

(in intensity or in time) within the powder. However, it should be noted that no TEM 665

28

observation (for example, of a potential intermediate morphology) supports either of these 666

hypotheses. Thus, alternatively, the smectite phase and the Si-rich phase could represent two 667

different reaction pathways from the initial olivine. In all cases, it is clear that the powder 668

grains have not all undergone the same alteration, or at least not with the same intensity. This 669

is further confirmed by the observation by TEM of grains with clean surfaces, obviously 670

intact despite the three months in the water bath (Fig. 8). This differential weathering suggests 671

that very low-scale (micrometers, or even nanometers) processes may have a strong influence, 672

and that interfacial conditions (pH, dissolved elements, L/R ratio) are not necessarily 673

representative of the bulk solution (e.g., King et al., 2010; Daval et al., 2011). 674

675

4.1.3. Fe-(oxy)hydroxides 676

677

No Fe-(oxy)hydroxide has been formally identified in our weathered samples by any 678

of the analytical methods employed and no reddening of the weathered powders has been 679

observed. This implies that such phase, if formed, must be limited to very minor amounts. 680

However, EDX data of the secondary phases (Fig. 9 and Table S2) provides some information 681

above the behavior of Fe during experimental weathering. For example, the Si-rich phase 682

formed in the “Mars reactors” has lost a significant amount of Fe compared to the olivine 683

from which it derives (Table S2). This means that the “missing” Fe must have been dissolved 684

or incorporated in a Fe-rich phase. In the “CO2-H2O2” reactor, Fe2+ released from olivine is 685

expected to be immediately oxidized and precipitated as Fe-(oxy)hydroxides, which is 686

consistent with its non-detection in the solution (Fig. 3). In the “CO2” reactor, reducing 687

conditions has allowed Fe2+ to stay in solution. However, the progressive decrease of Fe 688

concentration after 14 days (Fig. 3) along with the detection of H2 in the headspace suggest 689

that partial oxidation occurred through the dissociation of H2O molecules, possibly forming 690

29

some minor Fe-(oxy)hydroxides (Table 2). Alternatively, Fe may have been incorporated in 691

other Fe2+ or Fe3+-bearing phases, such as the smectite phase or siderite. 692

693

4.1.4. Carbonates 694

695

No carbonate has been formally identified in our samples by NIR spectroscopy, 696

despite the method being very sensitive to this group of minerals (e.g., Gaffey, 1986; Bishop 697

et al., 2001). Only a subtle enlargement toward the short wavelengths of the pre-existing 4.02-698

µm band was observed in the spectrum of the “CO2” sample (Fig. 4), suggesting the very 699

minor appearance of a carbonate mineral. Then, thermogravimetric analyses have revealed 700

traces of CO2 trapped in the same sample, which was confirmed by the TIC measurement. 701

The formation of siderite is possible, given the absence of O2 in this reactor. In any case, the 702

abundance involved must be very low (Table 2). 703

704

4.2. Effects of environmental parameters and implications for early Mars 705

706

Due to unavoidable differences in extrinsic (scale, timeframe) and intrinsic factors 707

(grain size, composition), laboratory experiments never exactly mimic natural processes (e.g., 708

Casey et al., 1993). For example, pedogenic processes – i.e., vertical mobility of dissolved 709

elements and pH variations within weathering profiles – are not reproduced in our 710

experimental setting, although they have a primary influence on the nature and abundance of 711

secondary phases (e.g., Gaudin et al., 2011). Nevertheless, our results allow us to isolate the 712

effects of several key environmental parameters on weathering processes, which are reviewed 713

below. 714

715

30

4.2.1. Effects of CO2 716

717

Our results emphasize the important effects of a CO2 atmosphere and associated low 718

pH on the weathering products of forsteritic olivine. Under terrestrial atmosphere, the main 719

(and only) secondary phase found is a Mg-rich smectite. In contrast, under CO2, the main 720

secondary phase found is a hydrated, Si-rich phase. However, an important result of the study 721

is that a CO2 atmosphere does not prevent the formation of a Mg-rich smectite from olivine. 722

The reasons for which the smectite phase is in lower abundance under CO2 compared to 723

terrestrial conditions (Table 2) are discussed below and in the next section. 724

The development of a Si-rich layer has been reported in numerous experiments of 725

olivine carbonation, even when the bulk solution was undersaturated with respect to 726

amorphous silica (e.g., Pokrovsky and Schott, 2000; Bearat et al., 2006; Garcia et al., 2010; 727

King et al., 2010; Daval et al., 2011). Such Si-rich phase is often described as a “passivating 728

layer”, which tends to decrease the dissolution rate of olivine and, in some cases, prevents the 729

formation of any other secondary phase (Giammar et al., 2005; Daval et al., 2011). This effect 730

would have been especially effective here, because the discontinuous, manual stirring of our 731

reactors is not expected to have been vigorous enough to remove such a passivating layer 732

from the grains during the course of the experiment. Therefore, it is possible that the 733

development of the Si-rich phase in the “Mars samples” have contributed – in addition to the 734

role of Fe discussed in the next section – to decrease the abundance of the smectite phase 735

compared to the “Earth samples”. 736

This same passivating effect of the Si-rich phase may also have limited the amount of 737

carbonates formed in our experiment. Indeed, the very low amount of carbonates is a major 738

discrepancy with other studies ran at high pressures of CO2 and/or higher temperatures (e.g., 739

Pokrovsky and Schott, 2000; Gerdemann et al., 2007; Garcia et al., 2010; King et al., 2010), 740

31

with the notable exception of the work of Giammar et al. (2005) and Daval et al. (2011) 741

mentioned previously. Interestingly, this low abundance of carbonates is consistent with the 742

overall secondary mineralogy of the martian surface, which is dominated by Fe/Mg smectites 743

and shows only rare and local occurrences of carbonates (Ehlmann et al., 2008b; Carter et al., 744

2013). 745

Taken together, the present results and previous studies of olivine carbonation are 746

consistent with a high production of silica through weathering of olivine-bearing primary 747

rocks on early Mars. This could possibly explain the high-silica rock compositions measured 748

within the northern plains (e.g., McLennan, 2003), as well as the recent detection of opaline 749

silica in numerous alluvial and delta fans (Carter et al., 2012). However, it is worth noting that 750

the Si-rich phase formed in our experiment has no diagnostic signature in the NIR domain 751

(Fig. 4), which implies that similar materials could be difficult to identify by orbital 752

spectroscopy. 753

Finally, a dense CO2 atmosphere implies more acidic meteoric water than on present-754

day Earth. As highlighted in section 3.1 and previous studies (e.g., Olsen and Rimstidt, 2007; 755

and references therein), this greatly enhances the dissolution rate of olivine. As a 756

consequence, a CO2 atmosphere would favor the leaching of soluble elements – Mg in 757

particular, which is ~30 times more abundant in the “CO2” solution than in the “Air” solution 758

at the end of the experiment (Fig. 3). Such an intense leaching would accelerate the 759

development of weathering profiles and facilitate the appearance of an upper Al-rich, 760

kaolinite-bearing layer from an Al-poor bedrock, without requiring (highly unlikely) tropical 761

conditions as on Earth (Gaudin et al., 2011). Moreover, a CO2 atmosphere – with low 762

production or shallow extinction depth of H2O2 (Zent, 1998) – would maintain reducing 763

conditions within these weathering profiles and thus increase the mobility of Fe (e.g., 764

Murakami et al., 2004). In such conditions, the upper zone of putative martian weathering 765

32

profiles would be less rich in Fe than their terrestrial counterparts (Gaudin et al., 2011; 766

Greenberger et al., 2012) and Al would be the only major element to remain “immobile”, 767

along with some Si to produce kaolinite. Therefore, the low Fe-oxide content generally 768

observed by remote instruments in Al-rich units on Mars does not ruled out pedogenesis as a 769

formation process for Al-rich clays over Fe/Mg-smectites superimpositions. 770

771

4.2.2. Effects of H2O2 772

773

Hydrogen peroxide was used in this experiment to represent the highly oxidizing 774

compounds formed by photochemical reactions in the Mars atmosphere (Clancy et al., 2004; 775

Encrenaz et al., 2004) and detected in the regolith by the Viking landers (Bullock et al., 1994; 776

Zent, 1998), which may have been more abundant on a “wetter” early Mars. Its effect on the 777

dissolution of olivine turns out to be weak in the timeframe of the experiment, as highlighted 778

by the ICP-OES analyses (Fig. 3). Similarly, Mössbauer spectra (Fig. S3) indicate that H2O2 779

has not enhanced the oxidation of our Mg-rich olivine in a measurable way. 780

On the other hand, we have noticed an important effect of this molecule regarding the 781

smectite phase, which formed in the “CO2” reactor but not in the “CO2-H2O2” one. Thus, 782

H2O2 seems to inhibit the formation of smectite under CO2 conditions. The only difference 783

between the two “Mars reactors” is the redox state of the system. Therefore, a hypothesis to 784

consider is that the inhibiting effect is related to the rapid oxidation of Fe2+ deriving from the 785

dissolution of olivine (as shown by Fig. 3, Fe2+ is present in the solution of the “CO2” reactor 786

but not in the one of the “CO2-H2O2” reactor, which suggests the precipitation of Fe-787

(oxy)hydroxides in the latter). Previous studies have shown that Fe2+ plays a strong role in the 788

formation of smectite at the low pH values associated with CO2 conditions (Murakami et al., 789

2004; Tosca et al., 2008). At pH ~4.6, Murakami et al. (2004) observed that Fe2+-rich 790

33

vermiculite or smectite precipitated at the edge of their Fe-bearing biotite under anoxic 791

conditions, while only Fe3+- and Al-(hydr)oxides were observed under oxic conditions. 792

Consistently, Tosca et al. (2008) reported that in anoxic conditions, Fe2+-bearing saponite 793

precursor can precipitate at pH as low as 5, whereas pure Mg-saponite form only at pH 9. 794

Moreover, this result is consistent with NIR spectra obtained by Dehouck et al. (2012) in their 795

weathering experiment conducted at low temperature and low L/R ratio. Indeed, two of their 796

silicate samples (olivine Ol1 and orthopyroxene OPx) weathered under CO2 showed a small 797

Mg-OH absorption band at 2.31 µm – attributable to a Mg-rich smectite phase, although not 798

confirmed due to the lack of TEM data – while the same samples weathered under CO2+H2O2 799

did not. 800

These observations imply that Fe2+ is required to ensure the stability of the smectite 801

phase under CO2 conditions (pH ~6 in our experiment) and suggest that H2O2 has prevented 802

its incorporation into this phase in the case of the “CO2-H2O2” reactor. Since our starting 803

olivine was relatively Fe-poor, this also explains the low abundance of the smectite phase in 804

the “CO2” sample (compared to “Air” and “Air-H2O2”), although the Si-rich phase may also 805

play a role here as discussed previously. This is also consistent with the EDX analyses, which 806

show that the filamentous phase of the “CO2” sample is richer in Fe than its counterpart 807

formed in the “Earth reactors” (Fig. 9). Hence, CO2 conditions seem to favor the formation of 808

Fe-bearing smectites rather than purely Mg-bearing ones. Therefore, Fe-rich starting material 809

(in laboratory experiments) or bedrock (in natural weathering) would probably tend to 810

produce proportionally more smectites than in this study. 811

Under terrestrial air, the role of Fe is less important than under CO2 because the pH is 812

higher (Fig. 3). Thus Fe is not required to ensure the stability of the smectite phase: as a 813

result, the smectite phase is more abundant and the effect of hydrogen peroxide is less 814

pronounced. However, a small inhibiting effect of H2O2 in terrestrial conditions cannot be 815

34

completely discounted since the 2.31-µm absorption band attributed to the smectite phase is 816

slightly shallower for the “Air-H2O2” sample than for the “Air” sample (Fig. 5 and 6). 817

Taken together, these results suggest that if the highly oxidizing compounds found on 818

present-day Mars were more abundant 3 to 4 billion years ago, they would have been an 819

obstacle for the formation of abundant clay minerals. However, images of the first drilling 820

operations performed by the Curiosity rover at Gale crater have shown that the smectite-821

bearing sediments of Yellowknife Bay are gray-colored at <2 cm depth (as shown by the 822

“mini-drill” of sol 180), contrasting with the reddish surface and suggesting only very shallow 823

oxidation at this place (Grotzinger et al., 2014; Vaniman et al., 2014). In addition, the 824

detection of reduced sulfur in these sediments is another indication of limited oxidation 825

(Grotzinger et al., 2014). The presence of abundant clays implies either that the production of 826

highly oxidizing compounds was limited on early Mars or that their inhibiting effect on the 827

formation of clays was counterbalanced by other mechanisms (e.g., Fe-rich source rock and/or 828

higher pH; Vaniman et al., 2014). Nevertheless, the role of H2O2 may have increase during 829

the subsequent eras (Dehouck et al., 2012) and ultimately become dominant (Huguenin, 1982; 830

Bibring et al., 2006) due to the progressive disappearance of other forms of weathering (i.e., 831

those involving liquid water). In addition, long term exposure of initial Fe2+-bearing smectites 832

to H2O2 may have cause their gradual transformation into Fe3+-bearing smectites (Beehr and 833

Catalano, 2012). 834

835

4.2.3. Effects of L/R ratio 836

837

In a previous weathering experiment, Dehouck et al. (2012) used several silicate 838

samples, including olivine with a composition similar to the one used here (Fo90Fa10). Thus, it 839

is possible to directly compare the results of these two experiments, which differ mainly by 840

35

their L/R ratio, low in Dehouck et al. (2012) (thin films of water condensed on the olivine 841

grains) while relatively high in the present study (30 g of olivine powder in 300 mL of water). 842

This gives the opportunity to evaluate the effect of the L/R ratio on the weathering of olivine 843

under CO2 conditions. 844

The only secondary phase formed from olivine formally identified in Dehouck et al. 845

(2012) was nesquehonite, a hydrated Mg-carbonate, along with possible Mg-smectite phase in 846

the samples weathered without H2O2, as mentioned above. In contrast, in the present study, 847

silica (along with smectite phase in the case of the “CO2” sample) dominates over carbonates. 848

Therefore, it appears that the L/R ratio has a major effect on the nature of the secondary 849

phases formed from olivine under CO2, with low ratios favoring carbonates and higher ratios 850

favoring silica. This is broadly consistent with results obtained at 150 °C and 150 bar by 851

Garcia et al. (2010), which obtained more carbonates for a L/R ratio of 0.1 than for a ratio of 852

10. Moreover, Bearat et al. (2006) noted that, for stirred experiments, lower L/R ratios can 853

lead to higher particle-particle abrasion, which in turn contribute to remove the Si-rich layer 854

and thus increase the formation of carbonates. 855

856

5. CONCLUSION 857

858

The laboratory experiment described in this paper has allowed us to compare the 859

weathering of olivine in simulated early martian conditions and terrestrial conditions, and to 860

evaluate the effect of hydrogen peroxide on this process. Our results show that the type of 861

atmosphere clearly has a primary influence on weathering pathways. The main (and only) 862

secondary phase formed under terrestrial air is a Mg-rich smectite phase. Under CO2, a 863

smectite phase appeared as well, but only in the absence of H2O2. Therefore, a pure CO2 864

atmosphere does not prevent the formation of smectites from olivine, although the addition of 865

36

highly oxidizing compounds could inhibit the process. However, the main secondary phase 866

formed under CO2 turns out to be a silica-rich phase with a non-diagnostic near-infrared 867

spectral signature. Finally, carbonate minerals are absent or nearly absent in all of our final 868

samples. 869

Several regions of Mars, such as Nili Fossae, are known to host olivine-rich bedrock 870

(Hoefen et al., 2003; Mustard et al., 2005; Poulet et al., 2007; Ody et al., 2013) and to have 871

undergone rock-water-atmosphere interactions (Ehlmann et al., 2008a; Gaudin et al., 2011). 872

Fe/Mg-smectites are the most common secondary phases found in these regions (e.g., 873

Mangold et al., 2007; Ehlmann et al., 2011). In Gale crater, the Curiosity rover has recently 874

found evidence for low grade diagenesis of lacustrine sediments, which resulted in the 875

formation of Mg-bearing smectite clays at the expense of olivine (Grotzinger et al., 2014; 876

McLennan et al., 2014; Vaniman et al., 2014). In these two contexts, our experimental results 877

are helpful to understand the weathering processes that occurred. They show that Fe/Mg-878

smectite clays can be formed at the low pH associated with a dense CO2 atmosphere and 879

highlight the critical importance of the redox conditions, the Fe-content of the source rock and 880

the formation of a passivating layer around the altered grains. 881

882

Acknowledgments. The authors thank two anonymous reviewers and the associate 883

editor Penelope King for their comments and suggestions which greatly improved the quality 884

of the manuscript. The authors alsothank all the people who brought them their scientific or 885

technical expertise and without whom this study would not have been possible: Hervé Loyen, 886

Laurent Lenta and Carole La from the LPGN laboratory (Nantes, France); Pierre-Emmanuel 887

Petit, Philippe Leone, Michel Suchaud, Nicolas Stéphant and Stéphane Grolleau from the 888

IMN laboratory (Nantes, France); Guillaume Blain, Francis Crumière and Massoud Fattahi-889

Vanani from the Subatech laboratory (École des Mines, Nantes, France); Damien Chaudanson 890

37

from the CINaM laboratory (Marseille, France); and Patrick Le Bescop from the CEA 891

laboratory (Saclay, France). Thanks to Steven Jaret for proofreading the manuscript. This 892

work was supported by the Centre National de la Recherche Scientifique (CNRS) through its 893

EPOV (Environnements Planétaires et Origines de la Vie) and PNP (Programme National de 894

Planétologie) programs, and by the Centre National d’Études Spatiales (CNES). 895

896

REFERENCES 897

898

Amram, K., Ganor, J., 2005. The combined effect of pH and temperature on smectite 899

dissolution rate under acidic conditions. Geochimica et Cosmochimica Acta 69, 2535-900

2546. 901

Ansan, V., Loizeau, D., Mangold, N., Le Mouélic, S., Carter, J., Poulet, F., Dromart, G., 902

Lucas, A., Bibring, J. P., Gendrin, A., Gondet, B., Langevin, Y., Masson, P., Murchie, 903

S., Mustard, J. F., and Neukum, G., 2011. Stratigraphy, mineralogy, and origin of 904

layered deposits inside Terby crater, Mars. Icarus 211, 273-304. 905

Bauer, A., Schafer, T., Dohrmann, R., Hoffmann, H., Kim, J.I., 2001. Smectite stability in 906

acid salt solutions and the fate of Eu, Th and U in solution. Clay Minerals 36, 93-103. 907

Bearat, H., McKelvy, M. J., Chizmeshya, A. V. G., Gormley, D., Nunez, R., Carpenter, R. 908

W., Squires, K., and Wolf, G. H., 2006. Carbon sequestration via aqueous olivine 909

mineral carbonation: Role of passivating layer formation. Environmental Science & 910

Technology 40, 4802-4808. 911

Beehr, A. R. and Catalano, J. G., 2012. Oxidation pathways of ferrous iron phyllosilicates: 912

insights into Martian phyllosilicate formation. Third Conference on Early Mars: 913

Geologic, Hydrologic and Climatic Evolution, abstract #7008. 914

38

Bibring, J. P., Langevin, Y., Gendrin, A., Gondet, B., Poulet, F., Berthe, M., Soufflot, A., 915

Arvidson, R., Mangold, N., Mustard, J., and Drossart, P., 2005. Mars surface diversity 916

as revealed by the OMEGA/Mars Express observations. Science 307, 1576-1581. 917

Bibring, J. P., Langevin, Y., Mustard, J. F., Poulet, F., Arvidson, R., Gendrin, A., Gondet, B., 918

Mangold, N., Pinet, P., and Forget, F., 2006. Global mineralogical and aqueous Mars 919

history derived from OMEGA/Mars express data. Science 312, 400-404. 920

Bishop, J. L., Lougear, A., Newton, J., Doran, P. T., Froeschl, H., Trautwein, A. X., Korner, 921

W., and Koeberl, C., 2001. Mineralogical and geochemical analyses of Antarctic lake 922

sediments: A study of reflectance and Mossbauer spectroscopy and C, N, and S 923

isotopes with applications for remote sensing on Mars. Geochimica et Cosmochimica 924

Acta 65, 2875-2897. 925

Bishop, J. L., Schiffman, P., Murad, E., Dyar, M. D., Drief, A., and Lane, M. D., 2007. 926

Characterization of alteration products in tephra from Haleakala, Maui: A visible-927

infrared spectroscopy, Mossbauer spectroscopy, XRD, EMPA and TEM study. Clays 928

and Clay Minerals 55, 1-17. 929

Bishop, J. L., Tirsch, D., Tornabene, L. L., Jaumann, R., McEwen, A. S., McGuire, P. C., 930

Ody, A., Poulet, F., Clark, R. N., Parente, M., McKeown, N. K., Mustard, J. F., 931

Murchie, S. L., Voigt, J., Aydin, Z., Bamberg, M., Petau, A., Michael, G., Seelos, F. 932

P., Hash, C. D., Swayze, G. A., and Neukum, G., 2013. Mineralogy and morphology 933

of geologic units at Libya Montes, Mars: Ancient aqueously derived outcrops, mafic 934

flows, fluvial features, and impacts. Journal of Geophysical Research: Planets 118, 935

487-513. 936

Bullock, M. A. and Moore, J. M., 2007. Atmospheric conditions on early Mars and the 937

missing layered carbonates. Geophysical Research Letters 34. 938

39

Bullock, M. A., Stoker, C. R., McKay, C. P., and Zent, A. P., 1994. A coupled soil-939

atmosphere model of H2O2 on Mars. Icarus 107, 142-154. 940

Carter, J., Poulet, F., Bibring, J. P., Mangold, N., and Murchie, S., 2013. Hydrous minerals on 941

Mars as seen by the CRISM and OMEGA imaging spectrometers: Updated global 942

view. Journal of Geophysical Research: Planets, 1-29. 943

Carter, J., Poulet, F., Mangold, N., Ansan, V., Dehouck, E., Bibring, J.-P., and Murchie, S., 944

2012. Composition of alluvial fans and deltas on Mars. Lunar and Planetary Science 945

Conference XLIII, abstract #1978. 946

Casey, W. H., Banfield, J. F., Westrich, H. R., and McLaughlin, L., 1993. What do 947

dissolution experiments tell us about natural weathering? Chemical Geology 105, 1-948

15. 949

Chevrier, V., Mathe, P. E., Rochette, P., Grauby, O., Bourrie, G., and Trolard, F., 2006. Iron 950

weathering products in a CO2+(H2O or H2O2) atmosphere: Implications for weathering 951

processes on the surface of Mars. Geochimica et Cosmochimica Acta 70, 4295-4317. 952

Chevrier, V. F., Dehouck, E., Lozano, C. G., Altheide, T. S., 2012. Mineral parageneses 953

resulting from weathering on Early Mars and the effect of CO2 vs SO2 atmospheres. 954

Third Conference on Early Mars: Geologic, Hydrologic and Climatic Evolution, 955

abstract #7080. 956

Clancy, R. T., Sandor, B. J., and Moriarty-Schieven, G. H., 2004. A measurement of the 362 957

GHz absorption line of Mars atmospheric H2O2. Icarus 168, 116-121. 958

Clark, R. N., King, T. V. V., Klejwa, M., Swayze, G. A., and Vergo, N., 1990. High spectral 959

resolution reflectance spectroscopy of minerals. Journal of Geophysical Research-960

Solid Earth and Planets 95, 12653-12680. 961

40

Clark, R. N., G. A. Swayze, R. Wise, E. Livo, T. Hoefen, R. Kokaly, and S. J. Sutley (2007), 962

USGS digital spectral library splib06a, U.S. Geological Survey Digital Data Series 963

231. Available from: http://speclab.cr.usgs.gov/spectral.lib06 964

Daval, D., Sissmann, O., Menguy, N., Saldi, G. D., Guyot, F., Martinez, I., Corvisier, J., 965

Garcia, B., Machouk, I., Knauss, K. G., and Hellmann, R., 2011. Influence of 966

amorphous silica layer formation on the dissolution rate of olivine at 90 degrees C and 967

elevated pCO2. Chemical Geology 284, 193-209. 968

Dehouck, E., Chevrier, V., Gaudin, A., Mangold, N., Mathe, P. E., and Rochette, P., 2012. 969

Evaluating the role of sulfide-weathering in the formation of sulfates or carbonates on 970

Mars. Geochimica et Cosmochimica Acta 90, 47-63. 971

Dehouck, E., Mangold, N., Le Mouélic, S., Ansan, V., and Poulet, F., 2010. Ismenius Cavus, 972

Mars: A deep paleolake with phyllosilicate deposits. Planetary and Space Science 58, 973

941-946. 974

Delvigne, J., Bisdom, E. B. A., Sleeman, J., and Stoops, G., 1979. Olivines, their 975

pseudomorphs and secondary products. Pédologie 3, 247-309. 976

Ehlmann, B. L., Mustard, J. F., Murchie, S. L., Bibring, J. P., Meunier, A., Fraeman, A. A., 977

and Langevin, Y., 2011. Subsurface water and clay mineral formation during the early 978

history of Mars. Nature 479, 53-60. 979

Ehlmann, B. L., Mustard, J. F., Fassett, C. I., Schon, S. C., Head, J. W., Marais, D. J. D., 980

Grant, J. A., and Murchie, S. L., 2008a. Clay minerals in delta deposits and organic 981

preservation potential on Mars. Nature Geoscience 1, 355-358. 982

Ehlmann, B. L., Mustard, J. F., Murchie, S. L., Poulet, F., Bishop, J. L., Brown, A. J., Calvin, 983

W. M., Clark, R. N., Des Marais, D. J., Milliken, R. E., Roach, L. H., Roush, T. L., 984

Swayze, G. A., and Wray, J. J., 2008b. Orbital Identification of Carbonate-Bearing 985

Rocks on Mars. Science 322, 1828-1832. 986

41

Ehlmann, B. L., Mustard, J. F., Swayze, G. A., Clark, R. N., Bishop, J. L., Poulet, F., Marais, 987

D. J. D., Roach, L. H., Milliken, R. E., Wray, J. J., Barnouin-Jha, O., and Murchie, S. 988

L., 2009. Identification of hydrated silicate minerals on Mars using MRO-CRISM: 989

Geologic context near Nili Fossae and implications for aqueous alteration. Journal of 990

Geophysical Research-Planets 114. 991

Ek, S., Root, A., Peussa, M., and Niinisto, L., 2001. Determination of the hydroxyl group 992

content in Silica by thermogravimetry and a comparison with 1H MAS NMR results. 993

Thermochimica Acta 379, 201-212. 994

Encrenaz, T., Bézard, B., Greathouse, T. K., Richter, M. J., Lacy, J. H., Atreya, S. K., Wong, 995

A. S., Lebonnois, S., Lefèvre, F., and Forget, F., 2004. Hydrogen peroxide on Mars: 996

evidence for spatial and seasonal variations. Icarus 170, 424-429. 997

Fernandez-Remolar, D. C., Sanchez-Roman, M., Hill, A. C., Gomez-Ortiz, D., Prieto 998

Ballesteros, O., Romanek, C. S., and Amils, R., 2011. The environment of early Mars 999

and the missing carbonates. Meteoritics & Planetary Science 46, 1447-1469. 1000

Forget, F., Wordsworth, R., Millour, E., Madeleine, J. B., Kerber, L., Leconte, J., Marcq, E., 1001

and Haberle, R. M., 2012. 3D modelling of the early martian climate under a denser 1002

CO2 atmosphere: Temperatures and CO2 ice clouds. Icarus 222, 81-99. 1003

Gaffey, S. J., 1986. Spectral reflectance of carbonate minerals in the visible and near-infrared 1004

(0.35-2.55 microns): calcite, aragonite, and dolomite. American Mineralogist 71, 151-1005

162. 1006

Galan, E., Carretero, M. I., Fernandez-Caliani, J. C., 1999. Effects of acid mine drainage on 1007

clay minerals suspended in the Tinto River (Rio Tinto, Spain). An experimental 1008

approach. Clay Minerals 34, 99-108. 1009

Garcia, B., Beaumont, V., Perfetti, E., Rouchon, V., Blanchet, D., Oger, P., Dromart, G., Huc, 1010

A. Y., and Haeseler, F., 2010. Experiments and geochemical modelling of CO2 1011

42

sequestration by olivine: Potential, quantification. Applied Geochemistry 25, 1383-1012

1396. 1013

Gaudin, A., Dehouck, E., and Mangold, N., 2011. Evidence for weathering on early Mars 1014

from a comparison with terrestrial weathering profiles. Icarus 216, 257-268. 1015

Gerdemann, S. J., O'Connor, W. K., Dahlin, D. C., Penner, L. R., and Rush, H., 2007. Ex situ 1016

aqueous mineral carbonation. Environmental Science & Technology 41, 2587-2593. 1017

Giammar, D. E., Bruant, R. G., and Peters, C. A., 2005. Forsterite dissolution and magnesite 1018

precipitation at conditions relevant for deep saline aquifer storage and sequestration of 1019

carbon dioxide. Chemical Geology 217, 257-276. 1020

Giorgetti, G., Marescotti, P., Cabella, R., and Lucchetti, G., 2001. Clay mineral mixtures as 1021

alteration products in pillow basalts from the eastern flank of Juan de Fuca Ridge: a 1022

TEM-AEM study. Clay Minerals 36, 75-91. 1023

Golubev, S. V., Pokrovsky, O. S., and Schott, J., 2005. Experimental determination of the 1024

effect of dissolved CO2 on the dissolution kinetics of Mg and Ca silicates at 25 1025

degrees C. Chemical Geology 217, 227-238. 1026

Golubev, S. V., Bauer, A., Pokrovsky, O. S., 2006. Effect of pH and organic ligands on the 1027

kinetics of smectite dissolution at 25 °C. Geochimica et Cosmochimica Acta 70, 4436-1028

4451 1029

Greenberger, R. N., Mustard, J. F., Kumar, P. S., Dyar, M. D., Breves, E. A., and Sklute, E. 1030

C., 2012. Low temperature aqueous alteration of basalt: Mineral assemblages of 1031

Deccan basalts and implications for Mars. Journal of Geophysical Research 117, 1032

E00J12. 1033

Grotzinger, J., Sumner, D. Y., Kah, L. C., Stack, K., Gupta, S., Edgar, L., Rubin, D., Lewis, 1034

K., Schieber, J., Mangold, N., Milliken, R., Conrad, P. G., DesMarais, D., Farmer, J., 1035

Siebach, K., Calef III, F., Hurowitz, J., McLennan, S. M., Ming, D., Vaniman, D., 1036

43

Crisp, J., Vasavada, A., Edgett, K. S., Malin, M., Blake, D., Gellert, R., Mahaffy, P., 1037

Wiens, R. C., Maurice, S., Grant, J. A., Wilson, S., Anderson, R. C., Beegle, L., 1038

Arvidson, R., Hallet, B., Sletten, R. S., Rice, M., Bell III, J., Griffes, J., Ehlmann, B., 1039

Anderson, R. B., Bristow, T. F., Dietrich, W. E., Dromart, G., Eigenbrode, J., 1040

Fraeman, A., Hardgrove, C., Herkenhoff, K., Jandura, L., Kocurek, G., Lee, S., 1041

Leshin, L. A., Leveille, R., Limonadi, D., Maki, J., McCloskey, S., Meyer, M., Minitti, 1042

M., Newsom, H., Oehler, D., Okon, A., Palucis, M., Parker, T., Rowland, S., Schmidt, 1043

M., Squyres, S., Steele, A., Stolper, E., Summons, R., Treiman, A., Williams, R., 1044

Yingst, A. and the MSL Science Team, 2014. A habitable fluvio-lacustrine 1045

environment at Yellowknife Bay, Gale crater, Mars. Science 343, doi: 1046

10.1126/science.1242777. 1047

Hänchen, M., Prigiobbe, V., Baciocchi, R., and Mazzotti, M., 2008. Precipitation in the Mg-1048

carbonate system - effects of temperature and CO2 pressure. Chemical Engineering 1049

Science 63, 1012-1028. 1050

Hänchen, M., Prigiobbe, V., Storti, G., Seward, T. M., and Mazzotti, M., 2006. Dissolution 1051

kinetics of fosteritic olivine at 90-150 degrees C including effects of the presence of 1052

CO2. Geochimica et Cosmochimica Acta 70, 4403-4416. 1053

Hoefen, T. M., Clark, R. N., Bandfield, J. L., Smith, M. D., Pearl, J. C., and Christensen, P. 1054

R., 2003. Discovery of olivine in the Nili Fossae region of Mars. Science 302, 627-1055

630. 1056

Huguenin, R. L., 1982. Chemical weathering and the Viking biology experiments on Mars. 1057

Journal of Geophysical Research 87, 10069-10082. 1058

Hunt, G. R., Salisbury, J. W., and Lenhoff, C. J., 1972. Visible and near-infrared spectra of 1059

minerals and rocks: V. Halides, phosphates, arsenates, vanadates and borates. Modern 1060

Geology 3, 121-132. 1061

44

Hunt, G. R., 1977. Spectral signatures of particulate minerals in visible and near infrared. 1062

Geophysics 42, 501-513. 1063

Hurowitz, J. A., Fischer, W., Tosca, N. J., and Milliken, R. E., 2010. Origin of acidic surface 1064

waters and the evolution of atmospheric chemistry on early Mars. Nature Geoscience 1065

3, 323-326. 1066

Hurowitz, J. A., Tosca, N. J., McLennan, S. M., and Schoonen, M. A. A., 2007. Production of 1067

hydrogen peroxide in Martian and lunar soils. Earth and Planetary Science Letters 1068

255, 41-52. 1069

Jones, C. L. and Brearley, A. J., 2006. Experimental aqueous alteration of the Allende 1070

meteorite under oxidizing conditions: Constraints on asteroidal alteration. Geochimica 1071

et Cosmochimica Acta 70, 1040-1058. 1072

Jones, J. B. and Segnit, E. R., 1969. Water in sphere-type opal. Mineralogical Magazine 37, 1073

357-361. 1074

King, H. E., Plumper, O., and Putnis, A., 2010. Effect of Secondary Phase Formation on the 1075

Carbonation of Olivine. Environmental Science & Technology 44, 6503-6509. 1076

Knauth, L. P. and Epstein, S., 1982. The nature of water in hydrous silica. American 1077

Mineralogist 67, 510-520. 1078

Kotra, R. K., Gibson, E. K., and Urbancic, M. A., 1982. Release of volatiles from possible 1079

Martian analogs. Icarus 51, 593-605. 1080

Komadel, P., and Madejová, J., 2006. Acid activation of clay minerals. In: Bergaya, F., 1081

Theng, B. K. G., and Lagaly, G., Eds., Handbook of clay science. Elsevier. 1082

Lantenois, S., Champallier, R., Beny, J. M., and Muller, F., 2008. Hydrothermal synthesis and 1083

characterization of dioctahedral smectites: A montmorillonites series. Applied Clay 1084

Science 38, 165-178. 1085

45

Le Deit, L., Flahaut, J., Quantin, C., Hauber, E., Mege, D., Bourgeois, O., Gurgurewicz, J., 1086

Masse, M., and Jaumann, R., 2012. Extensive surface pedogenic alteration of the 1087

Martian Noachian crust suggested by plateau phyllosilicates around Valles Marineris. 1088

Journal of Geophysical Research-Planets 117. 1089

Loizeau, D., Mangold, N., Poulet, F., Bibring, J. P., Gendrin, A., Ansan, V., Gomez, C., 1090

Gondet, B., Langevin, Y., Masson, P., and Neukum, G., 2007. Phyllosilicates in the 1091

Mawrth Vallis region of Mars. Journal of Geophysical Research-Planets 112. 1092

Loizeau, D., Mangold, N., Poulet, F., Ansan, V., Hauber, E., Bibring, J. P., Gondet, B., 1093

Langevin, Y., Masson, P., and Neukum, G., 2010. Stratigraphy in the Mawrth Vallis 1094

region through OMEGA, HRSC color imagery and DTM. Icarus 205, 396-418. 1095

Longhi, J., Knittle, E., Holloway, J. R., and Wänke, H., 1992. The bulk composition, 1096

mineralogy and internal structure of Mars. In: Kieffer, H. H., Jakosky, B. M., Snyder, 1097

C. W., and Matthews, M. S., Eds., Mars. The University of Arizona Press, Tucson. 1098

Mangold, N., Poulet, F., Mustard, J. F., Bibring, J. P., Gondet, B., Langevin, Y., Ansan, V., 1099

Masson, P., Fassett, C., Head, J. W., Hoffmann, H., and Neukum, G., 2007. 1100

Mineralogy of the Nili Fossae region with OMEGA/Mars Express data: 2. Aqueous 1101

alteration of the crust. Journal of Geophysical Research-Planets 112. 1102

McLennan, S. M., 2003. Sedimentary silica on Mars. Geology 31, 315-318. 1103

McLennan, S. M., Anderson, R. B., Bell III, J. F., Bridges, J. C., Calef III, F., Campbell, J. L., 1104

Clark, B. C., Clegg, S., Conrad, P., Cousin, A., Des Marais, D. J., Dromart, G., Dyar, 1105

M. D., Edgar, L. A., Ehlmann, B. L., Fabre, C., Forni, O., Gasnault, O., Gellert, R., 1106

Gordon, S., Grant, J. A., Grotzinger, J. P., Gupta, S., Herkenhoff, K. E., Hurowitz, J. 1107

A., King, P. L., Le Mouélic, S., Leshin, L. A., Léveillé, R., Lewis, K. W., Mangold, 1108

N., Maurice, S., Ming, D. W., Morris, R. V., Nachon, M., Newsom, H. E., Ollila, A. 1109

M., Perrett, G. M., Rice, M. S., Schmidt, M. E., Schwenzer, S. P., Stack, K., Stolper, 1110

46

E. M., Sumner, D. Y., Treiman, A. H., VanBommel, S., Vaniman, D. T., Vasavada, 1111

A., Wiens, R. C., Yingst, R. A., and the MSL Science Team, 2014. Elemental 1112

geochemistry of sedimentary rocks at Yellowknife Bay, Gale crater, Mars. Science 1113

343, doi: 10.1126/science.1244734. 1114

Mitov, I., Paneva, D., and Kunev, B., 2002. Comparative study of the thermal decomposition 1115

of iron oxyhydroxides. Thermochimica Acta 386, 179-188. 1116

Murakami, T., Ito, J. I., Utsunomiya, S., Kasama, T., Kozai, N., and Ohnuki, T., 2004. Anoxic 1117

dissolution processes of biotite: implications for Fe behavior during Archean 1118

weathering. Earth and Planetary Science Letters 224, 117-129. 1119

Murchie, S. L., Mustard, J. F., Ehlmann, B. L., Milliken, R. E., Bishop, J. L., McKeown, N. 1120

K., Noe Dobrea, E. Z., Seelos, F. P., Buczkowski, D. L., Wiseman, S. M., Arvidson, 1121

R. E., Wray, J. J., Swayze, G., Clark, R. N., Des Marais, D. J., McEwen, A. S., and 1122

Bibring, J.-P., 2009. A synthesis of Martian aqueous mineralogy after one Mars year 1123

of observations from the Mars Reconnaissance Orbiter. Journal of Geophysical 1124

Research 114, E00D06. 1125

Mustard, J. F., Poulet, F., Gendrin, A., Bibring, J. P., Langevin, Y., Gondet, B., Mangold, N., 1126

Bellucci, G., and Altieri, F., 2005. Olivine and pyroxene diversity in the crust of Mars. 1127

Science 307, 1594-1597. 1128

Neubeck, A., Duc, N. T., Bastviken, D., Crill, P., and Holm, N. G., 2011. Formation of H(2) 1129

and CH(4) by weathering of olivine at temperatures between 30 and 70°C. 1130

Geochemical Transactions 12. 1131

Noe Dobrea, E. Z., Bishop, J. L., McKeown, N. K., Fu, R., Rossi, C. M., Michalski, J. R., 1132

Heinlein, C., Hanus, V., Poulet, F., Mustard, R. J. F., Murchie, S., McEwen, A. S., 1133

Swayze, G., Bibring, J. P., Malaret, E., and Hash, C., 2010. Mineralogy and 1134

stratigraphy of phyllosilicate-bearing and dark mantling units in the greater Mawrth 1135

47

Vallis/west Arabia Terra area: Constraints on geological origin. Journal of 1136

Geophysical Research-Planets 115. 1137

Ody, A., Poulet, F., Bibring, J. P., Loizeau, D., Carter, J., Gondet, B., and Langevin, Y., 2013. 1138

Global investigation of olivine on Mars: Insights into crust and mantle compositions. 1139

Journal of Geophysical Research: Planets 118, 234-262. 1140

Olsen, A. A. and Rimstidt, J. D., 2007. Using a mineral lifetime diagram to evaluate the 1141

persistence of olivine on Mars. American Mineralogist 92, 598-602. 1142

Owen, T., 1982. The composition of the Martian atmosphere. Advances in Space Research 2, 1143

75-80. 1144

Paris, M., Fritsch, E., and Reyes, B. O. A., 2007. H-1, Si-29 and Al-27 NMR study of the 1145

destabilization process of a paracrystalline opal from Mexico. Journal of Non-1146

Crystalline Solids 353, 1650-1656. 1147

Pokrovsky, O. S. and Schott, J., 2000. Kinetics and mechanism of forsterite dissolution at 25 1148

degrees C and pH from 1 to 12. Geochimica Et Cosmochimica Acta 64, 3313-3325. 1149

Poulet, F., Bibring, J. P., Mustard, J. F., Gendrin, A., Mangold, N., Langevin, Y., Arvidson, 1150

R. E., Gondet, B., and Gomez, C., 2005. Phyllosilicates on Mars and implications for 1151

early martian climate. Nature 438, 623-627. 1152

Poulet, F., Gomez, C., Bibring, J. P., Langevin, Y., Gondet, B., Pinet, P., Belluci, G., and 1153

Mustard, J., 2007. Martian surface mineralogy from Observatoire pour la Minéralogie, 1154

l'Eau, les Glaces et l'Activité on board the Mars Express spacecraft (OMEGA/MEx): 1155

Global mineral maps. Journal of Geophysical Research-Planets 112. 1156

Prasad, S. V. S. and Sitakara Rao, V., 1984. Thermal transformation of iron (III) oxide 1157

hydrate gel. Journal of Materials Science 19, 3266-3270. 1158

48

Rondeau, B., Fritsch, E., Guiraud, M., and Renac, C., 2004. Opals from Slovakia 1159

("Hungarian" opals): a re-assessment of the conditions of formation. European 1160

Journal of Mineralogy 16, 789-799. 1161

Santiago Buey, C., Barrios, M. S., Romero, E. G., and Montoya, M. D., 2000. Mg-rich 1162

smectite "precursor" phase in the Tagus Basin, Spain. Clays and Clay Minerals 48, 1163

366-373. 1164

Schröder, C., Klingelhöfer, G., and Tremel, M., 2004. Weathering of Fe-bearing minerals 1165

under Martian conditions, investigated by Mossbauer spectroscopy. Planetary and 1166

Space Science 52, 997-1010. 1167

Smith, K. L., Milnes, A. R., and Eggleton, R. A., 1987. Weathering of basalt: formation of 1168

iddingsite. Clays and Clay Minerals 35, 418-428. 1169

Tosca, N. J., Milliken, R. E., and Michel, F. M., 2008. Smectite formation on early Mars: 1170

experimental constraints. Martian phyllosilicates: recorders of aqueous processes, 1171

abstract #7030. 1172

Vaniman, D. T., Bish, D. L., Ming, D. W., Bristow, T. F., Morris, R. V., Blake, D. F., 1173

Chipera, S. J., Morrison, S. M., Treiman, A. H., Rampe, E. B., Rice, M., Achilles, C. 1174

N., Grotzinger, J., McLennan, S. M., Williams, J., Bell III, J., Newsom, H., Downs, R. 1175

T., Maurice, S., Sarrazin, P., Yen, A. S., Morookian, J. M., Farmer, J. D., Stack, K., 1176

Milliken, R. E., Ehlmann, B., Sumner, D. Y., Berger, G., Crisp, J. A., Hurowitz, J. A., 1177

Anderson, R., DesMarais, D., Stolper, E. M., Edgett, K. S., Gupta, S., Spanovich, N. 1178

and the MSL Science Team, 2014. Mineralogy of a mudstone at Yellowknife Bay, 1179

Gale crater, Mars. Science 343, doi: 10.1126/science.1243480. 1180

Velbel, M. A., Long, D. T., and Gooding, J. L., 1991. Terrestrial weathering of Antarctic 1181

stone meteorites: Formation of Mg-carbonates on ordinary chondrites. Geochimica et 1182

Cosmochimica Acta 55, 67-76. 1183

49

Wilson, M. J., 2004. Weathering of the primary rock-forming minerals: processes, products 1184

and rates. Clay Minerals 39, 233-266. 1185

Wogelius, R. A. and Walther, J. V., 1992. Olivine dissolution kinetics at near-surface 1186

conditions. Chemical Geology 97, 101-112. 1187

Wordsworth, R., Forget, F., Millour, E., Head, J. W., Madeleine, J. B., and Charnay, B., 2012. 1188

Global modelling of the early martian climate under a denser CO2 atmosphere: Water 1189

cycle and ice evolution. Icarus 222, 1-19. 1190

Wray, J. J., Murchie, S. L., Squyres, S. W., Seelos, F. P., and Tornabene, L. L., 2009. Diverse 1191

aqueous environments on ancient Mars revealed in the southern highlands. Geology 1192

37, 1043-1046. 1193

Yen, A. S., Kim, S. S., Hecht, M. H., Frant, M. S., and Murray, B., 2000. Evidence that the 1194

reactivity of the martian soil is due to superoxide ions. Science 289, 1909-1912. 1195

Zent, A. P., 1998. On the thickness of the oxidized layer of the Martian regolith. Journal of 1196

Geophysical Research-Planets 103, 31491-31498. 1197

1198

50

FIGURES AND TABLES 1199

1200

Fig. 1. NIR spectrum of the initial olivine used in the experiment. Left panel: general aspect 1201

of the spectrum, exhibiting a typical olivine signature with a broad Fe2+-related absorption 1202

band at 1.04 µm. Left panel, inset: small apatite crystals (arrows) observed by SEM on the 1203

surface of some olivine grains. Middle panel: detail of the spectrum of the initial olivine, 1204

revealing a very shallow band at 2.31 µm (top), compared to a reference spectrum of apatite 1205

(bottom; sample LAAP03 of CRISM spectral library). Note the different reflectance scales. 1206

Right panel: another detail of the spectrum, revealing a small absorption band at 4.02 µm 1207

(top), which may also be due to apatite despite the slight shift in comparison with the 1208

reference spectrum (bottom). 1209

1210

Fig. 2. Schematic diagram and photograph of the experimental device, and summary of the 1211

weathering conditions tested. 1212

1213

Fig. 3. Chemical evolution with time of the experimental solutions. Left panel: evolution of 1214

pH (diamonds). Theoretical starting values (triangles) were calculated using the JCHESS 1215

geochemical software for pure water under CO2 partial pressures of 1 (CO2 atmosphere) and 1216

4.10-4 (terrestrial atmosphere). Error bars are added for unstable measures. Right panel: 1217

evolution of Si, Mg and Fe mass concentrations. Error bars include analytical uncertainties 1218

(provided by the SARM laboratory) and further uncertainties related to contaminations by the 1219

experimental device itself, which were determined thanks to a blank experiment (without 1220

solid phase). See section 2.4.4 for details about the protocols. 1221

1222

51

Fig. 4. NIR spectra of the final (weathered) samples of the experiment, compared to the 1223

spectrum of the initial olivine and to some library spectra. A: general aspect of the spectra 1224

between 1.0 and 2.6 µm. B: detailed view between 1.8 and 2.5 µm. C: detailed view between 1225

3.8 and 4.2 µm. Arrows in B and C indicate newly-appeared or deepened absorption bands. 1226

D: reference spectra of clay minerals and opal from the USGS library (Clark et al., 2007) 1227

(nontronite NG-1a, saponite SapCa-1, montmorillonite CM20; opal TM8896-hyalite). E: 1228

spectrum of a natural sample of nesquehonite (mixed with other Mg-carbonates) (Dehouck et 1229

al., 2012). F: reference spectra of carbonate minerals from the USGS library (hydromagnesite 1230

LACB28A, magnesite LACB03A, siderite LACB08A). 1231

1232

Fig. 5. Band depths calculations for the initial and weathered olivine powders. Error bars 1233

correspond to standard deviation around the average (diamonds) for six different 1234

measurements. For the 1.91-µm band, the continuum anchors are taken at 1.80 and 2.10 µm. 1235

For the 2.31-µm band, the continuum anchors are taken at 2.25 and 2.36 µm. In both cases, 1236

every values used in the calculation correspond to the median of three spectral channels to 1237

minimize the effect of instrumental noise. 1238

1239

Fig. 6. NIR spectra of the <2-µm fraction of the weathered olivine powders. The slight shift 1240

of the Fe2+-related band at 1.07 µm compared to Fig. 4 is an effect of the lower grain size. The 1241

detailed, continuum-corrected view on the right highlights the 2.31-µm absorption band, 1242

whose depth varies between samples. 1243

1244

Fig. 7. X-ray diffraction patterns (Cu Kα radiation) obtained from the oriented <2-µm fraction 1245

of the initial and weathered olivine powders. No diffraction peaks corresponding to clay 1246

minerals are observed in any sample. 1247

52

Fig. 8. TEM micrographs showing the different textures observed at the surface of the olivine 1248

grains (in dark) before and after the experiment. A: examples of clean surfaces resulting from 1249

grinding and typical of the initial olivine. B: examples of “cotton-like” textures, observed 1250

only in the “CO2” and “CO2-H2O2” samples. Note a nearby grain with clean surfaces in the 1251

top image. C: examples of filamentous textures, observed in the “Air”, “Air-H2O2” and 1252

“CO2” samples. D: comparison of the filamentous texture found in this study (left) with 1253

phyllosilicates (serpentine and saponite) produced by experimental alteration of olivine at 1254

200°C (Jones and Brearley, 2006) (right). The two micrographs are presented at the same 1255

scale. 1256

1257

Fig. 9. Chemical compositions measured by TEM-EDX in the initial and weathered olivine 1258

samples. A: ternary diagram Si-Mg-Fe showing individual compositions measured on initial 1259

grains and newly-formed phases (see caption). B to E: same data as in A, redrawn in the form 1260

of separate color ranges for clarity. 1261

1262

Fig. 10. Differential thermogravimetric (DTG) curves of the initial and weathered olivine 1263

powders. (“Air-H2O2” was not analyzed, having similar properties than “Air” in other 1264

datasets.) Samples were heated at 10 °C/min in dinitrogen. Mass losses are indicated by 1265

arrows and interpreted in terms of volatile releases (ads.: adsorbed). The mass loss at 760 °C 1266

in the curve of the “CO2” sample is probably a noise artifact. 1267

1268

Table 1. Chemical composition of the initial olivine used in the experiment, as determined by 1269

ICP-OES. LOD: limit of detection. LOI: loss on ignition (measured by thermogravimetry and 1270

heating at 1000 °C). 1271

1272

53

Table 2. Summary of analytical results and interpretations in terms of secondary phases. For 1273

NIR spectroscopy and TEM observations, the number of “+” symbols indicates the relative 1274

intensity of the absorption bands and the relative abundances of the newly-formed phases as 1275

estimated by eye, respectively. aEstimation based on the amounts of H2O and OH released 1276

from the corresponding samples, assuming a water content of 6 to 13 wt% for the Si-rich 1277

phase (Knauth and Epstein, 1982; Paris et al., 2007). The value may be slightly overestimated 1278

for the “CO2” sample, for which some H2O is borne by the smectite phase. bUpper limit of ~1 1279

wt% deduced from the lower abundance and thickness of the smectite/filamentous phase 1280

compared to the Si-rich/cotton-like phase, as observed by TEM. In addition, NIR and TEM 1281

indicate that the smectite phase is less abundant in the “CO2” sample than in the two “Earth 1282

samples”. cEstimation based on the amount of CO2 released in thermogravimetric analyses 1283

(attributed to magnesite and siderite, see text). dUpper limit estimated from the amount of 1284

dissolved Fe measured in the “CO2” reactor (24 mg/L at 14 days; Table S1), assuming that all 1285

of this Fe is precipitated as (oxy)hydroxide. ads.: adsorbed. n. a.: not analyzed. LOD: limit of 1286

detection.1287

54

1288

Fig. 1. NIR spectrum of the initial olivine used in the experiment. Left panel: general aspect 1289

of the spectrum, exhibiting a typical olivine signature with a broad Fe2+-related absorption 1290

band at 1.04 µm. Left panel, inset: small apatite crystals (arrows) observed by SEM on the 1291

surface of some olivine grains. Middle panel: detail of the spectrum of the initial olivine, 1292

revealing a very shallow band at 2.31 µm (top), compared to a reference spectrum of apatite 1293

(bottom; sample LAAP03 of CRISM spectral library). Note the different reflectance scales. 1294

Right panel: another detail of the spectrum, revealing a small absorption band at 4.02 µm 1295

(top), which may also be due to apatite despite the slight shift in comparison with the 1296

reference spectrum (bottom). 1297

1298

55

1299

Fig. 2. Schematic diagram and photograph of the experimental device, and summary of the 1300

weathering conditions tested. 1301

1302

56

1303

Fig. 3. Chemical evolution with time of the experimental solutions. Left panel: evolution of 1304

pH (diamonds). Theoretical starting values (triangles) were calculated using the JCHESS 1305

geochemical software for pure water under CO2 partial pressures of 1 (CO2 atmosphere) and 1306

4.10-4 (terrestrial atmosphere). Error bars are added for unstable measures. Right panel: 1307

evolution of Si, Mg and Fe mass concentrations. Error bars include analytical uncertainties 1308

(provided by the SARM laboratory) and further uncertainties related to contaminations by the 1309

experimental device itself, which were determined thanks to a blank experiment (without 1310

solid phase). See section 2.4.4 for details about the protocols. 1311

1312

57

1313

Fig. 4. NIR spectra of the final (weathered) samples of the experiment, compared to the 1314

spectrum of the initial olivine and to some library spectra. A: general aspect of the spectra 1315

between 1.0 and 2.6 µm. B: detailed view between 1.8 and 2.5 µm. C: detailed view between 1316

3.8 and 4.2 µm. Arrows in B and C indicate newly-appeared or deepened absorption bands. 1317

58

D: reference spectra of clay minerals and opal from the USGS library (Clark et al., 2007) 1318

(nontronite NG-1a, saponite SapCa-1, montmorillonite CM20; opal TM8896-hyalite). E: 1319

spectrum of a natural sample of nesquehonite (mixed with other Mg-carbonates) (Dehouck et 1320

al., 2012). F: reference spectra of carbonate minerals from the USGS library (hydromagnesite 1321

LACB28A, magnesite LACB03A, siderite LACB08A). 1322

1323

59

1324

Fig. 5. Band depths calculations for the initial and weathered olivine powders. Error bars 1325

correspond to standard deviation around the average (diamonds) for six different 1326

measurements. For the 1.91-µm band, the continuum anchors are taken at 1.80 and 2.10 µm. 1327

For the 2.31-µm band, the continuum anchors are taken at 2.25 and 2.36 µm. In both cases, 1328

every values used in the calculation correspond to the median of three spectral channels to 1329

minimize the effect of instrumental noise. 1330

1331

60

1332

Fig. 6. NIR spectra of the <2-µm fraction of the weathered olivine powders. The slight shift 1333

of the Fe2+-related band at 1.07 µm compared to Fig. 4 is an effect of the lower grain size. The 1334

detailed, continuum-corrected view on the right highlights the 2.31-µm absorption band, 1335

whose depth varies between samples. 1336

1337

61

1338

Fig. 7. X-ray diffraction patterns (Cu Kα radiation) obtained from the oriented <2-µm fraction 1339

of the initial and weathered olivine powders. No diffraction peaks corresponding to clay 1340

minerals are observed in any sample. 1341

1342

62

1343

Fig. 8. TEM micrographs showing the different textures observed at the surface of the olivine 1344

grains (in dark) before and after the experiment. A: examples of clean surfaces resulting from 1345

grinding and typical of the initial olivine. B: examples of “cotton-like” textures, observed 1346

only in the “CO2” and “CO2-H2O2” samples. Note a nearby grain with clean surfaces in the 1347

top image. C: examples of filamentous textures, observed in the “Air”, “Air-H2O2” and 1348

“CO2” samples. D: comparison of the filamentous texture found in this study (left) with 1349

phyllosilicates (serpentine and saponite) produced by experimental alteration of olivine at 1350

200°C (Jones and Brearley, 2006) (right). The two micrographs are presented at the same 1351

scale.1352

63

1353

Fig. 9. Chemical compositions measured by TEM-EDX in the initial and weathered olivine 1354

samples. A: ternary diagram Si-Mg-Fe showing individual compositions measured on initial 1355

grains and newly-formed phases (see caption). B to E: same data as in A, redrawn in the form 1356

of separate color ranges for clarity. 1357

1358

64

1359

Fig. 10. Differential thermogravimetric (DTG) curves of the initial and weathered olivine 1360

powders. (“Air-H2O2” was not analyzed, having similar properties than “Air” in other 1361

datasets.) Samples were heated at 10 °C/min in dinitrogen. Mass losses are indicated by 1362

arrows and interpreted in terms of volatile releases (ads.: adsorbed). The mass loss at 760 °C 1363

in the curve of the “CO2” sample is probably a noise artifact. 1364

1365

1366

65

1367

wt.%

SiO2 40.50

Al2O3 0.08

Fe2O3 10.44

MnO 0.13

MgO 50.09

CaO 0.10

Na2O 0.02

K2O <LOD

TiO2 <LOD

P2O5 <LOD

LOI -0.74

Total 100.63

1368

Table 1. Chemical composition of the initial olivine used in the experiment, as determined by 1369

ICP-OES. LOD: limit of detection. LOI: loss on ignition (measured by thermogravimetry and 1370

heating at 1000 °C). 1371

1372

66

1373

CO2-H2O2 CO2 Air-H2O2 Air

Analytical results

NIR bands 1.91 µm ++ ++ ++ ++

2.31 µm + ++ ++

TEM Cotton-like + ++

Filamentous + ++ ++

DTG mass

losses (%)

ads. H2O ~0.09 ~0.10 n. a. ~0.07

OH ~0.06 ~0.15 n. a.

CO2 ~0.02 n. a.

TIC (mg/g) <LOD 0.09 n. a. <LOD

Secondary phases (wt%)

Si-rich phasea ~1 – 3 ~2 – 4

Smectiteb <<1 <1 <1

Carbonatesc ~0.05

Fe-(oxy)hydroxidesd <0.1 <0.1

1374

Table 2. Summary of analytical results and interpretations in terms of secondary phases. For 1375

NIR spectroscopy and TEM observations, the number of “+” symbols indicates the relative 1376

intensity of the absorption bands and the relative abundances of the newly-formed phases as 1377

estimated by eye, respectively. aEstimation based on the amounts of H2O and OH released 1378

from the corresponding samples, assuming a water content of 6 to 13 wt% for the Si-rich 1379

phase (Knauth and Epstein, 1982; Paris et al., 2007). The value may be slightly overestimated 1380

for the “CO2” sample, for which some H2O is borne by the smectite phase. bUpper limit of ~1 1381

wt% deduced from the lower abundance and thickness of the smectite/filamentous phase 1382

compared to the Si-rich/cotton-like phase, as observed by TEM. In addition, NIR and TEM 1383

indicate that the smectite phase is less abundant in the “CO2” sample than in the two “Earth 1384

samples”. cEstimation based on the amount of CO2 released in thermogravimetric analyses 1385

(attributed to magnesite and siderite, see text). dUpper limit estimated from the amount of 1386

dissolved Fe measured in the “CO2” reactor (24 mg/L at 14 days; Table S1), assuming that all 1387

of this Fe is precipitated as (oxy)hydroxide. ads.: adsorbed. n. a.: not analyzed. LOD: limit of 1388

detection. 1389