HAL Id: hal-02938748
https://hal.archives-ouvertes.fr/hal-02938748
Submitted on 27 Nov 2020
HAL
is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire
HAL, estdestinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
During Serpentinization: Insights From the Oman Drilling Project
Benjamin Malvoisin, Chang Zhang, Othmar Müntener, Lukas Baumgartner, Peter Kelemen
To cite this version:
Benjamin Malvoisin, Chang Zhang, Othmar Müntener, Lukas Baumgartner, Peter Kelemen. Mea-
surement of Volume Change and Mass Transfer During Serpentinization: Insights From the Oman
Drilling Project. Journal of Geophysical Research : Solid Earth, American Geophysical Union, 2020,
125 (5), �10.1029/2019JB018877�. �hal-02938748�
Measurement of volume change and mass transfer
1
during serpentinisation: insights from the Oman
2
Drilling Project
3
Benjamin Malvoisin 1,2, Chang Zhang 1,3, Othmar M¨untener1, Lukas P.
4
Baumgartner1, Peter B. Kelemen 4,5, Oman Drilling Project Science Party
5
1Institut des Sciences de la Terre, Universit´e de Lausanne, Lausanne, Switzerland.
6 2Universit´e Grenoble Alpes, CNRS, ISTerre, 38000 Grenoble, France.
7 3State Key Laboratory of Lithospheric Evolution, Institute of Geology and Geophysics, Chinese Academy
8
of Sciences, Beijing 100029, China
9 4Lamont-Doherty Earth Observatory, Columbia University, 61 Route 9W, Palisades, NY 10964, United
10
States of America
11 5Department of Earth and Environmental Sciences, Columbia University, New York, NY 10027, United
12
States of America
13
Key Points:
14
• A dunite collected during the Oman Drilling Project records serpentinisation at
15
low temperature.
16
• The solid volume increases by more than 50 % during reaction as predicted for
17
a closed system.
18
• The primary olivine trace element zoning is preserved during serpentinisation.
19
Corresponding author: Benjamin Malvoisin,benjamin.malvoisin@unil.ch
Abstract
20
Serpentinisation plays a key role on the evolution of the physico-chemical properties
21
of the mantle lithosphere. The rate of serpentinisation reactions is controlled by the
22
transport of fluid which itself depends on volume change during reaction. Element
23
transfer can strongly modify the magnitude and sign of volume change. Here, we mea-
24
sure solid volume change and element transport perpendicular to a serpentine vein
25
in a serpentinised dunite collected at depth during the Oman Drilling Project. The
26
sample is extensively replaced (extent of reaction > 80 %) by a serpentine/brucite
27
mixture parallel to a main serpentine vein network. The Mg content of serpentine
28
and brucite indicates reaction with a small amount of fluid at temperatures below
29
100 ◦C. Concentrations of fluid-mobile trace elements (Na, Ca, Sr, Rb and Ba) de-
30
crease perpendicular to the main vein. Primary olivine contains parallel platelets of a
31
clinopyroxene/magnetite symplectite. Tomography at the nanoscale reveals that these
32
inclusions do not react during serpentinisation but are cracked and displaced. We use
33
these inert markers to measure a 59 to 74 % positive volume change, that is close to
34
the 52 % expected for reaction in a closed system. Chemical data indicate no change
35
in major element composition during reaction except for the addition of water. The
36
initial olivine zoning in Al, Ti, V, Sc and Cr is still preserved in serpentine and brucite.
37
Serpentinisation can thus be a local replacement process during which the solid volume
38
homogeneously increases at the micrometer scale and the transport of aqueous species
39
is limited.
40
1 Introduction
41
Ultramafic rocks react with water at slow spreading ridges, at ocean-continent
42
transitions, in subduction zones and in ophiolites. This process of ”serpentinisation”
43
raises the interest of the scientific community not only because it is ubiquitous but
44
also because it has a first-order impact in geodynamics and in biology (Kelley et
45
al., 2001; Sleep et al., 2004; Delescluse & Chamot-Rooke, 2008; Wada et al., 2008;
46
M¨untener, 2010; Russell et al., 2010; Reynard, 2013) since it produces serpentine
47
((M g, F e)3Si2O5(OH)4), brucite ((M g, F e)OH2), magnetite (F e3O4) and hydrogen
48
(H2) with physico-chemical properties (i.e. composition, density, magnetism, rheol-
49
ogy and energetic potential) strongly different from the primary olivine and pyroxene
50
(Moody, 1976; Escart´ın et al., 2001; Oufi, 2002; McCollom & Bach, 2009; Malvoisin,
51
Carlut, & Brunet, 2012).
52
Klein et al. (2009, 2013) developed thermodynamic models predicting the abun-
53
dance and composition of the products of serpentinisation as a function of tempera-
54
ture and reactant compositions. They predict that magnetite formation mainly occurs
55
above 200◦C whereas Fe-brucite and Fe-serpentine are the dominant iron-bearing phase
56
below this temperature (Klein et al., 2014). Experiments on powders provide serpen-
57
tinisation rates that are dependent on temperature, pressure, grain size and protolith
58
composition (Martin & Fyfe, 1970; Wegner & Ernst, 1983; Malvoisin, Brunet, et al.,
59
2012; Andreani et al., 2013). However, these constraints are not sufficient to quanti-
60
tatively model serpentinisation in natural systems. Experiments on rock aggregates
61
indeed revealed orders of magnitude slower kinetics compared to rates measured on
62
powders (Malvoisin & Brunet, 2014; Klein et al., 2015). Water transport thus plays
63
a key role on reaction rate and local thermodynamic equilibrium. Some peridotites
64
interacting with near-surface fluids have undergone rapid cooling. This generated fluid
65
pathways through thermal cracking (Demartin et al., 2004; Boudier et al., 2005) and
66
tectonic deformation (Roum´ejon & Cannat, 2014). Understanding the evolution of
67
these pathways and the generation of new pathways during serpentinisation requires
68
understanding of the couplings between reaction and deformation. Serpentine and
69
brucite have a lower density than olivine and pyroxene. Serpentinisation in a closed
70
system is thus expected to induce an increase in solid volume (ΔVV) of approximately
71
50 % (Macdonald & Fyfe, 1985; O’Hanley, 1992). This change in volume has two
72
opposite consequences for fluid pathway evolution during serpentinisation. It leads to
73
both decrease and increase in permeability through porosity filling by reaction prod-
74
ucts (Farough et al., 2015; Godard et al., 2013) and stress build-up leading to cracking
75
(Macdonald & Fyfe, 1985; O’Hanley, 1992; B. W. Evans, 2004; Jamtveit et al., 2009),
76
respectively. This latter process of reaction-induced cracking is constrained with ob-
77
servations in natural systems (Coleman & Keith, 1971; Loney et al., 1971; Pl¨umper et
78
al., 2012; Roum´ejon & Cannat, 2014; Malvoisin et al., 2017), thermodynamic theory
79
(Kelemen & Hirth, 2012), thermodynamic models of water rock reaction in natural
80
and simplified systems (Klein et al., 2009; Malvoisin, 2015; de Obeso et al., 2017),
81
physical and numerical models (Rudge et al., 2010; Ulven et al., 2014; Shimizu &
82
Okamoto, 2016; Malvoisin et al., 2017; O. Evans et al., 2018, n.d.) and experiments
83
(Zhu et al., 2016; Zheng et al., 2018; Lafay et al., 2018). Despite the key role of solid
84
volume change during serpentinisation, this parameter is not well constrained since it
85
has not been measured in natural samples yet and it is closely related to mass transfer
86
during reaction.
87
The transport of aqueous species during reaction can modify solid volume change
88
(Thayer, 1966; Gresens, 1967; Carmichael, 1987; Fletcher & Merino, 2001; A. Putnis &
89
Austrheim, 2010). For example, pseudomorphic replacement and microtexture preser-
90
vation are interpreted as evidence for isochoric replacement thanks to mass transfer
91
(Merino & Dewers, 1998; Centrella et al., 2015). Some serpentinised peridotites have
92
a different composition than mantle rocks. In addition to water incorporation, the
93
main change in major elements measured in abyssal serpentinised peridotites consists
94
in a decrease of theM gO/SiO2 ratio by 5 % on average (Snow & Dick, 1995; Niu,
95
2004; Monnier et al., 2006; Malvoisin, 2015; de Obeso & Kelemen, 2018). This de-
96
crease can be explained by Si gain or Mg loss during reaction (Malvoisin, 2015; J¨ons
97
et al., 2017; de Obeso & Kelemen, 2018). Enrichment in fluorine, boron and chlorine
98
are also measured in abyssal peridotites (Thompson & Melson, 1970; Bonatti et al.,
99
1984; Sharp & Barnes, 2004; Vils et al., 2008; Bonifacie et al., 2008; Debret et al.,
100
2014; Anselmi et al., 2014). Previous studies on trace elements in abyssal serpentinised
101
peridotites are limited to the hand-specimen scale without in-situ investigations. The
102
bulk geochemistry shows patterns in rare earth elements (REE) highly related to those
103
of their protoliths. Al, V, Sc and Cr are considered as ’immobile’ during hydrother-
104
mal alteration (Paulick et al., 2006). The high field strength elements (HFSEs) show
105
low solubility and mobility in aqueous fluids during hydrothermal alteration relative
106
to REEs. It was thus proposed to use REE and compatible elements to identify the
107
protoliths of serpentinites and to recover the processes experienced by these protoliths
108
(Deschamps et al., 2013). In contrast, U, Pb and Sr contents are characterized by
109
strong enrichments associated with alteration on the seafloor, and in particular with
110
the precipitation of carbonates (Seifert & Brunotte, 1996; Olivier & Boyet, 2006).
111
These changes in composition can influence the geochemical budgets at large scale
112
(Paulick et al., 2006). However, it is difficult to determine the respective contribu-
113
tion of serpentinisation and alteration on the seafloor to mass transfer and porosity
114
evolution in these samples.
115
A way to focus on the mineralogical processes associated with serpentinisation is
116
to work on samples collected by drilling. Tutolo et al. (2016) measured several percent
117
of porosity in serpentinised peridotites drilled in the Atlantis massif. The observed pore
118
diameters are orders of magnitude smaller than the pores formed during alteration on
119
the seafloor (10 μm (J¨ons et al., 2017)). Nanoscale porosity probably plays a key
120
role for fluid transport during replacement reactions but its mechanism of formation
121
remains poorly understood due to a lack of constraints on volume change and mass
122
transfer during reaction.
123
We measure here solid volume change associated with serpentinisation in a sam-
124
ple drilled during the International Continental Scientific Drilling Program (ICDP)
125
supported Oman Drilling Project. We couple these results with measurements of ma-
126
jor and trace element composition, thermodynamic modelling, and nanotomography.
127
The combination of these tools provides new constraints for determining the miner-
128
alogical processes at play during serpentinisation.
129
2 Sampling and methods
130
We study a core sample from Oman Drilling Project Hole BA4A (sample BA4A-
131
81-1-1-17), drilled into serpentinised peridotites of the Wadi Tayi massif of the Samail
132
Ophiolite, the largest sub-aerial exposure of oceanic lithosphere on Earth. The Samail
133
Ophiolite was formed by igneous accretion at a fast-spreading ridge and then obducted
134
onto the Arabian Continental margin (Boudier et al., 1988). Serpentinisation may have
135
occurred throughout the history of the ophiolite and is still ongoing (Neal & Stanger,
136
1983; Kelemen & Matter, 2008). The extent of hydration (serpentinisation) of core
137
from Hole BA4A well is generally more than 95 %, with only minor relicts of primary
138
mantle minerals. The sample we study comes from one of the freshest portions of the
139
core at a depth of 215 m. It is a dunite, principally composed of serpentine, brucite
140
and olivine in a mesh texture, which is cross-cut by a serpentine vein surrounded by
141
a reaction front. It thus seems ideal for quantifying mass transfer and volume change
142
during reaction, since it allows for comparison between fully and partly reacted rock
143
domains.
144
The variations of the measured values are given with errors corresponding to
145
their standard deviation. The composition and the microtexture of the samples were
146
first characterized at the University of Lausanne with a field-emission scanning elec-
147
tron microscope (Tescan Mira II LMU) operating at 10 kV and 23 mm working dis-
148
tance and equipped with an Energy Dispersive Spectrometer from Oxford Instruments.
149
Serpentine and brucite minerals were identified by Raman microspectrometry at the
150
University of Lausanne with a Horiba LabRAM HR800 equipped with a 532.1 nm
151
laser.
152
2.1 X-ray microtomography
153
A 8 mm and 23 mm high core is scanned with a Skyscan-1173 at the University
154
of Lausanne. 2200 X-ray images are acquired with a 360◦ rotation of the samples at a
155
step size of 0.225◦ with a time of acquisition of 800 ms per frame. The acquisition is
156
performed at 70 kV and 114μA by accumulating 33 frames with a voxel size of 11.07
157
μm. The 3D images reconstructed from the transmission images are segmented and
158
analysed with Blob3D (Ketcham, 2005). We develop a Matlab code to display the
159
result of the image analysis and to extract the extent of reaction from the segmented
160
3D image. The extent of reaction is calculated based on the measured volumes of
161
olivine, spinel and brucite/serpentine in a 350μmwide sliding window parallel to the
162
orientation of the vein cross-cutting the sample. The similar attenuation of olivine
163
and Fe-brucite (identified with Scanning Electron Microscopy) makes segmentation
164
difficult. These phases can be distinguished based on their aspect ratio (e) calculated
165
as the length ratio of the minimum and maximum axes of an ellipsoid fitted to each
166
segmented grain (Figure S1 in Supplementary materials). Fe-brucite has a lowerethan
167
olivine due to its platy crystal form. Therefore, the olivine volume used to calculate
168
the extent of reaction displayed in Figure 1 is determined by only considering grains
169
with e > 0.35. We also exclude from olivine grain segmentation groups of voxels
170
with a volume smaller than 1.5.105 μm3. The radius of a sphere of this latter volume
171
corresponds to three times the voxel size. The extent of reaction estimated with X-
172
-10 -5 0 5 Distance from the vein center (mm) 97.5
98 98.5 99 99.5 100
Extent of reaction (%)
Figure 1. Extent of reaction as a function of distance from the serpentine vein center. The extent of reaction illustrated in this figure is a maximum; it is calculated based on X-ray microto- mography (see section 2.1 for the computational method).
ray microtomography is a maximum due to these two constraints on olivine grain
173
identification.
174
2.2 Electron microprobe analysis and X-ray maps
175
Point analyses and X-ray maps were acquired with a JEOL JXA-8530F Hyper-
176
Probe at the University of Lausanne. All the data were acquired at 15 keV. The X-ray
177
maps were acquired at 40 nA with a dwell time of 50 ms and a resolution of 8.5μm.
178
Spot analyses of olivine, serpentine/brucite mixture and magnetite were acquired in
179
the same region as the maps to produce internal standards for converting the count
180
number of the X-ray maps into composition in oxide mass and then in mole fraction.
181
A mineral phase was then attributed to each pixel of the map based on composition
182
criteria (Figure 2). Point analyses reveal that serpentine and brucite are intermixed at
183
a scale of less than 1μm. The proportion of serpentine and brucite at each pixel identi-
184
fied as a serpentine/brucite mixture was determined assuming (M g, F e)3Si2O5(OH)4
185
and (M g, F e) (OH)2 as structural formulas for serpentine and brucite, respectively:
186
XBr=nMg+nF e−32nSi
nMg+nF e−nSi (1)
wherenMg,nF eandnSidenote the molar content of Mg, Fe and Si, respectively,
187
andXBr is the mole fraction of brucite.
188
The Mg number of olivine (XMgOl) is determined by averaging the composition of
189
the pixels attributed to olivine. The Mg numbers of serpentine and brucite (XMgS and
190
XMgB, respectively) are determined by fitting for the serpentine/brucite mixture, the
191
Mg number (XMg) as a function ofXBr at each pixel with the following relationship:
192
XMg = XBr
3−2XBrXMgB+3 (1−XBr)
3−2XBr XMgS (2)
Finally, the Mg numbers calculated for each phase are used to determine their mo-
193
lar volume with a linear combination of the end-member molar volumes taking the fol-
194
0.76 0.8 0.84 0.88 0.92 0.96
Mg/(Mg+Fe)
A B
Figure 2. Compositional maps of sample BA4A-81-1-1-17 obtained from X-ray maps ac- quired with the microprobe. A: Mg/(Mg+Fe) ratio. The white arrow points towards the main serpentine vein. The black arrows point towards the core of a mesh preserved after complete serpentinisation. B: phases. The serpentine/brucite mixture is displayed in red, olivine in orange, spinel in blue and magnetite in green. The scale bar is 1 mm.
Figure 3. Standard deviation ofXBr as a function of the logarithm of the sampling size (sl) and the distance from the main vein (l).
lowing values: 43.79, 46.39, 24.63, 26.43, 108.5, and 115cm3.mol−1for forsterite, fay-
195
alite, Mg-brucite, Fe-brucite, Mg-serpentine and Fe2+-serpentine, respectively (Chichagov
196
et al., 2001).
197
To track changes in composition during reaction (export or import of aqueous
198
species), we calculate the mean composition of the serpentine/brucite mixture (XmBr)
199
by weighting the mole contents in brucite and serpentine at each pixel with their mo-
200
lar volume. Averages are calculated for all the pixels identified as serpentine/brucite
201
mixtures in the maps. We also estimateXmBr with a smaller sampling size to deter-
202
mine the scale at which compositional changes occur (scale of homogenization). The
203
side length of the square sampling box (sl) varies from the pixel size (8.5μm) to the
204
map width (5.6mm; Figure 3). We generate 3,000 randomly located sampling boxes
205
for eachsl. We then sort these boxes as a function of the distance to the main vein
206
network (l) to also determine the variation of the standard deviation ofXmBr with
207
sland l.
208
2.3 Estimate of volume change during reaction
209
In the sample studied here, primary olivine contains thin mineral platelets of
210
magnetite and clinopyroxene. Such inclusions are generally formed by exsolution at
211
high temperature during olivine cooling and oxidation (thus prior to serpentinisation;
212
Olivine Serp/Brc
Olivine Olivine
Serp/Brc
A B C ǻ9
9 = 50 % ǻ9
9 = 10 %
Figure 4. Sketchs illustrating the effect on inert markers displacement of solid volume change during the formation of a serpentine/brucite mixture (Serp/Brc). A: initial distribution of straight and parallel inert markers in olivine. B and C: predicted displacement of the inert markers for changes in volume of 10 % and 50 %, respectively. The three main effects of volume change during reaction for inert markers are: bending at the olivine/reaction products contact (red arrows), rotation (white arrows) and increase of the spacing (blue arrows).
(Moseley, 1984; A. Putnis, 1979)). They display a topotaxial intergrowth with olivine
213
leading to parallel oriented platelets along a preferred olivine crystallographic orien-
214
tation (Zhang et al., 1999; Ashworth, 2000). Baronnet and Boudier (pers. comm.)
215
first proposed use of such non-reactive and oriented markers to quantify solid volume
216
change during serpentinisation (Figure 4). Their calculation is based on the measure-
217
ment of the angle change between the olivine surface and an initially straight marker.
218
Such markers are also expected to rotate and to space out as a result of volume change
219
during reaction (Figure 4). Here we extend the approach of Baronnet and Boudier
220
(pers. comm.) by considering these latter effects to estimate solid volume change by
221
measuring the displacement of several platelets of magnetite and clinopyroxene.
222
First, we determine the deformation and composition of these platelets in three
223
dimensions with focused ion beam tomography at the CIME-EPFL laboratory (Lau-
224
sanne). We use a Zeiss Crossbeam 540 apparatus equipped with a field emission scan-
225
ning electron miscroscope. We erode layer by layer with an ion beam a 18x15x19μm
226
cuboid containing a deformed platelet and located at the surface of an olivine grain.
227
After each milling of a 10 nm thick layer, a back-scattered electron image is acquired
228
at an acceleration voltage of 1.7 kV with a current of 2.5 nA, a dwell time of 15 μs
229
per pixel and a pixel size of 10x10 nm. The stack of images is then aligned with the
230
FIJI image processing software by running the MultiStackReg plug-in (Th´evenaz et
231
al., 1998). We segment the images based on pixel intensity with a home-made Matlab
232
code by first applying a median filter and then using dilatation/erosion techniques.
233
As the inclusions are parallel in each olivine grain, we can compare orientations in
234
olivine and in fully serpentinised domains around each olivine. We therefore estimate
235
solid volume change in two dimensions by integrating the displacement of several
236
inclusions. We acquire optical photomicrographs of primary olivine surrounded by
237
serpentine + brucite. We select olivine grains with surface(s) oblique to the inclusion
238
orientations and with inclusions both inside the grain and in the secondary phases
239
(Figure 5 A). The olivine rim is mapped in two dimensions with a set of segments.
240
We then compute regularly spaced lines parallel to these segments (spacing dini =d
241
between the lines) to generate a mesh (Figure 5 B). The inclusions outside the olivine
242
A B
C D
Ol
Serp/Brc Inclusion
Figure 5. Steps for determining solid volume change during serpentinisation. We measure the misorientation of inclusions contained in olivine (Ol) initially and in the serpentine/brucite mixture (Serp/Brc) after reaction. A: an olivine grain surrounded by serpentinisation reaction products is first selected. The grain surface must be oblique to the inclusions. B: the inclu- sions are mapped (red lines outside the olivine grain and blue lines inside the olivine grain) and projected on a 2-D mesh generated from the olivine surface. C: the inclusions are deformed by modifying the mesh spacing in the direction normal to the olivine grain surface. D: the best model (green lines) is selected from a set of models with different mesh spacing by minimizing the angle between the needles inside and outside the olivine grain using the least-squares method.
Comparing initial mesh spacing to the spacing in the best model provides an estimate of solid volume change during reaction (see text for details).
grains are projected on the mesh which can be deformed by modifying d. The least-
243
squares method is finally applied to determine the value of d, dbest, for which the
244
orientation of the needles outside the olivine grain best fits the orientation of the
245
needles inside the olivine grain (Figure 5D). The surface area of the mesh at dbest
246
(Sbest) can finally be compared to the surface area of the mesh before deformation
247
(Sini) to estimate the change in surface area during reaction (ΔSS = SiniS−Sbest
best ). Thin
248
sections obtained from perpendicular sections of the sample display the same mesh
249
textures with reaction zones of approximately uniform size around the grains. We
250
therefore assume homogeneous volume change for 3D calculations in the following.
251
For spherical grains, ΔSS does not depend on the orientation of the 2D thin section
252
used for measurement. The change in surface area can be converted in a change in
253
solid volume (ΔVV) with the following equation:
254
ΔV V =
ΔS S
1−(1−ξ)2/3
+ 1 3/2
−1
ξ (3)
withξthe local extent of reaction. ξcan be estimated from 2D images as:
255
ξ= 1− Sol
Sbest 3/2
(4)
withSol the surface area of the olivine grain. For spherical grains, Equation 4
256
only provides the actual extent of reaction if the grains are cut through their centers.
257
It overestimates ξ otherwise. Therefore, ΔVV determined with ξ from equation 4 is
258
a maximum. The limit of equation 3 as ξ tends towards zero is ΔVV = ΔSS, which
259
provides a minimum estimate for the change in volume.
260
2.4 Laser-ablation inductively coupled plasma mass spectrometry (LA-
261
ICPMS)
262
The chemical data are acquired with a sector-field Element XR inductively cou-
263
pled plasma mass spectrometer (ICP-MS) coupled with an Atlex 193 nm ArF excimer
264
laser housed inside an Australian Scientific Instruments (ASI) RESOlution system (in-
265
cluding an S155 dual volume sample cell) at the University of Lausanne. We use two
266
strategies to acquire the data: spot mode and continuous scanning. Both datasets are
267
acquired along the direction perpendicular to the main vein network. For each ele-
268
ment, one isotope is measured in a peak-hopping mode. The on-sample laser energy
269
densities are approximately 10 J.cm−2 with repetition rates at 10 Hz. The laser beam
270
shape of the spot mode is rectangular with a size of 100 x 25 μm. For continuous
271
analyses, the laser beam is a disk with a diameter of 100μmand the scanning rate is
272
10μm.s−1.
273
To reduce the results, glass standard BCR-2G was analyzed along with the sam-
274
ples to be used as an external calibration standard, while silica (SiO2 = 33 wt.%) is
275
used as an internal standard using LAMTRACE (Jackson, 2008) and Iolite (Version
276
3.11, The University of Melbourne) (Paton et al., 2011) software. An interval of 1.58
277
s is selected in the data reduction for continuous scanning. For smoothing and clarity,
278
we use a moving average on three points for plotting the results. The raw geochemical
279
data are given in the Supplementary Materials.
280
3 Results
281
3.1 Mineralogical distribution
282
The sample is cross-cut by a ∼5mm wide anastomosing network of serpentine
283
veins (referred to as ”black veins” in the core description (Kelemen et al., n.d.)). De-
284
spite complete serpentinisation in the vicinity of these veins, the typical mesh texture
285
of serpentinised peridotites is preserved (Figure 2 A). Approximately 10μmwide ser-
286
pentine veins surround mesh cores composed of a concentric alternation of brucite-rich
287
and serpentine-rich layers. Raman spectroscopy reveals that the serpentine minerals
288
are both chrysotile (main peak at 3699 cm−1and shoulder at 3689 cm−1) and lizardite
289
(main peak at 3684 cm−1 and secondary peak at 3705 cm−1) in the vein network. In
290
the mesh cores, lizardite is intermixed on the sub-micrometer scale with brucite (main
291
peak at 3636 cm−1 and secondary peak at 3652 cm−1). This latter peak corresponds
292
to the peak expected for pure brucite (M g(OH)2; (Speziale et al., 2005)). The shifting
293
towards lower frequencies of the main brucite peak suggests the presence ofF e(OH)2
294
(Speziale et al., 2005). Magnetite is only found as sub-micron sized grains dissem-
295
inated in serpentine and brucite. It mainly occurs in the wider serpentine veins of
296
the anastomosing network. Up to∼2 mm wide chromites are found throughout the
297
samples. They can react at their margin with the development of a 2μmwide reaction
298
zone mainly composed of magnetite. We also observe minor ∼50 μm wide clinopy-
299
roxenes in the rock matrix. They do not display evidence for reaction in agreement
300
with thermodynamic calculations (Klein et al., 2009).
301
≥3mmaway from the anastomosing vein network, olivine grains are found in the
302
mesh core (Figure 2). The transition from olivine-free to olivine-bearing serpentinised
303
peridotites occurs at approximately 4 mm from the main serpentine vein (Figure 1).
304
The extent of reaction is generally high with a minimum estimated at 97 % with X-ray
305
microtomography and at 80 % with X-ray maps acquired with the microprobe.
306
Olivine grains commonly contain ∼ 1μm wide, parallel inclusions (Figure 6).
307
We use these platelets to estimate solid volume change during serpentinisation in
308
section 3.4. Energy dispersive spectroscopy reveals that the inclusions are intergrowths
309
of clinopyroxene and magnetite, containing minor concentrations of Cr. FIB-SEM
310
nanotomography shows that the platelets have a symplectic texture (Figure 7 A). Its
311
composition is similar when it is included in the olivine or in the serpentine/brucite
312
mixture and thus the composition of both clinopyroxene and magnetite appear to
313
be largely unaffected by serpentinisation. The only effect of reaction appears to be
314
the generation of discontinuities in the platelet separating∼ 1 μm long segments of
315
clinopyroxene. The presence of traces of magnetite in the discontinuities as well as
316
their orientation in parallel to other magnetite grains suggest that the discontinuities
317
were formerly filled with magnetite which reacted during serpentinisation.
318
3.2 Mineralogical composition
319
Point EPMA analyses and X-ray maps indicate homogeneous iron distribution
320
in olivine with a Mg number (MgMg+F e mole fraction) of 0.895±0.011. NiO content in
321
olivine ranges from 0.25 to 0.33 wt.%.
322
For the brucite/serpentine mixture, XMg varies as a function of XBr along a
323
mixing line defined by equation 2 (Figure 8). The fit of the data leads to XMgS =
324
0.956±0.022 andXMgB = 0.724±0.022. XBr averaged over the entire mapped area
325
is 0.515±0.002. The mean value ofXMg in the serpentine/brucite mixture is 0.896.
326
Figure 3 gives the standard deviation ofXBr as a function of the sampling size (sl)
327
and the distance to the main serpentine vein (l). It thus provides an estimate of the
328
scale and extent of mass transfer in the sample. The standard deviation decreases
329
withsl and is high for sl below approximately 500 μm. Below this sl, the standard
330
20 μm
Ol Serp/Brc
Mag-Cpx intergrowth Mag
Figure 6. Back-scattered electron image of an olivine grain (Ol) reacted at its border to form a serpentine/brucite mixture (Serp/Brc). Olivine contains parallel inclusions of magnetite(Mag)- clinopyroxene(Cpx) intergrowths which are displaced in the serpentine/brucite mixture.
deviation highly depends onl and is 2 to 3 times higher in the first 2 mm near the
331
main serpentine vein than further away where olivine also occurs in the matrix (Figure
332
3).
333
3.3 Chemical composition
334
Compositional profiles, measured from both X-ray maps and LA-ICPMS, display
335
constant contents of FeO, MgO and SiO2 at a scale larger than several hundreds of
336
micrometers which is also the scale at which constant brucite contents are measured.
337
Figure 2 A reveals variations of more than 15 % in the Mg/(Mg+Fe) ratio at a smaller
338
scale and near the main serpentine vein. This is associated with brucite and serpentine
339
segregation in the mesh cores.
340
Spatial variations in Al2O3 and Cr2O3 are mainly associated with the primary
341
distribution of chromite in the sample. From a trace element point of view, there are
342
three distinct correlations between the abundances and the distances from the main
343
vein. Concentrations of fluid-mobile elements such as Na, Ca, Rb, Sr and Ba gradually
344
decrease with increasing perpendicular distance away from the main vein. Some spikes
345
in Ca and Na profiles may be due to the preservation of fine clinopyroxene grains or
346
inclusions (Figure 9).
347
Some elements display constant contents, such as B, Co, Ni, Cu, Zn and Yb
348
(Figure 10). Most of them are transition metal or heavy rare earth elements (HREEs).
349
The light-REEs, Nb, Pb and U are very low and their variation is difficult to
350
quantify (for example, the contents of Ce mostly cluster around 1 ppb, ignoring the
351
spikes possibly related to clinopyroxene). U is generally less than 1 ppb and provides
352
a possible line of evidence for the absence of carbonate formation. Finally, Al, Sc, Ti,
353
V and Sr show periodic increases at a length scale of 5 mm (Figure 11). The bulges
354
formed by these increases are rather symmetric with a flat plateau at the highest
355
concentration surrounded by sharp flanks where the concentration decreases.
356
Cpx
Ol Serp/Brc
A B
Figure 7. FIB-SEM nanotomography of a platelet which has partly experienced serpentinisa- tion. A: result of segmentation. The contact between olivine (Ol) at the bottom and the serpen- tine/brucite mixture (Serp/Brc) at the top is displayed with the yellow surface. The platelet is composed of clinopyroxene (Cpx; green) and magnetite (Mag; gray). It is displaced along cracks at some magnetite/clinopyroxene contacts in the serpentine/brucite mixture (red arrow). It is also discontinuous at its top (white arrows). B: angle between a reference orientation and the normal to the platelet surface. The reference orientation is chosen as the normal to the platelet when it is still included in the olivine grain (bottom part).
0 0.2 0.4 0.6 0.8 1 Brucite molar fraction (X
Br) 0.75
0.8 0.85 0.9 0.95
Mg number (X Mg)
2.4 2.6 2.8 3 3.2 3.4 3.6
Figure 8. Mg/(Mg+Fe) ratio (Mg number) of the serpentine/brucite mixture as a function of the brucite molar fraction (XBr) calculated with equation 1. Data are obtained at each pixel identified as a serpentine/brucite mixture. The density of data points is contoured (arbitrary units). The black line corresponds to the best fit of the data obtained with equation 2. 95 % and 68 % confidence bands are displayed with red and pink plain lines, respectively. We used the 68
% confidence bands to estimate the uncertainty onXMgS andXMgB.
Ol relics appear
Fine Cpx Fine Cpx
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
Sr ppm
89 7 6 5 4 3
'LVWDQFHIURP9HLQȝP
Rb ppb 100
101
10-1
Ca 103 ppm
1.0 2.0 3.0
0.7
103 104 Ca ppm
Na ppm
30 40 50
20
10
Figure 9. Concentration of water soluble elements (Na, Ca, Rb and Sr) as a function of the distance to the main vein network. The profiles are acquired perpendicular to the main vein
Ol relics appear 1
10 50
B ppm
101 102 103
Co ppmNi 103 ppm
1 2 3 4
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
Yb ppm
'LVWDQFHIURP9HLQȝP 10-1
10-2
Figure 10. Concentration of B, Co, Ni and Cr as a function of the distance to the main vein network. The profiles are acquired perpendicular to the main vein network. The first appearance
Original Ol grain Original Ol grain
10-1
V ppm
100
Ti ppm
30 25 20 15
10
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
101 102
Cr ppm
'LVWDQFHIURP9HLQȝP 100
Al ppm
102
101
Figure 11. Concentration of Al, Ti, V and Sc as a function of the distance from the main vein network. The profiles are acquired perpendicular to the main vein network. The first ap-
Figure 12. Change in solid volume (ΔVV) measured in sample BA4A-81-1-1-17. Minimum (blue bars) and maximum (orange bars) values of ΔVV are estimated for each photomicrograph.
The procedure for calculating these minimum and maximum bonds is described in the Method section. The mean (bold lines)±the standard deviation (thin lines) are displayed with dashed and plain lines for the minimum and maximum values of ΔVV, respectively. The bold dotted line corresponds to the change in volume calculated when considering no mass transfer during reaction.
3.4 Solid volume change
357
We used the misorientation of clinopyroxene/magnetite intergrowths initially in-
358
cluded in olivine to quantify volume change during serpentinisation. The platelet
359
analysed with FIB-SEM tomography is planar when it is included in olivine with less
360
than 5◦ of misorientation (Figure 7 B). It progressively bends when it is included in
361
the serpentine/brucite mixture (Figure 7 A and B) with misorientation values of more
362
than 15 ◦ the further away from the olivine surface. Bending preferentially occurs
363
along small cracks that are mostly located at clinopyroxene/magnetite contacts. The
364
acute angle between the platelet and the olivine surface increases with this bending.
365
This indicates a positive change in solid volume during reaction. Systematic differ-
366
ences in the orientation of ensembles of inclusions, between those hosted in olivine and
367
those hosted in surrounding zones of serpentine + brucite replacing olivine, are also
368
observed in scanning electron images (Figure 6).
369
To extract quantitative estimates of solid volume change during reaction, we use
370
the procedure described in section 2.3. We analyse 64 optical photomicrographs of
371
serpentinised olivine. The results of inclusion reorientation are summarized in Figure
372
12 and provided for each photomicrograph in the Supplementary materials. We mea-
373
sure positive ΔVV from 8 to 168 %. ΔVV minimum and maximum bonds have mean
374
values of 59 ± 29 and 74 ± 36 %, respectively (see section 2.3 for the definition of
375
these bounds).
376
4 Discussion and conclusion
377
The sample studied here appears to be ideal for studying mass transfer and
378
volume change during reaction. It comes from depth (215 m) and has thus experienced
379
relatively limited surface alteration; for example, we do not observe carbonates or
380
evidence for carbonation in U, Sr or Ca datasets consistent with the results of shipboard
381
core description (Kelemen et al., n.d.). Moreover, the NiO content of the remaining
382
olivine is below 0.32 wt. % indicating that the protolith is probably a dunite (Suhr,
383
1999).
384
Serpentinisation requires transport of ∼ 150 g of water per kilogram into an
385
initially impermeable peridotite. The sample is crossed by an anastomosing vein per-
386
pendicular to which a gradient in extent of reaction is measured (Figures 2 and 1). This
387
suggests that water migrated from the main serpentine vein into the peridotite matrix.
388
The observed mesh texture on the wall of the main vein indicates that a fracture net-
389
work with a typical spacing of 100μmpromoted fluid transport inside the peridotite.
390
Such an organised crack network may have formed by reaction-induced fracturing
391
(Jamtveit et al., 2009; Kelemen & Hirth, 2012; Pl¨umper et al., 2012; Malvoisin et al.,
392
2017) and/or thermal cracking (Demartin et al., 2004; Boudier et al., 2005). The origin
393
of the main serpentine vein is more difficult to constrain since tectonic processes could
394
also promote fracturing (Roum´ejon & Cannat, 2014). The fluid transport direction
395
is confirmed with compositional profiles in trace elements. A decrease in fluid-mobile
396
elements (Na, Ca, Sr, Rb and Ba) is indeed observed from and perpendicular to the
397
main vein. The history of the Oman ophiolite is complex with four main events dur-
398
ing which serpentinisation probably occurred: hydrothermal circulation at mid-ocean
399
ridges, interaction with metamorphic fluids during obduction (Searle & Malpas, 1980),
400
Cretaceous weathering of the ophiolite, followed by a marine transgression (de Obeso
401
& Kelemen, 2018), and modern ongoing low temperature serpentinisation through in-
402
teraction with rainwater (Neal & Stanger, 1983; Kelemen & Matter, 2008; Kelemen et
403
al., 2011). Fluids collected at surface seeps or in wells in the Oman ophiolite generally
404
display high Na, Cl, Sr and Ca concentrations (Chavagnac et al., 2013; Paukert et al.,
405
2012; Canovas et al., 2017; Paukert Vankeuren et al., 2019), the same as the elements
406
that display a progressive decrease from the main vein in the sample studied here.
407
Modern water sources and seawater both contain all these components at relatively
408
high concentration (Millero et al., 2008; Paukert Vankeuren et al., 2019). The time of
409
serpentinisation can thus not be determined with the data presented here. The enrich-
410
ment in fluid-mobile components in the Oman ophiolite fluids could be secondary and
411
acquired through dissolution of fluid-mobile elements previously trapped in gabbro or
412
in peridotites serpentinised on the seafloor (Miller et al., 2016). Meanwhile, the Sr
413
isotope ratios in peridotite-hosted carbonates, water, and leachates are substantially
414
more radiogenic than Cretaceous seawater (Kelemen et al., 2011). Several sources may
415
thus contribute to the modern composition of the fluids collected in Oman.
416
The standard deviation of the Mg/(Mg+Fe) ratio in the serpentine/brucite mix-
417
ture is three times higher for a sampling size below 100-500 μm than above (Figure
418
3). This indicates that major elements are redistributed at a scale below 100-500μm
419
corresponding to the mesh scale. This redistribution increases near the main vein
420
(Figure 3) indicating a larger scale segregation of brucite and serpentine. There is
421
thus mass transport probably associated with further fluid-rock interaction at a scale
422
of less than several hundreds of micrometers. This local mass transfer is probably the
423
incipient stage of the mass transport responsible for the change in chemical compo-
424
sition observed in serpentinised peridotites with a decrease of the M gO/SiO2 ratio
425
during more extensive alteration (Snow & Dick, 1995; Niu, 2004; Monnier et al., 2006;
426
de Obeso & Kelemen, 2018).
427
We do not measure spatial changes in major element composition at the cen-
428
timeter scale in microprobe and LA-ICPMS data. The Mg/(Mg+Fe) ratio in the
429
serpentine/brucite mixture is 0.896, similar to the one measured in the remaining
430
olivine grains (0.895±0.011). Brucite and serpentine display an equimolar distribu-
431
tion in the serpentine/brucite mixture (XBr = 0.515±0.002) and magnetite is rare
432
in the investigated area (Figure 2 B). This suggests that the initial (Mg+Fe)/Si ra-
433
tios of olivine are conserved during reaction. Excluding trace elements and at a scale
434
higher than several hundreds of micrometers, serpentinisation in the studied sample
435
corresponds to the isochemical hydration of olivine:
436
(M g0.90F e0.10)2SiO4+3H2O= (M g0.96F e0.04)3Si2O5(OH)4+(M g0.72F e0.28) (OH)2 (5)
where the compositions of serpentine and brucite are retrieved from microprobe
437
data with equation 2. The Mg numbers of brucite and serpentine can theoretically pro-
438
vide constraints on temperature and water to rock ratio. However, the uncertainties
439
on the thermodynamic parameters of Fe-brucite and Fe-serpentine preclude precise
440
quantitative estimates. The XMgB measured here is at the low end of the range for
441
thermodynamic predictions (Klein et al., 2009, 2013). The thermodynamic calcula-
442
tions indicate that more iron is incorporated in brucite at low temperature (T<100
443
◦C) and low water to rock ratio (< 0.5). The scarcity in magnetite in the sample is
444
also compatible with such conditions.
445
Al, Ti, V, Sc and Cr concentrations periodically vary at a typical length scale
446
of several millimetres without relationship to the extent of serpentinisation (Figure
447
11). Measured compositional profiles recall the diffusion profiles for the same ele-
448
ments measured in unaltered olivine-bearing xenoliths at a length scale typical of the
449
olivine crystal size in unaltered peridotites (Tollan et al., 2015). Therefore, we inter-
450
pret the measured composition as reflecting the primary olivine composition acquired
451
through diffusion or growth at high temperature. Preserving olivine primary composi-
452
tion requires that Al, Ti, V, Sc and Cr are immobile during the whole serpentinisation
453
process.
454
Our study of major and trace elements behaviour during serpentinisation indi-
455
cates very limited mass transfer except for elements associated with the fluid (Na, Ca,
456
Sr, Rb and Ba). In contrast, several studies have reported a decrease in MgO/SiO2in
457
serpentinised peridotites, compared to their mantle peridotite protoliths (Snow & Dick,
458
1995; Niu, 2004; Monnier et al., 2006; Malvoisin, 2015; de Obeso & Kelemen, 2018).
459
This decrease is either related to Mg loss during alteration on the seafloor (Snow &
460
Dick, 1995) or to Si metasomatism (Paulick et al., 2006; de Obeso & Kelemen, 2018).
461
Thermodynamic calculations predict a significant decrease in the M gO/SiO2 ratio
462
only when consideringSiO2gain. Mg loss is not predicted at equilibrium even at high
463
water to rock ratio (105) (Malvoisin, 2015). However, high-resolution tomography in-
464
dicates brucite dissolution and porosity formation in abyssal serpentinised peridotites
465
exposed on the seafloor (J¨ons et al., 2017). Mg loss would then be associated with
466
kinetic effects limiting the precipitation of Mg-bearing phases (e.g. carbonates) during
467
fluid flow. Brucite is not commonly observed in the samples dredged on the seafloor
468
and studying its fate in the cores drilled in Oman can thus provide insights into the
469
decrease in theM gO/SiO2 ratio measured in abyssal peridotite. As brucite contains
470
approximately 30 % iron, its reaction is expected to produce iron-bearing phases such
471
as magnetite while magnesium is either leached from the system with the fluid (Snow
472
& Dick, 1995) or combined with dissolved silica to form serpentine (Beard et al., 2009).
473
In our study, we observe tiny magnetite grains (<1μm) that could be associated with
474
incipient brucite alteration at low temperature (Figure 2). Magnetite production is
475
also expected to produce hydrogen according to experiments on samples from Oman
476
(Miller et al., 2017).
477