Download Serpentinite - Elements Magazine

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Plate tectonics wikipedia , lookup

Oceanic trench wikipedia , lookup

Mantle plume wikipedia , lookup

Transcript
Serpentinite:
What, Why, Where?
Bernard W. Evans1, Keiko Hattori2, and Alain Baronnet3
1811-5209/13/0009-99$2.50
R
DOI: 10.2113/gselements.9.2.99
ock-forming serpentine minerals form flat, cylindrical, and corrugated
crystal microstructures, which reflect energetically efficient layering
of alternate tetrahedral and octahedral sheets. Serpentinization of
peridotite involves internal buffering of the pore fluid, reduction of oxygen
fugacity, and partial oxidation of Fe2+ to Fe3+. Sluggish MgFe diffusion in
olivine causes precipitation of magnetite and release of H2 . The tectonic
environment of the serpentinization process dictates the abundance of fluidmobile elements in serpentinites. Similar enrichment patterns of fluid-mobile
elements in mantle-wedge serpentinites and arc magmas suggest a linkage
between the dehydration of serpentinite and arc magmatism.
The Serpentine Minerals
The serpentine minerals form
under a wide range of temperatures,
including Earth surface conditions
and hot hydrothermal temperatures. They have the approximate
formula Mg3Si2O5 (OH) 4. The basic
structural unit is a polar layer
0.72 nm thick (FIG. 1). A Mg-rich
trioctahedral sheet is tightly linked
on one side to a single tetrahedral
silicate sheet, notwithstanding
the 3–5% larger lateral lattice
KEYWORDS : serpentinization, antigorite, chrysotile, lizardite,
dimensions of the octahedral
buffering, subduction zone, recycling
sheet (Wicks and Whittaker 1975).
The second level of organization
into different serpentine species
INTRODUCTION
originates partly to compensate
the intralayer stress due to this dimensional misfit. Good
This article provides answers to questions that readers
might have with respect to some fundamental aspects of compensation results in a nearly constant layer curvature, with the larger octahedral sheet on the convex side.
serpentinization, our understanding of which has evolved
However, such curvature weakens the H-bonding between
over the last 20 years. It addresses what serpentinites are
the layers. H-bonding tries to maintain flat layers, but this
made of, why and how they formed the way they did, and
competes with the requirements of misfit compensation. As
where globally we fi nd them, including how their location
a result, the layers are locally either curved or flat.
influences their geochemistry.
WHAT?
Serpentinites are rocks consisting mostly of serpentinegroup minerals. They form by the hydration of olivinerich ultramafic rocks at relatively low temperatures. The
primary Mg-rich minerals—olivine, orthopyroxene, and
clinopyroxene—are replaced by 1:1 layered Mg silicate
hydrated minerals (the serpentines), magnetite, and in
some cases brucite [Mg(OH) 2 ]. The typical combination
of mineral products results in a fi ne-grained, dark green
to black rock (commonly crossed by veins of chrysotile or
calcite), whose density is significantly lower than that of
the primary peridotite, and whose magnetic susceptibility
is up to nearly two orders of magnitude greater. A comprehensive coverage of the serpentine minerals is given in
Deer et al. (2009).
1 Department of Earth and Space Sciences
University of Washington, Seattle, WA 98195-1310, USA
E-mail: [email protected]
2 Department of Earth Sciences, University of Ottawa
Ottawa, ON K1N 6N5, Canada
E-mail: [email protected]
Combined ball-and-stick and polyhedral atomic model
of two superimposed building layers of lizardite
serpentine. Oc = octahedral sheet; Te = tetrahedral sheet. Color
codes of atoms are on the upper left.
FIGURE 1
3 Centre Interdisciplinaire de Nanosciences de Marseille
CNRS, Aix-Marseille Université, 13288, Marseille, France
E-mail: [email protected]
E LEMENTS , V OL . 9,
PP.
99–106
99
A PR IL 2013
A
B
C
(A) High-resolution transmission electron microscopy
(TEM) micrograph of the flat lizardite microstructure
in which the basic structural unit is 0.72 nm thick. Inserts: enlarged
part of a polyhedral presentation of the lizardite microstructure at
the same scale. The microscope image shows only Mg and Si atoms
without O atoms. The small distance between Mg atoms makes
them appear as a continuous dark band, and each pair of Si atoms
shows up as a dark blob (right insert). (B) High-resolution TEM
image of the cross section of a chrysotile nanotube with a hollow
core. Inset: polyhedral model of the tube-in-tube cylindrical structure. Bar scale = 10 nm. (C) High-resolution TEM image of the wavy
structure of antigorite along its b-axis. Insert: approximate structure
represented as a polyhedral model at the same scale showing the
rows of Mg and Si atoms.
The three common serpentine minerals are lizardite, chrysotile, and antigorite. Lizardite has a flat crystal structure
(FIG. 2A) with the correct geometry of interlayer H-bonds.
This structure is favored by coupled substitutions of Al and
Fe3+ for VI Mg and IVSi. Owing to the many different ways
to stack lizardite layers, different polytypes are possible
(Bailey 1988). The most common of these is the one-layer
trigonal (1T) polytype, where successive layers are directly
superimposed without any lateral shift. The typical habit
of lizardite is a triangular plate truncated at each corner.
Lizardite is the principal mineral in the “mesh” cells that
commonly pseudomorph olivine (FIG. 3A, B).
reach several centimeters in length. Fibers are achiral, being
rolled up around the a-axis (normal chrysotile) or, less
commonly, around the b-axis (parachrysotile). Cylindrical
polytypes exist for chrysotile, resulting from tube-in-tube
layer glides having circumferential and/or axial components. Across the tube wall, there is a radial gradient of
stored elastic energy (Baronnet and Devouard 1996). This
tubular structure as a whole obeys five-fold symmetry
(Cressey and Whittaker 1993) and is a one-dimensional
crystal in that it is periodic only along its axis. The core
of the fibers may be empty or fi lled by amorphous matter.
Chrysotile is poor in Al and total Fe and occurs mainly
as a fi lling of fractures that crosscut serpentinite (FIG.
3C). Its parallel fiber orientation (cross, oblique, or slip) is
mostly determined by the repeated opening of fractures
(FIG. 3C). Chrysotile is the main constituent of commercial
FIGURE 2
Chrysotile forms multiwall nanotubes or nanoscrolls
(FIG. 2B). An undeformed chrysotile fiber is <100 nm in
outer diameter and ≥4 nm in inner diameter, and it may
A
B
C
D
E LEMENTS
Photomicrographs
of serpentine
minerals. (A) Lizardite pseudomorphing olivine with hourglass
and mesh textures in planepolarized light. The sample was
collected in the Northern
Serpentinite Belt in the
Dominican Republic. COURTESY OF
B ENOIT SAUMUR (B) The same as (A)
under crossed polarizers.
(C) Cross section of a chrysotile
“crack-seal” vein with parallel
growth banding (crossed polarizers). Serpentinite from the
Jeffrey Mine, Québec, Canada.
(D) Tabular, interlocking texture
of antigorite (crossed polarizers).
Serpentinite from the Piémont
zone, Western Alps, Italy.
FIGURE 3
100
A PR IL 2013
T
Fo + H
2
LA
O
Atg +
Brc
B
6
100
2O
Tlc +
H
Atg
Atg
Liz + Tlc
2
FA T
200
300
400
Fo +
4
B
Atg + Brc
Liz
Pressure (kbar)
8
500
o
Temperature ( C)
High-resolution TEM image of the cross section of
a fi fteen-sectored fiber of polygonal serpentine
surrounded by chrysotile fibers. Flat lizardite-like layers inside
sectors are linked by curved layers at their boundaries.
asbestos, a material with outstanding electric, thermal, and
phonic insulation properties, but its fi ne dust is thought to
be deleterious to human health (Fubini and Fenoglio 2007).
Antigorite displays curved, wavy layers similar to Roman
tiles on a roof (FIG. 2 C). The octahedral sheet is continuous and wavy, whereas the tetrahedral sheet undergoes
periodic reversals along the a-axis so that it connects to the
concave half-waves of adjacent octahedral sheets (Capitani
and Mellini 2004). The reversals allow serpentine layers
to be bound through strong, mainly covalent Si–O bonds.
This explains the lack of easy cleavage and the enhanced
hardness and seismic velocities of antigorite compared to
those of lizardite and chrysotile. The antigorite structure
is an as-grown modulated structure, and the wavelength
along the a-axis can be indicative of the composition and
P–T conditions of the metamorphic rocks containing the
mineral (Mellini et al. 1987). Antigorite has a triperiodic
unit cell, that is, it is a “normal crystal” like lizardite.
Antigorite contains Al and Fe3+ and is enriched in Si due
to the “talc-like” chemical contribution of its reversals.
Antigorite is considered to be a high-temperature phase
of serpentine. It commonly occurs in partially recrystallized and metamorphosed serpentinite. There, antigorite
commonly forms interlocking, elongated, blade-shaped
crystals that overprint earlier-formed serpentine (FIG. 3D).
In addition to these three familiar members, some microstructures have been recognized only recently with the
transmission electron microscope. Polygonal serpentine
reported by Mitchell and Putnis (1988) contains curved
and flat parts along individual layers, resulting in a microstructure with 15 (FIG. 4) or 30 sectors. It may be viewed
as a less metastable microstructure than that of chrysotile (Mitchell and Putnis 1988). Polyhedral serpentine as
facetted, onion-like nanospherules (Baronnet et al. 2007)
form from altered pyroxene in Al- and Fe3+ -bearing media
at T < 300 °C and P H2O = 0.7 kbar. Conical serpentine
nanofibers have recently been recognized in optically
isotropic veins of serpentinite (Andreani et al. 2007). They
are all scrolls with apical angles of <30°. Cones exhibit a
further loss of control of the layer curvature with respect to
cylindrical chrysotile, having an evolving curvature along
their length. This suggests that cone-shaped serpentine
may be even more metastable than chrysotile.
E LEMENTS
A possible phase diagram for the MgO–SiO2–H2O
system. To illustrate the uncertainty, two steep
H2O-conserved reactions are shown as wide gray bands (modified
after Evans 2004). Mineral compatibilities in the divariant fields are
shown on the MgO–SiO2 binary line after projection from H2O.
Abbreviations: A, Atg = antigorite; B, Brc = brucite; F, Fo = forsterite; L, Liz = lizardite; T, Tlc = talc.
FIGURE 5
FIGURE 4
Stability and Metastability of Serpentine
In the MgO–SiO2 –H 2O system, lizardite is more stable
than chrysotile at 200 and 300 °C and 0.7 kbar (Grauby
et al. 1998). Laboratory syntheses suggest that chrysotile
might be more stable than lizardite above 400 °C, and
chrysotile is certainly more common in recrystallized or
tectonized serpentinite (O’Hanley 1996). However, antigorite grows from lizardite and chrysotile with increasing
grade of metamorphism at T above about 320 °C. A calculated transition from lizardite to antigorite + brucite with
a steep negative dP/dT slope takes place at approximately
300 °C (FIG. 5). This renders chrysotile entirely metastable
at any temperature. This conclusion is ironic given that,
for many years, chrysotile was the only Mg serpentine
with listed thermodynamic properties—notwithstanding
the uncertainties caused by elastic strain-energy differences
among the layers in the structure.
Reversed laboratory experiments have shown that antigorite
breaks down at 640 °C and 2 GPa in the pure system and
at ≈700 °C when Al saturated; these conditions correspond
to a change in the dP/dT slope of the breakdown curve.
At this point, the reaction volume changes sign, although
the solid volume change stays negative. The reaction at
high pressure produces olivine + orthopyroxene and large
quantities of H 2O; consequently it is of great interest in
the context of subduction zone seismicity and arc magma
generation via fluid-fluxed melting. The P–T phase diagram
(FIG. 5) shows that the reaction terminal to the stability
field of forsterite + H2O at low T yields antigorite + brucite
(at 380 °C and 2 kbar, for example). This implies that the
almost universal formation of lizardite from olivine during
serpentinization signifies (1) major (down-T) overstepping
of the reaction, (2) a failure of antigorite to nucleate and
grow, or (3) errors in the thermodynamic data. In natural
systems with Al, Cr, Fe2+, and Fe3+, we have to contend
with possible reversals from the stability order of serpentines in the simple system. This perhaps explains the many
confl icting field observations in the literature of mutual
replacements among the serpentine minerals. The Gibbs
free energy, entropy, and volume differences among the
101
A PR IL 2013
1.0
in nature may be trivial despite the major decline in log
fO2. By contrast, the culminating serpentinization reaction
involves substantial progress at temperatures probably
several tens of degrees Celsius (for kinetic reasons) below
the stable isobaric invariant point where brucite joins the
mineral assemblage. The “paradox” is thus explained by
the internal (buffer) control of Fe oxidation, instead of
exposure to an external oxidizing fluid.
Fe3+/total Fe
0.8
0.6
Low Silica Activities
0.4
0.2
0.0
0
2
4
6
8
10
12
14
16
H2O
Whole-rock Fe3+/total Fe ratio in natural serpentinites
as a function of their H2O content (i.e. degree of
serpentinization) (MODIFIED AFTER EVANS 2008)
An isobaric phase diagram with aSiO2 shows that very low
silica activities also attend serpentinization (Frost and Beard
2007), as witnessed by the formation of minerals such as
brucite, awaruite, hydrogarnet, and, in rare cases, diaspore,
corundum, and perovskite. Again, the drive towards low
values of aSiO2 is the persistence to low T of MgFe olivine
together with aqueous pore fluid. The buffering reaction
in the MgO–SiO2 –H2O system is analogous to the previous
reaction:
FIGURE 6
3 Mg olivine + 4 H 2O + SiO2 aq = 2 Mg serpentine .
serpentine minerals are small, and so impurities matter.
There seems to be some consensus that chrysotile growth
is favored in fluid-fi lled voids.
WHY?
It has long been clear that the process of serpentinization
involves some oxidation of iron. At the same time, the pore
fluid becomes more reducing. This seems like a paradox,
but it isn’t. We try to clarify this below.
The Oxidation–Reduction Paradox
H 2 O gains pervasive access to peridotite along joints,
fractures, and grain boundaries. These openings may be
thermal in origin (cooling), tectonic, or hierarchical (due
to volume expansion). The hydration of olivine is accompanied by the oxidation of Fe2+ to Fe3+ and the release of
H2 gas:
MgFe olivine + H 2O → MgFe serpentine + magnetite
+ MgFe brucite + H2 .
Most of the ferric iron resides in magnetite, but a sizeable
proportion is also in the serpentine, which in most
instances is the lizardite variety. Typically 50 to 90% of
the Fe in lizardite is Fe3+, and it is for this reason that
advanced serpentinization (>10 wt% whole-rock H2O) is
commonly associated with whole-rock Fe3+/Fe2+ ratios in
many cases in excess of 2, which is larger than the ratio
in magnetite (FIG. 6). The percent molar Mg/(Mg + total
Fe) ratio (Mg#) of the serpentine is ordinarily greater than
that of the olivine.
Paradoxically, the oxidizing serpentinization reaction takes
place at very low oxygen fugacity and correspondingly high
hydrogen partial pressures. These conditions, which allow
the precipitation of native metals, such as awaruite (Ni3Fe)
and wairauite (CoFe), are reached by means of down-T (e.g.
down to ≈200 °C) mineral buffering of the grain-boundary
fluid. The principal driver of this buffer trend is the isobarically univariant redox reaction written conventionally as:
MgFe olivine + H 2O + O2 + SiO2 aq =
MgFe serpentine + magnetite .
The dlog fO2 versus T slope of this reaction on the phase
diagram is much steeper than the familiar curves for
experimental redox-buffer assemblages such as FMQ
(fayalite-magnetite-quartz) and MI (magnetite-iron). With
falling temperature, the reaction can proceed by the release
of H2 or by the addition of O2, but the extent of the reaction
E LEMENTS
In both cases, higher-T buffering reactions involving talc
rather than serpentine could also have been operative.
The Pore Fluid
The predictive success of the isobaric T–activity diagrams
suggests a general model for serpentinization involving
internal (mineral assemblage) control of the geochemistry
of the aqueous pore fluid. The unique fluid chemistry created
during serpentinization (high Ca, low Mg, high pH, high
f H2, and low aSiO2 ) has been recognized in many active
continental and oceanic hydrothermal systems. The pore
fluid’s geochemistry reflects mineral buffering. By addition
of Ca and extraction of Si and Na, this fluid is responsible
for converting mafic dike rocks enclosed in serpentinite
into rodingite, a conspicuous pink or white rock containing
hydrogrossular, diopside, prehnite, chlorite, clinozoisite,
vesuvianite, and/or wollastonite (Frost and Beard 2007).
Mg–Fe Partitioning
We can take the buffering logic further and argue that
the reacting mineral assemblage (principally olivine) also
controls the chemical potentials of Mg and Fe 2+ in the
pore fluid (Evans 2008). Thus, serpentine growing at the
olivine crystal surface is likely to be Mg rich because its
composition is guided by Mg/Fe2+ partitioning in olivine–
serpentine pairs. We assume that the partition coefficient,
K D, including its dependence on composition, is to a fi rst
approximation the same as for olivine + antigorite pairs in
fully recrystallized serpentinite (FIG. 7). Thus, an approach
to Mg/Fe2+ exchange equilibrium explains why the Mg#
[i.e. Mg/(Mg + Fe)] of the serpentine is generally greater
than that of olivine and orthopyroxene. Vein chrysotile,
derived by intergranular diffusion of aqueous Mg and Si
from host serpentinite, provides a geochemical test of this
model by showing a peak frequency at Mg# = 97 (Evans et
al. 2012, Fig. 14). Lizardite compositions have a similarly
placed frequency maximum (Mg# = 96–97), but in contrast
to chrysotile the histogram has a broad tail extending to an
Mg# of 86. The tail reflects the major and variable presence
of Fe3+ in lizardite. These high Mg#s are not a reflection
of any intrinsic, crystal-chemical limitation on Fe uptake
in serpentine; they are a response to local heterogeneous
phase equilibrium in a system of mantle-peridotite composition. With more Fe-rich olivine, such as in layered mafic
intrusions, we can fi nd a much more Fe-rich lizardite
(probably rich in Fe3+).
102
A PR IL 2013
0.06
tite are formed instead. This observation fi nds support
in the upward curvature of data points on the graph of
magnetic susceptibility relative to density (e.g. Beard et al.
2009). Rather than separate steps, we may be witnessing
enhanced diffusion between olivine and pyroxene microdomains, as cracks are opened by volume expansion or
tectonic deformation. Mg and Si are exchanged across
domains. This perhaps explains why “bastite” serpentine
replacing orthopyroxene is commonly not accompanied
by talc, which a local isochemical model would require.
0.04
Volume Increase
0.14
XFe Antigorite
0.12
0.10
0.08
0.02
0.00
0.00
0.05
0.10
0.15
X Fe Olivine
0.20
0.25
Partitioning of total Fe and Mg (as XFe) between
coexisting olivine and antigorite in metamorphosed
serpentinites. Roughly 10–20% of Fe in antigorite is Fe3+.
FIGURE 7
Why Magnetite and Hydrogen?
If the serpentinization reaction goes to completion, why
doesn’t the Mg# of lizardite end up being the same as the
Mg# of the original olivine, around 90? (Forget brucite
for the moment.) This would be the textbook model for
a solid-solution reaction, as depicted (oxygen conserved)
in a closed, isobaric T–XMg phase loop. What happens is
that residual olivine retains its primary composition (Mg#)
until it is completely consumed, and a result of this is that
lizardite is similarly restricted in its content of ferrous iron.
This behavior can be attributed to the very slow rate of
MgFe diffusion in olivine at low temperatures (Evans 2010).
Such diffusion is required in the textbook model; both
olivine and lizardite should become Fe enriched in unison
as their modal amounts change reciprocally. The resulting
problem of Fe mass balance is resolved by the precipitation
of magnetite, and the attendant oxidation is enabled by the
liberation of H 2 gas. The modal amount of magnetite and
the release of H 2 will be greatest when brucite, whose Mg#
is less than that of olivine, does not form (e.g. in harzburgite, a rock rich in orthopyroxene). Growth of Fe3+ -rich
lizardite will reduce the amount of magnetite but will have
little effect on hydrogen production.
Kinetics of Serpentinization
The progress of the serpentinization reaction can be readily
measured in the laboratory. The reaction may be sluggish
even at hydrothermal temperatures and occurs at temperatures lower than the equilibrium temperature by as much
as 50–60 °C. It is common to observe sizeable changes over
days or weeks, although rates fall off sharply with declining
temperature. These kinetics highlight the problem that
hydrous fluid has in nature to penetrate cracks and grain
boundaries towards its “target” in the temperature window
for reaction. Incompletely serpentinized peridotite and the
nearby co-occurrence of little-altered and highly altered
peridotite are both commonly observed. When we add to
these limitations those of MgFe diffusion in olivine, we
must recognize that low-T serpentinization of peridotite
does not proceed in nature under equilibrium conditions.
Equilibrium is both local and partial, and the extent of
serpentinization is generally limited by an insufficient
supply of H2O at the grain scale.
If we take the reaction mass balance at face value, serpentinization of peridotite implies a potential volume increase
as large as 40%. Isovolumetric serpentinization, on the
other hand, requires a major loss of components, principally Mg. Mid-20 th-century researchers noted that serpentinized peridotites fail to reveal convincing evidence of
expansion in their microstructures. However, whole-rock
major-element studies have generally not supported wholesale major-element depletion (O’Hanley 1996; Hattori and
Guillot 2007). Curiously, expansion cracks around partially
serpentinized olivine have most commonly been observed
in plagioclase in troctolites and gabbros (Fig. 2 in Evans
2004). The microstructure is very much like that in garnet
enclosing ultrahigh-pressure coesite that later inverted to
quartz. The development of chrysotile veinlets in cracks
cutting serpentinite rinds around outcrop-scale kernels of
fresh peridotite has been interpreted as caused by progressive expansion in the cores (O’Hanley 1996). Expansion
ordinarily requires a crystallization pressure to be exerted
by the growing mineral, serpentine in this case, utilizing
energy provided by the reaction affi nity of the overstepped
reaction at low temperatures (Kelemen and Hirth 2012).
Alternatively, given adequate pathways, this energy can
drive diffusion of elements away from the site of alteration.
Perhaps it is a question of enhanced water/rock ratio facilitating the diffusional exit of Mg and Si. Microscopic and
submicroscopic examination of fractured olivine grains
may eventually provide answers.
Ferric Iron in Serpentine
Our incomplete understanding of how Fe3+ partitions into
the serpentine structure is a problem in computing reaction
paths and products as a function of bulk composition
and water/rock ratio. The uptake of Fe3+ in lizardite can
take the form of a Mg cronstedtite component (a coupled
replacement by Fe3+ of Mg and Si on the M and T sites),
a vacancy-balanced M-site substitution, or a process
of deprotonation. Mg cronstedtite substitution can be
expected in SiO2 -poor rocks containing brucite or olivine,
whereas the M-vacancy trend should occur in rocks with
talc or pyroxene and in partially carbonated peridotites
(Streit et al. 2012). Mg cronstedtite substitution shows
as an inverse correlation (with slope approaching –0.5)
between T-site atoms (Si + Al/2 + Cr/2) and total Fe apfu.
On the other hand, an increase in the cation occupancy
ratio, T/(T + M), with an increase in Fe signifies M-vacancy
substitution. Oxidation of Fe2+ by deprotonation may well
be favored at the high pH associated with serpentinizing
fluids. The Mg cronstedtite substitution is probably the
dominant mode of Fe3+ uptake in serpentine. A mixedvolatile reaction linking oxidized and reduced Fe in Fe
serpentine suggests that low T should favor the formation
of the Mg cronstedtite component:
A two-step process of serpentinization has been suggested
by a number of authors, based upon petrographic study of
natural samples: an initial growth of brucite is reversed with
further serpentinization, and low-Fe lizardite and magneE LEMENTS
103
6 cronstedtite + H2 = 3 Fe2+ serpentine + 5 magnetite
+ 7 H 2O .
A PR IL 2013
Hopefully, the application of micro-XANES spectroscopy
will soon be able to tell us how the uptake of Fe3+ varies in
space and time in the serpentinization process.
Antigorite contains less Fe3+ than lizardite (Evans et al.
2012). Thus, during the lizardite-to-antigorite transition,
the cronstedtite component of ferrian lizardite cannot be
readily accommodated in antigorite, so it is likely to be
converted into magnetite and an Fe antigorite component,
as in the reaction above. This is a dehydration reaction that
will be accompanied by whole-rock reduction in Fe3+ (e.g.
Shive et al. 1988). The reaction either consumes residual
CH4 or H 2 or alternatively liberates oxygen or oxidized
species into the aqueous-fluid product.
Dehydration of Serpentinite
The prograde metamorphism of serpentinite ultimately
brings about the loss of H 2O and growth of secondary
olivine and talc (alternatively, orthopyroxene at high
pressure). Strict reversal of the serpentinization reaction
would require the reintroduction of hydrogen, something
that seems unlikely. Does this mean that metamorphosed
serpentinite retains its high Fe3+/Fe2+ ratio (FIG. 6)? Field
accounts of thermally deserpentinized ultramafic rock
from diverse tectonic environments reveal in many cases
the presence of newly generated olivine with a Mg# in
the unusual range of 98 to 93, together with abundant
magnetite. Although the data are sparse, the Mg# of
prograde olivine correlates well with analyzed whole-rock
Fe3+ contents, and so it can serve as a proxy indicator of
the extent of reduction in metaperidotite. Given the very
dilute concentration of O2 in the aqueous fluid and the
mass of potential magnetite reactant, any reduction in Fe3+
simply by release of O2 is likely to be very small. So there
may in some cases be a kind of hysteresis in petrophysical
properties in the peridotite–serpentinite cycle.
However, there are also field examples of progrademetamorphosed serpentinite where the reduction of Fe3+
apparently did take place and olivine compositions were
restored to mantle values (Mg# = 92–89). In these cases
we have to ask if there were sufficient modal amounts of
graphite and/or sulfide in the average serpentinite to react
with magnetite and release CO2 and SO2 in some form into
the fluid, or was an oxidizing fluid introduced? In practice,
we don’t always know the redox ratio (or the degree of
serpentinization) of the original serpentinite. This issue
is deserving of further study in regard to what kind of
released H 2O fluid we can expect from the eventual highpressure dehydration of antigorite serpentinite in subduction zones.
WHERE?
Serpentinites are volumetrically minor, but they occur
in all ancient orogenic belts. A suite of rocks consisting
of serpentinite, mafic volcanic rocks, and minor chert is
known in many mountain belts, and they are collectively
referred to as ophiolites. Serpentinites are also produced
at present in a variety of geological settings.
Locations of Present-Day Serpentinization
Peridotites are hydrated by seawater under a variety of
conditions, from the cool seafloor to hot ridge hydrothermal systems. Low-temperature seafloor serpentinization is common at the surface of oceanic lithosphere
produced at slow- to ultraslow-spreading ridges (Fig. 1 in
Guillot and Hattori 2013 this issue). Here, magma-starved
oceanic lithosphere exposes mantle peridotite on the
seafloor (Cannat et al. 2010). Because the Mid-Atlantic
Ridge has a slow spreading rate of ~2 cm/y, serpentinites
are exposed at numerous sites on the Atlantic Ocean
E LEMENTS
floor. Well-studied examples include the Kane area of the
Mid-Atlantic Ridge (MARK, ~15°20’ N), the Atlantic Massif
(~30° N), and the Iberian-margin serpentinites. Seawater is
able to penetrate to deep levels, forming serpentinite down
to ~7 km. Oceanic lithosphere at fast-spreading ridges, such
as the East Pacific Rise, is commonly covered by basaltic
rocks, although serpentinites are exposed along major
transform faults, fractures in trenches, and local rifts. The
serpentinites in the Hess Deep in the Pacific Ocean are a
good example. Another place for serpentinization is along
faults at outer rises near trenches, where incoming oceanic
plates are bent, producing deep fractures (Ranero et al.
2003). Serpentinization at low temperatures on and near
the seafloor is commonly accompanied by the formation
of carbonates and sulfides through microbial reduction of
marine sulfate within serpentinites (e.g. Alt et al. 2012).
Ocean-ridge hydrothermal activity results in the hydration
of abyssal peridotite near the seafloor at moderately high
temperatures (e.g. 350 °C). The Mid-Atlantic Ridge hydrothermal systems of Logatchev and Lost City have been well
reported since their discovery (e.g. Früh-Green et al. 2003).
Serpentinites formed by such hot fluids commonly show
positive Eu anomalies because of dissolution of plagioclase
in mafic igneous rocks during hydrothermal activity.
Forearc mantle peridotite is hydrated by water expelled
from the subducting slab, forming a layer of serpentinite
near the base of the mantle wedge. The existence of such
a hydrated layer is suggested by the low and anisotropic
seismic velocities in that location (Hirth and Guillot 2013
this issue).
Serpentinites in Orogenic Belts
Orogenic belts commonly contain bodies and lenses of
serpentinite (the classic Alpine serpentinites). Although
minor in volume, they provide information useful for
elucidating the history of orogenic belts. The protoliths of
these serpentinites have various origins; for example, they
may represent cumulates from a variety of mafic melts or
residual-mantle peridotites. Residual (i.e. depleted)-mantle
peridotites comprise abyssal and mantle-wedge peridotites.
The latter show a refractory geochemical signature since
they underwent a high degree of partial melting. The
absence of primary mantle minerals and the mobility of
most petrogenetically useful elements make it difficult to
characterize the protoliths. However, plots of the major
elements Mg, Al, and Si commonly provide information
related to the protoliths because the serpentinization is
primarily isochemical, except for Ca (e.g. Hattori and
Guillot 2007). During partial melting, the Al/Si ratio
decreases and the Mg/Si ratio increases in residual peridotites. This approach, however, is not useful for evaluating
the origin of hydrated dunite, whether of cumulate or
residual-mantle origin. The abundance of platinum-group
elements (PGEs) is found to be effective in differentiating
cumulates from residual-mantle dunite because partial
melting fractionates PGEs due to the different compatibilities of these elements.
Almost all large bodies of serpentinite are hydrated mantle
peridotite, but serpentinite produced by the hydration of
ultramafic cumulates form large blocks in massifs (Hattori
and Guillot 2007). Good examples are the base of the
Drakkarpo unit and the Nidar ophiolite in the western
Himalayas, and the Pelvas d’Abriès serpentinite in the
western Alps.
The abundance of immobile elements in bulk rocks and
the chemistry of relict Cr spinel have been successfully
used to discriminate abyssal serpentinite from mantlewedge serpentinite. Cr spinel is a robust mineral that
104
A PR IL 2013
commonly retains its primary composition in the cores
of grains, even in fully hydrated and deformed peridotite.
These studies show that regionally extensive serpentinites
are mostly composed of hydrated abyssal peridotite. They
were hydrated on or near the seafloor and obducted onto
the continents. Examples include the Jurassic Apennine
ophiolite, the Tertiary Alpine serpentinites, and the Late
Cretaceous to Mid-Eocene Northern Serpentinite Belt in
Cuba and Hispaniola.
Primitive mantle normalized
1000
Mantle-wedge serpentinites are found in forearcs, along
major deformation zones in subduction zones, and in
trenches. Their buoyancy (2.6 g/cm3 relative to 3.2 g/cm3
for anhydrous peridotite) allows them to rise from the
base of the mantle wedge to the surface (Saumur et al.
2010). As they ascend, they commonly bring with them
fragments of rocks, such as high-pressure rocks (e.g. Fryer
et al. 2006; Saumur et al. 2010). These serpentinites are
characterized by distinctly low concentrations of immobile,
incompatible elements compared to serpentinites formed
from oceanic mantle peridotite. Examples of mantle-wedge
serpentinites are the Mariana forearc serpentinites and the
serpentinites along major shear zones in Cuba and the
Dominican Republic (e.g. Hattori and Guillot 2007; Saumur
et al. 2010).
In subduction zones, sediments and altered oceanic lithosphere release water and fluid-mobile elements; this leads
to the enrichment of serpentinite in the overlying mantle
wedge in As, Sb, Pb, and Sr (e.g. Hattori and Guillot 2007).
Using the speciation of As and the presence of pseudomorphic textures, mantle-wedge serpentinites are interpreted
to show early incorporation of As at shallow depths and
later incorporation of Sb, Pb, and Sr at deeper levels (Hattori
et al. 2005; Deschamps et al. 2012).
Among these elements, As5+ replaces Si4+ in the tetrahedral
site of serpentine (Hattori et al. 2005). The distribution
of B also suggests residence in the tetrahedral site (Pabst
et al. 2011). Some elements form distinct phases, such
as sulfides. Among the fluid-mobile elements, alkalis are
not enriched in serpentinite (Hattori and Guillot 2007).
Their ionic radii are too large for them to be incorporated
into the serpentine structure, except for Li +, which has an
ionic radius similar to Mg2+. Furthermore, in a subduction
setting, alkalis are not released from the subducting slab
and then incorporated in serpentinite because they are
retained in phengite, a mineral that is stable at a depth
exceeding 300 km.
E LEMENTS
alay
100
as
10
crust
1
Mariana
Volcanic-front
Volcanic
front magmas
magmas
MORB normalized
100
10
MORB
1
B
As
Sb
Pb
Sr
Cu
Ce
Nd
Zr
Sm
Ti
Y
Al
Sc
Si
highly compatible
fluid soluble
Enrichment of fluid-mobile elements in mantle-wedge
serpentinites compared to their enrichment in
volcanic-front magmas (modified after Hattori and Guillot 2003).
The enrichment patterns are similar, which suggests a genetic link
between the two. The dehydration of serpentinites releases H2O
and fluid-mobile elements into the hot interior of mantle wedges,
which triggers partial melting and the generation of arc magmas.
FIGURE 8
Relations with Arc Magmas
It is well known that arc volcanoes are distributed in a
curve parallel to the trench, with a conspicuous lack of
volcanoes between the volcanic front and the trench. Also,
arc magmas contain high concentrations of fluid-mobile
elements, such as As, Sr, and Pb (FIG. 8; Noll et al. 1996).
There had been considerable debate regarding the formation of volcanic fronts. Hattori and Guillot (2003) showed
that the enrichment pattern of fluid-mobile elements in arc
magmas is similar to that of mantle-wedge serpentinites
(FIG. 8). The data explain both the lack of volcanoes in
the forearc and the occurrence of a volcanic front where
the subducted slab is around 100 km deep. Serpentinite
retains water in forearc mantle and transports fluid-mobile
elements incorporated at shallow depth to deep mantle
depths. Eventual dehydration of the serpentinite releases
water and fluid-mobile elements to the hot interior of the
mantle wedge, leading to partial melting and arc magmatism (Hattori and Guillot 2003).
Abyssal serpentinites formed at ridges and on the ocean
floor move to subduction zones by plate motion, which
results in the transfer of fluid-mobile elements from ocean
water to subduction zones (Vils et al. 2008; Deschamps et
al. 2012). Oceanic serpentinites and serpentinites in mantle
wedges are eventually dehydrated, releasing water and
fluid-mobile elements to the interior of the mantle wedges.
Thus, serpentinites play an important role in the recycling
of elements from the surface to the deep lithosphere and
back to the surface via mantle wedges and arc magmas.
105
al
yss
Ab
0.1
0.001
Role of Serpentinite in the Recycling
of Elements
Mantle peridotite loses incompatible elements during partial
melting, but the concentrations of these elements generally
increase during the progressive hydration of peridotite. In
a system with a low water/rock ratio, the concentrations
of soluble elements are enriched in the fluid as H 2O is
consumed by the hydration. Those elements dissolved in
the fluid are eventually incorporated in the serpentinite.
The composition of serpentinite is therefore affected by
the nature of the fluid and by the water/rock ratio. For
example, seawater contains high concentrations of B, U,
and S as sulfate, so that serpentinite formed in a marine
environment has high contents of these elements. As an
example, serpentinites in the MARK area are enriched in
B (up to 136 ppm) and U (up to 7 ppm) (Vils et al. 2008).
These values may be compared to primitive mantle values
of 0.3 and 0.0203 ppm, respectively.
Mantle-wedge
serpentinites
mantle
wedge serpentinites
Him
A PR IL 2013
Mg
CONCLUSIONS
Serpentinites have relevance to several important subjects
in the geosciences, including geodynamics, prebiotic life,
and the global recycling of elements, as this issue of
Elements shows. Future progress in understanding serpentinites will be enhanced when cutting-edge techniques,
such as TEM, reaction-path modeling, isotope geochemistry, and mineral spectroscopy, focus in fully integrated
REFERENCES
Alt JA, Garrido CJ, Shanks WC III,
Turchyn A, Padrón-Navarta JA, SánchezVizcaíno VL, Gómez Pugnaire MT,
Marchesi C (2012) Recycling of water,
carbon, and sulfur during subduction
of serpentinites: A stable isotope study
of Cerro del Almirez, Spain. Earth and
Planetary Science Letters 327-328: 50-60
Andreani M, Mével C, Boullier A-M,
Escartín J (2007) Dynamic control on
serpentine crystallization in veins:
Constraints on hydration processes
in oceanic peridotites. Geochemistry
Geophysics Geosystems 8: doi:
10.1029/2006GC001373
Bailey SW (1988) Polytypism of 1:1 layer
silicates. Reviews in Mineralogy 19: 9-27
Baronnet A, Devouard B (1996) Topology
and crystal growth of natural chrysotile
and polygonal serpentine. Journal of
Crystal Growth 166: 952-960
Baronnet A, Andréani M, Grauby O,
Devouard B, Nitsche S, Chaudanson D
(2007) Onion morphology and microstructure of polyhedral serpentine.
American Mineralogist 92: 687-690
Beard JS, Frost BR, Fryer P, McCaig A,
Searle R, Ildefonse B, Zinin P, Sharma
SK (2009) Onset and progression of
serpentinization and magnetite formation in olivine-rich troctolite from IODP
Hole U1309D. Journal of Petrology 50:
387-403
Cannat M, Fontaine F, Escartín J (2010)
Serpentinization at slow-spreading
ridges: Extent and associated hydrogen
and methane fluxes. In: Rona P,
Devey CW, Dyment J, Murton BJ (eds)
Diversity of Hydrothermal Systems on
Slow Spreading Ocean Ridge. American
Geophysical Union Geophysical
Monograph 188, pp 241-264
Capitani G, Mellini M (2004) The
modulated crystal structure of antigorite: The m = 17 polysome. American
Mineralogist 89: 147-158
Cressey BA,Whittaker EJW (1993)
Five-fold symmetry in chrysotile
asbestos revealed by transmission
electron microscopy. Mineralogical
Magazine 57: 729-732
Deer WA, Howie RA, Zussman J (2009)
Rock-Forming Minerals, Volume 3B.
Layered Silicates Excluding Micas and
Clay Minerals. Geological Society,
London, 314 pp
Deschamps F, Godard M, Guillot S,
Chauvel C, Andreani M, Hattori K,
Wunder B, France L (2012) Behavior of
fluid-mobile elements in serpentines
from abyssal to subduction environments: Examples from Cuba and
Dominican Republic. Chemical Geology
312: 93-117
E LEMENTS
fashion on the diversity of natural serpentinites from
well-chosen locations. These studies will illuminate the
roles of mineral buffering, kinetics, deformation, and the
constraints of local and partial equilibrium.
ACKNOWLEDGMENTS
We thank H. Austrheim, P. Kelemen, and G. Calas for their
critical reviews.
Evans BW (2004) The serpentinite
multisystem revisited: Chrysotile is
metastable. International Geology
Review 46: 479-506
Evans BW (2008) Control of the products
of serpentinization by the Fe2+ Mg-1
exchange potential of olivine and
orthopyroxene. Journal of Petrology 49:
1873-1887
Evans BW (2010) Lizardite versus antigorite serpentinite: Magnetite, hydrogen,
and life(?). Geology 38: 879-882
Evans BW, Dyar MD, Kuehner SM (2012)
Implications of ferrous and ferric iron
in antigorite. American Mineralogist 97:
184-196
Frost BR, Beard JS (2007) On silica
activity and serpentinization. Journal of
Petrology 48: 1351-1368
Früh-Green GL, Kelley DS, Bernasconi
SM, Karson JA, Ludwig KA, Butterfield
DA, Boschi C, Proskurowski G (2003)
30,000 years of hydrothermal activity
at the Lost City Vent Field. Science 301:
495-498
Fryer P, Gharib J, Ross K, Savov I, Mottl
MJ (2006) Variability in serpentinite
mudflow mechanisms and sources:
ODP drilling results on Mariana
forearc seamounts. Geochemistry
Geophysics Geosystems 7: G08014, doi:
10.1029/2005GC001201
Fubini B, Fenoglio I (2007) Toxic potential of mineral dusts. Elements 3:
407-414
Grauby O, Baronnet A, Devouard B,
Schoumacker K, Demirdjian l (1998)
The chrysotile-polygonal serpentine
suite synthesized from a 3MgO-2SiO2 excess H 2O gel (abstract). Terra Nova
Abstract Supplement 1: 24
Guillot S, Hattori K (2013) Serpentinites:
Essential roles in geodynamics, arc
volcanism, sustainable development,
and the origin of life. Elements 9: 95-98
Hattori KH, Guillot S (2003) Volcanic
fronts form as a consequence of serpentinite dehydration in the forearc mantle
wedge. Geology 31: 525-528
Hattori KH, Guillot S (2007) Geochemical
character of serpentinites associated
with high- to ultrahigh-pressure
metamorphic rocks in the Alps,
Cuba, and the Himalayas: Recycling
of elements in subduction zones.
Geochemistry Geophysics Geosystems
8: doi: 10.1029/2007GC001594
Hattori K, Takahashi Y, Guillot S,
Johanson B (2005) Occurrence of
arsenic (V) in forearc mantle serpentinites based on X-ray absorption
spectroscopy study. Geochimica et
Cosmochimica Acta 69: 5585-5596
Kelemen PB, Hirth G (2012) Reactiondriven cracking during retrograde
metamorphism: Olivine hydration
and carbonation. Earth and Planetary
Science Letters 345-348: 81-89
Mellini M, Trommsdorff V, Compagnoni
R (1987) Antigorite polysomatism:
behaviour during progressive metamorphism. Contributions to Mineralogy
and Petrology 97: 147-155
Mitchell RH, Putnis A (1988) Polygonal
serpentine in segregation-textured
kimberlite. Canadian Mineralogist 26:
991-997
Noll PD Jr, Newsom HE, Leeman WP,
Ryan JG (1996) The role of hydrothermal fluids in the production of
subduction zone magmas: Evidence
from siderophile and chalcophile trace
elements and boron. Geochimica et
Cosmochimica Acta 60: 587-611
O’Hanley DS (1996) Serpentinites:
Records of Tectonic and Petrologic
History. Oxford University Press,
Oxford, 277 pp
Pabst S, Zack T, Savov IP, Ludwig T, Rost
D, Vicenzi EP (2011) Evidence for boron
incorporation into the serpentine
crystal structure. American Mineralogist
96: 1112-1119
Ranero CR, Phipps Morgan J, McIntosh
K, Reichert C (2003) Bending-related
faulting and serpentinization at the
Middle America trench. Nature 425:
367-373
Saumur BM, Hattori KH, Guillot S (2010)
Contrasting origins of serpentinites
in a subduction complex, northern
Dominican Republic. Geological Society
of America Bulletin 122: 292-304
Shive PN, Frost BR, Peretti A (1988) The
magnetic properties of metaperidotitic
rocks as a function of metamorphic
grade: Implications for crustal magnetic
anomalies. Journal of Geophysical
Research 93B: 12187-12195
Streit E, Kelemen P, Eiler J (2012)
Coexisting serpentine and quartz from
carbonate-bearing serpentinized peridotite in the Samail Ophiolite, Oman.
Contributions to Mineralogy and
Petrology 164: 821-837
Vils F, Pelletier L, Kalt A, Müntener O,
Ludwig T (2008) The Lithium, Boron
and Beryllium content of serpentinized peridotites from ODP Leg 209
(Sites 1272A and 1274A): Implications
for lithium and boron budgets of
oceanic lithosphere. Geochimica et
Cosmochimica Acta 72: 5475-5504
Wicks FJ, Whittaker EJW (1975) A
reappraisal of the structures of the
serpentine minerals. Canadian
Mineralogist 13: 227-243
Hirth G, Guillot S (2013) Rheology and
tectonic significance of serpentinite.
Elements 9: 107-113
106
A PR IL 2013