You are on page 1of 50

Fluids in the forearc of subduction zones

Geophysical/mineralogical budgets

Seismological data and mineral/rock physics interpretation

Electrical conductivity from MT studies


Serpentinization and fluids in subduction zones

Cold forearc mantle wedge


T<700°C decoupled from plate
motion and convection

Reynard Lithos (2013)


modified from Schmidt and Poli (1998)

Serpentinization/mantle hydration is a major process in the forearc mantle


Can we detect and estimate serpentinization of the mantle wedge? Can we detect
fluids? using geophysical signal and property measurements
Interactions between mantle and water in forearc
subduction
Clay-like or mica-like magnesian minerals: serpentines (and chlorite)
Major water carrier (≈12 wt%) in subduction zones
Control on water budget and migration:

Olivine + Enstatite +H2O = Serpentine + Magnetite ± Chlorite, Brucite, H2…

Mantle + Water = Serpentine

Lizardite

Auzende et al. 2002

Antigorite
Chrysotile = nanotubes
Seismic evidence for serpentinite

Nikulin et al. (2009)


Large anisotropy
VP 5-6 km/s
Bostock et al. (2002) VS 2.5-3.5 km/s

Serpentine in the mantle wedge and plate interface


-serpentinization reaches 50-100% in hot SZ, <10% in cold SZ
-deformed serpentinite layer above the plate interface
-can explain aseismic behavior and decoupling of the plate interface
Elastic properties of antigorite single-crystal

Courtesy
S. Merkel

Bezacier et al. EPSL (2010); JGR (2013)

Natural crystal ≈100*150*15 µm


Brillouin spectroscopy to 9 GPa

Calculate velocities in isotropic (rocks with random orientations of


crystals) and anisotropic (deformed rocks) aggregates
Anisotropy of a deformed serpentinite
test2nr-Ant-pf.ai

(HKL) = 100 (HKL) = 010 [UVW] =001


N = 31940 Contours (x uni.) N = 31940 Contours (x uni.) N = 31940 Contours (x uni.)
6.23 9.05 24.34

8.0 21.0
5.0 7.0 18.0
4.0 6.0
15.0
5.0
X 3.0 X X 12.0
4.0
3.0 9.0
2.0
2.0 6.0

0.01 0.00 0.00


shading - inverse log shading - inverse log shading - inverse log
Max.Density = 6.23
Z Min.Density = 0.01 Max.Density = 9.05
Z Min.Density = 0.00 Max.Density = 24.34
Z Min.Density = 0.00
lower hemisphere Non-Polar data lower hemisphere Non-Polar data lower hemisphere Non-Polar data
pfJ = 3.12 Xs = 270.0 pfJ = 3.68

pole figures EBSD


Xs = 270.0 pfJ = 11.24 Xs = 270.0

Single-crystal
Antigorite
208.1(5.8) 66.4(1.2) 16.0(1.8) 0 5.5(0.1) 0
201.6(4.9) 4.9(0.5) 0 -3.2(0.1) 0
96.9(0.9) 0 1.6(0.1) 0
16.9(0.1) 0 -12.1(0.2)
Orientation map 18.4(0.1) 0

antigorite2-VpG.txt
65.6(0.5)

Vp Contours (km/s) AVs Contours (%) Vs1 Polarisation Planes


VRH rock aggregate
8.31 48.84 48.84

8.00
7.75 40.0
7.50
7.25
35.0
30.0 89 27 31 4 -3 -1
7.00 25.0
6.75 20.0
6.50
6.25
15.0
10.0
172 54 -7 7 13
6.00

5.62 0.87 0.87 163 -7 26 2


shading - inverse log shading - inverse log shading - inverse log
Max.Velocity = 8.31 Min.Velocity = 5.62 Max.Anisotropy = 48.84 Min.Anisotropy = 0.87
Anisotropy = 38.6 %
lower hemisphere lower hemisphere lower hemisphere
56 3 8
32 3
27
Anisotropy VP 39% VS 49%
Vs1 Contours (km/s) Vs2 Contours (km/s) dVs Contours (km/s)
4.72 4.10 1.84

1.60
1.40
4.20 1.20
4.00 3.60

Close to transverse isotropic


1.00
3.80 3.40 0.80

Bezacier et al. EPSL 2010


3.60 0.60
3.20
3.40 0.40

3.05 2.84 0.03


shading - inverse log shading - inverse log shading - inverse log
Max.Velocity = 4.72 Min.Velocity = 3.05 Max.Velocity = 4.10 Min.Velocity = 2.84 Max.Velocity = 1.84 Min.Velocity = 0.03
Anisotropy = 43.2 % Anisotropy = 36.3 %
lower hemisphere lower hemisphere lower hemisphere
Anisotropy of a deformed
test2nr-Ant-pf.ai

serpentinite
(HKL) = 100 (HKL) = 010 [UVW] =001
N = 31940 Contours (x uni.) N = 31940 Contours (x uni.) N = 31940 Contours (x uni.)
6.23 9.05 24.34

8.0 21.0
5.0 7.0 18.0
4.0 6.0
15.0
5.0
X 3.0 X X 12.0
4.0
3.0 9.0
2.0
2.0 6.0

0.01 0.00 0.00


shading - inverse log shading - inverse log shading - inverse log
Max.Density = 6.23
Z Min.Density = 0.01 Max.Density = 9.05
Z Min.Density = 0.00 Max.Density = 24.34
Z Min.Density = 0.00
lower hemisphere Non-Polar data lower hemisphere Non-Polar data lower hemisphere Non-Polar data

pole figures EBSD


pfJ = 3.12 Xs = 270.0 pfJ = 3.68 Xs = 270.0 pfJ = 11.24 Xs = 270.0

Antigorite

Orientation map
ULVL
antigorite2-VpG.txt

Vp Contours (km/s) AVs Contours (%) Vs1 Polarisation Planes


8.31 48.84 48.84

8.00
7.75 40.0
7.50 35.0
7.25 30.0
7.00 25.0
6.75 20.0
6.50 15.0
6.25 10.0
6.00

5.62 0.87 0.87


shading - inverse log shading - inverse log shading - inverse log
Max.Velocity = 8.31 Min.Velocity = 5.62 Max.Anisotropy = 48.84 Min.Anisotropy = 0.87
Anisotropy = 38.6 %
lower hemisphere lower hemisphere lower hemisphere

Anisotropy VP 39% VS 49%


Vs1 Contours (km/s) Vs2 Contours (km/s) dVs Contours (km/s)
4.72 4.10 1.84

1.60
1.40
4.20 1.20

Incidence angle °
4.00 3.60
1.00

Bezacier et al. EPSL 2010


3.80 3.40 0.80
3.60 0.60
3.20
3.40 0.40

3.05 2.84 0.03


shading - inverse log shading - inverse log shading - inverse log
Max.Velocity = 4.72 Min.Velocity = 3.05 Max.Velocity = 4.10 Min.Velocity = 2.84 Max.Velocity = 1.84 Min.Velocity = 0.03
Anisotropy = 43.2 % Anisotropy = 36.3 %
lower hemisphere lower hemisphere lower hemisphere
Anisotropy of deformed serpentinites and peridotite

pole figures EBSD

Antigorite

Orientation map

Reynard Lithos 2013


Seismic velocities and degree of serpentinization

Nearly all rocks fall on a


constant VP/VS ratio line

One observable instead of two

Dehydration of the oceanic crust


leads to eclogite that has no
seismological contrast with the
mantle (peridotites)

Lowering of P and S wave velocities in the mantle wedge can be


converted in % of serpentinization
Mantle wedge hydration
0

40

Depth(km)
1 1
80
2 2

3 3
120

160
6.0 6.5 7.0 7.5 8.0 8.5 9.0 3.50 3.75 4.00 4.25 4.50 4.75
Vp (km/s) Vs (km/s)

Tohoku NE Honshu Nakajima et al. 2009

Seismic velocities of subduction zone rocks


Bezacier et al. EPSL 2010, Tphy 2010, JGR 2013 Costa Rica DeShon and Schwartz GRL 2004

Extensive hydration of the mantle wedge in hot SZ, sluggish in cold SZ


Significant permeability of the subduction interface
Seismic evidence for serpentinite

Large anisotropy
VP 5-6 km/s
VS 2.5-3.5 km/s

Serpentine in the mantle wedge and plate interface


-can explain aseismic behavior and decoupling down to 90 km
-serpentinization reaches 50-100% in hot SZ, <10% in cold SZ
-deformed serpentinite layer at the subduction interface
Seismic velocities and degree of serpentinization

Anisotropy and fluids make velocities move out of the linear trend
Weakened plate
interface and heat
flow

Wada et al. JGR 2008


Global state of hydration

Abers et al. NGeo 2017


Imaging fluids in subduction?

Example of Costa Rica

Electromagnetic studies give


2D conductivity images
High conductivity in the cold
mantle wedge
(Worzewski et al. NGeo 2011)

Seismic velocities =
serpentinisation of the cold
wedge (DeShon and
Schwartz 2004)

High conductivity areas in the cold (<700°C) mantle wedge must correspond to
high salinity aqueous fluids
Electrical conductivity of serpentine
700 500 300°C
Conduction mechanism is similar to
that in olivine and pyroxene: hopping
Watson et al 2009
Dai and Karato 2009 of small polaron
Conductivity (S.m-1)

Similar activation energy


Positive pressure dependence
Dependence on iron concentration

“Dry” mantle ≈ 60-70% olivine +


30-40% orthopyroxene

Hydrated mantle ≈ 90-95% serpentine


+ <5% conductive minerals
(magnetite and sulfides) negligible
contribution (non connective
inclusions)
Reynard et al. EPSL 2011

Antigorite has very low conductivity close to those of “dry” minerals:


the cold mantle wedge rocks have very low conductivities < 10-4 S.m-1
Electrical conductivity model for forearc mantle

σbulk ≈ 2/3 F σfluid

Ryukyu hot
Costa Rica
Cascadia
Hokkaido
Tohoku cold

Serpentinized mantle has to contain a conductive aqueous fluid


More fluid or higher conductivity fluid in hot vs cold subduction
Need to add a highly soluble salt to water = NaCl
Saline aqueous fluids
Salinity-conductivity relationship: log(σfluid) ≈ 1.719 + 0.9455 log(m) Quist et al 1968, 1969

Ryukyu hot
Costa Rica
Cascadia
Hokkaido
Tohoku cold

Serpentinization reaction consumes pure water and enriches the residual fluid in salt (Scambelluri et al 2004)
Serpentinization of a given volume of mantle consumes about Vw = 20-30 volume% of water:
mf = mi* Vw /Ff : in the range 20-30 for 1% residual fluid fraction Ff
with mf ≈ 2 for cE-M mantle = 0.1-1 S.m-1, mi ≈ 0.05 = seawater diluted with pure water from the
dehydrating slab
Saline aqueous fluids
Salinity-conductivity relationship: log(cfluid) ≈ 1.719 + 0.9455 log(m) Quist et al 1968, 1969

Ryukyu hot
Costa Rica
Cascadia
Hokkaido
Tohoku cold

Serpentinization reaction consumes pure water and enriches the residual fluid in salt (Scambelluri et al 2004)
Serpentinization of a given volume of mantle consumes about Vw = 20-30 volume% of water.

Back calculation of initial salinity falls in the range expected for dehydration fluids from the slab (Ito et al. GCA
1983)
Saline aqueous fluids

Ryukyu hot
Costa Rica
Cascadia
Hokkaido
Tohoku cold

Serpentinization reaction increases the residual fluid salinity (Scambelluri et al 2004)


At 100% serpentinization, water from dehydration will flow through the system and dilute Cl
Likely in very hot subduction where fluid flux in the first 60 km is very important
Geochemical evidence for such saline fluids in subduction?
Regional variations in salinity/conductivity in response to "history" of the subduction
Geochemical record

Wallace JVGR (2005)

large variability of Cl/H2O ratio in primitive magmatic inclusions in arc


magmas can be explained by addition of Cl-enriched fluids produced by
serpentinization and added laterally to the arc magma source region
Interpreting 2D MT imaging in Cascadia

Reynard PEPS 2016


Wannamaker et al G3 2014
Imaging fluids in subduction?

Cold mantle wedge <700°C


decoupled from plate motion

Low salinity fluids expelled from the slab are enriched during progressive
serpentinization of the cold mantle wedge
Active in hot subduction where slab dehydrates at smaller depths than cold
subduction, addition of Cl-enriched to arc magmas
Permeability of the subduction interface

Ultra-low velocity layer ULVL


large anisotropy
VP 5.5-6.5 km/s
VS 2-3.5 km/s
Audet et al. Nature 2009 Nikulin et al. G3 2009
Song et al. Science 2009

Impermeable interface in
contact with the crust of
the overridding plate

Permeable interface in
contact with
serpentinized forearc
Bezacier et al. 2010ab 2013, Reynard PEPS 2016
mantle
RF imaging in Cascadia

Bostock et al Nature 2002

Tauzin et al EPSL 2016


RF imaging and MT in Cascadia

Tauzin et al EPSL 2017


Terrane faulting controls fluid circulation

Tauzin et al EPSL 2017


Deep seismicity and the water cycle

Kita et al., GRL 2006

(modified from Schmidt


and Poli EPSL 1998)
Feed-back between mantle hydration-dehydration
and water cycle
Olivine + Orthopyroxene + H2O = Serpentine ± Magnetite ± H2, CH4
Serpentinite (80-99%) up to 12-13 wt% H2O in solid form

Dehydration above 5 GPa is partial


Atg => phA + OPx + water

Phase A is a dense hydrous


magnesian silicate, the
assemblage can contain 3.5 wt %
« water »

Choke-point @ 550°C 5.5 GPa


Estimating water budget

Van Keken et al JGR 2011


Water budget

Van Keken et al JGR 2011


Water budget

Rüpke et al EPSL 2004


Earthquake mechanism in double Wadati-Benioff
zones: do we need water there?

-Double seismic zones (DSZ) observed in the slab lithospheric mantle


20-30 km below the plate interface
-Serpentine dehydration hypothesis of Peacock (2001) or mechanical
instabilities

What can experiments and geophysical data tell us?


Is serpentine dehydration the cause of seismicity on the
lower plane of double seismic zone?

Peacock, Geology 2001 Ulmer and Tromsdorff Nature 1995


Hilairet et al., GRL 2006
Double Wadati-Benioff seismic zones (DSZ)

Kita et al., GRL 2006

Upper plane down-dip compression


dehydration of the oceanic crust

Lower plane down-dip tension


slab unbending

World wide distribution of DSZ in


subduction zones

Separation varies with thermal structure


and follows isotherms around
500-700°C

If dehydration of serpentine, the slab


lithosphere has been hydrated
down to 30 km depth
Brudzinski et al Science 2007
Deformation experiments and dehydration
Dobson et al. Science 2002
Jung et al. Nature 2004
Jung et al. EPSL 2009

Gasc et al. PEPI 2011

Chernak & Hirth Geol 2011

Raleigh & Paterson JGR


1965
Murrell & Ismail Tφ 1976
Rutter & Brodie JGR 1988

Agreement on weakening
during dehydration

No consensus on evidence
for brittle failure due to
dehydration
Deformation and dehydration experiments

P = 4 GPa, T = 500-600°C

3D FIB shows connective porosity


No fracture cross-cutting the sample, deformation is distributed
Talc-like (10 Å phase) intermediate phase may weaken the aggregate

Compaction and extraction of fluid


Fracture may or may not occur depending on scale, texture and geometry,
and their link to built-up of pore-pressure
Shear instabilities

Braeck and Podladchikov PRL 2007

John et al 2009

Active for mantle rock rheology at


650-800°C

Associated with shear zones that


would cause seismic anisotropy
Kelemen and Hirth Nature 2007
Do we need serpentine or hydrous phases to
make earthquakes on the lower plane?

Kelemen and Hirth, Nature 2007


Experimental debate part 2
Experimental debate part 2

Ferrand et al Ncomm 2017 Ohuchi et al Ngeo 2017


Is there hydration in the lithospheric slab mantle?

Decreasing hydration
Ranero et al. Geology 2004
Contreras-Reyes et al. JGR 2008

Faccenda et al. NGeo 2009

Yes down to 5 km below


oceanic crust
Seismic tomography of N-E Japan
subduction

Nakajima et al., GRL 2009; Gondwana Res 2009

-Low velocity patches along the lower seismicity plane are a common feature in SZ
(also in Chile, Dorbath et al., G3, 2008)
-Occur also along faults e.g. Kushiro-oki 1993 and aftershocks
-VP is strongly affected while VS is not : anisotropy effect not serpentinization
VP-VS systematics for anisotropic rocks
Seismic properties of anisotropic rocks and slab mantle

Anisotropic peridotites account for observed lower plane seismic velocities


without hydration: dry slab mantle explains the data (<10% serpentine)
Why do we see anisotropy in tomography?

Ray paths used to construct the


tomographic image are clustered near
the lower seismicity plane, and will
sample only specific orientation of the
anistropic velocity tensor
Anisotropy induced by earthquake and shear zones

Suzuki and Kasahara, PEPI 1996


Ueda et al., Geology 2009

Aftershocks propagate on horizontal faults pseudotachylites


Anisotropy model

Reynard et al. GRL 2010

“Regional” anisotropy = oceanic extension at the ridge N160, horizontal then bent at
the trench (Tono et al., 2009)
“Local” anisotropy from shear zones in lower plane, orientation of fast axis
perpendicular to the trench in horizontal faults, strong because low T
Anisotropy model
Small water content in the lithosphere

Kaneshima et al., EPSL 2007; Iidaka and Suetsugu 1992; Kubo et al., EPSL 2009

Metastable olivine wedge implies that the transformation kinetics to high-pressure


spinel forms is slow, which requires water content of less than 150 ppm

You might also like