Subduction slab

1
Subduction Zone
• A convergent plate boundary
• Old, cold lithosphere plate descends into the Earth’s mantle
Surface expression
• Deep oceanic trench on the oceanic plate
• Line of volcanoes on the overriding plate

2
Seismic structure of the slabs
Cascadia subduction zone

Chile Trench and the Andes at 23oS

3
4
Earth is the only planet in the solar system, which has
subduction zones and plate tectonics

Mercury and Moon Tectonically and Magmatically dead

Venus dominated by thick lithosphere and mantle plume
Mars Plate tectonics and subduction zones may have
existed in the ancient past but no more

5
Subduction zones and convergent plate margins

6
Important feature of the subduction zone

7
 Details of major subduction zones

8
Thermal parameter = age of subducting plate × descent rate
9
 Seven strain classes of the subduction zones

1. Active backarc spreading

2. Very slow backarc spreading
3. Mildly tensional
4a. Neutral
4b. Mildly tensional to compressional
5. Mildly compressional
6. Moderately compressional
7. Strongly compressional

10
Three layers of the subduction slab

Mantle lithosphere: controls the physics of the subduction

Crust: controls both of the physics and chemistry of the subduction
Sediments: controls the chemistry of the subduction

11
Mantle Lithosphere

1-2% denser than the underlying asthenosphere

The density excess and its thickness increase with the lithosphere ages
Subduction occurs in the convergent margins up to 170 Ma

12
Mantle Lithosphere
• Mean age for a slab reaching a trench is 100 Ma.
• Upper portion of the mantle lithosphere is harzburgite with a lower
layer of lherzolite
• Compared to the underlying asthenosphere, mantle lithosphere
becomes negatively buoyant when it is 10-30 Ma old.

13
Oceanic Crust

Typical oceanic Serpentinized mantle does occur

crust structure along slow-spreading ridges
East Pacific Rise Atlantic, Arctic, and Indian oceans
14
Subduction of the Oceanic Crust
Hydrothermal effect greatly affect the uppermost 500 m of basalt, leading
to an increase in water, CO2, and K.

15
Serpentinite
Serpentinization normally occurs in the slow-spreading ridges

Slow spreading carries more

water than the fast-spreading
ridges

16
Subduction of Oceanic Crust
1. Normal oceanic crust is invariably subductable.
2. The density contrast between the crust and mantle should be more
than 0.1 g/cm3.
3. The thickness of the continental crust should be less than 15 km,
whereas it should be less than 30 km for the oceanic crust

17
Subduction of the Continental Crust

Coesite > 80 km

Diamond > 120 km

18
Discovery of stishovite

19
Ultrahigh-pressure metamorphic rocks

20
Ultrahigh-pressure
metamorphic rocks

21
Sediments
Thickness and composition of subducted sediments are functions of the
sedimentary processes occurring on the ocean floor and in the forearc
region as well as functions of the tectonic processes near the toe of the
rigid rock framework of the overriding plate.

Factors substantially reducing the volume of subducted sedimentary

material:
• Sediment compaction and fluid in the trench region
• Sediment off-scraping and incorporation in accretionary prisms
• Underplating
• Accretion to the core buttress of the overriding plate

Conversely, the occurrence of subduction erosion of the continental

margin may enhance the amount of sialic material involved in
subduction

22
Sediments Global subduction sediments

23
Sediments Thickness

Even though there is no rule of thumb to assess the amount of subcrustally

subducted sediments, the average thickness is commonly in the order of
only a few hundreds of meters.

Somewhere absent: Tonga

Somewhere km in thickness: Cascades

Mid- and South America subduction zones involve large volumes of

carbonate sediments, whereas subductions in the Northern Pacific are
substantially carbonate free but opal rich (hydrated amorphous form of
silica).

Sediment chemistry is substantially controlled by biogenic components, the

source of detrital material (岩屑), and the sedimentation rate

24
Sediments

• Terrigenous material (70–80 wt%)

• Carbonates (10–15 wt%)
• Opal (10–15 wt%) 25
Question
Can the sediments be subducted to the Earth’s deep interior?
If so, could you give the estimation for the subduction depth?

26
 Evidence for Sediments Subduction

• 10Be is a cosmogenic and

radioactive nuclide
• It is concentrated in young marine
sediments
• It does not build up in the
mantle over time
• The presence of 10Be is a great
indicator of the sediment
subduction

27
Thermal Structure of the Slab
Measurements of heat flow along a cross section perpendicular to a
subduction zone follow a standard pattern.
High Low
75-100×10-3 W/m2 40×10-3 W/m2

28
Thermal Structure of Subduction Slabs

The 95-km-thick plate is heated by

the surrounding mantle by
conduction
--- No shear, latent or radiogenic
heating of the slab is considered here

The modeled subducting slabs retain

low temperatures compared with
adjacent mantle, even at depths of
more than 700 km depth.

29
Thermal Structure of Subduction Slabs

Due to the low temperature, slabs

are normally characterized by high
seismic velocities and densities

• A slab subducting at 45o at 5

cm/yr would take only 20 Ma to
reach 700 km depth
• A slab subducting at 10 cm/yr will
reach 700 km depth only in 10 Ma

30
Factors affect the fine details of subduction-zone temperature
• Magnitude of frictional heating shear stresses
Normally in the range of 10-40 MPa based on the values of the heat flow
in the trench-volcanic-arc region (up to 100 Mpa)

• The amount of water

The presence of water can lower the melting temperature and cause
partial melting.
The heat is advected by rising magma beneath the volcanic arc and the
back-arc basin.

31
Factors affect the fine details of subduction-zone temperature

The phase transition occurs continuously with

the sinking slabs.

The olivine to wadsleyite phase transition is

wadsleyite
exothermic (heat-releasing) and at lower
temperatures occurs at a lower pressure.

The wadsleyite to ringwoodite phase

ringwoodite transition is endothermic (heat-absorbing)
and at lower temperatures occurs at a higher
pressure.
wadsleyite

32
Can a slab penetrate the 660-km discontinuity?

Enhances
Subduction

Opposes
Subduction

33
Can a slab penetrate the 660-km discontinuity?

Thermal structures of the slabs control

the occurrence of the phase transition
in the olivine system.

34
Sometimes but not
always mantle slabs
do not penetrate the
670 km discontinuity

35
Stagnant Slab

36
Fukao et al., 2009
Temperature is too high

37
Fukao et al., 2009
Some slabs flatten at the base of the transition zone
Izu-Bonin, Japan–southern Kurile, Ryukyu, Aleutian, and Tonga slabs
38
Goes et al. 2017
At subduction rates of 5 cm/yr and without significant stretching or
thickening, 1000 km corresponds to ~20 m.y. of subduction.

Comparison of tomography with plate reconstructions has led to estimates

of accumulation times, i.e., stagnation times:
~1000-km-long flat Izu-Bonin slab, 20–30 m.y
~1600 km-long Honshu flat slab, more than 15–20 m.y.
~1300 km-long Tonga flat segment, 30–40 m.y.
~500-km-long, Calabrian flat slab, 8–10 m.y.

Schmid et al. (2002) estimated that the Farallon material in the transition
zone corresponds (due to buckling) to subduction since 50–60 m.y. ago.

The deeper flat slab below Indonesia has been associated with opening of
the South China Sea 30–40 m.y. ago (Replumaz et al., 2004).

39
These numbers imply that material can indeed stagnate in the transition
zone for a few tens of millions of years.

However, there is no seismic evidence of older material in the upper

mantle and transition zone.

Because the time scales to diffuse the thermal anomaly of a cold slab and
hence make it seismically invisible are on the order of hundreds of
millions of years, material subducted in the early Cenozoic and Mesozoic
has apparently sunk into the deeper mantle.

40
Some slabs appear to pierce straight through to larger depths
Marianas, northern Kurile, Cocos plate
41
Goes et al. 2017
Stagnant slabs are common in the western Pacific, where very old
lithosphere is currently subducting (90–145 m.y.)

Some deeply penetrating slab sections are found along the eastern rim of
the Pacific, where the lithosphere currently subducting is relatively young
(0–50 m.y.,).

Further complexity:
The penetrating segment of the Izu-Bonin-Marianas slab corresponds to
the oldest lithosphere along this trench

At the Tonga-Kermadec and the Japan-Kuriles-Kamchatka trenches, the

youngest part of the lithosphere penetrates

All lithosphere currently subducting along these trenches has an age

(>90 m.y.) that exceeds that above which plates appear to no longer
thicken and increase their negative buoyancy
42
Although there is some correlation between absolute downgoing-plate
motions and age, there is no correlation between present-day trench
motions and slab morphology or dip
43
Goes et al. 2017
Length of the flat slab associates with the trench motion

Trench retreat
Trench migrates seaward
with time because of the
gravity

44
Goes et al. 2017
 Estimates of total trench retreat over the past ~50 m.y.

Most slabs that penetrate into the lower mantle subducted at relatively
stable trenches, while flat slabs all have a history of significant trench retreat
(>800 km). Goes et al. 2017
45
 Lower Mantle Slabs
Many studies have concluded that the lower-mantle anomalies correspond
to thicker volumes of material than the upper-mantle slabs, even
considering limitations of seismic resolution.

Several studies have estimated that some slabs can be thickens by:
4-5 times: the Cocos, Java, and Hellenic slabs to widths of ~400 km
3 times: lower-mantle slab anomalies below the Indian plate
even 400–700 km: slabs below North America

Such strong thickening has commonly been attributed to slab buckling that
starts in the transition zone

46
 Lower Mantle Slabs

The tomographic evidence shows that slabs experience significant

resistance at the base of the transition zone leading to flattening,
buckling, and thickening, and further deformation to less linear forms on
their journey through the lower mantle.

Comparisons with plate reconstructions constrain the time scales of slab

sinking into the lower mantle to 10–50 m.y. and through the rest of the
mantle to 150–200 m.y.

Even if many subduction zones have been long lived (e.g., along the
circum-Pacific), the slabs are generally not continuous over the full
mantle depth range, indicating that the sinking process is discontinuous,
which may reflect the influence of stagnation in the transition zone.

47
Earthquake in the subduction zone

~90% of the total earthquakes in the world occur in the subduction zone,
which are shallow and large.
48
49
 Large, shallow earthquakes in subduction zones contribute 90% of the
total seismic moment released worldwide

The focal mechanism of the

Large earthquakes is related to
the thrust faulting along
the subduction interface.

Seismogenic zone
• 2–5% of the total downdip
length of the Wadati-Benioff
Zone
• Generates large earthquake

Great (Mw ≥ 8.0) earthquakes from 1900 to 2016

presumed to be on subduction zone megathrusts (green) 50
Bilek et al. 2017
The biggest maximum earthquakes
• Beneath Andean-type arcs -- arcs built on continental crust
• Greater strength of continental crust compared to oceanic crust under
cold, wet forearc conditions

Earthquakes deeper than 50 km

• Controlled by the thermal state of the subducted lithosphere
• Relatively cold subducted slab
• It takes long time to reheat the slab which does not rupture along the
subduction

51
The magnitude of the largest earthquake in a given subduction zone also
seems to depend on the age of the subducted lithosphere and how fast it
is subducting
Young, fast lithosphere has strong coupling between the plates
---- More large earthquakes
Old, slow lithosphere sinks quietly weak coupling

52
Stern 2002
Subduction earthquakes occur in a temperature range of approximately
100oC to 150oC to approximately 350oC to 450oC .

The updip limit of the seismogenic subduction zone may be controlled

by diagenetic to low-grade metamorphic processes such as quartz
cementation and pressure solution

The downdip limit could correspond with the onset of plastic

deformation or a serpentinized mantle

53
Major hydrous minerals in the subduction slab

54
Transportation of hydrous minerals to the deep Earth

55
 Are the lower planes of double seismic zones caused by
serpentine dehydration in subducting oceanic mantle?

56
Peacock 2001
• Dehydration of serpentine may
cause the deep earthquake.
• The presence of serpentine may
also weaken the lithosphere and
help to bend the plate

57
 A cluster of low-frequency events in
the lower crust/uppermost mantle
beneath the volcanic arc seems to be
caused by deep magmatic activity of
mantle diapirs.

The deeper ones, however, clearly

delineate the descending lithospheric
plate and define two almost parallel
planes:
• Upper plane: earthquakes with
reverse faulting or down-dip
compressional stresses
• Lower plane: earthquakes with
down-dip extension stresses.
Cross sections perpendicular to the Japan Trench
in northeastern Japan

58
 Attenuation of the Tonga Subduction Zone

High Q: low attenuation

Low Q: high attenuation

<100
<200

>900

The low-Q region are roughly coincident with the low-velocity zones.
The high attenuation and low velocities above these subducting plates are
consequences of the dehydration of the subducting plate, the rise of water into
the mantle wedge, the partial melting and then the accumulation of melt at high
levels beneath and within the crust. 59
Deformation Between two Earthquakes

Using the 1964 Alaska earthquake

as an example:
• Mw=9.2
• Over 2 m of subsidence behind
• Over 6 m of uplift in front of the
epicenter.

Between earthquakes the seaward

edge of the plate is dragged
downwards while just inland the
flexural bulge means that there is
uplift.

During the earthquake, the locked

zone is suddenly released, causing
the seaward edge to rebound and
the uplifted bulge to subside.
60
Deep Earthquake

61
 Deep Earthquake and the Thermal Parameter

Thermal parameter = age of subducting plate × descent rate

Subduction zones with a thermal parameter less than 5000 km do not have deep
seismicity whereas those with a thermal parameter greater than 5000 km do.
62
Mechanism for the Deep Earthquake
1. Dehyderation Embrittlement

• A tensile stress σ3 opens a microcrack

perpendicular to σ3
• The microcrack concentrates tensile
stress at its tips, promoting further
• growth.
• Brittle shear failure occurs when such
microcracks interact and link up.

63
Double seismic zone

64
Double seismic zone

The upper layer is 15-20

km below the top surface

The bottom layer is ~30 km

below the top surface

65
 Mechanism for the Deep Earthquake
2. Anticrack faulting in a metastable phase

• A compressional stress σ1 promotes transformation to a denser phase in a

lenseshaped anticrack oriented perpendicular to σ1
• The anticrack concentrates compressional stress at its tips, promoting further
growth
• Anticracks can interact, link up, and coalesce to form a plane on which
macroscopic shear slip occurs at an oblique angle to the direction of greatest
principal stress 66
 Mechanism for the Deep Earthquake
3. Thermal Shear Instability

Under certain conditions the

localization of shear strain
increases and accelerates
exponentially, leading to
apparently abrupt failure on
a shear zone.

67
68
69
Number of Earthquakes in the subduction zone

The largest earthquakes occur at the

interface between the subducting and
overriding plates

Large amount of earthquakes occur

inside the sinking slab
--- Intraslab earthquakes

70
Hypothesis ---- Dehydration embrittlement
Slab sinking accompanies with the phase changes in the rocks where
hydrated minerals break down into denser anhydrous phases
• Water is released into the system
• The released water will increase pore pressure and reduce the
normal stress across potential planes of weakness

Although the high-pressure and temperature at greater depth would

normally prohibit earthquake rupture, the presence of water can trigger
earthquakes at intermediate depths (50 to 300 km).

71
Hydrous state of different slabs
Cold subduction
• Corner of the mantle wedge is usually
not serpentinized
• Limited fluid circulation in the core of
mantle wedge
• Yet very low seismic velocities requires
the presence of at least some fluids in
the forearc slab

Warm subduction
• Dehydration happen at shallow depths
• Large amounts of fluid enter the cold
corner of the mantle wedge
• Temperatures are too low for melting
• Fluids transform dry peridotite to
serpentine
72
Problem of Dehydration embrittlement for the cold subduction
In the cold subduction, the onset of dehydration lies at depths greater than
80 km.

Yet there a continuous band of earthquakes along the top of the forearc
slab
--- The segment of the slab extending from the trench down to the onset of
dehydration

Very low seismic velocities requires the presence of at least some fluids in
the forearc slab

Where these fluids come from if the only sources are dehydration reactions
that occur tens of kilometers downdip.

Whether the fluids from the sinking slabs are directly related to the
earthquakes or not
73
Newly relocated seismicity + seismic images + geodynamic models,
Earthquakes effectively track the updip flow of fluids along the forearc
slabs of cold subduction zones

Peloponnese region of the Hellenic subduction zone

Region of Interest
Western Greece
• 230-million year (Ma) old
oceanic part of the African
plate
• Subducts beneath the
Aegean microplate
• Rate: 35 mm/year

74
Subduction zone: weak interface coupling
Shallow megathrust earthquakes: do not exceed magnitude 7 km
Intermediate depth earthquakes: a maximum depth between 90 and 190 km

Peloponnese region of the Hellenic subduction zone

• Wide onshore forearc

region
• Well instrumented to
produce the
comprehensive data
necessary for wide-
aperture imaging and
earthquake relocations

from 2006 to 2017

75
Calculate earthquake locations by Inverting arrival times of seismic waves

Subduciton depth
Peloponnese region of the Hellenic subduction zone

Method
• The double-difference
method + three
dimensional tomographic
model for background
seismic velocities
• Average absolute
hypocenter errors of less
than 1.9 km in all directions

from 2006 to 2017

76
A single Wadati-Benioff zone: a dipping band of seismicity that follows the
slab directly below the plate interface

77
−10 to −15% VP and −8% VS in the oceanic crust relative to the surrounding
mantle
--- Hydrated basalts and gabbros (cannot rule out a contribution from pore
fluid being carried to depth)

78
No low-velocity subducted crust at a depth of 90 km
---- a transition from crustal rock to ecologite
---- releases up to 5 wt.% water to the system
Where does water move to?

79
Earthquakes in three different regions
Mantle wedge Interface Intraslab
• Exotic material of distinct composition and high viscosity in the mantle wedge
• Serpentine dehydration embrittlement
• Pulses of fluids released from the plate interface

80
The subducting crust directly updip from the mantle wedge earthquakes exhibits a
region of diminished seismic activity

81
Seismic results + Thermal-petrologic models
To determine the possible causes of mantle wedge earthquakes

Potential cause for the mantle wedge earthquakes

• Exotic material of distinct composition and high viscosity in the mantle

wedge
• Serpentine dehydration embrittlement
• Pulses of fluids released from the plate interface

X Because it is too cold for deserpentinization to occur

X Will cause velocity perturbations, not consistent with the fairly

uniform, high-velocity mantle wedge beneath western Greece

82
The areas of high Vp/Vs-ratio (>1.8)
in the mantle wedge above the
inferred locus of blueschist-
dehydration at 80 -140 km depth
---- Indicating the release of fluid
and formation of subarc melt.

83
• Most fluids escape directly upward into the mantle wedge through a
plate interface damaged by metamorphic reactions
• Some must make their way updip through the slab as the channel flow

84
Channel flow
The updip channel flow occurs under an intact plate interface
• Sealed owing to shear induced grain size reduction
• Or along a direction of minimum effective pressure that follows the slab
--- To explain intraslab seismicity and could produce the zones of
weakness required to facilitate rupture in the Hellenic forearc slab
updip of 80-km depth

85
To explain the earthquakes in the mantle wedge
A vent is required at the interface to divert the fluid to the mantle wedge
Diminished fluid circulation in the slab segment next to the vent
---- less prone rupture and aseismic segment

86
Aseismic segment in
the subducted crust
next to the vent

87
At shallower depth (<30-50 km), normal brittle failure causes
intraplate earthquakes

88
Conclusions and questions
Cold subduction zones
Intermediate depth earthquakes in the 40- to 80-km depth range is due to
fluids that migrate updip along the slab.

Question: Where does the fluid go?

No evidence: detectable volumes of fluids reach the surface
--- The fluids must react with, and become absorbed in, the overriding
crust and mantle wedge

Recent observations show low VP/VS ratio in the overriding crust

--- Caused by quartz precipitation from slab-derived fluids, and water can
easily become absorbed via serpentinization of cold peridotitic rocks (a
thin layer)

89
90
Hot Spot

91
• This intraplate volcanic center is far from the Japan Trench where
the Pacific Plate starts to penetrate into the mantle beneath Japan
• No deep mantle plume rooted in the base of the lower mantle
92
Group discussion
What is the potential cause for intraplate volcanos like Changbaishan?

93
Previous interpretation for the origin of the Changbaishan

Mantle upwelling and decompression melting at shallow depths

---- Opening of the Japan Sea and overall extension in northeast China causes
a large-scale asthenospheric upwelling

Dehydration of the stagnant slab and upwelling of hydrated mantle

--- Absence of any islandarc geochemical signature in the erupted rocks

Upwelling of normal asthenosphere and decompression melting

caused by the thickening of stagnant slab in the transition zone

Geodynamic modeling
Focused mantle upwelling can be generated both ahead of deep subducting
slab and around the lateral edges of slab

94
Northeast China Extended SeiSmic Array

A slow cylindrical
anomaly is observed
near that appears as a
hole within the
stagnant slab

95
• A slow cylindrical anomaly is observed near that appears as a hole within
the stagnant slab
• 8% velocity contrast between the slow anomaly and its surrounding
96
slabs
The velocity anomaly corresponds to a 800 K temperature difference
97
New mechanism for the Changbaishan

• Fed by a mantle upwelling

originating from below the
660-km discontinuity.

• The upwelling is a result of a gap

between stagnant slabs in the
transition zone that allows return
flow, including superadiabatic
asthenosphere, to rise.

98
Wei et al. 2020, S

99
Hotspot types
Midplate continental (Yellowstone …)
Midplate oceanic (Hawaii, Bermuda …)
On or near ridge (Iceland, Azores, Easter …)

100
 Big plume head generates
large igneous province:
Deccan Traps
Plume head
very large accumulations
(>100,000 km2) of mafic
igneous rocks, erupted or
emplaced at depth within
an extremely short
geological interval (a few
million years)

Reunion island
Plume tail
Mahoney et al., 2002 101
102
Where is the plume head? ---- need to understand origin of the mantle plume
Get subducted? ---Seems vary rare. Only Yakutat terrane southeast of Alaska was
found to be the oceanic plateau subduct to the deep mantle

103
Portnyagin et al. (2008) --- Accreted to the forearc of Kamchatka (rock composition)
Other seismic study and geodynamic modeling:
A 13-19 km thick oceanic plateau can subduct to 100-km depth or deeper upper
mantle

104
Stack SS precursors (SdS) from 45 years of global seismic data to detect
seismic reflectors in the lower mantle

A reflector with a width of 1000-km at 780-820 km depth

105
 810-km reflector?
30% basalt+70% harzburgite
(18%)
weaker than the observed
reflector

Have to be basalt
• Thickness 35 km
• Garnet to bridgmanite
phase transition

106
Oceanic plateau with a thickness of ~30 km

Stagnation: metastable garnet?

Temperature of the normal mantle

at ~800 km depth: 1900-2000 K
1400-1500oC adiabat

107
Phase transition of MORBs

720 km, 1700 K

1200 K

710 km, 1000 K

Litasov and Ohtani (2005)

108
 Along 1200oC adiabat: ~1700 K at 800-km depth
800-km depth Kubo et al. (2002)

1 mm grain size
Litasov and Ohtani (2015) only
observed metastable garnet at
1200oC and 25 GPa.
No metastable garnet exists at
1400-1500oC at 25-26 GPa.

109