Rocks and Earthquakes from Deep Subduction Zones: What can They Tell us About Water Recycling, Planf

1
Rocks and Earthquakes from Deep Subduction Zones
What can They Tell us About Water Recycling,
Planform of Mantle Convection, and Ocean Island
Basalts?

• Harry W. Green, II
• Institute of Geophysics and Planetary Physics and
  Department of Earth Sciences, University of
  California, Riverside

2
Outline

• Introduction
• Dehydration embrittlement
• Petrology of subducting lithosphere
• Tests for H2O in deep slabs
• Conclusions I Deep Slabs are Dry
• Sediments have been subducted to gt 350 km.
• Conclusion II They are the likely source of the
  geochemical signal in OIBs.
• Very high pressure phases in an ophiolite
• Speculation Carried from deep mantle?

3
Introduction

• We know that significant H2O is subducted and
  that much comes back in arc- and back-arc
  volcanism.
• We know that dehydration of hydrous phases can
  trigger earthquakes in subducting slabs indeed
  it is the only known viable mechanism for
  earthquakes between 100 and 350 km.
• It has been shown that tectonic stresses are not
  necessary to generate slab earthquakes -- local
  stresses, such as generated by dehydration of
  serpentine or breakdown of metastable olivine,
  are sufficient.
• Therefore, predictions can be made as to where
  earthquakes should occur and where they should
  not occur if slabs are wet.
• Here I examine these predictions and other
  observations and conclude that subducting slabs
  are wrung mostly dry by 400 km
• Recent discoveries in surficial rocks carry deep
  signals of subduction and perhaps plume transport
  from very deep.

4
Subduction Nicaragua Trench
5
Dehydration of Antigorite
6
Dehydration and Earthquakes
Crustal hydration at ridges 10-12 km deep
No equakes in mantle wedge
7
6 GPa
1.0 GPa
6 GPa
8
Horizontal Slab Sections Have Chaotic Focal
Mechanisms
Therefore, Local Stresses Sufficient for Equakes
9
Summary So Far

• Water goes down in slabs and comes back in arc-
  and back-arc volcanism
• Earthquake pattern follows antigorite dehydration
  boundary in slab and dehydration of antigorite
  under stress yields faulting.
• Surface faults dip toward trench
• Shallowest equakes are underthrusting on plate
  boundary
• Deeper equakes (gt 100 km) are inside the slab
  occur on subhorizontal faults (hence, NOT
  reactivated surface faults)
• Tectonic stresses are not necessary for
  earthquakes
• 7.

10
What Causes TZ Earthquakes?
11
Dense Hydrous Magnesium Silicates Incorporate
Progressively More H2O with Increasing Pressure
(hence probably no earthquakes) but can be stable
only after the anhydrous phases are
saturated.
Mg2SiO4 20 H2O
(After Angel et al., 2001)
12
Faulting Due to Dehydration of Nominally
Anhydrous Phases
P 3 GPa F 0.1
Faulting in wet eclogite
(Zhang et al., Nature, 2004)
13
Glass-Filled Mode I Cracks in Eclogite
14
Transformation-Induced (Anticrack) Faulting
Mechanism
Mg2GeO4 olivine Arrowheads point to anticracks
Green and Burnley (1989)
15
TRANSFORMATION- INDUCED FAULTING IN MANTLE
OLIVINE -- (Mg,Fe)2SiO4 -- AT 14 GPa (450
km) Arrowheads point to anticracks
Green et al., Nature (1990)
16
Summary So Far

• 7. Dehydration embrittlement is responsible for
  many, maybe all earthquakes between 50 and
  350 km.
• 8. Presence of dense hydrous magnesium silicates
  (DHMS) requires saturation of the olivine
  polymorph present (olivine, wadsleyite,
  ringwoodite).
• 9. However, if slabs contained sufficient water
  to do this, slabs would be too buoyant to enter
  the transition zone.
• 10. Therefore, if H2O is present in the
  transition zone it must be in wadsleyite and
  ringwoodite.
• 11. In the laboratory, both dehydration of
  nominally anhydrous phases and transformation of
  metastable olivine under stress lead to faulting.

17
THREE POPULATIONS OF EARTHQUAKES

• Dehydration
• Probably Dehydration
• Transition zone
• Dehydration?
• Transformation-
• induced faulting?

1
2
3
18
TONGA
Evidence strong that 60 of olivine pyroxene
remains untransformed in the detached slab
W-P Chen 1/01
Brudzinski Chen (2003)
19
Marianas Subduction Zone
20
Mariana Metastable Olivine Wedge
Kaneshima et al., EPSL (in press)
21
Summary So Far
11. Generation of ltlt1 fluid under stress is
sufficient to trigger earthquakes. 12. If 11 is
correct, dehydration of ringwoodite as the slab
enters the lower mantle should trigger
earthquakes or the H2O might be transferred to
Phase D -- in which case there should be a flurry
of earthquakes at 900 km when Phase D breaks
down. 13. Evidence strong for metastable olivine
in Tonga and Marianas
22
Mariana Deep Slab
The slab must be dry
23
Hypothetical Earthquake Distribution for Wet
Marianas Slab
24
Conclusions I

• H2O goes down in slabs and returns in arc- and
  back-arc volcanism.
• Surface faults dip toward trench.
• Equakes deeper than 100 km are inside the slab
  occur on subhorizontal faults (hence, NOT
  reactivated surface faults).
• Tectonic stresses are not necessary for
  earthquakes.
• If slabs contained sufficient water to stabilize
  DHMS, slabs would be too buoyant to enter the
  transition zone.
• Therefore, if H2O is present in deep slabs, it
  must be in the nominally anhydrous phases,
  principally wadsleyite and ringwoodite.
• Presence of significant H2O would preclude
  metastable olivine.
• Evidence for metastable olivine is strong in
  Tonga and Marianas.
• Presence of metastable olivine and absence of
  earthquakes in places predicted if H2O is present
  strongly suggest that subducting slabs lose
  essentially all H2O by 400 km depth.
• Therefore subduction does not recycle H2O to the
  deep mantle, at least not at the present time.

25
Can rocks currently at the surface contribute to
understanding of the deep mantle?

• Former stishovite in a deeply subducted pelite.
• Former stishovite (?) in an ophiolite.

26
Deep Subduction of Continental Material

• In the 1980s and 1990s it was established that
  continental material can be subducted to depths gt
  200 km during continental collision and shortly
  thereafter returned to the surface.
• It also has been shown by Irifunes group that
  such rocks subducted to 350 km become more dense
  than ambient mantle and will thus continue
  sinking to at least the bottom of the transition
  zone.
• We have now found the smoking gun of pelitic
  rocks subducted to the threshold of this point
  of no return.

27
Deep Subduction and Exhumation of Continental
Material
High Plateau
28
Subduction Setting
29
Natural Microstructures-Quartz
30
Precipitates Unrelated to Host Quartz
31
Coesite Wont Work Either, But Stishovite is
Possible Chemically
(e)
(b)
32
Connecting Stishovite to the Observations
2
001
1
225
631
631
361
225
053
361
361
7
053
3
4
361
5
6
4/m 2/m 2/m
33
All quartz domains have similar groups of ppts
that represent former stishovite crystals of
other orientations.
a
b
503
503
2
2
001
001
1
1
100
100
225
631
631
361
225
053
361
225
010
225
010
361
503
503
7
7
053
3
3
4
4
361
631
5
631
5
631
631
6
6
225
225
c
d
001
1
1
3
225
225
3
503
503
2
2
053
4
631
4
9
631
361
361
100
9
5
631
5
631
8
8
010
361
361
053
7
6
225
6
7
225
e
f
34
Conclusions II

• Quartz cannot explain chemistry, geometry or
  symmetry of observations.
• Precipitates cross high-angle quartz boundaries
  unaffected therefore quartz must be a secondary
  mineral formed after precipitate formation.
• Coesite cannot explain chemistry or symmetry of
  observations.
• Stishovite is consistent with all observations.
• Implied solubility of Al and Fe
• Geometry, symmetry and precipitate
  crystallography
• The simplest explanation of the data is that
  these pelitic sediments were subducted to 350 km
  and returned, making it virtually certain that
  other rocks have been subducted past the point
  of no return and continued on into the deeper
  mantle, demonstrating a likely mechanism for the
  continental signal in OIB geochemistry.

35
Very Deep Mantle Material Carried to the Surface?

• Diamond and coesite have been found in chromitite
  of an ophiolite that shows no evidence of
  subduction or shock (Yang et al., Geology (in
  press).
• The coesite appears to be pseudomorphic after
  stishovite.
• The diamond is included in OsIr alloy and the
  coesite is attached to an Fe-Ti alloy pellet
  showing intermetallic compounds that are not
  stable at low pressure.
• The coesite contains inclusions of BN (previously
  unknown in nature) and TiN (osbornite) that is
  found only in meteorites (both iron and stony)
  and 3 terrestrial occurrences - two of which are
  in carbonado.
• We have done a single experiment that shows
  osbornite is stable at 10 GPa.
• Preliminary data on N isotopes suggest a
  terrestrial origin.
• To the graybeards of the audience this may begin
  to sound like the josephinite saga of the 1970s.
• Could this really be deep material brought up in
  a plume?

36
Luobusa chromite deposit
Yalunzangbu ophiolite177-126 Ma ocean basin65Ma
close
37
Diamond in Os/Ir Alloy
c
38
Fe/Ti Alloy and Silicate Fragment
39
Coesite, Kyanite, Fe-Ti Alloys
40
Summer Solstice Eve in the Western Norway UHPM
Terrane
41
The End
42
Precipitate Orientations
43
Precipitate Accommodation
Kyanite 001 needle parallel to 001 in
Stishovite
Image simulated by Krassimir Bozhilov
44
Precipitate Accommodation
Kyanite 001 needle parallel to 001 in
Stishovite
Images simulated by Krassimir Bozhilov