Shun-ichiro Karato

Yale University

Department of Geology & Geophysics

New Haven, CT

​

*

1

Partial melting, water, rheological properties

and �the origin of the asthenosphere

*

2

The weak asthenosphere helps plate tectonics to operate.

why is the asthenosphere weak? (T effect, partial melting?)

​

Partial melting occurs in the asthenosphere beneath mid-ocean ridges.

• formation of the oceanic crust + the lithosphere

MORB (mid-ocean ridge) is homogeneous and modestly depleted.

• why?

​

​

Are they boring questions?

New observations (sharp and shallow LAB), a large rheological contrast between depleted

and undepleted materials 🡪 challenges to conventional models

A brief history of the study of the asthenosphere

• 1914 (Barrell): “the asthenosphere”
• 1926- (Gutenberg): the low velocity zone below the lithosphere (asthenosphere = partial melt)
• 1964 (Mizutani-Kanamori): experimental study on the elasticity of a partially melt
• 1973 (Gueguen-Mercier): importance of solid state relaxation for low velocity and high attenuation
• 1975 (Stocker-Gordon): importance of the geometry of melt (dihedral angle)
• 1979- (Waff, Faul, Kohlstedt): experimental studies on melt geometry
• 1984 (McKenzie): theory of compaction (difficulty of melt retention)
• 1984 (Cooper-Kohsltedt): modest effect of partial melting on creep
• 1986- (Karato, Kohlstedt: Paterson): strong weakening effects of hydrogen
• 1986 (Karato): partial melt hardening model (due to hydrogen removal)
• 1988 (Hofmann): a model of depleted upper mantle (residue of continental crust)
• 1992 (Plank-Langmuir): difficulty of partial melting away from the ridges
• 1992- (Jackson): experimental study on anelasticity of dunite
• 1995 (Karato): hydrogen weakening model of the asthenosphere
• 1996 (Hirth-Kohlsedt): extension of Karato (1986, 1995) model
• 1996 (Gaherty et al.): sharp LAB (lithosphere-asthenosphere boundary)
• 1998 (Karato-Jung): further extension of Karato (1995)
• 2003 (Holtzman et al.): deformation of partially molten peridotite
• 2003 (Bercovici-Karato): 410-km melting model for global material circulation
• 2007 (Yoshino et al.): complete wetting at high P
• 2009 (Kawakatsu et al.): a new partial melt model (based on Holtzman et al., 2003)
• 2010 (Jackson-Faul): a model of anelasticity including high-frequency relaxation
• 2010- (Tauzin, Karato): evidence for (global) 410-km melting

​

3

*

Summary

• Partial melting redistributes water 🡪 rheological contrast (Karato, 1986, 1995; Hirth-Kohlstedt, 1996)
• mixing of depleted and undepleted components is difficult
• large and sharp change in seismic velocities
• Direct mechanical effects of partial melting are small (if the melt fraction is small, <1 %, and if melt does not wet grain-boundaries)
• Seismological LAB (lithosphere-asthenosphere boundary) is caused by the sub-solidus processes.
• Geochemical character of the asthenosphere (moderately depleted and nearly homogeneous composition) is due to partial melting at ~410-km.

Mid-mantle melting is important.

*

4

A sharp and large velocity drop at the LAB�(shallow LAB in the old oceanic mantle)

5

*

• Revenaugh-Jordan (1991)
• Gaherty et al. (1996)
• Rychert et al. (2005)
• Rychert-Shearer (2009)
• Kawakatsu et al. (2009)

Key seismological observations on the lithosphere-asthenosphere system

*

6

• A sharp and large velocity drop at the LAB (shallow)
• dV-Q discrepancy (too large dV for the observed Q)
• Anisotropy
• depth-dependent, modest anisotropy (~2-4 %)
• fast direction ~// flow direction (in most regions)
• trench parallel flow (below some slabs) 🡪 decoupling
• low velocity region above 410-km

​

(Long-Silver, 2009) Tauzin et al. (2010)

What is the Asthenosphere?�[What do we need to explain?]

• Geophysical aspects
• low velocity, high attenuation (high electrical conductivity, low viscosity)
• a sharp and large velocity drop at the LAB (Lithosphere-Asthenosphere-Boundary)
• Decoupling between the lithosphere and the asthenosphere
• A thick low velocity layer above 410-km
• Geochemical aspects
• homogeneous, modestly depleted composition

[water content ~ 0.01 wt% (+/- a factor of 2)]

*

7

• Is partial melting needed or a likely mechanism to explain anomalies of the asthenosphere?
• Can continental crust formation explain the homogeneity of the asthenosphere?

​

• revisit sub-solidus model of the asthenosphere
• an alternative model to explain the geochemical characteristics of the asthenosphere
• possible mechanism of lubrication at the LAB

*

8

Models for geophysical aspects�

• Purely thermal model (Birch, 1952; Schubert et al., 1976; Faul and Jackson, 2005) 🡪 Diffuse (age-dependent) LAB 🡪 need “something else”
• Conventional model: asthenosphere = partially molten layer (Gutenberg, 1926; Lambert-Anderson, 1970)

but (1) partial melting is difficult away from the ridges

(2) no strong effect of partial melting on mechanical properties

*

9

• Extensive partial melting is likely beneath the ridge (above ~70 km), but the melt fraction is expected to be small in the asthenosphere away from the ridge.
• Melting at 60-80 km depth in the old oceanic upper mantle is difficult.

10

*

~70 km

ridge

A sub-solidus (hydrogen) model

• Karato-Jung (1998):

partial melting below a ridge

🡪 A sharp water content stratification:

water-rich asthenosphere, water-poor lithosphere

(by partial melting below ridges (Karato, 1986; Hirth-Kohlstedt, 1996))

• A sharp LAB at a constant depth ~70 km (age indep.)
• A small δV/V~1 % 🡨 absorption band model
• Don’t agree with obs.??
• A new partial melt model (Kawakatsu et al., 2009)

​

*

11

depleted

​

undepleted

ridge

~70 km

A new partial melt model

• Kawakatsu et al. (2009)

(based on Holtzman-Kohlstedt model)

• Partial melting below ~60-80 km
• Average melt fraction is small but

there are thin horizontal layers with

high melt fraction (very low velocity)

• low SV velocity, no melt segregation
• age-dependent LAB

​

• Is this model consistent with the observations and the physics of partial melt?

​

*

12

Can partial melting occur at 60-80 km depth (~900 C) ?�Is the LAB depth age-dependent ?

*

13

Kumar-Kawakatsu (2011)

dT=100 K 🡨🡪 dz=30 km

900 C !

Melting at 60-80 km is very difficult.

“Age-dependent LAB” is questionable.

*

14

observed structure

Kawakatsu et al. (2009) model

Holtzman et al. (2003), Kohlstedt-Holtzman (2009)

• Kawakatsu model is inconsistent with geophysical, petrological

and mineral physics observations.

Sub-solidus model?

• How can we explain a sharp and large velocity reduction?
• Karato-Jung (1998): sharp LAB but δV/V~1 % 🡨 absorption band model

​

• Absorption band model is inconsistent with seismological observations (Karato, 1977: Yang et al., 2007) 🡪 high-frequency relaxation mechanisms (confirmed by Jackson-Faul (2010))
• Are there any plausible high-frequency relaxation mechanisms that have large relaxation strength?

*

15

Possible role of high-frequency relaxation

*

16

If high-frequency relaxation mechanisms exist, then dV/V can be larger than

expected from Q.

Grain-boundary relaxation�

*

17

diffusional accommodation

grain-boundary

sliding

[based on Morris-Jackson (2009) model]

*

18

High-frequency peak

🡨🡪 5-10% δV/V

(Jackson-Faul (2010))

Water content is stratified 🡪 shift in the peak freq.

*

19

*

20

Sub-solidus model�(water content layering)

*

21

With plausible water effects (r=1-2), the velocity-depth profile is consistent with obs.

including anisotropy.

Geochemical aspects

*

22

• MORB has homogeneous, modestly depleted composition.
• MORB and continental crust have complementary trace element abundance pattern.

• asthenosphere (MORB source region) = a residue of extensive partial melting that formed the continental crust (Hofmann, 1988)

​

“After separation of the bulk of the continental crust, the residual portion of the mantle was rehomogenized, and the present-day internal heterogeneities between MORB and OIB sources were generated subsequently by processes involving only oceanic crust and mantle.” (Hofmann, 1988)

*

23

*

24

Can mixing occur effectively?

​

• Can mixing occur so effectively ??
• Depleted and un-depleted rocks have largely different viscosity (a factor of 10²-10³)

[preservation of the continental lithosphere]

• Materials with largely different viscosities do not mix

[For efficient mixing, strain larger than ~10 is needed.

For a viscosity contrast larger than 10², the hard materials (depleted materials) do not deform more than ~1 strain]

​

​

​

*

25

Manga (1996)

Homogeneous, modestly depleted asthenosphere by mid-mantle melting

• Evidence for 410-km melting
• Low velocity layer
• Water content layering

*

26

Karato (2011) Tauzin et al. (2010)

What happens after 410-km melting?

*

27

a thick low velocity layer

(due to complete wetting)

• Most of the upper mantle

is partially melted (with a

small melt fraction) 🡪 modest depletion

• Composition of the upper

mantle is controlled by solidus composition and homogeneous.

• Melt sinks in the deep upper mantle.
• Melt rises to the LAB in the shallow asthenosphere.

🡪 frozen wet gabbro

Trench parallel anisotropy helped by lubrication by a wet gabbro?

*

28

*

29

Melting in the lower mantle (~700 km)

*

30

(MgO-FeO-Al₂O₃-SiO₂)

Summary

• Conventional models of the asthenosphere (partial melt, residual of continental crust formation) are inconsistent with geophysics/mineral physics observations.
• Most of geophysical and geochemical characteristics of the asthenosphere can be explained as a result of indirect influence of partial melting (at ~410-km and at ~70 km) that redistributes water (hydrogen).
• Seismic properties (low velocity, anisotropy) are controlled mostly by water not by partial melting (except just above 410-km).
• Melting in the mid-mantle has important effects on the geochemical evolution of Earth.

​

*

31

Testing the model for the upper mantle

*

32

pyrolite (olivine+opx+pyrope), SIMS water calibration

​

[Dai and Karato (2009)]

A sharp boundary, but a small velocity reduction 🡪 need partial melting??

33

*

Water weakens grain-boundaries

*

34

Electrical conductivity and water in the mantle

*

35

Mineral physics model

Geophysical model