How Deep Mantle Structure Constrains the Temperature of Earths Core

1
How Deep Mantle Structure Constrains the
Temperature of Earths Core

• John Hernlund
• UCLA
• ASU DEEP Seminar
• March 9, 2005

2
Partners in crime
Tine Thomas (Liverpool)
Paul Tackley (UCLA)
Disclaimer Neither Tine nor Paul are to be held
legally responsible for any outrageous claims
made in this talkunless of course, said claims
are proven successful.
3
Talk Format

• Seismic discontinuities, the geotherm, and
  post-perovskite phase transition.
• Ultralow-velocity zones, stability, partial
  melting, core-mantle reactions, etc..
• Synthesis Implications for the temperature of
  Earths core.

Note Ive structured this assuming you have a
good background knowledge of lower mantle seismic
structure. As a result, some things might be
unclear, in which case I encourage questions!
4
Discontinuities and Phase Changes
5
Seismic Discontinuity atop D

• 2.5-3.0 jump in Vs 100-300 km above the CMB
• Many areas lack coverage
• Not laterally continuous?

Lay et al., 2003
6
Seismic velocity models for D

• Simple profiles that vary by region
• Usually, but not always, a negative gradient
  modeled below the discontinuity
• Vs jumps of 0.2 km/sec

Lay et al., 2003
7
Earth Structure, pre-2004
?
?
?
?
?
?
8
Poem What on Earth is D?

• A discontinuity seems to be there,
• Turning back waves from the top of the layer,
• But a constant appearance for this jump is rare,
  
• And as to its cause, well, we havent a prayer
• -Eddie Garnero Michael Wysession, Eos, 2000

9
Cause of Discontinuity in D?
Dense Layer
Thermal Gradients
Phase Change

• Sidorin et al., 1999 compared scenarios using
  convection models, and synthetic seismograms
• Best fit phase change with 6 MPa/K Clapeyron
  slope
• No suitable phase change was known at the time

10
The Post-Perovskite Transition

• Est. Clapeyron slope 7-10 MPa/K
• Density change 1

Oganov and Ono, 2004
Iitaka et al., 2004
11
Relationship to Geotherm?

• Predicted phase boundary can be hotter or colder
  than CMB
• If colder, then a double-crossing should occur
• Distinct observational consequences 1 or 2
  discontinuities?

12
Two Discontinuities in D

• Seismic migration technique applied to D
• Reveals additional discontinuity beneath Eurasia
  and the Caribbean region
• Good agreement with the hot core model,
  double-crossing

Christine Thomas Co., 2004
13
Proposed Model for D Structure

• Explains patchiness of discontinuity observations
• Explains double discontinuities
• Reveals thermal boundary layer structure

14
Synthetic Seismograms
Data (Eurasia)
Cold
Warm
Hot
15
Synthetic Migration
16
Cartoons to Convection Models

• Takashi Nakagawa and Paul Tackley
• Self-consistent generation of double-crossing
  structures
• Over time as core cools, will become
  single-crossing of phase boundary

17
Implied Thermal Structure of D

• Thermal gradient must be higher than phase
  boundary gradient for double-crossing
• 7-10 MPa/K
• 5.8-8.3 K/km
• Heat flux conducted along phase boundary is the
  minimum heat flux in places where discontinuities
  exist
• Using k10 W/m-K, this is 58-83 mW/m2
• Extrapolated CMB heat flow
• O(10 TW) or higher

18
First-Order Phase Structure
19
Implications Thermal Boundary Layer and Core
Temperature

• CMB T higher than pPv phase boundary
• Inner-outer core boundary T higher than 5200 K
• D Thermal boundary layer T change gt900 K

20
Implications Lateral Temperature Changes

• T changes of 2000 K recorded by seismic
  discontinuities.
• In Caribbean region, up to 1700 K over few
  hundred km
• In Eurasia region, up to 1300 K over few hundred
  km

21
Implications Fate of Subducted Slabs

• Observed range is in good agreement with
  temperature changes estimated from
  super-adiabatic slabs and plumes
• Additional evidence for subduction of slabs into
  D

22
D Discontinuities
Lithgow-Bertelloni and Richards scheme fair
match between expected slab locations from plate
motion history and discontinuities in D.
Mid-Pacific discontinuity may represent even
older (warmer) material swept up by
circum-Pacific slab driven circulation. Antarctic
D?
23
Post-perovskite Conclusions

• Post-perovskite phase transition accounts for two
  seismic discontinuities in D
• Absence of discontinuities in some regions can be
  explained by hot mantle
• Implied core-mantle boundary heat flow on the
  order of 10 TW
• Implied lateral temperature variations consistent
  with whole-mantle convection, and the subduction
  of oceanic lithosphere into lowermost mantle

24
On to ULVZ

• Some observations, and random thoughts
• Some dynamical constraints, models
• Assessing their origin

25
Where are ULVZ?
Thorne Garnero (2004)
Lay Garnero (2004)
26
Where should ULVZ be?
Anything existing in the lowermost mantle should
be swept around by mantle flow, and be
concentrated where the flow along the CMB
converges
27
Simple Stuff

• If they are stable features, they should maintain
  an isostatic balance, much like Earths crust.
• Note that this balance requires ULVZ density
  intermediate between mantle and outer core.

28
SurfaceVolume Ratio

• Consider cylindrical tablet of ULVZ stuff
• Mantle viscous stress keeps ULVZ from flattening
  out
• At equilibrium, buoyancy forces balance mantle
  forces

Isostatic Balance
Force Balance
Required Stress
(lower bound)
29
Simple Scenario, Contd

• Required stress proportional to height and
  density change.
• Must be supplied by focused and active upwelling
  flow w/thermal buoyancy.

Available Stress
30
Conical ULVZ

• Required stress to support a conical ULVZ has two
  equilibrium points, stable and unstable.
• A small aspect ratio structure is an unstable one.

Also The aspect ratio is a good proxy for
inferring density
31
Density Differences
Can we distinguish based upon length scales?
If ULVZ structure varies drastically according to
density of regional surrounding mantle, its
density variations must be comparable
32
ULVZ origin hypotheses

• Reaction products between core and mantle?
• Entrained core fluid?
• Sediments from outer core over-saturation?
• Partial melting of the mantle itself?
• Combination of these?

33
What are ULVZ, really?

• Patches of partial melt of limited horizontal
  extent?
• Alternatively, a highly variable thickness layer
  of slow Seis. Vel.
• How thick?

Thorne Garnero (2004)
34
Core-Mantle Reactions
Silicate reaction (Knittle and Jeanloz, 1991)
Aluminum reduction (Dubrovinski et al., 2001)

• In either case, reaction limited to small
  fractions
• Some reaction products are metallic, and liquid

35
Core-Mantle Reactions

• Requires core-mantle chemical disequilibrium
• Primordial disequilibrium Theyve always been
  that way from early Earth? (Geochemistry problem)
• Driven by cooling temperature decrease changes
  the equilibrium constant? (Only 100-200 K max)
• Inner core growth excludes light elements,
  saturating outer core? (Inner core only 5 of
  core mass! Outer core L.E. enrichment is only
  0.5)

36
Core-Mantle Reactions?

• Regardless of the particular reaction, the
  products will be more dense than mantle, less
  dense than outer core, i.e. self-shielding
• Advance of reaction front is thus limited to
  diffusion through any metallic phases in the
  products

37
Core-Mantle Reactions?

• Plus Produces mixture of liquidsolid
• Minus Associated with huge density increase
  relative to surrounding mantle, induces very
  small topography
• Minus Limited by chemical diffusion through
  reactant, which might not even present a pathway
  to fresh material
• Minus Increase in gravitational potential energy

38
Entrained Core Fluid?

• Perhaps outer core fluid is drawn upward by
  capillary action? (Kanda and Stevenson)
• Problem competing estimates of surface tension
  betwixt Fe and silicates
• Problem even with the most generous parameters,
  can only be drawn up less than one
  kilometerprobably too thin

39
Core Sediments?
(Buffett et al.)

• Knittle and Jeanloz rxn run backward?
• Over-saturation of light elements in outer core
  do to IC growth?
• Enrichment in OC light elements is very small
  thus far, (0.5)
• Grain size? Miscibility?

40
Partial Melting of Mantle?
Zerr et al.

• Simple mechanism solidus lt CMB temperature

41
Phase Diagram?

• Fe-Pe has smaller melting temperature at CMB
• Partial melt likely to be rich in FeO
• Melt density could be greater or smaller,
  depending on structure and FeO content

42
Partial Melting Everywhere?

• Outer core has constant T, so melting in one
  place implies melting everywhere?
• Perhaps there are fertile and refractory
  patches? (Jellinek and Manga)
• Perhaps partial melt layer is mostly very thin
  except for ULVZ? (Revenaugh, Ross, etc.)

43
What really happens with partial melting of the
lowermost mantle?

• Difficult to intuitively predict
• Some said whole D would be melted
• Others said it would be swept up in flow
• Lets look at some models to see

44
Transient Melting Models

• Transient, start cold, and allow TBL to grow
  until instability
• Light melt rises in the form of diapirs
• Dense melt lays in flat layers
• Segregation causes light melt to freeze quickly,
  dense melt to assume a lower profile along CMB

45
Transient Melting Models

• Note dramatic variations in thickness of
  partially molten zone
• Non-linear feedback between thermal convection
  and melt density distribution
• Melt distribution depends strongly on the
  relative density of surrounding mantle

46
Axi-symmetric steady cases
Resulting Melt Fields
Typical Temperature Field
Radius -gt
47
Density vs. Segregation
48
Partial Melting vs. CMB Rxn

• Partial melting front advances with thermal
  diffusion velocity, CMB Rxn front advances with
  chemical diffusion velocity
• Partial melt density changes can only be a few
  percent, CMB Rxn density changes up to 50

49
ULVZ Conclusions

• Partial melting provides a simple mechanism for
  explaining ULVZ
• Dramatic variations in partial melt distribution
  are produced by dynamic feedback
• Low rates of segregation required to avoid
  rising/freezing of melt, or sinking/rapid
  accumulation of melt at bottom of mantle

50
Synthesis
51
Synthesis
Double-crossing of phase boundary and mantle
melting imply that both the post-perovskite phase
boundary and solidus represent lower bounds on
the temperature of the CMB. Implied CMB
temperature consistent with Alfé et al. ab initio
study, and many others using shockwaves
52
And the temperature of Earths core-mantle
boundary

• 4300200 K

Disclaimer Uncertainties remain in the
post-perovskite phase boundary and solidus
values, however, many independent lines of
evidence are converging upon this range.
53
Acknowledgements
54
And MYRES, of course