Lecture 18: Chemical Geodynamics, or Mantle Blobology

1
Lecture 18 Chemical Geodynamics, or Mantle
Blobology

• Questions
• What can geochemistry tell us about the deep
  interior of the Earth?
• Is the mantle homogeneous and if not how many
  reservoirs are there? How long have they
  maintained their separate identities?
• How do we use radiogenic isotope ratios and trace
  element ratios in basalts to make such inferences
  about the mantle?
• Reading
• Albarède, Chapter 8

2
Summary of Earth Differentiation
(nucleosynthesis, mixing)
Solar Nebula
(volatiles)
(gas-solid equilibria)
(refractories)
Condensation and Accretion
(late veneer)
(continuing cometary flux?)
(siderophile chalcophile)
(melting gravity and geochemical affinity)
(atmophile)
(lithophile)
(lost due to impacts)
Core
Silicate Earth
Primitive Atmosphere
(freezing)
(catastrophic impact)
Moon
Primitive Mantle
Inner Core
Outer Core
(partial melting liquid-crystal partitioning)
(?)
degassing
Upper Mantle
Lower Mantle
Continental Crust
(plate tectonics partial melting, recycling)
(hotspot plumes)
degassing
Modern Ocean Atmosphere
Oceanic Crust
3
Geochemistry and Geodynamics

• A range of models have been proposed

4
Geochemistry and Geodynamics

• Our only data about the history of the Earths
  structure is derived from geochemical inference,
  because geophysics only samples the present
  (exception paleomag)
• However, geochemistry only samples the surface,
  so inferences about depths within the Earth are
  indirect, and must be supplemented by geological
  or geophysical constraints.
• In some cases, mantle samples are directly
  available as xenoliths or peridotite massifs, but
  mostly the mantle delivers its chemical signals
  to us in basaltic magmas.

5
Geochemistry and Geodynamics

• What information in a basalt can be taken as
  direct information about the source region?
• Not major element compositionpartial melting and
  shallow differentiation both separate major
  elements from one another in complicated ways
• Not trace element concentrationeven knowing all
  the partition coefficients, these are functions
  of extent and style of melting as well as source
  composition
• Stable isotopes, maybe, if high temperature
  fractionation is negligible
• Ratios of incompatible trace elementsyes. If
  both elements are sufficiently incompatible that
  they are quantitatively extracted, then liquid
  ratio equals source ratio.
• Ratios of heavy long-lived isotopesyes.
  Arguments based on diffusion strongly suggest
  that basalts are produced in isotopic equilibrium
  with their source.

6
Heterogeneity of Oceanic Basalts

• Observation while less diverse than continental
  rocks, oceanic basalts do display a significant
  diversity of isotopic compositions in 87Sr/86Sr.
• Focus on oceanic basalts because they are
  uncontaminated by continents.

MORB mid-ocean ridge basalt OIB ocean island
basalt
7
Isotopic Equilibrium and Disequilibrium

• So heterogeneous isotopic compositions come out
  of the mantle. What does this mean about the
  heterogeneity of the mantle itself?
• The essential argument for isotopic equilibrium
  between source and melt was presented by Hofmann
  and Hart (1978). Consider two cases
• (1) The mantle is uniform on a regional scale
  (10-1000 km3) due to efficient mechanical
  stirring, but not in chemical or isotopic
  equilibrium on a local (cm) scale due to
  inefficient diffusion.
• In case (1), isotope heterogeneity in erupted
  basalts might reflect, for example, different
  degrees of melting if radiogenic Sr accumulates
  in phlogopite and is contributed to the melt only
  as phlogopite melts.
• (2) The mantle contains regional inhomogeneities
  that have survived the stirring process for long
  times, but is isotopically equilibrated by
  diffusion on a local (100 m?) scale at least
  during melting.
• In case (2), isotope heterogeneity in erupted
  basalts reflects regional-scale difference in
  their source compositions only

8
Isotopic Equilibrium and Disequilibrium

• Case (1) Regional homogeneity, local
  disequilibrium
• http//wwwrses.anu.edu.au/gfd/members/davies/pages
  /passmovie.html
• Case (2) Regional heterogeneity, local
  equilibrium
• http//www.gps.caltech.edu/gurnis/Movies/MPegs/st
  irring.mpeg

9
Isotopic Equilibrium and Disequilibrium

• Isotope heterogeneity on the meter scale can
  certainly persist for long times in the solid
  state
• Typical diffusion coefficients of trace elements
  in mantle minerals are of order 1012 cm2/s at
  1200C.
• Hence typical timescale for diffusion across 1 m
  distances is t L2/D 3 x 108 years
• Stirring of viscous fluids stretches and thins
  heterogeneities but it also takes many millions
  of years to thin them to diffusive lengths.
• Isotope heterogeneity on cm scale probably cannot
  survive a melting episode
• Typical diffusion coefficients in silicate
  liquids are of order 10-7 cm2/s at 1200 C.
• Hence typical transport distance by diffusion is
  2 cm per year or 200 m in 10 ka.
• As soon as partial melt fills all the grain
  boundaries, the distance over which solid-state
  diffusion must act drops from the scale of
  heterogeneity to the size of a crystal!
• It follows that basalt liquids are expected to
  have isotope ratios that are faithful copies of
  their sources averaged over at least several
  meters.

10
Isotopic Equilibrium and Disequilibrium

• We can see evidence of this in the comparison of
  isotopic composition between basalts and
  associated residual peridotites
• Basalts are more homogeneous and more radiogenic
  than peridotite suites. Taken to imply that Nd
  (and Os) from a recycled component was in the
  source but is not sampled in the residual
  assemblage.
• Consistent with regional heterogeneity, local
  homogenization

11
Isotopes in Oceanic Basalts

• What then is the interpretation of the pattern of
  Sr isotope heterogeneity among MORB and OIB?
• Sr by itself is very hard to interpretwe dont
  know bulk earth value because Rb is volatile on
  accretion
• Sr and Pb isotope variations do not correlate in
  any simple way, which caused much gnashing of
  teeth 30-40 years ago
• It took the introduction of Nd isotope data to
  begin a real debate between meaningful models
• Sm and Nd are refractory, so we know CHUR
  composition and by inference BSE
• Sr and Nd isotopes in oceanic rocks do correlate,
  inversely
• MORB and crust are seen to be complementary
  (recall trace element story from lecture 2), but
  the meaning of OIB is ambiguous

the mantle array
12
The Sm-Nd mantle array

• The distribution of OIB data between MORB and
  Bulk Silicate Earth is consistent with at least
  three models

The standard model -- MORB samples the upper
mantle which is complementary to continental
crust extraction OIB samples the lower mantle
which is primitive the mantle array is the
result of mixing between depleted and
primitive. Or, different parts of the mantle may
have been depleted to various degrees and never
homogenizedthis would also generate an array of
data from depleted to primitive, but with a very
different spatial distribution of mantle
reservoirs! Or, there may be no primitive
reservoir involved at all, and OIB may be
mixtures between depleted MORB mantle and various
enriched components like recycled oceanic crust
or subducted sediment There might still be a
primitive mantle somewhere, but it might not ever
be sampled by volcanism
Hofmann and White model
13
Isotopic Mass Balance

• Knowing eNd for bulk silicate earth 0 eNd,
  Nd and mean age of continents and eNd for
  upper mantle, can we distinguish standard and
  whole-mantle models by mass balance? Lets
  calculate what volume fraction of the whole
  mantle must be depleted to balance the
  continents.

14
Isotopic Mass Balance

• For times short compared to the half-life of
  147Sm,
• Or, in epsilon notation, with initial eNd 0,
• IF
• There are only three reservoirs c, d, and p (and
  p is primitive)
• We know the Sm/Nd ratio of the crust, Nd of the
  crust, and the eNd of depleted mantle
• THEN we get a relationship between the age T of
  crust formation and the ratio of the masses of
  crust and depleted mantle
• The result, for T 2.5 Ga (which we get
  independently from ƒSm/Ndc and eNdc), is that the
  depleted mantle is 0.3 time the mass of the whole
  mantle.
• This fits beautifully with the standard model,
  since the upper mantle is 1/3 of the mantle.
• BUT if there is another large reservoir, namely
  stored subducted materials, this messes up the
  whole calculation. We can easily put enough
  enriched material with eNd gt 0 in this reservoir
  that the entire remainder of the mantle would be
  depleted mantle. So this is equally consistent
  with the Hofmann and White model!

15
The Sm-Nd mantle array

• How do we choose between these models? For
  starters, get more data and more isotope systems!
• Problem 1 with standard model with more data, we
  find that OIB extend beyond primitive mantle
  (PRIMA) composition, both to higher 87Sr/86Sr and
  lower eNd. Hence they must contain some enriched
  material.
• Problem 2 with standard model the array is not
  consistent with two-component mixingthe width of
  the trend is way outside analytical error and
  requires at least two enriched components.
• Problem 3 with standard model the MORB data are
  spatially organized by ocean, so the upper mantle
  is not homogenous either
• Problem 4 with standard model add other isotopes
  and the binary-ish mantle array breaks down
  altogether

16
The mantle isotope zoo

• So how many components do you need?
• For Sr-Nd-Pb-Pb-Pb space, at least four
• DMM depleted MORB mantle
• HIMU High U/Pb component
• EMI Enriched Mantle I (low Nd)
• EMII Enriched mantle II (high Sr)
• If 206Pb, 207Pb, 208Pb are not really
  independent, then four end members to span data
  in 3-space (Sr-Nd-Pb) is trivial, but the same
  components also bound data in Hf and Os space.

17
The Worm-o-gram

• How do the four bounding components mix with one
  another?
• Is there evidence of an internal component,
  that everything mixes towards? If so, what is it?
• Some authors see mixing towards particular
  locations, and argue that these represent common
  components with well-defined compositions FOZO,
  C, PREMA
• More on this when we talk about noble gas systems.

18
An oddity

• The DUPAL (Dupré and Allègre) anomaly nearly all
  the isotopically unusual hotspots are in a
  well-defined latitude band between 0 and 50S.
• If this has any geodynamic significance, nobody
  has figured out what it is!

Getting back to geodynamics...

• So what are DMM, HIMU, EMI, and EMII? Are they
  well-defined reservoirs with sensible histories
  and physical locations in the mantle, or merely
  arbitrary points in multi-isotope space?
• Before we can answer that we need to think more
  about trace elements, since parent-daughter
  ratios over time determine the isotope
  characteristics of the end members.

19
Trace Element Ratios

• Another kind of tracer of mantle sources should
  be ratios of incompatible elements in basalts,
  but one has to be careful to avoid effects of
  recent fractionation
• Two cases that do not work Sm/Nd and Lu/Hf

• Nearly all MORB samples plot above Bulk Earth in
  Hf and Nd isotopes, meaning their long term Lu/Hf
  and Sm/Nd ratios have been higher than
  chondritic. But nearly all MORB samples have
  subchondritic measured Lu/Hf and Sm/Nd ratios.
• It follows that Lu/Hf and Sm/Nd were fractionated
  recently (by the melting process itself), which
  turns out to requires garnet in the source (P gt
  2.5 GPa).

20
Trace Element Ratios

• Two that do work, for MORB OIB melting Nb/U
  Ce/Pb

Nb/U and Ce/Pb in oceanic basalts do not
correlate with Nb and Ce. This implies (1)
that the ratio in basalt does not depend on
extent of melting, and (2) that depleted and
enriched sources are equal also, so the ratio in
the residue does not get fractionated. Hence
either the elements have
equal partition coefficients or are both
incompatible enough to be totally extracted. But
the ratio in MORB and OIB is not chondritic! The
continent-forming process did fractionate these
element pairs (either because arc processes
involve oxidizing fluids or because of very small
extents of melting), and crust and mantle are
complementary reservoirs. ButOIB do not mix
towards primitive value, so there is no evidence
here of a primitive reservoir sampled by any
basaltic magma!
21
Trace Element Mass Balance

• If we know the Nb/U ratio of the primitive
  mantle, depleted mantle, and continental crust,
  we should be able to calculate the masses of each
  of these reservoirs.
• UCCXCC UDMMXDMM UBSE
• NbCCXCC NbDMMXDMM NbBSE
• XCC XDMM 1
• -gt
• (Nb/U)DMM 47, XCC (relative to whole silicate
  earth) 0.6, UCC 0.9-1.3 ppm.
• Conclusion it does not worksomething must be
  missing, because the continental crust appears to
  be 0.7 to 1.15 of the crustdepleted mantle
  system. Either there is a hidden reservoir of Nb
  or U somewhere, or some fraction of the mantle
  remains primitive and is not sampled by either
  MORB or OIB.
• Possible hidden reservoir is again subducted
  oceanic crust, perhaps eclogite with rutile to
  hold a lot of Nb

22
Anomalous fractionations involving continents

• Why are some trace element ratios different in
  continents than in mantle, even though basalt
  genesis does not fractionate them? Lets look at
  Ce/Pb again

• Which is the anomalous element, Ce or Pb? In the
  spidergram, Pb clearly stands out as high in CC,
  low everywhere else.
• Where do the continents get this signature?
  Where are continents made? In island arcs.

23
Anomalous fractionation

• In this case study of the Aleutians, we know
  Ce/Pb ratio and Pb isotope composition of the
  North Pacific sediment and ocean crust being
  subducted.
• In Pb-isotope space, the arc lavas appear to get
  all their Pb from mixtures of these two
  components.
• But the lavas are not a simple mix of MORB and
  sediment. The low 207Pb/204Pb component has a
  lowered Ce/Pb ratioPb must be preferentially
  extracted (relative to Ce) from the subducting
  basalt (but not from the sediment).
• Implication Pb is mobile in aqueous fluid,
  leading to low- Ce/Pb arc source and high-Ce/Pb
  residual slab.

24
Origin of the four mantle components

• DMM is easyit is ambient upper mantle, depleted
  2 Ga ago by extraction of the continents.
• However, MORB can be polluted by influence of
  nearby plumes (Schillings effect), so not all
  MORB plot right at DMM

Isotopic composition of mid-Atlantic ridge
samples near the Azores hotspot Begs questions
How well mixed is DMM reservoir? Is even pure
DMM recharged with a flux from somewhere? How
does the upper mantle stay fertile over time?
25
Origin of the four mantle components

• EMII is almost certainly recycled continental
  material, presumably subducted terrigenous
  sediment.
• Isotopic composition of young pelagic sediment is
  a pretty good match for EMII isotopes, but not
  perfectsediments must be aged for a while.
• As we saw, continents (and hence also
  continent-derived sediments) have very high Pb
  concentrations. Hence U/Pb is not very high and
  EMII does not evolve to especially enriched
  206Pb/204Pb. But Th/U is high (due to scavenging
  of Th from seawater), so 208Pb/204Pb increases
  faster.
• Because Sr/Pb and Nd/Pb ratios are lower than in
  other components, mixing arrays towards EMII
  should be strongly curve in isotope ratio-ratio
  space, as observed.
• Even though sediment signature is transferred to
  arc basalts at subduction zones, some sediment or
  some sediment-derived trace elements must be
  subducted, to arise elsewhere in OIBs.

26
Origin of the four mantle components

• HIMU is usually attributed to subducted, altered
  ancient oceanic crust.
• The preferential extraction of Pb from the
  basaltic part of slab at subduction zones leaves
  a high U/Pb residual component, which will evolve
  to high 206Pb/204Pb with time.
• But it is necessary that Rb also be removed
  relative to Sr during subduction, or HIMU would
  have wrong 87Sr/86Sr.
• Note that HIMU-DMM mixing arrays are linear,
  which implies Sr/ Pb and Nd/Pb ratios are similar
  in these end membersa problem?
• Other authors think HIMU is a component of
  metasomatically altered continental lithospheric
  mantleno agreement on this.
• Some even think HIMU has high U/Pb because of
  late segregation of Pb into the core...

27
Origin of the four mantle components

• EMI is problematic. All kinds of ideas are in
  play...
• It is close to Bulk Earth, except in eNd.
  Perhaps it is slightly modified BSE (modified
  how? Nobody says).
• It also resembles lower continental crust, from
  xenoliths and granulite terranes. Perhaps EMI
  and EMII are distinguished by intracontinental
  differentiation, and EMI is recycled by
  delamination whereas EMII is recycled by erosion
  and subduction.
• H2O-rich and CO2-rich fluids mobilize trace
  elements differently. It is possible that HIMU
  and EMI could be complementary products of
  migration of CO2-rich fluids from continental
  lithospheric mantle into lower continental crust.
• Another issue to revisit after we talk noble
  gases.

28
The Upper Mantle as an Open System

• For Pb, we can prove that there is continuing
  input to the upper mantle-ocean crust system from
  some other long-lived reservoir, probably the
  lower mantle. If this input balances Pb flux to
  continents at arcs, the upper mantle might be in
  steady-state for incompatible elements.

• This argument is based on Th/U ratios the
  continental crust has a chondritic Th/U ratio
  (3.9), but the MORB source has a much lower Th/U
  ratio (2.5).
• If input to upper mantle is chondritic in Th/U,
  and output to continents is chondritic, upper
  mantle could be in steady state, even with a
  different Th/U ratio, but this requires a short
  residence time of Pb in upper mantle.

29
Th/U ratios, Th isotopes and Pb isotopes

• Trying to match up Th/U ratio and 208Pb/206Pb
  composition of MORBs is a different exercise from
  the Sm/Nd and Lu/Hf problem presented above,
  because we can correct accurately for effects of
  melting and there is still a discordance.
• For a source in secular equilibrium, the activity
  of 230Th is equal to that of 238U. Hence the
  (232Th/230Th) activity ratio is a measure of the
  Th/U ratio of the source

• Since MORB has a Th excess due to melting
  processes, the measured value is an upper limit
  for the Th/U of the source. Data
• Mid-Atlantic Ridge kTh 2.5
• East Pacific Rise kTh 2.50.2
• Hawaii and Iceland kTh 3.0
• Tristan da Cunha kTh 3.7
• Here is a trace-element ratio indicator in which
  hotspots are closer to primitive!

30
Th/U ratios, Th isotopes and Pb isotopes

• The long-term history of Th/U in a source, on the
  other hand, is determined from the Pb isotopes
  (where T is the age of the Earth)

• Data
• Mid-Atlantic Ridge kPb 3.780.07
• East Pacific Rise kPb 3.730.06
• Indian Ridges kPb 3.890.11
• Hawaii and Iceland kPb 3.830.04
• Tristan da Cunha kPb 4.17
• Some hotspots have long-term Th/U higher than
  chondritic!
• SOthe maximum present day Th/U of the MORB
  source (from Th isotopes) is much less than the
  long-term Th/U average reflected in the Pb
  isotopes of the same source, and this is not a
  recent melting effect.
• The Pb in MORB cannot have been in a low Th/U
  reservoir for more than 600 Mathis must be the
  residence time of Pb in the upper mantle.

31
Th/U ratios, Th isotopes and Pb isotopes

• Where is the reservoir from which Pb is input to
  the upper mantle?
• Upper continental crust has chondritic Th/U and
  can be recycled by erosion (hence EMII flavored
  hotspots), but it has the wrong 207Pb/204Pb
  ratio, since its U was fractionated from Pb more
  than 1 Ga ago when 235U was more abundant.
• Continental lithospheric mantle might be the
  reservoir, but this would require their entire
  mass to exchange with the upper mantle every few
  hundred Ma, which is inconsistent with the
  long-term stability of cratonic lithosphere.
• That leaves only the lower mantle, which is so
  big and Pb-rich that over geologic time only half
  the mass of the upper mantle would have to be
  replaced by lower mantle to give the necessary
  flux (10 per Ga).
• Bottom line the more incompatible the element,
  the shorter its residence time in the upper
  mantle-oceanic crust system (200 Ma for the
  perfectly incompatible element)
• Hence DMM is roughly equal to BSE in Pb isotopes
  (which are replaced much faster than 238U decay),
  but quite different in Nd and Sr isotopes, since
  these elements are more compatible (especially in
  arcs).
• For the most incompatible elements the global
  system has evolved to a steady-state where output
  to the continents is balanced by input from lower
  mantle.
• Convective isolation (layering?) is necessary to
  explain long-term evolution of components, but it
  cannot be perfectit must be leaky.