Transcript Implications for global geodynamics

Courtesy of NASA/JPL-Caltech
Redefining the composition of the Earth:
Implications
for
global
geodynamics
In the beginning….
Matt Jackson
Department of Earth Sciences
Boston University
Where is “home”?
•
•
•
•
•
Following accretion, a deep terrestrial magma ocean…
Siderophile elements (Fe-Ni) to the core, leaving behind the early (primitive)
silicate mantle.
From the primitive silicate earth, the crust (continental and oceanic) was
extracted from the early primitive mantle.
Crust subducted back into the mantle & mixed/stirred.
Did portions of the earliest primitive mantle survive to the present day?
Courtesy of NASA/JPL-Caltech
©2001 Brooks/Cole Publishing/ITP
Brandenburg et al., EPSL 2008
Implications from Neodymium-142
Baffin lavas
-
Discovery: Boyet and Carlson (2005) found
that 142Nd/144Nd ratios in accessible modern
terrestrial lavas are 18±5 ppm higher than O
and C chondrites.
-
Background: Nd-isotopes, two “clocks”:
147Sm decays to 143Nd (t =106 Ga)
1/2
146Sm decays to 142Nd (t =103 Ma)
1/2
(de Leeuw et al., 2010)
- Implications:
1.
(Boyet and Carlson, Science, 2005)
142Nd
variation due to incomplete mixing of
s-, r-process nucleosynthetic products.
142Nd variation has nothing to due with 146Sm
decay.
OR….
2. 142Nd variation due to 146Sm decay. All
modern terrestrial samples evolved from a
mantle reservoir with a Sm/Nd ratio ~6% higher
than chondrites.
• New Earth model
(Earth is not chondritic)
6% diff.
18 ppm
1. Earth has higher
142Nd/144Nd than
chondrites.
2.Therefore, accessible
Earth has Sm/Nd 6%
higher than chondrites.
3.Therefore, Earth’s
143Nd/144Nd is higher
than chondritic.
6% diff.
+7 ε-units
“primordial” chondrite reservoir
(Ra)
Predicted parental mantle reservoir
overlaps with high 3He/4He reservoir
Reservoir
parental to
modern
terrestrial
mantle
Jackson et al.
(EPSL, 2007)
Jackson et al. (Nature, 2010)
Baffin and W. Greenland picrites
plot near the Geochron
Primitive
Material in
“LLSVPs”?
Torsvik et al. (Nature, 2010)
Jackson and Carlson (Nature, 2011)
Trace element composition
Non-chondritic Primitive
Mantle (C&B, 2008)
•
•
•
•
13TW of radiogenic heat (instead of 20 TW)
Urey ratio is a lot lower (0.15-0.2 convective Urey ratio).
To generate the continents, >70% of the mantle must be depleted.
DMM must be much larger in a non-chondritic world.
Jackson and Jellinek, in prep
Non-chondritic Primitive
Mantle (this study)
Non-chondritic
BSE (PREMA)
‘C’
FOZO
Early Earth & long-term evolution:
Themes perfectly suited for a CIDER
interdisciplinary approach.
• Geochemical evolution of a non-chondritic Earth: Continents, DMM, the
mantle “zoo”….reservoirs once considered depleted are now enriched!
This is more than just 20 ppm in 142Nd/144Nd!
• Coupled chemical and thermal evolution of new compositional Earth
models: Dynamicists, geochemists, seismologists.
• How to generate a non-chondritic Earth? Early differentiation, impact
erosion, hidden enriched reservoirs, etc.?
• Seismologists….How will this affect PREM? Extract a bit of melt, look at
how seismic structure changes?
• Composition of high 3He/4He (primitive?) mantle useful for modeling
seismic properties of LLSVPs? Superplume theme?
CIDER approach:
-A week or two of lectures, followed by “break-out” groups that develop ideas
organically. Maybe let this work a bit more organically than in past
years? Smaller groups? AGU workshops to follow-up on collaborations?
Gannoun et al., PNAS 2011
How did the Earth get higher Sm/Nd than chondrites?
143Nd/144Nd=
(ε143Nd=+7)
0.5130
Depleted accessible
mantle: 142Nd/144Nd
requires this is source of all
modern mantle reservoirs
“Hidden” early
enriched
reservoir
Early
differentiation
(melting) event
Age (Ga)
Problem: Terrestrial lavas with high 3He/4He don’t
plot on the Geochron!
Hawaii
Iceland
After Jackson et al. (EPSL, 2007)
Samoa
Galapagos
Jackson and Carlson (Nature, 2011)
Highest 3He/4He Baffin Island
lavas bracket the OJP
Relationship between flood basalts and a
primitive (non-chondritic) mantle
• Relics of the early Earth may not be so rare?
• Why would this reservoir be sampled by large
igneous provinces?
A.Primitive Mantle produces more heat, melts more.
B.Primitive Mantle is more fusible, melts more.
A recipe for producing extraordinary volumes of melt?
Old Reservoir, Old Idea
(new possibilities)
“The nominal value of εCHUR≈0 for the continental flood
basalts indicates they are derived from a reservoir which
has maintained an unfractionated, chondritic Sm/Nd
throughout the history of the earth.”
-DePaolo & Wasserburg, GRL
PREMA? (Prevalent Mantle)
PREMA
chondrite
• PREMA defined by the
most frequently
occuring 143Nd/144Nd in
global OIB dataset.
Zindler and Hart (1986)
• PREMA is isotopically
similar to the highest
3He/4He lavas.
• Is PREMA a surviving
portion of a nonchondritic Primitive
Mantle?
• Does PREMA source
flood basalts?
Range for nonchondritic primitive
mantle predicted by
142Nd/144Nd
Zindler and Hart, 1986
Zindler and Hart (1986):
CFB’s converge on PREMA
PREMA
Primitive Material in “LLSVPs”?
“Prospecting” for primitive mantle:
If any of the early-Earth survived,
what would it look like today?
1. Noble gas isotopes and abundances (high 3He/4He)
2. A primitive, mantle reservoir should have predictable
abundances (chondritic?) of the refractory, lithophile
elements (e.g., Sm and Nd).
3. Pb-isotopes will be on the Geochron, the locus of data in
Pb-isotope space that have had the same U/Pb for ~4.5 Ga.
•
Any mantle-derived melts satisfying these three
requirements? No!
House of cards?
Courtesy of NASA/JPL-Caltech
In the beginning….
4.568 Ga (Bouvier & Wadhwa, 2010)
Solar Nebula Theory:
1.
2.
3.
4.
5.
Cloud of gas and dust
Rotating disk
Gravitational collapse
Solar nebula with young sun
Planets accrete from rotating cloud
Survival of early-Earth
reservoirs in a chaotic mantle?
Davies, 2002
Brandenburg et al. (EPSL 2008)
Starting composition of the Earth—Chondritic?
1.) Carbonaceous (C) chondrites ≈ Sun
2.) C-chondrites and Earth came from
the same (homogeneous?) solar
nebula, and the sun represents over
99.9% of solar system’s mass.
3.) Therefore, C-chondrites ≈ Earth
(non-volatile, lithophile elements like
Sm and Nd)
4.) If the Earth is a C-chondrite, then
Earth and chondrites have the same
143Nd/144Nd. (147Sm  143Nd + 4He)
Comparison of solar-system abundances (relative to
silicon) determined by solar spectroscopy and by analysis
of carbonaceous chondrites (after Ringwood, 1979)
Primordial helium in Earth’s mantle?
•Helium in the Earth’s mantle:
-Two isotopes: 3He (lower abundance) and 4He (greater abundance)
-U and Th decay to Pb via alpha decay (4He nuclei production)
-Little 3He produced in the earth (mostly primordial)
-Therefore, 3He/4He in the earth decreases with time.
-Absolute 3He/4He ratios in the solar system are small (10-3 to 10-8),
we normalize to 3He/4He ratio in atmosphere (Ra, 1.38x10-6).
•The sun (solar wind) and the atmosphere of Jupiter have high 3He/4He.
High 3He/4He is thought to be primordial.
Lavas with primordial 3He/4He don’t
have primitive chondritic 143Nd/144Nd
Depleted
“primitive” chondrite reservoir
(Ra)
Enriched
Jackson et al.
(EPSL, 2007)
Problem: Terrestrial lavas with high 3He/4He don’t
plot on the Geochron!
Iceland
Hawaii
After Jackson et al. (EPSL, 2007)
Samoa
Galapagos
Jackson et al. (Nature, 2010)
Baffin and West Greenland picrites
plot near the Geochron
How does a portion of the
mantle survive for ~4.5 Ga?
• Solid-state convective stirring
is thought to process large
portions of the mantle on
geologic time-scales.
• Recent dynamic models
suggest that pristine portions
(up to 10-15%) of the mantle
might have escaped
differentiation and mixing
over the age of the Earth (in
convective “eddies”?).
Brandenburg et al. (EPSL, 2008)
A growing clamor….
Bottom line: Terrestrial oxygen isotopes
not like C and O-chondrites (Clayton)!
Implications:
1. DMM is >45-90% of the mantle (to
>1600 km depth). If primitive mantle
143Nd/144Nd is 0.5130 (instead of
0.51263) then much more than 25%
of the mantle needs to be depleted to
make DMM!
2. What was once considered depleted
may actually be enriched!
3. How to preserve for 4.5 Ga?
4. A whole new family of models are
needed!
Homogeneous?
Courtesy of NASA/JPL-Caltech
In the beginning….
In detail, chondrites aren’t “chondritic”
Amelin & Rotenburg (2004)
chondrules
Is the Earth chondritic?
Chondrules in primitive chondrites have highly variable 143Nd/144Nd.
Nature, 2004.
Are other
protoplanetary
disks
homogeneous?
(0-2 AU)
(2-20 AU)
Part 2: Trace Element composition
of a non-chondritic Earth
• How to construct the spidergram:
A. “Backbone” of the spidergram from
parent-daughter ratios of radiogenic
isotope systems. Start with 142Nd
result…Sm/Nd ratio ~6% higher than
chondrites).
B. “Pin” the spidergram to the abundance of
K in the Earth, which is given by the
amount of Ar in the Earth (no missing Ar)
• Generate continental crust and DMM
87Sr/86Sr
143Nd/144Nd.
from
Then calculate Rb/Sr.
Non-chondritic Earth
Non-chondritic Earth
176Hf/177Hf
143Nd/144Nd.
from
Then calculate Lu/Hf.
Non-chondritic Earth
Non-chondritic Earth
Summary of isotope plots
“Backbone” of the spidergram
• 3 segments based on parent-daughter ratios.
• Link the 3 segments with canonical ratios.
Rare Earth Elements
Splice the REE’s into the
“backbone”
What about K-U-Th-Pb?
•
•
•
•
K/La=constant in OIBs (330; Palme&O’Neill)
K/U=constant in OIBs (11,900; Arrevalo et al.)
U/Pb=0.1357 (Earth’s 238U/204Pb=8.5 to lie on geochron)
Th/U=3.9 (Earth’s 232Th/238U to pass through Baffin Is. lavas).
Filling in the “stragglers”
•
•
•
•
•
Ba/Rb= Canonical in oceanic basalts
Zr/Hf= Canonical in oceanic basalts
Y/Er= Y behaves like a REE, so “force” it to plot b/t REE’s
Ti/Gd=Ti has similar compatibility as Gd, “forced” to be smooth
Nb/U=Least constrained. “Forced” to be smooth b/t U and La.
Spidergram has a shape, but no
absolute concentrations!
Standard (chondrite) model for K
in the Earth: “Hidden” Ar reservoir
• K volatile, lost during accretion, so abundance is unknown.
• U is refractory, so it’s abundance can be estimated from
chondrites. ~20 ppb.
• K/U in Earth known from basalts (~1.2x104).
• Calc K abundance in chondritic primitive mantle is 250 ppm.
• 40K decays to 40Ar (t1/2=1.3 Ga).
• All 40Ar in the Earth radiogenic, accumulated since accretion.
• There should be ~140x1018 g 40Ar Earth.
• Only ~95x1018 g 40Ar Earth is accounted for (atm, CC, DM).
• Ergo, a “hidden” reservoir in the Earth hosts the “missing” 40Ar!
PM
Lyubetskaya & Korenaga (2007)
40Ar
Use
in Earth to “pin”
primitive mantle K concentration
• Turn the “problem” of “missing” Ar into a solution:
What we see is what we get: no missing Ar reservoir!
• Three 40Ar reservoirs:
1. Atmosphere: 66x1018g 40Ar (Turekian, 1959)
2. Continental Crust: 1-9x1018g 40Ar (say, 4.5x1018g 40Ar)
3. Depleted Mantle (2 independent estimates):
A. Assume K in DM: 23-45x1018g 40Ar
(say, 24x1018g 40Ar)
B. 40Ar flux at ridges: 0.25-25x1018g 40Ar
• Total 40Ar in Earth: 95x1018g
• Therefore, 160 ppm K in the Earth.
Spidergram with K “pinned” at 160 ppm
• U, Th and K all about 0.6 of chondrite model
Can we extract continental crust from
non-chondritic primitive mantle?
• Primitive Mantle = Continental Crust + Depleted Mantle
• New Primitive Mantle: Have to depleted >55% of the mantle
to make Continental Crust!
Prim Mantle (this study)
Previous
estimates for DM
DMM (this study)=PM-CC
Layered Mantle? 660km of DMM?
Not if the Earth is non-chondritic!
DMM (25% mantle)
DMM >55% mantle)
Hofmann,
1997
Thermal budgets of a
non-chondritic Earth
• U, Th and K are all 66% lower than the standard
model.
• So, 33% less radiogenic heat than the standard
model. Lower Urey ratio (radiogenic/total heat)
• The Earth produces 46 TW.
-- Standard model: ~20 TW is radiogenic heat
-- New model: ~13 TW is radiogenic heat.
• The rest is primordial heat. Is there enough
uncertainty in primordial heat that we can make a
non-chondritic Earth “work”?
Conclusions
• The specter of a non-chondritic bulk earth?
0.5130 is Primitive Mantle?!
• What was “depleted” may actually be primitive, or
even enriched!
• Geochemical evolution of the mantle…a whole
new family of models: Isotopic & thermal
evolution.
• If the Earth isn’t chondritic, then we don’t know
some of its fundamental properties (e.g., Urey
ratio, etc.)
• Without the chondrite model, the “road ahead”
has no map. These are exciting times!
What about initial 3He/4He on OJP?
(and the other old flood basalts?)
“….olivine phenocrysts in basalt are embedded in
basaltic groundmass that has much higher [U] and
[Th] than the olivine. Consequently, 4He from
alpha-decay of groundmass U is implanted into
the rims of the olivine grains.”
“ Who drank the ‘Cool-Aid’? ”
-S. Shirey, 2010
•
•
•
•
•
•
•
Bill White
Dave Graham
Hubert Palme
Hugh O’Neill
Bernard Bourdon
Guillaume Caro
Nobu Shimizu
Jackson and Carlson (Nature, TODAY)
Highest 3He/4He Baffin Island
lavas bracket the OJP
Relationship between flood basalts and a
primitive (non-chondritic) mantle
• Relics of the early Earth may not be so rare?
• Why would this reservoir be sampled by large
igneous provinces?
A.Primitive Mantle produces more heat, melts more.
B.Primitive Mantle is more fusible, melts more.
A recipe for producing extraordinary volumes of melt?
Old Reservoir, Old Idea
(new possibilities)
“The nominal value of εCHUR≈0 for the continental flood
basalts indicates they are derived from a reservoir which
has maintained an unfractionated, chondritic Sm/Nd
throughout the history of the earth.”
-DePaolo & Wasserburg, GRL
Average
MORB
chondrite
Jackson and Carlson (Nature, TODAY)
PREMA (Prevalent Mantle) = BSE?
If the large proportion of OIB lavas with present-day 143Nd/144Nd near 0.5130 reflects a
high proportion of non-chondritic primitive material in the mantle, then primitive material
must comprise a substantial portion of the modern terrestrial mantle.
•
•
•
•
•
Moon-forming event, and the survival of
a “hidden” early enriched reservoir?
146Sm-142Nd
systematics require that
an early differentiation event
happened within 30 million years of
accretion.
182W-182Hf systematic require that
Moon formation occurred 50 Ma or
more after accretion.
Thus, Moon formation (after 4.51
Ga)must have followed any early
differentiation event (before 4.53 Ga).
How would a “hidden” reservoir
remain hidden during a giant
impact event?
Also, how to keep a hidden
ENRICHED reservoir hidden?
Lots of U,Th,K, so hot….plumes!
Half an Hour After the Giant Impact, based
on computer modeling by A. Cameron, W.
Benz, J. Melosh, and others. Copyright
William K. Hartmann
The alternative? A non-chondritic Earth! Is neither option palatable?
Why do high 3He/4He lavas from other
localities plot off of the Geochron
(and have somewhat lower 3He/4He)?
• Recycled crust is rich in Pb,
U and Th.
• If recycled crust mixes with
ambient mantle, or
surviving pieces of primitive
mantle, the mixture will be
shifted away from the
geochron.
• U and Th in recycled crust
will generate 4He and will
reduce the 3He/4He of the
mixture.
Jackson et al. (Nature, 2010)
Jackson and Carlson (Nature, TODAY)
Starting composition of the Earth—Chondritic?
1.) Carbonaceous (C) chondrites ≈ Sun
2.) C-chondrites and Earth came from
the same (homogeneous?) solar
nebula, and the sun represents over
99.9% of solar system’s mass.
The next step requires a huge assumption:
3.) Therefore, C-chondrites ≈ Earth
(non-volatile, lithophile elements like
Sm and Nd)
4.) If the Earth is a C-chondrite, then
Earth and chondrites have the same
143Nd/144Nd. (147Sm  143Nd + 4He)
Comparison of solar-system abundances (relative to
silicon) determined by solar spectroscopy and by analysis
of carbonaceous chondrites (after Ringwood, 1979)
Starkey et al., 2009
Baffin Island
and West
Greenland picrites
• Samples are from
Padloping Island, east
coast of Baffin Island.
• Lavas erupted ~62Ma
as part of the protoIceland plume.
Jackson et al. (Nature, 2010)
Previous slide
Caveat: Crustal contamination
Average
Mantle
Jackson et al. (Nature, 2010)
Trace elements indicate no role for
continental contamination in our sample suite
Kent et al. (2004) obtained a trace element dataset on Baffin Island glasses
(pillow rims). The glasses are extremely fresh, give pristine Pb and U.
Jackson et al. (Nature, 2010)
Another Caveat: Radiogenic 4He
Jackson et al. (Nature, 2010)
Magmatic He is hosted in the olivine,
U and Th in the basalt matrix
1. Helium is massively degassed before and during eruption.
2.Following degassing of He, parent-daughter ratios (U/He &
Th/He) are increased by many orders of magnitude.
3.4He generated by U and Th decay diminishes 3He/4He ratio.
4.Lesson: Avoid measuring 3He/4He on old lavas (62 Ma)!
Olivine
Torsvik et al. (Nature, 2010)
Primitive Material in “superplumes”?
PREMA? (Prevalent Mantle)
PREMA
• PREMA is isotopically
similar to the highest
3He/4He lavas from
Baffin Island.
chondrite
• PREMA defined by the
most frequently
occuring 143Nd/144Nd in
global OIB dataset.
Zindler and Hart (1986)
Range for nonchondritic primitive
mantle predicted by
142Nd/144Nd
• Is PREMA a surviving
portion of a nonchondritic Primitive
Mantle?
Zindler and Hart, 1986
• Depleted MORB mantle and
continental crust are
geochemically complementary.
• If primitive mantle 143Nd/144Nd is
0.5130, instead of 0.51264, then
much more than 25% of the
mantle needs to be depleted to
make DMM!
• DMM is >45-90% of the mantle!
• DMM extends down to 1600 km?
Radial variation of the
RMS amplitude of
relative variation of bulk
sound and shear
wavespeed
van der Hilst et al., Science 1999
Extracting continental crust from a
non-chondritic primitive mantle
Nucleosynthetic anomalies?
Ranen & Jacobsen (2006): Measured anomalies in the abundance of 137Ba and 138Ba in a variety of
chondrites, and concluded that the difference in 142Nd/144Nd between chondrites and terrestrial
rocks reflects nucleosynthetic heterogeneity in the solar nebula. They argued that imperfect
mixing of the nucleosynthetic contributions from various stars thus could result in variations in
142Nd/144Nd that are not related to 146Sm decay.
1. These anomalies not confirmed in either previous (Hidaka et al. 2003) or more recent studies
(Andreasen & Sharma 2007; Carlson et al. 2007; Wombacher & Becker 2007).
2. Although excesses in 135Ba and 137Ba, which are related to variations in the ratio of r- to s-process
components, have been observed in carbonaceous chondrites, they have not been observed in ordinary
chondrites or eucrites (Hidaka et al. 2003; Andreasen & Sharma 2007; Carlson et al. 2007).
3. When Ba isotopic anomalies are measured in carbonaceous chondrites, they show little or no
correlation with the magnitude of 142Nd deficit measured in the same sample (Carlson et al. 2007). Ba
isotopic anomalies in carbonaceous chondrites appear to have little or no significance for the
interpretation of the 142Nd/144Nd difference between chondrites and terrestrial rocks.
Of greater concern is the discovery that carbonaceous chondrites contain approximately 100 ppm
deficits in 144Sm (Andreasen & Sharma 2006; Carlson et al. 2007), which, like 146Sm, is produced
by the p-process. This result indicates nucleosynthetic variability in C-chondrites.
1. It is possible to correct for this p-process deficit in C-chondrites. A 100 ppm deficit in 144Sm/152Sm
would translate into an 11 ppm deficit in 142Nd/144Nd due to the reduced abundance of 146Sm
(Andreasen & Sharma, 2006). Therefore, the correction brings the average C-chondrite 142Nd/144Nd
value to ~21 ppm below terrestrial, a value that is similar to that obtained for other meteorite groups.
2. P-process heterogeneity does not appear to be significant for O- and E-chondrites, basaltic eucrites or
lunar samples, as all these materials have the same 144Sm/152Sm as measured for terrestrial rocks
Conclusion: The observed difference between chondritic and terrestrial 142Nd/144Nd does not reflect
nucleogenic heterogeneity in the solar nebula, but instead is best explained by the decay of 146Sm.