Transcript A geochemical approach to the search for plate tectonic signatures

Geochemical Indicators of Plate
Tectonic Processes in Old Rocks
Julian Pearce
(Cardiff University)
Bob’s Smoking Guns for Archean Subduction
(Blueschists, UHP rocks, Ophiolites)
However, Subduction Fluxes are Forever
Even if the arc is overwritten by collision or eroded away,
inherited subduction signals remain in the mantle lithosphere and
can be reactivated later
Finding Evidence of Subduction is the Key to Knowing
when Plate Tectonics started
Pearce &
Peate
(1995)
But present-day subduction is a complicated processes; essentially mantle
flow and subduction input are the geochemical indicators of plate tectonics,
while crustal interactions tend to mask these indicators. The key to
identifying arc lavas in the Archaean is separating subduction signals from
crustal signals.
Geochemical Indicators of Plate Tectonic
Processes
•Indicators of Plate-driven Mantle Flow
•Indicators of Subduction
mantle depletion by episodic
melt extraction towards arc front
volcanic
arc rear-arc
seamount
back-arc
ridge
intraplate
volcano
lithosphere
F
LT
asthenospher
e
HT
B’
A’
C
B
UHT
A
At subduction
zones,
these two
processes act
together.
Geochemical Tracing of Subduction Input
Progressive
subduction leads to
sequential release of:
mantle depletion by episodic
melt extraction towards arc front
volcanic
arc rear-arc
seamount
back-arc
ridge
lithosphere
F
LT
asthenospher
e
HT
LT elements (Rb, Ba
etc)
intraplate
volcano
B’
A’
C
B
UHT
A
HT elements (LT
elements plus LMREE, Th, P)
UHT elements ( HT
elements plus Nb, Ta,
Zr, Hf).
In the Archean, we would
expect this sequence to
take place at shallower
depths than at present
Geochemical Tracers for Subduction
Input
100
a: Nb/Yb (mantle fertility and % melting)
proxies:
rock/MORB
d
10
e
b: Ba /Nb (total subd uction compon ent)
c: Th/Nb (de ep subd uction compon ent)
d: Ba/Th (shallow sub duction compo nent)
e ( U/Th) and f (Nb /Ta) have complex
properties as discussed in the text
Shallow subduction
components cannot be
investigated in
Archaean rocks
because of alterationsensitivity
b
shallow s ubduction
component
c
1
a
focu s of
this paper
0.1
deep subduction
component
f
mantle component
Rb Th Nb K Ce Pr P
Z r Sm Ti Tb Y Er Yb
Ba U Ta La Pb Sr Nd Hf Eu Gd Dy Ho Tm
Deep subduction
components are more
robust. Th/Nb (an
indicator of negative
Nb anomalies also an
effective tracer of
deep subduction input
that is robust to upper
amphibolite facies
Geochemical Tracing of Mantle Flow
Pearce (2005)
mantle depletion by episodic
melt extraction towards arc front
volcanic
arc rear-arc
seamount
back-arc
ridge
intraplate
volcano
lithosphere
F
asthenospher
e
B’
A’
C
B
A
Flowing mantle undergoing decompression can drastically change its
chemical composition
Geochemical Tracing of Mantle Flow
Tonga-Lau
System:
collabration w.
Pam Kempton, Jim Gill
Results indicate that mantle entering subduction systems
progressively loses incompatible elements by melt
extraction while flowing to the sub-arc region
Geochemical Tracers for Mantle Flow
Thus Nb/Yb acts as a
Good proxy
for mantle fertility
Nb/Yb gradients
provide
a means of tracing
mantle flow
100
Nb/Yb acts a proxy for mantle fertility
rock/MORB
Isotope ratios reach
plateaus, so trace
element ratios are
more effective for
mapping. Most
effective for
subduction systems
are VICE/MICE ratios
based on immobile
elements
10
Nb/Yb
enrich
(North ed MORB
Fi j i B a
sin)
1
depleted MORB
(Southern Lau Basin)
0.1
Rb Th
Ba U
K Ce Pr P Zr Sm Ti Tb Y Er
Ta La Pb Sr Nd Hf Eu Gd Dy Ho Tm
Th/Yb-Nb/Yb Fingerprinting
Pearce and Peate, 1995
At the present day, MORB and IOB plot in a well defined array,
along the mantle flow axis (Nb/Yb); arc lavas are displaced to
higher Th/Nb ratios. The overall dispersion of arc lavas is parallel
to the MORB array indicating the importance of melt extraction
during mantle flow in magma genesis.
Th/Yb-Nb/Yb Fingerprinting:
Role of Mantle Flow
10
IB
ar
r
ay
HT
Th/Ybsubduction
M
O
R
BO
1
melt
extraction
0.1
N-MORB
Nb/Yb
0.01
0.1
1
10
100
two components to a first
approximation
Th/Yb-Nb/Yb Fingerprinting:
Interaction of Mantle Flow and SZ input
Location of data and
shapes of trends indicate
process
10
ar
r
ay
Th/Yb
BO
IB
2
1
M
O
R
1
2= Add SZ component
during melt extraction
4
0.1
3
N-MORB
Nb/Yb
0.01
0.1
1
10
1= Add SZ component
before melt extraction
100
3. Add SZ component
without or after melt
extraction
4. UHT SZ component adds
Nb as well as Th
Types of Subduction Zone
North Tonga: Oceanic plumesubduction interaction
Cascades: Continental plumesubduction interaction
IBM Eocene: Intra-oceanic
subduction initiation
Japan Miocene: Intra-continental
subduction initiation
Various Localities: Ridge
subduction
Taiwan, SE Indonesia: Syncollision
Mariana: Arc Rifting
Anatolia: Subduction component
reativation
plus
steady state subduction
Each type of Subduction Zone shows a different
topology on the Th/Yb-Nb/Yb diagram
Example: W. Pacific Eocene
10
Boninites at Subduc tion Init iation
(Data of Pea rce et al., 1992)
IB
ar
ra
y
Th/Yb
M
OR
BO
1
0.1
N- MORB
Acoje, Philippines
Guam
Chichijima
Bonin For earc
Saipan
Mariana Forearc
Nb/Yb
0.1
1
10
100
EarliestArc
Example: North Tonga
10
N. To nga
(Plume-arc interaction)
IB
ar
ra
y
Th/Yb
M
O
RB
-O
1
Samoa
W Boninites
0.1
E Boninites
N-MORB
N Tofua arc
Data of Pearce
et al. (In press)
Nb/Yb
0.01
0.1
1
10
100
Example: Kamchatka
10
Kamchatka
(flow from rear-arc to arc front)
IB
ar
ra
y
Th/Yb
M
O
RB
-O
1
Distal
0.1
N-MORB
Proximal
Data compliation
Nb/Yb
0.01
0.1
1
10
100
Misinterpretation of Subduction Zones:
Amphibolite-facies metamorphism
HT fluids/melts from
sediments impregnated and
metamorphosed the lavas.
The result is that some
normally-immobile LIL elements
(e.g. Th) are enriched
10
Th/Yb
increasing
melt addition
1
N-MORB
Broken Hill cores
0.1
0.1
1
Ta/Yb
10
Misinterpretation of Subduction Zones:
granulite facies metamorphism
David Waters
The lower crust loses a melt
fraction (which can be seen in
places ‘escaping’ from the rock)
leaving a granulite residue
This depletes the residue in
LIL elements.
10
Th/Yb
increasing
melt addition
1
N-MORB
0.1
0.1
increasing melt
depletion
1
Ta/Yb
10
Misinterpretation of Subduction Zones: Crustal
Contamination
10
10
S E Anat olia Post-Collis ion
dat a from Pearce et al. (1990)
Greenland Margin - Iceland
M
M
O
O
R
R
BO
1
BO
1
IB
IB
ar
r
ar
ra
y
Th/Yb
ay
Th/Yb
0.1
0.1
N-MORB
N-MORB
Nb/Yb
0.01
0.1
1
10
100
0.01
0.1
Nb/Y b
1
10
100
The crustal contamination vector is typically parallel to the UHP
subduction vector.
Crustal Contamination
is also part of continental arc dispersion
10
Kamchatka
(AFC trend)
Th/Yb
ar
ra
y
R
1
RB
-O
IB
D
M
O
A
B
Distal
0.1
N-MORB
Proximal
Data compliation
Nb/Yb
0.01
0.1
1
10
100
Proposed Archean Subduction-Related Rocks
Komatiites
Boninites
BADR volcanic series
Adakite lavas and TTGs
Aim is to assess these subduction signals
Testing Archean Komatiite Subduction Models
10
10
Komati Formation, Barberto n (3.5Ga)
Komatiite/tholeiite
suites
ar
r
Th/Yb
M
M
O
O
R
BO
1
RB
-O
1
IB
IB
ar
ra
y
Th/Yb
ay
(Data of Chavagnac (2004) & Parman et al. ( 1997)
Data of Jochum
et al. (19 91) for
Onverwacht,
Kambalda, Monro,
Alexo, Belingwe
0.1
0.1
N-MORB
N-MORB
Nb/Yb
0.01
0.1
1
10
100
Nb/Yb
0.01
0.1
1
10
100
Parman et al. (1997, 2001), following the work of
Grove, controversially argue that Barberton lavas
were wet rather than hot – i.e. subduction related.
However, crustal contamination is more consistent
with the data
Testing Archean Boninite Subduction Models
10
10
Isua Greensto ne Belt
Finnish Greenstone Bet (2.8Ga)
Central Domain (boninites)
Data of Polat & cowor kers
IB
M
M
O
O
R
R
BO
1
BO
Data of Shchipansky et al. (2004)
1
IB
ar
r
Iringora
Other Domains
ar
r
ay
Th/Yb
Th/Yb
0.1
0.1
N-MORB
N-MORB
Nb/Yb
0.01
0.1
ay
Khi zovaara
1
10
100
Nb/Yb
0.01
0.1
1
10
100
Dispersion is along a crustal contamination vector, not mantle flow
vector; however the Isua boninites do require source depletion.
Alternative Model for Archean Boninites and
Related Rocks
Arculus et al. (1992)
from IODP Leg 125
Phanerozoic boninites result from shallow, wet melting
(opx = ol + siliceous melt)
Alternative Model for Archean Boninites and
Related Rocks
Komatiite
Arculus et al. (1992)
from IODP Leg 125
crust
But Archean boninites could be explained by komatiite-crust interaction
Testing Archean BADR Subduction Models
10
10
Super ior: Wawa
2.7Ga
Pilbara: Whundo
‘arc’ 3.12Ga
ar
ra
M
M
O
O
RB
-O
1
RB
-O
1
IB
IB
ar
ra
y
Th/Yb
y
Th/Yb
Thol. B asalts
Basalts
CA Basalts-And .
0.1
Andesites
0.1
Bonin ites
N-MORB
High-Mg andesite s
Nb-enriched bas.
N-MORB
Data of Smithies
et al. (200 5)
Nb-enriched bas.
Data of Polat
et al.
Nb/Yb
0.01
0.1
1
10
100
Nb/Yb
0.01
0.1
1
10
100
The proposed BADR series shows increasing Th/Nb with
increasing silica content and trend parallel to
crustalcontamination trends. Wawa lavas do not however have
an end member in the mantle array: a difficult call.
Turkish Analogue
Reactivation of sub-arc
lithosphere following collision
can be evaluated using Nb
anomalies
10
AFC
AFC
Average
Upper
Crust
Th/Yb
Post-collision magmas
erupted at the site of the
‘dead’ arc have subduction
signatures
RB
-O
IB
SZ
ar
ra
y
P.
(g M
t. el
fa tin
cie g
s)
1
S of Pontides
M
O
0.1
Foreland
N of Pontides
Average N-MORB
0.05
0.1
Ta/Yb
1
5
Post-collision magmas
erupted where there was
no arc have intraplate
signatures ( with crustal
assimilation)
Testing Archean Adakite/TTG Subduction
Models
All agree on melting of mafic material but how?
Flat subduction (lower crust melting)
Hot subduction (slab melting)
Delamination (lower crust melting)
Magma chambers (mid-crust melting)
Maybe all four! But subduction not essential.
Crustal Processing
With no plate tectonics, could Archean volcanic terranes have undergone
intracrustal reprocessing like present-day Collision Zones, so explaining
subduction-like chemical signatures?
Conclusions
Many Archean boninites and other evolved high-Mg magmas could be explained by
interactions between komatiite (and related) magmas and crust rather than by
subduction.
Archean basalt-andesite-dacite-rhyolite (BADR) series require substantial magmacrust interaction and may not all additionally have subduction components.
Adakites could (as is well known) involve melting of mafic rocks in the crust as well
as in subduction zones.
Unlike modern arc lavas, Archean ‘arc’ lavas do not exhibit a geochemical indication
of plate-driven flow into and within a mantle wedge.
If there was subduction in the Archaean, it must have involved variable ‘adakitic’
addition to a homogneous mantle
Personally, I would start subduction around 2.7 Ga but work still needs to be done
separating crustal and subduction signals.