Geochemical and Physical Processes Occurring Beneath Oceanic Ridgesand Islands, with Implications for Production of Basalts in the Archean Era

Aspects to Be Addressed:

-1- Composition of basaltic melts produced from fertile and depleted mantle-a- The mineralogy of fertile and depleted mantle-b- The major elements and the effects of eutectic and peritectic points and curves -c- Compatibility of trace elements for silicate minerals and melts

-2- Trace Element behaviour during partial melting of fertile and depleted mantle-a- Modelling the partial melting of fertile mantle (e.g., Primitive Mantle)-b- Modelling the partial melting of depleted mantle (e.g., DMM)

-3- Mixing of melts during extraction to the surface (lavas)-a- Extraction beneath MORS with implications for mechanisms of melt transport-b- Extraction of melt beneath Oceanic Islands and mechanisms of melt transport

-4- Identification of mixing components in MORBs and OIBs

-a- Depleted component in MORBs and OIBs-b- Fertile component in MORBs and OIBs

-5- The effects of mantle temperatures on MORB trace element compositions-a- Temperature regimes below rapidly and slowly spreading ridges-b- The NCR, EPR and SE Indian Ridge MORBs & relationship to mantle Temp.-c- The Archean basalts and relationship to mantle temperatures

Mineralogy of a Fertile Mantle and Effects of Partial Melting

Classification of the Ultramafic Rocksand effect of partially melting fertile mantle

an abyssal (ocean floor) lherzoliteestimated PUM mineralogical composition

Ol

Opx Cpx

Peridotites

Pyroxenites

Lherzolite

olivinewebsterite

harzbergite

olivineopxenite

websterite

clinopyroxeniteorthopyroxenite

wehrelite

dunite

olivinecpxenite

40%

5%

95%

DMM

Primitive Mantle Bulk Composition (a lherzolite) Oxide Wt.% SiO2 -- 45.0 TiO2 -- 0.2 Al2O3 -- 4.45 FeO -- 8.05 MgO -- 37.80 CaO -- 3.55 Na2O -- 0.36 K2O -- 0.03

Mode (vol.%)Fertile Mantle Ol: 53% Opx: 26% Cpx: 16% Sp: 5%

Mode (vol.%) DMM Ol: 57% Opx: 28% Cpx: 13% Sp: 2%

Effect of Partial Melting

Primitive mantle from McDonough & Sun (1995)Fertile mantle mode (calc. from bulk composition)DMM mode from Workman and Hart (2005)

Eutectic & Peritectic Points, and the Uniformity of Melt Compositions(with Respect to the Major Elements)

3d, 4d and 5d Transition Elements

Generalizations about Abundances of Elements in the Crust and Mantle

Groups IIIA&B

Group VIIIGroup 1

The Major and Minor Elements: Oxides of Na, K, Mg, Ca, Fe, Al, Si are generally are present at greaterthan about 1 wt.%. Oxides of Ti, P, S, Mn and sometimes K are commonly present at levels between0.1% and 1% elements and are g generally referred to as the minor elements.

Per

iod

Trace Elements Useful for Deciphering Igneous Processesin Recent and Ancient (e.g., Archean) Volcanic Rocks

The Rare Earth Elements (REEs) or Lanthanides

High Field Strength Elements (HFSE)

The Actinides

These are of high charge and form strong bonds with oxygen. They are very sparingly soluble and generally are not readily taken into aqueous solutions during weathering, hydrothermal alteration or metamorphism.

highly charged (e.g. La3+) andgenerally are sparingly solublein aqueous solutions.

XU forms strong aqueous complexeswith bicarbonate and carbonate ionsand can be very mobile if oxidized to the 6+ oxidation state.

Olivine(Mg,Fe)2SiO4

Orthopyroxene(Mg,Fe)SiO3

Melt

Partial Melting Leads to Trace Element Partitioning

Nb and Zr are preferentially partitioned into the melt phase relative to the solid phaseswhereas Y has a weak preference for the solids especially garnet & pyroxenes.

from Wark et al. (2003)

Mineralogy of the Mantle

Major minerals: Olivine, Ortho- and Clinopyroxenes.

Al-rich, minor phases: spinel, plagioclase and garnet.

Minor phases may have dramatic effect on trace elementabundances in melts.

Trace elements substituteon lattice sites and oninterstitial sites.

Stability of Al-Bearing Minerals in the Mantle and Adiabatic Melting of the Mantle

(100% liquid)

partiallymeltedmantle

first melting

Mantle Adiabate

100% solid

15%30%

geotherm

Consider a large volume of mantle (a cube at least 100km on a side) to be convected toward the surface. If it rises sufficiently rapidly it does not remain in thermal equilibrium with the surrounding mantle. The extreme is of this situation is adiabatic rise, where no heat is transferred to the surrounding mantle. Under these circumstances partial melting of the mantle will occur and does today at spreading centres.

Adiabatic Melting the Mantle

Aluminous Minerals Stabiltiy

Garnet: stable at high pressuresSpinel: stable at intermediate pressuresPlagioclase: stable at low pressures

Simulating Trace Element Behaviour During Partial Melting of the Mantle

The Distribution (or Partition) Coefficient

To Model Trace Element Behaviour in Melts Derived from Mantle Rocks

One must employ or know the following:

1) a Kd expression for each solid in the rock 2) Kd values for each must be known for the P and T conditions3) The total amount of each element “i” must be known4) The modal proportions of the minerals must be known as a function of the degree of partial melting.

Olivine Opx Cpx Garnet

Lu 0.02 0.052 0.439 4.5

Yb 0.017 0.047 0.432 4.18

Er 0.0087 0.041 0.422 3.2

Y 0.007 0.025 0.421 2.8

Ho 0.006 0.029 0.41 2

Tb 0.002 0.021 0.382 1

Gd 0.00099 0.016 0.37 0.8

Dy 0.004 0.025 0.402 1.4

Eu 0.00015 0.013 0.355 0.5

Hf 0.005 0.013 0.23 0.115

Sm 0.0006 0.01 0.28 0.115

Zr 0.0013 0.013 0.128 0.27

Nd 0.0002 0.007 0.19 0.06

Sr 0.008 0.009 0.096 0.003

Ce 0.00006 0.003 0.09 0.007

La 0.00005 0.0005 0.042 0.001

Nb 0.000041 0.0001 0.007 0.0042

U 0.00005 0.00005 0.0052 0.027

Th 0.00005 0.00005 0.003 0.0015

Rb 0.000045 0.00045 0.0006 0.00001

Ba 0.000043 0.00004 0.00068 0.00001

Partition or Distribution Coefficients

Values from Donnelly and Hart (2004)

Incipient meltfrom PUM

Batch and Fractional Melting of Fertile and Depleted Mantle

Incipient meltfrom DMM

0%

0.1%

0.2%

0.3%

1.0%

0%

0.1%

0.2%

0.3%

1.0%

DMMDepleted MORB Mantle (DMM) is adopted as depletedupper mantle (data from Workman & Hart, 2005)

Batch melting

Fractionalmelting

Batch melting

Fractional melting

Batch Melting: melt remains in equilibrium at all stagesuntil it is released to the surface (closed system melting).

Fractional Melting: melt is generated and immediatelyremoved from the solid source. (open system melting).

Dynamic Melting: melt is generated in equilibrium withthe source rock but separates after as small amount ofmelt is generated. (partially open system)

PUM meltingwith spinel stable

DMM meltingwith spinel stable

Primitive Upper Mantle (PUM) is adopted as fertile mantle (data from McDonough & Sun, 1995).

juvenile limb

senile limb

juvenile limb

senile limb

Fertile Mantle

Depleted Mantle

40 50 60 70 80 90 100Y Zr

Nb

20

40

Incipient meltderived from PUM Polynesian Islands (HIMU)

Samoan Islands Canary Islands+

30

Gt stable

Sp stable

av. E-MORB

PUMDMM av. N-MORB

fractional melting with 3 wt.% spinel

fractional melting with 5 wt.% garnet

OIB suites describelinear trends parallel to the juvenile limb

Deductions from modelling

-1- OIBs cannot be derived from depleted mantle.

-2- OIBs cannot be derived from fertile mantle if melting occurs in the spinel field.

-3- OIBs may be derived from fertile mantle if melting occurs in the garnet stability field.

-4- OIBs plot on the juvenile limb indicating that they are the result of low degrees of partial melting of the source rock.

Production of OIBs Through Mantle Melting

40 60 80 100Y Zr

Nb

20

40

60

DMMMelting of DMM

DMM

incipient meltfrom DMM

OIBfield

Confirmation that the Nb-Ar-Y modelling yields reasonable resultsThe slopes of the trends for the three OIB suites are similar to the trendspredicted for different degrees of partial melting of a fertile source.

Hypothesis Derived from Modelling ResultsThe difference among OIB trends may reflect the amount of garnet inthe mantle source rocks.

N

N

MORBs of the East Pacific Rise – A Spreading Ridge Devoid of Plume Contributions

Clipperton Transform Fault.

Siqueiros Transform

EPR

EPR

Central America

Sea Mount

Taken from Google Maps(originally from TerraMetrics, NASA)

Sea Mount

Samples arefrom this region

Seismic Studies (MELT) Define the Melting Region Beneath the EPR

Notes:Lateral variations in structure arerequired at depths exceeding 100 km,suggesting that the primary meltproduction in the up-welling mantlebegins at about 100 km and that traceamounts of melting occur to depths of50 km or more.

SOURCE: The MELT Seismic Team (1998) Imaging the Deep Seismic StructureBeneath a Mid-Ocean Ridge: The MELT Experiment. Science, v.280, p.1215-1218.

Cross-Section Across the East Pacific Rise at about 17oS

40 60 80 100Y Zr

Nb

20

40

60

40 60 80 100Y Zr

Nb

20

40

60PUM batch and fractional

melting (spinel stability field)

DMM batch and fractionalmelting (spinel stability field)

MORBs plot mostly on the Senile Limbof the partial melting trends. Implication: MORBs are the product ofgreater partial melting of the mantle thanare OIBs.

Partial Melting of PUMPUM batch and fractional melting do not reproducethe MORB data but the simulations mimic their trend.

Melting of DMM yields a better match to the MORBdata than the PUM simulations.

Production of EPR MORBs Through Mantle Melting

Av. E-MORB

PUMDMM

Av. N-MORB

Y Zr0

0

0

Nb/Y

Nb/Y

Zr/Y

0N

b/Y

Nb

Nb-Zr-Y Ternary Transformed to Orthogonal Axes

Orthogonal plots are simpler to use thanthan ternary diagrams. The Transformationof the Nb-Zr-Y ternary to the equivalentorthogonal plot is shown.

Av. E-MORB

PUM

DMM

DMM batch melting(in spinel field)

DMM fractional melting(in spinel field)

Partial Melting of Mantle to Produce of EPR MORBs

Trends described by MORBsThe fits to the EPR MORBs indicate arcuate trends between Nb/Y and Zr/Y. The relationship is explained if the melts to reflect different degrees of partial melting of mantle (note the simulations).

The Simulations are inadequate in:1) some samples are too great in Zr/Y to result from fractional melting of DMM.2) Most Nb/Y values are intermediate between batch and fractional melting trends.

Conclusion:The arcuate trends are most simplyexplained by partial melting of mantle.

Sources of Data:Yellow Dots: Le Roux et al. (2006)

Red Dots: Niu et al., (1999)

Grey Dots: Goss et al.(2010); Sims et al. (2002,2003); Hanan et al. (2000); Hall et al. (1996); Mahoney et al. (1994).

The Compositions of MORBs Plotting on Juvenile and Senile Limbsare determined by the region of the melt zone in which they are generated

Extensive Melting(melts plot on the Senile Limb)

Incipient melting(melts plot on the Juvenile Limb)

Question: How do the juvenile (incipient) melts make it to the top of the Ridge?

progressively more melt is formedas the mantle rises within themelt zone where it produces meltswhich plot on the Senile Limb

exponential fit todata of Le Roux et al.

A Problem: Melts from Two Sources Mix Beneath the EPR

143Nd/144Nd vs. Nb/Y Plot

The Conundrum: What is the nature of each source?

Data from Waters et al.

Av. E-MORB

PUM

DMM

Av. N-MORB

Conclusion of Waters et al. (2011):The EPR MORBs are produced by two components which mix beneaththe EPR to produce N-MORBs and E-MORBs.

One source is located on the senile limb of a mantle melting curve andthe other is located on the juvenile limb.

Data of Le Roux et al. (2006)

The half-life of 147Sm is ~ 109 years. The mantletraverses the EPR melting zone in ~ 150,000 yrs.There is insufficient time to alter the isotopic ratiodramatically. Conclusion: there are two sources to these MORBs

EPR MORBsfrom (Waters et al., 2011)

Processes:-1- Partial melting: partial melting produces magmas that fall on a senile limb which indicates extebsive partial melting.-2- Two Component Mixing Mixing of a depleted, senile limb component (~ Av. N-MORB) and an enriched component from a juvenile limb which indicates small degrees of partial melting.

The Big Questions:-1- What are the origins of the two components?-2- What physical properties allow mixing (ie., what is the plumbing within the melting regime beneath the EPR?

Major Processes Affecting MORBs of the EPR

EPR MORBs Red: Le Roux et al. (2006)Yellow: Niu et al. 1999Grey: as in a slide 15

A

B

Porous and Channel Flow Beneath Mid-Oceanic Ridges

Porous and Channel FlowPorous Flow was assumed (common knlowedge) before theresults of Kelemen et al. (1995). Subsequently, there havebeen many papers on the combined effects of porous and channel flow (e.g., Liang et al., 2010; Connolly et al. 2009)

ReferencesKelemen et al. (1995) Nature, v. 375, p. 747.Connoly et al. (2009) Nature, v. 462, p. 209Liang et al. (2010) Geochim. Cosmochim. Acta, v. 74, p. 321

Predominantly Porous Flow(low melt migration rates)

Mixing Region

Melting Zone Beneath the EPR(after the MELT Seismic Team, 1998)

Flow Regime below the EPR(after Waters et al., 2011)

Flow Regimes Beneath RidgesInitially flow is porous and melt migrates slowly upward,but as melt forms in greater amount it transitions tochannel flow and migration becomes very rapid.

Melt is Isolated from Source once in ChannelsOnce in channels, the melt has little contact with thebulk source rock because the channels are lined withOlivine (other phases have been removed by melting).

incipient melting

Channelextensivemelting

Mantle Flow Driven by Spreading at the EPR(Classic View: Mantle flow dominated by the spreading of a ridge)

Adiabatically Rising Plug

Conclusions and Implications-1- Flow is driven by ridge spreading with mantle flowing passively into the region beneath the ridge.-2- The scenario implies that passive flow dominates the regional mantle flow regime.-3- Model assumes that enriched magmas (E-MORBs) are derived from mantle heterogeneities.-4- Model assumes porous flow everywhere (melt formed below 100km mixes with more evolved melts.-5- Olivine orientation derived from MELT project does not conform to the passive flow model.

Global Mantle Flow Model Proposed by Conrad & Behn (2010)

The model considers:-1- plate motions,-2- Earth’s rotation-3- aesthenosphere viscosity-4- aesthenosphere anisotropy-5- density variations in mantle

Conrad and Behn (2010) in Geochim.Geophys. Geosys. v.11doi: 10.1029/2009GC002970

The A-B section crosses the EPR atabout 70o (see below) which and isalso close to the Melt seismic array.

Km

Note Reverse Flow

EPREPR

Flow of the Solid Mantle and Melt Beneath the EPRincorporating the Mantle Flow Model of Conrad & Behn (2010)

ReferencesKelemen et al. (1995) Nature, v. 375, p. 747.Connoly et al. (2009) Nature, v. 462, p. 209Liang et al. (2010) Geochim. Cosmochim. Acta, v. 74, p. 321

Mantle flow lines dominated by a regional mantle flow regime The melt flow regime differs fromthe solid mantle flow regime

(after Waters et al., 2011)

Incipient melting Zone: porous flowchanges to Channel Flow in theincipient melting zone

Flow lines of the Solid Mantle(melt flow lines in red and yellow)

Predominantly Porous Flow(low melt migration rates)

Mixing Region

DMMsource

Deep Mantle Source

Sampling Across the Central Indian Ridge (CIR)(Cordier et al., 2010)

N

Rodrigues Ridge

Reunion

Mauritius Rodrigues Is.

Central Indian Ridge (CIR) samplingtraverse

Egeria F.Z.

Africa

Madagascar

ReunionMauritius

N

CIR

N

Murton et al. (2005)

N

Cordier et al. (2010)

Nauret et al. (2006)

Trace Element Geochemical Trends Across the CIR

Distance from Ridge Crest NESW

Zr/

Y

n = 65

Ratios of Elements

PUM

DMM

Data from Cordier et al. (2010)

Explanation for the near-Random Generation of E-MORBs(Cordier et al., 2010)

Release of E-MORBs results from theirperiodic extraction from enriched partsof the mantle (a plume pudding view).

Release of E-MORBs rather than N-MORBs results from periodicentry of enriched Mantle into melt column (a layered view).

= E-MORBs

NIcelandorigin

Charlie Gibbs Fracture Zone

N. R

eykj

anes

Rid

ge

‘all satellite images’ from Google Maps

S. R

eykj

anes

Rid

ge

MORBs and Restrictions on the Sources of their Magmas

Northern MAR

SWZ

Nb–Zr–Y Systematics of the Reykjanes and Northern Mid-Atlantic Ridge(data from Hanan et al., 2000 and Murton et al., 2002)

The ‘origin’ is the southern-most tipof the Reykjanes Peninsula, Iceland

Av. N-MORB

PUM

DMM

Av. E-MORBSRRNRR

N-M

AR

SWZ

Sources of the Mixing Components

There are at least two sources, and enriched sourceand a depleted source.

The low Nb/Y source is derived from the Senile Limbof a Partial Melting curve (depleted magma)

Mid-Atlantic Ridge (MAR) MORBs Divorced from Large Hot Spots

MORBs from the MAR between 10 and 24 deg. N lat.

Kane F.Z.

17 deg N

Vema F.Z.

Legend

from Kane F.Z. to 17 deg N

from 17 deg N to Vema F.Z.

Av. E-MORB

PUM

DMM

There are at least two sources, and enriched sourceand a depleted source.

The low Nb/Y source is derived from the Senile Limbof a Partial Melting curve (depleted magma)

Sources of the Mixing Components

(after Cushman et al., 2004)

MORBs of the Galapagos Spreading Center (ridge)

There are at least two sources, and enriched sourceand a depleted source.

The low Nb/Y source is derived from the Senile Limbof a Partial Melting curve (depleted magma)

Sources of the Mixing Components

Nature of the Depleted Component of Archean Basalts (Superior Province)

Red LakePickle LakeVermillion LakeSchreiberHemloPickle CrowSteep RockLimby Lake

Greenstone Belts Plotted

Comparison with MORBs-1- Archean basalts plot on a trend similar to the trend described by MORBs plotting on the Senile Limb of Partial Melting Curves.

-2- MORBs plotting on the Senile Limb are considered to have been derived by extensive partial melting of DMM (which resides in the Upper Mantle).

Questions-1- Do these Archean rocks represent the basalts which have been derived from extensive melting of the Archean Upper Mantle.-2- If so the Archean Upper Mantle is much more enriched in Nb/Y than MORBs. -3- Perhaps the Archean Upper mantle had suffered much less partial melting than the modern upper mantle. -4- Alternatively, the Archean upper mantle may have been hotter than the modern upper mantle and this has produced these differences.

Thermal State of the Upper Mantle and MORB Compositions

Does the Thermal State of the Upper Mantle Affect MORB Melt Composition?Most emphatically Yes. The buoyancy difference of melt and solid mantle

affects the relative rates of ascent through the melting zone beneath a ridgeand must affect melt compositions through during transport and mixing.

Both derived from DMM

What is the Nature of the Enriched Mixing Component

Are the enriched sources the same as we observe today for Oceanic Island Basalts?

The Pickle Lake AssemblageNorthern Superior Province