Workshop on the Archean Mantle

8/8/2019 Workshop on the Archean Mantle

1/110

WORKSHOP ONTHE ARCHEAN MANTLE

d/,,5_ #o _4,

not tohorizontal scale T-IQO0-2000C

(NASA-CR-18_ISb) WnRKSHOP ON THE APCHEANMANTLF (Lunar and Planotary Inst.) 105 pCcCL 08G

N90-t47_o

Unc13sn3I_6 0253134

LPI Technical Report Number 89-05LUNAR AND PLANETARY INSTITUTE 33c3 NASA ROAD HOUSTON, TEXAS 77058-4399


2/110


3/110

WORKSHOP ONTHE ARCHEAN MANTLE

Edited byL. D. Ashwal

Organizing CommitteeL. D. Ashwal, K. Burke, I.D. MacGregor, A. J. Naldrett,

W. C. Phinney, F. Richter, and S. B. Shirey

Sponsored byLunar and Planetary InstituteNASA Johnson Space Center

January 11-13, 1989Houston, Texas

Lunar and Planetary Institute 3303 NASA Road 1

LPI Technical Report Number

Houston, Texas 77058-4399

89-05


4/110

Compiled in 1989 by theLUNAR AND PLANETARY INSTITUTE

The Institute is operated by Universities Space Research Association underContract NASW-4066 with the National Aeronautics and Space Administration.

Material in this document may be copied without restraint for library, abstractservice, educational, or personal research purposes; however, republication of anyportion requires the written permission of the authors as well as appropriateacknowledgment of this publication.

This report may be cited as:Ashwal L. D., ed. (1989) Workshop on The Archean Mant/e. LPI Tech. Rpt. 89-05.Lunar and Planetary Institute, Houston. 104 pp.

Papers in this report may be cited as:Author A. A. (1989) Title of paper. In Workshop on The Archean Mant/e(L. D. Ashwal, ed.), pp. xx-yy. LPI Tech Rpt. 89-05. Lunar and Planetary Insti tute,Houston.

This report is distributed by:

ORDER DEPARTMENTLunar and Planetary Institute

3303 NASA Road 1Houston, TX 77058-4399

Mail order requestors will be invoiced for the cost of shipping and handling.

Cover:Cartoon,modifiedslightlyfrom E.Nisbet (The YoungFamh, 1987,UnwinHyman, Inc.,Winchester,Massachuseets),showinga schematicmodelo[ the Archean uppermostmantle and crust. In this modeldeeply hydrated oceaniccrust is k_rmatiiticand thicker than today, possiblyby as much as 25 kin. In the hotter Archean system,firrcesdriving and resistingplate motion may all ha,a"been much smaller than today.Subduction zoneswould hate operated in a hotter environment:The main higherlevel productwould have been tonalitic. Any addition ofwdatiles to the system u_:ruldhave addedm_rremelt, not wetter meh. Continents wereprobably ccrmparablen thickness to moderncontinentalcrust c_"thicker--the depth o[ the oceansis also, by implication, comparable to or greater than today's. In tensional regimes,such as inback.arc settings, continentswould haverifted apart with initial eneptionof komatiitic liquid,folhnvedby marie liquid after the establishmentof high.levelfractionationchambers.In places,older lithospherewas up to/50 km (ormore)thick beneathcontinents,and containeddiamonds.Uppermost mantle is harzburgiteto dunite, with bu{ryantdunite floating cner a komatiiticmagma ocean located at depths below 250 kin.Note that ;aerticalcaleis simplyimaginationand horizontalscaledoesnot existat all in this cartoon.


5/110

ContentsIntroductionProgramSummary of Technical SessionsAbstractsKolar Amphibolites--Archean Analogues of Modern Basaltic Volcanism?

S. Balakrishnan, G. N. Hanson, and V. RajamaniGeochemistry and Nd-Sr Isotope Systematics of the Kamiskotia Area, Western AbitibiSubprovince, Canada: Implications for Mantle Processes During the Formation of theSouthern Superior Craton

C. T. Barrie and S. B. Shirey

Mendon Formation Komatiites: Extreme Al2O3/TiO 2 Variation in the UppermostOnvervacht Group of the Barberton Greenstone BeltG. R. 8yerly

Geochemistry and Geochronology of Late Archean Mafic-Ultramaflc SupracrustalAmphibolites from the Northeastern Sino-Korean Craton, China

W. G. Ernst and B.-M. JahnThe Geology and Tectonics of Venus and Some Possible Implications for theArchean Earth

J. W. HeadEvidence for Subduction and Spreading in the Archean Rock Record:Implications for Archean Tectonic Style and the Evolution of theSubcontinental Lithosphere

H. H. Helmstaedt and D. J. SchulzePrecambrian Mantle Beneath Montana: Geochemical Evidence from Eocene Volcanicsand Their Xenoliths

A. J. Irving,H. E. O'Brien, and I.S. McCallumGeochemistry and Isotopic Characteristics of Archean Basalts and Komatiites andTheir Inference on Early Crust-Mantle Differentiation

B..M. Jahn and G. Gn_u

Boundary Conditions for the Archean MantleJ. H. JonesMagma Evolution in the Stillwater Complex, Montana: REE, St, Nd, and Os IsotopicEvidence for Archean Lithospheric Interaction

D. D. Lambert, R.J. Walker, S. B. Shirey, R. W. Carlson, and J. W. MorganArchean Sedimentary Rocks and the Archean Mantle

S. M McLennan

13725

27

3O

33

36

39

42

45

47

50

53

57


6/110

The Diamond-Komatiite Paradox: Hot Mantle-thick LithosphereP. Morgan

Archean Crustal Recycling: Implications for Isotopic Heterogeneity in the MantleP. A. MueUer and J. L Wooden

Evidence for a Differentiated Mantle and Plate Tectonics During the Late ArcheanDeduced from Eclogite Xenoliths in the Bellsbank Kimberlite

C. R. Neal, L A. Taylor, A. N. HaUiday, P. Holden, J. P. David.son, R. I'4. Clayton, andT. K. Mayeda

The Kolar Schist Belt, South India: Implications to the Nature of the LateArchean Mantle

V. Rajamani, S. Balakrishnan, E. J. Krogstad, and G. N. HansonModels for the Thermal Evolution of the Earth

F. M. Richter

Mantle Xenoliths and Archaean Basalts from South Africa: Implications for LocalHeterogeneity in the Archaean MantleN. W. Rogers and J. S. Marsh

The Eclogite Component of the Subcontinental Lithosphere: Observations Bearing onIts Origin and Abundance

D. J. SchulzeThe Pb and Nd Isotopic Evolution of the Archean Mantle

S. B. Shirey and R. W. Carlson

Abundances of As, Sb, W, and Mo in Early Archean and Phanerozoic Mantle-derivedand Continental Crustal RocksK. W. Sims, H. E. Newsom, and E. S. Gladney

Mantle Heterogeneity as Evidenced by Archean Mafic and Ultramafic Volcanic Rocksfrom the Central Laramie Range, Wyoming

S. M SmaglikIsotopic Evolution of the Archaean Depleted Mantle

A D, Smith and J. N. LuddenMantle-Crust Relationships in the Eastern Superior Province Inferred from Hf and PbIsotope Studies

P. E. Smith and R. M. FarquharPetrogenesis of High Mg#, LILE-enriched Archean Monzodiorites and Trachyandesites{Sanukitoids) and Granodioritic Derivatives in Southwest Superior Province

R. A. Stem, G. N. Hanson, and S. B. ShireyRe-Os Isotopic Constraints on the Chemical Evolution of the Archean Mantle

.R. J. Walker, S. B. Shirey, R. W. Carlson, and J. W. Morgan1 ./st of Workshop Participants

6O

63

67

70

73

75

79

100103

97

94

91

88

85

82


7/110

IntroductionThe rationale for a Workshop on the Archean Mantle as part of

the Early Crustal Genesis Program comes from the seeminglyobvious conclusion that planetary mantles are the ultimate sourcesof planetary crusts. Consideration of the entire lithosphere system isappropriate, therefore, in broad-scale studies of how crust forms andevolves. Direct determination of mantle properties is possible onlyfor Earth in the present or recent past through studies of xenolithsand xenocrysts carried to the surface by kimberlites, lamproites, andalkali basalts, or through studies of exposures of upper mantle madepossible by the vagaries of tectonics. Indirect means are necessary todetermine the nature of Earth's ancient mantle and those of otherplanets. Characterization of Earth's Archean mantle, for example, ispossible by petrologic, geochemical, and isotopic study of ancientmantle-derived melts such as komatiite or basalt. Inferences can alsobe made from theoretical studies of thermal, chemical, and tectonicevolution of the early Earth. With these prospects in mind, theWorkshop on the Archean Mantle was organized with the objec-tives of considering and discussing evidence for the nature of Earth'sArchean mantle, including its composition, age and structure, influ-ence on the origin and evolution of Earth's crust, and relationshipto mantle and crustal evolution of the other terrestrial planets. Theworkshop was convened at the LPI on January 11-13, 1989, andwas attended by 41 participants from 5 countries. This volume con-tains extended abstracts of papers presented at the technical ses-sions, and extensive summaries of these sessions prepared by a fewcourageous "volunteers."

Lewis D. AshwalHouston, Texas

May, 1989


8/110

_ F_ j


9/110

ProgramWednesday Morning, January 11, 1989

SESSION I" ISOTOPIC AND CHEMICAL ASPECTS OF MANTLE EVOLUTIONChairman: G. N. HansonSummarizer: L. D. Ashwal

Boundary Conditions for the Archean MantleJ. H. Jones

The Pb and Nd Isotopic Evolution of the Archean MantleS. B. Shirey and R. W. Carlson

DiscussionIsotopic Evolution of the Archean Depleted Mantle

A. D. Smith and J. N. LuddenRe-Os Isotopic Constraints on the Chemical Ew_lution of the Archean Mantle

R. J. Walker, 5;. B. Shirey, R W. Carls_m, and J. W. M_rrganDiscussion

Wednesday Afternoon, January 11, 1989SESSION I: ISOTOPIC AND CHEMICAL ASPECTS OF MANTLE EVOLUTION (Continued)

Chairman: W. P. LeemanSummarizer: S. M. McLennan

Archean Crustal Recycling: Implications for Isotopic Heterogeneity in the MantleP. A. Mueller and J. L. Wo,den

Geochemical and Isotopic Constraints on the Composition of the Archean MantleC. Chauvel

Geochemistry and Isotopic Characteristics of Archean Basalts and Komatiites and Their Inference on Early Crust-Mantle DifferentiationB. M Jahn and G. Gruau

Discussion

SESSION II: CONSTRAINTS ON MANTLE EVOLUTION FROM SEDIMENTARY MATERIALSArchean Sedimentary Rocks anti the Archean Mantle

S. M. Mcb,,nnanAbundances of As, Sb, W and Mo in Early Archean and Phanerozoic Mantle-derived and Continental Crustal Rocks

K. W. Sims, H. E. Newsom, and E. S. GladneyDiscussion

Thursday Morning, January 12, 1989SESSION III: PHYSICAL ASPECTS OF MANTLE EVOLUTION

Chairman: K. BurkeSummarizer: P. Morgan

Mc, dels for the Thermal Evolution of the EarthF. M. Richter

The Diamond-Komatiite Paradox: Hot Mantle-Thick LithosphereP. M_rcgan

The Geology and Tectonics of Venus and _me Possible Implications for the Archean EarthJ. W. HeM

PRECEDII_G PAGE BLANK NOT FILMED p,__tNTF.NTIONk_,_:-_


10/110

4 Workshop on The Archean Mantle

Evidence for Suhduction and Spreading in the Archean Rock Record: Implications for Archean Tectonic Style and the Evolution of theSub-Cont inenta l Lithosphere

H, H. Helmsuwch and D.J. SchulzeDiscussion

SESSION IV: CONSTRAINTS ON MANTLE EVOLUTION FROM XENOLITHSChairman: A. J. NaldrettSummarizer: A. J. Irving

Precambrian Mantle Beneath Montana and British Columbia: Ge_u:hemica! Evidence from Eocene Volcanics and Their XenolithsA. J. Irving, H. E. O'Brien, avul I. S. McCaltum

The Archean Mantle Lithosphere Beneath the Wyoming ProvinceD. H. Eggter

Discussion

Thursday Afternoon, January 12, 1989SESSION IV: CONSTRAINTS ON MANTLE EVOLUTION FROM XENOLITHS (Continued)

The Ccmstitution of the "Real" Upper Mantle as Seen from XenolithsD, &'hulze

Evidence for a Differentiated Mantle and Plate Tectonics During the Late Archean Deduced from Eclogite Xenoliths in the BellsbankKimberlite

C. R. Neal, L A. Tayh,r, A. N. Halliday, P. Holden, and J. P. DavidsonDiscussion

SESSION V: CONSTRAINTS ON MANTLE EVOLUTION FROM MANTLE DERIVED MELTSChairman: S. B. ShireySummarizer: D. EIthon

Kolar Amphibolites-Archean Analogues of Mcdern Basaltic Volcanism?S. Balakrishnan, G. N. Hamcm, and V_ Rajamani

The Kolar Schist Belt, South India: Implications to the Nature of the Late Archean MantleV. Rajamani, S . Balakrishnan, E. J. KrogstM, and G. N. Hanson

DiscussionMendon Formation Komatiites: Extreme AI203/TiO2 Variation in the Uppermost Onverwacht Group of the Barbcrton Greenstone Belt

G. R ByerlyArchean FIt_.t _lsalts from South Africa: Implications for Local Heterogeneity in the Archaean Mantle

N. W. Rogers and J. S. MarshGec*chemistry and Geochronology of Late Archean Mafic-Ultramafic Supracrustal AmphiN31ites from the Northeastern Sino-KoreanCraton, China

W, G. Ermt and B.-M. JahnDiscussion

Friday Morning, January 13, 1989SESSION Vh CONSTRAINTS ON MANTLE EVOLUTION FROM MANTLE DERIVED MELTS(Continued)

Chairman: L. D. AshwalSummarizer:. C. T. Barrie

Petrogenesis of High Mg#, LILE-enriched Archean MonztxJiorites and Trachyandesites (Sanukitoids) and Granodioritic Derivatives inSouthwest Superio r Province

R A. Stern, G. N. Hanson, and S. B. Shirey


11/110

Technical Repor_ 89-05 5

Geochemistry and Nd-Sr Isotope Systematics of the Koamiskotia Area, Western Abitibi Subprovince, Canada: Implications for MantleProcesses During the Formation of the &mthern Superior Craton

C T. Barrie and S. B. ShireyDiscussionMantle-Crust Relationships in the Eastern Superior Province Inferred from Hf and Pb Isotope Studies

P. E. Smith and R. M FarquharMagma Evolution in the Stillwater Complex, Montana: REE, St, and Nd Isotopic Evidence for Archean Lithospheric Interaction

D. D. Lambert, R. J. Walker, S. B. Shirey, R. W. Carlson, and J. W. Mtr rganMantle Heterogeneity as Evidenced by Archean Mafic and Ultramafic Volcanic Rocks from the Laramie Range, Wyoming

S. M SmaglikDiscussionAdjourn Workshop


12/110


13/110

Summary of Technical SessionsThese summaries of presentations and discussions are

based on recordings made during the workshop and onnotes taken by those participants who kindly agreed toserve as summarizers. Discussion summaries are printed initalics. In most cases those participants who asked questionsor offered comments are identified, but this was not alwayspossible. We apologize to those we may have misquoted,misidentified, misinterpreted, or ignored.

SESSION I:ISOTOPIC AND CHEMICAL ASPECTS OF

MANTLE EVOLUTIONL D. Ashwal

After a brief introduction by L. Ashwal, the workshopopened with a talk by J. Jones on constraints that mightbe placed on the Earth's mantle at the beginning of theArchean. Formation of the Earth's core depleted the mantlein siderophile elements, as shown by the lower concen-trations of Ir, Au, Re, Os, etc., in mantle xenoliths comparedto chondritic meteorites (factor of 1-10-3). The constantCo/Mg of basaltic melts since the Archean shows that coreformation was complete before 3.0-3.5 Ga. Although theearliest mantle was likely to have been in equilibrium withFe-Ni metal (at or below the IW buffer), its oxidation statewas increased to QFM or thereabouts by 3_0-3.5 Ga. Therelatively constant Ni, Co, and Ir concentrations of mantlexenoliths suggest that a large-scale homogenization eventtook place at least as early as 3.0-3.5 Ga (from diamondinclusion ages), and possibly much earlier. Positive eNdvaluesof Archean mantle-derived magmas indicate that a large-scale (but incompletely understood) depletion event tookplace in the upper mantle between 4.4 and 3.9 Ga. Jonesbelieves, however, that some portions of the mantle mayhave escaped this differentiation, retaining, for example,undepleted heavy REE abundances as seen in some mantlexenoliths.

In discussion, H. Newsom asked Jones to elaborate on theexistence of undifferentiated mantle. Jones stated that thiswould be precluded by the "giant impact" theory wherebythe Moon formed by collision of a Mars-sized body with theearliest Earth, but that this theory is by no means proven.Other models could allow the preservation of "pristine" mantle.L Ashwal asked Jones to clarify whether this pristine mantlerepresented material undepleted in core-forming or crust-forming components, or both. Jones replied that he was referringto mantle undepleted in crust-forming components; refractorylithophile elements such as heavy REE are present in almost

"undisturbed" abundances. He pointed out that this couldalso be accounted for, although with difficulty, by complexrecycling of crust and mantle. Jones reiterated his view thatthe giant impact mechanism should result in a much moredifferentiated mantle, and "pristine" characteristics would notlikely be preserved. T. Irving pointed out that garnet lherzolitexenoliths do not show the undepleted heavy REE patterns.Jones stated to the concurrence of several participants thatsheared garnet lherzolites did show fiat-heavy REE. He alsostated that differentiated mantle materials are to be expected,but that it would be extremely important if undifferentiatedmaterials could be identified. B. Leeman asked how anyultramafic rock unaffected by differentiation or accumulationeffects could be uniquely identified, considering that manyxenoliths are probably crystal cumulates, and others are affectedby interaction with fluids or melts. He further stated thatit was amazing to see such homogeneity in siderophileabundances considering these effects. One possible explanationcould be a buffering effect related to the high partitioncoefficients of siderophiles in phases like olivine andorthopyroxene. Jones agreed that the siderophiles, in additionto the heavy REE behaved compatibly in this case but arguedthat their abundances represent primordial values. S. Shireyasked if the core could have buffered the composition of themantle over the course of geologic time considering the workof Jeanloz and coworkers indicating that the core.mantleboundary is extremely reactive. Shirey stated that if this isthe case and there is no apparent signature of buffering effects,then this might support models involving a stratified mantle.Jones replied that this depends entirely on how much 0 isin the core. An additional basis behind the idea of a highlyreactive core-mantle boundary relates to experiments showingthat at high temperatures U oxide dissolves in metal, implyingthat lithophile cations following 0 can voraciously entermetallic melts. Whether 0 is significantly abundant in thecore, however, is uncertain. B. Kaula pointed out that if thecore were so voracious, this would produce a chemicallydifferentiated layer at the bottom of the mantle. Whetheror not this is the case is a subject of much current debate.Jones replied that this depends on whether the core is saturatedin mantle components. Newsom stated to the agreement ofJones that if the apparent homogeneity of the mantle insiderophiles over time is due to buffering by the core, thenthis would imply the unlikely conclusion that the entire mantleinteracted with an O-rich core very early in Earth history.G. Ryder asked Jones about what was required to oxidizethe mantle to QFM conditions. Jones replied that it wouldtake a fair amount of material to change 8 wt.% FeO to90% FeO x 10% Fe203. This is not a problem in modelsinvolving accretion and later homogenization of a late.stage

PRECEDING PAGE BLANK NOT FILMED


14/110

8 Workshopon TheArcheanMantle

carbonaceous chondritic veneer, which would contain about20% H20. Leeman pointed out that H20 acting as a mantleoxidizing agent might have been more readily supplied bysubduction of oceanic crust and sediments. Jones wonderedwhere the original 1-120 in the crust came from. T. Naldrettmentioned that recent work on deep-sea glasses indicates thatthe mantle may be much less oxidized than QFM, perhapsQFM-2. This isconsistent with evidence frcrm magmatic sulfidedeposits, which crystallize pyrrhotite, not magnetite, indicatinglower f02 than QFM D. Eggler pointed out that the mantleisprobably not as reduced as IW (as indicated by intrinsicfo 2 measurements), nor as oxidized as QFM, and thereforemuch less early oxidation would be required. Jones expressedhis view that some change in the redox state of the mantlemust have taken place even if it is currently as reduced asIW Newsom asked about the timing of this oxidation event.Jones stated that it must have taken place very early, becausethe oldest samples measured have similar siderophile/lithophileelement ratios as modern materials, although once coreformation stopped, the mantle redox state could continue tochange by other processes. P. Morgan expressed the possibilitythat the Earth may still have been subjected to significantimpacts after coll ision with a Mars-sized object . He also pointedout that we must consider the energetics of the core-formingprocess--the Earth might be substantially different, forexample, if core formation continued to relatively late in Earthhistory. Jones replied that in terms of efiergetics most everythingwas tied up in the Mars-sized object. Regarding core-formingenergetics, this appears to be model-dependent. Whereas atotally molten Earth might be expected from rapid, early coreformation, models such as that of D. Stevenson, in which coreformation is delayed until the Earth is about Mars-sized couldresult in largely subsolidus heating. Jones clarified his positionthat the amount of post-Archean core formation is trivial.G. Ryder pointed out that he underst_ WetheriU's modelto show that the Mars.sized body probably would have beenthe among the last objects to hit the Earth. Jones stated thiswas model.dependent, but Kaula disagreed, stating that thereis a consensus among dynamicists that the main phase ofEarth's accretion involved impacts by large bodies (regardlessof how the Moon formed), and there would be a "tail-off'of smaller impacts lasting about lO0 Ma. He pointed outthat the giant impact hypothesis succeeds where others failin accounting for the Moon's depletion in Fe. Jones agreedthat although the bulk Moon is depleted in Fe, the silicateportionof the Moon is enriched in FeO compared to theEarth (unless D. Anderson is correct that Earth's lower mantleis twice as Fe-rich as the upper mantle). Newsom stated thatbetter evidence for the giant impact theory is the 23.5 tiltof the Earth's spin axis, but KauIa pointed out that this wasused by Safronov to argue against giant impact models. Leemanasked how lunar siderophile abundances compared to thoseof chondrites and Earth's mantle. Jones replied that lunar

si&'rophiles were also depleted compared to chondrites, withthe most highly siderophile elements showing the highestdepletions (factor of lO-S-IO-6for Ir, Re, and Os). lamarmaterials, therefore, behave as though they were in equilibriumwith metal. Leeman argued that since lunar samples are olderthan any Earth materials, the differentiation event involvingsiderophile depletion must have taken place very early. Jonespointed out that one fly-in-the-ointment is the commonly heldbelief that the Moon has a very small core. Ryder offeredanother: The Moon contains a substantial fraction of materialsfrom the Mars-sized impactor. Jones again said that this wasmodel-dependent. If, for example, the impactor had its owncore, then most of that material would wind up in the Earth.Kaula agreed and reminded the participants that anyaccretionary model must account for the equivalence of lunarand terrestrial oxygen isotope ratios. Jones added that ifRingwood is correct that the Moon is composed of


15/110

Technical Report 89-05 9

reservoirs and the type of convection operative (e.g., whole-mantle or layered) then are unknown. Ashwal asked if Shirey'smodel.would allow any early Archean continental crust. Shireyreplied that this was difficult to determine, but any continentalmaterial created very early must have been recycled back intothemantle wholesale. Models attempting to account for modernPb isotope distributions involve creation of "mature"continents, and if similar processesoperated during early Earthhistory, such continents should have been preserved. Newsomthen asked about later recycling of continental crust and theamount of crust present at the end of the Archean, referringto recent work by S.Jacobsen. Shirey finds it difficult to evaluatethis model (which involves complete recycling of continentalcrust with eN,tof-10 by 3.8 Ga) because of its assumptionsregarding interactions between discrete crust and mantlereservoirs. He expressed regret that Jacobsen was unable toattend the workshop. R. Stern asked if the light REE-enrichedbasaltic crust required by Shirey's model was reasonable. Shireyresponded that it was, considering that most of the extractionwould be governed by pyroxene.type distribution coefficients,resulting in slightly light REE-enriched basaltic mehs. Theamount of enrichment depends, of course, on the degree ofmelting. Stern then asked about the consequences of largelykomatiitic rather than basahic magmatism. Shirey replied thatin this case extraction must take place at sufficient depthsso that garnet played a significant role in controlling lightREE distributions. Considering that komatiitic magmatismrequires high degrees of partial melting, with residual garnetunlikely, this type of crust would probably be unsuitable forShirey's model. C. Chauvel asked about the thickness ofbasahiccrust required by Shirey's model. Shirey replied that a I0-20%basaltic fraction of upper mantle was required to produce thedepletion, and 20-25% of this must be recycled to accountfor the relative constancy of isotopic ratios during the Archean.Chauvel expressed skepticism about the constancy of Archeanisotopic data, suggesting, for example, that some of the samplesin Shirey's compilation may have been affected by crustalcontamination. Chauvel is convinced that there is, in fact,evidence for progressive depletion of the Archean mantle,expressed as an increase in ENain progressively younger rocks.Chauvel argued that the eNavalues ofsome of the lateArcheanrocks shown in Shirey's compilation were lowered by crustalcontamination, and that if these data points were removed,a progressive increase of e,a with time would be evident,although with considerable sca_ter reflecting mantleheterogeneity. Chauvel believes that at 2.7 Ga, eNa valuesvary between +2 and +8, the latter values being from theYilgarn block of Australia (e.g., Kambalda). Shirey askedChauvel if this would imply a single.stage extraction event,and Chauvel replied that this was possible. Shirey then askedabout the Kambalda samples, which show relatively uniformPb isotopic signatures but variable e,a. Chauvel replied thatthe Pb isotopic signature of these rocks is swamped by crustalPb even for the most depleted samples. Hanson commented

that the environment of greenstone belt formation is notsufficiently understood. We do not know, for example, whetherthe melts observedwerederived from subcontinental lit hosphereor oceanic mantle, and it may be erroneous, therefore, toconstruct mantle evolution models. Leeman concurred andstated further that la and K calculations are also model-dependent; single-stage la values may not be appropriate iflithospheric models are more complex. Shirey agreed, but addedthat it is useful to compare the Archean data set to thatfor modern rocks to see, for example, if the same level ofheterogeneity exists. Leeman stated that for many isotopicdatabases, there is inadequate characterization of traceelements and that until such work iscarriedout many questio_uwill remain unanswered. R. Lambert pointed out that two.stage, Pb isotopic evolution models might be too simplistic.A third stage of short duration involving pronouncedfractionation of U/Pb would cause "complete chaos" on Pbisotope evolution diagrams. Lambert cited evidence for thisfrom the PiIbara block of Australia where galenas and othersulfides associated with felsic rocks have higher # valuescompared to Pb associated with basic rocks. A third stageof I0-20 Ma duration, therefore, can cause significant shiftof apparent _1 values. He also stated that he published lalvalues of about 7.5 for Amitsoq gneiss (3.8 Ga), whereasShirey uses values of about 8.5 for early Archean rocks,suggesting a potential calculation problem. The lower valueswould plot on the Stacey-Kramers evolution curve.Shirey statedthat the Amitsoq gneiss is unique in being from ihe onlyearly Archean terrane that behaves coherently in terms ofPb; this is evidently not the case for other early Archeanterranes. With regard to Lambert's first point, Shirey statedthat there may be a problem in assuming that Pb isotopiccompositions of galenas represent those of the mantle. Shireyargued that galenas and other sulfides represent a separateupper crustal reservoir with elevated # values. J. Woodencommented that this was clearly recognizedas a crustal reservoirby Stacey and Kramers when they published their evolutionmodel. Shirey agreed, stating that he used this only as areference to show that according to his calculations, theArchean data did not show the early low-la stage. Thisprompted spirited discussionof Pb isotopic systematics amongseveral participants.B. Jahn then returned the discussion to the Nd isotopic

signature of the Archean mantle, stating that in consideringpossible causesfor the depleted signature of the Archean mantle,extraction of basaltic vs. continental-type crust are notmutually exclusive processes. Models of granitoid formationinvolve derivation from a basaltic precursor; and therefore weshould think of both types of crust-formation as part of thesame process. Shirey asked Jahn what mechanism he favoredfor destruction of continental rocks. Jahn replied that masstransfer of continental material back into the mantle tookplace by physical and chemical erosion and by subductionof the resulting sedimentary materials. Jahn stated that the


16/110

] 0 W_a'kshop on The Archean Mantle

eNa signature of Archean mantle-derived rocks includes asignificant recycled component, and a unique solution tounravel this effect does not yet exist. Shirey commented thatit is interesting that continental extraction does not seem toaffect the Pb isotopic signature considering the Pb, U, andTh concentrations in upper vs. lower continental crust andthe expectation that ancient continents weather from the topdown. The Pb isotopic signatures should be expected to bemore sensitive to these effects than Nd. Wooden commentedthat maybe we do see the signature of a recycled componentin Pb isotopic compositions, citing his work with MuelIer asa possible example. He stated fi_rther that it would not benecessary to recycle much continental material to account forthe Pb isotopic signature, but this was dependent uptm thenature of crust involved. Shirey agreed that this seems to workwell in explaining the data from the Wyoming Province, buthe expressed concern about the effects upon the Pb isotopesystematics of an early proto-crust. Jones asked if it werepossible that U and Pb are so incompatible compared to theREE that their signature in basaltic rocks represents largelyrecycled components. Shirey agreed that this was possible,referring to work by Galer and colleagues that suggests thatthe Pb concentration in the upper mantle is dominated bycontributions fiom other reservoirs such as the lower mantleor continental crust.Morgan commented that arguments against recyling of

continental _wust due to its inability to be subducted maybe misleading. It is thickness, not composition that is theimportant factor: Thin continental crust is just as subductibleas oceanic _:rust. Shirey agreed but added that in many placescontinental crust does not subduct because it is miderlain byrelatively buoyant, depleted mantle lithosphere. Morganreiterated that crustal thickness is the controlling factor insubductibility. Leeman pointed out that most workers considersuduction tg sediments rather than wholesale blocks ofcontinental crust, and there is abundant evidence for thisprocess. Shirey agreed but stated that in the early Earth, somemechanism must be called upon to destroy all of the extantcontinental crust, not just some fraction of it. G. Ernstcommented that the 4.1-4.2 Ga zircons from WesternAustralia imply the existence of at least some continentalcrust, which must have been subducted or otherwise removed.Shirey replied that all that remains _ continental materialfl'om the first 600 Ma of Earth history are about 100 zirconsand perhaps one rock from the Slave Province. Nearly allcontinental material, if it ever existed, must somehow havebeen destroyed. Leeman stated that Shirey's model relates tothe unresolved issue of whether the crust formed continuouslyor episodically. Shirey replied that it was surprising that nosignature of early continental "extraction was present in Pbisotopic systematics. Hanson terminated the discussion at thispoint arid adjourned the workshop for a coffee break.

The next talk was given by A. Smith, who discussedthe evolution of Archean-depleted mantle considering datafrom the Nd and Hf isotopic systems. His approach is totry to overcome potential problems with isochronsrepresenting mixing lines or reset ages by recalculating Ndinitial ratios using ages obtained from Pb-Pb or U-Pbsystems. This results in a large variation in initial Nd ratiosand conflicts with models favoring smooth evolution ofeNd with time. Instead, Smith and colleagues favor a two-stage model of mantle depletion since 4.55 Ga. The firststage established mantle heterogeneities, resulting eitherfrom shallow magma ocean fractionation or early basaltextraction. These are represented by Al-depleted komatiites(ADK) and ancient eclogite xenoliths (both showing thehighest end values) as well as Al-undepleted komatiites(AUK) and tholeiites (both showing lower t_q,t values). Thesecond stage, from the early Proterozoic t:o the present,shows relatively uniform increase in end of the upper mantleat a rate of about 2.2 t units per Ga. Smith suggestedthat the early Archean mantle was compositionally (andisotopically) stratified, but these heterogeneities werereduced at about the end of the Archean, perhaps by thegradual inception of convection.

In discussion, Jahn suggested that the difference betweenthe ADK and tholeiite sources may have been produced bygarnet fractionation, and Hf isotopes might better distinguishthe trends. Smith replied that there are Hf isotopic dataavailable for Pilbara and Abitibi rocks, although not for thesame samples run for Nd isotopes. The results show _HI valuesup to +8 firr 3.5 Ga Pilbara samples. Jahn commented thatthese data were not reliable because of dif ficulties withanalyt ical procedures and the presence of included zircons evenwithin komatiites. Jahn stated that he might present morereliable Hf isotopic data from Onverwacht (South Africa)samples in the afternoon session. Smith added that _Hfvaluesof up to +I2 for Abitibi (Superior Province, Canada) sampleswere reported by J. Patchett in 1981. Jahn wondered if Patchettwould still consider these data reliable. Smith stated that theAl-depIeted komatiites showing the highest end values havenot been anaylzed for Hf or Pb isotopes. Jahn questionedwhether the ADK and AUK source trends shown by Smithcould be distinguished unambiguously. Smith replied that ifall data are compiled, the ADK suite have consistently higher_Na values, which he il lustrated with a table. Jahn pointedout that changes in Sm/Nd ratios in these samples, for exampleby alteration, could cause'dramatic effects in calculated tN,tvalues. Smith replied that this was not compatible with theapparently consistent relationship between composit ion (ADKvs. AUK) and _,_a value. Chauvel asked Smith to explainhow ADK and AUK were distinguished. Smith replied thatthis was done on the basis of AI/Ti ratios and the amountof heavy REE depletion. Jones wondered if AUK could possibly


17/110

Technical Relxrrt 89-05 I i

represent ADK affected by crustal contamination effects. Smithreplied there were petrological arguments against this. Jonesdid not seem convinced. N. Rogers then suggested that detailedstudies of immobile trace elements were needed to evaluatewhether crustal contamination effects are important. Jahncommented that the Ta concentration of Onverwacht (SouthAfrica) komatiites shows a distinct crustal signature,interpretable either as contamination or source enrichmentwith crustal components. He added that many (but perhapsnot all) kcanatiites show some contamination effects. Smithagreed but emphasized that crustal contamination cannotaccount fi)r the very high end values of ADK, unless theirsource had even higher _ values. D. Elthon asked about thepossibility that some k{nnatiites may have had their Sm/Ndratios changed by interaction with pyroxene-rich cumulatematerials during magmatic scouring and corrosion duringemplacement. Smith said that this is possible, but that eNavalues would be reduced by only about 2 _ units if the effectsof such a processes were taken into account. Leeman raisedthe general question of the extent to which the" Sm/Nd ratiosof these rocks were changed by secondary processes such asmetamorphism, etc., and the effects this would have oncalculated eNa values. Smith replied that the rocks were atgreenschist grade and that REE mobilization was a possibility,but this process would have had to affect ADK and AUKselectively. An unidentified questioner wondered if this wasgeologically reasonable, i.e., whether they are found togetherin the same greenstone belt. Smith noted that most ADKare found mostly in early Archean greenstone belts, but inthe Norseman-Wiluna belt (Yilgarn Block, Western Australia)ADK and AUK are found together. Leeman asked if thetwo types were ever found interlayered with one another, andG. Byerly said this was the case in the Barberton greenstonebelt (South Africa). Smith added that except fi_r Norseman-Wiluna samples, isotopic analyses of ADK and AUK flomthe same greenstone belt were not available. Leemancommented that this would represent a good research project.The final talk in the morning session was given by R.

Walker, who summarized the geochemical and isotopesystematics of the Re-Os system and discussed potentialuses it may have for constraining early Earth processes andevents. The near equivalence in 187Re/lS6Os betweenchondrites (3.2) and Earth's mantle (3.3} is difficult toexplain. Possibili ties inch, de: similar Re and Os partitioningbetween core and mantle, retention of Re- and Os-richphases with chondritic Re/Os such as sulfides in the uppermantle during core formation, or post-core-formationaccretion of the upper mantle. The Re-Os isotopic systemmay also be used for geochronology, particularly for mafic-ultramafic rocks, as shown by data for Archean komatiitesfrom the Superior Province of Ontario. The decoupledbehavior of Re (highly incompatible) and Os (highlycompatible) suggests that this system can be used as a tracer

of crust-mantle evolution. Peridotite xenoliths from SouthAfrican kimberlites show marked depletions in 187Os withRe-Os model ages of about 2.8 Ga, possibly indicatingremoval of basaltic or komatiitic melts at or before thistime. With the exception of these rocks, the few samplesanalyzed to date for Re-Os isotopes plot at or near thechondritic evolution line and do not show the long-termisotopic depletion that is well established in other i.sotopicsystems such as Sm-Nd.

In discussion, Naldrett commented that the geochemistryof Re partitioning may depend on the amount of sulfide thatremains in the residual mantle source. Rhenium, fi_r example,may be left behind in residual sulfides during basalt genesis.Walker replied that this is not supported by data f(rc basaltsthat indicate Re concentrations as high as I-2 ppb. Naldrettthen suggested that the amount of sulfide left behind duringmagma genesis may have changed with time. Walker repliedthat more data would be needed to evaluate this. He alsomentioned the possibility that extraction of continentalmaterials, which can be expected to contain less Re than basalts,may tend to cause lesser Re depletions in the mantle. Jonesasked if it was possible that the depletion event that affectedthe Sm-Nd system may have predated the addition of Re.Os to the mantle. Walker replied that this was possible, butthat it is highly speculative at this point. Newsom commentedthat variability in incompatibility is documented in othergeochemical systems and that this would likely be discussedin the afternoon session. Shirey asked if Re-Os might behavelike the U-Pb system during basaltic recycling. Perhaps someRe could be added to the upper mantle in this way, and ifso, it might be fruitful to look fro" correlations between Pband Os isotopic compositions. Walker said that this waspossible but that most of the material would have to be mixedback in because the Re-Os system will be very sensitive tochanges in the Re/Os of the upper mantle. Schulze asked ifthe grouping of Os isotopic compositions among peridotitexenoli ths correlates with high-temperature vs. low temperaturexenolith types. Walker replied that there was no goodcorrelation between Os isotopic composition and high vs. lowtemperature types, but there is some correlation with mg'.He reiterated that the best correlation seems to be with theamount of basalt depletion in the xenoliths. Schulze pointedout that basalt-depleted xenoliths tend to be low-temperaturetypes and asked if it were lx)ssible to judge whether thosexenoliths with higher Os isotopic compositions may have hadcomponents added to them recently, fi;r example, bymetasomatic effects. Walker was not sure. D. Eggler wonderedif one of the data points with low Os isotopic compositionmight be from sample "161 1." Shirey replied that they didnot analyze that sample. An offer was made to provide it,and Walker cautioned the generous participant that they wouldneed a rather large quantity of sample. EIthon asked i f therewas any indication as to which mineral phases contained the


18/110

12 W, rksh_ponThe ArcheanMantle

Os. Walker said that this was a problem. Even if govd mineralseparate data were obtained, the possibility could not beeliminated that the Os was omtained in trace phases suchas osmiridium. He added that it is important to carry outpetrographic work on those samples selected for isotopicanalysis. Elthon asked if there was any evidence for mobilizationof Re or Os, such asdifferences in isotopic compositions betweenmetasomatized vs. nonmetasomatized xenoliths. Walker repliedthat there was no evidence in the isotopic compositions forsueh effects. In terms of concentrations, however, one sampIeof highly sheared spinel lherzolite from Lashaine in Tanzania(reported in the Basaltic Volcanism volume) has extremelylow concentrations _g Re and Os, but Walker asked if anyoneknew whether this is a metasomatized sample. We were toldthat this unusual sample shows trace element enrichment butmajor element depletion. Walker added that this sample hasa very low Os isotopic composit ion but doubted if this couldbe ttssumed representative cg metasomatized mantle. Walkerstated his expectation that metasomatism should not affectOs because of its immobility. At this point Chairman Hansonadjournedthe workshop for lunch.

SESSION I.'ISOTOPIC AND CHEMICAL ASPECTS OFMANTLE EVOLUTION (CONTINUED)

S. M McLennanAfter lunch the session continued with B. Leeman acting

as chairman. The first paper was given by P. Mueller, whoevaluated the Pb isotopic evidence for crustal recyclingand mantle heterogeneity during the Archean. Three typesof Archean cratons were distinguished on the basis of Pbisotopic ratios and included: (1) those falling on the mantlegrowth curve (Type I; new crust without components ofolder recycled crust, e.g., Superior Province); (2)thosefalling below the mantle growth curve (Type II; crustundergoing high grade metamorphism early in its history,e.g., Labrador and Southwest Greenland); and (3) thosefalling above the growth curve (Type III; crust that isdifferentiated early in its history, e.g., Wyoming Province).Recycling into the mantle of these different crustal typesmay impart considerable isotopic heterogeneity on themantle, although recognition of the type of craton thatmay be involved often is not straightforward; crust-mantleinteraction associated with development of Type Ill cratonsis most likely to produce recognizable enrichments. It isalso a difficult matter to estimate the timing of theintroduction of heterogeneities. Mr,tiler suggested thatyoung volcanic rocks found in the northwestern U.S.A.(e.g., Snake River Plateau) may have been derived in partfrom mantle that was enriched through interaction withType III cratonic material during the Archean.

The discussion began with a question from S. Shirey whowanted to know about the Pb and Nd isotope mass balancefor recycling of 3.3 Ga crust int_ the mantle, given the availableconstraints for the isotopic composition of the recycled crustalcomponent and the depleted mantle. Mueller responded thatthe generally accepted values for "depleted mantle" were&rived, to a large extent, from the southern Superior Provinceand it was not at all clear if such values were appropriatefor the Beartooth region. He pointed out that similar 3.2Ga rocks from west of the Beartooth gave higher e,a values(-1). In addition, Mueller noted that all rocks from theBeartooth region have essentially the same Nd-isotopicsignature, with no apparent evidence for subcratonic mantledepletion, over the period 32-2.7 Ga and that very efficientmixing must have been taking place to produce the limitedrange in isotopic composition during the late Archean.G. Hanson continued the questioning and asked what age

was produced by the regression of initial Pb isotopic ratios.Mueller responded that the data were highly scattered butthat for the rocks that were considered to be the oldest inthe region, Nd model ages, Pb-Pb whole rock scatterchrons,and Rb.Sr scatterchrons all produced ages of about 3.4 Ga.He also noted that a couple of Nd-model ages and some Hfages on zircons, published by J. Patchett, gave ages of about3.6 Ga.

While some minor technical difficulties with an overheadprojector were sorted out, a final questioner noted that anarc analogue was being proposed and questioned whether theisotopic- homogenization that was observed was consistent withisotopic systematics seen in present arcs. J. Wooden (coauthorof the paper) responded that the variations were comparableto modern arcs, except those that display extreme radiogenicsignatures. Leman noted that even in young arcs displayinghighly radiogenic isotopic signatures, such as the Lesser Antilles,the highly radiogenic component may be added at high levels(i.e., contamination) and that the material going down thesubduction zone may be relatively homogeneous. Muellerpointed out that once the Primitive Mantle separates intodifferent reservoirs, all interaction between reservoirs can beseen as contamination. He noted that, in general, a narrowrange in isotopic composition suggested deep efficient mixing,whereas a wide range suggested shallow, less efficient mixing.The second paper of the afternoon was presented by

C. Chauvel. She first reviewed the Nd isotopic evidencerelating to the Archean mantle from noncontaminatedvolcanic rocks and banded iron formations. She concludedthat the Archean mantle, from 3.8 Ga, was generallydepleted in nature but displayed considerable isotopicheterogeneity. This presented a dilemma because enrichedcontinental components appear to be minor. A model waspresented to explain these and other features of theArchean record. Prior to about 3.8 Ga, the model calls


19/110

Technicaleport9-05 l 3for a thick (


20/110

14 W_rrksh,_p on The Archean Mantle

out that detailed physical geology of various greenstone beltsalso differed considerably and that this should be taken intoconsideration when making comparisons.

SESSION II:CONSTRAINTS ON MANTLE EVOLUTION FROMSEDIMENTARY MATERIALS

S. M McLennanThe first paper of this brief session was presented by

S. McLennan, who suggested that the scarcity of negativeEu-anomalies in greenstone belt turbidites suggested thatintracrustal fractionation processes, in the crustalprecursors, were limited. Trace element and Nd-Sr isotopechemistry of modern continental margin turbidites indicatethat such sediments commonly bad a significant youngmantle-derived component. Comparisons were madebetween Archean and Recent turbiditts to see if therewere differences that could have implications for mantlecompositions. Archean turbidites tended to have higherEu/Eu*, suggesting that they had been less affected byintracrustal processes. Archean turbidites commonly hadhigh Gd/Yb ratios indicating a significant HREE-depletedcomponent; such a component is absent from Recentturbidites. He suggested that a HREE-depleted mantle-derived component would indicate P-T conditions of crustgeneration and/or mantle compositions may have differed.It was also noted that Recent active margin turbiditescommonly have low Th/U ratios (1.0-3.0), probablyreflecting a depleted mantle source. Such low ratios areseen to be absent from Archean turbidites, perhapsindicating that Archean mantle sources of the crust wereless depleted then at present.

The first questioner asked whether simple in situ decaycould explain the high Th/U ratios for the Archean samples.McLennan responded that he did not think that decay couldcause such large changes in the ratio (up to a fi_ctor of 3)but that in any case all of the mantle sources would equallysuffer from in situ decay; therefore, the differences betweenArchean and Recent Th/U ratios could not be explained insuch a way. S. Shirey then asked how many cratons wererepresented with Th/U data. McLennan responded that thePilbara, YiIgam, Kaapvaal were represented with a few samplesfrom the Superior Province. He further noted that much ofthe available data was by instrumental neutron activationanalysis and chat the quality of the Th/U data was oftendifficult to judge by this technique and therefore not included.C Chauvel then asked if there was a distinction betweenEarly and Late Archean samples in terms of Th/U ratio.McLennan could not recall specifically whet her such differencesexisted, but he had examined sedimentary trace element data

for secular variations and probably would have noted adifference in Th/U ratios.Another participant asked how a turbidite was best sampled.

McLennan suggested that a lower coarse-grained and upperfine-grained sample should be taken but that any pelagiccomponent preserved in the E-unit should be avoided. Inresponse to a query about whether significant differences existedbetween coarse and fine fractions, McLennan noted that insome cases, but not all, a separation of provenance componentsmay take place, resulting in significantly different Nd-isotopiccompositions. The next question centered on the role ofaccessory minerals in explaining differences in REE, Th/U,and other trace element abundances. McI_ennan replied thatthere is no special enrichment in heavy minerals that couldexplain such differences; for example, Zr abundances weregenerally low at


21/110

TechnicalReport89-05 15

mechanism may be in operation. Sims responded that the Pb/Cedata were from Hofmann's studies and that he suggested Pband Ba behaved similarly in early crustal formation but withlater fractionation Pb and Ce behaved similarly.H. Newsom then asked for comments on Hofmann's model,

which calls for early crust formation (with enriched Pb relativeto Ce) and a subsequent mantle homogenization event resultingin constant Pb/Ce ratios for ocean island basalt (OIB) andMORB reservoirs. He also noted that the model required littlecrustal formation or recycling after this time. A vigorous andwide.ranging discussion ensued. G. Hanson noted, and B.Leeman agreed, that the Kjs in melts are very low and, inorder to fractionate the highly incompatible elements, verylow degrees of partial melting would be required. He wenton to say that he agreed that movement of fluids, which likelyconstitutes a small proportion of the system, may be amechanism that would allow fractionation of these elements.Shirey pointed out that the average age of the MORB andOIB reservoirs is about 1.6-1.8 Ga and perhaps is dominatedby' Proterozoic and Phanerozoic history with the informationof Archean heterogeneity being lost by subsequent mixing.Hanson noted that calculations by F. Richter indicated thatheterogeneities were very rapidly lost in the mantle with 3-4cycles of convection, each on the order of 200 Ma.The discussion then turned to the question of crustal growth

rates. It was noted that certain ratios among highlyincompatible elements, such as Pb/Ce, were constant for OIBand MORB reservoirs, and this would not allow for significantcrustal growth since the Archean. This was apparentlyinconsistent with models calling for


22/110

16 Workshop on TheArcheanMantle

presented his latest models for thermal evolution of themantle. His calculations, constrained by the deductions thatcontinental geotherms have not changed significantly butthat the Archean global heat loss was greater than atpresent, indicate that the mantle has been cooling at arate of 50-100C/Ga, and the Archean mantle was about300(2 hotter than the modern mantle. He suggested thatthe only lithosphere preserved in the Archean was a distincttemperature-insensitive, chemically layered lithosphere.The Archean-Proterozoic boundary was interpreted to markthe time at which mantle temperatures dropped to a levelwhere all continental lithosphere was preserved.

K. Burke opened the questions for Richter's paper by askingif there are any chemical data to support the very refractorylithosphere required in Ri&ter's models (and previouslypublished tectosphere models of T. Jordan). Richter repliedthat xenolith data indicate refractory mantle to depths of 150km, but the interpretations of deeper sheared xenolith dataare uncertain. Shirey remarked upon the variabili ty in thicknessof l ithospheric roots beneath shietxts, to which Richter repliedthat seismic data used to develop these models averagelithospheric properties over large areas. G. Hanson asked foran interpretation of the upper-mantle seismic V_data in termsof mineralogy and chemistry. Richter responded by referringto publications of Jordan on the tectosphere, and summarizedby saying that the mantle lithosphere was highly depleted.Leeman asked why the 400-kin-thick tectosphere of Jordanbothered Richter, to which Richter replied that his modelsonly required a 200-kin-thick lithosphere and any greaterthickness was superfluous. Kaula questioned Richter aboutmechanisms to achieve the lateral upper-mantle heterogeneityrequired by the models. Richter replied that his models didnot yet address this problen_, although it was a worthy topicfirr future studies, and computational resources are justbecoming available that might make the problem tractablein two dimensions. R. Lambert commented that there is somecritical level to which the Earth must cool before a lithospherecan form, "something to do with the rheology." D. SchuIzecommented that sheared mantle xenoliths do not constraintithosphere thickness because of their possible association withthe asthenosphere and may not be a good sample of old mantlematerial. N. Rogers remarked that some xenoliths indicateextraction of a komatiitic component, and this extraction mayb_ signifcant in development of the lithosphere, to whichRichter replied that in his madels only the property that thematerial is relatively undeformabte at high temperatures isimportant. J.Jones requested information about the distributionof heat production in the models, to which Richter respondedthat some heat production was concentrated in the uppermostlayer (crust) but that the main part of the mantle was assumedto be adiabatic and well stirred, and thus relativelyhomogeneous.

In the next talk P. Morgan discussed his investigationsof the possibility that shield mantle may be stabilizedthermally to produce the lithosphere boundary layer. Inorder to explain a thick Archean continental lithosphere(as suggested by Archean-age diamonds) with high mantletemperatures (as suggested by Archean komatiites), heproposed a thermal model of the lithosphere in which asmall concentration of heat-producing radioisotopes in thelithospheric mantle cause the lowermost lithosphericgeotherm to be asymptotic to the asthenosphere adiabat.He demonstrated that such conditions were not theoret-ically unreasonable and may be a consequence ofmetasomatism and tectonics associated with stabilizationevents in the continental lithosphere.

D. Eggler opened the discussion with a remark concerningthe problems in determining mantle heat generation becauseof metasomatism of the samples. Morgan agreed with therestrictions caused by this problem but concluded that hismodels are consistent with reasonable interpretations ofavailable data, even accounting for metasomatism. F. Richtersuggested that Morgan should test the temporal stability ofhis thermal lithosphere models, to which Morgan respondedthat this would be a g ood next step in the study. The discussionof this paper continued with a request by K. Burke forclarification of the mechanism proposed by Morgan for irk'reasein mantle lithosphere heat generation by collisionalmechanisms, which Morgan explained in terms of pure lateralstrain in the lithosphere. R Lambert made the final remarkof the discussion by noting that the Atlantic cut throughportions of shield lithosphere.

An extraterrestrial analog for Earth's Archean heat lossand tectonics was suggested in an enveloping review ofVenus geology and tectonics by J. Head. He examined Venusin terms of possible zones of convergence and divergence,intraplate volcanism and tectonics, global patterns oftectonics, the forces driving tectonics, and the link betweentectonics and mantle convection. Similarities anddifferences between Venus and Earth suggest that Venusmay be used as a viable laboratory to study terrestrialprocesses under different conditions to modern Earth.Venus has a very high present surface temperature (560C),possibly resulting in a hot, thin lithosphere as may haveexisted on Earth in the Archean. Head concluded thatthe equatorial highlands of Venus may representfundamental aspects of deeper convection in the venusianinterior, perhaps a model for terrestrial mantle convection.

B. Kaula opened the short discussion of this paper witha comment that Head had shouts that there may be anAtlantic-type (slow) spreading ridge on Venus, but there wasno Pacific-type (fast) spreading ridge, unless it was in the


23/110

Technicaleport89-05 1720% of the planet not mapped by the Pioneer Venus altimeter.In this respect, Venus may be the normal pIanet and Earthmay be abnormal. The dominant heat loss mechanism fromVenus does not appear to be sea-floor spreading, and ahhoughmantle convection must exist to explain the high plateaus onVenus, the important question is whether it is impinging ona lithosphere or a crust. This discussion was cut short bythe chairman because of lack of time. S. Shirey asked if therewere any geochemical data f irr lshtar Terra, the continent-like area in the northcmt hemisphere of Venus. Head repliedthat the only geochemical data available firr Venus was fromSoviet Ianders in Beta Regio in the equatorial regions of theplanet. The data indicate high MgO, but additional datathat may be able to detect the effects of recycling are notyet available. Further questions were postponed to the generaldiscussion.

The final paper of the session, presented by D. Schuhe(for H. Helmstaedt), was a discussion of the role ofsubduction in the evolution of Archean subcontinentallithosphere. Lithologic, isotopic, mineralogic, and xenolithdata were presented to argue that ophiolites are preservedin Archean and Proterozoic greenstone belts, which aretangible evidence for early sea-floor spreading andsubduction. By linking higher Archean heat loss with higherrates of spreading, they assumed that Archean s'ubductionzones were characterized by high convergence rates andlow-angle subduction. They suggested that entire cratonsmust have been underplated by oceanic lithosphere,resulting in complex imbrication of the Archeansubcontinental upper mantle, probably somewhat homo-genized during high-grade metamorphism, partial melting,and later intrusions.

A. Smith opened the discussion with a question about whatwould happen to some of the ages given in the paper if theprimordial eNa value were not zero but positive. Schulzeresponded that these data were not his own, but that theywere not central to his arguments. T. Irving asked if Schulzehad any new data concerning carbon isotopes in diamon&.Schulze responded that these isotopes were different fromnormal mantle isotopic ratios, and probably indicated arecycling origin for the carbon in diamonds. G. Ernst askedif the diamonds were associated with eclogites, to which Schulzereplied that younger diamonds were associated with eclogites,the Archean diamonds were associated with peridotites. S.Shirey asked how it was possible to distinguish eclogite xenolithsfrom subduction zones from those of lowe'g crust. Schulzeexplained that they could be distinguished on the basis ofmineralogical equilibration temperatures and their verydifferent thermal histories. L. Ashwat asked if the grospyditesin the rocks discussed by Schulze may not originate from Ca-rich anorth_sites, although they would still have a subductionevent in their histories. Schulze replied that he could not be

certain of their protoliths. Jahn asked if there was a way tobe certain that the eclogites were of Archean age, to whichSchulze replied that he could not be certain; they were modelages that he presented, and they were not his data. R. Sternconcluded the discussion of this paper by asking what laybeneath the base of the ophiolites. Schulze referred the questionto his absent coauthor, but the audience offered the suggestionthat the ophiolites were underlain by granites.

SESSION IV:CONSTRAINTS ON MANTLE EVOLUTION

FROM XENOLITHSA. J. Irving

Two presentations by T. Irving and D. Eggler dealt withthe petrogenesis of Cretaceous-Tertiary potassic igneous"rocks in Montana, Wyoming, and Colorado and thegeochemical and xenolithic evidence for the existence ofancient (probably Archean) subcontinental mantlelithosphere beneath these regions. Irving presentedgeochemical data for Eocene minettes and related rocksfrom the Highwood Mountains and made comparisons withdata for other potassic rocks from the Wyoming province.He argued for the existence of three distinct sourcecomponents. The large negative end values and 1.87-2.1Ga Pb-Pb ages found for the Highwood and Smoky Butte.vokanics appear to be readily explained by interaction ofasthenospheric melts with Precambrian mantle lithosphererelated to the Wyoming craton. Irving also proposed a thirdcomponent related to Eocene subduction processes in orderto explain the Rb-Sr isotopic systematics.

In discussion, Eggler questioned such a direct link betweenactive subduction and the potassic volcanism, preferring insteada back arc, postsubduction model. The existence ofmetasomatized ancient mantle lithosphere beneath centralMontana was supported by Irving's descriptions of mica-bearing ultramafic xenoliths containing veins of glimmerite.He suggested that the veins could be dated isotopicaUy andmay turn out to be Precambrian metasomes, possibly relatedto ancient subduction events. S. Shirey recommended that thesamples definitely would be amenable to Rb-Sr datingtechniques.

Eggler summarized his extensive work on alkalic volcanicsfrom the Crazy Mountains and Absaroka Mountains inMontana and on peridotite and eclogite xenoliths fromColorado-Wyoming kimberlites. He further developed theidea of a thick mantle lithospheric keel composed ofmetasomatized, previously depleted perioditite beneath theWyoming province. In an intriguing comparison of present-day geophysically determined geotherms and fossilgeotherms deduced from thermobarometry calculations on


24/110

18 "#_,'kshopmhe Archean Mantlexenoliths, he concluded that much of the pre-Cretaceouslithospheric keel has been advectively thinned. As a resultthe only remaining thick portions are in central Montanaand southeastern Wyoming.

Discussion followed. J. Wooden made a plea for cautionin interpreting Pb-Pb whole rock ages. R. Lambert pointedout the evidence fi_r extension of the mantle keel to the northbeneath the Crows Nest w_lcanics.

Eclogite xenoliths found within kimberlite pipes werethe focus of two presentations by D. Schulze and C. Neal.The Sr and Nd isotopic studies of inch, sions within diamondshave indicated that some of these eclogites are Proterozoicto Archean in age. There is no question that these rocksare a component of the Precambrian mantle, representingpossibly frozen basaltic liquids, cumulates, or subducted andmetasomatized basaltic rocks, but there is much uncertaintyas to their abundance relative to peridotite. Schultzedescribed many examples of coesite and sanidine-bearingeclogites from South African and Siberian kimberlites andconcluded that they most likely represent subducted ancientocean floor basahs rather than fractionated basaltic liquids.He contended further that there is a sampling bias inxenolith suites from kimberlites because of the greatercohesiveness of eclogite relative to garnet peridotite, andthis appears to be borne out by his analysis of mineralconcentrates from the pipes. Thus eclogite might constituteonly 3-15% of the Precambrian mantle. Schultze madethe suggestion that garnet peridotite xenoliths might bemore readily disaggregated because of reaction between theirorthopyroxene and the kimberlite magma; however, Nealquestioned the viability of such a mechanism.Neal presented some exciting elemental and isotopic data

for a suite of eclogite xenoliths from a South Africankimberlite. The study is important because it is one ofthe first to obtain coordinated Sr-Nd-O isotopic data onultrapure mineral separates from these rocks. Three typesof eclogite were recognized and interpreted as mantlecumulates, subducted oceanic basalts, and subductedgabbros (preserving positive Eu anomalies in both garnetand clinopyroxene). The most startling result is theextremely high end and 2.3-2.5 Ga model ages for eclogitesof the second group, which Neal explained by proposinga late Archean, light REE-depleted MORB progenitor.Much discussion fl_llowed, particularly by C. Chauvel, P.

Morgan, and G. LiBerge. F. Richter expounded on the semanticpoint that the mounting evidence for a subducted ancientMORB origin for many eclogites did not necessarily arguecategorically for Archean plate tectonics, Further commentaryon this ensued, with the general consensus that s_me flrcm

of plate motion was the most reasonable mechanism .f(rcrecycling crustal material into the Archean mantle.

SESSION V:CONSTRAINTS ON MANTLE EVOLUTION FROM

MANTLE-DERIVED MELTSD. Ehhon

G. Hanson opened this session with a discussion of theP-T conditions of origin of komatiites and tholeiitic basahsand then turned to a discussion of the geochemical resultson the Kolar Schist Belt in southern India. He distinguishedfour different terranes in the region: (1) a belt of felsicgneisses in the west, (2) a western belt of amphibolites thathave LREE depletions, (3) an eastern belt of amphibolitesthat are LREE enriched, and (4) a belt of felsic gneissesin the east. The protoliths for the amphibolites areconsidered to be mostly tholeiitic/komatiitic basalts, witha few komatiites. The eastern and western belts differ intheir Nd and Pb isotopes. It was suggested that thesedifferent terranes were derived from different source regionsand were assembled into their present configuration at 2500Ma ago.

In discussion, K. Burke asked Hanson about the conditionsof melting and what differences there were from melting inthe present-day mantle. Hanson suggested that the komatiiteswere produced during the initial phase of melting at highpressures (50 kbar?), but the b_tsalts were produced by meltingat a later stage in the ascent of the mantle at lower pressures(20-25 kbars). T. Naldrett made a general comment thatkomatiites can assimilate a substantial amount of materialduring ascent because of their high temperatures and suggestedthat these effects should not be neglected. H. Newsom askedHanson about the percentages of melting invoh,ed in theproduction of magmas in the Kolar Schist Belt . Hanson repliedthat he, unlike others, did not advocate high extents of partialmelting but preferred a model invoh, ing a small percentageof melt production at high pressures. J. Jones noted that atvery high pressures, where the liquidus and solidus of peridotiteconverge, relatively small temperature changes can producesubstantial changes in the extent of melting.

The next presentation was by V. Rajamani in whichhe pursued the question of the relationship between thetholeiitic and komatiitic rocks from the "Kolar Schist Belt.He proposed that komatiites were produced by melting at>100 km depth, but that the tholeiitic basahs wereproduced by melting of a modified mantle at


25/110

TechnicalRel_rrt89-05 19

indicate that the tholeiites are not derived from eitherthe komatiites or from a common parental liquid, at leastby any relatively simple model. He concluded that thetholeiites were derived from a variety of Fe-rich mantlecompositions and that the tholeiites were not related toeach other by crystal fractionation processes.

In discussion, Ehhon noted that modern basahs, l ike thesesamples, typically have Fe/Mg too high to be in equilibriumwith the mantle and that most interpretations suggest thatthis is an effect of crystal fractionation, not melting of anFe-rich mantle. Rajamani and Hanson c_llectively noted thatthe compositions of these basahs do not lie along any commonliquid line of descent as would be expected if crystallizationdominated. They also noted that the geochemical data (e.g.,Ni vs. Mg) often clustered in a quite small region of a plotand that there were no associated cumulates fi)und in theregion. Burke, however, questioned whether cumulates mighthave been originally present but are no longer fi_und in theregion due to tectonic processes. P. Smith asked how closethe samples studied here were collected from areas of goldmineralization and noted that LREE enrichment is found inAbitibi samples associated with gold mineralization. Rajamaniresponded that these were collected 500-600 m from goldmineralization areas and did not feel that LREE enrichmentassociated with gold mineralization had influenced his samples.S. Shirey noted the HREE fractionation seen in the Kolarsamples was similar to that described for numerous Archeansamples the previous day by Jahn. Jahn agreed and notedthat Fe-rich basahs such as those described here are quitecommon among Archean samples. He also noted that theyare often associated with banded iron formations.The session continued with G. Byerly's discussion of the

geochemical relations within komatiites from the Mendonformation of the Barberton Greenstone Belt. He notedthat there have been several distinctive episodes ofvolcanism, but low angle faults and folds complicate thestratigraphy of the volcanic rocks. With the use of threedistinctive marker beds, however, a volcanic stratigraphyof the region has been developed in which six distinctiveflows have been identified. Komatiites from the Mendonfoimation are characterized by a high AI2OJTiO2 of 30to 100, whereas most other komatiites from Barberton havea ratio of _10. The AI and Ti abundances are wellcorrelated, particularly in the relatively fresh samples,suggesting that these elements are relatively immobile.Alteration is extensive in many of the samples, however,and it appears that many of the highly altered sampleslie along quartz-sericite mixing lines. Chromium spinels arefound in many samples and appear to reflect magmaticrather than metamorphic conditions; most spinels haveCr/(Cr+AI) of 0.55 to 0.90. The ratios of severalincompatible elements are quite variable, particularly in

the Mendon formation. For the freshest rocks, theabundances of Hf, Th, and Ta suggest that these sampleswere part of a magmatic arc terrane.

In discussion, Jones inquired about the conditions underwhich strong silica enrichment occurs. Byerly responded thatthese effects were probably related to hydrothermal convectioncells, possibly one in which periodic evaporation producedconcentrated brines. He suggested that the environment offormation may have been in a back arc plateau with highheat flow. Shirey asked if he had looked at the silica-richzones to determine their chemical relationships. Byerlyresponded that these zones are mixtures of silica, sericite, andcarbonate material. C. Chauvel asked whether these sampleswere the same as those that Wasserburg's group had studied.Byerly responded that they were from the same region butwere not the same samples.

N. Rogers briefly reviewed the types of petrological datathat are relevant to discussions of the composition of theArchean mantle in the context of evaluating thegeochemical variations in Archean Dominion Group basaltsfrom South Africa. In particular he noted that SouthAfrican garnet peridotites have lower FeO* than spinelperidotites and that the spinel lherzolites have a slightly(_1%) higher density as a consequence. He suggested thatparental MORBs have approximately the same FeO* asspinel peridotites, but that komatiites have higher FeO*and would be more suitably linked with the garnetperidotites. The Dominion Group basalts have low TiO2(often


26/110

20 Workshop onThe ArcheanMantleanticipated based on previous reports of Early Archeandates. The basaltic rocks were derived from depleted mantlethat had been depleted over approximately 1 Ga.

In discussion, Rajamani asked what the rocks were likepetrographically at present. Ernst replied that they had beenstrongly recrystallized with no pillowed or spinifex texturesremaining. The dominant minerals are hornblende, interme-diate plagiodase, and clinozoisite. Jahn noted that it ispossiblethat the oldest samples in the area may not have been sampledin this study. He indicated that rocks of 2.99 Ga ages havebeen reported from this general region. Burke and R. Lambertnoted that the geology of the region is very con_plex, with2400 Ma and 1800 Ma metamorphic events, and numerousintrusions and fauhs that cross-cut the region. Thesesubsequent events complicate and obscure the Archean geology.

SESSION VI:CONSTRAINTS ON MANTLE EVOLUTION FROMMANTLE-DERIVED MELTS (CONTINUED)

C. T. BarrieThe last session of the 'workshop opened with a talk

by R. Stern on an unusual suite of mildly alkaline,predominantly intrusive rocks found in the southwesternSuperior Province. The suite has chemical characteristicssimilar to LIL-and LREE-enriched, high-Mg andesitesknown as sanukites in the Setouchi volcanic belt of Japan.Accordingly, this suite has been termed a "sanukitoid" suite,along with spatially and chemically related monzodioritesand granodiorites. Using FeO or Ni vs. MgO plots andREE ratio diagrams, it is apparent that the more primitiverocks of the sam, kitoid suite cannot be derived from typicallamprophyric or tholeiitic compositions through anycombination of fractional crystallization, partial melting ofany single mantle source, or contamination by reasonablecrustal compositions. Stern suggested that the primitivesanukitoids were derived from a hydrous, LREE- and LIL-enriched mantle source by partial melting. This wouldexplain the rather constant Ce/Nd ratios over a wide rangeof Ce concentrations, which cannot be accounted for byvariable degrees of partial melting with a garnet-bearingresiduum. The more evolved granodiorites can be modeled-as fractionation products from sanukitoid parental magmas.

This talk was greeted with a very strong interest from theaudience. T. Naldrett requested clarification on the relationshipbetween the hydrous mantle source environment fi_r thesanukitoids and the trace-element enrichment process. Sternstated that the timing of the enrichment process must havebeen nearly synchronous with sanukitoid emplacement, aslocalized andesitic melts in the mantle would certainly rise

shortly after formation by virtue of their lower density andviscosity with respect to surrounding mantle material, t3.Leeman questioned whether the isotopic composition of therocks could constrain the timing of mantle extraction. Sternreferred to recent work by Shirey and Carlson that foundthat the sanukitoid suite has eNa values of +l to +2.5,suggesting that the source had a long-term history of LREEdepletion. However, the suite's LREE-enriched chemicalsignature would imply that the enrichment event occurredshortly before emplacement. H. Newsom wondered about thephysical characteristics of the suite and implications fire thesize and ge_nnetry of the mantle sources. Stern said that basedon his studies of the Roaring River C,,nplex, sanukitoid magmasprobably fractionated en route to supracrustal levels,accounting for the associated, relatively voluminousgranodiorites. Few lavas have been found that may be includedwithin the suite. L Ashwal mentioned that tholeiites in theregion have similar eNa values and queried whether they couldbe derived from the same source that had not been subjectedto local contamination immediately prior to partial meltingand extraction. Stem replied that this was possible in a generaIsense if you could account for variable MgO and FeO inthe source; one possibility to produce this variability wouldbe to call on earlier extraction events.At this time a forest of hands rose in the audience. D.

_hulze asked for clarif ication on the depth _g partial melting.The response was that it is constrained to 10-15 kbars byexperimental work on hydrous peridotite melts. T. lrvingwondered about the high sodium content. Stem replied thatthe suite does have unusually high sodium contents, perhapsattributable to melting of pIagioclase or jadeitic pyroxene, orto sodium behaving as an incompatible element in an enrichedfluid phase, t3. jahn asked about the high silica content _fmantle-derived melts and alluded to the possibility of silicabeing part of the fluid phase added to the mantle. Stern madethe analogy with boninites, which are commonly believed tobe mantle-derived and are silica rich, and agreed that highsilica contents could be attributed to a supercritical fluid meh.R. Lambert requested a petroIogical description for these rocks,considering the possibility of mixed magma sources, andwondered where the LREEs reside. The mineralogy is variablebut generally typical of diorites, with the bulk of the traceelements tied up in sphene and apatite. Jahn made an analogybetween the sanukitoid suite and syenodiorite suites in Chinaand France and wondered if alkali basalts could serve asparental magma. Stern said that there are no known alkalibasalts in the region, but lamprophyres have been consideredas possible parent magmas and they are easily dismissedconsidering their relatively high Mg numbers. In the Setouchiregion of Japan, alkali basalts are fi_und near the type localitiesbut are apparently unrelated and have isotopically distinctsignatures.


27/110

TechnicalReport89-05 21The session continued with a structural, geochemical,

and isotopic study of the Kamiskotia area in the westernmostAbitibi Subprovince by T. Barrie. Supported by highprecision U-Pb geochronology, the lithologies were dividedinto older siliceous and tholeiitic rocks, the Kamiskotiagabbro-volcanic suite that apparently formed in a back arc-like extensional regime (2707-2705 Ma), a series ofgranitoids formed in a predominantly compressional regime(2695-2694 Ma), and young alkalic dikes, several of whichpostdate late transpressive deformation in the region (2687Ma). The Nd and Sr isotopic data support the derivationof the Kamiskotia suite from a depleted mantle reservoirwith no evidence for interaction with an enriched crustalcomponent.A compilation of southern Superior Province ENdvalues

and U-Pb ages for mantle-derived magmas, including fivesuites from the Kamiskotia area, revealed that: (1) tholeiitesand komatiites occurred from 2740-2698 Ma, whereas mostalkaline rocks were emplaced from 2700-2670 Ma;(2) tholeiites and komatiites of the southern Abitibi areconsistently isotopically depleted, with end from +2 to +3,whereas those from the western subprovinces are morevariable from +I to +3; and (3)alkaline rocks have endpredominantly from +I to +2, slightly enriched with respectto the southern Abitibi-depleted mantle. Trace elementsignatures for southern Abitibi alkalic rocks are similar toalkalic suites found in Cenozoic destructive plate marginsettings and are consistent with a derivation from a hydrousmantle influenced by a subducted sedimentary component.As the alkalic rocks are relatively isotopically enriched, anolder sedimentary component may have been involved intheir genesis.K. Burke opened the discussion by noting his approval for

field.based structural studies that provide a tectonic frameworkfor follow.up geochemical and isotopic work. J. Wooden askedif it were possible to distinguish between slightly differentsources and contamination to produce the difference. Barrieresponded by agreeing that it is difficult to distinguish betweenthese possibilities for the mantle-derived magmas. However,the coincidence of a transpressive tectonic regime and spatiallyassociated late alkalic magmatism observed across 1200 kmof the southern Superior Province would imply that someuniform process affected the late Kenoran mantle underlyingthe entire region. A subducted sedimentary componentassociated with a subducted oceanic slab would be consistentwith this. Newsom asked if the Destor Porcupine Fault Zonerepresented a sedimentary lithologic break and whether goldmineralization is found along any preferred strata. Barrie saidthat there is some evidence that the major fault zones in thesouthern Abitibi were originally listric normal faults formedin an extensional tectonic regime, such as associated faultscarp(?) breccias and turbidites, h would appear that thesefaults were reactivated during subsequent compression and

transpression and served as mantle.tapping conduits for latealkaline magmatism and as conduits for gold-bearing fluidsfrom unknown sources.N. Rogers mentioned that the alkaline rocks of the central

Italian province were highly potassic and was curious aboutthe meaning and derivation of the more sodic rocks foundin the southern Abitibi. Leman also requested a character.ization of the alkaline rocks. The alkaline rocks discussed rangefrom lamproite dikes to the basanites and trachytes of theTimiskaming Group. Generally, the lamproite dikes have 50%Si02 Mg# - 65-80, NaeO from I% to 5%, KaO from 0.5%to 2%, relatively high LIL, LREE, and transition elements,and relatively low.HFS traces. Timiskaming volcanics havesimilar characteristics but as a suite aremore siliceous(48-61%Si02), more sodic (3.5-6.5% Na20), more potassic (I-7%K20), and lessmagnesian (45-60 Mg#).At this time discussion returned to sanukitology. Leemannoted the chemical similarities between the Archean sanukitoid

suite and the Eocene shoshonitic Challis volcanic rocks inthe western United States. He cautioned that close comparisonswith boninites have strong tectonic implications that may notbe appropriate for Archean rocks of the western SuperiorProvince. G. Hanson commented that considering chemistry,except for the silica content, the suite issimilar to alkali basalts.If partial melting of an alkali basalt sourceoccurs at shallowerdepths under hydrous conditions, one would expect the meltto be on a peritectic between olivine and silica, producinga higher silica content. Many people commented on therelatively common occurrence of alkaline rocks with similarchemistry but with different names, such as latites found inIdaho and Montana and syenodiorites found in China. Jonesrequested that we find a type locality with a shorter name.Hanson responded by saying that he and S. Shirey chose thename for the Archean suite after a literature search to finda name with no connotations.After a coffee break, R. Lambert filled in a vacant time

slot with a short talk on the thermal history of the mantle.First, he reviewed geotherm models for the modern mantledeveloped over the last 20 years, highlighting the differencesbetween subcontinental heat flux, which requires anattached lithospheric keel, and suboceanic mantle, whereconvection efficiently dissipates excess heat. Recent seismictomography has indicated a probable thermal discontinuityat approximately 700 km depth, with possible gradientsof 15/km over about 20-km intervals. These data arecommonly interpreted to represent the thermal boundarylayer between separate convection regimes in the upperand lower mantle. D. Anderson would suggest that onlya solid mantle is present below 700 km and that thereis very little chemical communication across the 700-kmboundary. Considering the higher radiogenic heatproduction at 2600 Ma, it would appear that the gradientwould be more like 25/kin across a 700-km thermal


28/110

22 W_rrkshop_l The ArcheanMantle

discontinuity, and this would intersect most estimates forthe mantle solidus resu!ting in extensive partial melting.Given these conditions this would imply that either amagma ocean existed in the Archean, or that an Archeanthermal boundary layer did not exist. An alternativepromoted by Anderson is that radiogenic heat-producingelements were not present in the Archean mantle and thata stratified mantle could have existed in Archean times.

Discussion was opened by Jones, who pointed out that thosewho believe in the origin of the Moon by fission from theEarth due to a giant impact promote an entirely molten Earthand the extraction of radiogenic elements from the lower mantlewould result as a consequence, consistent with Anderson's views.Lambert retorted that this would also result in the formationof cumulates from a magma ocean, result ing in highly variable_Na, values, which is not observed. Schulze reviewed the scantevidence for Archean diamonds and thermal gradients derivedfrom mineral pair geothermometers in diamond-bearingeclogites and sheared lherzolites, emphasizing that theconstructed Archean deep lithospheric and asthenosphericthermal gradients are tenuous at present.The session continued with Lambert's talk on trace

element and radiogenic isotope constraints on the evolutionof the Stillwater Complex. Previous workers have postulatedthat two parental magmas (U-type: ultramafic and A-type:anorthositic) underwent complex magma chamber processesto account for Stillwater's petrology, chemistry, andmineralization. Using REE isotope dilution for mineralseparates and Sr-Nd-Os isotope systematics on whole rocks,both U-type and A-type magmas were characterized indetail. The U-type magmas show LREE and INdenrichmentin comparison to the A-type magmas, which are more closelyassociated with the J-M reef. The Sm-Nd system was closedand behaved coherently, whereas the Rb-Sr system wasat least partly reset at 1390 Ma. New Re-Os data presentedshows that this system has remained relatively closed,producing a reasonable isochron that agrees with the Sm-Nd isochron for U-type derivatives, with a consistent 187Osenrichment at +23.5 gamma Os units (percent with respectto chondritic Os) for six of seven samples. The trace elementand radiogenic dat

Documents

Workshop on the Archean Mantle