Stabilisation of Archaean lithospheric mantle: A ReOs isotope study of peridotite xenoliths from the Kaapvaal craton

EPSL ELSEVIER

Earth and Planetary Science Letters 134 (1995) 341-357

Stabilisation of Archaean lithospheric mantle: A Re-0s isotope study of peridotite xenoliths from the Kaapvaal craton D.G. Pearson aY *, R.W. Carlson a, S.B. Shirey a, F.R. Boyd b, P.H. Nixon

’

aDepartment of Terrestrial Magnetism, Carnegie Institution of Washington, 5241 Broad Branch Rd, N. W. Washington, DC 20015, USA b Geophysical Laboratory,

Carnegie Institution of Washington, 5251 Broad Branch R4 N. W. Washington, DC 20015, USA ’ Department of Earth Sciences, Leeds Uniuersity, Leeds LS2 9JT, UK

Received 31 May 1994; accepted 30 June 1995

Abstract OS isotopic compositions of lithospheric peridotite xenoliths erupted by kimberlites in the Kaapvaal craton are almost exclusively less radiogenic than estimates of Bulk Earth (18’Os/ 1s80s as low as 0.106) and require long-term evolution in low Re/Os environments. Using Re depletion model ages which assume complete Re removal during formation, the data indicate that cratonic lithosphere stabihsation occurred at, at least, 3.5 Ga, when the lithosphere was over 200 km thick. This thick lithosphere persisted into the Phanerozoic to be sampled by kimberlites. Younger, Proterozoic and Phanerozoic Re depletion ages are interpreted to be largely the result of open system behaviour involving Re addition by metasomatic processes. Some of the younger ages may represent the addition of new lithospheric material during periods of major igneous activity. A mid-Archaean age for the Kaapvaal cratonic mantle concurs with Archaean Re-0s ages found in samples of Siberian and Wyoming cratonic mantle. Both shallow (spine1 facies) and deep (diamond facies) Kaapvaal peridotites have similar ages (3.3-3.5 Ga) suggesting that 150 km of mantle lithosphere may have accumulated very rapidly. OS isotope estimates for the timing of separation and stabilisation of Kaapvaal cratonic mantle overlap the main period of cratonic crust building and stabilisation (3.5-2.7 Ga). A similar overlap between crust and mantle stabilisation is evident for the Siberian craton. Archaean lithospheric mantle is compositionally different to that formed post-Archaean. The Kaapvaal peridotites have very low Fe0 compared to post-Archaean peridotites and show a large spread in Mg/Si. Some samples are anomalously Si-enriched compared with post-Archaean mantle samples. This compositional distinction and the varied Nd-0s isotope systematics are difficult to explain in terms of accepted models involving ancient melt depletion and subsequent metasomatism. Crystal segregation/cumulate processes have been suggested as a mechanism for forming the compositional range observed in Kaapvaal peridotites. This type of process may have occurred during harzburgite crystallisation from high-degree ( > 50%) mantle melts associated with Archaean plume activity. A role for hot mantle plumes in generating the thick lithospheric keels beneath the Kaapvaal and Siberian cratons is supported by the possibility of their rapid formation and their thermal stability with respect to post-Archaean lithosphere. The coincidence of mid-Archaean cratonic mantle differentiation with periods of major crust building and stabilisation on the Kaapvaal and Siberian cratons suggests a link between crust generation and stabilisation and lithospheric mantle

* Corresponding author’s present address: Department of Geological Sciences, Durham University, South Road, Durham, DHl 3LE, UK. 0012-821X/95/%09.50 0 1995 Elsevier Science B.V. All rights reserved SSDf 0012-821X(95)00125-5

342

D.G. Pearson et al. /Earth

and Planetary Science Letters 134 (1995) 341-357

formation in the Archaean. Thermal energy from the plume may have been the impetus for major crust building at the time of lithosphere stabilisation, possibly by underplating of basaltic magmas. Direct involvement of mantle plumes in episodes of major mantle and possibly crust differentiation would imply that modern style plate tectonics may not have been the primary mechanism of planetary differentiation in the early Earth. Archaean ages for peridotites originating up to 200 km deep suggest that the mechanical boundary layer beneath continents is at least this thick.

1. Introduction The initial generation and subsequent stabilisation of the first continents is poorly understood. Significant uncertainty remains regarding the processes that formed robust continental nuclei (termed “cratons”) and the rate at which continental lithosphere was extracted from the Archaean mantle. A clearer understanding of processes occurring in the Archaean mantle would help to constrain the dominant differentiation processes operating in the young Earth and to resolve the question of whether modern-style plate tectonics operated. Current models for Archaean continental growth (e.g., [4,7]) are almost exclusively constrained by evidence from crustal rocks. Archaean Nd model ages for diamond inclusion composites [8] are heavily relied upon for constraints on mantle thermal gradients and lithospheric age, but diamonds are restricted to the basal parts of lithospheric mantle and hence provide limited information on lithosphere formation rates. Moreover, Archaean diamond inclusion Nd model ages are enrichment ages, allowing the possibility of older protoliths. The longevity and stability of lithospheric mantle dictates its ability to generate distinctive isotopic characteristics and its likely role in continental tectonics [9,10]. Hence, constraining the long-term thermo-mechanical stability of continental lithospheric mantle has important implications for magma genesis and tectonics over a large part of Earth history. To assess the age of continental mantle and its rate of formation we require stratigraphic information comparable to that available for the crust. Kimberlite pipes provide a means of sampling the lithospheric mantle throughout much of its depth via their included xenoliths [11,12]. Depths of origin of peridotite xenoliths can be constrained by mineral thermobarometry [13]. Walker et al. [14] first showed that the Re-0s isotope system could reveal ancient ages in Kaapvaal peridotites. Building neering study we present new Re-0s

on that pioisotope data

for peridotites from the Kaapvaal craton and attempt to determine whether any coherent stratigraphy exists in the Kaapvaal cratonic lithospheric mantle and hence constrain its genesis.

2. Depth and composition of Archaean cratonic lithosphere Lithospheric depth can be estimated from mineral equilibria in peridotite xenoliths [11,13]. For example, in southern Africa the boundary between lowtemperature (low-T; equilibration temperatures < 12OO”C),magmaphile element (e.g., Ca, Al, Fe, Ti) depleted peridotites that lie along “Shield geotherms” in P/T space, and high-temperature (high-T: > 12OO”C),relatively fertile peridotites has commonly been taken by petrologists to represent the base of the lithospheric mantle, defined either as a mechanical or thermal boundary layer [11,15]. For cratonic areas, lithospheric depths defined on this basis are N 150-200 km and decrease to N 140 km or less in Proterozoic regions surrounding the cratons [11,13]. Although petrological estimates of the thickness of the mechanical boundary layer are in agreement with some of the seismological estimates (150200 km; e.g., [lo]) suggesting that lithospheric mantle beneath continents is physically and chemically distinct to at least these depths, many seismologists equate velocity anomalies persisting to much greater depths (300-400 km) with greater lithospheric thickness [9,16,17]. Thus there is an inconsistency between petrological and some seismological estimates of the “plate thickness” beneath cratons 1161. If the deeper seismological estimates of lithospheric depth are correct and we accept the derivation of kimberlite magmas from in excess of 300 km [18] we could seek to explain the depths of derivation of most kimberlite-borne peridotites in terms of the physics of kimberlite eruption. Brey et al. [19] have postulated that kimberlite magmas begin to

343

D.G. Pearson et al. /Earth and Planetary Science Letters 134 (I 995) 341-357

exsolve abundant CO, at depths of N 150 km. The transition to a highly fluid-charged magma could dramatically enhance the kimberlite’s ability to entrain and transport xenoliths at this depth and hence the xenolith inventory may not be representative of the entire lithosphere. However, petrological estimates of lithosphere thickness have the appeal of being able to explain the distribution of diamonds in southern Africa [ll] and it may be that deep lithospheric keels are not mechanically stable [lo]. Clearly, the problem is not yet resolved. Compositional studies of continental lithospheric mantle have been heavily biased towards peridotites from the Kaapvaal craton (e.g., 151).It has long been recognised that the low-T Kaapvaal peridotites are compositionally distinct from peridotites, predominantly of spine1 facies, sampled by alkali basalts and from “oceanic” peridotites [5,6,20,21]. This is particularly apparent in terms of the low Fe0 and Mg/Si of Kaapvaal peridotites (Figs. 1 and 2). Low Fe0 and modal garnet contents of Kaapvaal peridotites are mainly responsible for densities being - 1.7% less than fertile mantle [22,23]. This led Jordan (e.g., [9]> to postulate compositionally induced stabilisation of thick roots beneath cratons. Other samples of cratonic lithospheric mantle are available in Siberian kimberlites. There is substantial

Table 1 Rb-Sr and Sm-Nd

isotope compositions

Sample PHN5275

CPX

FRB909 CPX FREt9G9 GT

of peridotite

Rh

Sr

PPm

pm

0.0007

248. I

0.0508 0.0614

89.8 6.485

50

1.0

2.0

3.0

4.0

5.0

compositional overlap between Kaapvaal low-T and Siberian low-T peridotites (Figs. 1 and 2) [24]. Both peridotite suites are considerably more depleted in basaltic components than estimates of primitive mantle or other mantle-derived samples (Figs. 1 and 2). Their mean olivine mg numbers are identical (0.926) and higher than those for oceanic mantle (0.908) [5].

minerals

~'Rh'%r

87Sr/%r f

Sm

Nd

PPm

wm

t4’SdtaNd

raNdlr”‘tNd

&Nd (TJ

f

O.OCOOO8

0.703413

I1

4.155

14.02

0.1193

0.512414

9

-1.8

0.0016 0.0274

0.702679 0.713590

7 12

0.9478 0.9167

3.825 1.335

0.1498 0.4153

0.5 12235 0.514359

11 18

-0.8 0.6

22.29

0.0138

316.7

o.cQo12

0.702257

7

5.026

0.1363

0.512371

11

3.9

o.cO45

0.400

0.0326

0.717801

11

1.002

1.480

0.4095

0.514343

18

1.1

FRFi1350 CPX FRB135OGT FRB 1350 OPX

0.0586 0.0015 0.0229

193.5 0.0666 1.70

O.CQO88 0.0652 0.0399

0.703330 0.705105 0.703782

7 50 18

1.565 0.4135 0.0310

11.59 0.6419 0.1614

0.0816 0.3896 0.1161

0.511809 0.514132 0.512052

11 18 18

1.8 0.7 1.3

PHN5239 PHN5239

CPX GT

0.0583 0.0762

92. 0.904

0.0020 0.2279

0.703518 0.705870

7 8

1.653 1.443

5.996 1.804

0.1666 0.4838

0.512620 0.515107

9 18

4.3 4.8

PHN5273

GT

0.0056

F865 GT

CPX

0.0008

I

6.0

Fig. 1. FeO-MgO diagram for low-T Kaapvaal and Siberian peridotites compared to an average off-craton spine1 peridotite composition [20] and a range of primitive spine1 lherzolites (light shaded field) [63]. The diagram is contoured for temperature and degree of batch melting of primitive mantle at 3 GPa using the method described in [21].

FREi921 GT

FRB921

0.282

0.0575

0.707343

20

1.462

I .a99

0.4687

0.514823

19

1.7

0.392

0.0059

0.763818

20

1.180

1.554

0.4618

0.512952

lb

2.1

All samples from Premier except F865 GT (Finsch). Errors are 2u 2).

65

within run.

lNdtrj values calculated for kimberlite eruption ages (Table

344

D.G. Pearson et al. /Earth

and Planetary Science Letters 134 (1995) 341-357

Table 2 Re-0s isotope compositions of Kaapvaal peridotites Snmple

Locality

Kimbcrlitc Rc Ma PPb

Age

Or ‘s’R&%s PP”

‘*Os/‘ssOs m

‘ff’O.d’%k,

ym

TRD TMa Mg#

f

Ca

0.11820 0.11339 0.10808 0.11326 0.11588 0.11166

10 10 21 33 35 6

0.1171 -7.8 1.6 0.1132 -10.9 2.2 0.1077 -15.2 3.0 0.1099 -8.3 2.1 0.1143 0.1105

4.7 -7.8

-8.7 -14.2 -13.4 -0.3 -5.7 -14.0 4.3 -15.9 -13.2 -16.8 -7.2 -9.9

2.7 1.2 2.2 2.8 0.9 3.1 2.6 3.3 2.5 2.0

-5.0 -12.1 -2.3 -13.0 -2.7 -10.2

2.1 2.4 0.5 3.5 1.7 2.1

Ga

T’C

01

P GPa

Law-T spine1 lhcrzolites and hatzburgites ml3979#

FRBlOOl# PHN2600 Pm5266 PHN5254 PHN5275

Jagenfontein Jagersfontein Liqhobong Premier Premier Premier

86 86 90 1179 1179 1179

0.133 0.116 0.035 0.118 0.106 0.103

0.804 3.57 0.625 3.25 6.13

0.796 0.156 0.267 0.174

8.23

0.083 0.060

1.73 1.38 2.66 2.66 3.39 2.61 0.019 1.51 9.20 1.41 3.17 4.24

0.093 0.297 0.155 0.095 0.038 0.145 10.3 0.047 0.038 0.323 0.048 0.153

0.11596 0.10937 0.11012 0.12139 0.11370 0.10937 0.13668 0.10690

8 10 20 20 7 10 41 IO

0.11030 0.10614 0.11210 0.11465

17 20 5 30

0.1158 0.1088 0.1098 0.1195 0.1130 0.1092 0.1215 0.1068 0.1102 0.1057 0.1112 0.1 I44

I.69

0.144

5.94 1.26 1.54 5.33 2.45

0.06 1.09 0.31 I 0.057 2.02

0.11667 0.11175 0.12572 0.11040 0.11775 0.11687

I4 10 21 10 16 I7

0.1139 0.1117 0.1241 0.1043 0.1166 0.1140

2.0 2.6

- 0.935 3.5 0.941 8.5 0.955 3.8 2.2 2.8

0.921 0.921

637 571

Low-T gamer iherzolites and harzburgites

+F556 *F865 ‘F866 FRB921 FRB1350 PHN1569 *BD2125 PHN2302 PHN2823 PHN2825

Fiih

Fiih Fiirh Premier Premier ThabaPutsoa Mothae Liqhobong Liqhobong Liqhobong

119 119 I19 1179 1179 90 90 90 90 90

PHN5273 PHN5595

Premier Jngcrsfonlein

1179 86

0.034 0.086 0.081 0.053 0.026 0.079 0.041 0.015 0.073 0.095 0.032 0.135

I179 90 90 1179 1179 86

0.0s I 0.074 0.285 0.10 0.06 1.03

2.3 10.2 4.1 1.2 2.3 4.3 0.1

0.911 1116 0.929 1160# 0.929 1103# 1055 0.917 701 0.929 949 1032 3.5 0.928 994

2.9 0.93 I 15.6 0.956 2.7 0.922 3.2 0.932

5.9 6.2 5.8 4.2 2.1 3.8 4.7 4.3

627#

High-T gnmer lherzolites IXD909 PHN1596 PHN1611 PHN5239 PHN5267 K7-318

Prcmicr fhabn Putsoa Thaba Putsoa Premier Premier Jagersfontein

2.6 2.8 0.906 0.882 11.1 0.917 1.7 0.913 0.916

1395 1563 1472 1398 1375

6.2 6.9 6.5 6.7 6.6

* = diamond-bearing sample. T and P estimates using FE%/MC74, #OW79/MC74 (see 1131for details). FRB979 and 1001 are probably garnet facies on the basis of the Al content of their enstatite and the Cr content of their spinel. Kimberlite ages are from [37]. yOsi values calculated at eruption time of kimberlite (see text for parameters). Errors are 2a within run.

3. Trace element and isotopic characteristics Previous studies of Kaapvaal peridotites have shown that the low-T peridotites are enriched in light rare earth elements (LREE) relative to chondritic mantle (e.g., [Z]). Some mineral Sr-Nd isotope compositions are more “enriched” than Bulk Earth and indicate long-term isolation from the convecting mantle [14,26,27] although those from Premier are closer to Bulk Earth at the time of kimberlite eruption (Table 1). High-temperature Kaapvaal peridotites are not as enriched in LREE and have Sr-Nd isotopic compositions similar to depleted mantle ([14,27]; Table 1). Re-0s isotope data for peridotites from the Kaapvaal (Table 2) [14] together with those from the

Wyoming and Siberian cratons [13] show that cratonic peridotites in general have less radiogenic ls70s/ ‘*‘OS (negative yes ‘1 than average chondrites at the time of kimberlite eruption (Fig. 3). OS isotopic compositions of cratonic peridotites are gen-

1 yes is a measure of the difference, in percent, of the OS isotopic composition of a sample at a specified time relative to a reference mantle composition and is given by: -yes (Tj = [(1870s/ 1880s~,,,p,,~~~/~‘s70s/ 1880s),,,,,s(T) - 11 X 100. There is some doubt about the exact 18’Os/ ‘**OS of the presentday mantle and it may vary from 0.125 to 0.132 [14,28-301 but further back in Earth history this value is more tightly constrained by the precisely determined iron-meteorite initial ratio [31]. Here we use a present-day 1s70s/1880s of 0.12757 taken from the mean for carbonaceous chondrites [14].

34s

D.G. Pearson et al. /Earth and Planetary Science Letters 134 (1995) 341-357 1.6

,,,,,“,.,‘~

10

,,,,,,,,‘,,, Kaapvaal peridotites

8r 0

p

2

6

;

4

B

I"1

0.8

2 0

0.02

0.04

0.06

AI/Si

0.08

0.1

0.12

0

wt

Fig. 2. Mg/Si vs. Al/Si wt ratios of Siberian and Kaapvaal low-T peridotites compared to the range shown by primitive spine1 lherzolites erupted off craton. Siberian and Kaapvaal peridotite data are for large (300- > 500 g) fresh samples ([36]; Boyd, unpubl. data; other data from [63]).

erally much less radiogenic than abyssal and orogenie peridotites and ocean island basaltic magmas thought to come from the convecting mantle [28-301 (Fig. 3). Of terrestrial reservoirs capable of involvement in magma genesis, the lithospheric mantle appears to be uniquely unradiogenic in its OS isotopic composition. Low-T Kaapvaal peridotites have generally less radiogenic rg70s/ 1880s than the high-T peridotites but there is much overlap and some high-T samples have very low initial OS isotope compositions, comparable to the low-T peridotites (Fig. 3). Nd isotopic compositions of diopside or garnet vs. whole rock OS isotopic compositions show a much larger range than oceanic mantle (Fig. 4a). There is broad overlap of samples from Siberia and Kaapvaal, those with the least radiogenic initial Nd isotopic compositions have some of the least radiogenic OS isotope compositions. The data clearly cannot be reconciled with any single-stage history involving simple melt depletion (Fig. 4b). Unradiogenic initial OS and Nd isotopic compositions are indicative of ancient Re depletion and ancient LREE enrichment. These end-member isotopic characteristics could be generated by melt extraction followed by incompatible element enrichment. Preservation of very unradiogenic OS in samples with unradiogenic Nd requires that the event(s) that caused time-integrated unradiogenic Nd isotopic

1

I

I Siberian peridotites

4

i ij z” 2

0

I

Wvomine

6

I

Oceanic mantle peridotites

D e4 a 5 P 2

I

I

0 -20

-15

-10

-5

0

5

YOS Fig. 3. Histograms of initial OS isotopic composition at the kimberlite eruption time for Siberian [l] and Kaapvaal kimberlites compared to mantle xenoliths from Wyoming [3] and oceanic peridotites [28,64,65] (-yOsCTj= [(‘870s/ ‘880s),,,plc(T) /P’os/ ‘880s) ,,,&)11X 100. Kaapvaal data includes samples from [14].

346

D.G. Pearson et al. /Earth

and Planetary Science Letters 134 (1995) 341-357

compositions did not impart sufficient Re to significantly disturb the OS isotope characteristics. Minor (l-2%) interaction of melt depleted peridotite with a carbonatitic melt (v. high Nd, high Re, very low OS [l]), modeled as simple mixing, is capable of producing very large changes in Nd isotopic composition for minor changes in OS isotopic composition, especially if the interaction occurred in the Archaean (Fig. 4b). However, few of the data in Fig. 4b plot close to the present-day compositions expected for ancient melt residues re-enriched by small volumes of carbonatite at various times in the past (the end 20

I

I

I

1

r_________---_______-__-___~----

I

Oceanic

n M=

0

mantle )_~--~~~~~~------~~--~--~-----~~

kW B

.s 0

‘Nd -20

4. Evaluation of the residual mantle hypothesis

a

0

n 00

The depletion of low-T Kaapvaal peridotites in magmaphile constituents, most notably FeO, and their enrichment in MgO (Fig. 1) has led to the popular conviction that they represent the residues of Siberia

0 -60 -20

I

I

B

-10

0

10

0

-

20

'Ios 60

1

I

I

‘Nd

-60

I

-25

-20

points of the arrows in Fig. 4b, arrows are not mixing lines). The 10% or more carbonatite required to produce some of the Nd-0s isotopic compositions in Fig. 4 would result in marked mineralogical evidence of interaction. Combining variable degrees of melt extraction with metasomatism to explain some of the peridotite Nd-0s data raises problems because of the high, fairly uniform peridotite mg numbers (Fig. 5). Melt removal at greatly differing times also runs into problems of the compositional similarities of xenoliths within and between kimberlites. In detail, some samples may be interpreted in terms of melt extraction at 3.5 Ga followed by addition of u 10% or more basalt at, e.g., 3,2.5 or 1.5 Ga (Figs. 4 and 5). However, the large subhorizontal spread on Fig. 5 reveals that such models can only explain some of the data.

l -15

1

-10

yes

-5

I

0

5

10

Fig. 4. (a) q,,, vs. yoS for Kaapvaal and Siberian peridotites corrected to kimberlite eruption times (data from Table 2 and [1,14]). OS data on whole-rocks, Nd data on acid-washed mineral separates. The range for oceanic mantle is taken from [28-301. (b) Modeling the evolution of 0s and Nd isotopic compositions of depleted mantle lithosphere metasomatised by carbonatite and basalt at various times in the past. Diamonds are Kaapvaal and Siberian peridot&es. The depleted mantle is modeled as separating from the convecting mantle at 3.5 Ga. (solid evolution curve, no arrow). Arrows at the end of solid lines represent the present-day isotopic composition produced by interaction between various proportions of carbonatite-like metasomatic agent (percent labeled next to arrow) with a depleted peridotite at 3 Ga. The lines are nof mixing curves. Arrows at the end of dashed lines point to the present-day isotopic composition produced by metasomatism of peridotite by a basaltic magma at 3 Ga. Metasomatism is modeled as simple mixing. The locus of arrow points defines the mixing array as observed today. Dashed lines with arrows are basalt-peridotite mixtures at 3 Ga, arrow points to present-day composition of the mixture. Late-stage (present-day) interaction between metasomatised peridotite and a host Group I kimberlite is modeled as the solid curve labeled Kimberlite, proportions of host kimberlite given in percent. Re and OS concentrations for kimberlite and carbonatite from [l].

D.G. Pearson et al. /Earth and Planetary Science Letters 134 (1995) 341-357

extraction of ultramafic-komatiitic melt from fertile mantle. However, the low Mg/Si of some Kaapvaal peridotites, manifest as high modal enstatite (up to 46 wt%) is problematical. Boyd [5] proposed an origin by high-pressure partial melt extraction, but recognised a serious mass balance problem in forming Si-enriched peridotites by basaltic or komatiitic melt extraction. High-pressure melting experiments have shown that orthopyroxene stability on the solidus during melt extraction could account for up to N 31 wt% enstatite, the mean for Kaapvaal low-T peridotites, but not the entire spectrum of compositions [6,32,33]. Reaction of peridotite with silicate melt can generate Si-enrichment in peridotites [34] and hence anomalous enstatite contents. However, we concur with Herzberg [6] that this process is difficult to envisage on a lithosphere-wide scale and seems unlikely to produce residues with significantly higher average mg number (0.926) than other “residual” peridotites. While recent interaction with 10% or more basalt could explain some of the Nd-0s isotope variation (Fig. 4b and Fig. 51, the lack of correlation between ~0s and mg number in

347

most Kaapvaal samples is not consistent with such an explanation. In addition, we do not observe a correlation between enstatite content and incompatible element enrichment or high-field-strength-element depletion predicted by this type of interaction. Walker et al. [14] found a general correlation between peridotite mg number and OS isotopic composition. Better correlation between these parameters has been observed in peridotites from the Ronda massif [35] which show other geochemical evidence of being melting residues. The more extensive analyses of Kaapvaal peridotites reported here do not show good correlations between indices of melt depletion and OS isotopic composition (Fig. 51, or Re and OS abundances. Furthermore, the overall range of OS concentrations for the suite (0.1 to > 10 ppb) greatly exceeds that expected for residues of extensive melt removal. Discussion in the previous section indicated the difficulty in explaining the incompatible element (Nd)-compatible element (OS) isotope systematics of the Kaapvaal (and Siberian) peridotites by simple 2 or even 3 stage models involving melt depletion

0.96,....,....,....,....,....

. . 3.5 . Ga

n

2.5 Ga

1.5

Ga

Melting event

Fertile

--*2-.c

0.88

‘.\

*-.

Mantle

: . . . \ . I . . . . 1 . . . . #..9.,“‘l_..

Fig. 5. Olivine mg number (Mg #) vs. yes for Kaapvaal and Siberian peridotites. n = low-T peridotites; 0 = high-T samples. Solid curves show trends expected for basaltic melt extraction at various times in the past. Dotted lines are simple mixing between a residual peridotite with mg number = 94.2, ‘yes = - 18 and basaltic melt at 0 and 3 Ga. Fractions of the melt component are indicated.

348

D.G. Pearson et al. /Earth

and Planetary Science Letters 134 (1995) 341-357

and re-enrichment. The wide range of modal enstatite in low-T Kaapvaal peridotites led Boyd [.5,36] to suggest that crystal segregation or even metamorphic differentiation may be involved in their genesis. On the basis of recent high-pressure phase equilibria Herzberg [6] has proposed that many Kaapvaal low-T peridotites may be “cumulates” from highSi magmas formed by extensive melting ( > 50%) of normal mantle peridotite, or alternatively, they crystallised from a high-Si source such as chondrites during the Hadean (4.5-4.0 Ga). We will now try to better constrain the origin of the Kaapvaal peridotites in the light of age information revealed by OS isotopes.

5. Time scales of mantle lithosphere differentiation and correspondence with crustal ages. The Rb-Sr and Sm-Nd isotope systems first gave indications that cratonic mantle lithosphere is ancient [8,26,27] but although these systems are useful for assessing the metasomatic history of the lithospheric mantle, they do not appear to record the timing of lithospheric differentiation. Due to high ambient temperatures, the internal isochron approach for peridotites generally results in determination, to varying degrees of accuracy, of the emplacement age of the kimberlite pipe. For example, a garnet lherzolite (FRB1350) from the Premier pipe with a very low equilibration temperature, estimated with the diopside solvus to be < 800°C (Table 2) gives an Sm-Nd enstatite-diopside-garnet internal mineral isochron of 1150 f 41 Ma (2a errors; Fig. 6). This age is most likely a cooling age and is in good agreement with the previous Rb-Sr mica age determination for this kimberlite (1179 f 36 Ma; [37]). In contrast, the unradiogenic OS isotope signature of this peridotite indicates that it is at least 2.2 Gyr old. Thus, OS isotope signatures in the Kaapvaal peridotites are indicative of ancient Re depletion and can be used to constrain the time since magmatic differentiation. Interpreting the age information provided by OS isotopes requires some caution. Elemental and Sr-Nd isotopic data document evidence for widespread introduction of incompatible elements into kimberliteborne xenoliths during eruption [1,14,27]. Lack of correlation between ls7Re/ “‘0s and indices of melt depletion may indicate that peridotite Re contents are

Osll"'.""'.. 0

0.1

0.2

.""I 0.3

0.4

‘47Sm; 44Nd Fig. 6. Sm-Nd internal mineral isochron for FRB1350, a Premier low-T peridotite. Errors are 2~.

not a function of the magmatic event that caused their differentiation and may have been elevated by infiltration of either the host kimberlite magma or previous metasomatic events. Alternatively, Re or even OS loss may have occurred due to sulphide breakdown during eruption of the xenoliths [38]. There is no systematic correlation of highly incompatible elements such as Ba with Re to clearly indicate enrichment of Re, but a variety of differing metasomatic agents, together with chromatographic effects may not produce simple elemental systematits. In reality it is possible that both metasomatism and sulphide breakdown may have operated such that the measured Re/Os of the peridotites cannot be reliably used for conventional model age calculations. For example, several of the peridotite Re/Os model ages calculated using measured 18’Re/ 1880s (TM, in Table 2) either give geologically unreasonable ages or do not intersect the Bulk Earth evolution line at all. Walker et al. [14] reasoned that the highly refractory compositions of Kaapvaal peridotites could have led to complete removal of Re during formation. Using the assumption that Re/Os is zero it is possible to calculate a model age that is a minimum estimate of the age of the rock. Such a “Re depletion model age” represents the minimum period of time the rock has existed in a low Re/Os environment. If some Re remained after an initial melting event, or if the peridotites formed by a process other than melt removal, that retains significant Re, then the actual age will always be older than this minimum model age. Thus the term depletion merely

349

D.G. Pearson et al. /Earth and Planetary Science Letters 134 (1995) 341-357

reflects the fact that the rocks have evolved with lower Re/Os than Bulk Earth. If most kimberliteborne xenoliths have experienced some Re addition from the kimberlite we can correct for radiogenic OS generated since eruption. For Cretaceous southern African samples this makes little difference to the minimum model age but the correction is much more important for peridotites erupted by the Proterozoic Premier kimberlite. For this reason two of the Premier samples were replicated by Carius tube dissolution to ensure that the Re/Os values obtained from reducing acid dissolutions were not affected by spike-sample equilibration problems or incomplete dissolution (see the Appendix). Eruption age corrected Re depletion ages (T,, Table 2) for the two Premier Carius tube replicates agree well within the overall uncertainty ( m 200-300 Myr) with the reducing acid dissolution values (Table 3). Addition of 1% Group I kimberlite [1,14] is sufficient to raise the Re content of a Re-free xenolith to N lo-20 ppt. Given the ease with which the Re content of a peridotite may be enhanced, especially compared to its OS content, it is possible that only samples with the lowest Re contents could approximate their original Re contents. Samples FRB 1350, PHN5273 and PHN 2302 have the lowest Re con-

tents (15-30 ppt) and with their average OS contents (N 1.5-3 ppb) they should provide an indication of the effect of some indigenous Re on model ages. Re/Os model ages of these samples are within 200-400 Myr of their Re depletion (minimum) model ages. Even in these cases, anomalously high Ba (163 ppm) in FRB1350 indicates that significant incompatible element enrichment has taken place which may have enhanced Re. Re contents as low as 3-6 ppt have been recorded in apparently un-metasomatised peridotites from San Carlos [39] and Namibia (Pearson, unpubl. data) and suggest that even Re contents of 15 ppt in the Kaapvaal xenoliths may be dominated by introduced Re. If this is so then the oldest Re depletion ages (Tut, in Table 2) will be close to the true age of their formation, e.g., within 200-400 Myr. This is illustrated by PHN 2302. which has very low Re (15 ppt) and gives Re depletion and Re-0s model ages of 3.1 and 3.5 Ga, respectively. In contrast, incomplete Re removal during formation, or Re addition during metasomatism causes the Re depletion ages to be younger than the true differentiation age. It is also possible that Re loss linked to sulphide breakdown during eruption [38] may reduce Re/Os and cause underestimation of the TMa model age. Because of this uncertainty

Table 3 Replicate Re-0s analyses of peridotites Sample

Technique

Kimberlite Age, Ma

Re wb

0s wb

‘“Re1’880s

UV564/89

AD AD CT

350 350 350

0.469 0.473 0.459

1.58 1.40 1.52

1.44 1.63 1.46

0.12352 0.12356 0.12325

PHN2600

AD CT

90 90

0.035 0.044

0.625 0.679

0.267 0.312

PHN5239

AD CT

1076 1076

0.100 0.065

1.54 I .32

PHN5275

AD CT

1076 1076

0.103 0.085

PHN2825

AD OAD

90 90

BD2125

E

;oo

‘~70sl’s80s

yOs,

TRDGa

10 9 11

0.1153 0.1142 0.1149

-8.0 -8.9 -8.3

1.9 2.0 1.9

0.10808 0.10877

21 15

0.1077 0.1083

-15.2 -14.7

3.0 2.9

0.312 0.237

0.11040 0.11000

10 30

0.1043 0.1054

-13.0 -12.1

3.5 3.3

8.23 6.90

0.06 0.06

0.11166 0.11149

6 9

0.1106 0.1104

-8.2 -8.3

2.6 2.6

0.095

1.41 1.34

0.323

0.10614

20

0.1057

-16.8 _

3.3 _

0.041 0.038

0.019 0.021

10.3 8.72

0.13668 0.13625

41 20

0.1215 0.1234

-4.3 -2.9

0.9 0.6

reducing acid dissolution; CT= Carius Tube dissolution (see [62] for details); OAD = oxidising acid dissolution with chromic acid. Re determination and OS isotope measurement for OAD compromised by high Re blank in chromic acid, Errors are 2cr within run.

AD =

350

D.G. Pearson et al. /Earth

and Planetary Science Letters 134 (1995) 341-357

we will use the Re depletion model ages in the. following discussions because they do not rely on the Re (or OS) abundances. Re depletion model ages for Kaapvaal peridotites range widely, from Recent to 3.5 Ga (Table 2; Fig. 7). Since the Re depletion (T,) ages provide only minimum estimates of the formation ages of the peridotites, the Proterozoic ages for some xenoliths could reflect continued addition of material to the lithospheric mantle, for example during voluminous igneous activity associated with the Ventersdorp sequence between 2.6 and 2.8 Ga [40]. Alternatively, younger ages may simply reflect imperfect adherence of samples to the simplistic assumptions involved in the Tm calculation, i.e., that Re was quantitatively extracted from the peridotite in a single event and that no subsequent metasomatism has occurred. Because the Re depletion ages are minima we cannot rule out an origin for the Kaapvaal peridotites in the Hadean (> 4 Ga) [6]. We can try to constrain the possible effect of Re addition by metasomatism, or Re loss due to sulphide breakdown during eruption by using Re abundances from a suite of Ronda peridotites that are thought to be relatively undisturbed in terms of their Re abundances 1351. Using Re contents from the most depleted Ronda peridotites (which are not generally as major element depleted as Kaapvaal low-T peridotites) gives conventional Re-0s model ages (TM,) that are still Archaean, the oldest being 3.9 Ga, with most in the range 2.9-3.6 Ga. Thus we see no strong support for Hadean ages. The most significant features of the Kaapvaal Re depletion age distributions is that many are in the range 2.7-3.5 Ga and the oldest model ages for both the Kaapvaal and Siberian cratons are 3.2-3.5 Ga (Fig. 7). Similar maxima in the Re depletion ages from Kaapvaal and Siberian cratonic peridotites could support the view that these ages approximate the formation age of the mantle lithosphere in these regions. Furthermore, the oldest Re depletion ages of the Siberian peridotites are coincident with Re-0s model ages of 2.7-3.4 Ga in a suite of Siberian eclogite xenoliths [2]. Combination of Re depletion ages with mineral thermobarometry illustrates the age variation with depth through the Kaapvaal lithosphere (Fig. 8) and

from this we can try to constrain the rate of lithospheric mantle differentiation. For the Kaapvaal lithosphere, two samples, PHN2825, a garnet-spine1 peridotite originating from close to the garnet-spine1 transition, and PHN5239, a high-temperature garnet peridotite sampled from the lower Kaapvaal lithosphere at N 210 km, give Re depletion ages of 3.3-3.5 Ga. PHN2825 is from Liqhobong, northern Lesotho. A spine1 peridotite from the same pipe, probably originating from < 60 km deep, has a Re depletion age of 2.9 Ga, whereas garnet lherzolite PHN2302, also from Liqhobong, derived from a depth close to 140 km, has a Re depletion age of 3.1

1-1

Kaapvaal Craton

I

I Lower crustal xenoliths I-

1

Witwaterwand Osmkfdiums m

Greenstones -

7

Ancient Gneks Complex m 10

i

0

0.5

1

1.5

2

2.5

3

3.5

4

0

0.5

1

1.5

2

2.5

3

3.5

4

Model Age, Peridotites

Ga

Fig. 7. Histograms of Re depletion model ages (corrected for Re to the time of kimberlite eruption) for peridotites from the Kaapvaal and Siberian cratons compared to the age of major crust building events in each craton. Model ages of Witwatersrand osmiridiums shown on Kaapvaal histogram may document Archaean mantle magmatic products being eroded into the Witwatersrand basin.

351

D.G. Pearson et al. /Earth and Planetary Science Letters 134 (1995) 341-357

Ga. From these data there is no evidence for the shallowest mantle being significantly older than the deep Kaapvaal lithospheric mantle (Fig. 8). The oldest Re depletion ages throughout 150 km of Kaapvaal lithospheric mantle are between 3.0 and 3.5 Ga and could indicate that the entire lithospheric section represented by the xenoliths may be of very similar age, i.e., mid-Archaean. Similar conclusions may be drawn about the time scale of formation of the Siberian lithospheric mantle. Mid-Archaean isochron and model ages for a suite of diamondiferous eclogites and 3.0-3.2 Ga Re depletion ages for two diamondiferous dunite/harzburgites from the Udachnaya kimberlite indicate that the lithosphere beneath the Siberian craton had reached m 150 km thick by the mid-Archaean [2].

data place strong constraints on their genesis. The high-T peridotites show a range in OS isotopic compositions, comparable to low-T peridotites. High-T peridotites have very unradiogenic 1870s/ 1680s (yes - 12.1 to - 12.5) giving Re depletion ages of 2.4-3.5 Ga. This and previously published data [14] clearly indicate isolation of the high-T peridotites from the convecting mantle since the Proterozoic-Archaean and can not support the view that they represent samples of the underlying asthenosphere [11,25,27]. The high-T peridotites may come from the borders of a magmatic intrusion at the base of the lithosphere that caused heating, deformation and melt infiltration of what were previously low-T peridotites [41]. OS and Nd isotope systematics (Figs. 4 and 51, together with the more fertile nature of the high-T peridotites, allow the possibility that they represent the product of _ lo-20% basalt infiltration into low-T peridotites in the Archaean. Interaction of basaltic melt with peridotite can dissolve pyroxene components leaving a more olivine-rich residue [34], e.g., the high-T peridotites [36]. However, high-T Kaapvaal peridotites have higher modal diopside and garnet than low-T peridotites [36] so

6. The relationship between high-T and low-T peridotites A detailed evaluation of the relationship between high-T and low-T cratonic peridotites is beyond the scope of this paper but several aspects of the Re-0s

Proterozoic

Archaean Kaapvaal Craton J

West

FL

F

kimberlites

Proterozoic P

NL

0.09 0.12 eruption age 1.10

250

km

................ High-T

TBL Fig. 8. Schematic cross-section through the Kaapvaal craton and surrounding areas of Proterozoic crust, illustrating the variation of Re-depletion ages (Gal of peridotite xenoliths with depth (determined from mineral thermobarometry) for different kimberlites. J = Jagersfontein; P = Premier; F = Finsch; NL = North Lesotho; FL = Farm Lowrensia; EG = East Griqualand. Kimberlite pipe emplacement ages are given in Ga. The dashed boundary illustrates the division between low-T, dominantly coarse-grained peridotites, and high-T, dominantly sheared/porpyroclastic peridotites. The transition from lithosphere, defined as a mechanical boundary layer (MBL) to more transient mantle belonging to the thermal boundary layer (TBL) is taken to be below the depth of high-T xenoliths yielding ancient ages.

352

D.G. Pearson et al. /Earth

and Planetary Science Letters 134 (I 995) 341-357

this mechanism of olivine enrichment seems implausible. It is possible that the high-T peridotites represent oceanic lithosphere subducted beneath a pre-existing lithospheric keel in the Archaean [14,36]. However, the high equilibration temperatures and marked mineral chemical zonation in many high-T peridotites indicate recent interaction with a melt component that may have significantly altered the bulk composition of the majority of these rocks (e.g., [42]) and thus obscured their past history. Although our understanding of the genesis of high-T peridotites is incomplete, these rocks are isotopically distinct from Phanerozoic oceanic lithosphere and appear to have resided for billions of years in the lithospheric mantle.

7. Significance of the Premier peridotite ages Data from Premier peridotites (Table 1) with a variety of equilibration temperatures yield internal Sm-Nd mineral isochrons co-incident with the accepted age of the kimberlite eruption (e.g., Fig. 6). Initial eNd values for minerals from the Premier low-T peridotites (Table 1) are close to chondritic and yield variable, 0.6-2 Ga Sm-Nd model ages. We have shown above that the Re depletion model ages are more likely to give an indication of the formation age of the peridotites. One high-T peridotite (PHN5239) evidently formed in the midArchaean, concurring with Archaean ages for the low-T Kaapvaal samples. Although the remainder of the Premier samples show a broad range of Re depletion ages, from 1.2 to 2.7 Ga, 4 samples from this study and two analysed by Walker et al. [14] give Re depletion ages in the range 1.7-2.2 Ga, with a mean of 2 Ga, close to the age of the Bushveld intrusion at 2.05 Ga [40]. The Premier kimberlite erupts through the outcrop ring of the Bushveld intrusion. The magmatic activity associated with formation of the Bushveld Intrusion must have been a major magmatic event affecting the central part of the Kaapvaal lithosphere and may have led to widespread disturbance of lithospheric OS isotope systematics which now give Proterozoic ages. Considering the large magma volumes that must have been involved in the Bushveld event it is also possi-

ble that new lithospheric material could have been added at this time. The possibility of new lithosphere being generated in the area sampled by the Premier kimberlite has also been suggested from diamond inclusion ages [43]. A problem with the Premier peridotites representing new lithospheric material added in the Proterozoic is that they have similar bulk compositional characteristics to other, Archaean Kaapvaal peridotites yet distinct from Proterozoic peridotites such as those erupted by the Gibeon kimberlites of Namibia [44].

8. Mantle-crust relationships and the formation of the Kaapvaal cratonic lithospheric mantle Initiation of major crust building (and crust survival) for the Kaapvaal craton occurred at around 3.7 Ga and continued until w 2.7 Ga [4,45]. The Kaapvaal Shield provided a stable platform for the deposition of sedimentary rocks by _ 3.0 Ga [4]. The record of crustal growth on the Siberian craton is less detailed but ages in the range 3.5-3.0 Ga have been documented for the Anabar shield [46] 3.2 Ga tonalitic gneisses and Archaean greenstones are found in the Aldan shield [47,48]. The oldest Re depletion ages for the Kaapvaal and Siberian peridotites lie within these periods of major crust building and stabilisation (Fig. 7). In addition, 3.3 Ga Pb-Pb mineral and whole-rock isochrons have been reported recently for suites of lower-crustal xenoliths from Kaapvaal kimberlites [49]. The general correspondence between Re depletion model ages (minimum estimates) in the peridotites and largescale stabilisation of both upper and lower crust for the Kaapvaal craton, together with a similar correspondence between crust and mantle stabilisation ages for the Siberian craton, may indicate a much stronger physical link between crust and mantle processes during continent formation in the Archaean than has been documented previously. Alternatively, de Wit and Hart [50] have suggested that a lithospheric age of 3.2-3.3 Ga recorded in diamond inclusions simply records a major tectonic/metamorphic event on the Kaapvaal craton. The peridotite Re depletion age calculations are independent of parent/daughter ratios and record minimum esti-

f).G. Pearson et al. / Earth and Planetary Science Letters 134 (1995) 341-357

mates of melt removal. Tectono-metamorphic events are therefore unlikely to re-set these ages unless accompanied by significant volumes of magmatism. Estimates of lithospheric thicknesses from xenoliths are generally greater beneath Archaean cratons than the surrounding post-Archaean terrains [13]. Formation of over 200 km of Archaean lithospheric mantle simply by cooling of existing shallow mantle is unlikely considering the marked compositional contrast of cratonic lithosphere compared to fertile mantle (Figs. 1 and 2) and the long time scale required to thermally stabilise lithosphere to this depth [9]. In addition, anomalous thickening of lithospheric mantle by cooling requires a thicker crust in order to gravitationally compensate for the increase in density of the thermal boundary layer. In fact, crustai thicknesses on the Kaapvaai and other cratons are in general thinner than in the surrounding postArchaean terrains [51]. Therefore, the only likely stabilising mechanism for the thick keels beneath continents is compositional contrast [9]. Kaapvaal peridotites are compositionally distinct from peridotites erupted through post-Archaean crust (e.g., [5,6,20,21]). Work on Siberian peridotites also indicates that they are compositionally distinct from post-Archaean lithospheric mantle [24]. Hence, we require a mechanism for achieving sufficient compositional contrast to gravitationally stabilise lithospheric mantle beneath cratons that will also produce matlrerllv J diffment rneridntitec _____. *__I nf __ ___-_.._-_ UILLIIIII.

mmnncitinn ““.y,“““‘“”

tn .V thnw .II”“I

of post-Archaean lithosphere. From petrological and geochemical considerations it is evident that the Kaapvaal peridotites are not simply conjugate residues of crustal extraction during the midArchaean [5,6]. The Si-rich nature of many Kaapvaal low-T peridotites appears to preclude them being meiting residues from basaitic or komatiitic magma extraction [6,32] unless they have subsequently undergone metamorphic differentiation, or metasomatism [51. Melting of hydrous mantle to give orthopyroxene-rich residues is an alternative [50] although the experimental basis for this is not established. Because the Re depletion ages are minima we cannot rule out an origin for the Kaapvaal peridotites in the Hadean (> 4 Ga) [6]. However, overlap of peridotite Re depletion ages and the time span of crust building on the Kaapvaal craton (2.7-3.5 Ga) together with a similar relationship for the Siberian

353

craton [1,2] supports the case for an Archaean origin. If melting is not capable of producing the large range of modal enstatite in the Kaapvaal peridotites they may have achieved this by cumulate enrichment or metamorphic differentiation [5,6,36]. Formation of cratonic peridotites as “cumulates” from relatively high-Si melts produced by > 50% melting of mantle peridotite, is possible from phase equilibria constraints [6]. To create melt fractions > 50% it is likely that the melting process commenced at pressures in excess of 20 GPa and involved large-scale plume activity [6]. A major thermal and melt flux, associated with Archaean plume activity could promote rapid formation of lithosphere extending over 200 km deep and several 1000 km*. A hotter Archaean mantle may have provided conditions conducive to the generation of iocaiised large melt fractions as rising plumes would encounter the peridotite solidus at greater depths where the solidus and liquidus are closer and temperatures would remain super-solidus over longer depth ranges. Archaean plumes therefore may have been capable of producing the large (> 50%) melt fractions required for high-pressure crystallisation of harzburgitic “cumulates” similar in composition to the Kaapvaal low-T peridotites. If local melt fractions in excess of 50% require melting to commence at pressures greater than 20 GPa, olivine will be less dense than the equilibrium melt [52], hence, both residues and cum,,l.,tpr LlllllULVU

frnm Ll”lll

thnw maltc r&l1 ha ..na~&.c.l.r t1,u..iu I11UIb.J VT111 UC p”olrl”uy

l...n..nmt “u”yaur.

Given the complex dynamical situation and temperature gradients likely to occur in deeply derived “superplumes” it is conceivable that the compositionally diverse Kaapvaal lithosphere is a composite of high-pressure cumulates and melting residues. Thus Archaean lithospheric mantle may have formed by the coaiescence of buoyant cumulates and residues from extensive mantle melting. Although our data do not rule out progressive growth of lithosphere beneath cratons throughout geological time via mechanisms such as imbrication of subducted slabs 1531, they argue that the lithosphere beneath the Kaapvaal craton was at least 150-200 km thick by mid-Archaean times. Later interleaving of subducted oceanic lithosphere into a solid pre-existing lithospheric keel to explain some of the age variation in Fig. 8 does not appear likely considering the manner in which present day sub-

354

D.G. Pearson et al. /Earth

and Planetary Science Letters 134 (I 995) 341-357

ducting slabs are physically deflected by lithospheric keels [54]. Addition to the lithospheric keel from below is a likely possibility as is intrusion of magmatic products higher in the lithosphere, for instance during the voluminous Ventersdorp magmatic event and latterly the Bushveld intrusion. However, the petrological and geochemical similarity of the upper 150 km of Kaapvaal peridotites and their difference from Proterozoic peridotites [44] suggests that these processes may not have been a major factor after the Archaean but may have been important locally. The compositional distinction of the Kaapvaal peridotites and the facility for creating large melt fractions in the Archaean lead us to presently favour a model involving plumes. The overlap of Re depletion ages and crust generation for two cratons and new data showing Early Proterozoic Re depletion ages for peridotites beneath Early Proterozoic crust in Namibia [44] support a link between large-scale crust building and lithospheric mantle stabilisation. The genetic implications of this are currently unclear. Was the crust generated by the same thermal pulse that caused differentiation of the mantle or did the formation of a stable “raft” of lithosphere simply allow preservation of largescale continental crust for the first time? There is no strong chemical evidence at present that crustal rocks are genetically linked to the mantle below. However, the possible involvement of plumes in the generation of Archaean crust [55] and mantle suggests an alternative mechanism to convergent margin volcanism as the primary means of formation of the Earth’s first continents.

9. Stability of Archaean lithospheric mantle and implications for magma sources The eruption of Archaean lithospheric mantle xenoliths in Proterozoic and Phanerozoic kimberlites indicates isolation of thick cratonic lithospheric keels from the convecting mantle and their survival during continental drift for over 3 Gyr. Archaean crust and mantle ages for the Siberian and Kaapvaal cratons show long-term physical coupling of mantle and crust over 3 Gyr periods and illustrates the extreme robustness of Archaean lithospheric mantle to depths of at least 200 km. There have been several episodes

of possible plume impingement on the base of the Kaapvaal lithosphere during the Phanerozoic [56] with at least two major flood basalt episodes in and around the craton. However, comparison of the depths of equilibration of peridotites from Mesozoic kimberlites and those erupted by the Premier kimberlite in the Proterozoic indicate that the depth of lithosphere, at least to the level sampled by kimberlites, remained essentially the same over this period [13]. Thus, from the xenolith evidence, plume impingement on the base of the Kaapvaal lithosphere during the Phanerozoic does not appear to have eroded any substantial thickness. The longevity of Archaean lithospheric mantle indicates its non susceptibility to thermo-gravitational incorporation into the convecting mantle. More Fe-, Al-rich, denser post Archaean lithosphere may be more susceptible to delamination as it cools. In terms of basalt magma sources, melting of Archaean mantle lithosphere by younger plumes may only be possible if significantly hydrated [57]. Evidence from xenolith populations erupted by kimberlites in both the Proterozoic and Phanerozoic shows that primary hydrous phases are only locally abundant in a few areas in Archaean lithosphere and these occurrences may be the result of magmatic activity associated with kimberlites [58]. These observations suggest that if continental lithospheric mantle is a possible source for basaltic magmatism, then postArchaean lithospheric mantle is a more likely source than more refractory Archaean mantle [59]. The persistence of thick (> 200 km) lithosphere beneath some Archaean cratons for periods of 3 Gyr and its long-term coupling with the overlying crust indicates that the geophysical definition of the mechanical boundary layer as a coherent plate appears substantially underestimated beneath continents (Fig. 8). Acknowledgements We thank B. Byrd and D. Kuentz for technical support, D. James and P. Silver for numerous discussions, C. Hawkesworth and D. Canil for comments and L. Reisberg and M. de Wit for constructive reviews. Barry Dawson and Fanus Viljoen provided samples and N. Rogers gave up one of the last pieces of PHN 1611 for analysis. [CL]

D.G. Pearson et al. /Earth

and Planetary Science Letters 134 (1995) 341-357

Appendix A Detailed descriptions of all analytical techniques are given elsewhere [l]. Re-0s chemistry followed procedures similar to [14] but 9 N HBr was used as a trap in the distillation. OS was further purified using a single bead of Chelex 20 resin. Re was separated by anion exchange using 1 ml of AGl-X8 resin, Re being eluted with 4 N HNO,. Total analytical blanks (including filament blanks) were l-5 pg for OS and 5-15 pg for Re throughout the period of study. OS blanks were insignificant for all the samples studied. Some peridotites required Re blank corrections up to 30%. Re and OS were analysed by NTIMS [60,61] on the DTM 15” mass spectrometer (see [l] for details). None of the OS runs reported here required correction for Re. Lack of complete spike-sample equilibration has been documented for digestion of some samples using the combinations of reducing acids employed above [62]. To investigate any possible spike-sample equilibration problems or problems due to incomplete dissolution of acid resistant phases using the procedure employed here several samples were replicated using reducing acids, chromic acid digestion or oxidising dissolution at high temperature and pressure in a modified Carius tube [62]. The results of these tests are given in Table 3 and show excellent agreement between the HF-HCl dissolutions and those done in Carius tubes. The OS concentration for the chromian-spinel-rich peridotite PHN2825 dissolved in chromic acid agrees to within 5% of the value obtained by HF-HCl dissolution. The larger discrepancy for spine1 peridotite PHN5275 is accompanied by a reduction in Re concentration resulting in a very similar initial 0s isotopic composition. These differences therefore may reflect real powder heterogeneity rather than problems with spike-sample equilibration in the HF-HCl dissolution. The replicates from Table 3 show that the dissolution procedure employed for the majority of samples in this study gives reliable results that are in agreement with other digestion techniques.

[2]

[3]

[4]

[5] [6] [7] [S]

[9] [lo] [ll] [12] [13]

[14]

[15]

[16] [17]

[18]

References [19] [l] D.G. Pearson, S.B. Shirey, R.W. Carlson, F.R. Boyd, N.P. Pokhilenko and N. Shimizu, Re-Os, Sm-Nd and Rb-Sr

355

isotope evidence for thick Archaean lithospheric mantle beneath the Siberia craton modified by multi-stage metasomatism, Geochim. Cosmochim. Acta 59, 959-977, 1995. D.G. Pearson, G.A. Snyder, S.B. Shirey, L.A. Taylor, R.W. Carlson and N.V. Sobolev, Archaean Re-0s age for Siberian eclogites and constraints on Archaean tectonics, Nature 374, 711-713, 1995. R.W. Carlson and A.J. Irving, Depletion and enrichment history of suhcontinental lithospheric mantle: OS, Sr, Nd and Pb evidence for xenoliths from the Wyoming Craton, Earth Planet. Sci. Lett. 126, 457-472, 1994. M.J. deWit, C. Roering, R.J. Hart, R.A. Armstrong, C.E.J. deRonde, R.W.E. Green, M. Tredoeaux, E. Pederby and R.A. Hart, Formation of an Archaean continent, Nature 357, _553562, 1992. F.R. Boyd, Compositional distinction between oceanic and cratonic lithosphere, Earth Planet. Sci. L&t. 96, 15-26, 1989. CT. Herzberg, Lithosphere peridotites of the Kaapvaal craton, Earth Planet. Sci. Lett. 120, 13-29, 1993. A. Kroner, Evolution of the Archaean continental crust, Annu. Rev. Earth Planet. Sci. 13, 49-74, 1985. S.H. Richardson, J.J. Gurney, A.J. Erlank and J.W. Harris, Origin of diamonds in old enriched mantle, Nature 310, 198-202, 1984. T.H. Jordan, Continents as a chemical boundary layer, Phi10s. Trans. R. Sot. London A 301, 359-373, 1981. J. Polet and D. Anderson, Depth extent of cratons as inferred from tomographic studies, Geology 23, 205-208, 1995. F.R. Boyd and J.J. Gurney, Diamonds and the African lithosphere, Science 232, 472-477, 1986. J.J. Gurney, The diamondiferous roots of our wandering continents, S. Afr. J. Geol. 93, 423-437, 1991. A.A. Finnerty and F.R. Boyd, Thermoharometry for garnet peridotite xenoliths: a basis for upper mantle stratigraphy, in: Mantle Xenoliths, P.H. Nixon, ed., pp. 381-402, Wiley, London, 1987. R.J. Walker, R.W. Carlson, S.B. Shirey and F.R. Boyd, OS, Sr, Nd, and Pb isotope systematics of southern African peridotite xenoliths: Implications for the chemical evolution of subcontinental mantle, Geochim. Cosmochim. Acta 53, 1583-1595, 1989. P.H. Nixon and F.R. Boyd, Petrogenesis of the granular and sheared ultrabasic nodule suite in kimberlite, in: Lesotho Kimberlites, P.H. Nixon, ed., pp. 48-56, Cape and Transvaal, Cape Town, 1973. A.L. Lerner-Lam and T.H. Jordan, How thick are the continents, J. Geophys. Res. 92, 14,007-14,026, 1987. S.P. Grand, Tomographic inversion for shear velocity beneath the North American plate, J. Geophys. Res. 92, 14,065-14,090, 1987. A.D. Edgar and HE Charbonneau, Melting experiments on a Si02-poor, CaO-rich aphanitic kimberlite from S-10 GPa and their bearing on sources of kimherlite magmas, Am. Mineral. 78, 132-142, 1993. G.P. Brey, L.N. Kogarko and I.D. Ryabchikov, Carbon dioxide in kimberlitic melts, Neues Jahrb. Mineral. Monatsh. 4, 159-168, 1991.

356

D.G. Pearson et al. /Earth

and Planetary Science Letters 134 (1995) 341-357

[20] S. Maaloe and K. Aoki, The major element composition of the upper mantle estimated from the composition of lherzolites, Contrib. Mineral. Petrol. 63, 161-173, 1977. 1211 G.N. Hanson and C.H. Langmuir, Modeling of major elements in mantle-melt systems using trace element abundances, Geochim. Cosmochim. Acta 42, 725-741, 1978. [22] F.R. Boyd and R.H. McCallister, Densities of fertile and sterile garnet peridotites, Geophys. Res. Len. 3, 509-512, 1976. [23] C.J. Hawkesworth, P.D. Kempton, N.W. Rogers, R.M. Ellam and P.W. v. Calsteren, Continental mantle lithosphere, and shallow level enrichment processes in the Earth’s mantle, Earth Planet. Sci. Lett. 96, 256-268, 1990. [24] F.R. Boyd, D.G. Pearson, N.P. Pokhilenko and S.A. Mertzman, Cratonic mantle composition: Evidence from Siberian xenoliths, EOS Trans. Am. Geophys. Union 74, 321, 1993. [25] P.H. Nixon, N.W. Rogers, LL. Gibson and A. Grey, Depleted and fertile mantle xenoliths from southern African kimberlites, Annu. Rev. Earth Planet. Sci. 9, 285-309, 1981. [26] M. Menzies and R.V. Murthy, Enriched mantle: Nd and Sr isotopes in diopsides from kimberlite nodules, Nature 283, 1980. [27] S.H. Richardson, A.J. Erlank and S.R. Hart, Kimberlite-borne garnet peridotite xenoliths from old enriched subcontinental lithosphere, Earth Planet. Sci. Lett. 75, 116-128, 1985. [28] C.E. Martin, Osmium isotopic characteristics of mantle-derived rocks, Geochim. Cosmochim. Acta 55, 1421-1434, 1991. [29] E. Hauri and S.R. Hart, Re-0s isotope systematics in HIMU and EM11 ocean island basalts, Earth Planet. Sci. Lett. 114, 253-271, 1993. [30] L. Reisberg, A. Zindler, F. Marcantonio, W. White, D. Wyman and B. Weaver, OS isotope systematics in ocean island basalts, Earth Planet. Sci. Lett. 120, 149-167, 1993. [31] M.F. Horan, J.W. Morgan, R.J. Walker and J.N. Grossman, Rhenium-Osmium isotope constraints on the age of iron meteorites, Science 255, 1118-1121, 1992. [32] D. Canil, Orthopyroxene stability along the peridotite solidus and the origin of cratonic lithosphere beneath southern Africa, Barth Planet. Sci. Len. 111, 83-95, 1992. [33] M.J. Walter and CM. Bertka, Peridotite solidus from 2-7 GPa: Lithology and melting reactions, EOS Trans. Am. Geophys. Union 75, 192, 1994. [34] P.B. Keleman, H.J.B. Dick and J.E. Quick, Formation of harzburgite by pervasive melt/rock reaction in the upper mantle, Nature 358, 635-641, 1992. [35] L.C. Reisberg., J. Allegre and J.-M. Luck, The Re-0s systematics of the Ronda ultramafic complex in southern Spain, Earth. Planet. Sci. Len. 105, 196-213, 1991. [36] F.R. Boyd and S.A. Mertzman, Composition and structure of the Kaapvaal lithosphere, southern Africa, in: Magmatic Processes: Physicochemical Principles, B.O. Mysen, ed., pp. 13-24, Geochem. Sot., Penn State, Spec. Publ. 1, 1987. [37] H. Allsopp, J.W. Bristow, C.B. Smith, R. Brown, A.J.W. Gleadow, J.D. Kramers and 0. Garvie, A summary of radiometric dating methods applicable to kimberlites and related rocks, in: Kimberlites and Related Rocks: Their Composi-

[38]

[39]

[40]

[41]

[42]

[43]

[44]

[45]

[46]

[47]

[48]

[49]

[50]

[51]

[52] [53]

tion, Occurrence, Origin and Emplacement, J.L. Ross, ed., pp. 349-357, Blackwell, Oxford, 1989. J.P. Lorand, Are spine1 lherzolite xenoliths representative of the abundance of sulphur in the upper mantle?, Geochim. Cosmochim. Acta. 54, 1487, 1990. J.W. Morgan, G.A. Wandless, R.K. Petrie and A.J. Irving, Composition of the Earth’s upper mantle -1. Siderophile trace elements in ultramafic nodules, Tectonophysics 75, 47-67, 1981. F. Walraven, R.A. Armstrong and F.J. Kruger, A chronostratigraphic framework for the north-central Kaapvaal craton, the Bushveld Complex and the Vredefort structure, Tectonophysics 171, 23-48, 1990. J.J. Gurney and B. Harte, Chemical variations in upper mantle nodules from southern African kimberlites, Philos. Trans. R. Sot London A 297, 273-293, 1980. D. Smith, W.L. Griffin and C.G. Ryan, Compositional evolution of high-temperature sheared lherzolite PHN1611, Geochim. Cosmochim. Acta 57, 605-613, 1993, S.H. Richardson, J.W. Harris and J.J. Gurney, Three generations of diamonds from old continental mantle, Nature 366, 256-258, 1993. D.G. Pearson, F.R. Boyd, K.E.O. Hoal, B.G. Hoal, P.H. Nixon and N.W. Rogers, A Re-0s isotopic and petrological comparison of Proterozoic and Archaean mantle: Evidence for long-term crust-mantle coupling beneath contintents and changing modes of Earth differentiation, Mineral. Mag. 58A, 1994. A. Kroner, W. Compston and IS. Williams, Growth of early Archaean crust in the Ancient Gneiss Complex of Swaziland as revealed by single crystal zircon dating, Tectonophysics 161, 271-298, 1989. S.I. Turchenko, The Anabar Shield, in: Precambian Geology of the USSR, D.V. Rundqvist and F.P. Mitrofanov, eds., pp. 247-264, Elsevier, Amsterdam, 1993. A.P. Nutman, I.V. Chemyshev, H. Baadsgaard and A.P. Smelov, The Aldan Shield of Siberia, USSR: the age of its Archaean components and evidence for widespread reworking in the mid-Proterozoic, Precambrian Res. 54, 195-210, 1992. IS. Puchtel, D.Z. Zhuravlev, A.V. Samsonov and N.T. Arndt, Petrology and geochemistry of metamorphosed komatiites and basalts from the Tungurcha greenstone belt, Aldan Shield, Precambrian. Res. 62, 399-417, 1993. Y. Huang, P.W. v. Calsteren and C.J. Hawkesworth, The evolution of the lithosphere in southern Africa: A perspective on the basic granulite xenoliths from kimberlites, Geochim. Cosmochim. Acta, in press. M.J. deWit and R.A. Hart, Earth’s earliest continental lithosphere, hydrothermal flux and crustal recycling, Lithos 30, 309-335, 1993. R.J. Durrheim and W.D. Mooney, Archean and Proterozoic crustal evolution -Evidence from crustal seismology, Geology 19, 606-609, 1991. C.B. Agee and D. Walker, Olivine flotation in mantle melt, Earth Planet. Sci. Len. 114, 315-324, 1993. H.H. Helmstaedt and D.J. Schulze, Southern African kimber-

D.G. Pearson et al. /Earth and Planetary Science Letters 134 (1995) 341-357

1541

[55] [56] [57]

[58]

[59]

lites and their mantle sample-implication for Archean tectonics and lithosphere evolution, in: Kimberlites and Related Rocks, 1, Geol. Sot. Aust. Spec. Pub!. 14, 358-368, 1989. IS. Sacks and J.A. Snoke, The use of converted phases to infer the depth of the lithosphere asthenosphere boundary beneath south America, J. Geophys. Res. 82, 2011-2017, 1977. A. Kriiner and P.W. Layer, Crust formation and plate motion in the early Archean, Science 256, 1405-1412, 1992. W.J. Morgan, Plate motions and deep mantle convection, Geol. Sot. Am. Mem. 132, 7-22, 1972. K. Gallagher and C.J. Hawkesworth, Dehydration melting and the generation of continental flood basal&, Nature 358, 57-59, 1992. P.D. Kinny and J.B. Dawson, A mantle metasomatic injection event linked to late Cretaceous kimberlite magmatism, Nature 360, 726-728, 1992. M.A. Menzies, The lower lithosphere as a major source for continental flood basalts: a re-appraisal, Geol. Sot. Spec. Pub!. 68. 31-39, 1992.

357

[60] R.A. Creaser, D.A. Papanastasio and G.J. Wasserburg, Negative thermal ion mass spectrometry of osmium, rhenium and iridium, Geochim. Gxmochim. Acta 55, 397-401, 1991. 1611 J. Volkening, T. Walczyk and K.G. Heumann, Osmium isotope ratio determinations by negative thermal ion mass spectrometry, Int. J. Mass Spectrom. Ion Process. 105, 147159, 1991. [62] S.B. Shirey and R.J. Walker, Carius tube digestions for Re-0s chemistry: An old technique applied to new problems, EOS Trans. Am Geophys. Union 75, 355-356, 1994. [63] E. Jagoutz, H. Palme, H. Baddenhausen, K. Blum, M. Cendales, G. Dreibus, B. Spettel, V. Lorenz and H. Wanke, The abundances of major, minor and trace elements in the Earth’s mantle as derived from primitive ultramafic nodules, Proc. Lunar Sci. Conf. 10, 2031-2050, 1979. [64] J.M. Luck and C.J. Allbgre, Osmium isotopes in ophiolites, Earth Planet. Sci. Lett. 107, 406-415, 1991. [65] E.H. Hauri, N. Shim& J.J. Dieu and S.R. Hart, Evidence for hotspot-related carbonatite metasomatism in the oceanic upper mantle, Nature 365, 221-227, 1993.