Modal mineralogy and geochemistry of Kaapvaal peridotites; the origin
of garnet and diopside and implications for craton stability
D. G. Pearson1, F.R. Boyd2 & N.S.C. Simon3
1
Dept. Geological Sciences, Durham University, South Rd, Durham, UK
2
Geophysical Laboratory, Carnegie Institution of Washington, 5251 Broad Branch Road,
N.W., Washington, D.C. 20015, U.S.A.
3
Faculty of Earth Science, University of Amsterdam, De Boelelaan 1085, 1181 HV
Amsterdam, The Netherlands.
Introduction
The origin of Kaapvaal garnet peridotites has long been a subject of controversy [1-6].
Most of the debate has centred around the cause of the anomalous orthopyroxene
enrichment compared to peridotites from other tectonic settings while less attention has
been paid to the origin of the garnet and diopside in these rocks [7]. Garnet exerts the
largest control on the bulk density of peridotitic mantle [8], hence knowing how and
when the garnet in lithospheric peridotites formed is a fundamental constraint on the
temporal evolution of the bulk density of the lithosphere, at least in the region of
kimberlite eruption. In addition, garnet and diopside dominate the trace element and
incompatible element radiogenic isotope signature of these rocks and thus control the
chemical evolution of the lithosphere through time. Here we use modal and elemental
data to evaluate the likelihood of the garnet and diopside being of primary origin in
Kaapvaal peridotites. An accompanying presentation [9] will address more directly the
timing of formation of these two phases from a trace element perspective.
Possible origins for garnet and diopside in Kaapvaal cratonic peridotites:
Some of the end-member models for formation of the garnet and diopside within
Kaapvaal peridotites are listed below together with observations from mineral chemistry
that bear on these hypotheses. Many other models have been produced that are hybrids
of these “end-members” but are omitted here for brevity.
Model 1) As residual phases after extraction of large degrees of melt extraction.
This model involves single or multiple-stage melt depletion and produces modal
mineral variations that are a function of both the amount of melt extracted from
the rock and the pressure at which the melting takes place.
Observations: Garnet and diopside are overabundant for rocks of such high mgnumber if the peridotites represent low pressure (<3 GPa) melt residues
(experimental data of Kinzler & Grove; [10]; Fig. 1). Plots of mineral abundance
versus mg-number of garnet for Low-T peridotites produce scatter rather than the
correlations expected for melt removal processes. Garnet and diopside
abundances could be consistent with being residual phases following 6-7 GPa
melting of peridotites (Fig. 1) using the experimental data of Walter [11], but
residues formed at this pressure are dunites or orthopyroxene poor peridotites i.e.,
dissimilar to Kaapvaal peridotites. Both diopsides and garnets are always LREE
enriched in low-T peridotites, inconsistent with a residue model. Finally, Sr-Nd
isotope ratios in both phases rarely show evidence of long-term parent-daughter
element depletion.
Model 2)
As residual phases that have suffered cryptic enrichment from
infiltrating melts and later re-equilibration.
This model involves the melt depletion event(s) in model 1 followed by
subsequent additions of incomaptible elements in fluids/melts
Observations: Unless cryptic enrichment added most of the volume of garnet and
diopside these phases are still too abundant for any plausible low-P residues and
the garnet+diopside/orthopyroxene ratio is too high for high-pressure residues.
The consistent shape of REE patterns in garnets and diopsides does not suggest
equilibration between depleted cores and enriched rims of varying REE patterns.
Any trace element zonation patterns usually involve LREE depleted garnet rims
and LREE enriched garnet cores [9], i.e., the opposite of that expected for simple
enrichment of a pre-existing residual phase.
Model 3) Exsolution products from high temperature/pressure opx [12,13, 14]
Statistical correlations between the abundance and location of opx, garnet and
diopside [12-14] and high P/T experiments suggest that a high-P aluminous
orthopyroxene is a possible parent phase that exsolves garnet and diopside
[13,14].
Observations: However, subtle major element disequilibrium between garnet and
diopside in numerous Low-T peridotites [15] together with the incompatible
element rich nature of garnets and diopside and trace element disequilibrium [9]
indicate that a simple exsolution mechanism is unlikely for most garnet and
diopside. Isotopic disequilibrium in many rocks [16-18] also argues against this.
Model 3) Products of precipitation directly from a melt [19-22]
This model suggests that garnet and diopside crystallise during melt/fluid metasomatism
of a garnet-poor harzburgite parent.
Observations: Major and trace element studies indicate that garnet and diopside probably
do not crystallise together in many rocks [9, 15]. Garnet and diopside compositions are
not consistent with melt extraction residues. Excess modal diopside and garnet at very
depleted olivine mg-numbers can be explained if garnet and diopside are introduced later
via a melt. Major element differences between peridotite phases and experiments could
be accounted for. LREE-enriched REE patterns would be expected. The apparent major
element disequilibrium between garnet and diopside in numerous Low-T rocks [15], the
lack of trace element [7,9] and isotopic [16-18] equilibrium can be explained if they
crystallised at different times.
Timing and origin
Several features indicate that the crystallisation of garnet and diopside in many Kaapvaal
Low-T peridotites could have occurred relatively recently, as proposed by Shimizu [7].
Some grains of garnet and diopside are very lobate and do not form grain boundaries
indicating long-term textural equilibrium [7,9]. Major and trace element zonation in both
garnet and diopside in Low-T peridotites is indicative of either recent formation of the
whole grain or recent addition of the rims without complete equilibration (<1000 years at
mantle temperatures). Gross Nd isotope disequilbrium between diopside and garnet in
clusters within Kimberley peridotites [16] could suggest recent crystallisation of both
phases.
Our preferred model for the origin of diopside and garnet in Kaapvaal peridotites is via
introduction (patent metasomatism) significantly after initial melt depletion i.e., option 4.
Major element and trace element equilibria suggest that the diopside in many Low-T
rocks formed after the garnets, or possibly at the same time as some garnet rims. One
question that arises from this hypothesis is the nature of the medium that these phases
crystallised from.
Burgess and Harte [20, 21] propose that the garnets in Low-T peridotites crystallise from
a fractionated melt that crystallised the low-Cr megacryst suite. Our new trace element
data for Lesotho peridotites supports this model [9]. The diopsides may have crystallised
from a different medium. Boyd & Mertzman [23] have proposed that fine-grained,
secondary diopside crystallised along with mica in Jagersfontein Low-T peridotites,
possibly from a supercritical fluid phase via the reaction:
Garnet + Orthopyroxene + Fluid = Phlogopite + Diopside
Modal abundance evidence indicates a relationship between coarse phlogopite and
diopside in other peridotite suites that may imply formation of coarse diopside via similar
reactions. Strong correlations between the presence of diopside and coarse mica are
observed for lherzolites (both garnet and spinel-bearing) from the Premier kimberlite
(Boyd, unpublished data) whereas harzburgites from the same pipe are almost mica-free.
A similar relationship is evident for peridotites from Kimberley. Hence, both modal
abundance data and trace element data [9] support a separate origin for diopsides,
possibly related to a hydrous fluid or the kimberlitic melt [9]. A similar relationship has
been proposed for diopsides from the Wesselton peridotite suite [24].
Implications
From the evidence presented above we agree with the suggestion of Shimizu [7] that a
large fraction of the garnet and diopside in Kaapvaal low-T peridotites crystallised
relatively late in the evolutionary history of the peridotites. We concur with Shimizu [7]
that this has major implications for the density and chemical evolution of the lithospheric
keel beneath the Kaapvaal craton. Garnet is a major influence of the density of
peridotite. In addition, the inventory of incompatible elements and their daughter
isotopes is almost entirely contained in garnet and diopside. The longevity of the
Kaapvaal cratonic keel (e.g. [22]) would be promoted by its existence as a garnet and
diopside-poor harzburgite. The density of this keel may have increased significantly in
post-Archean times, possibly related to melt infiltration associated with Mesozoic
plumes. Some models have suggested replacement of the base of the Kaapvaal
lithosphere with hot asthenosphere [25] to cause regional the uplift inferred from fissiontrack studies [26]. The most recent seismic studies indicate cool lithosphere down to in
excess of 250 km for most of the Kaapvaal craton [27]. A sudden increase in density due
to melt infiltration and mineral growth may have been sufficient to cause delamination of
the lower fraction of the Kaapvaal lithospheric mantle (below 250 km). This could then
induce the recent (90 Ma) regional uplift event.
References
1.
Boyd, F.R. Earth Planet. Sci. Lett. (1989) 96, 15-26.
2.
Canil, D. earth Planet. Sci. Lett. (1992) 111, 83-95.
3.
Herzberg, C.T. (1993) Earth Planet. Sci. Lett. 120, 13-29.
4.
Walter, M.J. (1999) in Mantle Petrology: Field observations and high pressure experimentation
ed. Y. Fei, C.M.B.B.O.M.) 225-239 (Spec. Pub. Geochem. Soc. No. 6, Houston,.
5.
Keleman, P.B., Dick, H.J.B. & Quick, J.E. Nature 358, 635-641 (1992).
6.
Keleman, P.B., Hart, S.R. & Bernstein, S. Earth Planet. Sci. Lett. 164, 387-406 (1998).
7.
Shimizu, N. in Mantle Petrology: Field observations and high pressure experimentation: A tribute
to Francis R. (Joe) Boyd (ed. Y. Fei, C.M.B.B.M.) 47-55 (The Geochemical Society, Houston,
1999).
8.
Boyd, F.R. & McCallister, R.H. Geophys. Res. Lett. 3, 509-512 (1976).
9.
Simon, N.S.C., Pearson, D.G., Carlson, R.W. & Davies, G.R. Slave-Kaapvaal craton workshop
This volume (2001).
10.
Kinzler, R.J. & Grove, T. (1999), in Proceedings of the VIIth international kimberlite conference
(eds. Gurney, J.J., Gurney, J.L., Pascoe, M.D. & Richardson, S.H.) 437-443 (Red Roof Design,
Cape Town, 1999).
11.
Walter, M.J. (1998) J. Petrology, 39, 29-60.
12.
Cox, K.G., Smith, M.R. & Beswetherhick, S. (1987) In: Nixon, P.H. (ed) Mantle Xenoliths, New
York: Wiley & Sons.
13.
Canil, D. (1991) Earth Planet Sci. Lett.
14.
Saltzer. R.L., Chatterjee, N. & Grove, T.L. (in press) Contrib. Mineral. Petrol.
15.
Brey, G.P. (1989) Habilitations Thesis. Tech. Hochsch. 217 (Darmstadt, Germany, 1989).
16.
Gunther, M. & Jagoutz, E. Proc. 5th International Kimberlite Conference, (1994).
17.
Gunther, M. & Jagoutz, E. Russian Geol. Geophys. 38, 229-239 (1997).
18.
Richardson, S.H., Erlank, A.J. & Hart, S.R. (1985) Earth Planet Sci Lett 75, 116-128.
19.
Matthews, M.B., Harte, B. & Prior, D. Geochim. Cosmochim. Acta 56, 2633-2642 (1992).
20.
Burgess, S.R. & Harte, B. in Proceedings of the VIIth international kimberlite conference (eds.
Gurney, J.J., Gurney, J.L., Pascoe, M.D. & Richardson, S.H.) 66-80 (Red Roof Design, Cape
Town, 1999).
21.
Burgess, S. & Harte, B. (in press) J. Petrology .
22.
Pearson et al., (in press) in “The Early Earth” Fermor Volume, Spec. Pub. Geol. Soc. London.
23.
Boyd & Mertzman (1987) in: B.O. Mysen (ed). Magmatic Processes: Physiochemical Principles.
Spec. Pub. The Geochemical Society no. 1, 13-24.
24.
Griffin, W.L., Shee, S.R., Ryan, C.G., Win, T.T. & Wyatt, B. (1999) Contrib. Mineral. Petrol.
134: 232-250.
25.
Griffin, W.L. et al. In: Braun et al (eds) The structure and evolution of the Australian continent.
Geodynamics volume 26, Am. Geophys. Union. Washington DC, p1-26.
26.
Brown, R.W., Gallagher, K., Griffin, W.L., Ryan, C.G., de Wit, M.C.J., Belton, D.X. & Harman,
R. In: Ext. Abstracts 7th International Kimberlite Conference, Cape Town, p105-107.
27.
James, D.E. et al. Geophys. Res. Lett. (2001) 28, p2485-2488.
30.0
Modal garnet %
Walter 1998 Data
20.0
Kaapvaal Gt
7GPa
Gt % 3 GPa
6GPa
Gt % 5 GPa
10.0
Gt % 6 GPa
5GPa
Gt % 7 GPa
3 GPa
0.0
90.0
92.0
94.0
mg-number olivine
Modal orthopyroxene %
40.0
Walter 1998
30.0
6 GPa
Kaapvaal Low-T
20.0
4-5 GPa
10.0
0.0
90.0
92.0
94.0
mg-number olivine
Fig. 1: Plots of olivine mg-number versus modal abundance of garnet and orthopyroxene for Kaapvaal
Low-T peridotites compared to experimental studies of melt extraction from fertile peridotite [10,11]. The
data are incompatible with any simple low or high-P melt extraction history.