Between a Rock and a Hard Place

Using garnet geochemistry to investigate the lithospheric mantle beneath northern Tanzania

Sorcha McMahon is a third year PhD student in the School of Earth Sciences at the University of Bristol. Sorcha is investigating how strange igneous rocks called carbonatites may have formed, using both natural samples and high-pressure experiments.

As part of my undergraduate MSci course at the University of Cambridge, I carried out a project investigating a collection of thin sections from peridotite xenoliths from northern Tanzania. The main aim of this research was to constrain the petrogenetic evolution of the lithospheric mantle beneath the East African Rift (EAR) and adjacent Tanzanian Craton. Techniques employed included the electron microprobe (EMPA) and LA-ICP-MS to characterise the major and trace element contents of constituent minerals in garnet-bearing lherzolite and harzburgite mantle xenoliths. For the first time globally, we found ultra-depleted pyrope garnets from Lashaine peridotites displaying rare earth element (REE) patterns similar to those of hypothetical garnets proposed to have formed in the subcontinental lithospheric mantle prior to any metasomatism*.

Lashaine volcano is a small tuff cone crater of Quaternary age located near the eastern margin of the Tanzanian Craton in east Africa (3o22’S, 36o26’E). In this study, we focus on lherzolite and harzburgite xenoliths erupted in olivine melilitite and carbonatite scoria.

Map of northern Tanzania

Map of East Central Africa showing Rift valleys, the craton and the surrounding early to late Proterozoic mobile belts, modified from Koornneef et al. (2003). E-EAR and W-EAR are eastern branch and western branch of the East African Rift, respectively. Stars indicate Labait, Lashaine and Mwadui xenolith localities. Credit: Sorcha McMahon

The Lashaine rocks are exceptionally fresh and display typical cratonic mineralogy: forsteritic olivine, enstatitic orthopyroxene (opx), emerald-green Cr-diopside (clinopyroxene, cpx) and purple pyrope garnet. Phlogopite mica is present in some sections. Textures approaching equilibrium (dihedral angles of 120o) are most common. Fresh pyrope-garnets appear pink in thin section and generally range from 1 to 6 mm. They are invariably replaced by fine-grained aggregates of clinopyroxene, spinel, orthopyroxene and brown glass, and reaction coronae are common indicating rapid decompression during xenolith eruption.

Garnet reaction corona

A photomicrograph of a reaction corona of garnet in Lashaine xenolith BD812, annotated with zones described in Dawson et al. (1970) and Reid and Dawson (1972). Credit: Sorcha McMahon

Some of the Lashaine garnet harzburgites (harzburgites are defined as containing less than 5 % modal cpx) display a banded texture, in which there are ~1 cm wide, continuous bands of equant crystals of pyrope garnet (2 – 4 mm in diameter), regularly occurring at 1 cm spaced intervals. ‘Necklace’ textures are also present where the garnets form an interconnected network around large (up to 6 mm) equant grains of orthopyroxenes.

Banded Garnet Harzburgite BD3928

(a) Image of Lashaine banded garnet harzburgite BD3928 showing the textural relationship between ultra-depleted garnet (Gt), orthopyroxene (Opx) and olivine (Ol). (b) garnet harzburgite BD3928 showing the textural relationship between ultra-depleted garnet (Gt), orthopyroxene (Opx) and olivine (Ol). (b) Schematic illustration of (a) to show the close spatial relationship between garnet and orthopyroxene. (c) Higher magnification image of the ‘necklace’ texture shown in (a) and (b). The location of the image shown in (c) is outlined in (a) by the white rectangle. Credit: Sorcha McMahon

Using major element concentrations measured on the electron microprobe, mineral pairs in equilibrium are used to calculate equilibration pressures and temperatures for each garnet-bearing sample. The two-pyroxene thermometer of Taylor (1998) and the Al-in-orthopyroxene barometer of Nickel & Green (1985) are used in conjunction with crustal thickness data (Julia et al. 2005) and the FITPLOT program (McKenzie et al. 2005), to conclude the following:

(1) The surface heat flux is low (~41 mW m^2).

(2) Assuming an elevated TP of 1400 C (owing to the presence of the East African plume) the lithosphere is 186 km thick, and

(3) the base of the mechanical boundary layer (MBL) resides at 170 km.

Relatively calcic garnets (CaO 4.8 – 6.7 wt. %) are generally found in the higher temperature lherzolites, and have normal to mildly sinusoidal chondrite-normalised REE patterns (as measured by LA-ICP-MS). Lower temperature Lashaine harzburgites are estimated to have come from shallower depths (140 – 110 km) and have sinusoidal to ‘square-root-symbol’-shaped chondrite-normalized REE patterns. Low temperature (~1150 C) ultra-depleted garnets in banded harzburgites display extremely low CaO contents (<0.35 wt. %) and ‘square-root-symbol’-shaped REE patterns. These highly magnesian (Mg# 91.5 – 92.5) garnets coexist with high-Mg# orthopyroxene (up to 96.4) and olivine (Fo95.6).

Garnet REE plots

Chondrite-normalized REE patterns of Lashaine garnets (using factors from McDonough & Sun, 1995). Bold continuous line shows the hypothetical composition of pre-metasomatic garnet after Stachel et al. (2004). Dashed lines in (d) illustrate the effect of adding very small amounts of high-Mg carbonatitic fluid (ON- KAN-389; Weiss et al., 2009) to a pre-metasomatic garnet. Credit: Sorcha McMahon

The REE patterns of garnets in these banded harzburgites are remarkably similar to examples of pre-metasomatic garnets as hypothesised by Stachel et al. (2004). Differences in LREE (Light REE, to the left of the graphs above) between hypothetical pre-metasomatic garnets and our banded garnets may be due to a very small enrichment by a high-Mg carbonatitic fluid. Combining evidence from major and trace element contents, we propose that the decrease in Ca, REE and temperature (and depth) represents decreasing enrichment by percolating metasomatic agents.

The ‘necklace’ texture displayed in the banded garnet harzburgites, coupled with the garnets’ compositions, is interpreted to result from the isochemical exsolution of garnet from orthopyroxene. The development of garnet around the margins of large orthopyroxene grains may have occurred via an Ostwald ripening mechanism, similar to that proposed by Dawson (2004). In previous studies of such a phenomenon, the process initially involves strain-induced exsolution of garnet from orthopyroxene along planar interfaces and the formation of lamellae. These lamellae coarsen with time, until the diffusion distance is the same as the grain size, and equant grains of garnet start to crystallize along orthopyroxene grain boundaries and develop necklace textures. In contrast to an exceptionally preserved example of exsolution lamellae (150 – 200 um at Monastery Mine, Kaapvaal craton), the Lashaine garnets are significantly larger, suggesting that they formed over longer timescales (assuming formation by the same exsolution mechanism).

Such pre-metasomatic garnets are rare on a global scale, likely due to their formation at a relatively early stage in craton evolution and then probable overprinting by subsequent metasomatism.

For a more thorough (and slightly lengthier!) review of this study, please read http://petrology.oxfordjournals.org/content/54/8/1503 (closed access journal).

This paper was written with Barry Dawson, who sadly passed away earlier this year. He was a great geologist and it was a pleasure to meet him and work on samples he had collected in the 1960s as part of the (then) Tanganyika Geological Survey.

* The Collins English Dictionary defines metasomatism as “the change in the composition of a rock or mineral by the addition or replacement of chemicals”. This change is typically the result of percolating melts or fluids, and can result in ‘modal’ (replacement/addition of minerals present) or ‘cryptic’ metasomatism (changes reflected in the geochemistry).

Charly completed a PhD in experimental petrology. She used to make pretend volcanoes; now she works in renewable energy. Charly tweets at @C_Stamper.