Journal of Petrology Advance Access published online on December 4, 2008
Journal of Petrology, doi:10.1093/petrology/egn057
| ||||||||||||||||||||||||||||||||||||||||||||||||||||
Li, B and Be Contents of Harzburgites from the Dramala Complex (Pindos Ophiolite, Greece): Evidence for a MOR-type Mantle in a Supra-subduction Zone Environment
1Institute of Geology and Hydrogeology, University of Neuchatel, Emile Argand 11, CP158, 2009 Neuchatel, Switzerland
2Institute of Isotopes, Hungarian Academy of Sciences, po BOX 77, 1525 Budapest, Hungary
Received January 12, 2007; Revised typescript accepted October 14, 2008
 |
Abstract |
|---|
The Pindos ophiolite represents oceanic lithosphere obducted during the Jurassic. The Dramala mantle section mainly consists of highly depleted spinel harzburgite and minor plagioclase-bearing harzburgite. Textural observations and major element compositions of minerals indicate that the harzburgites experienced impregnation by a mafic, depleted melt and subsequent high-temperature (high-T) hydration and cooling (>750°C) forming pargasite and edenitic hornblende. During further cooling (from 
350–400°C to < 100°C), talc + tremolite ± serpentine ± olivine, serpentine + magnetite, and finally plagioclase alteration phases formed. To test the hypothesis of a supra-subduction zone origin for the Dramala mantle, we measured Li, B and Be contents of minerals by secondary ion mass spectrometry. Whole-rock contents were measured using inductively coupled plasma–mass spectrometry and prompt gamma neutron activation analysis. We observe low Li and B contents of primary minerals (olivine, orthopyroxene, clinopyroxene) consistent with values for unmetasomatized mantle minerals; only Li contents of clinopyroxene (up to 3·7 µg/g) are slightly elevated. The bulk Li contents (0·5–1·1 µg/g) are in the upper range of values for unmetasomatized mantle, whereas B contents (<0·04–1·1 µg/g) are variable and slightly elevated compared with the unmetasomatized mantle as a result of serpentinization. Beryllium abundances in all minerals are very low (<0·005 µg/g), except for pargasite, where a maximum Be content of 0·012 µg/g was measured. The selective addition of Li to clinopyroxene can be related to the interaction with a depleted melt, and/or to partitioning of Li into clinopyroxene upon cooling. During high-T hydration and cooling, the fluid calculated to be in equilibrium with the pargasite or edenitic hornblende (based on Li, Be and B) could have been reaction-modified seawater. Low-T hydration may have led to a very minor increase in bulk B content of most samples and to the formation of serpentine with highly variable B contents (0·1–28 µg/g). Low-T hydration decreased the Li content of orthopyroxene, and Li was probably leached from some samples. The lack of correlation between degree of serpentinization and bulk B contents as well as the presence of high- and low-B serpentine can be explained by low fluid–rock ratios, decreasing T during serpentinization and lack of equilibrium as a result of fast obduction–exhumation. The low light-element contents of primary minerals and whole-rock samples clearly argue against a supra-subduction zone (SSZ) origin of the Dramala mantle section, and against the previous hypothesis of hydrous melting of the Pindos mantle above a subduction zone. We therefore conclude that the Dramala harzburgites represent a mid-ocean ridge (MOR)-type mantle, and not an SSZ-type mantle, juxtaposed with MOR-type and SSZ-type oceanic crust, either in a back-arc or in an intra-oceanic subduction zone setting.
KEY WORDS: light elements; melt impregnation; peridotite; supra-subduction zone ophiolite; MOR-type mantle; lithium; beryllium; boron
*Corresponding author. Present address: EMPA, Ueberlandstrasse 129, CH-8600 Dübendorf, Switzerland. E-mail: laure.pelletier{at}empa.ch