Outils personnels

Vous êtes ici : Accueil / L'institut / Actualités / Soutenance de thèse en géologie : Irène PEZZALI

Soutenance de thèse en géologie : Irène PEZZALI

"Composition et évolution du manteau lithosphérique nord-africain : évidences pétrologiques et géochimiques à partir des enclaves de manteau échantillonnées par le volcanisme Cénozoïque intraplaque du Moyen Atlas (Maroc)".

Tutors: Prof. R. Vannucci (Università degli Studi di Pavia)
Prof. G. Chazot (Université de Bretagne Occidentale)

The study of peridotite xenoliths brought to the surface by alkaline basaltic melts in the
Azrou Timahdite volcanic district (Middle Atlas, Morocco) has revealed extreme lithological and
chemical heterogeneity in the lithospheric mantle beneath the NE-SW volcanic alignment parallel
to the Trans-Moroccan fault system. Most of the investigated xenoliths are lherzolites, with
subordinate harzburgites and pyroxenites. In particular, this Ph.D. Thesis was mainly focused on the
pyroxenite rocks in order to understand whether these rocks represent magmatic events and/or
sectors of subducted crust recycled in the lithospheric mantle. The study was also aimed at
characterizing the composition and the geodynamic evolution of the North-Africa lithospheric
The studied lherzolite and harzburgite samples show a wide heterogeneity as for
petrography, mineral and whole rock chemistry as well. In particular, the study of clinopyroxene
trace element signatures has highlighted that lherzolite samples can be divided in three main
• Group 1: it comprises three samples (2003-1, 2003-2 and HEBRI3), which are
characterized by spoon-shaped clinopyroxene REE patterns, with progressive LREE
enrichment and an almost flat pattern in the MREE and HREE region.
• Group 2: it is formed by six Amp-free lherzolites (TAK16, TAK19, KAD15, BOU1,
SAI1 and SAI2) and two Amp-bearing lherzolites (KAD5 and KAD14); their
clinopyroxenes show LREE (≥ 100xC1) and MREE (> 10xC1) enrichment over
HREE (≈ 10xC1), thus displaying an overall large REE fractionation.
• Group 3: it is constituted only by KAD1 sample; its clinopyroxene has a depleted
LREE pattern (below 10xC1), with almost flat pattern in the MREE and HREE
region (about 10xC1).
Clinopyroxenes from harzburgite samples, moreover, are characterized by relative enrichment in
LREE and MREE over HREE. These patterns are, however, similar to those described for the group
2 lherzolites. The geochemical data for the ultramafic samples confirmed the evolutionary scenario
for the lithospheric mantle beneath the Middle Atlas proposed by Raffone et al. (2009). Only few
samples still record old melt-extraction processes, documented by DMM- or MORB-like isotopic
signatures. After their accretion to the lithosphere, these mantle sectors were exhumed (during late
Cretaceous times) by asthenospheric melts with HIMU-like isotopic signatures. These melts caused
a strong metasomatic overprint in the lithospheric mantle. Accordingly, some lherzolite and
harzburgite samples record the chromatographic effects of melt percolation (group 1 and group 3
samples), whereas other xenoliths (group 2 samples) progressively approached chemical
equilibrium with the migrating melts and were extensively re-fertilized. This heterogeneity possibly
resulted from the difference location of samples with respect to the melt percolation front. Relative
to group 2 samples, group 1 and group 3 samples were located far from the melt sources. It is worth
to note that the lithosphere beneath the Middle Atlas record a long process of thermo-chemical
erosion, also document by the isotopic signatures of the xenoliths. With the exception of
subordinate lherzolites, that have DMM-like isotopic affinity, the great majority of the xenoliths
have a clear HIMU fingerprinting, similar to that of the erupted alkaline lavas.
The petrographic study (chapter 2) showed that the Moroccan pyroxenites could be divided
in four main lithologies (Grt- free and Grt-bearing clinopyroxenite, Grt-websterite and Olwebsterite),
without significant petrographic differences within each group. Despite their
petrographic similarities, the geochemical data (major and trace elements, bulk rock composition,
Sr-Nd isotopic signature, Li and O isotopic fingerprint) revealed that Middle Atlas pyroxenites have
arisen due to different processes:
• TAK4 Grt-clinopyroxenite represents the product of recycling of older oceanic crust.
In particular, the high Al2O3 bulk rock content, the positive Eu anomaly described
for both the whole rock and the mineralogical phases (clinopyroxene and garnet), the
high garnet percentage and the presence of sapphirine strongly support the
hypothesis that the protolith was a plagioclase-rich cumulate, i.e. plagioclase-rich
troctolite. Similar high Al2O3 bulk rock contents were reported by Bach et al. (2001)
for the altered gabbros from the South-Western Indian Ridge (SWIR). For this
reason, it is not excluded that the protolith was an altered plagioclase-rich cumulate.
• TAK13 and TAK17 Ol-websterites also represent recycled oceanic crust. The
marked positive Eu anomalies (described both for the bulk rock and for the
clinopyroxenes REE patterns) strongly support this hypothesis. The low Al2O3 bulk
rock content, similar to that reported by Coogan et al. (2001) for the fresh gabbros of
SWIR, suggests that the protolith was a Pl-poor cumulate (i.e. Ol-rich troctolite).
• TAK3 and TAK5 Grt-websterite were, most probably, formed by reactions between
mantle melt and older mafic layers (most probably older pyroxenites). The lack of
positive Eu anomalies (i.e. the evidence in favour of former plagioclase) seems to
rule out an origin as recycled oceanic crust. On the other hand, the high garnet and
spinel contents associated with the high Al2O3 bulk rock values are incompatible
with a simple crystal accumulation processes. These characteristics, therefore,
suggest that the older pyroxenites with which the mantle melts reacted were made by
clinopyroxene, spinel and garnet. The involvments of a crustal component in the
origin of TAK3 sample is suggested by the low δ7Li values. The origin of TAK5
sample is more controversial. The high Al2O3 content and similarity of REE spectra
with some A-group pyroxenites from Ronda (Garrido & Bodinier, 1999) favours a
melt reacted origin. However, if the high Al2O3 content is disregarded, REE patterns
and other incompatible elements and Li isotope as well, do not allow a clear
distinction from group 3 pyroxenites.
• TAK7 Ol-websterite shows a clinopyroxene REE pattern similar to those described
for the clinopyroxene grains of the Morocco lherzolites of first group. This
similarity, associated with the higher bulk rock MgO content, lead to hypothesize
also for TAK7 sample an origin due to open system magmatic replacement of preexisting
mantle ultramafic rocks. TAK7 mineral mode (i.e. high orthopyroxene and
olivine mineral percentages, with low clinopyroxene content and subordinate spinel)
implies that the ultramafic protolith was a mantle peridotite, probably harzburgite in
• TAK6, TAK8, TAK9, TAK10, TAK12 and IBA21c Grt-websterites show similar
geochemical characteristics. These samples are better interpreted as the result of
magmatic crystallization from enriched melts at mantle depth. After their formation
in the spinel stability field, pyroxenites re-equilibrated in the garnet stability field
thus forming the Grt-websterites.
• BOU4 clinopyroxenite shows cumulate features. The petrographic characteristics
together with all the geochemical data confirm that BOU4 clinopyroxenite
crystallized from alkaline melts with HIMU component.
An overall geodynamic scenario for the Middle Atlas (northwestern Morocco) is illustrated
in figure 8.1, where the lithosphere-asthenosphere interaction processes beneath the area are
described in a simplied form. Based on the discussion reported in Charter 7 (§ 7.1), the sectors of
rejuvenated lithosphere formed by intra-cratonic upwelling and accretion of asthenospheric material
are shown, along with the location of group 1, group 2 and group 3 lherzolite xenoliths. The latter
are located at distinct structural levels depending on their relative position with respect to the melt
percolation front.
Within this scenario the origin and the age of pyroxenites interpreted as fragments of
recycled oceanic crust (TAK4, TAK13 and TAK17) and as products of melt-rock interaction
processes (TAK 3 and TAK55) are not completely understood. By analogy with Ronda and Beni
Bousera these pyroxenites could represent old mafic rocks that they have been isolated in the
subcontinental lithospheric mantle for very long time spans. On this ground, the fragments of
Figure 8.1 – Schematic illustration of the geodynamic scenario beneath the Middle
Atlas (northwesetrn Morocco). The location of the different group of lherzolites and
pyroxenites and the sectors of rejuvenated lithosphere formed by intra-cratonic
upwelling and accretion of asthenospheric material are shown (modified after Raffone
et al., 2009).
Oceanic material and the crustal components recorded by pyroxenites may be tentatively related to
subduction processes occurred during Pan-African times (Neoproterozoic).
If so, these pyroxenites maintained for long time their pristine geochemical signatures
withourt marked changes, which are expected at least for stable isotopes (and particularly Li). An
alternative interpretation to be still explored is that these pyroxenites contain crust-derived
components involved in or released from recent (<85 ma) subduction processes occurred in western
Mediterranean. According to several Authors (Gelabert et al., 2002; Duggen et al., 2003, 2005;
2009; see Chapter 1; § 1.1.1), subduction processes started since the late Cretaceous and developed
until 10-8 ma, when slab rollback and continental age delamination caused subduction suction. In
this scenario, fragments of oceanic crust could have been temporarely stored in the lithosphere
before their uptake by plume-related Plio-Quaternary alkaline lavas that brought them to the

Photo du mois


(C) Pascale Lherminier / Ifremer