Journal of Petrology Advance Access originally published online on April 17, 2007
Journal of Petrology 2007 48(6):1079-1118; doi:10.1093/petrology/egm011
© The Author 2007. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oxfordjournals.org
Entrained Macrocryst Minerals as a Key to the Source Region of Olivine Nephelinites: Humberg, Kaiserstuhl, Germany
Alexey Ulianov1,*,
Othmar Müntener1,
Peter Ulmer2 and
Thomas Pettke3
1University of Lausanne, Institute of Mineralogy and Geochemistry, Anthropole, CH-1015 Lausanne, Switzerland
2Institute of Mineralogy and Petrography, Eth Zürich, Eth Zentrum, Sonneggstrasse 5, CH-8092 Zürich, Switzerland
3Institute of Geological Sciences, University of Bern, Baltzerstrasse 1–3, CH-3012 Bern, Switzerland
RECEIVED
MAY 18, 2006;
ACCEPTED
FEBRUARY 27, 2007
 |
ABSTRACT
|
|---|
Olivine nephelinites commonly contain macrocrysts of olivine
and clinopyroxene. Some of these macrocrysts might represent
fragments of the source region of the host magma transported
to the Earth's surface. If this hypothesis is correct these
fragments can be used to characterize the composition of the
source region and to put constraints on the magma generation
process. In this study, we investigate the origin of macrocrysts
and mineral aggregates from an olivine nephelinite from the
Kaiserstuhl, Germany. We focus on clinopyroxenes (Cpx), which
can be divided into three groups. Cpx I is relict Cpx from aggregates
with deformed olivine that is depleted in Ca and characterized
by strong light rare earth element (LREE) fractionation, low
Ti/Eu and negative high field strength element (HFSE) anomalies.
Its geochemical signature is consistent with formation by carbonatite
metasomatism and with equilibration in the presence of orthopyroxene.
Cpx II is Ca-rich Cpx, forming both aggregates with deformed
olivine and individual macrocrysts. The LREE, as for Cpx I,
are strongly fractionated. Convex REE patterns may be present.
The depletion in HFSE is less pronounced. Cpx III is oscillatory
zoned Cpx phenocrysts showing enrichment in Ca, convex REE patterns
and no HFSE anomalies. The transition in the trace element abundances
between the Cpx of the three groups is gradual. However, Cpx
I and II did not crystallize from the host magma, as demonstrated
by the presence of kink-bands and undulose extinction in the
associated olivine and by the composition of alkali aluminosilicate
glass inclusions in Cpx II. Based on the Cpx relationships,
we interpret the studied suite of macrocrysts and mineral aggregates
as a mixture of disintegrated fragments of the source region
of the host olivine nephelinite. The process of melt generation
was multi-stage. A primary carbonatite melt ascending from deeper
levels in the mantle, probably from the dolomite–garnet
peridotite stability field, reacted with mantle peridotite along
the solidus ledge in the system lherzolite–CO
2 (< 20–22
kbar) and started to crystallize carbonate minerals. Because
of its low solidus temperature, the resulting carbonate-wehrlite
assemblage melted incongruently with the formation of additional
clinopyroxene. The carbonatite melt evolved during crystallization
of carbonate minerals and concomitant incongruent melting of
the carbonate-wehrlite, accompanied by the segregation of incipient
alkali aluminosilicate melts. As a consequence of fast reaction
rates in the presence of a carbonatite melt, this process probably
took place under disequilibrium conditions. Further melting
of the assemblage wehrlite + alkali aluminosilicate melt led
to the generation of the olivine nephelinite magma. It entrained
fragments of the wehrlite and brought them to the surface.
KEY WORDS: carbonatite; metasomatism; source region; clinopyroxene macrocrysts; wehrlite; olivine nephelinite; Kaiserstuhl
|
INTRODUCTION
|
|---|
The understanding of magma generation processes requires characterization
of the source region that undergoes partial melting. Constraints
on the phase assemblage of the source region, the major and
trace element compositions of its mineral and volatile components
and conditions of partial melting are actively debated among
petrologists. Except for the study of
in situ examples of partial
melting during high-
T metamorphism, unravelling the source region
characteristics is based on indirect approaches. In the case
of alkaline magmatic rocks, a wealth of geochemical and some
phase equilibrium constraints can be derived from their major
and especially trace element geochemistry and radiogenic isotope
compositions (e.g. Greenough, 1988
; Wedepohl
et al., 1994
; Furman,
1995
; Späth
et al., 2001
). Studies of entrained mantle
xenoliths have revealed minimum
P–
T conditions of formation
of alkaline mafic and ultramafic magmas (e.g. Boyd, 1973
; Henjes-Kunst
& Altherr, 1992
; Akinin
et al., 1997
; Lee & Rudnick,
1999
; MacKenzie & Canil, 1999
; Kopylova & Caro, 2004
).
Additionally, a number of experimental investigations have been
conducted that focused on their high-pressure liquidus phase
relations (e.g. Bultitude & Green, 1971
; Green, 1973
; Brey
& Green, 1977
; Adam, 1990
; Girnis
et al., 1995
; Ulmer &
Sweeney, 2002
). Liquidus experiments can provide direct insights
into the nature of the phase assemblages in the source regions.
However, they commonly assume homogeneity and thermodynamic
equilibration of the source region and a given set of
P–
T conditions, and the fluid composition is determined
a priori.
Although either approach - geochemical, xenolith-based, and
experimental - has its strengths and weaknesses, neither allows
study of the source region mineral assemblages directly in a
natural environment.
Alkaline mafic and ultramafic magmas are generally considered to rise rapidly to the Earth's surface. Their ascent to the surface is often accompanied by the entrainment of significant amounts of disintegrated rocks from the walls of the magma conduit as xenoliths and xenocrysts. Typical magmas that bear such debris are kimberlites, foidites and alkali basalts (for reviews, see Sobolev, 1977; Dawson, 1980; Nixon, 1987; Menzies, 1990; Griffin et al., 1999). At the same time, many kimberlite, foidite and alkali basalt intrusions are almost devoid of xenoliths but contain abundant macrocrysts of olivine (kimberlites) or olivine and clinopyroxene (olivine nephelinites, basanites) showing strong indications of solid-state deformation such as kink-bands, undulose extinction, recrystallization and internal cracks (e.g. Skinner, 1989; Mitchell, 1995). These minerals are neither the products of fragmentation of mantle xenoliths, because of the lack of other minerals typical of common mantle lithologies, nor could they have crystallized from the host magma, because of their deformation features. This raises the question of whether they are entrained directly from the source region of the host magma. If so, they could provide direct constraints on the phase composition, trace element characteristics and, potentially, the P–T conditions of the source region.
This study focuses on a suite of macrocrysts and clinopyroxene–olivine aggregates from a single specimen of olivine nephelinite from the Humberg, Kaiserstuhl (Baden, Germany). The suite is dominated by clinopyroxene and olivine. Deformation features in olivine from the clinopyroxene–olivine aggregates, as well as the chemical composition of silicate melt inclusions in the clinopyroxene, preclude their crystallization from the host magma. Compared with early formed clinopyroxene phenocrysts that display unequivocal relationships with the host magma (oscillatory zoning, euhedral shape, crystal clots), clinopyroxene macrocrysts and most clinopyroxenes intergrown with deformed olivine show clear similarities in major and trace element composition. This study targets this potential source region material. We present and discuss constraints on the genesis of the host olivine nephelinite magma and the history of formation of its source region via the fragments preserved in the rock. Based on this approach, we show that the process of carbonatite metasomatism of mantle peridotite lithologies defined by the solidus reactions in the system lherzolite–CO2 under uppermost mantle conditions (<20–22 kbar) can yield an appropriate source region for the olivine nephelinite magma, a hypothesis that has previously been suggested by Dunworth & Wilson (1998) to explain the origin of olivine melilitites of the Upper Rhine graben volcanic province.
|
GEOLOGY
|
|---|
The Kaiserstuhl volcanic complex has a long history of geological
investigation (von Dietrich, 1783
; see reviews by Wimmenauer,
1957
; Keller
et al., 1990
; Sigmund, 1996
). It is situated within
the Upper Rhine Graben (Southern Germany,
Fig. 1a) and is one
of numerous manifestations of Tertiary to Quaternary volcanism
within western and central Europe (Wilson & Downes, 1992
;
Wedepohl
et al., 1994
). The main phase of volcanic activity
in the Kaiserstuhl occurred in Miocene times between 18 and
13 Myr ago (Lippolt
et al., 1963
; Schleicher & Keller, 1991
;
Kraml
et al., 1995
). Major rock types emplaced during this period
are olivine nephelinite and basanite (including type-locality
limburgite), tephrite and essexite, phonolite, sövite and
alvikite carbonatite (Wimmenauer, 1957
, 1959
, 1962
, 1963
; Keller,
1984
; Keller
et al., 1990
; see
Fig. 1b). Volumetrically insignificant
carbonate-rich melilite dike rocks (bergalites) are also present.
The olivine nephelinites are regarded as primitive mantle-derived
magmas, whereas limburgites are interpreted as their differentiates
(Keller, 1978
; Keller
et al., 1990
). Whole-rock Nd and Pb isotopic
data indicate disequilibrium between the olivine nephelinites
and carbonatites, whereas their Sr isotopic compositions are
fairly similar (Schleicher
et al., 1990
, 1991
; Schleicher &
Keller, 1991
). The bergalites have isotope characteristics similar
to the carbonatites (Hubberten
et al., 1988
; Schleicher
et al.,
1990
, 1991
).
The olivine nephelinites sometimes contain peridotite xenoliths
(Keller
et al., 1990
, 1997
; Sigmund, 1996
). Most of these are
spinel peridotites; garnet–spinel peridotites are exceptionally
rare. Porphyroclastic textures are not observed, although clear
indications of strain (undulose extinction, cracked zones and
cataclastic zones at the rims of large mineral grains) are common
(Sigmund, 1996
; Keller
et al., 1997
). Geothermobarometric estimates
indicate temperatures of 950–1100°C (Sigmund, 1996
;
Keller
et al., 1997
). The maximum pressure range for spinel-bearing
lithologies, calculated based on Cr contents in spinel, does
not exceed 22 kbar (Sigmund, 1996
; Keller
et al., 1997
). Garnet-bearing
xenoliths yield temperatures of
c. 1000–1050°C at
pressures of 16–24 kbar, depending on the geobarometer
used (Sigmund, 1996
; Keller
et al., 1997
). The Al-in-Opx geobarometer
of Brey & Köhler (1990
) confines this range to
c. 16–18
kbar, or 53–60 km depth, which probably reflects the limit
of garnet stability in mantle peridotites beneath the Kaiserstuhl
(Sigmund, 1996
; Keller
et al., 1997
).
The Moho beneath the Kaiserstuhl has been detected by seismic refraction studies at a depth of only c. 24 km; there are clear indications of a broad mantle upwarp in the region of the Kaiserstuhl corresponding to the Black Forest–Vosges dome in the surface geology (Edel et al., 1975; Koch, 1993; see Fig. 1a). Gravity modelling (Kahle & Werner, 1980) and electromagnetic deep soundings (Haak et al., 1970; Scheelke, 1974; Winter, 1974) also suggest that a mantle upwarp, centred beneath the Kaiserstuhl, underlies the Upper Rhine Graben (Illies, 1970, 1975). However, available seismic tomography data do not allow assignment of this dome-like structure to an actively upwelling upper mantle diapir and reveal a heterogeneous zone in the depth range of 25–50 km ascribed to partial melts that intruded the mantle lithosphere during the middle and late Miocene and have cooled and solidified since then (Glahn & Granet, 1992). This zone shows a rather smooth transition to the upper asthenosphere below 50 km (Glahn & Granet, 1992).
This study is based on a single drill-core sample (70 mm x 53 mm) of olivine nephelinite from the locality of Humberg situated in the western part of the Kaiserstuhl complex (Fig. 1b). This locality is known for the presence of essexite rocks and weathered tephrite lavas and tuffs, formerly mined in several quarries, which are now derelict. Olivine nephelinite has not been reported in surface outcrop. However, it crops out c. 4
km north of the Humberg, in the area of Lützelberg and Limberg, near Sasbach (Fig. 1b). The Lützelberg is the largest and most prominent occurrence of olivine nephelinite rocks in the Kaiserstuhl. It contains abundant spinel peridotite xenoliths (Keller, 1984; Keller et al., 1990). Both the Lützelberg and the Humberg represent paleovolcanic edifices elongated in a NNE–SSW direction, following the orientation of lineaments of the Upper Rhine graben (Fig. 1a and b). The study sample has a fresh and compact appearance. It does not show indications of atmospheric weathering, unlike samples from most local surface outcrops. In terms of chemical composition, it exhibits a remarkable similarity to the Lützelberg olivine nephelinite (Table 1; see Keller et al., 1990). Based on the geological situation (Fig. 1b) and available geochemical data (Table 1), we consider that it may come from the feeder dike system of the Lützelberg olivine nephelinite volcano.
View this table:
[in this window]
[in a new window]
|
Table 1: Major and trace element abundances in the studied olivine nephelinite (H); for comparison, major element abundances in olivine nephelinite from the Lützelberg (L) are also reported
|
|
The drill-core sample investigated in this study was taken from
the collection of alkaline volcanic rocks initially belonged
to Eugen Wegmann and now kept at the University of Neuchâtel,
Switzerland. The archiving of specimens in this collection dates
to 1961.
|
ANALYTICAL TECHNIQUES
|
|---|
Major element compositions of minerals were determined by wavelength-dispersive
analysis using a four-spectrometer CAMECA SX50 electron microprobe
at the Institute of Geological Sciences, University of Bern.
The microprobe was operated at 15 kV. Beam currents were 20
nA for most silicate and oxide minerals and 10 nA for carbonate
minerals. Compositions of olivine, clinopyroxene and spinel
were obtained using spot analyses. For phyllosilicates and carbonate
minerals, the electron beam was rastered over an area up to
10 µm
x 15 µm. Major element compositions of quenched
carbonatite melt and silicate glass were determined by wavelength-dispersive
analysis using the electron microprobe at the University of
Bern, and a five-spectrometer JEOL JXA-8200 electron microprobe
at the Institute of Mineralogy and Petrography, University of
Lausanne. Operating conditions involved a 15 kV accelerating
voltage and a 6–10 nA beam current. The beam was rastered
over an area 10 µm in diameter. Count times were 10 s
for alkalis and silicon and 15–20 s for other elements.
Intensity vs time scans performed under these conditions did
not reveal any appreciable alkali loss. Data were reduced with
the PAP program. Trace element contents of minerals were analysed
using a laser ablation inductively coupled plasma mass spectrometry
(LA-ICPMS) instrument equipped with a 193 nm ArF excimer laser
(Lambda Physik, Germany) and an ELAN 6100 quadrupole ICPMS system
(Perkin Elmer, Canada) at the Institute of Isotope Geochemistry
and Mineral Resources, ETH Zürich (see Günther
et al., 1997
; Heinrich
et al., 2003
). Operating conditions were
similar to those described by Pettke
et al. (2004
). An SRM 610
external standard from NIST was used. Electron microprobe data
were used for internal standardization. Whole-rock major element
abundances in the host olivine nephelinite were determined using
a Philips PW2400 X-ray fluorescence (XRF) spectrometer (Institut
de Minéralogie et Pétrographie, University of
Fribourg). Trace element abundances were determined on flux-free
glass pills using the same LA-ICPMS technique as for minerals.
Additionally, some trace elements were analysed by XRF at the
University of Fribourg.
|
BULK CHEMICAL COMPOSITION AND CLASSIFICATION OF THE OLIVINE NEPHELINITE
|
|---|
The sample is classified as an olivine nephelinite because of
the frequent presence of olivine macrocrysts, high amount of
modal nepheline and lack of modal feldspar, as suggested by
the IUGS classification for volcanic rocks (Le Maitre
et al.,
2002
). The bulk-rock major and trace element contents are given
in
Table 1. In terms of major elements, the rock studied broadly
compares with nephelinitic rocks from other localities (Le Bas,
1989
, and references therein). Compared with the average olivine
(mela)nephelinites of Nockolds (1954
) and Le Bas (1987
), it
is characterized by a somewhat elevated MgO content and a depletion
in alkalis, in particular K
2O. The alkali depletion of the studied
specimen is also evident when compared with the Lützelberg
olivine nephelinite, Kaiserstuhl (
Table 1). This is related
to the partial replacement of nepheline by analcime in the rock
matrix. The elevated MgO content probably reflects the presence
of large amounts of olivine macrocrysts. The elevated abundances
of incompatible trace elements generally match those for nephelinitic
rocks from other occurrences (e.g. Wedepohl
et al., 1994
; Furman,
1995
; Späth
et al., 2001
). The relatively low Cr content
indicates the lack of extensive contamination by xenocrysts
derived from depleted mantle lithologies, and the elevated content
of Ni reflects the abundance of olivine in the rock.
|
PETROGRAPHY OF THE OLIVINE NEPHELINITE
|
|---|
The studied rock sample consists of clinopyroxene, olivine and
Cr-spinel–Ti-magnetite macrocrysts set in a fine-grained
matrix. The matrix is composed of olivine microphenocrysts,
clinopyroxene, phlogopite, nepheline, analcime, Ti-magnetite,
ilmenite, apatite and calcite. Dominant phases are clinopyroxene,
nepheline and analcime. Clinopyroxene forms microphenocrysts
(40–50 µm) and microlites (<10–15 µm).
The latter are abundant and define the texture of the matrix.
Nepheline and analcime occur as interstitial minerals. Nepheline
is partly replaced by analcime. Olivine is subordinate and partly
or completely replaced by saponite, serpentine and other secondary
phyllosilicates. Poikilitic, texturally late phlogopite occurs
infrequently. Ti-magnetite and less common ilmenite constitute
the major oxide minerals of the matrix and often form intergrowths.
Rare segregations and veinlets of calcite are present. Representative
microprobe analyses of the matrix minerals are provided in
Table 2.
|
MACROCRYSTS AND MINERAL AGGREGATES
|
|---|
The studied suite of macrocrysts and mineral aggregates is dominated
by olivine and clinopyroxene. Cr–Al-spinel and titanomagnetite
are minor phases (
c. 5% of all macrocrysts), with titanomagnetite
often forming overgrowths on Cr–Al-spinel. The macrocrysts
and aggregates do not exceed 3–4 mm in size (
Fig. 2a–f).
They account for
c. 15% of the rock. Individual macrocrysts
are more common than aggregates. Core areas of the macrocrysts
show distinct chemical compositions, whereas their rims are
indistinguishable from matrix microphenocrysts.
View larger version (149K):
[in this window]
[in a new window]
[Download PowerPoint slide]
|
Fig. 2. Photomicrographs and backscattered electron image of macrocrysts and mineral aggregates in the studied olivine nephelinite. (a) Olivine (cross-polarized light). Undulose extinction and kink bands are well developed. (b) Clinopyroxene–olivine aggregate 997/274 (plane-polarized light; one olivine grain from a superimposed complementary photomicrograph in cross-polarized light). The inner part of the aggregate is formed of relict Cpx I that is partly replaced by high-Cr Cpx II containing frequent inclusions. Along the outer rim of the aggregate, Cpx II is mantled by a zone of pale brownish Ti–Al-rich clinopyroxene chemically indistinguishable from the matrix clinopyroxene. Olivine is deformed (kink-bands). Black lines indicate locations of microprobe profiles (see Fig. 5a). (c) Backscattered electron image of a Ti-magnetite macrocryst with a Cr–Al-rich spinel core. (d) High-Cr Cpx II, macrocryst 1426 (cross-polarized light). Clinopyroxene domains with different optical orientations are well developed owing to the angular shape of the core area; their propagation inwards along a crack is visible. (e, f) High-Cr and low-Cr Cpx II, macrocrysts 1610 and 3484, respectively (plane-polarized light). Colourless core areas with abundant inclusions are clearly distinguishable, as well as pale brownish Ti–Al-rich mantles.
|
|
Large mantle xenoliths and megacrysts are absent, as well as
clinopyroxene macrocrysts compositionally similar to clinopyroxene
from common peridotite lithologies. Garnet, orthopyroxene, ilmenite,
amphibole and other minerals that can potentially occur as macrocrysts
in alkaline mafic and ultramafic magmas have not been found.
Green-core clinopyroxene, which often forms macrocrysts in alkaline
basalts (e.g. Duda & Schmincke, 1985
; Dobosi & Fodor,
1992
; Pilet
et al., 2002
), is similarly absent.
Olivine and spinel
Olivine is abundant among the macrocrysts (Fig. 2a) and is also a major constituent of clinopyroxene–olivine aggregates (Fig. 2b). In both cases, deformed olivine with kink-bands and undulose extinction is common. Large olivine grains (>1 mm) often contain homogeneous cores (Mg-number 90–91·5, 0·08–0·14 wt % CaO). Towards the contact with the host olivine nephelinite, strong chemical zoning is observed. The Mg-numbers and Ni content decrease, whereas the Mn and Ca contents increase (Table 3).
Rare macrocrysts of Cr–Al-spinel–Ti-magnetite consist
of anhedral cores of Cr–Al-rich, low-Ti spinel rimmed
by a zone of Cr–Al-poor Ti-magnetite (
Fig. 2c;
Table 4).
The cores are usually homogeneous. At the contact between the
core and the rim, strong chemical zoning is observed. Microprobe
profiles of olivine and spinel are available as an Electronic
Appendix at
http://www.petrology.oxfordjournals.org (Figs A1
and A2, respectively).
Clinopyroxene
Clinopyroxenes can be divided in three groups. Their main characteristics
are summarized in
Table 5. Chemical compositions are given in
Tables 6–9. Petrographic and major and trace element geochemical
data are presented in
Figs 2–8. A comprehensive set of
petrographic data and electron microprobe profiles is available
as an Electronic Appendix at
http://www.petrology.oxfordjournals.org (Figs A3–A10, Table A1).
View this table:
[in this window]
[in a new window]
|
Table 6: Representative major element analyses of core areas of clinopyroxene macrocrysts and clinopyroxenes from Cpx–Ol aggregates
|
|
View this table:
[in this window]
[in a new window]
|
Table 7: Representative trace element analyses of clinopyroxene macrocrysts and clinopyroxenes from Cpx–Ol aggregates
|
|
View larger version (136K):
[in this window]
[in a new window]
[Download PowerPoint slide]
|
Fig. 3. Small clinopyroxene macrocryst exhibiting a well-developed domain texture over the entire core area. (a) Photomicrograph (cross-polarized light). (b) Element map obtained by energy-dispersive analysis and showing the distribution of Ti in a fragment of the core. Ti-rich inclusions (white) are late-stage Ti-rich magnetite crystals on the walls of opened fluid and melt-fluid inclusions.
|
|
View larger version (138K):
[in this window]
[in a new window]
[Download PowerPoint slide]
|
Fig. 4. Clinopyroxene phenocrysts (cross-polarized light). (a) Four individual phenocrysts forming a crystal clot. (b) Single phenocryst showing oscillatory zoning. Numbered spots correspond to laser ablation pits (see Table 9). (c, d) Element maps obtained by energy-dispersive analysis and showing the distribution of Al and Ti inside phenocrysts forming the crystal clot in (a). Oscillatory zoning in the distribution of Al and Ti is visible. Location of a microprobe profile in (c) is indicated by the white line with arrow (Fig. 5d, 300 µm profile).
|
|
View larger version (40K):
[in this window]
[in a new window]
[Download PowerPoint slide]
|
Fig. 5. Microprobe profiles illustrating the zoning patterns of clinopyroxenes. (a) Relict Cpx I partly replaced by high-Cr Cpx II in clinopyroxene–olivine aggregate 997/274. (b) High-Cr Cpx II, macrocryst 1610. (c) Low-Cr Cpx II, macrocryst 3484. (d, e) Cpx III, phenocrysts 300 and 585. (f) Relatively small clinopyroxene macrocryst (1317) with a well-developed domain texture. Clinopyroxene grain numbers correspond to the length of the electron microprobe profiles carried out for each of the grains.
|
|
View larger version (64K):
[in this window]
[in a new window]
[Download PowerPoint slide]
|
Fig. 7. Normalized REE and multi-element patterns of clinopyroxene (core-to-rim zoning and microphenocrysts): (a, b) core-to-rim zoning of clinopyroxene macrocrysts; (c, d) core-to-rim zoning of clinopyroxene phenocrysts and composition of clinopyroxene microphenocrysts from the olivine nephelinite matrix. Normalization values as in Fig. 6.
|
|
View larger version (19K):
[in this window]
[in a new window]
[Download PowerPoint slide]
|
Fig. 8. Binary diagrams illustrating the major element variations in clinopyroxene. A gradual transition between high-Cr Cpx II, low-Cr Cpx II and Cpx III is observed. Error bars correspond to 1 ; if not shown, they are smaller than symbol size.
|
|
Cpx I is very rare and forms partially resorbed grains in the
inner parts of some clinopyroxene–olivine aggregates (
Fig. 2b).
Such relict grains are almost devoid of inclusions of other
phases and are characterized by a distinctive chemical composition.
Cpx I is magnesian, enriched in Cr and Na, depleted in Ca and
strongly depleted in Ti (
Tables 5 and
6). It is fairly homogeneous
in terms of its major element composition (
Fig. 5a). In terms
of trace elements, it displays strong light rare earth element
(LREE) enrichment and heavy REE (HREE) depletion (
Table 7),
resulting in an almost ‘straight-line’ appearance
of the REE patterns (
Fig. 6a). In the multi-element patterns,
pronounced negative anomalies of Nb, Zr, Hf and Ti are present
(
Fig. 6b).
Cpx II is common and clearly dominates the studied assemblage. It forms macrocrysts and occurs in clinopyroxene–olivine aggregates, partially replacing relict Cpx I when the latter is also present. A typical macrocryst consists of a colorless, anhedral-to-subhedral, resorbed core and a subhedral-to-euhedral pale brownish mantle (Fig. 2d–f). No oscillatory zoning in the core area is observed. The mantle can sometimes show oscillatory zoning. In larger macrocrysts, the core area is well developed and optically homogeneous. Closer to the mantle, clinopyroxene forms microscopic domains with different optical orientation, which penetrate the core and define an intermediate zone between the core and the mantle (Fig. 2d). The domain texture tends to spread along cracks and in the corner parts of the core (Fig. 2d), and is therefore unlikely to be a result of direct magmatic crystallization; it appears most consistent with an origin by fast reaction of the clinopyroxene cores with a melt. Small macrocrysts can be entirely overprinted by this process (Fig. 3a and b). Detailed element maps demonstrate that the domain texture is due to the presence of domains characterized by variable chemical compositions (Fig. 3b; for more detail, see Fig. A4 of the Electronic Appendix).
Cpx II from clinopyroxene–olivine aggregates and from the core areas of the macrocrysts are similar. However, the domain texture and outer mantle typical of single Cpx II macrocrysts are not developed in direct contact of Cpx II with olivine, although they are present at its contact with the host olivine nephelinite.
The clinopyroxene cores contain abundant inclusions. Inclusions of fluid, melt and carbonate phases dominate; olivine is common but mostly replaced by secondary minerals (Fig. 2d–f). Spinel may be present. The mantle hosts an entirely different set of inclusions (tiny needles of rutile, Ti-magnetite).
The core areas of Cpx II are Ca-rich, relatively magnesian, depleted in Na and characterized by low to moderate Al contents (Tables 5 and 6). Cpx II displays two modes in Cr content corresponding to high-Cr and low-Cr sub-types (Table 6). High-Cr Cpx II is typical of clinopyroxene–olivine aggregates (Figs 2b and 5a) and also occurs among the macrocrysts (Figs 2e and 5b). It displays a wide range of Ti contents (0·1–1·2 wt %). Low-Cr Cpx II (Figs 2f and 5c) accounts for c. 60–70% of Cpx II macrocrysts, but is absent in clinopyroxene–olivine aggregates. This clinopyroxene is relatively rich in Ti (0·9–1·4 wt %). Between high- and low-Cr Cpx, transitional varieties occur (Table 6). The core areas of large Cpx II grains are more or less homogeneous in terms of chemical composition (Fig. 5a–c). Small grains show a significant scatter of chemical compositions in microprobe traverses, as a result of the development of the domain texture (Fig. 5f). Closer to the core–rim boundary, positive Cr anomalies and often some increase in Mg-number are observed, followed by strong chemical zoning towards the composition of matrix clinopyroxene within the rim zone. The Al and Ti contents increase, whereas the Mg-number decreases (Fig. 5a–c).
Cpx II are enriched in LREE (Table 7, Fig. 6c and e). Convex REE patterns are present (Fig. 6c and e). The multi-element patterns of the core parts of high-Cr Cpx II display variable negative anomalies of Zr, Hf and Ti (Fig. 6d). A considerable scatter in the trace element abundances over the whole set of the analysed grains of high-Cr Cpx II is observed (Fig. 6c and d). The core areas of low-Cr Cpx II show more uniform trace element patterns that are less depleted in Zr and Hf; the Ti depletion disappears (Fig. 6f).
Complementary to the major element zoning data, Cpx II show strong trace element zoning within the outer rim, resulting in enrichment in almost all incompatible trace elements rimwards (Table 7, Fig. 7a and b). Within the core, no significant variations between individual LA-ICPMS analyses are observed. In the intermediate zones in contact with the domain texture and corresponding to the Cr-peaks in the electron microprobe profiles (e.g. Fig. 5c), the trace element patterns are only marginally re-equilibrated, as shown by convex REE distributions and somewhat reduced Zr and Hf anomalies.
Cpx III represents early phenocrysts occasionally forming clots. Most of these phenocrysts consist of an euhedral core and a well-developed, broad mantle grading into the core with no sharp divide but, unlike the core, showing oscillatory zoning (Fig. 4a–d). Some phenocrysts do not have individualized core areas and show oscillatory zoning in the innermost part. No inclusions are observed either in the core or the mantle, except Ti-magnetite and tiny rutile needles within the outermost rim or along cracks. The outermost rim of Cpx III is indistinguishable from that of Cpx II.
The core areas of Cpx III are Ca-rich, relatively depleted in Na and characterized by low to moderate alumina contents (Table 8). They are depleted in Ti and often enriched in Cr compared with the intermediate and outer rims (Fig. 5d). The core areas of Cpx III display strong REE enrichment and uniform, well-defined convex REE patterns (Table 9, Fig. 6g). The multi-element patterns show a marked depletion in Sr that is not found in other clinopyroxene types. Weak negative Zr and positive Hf anomalies and the absence of any Ti anomaly are observed (Fig. 6h).
The core-to-rim trace element patterns of Cpx III phenocrysts and those of relatively large clinopyroxene microphenocrysts (30 µm x 80 µm) from the rock matrix are shown in Fig. 7c and d. The phenocrysts exhibit enrichment in almost all incompatible elements rimwards. The microphenocrysts display a large scatter in trace element abundances, encompassing compositions typical of core areas of large phenocrysts as well as those characteristic of the rim zones of all clinopyroxenes.
Geochemical relationships between different clinopyroxene types
In the following discussion, all clinopyroxenes are considered in the context of the chemical composition of their core areas. Their averaged major element compositions are plotted in Fig. 8. Relict Cpx I form a distinct group. They could be compared with clinopyroxene from depleted mantle rocks. Cpx I grain 997/274 with 1·5 wt % Cr2O3 is one of the most Cr-rich clinopyroxenes ever found in mantle xenoliths and macrocrysts from the Kaiserstuhl (see Keller, 1977, 1984; Keller et al., 1990; Sigmund, 1996). In terms of major elements, Na and Ca in particular, there is no continuum between such relicts and Cpx II (Figs 5a and 7). Between high- and low-Cr Cpx II and Cpx III, the transition in major elements except for Cr is gradual (Fig. 8, Tables 6 and 8). The Cr contents decrease rather sharply from high- to low-Cr Cpx II, and then increase again in the core parts of early Cpx III phenocrysts (Fig. 8, Tables 6 and 8).
In terms of most trace elements, the transition between the clinopyroxene types is gradual. Trace element patterns of the core areas of high-Cr Cpx II retain many features typical of relict Cpx I. For instance, relict clinopyroxene 997/274 is partially replaced by high-Cr clinopyroxene 997, with a sharp boundary between them being observed both optically (Fig. 2b) and in the electron microprobe profiles (Fig. 5a). Their trace element patterns are similar (compare patterns 997/274 and 997 in Fig. 6a–d). However, in general, the ‘straight-line’ REE patterns of Cpx I grade into the convex patterns in Cpx II (Fig. 6c). The negative anomalies of Zr and Hf are less pronounced than in Cpx I, and the negative anomaly of Ti in many cases disappears (Fig. 6d). The core areas of low-Cr Cpx II display trace element patterns that are intermediate between those of high-Cr Cpx II and Cpx III phenocrysts. The REE elements are strongly fractionated and sometimes show convex patterns (Fig. 6e). The negative Zr and Hf anomalies are less expressed than in high-Cr Cpx II (Fig. 6f). The Ti depletion disappears (Fig. 6f). Cpx III is characterized by well-defined convex REE patterns (Fig. 6g). Compared with Cpx II, the middle REE (MREE) to HREE, Y, Zr, Hf and Ti are enriched (Fig. 6h). The only discontinuity between the trace element compositions of different clinopyroxene types is the depletion of Cpx III in Sr (Fig. 6h).
All clinopyroxenes exhibit strong negative Rb, K and Ba anomalies that are consistent with the incompatibility of these large cations in the clinopyroxene lattice. The negative anomalies of Rb and K are matched in the multi-element pattern of the host olivine nephelinite (Fig. 6).
Inclusions in clinopyroxene
The core areas of Cpx II contain frequent solid, melt and fluid inclusions (Fig. 2b, d–f). The solid inclusions are represented by olivine, carbonate minerals, spinel, apatite and graphite (Fig. 9a–h). Olivine and particularly carbonate minerals are common. Apatite and graphite are rare; they occur in some multi-phase melt–fluid inclusions. The melt inclusions are represented by quenched carbonatite melt and alkali aluminosilicate glass.
View larger version (97K):
[in this window]
[in a new window]
[Download PowerPoint slide]
|
Fig. 9. Backscattered (a–f, h) and secondary (g) electron images of inclusions in the core areas of clinopyroxene grains. (a) Calcite, Sr-aragonite, glass and fluid inclusions, macrocryst 3484. (b) Olivine, Cr–Al-spinel and alkali aluminosilicate glass II, macrocryst 1932. (c) Multi-phase inclusion of Mg–Mn-calcite, olivine and alkali aluminosilicate glass II, macrocryst 1774. (d) Multi-phase inclusion of Mg–Mn-calcite and alkali aluminosilicate glass II, macrocryst 3484. (e) Inclusion of quenched carbonatite melt, macrocryst 2308. (f) Multi-phase inclusion containing calcite, quenched carbonatite melt and alkali silicate glass I, macrocryst 1827w. (g) Opened fluid–melt inclusion with Ti–Al-rich magnetite on the wall containing alkali aluminosilicate glass II, macrocryst 3484. (h) Opened fluid–melt inclusion with graphite and alkali aluminosilicate glass II, macrocryst 3484.
|
|
Olivine forms anhedral to subhedral inclusions ranging from
20 to 150 µm in size. They are often associated with calcite
and, in the case of smaller inclusions, with alkali aluminosilicate
glass and fluid (
Fig. 9b and c). Large olivine inclusions are
replaced by saponite, serpentine and calcite, whereas small
ones (<30 µm) can often be found unaltered. The chemical
composition of such inclusions approaches that of the core areas
of large olivine macrocrysts.
Carbonate minerals are represented by calcite and Sr-aragonite forming small (normally <35–40, rarely up to 120 µm) drop-like inclusions in the core parts of clinopyroxene macrocrysts (Fig. 9a). Less typically, they occur as thin veinlets in the latter. Calcite may also be present in complex multi-phase inclusions where it is associated with alkali aluminosilicate glass and/or olivine (Fig. 9c and d). Two types of calcite can be distinguished. More commonly, pure calcite forms drop-like inclusions and, less frequently, veinlets. Mg–Mn-rich calcite occurs only in small (<20–35 µm) isolated inclusions. Sr-aragonite forms individual inclusions, thin veinlets and incrustations along cracks. Larger inclusions of Sr-aragonite are often inhomogeneous, made of domains with variable Sr contents. The major and trace element compositions of calcite and Sr-aragonite are given in Table 11. Pure and Mg–Mn-rich calcite are depleted in Sr (SrO <0·5 wt %), whereas Sr-aragonite is characterized by very low contents of Mn, Fe and Mg (Table 10). The chemical composition of the Mg–Mn-rich calcite inclusions is highly variable in both single macrocrysts and over the whole set of the studied macrocrysts. Both very high (9·0 wt %) and relatively low (1·45 wt %) MgO contents are observed in nearby inclusions. Both types of calcite, as well as Sr-aragonite, have very low abundances of REE and high field strength elements (HFSE). Normally, the total REE abundance does not exceed 2–3 µg/g (Table 11). Only in heterogeneous inclusions where calcite is in contact with secondary phyllosilicates replacing olivine, may the 
REE reach 10–11 µ g/g.
Spinel forms tiny (<7–8 µm) euhedral and subhedral
inclusions in both high- and low-Cr Cpx II. The inclusions are
commonly associated with alkali aluminosilicate melt, although
individual inclusions are also present (
Fig. 9b). Spinel inclusions
are much more common in high-Cr Cpx II than in low-Cr Cpx II.
The composition of the inclusions ranges from Cr–Al-rich
spinel (picotite) in high-Cr Cpx II to very aluminous low-Cr
spinel (hercynite) in low-Cr Cpx II (
Table 11).
Inclusions (up to 20–30 µm in size) of quenched carbonatite melts rarely occur in the studied clinopyroxenes. In the core parts of two Cpx II macrocrysts, we found seven unaltered inclusions that can be referred to as quenched carbonatite melt based on their texture (Fig. 9e and f) and major element composition (Fig. 10; Table 11). Their chemical compositions correspond to silicic, calcic carbonatite melts with relatively low abundances of MgO and FeO and alkalis (Table 11). Some inclusions show significant inhomogeneity as a result of the segregation of silicate and carbonate phases in the quenched aggregate (Fig. 9f; to a lesser degree, present in Fig. 9e).
Irregular or rounded inclusions of alkali aluminosilicate glass
mostly do not exceed 20 µm in size (
Fig. 9a–d and
f–h). When imaged using a scanning electron microscope
(magnification
x 2500–3000), small inclusions appear homogeneous.
Larger inclusions (20–25 µm) often contain microcrystals
of nepheline. Two glass types can be distinguished. Glass I
was detected in three inclusions, two of them also containing
quenched carbonatite melt and crystalline calcite (
Fig. 9f).
It is tephriphonolitic in composition (
Table 11; also see
Fig. 10).
Glass II is far more common. It often occurs in heterogeneous,
often fluid-bearing, inclusions that do not contain quenched
carbonatite melt (
Fig. 9a–d, g and h). Large variations
in the modal proportion of melt in the fluid-bearing inclusions
suggest that both fluid and melt represent individual phases
during the entrapment of the inclusions; that is, the entrapment
was heterogeneous. Glass II is often associated with Mg–Mn-rich
calcite and Mg-rich olivine (
Fig. 9c and d). Rarely, Cr–Al-spinel
and apatite are present. Secondary phases observed in some inclusions
are Ti-phlogopite and Ti-magnetite; the former typically develops
at the border between the glass and the fluid, whereas the latter
forms crystal precipitates on the walls of the fluid cavities.
Primary phlogopite and amphibole are absent. In terms of major
elements, glass II closely matches the theoretical composition
of persilicic nepheline (A. Ulianov
et al., unpublished data).
Its chemical composition is characterized by a low Fe content
and extreme depletion in Ti, Mg and, in most cases, Ca (
Fig. 10;
Table 11). These features might reflect the nepheline-like structure
of the corresponding melt, which is poorly compatible with such
cations (e.g. Simakin
et al., 2005
). Glass II is rich in incompatible
elements (Sr, Cs, Rb, Nb, Ta, Zr, LREE; A. Ulianov
et al., unpublished
data).
Glass II chemical compositions appear primary. Except for the presence of nepheline microcrystals in large inclusions, no convincing evidence for post-entrapment chemical modification of the glass was found. Detailed element maps show that the clinopyroxene host in contact with the glass does not change its chemical composition, whereas olivine crystals associated with the glass in multi-phase inclusions are highly magnesian.
Fluid inclusions up to 30–35 µm in size are common. They are rounded or, less frequently, show negative crystal shapes. Microscopically, three phases can be distinguished in the inclusions: a gas, a liquid, and an oxide phase with high reflection index. The last is a Ti-rich magnetite, clearly visualized by SEM (Fig. 9g). The study of opened inclusions by SEM has also revealed a close association of the fluid with glass II, with more than half of all inclusions either showing a contact of the fluid cavity with ‘massive’ glass or containing thin glass coatings on the walls (Fig. 9g and h). Several inclusions contain apatite. One (possibly two) inclusions contain graphite (Fig. 9h).
|
DISCUSSION
|
|---|
The relationship between macrocrysts and the host magma
One of the prime aims of this study is to show that the studied
suite of macrocrysts and mineral aggregates represents fragments
of the mantle source region entrained by the host magma during
its formation. Comparing the major and trace element compositions
of the clinopyroxene macrocrysts with those of the ‘true’
phenocrysts of clinopyroxene (with oscillatory zoning) seems
to be the best way to evaluate whether or not the macrocrysts
are related to the host magma. Other options include comparing
the trace element characteristics of the macrocrysts with those
of the host rock, provided that mineral–melt partition
coefficients are well known, or showing that the macrocrysts
are in isotopic equilibrium with the host rock. The first option
implies that no immiscible liquids or fluids separated from
the host magma
en route to the surface, and both require the
host rock to be fresh. The second option also faces difficulties
because of possible isotope re-equilibration between the host
magma and the macrocrysts.
In general, the major element data show a gradual transition from Cpx II to Cpx III, and the trace element data demonstrate a gradual transition between all Cpx types. Evolutionary changes include systematic enrichment in Ti, Zr and Hf from Cpx I to high- and low-Cr Cpx II and further to Cpx III, accompanied by some depletion in La and Ce resulting in convex REE patterns. The gradual transition in the chemical composition from Cpx II to Cpx III could be consistent with the derivation of Cpx II from the source region of the olivine nephelinite magma that precipitated early phenocrysts of Cpx III.
Formation of the source region: metasomatism and relationship with carbonatite magmatism
Metasomatism: geochemical characteristics of relict Cpx I
It has been widely proposed that mantle metasomatism can be a precursor to alkaline magmatism (e.g. Norry et al., 1980; Wass & Rogers, 1980; Lloyd, 1981; Ionov et al., 1994; Furman, 1995; Späth et al., 2001; Dawson, 2002). The studied Cpx macrocrysts and Ol–Cpx aggregates do not represent fragments derived from ‘pristine’ (i.e. non-metasomatized) mantle rocks. Cpx I is characterized by marked LREE enrichment (12–14 times chondrite), negative anomalies of Zr, Hf and Ti and low Ti/Eu ratios (Fig. 11a and b; Table 7). These features are typical of clinopyroxene inferred to be formed by carbonatite metasomatism of the mantle, as shown by both natural mineral assemblages from xenoliths (e.g. Hauri et al., 1993; Norman, 1998; Powell et al., 2004) and experiments in carbonatite systems (e.g. Klemme et al., 1995; Blundy & Dalton, 2000; Adam & Green, 2001). They are also consistent with the general bulk-rock geochemical characteristics of the metasomatized xenoliths (Yaxley et al., 1991, 1998; Ionov et al., 1993; Rudnick et al., 1993; Powell et al., 2004), and of many carbonatites (e.g. Nelson et al., 1988; Woolley & Kempe, 1989). Moreover, the occurrence of carbonatite melt inclusions found in Cpx II macrocrysts testifies to the presence of a carbonatitic component.
View larger version (46K):
[in this window]
[in a new window]
[Download PowerPoint slide]
|
Fig. 11. (a, b) Normalized trace element patterns of relict Cpx I compared with trace element data for clinopyroxenes from a representative set of uppermost mantle xenoliths from different occurrences worldwide for which carbonatite metasomatism has been inferred (Hauri et al., 1993; Ionov, 1998; Norman, 1998; Coltorti et al., 1999; Ionov et al., 2002; Neumann et al., 2002; Powell et al., 2004). However, we are uncertain whether all of these clinopyroxenes formed in equilibrium with primary carbonatite melts. Some of them are characterized by marked Sr depletion (Hauri et al., 1993; Coltorti et al., 1999) that might be caused by the fractionation of carbonates from primary carbonatite melt in the mantle (e.g. Ionov & Harmer, 2002). Evolved carbonatite melts with elevated silica and alkali contents and even strongly carbonated silicate melts may account for the geochemical characteristics of some of the clinopyroxenes used for comparison.
|
|
Phase relations in the system peridotite–CO2
Chemical reactions relevant to the process of carbonatite metasomatism
in the mantle were investigated experimentally in the model
system CaO–MgO–SiO
2–CO
2 (CMS–CO
2; see
Eggler, 1974
, 1976
, 1978
; Wyllie & Huang, 1976
; Brey
et al., 1983
; White & Wyllie, 1992
; Lee & Wyllie, 2000
;
Lee
et al., 2000
) and in natural peridotite compositions (e.g.
Olafsson & Eggler, 1983
; Wallace & Green, 1988
; Falloon
& Green, 1989
; Dalton & Wood, 1993
; Yaxley & Green,
1998
). Two subsolidus reactions for carbonate–lherzolite
and five solidus curves for lherzolite, wehrlite, harzburgite
and websterite plus CO
2–carbonate in the system peridotite–CO
2 are shown in
Fig. 12a. The reaction Opx + Dol = Cpx + Ol + CO
TOP
ABSTRACT
INTRODUCTION
