Journal of Petrology Advance Access originally published online on February 25, 2005
Journal of Petrology 2005 46(7):1345-1366; doi:10.1093/petrology/egi018
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Crustal Evolution of Island-Arc Ultramafic Magma: Galmoenan Pyroxenite–Dunite Plutonic Complex, Koryak Highland (Far East Russia)
1 MAX-PLANCK-INSTITUT FÜR CHEMIE, ABT. GEOCHEMIE, POSTFACH 3060, 55020 MAINZ, GERMANY
2 VERNADSKII INSTITUTE OF GEOCHEMISTRY AND ANALYTICAL CHEMISTRY, RUSSIAN ACADEMY OF SCIENCES, KOSYGIN STR. 19, 117975, MOSCOW, RUSSIA
3 INSTITUTE OF ORE DEPOSITS GEOLOGY, PETROGRAPHY, MINERALOGY AND GEOCHEMISTRY, STAROMONETNYI, 35, MOSCOW, 109017 RUSSIA
4 SCHOOL OF EARTH SCIENCES AND CENTRE FOR ORE DEPOSIT RESEARCH, UNIVERSITY OF TASMANIA, HOBART, TAS. 7001, AUSTRALIA
5 INSTITUTE OF PRECAMBRIAN GEOLOGY AND GEOCHRONOLOGY (IGGD RAS), MAKAROV NAB. 2, ST. PETERSBURG, 199034, RUSSIA
RECEIVED OCTOBER 17, 2003; ACCEPTED JANUARY 26, 2005
 |
ABSTRACT |
|---|
TOP
ABSTRACT
INTRODUCTIONAlaskan-type platinum-bearing plutons and potassium-enriched mafic to ultramafic volcanic rocks are temporally and spatially associated within the Late Cretaceous–Paleocene Achaivayam–Valaginskii intra-oceanic palaeo-arc system, allochthonously present in the Koryak Highland and Kamchatka Peninsula (Far East Russia). The compositions of the parental magmas to the Alaskan-type complexes are estimated using the Galmoenan plutonic complex as an example. This complex, composed of dunites, pyroxenites and minor gabbros, is the largest (
20 km3) in the system and the best studied owing to associated platinum placer deposits. The compositions of the principal mineral phases in the Galmoenan intrusive rocks [olivine (Fo79–92), clinopyroxene (1–3·5 wt % Al2O3, 0·1–0·5 wt % TiO2), and Cr-spinel (5–15 wt % Al2O3 and 0·3–0·7 wt % TiO2)] are typical of liquidus assemblages in primitive island-arc magmas in intra-oceanic settings, and closely resemble the mineral compositions in the Achaivayam–Valaginskii ultramafic volcanic rocks. The temporal and spatial association of intrusive and extrusive units, and the similarity of their mineral compositions, suggest that both suites were formed from similar parental magmas. The composition of the parental magma for the Galmoenan plutonic rocks is estimated using previously reported data for the Achaivayam–Valaginskii ultramafic volcanic rocks and phenocryst-hosted melt inclusions. Quantitative simulation of crystallization of the parental magma in the Galmoenan magma chamber shows that the compositions of the cumulate units are best modelled by fractional crystallization with periodic magma replenishment. The model calculations reproduce well the observed mineral assemblages and the trace element abundances in clinopyroxene. Based upon the estimated composition of the parental magmas and their mantle source, we consider that fluxing of a highly refractory mantle wedge (similar to the source of boninites) by chlorine-rich aqueous fluids is primarily responsible for both high degrees of partial melting and the geochemical characteristics of the magmas, including their enrichment in platinum-group elements. KEY WORDS: subduction; platinum-group elements; clinopyroxene; trace elements; fractional crystallization; Alaskan-type plutons
|
INTRODUCTION |
|---|
Alaskan-type plutonic complexes were recognized in SE Alaska as a distinct class of intrusions in the 1960s, based on their tectonic setting, composition, internal structure and mineralization (Noble & Taylor, 1960
; Taylor & Noble, 1960). In addition to the type locality, similar intrusions have been documented in convergent plate margin settings worldwide—in British Columbia (e.g. Nixon et al., 1990), Oregon (Gray et al., 1986), Venezuela (Murray, 1972), New South Wales, Australia (Johan et al., 1989), Southland, New Zealand (Spandler et al., 2000, 2003), the Ural Mountains (Taylor, 1967) and the Koryak Highland, Northern Kamchatka, Russia (Batanova, 1991; Batanova & Astrakhantsev, 1992, 1994). Alaskan-type plutons are often concentrically zoned and composed (from core to margin) of dunite, wehrlite, olivine clinopyroxenite and hornblende clinopyroxenite, and minor gabbro (Himmelberg & Loney, 1995). The most important characteristic of these intrusive ultramafic rocks is a primary magmatic enrichment in minerals containing platinum-group elements (PGE) with the Pt–Fe alloys (isoferroplatinum) dominating the assemblage (e.g. Slansky et al., 1991; Cabri et al., 1996).
It has been proposed that Alaskan-type intrusions are cumulates derived from the crystallization of hydrous mafic and ultramafic magmas (e.g. Murray, 1972; Irvine, 1974; Himmelberg & Loney, 1995; Helmy & El Mahallawi, 2003). However, the composition and crystallization history of such magmas remain controversial. Irvine (1973, 1974) argued for fractionation of alkaline ultramafic parental magmas (ankaramite) based on the study of subduction-related, K-enriched magnesian rocks from the Juneau area (Alaska). Later research established the origin of the parental magmas of the Alaskan intrusives in a supra-subduction zone tectonic setting, and suggested a subalkaline, orthopyroxene-normative composition (e.g. Nixon et al., 1990; Kepezhinskas et al., 1993; Sha, 1995). In a recent review, Himmelberg & Loney (1995) advocated fractional crystallization of a H2O-saturated subalkaline island-arc basalt magma for the origin of Alaskan-type complexes based on comparison of their chemical and mineral compositions and the compositions of clinopyroxenite and gabbro xenoliths in the Aleutian arc lavas.
The true nature of the magmas parental to Alaskan-type intrusions and their crystallization paths remains a petrological challenge, particularly in relation to the specific enrichment of these magmas in PGE. Several approaches are used to estimate the compositions of magmas parental to mafic–ultramafic intrusives. Some qualitative methods use the compositions of cumulate minerals, their relative abundances and the order of crystallization (e.g. Thy et al., 1989). Other semi-quantitative methods are based on the use of compositions of chilled margins and aphyric sills and dykes (Hoover, 1989; Greenwood et al., 1990), the compositions of individual minerals and crystal–melt partition coefficients (e.g. Ross & Elthon, 1993), and melt inclusion studies (Batanova et al., 1996; Spandler et al., 2000). An alternative approach to constrain the parental magma compositions of the plutonic rocks is the study of spatially and temporally associated volcanic units, as suggested by Irvine (1973). Based on age and Sr–Nd isotope data, Tistl et al. (1994) proposed that spatially associated Alaskan-type ultramafic intrusions and high-K primitive basalts in the Western Cordillera of Colombia are genetically related. However, the chemical and physical parameters of crystal fractionation and accumulation in this case were not quantitatively constrained.
In this study we estimate the compositions of the magmas parental to the Alaskan-type complexes that are widespread in the Kamchatka–Koryak Highland region (NE Russia) using the Galmoenan plutonic complex as an example. This intrusive complex is typical of a regional palaeo island-arc system, formed in the Late Cretaceous–Paleocene (Astrakhantsev et al., 1991; Batanova & Astrakhantsev, 1992, 1994; Kepezhinskas et al., 1993), and is well studied because of associated platinum placer deposits (Mochalov et al., 2002; Nazimova et al., 2003). An important component of the same magmatic palaeosystem is primitive ultramafic volcanic rocks (pillow lavas and dykes), which are highly enriched in large-ion lithophile elements (LILE; Sobolev et al., 1989; Kamenetsky et al., 1995). The occurrence of an extrusive series in close temporal and spatial association with the intrusive complexes is used in this study for the quantitative reconstruction of the composition of the Galmoenan parental magmas. The COMAGMAT 3.5 phase equilibria model (Ariskin et al., 1993; Ariskin, 1999) is applied to constrain the crystallization sequence and mechanisms responsible for the origin of the substantial volumes of ultramafic cumulates, including dunites.
|
BACKGROUND TO THE GEOLOGY OF THE AREA |
|---|
The Kamchatka Peninsula and the Koryak Highland form part of the Mesozoic–Cenozoic active continental margin of NE Asia (Fig. 1), where different continental and subduction-related magmatic, sedimentary and metamorphic terranes have been thrust over Late Crataceous–Eocene continental-derived turbidite formations (Konstantinovskaia, 2001
, and references therein). One of these terranes, the Achaivayam–Valaginskii Terrane (AVT), includes a number of suites formed in an intra-oceanic arc setting in the Late Cretaceous–Paleocene, and accreted into their current position during the Paleocene–Miocene (Zinkevich & Tsukanov, 1992; Konstantinovskaia, 2001; Kovalenko, 2001). The basement of the AVT is composed of Albian to Late Campanian, interbedded mid-ocean ridge basalt (MORB)-type volcanic rocks, radiolarian cherts, red jaspers and hyaloclastites, covered by volcaniclastic rocks and lavas of Late Campanian to Paleocene age with strong island-arc geochemical affinities (Astrakhantsev et al., 1987; Konstantinovskaia et al., 1993; Batanova & Astrakhantsev, 1994). Several suites, varying in composition from tholeiitic to high-K (shoshonitic) basalts, have been recognized among the AVT volcanics (Fedorov, 1990; Magakyan et al., 1993). Ultramafic members of the high-K series (pillow lavas, dykes, sills, and tuffs) are present in several localities in eastern Kamchatka and the southern Koryak Highland (Sobolev et al., 1989; Fedorov, 1990; Zinkevich et al., 1991; Kamenetsky et al., 1995), and are considered to be the most primitive representatives of intra-oceanic island-arc high-K magmas (Kamenetsky et al., 1995).
|
Spatially and temporally associated with the volcanic complexes within the AVT is a belt of more than 20 intrusive bodies, interpreted to have been emplaced in island-arc crust at different levels (Astrakhantsev et al., 1991
; Batanova, 1991; Batanova & Astrakhantsev, 1992, 1994; Kepezhinskas et al., 1993). The Galmoenan complex (14 km x 3 km), located in the southern Koryak Highland (Figs 1 and 2) is one of the largest and best exposed of these intrusions. The host rocks are Albian–Campanian silicified tuffs, argillites and black cherts formed in an ocean-floor environment transitional to an island-arc stage. The intrusive rocks were emplaced before the folding and thrusting of the country rocks, and were significantly affected by the latter processes, resulting in plastic and brittle deformation and formation of a serpentinite mélange. The intrusion is concentrically zoned with dunite in the core, followed by wehrlite, olivine clinopyroxenite, hornblende clinopyroxenite, gabbro, and finally orthopyroxene-bearing quartz hornfels at the contact with the country rocks (Fig. 2; Astrakhantsev et al., 1991; Batanova, 1991; Batanova & Astrakhantsev, 1994). The main lithological units of the Galmoenan complex [which are structurally inverted (Astrakhantsev et al., 1991)] are listed in Table 1 and shown in Fig. 2. A dunite body forms the upper part of the ultramafic section in the present-day structure (sections A–B and C–D in Fig. 2); clinopyroxenite commonly underlies the dunite. Small gabbronorite bodies (clinopyroxenite–hornblendite–gabbro unit, see Fig. 2) form the outermost zones of the complex. Several relatively thick (1–5 m) mafic dykes were found within the dunite, clinopyroxenite and country rocks.
|
|
|
STRUCTURE AND PETROGRAPHY OF THE GALMOENAN COMPLEX |
|---|
Dunite unit
A thick unit of dunites (650 m) dominates the Galmoenan plutonic complex. The dunite is composed of olivine and Cr-spinel, and characterized by a tectonic fabric produced during high-temperature plastic deformation and recrystallization, coeval with emplacement and later thrusting of the whole complex. Most common is a subhorizontal planar fabric with pronounced foliation of olivine porphyroclasts and relic lenses and layers of coarse-grained polygonal and pegmatoid varieties. The planar structures are sub-parallel to the dunite–clinopyroxenite boundary. A range of microstructures (porphyroclastic, granoblastic and mosaic) reflect different degrees of deformation and recrystallization (e.g. Fig. 3a). Veins of olivine clinopyroxenite (1–10 cm thick) are present in the outer part of the dunite unit, close to the contact with the clinopyroxenite.
|
Dunite–clinopyroxenite transition
The boundary between the dunite and clinopyroxenite units is visually sharp, although clinopyroxene starts to appear in the dunite before the contact. The amount of clinopyroxene gradually increases over a distance of 100–200 m across the contact. Olivine clinopyroxenite veins (up to 1·5 m thick) are abundant in this transition zone. These veins are surrounded by wehrlite and clinopyroxene-bearing dunite (Fig. 3b). Also, several lens-like olivine clinopyroxenite layers up to 10 m thick occur inside the dunite, above the main dunite–clinopyroxenite transition zone. The transitional zone in the northern part of the complex is characterized by dunite–wehrlite–clinopyroxenite inter-layering (up to 100 m thick). The wehrlite of the dunite–clinopyroxenite transition zone contains abundant spinel, whereas the wehrlite within the clinopyroxenite unit is typically spinel-free. The high-temperature ductile deformation observed in the dunite unit is also characteristic of the rocks in the transition zone, represented by a porphyroclastic textures (Fig. 3c).
Clinopyroxenite unit
The rocks of the clinopyroxenite unit (up to 450 m thick) show no regular stratification, either compositionally or texturally. Bimineralic, coarse- to medium-grained olivine–clinopyroxene rocks are dominant; subordinate magnetite-bearing lithologies with sideronitic textures (xenomorphic magnetite in the matrix cementing clinopyroxene) form layers (up to 50 m thick) throughout the clinopyroxenite unit. There is only a weak planar fabric but local modal layering conforms to the general dunite–clinopyroxenite boundary. Cumulate hypidiomorphic–granular textures are often preserved in the rocks (Fig. 3d), and intracrystalline deformation is developed in mineral grains (sectoral extinction of clinopyroxene, kink-banding in olivine). Clinopyroxene often contains inclusions of phlogopite. Several micro-veins of dunite (up to 5 cm thick) are found within the magnetite-bearing olivine clinopyroxenite.
Clinopyroxenite–hornblendite–gabbro unit
A clinopyroxenite–hornblendite–gabbro unit (up to 500–600 m thick) is located in the outermost zone of the complex. This unit displays a large variety of petrographic and structural features. The clinopyroxenites include hornblende- (Fig. 3e), plagioclase- and magnetite-bearing types, with up to 25% of magnetite in the latter. Close to the contact with the country rocks, the unit is composed of plagioclase- and phlogopite-bearing (± K-feldspar) olivine clinopyroxenite, olivine gabbro, gabbronorite and monzogabbro. These lithologies are characterized by porphyritic textures (Fig. 3f). The contact rocks are aphyric microgabbronorites, micronorites (chilled margin) and orthopyroxene-bearing quartz hornfels (metamorphosed wall rock).
|
AVT ULTRAMAFIC VOLCANIC ROCKS |
|---|
Ultramafic volcanic rocks have been previously described from the Valaginskii Range (VR), Tumrok Range (TR) and Natalia Bay (NB) (e.g. Markovsky & Rotman, 1971
; Sobolev et al., 1989; Fedorov, 1990; Kamenetsky et al., 1995) (Fig. 1). These rocks are strongly porphyritic picrites (up to 40–75 vol. % of olivine and up to 20 vol. % of clinopyroxene phenocrysts) with a glassy to microcrystalline matrix. The phenocrysts commonly contain inclusions of Cr-spinel and melt. The glassy matrix has a spinifex texture, and is composed of altered glass, clinopyroxene, amphibole, phlogopite and titanomagnetite. Previously studied samples of picrite from TR (DAN-51, DAN-57) and VR (KB-189) and picrobasalt from NB (F815-9) were chosen for comparison with the Galmoenan olivine–clinopyroxene cumulates. |
MINERAL COMPOSITIONS OF THE GALMOENAN COMPLEX AND AVT ULTRAMAFIC VOLCANICS |
|---|
Olivine
Olivine is present in all ultramafic and most mafic suites of the Galmoenan complex. The olivine is most magnesian in the dunite (Fo90–92, Table 2; see Appendix for analytical techniques) and becomes progressively less forsteritic in the rocks surrounding the dunite core: Fo89 in olivine clinopyroxenite and clinopyroxenite veins in dunite, Fo88–84 in clinopyroxenite, and Fo79 in plagioclase clinopyroxenite and olivine gabbro. Olivine phenocrysts in the AVT ultramafic volcanic rocks range in compositions from Fo95 to Fo82, but most samples are Fo92–90; several different populations of olivine, varying in CaO content at a given Fo content, have been described (Kamenetsky et al., 1995
).
|
Spinel
Spinel group minerals in the Galmoenan rocks are represented by Cr-spinel (dunite), Cr-magnetite (wehrlite, olivine clinopyroxenite) and magnetite (clinopyroxenite–hornblendite–gabbro unit). Cr-spinel is characterized by high Cr/(Cr + Al) and Fe3+/(Fe3+ + Cr + Al) and low Mg-number [100 x Mg/(Mg + Fe2+)], TiO2 and Al2O3 (Table 3). Such compositions are typical of spinel in the Alaskan-type complexes and primitive subduction-related volcanic rocks (Figs 4 and 5). Spinel inclusions in olivine from the AVT ultramafic–mafic volcanic rocks are of similar composition, but are more magnesian owing to the higher temperatures of final equilibration with olivine (Figs 4 and 5).
|
|
|
Clinopyroxene
Major elements
Clinopyroxene in all Galmoenan plutonic units is diopsidic in composition (Table 4, Fig. 6) with a high wollastonite component (45–50%), and relatively low Al2O3 (1–3·5 wt %) and TiO2 (0·1–0·5 wt %) (Fig. 7a and b). Similar compositions are reported for clinopyroxene from the studied picrites KB-189, DAN-51, DAN-57 (Fig. 7, Table 5), and from the AVT ultramafic volcanic rocks in general. The Mg-number decreases progressively from 91 in clinopyroxene in the ultramafic cumulates to 78 in clinopyroxene in the mafic rocks (Fig. 7). The increase in Al2O3 and TiO2 with decreasing Mg-number of clinopyroxene (Fig. 7) reflects the crystallization trend in H2O-bearing subduction-related magmas (Conrad & Kay, 1984
; DeBari & Coleman, 1989; Loucks, 1990), and is also typical of Alaskan-type complexes (Himmelberg & Loney, 1995) and of the AVT ultramafic volcanic rocks (Kamenetsky et al., 1995).
|
|
|
|
Clinopyroxene in the ultramafic units is usually homogeneous, whereas the clinopyroxene phenocrysts in the gabbros and dykes are often zoned. Two generations of clinopyroxene are found in the plagioclase-bearing rocks. The earlier generation comprises large zoned phenocrysts (up to 1 cm) with compositions similar to those in the clinopyroxenite unit (e.g. Mg-number 87–88 mol %). The later generation of smaller grains (0·5–1 mm) is identical in composition (Mg-number 72–80 mol %) to the rims of the phenocrysts of the earlier generation (Table 4). Petrographic observations suggest that the cores of phenocrysts formed prior to plagioclase crystallization, whereas the rims of the phenocrysts and later generation of clinopyroxene crystallized with plagioclase.
Trace elements
Clinopyroxenes from different lithological units of the Galmoenan complex have similar C1-chondrite normalized incompatible trace element patterns (Fig. 8a and b). The characteristic feature of the Galmoenan clinopyroxenes is high Sr and low Zr normalized abundances and relatively flat rare earth element (REE) patterns. The trace element abundances (e.g. CeN, NdN, YbN) increase as Mg-number of the clinopyroxene decreases (Table 4, Fig. 9). The exception is Sr, the concentration of which is nearly constant in clinopyroxene from all lithologies (Fig. 8a and b).
|
|
Clinopyroxenes from the olivine clinopyroxenite veins in dunite are characterized by the highest Mg-number and the normalized concentrations of the middle rare earth elements (MREE) and heavy rare earth elements (HREE) similar to those in the clinopyroxenite unit. At relatively constant HREE, La and Ce are highly variable (Fig. 8a). Clinopyroxene in the plagioclase-bearing lithologies is characterized by compositional zoning. Usually the cores of the clinopyroxene phenocrysts have the same trace element patterns as clinopyroxene from the olivine–clinopyroxene cumulates, whereas the rims, as well as grains composing the matrix, are markedly enriched in incompatible elements (Fig. 8b).
Clinopyroxene phenocrysts from the AVT ultramafic volcanic rocks exhibit trace elements patterns closely similar to those observed for clinopyroxene from the Galmoenan plutonic units (Fig. 8c). Clinopyroxenes from the Galmoenan cumulates and the AVT volcanic rocks are well matched in terms of abundances of moderately incompatible elements (Al, Ti, HREE) (Figs 7 and 9), whereas contents of highly incompatible elements (La, Ce, Nd, Sm) are clearly different at a given Mg-number of clinopyroxene. The VR clinopyroxene is most enriched compared with clinopyroxene from TR and NB (consistent with differences in VR and TR whole-rock and melt inclusion compositions; Kamenetsky et al., 1995), whereas the Galmoenan clinopyroxene is compositionally transitional. It is worth noting that clinopyroxene from the olivine clinopyroxenite veins in dunites and later generation (rims of phenocrysts and matrix grains) clinopyroxene in gabbros is as enriched in most incompatible elements as the VR clinopyroxene (Fig. 9).
Hornblende and phlogopite
Representative compositions of hornblende and phlogopite from the Galmoenan complex are given in Table 6. Hornblende ranging in composition from pargasite to edenite is ubiquitous in the clinopyroxenite–hornblendite–gabbro unit. Phlogopite is present in clinopyroxene-bearing dunites of the dunite–clinopyroxenite transition zone as rare interstitial grains (Fig. 3b), and becomes common in the more evolved (less magnesian) rocks, appearing as inclusions in clinopyroxene. In gabbros, phlogopite is an abundant intercumulus phase, surrounding clinopyroxene and plagioclase (Fig. 3f).
|
|
CRYSTALLIZATION MODELLING |
|---|
Parental melt and input parameters
To model the crystallization processes responsible for the formation of the ultramafic–mafic cumulates we need to define the composition of the parental melt. The inferred genetic link between the Galmoenan plutonic units and the AVT ultramafic volcanic rocks suggests that they could be the products of chemically similar parental magmas. As the composition of the Galmoenan clinopyroxene is most similar to the compositions of volcanic clinopyroxene from TR and NB (Fig. 9) the primary melt defined for the TR ultramafic volcanic rocks (Table 7, composition 1; Kamenetsky et al., 1995
) was chosen to represent the parental magma composition of the Galmoenan complex. Accordingly, the Galmoenan parental melt is ultramafic (23·8 wt % MgO), with relatively high H2O (0·6 wt %), Cl (0·14 wt %) and total alkali contents (2·2 wt %). Its trace element composition was estimated using the compositions of melt inclusions trapped in TR primitive olivine phenocrysts (sample DAN-51; Kamenetsky et al., 1995; Table 7). The degree of oxidation of Fe in the melt [(Fe2+/Fe3+)L = 3] was calculated using the Fe2+/Fe3+ of Cr-spinel and the model of Maurel & Maurel (1982).
|
Preliminary modelling using the COMAGMAT algorithm (Ariskin et al., 1993
) shows that the earliest Galmoenan clinopyroxene (Mg-number 90–91) could have crystallized from the model primary melt at a pressure of 3–4 kbar at the given oxidation conditions.
Crystallization models
The ultramafic parental melt (Table 7) is in equilibrium with olivine + Cr-spinel (Kamenetsky et al., 1995). Olivine fractionation was simulated using the PETROLOG 2.1 package (Danyushevsky, 2001) until saturation with clinopyroxene was reached (17% of olivine fractionation leading to high-Mg basaltic melt, Table 7, composition 2). The simulation of further cotectic crystallization can be performed using either the MELTS algorithm (Ghiorso & Sack, 1995) or COMAGMAT (Ariskin et al., 1993). It has been demonstrated (Yang et al, 1996; Slater et al., 2003; Thompson et al., 2003) that both algorithms produce comparable results. However, MELTS tends to overestimate the clinopyroxene crystallization temperature, especially at elevated pressures. On the other hand, COMAGMAT calculates more realistic proportions of cotectic olivine, clinopyroxene and plagioclase, and this explains our preference for COMAGMAT (version 3.5) in the crystallization modelling of the thick (>450 m) olivine–clinopyroxene units of the Galmoenan complex.
Modelling of simple fractional crystallization (Fig. 10) results in a large range of clinopyroxene compositions in the cumulates, but limits crystallization of olivine once clinopyroxene is on the liquidus. Cessation of olivine crystallization is caused by the reaction under given oxidation conditions: olivine + O2 = pyroxene + magnetite (Ariskin, 2003). Therefore, this model does not explain the presence of olivine in the Galmoenan olivine–clinopyroxene cumulate rocks.
|
The second model simulates fractional crystallization (10%) of a high-Mg basalt followed by periodic replenishment of the magma chamber by, and mixing with, small batches (5%) of the parental ultramafic melt (Fig. 11). As expected, the results of this modelling at an early stage of crystallization are not significantly different from simple fractional crystallization (compare Figs 10 and 11). However, after the second cycle of modelled crystallization and replenishment, clinopyroxene follows olivine within every cycle (Fig. 11). This model creates a series of olivine-bearing, clinopyroxene-rich crystallization products with progressively changing Mg-number of olivine (93 to 80) and clinopyroxene (91 to 79). This is consistent with the actual mineral compositions in the Galmoenan cumulates (olivine Fo92 to Fo79, clinopyroxene Mg-number 91 to 80). Also this model reproduces well the advanced stages of crystallization [olivine-bearing plagioclase-rich (gabbroic) rocks; see Fig. 11] and describes best the Galmoenan plutonic crystallization trends.
|
Modelling of the trace element composition of clinopyroxene
A model of fractional crystallization combined with periodic replenishment of new magma was used to simulate the trace element compositions of liquidus clinopyroxene, and thus to test the applicability of the model to the Galmoenan cumulates. We used the clinopyroxene/melt partition coefficients and their temperature dependence as defined by Sobolev et al. (1996)
for an island-arc H2O-bearing basaltic melt and low-Al liquidus clinopyroxene compositions (Table 8). The other commonly used clinopyroxene/melt partition coefficients (e.g. Wood & Blundy, 1997) correspond to more Al-rich clinopyroxene and dry conditions and predict much higher values for HREE.
|
The COMAGMAT algorithm calculates temperatures and modal proportions of liquidus phases for each crystallization increment (1%). This permits calculation of the bulk crystal/melt partitioning of selected trace elements at a given temperature. The crystallization of olivine–clinopyroxene (± magnetite) cumulates suggests that the varying clinopyroxene/melt partition coefficients (Sobolev et al., 1996
) and the varying clinopyroxene fraction among the accumulating crystals should principally control the behaviour of Sr, Y and REE, as the abundances of these elements in olivine and spinel are insignificant relative to those in the coexisting clinopyroxene (e.g. Green, 1994; Nielsen et al., 1994). The subsequent crystallization of plagioclase should strongly affect Sr partitioning, whereas REE partitioning remains relatively unchanged (e.g. Bindeman et al., 1998; Lesnov, 2001). Figure 12 shows that the compositions of clinopyroxene calculated using the model of fractional crystallization with periodic melt replenishment are indeed within the measured range of the Galmoenan clinopyroxene and compositionally analogous clinopyroxene from some AVT ultramafic volcanics (Tumrok Range and Natalia Bay). As we previously pointed out, clinopyroxene from the Valaginskii Range, although similar in HREE abundances, is more enriched in LREE and MREE and thus does not belong to the calculated trend.
|
|
DISCUSSION AND CONCLUSIONS |
|---|
The temporal and spatial association of both the Alaskan-type Galmoenan complex and the ultramafic extrusive suites of the Achaivayam–Valaginskii terrane within the same Late Cretaceous–Paleocene intraoceanic island-arc system, allochthonously emplaced in the Koryak Highland and Kamchatka Peninsula in NE Russia, suggests a genetic affinity. The comagmatic nature of the intrusive and extrusive rocks is supported by the composition of their primitive liquidus assemblage—high-Mg olivine, high-Cr and low-Ti spinel, and diopsidic clinopyroxene depleted in Al and Ti (Figs 4 –7). This allows us to assume that both suites crystallized from similar parental magmas. In other words, the magmas previously identified as parental to the AVT volcanic rocks (Sobolev et al., 1989
; Kamenetsky et al., 1995) could be parental to the cumulate rocks of the Galmoenan complex. These ultramafic melts (19–24 wt % MgO), should be recognized as a new magma type within the island-arc compositional spectrum. Unlike common subduction-related magmas (calc-alkaline, tholeiitic and boninitic series) they are exceptional in having significant enrichment in potassium, chlorine and LILE relative to REE, strongly depleted high-field strength elements of similar incompatibility, and MORB-like neodymium isotope ratios (Kamenetsky et al., 1995). The high potassium contents of the Galmoenan parental magmas are independently supported by the presence of phlogopite in the ultramafic rocks and gabbros and K-feldspar in the gabbros. We note that arc-related high-Mg parental magmas enriched in potassium have also been proposed for at least two other Alaskan-type ultramafic complexes (Irvine, 1973; Tistl et al., 1994).
The parental magma composition, proposed here to be common for the Galmoenan plutonic units and the AVT ultramafic volcanic rocks, is independently substantiated by consideration of the trace element compositions of clinopyroxene from both intrusive and extrusive suites (Figs 8 and 9). First of all, their overall similarity argues for crystallization from compositionally similar parental melts, assuming that the partitioning of trace elements between clinopyroxene and melt was similar in both cases. Second, the calculated melt compositions in equilibrium with clinopyroxene from the Galmoenan cumulates and the AVT ultramafic volcanic rocks show close similarity to the trace element compositions of the inferred parental magmas of the AVT ultramafic volcanic rocks, as represented by olivine-hosted melt inclusions (Fig. 13). This is the most compelling evidence for the formation of the Galmoenan cumulates from ultramafic island-arc magmas, similar to those identified previously for the extrusive sequences of the AVT (Kamenetsky et al., 1995).
|
The simulation of crystallization in the Galmoenan magma chamber from the parental magma (defined based on the comparison of AVT ultramafic volcanic rocks and their melt inclusions) shows that the compositions of the cumulate units are best modelled by a process of fractional crystallization with periodic magma replenishment. This provides reasonable explanations for the large amounts of dunite present (
30 wt % of the model parental melt; Fig. 11), the significant interval of co-crystallization of olivine, clinopyroxene and later plagioclase, and also the range of compositions (in terms of Mg-number) of the liquidus mafic minerals. An additional test for the validity of the crystallization models can be provided by relationships between the Mg-number and incompatible elements in the calculated and actual clinopyroxene compositions (Fig. 12). The model of fractional crystallization with replenishment produces an instantaneous increase in both Mg-number and trace element abundances in the clinopyroxene at each ‘episode’ of replenishment, and thus approximates the actual compositions of clinopyroxene from clinopyroxenites and gabbros better than simple fractionation (Fig. 12). The model curves have the ‘sawtooth’ shape reflecting the changes in the clinopyroxene composition caused by mixing with the primitive magma. Simple fractional crystallization produces a similar overall trend (decreasing Mg-number with increasing REE); however, the real REE concentrations in clinopyroxene are not modelled well. We thus assume a dominant role of fractional crystallization and replenishment of the magma chamber with small batches of primitive melt. Although the actual variations in composition, mass and frequency of melt injections cannot be quantified, this model reflects the general impact of the refilling on the final crystallization products.
The systematically higher trace element abundances at a given Mg-number in clinopyroxene from the veins in dunites, and in clinopyroxene belonging to later generation (rims of phenocrysts and matrix grains) in the plagioclase-bearing rocks resemble the compositions of clinopyroxene from the Valaginskii Range ultramafic volcanic rocks (Figs 9 and 12). This, taken together with the previously described stronger enrichment in incompatible trace elements of the Valaginskii Range (VR) magmas compared with the Tumrok Range (TR) magmas (Kamenetsky et al., 1995), may argue for a temporal change in the magma composition during plutonic evolution of the Galmoenan magmatic system. In other words, there exists a possibility that the Galmoenan magma chamber initially filled with a TR-type melt was subsequently replenished with more enriched VR-type melts. It has been suggested that the mantle source for the ATV ultramafic magmas was a highly refractory peridotite, fluxed with subduction-derived enriched melts and fluids at a depth of at least 100 km (Kamenetsky et al., 1995). We speculate that the observed evolution of the Galmoenan magmatic system reflects the change in the subduction-derived components (amount and composition) in the mantle source.
The formation of magmatic platinum-group minerals is a first-order function of the absolute quantity of PGE available in the melt; thus we need to understand the factors responsible for PGE enrichment in magmas. The composition of the mantle source and the conditions of melting are clearly important in constraining the PGE enrichment of the parental magmas forming the Galmoenan and other Alaskan-type intrusive complexes. The mantle source proposed for the Galmoenan parental magmas is a highly refractory peridotite, strongly depleted in ‘basaltic’ components but subsequently metasomatized by reaction with the trace element enriched fluids and melts. This source could have been ‘pre-enriched’ in PGE by previous melting episodes. Refractory mantle peridotite (harzburgite) has been proposed as the source of primitive boninite magmas in subduction-related environments (e.g. Hickey & Frey, 1982; Cameron et al., 1983; Crawford et al., 1989; Kamenetsky et al., 2002) and we note that the AVT ultramafic volcanic rocks have a ‘boninitic’ magma component represented by low-Ca olivine phenocrysts and olivine-hosted clinoenstatite inclusions (Kamenetsky et al., 1995). The origin of the PGE mineralization in the Bushveld and Stillwater igneous complexes has been ascribed to crystallization (in part) from high-Mg boninitic magmas (e.g. Sharpe & Hulbert, 1985; Hatton & Scharpe, 1989; Boudreau et al., 1997). Although the compatible behaviour of the PGE during melting causes progressively increasing PGE abundances in the refractory mantle residue, the preferential extraction of PGE into partial melts of harzburgite requires additional constraints. These can be provided by numerous studies on the role of chlorine-rich aqueous fluids in the redistribution of PGE within their host intrusive rocks (e.g. Boudreau et al., 1986; Hsu et al., 1991; Boudreau, 1993; Fleet & Wu, 1993) and transport at magmatic conditions (Sassani & Shock, 1998). In accord with the conclusions of Willmore et al. (2002), and the given high chlorine contents in the AVT parental magmas, we envisage that fluxing of a refractory mantle wedge in a supra-subduction zone tectonic setting by chlorine-rich aqueous fluids is primarily responsible for the formation of PGE-enriched magmas.
|
APPENDIX |
|---|
Analytical methods
Electron microprobe analyses of minerals were performed using Cameca Camebax-Microbeam (Vernadsky Institute of Geochemistry and Analytical Chemistry, RAS Moscow) and JEOL Superprobe 8200 (Max Planck Institut für Chemie, Abt. Geochemie, Mainz) electron microprobes using routine procedures. Operating conditions were 15 kV and 5–20 nA. International natural mineral standards from the Smithsonian Institution were used (Jarosevich et al., 1980
). The ZAF correction procedure was applied.
Trace element abundances in the clinopyroxenes (Table 4) were analysed using a Cameca IMS-4f ion microprobe (Institute of Microelectronics, RAS, Yaroslavl, Russia) following the method of Shimizu & Hart (1982). To improve beam stability and avoid mass superposition, the 10 kV O2– was used instead of O– as a primary beam. A primary beam current of 4 nA was focused on the sample over a spot diameter of about 25 µm. Secondary ions were collected from the imaged field of 25 µm in diameter and energy filtered using a sample offset voltage of –100 V and energy window of 50 eV (Sobolev, 1996). The analytical error was <10% for most elements present in amounts higher than 0·1 ppm and 30–50% for elements at lower concentration.
Trace element abundances in the clinopyroxenes (Table 5) were obtained using a laser ablation inductively coupled plasma-mass spectrometry (ICP-MS) system at the Max Planck Institut für Chemie, Mainz. This system includes a New Wave, Merchantek UP213 UV Nd–YAG laser coupled to a Finnigan-MAT Element-2 magnetic sector field ICP-MS system. Samples were ablated using 90 µm spots, a repetition rate of 10 Hz and a laser energy of 6 J/cm2 in He atmosphere. The measurements were calibrated using the NIST SRM 612 and KL2-G reference glasses (Jochum et al., 2000). Typical external precision is better than 4% (RSD) for most elements.
|
ACKNOWLEDGEMENTS |
|---|
We acknowledge P. Fedorov for providing the picrobasalt sample (F815), and S. Simakin, D. Kuzmin and B. Stoll for help with analytical studies. An anonymous reviewer is thanked for thoughtful comments. We greatly appreciate the editorial handling and insightful reviews by R. J. Arculus and M. Wilson. This work was supported by the Wolfgang Paul Award (Alexander von Humboldt Stiftung, Germany) to A.V.S. V.S.K. acknowledges the Friedrich Wilhelm Bessel Award (Alexander von Humboldt Stiftung, Germany) that provided funding for his stay in Mainz.
* Corresponding author. Telephone: +49-6131-305570. Fax: +49-6131-371051. E-mail: batanova{at}mpch-mainz.mpg.de
|
REFERENCES |
|---|
Anders, E. & Grevesse, N. (1989). Abundance of the elements: meteoritic and solar. Geochimica et Cosmochimica Acta 53, 197–214.[CrossRef][ISI]
Ariskin, A. A. (1999). Phase equilibria modeling in igneous petrology: use of COMAGMAT model for simulating fractionation of ferro-basaltic magmas and the genesis of high-alumina basalt. Journal of Volcanology and Geothermal Research 90, 115–162.[CrossRef][ISI]
Ariskin, A. A. (2003). The compositional evolution of differentiated liquids from the Skaergaard Layered Series as determined by geochemical thermometry. Russian Journal of Earth Sciences 5, 1–29.
Ariskin, A. A., Frenkel, M. Ya., Barmina, G. S. & Nielsen, R. L. (1993). COMAGMAT: a Fortran program to model magma differentiation processes. Computers and Geosciences 19, 1155–1170.[CrossRef]
Astrakhantsev, O. V., Kazimirov, A. D., Krylov, K. A. & Fedorov, P. I. (1987). Tectonics and structure of the Vatyna nappe frontal