Journal of Petrology Advance Access originally published online on January 17, 2008
Journal of Petrology 2008 49(2):267-294; doi:10.1093/petrology/egm081
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Magma Ascent and Crustal Accretion at Ultraslow-Spreading Ridges: Constraints from Plagioclase Ultraphyric Basalts from the Arctic Mid-Ocean Ridge
1Reslab Integration as, Kokstadflaten 19B, N-5257 Kokstad, Norway
2Department of Earth Science, University of Bergen, Allegt 41, 5007 Bergen, Norway
RECEIVED APRIL 7, 2006; ACCEPTED NOVEMBER 30, 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMOHNS AND KNIPOVICH RIDGESSEAFLOOR SAMPLINGPETROGRAPHYGEOCHEMISTRYDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Plagioclase ultraphyric basalts (PUBs) with up to 54% plagioclase phenocrysts were dredged in the rift valley and adjacent flanks of the ultraslow-spreading Mohns and Knipovich ridges. The PUBs show large variations in crystal morphologies and zoning. The large variations suggest that single basalt samples contain a mixture of plagioclase crystals that aggregated at different levels in the magma conduits. Resorbed crystals and repeated reverse zones suggest that the magma reservoirs were replenished and heated several times. Thin concentric zones with melt inclusions, and sharp reductions in the anorthite content of 3–7%, are common between the reverse zones. These zones, and skeletal crystals with distinctly lower anorthite contents than massive crystals, are interpreted to be the result of rapid crystalliztion during strong undercooling. The changes between short periods of cooling and longer periods with reheating are explained by multiple advances of crystal-rich magma into cool regions followed by longer periods of gradual magma inflow and temperature increase. The porphyritic basalts are characterizd by more depleted and more fractionated compositions than the aphyric basalts, with lower (La/Sm)N, K2O and Mg-numbers. This relationship, and the observation that PUBs are sampled only close to segment centres along these ridges, suggests that the PUBs formed by higher degrees of melting and evolved in more long-lived magma reservoirs. We propose that the zoning patterns of plagioclase crystals and crystal morphologies of these PUBs reflect the development and flow of magma through a stacked sill complex-like conduit system, whereas the aphyric equivalents represent later flow of magma through the conduit. The formation of voluminous higher-degree melts may trigger the development of the magma conduits and explain the generally depleted compositions of PUB magmas.
KEY WORDS: basalt; mineral chemistry; MORB; magma mixing; magma chamber; major element
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMOHNS AND KNIPOVICH RIDGESSEAFLOOR SAMPLINGPETROGRAPHYGEOCHEMISTRYDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Mid-ocean ridges spreading at 20–180 mm/year (slow- to fast-spreading) are commonly segmented by transform faults that are rigid plate boundaries separating first-order segments. At spreading rates below these, the ridges may be characterized by linked amagmatic and magmatic accretionary segments and a lack of transform faults (Dick et al., 2003
). Examples of such ridges are the Arctic ridges north of Jan Mayen, the eastern part of the South West Indian Ridge between the Melville Fracture Zone and the Rodriguez Triple Junction, and the Gulf of Aden spreading centre. The importance of slow- to ultraslow-spreading ridges (<20 mm/year) in understanding mid-ocean ridge processes has been emphasized in recent years; however, the nature of the lower crust and the magma reservoirs is still poorly understood (e.g. Mendel et al., 1997; Parson et al., 1997; Patriat et al., 1997; Robinson et al., 2001; White et al., 2001; Jokat et al., 2003; Michael et al., 2003).
The more primitive basalt compositions at slow- to ultraslow-spreading ridges relative to those of basalts from faster-spreading ridges indicate that the magmas have erupted relatively shortly after segregation from the melting region (Flower, 1981; Sinton & Detrick, 1992). Gradual increases in P-wave velocities with depth may indicate a gradual mantle–crust transition with partially frozen reservoirs (mush zones), serpentinized mantle rocks, or gabbroic rocks representing frozen and fractionated magmas in the lower crust and uppermost mantle (e.g. White et al., 1992; Sleep & Barth, 1997; Klingelhöfer et al., 2000a, 2000b). There is also evidence of substantial fractionation in magma reservoirs from first-order observations of thick gabbro sequences at the South West Indian Ridge (Dick et al., 2000; Coogan et al., 2001).
Aphyric and plagioclase-phyric basalts were sampled during the SUBMAR-2000 and -2002 cruises to the Mohns and Knipovich ridges (Fig. 1). The plagioclase-phyric basalts have plagioclase phenocryst contents between 10 and 54% and olivine and clinopyroxene phenocryst contents <2%. The plagioclase-phyric samples are classified as plagioclase ultraphyric basalts (PUBs) following Cullen et al. (1989). In this paper we report morphologies and compositional zoning patterns of plagioclase crystals together with glass compositions, and discuss the results in terms of the functioning of magma reservoirs and conduit systems at ultraslow-spreading ridges.
|
|
MOHNS AND KNIPOVICH RIDGES |
|---|
|
TOPABSTRACTINTRODUCTIONMOHNS AND KNIPOVICH RIDGESSEAFLOOR SAMPLINGPETROGRAPHYGEOCHEMISTRYDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The Mohns and Knipovich ridges, located in the Norwegian–Greenland Sea, represent two super-segments of the ultraslow-spreading Arctic Mid-Ocean Ridge (Fig. 1). The Mohns Ridge has a WSW–ENE trend and extends
575 km from the Jan Mayen Fracture Zone (10°W, 72°N) to the Knipovich Ridge (7°E, 73°30'N). The spreading centre changes direction to the north–south-trending Knipovich Ridge, which extends 550 km to the Molloy Fracture Zone at 78°30'N. The spreading along the Mohns Ridge is 30° oblique, whereas the Knipovich Ridge changes from 60° oblique spreading in the southern part to 30° north of 75°N. The ridges are spreading at full rates of 15–17 mm/year (Eldholm et al., 1990).
A well-defined axial valley characterizes the ridges, with depth variations generally between 2·8 and 3·8 km, increasing from the Jan Mayen area to the Knipovich Ridge (Crane et al., 2001; Hellevang, 2003). The axial crustal thickness on the central Mohns Ridge has been examined by both REE-inversion methods and seismic measurements (Klingelhöfer et al., 2000a). Klingelhöfer et al. (2000a) concluded that the crustal thickness is highly variable, with a mean of 4 km and maximum thicknesses below the axial volcanic ridges (AVRs) reaching 6 km; the latter is comparable with average crustal thicknesses on faster-spreading ridges (Cannat, 1996).
The spreading centres are segmented with AVRs arranged en echelon. The AVRs rise 600–1300 m above the rift valley at distances (center to center) of 20 to >100 km (Crane et al., 2001; Okino et al., 2002; Hellevang, 2003). Their lobate (as opposed to rectilinear and hence probably fault-controlled) margins and occurrence of fresh basaltic glass on the slopes of these highs indicate a volcanic origin (Hellevang, 2003). This is also favoured by distinct mantle Bouguer anomaly (MBA) lows, and high reflectivity on SeaMARK II side-looking sonar images, suggesting thicker crust and enhanced volcanic activity, respectively, at the AVRs (Géli et al., 1994; Crane et al. 2001; Okino et al., 2002). Deep basins between the AVRs have been interpreted to represent transfer zones that gradually offset the rift valley (Dauteuil & Brun, 1993). Dredge sampling shows that the seafloor within these basins is of volcanic origin (Hellevang, 2003; Hellevang & Pedersen, 2005). Géli et al. (1994) showed that the axial part of the Mohns Ridge is situated above a low-velocity zone. This low-velocity zone is widest beneath the AVRs (10–15 km) and narrows towards the transfer zones (0–5 km).
The most pronounced AVRs on the northern Knipovich Ridge are in line with discontinuous off-axial highs. This may suggest that the segmentation in this area has been stable for more than 7–8 Myr, based on a continuous half-spreading rate of 8 mm/year (Hellevang & Pedersen, 2005).
|
SEAFLOOR SAMPLING |
|---|
|
TOPABSTRACTINTRODUCTIONMOHNS AND KNIPOVICH RIDGESSEAFLOOR SAMPLINGPETROGRAPHYGEOCHEMISTRYDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Aphyric and plagioclase ultraphyric basalts were dredged on the neovolcanic zone and the flanks of the Mohns and Knipovich ridges during the SUBMAR-2000 and -2002 cruises (Fig. 1a and b). Sampling was also carried out using a remotely operated vehicle (ROV) at depths between 600 and 1500 m on the northwestern flank of the Mohns Ridge. Samples were gathered from 64 dredge hauls and two ROV dives. Of these, 5% were porphyritic basalts (Fig. 2), 77% aphyric basalts, 1% hyaloclastite, and 17% ice-dropped material. Both aphyric and porphyritic basalts were recovered from the AVRs, whereas dredges from the deeper basins between the AVRs were dominated by aphyric basalts.
|
|
PETROGRAPHY |
|---|
|
TOPABSTRACTINTRODUCTIONMOHNS AND KNIPOVICH RIDGESSEAFLOOR SAMPLINGPETROGRAPHYGEOCHEMISTRYDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The proportion of phenocrysts in the porphyritic rocks varies from
10% for sample Mo-25D/1 to a maximum of 54% for samples Mo-6D/49 and Mo-9D/2. Logarithmic plots of plagioclase phenocryst size distributions (CSDs) show curvilinear trends for the smaller crystal sizes (Fig. 3). These curvilinear trends comprise more than 90% of the crystals at crystal sizes up to 4–7 mm. Some of the CSDs show a bend or kink towards a lower slope for the larger crystals. These bimodal crystal size distributions can be interpreted as representing mixing of crystal populations in the melts (Marsh, 1988), and may suggest a xenocrystic origin for the larger crystals. Crystal clusters composed of plagioclase only or plagioclase and olivine and/or clinopyroxene are common (Fig. 4c and e). These crystal clusters may be autoliths of gabbroic or troctolitic modal composition. Growth relationships between phenocryst phases within these crystal clusters suggest a crystallization order of olivine and plagioclase prior to clinopyroxene.
|
|
Plagioclase is by far the dominant phenocryst phase and occurs as discrete single crystals (megacrysts) or in crystal clusters. The plagioclase phenocrysts show strong variations in morphology, with tabular, skeletal, acicular, resorbed or euhedral crystals, and massive crystals with no melt inclusions (Figs 2 and 4). Single samples may show several of these plagioclase morphologies coexisting; for example, skeletal crystals coexisting with massive crystals, and resorbed crystals coexisting with euhedral crystals (Figs 2 and 4g).
Concentric bands with internal variations in the extinction angles are common (Fig. 4a and d), although many of the crystals are optically more or less unzoned (Fig. 4f). These concentric bands probably reflect variations in growth conditions (e.g. growth rate). The bands vary in thickness from less than 20 µm to more than 1 mm and show variations from euhedral shapes forming sector-zoned crystals (Fig. 4a and d) to highly rounded zones indicating dissolution (Fig. 4d). The bands frequently contain trapped melt inclusions together with mineralized and unmineralized vapour cavities (Figs 4a and 5).
|
Plagioclase crystals typically have abundant melt inclusions (Figs 4a, b, g, h, and 5). Inclusions may occur within thin zones (<50 µm) or be distributed throughout the crystals, resulting in a dusty appearance (Fig. 4b and h). Melt inclusions that occur in distinct euhedral zones are probably formed during crystal growth. Some of the phenocrysts, especially in sample Mo-25D/1, show amoeboidal melt inclusions that fill up as much as 25% of the plagioclase crystals (Fig. 4b). The amoeboidal structures crosscut growth faces and contain fresh glass when present in the glassy margin of the sample, indicating that these structures formed by preferential dissolution of the plagioclase crystal.
Olivine occurs as a minor phenocryst phase in the PUBs (<1%), forming 1–3 mm highly resorbed to euhedral crystals. Clinopyroxene is also a minor phenocryst phase (<1%) and occurs as up to 8 mm, partly resorbed to highly resorbed, transparent phenocrysts. Sample Mo-6D/30 contains large poikilitic clinopyroxene crystals ranging in size from <1 mm to 8 mm. These phenocrysts have abundant inclusions of resorbed and euhedral plagioclase and highly resorbed olivine crystals (Fig. 6). Some of the resorbed plagioclase inclusions are surrounded by mafic cryptocrystalline material suggesting that they were trapped in the poikilitic clinopyroxene crystals together with surrounding melt. Clinopyroxene crystals are generally unzoned, although chemically and optically discrete zones occur in some of the larger crystals (Fig. 6).
|
|
GEOCHEMISTRY |
|---|
|
TOPABSTRACTINTRODUCTIONMOHNS AND KNIPOVICH RIDGESSEAFLOOR SAMPLINGPETROGRAPHYGEOCHEMISTRYDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Analytical methods
Basaltic glass and mineral phases were analysed for major element compositions using an ARL SEMQ microprobe at the University of Bergen. Only samples that contained pure glass without visible alteration were analyzed. Analytical conditions were 15 kV accelerating voltage, 1 nA specimen current, a focused beam of 5 µm, and commonly 20 s counting time, although some clinopyroxenes were analysed with 60 s counting time to obtain more precise measurements of Cr and Ti. Standards for calibration were synthetic and natural minerals [wollastonite (Ca), quartz (Si), corundum (Al), orthoclase (K), albite (Na), and rutile (Ti)], MgO (Mg), and pure metal phases (Fe, Cr, Mn). Indian Basalt glass (USNM 113716) was used to control the calibration for basaltic glass analyses. The glass compositions are presented in Table 1.
|
Repeated analyses of single spots on crystals generally show a precision of <±1% An and <±0·5% Fo. Analyses of matrix microcrysts yielded larger analytical variations, commonly with standard deviations (SD) of 1–5% anorthite in plagioclase, 2–4% enstatite in clinopyroxene, and <1% for the forsterite component of olivine (see Table 4). Matrix plagioclase microcrysts also show elevated concentrations of minor elements such as FeO, MgO and TiO2. The relatively large variations and high contents of minor elements are probably caused by beam overlap with glass or matrix during analyses, because of the small size of the matrix crystals, which have minimum dimensions generally <10–15 µm. The phenocrysts were analysed with 21–118 point analyses along core to rim profiles, with point analyses being separated by 25–120 µm; hence fine-scale oscillatory zoning on 1–20 µm scale is not reproducible. Repeated core to rim analyses of single phenocrysts commonly show consistent variations, as shown by plagioclase crystals in Mo-5R (pl1 and pl2) (Fig. 7). Average compositions of microcrysts and phenocrysts of porphyritic basalts are presented in Table 4.
|
Hand-picked glasses were analysed for trace elements by inductively coupled plasma mass spectrometry (ICP-MS) at the Memorial University of Newfoundland. The analytical procedure has been described by Jenner et al. (1990
) and included the HF–HNO3 dissolution of a 0·1 g (where available) sample aliquot, followed by ICP-MS analysis using the method of standard addition to correct for matrix effects. For quality control, one or more geological reference standards were prepared and analysed with the samples, together with a reagent blank to measure the reagent contribution to the values. Selected samples were dissolved (where sufficient sample was available) and analysed in duplicate, to further control the precision of the measurements. Trace element data are presented in Table 2.
|
Basaltic glass samples were analysed for Sr and Nd isotopes by thermal ionization mass spectrometry using a MAT262 system housed at the University of Bergen. Basaltic glass was handpicked and carefully cleaned in an ultrasonic bath before dissolution. Samples were dissolved in HF and HNO3, and REE and Sr were separated by specific extraction chromatography. Nd was subsequently separated using a low-pressure ion-exchange chromatographic setup with HDEHP-coated Teflon powder (100 µm) as the ion-exchange resin. Nd isotopic ratios were corrected for mass fractionation using a 146Nd/144Nd ratio of 0·7219, and repeated measurements of Johnson and Matthey NdO3 batch No. S819093A yielded an average 143Nd/144Nd of 0·511107 ± 5 (2SE, n = 90). Sr was loaded on a double filament and analysed in static mode. Sr isotopic ratios were corrected for mass fractionation using an 88Sr/86Sr of 8·3752, and repeated measurements of the NBS987 standard yielded an average 87Sr/86Sr of 0·710240 ± 8 (2SE). Isotope data are presented in Table 3.
|
|
Element compositions of basaltic glasses
Most dredges within the neovolcanic zone in the central rift valley contained basalts with fresh glass, whereas some of the dredges at the flanks of the ridges contained basalts with devitrified and palagonized glass. Only plagioclase ± olivine and clinopyroxene crystals have been analysed for these samples (e.g. Mo-5R, Mo-6D/30, Mo-6D/49 and Mo-9D/2).
The basaltic glasses from the Mohns and Knipovich ridges have MgO concentrations between 6·9 and 9·1 wt % (n = 20) and 7·1 and 8·6% wt % (n = 31), respectively (Table 1). The Mg-numbers [Mg-number = Mg2+/(Mg2+ + Fe2+(tot)) x 100] of the Mohns Ridge basalts vary between 53·6 and 63·9 with an average of 60·3 (Table 1). The Knipovich Ridge basalts have slightly higher Mg-numbers between 57·3 and 66·3, and an average of 61·8 (Table 1). The Mohns and Knipovich samples display well-defined trends of Al2O3 and FeO vs Mg-number, whereas TiO2, CaO, Na2O and K2O display more scattered trends (Fig. 8). Basalts from the Knipovich Ridge are generally higher in the moderately to highly incompatible major elements TiO2, K2O and Na2O relative to the Mohns Ridge basalts. They are also generally higher in Al2O3 and lower in CaO (Fig. 8). The FeO contents vary highly for both ridges although both show a common, distinct low-FeO group defined by an 1 wt % lower FeO content at the same Mg-number as the other samples. The low-FeO group comprises aphyric basalts from both ridges, whereas all ultraphyric basalts have distinctly higher FeO contents.
|
The differences in the incompatible major elements between the Mohns and Knipovich basalts are accompanied by differences in incompatible trace element compositions (Figs 9 and 10). The Knipovich basalts generally have higher light rare earth element (LREE) concentrations and chondrite-normalized La/Sm ratios (La/SmN) than the Mohns basalts, with average La/SmN of 1·31 and 1·13, respectively.
|
|
The PUBs can be distinguished from the aphyric basalts by their generally higher FeO contents and lower K2O contents (Fig. 8), distinctly more LREE-depleted and flatter REE patterns (Fig. 9), and distinctly lower La/SmN ratios (PUBs: average 0·82; aphyric basalts: average 1·4; Fig. 10).
Plagioclase compositions are strongly controlled by the Ca-number [Ca/(Ca + Na) x 100] and Al-number [Al/(Al + Si) x 100] of the melt (Panjasawatwong et al., 1995). The Knipovich Ridge basalts have lower average Ca-number (Ca-number 68·9 vs 72·8), and slightly higher average Al-number (27·3 vs 26·2) compared with the Mohns Ridge basalts (Table 1). The Knipovich Ridge basalts have generally higher Mg-numbers than the Mohns Ridge basalts, and the Knipovich Ridge basalts are shifted to the right relative to the Mohns Ridge basalts in Ca-number–Mg-number diagrams (Fig. 11a). The Mohns and Knipovich basalts appear to follow a common Al-number–Mg-number trend; the aphyric basalts of the Knipovich Ridge occupy the high-Al-number–high-Mg-number end of the trend (Fig. 11b).
|
The Mohns Ridge PUBs show a positive Ca-number–Mg-number trend that is steep compared with variations defined by aphyric samples from the Mohns Ridge (Fig. 11a). Average Ca-numbers for PUBs from the Mohns and Knipovich ridges are 73·4 and 69·6, respectively; this difference is primarily a function of the higher Mg-numbers of the Mohns Ridge PUBs, and the Knipovich Ridge PUBs plot at only slightly lower Ca-numbers than the trend defined by the Mohns Ridge PUBs. The Al-numbers of Mohns and Knipovich PUBs are more or less similar, with average values of 26·0 and 26·2, respectively.
Isotope composition of basaltic glass
Basaltic glasses from the Mohns Ridge yielded 87Sr/86Sr between 0·7031 and 0·7035 and 143Nd/144Nd between 0·513010 and 0·513123 (Table 3, Fig. 10). The Knipovich Ridge basalts are generally more depleted isotopically with higher 143Nd/144Nd (0·513046–0·513160), and lower 87Sr/86Sr (0·7028–0·7033) (Fig. 10).
The PUBs from the Knipovich Ridge have higher 143Nd/144Nd and slightly lower 87Sr/86Sr than the aphyric samples (Fig. 10a). In general, the Mohns Ridge PUBs show a similar relationship, with distinctly higher 143Nd/144Nd compared with the aphyric samples. Together, the Mohns and the Knipovich PUBs define a trend in Fig. 10a that is displaced to the right (higher Sr-isotope ratios at similar Nd-isotope ratios) of the aphyric basalts. One Mohns Ridge PUB exhibits distinctly higher 87Sr/86Sr and lower 143Nd/144Nd and defines an end-member of the PUB trend. The fact that the corresponding aphyric basalt trend does not extend to similarly high 87Sr/86Sr values is probably a sampling effect, as aphyric basalts with similar or higher Sr-isotopic ratios have been sampled from nearby segments of the Mohns Ridge.
La/SmN vs 143Nd/144Nd (Fig. 10b) shows that the PUBs in general define the more depleted, low La/SmN–high 143Nd/144Nd end of local trends at both ridges. The PUBs appear accordingly to have trace and isotopic signatures that are distinct from those of the associated aphyric basalts.
Mineral compositions
Olivine phenocrysts in the PUBs show compositional variations between Fo88·1 and Fo84·1 (Table 4). The clinopyroxene phenocrysts are chromian augites and endiopsides with Mg-numbers between 90·2 and 86·3, Al2O3 between 2·55 and 3·52 wt %, and Cr between 0·28 and 1·10 wt %. These crystals commonly show minor compositional variations (Table 4). However, one large (8 mm) zoned clinopyroxene with substantial core to rim variations in Cr and Al2O3 was observed (Fig. 6). The core of this crystal shows a normal zoning pattern with a decrease in the Cr content outwards from 0·80 to 0·25 wt %, and a decrease in the Al2O3 from 3·10 to 2·70 wt %. The normal zoning is not reflected in the Mg-number, which shows more or less constant values of 89 close to the core (inner 2 mm), and higher values of 90 in the outer 1 mm of the core. The core is surrounded by an optically distinct mantle, exhibiting Cr contents between 0·98 and 1·11 wt % and Al2O3 between 3·23 and 3·52 wt %. Both Cr and Al2O3 decrease outwards in the mantle, whereas Mg-number increases slightly from 88·4 to 89·1. The differences in composition between the outer core and the mantle clearly reflect crystallization in different magmas.
Plagioclase microcrysts in the matrix are considerable more albitic than the phenocrysts and range in composition between An67·8 and An73·2 (Table 4). These crystals are generally high in minor elements such as FeO and MgO compared with the phenocrysts, and average FeO and MgO contents range from 0·60 to 0·98 wt % (phenocrysts: 0·28–0·41 wt %) and 0·22–0·43 wt % (phenocrysts: 0·15–0·27 wt %), respectively. Analyses of microcrysts from the matrix also show elevated concentrations of TiO2 (0·03–0·07 wt %) compared with the phenocrysts, which generally contain <0·01 wt % TiO2, although the plagioclase phenocrysts of Mo-6D/30 have slightly higher TiO2 concentrations of 0·03 wt %. The high minor element contents of the microphenocrysts are probably an artefact of the analyses, as a result of partial beam overlap with glass or matrix during microprobe analyses.
Olivine and clinopyroxene microcrysts have distinctly more evolved compositions compared with the phenocrysts (Table 4). Olivine microcrysts show variations between Fo84·7 and Fo82·5, whereas the clinopyroxene microcrysts show variations between Mg-number 83·9 and 78·5, respectively.
Compositional profiles of plagioclase phenocrysts
Compositional profiles across plagioclase phenocrysts show significant core to rim variations, and single phenocrysts show variations of up to 15% An (Figs 7, 12 and 13). The phenocrysts display normal, reverse and more complex variations, and considerable differences in zoning patterns and morphologies occur between coexisting phenocrysts in a single sample. The overall variations from core to rim of the plagioclase phenocrysts are more commonly reverse than normal, however.
|
|
The plagioclase analyses show that core to rim variations can be described and subdivided into first- and second-order variations. First-order variations are major core to rim changes in the An content, and commonly one to five such zones can be distinguished at wavelengths of 100 to more than 1000 µm. The second-order variations are defined by zig-zag variations of 1–2% An that occur within all first-order zones. Repeated core to rim analyses of single plagioclase crystals show that the first-order variations can be reproduced, whereas the second-order variations cannot be reproduced (Fig. 7a and b). The latter may be due to analytical errors on the microprobe analyses of the order of ±1% An, and dense oscillatory zoning at wavelengths that are less than the typical separation of analyses in this study of >50 µm.
The first-order zoning is commonly rhythmic, with repeated reverse zones separated by thin normal zones. Up to four such repetitions occur in the large phenocrysts, and these repeated reverse zones may be traced in several coexisting phenocrysts in some of the basalt samples (Fig. 7). The plagioclase crystals from the Knipovich PUBs show similar core to rim variations to the Mohns Ridge PUBs, with alternating reverse and normal zones, although with generally more sodic compositions (An84·9–71·4).
Some of the phenocrysts show dense (< 20 µm) optical zoning that can be distinguished by different extinction angles (Figs 7a and 12b). This dense zoning is referred to as oscillatory zoning in the following discussion. Melt inclusions are present within some of the optically distinct zones (Fig. 7). Although not examined in detail in this study, the short-wavelength dense optical zones may be associated with the second-order geochemical variations that are observed in most of the analysed crystals. Zones that contain melt inclusions, on the other hand, commonly occur where there are substantial variations in the An content of the plagioclase crystals, typically representing transitions between first-order zones. These transition regions are either relatively thin (10–50 µm), marking the step from one reverse zone to another reverse zone (Figs 7b and 13c, d), or more gradual >100 µm (Figs 7a, c and 13a), and are commonly several per cent lower in An than the surrounding plagioclase.
There is a systematic relationship between plagioclase composition and the morphology of the crystals; skeletal crystals and crystals with melt inclusions are more sodic than massive, melt inclusion-free crystals (Fig. 14). Compositional profiles across skeletal crystals show mean compositions between An79·8 and An82·5 (Table 4). These values are well below the mean compositions of massive plagioclase crystals (An84·0–87·5). Plagioclase crystals may exhibit thin concentric bands with melt inclusions, or melt inclusions occupying a major part of the crystal. In both cases, areas with melt inclusions are associated with more sodic plagioclase compositions. This relationship between morphology and composition is well illustrated by the crystals in samples Mo-5R (Fig. 7), Mo-6D (Fig. 14a), Mo-9D (Fig. 14b), and Mo-25D/1 (Fig. 13b–d). The crystals from sample Mo-5R show thin concentric bands with melt inclusions. These zones are 10–100 µm thick, and are typically associated with reductions in the An content by 3–7%. Back-scattered electron images of one zone with melt inclusions within crystal Mo-5R pl3 show a 40–50 µm thick band with a more sodic composition (darker grey) containing 10 µm spherical inclusions (Fig. 5a).
|
The crystals in sample Mo-25D/1 show both concentric bands with melt inclusions and larger areas with melt inclusions. One crystal (Mo-25D/1 pl2) possesses a core with many small inclusions and larger amoeboidal melt inclusions, resulting in a dusty appearance (Fig. 13b). This core is surrounded by a massive inclusion-free mantle. The core and mantle show distinct compositional differences with mean values of An82·5 and An87·8, respectively (Fig. 13b, Table 4), and there is a clear difference in the extinction angle between the zones.
Several totally different core to rim variations may be present in a single sample. Figure 7 shows three crystals from sample Mo-5R. These crystals all show similarities as they exhibit repeated reverse zones, separated by normal zones associated with melt inclusions. The crystals also show more or less similar ranges in absolute values. The core to rim variations are, however, different with respect to the number of reverse zones. The outermost 1–1·5 mm part shows similar compositional variations, similar zoning patterns, and also a band of melt inclusions associated with the lowest An value. The outermost parts of the crystals in this sample thus apparently reflect a similar growth history, whereas their earlier history may be different. Plagioclase crystals from other PUBs may show different morphologies and zoning patterns, as illustrated by sample Mo-25D/1 from the Mohns Ridge (Fig. 13) and Kn-62D/17 from the Knipovich Ridge (Fig. 12). These samples include crystals that have different zoning patterns and thus different crystallization histories.
MgO is present as a minor element in plagioclase crystals (Table 4), and the mean concentration varies between 0·18 and 0·43 wt %, with higher concentrations measured in the microphenocrysts in the matrix (although, as noted above, this may partially be due to beam overlap with matrix or glass during microprobe analysis). Variations are also observed within single phenocrysts, where higher MgO contents are associated with the more sodic inclusion-rich zones. This is well illustrated by crystal Mo-5R pl1, which shows three distinct zones with melt inclusions (Fig. 15). The MgO of the crystal varies between 0·17 and 0·38 wt %, with an average of 0·24 wt % (Fig. 15b). There is a clear relationship between the MgO content and the An content, well illustrated in the outermost 1 mm of the crystal, where higher MgO is associated with low An. The outermost two zones with melt inclusions show peaks in MgO of 0·33 and 0·38 wt %, respectively, both corresponding to the lowermost An contents within the first-order zones.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMOHNS AND KNIPOVICH RIDGESSEAFLOOR SAMPLINGPETROGRAPHYGEOCHEMISTRYDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Difference in plagioclase compositions between the Mohns and the Knipovich ridges
The average An contents of the plagioclase phenocrysts in the Mohns Ridge PUBs are 6–8% higher than those of the plagioclase phenocrysts in the Knipovich PUBs. Plagioclase–melt equilibrium experiments show that there is a strong relationship between melt composition and the composition of plagioclase crystals that are in equilibrium with these melts (e.g. Housh & Luhr, 1991
; Panjasawatwong et al., 1995). Besides melt water content and depth of crystallization, both Ca-number and Al-number in the melt have a strong effect on the composition of the plagioclase precipitating.
H2O analyses of Knipovich basalts show that the porphyric basalts contain 0·5% water, which is less than the aphyric basalts (Hellevang & Pedersen, 2005). The Mohns Ridge basalts have not been analysed for H2O, but a well-defined positive correlation between H2O and K2O contents for the Knipovich Ridge (Hellevang & Pedersen, 2005) and the similar K2O contents of the porphyritic basalts from the two ridges suggest that the water content at the Mohns Ridge will generally be as low as it is at Knipovitch. It seems, therefore, unlikely that differences in water content in the porphyritic magmas can be responsible for the difference in plagioclase composition between the ridges.
The An content of equilibrium plagioclase increases 1%/kbar with decreasing crystallization pressure at constant MgO (Pringle et al., 1974; Longhi et al., 1993). The general geochemical variations within the Knipovich Ridge suite suggest relatively shallow fractionation depths (Hellevang & Pedersen, 2005), and so differences in crystallization pressure appear unrealistic as an explanation for the differences in plagioclase compositions between the Mohns and Knipovich ridges.
Higher magmatic Ca-number and Al-number both lead to higher An content in crystallizing plagioclase. The Al-numbers of basaltic glass from both the Mohns and Knipovich ridges follow a similar positive correlation with Mg-number, and show no apparent difference between the PUBs from these two ridges. The Knipovich Ridge PUBs and aphyric basalts generally have lower Ca-numbers than the Mohns Ridge basalts at the same Mg-numbers. This difference is related to distinctly lower Na2O contents and slightly higher CaO contents in the Mohns Ridge basalts relative to the Knipovich Ridge basalts (Fig. 8).
The difference in average plagioclase compositions between the Mohns and Knipovitch Ridges therefore seems to be controlled by differences in the major element magma compositions rather than to differences in crystallization conditions or volatile contents of the magmas.
Formation of extreme plagioclase-enriched magmas
Plagioclase ultraphyric basalts have been described from a wide range of volcanic environments including divergent and convergent plate margins, oceanic islands and continental flood basalt provinces (Flower, 1980; Kuo & Kirkpatrick, 1982; Cullen et al., 1989; Sinton et al., 1993; le Roex et al., 1996; Hansen & Grönvold, 2000). Several workers have suggested that plagioclase-rich magmas are formed when magmas mix with mush zones, or plagioclase cumulates (Cullen et al., 1989; le Roex et al., 1996; Hansen & Grönvold, 2000).
In the PUBs studied here, the phenocryst morphologies and compositions vary strongly between and within samples, indicating that the crystals experienced different growth histories even when they were erupted in the same magma. The olivine phenocrysts support this conclusion: using the Mg/Fe exchange coefficients of Roeder & Emslie (1970), the olivine phenocrysts should have been in equilibrium with melts with Mg-numbers between 67 and 71. This is higher than any basaltic glass yet analyzed from the Mohns and Knipovich ridges, and thus the olivine crystals must have been in disequilibrium with their host magmas at eruption. The resorbed surfaces of clinopyroxene and olivine crystals are in agreement with this conclusion. The preservation of this chemical disequilibrium despite the rapid diffusion of Fe and Mg in olivine (Gaetani & Watson, 2000) suggests that the residence time of olivines in the more evolved magma prior to eruption was relatively short—probably of the order of days to months.
The high content of plagioclase crystals in the PUBs compared with olivine and clinopyroxene may be due to extreme plagioclase crystallization, or due to mixing of magma with a plagioclase-dominated cumulate. Extreme plagioclase fractionation may occur if the magmas have extremely high Al2O3 and CaO contents (Sinton et al., 1993; Nielsen et al., 1995); however, this results in steep liquid lines of descent (LLDs), especially with respect to Al2O3 and CaO. We observe relatively flat and well-defined LLDs for elements such as FeO that can be explained only by the corresponding fractionate having MgO contents >10%, suggesting >20% olivine fractionation (Hellevang & Pedersen, 2005). The low abundance of clinopyroxene and olivine in the erupted magmas seems, therefore, likely to be connected with physical separation of phenocryst phases (i.e. plagioclase flotation and/or resorption of olivine and clinopyroxene). A number of studies have shown that plagioclase crystals have lower densities than the host melts, suggesting that plagioclase flotation may occur and that plagioclase-rich fractionates may develop at the top of the magma reservoirs (Campbell et al., 1978; Flower, 1980). Augite and olivine crystals, on the other hand, are denser than the melt and will settle to the base of the magma reservoir (e.g. Batiza & Niu, 1992; le Roex et al., 1996, 2002). The olivine and clinopyroxene phenocrysts generally show evidence of resorption, suggesting that dissolution may have also contributed to the relative low abundance of these minerals.
The curvilinear CSD trends for the smaller crystals (up to 4–6 mm in length, Fig. 3) may suggest steady nucleation and crystallization rates for these relatively small crystals (e.g. Marsh, 1988; Cashman, 1993). The larger crystals deviate from these trends and define lower slopes or even flat CSDs. These larger crystals therefore seem to stem from separate magmas that followed a more protracted crystallization history. The plagioclase enrichment in these rocks seems, accordingly, to be the result of the aggregation of crystals formed in different magma bodies with contrasting crystallization histories.
Plagioclase growth rates
The growth rates of plagioclase have been investigated during growth experiments (e.g. Swanson, 1977; Kirkpatrick et al. 1979), by interpreting CSDs (Cashman, 1993), and by numerical simulations (e.g. Loomis, 1982). These studies reported growth rates ranging between 10–5 and 10–11 cm/s. Cashman (1993) showed that variations in plagioclase CSDs suggest variations in growth rates across dikes, and argued that for near-surface basaltic systems, expected average growth rates will be of the order of 10–10–10–11 cm/s. These rates are also favoured by studies of long-lived (>200 kyr) volcanoes in Mexico, where the approximate crystallization histories and timing can be constrained (Tepley et al., 2000). Rapid crystallization (10–6–10–8 cm/s), on the other hand, might be expected close to the rapidly cooling margins of dikes or intrusions (Cashman, 1993, and references therein). The crust at slow- to ultraslow-spreading ridges is expected to be relatively cold because 0f low magmatic activity and volcanic activity, and hence the cooling of the magma within the crust. Consequently, crystallization might be expected to be relatively fast at the Mohns and Knipovich ridges.
The crystallization rate is controlled by the degree of cooling below the liquidus, with higher rates at higher degrees of undercooling. The phenocrysts of the PUBs reported here show morphologies indicating changes between rapid growth and slower growth and even partial melting of crystals; consequently, any estimates of growth rates will obviously be somewhat uncertain. The dominance of reverse zones or compositionally uniform core to rim profiles in plagioclase phenocrysts may suggest that the crystals grew in magma reservoirs where the temperature was constant or periodically raised, favouring slow crystallization. The melt inclusion-rich zones, on the other hand, most probably formed during periods of strong undercooling and rapid growth; these are generally less than 1 mm across and may be as thin as 40 µm. Using growth rates of 10–6–10–7 cm/s more typical for rapidly crystallizing dike margins suggests that these zones may have formed over time spans ranging from minutes to tens of hours, and therefore reflect very short-term events that rapidly cooled the magma.
Melt inclusions trapped during rapid growth
Rapid growth of plagioclase during high degrees of undercooling may result in the trapping of melt inclusions. Such growth-related melt inclusions tend to congregate along growth surfaces within plagioclase crystals (Lofgren, 1974), or may alternatively be disseminated throughout a major volume of the crystal during rapid skeletal growth. Melt inclusions may also be trapped during changes in the growth conditions if partial melting at the surface of the crystal is followed by rapid crystallization. Evidence for both skeletal and rapid zoned growth is common in the studied PUBs (Figs 7 and 12–14
).
Many of the phenocrysts in this study also show evidence for partial dissolution (resorption) during transport to the surface, resulting in rounded crystal surfaces. This is illustrated by the plagioclase phenocrysts of sample Mo-25D/1, which show abundant rounded surfaces of plagioclase crystals and also rounded large melt inclusions within these phenocrysts. This suggests that the large phenocrysts were in disequilibrium with the melt during magma ascent. The melt inclusions were probably trapped during a period of rapid growth, forming a skeletal structure, and later modified by resorption at the surface of the inclusions, forming the amoeboidal shape (Figs 5 and 13). This latter process is especially important in growth surfaces or volumes within phenocrysts of relatively low An composition, which may dissolve because the temperature rises above the liquidus temperature for such a plagioclase composition (Stewart & Fowler, 2001). This is in accordance with the observations that melt inclusions in the Mo-25D/1 plagioclase crystals are concentrated in the more albitic part of the crystals, which is up to 6% more albitic than the melt-inclusion-free parts (Fig. 13b and d).
The relationship between plagioclase composition, crystal morphology, and magma reservoir dynamics
First-order zoning within plagioclase crystals, observed as long-wavelength variations of up to 10% An, are characteristic of Mohns and Knipovich PUBs. These first-order zones are further subdivided into dense second-order zoning observed as compositional zig-zag variations with wavelengths of <20 µm and variations in the plagioclase composition commonly <2% An. The second-order zoning seems to be linked to a dense optical oscillatory zoning pattern with wavelengths of <20 µm. Several workers have suggested that oscillatory zoning is controlled by near-equilibrium growth when a boundary layer, depleted in the main components of the crystal, forms close to the surface (Pringle et al., 1974; Haase et al., 1980; Anderson, 1984; Singer et al., 1995; Bottinga et al., 1966). The oscillatory zoning observed in some of the phenocrysts therefore suggests longer periods when the magma reservoirs were crystallizing at low degrees of undercooling.
The long-wavelength zoning is characterized by multiple reverse- to evenly zoned parts separated by narrow bands with normal zoning that are rich in melt inclusions. Large compositional variations at long wavelengths in plagioclase crystals are generally interpreted to reflect either non-equilibrium compositional changes of the magma, for example, due to magma mixing or fractional crystallization (Panjasawatwong et al., 1995), or changes in the physical parameters, for example, changing pressure (Longhi et al., 1993; Panjasawatwong et al., 1995), or variations in temperature (e.g. Lofgren, 1974). As the pressure dependence of plagioclase composition is of the order of 1% An/kbar (Pringle et al., 1974; Longhi et al., 1993), it seems unlikely that the large-scale zones in the crystals (commonly higher than 1% An, both normal and reversed zoning) result from pressure effects.
Temperature and compositional effects are difficult to separate. Magma mixing will lead to phenocrysts experiencing both rapid cooling (if they were growing in the hotter magma) and heating (if originally present in the cooler magma) in a changing chemical environment. Only in the situation where small magma pockets are rapidly cooled in marginal zones of the reservoir by the much colder roof or walls might cooling occur without concomitant bulk chemical change.
The normally zoned parts of the phenocrysts studied here are rich in melt inclusions, suggesting rapid growth during strong undercooling. This is further supported by two lines of evidence: (1) the more sodic nature of these zones, as strong undercooling is known to lead to the crystallization of albite-rich plagioclase (Lofgren, 1974); (2) their higher MgO concentrations (Fig. 15), as incompatible elements in plagioclase (MgO, FeO, etc.) are thought to concentrate in a boundary layer at the crystal–melt interface during rapid growth (Bottinga et al., 1966), and may be incorporated into plagioclase by continued growth. The long-wavelength reverse- to evenly zoned areas suggest, on the other hand, constant to gradually higher temperatures during crystallization, perhaps coupled with gradually more primitive magma compositions with higher Ca-numbers. This can be explained by replenishment and gradual hybridization of the reservoir. Plagioclase crystallization at low degrees of undercooling and at slow rates is expected during such temperature increases, consistent with a general lack of melt inclusions in these zones.
Taken together, the long-wavelength even to reverse zoning of crystals and the evidence for resorption discussed above suggest that cyclic replenishment and mixing is a characteristic feature of the magma reservoirs that produced these PUBs. The thin long-wavelength normal zones with melt inclusions and skeletal crystals suggest high degrees of undercooling and short-term rapid cooling events. The cyclic evolution recorded by the plagioclase crystals suggest long periods of slow heating–mixing, and short cooling events. A similar cyclic evolution of magma reservoirs, with rapid heating or cooling interspersed with longer periods of slower cooling and fractionation has been inferred from the rhythmic layering seen in layered intrusions and ophiolite complexes (e.g. Brown, 1956; Irvine, 1981). In contrast, the zoning pattern of the studied plagioclase phenocrysts records long periods of slow heating or cooling and very short periods of strong undercooling.
The nature of the magma reservoirs and the conduit systems
The plagioclase phenocrysts studied here have grown in dynamic environments, under physical conditions changing between rapid cooling events, slow cooling, and slow reheating. This points towards open, periodically replenished magma reservoirs, combined with a process leading to intermittent, rapid undercooling of the magma.
The high modal plagioclase content and the range of crystal morphologies suggest that the erupted crystals represent several generations of phenocrysts that mixed with the host melt prior to eruption. The high crystal content implies eruption from regions of the reservoir where crystals had accumulated. The viscosity of magmas increases by several orders of magnitude when the content of phenocrysts exceeds 40–50%, as crystals start to interact. The viscosity increase is larger for tabular crystals such as plagioclase (Sato, 2005), and a crystal mush dominated by plagioclase phenocrysts may, therefore, behave more or less rigidly and have the character of a partially frozen mush-zone (Cullen et al., 1989; Hansen & Grönvold, 2000).
The mixed plagioclase phenocryst assemblages typical of these PUBs comprise both highly resorbed phenocrysts that indicate disequilibrium between the crystals and the melt, and melt inclusion-rich zones and skeletal crystals that probably formed during rapid crystal growth caused by high degrees of undercooling. Clearly, these crystals have experienced different thermal histories and must have been spatially separated before they were brought together during eruption. They may represent a collection of crystals that aggregated at different levels in the conduit systems, or were brought together as a result of multiple mixing of phenocryst-loaded magmas.
It is essential to determine the cause of the intermittent cooling events, if we are to understand the cyclic pattern recorded in these crystals. Magma convection could transport the crystals into cooler regions of the reservoir, resulting in intermittent undercooling. Hydrothermal systems probably play an important role in the cooling of magma reservoirs at spreading centres, and intermittent changes in the hydrothermal systems may possibly also result in fluctuations in the degree of undercooling of the magma. The inclusion-rich bands showing normal zoning probably reflect short-term, abrupt cooling events (hours to days). It may be questioned if the above mechanism would give rise to such short-term fluctuations in the degree of undercooling. Advances of magma into cooler regions of the crust would be expected to result in abrupt and strong undercooling of the magma, as the temperature difference between the wall-rocks and the magma is high, and the volume of melt relative to the surface of the conduit wall may be low.
Our favoured model is, therefore, that the cyclic zoning pattern shown by many of the studied plagioclase crystals formed as a result of abrupt advances of crystal-rich magma into cool regions followed by longer periods (hundreds of years?) of gradual magma inflow and temperature increase (Fig. 16). This would also well explain the mixing of crystals formed under different degrees of undercooling. The plagioclase phenocrysts may accordingly reveal how the magma conduit propagates to the surface; up to four major magma advances may have been recorded in some of the crystals. Each advance may be triggered by episodes of major magma replenishment into crystal-rich magma reservoirs.
|
Relationship between PUB formation and the degree of melting
Basalts from the northern Knipovich Ridge show distinct along-axis geochemical variations that have been attributed to higher degrees of melting below magmatic segment centres (Hellevang & Pedersen, 2005
). The magmatic segment centres are observed as the shallowest regions within the rift valleys of the ridges, defining axial volcanic ridges (AVRs). These ridges are spaced at 30–100 km along the Mohns and Knipovich ridges and are commonly separated by deep troughs (Okino et al., 2002; Hellevang, 2003; Hellevang & Pedersen, 2005). Mantle Bouguer anomaly lows, relatively strong positive magnetic anomalies, and increased thickness of the extrusive sequence on the AVRs suggest that these ridges are centres of enhanced volcanic and magmatic activity (Klingelhöfer et al., 2000a; Okino et al., 2002). PUBs were sampled only close to the AVRs, and plagioclase ultraphyric magmas appear preferentially to erupt in areas with higher magmatic activity along the ridge.
The PUBs show lower concentrations of the more incompatible trace elements such as K2O and exhibit more depleted REE patterns than the aphyric basalts (Figs 8 and 9). Sr- and Nd-isotope data confirm that the PUBs formed from a slightly more depleted mantle source than the aphyric basalts that were sampled from the same areas. The trace element and the isotope data indicate that the PUBs may have been derived from a separate, more depleted mantle source than the mantle source of the aphyric basalts. However, the trace element and isotope systematics are also compatible with both types being derived from the same heterogeneous source, with the PUB magmas forming by the highest degrees of melting (Hellevang & Pedersen, 2005).
The PUBs are more evolved than their aphyric counterparts, which is compatible with longer residence times in the conduit systems (e.g.; Flower, 1981; Sinton & Detrick, 1992; Niu & Batiza, 1993; Bideau & Hekinian, 1995). The composition and morphology of the plagioclase phenocrysts show that the PUB magmas were repeatedly close to solidifying. This combined with steady or increasing anorthite contents from cores to rims may, as discussed above, record the development of magma conduits by intermittent propagation of magma into cooler regions. These observations can be explained in terms of the formation of a stacked sill complex, as illustrated schematically in Fig. 16.
The spatial and temporal relations between the aphyric lavas and the PUBs are poorly constrained. The aphyric basalts are by far the most abundant type of the two in dredged samples. It is unclear what the history recorded by the plagioclase megacrysts tells about magmatic accretion at ultraslow-spreading ridges in general. One possible relation between the two may be that the PUBs form initially in the magmatic cycles as the melt propagates slowly to the surface and establishes new conduits; and that the aphyric basalts erupt later by more rapid extraction of magmas formed by lower degrees of melting through the already established conduit systems.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMOHNS AND KNIPOVICH RIDGESSEAFLOOR SAMPLINGPETROGRAPHYGEOCHEMISTRYDISCUSSIONCONCLUSIONSREFERENCES |
|---|
- Plagioclase ultraphyric basalts sampled from the ultraslow-spreading Arctic Mid-Ocean Ridge contain up to 54% of plagioclase phenocrysts. Large variations in crystal morphologies and thermal history show that single basalt samples contain a mixture of plagioclase crystals that must have aggregated at different levels in the magma conduits.
- PUBs were preferentially sampled at, or close to segment centres, where their parental magmas formed by higher degrees of melting and evolved more than their aphyric counterparts. The PUBs appear accordingly to be associated with the most robust and long-lived magmatic systems at these ultraslow-spreading ridges. ‘Robust and long-lived’, however, are relative terms. The composition and morphology of the phenocrysts signal that even at the magmatically most active parts of the ridges, the magma conduits repeatedly were at the verge of freeze-up.
- The plagioclase crystals of the PUBs appear to have formed under physical conditions changing between rapid cooling events (hours to days) and slower reheating. Multiple, narrow zones with normal zoning that are rich in trapped melt inclusions are explained by intermittent short periods with strong undercooling and rapid crystal growth, whereas wider zones with reverse zoning (increasing An contents) are explained by temperature increases during magma replenishment.
- We propose that this cyclic zoning pattern results from multiple abrupt advances of crystal-rich magma into cool regions followed by longer periods of gradual magma inflow and temperature increase. This pattern may reflect the development and flow of magma through a stacked sill complex.
- The lower incompatible trace element contents of the PUBs relative to the aphyric basalts suggest that the porphyritic magmas formed by the highest degrees of melting. We speculate that the formation of the PUB magmas triggered the formation of conduit systems, allowing a subsequent extraction to the surface of lower-degree melts from the mantle.
|
ACKNOWLEDGEMENTS |
|---|
This work was supported by the Norwegian Research Council. We thank Ole Tumyr and Yuval Ronen at the University of Bergen for help with microprobe and isotope analyses, and Mike Tubrett at Memorial University of Newfoundland for help with REE analyses. We also thank Colin Devey, Laurence Coogan, Astri Kvassnes and an anonymous reviewer for constructive reviews of the manuscript.
*Corresponding author. E-mail: bjarte.hellevang{at}ri.reslab.no
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMOHNS AND KNIPOVICH RIDGESSEAFLOOR SAMPLINGPETROGRAPHYGEOCHEMISTRYDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Anderson AT. Probable relations between plagioclase zoning and magma dynamics, Fuego Volcano, Guatemala. American Mineralogist (1984) 69:660–676.[Abstract]
Batiza R, Niu Y. Petrology and magma chamber processes at the East Pacific Rise 9°30'N. Journal of Geophysical Research (1992) 97(B5):6779–6797.
Bideau D, Hekinian R. A dynamic model for generating small-scale heterogeneities in ocean floor basalts. Journal of Geophysical Research (1995) 100:10141–10162.[CrossRef]
Bottinga Y, Kudo A, Weill D. Some observations on oscillatory zoning and crystallisation of magmatic plagioclase. American Mineralogist (1966) 51:792–806.[Web of Science]
Brown GM. The layered ultrabasic rocks of Rhum, Inner Hebrides. Philosophical Transactions of the Royal Society of London, Series B (1956) 240:1–53.
Campbell IH, Roeder PL, Dixon JM. Plagioclase buoyancy in basaltic liquids as determined with a centrifuge furnace. Contributions to Mineralogy and Petrology (1978) 67:369–377.[CrossRef][Web of Science]
Cannat M. How thick is the magmatic crust at slow spreading oceanic ridges? Journal of Geophysical Research (1996) 101:2847–2857.[CrossRef]
Cashman KV. Relationship between plagioclase crystallization and cooling rate in basaltic melts. Contributions to Mineralogy and Petrology (1993) 113:126–142.[CrossRef][Web of Science]
Coogan LA, MacLeod CJ, Dick HJB, Edwards SJ, Kvassnes A, Natland JH, Robinson PT, Thompson G, O’Hara MJ. Whole-rock geochemistry of gabbros from the Southwest Indian Ridge: constraints on geochemical fractionations between the upper and lower oceanic crust and magma chamber processes at (very) slow-spreading ridges. Chemical Geology (2001) 178:1–22.[CrossRef][Web of Science]
Crane K, Doss H, Vogt P, Sundvor E, Cherkashov G, Poroshina I, Joseph D. The role of the Spitsbergen shear zone in determining morphology, segmentation and evolution of the Knipovich Ridge. Marine Geophysical Researches (2001) 22:153–205.[CrossRef][Web of Science]
Cullen A, Vicenzi E, McBirney AR. Plagioclase-ultraphyric basalts of the Galapagos Archipelago. Journal of Volcanology and Geothermal Research (1989) 37:325–337.[CrossRef][Web of Science]
Dauteuil O, Brun JP. Oblique rifting in a slow-spreading ridge. Nature (1993) 361:145–148.[CrossRef]
Dick HJB, Natland JH, Alt JC, et al. A long in situ section of the lower ocean crust: results of ODP Leg 176 drilling at the Southwest Indian Ridge. Earth and Planetary Science Letters (2000) 179:31–51.[CrossRef][Web of Science]
Dick HJB, Lin J, Schouten H. An ultraslow-spreading class of ocean ridge. Nature (2003) 426:405–412.[CrossRef]
Eldholm O, Karasik AM, Reksnes PA. The North American Plate boundary. In: The Arctic Ocean Region. The Geology of North America L—Grantz L, Sweeney JF, eds. (1990) Boulder, CO: Geological Society of America. 171–184.
Flower MFJ. Accumulation of calcic plagioclase in ocean-ridge tholeiite: an indication of spreading rate? Nature (1980) 287:530–532.[CrossRef]
Flower MJF. Thermal and kinematic control on ocean-ridge magma fractionation: contrasts between Atlantic and Pacific spreading axes. Journal of the Geological Society, London (1981) 138:695–712.
Gaetani GA, Watson EB. Open system behaviour of olivine-hosted melt inclusions. Earth and Planetary Science Letters (2000) 183:27–41.[CrossRef][Web of Science]
Géli L, Renard V, Rommevaux C. Ocean crust formation processes at very slow spreading centers: A model for the Mohns Ridge, near 72°N, based on magnetic, gravity, and seismic data. Journal of Geophysical Research (1994) 99:2995–3013.[CrossRef]
Haase CS, Chadam J, Feinn D, Ortoleva P. Oscillatory zoning in plagioclase feldspar. Science (1980) 209:272–274.
Hansen H, Grönvold K. Plagioclase ultraphyric basalts in Iceland: the mush of the rift. Journal of Volcanology and Geothermal Research (2000) 98:1–32.[CrossRef][Web of Science]
Hellevang B. Segmentation and crustal accretion along the ultra-slow spreading Mohns and Knipovich ridges, Norwegian–Greenland Sea. (2003) University of Bergen. Doctoral thesis.
Hellevang B, Pedersen RB. Magmatic segmentation of the northern Knipovich Ridge: Evidence for high-pressure fractionation at an ultraslow spreading ridge. Geochem. Geophys. Geosyst. (2005) 6:Q09007.[CrossRef]
Housh TB, Luhr JF. Plagioclase–melt equilibria in hydrous systems. American Mineralogist (1991) 76:477–492.[Abstract]
Irvine TN. Terminology for layered intrusions. Journal of Petrology (1981) 23:127–162.[Web of Science]
Jenner GA, Longerich HP, Jackson SE, Fryer BJ. ICP-MS—a powerful tool for high-precision trace-element analysis in earth sciences—evidence from analysis of selected USGS reference samples. Chemical Geology (1990) 83:133–148.[CrossRef][Web of Science]
Jokat W, Ritzmann O, Schmidt-Aursch MC, Drachev S, Gauger S, Snow J. Geophysical evidence for reduced melt production on the Arctic ultraslow Gakkel mid-ocean ridge. Nature (2003) 423:962–965.[CrossRef]
Kirkpatrick RJ, Klein L, Uhlmann DR, Hays JF. Rates and processes of crystal growth in the system anorthite–albite. Journal of Geophysical Research (1979) 84:3671–3676.
Klingelhöfer F, Géli L, Matias L, Steinsland N, Mohr J. Crustal structure of a super-slow spreading centre: a seismic refraction study of Mohns Ridge, 72°N. Geophysical Journal International (2000a) 141:509–526.[CrossRef][Web of Science]
Klingelhöfer F, Géli L, White RS. Geophysical and geochemical constraints on crustal accretion at the very-slow spreading Mohns Ridge. Geophysical Research Letters (2000b) 27:1547–1550.[CrossRef][Web of Science]
Kuo L-C, Kirkpatrick R. Pre-eruption history of phyric basalt from DSDP Legs 45 and 46: evidence from morphology and zoning patterns in plagioclase. Contributions to Mineralogy and Petrology (1982) 79:13–27.[CrossRef][Web of Science]
le Roex AP, Frey FA, Richardson SH. Petrogenesis of lavas from the AMAR Valley and Narrowgate region of the FAMOUS Valley, 36–37N on the Mid-Atlantic Ridge. Contributions to Mineralogy and Petrology (1996) 124:167–184.[CrossRef][Web of Science]
le Roux PJ, le Roex AP, Schilling J.-G. Crystallization processes beneath the southern Mid-Atlantic Ridge (40–55°S), evidence for high-pressure initiation of crystallization. Contributions to Mineralogy and Petrology (2002) 142:582–602.[Web of Science]
Lofgren G. An experimental study of plagioclase crystal morphology: isothermal crystallization. American Journal of Science (1974) 274:243–273.[Abstract]
Longhi J, Fram MS, Vander Auwera J, Montieth JN. Pressure effects, kinetics, and rheology of anorthositic and related magmas. American Mineralogist (1993) 78:1016–1030.[Abstract]
Loomis TP. Numerical simulations of crystallisation processes of plagioclase in complex melts: the origin of major and oscillatory zoning in plagioclase. Contributions to Mineralogy and Petrology (1982) 81:219–229.[CrossRef][Web of Science]
Marsh BD. Crystal size distribution (CSD) in rocks and the kinetics and dynamics of crystallization. I. Theory. Contributions to Mineralogy and Petrology (1988) 99:277–291.[CrossRef][Web of Science]
Mendel V, Sauter D, Parson L, Vanney J.-R. Segmentation and morphotectonic variations along a super slow-spreading center: the Southwest Indian Ridge (57°E–70°E). Marine Geophysical Researches (1997) 19:505–533.[CrossRef][Web of Science]
Michael PJ, Langmuir CH, Snow JE, Goldstein SL, Graham DW, Lehnert K, Kurras G, Jokat W, Muhe R, Edmonds HN. Magmatic and amagmatic seafloor generation at the ultraslow-spreading Gakkel ridge, Arctic Ocean. Nature (2003) 423:956–961.[CrossRef]
Nielsen RL, Crum J, Bourgeois R, Hanscall K, Forsythe LM, Fisk MR, Christie DM. Melt inclusions in high-An plagioclase from the Gorda Ridge: an example of the local diversity of MORB parent magmas. Contributions to Mineralogy and Petrology (1995) 122:34–50.[CrossRef][Web of Science]
Niu Y, Batiza R. Chemical variation trends at fast and slow spreading mid-ocean ridges. Journal of Geophysical Research (1993) 98:7887–7902.
Okino K, Curewitz D, Asada M, Tamaki K, Vogt P, Crane K. Preliminary analysis of the Knipovich Ridge segmentation: influence of focused magmatism and ridge obliquity on an ultraslow spreading system. Earth and Planetary Science Letters (2002) 202:275–288.[CrossRef][Web of Science]
Panjasawatwong Y, Danyushevsky LV, Crawford AJ, Harris KL. An experimental study of the effects of melt composition on plagioclase–melt equilibria at 5 and 10 kbar: implications for the origin of magmatic high-An plagioclase. Contributions to Mineralogy and Petrology (1995) 118:420–432.[CrossRef][Web of Science]
Parson L, Sauter D, Mendel V, Patriat P, Searle R. Evolution of the axial geometry of the Southwest Indian Ridge between the Melville Fracture Zone and the Indian Ocean Triple Junction—Implications for segmentation on very slow-spreading ridges. Marine Geophysical Researches (1997) 19:535–552.[CrossRef][Web of Science]
Patriat P, Sauter D, Munschy M, Parson L. A survey of the Southwest Indian Ridge axis between Atlantis II Fracture Zone and the Indian Ocean Triple Junction: regional setting and large scale segmentation. Marine Geophysical Researches (1997) 19:457–480.[CrossRef][Web of Science]
Pringle GJ, Trembath LT, Pajari GE, Jr. Crystallisation history of a zoned plagioclase. Mineralogical Magazine (1974) 39:867–877.[CrossRef][Web of Science]
Robinson CJ, Bickle MJ, Minshull TA, White RS, Nichols ARL. Low degree melting under the Southwest Indian Ridge: the roles of mantle temperature, conductive cooling and wet melting. Earth and Planetary Science Letters (2001) 188:383–398.[CrossRef][Web of Science]
Roeder PL, Emslie RF. Olivine–liquid equilibrium. Contributions to Mineralogy and Petrology (1970) 29:275–289.[CrossRef][Web of Science]
Sato H. Viscosity measurement of subliquidus magmas: 1707 basalt of Fuji volcano. Journal of Mineralogical and Petrological Sciences (2005) 100:133–142.[CrossRef][Web of Science]
Singer BS, Dungan MA, Layne GD. Textures and Sr, Ba, Mg, Fe, K, and Ti compositional profiles in volcanic plagioclase: Clues to the dynamics of calc-alkaline magma chambers. American Mineralogist (1995) 80:776–798.[Abstract]
Sinton CW, Christie DM, Coombs VL, Nielsen RL, Fisk M. Near-primary melt inclusions in anorthite phenocrysts from the Galapagos Platform. Earth and Planetary Science Letters (1993) 119:527–537.[CrossRef][Web of Science]
Sinton JM, Detrick RS. Mid-ocean ridge magma chambers. Journal of Geophysical Research (1992) 97:197–216.
Sleep NH, Barth GA. The nature of oceanic lower crust and shallow mantle emplaced at low spreading rates. Tectonophysics (1997) 279:181–191.[CrossRef][Web of Science]
Stewart ML, Fowler AD. The nature and occurrence of discrete zoning in plagioclase from recently erupted andesitic volcanic rocks, Montserrat. Journal of Volcanology and Geothermal Research (2001) 106:243–253.[CrossRef][Web of Science]
Sun S-S, McDonough WF. Chemical and isotopic systematics of ocean basalts: implications for mantle composition and processes. Magmatism in the Ocean Basins. Geological Society, London, Special Publications—Saunders AD, Norry MJ, eds. (1989) 42:313–345.
Swanson SE. Relation of nucleation and crystal-growth rate to development of granitic textures. American Mineralogist (1977) 62:966–978.[Abstract]
Tepley FJI, Davidson JP, Tilling RI, Arth JG. Magma mixing, recharge and eruption histories recorded in plagioclase phenocrysts from El Chichon Volcano, Mexico. Journal of Petrology (2000) 41:1397–1411.
White RS, McKenzie D, O’Nions KR. Oceanic crustal thickness from seismic measurements and rare earth element inversions. Journal of Geophysical Research (1992) 97:19683–19715.[CrossRef]
White RS, Minshull TA, Bickle MJ, Robinson CJ. Melt generation at very slow-spreading oceanic ridges: constraints from geochemical and geophysical data. Journal of Petrology (2001) 42:1–26.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||