| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology | Volume 43 | Number 4 | Pages 607-629 | 2002
© Oxford University Press 2002
Evolution of Crystallizing Interstitial Liquid in an Arc-Related Cumulate Determined by LA ICP-MS Mapping of a Large Amphibole Oikocryst
GEOVETARCENTRUM, GÖTEBORGS UNIVERSITET, BOX 460, SE-405 30, GÖTEBORG, SWEDEN
Received November 27, 2000; Revised typescript accepted October 15, 2001
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONGEOLOGIC SETTING AND...METHODS AND RESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
High-resolution chemical mapping of a large, single, oikocrystic amphibole grain from the Eriksberg gabbro, by laser ablation inductively coupled plasma mass spectrometry (LA ICP-MS), documents the progressive evolution of interstitial liquid in a hydrous basaltic system. The amphibole has a nearly constant mg-number and only minor variations in most major elements, in part because of the resorption of olivine buffering the liquid composition. However, non-buffered trace elements including Sc, V, Nb, Zr, Th, U, and the rare earth elements (REE) show at least an order of magnitude variation. For example, V varies from <1 to >1000 ppm, Zr from 4·3 to >1000 ppm, Nb from 0·14 to 12·3 ppm, and the REE patterns range from bowed up with a negative Eu anomaly to bowed down with a positive Eu anomaly—all within a single crystal. The distribution of amphibole compositions indicates that the interstitial liquid was not uniformly distributed as crystallization proceeded. Rather, the compositional variations reflect progressively more channelized flow of interstitial liquid during compaction. When the interstitial liquid evolved so that the crystallizing amphibole had
300 ppm V, vapor saturation was reached and the behavior of the incompatible trace elements changed markedly. The final amphibole to crystallize is enriched in soluble incompatible elements and depleted in Zr and Nb—a relationship that is consistent with crystallization from a fluid. Where compaction returns such evolved liquid or fluid to an overlying magma reservoir, resulting fractionation trends will show the relative depletions in Ti, Nb, and Zr seen in arc-related basalts. KEY WORDS: interstitial amphibole; cumulates; arc magmatism; trace elements; LA ICP-MS
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGIC SETTING AND...METHODS AND RESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The most commonly studied cumulus rocks are those formed by crystallization of relatively dry basaltic liquids (e.g. Parsons, 1986
; Cawthorn, 1996). These include most layered intrusions, ophiolites, oceanic cumulates, and most cumulates recovered from the Moon and in meteorite collections. Studies of these cumulates have included efforts to understand how the crystals are segregated from the liquid (cumulus processes), as well as how the resulting pile of crystals plus liquid evolves during slow cooling (postcumulus processes). By comparison, very little effort has been directed toward understanding these processes in the crystallization of relatively hydrous basaltic liquids such as those associated with arc magmatism. The presence of a significant amount of water in the liquid is especially important for the postcumulus evolution of a crystal pile, as it will stabilize hydrous minerals and cause vapor saturation while a significant amount of interstitial liquid remains.
The importance of fractional crystallization in producing compositional variability in crystallizing silicate liquids has been well established since Bowen’s experimental work (Bowen, 1928). During perfect fractional crystallization, the solids are isolated from the liquid before the liquid evolves appreciably and thus they provide an instantaneous record of the liquid evolution. Incorporation of liquid in a crystal pile can change both the composition of the solids and the evolution of the supernatant liquid (initially the parental liquid). If the interstitial liquid crystallizes where it was initially incorporated in the crystal pile (closed-system crystallization in the crystal pile), then only the solids will record the presence of this liquid. However, if the interstitial liquid is returned to the supernatant liquid by compaction for example (open-system crystallization in the crystal pile), then its compositional evolution can be modified. This is because the evolution of the interstitial liquid can be buffered by, or react with, the solid assemblage (Meurer & Boudreau, 1998). Simplified versions of this process have been considered theoretically (O’Hara & Fry, 1996a, 1996b) and modeled for the fractionation of the Kiglapait intrusion (Langmuir, 1989). The chemical consequences of open-system crystallization could be pronounced for arc cumulates as this process will involve amphibole fractionation (in the crystal pile) much earlier in the crystallization of a batch of magma than would perfect fractional crystallization.
Here we use the chemical variation of a large oikocrystic amphibole from the Eriksberg gabbro of southern Sweden to model the chemical evolution of interstitial liquid in an arc-related olivine–plagioclase cumulate. Textural relations suggest that amphibole joined the crystallizing assemblage of the interstitial liquid at the expense of olivine and possibly plagioclase. The compositional evolution of the interstitial liquid is reflected in the variations in the composition of the amphibole, determined by laser ablation inductively coupled plasma mass spectrometry (LA ICP-MS). Similar chemical variations in amphibole from elsewhere in the Eriksberg gabbro and from the Rymmen gabbro (southern Sweden) are indicated by TiO2 variations, suggesting the processes described are general features of arc cumulates.
|
GEOLOGIC SETTING AND PETROGRAPHIC RELATIONS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGIC SETTING AND...METHODS AND RESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The Eriksberg gabbro is located in the Småland–Värmland granitoid batholith of southern Sweden (Fig. 1). It preserves continental-arc cumulates related to the formation of the Transscandinavian Igneous Belt during the Proterozoic, 1·79 Ga (Claeson, 2001
). The Eriksberg gabbro and the surrounding Filipstad granite are undeformed and are interpreted to have been essentially unaffected by regional metamorphism (Jarl & Johansson, 1988; Claeson, 2001). Rock types in the Eriksberg gabbro range from troctolite to leuco-tonalite, with postcumulus (oikocrystic) amphibole present in nearly all of the earliest formed cumulates, and cumulus (granular) amphibole found in some of the more evolved rocks.
|
This study focuses on a single thin-section from a primitive portion of the Eriksberg gabbro. It is composed of cumulus plagioclase (46%) and olivine (28%), and intercumulus clinoamphibole (20%), orthopyroxene (4%), and clinopyroxene (2%) with trace amounts of igneous biotite, orthoamphibole, ilmenite–magnetite, and apatite (Fig. 2). Igneous orthoamphibole is rare and its occurrence has been described in detail elsewhere (Claeson & Meurer, 2002), and as it is not central to this study we refer to the clinoamphibole simply as amphibole. Alteration of the sample is restricted to partial serpentinization of olivine and limited saussuritization of plagioclase.
|
The thin-section is divided into three regions based on the dominant oikocrystic mineral. One end of the slide has a large clinopyroxene oikocryst that fills most of the interstitial space but is almost always separated from the olivine and plagioclase by a thin layer of amphibole. All of the clinopyroxene in the section is optically continuous, and the textures suggest it grew before the amphibole. However, a very limited amount of clinopyroxene may have been converted to amphibole by reaction with the liquid, but textural and chemical evidence to support this interpretation is ambiguous at best. We refer to this portion of the slide as the ‘clinopyroxene region’ (of the slide). At the opposite end of the slide is a region dominated by a large orthopyroxene oikocryst with smaller oikocrystic orthopyroxene crystals near it. This region is nearly free of amphibole and is therefore referred to as the ‘dry region’. Separating these two regions is an area that is essentially free of any interstitial phase except amphibole. Greater than 85% of the biotite in the slide and all of the orthoamphibole occur in a small area near the dry region. Five apatite grains, all <150 µm in maximum dimension, were located using a combined backscatter–X-ray mapping approach [described by Meurer & Boudreau (1996)]. Four of these occur near the margin of the dry region, and the other occurs in the middle of the slide, in an area with large plagioclase crystals (Fig. 2).
|
METHODS AND RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGIC SETTING AND...METHODS AND RESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Major-element compositions for plagioclase, olivine, clinopyroxene, orthopyroxene, amphibole, and biotite were determined by scanning electron microscopy (SEM) using a Zeiss DSM 940 system with a quantitative energy-dispersive system (EDS) from Link at the Earth Sciences Center, Göteborgs Universitet. An excitation voltage of 25 kV, sample current of 1 nA, and live counting time of 100 s were used. Additional analyses of amphibole and biotite, and analyses of apatite, were collected at Duke University using a Cameca Camebax system. Operating conditions were 15 kV excitation voltage and 15 nA beam current. In both instances natural minerals were used as standards. The procedures outlined by Meurer & Boudreau (1996)
were used to optimize analyses of the halogens. Trace-element analyses were collected using a Cetac ASX-200 Nd–YAG UV laser for sampling and an HP-4500 quadrupole ICP-MS system for analysis at Göteborgs Universitet. Analyses were conducted in both spot and line-traverse modes and beam diameters of 50–200 µm were used, with minimum counting times of 30 s and maximum times of 160 s. Because a standard thin-section with a thickness of 30 µm was sampled, all analyzed spots had depth:width <1. Most elements were blank corrected using averages of 200 s gas-blanks collected throughout the period of analyses. Si and Ca show systematic drift during runs and were corrected using a 20 s gas-blank collected at the start of each analysis. The raw counts were normalized to both 29Si and 44Ca (when possible). Both normalizations give similar results so the 29Si data are used for consistency. The normalized counts were divided into three time blocks that were averaged and reduced then used for assessing homogeneity and analysis quality. NIST 612 glass was used as a calibration.
Assessing the quality of laser ICP-MS analyses is complicated by the lack of homogeneous, trace-element mineral standards. This is problematic because minerals ablate with different efficiencies so that for the same concentration of an element in a mineral that ablates easily (e.g. plagioclase), the signal can be substantially stronger than for a mineral that is more resistant (e.g. olivine). An additional complication is introduced when different spot sizes and energies for the laser are used. In the present study we have used 5 mJ for all analyses, but spot sizes have been varied both within and between minerals. To assess the accuracy and precision of the method used, we have collected 20 replicate analyses of the NIST 612 and 614 glasses using the same laser settings as for most of the unknowns. For the NIST 612 glass (with trace-element concentrations of 40 ppm; see Table 5, below) the percent relative standard deviation (%RSD) is 
5 for all elements and for NIST 614 glass (with trace-element concentrations of 1 ppm; see Table 5) the %RSD is 12 (Fig. 3). The average values of the NIST 614 calibrated against NIST 612 are within 2% of the expected values. It should be noted that our laser does not couple well with the colorless NIST 614 glass so ablation efficiencies are lower than for any of the minerals analyzed and therefore the %RSD values are upper limits for these concentrations. On the basis of these tests, and the %RSD of the three time blocks from the mineral analyses, we suggest the following conservative estimates of precision for an element (x) with a given concentration range: x 
10 ppm (5%), 0·25 x < 10 ppm (10%), 0·05 x < 0·25 ppm (20%), < 0·05 ppm (20–50%).
|
|
Olivine and plagioclase show little major-element compositional variability (Table 1). Both are relatively primitive with average Fo = 77 (range 76·6–77·2) and average An = 88 (range 85·0–92·5). Clinopyroxene is magnesian (average mg-number = 87) and relatively uniform in major-element composition (Table 1). Although the mg-number of orthopyroxene shows limited variation (average 79·5; range 78·0–80·6), higher Al2O3 is associated with lower SiO2 and higher FeO* (Table 2). Amphibole also shows some limited major-element variability with positive correlations between SiO2, TiO2, and CaO and all of these being negatively correlated with Al2O3 and Na2O (Table 3). By far, TiO2 shows the largest variation, ranging from near the detection limit (0·02%) to >2·7% and is discussed with the trace-element data. Orthoamphibole and biotite occur in small amounts, are magnesian (average mg-numbers = 75 and 86, respectively), and relatively uniform in major-element composition [see Claeson & Meurer (2002) for details]. Apatite compositions show significant variation only in their halogen contents. One grain that appears to be altered has low total halogens, an expected consequence of low-temperature alteration (Boudreau & McCallum, 1990), whereas the others have XCl > XOH > XF with one sample having XCl 0·85 (Table 4).
|
|
|
|
No effort was made to systematically investigate trace-element compositional variations of all minerals present. Analyses of plagioclase, olivine, and clinopyroxene were collected for comparison and modeling purposes (Table 5). We focused on documenting the compositional variation in amphibole (i.e. the clinoamphibole; Table 6) and to a lesser extent orthopyroxene (Table 5), as they show the largest major-element variations. Trace-element variations in orthopyroxene mimic those in amphibole although in general the concentrations are much lower.
|
The trace-element content of amphibole varies systematically for most elements. Sc, TiO2, and V are positively correlated and show variations that extend over more than two orders of magnitude and over three orders of magnitude in the case of V (Fig. 4a). Because of its high partition coefficient into amphibole (7) and extreme variability V is compared with other trace elements as a way of understanding how they varied during fractionation of amphibole from the liquid. Ni and Co are positively correlated and vary by only a factor of two, but they are not correlated with V (Fig. 4b). Ba and Sr are positively correlated with each other and with V and vary by about an order of magnitude (Fig. 4c). Zr and Nb are not correlated with V in a simple way. Amphibole analyses with >300 ppm V form a coherent positive correlation in Zr–Nb space, whereas analyses with <300 ppm V form a separate and exponentially increasing trend in Zr (Fig. 4d). U and Th show similar relations as Zr and Nb, with samples with >300 ppm V defining a tight group in U–Th space. Those with <300 ppm V span a compositional range including both lower and very much higher U and Th concentrations (Fig. 4e). The rare earth elements (REE) vary systematically with V concentrations. The chondrite-normalized ratio of Ce/Yb plotted against V concentration shows a nearly asymptotic increase as the V concentration tends to zero (Fig. 4f).
|
A detailed examination of the REE patterns as a function of the V content reveals variation in not only the slope of the pattern but in its shape and concentrations as well (Fig. 5). Amphibole analyses with high V have the highest overall normalized REE contents and patterns that are bowed up in the middle with a pronounced negative Eu anomaly (Fig. 5a). As the V concentration decreases the patterns shift downwards, especially the heavy REE (HREE), and the Eu anomaly decreases (Fig. 5b and c). At low V the patterns lose the negative Eu anomaly and show a continuous decrease from Nd in both directions (Fig. 5d). For samples with <300 ppm V, the HREE decrease rapidly. Coupled with a more modest decrease in Nd and Sm, this produces a positive Eu anomaly (Fig. 5e). At the lowest V contents, the decrease in the REE from Gd to Er is larger than that of Yb and Lu, producing an upturn in the end of the patterns (Fig. 5e and f). The three analyses with the least V have patterns with HREE that fall below chondritic concentrations (Fig. 5f).
|
The REE concentrations of the other major phases are all less than those of most amphibole (Fig. 6). Clinopyroxene patterns are similar to the patterns of high-V amphiboles. Plagioclase patterns show limited variability, a pronounced negative slope, and large positive Eu anomaly. Orthopyroxene contents of the REE are near detection limits, especially for the light REE (LREE). A small number of the orthopyroxene patterns have positive slopes and negative Eu anomalies, but most have a ‘W’ shape with La and Ce higher than Nd and Sm, a slight positive Eu anomaly, and a positive slope in the HREE. The shape of these patterns is similar to those of amphiboles with low V contents.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGIC SETTING AND...METHODS AND RESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Assuming the composition of the amphibole reflects that of the crystallizing interstitial liquid, then the distribution of more primitive or evolved amphibole provides insight into how the crystallization of the interstitial liquid proceeded. As discussed below, we interpret the amphibole to have crystallized while interstitial liquid moved through the sample. However, we begin by modeling the amphibole crystallization as though it took place in a closed system, as this model can be used as a measure of how much open-system exchange occurred. The model is based on estimates from Table 7 of the proportions of phases involved in reactions (both crystallization and resorption) after amphibole saturation. These proportions are used to model the trace-element evolution of the system, and the results are compared with the amphibole compositional variation. We then consider the importance of open-system behavior and vapor saturation, and propose a model for the crystallization of the interstitial liquid in this system, and the potential importance of amphibole crystallization on the fractionation path of arc-related liquids. Last, we present preliminary results from additional samples from the Eriksberg and from a similar intrusion (the Rymmen gabbro; Claeson, 2001
). These data and published microanalyses of amphiboles indicate that the large amount of variability documented here is a general feature of interstitial amphibole in cumulates.
|
|
A closed-system trace-element model
We construct a model for the trace-element composition of the amphibole that crystallized during progressive fractionation in a closed system using estimated proportions of phases involved in its crystallization. These are obtained by a mass-balance calculation such that the interstitial liquid plus any resorbed minerals must equal the composition of those minerals that crystallized after amphibole saturated. The crystallized minerals include amphibole, orthopyroxene, and trace amounts of biotite, orthoamphibole, and apatite. The trace minerals are neglected in the model. Experimental studies of amphibole stability in mafic liquids suggest that the onset of amphibole crystallization can be associated with the reaction of olivine and plagioclase with the liquid (Helz, 1991
), and this is confirmed by textural relations (Fig. 7). However, there is also textural evidence that plagioclase began crystallizing later during amphibole crystallization (Table 7). Thus, it is not clear if plagioclase should be included with olivine and liquid on the reactant’s side of the mass balance or with amphibole and orthopyroxene on the product’s side. If plagioclase is taken as a reactant, then the composition of the interstitial liquid required for the mass balance has a low Al2O3 content, which is inconsistent with the crystallization of the observed Al2O3-rich orthoamphibole gedrite (Claeson & Meurer, 2002). Thus, we infer that net plagioclase crystallization took place after amphibole saturated (i.e. more plagioclase ultimately crystallized than was resorbed). The corresponding mass-balance equation for an element j and phase proportions a–e is
 |
The observed mineral compositions can be used in this equation. However, the composition of the interstitial liquid is unknown as are the modal proportions except for amphibole (although these must sum to unity on each side of the equation). The mg-number of the olivine and the amphibole are nearly the same, and the mg-number of the amphibole is essentially invariant (Tables 1 and 3), requiring that essentially all of the Fe and Mg in the amphibole be derived from resorption of olivine. This problem is exacerbated by the amount of Fe3+ in the liquid, but can be remedied if we assume that the initial olivine mg-number was higher than the observed mg-number and was shifted downward by reaction with the liquid (Barnes, 1986).
The derived mass-balance equation is solved iteratively with the following constraints or assumptions: (1) the initial olivine Fe–Mg composition must be in equilibrium with the liquid (Roeder & Emslie, 1970); (2) the liquid Ca–Na content must be in equilibrium with the average plagioclase (Sisson & Grove, 1993); (3) the liquid must have the minimum water and alkali contents needed to stabilize amphibole (Cawthorn & O’Hara, 1976; Helz, 1991); (4) the liquid cannot have more H2O than saturation at mid- to lower-crustal levels (Burnham, 1979); (5) 63% of the observed orthopyroxene crystallized with the amphibole.
The solutions are a family of liquid compositions for a given amount of plagioclase crystallized that constrain the composition on the product’s side. The liquid composition selected is intermediate between primitive high-Al basalts and low-Mg high-Al basalts (Table 8), in accordance with Claeson (2001). The resulting proportions of crystallized phases are 10% plagioclase, 11% orthopyroxene, and 79% amphibole (making up 26% of the final mode), formed by the crystallization of 69% liquid and the resorption of 31% olivine. It should be noted that these proportions are not those of an instantaneous crystallizing assemblage but are the product of the crystallization interval from the point of amphibole saturation until solidification.
|
The trace-element composition of the liquid used in the model is obtained by inverting the olivine, plagioclase, and clinopyroxene compositions, all of which were in equilibrium with the liquid when amphibole first saturated (Table 9). Distribution coefficients (Kd values) for these inversions are from the compilation of Bédard (1994), although we use a lower Eu Kd for plagioclase (KdpluEu = 0·22) to be consistent with the observed plagioclase REE patterns (Fig. 6). Co and Ni concentrations are derived from olivine, alkali and alkaline metals from plagioclase, and all other elements from clinopyroxene except for the REE, which are based on the average of plagioclase and clinopyroxene. The clinopyroxene inversion suggests 35% lower LREE than the plagioclase inversion but essentially the same HREE concentrations. Amphibole Kd values can vary substantially as a function of both solid and liquid composition (Klein et al., 1997; Bottazzi et al., 1999; Hilyard et al., 2000). Therefore we estimate them using the calculated liquid and a primitive amphibole composition, derived by averaging the five analyses with the highest V (Table 9).
|
The trace-element evolution of the system is modeled by removing constant proportions of minerals based on the major-element mass balance from the calculated liquid composition. We modeled the system using two contrasting assumptions: (1) all olivine is resorbed at the onset of amphibole crystallization; (2) olivine is uniformly resorbed throughout the amphibole crystallization. The results presented are for early olivine resorption but continuous resorption produces similar results, with rates of incompatible-element enrichment and compatible-element depletion shifted slightly relative to each other.
The model predicts the amphibole compositions best for the REE (Fig. 8). However, the predicted continuous increase in the LREE is not seen in the amphibole patterns. Progressive fractionation of amphibole depleted the liquid in middle REE (MREE), producing the change from a negative to a positive Eu anomaly. For a closed system, the REE indicate that the most evolved amphibole crystallized with 1% interstitial liquid remaining.
|
Apart from the REE, the closed-system model does not fit the observed compositions well. Compatible elements are predicted to be much more rapidly depleted than the amphibole analyses indicate. For example, V concentrations in the amphibole are expected to fall below 200 ppm after only 20% crystallization; but over 60% of the amphibole analyses contain >200 ppm V. Rapid depletions are predicted for Sc and TiO2 but are not observed. Ni and Co, which show very limited variation and do not correlate with V (Fig. 4b), do not fit with the model. The bulk distribution coefficients for Ba and Sr used in the model are less than unity, suggesting that both elements should be enriched during fractionation. However, both systematically decrease with decreasing V (Fig. 4c). Many of the analyses of the incompatible elements Zr, Th, and U fit the model, but other analyses are either over-enriched for a given V content (U) or, in the case of Zr and Th, over-enriched in an absolute sense. Zr contents are particularly difficult to fit with the closed-system model because concentrations that are both too high and too low are found. Changes to the initial liquid composition can improve the model fit for some elements but cannot resolve many of the misfits.
Open-system behavior
Crystallization of amphibole during compaction or compositional convection within a crystal mush allows open-system chemical behavior (Sparks et al., 1985; Shirley, 1987). The V concentration can be used to distinguish amphibole that crystallized early (high V) from late (low V) and to interpret the progression of crystallization. High V concentrations are found throughout most of the slide, with an area of low concentration found immediately adjacent to the dry region and another near the clinopyroxene region (Fig. 9a). That most of the slide contains high-V amphibole supports the conclusion that the V concentrations are not depleted as rapidly as predicted by the closed-system fractional crystallization model. Whereas V concentrations show definite spatial patterns, Ni, which does not correlate with V, does not (Fig. 9b), as expected if olivine buffers the Ni and Co concentrations in the liquid. Low Sr and Ba concentrations are found in the areas of low V (Fig. 9c) as indicated in Fig. 4c. The REE patterns also show a systematic distribution as expected from their correlation with V concentrations (Fig. 5). Most analyses with small or positive Eu anomalies are found near the dry region or in the low-V cluster near the clinopyroxene region (Fig. 9d). The concentrations and spatial patterns defined by V–Sc–TiO2, Sr–Ba, and the REE are all consistent with crystallization beginning from a larger initial volume of interstitial liquid than is recorded by the amount of interstitial amphibole. The initially larger volume of liquid allowed crystallization of more high-V amphibole than predicted by the closed-system model. Much of the evolved liquid that was produced was expelled during compaction. It is also possible that more anorthositic layers deeper in the crystal pile could have contributed liquid that crystallized higher-V amphibole upon encountering and reacting with the olivine. This is because amphibole stability in basaltic systems is enhanced by reaction with olivine (Helz, 1976, 1991; Anderson, 1980). Thus, liquid that was undersaturated in amphibole in a plagioclase-rich cumulate a few meters below our sample could become amphibole saturated when it encountered olivine. In general, this would enhance the early crystallization of interstitial liquid in portions of the crystal mush that are richer in olivine.
|
If the idea that the amphibole crystallized before and during compaction of a crystal mush is accepted, then the analyses with the highest concentrations of incompatible and lowest concentrations of compatible elements should correspond to the last open pathways for evolved liquids. Four elements (U, Th, Zr, and Nb) that behave incompatibly can be used to test this idea. Spatial patterns of U concentrations are similar to those of Th, with both being the highest near the margin of the dry region (Fig. 9e and f). Th is slightly higher in the area of low V near the clinopyroxene region but U is not. The highest Zr concentrations are all found in the low-V amphibole near the margin of the dry region, but some of the analyses with the lowest V also have the lowest Zr (Fig. 9g). Likewise, both the highest and lowest Nb concentrations are found in the low-V amphibole near the margins of the dry region (Fig. 9h).
Both the V and the incompatible-element concentrations are consistent with much of the area at the margin of the dry region having been a pathway for evolved liquids. We suggest that the orthopyroxene oikocryst acted as a small impermeable region, forcing the flow of interstitial liquid around it. This slightly higher liquid flux allowed the maintenance of a higher permeability pathway during crystallization of amphibole. Thus, much of the compositional variation in the amphibole can be ascribed to crystallization from an increasingly channelized flow of evolving liquid through the crystal mush. However, this model does not explain the low concentrations of Zr and Nb found in low-V amphibole near the clinopyroxene region and near the dry region or the high U and Th concentrations in high-V amphibole near the dry region.
Vapor saturation and return of material to the supernatant liquid
The complex spatial patterns of the incompatible elements and their correlations with high and low V concentrations in the amphibole are best explained by a model that includes a vapor-saturation event as the interstitial liquid evolves to higher water contents. We envision the initial vapor saturation to have produced a Cl-rich supercritical fluid (hereafter simply fluid), as Cl strongly partitions into the vapor phase (Holland, 1972; Kilinc and Burnham, 1972; Candela, 1986). Th and U were strongly partitioned into the fluid whereas Nb and Zr remained in the liquid. Amphibole that crystallized from the most evolved liquid in equilibrium with the fluid has low V and high concentrations of Nb, Zr, Th, and U. Amphibole crystallization from the fluid continued after the liquid was exhausted, producing amphibole with high U and Th but with the lowest Zr and Nb concentrations observed. The fluid was also able to equilibrate with some of the earlier crystallized amphibole near the dry region, producing high-V amphibole with relatively high U and Th, with the larger area of enrichment of U suggesting a chromatographic process (Fig. 9e and f). On the basis of this model, we suggest that the amphibole with low V, low Zr and Nb, and high U and Th near the dry region defines the last portion of the sample to crystallize.
Support for the proposed vapor saturation model is found in the distribution and composition of apatite in the sample. Although only five grains of apatite were found, four of these define a girdle distribution along the margins of the dry region, suggesting that P was concentrated in the upward moving fluid. The high Cl content of these apatite grains (Table 4) is indicative of equilibration with a Cl-rich fluid (Meurer & Boudreau, 1996). The amphibole is not Cl rich, but its high mg-number causes a strong preference for F or OH into the hydroxyl-site, so the amphibole provides a much less simple measure of the volatile ratios than does the apatite (Volfinger et al., 1985).
The importance of open-system crystallization in a cumulus pile to the fractionation path of the supernatant liquid depends upon how much material is returned from the cumulus pile and how compositionally distinct it is. Assuming amphibole saturated with at least 25% interstitial liquid present, then 40–50% fractionation of this liquid would produce a liquid composition sufficiently distinct that its efficient return to the supernatant liquid would have a discernible effect. For the sample studied, retention of the HREE, Sc, TiO2, and V by the amphibole would make these elements appear to be much more compatible in the fractionating assemblage (olivine and plagioclase) than they are. If more extreme differentiates in the crystal pile are considered, then the return of fluid could be important. For example, return of a fluid phase, rich in many incompatible elements, but nearly devoid of Nb and Zr, could contribute to the distinctive depletion of Nb and Zr in arc-related liquids. It has been proposed that amphibole might be involved in producing the characteristic geochemical signature of arc-liquids (Tiepolo et al., 2000). We suggest that its early crystallization from the interstitial liquid could produce this effect in a supernatant liquid that is not saturated in amphibole.
Compositional variability of interstitial amphibole in cumulates
Our preliminary work on additional samples from the Eriksberg and the Rymmen gabbros and the limited number of published microanalytical studies of amphibole indicate that the large variability documented here is a general feature of interstitial amphibole in cumulates. Individual oikocrysts from two additional samples of more primitive cumulates from the Eriksberg have large variations in TiO2 (from <0·01 to 1·35% and from 0·02 to 1·92%). Likewise, analysis of a large oikocryst from the Rymmen gabbro, a gabbroic body similar to the Eriksberg (Claeson, 2001), shows a large range of TiO2 (from <0·01 to 3·12%). The TiO2-rich and -poor portions of these amphiboles show complex spatial patterns (Fig. 10), similar to the amphibole oikocryst described in detail here, and we predict that they have similar trace-element variability. Tribuzio et al. (2000) analyzed amphibole in gabbroic rocks from the Apennines and despite being limited to at most three analyses per sample these data reveal 2–3 times variations in some incompatible elements (most notably Zr and Nb). Gillis & Meyer (2001) examined major element and REE variability in amphibole in oceanic cumulates. Their data reveal over an order of magnitude variability in the REE in some samples. This variability is interpreted to reflect growth of amphibole at different times from interstitial liquid, migrating fluids, and local mineral–mineral reactions, but might simply reflect progressive fractionation and vapor saturation as documented here.
|
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGIC SETTING AND...METHODS AND RESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Laser ablation ICP-MS analyses of a single large amphibole oikocryst reveal substantial spatial variations in its trace-element concentrations, despite its relatively homogeneous major-element composition. This result suggests that caution should be used in characterizing the trace-element concentrations in interstitial amphiboles in cumulates using microanalysis techniques. Using a limited number of analyses or closely spaced analyses may not adequately sample amphibole compositions and has the potential for being misleading.
Large portions of the sample contain relatively primitive amphibole, characterized by high concentrations of compatible elements (e.g. TiO2, V) and low concentrations of incompatible elements (e.g. Zr, U), indicating that much of the amphibole crystallized from a relatively uniform interstitial liquid composition. The initial volume of interstitial liquid was larger than is indicated by the observed proportions of cumulus and intercumulus minerals. This allowed more primitive amphibole to crystallize than expected for a closed system with the inferred proportion of interstitial liquid. Compaction of the sample expelled some of the more evolved liquid. The early growth of a large orthopyroxene oikocryst at one end of the sample created a low-permeability region that focused flow around it, helping to maintain a higher permeability. The spatial distribution of amphibole compositions indicates that the interstitial liquid was not uniformly distributed as crystallization proceeded. Rather, it was progressively concentrated into channels that facilitated migration of liquid at even small volume fractions. The pore space was eventually filled by amphibole, with complex compositional relations between compatible and incompatible elements that reflect both progressive fractionation and elemental partitioning between a silicate liquid and an aqueous fluid.
The crystallization of interstitial amphibole in the crystal pile of hydrous basaltic liquids and the eventual saturation of fluid can result in unexpected modifications of the trace-element evolution of the supernatant liquid where evolved materials are returned. We suggest that crystallization of amphibole in the crystal pile might produce arc-type trace-element characteristics even in liquids not saturated in amphibole.
|
ACKNOWLEDGEMENTS |
|---|
J. H. Bédard, C. E. Lesher, and C. Tegner provided insightful and constructive reviews that greatly improved this manuscript and the ideas presented. The editorial work of D. Geist resulted in a greatly improved manuscript. We thank M. E. S. Meurer and D. Cornell for suggestions and ideas. We specially thank M. E. S. Meurer for comments on numerous drafts of this manuscript. Financial support for this work was provided by NSF grant EAR-9725394 to W.P.M.
|
FOOTNOTES |
|---|
*Corresponding author. Present address: Department of Geosciences, University of Houston, 312 Science and Research Building, 1, Houston, TX 77204-5007, USA. Telephone: +1 731 743 0214. Fax: +1 713 748 7906. E-mail: wpmeurer{at}mail.uh
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGIC SETTING AND...METHODS AND RESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Anderson, A. T., Jr (1980). Significance of hornblende in calc-alkaline andesites and basalts. American Mineralogist 65, 837–851.[Abstract]
Barnes, S. J. (1986). The effect of trapped liquid crystallization on cumulus mineral compositions in layered intrusions. Contributions to Mineralogy and Petrology 93, 524–531.[Web of Science]
Bédard, J. H. (1994). A procedure for calculating the equilibrium distribution of trace elements among the minerals of cumulate rocks, and the concentration of trace elements in the coexisting liquids. Chemical Geology 118, 143–153.
Bottazzi, P., Tiepolo, M., Vannucci, R., Zanetti, A., Brumm, R., Foley, S. F. & Oberti, R. (1999). Distinct site preferences for heavy and light REE in amphibole and the prediction of Amph/LDREE. Contributions to Mineralogy and Petrology 137, 36–45.
Boudreau, A. E. & McCallum, I. S. (1990). Low temperature alteration of REE-rich chlorapatite from the Stillwater complex, Montana. American Mineralogist 75, 687–693.[Abstract]
Bowen, N. L. (1928). The Evolution of the Igneous Rocks. New York: Dover.
Burnham, W. C. (1979). The importance of volatile constituents. In: The Evolution of the Igneous Rocks: Fiftieth Anniversary Perspectives. Princeton, NJ: Princeton University Press, pp. 439–482.
Candela, P. A. (1986). Toward a thermodynamic model for the halogens in magmatic systems: an application to melt–vapor–apatite equilibria. Chemical Geology 57, 289–301.
Cawthorn, R. G. (1996). Layered Intrusions. New York: Elsevier.
Cawthorn, R. G. & O’Hara, M. J. (1976). Amphibole fractionation in calc-alkaline magma genesis. American Journal of Science 276, 309–329.
Claeson, D.T. (2001). Investigation of gabbroic rocks associated with the Småland–Värmland granitoid batholith of the Transscandinavian Igneous Belt, Sweden. Ph.D. thesis, Earth Sciences Centre, Göteborgs Universitet.
Claeson, D.T. & Meurer, W. P. (2002). An occurrence of igneous orthorhombic amphibole; Eriksberg gabbro, southern Sweden. American Mineralogist, in press.
Gillis, K. M. & Meyer, P. S. (2001) Metasomatism of oceanic gabbros by late stage melts and hydrothermal fluids: evidence from the rare earth element composition of amphiboles. G3—Geochemistry, Geophysics, Geosystems, 2, Paper Number 2000GC000087.
Helz, R. T. (1976). Phase relations of basalts in their melting range at PH2O = 5 kb. Journal of Petrology 14, 249–302.
Helz, R. T. (1991). Phase relations and compositions of amphiboles produced in studies of the melting behaviour of rocks. In: Veblen, D. R. & Ribbe, P. H. (eds) Amphiboles: Petrology and Experimental Phase Relations. Mineralogical Society of America, Reviews in Mineralogy 9B, 279–353.
Hilyard, M., Nielsen, R. L., Beard, J. S., Patinõ-Douce, A. & Blencoe, J. (2000). Experimental determination of the partitioning behavior of rare earth and high field strength elements between pargasitic amphibole and natural silicate liquids. Geochimica et Cosmochimica Acta 64(6), 1103–1120.
Holland, H. D. (1972). Granites, solutions and base metal deposits. Economic Geology 67, 281–301.
Jarl, L.-G. & Johansson, Å. (1988). U–Pb zircon ages of granitoids from the Småland–Värmland granite–porphyry belt, southern and central Sweden. GFF 110, 21–28.
Kilinc, I. A. & Burnham, W. C. (1972). Partitioning of chloride between a silicate melt and coexisting aqueous phase from 2 to 8 kilobars. Economic Geology 67, 231–235.
Klein, M., Stosch, H.-G. & Seck, H. A. (1997). Partitioning of high field-strength and rare-earth elements between amphibole and quartz-dioritic to tonalitic melts: an experimental study. Chemical Geology 138, 257–271.
Langmuir, C. H. (1989). Geochemical consequences of the solidification of magma chambers through ‘in situ’ crystallization. Nature 340, 199–205.
Meurer, W. P. & Boudreau, A. E. (1996). Evaluation of models of halogen variations in apatite from layered intrusions using a detailed study of apatite from OB-III and OB-IV of the Middle Banded series, Stillwater complex, Montana. Contributions to Mineralogy and Petrology 125, 225–236.
Meurer, W. P. & Boudreau, A. E. (1998). Compaction of igneous cumulates Part I—whole-rock compositions as an indicator of the trapped liquid proportions in the Stillwater complex, Montana. Journal of Geology 106, 281–292.[Web of Science]
O’Hara, M. J. & Fry, N. (1996a). Geochemical effects of small packet crystallization in large magma chambers—further resolution of the highly compatible element paradox. Journal of Petrology 37, 891–925.
O’Hara, M. J. & Fry, N. (1996b). The highly compatible trace element paradox—fractional crystallization revisited. Journal of Petrology 37, 859–890.
Parsons, I. (1986). Origins of Igneous Layering. NATO ASI Series C: Mathematical and Physical Sciences, 196. Dordrecht: D. Reidel.
Roeder, P. L. & Emslie, R. F. (1970). Olivine–liquid equilibrium. Contributions to Mineralogy and Petrology 29, 275–289.[Web of Science]
Shirley, D. N. (1987). Differentiation and compaction in the Palisades sill. Journal of Petrology 28, 835–865.
Sisson, T. W. & Grove, T. L. (1993a). Experimental investigations of the role of H2O in calc-alkaline differentiation and subduction zone magmatism. Contributions to Petrology and Mineralogy 113, 143–166.[Web of Science]
Sisson, T. W. & Grove, T. L. (1993b). Temperatures and H2O contents of low-MgO high-alumina basalts. Contributions to Mineralogy and Petrology 113, 167–184.[Web of Science]
Sparks, R. S. J., Huppert, H. E., Kerr, R. C., McKenzie, D. P. & Tait, S. R. (1985). Postcumulus processes in layered intrusions. Geological Magazine 122, 555–568.[Abstract]
Tiepolo, M., Vannucci, R., Oberti, R., Foley, S., Bottazzi, P. & Zanetti, A. (2000). Nb and Ta incorporation and fractionation in titanian pargasite and kaersutite: crystal-chemical constraints and implications for natural systems. Earth and Planetary Science Letters 176, 185–201.[Web of Science]
Tribuzio, R., Tiepolo, M. & Thirlwall, M. F. (2000). Origin of titanian pargasite in gabbroic rocks from the Northern Apennine ophiolites (Italy): insights into the late-magmatic evolution of a MOR-type intrusive sequence. Earth and Planetary Science Letters 176, 281–293.
Volfinger, M., Roberts, J. L., Vielzeuf, D. & Neiva, A. M. R. (1985). Structural control of the chlorine content of OH-bearing silicates (micas and amphiboles). Geochimica et Cosmochimica Acta 49, 37–48.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  R. F. Martin Amphiboles in the Igneous Environment Reviews in Mineralogy and Geochemistry, October 1, 2007; 67(1): 323 - 358. [Full Text] [PDF]
|
|
|
|
 |
|
 M. B. Holness, A. T. Anderson, V. M. Martin, J. Maclennan, E. Passmore, and K. Schwindinger Textures in Partially Solidified Crystalline Nodules: a Window into the Pore Structure of Slowly Cooled Mafic Intrusions J. Petrology, July 1, 2007; 48(7): 1243 - 1264. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. T. Claeson, W. P. Meurer, K. J. Hogmalm, and S.-A. Larson Using LA-ICPMS Mapping and Sector Zonation to Understand Growth and Trace-Element Partitioning in Sector-Zoned Clinopyroxene Oikocrysts from the Norra Ulvo Gabbro, Sweden J. Petrology, April 1, 2007; 48(4): 711 - 728. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||