Journal of Petrology 45(6) © Oxford University Press 2004; all rights reserved
The Lower Zone–Critical Zone Transition of the Bushveld Complex: a Quantitative Textural Study
1 103 OLD CHEMISTRY BLDG, DUKE UNIVERSITY, DURHAM, NC 27708, USA
2 HUGH ALLSOPP LABORATORIES, WITS, JOHANNESBURG, SOUTH AFRICA
RECEIVED MARCH 24, 2003; ACCEPTED DECEMBER 1, 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONGEOLOGICAL SETTING AND SAMPLE...METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
The Lower Zone–Critical Zone boundary of the Bushveld Complex is an intrusion-wide, major stratigraphic transition from ultramafic harzburgite and pyroxenite in the Lower Zone to increasingly plagioclase-rich pyroxenites and norites in the Critical Zone. Quantitative textural and compositional data for 29 samples through this transition show the following: Lower Zone orthopyroxene grains are larger, have higher aspect ratios, are better foliated and have a lower trapped liquid component than those of the Critical Zone. The larger grain size of the Lower Zone results in crystal size distribution plots that are rotated to lower slopes and intercepts relative to those in the Critical Zone. Although all rocks show differing amounts of foliation, mineral lineations are weak to absent. These data are consistent with significant compaction-driven recrystallization in the study section. Numerical modeling of concurrent compaction and crystallization provides a quantitative model of how the Lower Zone–Critical Zone transition may have formed: plagioclase is rare in the Lower Zone because compaction removes interstitial liquid before it reaches plagioclase saturation. However, as the crystal pile grows, plagioclase saturation is reached in the interstitial liquid before compaction is complete in more evolved pyroxenites, producing more abundant but still modest amounts of plagioclase characteristic of the Lower Critical Zone. It is concluded that both the textures and the modal mineralogy are largely controlled by compaction and compaction-driven recrystallization; primary magmatic textures are not preserved.
KEY WORDS: Bushveld Complex; compaction; crystal size distributions; crystal aging; igneous textures
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND SAMPLE...METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
The role of compaction in the chemical fractionation of igneous bodies, from thick lava flows and lava lakes to large intrusions and mid-ocean ridge magma chambers, has been the subject of many recent studies (e.g. Philpotts et al., 1998
; Rabinowicz et al., 2001), and although such a process is well documented in some settings [e.g. Stillwater Complex (Meurer & Boudreau, 1998), Skaergaard Layered Series (McBirney, 1995, 2002), Holyoke Flows (Boudreau & Philpotts, 2002), Palisades Sill (Shirley, 1986)], its role is debated for others. In the Bushveld Complex, for example, some workers argue that there was no significant crystal–liquid mush zone in which compaction could have occurred (e.g. Cawthorn, 1999), whereas others suggest it played a critical role (e.g. Mathez, 1995, 1997). Determining whether compaction was a factor in the chemical and textural evolution of the Bushveld Complex will clarify whether or not large intrusions have a significant thickness of compacting crystals.
In addition, a fundamental question in modern studies of magma chambers in general, and layered intrusions in particular, is how chemical fractionation occurs. Recent models suggest that a reasonable means to achieve chemical fractionation where in situ growth predominates is for the highly evolved interstitial liquid fraction to be returned to the magma chamber (e.g. Langmuir, 1989). Compaction of the crystal pile is one mechanism by which evolved liquids can be returned to the main body of magma (e.g. Irvine, 1980; McBirney, 1995; Meurer & Boudreau, 1998).
To date, most evidence cited for compaction in igneous bodies has been geochemical. For example, a stratigraphic offset in trace element concentrations can be explained as the geochemical signature of late, upward migrating interstitial liquids expelled from the compacting crystal pile (e.g. Irvine, 1980; Godard et al., 1995; McBirney, 1995; Philpotts et al., 1996). However, these geochemical studies suffer from a lack of uniqueness; such trace element differences have also been attributed to multiple magma pulses (e.g. Eales & Cawthorn, 1996), or to late re-equilibration with interstitial liquids (Cawthorn, 1995).
The stratigraphic section from the Lower Zone to the Upper Critical Zone of the Bushveld Complex is a major, intrusion-wide, stratigraphic transition from ultramafic harzburgite and pyroxenite in the Lower Zone to increasingly plagioclase-rich pyroxenites and norites in the Critical Zone. This stratigraphic section is of interest in understanding crystal mush-zone compaction because of the change from one (orthopyroxene) to two (orthopyroxene + plagioclase) major modal minerals. For example, compaction-driven recrystallization might be expected to be more efficient in monomineralic assemblage where mass transport distances are small. In addition, one might expect compaction efficiency to decrease once plagioclase becomes a cotectic phase, all other parameters being equal (Meurer & Boudreau, 1998).
Recent advances in the quantification of igneous textures allow us to address these issues. Textural studies of igneous rocks are gaining favor as a means to decipher crystallization history, magma chamber processes, and textural evolution of igneous systems as preserved in the solid rock (e.g. Marsh, 1998; Higgins, 2002a).
For example, crystal size distributions (CSDs) are usually shown as a semi-log plot of population density, that is, number of crystals per unit volume, against crystal size, and provide a direct method of determining some of the crystallization kinetics in a system independent of experimental approaches or theoretical thermodynamic or kinetic modeling. The basic precept of CSD analysis of igneous rocks is that easily measured parameters such as crystal size, crystal number and other crystal population characteristics can be related to rates of crystal nucleation and growth. Thus, if growth rate is constant and nucleation rate increases exponentially with time a typical log–linear CSD plot is produced (Marsh, 1988, 1998). Changes of the crystallization behavior are reflected in the shape of the CSD plot. For example, if liquid is expelled from the system, one would expect a decrease in nucleation rate and a corresponding decrease in small size fractions (Marsh, 1998).
Crystal size distribution (CSD) techniques have been used to establish whether solidification occurred as the result of single system (batch, or non-fractional) crystallization or whether a population of crystals represents a mixture of crystals from two different crystal-phyric magmas (e.g. Resmini & Marsh, 1995; Marsh, 1998; Garrido et al., 2001). In addition, CSD studies have been used to provide information on crystal growth rate and residence time of crystals in the liquid (Marsh, 1988), nucleation density and magmatic processes such as crystal accumulation and crystal removal (Cashman & Marsh, 1988), and post-nucleation crystal aging (also called annealing, Ostwald ripening and textural coarsening in the literature: e.g. Waters & Boudreau, 1996; Higgins, 2002). For example, the timing of aging in an equilibrating crystal pile has also been established through studies of chadocryst populations whose growth was sequentially arrested by the enclosing oikocryst (e.g. Higgins, 1998). These processes are illustrated schematically in Fig. 1.
|
Other workers have measured the effect of compaction using statistical determinations of alignment of crystals in a system (Higgins, 1991
; Meurer & Boudreau, 1998; Gray et al., 2001). A quantitative method for determining the extent of alignment of tabular grains is based on the eigenvectors of the major axes of the grains of interest (e.g. Benn & Allard, 1989). For example, in the Stillwater Complex, these measurements were used to show systematic variations in foliation with stratigraphic height (Meurer & Boudreau, 1998). Combining mineral foliation data with other factors, such as trace element distribution or modal changes with stratigraphic position, can be useful in gaining insight into how foliation development is related to compaction and redistribution of late interstitial liquids. Furthermore, alignment factors have been compared with mineral aspect ratios to determine to what extent foliation is related to recrystallization under uniaxial compression (Meurer & Boudreau, 1998).
Analysis of the spatial distribution of grains in rocks is another method of quantifying igneous texture (e.g. Jerram et al., 1996). Such analysis can be used to understand whether crystals are clustered or randomly distributed, and has implications for discerning between differing models of the crystallization of igneous rocks (Jerram et al., 1996, 2003). Furthermore, experimental work has identified characteristic trends for mechanical compaction, sorting and overgrowth on spatial distribution plots (Jerram et al., 1996). Comparison of spatial distribution data with these established results provides another tool for determining the relative importance of various crystallization mechanisms during the formation of the Bushveld Complex.
In the Bushveld Complex, evidence for crystal aging and recrystallization has been illustrated in textural studies of chromitites of the Lower and Critical Zones, where the resultant sintered texture was interpreted as an accommodation of high local stresses at touching chromite corners (Hulburt & Von Gruenewaldt, 1985). However, the source of these locally high stresses was not addressed in the study. Here we will propose that a similar process was functioning during the long crystallization history of the Lower and Critical Zones of the Bushveld Complex, and that the stresses that caused aging in our samples were due primarily to pure shear from compaction in a mush zone. In addition, a numerical model of concurrent crystallization and compaction is presented to illustrate the development of the major features of the Lower and Critical Zones.
|
GEOLOGICAL SETTING AND SAMPLE DESCRIPTIONS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND SAMPLE...METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
The Bushveld Complex of the Republic of South Africa is the world's largest continental layered intrusion, with an aerial extent of about 66 000 km2 (Fig. 2). Its general geology and igneous stratigraphy have been well described and are only briefly summarized here. A more complete discussion has been given by Eales & Cawthorn (1996
, and references therein).
|
The layered rocks of the 2·06 Ga (Walraven, 1988
) Bushveld Complex form a lopolithic intrusion with an average thickness of about 7–9 km. The stratigraphy of the Bushveld Complex has been subdivided into a basal Marginal Zone, overlain sequentially by the Lower Zone, Critical Zone, Main Zone and Upper Zone (Fig. 2). The Marginal Zone comprises sills of locally quench-textured pyroxenites and micropyroxenites whose compositions have been interpreted as parental to the main Bushveld magmas (Sharpe, 1985). The Lower Zone is approximately 800–1000 m thick and consists primarily of layers of pyroxenite, dunite, and harzburgite with some chromite with small amounts of interstitial plagioclase, biotite, and clinopyroxene. The boundary between the Lower Zone and the Critical Zone has been defined by what is, to the casual observer, a rather modest increase from about 2% to about 6% in the modal abundance of interstitial plagioclase (Cameron, 1978). Alternatively, the Lower Zone–Critical Zone boundary has been defined by a decrease in olivine abundance (Teigler & Eales, 1996). Here we follow the Cameron (1978) definition, which was based on the Jagdlust section of the Bushveld, the section sampled for this study. The Critical Zone is subdivided into the Upper Critical and Lower Critical Zones, with the boundary between the two occurring where plagioclase changes from an interstitial mineral to an abundant, euhedral phase. Thirteen major chromitite seams have been identified in the Critical Zone, namely the Lower Group (LG) seams 1–7, four middle group (MG1–4) seams, and two upper group (UG) seams.
The boundary between the Critical Zone and the Main Zone is usually taken to be the top of the giant mottled anorthosite layer, an anorthosite with large oikocrysts of pyroxene. It consists of a thick succession of norites and gabbronorites devoid of olivine or chromium spinel. The base of the Upper Zone is generally taken as occurring at the Pyroxenite Marker horizon (Kruger, 1990), and is also marked by the appearance of euhedral magnetite (Eales & Cawthorn, 1996). This zone is characterized by generally well-defined modal layering, with particularly well-developed magnetite and anorthosite layers.
Sampling
In this study we focus on the stratigraphic interval from the middle of the Lower Zone to the middle of the Upper Critical Zone (Fig. 2) to better constrain the nature of the transition. Our samples were collected from the Jagdlust section of the Eastern Bushveld Complex, where outcrop exposure is good. In the Lower Zone, sampling focused on orthopyroxenites, as harzburgites were more weathered and fresh samples were not easily obtained. Throughout the study section, sampling otherwise focused on typical Bushveld rock types. Samples collected can be subdivided into four groups based on modal mineralogy and textural characteristics.
Group 1
The samples in the Lower Zone (Fig. 3) are primarily orthopyroxenites, with an average orthopyroxene content of 85·5 mode %, along with varying amounts of olivine and clinopyroxene. Plagioclase content is negligible. These samples (numbers 10, 12, 13, 14, 15, 17) were collected over a vertical range of about 150 m. Orthopyroxene grains are generally sub- to euhedral, with grains up to 1·5 cm in length observed in the field. An orthopyroxene-defined mineral foliation is visible at hand-sample scale, and mineral foliation is parallel to macroscopic layering. In these and other samples, there is no obvious mineral lineation. Petrographic analysis of these samples suggests that post-crystallization textural modification has occurred (Fig. 4a–f): bent and broken grains are common, as are grains that appear to have incorporated sub-grains.
|
|
Group 2
Samples above the Lower Zone–Critical Zone boundary have increasing amounts of interstitial plagioclase. Critical Zone samples whose minerals are orthopyroxene (68–81%) and plagioclase (5–16%) with little or no euhedral chromite include samples 73, 28, 22, 29, 6, 35 and 4. These samples are characterized by orthopyroxene grains that are clearly separated to just touching in a plagioclase matrix, in which plagioclase does not form euhedral grains (Fig. 3). Samples 22, 28 and 29 were collected from about 10, 130 and 150 m stratigraphically beneath the LG6 chromitite layer, respectively, whereas sample 6 was collected from a feldspathic bronzitite layer about 2·5 m above the LG6 chromitite layer. Sample 73 was collected just beneath the MG chromitite layers, and sample 35 was collected from between the MG1 and MG3 chromite layers. In the horizon that hosted sample 73, thin lenticular segregations of plagioclase were observed, although plagioclase is still generally interstitial and makes up only a very few percent of the rock.
Group 3
The third textural group is similar to Group 2 with respect to relative proportions of orthopyroxene and plagioclase, but also contains 7–45% subhedral to euhedral chromite grains (Fig. 3). This group includes samples 7, 8, 33, 27, 30, 34 and 5. The chromite grains tend to form chains similar to those described for these rocks in the Potgietersrus section of the Bushveld Complex (Hulbert & Von Gruenewaldt, 1985). Samples 5, 7 and 8 occur stratigraphically just above the top of the LG chromite layers, whereas samples 33, 27 and 30 were collected from the stratigraphic interval between the LG chromite layers and the MG chromites.
Group 4
Samples 38, 41, 42, 43, 57, 58, 61, 63, 64, 66 and 67 are Upper Critical Zone norites and anorthosites. Plagioclase forms a euhedral phase in these samples and accounts for 20–81% of the modal assemblage (Fig. 3). These samples were collected upwards from the MG chromitite layers that mark the Lower–Upper Critical Zone boundary. In these samples, pyroxene grains are generally isolated from other pyroxene grains in two dimensions; however, pyroxenes locally form a loose framework in a matrix of plagioclase. Plagioclase grains can have twins at an angle to the long grain axis, again consistent with some degree of post-accumulation recrystallization (Fig. 4g–j).
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND SAMPLE...METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
A suite of 40 accurately located and oriented samples were obtained from an
1500 m interval bracketing the transition between the Lower and Critical Zones. In this study we disregarded dunite and stratigraphically minor chromitite samples, and focus on typical orthopyroxene ± plagioclase-bearing lithologies. Thin sections from 29 samples from the Lower Zone–Critical Zone transition interval were analyzed as follows. A digital image of the thin section in cross-polarized light was obtained using a 35 mm film slide scanner and printed at high resolution. This image was then compared petrographically with the thin section proper, and separate manual tracings of the major mineral phases (orthopyroxene ± plagioclase) were made by overlays on a light table. Each tracing comprises a population of between about 100 and 1900 mineral grains, depending on phase proportions and grain size. These tracings were then scanned, converted to binary (black and white) digital images, and stray marks were removed in Adobe Photoshop. Examples of original thin-section images and corresponding binary images for orthopyroxene and plagioclase are shown in Fig. 3. Binary images used in CSD analysis are available for downloading from the Journal of Petrology website (http://www.petrology.oupjournals.org). The binary image was loaded into the NIH PC-based image analysis program, Scion Image, to obtain crystal major and minor axis lengths, crystal areas and the angles of major axes within an arbitrary reference frame in the plane of the thin section, which is oriented either parallel or perpendicular to foliation or layering. The smallest grain size measured is given by the width of the tracing pen on the scale of the enlarged print of the thin section, generally about 0·025 mm. The parameters in Scion Image were set to ignore particles smaller than 5 pixels, the size of a dot made with the tracing pen. The smallest grains observed at 25x magnification were easily large enough to be counted in the analysis, thus the smallest grain size reported for each sample can be taken as the limit for that sample. From these data, alignment factors (AF), crystal size distribution (CSD), and R-value (spatial distribution pattern parameter) calculations were made.
We also collected data for aspect ratio (AR), maximum grain size (Lmax), and average grain size (Lavg) for both orthopyroxene and plagioclase. Lmax was taken as the average of the four largest grains of the sample population, and Lavg represents the average length of all grain measurements with grain size measured in Feret length (the length of the side of a square of equal area to the measured grain area). Using Feret length minimizes differences caused by irregular grain shape and allows for accurate calculation of crystal size distributions as described below.
Alignment factor
Major axis lengths and orientation of individual grains were resolved into their x–y components in an arbitrary x–y coordinate system from samples cut both normal and parallel to dominant mineral foliation, where possible. Because foliation is defined by the larger grains, only the 40 longest major axis lengths were used for each sample. From these, a statistically average orientation can be described by a normalized orientation tensor (Harvey & Laxton, 1980; Benn & Allard, 1989; Meurer & Boudreau, 1998):
 | (1) |
 | (2) |
). Figure 3 shows examples of thin sections cut normal and parallel to foliation for alignment factors ranging from about four to 70.
Crystal size distribution
The relationship between the population density and crystal size of a population of crystals, as shown in a CSD plot, contains important information about crystal growth, nucleation, aging and mixing of crystal populations. Derivations of mathematical relationships established in CSD methodology have been summarized by Marsh (1998, and references therein). A number of recent papers (e.g. Coogan et al., 2002; Jerram et al., 2003) have used the program CSDCorrections by Higgins (Higgins, 2000) to calculate CSD plots. This program has a number of useful features, including corrections for cut-section effects and rock fabric. However, two features of the program make it less than ideal for Bushveld Complex minerals. First, the program assumes crystals have well-formed, idiomorphic habits with fixed aspect ratios. Although this may be an adequate approximation for minerals in volcanic rocks, Bushveld minerals do not have a consistent shape. Larger grains are typically more elongated (have a higher aspect ratio) than smaller grains, and all grains can have irregular to partly convex boundaries. Because of this, and because the program uses only one measure of crystal size (length or width) to calculate the CSD and does not correct the CSD to actual volume of crystals in the rock, it can significantly miscalculate the total mineral volume by as much as 100%. CSDCorrections should give similar volumes regardless of whether major axes or minor axes are input; however, for irregularly shaped grains there are significant discrepancies in the results for both.
Second, CSDCorrections plots CSD data using the longest grain dimension for non-spherical grains. Thus, a sample with high aspect ratio grains will have a flatter CSD than a sample with more equant grains, even though the individual grains in the two sample populations have identical volumes. Because of this, the program can mask differences that are caused by processes such as crystal aging that actually reduce the number density if the aspect ratio is also changing.
To avoid these problems with CSDCorrections but still take advantage of its other features, CSD plots were constructed as follows. Measured grain size was calculated as ‘Feret length’, which is the length of a square with an area equal to the measured area of the grain. These lengths were then input into CSDCorrections. CSD plots were then calculated for grains with a 1:1:1 aspect ratio (cubes) with no rounding and with the degree of foliation defined by the alignment factor defined above (as discussed below, the rocks do not have any significant lineation). This method calculates the total volume correctly and the grain size is a function of grain volume alone. However, an uncertain error is introduced in the cut section and fabric correction by the assumption of cubic shape. (True aspect ratios are approximately 1:1:L, where L is the long dimension and ranges from <2 to about 3·5.)
A comparison of CSD plots as calculated by CSDCorrections with those produced by the method of Marsh (1988) is shown in Fig. 5. For the latter, the CSD is produced by assuming that (grain volume) = (grain area)3/2 and is corrected only by the ratio (area of mineral phase)/(total area). Those trends produced by CSDCorrections tend to be flatter. Because of the variable nature of grain shapes in our sample suite, major and minor axes yield entirely inappropriate volumes as calculated by CSDCorrections. For example, in sample 73R, the actual orthopyroxene volume is 75%, whereas CSD calculations in CSDCorrections yield 28% and 176% calculated from minor and major axes, respectively, using average measured aspect ratio for the shape factor. However, when Feret length is used, reasonable volumes are obtained in CSDCorrections for all samples (Fig. 5b, inset).
|
, orthopyroxene data; 
, plagioclase data.For the relative variations discussed below, conclusions would be similar using either corrected or uncorrected data. Although we have used CSDCorrections, it is unclear that corrected CDS plots will represent any true improvement for comparative studies such as this until the algorithms can more accurately account for variably shaped grains.
Spatial distribution patterns
The spatial distribution of grains in rocks can be useful in interpreting their crystallization history (Kretz, 1969; Jerram et al., 1996; Jerram & Cheadle, 2000). A method for the analysis of spatial distribution of crystals in rocks was developed by Jerram et al. (1996), and is based on a ratio of the mean nearest neighbor distance for a population of crystals to the mean nearest neighbor diatance for a random distribution of grains. The mean nearest neighbor distance for the sample, rA, is defined as
 | (3) |
 | (4) |
is the number density of the observed distribution. The R-value is the parameter used to describe the spatial distribution of crystals and is described by
 | (5) |
).
Geochemical methods
Whole-rock composition data were collected from 34 samples across the study section. Generally, the same samples were used for geochemical analysis as were used for textural analysis. However, in some instances, samples used for textural study were not appropriate for geochemical analysis (i.e. samples were too weathered for reliable geochemical analysis), or vice versa. In those cases samples from a similar stratigraphic height were analyzed. Major element abundances were determined by direct current plasma optical emission spectrometry (DCP-OES). Samples were prepared by the methods of Klein et al. (1991) on an ARL Fisons spectroscan 7 DCP at Duke University. Trace element abundances were obtained by inductively coupled plasma mass spectrometry (ICP-MS) using a VG-Elemental Quadropole-3 system at Duke University. Sample digestion and operating parameters have been described by Meurer & Boudreau (1999b).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND SAMPLE...METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Textural data
Textural data are summarized in Figs 6–14 and Table 1. Data were collected for orthopyroxene in all samples, and additionally for plagioclase in the Upper Critical Zone samples.
|
|
|
|
|
|
|
|
|
|
Alignment factor
AF measured on orthopyroxene grains decreases with stratigraphic height, and foliation is best developed in the pyroxene-rich harzburgite of the Lower Zone, with AF values ranging from about 40 to 72, and an average value of about 64. In the Critical Zone, AF values are consistently lower for orthopyroxene, ranging from seven to 72 and averaging about 42. However, Upper Critical Zone plagioclase grains exhibit a greater extent of foliation than do Upper Critical Zone orthopyroxene grains, with the alignment factor ranging from 24 to 74, with an average value of about 57 (Fig. 6).
There is generally a positive correlation between the quality of the foliation and mineral aspect ratio in all samples analyzed (Fig. 7a), and the four textural groups we have described plot in distinct but overlapping fields. The Lower Zone samples (Group 1) show a strong positive correlation between aspect ratio of orthopyroxene grains and alignment factor. However, the relationship is less well defined in the other groups. The fields defined by the Group 2 (LCZ, no chromite) and Group 3 (LCZ, chromite abundant) samples show little change in aspect ratio with increased alignment factor. Plagioclase appears as a euhedral phase in the Upper Critical Zone (Group 4), and it has higher aspect ratios than does coexisting orthopyroxene. Data for orthopyroxene in this group show a positive correlation between aspect ratio and alignment factor. In contrast, plagioclase data for Group 4 appear to show a flat to nearly negative trend where aspect ratio does not change significantly with alignment factor. However, given that plagioclase data are from relatively few samples and span a limited stratigraphic range, they are otherwise within the error of the trend for all samples. Aspect ratio varies generally with stratigraphic position in the intrusion (Fig. 6), with Lower Zone orthopyroxene samples having consistently higher values than orthopyroxene in the Critical Zone.
To determine the strength of lineation, alignment factors were calculated for surfaces in the plane of foliation. A comparison of alignment factors on surfaces parallel to mineral foliation with alignment factors on surfaces perpendicular to mineral foliation in samples where mineral foliation is obvious at hand-sample scale (Fig. 7b) shows no correlation between the two. Even strongly foliated samples show no significant lineation of mineral grains.
Alignment factor data for plagioclase show no significant variation, either with stratigraphic height or in comparison with orthopyroxene textural data. However, plagioclase aspect ratios are greater than those of orthopyroxene in Upper Critical Zone rocks, an expected result because of the tabular habit of plagioclase. Finally, grain sizes for plagioclase are generally similar to those for orthopyroxene in rocks where both phases are modally significant (Fig. 6).
Crystal size distributions
Orthopyroxene
Figure 8 shows the CSDs for orthopyroxene plotted against stratigraphy and modal proportions. The most obvious result to emerge from Fig. 8 is that all CSDs are log–linear for the larger sizes and tend to have a convex-upwards shape at small size fractions, regardless of mineralogy or stratigraphic position. CSD slopes and projected intercepts are calculated for a best-fit line through the log–linear portion of the plot (filled squares in Fig. 8). Figure 6 summarizes CSD slope and intercept results for orthopyroxene as a function of stratigraphic height.
Quality of CSD data can be ascertained by examining closure limits; in other words, by confirming that calculated crystal content of a rock does not exceed 100% (Higgins, 2002b). A comparison of our CSD data with empirically calculated closure limits is shown in Fig. 9a and b, and illustrates that our data are well within range. A plot of orthopyroxene CSD slopes vs intercepts (Fig. 9c) shows that the data fall into two linear arrays, the first composed of the orthopyroxene-rich samples of groups 1–3 (r2 = 0·96) and the second composed of the plagioclase-rich norites of group 4 with r2 = 0·76. These trends are the result of CSD closure: for rocks with similar modal portions and with log–linear distributions, the slope and intercept are negatively correlated (Higgins, 2002b). Thus, the offset trend in Fig. 9c for the plagioclase-rich samples reflects the distinctly lower modal abundance of orthopyroxene in these samples. Within the orthopyroxene-rich samples, the typically larger grain size of the Lower Zone samples is reflected in their shallower slopes and lower intercept values (filled squares of Fig. 9c). In all samples below the Lower Zone–Critical Zone boundary, CSDs have slopes greater than –3·0 mm–1 (with an average of –2·9 mm–1) and intercepts less than 3·0 ln (mm–4) with an average of 2·45 ln (mm–4). Conversely, above the boundary, slopes are generally less than –4·0 mm–1, with an average of –4·66 mm–1, and intercepts are greater than 3·0 ln (mm–4), with an average of 3·75 ln (mm–4).
Average grain size, Lavg, for the Lower Zone orthopyroxene is 1·13 mm and average Lavg for the Critical Zone is 0·99 mm (Fig. 6). Maximum grain size, Lmax, is shown in Figs 6 and 10. There is a general decrease in Lmax with stratigraphic height: average Lmax in the Lower Zone is 4·1 mm, whereas average Lmax in the Critical Zone is 3·0 mm. It should be noted that Lmax is measured as Feret length, which mutes any additional increases in major axis length that are represented by changes in aspect ratio with stratigraphic position. When Lmax is plotted against CSD slope and intercept (Fig. 10a and b), data from all four textural groups form a general trend of shallower slope with increased grain size (Fig. 10a), with each group clustering within the larger dataset: Group 4 shows the most scatter. Similar results are observed when CSD intercept is plotted against Lmax (Fig. 10b); larger grain sizes are associated with lower intercepts and more scatter is seen in the data where more phases are present (e.g. groups 3 and 4).
The relationship between Lmax and CSD slope and intercept reflects the slope–intercept relationship that is a function of closure. In these plots, however, the flatter log–linear trends in the Lower Zone are also emphasized: Lower Zone samples have generally larger grain sizes producing shallower slopes and lower intercepts. Samples from other groups show a similar, although less pronounced, correlation between slope and intercept and Lmax. Plots of Lmax vs slope and intercept have also been used to determine diagnostic relations for batch and open magmatic systems (Marsh, 1998). For Bushveld CSD data, where a plot of slope against Lmax has a positive trend, coupled with a plot of slope vs intercept having a negative trend, the system reflects a single population of crystals. (However, see below on the discussion of sample 42.)
Plagioclase
Figures 6 and 11 summarize CSD results for plagioclase. Plagioclase CSDs have a generally consistent shape; again with a characteristic convex-upwards curve at small size fractions. Plagioclase CSD slopes are generally steeper than orthopyroxene slopes in which plagioclase content is greater than about 50 mode %, but where plagioclase is between about 10 and 50%, plagioclase slopes vary relative to orthopyroxene slopes. However, no corollary relationship is seen in plagioclase CSD intercepts as compared with orthopyroxene; again, this may reflect the limited stratigraphic coverage of plagioclase samples. A plot of plagioclase CSD slopes vs intercepts shows correlation with an r2 value of 0·6 (Fig. 11b).
CSD plots for samples 42 and 63 represent unique crystal populations: sample 42 is glomerocrystic with regions of much larger plagioclase in a matrix of fine-grained plagioclase. To investigate this characteristic further, a CSD plot for sample 42 was generated by the method of Marsh (1988), where bin size was set to 0·1 mm to catch all variation within the slope of the CSD calculation (Fig. 12a). In this more detailed CSD two distinct slopes are observed, each representing a different crystal population. The larger bin, shallower slope portion of the CSD represents glomerocrystic plagioclase (Fig. 12b), which has grain sizes about double those of the matrix plagioclase (about 0·5 mm in length). The steeper slope at smaller grain sizes describes the plagioclase crystal population that forms the matrix of the rock.
CSDs for all samples were calculated according to this simpler method (Marsh, 1988) with smaller bin sizes and fewer shape corrections to check for variations that might not be evident in the results from CSDCorrections. However, only the glomerocrystic sample 42 had a composite CSD with two slopes.
Sample 63 has a more complex story. Figure 13a shows that this rock consists of three predominant mineral populations: oikocrysts of orthopyroxene, plagioclase contained as chadocrysts in the oikocrysts, and non-chadocrystic or matrix plagioclase. In addition, the oikocrysts show a preferred orientation of their long axes with an AF of 58·7 (Fig. 13d). Initially, it was thought that the growth of the orthopyroxene oikocryst had arrested the growth and aging of plagioclase crystals as it overtook them, in the style of the olivine oikocrysts of the Lac-St-Jean anorthosite in Canada (Higgins, 1998). However, although chadocrystic plagioclase has steeper CSD slopes and higher nucleation densities than matrix plagioclase (Fig. 13b), the actual population densities are similar for all plagioclase: oikocryst 1 has a population density of 5·18 grains/mm2, oikocryst 2 has a population density of 4·43 grains/mm2, and the matrix plagioclase has a population density of 4·58 grains/mm2. If the growing oikocryst had trapped early plagioclase population before crystal aging, the population density should be higher within the oikocrysts than outside them. An examination of typical plagioclase–plagioclase grain boundaries within the oikocrysts (Fig. 13c) shows that instead of having straight grain boundaries, they appear to have reacted away from their common boundaries and are smaller because they have been partially resorbed.
Spatial distribution pattern
Figure 14 shows plots of the spatial distribution parameter, or R-value, plotted against (a) porosity (= 100% – volume % of plotted mineral), (b) aspect ratio and (c) the CSD slope. The fields shown in Fig. 14a (ordered vs clustered distributions, and touching vs non-touching frameworks) are defined for distributions of spherical grains, and hence are not directly applicable to tabular Bushveld Complex grains and are shown for reference only. (The non-applicability of these fields for tabular minerals is evident in that the non-touching orthopyroxene that is characteristic of Groups 2 and 3 none the less falls in the touching framework field.) The trends for grain distributions that have undergone different degrees of size sorting, mechanical compaction (= simple grain rearrangement), deformational compaction or overgrowth have been determined (Jerram et al., 1996) and are shown as the vector array in Fig. 14a. The vector for sorting shows the trend expected for increasingly better size sorting, and the overgrowth vector reflects the movement of grain centers with grain overgrowth of neighboring grains. The mechanical compaction trend reflects an increase in R-value coupled with a decrease in porosity as grains undergo a simple re-ordering and move into contact with each other in response to loss of interstitial liquid. The deformational compaction trend results from a decrease in R-value as the pure shear deformation of initially spherical grains causes the grain centers to move apart in a plane normal to the principal stress, along with a further decrease in porosity (Jerram et al., 1996).
We also determined porosity and R-value for a series of experimentally aged equant grains of Sn–Pb alloys in a eutectic liquid matrix produced by Hardy & Voorhees (1988), and found that they plot in a cluster at both the intersection of the clustered–ordered fields and the intersection of the touching–non-touching framework fields. Aging assemblages evolve to characteristic size–frequency distributions that remain constant with time even as the average size increases with time (e.g. Hardy & Voorhees, 1988). As porosity changes would be minor in an aging assemblage, aging crystals should have a narrow range when plotted in Fig. 14a.
Most of the Bushveld Complex spatial distribution data define a positive trend in Fig. 14a. Those furthest from this field are those in which the mineral of interest is in low modal abundance. The positive trend seen in many of the data can be the result of either variable degrees of sorting or variable deformational compaction, or both. The progressive deformation of initially spherical grains will produce grains that have increasingly higher aspect ratios and become more similar to tabular grains in shape. Hence one would expect R-values to show a negative correlation when plotted against the aspect ratio if deformation were a cause of the positive trend in Fig. 14a. In contrast, if the trend were the result of better sorting, one might expect the R-value to correlate negatively with CSD slope; poorly sorted samples should have a more uniform size distribution and hence a flatter slope. Plots of R-value against the aspect ratio (Fig. 14b) and CSD slope (Fig. 14c) for the Bushveld Complex samples show a relatively modest negative correlation for both.
Geochemical data
Major element composition data are presented in Table 2 and are shown plotted against stratigraphy in Fig. 15. Major and trace element data introduced here will be presented in full and treated more completely in a subsequent paper.
|
|
Whole-rock Mg number [Mg/(Mg + Fetotal)] and An number (calculated from normative plagioclase) are plotted against Bushveld mode and stratigraphy in Fig. 15. Mg number is approximately constant in the Lower Zone at about 0·75, and decreases moderately in the Lower Critical Zone. At the Lower Critical Zone–Upper Critical Zone boundary, there is a great deal of variation in Mg number, which is lowest in rocks with low modal mafic minerals. However, Mg number continues to decrease gradually throughout most of the Upper Critical Zone, ranging from about 0·65 to about 0·55. Whole-rock An number increases in the Lower Critical Zone from about 0·85 to about 0·95. Again at the Lower Critical Zone–Upper Critical Zone boundary there is a good deal of variation in the data. However, An number stays fairly constant through the Upper Critical Zone, exhibiting a decrease of only 0·05 (from about 0·77 to 0·72) across 1000 m of section.
Whole-rock data for igneous cumulates can be difficult to interpret as they represent a mixture of cotectic crystals and possibly evolved and modified fractionated interstitial liquid. Further complications can occur if interstitial liquids migrate prior to complete solidification (e.g. Langmuir, 1989); thus traditional methods of calculating the trapped liquid proportion (% TLP; e.g. Maier & Barnes, 1998) based on incompatible elements may be erroneous. With this caveat, however, an estimation of % TLP is invaluable to the first-order understanding of the geochemical evolution of the Bushveld Complex.
The problem of estimating % TLP in cumulates has been addressed in the Stillwater Complex by using Y (Meurer & Boudreau, 1998), which provides a reasonable estimate that is broadly in range with previous estimates of Bushveld Complex residual porosities of 5–50% (average 20%) in the Upper Critical Zone and 10–20% (average 15%) in the Lower Zone (e.g. Maier & Barnes, 1998; Cawthorn, 1999). Calculated Y % TLP data show about 5% TLP in the Lower Zone with % TLP increasing up-section to between 5 and 20%, and with increased spread in data up-section as well (Fig. 15a). Phosphorus data for the same study section (Willmore et al., 2000) have also been used as a proxy for residual porosity and P and Y % TLP agree well, showing lowest calculated residual porosity in the Lower Zone and more variable but generally higher calculated porosity in the Critical Zone.
Y % TLP plotted against alignment factor (Fig. 16) shows a noisy but generally negative correlation: alignment factor is higher in rocks with low calculated trapped liquid component. Significantly, the nearly monomineralic rocks of the Lower Zone show a clear clustering of the low porosity with high AF.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND SAMPLE...METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
The textural data presented above show consistent variations between the Lower and Critical Zones. Lower Zone orthopyroxene grains are larger, have higher aspect ratios and are better foliated than those of the Critical Zone. Although all rocks show differing amounts of foliation, mineral lineations are weak to absent. The coarser-grained, more foliated rocks of the Lower Zone also contain the lowest calculated residual liquid fraction. Except for the glomerocryst-bearing sample already noted, all CSD plots show trends typical of a single crystal population and are overturned at small grain sizes. The larger overall grain size of the Lower Zone results in flatter slopes and lower intercept values in the log–linear part of the CSD trends relative to those in the Critical Zone. Spatial distribution R-values are, at least in part, a function of aspect ratio.
Hypothesis
It is our contention that most of the modal compositional and textural data are explained as the result of concurrent compaction and recrystallization–aging during the solidification of the Lower and Critical Zones. In a compacting mush zone, the pure shear of compaction in the crystal pile results in both preferential dissolution of unfavorably oriented grains and their subsequent recrystallization in more favorable orientations in the uniaxial stress regime, as well as rotation of tabular grains into alignment with each other normal to the direction of compaction. These processes can account for the observed high orthopyroxene aspect ratios, which correlate with well-developed foliation in the Lower Zone, and the generally low trapped liquid component. The lack of lineation indicates that there was no slip component arising from either magmatic deposition or subsequent compaction and recrystallization.
At the base of the Bushveld Complex, where cooling was most rapid, the compacting mush zone would crystallize before much compaction and fractionation could occur. This produced the Marginal Zone rocks whose compositions are considered to be quenched liquid parental to the Lower Zone. However, in the middle of the Lower Zone, where our sampled section begins, the rocks are nearly monomineralic orthopyroxenites situated far enough from any chill zone not to be affected by rapid freezing effects. The magma in the Lower Zone was crystallizing orthopyroxene and both the crystallizing magma and the initial interstitial liquid between the orthopyroxene grains were far from plagioclase saturation (Fig. 17a). Before the interstitial liquid could cool to the point of plagioclase saturation, compaction led to it being nearly entirely expelled. Extensive compaction-driven recrystallization led to the development of well-foliated rocks with high aspect ratios and low residual porosity in the Lower Zone. Orthopyroxene grains were in direct contact with one another and could readily recrystallize and coarsen.
|
|
At higher stratigraphic levels (the Lower Critical Zone), the more differentiated magma was still saturated only in orthopyroxene, but was closer to plagioclase saturation. Crystallization of this interstitial liquid allowed it to reach plagioclase saturation prior to being expelled by compaction (Fig. 17b). The presence of this increasingly abundant interstitial plagioclase inhibited growth and recrystallization of orthopyroxene both by pinning grain boundaries and by maintaining longer mass transport distances between orthopyroxene grains. Orthopyroxene grains were no longer in physical contact with each other, grain boundaries could not migrate as easily, diffusion distances became longer and thus textural maturation was retarded. This process resulted in lower alignment factors, lower CSD slopes and lower aspect ratios of the Critical Zone orthopyroxene.
This study postulates that both crystal aging and recrystallization were controlling factors in the development of the Lower Zone and Critical Zone textures. Crystal aging is a process by which larger grains grow at the expense of smaller or otherwise energetically less favored grains (e.g. Hunter, 1986; Park & Hanson, 1999). Crystal size changes to reduce the interfacial surface energy; large grains have the thermodynamic advantage and grow at the expense of smaller grains, thus causing the entire system to coarsen and decreasing the total number of grains in the system (Hunter, 1986; Higgins, 1998; Park & Hanson, 1999). The aging phenomenon has been shown to be important in igneous systems at even small degrees of crystallization, before growing crystals begin to impinge on each other (Park & Hanson, 1999). Furthermore, monomineralic sequences in layered rocks are generally observed to be coarser than adjacent polymineralic sequences (e.g. Jackson, 1961; Dick & Sinton, 1979), suggesting than aging is more efficient in monomineralic rocks. In fact, grain boundary migration that produces aging is limited when more than one phase is present, as boundaries are pinned at the triple junctures by the presence of a second phase (Hunter, 1986). Grain growth is more rapid when atoms need only to move across a grain boundary (as is the case in monomineralic rocks) versus when atoms must diffuse along a grain boundary in the solution–reprecipitation process of grain boundary migration that is dominant in bi- or multimineralic rocks (Hunter, 1986; Hardy & Voorhees, 1988).
On the other hand, recrystallization is a process whereby grain boundaries shift as a result of, for example, anisotropic stress. Pressure-solution recrystallization is a form of diffusive mass transport that is favored under a deviatoric stress regime in the presence of interstitial fluid (Durney, 1976; Dick & Sinton, 1979). Dislocations in the crystal structure occur in response to stress, and the stressed crystal has a higher free energy than an unstrained one (Raghavan & Cohen, 1975). Crystals whose lattices are favorably oriented relative to the stress regime grow at the expense of crystals that are unfavorably oriented (Nicolas & Poirier, 1976). This is accomplished primarily by diffusive atomic transport along grain boundaries, or through an intercrystalline fluid phase from unfavorably oriented grains to favorably oriented ones (Nicolas & Poirier, 1976). In this study, textural maturation and recrystallization refer to the combined effects of crystal aging and stress-induced recrystallization.
Alignment factor data
In a compacting system, one mechanism for crystals to absorb uniaxial stress is by physically rotating elongated grains into alignment with each other, generating a foliation without affecting individual crystal shapes. Crystals also respond to uniaxial stress by preferential dissolution and reprecipitation at more favorable orientations (e.g. Green, 1966, 1967). This second process results in foliation development corresponding to higher aspect ratios. However, igneous foliation can be developed in response to other mechanisms as well. In a textural study of the Skaergaard Layered Series (Boudreau & McBirney, 1997; McBirney & Nicolas, 1997), layering was divided into two categories; namely, layering formed by dynamic processes such as magmatic flow, and layering formed by non-dynamic processes that occur in situ as crystallization proceeds in the crystal pile. For example, if foliation results from the alignment of crystals owing to magmatic flow, we would expect to see a strong lineation developed as well (Boudreau & McBirney, 1997); where layering results from the pure shear of compaction, mineral lineation should be seen only where compaction is overprinting pre-existing lineations (McBirney & Nicolas, 1997). Alternatively, if compaction occurred along a tilted intrusion floor, for example, a component of slip can develop.
If there were a slip component in the Bushveld Complex samples, there should be a correlation between an alignment factor measured in the plane of the layering and one perpendicular to it. However, Fig. 7 illustrates that there is no correlation between the quality of foliation developed and the presence of lineation in our sample suite. Thus, the foliation observed in our samples is consistent with pure shear developed from simple compaction alone, with no slip component.
Although the observed positive correlation between the foliation and aspect ratio could be a function of the original grain shape, several lines of evidence argue against the grain shape being unmodified. First, because the Lower Zone rocks have low residual porosity, as shown by our Y % TLP and P data, it is unlikely that there was no secondary textural development: grains in direct contact easily undergo coarsening (e.g. Hunter, 1986). In contrast, initial porosity from the mechanical accumulation of grains in crystal mushes, which have undergone no deformational compaction or recrystallization, has been estimated at about 20–50 vol. % (e.g. Jackson, 1961). Similarly, maximum close packing of spheres yields a minimum porosity for mechanical compaction (of spheres) of 36·4% (Finney, 1968, cited by Jerram et al., 1996). Second, a relationship between foliation and aspect ratio that was dependent on original grain shape would require that initial orthopyroxene grain shape be a function of the other phases that are crystallizing. Furthermore, we see abrupt changes in the AF and mineral aspect ratio across the Lower Zone–Lower Critical Zone transition, where the only petrographic change is a modest increase in the amount of interstitial plagioclase. If the foliation is a function of the original grain shape only, this would appear to require that the original orthopyroxene anticipated the eventual crystallization of the interstitial plagioclase.
The positive correlation between mineral aspect ratio and AF (Fig. 7) supports the interpretation that recrystallization (crystal aging) occurred in a regime of uniaxial stress, as would be expected in a compacting crystal pile. Both selective grain resorption of unfavorably oriented grains and uneven crystal growth caused by uniaxial stress of compaction result in grains with high aspect ratios (Maaloe, 1976; Mathison, 1987; Meurer & Boudreau, 1998).
Foliation of orthopyroxene is significantly less well developed in the Critical Zone of the Complex. This result is consistent with compaction: foliation in a compacting mush zone is best developed in monomineralic systems (Hunter 1987; Meurer & Boudreau, 1996). Where plagioclase or chromite becomes a major phase, orthopyroxene grain boundaries become essentially pinned in place, inhibiting further efficient accommodation of compaction by pure shear. Recrystallization is slower where orthopyroxene grains are not in contact with one another. Furthermore, plagioclase has been shown to form linked networks of grains that uniquely affect rheology at crystallinities as low as 30% (Philpotts et al., 1998). The effectiveness of plagioclase in changing crystallization dynamics has been modeled, and can lead to the development of fluid overpressure at the Critical Zone boundary (Boorman et al., 2003). In Upper Critical Zone norites where plagioclase grains are in physical proximity, foliation is again well developed, but is defined by plagioclase grains instead of by orthopyroxene grains.
CSD data
Two dominant results are immediately obvious from the orthopyroxene CSD data in this study: (1) CSDs throughout the section exhibit a characteristic convex-upwards shape at small size fractions; (2) CSD slopes are consistently flatter, and intercepts are lower, in the Lower Zone than in the Critical Zone.
The convex-upwards shape of the plots at small size fractions (Fig. 9) can be explained as the loss of small size fractions owing to crystal aging or annealing where the small size fractions are consumed at the expense of the growth of the larger crystals to minimize the free energy of the system (Hunter, 1986; Resmini, 1993; Waters & Boudreau, 1996; Higgins, 1998, 1999, 2002a). Overturned CSDs can also be explained as the loss of small size fractions as recently nucleated crystals are expelled from the compacting pile carried with late liquid fractions; and, as the late liquid is expelled, it no longer serves as a source of new crystal nucleation for the crystal system from whence it came (Marsh, 1998). However, there is no evidence in our sample section that the interstitial liquid transported crystal nuclei and deposited them at other stratigraphic levels. If that were the case, we might observe an upward kink at small grain sizes up-section to correspond to the downturn at small grain sizes in the sampled section (Marsh, 1998).
The flatter CSD slopes of orthopyroxene in the Lower Zone relative to the overlying zones are not consistent with simple magmatic cooling and crystallization, as it implies slower cooling near the bottom contact than for the overlying rocks. For example, relatively shallow CSD slopes have been interpreted as indicating slower cooling in a vertical section through the Lower Gabbros of the Oman ophiolite (e.g. Garrido et al., 2001). In that case, steeper slopes were interpreted to indicate rapid cooling by hydrothermal convection through the ocean crust. If the cooling rate were the main control on CSD slopes, then rocks near the bottom of the Bushveld Complex should cool and crystallize more rapidly than the hotter interior of the intrusion, and thus have steeper CSD slopes. However, the opposite is observed across the Lower and Lower Critical Zones of the Bushveld Complex. It is concluded that grain size was controlled mainly by post-crystallization, mush-zone processes (e.g. crystal aging and compaction-driven recrystallization), rather than by initial crystallizing conditions of the magma chamber.
The effect of crystal aging on CSD shape has been demonstrated for the Stillwater Complex chromites (Waters & Boudreau, 1996), in plagioclase in the Lac-St-Jean anorthosite (Higgins, 1998), in plagioclase, olivine and clinopyroxene in the Kiglapait intrusion (Higgins, 2002a) and in modeled scenarios (Cashman, 1990; Higgins, 1998). As large crystals grow at the expense of small crystals, the resultant CSDs are rotated counter-clockwise to flatter slopes (Higgins, 1998, 2002a; Marsh, 1998). The Lower Zone CSD plots are similarly ‘rotated’ relative to the overlying zones such that slopes are consistently shallower than in the Critical Zone.
A re-examination of the question of the convex-upwards shape of the CSD in light of the changes of slope and intercept with stratigraphy leads to the following deduction: if physical separation of small size fractions from the system were the only mechanism causing the convex-upwards shape of Bushveld CSDs, the CSD slopes need not vary systematically with stratigraphy or mineral mode. If the overturning of the CSDs were due only to the physical separation of smallest fractions from the system, the linear portions of the CSD trends need not be affected and slopes could be constant across the Lower Zone–Critical Zone transition. Finally, if crystal settling were the predominant mechanism, accumulation of similar size crystals would be reflected by a flattening of the CSD shape at larger size fractions (Marsh, 1988), a feature not seen in our results. Thus, we infer that the same processes that led to more recrystallization in the Lower Zone also led to more extensive aging in the Lower Zone than in the Critical Zone.
A final important result to emerge from examining the shapes of orthopyroxene CSDs from our sample suite is that their relationships and shapes do not in themselves show evidence of multiple controls on crystallization dynamics. For example, if multiple magma pulses were mixing, especially if crystal-phyric magmas were added, the CSDs should reflect these different population mixtures, assuming that textural effects were not subsequently annealed out. Mixing of distinct crystal populations would result in kinked or offset CSDs (e.g. Marsh 1998). For example, although the thickness of the Lower Zone has generally been explained as resulting from multiple injections of fresh magma preventing the system from reaching plagioclase saturation, Lower Zone CSDs are consistently uniform in shape. Only the glomerocrystic Upper Critical Zone plagioclase sample 42 shows evidence for mixing of two crystal populations. Furthermore, a plot of CSD slopes vs intercepts for our sample suite forms a linear trend, a result consistent with a single evolving crystal population. Any deviations from that trend can be explained as resulting from the effects of variations in modal mineralogy (closure). We conclude that if magma mixing added different crystal populations into the system or modified existing populations, then the record of this has been overprinted by subsequent recrystallization (Higgins, 2002).
Plagioclase vs orthopyroxene
Although most CSD plots for orthopyroxene and plagioclase have similar shapes, CSD slopes are generally steeper for plagioclase than for orthopyroxene, indicating more rapid crystallization, higher nucleation rates, or both. This relationship is especially true where plagioclase forms more than 50% of the modal assemblage. Plagioclase CSD slopes correlate with CSD intercepts as expected by closure (Fig. 11b), but the shallow slopes and low intercepts do not vary consistently with stratigraphy; rather it appears that recrystallization processes are most effective in samples with higher proportions of plagioclase. This interpretation is supported by alignment factor data for plagioclase: AF is significantly better developed where modal plagioclase is greater than 50% (Fig. 6). As for orthopyroxene, plagioclase foliations are best developed where plagioclase grains are in contact with each other.
Geochemistry
Geochemical data, like the textural data, tend to be well clustered in the Lower Zone and exhibit a spread at the Lower Critical Zone–Upper Critical Zone boundary. This spread is seen in both the bulk-rock data and in calculated trapped liquid proportions.
Numerical simulations of compaction of density stratified cumulates (e.g. Meurer & Boudreau, 1996) show that compaction in a crystal pile can cause uneven distribution of liquids. In these simulations, movement of interstitial liquid was greatly affected when migrating through rocks with sharp density contrasts. The Upper Critical Zone boundary represents such a sharp density contrast as modal plagioclase increases abruptly at that contact. These abrupt density contrasts are associated with porosity peaks and spikes in trace element data in the Munni Munni Complex and the Jimberlana Complex (Meurer & Boudreau, 1996). The effect of compaction on trapped liquid composition and distribution in the Bushveld Complex are the subject of a future paper.
Quantitative modeling of the Lower Zone–Critical Zone transition
The program IRIDIUM (Boudreau, 2003) has been used to illustrate how the interplay of concurrent compaction and crystallization can give rise to the modal changes of the Lower Zone–Critical Zone transition. Briefly, the IRIDIUM program links a phase equilibration routine based on the MELTS program (Ghiorso & Sack, 1995) with the compaction-driven mass transport equations of McKenzie (1984) as used by Shirley (1986) as well as standard heat transport equations. Further program details have been given by Boudreau (2003).
For the simulation, the composition of a Bushveld B1 sill (Maier & Barnes, 1998; Maier et al., 2000) was used as the starting composition. This composition has the crystallization order opx 
plag cpx. The initial state of the system is assumed to consist of an initially uniform liquid in a chamber 2 km thick with a small crystal pile at the base composed of 40% opx, 60% liquid. Crystallization of the chamber and growth of the pile is accomplished by allowing the top of the chamber to cool at a rate of 0·005°C/day. The magma is otherwise continuously mixed adiabatically (isentropic). Crystallization of the liquid in the chamber leads to crystals accumulating on the top of the crystal pile.
To mimic a cooling front moving upward through the crystal pile, the base is cooled from the initial near-liquidus temperature at a rate of 0·001°C/day until the bottom temperature reaches 500°C, at which point the temperature is held constant. Also, compaction and crystallization in the pile are ignored (not calculated) once the liquid fraction falls below 0·1. This residual liquid fraction can be thought of as a ‘trapped liquid’ component that would crystallize both more of the observed saturated phases as well as other, late crystallizing minor minerals. Finally, to better illustrate the effects of crystallization alone on compaction behavior, the liquid viscosity is set constant at 11 Pa s and the solid–liquid density difference constant at 0·3 g/cm3. The effective solid (matrix) viscosity is set to 5·0 x 1015 Pa s. These values are in part arbitrary and selected for illustrative purposes of the calculation.
The growth of the crystal pile is shown at the time steps of 0, 1021, 4042, 5063 and 6652 years in Fig. 17. For each pair of graphs at each time step, the graph on the left shows the temperature profile within the crystal pile, and the graph on the right shows the weight percent of the phases within the pile. The dashed horizontal line in each graph shows the top of the pile. The initial state of the system is shown at time zero, with a uniform T and liquid fraction in the crystal pile.
After 1021 years, compaction and crystallization resulting from cooling through the base of the crystal pile has reduced the liquid fraction to the computational limit in the lower part of the crystal pile. Up to this time, growth of the crystal pile by crystallization of additional pyroxene has approximately equaled the compaction subsidence rate such that the top of the crystal pile has not changed its location significantly. It should be noted that although the bottommost part of the crystal pile is cooling, the interstitial liquid does not reach plagioclase saturation. This is because the interstitial liquid is initially far from plagioclase saturation and the crystallizing interstitial liquid is essentially ‘compacted-out’ from the base of the crystal pile before it reaches plagioclase saturation.
After 4042 years, the main magma is still saturated only in orthopyroxene. However, both the liquid in the chamber and the higher-level interstitial liquids are approaching plagioclase saturation. At this time, the cooler, partly crystallized interstitial liquid about halfway up the pile crystallizes to the point of plagioclase saturation before the porosity is reduced by compaction. The solid assemblage now consists of mostly opx but with about 5% plagioclase. This amount of interstitial plagioclase increases only slightly with height until the point where the magma chamber becomes saturated in plagioclase.
Soon after 4042 years, the magma in the chamber becomes saturated in plagioclase (at the point marked plin on the 5063 years graph) and both plagioclase and orthopyroxene are deposited at the top of the pile. Once plagioclase becomes a cotectic phase, the pile grows relatively quickly such that, by 6652 years, the top of the crystal pile is at about 1600 m. Pile growth is so fast that hot crystals are buried by later, cooler crystals such that diffusive heat loss is through the top of the pile as well as through the base. Also, at some point between 5062 and 6652 years the magma has become saturated in cpx (at the point marked cpxin on the 6652 years graph). Compaction and crystallization has moved the effective bottom of the crystal pile to about 750 m.
The ‘haplo-Bushveld’ profile as it has developed after 6652 years mimics the general features of the Lower and Critical Zones of the Bushveld Complex. Thus the point marked LCZ is the model equivalent to the base the Lower Critical Zone, below which the orthopyroxenites contain very little plagioclase and above which the orthopyroxenites contain a higher but still relatively minor plagioclase component. Similarly, the point marked UCZ is equivalent to the base of the Upper Critical Zone; above this point the rocks are norites (and later gabbronorites) with relatively abundant plagioclase.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND SAMPLE...METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Results of this study are consistent with an increasing number of CSD studies (e.g. Waters & Boudreau, 1996
; Higgins, 2002), which show that igneous textures in layered intrusions can be strongly modified by crystal aging and compaction-driven recrystallization. Because of the proven importance of these processes, one cannot assume that observed textures in layered intrusions are in any remote sense primary. Our results are consistent with those of Higgins (2002), who noted that if there were any grain size evidence of phase segregation by crystal settling in the Kiglapait intrusion then it has been completely erased by crystal aging and recrystallization. Anyone who uses observed textures to infer magmatic processes (such as crystal settling) must first demonstrate that any observed textures have escaped any modification. |
SUPPLEMENTARY DATA |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND SAMPLE...METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Supplementary data for this paper are available on Journal of Petrology online.
|
ACKNOWLEDGEMENTS |
|---|
We thank Chris Beater of Samancor and Anglo American Platinum Corporation for access to the Cameroon study section. We also gratefully acknowledge the reviews of A. R. McBirney, M. D. Higgins and L. A. Coogan, whose suggestions greatly improved this paper. This work was supported by National Science Foundation grants EAR 99-02183 and EAR 02-06905.
|
FOOTNOTES |
|---|
* Corresponding author. E-mail: boudreau{at}duke.edu
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND SAMPLE...METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Benn, K. & Allard, B. (1989). Preferred mineral orientations related to magmatic flow in ophiolite layered gabbros. Journal of Petrology 30, 925–946.
Boorman, S. L., McGuire, J. B., Kruger, F. J. & Boudreau, A. E. (2003). Fluid overpressure in layered intrusions, Part II: Formation of a breccia pipe in the eastern Bushveld Complex, Republic of South Africa. Mineralium Deposita 38, 356–369.[Web of Science]
Boudreau, A. E. (2003). IRIDIUM—a program to model reaction of silicate liquid infiltrating a porous solid assemblage. Computers and Geosciences 29, 423–429.[CrossRef]
Boudreau, A. E. & McBirney, A. R. (1997). The Skaergaard Layered Series. Part III. Non-dynamic layering. Journal of Petrology 38, 1003–1020.[CrossRef][Web of Science]
Boudreau, A. E. & Philpotts, A. R. (2002). Quantitative modeling of compaction in the Holyoke flood basalt flow, Hartford Basin, Connecticut. Contributions to Mineralogy and Petrology 144, 176–184.[Web of Science]
Cameron, E. N. (1978). The Lower Zone of the Eastern Bushveld Complex in the Olifants river trough. Journal of Petrology 19, 437–462.
Cashman, K. V. (1990). Textural constraints on the kinetics of crystallization of igneous rocks. In: Nicholls, J. & Russell, J. K. (eds) Modern Methods of Igneous Petrology: Understanding Magmatic Processes. Mineralogical Society of America, Reviews in Mineralogy 24.
Cashman, K. V. & Marsh, B. D. (1988). Crystal size distribution (CSD) in rocks and the kinetics of crystallization II. Makaopuhi lava lake. Contributions to Mineralogy and Petrology 99, 292–305.[CrossRef][Web of Science]
Cawthorn, R. G. (1995). Models for incompatible trace element abundances in cumulus minerals and their application to plagioclase and pyroxene in the Bushveld Complex. Contributions to Mineralogy and Petrology 123, 109–115.
Cawthorn, R. G. (1999). Permeability of the footwall cumulates to the Merensky Reef, Bushveld Complex. South African Journal of Geology 102, 293–310.[Abstract]
Coogan, L. A., Thompson, G. & MacLeod, C. J. (2002). A textural and geochemical investigation of high level gabbros from the Oman ophiolite: implications for the role of the axial magma chamber at fast spreading ridges. Lithos 63, 67–82.[CrossRef][Web of Science]
Dick, H. J. B. & Sinton, J. M. (1979). Compositional layering in alpine peridotites: evidence for pressure solution creep in the mantle. Journal of Geology 87, 403–416.[Web of Science]
Durney, D. W. (1976). Pressure-solution and crystallization deformation. Philosophical Transactions of the Royal Society of London 283, 229–240.
Eales, H. V. & Cawthorn, R. G. (1996). The Bushveld Complex. In: Cawthorn, R. G. (ed.) Layered Intrusions. Amsterdam: Elsevier, pp. 181–229.
Eales, H. V., Maier, W. D. & Teigler, B. (1991). Corroded plagioclase feldspar inclusions in orthopyroxene and olivine of the Lower and Critical Zones, Western Bushveld Complex. Mineralogical Magazine 55, 479–486.[Web of Science]
Finney, J. (1968). Random packing and the structure of the liquid state. Ph.D. thesis, University of London.
Garrido, C. J., Keleman, P. B. & Hirth, G. (2001). Variation in cooling rate with depth in lower crust formed at an oceanic spreading ridge: plagioclase crystal size distributions in gabbros from the Oman ophiolite. Geochemistry, Geophysics, Geosystems 2, 2000GC00136.
Ghiorso, M. S. & Sack, R. O. (1995). Chemical mass transfer in magmatic processes IV. A revised and internally consistent thermodynamic model for the interpretation and extrapolation of liquid–solid equilibria in magmatic systems at elevated temperatures and pressures. Contributions to Mineralogy and Petrology 119, 197–212.[Web of Science]
Godard, M., Bodinier, J. L. & Vasseur, G. (1995). Effects of mineralogical reactions on trace element distributions in mantle rocks during percolation processes: a chromatographic approach. Earth and Planetary Science Letters 133, 449–461.[CrossRef][Web of Science]
Gray, N. H., Philpotts, A. R. & Dickson, L. D. (2003). Quantitative measures of textural anisotropy resulting from magmatic compaction illustrated by a sample from the Palisades sill, New Jersey. Journal of Volcanology and Geothermal Research 121, 293–312.[CrossRef][Web of Science]
Green, H. W. (1966). Preferred orientation of quartz due to recrystallization during deformation. American Geophysical Union Transactions 47, 491.
Green, H. W. (1967). Quartz: extreme preferred orientation produced by annealing. Science 157, 1444–1447.
Hardy, S. C. & Voorhees, P. W. (1988). Ostwald ripening in a system with a high volume fraction of coarsening phase. Metallurgical Transactions Acta 19A, 2713–2721.
Harvey, P. K. & Laxton, R. R. (1980). The estimation of finite strain from the orientation distribution of passively deformed linear markers; eigenvalue relationships. Tectonophysics 70, 285–307.[CrossRef][Web of Science]
Higgins, M. D. (1991). The origin of a laminated and massive anorthosite, Sept Iles layered intrusion, Quebec, Canada. Contributions to Mineralogy and Petrology 106, 340–354.[CrossRef][Web of Science]
Higgins, M. D. (1998). Origin of anorthosite by textural coarsening: quantitative measurements of a natural sequence of textural development. Journal of Petrology 39, 1307–1323.[CrossRef][Web of Science]
Higgins, M. D. (1999). Origin of megacrysts in granitoids by textural coarsening: a crystal size distribution (CSD) study of microcline in the Cathedral Peak granodiorite, Sierra Nevada, California. In: Castro, A., Fernandez, C. & Vigneresse, J. L. (eds) Understanding Granites: Integrating New and Classical Techniques. Geological Society, London, Special Publications 158, 207–219.
Higgins, M. D. (2000). Measurement of crystal size distributions. American Mineralogist 85, 1105–1116.
Higgins, M. D. (2002a). The role of textural coarsening in the development of the Kiglapait layered mafic intrusion, Labrador, Canada: a crystal size distribution study. Contributions to Mineralogy and Petrology 144, 314–330.[Web of Science]
Higgins, M. D. (2002b). Closure in crystal size distributions (CSD), verification of CSD calculations and the significance of CSD fans. American Mineralogist 87, 171–175.
Hulburt, L. J. & Von Gruenewaldt, G. (1985). Textural and compositional features of chromite in the Lower and Critical Zones of the Bushveld Complex, south of Potgietersrus. Economic Geology 80, 872–895.
Hunter, R. (1986). Textural equilibrium in layered igneous rocks. In: Parsons, I. (ed.) Origins of Igneous Layering. Dordrecht: D. Reidel, pp. 473–503.
Irvine, T. N. (1980). Magmatic infiltration metasomatism, double diffusive fractional crystallization and adcumulus growth in the Muskox Intrusion and other layered intrusions. In: Hargraves, R. B. (ed.) Physics of Magmatic Processes. Princeton, NJ: Princeton University Press, pp. 325–384.
Jackson, E. D. (1961). Primary textures and mineral associations in the Ultramafic Zone of the Stillwater Complex, Montana. US Geological Survey Professional Paper 358, 1–106.
Jerram, D. A. & Cheadle, M. J. (2000). On the cluster analysis of crystals in rocks. American Mineralogist 85, 47–67.
Jerram, D. A., Cheadle, M. J., Hunter, R. H. & Elliot, M. T. (1996). The spatial distribution of grains and crystals in rocks. Contributions to Mineralogy and Petrology 125, 60–74.[CrossRef][Web of Science]
Jerram, D. A., Cheadle, M. J. & Philpotts, A. R. (2003). Quantifying the building blocks of igneous rocks: are clustered crystal frameworks the foundation? Journal of Petrology 44, 2033–2051.
Klein, E. M., Langmuir, C. H. & Staudigel, H. (1991). Geochemistry of basalts from the southeast Indian Ridge, 115 E–138 E. Journal of Geophysical Research 96, 2089–2107.
Kretz, R. (1969). On the spatial distribution of crystals in rocks. Lithos 2, 39–65.
Kruger, F. J. (1990). The stratigraphy of the Bushveld Complex: a reappraisal and the relocation of the Main Zone boundaries. South African Journal of Geology 93, 376–381.[Abstract]
Langmuir, C. H. (1989). Geochemical consequences of in situ crystallization. Nature 340, 199–205.[CrossRef]
Maaloe, S. (1976). The zoned plagioclase of the Skaergaard intrusion, East Greenland. Journal of Petrology 17, 398–419.
Maier, W. D. & Barnes, S. J. (1998). Concentrations of rare earth elements in silicate rocks of the Lower, Critical and Main Zones of the Bushveld Complex. Chemical Geology 150, 85–103.[CrossRef][Web of Science]
Maier, W. D., Arndt, N. T. & Curl, E. A. (2000). Progressive crustal contamination of the Bushveld Complex: evidence from Nd isotopic analyses of the cumulate rocks. Contributions to Mineralogy and Petrology 140, 316–327.[CrossRef][Web of Science]
Marsh, B. D. (1988). Crystal size distribution (CSD) in rocks and the kinetics and dynamics of crystallization I. Theory. Contributions to Mineralogy and Petrology 99, 277–291.[CrossRef][Web of Science]
Marsh, B. D. (1998). On the interpretation of crystal size distributions in magmatic systems. Journal of Petrology 39, 553–599.[CrossRef][Web of Science]
Mathez, E. A. (1995). Magmatic metasomatism and the formation of the Merensky Reef, Bushveld Complex. Contributions to Mineralogy and Petrology 199, 277–286.[CrossRef]
Mathez, E. A., Hunter, R. H. & Kinzler, R. (1997). Petrologic evolution of partially molten cumulate: the Atok section of the Bushveld Complex. Contributions to Mineralogy and Petrology 129, 20–34.[CrossRef][Web of Science]
Mathison, C. I. (1987). Pyroxene oikocrysts in troctolitic cumulates—evidence for supercooled crystallization and post-cumulus modification. Contributions to Mineralogy and Petrology 97, 228–236.[CrossRef][Web of Science]
McBirney, A. R. (1995). Mechanisms of differentiation of the Skaergaard Intrusion. Journal of the Geological Society, London 152, 421–435.
McBirney, A. R. (2002). The Skaergaard Layered Series; Part VI, excluded trace elements. Journal of Petrology 43, 535–556.
McBirney, A. R. & Nicolas, A. (1997). The Skaergaard Layered Series, Part II. Magmatic flow and dynamic layering. Journal of Petrology 38, 569–580.[CrossRef][Web of Science]
McKenzie, D. (1984). The generation and compaction of partially molten rock. Journal of Petrology 25, 713–765.
Meurer, W. P. & Boudreau, A. E. (1996). The petrology and mineral compositions of the Middle Banded series of the Stillwater Complex, Montana. Journal of Petrology 37, 583–607.
Meurer, W. P. & Boudreau, A. E. (1998). Compaction of igneous cumulates part II: Compaction and the development of igneous foliations. Journal of Geology 106, 293–304.[Web of Science]
Meurer, W. P., Klaber, S. A. & Boudreau, A. E. (1997). Discordant bodies from olivine-bearing zones III and IV of the Stillwater complex, Montana—evidence for post-cumulus fluid migration in layered intrusions. Contributions to Mineralogy and Petrology 130, 81–92.[CrossRef][Web of Science]
Mock, A., Jerram, D. A. & Breutkreuz, C. (2003). Using quantitative textural analysis to understand the emplacement of shallow-level rhyolitic laccoliths—a case study from the Halle Volcanic Complex, Germany. Journal of Petrology 44, 833–849.
Nicolas, A. & Poirier, J. P. (1976). Crystalline Plasticity and Solid State Flow in Metamorphic Rocks. London: John Wiley, 444 pp.
Park, Y. & Hanson, B. (1999). Experimental investigations of Ostwald ripening rates of forsterite in the haplobasaltic system. Journal of Volcanology and Geothermal Research 90, 103–113.[CrossRef][Web of Science]
Philpotts, A. R., Carroll, M. & Hill, J. M. (1996). Crystal-mush compaction and the origin of pegmatitic segregation sheets in a thick flood basalt flow in the Mesozoic Hartford Basin, Connecticut. Journal of Petrology 37, 811–836.
Philpotts, A. R., Shi, J. & Brustman, C. (1998). Role of plagioclase chains in the differentiation of partly crystallized basaltic magma. Nature 395, 343–346.[CrossRef]
Rabinowicz, M., Getnthon, P., Ceuleneer, G. & Hillairet, M. (2001). Compaction in a mantle mush with high melt concentrations and the generation of magma chambers. Earth and Planetary Science Letters 188, 313–328.[CrossRef][Web of Science]
Raghavan, V. & Cohen, M. (1975). Solid state transformations. In: Hannay, N. B. (ed.) Treatise on Solid State Chemistry, Volume 5: Changes of State. New York: Plenum, pp. 67–127.
Resmini, R. G. & Marsh, B. D. (1995). Steady-state volcanism. Paleoeffusion rates, and magma system volume inferred from plagioclase crystal size distributions in mafic lavas: Dome Mountain, Nevada. Earth and Planetary Science Letters 68, 273–296.
Sharpe, M. R. (1985). Strontium isotope evidence for preserved density stratification in the main zone of the Bushveld Complex, South Africa. Nature 316, 119–126.[CrossRef]
Shirley, D. N. (1986). Compaction of igneous cumulates. Journal of Geology 94, 795–809.[Web of Science]
Teigler, B. & Eales, H. V. (1996). The lower and critical zones of the western limb of the Bushveld Complex as intersected by the Nooitgedacht boreholes. Geological Survey of South Africa Bulletin 111.
Walraven, F. (1988). Notes on the age and genetic relationships of the Makhutso granite, Bushveld Complex, South Africa. Chemical Geology 72, 17–28.
Waters, C. & Boudreau, A. E. (1996). A reevaluation of crystal size distributions in chromite cumulates. American Mineralogist 81, 1452–1459.[Abstract]
Willmore, C. C., Boudreau, A. E. & Kruger, F. J. (2000). The halogen geochemistry of the Bushveld Complex, Republic of South Africa: implications for chalcophile element distribution in the Lower and Critical Zones. Journal of Petrology 41, 1517–1539.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  B. O'Driscoll, C. T. E. Stevenson, and V. R. Troll Mineral Lamination Development in Layered Gabbros of the British Palaeogene Igneous Province: A Combined Anisotropy of Magnetic Susceptibility, Quantitative Textural and Mineral Chemistry Study J. Petrology, June 1, 2008; 49(6): 1187 - 1221. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. B. Holness, C. Tegner, T. F. D. Nielsen, G. Stripp, and S. A. Morse A Textural Record of Solidification and Cooling in the Skaergaard Intrusion, East Greenland J. Petrology, December 1, 2007; 48(12): 2359 - 2377. [Abstract] [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]
|
|
|
|
|
|
B. O'Driscoll, C. H. Donaldson, V. R. Troll, D. A. Jerram, and C. H. Emeleus An Origin for Harrisitic and Granular Olivine in the Rum Layered Suite, NW Scotland: a Crystal Size Distribution Study J. Petrology, February 1, 2007; 48(2): 253 - 270. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. G. CAWTHORN and K. BOERST Origin of the Pegmatitic Pyroxenite in the Merensky Unit, Bushveld Complex, South Africa J. Petrology, August 1, 2006; 47(8): 1509 - 1530. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||