|
|
 |
|||||||||||||||||
|
|||||||||||||||||
 |
|||||||||||||||||
| JOURNAL HOME | HELP | CONTACT PUBLISHER | SUBSCRIBE | ARCHIVE | SEARCH | TABLE OF CONTENTS |
1 GeoForschungsZentrum Potsdam, Telegrafenberg, D-14473 Potsdam, Germany
2 Frei Universität Berlin, Institut für Geologische Wissenschaften, Malteserstr. 74-100, D-12249 Berlin, Germany
Correspondence: * E-mail: dharlov{at}gfz-potsdam.de
 |
ABSTRACT |
|---|
 Top Abstract IntroductionExperimental procedure and...ResultsDiscussionEvaluation of kinetic dataAcknowledgmentsReferences cited |
|---|
Cnd + Qtz is, at some point, in direct competition with Cnd + Qtz 
Ky and Sil Ky. Early during the experiment it may be assumed that a steady state between dissolution and growth rates is established. However, due to the sluggish nucleation of kyanite, there is a P-T dependent induction period during which Cnd + Qtz Sil is the controlling reaction. Once kyanite does appear, the reaction proceeds very fast to kyanite via reactions Cnd + Qtz Ky and Sil Ky. The kyanite surface area is probably a major factor in controlling the overall reaction rates. Under constant P and T, the system evolves from metastable sillimanite formation to sillimanite consumption, which is only dependent on the kyanite surface area. Similar competition between stable and metastable reactions could occur during contact metamorphism where metastable mineral growth is observed. The relative sluggishness of all three reactions under relatively dry conditions could help to explain the persistence of metastable corundum + quartz ± Al2SiO5 assemblages in nature.
Key Words: Sillimanite • kyanite • corundum • quartz • rate of reaction • experimental petrology • Rietveld refinement • aluminosilicates
|
INTRODUCTION |
|---|
|
TopAbstractIntroductionExperimental procedure and...ResultsDiscussionEvaluation of kinetic dataAcknowledgmentsReferences cited |
|---|
From a purely experimental standpoint, the location of the metastable Sil-Cnd-Qtz equilibrium within the kyanite stability field has never been located in P-T space. This fact, as well as the occurrence and possible stability of the assemblage corundum + quartz ± sillimanite in nature, has motivated a new series of experiments involving sillimanite in the Al2O3-SiO2 system with two specific goals. The first of these was to experimentally define the approximate location in P-T space of the metastable equilibrium of sillimanite relative to corundum and quartz in the kyanite stability field, over a series of temperatures within the time period of a few hours to a few weeks. The second goal was to relate the various reaction rates between kyanite, sillimanite, corundum, and quartz, within the kyanite stability field, for a series of specific temperatures and pressures under both H2O-rich and H2O-poor conditions. The metastability of the assemblage Sil-Cnd-Qtz implies that the reaction Cnd + Qtz Sil is always in competition with Cnd + Qtz Ky and Sil Ky (Fig. 1
). However, there should be an induction period of a certain duration before kyanite nucleates that is dependent on temperature, pressure, and the amount of flux, i.e., H2O, available. As such, the induction period then provides an opportunity to study the relative rates of the sillimanite and kyanite forming parallel reactions as a function of temperature, pressure, and quantity of H2O. In a broader context, an experimental study of the competition between stable and metastable reactions allows for further insight to be gained with regard to the role of metastable reactions during metamorphism and their subsequent influence on the extent and persistence of mineral phases in rock fabrics.
|
|
EXPERIMENTAL PROCEDURE AND EVALUATION |
|---|
|
TopAbstractIntroductionExperimental procedure and...ResultsDiscussionEvaluation of kinetic dataAcknowledgmentsReferences cited |
|---|
-Al2O3 (1–2 µm) and a synthetic, dry quartz crushed in ethanol to a grain size of 10–20 µm. Two varieties of sillimanite were used. The first consisted of gem quality, clear, pale blue crystals from Sri Lanka with no evidence of intergrown sheet silicates or other foreign phases such as quartz. The only contaminants were very minor amounts of goethite along some cracks and in rough areas on the surface of the sillimanite crystals. The sillimanite was first broken into large fragments, which were cleaned ultrasonically in an ethanol bath. Fragments of clean sillimanite totally free of goethite on the surface or along cracks were then hand picked under the binocular microscope (50x). The second variety of sillimanite consisted of gem quality, pale-brown crystals, free of sheet silicates or inclusions, hand-picked [binocular microscope (50x)] from a metapelitic pegmatoid associated with a charnockitic magmatic intrusion located in the Reinbolt Hills, Princess Elisabeth Land, East Antarctica (Grew 1980; Nicols and Berry 1991). Subsequent crushing of both sillimanite samples in ethanol to a grain size 15–50 µm and mounting in refractive index oil (n = 1.540) showed no foreign phases at the microscopic level.
Electron microprobe (EMP) analysis of both sillimanites was made using the CAMECA SX50 electron microprobe at the GeoForschungsZentrum Potsdam (Table 1). EMP operating conditions were 15 kV and 50 nA, using a 1 µm diameter electron beam with counting times of 90 s on each element. Standards included kyanite, magnetite, ilmenite, and chromite. Detection limits were ~200–300 ppm for all elements. EMP analysis of the Sri Lanka sillimanite indicated an Fe2O3 content of <0.30 wt% and for the Reinbolt Hills sillimanite, an Fe2O3 content of ~1.25 wt% (Table 1). Analysis of Cr and Mn indicated these elements to be below EMP detection limits.
|
). Experiments using the Sri Lanka sillimanite were labeled SCQ, whereas experiments using the Reinbolt Hills sillimanite were labeled FSCQ. Ten milligrams of the Sil-Cnd-Qtz mix, plus H2O as a flux, were loaded into 3 mm wide, 1.3 cm long Pt capsules. While partially immersed in an ice water bath, the Pt capsules were arc-welded shut using an argon plasma torch. After folding, the capsules were checked for leaks by first weighing, then placing them in a 105 °C oven for 24 h, and then re-weighing. The amount of H2O used consisted of either 0.4–0.6 or 2–3 mg (Table 3).
|
|
, the Pt capsule needed to remain sealed during the experiment. In addition, the set-up was required to be mechanically perfect with regard to the placement of the Pt capsule and the thermocouple tip. Last, any experiment in which there were any significant shifts in either pressure or temperature during the run was discarded.
Evaluation of XRD spectra
Growth or decrease of a phase was determined using powder X-ray diffractometer (XRD) spectra and subsequent Rietveld analysis of these spectra. For powder XRD spectra, 2–3 mg of the charge were ground in acetone in an agate mortar to a grain size of <5–10 µm. The powder was diluted with Elmer’s white glue and mounted on a circular acetate foil. Powder XRD patterns were recorded via transmission through the rotating foil using a fully automated STOE STADI P diffractometer (CuK1 radiation), equipped with a primary monochromator and a 7° position-sensitive detector. Operating conditions were 40 kv and 40 mA, with a take off angle of 6°. The spectra were recorded in the range of 5–125° (2
) using a step interval of 0.1°. The resolution of the PSD was set to 0.02°. Counting times were selected to yield a maximum intensity of 2000 to 3000 counts for each sample, resulting in 5 to 20 s per detector step. Multiple XRD measurements of the same sample gave good reproducibility.
The Rietveld refinements were performed using the GSAS software package (Larson and von Dreele 1987). This allowed for both the characterization and, more importantly, the determination of the relative molar amounts of each crystalline phase present, here sillimanite, kyanite, corundum, and quartz. The number of profile parameters used ranged from 13 to 15. These consisted of 8 to 10 parameters to fit the background using a real space correlation and 5 parameters to define the peak form as a pseudo-Voigt with a variable Lorentzian character. No parameters describing peak asymmetry were necessary because the peak shape is highly symmetric due to the geometry of the STOE STADI P diffractometer. The preferred orientation was corrected using spherical harmonic functions. During the refinement of the XRD spectra, scale-factor, background, zero-point correction, unit-cell parameters, phase proportions, preferred orientations, profile parameters, atomic positions, and fractions were all taken into account. The isotropic displacement factors were not refined.
In Table 2, the output from each Rietveld refinement, in terms of relative molar amounts of each phase, is listed for a random sample from each of the mixes, i.e., SCQ-I, SCQ-II, SCQ-III, and FSCQ, for a series of test samples from the SCQ-I and SCQ-II mixes (SCQ-I-1 through SCQ-II-4 and SCQ-II-1 through SCQ-II-6), and for a series of individually mixed 10 mg size samples (TSCQ-1, TSCQ-2, and TSCQ-3). In Table 3, the output from each Rietveld refinement, again in terms of relative molar amounts of each phase, is listed for each successful experiment. In Table 3, sillimanite growth is considered to have occurred if the sillimanite molar fraction is greater than or equal to 0.37.
|
RESULTS |
|---|
|
TopAbstractIntroductionExperimental procedure and...ResultsDiscussionEvaluation of kinetic dataAcknowledgmentsReferences cited |
|---|
; Fig. 2). The upper bound of sillimanite metastability tentatively outlines the lower pressure limit, for the metastable Sil-Cnd-Qtz equilibrium. At pressures above this lower pressure limit, there was a partial to massive reaction to kyanite with subsequent complementary reduction in sillimanite (and corundum + quartz) over the same approximate time period seen for sillimanite growth at 50 MPa less pressure (see Table 3 and Fig. 2). Using the same criteria with respect to whether sillimanite growth had occurred or not, experiments using the Fe-rich Reinbolt Hills sillimanite at 700 and 800 °C shifted the metastable Sil-Cnd-Qtz equilibrium up 50 MPa at 800 °C but at 700 °C, any shift in pressure was indeterminate (Table 3). Sillimanite growth took the form of either plates (Fig. 3a) or new, needle-like crystals (Fig. 3b) growing on the sillimanite grains from the original Sil-Cnd-Qtz mix.
|
|
). At 600 °C, the sillimanite- and kyanite-forming reactions are extremely slow on the laboratory time scale such that measuring reaction rates at different pressures is not practicable. At 900 °C, kyanite nucleates readily and the kyanite-forming reactions are very fast such that this temperature is also problematic for a kinetic interpretation of sillimanite growth.
For a study of the kinetics of the metastable reaction Cnd + Qtz Sil, only those runs were selected in which no kyanite was present in the reaction products. Experiments with low H2O content (0.4–0.6 mg) at 700 and 800 °C were either indeterminate with respect to reaction or the conversion was so small that no reliable reaction rates could be derived. Therefore, only experiments with high H2O contents (2–3 mg) were considered for this evaluation. Conversion to sillimanite is the difference between the molar amount of sillimanite in the batch before and after the experiment relative to the molar amounts of the reactants, i.e., quartz plus corundum in the starting mixture. Reaction rates are tentatively calculated to vary linearly with time in accordance with surface-controlled reaction kinetics, since for moderate ranges of conversion the surface area of each reactant or product mineral (all initially present in equal molar amounts) can be taken as constant. The linear extrapolation to zero reaction rates then should point to equilibrium between reactant and product minerals.
Results from experiments at 700 °C (<200 h) reveal a weak trend for declining reaction rates at higher pressures approaching the proposed limit of metastable existence of Sil relative to Cnd + Qtz (Fig. 4). However, reaction rates are too small to allow extrapolation to an equilibrium pressure. In the 800 °C experiments, there was clearly reaction to sillimanite at 1800 MPa. Linear extrapolation points to an equilibrium pressure near 1900 MPa; however, in one experiment at 1850 MPa (SCQ-50) there was already almost total conversion to kyanite after 12 h. At 900 °C (not plotted), the reaction kinetics were fast enough to detect strong effects even in low H2O experiments (0.4–0.6 mg) after 3 h (Table 3). At 1900 MPa, after 3 h essentially no reaction had occurred, indicating that these P-T conditions are near the metastable Cnd + Qtz Sil equilibrium. At 1950 and 2000 MPa (SCQ-20, SCQ-17), increasing amounts of kyanite are present after 3 h.
|
, the molar amounts of corundum, quartz, sillimanite, and kyanite (if present) are shown for experiments at 800 °C and pressures from 1600 to 2000 MPa. In the starting mixtures corundum, quartz, and sillimanite are present in equal molar amounts within weighing error. Thus, if no reaction had occurred, the three of them should cluster around 0.333. Any difference between quartz and corundum cannot be due to reaction but is a measure of the cumulative error resulting from preparation and analysis procedures. Figure 5a depicts the situation after 24 h run time, and Figure 5b after 48 h. In any case, the reactions are significantly slower for H2O-poor conditions (0.4–0.6 mg H2O; open symbols) than at more hydrous conditions (2–3 mg H2O; filled symbols). In H2O-poor experiments at 1600 MPa, there is a weak signal for sillimanite growth. At 1700 to 1800 MPa, there is virtually no reaction or, depending on the onset of kyanite nucleation, all reactants become consumed in favor of kyanite. Above 1800 MPa, kyanite becomes the dominant phase and the metastable reaction to sillimanite cannot be observed anymore.
|
illustrates the results from isothermal experiments at 800 °C for 1800, 1700, and 1600 MPa, all of which are below the proposed metastable Sil-Cnd-Qtz equilibrium. These diagrams indicate that there is generally a slight or significant amount of sillimanite growth before the onset of kyanite formation. The first appearance of kyanite in individual runs for a given P-T condition is subject to some scatter; however, there is a systematic trend for kyanite to occur earlier in the experiments the higher the pressure. Once kyanite appears, the reactions Sil Ky and Cnd + Qtz Ky soon outpace Cnd + Qtz Sil. Nucleation of kyanite becomes distinctly sluggish with decreasing temperature.
|
|
DISCUSSION |
|---|
|
TopAbstractIntroductionExperimental procedure and...ResultsDiscussionEvaluation of kinetic dataAcknowledgmentsReferences cited |
|---|
indicates the lower pressure limit for any detectable sillimanite growth in the experiments. There are two ways to explain this line, one with a thermodynamic and the other with a kinetic bias. In the thermodynamic interpretation, this line represents the equilibrium Sil Cnd + Qtz. This assumption, however, is challenged by the fact that only sillimanite growth at pressures below this boundary could be measured, but reversals above this line failed. Instead of Cnd + Qtz growth, growth of kyanite consumes all other reactants. The reason for the absence of Cnd + Qtz-producing experiments could be the general sluggishness of the studied metastable reaction as illustrated by all H2O-poor experiments at 600 °C, most experiments at 700 and 800, and even some at 900 °C, which revealed no measurable reaction before the onset of kyanite growth. Cnd + Qtz is metastable with respect to sillimanite on the high-pressure side of their equilibrium where reaction to kyanite is even more favorable than on the low-pressure side, so that it is unlikely or even impossible to find experimental conditions where the metastable reaction to Cnd + Qtz can be detected before its products are consumed by formation of the stable phase, kyanite.
From a purely kinetic interpretation, this boundary has no thermodynamic meaning at all, but is simply a line in P-T space that indicates where the rate of nucleation and growth of kyanite begins to outpace sillimanite formation. The decreasing induction period at higher pressures, before the onset of kyanite growth, favors this interpretation. As Figure 6 shows, at 800 °C kyanite was a minor reaction product after 144 h at 1600 MPa, just started to appear after 36 h at 1700 MPa, and was already dominant after 24 h at 1800 MPa. Consequently, at higher pressures kyanite must almost instantly be present and prevent the progress of any metastable reactions. On the other hand, observations at 800 °C (Figs. 5 and 6) and at 900 °C (runs SCQ-16, -17, and -20) indicate that at the boundary in Figure 2, sillimanite growth had already ceased or declined below detection limits before kyanite entered the scene. This would mean that the positively sloped boundary, in fact, represents the limit of sillimanite metastability relative to quartz and corundum. We therefore assert that both (meta-) stability and reaction rates must be considered to interpret the experimental data. It is probably not accidental that the P-T-area where kyanite nucleation and growth starts to dominate over sillimanite growth seems to coincide with the metastable Sil-Cnd-Qtz equilibrium. At P-T conditions near the Sil-Cnd-Qtz metastable equilibrium, sillimanite growth is very slow, such that dissolved Al is more easily available for nucleation of kyanite than at lower pressures. Since sillimanite is unstable with respect to kyanite as well as Cnd + Qtz, it will break down rapidly and thereby trigger kyanite formation.
In the H2O-rich experiments, the incongruent dissolution of sillimanite and higher solubility of SiO2 than Al2O3 could theoretically shift the molar amounts of the minerals in the experimental mixes to a measurable extent. Dissolved sillimanite or kyanite would probably be present after the experiment as quench quartz and corundum, and would therefore increase the apparent amounts of the reactants. The increase in quartz then would be larger than in corundum. However, in the experiments where Sil or Ky growth could not be detected, the reaction (or apparent reaction) Sil Qtz + Cnd was never observed, i.e., the experimental mixes seem to remain unchanged within the detection limits of the Rietveld refinement. Also, there is no systematic increase in Qtz vs. Cnd in the run products of these experiments. Incongruent dissolution therefore did not play any role in increasing or decreasing the phase fractions in the experimental products.
Thermodynamic evaluation
In nature, Cnd + Qtz could be stable with respect to both kyanite and sillimanite if both the metastable Sil-Cnd-Qtz and Cnd-Qtz-Ky equilibria intersected in P-T space. Such an intersection would be possible if the metastable Sil-Cnd-Qtz equilibrium had a negative slope. However, equilibria calculated using enthalpy, entropy, and volume data for sillimanite from the most current internally consistent thermodynamic databases, e.g., Berman (1988) [–2586.09 kJ/mol; 95.93 J/(K·mol); 4.983 J/bar], Gottschalk (1997) [–2586.09 kJ/mol; 95.39 J/(K·mol);4.990 J/bar], and Holland and Powell (1998) [–2585.89 kJ/mol; 95.50 J/(K·mol); 4.986 J/bar] indicate that the calculated metastable Sil-Cnd-Qtz equilibrium has a positive slope in P-T space. This result means that for geologically relevant pressures and temperatures, the Sil-Cnd-Qtz and Cnd-Qtz-Ky equilibria plot approximately parallel to each other (Fig. 2). This observation is experimentally supported by the results from this study, which indicate that the lower pressure limit for the Sil-Cnd-Qtz equilibrium in P-T space has a positive slope that broadly parallels these three calculated Sil-Cnd-Qtz equilibria in Figure 2. The fact that the lower pressure limit for the Sil-Cnd-Qtz metastable equilibrium is roughly outlined by these calculated Sil-Cnd-Qtz equilibria would suggest that it is probably within the vicinity of the actual Sil-Cnd-Qtz metastable equilibrium. This interpretation is supported by the fact that, because the Sil-Cnd-Qtz equilibrium is the sum of the positively sloped Sil-Ky and Cnd-Qtz-Ky equilibria, intuitively, the Sil-Cnd-Qtz equilibrium must by default have a positive slope.
The relatively wide range of calculated equilibria for Sil-Cnd-Qtz in P-T space (Fig. 2), utilizing the data of Berman (1988), Gottschalk (1997), and Holland and Powell (1998), is the result of an extremely small 
Sr (3.65 J/(K·mol); 1 bar, 298.15 K) and vr (0.156 J/b; 1 bar, 298.15) for this equilibrium (see discussion in Berman 1988). These terms, coupled with the expansivity and compressibility functions for the adjustment of Vr and the adjustment of CP as a function of temperature in tandem with the subsequent adjustment of Hr and Sr, make both the Gibbs free energy of the Sil-Cnd-Qtz equilibrium:
 | (1) |
where Po = 1 bar, as well as the slope of this equilibrium:
 | (2) |
extremely sensitive to small perturbations in the free energy function for sillimanite, corundum, and/or quartz. As a consequence, small changes, well within experimental uncertainty, in the volume (expansivity and compressibility) and CP functions for sillimanite, corundum and/or quartz can result both in large shifts in P-T space for the Sil-Cnd-Qtz equilibrium over hundreds of degrees (Celcius) or even change the slope of this equilibrium from positive to negative (see discussion in Berman 1988).
The calculated positive slope for the Sil-Cnd-Qtz equilibrium, as well as its experimental confirmation in this study (cf. Fig. 2), run contrary to the calculated equilibrium of Anovitz et al. (1993). In that study, Anovitz et al. (1993) derived a negatively sloped Sil Cnd + Qtz equilibrium by combining two reversed equilibria, i.e., almandine + sillimanite = hercynite + quartz (Bohlen and Liotta 1986) and almandine + corundum = hercynite + sillimanite (Shulters and Bohlen 1989). The intersection of this calculated negatively sloped equilibrium for Sil-Cnd-Qtz with the positively sloped Cnd-Qtz-Ky equilibrium (cf. Fig. 2; Harlov and Milke 2002) suggested that Cnd + Qtz could be a stable assemblage at 1600 MPa and T > 1100 °C. Later, studies of natural occurrences of Cnd + Qtz, e.g., Guiraud et al. (1996), Shaw and Arima (1998), and Mouri et al. (2003, 2004), have utilized the calculated Cnd + Qtz stability field determined by Anovitz et al. (1993), to help explain the presence of co-existing Cnd + Qtz in relatively dry, high-grade rocks. In these studies, the original Cnd + Qtz P-T stability field has been substantially extended to values lower than those envisioned by Anovitz et al. (1993). This extension was accomplished by taking into account various additional factors such as uncertainties in the enthalpies of formation for sillimanite, corundum, and quartz (and subsequent error propagation) as well as the effect of minor components in the sillimanite and/or corundum such as the presence of Fe2O3.
The apparent positive slope of the metastable Sil-Cnd-Qtz equilibrium makes it approximately parallel to both the Cnd-Qtz-Ky and the Sil-Ky equilibrium. As a consequence, intersection (when possible) of the experimentally reversed Cnd-Qtz-Ky equilibrium with the calculated Sil-Cnd-Qtz equilibrium (cf. Berman 1988; Gottschalk 1997; Holland and Powell 1998) or the experimentally determined lower pressure limit for the Sil-Cnd-Qtz equilibrium from this study (cf. Fig. 2), could only occur at pressures and temperatures well beyond those typical of the lower crust and upper mantle of the Earth. This result would negate any possible stability field for Cnd + Qtz in either of these regions. It would imply that in nature, the presence of corundum in direct contact with quartz in crustal rocks, either with or without a sillimanite rim along the quartz-corundum boundary, as described by Mouri et al. (2003), Shaw and Arima (1998), Guiraud et al. (1996), Motoyoshi et al. (1990), etc., is indicative of low H2O activity conditions in the rock coupled with relatively sluggish reaction kinetics between corundum and quartz.
|
EVALUATION OF KINETIC DATA |
|---|
|
TopAbstractIntroductionExperimental procedure and...ResultsDiscussionEvaluation of kinetic dataAcknowledgmentsReferences cited |
|---|
):
 | (3) |
 | (4) |
 | (5) |
Changes in the amount of each of the product and reactant phases are therefore dependent on the extent of two reactions from reactions 3 to 5 or:
 | (6) |
 | (7) |
 | (8) |
The experiments summed in Table 3 indicate that the system goes through a reaction sequence. In the first stage, either no reaction is detectable or metastable reaction 3 occurs exclusively. Kyanite nucleation sets in after a certain induction period (cf. Fig. 6), which is a function of P-T and the amount of H2O present. Once reactions 4 and 5 start to proceed, the system transforms relatively fast almost totally into the stable phase kyanite (Fig. 6). Initially reactions 4 and 5 are equally fast, so that the ratio of (Cnd + Qtz)/Sil remains unchanged (compare Figs. 5 and 6). Once kyanite has become the dominant phase, reaction 4 gradually becomes impeded, such that sillimanite diminishes faster than Cnd + Qtz (compare Figs. 5 and 6). This finding indicates that the spatial separation between the two reactants and therefore the transport of dissolved components becomes effective for controlling the rate of reaction 4.
Minimum values for the rate of reaction 5 can be extracted by considering both parallel reactions equally fast. At 800 °C and 1600 MPa under H2O-rich conditions (SCQ-54), total conversion to kyanite occurred in less than 48 h, indicating a rate 
0.021/h for reaction 4. Reaction 4 was almost complete after 12 h at 1850 MPa (SCQ-50), indicating a reaction rate 0.077/h. After 3 h at 900 °C, reaction rates under H2O-poor conditions at 1950 MPa (SCQ-20) and 2000 MPa (SCQ-17) were 0.037 and 0.096/h, respectively (cf. Table 3). These rates show good agreement with previously published measurements of the Sil Ky conversion kinetics at 800 °C (1500 MPa) and 900 °C (1680 MPa) of 0.033 and 0.111/h, respectively (Ostapenko et al. 1991). The lack of complete agreement in such a comparison takes into account that the solid-fluid ratios in the experiments of Ostapenko et al. (1991) were not exactly the same as those in the present study, that there was an induction period in our unseeded experiments, and that the kyanite surface area in our experiments evolved in an uncontrolled manner. Ostapenko et al. (1991, 1992) claimed that the attachment of dissolved particles to the crystal surface is the rate-determining step for transformation reactions between the Al2SiO5 phases. However, they did not vary solid-fluid ratios in their experiments to test whether the varied transport capacity of the aqueous fluid is also a rate-controlling factor. In the present study in which the amount of H2O in the Pt capsule was varied, experiments revealed that this is clearly the case for reaction 3. Reaction 5 seems to be enhanced by a lower solid/fluid ratio in the same way. In the H2O-rich experiments at 800 °C (Fig. 6), sillimanite was completely converted to kyanite within less than 24 h after the first appearance of kyanite, whereas in the H2O-poor experiments, the time interval between the first appearance of kyanite and total conversion was much longer. Also, the observation that reaction 4 becomes slower than 5, once corundum and quartz become spatially isolated, is strong evidence for transport steps being effective in governing kyanite growth kinetics.
It is an intriguing problem by which mechanism the amount of H2O present in the experiment exerts control over reaction kinetics. At high pressure, quartz, corundum, and Al2SiO5 are relatively soluble minerals (Manning 1994a, 1994b; Tropper and Manning 2004). Diffusion of cationic species in H2O at the elevated temperatures of these experiments (Watson and Wark 1997) is fast such that homogeneous steady state concentrations within the fluid would be expected to evolve irrespective of the particular solid-fluid ratio. However, the dependence of reaction rates on the solid-fluid ratio in H2O-rich systems has been observed in earlier high P-T studies (Matthews and Goldsmith 1984; Heinrich et al. 1989; Milke and Metz 2002). It should be noted that for most experimental studies, which have identified a reaction as surface-controlled because the reaction rate proved to vary linearly with the surface area of one single reactant or product phase, the dependence of the reaction rate on the solid-fluid ratio was not tested. Hence, the dependence of the reaction rate on the amount of fluid present might be a much more common phenomenon than previously realized. In strictly surface-controlled reactions, variations in the solid-fluid ratio should not affect the reaction kinetics. However, if transport in the aqueous fluid were an equal factor, variation in the amount of H2O present would mean a variation in the amount of the dissolved and thus diffusing material. In such a case, a positive correlation between the amount of H2O present and the reaction rate would be expected.
Transport control for the reaction rate implies the existence of concentration gradients in the fluid. Concentration gradients are supported by direct experimental evidence for microsystems with locally enhanced dissolution and growth rates in hydrothermal experiments (Heinrich et al. 1989; Lüttge and Metz 1991; Milke and Metz 2002; Harlov et al. 2002, 2003, 2005). Evidently, under high P-T conditions, the aqueous fluid in these reacting systems is not homogeneous, but concentration gradients are established in the form of surface boundary layers or halos around growing or dissolving grains. The amount of dissolved material then becomes as important as the surface area of the involved minerals in governing reaction kinetics (see discussion in Lasaga 1986). The distinct dependence of sillimanite and kyanite growth rates on the solid-fluid ratio is clear evidence that reactions 3, 4, and 5 are not strictly surface controlled. Instead, it is evident that concentration gradients evolved within the fluid and that diffusive transport within these gradients strongly affected both sillimanite and kyanite growth kinetics. The studied reactions Cnd + Qtz Sil and Cnd + Qtz Ky thus provide one more example that reaction kinetics need not be controlled by one rate-determining step, but an interplay of several factors (Lasaga 1986). For a higher solid/fluid ratio, this combined rate control will, at some point, evolve into pure transport control (Dohmen and Chakraborty 2003).
The most startling observation in the kinetic analysis is the effect of metastable sillimanite on the reactivity of Cnd + Qtz deep within the kyanite stability field. Although sillimanite does not occur in reaction 4, its persistence largely controls kyanite nucleation and thereby the kinetics of kyanite formation. These same observations would also be applicable in very dry, deep-crustal rocks, though on a much larger time scale.
An important conclusion from this study is that the increasingly common observation in high-grade rocks of sillimanite rims on quartz in contact with corundum does not necessarily imply that the sillimanite grew under P-T conditions in the sillimanite stability field but could have just as easily have formed metastably in the kyanite stability field. In addition, the host rocks could have been moved in and out of the kyanite and sillimanite stability fields multiple times without any real effect on the subsequent persistence of the sillimanite rims. The overall implication here is that mineral assemblages in general do not guarantee a complete reflection of the P-T conditions under which they formed or the P-T path that they have subsequently followed during the history of the rock (see discussion in Wain et al. 2001). As such, the presence or near absence of fluids, the fluid chemistry, as well as metastable mineral reaction rates, can have dramatic consequences on how the petrofabric of a rock both forms and subsequently evolves or persists during the P-T history of the rock.
|
ACKNOWLEDGMENTS |
|---|
|
TopAbstractIntroductionExperimental procedure and...ResultsDiscussionEvaluation of kinetic dataAcknowledgmentsReferences cited |
|---|
|
Footnotes |
|---|
MANUSCRIPT RECEIVED March 30, 2007; MANUSCRIPT ACCEPTED October 23, 2007
|
REFERENCES CITED |
|---|
|
TopAbstractIntroductionExperimental procedure and...ResultsDiscussionEvaluation of kinetic dataAcknowledgmentsReferences cited |
|---|
Anovitz, L.M., Essene, E.J., Metz, G.W., Bohlen, S.R., Westrum Jr., E.F., and Hemingway, B.S. (1993) Heat capacity and phase equilibria of almandine, Fe3Al2Si3O12. Geochimica et Cosmochimica Acta, 57, 4191–4204.[CrossRef][ISI][GeoRef]
Berman, R.G. (1988) Internally-consistent thermodynamic data for minerals in the system Na2O-K2O-CaO-MgO-FeO-Fe2O3-Al2O3-SiO2-TiO2-H2O-CO2. Journal of Petrology, 29, 445–522.
Bohlen, S.R. and Liotta, J.J. (1986) A barometer for garnet amphibolites and garnet granulites. Journal of Petrology, 27, 1025–1034.
Bohlen, S.R., Montana, A., and Kerrick, D.M. (1991) Precise determination of the equilibria kyanite = sillimanite and kyanite = andalusite and a revised triple point for Al2SiO5 polymorphs. American Mineralogist, 76, 677–684.[Abstract][ISI][GeoRef]
Dasgupta, S. and Ehl, J. (1993) Reaction textures in a spinel-sapphirine granulite from the Eastern Ghats, India and their implications. European Journal of Mineralogy, 5, 537–543.
Dohmen, R. and Chakraborty, S. (2003) Mechanism and kinetics of element and isotopic exchange mediated by a fluid phase. American Mineralogist, 88, 1251–1270.
Gottschalk, M. (1997) Internally consistent thermodynamic data for rock forming minerals in the system SiO2-TiO2-Al2O3-Fe2O3-CaO-MgO-FeO-K2O-Na2O-H2O-CO2. European Journal of Mineralogy, 9, 175–223.
Grew, E.S. (1980) Sillimanite and ilmenite from high-grade metamorphic rocks of Antarctica and other areas. Journal of Petrology, 21, 39–68.
Guiraud, M., Kienast, J.-R., and Ouzegane, K. (1996) Corundum-quartz-bearing assemblage in the Ihouhaouene area (In Ouzzal, Algeria). Journal of Metamorphic Geology, 14, 755–761.[CrossRef][ISI][GeoRef]
Harlov, D.E. and Förster, H.J. (2003) Fluid-induced nucleation of (y +REE)-phosphate minerals within apatite: nature and experiment. Part II. Fluorapatite. American Mineralogist, 88, 1209–1229.
Harlov, D.E. and Milke, R. (2002) Stability of corundum and quartz relative to kyanite and sillimanite at high temperature and pressure. American Mineralogist, 87, 424–432.
Harlov, D.E. and Newton, R.C. (1993) Reversal of the metastable kyanite + corundum + quartz and andalusite + corundum + quartz equilibria and the enthalpy of the formation of kyanite and andalusite. American Mineralogist, 78, 594–600.[Abstract][ISI][GeoRef]
Harlov, D.E., Forster, H-J., and Nijland, T.G. (2002) Fluid induced nucleation of (y + REE)-phosphate minerals in apatite: nature and experiment Part I. chlorapatite. American Mineralogist, 87, 245–261.
Harlov, D.E., Wirth, R., and Förster, H.-J. (2005) An experimental study of dissolution-reprecipitation in fluorapatite: fluid infilitration and the formation of monazite. Contributions to Mineralogy and Petrology, 150, 268–286.[CrossRef][ISI][GeoRef]
Harris, N. (1981) The application of spinel-bearing metapelites to P/T determination: an example from south India. Contributions to Mineralogy and Petrology, 76, 229–233.[CrossRef][ISI][GeoRef]
Heinrich, W., Metz, P., and Gottschalk, M. (1989) Experimental investigation of the kinetics of the reaction: 1 tremolite + 11 dolomite = 8 forsterite + 13 calcite + 9 CO2 + 1 H2O. Contributions to Mineralogy and Petrology, 102, 163–173.[CrossRef][ISI][GeoRef]
Hiroi, Y., Ogo, Y., and Namba, K. (1994) Evidence for prograde metamorphic evolution of Sri Lankan pelitic granulites, and implications for the development of continental crust. Precambrian Research, 66, 245–263.[CrossRef][ISI][GeoRef]
Holland, T.J.B. and Powell, R. (1998) An internally consistent thermodynamic data set for phases of petrological interest. Journal of Metamorphic Geology, 16, 309–343.[CrossRef][ISI][GeoRef]
Johannes, W. (1973) Eine vereinfachte Piston-Zylinder-Apparatur hoher Genauigkeit. Neues Jahrbuch für Mineralogie, 1973 H7/8, 337–351.
Johannes, W., Bell, P.M., Mao, H.K., Boettcher, A.L., Chipman, D.W., Hays, J.F., Newton, R.C., and Seifert, F. (1971) An interlaboratory comparison of piston-cylinder pressure calibration using the albite-breakdown reaction. Contributions to Mineralogy and Petrology, 32, 24–38.[CrossRef][ISI][GeoRef]
Lal, R.K., Ackermand, D., and Upadhyay, H. (1987) P-T-X relationships deduced from corona textures in sapphirine-spinel-quartz assemblages from Paderu, Southern India. Journal of Petrology, 28, 1139–1168.
Larson, A.C. and von Dreele, R.B. (1987) Generalized structure analysis system (GSAS). Los Alamos National Laboratory Report No. LA-UR-86-748
Lasaga, A.C. (1986) Metamorphic reaction rates laws and the development of isograds. Mineralogical Magazine, 50, 359–373.[ISI][GeoRef]
Lüttge, A. and Metz, P. (1991) Mechanism and kinetics of the reaction 1 dolomite + 2 quartz = 1 diopside + 2 CO2 investigated by powder experiments. Canadian Mineralogist, 29, 803–821.[ISI]
Manning, C.E. (1994a) The solubility of quartz in H2O in the lower crust and uppermantle. Geochimica et Cosmochimica Acta, 58, 4831–4839.[CrossRef][ISI][GeoRef]
——— (1994b) Experimental determination of the solubility’s of quartz, kyanite and corundum in H2O in the lower crust and upper mantle. Mineralogical Magazine, 58A L-Z, 555–556.
Matthews, A. and Goldsmith, J.R. (1984) The influence of metastability on reaction kinetics involving zoisite formation from anorthite at elevated pressures and temperatures. American Mineralogist, 69, 848–857.[Abstract][ISI][GeoRef]
Milke, R. and Metz, P. (2002) Experimental investigation of the kinetics of the reaction wollastonite + calcite + anorthite = grossular + CO2. American Journal of Science, 302, 312–345.
Motoyoshi, Y., Hensen, B.J., and Matsueda, H. (1990) Metastable growth of corundum adjacent to quartz in a spinel-bearing quartzite from the Archaean Napier Complex, Antarctica. Journal of Metamorphic Geology, 8, 125–130.[ISI][GeoRef]
Mouri, H., Andreoli, M.A.G., Kienast, J.R., Guiraud, M., and de Waal, S.A. (2003) First occurrence of the rare "corundum + quartz" assemblage in the high-grade zone from the Namaqualand Metamorphic Complex, South Africa: Evidence for higher P,T metamorphism? Mineralogical Magazine, 67, 1015–1021.
Mouri, H., Guiraud, M., and Osanai, Y. (2004) Review on "corundum + quartz" assemblage in nature: Possible indicator of ultra-high temperature conditions? Journal of Mineralogical and Petrological Sciences, 99, 159–163.[CrossRef][ISI][GeoRef]
Nichols, G.T. and Berry, R.F. (1991) A decompressional P-T path, Reinbolt Hills, East Antarctica. Journal of Metamorphic Geology, 9, 257–266.[CrossRef][ISI][GeoRef]
Ostapenko, G.T., Gorogotskaya, L.I., and Timoshkova, L.P. (1991) Sillimanite-kyanite conversion kinetics. Geochemistry International, 28, 79–85.
Ostapenko, G.T., Gorogotskaya, L.I., Timoshkova, L.P., and Chibisov, A.I. (1992) Kyanite andalusite, kyanite sillimanite, and andalusite sillimanite transformation kinetics. Geochemistry International, 29, 1–10.
Perchuk, L., Gerya, T., and Nozhkin, A. (1989) Petrology and retrograde P-T path in granulites of the Kanskaya formation, Yenisey range, eastern Siberia. Journal of Metamorphic Geology, 7, 599–617.[ISI][GeoRef]
Powers, R.E. and Bohlen, S.R. (1985) The role of synmetamorphic igneous rocks in the metamorphism and partial melting of metasediments, Northwest Adirondacks. Contributions to Mineralogy and Petrology, 90, 401–409.[CrossRef][ISI][GeoRef]
Santosh, M. (1987) Cordierite gneisses of southern Kerala, India: Petrology, fluid inclusions and implications for crustal uplift history. Contributions to Mineralogy and Petrology, 96, 343–356.[CrossRef][ISI][GeoRef]
Shaw, R.K. and Arima, M. (1998) A corundum-quartz assemblage from the Eastern Ghats Granulite Belt, India: evidence for high P-T metamorphism? Journal of Metamorphic Geology, 16, 189–196.[CrossRef][ISI][GeoRef]
Shulters, J. and Bohlen, S.R. (1989) The stability of hercynite-gahnite spinels in corundum- or quartz-bearing assemblages. Journal of Petrology, 30, 1017–1032.
Tracy, R.J. and McLellan, E.L. (1985) A natural example of the kinetic controls of compositional and textural equilibration. In A.B. Thompson and D.C. Rubie, Eds., Metamorphic Reactions, Kinetics, Textures, and Deformation, p. 118–137. Advances in Physical Geochemistry, New York.
Tropper, P. and Manning, C.E. (2004) The solubility of rutile and corundum in H2O at high P and T: Constraints on Ti and Al mobility during high-P metamorphism. Lithos, 73, Supplement 1, S 113.[ISI][GeoRef]
Wain, A.L., Waters, D.J., and Austrheim, H. (2001) Metastability of granulites and processes of eclogitisation in the UHP region of western Norway. Journal of Metamorphic Geology, 19, 609–625.[CrossRef]
Waters, D.J. (1991) Hercynite-quartz granulites: Phase relations and implications for crustal processes. European Journal of Mineralogy, 3, 367–386.
Watson, E.B. and Wark, D.A. (1997) Diffusion of dissolved SiO2 in H2O at 1 GPa, with implication for mass transport in the crust and upper mantle. Contributions to Mineralogy and Petrology, 130, 66–80.[CrossRef][ISI][GeoRef]
| |||||||||||||||||||||||||||||||||||||||
