Journal of Petrology Advance Access originally published online on December 10, 2007
Journal of Petrology 2008 49(2):225-238; doi:10.1093/petrology/egm078
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Experimental Investigation of the Upper Thermal Stability of Mg-rich Actinolite; Implications for Kiruna-Type Iron Deposits
Department of Geological Sciences and Environmental Studies, Binghamton University, Binghamton, NY 13902-6000, USA
RECEIVED JANUARY 9, 2007; ACCEPTED NOVEMBER 14, 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
The occurrence of actinolite in magnetite deposits of possible magmatic origin has prompted an experimental investigation of the upper thermal stability of Mg-rich actinolite to determine how the stability of actinolite changes with increasing Fe content. Experiments were carried out primarily on the compositional re-equilibration of natural tremolite [molar Fe/(Fe + Mg) = Fe-number = 0·014] in the presence of synthetic clinopyroxene (Ca0·80Fe0·67Mg0·54Si2·00O6), synthetic pigeonite/orthopyroxene (Ca0·08Fe1·19Mg0·70Si2·02O6), quartz, and water to a more Fe-rich actinolite over the range of 600–880°C, 1 and 4 kbar, at the Ni–NiO oxygen buffer, for durations of 1–2 weeks. The bulk composition of the mineral mixture is close to actinolite with Fe-number = 0·5. These experiments constitute a half-reversal of the amphibole composition, which, when approached from a Mg-rich starting composition, provides information on the minimum Fe content of actinolite at a given temperature. Compositional changes were monitored by electron microprobe analysis of amphibole rim compositions and/or overgrowths on the original tremolite. At 4 kbar and 880–800°C, tremolite shows strong re-equilibration with overgrowths of an Fe-rich but low-Ca (1·7 > Ca > 1·4) actinolite; Fe-rich cummingtonite (Ca <0·7) begins to nucleate at 860°C. At 800–700°C, tremolite shows weak compositional re-equilibration but strong nucleation of Fe-rich cummingtonite. Similar results were observed at 1 kbar, with tremolite showing strong re-equilibration to low-Ca actinolite at 790–600°C with cummingtonite nucleation at 800°C and below. The wide variation in Ca contents of the re-equilibrated amphiboles was unexpected. Additional univariant reversal experiments were carried out on the thermal decomposition of a natural actinolite (Fe-number = 0·22) from Pleito Melón, Chile, indicating the breakdown of actinolite to clinopyroxene, orthopyroxene, quartz, and water at 780°C and 1 kbar, and 850°C and 4 kbar. Considering only amphiboles with Ca >1·7 a.p.f.u., the thermal stability of actinolite is observed to decrease in a linear manner over the P–T range investigated with a dT/dFe-number slope of –372°C/Fe-number at 1 kbar and –546°C/Fe-number at 4 kbar. The high thermal stabilities (750–900°C) of actinolites with Fe-numbers in the range of 0–0·4 overlap with the range of water-saturated melting for a typical andesite or tonalite. These conditions also overlap the field of experimental Fe–P-rich melt formation, suggesting that actinolite may have an igneous origin in Kiruna-type ore deposits.
KEY WORDS: actinolite; mineral stability; Kiruna deposits, thermodynamic values; cummingonite
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Actinolite is one of the principal minerals of the greenschist facies and typically occurs in metamorphic rocks of mafic composition. Actinolite is also the most common silicate found in Kiruna-type ore deposits. It occurs in most Fe deposits from the Chilean Cretaceous Iron Belt and is particularly abundant in Sierra La Bandera, Chile. Actinolite has been found in the Chilean Cretaceous Iron Belt as 10 cm long tabular and fibrous crystals in pegmatitic dikes (Fig. 1a) composed almost entirely of actinolite and magnetite with minor apatite (Fig. 1b), as phenocrysts in magnetite-lavas and pyroclastic deposits, as orbicules in magnetite ore (Fig. 2a and b), and also as fibrous crystals in hydrothermal veins. Based on the textures mentioned above, at least part of the actinolite appears to be magmatic in origin.
|
|
The origin of Kiruna-type deposits is still a matter of debate. The massive structure and the textural resemblance to lava flows, dikes, and pyroclastic deposits led Geijer (1931
) and Park (1961) to postulate a magmatic origin for these deposits in Kirunavaara, Sweden, and El Laco, Chile, respectively. They suggested that Kiruna-type deposits were formed by the eruption or intrusion of an unusual Fe-rich magma. Because of the spectacular exposure and lack of overprinting events, most of the discussion about these deposits has centered on the exposures at El Laco, Chile. Some workers (Nyström & Henríquez, 1994; Broman et al., 1999; Naslund et al., 2002; Henríquez et al., 2003) have reported textural, fluid inclusion microthermometry (in pyroxene), and geochemical (V-rich magnetite) evidence that supports a magmatic origin, whereas others (Sheets et al., 1997; Rhodes & Oreskes, 1999; Rhodes et al., 1999; Sillitoe & Burrows, 2002, 2003) have presented fluid inclusion microthermometry (in apatite), geochemical (Ti-poor magnetite), isotopic, and rare earth element (REE) evidence that supports a hydrothermal origin, by the replacement of pre-existing volcanic rocks. The reader is directed to the references mentioned above for a more detailed discussion of the evidence supporting a magmatic vs hydrothermal origin for Kiruna-type deposits. The presence of actinolite in Kiruna-type deposits with textural evidence of a possible magmatic origin raises the following basic questions. (1) What is known about the stability of actinolite by itself? (2) Could it be of magmatic origin?
Experimental studies on the upper thermal stability of synthetic and natural tremolite (e.g. Jenkins & Clare, 1990; Jenkins et al., 1991; Welch & Pawley, 1991; Chernosky et al., 1998) and on the upper thermal stability of synthetic ferro-actinolite (Ernst, 1966; Hellner & Schürmann, 1966; Jenkins & Bozhilov, 2003) exist in the literature. Apart from the studies of Hellner & Schürmann (1966) on the tremolite–ferro-actinolite join and of Cameron (1975) on synthetic actinolite formed from an actinolite bulk composition with an atomic ratio of Fe/(Fe + Mg) (= Fe-number) of 0·5, no other experimental research exists on the thermal stability of actinolite of intermediate compositions. The most recent study bearing on the issue of the thermal stability of intermediate actinolite is that of Ghiorso & Evans (2002). Those workers thermodynamically modeled the stability–composition relationships of both orthorhombic and monoclinic amphiboles within the Ca–Mg–Fe2+ composition quadrilateral using a wide range of experimental and thermodynamic data, including data for the cummingtonite–grunerite join, anthophyllite–ferroanthophyllite join, cation-ordering in actinolite, tieline orientations of coexisting natural actinolite–cummingtonite assemblages from many localities, and the experimental studies on monoclinic amphiboles mentioned above. The study of Ghiorso & Evans (2002) is extremely valuable in that it models the extant experimental, thermodynamic, and natural paragenesis information for quadrilateral amphiboles with an internally consistent set of thermochemical data and thermodynamic–activity expressions involving intra-crystalline as well as inter-crystalline cation exchange reactions. It must be noted, however, that there still remains a dearth of experimental data on the stability–composition relationships of the intermediate actinolites against which the calculations of thermodynamic models can be tested or calibrated.
This study provides experimental data on the upper thermal stability of intermediate, Mg-rich actinolite by two independent approaches. First, half-reversals of actinolite composition were obtained via the divariant reaction
|
| (1) |
|
, bulk composition of the mixture; •, starting compositions of the actinolite (Act), low-Ca pyroxene (Opx/Pig), and clinopyroxene (Cpx). Arrows indicate the compositional re-equilibration of the amphibole expected at a higher (T1) and lower (T2) temperature at a given pressure and oxygen fugacity.Second, the univariant dehydration reaction of a natural actinolite from one of the Cretaceous iron-belt deposits in Chile (Pleito Melón) with an Fe-number of 0·22 was studied at similar P–T conditions. This reaction, involving Fe–Mg solid solution in the amphibole and pyroxene, is modeled by the component reaction
|
| (2) |
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Apparatus
The experiments were conducted in internally heated gas vessels using Ar as the pressure medium. Temperatures were recorded with dual Inconel-sheathed chromel–alumel thermocouples positioned near each end of the sample. The thermocouples were calibrated against the melting points of LiCl (605°C) and NaCl (800·5°C). Temperature uncertainty includes thermocouple accuracy (±2°C), controller variation (±1°C), and the thermal gradient across the sample, which was variable but typically less than 6°C (Table 1). Pressures were monitored with both a Bourdon-tube gauge and a factory-calibrated Harwood manganin cell and are considered accurate to ± 0·05 kbar.
|
The experiments were conducted at conditions of 1 and 4 kbar, between 600 and 880°C, and at the oxygen fugacity of the Ni–NiO buffer. This oxygen buffer was considered appropriate because it is slightly above the fayalite–magnetite–quartz (FMQ) buffer (about 1 logarithmic unit) and it has the advantage that it takes relatively small quantities of buffer to achieve a given hydrogen fugacity (low buffering capacity; Chou, 1987
), achieves equilibrium fast, and is commonly used in experimental research, which facilitates comparison of these results with other studies. In addition, according to Jenkins & Bozhilov (2003), the highest yields of synthetic ferro-actinolite were obtained using the Ni–NiO buffer, suggesting that this buffer may promote incorporation of the ferro-actinolite component in actinolite in these experiments. All of the experiments were performed using a conventional double-capsule technique (e.g. Chou & Cygan, 1990) to control the oxygen fugacity (via the hydrogen partial pressure) in the sample. The starting materials were sealed in Ag50Pd50 inner capsules (to minimize Fe absorption by the capsule) and the Ni–NiO oxygen buffer was sealed in a Pt outer capsule. Platinum was used for the outer capsule in these experiments rather than gold because of the high degree of alloying between gold and the inner AgPd capsule that occurred at the higher temperatures of this investigation. To minimize the rapid loss of hydrogen from the platinum outer capsule (Chou, 1986), ferrocene (FeC10H10) was placed in an open capsule inside the gas vessel next to the double capsule containing the sample to provide a source of hydrogen. As discussed by Driscall et al. (2005), ferrocene breaks down to a mixture of Fe, C, and hydrogen, with the last being released into the vessel to create a certain partial pressure of hydrogen. The hydrogen partial pressure was not measured directly but instead was calculated from the known mass of ferrocene placed inside the vessel, the mass of argon pumped into the vessel (calculated from the known volume occupied by gas inside the vessel and the P–V–T properties of argon; Robertson et al., 1969), and the fugacity coefficient of hydrogen at elevated P–T conditions (Shaw & Wones, 1964). Sufficient ferrocene was introduced inside the vessel to keep the partial pressure of hydrogen in the gas pressure medium near that of the Ni–NiO buffer.
Starting materials
Two different starting materials were used for the experimental investigation of divariant reaction (1). Most of the experiments were performed with a mixture of natural tremolite and synthetic pyroxenes. Natural tremolite was obtained from a calcite marble from Barrie Township, Ontario, Canada, with a chemical composition of K0·04Na0·16Ca1·89Fe0·07Mg4·93Al0·07Si7·89O22(OH,F0·2)2 (Jenkins, 1987). The tremolite was hand picked, avoiding the calcite, and then treated with cold 1·2M HCl to remove any adhering calcite. Synthetic clinopyroxene with a bulk composition Ca0·80Fe0·67Mg0·54Si2·00O6 was prepared from a mixture of reagent-grade MgO, CaCO3, Fe° (grain size of 2–6 µm), and SiO2. This mixture was decarbonated by heating in air at 900°C for 15 min and then treated hydrothermally with 10 wt % water in the gas vessel at 3·9 kbar and 950°C for 95 h. The run products were characterized by X-ray diffraction (XRD), which indicated an almost pure yield of clinopyroxene. The sample was ground and treated again at 950°C and 1·8 kbar for 72 h, giving a pure yield of well-crystallized clinopyroxene. Synthetic low-Ca pyroxene with a bulk composition Ca0·08Fe1·19Mg0·70Si2·02O6 was prepared in a similar way at 925°C and 2 kbar for 72 h; the run product was a mixture of cristobalite, olivine and orthopyroxene. The sample was treated again at 909°C and 2·4 kbar for 64·5 h; the run products were pigeonite, orthopyroxene, and trace amounts of olivine. An additional 2 wt % of SiO2 was mixed with these synthesis products to insure that the olivine reacted out, and the mixture was treated one more time at 894°C and 1·7 kbar for 162·5 h; the run products were pigeonite, orthopyroxene, and quartz. The presence of pigeonite along with orthopyroxene is a consequence of trying to maximize the Fe content of the orthopyroxene (Lindsley, 1983). Although not desired, the presence of pigeonite should not significantly affect the reaction because of the relatively small thermochemical difference (
5 kJ in enthalpy) between orthopyroxene and clinopyroxene of essentially the same bulk composition (e.g. Sack & Ghiorso, 1994). The bulk composition of the tremolite–pyroxenes mixture was approximately that of the actinolite composition Ca2Mg2·5Fe2·5Si8O22(OH)2.
A few experiments were also performed using a mixture of natural ferro-actinolite [Fe-number = 0·84, sample 94 of Verkouteren & Wylie (2000)] with synthetic fayalite, talc, and quartz with a bulk composition close to actinolite with Fe-number = 0·50 and Ca 1·0 a.p.f.u. This mixture was used to attempt to re-equilibrate an Fe-rich amphibole to an Fe-poorer composition and thereby define the maximum Fe content in amphibole. These experiments were possible only at the lowest temperatures investigated (600°C and 1 kbar) because of the inherent instability of ferro-actinolite at high temperatures (e.g. Jenkins & Bozhilov, 2003). The coexisting phases fayalite, quartz, and talc were chosen because they would be stable at these low temperatures and pressures and were compositionally equivalent to a mixture of ferrosilite and enstatite. Synthetic fayalite was prepared from a dry mixture of reagent-grade Fe°, Fe2O3, and SiO2 sealed under vacuum in a silica tube and placed for 3 days in a furnace at 900°C. Synthetic talc was prepared from a mixture of reagent-grade MgO and SiO2 and treated hydrothermally with 20 wt % water in the gas vessel at 690°C at 3·2 kbar for 1 day, then ground and treated again at the same conditions for 3 days.
Starting materials for univariant reaction (2) consisted of a natural actinolite and the breakdown products of this actinolite. The actinolite was obtained from an actinolite–magnetite dike, Pleito Melón, Chile (Aguirre, 2001), which was concentrated using a Franz magnetic separator and then hand picked to avoid any residual magnetite. The actinolite has the composition K0·01Na0·06Ca1·97Fe1·11 Mg3·86Al0·15Si7·87O22(OH)2 with an Fe-number of 0·22 (Lledo, 2005). The breakdown products of the natural actinolite (clinopyroxene, orthopyroxene, and quartz) were formed at 950°C and 4·1 kbar after 52 h. These were mixed in equal proportions by weight with the actinolite to make a well-seeded reversal mixture.
Unless otherwise indicated, all the experiments were conducted with the inner capsule containing 7·5 wt % water, and the external capsule (buffer) containing 15 wt % water.
Analytical methods
For electron microprobe analysis, samples were mounted in epoxy, polished, carbon coated, and analyzed with a JEOL 8900 superprobe at 15 kV accelerating voltage, beam current of 30 nA, and spot size less than 1 µm. The standards used were: hematite for FeO, corundum for Al2O3, Amelia Court House, Virginia, albite for Na2O and SiO2, orthoclase for K2O, and diopside for MgO and CaO.
Powder XRD analysis was carried out as follows. Prior to analysis, all samples were milled by hand in an agate mortar under ethanol. The powders were mounted on a zero-background oriented quartz plate and analyzed using a Philips PW3040-MPD automated diffractometer with Cu K
radiation, operated at 40 kV and 20 mA and fitted with a graphite diffracted-beam monochrometer. Most samples were analyzed with the full-pattern Rietveld structure refinement method on continuous scans over the range 8 to 50° 2
at increments of 0·02° 2 and counting times of 1 s per increment. The Rietveld refinements were calculated using the program DBWS-9807 by Young (1995), which provides an estimate of the proportions of the phases by weight.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Results for divariant reaction (1)
In the divariant experiments performed at 4 kbar, XRD analysis (Table 2) shows that tremolite abundance strongly decreases at 880°C. At 860°C tremolite re-equilibrates, forming overgrowths (or replacements) of an Fe-rich and Ca-poor amphibole (cummingtonite). At temperatures at or below 800°C microprobe analysis, back-scattered electron (BSE) images (Fig. 4), XRD analysis, and Rietveld refinement (Table 2) show growth of cummingtonite that increases with decreasing temperatures (700°C) (Fig. 5, Table 2).
|
|
|
The microprobe analyses of the amphiboles are listed in the Electronic Appendix (available for downloading at http://www.petrology.oxfordjournals.org) and are summarized in Figs 6 and 7. In experiments performed at 4 kbar (Fig. 6), there was a strong Fe enrichment in the amphibole composition but there was also a strong Ca depletion with decreasing T, which was unexpected. Similarly, for the experiments performed at 1 kbar, microprobe analysis (Fig. 7) shows that tremolite re-equilibrates to a more Fe-rich amphibole with decreasing temperature. A slight Ca depletion was detected by XRD and Rietveld refinement (Table 2), which indicates cummingtonite growth at temperatures of 800°C or below. The strong Ca re-equilibration in the amphibole was unexpected and problematic. To study the Fe variation as a function of temperature, independent of variations in the Ca content, only amphiboles with Ca contents between 1·7 and 2·0 a.p.f.u. (Fig. 8a and b) were included in the thermodynamic analysis. This range of Ca is more restrictive than the range specified by Leake et al. (1997
) for actinolite, but more representative of the majority of natural actinolites (e.g. Verkouteren & Wylie, 2000) and more consistent with the Ca contents deduced from Rietveld analyses (Table 2).
|
|
|
1·7 a.p.f.u. 
, compositional rims on tremolite seed crystals re-equilibrated in the presence of clinopyroxene and orthopyroxene according to divariant rection (1); 
, compositions of natural actinolite (Fe-number = 0·22) equilibrated in the presence of clinopyroxene, orthopyroxene, quartz, and water according to univariant reaction (2); Compositional re-equilibration of tremolite, occurring largely as amphibole overgrowths on original tremolite grains (Figs 4 and 5), typically produces a continuum of compositions that range from nearly Fe-free tremolite to Fe-rich actinolite or cummingtonite. As discussed by Jenkins (1994
) for the case of aluminous tremolite re-equilibration, it is reasonable to assume that the composition of amphibole closer to equilibrium is that corresponding to the highest Fe content attained in the overgrowths. The Fe contents of these overgrowths are shown in Fig. 8a and b as open triangles. Not surprisingly, the extent of compositional re-equilibration was greater at 4 kbar (Fig. 8a) than at 1 kbar (Fig. 8b) because of the enhanced thermal stability of amphibole at the higher pressure. Compositions that are very low in Fe (near the left edge of Fig. 8a and b) represent analyses of amphibole that experienced very little compositional re-equilibration; that is, they represent essentially the starting material tremolite. Compositions that move towards the middle of the diagram represent rims enriched in Fe. The arrows shown on the figures show the maximum Fe content achieved at a given temperature. The open squares in Fig. 8b show the compositional re-equilibration of natural ferro-actinolite with an initial Fe-number of 0·84 to somewhat lower Fe contents in the presence of talc, fayalite, and quartz treated at 600°C and 1·0 kbar for 295 h. The run products of this assemblage, based on XRD, are, in order of decreasing abundance: ferroactinolite, cummingtonite, olivine, talc and traces of magnetite. The filled triangles in Fig. 8 are microprobe analyses of amphiboles from the experiments performed on univariant reaction (2) (discussed below). In general, there is good agreement between the compositions of amphibole obtained by compositional re-equilibration of tremolite via reaction (1) and the upper thermal stability of natural actinolite via reaction (2).
Results for univariant reaction (2)
The results of the univariant reversals on reaction (2) are summarized in Table 3 and shown in Fig. 9. Reaction direction was estimated using a simple comparison of X-ray diffractometer peak heights, which was sufficient in view of the strong reaction observed crossing from one side of the reaction boundary to the other. The following X-ray reflections were found to be most useful for determining the reaction direction: the (101) of quartz, (
40) of actinolite, (221) of orthopyroxene, and (21) of clinopyroxene. Some reaction products (Act880,4; Act870,4; Act860,4) show subtle Mg enrichment with increasing temperature (Fig. 10a), just after crossing, or in the vicinity of, the stability boundary at 4 kbar. Additionally, some products (Act830,4) show Ca depletion with values of Ca <1·7 a.p.f.u. At 1 kbar, Mg enrichment was also detected for the amphiboles in samples Act780,1; Act800,1, and Act790,1 but without Ca depletion (Fig. 10b).
|
, pyroxene growth. Dashed curves show the calculated locations of isopleths of constant Fe-number in the actinolite (Act) coexisting with clinopyroxene (Cpx), orthopyroxene (Opx), quartz (Qtz), and water (V). The continuous curve is the isopleth for an actinolite of Fe-number 0·22.
|
) in average microprobe values are approximately the size of the symbol. The data points correspond to the average of the rim compositions in mole per cent calculated from data presented in the Electronic Appendix
|
Based on the results of XRD analysis as well as the decrease of Fe-number with increasing temperature detected by electron microprobe in some samples, it can be established that natural actinolite (Fe-number = 0·22) has an upper thermal stability of approximately 850°C at 4 kbar and 780°C at 1 kbar.
The results from this study indicate that the upper thermal stability of actinolite with Ca >1·7 a.p.f.u. and with Fe-numbers of 0–0·4 varies essentially linearly over the range 945–780°C at 4 kbar with a dT/dFe-number slope of –546°C/Fe-number and 860–720°C at 1 kbar with a dT/dFe-number slope of –372°C/Fe-number (Fig. 8a and b).
Thermodynamic modeling
In our thermodynamic treatment, we assume that the compositions of the amphiboles and pyroxenes are closely modeled by the system CaO–MgO–FeO–SiO2–H2O with minimal involvement of Fe3+. This assumption is supported by the minor amounts of Fe3+ (<10% of all Fe) in natural Al-, Na-, and Ti-poor actinolite (e.g. Clowe et al., 1988; Popp et al., 1995; Young et al., 1997; Evans & Yang, 1998) and in pyroxenes poor in Na (e.g. Robinson, 1980). In addition, use of the Ni–NiO buffer should at least minimize, if not prevent, the formation of Fe3+ in clinoamphibole (Clowe et al., 1988; Popp et al., 1995) and permit corrections to be made in this study for the presence of Fe3+ once this becomes known as a function of fO2.
The phases considered in this study are amphibole, orthopyroxene, clinopyroxene, quartz, and water. Amphibole was considered to exhibit mixing on two sites, involving ideal mixing of Fe and Mg on the five combined M1, M2, and M3 sites (M123) and non-ideal mixing of Ca, Mg, and Fe on the two M4 sites. This two-site simplification of cation mixing in actinolite is warranted in view of the nearly identical proportions of Fe and Mg at the M1, M2, and M3 sites (e.g. Evans & Yang, 1998). Orthopyroxene and clinopyroxene were considered to be two-site mixtures, with ideal mixing of Fe and Mg on the M1 site and non-ideal mixing of Ca, Mg, and Fe on the M2 site. Intersite mixing of Fe and Mg was modeled using the expressions of Driscall et al. (2005) for amphibole, Stimpfl et al. (1999) for orthopyroxene, and Brizi et al. (2000) for clinopyroxene as listed in Table 4. Non-ideal mixing on the M4 site of amphibole and on the M2 site of pyroxene were modeled as ternary regular solutions using the interaction (Margules) parameters listed in Table 4.
|
The boundaries shown in Figs 8 and 9 were calculated by finding the compositions of the phases that satisfied the Fe–Mg intersite mixing expressions of Table 4 for amphibole, orthopyroxene and clinopyroxene, along with the dehydration and intra-phase exchange reactions listed in Table 5. Allowing complete freedom of cation mixing on all sites generally made it difficult to achieve a stable numerical solution; however, by fixing the Ca content on the M4 site of amphibole to a specific value, such as 1·8 a.p.f.u., and setting Ca to zero on the M2 site of orthopyroxene, which is expected to be minor at the low temperatures of this study (e.g. Lindsley, 1983
; Ghiorso & Evans, 2002), but allowing Ca to vary in clinopyroxene, stable numerical solutions were readily achieved. With these restrictions, there were a total of seven independent compositional variables in three phases, which could be solved for by using the three Fe–Mg intersite exchange reactions in Table 4 and the four reactions listed in Table 5 and using the Newton method (Solver function of ExcelTM) for finding the solution to these non-linear equations. Thermodynamic data of Holland & Powell (1998) were used for all of the phases in Table 5 except for (1) the (298 K and 1 bar) values of 
and S° of tremolite, which were adjusted to –12311·0 kJ and 0·5515 kJ/K, respectively, to better fit the stability curve of natural tremolite reported by Jenkins & Clare (1990), and (2) the values of , S°, and V ° of ferro-actinolite, which were changed to –10500·0 kJ, 0·71924 kJ/K, and 28·42 kJ/kbar, respectively, based on the experimental work of Jenkins & Bozhilov (2003). Clinoenstatite (Mg2Si2O6) is not included in the database of Holland & Powell (1998); therefore, the , S°, and V ° of clinoenstatite were estimated as –3085·24 kJ/mol, 135·32 J/K mol, and 63·24 cm3/mol, respectively, by increasing the , S°, and V° of enstatite cited by Holland & Powell (1998) until the differences between clinoenstatite and enstatite matched the differences between these phases reported by Sack & Ghiorso (1994). The heat capacity, thermal expansivity, and compressibility data for clinoenstatite were assumed to be the same as those of enstatite.
|
Figure 8a and b shows the calculated values of Fe-number for coexisting amphibole (continuous curve), clinopyroxene (dash–dot curve), and orthopyroxene (dashed curve) at 4 and 1 kbar, respectively. The calculated curves for amphibole are a reasonably good fit to the observed Fe-number at 4 kbar (Fig. 8a), particularly above 800°C. At 750°C the calculated curve overestimates the observed Fe content, which may be the result of sluggish re-equilibration kinetics at this temperature. At 1 kbar (Fig. 8b), the calculated curve tends to underestimate the Fe-number at T > 750°C and overestimate the Fe-number at about 600°C. The effect of fixing the Ca content in actinolite to a specific value can also be calculated. At an Fe-number of 0·5, the location of the actinolite curve decreases by about 25°C as the Ca content increases from 1·6 to 1·9 a.p.f.u., which is about the level of uncertainty in the experimentally determined location of this curve.
Figure 11a and b shows the calculated locations of the T–Fe-number curves for coexisting actinolite, clinopyroxene, and orthopyroxene at 2 and 5 kbar, respectively. In Figure 11a, the boundaries at 2 kbar are compared with the results of experiments of Cameron (1975), which were performed at a fixed bulk Fe-number of 0·5 but with variable bulk Ca contents. Only experiments with bulk Ca contents of 1·6–2·0 a.p.f.u. are shown. Even though it was not possible for Cameron (1975) to define any phase boundaries by reversal experiments, there is general agreement between the location of actinolite-bearing (filled squares) and actinolite-absent (open circles and squares) experiments. It should be noted that cummingtonite is present in the lower temperature experiments of Cameron (1975) (filled squares and open circles in Fig. 11a).
|
, Cpx + Opx + Qtz. (b) Comparison of the calculated T–Fe-number boundaries at 5 kbar for coexisting actinolite, clinopyroxene, and orthopyroxene using the thermodynamic treatment of this study (continuous curves) compared with that used by Ghiorso & Evans (2002At both 2 and 5 kbar clinopyroxene and orthopyroxene are calculated to be more enriched in Fe than the coexisting amphibole, with orthopyroxene being much more iron rich than either amphibole or clinopyroxene (Fig. 11). This sequence of Fe enrichment (i.e. amphibole < clinopyroxene << orthopyroxene) is the same as that predicted by Ghiorso & Evans (2002
) at 5 kbar (Fig. 11b, dashed curves). As indicated in Fig. 11b, our calculated boundaries converge towards the Fe-free axis at a temperature nearly 45°C higher than that predicted by Ghiorso & Evans (2002). This is primarily caused by the differences in the and S° for tremolite, which, as mentioned above, were chosen in this study to match the breakdown boundary observed by Jenkins & Clare (1990) for the same natural tremolite as used in the starting material in this study. Apart from this difference, the two sets of calculated curves show nearly identical compositions for the coexisting phases, particularly at the lower temperature limit of about 825°C, where cummingtonite was predicted to appear by Ghiorso & Evans (2002). |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Igneous origin of Mg-rich actinolite
The results from this study indicate that the upper thermal stability of actinolite with Ca >1·7 a.p.f.u. and Fe-numbers of 0–0·4 varies essentially linearly over the range of conditions investigated. As a result, most natural actinolites, which generally have Fe-numbers in the range of 0–0·4 (Verkouteren & Wylie, 2000
), are sufficiently stable to exist well above the water-saturated andesite solidus of Stern et al. (1975), as shown in Fig. 12. We would like to stress, however, that the simple comparison of dehydration maxima with solidus and liquidus boundaries in Fig. 12 neglects the important differences in bulk composition between andesite and the actinolite–pyroxene assemblage studied here. The presence of 15–18 wt % Al2O3 and 2–3 wt % Na2O in typical andesites will promote the formation of amphiboles that are pargasite, magnesiohornblende or edenite (i.e. ‘hornblende’) in composition (Carroll & Wyllie, 1990; Beard & Lofgren, 1991) rather than amphiboles belonging to the tremolite–ferro-actinolite join. Indeed, most of the field occurrences of actinolite in andesitic or dioritic rocks have textures suggesting that the actinolite is a subsolidus conversion of pre-existing pyroxene or hornblende at greenschist-facies conditions (e.g. Johnson et al., 1989; Ernst, 2002). However, based on the textures observed in the field (Figs 1 and 2), the relatively high thermal stability of actinolite demonstrated in this study, and the Al-poor, Fe–P-rich melts experimentally produced from andesitic magmas (Lledo, 2005), it is suggested that actinolite could precipitate as an igneous phase in Kiruna-type ore deposits.
|
Role of cummingtonite
The strong tendency for cummingtonite to nucleate in our experiments (Figs 4 and 5) and the nearly continuous range of Ca contents observed here for amphibole (Figs 6 and 7) were somewhat surprising. Cameron (1975
) showed a miscibility gap for an intermediate actinolite composition (Fe-number = 0·50) with cummingtonite having 10 mol% of actinolite component (0·2 a.p.f.u. of Ca) in solid solution, coexisting with actinolite having 85 mol% of actinolite component (1·7 a.p.f.u. of Ca) at 650°C and 2 kbar. The results of the microprobe analysis reported by Cameron (1975) show a broader range of solid solutions than was apparent from their X-ray data, probably indicating intergrowth of actinolite and cummingtonite. The same problem may exist in the set of experiments reported in the present work, although the data shown in Fig. 8a and b exclude amphiboles with less than 1·7 Ca a.p.f.u. in an attempt to minimize microprobe analyses biased by intergrown cummingtonite. The data from this study suggest that lowering the Ca content in actinolite expands the thermal stability field of actinolite and cummingtonite. There is noticeable closure of the miscibility gap between actinolite and cummingtonite at 4 kbar at higher temperatures that could be metastable; further study is required to confirm this.
According to Ghiorso & Evans (2002), the reaction
|
| (3) |
) would predict in their thermodynamic calculations, except that the positions of actinolite and clinopyroxene in reaction (3) are in the reverse order. In fact, the results of this study show strong cummingtonite nucleation in all samples below 800°C (at 4 kbar) and below 700°C (at 1 kbar), suggesting that the reaction
|
| (4) |
), reaching a maximum temperature of about 840°C at 5 kbar with an Fe-number = 0·3 when coexisting with actinolite and orthopyroxene (Ghiorso & Evans, 2002). This thermal maximum is essentially in the middle of the compositional range of the amphibole re-equilibration experiments described here (Fe-number of 0–0·5) and may explain why cummingtonite nucleates so frequently in these experiments. An interesting possibility that arises from this study is that the conditions leading to the igneous origin of actinolite may also lead to the igneous origin of cummingtonite in Kiruna-type ore deposits. |
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Compositional half-reversals starting from nearly Fe-free tremolite in the presence of Fe-rich clinopyroxene and orthopyroxene were feasible down to temperatures of about 750°C. Additional compositional re-equilibration of a natural actinolite with moderate Fe content (Fe-number = 0·22) in the presence of orthopyroxene, clinopyroxene, and quartz, and one with high Fe content (Fe-number = 0·84) in the presence of fayalite, talc, and quartz to lower Fe contents provide some experimental constraints on the maximum Fe content of actinolite. Experimental results indicate a nearly linear decrease in thermal stability with an average dT/dFe-number slope of about –400°C/Fe-number [or –4°C/(mol% Fe)] from 1 to 4 kbar. Thermal breakdown of the natural actinolite with Fe-number = 0·22 to orthopyroxene, clinopyroxene, quartz, and water was found to occur at 780°C at 1 kbar and at 850°C at 4 kbar. These results provide a direct determination of the thermal stability of this actinolite.
Spontaneous nucleation of cummingtonite was frequently observed, with extensive growth of cummingtonite occurring at 800°C or lower at 4 kbar and at 1 kbar. Microprobe analyses revealed that amphiboles with essentially a continuum of Ca contents existed (Figs 6 and 7), making the boundaries of any miscibility gap between actinolite and cummingtonite indistinct. It should be noted that cummingtonite nucleation was observed over a wide range of temperatures and may be stable up to the maximum temperatures investigated (880°C).
The high thermal stabilities (750–900°C) of actinolites with Fe-numbers in the range of 0–0·4 overlap with the range of water-saturated melting for a typical andesite or tonalite and of experimentally produced Fe–P-rich melt formation (Lledo, 2005), suggesting that an igneous origin of actinolite in Kiruna-type ore deposits is at least feasible.
|
SUPPLEMENTARY DATA |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Supplementary data for this paper are available at Journal of Petrology online.
|
ACKNOWLEDGEMENTS |
|---|
Special thanks go to Ramón Aguirre, who provided an actinolite sample that was used for the experimental research. Thanks go to Bill Blackburn for his help with the microprobe analysis, and to H. R. Naslund, A. Philpotts, and J. R. Graney for their comments on the PhD thesis that served as the basis for this work. We thank the National Science Foundation (grants NSF EAR-9909452 and NSF EAR-0228975) for financial support. This manuscript greatly benefited from critical reviews by three anonymous reviewers and Reto Gieré.
*Corresponding author. Present address: Department of Geoscience, University of Nevada, Las Vegas, Las Vegas, NV 89154-4010, USA. E-mail: haroldo.lledo{at}unlv.edu
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Aguirre R. Magmatic iron ore breccias in the Chilean Coastal Cordillera La Falda and Bronces Sur mines Pleito–Melon district, III Region, Chile. In: MA thesis (2001) Binghamton, NY: Binghamton University. 102.
Beard JS, Lofgren GE. Dehydration melting and water-saturated melting of basaltic and andesitic greenstones and amphibolites at 1, 3, and 6·9 kbar. Journal of Petrology (1991) 32:365–401.
Brizi E, Molin G, Zanazzi PF. Experimental study of intracrystalline Fe2+–Mg exchange in three augite crystals; effect of composition on geothermometric calibration. American Mineralogist (2000) 85:1375–1382.
Broman C, Nyström JL, Henríquez F, Elfman M. Fluid inclusions in magnetite–apatite ore from a cooling magmatic system at El Laco, Chile. GFF (1999) 121:253–267.[Web of Science]
Cameron KL. An experimental study of actinolite–cummingtonite phase relations with notes on the synthesis of Fe-rich anthophyllite. American Mineralogist (1975) 60:375–390.[Web of Science]
Carroll MR, Wyllie PJ. The system tonalite–H2O at 15 kbar and the genesis of calc-alkaline magmas. American Mineralogist (1990) 75:345–357.[Abstract]
Chernosky JV, Berman RG, Jenkins DM. The stability of tremolite; new experimental data and a thermodynamic assessment. American Mineralogist (1998) 83:726–739.[Abstract]
Chou IM. Permeability of precious metals to hydrogen at 2 kb total pressure and elevated temperatures. American Journal of Science (1986) 286:638–658.
Chou IM. Oxygen buffer and hydrogen sensor techniques at elevated pressures and temperatures. In: Hydrothermal Experimental Techniques—Ulmer GC, Barnes HL, eds. (1987) New York: John Wiley. 61–99.
Chou IM, Cygan GL. Quantitative redox control and measurement in hydrothermal experiments. In: Fluid–Mineral Interactions: A Tribute to H. P. Eugster. Geochemical Society, Special Publication—Spencer RJ, Chou I.-M, eds. (1990) 2:3–15.
Clowe CA, Popp RK, Fritz SJ. Experimental investigation of the effect of oxygen fugacity on ferric–ferrous ratios and unit-cell parameters of four natural clinoamphiboles. American Mineralogist (1988) 73:487–499.[Abstract]
Driscall J, Jenkins DM, Dyar MD, Bozhilov KN. Cation ordering in synthetic low-calcium actinolite. American Mineralogist (2005) 90:900–911.
Ernst WG. Synthesis and stability relations of ferrotremolite. American Journal of Science (1966) 264:37–65.[Abstract]
Ernst WG. Paragenesis and thermobarometry of Ca-amphiboles in the Barcroft granodioritic pluton, central White Mountains, eastern California. American Mineralogist (2002) 87:478–490.
Evans BW, Yang H. Fe–Mg order–disorder in tremolite–actinolite–ferro-actinolite at ambient and high temperature. American Mineralogist (1998) 83:458–475.[Abstract]
Geijer P. The iron ores of the Kiruna-type. Economic Geology (1931) 73:478–485.
Ghiorso MS, Evans BW. Thermodynamics of the amphiboles: Ca–Mg–Fe2+ quadrilateral. American Mineralogist (2002) 87:79–98.
Ghiorso MS, Evans BW, Hirschmann MM, Yang H. Thermodynamics of the amphiboles: Fe–Mg cummingtonite solid solutions. American Mineralogist (1995) 80:502–519.[Abstract]
Hellner E, Schürmann K. Stability of metamorphic amphiboles; the tremolite–ferroactinolite series. Journal of Geology (1966) 74:322–331.[Web of Science]
Henríquez F, Naslund HR, Nyström JO, Vivallo W, Aguirre R, Dobbs FM, Lledó H. New field evidence bearing on the origin of the El Laco magnetite deposit, northern Chile—a discussion. Economic Geology (2003) 98:1497–1500.
Holland TJB, Powell R. An internally consistent thermodynamic data set for phases of petrological interest. Journal of Metamorphic Geology (1998) 16:309–343.[CrossRef][Web of Science]
Jenkins DM. Synthesis and characterization of tremolite in the system H2O–CaO–MgO–SiO2. American Mineralogist (1987) 73:707–715.[Web of Science]
Jenkins DM. Experimental reversal of the aluminum content in tremolitic amphiboles in the system H2O–CaO–MgO–Al2O3–SiO2. American Journal of Science (1994) 294:593–620.
Jenkins DM, Clare AK. Comparison of the high-temperature and high-pressure stability limits of synthetic and natural tremolite. American Mineralogist (1990) 75:358–366.[Abstract]
Jenkins DM, Holland T, Clare AK. Experimental determination of the pressure–temperature stability field and thermochemical properties of synthetic tremolite. American Mineralogist (1991) 76:458–469.[Abstract]
Jenkins DM, Bozhilov KN. Stability and thermodynamic properties of ferro-actinolite; a re-investigation. American Journal of Science (2003) 303:723–752.
Johnson GI, Cooper JA, Blight DF. The geology and geochronology of a Proterozoic trachyandesite plug, Murchison Province, Yilgarn Block, Western Australia. Australian Journal of Earth Sciences (1989) 36:319–336.[CrossRef][Web of Science]
Leake BE, Woolley AR, Arps CES, et al. Nomenclature of amphiboles: Report of the subcommittee on amphiboles of the International Mineralogical Association, commission on new minerals and mineral names. American Mineralogist (1997) 82:1019–1037.[Abstract]
Lindsley DH. Pyroxene thermometry. American Mineralogist (1983) 68:477–493.[Abstract]
Lledo H. Experimental studies on the origin of iron deposits; and mineralization of Sierra La Bandera, Chile. In: PhD dissertation (2005) Binghamton, NY: Binghamton University. 271.
Naslund HR, Henríquez F, Nyström J, Vivallo W, Dobbs FM. Magmatic iron ores and associated mineralisation: Examples from the Chilean High Andes and Coastal Cordillera. In: Hydrothermal Iron Oxide Copper–Gold and Related Deposits: A Global Perspective, Volume 2—Porter TM, ed. (2002) Adelaide: PGC Publishing. 207–226.
Nyström JO, Henríquez F. Magmatic features of iron ores of the Kiruna type in Chile and Sweden: ore textures and magnetite geochemistry. Economic Geology (1994) 90:473–475.[Web of Science]
Park CF Jr. A magnetite ‘flow’ in northern Chile. Economic Geology (1961) 56:431–436.
Popp RK, Virgo D, Yoder H. S. Jr, Hoering TC, Phillips MW. An experimental study of phase equilibria and Fe oxy-component in kaersutitic amphibole: Implications for the fH2 and aH2O in the upper mantle. American Mineralogist (1995) 80:534–548.