Journal of Petrology Advance Access originally published online on October 22, 2004
Journal of Petrology 2005 46(2):255-273; doi:10.1093/petrology/egh070
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology vol. 46 issue 2 © Oxford University Press 2004; all rights reserved
Age, Origin and Cooling History of the Coronel João Sá Pluton, Bahia, Brazil
1 DEPARTMENT OF GEOLOGICAL SCIENCES, UNIVERSITY OF TEXAS, AUSTIN, TX 78712-0254, USA
2 NEG-LABISE, DEPT. GEOLOGÍA, UNIVERSIDADE FEDERAL DE PERNAMBUCO, CP 7582, 50670-000 RECIFE PE, BRAZIL
RECEIVED APRIL 2, 2002; ACCEPTED AUGUST 8, 2004
 |
ABSTRACT |
|---|
TOP
ABSTRACT
REGIONAL SETTINGIn north-east Brazil, Archean and Paleoproterozoic cratonic blocks are enclosed within a network of Brasiliano-age (0·7–0·55 Ga) metasedimentary foldbelts. The unfoliated Coronel João Sá granodiorite pluton, which contains magmatic epidote and strongly resorbed clinopyroxene, intrudes the Sergipano Foldbelt. Zircons yield a concordant U–Pb crystallization age of 625 ± 2 Ma; titanite ages are approximately 621 Ma. Discordant zircons suggest inheritance from at least two magma sources of ages <1·8 and >2·2 Ga. Model calculations based on diffusion parameters and Rb–Sr isotope data from separated minerals indicate that the pluton cooled at a rate of
36°C/Myr. Whole-rock element compositions and initial Sr–Nd isotopic compositions that are heterogeneous on all length scales suggest magma mixing. Trace-element concentrations and Nd isotope data argue against a contribution from a contemporaneous mantle-derived magma. Values of magmatic 
Nd (at 625 Ma) resemble contemporary Nd for local supracrustal rocks and basement, compatible with anatexis of a crustal source. In north-east Brazil, cratonic blocks could have amalgamated with foldbelts that originated as: (1) a mosaic of island arcs and arc basins (traditional allochthonous model), or as (2) extensional continental sedimentary basins (but not oceanic crust) later involved in collision (autochthonous model). The Coronel João Sá isotopic and chemical data support an autochthonous origin. KEY WORDS: Brasiliano Orogeny; granodiorite pluton; Rb–Sr isotopes, Sm–Nd isotopes; U–Pb isotopes, magma cooling rate
|
REGIONAL SETTING |
|---|
Western Gondwana was amalgamated between 0·7 and 0·55 billion years ago during the Pan African Orogeny, which is known as the Brasiliano Orogeny in South America. Cratonic blocks that accreted together to form Western Gondwana included the Congo–São Francisco Craton, the West African–São Luis Craton and the Amazon Craton (Fig. 1, inset). Today, the bedrock geology in north-east Brazil consists of Archean and Paleoproterozoic cratonic blocks, informally referred to as massifs (basement inliers), within a network of Pan African–Brasiliano age metasedimentary foldbelts that have been intruded by numerous syn- to post-tectonic granitoids (Sial, 1986
, 1987).
|
Isotope geochronology suggests that four major tectonothermal events affected north-east Brazil at
2·6 Ga (Archean), 2·2–2·1 Ga (Transamazonian), 1·1–0·95 Ga (Cariris Velhos) and 0·7–0·55 Ga (Pan African–Brasiliano). The massifs yield Archean and Transamazonian Rb–Sr and U–Pb ages. Zircons from pre-Brasiliano metavolcanic rocks and granitoids within the foldbelts yield 1·1–0·95 Ga Cariris Velhos U–Pb ages (Van Schmus et al., 1995).
Two contrasting models for the tectonic development of north-east Brazil have been proposed. The ‘allochthonous’ model proposes that during the Brasiliano Orogeny, unrelated terranes composed of cratonic blocks, island arcs and arc basins amalgamated together, becoming north-east Brazil (Caby & Arthaud, 1986). According to an alternative ‘autochthonous’ model, north-east Brazil developed from a continental basin-and-range-type terrane in which the foldbelts originated as extensional continental sedimentary basins. Van Schmus et al. (1995) suggested that extension occurred during the Cariris Velhos event. The rocks in these basins were folded later, during the Brasiliano collision.
The roughly triangular Sergipano Foldbelt is the southernmost foldbelt in north-east Brazil. The characteristics of the Sergipano Foldbelt, which wraps around the northern edge of the São Francisco Craton (Fig. 2), have been used as a basis to argue for both of the above-mentioned models (Dos Santos & Silva Filho, 1975; Brito Neves et al., 1977; Jardim de Sá, 1984; Campos Neto & Brito Neves, 1987; Davison & Dos Santos, 1989; Caby et al., 1991; Van Schmus et al., 1995). During the Brasiliano collision, rocks of the Sergipano Foldbelt were displaced internally by large-scale sinistral strike-slip movements while being thrust obliquely against the São Francisco Craton. These faults separate a succession of tectono-stratigraphic units, referred to as ‘domains’, that are progressively more deformed and metamorphosed from the south, to the north where the foldbelt is adjacent to the Pernambuco–Alagoas Massif (Davison & dos Santos, 1989) (Fig. 2). In the south, the protoliths consist of clastic and carbonate sediments, which persist toward the north but with increasing predominance of mafic and ultramafic igneous protoliths. In the Macururé domain, which is host to the Coronel João Sá pluton, the common protoliths are greywacke, siltstone and shale rhythmically bedded on a centimeter scale (turbidites?) (Davison & dos Santos, 1989).
|
Unfoliated (post-orogenic) granitoid bodies have intruded much of the Sergipano Foldbelt. The Coronel João Sá pluton (Fig. 3), named after a village situated upon it (37·93°W, 10·28°S), consists of a larger southern body, and a smaller northern body with the same characteristic lithology though with slight chemical and Sr–Nd isotopic differences. The host rock to the pluton is a brown, fine- to medium-grained biotite–garnet schist, in which staurolite is present in the contact aureole. To the west, the Coronel João Sá pluton is in fault contact with a graben filled by undeformed molassic Juá Conglomerate. The pluton is heavily sheared and altered with chlorite veining where a branch of the São Miguel do Aleixo fault zone cuts across its south-western part. Aplite dikes are present in the easternmost exposures of the pluton.
|
This study consists of a petrographic, geochemical and isotopic examination of the Coronel João Sá pluton. Its purpose is to characterize the source and cooling history of the magma, and to infer whether the Sergipano Foldbelt originated from a continental or oceanic basin.
|
PETROGRAPHY AND CLASSIFICATION |
|---|
The presence of small, hand-worked quarries facilitated the collection of fresh samples within an environment of tropical weathering. Chaves (1991)
, McReath et al. (1993) and McReath et al. (1998) have previously reported aspects of the geology of the Coronel João Sá pluton that are briefly summarized here. The pluton consists of granodiorite with minor granite (IUGS classification, Le Bas & Streckeisen, 1991). In decreasing abundance, the mineral assemblage is plagioclase (An20–42) + quartz + microcline (Or89–94) + biotite + hornblende + epidote + titanite ± clinopyroxene, with accessory zircon, allanite and apatite. Granodiorite in the easternmost part of the pluton is finer-grained than elsewhere and crosscut by numerous aplite dikes composed of quartz, low-An plagioclase (An4–24) and alkali feldspar, and minor mafic minerals (hornblende, biotite, epidote). Iron–titanium oxide minerals have not been observed within any of the facies of the pluton. Clinopyroxene (augite–salite) is present in the northern body and in the western 80% of the southern body, which passes eastward through a facies transition where clinopyroxene disappears and cm-sized poikilitic K-feldspar phenocrysts appear enclosing hornblende, biotite, titanite, plagioclase and quartz. Clinopyroxene is severely resorbed, as evidenced by former euhedral laths up to 1·5 cm long that have been overgrown or replaced by biotite, hornblende and epidote, leaving only scattered remnant grains of clinopyroxene exhibiting optical continuity (Fig. 4).
|
Frost et al. (2001)
proposed a geochemical classification of granitoids based upon Fe* = [FeOtot/(FeOtot + MgO)], a modified alkali–lime index (MALI = Na2O + K2O – CaO) and an aluminum saturation index [ASI = Al/(Ca – 1·67P + Na + K)] (molecular). According to this classification, the Coronel João Sá granitoid is magnesian (Fe*), calc-alkalic (MALI), with ASI = 1·03 ± 0·03. This rock resembles the ‘Cordilleran’ granitoids, the dominant type on Earth.
Green amphibole-rich inclusions, or ‘clots’ (poikiloblastic actinolite to magnesio-hornblende) up to 5 cm across, are ubiquitous in all facies of the granodiorite. They are mantled by biotite, suggesting that biotite was a product of reaction between the clots and surrounding magma. Locally abundant microgranular tonalitic enclaves are of variable size (centimeters to meters), ovoid and commonly in parallel alignment. Higher abundances of mafic minerals in the interior of the enclaves merge into lower abundances in the marginal zones. Enclaves are more mafic than the host granodiorite and contain less K-feldspar, although both enclave and host contain the same minerals, with similar grain sizes, textures and chemical compositions. Clinopyroxene is absent in the enclaves from the eastern facies, just as it is absent in the host rock, though it may have been formerly present as indicated by textures of replacement hornblende and biotite. Patchy zoning indicates relict An-rich cores in plagioclase. The acicular texture of apatite in the tonalitic enclaves suggests rapid quenching, although the enclaves do not have chilled rims. Enclaves within granitoid plutons typically are attributed to injection of hot mafic magma into cooler, more viscous felsic magma, followed by physical (mingling) and chemical interactions (e.g. Poli & Tommasini, 1991). Sial et al. (1998) described the mafic inclusions in the granitoids of north-east Brazil in more detail.
|
PETROLOGICAL INTERPRETATION: CONSTRAINTS ON INTENSIVE PARAMETERS |
|---|
Abundant titanite in the granitoid (approximately 1% by volume) displays a euhedral crystal form indicating a magmatic rather than a metamorphic origin. Plagioclase is fresh, suggesting the absence of weathering or low-temperature (subsolidus) alteration. The textural and chemical features of epidote, such as the presence of allanite cores, euhedral crystal habit against biotite but vermicular texture against plagioclase and quartz, resorption replacement of hornblende and pistacite contents [Fe3+/(Fe3+ + Al3+)] from 0·18 to 0·25, indicate a magmatic origin for this mineral (Zen & Hammarstrom, 1984
; Schmidt & Thompson, 1996). Magmatic epidote occurs commonly throughout Brasiliano-age plutons in north-east Brazil and was described in detail by Sial et al. (1999).
The absence of oxide minerals might be explained by an unusually low fO2 and correspondingly low Fe3+/(Fe3+ + Al3+), yet the presence of magmatic epidote requires a significant Fe3+ component in the melt. In the study of Schmidt & Thompson (1996) on the stability of epidote in tonalite and granodiorite magmas, oxide phases were present only in experiments in which epidote was absent from the mineral assemblage. Those workers cited field observations that magnetite is considerably more abundant in epidote-absent than in epidote-bearing granitoids, and concluded that magnetite is the main Fe3+-bearing phase at temperatures above epidote stability, whereas Fe3+ tends to enter epidote at lower temperatures. In the Coronel João Sá pluton, epidote and titanite accommodate Fe3+ and Ti4+, respectively.
Quantification of fO2 in the Coronel João Sá magma is difficult in the absence of oxide minerals. Liou (1973) found that the presence of magmatic epidote requires fO2 > QFM (quartz–fayalite–magnetite buffer) and the pistacite content of the epidote suggests that fO2 was buffered between QFM and NNO (nickel–nickel oxide) oxygen buffers. Values of Fe/(Fe + Mg) in biotite and hornblende (0·5) also suggest fO2 in the QFM to NNO range (Czamanske & Wones, 1973).
Unfortunately, no geothermometer or geobarometer can be applied to the Coronel João Sá pluton that utilizes magnetite, ilmenite, garnet, orthopyroxene, fayalite or muscovite, as all of these minerals are absent. The absence of an equilibrium Fe–Ti oxide phase precludes the quantitative application of the aluminum-in-hornblende geobarometer proposed by Hammarstrom & Zen (1986). The concentration of Al in hornblende increases systematically from west to east across the pluton, suggesting qualitatively that paleopressure increased in that direction (Fig. 5). However, both the absence of an Fe–Ti oxide mineral in the assemblage and the low abundance of Fe3+ in hornblende may make this apparent trend spurious. Calculations of Fe3+ and Fe2+ in hornblende by the procedure recommended by Cosca et al. (1991) and Anderson & Smith (1995) indicate very low values of Fe3+/(Fe3+ + Fe2+), from 0·18 to as low as 0·02. Anderson & Smith (1995) showed that in this situation, Al-in-hornblende pressures are much higher than pressures independently estimated, the Fe3+ deficiency being balanced by excess Al3+ in the Fe3+–(6)Al3+ exchange vector, thus increasing the apparent paleopressure.
|
We may estimate some general constraints by combining a thermometer, such as hornblende–plagioclase, with information implied by the presence of epidote and absence of oxide minerals. Figure 6, a phase diagram from Schmidt & Thompson (1996)
for water-saturated granodiorite magma with NNO buffering of fO2, exhibits a field of epidote stability. At lower fO2, such as with QFM buffering, the epidote stability field shrinks to higher pressures and lower temperatures. Schmidt & Thompson (1996) performed experiments over a range of high pressures (7–18 kbar). Consequently, the low-pressure intersection of the epidote-out boundary with the granodiorite solidus can be estimated only by extrapolation, and this value of about 4 kbar may provide a reasonable minimum pressure for the Coronel João Sá magma. The presence of staurolite in the contact aureole does not constrain the pressure estimate because staurolite is stable over a wide range of pressure (Richardson, 1968).
|
Superimposed upon Fig. 6 are temperature calculations using the Holland & Blundy (1994)
hornblende–plagioclase geothermometers A and B, which have opposing slopes in P–T space. Holland & Blundy cautioned not to attribute pressure significance to the intersections of ‘A’ lines with ‘B’ lines, as the coefficients in the calculations are not optimized for this purpose. All of the lines lie well inside the epidote stability field; i.e. the P–T estimates based upon data from hornblende, plagioclase and stability of epidote are self-consistent.
The presence of clinopyroxene within the granitoids suggests a geobarometer based upon equilibrium exchange with the aluminum-bearing ‘Tschermak's molecule’ component of pyroxene:
 | (1) |
have summarized data from experiments and natural rocks to recalibrate the geobarometer and evaluate its potential. Table 1, which provides a representative chemical analysis of plagioclase and clinopyroxene (sample CJS 20), illustrates the problem of applying the SCAn geobarometer to Coronel João Sá rocks. Analytical uncertainties in the electron microprobe analyses (several percent for each oxide, Table 1) must also be taken into account in calculating the activity (aCaTs). For this sample, aCaTs = 0·0004 ± 0·0009, which corresponds to an uncertainty of ±5 kbar in calculated paleopressure. Microprobe data for samples CJS 09, CJS 13 and CJS 25 likewise fail to provide meaningful calculations of pressure.
|
|
MAJOR- AND TRACE-ELEMENT GEOCHEMISTRY |
|---|
Geología e Sondagens Ltda (GEOSOL) in Belo Horizonte, Minas Gerais, made whole-rock chemical analyses of 15 samples and the GeoAnalytical Lab, Washington State University, Pullman, WA analyzed 10 samples, including five that are duplicates. Major-element data from the two laboratories are very similar. Table 2 provides a tabulation of all chemical data.
|
Plots of selected major- and trace-element concentrations vs SiO2 (Harker diagrams, Fig. 7) exhibit distinct negative trends with large negative values of the correlation coefficient, r. These trends are consistent with processes of crystal fractionation, magma mixing or inhomogeneous distribution of accessory restite minerals. In Coronel João Sá granitoids, Ti is sequestered in titanite, P is sequestered in apatite, Mg in mafic minerals, Ba in microcline, and Zr and Hf in zircon. Plots (not shown) of these elements vs CaO or FeOtotal (used as indices of differentiation) are also highly correlated.
|
In variation diagrams in which whole-rock abundances of selected trace elements (Fig. 8a) or rare earth elements (Fig. 8b) are normalized to primitive mantle, the data form closely parallel patterns. Moreover, the concentration of any given trace element is highest in the western facies of the southern body, which has generally the lowest wt % SiO2; the concentration is lower in the eastern facies (intermediate SiO2) and lowest in rocks of the northern body, which are geochemically the most highly evolved (highest SiO2). Negative Eu anomalies (Eu/Eu*
0·9) are small.
|
|
ISOTOPE GEOCHEMISTRY |
|---|
We analyzed a representative suite of whole-rock samples by the U–Pb, Sm–Nd and Rb–Sr methods. The samples, collected with wide spatial coverage (Fig. 3), are mostly granodiorite but also include two tonalitic enclaves, two amphibole-rich xenoliths and one aplite dike. Fresh rock in a working quarry centrally located in the pluton (Fig. 3) provided sample CJS 20. Whole-rock and purified concentrates of zircon, titanite, plagioclase, K-feldspar, apatite and biotite from this sample were analyzed isotopically.
U–Pb
Zircon crystals from sample CJS 20 were hand-picked according to morphology—whether common prismatic forms (Fig. 9) or rare needle-like or rounded grains. Cores are present in nearly all zircon grains, from which small populations having a minimum of visible inherited material were selected. Euhedral titanite crystals were selected on the basis of greatest clarity and absence of inclusions. Both minerals were air-abraded until crystal faces were no longer visible, then dissolved, spiked with 205Pb and 235U, and analyzed as described by Connelly et al. (1996) using a Finnigan MAT 261 multicollector mass spectrometer at the University of Texas at Austin.
|
The U–Pb and Pb–Pb ages of zircon fraction Z5 (Table 3, Fig. 10) are highly concordant: t206/238 =t207/235 = t207/206 = 625 Ma (±2 Ma including 2
analytical uncertainty). Discordia lines projected from concordant point Z5 through discordant points Z1, Z2 and Z4 intersect concordia at 1·67 ± 0·25, 2·23 ± 0·03 and 1·76 ± 0·03 Ga, respectively (Fig. 10). The latter samples must also contain radiogenic Pb inherited from older sources. Moreover, they are multi-grain samples that almost certainly represent mixtures of these ancient sources, causing the apparent ages to be partially homogenized. Therefore, 1·67 ± 0·25 Ga is a maximum estimate of the age of the younger source(s) and 2·23 ± 0·03 Ga is a minimum age estimate for the older source(s).
|
|
Data for titanite samples T1 and T2 plot 1·0 and 1·2% discordantly, but their error ellipses overlap concordia (Fig. 10). A weighted average of T1 and T2 data gives t206/238 = 620·9 ± 1·4, t207/235 = 622·7 ± 2·0 Ma. In consideration of the 2
error ellipses, there is a 95% probability that the ages of titanite and zircon Z5 are distinctly different. The closure temperature for the U–Pb system in zircon is >800°C, and for titanite it is estimated at 660–700°C. We interpret that the pluton crystallized at 625 Ma (concordant zircon age), followed by slow cooling to the titanite closure temperature at 621 Ma. The Coronel João Sá pluton is not foliated or recrystallized, and it cross-cuts host rock whose Brasiliano age of deformation and metamorphism is latest Proterozoic. Isotopic inheritance, not later geochemical disturbance, was responsible for the pattern of apparent ages. The much more ancient sources themselves must lie at depth beneath the pluton, or the Coronel João Sá magma was contaminated by zircon from the local host rock that was inherited from these more ancient sources.
Sm–Nd and Rb–Sr
Our whole-rock Rb–Sr isotope data (Table 4) are generally compatible with data previously reported by Santos et al. (1988) and McReath et al. (1998). A Rb–Sr isochron plot (Table 4, Fig. 11) or Sm–Nd (Table 5, isochron plot not shown) have large values of the MSWD statistical parameter, i.e. the scatter of points about a best-fit line exceeds the scatter as a result of analytical error alone. Limiting the isochron calculation to a subset of the data—e.g. only data from granodiorite or only from the western facies of the southern body—does not improve the situation. Scatter is evident in Fig. 12, showing Sr and Nd initial ratios back-calculated to the 625 Ma concordant U–Pb age of zircon. The scatter is not explained by a later geochemical disturbance, for which there is no field or petrographic evidence. Geochemical disturbance might have affected the chemically mobile elements Rb and Sr, but probably not the immobile, geochemically coherent Sm and Nd (e.g. see Barovich & Patchett, 1992).
|
|
|
|
Our favored explanation is that the scatter of data at Coronel João Sá was an initial condition. Aplites and granodiorites from the western facies and eastern facies were sampled from a single continuous pluton. A plot of initial 87Sr/86Sr vs
Nd at 625 Ma suggests a mixing relationship for these rocks (Fig. 13). All of the initial ratios reflect a continental crust signature (negative Nd, elevated 87Sr/86Sr). Initial Sr ratios also correlate positively and initial Nd ratios correlate negatively with SiO2. Crystal fractionation acting alone cannot account for the range of variation in initial Sr and Nd isotopic compositions. More probably, several sources contributed isotopically distinct magmas that became the northern body and the western and eastern facies of the southern body. Sr–Nd isotope data from these bodies are clustered (Fig. 12). In particular, initial Sr ratios in the northern body were consistently high (Fig. 12a).
|
At least two magma types are represented by tonalitic enclaves and by granodiorite that is itself inhomogeneous. Magmas in intimate contact may mingle, but differences in viscosity and liquidus temperatures can be an effective barrier to mixing. Enclaves form when the hotter, more mafic magma makes up less than 30–40% of the mass (as at Coronal João Sá), whereas mingling or mixing occur with higher abundances of mafic magma (Frost & Mahood, 1987
).
End-members of mixing curves such as Fig. 13a are commonly modeled as ‘mafic’ and ‘felsic’ components. The aplite composition serves as a plausible crustal end-member, but the chemistry of the more mafic end-member does not resemble the composition of a mantle-derived mafic magma. As continental crust is differentiated from mantle, the crust is enriched in silica; it is also enriched in Rb preferentially to Sr, and in Nd preferentially to Sm, which causes negative values of Nd to develop. This process, which pertains to the source rocks of Coronel João Sá granitoids, is expected to produce a negative correlation of SiO2 vs Nd and a positive correlation of SiO2 vs 87Sr/86Sr at the time of emplacement, 625 Myr ago, as observed in Fig. 13b and c. Differentiation would also relatively enrich the crust in large-ion lithophile elements (Ba, Sr) and high field strength elements (Zr, Hf). Plots of these elements vs Nd should produce negative correlations, but this is contrary to the positive correlations seen in Fig. 14. Arrows point toward estimated Nd (+10) and concentrations of trace elements in continental lithospheric mantle (CLM, Fig. 14). Compositions of postulated CLM end-members are distant from Coronel João Sá compositions and they lie off the data trends. Thus, the immediate sources of Coronel João Sá magma were not directly related to the mantle but must already have experienced multiple stages of development.
|
|
Rb–Sr SYSTEMATICS IN MINERALS |
|---|
Table 6 presents Rb–Sr isotope data for titanite, plagioclase, K-feldspar, apatite and biotite separated from sample CJS 20. In an isochron diagram (Fig. 11), the datum for highly radiogenic biotite controls the slope of the isochron. The slope is consistent with an age of about 610 Ma but the data points scatter substantially (MSWD = 9·4). This is to be expected; an isochron assumes an ‘instantaneous’ magmatic event, whereas the Rb–Sr systematics of these plutonic minerals developed over a prolonged interval of slow cooling. The following discussion uses petrographic, chemical and isotopic data as a basis to calculate the cooling rate of the Coronel João Sá pluton.
|
Could the individual mineral phases analysed from CJS 20 have inherited non-uniform initial Sr isotope compositions? We have already noted initial 87Sr/86Sr in whole rocks that is variable over a scale of meters to kilometers. Initial Sr isotope compositions in igneous rocks can be inhomogeneous on the scale of a hand sample, or even on a micrometer scale in zoned crystals of volcanic plagioclase (Davidson et al., 2001
). However, diffusion rapidly levels any initial variability when magmatic temperatures are sustained. A characteristic diffusion distance is expressed by x = (Dt)0·5, where x is distance, D is the diffusion coefficient (distance2/time) and t is time (Crank, 1975). According to the data in Table 6, Sr would diffuse out of a typical grain of plagioclase within a few thousand years at an assumed temperature of 800°C. This time-scale is orders of magnitude shorter than the millions of years' cooling period for a deeply buried pluton. Once the diffusing Sr reaches the grain margin, grain-boundary diffusion transports it extremely rapidly throughout the local volume of rock (Jenkin et al., 1995).
As the temperature drops through the closure temperature for a mineral species, that mineral begins to retain radiogenic Sr. Dodson (1973) derived a mathematical relationship between the closure temperature and the diffusion coefficient, activation energy, effective diffusion radius of the crystal (assumed to be the physical grain size) and rate of cooling (closure temperature being higher if cooling is more rapid). Once a given mineral has become a closed system, only the remaining mineral assemblage continues to communicate by diffusion until the closure temperature of the next mineral species is reached, and so on (Giletti, 1991). Finally, after the ‘residual whole rock’ has been reduced to only two participating minerals, communication ceases when the mineral with the higher closure temperature becomes closed, regardless of the intrinsic diffusion properties of the other mineral. We have applied this analysis to sample CJS 20, subject to the assumptions and qualifications provided in the Appendix.
The objective of modeling is to devise a time–temperature history that is consistent with the present-day 87Sr/86Sr in each mineral, and with other isotopic and petrographic data. Following Giletti (1991, appendix), we assume a cooling rate and a starting temperature, which Giletti equated with the solidus. During cooling, 87Sr/86Sr increases uniformly and homogeneously throughout the rock by rapid diffusion until the first mineral closes, following which the Sr isotope ratio in that mineral continues to increase until the present day at a rate proportional to its Rb/Sr. We recalculate Rb/Sr and 87Sr/86Sr in the mineral assemblage in the residual whole rock prevailing at the time of first mineral closure, and 87Sr/86Sr continues to increase homogeneously until the second mineral closes, and so forth. Each mineral species became a closed system at a different time, and with different initial 87Sr/86Sr. The cooling rate influences the Dodson (calculated) closure temperature (Tc) and also the intervals between closure ages—the slower the cooling, the greater the differences between them. The starting temperature determines how far the temperature must drop (i.e. how much time passes) before the first mineral closes; the value of the starting temperature adjusts the range of closure ages up or down in an absolute sense. The interval of cooling, even at a slow rate, is likely to be brief compared with the age of the Coronel João Sá pluton.
We used an interactive spreadsheet approach to investigate combinations of starting temperatures between 650 and 900°C, with cooling rates between 20 and 100°C/Myr, to determine a statistical best fit between calculated 87Sr/86Srtoday vs 87Sr/86Sr as actually measured. Unfortunately, the calculations are compatible with a range, not a single unique combination of these variables. Largely this is because the growth of 87Sr/86Sr in titanite, apatite and plagioclase (low Rb/Sr minerals) is so insensitive to the passage of time that the data fit various cooling scenarios almost equally well.
We can use the U–Pb data (Table 3) indicating closure of titanite at 621 Ma and a titanite Tc of 660–700°C (Scott & St-Onge, 1995) to further limit the optimum combination of cooling parameters. Taken together, the Rb–Sr and U–Pb data provide an excellent fit with a starting temperature of 840°C, and a cooling rate of 36°C/Myr. Under these conditions, about 4 Myr (from 625 to 621 Ma) would separate the closure of the U–Pb systems in zircon and titanite. The Rb–Sr system in titanite closed at about the same time and temperature. Rb–Sr data, which are lacking for epidote, hornblende and clinopyroxene, would have resembled the inconclusive data for titanite, apatite and plagioclase, and would not have changed the interpretation. Table 7 summarizes the calculated cooling history.
|
Note that the inferred starting temperature (840°C) is well above the solidus. Naney (1983)
investigated the mineral phase relationships of a synthetic granodiorite composition at pressures of 2 and 8 kbar. By analogy, the solidus of the Coronel João Sá granodiorite, which is of similar bulk-rock composition, would have been approximately 650°C at 4 kbar. At 840°C and 4 kbar, sample CJS 20 would have consisted of plagioclase, hornblende, biotite, quartz, titanite(?), possibly K-feldspar and pyroxene (depending upon wt % H2O), and melt. The Giletti procedure for calculating successive closure temperatures does not require that the whole rock is entirely solid, but only that it is a closed system. Minerals that closed late during cooling may not even have existed when early-crystallized minerals became closed.
Qualitative estimates of paleopressure are of the order of 4–5 kbar (12–15 km paleodepth). We do not know the temperature–depth regime that prevailed at the end of the Brasiliano Orogeny, nor the three-dimensional configuration of the Coronel João Sá pluton, nor where the sampled rocks lay within it. Model calculations using the heat conduction equation (Crank, 1975) are roughly consistent with a cooling rate of 36°C/Myr at a burial depth of 15 km.
|
SOURCES OF THE CORONEL JOÃO SÁ MAGMA |
|---|
Modeling of the Sm–Nd isotope data provides insight into possible magma sources. Figure 15 displays whole-rock
Nd back-calculated to 625 Ma. Values of 625 are –7 to –5, which could have been inherited by the magma from a single (somewhat inhomogeneous) source, or averaged from a mixture of sources with variable Nd. A plausible single-source model consists of two stages. In the first stage, the source becomes separated from depleted mantle (Fig. 15), following which Nd develops with f = –0·4, which is an average value measured in continental crust. (f is [(147Sm/144Ndsource/147Sm/144NdCHUR) – 1].) At 625 Ma, partial melting creates the Coronel João Sá magma with fractionated (somewhat diminished) Sm/Nd. The gray band (Fig. 15) encloses the stage 1 growth lines and it intersects the depleted mantle curve at a tightly constrained 2·4 ± 0·1 Ga. This age is in good agreement with the 2·5 Ga Rb–Sr age of gneissic basement to the Sergipano Foldbelt, which is exposed in several localities such as the Itabiana Dome (Fig. 2; D'el-Rey Silva, 1995).
|
In an alternative model, magma is generated in unrelated sources of diverse ages. Continental to shallow marine, siliciclastic–carbonate supracrustal metasediments of the Sergipano Foldbelt are interbedded with volcanic rocks whose U–Pb zircon ages are 1·05–1·0 Ga, and intruded by granitoids with 0·98–0·92 Ga U–Pb ages (Van Schmus et al., 1995
). Brito Neves et al. (1995) proposed the name ‘Cariris Velhos’ for this period of rifting and associated magmatism. Figure 15 compares Coronel João Sá data with Sm–Nd analyses of Archean and Paleoproterozoic gneissic basement, and Mesoproterozoic Cariris Velhos supracrustal rocks (Van Schmus et al., 1995). This second model is consistent with the discordant U–Pb ages of zircon, which suggest sources of both Proterozoic (metasedimentary?) and possibly Archean age. The ranges of Nd indicate that 80% could have been provided from a source with Cariris Velhos isotopic characteristics. |
DISCUSSION |
|---|
This study of the Coronel João Sá pluton was initiated in an attempt to constrain the tectonic origin of the Sergipano Foldbelt and, by extension, of other foldbelts in north-east Brazil. The intercratonic and intracratonic models were tested by examination of potential magma sources for this pluton. If the geochemical characteristics of the source were juvenile, more ‘primitive’ material, then this observation would indicate an intercratonic origin as an ocean basin that closed through subduction. If the source consisted of older continental crustal material, then this observation would support an intracratonic origin as a continental extensional basin.
Petrographic and mineral chemistry data show evidence for both the process of magma mixing and incorporation of minor residual phases—e.g. inherited core material in zircon crystals. The presence of tonalitic enclaves shows that mixing between magmas of different compositions had occurred. A plausible explanation of the amphibole-rich xenoliths and presence of clinopyroxene (which is in severe disequilibrium with the magma) is dehydration-melting of an amphibolitic source rock to produce melt + residual clinopyroxene. Experiments by Beard & Lofgren (1991), Rushmer (1991), Wolf & Wyllie (1994), Patiño Douce & Beard (1995) and others have investigated the dehydration-melting of amphibolite (i.e. H2O is supplied only by breakdown of hydrous metamorphic minerals) or melting with insufficient externally added H2O to achieve saturation. In some experiments, amphibole and quartz were consumed to create clinopyroxene. Dehydration-melting of amphibolite, which is thought to be common in the lower continental crust, produces variable magma compositions according to the prevailing intensive variables (P, T, etc.) and also to the bulk compositions of the available source rocks. Experimentally determined temperatures for dehydration-melting are in the range of 850–1000°C (Rushmer, 1991), compatible with the 840°C initial temperature inferred from our calculation of the cooling history of the Coronel João Sá pluton.
As a hydrous mineral (such as amphibole) breaks down in the magma source, the melt volume abruptly increases; formation of stress fractures would permit batches of magma to drain off and the process could repeat by melting of another protolith with a somewhat different composition (Rushmer, 1991). The source of the Coronel João Sá magma is inferred to be mixed, the amphibolite xenoliths signifying a contribution from lower crust of basaltic composition, whereas the presence of zircons with inherited cores implies a recycled (sedimentary?) component. These diverse sources could explain differences among the western facies of the southern body (Fig. 3) which contains clinopyroxene, the eastern facies (cpx absent) and the separate northern body. These considerations support the intracratonic model for the origin of the foldbelt. The depth of crystallization at a minimum of 12 km, indicated by the presence of magmatic epidote, is also consistent with this model.
Linear major- and trace-element correlations on Harker variation diagrams, combined with variable initial Sr–Nd isotopic compositions, suggest magma mixing of at least two end-members derived from protoliths of diverse ages as indicated from inherited zircon cores. Likely candidates of appropriate age and Sm–Nd isotope systematics are Archean/Transamazonian basement, and Cariris Velhos age supracrustal material making up the foldbelt. Isotopic and trace-element data argue against a significant melt contribution from the mantle. Available data are not conducive to quantitative modeling of the end-members. There are no bulk chemical analyses of candidate source lithologies in the general region. Sm–Nd isotope data from the pluton and reconnaissance regional data (Van Schmus et al., 1995) do not plot as defined trends. Together, the data patterns indicate that the granodiorite magma was the product of partial melting of local heterogeneous crustal sources.
|
APPENDIX |
|---|
Diffusion calculations for the Rb–Sr system in minerals separated from sample CJS 20 are based upon the following assumptions.
- Cooling must have been slow (degrees per million years), as would be appropriate for the deeply buried Coronel João Sá pluton.
- Sr isotopes diffuse, but Rb and Sr concentrations in each mineral species are assumed to remain unchanged during cooling. Volume diffusion is the mechanism of transport. Exsolved phases are nearly absent in CJS 20 feldspar (the diffusion distance in perthite would have been much shorter than the bulk crystal dimension assumed in modeling).
- Dodson (1973) and Giletti (1991) assumed that diffusion occurs between a mineral and an ‘infinite reservoir’ that itself does not experience a change of Sr isotopic composition. The diffusion measurements of Giletti (1991), Giletti & Casserly (1994) and Cherniak & Watson (1994) took great precaution to assure communication with an infinite reservoir. Jenkin et al. (1995) and Jenkin (1997) considered the more realistic case of isotope exchange in a finite closed whole-rock system and showed that closure temperature depends in a complex manner upon the relative sizes and diffusion properties of all the participating mineral reservoirs.
A material balance calculation based upon mineral Rb and Sr concentrations and modal abundances (Table 6) shows that sample CJS 20 minerals, analyzed individually, account for nearly 93% of the whole-rock Rb (a reasonably good match), but only about 61% of the Sr. We did not analyze epidote, hornblende or clinopyroxene, which would be low-Rb, high-Sr minerals. However, their low modal abundances (epidote 1·8%, hornblende 2·9%, clinopyroxene 0·3%) suggest that these minerals can only partially account for the remaining 39% of the Sr; possibly, the analyses of purified mineral aliquots are not precisely representative values. In view of our ignorance of the Sr diffusion roles played by these minor mafic minerals, we have adopted the simple Dodson–Giletti assumption of an infinite exchange reservoir.
- Diffusion coefficients (D0) for biotite, apatite, titanite and K-feldspar (Table 6) are from fig. 7 of Giletti (1991), using D values pertaining to 800°C. D0 was calculated using D/D0 = exp(–E/RT), where E is activation energy (Table 6), R is the gas constant (8·314 joule/degree mole) and T = 1073·15 K. For CJS plagioclase, we used the Giletti & Casserly (1994) relationship between D0 and plagioclase composition (An31). [D values for plagioclase determined by Cherniak & Watson (1994) using an independent technique are very similar.] The D vs temperature relationship is assumed to extrapolate smoothly to temperatures that are much lower than the temperature range of experimental measurement. The influence of pressure upon diffusion rates is assumed to be inconsequential.
- We assume that closure of the U–Pb system in CJS 20 zircon occurred early, at high temperature. The 625 Ma U–Pb age is used to back-calculate the value of whole-rock initial 87Sr/86Sr (0·70805).
- There is no evidence that a later geochemical disturbance such as regional reheating had affected the mineral Rb–Sr systems.
|
ACKNOWLEDGEMENTS |
|---|
We thank Harlan Nicholas Aguiar for assistance in the field and with sample preparation. We also thank James Connelly for providing the lab facility for zircon analysis, and Kathy Manser for her patience and guidance in culling the best out of seemingly countless zircon grains. Larry Mack, Todd Housh and Eric James provided assistance with Rb–Sr and Sm–Nd analyses, and Fangqiong Lu helped with the microprobe. We held valuable discussions with Dan Barker, Martin Dodson, Bruno Giletti, Doug Smith, Jon Davidson, Ian McReath and Ben Castellana. Critical comments from Calvin Barnes, Randy Van Schmus, Susan DeBari and Marjorie Wilson prompted us to re-examine assumptions and make additional model calculations. This paper is based on the M.S. thesis of C. H. Castellana, who thanks the Geological Society of America for a travel grant to Brazil. PADCT/FINEP grant 65.930.619-00 to Sial helped to defray field and lab expenses.
|
FOOTNOTES |
|---|
* Present address: Weston Solutions, Inc., 14724 Ventura Boulevard, Suite 1000, Sherman Oaks, CA 91403-3501, USA.
Corresponding author. Telephone: (512) 471-7562. Fax: (512) 471-9425. E-mail: leonlong{at}mail.utexas.edu
|
REFERENCES |
|---|
Anderson, J. L. & Smith, D. R. (1995). The effects of temperature and fO2 on the Al-in-hornblende barometer. American Mineralogist 80, 549–559.[Abstract]
Barovich, K. M. & Patchett, P. J. (1992). Behavior of isotopic systematics during deformation and metamorphism: a Hf, Nd and Sr isotopic study of mylonitized granite. Contributions to Mineralogy and Petrology 109, 386–393.[CrossRef][Web of Science]
Beard, J. S. & Lofgren, G. E. (1991). Dehydration melting and water-saturated melting of basaltic and andesitic greenstones and amphibolites at 1, 3, and 6·9 kb. Journal of Petrology 32, 365–401.
Brito Neves, B. B. de, Sial, A. N. & Albuquerque, J. P. T. de (1977). Vergência centrifuga residual no sistema de dobramentos Sergipano. Revista Brasileira de Geociências 7, 102–114.
Brito Neves, B. B. de, Van Schmus, W. R., Santos, E. J. dos, Campos Neto, M. C. & Kozuch, M. (1995). O evento Cariris Velhos na Provincia Borborema: integração de dados, implicaçoes e perspectivas. Revista Brasileira de Geociências 25, 279–296.
Caby, R. & Arthaud, M. (1986). Major Precambrian nappes of the Brazilian Belt, Ceará, Northeast Brazil. Geology 14, 871–874.
Caby, R., Sial, A. N., Arthaud, M. & Vauchez, A. (1991). Crustal evolution and the Brasiliano Orogeny in Northeast Brazil. In: Dallmeyer, R. D. & Lécorché, J. P. (eds) The West African Orogens and Circum-Atlantic Correlatives. Berlin: Springer, pp. 373–397.
Campos Neto, M. C. & Brito Neves, B. B. de (1987). Consideraçoes sobre a organização e geometria do sistema de dobramentos Sergipano. Abstract Boletim Primeiro Simpósio Nacional de Estudos Tectônicos, Salvador, pp. 90–93.
Chaves, J. M. (1991). Maciços Coronel João Sá e Glória. Petrología e geoquímica de granitoides no domíno Macururé, faixa Sergipana (NE do Brasil). M.S. thesis, Universidade Federal da Bahia, 158 pp.
Cherniak, D. J. & Watson, E. B. (1994). A study of strontium diffusion in plagioclase using Rutherford backscattering spectroscopy. Geochimica et Cosmochimica Acta 58, 5179–5190.[CrossRef][Web of Science]
Connelly, J. N., Berglund, J. & Larson, S. A. (1996). Thermotectonic evolution of the eastern segment of southwestern Sweden: tectonic constraints from U–Pb geochronology. In: Brewer, T. S. (ed.) Precambrian Crustal Evolution in the North Atlantic Region. Geological Society, London, Special Publications 112, 297–313.
Cosca, M. A., Essene, E. J. & Bowman, J. R. (1991). Complete chemical analyses of metamorphic hornblendes: implications for normalizations, calculated H2O activities, and thermobarometry. Contributions to Mineralogy and Petrology 108, 472–484.[CrossRef][Web of Science]
Crank, J. (1975). The Mathematics of Diffusion, 2nd edn. Oxford: Clarendon Press, 414 pp.
Czamanske, G. K. & Wones, D. R. (1973). Oxidation during magmatic differentiation, Finnmarka Complex, Oslo Area, Norway: Part 2, the mafic silicates. Journal of Petrology 14, 349–380.
Davidson, J., Tepley, F., III, Palacz, Z. & Meffan-Main, S. (2001). Magma recharge, contamination and residence times revealed by in situ laser ablation isotopic analysis of feldspar in volcanic rocks. Earth and Planetary Science Letters 184, 427–442.[CrossRef][Web of Science]
Davison, I. & dos Santos, R. A. (1989). Tectonic evolution of the Sergipano Fold Belt, NE Brazil, during the Brasiliano Orogeny. Precambrian Research 45, 319–342.[CrossRef][Web of Science]
D'el-Rey Silva, L. J. H. (1995). The evolution of basement gneiss domes of the Sergipano Fold Belt (NE Brazil) and its importance for the analysis of Proterozoic basins. Journal of South American Earth Sciences 8, 325–340.[CrossRef][Web of Science]
DePaolo, D. J. (1981). Neodymium isotopes in the Colorado Front Range and crust–mantle evolution in the Proterozoic. Nature 291, 193–196.[CrossRef]
Dodson, M. H. (1973). Closure temperature in cooling geochronological and petrological systems. Contributions to Mineralogy and Petrology 40, 259–274.[CrossRef][Web of Science]
Frost, T. & Mahood, G. A. (1987). Field, chemical, and physical constraints on mafic–felsic magma interaction in the Lamarck Granodiorite, Sierra Nevada, California. Geological Society of America Bulletin 99, 272–291.
Frost, B. R., Barnes, C. G., Collins, W. J., Arculus, R. J., Ellis, D. J. & Frost, C. D. (2001). A geochemical classification for granitic rocks. Journal of Petrology 42, 2033–2048.
Giletti, B. J. (1991). Rb and Sr diffusion in alkali feldspars, with implications for cooling histories of rocks. Geochimica et Cosmochimica Acta 55, 1331–1343.[CrossRef][Web of Science]
Giletti, B. J. & Casserly, J. E. D. (1994). Strontium diffusion kinetics in plagioclase feldspars. Geochimica et Cosmochimica Acta 58, 3785–3793.[CrossRef][Web of Science]
Hammarstrom, J. M. & Zen, E. (1986). Aluminum in hornblende: an empirical igneous geobarometer. American Mineralogist 71, 1297–1313.[Abstract]
Holland, T. & Blundy, J. (1994). Non-ideal interactions in calcic amphiboles and their bearing on amphibole–plagioclase thermometry. Contributions to Mineralogy and Petrology 116, 433–447.
Jardim de Sá, E. F. (1984). A evolução Proterozóica da Província Borborema. XI Simpósio de Geologia do Nordeste, Natal, pp. 33–45.
Jenkin, G. R. T. (1997). Do cooling paths derived from mica Rb–Sr data reflect true cooling paths? Geology 25, 907–910.
Jenkin, G. R. T., Rogers, G., Fallick, A. E. & Farrow, C. M. (1995). Rb–Sr closure temperatures in bi-mineralic rocks: a mode effect and test for different diffusion models. Chemical Geology 122, 227–240.[CrossRef][Web of Science]
Le Bas, M. J. & Streckeisen, A. L. (1991). The IUGS systematics of igneous rocks. Journal of the Geological Society, London 148, 825–833.
Liou, J. G. (1973). Synthesis and stability relations of epidote, Ca2Al2FeSi3O12(OH). Journal of Petrology 14, 381–413.
McCarthy, T. C. & Patiño Douce, A. E. (1998). Empirical calibration of the silica–Ca Tschermak's–Anorthite (SCAn) geobarometer. Journal of Metamorphic Geology 16, 675–686.[CrossRef][Web of Science]
McReath, I., LaFon, J. M., Chaves, J. M. & Conceição, H. (1993). Strontium isotope and geochemical evidence for complex petrogenesis in the Coronel João Sá intrusion, Sergipe Fold Belt. Proceedings of the 4th Congresso Brasileiro de Geoquimica, pp. 86–89.
McReath, I., LaFon, J. M., Davison, I., Chaves, J. M. & Conceição, H. (1998). Brasiliano-age granitoids in the Sergipana Fold Belt, NE Brazil: the Coronel João Sá pluton. Journal of South American Earth Sciences 11, 51–66.[CrossRef][Web of Science]
Ministério das Minas e Energia, Departamento Nacional da Produção Mineral (1988). Programa Levantamentos Geológicos Básicos do Brasil: Carta Geológica Folha Carira.
Naney, M. T. (1983). Phase equilibria of rock-forming ferromagnesian silicates in granitic systems. American Journal of Science 283, 993–1033.
Patiño Douce, A. E. & Beard, J. S. (1995). Dehydration-melting of biotite gneiss and quartz amphibolite from 3 to 15 kbar. Journal of Petrology 36, 707–738.
Poli, G. E. & Tommasini, S. (1991). Model for the origin and significance of microgranular enclaves in calc-alkaline granitoids. Journal of Petrology 32, 657–666.
Richardson, S. W. (1968). Staurolite stability in a part of the system Fe–Al–Si–O–H. Journal of Petrology 9, 467–488.
Rushmer, T. (1991). Partial melting of two amphibolites: contrasting experimental results under fluid-absent conditions. Contributions to Mineralogy and Petrology 107, 41–59.[CrossRef][Web of Science]
Santos, E. J. dos & Silva Filho, M. A. (1975). Ensaio interpretativo sobre as evolução da Geosinclinal de Propriá, Nordeste do Brasil. Revista Mineração e Metalurgia 367, 3–22.
Santos, R. A. dos, Menezes Filho, N. R. de & Souza, J. D. de (1988). Programa levantamentos geológica básicos do Brasil. Carira. Folha SC24-Z-A-III. Texto Explicativo. Brasília: DNPM/MME.
Schmidt, M. W. & Thompson, A. B. (1996). Epidote in calc-alkaline magmas: an experimental study of stability, phase relationships, and the role of epidote in magmatic evolution. American Mineralogist 81, 462–474.[Abstract]
Scott, D. J. & St-Onge, M. R. (1995). Constraints on Pb closure temperature in titanite based on rocks from the Ungava Orogen, Canada: implications for U–Pb geochronology and P–T–t path determinations. Geology 23, 1123–1126.
Sial, A. N. (1986). Granite types in northeast Brazil: current knowledge. Revista Brasileira de Geociências 16, 54–72.
Sial, A. N. (1987). Granitic rocks of northeast Brazil. Proceedings, International Symposium on Granites and Associated Mineralisations (ISGAM), Salvador, pp. 61–69.
Sial, A. N., Ferreira, V. P., Fallick, A. E. & Cruz, M. J. M. (1998). Amphibole-rich clots in calc-alkalic granitoids in the Borborema Province, Northeastern Brazil. Journal of South American Earth Sciences 11, 457–471.[CrossRef][Web of Science]
Sial, A. N., Toselli, A. J., Saavedra, J., Parada, M. A. & Ferreira, V. P. (1999). Emplacement, petrological and magnetic susceptibility characteristics of diverse magmatic epidote-bearing granitoid rocks in Brazil, Argentina and Chile. Lithos 46, 367–392.[CrossRef][Web of Science]
Stacey, J. S. & Kramers, J. D. (1975). Approximation of terrestrial lead isotope evolution by a two-stage model. Earth and Planetary Science Letters 26, 207–221.[CrossRef][Web of Science]
Sun, S.-s. & McDonough, W. F. (1989). Chemical and isotopic systematics of ocean basalts: implications for mantle composition and processes. In: Saunders, A. D. & Norry, M. J. (eds) Magmatism in the Ocean Basins. Geological Society, London, Special Publications 42, 313–345.
Van Schmus, W. R., Brito Neves, B. B. de, Hackspacher, P. & Babinski, M. (1995). U/Pb and Sm/Nd geochronologic studies of the Eastern Borborema Province, Northeastern Brazil: initial conclusions. Journal of South American Earth Sciences 8, 267–288.[CrossRef][Web of Science]
Wolf, M. B. & Wyllie, J. (1994). Dehydration-melting of amphibolite at 10 kbar: the effects of temperature and time. Contributions to Mineralogy and Petrology 115, 369–383.[CrossRef][Web of Science]
Zen, E. & Hammarstrom, J. M. (1984). Magmatic epidote and its petrologic significance. Geology 12, 515–518.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||