In the Paraná volcanic province (Figure 1), there are occurrences of paralavas that are genetically related to remelting of the holocrystalline host rock at 1,600 ºC. These rocks were previously denominated pegmatites or segregation sheets and occur in all continental volcanic provinces and plutonic rocks of basaltic composition from many provinces. The formation of these rocks is usually ascribed to magmatic processes that involve crystal fractionation and the filter-pressing ascent of the fractionated liquid to form dikes and sills within the mother flow or intrusion. The universal presence of long, thin, hollow crystals in a glassy matrix is a common description of paralavas that resulted from quenching during pyrometamorphism (Grapes et al. 2013).
The presence of the voluminous, glassy matrix in coarse-grained basalt is a recurring feature in large volcanic provinces but is still a paradox that remains to be solved. Several questions arise and remain unanswered from the presence of quenched paralavas in thick basalt flows: (1) Processes that caused a strong, local increase in temperature, so the nearly solid basalt remelted; (2) The chemical fractionation of the paralava as compared to the host basalt; (3) The rise of the newly-formed magma through the solid, hot basalt; (4) The quenching of the ascending liquid due to strong thermal disequilibrium with the host basalt; and (5) The initial, strongly reduced composition of the paralava. We presently address all these questions and interpret them in a unified genetic model.
In occurrences of gabbroic pegmatites from Skaergaard (Larsen et al. 1992), fluids are dominated by H2O, CO2 and CH4 as inclusions in plagioclase, quartz and apatite. The source of the carbon component was attributed to assimilation of sedimentary blocks from the volcanoclastic sequence, but the hypothesis of magmatic origin of carbon in silicate glass inclusions (MORB type magma) was not ruled out. Although this evaluation is close to the actual process involved in the formation of these rocks, it is likely that the processes involved in their origin are variable, particularly the types of fluids present. All evaluations attribute the mobility of the fractionated fluid to the presence of gases, but the provenance of the gases is considered the degassing of the host magma (Puffer and Horter 1993). The evidence evaluated thus far relies on a single source of energy, namely the heat liberated from the cooling magma. An additional source of energy seems necessary to remelt the host basalt, adding complexity to the processes involved. We propose a unified description of processes, including field evidence and the interpretation of geochemistry in a scenario of methane generation in the basin and its ascent and reaction in the hot, solid lava. We found the evidence in outcrops of the Paraná volcanic province where reactive relationships are present between the host basalt and the surrounding paralava. We describe here this evidence from Francisco Beltrão (near the depocenter of the Paraná Basin), including field relationships, petrography, rock geochemistry, electron microprobe and fluid inclusions in quartz and plagioclase. The origin of these rocks constitutes a significant problem not yet solved in igneous petrology. The copper enrichment in the paralavas gives economic significance to these rocks.
MATERIALS AND METHODS
The study of paralavas in the Paraná volcanic province derived from several years of research of hydrothermal mineralization hosted in volcanic rocks (e.g., Duarte et al. 2009, 2011, Hartmann et al. 2012a, b, Rosenstengel and Hartmann 2012) and sand injectite structures (Hartmann et al. 2013). Field studies for the present study focused in two areas (Figure 1) that provided significant evidence for the genetic interpretation. The first area is located in southwestern Paraná State that was previously described by Wildner et al. (2006); their samples AS, KN and WW are used here. Dalba quarry in the municipality of Francisco Beltrão is within this area. Samples P1b, R2Pb are from the host basalt, while PD1, P2p, P3p and P4 are from the paralavas in this quarry. Sample PD1 was collected for fluid inclusions and contains samples P2p and P3p, while P4 is a green core collected for XRD. The second area is the Rochasul quarry, located in the town of Jardim, west-central Mato Grosso do Sul State, where sample MS36 was collected.
The methodology included evaluation and description of paralava outcrops indicating the elevation and geographical coordinates (Garmin GPS). The sampled host volcanic rocks (n = 13) and paralavas (n = 14) were analyzed for major and trace elements in the ACME laboratories, Canada. Thin sections from these samples were prepared at UFRGS and at the Geological Survey of Brazil (CPRM/PA). X-ray diffraction (n = 1) was performed in the laboratories of UFRGS with a Siemens Bruker-AXS D5000 with 2θ goniometer, radiation Kαl copper tube. Electron microprobe analyses and backscattered electron images were obtained for magnetiteilmenite, augite-pigeonite and feldspar with a Cameca SX-100 equipment housed at the Institut für Mineralogie und Kristallchemie, Stuttgart University, Germany. The routine measurement protocols used were 15 kV accelerating voltage, 15 nA beam current and a beam size approximately 1 μm. The fluid inclusions study was undertaken at the Núcleo de Apoio à Pesquisa Geoanalítica, Instituto de Geociências, Universidade de São Paulo (Geoanalítica - USP Facility). Four doublepolished thin sections were prepared from sample PD1, suitable for fluid inclusions analyses. During petrographic description using a microscope Leitz Wetzlar, the fluid inclusions were assessed for their characterization, distribution and mode of occurrence to identify their possible origin (primary, secondary or pseudo-secondary), and also for the selection of the best fields for microthermometric analysis. The microthermometry of fluid inclusions was carried out in a CHAIXMECA MTM 85 heating and freezing stage calibrated with chemical compounds of Merck MSP for temperatures higher than 40 °C. The calibration also used synthetic standard Syn Flinc (Synthetic Fluid Inclusions). At -56.6 °C, the composition of standard 1 was 75 mol % H2O and 25 mol % CO2. At -21.2 °C, the standard 2 has Eutectic composition of H2O + NaCl system with 23.2 wt.% NaCl. At -10.7 °C, Standard 3 had eutectic composition of H2O + KCl system with 19.65 wt.% KCl. At 0 °C and 374.1 °C, the standard 4 was composed of pure water (triple point and critical point, respectively). The standard 5, transition of α quartz to β quartz, was used at 573.0 °C.
Information processing used freeware GCDKit (http://www.gcdkit.org/) for the chemical analyses, DIFFRACPLUS software for X-ray diffraction (standards of PDF2 database) and Microsoft Excel worksheet CalcMin (Brandelik 2009) for EPMA data. Interpretations with the determination of fluid inclusions salinity were performed by thermodynamic equations inserted into the software FLUIDS (Bakker 2003). CorelDraw software was used for preparation of figures.
The largest number of occurrences of paralavas described in the Paraná volcanic province (n = 38) is located in southwestern Paraná state, previously described as pegmatites (Arioli 2008). Fewer occurrences are known in other portions of the province. The paralavas occur mostly in thick (~70 m) lava flows and usually form small (some 0.1-2.0 m thick, most 1-10 m and a few reaching 100 m in length). They form vertical, inclined or horizontal bodies, including stockworks. The contacts of the intrusive bodies may be sharp but are mostly gradational with the host rocks.
Different names have been used to describe these rocks, most commonly pegmatite (Table I). But the common presence of glass in the groundmass (intersertal texture) precludes the use of the word pegmatite for classification. The presence of long (10 cm), curved, skeletal and hollow crystals immersed in glass requires quenching of the lava. Paralava occurrences have been described in basalt, basaltic andesite and andesite. In this study, we refer to the host rocks as basaltic rocks. The study by Wildner et al. (2006) indicates the presence of 2.5 vol.% silicic rocks in the Paraná volcanic province (e.g., rhyodacite, rhyolite, latite, quartz latite, dacite). However, paralavas have not been described in these rocks. Overall, field relationships of the paralavas are comparable to the description of segregation sheets in the Holyoke flood-basalt flow in the Hartford Basin, Connecticut (Philpotts et al. 1996). Similarly, paralavas in the Paraná volcanic province were generated in the lower third of the flow core and occur only in intrusions in thick lava flows displaying vertical cooling joints (Figure 2a, b). This seems a requirement for the formation of the paralavas. Rapid cooling in thinner flows precludes the formation of significant volumes of paralavas.
There are no occurrences of injected paralavas into the columnar jointing of the host basalt, suggesting that the formation of paralavas was earlier (higher temperature) than the columnar joints. The contacts between the paralavas and host basalt are locally sharp but mostly gradational against a coarse-grained, holocrystalline host basalt. Also, the paralavas do not crosscut the lower or upper contact of the flow and remain confined to the flow.
|Pegmatite||North Mountain basalt, Nova Scotia Canada||Walker (1953), Greenough and Dostal (1992b)|
|Pegmatitic segregation vein||Flood basalt||Puffer and Horter (1993)|
|Pegmatitic segregation sheet||Flood basalt flow||Philpotts et al. (1996)|
|Pegmatoid gabbro||Basalt, Serra Geral Group||Vasconcellos et al. (2001)|
|Mafic pegmatite||North Mountain basalt, Nova Scotia Canada||Kontak et al. (2002)|
|Basic pegmatite||Basalt, Serra Geral Group||Arioli (2008), Silva (2011), Ferreira (2011), Ferreira et al. (2014)|
|Paralava||Flood basalt and intrusives||This work|
For the full understanding of the genesis of the paralavas, we examine the sedimentary rocks of the Paraná Basin below the Paraná volcanic province. These rocks are composed mostly by fine to medium-grained sandstones with some siltites, and minor limestones. However, the Ponta Grossa Formation (states of Paraná and São Paulo) and the Irati Formation (entire basin) have thick layers of bituminous shales (e.g., Goulart and Jardim 1982). For example, "the Irati Formation, a wellknown unit for its oil-prone rocks ..., extends almost throughout the entire Paraná Basin. It has an average thickness of 40 m, with peaks of about 70 m..." (Holz et al. 2010). Permian sequences (e.g., Irati Formation) of the Paraná Basin are well represented in the underburden of the Paraná volcanic province in the Rochasul quarry and in the western border of the basin (e.g., Simas et al. 2012, their Figure 1). An example along the eastern border is the description of a drill core near Torres, Rio Grande do Sul (Aborrage and Lopes 1986). At 600 m depth, the core displays 35 m of dark siltites and clayey limestones of the Permian Irati Formation. It is most relevant for the origin of the paralavas that the Irati Formation contains pyrobituminous shales, from which kerogen is extracted commercially by PETROBRAS in São Mateus do Sul (Paraná).
Kerogen (particularly type III) is known to liberate large volumes of methane when intensely heated (Grapes 2011). Twenty-six occurrences of oil in the eolian sandstones near the top of the Paraná Basin in the state of São Paulo had their formation attributed to the intrusion of diabase sills into the black shales of the Irati Formation deeper in the basin (Araújo et al. 2004). As indicated by Aarnes et al. (2010), oil is formed in bituminous shales at a larger distance from the sill and methane is formed closer to the sill contact during the same pyrometamorphic event.
The lower Taquaral Member of the Irati Formation comprises siltstones and gray to black mudstones, and the upper Assistência Member has organic-rich mudstones and shales interbedded with limestones (Holz et al. 2010). The upper Assistência Member is rich in total organic carbon (TOC = 10-25 %; e.g., high in the Lauro Müller region, TOC = 20%). This high-TOC kerogen is known to crack into 95% methane when subjected to temperature >200 °C. Several other formations in the basin also contain black shales. For example, the Serra Alta Formation overlies the Irati Formation and also contains marine shales that contributed to the kerogen budget for methane generation during basalt sill injection. In the Ponta Grossa Formation, TOC seems to be lower, between 0.1-0.6% (Goulart and Jardim 1982).
The host basalts of paralavas in the province are holocrystalline, fine to medium-grained. In the Dalba quarry, the thick basaltic flow has paralava intrusions. Both paralavas and host basalt are cut by the cooling joints indicating that the formation of paralavas was prior to the formation of cooling structures of the volcanic flow. The geological relationships between the host basalt and the paralavas exhibit reaction rims (Figure 2c, d) that are distinct in the field, in petrography and in chemical composition. The reaction rims include a green core and two intervening light and dark grey portions close to the host basalt. The light portion has mostly plagioclase and evolves gradually into a dark grey portion with opaque minerals, plagioclase and pyroxene. The constitution of the green core is mostly celadonite, as determined by x-ray diffraction. However, Arioli (2008) described these portions as dark glass containing about 5 wt.% H2O. The alteration of glass into celadonite is a well known volcanic process. The largest body of paralavas observed in the Paraná volcanic province occurs in the Rochasul quarry (Figure 2e, f). The exposed dimensions are approximately 20 m in thickness by 200 m in length (Brückmann et al. 2013). The paralavas have typical long and curved clinopyroxene up to 10 cm long. The quarry is entire taken by the paralava without exposure of the host rock. The host rock is interpreted as a sedimentary formation, because the geological contact of the first basalt flow of the Paraná volcanic province, in direct contact with an eolian sandstone of the Botucatu Formation, occurs in a higher topographic level along the BR267 highway, eight kilometers distant from the quarry.
The host basalt is a mesocratic rock, greenish gray, fine grain size. The mineralogical composition (Figure 3) is approximately 48 vol.% plagioclase (mainly labradorite), ~ 26% clinopyroxene (augite and pigeonite), ~16% opaque minerals (magnetite and ilmenite) and ~10% clay minerals (smectite and celadonite). Plagioclase forms elongated crystals (≤0.3 mm) subhedral to euhedral with intergranular texture, whereas the clinopyroxene and opaque minerals occupy the interstices. Plagioclase crystals show corroded borders in direct contact with the smectite or celadonite and no rim alteration in contact with clinopyroxene and opaque minerals. Clinopyroxene forms subhedral to anhedral crystals (0.2-0.3 mm) and has poikilitic inclusions of opaque minerals. Clinopyroxene is commonly fractured and altered to clay minerals. Opaque minerals range in size from 0.2 to ≤0.5 mm, and may be euhedral (lozenge to hexagonal) or spherical and even anhedral with rim alteration to red iron oxide (hematite). This alteration occurs along fractures and neighboring crystals. Texture varies from subophitic to ophitic. The opaque minerals are euhedral (hexagonal or rectangular), ~0.3 mm in size, mostly without rim alteration. Olivine and apatite occur as accessory minerals in the host rocks.
The paralavas comprise either horizontal or vertical thin bodies, in some cases forming a stockwork interconnected by veins. In the Dalba quarry (sample PD1), dark and light reaction rims are observed. The mineralogical composition of the dark reaction rim is similar to the host basalt with 45 vol.% plagioclase, ~20% clinopyroxene, ~22% opaque minerals, ~10% clay minerals (smectite and celadonite) and ~3% accessory minerals (apatite and quartz), in addition to traces of carbonates. The opaque minerals (0.1 to 0.2 mm long) are concentrated in the matrix and also in the surrounding pockets of celadonite. Phenocrysts of opaque minerals with anhedral and also skeletal habit show alteration to red iron oxide (hematite). Plagioclase occurs in three distinct habits. The most significant is made up of very fine grained, anhedral crystals associated with quartz that forms a cryptocrystalline matrix in the crystal interstices. This mass of plagioclase is common around the blobs of celadonite. The second plagioclase habit has elongated crystals embedded in the matrix (0.1 to 0.2 mm, whereas the third corresponds to fractured phenocrysts immersed in the matrix. The clinopyroxene (0.1 to 0.2 mm long) is associated with plagioclase and opaque minerals, and occurs also as sparse phenocrysts showing intense fracturing. The celadonite is also scattered in the rock. The mineralogy of the light reaction rim of the paralava (Figure 3) comprises ~ 50 wt.% plagioclase (mainly andesine), ~23% of clinopyroxene (augite), ~20% opaque minerals (magnetite and ilmenite), ~5% celadonite, <1% native copper and traces of apatite and quartz. This mineral assemblage has typically phenocrysts of plagioclase, clinopyroxene and opaque minerals larger than 0.5 mm. The predominant textures are sub-ophitic, ophitic and intersertal. Plagioclase is elongated with the major axis longer than 1 mm; the mineral does not show reaction rims. The clinopyroxene crystals are intensely fractured, whereas the opaque minerals are characterized by a skeletal habit. Apatite is an accessory mineral and stands out in the paralavas for a higher volume (greater than the host basalts), by the needle shape or hexagonal shape in basal (001) sections. Celadonite occurs in green pockets, and veins of celadonite link these pockets with fractures and in many cases cross the fractures of crystals to interconnect with other pockets of celadonite. The native copper mineralization in the paralava (Figure 4) occurs commonly associated with this reaction rim, filling millimetric cavities in association with smectite or celadonite (Figure 4a, b, c). Native copper may be also associated with the clinopyroxene (Figure 4d, e, f). The copper content of the light grey reaction rim is 600 ppm, although the analyzed sample did not have visible metal. The copper content may vary to higher contents due to the nugget effect.
The preparation of sample P4 included scraping, separation and concentration of the green core material. The mineralogical determination was performed by the whole rock, X-ray diffraction powder method in the scan range (2θ) from 2° to 72°. The evaluation of these results was based on the peak intensity and the interplanar distances. The main peak intensity had interplanar distances of 9.97 Å, 4.53 Å, 4.35 Å, 3.63 Å, 3.08 Å, 2.57 Å, 2.40 Å e 1.51 Å which identifies celadonite. Other peaks confirmed the presence of plagioclase, quartz and calcite.
The evaluation of the geochemistry of paralavas requires the understanding of the geochemistry of the host basalts, because they occur only in thick flows (>70 m). The presence of reaction rims between the host basalt and the paralavas in the Dalba quarry motivated us to analyze the reaction rims. In the present evaluation, we make a comparison between the geochemistry of the host basalt and the paralavas. The chemical analyses are listed in Tables II and III. In addition, we used the chemical analyses from Ferreira (2011) in samples close to the study area. According to their chemical compositions, the host rocks are basalts, both in the diagram of Cox et al. (1979) and Middlemost (1994), available in Figure 5. The basalts are of the Paranapanema intermediate-Ti chemical type (Nakamura et al. 2003) whereas the paralavas have higher SiO2 content similar to basaltic andesite and basaltic trachyandesite and also higher content of incompatible elements, including TiO2. Similar relationship is observed in the samples of Ferreira (2011). On the other hand, the geochemical compositions of host basalts and paralavas are also similar to the study of D'Oriano et al. (2013), who remelted basalt tephra in the laboratory and obtained liquids enriched in incompatible elements (Figure 5). Comparatively, the analyses on samples from the Dalba quarry show that the average content of SiO2 from the host basalt is 50.17 wt.%, similar to other basalts from the Paraná volcanic province. However, the reaction rims of paralava have increments of 2.6 wt.% in the dark reaction rim and 9.5 wt.% in the light reaction rim with contents of 51.47 wt.% and 54.96 wt.% respectively. Corresponding geochemical distribution occurs in K2O, P2O5, Ba, Y, Zr, Th, U, Cu, La and other REE. All these elements are enriched in the paralavas. In the Dalba quarry, this behavior is evident and shows enrichment from the host basalts to the dark reaction rim, and even higher in the light reaction rim where plagioclase phenocrysts occur. In contrast, MgO and CaO exhibit opposite behavior. The MgO content (Figure 5) in the host basalts ranges from 6.87 wt.% to 4.93 wt.% and in the paralavas decreases to values between 4.93 wt.% and 1.38 wt.%. Analyses form Dalba quarry samples show the host basalt contents of MgO with an average approximately 4.84 wt.%. MgO gradually decreases in the paralava with 3.97 wt.% in the dark reaction rim and 2.62 wt.% in the light reaction rim. In other words, the decrease is larger in portions where plagioclase phenocrysts are present. The same behavior is observed in CaO that has the highest concentration in the host basalts, varying between 10.97-9.05 wt.% while in the paralavas this content ranges from 7.89 to 5.5 wt.%. In samples from the Dalba quarry, the content of CaO in the host basalts showed an average of 9.5 wt.% whereas it successively decreases from the dark reaction rim (6.17 wt.%) to the light reaction rim (4.49 wt.%).
Loss on ignition of the host basalts varies between 0.5-1.8 wt.% and between 0.9-2.4 wt.% in the paralavas. In the Dalba quarry, the host basalts display an average loss on ignition of 1.6 wt.%. In the dark reaction rim, loss on ignition is 2.4 wt.% and in the light reaction rim 1.5 wt.%. This distribution shows that these rocks were affected by the H1, H2 and H3 hydrothermal events described in detail by Hartmann et al. (2012a, b, 2013).
The electron microprobe analyses were performed in samples of host basalts and paralavas in order to compare their mineralogy. The composition of feldspar is plotted in the ternary classification diagram Ab-An-Or (Figure 6a, b) and representative EMPA results are listed in Table IV. Overall, the feldspar of host basalts ranges in composition from An50 to An70 (labradorite), but several analyses are more sodic. On the other hand, the plagioclase in the paralavas presents compositions between An30 and An50 (andesine) and some analyses are labradorite, a more sodic composition than in the host basalts. The alkali feldspar of host basalt and paralavas is essentially composed by sanidine. Most analyses of KF in the host basalt are sanidine compositions between Or55-80. Some analyses are Or99 and some individual analyses plot in the orthoclase field. In the paralavas, the sanidine displays Or35-85 and a few analyses plot in the orthoclase field.
Clinopyroxene was observed in both host basalt and paralava. According to the classification of Morimoto et al. (1988), augite and pigeonite are present in the host basalt, and only augite in the paralavas (Figure 6c, d; Table V). In the host basalt, the clinopyroxene is present in the matrix and also as sparse microphenocrysts; in the paralavas, clinopyroxene occurs only as phenocrysts. The variation of pigeonite composition in the host basalt is Wo7-20 (Figure 6d). The augite composition in the host basalt ranges from Wo26-41 with slightly higher Mg content than that observed in the augite from paralavas. On the other hand, the augite composition in the paralavas displays compositions between Wo27-42 and has in general higher iron content than augite from the host basalt.
The opaque minerals are approximately 0.1 to 0.2 mm in size in the matrix and also occur as anhedral phenocrysts with skeletal habit and showing some alteration. The electron microprobe analyses of magnetite and ilmenite are plotted in the binary diagram FeO versus TiO2 (Figure 6e, f) that shows positive correlation between the two minerals. Representative EMPA results are listed in Table VI. The magnetite-ilmenite pair has similar compositional distribution in both the host basalts and the paralavas. However, the magnetite from paralavas has slightly higher FeO and TiO2 content than the host basalt. The analyses of ilmenite from paralavas show a wide distribution of FeO and TiO2 contents but in the high FeO and high TiO2 analyses, the ilmenite of paralavas has higher FeO and similar TiO2 content than the host basalts. The occurrence of magnetite and ilmenite is linked to the presence of native copper in paralavas (Figure 7) similar to the relationship of copper mineralization with clinopyroxenes, smectite and celadonite.
Four double polished thin sections of sample PD1 (P1C, P5, P6 and P7) were prepared for analyses of fluid inclusions. The most common size of fluid inclusions is 5-10 μm. In P1C, 8 (eight) fluid inclusions were evaluated, although the data were obtained in only 4 fluid inclusions due to their tiny dimensions. In these inclusions, salinities range from 5.6 to 7.3 wt.% eq. NaCl, but the results were disregarded because many of these inclusions have suggestive behavior of metastability. In thin section P5 57 fluid inclusions were investigated. Although the majority has small dimensions between 5 and 10 μm some reach 12 μm, 17 μm and 25 μm. Some have suggestive features of melt inclusions (Figure 8a), while the largest fluid inclusions have leakage features and necking down and were not measured. Part of the fluid inclusions is composed of aqueous saline solutions, although some of them are metastable. The Tfg (ice melting temperature) in negative temperature was repeated in several runs. In these cases, the salinity ranged from 13.9 to 16.1 wt.% eq. NaCl. Some of the fl uid inclusions are dark and distorted, but repeated cooling to -180 °C showed no change in phase. This seems to indicate the presence of CO2 or mixtures of CO2 with other volatiles, unless the inclusions have extremely low densities such as water vapor, CO2 gas or CH4 gas. Phase changes were not observed with microthermometry. In thin section P6, 43 fl uid inclusions were analyzed. The fl uid inclusions are rounded and a few are black, elongated or distorted (Figure 8b). The main size of fl uid inclusions is from 5-15 μm, but some are 20 to 40 μm. The microthermometric measurements in these fluid inclusions presents dubious results. The system H2O + NaCl + CaCl2 (+/- MgCl2) is cautiously suggested with salinity ranging approximately from 14 to 16 wt.% eq. NaCl. A clear single-phase inclusion cooled down to -180 °C and presented no changes, showing that it may consist of a solid or of a lowdensity fl uid undetectable by microthermometry (CO2 gas, CH4 gas or water vapor). There is also a secondary fl uid inclusion (~100 μm), entrapped at low temperature with suggestive leakage features, and displaying very different low salinity (2.30 wt.% eq. NaCl), morphology, behavior and size from other fl uid inclusions observed.
In thin section P7, 18 fl uid inclusions were analyzed. The sizes vary from 5 to 10 μm and only one is 17 μm with suggestive features of melt. The bubbles within the fluid inclusions are small. During the heating of these fluid inclusions, the temperature was raised to 600 °C with no change in any of the inclusions analyzed. In all thin sections, the eutectic temperatures of the aqueous phase ranged mainly from - 60 °C to - 50 °C, suggestive of an aqueous system comprising NaCl + H2O + CaCl2 (+/- MgCl2). Nevertheless, these results are often dubious due to the small size of the inclusions and the occurrence of metastability.
The fi eld observation of the formation of paralavas as reaction rims within the host basalt precludes any interpretation of the paralavas as products of fractional crystallization. Although the products of remelting of the host basalt may be chemically similar to fractional crystallization, the process is overall comparable to the remelting of volcanic blocks entrained in lava (D´Oriano et al. 2013). All previous investigations considered fractional crystallization as the main process responsible for the enrichment of the lava in incompatible elements and its ascent to the upper part of the lava core (Table VII). But some geochemical features cannot be explained entirely with use of fractional crystallization. For instance, the study of Greenough and Dostal (1992a) showed that the presence of the rhyolite layers and pegmatite in the North Mountain basalt flows cannot be explained by fractional crystallization. Mass balance calculations of the major elements have missing phases, which in that study is represented by stilpnomelane. The authors suggested that the rhyolite layers and the pegmatite are product of silicate liquid immiscibility. The study of Hartley and Thordarson (2009) concluded that although "...crystal fractionation exerts a first-order control on the Grande Ronde basalt composition, it is generally agreed that fractional crystallization alone is not sufficient to explain the compositional and isotopic variations observed in the Grande Ronde basalt. Both mantle heterogeneity and crustal contaminations are advocated as possible explanations for these variations."
The abundant glass observed in the groundmass of all pegmatite lenses and the presence of large (10 cm), curved, hollow crystals requires quenching of the lava. This texture is observed in all pegmatite lenses described in continental basalts in other provinces and in the Paraná volcanic province.
For instance, in Hawaii, segregation veins occur in entirely crystalline diabase and contain ~30 vol.% glass matrix (Fodor and Bauer 2014). The geochemistry of these veins and host basalts is similar to the Paraná volcanic province occurrences. The reason for the presence of undevitrified glass in these pyrometamorphic rocks remains largely unexplained in general (Grapes 2011).
There is no reasonable explanation for the presence of higher temperature liquid lava in the lower core of a thick, solid lava flow, or for its quenching against comparable-temperature host rock. There is also no known fractional crystallization process that would raise the temperature of cooling basalt above its liquidus temperature to cause remelting of the lower core. An extraneous source of energy seems required to cause remelting of the crystalline basalt. We suggest the combustion of methane as the cause of remelting of basalt and formation of the paralava. It is therefore necessary to examine the steps involved in the process, including the source of the methane, its non-reactive ascent in the crust, spontaneous combustion in the lower core of the lava, the temperature attained during combustion, remelting of the crystalline basalt (paralava formation), and injection of the paralava upwards into the core and quenching of the paralava.
The methane present in sedimentary basins in general may have originated from mantle degassing, anaerobic bacterial reduction of organic matter or thermal cracking of kerogen. The Paraná Basin contains several sedimentary formations rich in kerogen under the Paraná volcanic province. The two most significant ones are the coal-rich, Permian Rio Bonito Formation present in 10% of the basin in its southeastern portion, and the Devonian Ponta Grossa Formation and Permian Irati Formation present along the entire basin. Bituminous shales are also present in the Serra Alta Formation higher up in the stratigraphy.
The injection of basaltic sills into coal seams turns kerogen mostly into CO2. "Humic coals are mainly composed of Type III kerogen from which 10-25% of the carbon mass can be converted into gas" whereas "...type I and II kerogen commonly found in organic-rich shales have the potential of converting up to 95% of the TOC to hydrocarbons..." (Aarnes et al. 2010). Because the coal seams cover a minor area of the Paraná Basin and coal generates mostly CO2, the bituminous shales of the Irati Formation are more significant for our modelling. In several continental volcanic provinces, "in shales with total organic carbon content (TOC) >5 wt.%, CH4 is the dominant volatile ... generated through organic cracking, relative to H2O generation from dehydration reactions ..." (Aarnes et al. 2010). Type I and II kerogen is commonly found in organic-rich shales such as the Irati Formation and may convert up to 95% of the TOC to hydrocarbon. Low oxygen fugacity in organic-rich shales will originate fluid dominated by CH4-H2O rather than H2O-CO2 in contact aureoles (Aarnes et al. 2010). It can be assumed therefore that kerogen will convert mostly into CH4 rather than CO2 in contact metamorphism of shale. CH4 will dominate for TOC contents of >5 wt.%, while H2O will dominate for TOC contents of <1 wt.%. In the Paraná Basin, particularly in the state of Paraná, TOC contents of the bituminous shales of the Assistência Member (Irati Formation) have TOC contents between 10-25% (Pereira 2013).
Basaltic sills are present under the Paraná volcanic province in a large volume. Mariani et al. (2013) concluded from gravimetric modeling that sills up to 10,000 m in integrated thickness are present in the crust below the 1,500 m thick lavas. Extensive drilling (n = 1424 holes) for coalseam evaluation by the Geological Survey of Brazil in 1986 resulted in a clear picture of the number and thickness of sills (Table VIII) injected into the sedimentary basin (southeastern coal-bearing portion). The examination of 835 logs that include the Irati Formation and the Rio Bonito Formation shows that basaltic sills are present dominantly in the Irati Formation, followed by the Rio Bonito Formation and the other formations. The field observation in quarries confirms the dominance of sills in the Irati Formation over the other formations. The fissile nature of the Irati Formation shales seems to have facilitated the injection of the magmas (e.g., Araújo et al. 2004) compared to the dominantly fine-grained sandstones of the other formations. The average thickness of the sills is near 20 m for all formations. The Irati Formation thus emerges as one target for the observation of pyrometamorphism of bituminous shales in the Paraná Basin.
|Formation||Number||Average thickness, m|
The sedimentary rocks of the Paraná Basin contain therefore the necessary large volume of kerogen, spread over the entire basin, as potential generator of methane; part of the kerogen was cracked and a large volume still remains in the sedimentary basin. The large number of basaltic sills injected into the bituminous beds offers additionally the necessary heat for the secondary cracking of kerogen into methane. The newlyformed gas is modelled to migrate upwards unreactive through the sedimentary and the consolidated (cool) volcanic pile. The temperature of the sedimentary rocks, and contained (and ascending) fluids including methane, is estimated at 60 °C above the temperature expected for intraplate basins. This additional heat probably originated in volcanism, mostly from sills. Evidence in this sense is found in the microthermometric studies of primary fluid inclusions in calcite from fractures (Cesário Lange region, São Paulo state) indicating that aqueous fluids were trapped at "70-141 °C, with the highest mode at 105 °C; salinity is between 5.33-0.0 wt.% eq. NaCl" (Sawakuchi et al. 2011). The highest temperature attained is constrained by mineral assemblies and oxygen isotopes at <200 °C. At this temperature, methane is not ignited and flows through the sedimentary layers toward the top; a much higher temperature is required to ignite methane. The autoignition temperature of methane is 580 °C in air, which is the minimum temperature required to ignite the gas in air without a spark or flame being present (Wikipedia). As mentioned by Caron et al. (1999), "the autoignition temperature ... is strongly pressure-dependent and decreases with increasing pressures", but is not expected to decrease down to the temperature present in the Paraná Basin sedimentary rocks below the Paraná volcanic province. In our model, methane gas was capable to migrate upwards through the sedimentary package without spontaneous reaction with oxygen present in pores and in mineral lattices.
As the gas reached the base of the uppermost flow that was still cooling at a temperature around 1000 ° C (without having yet developed columnar jointing) the scenario became entirely different. A coincidence of timing occurred between the processes of methane generation in the Irati Formation and its arrival at the base of the uppermost cooling basalt. According to Aarnes et al. (2010), "the generation of volatiles is occurring on a timescale of 10-1000 years within an aureole of a single sill..." This is the same time-scale of cooling of a basalt lava. In the studies of a 45 m thick basaltic andesite flow from the Paraná volcanic province, Schenato et al. (2003) estimated the time for complete solidification as 35 years, corresponding to cooling processes between the liquidus (1200 °C) and the solidus (1000 °C) of the lava. During cooling, the two solidus curves approaching from above and below intersected at two-thirds distance from the top of the flow; the temperature reached 1000 °C and the flow became solid. This level coincides with the position commonly attributed by many authors for the origin of the segregation melts. Cooling of the 45 m thick lava to ambient temperature probably lasted 560 years (Schenatto et al. 2003).
Only very few thin basalt lavas have pegmatite seams, and these are of the order of millimeters in thickness; we interpret this as due to the faster cooling of the lava. Methane did not ignite as it passed through this thin flow. Thicker (>70 m) lavas and mafic plutonic complexes remain at a high temperature for a longer period of time. Because the vertical cooling joints of thick lava flows begin to form at 1,000-900 °C and do not contain injected paralavas, these must have formed at a temperature >900 °C. In one study, the initiation of surface cracks in a cooling lava lake from Hawaii was observed to occur at 900 °C for a lava emplacement temperature of 1090 °C (Peck and Minakami 1968).
Methane is extremely flammable and is violently reactive with oxidizers (e.g., Wikipedia). In the cooling basalt, little free oxygen (the prime example of oxidizer agent) is present so methane reacts with the oxygen in the mineral lattices in a reaction of "reduced burning". This is a strongly exothermic chemical reaction. The local temperature in the holocrystalline, cooling (1,000 °C) basalt is raised because of CH4 oxidation. Several geological estimates have been made on the temperature attained during methane burning. For instance, Grapes (2011) estimates the temperature in air up to 1,600 °C, whereas Sokol et al. (2010) refer to ultrahigh temperature up to 1,500 °C in ignition foci of sedimentary rocks by combustion metamorphism. A detailed study of mud melting by methane ignition established the temperature at 1,400-1,800 °C, with the resultant formation of glass at 1,400 °C and lower.
We thus model the remelting of the base of the basalt lava core as a consequence of temperature rise from the natural cooling lava at 1,000 °C up to 1,600 °C. At this very high temperature, the affected basalt would melt extensively, possibly 75% liquid with 25% remaining solid. This is the inverse proportion commonly mentioned for the extent of fractional crystallization required to generate the pegmatite sheets in other provinces. The description of segregation sheets by Philpotts et al. (1996), for example, concludes from fractional crystallization modelling that "the composition of the segregations corresponds to liquids that can form by as little as 25% crystallization of the initial basalt." The resultant new magma may be overall similar by the two processes. The mixed rock that remains in place would be composed of a large portion of crystallized new magma and a small portion of remnant, unfused solid. In the terminology of Grapes (2011), this may be called a buchite, which is a partially fused rock by pyrometamorphism.
In a relevant study for the present interpretation, D'Oriano et al. (2013) remelted Etna, Vesuvius and Stromboli basalt tephra in the laboratory and obtained liquids enriched in incompatible elements and SiO2 and impoverished in Mg. In the present study, similar results were observed in EMPA analyses of magnetite, which displays higher contents of TiO2 in the paralavas than in the host basalt. The chemical composition of samples P1 (host basalt), P2 and P3 (reaction rims of paralavas) have similar behavior, including Zr and Y. This observation supports the comparison between remelting in paralavas of Paraná volcanic province with those obtained in the laboratory.
The overall process is shown in the model (Figure 9) that explains the origin of the newlyformed liquid in the lower portion of the lava core. The paralava contains a higher proportion of light elements, so it is more "differentiated" and rises buoyantly through stockworks to the central and upper portions of the core. Dikes and sills of paralava are thus formed in the core of thick basaltic flows. The columnar jointing will be formed later by cooling of the flow below 900 °C, with the paralava already solid with intersertal texture and part of the coarse-grained host basaltic rock.
We envisage the occurrence of similar processes of basalt or gabbro remelting to form paralavas in other continental provinces whenever methane streaming occurred through solid rock portions at 1000 °C and methane autoignition occurred. The recognition of the process is most significant for petrology and economic geology. The presence of CH4 in fluid inclusions (Larsen et al. 1992) was attributed to carbon assimilation from sedimentary blocks. In the fluid inclusions evaluated in the paralavas from the Paraná volcanic province, the NaCl+H2O+CaCl2 system was suggested as a hypothesis. This system can be observed in a wide range of geological environments and also in the sedimentary basins with hydrocarbon occurrences (Steele-MacInnis et al. 2011) such as the Paraná Basin. Overall, the evaluation of this study indicates at least two distinct phases of trapped fluid inclusions. The older generation was trapped at higher temperatures from a higher salinity fluid, which may contain a volatile of low-density, besides single-phase inclusions consisting only of this volatile species. Fluid inclusions of later generation were entrapped under low temperature conditions from low salinity solutions which may be related to the Cretaceous hydrothermal events described by Hartmann et al. (2012a) and superimposed on these rocks. On the other hand, the presence of CH4 in the fluid inclusions in the paralavas of Paraná volcanic province has not been definitively identified because of the small size of inclusions. However, the cooling results indicate that the fluid inclusions may be constituted by a low density fluid such as water vapor, CO2 or CH4 undetectable by microthermometry. Further Raman studies of fluid inclusions with appropriate size may confirm the presence of CH4.
The economic importance of paralavas opens a new frontier for studies related to mineral exploration. This can be illustrated with the copper mineralization in the paralava of Dalba quarry, in which the copper average in the host basalts is 141 ppm (n = 2), slightly below the average (152 ppm) established for the volcanic province (Crocket 2002). On the other hand, in the dark reaction rim, the copper content increases to 236 ppm, reaching 600 ppm in the light reaction rim. This shows that the concentration of metal was related with the paralava generation process. The host basalt presents LOI = 1.6 wt.%, similar to the LOI of the light reaction rim (1.5 wt.%), while the LOI of the dark reaction rim is 2.4 wt.%. The native copper mineralization in the host basalt is associated to the dendrites at the surface of columnar joints and is consistent with the epigenetic hydrothermal mineralization generated during the H1, H2 and H3 Cretaceous hydrothermal events (for details see Hartmann et al. 2012a, b) which promoted hydrothermal alteration of the host basalts and paralavas. The native copper occurrences observed in the paralavas filled microcavities in the light reaction rim and the higher Cu contents are associated with paralavas bodies.
The geological and mineralogical evidence thus far described in pegmatites and some segregation sheets in thick basalt lavas and in intrusions were previously attributed to fractional crystallization, but may be alternatively explained by partial melting process, including the Paraná basaltic province and possibly other similar flood basalt and mafic intrusions in the world. We thus use "paralava" for these geological bodies, as an interpretation of the remelting of the host basalt by the heat generated from combustion of methane confined within a thick, cooling basalt flow or intrusion.
Rocks from the Paraná volcanic province previously designated pegmatites or segregation sheets are presently interpreted as paralavas, the result of melting of hot host basalts by methane combustion. The paralavas generated by remelting (1,600 ºC) of the lower portion of the still-hot (1,000 ºC) core of the host basalt and rose buoyantly to inject the core of the flow. The external source of energy required to raise the temperature was the reaction of methane with the hot minerals (silicate and oxides). The methane was originated deeper in the Paraná Basin from the cracking of kerogen in bituminous shales ( Ponta Grossa, Irati Formation and others ).