Évolution spatio-temporelle de l’hydrothermalisme dans la plaque supérieure de l’arc des Petites Antilles en Guadeloupe

Mineral equilibria modelling

 First of all, and before any quantitative analysis of the P-T conditions of formation of Terre-de-Haut hydrothermal system, it must be specified that a mineral association involving the co-stability of chlorite+quartz+epidote+muscovite+albite+titanite+biotite+/-actinolite, and pyrite is stable, in volcanic rocks transformed in hydrothermal and geothermal systems, for a minimal temperature of about 300°C (Henley & Ellis, 1983; Rancon, 1983; White & Hedenquist, 1995; Lagat, 2014).

Chlorite geothermometry 

Chlorite is a common phase in low-grade metamorphic rocks, and Cathelineau (1988) first proposed a geothermometer based on chemical composition of chlorite. Later on, more elaborated versions of this thermometer have been proposed, that are particularly well adapted for hydrothermalized rocks (Lanari et al., 2014; Bourdelle & Cathelineau, 2015). Following these recent investigations, for Sipoor and Al-rich chlorites, the more appropriate calibration is the one proposed by Lanari et al. Chapitre V 116 (2014), while for Si-rich chlorites the calibration proposed by Bourdelle and Cathelineau (2015) is more adapted. Because chlorite is stable over a wide temperature range in geothermal systems (< 100°C to > 350°C, Henley & Ellis, 1983; Reyes, 1990; White & Hedenquist, 1995; Lagat, 2014), it is critical to distinguish the possible variations of chlorite compositions with respect to their various microstructural crystallization sites. Indeed, we found the occurrence of chlorite in schistose corridors as well as in late (i.e.post-schistosity) veins or fractures. Out of the main schistose corridors, pale green chlorites are frequently observed, in association with low temperature minerals, within micro-cracks or vacuoles. Figure V-15: Chlorite morphologies. A) Ferromagnesian phenocryst pseudomorphosed by fibrous muscovite and grey automorphous crystals of chlorite (type I), in a meta-rhyodacite rock. Sample from north of Grande Anse beach (16STH19). B) Ferromagnesian phenocryst pseudomorphosed by grey automorphous crystals of chlorite (type I) and green-grey to yellow fibrous grains (type II), in a metadacite rock. Sample from north of Grande Anse beach (17TH16). Evolution des systèmes hydrothermaux 117 Table 

 Microprobe analysis and chlorite thermometry on three samples from north of Grande

 Anse beach: 16STH19 (meta-rhyodacite), 17TH16 (meta-dacite) and SA-14 (meta-andesite). Chapitre V 118 In order to quantify the maximal temperature suffered by the Terre-de-Haut hydrothermal system, we selected three samples from north of Grande Anse beach (16STH19, 17TH16, SA-14), thus in a significantly deformed domain, with well-developed schistose corridors. We also selected this area because it is a domain without evidence of supergene alteration (Figure V-3-B and Figure V-14). Moreover, in these selected samples chlorite developed in different microstructures as (1) grey automorphous crystals (Figure V-15-A), associated with muscovite, quartz and epidote, developed at the expense of magmatic pyroxenes, (2) green-grey to yellow fibrous grains (Figure V-15-B) associated with fibrous white micas, quartz, calcite, pyrite and/or talc, preferentially located at the rims of pseudomorphosed pyroxenes. Interestingly, chlorite compositions evolve in relation with their microstructural position (see Table V-2), and even the type (1) grey automorphous grains display, in many cases, composition zoning from cores to rims. An increase of Si and a decrease of Al content is depicted from type (1) automorphous to type (2) fibrous chlorites. Cores compositions of type (1) chlorites are Si-poor, thus we used the Lanari et al. (2014) calibration and this yield a temperature range of 255-380°C, with a calculated average temperature, taking into account the different samples, of 355°C (standard deviation=27). The rims of type (1) chlorites have higher Si-contents, thus in that case, and with the assumption that FeTotal = Fe2+, the Bourdelle and Cathelineau (2015) calibration is the most adapted. With this calibration the calculated temperatures range is 180-350 °C and the calculated average temperature, considering the different samples, of 252°C (standard deviation=57). Type (2) fibrous chlorites have also high Si-contents, thus using the Bourdelle and Cathelineau (2015) calibration we obtained a temperature range of 115-250°C with a calculated average temperature, considering the different samples, of 175 °C (standard deviation=33). All together, chlorite thermometers provide coherent estimation of the progressive cooling registered by the various generations of chlorites during their hydrothermal evolution through time.

Pseudosection modelling

 The intensive parameters (pressure and temperature) corresponding to the high temperature hydrothermal stage, were calculated with the free energy minimization program Theriak-Domino (de Capitani & Petrakakis, 2010, updated software v. 4 February 2017) and the internally consistent thermodynamic “tcdb55c2d” database (Holland & Powell, 2004). Mixing models used for solid solutions are presented in appendix (Table X-5). In order to be as close as possible to the equilibrium conditions at the scale of the whole rock, which are essential for the application of this type of method, we selected a highly transformed sample with a maximum of high temperature new phases expressed, but without mineralogical evidence of Evolution des systèmes hydrothermaux 119 supergene alteration. The best candidate for pseudosection modelling is, in such a case, a sample of rhyodacite (16STH19) selected from the most hydrothermalized area (north of Grande-Anse beach). P-T pseudosections were calculated in the system SiO2-Al2O3-TiO2-FeO-MnO-MgO-CaO-Na2OK2O-H2O, using the whole rock composition of the selected sample (Table V-3). The observed mineral assemblage involves albite + chlorite + muscovite + biotite + clinozoisite + epidote s.s. + quartz. Under tropical conditions, the iron oxidation state is always difficult to estimate. In order to test the effect of Fe2O3 in our models we calculated pseudosections with various amounts Fe3+. When the later reaches 5%, we never find the mineral assemblage observed in the natural sample. Consequently, we consider that a model with total iron considered as FeO is the best approximation for pseudosections calculation. Table V-3: Whole-rock composition of 16STH19 sample and corresponding input in Theriak-Domino program. We considered the fluid phase in excess to be consistent with the conditions of a fluid-saturated hydrothermal system. However, in high-enthalpy geothermal systems, fluids have different origins (supergene waters, with or without sea-water component, volcanic –derived fluids and gases, with variable amounts of CO2 or SO2), it seems thus critical to test the effect of fluid composition on the modeled system. In a first approach, a pseudosection was calculated with 100% of H2O for the fluid composition. In this case, the observed mineral assemblage corresponds to the calculated field (360- 375°C and 0.90-1.40 kbar) identified with bold-black characters in Figure V-16-A and is in agreement with the highest temperature values obtained by chlorite-geothermometry on type-1 chlorite cores obtained on this sample (16STH19, see Figure V-16-B and Table V-2). A temperature of about 365°C Chapitre V 120 Figure V-16: Mineral abbreviations are from Kretz, 1983. A) Calculated P‐T pseudosection for meta‐ rhyodacite sample (16STH19) with H2O saturation and total iron as Fe2+ with Theriak-Domino program (de Capitani & Petrakakis, 2010, updated software v. 4 February 2017) and “tcdb55c2d” database (Holland & Powell, 1998, 2004; updated Nov. 2003). The rock composition, given as mol.% oxide. Fields are colored with respect to their variances, darker colours indicate higher‐variance assemblages. The stability field written in bold correspond to the observed assemblage. B) Simplified representation of P-T pseudosection with the stability field in grey and the main geothermal gradient consistent with regional thermal regime (Manga et al., 2012; Verati et al., 2018; Favier et al., 2019). C) P-X(H2O) equilibrium phase diagram calculated at 365°C with the same composition in Figure 18-A, performed between 0 (H = 0) and 100 moles (H = 100). The field colored in grey corresponds to the observed assemblage. seems a coherent estimation to remain consistent with both methods. In a second time a specific PXH2O pseudosection (with XH2O representing the variability of amount of water in the considered system) was calculated in order to decipher the effect of the amount of H2O, in the fluid present in excess, on the pressure estimates (Figure V-16-C). The observed mineral assemblage is stable for a large range of XH2O values, i.e. between 12% and 100%. Moreover, in this specific sample, the chemical components of the LOI were analysed in the SARM Laboratory in Nancy (see Table V-3). This kind of analysis allows an estimation of the components in the fluid phase with respect to the bulk-rock chemistry of the analysed sample. In the 16STH19 sample, the amount of H2O, in the fluid phase, is estimated at around 59%, with a CO2 content of about 27%. The result of our thermodynamic modeling is totally consistent with this estimation, and indicates that regardless of the XH2O value, Evolution des systèmes hydrothermaux 121 if it is greater than 12%, the observed mineral assemblage remains stable. However for the XH2O values of interest, and for a temperature of about 365°C, the possible pressure range is restricted between 0.95 and 1.15 kbar. Indeed, in such a chemical system, for pressures lower than 0.9 kbar wairakite, commonly described in various geothermal fields, will be systematically present in the mineral association. Altogether, the combination of chlorite thermometry and thermodynamic modeling yields to maximal ranges of P-T conditions of 360-375°C and 0.90-1.40 kbar (i.e. Greenschist facies conditions) for the development of the observed high temperature mineral assemblage in the metarhyodacite from the most hydrothermalized area of Terre-de-Haut Island. 

 Discussion

Relationship between deformation and hydrothermalism: implications for fluid transfer pathways In the present day situation we observed in the field a system of brittle faults and in a general crosssection of Terre-de-Haut Island (Figure V-17), the main hydrothermal zone is localized within a WNW-ESE graben compatible with the regional transtensional context as proposed by Feuillet et al. (2001, 2002) and Leclerc et al. (2014, 2016). Figure V-17: Cross-section NE-SW from Terre-de-Haut Island displaying brittle-ductile transition with faults network and schistosity planes, with the central part the highly hydrothermalized area in the WNWESE graben. However, the comparison between the structural and the mineralogical maps (Figure V-8 and Figure V-14) demonstrates that high-temperature hydrothermal phases are preferentially developed along the discontinuous and spaced cleavages concentrated in schistose corridors, indicating that their crystallization is contemporaneous with the development of such tectonic structures. As illustrated by the regional cross section (Figure V-17), brittle faults clearly crosscut the schistose corridors. The observed schistosity planes being generated by pressure solution processes, the schistose corridors are markers of the circulation of high-temperature hydrothermal fluids. As shown in the detailed structural map (Figure V-8) and the regional cross-section (Figure V-17), these schistose domains show very specific geometries. In some cases, for example east of Fond Curé, schistose zones tend to anastomose while in other domains, like west of Morne Rouge, sub-vertical schistose corridors clearly intersect other moderately dipping zones. However, and whatever the geometrical pattern is, in this hydrothermalized area, the primary volcanic structures are severely erased. In such a case we can propose, as a working hypothesis, that the poorly dipping schistosity planes developed at the expense of pre-existing, sub-horizontal, volcanic structures, particularly magmatic fluidality and lithological interfaces. Following this interpretation, the least dipping schistose corridors must be regarded as the result of tectonic reactivation of volcanic structures. On the other hand, the most dipping schistose corridors are neo-formed tectonic structures and they evolve, during progressive cooling of the studied area, from ductile to brittle structures. However, and whatever the mechanism of formation of schistose zones is, the observed finite structural pattern is particularly favourable to both vertical and lateral fluid transfers if the schistose zones are connected. Through time, during the thermal evolution of the hydrothermal system, low-temperature hydrothermal phases continue to crystallize in schistose zones but also in fractures, veins and brittle faults. This means that during the cooling history of the hydrothermal system, tectonic structures continue to be preferential drains allowing the circulation of fluids. Moreover, in this specific area a recent study (Navelot et al., 2018) revealed also fluid transfers at the lithological interfaces between alternating lavas and debris flows. In these rocks, mineral transformations have a significant impact on their petrophysical properties, and the pyroclastic and debris flows are suitable horizons for storage and fluids flow.

 Terre-de-Haut hydrothermal domain: an exhumed paleo-geothermal system?

 A geothermal system represents a thermal anomaly, thus a significant deviation with respect to the regional standard conductive geothermal gradient. In the Guadeloupe archipelago, recent study (Manga et al., 2012) has revealed a steady-state conductive gradient between 69.3 ± 1.5 and 98.2 ± 8.8 °C/km. Furthermore, in the Basal Complex from Basse-Terre, thus in the most eroded piece of the active arc, arc-related metamorphism, dated between 4 and 2 Ma, has been recently evidenced, compatible with paleo-conductive geothermal gradient in the order of 70-100°C/km (Verati et al., 2018; Favier et al., 2019). Therefore, this type of conductive gradient remained stable during the last 4 Ma. In contrast to these regional standard values, in the Bouillante geothermal field, temperatures in the range of 230- 255°C are measured at depths between 475 m and 1.2 km (Sanjuan et al., 2001, 2004; Guillou-Frottier, 2003; Mas et al., 2006; Bouchot et al., 2010). In the highly hydrothermalized zone from Terre-de-Haut Island, we discovered the occurrence of high-temperature hydrothermal mineral parageneses that are stable under a temperature around 365 °C and very low-pressure conditions (Figure V-16-A-C). These conditions represent a significant thermal anomaly with respect to the regional standard geothermal gradient. Moreover, some of the observed mineral Evolution des systèmes hydrothermaux 123 associations as well as the microstructures of the reaction sites (i.e. progressive and/or complete replacement of pyroxenes by epidotes, chlorites, white micas, …) are similar to the ones described in hydrothermal rocks from the Bouillante well (Mas et al., 2006). But, the temperatures recorded by the Terre-de-Haut fossil hydrothermal system are higher than the temperatures proposed for the active Bouillante field. However, they are compatible with temperature ranges measured in drillings in different geothermal fields around the world (Browne, 1978; Del Moro et al., 1982; Battaglia et al., 1991; White & Hedenquist, 1995; Mas et al., 2006; Bogie et al., 2008; Bouchot & Genter, 2009). In addition to bearing the mineralogical traces of a strong thermal anomaly, the studied area is all the more comparable to a geothermal reservoir as it is a place of strong circulation of huge amount of hydrothermal fluids. These fluids are not exclusively composed with H2O, but a minimal value of 12 moles of H2O in the fluids is necessary to produce the observed hydrothermal mineral assemblage (Figure V-16-C). High-temperature hydrothermal fluids circulation is effective enough to totally erase the primary volcanic structures, and as discussed in the previous section, tectonic structures constitute efficient drains for hydrothermal fluids circulation. All together, these data support the idea that the Terre-de-Haut hydrothermal domain constitutes an eroded, and thus exhumed, paleogeothermal system. The latter can be used as an analogue of the deepest parts of active geothermal systems in the Lesser Antilles volcanic arc and can therefore serve to better constrain their operation modes.