Microscopía electrónica de barrido (SEM)

The δ18_{O signatures observed at the mineral scale and in the bulk are the cumulative result of a}

sequence of processes, a general model presenting three stages of metasomatism is presented as shown first in Figure 26. The signatures present in the sedimentary and magmatic protoliths are first modified by the results of seafloor alteration (stage 0), mechanical or chemical mixing of sedimentary and magmatic signatures either by the deposition of volcano-sedimentary layers or by melange-formation processes in the accretionary prism (stage 1), and finally fluid-rock interaction during subduction (stages 2 and 3). These processes will be investigated by discussing first the pre-metamorphic processes (current section), and then the high-pressure fluid circulations (section 5.3).

The origin of protoliths in the Halilbağı unit was investigated in Chapter 3. Based on WR geochemistry and zircon U-Pb dating it was concluded that the Halilbağı unit is made of an assemblage of rock types of different affinity and ages, of oceanic and continental margin origin.

The Halilbağı rocks are geochemically and geochronologically similar to the less-metamorphic oceanic complexes of the Tavşanlı area, which can be related to the Izmir-Ankara ophiolitic complexes. The only oxygen isotope data available for the Tavşanlı zone are not directly relevant for this study as they are on different rock types: magnesite veins in the serpentinite of the ophiolite unit (Kahya 2014), or on the post-continental magmatic rocks (Mutlu 2012). One previous study measured δ18_{O (VSMOW) of 12 to 17 ‰ for the OIB pillow lavas in the Kılıçlar}

locality of the ophiolitic melange of the Izmir-Ankara suture, interpreted as a fore-arc accretionary prism (Gökten and Floyd 2007). In this case, the pillow lavas in prehnite-pumpellyite facies retain a magmatic 87_{Sr /}86 _{Sr ratio at ca. 0.704, which was interpreted as an indication that the pillow-}

lavas were not mechanically mixed with sediments, but rather hydrothermally altered. Such elevated δ18_{O values of 12 to 17 ‰ are at the upper range of what is expected for seafloor mafic}

lithologies altered at low temperatures in the upper part of the oceanic sequence (Gregory and Taylor 1981), and could be due to further fluid circulations in the accretionary prism.

SHS44A and SHB45 magmatic zircon cores provide direct constraints on the magmatic δ18_{O of}

this leucogabbro and alkali basalt: zircon δ18_{O is of 5.2 ± 0.4‰ for SHS44A and 4.5-6 ‰ for}

SHB45 (stage 0 in Figure 4 - 27). These two zircon values correspond to a modelled WR of 5.7 at 6 ‰ respectively, using an empirical formula by Valley (2003). This indicates that at magmatic temperatures, the protolith of these two rocks had mantle-like oxygen composition. These magmatic values can be extrapolated to other rocks of similar magmatic MORB and OIB immobile trace-element signatures identified in Chapter 3: SHB05, SHB08, SHB12B and SHB44B.

In contrast to these mantle-like values, the modelled δ18_{O values for the metabasites based on the}

analysed garnet cores range from 8 to 16 ‰ (following Stage 1 in Figure 4 - 27). The highest modelled δ18_{O values are from the south of Halilbağı (SHS44A and B, modelled WR 17.0 and}

16.3 ‰), followed by rocks from central Halilbağı at 14.4 ‰ (boulder SHB05) see map in Figure 3 - 5. Similarly, in eclogite SHB45 and blueschist SV01-75 garnet cores yield modelled WR at 13.3 ‰ and ca. 11.5 ‰ respectively. The garnet cores in eclogite SHB12B provide the lowest modelled WR value of ca. 8‰, which still is far from the mantle value of ~ 5.5‰. Garnet is not a magmatic mineral, but grows early in the metamorphic evolution and thus its core δ18_O

composition is likely to represent the WR value at the early stage of metamorphism. These 8‰ to 16 ‰ signatures are interpreted as reflecting variable early alteration of the metabasalts (see also Putlitz et al. 2000).

Figure 4 - 26. a. Schematic representation of fluid pulses documented in the Halilbağı unit with regard to the subduction zone. b. P-T path with indications of fluid-rock interaction stages. The P-T

field for the Accretionary units OC1 and OC2 and the main reactions are from Plunder et al. (2015), the P-T field for the Halilbağı unit main assemblages is a compilation of Çetinkaplan et al. (2008)

and Davis and Whitney (2006,2008).

In summary, relicts of magmatic oxygen isotope signatures in Halilbağı rocks only present in rare zircon cores in SHS44A and SHB45; mantle-like signatures that were likely original in all the MOR and OIB mafic rocks were obliterated before the crystallisation of the high-pressure (blueschist or eclogite) assemblage. The result is an elevated δ18_{O between 11 and 17 ‰ for all}

measured WR of mafic samples (Figure 4 - 27). The measured values correspond to the higher end of seafloor altered basalts, similar to the Gökten and Floyd (2007) Izmir-Ankara suture prehnite-pumpellyite pillow basalts. Ophiolites and oceanic drilling studies (Muehlenbachs and Clayton 1972; Gregory and Taylor 1981; Staudigel et al. 1995) show that heavy values δ18_{O in}

basalts can be attained by circulation of seawater at surficial levels where temperature is low and can reach even higher δ18_{O when sediments are present (e.g. Fouillac and Javoy 1988). Oceanic}

alteration, especially in the presence of sediments, is thus a likely process for the enrichment in

18_{O (stage 0 in Figure 4 - 26a). An alternate setting for the enrichment of} 18_{O in Halilbağı}

metabasites is the accretionnary prism and early (pre-garnet) subduction (stage 1 in Figure 4 - 26a, b). An enrichment trend has previously been observed by Bebout and Barton (1989) and Miller et al. (2001) in the WR of metabasites in sediment-rich subducted terranes of the Catalina Schist and Corsica, respectively. Miller et al. (2001) found a large span of δ18_{O values for metabasites}

and meta pillow-lavas, from 6 to 16 ‰, which they ascribe to hydrothermal alteration, with minor modification from subduction-related fluids, especially present at lithological interfaces (Martin et al. 2011; Vitale Brovarone et al. 2014). Bebout and Barton (1989) found various degrees of homogenisation for metasedimentary and metabasitic garnet-free 400°C blueschists, which δ18_O

The serpentinite SHS26 also yields heavy δ18_{O with a measured WR of 13.8 ‰, which is}

consistent with its main mineral antigorite, measured at 13.4 ‰. These serpentinite pods have previously been interpreted as mantle wedge slices (Whitney et al. 2014), in which the high δ18_O

value would imply that mantle wedge was serpentinised by sedimentary fluids with high δ18_{O. A}

simpler explanation is that these lenses represent oceanic serpentinites such as present in lower grade oceanic complexes (Plunder et al. 2013) that equilibrated with seawater at low temperatures or sediments in the accretionary prism, similarly to the metabasites. The latter hypothesis is supported by the REE and trace element signature of this rock that resemble that of an olivine- rich cumulate, in line with an oceanic protolith (see Chapter 3) and not a mantle wedge peridotite protolith.

Figure 4 - 27. Overview of measured WR δ18_{O and value for water equilibrated with mineral zones at} 500°C, following fractionation coefficients by Zheng (1993a, 1996) at 500°C. Straight ties indicate core-rim relationship in a single sample.

WR REE and trace-element patterns (see Chapter 3) indicate that intermediate lithologies SHB53, SIB50B, SIB32 are likely a mix between volcanic and sedimentary components (e.g volcanoclastic). These samples yield a measured WR δ18_{O between 12.2-13.7 ‰, which is within}

the range of δ18_{O measured in rocks that yield magmatic-like REE patterns and magmatic relicts.}

Similar to purely magmatic metabasites, heavy signature can be traced back to at least early garnet growth in SIB50B and SV03-103, SV12-13F, where garnet cores yield modelled WR δ18_{O of ca}

13 ‰, 11 ‰ and 15 ‰ respectively. The intermediate lithologies studied here are similar in δ18_O

to the pillow breccias of Miller et al. (2001), that yield a δ18_{O of 12-15 ‰, values that are higher}

than the pillow-lavas in the same sequence.

The metasediments co-deposited with these metabasites yield a range of elevated values, from 18.9 to 22.5 ‰ with the exception of the Günyüzü marble SGM21 which yields a higher value at 26.5 ‰. These values are similar to oceanic sediments analysed by Bebout (1991) in the Catalina schist, or the sediments in the Corsica HP ophiolite analysed by Miller et al. (2001). As sediment

δ18_{O signatures can span a wide range in un-metamorphosed sequences, these rocks do not}

provide additional information on the pre- or early-subduction hydrothermal phase.