Espectroscopía Infrarroja por Transformada de Fourier (FTIR)

In order to relate mineral δ18_{O composition to bulk-rock δ}18_{O, and assess if some of the mineral}

variations are purely temperature-related fractionation along the T variation experienced by the rocks, simple models of δ18_{O fractionation within the rock assemblage are presented for the main}

rock types. These models aim at answering questions such as: is a measured mineral in equilibrium with the measured WR of the sample? Is the core-rim variation measured in a mineral attributable to temperature changes in a closed system? Or does it suggest an open system resulting in a change in WR δ18_O?

As no P-T pseudosections were produced during this work, the effect of phase changes is not taken into account here, to not add additional sources of uncertainty. These simple models mostly reflect the fractionation related to the overall rock content in SiO2, and the variation in SiO2

content between coexisting phases, as SiO2 forms a strong bond that retains heavy oxygen

preferentially. Effects linked to specific mineral structures (e.g. reaction with albite, see also effects of reactions modelled in Dora Maira Whiteschists, Chapter 1) are usually secondary in magnitude and are not examined here. Specifically at temperatures outside of the stability field of the assemblage preserved in the thin section, the preserved assemblage is used as a proxy for the “true” assemblage as its chemical composition is identical and thus the fractionation behaviour similar. Also, as is expanded in the discussion, each mineral is considered open for δ18_O

exchanges in the temperature range (400-600°C). This is likely not the case on the retrograde path as investigated by previous studies involving more sophisticated models involving diffusion (Eiler et al. 1992), but is probably a good assumption during prograde reactions when most of the assemblage recrystallizes. Moreover, diffusion might impact the signature recorded by the measured minerals. Garnet and zircon retain high-temperature signatures for oxygen (Watson and Cherniak 1997; Vielzeuf et al. 2005b), it thus gives a reliable record of the reactive bulk. Less is known about lawsonite and apatite.

These models are based on the modes of major minerals observed in thin section, and are constructed using the fractionation coefficients of Zheng (1993a, 1993b, 1996). As coefficients for lawsonite are not available, epidote is used as a proxy because of its similar composition (as done by Martin et al. 2014b. reproducing measured whole-rocks successfully). The inter-mineral fractionation is calculated, and then the relative values are adjusted either to a measured mineral composition or to the measured bulk-rock to obtain absolute values. More details about the modelling procedure can be found in Gauthiez-Putallaz et al. (2016) and in Chapter 5. The table of mineral modes used for the calculation is available inAppendix table A4 – 14.

First, constant WR models were constructed for three main rock types (calcsilicate metasediment, mafic and serpentinite, Figure 4 - 22) for a temperature range that overestimates the stability of minerals, allowing to estimate the maximum fractionation difference. In a mafic eclogite, the temperature-related fractionation is minimal as all minerals have similar SiO2 contents; see also

Russell et al. (2013). In this case, the strongest variation in δ18_{O over the T range can be expected}

from garnet and will be <0.5 ‰ during the temperatures where garnet is stable (between 450 and 600°C according to P-T estimates by Davis and Whitney, 2006 and Cetinkaplan et al., 2008). For apatite and epidote, the variation over the chosen temperature range is <0.5 ‰. In this rock, apatite and epidote marginally decrease in δ18_{O from 400 to 700°C, inversely to garnet. In the}

metasediment SHS27, due to the presence of 30% quartz and 30% calcite that concentrate heavy δ18_{O, the expected fractionation is much larger. Hypothetical garnet crystallising in this}

metasediment would see an increase in its δ18_{O value of 2 ‰ from 400 to 700°C. Similarly,}

increase in apatite and epidote δ18_{O of around 1 ‰ with increasing temperature can be expected}

in the calcite-bearing metasediment. The serpentinite model highlights that the silicate, oxide and carbonate fractions of the rock would yield radically different δ18_{O values at equilibrium (more}

than 10 ‰ spread at 400°C, 5 ‰ at 700°C), although the antigorite would only decrease ~1 ‰ in δ18_{O over the T range, as it represents the most abundant mineral. To sum up, in quartz-poor}

eclogites, δ18_{O changes related to temperature are expected to be minimal in garnet, apatite and}

lawsonite, which indicates that core-rim zoning present in metabasite garnets (SHB12B, SHB45, SV01-75 and SV12-13F) and SHB12B apatite cannot be attributed to temperature and are thus the sign of an open system. In quartz or calcite-rich metasediments, changes of up to 2 ‰ are expected in garnet, and up to 1 ‰ in apatite and lawsonite, zoning observed in SHS03 apatites is of ca. 2.5 ‰ and is thus also likely the sign of an open system. Only antigorite was measured in serpentinite, and no more than 1 ‰ variation in its mineral value is expected with temperature change.

Figure 4 - 22. Modelling of equilibrium fractionation between metamorphic phases at constant WR, using assemblage and modes as observed in thin section. Fractionation coefficients used are from

Zheng 1993a; Zheng 1993b; Zheng 1996. Phase stability is overstated in these diagrams, with the

aim of providing maximal changes in δ18_{O according to temperature.}

Table 4 - 3. Summary of WR recalculations based on equilibrium fractionation. Magmatic WR recalculated from zircon was produced using the relationship by Valley 2003.

Sample

Measured values Modelled WR

zone zircon apatite garnet

apatite zircon garnet WR mag 400°C- 500°C 400°C - 550°C SHB05 15.3 13 15.4 14.1 - 14.3 14.4 - 14.0 SHB08 15.7 15.4 14.8 - 15.0 SHB12B core 10.0* 6.3 11 8.7 - 8.9 7.7 - 7.3 rim 15.0* 13.0* 13.7 - 13.9 14.3 - 14.0 SHB45 core 16 5* 12.2 15.6 5.4 14.5 - 14.8 13.4 - 13.1 rim 13.9 15.1 - 14.8 SHS44A core 16.2 5.3 14.6 16.9 6.1 15.8 - 15.9 17.0 - 16.3 rim 12.0* SHS44B 16.6 14.7 17 15.5 - 15.7 16.3 - 15.9 SHB53 12.7 12.2 11.5 - 11.7 SIB32 14.1 12.2 13.3 - 13.4 SIB50B 14.2 11.5 13.7 13.4 - 13.5 13.4 - 12.9 SGM21 26.1 26.5 28.1 - 27.7 SHM23B 15.9 19.9 18.9 -18.3 SHS27 20.3 19.9 22 - 21.7 SHS3 core 19.5* 14.6 20 21.7 - 21.3 19.5 - 18.2 rim 17.0* 18.2 - 17.8 SIS52 20.2 22.5 22.9 SIS53 16.3 18.9 18.3

* representative value for a chosen mineral zone (not an average of all analyses, non-normal distribution) Values deviating by more than 1.5 ‰ from the measured WR are indicated in bold.

Second, the hypothetical WR in equilibrium with measured mineral values are modelled and presented in Table 4 - 3. In this model, mineral modes are assumed to be as observed in the thin section for each sample, and temperature is set to values that are approximate brackets to the crystallisation interval for the given mineral. As the temperature profiles show, this procedure is

more reliable in metabasites where the uncertainty related to the assumption made for temperature is small. In metasediments, the uncertainty of a garnet-based calculation is larger, and the modelled value has to be taken as an indication only if the temperature of equilibration cannot be constrained.

The main results are that in most samples, the modelled WR from each measured mineral is similar to the measured WR within the estimated uncertainty of ca. 1 ‰. This indicates that in these samples, there is no indication for oxygen isotope disequilbrium. For a subset of samples that show mineral zoning (SHB12B, SHB45, SHS03, SHS44A), this modelling reinforces the hypothesis of an open-system: the largest contrasts are present in eclogite SHB12B, with a WR modelled of 7.3-7.7 ‰ and 14.0-14.3 ‰ from the measured garnet cores and the garnet rims respectively, that are both distinct of the measured WR value of 11.0 ‰. Hints for an open-system are also identified in SHS27, where the WR values modelled from apatite at 21.7-22.0 are higher than the measured WR value at 19.9 ‰, however this is subject to a larger uncertainty due to the magnitude of fractionation in quartz and calcite-rich rocks.

5 Discussion

The measurement of in situ oxygen in metamorphic minerals allows uncovering several steps of the fluid evolution of the Halilbağı rocks from their oceanic protoliths to peak P and T at about 23 kbar, 550°C, and the start of exhumation. The aim of this study is to combine the records present in zircon, garnet, apatite and lawsonite to track δ18_{O changes along the subduction and}

exhumation paths. The oxygen isotope or elemental zoning in metamorphic minerals is discussed below for P-T and fluid composition indications. Any relict of protolith signatures is discussed considering also to what extent these signatures are altered at low T before the growth of the main metamorphic assemblage. A fluid history is assembled with information gathered across minerals and rock types. This approach allows reconstructing which rock types in the Halilbağı unit were affected by HP fluid events, and gives indication of what the fluid source and composition might have been. Finally, the effect of fluid-rock interaction on WR composition is explored.