Sulfur isotope behavior during metamorphism and anatexis of Archean sedimentary rocks: A case study from the Ghost Lake batholith, Ontario, Canada

Recycling of surface-derived sulfur into the deep earth can impart distinct sulfur isotope signatures to magmas. The details of sulfur transfer from sedimentary rocks to magmas (and ultimately igneous rocks) through metamorphism and devolatilization and/or partial melting, however, is difficult to trace. To understand this process in detail we studied multiple-sulfur isotope compositions of sulfides in the Archean (c. 2685 Ma) Ghost Lake batholith (GLB) and its surrounding host metasedimentary rocks of the Superior Craton (Ontario, Canada) by high spatial resolution secondary ion mass spectrometry, complemented by high-precision gas source isotope ratio mass spectrometry measurements. The GLB comprises strongly peraluminous biotite+cordierite, biotite+muscovite, and muscovite+garnet+tourmaline granites to leucogranites, which are thought to represent the partial melts of surrounding metagreywackes and metapelites. The metasedimentary rocks display a range of metamorphic grades increasing from biotite-chlorite (280-380 degrees C) at similar to 5 km away from the GLB to sillimanite-K-feldspar grade (similar to 660 degrees C) immediately adjacent to the batholith, thus providing a natural experiment to understand sulfur isotope variations from low- to high-grade Archean sedimentary rocks, as well as granites representative of their partial melts. We find that metasedimentary sulfide delta S-34 values increase with progressive metamorphism at most 2-3 parts per thousand (from -1 parts per thousand up to +1 to +2 parts per thousand). An increase in delta S-34 values in pyrrhotite during prograde metamorphism can be explained through Rayleigh fractionation during pyrite desulfidation reactions. Pyrite from all but one of the granite samples preserve delta S-34 values similar to that of the high-grade metasedimentary rocks, indicating that partial melting did not result in significant fractionation of delta S-34. The exception to this is one granite sample from a part of the batholith characterized by abundant metasedimentary inclusions. This sample contains pyrite with heterogeneous and low delta S-34 values (down to -16 parts per thousand) which likely resulted from incomplete homogenization of sulfur between the granitic melt and metasedimentary inclusions. Small (several tenths of a permil), mostly positive Delta S-33 are observed in both the metasedimentary rocks and granites. Our results suggest that Archean strongly peraluminous granites could be a high-fidelity archive to quantify the bulk sulfur isotope composition of the Archean siliciclastic sediments. Further, our findings indicate that subduction of reduced sulfur-bearing sediments in the Archean with delta S-34 at or near 0 parts per thousand should result in release of sulfur-bearing fluids in the mantle wedge with similar values (within a few permil). S-MIF (if initially present in Archean surface material) may be preserved during this process. However, the absence of S-MIF in igneous rocks does not preclude assimilation of Archean sedimentary material as either S-MIF may not be originally present in the Archean sedimentary sulfur and/or homogenization or dilution could obscure any S-MIF originally present in assimilated Archean sediments. (C) 2020 Elsevier B.V. All rights reserved.