Uncovering and quantifying the subduction zone sulfur cycle from the slab perspective

The study addresses how sulfur is transported and transformed in subduction zones, focusing on the speciation, flux, and isotopic composition of sulfur released by dehydrating slabs. Prior observations of high sulfur contents and positive δ34S in some arc magmas have been attributed to slab-derived sulfate, implying oxidation of the mantle wedge. However, high-pressure rocks rarely contain sulfate, and experiments often indicate reduced sulfur species dominate slab fluids. The purpose is to directly constrain the redox state, concentrations, and δ34S of sulfur in slab-derived fluids by examining exhumed high-pressure rocks and associated veins from SW Tianshan, thereby clarifying slab-to-arc sulfur transfer and its role in arc magma redox and sulfur isotope signatures.

• Arc lavas, particularly in parts of the Western Pacific, can exhibit high sulfur concentrations (up to 3000 µg g−1) and positive δ34S (+5 to +11‰), which have been interpreted as slab-derived sulfate addition via fluids. Some deep arc cumulates (e.g., Eastern Pacific) show mantle-like δ34S, suggesting limited slab sulfate input and potential crustal assimilation to explain positive δ34S.
• Experiments suggest slab-derived aqueous fluids efficiently transport sulfur and may be sulfate-dominated; yet high-pressure rocks rarely preserve sulfate and other experiments indicate reduced sulfur species dominate under slab conditions.
• In situ δ34S in HP eclogites and serpentinites reveal isotopic heterogeneity, pointing to complex sulfur behavior during metamorphism and metasomatism.
• Subducted sulfur reservoirs include sediments, altered oceanic crust (AOC), and serpentinites; HP veins provide fossilized fluid pathways recording fluid chemistry. These contrasting lines of evidence motivate a field- and thermodynamics-based reassessment of sulfur speciation and flux from the slab.

Field and samples: 10 high-pressure (HP) rocks (2 metapelites, 5 metabasites, 3 serpentinites) and three sulfide-bearing HP veins (Vein_1 JTS; Vein_2 L1422; Vein_3 L1013) from the SW Tianshan (China) (U)HP/LT belt were studied. Veins formed at eclogite-facies conditions (~510–550 °C, ~2.1–3.0 GPa) and likely represent fluids released at 70–90 km depth.

Petrography and microanalysis: Sulfide occurrences and mineral associations were documented. Co-Ni element maps of pyrite were obtained by electron microprobe (CAMECA SXFive FE; 20 kV, 100 nA). In situ trace elements in sulfides were analyzed by LA-ICP-MS.

Bulk-rock sulfur geochemistry: Whole-rock sulfur content ([S]WR) and δ34SWR were determined by sequential extraction of acid volatile sulfide (AVS), chromium reducible sulfide (CRS), and sulfate, followed by isotope ratio mass spectrometry (Thermo MAT 253 with EA). [S]WR was calculated by summing S in AVS, CRS, sulfate; δ34SWR by mass-weighting the fractions.

In situ sulfur isotopes: SIMS δ34S measurements of sulfides were done on a Cameca IMS 1280 (NORDSIM, Stockholm) and at IGGCAS (Beijing), using a 133Cs+ primary beam over 10×10 µm rasters. Results referenced to V-CDT.

Thermodynamic modeling (DEW): Sulfur concentrations and speciation in slab fluids ([S]fluid) were computed using the Deep Earth Water (DEW) model coupled with EQ3 across a subduction thermal gradient at 60, 75, 90, 120, 150 km (2–5 GPa; 400–800 °C). Oxygen fugacity (relative to FMQ) was assigned by lithology and depth (AOC/metasediments from FMQ+1 at shallow to FMQ−3 deep; serpentinite 1–2 log units higher at corresponding depths). Modeled aqueous species include H2S, HS−, SO4^2−, HSO4−, and complexes; equilibrium with pyrite/pyrrhotite imposed.

Isotope fractionation modeling: fO2–pH diagrams were computed to evaluate sulfur isotope fractionation among species and between fluid and pyrite at relevant P–T–fO2–pH, and to compare closed- vs open-system precipitation at 550 °C.

Global mass-balance: Sulfur influx (FS) into subduction zones was estimated using global trench length, convergence rate, slab stratigraphy thickness/densities, and average [S] and δ34S of sediments, volcanic layers, sheeted dikes, gabbro, and serpentinite. Sulfur outflux (fS) was calculated as [S]fluid × slab water flux (depth-dependent H2O release) across 30–230 km. Net δ34S of slab-released fluids was computed by flux-weighting lithology-specific δ34Sfluid, with adjustments for limited fluid–rock isotopic exchange during ascent.

• Sulfur in slab-derived fluids is dominated by reduced species (H2S and HS−) at sub-arc depths; sulfate is rare in HP rocks, occurring only in late retrograde fractures.
• Distinct δ34S signatures by reservoir: metasediment-derived fluids ~ −8‰; altered oceanic crust ~ −1‰ (MORB-like to slightly negative); serpentinite-derived fluids ~ +8‰.
• Bulk-rock data: metapelites [S]WR 1101–5612 µg g−1, δ34SWR −7.2 to −7.9‰; metabasites [S]WR 841–3978 µg g−1, δ34SWR −7.2 to +3.6‰ (avg −2.7‰); serpentinites [S]WR 124–422 µg g−1, δ34SWR +3.6 to +12‰; S6+ fractions generally <6% of S.
• Vein transects record fluid–rock interaction: in Vein_1, [S]WR increases toward the vein (to 2183 µg g−1) while δ34SWR decreases to −0.98‰; vein sulfides ~ −1‰. In Vein_2, vein [S]WR 9251 µg g−1, δ34SWR −0.7‰; host blueschist δ34SWR −11‰; pyrite shows multiple growth generations with δ34S from +8.6‰ (cores) to −6.7‰ (rims). Vein_3 pyrite cores ~ −8‰ with thin rims ~ −5‰; host eclogite pyrite cores MORB-like (−1.3 to +0.5‰) overgrown by −8‰ rims.
• DEW modeling: [S]fluid is generally low (<0.1 molal) but peaks at 0.20–0.35 molal near 3.0 GPa (~90 km), for metasediment, metabasalts, and serpentinite alike, indicating a sulfur release pulse. Sulfur species are consistently reduced across modeled conditions; ±1 log unit fO2 variations mainly change species proportions without large [S]fluid changes.
• Sulfur isotope fractionation between H2S-bearing fluids and pyrite is small (<1.3‰ at relevant P–T–fO2–pH; <1‰ at 550 °C), so vein sulfide δ34S closely tracks fluid δ34S.
• Global mass balance: subduction sulfur influx FS ≈ 4.65 × 10^13 g yr−1 with bulk δ34S ≈ −3.60‰; major input reservoirs are gabbro (49%) and sediment (23%).
• Slab sulfur outflux fS between 30–230 km is ≈ 2.91 × 10^12 g yr−1 (6.3% of FS). The dominant release window is 70–100 km with ≈ 2.46 × 10^12 g yr−1 (5.3% of FS), coincident with pyrite→pyrrhotite breakdown and peak H2O release.
• Net δ34S of slab fluids: at 70–100 km, after accounting for limited isotopic exchange during ascent, δ34S ≈ −2.5 ± 3‰; integrated over 30–230 km, δ34S ≈ −2.1‰. Even with model uncertainties, the slab fluid δ34S remains slightly negative and sulfate flux is negligible.
• Implications: Modest slab-to-wedge sulfur transfer can sustain elevated S in arc mantle sources but does not supply sufficient sulfate to oxidize the mantle wedge or to generate positive δ34S in arc magmas. Most sulfur (>80%) with negative δ34S is retained and subducted into the deep mantle, potentially raising the long-term δ34S of surface reservoirs.

The findings demonstrate that slab-derived fluids at sub-arc depths are dominated by reduced sulfur (H2S/HS−) with small isotopic fractionation relative to precipitated sulfides, allowing vein pyrite to record fluid δ34S. The DEW-modeled [S]fluid peak at ~90 km, combined with global H2O fluxes, constrains a modest but focused sulfur release (primarily at 70–100 km). The flux-weighted negative δ34S of slab fluids (−2.5 ± 3‰) shows that slab fluids cannot be the source of positive δ34S signatures observed in some arcs, nor are they a significant sulfate-based oxidant of the mantle wedge. Instead, the high fO2 of certain arc magmas likely reflects processes associated with H2O/CO2 addition and redox reactions in the mantle wedge unrelated to sulfate delivery. The majority of subducted sulfur with negative δ34S is retained beyond arc depths and transported into the deep mantle, consistent with isotopic evolution of surface reservoirs and some OIB signatures.

This study provides quantitative constraints on sulfur speciation, concentration, flux, and isotopic composition of slab-derived fluids using exhumed HP rocks and veins combined with thermodynamic modeling. Key contributions include: (1) demonstration that reduced sulfur species dominate slab fluids; (2) identification of a focused sulfur release at 70–100 km depth; (3) determination that only ~6.4% (up to 20% maximum) of subducted sulfur is released between 30–230 km; and (4) establishment of a slightly negative net δ34S (−2.5 ± 3‰) for slab fluids to the mantle wedge. These results imply limited slab sulfate flux, minimal direct oxidation of the wedge by sulfate, and that positive δ34S in arc magmas must arise from alternative processes during melting and magma evolution. Future work should refine sulfur speciation in fluids at high P–T (including potential unmodeled species), better constrain variability in slab redox and sediment compositions, and investigate mechanisms generating heavy δ34S in arc magmas.

• Thermodynamic uncertainties: DEW equilibrium constants carry uncertainties (±0.3–0.5 logK), and unrecognized sulfur species could cause [S]fluid underestimation.
• Redox and lithologic assumptions: fO2 paths and slab stratigraphy are generalized; local variations (e.g., degree of serpentinization, sediment types/redox) may alter fluid compositions and fluxes.
• Isotope exchange during transport: Although constrained to be limited, some fluid–rock δ34S exchange likely occurs along pathways, potentially modifying primary signals.
• Global budgets: Inputs rely on average global parameters (trench length, convergence rates, layer thickness/densities, [S], δ34S), introducing uncertainties in flux estimates.
• Sample representativeness: Results are based on SW Tianshan rocks/veins; while conditions are analogous to many cold subduction zones, variability among arcs and slabs may exist.