• Explore more than 100 years of PNAS content!
  • Sign-up for PNAS eTOC Alerts

Evidence of magnetic isotope effects during thermochemical sulfate reduction

  1. James Farquhara,1
  1. aDepartment of Geology and Earth System Science Interdisciplinary Center (ESSIC), University of Maryland, College Park, MD 20742;
  2. cDepartment of Chemistry and Biochemistry, University of Maryland, College Park, MD 20742;
  3. dGeophysical Laboratory, Carnegie Institution of Washington, 5251 Broad Branch Road NW, Washington, DC 20015; and
  4. bDepartment of Earth, Atmospheric, and Planetary Sciences, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139-4307
  1. Edited by Mark H. Thiemens, University of California at San Diego, La Jolla, CA, and approved August 23, 2011 (received for review May 24, 2011)


Thermochemical sulfate reduction experiments with simple amino acid and dilute concentrations of sulfate reveal significant degrees of mass-independent sulfur isotope fractionation. Enrichments of up to 13‰ for 33S are attributed to a magnetic isotope effect (MIE) associated with the formation of thiol-disulfide, ion-radical pairs. Observed 36S depletions in products are explained here by classical (mass-dependent) isotope effects and mixing processes. The experimental data contrasts strongly with multiple sulfur isotope trends in Archean samples, which exhibit significant 36S anomalies. These results support an origin other than thermochemical sulfate reduction for the mass-independent signals observed for early Earth samples.

Since the report by Farquhar et al., (1) that significant deviations from the terrestrial fractionation line are observed in samples older than approximately 2.32–2.45?Ga (2, 3), considerable effort has been dedicated to identifying the origin and significance of this signal (410). The mass-independent signal in these ancient samples is expressed as variations in both Δ33S and Δ36S (1).* Given the observations that gas-phase reactions can produce mass-independent signals for both Δ33S and Δ36S, the first studies on this subject attributed this ancient signal to photolytic reactions in the early atmosphere. Subsequent studies also pointed out that the mass-independent reactions may also be produced by variations in the spectrum of light that drives atmospheric photolytic reactions (1012), and other studies speculated that liquid phase reactions involving weakly bound transition states may account for these variations (7, 13).

In a recent report, Watanabe et al. (7) demonstrated that high temperature reduction of sulfate using alanine and glycine as organic substrates caused moderate mass-independent sulfur isotope fractionations. These authors did not identify the origin of the effect, but suggested that it was either a magnetic isotope effect (MIE) (14) or another type of isotope effect accompanying heterogeneous reactions such as adsorption of S-bearing compounds on surfaces of solids (13). Magnetic isotope effects are expressed in rare cases for isotopes with nuclear magnetic moments, like 13C, 17O, 29Si, 199Hg, 201Hg, 73Ge, 235U, and 33S (14, 15, 16). The effect is expressed when the lifetime of a radical pair is sufficient for hyperfine coupling between magnetic nuclei and unpaired electrons to influence interconversions between singlet and triplet states. This coupling in turn changes the proportion of reactive intermediates that can participate in spin-selective reactions. The 33S nucleus has a spin of 3/2 and a magnetic moment of 0.643 nuclear magnetons and has been implicated in at least one well-characterized example of a 33S MIE (14, 1719). The alternative suggestion relates to a proposal that anomalous isotope effects may be associated with heterogeneous reactions as a result of possible missing vibrational levels involving very weakly bound transition states (13). This proposal has been contested by Balan et al. (20), who argue that no effect exists when a more complete treatment of the reaction mechanisms is undertaken. Uncertainty in ascribing the origin of the effect to an MIE arose because two of their samples possessed Δ36S that was different from that of the starting materials. Uncertainty, ruling out an MIE (and demonstrating a different type of anomalous isotope effect) also arose because other processes relevant to the complex reaction pathways of thermochemical sulfate reduction (TSR) in their system involve mixing and can generate mass conservation effects (21) that have been shown to produce small variations in Δ33S and more significant variations in Δ36S (e.g., ≥2‰ in biological and biogeochemical systems (22, 23).

Here we report results from two sets of high temperature experiments (i.e., flow-reactor and Carius tube experiments) that suggest the observation of 33S anomalies in these reactions are related to a magnetic isotope effect in the reaction products. The mechanistic aspects for the origin of 33S anomalies via disulfide ion-radical pair reaction routes have been proposed through multiple sulfur isotope measurements and 33S electron spin resonance (ESR) spectroscopic evidence (24, 25).

Results and Discussion

The flow through reactor experiments yielded two distinct sulfur products as seen in Table?S1 [gaseous H2S and chromium-reducible sulfur (CRS)] with anomalous 33S enrichments (Δ33S from +0.25 to +13.1‰), but no significant change in 36S composition from starting sulfate (Fig.?1 A and B). In Carius tube experiments, 33S enrichments in acid volatile sulfide (AVS) and CRS products were small to undetectable except when S8 or Na2S2O4 was added (see Table?S2) to catalyze reduced sulfur production in the reaction products with no effect to the S-radical chemistry. We speculate that the observed isotope effect originates from ion-radical pair (RS?H+/RS?SH) intermediates (24, 26) through the following sequence of reactions:

  • 1. Prolonged heating of glycine (mp?=?262?°C) affords three major classes of compounds: (i) small neutral molecules, (ii) carbon and other carbon-based polymers, and (iii) radical intermediates (both heteroatom and carbon-centered radicals) as indicated in Scheme?1 below (27, 28).

    Scheme 1.

    Pyrolytic decomposition products of glycine. Solid-state NMR and high resolution ESI-MS (in positive mode) confirm the presence of polymers in the reaction mixture.

  • 2. Sodium sulfate can then be reduced by some of the reducing gases produced in Scheme?1 (e.g., H2, CO, etc.) to give sodium sulfide (1, 2). Trace metal impurities in sodium sulfate can catalyze this reaction. Solid carbon or carbon-based polymers, generated during pyrolysis can also reduce sodium sulfate to sodium sulfide (29, 30) (3):Formula[1]Formula[2]Formula[3]

  • 3. Sodium sulfide in the presence of trace acid and/or water and heat will generate hydrogen sulfide (H2S) (4). The acid/water would come from the H2O, HCN, and COOH generated during glycine pyrolysis (see Scheme?1) or even from the glycine starting material, Formula[4]

  • 4. Hydrogen sulfide can undergo thermolysis (31, 32), which can react with radical intermediates generated during glycine pyrolysis (see Scheme?1) to give thiyl radicals as in [5] and [6] (33). Note that these reactions are not spin-selective so no 33S anomaly will result, Formula[5]Formula[6]

  • 5a. The thiyl radical (HS?) can recombine with other radicals to give neutral, sulfur-containing molecules. Reactions between free radicals (e.g., 79) can be spin-selective (34). The absence of measureable sulfur-33 enrichment in the carbon-bound sulfur (Raney Ni fraction) suggests however that this reaction is not the origin of the isotope effect in these experiments, Formula[7]Formula[8]Formula[9]

  • 5b. The products of [7], [8], and [9] can undergo further homolytic cleavage of S-H bonds shown in Scheme?2 to give thiyl radical moieties (?SH, ?SY, and ?SR). These thiyl radicals have strong reactivity and can also react with other radicals to form polysulfide products via sulfur polymerization (10). Such reactions will not produce sulfur-33 enrichments due to strong spin-orbit coupling (see below), Formula[10]

    Scheme 2.

    Proposed ion-radical pair mechanism showing spin evolution between triplet and singlet states during thermochemical sulfate reduction.

  • 6. The thiyl radical can also abstract hydrogen from C-H bonds to give carbon-centered radicals. These reactions will not be spin-selective,Formula[11]Formula[12]

Fig. 1.

S-isotope plots of Δ36S versus Δ33S (A) and δ33S versus δ34S (B) for Carius tube and flow reactor experiments, abbreviated as GSW (Graphic), GSSW (Graphic), and GSDW (Graphic). Typical mass-dependent arrays are plotted in both panels. Most data follow a tightly constrained, mass-dependent relationship of δ33S?=?0.515(?±?0.008)?×?δ34S (B). Deviations from this array and the Δ36S versus Δ33S array are interpreted as magnetic isotope effects. The MIE trends are distinct from sulfur photoexcitation experiments and are not a likely explanation for the Archean sulfur isotope record. Error bars represent 1σ analytical uncertainties of 0.02 and 0.2 for Δ33S and Δ36S, respectively.

Based on the above sequence of reactions, we propose Scheme?2, which describes the pathway envisioned for production of the observed sulfur-33 enrichments in Cr-reducible sulfur and hydrogen sulfide. We have generalized this sequence of reactions by writing it for only R (and not Y and H sulfur bonded radicals). During accidental encounter of free radicals (reactions A and B) in Scheme?2, the statistical distribution of radical pair spin states will be one-fourth singlet and three-fourths triplet states. MIE may occur if this 1∶3 proportion is altered by the rapid formation of singlet products upon initial encounter, leaving an excess (>three-fourths) of triplet radical pairs that undergo triplet to singlet conversion. However, these reactions are an unlikely source of MIE because it has been shown that thiyl radicals exhibit strong spin-orbit coupling (35, 36) and therefore are expected to experience rapid spin flipping independent of hyperfine coupling.

We instead suggest that the observation of 33S enrichments in the CRS fraction implicates MIE associated with the formation of polysulfide species [reaction C (Scheme?2) and the subsequent network highlighted in the red box]. We suggest this polymerization reaction of sulfur product is mediated by an ion-radical mechanism similar to that proposed by Buchachenko et al. (26). Here RSH acts as an electron donor to RSSH, forming an ion-radical pair intermediate that is initially in a singlet state. Coulombic attraction inhibits dissociation of the radical pair. Here, the radical pair may either (i) reform the original reactants by back electron transfer, (ii) slowly lose H2Sn (where n?=?0 or 1) in a non-spin-selective manner, or (iii) undergo singlet to triplet conversion via 33S hyperfine coupling. Back electron transfer from the triplet state is spin forbidden, and therefore the radical pair must lose H2Sn to form an Graphic radical pair. Subsequent triplet-singlet conversion allows for the radical pair to combine and thus form a polysulfide (RSSSnR) product. Such products that are formed through the triplet pathway are therefore enriched in 33S. Our mechanism is supported by prior ESR measurements (24, 25), which show that disulfide radicals have 33S hyperfine structure values (~10?G) similar to sulfur nuclei that exhibit MIE. The experimental products also preserve evidence for significant mixing and classical isotope effects that influence Δ36S in a mass-dependent manner (21, 23), supporting this small anomaly as the cause of 36S variation reported in ref.?7).

The Eschka sulfur and the Cr-reducible sulfur appear to be isotopically fractionated relative to the residual sulfate by two distinct processes. The Eschka sulfur has a mass-dependent 34S enrichment with respect to the starting composition, and the Thode fraction (which include residual sulfate), which is, in turn, 34S-depleted relative to the starting composition. We infer that these observations indicate the principal loss pathway for sulfate is the mass-dependent formation of product sulfur in the Eschka fraction. The Cr-reducible sulfur fraction is 33S-enriched and 34S-depleted, consistent with an MIE following the mechanism described that would yield a smaller secondary product fraction, and a residue with a small 33S-depletion. We interpreted the absence of a measurable fractionation in the Eschka and Raney Ni fractions to reflect dilution of R-SH that formed from non-spin-selective reactions.


We conclude that the origin of MIE is related to sulfur radicals generated by thiyl-mediated thermolysis of H2S that undergo a rapid ion-radical pair polymerization reaction to form the chromium (II) reducible sulfide product. Because the MIE captured in these TSR experiments principally affects Δ33S without significantly affecting Δ36S, the relevance of these reactions as an explanation for mass-independent sulfur isotope effects reported from Earth’s most ancient rocks (where deviations from mass-dependent arrays are noted for both 33S and 36S) is limited. Moreover, the absence of sulfur isotope mass-independent fractionation in post-Archean organic-rich rocks suggests that TSR is not a widespread source of fractionations in typical sediments and further supports the assertion that the early record does not reflect this chemistry. Thermal reactions have, however, been proposed as a mechanism for formation of sulfur-containing compounds as well as their radical species in a variety of natural systems where organic matter and sulfur radicals are present [e.g., where sulfur radicals control petroleum maturation (37)]. It is possible therefore that sulfur MIE are generated in some settings, and evidence for this effect should be sought.

Experimental Procedure and Methods

Two sets of experiments were undertaken to monitor the products of TSR: (i) flow-reactor experiments and (ii) Carius tube experiments. Reagent grade sodium sulfate (~0.5?mol/L) and powdered glycine were used in both experiments, Formula[13]For the flow reactor experiments, glycine and 1.0?mL sodium sulfate solution were added to a reaction flask, matching the stoichiometry of [13], which was heated continuously at approximately 300?°C for 340?h under 15?bubbles/?min nitrogen flow. Water lost to evaporation was replenished by injecting 0.5?mL of Milli-Q water through a septum in the reaction flask [three or four times per experiment (Table?S1)]. Product hydrogen sulfide carried by the nitrogen flow was isolated by trapping with a Zn-acetate buffer, yielding a white crystalline ZnS precipitate. Solid and liquid residues in the reaction flask were treated by procedures outlined for the Carius tube experiments.

High-purity Pyrex glass Carius tubes (dimension 35.5-cm long, 12-mm outer diameter, ~1.2-mm wall thickness) were loaded with 0.5?mL sodium sulfate solution and glycine to match the stoichiometry of [13]. Two experiments also included sulfur intermediate species, S8 and Na2S2O4 (Table?S3), to catalyze the production of sulfide in reaction products. Sample tubes were placed in a stainless steel jacket before being heated in a muffle furnace at temperatures listed in Table?S1. After heating, the Carius tubes were chilled with liquid-nitrogen, crack-opened, and zinc acetate added to fix sulfide.

Solid and liquid fractions were isolated from the Carius tubes and placed into a flask for sequential extraction. The sulfur from reaction products were extracted using sequential reaction with 5?N HCl for AVS; Cr(II) acid distillation in ethanol for S-Sn (where n≥1) fractions (CRS); Raney Ni desulfurization for carbon bonded sulfur; Thode reducible sulfur for sulfate; and Eschka oxidized sulfur for total organic sulfur (methods described in ref.?38). Recovery was incomplete because some material adhered to the Carius tube walls, but upper estimates of the fraction of product Cr-reducible sulfur and Raney Ni reducible sulfur is provided in Table?S2. The proportion of Cr-reducible and Raney Ni reducible sulfur relative to Eschka- and Thode-sulfur was determined by the Cline method (39) using a UV-visible double beam (model UVD-3200) scanning spectrophotometer (Labomed Inc.) before converting sulfur into Ag2S for fluorination in Ni bombs, conversion to SF6 by heated reaction with F2, and subsequent S-isotopic analysis in a dual inlet ThermoFinnigan-253 mass spectrometer.

A solid-state NMR was acquired for residual solid fractions after the experiments using a Varian/Chemagnetics Infinity 300 solid-state NMR spectroscopy. High resolution electrospray ionization (ESI)–MS (resolving power 6,000?fwhm) were also taken for liquid fractions in both a positive ion mode using an AccuTOF (JEOL USA, Inc.) TOF–MS. The spray voltage was set to 2.3?kV, and the capillary and orifice temperatures were maintained at 250?°C and 80?°C, respectively. The instrument was typically operated at the following potentials: orifice 1?=?30?V, orifice 2?=?5?V, ring lens?=?10?V. The rf ion guide voltage was generally set to 1,000?V to allow detection of ions greater than m/z?=?100. Both solid-state NMR and ESI–MS analyses confirm the presence of neutral molecules, complex carbon-based macromolecules, and polymers that were formed through radical condensation reactions.


We thank R. Lawler for his technical assistance and guidance on refining our proposed mechanism. We also thank W. Guo, Y. Li, S.-T. Kim, D. Bentil, and A. Amrani for their comments and discussions that motivated this study. This study was supported by grants from the National Aeronautics and Space Administration Astrobiology Institute, Carnegie Institute of Washington team (to J.F.).


  • 1To whom correspondence may be addressed. E-mail: hoduro{at}umd.edu, hoduro{at}mit.edu, or jfarquha{at}glue.umd.edu.
  • Author contributions: H.O. and J.F. designed research; H.O. and J.F. performed research; H.O. and J.F. contributed new reagents/analytic tools; H.O., B.H., H.O.S., A.J.K., G.C., and J.F. analyzed data; and H.O., B.H., H.O.S., and J.F. wrote the paper.

  • The authors declare no conflict of interest.

  • This article is a PNAS Direct Submission.

  • This article contains supporting information online at www.danielhellerman.com/lookup/suppl/doi:10.1073/pnas.1108112108/-/DCSupplemental.

  • ?*Here, Δ33S?=?(33S/32S)sample/(33S/32S)cdt-[(34S/32S)sample/(34S/32S)cdt]0.515 and Δ36S?=?(36S/32S)sample/(36S/32S)cdt-[(34S/32S)sample/(34S/32S)cdt]1.90. Note, this definition is different than that used in ref.?7. The definition used here is consistent with definitions that normalize to a reference array defined by single-step equilibrium isotope exchange reactions and does not impact the conclusions of this study.


  1. ?
  2. ?
  3. ?
  4. ?
  5. ?
  6. ?
  7. ?
  8. ?
  9. ?
  10. ?
  11. ?
  12. ?
  13. ?
  14. ?
  15. ?
  16. ?
  17. ?
  18. ?
  19. ?
  20. ?
  21. ?
  22. ?
  23. ?
  24. ?
  25. ?
  26. ?
  27. ?
  28. ?
  29. ?
  30. ?
  31. ?
  32. ?
  33. ?
  34. ?
  35. ?
  36. ?
  37. ?
  38. ?
  39. ?

Online Impact

                                      1. 99132880 2018-01-23
                                      2. 802899879 2018-01-23
                                      3. 295573878 2018-01-23
                                      4. 352668877 2018-01-23
                                      5. 984633876 2018-01-23
                                      6. 545928875 2018-01-23
                                      7. 976569874 2018-01-23
                                      8. 871324873 2018-01-23
                                      9. 263462872 2018-01-23
                                      10. 577161871 2018-01-23
                                      11. 255603870 2018-01-23
                                      12. 117346869 2018-01-23
                                      13. 90982868 2018-01-23
                                      14. 663415867 2018-01-23
                                      15. 793874866 2018-01-23
                                      16. 843582865 2018-01-23
                                      17. 864971864 2018-01-22
                                      18. 258841863 2018-01-22
                                      19. 957295862 2018-01-22
                                      20. 553518861 2018-01-22