An international journal of inorganic chemistry www.rsc.org/dalton

Number 11  |  21 March 2008  |  Pages 1385–1508

ISSN 1477-9226

HOT ARTICLE Alain Manceau and Kathryn L. Nagy Relationship between Hg(ii)–S bond distance and Hg(ii) coordination in thiolates

PERSPECTIVE Takahiro Sasamori and Norihiro Tokitoh Doubly bonded systems between heavier Group 15 elements

1477-9226(2008)11;1-9

PAPER

www.rsc.org/dalton | Dalton Transactions

Relationships between Hg(II)–S bond distance and Hg(II) coordination in thiolates Alain Manceau*a and Kathryn L. Nagyb Received 28th November 2007, Accepted 10th January 2008 First published as an Advance Article on the web 12th February 2008 DOI: 10.1039/b718372k Extended X-ray absorption fine structure (EXAFS) spectroscopy is a sensitive structural probe of the coordination environments of Hg(II) with thiol (sulfhydryl) groups, and is equally applicable to solid and aqueous organic or inorganic matter. Information on the number and geometric arrangement of S ligands can be derived from metal–ligand distances because the distances vary with Hg stereochemistry and can be accurately measured by the EXAFS technique. To improve the reliability of determining coordination structures of Hg with thiol groups, correlations among Hg–S bond distance, Hg coordination, and S–Hg–S angle in homoleptic Hg(II)-thiolates were calculated from analysis of the structures of the 162 Hg(SR)n motifs (n = 2, 3, 4) contained in the Cambridge Structural Database v. 5.28. Graphical correlations of bond distance with coordination number and with bond angle show distinct ranges of values characteristic of specific structural configurations.

Introduction Mercury has a high thiophilicity, which is exemplified in nature by its occurrence as cinnabar (a-HgS) and metacinnabar (b-HgS) in ore deposits and anoxic sediments and sedimentary rocks, as well as its high affinity for thiolate ligands (SR) in living and detrital organic matter. In biota, Hg(II) is transferred from one protein to another of higher affinity when it is transported into the cell, and can be ultimately sequestered in an insoluble form or reduced to Hg(0) and vaporized.1–3 The affinity of Hg for shuttling proteins is controlled mainly by the strength of its binding to sulfur, which depends on the number and topology of the SR groups.4 Mercury can adopt a large range of coordination modes, the linear, trigonal planar and tetrahedral being the most common.5,6 Another intrinsic property of the Hg(II) atom is to form asymmetrical Hg–S bonds when the number of S ligands exceeds two; the T-shape geometry of some coordination compounds represents in this respect the most extreme case of trigonal distortion.7–9 The distorted trigonal planar configuration is believed to be responsible for Hg(II) detoxification in living organisms because it is the most stable, followed by the symmetrical dicoordinate S– Hg–S configuration.3,10 The binding mechanism of Hg in natural organic matter (NOM) likely also varies in strength with the Hg/S ratio, thereby controlling the fate and bioavailability of Hg(II) in the environment. However, in contrast to proteins, the coordination environment of Hg(II) in NOM cannot be obtained by diffraction and NMR spectroscopy, because NOM cannot be crystallized and has a molecular weight too high for NMR.11 Instead, Hg–S bond lengths derived from analysis of EXAFS spectroscopic data can be compared to the dependence of Hg– S bond length on Hg(II) coordination chemistry in reported a Maison des G´eosciences, CNRS and Universit´e J. Fourier, 38041, Grenoble Cedex 9, France. E-mail: [email protected] b Department of Earth and Environmental Sciences, 845 W. Taylor St., MC186, University of Illinois at Chicago, Chicago, IL, 60607, USA. E-mail: [email protected]

This journal is © The Royal Society of Chemistry 2008

simple structures to assess mercury’s binding environment in more complex materials such as NOM. Crystallographic data surveyed 15 to 20 years ago showed ˚ for each that the Hg–S distance increases on average by 0.1 A added ligand in the series of mononuclear Hg(SR)n complexes, where n = 2, 3 or 4.6 Therefore, with an accuracy on first-shell ˚ by EXAFS spectroscopy,12 the number of distances of ca. 0.02 A S ligands bonded to Hg is typically determinable from EXAFSderived bond lengths, alone.13–18 This number is more difficult to obtain with derived coordination numbers (CN) because they are meaningful to only ca. 20% in symmetrical environments, and their accuracy can be as low as 50% in highly disordered structures.19 The same crystallographic data also showed that ˚ (Rav. = the mean Hg–S distances range from 2.32 to 2.36 A ˚ ˚) ˚ 2.34 A) for Hg(SR)2 complexes, 2.40 to 2.51 A (Rav. = 2.44 A ˚ ˚ for Hg(SR)3 complexes, and 2.50–2.61 A (Rav. = 2.54 A) for Hg(SR)4 complexes.6,20 The separation in bond distances between the Hg(SR)3 and Hg(SR)4 complexes, and to some extent the Hg(SR)2 and Hg(SR)3 complexes, may in some cases be too low to discern the Hg coordination environment in a statistically meaningful way by EXAFS spectroscopy. Taking into account the accuracy on EXAFS distances, this situation would occur typically for compounds with experimental bond distances of ˚ . At the time these data sets for Hg(II)∼2.38 and 2.48–2.53 A thiolates were analyzed comparatively, only a few structures of homoleptic mercury sulfides had been refined (approximately 13). Today, Rav values and their standard deviations can be calculated more precisely from a much larger dataset of 162 structures. In addition to the R–CN relationship, there is also a relationship within the set of three-coordinate, or Y-shape, complexes between the S–Hg–S angle and the length of the opposing Hg–S bond.6,21 Hg(SR)3 complexes always show significant distortion towards a linear geometry. As the angle between two bonds is increased from 120◦ , the lengths of these two bonds are reduced, and the Rav value of the Hg(SR)3 complex is preserved by lengthening the third bond. The easy deformability of the coordination sphere, and thus Dalton Trans., 2008, 1421–1425 | 1421

Fig. 1 Mean bond-length distances for two-, three- and four-coordinate Hg(SR)n complexes. Hg is black in Hg(SR)2 , blue in Hg(SR)3 , and red in Hg(SR)4 complexes, and S is always yellow. The blue square is the NOM compound.26 Source: Cambridge Structural Database, v. 5.28 (Nov. 2006).23

variability of coordination numbers, have been described in other M(SR)3 compounds of d10 metals, such as Cu(I).5 The distortions are considered to result from steric interactions between the thiolate ligands and S · · · S repulsions.5,22 In this study, the Rav values and their standard deviations for Hg(SR)n (n = 2, 3, 4) complexes were calculated from the crystallographic structures of all homoleptic Hg(II)-thiolate compounds contained in the Cambridge Structural Database.23 Graphical correlations between bond distance and S–Hg–S bond angle also are presented for the Hg(SR)3 complex. These results are used to assess the geometry of the current model for the complexation of Hg(II) to NOM.

Correlation between Hg–S bond distance and coordination number In total, 96 bond lengths from 48 Hg(SR)2 motifs, 33 from 11 Hg(SR)3 motifs, and 412 from 103 Hg(SR)4 motifs were calculated from the CSD v. 5.28 (Fig. 1). Dicoordinate Hg always occurs at the crystallographic center of symmetry of the S–Hg–S complex ˚ . In Hg(SR)3 and Hg(SR)4 complexes, the three or to within 0.02 A four Hg–S bonds are never equal regardless of the identity of the ligand molecules, and thus none of these complexes are perfectly symmetric. The maximum range of distances among the three ˚ , except for two THg–S bonds of a Hg(SR)3 complex is 0.12 A ˚9 shaped Hg(II)-thiolates in which one bond is stretched by 0.46 A ˚ 7 relative to the two other bonds. These two coordinaor 0.56 A tion complexes, referred to as UCEYAD ([Hg(Tab)2 ](PF6 )2 ) and FETTUT ([Hg(S2 COCH3 )2 ]) in the CSD, were excluded from the calculations described below because they are structurally unusual 1422 | Dalton Trans., 2008, 1421–1425

and not compositionally representative of expected molecular subunits in NOM. Overall, the distribution in distances is broader in the Hg(SR)4 complexes, with differences among bond lengths as ˚. high as 0.5–0.6 A The mean Hg–S distances fall in three groups of constant coordination number without overlap in distance, except for one Hg(SR)2 complex, in which Hg is coordinated to two thiocyanate groups, and the T-shaped FETTUT molecule. The average (Rav ± rav ) and median (Rm ) values of the mean distances for the three ˚ and Rm = 2.342 A ˚ for types of complex are Rav = 2.345 ± 0.025 A ˚ ˚ Hg(SR)2 , Rav = 2.446 ± 0.018 A and Rm = 2.443 A for Hg(SR)3 , ˚ and Rm = 2.553 A ˚ for Hg(SR)4 . Average and Rav = 2.566 ± 0.047 A ˚ when a third ligand is added, bond distance increases by 0.10 A ˚ when a fourth is added, confirming and again by 0.11–0.12 A the analysis of earlier data by Wright et al.6 More importantly, the separations between the three sets of Rav and Rm values are statistically large enough to use the EXAFS distance (R) to derive the correct coordination number at the 95% confidence level (2r). The standard deviations of the Rav values (rav ) can be compared to the dispersion of all individual bond lengths in all complexes of the same type (rtot ) to determine if inter (rav ) or intra (rtot ) ˚ and molecular disorder is higher. For Hg(SR)2 , rtot = 0.024 A ˚ ; for Hg(SR)3 , rtot = 0.038 A ˚ and rav = 0.018 A ˚; rav = 0.025 A ˚ and rav = 0.047. In the Hg(SR)3 and, for Hg(SR)4 , rtot = 0.128 A and Hg(SR)4 complexes, rtot > rav , meaning that the distribution in distances is statistically larger within a complex than between any two complexes. This result indicates that the increase in length of one Hg–S bond is counterbalanced in a Hg(SR)n center by shortening of the remaining Hg–S bonds.24 In other words, all bond distances at the Hg center average out to approximately ˚ for Hg(SR)4 ), ˚ for Hg(SR)3 and 2.566 A the same value (2.446 A This journal is © The Royal Society of Chemistry 2008

independent of the breadth of their distribution in any one compound.

Correlation between Hg–S bond distance and S–Hg–S angle The S–Hg–S angle varies from 161.4 to 180◦ in the Hg(SR)2 complexes (average 175.6 ± 5.1◦ ), and is independent of the bond distance (Fig. 2(a)). The range of variation is larger for the Hg(SR)3 complexes (91–173◦ ), deviating by as much as −29 and +53◦ from the ideal 120◦ value in the regular trigonal planar geometry. However, the sum of the three angles from the same complex is always close or equal to 360◦ (average 359.7 ± 0.7◦ ), indicating that all three-coordinate centers are planar. In addition, there is a correlation (r = 0.80) between the S–Hg–S angle and the length of the opposite bond (Fig. 2(b)), except for the FETTUT and UCEYAD T-shaped clusters whose largest S–Hg–S angle (165 and 173◦ , respectively) falls in the Hg(SR)2 angular domain (Fig. 2(a)). The coefficients of the regression equation of the bond distance ˚ −1 vs. S–Hg–S angle correlation h = aR + b are a = 262.5 deg A ◦ and b = −522 deg (rh = 7.6 ).

Fig. 2 (a) S–Hg–S bond angle vs. Hg–S bond distance graph for two- and three-coordinate Hg(SR)n complexes. The number of points for each set of complexes is multiplied by the number of Hg–S bonds in each set, e.g., the Hg(SR)3 series has three times more points than in Fig. 1 because each complex has three sets of bond angle–distance values. The points denoted by NOM are from the T-shape model for the binding of Hg(II) to natural organic matter derived from EXAFS data.26 (b) S–Hg–S bond angle vs. Hg–S bond distance correlation in the Hg(SR)3 series. The two T-shape complexes were excluded from the calculation.

This journal is © The Royal Society of Chemistry 2008

The T-shape geometry In which circumstances does Hg adopt a T-shape geometry? Based on the relationship between the S–Hg–S angle and the length of the opposite Hg–S bond, this occurs when Hg is coordinated strongly to two S atoms as indicated by short-bond distances and weakly to a third one as indicated by a significantly longer bond distance. In the FETTUT complex, Hg is bonded to two xanthate ligands (S2 COCH3 ) at short distances, and to a third S ligand from a neighboring polymeric chain at a longer distance ˚ ) (Fig. 3). This coordination geometry is somewhat similar (2.92 A to that of a-HgS (cinnabar), in which –S–Hg–S–Hg– chains (R = ˚ , S–Hg–S = 173◦ ) running through the structure are linked 2.37 A approximately perpendicularly by weak inter-chain Hg–S bonds ˚ ,25 allowing cinnabar to be easily cleaved along of 3.09 and 3.29 A planes parallel to the chain directions. In contrast to the FETTUT complex, the long bond does not result from intermolecular interaction in the UCEYAD complex, but from the coordination ˚ ). The two short of Hg to one weak SCN− ligand (R = 2.80 A ˚ ). The T-shape distances are with thiolate ligands (R = 2.34 A geometry, though unusual among all known homoleptic Hg(II)thiolates, has been proposed recently as the strongest bonding environment of divalent mercury to NOM, suggesting that it may be common in nature.26 Using derived distances from EXAFS data, Hg was shown to be bonded linearly to two S groups (likely ˚ , and through intermolecular interaction to a thiols) at 2.33 A ˚ suggested to be a methionine sulfur or an third S at 2.92–3.08 A exogeneous S group. The linear arrangement of the two short bonds was deduced from the intense multiple scattering (MS) contribution in the EXAFS data that occurs at twice the short ˚ , Fig. 4). Based on the results metal–ligand distance (2R = 4.66 A presented here, the long Hg–S distance also would be diagnostic for the presence of linear S–Hg–S bonds with two other sulfur ligands at short distance using the correlation between the S– Hg–S angle and length of the opposite Hg–S bond. Therefore, the tri-coordinate geometric model proposed for the binding

Fig. 3 Bonding environment of Hg in the two T-shaped Hg(SR)3 complexes whose structures were refined by diffraction.7,9

Dalton Trans., 2008, 1421–1425 | 1423

as derived, for example, from EXAFS spectroscopy. The three main coordination modes of Hg(II) can be distinguished in most cases by this technique because the accuracy on absorber–scatterer ˚ , provided the data were measured over distance is as high as 0.02 A ˚ −1 ) with little noise. a large wavevector k range (typically kmax ≥ 12 A Discriminating the two sets of angles in the Y-shape geometry is more difficult because the bond-length resolution in Hg-EXAFS is ˚ . This resolution, which corresponds rarely better than dR = 0.11 A to a difference in angles of 29◦ , is obtained for kmax = p/2dR = ˚ −1 , a value difficult to attain experimentally on diluted 14.2 A samples.12 If dR is larger than the difference between the shortest and longest Hg–S bonds at the Hg center, no information is obtained on the three bond angles, other than that they average to 120◦ . If the asymmetry of the complex is large enough for EXAFS to see more than one average Hg–S distance, then the value of the angle opposite to each EXAFS distance can be estimated within an uncertainty of 7.6◦ from the graphical correlation between bond length and bond angle.

Acknowledgements Fig. 4 Relative amplitude of the three non-equivalent multiple scattering paths in EXAFS spectroscopy for the S–Hg–S unit as a function of bond angle. The amplitude values are normalized to the amplitude of the Hg–O single scattering path, which is set to 100.28 Calculation performed with ˚. R = 2.35 A

mechanism of Hg on NOM is supported by two independent lines of reasoning. Similarly to the long Hg–S distance, the amplitude of the MS contribution to EXAFS in coordination complexes varies with the S–Hg–S bond angle, and thus can be used to obtain geometric information as well.27 The sensitivity of the MS amplitude to the dihedral angle has been calculated here with the FEFF 7.01 code28 to estimate if this dihedral angle is slightly bent in NOM, as in FETTUT (165◦ ) and UCEYAD (173◦ ), or straight as in any Hg(SR)2 complex. The MS contribution measured in EXAFS comprises three non-equivalent scattering paths with the same effective distance: a three-legged path (Hg→S1→S2→Hg, MS3), a fourlegged path involving two different S atoms (S1⇔Hg⇔S2, MS41), and a second four-legged path involving the same atomic pair (Hg→S1→Hg→S1→Hg and Hg→S2→Hg→S2→Hg, MS4-2) (Fig. 4). The second four-legged path has no angular dependence because the photoelectron is scattered twice by the same S atom, as confirmed by calculation. Fig. 4 shows a 12% reduction of the total MS intensity at 170◦ and 41% at 160◦ , and little effect at angles lower than 160◦ . Therefore, the strong intensity of the MS contribution in organic matter indicates that the S–Hg–S angle ˚, is linear and, together with the long Hg–S bond at 2.92–3.08 A that mercury has an undistorted T-shape geometry. The reason why divalent mercury would acquire this uncommon geometry in NOM, and by implication be widespread in the environment, is as yet unknown.

Concluding remarks Despite the versatility of the Hg(II)–S bonding environment, reliable geometric information can be obtained from bond distances, 1424 | Dalton Trans., 2008, 1421–1425

Support was provided to A. Manceau from the ANR/CNRSECCO program, and to K. L. Nagy from U.S. National Science Foundation grant EAR-0447310.

References 1 T. V. O’Halloran, B. Frantz, M. K. Shin, D. M. Ralston and J. G. Wright, Cell, 1989, 56, 119. 2 J. L. Hobman, J. Wilkie and N. L. Brown, BioMetals, 2005, 18, 429. 3 J. Qin, L. Song, H. Brim, M. J. Daly and A. O. Summers, Microbiology, 2006, 152, 709. 4 R. A. Steele and S. J. Opella, Biochemistry, 1997, 36, 6885. 5 I. G. Dance, Polyhedron, 1986, 5, 1037. 6 J. G. Wright, M. J. Natan, F. M. McDonnell, D. M. Ralston and T. V. O’Halloran, Prog. Inorg. Chem., 1990, 38, 323. 7 E. R. Tiekink, Acta Crystallogr., Sect. C, 1987, 43, 448. 8 S. W. A. Fong, W. T. Yap, J. J. Vittal, T. S. A. Hor, W. Henderson, A. G. Oliver and C. E. F. Rickard, J. Chem. Soc., Dalton Trans., 2001, 1986. 9 J. X. Chen, W. H. Zhang, X. Y. Tang, Z. G. Ren, Y. Zhang and J. P. Lang, Inorg. Chem., 2006, 45, 2568. 10 D. Ghosh, K. H. Lee, B. Demeler and V. L. Pecoraro, Biochemistry, 2005, 44, 10732. 11 L. M. Utschig, J. W. Bryson and T. V. O’Halloran, Science, 1995, 268, 380. 12 J. E. Penner-Hahn, Chem. Rev., 2005, 249, 161. 13 J. G. Wright, H. T. Tsang, J. E. Penner-Hahn and T. V. O’Halloran, J. Am. Chem. Soc., 1990, 112, 2434. 14 M. Matzapetakis, B. T. Farrer, T. C. Weng, L. Hemmingsen, J. E. Penner-Hahn and V. L. Pecoraro, J. Am. Chem. Soc., 2002, 124, 8042. 15 F. Jalilehvand, B. O. Leung, M. Izadifard and E. Damian, Inorg. Chem., 2006, 45, 66. 16 K. M. Zaman, L. Y. Blue, F. E. Huggins and D. A. Atwood, Inorg. Chem., 2007, 46, 1975. 17 D. T. Jiang, S. M. Heald, T. K. Sham and M. J. Stillman, J. Am. Chem. Soc., 1994, 116, 11004. 18 B. O. Leung, F. Jalilehvand and V. Mah, Dalton Trans., 2007, 4666. 19 A. Manceau, M. C. Boisset, G. Sarret, J. L. Hazemann, M. Mench, P. Cambier and R. Prost, Environ. Sci. Technol., 1996, 30, 1540. 20 L. M. Utschig, J. G. Wright and T. V. O’Halloran, Methods Enzymol., 1993, 226, 71. 21 T. Alsina, W. Clegg, K. A. Fraser and J. Sola, J. Chem. Soc., Dalton Trans., 1992, 1393.

This journal is © The Royal Society of Chemistry 2008

22 E. S. Gruff and S. A. Koch, J. Am. Chem. Soc., 1990, 112, 1245. 23 F. H. Allen, Acta Crystallogr., Sect. B, 2002, 58, 380. 24 N. Govindaswamy, J. Moy, M. Millar and S. A. Koch, Inorg. Chem., 1992, 31, 5343. 25 P. Auvray and F. Genet, Bull. Soc. Fr. Mineral. Cristallogr., 1973, 96, 218.

This journal is © The Royal Society of Chemistry 2008

26 U. Skyllberg, P. R. Bloom, J. Qian, C. M. Lin and W. F. Bleam, Environ. Sci. Technol., 2006, 40, 4174. 27 A. Van DerGaauw, O. M. Wilkin and N. A. Young, J. Chem. Soc., Dalton Trans., 1999, 2405. 28 A. L. Ankudinov, B. Ravel, J. J. Rehr and S. D. Conradson, Phys. Rev. B, 1998, 58, 7565.

Dalton Trans., 2008, 1421–1425 | 1425