Synthesis, structure determination, and spectroscopic/computational characterization of a series of Fe(II)-thiolate model complexes: Implications for Fe-S bonding in superoxide reductases

Published on Web 01/19/2005

Synthesis, Structure Determination, and Spectroscopic/

Computational Characterization of a Series of Fe(II)-Thiolate

Model Complexes: Implications for Fe-S Bonding in

Superoxide Reductases

Adam T. Fiedler,^‡Heather L. Halfen,^†Jason A. Halfen,*^,†and Thomas C. Brunold*^,‡

Contribution from the UniVersity of Wisconsin-Eau Claire, Department of Chemistry,

105 Garfield AVenue, Eau Claire, Wisconsin 54702, and UniVersity of Wisconsin-Madison,

Department of Chemistry, 1101 W. UniVersity AVenue, Madison, Wisconsin 53706

Received May 24, 2004; Revised Manuscript Received November 10, 2004; E-mail: halfenja@uwec.edu; brunold@chem.wisc.edu

Abstract: A combined synthetic/spectroscopic/computational approach has been employed to prepare and

characterize a series of Fe(II)-thiolate complexes that model the square-pyramidal [Fe(II)(N_His)₄(S_Cys)]

structure of the reduced active site of superoxide reductases (SORs), a class of enzymes that detoxify

superoxide in air-sensitive organisms. The high-spin (S ) 2) Fe(II) complexes [(Me₄cyclam)Fe(SC₆H₄-p-

OMe)]OTf (2) and [FeL]PF₆(3) (where Me₄cyclam ) 1,4,8,11-tetramethylcyclam and L is the pentadentate

monoanion of 1-thioethyl-4,8,11-trimethylcyclam) were synthesized and subjected to structural, magnetic,

and electrochemical characterization. X-ray crystallographic studies confirm that 2 and 3 possess an N₄S

donor set similar to that found for the SOR active site and reveal molecular geometries intermediate between

square pyramidal and trigonal bipyramidal for both complexes. Electronic absorption, magnetic circular

dichroism (MCD), and variable-temperature variable-field MCD (VTVH-MCD) spectroscopies were utilized,

in conjunction with density functional theory (DFT) and semiemperical INDO/S-CI calculations, to probe

the ground and excited states of complexes 2 and 3, as well as the previously reported Fe(II) SOR model

[(L⁸py₂)Fe(SC₆H₄-p-Me)]BF₄(1) (where L⁸py₂is a tetradentate pyridyl-appended diazacyclooctane

macrocycle). These studies allow for a detailed interpretation of the SfFe(II) charge transfer transitions

observed in the absorption and MCD spectra of complexes 1-3 and provide significant insights into the

nature of Fe(II)-S bonding in complexes with axial thiolate ligation. Of the three models investigated, complex

3 exhibits an absorption spectrum that is particularly similar to the one reported for the reduced SOR enzyme

(SOR_red), suggesting that this model accurately mimics key elements of the electronic structure of the enzyme

active site; namely, highly covalent Fe-S π- and σ-interactions. These spectral similarities are shown to

arise from the fact that 3 contains an alkyl thiolate tethered to the equatorial cyclam ring, resulting in a

thiolate orientation that is very similar to the one adopted by the Cys residue in the SOR_redactive site.

Possible implications of our results with respect to the electronic structure and reactivity of SOR_redare

discussed.

Introduction

Structural studies have revealed that, in all SORs, the active

site consists of a mononuclear Fe center ligated by four

equatorial histidines and an axial cysteine in a square-pyramidal

geometry.^4-6In the ferric (S ) ⁵/₂) form of the enzyme (SOR_ox),

the coordination site trans to the thiolate is occupied by a

glutamate residue that dissociates upon conversion to the ferrous

(S ) 2) state (SOR_red).⁴Based on the ability of small molecules

(N₃^-, CN^-, NO) to coordinate to the Fe center in SOR,^6,7the

reaction with superoxide is believed to proceed by an inner-

sphere mechanism. Furthermore, putative Fe(III)-(hydro)peroxo

intermediates in the catalytic cycle have been generated in pulse-

In a wide variety of organisms, the detoxification of super-

oxide (O₂^•-) is accomplished via its dismutation to hydrogen

peroxide (H₂O₂) and molecular oxygen (O₂).¹In anaerobic

bacteria and archaea, though, an alternative strategy has been

found that involves only the one-electron reduction of super-

oxide to H₂O₂, a reaction that is catalyzed by a class of non-

heme iron enzymes known as superoxide reductases (SORs).^2,3

^†University of Wisconsin-Eau Claire.

^‡University of Wisconsin-Madison.

(1) (a) Valentine, J. S.; Wertz, D. L.; Lyons, T. J.; Liou, L.; Goto, J. J.; Gralla,

E. B. Curr. Opin. Chem. Biol. 1998, 2, 253-262. (b) Fridovich, I. Annu.

ReV. Biochem. 1995, 64, 97-112. (c) Miller, A. F. Comments Mol. Cell.

Biophys. 1997, 9, 1-48.

(4) Yeh, A. P.; Hu, Y.; Jenney, F. E., Jr.; Adams, M. W. W.; Rees, D. C.

Biochemistry 2000, 39, 2499-2508.

(2) Jenney, F. E., Jr.; Verhagen, M. F. J. M.; Cui, X.; Adams, M. W. W. Science

1999, 286, 306-309.

(5) Coelho, A. V.; Matias, P.; Fulop, V.; Thompson, A.; Gonzalez, A.;

Coronado, M. A. J. Biol. Inorg. Chem. 1997, 2, 680-689.

(3) Jovanovic, T.; Ascenso, C.; Hazlett, K. R. O.; Sikkink, R.; Krebs, C.;

Litwiller, R.; Benson, L. M.; Moura, I.; Moura, J. J. G.; Radolf, J. D.;

Huynh, B. H.; Naylor, S.; Rusnak, F. J. Biol. Chem. 2000, 275, 28439-

28448.

(6) Clay, M. D.; Jenney, F. E., Jr.; Hagedoorn, P. L.; George, G. N.; Adams,

M. W. W.; Johnson, M. K. J. Am. Chem. Soc. 2002, 124, 788-805.

(7) Clay, M. D.; Cosper, C. A.; Jenney, F. E., Jr.; Adams, M. W. W.; Johnson,

M. K. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 3796-3801.

9

J. AM. CHEM. SOC. 2005, 127, 1675-1689

1675

A R T I C L E S

Fiedler et al.

radiolysis^8-10and mutagenesis experiments¹¹and studied with

spectroscopic and computational¹²methods. Although not

confirmed experimentally, it seems likely that the substrate binds

to the Fe(II) center in the vacant position trans to the axial

cysteine ligand.

Scheme 1

The axial cysteine residue in the SOR active site is thought

to play a major role in facilitating superoxide reduction and

product release, primarily through charge donation to the Fe

center. Specifically, spectroscopic studies by Johnson and co-

workers have revealed the presence of a strong Fe-S π-bonding

interaction between one of the S_Cys-based lone pairs and an

Fe(d_π) orbital, which is particularly evident from the intense,

low-energy S_CysfFe(III) charge transfer (CT) transition at

15 150 cm^-1in the electronic absorption spectrum of oxidized

P. furiosus SOR.^6,13The corresponding transition in the reduced

state is found at 31 200 cm^-1.⁶It has been suggested that this

covalent Fe-S π-interaction facilitates electron transfer to the

singly occupied π* acceptor orbital of superoxide and then

promotes dissociation of product peroxide via a trans influence

and/or effect.¹⁴The role of the conserved active-site glutamate

residue is less clear. It was initially proposed that the existence

of a six-coordinate ferric center prevents superoxide binding to

the oxidized enzyme, thus avoiding the possibility of superoxide

dismutation that would be detrimental for the organism.⁴Yet

further studies have found no significant superoxide oxidation

activity upon substitution of this glutamate residue, and the rates

of superoxide reduction displayed by the variants were virtually

unchanged.^8,9Alternatively, mutation of a conserved second-

sphere lysine residue resulted in a large (20-30-fold) reduction

of SOR activity, and it is believed that this residue functions in

proton donation to the nascent peroxo intermediate.⁹

which contains an Fe(II) center supported by the tetradentate

ligand 1,5-bis(2-pyridylmethyl)-1,5-diazacyclooctane (L⁸py₂) in

its equatorial plane and an axial p-toluenethiolate ligand. In an

effort to expand the range of structural and electronic properties

exhibited by SOR_redmodels, we have prepared complexes 2

and 3 (Scheme 1), which are based on the macrocyclic N₄-

donor 1,4,8,11-tetraazacyclotetradecane (cyclam). The synthesis

and structural characterization of these second- and third-

generation SOR_redmodels (2 and 3, respectively) is described

herein.

Despite the success in generating structural and functional

models of the SOR active site, there exist very few studies that

have examined in detail the electronic properties of mononuclear

Fe(II) complexes with mixed N/S ligation, such as 1-3. Given

the importance ascribed to axial thiolate ligation with respect

to SOR catalysis, these complexes provide ideal systems through

which to explore the effects of structural variations on Fe(II)-

thiolate bonding. Therefore, we have engaged in a detailed

investigation of models 1-3 using a combination of spectro-

scopic and computational methods. The electronic absorption

spectra of these models are dominated by SfFe(II) CT

transitions in the visible and near-UV regions that also produce

the major contributions to the corresponding magnetic circular

dichroism (MCD) spectra. To estimate the zero-field splitting

(ZFS) parameters for these complexes, both a quantitative

analysis of variable-temperature variable-field MCD (VTVH-

MCD) data and semiemperical INDO/S-CI calculations were

performed. Density functional theory (DFT) calculations were

used to provide quantitative bonding descriptions of the

Fe(II)-thiolate complexes that were validated on the basis of

our spectroscopic data. Among other findings, these experi-

mentally calibrated electronic structure descriptions establish

an interesting relationship between structural constraints and

Fe(II)-S bonding. The implications of these results with respect

to the current understanding of Fe(II)-ligand interactions in

the SOR_redactive site are also explored.

The emergence of this novel superoxide detoxification

pathway has stimulated the synthesis of structural and functional

models of the SOR active site.¹⁵Kovacs and co-workers have

prepared a five-coordinate Fe(II) model with a N₄S ligand set

that reacts with superoxide to yield H₂O₂.¹⁶When the reaction

is performed at low temperature (-90 °C), they observe a six-

coordinate [Fe(III)(η¹-OOH)] intermediate with a low-spin

1

(S ) /₂) ferric center,¹⁷in which the hydroperoxo ligand is

likely trans to one of the N-donor ligands rather than to the

thiolate as in the putative protein intermediate. In a previous

study, we presented the synthesis of a series of square-pyramidal

Fe(II)-thiolate models that feature a vacant coordination site

trans to the axial thiolate ligand.¹⁸This first generation of SOR_red

model complexes is represented in Scheme 1 by complex 1,

(8) Coulter, E. D.; Emerson, J. P.; Kurtz, D. M., Jr.; Cabelli, D. E. J. Am.

Chem. Soc. 2000, 122, 11555-11556.

(9) Lombard, M.; Houee-Levin, C.; Touati, D.; Fontecave, M.; Niviere, V.

Biochemistry 2001, 40, 5032-5040.

(10) (a) Abreu, I. A.; Saraiva, L. M.; Soares, C. M.; Teixeira, M.; Cabelli, D.

E. J. Biol. Chem. 2002, 276, 38995-39001. (b) Niviere, V.; Lombard, M.;

Fontecave, M.; Houee-Levin, C. FEBS Lett. 2001, 497, 171-173.

(11) Mathe, C.; Mattioli, T. A.; Horner, O.; Lombard, M.; Latour, J.-M.;

Fontecave, M.; Niviere, V. J. Am. Chem. Soc. 2002, 124, 4966-4967.

(12) Silaghi-Dumitrescu, R.; Silaghi-Dumitrescu, I.; Coulter, E. D.; Kurtz, D.

M., Jr. Inorg. Chem. 2003, 42, 446-456.

Experimental and Computational Procedures

(13) Clay, M. D.; Jenney, F. E., Jr.; Noh, H. J.; Hagedoorn, P. L.; Adams, M.

W. W.; Johnson, M. K. Biochemistry 2002, 41, 9833-9841.

(14) Adams, M. W. W.; Jenney, F. E., Jr.; Clay, M. D.; Johnson, M. K. J. Biol.

Inorg. Chem. 2002, 7, 647-652.

Materials and Methods. All reagents, including the macrocyclic

ligand Me₄cyclam, were obtained from commercial sources. [Fe-

(MeCN)₂(OTf)₂] was prepared by the reaction of anhydrous FeCl₂with

a stoichiometric quantity of triflic acid in anhydrous MeCN, followed

by precipitation with anhydrous diethyl ether.¹⁹Solvents were purified

(15) Kovacs, J. A. Chem. ReV. 2004, 104, 825-848.

(16) Shearer, J.; Nehring, J.; Lovell, S.; Kaminsky, W.; Kovacs, J. A. Inorg.

Chem. 2001, 40, 5483-5484.

(17) Shearer, J.; Scarrow, R. C.; Kovacs, J. A. J. Am. Chem. Soc. 2002, 124,

11709-11717.

(18) Halfen, J. A.; Moore, H. L.; Fox, D. C. Inorg. Chem. 2002, 41, 3935-

3943.

(19) Dixon, N. E.; Lawrance, G. A.; Lay, P. A.; Sargeson, A. M.; Taube, H.

Inorg. Synth. 1990, 28, 70-76.

9

1676 J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005

Synthesis/Characterization of Fe(II)−Thiolate Complexes

A R T I C L E S

according to standard methods. All reactions were conducted, and

products were handled, under an inert atmosphere using a Vacuum

Atmospheres glovebox. Electrochemical experiments were conducted

with a BAS Epsilon potentiostat, using a platinum disk working

electrode, a silver wire pseudo-reference electrode, and a platinum wire

auxiliary electrode. Cyclic voltammograms were obtained in MeCN

(0.1 M n-Bu₄NClO₄) with an analyte concentration of 1 mM. Ferrocene

was used as an internal standard, and its redox potential under the

experimental conditions was adjusted to the literature value (+380 mV

vs SCE).²⁰Other analyte redox potentials were shifted accordingly.

Solid-state magnetic susceptibilities were determined at room temper-

ature using a Johnson-Matthey magnetic susceptibility balance.

Elemental analyses were performed by Atlantic Microlabs (Norcross,

GA).

[(Me₄cyclam)Fe(SC₆H₄-p-OMe)]OTf (2): A solution of Me₄cyclam

(51 mg, 0.199 mmol) dissolved in THF (5 mL) was treated with [Fe-

(MeCN)₂(OTf)₂] (84 mg, 0.193 mmol). After stirring for 20 min, solid

NaSC₆H₄-p-OMe (34 mg, 0.210 mmol) was added to the solution,

resulting in the development of a light yellow color. After 2 h, the

solvent was removed from the mixture under vacuum, and the solid

residue was extracted into CH₂Cl₂(5 mL) and then filtered through a

glass wool plug. Addition of Et₂O to the filtrate provided the crude

product as an off-white solid, which was recrystallized from CH₂Cl₂/

Et₂O to afford pure 2 as very light pink prisms, 55 mg (47%). Anal.

Calcd for C₂₂H₃₉F₃FeN₄O₄S₂: C, 44.00; H, 6.54; N, 9.33. Found: C,

44.01; H, 6.46; N, 9.27.

[LFe]PF₆(3): A solution of 1-thioethyl-4,8,11-trimethylcyclam (127

mg, 0.420 mmol)²¹in MeOH (5 mL) was treated with LiOH‚H₂O (17.5

mg, 0.417 mmol). Solid [Fe(MeCN)₂(OTf)₂] (182 mg, 0.418 mmol)

was added after stirring for 30 min, and stirring continued for an

additional 30 min. The cloudy mixture was then filtered through a Celite

pad, and a solution of NH₄PF₆(215 mg, 1.32 mmol) in MeOH (3 mL)

was added, resulting in the deposition of a crystalline solid. The mixture

was filtered, and the solid was redissolved in a mixture of CH₂Cl₂and

MeCN (2:1, 10 mL). This solution was filtered through a Celite pad,

and the filtrate was treated with excess diethyl ether to produce 3 as a

colorless crystalline solid, 125 mg (60%). Anal. Calcd for C₁₅H₃₃F₆-

FeN₄PS‚0.25CH₂Cl₂: C, 34.98; H, 6.45; N, 10.70. Found: C, 35.08;

H, 6.45; N, 10.79. The presence of fractional solvent in the crystal

lattice of 3 was confirmed by X-ray crystallography.

X-ray Crystallography. Single crystals were mounted in thin-walled

glass capillaries and transferred to a Bruker-Nonius MACH3S X-ray

diffractometer for data collections at -150(2) °C (2) or -100(2) °C

(3) using graphite monochromated Mo KR (λ ) 0.710 73 Å) radiation.

Unit cell constants were determined from a least-squares refinement

of the setting angles of 25 intense, high angle reflections. Intensity

data were collected using the ω/2θ scan technique to a maximum 2θ

value of 53.94° (2) or 49.94° (3). The data were corrected for Lorentz

and polarization effects and converted to structure factors using the

teXsan for Windows crystallographic software package.^22aSpace groups

were determined based on systematic absences and intensity statistics.

Successful direct-method solutions were calculated for each compound

using the SHELXTL suite of programs.^22bNon-hydrogen atoms that

were not identified from the initial E-map were located after several

cycles of structure expansion and full matrix least squares refinement

on F². Hydrogen atoms were added geometrically. All non-hydrogen

atoms were refined with anisotropic displacement parameters, while

hydrogen atoms were refined using a riding model with group isotropic

displacement parameters. Relevant crystallographic results for 2 and 3

are summarized in Tables S1 and S2. Complete crystallographic data

for each compound in CIF format are provided as Supporting

Information.

The asymmetric unit of 2 includes two crystallographically inde-

pendent formula units. One of the two CF₃SO₃ions is rotationally

-

disordered about the C-S bond; a split-atom model featuring two

independent conformations for the anion, with refined occupancies of

0.597(7) and 0.403(7), was used to account for the disorder. Bond length

and angle constraints were applied to the disordered anion. The absolute

structure of the sample was assigned by consideration of the Flack x

parameter, in this case 0.00(2). The X-ray crystal structure of 3 contains

a disordered cation, two partially occupied yet ordered anion positions,

and a disordered, half-occupied CH₂Cl₂solvate that resides on a special

position possessing 3h site symmetry. The Fe center in 3 resides on a

crystallographic 2-fold axis, necessitating 2-fold rotational disorder for

the cation. In addition, there is pseudo-mirror disorder within the cation,

resulting in each alkyl spacer between adjacent nitrogen donors

appearing as a statistical combination of ethylene (-C₂H₄-) and

propylene (-C₃H₆-) groups. Bond length and angle constraints were

applied to all of the disordered alkyl groups. Analysis of the final refined

structure using the PLATON suite of programs²³revealed a potential

alternative space group assignment (P6₃/mcm). However, the structure

of 3 could not be solved in this alternative space group, and an attempt

to transform the structure that was solved and refined in P3hc1 into

P6₃/mcm resulted in an unstable refinement. Thus, we conclude that

the appropriate space group for 3 is P3hc1, and that the alternative space

group identified by PLATON is disallowed by the substantial cation

disorder. Attempts to disrupt the cation disorder in 3 by anion metathesis

(ClO₄^-, CF₃SO₃^-, BPh₄^-) resulted in either more significant disorder

or in poorly diffracting samples. No attempts were made to locate or

calculate the positions of the hydrogen atoms of the half-occupied,

disordered CH₂Cl₂solvate. These atoms are omitted from the empirical

formula and the calculations of F₀₀₀and d_calcdfor 3.

Absorption and MCD Spectroscopies. Room-temperature elec-

tronic absorption spectra of 1-3 were measured using a Hewlett-

Packard 8453 spectrophotometer. Variable temperature absorption and

MCD spectra were recorded using a Jasco J-715 spectropolarimeter in

conjunction with an Oxford Instruments SM-4000 8T magnetocryostat.

The solid-state samples of 1-3 used for low-temperature absorption

and MCD studies were prepared as uniform mulls in poly(dimethyl-

siloxane) (Aldrich). Frozen solution samples were prepared by dis-

solving the complexes in a 2:1 mixture of MeOH and glycerol.

VTVH-MCD data were analyzed using the software program

developed by Dr. Frank Neese (MPI Mu¨lheim an der Ruhr, Germany).²⁴

To locate the best overall fit for each model 1-3, the D and E/D values

were systematically varied over the ranges of -20 cm^-1< D < 20

cm^-1and 0 < E/D < 0.33. The g-values used in the fitting routine

were obtained from INDO/S-CI calculations performed on 1-3 (vide

infra).²⁵For each set of ZFS parameters the three transition moment

products (M_xy, M_xz, and M_yz) were adjusted to achieve the best agreement

between the experimental and calculated VTVH-MCD data, as quanti-

fied by the goodness-of-fit parameter (ø²):

N

ø²)

(f⁽_i^calcd)- f_i^{(expt) 2}

)

∑

i)1

Computational Methods. (a) Computational Models. For complex

1, the computational model used in DFT and INDO/S-CI calculations

was based completely on the X-ray crystal structure¹⁸and included

the entire complex except for the counteranion. To obtain a computa-

tional model of complex 2, the coordinates of the first-sphere atoms

(consisting of the FeSN₄core, along with C₁of the aryl ring) of the

two structurally distinct cations in the X-ray structure were averaged

to produce a composite structure, and the positions of all remaining

atoms in the cyclam and aryl rings were geometry-optimized. Similarly,

(20) Connelly, N. G.; Geiger, W. E. Chem. ReV. 1996, 96, 877-910.

(21) Halfen, J. A.; Young, V. G., Jr. Chem. Commun. 2003, 2894-2895.

(22) (a) TeXsan for Windows V 1.02; Molecular Structure Corporation, Inc.:

The Woodlands, TX. (b) SHELXTL V. 5.1 for Windows NT; Bruker AXS:

Madison, WI.

(23) Spek, A. L. J. Appl. Crystallogr. 2003, 36, 7-13.

(24) (a) Neese, F.; Solomon, E. I. Inorg. Chem. 1998, 37, 6568-6582. (b) Neese,

F.; Solomon, E. I. Inorg. Chem. 1999, 38, 1847-1865.

9

J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005 1677

A R T I C L E S

Fiedler et al.

Scheme 2

the crystallographic disorder in the positions of the cyclam ring atoms

of complex 3 required optimization of all atoms beyond the first sphere.

All geometry optimizations were performed with the Amsterdam

Density Functional (ADF) 2002.03 software package²⁶on a cluster of

20 Intel Xeon processors (Ace computers) using ADF basis set IV

(triple-ú with single polarization on the ligand atoms), an integration

constant of 4.0, and the Vosko-Wilk-Nusair²⁷local density ap-

proximation with the nonlocal gradient corrections of Becke²⁸and

Perdew.²⁹The Cartesian coordinates for the DFT geometry-optimized

models of 2 and 3 are provided in Tables S3 and S4 (see ref 18 for

structural parameters of 1).

(b) DFT Calculations. All single-point DFT calculations were

performed using the ORCA 2.2 software package developed by Neese.³⁰

Computations on models 1-3 were carried out using a spin unrestricted

formalism, while those on the free ligands were performed spin

restricted. For the first-sphere atoms (Fe, N, S), Ahlrichs’ valence

triple-ú basis set³¹with one (N) or two (Fe, S) sets of polarization

functions³²were used. The remaining atoms utilized Ahlrich’s split

valence basis set³¹with one set of polarization functions on all non-

hydrogen atoms, along with the corresponding auxiliary basis set (SV/

J).³³The calculations were performed with the local density approxi-

mation of Perdew and Wang³⁴and the nonlocal corrections of Becke²⁸

and Perdew,²⁹using an integration grid of 4.0. The gOpenMol program

developed by Laaksonen³⁵was used to generate boundary surface plots

of molecular orbitals (using an isodensity value of 0.05 b^-3).

by the tetradentate ligand L⁸py₂in its equatorial plane and an

axial p-toluenethiolate ligand, complexes 2 and 3 (Scheme 1)

utilize the macrocyclic N₄-donor 1,4,8,11-tetraazacyclotetrade-

cane (cyclam). The coordination sphere of 2 includes the tetra-

N-methylated macrocycle Me₄cyclam and is completed by a

4-methoxybenzenethiolate ligand, whereas complex 3 is coor-

dinated by a single pentadentate ligand, derived from cyclam,

which features a pendant thioethyl group.²¹

(c) Semiemperical INDO/S-CI Calculations. Semiemperical cal-

culations employing the INDO/S model developed by Zerner and co-

workers³⁶were also performed using the ORCA program.³⁰The

calculations used the valence-shell ionization potentials and Slater-

Condon parameters listed by Bacon and Zerner,³⁷the standard interac-

tion factors f_pσpσ) 1.266 and f_pπpπ) 0.585, and the following spin-

A convenient one-pot synthesis was used to prepare complex

2, a methodology that is equally applicable to related complexes

containing other aryl- or alkylthiolate ligands (Scheme 2).

Specifically, equimolar amounts of Me₄cyclam and [Fe(MeCN)₂-

(OTf)₂] were combined in THF, generating [(Me₄cyclam)Fe-

(OTf)]OTf in situ.³⁸Addition of sodium 4-methoxybenzeneth-

iolate, followed by recrystallization of the crude product from

CH₂Cl₂, afforded the SOR_redmodel complex 2 as a colorless

crystalline solid in moderate yield (47%).³⁹The synthesis of 3,

which possesses a thioethyl-pendant trimethylcyclam ligand, was

achieved following a different strategy. The pentadentate ligand

suffers from facile aerobic degradation, and as a result, small

quantities of the ligand were prepared by the reaction of

trimethylcyclam with thirane immediately prior to metalation.

Deprotonation of the thiol-pendant ligand with LiOH in MeOH,

followed by metalation with [Fe(MeCN)₂(OTf)₂] and anion

metathesis with NH₄PF₆, afforded complex 3 as a colorless solid

in yields approaching 60% (Scheme 2). Interestingly, 3 could

be isolated in reasonable yield only when prepared in a protic

solvent (MeOH); attempts to prepare 3 under anhydrous

conditions (deprotonation using NaH followed by metalation

in THF) led to substantially reduced yields of product. Appar-

ently, the basic nature of the reaction medium is sufficient to

allow metalation, rather than protonation, of the pentadentate

ligand.

orbit coupling constants: ú_3d(Fe) ) 400 cm^-1, ú_4p(Fe) ) 445 cm^-1

ú_3p(S) ) 250 cm^-1, ú_2p(N) ) 76 cm^-1, and ú_2p(O) ) 150 cm^-1

,

.

Restricted open-shell Hartree-Fock (ROHF) SCF calculations were

converged onto the spin quintet ground states of models 1-3, which

served as the reference states for configuration interaction (CI)

calculations. Descriptions of the active spaces used for all INDO/S-CI

calculations of 1-3 are provided in the Supporting Information.

Results and Analysis

Complex Design and Syntheses. Our first generation of

SOR_redmodel complexes is represented by complex 1 (Scheme

1), the synthesis of which was described earlier.¹⁸While

complex 1 contains a square-pyramidal Fe(II) center supported

(25) In all cases, the g-values were in the range 2.01-2.08, and variations in

these values had only negligible effects on the quality of the resulting fits.

Specifically, our analysis revealed that the overall shapes of the VTVH-

MCD magnetization curves are determined principally by the ZFS

parameters, not the g-values.

(26) (a) Baerends, E. J.; Ellis, D. E.; Ros, P. Chem. Phys. 1973, 2, 41. (b)

Versluis, L.; Ziegler, T. J. Chem. Phys. 1988, 88, 322-328. (c) te Velde,

G.; Baerends, E. J. J. Comput. Phys. 1992, 99, 84-98. (d) Guerra, C. F.;

Snijders, J. G.; te Velde, G.; Baerends, E. J. Theor. Chem. Acc. 1998, 99,

391-403.

(27) Vosko, S. H.; Wilk, L.; Nusair, M. Can. J. Phys. 1980, 58, 1200-1211.

(28) Becke, A. D. J. Chem. Phys. 1986, 84, 4524-4529.

(29) Perdew, J. P. Phys. ReV. B: Condens. Matter 1986, 33, 8822-8824.

(30) Neese, F. ORCA, version 2.2.; an ab initio, density functional, and

semiempirical program package; Max-Plack Institut fu¨r Bioanorganische

Chemie: Mu¨lheim an der Ruhr, Germany, 2001.

Structural, Magnetic, and Electrochemical Characteriza-

tion. The solid-state structures of 2 and 3 were determined by

X-ray crystallography. A summary of the crystallographic results

is presented in Table S1, and significant interatomic distances

(31) Schaefer, A.; Horn, H.; Ahlrichs, R. J. Chem. Phys. 1992, 97, 2571.

(32) Ahlrichs, R. Unpublished results.

(33) (a) Eichkorn, K.; Treutler, O.; Ohm, H.; Haser, M.; Ahlrichs, R. Chem.

Phys. Lett. 1995, 240, 283. (b) Eichkorn, K.; Weigend, F.; Treutler, O.;

Ahlrichs, R. Theor. Chem. Acc. 1997, 97, 119.

(38) Rohde, J.-U.; In, J. H.; Lim, M. H.; Brennessel, W. W.; Bukowski, M. R.;

Stubna, A.; Munck, E.; Nam, W.; Que, L., Jr. Science 2003, 299, 1037-

1039.

(34) Perdew, J. P.; Wang, Y. Phys. ReV. B: Condens. Matter 1992, 45, 13244-

13249.

(35) (a) Laaksonen, L. J. Mol. Graphics 1992, 10, 33. (b) Bergman, D. L.;

Laaksonen, L.; Laaksonen, A. J. Mol. Graphics Modell. 1997, 15, 301.

(36) (a) Ridley, J.; Zerner, M. C. Theor. Chem. Acc. 1973, 32, 111. (b) Zerner,

M. C.; Loew, G. H.; Kirchner, R. F.; Mueller-Westerhof, U. T. J. Am.

Chem. Soc. 1980, 102, 589.

(39) The use of hydrated Fe(II) sources or protic solvents during the metalation

of Me₄cyclam led to very low yields of the final Fe(II)-thiolate complexes,

presumably due to protonation of the macrocyclic ligand. This synthetic

difficulty was not encountered during the preparation of 1 or related SOR_red

models supported by L⁸py₂, which reflects the enhanced proton affinity of

Me₄cyclam over the pyridyl-pendant diazacyclooctane ligand.

(37) Bacon, A. D.; Zerner, M. C. Theor. Chem. Acc. 1979, 53, 21.

9

1678 J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005

Synthesis/Characterization of Fe(II)−Thiolate Complexes

A R T I C L E S

geometry, as indicated by the shape-defining parameter τ (0.57

and 0.46 for the two crystallographically independent cat-

ions).^40,41The distortions of these Fe(II)-thiolate cations toward

a more trigonal-bipyramidal geometry gives rise to asymmetry

in their Fe-N bond lengths, such that the Fe-N bonds in the

pseudo-trigonal planes (including the sulfur donors) are shorter

than the “axial” Fe-N bonds. Differences in the metrical

parameters between the two independent cations illustrate the

structural flexibility provided by the Me₄cyclam macrocycle.

While differences in bond lengths between pairs of equivalent

Fe-ligand bonds in the two cations are generally small (0.01-

0.03 Å), some equivalent N-Fe-N bond angles vary by as

much as 6°, and bond angles involving the sulfur donor differ

by as much as 10°.

As described in the Experimental Section, substantial 2-fold

rotational and pseudo-mirror disorder in the Fe(II) cation of 3

precludes a detailed analysis of the structural properties of this

complex. However, a cursory examination of the structure of 3

reveals some of the same structural features identified in the

X-ray crystal structure of 2. Specifically, the Fe(II) center in 3

adopts a distorted pentacoordinate geometry (τ ) 0.50)⁴¹with

the pendant thioalkyl arm residing on the same face of the

macrocycle as the three N-methyl groups (Figure 2). The

macrocyclic ligand is again found in the trans-I conformation,

and the asymmetry in the Fe-N bond lengths identified in 2

persists in 3. While the Fe-S bond length in 3 is similar to

that in 2, the constrained nature of the five-membered chelate

ring that includes the thiolate donor causes the Fe-S-C bond

angle in 3 to be significantly more acute than that in 2:

98.0(5)° in 3 vs 115.9(2)° and 125.3(2)° for the two independent

cations in 2.

Figure 1. Thermal ellipsoid representation (50% probability boundaries)

of one of the two crystallographically independent cations in the X-ray

crystal structure of 2, with hydrogen atoms omitted for clarity. Significant

interatomic distances (Å) and angles (°) include the following: Fe1-N1,

2.183(6); Fe1-N2, 2.279(6); Fe1-N3, 2.186(5); Fe1-N4, 2.237(5); Fe1-

S1, 2.3220(18); N1-Fe1-N2, 90.9(2); N1-Fe1-N3, 136.7(2); N1-Fe1-

N4, 84.4(2); N2-Fe1-N3, 81.4(2); N2-Fe1-N4, 164.1(2); N3-Fe1-

N4, 91.6(2); S1-Fe1-N1, 113.33(15); S1-Fe1-N2, 92.30(15); S1-Fe1-

N3, 109.58(15); S1-Fe1-N4, 103.53(15); Fe1-S1-C15, 125.3(2).

Similar to 1, both 2 and 3 contain high-spin Fe(II) centers as

suggested by their Fe-N bond lengths⁴²and verified by room-

temperature magnetic susceptibility measurements (2, µ_eff) 4.9

â; 3, µ_eff) 5.1 â). The high-spin, S ) 2 ground states of these

Fe(II) complexes parallel that of the SOR_redactive site.^3,6

Complex 1 was previously shown to undergo irreversible

oxidation at a potential (+615 mV vs SCE) that renders it inert

toward aerobic oxidation.¹⁸The overall electrochemical proper-

ties of 2 and 3 are similar to those of 1 in that both complexes

are irreversibly oxidized, albeit at potentials slightly depressed

relative to that of 1 (2, +570 mV; 3, +530 mV vs SCE). The

lower redox potentials may be due to the more electron-donating

thiolate ligands and/or more basic macrocycles in complexes 2

and 3, factors that would be expected to aid in the stabilization

of the Fe(III) state. However, the oxidized counterparts of 2

and 3 have yet to be isolated as stable materials.

Figure 2. Ball and stick representation of the cationic portion of the X-ray

crystal structure of 3 with hydrogen atoms omitted for clarity. Significant

interatomic distances (Å) and angles (°) include the following: Fe1-N1,

2.176(5); Fe1-N2, 2.225(5); Fe1-S1, 2.297(3); N1-Fe1-N2, 87.4(2);

N1-Fe1-N1A, 137.6(3); N1-Fe1-N2A, 88.0(2); N2-Fe1-N2A,

167.2(3); N1A-Fe1-N2A, 87.4(2), S1-Fe1-N1, 111.2(4); S1-Fe1-N2,

89.5(2); S1-Fe1-N1A, 110.8(4); S1-Fe1-N2A, 103.3(2); Fe1-S1-C1,

98.0(5).

Electronic Absorption Spectroscopy. The room temperature

(rt) electronic absorption spectra of complex 1 in MeCN and

complexes 2 and 3 in CH₂Cl₂are shown in Figure 3. Although

the bands of 2 are uniformly more intense and ∼4000 cm^-1

higher in energy than those of 1, the spectra of 1 and 2 bear a

close resemblance to each other in terms of their overall

and angles are presented in the captions of Figures 1 and 2.

Full details of the structure determinations (in CIF format) are

provided as Supporting Information.

The asymmetric unit of 2 contains two compositionally

identical formula units that exhibit some slight metrical differ-

ences; a thermal ellipsoid representation of one of the two

independent Fe(II)-thiolate cations in 2 is shown in Figure 1.

Each of the two cations contains a pentacoordinate Fe(II) ion

ligated by the Me₄cyclam macrocycle and a single thiolate

ligand. In each case, the macrocycle is found in the trans-I

conformation, with the thiolate ligand and the four N-methyl

groups occupying the same face of the complex. Both cations

exhibit significant distortions from an idealized square-pyramidal

(40) The geometric parameter τ is defined as τ ) |(â - R)|/60, where R and â

are the two trans basal angles in psuedo-square pyramidal geometry (see

ref 41). In idealized square-planar geometries, R ) â, but trigonal distortions

result in R * â.

(41) Addison, A. W.; Rao, T. N.; Reedijk, J.; van Rijn, J.; Verschoor, G. C. J.

Chem. Soc., Dalton Trans. 1984, 1349-1356.

(42) (a) Blakesley, D. W.; Payne, S. C.; Hagen, K. S. Inorg. Chem. 2000, 39,

1979-1989. (b) Diebold, A.; Hagen, K. S. Inorg. Chem. 1998, 37, 215-

223.

9

J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005 1679

A R T I C L E S

Fiedler et al.

Figure 3. Room-temperature electronic absorption spectra of complex 1

in MeCN solution and of complexes 2 and 3 in CH₂Cl₂solution.

Table 1. Summary of Electronic Absorption Features for 1-3

1

1 c

complex

solvent^a(T)^b

λ (ꢀ )

_max, cm^-, M^-1cm^-

1

2

MeCN (rt)

CH₂CH₂(rt)

MeOH (rt)

29 200 (1800), sh 24 000 (400)

32 600 (4000), sh ∼28 000 (600)

34 000 (1700), sh ∼31 500 (900),

sh ∼29 000 (200)

solid state (15 K)

CH₂CH₂(rt)

31 100, sh ∼27 000

3

38 100 (2000), 34 700 (1400),

30 400 (1800)

MeOH (rt)

39 500 (2600), 35 900 (1500),

31 200 (1800)

solid state (15 K)

38 000, 35 000, 30 500, sh 28 900

^aSolid state ) mull in poly(dimethylsiloxane). ^brt ) room temperature.

^cꢀ values are not available for solid-state data.

appearance. Both spectra are comprised of two main features:

(i) a low-energy shoulder of weak intensity and (ii) a more

intense band to higher energy (Table 1). In contrast, the

absorption spectrum of complex 3 is composed of three features

of approximately equal intensity centered at 30 400, 34 700, and

38 000 cm^-1. As none of these absorption features of 1-3 are

present in the precursor Fe(II)N₄complexes that lack thiolate

ligation, they can be attributed to SfFe(II) charge transfer (CT)

transitions.⁴³At energies > 40 000 cm^-1, the complexes exhibit

intense ligand-based and NfFe(II) CT transitions.

The absorption spectra of complexes 2 and 3 were also studied

in MeOH (Figure S1), a solvent with the potential of coordinat-

ing to the Fe(II) center. For 3, the major features exhibit blue-

shifts of 800-1400 cm^-1when the CH₂Cl₂solvent is replaced

by MeOH, although the general appearance of the spectrum

remains the same (Table 1). For 2, this change in solvent results

in a 1400 cm^-1blue-shift of the most intense absorption band

and a dramatic overall reduction in band intensities, implying

a more significant perturbation of the ligand environment of

the Fe(II) center. Based on these results, we conclude that MeOH

is coordinating to the Fe(II) centers of both complexes, although

the interaction appears to be stronger for complex 2 than for 3.

To probe the electronic structures of complexes 1-3 in the

solid state where possible complications due to solvent coor-

dination are avoided, the corresponding 15 K mull absorption

spectra were also obtained (Figure S2). While the positions of

the low-energy bands remain essentially unchanged for 1, the

most intense absorption feature of 2 now appears at 31 100

Figure 4. Gaussian-resolved (‚‚‚‚) solid-state MCD spectra (s) of

complexes 1-3. Spectra were obtained with a magnetic field of 7 T at a

temperature of 4 K for 1 and 2 and 20 K for 3. Band energies and parameters

are summarized in Table 2. Note that each spectrum is plotted over a

different energy range (x-axes are not the same for 1-3).

cm^-1, compared to 32 600 cm^-1in CH₂Cl₂and 34 000 cm^-1

in MeOH (Table 1). In the solid-state spectrum of complex 3,

the sharpening of the transitions that give rise to the low-energy

band reveals the presence of an additional weak feature at 28 900

cm^-1, alongside the dominant band at 30 500 cm^-1. The two

higher-energy bands appear at 35 000 and 38 000 cm^-1in the

solid state (Table 1).

Solid-State (Mull) MCD Spectroscopy. Figure 4 displays

the solid-state (mull) MCD spectra obtained for complexes 1-3.

All band intensities increase with decreasing temperature (data

not shown), consistent with the expected C-term behavior of

these paramagnetic S ) 2 spin systems. Simultaneous fitting

of the low temperature (LT) absorption and MCD spectra

required 8-10 Gaussian bands (dotted lines in Figure 4) of

roughly equal bandwidths, with peak positions and intensities

as indicated in Table 2 (see Figure S3 for Gaussian fits of the

absorption data). The MCD spectrum of 1 is characterized by

two broad positive features that coincide with the two dominant

CT bands in the absorption spectrum. While the MCD spectrum

of 2 also exhibits a positively signed low-energy feature, the

high-energy region instead contains large contributions from

negatively signed transitions.⁴⁵The transitions that give rise to

the most intense peak in the LT absorption spectrum of 2 appear

(43) The ligand field (LF) transitions of high-spin Fe(II) complexes, which

typically appear in the NIR region, are very weak in both absorption and

MCD spectra.⁴⁴Consequently, our absorption and solid-state MCD studies

of 1-3 failed to detect any such transitions. In constrast, MCD spectra

obtained with a concentrated (10 mM) solution of 1 in 1:1 propionitrile/

butyronitrile solvent did reveal a LF band at ∼15 000 cm^-1(data not

shown).

(44) (a) Pavel, E. G.; Kitajima, N.; Solomon, E. I. J. Am. Chem. Soc. 1998,

120, 3949-3962. (b) Solomon, E. I.; Pavel, E. G.; Loeb, K. E.;

Campochiaro, C. Coord. Chem. ReV. 1995, 144, 369-460.

(45) Although the solid-state sample of 2 contains two crystallographically

distinct complexes (cations I and II), the structural differences are too small

to cause the appearance of two sets of bands that can be resolved in the

MCD spectrum but may possibly lead to a general broadening of the bands.

9

1680 J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005

Synthesis/Characterization of Fe(II)−Thiolate Complexes

A R T I C L E S

Table 2. Fit Parameters from Gaussian Deconvolutions of the

Experimental Absorption and MCD Spectra of Complexes 1-3

Complex 1

energy

ꢀ^a

osc. str.

rel. MCD

intensity^c

1

band

(cm^-

)

(M^-¹cm^-

)

(f_exp

×

10³)^b

1

2

3

4

5

6

7

8

9

21 000

22 400

23 750

25 850

27 500

28 950

30 600

32 250

34 500

35 600

290

300

270

480

850

1270

970

800

470

410

2.7

2.8

2.5

4.4

7.2

12.7

9.7

8.0

4.7

3.8

70

35

21

-10

15

41

100

-7

32

10

-38

Complex 2

energy

ꢀ^a

osc. str.

rel. MCD

intensity^c

1

band

(cm^-

)

(M^-¹cm^-

)

(f_exp

×

10³)^b

1

2

3

4

5

6

7

8

9

25 500

27 000

28 200

29 650

30 850

32 050

33 100

34 100

35 600

390

690

3.6

6.4

30

100

37

-50

79

-65

11

-73

55

1260

1540

2270

2470

1060

2020

1600

11.6

14.2

20.9

26.5

8.9

18.6

16.0

Complex 3

Figure 5. Experimental VTVH-MCD data collected at the indicated

wavelengths for complexes 1-3 (solid lines), plotted against âH/2kT.

VTVH-MCD experiments were performed by measuring the signal intensity

at four different temperatures (2, 4, 8, and 15 K) as a function of magnetic

field between 0 and 7 T. Theoretical fits (X) were obtained using the “best

fit” ZFS values listed in Table 3.

energy

ꢀ^a

osc. str.

rel. MCD

intensity^c

1

band

(cm^-

)

(M^-¹cm^-

)

(f_exp

×

10³)^b

1

2

3

4

5

6

7

8

28 750

30 450

31 450

34 650

35 700

36 650

38 000

41 250

400

1350

740

960

410

200

1370

560

3.4

11.4

6.3

8.1

3.5

1.8

13.6

6.0

33

100

-92

43

-71

35

(30 300 cm^-1) and 470 nm (21 300 cm^-1) for 1, 307 nm (32 600

cm^-1) and 370 nm (27 000 cm^-1) for 2, and 281 nm (35 600

cm^-1) and 314 nm (31 800 cm^-1) for 3. These positions were

selected so as to coincide with the dominant CT absorption

features exhibited by the three complexes in the solid state.

Figure 5 shows the low-energy VTVH-MCD data sets obtained

for 1-3 at the wavelengths indicated; the corresponding high-

energy data sets are shown in Figure S4.

-5

24

^aEpsilon values (ꢀ) for the low-temperature mull spectra were estimated

by comparison to the corresponding rt solution spectra, for which ꢀ-values

are available. f_exp) (4.61 × 10^-9)ꢀ_maxν_1/2

.

^cSince ∆ꢀ values are not

b

available for mull spectra, only relative intensities are reported. The largest

peak is scaled to 100.

As discussed in seminal papers by Neese and Solomon, the

VTVH-MCD saturation behavior of S > /₂systems depends

in the MCD spectrum as a derivative-shaped feature centered

at 31 600 cm^-1, consisting of two bands (bands 5 and 6) of

nearly equal intensity but opposite sign. Such features in MCD

spectra are commonly referred to as pseudo-A terms and require

that the two transitions involved be differently polarized. In the

high-energy region, the MCD spectra of both 1 and 2 exhibit

similar pseudo-A terms centered near 35 000 cm^-1(bands 9/10

and 8/9, respectively).

1

on the ZFS parameters (D, E/D), the three g-values, and the

polarization of the corresponding transition with respect to the

molecular coordinate system (as defined by the D-tensor).²⁴As

is evident in Figure 5, the VTVH curves obtained for complexes

1 and 3 are quite similar in terms of general appearance but

distinctly different from those obtained for complex 2, indicating

a substantial difference in ZFS parameters and/or transition

polarizations. For 1 and 3, the MCD signal intensity saturates

slowly with magnetic field (H), and the curves obtained at

different temperatures exhibit a large spread (i.e., the curves

are highly “nested”), while the corresponding curves for 2

saturate quickly and display little nesting. By comparison to

other Fe(II) complexes, the VTVH-MCD results for 1 and 3

are characteristic of systems with positive ZFS (D > 0 cm^-1),

while those for 2 suggest a negative ZFS (D < 0 cm^-1).⁴⁴These

qualitative conclusions were verified by quantitatively analyzing

the VTVH-MCD data within the theoretical framework devel-

oped by Neese and Solomon.²⁴With this type of analysis it is

possible to accurately estimate the ZFS parameters of a given

system, as we recently demonstrated in a combined high-field

The MCD spectrum of 3 reveals three closely spaced bands

in the low-energy region: a weak positive feature at 28 750

cm^-1(band 1) and an intense pseudo-A term feature centered

at 30 900 cm^-1(bands 2 and 3). These three transitions combine

to produce the intense band at 30 400 cm^-1in the corresponding

rt absorption spectrum. Similarly, the two transitions that

produce the pseudo-A term centered at 35 000 cm^-1(bands 4

and 5) are the primary contributors to the second absorption

band. Notably, the transition that gives rise to the highest-energy

absorption band at 38 000 cm^-1(band 7) carries very little

intensity in the MCD spectrum of 3.

VTVH-MCD Data Analysis. For each complex, VTVH-

MCD data sets were collected at two wavelengths: 330 nm

9

J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005 1681

A R T I C L E S

Fiedler et al.

Table 3. Comparison of Experimental and INDO/S-CI Calculated ZFS Parameters for Complexes 1-3

experimental ZFS parameters^a

INDO/S-CI

probe

D

(cm^-

1

complex

wavelength (wavenumbers)

D (cm^-

)

E/D

)

E/D

1

470 nm

best fit

range^b

best fit

range^b

best fit

range^b

+ 9.0

0.20

+3.7

-4.7

-3.8

0.24

(21 300 cm^-1

370 nm

)

D > +4.0

-12.0

0.10 < E/D <0.25

0.20

0.15 < E/D <0.30

0.05

0.04 < E/D < 0.15

2

3

0.12

0.23

(27 000 cm^-1

314 nm

D < -9.0

+5.0

+4.0 < D < +11.0

(31 800 cm^-1

^aExperimental values were obtained via analysis of VTVH-MCD data measured at the indicated probe wavelength. ^bThe D and E/D values within the

given ranges (or limits) provide fits of the VTVH-MCD data in which the resulting ø²-value is e1.5 times the ø²-value obtained with the “best fit” values.

high-frequency EPR (HFEPR)/MCD study of a high-spin Co-

(II) system.⁴⁶The results from our VTVH-MCD data analysis

for complexes 1-3 are summarized in Table 3.⁴⁷

are similar in the solid-state and solution spectra, the bands are

blue-shifted by ∼2000 cm^-1upon solvation of 3. This result

suggests that the binding of MeOH (or glycerol) to the Fe center

is sufficiently strong to perturb the absolute band energies but

not strong enough to alter the overall appearance of the spectrum

(i.e., the band intensities and relative energies), which is

determined primarily by the N₄S ligand set.

As expected on the basis of qualitative considerations,

acceptable fits of the VTVH-MCD data of complex 1 could be

obtained only for D > 0 cm^-1. The best overall fit of the 470

nm data set, which probes the VTVH-MCD behavior of band

1, was achieved with the following ZFS parameters: D ) +9

cm^-1and E/D ) 0.20. However, the variance of the ø²parameter

with respect to D reveals that this optimal fit corresponds to a

shallow minimum, and we can only state with certainty that D

> +4 cm^-1. In contrast, the ø²parameter exhibits a stronger

dependence on the E/D ratio, leading to acceptable fits only

for moderately rhombic values (Table 3). Additionally, the best

overall fits of the 470 nm VTVH data set suggest that the

corresponding transition (band 1) is polarized perpendicular to

the x-axis. Since this transition has predominately SfFe(II) CT

character, it should be primarily polarized along the Fe-S bond

vector. Consequently, as band polarizations relate to the principal

axes of the D-tensor of the complex, our VTVH-MCD data

suggest that in 1 the Fe-S bond vector is oriented within the

yz-plane of the D-tensor.

The VTVH-MCD saturation curves measured for 3 also

exhibit significant changes upon solvation of the complex

(Figure S5(b)). Compared to the solid-state data, the solution

MCD magnetization curves saturate more rapidly with field and

exhibit a lesser degree of nesting. Fitting of the VTVH-MCD

solution data indicates that the D value increases upon solvation

and that the E/D ratio approaches the rhombic limit. The best

fit was obtained with D ) +12 cm^-1and E/D ) 0.25, although

good fits were also found over a range of large D values (|D|

> 9 cm^-1) near the rhombic limit.

For complex 3, additional ligand-binding studies were carried

out using azide, a close electronic mimic of superoxide. The

results of these studies are summarized in the Supporting

Information.

Computations. (a) Spin-Hamiltonian Parameters. INDO/

S-CI calculations performed on complexes 1 and 2 predict ZFS

values consistent with those provided by our analysis of the

VTVH-MCD data (Table 3). While the computed D value for

1 (D ) +3.7 cm^-1) is too small, the INDO/S-CI calculations

predict a rhombic system (E/D ) 0.24) with a positive ZFS

parameter D, as determined experimentally. Interestingly, the

calculations provide an unusual orientation of the D-tensor in

which none of the principal axes are collinear with the Fe-

ligand bonds. For complexes such as 1, with approximate

square-pyramidal geometry, one of the principal axes would

be expected to coincide with the pseudo-C₄axis, in this case

the Fe-S bond vector. Instead, the computed D-tensor is rotated

such that one axis forms a 24° angle with the Fe-S bond axis,

while another bisects the N_pyr-Fe-N_pyrbond vectors (Figure

6). Since the Fe-S bond lies within the yz-plane of the computed

D-tensor, this prediction is fully consistent with the VTVH-

MCD polarization data presented above. This unexpected

D-tensor orientation reflects key elements of the electronic

structure of 1 that will be explored in the next section.

A similar analysis was performed on the VTVH-MCD data

of complex 2, confirming that D < 0 cm^-1(Table 3). In sharp

contrast to the 470 nm band of complex 1, the transition at 370

nm of complex 2 is predominantly z-polarized, with only

marginal contributions from x- and y-polarizations. For 3, the

only acceptable fits of the 314 nm VTVH-MCD data set were

found for D > 0 cm^-1, again consistent with our qualitative

considerations. The best fit was obtained with D ) +6.0 cm^-1

and E/D ) 0.05, although acceptable fits also resulted for ZFS

parameters within the ranges +4 < D < +11 cm^-1and 0.04 <

E/D < 0.15 (Table 3). Regardless of the D and E/D values

chosen, the fitting procedure consistently required that the

dominant contribution to the 314 nm MCD feature involve the

M_yztransition moment product, indicating that the corresponding

transition is polarized perpendicular to the molecular x-axis.

Ligand-Binding Studies. As noted above, absorption results

suggest that MeOH is capable of coordinating to the Fe(II)

centers of complexes 2 and 3 (Table 1). Figure S5(a) displays

the low-temperature MCD spectrum of 3 in a frozen solution

of 2:1 MeOH/glycerol. Although the overall patterns of bands

INDO/S-CI calculations for complex 2 provide ZFS values

of D ) -4.9 cm^-1and E/D ) 0.12, reproducing the negative

ZFS parameter D and moderate rhombicity determined experi-

mentally (Table 3). Compared to complex 1, the calculated

D-tensor orientation of 2 is more readily understood. The z-axis

lies directly along the Fe-S bond, and the nearly linear N₂-

(46) Krzystek, J.; Zvyagin, S. A.; Ozarowski, A.; Fiedler, A. T.; Brunold, T.

C.; Telser, J. J. Am. Chem. Soc. 2004, 126, 2148-2155.

(47) For all three complexes, the most detailed analysis was performed on the

low-energy data sets, since the high-energy data were somewhat affected

by temperature-independent (B-term) contributions that became apparent

at energies > 30 000 cm^-1. However, in all instances, the ZFS parameters

that provided the best fits of the low-energy VTVH-MCD data sets also

provided excellent fits of the corresponding high-energy data sets.

9

1682 J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005

Synthesis/Characterization of Fe(II)−Thiolate Complexes

A R T I C L E S

Figure 7. (Top) Isosurface plots of the π_ipand π_opMOs obtained from

DFT calculations on the free p-tolyl thiolate ligand. (Bottom) Plots of the

two lowest-energy unoccupied MOs obtained from DFT calculations on

the free L⁸py₂ligand.

Figure 6. Structures of complexes 1 and 2 illustrating the D-tensor

orientations provided by INDO/S-CI calculations. For 2, the alternate

coordinate system considered in our DFT studies is indicated in parentheses.

(c) Bonding Description for Fe(II)-Thiolate Models.

Single-point DFT calculations were also performed on the full

complexes 1-3 except for the counterions, as detailed in the

Computational Methods section. Table S5 summarizes the

energies and compositions of the relevant spin-down molecular

orbitals for complexes 1-3, and the corresponding MO energy-

level diagrams are shown in Figure 8.⁵⁰

Fe-N₄unit defines the x-axis (Figure 6), again consistent with

our VTVH-MCD results.

Although the VTVH-MCD data unambiguously require a

positive D value for complex 3, INDO/S-CI calculations predict

a negative D value of -3.8 cm^-1and an E/D ratio of 0.23.

This discrepancy is likely due to the fact that the calculation,

even after inclusion of a large CI active space, provided a

ground-state Fe 3d electron configuration that is inconsistent

with both our spectroscopic results and DFT calculations for 3

(vide infra).

(b) Ligand Frontier Orbitals. To identify the relevant

frontier orbitals of the ligands present in complexes 1-3, DFT

calculations were first performed on the free L⁸py₂and Me₄-

cyclam macrocycles, as well as the aryl thiolate ligands of 1

and 2. In agreement with earlier computational studies,⁴⁸the

two highest-energy occupied MOs (HOMOs) of the aryl

thiolates have primarily sulfur 3p orbital character but also

possess considerable contributions from the aromatic ring

(Figure 7, top). Using previously established nomenclature,⁴⁹

these thiolate π-orbitals are labeled according to their orientation

with respect to the benzene ring: π_iplies in the plane of the

aromatic ring, while π_opis perpendicular to this plane. The π_ip

and π_oporbitals are nearly degenerate in the isolated ligand but

differ substantially with respect to the extent of electron

delocalization onto the aryl ring, amounting to 30% for the π_op

orbital, compared to only 10% for the π_iporbital.

Single-point DFT calculations on the free L⁸py₂and Me₄-

cyclam ligands were performed with the macrocycles in the

same conformation as adopted in complexes 1 and 2. As

expected, the four highest-energy occupied MOs of both the

Me₄cyclam and L⁸py₂macrocycles possess almost exclusively

N_σorbital character. Yet unlike Me₄cyclam, the L⁸py₂macro-

cycle features two unoccupied π*-based MOs localized on the

pyridine moieties (Figure 7, bottom). These π* MOs lie only

∼18 000 cm^-1above the HOMO and are thus available for

π-backbonding upon binding of L⁸py₂to low-valent (“electron

rich”) metals.

The electronic structure of 1 was computed with the molecule

aligned according to the D-tensor provided by INDO/S-CI

calculations (vide supra), in which the z-axis is rotated away

from the Fe-S bond vector by 24° and the y-axis bisects the

N_pyr-Fe-N_pyrbond angle (Figure 6).⁵¹The reason for this

unusual D-tensor orientation becomes apparent upon examina-

tion of the orbital compositions of the Fe d-based MOs. MOs

123 and 124 (Fe d_xz- and d_yz-based, respectively) reveal very

strong bonding interactions between the Fe(II) orbitals and the

π-acceptor orbitals of the L⁸py₂macrocycle, with 28% and 42%,

respectively, of the electron density residing on the pyridyl rings

(Table S5). This π-backbonding is also evident from MOs 126

and 127 (the antibonding counterparts of MOs 123 and 124)

that, although primarily localized on the L⁸py₂ligand, contain

substantial Fe d-orbital character. These results indicate that the

Fe d-orbitals in complex 1 are oriented such as to maximize

their π-interactions with the pyridyl acceptor orbitals. Impor-

tantly, both the orbital compositions and the D-tensor alignment

reveal that the dominant Fe-ligand interactions in complex 1

involve the equatorial pyridyl moieties, not the axial thiolate

ligand.

The computed Fe d-orbital splitting pattern for 1 is consistent

with an Fe(II) system with D > 0 cm^-1, as the doubly occupied

2

d_{x -y}-based MO (#122) is lower in energy than the d_xz/d_yz-

(50) For these S ) 2 complexes, spin polarization lowers the energies of the

majority-spin (i.e., spin-up) Fe d orbitals relative to their minority-spin (i.e.,

spin-down) counterparts, resulting in substantial mixing of the former with

ligand-based orbitals. The natures of the Fe-ligand bonding interactions

are therefore inferred from the ligand- and Fe d-based spin-down MOs

that are considerably less mixed than their spin-up counterparts.

(51) Initial calculations were performed with the Fe-S bond oriented along the

z-axis and with the four N atoms lying approximately in the xy-plane,

accounting for the approximate C_4Vsymmetry of the complex. However,

this resulted in a set of Fe d-based MOs with similar contributions from

multiple d orbitals, thus complicating analysis of the electronic structure.

In constrast, the orientation provided by INDO/S-CI calculations resulted

in Fe d-based MOs with a single dominant contributor orbital, although

some d-orbital mixing remained to maximize the bonding interactions with

both the thiolate and L⁸py₂-based orbitals.

(48) McNaughton, R. L.; Tipton, A. A.; Rubie, N. D.; Conry, R. R.; Kirk, M.

L. Inorg. Chem. 2000, 39, 5697-5706.

(49) Davis, M. I.; Orville, A. M.; Neese, F.; Zaleski, J. M.; Lipscomb, J. D.;

Solomon, E. I. J. Am. Chem. Soc. 2002, 124, 602-614.

9

J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005 1683

A R T I C L E S

Fiedler et al.

obtained with this coordinate system are also indicated (in

parentheses) in Table S5 and Figure 8. This alternate orientation

provides a d-orbital splitting pattern typical of trigonal bipyra-

midal systems, which are characterized by D < 0 cm^-1and

2

d-orbital energies following the sequence (d_xz, d_yz) < (d_{x -y}

,

2

d_xy) < d_z. Since the VTVH-MCD data also indicate a negative

D value for 2, it appears that the “effective” electronic symmetry

of this complex is indeed closer to trigonal bipyramidal than

square pyramidal. Nevertheless, the Fe d-orbital labels obtained

by orienting 2 according to the INDO/S-CI computed D-tensor

will be used for the remainder of the paper.

For complexes 1 and 2 with aryl thiolates, the bonding

interactions between the π_ipand π_oporbitals and the Fe d-orbitals

depend on two structural parameters: (i) the Fe-S-C₁bond

angle and (ii) the Fe-S-C₁-C₂dihedral angle (see Figure 7

for labeling scheme used). The X-ray crystal structures of

complexes 1 and 2 reveal Fe-S-C₁angles of 98° and 119°,

respectively, and dihedral angles of 111° and 97°.⁵⁴Since for

both complexes the two sets of angles are much closer to 90°

than 180°, the π_oporbital points primarily along the Fe-S bond

and forms σ-bonds with the Fe d-orbitals, while the π_iporbital

is largely perpendicular to the Fe-S bond vector and thus

develops π-bonding interactions with the Fe d-orbitals.⁵⁵For

complex 1, the Fe-S bonding interactions primarily involve

2

the Fe d_xz/π_ipand d_z/π_oporbitals, as reflected in the Fe d-based

MOs 123 and 125 (Table S5). The π_ip-based MO (#120) is

∼4000 cm^-1lower in energy than the π_op-based MO (#121),

suggesting that the Fe-S π-interaction is considerably stronger

than the Fe-S σ-interaction. For complex 2, the Fe-S bonding

interactions are reflected in the Fe d-based MOs 119-121. The

Figure 8. MO energy diagrams for complexes 1-3 obtained from spin-

unrestricted DFT calculations, showing the energies of the spin-down MOs

relative to that of the corresponding HOMO. MOs are labeled according to

their principal contributors.

derived pair of MOs (#123/124), which are split by ∼1000

2

Fe-S σ-interaction principally involves the Fe d_zand π_op

cm^{-1 52}

This splitting of the d_xz- and d_yz-based MOs (∆E_π) is

.

orbitals, while both the Fe d_xzand d_yzorbitals engage in

π-interactions with the π_iporbital. As with complex 1, the Fe-S

σ-interaction is much weaker than the π-interaction, which is

responsible, in large part, for the substantial rhombicity exhibited

by 1 (Table 3). The most unusual electronic feature of this

complex is the 12 000 cm^-1energy gap (∆E_σ) between the Fe

2

evidenced by the fact that the Fe d_z-based MO (#120) is ∼2500

2

cm^-1lower in energy than the Fe d_yz-based MO (#121).

Collectively, these results suggest that the aryl thiolate is a poor

σ-donor, albeit an effective π-donor.

d_z- and d_xy-based MOs (#125 and #130, respectively) that

constitute the e_g(σ*) set of MOs in the parent O_hsymmetry.

This enormous splitting suggests that the nitrogen “lone pairs”

of the L⁸py₂ligand are much better σ-donors than the axial aryl

thiolate ligand.⁵³

In contrast to 1 and 2, complex 3 contains an alkyl thiolate

ligand whose S 3p orbitals are not involved in intraligand

π-interactions. For systems involving metal-(alkyl)thiolate

bonds, such as the SOR, blue copper, and sulfite oxidase active

sites, the nature of the metal-sulfur bonding interaction is

determined by the M-S-C₁bond angle.⁵⁶If this angle is near

180°, then the two S 3p-based lone pairs interact with the metal

d-orbitals in a π fashion, but as the bond angle is reduced from

180° to 90°, one lone pair starts developing a pseudo-σ

interaction with the metal center. The crystal structure of 3

reveals an Fe-S-C₁bond angle close to 90° (98°), suggesting

that the thiolate possesses one almost purely σ-donating orbital

(S_σ) and one purely π-donating orbital (S_π). Furthermore, the

constraints imposed on the pendant thiolate by the cyclam ring

result in a N₂-Fe-S-C₁dihedral angle (Figure 2) of nearly

0° (3.3°). As a result, the S_πorbital interacts exclusiVely with

the Fe d_yzorbital, while the Fe d_xzorbital is orthogonal to both

S 3p-based lone pairs (Scheme 3). This situation is in contrast

DFT studies of 2 were performed with the model aligned

according to the D-tensor orientation provided by INDO/S-CI

calculations (Figure 6). This resulted in an Fe d-orbital splitting

pattern intermediate between those expected for trigonal bi-

pyramidal and square pyramidal coordination geometries,

consistent with the Fe ligand environment in complex 2 (Figure

1). In particular, the trigonal distortion exhibited by this species

results in a significant destabilization of the Fe d_yzorbital through

σ-antibonding interactions with the nitrogen-based lone pairs

of the Me₄cyclam ligand and, consequently, in a 3900 cm^-1

splitting of the Fe d_π-based MOs (#119 and 121). A more

conventional splitting pattern is obtained when the z-axis and

y-axis are oriented along the Fe-N₂and Fe-S bond vectors,

respectively, as shown in parentheses in Figure 6. The MO labels

(52) The coordinate system for complex 1 shown in Figure 6 allows for pure

2

d_xzand d_yzorbitals but interchanges the identities of the d_xyand d_{x -y}

orbitals.

2

(53) Two additional factors contribute to the large ∆E_σvalue for 1: (i) the d_z

(54) For complex 2, the angles reported in the text are averaged over both

crystallographically independent cations.

orbital is not oriented directly along the Fe-S bond axis and thus does not

experience the full effect of the destabilizing σ* interaction, and (b) the

pyridyl rings are tilted slightly out of the N_pyr-Fe-N_pyrplane, allowing

(55) Note that for both complexes deviations of these angles from 90° give rise

to somewhat mixed σ- and π-bonding interactions.

2

the d_zorbital to engage in weak π-backbonding (i.e. stabilizing) interactions

(56) Solomon, E. I.; Szilagyi, R. K.; George, S. D.; Basumallick, L. Chem. ReV.

2004, 104, 419-458.

with the L⁸py₂π*-acceptor orbitals.

9

1684 J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005

Synthesis/Characterization of Fe(II)−Thiolate Complexes

A R T I C L E S

Scheme 3

Figure 9. Qualitative energy level diagram illustrating the effect of MeOH

coordination on Fe d-orbital energies of complex 3.

to the one encountered above for complex 2, wherein the

orientation of the untethered aryl thiolate results in significant

bonding interactions between the π_iporbital and both the Fe d_xz

and d_yzorbitals.

The MO diagram of 3 shown in Figure 8 reflects the strong

σ- and π-donating ability of the alkyl thiolate. In particular,

2

the interaction between the S_σand Fe d_zorbitals is very

covalent, as judged on the basis of both MO energies and

compositions (Table S5). While for both 1 and 2 the thiolate-

based MO principally involved in Fe-S π-bonding (π_ip) is

several thousand wavenumbers lower in energy than the

corresponding σ-bonding MO (π_op), for 3 this pattern is reversed

with the S_σ-based MO (#91) now stabilized by ∼6000 cm^-1

2

relative to the S_π-based MO (#92). Furthermore, the Fe d_z-

Figure 10. INDO/S-CI calculated spectrum for 1 (s) obtained using the

crystallographically determined structure, along with the experimental

absorption spectrum of 1 in MeCN solution at room temperature (‚‚‚).

based MO (#96) of 3 is strongly destabilized relative to its

counterpart in complex 2, suggesting that alkyl thiolates are

much stronger σ-donors compared to aryl thiolates. These

differences between the two types of thiolate ligands are likely

due to the highly delocalized nature of the π_oporbital in aryl

thiolates (Figure 7), which prevents strong σ-bonding inter-

actions with the Fe centers in 1 and 2.

DFT and INDO/S-CI calculations appear largely consistent with

our spectroscopic results, it is reasonable to use these methods

for assigning key features in the corresponding electronic

absorption and MCD spectra. To accomplish this goal, the

INDO/S-CI method was used to interpret the absorption

spectrum of complex 1. As shown in Figure 10, the calculated

absorption spectrum for 1 consists of two prominent bands that

correspond to the two principal CT features in the experimental

spectrum.⁵⁷On the basis of this calculation, the low-energy

feature in the experimental spectrum (bands 1 and 2) is assigned

Our DFT results indicate that the singly occupied Fe d-based

spin-down MO of 3 (#93) has primarily Fe d_xzcharacter, which

provides an interesting contrast to the situation encountered for

complexes 1 and 2 where the corresponding MOs lie in the

2

equatorial plane (d_xyor d_{x -y}character). This change is due to

the unique structural features of 3 described above, which ensure

that the Fe d_xzorbital is essentially nonbonding. The computed

Fe d-orbital splitting pattern for 3 provides insight into the origin

of the changes in the ZFS parameters accompanying solvation

of this species in MeOH. As noted above, upon weak solvent

binding to the Fe(II) center of 3 the D value increases and the

E/D ratio approaches the rhombic limit. For such tetragonal

Fe(II) systems, these results suggest that the energy gap between

the doubly occupied Fe d-orbital and the other d-orbitals of the

parent t_2gset is diminished upon solvation and that the splitting

pattern becomes more characteristic of a system with negative

D.⁴⁴This scenario is only possible if E(d_xz) < E(d_xy) in the solid

state, since MeOH binding is expected to destabilize the d_xz

and d_yzorbitals, while leaving d_xylargely unaffected (Figure

9).

2

to the S(π_op)fFe(d_z) (σfσ*) CT transition, while the more

intense feature at higher energy (band 6) is attributed to the

S(π_ip)fFe(d_xz) (πfπ*) CT transition (see Table 2). Band 7 at

30 600 cm^-1is then assigned to the S(π_ip)fFe(d_z) transition,

2

where the considerable absorption intensity of this band suggests

that the π_iporbital experiences a significant σ-interaction with

the Fe d_zorbital.⁵⁸Note that the spectroscopically determined

2

energy splitting of 1650 cm^-1between the Fe d_xzand d_zorbitals

2

is in remarkable agreement with the DFT-calculated splitting

of 1630 cm^-1(Table S5). Figure 11 (left) summarizes these

(57) The calculated spectrum in Figure 10 has been red-shifted by 4000 cm^-1

to facilitate comparison with the experimental spectrum.

(58) It is important to note that bands that are intense in the absorption spectrum

are often weak in the MCD spectrum, and it is therefore not surprising

that the relative intensities of bands 6 and 7 are dramatically different in

the two spectra.

Spectral Assignments. In light of the fact that the electronic

structure descriptions of the Fe(II) models 1-3 provided by

9

J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005 1685

A R T I C L E S

Fiedler et al.

intense bands that lie within the absorption feature at 35 000

cm^-1(bands 4-6) are difficult to assign conclusively, it is

obvious that the intense band at 38 000 cm^-1(band 7) must

2

arise from the S_σfFe(d_z) (σfσ*) transition. Even when

accounting for the fact that this band is superimposed on the

shoulder of a high-energy band with a maximum at .40 000

cm^-1, it remains twice as intense as the corresponding σfσ*

transition of 2, confirming the highly covalent nature of the

Fe-S σ-bonding interaction in complex 3.

Discussion

A more complete understanding of the catalytic activity of

SORs requires the synthesis and characterization of relevant

model complexes that accurately mimic the geometry, reactivity,

and electronic structure of the SOR_redactive site. This report

describes the synthesis of two novel Fe(II) complexes, 2 and 3,

that possess the N₄S donor set of ligands found in SOR_red. Unlike

the first-generation of SOR models represented by 1, which

utilizes the tetradentate ligand L⁸py₂to model the four equatorial

His residues found for the enzymatic site,¹⁸these second- and

third-generation models are based on the macrocyclic N₄-donor

cyclam ligand (Scheme 1). Our X-ray crystallographic results

show that the molecular structures of both 2 and 3, unlike that

of 1, display large trigonal distortions from an idealized square-

pyramidal geometry, with τ parameters in the range of 0.46 to

0.57. The three complexes also differ with respect to the nature

and position of the axial thiolate ligand. While 1 and 2 feature

unconstrained aryl thiolates, complex 3 contains an alkyl thiolate

that is tethered to the cyclam ring (Figure 2), resulting in a

restricted thiolate orientation.

To explore the electronic structures of models 1-3 in detail,

we have employed electronic absorption, MCD, and VTVH-

MCD spectroscopies in conjunction with DFT and INDO/S-CI

calculations. By combining the insights gained from our

electronic structure calculations with the results obtained from

our spectroscopic studies, we have succeeded in developing

experimentally validated bonding descriptions for all three

complexes. These bonding descriptions greatly enhance our

understanding of the dominant Fe(II)-thiolate bonding inter-

actions in SOR_red-like species, as well as their dependencies on

the identities of the equatorial N-donor ligands.

Figure 11. Energy-level diagram displaying the experimentally determined

transition energies (in cm^-1) within the MO framework provided by DFT

calculations for 1-3. Transitions are labeled according to the numbering

scheme used in Figure 4 and Table 2.

spectral assignments within the MO framework provided by

DFT calculations for 1.

INDO/S-CI calculations for complex 2 afforded results very

similar to those discussed above for 1, leading to closely

analogous assignments of key spectral features (Figure 11,

center). Specifically, the weak shoulder in the absorption

spectrum, corresponding to the positive MCD feature at ∼27 000

cm^-1(band 2), arises from the S(π_op)fFe(d_z) (σfσ*) CT

2

transition. Note that this assignment concurs nicely with the

z-polarization of band 2 as determined by our VTVH-MCD

analysis. The intense low-temperature absorption feature at

31 100 cm^-1(bands 5 and 6) has contributions from two CT

transitions, involving electronic excitations from the π_ip-based

2

MO into the Fe d_z- and d_yz-derived orbitals (Figure 11). The

separation of bands 5 and 6 thus provides an energy splitting

of 1200 cm^-1between the Fe d_zand d_yzorbitals, which is

2

somewhat smaller than the DFT-calculated value of 2500 cm^-1

(Table S5).⁵⁹

Comparisons to SOR Active Site. Like the enzymatic

system, all three generations of SOR models included in this

study feature a N₄S ligand set around a high-spin (S ) 2)

Fe(II) center, and also exhibit vacant coordination sites trans

to the thiolate ligand.^4-6The first-generation of models (1)

provide the best structural mimics of SOR_red, as the Fe(II)

coordination environment is clearly square-pyramidal (τ ≈ 0.15).

In contrast, the Fe(II) coordination environment in the second-

and third-generation models is intermediate between square-

pyramidal and trigonal-pyramidal (τ ≈ 0.50).

The solid-state MCD spectrum of 3 reveals that the low-

energy absorption feature consists of three bands (Figure 4).

As DFT calculations indicate that E(S_σ) , E(S_π), we conclude

that all three bands arise from S_πfFe(II) CT transitions. Band

2, which is intense in both the electronic absorption and MCD

spectra, is assigned to the S_πfFe(d_yz) (πfπ*) transition. The

weak absorption features at 28 750 cm^-1(band 1) and 31 450

cm^-1(band 3) are assigned to the S_πfFe(d_xy) and S_πfFe(d_z)

2

transitions, respectively.⁶⁰Despite its weak absorption intensity,

Although complex 1 is structurally the best model of the

SOR_redactive site, the spectroscopic and computational results

2

the S_πfFe(d_z) transition acquires considerable MCD intensity

due to excited-state mixing via SOC with the S_πfFe(d_yz)

(πfπ*) transition, resulting in the pseudo-A term feature

centered at 30 900 cm^-1(Figure 4). While the three moderately

(60) Based on these assignments, it appears that the DFT calculation for 3

2

underestimates the splitting between the Fe d_yz- and d_z-based MOs (#95

and #96, respectively, in Figure 8/Table S3), which is found to be 1000

cm^-1experimentally, as compared to 160 cm^-1predicted computationally.

Moreover, the 1700 cm^-1experimental splitting between the Fe d_xy- and

d_yz-based MOs (#94 and #95, respectively) is much smaller than the

calculated splitting of 4020 cm^-1. It is therefore apparent that, although

DFT provides a qualitatively correct Fe d-orbital splitting pattern for 3,

the experimental Fe d-orbital splitting pattern is, in fact, reminiscent of an

“effective” electronic symmetry intermediate between trigonal bipyramidal

and square pyramidal.

(59) Based on our INDO/S-CI computations, the S(π_ip)fFe(d_yz) (πfπ*)

transition is strongly polarized along the Fe-S bond (z-axis), while the

2

S(π_ip)fFe(d_z) transition is polarized along the y-axis. Mixing of the

corresponding excited states via spin-orbit coupling (SOC) should thus

give rise to a pseudo-A term feature, which is indeed observed in the MCD

spectrum of 2 at 31 500 cm^-1(Figure 4).

9

1686 J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005

Synthesis/Characterization of Fe(II)−Thiolate Complexes

A R T I C L E S

presented herein indicate that electronically it differs quite

substantially. This difference is evident even on a qualitative

level; namely, while SOR_red(as well as complexes 2 and 3) is

colorless,⁶1 exhibits a bright yellow color both in the solid

state and in solution due to the presence of SfFe(II) CT

transitions in the visible region. Expressing this observation in

quantitative terms, the onset of the CT manifold appears at

∼29 000 cm^-1in the SOR_redabsorption spectrum⁶but at an

energy of ∼20 000 cm^-1in the spectrum of complex 1.

SfFe(II) CT transitions formally produce an Fe(I) center in

the excited state, and since this low Fe oxidation state is highly

unstable in most coordination environments, these transitions

normally appear far in the UV region. However, our spectro-

scopic and computational studies of complex 1 have revealed

strong bonding interactions between the Fe(II) d_πorbitals and

the pyridine π* orbitals, which allow for a ∼4000 cm^-1

stabilization through π-backbonding of the transiently formed

Fe(I) center in these CT excited states. This result has implica-

tions for the competence of 1 as a functional SOR mimic, since

it is widely believed that the destabilization of the Fe d_πorbitals

via Fe-S π-antibonding interactions is crucial for facile electron

transfer to the superoxide anion.¹⁴In the SOR_redactive site, such

a stabilization of the Fe d_πorbitals via π-backbonding is not

possible, as the imidazole rings of the His residues are weak

π-donor ligands,⁶¹not π-acceptors such as L⁸py₂.

by complex 3 gives rise to certain electronic features that may

not be shared by the enzyme active site; most importantly, a

significant stabilization of the doubly occupied, “redox active”

Fe d_xz-based MO (#93) relative to the other Fe d-based MOs,

leading to a value of ∆E_πin excess of 2000 cm^-1. When Fe(II)

exists in a more rigorous square-pyramidal coordination envi-

ronment, as exemplified by complex 1 and the SOR_redactive

site, the redox active MO lies in the equatorial plane defined

by the N ligands, and the value of ∆E_πis much smaller than

that for 3.¹²

Although 2 and 3 share many key structural features, the

electronic properties of the two complexes are in fact quite

distinct. Specifically, while 2 fails to reproduce several key

elements of the electronic structure of SOR_red, our spectroscopic

and computational results suggest that complex 3 adequately

models the Fe-S bonding interactions present in the SOR active

site. These similarities between 3 and SOR_redbecome particu-

larly apparent when one compares their electronic absorption

and MCD spectra. Both species exhibit multiple SfFe(II) CT

transitions, one group carrying moderate intensity (ꢀ ≈ 2000

M^-1cm^-1) near 31 000 cm^-1(30 450 cm^-1for 3 and 31 200

cm^-1for SOR_red) that are assigned to πfπ* transitions, and

another group near 38 500 cm^-1(38 000 cm^-1for 3 and 38 900

cm^-1for SOR_red) arising from σfσ* CT transitions.⁶Thus,

complex 3 almost quantitatively reproduces the SfFe(II) CT

energies of SOR_redand, unlike 1 and 2, also yields the proper

ordering E(σfσ*) . E(πfπ*) found for the enzyme active

The Fe d-orbital splitting pattern determined for 1 is consistent

with both the square-pyramidal coordination geometry of this

complex and the positive D value found through analysis of

VTVH-MCD data. An interesting aspect of the electronic

structure of 1, as elucidated by our DFT calculations, is the

site. Furthermore, the ∆E_σ(3) value of 9500 cm^-1is near the

6

range of 7400 to 9000 cm^-1reported for SOR_red

.

Our studies permit identification of two structural features

of model 3 that make its electronic properties similar to those

of SOR_red: (i) the presence of an alkyl thiolate instead of an

aryl thiolate as in 1 and 2 and (ii) the geometric constraints

imposed on the thiolate orientation by the intramolecular -CH₂-

CH₂- linkage to the cyclam ring, which results in an Fe-S-

C₁angle of 98° and a N₂-Fe-S-C₁dihedral angle of 3°

(Figure 2). As noted above, this orientation of the thiolate ligand

leads to a highly covalent Fe-S σ-bonding interaction and limits

the π-bonding interaction to a single Fe d_π-orbital, causing a

substantial splitting (∆E_π) between the two Fe d_πorbitals. Given

the striking electronic similarities between 3 and SOR_red, it is

not surprising that crystallographic studies⁴have revealed that

the SOR_redactive site adopts a similar geometry with respect

to the Fe-thiolate unit, with an Fe-S-C₁angle of 116° and a

N(His₁₆)-Fe-S-C₁dihedral angle of 6°, averaged over the

crystallographically inequivalent protein subunits.⁶²This ori-

entation of the active-site Cys residue largely restricts the

π-bonding interaction to a single Fe d_π-orbital, likely resulting

in a substantial ∆E_πsplitting for SOR_red. Collectively, the

structural and spectral parallels between 3 and SOR_redindicate

that the Fe-S bonding interactions in these two species are quite

similar.

2

large energy gap between the Fe d_z- and d_xy-based MOs, ∆E_σ-

(1) ) 11 800 cm^-1. Near-IR MCD studies by Johnson and co-

workers have provided a ∆E_σvalue for SOR_redin the range

7400 to 9000 cm^-1.⁶The fact that ∆E_σ(SOR_red) is significantly

smaller than ∆E_σ(1) initially appears somewhat surprising, given

that the Fe-S bond distance in the protein is larger than that in

the model; i.e., r(Fe-S) ) 2.32 Å for 1,¹⁸as compared to

r(Fe-S) ) 2.42 Å (X-ray structure⁴) or 2.37 Å (EXAFS⁶) for

SOR_red. However, these ∆E_σvalues are consistent with our

spectroscopic and computational results that consistently indicate

that the aryl thiolate ligand of 1 is a relatively poor σ-donor

compared to the alkyl thiolate found in the SOR_redactive site.

Interestingly, the energies and intensities of some of the

SfFe(II) CT transitions of complex 2 compare quite favorably

with those reported for SOR_red. Most notably, the S(π_ip)f

Fe(II) (πfπ*) transition that occurs at 32 050 cm^-1(ꢀ ) 2470

M^-1cm^-1) for 2 resembles the prominent SfFe(II) (πfπ*)

6

transition at 31 200 cm^-1(ꢀ ≈ 2000 M^-1cm^-1) for SOR_red

.

Despite this similarity, the coordination geometry of 2 deviates

substantially from square pyramidal, profoundly affecting the

electronic structure and, in turn, the ZFS parameters of this

complex. Our VTVH-MCD analysis reveals a negative D value

of ∼ -9 cm^-1for 2, in contrast to the positive D value of ∼ +10

cm^-1reported for SOR_red.⁶This difference can be understood

in terms of the Fe d-orbital splitting pattern obtained for 2 in

Most discussions of the electronic structure of the SOR_red

active site have focused on the Fe(d_π)-S_πinteraction, which is

presumed to facilitate electron transfer from the Fe(II) center

to the π*-acceptor orbital of the substrate superoxide and

2

the alternate coordinate system, (d_xz, d_yz) < (d_{x -y}, d_xy) < d_z,

which is characteristic of a system with trigonal bipyramidal

geometry (Figure 6). Likewise, the trigonal distortion exhibited

(62) In the crystal structure of the homotetrameric SOR from P. furiosus, two

of the subunits (B and C) suffered from low Fe occupancies and provided

less accurate active-site geometries. Thus, the bond angles reported in the

text were obtained by examination of only the A and C subunits of SOR_red

(see ref 4).

(61) (a) Holm, R. H.; Solomon, E. I. Chem. ReV. 1996, 96, 2237-2237. (b)

Guckert, J. A.; Lowery, M. D.; Solomon, E. I. J. Am. Chem. Soc. 1995,

117, 2817-2844.

9

J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005 1687

A R T I C L E S

Fiedler et al.

promote dissociation of the nascent peroxide anion via a trans

influence.^7,14This emphasis on the Fe-S π-interaction as being

instrumental for SOR function largely stems from the fact that

the absorption spectrum of the oxidized form of SOR (SOR_ox)

is dominated by an intense SfFe(III) CT transition of πfπ*

character at ∼15 000 cm^-1, while the σfσ* counterpart is

hidden under the broad absorption feature at ∼30 000 cm^{-1 6}

.

Consequently, the importance of the Fe-S σ-interaction has

been largely ignored in discussions of electronic-structure

contributions to SOR function. Yet our spectroscopic and

computational studies of complex 3, an excellent model of

SOR_redwith respect to Fe(II)-S bonding, suggest that the Fe-S

σ-interaction in SOR_redis likely to be even stronger and more

covalent than the corresponding π-interaction. In particular, the

specific orientation of the Cys residue within the SOR active

site, which leads to an Fe-S-C₁angle (116°) much closer to

90° than 180°, appears designed to maximize the σ-donor ability

of the thiolate ligand. Together, these strong σ- and π-bonding

interactions result in substantial charge donation from the anionic

thiolate ligand to the Fe(II) center, lowering the effective nuclear

charge and, thus, decreasing the redox potential (E°) of the Fe

center in SOR. Consequently, SOR may be viewed as another

member of the family of metal-thiolate containing proteins

(such as blue copper⁵⁶and sulfite oxidase^48,63) for which the

exact positioning of the cysteine residue with respect to the metal

center is crucial for optimal enzymatic activity.

Figure 12. Two possible scenarios for inner-sphere electron transfer (ET)

from the Fe(II) center of SOR to the superoxide radical (O₂^•-). (a) The

“boxed” electron in the Fe d_π/S_πorbital is transferred to the π* acceptor

orbital of O₂^•-, yielding an Fe(III) S ) ³/₂intermediate. (b) Orbital mixing

“promotes” the boxed electron from the d_xyorbital to the nearby d_πorbital,

followed by ET to O₂^•-. This mechanism directly results in the S )

state of the oxidized enzyme.

/

5

2

crystallographic data indicate that the pentacoordinate Fe(II)

center in SOR_reddoes not bind an axial water molecule,^4,5and

a computational study predicted a lengthy Fe-O(H₂) bond of

2.8 Å.¹²These observations can be rationalized in terms of a

thiolate trans influence; i.e., for solvent to coordinate to the Fe

center, it would have to develop bonding interactions with the

Fe d_πand d_σorbitals already involved in Fe-S bonding. The

Fe-S σ-interaction likely plays the dominant role in modulating

the affinity of the Fe center toward exogenous ligands, because,

unlike the π-interaction, it is directed along the Fe-S bond and

discourages σ-donating ligands, such as H₂O, from binding. In

our studies, the trans influence associated with the strong Fe-S

σ-interaction is evident in the weak affinity of the Fe(II) center

of complex 3 toward solvent and azide, as determined spectro-

scopically. By analogy, the thiolate trans influence may also

labilize the Fe(III)-peroxide interaction after the electron is

transferred to coordinated superoxide. Specifically, the strong

Fe-S(Cys) bonding interaction may weaken the Fe(III)-

OO(H) bond and thus ensure rapid enzymatic turnover by

facilitating H₂O₂dissociation (step 3 of the catalytic cycle).

The SOR catalytic cycle is believed to proceed through an

inner-sphere mechanism, in which the superoxide anion first

binds directly to the Fe(II) center (step 1), followed by electron

transfer (ET) to the substrate (step 2). For step 2 to proceed

•-

rapidly, the singly occupied π* acceptor orbital of O₂must

interact electronically (via orbital overlap in a π-fashion) with

the “redox-active” donor d-orbital(s) of the Fe(II) center. It has

been previously suggested that the Fe d_πorbital that participates

in the strong Fe-S π-interaction serves as the redox-active

orbital.^7,14However, since this orbital (d_π/S_π) is singly occupied

in the SOR_redactive site, its direct involvement in ET to the

substrate would initially produce a high-energy Fe(III) S ) ³/₂

intermediate (Figure 12a). To avoid such an unfavorable

intermediate-spin state, the redox-active orbital of SOR_redwould

need to be the doubly occupied Fe d_xy-based MO, although this

orbital lacks the proper orientation for significant overlap with

the superoxide π* acceptor orbital. An additional possibility,

shown in Figure 12b, involves transfer of the spin-down (â) Fe

electron from the d_xyorbital via one of the d_πorbitals

(superexchange mechanism).⁶⁴Efficient mixing between d_xyand

the relevant d_πorbitals would require these orbitals to be in

close energetic proximity. Indeed, the position of the Cys residue

within the SOR_redactive site appears to promote this type of

mechanism because, as noted above, the S_πorbital is essentially

orthogonal to one of the Fe d_πorbitals, a configuration that

maximizes ∆E_πwhile minimizing the energy gap between d_xy

and the lowest-energy d_πorbital. Thus, the axial thiolate ligand

is able to depress the enzymatic redox potential via strong Fe-S

π- and σ-interactions without discouraging inner-sphere electron

transfer to the superoxide π* acceptor orbital.

Conclusion

In this study we have prepared and thoroughly characterized

a series of Fe(II)-thiolate complexes, 1-3, that possess various

types of axial thiolates (aryl and alkyl), equatorial N₄-macro-

cycles (L⁸py₂and Me₄cyclam), and coordination geometries.

This approach permitted us to probe the influence of each of

these variables on the overall electronic properties of these

SOR_redmodels, with a particular focus on the nature of the

Fe(II)-S bonding interactions. On the basis of our spectroscopic

and computational analyses, we conclude that all three com-

plexes mimic important features of the electronic structure of

SOR_red, with complex 3 replicating the Fe-S bonding interac-

tions present within the reduced active site particularly well.

This similarity between complex 3 and SOR_redis due, in large

part, to the constrained orientations of the pendant alkyl thiolate

ligand in the synthetic model and the cysteine residue in the

enzyme.

It is interesting to note that while the SOR active site is

located near the protein surface and thus exposed to solvent,

(63) McNaughton, R. L.; Helton, M. E.; Cosper, C. A.; Enemark, J. H.; Kirk,

M. L. Inorg. Chem. 2004, 43, 1625-1637.

(64) Newton, M. D. Chem. ReV. 1991, 91, 767-792.

9

1688 J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005

Synthesis/Characterization of Fe(II)−Thiolate Complexes

A R T I C L E S

Acknowledgment. This research was supported by grants

from the National Science Foundation (CHE-0243951 to J.A.H.),

the Camille and Henry Dreyfus Foundation (Henry Dreyfus

Teacher-Scholar Award to J.A.H.), the University of Wisconsin-

Eau Claire, the University of Wisconsin and the Sloan Founda-

tion Research Fellowship Program (T.C.B.), and the NSF

Graduate Research Fellowship Program (A.T.F.). We thank Dr.

Victor G. Young (University of Minnesota) for his assistance

with the structure determination of 3.

energies and compositions of relevant molecular orbitals from

DFT calculations for complexes 1-3 (Table S5), electronic

absorption spectra in MeOH solution (Figure S1), low-temper-

ature solid-state absorption spectra (Figure S2), Gaussian-

resolved solid-state absorption spectra (Figure S3), and high-

energy VTVH-MCD data sets (Figure S4) for all three complexes

1-3, MCD spectra of 3 in MeOH solution (Figure S5), results

from azide-binding experiments, INDO/S-CI active spaces, and

complete crystallographic data for 2 and 3 in CIF format. This

material is available free of charge via the Internet at

http://pubs.acs.org.

Supporting Information Available: X-ray crystallographic

data for 2 and 3 (Table S1), relevant structural parameters for

2 and 3 (Table S2), Cartesian coordinates for DFT geometry-

optimized models of 2 and 3 (Tables S3 and S4, respectively),

JA046939S

9

J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005 1689

Article Doi

Synthesis, structure determination, and spectroscopic/computational characterization of a series of Fe(II)-thiolate model complexes: Implications for Fe-S bonding in superoxide reductases

DOI: 10.1021/ja046939s

Source and publish data:

Authors:

Article abstract of DOI:10.1021/ja046939s

Full text of DOI:10.1021/ja046939s

Products guided by the article

R&D Labs maybe for 847749-32-0

Relevant to this article

Hot Product