Structure and Vibrational Dynamics of Model Compounds of the [FeFe]-Hydrogenase Enzyme System via Ultrafast Two-Dimensional Infrared Spectroscopy

Transcription

1 J. Phys. Chem. B 2008, 112, Structure and Vibrational Dynamics of Model Compounds of the [FeFe]-Hydrogenase Enzyme System via Ultrafast Two-Dimensional Infrared Spectroscopy A. I. Stewart, I. P. Clark, M. Towrie, S. K. Ibrahim, A. W. Parker, C. J. Pickett, and N. T. Hunt*, Department of Physics, UniVersity of Strathclyde, SUPA, 107 Rottenrow East, Glasgow G4 0NG, U.K., Central Laser Facility, Central Laser Facility, Science & Technology Research Council, Rutherford Council, Rutherford Appleton Laboratory, Harwell Science and InnoVation Campus, Didcot, Oxfordshire, OX11 OQX, U.K. and School of Chemical Sciences and Pharmacy, UniVersity of East Anglia, Norwich NR4 7TJ, U.K. ReceiVed: April 15, 2008; ReVised Manuscript ReceiVed: June 2, 2008 Ultrafast two-dimensional infrared (2D) spectroscopy has been applied to study the structure and vibrational dynamics of (µ-s(ch 2 ) 3 S)Fe 2 (CO) 6, a model compound of the active site of the [FeFe]-hydrogenase enzyme system. Comparison of 2D-IR spectra of (µ-s(ch 2 ) 3 S)Fe 2 (CO) 6 with density functional theory calculations has determined that the solution-phase structure of this molecule is similar to that observed in the crystalline phase and in good agreement with gas-phase simulations. In addition, vibrational coupling and rapid (<5 ps) solvent-mediated equilibration of energy between vibrationally excited states of the carbonyl ligands of the di-iron-based active site model are observed prior to slower ( 100 ps) relaxation to the ground state. These dynamics are shown to be solvent-dependent and form a basis for the future determination of the vibrational interactions between active site and protein. I. Introduction The technique of ultrafast two-dimensional infrared spectroscopy (2D-IR) has become well-established in recent years as a powerful probe of molecular structure and vibrational dynamics. By spreading the molecular response over two frequency axes to produce a spectrum not unlike that obtained via 2D-NMR spectroscopy, in which the 1D infrared signature is present along the spectrum diagonal with off-diagonal peaks revealing vibrational coupling and population transfer between individual modes, 2D-IR circumvents the main drawback of conventional infrared spectroscopy, namely, the loss of information due to the projection of an ensemble-averaged response onto one frequency axis. 1 Furthermore, it has been shown that ultrafast 2D-IR spectroscopy can be used to provide a direct probe of molecular structure and solute-solvent interaction dynamics and to monitor fast processes such as chemical exchange in hydrogen-bonded systems. 2,3 As the field of ultrafast 2D-IR spectroscopy has evolved, two parallel methodologies have been established, that using the time-domain interferometric detection of infrared photon echo signals, 1,2,4 analogous to techniques used in multidimensional NMR spectroscopy, 5 and a quasi-frequency domain double-resonance technique. 6 Though differing in their approach, it has been demonstrated that the information obtained by both methods is comparable. 7 This arises from the fact that both experiments probe the same third-order response function of the material in question in the small field regime, i.e., in which a linear expansion of the nonlinear response in terms of the electric field of the incident laser pulses is justified. Thus, in the main it can be shown that the two measured signals are related through a simple Fourier * To whom correspondence should be addressed. nhunt@ phys.strath.ac.uk. University of Strathclyde. STFC Rutherford Appleton Laboratory. University of East Anglia. transform relationship, 7 though the more experimentally complex photon echo approach is more flexible and can allow access to certain experimental parameters and radiation-matter interaction pathways that are not resolvable using the more straightforward double-resonance methodology. To date, ultrafast 2D-IR spectroscopy has been applied to a range of problems including the determination of the structure of both short and long chain peptides 6,8 15 and the elucidation of solute-solvent interaction dynamics, 1,16 18 including the role played by hydrogen bonding In addition, 2D-IR has been used to measure the rates of exchange processes occurring in hydrogen-bonded systems such as N-methylacetamide and hydrogen-bonded complexes Ultrafast 2D-IR spectroscopy has also been recently extended to study nonequilibrium systems such as the observation of transition states in fluxional molecules, 28 the unfolding of strained peptide chains, and excitedstate solvation dynamics through the development of transient 2D-IR methods. 3,29 31 Two-color 2D-IR methods have also been applied to observe intramolecular vibrational energy relaxation processes. 12,32,33 Finally, there have been recent developments in the experimental methodology in order to enhance sensitivity and decrease the time required for data acquisition such as the use of frequency up-conversion of infrared photon echo signals 34,35 and the use of pulse-shaping technology. 36,37 The technique has been the topic of recent review articles. 2,3 This Article reports the first application of ultrafast 2D-IR spectroscopy to the study of model systems of an enzyme active site in an attempt to determine the solution-phase structure, the intramolecular ultrafast vibrational dynamics, and the role played by the solvent bath in these processes. The hydrogenases are a family of metalloprotein-based enzymes. Their function is to catalyze the reversible activation of molecular hydrogen according to the reaction H 2 T 2H + + 2e -. Structurally, the hydrogenases can be divided into two subsets, the Ni-Fe and the Fe-only hydrogenases, based on the composition of the active site. This consists of a dimetallic /jp803338d CCC: $ American Chemical Society Published on Web 07/23/2008

2 10024 J. Phys. Chem. B, Vol. 112, No. 32, 2008 Stewart et al. Figure1. Schematicdiagramoftheactivesiteofthe[FeFe]-hydrogenase enzyme (a) and a typical model compound (µ-s(ch 2) 3S)Fe 2(CO) 6 as studied here (b). 38 An alternative three-dimensional representation of (µ-s(ch 2) 3S)Fe 2(CO) 6 can be seen in Figure 4. cluster (either Ni-Fe or Fe-Fe) bridged by two sulfur atoms. In addition, the metal atoms are coordinated by carbonyl (bridging and terminal) and cyanide ligands (Figure 1). Interest in these species is driven by the view that understanding the chemistry involved will lead to new electrocatalytic systems for hydrogen production or uptake. Such technology will result in new materials allowing the replacement of the expensive platinum metal catalysts used in fuel cells with those made from the more abundant iron and nickel. While the Ni-Fe enzymes have been extensively studied it is the Fe-Fe system that has been the focus of considerably more synthetic model studies resulting from the structural similarity of the active site to complexes of the form (µ-srs)fe 2 (CO) 6. 43,44 The advantage of the latter lies in the ability to study analogues of the active site of these enzymes without the complication of the surrounding protein. Furthermore, it has been shown that such synthetic analogues are capable of catalyzing the reduction of protons, a phenomenon that has been extensively explored using not only (µ- S(CH 2 ) 3 S)Fe 2 (CO) 6, the subject of this study (see Figure 1b), which shares a common topology with the active site, but also with more precise models of this site, including the substitution of cyano ligands for carbonyl groups. 43 While the synthesis and electrochemistry of these species is well understood, however, to date there has been no direct study of either the solutionphase structure or the ultrafast vibrational dynamics of these model compounds. An understanding of these is of two-fold benefit, both in the exploitation of model compounds as technological systems for hydrogen production and uptake and as a first step toward future studies of the full enzyme system. The major benefit of model systems such as that studied is the capacity to separate the active site from the surrounding protein scaffold. As such, comparisons of the data obtained herein on model compounds in a range of solvents with future studies of the vibrational structure and dynamics of the active site of the enzyme will be of value in determining the nature of vibrational relaxation mechanisms present and also to explicitly determine the role played by the surrounding protein pocket. The latter is of particular importance because in the enzyme the active site occupies a hydrophobic pocket and as such must rely on the surrounding protein for vibrational relaxation during the reactive cycle. 45 The remainder of the paper is organized as follows. Section II relates the experimental methodology employed, Section III is dedicated to the results and discussion thereof, and conclusions are drawn in Section IV. Figure 2. (a) FTIR spectrum of a 1 mm solution of (µ- S(CH 2) 3S)Fe 2(CO) 6 dissolved in heptane. (b) 2D-IR spectrum of the same sample with a pump-probe time delay of 5 ps. (c) As with b, but with a pump-probe time delay of 25 ps. The dashed lines in the 2D-IR spectra are to guide the eye (see text). II. Experimental Section 2D-IR Spectroscopy. The 2D-IR spectra presented here were obtained using the quasi-frequency domain double-resonance technique, though with some modifications to the established methodology. 6,7 The ultrafast time-resolved infrared experiment has been described in detail elsewhere. 46 The basis of the experiment was an 80 MHz, mode-locked Ti:sapphire oscillator, which was used to seed a two-stage, regenerative, and multipass amplifier. The combination produced 5 mj per pulse output at 800 nm with a repetition rate of 1 khz and a pulse duration of 120 fs. This output was used to pump two white-light-seeded optical parametric amplifiers (OPA) to produce the mid-infrared pump and probe radiation. Both OPAs were equipped with difference frequency generation (DFG) in order to generate radiation centered at 5 µm (2000 cm -1 ), resonant with the carbonyl stretching modes of the model compound, with a bandwidth of 200 cm -1. The use of two OPAs has two advantages. First, it allows greater amounts of mid-ir intensity to be used in the pump beam to counter the detrimental effect on power of the narrow-band filter, vide infra. Second, the use, not employed here, of two-color 2D-IR experiments is made feasible. Following the DFG stage, the pump radiation was directed into a pulse-shaping device. Upon entering the pulse-shaper, the pump pulse was dispersed by a grating before being directed onto a spherical mirror (focal length ) 500 mm) and a large diameter (50 mm) plane mirror. Each of these three elements was separated by the focal length of the spherical mirror to ensure that the return beam was recollimated prior to exiting the pulse shaper. A slit placed in front of the plane mirror was used for wavelength and bandwidth selection from the dispersed pump pulse envelope. Computer control of the horizontal position of the slit enabled the pump wavelength to be scanned

3 Compounds of the [FeFe]-Hydrogenase Enzyme System J. Phys. Chem. B, Vol. 112, No. 32, during acquisition of 2D-IR data. Typically, the bandwidth of the pump pulses used, as set by the slit width, was ca. 10 cm -1 fwhm. This approach is the most significant deviation from the established double-resonance methodology, which normally employs a Fabry-Perot filter in the pump beam. 7 The major advantage to this modification is that the mid-ir pump pulses maintain an approximately Gaussian temporal profile as opposed to the single-sided exponential produced by the Fabry-Perot filter, making pump-probe time delays closer to zero accessible without concerns over effects caused by the temporal overlap of the pulse envelopes. Following wavelength selection, the pump radiation was modulated at 500 Hz using an optical chopper, directed onto an optical delay line to control the temporal delay between pump and probe pulses, and thence sent through a mid-ir λ/2 plate to control the relative polarization direction of the pump pulse with respect to that of the probe pulse. It was then focused into the sample where spatial overlap with the probe radiation was achieved using a calcium fluoride lens (focal length ) 300 mm). Prior to being sent to the sample, the probe radiation was passed through a beamsplitter with a transmitted/reflected beam intensity ratio of 50:50. One portion was diverted into a spectrometer and detected using a 64 element, liquid-nitrogen-cooled HgCdTe array to act as a reference signal. After passing through the sample, the remaining probe light was dispersed and detected by an identical spectrometer/array detector combination. 2D-IR spectra were thus obtained at fixed pump-probe time delays by scanning the slit position in the pulse shaping unit recording narrow-band pump and broad-band probe spectra at a range of pump frequencies. The latter are obtained from the difference between the probe signal obtained in the presence and absence of the pump radiation, normalized by the reference signal to account for laser intensity fluctuations. In the work presented here, the samples were held between two CaF 2 windows separated by 90 µm using a PTFE spacer. It was not necessary to flow the sample, though the air-sensitive nature of the metal carbonyl-derivative samples required that the 1 mm solutions in each solvent be produced in an inertatmosphere environment following freeze-thaw degassing of the solvents. The concentration of the solutions was selected to give a peak optical density in the carbonyl stretching region of the mid-ir region of around 0.4, sufficiently high to yield good signal-to-noise ratios but not high enough to result in concentration-related 2D-IR line shape distortions. The (µ-s(ch 2 ) 3 S)- Fe 2 (CO) 6 used herein was produced using established methods, 43 and all solvents were obtained from Sigma-Aldrich, Ltd., and used without further purification. Samples were filtered using 0.2 µm filters to reduce the experimental noise arising from scattered pump radiation. Density Functional Theory Calculations. All density functional theory (DFT) calculations were carried out using the Gaussian 03 package. 47 Structural optimization calculations of (µ-s(ch 2 ) 3 S)Fe 2 (CO) 6 were carried out using the established crystal structure as a starting point prior to calculation of the 1D infrared spectrum. All simulations were performed on the gas-phase (µ-s(ch 2 ) 3 S)Fe 2 (CO) 6 molecule using the B3LYP functional 48 and the LanL2DZ basis set. The latter employs the Dunning/Huzinaga valence double-ζ D95V 49 basis set for firstrow atoms and the Los Alamos effective core potential plus DZ on atoms from Na-Bi No scaling or correction factors were applied. III. Results and Discussion Spectroscopy: (µ-s(ch 2 ) 3 S)Fe 2 (CO) 6 in Heptane. The 1D, Fourier transform infrared (FTIR) spectrum ofa1mmsolution of (µ-s(ch 2)3 S)Fe 2 (CO) 6 in heptane is shown in Figure 2a alongside 2D-IR spectra recorded with pump-probe time delays of 5 and 25 ps (Figure 2b,c). The FTIR spectrum of (µ-s(ch 2 ) 3 S)Fe 2 (CO) 6 is in excellent agreement with previous studies; 44,53 furthermore, the 1D spectrum of the sample used in the 2D-IR experiments was recorded before and after 2D-IR measurements were taken in order to establish the quantity, if any, of breakdown of the solute caused by either the laser or residual oxygen dissolved in the solvent following degassing. In every case, no change of the FTIR spectrum was observed, and no laser power-dependence of the shape or position of the recovered 2D-IR signals was observed. By examination of the 2D-IR spectrum with parallel pump and probe polarizations and a pump-probe time delay of 5 ps (Figure 2b), it can clearly be seen that negative peaks, corresponding to bleach and stimulated emission signals due to the four main ground-state carbonyl stretching modes of this compound at 2075, 2036, 2006, and 1991 cm -1 are observed on the diagonal of the 2D-IR spectrum as would be expected. These are indicated in Figure 2b by the crossing points of the diagonal dashed line with the horizontal and vertical dashed lines. A small bleach signal due to the weakest of the CO stretching modes is also observed at 1982 cm -1. Additionally, each of these bleach signals is accompanied by a positive, transient absorption peak assignable to the V)1-2 excitedstate absorption of each of these transitions. These peaks are shifted to lower probe wavenumber by the anharmonicity of the carbonyl stretching potential function, as will be discussed in more detail below. In addition to each of the pairs of resonances observed on the diagonal of the spectrum, further such pairs are observed in the off-diagonal region of the spectrum, indicating the presence of vibrational interactions between each of the carbonyl stretching modes in this molecule, as is consistent with previous studies of metal carbonyl derivatives in solution. 1,7 The positions of these cross peaks are marked in Figure 2 by the crossing points of the horizontal and vertical dashed lines. These interactions arise from a combination of vibrational coupling and population transfer effects, both of which have an impact on the offdiagonal peak intensities, but in different ways which will be revisited in detail below. It should be noted that, due to the range of peak amplitudes observed in the 2D-IR spectra, not all of the peaks are clearly observable in Figure 2a, though cross sections through the 2D-IR data unambiguously reveal the peaks marked by intersections of the dashed lines. The 2D-IR spectrum of (µ-s(ch 2 ) 3 S)Fe 2 (CO) 6 in heptane is presented in Figure 2c, again with parallel pump and probe polarization directions but in this case with a pump-probe delay time of 25 ps. A general decrease in amplitude of each of the peaks is observed, and while there is no discernible change in the peak positions or lineshapes, the relative intensities of the cross peaks have increased with respect to the diagonal peaks. The overall decrease is caused by a combination of molecular rotation and vibrational relaxation, and information regarding the molecular structure and the vibrational dynamics can be extracted from the polarization and time dependence of the amplitudes of these peaks. The ultrafast dynamics observed will be discussed further below, but it is initially instructive to quantify the peak positions obtained via 2D-IR spectroscopy in order to obtain information relating to the anharmonic

4 10026 J. Phys. Chem. B, Vol. 112, No. 32, 2008 Stewart et al. TABLE 1: Results of Fitting the Slice through the 2D-IR Spectrum, Shown in Figure 3, to the Sum of 10 Gaussian Functions a pump frequency ) 2036 cm -1 Figure 3. Comparison of experimental data and a fit to a sum of 10 Gaussian functions (see eq 1) for a slice through the 2D-IR spectrum of (µ-s(ch 2) 3S)Fe 2(CO) 6 in heptane taken at a pump frequency of 2036 cm -1. Figure 4. Results of DFT calculations of the gas-phase structure of the (i-s(ch 2) 3S)Fe 2(CO) 6 (inset) and simulated 1D-IR spectrum. The molecule is predicted to belong to the c s point group. FTIR τ pump-probe ) 5ps τ pump-probe ) 25 ps a ω 1 (cm -1 ) σ 1 (cm -1 ) a ω 1 (cm -1 ) σ 1 (cm -1 ) a ω 2 (cm -1 ) σ 2 (cm -1 ) a ω 2 (cm -1 ) σ 2 (cm -1 ) a ω 3 (cm -1 ) σ 3 (cm -1 ) a ω 3 (cm -1 ) σ 3 (cm -1 ) a ω 4 (cm -1 ) σ 4 (cm -1 ) a ω 4 (cm -1 ) σ 4 (cm -1 ) a ω 5 (cm -1 ) σ 5 (cm -1 ) a ω 5 (cm -1 ) σ 5 (cm -1 ) a Also shown are the results of fitting the FTIR spectrum of (µ-s(ch 2) 3S)Fe 2(CO) 6 (Figure 2a) to a series of Gaussian functions. vibrational potentials of the carbonyl stretching modes. In this case, this can be achieved by fitting slices (or narrow-band pump-broadband probe spectra) through the 2D-IR spectra at pump frequencies corresponding to the peaks observed on the diagonal of the spectrum to the sum of 10 Gaussian lineshapes (5 each for transitions originating in the ground and excited vibrational states of each of the CO stretching modes respectively) given by n[ I n (ω) ) a ( exp -4ln2(ω-ω )] n )2 (1) 2 σ n where ω n and σ n represent the central frequency and width parameter of the n th Gaussian function, respectively. An example of the fit to the data obtained for a pump frequency of 2036 cm -1 is shown in Figure 3, and the results are shown in Table 1 for pump-probe delays of 5 and 25 ps. Data for a pump frequency of 2036 cm -1 is used for illustration purposes because it represents the largest of the observed 2D-IR peaks, however, similar results were obtained at each of the pump frequencies corresponding to the four most intense carbonyl stretching modes observed. Table 1 also shows the results of fitting the FTIR spectrum of (µ-s(ch 2 ) 3 S)Fe 2 (CO) 6 to a sum of five Gaussian functions for comparison purposes. The labeling of the 10 Gaussian functions used for the fit shown in Figure 3 employs the subscripts 1-5 to allow comparison between the negative, bleach and stimulated emission lineshapes with those present in the FTIR spectrum because these correspond to transitions to or from the vibrational ground state (V ) 0). Transient absorption (positive) peaks originating from V)1 are signified using 1-5, indicating that they form part of a positive-negative pair with the negative peak of the same number. It is clear from Table 1 that the agreement between the FTIR frequencies and those of the negative, bleach peaks is very good. The one discrepancy occurs at the very highest frequency peak at 2075 cm -1. The 2D-IR peak is observed at 2070 cm -1, though this is likely to have been caused by small errors in extrapolating the frequency calibration of the pixels in the HgCdTe array to the extent of the detector. The data in Table 1 also reinforce the observation that there is little change in the peak lineshapes or positions as a function of pump-probe delay time. It is noticeable that there is a general small increase in the observed line width parameter in the 2D- IR spectrum in comparison with that observed in the FTIR spectrum, though this is likely to be a result of the lower resolution of the detection method used as the grating and array detector pair limit the resolution of the 2D-IR spectrometer to around 4 cm -1. From the results of fitting the slices through the 2D-IR spectra, it is possible to shed some light upon the anharmonic frequency shifts that occur between the V)0-1 and 1-2 transitions of this molecule. These can be obtained from the relative peak positions of the pairs of negative and positive peaks obtained from the fitting process. Fitting is required to account for any errors caused by the overlap of positive and negative peak envelopes, which may distort the lineshapes. It should be noted that without this anharmonicity it would be impossible to

5 Compounds of the [FeFe]-Hydrogenase Enzyme System J. Phys. Chem. B, Vol. 112, No. 32, TABLE 2: Results of DFT Calculations of the Vibrational Modes of (µ-s(ch 2 ) 3 S)Fe 2 (CO) 6 in the CO Stretching Region a mode (cm -1 ) 2037 ( ) 1985 ( ) 1982 ( ) 1967 ( ) expt (cm -1 ) assignment (CO) 6 sym str (CO) 3FeFe(CO) 3 asym str basal CO asym str apical CO sym str Fe(CO) 3 asym str a Table includes calculated frequencies, assignments, and angles between transition dipole moment vectors for each mode. observe the 2D-IR spectra shown in Figure 2 because the coincidence of the bleach, stimulated emission, and transient absorption frequencies would lead to cancellation of the observed signals. From the data in Table 1, it can be seen that the 0-1 transitions and anharmonic shifts for the five carbonyl stretching modes are 2075, 2036, 2006, 1991, and 1982 cm -1 and 2.7, 5.9, 5.9, 4.1, and 4.7 cm -1, respectively. The anharmonic shift values have been obtained via an averaged value taken from the 5 and 25 ps 2D-IR spectra in Figure 2. In order to assign these vibrational frequencies to stretching modes of the carbonyl ligands of (µ-s(ch 2 ) 3 S)Fe 2 (CO) 6, DFT calculations were carried out to obtain estimates of the equilibrium structure and simulations of the 1D infrared spectrum. The results of these calculations are shown in Figure 4 and Table 2. It can be seen from Figure 4 that the simulated spectrum is in close agreement with the measured 1D spectrum, to an accuracy of 98% in the frequency of each mode. It is slightly surprising perhaps that the calculated frequencies consistently underestimate the measured ones, though the necessary use of the effective core potential-based basis set due to the presence of very heavy atoms will lead to lower accuracy than might normally be expected. It should be noted that the mode assignments given in Table 2 are based upon the normal representation of each Fe atom as having an approximately square-based pyramidal structure in which two carbonyl groups and two sulfur ligands form the base of the pyramid (Figure 1), hence the distinction between the two basal and one apical carbonyl groups. While the agreement obtained through these calculations is good, there remains the possibility that the ordering of the five calculated carbonyl stretching modes is different in the calculations from those observed via infrared spectroscopy. It is possible to gain further insight into this by comparing the calculated angles between the transition dipole moments of each of the modes with those obtained directly from 2D-IR spectroscopy. This will also allow some light to be shed on the comparability of the solution-phase and calculated gasphase structure of (µ-s(ch 2 ) 3 S)Fe 2 (CO) 6. In order to carry this out, it is necessary to investigate the ultrafast dynamics of the peaks observed in the 2D-IR spectra in Figure 2. Ultrafast Dynamics. In order to obtain detailed information on the vibrational dynamics and molecular structure of (µ- S(CH 2 ) 3 S)Fe 2 (CO) 6 in heptane, a series of spectra were recorded in which the pump frequency was fixed to coincide with each of the three most intense carbonyl-stretching resonances of (µ- S(CH 2 ) 3 S)Fe 2 (CO) 6 in turn (2035, 2006, and 1992 cm -1, respectively); the highest and lowest frequency bands were not used due to lower signal-to-noise ratios obtained when using the weaker transitions. Narrow-band pump, broad-band probe spectra were then recorded at a series of pump-probe time delays. This type of experiment is one of the strengths of the Figure 5. Examples of the ultrafast dynamical data obtained for each of the transitions observed in a slice through the 2D-IR spectrum of (µ-s(ch 2) 3S)Fe 2(CO) 6 in heptane with a pump frequency of 2006 cm -1. (a, top) Variation in peak amplitude with pump-probe time delay using the magic angle pump-probe polarization geometry. (b, bottom) Temporal evolution of the anisotropy of each of the observed peaks. double-resonance technique, effectively allowing slices through the 2D-IR spectrum to be recorded at high time resolution without the need to obtain the whole 2D-IR spectrum, which is significantly less efficient. Furthermore, obtaining the above time traces with both parallel and perpendicular pump-probe polarization geometries allows subsequent calculation of both the magic angle and anisotropy temporal profiles of the lines observed, facilitating the separation of the effects of molecular rotation from vibrational relaxation and also a determination of the angles between the transition dipole moments of the observed bands, respectively. 1,54 Examples of this data are shown in Figure 5, which displays the results of the experiments carried out at a pump frequency of 2006 cm -1. Figure 5a shows the time-dependence of the magic angle polarization signal for each of the observed peaks, which are labeled by their observed central frequency. These are clearly separable into the bleach and transient absorption signals and have been color-coded into the four observed V) 0-1 and V)1-2 transition pairs as discussed above. In order to quantify this data, the profiles have each been fit to multiexponential functions over pump-probe time delays ranging from 2 to 150 ps. In the majority of cases, a singleexponential decay time (be it of bleach or transient absorption) was all that was required; however, certain lines showed timedomain dynamics that could not be well-represented by such a simple function. Some required a biexponential decay function, while others displayed a clear risetime followed by a slow single-exponential decay. The results are summarized for all studied pump frequencies in Table 3. When the resonances at 2006 or 2035 cm -1 were pumped, the observed diagonal peak pair, i.e., the V)0-1 and 1-2 transitions corresponding to the pumped mode, both display a biexponential decay function with a fast decay of 6 ps followed by a slow decay of 120 ps. In contrast, the remaining three bleach signals show single-exponential dynamics with a decay time of around 130 ( 10 ps. It should be noted that the

6 10028 J. Phys. Chem. B, Vol. 112, No. 32, 2008 Stewart et al. TABLE 3: Results of Fitting Dynamical Data Such as That Shown in Figure 5 to Exponential Functions a pump υ (cm -1 ) probe υ (cm -1 ) B/A τ R heptane τ D1 τ D2 τ R hexadecane τ D1 τ D B B B B A A A A B B B B b A A A A B b B B B b A b b b A A b b b 1987 A b a The combination of the three time constants used reflects the type of function that produced the best fit, τ Dn are the decay constants for a single and biexponential function (n ) 1,2), while τ R is the rise time, if required. Data obtained for (µ-s(ch 2) 3S)Fe 2(CO) 6 dissolved in hexadecane is shown alongside heptane data for comparison purposes. Note that B/A correspond to a bleach or transient absorption signals, respectively. b Undefined due to poor S/ N ratio. spread of observed values results from the reduction in signalto-noise levels beyond pump-probe time delays of 150 ps, which increases the error in the fitted parameters; this is particularly true of the less intense absorptions. In contrast to the bleach signals, the dynamics observed for the off-diagonal transient absorption signals are much more complex. Those lying to the blue side of the pumped transition show a clear, fast risetime on the order of 5-10 ps followed by a single-exponential decay in the region of 100 ps, while the transient absorption lying closest in frequency to the red side of the pumped mode shows a biexponential decay with parameters of 5 and 100 ps, respectively. Those lying further to the red again show a risetime and decay profile. That the presence of a biexponential decay was only observed in the modes closest in frequency to the pumped transition could be attributable to minor overlap with the pump pulse bandwidth, though this effect is seen in modes separated by >30 cm -1 in the case of the 2035 cm -1 pump experiment, suggesting that another mechanism may be responsible. When the weaker 1992 cm -1 mode was pumped, the bleach again shows a biexponential decay profile with time scales similar to those above, but in this case two of the three bleaches lying to the blue side of this mode showed a risetime of around 5 ps followed by a 145 ps decay. The diagonal transient absorption associated with V)1-2 transition of this pumped mode shows a biexponential (albeit poorly defined due to the signal-to-noise ratio for this transition) decay, and the modes lying to the blue side that produced good quality fits to the data again show a 5 ( 1 ps risetime followed by a 130 ( 10 ps decay, in keeping with the major patterns above. The data observed would appear to be consistent with a model in which the carbonyl stretching modes of (µ- S(CH 2 ) 3 S)Fe 2 (CO) 6 are coupled, as each of the off-diagonal peaks is apparently nonzero at zero pump-probe time delay (Figure 5a). This observation is somewhat complicated here by effects arising from coincident pump and probe pulses on the sample at time zero 7 but can be reasonably asserted based upon prior 2D-IR studies of metal carbonyl derivatives. 1,7 In the case of the bleaches, the off-diagonal bleaches then recover with a single-exponential relaxation time of around 120 ps, which would appear to indicate the vibrational lifetime of these modes in heptane. Such a figure is not inconsistent with previous observations of similar molecules. 55,56 Further support for this assignment arises from the fact that all modes, whatever the ultrafast dynamics near time zero, ultimately relax with a similar time constant. In the case of the bleach corresponding to the pumped mode, these consistently show a biexponential decay that is also seen in the decay of the associated transient absorption corresponding to the V)1-2 transition of the pumped mode. This biexponential features a fast decay of around 5 ps followed by the longer 120 ps component common to all bleaches. This would appear to be consistent with a quantity of fast vibrational population transfer among the carbonyl stretching modes occurring immediately after the pumping of one of the modes. Such an effect has been observed previously in smaller carbonyl systems when employing narrow-band pump, broad-band probe spectroscopy, with similar time scales reported. 55,56 The concept of population transfer is also consistent with the rise times observed for the transient absorption signals, the higher frequency modes each showing a 5-10 ps risetime as population transfer occurs, followed by the universal ps relaxation dynamics. Similarly, the observation of a risetime for the bleaches when the 1992 cm -1 mode was pumped can be explained using this model. It should be considered that the signals in this region of the spectrum are smaller than those measured with pump frequencies of 2006 and 2035 cm -1.As such, it is feasible that population transfer leads to increased stimulated emission from the V)1 state leading to the observed risetime. This would be less obvious in the presence of stronger bleach signals. The lower signal-to-noise ratios also account for the observed slightly longer decay constant. The observation of only two (within experimental error) time constants would also suggest a relatively straightforward vibrational decay mechanism despite the complex spectroscopy of this system. Such vibrational energy transfer pathways arise through the actions of the solvent bath upon the excited molecule. The long vibrational lifetime of this species arises from the fact that interactions between the carbonyl stretching modes of the solute and the solvent are weak. Furthermore, there are no solvent absorptions close in energy to the excited modes of the solute to facilitate rapid vibrational energy transfer into the surrounding solvent bath. This must therefore occur through an overtone or combination band modes of the solvent. With this in mind, however, it is apparent that the vibrational lifetime of the carbonyl modes of (µ-s(ch 2 ) 3 S)Fe 2 (CO) 6 is somewhat shorter than that observed in other simpler carbonyl systems This may arise from the greater number of carbonyl stretches of similar frequency, allowing a greater flexibility in terms of vibrational relaxation pathway.

7 Compounds of the [FeFe]-Hydrogenase Enzyme System J. Phys. Chem. B, Vol. 112, No. 32, The population transfer process, especially to modes lying to higher energy than the pumped mode, has also been studied previously. 58 This is thought to arise from interactions with the low-frequency density of states of the surrounding solvent bath. Most liquids possess a broad band of inter- and intramolecular vibrational modes in the range of cm -1, though this can be broader in more strongly interacting liquids Thus, population transfer to higher and lower frequency modes within 150 cm -1 of the pumped mode is facilitated by scattering mechanisms involving phonon modes of the solvent bath allowing rapid equilibration of the excited vibrational population among the carbonyl stretching modes. 58 The observation of such vibrational relaxation mechanisms in model compounds of the hydrogenase enzyme system is of significance for the full enzyme. In particular, it is interesting to note that the active site of the enzyme exists within a pocket provided by the protein scaffold into which no aqueous solvent can penetrate. This implies that any such vibrational relaxation of the enzyme would have to be provided by the surrounding scaffolding. As such, future comparisons with similar measurements involving the enzyme would be instructive. In addition to obtaining information relating to the vibrational dynamics and energy transfer pathways of this molecule, the 2D-IR dynamical studies above also provide the opportunity to establish the molecular structure of the carbonyl ligands in solution. While the crystalline structure of (µ-s(ch 2 ) 3 S)- Fe 2 (CO) 6 is known, 63 the solution-phase structure remains unclear. With the use of the anisotropy of the off-diagonal peaks in the 2D-IR spectra, it is possible to compare the observed angular relationships between the transition dipole moments of the carbonyl modes with those calculated using DFT. Good agreement with the simulations would confirm that the calculated gas-phase structure, which is very similar to that observed from X-ray crystallography, 63 and the solution-phase structure are similar and also confirm the ordering of the modes in the simulated FTIR spectrum and thus their assignment. Furthermore, the time-dependence of the anisotropy allows the determination of the molecular rotation times of (µ-s(ch 2 ) 3 S)- Fe 2 (CO) 6 in solution. The data in Figure 5b shows the anisotropy for each of the observed bands when a pump frequency of 2006 cm -1 is used. In this case, it can be seen from the anisotropy values at time zero that the pumped mode (and the associated V ) 1-2 transition), in red, possess transition dipole moments that are parallel to each other, as would be expected, and furthermore, that it lies perpendicular to those of the other three most intense carbonyl stretching modes, as indicated by anisotropy values of 0.4 and -0.2, respectively. Repeating this operation for each of the three pumped modes at 2035, 2006, and 1992 cm -1 reveals the transition dipole angular relationships for each of the four main carbonyl stretches. These are all consistent with the calculated values in Table 2, confirming that the point group in the solution phase is indeed c s and the validity of the mode assignments calculated by DFT. Fitting of the temporal decay of the anisotropy parameters in Figure 5b shows a single-exponential relationship with a decay time constant of around 17 ps. This indicates the rotational time scale of the molecule in heptane. This value is somewhat smaller than may be expected for such a large molecule; however, it must be taken into account that the molecule is approximately spherical in overall shape and that the viscosity of the solvent is very low (0.387 mpa s at 298 K). 64 An additional factor that must also be taken into consideration is the limited interaction with the surrounding solvent, which will Figure 6. FTIR (top) and 2D-IR (bottom) spectra of (µ- S(CH 2) 3S)Fe 2(CO) 6 dissolved in CH 3CN. The pump-probe time delay was 5 ps, and the pump-probe polarization geometry was parallel when recording the 2D-IR spectrum. Dashed lines are used to guide the eye. Small deviations of the line positions from the diagonal are caused by calibration errors of the pump frequency, which are exacerbated by the broader lineshapes observed in acetonitrile. occur mainly through dispersive interactions. As such, it may be expected that slip boundary conditions may be more applicable than stick conditions, leading to a decoupling of the molecular rotational motion from the surrounding solvent. Other Solvents. In an effort to further determine the effects of the solvent bath upon the phenomena described above, the molecule (µ-s(ch 2 ) 3 S)Fe 2 (CO) 6 was also dissolved in acetonitrile (CH 3 CN) and hexadecane. The former is significantly more polar than both heptane and hexadecane, while hexadecane offers an almost order of magnitude increase in solvent viscosity. The results obtained in hexadecane were largely in excellent agreement with those obtained for the heptane sample. The 1D and 2D-IR data showed little change, as would be expected for such similar solvents. The major change was observed when studying the dynamical data, i.e., that obtained by measuring the time-dependence of the vibrationally excited states of each of the carbonyl stretching modes as slices through the 2D-IR spectrum. While the vibrational lifetime data as shown in Table 3 was observed to be identical to that of the heptane sample, within experimental error (see Table 3), the anisotropy relaxation time increased to 53 ps. This is as might be expected given a more viscous solvent (viscosity of hexadecane ) 3.03 mpa s at 298 K 64 ) although it is interesting to note that the increase is not proportional to the change in viscosity, as suggested by the Debye-Stokes-Einstein relation. The latter observation may be due in large part to the lack of interaction between solvent and solute leading to weak coupling between the rotational relaxation time and the macroscopic solvent viscosity as discussed above. The 1D and 2D-IR spectra of (µ-s(ch 2 ) 3 S)Fe 2 (CO) 6 dissolved in CH 3 CN are shown in Figure 6. The 2D-IR spectrum shown was obtained using a pump-probe time delay of 5 ps and parallel pump-probe polarization geometry, though as with the heptane and hexadecane samples, no evolution of the 2D- IR peak shapes or positions was observed with either pump-probe delay time or polarization geometry.

8 10030 J. Phys. Chem. B, Vol. 112, No. 32, 2008 Stewart et al. TABLE 4: Results of Fitting Magic Angle Dynamical Data for (µ-s(ch 2 ) 3 S)Fe 2 (CO) 6 Dissolved in CH 3 CN a solvent pump υ (cm -1 ) probe υ (cm -1 ) B/A a 1 τ D2 CH 3CN B B B A A b 1974 A Figure 7. Vibrational lifetime dynamics for (µ-s(ch 2) 3S)Fe 2(CO) 6 dissolved in CH 3CN. Data shown is magic angle relaxation of signals observed when the pump wavelength was 2032 cm -1. The effects of the polar solvent are clearly apparent upon the FTIR spectrum of (µ-s(ch 2 ) 3 S)Fe 2 (CO) 6, the line broadening effect leading to a dramatic loss of spectral resolution. However, it is interesting to note that this does not manifest itself as inhomogeneous broadenening of the 2D-IR lineshapes. Such broadening would be expected to lead to a diagonal elongation of the peaks in the 2D-IR spectrum, but this is not the case. 18 A slight elongation along the pump frequency axis direction is observed, though this is likely to be due to the combination of increased line width as compared with the heptane and hexadecane samples with the finite bandwidth of the pump pulse leading to this apparent broadening effect. A similar phenomenon has been previously observed when studying broad lineshapes with the double-resonance 2D-IR method. 7 One possible explanation for the lack of diagonal elongation of the lineshapes is that the time scale of any inhomogeneous broadening is faster than the experimental time resolution. This would lead to complete spectral diffusion within the pump-probe time delay period and hence circular lineshapes at all pump-probe time delays, as have been observed. Indeed, diagonally elongated lineshapes are most often observed in hydrogen-bonded systems such as aqueous peptide solutions and indeed in water itself. 65,66 As no hydrogen bonding is present in the CH 3 CN/(µ- S(CH 2 ) 3 S)Fe 2 (CO) 6 solution, it is possible that any interactions are weaker and thus more transient than hydrogen-bonding processes. It is also instructive to consider the vibrational lifetime data obtained for the CH 3 CN solution. As in the case of hexadecane and heptane, the temporal dependence of the magic angle dynamics and anisotropy of each of the modes has been studied at pump frequencies corresponding to each of the three bands observed in the 1D infrared spectrum at 2074, 2032, and 1995 cm -1. In the case of the anisotropy decay, a value of 21 ps was observed in all cases, which corresponds well with the value of 17 ps in heptane given the similar viscosity values for the two solvents (CH 3 CN viscosity ) mpa sat298k 64 ). It should be noted that anisotropy parameters were obtained only for the two higher frequency carbonyl stretching modes of (µ- S(CH 2 ) 3 S)Fe 2 (CO) 6 in CH 3 CN as the broadened linewidths will lead to convolution effects where the lower frequency absorptions overlap. The dynamics observed for each of the bands observed in CH 3 CN when pumped at 2032 cm -1 are displayed in Figure 7. This data corresponds to the magic angle polarization geometry and as such relates to the vibrational lifetime of the transitions studied. As in the case of heptane above, these data were quantified by fitting to exponential functions. In contrast to the heptane and hexadecane samples, however, a single-exponential decay profile sufficed in each case, providing good agreement B B B A A A B B B A A A a A single-exponential function was used in each case. Note that B/A correspond to a bleach or transient absorption signal, respectively. b Undefined due to poor S/N ratio. with the observed decay profiles. The results of this fitting process are shown in Table 4. In each case, a decay time scale of 110 ( 20 ps was observed. It is interesting to note that the observed biexponential decays exhibited by the pumped transitions are no longer observed, similarly that the rise times shown by the transient absorptions lying to the blue side of the pumped bands are also no longer evident. The red traces in Figure 7, corresponding to the pumped transition and its V)1-2 transient absorption partner, do appear to show a biexponential character, but this appears to be extremely fast and was not reliably resolvable during the fitting process. These observations imply that the solvent-solute interactions are somewhat stronger in CH 3 CN, as might be anticipated from the broader lineshapes and more polar nature of the solvent. This apparently speeds up the process of vibrational population transfer between the carbonyl modes, hence the fact that the 5 ps processes observed in the alkane solvents are no longer observed. In light of this, it is interesting to note that the overall vibrational relaxation time of the carbonyl modes is similar to that obtained in the alkane solvents. It might be expected that this process would be faster in a more polar solvent; however, metal carbonyl transitions are well separated from most solvent bands, even in the case of CH 3 CN, which possesses a CN stretching mode at 2250 cm -1. The reason for this is that, although closer than any alkane solvent vibrational modes, this band is still some 176 cm -1 above the highest frequency carbonyl stretching mode of (µ-s(ch 2 ) 3 S)Fe 2 (CO) 6. Thus, given that the involvement of the low-frequency density of states of the solvent would be required in order to utilize this mode as a relaxation pathway, the fact that the CH 3 CN density of states does not have significant amplitude at 176 cm -1 may well be prohibitive. 58 IV. Conclusions Ultrafast 2D-IR spectroscopy has been applied for the first time to study the solution-phase structure and vibrational dynamics of a model compound specifically relating to the

9 Compounds of the [FeFe]-Hydrogenase Enzyme System J. Phys. Chem. B, Vol. 112, No. 32, hydrogenase enzyme system. 2D-IR spectra in conjunction with DFT calculations have shown that the structure of (µ- S(CH 2 ) 3 S)Fe 2 (CO) 6 in solution is very similar to that predicted in the gas phase and measured using X-ray crystallography. Ultrafast dynamics obtained from 2D-IR spectroscopy reveal that vibrational relaxation following excitation of individual carbonyl modes occurs on two separate time scales in alkane solvents: Vibrational population transfer among the carbonyl modes occurs on a time scale of around 5 ps followed by ps vibrational relaxation to the ground state. While both these effects might be expected to be solvent-mediated, the population transfer rate appeared to be most strongly influenced by solvent polarity. Data recorded for (µ- S(CH 2 ) 3 S)Fe 2 (CO) 6 in CH 3 CN showed only a ps decay time scale, suggesting that the population transfer rate had become too fast to be reliably resolved. This is supported by the observation of 2D-IR lineshapes apparently indicating non-inhomogeneously broadened transitions. Conversely, little effect upon the vibrational relaxation time was observed. The reason for this discrepancy arises from the greater interactions with the polar solvent through the low-frequency density of states of CH 3 CN, allowing more effective population transfer among closely spaced carbonyl modes as compared with that of alkane solvents. However, the increased width of the density of states of CH 3 CN in comparison with that of alkane solvents is insufficient to bridge the gap between the carbonyl modes and high-frequency vibrational absorptions of CH 3 CN, which would facilitate a significantly shorter vibrational lifetime. It is anticipated that the results obtained herein will be of benefit as comparison points for future studies of the structure and vibrational dynamics of the full enzyme system, in which the ability to separate the active site and protein scaffold will provide new insights into the role of the protein pocket. Furthermore, they will contribute to the knowledge base surrounding applications of hydrogenase enzyme-based technologically relevant systems for hydrogen production and utilization. Acknowledgment. The authors would like to acknowledge the Engineering and Physical Sciences Research Council of the U.K. (EPSRC) for an Advanced Research Fellowship (N.T.H.) and postgraduate studentship (A.I.S.). Funding for this work has also been provided by the U.K. Science and Technology Facilities Council (STFC) for work carried out at the Central Laser Facility, Rutherford Appleton Laboratory. CJP and SKI wish to thank the BBSRC and EPSRC (Supergen V) for supporting this work. References and Notes (1) Khalil, M.; Demirdoven, N.; Tokmakoff, A. J. Phys. Chem. A 2003, 107, (2) Finkelstein, I. J.; Zheng, J. R.; Ishikawa, H.; Kim, S.; Kwak, K.; Fayer, M. D. Phys. Chem. Chem. Phys. 2007, 9, (3) Bredenbeck, J.; Helbing, J.; Kolano, C.; Hamm, P. ChemPhysChem 2007, 8, (4) Asplund, M. C.; Zanni, M. T.; Hochstrasser, R. M. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, (5) Hahn, E. L. Phys. ReV. 1950, 80, xxx. (6) Hamm, P.; Lim, M.; De Grado, W. F.; Hochstrasser, R. M. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, (7) Cervetto, V.; Helbing, J.; Bredenbeck, J.; Hamm, P. J. Chem. Phys. 2004, 121, (8) Demirdoven, N.; Cheatum, C. M.; Chung, H. S.; Khalil, M.; Knoester, J.; Tokmakoff, A. J. Am. Chem. Soc. 2004, 126, (9) Cheatum, C. M.; Tokmakoff, A.; Knoester, J. J. Chem. Phys. 2004, 120, (10) Chung, H. S.; Khalil, M.; Tokmakoff, A. Biophys. J. 2004, 86, 526A. (11) Smith, A. W.; Cheatum, C. M.; Chung, H. S.; Demirdoven, N.; Khalil, M.; Knoester, J.; Tokmakoff, A. Biophys. J. 2004, 86, 619A. (12) Rubtsov, I. V.; Wang, J.; Hochstrasser, R. M. J. Chem. Phys. 2003, 118, (13) Hamm, P.; Lim, M.; Hochstrasser, R. M. J. Phys. Chem. B 1998, 102, (14) Kim, Y. S.; Wang, J.; Hochstrasser, R. M. J. Phys. Chem. B 2005, 109, (15) Fang, C.; Senes, A.; Cristian, L.; De Grado, W. F.; Hochstrasser, R. M. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, (16) Khalil, M.; Tokmakoff, A. Chem. Phys. 2001, 266, 213. (17) Okumura, K.; Tokmakoff, A.; Tanimura, Y. Chem. Phys. Lett. 1999, 314, 488. (18) Roberts, S. T.; Loparo, J. J.; Tokmakoff, A. J. Chem. Phys. 2006, 125, (19) Tokmakoff, A. Science 2007, 317, 54. (20) Fecko, C. J.; Loparo, J. J.; Roberts, S. T.; Tokmakoff, A. J. Chem. Phys. 2005, 122, (21) Steinel, T.; Asbury, J. B.; Corcelli, S. A.; Lawrence, C. P.; Skinner, J. L.; Fayer, M. D. Chem. Phys. Lett. 2004, 386, 295. (22) Fayer, M. D. Abstr. Pap. Am. Chem. Soc. 2004, 227, 1. (23) Asbury, J. B.; Steinel, T.; Kwak, K.; Corcelli, S. A.; Lawrence, C. P.; Skinner, J. L.; Fayer, M. D. J. Chem. Phys. 2004, 121, (24) Asbury, J. B.; Steinel, T.; Fayer, M. D. J. Phys. Chem. B 2004, 108, (25) Woutersen, S.; Mu, Y.; Stock, G.; Hamm, P. Chem. Phys. 2001, 266, 137. (26) Kwak, K.; Zheng, J. R.; Cang, H.; Fayer, M. D. J. Phys. Chem. B 2006, 110, (27) Zheng, J. R.; Kwak, K.; Asbury, J. B.; Chen, X.; Piletic, I. R.; Fayer, M. D. Science 2005, 309, (28) Cahoon, J. F.; Sawyer, K. R.; Schlegel, J. P.; Harris, C. B. Science 2008, 319, (29) Bredenbeck, J.; Helbing, J.; Hamm, P. Phys. ReV. Lett. 2005, 95, (30) Bredenbeck, J.; Helbing, J.; Behrendt, R.; Renner, C.; Moroder, L.; Wachtveitl, J.; Hamm, P. J. Phys. Chem. B 2003, 107, (31) Kolano, C.; Helbing, J.; Kozinski, M.; Sander, W.; Hamm, P. Nature 2006, 444, 469. (32) Naraharisetty, S. R. G.; Kasyanenko, V. M.; Rubtsov, I. V. J. Chem. Phys. 2008, 128, (33) Kurochkin, D. V.; Naraharisetty, S. R. G.; Rubtsov, I. V. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, (34) DeCamp, M. F.; DeFlores, L. P.; Jones, K. C.; Tokmakoff, A. Opt. Express 2007, 15, 233. (35) Nee, M.; McCanne, R.; Kubarych, K.; Joffre, M. Opt. Lett. 2007, 32, 713. (36) Strasfeld, D. B.; Shim, S. H.; Zanni, M. T. Phys. ReV. Lett. 2007, 99, (37) Shim, S. H.; Strasfeld, D. B.; Ling, Y. L.; Zanni, M. T. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, (38) Yang, X.; Razavet, M.; Wang, X. B.; Pickett, C. J.; Wang, L. S. J. Phys. Chem. A 2003, 107, (39) Frey, M. Struct. Bonding (Berlin) 1998, 90, 98. (40) Fontecilla-Camps, J. C. Coord. Chem. ReV. 2005, 249, (41) de Lacey, A. L.; Fernandez, V. M.; Rousset, M. Coord. Chem. ReV. 2005, 249, (42) Frey, M. ChemBioChem 2002, 3, 153. (43) Liu, X. M.; Ibrahim, S. K.; Tard, C.; Pickett, C. J. Coord. Chem. ReV. 2005, 249, (44) Borg, S. J.; Tye, J. W.; Hall, M. B.; Best, S. P. Inorg. Chem. 2007, 46, 384. (45) Nicolet, Y.; Cavazza, C.; Fontecilla-Camps, J. C. J. Inorg. Biochem. 2002, 91, 1. (46) Towrie, M.; Grills, D. C.; Dyer, J.; Weinstein, J. A.; Matousek, P.; Barton, R.; Bailey, P. D.; Subramanian, N.; Kwok, W. M.; Ma, C. S.; Phillips, D.; Parker, A. W.; George, M. W. Appl. Spectrosc. 2003, 57, 367. (47) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson,