The spectral modeling was developed on the basis of the coordination chemical principles of weak field Fe-S cluster chemistry. A good review of weak-field Fe-S complexes can be found by Holm et al. DOI: 0.1021/cr020615+ .
In our reconstitution experiments, the [Fe(SR)4]2-/-, rubredoxin like complexes are considered as unwanted, byproducts given that we are exclusively focused on [4Fe-4S]-maquettes. However, these complexes provide a simple and clear first step for the development of the spectral model. UV-vis or electronic absorption spectroscopy is a valence excited state technique, which utilizes the 200 – 800 nm (12,500 – 50,000 cm-1) region of electromagnetic radiation. This energy range allows for the excitation of electrons from the occupied to the unoccupied frontier orbitals in the ~1.5 – 6.2 eV energy window.
For transition metal complexes, assuming innocent coordination environment and normal M–L chemical bonding, the frontier orbitals are dominantly metal d-orbital based. A simple bonding model that is already insightful is ligand field theory which considers explicitly the metal d-orbitals and implicitly the ligand environment through the Racah parameters.
The ligand field splitting for a weak field coordination environment of a transition metal ion is small due to the s- and p-donation of the ligand (thiolates). The ideal tetrahedral splitting is shown in Figure 1 with the symmetry labels of the orbitals. The small ligand field splitting (D) practically renders the d-orbitals behave as atomic orbitals at the weak field limit and thus Hund's rule apply, i.e. 1) max ms and 2) max ml quantum numbers. This renders the complexes high spin (Figure 2).
Figure 1. Td ligand field splitting Figure 2. High spin d6 configuration
Considering the occupation of the d-orbital of the central ion, even in the case of homoleptic complex such as the rubredoxin-like tetrathiolate complex, the symmetrical orbital splitting may be distorted for non-degenerate occupation of energetically degenerate orbitals (Jahn-Teller distortion force lowering the electronic energy). However, since both distortions shown in Figure 3 are equally possible, the molecular structure and even the room temperature spectra will actually not display the results of the distortion (dynamic Jahn-Teller effect) other then broadening of the spectral features.
Figure 3. Equivalent electronic distortions (dynamic JT) of the ligand-field for d6 tetrahedral complexes.
In the ligand-field set, the only allowed excitation involves the spin-down (ß) electron in the lowest lying d-orbital into any of the higher lying vacant ß-electron holes. Notably, the excitation energy within the e-set will be independent from the ligand field strength, while it can be a measure of the Jahn-Teller distortion force, if resolved. The e→t2 excitation is the so-called d→d or ligand-field transitions (LFT), which commonly occur at longer wave length region of the UV-vis, often in the near-IR region.
Now considering non-metal based orbitals below and above the ligand field set (Figure 4) we can define all possible excitations that we will model by fitting the UV-vis spectra. For the sake of generality, the degeneracy of d-orbitals are lifted as expected for the rubredoxin-like Fe-thiolate complex, given that the Fe-S(thiolate) bond rotation will eliminate the symmetry of the coordination environment.
Figure 4. Complete molecular orbital energy diagram of the [Fe(SR)4] complex
The molar extinction coefficients (oscillator strength) and the excitation energies clearly classify the various adsorption bands that can be used (as envelop of numerous excitations) to fit and model the spectral features of rubredoxin-like, mononuclear Fe-thiolate complexes.
Building on the above electronic structure model we can extend the molecular orbital diagram of the mononuclear Fe-thiolate complex by considering a new ligand environment as created by the sulfide ions. Sulfides are even more weak field ligands than the S(thiolate) due to the availability of three S 3p2 lone pairs for s- and p-donation to both Fe centers they are connected to (Figure 5).
Figure 5: Fe-S clusters of plant and bacterial ferredoxins
For the construction of the molecular orbital diagram, it is important to consider the effective nuclear charge of the S atoms seen by the S 3p orbitals in the sulfide and in the thiolate. Due to the more negative charge of the sulfides, its 3p-based lone pairs are located at higher energy than the lone pairs of the S(thiolate). Thus, the LMCT bands in of Figure 4 will split into two groups; the lower energy will correspond to the sulfide → Fe while the higher energy will be due to thiolate S → Fe charge transfers. Figure 6 shows the basis of the spectral modeling that we will use for quantitation of the [4Fe-4S]-maquette reconstitution yields.
Figure 6. Schematic blocks of sulfide and thiolate donor and Fe-based acceptor orbitals
Thus, the UV-vis modeling will be accomplished by representative fits to the following features
low energy, weak (e ≈ 10) d→d LFT with one broad envelop
intermediate energy, strong (e ≈ 103) LMCT associated with sulfides → Fe
high energy, strong (e ≈ 103) LMCT associated with thiolate S → Fe
at the low l limit of the UV-vis range strong (e ≈ 105) LLCT or LL'CT associated with sulfide/thiolates
Caveats:
due to the presence of FeS(aq) nanoparticles (FeS)n<150, which appears as turbidity at the high l limit of the spectrum, the LFT feature cannot be used for analytical quantitation because this may significantly vary from sample to sample depending on the success of reconstitution experiment.
similarly at the high energy end, the LLCT or LL'CT excitations are ill-defined and these may “steal” intensity away from the thiolate LMCT band.
more caveats will be listed here as they are identified.
We have considered numerous fitting strategies to find an optimum number of peaks to describe the smooth, undulating features of the UV-vis spectra of [4Fe-4S]-maquettes. When the non-linear wave length scale is converted to the linear wave number scale, the features become broader and the fitting is even more challenging. Only 4 peaks as single features for d→d LFT, S2- → Fe LMCT, RS- → Fe LMCT, and LLCT does not give satisfactory fits due to the broad, envelop like features that cannot be described with a single Voigt line-shape. Any fits with less then 6 peaks (double for the LMCT features) did always result in high residuals or fits that were visibly not acceptable. Fits composed of more than 6 peaks cannot be justified due to the peak and shoulder resolution, since iterative peak fitting routinely collapsed into 6 peaks. The extra peaks formed small peaks under dominant ones. In the case of chemical speciation (presence of the rubredoxin-like feature) or high turbidity, additional peaks have to be introduced in order to maintain the validity of the 6-peak model.
The steps of our standard operating procedures for the UV-vis modeling are as follows:
obtain the CSV/DAT file from the spectrometer
since
wavelength is NOT a linear scale, fitting needs be done in
wavenumber scale; thus convert wavelength to wavenumber scale (
=
1E7/l(in nm) gives 1/cm wavenumber)
for PeakFit, which can only handle the old 2004-2007 Excel, save the spectrum to XLS format
open up the XLS file in PeakFit and prepare the data by cutting the spectrum down (apply to new) to 12,500 – 37,500 cm-1 range, which is was selected by considering the spectra of all [4Fe-4S]-maquettes in our sample library
select Fitting Method (I) that focuses on minimizing residuals
open up the 6-peaks_abs.scn template. The fit should already be fairly close, but the intensity likely be off depending on the success of the reconstitution process
although it is not absolutely needed, but iterative curve fitting procedure can be enhanced by manually adjusting intensity of peaks or even gently the peak positions to achieve a better initial values for fitting; this also avoids shifting the peaks that may redistribute the peak intensities (adjacent peaks steal intensity from each other)
start fit once and use graphical update to see what is happening; generally the fitting finishes quick - save the fit to SCN file.
Start the fit again by requesting additional adjustment in order to evaluate how robust is the fit, whether there are peaks that collapse or whether more peaks needed due to speciation. If there are large scale changes, visually detectable intensity redistributions, the model needs to be rejected and the fit obtained in Step 8 needs to be used
review fit:
print fit as a postscript PS file that can be readily converted to PDF, use white background w/colored traces of components
export fit results to XLS for the extended range 11,000 – 40,000 using 5 cm-1 fine steps
numerical results in TXT of the detailed summary of fitting
tabulate data in TAB: results of fit for every data point
print residuals as a postscript PS file, use white background w/colored traces of components
exit from review
save the fit to SCN file
save the truncated spectrum as .DAT file
open up the master spreadsheet UV-vis.xlsx and calculate the chromophore, [4Fe-4S]-maquette concentration relative to three standards from literature. Use the Integrated Area values instead of the Analytical Area, since the latter may reach outside of the 12,500 – 37,500 cm-1 wave number range used for fitting. As shown on the top of the UV-vis.xlsx sheet, the numerical integrated area were used to convert the intensity values to concentration for the literature-based, reference compounds [Fe4S4(SEt)4]2- in acetonitrile [DOI 10.1021/ja00350a028], [Fe4S4(ßME)4]2- in 2-(cyclohexylamino)ethanesulfonic acid buffered aqueous solution [DOI: /10.1073/pnas.94.13.6635], and buffered aqueous solution of the as prepared, oxidized PFL-AE radical SAM metalloenzyme from the Broderick laboratory at MSU [DOI: 10.1021/ja9711425].
Step 1
Step 2
Step 3
Step 4
Step 5
Step 6
Step 7
Step 8
Step 9
Step 10
Step 11
Step 12
Step 13
