UV-Vis Spectroscopy, Electrochemical and DFT Study of Tris(β-diketonato)iron(III) Complexes with Application in DSSC: Role of Aromatic Thienyl Groups

A series of tris(β-diketonato)iron(III) complexes, with the β-diketonato ligand bearing different substituent groups, have been synthesized and characterized by Fourier transform infrared (FT-IR), ultraviolet-visible (UV-Vis) and mass spectroscopic methods. The maximum band UV-Vis absorption wavelengths of the tris(β-diketonato)iron(III) complexes were in the range of 270–380 nm. The complexes have very good solubility in various solvents such as chloroform, dichloromethane, ethyl acetate, tetrahydrofurane, dimethylsulphoxide and dimethylformamide. After the syntheses and characterization processes, spectroscopic and electrochemical properties of these tris(β-diketonato)iron(III) complexes were investigated. A density functional theory (DFT) study related to the spectroscopic and electrochemical properties of the tris(β-diketonato)iron(III) complexes was used to investigate the possible application of these complexes as dye sensitizers or redox mediators in dye-sensitized solar cells.

Iron(II) and iron(III) complexes have strong absorption bands in the near ultraviolet region. Iron porphyrins, for example, have a typical Soret band at ca. 400 nm [10], whereas tris(β-diketonato)iron(III) complexes have a strong absorption band at ca. 300 nm [11,12]. The intensity and position of the bands change as the iron complexes are reduced or oxidized [10,[13][14][15]. Tris(β-diketonato)iron(III) complexes undergo a single one-electron ironrelated electrochemically and chemically reversible reduction reaction. The electrochemical data of a series of tris(β-diketonato)iron(III) complexes were previously reported [16]. To compliment the electrochemical study of the tris(β-diketonato)iron(III) complexes, a spectroscopic and spectroelectrochemical (SEC) study was conducted on the complexes. SEC combines electrochemistry and spectroscopy to show the influence of the redox chemistry of the tris(β-diketonato)iron(III) complexes on their UV-Vis absorption spectra. The oxidation state of the iron complexes is changed electrochemically, while the spectra of the product of the redox transformation are then simultaneously monitored in situ by UV-Vis spectroscopy.
In this study, we thus present a spectroscopic, electrochemical and computational chemistry study of the series of tris(β-diketonato)iron(III) complexes shown in Scheme 1. The influence of the type of substituent groups (methyl, trifluoromethyl, furyl, phenyl and thienyl) on the observed spectroscopic and electrochemical behaviour is evaluated.

Results and Discussion
The molecular structure of complexes 2-6 and 8, containing β-diketonato ligands with two different substituent groups, can be a fac-or a mer-isomer [17,18]. The difference in the density functional theory (DFT) calculated electronic energies (E), zero-point corrected electronic energies (ZEE) and free energies (G) of the fac-and mer-isomers of a complex are generally very small ( Table 1), implying that both isomers will exist in an experimental sample of a complex. The experimentally measured UV-Vis and redox potential are thus expected to be the result of a mixture of fac-and mer-isomers.

Results and Discussion
The molecular structure of complexes 2-6 and 8, containing β-diketonato ligands with two different substituent groups, can be a facor a mer-isomer [17,18]. The difference in the density functional theory (DFT) calculated electronic energies (E), zero-point corrected electronic energies (ZEE) and free energies (G) of the facand mer-isomers of a complex are generally very small (Table 1), implying that both isomers will exist in an experimental sample of a complex. The experimentally measured UV-Vis and redox potential are thus expected to be the result of a mixture of facand mer-isomers.  3 ] complexes 1-10. The density functional theory (DFT) calculated electronic energy difference (∆E in eV), zero-point-corrected electronic energies difference (∆ZEE in eV) and free energy difference (∆G in eV) between the facand mer-isomers are also given. The experimental spectra of complexes 1-9 have a strong absorbance peak in the range of 270-376 nm, with one or more lower energy weak absorbance bands in the 400-600 nm region (see Figure 1). Comparing the UV-Vis of complexes 1-9, it is clear that λ A,max,exp becomes red-shifted as more aromatic groups are attached to the β-diketonato ligands in the [Fe(β-diketonato) 3 ] complexes 1-9. The introduction of one Ph or two Ph side groups per β-diketonato ligand successively causes a redshift of~30 nm: Fe(acac) 3  The absorbance maxima of the complex with six thienyl groups is the most redshifted. Complexes 1-9 can thus be grouped into three groups according to the influence of aromatic groups on the wavelength of the maximum absorbance λA,max,exp, in the 250-400 nm region: UV-Vis group 1: λA,max,exp ≈ 270 nm, complexes 1 and 2, containing only CF3 or CH3 (no aromatic) substituent groups on the β-diketonato ligands.

TDDFT
To get insight in the type of charge transfer (CT) bands observed in the experimental UV-Vis spectra of 1-10, a time-dependent density functional theory (TDDFT) study was performed. To validate the TDDFT method, different functionals and basis sets were used to determine the artificial spectra, excitation energies and oscillator strengths associated with the different absorbance bands for [Fe(acac)3], complex 1. All excitations above 200 nm were determined. The wavelength (λA,max) corresponding to the maximum intensity (oscillator strength f) and the corresponding f data are provided in Table S1. The difference between the DFT-calculated maximum wavelength (λA,max(calc) in nm) and the experimental value λA,max(experimental) = 270 nm of [Fe(acac)3], complex 1, using a selection of functionals and basis sets that have been previously proven to give good agreement between theory and experiment [19], is shown in Figure 2. The B3LYP and M06 functionals, both using the CEP-121G basis set, gave the best agreement with the experiment, namely within 9.7 and 15.1 nm of the experimental wavelengths, respectively. Adding Grimme's dispersion D3 to B3LYP using the CEP-121G basis set resulted in  Similarly, the introduction of one or two thienyl side groups per β-diketonato ligand successively causes a redshift of more than 40 nm: The absorbance maxima of the complex with six thienyl groups is the most red-shifted. Complexes 1-9 can thus be grouped into three groups according to the influence of aromatic groups on the wavelength of the maximum absorbance λ A,max,exp , in the 250-400 nm region: UV-Vis group 1: λ A,max,exp ≈ 270 nm, complexes 1 and 2, containing only CF 3 or CH 3 (no aromatic) substituent groups on the β-diketonato ligands.

TDDFT
To get insight in the type of charge transfer (CT) bands observed in the experimental UV-Vis spectra of 1-10, a time-dependent density functional theory (TDDFT) study was performed. To validate the TDDFT method, different functionals and basis sets were used to determine the artificial spectra, excitation energies and oscillator strengths associated with the different absorbance bands for [Fe(acac) 3 ], complex 1. All excitations above 200 nm were determined. The wavelength (λ A,max ) corresponding to the maximum intensity (oscillator strength f) and the corresponding f data are provided in Table S1. The difference between the DFT-calculated maximum wavelength (λ A,max (calc) in nm) and the experimental value λ A,max (experimental) = 270 nm of [Fe(acac) 3 ], complex 1, using a selection of functionals and basis sets that have been previously proven to give good agreement between theory and experiment [19], is shown in Figure 2. The B3LYP and M06 functionals, both using the CEP-121G basis set, gave the best agreement with the experiment, namely within 9.7 and 15.1 nm of the experimental wavelengths, respectively. Adding Grimme's dispersion D3 to B3LYP using the CEP-121G basis set resulted in λ A,max (calc) = 257.7 nm for 1, with a larger deviation from the experiment (12.3 nm) than using B3LYP without D3.
The results for λ A,max (calculated), using the B3LYP/CEP-121G and M06/CEP-121G methods for complexes 1-10, are summarized in Table 1. Results that illustrate the influence of a long-range corrected functional, using CAM-B3LYP/CEP-121G, are also included in Table 1. B3LYP reproduced the λ A,max (experimental) for 1-9 with an average deviation (AD) and a mean absolute deviation (MAD) from experiment of 7.8 and 3.8 Å, while M06 was slightly less accurate with AD = 11.0 Å and MAD = 4.3 Å. λ A,max values determined with CAM-B3LYP/CEP-121G were blue-shifted compared to the experimental values, presenting much larger values of AD = 18.4 Å and MAD = 11.6 Å. Furthermore, it is observed that, in most cases, λ A,max (calc) of the facand mer-isomers of a complex are very similar. The experimental and B3LYP-calculated spectra of 1-9 are compared in Figure 3. Both experimental and calculated spectra show three main bands (indicated as bands 1-3 in Figure 3), namely the low energy band in the visible region, the strong absorbance peak and another higher energy (smaller wavelength) band in the UV region. λA,max(calc) = 257.7 nm for 1, with a larger deviation from the experiment (12.3 nm) than using B3LYP without D3. The results for λA,max(calculated), using the B3LYP/CEP-121G and M06/CEP-121G methods for complexes 1-10, are summarized in Table 1. Results that illustrate the influence of a long-range corrected functional, using CAM-B3LYP/CEP-121G, are also included in Table 1. B3LYP reproduced the λA,max(experimental) for 1-9 with an average deviation (AD) and a mean absolute deviation (MAD) from experiment of 7.8 and 3.8 Å, while M06 was slightly less accurate with AD = 11.0 Å and MAD = 4.3 Å. λA,max values determined with CAM-B3LYP/CEP-121G were blue-shifted compared to the experimental values, presenting much larger values of AD = 18.4 Å and MAD = 11.6 Å. Furthermore, it is observed that, in most cases, λA,max(calc) of the fac-and mer-isomers of a complex are very similar. The experimental and B3LYP-calculated spectra of 1-9 are compared in Figure 3. Both experimental and calculated spectra show three main bands (indicated as bands 1-3 in Figure 3), namely the low energy band in the visible region, the strong absorbance peak and another higher energy (smaller wavelength) band in the UV region.  The results for λA,max(calculated), using the B3LYP/CEP-121G and M06/CEP-121 methods for complexes 1-10, are summarized in Table 1. Results that illustrate th influence of a long-range corrected functional, using CAM-B3LYP/CEP-121G, are als included in Table 1. B3LYP reproduced the λA,max(experimental) for 1-9 with an averag deviation (AD) and a mean absolute deviation (MAD) from experiment of 7.8 and 3.8 Å while M06 was slightly less accurate with AD = 11.0 Å and MAD = 4.3 Å. λA,max valu determined with CAM-B3LYP/CEP-121G were blue-shifted compared to th experimental values, presenting much larger values of AD = 18.4 Å and MAD = 11.6 Å Furthermore, it is observed that, in most cases, λA,max(calc) of the fac-and mer-isomers of complex are very similar. The experimental and B3LYP-calculated spectra of 1-9 a compared in Figure 3. Both experimental and calculated spectra show three main band (indicated as bands 1-3 in Figure 3), namely the low energy band in the visible region, th strong absorbance peak and another higher energy (smaller wavelength) band in the U region.   To get insight into the type of charge transfer (CT) bands observed in the experimental UV-Vis spectra of 1-10, the MOs involved in the transitions were evaluated. In Table 2 and  Table S2, selected calculated excitation energy (E), wavelength (λ), oscillator strengths (f) and assignments of main transitions involved in the excitation of complexes 1, 7 and 9 are given with the MOs involved in the transitions, visualized in Figures 4-6. For [Fe(acac) 3 ] (1), [Fe(dbm) 3 ] (7) and [Fe(dtm) 3 ] (9), the maximum absorbance transition (band 2, λ A,max ) involves excitation from ligand-based occupied MOs to mainly metal-based unoccupied MOs, thus mainly ligand-to-metal charge transfer (LMCT). Transitions at energies higher than that of the maximum absorbance transition (band 3) also have some ligand-to-ligand charge transfer (LLCT) characteristics. Similarly, band 1 and 2 transitions of complexes 2-8 and 10 are mainly LMCT, and band 3 are LLCT. For further insight into the intramolecular charge transfer process during excitation of the maximum absorbance band, electronic density difference (EDD) plots to show the direction of the charge transfer between the ground and excited state of the maximum absorbance peak were determined and visualized in Figure 7. The region of electron density depletion (indicated with red) for all complexes is localized on the β-diketonato ligands. The region where the electron density increases (green color) is mainly on Fe(III), but also partially on the β-diketonato ligands. The EDD plots thus confirm that the maximum absorbance excitations are mainly LMCT. For the UV-Vis group 2 and 3 complexes that contain aromatic groups, the region of electron density decrease also occurs on the aromatic groups, all of pi character. Due to the symmetrical nature of the EDD plots, excitation would result in a serious intramolecular electron recombination, in agreement with the short excited state lifetimes calculated for the excited states (see Table 3   For further insight into the intramolecular charge transfer process during excitation of the maximum absorbance band, electronic density difference (EDD) plots to show the direction of the charge transfer between the ground and excited state of the maximum absorbance peak were determined and visualized in Figure 7. The region of electron density depletion (indicated with red) for all complexes is localized on the β-diketonato ligands. The region where the electron density increases (green color) is mainly on Fe(III), but also partially on the β-diketonato ligands. The EDD plots thus confirm that the maximum absorbance excitations are mainly LMCT. For the UV-Vis group 2 and 3 For further insight into the intramolecular charge transfer process during excitation of the maximum absorbance band, electronic density difference (EDD) plots to show the direction of the charge transfer between the ground and excited state of the maximum absorbance peak were determined and visualized in Figure 7. The region of electron density depletion (indicated with red) for all complexes is localized on the β-diketonato ligands. The region where the electron density increases (green color) is mainly on Fe(III), but also partially on the β-diketonato ligands. The EDD plots thus confirm that the maximum absorbance excitations are mainly LMCT. For the UV-Vis group 2 and 3   agreement with the short excited state lifetimes calculated for the excited states (see Table  3 and the discussion in Section 2.1.3).
One of the requirements for dyes to effective is that the unoccupied MOs (MOs to where the charge is transferred upon excitation) need to lie above the conduction band (CB) of the semiconductor (to provide a driving force for dye injection into the semiconductor), and the occupied MOs need to lie below the redox potential of the redox electrolyte (to provide a driving force for dye regeneration [26]). Although the CF3containing complexes 2, 4-6 and 10 have low lying LUMOs, the unoccupied MOs involved in the maximum absorbance excitation do lie sufficiently high enough to provide a driving force for dye injection: ΔGinject > 0.2 eV required for a DSSC to be effective [27,28] (see Table  3). The calculated positive ΔGinject and ΔGregenerate values obtained for the maximum absorbance excitation for 1-10 indicate, according to definition, that electron injection and

Application as Dye to DSSC
For a complex to be used as a dye, the absorption spectra of the dye should have strong absorption peaks in the ultraviolet-visible (UV-Vis, ca. 350-750 nm) regions of the solar spectrum [20][21][22]. The experimental spectra of complexes 1-9 have a strong absorbance peak between 270-376 nm, with a lower energy LMCT in the 400-600 nm region. To further determine if the [Fe(β-diketonato) 3 ] complexes 1-10 could qualify as dyes in DSSC, theoretically calculated properties for potential dyes such as light harvesting efficiency (LHE), excited state lifetime (τ in ns), HOMO energies (E HOMO ), LUMO energies (E LUMO ), ∆G inject (eV) and ∆G regenerate (eV) values are determined and presented in Table 3 for complexes 1-10, for a DSSC with iodide/triiodide (I − /I 3 − ) as a redox mediator (redox potential = −4.8 eV vs. vacuum, or 0.3 eV vs. NHE [23]) and anatase (TiO 2 with E CB = −4.0 eV vs. vacuum or −0.5 eV vs. NHE [24,25]) as a semiconductor.
One of the requirements for dyes to effective is that the unoccupied MOs (MOs to where the charge is transferred upon excitation) need to lie above the conduction band (CB) of the semiconductor (to provide a driving force for dye injection into the semiconductor), and the occupied MOs need to lie below the redox potential of the redox electrolyte (to provide a driving force for dye regeneration [26]). Although the CF 3 -containing complexes 2, 4-6 and 10 have low lying LUMOs, the unoccupied MOs involved in the maximum absorbance excitation do lie sufficiently high enough to provide a driving force for dye injection: ∆G inject > 0.2 eV required for a DSSC to be effective [27,28] (see Table 3). The calculated positive ∆G inject and ∆G regenerate values obtained for the maximum absorbance excitation for 1-10 indicate, according to definition, that electron injection and dye regeneration are spontaneous in a TiO 2 -(I − /I 3 − ) DSSC. The calculated LHE (the fraction of light intensity absorbed by the dye at the specific wavelength) of 1-10 varies between 0.4 and 0.9. The LHE are the highest for complexes 5-9, with one or two aromatic substituent groups on each β-diketonato ligand. The increasing π-conjugation between the β-diketonato backbone into the aromatic substituent of the donor subunit seems to enhance the LHE. The calculated excited state lifetimes (τ) of 1-10 are low (2-5 ns), and may not be sufficiently long enough to retard the charge recombination process to enhance the efficiency of the DSSCs [29]. Reported excited state lifetime values for known dyes are higher, e.g., 27 ns for CYC-B11 [30] and 11.7 ns for YD2-o-C8 [31], experimentally known efficient Ru [30] and Zn-porphyrin [32] based dye sensitizers, respectively.

Electrochemistry
The reduction of [Fe(β-diketonato) 3 ] complexes can be considered as the acceptance of an electron into the LUMO of the complex. Since the LUMOs of 1-10 are iron-based (see Figures 5 and 6 for complexes 7 and 9 as examples), the reduction is iron-based, and thus is a Fe(III/II) redox process.

SEC
The values of Fe(III/II) redox couple of complexes 1-9 have previously been reported by us [16]. To explore the UV-Vis spectral changes associated with the reduction of [Fe(βdiketonato) 3 ] complexes 1-9, an in situ spectroelectrochemical (SEC) study was conducted on the complexes. The spectral changes during the reduction of [Fe(β-diketonato) 3 ] complexes 1-9 are shown in Figure 8        In Figure 9a, the spectral changes during the reduction of [Fe(ba) 3 ] (3) are shown. This complex has a strong band at 298 nm and a smaller band at 250 nm. Upon reduction of the complex, the 298 and 250 nm bands, respectively, decrease and increase in intensity. Two isobestic points are visible at 230 and 265 nm (between 200 and 250 nm). In Figure 9b, the spectral changes during the reduction of [Fe(tfba) 3 ] (4) are shown. This complex has a strong band at 304 nm and a shoulder at 259 nm. Upon reduction of the complex, the 304 and 259 nm bands, respectively, decrease and increase in intensity. Two isobestic points are visible at 230 nm and 275 nm (between 200 and 300 nm). In Figure 9c, the spectral changes during the reduction of [Fe(tffu) 3 ] (5) are shown. This complex has a strong band at 333 nm, two smaller bands at 230 and 385 nm and a shoulder at~520 nm. Upon reduction of the complex, the 333, 385 and~520 nm bands disappear completely. A new strong band at 295 nm and a shoulder at 350 nm appear. The 230 nm band increases with a redshift of 5 nm. Two isobestic points are visible at 230 and 300 nm. In Figure 9d, the spectral changes during the reduction of [Fe(tta) 3 ] (6) are shown. This complex has a strong band at 333 nm, a smaller band at 270 nm and a shoulder at 385 nm. Upon reduction of the complex, the 333 nm band decreases in intensity to form a band at 320 nm with a shoulder at 350 nm. The shoulder at 385 nm also decreases in intensity. The band at 270 nm slightly increases in intensity and a new shoulder at 230 nm appears. Two isobestic points are visible at 225 and 300 nm.
In Figure 10a, the spectral changes during the reduction of [Fe(dbm) 3 ] (7) are shown. This complex has a strong band at 336 nm, a smaller band at 251 nm and a shoulder at 415 nm. Upon reduction of the complex, the 336 nm band decreases in intensity and the shoulder at 415 nm disappears completely. The 251 nm band increases in intensity with a slight blueshift of 3 nm. One isobestic point is visible at 270 nm. In Figure 10b, the spectral changes during the reduction of [Fe(bth) 3 ] (8) are shown. This complex has a strong band at 361 nm, a smaller band at 245 nm and a shoulder at 275 nm. Upon reduction of the complex, the 361 nm band decreases in intensity. The 245 nm band initially slightly decreases with a redshift of 10 nm, and then strongly increases at 255 nm. The shoulder at 275 nm increases with a redshift of 25 nm. One isobestic point is visible at~300 nm. In Figure 10c, the spectral changes during the reduction of [Fe(dtm) 3 ] (9) are shown. This complex has a band at 260 nm and four shoulders at 295, 350, 375 and 445 nm. Upon reduction of the complex, the 350, 375 and 445 nm shoulders decrease in intensity and the 260 nm band with the 295 nm shoulder increases in intensity. One isobestic point is visible at 320 nm.
In comparing the spectral changes during the one-electron Fe(III)/Fe(II) reduction of complexes 1-9, the maximum absorbance peak of all the Fe(III) complexes decreases (indicated with red arrows in the Figures 8-10). A new maximum absorbance peak associated with Fe(II) forms at a lower wavelength (indicated with blue arrows in the Figures 8-10). For the thienyl-and furyl-containing complexes 5, 6, 8 and 9, two new absorbance peaks associated with Fe(II) form at the lower wavelength. The absorbance peak associated with Fe(II) that forms at the lower wavelength decreases in intensity for the UV-Vis group 1 complexes, while it increases in intensity for the UV-Vis group 2 and group 3 complexes. This peak might be related to the pi orbitals on the aromatic groups of complexes 3-9.
In Figure 11, the experimental and TDDFT UV-Vis of [Fe III (β-diketonato) 3 ] and [Fe II (βdiketonato) 3 ] − are compared for β-diketonato = acac (complex 1, without any aromatic group) and dbm (complex 9, containing aromatic groups). Both the experimental and calculated spectra of [Fe(acac) 3 ] show similar features (Figure 11a), namely: (i) the two strong absorbance peaks in the 200-300 nm region decrease in intensity and redshift upon reduction of Fe(III), and (ii) the lower energy, higher wavelength band above 400 nm disappears upon reduction of Fe(III). Only the redshift of the small experimental peak at ca. 350 nm could not be identified in the calculated spectra. For [Fe(dbm) 3 ], the calculated spectra reproduced the main features of the experimental spectra (Figure 11b), namely: (i) the strong absorbance peaks below 300 nm increase in intensity and blueshift upon reduction of Fe(III), (ii) the strong absorbance peaks at ca. 350 nm decrease in intensity and redshift upon reduction of Fe(III), and (iii) the lower energy, higher wavelength band above 400 nm disappears upon reduction of Fe(III). Figures 8-10). For the thienyl-and furyl-containing complexes 5, 6, 8 and 9, two new absorbance peaks associated with Fe(II) form at the lower wavelength. The absorbance peak associated with Fe(II) that forms at the lower wavelength decreases in intensity for the UV-Vis group 1 complexes, while it increases in intensity for the UV-Vis group 2 and group 3 complexes. This peak might be related to the pi orbitals on the aromatic groups of complexes 3-9. In Figure 11, the experimental and TDDFT UV-Vis of [Fe III (βdiketonato)3] and [Fe II (β-diketonato)3] − are compared for β-diketonato = acac (complex 1, without any aromatic group) and dbm (complex 9, containing aromatic groups). Both the experimental and calculated spectra of [Fe(acac)3] show similar features (Figure 11a), namely: (i) the two strong absorbance peaks in the 200-300 nm region decrease in intensity and redshift upon reduction of Fe(III), and (ii) the lower energy, higher wavelength band above 400 nm disappears upon reduction of Fe(III). Only the redshift of the small experimental peak at ca. 350 nm could not be identified in the calculated spectra. For [Fe(dbm)3], the calculated spectra reproduced the main features of the experimental spectra (Figure 11b), namely: (i) the strong absorbance peaks below 300 nm increase in intensity and blueshift upon reduction of Fe(III), (ii) the strong absorbance peaks at ca. 350 nm decrease in intensity and redshift upon reduction of Fe(III), and (iii) the lower energy, higher wavelength band above 400 nm disappears upon reduction of Fe(III). The main features of the spectral changes during the one-electron Fe(III)/Fe(II) reduction are thus: (i) The maximum absorbance peak is at a higher wavelength of all the Fe(III) complexes that decrease (indicated with red arrows in the  and; (ii) The new maximum absorbance peak is associated with Fe(II) that forms at a lower wavelength (indicated with blue arrows in the Figures 8-10).
The EDD plots between the ground and excited state of these two maximum absorbance peaks were determined for complex 1 (representative of non-aromatic-containing complexes) and complex 7 (representative of complexes containing aromatic groups), and illustrated in Figure 12. For [Fe II (acac) 3 ] − , the region of electron density depletion upon excitation (indicated with red) is on Fe(II) and the ligands. The HOMO of [Fe II (acac) 3 ] − is of Fe-d character (the Fe-d-based MO that accepted the electron upon reduction of Fe(III)), and this electron density depletes upon Fe(II) excitation, leading to the decrease of intensity of both absorbance maxima peaks (at 227.18 and 268.09 nm) compared to Fe(III) (Figure 11a).
For [Fe II (dbm) 3 ] − , however, the region of electron density depletion upon excitation (indicated with red in Figure 12) is mainly on the aromatic groups for the lower wavelength absorbance maxima (279.01 nm), and on the ligand (with a small fraction on Fe(II)) for the higher wavelength absorbance maxima (345.84 nm). Both peaks thus involve mainly intra ligand electron transfer, leading to an increase of the lower wavelength absorbance maxima (279.01 nm), and decrease of higher wavelength absorbance maxima peak (345.84 nm) (Figure 11b). The increase in intensity of the lower wavelength absorbance maxima of complexes 3-9, containing aromatic groups, are thus related to charge transfer from the aromatic groups upon excitation. Fe(II)) for the higher wavelength absorbance maxima (345.84 nm). Both peaks thus mainly intra ligand electron transfer, leading to an increase of the lower wav absorbance maxima (279.01 nm), and decrease of higher wavelength absorbance peak (345.84 nm) (Figure 11b). The increase in intensity of the lower wav absorbance maxima of complexes 3-9, containing aromatic groups, are thus re charge transfer from the aromatic groups upon excitation.
[ In DSSCs, the redox mediator needs to regenerate the oxidized dye. The wel (I3 − /I − ) redox couple has been used as redox couple in DSSC for many years [34]. A redox couple that is considered may have a redox potential up to 0.5 V more posit that of (I3 − /I − ) [34]. The Fe(III/II) redox couple [16] of complexes 1-10 are compare redox potential of the (I3 − /I − ) redox couple (0.3 V vs. NHE [23]) in Figure 13. Comp 10 can be grouped into three groups according to the experimentally measure value of the Fe(III/II) redox couple: Redox group 1: E 0′ < −0.18 V vs. NHE, complexes 1, 3 and 7-9, containing substituent groups on the β-diketonato ligands.

Application as Redox Mediator in DSSC
In DSSCs, the redox mediator needs to regenerate the oxidized dye. The well-known (I 3 − /I − ) redox couple has been used as redox couple in DSSC for many years [34]. Any other redox couple that is considered may have a redox potential up to 0.5 V more positive than that of (I 3 − /I − ) [34]. The Fe(III/II) redox couple [16] of complexes 1-10 are compared to the redox potential of the (I 3 − /I − ) redox couple (0.3 V vs. NHE [23]) in Figure 13. Complexes 1-10 can be grouped into three groups according to the experimentally measured redox value of the Fe(III/II) redox couple: Redox group 1: E 0 < −0.18 V vs. NHE, complexes 1, 3 and 7-9, containing no CF 3 substituent groups on the β-diketonato ligands.
Considering a DSSC with the organic LEG4 dye [35,36] (redox potential = 1.07 V vs. NHE in CH 3 CN [37], excited state = −0.97 vs. NHE [38]) adsorbed onto a film of TiO 2 semiconductor, only the redox potentials of redox group 2 are suitable for use as a redox mediator. However, with another dye and semiconductor, more complexes may qualify as redox mediators in DSSC [39]. For example, [Fe(acac) 3 ] complex 1 showed promising results as a redox mediator in p-type DSSCs in conjunction with a perylene-thiophenetriphenylamine sensitizer and NaO as a semiconductor [2].
mediator. However, with another dye and semiconductor, more complexe as redox mediators in DSSC [39]. For example, [Fe(acac)3] complex 1 show results as a redox mediator in p-type DSSCs in conjunction with a perylen triphenylamine sensitizer and NaO as a semiconductor [2].  [37,38] in CH3CN) and the Fe(III/II) redox potentials of complexes 1-10 (black or pu potential of complexes 2, 4, 5 and 6, suitable for use as redox mediators in a DSSC TiO2 semiconductor, are indicated in purple. The redox potential of I − /I3 − (light brow comparative purposes.

General
Complexes 1-10 were synthesized and characterized as reported in the Characterization data agree with reported data and are provided in the s information. UV-Vis spectra were recorded on a Varian Cary 50 Conc ultra spectrophotometer.
Spectroelectrochemical measurements were performed on 0.003 mol of the complex dissolved in CH3CN as a solvent, containing 0.200 mol butylammonium hexafluorophoshate as the supporting electrolyte. transparent thin layer electrochemical (OTTLE) Omni cell system, fitted wi Omni windows, was filled with the solution. The OTTLE cell was connecte Conc ultra-violet/visible spectrophotometer, as well as a BAS100B e analyzer (linked to a personal computer utilizing the BAS100W Version Spectra were collected on the spectrophotometer every 5 min between 200min or till no spectral changes occurred) while scanning at a rate of 500 electrochemical analyzer from the resting potential of the iron(III) complex u the Fe(III/II) reduction potential of the complex. Spectral (absorbance vs. wa were collected, and were exported as csv data and imported into Microso for analysis. Spectral processing and visualization were also done with M Excel.

General
Complexes 1-10 were synthesized and characterized as reported in the literature [16]. Characterization data agree with reported data and are provided in the supplementary information. UV-Vis spectra were recorded on a Varian Cary 50 Conc ultra-violet/visible spectrophotometer.
Spectroelectrochemical measurements were performed on 0.003 mol dm −3 solutions of the complex dissolved in CH 3 CN as a solvent, containing 0.200 mol dm −3 tetra-nbutylammonium hexafluorophoshate as the supporting electrolyte. An optically transparent thin layer electrochemical (OTTLE) Omni cell system, fitted with NaCl liquid Omni windows, was filled with the solution. The OTTLE cell was connected to a Cary 50 Conc ultra-violet/visible spectrophotometer, as well as a BAS100B electrochemical analyzer (linked to a personal computer utilizing the BAS100W Version 2.3 software). Spectra were collected on the spectrophotometer every 5 min between 200-600 nm (for 90 min or till no spectral changes occurred) while scanning at a rate of 500 µV s −1 on the electrochemical analyzer from the resting potential of the iron(III) complex until 0.5 V after the Fe(III/II) reduction potential of the complex. Spectral (absorbance vs. wavelength) data were collected, and were exported as csv data and imported into Microsoft Office Excel for analysis. Spectral processing and visualization were also done with Microsoft Office Excel.