Complexes with Tunable Intramolecular Ferrocene to TiIV Electronic

Feb 16, 2016 - Department of Chemistry, Furman University, Greenville, South ... Department of Physics, Marmara University Göztepe Kampus, 34772 Ista...
1 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/IC

Complexes with Tunable Intramolecular Ferrocene to TiIV Electronic Transitions: Models for Solid State FeII to TiIV Charge Transfer Michael D. Turlington,†,¶ Jared A. Pienkos,†,¶ Elizabeth S. Carlton,† Karlee N. Wroblewski,† Alexis R. Myers,† Carl O. Trindle,‡ Zikri Altun,§ Jeffrey J. Rack,∥ and Paul S. Wagenknecht*,† †

Department Department § Department ∥ Department ‡

of of of of

Chemistry, Furman University, Greenville, South Carolina 29613, United States Chemistry, University of Virginia, Charlottesville, Virginia 22904, United States Physics, Marmara University Göztepe Kampus, 34772 Istanbul Turkey Chemistry and Chemical Biology, University of New Mexico, Albuquerque, New Mexico 87131, United States

S Supporting Information *

ABSTRACT: Iron(II)-to-titanium(IV) metal-to-metal-charge transfer (MMCT) is important in the photosensitization of TiO2 by ferrocyanide, charge transfer in solid-state metal-oxide photocatalysts, and has been invoked to explain the blue color of sapphire, blue kyanite, and some lunar material. Herein, a series of complexes with alkynyl linkages between ferrocene (Fc) and TiIV has been prepared and characterized by UV−vis spectroscopy and electrochemistry. Complexes with two ferrocene substituents include Cp2Ti(C2Fc)2, Cp*2Ti(C2Fc)2, and Cp2Ti(C4Fc)2. Complexes with a single ferrocene utilize a titanocene with a trimethylsilyl derivatized Cp ring, TMSCp, and comprise the complexes TMSCp2Ti(C2Fc)(C2R), where R = C6H5, pC6H4CF3, and CF3. The complexes are compared to Cp2Ti(C2Ph)2, which lacks the second metal. Cyclic voltammetry for all complexes reveals a reversible TiIV/III reduction wave and an FeII/III oxidation that is irreversible for all complexes except TMSCp2Ti(C2Fc)(C2CF3). All of the complexes with both Fc and Ti show an intense absorption (4000 M−1cm−1 < ε < 8000 M−1cm−1) between 540 and 630 nm that is absent in complexes lacking a ferrocene donor. The energy of the absorption tracks with the difference between the TiIV/III and FeIII/II reduction potentials, shifting to lower energy as the difference in potentials decreases. Reorganization energies, λ, have been determined using band shape analysis (2600 cm−1 < λ < 5300 cm−1) and are in the range observed for other donor−acceptor complexes that have a ferrocene donor. Marcus−Hush-type analysis of the electrochemical and spectroscopic data are consistent with the assignment of the low-energy absorption as a MMCT band. TD-DFT analysis also supports this assignment. Solvatochromism is apparent for the MMCT band of all complexes, there being a bathochromic shift upon increasing polarizability of the solvent. The magnitude of the shift is dependent on both the electron density at TiIV and the identity of the linker between the titanocene and the Fc. Complexes with a MMCT are photochemically stable, whereas Cp2Ti(C2Ph)2 rapidly decomposes upon photolysis.

■

photocatalysis.5−8 In particular, several investigations of MMCT excited states in solid state Ti−O−M architectures involving TiIV acceptors oxo-bridged to donors (M) such as FeII, CrIII, and MnII have recently appeared,6 leading to interest in molecular examples of such Ti−O−M structures.7 Of additional relevance is that ferrocyanide photosensitizes TiO2, with the photosensitization being ascribed to an FeII to TiIV charge transfer.8 Furthermore, such an FeII to TiIV MMCT has been implicated as being responsible for the color observed in minerals such as blue sapphire9 (a corundum-based gemstone) and blue kyanite9e,10 (an aluminosilicate-based gemstone). FeII to TiIV charge transfer is also reported to contribute to the relative blueness of the lunar continuum in

INTRODUCTION Charge-transfer (CT) excited states where charge is separated over a long distance are interesting from a fundamental standpoint and for applications involving solar energy conversion and photocatalysis.1−8 For example, polypyridyl complexes of ruthenium(II) have been exploited in dyesensitized solar cells (DSSCs). Here, the metal-to-ligand CT (MLCT) excited state serves to inject an electron into the conduction band of the TiO2 semiconductor.2 Recently, much effort has been devoted toward studying CT excited states in completely organic systems where a donor group is bridged to an acceptor through a π-conjugated bridge (termed donor-πbridge-acceptor systems, D−π−A). Such systems are promising candidates to replace the expensive ruthenium-based dyes.3 Metal-to-metal CT (MMCT) excited states have also been of general fundamental interest4 and for applications involving © XXXX American Chemical Society

Received: November 9, 2015

A

DOI: 10.1021/acs.inorgchem.5b02587 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry telescopic reflectance measurements.11 Consistent with these assignments, the blue color of anhydrous FeTi(SO4)3 has also been attributed to an FeII to TiIV MMCT.12 Evaluation of spectroscopic data indicates that in solution this species exists as FeTiO(SO4)32− and EPR intensity data suggests that the FeII/TiIV state predominates.13 However, recent DFT theoretical evaluations on blue sapphire suggest that in the corundum lattice, the ground state is FeIII/TiIII and a MMCT results in FeII/TiIV.14 In spite of the possible technological significance of the FeII to TiIV MMCT and the questions surrounding its importance in mineralogy, very few examples exist of molecular complexes displaying such an optical transition (Figure 1).15−17 Our recent interest in electron transfer through alkynyl bridged organometallic frameworks18 led us to investigate the Cp2Ti(C2nFc)2 architecture.

by 1H NMR after precipitation from THF and hexanes, elemental analysis revealed that LiCl was present. Sonication of the products in water did not remove LiCl, suggesting Li+ coordination to the alkynyl-complex, similar to the tweezer complexes described by Lang et al.17 Attempted purification on untreated silica gel resulted in decomposition of the titanocene. For example, chromatographing Cp2Ti(C2Fc)2 on silica yielded 1,4-bis(ferrocenyl)butadiyne (orange band) as the major product. These compounds also decomposed on neutral alumina. Elution through base-treated (triethylamine) silica gel prevented the aforementioned decomposition and also resulted in removal of the LiCl impurity. Synthesis of the monosubstituted complexes, TMSCp2Ti(C2Fc)(C2R), proceeds through a common intermediate, TMS Cp2Ti(C2Fc)Cl. Lang et al. demonstrated that substituted Cp rings are necessary to prevent disproportionation of Cp2Ti(C 2R)Cl to Cp2TiCl2 and Cp2 Ti(C 2R)2 .19 Their synthetic procedure for TMSCp2Ti(C2Ph)Cl involved slow addition of a solution containing 1 eq of lithiated ethynylbenzene to TMSCp2TiCl2 in Et2O. We have found that this reaction can also be controlled by maintaining a temperature of −15 °C, thus obviating the need for slow addition. TMS Cp2Ti(C2Fc)Cl decomposes even on base-treated silica gel, so optimal elemental analysis results were not obtained despite clean 1H NMR spectra. However, the asymmetric titanocene products, TMSCp2Ti(C2Fc)(C2R), obtained from this starting material exhibited excellent purity following elution through base treated silica gel and recrystallization from hexanes. In the solid state, the complexes are air stable at room temperature for days to weeks, but slowly decompose over a period of months if not stored at low temperature. In roomtemperature THF, CH2Cl2, or toluene solution that has neither been dried nor degassed, most of the complexes are thermally and photochemically stable and thus can be handled in solution over periods of minutes to hours without protection from room light. Only Cp*2Ti(C2Fc)2 in CH2Cl2 solution is thermally unstable, undergoing complete decomposition within 4 h. Moreover, all of the bimetallic complexes studied (with the exception of Cp*2Ti(C2Fc)2 in CH2Cl2) are stable to direct photolysis in dried degassed THF, CH2Cl2, or toluene. In contrast, the monometallic complex, Cp2Ti(C2Ph)2, is unstable with respect to photolysis (vide inf ra). Finally, all of the complexes are subject to acid hydrolysis. Because of this sensitivity, all CDCl3 used for spectroscopy was passed through activated alumina prior to sample preparation. Electrochemistry. The cyclic voltammograms (vs FcH+/0) of the bimetallic Cp2Ti(C2nFc)2 complexes (Figure 3) show

Figure 1. Existing complexes displaying an FeII to TiIV MMCT where Fc = ferrocenyl; n = 0 and X = H,15 n = 2 and X = Me3Si,16 n = 1 and X = Me3Si.17

Such systems were studied by Wakatsuki (n = 2)16 for ground-state investigations into long-range electron-transfer and by Lang (n = 1)17 as molecular tweezers for cations. For both complexes, an intense absorption band in the red region of the visible spectrum is mentioned only briefly, it being ascribed to an FeII to TiIV MMCT band based on its intensity.15,17b Because such an excited state might be exploited for applications involving long-range electron-transfer, we report here an in-depth chemical and photophysical characterization of a series of FeII/TiIV complexes (Figure 2). The data herein clearly support the presence of an FeII to TiIV charge transfer transition and represent the first significant investigation into such a transition in molecular species.

■

RESULTS AND DISCUSSION Synthesis. The bisalkynylferrocenyl complexes, Cp2Ti(C2Fc)2, Cp2Ti(C4Fc)2, and Cp*2Ti(C2Fc)2 (Figure 2) were prepared following a modified procedure for the trimethylsilyl substituted analogues, TMSCp2Ti(C2nFc)2.16,17 In all cases, addition of the appropriate titanocene dichloride to an Et2O solution containing 2 equiv of the lithiated alkyne resulted in a deeply colored precipitate. Although the product appeared pure

Figure 2. General synthetic route for D−π−A complexes with a ferrocene donor and titanocene acceptor. B

DOI: 10.1021/acs.inorgchem.5b02587 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

There is also a cathodic shift of the 1e− TiIV/III reduction of Cp2Ti(C2Fc)2 vs Cp2Ti(C4Fc)2. This can be explained by the relative proximity of the electron-rich ferrocene to TiIV. Not surprisingly, replacement of the two Cp ligands on titanocene with Cp* (more electron-rich) results in a significant cathodic shift of the TiIV/III reduction and a milder shift of the initial Fc0/+ oxidation (Figure 3, bottom). The second oxidation wave remains unshifted, consistent with assignment of this wave to oxidation of the resulting FcC4Fc product. The cyclic voltammograms for all complexes (including the monoferrocenyl complexes discussed below) in THF are qualitatively similar (Supporting Information). In an attempt to suppress the elimination reaction following 2e− oxidation, complexes with a single ethynylferrocene, TMS Cp2Ti(C2Fc)(C2R), were investigated. Cyclic voltammetry demonstrates that decomposition was suppressed only for Cp2Ti(C2Fc)(C2CF3) where now a one-electron ferrocenyl oxidation wave is observed with the same peak current as the TiIV/III reduction wave (Figure 4, top). For R = C6H5 and p-

Figure 3. Cyclic voltammograms of complexes with two alkynylferrocenyl ligands appended to the titanocene. Conditions: [Ti] ∼ 1 mM in CH2Cl2 (0.1 M TBAP), glassy carbon working electrode, Ag/Ag+ reference electrode. The x-axis reports voltages vs ferrocene as determined by running a voltammogram of FcH before and after data collection. Voltage sweeps are initiated in the cathodic direction.

features representative of both the chemically reversible oneelectron TiIV/III reduction and of the chemically irreversible 2e− oxidation of the two ethynylferrocenyl ligands. For the closely related TMSCp2Ti(C2Fc)217a and TMSCp2Ti(C4Fc)216a complexes, the irreversible nature of this 2e− oxidation has been ascribed to subsequent rapid heterolytic Ti−C bond cleavage resulting in elimination of Fc(C2n)2Fc (eq 1). This hypothesis was confirmed by isolating the Fc(C2n)2Fc

product following 2e− oxidation.16a,17a During this cleavage, 1e− from each Ti−C bond serves to reduce Fc+ back to Fc, while the remaining e− is involved in formation of the new C−C bond. In the voltammogram of Cp2Ti(C2Fc)2, the reversible oxidation of the FcC4Fc product appears as two nearly superimposed 1e− waves slightly anodic of the first 2e− Fc oxidation (confirmed by comparison with an authentic sample of FcC4Fc, Supporting Information). The fact that the first irreversible 2e− wave is unsplit indicates a lack of electronic communication between the two ferrocenes via the C2TiIVC2 linkage, whereas the 90 mV separation of oxidation peaks for FcC4Fc (the second oxidation wave) indicates electronic communication through the organic alkynyl bridge.17a,20 This separation has been confirmed by differential pulse voltammetry (Supporting Information). Finally, increasing scan rates to 1000 mV/s did not result in observation of a return wave for the first 2e− oxidation step. In contrast to the ethynylferrocene complex, the voltammogram of the butadiynylferrocene complex, Cp2Ti(C4Fc)2, (Figure 3, middle) reveals that the irreversible 2e− oxidation occurs at nearly the same potential as the oxidation of the product FcC8Fc.16a The fact that the initial 2e− oxidation of the ethynylferrocene complex, Cp2Ti(C2Fc)2, is more cathodic than that of Cp2Ti(C4Fc)2 suggests that the TiIV-alkynyl bond is electron-rich and its proximity facilitates Fc oxidation. This implies ionic character of the TiIV-alkynyl bond, with significant negative charge density remaining on the terminal alkynyl carbon. This is consistent with a slight upfield shift of the Fc resonances in the 1H NMR of Cp2Ti(C2Fc)2 vs Cp2Ti(C4Fc)2.

Figure 4. Cyclic voltammograms for complexes with a single ethynylferrocene ligand on the TiIV, TMSCp2Ti(C2Fc)(C2R). Experimental conditions are identical to Figure 3.

C6H4CF3, the Fc oxidation remains chemically irreversible. Note that the second oxidation wave is not split, indicating that the Fc-containing product does not contain two Fc moieties in communication with one another. This suggests 1-aryl-4ferrocenylbutadiyne as the main product. Indeed, oxidation of TMS Cp2Ti(C2Fc)(C2Ph) with one equivalent of AgPF6 resulted in FcC4Ph as the main isolated organic product, with trace amounts of FcC4Fc also forming (indicating the possibility of a minor bimolecular elimination pathway). FcC4Ph was identified by mass spectrometry and by comparison of the 1H NMR spectrum with an independently prepared sample.21 Such a product trajectory has precedent in that one e− oxidation of cis(FcC2)PhPt(dppe) involves Fc oxidation followed by elimination of FcC2Ph.22 The fact that the Fc oxidation wave is chemically reversible for R = CF3 suggests that the reversibility is impacted by the electron density at TiIV. The presence of the electronwithdrawing trifluoropropynyl ligand decreases the electron density at TiIV, and might increase the degree of ligand to metal σ- and π- donation from the ethynylferrocene ligand, strengthening that bond. In further support of this hypothesis, the TiIV/III reduction wave of the synthetic intermediate, C

DOI: 10.1021/acs.inorgchem.5b02587 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry TMS

Cp2Ti(C2Fc)Cl, (Supporting Information) is anodically shifted to nearly the same extent as the complex with R = CF3, and also shows a chemically reversible Fc oxidation. However, it would appear that simply decreasing the e− density at Ti is not sufficient to suppress the elimination reaction given that such an elimination still occurs after 2e− ferrocenyl oxidation for the bis-butadiynylferrocene complex, Cp2Ti(C4Fc)2. UV−vis Spectroscopy. Cp2Ti(C2Ph)2 is bright orange in appearance with no absorbance features in the red region of the spectrum. Replacement of both phenylethynyl ligands with alkynylferrocenes results in deeply colored compounds, some appearing almost black in the solid state. The UV−vis spectra in CH2Cl2 (Figure 5) clearly show a low energy (LE) absorbance feature for the bimetallic complexes (centered near 580 nm) that is absent for Cp2Ti(C2Ph)2.

Figure 6. UV−vis absorption spectra of Cp*2Ti(C2Fc)2 and TMS Cp2Ti(C2Fc)(C2R) complexes in CH2Cl2. The LE absorption band shifts to higher energy with electron-rich ligands on Ti (Cp*) and to lower energy with electron-poor ligands on Ti (C2R = C2CF3). Line color approximates observed complex color.

between the Fc0/+ oxidation and the TiIV/III reduction. Due to the chemically irreversible nature of the Fc oxidation, its thermodynamic potential can only be estimated for most complexes herein (Table 1). Regardless, with the exception of Cp2Ti(C4Fc)2 (2nd entry in Table 1), there is a clear increase in the energy of the LE band as the difference between the TiIV/III and Fc+/0 redox potentials (ΔE1/2 Table 1) increases. Furthermore, the increase in the LE band energy upon switching solvent from CH2Cl2 to THF is mirrored by the increase in ΔE1/2 values between these two solvents (Table 1), thus providing additional support for the MMCT assignment. Explaining the anomalous behavior of Cp2Ti(C4Fc)2 requires consideration of the potential well diagram (Figure 7) showing the relationship between the energy of the optical MMCT transition, Eop, the energy of the thermal transition, ΔG°, and the reorganization energy, λ. Here, λ is the excess energy required (over and above ΔG°) to reach the excited state configuration with the same molecular and solvation geometry as the ground state, thus ensuring a vertical transition. This energy can be estimated from the full-width at half-maximum, Δν1/2, of the LE band (eq 2).4a,b,28

Figure 5. UV−vis absorption spectra of Cp2Ti(C2R)2 complexes in CH2Cl2. Probable assignments given. Line color approximates observed complex color.

As mentioned above, this LE band has been ascribed to an FeII to TiIV MMCT. Another possibility is that this LE band is a ferrocene d−d band that borrows intensity from a higherenergy CT band. Such a model was originally suggested for the LE absorption in a ferrocene-π-bridge-acceptor complex, cis-1ferrocenyl-2-(4-nitrophenyl)ethylene, studied as a second-order nonlinear optical chromophore.23 Later studies suggested that metal to nitrophenyl CT was more consistent with the experimental data.24 A more recent TDDFT study suggests that the LE band for this complex has both d−d and CT character.25 Thus, it seems important to experimentally interrogate the FeII to TiIV MMCT hypothesis for the bimetallic complexes herein. If the LE band is due to a MMCT, then only a single ferrocene should be necessary. Indeed, the TMSCp2Ti(C2Fc)(C2R) complexes show the same LE absorption feature observed for the bis-alkynylferrocene complexes, albeit with about half the molar absorptivity (Figure 6). This observation is consistent with the theoretical treatment by Parker and Crosby for two degenerate acceptors and a single donor (A−D−A).26 The model demonstrates that for a centrosymmetric (linear) system, one expects twice the intensity for the A−D−A charge transfer than for the nondegenerate D−A system. For nonlinear A−D−A complexes, the model predicts two transitions whose intensities sum to twice that of the D−A system. However, often only a single transition is observed, likely due to the magnitude of the matrix element coupling the two states.27 If the LE band has significant MMCT character, then its energy should be directly related to the potential difference

λ = (Δv1/2)2 /2310 cm−1

(2)

The values thus calculated for the reorganization energy (Table 2) fall into or near the range of values in the literature for D−π−A complexes in CH2Cl2 involving a Fc donor with either organic or inorganic acceptors (3800 to 7000 cm−1).29,30 The resulting values can be used to calculate ΔG° (eq 3). ΔG° = Eop − λ

(3)

These results are compared with ΔE1/2 (Table 2). Whereas ΔE1/2 is a reasonable estimate for ΔG°, and is often used as such, it fails to take into account the modified electrostatic interactions in the excited state.29,31 The difference between ΔE1/2 and ΔG°, a Coulombic energy correction work term, w, should decrease with distance between the donor and acceptor, and thus, it is not surprising that the smallest calculated value of w occurs for the butadiynyl complex, Cp2Ti(C4Fc)2. These values for w appear relatively solvent independent and are in a very similar range to those calculated for complexes with an ethylene bridge between ferrocene or permethylferrocene donors and a polychlorotriphenylmethyl radical acceptor (0.04−0.2).29 As expected, the trend of energy of the LE band D

DOI: 10.1021/acs.inorgchem.5b02587 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry Table 1. Comparison of Redox Potentials with Wavelength of the CT Bands E1/2 (TiIV/III)a

a

ΔE1/2

E1/2 (Fc+/0)a

Cp to TiIV (nm)

MMCT (nm)

complex

CH2Cl2

THF

CH2Cl2

THF

CH2Cl2

THF

CH2Cl2

THF

CH2Cl2

THF

Cp*2Ti(C2Fc)2 Cp2Ti(C4Fc)2 TMS Cp2Ti(C2Fc)(C2Ph) Cp2Ti(C2Fc)2 TMS Cp2Ti(C2Fc)(C2PhCF3) TMS Cp2Ti(C2Fc)(C2CF3)

−2.28 −1.45 −1.72 −1.75 −1.67 −1.54

−2.13 −1.50 −1.76 −1.82 −1.69 −1.55

−0.22 0.12b −0.02b −0.08b 0.00b 0.05

−0.08 0.14b 0.04b −0.06b 0.04b 0.05

2.06 1.57 1.70 1.67 1.67 1.59

2.05 1.64 1.80 1.76 1.73 1.60

544 581 583 584 593 630

542 563 573 571 586 623

372 441 420 408 418 425

373 439 417 408 416 425

b

b

Volts vs FcH+/0. bIrreversible wave. E1/2 estimated as the half-wave potential.

Computational Modeling. To further interrogate the CT assignments discussed above, DFT and TD-DFT33 calculations have been performed on the parent bimetallic complex, Cp2Ti(C2Fc)2, the more electron-rich Cp*2Ti(C2Fc)2, and Cp2Ti(C2Ph)2. Molecular structures were obtained by geometry optimization with the ωB97XD/def2-TZV functional34 and basis35 combination, which in our experience, reproduces bond lengths within 30 pm. Our choice of density functional and basis for TD-DFT spectroscopic modeling is guided by benchmarks that suggest either PBE036 or B3PW9137 in a Pople basis can represent the spectra of ferrocenyl systems well.25,38 The medium is represented by a Tomasi polarizable continuum39 assigned the macroscopic dielectric constant of CH2Cl2. All geometry optimization and spectroscopic calculations include medium effects. Some of the frontier orbitals for Cp2Ti(C2Fc)2 are depicted in Figure 8 and Supporting Information. The HOMO has a dominant iron d-orbital contribution with significant contribution also from the ethynyl linkage. The HOMO−1, HOMO−2, and HOMO−3 are predominantly iron d-orbital based and nearly degenerate (within 0.15 eV), consistent with MOs on isolated ferrocene. The LUMO is dominated by the titanium dorbitals, with little amplitude on the titanocene Cp rings, significant participation by the ethynyl linkage, and minor amplitude on the iron d-orbitals. The LUMO+1 and LUMO+2 are largely titanocene based. Frontier orbitals for the more electron-rich Cp*2Ti(C2Fc)2 are qualitatively similar (Supporting Information) with the LUMO lying 0.51 eV higher in energy than that for Cp2Ti(C2Fc)2, consistent with the replacement of Cp with the more electron-rich Cp*. The vertical transition energies and oscillator strengths predicted by TD-DFT for Cp2Ti(C2Fc)2, (Figure 9 and Supporting Information) are in reasonable agreement with the experimental spectrum but slightly underestimate the CT transition energies owing to poor long-range behavior of the functional.40 In agreement with experimental results, the calculated spectrum of Cp2Ti(C2Fc)2 shows a low-energy

Figure 7. Potential well diagram showing relationship between observed electrochemical and optical parameters, assuming a MMCT model.

with ΔG° rather than ΔE1/2 is more consistent, with Cp 2Ti(C 4Fc) 2 no longer being an outlier. The slight discrepancy of Cp2Ti(C2Fc)2 with respect to this trend can readily be explained by its relatively low value for λ. It is worth reiterating that the spectroscopically determined values for ΔG° (eq 3) are slightly smaller than the electrochemically determined ΔE1/2 values, but the difference corresponds to an expected Coulombic energy correction work term that is consistent with those in related systems.29 This correspondence of the electrochemically determined energy difference between the FeII/TiIV ground state and FeIII/TiIII excited state with the spectroscopically measured parameters offers further evidence that the observed LE band has significant MMCT character. Finally, some caution should be exercised in interpreting these results given that the E1/2 values for the Fc oxidation were estimated from the irreversible waves in all cases except for TMSCp2Ti(C2Fc)(C2CF3). Lastly, in agreement with previous reports of titanocenes,32 the absorption feature centered between 370 and 440 nm is assigned to the Cp to TiIV LMCT. Consistent with this assignment, a general decrease in the energy of this feature accompanies anodic shifts of the TiIV/III reduction (Table 1).

Table 2. Spectroscopic Parameters and Calculated Coulombic Energy Work Term, w λ (cm−1)b

Δν1/2a

ΔG° (cm−1)c

ΔG° (V)

complex

CH2Cl2

THF

CH2Cl2

THF

CH2Cl2

THF

CH2Cl2

THF

Cp*2Ti(C2Fc)2 Cp2Ti(C4Fc)2 TMS Cp2Ti(C2Fc)(C2Ph) Cp2Ti(C2Fc)2 TMS Cp2Ti(C2Fc)(C2PhCF3) TMS Cp2Ti(C2Fc)(C2CF3)

2540 3320 3300 3050 3400 3450

2540 3220 3260 3070 3300 3470

2790 4770 4710 4030 5000 5150

2600 4490 4600 4080 4710 5210

15600 12400 12400 13100 11900 10800

15850 13270 12790 13400 12350 10950

1.93 1.54 1.54 1.62 1.48 1.34

1.97 1.65 1.59 1.66 1.53 1.36

ΔE1/2 (w)d CH2Cl2 2.06 1.57 1.70 1.67 1.67 1.59

(0.13) (0.03) (0.16) (0.05) (0.19) (0.25)

THF 2.05 (0.08) 1.64 (−0.01)e 1.80 (0.21) 1.76 (0.10) 1.73 (0.20) 1.60 (0.24)

Estimated from a Gaussian fit of the low-energy side of the MMCT. bDetermined using eq 2 cDetermined using eq 3. dValue of w determined from ΔE1/2 − ΔG° (V). eValues for w cannot be less than zero. The negative value here results from experimental uncertainty.

a

E

DOI: 10.1021/acs.inorgchem.5b02587 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

the more electron-rich Cp*2Ti(C2Fc)2, again consistent with experimental results. The assignment of this LE band is most germane to this study. TD-DFT calculations suggest that this transition cannot be represented as a simple excitation from an origin MO to a destination MO. In this case, natural transition orbitals (NTOs)41 (Figure 10) can provide a clearer picture of the

Figure 10. Dominant destination (left) and origin (right) NTOs for the LE transition of Cp2Ti(C2Fc)2. The occupation value represents the contribution of the NTOs shown.

charge transfer. The origin NTOs (including two minor components, Supporting Information) are entirely located on the ferrocene fragments with major population on the Fe d set and substantial admixture of the Cp rings’ pi set. The destination NTO with the largest contribution to the LE absorption band has a large presence on the Ti fragment (predominantly of Ti d-orbital character) but has considerable population on the ferrocenes as well. The destination NTO with the second largest contribution is entirely located on the ferrocene. This is also the case for the two minor components (Supporting Information). We can infer that the LE band is largely FeII to TiIV CT (in agreement with the experimental results above), but with an important contribution of local excitation on the ferrocenes. This is consistent with the aforementioned cis-1-ferrocenyl-2-(4-nitrophenyl)ethylene, where TDDFT calculations suggest that the LE band for this complex has both d−d and CT character.25 NTO occupation numbers indicate the relative amounts of charge involved in each origin and destination NTO and thus sum to 1.00. They are not an indication of the relative contribution of each NTO to the overall intensity of the given transition. In the case of the composite LE band, although it contains significant d−d character, the MMCT character is likely the dominant contributor to the intensity owing to the relative oscillator strengths of CT transitions vs d−d transitions. NTO analysis of the bands computed to lie at 450 and 485 nm for Cp2Ti(C2Ph)2 indicate that these absorptions are each mixtures of two LMCT single excitations, phenyl to Ti and Cp

Figure 8. Frontier MOs for Cp2Ti(C2Fc)2. Additional MOs appear in the Supporting Information.

Figure 9. Predicted vertical transition energies, oscillator strengths, and UV−vis spectra for Cp2Ti(C2Ph)2 (top) and Cp2Ti(C2Fc)2 (bottom). The UV−vis spectrum is obtained by assuming a Gaussian width of 1000 cm−1 for each transition.

(LE) band not present for Cp2Ti(C2Ph)2 (Figure 9). This LE band blue-shifts by approximately 0.2 eV (620 to 560 nm) for F

DOI: 10.1021/acs.inorgchem.5b02587 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

with R2 values ≤0.25 (Supporting Information). Though the fit to π* was very poor (R2 = 0.05), we attempted to further refine that model using Kamlet and Taft’s multivariable model (eq 4)50

to Ti (Supporting Information). The ethynyl bridge is involved in each transition. Likewise, NTO analysis for the intense band at 485 nm for Cp2Ti(C2Fc)2 demonstrates that the dominant destination NTO (occupation = 0.65) is predominantly Ti dorbital in character and that the dominant origin state is Cp (on the Ti) ligand in character but with some contribution from ethynyl bridge and Fe d-orbitals. This again suggests predominant Cp to Ti LMCT character for this absorption band. The minor destination and origin states are localized on the ferrocenes. The model choice B3PW91/6311G(d,p) recommended by Barlow25 has produced reasonable transition frequencies and qualitatively useful relative intensities for the lower energy transitions in the Cp2Ti(C2Fc)2, Cp*2Ti(C2Fc)2, and Cp2Ti(C2Ph)2 systems. We explored the ωB97XD and CAM-B3LYP functionals with the def2-TZV basis as well. Despite the greater sophistication of these functionals, which contain range corrections and dispersion energies, their results were not even qualitatively informative. Solvatochromism. Solvatochromism has been reported for the CT bands in many Fc−π−A complexes. For complexes where the acceptor is cationic, negative solvatochromism (a shift to higher energy with increased solvent polarity) is typically observed,24,30,42−44 whereas for complexes with neutral acceptor moieties, positive solvatochromism is typically observed.24,29,42,45−47 It is worth noting that some of these conclusions are based on just two or three solvents. Analyses of extensive solvent sets with such Fc−π−A complexes sometimes results in irregular fits to any known solvent parameters.43,47 For Cp2Ti(C2Fc)2, the MMCT band is sensitive to solvent, apparently shifting to lower energy with increasing solvent molar mass. This trend led us to investigate less common but higher molar mass solvents, 1,2-dibromobenzene and 1,2diiodobenzene. The observed trend of a lowered energy for the MMCT with increased solvent molar mass continued (Table 3). Higher polarity solvents such as alcohols cannot be included due to poor solubility of the complexes. In contrast, correlations with solvent scales such as the solvent dielectric constant, Kosower’s Z,48 Dimroth and Reichardt’s ET(30),49 Kamlet’s π*,50 Gutmann’s AN and DN,51 and Swain’s acity and basity scales52 were very poor,

E(cm−1) = E° + s(π * + dδ) + aα + bβ

where π* combines both the polarity and polarizability of the solvent, α describes the hydrogen bond donor acidity of the solvent, β describes the hydrogen bond acceptor basicity, and δ is a polarizability correction term (0 for all nonhalogenated aliphatic solvents, 0.50 for all polyhalogenated aliphatics, and 1.00 for all aromatic solvents).50 The coefficients s, d, a, and b, determined from multiple linear regression, describe the susceptibility of the transition energy to each of the solvent parameters. Statistically, increasing the number of fit parameters will always increase the quality of the fit. Thus, we separately tested the addition of each parameter to π* to interrogate which variable had the greatest impact on the goodness of fit (Supporting Information). Whereas π* (R2 = 0.05) and the addition of α (R2 = 0.08) do very little to explain the trend in solvatochromism, improvements in the correlation associated with the addition of β (R2 = 0.51), or δ (R2 = 0.64) are statistically significant, with the polarizability correction term δ having a greater impact. Thus, the aforementioned correlation of MMCT transition energy with molar mass is likely due to the known correlation of polarizability with molar mass (due to the increased number of e− in higher molar mass materials). The above analysis suggests investigating the correlation of the MMCT energy with solvent polarizability using Catalán’s SP scale.53 Clearly the correlation between the MMCT energy and this scale is also poor (Table 3, Figure 11), but the R2 value

Table 3. Solvatochromism for Cp2Ti(C2Fc)2 and Cp2Ti(C4Fc)2 Cp2Ti(C2Fc)2 solvent o-C6H4I2 o-C6H4Br2 CCl4 CHCl3 C6H5NO2 toluene CH2Cl2 pyridine DMF THF acetone Δa (cm−1)

MW (g/mol)

SP

329.9 235.9 153.8 120.5 120.1 92 85 79.1 73.1 72 58

0.768 0.783 0.891 0.782 0.761 0.842 0.759 0.714 0.651

LMCT (nm)

MMCT (nm)

LMCT (nm)

598 592 588 586 584 583 584 581 573 572 567 910

411 405 410 410

602 588 584 585 584 578 585 577 566 563 557 1340

451 443 444 444 441 442 441 442 437 439 436 760

410 408 409 407 407 406 300

Figure 11. Energy of the MMCT transition as a function of Catalán’s SP scale. The halocarbon solvents, CCl4, CHCl3, and CH2Cl2, are indicated by red triangles.

Cp2Ti(C4Fc)2

MMCT (nm)

(4)

(0.46) was the highest of all the single parameter models tested (other than molar mass, where R2 = 0.71). Closer analysis of the plotted data indicates that the halocarbon solvents (CCl4, CHCl3, and CH2Cl2: red triangles) are largely responsible for the poor correlation. Though we have no current explanation for why this solvent class might not fit this model, their exclusion increases R2 to 0.86. This again points to a model where solvent polarizability dominates in the solvatochromism, suggesting that the excited state is more polarizable than the ground state.53a Such a bathochromic shift with increasing solvent polarizability has also been observed for another D−π−A system with a Fc donor, namely, 4-(ferrocen-1-yl)benzylidene-malonitrile, albeit with only three solvents.54

Energy difference (cm−1) between absorption maxima in acetone and C6H4I2 a

G

DOI: 10.1021/acs.inorgchem.5b02587 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry Table 4. Solvatochromism for

TMS

Cp2Ti(C2Fc)(C2R) and Cp*2Ti(C2Fc)2 Complexes

TMS

Cp2Ti(C2Fc)(C2PhCF3)

a

TMS

Cp2Ti(C2Fc)(C2CF3)

TMS

Cp2Ti(C2Fc)(C2Ph)

Cp*2Ti(C2Fc)2

solvent

MW (g/mol)

MMCT (nm)

LMCT (nm)

MMCT (nm)

LMCT (nm)

MMCT (nm)

LMCT (nm)

MMCT (nm)

LMCT (nm)

C6H4I2 CCl4 CH2Cl2 THF acetone Δb (cm−1)

329.9 153.8 85 72 58

612 602 593 586 580 900

423 418 418 416 416 400

650 638 629 619 611 980

431 424 424 425 423 440

602 586 581 575 570 930

423 419 420 418 417 340

552 548 543 542 538 470

-a 374 372 373 370 -

Due to a limited solvent window, this peak was not observed. bEnergy difference (cm−1) between absorption maxima in acetone and C6H4I2.

The magnitude of the solvatochromism increases with increased distance between the donor and acceptor. Chiefly, between the two solvents that generate the largest solvatochromic shift, acetone and o-diiodobenzene, there is a shift of 910 cm−1 for Cp2Ti(C2Fc)2. This increases to 1340 cm−1 for Cp2Ti(C4Fc)2 where an additional ethynyl linkage has been inserted between the donor and acceptor. For the three monosubstituted complexes, TMSCp2Ti(C2Fc)(C2R), the magnitude of the solvatochromic shift for the MMCT is very similar to that for Cp2Ti(C2Fc)2, ranging from 900 to 980 cm−1 (Table 4). Note that these four complexes all have TiIV/III reduction potentials within 0.21 V of one another. In contrast, Cp*2Ti(C2Fc)2 has a TiIV/III reduction potential more than 0.5 V further negative and has a significantly smaller magnitude for the solvatochromic shift (470 cm−1). Clearly the charge distribution in the ground state has a significant impact on the magnitude of the solvatochromism and may also play a role in the larger magnitude of solvatochromism observed for Cp2Ti(C4Fc)2. The magnitude of the solvatochromic shift for the LMCT is smaller than that of the MMCT in every case, likely reflecting the shorter distance of charge separation in this excited state. Additional TDDFT calculations are being pursued to characterize these charge distributions and their relationship to polarizability and solvent influences. Photochemistry. The bimetallic Fe−Ti systems are stable to photolysis with broad band irradiation from a tungsten lamp, contrasting the related Cp2Ti(C2Ph)2 complex, which rapidly decomposes in THF, benzene, or toluene solution upon photolysis. The chief organic product of photolysis is III (Figure 12), as demonstrated by comparison of the 1H NMR of the product with an independently prepared sample.55 This product likely forms from reductive elimination of two phenylethynyl ligands to give I, followed by reduction of I with the resulting “titanocene”, II. Reductive elimination is precedented in that photolysis of Cp2TiPh2 results in biphenyl and a TiII product.56 The so-called “titanocene” resulting from such a reaction has been demonstrated to have a dihydride structure, II,57 and has been shown to act as a reducing agent toward ethylene and CO2.58 The hydrides in II result from the coupling of the two Cp rings to the fulvalene structure. To gain further evidence that the reducing hydrogen equivalents are coming from the starting material, the photolysis was performed in d6-benzene and d8-THF. This did not result in deuterium incorporation into the product. Also, the substituted butadiyne, I, was not observed, suggesting either a very efficient reduction of I by II, or perhaps even a concerted process to product III. One explanation of the relative photolytic stability of the bimetallic complexes vs Cp2Ti(C2Ph)2 is that the MMCT state is a bound state, whereas one or more of the LMCT states are dissociative states.59 UV−vis spectroscopy and TD-DFT

Figure 12. Scheme for the photochemical decomposition of Cp2Ti(C2Ph)2.

analysis suggest an energy ordering where Cp− to TiIV and PhCC− to TiIV LMCT states are higher in energy than the MMCT states. In the bimetallic complexes, the LMCT state(s) may undergo rapid internal conversion to the lower energy MMCT state before dissociation. Even the complex with one ethynylferrocene and one ethynylbenzene ligand is quite photostable, suggesting that the PhCC− to TiIV LMCT state rapidly funnels into the MMCT prior to reaction out of the LMCT. As a further test of this, irradiation of dilute absorbance matched (A419 = 1.0) solutions of Cp2Ti(C2Ph)2, Cp2Ti(C2Fc)2, and TMSCp2Ti(C2Fc)(C2Ph) in THF was conducted at 419 nm, a wavelength that excites into the LMCT state(s). In the time required for Cp2Ti(C2Ph)2 to undergo complete photobleaching, the bimetallic complexes showed