Article pubs.acs.org/JPCA
Experimental and Computational Exploration of Ground and Excited State Properties of Highly Strained Ruthenium Terpyridine Complexes Paul J. Vallett and Niels H. Damrauer* Department of Chemistry and Biochemistry, University of Colorado Boulder, Boulder, Colorado 80309, United States S Supporting Information *
ABSTRACT: Dissociative electron transfer reactions are prevalent in one-electron reduced aryl halides; however, calculations applied to charge-transfer excited states of metal complexes suggest that this reaction would be strongly endergonic unless attention is paid to specific structural details. In this current study, we explore the effect of introducing intramolecular strain into a series of halogenated ruthenium(II) polypyridyls. Parent [Ru(tpy) 2 ] 2+ (1) (tpy = 2,2′:6′,2″-terpyridine) is compared with two complexes, [Ru(6,6″Br2-tpy)(tpy)]2+ (2) and [Ru(6,6″-Br2-tpy)2]2+ (3) (6,6″-Br2-tpy = 6,6″-dibromo-tpy) that incorporate interligand van der Waals strain derived from the large halogen substituents. DFT calculations and the crystal structure of 3 show how this strain distorts the geometry of 3 as compared to 1. Time-dependent DFT calculations are used to explain the effect of this strain on electronic absorption spectra where, in particular, a transition observed in 3 is attenuated in 2 and absent in 1 and heralds interligand charge transfer mediated by the halogen substituent. Ultrafast transient absorption spectroscopy reveals coherent vibrational dynamics particularly in 3 but also in 2 that is interpreted as reflecting heavy-atom motion. Surprisingly, in spite of the additional strain, the excited-state lifetime of 3 is observed to be approximately a factor of 6 longer than 2. Constrained-DFT calculations show that while the excited behavior of 2 is similar to 1, the strain-induced geometric distortions in 3 cause a nesting of excited state triplet surfaces resulting in a longer excited state lifetime. This may afford the additional time needed to engage in photochemistry, and kinetic evidence is observed for the breaking of a C−Br bond in 3 and formation of a contact ion pair state.
■
INTRODUCTION Energy-wasting charge recombination is a limiting factor in the use of chromophoric materials in photochemical or photoredox transformations such as those needed in solar energy conversion schemes.1−3 Many photoactive systems ranging from natural photosynthetic constructs to artificial material assemblies within organic photovoltaics or within dye-sensitized solar cells expend significant redox potential as electron transfer (ET) driving force in order to rapidly separate charge carriers by large distances to avoid recombination.4−10 However, this necessarily reduces the efficiency of such systems.7,11 Our research has previously focused on utilizing ligand structure and nuclear motions to control the photophysics of materials with the goal of delaying the charge recombination process after ET without relying exclusively on expending driving force.12−14 Recently we have explored a different approach and computationally considered ruthenium polypyridyl systems designed to store energy via photoinduced dissociative ET (DET) inclusive of scission of a carbon−halogen bond.15 While photoinduced DET in small organic systems has been readily observed,16−18 we determined that in typical Ru(II) polypyridyl systems containing aryl-halogen bonds within the ligand framework, the presence of the charged metal center significantly modifies the reaction thermodynamics such that the desired DET product is © XXXX American Chemical Society
higher in energy than the metal-to-ligand charge-transfer (3MLCT) excited state from which the reaction would originate. After examining the reaction as a thermochemical cycle, we focused on reducing just one step, namely, the energy required to homolytically break the carbon−halogen bond. One promising and relatively straightforward way to accomplish this is to incorporate intramolecular strain into the metal complex structure that is released upon halogen bond scission thereby decreasing the net energy of the desired photoreaction. One way to introduce strain in metal−ligand complexes is via the attachment of bulky functional groups such as bromine substituents in the ligands themselves and in particular at the 6 and 6″ positions of a 2,2′:6′,2″-terpyridine (tpy). Previous density functional theory (DFT) results by our group15 as well as those discussed herein indicate that, when such ligands are incorporated in bis-terpyridyl complexes of Ru(II), the bromine substituents are in van der Waals contact with the opposing terpyridine ligand, causing interligand strain. The release of strain upon carbon−halogen bond scission causes the Received: April 29, 2013 Revised: June 26, 2013
A
dx.doi.org/10.1021/jp404248z | J. Phys. Chem. A XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry A
Article
then precipitated. The orange solid was collected on a frit and washed with excess water. The compound was purified on a silica gel column using an 8:1:1 CH3CN/H2O/saturated aqueous KNO3 solution. Yield 13% (13.2 mg). 1H NMR (500 MHz, CD3CN) δ 8.76 (d, J = 8.3 Hz, 2H), 8.66 (d, J = 8.1 Hz, 2H), 8.52−8.44 (m, 5H), 8.37 (t, J = 8.2 Hz, 1H), 7.98 (t, J = 7.7 Hz, 2H), 7.74 (t, J = 7.9 Hz, 2H), 7.53 (d, J = 5.5 Hz, 2H), 7.38 (d, J = 8.0 Hz, 2H), 7.29−7.23 (m, 2H). Accurate mass: as [A2+ PF6−] calcd. 870.9362 g/mol; measured 870.9396 g/mol. [Ru(6,6″-Br2-tpy)2](PF6)2 (3). First, 78.2 mg (0.2 mmol) of 6,6″-dibromo-2,2′:6′,2″-terpyridine was added to a flask containing 10 mL of ethylene glycol. Again, to completely dissolve the ligand the mixture was stirred constantly and heated to ∼200 C°. Then, 20.7 mg (0.1 mmol) of RuCl3 was added to the reaction mixture, and the heat was turned off. The reaction mixture was allowed to stir for an additional 20 min. Purification was accomplished using the same procedure as outlined above for 2. Yield 11% (12.9 mg). 1H NMR (500 MHz, CD3CN) δ 8.73−8.68 (m, 4H), 8.53 (dd, J = 8.0, 1.3 Hz, 4H), 8.45−8.40 (m, 2H), 7.80 (t, J = 8.0 Hz, 4H), 7.48 (dd, J = 8.0, 1.3 Hz, 4H). Accurate mass: as [A2+ PF6−] calcd. 1028.7185 g/mol; measured 1028.7201 g/mol. X-ray Diffraction. To grow crystals, 3 mL of a 0.01 M solution of 3 was dissolved in acetonitrile and placed in an uncovered 20 mL glass scintillation vial, which was then placed inside a 250 mL glass bottle that contained approximately 15 mL of diethyl ether. The bottle was sealed and kept at ∼10 °C for 72 h. Crystals of 3 obtained were cubic measuring approximately 450 μm along an edge. Single crystals were coated with Paratone-N oil and mounted under a cold stream of dinitrogen gas. Single crystal X-ray diffraction data were acquired on a Bruker Kappa APEX II CCD diffractometer with Mo Kα radiation (λ = 0.71073 Å) and a graphite monochromator. Initial lattice parameters were obtained from a least-squares analysis of more than 100 reflections; these parameters were later refined against all data. None of the crystals showed significant decay during data collection. Data were integrated and corrected for Lorentz and polarization effects using Bruker APEX2 software, and semiempirical absorption corrections were applied using SCALE with the aid of numerical face indexing.53 Space group assignments were based on systematic absences, E statistics, and successful refinement of the structures. Structures were solved by Direct Methods and were refined with the aid of successive Fourier difference maps against all data using the SHELXTL 6.14 software package.54 A solvent molecule was found in Fourier difference maps to be disordered. After numerous attempts to model the remaining disorder failed to improve agreement factors, SQUEEZE55 was used to remove the remaining disordered components. Thermal parameters for all nonhydrogen atoms were refined anisotropically. All hydrogen atoms were assigned to ideal positions and refined using a riding model with an isotropic thermal parameter 1.2 times that of the attached carbon atom (1.5 times for methyl hydrogens). All other geometric parameters can be found in the CIF file included with the Supporting Information. Crystal structure analysis and image generation was accomplished using the Mercury 3.0 (Cambridge Crystal Structure Centre) software package. Transient Absorption. Single wavelength kinetics were obtained using the following system: A 1 kHz pulse train centered at 800 nm (50 fs full width at half-maximum (fhwm))
calculated bond dissociation energy (BDE) to be reduced by 0.45 eV as compared to a related unstrained system.15 While ruthenium polypyridyl complexes are widely used platforms for exploration of photoinduced ET,19−22 energy transfer (EnT) arrays,19,23−30 and sensitizers for semiconductors,9,31−43 their ligand modification, when necessary, typically occurs at positions that are para to the chelating nitrogen atoms due to synthetic expediency and control over ET or EnT dyad architecture.44,45 Examples of functionalization at the 6 and 6′ positions of bipyridine or the 6 and 6″ positions of terpyridine in metal complexes are relatively scarce in the literature.46−48 Researchers have recently seen success exploiting a decrease in photostability that accompanies substitution of steric bulk in these ligand positions to create a potent light-triggered anticancer agent. However, in contrast to the bond-breaking reactions we are seeking, their photochemical reaction involves full ligand-loss processes.49 In this work, we examine the effect of strain and short interligand distances on structure, ground state electronic energy levels, linear light absorption, and excited state kinetics. In addition, density functional theory (DFT), time-dependent calculations (TD-DFT), and spin constrained DFT (C-DFT) techniques are used to aid in the explanation of ground state electronic effects and excited state dynamics. We also show evidence for the occurrence of the desired bond scission reaction.
■
EXPERIMENTAL AND THEORETICAL METHODS All reagents and materials from commercial sources were used as received. Solvents were purchased from Sigma-Aldrich Chemical Co. or Mallinckrodt Chemicals. All deuterated solvents were obtained from Cambridge Isotope Laboratories, Inc. Ruthenium(III) chloride hydrate (RuCl3) was obtained from Strem Chemicals, Inc. 2,2′:6′,2″-terpyridine (tpy), 6,6″dibromo-2,2′:6′,2″-terpyridine (6,6″-Br2-tpy), ammonium hexafluorophosphate, and potassium nitrate were purchased from Sigma-Aldrich Chemical Co. The 5,5″-dibromo-2,2′:6′,2″terpyridine (5,5″-Br2-tpy) was obtained from HetCat as an impure solid and subsequently recrystallized in hot methanol. The precursor complex Ru(tpy)Cl3 was prepared according to a previously published procedure50 and the parent complex [Ru(tpy)2](PF6)2 (1) was prepared by modifying a previously published procedure51 with triethylamine used as the reductant and AgCF3SO3 as a chloride scavenger. The [Ru(5,5″-Br2tpy)2](PF6)2 (5) complex was prepared according to a previously reported procedure52 followed by purification on a silica gel column (CH3CN/H2O/sat. aqueous KNO3 solution, gradient 8:1:1 to 3:3.5:1.7). All NMR spectra were recorded using a Bruker Avance-III 300 MHz or a Varian Inova 500 MHz spectrometer. Accurate Mass measurements were commissioned in house from the University of Colorado’s Central Analytical Laboratory. Experimental absorption spectra were measured with a Hewlett-Packard HP8452A diode array UV− vis spectrophotometer. [Ru(6,6″-Br2-tpy)(tpy)](PF6)2 (2). First, 39.1 mg (0.1 mmol) of 6,6″-dibromo-2,2′:6′,2″-terpyridine was added to a flask containing 10 mL of ethylene glycol. To completely dissolve the ligand, the mixture was stirred constantly and heated to ∼200 C°. Then, 44.0 mg (0.1 mmol) of Ru(tpy)Cl3 was added to the reaction mixture, and the heat was turned off. The reaction mixture was allowed to stir for an additional 20 min. To the crude reaction mixture, 10 mL of a saturated aqueous NH4PF6 solution was added and the desired product B
dx.doi.org/10.1021/jp404248z | J. Phys. Chem. A XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry A
Article
number of unpaired elections present on the metal center. The lowest 50 singlet transitions of the ground state were calculated using time-dependent (TD) DFT. The character and orbital contribution to each transition was determined using natural transition orbital (NTO)71−73 and attachment−detachment (AD) density analyses.74 These techniques have been previously shown to be useful in describing d6 transition metal complexes along with their electronic absorption transitions.15,75,76 All calculations were performed with the QChem 4.0.0.2 software package.77 Geometries and surfaces were visualized and analyzed using Molekel version 5.4.0.8.78 XYZ coordinates for optimized geometries and tabulated TD-DFT transitions are reported in the Supporting Information. Simulated electronic absorption spectra were generated from the TD-DFT data using a home-built LabView 11.0 program. This program places a Gaussian function centered at the calculated wavelength for each of the transitions. These Gaussians all have a fwhm (νfwhm) of 0.15 eV (1205 cm−1), and the maximum value of each Gaussian, the maximum absorption extinction coefficient (α max ), are scaled in proportion to the calculated oscillator strength ( f) according to the relationship shown in eq 1.79 All of the Gaussians are then summed to obtain the final simulated absorption spectrum. Transitions with oscillator strength less than 0.02 were excluded from the simulated spectrum. All transitions, including those with zero oscillator strength, are presented in the Supporting Information. Vibrations were visualized using the Jmol version 12.2.33 software package.80
was derived from a Quantronix Odin Ti:Sapphire multipass amplifier seeded by a K&M Ti:Sapphire oscillator. Approximately 200 μJ of the available 1 mJ was used to pump a homebuilt noncollinear optical parametric amplifier (NOPA)56,57 with a series of irises to ensure a high quality spatial mode. In this work, the NOPA was tuned to 510 nm with a fwhm bandwidth of 25 nm. The resultant visible pulse train was compressed with a pair of fused silica prisms and then passed through a pulse shaper, which consists of a home-built, all reflective zero dispersion compressor58 paired with a dual mask CRI 640 spatial light modulator (SLM). The output of the pulse shaping apparatus was mechanically chopped at 500 Hz, and focused onto the sample with a 125 mm lens (160 μm spot diameter). Autocorrelation was performed at the sample position, and the prisms were adjusted until the pulses were near transform limited. Pulses were further compressed to account for additional dispersion using the pulse shaping apparatus by maximizing the peak of the autocorrelation signal using a home-built covariance matrix adaptation evolutionary strategy optimization program on a restricted phase space composed of sixth order Chebyshev polynomials.59−61 Final pulses were sub 45 fs fwhm and attenuated to be 250 nJ/pulse. The probe beam was produced in a CaF2 window and focused with the same lens as the pump beam at an angle of ∼7°. Pump and probe beams were set to a relative polarization of magic angle (54.7°) by rotating the pump beam directly prior to the sample with an achromatic half-wave plate (Thorlabs; AHWP05M-600) to ensure that only population dynamics are being monitored. The probe was spectrally resolved in an Action 2300i monochromator. The detected signal was coupled to a Stanford Research SR250 boxcar integrator. The output was then detected at 500 Hz with a Stanford Research SR810 lockin amplifier. Transient absorption spectra were obtained on a similar system that has been recently described in detail.62 Plotting and analysis of the experimental data was accomplished using the commercially available software Igor Pro 6.20B02 (WaveMetrics) along with a home-built program for Fourier transform analysis and curve fitting (LabView 2010, National Instruments). The transient kinetics presented in this report were plotted in terms of negative change in normalized transmittance of the probe beam, −ΔT, while the transient spectra were plotted in terms of −ΔT/T. Both can be interpreted in the same way as a ΔA signal with positive features corresponding to a net transient absorption and negative signals corresponding to net transient bleach. DFT Calculations. All complexes were studied in the gasphase, i.e., without including explicit solvent molecules or solvent models, using the PBE063−65 functional. Optimized geometries were obtained without symmetry constraints. A brief discussion about how we chose this functional among a number of possibilities can be found in the Supporting Information. The 6-31g(d)66,67 basis set was used for all atoms except Ru for which the SDD effective core pseudopotential and associated basis set were used.68,69 Ground state singlet and excited state triplet geometries were confirmed as stable through frequency calculations that returned zero imaginary frequencies. Reported vibrational frequencies are scaled by 0.9519 to account for a known systematic underestimation by the PBE0 functional.70 Assignment of a triplet optimized geometry to either metal-centered (3MC) or 3 MLCT character was accomplished through examination of the Ru center atomic Mulliken spin population value of approximately 2 or 1, respectively, corresponding to the
αmax = 2.316 × 108
■
⎛ ln 2 ⎞ f L × ⎜2 ⎟ 2 π ⎠ Δνfwhm mol m ⎝
(1)
RESULTS AND DISCUSSION Ground State Structural Analysis. As mentioned in the Introduction, we have identified intramolecular van der Waals (VDW) strain as a structural handle useful for lowering a specific C−X bond dissociation energy (BDE). As this bondbreaking step is one of a larger multistep thermodynamic cycle, reducing the energetic cost of the BDE serves to decrease the energy required for the desired photoinduced DET reaction. We therefore embarked upon synthesis of compounds 2 and 3 as presented in Scheme 1 that used the 6,6″-Br2-tpy ligand to incorporate such strain through interligand steric interactions within bis-terpyridyl metal complexes. Scheme 1. Structures for Complexes Explored: RR′H (1), RBr R′H (2), and RR′Br (3)
In order to observe the effects of strain on the geometry of these complexes, the crystal structure of complex 3 was determined by XRD analysis and is compared to the published crystal structure of 181 in Figure 1. The unit cell of 3 contains two conformers, one highly distorted from the approximate D2d symmetry seen in 1 (shown in Figure 1, referred to as 3A) and C
dx.doi.org/10.1021/jp404248z | J. Phys. Chem. A XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry A
Article
DFT calculations (vide infra), and we believe it best represents the solvated geometry of 3. Conformer 3B likely arises as a result of the greater number of counterion close contacts present in the crystal structure. Selected geometric parameters for 3A are compared to 1 in Table 1. See Supporting Information for a comparison of 3A to 3B (Figure S1 and Table S2). It is clear from an examination of Figure 1 that 3A is highly distorted as compared to 1. The distortion primarily takes the form of a twist that moves the terminal pyridyl rings out of the plane of the central pyridyl ring, which is evident in the increased N1−C2−C1−N2 dihedral angle and deviation from 90° of the C2−N1−Ru−N2′ dihedral angle. In addition, the brominated ligands have a more opened bite that allows for an increase in the Br−N1′ distance as evidenced by the increased C4−C3−C5 angle along the backbone of the ligand. While the central Ru−N1 bond distance is slightly shortened by an average of 0.7%, the changes described above cause a lengthening of the Ru−N2 bond distances by 4.3% in 3A as compared to 1. Driving the distortion is the steric interaction between the 6 and 6″ bromine substituents and the central pyridyl unit of the opposing terpyridine ligand. These close intramolecular contacts are visualized in Figure 2 as red lines indicating
Figure 1. Crystal structure of 181 (inset) compared to the distorted structure observed for 3A as determined by XRD (top) and by gasphase DFT (bottom). Counterions and solvent molecules have been removed from crystal structures for clarity.
one less distorted (3B). Present in the crystal structure are a number of short intermolecular contacts between the two conformers and between the counterions and the conformers. Conformer 3B has 10 short contacts arising from proximal counterions, while 3A has only three. Conformer 3A is most similar to the only stable conformer determined from gas-phase
Figure 2. Crystal structure of 3A showing 40% probability ellipsoids with counterions and hydrogen atoms removed for clarity. Red dashed lines indicate short intramolecular contacts that are less than the sum of the VDW radii of the two atoms that are involved. The elongated bromine ellipsoids indicate larger disorder in the location of the bromines.
Table 1. Selected Average Distances and Angles for Experimental Crystal Structures and DFT Calculated Ground State Structures.a [Ru(tpy)2]2+ (1) parameter Bond Length (Å) Ru−N1 Ru−N2 H−N1′ Br−N1′ Br−C2′ Angle (deg) N1−Ru−N2 C4−C3−C5 Dihedral (deg) N1−C2−C1−N2 C2−N1−N1′−C2′ C2−N1−Ru−N2′
crystal
b,c
1.985 (0.009) 2.067 (0.004) 2.886 (0.111)
[Ru(6,6″-Br2-tpy)(tpy)]2+ (2)
[Ru(6,6″-Br2-tpy)2]2+ (3)
DFT
DFT-tpy
DFT-Br2-tpy
crystal Ac
DFT
1.992 2.086 2.801
1.992 2.089 2.806
2.004 2.179
1.976 (0.007) 2.157 (0.050)
2.002 2.169
3.124 3.357
3.076 (0.062) 3.317 (0.124)
3.140 3.381
79.20 (0.85) 84.26 (0.51)
78.86 84.45
78.89 84.58
78.24 89.11
80.64 (10.10) 89.03 (0.36)
78.36 88.75
1.47 (2.66) 88.23(4.70) 87.06 (4.31)
0.00 90.00 90.00
0.32 85.37 87.78
3.14
8.64 (10.42) 74.58 (4.69) 82.67 (5.13)
6.26 76.01 81.74
86.49
a c
Atom numbering corresponds to those given in Scheme 1; prime marks indicate the same atom position on the opposing ligand. bReference 81. Uncertainties are given in parentheses as ±2σ by determining the standard deviation of all similar measurements within the crystal structure. D
dx.doi.org/10.1021/jp404248z | J. Phys. Chem. A XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry A
Article
have nearly identical Ru−N lengths, C4−C3−C5 angles, and N1−C2−C1−N2 dihedrals indicating no strain as expected. The brominated terpyridine geometrical values match more closely to those of 3. Comparing the brominated terpyridine ligand to the nonbrominated terpyridine ligand on 2 reveals elongated bond lengths for Ru−N1 (2.004 Å vs 1.992 Å) and for Ru−N2 (2.179 Å vs 2.089 Å), a larger C4−C3−C5 ligand angle (89.11° vs 84.58°), and a larger N1−C2−C1−N2 dihedral (3.14° vs 0.32°). The C2−N1−N1′−C2′ dihedral that shows the relative angle between the ligands is 85.37, which lies between 90.00° and 76.01° for 1 and 3, respectively. These data as a whole indicate that the qualitative nature of the distortions due to the 6,6″-Br2-tpy ligand is the same for 2 as was seen in 3, but that the magnitude of the strain in the complex is less. DFT Ground State Electronic Structure. It appears from the intramolecular close contacts observed in 3 that the bromine atoms associated with 6,6″-Br2-tpy are in a position to affect the ground state electronic structure of complexes containing this ligand. Certainly changes in the π-system of the ligands or in the coordination geometry brought about by interligand VDW strain would as well. In this context, we have made efforts to separately determine the electronic effects of bromine substituents from the distortion and close contact effects. With this aim, it was useful to compare the strained complexes 2 and 3 to analogous complexes with 5,5″- and 4,4″dibromo-terpyridine ligands where interligand VDW strain is absent. The ground state geometries for [Ru(5,5″-Br2-tpy)(tpy)]2+ (4), [Ru(5,5″-Br2-tpy)2]2+ (5), [Ru(4,4″-Br2-tpy)(tpy)]2+ (6), and [Ru(4,4″-Br2-tpy)2]2+ (7) were also calculated using DFT. The geometries closely match that of 1 (D2d and C 2v symmetry for homo- and heteroletpic complexes, respectively) and no distortion effects were observed. Full structures are reported in the Supporting Information. Energies for the frontier orbitals HOMO−4 through LUMO +3 for complexes 1−5 are displayed in Figure 3. Data for complexes 6 and 7 are not displayed because the bromine substituents in the 4 and 4″ position exist on a node for the terpyridine π and π* orbitals. However, these substituents at
atoms that are closer than the sum of their VDW radii. Even with the distorted geometry, the average interligand Br−N1′ distance for 3 is 3.076 Å, which is 10% less than 3.40 Å, the sum of the bromine and sp2 hybridized nitrogen VDW radii.82,83 The bromine substituents are also in close contact with the two carbon atoms adjacent to the N1′ nitrogen with an average Br− C2′ distance of 3.317 Å. This is 8% shorter than the Br−C VDW contact distance of 3.62 Å that employs the aromatic carbon VDW radii.82,83 As a comparison, the analogous interligand H−N1′ distance in 1 is 2.886 Å, which is 11% larger than the sum of the hydrogen and nitrogen VDW radii (2.60 Å).82 While X-ray quality crystals of 2 could not be obtained, we suspect that these same short contacts observed for 3 exist and cause similar distortions. The geometry of 2 is explored in the DFT calculations presented below. DFT Ground State Geometry. DFT ground state structures for 1−3 were calculated, and selected geometrical values are compared to the relevant crystallographic data in Table 1. The optimized geometry for 1 as calculated by DFT closely matches the published crystal structure. Slight deviations in geometric parameters are expected due to intermolecular effects in the crystal that are not present in the gas-phase calculations. The central Ru−N1 distance is well reproduced with the PBE0 functional. This bond length is calculated to be only 0.3% larger than the crystallographic value. The calculated Ru−N2 bond lengths are overestimated relative to the crystal structure by only 0.9%, which represents a significant improvement relative to the typical 2−3% overestimation of Ru−N distances that result from B3LYP calculations of similar complexes.15,76,84 The calculated structure of 3 shows a remarkable reproduction of geometric distortions present in the crystal structure for 3A as can be observed in Figure 1. The central Ru−N1 distances are overestimated by 1.3%, moderately worse than the error for the same bond length in 1, while the Ru−N2 distances are better reproduced with only 0.6% overestimation. The same trends in geometrical values that are indicative of the interligand steric-induced strain in the crystal structure are also observed in the DFT geometry: an opening of the C4−C3−C5 ligand angle (88.75° DFT vs 89.03° crystal to be compared with 84.45° DFT for 1), a nonzero N1−C2−C1−N2 dihedral representing the twisting of the terminal pyridyl rings (6.26° DFT vs 8.64° crystal to be compared with 0.00° DFT for 1), and a non-right dihedral angle C2−N1−N1′−C2′ between the two ligands (76.01° DFT vs 74.58° crystal to be compared with 90.00° DFT for 1). That the gas-phase DFT calculation reproduces geometrical parameters of the 3A crystal structure with high fidelity suggests the second structure in the crystal unit cell (3B) is higher in energy. We suspect that packing forces caused by close intermolecular contacts between conformers and between 3B and nearby counterions are responsible and that the less twisted geometry is not indicative of a lack of strain, as the C4−C3−C5 ligand angle and N1−C2−C1−N2 ligand dihedral still deviate from the crystal structure of 1 (see Table S2, Supporting Information). While X-ray quality crystals for 2 have not been grown, the calculated geometry should provide a good indication of interligand steric-induced strain present in this complex relative to the other two complexes. The nonbrominated terpyridine ligand exhibits the same geometry as calculated for the terpyridine ligands on 1. Both 1 and the tpy ligand on 2
Figure 3. Orbital energies for LUMO to LUMO+3 (blue) and HOMO to HOMO−4 (red) for complexes 1 (R = H), which has no bromines and no VDW strain, 4 and 5 (R = 5,5″-Br2), which include bromines but no VDW strain, and 2 and 3 (R = 6,6″-Br2), which include bromines and VDW strain, as calculated for gas phase DFT optimized ground state geometries. Dashed lines indicate similar orbital character as determined by visual inspection. E
dx.doi.org/10.1021/jp404248z | J. Phys. Chem. A XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry A
Article
the 5 and 5″ positions exist on the same orbital lobes as are relevant for the 6 and 6″ positions in these frontier orbitals, thus allowing for a more direct comparison. A comparison between homoleptic complexes 1 and 5 shows the electronic effects of introducing bromines on the frontier orbitals without interligand VDW strain. The addition of the halogen causes the occupied Ru d-orbitals and the unoccupied ligand π*-orbitals to be stabilized by ∼0.2 eV, presumably due to the inductive electron withdrawing character of the bromine substituents.44 The occupied π orbitals of the ligand are slightly destabilized, presumably by the π-electron donating ability of the bromine. With electronic effects due to Br-substituents accounted for, the effect of strain on the frontier orbitals in 3 relative to 5 can be discerned. The relaxing of symmetry in 3 can be observed in the lifting of the degeneracy of the LUMO and LUMO+1 and the HOMO−1 and HOMO−2 set of orbitals. The LUMO and LUMO+1 are destabilized by 0.1 and 0.2 eV, respectively. This effect cannot be explained as being due to an electronic perturbation that arises when the bromine substitutions are changed from the 5 and 5″ positions to the 6 and 6″ positions. Calculations of the free ligands (locked to a geometrical configuration appropriate for tridentate binding of a metal center) show that the LUMO and LUMO+1 of 6,6″-Br2-tpy are actually lowered by 25 meV compared to those same orbitals in 5,5″-Br2-tpy. We expect then that the destabilization of the LUMO through LUMO+3 orbitals in 3 as compared to 5 arises from a ruffling or deplanarization of the ligand π-system as well as from interligand interactions between the electron-dense bromine located proximal to these π orbitals. This latter point, and especially the distinction of interligand effects from ruffling effects, is made more clearly below in the context of the heteroleptic species. Of final note in the homoleptic comparison, the metal-centered HOMO remains unchanged from 5 to 3 leading to an increased HOMO−LUMO gap, which assists in explaining the blue-shifted metal-to-ligand charge-transfer band in the electronic absorption spectrum (vide infra). The effect of interligand steric-induced strain can also be observed on the heteroleptic complexes through a comparison of unstrained 4 with strained 2. In 4, the LUMO and LUMO+2 orbitals are localized on the brominated terpyridine ligand and both are ∼0.1 eV lower than their LUMO+1 and LUMO+3 counterparts localized on the nonbrominated terpyridine. Again, we expect that this is due to electron-withdrawing properties of the Br substituents comparable to what was described above. Upon addition of strain, all orbitals localized on the nonbrominated terpyridine (HOMO−4, LUMO+1, and LUMO+3 within 4) are destabilized by ∼0.2 eV on going to 2, while the brominated terpyridine orbitals are conversely stabilized slightly by ∼0.05 eV. This is a clear indication that the energy of the tpy ligands is increased by interactions with the bromine substituent of the other ligand and not from the ruffling of the ligand caused by distortions. The origin of the calculated stabilization of the orbitals on the brominated ligand is unknown but could arise from slight differences in orbital coefficients at the 5,5″ and 6,6″ tpy positions, as observed for the ligand calculations (vide supra). Electronic Absorption Spectra. The effect of the bromine substituents in the 6 and 6″ positions on electronic structure has been accessed through electronic absorption spectroscopy, shown in Figure 4. All three complexes have broad absorption in the visible region from 400 to 550 nm, which is assigned as
Figure 4. Experimental (top) and simulated (bottom) electronic absorption spectra of 1 (blue), 2 (green), and 3 (red). Experimental spectra were collected at room temperature in acetonitrile solution.
metal-to-ligand-charge-transfer (MLCT) in character.85 Typically, substitution at the 4 and 4′ positions of the terpyridine with electron withdrawing groups results in a red-shifting of the MLCT absorption band due to stabilization of the terpyridine π* orbitals.19,44 However, substitution at the 6 and 6″ positions in 3 blue-shifts the MLCT peak, a result that is supported by the observed destabilization of the terpyridine π* orbitals as seen in Figure 3. The MLCT band for heteroleptic 2 is broader than that seen for 1 or 3. The separation in energy of the π* orbitals on each ligand in the heteroleptic species helps explain this with MLCT excitation partitioning between the two nonisoenergetic ligands.62 Both 2 and 3 have MLCT extinction coefficients that are approximately half that of 1, a result that qualitatively matches what has been seen in other sterically hindered metal complexes.47 We anticipate that the explanation in our case lies primarily in strain-induced lengthening of Ru− N bonds, which diminishes metal−ligand orbital overlap needed for intense MLCT. All three complexes have strong absorptions near 310 nm, typically assigned to intraligand π* ← π transitions.19,86 As with the MLCT transition, this band appears with approximately half the extinction coefficient in 2 and 3 as compared to 1. Interestingly, both 2 and 3 show a new intense and sharp transition appearing in the 350 nm region that is absent in 1. To gain insight into the nature of these transitions, it is useful to compare our data with tabulated absorption peaks from the analogous unstrained complexes with 5,5″ and 4,4″ dibromoterpyridines: complexes 4−7 as introduced earlier (Table 2).52,87 From the data, it is clear that while the 4,4″-Br2 and 5,5″-Br2 complexes exhibit a number of qualitatively similar absorption peaks to 1, none exhibit pronounced features in the 330−370 nm region. The full absorption spectrum for 5 can be seen in Figure S2, Supporting Information, and while there is a shoulder at 340 nm, there are no sharp peaks apparent. The comparative data then suggest that the peaks appearing near the 350 nm region for 2 and 3 are a consequence of the unique geometry of the complex and the close intramolecular contacts associated with the 6 and 6″ bromine substituents. TD-DFT Calculations. Gas phase TD-DFT calculations were undertaken to explore the character of the unknown transitions that appear in the 330−370 nm region for 2 and 3 F
dx.doi.org/10.1021/jp404248z | J. Phys. Chem. A XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry A
Article
Table 2. Tabulated Electronic Absorption Peaks for Complexes in Acetonitrile Solution at Room Temperature peaks (nm) (extinction coefficient/104 M−1 cm−1)
complex assignment (1) (2) (3) (4) (5) (6) (7) a
MLCT
2+
[Ru(tpy)2] [Ru(6,6″-Br2-tpy)(tpy)]2+ [Ru(6,6″-Br2-tpy)2]2+ [Ru(5,5″-Br2-tpy)(tpy) ]2+a [Ru(5,5″-Br2-tpy)2]2+ [Ru(4,4″-Br2-tpy)(tpy)]2+b [Ru(4,4″-Br2-tpy)2]2+b
474 474 466 474 475 478 481
interligand
(1.6) (0.68) (0.73) (1.1) (1.1)
intraligand/MLCT 308 304 306 320 317 311 312
350 (1.1) 354 (2.6)
(6.4) (3.0) (3.2) (4.9) (6.6)
270 (4.2) 272 (1.9) 282 (3.7) 282 (4.8)
270 (3.8) 260 (4.3) 272 263
Reference 52. bReference 87. Extinction coefficients were not reported in this reference.
Table 3. Character of Selected Important TD-DFT Transitions As Determined by NTO Analysis E (eV) (1) [Ru(tpy)2]2+ 3.20 4.45
λ (nm)
f
388 279
0.185 1.042a
Ru dxy,yz,xz→ tpy π* tpy π→ tpy π* Ru dxy,yz→ tpy π*
(98.5%) (52.7%) (33.2%)
MLCT intraligand MLCT
0.083 0.127
Ru dxy→ tpya π* tpya π, Br px, tpyb→ tpya π* Ru dxy→ tpya π* tpyb π, Br px, tpya→ tpyb π* Ru dxy→ tpyb π* Ru dxy→ tpya π* tpya π, Br px→ tpya π* Ru dxy,yz→ tpyb π* tpyb π→ tpyb π*
(98.8%) (81.1%) (16.1%) (72.6%) (16.0%) (70.0%) (20.0%) (70.2%) (21.1%)
MLCT inter/intraligand MLCT inter/intraligand MLCT MLCT intraligand MLCT intraligand
Ru dxy,yz→ tpy π* tpy π, Br px→ tpy tpy π, Br px→ tpy Ru dxy→ Ru dz2 tpy π, Br px→ tpy Ru dxy→ tpy π* tpy π→ tpy π* Ru dxy→ tpy π* tpy π, Br px→ tpy
(95.5%) (90.4%) (66.7%) (24.1%) (87.4%) (78.9%) (18.4%) (77.9%) (19.5%)
MLCT inter/intraligand inter/intraligand metal centered intraligand MLCT intraligand MLCT intraligand
(2) [Ru(6,6″-Br2-tpya)(tpyb)]2+ 3.15 394 3.88 320 4.07
305
0.079
4.26
291
0.345
4.54
273
0.363
(3) [Ru(6,6″-Br2-tpy)2]2+ 3.09 401 3.82 325 3.91 317
0.043 0.148 0.094
4.02 4.40
308 282
0.084 0.246
4.42
281
0.243
orbital character
π* π* π*
π*
(% contribution)
assignment
a
This transition is the sum of two transitions from TD-DFT results with the identical energy, oscillator strength, and NTO results. Here and in Figure 4, it is presented as one transition with twice the oscillator strength for clarity. See Table S3, Supporting Information, for full TD-DFT results.
and to confirm the reduction in strength of the other bands. The simulated absorption spectra for 1−3 are presented in Figure 4 (bottom). Important transitions that represent the majority of a given observed band were selected and characterized using natural transition orbital (NTO) analysis and are listed in Table 3. An attachment/detachment (AD) density analysis is discussed below. A full list of calculated transitions for complexes 1−7 (Table S3) including zero oscillator strength transitions, NTO analysis of complexes 4−7 (Table S4), and the simulated spectra for complexes 4−7 (Figure S2) is available in the Supporting Information. Qualitatively, the simulated spectra for 1−3 capture the same features as the experimental spectra: a MLCT band on the red edge, a strong UV peak, and new transitions in between those bands from 300 to 350 nm in 2 and 3 that are not present in 1. In addition, the relative strength of the bands matches experimental results with the MLCT and UV bands in 2 and 3 at roughly half the magnitude of 1. Quantitatively, however, all of the simulated peaks are strongly blue-shifted from experimental values. The calculation overestimates the
experimental values by 0.4 and 0.6 eV for the UV and MLCT peaks, respectively. This is an issue that appears to be unique to TD-DFT calculations of ruthenium polypyridyl compounds and is observed with hybrid functional TD-DFT calculations for similar complexes in the literature; this is discussed further in the Supporting Information.88 Regardless, the qualitative features of the spectra are well reproduced, and the results can be used to determine the character of the unknown transitions. The NTO results in Table 3 reveal interesting effects on the orbital content of the transitions that contribute to the observed reduction in strength of absorption bands in 2 and 3 as compared to 1. For example, examining the character of the strongest calculated transition in the UV region at 4.45 eV for 1 reveals that the majority (52.7%) is ligand-based π* ← π with a minority character (33.2%) that is MLCT. Conversely, the highest oscillator strength transitions in the same region for 2 and 3 (4.54 and 4.42 eV, respectively) show that the situation is reversed: the majority character (>70%) is MLCT, while the minority character (
