Unidirectional Photo-induced Charge Separation and Thermal Charge Recombination of Cofacially Aligned Donor-Acceptor System Probed by Ultrafast Visible-Pump/Mid-IR-Probe Spectroscopy

Abstract

A new ${\pi}$-stacked donor-acceptor (D-A) system, [Ru(1-([2,2'-bipyridine]-6-yl-methyl)-3-(2-cyclohexa-2',5'-diene-1,4-dionyl)-1H-imidazole)(2,2':6',2"-terpyridine)]$[PF_6]_2$ (ImQ_T), has been synthesized and characterized. Similar to its precedent, [Ru(6-(2-cyclohexa-2',5'-diene-1,4-dione)-2,2':6',2"-terpyridine)(2,2':6',2"-terpyridine)]$[PF_6]_2$ (TQ_T), this system has a cofacial alignment of terpyridine (tpy) ligand and quinonyl (Q) group, which facilitates an electron transfer through ${\pi}$-stacked manifold. Despite the presence of lowest-energy charge transfer transition from the Ru-based-HOMO-to-Q-based-LUMO (MQCT) predicted by theoretical calculations by using time-dependent density functional theory (TD-DFT), the experimental steady-state absorption spectrum does not exhibit such a band. The selective excitation to the Ru-based occupied orbitals-to-tpy-based virtual orbital MLCT state was thus possible, from which charge separation (CS) reaction occurred. The photo-induced CS and thermal charge recombination (CR) reactions were probed by using ultrafast visible-pump/mid-IR-probe (TrIR) spectroscopic method. Analysis of decay kinetics of Q and $Q^-$ state CO stretching modes as well as aromatic C=C stretching mode of tpy ligand gave time constants of <1 ps for CS, 1-3 ps for CR, and 10-20 ps for vibrational cooling processes. The electron transfer pathway was revealed to be Ru-tpy-Q rather than Ru-bpy-imidazol-Q.

Introduction

Vectorial electron transfer is of primary interest in artificial photosynthesis.1-10 The ultimate goal is to generate a long-lived CS state in which an electron donor (D) and acceptor (A) is separated by a controllable distance enough to perform a variety of chemical reactions with the driving force obtained by the CS. In the early event of natural photosynthesis, CS through protein membrane occurs via only one direction (L-branch, vide infra) between the macroscopic C2 symmetric L- and M- branches, (L) the special pair-bacteriochlorophyll (BChlL)-bacteriopheophytin (BPheoL)-menaquinone (Qa)-ubiquinone (Qb) and (M) the special pair-bacteriochlorophyll (BChlM)-bacteriopheophytin (BPheoM)-ubiquinone (Qb)-menaquinone (Qa).11-20 Extensive efforts have been devoted to elucidate the background of this phenomenon; modeling the competition between through-space and through-bond electron transfer is one of those efforts.

We have recently reported the photo-induced CS and thermal CR reactions via ultrafast TrIR spectroscopy for the [Ru(6-(2-cyclohexa-2',5'-diene-1,4-dione)-2,2':6',2'' -terpyridine)( 2,2':6',2''-terpyridine)][PF6]2 (TQ_T) system where the orthogonal alignment of Q to tpy ligand imposes this unit juxtaposed cofacially on the central pyridyl ring of second tpy ligand (Scheme 1).21 CS reaction occurs almost instantaneously and CR reaction does with time constant of 22 ps. Due to the fact that the CS reaction is ultrafast in nature, and the authentic generation of Ru-to-tpy1 MLCT state was not available, the detailed elucidation of electron transfer pathway could not be determined. The major question is whether ET reaction undergoes through the Ru-tpy2-Q pathway via π-stacked D-A manifold or through Ru-tpy1-Q via throughbond mechanism (Scheme 1). In order to decipher this fundamental question, we have newly designed and synthesize ImQ_T, in which D and A topology is as same as TQ_T but the methylene bridge between bpy and imidazolyl moiety blocks the electronic delocalization of the ligand and thus an efficient charge transfer through this unit is inherently inhibited. The design of ImQ_T is largely indebted to its archetype, [ Ru(3-(2,2'-bipyridine-6-yl-methyl)-1-methyl-1H-imidazole)( 2,2':6',2''-terpyridine)] [ PF6]2 (Ru(tpy)(b^im)).22 The characteristic feature of Ru(tpy)(b^im) is its electronic structure where the imidazolyl-Ru moiety constitutes HOMO and the tpy moiety does LUMO. Due to this spatial separation of frontier orbitals, selective excitation toward a particular ligand, e.g. tpy, became available, from which a consecutive CS reaction can be manipulated.

Scheme 1

Here we report synthesis and geometry, ground- and excited state electronic structures, and CS and CR reaction dynamics of ImQ_T molecule.

 

Experimental

Materials. All reactions were carried out under a nitrogen atmosphere unless otherwise noted. Standard Schlenk techniques were employed to manipulate air-sensitive solutions, while workup procedures were done in air. Tetrahydrofuran (THF) were purchased from Fischer Scientific (HPLC grade) and dried over Na/benzophenone and were subsequently distilled under nitrogen prior to use. Acetone (Kanto, HPLC) and acetonitrile (Samchun, 99.5%) were distilled over CaH2 prior to use. Methanol (Fischer Scientific, absolute), ethylene glycol (Aldrich) and hydrobromic acid (Samchun, 48 wt %) were used without further purification. Imidazole (99%), 2,5-dimethoxyphenylboronic acid, 2-hydroxy-6-methylpyridine (97%), n-butyllithium solution (2.5 M in hexane), diisopropylamine, trifluoromethanesulfonic anhydride (99%), chlorotrimethylsilane (97%), hexachloroethane (99%), copper(Ⅱ) acetate (98%), lithium chloride, 2,2':6',2''-terpyridine (98%), ammonium hexafluorophosphate (NH4PF6) and 2,3-dichloro-5,6-dicyano-p-benzoquinone (DDQ) were purchased from Aldrich Chemical Co. Tetrakis(triphenylphosphine) palladium(0) was purchased from Pressure Chemical Co. Sodium bicarbonate (98%) was purchased from Samchun Chemical Co. 6-Chloromethyl-2,2'-bipyridine23 and (2,2':6',2''-tepyridine)(trichloro)Ru (III) (Ru(tpy)Cl3)24 were prepared according to literature procedures. Column chromatography was performed on silica gel 60 (230-400 Mesh, Merck).

Instrumentation. 1H and 13C NMR spectra were recorded with Bruker (75 MHz for 13C NMR), Agilent (400 MHz and 100 MHz for 1H and 13C NMR, respectively) and Agilent (500 MHz and 125 MHz for 1H and 13C NMR, respectively) spectrometers. 1H NMR spectra were taken in CDCl3 and DMSO-d6 and were referenced to residual CDCl3 (7.26 ppm) and DMSO-d6 (2.50 ppm), respectively. Chemical shifts of the 13C NMR spectra were measured relative to CDCl3 (77.16 ppm) or DMSO-d6 (39.52). High-resolution mass spectrometry (HRMS) data were obtained at the Korea Basic Science Institute (Daegu). Electronic absorption spectra were recorded on a Beckman Du-650 spectrophotometer Cyclic voltammograms were obtained with a CH Instrument voltammetric analyzer. Measurements were performed after the acetonitrile (spectroscopic grade) solution was purged with dry nitrogen gas for 30 min. The supporting electrolyte was 0.1 M tetrabutylammonium hexafluorophosphate (TBAPF6). Glassy carbon and Ag/Ag+ (0.01 M AgNO3) were used as working and reference electrodes, respectively. The scan rate was maintained at 100 mV/s.

Synthesis.

1-(2,5-Dimethoxyphenyl)-1H-imidazole (1): Imidazole (0.68 g, 10 mmol) and 2,5-dimethoxyphenylboronic acid (2.18 g, 12 mmol) were added to a solution of copper(II) acetate (0.15 g, 8 mol %) in 20 mL methanol and the mixture was allowed to stirred at 65 ℃ for 36 h under an oxygen atmosphere. The reaction mixture was concentrated by rotary evaporation and extracted with dichloromethane. Purification of the resulting residue by silica-gel chromatography (dichloromethane: methanol = 24:1) provided orange oil. (Yield: 82%, 1.68 g) 1H NMR (500 MHz, CDCl3) δ 7.74 (s, 1 H), 7.16 (s, 1 H), 7.10 (s, 1 H), 6.92 (d, J = 8.8 Hz, 1 H), 6.83- 6.79 (m, 2 H), 3.73 (s, 3 H), 3.71 (s, 3 H); 13C NMR (125 MHz, CDCl3) δ 153.7, 146.5, 137.7, 128.8, 127.0, 120.1, 113.6, 113.1, 111.7, 56.4, 55.8; HRMS (FAB+), m/z [M+H]+ found (calc): 205.0976 (205.0977).

2-(1H-imidazol-1-yl)benzene-1,4-diol (2): To a schlenk flask containing 1 (1.02 g, 5 mmol), 30 mL of 48 wt % hydrobromic acid was added. The reaction mixture was refluxed overnight. The solvent was removed by distillation and concentrated to 5 mL. Solid NaHCO3 was added until the solution was neutralized. Precipitate was filtered and dissolved with methanol. Purification of the resulting residue by silica-gel chromatography (dichloromethane:methanol = 15:1) gave white solid (Yield: 94%, 0.82 g). 1H NMR (400 MHz, DMSO) δ 9.42 (br, 2 H), 7.95 (s, 1 H), 7.43 (t, J = 1.2 Hz, 1 H), 7.04 (s, 1 H), 6.89 (d, J = 8.7 Hz, 1 H), 6.74 (d, J = 2.8 Hz, 1 H), 6.67 (dd, J = 8.7, 2.8 Hz, 1 H); 13C NMR (100 MHz, DMSO) δ 150.3, 142.5, 137.4, 128.0, 124.9, 120.3, 117.8, 115.1, 111.7; HRMS (FAB+), m/z [M+H]+ found (calc): 177.0665 (177.0664).

1-([2,2'-Bipyridine]-6-yl-methyl)-3-(2,5-dihydroxyphenyl)-1H-imidazol-3-ium chloride (3): A solution of 2 (0.35 g, 2.0 mmol) and 6-(chloromethyl)-2,2'-bipyridine (0.49 g, 2.4 mmol) in 10 mL of acetonitrile was refluxed for 36 h. After cooling to room temperature, solvent was removed by rotary evaporator. Resulting solid was dissolved by minimum amount of dichloromethane and dropped to 100 mL of diethyl ether and stirred for 30 min. White precipitate was filtered and dried. (Yield: 92%, 0.70 g) 1H NMR (400 MHz, DMSO) δ 10.31 (s, 1 H), 9.80 (s, 1 H), 9.56 (s, 1 H), 8.70- 8.66 (m, 1 H), 8.37 (d, J = 7.9 Hz, 1 H), 8.24 (d, J = 7.9 Hz, 1 H), 8.10 (s, 1 H), 8.05 (dd, J = 8.9, 6.7 Hz, 2 H), 7.96 (td, J = 7.6, 1.2 Hz, 1 H), 7.60 (d, J = 7.7 Hz, 1 H), 7.46 (dd, J = 7.5, 4.8 Hz, 1 H), 7.08 (d, J = 8.8 Hz, 1 H), 6.96 (d, J = 2.7 Hz, 1 H), 6.88 (dd, J = 8.8, 2.4 Hz, 1 H), 5.78 (s, 2 H); 13C NMR (100 MHz, DMSO) δ 155.0, 154.5, 153.1, 150.3, 149.4, 143.0, 138.7, 138.1, 137.5, 124.5, 123.4, 123.1, 122.5, 122.1, 120.6, 120.0, 118.0, 117.9, 112.0, 53.1; HRMS (FAB+), m/z [M]+ found (calc): 345.1353 (345.1352).

[Ru(1-([2,2'-Bipyridine]-6-yl-methyl)-3-(2,5-dihydroxyphenyl)-1H-imidazole)(2,2':6',2''-terpyridine)][PF6]2 (Im-QH2_T, 4): A mixture of Ru(tpy)Cl3 (132 mg, 0.3 mmol) and 3 (114 mg, 0.3 mmol) in 5 mL of ethylene glycol was heated at 160 ℃ for 4 h. The reaction mixture was allowed to cool slowly to room temperature. The solution was dropped to a saturated NH4PF6 aqueous solution. The brown precipitate was filtered and purified with silica-gel column chromatography. Elution with CH3CN/0.5 M NaNO3 (9:1) gave a product as an orange band. The collected orange band was concentrated to ca. 5 mL and was triturated in a saturated NH4PF6 aqueous solution. The precipitated orange solid was filtered and washed several times with ether, water and dried in vacuo. (Yield: 69%, 200 mg) 1H NMR (400 MHz, DMSO) δ 8.90 (d, J = 8.0 Hz, 1 H), 8.81 (s, 1 H, –OH), 8.68 (d, J = 8.2 Hz, 1 H), 8.50 (s, 1 H, –OH), 8.47-8.38 (m, 5 H), 8.18 (d, J = 7.8 Hz, 1 H), 8.05–7.88 (m, 4 H), 7.76 (d, J = 5.2 Hz, 1 H), 7.65 (d, J = 1.8 Hz, 1 H), 7.54 (d, J = 5.3 Hz, 1 H), 7.32-7.23 (m, 3 H), 6.91 (d, J = 1.8 Hz, 1 H), 6.68 (d, J = 5.1 Hz, 1 H), 6.48 (dd, J = 8.8, 2.9 Hz, 1 H), 6.24 (d, J = 8.8 Hz, 1 H), 6.04 (d, J = 3.2 Hz, 2 H), 5.18 (d, J = 2.8 Hz, 1 H); 13C NMR (100 MHz, DMSO) δ 178.6, 158.3, 157.9, 157.2, 155.9, 155.4, 155.3, 154.7, 153.3, 152.4, 149.1, 147.2, 145.2, 137.7, 137.7, 137.5, 137.4, 134.5, 127.5, 127.1, 127.1, 126.3, 125.7, 124.8, 123.8, 123.7, 123.5, 123.5, 123.4, 123.0, 122.6, 117.4, 116.2, 113.5, 53.5; HRMS (FAB+), m/z [M+PF6]+ found (calc): 824.0914 (824.0921).

[Ru(1-([2,2'-Bipyridine]-6-yl-methyl)-3-(2-cyclohexa-2',5'-diene-1,4-dionyl)-1H-imidazole)(2,2':6',2''-terpyridine)] [PF6]2 (ImQ_T, 5): 4 (145 mg, 0.15 mmol) and 10 equivalents of DDQ (0.34 g) were dissolved in 10 mL of distilled acetone and stirred under nitrogen for 12 h. The solution was dropped to 100 mL of diethyl ether. Brown solids were filtered and washed with diethyl ether several times. (Yield: Quantitative) 1H NMR (400 MHz, DMSO) δ 8.93 (d, J = 8.0 Hz, 1 H), 8.77 (d, J = 8.1 Hz, 1 H), 8.69 (dd, J = 16.6, 8.1 Hz, 3 H), 8.47 (dd, J = 17.2, 8.3 Hz, 2 H), 8.18 (dd, J = 13.4, 7.7 Hz, 2 H), 8.05 (t, J = 7.8 Hz, 1 H), 7.95 (t, J = 7.2 Hz, 3 H), 7.70 (s, 1 H), 7.45 (d, J = 5.4 Hz, 1 H), 7.37 (t, J = 8.0 Hz, 1 H), 7.31 (t, J = 8.0 Hz, 1 H), 7.26 (t, J = 8.0 Hz, 1 H), 7.07 (s, 1 H), 6.86 (dd, J = 10.2, 2.3 Hz, 1 H), 6.79 (d, J = 5.1 Hz, 1 H), 6.63 (d, J = 10.2 Hz, 1 H), 6.20 (d, J = 2.4 Hz, 1 H), 6.05 (dd, J = 54.6, 16.3 Hz, 2 H); 13C NMR (100 MHz, DMSO) δ 187.1, 180.2, 179.4, 157.8, 157.5, 157.1, 157.0, 156.9, 156.0, 155.0, 154.3, 153.2, 147.9, 142.4, 138.6, 138.6, 138.2, 137.2, 136.3, 136.2, 132.3, 128.8, 128.4, 127.7, 126.9, 125.0, 125.0, 124.8, 124.5, 124.3, 124.2, 124.1, 124.1, 124.0, 53.8; HRMS (FAB+), m/z [M+H]+ found (calc): 678.1190 (678.1201).

Pump-Probe Transient Absorption Spectroscopic Measurements. The details of the time-resolved vibrational spectrometer are described elsewhere.25-27 Briefly, two identical home-built optical parametric amplifiers (OPAs), pumped by a commercial Ti:sapphire regenerative amplifier (Hurricane, Spectra Physics) with a repetition rate of 1 kHz, are used to generate a visible pump pulse and a mid-IR probe pulse. Pump pulse at 575 nm with 3.0 mJ of energy was generated by frequency doubling of a signal pulse of one OPA. Tunable mid-IR probe pulse was generated by difference frequency mixing of the signal and idler pulse of the other OPA. The polarization of the pump pulse was set at the magic angle (54.7°) relative to the probe pulse to recover the isotropic absorption spectrum. The broadband transmitted probe pulse was detected with a 64-elements N2(l)-cooled HgCdTe array detector. The array detector is mounted in the focal plane of a 320 mm monochromator with a 120 l/mm grating, resulting in a spectral resolution of ca. 1.3 cm-1 / pixel at 1600 cm-1. The signals from each of the detector elements were amplified with a homebuilt 64-channel amplifier and digitized by a 12-bit analog-to-digital converter. Chopping the pump pulse at half the repetition frequency of the laser and computing the difference between the pumped and the unpumped absorbance determine the pump-induced change in the absorbance of the sample, ΔA. Due to the excellent short-term stability of the IR light source (< 0.5% rms), less than 1 × 10-4 rms in absorbance units after 0.5 sec of signal averaging is routinely obtained without single shot referencing with an independent detector. The pump spot was made sufficiently larger than the probe spot to ensure spatially uniform photoexcitation across the spatial dimensions of the probe pulse. The instrument response function was typically 180 fs.

Computational Method. All calculations were performed using the Gaussian 09 program package.28 All the results were obtained using a spin-restricted formalism at the DFT level of theory29-33 using the B3LYP hybrid functional.34 The ruthenium atom was described by using the LANL2DZ basis set, which includes the relativistic effective core potential (ECP) of Hay and Wadt35,36 for the inner electrons and a double-ζ basis set for the outer electrons. The standard 6-31G(d) basis set37 was used for the remaining atoms. All geometry optimization procedure was done without any symmetry restriction. Frequency calculations were performed to extract the IR frequencies and intensities with the optimized geometries of both ground- and S1 excited-states. The excitation energies and oscillator strengths at the optimized geometry in the ground state are obtained by TD-DFT calculations with the same basis sets as those for the ground state.38,39

 

Results and Discussion

Synthesis and Structure. The design strategy of ImQ_T complex was inspired by the TQ_T topology21 that satisfies two conditions; (1) electron acceptor, Q, should be placed juxtaposed to the primary ligand (tpy) to facilitate π-π interaction, (2) the direct conjugation between the secondary ligand (bpy) and the electron acceptor moiety should be minimized. In our previous work, we observed that the electronic population of the LUMO of [Ru(bip)2]2+ (bip = 2,6-bis(3-methylimidazol-1-yl)pyridine) is mainly localized in the central pyridyl ring indicating that the electron is not delocalized over the whole bip ligand but is confined within the central pyridyl ring.40 Furthermore, we observed that LUMO energy levels were leveled in the series of heteroleptic [Ru(tpy)L]2+ complexes in which L is either NNC or NN^C structural motif.22 These works demonstrate that an electronic delocalization between the pyridyl ring and the imidazolyl one in bip ligand or that between the bipyridyl ring and the imidazolyl one in NNC or NN^C type ligands is inherently blocked regardless the presence of a methylene bridge between two units. Thus the electronic population in the Ru-to-bpy MLCT state of ImQ_T can be segregated from the Q moiety. The photo-induced CS as well as thermal CR reaction pathways would be determined by the competition between through-space Ru-tpy-Q route and throughbond Ru-bpy-methylene bridge-imidazole-Q counterpart if the laser excitation generates mixed Ru-to-tpy/Ru-to-bpy MLCTs. However, if a selective preparation of either one of such MLCTs is possible, the elucidation of electron transfer dynamics for a particular pathway becomes available.

The synthesis of 1-([2,2'-bipyridine]-6-yl-methyl)-3-(2,5- dihydroxyphenyl)-1H-imidazol-3-ium chloride (3) was accomplished by the reaction sequence shown in Scheme 2. In the presence of catalytic amount of copper acetate, the coupling reaction between imidazole and 2,5-dimethoxyphenylboronic acid was performed in methanol under oxygen atmosphere, which affords 1-(2,5-dimethoxypheny)-1H-limidazole (1) in 82% yield. A subsequent demethylation of 1 with HBr gave 2 in 94% yield. Reaction of 2 with 6-(chloromethyl)-2,2'-bipyridine in refluxing acetonitrile gave the hydroquinonyl substituted imidazolium compound, 3, in 92% yield.

In order to attach the prepared ligand to the Ru metal, we employed [Ru(tpy)Cl3] for the metallating agent. The reaction requires high temperature; 4 h of the reaction in ethylene glycol at 160 ℃ followed by the anion exchange with NH4PF6 afforded the heteroleptic ruthenium complex, ImQH2_T (4) in 69% yield. The subsequent oxidation of hydroquinone to 1,4-benzoquinone was done by using DDQ as an oxidizing agent. The desired final complex was obtained quantitatively.

Scheme 2.Synthesis of ligand.

Scheme 3.Synthesis of Ru-Q complex.

Figure 1 illustrates the geometry of ImQ_T optimized by DFT calculation at B3LYP/6-31g(d)-LANL2DZ level. The geometry of TQ_T benchmark reported in our previous work21 is also displayed for the comparison. Due to the methylene bridge, bpy and imidazolyl ring is not coplanar. Imidazolyl ring is distorted from the bpy plane and thus the Q ring is directed outward from the vertical plane that bisects the tpy ligand. As a result, the centroid of Q ring plane is laterally shifted toward the peripheral pyridyl ring from the top of the central pyridyl ring of the tpy ligand. The distance between Ru metal and the centroid of Q ring is 4.81 Å, which is 0.12 Å longer than that of TQ_T. The distances between the Q centroid and tpy ring plane of ImQ_T and TQ_T are, however, similar each other; 3.54 and 3.58 Å, respectively. The slightly shorter plane-to-plane distance observed in ImQ_T is due to the larger bite angle of Npyr- Ru-Cim 41 compared to that of Npyr-Ru- Npyr in TQ_T.

The degree of π-π stacking interaction can be scaled by the chemical shifts of 1H-NMR spectra of corresponding proton peaks.42,43 The characteristic splitting patterns of Q ring protons unambiguously displays their peak positions at 6.79, 6.63, and 6.20 ppm. Considering the fact that chemical shifts of corresponding TQ_T Q proton peaks were observed at 6.68, 6.41. and 5.56 ppm, the π-π stacking interaction between tpy and Q of ImQ_T is seemingly less than that of TQ_T. Since the Q plane-to-tpy plane distance of ImQ_T is slightly shorter than that of TQ_T according to the calculated geometry, the concomitant down field shifts of Q proton peaks of ImQ_T relative to TQ_T are contradictory. The background of this result can be explained by the characteristic structural feature of ImQ_T; the Q ring is not exactly cofacial with the central pyridyl ring of tpy ligand. Therefore, the Q protons are less affected by the ring current imposed by the aromatic ring of tpy ligand and thus the peak positions of Q protons are going back to their original positions. Accordingly, the D-A electronic coupling in ImQ_T is expected to be decreased relative to TQ_T.

Figure 1.Geometries of TQ_T (a) and ImQ_T (b) with their top views (c and d). Metal-to-quinonyl plane centroid distances are shown in red. Quinonyl plane-to-terpyridyl plane distances are shown in blue.

Table 1.aExperimental conditions: solvent = acetonitrile, temperature = 23 ± 1 oC. bExperimental conditions: [compound] = 5 mM; [TBAPF6] = 0.1 M; solvent = acetonitrile; temperature = 23 ± 1 oC; scan rate = 50 mV/s; reference electrode = Ag/Ag+; working electrode = glassy carbon. All potentials are referenced to a ferrocene/ferrocenium redox couple as an internal standard and converted to NHE by the relation ferrocene/ ferrocenium vs. NHE = +0.64 V.

Electrochemistry and Molecular Orbitals. The electrochemical redox potentials of the ImQ_T in acetonitrile were recorded and their values are listed in Table 1. Both one electron oxidation and reduction occur reversibly at 1.46 and 0.11 V vs. NHE, respectively. The latter is shifted positively by 0.13 V relative to the value observed for TQ_T benchmark (-0.02 V vs. NHE) while the former is 0.18 V shifted negatively. The difference between the oxidation and reduction potentials of ImQ_T is 0.31 V smaller than that of TQ_T, and thus the energy level of charge separated state of ImQ_T is expected to be lower by the same amount than that of TQ_T. This difference gives rise to the change of the driving force of electron transfer reaction.

The DFT calculation result supports the electrochemical data. The calculated frontier molecular orbital energies and their corresponding isosurfaces are shown in Figure 2. The electronic populations of LUMOs of both ImQ_T and TQ_T are confined in Q moieties. On the other hand, those of HOMOs of two systems are mainly localized in Ru metal with a small amount being spread over the ligand system. For TQ_T, the electronic population of HOMO at Ru metal is ca. 75% and the rest 25% is equally distributed in two tpy ligands (Table 2). However, the electronic population of HOMO at Ru metal in ImQ_T is only 64.4% and those at tpy and imidazolyl ring are 10.5 and 20.3%, respectively.

Table 2.aTpy ligand coplanar to Q. bTpy ligand where Q is attached. c0 and 1 indicate HOMO and LUMO, respectively. –1, –2, ..., –4 correspond to HOMO-1, HOMO-2, …, HOMO-4, respectively while 2 and 3 represent LUMO+1 and LUMO+2, respectively.

Figure 2.3-Dimensional representations of frontier molecular orbital isosurfaces calculated at the B3LYP/LANL2DZ level. The MO isosurfaces of TQ_T molecule is from ref. 21.

More specifically, HOMO is constituted of Ru d orbitals and ligand π orbitals implying that the imidazol group provides a strong π-donation effect besides a σ-donation effect. As a result, the HOMO energy level of ImQ_T is destabilized compared to TQ_T, which gives rise to the negative shift of electrochemical oxidation potential. The calculated HOMO energy levels of ImQ_T and TQ_T are –6.04 and –6.31 eV; the difference of 0.27 eV is similar to that observed in the electrochemical oxidation potentials of two systems (0.18 V). The calculated value of LUMO energies of ImQ_T and TQ_T are -3.97 and -3.85 eV, respectively. Considering the fact that the electrochemical reduction potentials of two systems were 0.11 and –0.02 V, which differ by 0.13 V, 0.12 V of difference obtained from the calculation is in excellent agreement with the experimental result.

Figure 3.Steady state absorption spectra of ImQ_T in CH3CN at 25 ℃. Calculated transition energies and their corresponding oscillator strengths, f, of singlet → singlet transitions by TD-DFT method are depicted as vertical lines in wavenumber (top abscissa). MQCT transitions are emphasized by red arrows. The wavelength corresponding to the laser excitation (575 nm) is marked by green arrow. Note that the S4 state having MLCT in character is expected to reside near λex. The abscissa on top was shifted to its value multiplied by 1.1 to match with the experimental spectrum.

Steady State Absorption Spectrum. The UV-vis absorption spectrum of ImQ_T is depicted in Figure 3. The spectral feature resembles that of TQ_T benchmark; major MLCT bands appear in the visible region with λmax value of 477 nm and, importantly, there was no MQCT signature. Further examination using TD-DFT method demonstrates the nature of the electronic transitions of the absorption spectra. We calculated fifty singlet excited states and lower energy transitions (S1-S19) are listed in Table 3. These transitions (S1-S17) are co-plotted with the experimentally obtained results (Figure 3). Due to the intrinsic under-estimation of excited state energies by solvation-corrected TD-DFT calculation, we multiplied a correction factor of 1.1 to the calculated values.

The calculation results are in good agreement with the experimental data as shown in Figure 3. However, similar to the case of TQ_T, the lowest energy MQCT bands corresponding to Ru based occupied orbitals-to-quinone based LUMO transitions (S1-S3) does not appear in the absorption spectrum. The calculated oscillator strengths ( f ) of these transitions are 0.0069, 0.0006, and 0.0003 for S1, S2, and S3, respectively. The latter two are negligible and thus unlikely to appear in the spectrum. However, 0.0069 of f value for the first one is substantial, which is big enough to show up in the spectrum. Nevertheless, this charge transfer band does not appear in the spectrum indicating that the direct optical transition to the Ru+-Q- charge separated state is not viable. It is important to note that the transition to the S4 state has significantly large f value (0.0127), which gives rise to the lower energy tailing in the spectral envelope. Interestingly, the peak position corresponding to S4 state predicted by TDDFT calculation is at least 730 cm-1 separated from its closest neighbor, S5 state. Moreover, this peak is actually the lowest energy absorption band observable in the experimentally determined absorption spectrum. Thus we choose this wavelength (575 nm) for the laser excitation in order to generate a genuine S4 state as a reactant state for the photoinduced CS. The compositions of the S4 state wavefunction is listed in Table 3. All constituent single electron transitions are Ru-based occupied orbitals (HOMO and HOMO-2)-to-tpy-based virtual orbital (LUMO+1). LUMO+1 orbital has ca. 80% of electronic populations at tpy ligand. S4 transition can thus be regarded as a Ru-to-tpy MLCT.

Photo-induced Ultrafast Charge Separation and Thermal Charge Recombination Reaction Dynamics Probed by Vis-Pump/IR-probe Spectroscopy. As shown in our previous report, the dynamics of photo-induced CS and thermal CR reaction can best probed by the CO stretching mode of Q and Q-.21,44 The ground-state FTIR spectrum of the CO stretching mode of ImQ_T shows a clear absorption band at 1669 cm-1 (Figure 4). Upon electronic excitation at 575 nm in which low energy MLCT tailing prevails, the characteristic Q- mode appeared at 1518 cm-1 instantly with concomitant bleaching at 1669 cm-1 (Figure 4).

The correlation between transient absorption and bleaching was confirmed by their time-resolved kinetic behaviors (Figure 5). It is important to note that the rise of Q- mode CO stretching band was too fast to monitor precisely and was beyond our detection limit. Thus we only fit the kinetics of decay component. Contrary to the TQ_T case, the decay kinetics cannot be fitted by the monoexponential function described by Eq. (1); only by using a biexponential function could the data be satisfactorily modeled (eq. 2).

The fitting parameters of TrIR decay kinetics of Q- mode were A0 = 0.11, τ1 = 1.3 ± 0.5 ps with A1 = 1.33 and τ2 = 10.8 ± 1.1 ps with A2 = 6.72. Those of Q mode gave A0 = 0.03, τ1 = 3.3 ± 0.5 ps with A1 =-1.65 and τ2 = 13.7 ± 0.9 ps with A2 =-4.47, which are close to those of Q- mode kinetics. Therefore, two transient IR band at 1669 and 1518 cm-1 can be assigned to the CO stretching mode of ground and CT states, respectively. The background of the biexponential decay kinetics is not clear at this time. We suggest two scenarios: One is that the fast component is for the vibrational relaxation at the CS state and the slower one is for the charge recombination process. Another one is the charge recombination for the fast component and the vibrational relaxation from the hot ground state for the slow component.

In order to analyze TrIR spectrum as well as the electron transfer dynamics further, we calculated vibrational frequencies of both ground- and CS states (S0 and S1). In order to reproduce a ground state vibrational spectrum, we performed frequency calculation based on the optimized geometry of ground state by using DFT method. For the spectrum of S1 state, vibrational frequencies were calculated with the geometry optimized at S1 state by using TD-DFT method. Simulated vibrational spectra for the S0 (d) and S1 (c) with each peak being convoluted with Gaussian function are shown in Figure 6. Full-widths at half maximum (FWHM) of each peak are equally assumed to be 5 cm-1. The difference spectrum of two states (S1-S0) (b) is compared with the experimental TrIR spectrum at 1 ps time delay (a). Vibrational modes of representative peaks are illustrated in Figure 6(e). The experimental and simulated spectra are in excellent agreement with the latter being shifted to the red by ca. 75 cm-1. With these well matched spectra, we could analyze the experimental TrIR spectrum in more detail.

Table 3.aSubscripts correspond to the following orbitals: The highest occupied orbitals have index 0, i.e. 0 = HOMO, while all other occupied orbitals have index −1, −2. …, −n, which correspond to HOMO-1, HOMO-2, …, HOMO-n, respectively. LUMO = 1, LUMO+1 = 2, LUMO+2 = 3, and so on. bOscillator strength.

Figure 4.TrIR spectra of ImQ_T. FTIR spectra are shown in dotted lines (grey). Experimental conditions: λex = 575 nm, solvent = CD3CN. Temp = 23 ± 1 ℃.

Figure 5.Decay kinetics of Q- (top) and Q (bottom) CO stretching mode. Experimental conditions: λex = 575 nm, solvent = CD3CN. Temp = 23 ± 1 ℃.

Figure 6.Experimental (a) and simulated by TD-DFT calculation (b) TrIR spectra. Simulated S1 (c) and S0 (d) state vibrational spectra. Spectrum on (b) was obtained by (c)-(d). Peak positions are guided by blue (bleaching) and orange (transient absorption) dotted lines. Correlations of vibrational modes are connected by green (Q), red (tpy), and blue (bpy) lines. Blue Representative vibrational modes are illustrated in (e) with vibrational vectors in blue arrows.

Two CO stretching modes are predicted to appear at 1743 (asymmetric) and 1750 (symmetric) cm-1 with the former having much higher intensity. These two modes were shown in the experimental IR spectrum at 1669 and 1674 cm-1 (as a shoulder), respectively. These bands were bleached upon laser excitation at the same positions. In S1 state, these two stretching modes are shifted to 1587 and 1538 cm-1, respectively. The intense transient absorption appeared at 1518 cm-1 in the experimental TrIR spectrum corresponds to the former. The aromatic C-C stretching modes of Q are predicted to appear at 1706 (symmetric) and 1666 (asymmetric) cm-1 in the ground state and 1645 and 1578 cm-1 in the S1 state. In the experimental TrIR spectrum, these two bands were observed as weak bleaches at 1639 and 1595 cm-1, respectively. Transient absorption peaks corresponding to these modes were not exhibited clearly. The fingerprints corresponding to the aromatic stretching modes of the ligands are broadly distributed in the 1610~1660 cm-1 region in the simulated spectrum. Among these, notable bands are at 1621 and 1627 cm-1 for the stretching modes of tpy and bpy, respectively, which were slightly shifted to the blue in the S1 state; 1627 cm-1 for the former and 1630 for the latter. In the experimental TrIR spectrum, a moderately intense broad transient absorption was monitored at ~1559 cm-1. Given the fact that the excitation wavelength was 575 nm, which corresponds to the Ru-to-tpy MLCT, the origin of such a transient absorption stems most likely from stretching mode of tpy.

Figure 7 displays a kinetic profile probed at 1559 cm-1. The kinetic trace cannot be fitted with conventional multiexponential function. Thus we fitted the curve by three separate regions. In early time region up to 1 ps, transient absorption decays ultrafast with time constant of ca. ~0.1 ps. The instrumental response function of our laser system is 180 fs. Thus the time constant of this decay cannot be measured precisely and 0.1 ps of time constant is only an estimate. However, it is obvious that transient absorption instantly generated upon laser excitation decays in ultrafast manner indicating that the tpy-located electronic population moves to other place very fast. We tentatively attribute this component to the tpy-to-Q CS process. More specifically, it is the CS from the MLCT state to MQCT state. The reduced transient absorption peak stays for a while (~2 ps) and slightly rise again with time constant of ~1 ps indicating that the electronic population is rebuilt in the tpy ligand. The time constant of this process coincides with that of the fast component of the charge recombination process probed with Q and Q- CO stretching modes. Therefore, it is reasonable to assign such a component as a CR process from Q to tpy ligand. Then the decay of transient absorption at longer period with 22 ps of time constant is thus a vibrational cooling in the ground state. This value is well agreement with the longer component of charge recombination process probed with Q and Q- CO stretching mode.

Figure 7.Decay kinetics at 1559 cm-1. Experimental conditions: λex = 575 nm, solvent = CD3CN. Temp = 23 ± 1 ℃.

Figure 8.Normalized TrIR spectra of Q CO stretching mode of ImQ_T obtained at a series of time delays. Experimental conditions: λex = 575 nm, solvent = CD3CN. Temp = 23 ± 1 ℃.

The time constant of the vibrational cooling process estimated in this work is similar to that observed for the intramolecular CS and the following CR reaction in bis(η5- cyclopentadienyl)molybdenum coordinated to an ene-1,2- dithiolatenaphthalenetetracarboxylicdiimide ligand system.45 Figure 8 displays the normalized spectra in the Q mode bleaching area, which clearly demonstrates the evidence of vibrational cooling. The initial peak position corresponding to the Q mode bleaching at 1669 cm-1 gradually blue-shifted with concomitant narrowing of the bandwidth. At 56.2 ps when the CR and the vibrational cooling processes are almost finished, the peak position was 1670.3 cm-1, which was 1.3 cm-1 blue-shifted from its original position. This result is surprising because the time scale of CR reaction in TQ_T system was estimated to be ~20 ps. It was not clear whether such a time scale is for the sole CR reaction or for the processes including CR and vibrational cooling. However, the analysis of electron transfer dynamics with ImQ_T system clearly reveals that the CR reaction takes only 1-3 ps followed by 10-20 ps of vibraitional cooling process.

 

Conclusion

In sum, a new D-A system, ImQ_T, in which [Ru(tpy)-(NHC)]2+ donor and quinone acceptor are juxtaposed in a van-der-Waals contact has been synthesized and characterized. The ground- and excited-state electronic structures were determined by both spectroscopic and electrochemical method and were further confirmed by theoretical calculation by using DFT and TD-DFT method. The photoinduced charge separation and thermal charge recombination reactions were probed by ultrafast visible-pump/mid-IRprobe spectroscopic method. The transient IR absorption spectrum was analysed by the aid of ground- and excited state frequency calculation. Selective excitation to tpylocalized MLCT state gives rise to the ultrafast charge separation through π-stacked manifold. The time scale of the CS reaction could not be determined precisely and the estimated value is ca. 0.1 ps. Thermal CR reaction takes place within 1-3 ps range and the following vibrational cooling process takes 10-20 ps. This work demonstrates that the intramolecular photo-induced CS reaction can take place via π-stacked manifold of van der Waals contact in ultrafast manner. Detailed time constant of CS, CR and vibrational cooling processes were also determined.