Solvent effects on the photophysical properties of Bu 4 N[(4,4 0 -bpy) Re(CO) 3 (bpy-5,5 0 -diCOO)] complex. A combined experimental and computational study

.


Introduction
The nature of the axial X ligand in fac-ReX(CO) 3 (a-diimine) complexes determines the fact that these compounds may or may not be strong luminophores, either in fluid solutions or at lowtemperature glasses. The accessible excited states, Re(CO) 3 to adiimine metal-ligand-to-ligand charge transfer (MLLCT), ligand-toligand charge transfer (LLCT), and/or intraligand (IL) excited states, are generally involved with their observed luminescence at room temperature. Due to the fact that they are thermally and photochemically stable as well as rich in their excited-state behavior and redox chemistry these complexes have been used in broad research areas such as electron transfer studies [1], solar energy conversion [2e4], catalysis [5], as luminescent sensors [6e8], and as labeling reagents and non-covalent probes for biomolecules and ions [9e11]. For instance, biochemical applications based on the formation of adducts between these rhenium complexes and biomolecules like DNA have emerged [12,13]. A major drawback, however, relies on their limited solubility in aqueous media. In fact, most of the fac-ReX(CO) 3 (a-diimine) complexes which annually appear on the literature reports are exclusively soluble in organic solvents while only a few of them can be managed in aqueous solutions at physiological pHs [14e21]. In previous work we have synthesized and characterized a water soluble fac-ReX(CO) 3 (a-diimine) complex coordinating the ligands 2,2 0 -bipyridine-5,5 0dicarboxylate (bpy-5,5 0 -diCOO) and 4,4 0 -bipyridine (4,4 0 -bpy) [22]. In the Bu 4 N[(4,4 0 -bpy)Re(CO) 3 (bpy-5,5 0 -diCOO)] complex structure the two negative charges of the carboxylate groups are balanced by the positive charges of the Re(I) center and that of a tetrabutyl ammonium cation (See Scheme 1). The pH dependent UV-vis spectroscopy of the Re(I) complex was compared with TD-DFT calculations on the different acid-base species existing in aqueous solutions and the nature of the responsible electronic transitions was established [22]. A switch between a MLLCT ReðCOÞ 3 /bpyÀ5;5 0 ÀdiCOO to a MLLCT ReðCOÞ 3 /4;4 0 Àbpy character of the lowest energy excited state occurred when the 4,4 0 -bpy ligand was protonated explaining the overall spectral changes in the 350e500 nm range after protonation of the Re(I) complex [22]. With the aim of further deepening the understanding of the role of the different excited states on the electronic structure of the complex, in this paper we continue with the study of the photophysical properties of Bu 4 N[(4,4 0 -bpy)Re(CO) 3 (bpy-5,5 0 -diCOO)]. For this purpose, laser induced optoacoustic spectroscopy (LIOAS), steady-state and time-resolved luminescence techniques as well as time-dependent density functional theory (TD-DFT) calculations were performed in both protic and aprotic media. The luminescence behavior of the Re(I) complex showed a strong solvent dependence. When the spectrum was taken in water the emission occurred peaking at l max ¼ 435 nm. In the aprotic solvents the maximum of the luminescence spectrum occurred at 604, 595 and 570 nm for CH 3 CN, CH 2 Cl 2 and CHCl 3 , respectively. In MeOH and EtOH, however, two peaks were observed; one at 425 and the other around 600 nm. Depending on the solvent, two or three exponential functions were needed to achieve a satisfactory fit of the time resolved luminescence decays. The presence of O 2 in the solutions quenched both the steady state and time resolved luminescence of the Re(I) complex in all the solvents. The calculated dipole moment, m, increased monotonically with the solvent dielectric constant, ε r . While the calculated energy of the MLLCT ReðCOÞ 3 /4;4 0 Àbpy transition increased with the increase of m, the calculated energy of the MLLCT ReðCOÞ 3 /bpyÀ5;5 0 ÀdiCOO transition remained essentially constant. In addition, unrestricted TD-DFT calculations on the triplet state were performed. These calculations showed that the S 0 -T n transition should be viewed as a delocalized MLLCT transition where the electronic charge is transferred from Re(CO) 3 moiety to both 4,4 0 -bpy and bpy-5,5 0 -diCOO ligands. All the experimental results as well as the theoretical calculations indicate that solvent effects on the steady state and time resolved luminescence of the Bu 4 N[(4,4 0 -bpy)Re(CO) 3 (bpy-5,5 0 -diCOO)] complex can be accounted by the coexistence of 3 MLLCT ReðCOÞ 3 /4;4 0 Àbpy , 3 MLLCT ReðCOÞ 3 /bpyÀ5;5 0 ÀdiCOO and 1 IL excited states.

Photophysical measurements
The UV-visible absorption spectra were obtained with a Shimadzu UV-1800 spectrophotometer. Steady-state fluorescence measurements were performed using a computer-interfaced Near-IR Fluorolog-3 Research Spectrofluorometer, and were corrected for differences in spectral response and light scattering. Both measurements were performed at room temperature (ca. 25 C) in quartz cells (1 cm path length). Solutions were deaerated with O 2free nitrogen in a gas-tight apparatus before recording the spectra.
Emission quantum yields, F em , were calculated with eq. (1), by using solutions of a reference compound (Rhodamine B in ethanol, where n s and n ref are the refractive indexes of the solutions containing the sample and the reference compound, respectively, A is the absorbance of the sample or reference at the excitation wavelength (A s and A ref < 0.1), and I is the integral of the emission spectrum and was used as a relative measure of the respective intensities of the luminescence. Lifetime measurements were performed at room temperature using the Time-correlated Single-Photon Counting (TCSPC) unit of the Fluorolog-3 with 341 nm NanoLED excitation source.
Photoacoustic measurements were performed by using a set-up already described [24]. Basically, a Q-Switched Nd:YAG laser (Surelite II, Continuum, 7 ns FWHM) operating at 355 nm was used as excitation source. The fluence of the laser pulses was varied using a neutral density filter, and the energy values were measured with a pyroelectric energy meter (Laser Precision Corp. RJ7620 and RJP-735). The laser beam was shaped by a 1 mm diameter pinhole in front of the cuvette, so that the resolution time in our experimental set-up, t R , was ca. 800 ns [25]. The detecting system consisted of a 4 mm thick x 4 mm in diameter home-made ceramic piezoelectric transducer (PZT), pressed against a cuvette side wall parallel to the laser beam direction. The detected acoustic signals were amplified, digitized by a digital oscilloscope (TDS 3032, Tektronix), and stored in a personal computer for further treatment of the data. New Coccine or 2-HBP were used as calorimetric reference (CR) compounds in the buffer and in CH 3 CN solutions, respectively [26,27]. A at l exc for the CR and the sample were matched within 5%.
Under the same experimental conditions for the sample and the CR, the signals generated by 64 laser shots were averaged to obtain a better signal to noise ratio. The absorption spectrum of the solution was checked before and after each set of laser shots, in order to detect possible sample degradation. Solutions were deaerated by bubbling N 2 or O 2 for 15 min before each experimental run. Given an excited species with a lifetime t, if t 1 / 5 t R then this species releases its heat content as prompt heat. On the other hand, when t > 5t R the excited species functions as heat storage within the time resolution of the LIOAS experiment.
The peak to peak amplitude of the first optoacoustic signal (H) is related to the fraction of the excitation laser fluence (F) absorbed by the sample by Eq. (2) [25], which was used for the handling of the LIOAS signals where the experimental constant K contains the thermo-elastic parameters of the solution as well as instrumental factors. A and a represent, respectively, the absorbance of the sample at the excitation wavelength and the fraction of the energy released to the medium as prompt heat.
The efficiency of the Re(I) complex toward singlet oxygen sensitization was assessed by the direct measurement of the 1 O 2 ( 1 D g ) near-infrared luminescence. After the irradiation of aerated solutions of the complex the generation of 1 O 2 ( 1 D g ) was evidenced by the appearance of the characteristic 1 O 2 ( 1 D g ) / 3 O 2 phosphorescence at 1270 nm. Time resolved phosphorescence detection was used for singlet oxygen detection. The near IR luminescence of 1 O 2 ( 1 D g ) was observed at 90 geometry through a 5 mm thick anti reflective coated silicon metal filter with a wavelength pass > 1.1 mm and an interference filter at 1.27 mm by means of a pre-amplified (low impedance) Ge-photodiode (Applied Detector Corporation, time resolution 1 ms). Simple exponential analysis of the emission decay was performed with the exclusion of the initial part of the signal. The quantum yield of 1 O 2 ( 1 D g ) formation, F D , was determined by measuring its phosphorescence intensity using an optically matched solution of phenalenone (F D ¼ 0.98) [28] as a reference sensitizer.

Computational details
The electronic structure of the rhenium complex was studied by performing Density Functional Theory [29e31] (DFT) and Time Dependent DFT [32e34] (TD-DFT) calculations using Gussian 09 software [35]. The ground state structure of the complex was optimized by DFT calculations using B3LYP functional and LanL2DZ basis set. Vibrational frequencies were computed at the same level of theory to confirm that these structures were minima on the energy surfaces. Solvents effects (CHCl 3 , CH 2 Cl 2 , CH 3 CN, EtOH, MeOH and H 2 O) on the optimization of the ground state structure were taken into account by means of the Polarizable Continuum Model (PCM) [36e38]. At the optimized ground state geometries, a set of 200 vertical excitations were computed with either the B3LYP, M06 and PBE0 hybrid functional in the six solvents mentioned above. LanL2DZ basis set was used for all the atoms in TD-DFT calculations with the B3LYP and M06 functional. TD-DFT calculations with the PBE0 functional were performed using the 6-311þþG(d,p) basis set for C, N, O and H atoms while LanL2TZ(f) [39,40] (triple zeta basis set designed for an ECP plus f polarization) was used for Re atom. Percentage compositions of different molecular fragments to molecular orbitals (MOs) from output files generated from Gaussian 09 were calculated using the AOMix program [41]. Absorption spectra were simulated with Gaussian distributions with a full-width at half-maximum (fwhm) set to 3000 cm À1 with the aid of GaussSum 2.2.5 program.
For characterization of the electronic transitions as partial charge transfer (CT) transitions, the following definition of the CT character can be used, eq. (3) [42]: where P g (M) and P I (M) are electronic densities on the metal in the electronic ground state and the I-th excited state, respectively. Positive CT I (M) values correspond to MLCT transitions, negative CT I (M) values correspond to LMCT transitions [42]. We can rewrite this definition by using the atomic orbital contribution to a particular MO. Therefore, the CT character for a HÀn / L þ m excitation is, eq. (4): If the excited state is formed by more than one one-electron excitation, then the metal CT character of this excited state is expressed in eq. (5) as a sum of CT characters of each participating excitation, i/j [42].
where C I (i/j) are the appropriate coefficients of the I-th transition giving the percentage contribution of a configuration to the resulting excited state TD-DFT wave function.

Absorption spectroscopy
The UV-vis spectrum of the Re(I) complex in aqueous solutions at pH ¼ 7 consists of one intense absorption band centered at l max ¼ 252 nm, a set of three intense bands centered at l max ¼ 293,  Table 1     bipyridine-5,5 0 -dicarboxylate (bpy-5,5 0 -diCOO). Table 2 lists the orbital percentage composition of the relevant MOs of [(4,4 0 -bpy) Re(CO) 3 (bpy-5,5 0 -diCOO)]in the six solvents studied based on the fragments defined above. With the aid of those percentage compositions we calculated the percentages of charge transfer (CT(%)) of Table 1 using eq. (3)e(5). The CT(%) for both MLLCT ReðCOÞ 3 /4;4 0 Àbpy and MLLCT ReðCOÞ 3 /bpyÀ5;5 0 ÀdiCOO decreases as the solvent polarity increases, see Table 1 and Fig. S1. This is due to two facts. On the one hand, the percentage composition of Re orbitals in H-7 is higher than in H-6 in H 2 O, CH 3 Fig. S2 shows the simulated absorption spectra for all the solvents under study. No significant differences in the calculated absorption spectra were encountered when the level of theory was switched between B3LYP/LanL2DZ, M06/LanL2DZ and PBE0/6-311þþG(d,p)/LanL2TZ(f). The comparison with the experimental spectra of Fig. 1 is quite satisfactory and the simulated spectra follow the observed absorptions with reasonable accuracy both in position and relative intensities. Moreover, the bathochromic shift experienced by the wavelength corresponding to the lowest energy band of Fig. 1 is reproduced by the spectra of Fig. S2. Interestingly, the solvent effect on the highest energy band of Fig. 1 (a bathochromic shift on l max as the polarity of the solvent is decreased from 252 nm in H 2 O to 256 nm in CHCl 3 ) is also observed in Fig. S2. However, the calculated l max for the highest energy band in CH 3 CN and EtOH deviates from that trend and appear at the shortest wavelengths. This disparity was observed either with B3LYP, M06 or PBE0 functional.

Steady state and time resolved luminescence
The steady state luminescence of Bu 4 N[(4,4 0 -bpy)Re(CO) 3 (bpy-5,5 0 -diCOO)] shows a strong solvent dependence. Fig. 4 shows the emission spectrum of N 2 -deaerated solutions of the Re(I) complex in the six solvents with l exc ¼ 360 nm. When the spectrum was taken in water the emission occurred peaking at l max ¼ 435 nm. In the aprotic solvents the maximum of the luminescence spectrum occurred at 604, 595 and 570 nm for CH 3 CN, CH 2 Cl 2 and CHCl 3 , respectively. However, in MeOH and EtOH, two peaks were observed; one at 425 and the other around 600 nm. On the other hand, Bu 4 N[(4,4 0 -bpy)Re(CO) 3 (bpy-5,5 0 -diCOO)] is a weak luminophore with F em ranging between 10 À3 and 10 À2 .
A sum of two exponentials were required to achieve a satisfactory fit of the emission decay profiles of the Re(I) complex solutions in H 2 O, MeOH, EtOH and CH 3 CN, while 3 components were required for the satisfactory analysis of the traces in CH 2 Cl 2 and     Table 3.

Singlet oxygen generation and LIOAS experiments
Singlet oxygen generation in D 2 O or CH 3 CN solutions could not be detected under our experimental conditions. The fact that 1 O 2 ( 1 D g ) is not generated with this complex is related to the low 3 MLLCT emission in those solvents. This is in agreement with our previous results in relation to the photosensitized generation of singlet oxygen from structurally related Re(I) complexes: 1 O 2 ( 1 D g ) generation was not observed when the luminescence quantum yields of the Re(I) complexes were as low as F em~1 0 À3 [23,24].
The photoacoustic signal in both solvents showed the same behavior: no time shift or changes of shape, with respect to the calorimetric reference signal (See Inset of Fig. 5 for aqueous solutions). Linear relationships in both solvents were obtained between the amplitude of the first optoacoustic signal (H) and the excitation fluence (F) for samples and references at various A, in a fluence range between 1 and 30 J/m 2 . The ratio between the slopes of these lines for sample and reference yielded the values of a for the samples. For the complex in CH 3 CN solution, see Fig. 5, the slopes were independent on the specific atmosphere and were the same that the CR measured at the same experimental conditions, as  These values combined with fluorescence data and singlet oxygen quantum yield production fit the energy balance of eq. (6) [24,25]: where E em is the "0e0" luminescence energy, E a is the molar energy of the laser pulse (hc/l exc ), E st is the molar energy content of the species formed with a quantum yield F st which stores energy for a time longer than the heat integration time and decays with a lifetime t. When singlet oxygen acts as storing species, the corresponding values for this species are F st ¼ F D and t ¼ t D .

Triplet energy calculations
Triplet emission energies were computed by optimizing the geometry at their excited state with analytic TD-DFT calculations utilizing B3LYP functional and LanL2DZ basis set. This approach provided the vertical transition energy of the lowest triplet excited state at its optimal geometry. In addition, we also used a D(SCF) method. While TD-DFT directly computes the vertical transition energy of the lowest triplet state at its optimal geometry, the D(SCF) method is based on the difference in total energies between selfconsistent calculations of the ground singlet and triplet electronic configurations and approximates "0e0" transitions. The validation of both techniques relies on the consistency between D(SCF) and TD-DFT results [46,47]. All calculations were carried out in the presence of CHCl 3 as a solvent through the PCM to better simulate experimental results. In unrestricted TD-DFT calculations, the choice of the triplet of interest (T n ) was based on the nature of the lowest energy singlet-singlet transition (HÀ7 / L see Table 1) in CHCl 3 . Then, the optimized geometry of T n and the calculated wavelength, l calc ¼ 491.4 nm, corresponding to a HÀ6 / (H a , Lþ1 b ) vertical transition were obtained. Fig. 6 shows frontier MO plots of the triplet's H a and Lþ1 b along with the singlet's HÀ6. H a is a MO centered on the bpy-5,5 0 -diCOO ligand while Lþ1 b is centered on the 4,4 0 -bpy ligand. Therefore, the S 0 -T n transition should be viewed as a delocalized MLLCT transition where the electronic charge is transferred from Re(CO) 3 moiety to both 4,4 0 -bpy and bpy-5,5 0 -diCOO ligands. The S 0 -T n energy difference computed as D(SCF) energy where T n and the S 0 structures are at the S 0 (DE 1 (T n @T n :S 0 @S 0 )) and T n (DE 2 (T n @T n :S 0 @T n )) relaxed geometry are reported in Table 4, at the B3LYP/LanL2DZ/PCM(CHCl 3 ), PBE0/ LanL2DZ/PCM(CHCl 3 ) and M06/LanL2DZ/PCM(CHCl 3 ) level of theory. DE 1 values are too "blue" compared to the experimental ones (vide supra) while DE 2 are higher by 0.2e0.3 eV. Table 5 shows selected bond lengths and angles around Re atom for the ground (S 0 ) and triplet (T n ) state computed at the B3LYP/LanL2DZ/ PCM(CHCl 3 ) level of theory (See Scheme 1 for the numbering of atoms in Table 5). It is observed that in the triplet state the Re-N1, Re-N2 and Re-N24 distances are shortened while Re-C46, Re-C47 and Re-C48 distances are elongated relative to the ground state. This is in agreement with the nature of the 3 MLLCT where formally the charge on the Re center is 2 þ and a negative charge is located over the azine ligand, increasing thus the electrostatic attraction between the Re center and the coordinated ligand, with a concomitant decrease in ReeN bonds distances. The increase of ReeC bonds distances points to a steric effect imposed by the approaching of the azine ligands. It is interesting to note that the Re-N1 and Re-N2 distances are shortened by 0.062 Å while the Re-N24 distance is only shortened by 0.023 Å. The elongation of Re-C46, Re-C47 and Re-C48 distances is 0.040, 0.039 and 0.037 Å, respectively, i.e. very similar for the three ReeC bonds.

Conclusions
The absorption spectrum as well as the steady state and time resolved luminescence of the Bu 4 N[(4,4 0 -bpy)Re(CO) 3 (bpy-5,5 0 -diCOO)] complex displays a marked solvent effect. The high energy absorption bands experience a bathochromic shift as the polarity of the solvent decreases. In addition to the bathochromic shift, the lowest energy band broadens. TD-DFT calculations allowed us to identify the main electronic transitions in the low energy region as 1 MLLCT ReðCOÞ 3 /4;4 0 Àbpy and 1 MLLCT ReðCOÞ 3 /bpyÀ5;5 0 ÀdiCOO . The magnitude of the calculated dipole moment increases with the polarity of the solvent. Besides, the energy of 1 MLLCT ReðCOÞ 3 /4;4 0 Àbpy increases also with solvent polarity increase. However, the energy of the 1 MLLCT ReðCOÞ 3 /bpyÀ5;5 0 ÀdiCOO transition is rather insensitive to the solvent polarity. This disparity is attributed to the fact that the 1 MLLCT ReðCOÞ 3 /4;4 0 Àbpy transition is nearly parallel to the orientation of the dipole moment while the 1 MLLCT ReðCOÞ 3 /bpyÀ5;5 0 ÀdiCOO transition is almost perpendicular to it. Both electronic transitions show a decrease of their CT(%) with increasing solvent polarity. The solvent effect on the position of the luminescence maximum and lifetime is consistent with the radiative and non-radiative deactivation of 1 IL, 3 MLLCT ReðCOÞ 3 /4;4 0 Àbpy and 3 MLLCT ReðCOÞ 3 /bpyÀ5;5 0 ÀdiCOO excited states. No singlet oxygen generation was detected either in D 2 O or CH 3 CN. LIOAS experiments showed that after photonic excitation all the absorbed energy was released to the medium as prompt heat in agreement with the low quantum yield of luminescence and the absence of any long-lived energy storage species. Unrestricted TD-DFT calculations were successfully applied to the triplet species. It is observed that in the triplet state the ReeN distances are shortened while ReeC distances are elongated relative to the ground state, in agreement with the CT nature of the triplet state. The calculated D(SCF) energies were somewhat "blue" shifted relative to the experimental ones. Table 4 Calculated wavelengths, l 1 and l 2 , corresponding to the energy differences DE 1 and DE 2 , respectively. DE(S 0 /T n ) are computed as D(SCF) energy where T n and the S 0 structures are at the S 0 (DE 1 (T n @T n :S 0 @S 0 )) and T n (DE 2 (T n @T n :S 0 @T n )) relaxed geometry. All calculations were performed at the at B3LYP/LanL2DZ, PBE0/LanL2DZ or M06/LanL2DZ level of theory including solvents effects (CHCl 3 ) through the PCM.  Table 5 Optimized geometrical parameters of S 0 and T n for [(4,4 0 -bpy)Re(CO) 3 (bpy-5,5 0 -diCOO)] À . Bond lengths (Å) and angles ( ) around the Re ion were calculated at the uB3LYP/LanL2DZ/PCM (CHCl 3 ) level of theory.