Article pubs.acs.org/JPCB
Dynamics of Ground and Excited State Vibrational Relaxation and Energy Transfer in Transition Metal Carbonyls Milan Delor,† Igor V. Sazanovich,‡ Michael Towrie,‡ Steven J. Spall,† Theo Keane,† Alexander J. Blake,§ Claire Wilson,§ Anthony J. H. M. Meijer,*,† and Julia A. Weinstein*,† †
Department of Chemistry, University of Sheffield, Sheffield S3 7HF, U.K. Central Laser Facility, Research Complex at Harwell, Rutherford Appleton Laboratory, Harwell Science and Innovation Campus, STFC, Chilton, Oxfordshire, OX11 0QX, U.K. § School of Chemistry, University of Nottingham, Nottingham NG7 2RD, U.K. ‡
S Supporting Information *
ABSTRACT: Nonlinear vibrational spectroscopy provides insights into the dynamics of vibrational energy transfer in and between molecules, a crucial phenomenon in condensed phase physics, chemistry, and biology. Here we use frequency-domain 2-dimensional infrared (2DIR) spectroscopy to investigate the vibrational relaxation (VR) and vibrational energy transfer (VET) rates in different solvents in both the electronic ground and excited states of Re(Cl)(CO)3(4,4′-diethylester-2,2′-bipyridine), a prototypical transition metal carbonyl complex. The strong CO and ester CO stretch infrared reporters, located on opposite sides of the molecule, were monitored in the 1600−2100 cm−1 spectral region. VR in the lowest charge transfer triplet excited state (3CT) is found to be up to eight times faster than in the ground state. In the ground state, intramolecular anharmonic coupling may be solvent-assisted through solvent-induced frequency and charge fluctuations, and as such VR rates are solvent-dependent. In contrast, VR rates in the solvated 3CT state are surprisingly solvent-insensitive, which suggests that predominantly intramolecular effects are responsible for the rapid vibrational deactivation. The increased VR rates in the excited state are discussed in terms of intramolecular electrostatic interactions helping overcome structural and thermodynamic barriers for this process in the vicinity of the central heavy atom, a feature which may be of significance to nonequilibrium photoinduced processes observed in transition metal complexes in general.
■
INTRODUCTION Vibrational relaxation (VR) and energy transfer (VET) are ubiquitous and of central importance to photoinduced processes in the condensed phase.1−3 Vibrational dynamics occurs over a wide range of time scales, ranging from femtoseconds to hundreds of picoseconds,2 influencing the rates and mechanisms of a vast number of chemical reactions.3 Time-resolved two-dimensional infrared (2DIR) spectroscopy allows the direct measurement of population and relaxation dynamics of vibrational modes in molecules, providing insight on relaxation pathways, energy flow, and coupling mechanisms over significant distances.3−8 Transition metal complexes often exhibit nonequilibrium photoinduced processes such as ultrafast intersystem crossing (ISC) and electron transfer from nonthermalized excited states.9−15 It is known that both intra- and intermolecular structural reorganization play crucial roles in defining the rates and pathways of excited state reactions. For example, optically driven vibrational modes can explore the crossing point of spincrossover potential surfaces and drive ISC reactions along a specific coordinate,9−11,13 while solvent−solute interactions influence the shape of the potential energy landscape.16,17 Similarly, electron−hole recombination in charge transfer © 2014 American Chemical Society
reactions can be facilitated through high frequency vibrational modes coupling to the electronic excited state potential surfaces.16,18 Investigating VR, intramolecular vibrational redistribution (IVR), VET and associated coupling mechanisms in transition metal systems in both ground and excited electronic states may provide insights into the complex factors controlling their excited state dynamics and how to manipulate them. Here we use a narrowband infrared pump−broadband infrared probe 2DIR arrangement3,5,19,20 to initiate and follow VR and VET in the electronic ground and excited states of a prototypical transition metal complex. The excited state is accessed by introducing a UV pump before the IR pump− probe pulse sequence (Figure 1A).21 This pulse sequence allows unambiguous isolation of vibrational relaxation and energy transfer from electronic relaxation and solvation dynamics because the long (>nanosecond) lifetime of the excited state under investigation allows these processes to fully relax before introducing the IR pump−probe sequence. This Received: June 25, 2014 Revised: August 26, 2014 Published: September 8, 2014 11781
dx.doi.org/10.1021/jp506326u | J. Phys. Chem. B 2014, 118, 11781−11791
The Journal of Physical Chemistry B
Article
MeCN) allows us to investigate the effects of both intra- and intermolecular interactions, as well as how charge distribution affects vibrational coupling in this model transition metal carbonyl complex.
■
EXPERIMENTAL DETAILS The TRIR and 2DIR two/three-pulse experiments were performed on the ULTRA system at the STFC Central Laser Facility, Rutherford Appleton Laboratories, described elsewhere in detail.41,42 Three-pulse experiments use optical choppers modulating the repetition rate of the UV pump at 5 kHz and IR pump at 2.5 kHz while the probe pulse is at 10 kHz, facilitating the simultaneous collection of background, UV pump−IR probe (TRIR), IR pump−IR probe (ground state 2-dimensional IR) and UV pump−IR pump−IR probe (excited state 2dimensional IR or IR previbrationally excited TRIR depending on the order of the two pumps) data. The tunable IR pump (∼12 cm−1 bandwidth) and probe (400 cm−1 bandwidth) pulses used for double-resonance 2D-IR spectroscopy were generated by two optical parametric amplifiers pumped by synchronized Ti:sapphire-based regenerative amplifiers with ps and fs pulse duration, respectively. The 400 nm pump beam was generated from the second harmonic of the fs laser. The probe spectrum was recorded via a HgCdTe array detector and spectrometer combination that yielded a spectral resolution of 2 cm−1. For the data shown in the text, beam energies were approximately 0.5 μJ/pulse for both UV and IR pumps, with spot sizes of approximately 80, 100, and 150 μm for IR probe, IR pump, and UV pump, respectively. The pump−probe polarization relationships were set to magic angle to remove the effects of molecular rotation in all 2-pulse experiments. In 3pulse experiments, UV and IR pumps were set parallel to each other and the probe beam was set at magic angle with respect to these pumps. Samples were flowed and rastered in Harrick cells. UV−vis absorption and FTIR spectra were taken before and after every experiment to check for any decomposition. Optical densities at the excitation wavelength (400 nm) were kept at approximately 0.5. Synthesis of ReCOe. ReI(CO)5Cl (290 mg, 0.80 mmol) and 4,4′-diethylester-2,2′-bipyridine (245 mg, 0.82 mmol) were put in a Schlenk tube, and placed under Ar atmosphere by three vacuum−Ar cycles; 60 cm3 of dry Ar-saturated toluene was added, and the mixture was left to reflux for 15 h resulting in an orange reaction solution. The solution was then dried using a rotary evaporator, and the product obtained was purified using a silica packed column with 99:1 DCM/methanol mobile phase. The amount obtained was 293 mg (0.51 mmol, 64% yield). 1H NMR (250 MHz, CDCl3, ppm) δ = 9.23 (dd, 5.7 Hz, 0.6 Hz; 2H), 8.85 (d, 0.9 Hz, 2H), 8.11 (dd, 5.7 Hz, 1.6 Hz; 2H), 4.57 (q, 7.1 Hz; 2H; CH2), 1.51 (t, 7.1 Hz; 3H, CH3). The UV−vis spectrum is shown in Supporting Information, Figure S4. Single crystals suitable for X-ray diffraction were grown by slow diffusion of Et2O into a solution of the complex in THF at room temperature (rt), and an orange block crystal, 0.32 × 0.16 × 0.15 mm3 was analyzed. The details of X-ray structure determination are given in the Supporting Information (Tables S2−S6, Figure S5), and a ball-and-stick representation of the structure is shown in Figure 1B. Density Functional Theory (DFT). All calculations were performed using Gaussian 09, version C.01,43 compiled using Portland compiler v 8.0-2 with the Gaussian-supplied versions of ATLAS and BLAS,44 using the B3LYP functional of DFT.45
Figure 1. (A) Pulse sequence used in the excited state 2DIR experiments; t1 is kept at 600 ps. The ground state 2DIR experiments are performed without the UV pump. (B) X-ray structure of ReCOe; the structure of the DFT-optimized singlet state matches the X-ray structure satisfactorily (the overlap is shown in Supporting Information, Figure S6). (C) Ground state Fourier transform infrared (FTIR) spectrum of ReCOe in CD2Cl2. (D) Time-resolved IR (TRIR) spectrum 600 ps after 400 nm excitation. Asterisks denote excited state bands. Blue lines are calculated spectra with an anharmonic scaling factor of 0.9763, a fwhm of 7 cm−1, and with intensities scaled to the ground state a(es) band. The calculated TRIR spectrum is obtained by subtracting the ground singlet state IR spectrum from the lowest triplet state IR spectrum. See text for details of assignments.
task cannot be achieved using one-dimensional time-resolved infrared spectroscopy. The model system chosen to be analyzed in this study is Re(Cl)(CO)3(4,4′-diethylester-2,2′-bipyridine), ReCOe (Figure 1B). Carbonyl complexes of ReI with diimine ligands have been extensively investigated because of their exceptionally rich photophysics11,17,22 leading to applications as photosensitizers,23,24 photocatalysts for CO2 reduction,25,26 and triggers for electron transfer reactions27 stemming from the chargetransfer nature of their excited states. Their high-frequency CO stretching vibrations are excellent infrared reporters, and as such they have served as model systems for 2DIR spectroscopy.21,26,28−31 Recently, 2DIR has been used to study VR and VET in the electronic ground state of several other metallocarbonyl complexes,28,32,33 while excited state 2DIR was applied to study solvation dynamics,34 as well as solute reorientational and vibrational dynamics of hot photoproducts formed by UV photolysis.35−40 To the best of our knowledge, direct measurements and comparison of ground and solvated excited state vibrational dynamics have yet to be undertaken. In ReCOe, a CO group of the ester functionality on the bipyridine unit serves as an infrared reporter on the opposite side of the molecule to the CO groups. CO and CO stretching frequencies are separated in space by approximately 8−9 Å, but are separated in energy by some 300 cm−1, thus giving an opportunity to probe long-range VET across the molecule with a single broadband pulse. The direct comparison of vibrational dynamics in the ground and excited states of ReCOe in different solvents (CD2Cl2, CH2Cl2, toluene, 11782
dx.doi.org/10.1021/jp506326u | J. Phys. Chem. B 2014, 118, 11781−11791
The Journal of Physical Chemistry B
Article
In all cases an extensive basis set was used, consisting of 6311G(d,p)46 on all elements apart from rhenium, which was described using a Stuttgart−Dresden pseudopotential.47 In previous work it was found that this results in a reasonably accurate description of transition-metal complexes and their properties,48,49 allowing for semiquantitative comparison with experiment. In all calculations the bulk solvent was described using PCM,50 whereby the standard parameters for dichloromethane as supplied in Gaussian were used. An ultrafine integral grid was used. All structures were confirmed to be minima through the absence of imaginary frequencies, which were calculated in the harmonic approximation.
■
RESULTS The Fourier transform infrared (FTIR) spectrum in Figure 1C shows the ground state IR absorbance of ReCOe in CD2Cl2. The annotated assignments are based on published material,31 as well as DFT calculations which agree with previous assignments and reproduce the IR spectrum well. The bands at 2025, 1927, 1904, and 1733 cm−1 are assigned to a′(1) (symmetric stretching of all CO groups), a″ (antisymmetric stretching of equatorial COs), a′(2) (antisymmetric stretching of axial vs equatorial COs) and a(es) (two almost overlapping transitions corresponding to symmetric and antisymmetric stretching of ester COs, with the antisymmetric contribution dominating), respectively. Figure 1D shows the time-resolved IR (TRIR) spectrum 600 ps after 400 nm excitation of ReCOe in CD2Cl2, that is, once the solute is vibrationally relaxed and fully solvated. The excited state evolution of very similar complexes of the type Re(diimine)(CO)3X is well documented.11,17,22,51−54 The spectrum at 600 ps corresponds to the lowest triplet, a charge transfer (3CT) state of predominantly Re/Cl → bpy character. This is consistent with the observed positive shift of the ν(C O) bands and negative shift of ν(CO) in this excited state, originating from a shift of electron density away from the metal center toward the 4,4′-diethylester-2,2′-bipyridine unit which reduces π back-donation to the antibonding π* orbital of the −CO ligands and increases population of the antibonding orbital of the −CO ligands.51,54−56 Calculated infrared frequencies for the lowest triplet state within the harmonic approximation are in good agreement with experiment. The assignments are annotated in Figure 1D. Anharmonic calculations show that there are several combination bands in the area of interest in the lowest triplet state, in particular those at 2030 and 2054 cm−1 corresponding to [backbone torsional + a′(2)*] and [backbone torsional + a″*] modes. These combination bands are detailed in the Supporting Information. Vibrational mode frequencies and line widths are highly sensitive to their environments.4 Consequently, the position of the intense and well-defined high frequency CO modes in metal carbonyls have been extensively used as probes of electrostatic and solvation microenvironments.22,28,51−53,57−59, The larger line widths in the CT state can be, among other factors, an indication of increased frequency fluctuations and of reduction in symmetry.56,57,60 Figure 2 shows ground state 2-dimensional IR (GS2DIR) maps of ReCOe in CD2Cl2 at 1 and 10 ps delays (A,B), and 1D pump−probe slices (several delay times at a single pump frequency) upon excitation of the a′(1) band at 2025 cm−1 (C,D) and of the a(es) band at 1730 cm−1 (E). When the IR pump is resonant with the 0 → 1 transition of a particular vibration, the ν = 0 mode of that vibration is partially depleted
Figure 2. Ground state 2DIR spectra of ReCOe in CD2Cl2. (A, B) 2D maps at 1 and 10 ps IR pump−IR probe delay, respectively. Colors correspond to −25 mOD (blue) and +25 mOD (red). (C) Pump− probe 1D spectrum at representative delay times after exciting a′(1) (ca. 2025 cm−1). (D) Close-up of the a(es) probe region at early time delays following a′(1) excitation. (E) Pump−probe 1D spectrum at representative delay times after exciting a(es) (ca. 1730 cm−1). (F) Normalized spectra of the a(es) response (probe) to the excitation (pump) of several different vibrations. The a(es) diagonal peak pair clearly changes shape between 1 and 5 ps, indicating a shift from diagonal to off-diagonal character.
and the ν = 1 mode is populated. This results in a bleach of the ν = 0 → 1 absorption due to depopulation of the ground state and stimulated emission from ν = 1 → 0, and to a transient ν = 1 → 2 absorption anharmonically shifted to lower energies, giving the characteristic differential absorption profiles of Figure 2. This is usually referred to as a diagonal peak pair (pump frequency ≈ probe frequency). If the pumped vibration is coupled to other modes in the molecule, these other modes will respond to the excitation and show cross-peak or off-diagonal bleach-transient pairs separated by the off-diagonal anharmonicity, which corresponds to the frequency shift of the combination band of the two vibrations. Thus, the distinction between diagonal and off-diagonal signals can be made by analyzing the band shapes and peak positions. 2DIR spectral analysis has been extensively described in several publications, see for example refs 3, 4, 19, and 21. Figure 2F shows how the a(es) bleach/transient peak pair separation depends on which mode is excited. The diagonal response peak pair (blue) is clearly different to the off-diagonal response (red and black). However, the diagonal a(es) band shape changes with time, evolving into off-diagonal character (green) within 5 ps of excitation. This indicates that the ester CO stretch modes undergo IVR rapidly to populate spatially 11783
dx.doi.org/10.1021/jp506326u | J. Phys. Chem. B 2014, 118, 11781−11791
The Journal of Physical Chemistry B
Article
Figure 3. Ground State 2DIR kinetic traces showing the diagonal and cross-peak population dynamics of different vibrations to (A) a′(1) excitation and (C) a(es) excitation. Panel (B) shows the kinetic model used to deconvolute the a(es) cross-peak dynamics from the rate-limiting a′(1) decay (see text for details). Solid lines represent best fits using mono or biexponential decays deconvolved from a Gaussian instrument response function. Fitting parameters are summarized in Table 1.
Table 1. Summary of CO and CO Diagonal and Cross-Peak Lifetimes in Picoseconds from GS2DIR Dataa probe pump
a(es) a″ a′(1)
a(es)
a′(2)
a″
a′(1)
1.1 ± 0.3 (91%) 9 ± 2 (9%) (−) 14 ± 2 40 ± 10 (−) 5 ± 1b 11 ± 2b
(−) 7 ± 3 21 ± 3 48 ± 4
(−) 8 ± 1 14 ± 1 45 ± 1
(−) 3 ± 2 46 ± 3
(−) 3 ± 2 41 ± 3
(−) 6 ± 1 15 ± 1 (−) 4 ± 2 47 ± 4 0.7 ± 0.5 (42%) 48 ± 6 (58%)
a Kinetic traces were fitted to a sum of exponentials convoluted with a Gaussian instrumental response function which was determined to be between 1.5 and 1.7 ps depending on the experimental conditions. Negative signs in brackets indicate the lifetimes correspond to grow-in of the signals. For bi-exponential decays, relative component contributions are shown in brackets. Quoted lifetimes are averages ±1 standard deviation from several fits of the bleach dynamics of peak pairs, checked against integrated transient band dynamics. bRise−decay deconvolved from the rate-limiting a′(1) decay. See Figure 3B and text for details.
response may arise from the propagation of vibrational energy from the excited CO modes through anharmonically coupled low-frequency modes across the molecule.7,8,62 This VET eventually reaches low-frequency modes that are in spatial proximity to the CO groups and are anhamonically coupled to a(es), giving rise to the off-diagonal signal observed. However, the apparent 27 ps arrival time of the probed a(es) mode is clearly shorter than the 48 ps vibrational decay lifetime of the pumped a′(1) mode. This response time shorter than the decay of the pumped mode occurs because the 48 ps lifetime of the CO stretches is the rate limiting step in this experiment: a′(1) relaxation is slower than subsequent VET across the molecule and equilibration of intramolecular low-frequency modes with the surrounding solvent bath (vibrational cooling). The convolution of the slow a′(1) ν = 1 exponential decay with the probed response dynamics is distorting the off-diagonal a(es) peak kinetics. By deconvolving the a(es) cross-peak dynamics from the slow a′(1) decay, one can extract the response dynamics as if the a′(1) ν = 1 mode was infinitely short-lived.63 This is shown in Figure 3B. In the black curve, the rise (τ = 5 ps) represents the actual VET rate, and the decay (τ = 11 ps) corresponds to energy equilibration and vibrational cooling to the surrounding solvent bath.1,64 When a(es) is excited, the evolution depicted in Figure 2F is observed: an initial diagonal peak pair decays with a 1.1 ps lifetime, giving rise to an off-diagonal signal which decays with a 9 ps lifetime. This shows that the a(es) ν = 1 mode is shortlived (τ = 1.1 ps), quickly undergoing IVR to spatially close lower-frequency modes which couple to a(es). These lowfrequency modes then undergo vibrational relaxation on a 9 ps time scale. a′(1), a′(2), and a″ cross-peaks have an
nearby low-frequency modes which are anharmonically coupled to a(es), giving rise to the observed off-diagonal signal.3 Investigating the spectral dynamics of intermode coupling provides insight into the mechanisms responsible for these interactions. Figure 3 shows the diagonal and cross-peak dynamics of different modes following a′(1) excitation (A) and a(es) excitation (C). All subsequent kinetic traces were obtained from the bleaches of the peak pairs and fitted to a sum of exponentials convoluted with a Gaussian instrumental response function; Table 1 gives a summary of the obtained lifetimes. Note that these experiments have a time resolution of approximately 1.5 ps (limited by the ca. 12 cm−1 fwhm pump requirement), and therefore nothing quantitative can be inferred from kinetic components of