Femtosecond Dynamics of Excited-State Evolution in [Ru(bpy)3]2+

See allHide authors and affiliations

Science  03 Jan 1997:
Vol. 275, Issue 5296, pp. 54-57
DOI: 10.1126/science.275.5296.54


Time-resolved absorption spectroscopy on the femtosecond time scale has been used to monitor the earliest events associated with excited-state relaxation in tris-(2,2′-bipyridine)ruthenium(II). The data reveal dynamics associated with the temporal evolution of the Franck-Condon state to the lowest energy excited state of this molecule. The process is essentially complete in ∼300 femtoseconds after the initial excitation. This result is discussed with regard to reformulating long-held notions about excited-state relaxation, as well as its implication for the importance of non-equilibrium excited-state processes in understanding and designing molecular-based electron transfer, artificial photosynthetic, and photovoltaic assemblies in which compounds of this class are currently playing a key role.

Many of the photochemical and photophysical properties of molecules depend upon the kinetics of excited-state processes that occur after the absorption of a photon. Therefore, it is important to understand how excited states behave as a function of time. The conventional view of this temporal evolution holds that photoreactivity is largely dictated by the characteristics of the lowest energy excited state of a molecule. Thus, higher energy excited states are presumed to convert to this lowest energy state and in so doing are removed from any functional role in photochemical and photophysical transformations. Femtosecond time-resolved spectroscopy (1) has resulted in experimental observations that call into question the validity of this model; striking examples include the 200-fs cis-to-trans isomerization of rhodopsin (2), rapid photodissociation of CO from myoglobin-CO (3), and ultrafast electron injection into dye-sensitized semiconductor electrodes (4). These cases among others reveal a pattern of photoreactivity arising from nonthermalized excited states in which structural rearrangement and electron transfer can kinetically compete with processes such as intramolecular vibrational relaxation (IVR), internal conversion (IC), and intersystem crossing (ISC).

The inference that nonequilibrated excited states can play a chemically significant role in photoinduced transformations could have important consequences in a variety of areas ranging from design principles for electron-transfer assemblies and photochemical energy storage devices to the formulation of new theoretical models for molecular-based energy conversion and excited-state relaxation dynamics. Although much of the work in the ultrafast dynamics community has concentrated on either small molecules or biological systems, our research focuses on transition metal compounds (5, 6). Considerable effort is being expended in many laboratories to incorporate such complexes into schemes for artificial photosynthesis (7), photocatalysis (8), and the development of molecular-based photovoltaic and opto-electronic devices (9). In addition, the importance of ISC and IC processes in the photoinduced properties of metal-containing complexes makes such systems of interest for ultrafast dynamical studies of their excited-state behavior (10). We have obtained results that are not consistent with conventional models for describing photoinduced dynamics in transition metal complexes, suggesting the need to reevaluate currently accepted views of their excited-state behavior.

Tris-(2,2′-bipyridine)ruthenium(II), or [Ru(bpy)3]2+, is representative of a class of molecules that has played a central role in the development of inorganic photophysics in addition to providing the underpinning for the last two decades of research on transition metal-based photosensitization, charge separation, and photoinduced electron transfer chemistry (11). We have therefore chosen it as a prototype for our study of the ultrafast dynamics of metal complexes. The strong visible absorption characteristic of this molecule (Fig. 1) can be described as a metal-to-ligand charge transfer (1MLCT ← 1A1), in which an electron located in a metal-based d-orbital is transferred to a π* orbital of one of the bpy ligands (hν, photon energy) (12). The excited-state species that is eventually

Embedded Image

formed (a 3MLCT state) is well known to engage in both oxidative and reductive chemistry (11). This capability, coupled with its relatively long lifetime in fluid solution (τ ≈ 1 μs), near unity quantum yield of formation (13), the high visible absorptive cross section of the ground state, and the overall photochemical stability of this molecule and its derivatives makes them amenable to a wide variety of applications (14, 15). We have used femtosecond absorption spectroscopy to time resolve the formation of the 3MLCT state in [Ru(bpy)3]2+ (16) and have observed the initial evolution of the Franck-Condon state.

Fig. 1.

Electronic absorption spectrum of [Ru(bpy)3](PF6)2 in CH3CN solution at 298 K.

The laser system used has been described in detail elsewhere (17, 18). Excited-state difference spectra at various time delays Δt (Fig. 2) show that spectral changes in the 450- to 490-nm range are quite dramatic: A bleach begins to evolve at λ = 470 nm near Δt = 0 fs and grows substantially in intensity by Δt = 50 fs along with what appears to be a weak excited-state absorption at higher energy. The transient exhibits both a marked blue shift and changes in its spectral profile in all of the early time data until Δt ≈ 300 fs, after which most of this spectral shifting appears to have ceased. Single-wavelength kinetics traces, which were obtained by passing the probe beam through a monochromator after the sample, likewise show the growth of a bleach signal in the 450- to 490-nm region for the first 200 to 300 fs as well as a very weak excited-state absorption for λ > 500 nm. These changes at early times are complete at all wavelengths for Δt > 300 fs, consistent with the spectral traces illustrated in Fig. 2. Given that the lifetime of the 3MLCT state under these experimental conditions is on the order of 1 μs, the most critical aspect of the femtosecond difference spectra with regard to the formation of the 3MLCT state is the point at which the spectra stop changing. The data show that this occurs by ∼300 fs after the initial excitation: There is no evidence of any additional significant changes in the absorptive properties of the molecule in the spectra collected from Δt = 300 fs to 5 ps.

Fig. 2.

Femtosecond time-resolved excited-state-ground-state absorption difference spectra for [Ru(bpy)3](PF6)2 in CH3CN solution at 298 K (17). The probe pulse gave adequate intensity for probing in the 450- to 530-nm range; below 450 nm, the relative amplitude of the signal dropped due to the vanishing intensity of the probe pulse in this region and the strong ground-state absorbance of the sample. The resulting poor signal-to-noise ratio resulted in some uncertainty in the baseline correction and hence the amplitude of the differential absorption in this portion of the spectrum. The orientation of the probe beam was set at the magic angle (∼55°) relative to the parallel pump beam to minimize polarization effects. Detection was accomplished with an optical multichannel analyzer (OMA), and wavelength calibration was verified with a HeNe laser. The time delays were effected with an optical delay line. The position of Δt = 0, defined as the maximum of the pump-probe cross correlation, was checked both before and after each full scan, and in no case was the drift over the course of the experiment greater than ∼5 fs. The dotted line in each spectrum corresponds to ΔA = 0, and the inset numbers indicate probe beam delay times from Δt = 0. The data were smoothed with a gaussian smoothing function with a bandwidth of 2 nm; a somewhat broader filter was applied to the data between 470 and 480 nm to compensate for an artifact associated with the OMA.

We verified that the spectrum established by Δt = 300 fs corresponds to that of the 3MLCT state by obtaining nanosecond time-resolved data. The details of the laser spectrometer used to collect these data will be published elsewhere (19). The excited-state-ground-state absorption difference spectrum for [Ru(bpy)3]2+ in CH3CN obtained from this experiment is illustrated in Fig. 3A. Excited-state-ground-state isosbestic points (that is, the change in absorbance ΔA = 0) present at 400 and 500 nm and the strong bleach in the 400- to 500-nm range are characteristic of the thermalized 3MLCT state (20). An overlay of the difference spectra collected at Δt = 500 fs and 5 ps with the nanosecond data collected on [Ru(bpy)3]2+ in the region from 440 to 520 nm (Fig. 3B) shows that the three spectra are essentially superimposable within the noise level of the femtosecond data; the mismatch in the spectra at λ < 450 nm is likely because of the decreased signal-to-noise ratio in this region for the femtosecond data (see caption to Fig. 2). The comparison in Fig. 3B in conjunction with the data presented in Fig. 2 provides strong evidence that the excited state probed on the nanosecond time scale and at Δt > 300 fs are the same. Unfortunately, the weak nature of the excited-state absorption for λ > 500 nm makes it difficult to observe the isosbestic cleanly in the femtosecond transient spectra, but the single-wavelength traces verify its presence. This result provides additional support for our assignment, as we consider it extremely unlikely that additional states would have an isosbestic point coincident with the 1A1/3MLCT isosbestic of [Ru(bpy)3]2+ and show the same absorption profile. All of the kinetic and spectroscopic data are therefore consistent with the system being essentially established in the 3MLCT state in ∼300 fs, implying a half-life for the formation of this state on the order of 100 fs.

Fig. 3.

(A) Excited-state-ground-state absorption difference spectrum of [Ru(bpy)3](PF6)2 in CH3CN following nanosecond excitation at 475 nm. The spectrum was obtained point-by-point by plotting the amplitude from a single-exponential fit of the excited- state relaxation data as a function of probe wavelength. The absorptive feature in the ultraviolet (λmax = 370 nm) is due to ligand-based transitions of the bpy chromophore. The bleach in the region from ∼400 to 500 nm is a superposition of absorptions associated with the 3MLCT state and loss of the strong ground state 1MLCT ← 1A1 absorption. Extremely weak absorptive features for λ > 500 nm are due either to bpy transitions or LMCT transitions of the RuIII excited-state chromophore. (B) Overlay of the spectrum of [Ru(bpy)3]2+ obtained at Δt = 500 fs and 5 ps after femtosecond excitation with the absorption difference spectrum of the compound after nanosecond excitation.

The complex evolution of the spectra between Δt = 0 and 300 fs evident from the data in Fig. 2 indicate that there are dynamic processes occurring prior to the establishment of the long-lived excited state. The shift of the transient tracks the formation and thermalization (21) of the long-lived excited state. The overall spectral evolution of the signal is somewhat difficult to interpret in terms of molecular dynamics because it represents a superposition of both ground-state depletion and excited-state absorption or absorptions. Although we anticipate that the ground-state bleach will be instantaneous, excited-state features and hence the superposition spectrum will evolve as the molecule relaxes. The undulations that are apparent superimposed on the bleach signal and at shorter wavelengths do not appear in the solvent blank and therefore must be due to the sample. At present, we are uncertain as to the origin of these features. In terms of solvent contributions to the overall relaxation process, Fleming and co-workers (22) among others (23) have described the ultrafast molecular dynamics of CH3CN in detail and showed that the inertial contribution to the solvent response of CH3CN occurs on the 100-fs time scale. Therefore, given the time scale on which our spectra are changing and that the charge-transfer transition results in the formation of an excited state with a large dipole (24), solvent dynamics are likely having a profound influence on the intramolecular excited-state dynamics and, consequently, the spectral features at early times. In addition, IVR is undoubtedly occurring concurrent with solvent reorganization and ISC and may be contributing to changes in spectral profiles as well.

Although the details pertaining to the earliest time scale response in this system are not yet completely understood, the overall time scale for the formation of the 3MLCT state has important implications for understanding the photoinduced dynamics of these types of systems. The first of these relates to the models which have been developed for describing excited-state relaxation (5, 6). It is tacitly assumed that the fastest process occurring in the course of excited-state relaxation is IVR, then IC, then finally ISC: the rate constants for intramolecular relaxation are therefore ordered as kIVRkICkISC. This anticipated trend is largely based upon the spin-allowed nature of IC versus the spin-forbidden ISC, as well as the expectation that the surface-to-surface crossings characteristic of both IC and ISC will be slower than single-surface processes such as IVR. The net result is a model that invokes a kind of relaxation cascade through the various excited electronic states of the system, with the excitation wave packet sampling the various potential energy surfaces that lie between the Franck-Condon state and the lowest energy excited state of the molecule.

However, the rapid formation of the 3MLCT state after 1MLCT ← 1A1 excitation appears to necessitate motion of the wave packet away from the Franck-Condon region directly to a region of overlap between the initial 1MLCT and final 3MLCT states; the observed time scale dictates that this likely occurs without significant evolution on the initial surface. We therefore suggest that the results of our femtosecond measurements on [Ru(bpy)3]2+ preclude the possibility of there being any well-defined establishment of the wave packet on any potential energy surface other than the lowest energy 3MLCT state in the course of excited-state relaxation. Further support of this notion may come from the absence of vibrational coherence in the single-wavelength time traces, although this result may also be a consequence of the large number of modes in the molecule or that intramolecular relaxation occurs through coupling to high-frequency modes such that the oscillations are not temporally resolvable (25). To the extent that such processes as IVR, IC, and ISC can be distinguished from each other on these time scales (which may not be the case), our data suggest that all of these processes are occurring in concert with each other and with solvent reorganization as the system evolves in time. We believe that this represents a significant change in the conventional model for excited-state relaxation, one in which excited-state evolution is best described in terms of a direct transition from the initial surface to the final surface as opposed to a cascade through various well-defined vibronic levels of the system (26). Such intermediate energy levels may be present, but the wave packet only becomes stationary in the lowest energy vibronic state of the system in the course of excited-state relaxation.

Our results add to a growing body of evidence which shows that non-equilibrated excited states are of fundamental importance in the relaxation dynamics of transition metal complexes. We believe the details of many photophysical processes and indeed the identity and distribution of photoproducts are likely being determined in the earliest moments after photoexcitation (27). This leads to an important final point concerning how one might use this information in the design of molecular-based photolytic assemblies. The idea that dynamics other than intramolecular relaxation can occur prior to excited-state thermalization suggests that it might be possible to access the stored energy in the absorptive state to carry out photoinduced transformations. Such systems would have vastly improved efficiencies because intramolecular energy redistribution is largely responsible for reducing the quantum yields of most photochemical and photophysical processes. This may be what is occurring in many electron donor-acceptor complexes, for example, evidenced by the fact that the initial charge separation is generally too fast to be observed on the picosecond time scale. We suggest that the nature of molecular systems at or near the Franck-Condon state can play an important if not dominant role in photoinduced dynamics, and therefore should be considered in both the analysis of photophysical processes as well as the design of photochemical assemblies that incorporate transition metal complexes.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
View Abstract

Navigate This Article