|Home | About | Journals | Submit | Contact Us | Français|
The lifetime of the lowest excited singlet (S1) state of peridinin and many other carbonyl-containing carotenoids and polyenes has been reported depend on the polarity of the solvent. This effect has been attributed to the presence of an intramolecular charge transfer (ICT) state in the manifold of excited states for these molecules. The nature of this ICT state has yet to be elucidated. In the present work, steady-state and ultrafast time-resolved optical spectroscopy have been performed on peridinin and three synthetic analogues, C33-peridinin, C35-peridinin, and C39-peridinin which have different numbers of conjugated carbon-carbon double bonds. Otherwise, the molecules are structurally similar in that they posses the same functional groups. The trends in the positions of the steady-state and transient spectral profiles for this systematic series of molecules allow an assignment of the spectral features to transitions involving the S0, S1, S2 and ICT states. A kinetics analysis reveals the lifetimes of the excited states and the dynamics of their excited state deactivation pathways. The most striking observation in the data is that the lifetime of the ICT state converges to the same value of 10.0 ± 2.0 ps in the polar solvent, methanol, for all the peridinin analogues regardless of the extent of π-electron conjugation. This suggests that the ICT state is highly localized on the lactone ring which is a common structural feature in all the molecules. The data further suggest that the S1 and ICT states behave independently and that the ICT state is populated both from both S1 and S2, the rate and efficiency from S1 being dependent on the length of the π-electron chain of the carotenoid and the solvent polarity.
For polyenes and carotenoids, transitions to and from the ground state, S0, to the lowest-lying excited state, S1, are both symmetry and parity forbidden.1–5 The forbiddenness of the S0 ↔ S1 transitions has been explained theoretically by a model assigning Ag− symmetry to both states, and supported experimentally by the lack of a solvent effect on both the (weak) S1 → S0 fluorescence spectrum and the S1 lifetime.6–8 However, for polyenes and carotenoids possessing a carbonyl functional group, a profound effect of solvent polarity on the lifetime of the lowest excited singlet state has been reported to be as large as one or two orders of magnitude for some carotenoids and apo-carotenal molecules.9–15 It has been proposed that these findings are consistent with the presence of an intramolecular charge transfer (ICT) state due to the presence of the carbonyl group in conjugation with the π-electron system of double bonds.8–10 It has been argued that changes in the position of the ICT state relative to the S1 state rationalize the dependence on solvent polarity of the S1 lifetime.8,10 However, the precise nature of the ICT state has yet to be elucidated. Experiments on carbonyl-containing carotenoids having different extents of π-electron conjugation have shown that the effect of solvent becomes more pronounced with decreasing number of carbon-carbon double bonds, N.12,13,15 Thus, whatever perturbation is responsible for the solvent effect, it becomes more pronounced as the conjugated system of π-electron double bonds is shortened. Proposals for the nature of the ICT state include it being a separate electronic state from S1,9,16–19 quantum mechanically mixed with S1,20,21 or simply S1 itself but possessing a large intrinsic dipole moment due to coupling with S2.22
To explore the nature of the ICT state in carbonyl-containing carotenoids, we synthesized a series of peridinin analogues having different numbers of conjugated carbon-carbon double bonds (Fig. 1). Naturally-occurring peridinin has a C37 carbon skeleton and N=8 with a carbonyl group in the conjugated system. The synthetic analogues are C33-peridinin, which has two fewer double bonds than peridinin, C35-peridinin, which has one less double bond than peridinin, and C39-peridinin, which has one more double bond than peridinin. In all other ways, these molecules are structurally similar to peridinin (Fig. 1) in that they posses the same functional groups. The trends in the spectroscopic and kinetic properties exhibited by this systematic series of peridinin analogues are consistent with a model in which the ICT state behaves independently from S1, and accepts population from S1 at a rate that depends on both the length of the conjugated π-electron chain and the polarity of the solvent.
Peridinin was extracted from Amphidinium carterae cells as previously described.23,24 The synthesis of the C33-, C35- and C39-peridinin analogues will be described elsewhere. The analogues were supplied as dried samples. Prior to the spectroscopic experiments, all molecules were purified using a Millipore Waters 600E high-performance liquid chromatograph (HPLC) employing a YMC-Carotenoid C30 column and a Waters 996 single diode-array detector. The isocratic mobile phase consisted of 87/10/3 v/v/v acetonitrile/methanol/water at a flow rate of 0.8 mL/min. HPLC peaks corresponding to the all-trans molecules were collected, dried under a gentle stream of gaseous nitrogen, and stored at −80°C until used in the spectroscopic experiments. The molecules were dissolved in solvents with increasing polarity, P(ε), but with similar polarizability, P(n): n-hexane (P(ε)=0.229, P(n)=0.228, Fisher Scientific), methyl t-butyl ether (MTBE, P(ε)=0.526, P(n)=0.226, Fisher Scientific), ethyl acetate (P(ε)=0.626, P(n)=0.226, Aldrich Chemicals), 2-propanol (P(ε)=0.852, P(n)=0.230, Fisher Scientific) and methanol (P(ε)=0.913, P(n)=0.203, Aldrich Chemicals).
Steady-state absorption spectra were recorded using a Varian Cary 50 UV-visible spectrophotometer. Steady-state fluorescence measurements were done using a Jobin-Yvon Horiba Fluorolog-3 equipped with a Hamamatsu R928P detector, and a 450 W ozone-free Osram XBO xenon arc lamp. The fluorescence was monitored at a right-angle relative to the excitation. Excitation and emission monochromator slits were set to a bandpass of 5 nm. All fluorescence spectra were corrected for the instrument response profiles using correction factors generated using a standard lamp.
Transient pump-probe absorption experiments were carried out using a femtosecond transient absorption spectrometer system previously described.25 Briefly, the system is based on an amplified, 1 kHz Ti:Sapphire laser (Spectra-Physics). Pump pulses with a duration of ~60 fs were obtained from an OPA-800C optical parametric amplifier (Spectra-Physics). Probe laser pulses were derived from a white light continuum (450 – 800 nm in the visible region, and 800 – 1450 nm in the NIR) generated by a 3 mm Sapphire plate (Ultrafast Systems LLC). For detection in the visible spectral range, a charge-coupled detector S2000 with a 2048 pixel array from Ocean Optics was used. In the NIR region, a 512 pixel array SU-LDV high resolution InGaAs Digital Line Camera from Sensors Unlimited was used. The pump and probe beams were overlapped at the sample at the magic-angle (54.7°) polarization. The signals were averaged over 5 seconds. The samples were pumped as close as possible into the 0-0 vibronic band of the S0 → S2 steady-state absorption spectrum. The pump wavelengths are listed in Table 1. The energy of the pump beam was set to 1 μJ/pulse in a spot size of 1 mm diameter corresponding to an intensity between 3.0 and 3.3 × 1014 photons/cm2 per pulse. The full width at half maximum of the cross correlation in methanol for excitation pulses at 485 nm and probe pulses at 565 nm was determined to be ~170 fs according to the procedure of Ziolek et al.26 and was assumed to be same for other solvents due to the similarity of refractive indices. The pump wavelengths were 470 nm for C33-peridinin and C35-peridinin, 485 nm for peridinin, and 501 nm for C39-peridinin. The samples were adjusted to an optical density of 1.5–2.5 at the excitation wavelength in a 1 cm cuvette, and were then transferred to a 2 mm path length cuvette in the spectrometer, where they were mixed continuously using a magnetic micro-stirrer to prevent photo-degradation. The integrity of the samples was checked by taking steady-state absorption spectra before and after every experiment. Chirp correction of the transient absorption spectra was performed using Surface Explorer (v.1.0.6) software (Ultrafast Systems LCC) by building a dispersion correction curve from a set of initial times of transient signals obtained from single wavelength fits of representative kinetics from a pure solvent sample. The number of principle kinetic components was determined by singular value decomposition. Transient absorption kinetics were analyzed at specific wavelengths by fitting the temporal profiles to a sum of exponentials equation incorporating a Gaussian instrumental response function using the Surface Explorer software.
Fluorescence lifetime measurements were done using a time correlated single photon counting (TCSPC) module installed on a Jobin-Yvon Horiba Fluorolog 3 spectrometer. The system consisted of a single photon counting controller FluoroHub 2.0 (J-Y Horiba), a Hamamatsu R928P detector, and a pulsed NanoLed-470L diode as the excitation light source that provided 466 nm excitation having a pulse duration less than 200 ps. The fluorescence was monitored at 490 nm for C33-peridinin and at 550 nm for C35-peridinin. Fitting of the fluorescence kinetics was carried out using Data Analysis Software DAS version 6.4 (JY Horiba).
Steady-state absorption spectra of peridinin and C33-, C35-, and C39-peridinin analogues in n-hexane, MTBE, ethyl acetate, 2-propanol and methanol are shown in Fig. 2. In all solvents the spectra shift systematically by ~20 nm to longer wavelength with increasing π-electron conjugation chain length. In the non-polar solvent, n-hexane (Fig. 2A), the absorption spectral lineshapes of all the molecules exhibit resolved vibronic bands. The vibronic features become less pronounced with increasing solvent polarity (Figs. 2B–D). The vibronic bands are absent from the absorption spectra taken in the highly polar solvent, methanol (Fig. 2E) resulting in broad unstructured lineshapes for all the molecules except C33-peridinin (thin solid line in Fig. 2E) where small inflections attributable to residual vibronic bands are evident.
Fluorescence spectra of peridinin and C33-, C35-, and C39-peridinin analogues are shown in Fig. 3. For peridinin and the C33- and C35-peridinin analogues, the spectra are broad and substantially red-shifted relative to their respective absorption spectra. This indicates that the emission originates primarily from the S1 state rather than the S2 state for the molecules dissolved in non-polar solvents. The assignment of the emission as S1-like is supported by the fact that the maxima in the spectral traces do not shift substantially when the molecules were dissolved in the highly polarizable solvent, carbon disulfide (P(n) = 0.357), compared to their position in n-hexane (P(n) = 0.229). (See Fig. S1.) This is also consistent with the idea that the S0 → S1 transition has a negligible dipole moment, and consequently experiences only very small interactions with the solvent environment. Vibronic features in the S1 fluorescence spectra are most evident for the molecules dissolved in the non polar solvent, n-hexane (Fig. 3A). These features become better resolved as the π-electron conjugation chain length is increased; e.g. compare C33-peridinin in MTBE (thin solid line in Fig. 3B) with C39-peridinin in the same solvent (dotted line in Fig. 3B). In general, the vibronic features diminish with increasing solvent polarity and additional fluorescence intensity appears at longer wavelength. This additional fluorescence is most noticeable for the molecules dissolved in methanol and is attributable to emission from the ICT state that becomes evident as it is stabilized below S1. (See below.)
C39-peridinin is unique in the series in that its fluorescence spectrum in the visible region (dotted lines in Fig. 3) contains emission originating from both the S1 and S2 states. In n-hexane, emission from the S2 state is more intense than emission from S1 (dotted line in Fig. 3A). In the more polar solvents, the S1-like emission dominates.
Transient absorption spectra of peridinin, and C33-, C35- and C39-peridinin analogues taken in different solvents at various delay times after excitation into the S2 state are shown in Fig. 4. Excitation of the shortest molecule in the series, C33-peridinin (Figs. 4A–E), results in a rapid (~200 fs) build-up of excited state absorption (ESA) in the wavelength range 500–700 nm. The transient absorption spectra taken at the earliest times are broad and asymmetric in all solvents and shift from longer to shorter wavelength within a few hundred femtoseconds. This behavior is exemplified in the spectra of C33-peridinin in ethyl acetate (Fig. 4C) where the 0 ps ESA band exhibits a broad asymmetric lineshape that narrows within 200 ps. In addition, the 0 ps time spectrum of C33-peridinin in ethyl acetate (Fig. 4C) shows a slight dip in the spectrum at ~600 nm. The dip and/or narrowing are noticeable in all the early time traces for C33-peridinin in all solvents. Upon closer examination the narrowing is caused by the signal on the long wavelength side of the dip decaying in a few hundred femtoseconds, while the signal on the short wavelength side of the dip increases in intensity, ultimately resulting in a less broad, symmetric, blue-shifted band; e.g. at around 550 nm in ethyl acetate (Fig. 4C). The precise wavelength maximum of this remaining band depends on the solvent and shifts to shorter wavelength as the solvent polarity increases. Note, it shifts from ~625 nm in n-hexane (green trace in Fig. 4A) to ~520 nm in methanol (green trace in Fig. 4E).
Excitation of the next molecule in the series, C35-peridinin, in all solvents shows an immediate rise of a long wavelength band at ~700 nm (red traces in Figs. 4F–J) that disappears after a few hundred femtoseconds. Subsequently, a broad band between 500 and 700 nm appears (green traces in Figs. 4F–J). In n-hexane this broad band exhibits structural features (green trace in Fig. 4F). In addition, in n-hexane (Fig. 4F), a strong band appears at 490 nm. This band decreases in intensity with increasing solvent polarity, but it can be seen in C35-peridinin as a small shoulder in MTBE (Fig. 4G), as an inflection in ethyl acetate (Fig. 4H) on the short wavelength side of the broad band, and in the longer time (≥ 10 ps) traces for C35-peridinin in 2-propanol (Fig. 4I) and methanol (Fig. 4J).
Note that the 1 ps traces for C33-peridinin and C35-peridinin in the polar solvents (green traces in Figs. 4B–E and 4G–J) dip below the baseline in the region 650–800 nm. This is due to a tail of stimulated emission that appears in the NIR region and originates from the ICT state.8,27 This assignment is supported by transient absorption experiments probed in the NIR region between 800 and 1250 nm (Fig. S2) which show pronounced emission extending below 800 nm for all the molecules dissolved in methanol except C39-peridinin.
Transient absorption spectra of the third molecule in the series, peridinin, have been described in several previous publications.8,9,18,19 Similar to what is observed for C35-peridinin, in all the solvents there is an immediate rise of a long wavelength band between 700 and 800 nm (black traces in Figs. 4K–O) that disappears after a few hundred femtoseconds. Its very short lifetime strongly suggests that this signal is due to an S2 → Sn transition. Subsequently, both a broad, long wavelength band, and a sharper, short wavelength band appear (green traces in Figs. 4K–O). The intensity of the broad long wavelength band increases and shifts to the blue with increasing solvent polarity relative to that of the sharper short wavelength band. As will be discussed in more detail below, the short wavelength band can be attributed to an S1 → Sn transition, and the long wavelength band is assigned to a transition originating from the ICT state whose energy is stabilized in polar solvents.9 This accounts for the blue shift of the ICT → Sn transition with increasing solvent polarity.
Excitation of the longest molecule in the series, C39-peridinin, results in the rapid build-up of a narrow ESA band at short wavelengths and the appearance of broad ESA features at longer wavelengths (green traces in Figs. 4P–T). The longer wavelength features become more prominent as the polarity of the solvent increases. In the non-polar solvent, n-hexane (Figs. 4P), the narrow, short wavelength band dominates the broad, long wavelength features, whereas in the polar solvent, methanol (Fig. 4T), the two signals have comparable intensity.
In order obtain the excited state dynamics of the molecules, transient profiles corresponding to the maxima of the S1 → Sn and ICT → Sn spectra shown in Fig. 4 were fit to a sum of exponentials function. Table 1 summarizes the results of the kinetics analysis. Figure 5 shows the solvent dependence of the decay kinetics of the ESA signals corresponding to the longer wavelength (ICT → Sn) feature in the spectra. The solid lines represent the fits obtained from the kinetics analysis. For C33-peridinin (Fig. 5A) there is a short-lived spike at very early times in the traces. This is due to the rapid appearance and decay of the S2 → Sn absorption band in this wavelength region. At longer times, the decay dynamics of C33-peridinin show an extreme sensitivity to solvent polarity. The lifetime of the excited state is slowest in n-hexane, and faster by a factor of ~500 in methanol. This is the largest solvent effect on the excited state lifetime of a carotenoid yet reported. The lifetime of this same component for the other molecules becomes both faster and less sensitive to solvent polarity as the extent of π-electron conjugation increases. (Note the decreasing time values on the horizontal scale of Figs. 5A–D.) The data show that the threshold at which the kinetics change with solvent polarity increases with increasing length of the analogue. The lifetime of C33-peridinin (Fig. 5A) is 4.2 ns in n-hexane and drops to 200 ps upon changing the solvent to MTBE, then ultimately drops by more than an order of magnitude to 10 ps in methanol. In contrast, the lifetime of C39-peridinin (Fig. 5D) is a constant 41 ps in n-hexane, MTBE and ethyl acetate, then drops to ~30 ps in 2-propanol, finally reaching 9 ps in methanol. Previous studies on peridinin have shown that below a solvent polarity value, P(ε) ≈ 0.5, the lifetime is constant at ~165 ps, whereas above this polarity value, the lifetime decreases linearly with solvent polarity.10 One additional noteworthy point about these kinetics data is that while the lifetimes of the molecules in the non-polar solvent, n-hexane, are strikingly different, varying from 4.2 ns for C33-peridinin to 41 ps for C39-peridinin in n-hexane, the lifetimes of all of the molecules converge to essentially the same value of 10.0 ps (within a range ± 2.0 ps) in the polar solvent, methanol. Previous studies on apo-carotenals and apo-carotenoic acids have shown similar trends and convergences.12–14
Figure 6 presents an overlay of the kinetics of the short (S1 → Sn) and long (ICT → Sn) wavelength ESA signals taken from C35-peridinin, peridinin and C39-peridinin in methanol. The decay kinetics of the S1 → Sn ESA signals were observed to be slower than those associated with the ICT → Sn signals which suggests that the S1 and ICT states not only have distinct spectral profiles but also deactivate independently of one another. The values of the kinetic parameters are summarized in Table 1.
Due to the fact that the lifetime of the lowest excited singlet state of C33-peridinin and C35-peridinin in n-hexane is either comparable to or exceeds the long-time resolution limit of the transient laser spectrometer, the values were also measured using TCSPC fluorescence methods. The results are shown in Fig. 7. The strong emission bands from the S1 states of these molecules in n-hexane (Fig. 3A) facilitated the detection of the fluorescence transient decay signals. The signals were monitored in the (0-0) vibronic bands at 490 nm for C33-peridinin and at 550 nm for C35-peridinin of their S1 steady-state fluorescence spectra (Fig. 3A). The lifetimes obtained from fitting the data to a single exponential decay function were 4.2 ± 0.2 ns for C33-peridinin and 1.0 ± 0.1 ns for C35-peridinin.
In addition to a red-shift in their absorption spectra with increasing π-electron conjugation, the series of peridinin analogues studied here exhibit substantial line-broadening and loss of vibronic resolution when the molecules are dissolved in increasingly polar solvents (Fig. 2). This behavior is typical of carotenoids possessing a carbonyl group in conjugation with the π-electron system of carbon-carbon double bonds, and can be attributed to an increase in the number of conformational isomers formed when the molecules are dissolved in polar solvents. Each of the conformational isomers will have a slightly different absorption spectrum, so that the ensemble average results in a broad lineshape.22,28 Enhanced spectral broadening and the loss of vibronic features with increasing solvent polarity are also observed in the S1 → S0 fluorescence spectra of the molecules (Fig. 3). This can also be accounted for on the basis of enhanced conformational disorder of the molecules in the polar solvents.
With the exception of C39-peridinin, the emission spectra of all the molecules in all solvents are dominated by S1-like fluorescence (Fig. 3) which is typical of short (N ≤ 8) carotenoids. Dominant S1 emission occurs due to a small S2 - S1 energy gap which promotes nonradiative internal conversion from S2 leading to a diminished yield of fluorescence from the S2 state.5,29 For longer carotenoids (N ≥ 10) which have larger S2 - S1 energy gaps, the rate of S2 → S1 internal conversion is decreased, and this enhances the probability of S2 emission. C39-peridinin behaves like an intermediate length (10 ≥ N ≥ 8) carotenoid and displays dual S1 and S2 state emission in all solvents (Figs. 3A–E). Dual emission is typically seen in moderately long carotenoids and is due to the fact that with increasing π-electron chain length, the energy gap between S2 and S1 increases which slows the rate of nonradiative decay from S2 to the point where excited state deactivation by radiative means becomes competitive with S2 → S1 internal conversion. However, C39-peridinin is unique in that the yield of emission from S2 relative to S1 changes dramatically with the polarity of the solvent (dotted lines in Fig. 3). C39-peridinin has a higher amplitude of S2 emission in the non-polar solvent, n-hexane. In more polar solvents, the S1-like emission dominates. This indicates that the rate of internal conversion from S2 increases with solvent polarity leading to a reduced emission yield relative to that of S1. The small dependence on solvent polarity of the energies of the S1 and S2 states is not sufficient to account for this distinctive effect. Most likely the observation can be traced to the solvent-induced modulation of the energy of the ICT state near S1 and S2 because the energy of the ICT state does depend strongly on solvent polarity. A diminished yield of S2 fluorescence from C33-peridinin with increasing solvent polarity will be seen if the rate of populating the ICT state from S2 also increases with solvent polarity. This may seem counterintuitive since the energy gap between S2 and the ICT state is likely to become larger as the ICT state is stabilized with increasing solvent polarity. However, the effect can be rationalized if the controlling factor in depopulating S2 via the ICT state is not the magnitude of the energy gap between these states, but rather the size of the apparent activation barrier at the crossing point between the S2 and ICT potential energy surfaces. In this case, increasing solvent polarity would stabilize the ICT state, lower the apparent activation barrier for population transfer from S2 to the ICT state, and lead to faster nonradiative decay from S2 which would diminish the relative yield of S2 emission as observed.
The kinetics of the S1 state for these molecules have been obtained using TCSPC techniques (for C33-peridinin and C35-peridinin) and single wavelength analyses of the transient absorption data. In a given solvent, the S1 lifetime of the molecules was found to decrease with increasing N. For example, in n-hexane, the S1 lifetime for the series of molecules decreased from 4.2 ns to 41 ps in going from C33-peridinin to C39-peridinin. This is attributable to the fact that the S1 energy decreases with increasing N. It has been demonstrated for carotenoids30 that rate of internal conversion increases exponentially with decreasing energy gap between the S1 and S0 states in accordance with the energy gap law for radiationless transitions.31 In addition, the present data show clearly that the shorter the peridinin analogue, the stronger the effect of polarity on the excited state lifetime. This effect is central to the question of the origin of the ICT state whose properties can be explained by a consideration of its position relative to the energies of the S1 and S2 states.
The most striking observation in the kinetics data is that the lifetime of the ICT state is essentially the same (10.0 ± 2.0 ps) for all four molecules examined in the polar solvent, methanol. This convergence to a common value for these four molecules, which differ in their extents of π-electron conjugation, indicates that the ICT state must be localized away from the extended π-electron chain and on or near the lactone-ring, which remains a constant structural component for all of the molecules. The obligatory requirement of the carbonyl group for inducing the effect of solvent on the excited state lifetime of carotenoids has been demonstrated previously,9,10 and also has been reported for apo-carotenals and apo-carotenoic acids.12–14 The lack of sensitivity to the extent of the π-electron conjugation seen in these series of molecules suggests not only that the ICT state is highly localized, but also argues that the ICT state decays independently from the S1 state whose energy changes by ~2000 cm−1 with each incremental change in N. This idea is supported by the observation that the S1 → Sn and ICT → Sn transitions display different spectra (Fig. 4) and decay kinetics (Fig. 6, Table 1). However, it cannot be stated for certain whether these differences result from the S1 and ICT states being uncoupled or from a situation where the two states are strongly coupled and correspond to different minima on the same potential energy surface.
The trends observed in the spectral features and dynamics for this series of molecules can be accounted for on the basis of changes in the relative energies of the S1, S2 and ICT excited states with changing N and solvent polarity. As the π-electron conjugation chain length of the molecules increase, both the S1 and S2 states decrease in energy. As the polarity of the solvent increases from n-hexane to methanol, the ICT state is stabilized. Thus, changes in chain length and solvent polarity can lead to differences in the population of the S1 and ICT states evidenced by the appearance of different ESA signal intensities associated with transitions from these states (Fig. 4).
From the ESA traces presented for all four molecules and shown in Fig. 4, it is clear that C39-peridinin in methanol (blue trace in Fig. 4T), peridinin in MTBE (blue trace Fig. 4L), and C35-peridinin in n-hexane (red trace in Fig. 4F) have very similar lineshapes in that they show comparable amplitudes associated with the S1 → Sn and ICT → Sn transitions. For all other molecule/solvent pairings displayed in Fig. 4, features associated with one or the other of these transitions dominates the profiles suggesting that either the S1 state or the ICT state is lower in energy than the other. Under conditions where the ICT state lies below the S1 state, the S1 → Sn spectrum is broadened and the lifetime of the S1 state is shortened (Table 1) due to transfer of population from S1 to the ICT state. Under conditions where the ICT state lies above S1, little if any effect on the S1 → Sn spectrum and S1 lifetime is seen. Therefore, it is proposed that for the three special molecule/solvent pairs, C39-peridinin in methanol (blue trace in Fig. 4T), peridinin in MTBE (blue trace Fig. 4L), and C35-peridinin in n-hexane (blue trace in Fig. 4F), the ICT and S1 state energies are very close in energy. Because the S1 energies of the molecules can be determined from the spectral origins of the S1 → S0 fluorescence spectra (Fig. 3), the ICT state energies of the three molecules in these solvents are then simultaneously determined. A plot of these three values versus solvent polarity (open circles in Fig. 8) yields a precisely straight line along which the entire range of energy values for the ICT states of all four molecules in all solvents including MTBE and 2-propanol (filled circles in Fig. 8) can be determined. The plot shows that the highest energy the ICT state can achieve is 18,300 cm−1 in the non-polar solvent, n-hexane. The minimum energy of the ICT state is 15,150 cm−1 in the polar solvent, methanol. A plot of the energies of the S1 and S2 states of the four molecules determined from their steady-state absorption (Fig. 2) and fluorescence (Fig. 3) spectra shows that the S1 and S2 state energies follow a linear dependence according to the function, 1/(1±2N) (Fig. 9). Moreover, the S1 state energy decreases more rapidly with increasing N than does the S2 state energy. Comparing the range of possible ICT state energies (obtained from Fig. 8 and represented by the shaded region in Fig. 9) with the energies of the S1 and S2 states shows clearly that for the shortest molecule in the series, C33-peridinin, the ICT state energy is always lower than the S1 state energy of this molecule regardless of solvent polarity. On the other extreme, for the longest molecule examined here, C39-peridinin, the ICT state energy lies above its S1 state energy for all solvents except the most polar solvent, methanol, in which the two states are isoenergetic. For C35-peridinin the S1/ICT energy equivalence point is achieved when the molecule is dissolved in the non-polar solvent, n-hexane. For peridinin, the ICT and S1 states will have roughly equivalent energies in solvents having a polarity index midway between n-hexane and methanol.
Figure 9 also shows S3 state energies (open squares) obtained from the positions of the cis-peaks in the steady-state absorption spectra in methanol (Fig. S3) fit to a straight line. Cis-peaks corresponding to S0 → S3 transitions are readily observed in the spectra of cis-carotenoids due to less stringent selection rules for light absorption for molecules having undergone trans-to-cis isomerization.32,33 The filled triangles in Fig. 9 indicate the sum of the energy of the ICT state in methanol (15,150 cm−1) and the energy of the ICT → Sn transition observed in the excited state absorption spectra also taken in methanol (Figs. 4E, J, O and T). The fact that these two sets of points (open squares and filled triangles) are in such good agreement indicates that the end state of the ICT → Sn transition is S3. The strong allowedness of the ICT → Sn (S3) transition (Fig. 4) can then be thought of as deriving from asymmetry in the wavefunction of the ICT state that relaxes the selection rules for light absorption between the ICT and S3 states, analogous to what occurs for the S0 → S3 transition.
This analysis provides an explanation why C33-perinidin shows only very little evidence in its transient absorption spectra (Figs. 4A–E) of transitions associated with the S1 state. The ICT state in C33-peridinin is so low in energy in all solvents, that not only is it rapidly populated directly from the S2 state after photo-excitation, but it also depopulates the S1 state very rapidly. The ESA spectra obtained shown in Figs. 4A–E and are consistent with this conclusion. The data indicate that the very short-lived S2 → Sn transition appears simultaneously with the ICT → Sn transition showing that the ICT state is populated within the time of the laser excitation pulse. After the S2 state decays in < 170 fs, the ICT state remains, but the rapid population of the ICT state has left it vibronically hot as evidenced by the fact that after ~1 ps depending on the solvent, the ICT → Sn transition narrows and blue-shifts (red and blue traces in Figs. 4A–E). The remaining, vibronically-relaxed ICT → Sn band (blue traces in Figs. 4A–E) persists for a time ranging from > 1 ns in the nonpolar solvent, n-hexane, to 10 ps in the polar solvent, methanol. In n-hexane, MTBE and ethyl acetate, an additional minor decay component is required to achieve a satisfactory fit to the datasets for C33-peridinin. The origin of this component remains unclear at this time, and it was not needed for a good fit to the kinetic data from the other molecules.
Like C33-peridinin, the S2 state of C39-peridinin decays directly into both the S1 and the ICT states in < 170 fs. This is clear from the rapid rise of ESA associated with both states. In non-polar solvents there occurs a slight narrowing and blue-shifting of the resulting S1 → Sn and ICT → Sn transitions which can be attributed to vibronic cooling. As the solvent polarity increases (Figs. 4P–T), features associated with the ICT state are seen, and in methanol (Fig. 4T) this kinetic component decays in 9 ps (Table 1). Similar conclusions regarding the kinetic behavior of C35-peridinin and peridinin can be drawn. For these two molecules, it is clear from the ESA traces presented in Figs. 4F–O and the data in Table 1 that in most cases the S2 state decays in < 170 fs and directly populates both the S1 and the ICT states. Subsequently, ESA profiles associated with either the S1 → Sn or ICT → Sn transitions are seen depending on the relative energy ordering of the states.
The energies of the transitions and the apparent activation barriers for the transfer of population between the states are summarized in a series of potential energy surface diagrams given in Fig. 10. The figure is drawn as if the states are quantum-mechanically uncoupled, but the data presented here do not distinguish between this possibility and the alternative that the states are strongly coupled and represent different minima on the same potential energy surface. For C33-peridinin in both nonpolar and polar solvents, the apparent activation barrier for the transfer of population from S1 to the ICT state is likely to be very small. This would explain why there is little evidence in the transient absorption spectra (Figs. 4A–E) of features associated with an S1 → Sn transition. On the other extreme for C39-peridinin, efficient transfer of population from S1 to the ICT state occurs only when the molecule is dissolved in highly polar solvents and when the two states are close in energy.
In the present work, steady-state and ultrafast time-resolved optical spectroscopy have been performed on peridinin and three synthetic analogues, C33-peridinin, C35-peridinin, and C39-peridinin which differ in their extents of π-electron conjugation. The trends in the positions of the steady-state and transient spectral profiles for this systematic series of molecules have allowed an assignment of the spectral features to transitions involving the various excited electronic singlet states including the ICT state. A kinetics analysis revealed that the dependence on solvent polarity of the excited state lifetime gets stronger as the extent of π-electron conjugation of the carotenoid is reduced. However, the most striking observation in these data is that the lifetime of the ICT state converges to the same value of 10.0 ± 2.0 ps in the polar solvent, methanol, for all the peridinin analogues regardless of the extent of π-electron conjugation. This suggests that the ICT state is localized on the lactone ring, which is the common, carbonyl-containing, structural feature in all four molecules. In addition, the kinetics data are best explained assuming the S1 and ICT states deactivate independently, the rate of which is found here to depend strongly on both solvent polarity and the extent of π-electron conjugation of the carotenoid.
The authors wish to thank Professors Robert Birge and Tomáš Polívka for many useful discussions. This work has been supported in the laboratory of HAF by grants from the National Institutes of Health (GM-30353), the National Science Foundation and the University of Connecticut Research Foundation. We also thank Dr. Thomas Netscher of DSM Nutritional Products, Ltd., for the donation of (−)-actinol. This work has been supported in the laboratory of SK by a Grant-in-Aid for Science Research on Priority Areas 16073222 from the Ministry of Education, Culture, Sports, Science and Technology, and Matching Fund Subsidy for a Private University, Japan.
Supporting Information available: Overlay of the fluorescence spectra of C33-peridinin, C35-peridinin and peridinin taken at room temperature in carbon disulfide and n-hexane, NIR transient spectra of all four analogues taken at room temperature in methanol, and steady-state absorption spectra of C33-peridinin, C35-peridinin, peridinin and C39-peridinin taken at room temperature in methanol, extended to high energy to show the “cis-peak” region between 27,000 and 40,000 cm−1. This material is available free of charge via the Internet at http://pubs.acs.org.