1. Introduction

Chlorosome has been recognized as the largest light-harvesting complexes in nature, which is composed of hundreds to thousands of bacteriochlorophyll (BChl) c, d, and/or e molecules self-assembled into in highly ordered suprastructures without any direct involvement of a protein scaffold [1,2]. The ultrafast energy transfer ability via pigment-pigment interactions and extremely fast exciton mobility due to its unique aggregate structure make it to be one of the most efficient photosynthetic light-harvesting complexes [3–6]. Therefore, chlorosome has attracted lots of researches about its possible applications on supramolecular photonic, electronic or photocatalytic devices for molecular electronics or solar energy conversion because of its high efficient light-harvesting capability [7,8]. Recently, synthetic chlorins self-aggregated into different structures have been fabricated successfully as good models for light harvesting antennas composed of BChls-c, d, e in green photosynthetic bacteria for further research of the photosynthetic light-harvesting complexes [9]. In addition, it has been reported that the photostability of chlorophyll-a and artificial BChl aggregates could be effectively enhanced by replacing the center magnesium ion with a zinc ion, while their optical properties such as the steady-state absorption spectra, the transient spectra including excited state absorption and induced emission, could be maintained similar with the natural ones [8–10]. Therefore, the zinc photosynthetic aggregates are expected to be more promising in the artificial photosynthesis or photovoltaics systems, and to provide a useful model to investigate the dynamics of photo-induced energy transduction in the chlorosomes.

Due to the significant attention paid to the natural chlorosomes and/or their artificial substitutes over the last few decades, a detailed understanding of electronic and vibrational dynamics simultaneously after photo-excitation is highly desirable for practical reasons, and fast photo-physical processes within chlorosomes and/or chlorosomes-like aggregates have been studied by several groups [5–13]. For example, coherent oscillation in chlorosomes has been elucidated by using the two-dimensional electronic spectroscopy, and the dephasing time of 60 fs at 77 K has been clarified [5]. Ultrafast energy transfer on a sub-100 fs time scale has been resolved due to the exciton diffusion in the highly disordered interior of the chlorosome [11]. Excitation energy transfer in chlorosomes is driven by environmental interaction and the calculated time of energy transfer in intra-chlorosome is from 200 fs to 500 fs [12]. And in the case of chlorosomes-like aggregates, the excitonic relaxation and coherent vibrational has been investigated in stair-shaped zinc chlorin aggregates by our group [13]. However, up to present, only few studies are carried out on the zinc aggregates.

In the present study, the dynamics of coherent molecular vibration coupled to the excitonic transition in a layer-structured zinc chlorin aggregates have been investigated simultaneously. In comparison with the previously studied stair-shaped zinc chlorin aggregates, the chlorophyllous pigment molecules interact more tightly with the multiple interactions in the current layer-structured aggregate [13]. By using the ultrashort laser pulses with pulse duration of 7.1 fs and broad spectrum, the vibrational dynamics in the broad spectrum covered by the laser can be studied from the time-dependent change in the electronic transition probability and/or electronic spectral shape induced by femtosecond lasers. Hence more reliable data can be obtained than step-by-step changing single-wavelength study of excitonic or vibrational relaxations. The wavelength dependencies of the difference absorption and the vibrational amplitude have been found to be similar with each other. This can be interpreted by a transition dipole moment modulated by a dynamic intensity borrowing between the B transitions (S₂) and the Q transition (S₁) through the vibronic interaction mediated by molecular vibrations.

2. Experimental method and sample preparation

Both the pump and probe beams are obtained from a noncollinear optical parametric amplifier (NOPA) seeded by a white-light continuum [14,15]. The laser source of NOPA system is a commercially supplied regenerative amplifier (Spectra Physics, Spitfire), whose central wavelength, pulse duration, repetition rate, and average output power are 800 nm, 50 fs, 5 kHz, and 950 mW, respectively. A pair of prisms with an apex angle of 68° together with chirp mirrors are used to compress the pulse duration to be 7.1 fs. Both the pump and probe pulse cover the spectral range extending from 576 to 771nm as is shown in Fig. 1(b). The pump and probe pulse energies are about 45nJ and 6nJ, respectively. We applied polychromator and multichannel lock-in amplifier to detect the pump-probe signals, which are dispersed by the polychromator (300 groove/mm, 500nm blazed) and finally guided to the 128-channel photodetector by a bundle fiber. The pump-probe signal is measured at the delay time from −200 to 2800fs with a delay time step of 1.2 fs. The spectral resolution of the total system is about 1.5 nm and all experiments are performed at room temperature (293 ± 1 K).

 

Fig. 1 (a) Structure of the Zn Chl Aggregate. (b) the normalized spectrum: NOPA laser spectrum (black line), stationary absorption spectrum (red line), fluorescence spectrum (blue line).

Download Full Size | PDF

The sample studied here is a zinc 3¹-hydroxy-chlorin (exact compound name: zinc 3-devinyl-3-hydroxymethylpyropheophorbide a) aggregate, shown in Fig. 1 (a), which is prepared from naturally occurring chlorophyll a according to reported procedures [16]. The chlorophyllous pigment was mixed with Triton X-100 (Nacalai Tesque, Kyoto, Japan) in methanol (1.2 mL), and the mixture was injected into water (18.8 mL) and shaken vigorously. Final concentrations of the pigment and surfactant were 160 mM and 1.3 mM, respectively [17]. The size of zinc chlorin aggregate is about several hundred nanometers.

To avoid the problem such as sample damage and heating, in the real-time spectroscopy experiment, we adopted a 0.5 mm flow cell together with the micro annular gear pump (mzr-2905; HNP Mikrosysteme GmbH), which provided a flow rate of 15 ml/min. Considering the dispersion induced by the cell windows around the sample, we use a quartz plate whose thickness is the same as the quartz cell before the autocorrelation BBO crystal when measuring the pulse duration. So the pulse duration measured here is consistent with the one at interaction focus within the sample.

3. Results and discussion

The normalized stationary absorption and fluorescence spectra of the zinc chlorin aggregate together with the laser spectrum are shown in the inset of Fig. 1(b). The stationary absorption peaks are around 449nm and 742 nm, corresponding to the B-band and the Q-band of the zinc chlorin aggregates, respectively. And the fluorescence spectrum has a peak around 767 nm, which is almost at the edge of our laser spectrum. Figure 2(a) depicts the two-dimensional plot of the differential absorbance $\Delta A(\lambda ,t)$ as a function of the probe wavelength λ and pump-probe delay time t. The isosbestic point is around 728 nm, while the most intense positive as well as the negative signals are located at ~710nm and ~745nm, respectively. We also note that both the negative and positive signals are asymmetric, which is due to the mixed contribution from the ground state bleaching (GSB), the photo-induced emission (PIE), and photo-induced absorption (PIA) overlapped in the 720-730nm range. All these processes complete with the others resulting in the final positive and negative $\Delta A(\lambda ,t)$ signal.

 

Fig. 2 (a) Two-dimensional plot of the absorbance changes (probe wavelength versus delay time). (b) The time dependence of the absorbance change at five wavelengths.

Download Full Size | PDF

To have a detailed investigation on the excitonic relaxation, the absorption change $\Delta A(\lambda ,t)$ has been fitted by the sum of two exponential-delay functions with signal amplitude parameters a(λ), b(λ), and c(λ) and time constants τ₁ and τ₂ in the following equation.

Figure 3 shows the fitted lifetime constants τ₁ and τ₂ together with the corresponding a(λ), b(λ), and c(λ). It is noteworthy that in the entire range from 650 to 765nm, the pump-probe signal shows a faster recovery in the range of 100 ± 5 fs (τ₁), and then a subsequent slower relaxation process of 863 ± 70 fs (τ₂).

 

Fig. 3 (a) Lifetime τ₁ and τ₂. (b) Time resolved difference absorption spectrum. (c) Spectral a(λ), b(λ), and c(λ).

Download Full Size | PDF

In the following discussion, we will focus on the interpretation of the observed ultrafast excitonic dynamics. As we have discussed above, the fluorescence spectrum is nearly at the edge of our laser probe spectrum, so that the PIE contribution is nearly negligible. Since the lifetime of Q-exciton (S₁-exciton) in Zn chlorin aggregates is on the order of several hundred picoseconds [18], the PIA from Q-state to B-state and the GSB could be assumed to be constant in the present study, and both lifetimes of 100 ± 5 fs and 863 ± 70 fs are too short to be assigned to them. At first glance, the internal conversion (IC) from B-band (S₂) to Q-band (S₁), which was reported to be less than 200 fs [19], might be corresponding to the above fast decay process of ~100fs. However, as shown in Fig. 1(b), the S₂ state is too far to be excited by one-photon excitation using the current pump laser ranging from 576 to 771 nm. Two-photon excitation to S₂ state could also be excluded because the pump-probe signal is nearly linearly dependent on the pump pulse intensity. Therefore, the fast relaxation of ~100 fs cannot be explained by IC. Similarly, the assignment of biexciton state (not the two-exciton or multiexciton state) can also be eliminated because of the linear dependence between the pump-probe signal and the pump laser intensity.

Another possible assignment is the relaxation from multiexciton states (MES), which has also been reported to be in the time scale of ∼100 fs in a stair-structured Zn chlorin aggregates [13,20]. A typical spectral signature for the relaxation from MES to one-exciton state is the shifted spectrum due to the Pauli Exclusion Principle. Specifically, the creation of the excitons in the aggregate requires higher photon energy as the number of the exciton is increased, therefore the corresponding PIA spectrum should be blue-shifted as the number of the exciton is increased [21–23]. As shown in Fig. 3(c), the PIA spectra for the 100 fs decay component a(λ) is blue shifted compared with b(λ), and c(λ). In such short decay time range of 100 fs, the PIA from Q-state to B-state and GSB component could be regarded to be constant because of the rather longer Q-exciton lifetime (several hundreds of ps) in Zn chlorin aggregates [18]. Therefore, the blue-shifted feature in a(λ) compared with b(λ) and c(λ) can be assigned to the blue-shifted PIA spectra due to MES only. In other words, the PIA in the 100 fs component a (λ) corresponds to the transition from MES |n, S₁〉 to |n + 1, S₁〉 (n = 2, 3…), while the PIAs with longer lifetime in b(λ), and c(λ) corresponds to the transition from |1, S₁〉 to |2, S₁〉. Hence, the ultrafast relaxation of ~100 fs is attributed to the fast relaxation from higher MES to the one-exciton state. Similar relaxation process with decay-time constants of 200 fs and 320 fs from MES to |1, S₁〉 have also been observed in the PIC J-aggregates and porphyrin J-aggregates, respectively [22,23].

As for the slower decay with time constant ~800 fs, one possible explanation is the exciton-exciton annihilation (EEA) in aggregates, which has been reported to be in the order of subpicosecond time scale [24,25] and is usually given via a two-step process [25,26]. At first, two excitation beings in the S₁-state have to move close together so that their excitation energy can be used to create a higher excited S_n-state (n>1), and then an ultrafast relaxation from the higher excited S_n-state to S₁-state will be realized via IC process. However, it is unlikely to assign the slow relaxation to the exciton-exciton annihilation, because it appears in both the lower and higher excitation energy ranges and the relaxation is not accompanied by a red shift typical for an annihilation process [24]. Another possible explanation is the relaxation to a nonfluorescent charge transfer (CT) state. In similar aggregate system, protochlorophyllide aggregates (N = 4), the relaxation process with a time constant of 500 fs has been ascribed to the relaxation from exciton states to the nonfluorescent CT state [24]. Therefore, it is reasonable to assign the ~800 fs decay component observed in our layer-structured Zn chlorin aggregates to the transition from Q-exciton state to the dark CT state, which can be intrinsically facilitated by the structure of the aggregates. Finally, the constant component c(λ) can be explained by several relaxations related to the Q-exciton state.

As shown in Fig. 2(b), besides the slow-dynamics component due to excitonic relaxation, the time-dependent absorbance change is composed of the highly oscillating component ($\delta \Delta A(\omega ,t)$) due to molecular vibrations. To get a detailed investigation on the molecular vibrations, Fast Fourier transform (FFT) analysis is used to obtain the vibrational frequencies and its corresponding amplitudes. The two-dimensional plot of the Fourier amplitude against the molecular vibrational frequency and probe wavelength is presented in Fig. 4(a). Molecular vibrations with frequencies of 91,145, 271, 366, 580, 746, 983 and 1227 cm⁻¹ have been detected. The low-frequency (below 200 cm⁻¹) modes are due to intramolecular out-of-plane deformations of the chlorin macrocycle rather than intermolecular modes, as reported by the resonance Raman study [27], which indicates the Q-exciton transition of the aggregates has out-of-plane character. All vibrational modes have the similar spectral features in accordance with the time-resolved difference absorption spectrum in Fig. 2(a) both in the isosbestic point and in the peak positions. To make the point clear, the time-resolved ΔA at 500 fs is shown on the top of Fig. 4(a) for comparison.

 

Fig. 4 (a) Two-dimensional plot of FT amplitude spectra of the pump-probe signal (bottom figure), probe wavelength dependence of difference absorption at 500 fs (top figure). (b) Probe delay time dependence of integrated difference absorption in the Q band. (c) FFT power spectra of the integrated difference absorption.

Download Full Size | PDF

The probe wavelength dependence of the vibrational amplitudes can be explained in the following. In pump probe experiment, the modulation of the absorbance change ($\delta \Delta A(\omega ,t)$) can be expressed as a function of the probe photon energy ω and pump-probe delay time t [28,29].

Here, $\overline{\Delta A(\omega ,t)}$ is the difference absorption spectrum without molecular vibrations, μ is the transition dipole moment, δω is the change of the excitonic transition energy, and δΔω is the change in the bandwidth of ΔA spectrum, ω_v is the vibrational frequency. The second term in Eq. (2) corresponds to the case when the potential minimum is displaced along the normal coordinate with the relevant vibrational frequency on the potential surface of the excited state with respect to the ground state. If the displacement is small, the probe wavelength dependence of coherent vibrations will follow the first derivative of the corresponding absorption spectra [13,29]. The third term presents the effect in remnant case of no displacement between the ground and excited states. Both of second and third term can be excluded in the present results, because only the molecular vibrations resulted from the first term will follow the spectral feature of ΔA spectrum, as shown in Fig. 4(a).

The first term in Eq. (2) represents the modulation of the transition dipole moment μ of the relevant electronic transition, which has deviated from the Condon approximation [26]. The deviation is usually introduced by considering a third electronic state [22], which provides time-dependent contribution varying periodically with the vibrational frequency to the other two states, between which the transition intensity is being monitored. In the case of Q-transition in Zn chlorin aggregates, the B band can be considered as the third electronic state, as illustrated in Fig. 5. As shown in Fig. 4(b), the total signal intensity associated with the Q exciton of the Zn chlorin aggregates evaluated by integrated intensity over the whole spectral region of the Q band given by $\Delta A(t)={\displaystyle {\int }_Q\Delta A(\omega ,t)}d\omega$ is not constant, which indicates the dynamic intensity borrowing between the Q- and B-transition can result in the modulation of the transition dipole moment of Q-band [22,23]. In Fig. 4(c), by using FFT analysis, vibrational modes with frequency of 91, 145, 271, 366, 580, 746, 983 and 1227cm⁻¹ are found to play a significant role in the modulation of the transition dipole moment μ of the Q-transition. All the vibrational modes are consistent with those shown in Fig. 4(a), and most of them are consistent with the frequencies observed in resonance Raman experiment [27]. Therefore, as shown in Fig. 5, the experimental results can be explained in terms of transition dipole moment of Q-band modulated by dynamic intensity borrowing from B transition to Q transition through the vibronic interaction. After comparing the previous work [13] interpreted by the second term in Eq. (2) with the present work, we ascribe the two explanations to the structural difference. The stair-shaped aggregates possess a methoxy (-OCH₃) group at the 3¹-position, which would contribute intermolecular coordination bonding to make a supramolecular J aggregate. In contrast, the present sample possesses a 3¹-hydroxy (-OH) group, which would form additional intermolecular hydrogen bonding with a carbonyl group of the neighboring molecule. Therefore, the chlorophyllous pigment molecules are expected to interact much more tightly with the multiple interactions.

 

Fig. 5 Energy diagram of the zinc chlorin aggregates.

Download Full Size | PDF

4. Conclusion

Using 7.1 fs broadband real-time vibrational laser spectroscopy, both excitonic relaxation and coherent vibrational dynamics in an artificial layer-structured zinc chlorin aggregates have been studied simultaneously. We ascribe the fast decay process of 100 ± 5 fs to the relaxation from MES to the one-exciton state and the slow one of 863 ± 70 fs to the relaxation from Q-exciton state to the dark nonfluorescent CT state. The vibrational amplitudes show a spectral profile of the zeroth derivation of the difference absorption spectrum. It is interpreted by the modulation of the transition dipole moment, which originates from dynamic intensity borrowing between the Q- and B-transitions through vibronic interaction. Hopefully, the vibronic dynamics observed in our zinc chlorin aggregates could provide a key to the solution for complex photosynthesis process coupled with molecular vibrations in nature.