Letter pubs.acs.org/JPCL
Quantum Beats and Phase Shifts in Two-Dimensional Electronic Spectra of Zinc Naphthalocyanine Monomer and Aggregate Ki-Hee Song,† Munui Gu,§ Min-Seok Kim,†,‡,§ Hyeok-Jun Kwon,§ Hanju Rhee,*,† Hogyu Han,*,§ and Minhaeng Cho*,‡,§ †
Space-Time Resolved Molecular Imaging Research Team, Korea Basic Science Institute (KBSI), Seoul 136-713, Korea Center for Molecular Spectroscopy and Dynamics, Institute for Basic Science (IBS), Korea University, Seoul 136-701, Republic of Korea § Department of Chemistry, Korea University, Seoul 136-701, Korea ‡
S Supporting Information *
ABSTRACT: The origin of quantum coherence in two-dimensional (2D) electronic spectra of molecular aggregates and light-harvesting complexes still remains an open question. In particular, it could be challenging to distinguish between electronic and vibrational coherences for a coupled system, where both degrees of freedom can be simultaneously excited. In this Letter, we examine quantum beats in the 2D spectra of zinc naphthalocyanine (ZnNc) aggregate and monomer, and compare their characteristic features in terms of the frequency and relative phase of diagonal and off-diagonal amplitude oscillations. The longlasting oscillating components (>1 ps) at 600−700 cm−1 observed in both the aggregate and monomer are found to be attributed to the vibrational coherence. The wide phase variations of the 2D spectral amplitude oscillations are observed not just in the aggregate but also in the monomer state. This suggests that the unusual 90° phase shift may be attributed to neither quantum population-tocoherence transfer nor vibronic exciton coupling.
Q
In addition to the remarkably long lifetime of quantum beats in light-harvesting complexes, the significance of the phase shift (Δϕ) between oscillatory components at different frequency regions in a given 2D spectrum has been highlighted to account for the so-called quantum transport mechanism in the FennaMatthews-Olson (FMO) complex.5,8 It was suggested that an unusual 90° phase shift between diagonal and cross peak oscillations, which cannot simply be explained by taking all the possible Feynman pathways into account, would be a consequence of the coupling between population and coherence of the delocalized exciton states. Furthermore, they argued that the surrounding protein matrices around chlorophyll pigments could play an active role in achieving the population-to-coherence transfer, which can in turn give rise to the prolonged electronic coherence and allows the system to search the most efficient energy transport pathway. Despite that this could be a plausible explanation with highlighting only the electronic couplings between multiple pigments, the interplay between the electronic and nuclear degrees of freedom was not fully taken into consideration.7,14,18−20 A recent theoretical study revealed that such 90° phase shift between 2D peak oscillations could arise from a
uantum coherence in two-dimensional (2D) electronic spectroscopy (ES) of photosynthetic light harvesting complex (LHC) has drawn great attention over the past few years.1−8 Notably, it was suggested that the long-lived electronic coherence (EC) could be direct evidence of coherent wavelike energy transfer between constituent pigments, and play an important part in achieving long-range exciton migration from the LHC to the reaction center (RC) with remarkable efficiency.1,5 This argument is based on the fact that the observed amplitude oscillations of the 2D peaks originate from a coherent superposition of multiple electronic exciton states resulting from interpigment electronic couplings. However, there have been debates on the origin of such quantum beats in the 2D ES, since even a single chlorophyll pigment itself can have vibronically coupled nuclear degrees of freedom (DOF) in the low frequency region.9 Vibrational coherence (VC), created by the coherent excitation of those vibrational DOF on the same pigment, can also show similar long-lived oscillatory signatures, which has been demonstrated in the 2D ES of a monomeric dye molecule.10−12 Although various approaches aiming at identifying purely EC signatures have been proposed theoretically,13,14 the origin of the observed 2D spectral amplitude oscillations still remains an open question for coupled multichromophore systems,2,4,15−17 where both electronic and vibronic transitions can be simultaneously excited. © XXXX American Chemical Society
Received: September 14, 2015 Accepted: October 14, 2015
4314
DOI: 10.1021/acs.jpclett.5b02030 J. Phys. Chem. Lett. 2015, 6, 4314−4318
Letter
The Journal of Physical Chemistry Letters
Figure 1. (a) The molecular structure of ZnNc and (b) the energy level schemes of the monomer and aggregate. (c) The absorption spectra of the monomer (black) and aggregate (red), and the pulse spectrum (green). (d) The 2D electronic spectra of the monomer (upper) and aggregate (lower) at T = 90 fs. (e) The amplitude oscillations of the CP21 peaks (red arrows in (d)) of the monomer (upper) and aggregate (lower) as a function of T. (f) The FT spectra of the CP21 peak oscillations shown in (e) for the monomer (top) and aggregate (middle) and the spontaneous Raman spectrum (bottom) of the ZnNc powder.
spectra of the monomer and aggregate at ∼50 μM are compared in Figure 1c. It is noted that the absorption peaks of the aggregate are slightly blue-shifted from those of the monomer probably due to a different solvation effect, i.e., solvatochromism. The high energy peak of the aggregate (red) at 13900 cm−1 increasing with the ZnNc concentration (Figure S1) can be assigned to one of the exciton bands (gv0−e″v0 transition in Figure 1b) that is blue-shifted from the 0−0 peak (gv0−ev0 transition) of the monomer at 12950 cm−1 due to faceto-face (H-type) interaction. Considering the overall blue-shift of the peaks shown in the aggregate, the 0−1 peak (gv0−e′v1 transition) involving the lower electronic excited state of the aggregate is likely to be blue-shifted a bit from that (gv0−ev1 transition) of the monomer, which appears at 13660 cm−1 (black), and partly overlaps with the higher exciton 0−0 peak of the aggregate at 13900 cm−1 (gv0−e″v0 transition). Consequently, a significant vibronic exciton coupling can be induced as theoretically suggested by Perlik et al. for the vibronic dimer.14 For 2D ES experiments, we implemented the method based on a two-beam geometry using an optical pulse shaper (see Figure S4 for the detailed optical setup) instead of a conventional boxcar configuration where three input pulses (k1, k2, k3) are noncollinearly incident onto the sample.23,24 Briefly, a 30 fs laser pulse whose spectrum covers the coupled exciton and vibronic bands is split into two beams, which are used as a pump and a probe, respectively. By the acousto-optic pulse shaper (Dazzler, Fastlite), the pump pulse is duplicated (k1, k2) in a single beam, and they are separated in time by t1 (coherence time). Then, they interact with the other probe beam (k3) that is time-delayed from k2 by T (waiting time) at the sample. Since the replica pump pulses are identical (k1 = k2) and collinearly propagate, a phase-matching photon echo signal (ks) is created into the same direction as k3 (//ks = −k1
vibronic exciton coupling without relying on the population-tocoherence transfer argument previously invoked to interpret experimental observations for FMO.14 In a vibronic dimer where the vibronic levels of two different electronic states are nearly resonant, it was proposed that the induced vibronic exciton coupling can lead to a deviation from Δϕ = 0° or 180°, and it can even allow Δϕ to have arbitrary values depending on the electronic coupling strength (J). Therefore, they provided a compelling argument that the population-to-coherence transfer involving the purely electronic states may not be the only scenario to explain the observed 90° phase difference in the 2D ES. Nonetheless, both studies on phase shifts of quantum coherence oscillations considered only electronically coupled chromophores with or without vibrational transitions. In this regard, a comparative study of monomeric and coupled multichromophore systems with the same pigment would be essential to gain a better understanding of the vibronic exciton coupling effect on the coherent oscillations and their phase shifts in the 2D ES. Here, we examine the 2D ES of two different molecular systems, a monomer without interchromophore electronic coupling (purely vibronic) and its aggregate whose excitonic coupling energies and monomeric vibrational transition energies significantly overlap each other (vibronic exciton coupled). The chromophore is zinc naphthalocyanine (ZnNc) derivative with four peripheral binaphtholate groups (Figure 1a). The chromophores are dissolved in benzonitrile, where they are in a monomeric state, whereas they form aggregates in tetrahydrofuran (THF).21,22 We confirmed this by examining the concentration-dependent normalized absorption spectra of ZnNc in THF and benzonitrile. Note that the spectral line shape does not change with the ZnNc concentration in benzonitrile, whereas it changes dramatically upon increasing ZnNc concentration in THF (see Figure S1). The absorption 4315
DOI: 10.1021/acs.jpclett.5b02030 J. Phys. Chem. Lett. 2015, 6, 4314−4318
Letter
The Journal of Physical Chemistry Letters
decay faster than the time resolution of our measurement. In general, the EC between two exciton states can be rapidly destroyed by two dephasing processes: population and pure phase relaxations of those superposition states. A recent 2D ES study of a covalently linked molecular dimer revealed that the intramolecular radiationless transition (population transfer) from the upper to lower exciton states is the dominant exciton relaxation dynamics in such a strongly coupled chromophore system,27 in contrast to photosynthetic proteins where the electronic coupling of the embedded pigments is relatively weak. In that experiment, the population transfer between the two exciton states, which occurs very fast in 10 ps) without ultrafast decay components, but there is a fast rise (∼70 fs) in the off-diagonal region (see Figure S3). Nonetheless, it is believed that the possibility of ultrafast radiationless transition between exciton states in a strongly coupled multichromophore system needs to be further investigated. On closer inspection of the cross peak oscillations of the monomer and aggregate in Figure 1e, however, there is a small yet noticeable difference between their periods. Note that the two oscillations initially close to in-phase have an opposite phase at a longer time (see the blue vertical lines). Second, the two oscillation frequencies of the monomeric ZnNc, which are 650 and 710 cm−1 (see Figure 1f), are slightly higher than those of the aggregate (620 and 680 cm−1) by ∼30 cm−1. The reason that the latter values are closer to the Raman shift frequencies of the powder sample than the former values of monomers in benzonitrile are is probably because the H-aggregate dominantly forms in the ZnNc powder as well as in THF. The ∼30 cm−1 frequency difference between the aggregate and monomer can simply be explained by vibrational solvatochromic frequency shift.28,29 In general, the vibrational frequency of a polyatomic molecule in condensed phase can shift upon solvation due to changes in solute−solvent interactions. In the aggregate, one ZnNc is surrounded by other ZnNc’s and THF molecules, whereas only by benzonitrile molecules in the monomer. As a result, each ZnNc experiences different local electrostatic fields and intermolecular potentials in the aggregate and monomer, which may be responsible for the frequency shift of ∼30 cm−1 of their 2D peak oscillations observed here. In the previous work on the vibronic dimer, it was theoretically predicted that a wide phase variation of 2D peak oscillations could arise from an increased overlap of neighboring peaks with opposite phase by the vibronic exciton coupling.14 This implies that the 90° phase difference observed in the FMO can be explained simply by vibronic exciton coupling without invoking population-to-coherence transfer considering the purely exciton states. To address this issue, we examine the phase property of the 680 cm−1 oscillating component of several diagonal and off-diagonal spectral amplitudes on the 2D spectra of the ZnNc aggregate. Note that the ∼680 cm−1 frequency component, which can be definitely attributed to the oscillation by VC as described above, is found in the entire region of the aggregate 2D spectrum displayed in Figure 1d. As ω1 varies from 13412 to 13912 cm−1 with a fixed ω3 = 13112 cm−1 (horizontal off-diagonal line, H,
+ k2 + k3) and thus self-heterodyne detected with the transmitted k3. The heterodyned signal spectrum normalized by the probe, S(t1, T, ω3), is measured at each t1 with a CCD detector, where the pulse shaper and detector are both synchronized with the laser (500 Hz repetition rate). Then, the Fourier transformation of S(t1, T, ω3) along t1 finally yields the 2D spectrum S(ω1, T, ω3). One advantage of using the programmable pulse shaper is the rapid t1-scan capability,25 which speeds up the data acquisition significantly so that a single 2D spectrum at a given T is obtained just in a few seconds. In our two-beam geometry where k1 and k2 collinearly propagate, due to the indistinguishability of the time ordering between k1 and k2, the emitted signal carries both the rephasing and nonrephasing responses, consequently yielding the absorptive signal only.26 Figure 1d depicts the real parts of the absorptive 2D spectra at T = 90 fs for the monomer (upper) and aggregate (lower). In the aggregate 2D spectrum, the upper left cross peak (ω1 = 13100 cm−1, ω3 = 13900 cm−1) instantaneously appears at initial times upon photoexcitation, indicating that the corresponding cross peak is attributed to the exciton coupling in the aggregate, rather than to a dynamical exchange between the monomer and aggregate. Although the 0−1 vibronic peak in the monomer absorption is not prominently shown at the upper cross peak probably due to a cancellation effect by the excited state absorption, it is clearly seen at the lower cross peak together with a weak exciton peak slightly blue-shifted from it along ω1. The similar vibronic features are observed in the monomer 2D spectrum (top of Figure 1d) except for the absence of the exciton coupling signal. It should be noted that the small vibronic peak of the lower electronic state, which is significantly covered by the higher exciton band in the 1D absorption spectrum, is well resolved in the cross peaks of the 2D spectrum. The amplitude oscillations of the 2D cross peaks (CP21, red arrows in Figure 1d) of the aggregate and monomer as a function of T are directly compared in Figure 1e. In both cases, slow beatings with a temporal period of ∼400 fs are observed in the highly oscillating signals that persist up to >1 ps. Their Fourier transform (FT) spectra show that two components with slightly different frequencies of around 600−700 cm−1 mainly contribute to the oscillations. Not only the corresponding peak frequencies of the monomer and aggregate but also the relative intensity ratios of the high to low frequency peaks in the FT spectra are similar to each other (Figure 1f). This led us to conclude that the long-lasting oscillations originate primarily from the VC of the localized ZnNc moiety, rather than the EC between the delocalized exciton states over the entire multiple chromophores of the aggregate, for the following reasons: (1) Their frequencies are not consistent with the energy difference (∼820 cm−1) between the two electronic absorption peaks of the aggregate, expected as an oscillating frequency of the EC. (2) The lifetime of the oscillation is quite long (>1 ps) compared with the typical electronic dephasing time (
