Carotenoid radical formation after multi-photon excitation of 8′-apo-β-carotenal

Abstract

Carotenoids containing a conjugated C=O group exhibit complex excited-state dynamics that are influenced by solvent polarity due to the involvement of an intramolecular charge transfer (ICT) state. Our study explores the excited-state behavior of 8′-apo-β-carotenal under multi-photon excitation conditions. Using near-infrared (1300 nm) multi-photon excitation, we observe the formation of a cation radical of 8′-apo-β-carotenal, a process distinct from those following one-photon visible or UV excitation. Our findings suggest that this radical formation results from multi-photon excitation involving a higher-lying dark state, supported by intensity-dependent experiments. This work demonstrates that radical generation is a characteristic of this higher excited state and is not produced during relaxation from the S₁/ICT state. The results open new pathways for understanding carotenoid radical formation mechanisms under intense multi-photon excitation.

---

Introduction

Keto-carotenoids containing a conjugated C=O group have been extensively studied since the seminal study by Bautista et al., who demonstrated that the excited-state dynamics of these carotenoids depend on solvent polarity.¹ This dependence was attributed to the presence of an intramolecular charge transfer (ICT) state that is coupled to the lowest carotenoid excited state, S₁, forming an S₁/ICT state, thereby modifying the lifetime of the lowest excited state.^1–5 The S₁/ICT lifetime shortening in polar solvents is always accompanied by new spectral bands in transient absorption spectra associated with the ICT state: the ICT excited state absorption band that is red shifted from the main S₁ → S_N transition^1,6 and ICT stimulated emission band in the near-IR spectral region.² Polarity-induced effects were also reported for the keto-carotenoid absorption band associated with the S₀ → S₂ transition. While in non-polar solvents the characteristic three-peak absorption spectrum with well-resolved vibrational bands is observed, in polar solvents the resolution of the vibrational bands is significantly diminished, and the absorption band is broadened.^3,7–10

While the first reports on polarity-induced effects in keto-carotenoids targeted the biologically relevant keto-carotenoids peridinin and fucoxanthin, which occur in photosynthetic antenna of dinoflagellates and diatoms,^11,12 another prototypical keto-carotenoid is 8′-apo-β-carotenal, which also exhibits a strong polarity-induced effect¹³ and is readily available as it is used as a food colorant.^14–16 Excited state processes of 8′-apo-β-carotenal were first reported in 1986¹⁷ and detailed exploration of the dependence of the excited state dynamics on solvent polarity was provided later:^13,18–20 the lifetime of the lowest excited state in nonpolar solvents is ∼25 ps, and it is shortened to ∼8 ps in polar solvents such as methanol or acetonitrile. In polar solvents, a pronounced ICT band appears around 650 nm, reaching about 70–80% of the intensity of the S₁ → S_N band peaking at ∼560 nm.^13,20

All information about excited-state properties of 8′-apo-β-carotenal is obtained from transient absorption experiments exciting the S₂ state because of the forbidden nature of the S₀ → S₁ transition, preventing direct excitation of the S₁ (or ICT) state by one-photon transition. Thus, the S₁/ICT state, which is typically described as a double-well potential surface at which the S₁-like and ICT-like minima are separated by a polarity-dependent barrier,⁵ is consistently formed through relaxation from the S₂ state. This complicates data interpretation as the S₁/ICT is populated with excess energy and the distribution of population between the S₁ and ICT minima may depend on the excitation wavelength.²¹ Thus, it is of interest to design experiments in which the S₁/ICT could be excited directly.

Direct excitation of the S₁ state in the transient absorption experiment can be in principle achieved via two-photon excitation as demonstrated in a few experiments reported earlier.^22–26 These studies showed that excitation in the 1100–1300 nm spectral region generates typical transient absorption signals associated with the S₁ state, proving that the S₁ state is indeed possible to excite directly via two-photon excitation. These experiments also identified subtle differences between one- and two-photon excited S₁ signals, such as generation of carotenoid triplets or radicals in some cases.^23–25 Detailed excitation intensity-dependent studies even demonstrated that multi-photon excitation is likely involved in the generation of signals after intense excitation in the near-IR spectral region.²³ A two-photon excited transient absorption experiment was also reported for the keto-carotenoid fucoxanthin,²⁷ showing that tuning the two-photon excitation wavelength has the ability to selectively excite either the S₁ or ICT part of the S₁/ICT potential surface.

Here, we report on the near-IR (1300 nm) excitation study of the keto-carotenoid 8′-apo-β-carotenal. Contrary to previous experiments, we have extended the probing window to the near-IR spectral region, which has allowed us to identify and monitor the formation of a cation radical of 8′-apo-β-carotenal. Detailed intensity dependence analysis demonstrated that the carotenoid radical is generated via multi-photon excitation of a previously unidentified dark excited state.

Materials and methods

Experimental details

A Spitfire Ace regenerative amplifier system (Spectra Physics), seeded with a Ti:sapphire oscillator (MaiTai SP, Spectra Physics) and pumped by an Nd:YLF laser (Empower 30, Spectra Physics), producing 4 mJ pulses centered at 800 nm with a 1 kHz repetition rate and a pulse duration of ∼100 fs, was used for generating excitation (pump) and probe pulses. The near-IR excitation beam was generated through parametric amplification (TOPAS Prime, Light Conversion). The pump beam was focused on the sample by a lens with a focal length of 500 mm. The pump intensity was set to 7 × 10¹⁶ photons per pulse cm², corresponding to 15 mW excitation power at 1300 nm. A fraction of the 800 nm beam from the Spitfire Ace was sent to the home-built non-collinear optical parametric amplifier in order to produce a 1250 nm beam used to generate the white-light supercontinuum (WLC) probe beam by focusing it to a 2 mm sapphire plate. The polarization between pump and probe beams was set parallel by a λ/2 waveplate. The signal-to-noise ratio was enhanced by splitting the WLC to reference and probe beams, both detected by a double CCD array in a prism spectrograph. This detection system allowed us to collect the data in a broad spectral range from 500 to 900 nm. The coherent artifact region (ca. −0.04 to 0.7 ps) was removed from the data. The same arrangement was used for the one-photon excitation (1PE) experiments with excitation wavelengths at 280 and 480 nm except that the 1PE intensity was set to ∼10¹⁴ photons per pulse cm². The obtained data sets were fitted using CarpetView software (Light Conversion, Lithuania) with a sequential model providing evolution-associated difference spectra (EADS).²⁸ The pump power dependence experiments were carried out to test nonlinearity of the obtained signals.

Sample preparation

8′-apo-β-carotenal was purchased from Sigma-Aldrich (≥96%, UV). The sample was stored in the dark at −40 °C prior to the experiment. For the transient absorption measurement, the sample was dissolved in n-hexane and methanol to yield OD ∼ 0.75 mm⁻¹. The experiment was carried out at room temperature with the sample mixed by a magnetic stirrer bar within a 2-mm quartz cuvette. Steady-state absorption spectra were measured before and after experiments to check for photodegradation of the sample. Some degradation was observed in the polar methanol (∼15%), a part of which could be due to trans–cis isomerization due to a slight increase of absorption in the 330–350 nm spectral region, while almost no degradation occurred when the carotenoid was dissolved in hexane (Fig. S1, ESI†).

Results

Ground state absorption

Fig. 1 shows the absorption spectra of 8′-apo-β-carotenal dissolved in two different solvents: non-polar n-hexane and polar methanol. Solvent polarity induces the effects described earlier.¹³ In n-hexane, the spectrum is narrower and shows three well resolved 0-0, 0-1, and 0-2 vibronic bands related to the S₀ → S₂ electronic transition. This contrasts with the spectrum in methanol, where these vibrational bands are much less defined, which is a common characteristic for carbonyl-containing carotenoids dissolved in polar solvents.^6,29 However, the solvent change does not notably alter the energy level of the S₂ state. No other absorption bands, such as a cis peak, are evident in the spectra. Even though 8′-apo-β-carotenal does not possess the ideal C_2h symmetry, there are no signs of the S₀ → S₁ transition found in either of the steady-state absorption spectra.

Fig. 1 (a) Energy level scheme of a carbonyl carotenoid with characteristic transitions, where A, ESA, and IC mean absorption, excited state absorption, and internal conversion, respectively; the structure of 8′-apo-β-carotenal is also shown. (b) Steady-state absorption spectrum of the 8′-apo-β-carotenal in methanol and n-hexane. The narrow gray rectangle marks the expected energy of the 0-0 band of the S₀ → S₁ transition. The solid, dashed, and dotted vertical lines represent the two-, three-, and four-photon excitation energies, respectively.

Excited state absorption and dynamics

Previous research on carotenoids with similar conjugation lengths, which studied weak S₁ emission and S₁ → S₂ transition energies, allows us to estimate the peak of the 0-0 band of the S₀ → S₁ transition between 14 880 and 15 580 cm⁻¹ (642 to 672 nm) for 8′-apo-β-carotenal.⁷ This range is highlighted with a gray rectangle in Fig. 1. Given that the energy of the S₁ state of 8′-apo-β-carotenal can only be estimated we have selected the excitation wavelength 1300 nm to reach the two-photon energy close to the expected peak of the 0-0 band of the S₀ → S₁ transition. This wavelength also falls into the spectral region where both solvents have minimal absorption (Fig. S2, ESI†).

Transient absorption spectra measured after excitation of 8′-apo-β-carotenal at 1300 nm in n-hexane and methanol are shown in Fig. 2. In n-hexane, the main peak is at ∼550 nm and is due to the S₁ → S_N transition as the spectral profile of this peak is comparable to that obtained in previous one-photon excitation studies of 8′-apo-β-carotenal in n-hexane,^13,20 proving that the 1300 nm excitation generates S₁ population. On the high energy side of the S₁ → S_N peak, there is a shoulder usually attributed to the S* state, observed previously in both one- and two-photon excitation experiments.^25,30–32 In the 600–700 nm region, two weak bands, which are associated with asymmetry of the conjugated system induced by the carbonyl group, are clearly visible even in the non-polar n-hexane.^13,20 In the near-IR spectral region, another weak positive band is located at 880 nm. It has a lifetime longer than the S₁ → S_N band as it decays only slightly over the first 100 ps and likely corresponds to a cation radical of 8′-apo-β-carotenal. Moreover, the 100 ps transient spectrum also contains a weak positive feature peaking around 530 nm, which does not decay on the time scale of our experiment. Based on the spectral position and long lifetime of this band, we attribute this signal to a triplet state.

Fig. 2 Transient absorption spectra of 8′-apo-β-carotenal in (a) n-hexane and (b) methanol after 1300 nm multi-photon excitation with 15 mW pump power. To mitigate data noise, the n-hexane 100 ps curve was derived by averaging the values between 95 and 115 ps.

The transient absorption spectrum changes significantly when 8′-apo-β-carotenal is dissolved in methanol. The S₁ → S_N transition band is red shifted to 565 nm at 1 ps, but it is not the dominant peak anymore. The increased polarity enhances the bands in the 600–700 nm region, generating the characteristic broad ICT band, which is even stronger than the S₁ → S_N transition. This contrasts with earlier data obtained after direct excitation of the S₂ state,¹³ but the enhanced amplitude of the ICT band observed here is due to overlap with the blue part of the strong spectral band peaking at 800 nm. Enhancement of this band in the polar solvent methanol supports the assignment of this feature to the 8′-apo-β-carotenal radical, because polar solvents stabilize the charge on the carotenoid cation radical. It is stabilized during the first 10 ps as it exhibits a blue shift from 820 to 800 nm between 1 and 10 ps. The carotenoid radical band almost does not decay within the time window of the experiment (Fig. 2).

Not only the spectral features but also the lifetimes of individual states of the carbonyl carotenoids are affected by the solvent polarity, as reported earlier also for the 8′-apo-β-carotenal.^13,20 The typical behavior of carbonyl carotenoids is a shortening of the S₁/ICT lifetime in polar solvents. This also applies for 8′-apo-β-carotenal excited at 1300 nm and the decays of the key spectral features are shown in Fig. 3. The prolongation of the S₁/ICT lifetime in n-hexane compared to methanol is obvious as well as the long lifetime of the cation radical in the near-IR spectral region. Interestingly, in methanol, the slow rise of the cation radical signal at 800 nm matches the decay of the transient S₁ → S_N band at 555 nm.

Fig. 3 Kinetics of 8′-apo-β-carotenal in (a) n-hexane and (b) methanol after 1300 nm multi-photon excitation with 15 mW pump power. The kinetics are shown at the wavelengths corresponding to the maxima of the S₁ → S_N, ICT → S_N, and D₀ → D₂ transitions.

To obtain lifetimes of individual processes, the global fitting procedure using a sequential relaxation scheme was employed. The results of the analysis are shown in Fig. 4. Each dataset required four time components to accurately fit the spectral evolution, represented by the evolution-associated difference spectra (EADS). In all datasets, the first EADS is related to the S₁/ICT state, and as in the case of two-photon excitation, it should be excited directly without depopulation from higher energy states. Also, due to significant interference from the coherent artifact within the first 0.5 ps, we cannot reliably assign sub-picosecond dynamics. In methanol, the S₁ band is accompanied by the ICT band typical for carbonyl carotenoids in polar solvents. As expected, the ICT amplitude is likely affected by the strong 800 nm cation radical band, which is also present in the first EADS spectrum, meaning that the radical must be generated within the first picosecond. Both S₁ and 800 nm bands exhibit signs of relaxation on the picosecond time scale. The relaxation process can be inferred from the peak narrowing and blue shift observed between the first and second EADS. The second EADS can be considered as depicting a relaxed S₁/ICT state with a polarity-dependent lifetime. In the non-polar n-hexane, the lifetime is 28 ps, and the polar methanol shortens the lifetime to 7 ps in agreement with earlier data obtained after one-photon excitation of the S₂ state.^13,18 It is important to note that the 7 ps process characterized by transition from the second (red) to third (blue) EADS also forms the final shape of the cation radical band. Since there is also an isosbestic point around 710 nm, clearly visible in both data (Fig. 2) and fits (Fig. 4), it may suggest that a part of the cation radical signal could be generated from the S₁/ICT state. Yet, since the signal from the carotenoid radical is clearly visible already in the first EADS, it is likely that similarity of the dynamics of the S₁/ICT decay and cation radical stabilization is rather coincidental. The third and fourth EADS are free of S₁ and ICT signatures and the only spectral features remaining are the 800 nm peak and the carotenoid triplet.

Fig. 4 EADS obtained from global fitting of multi-photon excitation transient absorption data of 8′-apo-β-carotenal in n-hexane (a), and methanol (b) after 1300 nm excitation. Both samples were excited by the 15 mW pump.

To justify the expected nonlinearity of the signal generated after excitation at 1300 nm, we have carried out excitation intensity dependence of the signal at the three key wavelengths corresponding to the maxima of the transient S₁ → S_N band in both solvents and the peak of the radical cation in methanol. The results are presented in Fig. 5. The data clearly indicate that the slope of the linear fit in the logarithmic plot is approximately four, and thus does not support the notion that the signal after 1300 nm excitation is due to two-photon excitation of the S₁/ICT state. Instead, the value rather points to a four-photon excitation of some higher-lying state, a picture comparable to that reported earlier for β-carotene.^23,33

Fig. 5 Dependence of signal amplitude on pump intensity at several wavelengths after excitation at 1300 nm at 2 ps in n-hexane and methanol. The numbers indicate the slope of the respective fitting curves.

Discussion

The data presented in previous paragraphs demonstrate that in addition to the typical features observed in transient absorption spectra of 8′-apo-β-carotenal related to the S₁ and ICT bands, excitation at 1300 nm also produces a carotenoid radical and, solely in n-hexane, a weak signal from the carotenoid triplet. These new species, which are absent in experiments using excitation via the allowed S₀ → S₂ transition, therefore must be related to the multi-photon nature of excitation as 8′-apo-β-carotenal has no linear absorption at 1300 nm. Surprisingly, the excitation intensity dependence shown in Fig. 5 indicates that the observed signal is not produced via two-photon excitation of the S₁/ICT state, but a dependence close to four-photon excitation occurs instead. It is important to stress that in our earlier experiments on lutein using the same experimental setup and excitation wavelengths, and comparable excitation intensities, we have observed intensity dependence identifying a clear two-photon excitation.²⁶ Thus, whether two- or multi-photon excitation dominates in a particular experiment is likely dependent on the carotenoid structure.

The observation that the signal measured here is a result of a four-photon excitation changes our assumption that we directly excite the S₁/ICT state. Instead, some higher-lying excited state located around 325 nm (∼30 770 cm⁻¹), thus far higher than the S₂ state, is primarily excited. Clearly, this excited state is forbidden for one photon transition from the ground state as evidenced by minimal ground state absorption in this spectral region (Fig. 1). The observation of the strong signal from a carotenoid radical already at 1 ps after 1300 nm excitation (Fig. 2) provides further evidence that the state primarily excited in this experiment is not the S₁/ICT, because this state, if populated via relaxation from the S₂ state, does not produce a carotenoid radical as reported in a number of previous experiments.^29,34

Since our data points to the conclusion that the carotenoid radical is formed from a dark excited state located above the S₂ state, the radical could be also generated if a higher-lying excited state is excited directly via one-photon excitation in the UV spectral region. This is indeed possible as essentially all carotenoids have a UV absorption band located in the 250–300 nm spectral region corresponding to a transition to the S_UV state.³⁴ For 8′-apo-β-carotenal the S_UV band has the lowest vibrational peak at ∼280 nm in both methanol and n-hexane (Fig. 1). Thus, to test this hypothesis, we have excited the S_UV band at 280 nm and transient absorption spectra obtained in this experiment are in Fig. 6 compared to those measured after multi-photon excitation. No carotenoid radical is produced by the 280 nm excitation, and transient spectra are nearly identical to those measured when the S₂ state is excited at 480 nm (Fig. S3, ESI†). The excited-state dynamics after 280 nm excitation are similar to those obtained earlier after 480 nm excitation as evidenced by global fitting analysis of the UV excited data (Fig. S4, ESI†). Absence of any signal from the carotenoid radical after 1PE of either the S_UV or S₂ state further supports the hypothesis that carotenoid radicals cannot be generated from the S₁/ICT state, even though the time scales related to decay if the S₁/ICT state and stabilization of the carotenoid radical are similar.

Fig. 6 Comparison of transient absorption spectra of 8′-apo-β-carotenal in (a) n-hexane and (b) methanol after 1300 nm multi-photon excitation (solid lines) and UV excitation at 280 nm (dashed lines). To mitigate noise, the 100 ps curve obtained after multi-photon excitation of 8′-apo-β-carotenal in n-hexane is shown as the average of all spectra measured between 95 and 115 ps.

While this agrees with previous experiments using UV excitation of carotenoids in which a carotenoid radical has never been reported,^34–36 it is rather surprising in the context of the experiment reported here. Obviously, the state excited at 280 nm should have enough energy to generate radicals as it must be higher than the dark state excited via the four-photon excitation at 1300 nm (∼325 nm). However, the radical is not produced, meaning that there is no pathway to radical formation from the S_UV state, and relaxation of the S_UV state excited at 280 nm bypasses the radical precursor, which can be achieved only via multi-photon excitation.

A hypothesis offered by Buckup et al.²³ suggested that multi-photon excitation may directly excite a doublet state, which would explain why the radical is not observed after one-photon excitation of the S_UV state. It is also worth noting that carotenoid radicals were reported after the ‘ladder’ excitation of astaxanthin in a pump–repump experiment called resonant-2-photon-2-color-ionization (R2P2CI).^37,38 In this experiment, the first excitation pulse at around 500 nm excites the S₂ state, while the second pulse arriving within the 100 fs is at around 1000 nm and thus resonant with the S₂ → S_N. In such arrangement, which excites an upper excited state located around 30 000 cm⁻¹, radicals of a few different carotenoids have been reported.^37–39 As well as in our experiment described here, these experiments provide evidence that carotenoid radicals can be generated from a state having energy of ∼30 000 cm⁻¹, thus below the S_UV state which, when excited directly via one-photon excitation in UV, does not produce carotenoid radicals (Fig. 6).^34,36 Interestingly, no carotenoid radical was reported in a pump–repump experiment in which the S₁ → S_N transition was re-excited.³⁶

Thus, a comparison of our data with earlier experiments suggests that the production of a carotenoid radical relates to multi-photon or ‘ladder’ excitation of an excited state located around 30 000 cm⁻¹, while one-photon excitation, even with higher photon energies, cannot produce the radical. This could be due to a configuration of the relaxed ground state, which could be unfavorable to produce radicals. Upon ladder excitation, however, the configuration of the intermediate state may enable a pathway that leads to the formation of the carotenoid radical. If this hypothesis is valid though, it would mean that in our multi-photon excitation experiment, we must be in the condition of a ‘pseudo-ladder’ excitation involving population of some intermediate state. Since our data does not allow for analyzing the very early delay times after excitation due to strong coherent artifacts in the first 500 fs, we cannot verify this hypothesis. The verification awaits for future multi-photon experiments, which should target the early dynamics after excitation.

Conclusions

Based on analysis of broadband transient absorption data obtained for 8′-apo-β-carotenal after NIR excitation at 1300 nm we conclude that, in contrast to the original assumption that NIR excitation induces a two-photon excitation of the S₁/ICT state, such excitation conditions lead predominantly to a multi-photon excitation of a dark excited state located above the strongly absorbing S₂ state. This state is a precursor of a carotenoid radical, though in the nonpolar n-hexane, in which the radical is not stabilized, it also opens a pathway to the triplet state. The radical precursor state must have a specific configuration as it cannot be excited by one-photon excitation from the ground state, even if UV excitation with energy above the radical precursor state is used. Our results show that in any multi-photon excitation experiments on carotenoids, intensity-dependent data are critical for data analysis and interpretation. Whether the obtained data results from two-photon^22,26,40 or multi-photon²³ excitation can be decided only from analysis of the intensity-dependent data. Clearly, carotenoid structure as well as the used solvent can be decisive factors as to whether two- or multi-photon excitation dominates.

Data availability

The key data are presented directly in the article. Raw, unprocessed spectroscopic datasets are available directly from the authors upon request.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

Financial support was provided by the Czech Science Foundation grant 19-28323X.

Notes and references

1. J. A. Bautista, R. E. Connors, B. B. Raju, R. G. Hiller, F. P. Sharples, D. Gosztola, M. R. Wasielewski and H. A. Frank, J. Phys. Chem. B, 1999, 103, 8751–8758.
2. D. Zigmantas, T. Polívka, R. G. Hiller, A. Yartsev and V. Sundström, J. Phys. Chem. A, 2001, 105, 10296–10306.
3. D. Zigmantas, R. G. Hiller, A. Yartsev, V. Sundström and T. Polívka, J. Phys. Chem. B, 2003, 107, 5339–5348.
4. K. Redeckas, V. Voiciuk and M. Vengris, Photosynth. Res., 2016, 128, 169–181.
5. R. G. West, M. Fuciman, H. Staleva-Musto, V. Šebelík, D. Bína, M. Durchan, V. Kuznetsova and T. Polívka, J. Phys. Chem. B, 2018, 122, 7264–7276.
6. H. A. Frank, J. A. Bautista, J. Josue, Z. Pendon, R. G. Hiller, F. P. Sharples, D. Gosztola and M. R. Wasielewski, J. Phys. Chem. B, 2000, 104, 4569–4577.
7. T. Polívka and V. Sundström, Chem. Rev., 2004, 104, 2021–2071.
8. P. O. Andersson, T. Gillbro, L. Ferguson and R. J. Cogdell, Photochem. Photobiol., 1991, 54, 353–360.
9. H. Nagae, M. Kuki, R. J. Cogdell and Y. Koyama, J. Chem. Phys., 1994, 101, 6750–6765.
10. M. Kuki, H. Nagae, R. J. Cogdell, K. Shimada and Y. Koyama, Photochem. Photobiol., 1994, 59, 116–124.
11. K. Seth, A. Kumar, R. P. Rastogi, M. Meena, V. Vinayak and H. Harish, Algal Res., 2021, 60, 102475.
12. Y. Bai, T. Cao, O. Dautermann, P. Buschbeck, M. B. Cantrell, Y. Chen, C. D. Lein, X. Shi, M. A. Ware, F. Yang, H. Zhang, L. Zhang, G. Peers, X. Li and M. Lohr, Proc. Natl. Acad. Sci. U. S. A., 2022, 119, e2203708119.
13. M. Durchan, M. Fuciman, V. Šlouf, G. Keşan and T. Polívka, J. Phys. Chem. A, 2012, 116, 12330–12338.
14. A. Mortensen, Pure Appl. Chem., 2006, 78, 1477–1491.
15. W. Jang, C. Lee, H. J. Suh and J. Lee, Food Sci. Biotechnol., 2023, 32, 1501–1513.
16. M. Rodriguez-Concepcion, J. Avalos, M. L. Bonet, A. Boronat, L. Gomez-Gomez, D. Hornero-Mendez, M. C. Limon, A. J. Meléndez-Martínez, B. Olmedilla-Alonso, A. Palou, J. Ribot, M. J. Rodrigo, L. Zacarias and C. Zhu, Prog. Lipid Res., 2018, 70, 62–93.
17. M. R. Wasielewski and L. D. Kispert, Chem. Phys. Lett., 1986, 128, 238–243.
18. F. Ehlers, D. A. Wild, T. Lenzer and K. Oum, J. Phys. Chem. A, 2007, 111, 2257–2265.
19. M. Kopczynski, F. Ehlers, T. Lenzer and K. Oum, J. Phys. Chem. A, 2007, 111, 5370–5381.
20. K. Horiuchi, C. Uragami, R. Tao, D. Kosumi, R. J. Cogdell and H. Hashimoto, Molecules, 2023, 28, 4424.
21. D. Kosumi, T. Kusumoto, R. Fujii, M. Sugisaki, Y. Iinuma, N. Oka, Y. Takaesu, T. Taira, M. Iha, H. A. Frank and H. Hashimoto, Phys. Chem. Chem. Phys., 2011, 13, 10762–10770.
22. P. J. Walla, P. A. Linden, C. P. Hsu, G. D. Scholes and G. R. Fleming, Proc. Natl. Acad. Sci. U. S. A., 2000, 97, 10808–10813.
23. T. Buckup, A. Weigel, J. Hauer and M. Motzkus, Chem. Phys., 2010, 373, 38–44.
24. D. Kosumi, K. Abe, H. Karasawa, M. Fujiwara, R. J. Cogdell, H. Hashimoto and M. Yoshizawa, Chem. Phys., 2010, 373, 33–37.
25. V. Šebelík, M. Fuciman, R. G. West and T. Polívka, Chem. Phys., 2019, 522, 171–177.
26. V. Šebelík, V. Kuznetsova, H. Lokstein and T. Polívka, J. Phys. Chem. Lett., 2021, 12, 3176–3181.
27. D. Kosumi, T. Kusumoto, R. Fujii, M. Sugisaki, Y. Iinuma, N. Oka, Y. Takaesu, T. Taira, M. Iha, H. A. Frank and H. Hashimoto, Chem. Phys. Lett., 2009, 483, 95–100.
28. I. H. M. Van Stokkum, D. S. Larsen and R. Van Grondelle, Biochim. Biophys. Acta, Bioenerg., 2004, 1657, 82–104.
29. D. Zigmantas, R. G. Hiller, F. P. Sharples, H. A. Frank, V. Sundström and T. Polívka, Phys. Chem. Chem. Phys., 2004, 6, 3009–3016.
30. C. C. Gradinaru, J. T. M. Kennis, E. Papagiannakis, I. H. M. Van Stokkum, R. J. Cogdell, G. R. Fleming, R. A. Niederman and R. Van Grondelle, Proc. Natl. Acad. Sci. U. S. A., 2001, 98, 2364–2369.
31. V. Balevičius, D. Abramavicius, T. Polívka, A. Galestian Pour and J. Hauer, J. Phys. Chem. Lett., 2016, 7, 3347–3352.
32. T. Lenzer, F. Ehlers, M. Scholz, R. Oswald and K. Oum, Phys. Chem. Chem. Phys., 2010, 12, 8832–8839.
33. D. Kosumi, M. Fujiwara, H. Hashimoto and M. Yoshizawa, J. Phys. Soc. Jpn., 2009, 78, 104715.
34. T. Khan, R. Litvín, V. Šebelík and T. Polívka, ChemPhysChem, 2021, 22, 471–480.
35. H. H. Billsten, J. Pan, S. Sinha, T. Pascher, V. Sundström and T. Polívka, J. Phys. Chem. A, 2005, 109, 6852–6859.
36. V. Kuznetsova, M. Fuciman and T. Polívka, Phys. Chem. Chem. Phys., 2023, 25, 22336–22344.
37. S. Amarie, J. Standfuss, T. Barros, W. Kühlbrandt, A. Dreuw and J. Wachtveitl, J. Phys. Chem. B, 2007, 111, 3481–3487.
38. S. Amarie, U. Förster, N. Gildenhoff, A. Dreuw and J. Wachtveitl, Chem. Phys., 2010, 373, 8–14.
39. E. Papagiannakis, M. Vengris, D. S. Larsen, I. H. M. Van Stokkum, R. G. Hiller and R. Van Grondelle, J. Phys. Chem. B, 2006, 110, 512–521.
40. D. A. Gacek, A. Betke, J. Nowak, H. Lokstein and P. J. Walla, Phys. Chem. Chem. Phys., 2021, 23, 8731–8738.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4cp04373a

---