1. Introduction

Carotenoids, along with chlorophylls and bacteriochlorophylls, are the most prevalent light absorbing pigments in nature and have many functions in photosynthetic systems [1–5]. Carotenoids are known to have mainly two vital functions within photosystems. One of them is to enhance photosynthesis due to its ability to transfer excitation energy to chlorophylls. The other is that they can quench highly reactive singlet oxygen and protect the photosystem [6–10]. Carotenoids achieve these functions through their excited states [11–22].

Nevertheless, there are still some unsolved problems in the steady-state spectroscopy of the β-car. One of them is the double components feature of the C = C stretching band (v₁ mode) in resonance Raman scattering [23–26]. In the theoretical aspect, the resonance Raman spectrum can be interpreted through Kramers-Heisenberg-Dirac (KHD) formula and efficiently calculated by Tannor-Heller theory based on time-dependent wavepacket propagation [27]. The computational result suggests that the small blueshift of the v₁ mode can be only observed with the geometry change through electronic excitation and the mode mixing in the excited state (Duschinsky rotation) should be considered [28–30]. It is known that the solvents have significant influences on the π-electron delocalization and the electron-phonon coupling, which finally change the corresponding spectroscopic properties of the β-car [31–34]. Using resonance Raman spectroscopy with changing external conditions, we can preliminarily study the influence of environmental factors on the ground state and excited state of the carotenoid. The results obtained are complementary to the investigation of excited states and time-resolved spectroscopy.

The generalized two-dimensional correlation spectroscopy (2DCOS) is another powerful tool to study the dynamics of the molecular systems in a relatively slow-changing variable comparing to the time-resolved technique [35]. The combined use of resonance Raman and 2DCOS, namely 2DRRCOS, can reveal the dynamics of the excited state of carotenoid, perturbed by changing environmental factors.

Here, we focus on the effects of environmental factors (solvent, temperature, aggregation, and electric field) on the resonance Raman scattering of v₁ mode and reveal the evidence of v₁ mode double components feature under the influence of environments. Also, the independent influence of each environmental factor on the β-car excited state and the decisive role of the solvent are discussed.

2. Experimental

The β-car was purchased from Tokyo Chemical Industry Co., Ltd. (TCI, Shanghai) and was storing at −20 °C in the dark. The purity of the β-car was ≥97.0% and we used without further purification. The absorption spectra measurements were carried out on a Purkinje General TU-1901 spectrophotometer with 1 nm step. The solvents were contained in 1 cm × 1 cm quartz cell. All of the experiments were performed at room temperature in a dark room. The Raman spectra of β-car in different solvents were backscattered and collected using Renishaw inVia Raman microscope with a 514.5 nm excitation from the Ar⁺ laser (Spectra Physics163-M42) working at 5.0 mW to avoid the local heating effect. The excitation beam was positioned on a 50× objective lens (LEICA DMLM 0.12NA) to graze the surface of the capillary. The capillary's inner diameter is 0.1 cm and placed perpendicular to the laser beams. The spectra presented here were all taken with a scanning speed of 10 cm⁻¹/min (CCD integration times). The Raman spectra of the β-car-benzene solution was measured at different temperatures in a Linkam thms600 system. To prevent evaporation, the temperature was controlled at 273 K immediately after the capillary was set into the cryostat. The laser beam then focuses on the surface of the capillary to obtain the largest Raman gain. For the electric field tests, the β-car was dissolved in toluene and then spin-coated on the indium tin oxide (ITO)/glass substrate, and the solvent was evaporated at room temperature. Then the Ag film covered the β-car by using the vacuum evaporation technique. The thin film was subjected to optical measurement immediately after fabrication. To apply the direct-current electric field, a direct-current power supply with the voltage range from 1-15 V was used. The Raman spectra of the β-car thin film were measured immediately after the film was prepared. The 2D correlation spectroscopy based on Noda’s algorithm was programmed using Matlab R2015b (The Math Works Inc., Natick, MA, USA) [35]. The schematic of the 2DRRCOS is shown in Fig. 1.

 

Fig. 1. Schematic of 2DRRCOS experimental setup.

Download Full Size | PDF

3. Results and discussion

The β-car has three prominent Raman bands named v₁, v₂, and v₃ located at ∼1524 cm^-1, ∼1157 cm^-1, and ∼1005 cm^-1 respectively. Previous attempts to reveal the double peaks feature left questions either on broad resonance Raman excitation profile or peak fit by two Gaussian components [25,36]. The former can be explained by large damping factors, causing the lack of fine excitation profile structure. The latter will be more reasonable based on the 2DRRCOS [Fig. 1(a)]. Firstly, we investigated the double components feature of β-car as a function of solvent polarizability, which had a significant influence on the absorption 0-0 peak position of β-car. The 1444 cm^-1 of cyclohexane was set as the internal label and the Raman intensity was normalized.

From Fig. 2(a), the asynchronous peaks (1516 cm⁻¹, 1522 cm⁻¹) and (1516 cm⁻¹, 1528 cm⁻¹) indicate that there are two modes that vary out-of-phase with each other and we found more information in the asynchronous cross peak at (∼1520 cm⁻¹, ∼1155 cm⁻¹). As shown in Fig. 2(b), there are two cross peaks located at (1528 cm⁻¹, 1159 cm⁻¹) and (1516 cm⁻¹, 1153 cm⁻¹), respectively. The interpretation of these two bands can be straightforward. Given that the electronic transition of β-car mainly depends on the polyene chain properties named bond length alternation (BLA), the conjugation length of β-car in the excited state can be measured as a function of the intensity ratio of the C = C and C-C stretching modes (i.e., the intensity ratio Iv₁/ Iv₂) [37]. Figure 2(b) illustrates that it should have two kinds of β-car with different conjugation lengths when changing solvent polarizability. The additional conjugation species should be attributed to the mode mixing or Duschinsky effects in the excited state and the β-car system obtains momentum along the asymmetric coordinate after excitation. Therefore, the cross peaks at (∼1520 cm⁻¹, ∼1155 cm⁻¹) are the signature of the excited state mode mixing by changing solvent polarizability.

 

Fig. 2. (a) The asynchronous 2DRRCOS of the ∼1520 cm⁻¹ mode. The color bar represents the correlation intensity. (b) Based on the 2DRRCOS results, the ∼1520 cm⁻¹ band can be deconvoluted with two strong peaks located at 1519 cm⁻¹ and 1525 cm⁻¹. (c) These two modes show subtle frequency shift by changing the solvent. (d) The relative intensity ratio of the 1519 cm⁻¹ and 1525 cm⁻¹ mode with the change of the solvent polarizability.

Download Full Size | PDF

The double components feature of v₁ mode is attributed to the mode mixing in the excited state, whose intensity is in the Franck-Condon region corresponding to the dynamics of first tens of femtoseconds, leading to the vibrational mode shifts to higher wavenumbers [38]. This can be explained by adding the Duschinsky rotation effect into the modeling of the lowest potential energy surface between the ground and excited states [39]. From recent molecular dynamics study, it also emphasized the decisive role of solvent polarizability. While the dielectric constant (i.e. the polarity of the solvent) induces subtle frequency shift to the v₁ mode [40–43]. In contrast, the electron-phonon coupling constant V₁ and V₂ decrease obviously with the increasing solvent polarizability [12]. This constant has the similar meaning to the so-called reorganization energy, which is an effective measure of the electron-phonon coupling. All the vibrational modes contribute to the reorganization energy and each of them is proportional to its Huang-Rhys factor [43]. The reorganization energy reduces with the increase of solvent polarizability and will influence the correlation function in frequency domain which finally decides the linear absorption profile.

Based on the 2DRRCOS, we deconvoluted the v₁ band into v_1a and v_1b with Lorentzian peak profile, the best fitting achieve by v_1a and v_1b modes locating at 1519 cm⁻¹ and 1525 cm⁻¹, respectively [Fig. 2(c)]. Figure 2(d) clearly presents that the v_1a to v_1b intensity ratio reach to the top as the solvent polarizability get closer to 0.27, and the corresponding absorption peak is about 500 nm. Considering the fixed excitation wavelength, the decrease of solvent polarizability is moving the electronic transition away from the resonance condition, which is equivalent to change excitations in the investigation of the resonance Raman profile. Such a peak indicates the v_1a to v_1b intensity ratio depends strongly on the wavelength difference to the excitation. Interestingly, these two modes do not shift at all in the different solvent, but their intensity ratio changes dramatically leading to the small shift of the v₁ band. The largest ratio doesn’t obey the rigorous resonance condition and slightly blueshift to the shorter wavelength. This can be explained by associating the resonance effect with the energy relaxation that a more polarizable solvent leads to faster dynamics in an excited state at first tens of femtoseconds after excitation [44,45]. Also, the small shift of v_1a to v_1b mode can be examined by introducing the amplitude mode theory, which has successfully predicted the energy gap modulated by the electron-phonon coupling and the ground state geometry [46]. The absorption energy depends on the electron-phonon coupling and the reduced form can be expressed as $\delta E/E_g = \left( {1/2\lambda } \right)\delta \lambda /\lambda$. The $\lambda$ is electron-phonon coupling constant and is decided by the “product rule”-$\mathop \prod \nolimits_{n = 1}^N \left( {\omega _n{\rm /}\omega _n^0 } \right)^2 = 2\tilde{\lambda }$. In long chain approximation, there is $\lambda \cong \tilde{\lambda }$. That is to say, the frequency shift should be accompanied by energy gap variation. However, with the redshift of the absorption band, there are almost no frequency shifts between the two modes v₁ and v₂ in resonance Raman spectra of β-car. This result is mainly due to the weakened inter-chain interaction during solvation, and the ground state geometry of β-car has little changes when it dissolves well in different solvents.

The Franck-Condon active modes are sensitive to temperature effects due to changes in either equilibrium position or vibrational frequency upon excitation [39,45]. The asynchronous plot at (∼1520 cm⁻¹, ∼1155 cm⁻¹) shows a similar pattern as that in changing polarizability [ Fig. 3(a)]. This suggests that there also exists the double components feature of β-car in solid benzene. However, the asynchronous peaks at (∼1520 cm⁻¹, ∼1520 cm⁻¹) present a typical butterfly pattern since the v_1a and v_1b bands shift to the higher wavenumber with the decrease temperature [Figs. 3(b) and 3(c)]. This shift indicates that the equilibrium β-car geometry changes in both ground and excited states. Thus, the polarizability, i.e. the refractive index of the system, should firstly decide the energy gap, while the temperature effect on β-car leads to a slight geometry change on the polyene chain of β-car. As a result, the effect of temperature induces a slight shift to the absorption of β-car dissolved in various solvents. This is consistent with the amplitude theory because the ground state geometry shifts about 4 cm⁻¹. Thus the equation $\delta E/E_g = \left( {1/2\lambda } \right)\delta \lambda /\lambda$ contributes to the slight energy gap shift with the temperature induced vibrational shift. Therefore, these two factors affect β-car from different aspects: the solvent polarizability mainly influences the displacement of the lowest points of the potential energy surface, while the temperature affects the excited state geometry of β-car.

 

Fig. 3. (a) Asynchronous plot at (∼1520 cm⁻¹, ∼1155 cm⁻¹) with decreasing temperature of β-car dissolved in benzene. (b) The v₁ band evolution as a function of temperature. Inset: the Raman intensity ratio between v₁ (C = C) and v₂ (C-C) bands. The red line is the linear fitting of the intensity ratio as a function of temperature. (C) The fitting Raman shift of v_1a to v_1b modes at 77 K. (d) Intensity ratio between v_1a and v_1b bands at various temperatures.

Download Full Size | PDF

The intensity ratio of v₁ and v₂ bands changes from 1.20 at 273 K to 1.46 at 77 K and the slope looks linear when the temperature decreases, which indicates a linear redshift on absorption 0-0 peak [Fig. 3(b) inset]. This slight redshift can be shown by the relative intensity ratio of v_1a or v_1b band to the 992 cm⁻¹ band of benzene [Fig. 3(d)]. Although the v_1a to v_1b intensity ratio fluctuates with the decreasing temperature, the effect of temperature on the intensity ratio is less significant than that by changing polarizability. So far, based on the 2DRRCOS and peak fitting, we not only deconvoluted the v_1a and v_1b bands influenced by temperature and solvent polarizability but also distinguished each parameter’s contribution to the electronic and vibrational properties of β-car.

We know that the aggregation induces so-called singlet-fission to the excited state of β-car by using a binary solution with high polarities, such as an ethanol-water solvent mixture [13]. It is so wired that the v₁ band hardly changes under the effect of temperature, even though that the aggregation leads to significant modification on the excited state geometry, as shown in Fig. 4(a). The phase transition of the solvent always leads to the downshift of the v_1a to v_1b bands, but they shift to higher wavenumber with decreasing temperature. As a result, the relative intensity of v_1b band is greatly enhanced at 77 K due to the large blueshift bring by singlet fission and temperature effect. The solvent polarity shapes the ground and excited state geometry, and the temperature effect is similar to that in benzene. When the ground state of β-car changes frequently with the changing environmental effects, the asynchronous plot at (∼1520 cm⁻¹, ∼1520 cm⁻¹) shows at least three peaks [Fig. 3(b)]. This result indicates the hypothesis above, the cumulative effects on the intensity ratio of v_1a to v_1b bands, such as temperature and phase transition, should be a reasonable explanation for the unchanged band position. The reversed intensity ratio is, therefore, an indicator of a larger blueshift away from the resonance region.

 

Fig. 4. (a) Raman spectra of β-car aggregate in ethanol-water solution, temperature range: 273K-77 K. (b) The asynchronous plot of (∼1520 cm⁻¹, ∼1520 cm⁻¹).

Download Full Size | PDF

A very recent report by molecular dynamics simulation suggested that the polarity should not the deciding parameter influencing the excited state of carotenoid since the S₀-S₂ transition and v₁ mode in the lowest dielectric solvent were the same as that in the highest one [40,41]. The electric field strength reached about 10⁶ V/cm, which was equal to the reaction electric field of a polar solvent and could modulate the BLA [47]. It is the ground state BLA of β-car that determines the vibrational frequency and the energy gap. Therefore, the thin film leads to a lower Raman shift of v₁ mode and a lengthened effective conjugation length [Fig. 5(a)], causing the redshift of 0-0 absorption peak [48]. This cooperative effect leads to a similar 2DRRCOS asynchronous pattern with more peaks at (∼1520 cm⁻¹, ∼1520 cm⁻¹) region [Fig. 5(b)]. At the same time, it is noteworthy that the v_1a and v_1b modes have subtle shift by increasing voltage but the rapid increase relative intensity of v_1b mode should be the most important factor that influences the v₁ band position [Figs. 5(c) and 5(d)]. The increase of external electric field induces a solvation-like Raman spectrum of β-car thin film. Besides, the thermal effect can be excluded since the v₁ and v₂ modes shift toward higher and lower wavenumbers, respectively. Thus, this result can be interpreted by increasing BLA in the β-car ground state that the electric field leads to the lengthening of C = C double bond and the shortening of C-C single bond, which shows a more “polyene” feature. The external electric field plays two major roles. One of them is the solvation-like effect that induces the change of polarizability. The other is changing the ground state morphology by the increasing voltage, which provides enhanced polarity.

 

Fig. 5. (a) Raman spectra of v₁ and v₂ modes with the increase external electric field. (b) 2DRRCOS asynchronous plot of (∼1520 cm⁻¹, ∼1520 cm⁻¹) region. (c) v_1b band gains intensity at larger external electric field. (d) The evolution of the relative Intensity ratio of v_1a and v_1b bands as a function of voltage.

Download Full Size | PDF

4. Conclusion

In conclusion, we used 2DRRCOS to reveal the v_1a and v_1b bands evolution under various environmental effects. We observed that the polarizability mainly decided the excited state dynamics, leading to the change of relative intensity ratio of v_1a and v_1b bands. Solvent effects, such as polarity and polarizability, only change the excited state dynamics, while the temperature induces changes on both excited and ground state mode mixing. Further, we applied the electric field on the β-car thin film and demonstrated that the increased strength of the electric field would not dramatically change the Raman shift of the v_1a and v_1b modes. The result was consistent with the fact that the energy of the S₀-S₂ transition was less influenced by the solvent polarity of a well-dissolved β-car solution. The investigation on β-car aggregate indicates that the morphology may induce significant changes in the excited state, and the evolution of v_1a and v_1b modes under environmental effects will become more complicated.

By introducing environmental perturbation, the mode mixing in the excited state can be modulated. The electric field effect on β-car provides changes in polarity and is similar to the solvation effect. Our results emphasize the significance of excited state mode mixing influenced by environmental effects and demonstrate the ability of the 2DRRCOS for studying the structural and dynamics of mode mixing effect on resonance Raman scattering.