Two- dimensional electronic spectra: energy transfer dynamics
Two-dimensional electronic spectroscopy (2DES)30-32 is a four-wave mixing spectroscopic technique capable of unravelling ultrafast energy transfer dynamics and the presence of coherences with high temporal and spectral resolution. The utilization of three broadband (90 nm fwhm centered at 695 nm, 2000 cm-1) and ultrashort (15 fs) laser pulses allows to excite a coherent superposition of all the electronic and vibrational states covered by the laser spectrum (all the Chl QY absorption bands) (Fig. 1a) and to follow both the evolution of the initially created coherences and the ET processes. The first and second pulses create coherent superpositions of states, the third pulse induces the emission of a photon echo signal, and the fourth pulse interferes with the emitted signal to allow the measurement of both the amplitude and the phase of the signal. By sequentially delaying the first and second pulses by the coherence time (τ) for a fixed population time (T, time between the second and third pulses) and by Fourier transform along τ, the excitation dimension of the 2D spectrum is obtained whereas the detection axis is acquired by dispersing the photon echo signal for each τ in a spectrometer. Consequently, the 2D spectra correlates emission with excitation events as a function of T: the populations of the initially excited states appear on the diagonal [excitation wavelength (λexc) = detection wavelength (λdet)] as positive features representing ground state bleach (GSB) and stimulated emission (SE), the excited state absorption (ESA) appears as a negative band above diagonal, and ET processes are directly visualized as population transfer between a donor and an acceptor as below diagonal features (cross-peaks, CP). Furthermore, coherences between two states (A and B) are observed only at the CP positions (λA,λB) or (λB,λA) with their 2D signal amplitude oscillating as a function of T with a frequency corresponding to the difference in energy between the states involved (ωA – ωB)30-32. The 2DES data has been collected with T from -1 ps to 1 ns, with the range between -104 to 2000 fs scanned with 8 fs steps to investigate whether the ultrafast ET processes occur via a coherent mechanism.
To assess the pathways and mechanism of ET in the PSII core antenna, we collected the 2D spectra of CP43, CP47, and monomeric Chl at 77 K. Figure 1a shows the linear absorption spectra of CP43, CP47 and Chl at 77 K together with the laser spectral profile utilized in the 2DES experiments. The laser light excites all the QY electronic states involved in ET and extends to the red side of the absorption spectra to allow the investigation of the low-lying energy states properties. The real rephasing 2D spectra for CP43, CP47 and Chl at 77 K are shown in Figure 1b for T = 312 fs after excitation, a population time at which, according to previously published transient absorption data17-19, the ultrafast phase of ET in CP43/CP47 is still in progress; hence, excited state population as well as energy transfer spectral features are present. Both the CP43/CP47 2D spectra show a rich diversity of spectral bands with diagonal and CP features whereas monomeric Chl shows a single band with the positive GSB and SE around the diagonal and the negative ESA above the diagonal, as expected for a monomeric molecule32,51,52. For CP43 (Fig. 1b, Table 1), the diagonal band shows two maxima centered at (λexcitation, λdetection) (667.7,669.1) nm with an amplitude of 0.80, and (681.4,682.7) with an amplitude of 0.92 which correspond to the GSB and SE of the main bands present in the linear absorption spectrum at 669 and 682.5 nm. These central wavelengths are taken at the minima of the second derivative of the absorption spectrum which shows bands at 682.5, 677.5 (weak), 669, and 660.5 nm. In the following, we will use the notation [(λexcitation, λdetection) nm, amplitude] to describe the spectral features (note that all 2D spectra has been normalized to one at the maximum of the 2D amplitude at T = 104 fs.) (Table 1). Below diagonal, three CPs are readily visible at [(675.4,682.7) nm, 0.45], [(669.2,682.7) nm, 0.42], and [(664.5,682.2) nm, 0.38]. These CPs are signatures of ET and/or coherences between the 664.5, 669.2, and 675.4 nm states and the low-energy 682.5 nm state. Note that all these states are present in the absorption spectrum yet the exact band position differs from absorption to 2DES data due to band overlap. For CP47 (Fig. 1b, Table 2), one maximum [(675.8,676.6) nm, 0.73] and two shoulders at [(681.4,682.7) nm, 0.60] and [(670.3,670.5) nm, 0.60] are distinguishable on the diagonal corresponding to the GSB and SE of the 683, 676.5 and 671 nm bands observed in the linear absorption spectrum (an additional band appears in the second derivative of the absorption at 660.5 nm). In this case, due to stronger spectral overlap (with respect to CP43) the CPs are not as well-resolved as in CP43, yet in CP47 the CPs are also clearly visible below diagonal along two detection wavelengths: 676 and 683 nm (see Fig. 2b for better resolved CPs at longer T). These CPs point towards the presence of ET and/or coherences among the 683, 676.5 and 671 nm states.
Table 1. CP43 real rephrasing 2D spectral features
T
|
Diagonal682
|
Diagonal669
|
CP675-682
|
CP669-682
|
CP665-682
|
CP659-682
|
CP654-682
|
112 fs
|
681.1,682.2
1.00
|
667.7,669.1
1.00
|
675.1,682.2 0.49
155 cm-1
|
669.5,682.2
0.37
280 cm-1
|
664.8,682.2
0.26
385 cm-1
|
659.1,681.8 0.14
505 cm-1
|
653.9,681.3 0.08
615 cm-1
|
312 fs
|
681.4,682.7
0.92
|
667.7,669.1
0.80
|
675.4,682.2
0.47
150 cm-1
|
669.5,682.7 0.39
295 cm-1
|
664.5,682.2 0.34
390 cm-1
|
659.1,681.8
0.21
505 cm-1
|
653.9,681.3
0.13
615 cm-1
|
496 fs
|
681.4,682.7
0.90
|
667.7,669.1
0.67
|
675.4,682.7
0.45
160 cm-1
|
669.2,682.7 0.42
295 cm-1
|
664.5,682.2 0.38
390 cm-1
|
659.1,681.8 0.25
505 cm-1
|
653.9,681.3
0.15
615 cm-1
|
992 fs
|
681.8,683.2
0.88
|
667.7,668.7
0.46
|
675.8,683.2 0.42
160 cm-1
|
669.2,682.7 0.45
295 cm-1
|
664.8,682.2 0.44
385 cm-1
|
659.1,681.8 0.31
505 cm-1
|
653.9,681.3 0.17
615 cm-1
|
10 ps
|
681.8,683.7
0.60
|
666.6,667.8
0.11
|
675.8,683.7
0.29
245 cm-1
|
666.6,683.2
0.53
365 cm-1
|
659.1,681.7 0.37
515 cm-1
|
653.9,681.3
0.24
615 cm-1
|
The excitation and detection wavelengths are given in nm separated by a comma (excitation,detection). The features for which the maximum precise position could not be determined are indicated in grey. In these cases, the position was chosen based on the nearby diagonal and CP positions. The amplitude of each feature is displayed below its position. The energy difference between all CPs is indicated in cm-1.
Table 2. CP47 real rephrasing 2D spectral features
T
|
Diagonal683
|
Diagonal676
|
Diagonal670
|
CP676-683
|
CP670-683
|
CP665-683
|
CP670-676
|
CP665-676
|
112 fs
|
681.4,682.7
0.66
|
675.8,676.6
0.93
|
670.0,670.0
0.92
|
675.8,682.7
0.40
150 cm-1
|
671.7,682.7 0.29
240 cm-1
|
667.0,682.7
0.18
345 cm-1
|
670.0,676.1
0.50
135 cm-1
|
662.3,676.1
0.22
310 cm-1
|
312 fs
|
681.4,682.7
0.60
|
675.8,676.6
0.73
|
670.3,670.5
0.60
|
675.8,682.7
0.33
150 cm-1
|
671.7,682.7 0.27
240 cm-1
|
667.0,682.2
0.21
335 cm-1
|
670.0,676.1
0.38
135 cm-1
|
662.3,676.1
0.22
310 cm-1
|
496 fs
|
681.8,682.7
0.48
|
675.8,676.1
0.56
|
670.3,670.5
0.50
|
675.8,682.7
0.28
150 cm-1
|
671.7,682.7 0.26
240 cm-1
|
667.0,682.7 0.22
345 cm-1
|
670.0,676.1 0.30
135 cm-1
|
662.3,676.1
0.20
310 cm-1
|
2.5 ps
|
683.0,684.1
0.35
|
676.2,676.6
0.33
|
669.2,669.6
0.27
|
676.9,684.1
0.25
155 cm-1
|
670.3,683.7
0.26
295 cm-1
|
662.3,682.2
0.20
440 cm-1
|
669.2,676.1 0.19
150 cm-1
|
662.3,676.1
0.16
310 cm-1
|
20 ps
|
683.0,684.1
0.18
|
675.1,675.6
0.13
|
669.5,670.1
0.14
|
675.4,684.6
0.15
200 cm-1
|
670.3,683.7
0.17
295 cm-1
|
662.3,682.2
0.14
440 cm-1
|
668.1,675.6
0.08
165 cm-1
|
663.0,676.1
0.08
290 cm-1
|
The excitation and detection wavelengths are given in nm separated by a comma (excitation,detection).
The features for which the maximum precise position could not be determined are indicated in grey. In these cases, the position was chosen based on the nearby diagonal and CP positions. The amplitude of each feature is displayed below its position. The energy difference between all CPs is indicated in cm-1.
Remarkably, the application of 2DES at 77 K allows the observation of well-resolved diagonal and CP spectral features that remain hidden when employing other spectroscopic methods such as transient absorption. Here, the additional detection dimension of the 2DES data allows us to directly visualize ET processes between closely spaced electronic states, with energy differences as low as 135 cm-1. After identifying the main spectral features in the 2D spectra, we analyze the ET dynamics. Figure 2a/b shows the CP43/CP47 real rephasing 2D spectra at four representative T (112 fs, 496 fs, 1 ps, 10 ps)/ (112 fs, 496 fs, 2.5 ps, 20 ps). The 2D spectra for monomeric Chl as a function of T is not shown because, as expected, no significant spectral evolution is observed43-46 but only the partial decay of the GSB, SE and ESA in the time scale of our experiment (the longest probed T is 1 ns). At T = 0 fs, coherence is the only origin of CPs since no ET can occur right upon excitation. Yet, strong coherent artefacts that cannot be eliminated are present until T ≈ 100 fs due to the wave-mixing nature of 2DES32, resulting in T = 104 fs being the first T free from artefacts in our datasets. As the system evolves over time, the coherence usually disappears through intramolecular and intermolecular (mostly pigment-environment) interactions32.
The position and amplitude of the main diagonal and CP features for CP43/CP47 at five representative T are summarized in Table 1 and Table 2, respectively (note that the features that appear as shoulders not as maxima are shown in grey). In addition, the difference in energy between the states involved in each CP is indicated. For CP43 (Fig. 2a, Table 1), two diagonal maxima are observed at 682 and 669 nm, together with five CPs connecting the states absorbing at 675, 669, 665, 659 and 654 nm with the low-lying 682.5 nm state. It is worth noting that even though the CP659-682 and the CP654-682 do not appear as separated CPs along the excitation axis due to band overlap, they do emerge as maxima along the detection axis. The overall spectral evolution consists of (from 112 to 992 fs): the fast decay of the Diagonal669 amplitude (from 1.00 to 0.46), the slow decay of the Diagonal682 amplitude (from 1.00 to 0.88), the moderate decrease of the CP675-682 (from 0.49 to 0.42) and the increase of the other CPs: CP669-682 (from 0.37 to 0.45), CP665-682 (from 0.26 to 0.44), CP659-682 (from 0.14 to 0.31), and CP659-682 (from 0.08 to 0.17). The evolution of the spectral features reveals ET from the 675, 669, 665, 659, and 654 nm states to the 682.5 nm state. The Diagonal669 band decays into the Diagonal682 feature via the CP669-682: the Diagonal669 amplitude decreases while the CP669-682 amplitude increases due to population transfer to the 682.5 nm state. At the same time, the Diagonal682 amplitude should increase, yet it decays slightly (from 1.00 to 0.88). This could be explained by that fact that two states contribute to the Diagonal682: the broad [680 nm (11 nm fwhm)] and the narrow [682.5 (3 nm fwhm)] low-energy states since both are excited by the laser pulses. In this case, it is reasonable to consider that the decay of the 680 nm state GSB and SE hinders the observation of the increase of the Diagonal682 amplitude due to energy transfer from the 669 nm to the 682.5 nm state (see below). Furthermore, and even though the Diagonal675 and the Diagonal665 are not resolved as separated bands due to spectral overlap, the presence of the CP675-682 and the CP665-682 implies that ET from both the 665 and the 675 nm states to the 682.5 nm state occurs on an sub-picosecond time scale (the moderate decrease, instead of increase, of the CP675-682 is most likely caused by band overlap with the Diagonal682 band that decays slightly in the same time scale). Moreover, the high sensitivity of 2DES allows us to visualize ET from higher energy states (absorbing at 659 and 654 nm) via the presence of the CP659-682 and CP654-682. At longer times, T = 10 ps, the Diagonal669 feature has disappeared almost completely (amplitude 0.11), while the Diagonal682 is clearly visible (amplitude 0.60) an indication that this state has been populated by ET from higher-energy sates.
Remarkably, the ET from the 665, 669 and 675 nm states to the narrow (3 nm fwhm) low-energy 682.5 nm state is revealed by two facts: the CPs central wavelength (682.5 nm) and their width (around 3 nm). It is worth noting that even though the Diagonal682 is broad (at least 10 nm) owing to the population of both broad and narrow low-energy states [680 nm (11 nm fwhm) and 682.5 (3 nm fwhm)] by the laser excitation, the fact that the CPs are as narrow as the 682.5 nm state and centered at the same wavelength demonstrates that only this state receives excitation energy from higher energy states. The observation of three well-resolved CPs reveals the existence of, at least, three distinct ET pathways from the different Chls pools absorbing around 675, 669 and 665 nm to the 682.5 nm state, whereas the broad low-energy 680 nm state does not receive energy (i.e., remains unconnected) from the high-energy states.
For CP47 (Fig. 2b, Table 2), the overall spectral evolution is similar to CP43 yet with significant differences. In this case, due to stronger spectral overlap half of the spectral features displayed in Table 2 do not appear as separated maxima; therefore, and in order to analyze their amplitudes, their positions have been taken from better-resolved 2D spectra (for instance at T = 496 fs for the CPs present along the 683 nm detection wavelength at and T = 2.5 for the CPs along the 676 nm detection wavelength). At T = 112 fs, the GSB and SE features have maximum amplitude (≈ 0.94) between 667 and 677 nm along the diagonal and extend to the red up to around 685 nm with half the maximum amplitude (≈ 0.47). The CPs start to be apparent below diagonal along the detection axis around 676 and 683 nm although to a lower extent than for CP43. At longer T, 496 fs and 2.5 ps, the CPs are better resolved and clearly visible. In fact, at these population times CP47 displays two kinds of CPs, three CPs connecting the 676, 670, and 665 nm states (CP676-683, CP670-683, and CP665-683) with the low-energy 683 nm state (similar to CP43); and two CPs relating the 670 and 665nm states (CP670-676 and CP665-676) with the 676 nm state. The latter CPs appear as separated bands at 2.5 ps. At longer times, T = 20 ps, three diagonal features are visible at 683, 676 and 670 nm; and most of the CPs emerge as maxima. Unfortunately, in this case the strong band overlap complicates the analysis of the 2D spectral features amplitude, which in almost all cases decreases as function of time, most likely due to pigment-environment interactions32.
Nevertheless, and despite the strong spectral congestion, the CP47 spectral features demonstrate the presence of multiple ET pathways with various donor and acceptor states: the lowest-energy state (683 m) receives energy from higher-energy states (676, 670, and 665 nm), the high-energy states (670 and 665 nm) transfer energy to the 676 nm state, and the 676 nm state receives and transfers energy.
Two-dimensional frequency maps: coherences between states involved in ET
After analyzing the ET dynamics, now we explore whether ET in CP43/CP47 proceeds via a coherent mechanism. As it was previously indicated, coherence between two states (A and B) emerges in the real rephasing 2D spectra as a CP [(λA,λB) or (λB,λA)] which signal amplitude oscillates as a function of T with a frequency equivalent to the difference in energy between the states (ωA – ωB)30-32. To examine the presence of coherent ET in CP43/CP47, the quantum beats (2D signal amplitude oscillations as a function of T) have been isolated from the 2D traces by subtracting the ET dynamics (applying and subtracting a polynomial fit to the traces) and analyzed by Fourier transform (FT) in order to retrieve the oscillation frequencies present in the datasets (CP43/CP47 and Chl). The FT has been performed on the traces from 104 to 1504 fs which yields a 40 cm-1 frequency resolution (± 20 cm-1) with the maximum sampled frequency being ≈ 2100 cm-1 (8 fs step employed within the FT range). Figure 3 shows the results of the FT analysis up to 850 cm-1. Plotting the maximum amplitude of each frequency found in the complete 2D dataset for each studied system (Fig. 3a) provides a general view of the dominant frequencies present in the data. Yet, a detailed physical picture is obtained by analyzing the 2D frequency maps (Fig. 3b, c, and d). Differently from the 2D spectra which contain contributions from all absorbing and emitting states, the frequency maps display only the states which oscillate at a specific frequency, hence these maps reveal coherences between electronic/vibrational states. The major frequencies for CP43 are (± 20 cm-1): 150, 265, 300, 345, 400, 520, and 750 cm-1 (Fig. 3b) (the 2D frequency maps for 300, 400, and 520 cm-1 are shown in the Supplementary Fig. S1). For CP47 the dominant frequencies are (± 20 cm-1): 130, 265, 300, 345, and 740 cm-1 (Fig. 3c) (the 2D frequency maps for 300 cm-1 is shown in the Supplementary Fig. S1). In the case of Chl, the observed frequencies related to the ones found for CP43/CP47 are (± 20 cm-1): 170, 265, 305, 345, 395, 520 and 740 cm-1 (Fig. 3d) (the 2D frequency maps for 305, 395, and 520 cm-1 are shown in the Supplementary Fig. S1). It is interesting to note that for the three samples almost all frequencies retrieved by FT of the 2DES data have been previously observed in their FLN spectra (except for the frequencies below 250 cm-1 in CP43/CP47 that overlap with the strong phonon wing in FLN) (CP438, CP4711, and Chl53). Since the CP43/CP47 oscillation frequencies have their counterparts in the FLN spectrum of isolated Chl, we can conclude that the frequencies observed in 2DES correspond to Chl intramolecular vibrations. The information about the coherences present in the 2D frequency maps can be understood as follows. Electronic and vibronic coherence between two states (A and B) appears as a CP [(λA,λB) or (λB,λA)] in the 2D frequency map corresponding the difference in energy between the states (ωA – ωB)30-32. For vibrational coherence the chair-type structure emerges3,32,34,44,45 in the real rephasing 2D frequency maps as a combination of vibrational coherence generated via the GSB and SE [see Fig.3d (Chl 740 cm-1 map) for an example], whereas the vibrational coherence generated only via the SE emerges with the amplitude distribution of the Chl maps at 265 and 345 cm-1 (Fig.3d) where the CP below the diagonal feature is absent32.
The interpretation of the 2D frequency maps (Fig. 3b, c, and d) is assisted by: i) the calculation of the energy difference between the electronic states involved in each ET step (states connected by CPs) (Table 1/2 for CP43/CP47), ii) the indication of the electronic states center wavelength as horizontal and vertical back dotted lines, iii) the inclusion of the below and above diagonal lines shifted by the frequency of each map as thin black lines.
For CP43 (Fig. 3b), the low frequency 150 cm-1 map corresponding to the CP675-682 displays minor but visible amplitude at the CP position, indicating that ET from the 675 to the 682 nm state may proceed via a vibronic coherent mechanism. The related monomeric Chl map at 170 cm-1 clearly deviates from the chair-type structure and from the structure observed for CP43/CP47, still, a virtually identical 2D map was observed recently for monomeric bacteriochlorophyll a at 77 K50 (in isopropanol as used here). The 265 cm-1 map which corresponds to the CP669-682 displays a complicated amplitude distribution. In this case, it is extremely useful to compare the CP43 map with the related Chl map which shows the amplitude distribution of the vibrational coherence generated via SE32. In fact, the CP43 map resembles the Chl map, it contains the above diagonal, the diagonal and the two below diagonal features as the Chl map yet with a significant difference: the intense CP669-682 feature. Consequently, we conclude that for CP43 the 265 cm-1 map represents both vibrational and vibronic coherence, the latter indicating that ET from the 669 to the 682 nm state occurs via a coherent mechanism. The 345 cm-1 map (CP665-682) shows a dominant signal at the CP position both below and above diagonal (note that the above diagonal CP feature is not exactly at the CP682-665 position most like likely due to spectral overlap with the ESA signal), with a minor contribution from the vibrational coherence via SE, indicating the presence of vibronic coherence between the 665 to the 682 nm state which implies that this ET channel also exploits the advantages of coherence. The 750 cm-1 map reveal the chair-type structure corresponding to purely vibrational coherence, that is, this vibration is too high in energy to couple to the electronic energy gaps. For CP47 (Fig. 3c), the 130 cm-1 map corresponds to the CPs connecting the 676 and the 683 nm states (CP676-683) as well as the 670 and the 676 nm states (CP670-676). The map shows a clear feature at the CP676-683 position, indicating that ET between these states occurs via a coherent mechanism. In addition, the CP feature extends to the CP670-676 suggesting that also these two states might be connected by coherent ET. The 265 cm-1 frequency is close in energy to the energy gaps between the 670 and 683 nm states as well as between the 665 and 676 nm states (CP670-683 and CP665-676, respectively), and indeed, both CPs appear in the 265 cm-1 frequency map with significant amplitude. Again, the shape of the amplitude distribution in the frequency map strongly indicates that ET between the states involved in the above-mentioned CPs proceeds via a coherent mechanism. The 345 cm-1 map (CP665-683) displays intense amplitude features both below and above diagonal at the CP position. Similarly to CP43, the minor participation of vibrational coherence contributes to the diagonal and above diagonal CP and the overlap with the ESA band may shift the possible vibronic above diagonal feature slightly to the blue. In any case, the strong signal at the CP665-683 position provides compelling evidence of coherent ET from the 665 to the 683 nm state. The 740 cm-1 map display the chair-type amplitude shape consistent with purely vibrational coherence, therefore, we conclude that this high energy vibration is not able to couple to the electronic energy gaps.