2.2. Femtosecond Transient Absorption Spectroscopy
Femtosecond transient absorption (TA) measurements were first performed to obtain the excited state dynamics of BBM molecule.
Figure 2a and S2a show ultrafast two-dimensional TA spectra of BBM molecules in DMSO and THF under excitation at 350 nm with a power intensity of 70 nJ/pulse. The TA spectrum against the delay time from −1 ps to 8 ns exhibits intricate excited state features between 350 nm and 650 nm including three excited state absorption bands (ESA1 band center at 380 nm; ESA2 band center at 440 nm; ESA3 band of 460 nm - 640 nm), the stimulated emission band (SE center at 450 nm) and the ground-state bleach at the blue edge. Moreover, global analysis with a parallel model was employed to obtain the decay-associated difference spectra (DADS, see
Figure 2b), yielding four characteristic lifetimes with time constants of τ
1 = 0.3 ps, τ
2= 3.0 ps, τ
3 = 1.6 ns, and τ
4 = infinitely long lifetime in DMSO.
As shown in
Figure S3, between the 0 and 0.6 ps time scales, the BBM temporal trace in DMSO at 425 nm (ESA 2 band) exhibits initial rapid decay dynamics with a time constant of 0.3 ps, and the ESA 3 band at 540 nm rise significantly on the same timescale (see
Figure 2c). This evolution is attributed to the ESIPT from the excited state of BBM-enol* to BBM-keto*. After 0.6 ps, as shown in
Figure 2d, the ESA3 band at 540 nm exhibits a dominant two exponential decay with time constants of τ
2 = 3.1 ps and τ
4 = infinitely long lifetime. Notably, the decay of the ESA3 band is followed by the increase of ESA 1 band signal at 383 nm in 3.1 ps (see
Figure 2d). The ESA 1 band at 383 nm then decays with two exponential processes of τ
3 = 1.6 ns, and τ
4 = infinitely long lifetime (see
Figure S4). The SE band at 450 nm reveals 1.6 ns decay dynamics indicating that the 1.6 ns is attributed to the spontaneous emission lifetime from the excited state of BBM-keto*, which was further confirmed using the fs up-conversion fluorescence spectroscopy at an excitation of 350 nm, as shown in
Figure 3. We utilized EOS-TA measurements in the ns ~ μs time range to determine the longest-lived components τ
4 in TA spectra. As shown in
Figure S5a,c in the Supplementary Material, a broad ESA band in the range of 370 - 850 nm exhibits single exponential decay with a time constant of 66 ns, indicating depopulation from BBM-keto triplet state T
1 state [
9,
29].
The previous reports have demonstrated that the significant role of twisting motion pivot on C23-C25 (C14-C15) bond was observed from the HBI-keto*, which results in fluorescence quenching via cis-trans photoisomerization [
6,
9,
30]. To probe possible photoisomerization of BBM in the course of the primary photo-processes, the sample was measured in THF with low viscosity using TA. As shown in
Figure S2b, the DADS of BBM in THF reveal four exponential decay components of 0.3 ps (τ
1), 2.6 ps (τ
2), 1.7 ns (τ
3) and an infinite lifetime (τ
4), the longest τ
4 time components are well resolved with a time constant of 862 ns, which is assigned to the triplet keto-T
1 state lifetime (see
Figure S5b,d). The spectral shape of each DADS component does not show differ from that in DMSO solvent, indicating that the intermediate states of the BBM molecule remain unchanged in THF (see
Figure 2b and
Figure S2b). Notably, the second component τ
2 has shorter lifetimes in the low viscous THF solvent versus that in DMSO, indicating that a conformational change occurs after ESIPT processes.
2.3. Femtosecond Fluorescence Up-Conversion Spectroscopy
Our TA spectra thus far suggest enol to keto ESIPT which is followed by the photoinduced conformational change of BBM chromophore. However, two open questions here remain unclear. First, is the conformational change caused by the C23-C25 and/or C14-C15 bond cis-trans photoisomerization or other types of distortion? Second, does the ESIPT occur at both HBI halves of BBM molecule? To address the first question, we obtained time-resolved up-conversion fluorescence spectra for BBM following 350 nm excitation (see
Figure 3 and S6). As shown in
Figure 3a, in DMSO, a blue emission band maximize at approximately 438 nm at time zero, as time elapsed, the emission band redshifts to a final position nearly corresponding to the steady-state emission spectrum with a maximum of 445 nm. The DADS spectra yield three components including τ
1 = 0.3 ps rising dynamics followed by τ
2 = 2.9 ps, and τ
3 = 1.3 ns (see
Figure 3b). The ultrafast rise with a time of 0.3 ps reflects the formation of the BBM-keto* component via the ESIPT. In particular, the transient intensity of emission on the blue side of the emission band (420 nm) exhibits an ultrafast rise (~ 0.3 ps) followed by a dominant fast decay component of 2.7 ps and a long decay lifetime of 1.3 ns (see
Figure 3c); at the longest emission wavelength of 480 nm, the kinetics is well fitted by a fast rise of 2.7 ps and a 1.3 ns single exponential decay components, in which the rise of 2.7 ps matches the decay component of the short emission wavelength at 420 nm and is consistent with the τ
2 component obtained from TA spectra. In THF solvent, as shown in
Figure S6b, similar excited-state features were observed as DMSO, three components of τ
1 = 0.3 ps, τ
2 = 1.6 ps, and τ
3 = 1.2 ns were required for best fit.
The transient amplitude and frequency change of the second component τ
2 in two solvents show distinct dynamics differences (see
Figure 3c, S7 and S8). Note that the solvent relaxation time of DMSO and THF is 1.8 ps and 0.9 ps, respectively [
31], in both solvents, as shown in
Figure S8 the prolonged transient dynamic stokes shift (3.9 ps in DMSO; 1.5 ps in THF) indicate that solvation and conformational change might occur simultaneously until the system is completely solvated, and that afterward, conformational change dynamics primarily control the excited state relaxation [
21,
32]. Additionally, the DADS spectrum of τ
2 displays a dispersive lineshape indicating a dynamic redshift with enhanced fluorescence intensity (see
Figure 3d). This phenomenon cannot be attributed to
cis-trans photoisomerization along C23-C25 or C14-C15 bonds in BBM-keto* form, as such photoisomerization would effectively quench fluorescence intensity and result in non-radiative decay. [
6,
9]. There is, however, a flexible point at the C4-C7 central bridge which may facilitate conformational changes by rotating the dihedral angle between two HBI halves of BBM-keto* form [
33,
34,
35,
36].
2.4. Femtosecond Raman Stimulated Spectroscopy
To summarize our findings thus far, the TA experimental and upconversion fluorescence results provide evidence for ESIPT and conformational change evolution in the BBM molecule. However, further support is required to precisely assign the τ
1 and τ
2 time constants. For τ
1, does the ESIPT occur at two HBI halves or only one of the HBI half in BBM? For τ
2, does the conformational change derived from the rotation through C
4-C
7 bond? To gain more experimental insights into conformational change during ESIPT, FSRS was used to track local vibrational marker bands on the excited states [
20,
21,
22,
24,
25,
37,
38]. We strategically selected 600 and 480 nm Raman pump, which correspond to the red shoulder of BBM-keto* and BBM-enol* ESA bands. This pre-resonantly enhances the FSRS signal on the Stokes side (see
Figure S10), providing the excited-state Raman signals that reveal intricate dynamics tracking ESIPT and the structural change of BBM molecules [
20,
21,
22].
Figure 4a reveals a 2D contour plot of FSRS spectra for BBM across the frequency range of 850 cm
-1 - 1700 cm
-1 in DMSO with Raman pump excitation of 600 and 480 nm (baseline drawn and FSRS raw data are shown in
Figure S9 in supplementary material). Five excited-state Raman peaks were observed at 953 cm
-1(I), 1130 cm
-1(II), 1358 cm
-1(III), 1565 cm
-1(IV), and 1587 cm
-1(V) under the Raman pump excitation of 480 nm (BBM-enol* form). A more detailed description of Raman modes at BBM-enol form was achieved through density functional theory (DFT) with the optimized geometry at the dihedral angles between two HBI halves at Θ = 39° and between benzene and imidazole rings at Φ = 0°. As illustrated in
Figure S11 and
Table S1, the observed Raman modes at 1565 cm
-1, and 1587 cm
-1 are mainly attributed to the C14-C15 (C23-C25) stretching accompanied by the benzene and imidazole rings deformation; the mode at 1358 cm
-1 is assigned to the benzene and imidazole rings stretching and C-H in-plane waging; 1130 cm
-1 is attributed to C-H in-plane waging motions of two benzene rings; 953 cm
-1 is attributed to C-H out-of-plane bending of two benzene rings. Those modes shift to 939 cm
-1(I), 1122 cm
-1(II), 1348 cm
-1(III), 1534 cm
-1(IV), and 1568 cm
-1(V) at the Raman pump of 600 nm (BBM-keto* form).
Figure 4b presents the transient amplitude of Raman modes (IV) and (V) of BBM in DMSO at different Raman pump wavelengths. Upon the excitation with a 480 nm Raman pump, both Raman modes exhibit two exponential decays with a time constant of τ
1 = 320 fs and an infinite long lifetime. The τ
1 is consistent with the ESIPT rate observed in TA and up-conversion fluorescence, indicating resonantly enhances excited state Raman signal of BBM-enol* form under the excitation of 480 nm Raman pump; the subsequent infinite long lifetime can be attributed to the decay of fluorescence emission from BBM-enol* state. By using 600 nm Raman pump, the transient amplitude of these two Raman modes exhibits a rising dynamics of τ
1 = 260 fs followed by two exponential decays with time constants of τ
2 = 3.6 ps and an infinite long lifetime. Moreover,
Figure S12 illustrates that τ
2 displays significant viscosity-dependent dynamics in THF (2.6 ps) and corroborates the aforementioned conformational change dynamics observed in TA and fs up-conversion spectra. Importantly, as shown in
Figure 4c, the difference in peak frequency between modes (IV) and (V) is 22 cm
-1 in BBM-enol* form (with Raman pump of 480 nm), which increases to 34 cm
-1 in BBM-keto* form (with Raman pump of 600 nm). As presented in
Table S1, the DFT calculations demonstrate a slight decrease in the peak frequency difference of the two Raman modes from BBM-enol (4 cm
-1) to BBM-keto (3 cm
-1) when the ESIPT occurs in both HBI halves. Conversely, if the ESIPT only takes place in one HBI half of BBM, there is a significant increase in the peak frequency difference from 4 cm
-1 in BBM-enol* form to 38 cm
-1 in BBM-keto* form, which aligns with the experimental observation. Additionally,
Figure 4c demonstrates that the FSRS spectra at 480 nm Raman pump excitation exhibit equal peak intensity for the two Raman modes (IV) and (V) with Gaussian line shape in the BBM-enol* form. However, in the BBM-keto* form (600 nm Raman pump), there is a difference in peak intensity between these two Raman modes due to ESIPT occurring in one HBI half of BBM, resulting in different Raman polarizability.
Figure 4d illustrates the transient frequency shift of mode (I) and (II) in BBM-keto* form with Raman pump excitation of 600 nm. The 934 cm
-1 mode rapidly blueshifts to 941 cm
-1 in 0.3 ps, followed by a slow redshift to 936 cm
-1 with time constants of 3.8 ps; while mode (II) exhibits two exponential frequency blueshifts from 1117 cm
-1 to 1128 cm
-1 with time constants of 0.3 ps and 3.8 ps. The initial blueshift of the two Raman modes can be attributed to the vibrational cooling process accompanied by ESIPT. To investigate the second transient frequency change in those modes, TD-DFT calculations were performed for the BBM-keto* form with ESIPT occurring in one HBI half of BBM. The optimized geometry at the S1 electronic excited state of BBM-keto* reveals the dihedral angle of Θ = 26° and Φ = ~ 0°. By reducing the Θ from 39° (the dihedral angle on S1 Franck-Condon state) to 26° while keeping the Φ = 0°, it was observed that the Raman modes (I) and (II) exhibit red shift and blue shift, respectively, during the twisting of Θ (see
Figure S13). This suggests that the second transient frequency change is caused by rotation of the dihedral angle Θ between two halves of BBM, ultimately leading to a planarized BBM-keto* isomer.
At last, to clarify the electron transitions responsible for the dynamic red shift fluorescence spectra observed in fs fluorescence up-conversion spectra, we used TD-DFT to calculate the energy gap and the oscillator strength of the S1
→ S0 transition of BBM-keto* form (ESIPT occurs in one HBI half of BBM) at different torsion angles Θ = 26°, 30°, and 35° in the S1 excited state. The results presented in
Table S2 demonstrate an evident fluorescence spectral redshift and an increase in oscillator strength as the Θ is tuned from 39° to 26°. This finding is consistent with the experimental observation of fs fluorescence up-conversion spectra, further confirming that the planarization via Θ dihedral angle torsion plays a crucial role in the excited state relaxation of BBM keto* state.