Palak
Mandal
and
Aditya N.
Panda
*
Department of Chemistry, Indian Institute of Technology Guwahati, Guwahati, India. E-mail: anp@iitg.ac.in
First published on 11th February 2026
We report on non-adiabatic surface-hopping dynamics of exciton localization and delocalization processes in two 2-phenylpyridine oligomers: a linear and a helical one. In the linear pentamer, excited-state relaxation involves the delocalization of the exciton across multiple monomer units. In contrast, the relaxation in the helical pentamer proceeds with the localization of the exciton on a single monomeric unit.
Modeling the photo-induced processes mentioned above poses significant challenges, as it involves calculating the energies of multiple electronic states and the non-adiabatic coupling between them. In this regard, trajectory-based quantum–classical methods, such as Ehrenfest dynamics and trajectory surface hopping (TSH) dynamics,17,18 have been useful for addressing the above issue. TSH is the most widely used method for simulating non-adiabatic dynamics, due to its simplicity, practicality, and scalability.19 This method requires the calculation of electronic energies, gradients and couplings between excited states on-the-fly, which is computationally very intensive for larger conjugated systems, even when employing commonly used time-dependent density functional theory (TD-DFT) methods. In recent times, many semiempirical quantum methodologies have been developed to alleviate the problem of resources and time for larger systems.13,20–22 In this context, the semiempirical time-dependent density functional tight binding (TD-DFTB) method, which is analogous to TD-DFT but employs parameterized integrals with smaller basis sets, has also become very useful.23 In the past few years, photo-induced dynamics employing TSH with TD-DFTB have been studied for a large number of conjugated systems, including oligomers of phenylenes, and acene derivatives.24–28
Oligophenylene and its conformational isomers, such as ortho, meta, and para-oligophenylene, have long been studied due to their interesting optical and electronic properties.29–33 However, recent studies have shown that the presence of electron-withdrawing pyridine rings in oligomers results in significant alterations in the electrochemical and optical properties compared to the homo-oligomeric forms.34–37 This suggests that the introduction of pyridine rings into oligophenylenes may also lead to significant changes in their electronic and optical properties. In this work, we have considered two different conformers A and B of oligomers of 2-phenylpyridine (PhPy), shown in Fig. 1, to investigate how the excited states evolve after photoexcitation. This follows our previous study on the oligomers of PhPy,38 which discussed the static excited state properties of three different conformers of oligomers. In conformer A, the oligomers are formed by connecting the para-position of the phenylene ring of one PhPy unit with the meta-position of the pyridine ring of another PhPy unit. On the other hand, the oligomers of conformer B are formed by connecting the para- and ortho-positions of the phenylene and pyridine rings of two different PhPy units, respectively. Conformer A remains linear as we move from (PhPy)2 to (PhPy)5, while conformer B adopts a helical-like structure for (PhPy)5 (Fig. 1). The TSH dynamics was carried out in combination with the TD-DFTB method. In addition to TD-DFTB, the vertical excitation energies (Eg) are also calculated at the long-range corrected (LC) version of TD-DFTB (LC-TD-DFTB),39,40 and these results are compared with the results obtained at the algebraic diagrammatic construction scheme to the second order (ADC(2)) level.41–43 The details of computational methods are mentioned in the SI.
![]() | ||
| Fig. 1 Ground state optimized geometries of dimers and pentamers of conformers A and B obtained at the B3LYP-D3BJ/def2-TZVP level. | ||
The Eg values and the corresponding oscillator strengths (fosc) for the first ten excited states calculated on the previously optimized ground state geometries38 are tabulated in Table S1. In the case of A, the TD-DFTB results show a good match with the RI-ADC(2) results in both dimer and pentamer. The Eg values for the dimer obtained at the TD-DFTB level differ by 0.06 to 0.27 eV from the RI-ADC(2) results, while this energy difference is slightly higher in the pentamer (0.47–0.70 eV). Both the TD-DFTB and RI-ADC(2) results predict a bright S1 state with considerable fosc value in the dimer and pentamer; however, the LC-TD-DFTB predicts a small fosc value for the S1 state in the dimer (see Table S1). In B, though Eg values for the first ten excited states found using TD-DFTB are close to those observed in the RI-ADC(2) results, the values for fosc differ from the RI-ADC(2) results. For example, in the dimer, S3 is the brightest state in the TD-DFTB results, whereas the RI-ADC(2) results indicate S1 to be the brightest state with an fosc value of 0.882. For the pentamer, the RI-ADC(2) results show the highest fosc value of 1.465 for the S2 state, whereas the TD-DFTB results show small fosc values for all excited states. A similar observation is also noted in the case of the LC-TD-DFTB results.
The photo-induced dynamics of the two dimers starting from the S3 bright states are shown in Fig. 2 (top panel). As the figure shows, conformer A undergoes ultrafast internal conversion within 10 fs from the initially populated S3via the S2 state to the S1 state. Conformer B shows slightly different behavior as the population of the initially populated excited state (i.e., S3) gets transferred to a higher excited state (S4) at first, before relaxing to the S1 state. This is due to the strong non-adiabatic coupling between the S3 and S4 states in that region. For both conformers, after the initial 50 fs, 60% of the total population remains in S1 and the remaining 40% is shared among the S2, S3, and S4 states, within the considered time frame. The bottom panel of Fig. 2 shows the evolution of average dihedral angles ϕ1, ϕ2, and ϕ3 (indicated in Fig. 1) of the two conformers with time. For A, at t = 0, the molecule has a non-planar backbone with average ϕ1, ϕ2, and ϕ3 values around −20°, 36°, and −19°, respectively. As the time progresses, the ϕ values decrease, and at around 150 fs the values of all three dihedral angles become ≈0°. In the case of B, the dynamics started again with a non-planar geometry (with ϕ1, ϕ2, and ϕ3 values of ≈24°, −20°, and 22°, respectively). However, the geometry never reaches planarity during the entire time, although the values of ϕ2 and ϕ3 become close to 0° around 140 fs.
![]() | ||
| Fig. 2 Average populations of the first four excited states (top panel), and average dihedral angles (bottom panel) vs. time for the dimers. | ||
To gain insight into the nature of the excited states during structural evolution, we performed an excitonic analysis of the geometries at various time frames extracted from a single representative trajectory. The charge transfer parameter (ωCT) values, electron–hole (e–h) correlation plots, and natural transition orbitals (NTO) have been considered for describing the characters of the excited states. Detailed descriptions of the ωCT and e–h correlation plots can be found in the SI. The values of Eg, fosc, and ωCT corresponding to the current state at two different time frames for (PhPy)2-A and B are tabulated in Table S2. A closer look at the ωCT values of the first ten excited states, including the current state, indicates that the ωCT values calculated at the TD-DFTB level are somewhat larger compared to those obtained at the RI-ADC(2) level. This is expected as the conventional DFTB is parameterized for the generalized-gradient approximation (GGA) functional, which has a problem in describing the CT character.44 Considering this, the characterization of the excited state is performed considering the RI-ADC(2) results. The TD-DFTB results are presented in the SI.
At 0° fs, the RI-ADC(2) results show that the current states (i.e., S3) of (PhPy)2-A and B have ωCT values of 0.16 and 0.10 (see Table S2), respectively, indicating their local excitation (LE) characters. The NTOs involved in these states are depicted in Fig. 3. For A, at 0 fs, both the h and the e orbitals are delocalized over the entire dimer backbone, showing the delocalized LE character of this state. When the conformer A becomes planar at 138 fs, the current state (S1) still exhibits a small ωCT value of 0.07. However, the NTOs reveal that the h and e orbitals are now localized on a single monomeric unit, indicating the localization of the exciton during structural relaxation. At 0 fs, the e–h correlation plot (Fig. S3) shows contributions from both the diagonal blocks, whereas at 138 fs, only one diagonal block contributes, highlighting the shift from delocalized to localized LE characteristics of the current states. In contrast, for conformer B, both at 0 and 141 fs, the h and e orbitals are mostly confined to the same monomeric unit (Fig. 3), indicating the localized LE characters of the states. The same is observed from the e–h correlation plots (Fig. S3), where single intense diagonal blocks are observed at both 0 and 141 fs.
![]() | ||
| Fig. 3 Geometries and NTOs involved in the current states of a single representative trajectory of (PhPy)2-A and B obtained at the RI-ADC(2) level. | ||
The excited-state dynamics results for the pentamers are shown in Fig. 4. The photo-induced dynamics of A and B start from S4 and S3 bright states, respectively. As in the dimers, here too, both the conformers undergo ultrafast internal conversion within 10 fs after photo-excitation from the initial excited states (S4 or S3) to the S1 states. Beyond this, constant fluctuations in the electronic population are observed for both conformers. This population fluctuation is primarily due to the proximity of the energies of adjacent electronic states in the pentamers. The potential energies for the first four excited states of a single representative trajectory for (PhPy)5-A and B are presented in Fig. S4 and S5, respectively. The figures show that the energy difference between the adjacent excited states is so small that the four states appear to almost coincide.
![]() | ||
| Fig. 4 Average populations of the first four excited states (top panel), and average dihedral angles (bottom panel) vs. time for pentamers. | ||
The evolution of the average dihedral angles (ϕ1–ϕ9) for the two conformers is also depicted in Fig. 4 (bottom panel). In conformer A, the molecule initially displays a non-planar geometry, with the angles ϕ1, ϕ3, ϕ5, and ϕ7 ranging from −15° to −20°, while ϕ2, ϕ4, ϕ6, and ϕ8 are approximately between 30° and 45°. On the other hand, the value of the terminal dihedral angle ϕ9 is around 8°. As the dynamics progress, all the dihedral angles tend to converge towards 0°. In (PhPy)5-B, on the other hand, the value of ϕ1 remains ≈20° over the whole 300 fs range, while other angles approach 0°.
The values of ωCT corresponding to the current states at different times during the excited state relaxation process for (PhPy)5-A and B are tabulated in Tables S3 and S4, respectively. As expected, the ωCT values corresponding to the first ten excited states, including the current states, calculated at the TD-DFTB level, are significantly larger than those obtained at the RI-ADC(2) level (see Tables S3 and S4). Therefore, we focus solely on the RI-ADC(2) results. For (PhPy)5-A, as shown in Table S3, the ωCT values for the current states at times 0 fs (S4), 2 fs (S3), 13 fs (S1), and 101.5 fs (S1) are 0.51, 0.48, 0.44, and 0.51, respectively. Notably, at 13 fs, the fosc values for the S2 and S1 states (Table S3) indicate that the energetic order of these states is reversed in the RI-ADC(2) results compared to the TD-DFT results. The NTOs corresponding to the current states during excited state relaxation are presented in Fig. S6. At 0 fs, the two pairs of NTOs suggest that both the e and h orbitals are localized on the terminal PhPy units, indicating a localized LE state. The e–h correlation plot shown in Fig. S7 supports this observation, showing a single intense diagonal block that signifies the localized LE state. As the dynamics progress, the h and e orbitals begin to delocalize across multiple PhPy units. At 2 fs, the h and e orbitals are delocalized over one side of the pentamer unit; at 13 fs, they are delocalized over the middle PhPy units; and at 101.5 fs, they are delocalized over the left-side PhPy units. The e–h correlation plots further corroborate these findings (see Fig. S7). For (PhPy)5-B, the excited state results are different from the results of (PhPy)5-A. In this case, the NTOs corresponding to the current states indicate that the h and e orbitals are confined over a single monomeric unit, but their locations are different at different times, as shown in Fig. S8. The e–h correlation plots also suggest the same, as only a single diagonal block contributes to the electronic excitation shown in Fig. S9.
The overall excited state relaxation process in (PhPy)5-A and B is presented in Fig. 5. In A, after photoexcitation, the excited state relaxation process proceeds via delocalization of the exciton from an initially localized LE state S4 to the lowest energy excited state S1. As the figure shows, at 0 fs, the S4 state is localized on a single PhPy unit indicated by an orange block at the top right. When the system relaxes to S3 from the S4 state at 2 fs, the exciton starts to delocalize over the right-side PhPy units, indicated by a large diagonal width from the top right to the middle blocks. At 13 fs, the exciton is delocalized over the middle three blocks, and finally, when the molecule relaxes to the lowest energy S1 state, the exciton gets delocalized over the bottom left PhPy units. In contrast to A, the exciton is localized in a single PhPy unit during excited state relaxation in B. Initially at 0 fs, the exciton is localized over one of the terminal PhPy units, indicated by an intense diagonal block at the top right. Because of its helical structure, the localized exciton is transferred from one terminal PhPy unit to another terminal PhPy unit during S3–S2 relaxation at 3 fs (indicated by a bottom left intense block). Relaxation to the S1 results in shifting of the localized exciton from the terminal PhPy unit to the adjacent PhPy unit.
![]() | ||
| Fig. 5 Schematic representation of the excited state relaxation process in pentamers of conformers A and B. | ||
In this work, we aimed to understand the excited-state dynamics of two different conformers, A and B, of the PhPy oligomers. On-the-fly non-adiabatic surface hopping dynamics was carried out using TD-DFTB. In dimers, excited state relaxation leads to a planar structure in conformer A, while conformer B remains non-planar for up to 300 fs. In (PhPy)2-A, structural relaxation results in localization of the exciton in the S1 state from a delocalized LE state S3. Meanwhile, in conformer B, the exciton stays localized throughout the excited relaxation process. As the chain length increases, the linear conformer (PhPy)5-A shows exciton delocalization across multiple monomeric units during relaxation from the high-energy excited state (S4) to the lowest energy excited state (S1). In comparison, the helical conformer (PhPy)5-B exhibits a movement of the localized exciton from one monomeric unit to another during the relaxation from the S3 to S1 states. The delocalization of the exciton in the linear conformer indicates that this conformer may act as a better exciton transport material over the helical form in the field of organic electronics. Our present results could be useful in understanding the exciton (de)localization processes in similar oligomers and polymers, and in designing devices with improved performances.
ANP conceived the project; PM carried out the calculations; PM and ANP analyzed the results; PM and ANP wrote the manuscript.
| This journal is © The Royal Society of Chemistry 2026 |