Probing the electronic structure and photophysics of thiophene–diketopyrrolopyrrole derivatives in solution

Diketopyrrolopyrroles are a popular class of electron-withdrawing unit in optoelectronic materials. When combined with electron donating side-chain functional groups such as thiophenes, they form a very broad class of donor–acceptor molecules: thiophene–diketopyrrolopyrroles (TDPPs). Despite their widescale use in biosensors and photovoltaic materials, studies have yet to establish the important link between the electronic structure of the specific TDPP and the critical optical properties. To bridge this gap, ultrafast transient absorption with 22 fs time resolution has been used to explore the photophysics of three prototypical TDPP molecules: a monomer, dimer and polymer in solution. Interpretation of experimental data was assisted by a recent high-level theoretical study, and additional density functional theory calculations. These studies show that the photophysics of these molecular prototypes under visible photoexcitation are determined by just two excited electronic states, having very different electronic characters (one is optically bright, the other dark), their relative energetic ordering and the timescales for internal conversion from one to the other and/or to the ground state. The underlying difference in electronic structure alters the branching between these excited states and their associated dynamics. In turn, these factors dictate the fluorescence quantum yields, which are shown to vary by ∼1–2 orders of magnitude across the TDPP prototypes investigated here. The fast non-radiative transfer of molecules from the bright to dark states is mediated by conical intersections. Remarkably, wavepacket signals in the measured transient absorption data carry signatures of the nuclear motions that enable mixing of the electronic-nuclear wavefunction and facilitate non-adiabatic coupling between the bright and dark states.


Transient Absorption Spectroscopy (a) Wavepacket analysis of Toluene
Vibrational wavenumbers associated with toluene averaged over the 550-700 nm window. A list of the observed wavenumbers associated with toluene signals is given in Table S1.

Table S1
Wavenumbers associated with peaks in the averaged toluene spectrum shown in Fig. S2 compared to previously experimentally determined Raman wavenumbers of toluene (from refs 1,2

(b) DPPDTT in Toluene
The main manuscript details transient measurements of DPPDTT acquired in chloroform solution. The equivalent data acquired in toluene are displayed in Fig. S3(a). Due to the low solubility of the polymer and some precipitation in toluene, significant scattering is observed in the associated TA data, resulting in far lower signal-to-noise ratios compared to data obtained in chloroform (Fig. 7). To alleviate some of these issues, the frequency and time domains were smoothed by adjacent averaging and the use of a Savitzky-Golay filter, respectively. A comparison of the kinetics for DPPDTT in the two solvents for three probe wavelengths are shown in Fig. S3(b-d), and show clear similarities.

(c) DPPDTT Rate Model
A rate model for both chloroform and toluene measurements was constructed to dissect the complex polymer dynamics: the photoexcited kinetics of DPPDTT were described using a generic rate model (eq. S1), where y(n) and y(m) are the population of the n th and m th states. km is the decay rate constant for y(m), which also describes the rate of y(n) formation. The population of y(n) subsequently decays with rate constant kn. Combined, the time dependent population of y(n) can be described: .
(S1) There are two possible origins for the range of lifetimes observed in the polymer, arising from either large-scale conformational inhomogeneity, such as planar vs. twisted polymer chains, or differences in conjugation lengths on different chain segments. As discussed in the main text, because the absorption maximum of the polymer matches that of the dimer, the delocalisation is considered to be spread over ~2 repeat units ( Fig. 1(d))-a conclusion supported by a recent The kinetic fits to chloroform data are displayed/discussed in the main text (Fig. 7). A similar model was constructed for DPPDTT in toluene, and the results from this analysis are displayed in Fig. S4. Notably some of the time constants are altered, which is proposed to be due to slight changes in the relative energies of the electronic states, however no additional kinetic components are required to describe the data. Given the excellent match between experimental data and modelled kinetics, it is evident that a three-state model robustly describes the excited state dynamics of DPPDTT.

Kinetic Fitting Parameters
All data were chirp corrected prior to kinetic analysis. The fits to data shown in the main manuscript were modelled using an analytical expression comprised from the convolution of a gaussian IRF with multiple exponential rise/decays. The instrument response function was locked to the pump pulse duration determined by PG-FROG.

TDPP-v-TDPP and TDPP synthetic methods
Commercially available reagents were purchased and used without further purification unless otherwise stated. 3-(5-Bromothiophen-2-yl)-2,5-bis(2-ethylhexyl)-6-(thiophen-2-yl)-2,5dihydropyrrolo [3,4-c]pyrrole-1,4-dione was synthesized following the published route. 5      For the monomer and R = -H, calculations predicted the global ground state minimum energy geometry to be planar and C2h (see Table S3). Upon changing the R substituent to -CH2CH(CH3)2, a sidechain group that is more representative of the true branched alkyl chains (see Fig. 1 in main paper), an added complication arises from the conformational complexity associated with the isobutyl substituents. In geometries where the hydrogens on the -CH2 group of the isobutyl substituents point towards the thiophene ring ( Fig. S19(a,b)), the terminal thiophene rings twist away from the planar DPP core in opposite directions (a = 32°), thus forming a structure which has overall Ci symmetry. While, if the alkyl chains are rotated to avoid steric interactions with the aromatic core of the molecule (Fig. S19(c,d)), a reduces to 6 °. Locking this structure to planarity (C2h) and re-optimisation returns a ground state minimum structure with an associated minimum energy only 11 meV above that of the twisted Ci geometry (a = 6 °). Thermal energy at 298 K is 26 meV, and therefore it is likely that the ground state geometry will quickly interconvert (depending on the barrier height) between these two (and potentially other) minima. Thus, in a time-averaged picture, the planar C2h minima is representative of ground state TDPP-Br.
Calculations for the dimer (full results detailed in Table S3) with methyl and isobutyl R groups confirmed similar behaviour to the monomer: the C2h minimum (R = -isobutyl) was calculated to be 0.8 meV above that of the global Ci minimum energy geometry-a value which is insignificant compared to solvent fluctuations or thermal energy at 298 K, again justifying approximating a C2h structure for TDPP-v-TDPP.   Figure S21(a,b), and S21(c,d) respectively.
TD-DFT calculations (again using B3LYP/6-311G(d,p)) returned similar results for the excited bright state (1Bu) with isobutyl substituents, where the C2h structure was 14 meV higher in energy than the Ci minima. Calculations for the dimer returned a 30 meV energy difference.
These global minimum energy geometries were used for vibrational normal mode calculations used to assign observed wavenumbers from the wavepacket FT analysis. For these vibrational calculations, R = -CH3 was used to minimise the computational cost. For TDPP-Br and TDPPv-TDPP, the vibrational normal modes and associated wavenumbers the S0 (1Ag) and bright excited (1Bu) electronic states were calculated using DFT and TD-DFT, respectively. The ground state minimum energy geometries were optimised and vibrational wavenumbers calculated with the DFT exchange-correlation functional B3LYP and a 6-311G(d,p) basis set.
The same basis function and basis set were used with TD-DFT to calculate the associated vibrational wavenumbers of the 1Bu excited states.