Is the Chemical Strategy for imbuing ‘Polyene’ Character in Diketopyrrolopyrrole Based Chromophores sufficient for Singlet Fission?

. In this work, we have rationally designed and synthesized a novel thiophene diketopyrrolopyrrole (T DPP ) - vinyl based dimer. We have investigated the optical and electronic properties and have probed the photophysical dynamics using transient absorption to investigate the possibility of singlet exciton fission. These revealed extremely rapid decay to the ground state (< 50 ps), which we confirm is due to intramolecular excitonic processes rather than large scale conformational change enabled by the vinyl linker. In all cases the main excited state appears to be ‘dark’, suggesting rapid internal conversion into a dark 2A g - type singlet state. We found no evidence of triplet formation in TDPP - V - TDPP under direct photo excitation. This may be a consequence of significant singlet stabilization in the dimer, bringing it below the energy needed to form two triplets. Our studies on this model compound set valuable lessons for design of novel triplet - forming materials, and highlight the need for more broadly applicable design principles.

KEYWORDS singlet fission, donor-acceptor, vinyl, DPP, polyene, internal conversion, transient absorption Considerable efforts have been directed towards the development of efficient excitonic solar cells based on organic materials, through a multitude of advancements in the design of appropriate molecular backbones, thin-film processing conditions and device-architectural engineering. 1, 2, 3 ,4, 5 To date, the highest power conversion efficiency achieved in organic solar cells is in the range of ~10%. 6 One critical limitation for these and indeed all other singlejunction solar cell technologies arises from fundamental thermodynamic processes such as relaxation of high-energy carriers/excitons to the band edge. The consideration of such processes yields the Shockley-Quiesser efficiency limit of ~33% for single junction solar cells. 7,8 One of the ways to circumvent this limitation is by invoking and implementing the concept of singlet exciton fission (SEF). 8,9 This process involves the conversion of a singlet exciton into a pair of low-energy triplet excitons, which are initially coupled into an overall singlet state. This maintenance of singlet spin enables SEF to proceed on ultrafast time scales to out-compete other loss channels. 10 Efficient SEF relies on the correct alignment of the singlet and triplet energy states, such that 2 E (T 1 ) ≤ E (S 1 ). 11 By thereby splitting a high-energy exciton into two low-energy excitons, it becomes possible to harvest a significant proportion of the energy in 'visible spectrum' photons typically lost to thermalization, resulting in a new maximum efficiency limit of ~44%. 12 In the recent years, quantum chemical and experimental studies have demonstrated that acenes, for example pentacene, exhibit efficient SEF with ~200% triplet yield on timescales as fast as 80 fs. [13][14][15][16][17][18][19] SEF proceeds either through the intermolecular or intramolecular mechanism. [20][21][22][23][24] For both mechanisms, SEF rates and efficiencies can be controlled by fine-tuning the strength of interchromophore coupling, which in the case of intermolecular fission entails the difficult task of optimizing chromophore packing within the film. 18,24,25 Furthermore, several groups have demonstrated that coupling between singlet and triplet-pair manifolds can be very effectively mediated by intermediate charge-transfer (CT) states. [21][22][23]26 This insight provides design criteria for the selection of building blocks for new intramolecular SEF materials that satisfy two key requirements for efficient SEF: an increased S 1 -T 1 gap (to achieve 2x E (T 1 ) ≤ E (S 1 )), and a lowest-lying optical excitation with significant chargetransfer character that can mediate the SEF process. [27][28][29][30]31 There is little knowledge about the degree of interchromophore coupling needed to make SEF competitive with fluorescence and to yet allow the component triplets of the singletcoupled triplet pair to 'escape' and behave as independent excitations. 10,32,33 It is evident, however, that these processes require a delicate balance of chemical design parameters. These processes depend largely on the crystal environment, whether by positioning the molecules favourably for direct interaction or by stabilizing the low-energy charge transfer states which serve as virtual states in the coupling process. 26,34 The interchromophore interaction can range from mere physical contact through indirect covalent coupling to direct covalent bonding that permits strong -conjugation. The extent of coupling has direct effects on the SEF rate, but also indirectly affects the process by modifying the energy of the states of the isolated monomers and thus the exoergicity of SEF. 33,35 This interchromophore coupling can be controlled by tailoring the chemical structure. 10,18,24,36,[37][38][39][40] One rational design approach to optimize this coupling for intramolecular SEF materials is based on alternately coupling electron-rich and electrondeficient moieties to form a quinoidal D-A-D backbone imparting significant charge-transfer character to the lowest energy optical excitation. 18,26,32,41 The resulting charge-transfer state possesses a large absorption cross-section and may couple more strongly to multi-exciton triplet states than pure singlet Frenkel excitons. 27,42 We highlight in particular the design strategy proposed with this motif for conjugated polymers and oligomers. In that work, the oxidized thiophene acceptor found to be necessary to enable SEF simultaneously endows materials with significant oligoene/polyene character, suggesting a judicious balance of charge-transfer and polyene character is needed. 26,43 Indeed, the property of intramolecular SEF can also be achieved by incorporating oligoene-like moieties. 44,45 In conventional polyenes, the lowest 2A g states lie below the 1B u states, enabling rapid (few ps) internal conversion to this 'dark' singlet that renders them largely non-emissive. It has long been considered that the 2A g state is crucial to enable SEF, as it can be alternatively described as a bound triplet pair. 10,46 However, more recent experimental evidence in oligoene aggregates 45,47 and conjugated polyenes 48 has revealed direct triplet formation instead from the bright 1B u state. It thus remains unclear what aspect of oligoene character enables SEF and precisely how the coupling between manifolds is achieved, though it is clear empirically that it is a powerful 'ingredient' for SEF materials. 26,48 Among the reported intramolecular SEF materials, vinylene-containing polymers have typically exhibited some form of optical activation, with fission enhanced by or even requiring excitation with a photon above the bandgap. 26,48 The lack of intramolecular SEF in shorter oligoenes such as the carotenoids in spite of suitable energetics suggests that there is also a minimum size requirement for this SEF motif. On the other hand, non-activated intramolecular singlet fission has been observed in several pentacene and tetracene dimers, with efficiencies as high as 200%. 21-23, 24, 28, 38, 39 These results indicate that intramolecular SEF may be possible as long as two distinct sites for the product triplets are available.
The chemical design of charge-transfer SEF chromophores can be challenging, as it not only involves increasing the S 1 -T 1 gap, but also endowing the S 1 state with sufficient chargetransfer character to enhance its coupling with the multiexcitonic triplet state. The former involves either the destabilization of the S 1 state or stabilization of the T 1 state by fine-tuning chromophore structural motifs. The latter is feasible when there is a large electronic coupling between donor and acceptor leading to a large HOMO-LUMO spatial overlap, hence large exchange energies. 49 The present understanding and the scarcity of appropriately designed materials provide little insight into the possible ways of harvesting multiple excitons in molecular materials. Diketopyrrolopyrrole (DPP)-based polymers and oligomers have attracted attention by virtue of high efficiencies in solar cells (~10%) and high hole (~20 cm 2 V -1 s -1 ) and electron mobilities (~12 cm 2 V -1 s -1 ) in OFETs. 50,51 Their attractive properties such as high electron affinities and relatively low triplet energies can lead to the exploration of novel design principles for next-generation singlet fission materials based on DPP oligomers and polymers.
Recent work has suggested that DPP derivatives can undergo SEF in the solid state 34  In keeping with the aforesaid design principle of combining charge-transfer and polyene character, we coupled together two thiophene-DPP molecules (TDPP) with a vinyl linker to control the interchromophore coupling, yielding the dimer TDPP-V-TDPP. UV-Vis absorption measurements reveal substantial interaction between the DPP units, while photoluminescence (PL) spectroscopy shows the molecule to be non-emissive as typically observed in polyenes.
Cyclic voltammetry (CV) measurements have been performed to elucidate the ionization potential and electron affinity. In order to probe the excited-state dynamics and better understand the interplay of bright and dark states, we performed transient absorption measurements on TDPP-V-TDPP in solution as well as in rigid polystyrene matrices. We observe in all cases that the initial 'bright' excited state appears to rapidly decay by internal conversion into a dark and extremely short-lived 2A g -type singlet, analogous to 'polyene' photophysics, because of the vinyl bridging unit. There is no evidence of triplet or charge formation, nor does substantial conformational change drive the rapid excited-state deactivation, which we consider instead to be analogous to the fast vibrationally-mediated decay active in carotenoids and other polyenes.
Our experiments demonstrate that SEF is inactive in this material, which is likely a consequence of the strongly reduced singlet energy upon dimerization. Our study nonetheless sets important lessons for the design principles of DPP-based oligomers for future applications as nextgeneration SEF materials.
TDPP-V-TDPP was synthesized using a Stille coupling reaction involving a monobromo derivative of DPP and 1,2-bis(tri-sec-butylstannyl)ethene in molar ratios of 2:1 using catalyst Pd 2 (dba) 3 and ligand P(o-tol) 3 in dry toluene as solvent. The synthesis of TDPP-V-TDPP is shown in Scheme 1. DPP was alkylated with the short hexyl chain to enhance the crystallinity.
The intermediates and the final compounds were purified by silica gel chromatography. The  Instead, we attribute these bands to twisted conformers in which the conjugation between TDPP units is minimal. Heating the solution causes a reduction in the 0-0/0-1 vibronic ratio, consistent with increased torsional disorder on average, but the relative weight of these higher-energy peaks does not vary (Figure S4, SI). The barrier for interconversion between these states must be substantial, suggesting that these monomer-like bands represent a different stereoisomer which was kinetically trapped in a twisted configuration during synthesis.
We could detect no luminescence from TDPP-V-TDPP, whether in a standard fluorimeter (excitation from 400 nm to 750 nm) or following laser excitation at 550 nm or 650 nm. We thus conclude that the PL quantum efficiency in the dimer is extremely low, due to a short singlet excited-state lifetime. This result suggests similar behaviour to the polyenes, where rapid (<200 fs) deactivation of the bright 1B u state through internal conversion to 2A g results in PL quantum yields below 10 -5 . The cyclic voltammograms are shown in Figure S3, SI. The extracted oxidation and reduction potentials are -5.00 eV and -3.44 eV respectively and the electrochemical gap is 1.56 eV. TDPP-V-TDPP exhibits reversible reduction cycle and a quasireversible oxidation cycle.
The low PL efficiency of TDPP-V-TDPP means that other spectroscopic tools are needed to build a comprehensive picture of the excited-state decay processes, namely transient absorption (TA) spectroscopy using a system described previously. 48 In this technique, the sample is excited with a short, tuneable pump pulse, followed at a precisely controllable time Because this technique detects excited-state absorption, it is particularly well suited to materials and states which do not strongly emit, such as TDPP-V-TDPP.
In Figure 2, we present TA spectra (a, c) and decay kinetics (b, d) of TDPP-V-TDPP in chlorobenzene solution. Under band-edge excitation (Figure 2, a-b), we observe behaviour reminiscent of polyenes. The dominant T/T > 0 bands closely agree with the absorption spectrum (grey dashed), enabling assignment to ground-state bleach (GSB). Due to the absence of any additional positive features that could be attributed to stimulated emission, we consider the primary species observed (colored traces) to be an optically 'dark' state. On the very earliest timescales (thin black line) we detect additional spectral signatures such as a broadened photoinduced absorption (PIA) in the NIR, with distinct absorption >1075 nm, and a crossing point between positive and negative TA signals that is ~10 nm shifted to the red. The presence of this shifted crossing point is a potential sign of stimulated emission of photons from the bright initially excited state, characterized by extremely short lifetime. These features we assign as signatures of the initial singlet excited state, which converts on ultrafast timescales to a shortlived dark state, analogous to the rapid 1B u 2A g internal conversion well known in the polyenes. The timescale of this process is within our instrument resolution and should be <200 fs; the existence of such rapid internal conversion is consistent with the absence of detectable fluorescence. The lifetime of the second state is well described with biexponential decay, with the primary decay attributed to the 3.6 ps component. There is no evidence of changing spectral shape over the lifetime of this state, suggesting that the two time constants do not reflect internal conversion between different excited states.
Similar basic features are evident following excitation with excess pump photon energy, as at 540 nm (Figure 2, c-d). The same signatures of initial (thin black) and longer-lived (blue) Figure 2. Sub-ps transient absorption measurements of TDPP-V-TDPP in solution with excitation (a, b) at the band edge and (c, d) well above the band edge reveal rapid internal conversion and excited-state quenching. (a) The initial excited state (thin black) converts within the instrument response to a second 'dark' state which exhibits no sign of stimulated emission. No further spectral evolution is detected during the rapid decay to the ground state. By 100 ps no signal can be detected. The UV-vis absorption spectrum (grey dashed) shows a close match to the positive TA signals observed, suggesting none can be attributed to stimulated emission. (b) Decay kinetics integrated over the bands indicated in (a). Following the initial fast kinetics, best observed in the 950 and 1150 nm traces, the kinetics can be well described with biexponential decay with the indicated time constants. (c) Excitation with significant excess energy results in the same overall spectral behaviour, but with slight broadening and a change in the shape of the ground-state bleach. New long-lived features are apparent, with faint bleach and accompanying photo-induced absorption detected at 100 ps and beyond. These features agree well with those of TDPP monomers. (d) The decay kinetics following excess-energy excitation reveals the same primary decay processes, but an additional decay component must be included in the fit to describe the new long-lived state. A very small fraction of the population persists beyond the limits of the measurement, into the nanosecond regime. Fits compared to raw data in Figs. S6-7, SI. states can be observed, with only a slight difference in the shape of the GSB. However, on the longest timescales (100 ps and beyond) we can distinguish a faint GSB band at 600 nm and corresponding PIA at 700-800 nm. These are signatures of a new longer-lived state. We find an additional exponential component (126 ps) is needed to describe the decay dynamics, and a very small fraction of the population persists beyond the 1.5 ns range of our measurements. Such effects could be consistent with optically activated singlet fission, as has been observed in other polyenes, or the activation of other decay channels through excess-energy excitation. To clearly establish the nature of this behaviour, we systematically measured the pump wavelength dependence. Figure 3, we detect qualitatively similar spectral effects of increased pump Figure 3. TA measurements at a range of excitation wavelengths reveal systematic spectral changes detectable at (a) early and (b) longer pump-probe delays. As the pump photon energy is increased, we detect an increase in the PIA signal in the NIR relative to GSB, and a shift in the peak ratio of the GSB spectrum (increasing 0-1 contribution). On longer timescales the enhancement of PIA and GSB features is still more pronounced, as excess photon energy slightly enhances the excited-state lifetime as well. (c) Integrated decay kinetics in the NIR show a small, systematic increase in the excited-state lifetime with pump photon energy until a sharp increase for 515 nm and 500 nm. This dramatic change is attributed to monomer-like TDPPs, whether arising from a small fraction of uncoupled precursors or from particular dimer conformations in which the TDPP units are not well conjugated. The smaller, systematic changes in the lifetime and TA spectra (particularly the GSB) reflect the pump-wavelength selectivity of particular sub-populations of TDPP-V-TDPP with varying conjugation length (i.e. torsional disorder). Full data presented in Fig. S8, SI. photon energy both at (a) early and (b) long pump-probe delays. The primary effects are an increase in the relative weight of the PIA in the NIR and a change in the vibronic peak ratio in the GSB, with a progressive increase and broadening of the 0-1 peak.

As shown in
These changes are particularly pronounced on longer timescales, as increased pump photon energy also increases the excited-state lifetime. In addition to this subtle but systematic effect in the decay kinetics (Figure 3c), we find a sharp change in behaviour for the two highestenergy pump wavelengths (500 and 515 nm). At these wavelengths a substantial amount of the longer-lived excited state is generated. We find no evidence, however, for an assignment to optically activated SEF, as the triplet spectrum obtained through an established sensitization method using N-methylfulleropyrrolidine as triplet donor 48 exhibits markedly different PIA (Figure 4a). Likewise, it is clear that this longer-lived species is not a charge-transfer state, as there is a poor match to the spectrum from chemical doping (Figure 4a). The match of all spectra at the short-wavelength edge of the excited-state absorption is intriguing. A similar shoulder ~700-800 nm is seen under all direct excitation conditions, as well as in the sensitized triplet spectrum and following chemical oxidation, albeit with much lower intensity relative to the GSB. While the onset of this band is largely governed by the position of the ground-state absorption, the similarity of the shoulder suggests a more fundamental relationship between the origin of the bands. We propose that the reason lies in the underlying charge-transfer character of excitations in DPP materials in general. This shoulder is evident in polaronic states (chemical reduction) and may be taken as a signature of charge-transfer character. Mixing of S 1 and T 1 states with charge-transfer configurations may then endow the PIA with such signatures, analogous to the singlet-triplet mixed features reported in bound 1 (TT) states in TIPS-tetracene solutions. 52 Comparison of the long-lived spectra following excitation < 600 nm (Figure S4 and S5, SI) with published data reveals a striking similarity to the bright singlet state in a monomer TDPP derivative in solution. 34 Indeed, we find the largest effects in the pump wavelength dependence can all be accounted for as selective excitation at high pump photon energies of a sub-population of monomer-like molecules. As discussed above in the context of the absorption spectrum, we expect that these molecules are not chemical monomers, but rather conformers in which the conjugation between TDPP units is ineffective. We find that selectivity of different degrees of torsional disorder can explain the more subtle, systematic changes in TA spectra and kinetics observed across the entire pump wavelength series. Lower pump photon energies selectively excite molecules with more extended conjugation and presumably more planalized structure, as reflected in the 0-0/0-1 vibronic ratio observed in the GSB (a direct measure of the 'type' of molecules excited). Molecules in such geometry are likely closer to the equilibrium excited-state geometry, and thus have a greater radiative rate and need less time for vibrational relaxation, resulting in a slightly shorter excited-state lifetime. The pump wavelength dependence does not explain the surprisingly short primary excited-state lifetime, which is well described by the same 3.6 ps and 10 ps time constants in all cases (discounting the effects of the monomer-like species). As one of the most likely decay channels for such a covalent dimer is non-radiative decay via torsional relaxation about the central bond, 53,54 we investigated the effects of geometric constraint in high viscosity polystyrene (PS) solution (95 mg/mL in chlorobenzene) and rigid PS matrix (Figure 4, b-e).
In PS solution we observe the same absorption line shape as in dilute chlorobenzene, while the spin-cast PS matrix exhibits pronounced broadening and a red-shift. These changes could suggest the presence of intermolecular interactions in spite of the relatively low TDPP-V-TDPP loading (50 µg/mL). However, there is no corresponding change in the excited-state dynamics. We propose instead that the broadened and strongly red-shifted spectrum reflects the planarizing effects of the PS matrix, resulting in some molecules with significantly enhanced conjugation length. In viscous solution (Figure 4c) we observe the same spectral shapes and same pump wavelength dependence identified previously: an excitation-dependent vibronic structure in the GSB, no sign of stimulated emission and a PIA towards the NIR. While the TA spectra of the polymer matrix (Figure 4d) appear very different in shape, comparison with the absorption spectrum confirms that the entire spectral shape is consistent with GSB. There no new features corresponding to longer-lived excited states or a change in decay pathway due to conformational restriction. In fact, the overall behaviour appears to be entirely consistent with measurements in 'free' solution: as previously, excitation with higher-energy photons results in an enhancement of the GSB on the 'blue' side and a reduction on the 'red' side, pointing to the same selectivity of torsional conformers.
In no case do we observe a significant change in the excited-state lifetime (Figure 4e).
The decay in PS solution and even PS matrix is almost indistinguishable from that in chlorobenzene solution, ruling out fast large-scale conformational change as the primary decay pathway. We thus consider that the photophysics observed here are most consistent with purely electronic dynamics, following the polyene model in which ultrafast internal conversion from the initial bright state forms a dark state of A g symmetry. This state in turn exhibits very strong coupling to the ground-state vibrational manifold and rapidly decays through multi-phonon emission.
Our chemical strategy for imbuing 'polyene' character in a donor-acceptor-based covalent dimer appears to be successful, but results in no singlet fission. We expect that the primary reason for this behaviour lies in the energy balance. Dimerization into TDPP-V-TDPP results in a significant stabilization of the singlet energy, by ~400 meV. It is evident that there is no comparable reduction in the triplet energy, which is estimated at 1.1 eV in the monomer; the Figure 4. A series of control measurements were performed to identify the source of the ultrafast dynamics in TDPP-V-TDPP. (a) Chemical oxidation and triplet sensitisation spectra reveal no similarity to the observed transient absorption spectra following direct excitation, nor to the difference between band-edge and excess-energy excitation. There is thus no evidence for singlet exciton fission or ultrafast charge formation. Full sensitization data presented in Fig. S9, SI. (b) UV-Vis absorption spectra of TDPP-V-TDPP in highviscosity solution and rigid polymer matrix. (c) TA spectra in concentrated PS solution are essentially indistinguishable from those obtained in chlorobenzene solutions, following both band-edge (dark red, 700 nm) and excess-energy (blue, 540 nm) excitation. (d) TA spectra in PS matrix reveal the same qualitative differences between excitation at 540 nm (blue) and 700 nm (dark red). (e) Integrated decay kinetics at the GSB peak reveal the same behaviour in viscous solution (thick) and PS matrix (thin) previously observed in chlorobenzene solutions: a slight enhancement of the lifetime following excess-energy excitation (blue), with the decay dominated by the same 3.6 ps and 10 ps time constants. net result of this would be that triplet-pair formation in energetically inaccessible. Instead, inclusion of the vinyl linker creates the possibility of a dark, lower-energy state analogous to the 2A g state analogous to polyenes, which decays rapidly and non-radiatively back to the ground state. This surprising effect reveals some of the possible pitfalls of small structural changes to design SEF dimers. However, the results also demonstrate that combining polyene character with charge-transfer chromophores can be straightforwardly achieved, and the features observed in the PIA spectra suggest that the charge-transfer character is maintained in the excited states.
The remaining challenge, then, is to modulate the exchange energy to make SEF energetically accessible without too significantly perturbing the beneficial CT mixing. Destabilization of the S 1 state should reduce the of S 1 and CT states whilst simultaneously increasing the S 1 -T 1 gap. 26,49 This can be modulated by incorporating a covalent linker to control the extent of HOMO-LUMO spatial overlap in the chromophore. Rational choice or design of covalent linkers such as phenyls, biphenyls 55 or their substituted analogues can increase the energy of the S 1 state by steric or hyperconjugation effects. End-capping the oligomer with strong electron withdrawing groups such as cyano, etc. will increase the proportion of quinoidal forms and form biradicals that stabilize the triplet state. 56