Solvent‐Dependent Singlet Fission in Diketopyrrolopyrrole Dimers: A Mediating Charge Transfer versus a Trapping Symmetry‐Breaking Charge Separation

Three diketopyrrolopyrrole (DPP) dimers, linked via different dithienylphenylene spacers, ortho‐DPP (o‐DPP), meta‐DPP (m‐DPP), and para‐DPP (p‐DPP), are synthesized, characterized, and probed in light of intramolecular singlet fission (i‐SF). Importantly, the corresponding DPP reference (DPP‐Ref) singlet and triplet excited state energies of 2.22 and 1.04 eV, respectively, suggest that i‐SF is thermodynamically feasible. The investigations focus on the impact of the relative positioning of the DPPs, and give compelling evidence that solvent polarity and/or spatial overlap govern i‐SF dynamics and efficiencies. Polar solvents make the involvement of an intermediate charge transfer (CT) state possible, followed by the population of 1(T1T1) and subsequently (T1 + T1), while spatial overlap drives the mutual interactions between the DPPs. In o‐DPP, the correct balance between polar solvents and spatial overlap leads to the highest triplet quantum yield (TQY) of 40%. Notable is the superimposition of CT and triplet excited states, preventing an accurate TQY determination. For m‐DPP, poorer spatial overlap correlates with weaker CT character and manifests in a TQY of 11%. Strong CT character acts as a trap and prevents i‐SF, as found with p‐DPP. The DPP separation is decisive, enabling a symmetry‐breaking charge‐separated state rather than CT formation, shutting down the formation 1(T1T1).


Theory
Density functional theory (DFT) and its time-dependent (TD-DFT) counterpart were used to investigate the ground (S 0 ) and excited state (S 1 , T 1 , T 2 ) properties of DPP-Ref, o-DPP, m-DPP, and p-DPP. The S 0 geometries were found using the B3LYP hybrid functional with Grimme's dispersion correction (D3) and Becke-Johnson dampening (BJ), [73][74][75][76] combined with the 6-311G(d) basis set for the DPP-Ref structure, or the split valence 6-31G(d) for the dimers. [77,78] It was found that substitution of the nitrogen atoms with at least isobutyl groups was necessary to reproduce the geometry and vertical excitations (VE) of DPP-Ref. The dimers were therefore optimized with isobutyl substituents ( Figure S29 and Table S3, Supporting Information).
Comparison of different conformations of DPP-Ref indicated the thiophene-trans configuration to be lowest in energy, as also found for the dimer conformers ( Figure S30, Supporting Information). Subsequent analysis of the o-DPP conformers revealed that stacking of the DPP monomers resulted in the lowest energy structure, whereas m-DPP and p-DPP preferred monomer separation (Figure 1). The separation of monomer planes in o-DPP was found to be 5.7 Å and in contrast, the m-DPP and p-DPP dimers are significantly separated such that spatial overlap is minimized or completely eliminated respectively.
An approximation of the SF thermodynamics of the DPP dimers was obtained by calculating TD-DFT VE (S 1 , T 1 , T 2 ) at the M06-2X/def2-TZVP level of theory (Table S5, Supporting Information). The S 0 -S 1 vertical excitations were found to be in the order of 2.61-2.27 eV, and tended toward lower energies in the order DPP-Ref > o-DPP > m-DPP > p-DPP, in agreement with the steady-state absorption-vide infra. Subsequently, the SF process was predicted to be exothermic for all dimers, with S 1 /T 1 (VE) ratios of 2.34, 2.40, and 2.38 for o-DPP, m-DPP, and p-DPP, respectively. In contrast, the TTA-UC process was found to be endothermic, with T 2 /T 1 ratios of 2.52, 2.50, and 2.44, respectively, in turn ruling out the presence of delayed fluorescence. Both S 1 /T 1 and T 2 /T 1 for all dimers were found to be of a similar value, and should therefore not impact on the SF dynamics of the dimers differently.
The charge transfer/resonance character of SF dimers has recently been highlighted to enhance the formation of the correlated triplet pair significantly. [79] Studying the spin distribution in DPP-linked pentacene dimers revealed that strong charge transfer/resonance character of the S 1 state correlated with 1 (T 1 T 1 ) exciton formation and subsequent triplet quantum yields (TQY). The DPP dimers investigated were found to have stable ground-state wavefunctions, namely, non-biradicals with satisfactory closed-shell description of the singlets. Subsequent analysis of the TD-DFT S 0 -S 1 VE in all dimers indicated S 1 to be a linear combination of the frontier molecular orbitals (FMOs) ± 1 ( Figures S37-S40, Supporting Information).
The FMOs in all dimers were delocalized over both DPPs and only subtle differences were noted. Briefly, the o-DPP highest occupied molecular orbitals (HOMOs) and lowest occupied molecular orbitals (LUMOs) were equally and unequally distributed across the subunits, respectively, with minimal delocalization onto the benzene linker. In contrast, m-DPP and p-DPP FMOs were equally distributed and included the orbitals of the benzene linker. In addition, analysis of the FMO's contributions to the TD-DFT S 0 -S 1 VE indicated minimal redistribution of electron density, even in the case of o-DPP, where the multiple contributions were almost equally distributed, minimizing the electron density localization. As a result, the S 0 -S 1 VE can primarily be described as π-π* in character, with a minimal degree of charge transfer/resonance, and should hinder the 1 (T 1 T 1 ) exciton formation, resulting in the low TQYs observed-vide infra.

Steady-State Characterization
Steady-state absorption measurements were performed to explore the ground state of the DPPs. Measurements were conducted in solvents of different polarity, that is, toluene and benzonitrile, to investigate the influence of polarity on the SFmechanism and its dynamics. The DPP-Ref absorptions in toluene included a set of well-resolved maxima between 400 and 600 nm, which represent the 0-*0 and 0-*1 transitions to the lowest singlet excited states. In addition, maxima were seen between 300 and 400 nm, representing the corresponding transitions to higher singlet excited states. Extinction coefficients as high as 30 000 m −1 cm −1 were found in toluene. Changing the solvent to benzonitrile led to no significant differences.
Analysis  o-DPP < m-DPP < p-DPP. It is safe to postulate that the electronic couplings, which differ in the DPP dimers, are sufficient to govern i-SF-vide infra. [80] Steady-state fluorescence measurements provided first indications of differences in the excited states. For DPP-Ref, the fluorescence maximum lies between 500 and 750 nm and the fluorescence quantum yield (FQY) between 69.5% in toluene and 66.7% in benzonitrile. Common to o-DPP, m-DPP, and p-DPP are hypsochromic shifts, which coalesced with the corresponding shifts seen in the absorption measurements (Figure 2b), and significant FQY quenching. Starting with m-DPP, a 46.6% FQY suggested moderate inter-DPP interactions. o-DPP is subject to a much stronger fluorescence quenching with an FQY of 12.1% and the strongest quenching is found for p-DPP with a 1.4% FQY. In contrast to DPP-Ref, changing the solvent from toluene to benzonitrile resulted in further FQY quenching: 5.3% (m-DPP), 2.1% (o-DPP), 1.2% (p-DPP) ( Figure S1 and Table S1, Supporting Information).
The steady-state absorption and fluorescence measurements indicated that for o-DPP, the intensity increase of the 0-*1 absorption is accompanied by a quenched FQY, when compared to DPP-Ref. Such trends are in line with intramolecular forces between slightly displaced, parallel arranged DPPs, as in H-type aggregates. Concentration independence underlines the intrarather than inter-molecular nature of the DPP interactions in o-DPP. The solvent-dependent FQYs suggest an excited-state deactivation involving the formation of either charge transfer (CT) state, relating to the partial transfer of electron density, or symmetry-breaking charge-separated (SBCS) state, relating to the full transfer of electron density and, in turn, the full separation of electrons and holes. Involvement of a CT/SBCS state induces red-shifts in the H-type aggregates, which usually feature blue-shifts. [81][82][83] Similarly, a deactivation via a CT state and/or a SBCS state in H-aggregates was reported for the case, in which a xanthene bridge was used in DPP dimers. [50] For p-DPP, the broadened absorption goes hand-in-hand with the most strongly quenched fluorescence among the DPP dimers. Again, the quenched and, more importantly, the solvent-dependent, albeit weak, FQYs suggest the involvement of either CT state or SBCS state. The broadening of the absorption features is attributed to the free and unhindered rotation of the DPPs, due to the minimal steric interaction of the paraposition, which was supported by theory. Finally, for m-DPP both absorption and fluorescence resemble what is seen for the reference. The overall strength of the interaction is expected to be the weakest among the DPP dimers, as the phenyl linker affords better electronic communication in case of ortho-and para-substitutions. [84,85] In low-temperature measurements with DPP-Ref at 80 K in 2-methyltetrahydrofuran (MeTHF) and in the presence of ethyliodide (EtI) to enhance the triplet formation, a 1185 nm feature was ascribed to phosphorescence. Correspondingly, a triplet excited state (T 1 ) energy of 1.04 eV was derived- Figure S2

Electrochemical and Spectroelectrochemical Characterization
Electrochemical properties of the three dimers were examined by cyclic voltammetry (CV) using CH 2 Cl 2 as solvent containing 0.1 m tetrabutylammonium hexafluorophosphate (TBAPF 6 ) as supporting electrolyte (Figure 3, Table 1). The three dimers present a similar band gap owing to the fact that the electron donor and the acceptor fragments are the same. However, p-DPP presents the lowest E g cv among the three different dimers. Based on a push-pull structure, the LUMO energy level is controlled by the electron accepting (deficient) unit, while the HOMO energy level is governed by the electron rich (donating) unit. DPP-Ref has the largest band gap with a value of 2.2 eV when compared to the DPP dimers, which are between 2.0 and 2.1 eV.
Spectroelectrochemical (SEC) measurements were conducted to register the optical features of both the one-electron reduced and oxidized forms of the DPP dimers and to use them to interpret the time-resolved transient absorption measurements. Experiments were conducted in a toluene/ acetonitrile mixture (4/1 v/v) with 0.1 m TBAPF 6 as supporting electrolyte. In the case of o-DPP, applying +1.0 V resulted in fully reversible spectroscopic features of its one-electron oxidized form in the 650-1000 nm range with maxima at 780 and 900 nm (Figure 4a). In contrast, a potential of −1.5 V resulted in the formation of its semi-stable one-electron reduced form with only a single maximum at approximately 480 nm. Unfortunately, the ground-state absorption superimposes the feature of the one-electron reduced form, which, in turn, could not be fully resolved (Figure 4b). Similar results were gathered for m-DPP and p-DPP ( Figures S4 and S5, Supporting Information).

Time-Resolved Characterization
Time-correlated single photon counting (TCSPC) measurements, with a 530 nm excitation wavelength, revealed for DPP-Ref a single fluorescent component with a lifetime of 5.5 ns in toluene. o-DPP, m-DPP, p-DPP gave rise to two fluorescent components, a short-lived one on the picosecond timescale and a long-lived one on the nanosecond timescale. [86] In detail, the lifetimes in toluene are 832 ps and 2.4 ns for o-DPP, 490 ps and 3.6 ns for m-DPP, and <200 ps and 2.5 ns for p-DPP. A change to benzonitrile only affected the short-lived component without, however, impacting the long-lived one ( Figure S6, Supporting Information). A summary is given in Table 2.
Subsequently, time-resolved transient absorption spectroscopy on the femto-(fsTAS) and nanosecond (nsTAS) timescales were performed following 530 nm excitation. Lifetimes were obtained by means of GloTarAn target analyses. In DPP-Ref, excitation is immediately followed by the formation of the singlet excited state (S 1 ). In toluene, maxima are found at 435, 710, and 885 nm, while minima evolved at 513, 562, and 616 nm ( Figure S8a,c, Supporting Information). The former minima relate to ground-state bleaching (GSB) and the latter to stimulated emission (SE). All of these features were replaced within 4.5 ps ( Figure S8b,d, Supporting Information) by an intermediate state, which was characterized by subtle shifts of the 885 nm maximum and the 562 and 616 nm minima. The underlying process corresponds to solvent reorganization, which was due to increasing the solvation energy of the excited state leading to (S 1 ) Sol . [29,87] From (S 1 ) Sol , the complete reinstatement of the ground state (S 0 ) is reached within 6.0 ns ( Figure S10, Supporting Information). Similar results were gathered for DPP-Ref in benzonitrile ( Figures S9 and S11, Supporting Information).
An initial analysis of o-DPP, m-DPP, and p-DPP showed that all of them featured in toluene the same initial singlet excited state (S 1 S 0 ) with lifetimes of 9.9, 7.8, and 15.5 ps, respectively. Common to all (S 1 S 0 ) was the solvent reorganization to afford    Interesting is the fact that the third species bears similarities with the aforementioned (S 1 S 0 ) Sol signatures. In other words, it has still some singlet excited state character. Next to the (S 1 S 0 ) Sol signatures, transients at 460, 650, and 1000 nm resemble the fingerprints seen in SEC (Figure 4 and Figures S4 and S5, Supporting Information). We postulate in line with recent work mixing of the singlet excited state with a CT state and defining the third species as (S 1 S 0 ) CT . [25,29,88] Noteworthy, the lifetimes of (S 1 S 0 ) Sol and (S 1 S 0 ) CT match those of the short-and long-lived fluorescent components detected in the TCSPC measurements -vide supra ( Table 2). The only notable differences between o-DPP, m-DPP, and p-DPP are the expected shifts in the GSB and SE features. Importantly, no appreciable population of any triplet excited state is observed in toluene (Figures S12-S15, S18,S19, Supporting Information). The strong impact of solvent polarity changes on the FQYs suggested a subtle interplay between the different states in the excited state deactivation of o-DPP, m-DPP, and p-DPP. A comparison with toluene corroborates that the first, second, and third species are still involved in the excited-state deactivation. The longer (S 1 S 0 ) lifetimes in benzonitrile are due to the higher viscosity, which slows down solvent reorganization.
Analysis of the results for o-DPP showed that, after solvent reorganization (S 1 S 0 ) Sol , a stronger CT contribution in (S 1 S 0 ) CT is derived from the positive transient at 650 nm (Figure 5a,c). Likewise, a shorter (S 1 S 0 ) CT lifetime relative to toluene stems from the CT-stabilization in more polar solvents. (S 1 S 0 ) CT also experienced a stronger CT-mixing, which is manifested in an intense peak at 480 nm and depletion of GSB. Both characteristics are in line with the SEC features seen for the one-electron reduced form of o-DPP (Figure 4b). Lifetimes of (S 1 S 0 ) Sol and (S 1 S 0 ) CT (Figure 5b,d and Table 2) match the corresponding TCSPC lifetimes in benzonitrile (Table 2)-vide supra. Contrary to the results in toluene, the (S 1 S 0 ) CT deactivation is linked to the population of two additional states. Of great importance is the fact that none of these two additional states resemble the CT character seen, for example, in (S 1 S 0 ) CT , or the (S 1 S 0 ) Sol features. In particular, maxima are found at 470 (broad) and 555 nm, which are complemented by minimum at 585 nm. Of equal importance is the fact that they are spectrally nearly identical.
To clarify the nature of these two additional states, we turned to triplet-triplet sensitization measurements with N-methylfulleropyrrolidine (N-MFP, Figure S23, Supporting Information) as a triplet-triplet energy transfer (TTET) donor and o-DPP as TTET acceptor. Photoexcitation was performed at 430 nm. In the case of o-DPP, again maxima at 470 (broad) and 555 nm and a minimum at 585 nm were seen to grow in at the expense of the triplet excited state maximum of N-MFP at 700 nm. The correspondingly formed (T 1 S 0 ) decays with approximately 20 µs to the ground state. Considering the appreciable resemblance between the (T 1 S 0 ) signature in TTET and the spectroscopic signature of the first of the two additional states, we infer the population of 1 (T 1 T 1 ) in o-DPP, that is, two triplet excited states are formed (Figure 6a,c). The reintensification of the GSB was emphasized in Figure S22, Supporting Information. In contrast to the approximately 20 µs lifetime of (T 1 S 0 ), 1 (T 1 T 1 ) is short lived. It is within 42.0 ns that 1 (T 1 T 1 ) transforms in the second of the two additional states. [89] We rationalize the transformation to the decoherence of 1 (T 1 T 1 ) (Figure 6b,d and Table 2) and the formation of the uncorrelated triplet excited state (T 1 + T 1 ) (Figure 6a,c). The lifetime of the latter is 15.0 µs and agrees well with that derived for (T 1 S 0 ) (Figure 6b,d and Table 2). In short, the existence of these two triplet-based species, inferred by a reasonable coupling of the thiophene-phenyl-thiophene bridges due to rotational freedom/heteroatom effect, corroborated i-SF, as conventional ISC would lead to only one triplet excited state. Despite the lack of CT character in 1 (T 1 T 1 ) and (T 1 + T 1 ), an alternative deactivation route should be mentioned: two subsequently formed CTs with singlet and triplet spin, respectively.
The TQY was calculated using the method, which incorporates the singlet (ε S1* ) and triplet (ε T1* ) excited state molar extinction coefficients. As such, a value of 40% was derived ( Figure S28a and Table S2, Supporting Information). [28,29,31] Note that the spectroscopic features stemming from the oneelectron reduced form superimpose those of 1 (T 1 T 1 ) and (T 1 + T 1 ) making an accurate calculation of the TQYs difficult.
m-DPP acts like o-DPP with a biphasic decay of (S 1 S 0 ) CT via 1 (T 1 T 1 ) and (T 1 + T 1 ), but with a shorter (S 1 S 0 ) Sol lifetime of www.advenergymat.de www.advancedsciencenews.com 298 ps (Figures S16a,c and S17a,c, Supporting Information and Table 2). Overall, the CT character in m-DPP is weaker than in o-DPP and the TQY is 11%. A likely rationale implies that the inter-DPP electronic coupling is weaker in m-DPP across the meta-position of the phenyl linker than across the ortho-position in o-DPP and, in turn, reduces TQY. [84,85] Finally, for p-DPP the shortest (S 1 S 0 ) Sol lifetime and subsequent population of (S 1 S 0 ) CT was observed. Despite this finding, no evidence for any significant triplet excited-state population was gathered. At first glance, (S 1 S 0 ) CT revealed an even stronger CT character for p-DPP than for o-DPP. A closer look reveals, however, that the feature of the oneelectron reduced form covers the range from 400 to 650 nm and, in turn, masks the GSB at around 600 nm ( Figures  S20a,c and S21a,c, Supporting Information and Table 2). From the strong resemblance of the transient with the SEC fingerprint and the low FQYs in the steady-state assays, a SBCS rather than (S 1 S 0 ) CT formation with a significant CT character is inferred. [50] Minimizing the spatial overlap in o-DPP, m-DPP, and p-DPP enables SBCS. SBCS is, however, detrimental to SF. Therefore, the i-SF efficiency follows the trend of o-DPP > m-DPP >>> p-DPP and a schematic diagram of the kinetic models is shown in Figure 7.

Conclusion
A novel set of three different dimers of DPPs, linked via dithienylphenylene spacers (o-DPP, m-DPP, and p-DPP) was synthesized, characterized, and examined in light of intramolecular singlet fission (i-SF). Steady-state absorption and fluorescence measurements indicated the involvement of a charge transfer (CT) and/or SBCS state as part of the SF cascade. Solvent polarity plays the most important role in either CT or SBCS state population. Time-resolved transient absorption spectroscopy revealed two limiting scenarios. On the one hand, CT mediated i-SF. Here, non-polar solvents such as toluene were found to be detrimental to the population of any triplet excited states. It is only polar solvents such as benzonitrile, which enabled the population of 1 (T 1 T 1 ) and subsequent (T 1 + T 1 ). On the other hand, the spatial arrangement of the DPPs was important to govern i-SF. o-DPP had the highest i-SF efficiency with a TQY of 40% in benzonitrile. The TQYs were as low as 11% for m-DPP and 0% for p-DPP when benzonitrile was used as a solvent. As such, with the electronic coupling between the two DPPs, in the case of o-DPP, it is sufficiently strong to promote i-SF and, at the same time, weak enough to ensure 1 (T 1 T 1 ) decoherence and avoid quantitative triplet-triplet annihilation. An alternative mechanism, which is based on the unlikely transient population of two subsequent CT states, namely of singlet and triplet spin 1 CT and 3 CT, respectively, before just one T 1 is formed, cannot entirely be ruled out. The i-SF was quantitatively suppressed in p-DPP, where the lack of spatial overlap resulted in a SBCS and shut down the population of 1 (T 1 T 1 ) and (T 1 + T 1 ). Therefore, this study has highlighted that the strength of the CT character, controlled by the solvent polarity, is decisive in balancing the electronic coupling and determining whether it mediates i-SF in DPP dimers or blocks i-SF due to trapping.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.