Chain Conformation and Exciton Delocalization in a Push–Pull Conjugated Polymer

Linear and nonlinear optical line shapes reveal details of excitonic structure in polymer semiconductors. We implement absorption, photoluminescence, and transient absorption spectroscopies in DPP-DTT, an electron push–pull copolymer, to explore the relationship between their spectral line shapes and chain conformation, deduced from resonance Raman spectroscopy and from ab initio calculations. The viscosity of precursor polymer solutions before film casting displays a transition that suggests gel formation above a critical concentration. Upon crossing this viscosity deflection concentration, the line shape analysis of the absorption spectra within a photophysical aggregate model reveals a gradual increase in interchain excitonic coupling. We also observe a red-shifted and line-narrowed steady-state photoluminescence spectrum along with increasing resonance Raman intensity in the stretching and torsional modes of the dithienothiophene unit, which suggests a longer exciton coherence length along the polymer-chain backbone. Furthermore, we observe a change of line shape in the photoinduced absorption component of the transient absorption spectrum. The derivative-like line shape may originate from two possibilities: a new excited-state absorption or Stark effect, both of which are consistent with the emergence of a high-energy shoulder as seen in both photoluminescence and absorption spectra. Therefore, we conclude that the exciton is more dispersed along the polymer chain backbone with increasing concentrations, leading to the hypothesis that polymer chain order is enhanced when the push–pull polymers are processed at higher concentrations. Thus, tuning the microscopic chain conformation by concentration would be another factor of interest when considering the polymer assembly pathways for pursuing large-area and high-performance organic optoelectronic devices.


INTRODUCTION
The most fundamental aspect of the materials science of semiconductor polymers concerns the relationship between solid-state microstructure, macromolecular conformation, and the electronic and optical properties of these materials.][5][6][7][8][9][10] These models have been extended to push-pull polymers, [11][12][13] in which the role of charge-transfer interactions are evident.In this article, we examine absorption, photoluminescence (PL), and transient absorption optical lineshapes in films of poly [2,5-(2-octyldodecyl)-3,6-diketopyrrolopyrrole-alt-5,5-(2,5-di(thien-2-yl)thieno-[3,2-b]-thiophene)] (DPP-DTT), a push-pull polymer designed for applications in thin-film transistors. 14,15The films are cast from solutions of various concentrations; we find that the optical lineshapes of the films show a strong dependence on the precursor concentration, and correlate the spectral shapes with chain conformation derived from resonance Raman measurements and the corresponding ab initio calcuations.We conclude that when solutions are cast from gel-like solutions, excitons are more highly delocalized along the polymer backbone, driven by more planar, less torsionally disordered backbones.
A subtle control of the polymer aggregation could tune the polymer chain conformations, which determines the arrangement of the chromophores.It is crucial to build the correlations between the microscopic conformations and photophysical properties.Although previous studies have shown that photophysical and electronic properties of conjugated homopolymers, like poly-(3-hexyl)thiophene (P3HT) or polyphenylenevinylenes, are greatly influenced by the long-range order, 7,[16][17][18] where the exciton dissociation and charge migrations are impacted by the polymeric aggregate fractions, microstructural conformations, defects and etc., [19][20][21] the photophysical aggregates and nonaggregates of short-range order could also play a role in the photogenerated charges. 22,23Furthermore, in P3HT-like derivatives, the exciton coherence lengths, which indicates the spatial span of the exciton wave, are varied with different molecular weights. 16Specifically, the intrachain exciton extending more units along the chain backbone in the high molecular-weight P3HTs give rise to longer PL lifetime, compared to their low molecular-weight counterparts. 175][26][27] Previous work showed that the electron push-pull nature enhances exciton-exciton annihilation, which gives rise to long-lived bound charge pairs. 28Besides, intra-and interchain charge-transfer interactions, representing the wavefunction overlaps between the electron-sufficient and -deficient moiety along and across the polymer chains, could be subject to the polymer chain backbone orders and π-π stacking, respectively. 12,24,29,30Furthermore, the charge-transfer character in the electron push-pull polymers renders a permanent dipole and induces a strong overlap of the electron cloud between push and pull chromophores, which breaks the Kasha approximation of only transition dipole moments interacting. 31Therefore, a different energetic landscape from that of homopolymers polymers might be expected in electron push-pull materials.
3][34][35][36][37] In either way, the interchain excitonic interactions are considered to be reduced.However, most studies introduced new components which might lead to ambiguity regarding polymer chain order, microstructures, dielectric environment and unwanted bath-system interactions.Induced gel formation, on the other hand, only exploited either molecular weightor concentration-dependence, which are intrinsic properties of polymers in their solution and solid state. 38,39Here, we contribute a detailed understanding of the impact of photophysical aggregation on excition delocalization as well as chain planarization through gel formation process in push-pull polymers.

RESULTS
Using the electron push-pull polymer, DPP-DTT, as the material of interest displayed in Fig. 1a, thin films were prepared from solutions below and above the viscosity deflecting concentration, c * as shown in Fig. 2a.The film preparation method was described previously. 40To account for the perturbation of the Coulombic interactions within the excitation band, the absorption spectra were fit to a Franck-Condon (FC) progression modified by the contributions of exciton bandwidth, where the effective Huang-Rhys parameter was set to be 0.73 (Fig. 1b). 13With an increase in concentration, the ratio of A 0−0 and A 0−1 absorption peaks decreases, which corresponds to an increase in interchain exciton bandwidth from 16 to 54 meV as shown in Fig. 2b, accompanied with a small blue shift around 15 meV of the A 0−0 peak (see Table S1 and Fig. 1b).The magnitude of the exciton bandwidth falls well under the weakly-coupled HJ-aggregate limit. 7A direct consequence of larger excitonic interaction in the H-aggregate is an increasing Stokes shift when examining the steady-state photoluminescence (PL) and absorption measurements simultaneously (Fig. 1b).Interestingly, in addition to the redshift of the PL peaks, a trend of linewidth narrowing is also observed as shown in Fig. 2d, where the major peak can be simply fitted with a standard Gaussian distribution.Although enhanced interchain excitonic interaction could lead to the redshifting behavior in the emission of H-aggregate, it will not result in drastic line narrowing in the PL line shapes.Based on the model developed by Knapp, 41 Knoester 42 and Spano, 6,10 the line narrowing effects seen in aggregate PL spectra can be explained by the motional narrowing effect, where the distribution of static disorder is averaged out due to the fast-moving excitons.Specifically, the narrowing effect could be attributed to a few physical factors; an increase in the magnitude of static disorder, (inhomogeneous broadening linewidth, σ d ), growing aggregate size, (number of chromophores in each aggregate, N ), and/or shorter exciton coherence length across chains, in the weakly-coupled H aggregate. 6,10,16 We interpret these observations as a weakly varying width of total disorder with processing concentrations.
The relevant length scales for the photophysical aggregate are highly microscopic (of order nearest molecular neighbor), while X-ray crystallography samples length scales are relevant to crystalline order.Nevertheless, we can compare this information with that derived from grazing-incidence wide-angle X-ray scattering (GIWAXS) measurements.Recent measurements on this material 39 demonstrated invariant d-spacing values for π − π stacking of 3.7 Å and lamellar spacing of 19.6 ± 0.2 Å, consistent with other DPP-based copolymers as shown in previous literature. 30,38,43Furthermore, consistent full width at half maximum (FWHM) linewidths for each (010) scattering peak when comparing all samples, indicating similar paracrystalline static disorder among all samples. 44To compare the aggregate sizes of the thin films for different concentrations, differential scanning calorimetry (DSC) was performed as shown in Fig. S3.A melting temperature of 376 ± 1 • C is found among all samples, and a consistent curve shape for the melting peak is observed.Combining both GIWAXS and DSC measurements, we deduce that the aggregate size and the lattice disorder are probably not the dominant factors for the drastic red shift as observed in the steady-state PL spectra.Since the intra-and interchain exciton delocalization are countering each other, 16,45 the motional narrowing effect could be ascribed to the variation of the exciton dispersion due to the change of polymer chain order in the HJ-type photophysical aggregates, as will be demonstrated later on.
Another distinctive feature from the PL spectra is the emerging side peak on the high energy shoulder when the concentration crosses c * .As only separated from the dominant Gaussian peak by 100 meV, we exclude the possibility of it being a vibronic satellite of the typically dominant 160-180 meV modes.A similar trend is also observed within the absorption spectra (Supplemental Fig. S1); the deviation from the aggregate spectra on the high-energy side (from 1.43 to 1.68 eV) is partially ascribed to the absorption of the non-aggregates. 7,43To more clearly demonstrate the component ratio of the non-aggregates relative to the aggregates, the relative optical contribution from both species is calculated and shown in Fig. 2c.A constant ratio is observed before c * while the relative contribution from the non-aggregates starts to increase drastically after the critical concentration.Therefore, the new PL side peaks can be correspondingly assigned to the emission of such photophysical non-aggregates.
The contributions of the non-aggregate were probed via transient absorption (TA) spectroscopy across the viscosity deflecting concentration.In Fig. 3 we display TA spectra for 3, 4, 6 and 8 g/L films, with the pump pulse centered at 760 nm, and with fluence under 5 µJ/cm 2 .(Measurement for films prepared from 5 g/L solutions are presented in Fig. S5 in the Supplementary Information.)The four TA plots show two domains where the red and blue represent the ground state bleaching (GSB) and photoinduced absorption (PIA), respectively (Fig. 4).All subplots show a dominant GSB signal and the high absorption coefficient is commonly observed due to the long persistence length of the conjugated copolymers. 46e lifetimes of GSB are around 18 ps (see SI Fig. S6).The PIA signals, on the other hand, show a comparable lifetime to that of the GSB, which suggests that the high-lying excited states relax to the lowest excited vibronic state on a comparable timescale.
To avoid arbitrary spectral drifting at early times, the spectra were averaged by taking the temporal cuts from 0.15 to 1 ps as displayed in Fig. 5. Within the polymer films deposited from higher concentrations, a new excited-state feature emerges in the PIA region.The zero-cross points, defined as photon energies where the differential transmission signal is zero, indicate counteracting contributions between positive signal (GSB) and negative signal (PIA).The zero cross points (black circles) with respect to the concentration are plotted in Fig. 2e.The zero-cross points are observed to have a slight decrease up to c * followed by a drastic increase.The measurement of 5-g/L film shows a clear small peak around 1.33 eV (932 nm) due to stimulated emission (SE), which also gives rise to the red shift of the zero-cross points in addition to GSB.As noted, the zero-cross points for samples of 4 and 6 g/L might also have contributions from SE, even though they are much weaker than 5-g/L sample.The relative integral ratio of the PIA feature and the GSB of the excitons shows a similar trend by comparing the integral of the absorption in these two regions (shown as the shaded area in Fig. 5a).A direct absorption from the high-energy excited states in the non-aggregates could contribute to this enhanced PIA signature, which is consistent with a growing content of non-aggregates in samples prepared from high concentrations as mentioned earlier.It is also worth pointing out that the new PIA features resemble the derivative-like Stark effect features, 47,48 as the differential spectra are perturbed by the induced electric field from the accumulating photogenerated charges at the interfaces of aggregates and non-aggregates. 22The degenerate states are lifted by the induced electric field to give such new feature, especially when comparing the differential line shapes with the linear transmission spectra.Below c * , the GSB transition very much follows the linear transmission, while above c * , the differential spectra have a sharper transition.The two above mentioned factors could both contribute to the new PIA features.To further understand the polymer chain order and local conformations, we performed resonance Raman spectroscopy to study the on-chain vibrational modes, which are coupled to the electronic transitions.As a more torsionally ordered polymer chain backbone could sustain a longer exciton coherence length, 16 a more delocalized electron density would weaken the resonance Raman intensity. 17,49Here, the high-energy band in DPP-DTT at 488 nm was excited, which contributes to a delocalized π−π * transition as demonstrated by Wood et al. 49 The Raman spectra at high and low frequency range are shown in Fig. 5c and d, respectively.
The most significant change is observed at 1410 cm −1 within the high-frequency range from 1200 to 1600 cm −1 displayed in Fig. 5c, whereas the more substantial variations occur in the low-frequency range as shown in Fig. 5d.
The Raman-active modes are assigned based on density functional theory (DFT) calculations.These calculations were performed on a DPP-DTT trimer, truncated with a fourth DPP-thiophene (T) unit.The additional DPP-T unit is utilized to allow for a symmetric charge distribution.The calculated Raman spectra are displayed in Fig. 5a and b.As demonstrated by Chaudhariet al., two different geometries might coexist in the ensemble, both contributing to the Raman cross section.In geometry 1 (G1), the oxygen atoms of the DPP unit are oriented close to the sulfur atoms of the neighboring T units, while in geometry 2 (G2), they are instead oriented near the hydrogen atoms of the neighboring T.
The optimized structures of G1 and G2 are shown in Fig. S7.Based on DFT calculations, the ground-state energy of G1 is lower than that of G2 by approximately 41.2 kJ mol −1 .The calculated Raman spectra of both geometries are shown in Fig. 5; both spectra are normalized to the peak at 1525 cm −1 in G1.In the following discussion, all Raman shifts refer to experimental results unless specified otherwise.A complete assignment of the Raman modes is presented in Table 1.
To allow us to compare the change of Raman intensities quantitatively among different samples, a vibrational mode that is not significantly influenced by the torsional order of the polymer backbone (i.e., a local or intraunit vibrational mode) is used as a benchmark.Herein, such mode is chosen to be the local C=C stretching in the highly rigid DPP unit at 1525 cm −1 , coupled with a local, asymmetric DTT ring deformation, as shown in the vector diagram in Fig. 6a.Such localized C=C stretching mode on the DPP unit is also observed in a related DPP-based copolymer. 49Another localized stretching mode at 1366 cm −1 associated with the two bridgehead carbon atoms on the DPP units, which shows constant intensity among all samples, supporting our rationalization.In contrast, the greatest changes in intensity are observed at 1416 cm −1 , where the DTT ring demonstrates strong deformation as shown in Fig. 6b.Interestingly, the corresponding simulated Raman peaks are calculated to be doubly degenerate, with one mode being the ring deformation of the second and third DTT unit and the other mode being that of the first and fourth DTT unit.Such 'globally' delocalized DTT ring deformation directly contributes to a more delocalized exciton wave along the polymer chain backbone, leading to a weaker vibronic coupling strength and decrease of the Raman intensity.A more quantitative demonstration is displayed in Fig. 2f, where the intensity ratio of Raman peaks at 1410 and 1366 cm −1 is shown to decrease, with clear two-regime behavior.
The other two globally delocalized vibrational modes, at 706 and 467 cm −1 , show a similar decreasing trend in intensity, where the former mode is a gentle ring breathing motion among all DPP and DTT units and the latter is a degenerate ring deformation mode in all DTT units (Fig. 6f).Aside from the stretching modes, the dynamically disordered thienothiophene unit also experiences a strong torsional motion as observed at 494 cm −1 (Fig. 6e).
Besides the two different trends and their according vibrational modes, an intriguing increasing trend is observed at 810 cm −1 , colored in purple in Fig. 5c.This mode corresponds to the global symmetric C-N stretch in the DPP units (Fig. 6d), which localizes the exciton wave.Therefore, when the polymer chain backbone becomes more planarized, which sustains a more delocalized exciton, the vibrational mode at 810 cm −1 is not only able to collect the electron density, but also redistribute that density in an almost perpendicular direction relative to the polymer chain.The final result is an increase in intensity for higher concentration samples.
Such observation of the Raman intensity variations aligns well with the observation of the PL motional narrowing effect, as mentioned above, indicating the polymer chain planarization in samples prepared higher than the gel formation concentration.Besides, the detailed analysis of the absorption and PL spectra show the increased exciton bandwidth, indicating stronger interchain interaction.Another surprising finding of higher optical contributions from the non-aggregates with increasing concentrations surfaces when comparing the lineshape and oscillator strength from the linear absorption and nonlinear TA measurements.The strong correlations between the viscosity and spectroscopic results can now be well established (Fig. 2).We note that the abosption spectra were previously published in Ref. 40.The PL peaks are fitted with a single Gaussian distribution (red dash line), while the absorption vibronic replica are simulated with modified Franck-Condon progression (solid black line).The spectra shift for PL and absorption are denoted with the dashed black lines with increasing concentrations.A pronounced red shift can be readily observed in the PL spectra, meanwhile the absorption spectra also display a small blue shift of around 15 meV (see Table S1 in SI for quantitative results.)

DISCUSSION
Electron push-pull polymers, unlike conjugated homopolymers, exhibit strong charge-transfer character between the electron sufficient and deficient unit within chains.1][52][53] As further demonstrated by Chang et al., the dominant interchain charge transfer interaction in DPP-based systems could lead to abnormal red-shifted H-type aggregate behavior, when the electron and hole transfer integral is out-of-phase, which has been shown experimentally. 436][57] Of particular relevance, Chaudhari et al.
showed that DPP-DTT polymer aggregates adopt a geometry where the donor and acceptor units between chains are alternating stacked when the films are prepared by spin coating.However, they also point out that segregated donor-on-donor or acceptor-on-acceptor stacking arrangement might have slightly lower energy compared to the slip-on geometry.Such stacking order could be achieved when certain processing conditions are met.Under the influence of solution concentration, the resonance Raman spectra displayed the most changes in the low-frequency range, where most of the contributions originate from the stretching and torsional modes of the DTT unit.Such conformational disorder stems from the rotational invariance of the mesogenic groups (i.e.thienothiophene component), and the favorable lowest-energy backbone conformation could be adopted easily. 58Indeed, previous DFT calculations performed on DPP-DTT have shown that the different conformations of the DTT unit (e.g.different twisting angles between the thienothiophene and monomeric thiophene rings) have more minor energy difference, compared to that of the DPP unit, implying that the disorder source possibly originates from the former. 30In contrast, the DPP units are highly planarized and rigid, verified by the localized vibrational modes seen in Fig. 5d in the high-frequency region.Note that whether the Raman intensities increase, decrease, or remain constant depends on the nature of each vibrational mode; some of the thienothiophene ring deformations along the chain backbone will aid in dispersing the exciton wave, while specific DPP ring deformation modes will lead to exciton wave localization, perpendicular to the polymer chain backbone.This direction dependency reflects the tensor aspects of the Raman polarizability.The large repeat unit will have not only diagonal elements but also off-diagonal elements in the polarizability expression.
Current studies do not allow us to quantitatively determine the spatial correlation of site energies, β, which can be quantified by the I 0−0 /I 0−1 PL ratio, 16 specifically, due to limitations in the PL detection range.However, as mentioned earlier, the 0-0 peaks become more red-shifted and narrowed with increasing concentration.Furthermore, from the resonance Raman spectra, we are able to conclude that a more dispersed exciton wave is achieved along the polymer chain backbone with increasing concentration.Therefore, it is reasonable to deduce that the polymer chain backbone is more planarized when the processing concentrations increase.Combining both observations, we formulate the hypothesis that the more dispersive exciton wave along the polymer chain leads to a greater extent of spatial correlations of energies, which was described in the J-aggregates as shown by Knapp 41 and Knoester. 42It is worth noting that the spatial correlation function of site energies should be two-dimensional, both along the polymer chain backbone and across the chains in the π-stack, which was demonstrated in P3HT by Spano et al. 10,16,59 Specifically, in the high M w P3HT, the extent of exciton coherences along (across) the chains are higher (lower) than that of low M w P3HTs. 16To accurately account for the exciton coherences of DPP-DTT, rigorous calculations of the two-dimensional correlation functions and a larger range of PL measurements are needed. 16

CONCLUSIONS
We demonstrate that the exciton in push-pull polymers is more delocalized in films cast from solutions in which the concentration surpasses the viscosity threshold for gelation.We hypothesize that this phenomenon is attributed to enhanced chain backbone planarization, as indicated by resonance Raman spectroscopy and DFT calculations.Analysis of the absorption and steady-state PL spectra is consistent with a more highly delocalized exciton along the chain backbone, and more chromophores uncoupled to photophysical aggregates are also formed above the gel formation concentration.The contributions to the transient absorption spectra at the timescale of a few picoseconds are likely two-fold: derivative-like spectral line shapes due to accumulating photogenerated charges at the aggregate/non-aggregate interfaces 17 and direct excited-state absorption from the non-aggregate chromophores.We demonstrate the importance of understanding the short-range polymer chain order and excitonic interactions.Manipulating such short-range length scales could further our understanding in preaggregate assembly and favorably accelerate the development of next-generation organic optoelectronics.

EXPERIMENTAL METHODS
Sample Preparation.DPP-DTT (M w = 290,000 g mol −1 , dispersity = 2) was purchased from Ossila Limited.For sample preparation, a stock solution of 10 g/L DPP-DTT in chlorobenzene (anhydrous, Sigma-Aldrich) was prepared by heating at 100 • C for around 4 hours, followed by heating at 60 • C overnight.Then solutions with lower concentrations are diluted from the stock solutions.The DPP-DTT thin films are prepared by wire-bar coating the solutions of different concentrations on fused-silica substrates at 56 • C, followed by annealing for 10 min.
Vis-NIR Absorption Spectroscopy.The Vis-NIR absorption measurements were performed using the Cary 5000 UV-Vis-NIR spectroscopy.
Photoluminescence Spectroscopy The steady-state photoluminescence spectroscopy is performed using the inVia Renishaw Spectrometer in the back-scattering configuration.The samples are illuminated by a 785 nm red laser.
Ultrafast Transient Absorption Spectroscopy.The transient absorption measurements are performed using an ultrafast laser system (Pharos Model PH1-20-02-10, Light Conversion).Tunable wavelengths are generated using a laser fundamental of 1030 nm at 100 kHz repetition rate.The integrated transient absorption was measured in a commercial setup (Light Conversion Hera).The wavelengths of the pump can be tuned from 360 to 2600 nm by feeding 10W laser output to a commercial optical parametric amplifier (Orpheus, Light Conversion, Lithuania) while probe beam is generated by sending 2 W to a sapphire crystal to obtain a single-filament white-light continuum in the spectral range of 490-1060 nm.The probe beam was collected by an imaging spectrograph (Shamrock 193i, Andor Technology Ltd., U.K.) coupled with a multichannel detector (256 pixels, 200-1100 nm wavelength range) after transmitting through the sample.All the samples were measured in a homemade vacuum chamber.
Resonance Raman Spectroscopy.The resonance Raman spectra are measured using the inVia Renishaw Raman spectrometer, where the samples are excited with 488 nm laser with a back-scattering configuration.
Quantum Chemistry Calculations.Density functional theory (DFT) and timedependent DFT (TDDFT) calculations were performed at the LC-ωHPBE/6-311G(d) level of theory using the Gaussian 16 Rev.A.03 software suite. 60Empirical gap tuning was performed for the two oligomer geometries following the method of Sun et al., [61][62][63][64][65] obtaining converged range-separation parameters of ω 1 = 0.1295 (for G1) and ω 2 = 0.1216 (G2).Following optimization, vibrational frequency analysis was performed on the oligomer, with vibrational scaling factors of 0.995 (G1) and 0.968 (G2) applied to the calculated Raman frequencies.The Raman activities were converted to intensities consistent with prior literature; 66,67 further details are available in the Supplementary Information.A TDDFT calculation was performed to obtain the excitation wavelength, 439 nm (22780 cm −1 ), corresponding to the wavelength used in the associated experimental spectroscopy.From left to right are the widths of Gaussian inhomogeneous distribution used to describe the energetic disorder, σ; the energy of the 0-0 vibronic transition, E 0 ; exciton bandwidth, W; the energy of the vibrational mode coupled to the electronic transition, E p and the fit constant.

Differential Scanning Calorimetry
The differential scanning calorimetry is conducted with TA Instruments DSC250.The thermograph is shown in Figure S3.With a cooling and heating rate of 20

Transient absorption
The transient absorption spectra measured under the lowest (Fig.

Figure 4 :Figure 5 :
Figure 4: Normalized transient absorption spectra integrated from 0.15 to 1.0 ps measured for samples of 3, 4, 5, 6, and 8 g/L.The cutoff from 1.57 eV to 1.71 eV is due to the leak of pump beams.The shaded area is integrated to estimate the ratio between GSB and PIA.The differential transmission spectra are overlapped with the linear transmission spectra (dashed line) converted from Figure 1.

Figure 1 :
Figure 1: (a) Molecular structure of DPP-DTT.(b) Normalized absorption spectra (open circle), and steady-state PL (solid circle) of thin films deposited from different solution concentrations.We note that the abosption spectra were previously published in Ref.40.The PL peaks are fitted with a single Gaussian distribution (red dash line), while the absorption vibronic replica are simulated with modified Franck-Condon progression (solid black line).The spectra shift for PL and absorption are denoted with the dashed black lines with increasing concentrations.A pronounced red shift can be readily observed in the PL spectra, meanwhile the absorption spectra also display a small blue shift of around 15 meV (see TableS1in SI for quantitative results.)

Figure 2 :
Figure 2: (a) Viscosity measurement on DPP-DTT-chlorobenzene solutions performed at 56 • C, reproduced from Ref 40.The two-regime behavior is visualized by the two linear lines with different slopes (b) The A 0−0 /A 0−1 ratio (black open squares) and the effective interchain exciton bandwidth (red solid squares) acquired from FC simulations as a function of concentrations.The error bars associated with exciton bandwidths are also indicated.(c) The ratio of the optical absorption area of the aggregate and nonaggregate (open hexagon).(d) The peak positions (black open triangles) and widths (purple solid triangles) acquired from Gaussian distributions.(e) The integral ratio (blue open pentagons) of the shaded area in PIA and GSB, as shown in Fig. 4. The zero cross points (ZCP) are denoted in blue solid pentagons.(f) The ratio of resonance Raman peak at 1411 and 1366 cm −1 , with all spectra simultaneously normalized at peak 1525 cm −1 .The dashed line is the guide for eye of the critical concentration point.

Figure 3 :
Figure 3: Transient absorption spectral maps showing the differential transmission signals of (a) 3, (b) 4, (c) 6, and (d) 8 g/L samples pumped with 760-nm pulsed beam under 5-µJ/cm 2 fluence.The transmission PIA and GSB signals are colored in blue and red, respectively.The dash line lies in the zero-amplitude cross points for each sample, where a clear shift is observed.

(i) a
The whole spectrum is normalized at 1525 cm −1 ; b Essentially degenerate modes; (ii) L: local, D: delocalized; 1, 2, 3 and 4 label the order for DPP or DTT unit displayed in Figure6;(iii)C t , C s , C p are ternary, secondary and primary carbon atom, respectively; b-C: bridgehead carbon of polycyclic rings; m-Th: monomeric thiophene ring.

Figure 6 :
Figure 6: Vector diagrams for the Raman modes of a symmetrically truncated DPP-DTT trimer in geometry 1. (Raman modes for geometry 2 are shown in Figure S8.)Alkyl side chains are substituted with methyl groups for computational efficiency.Atomic color scheme: gray = carbon, white = hydrogen, yellow = sulfur, blue = nitrogen, red = oxygen.The blue and red numbers at the top indicate the indices of the DPP and DTT units, respectively.Atomic displacements are indicated by the arrows.

Figure S2 :
Figure S2: The Gaussian fit (blue dashed line) for PL spectra of thin film samples prepared from each concentration.The high energy shoulder is indicated by the red line.

Figure S3 :
Figure S3: First heating curves of drop-cast films prepared from concentrations of 2, 4, 6 and 8 g/L with endo-down heating flow.
) and highest pump fluences (Figure S5) that are feasible for clear signals, respectively.The fluences are displayed in each spectrum.

Figure S4 :
Figure S4: Normalized transient absorption time-energy map collected at the lowest pump fluences for 3, 4, 5, 6 and 8 g/L.The fluences are indicated in the annotation.

Figure S5 :Figure S6 :
Figure S5: Normalized transient absorption time-energy map collected at the highest pump fluences for 3, 4, 5, 6 and 8 g/L.The fluences are indicated in the annotation.

Figure S7 :
FigureS7: DFT-optimized molecular structures of (a) geometry 1 (G1) and (b) geometry 2 (G2).Notice that in G1, the nitrogen atom in DPP unit is close to the sulfur atom in the neighboring thiophene unit, while in G2, the nitrogen atom is in the vicinity of the hydrogen atom in the neighboring thiophene unit.

Figure S8 :
Figure S8: The vector diagrams for the modes of interest in geometry 2, corresponding to the geometry 1 modes shown in Figure 6 in the main article.

Table 1 :
Comparison between the experimental and simulated resonance Raman modes.Simulation results correspond to DPP-DTT geometry 1.