Template for Electronic Submission to ACS Journals

Control of donor-acceptor photophysics through structural modification of a “twisting” push-pull molecule

Thomas R. Hopper†, Deping Qian‡, Liyan Yang§, Xiaohui Wang¶, Ke Zhou¶, Rhea Kumar†, Wei Ma¶, Chang He§, Jianhui Hou§, Feng Gao‡ and Artem A. Bakulin†*

†Centre for Plastic Electronics and Department of Chemistry, Imperial College London, London W12 0BZ, United Kingdom‡Department of Physics, Chemistry and Biology (IFM), Linköping University, Linköping SE-58183, Sweden§State Key Laboratory of Polymer Physics and Chemistry, Institute of Physics and Chemistry, Chinese Academy of Sciences, Beijing 100190, China¶State Key Laboratory for Mechanical Behavior of Materials, Xi’an Jiaotong University, Xi’an 710049, China*Corresponding author email: a.bakulin@imperial.ac.uk

ABSTRACT: In contemporary organic solar cell (OSC) research, small A-D-A molecules comprising electron donor (D) and acceptor (A) units are increasingly used as a means to control the optoelectronic properties of photovoltaic blends. Slight structural variations to these A-D-A molecules can result in profound changes to the performance of the OSCs. Herein, we study two A-D-A molecules, BTCN-O and BTCN-M, which are identical in structure apart from a subtle difference in the position of alkyl chains, which force the molecules to adopt different equilibrium conformations. These steric effects cause the respective molecules to work better as an electron donor and acceptor when blended with benchmark acceptor and donor materials (PC71BM and PBDB-T). We study the photophysics of these “D:A” blends and devices using a combination of steady-state and time-resolved spectroscopic techniques. Time-resolved photoluminescence reveals the impact of the molecular conformation on the quenching of the A-D-A emission when BTCN-O and BTCN-M are blended with PBDB-T or PC71BM. Ultrafast broadband transient absorption spectroscopy demonstrates that the dynamics of charge separation are essentially identical when comparing BTCN-M and BTCN-O based blends, but the recombination dynamics are quite dissimilar. This suggests that the device performance is ultimately determined by the morphology of the blends imposed by the A-D-A conformation. This notion is supported by X-ray scattering measurements on the “D:A” films, and electroluminescence data and pump-push-photocurrent spectroscopy on the “D:A” devices. Our findings provide insight into the remarkable structure-function relationship in A-D-A molecules, and emphasize the need for careful morphological and energetic considerations when designing high-performance OSCs.

Push-pull molecules are conjugated organic molecules comprising electron-donating and electron-withdrawing groups separated by a π-system. This molecular configuration is arguably best recognized in the form of donor-acceptor (D-A) copolymers, which are commonly used in organic solar cells (OSCs) because their optical and electronic properties can be fine-tuned by adjusting the nature of intramolecular charge transfer (CT) between the D and A units.1 This is typically achieved by altering the donating and/or accepting ability of the respective units through chemical modification (i.e. the substitution of functional groups).

Since the first reported synthesis in 2008,2 the benzodithiophene (BDT) unit has become a popular building block for D-A copolymers in the OSC community.3–8 The planar conjugated structure provides strong visible absorption as well as good π-stacking; both of which, in addition to solubility, stability and electronic properties, can be readily tailored by appropriate modifications to the backbone and side chains.9 In the seminal article, the BDT-based polymer was synthesized with alkoxyl side chains,2 however later work showed that substituting these side chains for conjugated alkylthiophene groups significantly improved the thermal stability, hole mobility and general photovoltaic performance.10 The alkylthiophene-substituted BDT building block has since been utilized in small molecule OSCs, since the well-defined structure of the small molecule permits enhanced solution processability and more consistent blend morphologies than the polymeric counterparts.11

While the molecular structure of BDT-based small molecules can be extensively customized to achieve different optoelectronic properties,9 Liu et al. recently reported two, almost structurally identical, A-D-A molecules (BTCN-O and BTCN-M) that exhibit opposite D and A character in the presence of benchmark photovoltaic materials; the polymer donor poly[(2,6-(4,8-bis(5-(2-ethylhexyl)thiophen-2-yl)-benzo[1,2-b:4,5-b’]dithiophene))-alt-(5,5-(1’,3’-di-2-thienyl-5’,7’-bis(2-ethylhexyl)benzo[1’,2’-c:4’,5’-c’]dithiophene-4,8-dione)] (PBDB-T) and the fullerene acceptor [6,6]-Phenyl-C71-butyric acid methyl ester (PC71BM).12 The A-D-A molecules differ only in the position of octyl chains situated on the peripheral thiophene units attached to the BDT-core; the 4- and 5-positions (“ortho” substitution) in the case of BTCN-O, and the 3- and 5-positions (“meta” substitution) in the case of BTCN-M. This small difference in alkyl substitution causes the steric hindrance of a twisting motion in BTCN-M that is not present in BTCN-O, allowing BTCN-O to access a more planar conformation. Consequently, BTCN-O exhibits more prominent π-π stacking in the solid phase and acts more as an electron donor, while the relatively twisted BTCN-M behaves more as an electron acceptor.

In this contribution, we rationalize the differences in the photophysics and performance of the BTCN-O and BTCN-M based devices by linking the nanostructure of the D:A blends to the formation and recombination dynamics of free carriers. Time-resolved photoluminescence measurements show that quenching of the A-D-A emission in blends with molecular donors and acceptors is determined by both (i) the alignment of frontier molecular orbitals, and (ii) the blend morphology imposed by the A-D-A conformation. X-ray scattering measurements reveal enhanced π-π stacking and more crystalline domains in the BTCN-O based blends. Electroluminescence (EL) data confirm that the molecular interfaces are dissimilar and play very different roles in the materials systems based on BTCN-O and BTCN-M. Transient absorption and pump-push-photocurrent spectroscopy demonstrate that the dynamics of charge separation are essentially the same when comparing BTCN-M and BTCN-O based blends, but the charge recombination dynamics have noticeable differences. We propose that the fate of these charges and the performance of the devices are controlled by morphological differences that stem from the degree of twisting in the A-D-A structure. Our findings highlight the complex interplay of energetic and morphological effects in OSCs containing A-D-A small molecules.

Figure 1. (a) Chemical structures and (b) thin-film UV-Vis absorption spectra of the pristine components. (c) Thin-film UV-Vis absorption spectra of the blends. (d) Band diagram for all the pristine components and device electrodes/charge transport layers. Energy levels were obtained from the literature.12

The chemical structures of BTCN-M and BTCN-O are shown in Figure 1(a). Both molecules are comprised of an electron-rich BDT core flanked by 2-(3-oxo-2,3-dihydroinden-1-ylidene) units. These side units have strong electron-acceptor character emanating from the electron-deficient cyano and carbonyl functional groups. The only structural difference between these regioisomers are the positions of the octyl (-R) chains connected to the thiophene branches. The structures of PBDB-T and PC71BM are also shown.

Figure 1(b) contains the normalized UV-Vis absorption spectra for the pristine component thin films. In comparison to BTCN-M, the broad and redshifted absorption from BTCN-O is consistent with the notion of increased conjugation, and more importantly, enhanced intermolecular π-π stacking due to the sterically unhindered planarity of the molecule. This is further corroborated by the comparison of the absorption spectra for thin films and solutions performed by Liu et al. in a separate work.12 Apart from these differences, BTCN-O and BTCN-M share quite similar absorption profiles; notably a “double-humped” optical motif above the absorption onset, which can also be observed in PBDB-T. This feature is inexorably associated with the BDT moiety, which is present in all three molecules. Indeed, the same optical motif can be seen in numerous BDT-based organic molecules.3–8,11–14 In the case of PBDB-T, the low- and high-energy peaks comprising this feature were previously assigned to π-π* transitions along and between the polymer chains, respectively.5 The normalized optical spectra for the blend thin films are shown in Figure 1(c). Both BTCN-O blends exhibit an enhanced optical response in the 730-800 nm region with respect to their BTCN-M counterparts. The BTCN-O blends also give rise to a more structured optical response, which is indicative of the ordered molecular packing in these systems. The impact of these effects on the photophysics and performance of the devices is discussed later on.

Figure 1(d) depicts the band diagram for an inverted geometry device. For the sake of comparison, all the photoactive components are displayed alongside the cathode (Al), hole-transport layer (MoO3), anode (ITO) and electron-transport layer (ZnO). The energies of the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) for the photoactive components were taken from cyclic voltammetry measurements reported elsewhere.12 According to these measurements, the LUMO for BTCN-O and BTCN-M are equivalent in energy (-3.95 eV), while the HOMO of BTCN-M (-5.69 eV) is marginally deeper (0.1 eV) than that of BTCN-O (-5.59 eV). Consequently the bandgap of BTCN-O is slightly narrower, which is consistent with our optical measurements. Quantum chemical calculations by Liu et al. ascribe this to HOMO-delocalization across the thiophenes in the comparatively planar BTCN-O molecule.12 Based on the almost identical energetics of BTCN-O and BTCN-M, one would expect these molecules to behave equally well as electron donors and acceptors. Also, according to the conventional wisdom for OSC operation, one might expect more efficient D-A charge separation when the energetic offset between the frontier molecular orbitals is greater,15,16 i.e. in the PBDB-T based blends, where the driving forces for both electron and hole transfer are >0.3 eV.

Table 1. Performance characteristics for the inverted geometry “D:A” blend devices. VOC: open-circuit voltage; JSC: short-circuit current; FF: fill factor; PCE: power conversion efficiency.

“D:A” pairing

VOC

V

JSC

mA cm-2

FF

PCE

(%)

D

A

BTCN-M

PC71BM

0.935

0.506

0.36

0.17

BTCN-O

PC71BM

0.969

11.34

0.59

6.27

PBDB-T

BTCN-M

0.960

11.32

0.46

4.99

PBDB-T

BTCN-O

0.924

4.10

0.32

1.23

The device performance characteristics for the four blends are summarized in Table 1. The corresponding current-density-voltage (J-V) curves can be found in Figure S1. By comparing the performance of the PC71BM blends, we compare the performance of BTCN-M and BTCN-O as donors. BTCN-M:PC71BM and BTCN-O:PC71BM give fairly similar open circuit voltages (VOC), however the former blend gives a significantly lower short-circuit current (JSC) and fill factor (FF), which amounts to a very poor power conversion efficiency (PCE). The near-zero JSC of BTCN-M:PC71BM is of particular concern, and highlights that almost no charges are extracted from this blend. Liu et al. attribute this to the very low efficiency of exciton dissociation.12 This is reflected by the flat, almost-zero response in the external quantum efficiency (EQE) spectrum in Figure S1.

The electron-accepting ability of BTCN-M and BTCN-O can be assessed by contrasting the performance of the PBDB-T based blends. PBDB-T:BTCN-M and PBDB-T:BTCN-O give similar VOC values, but different JSC and FF values. Although not as stark as for the PC71BM blends, the large difference in JSC values appears to dominate the relative performance of these blends. The EQE spectra in Figure S1 show that while BTCN-O provides additional photocurrent from the extended absorption in the 700-800 nm range, the integrated EQE response for the PBDB-T:BTCN-M blend is much higher, indicating more efficient exciton dissociation in this blend.

The similarity in VOC values when comparing BTCN-M:PC71BM with BTCN-O:PC71BM BTCN-O, and PBDB-T:BTCN-M with PBDB-T:BTCN-O is expected given the negligible differences between the frontier energy levels of BTCN-M and BTCN-O. However, the CT state energies, and hence the VOC ought to be higher in the low-offset PC71BM blends than in the high-offset PBDB-T blends. We can only speculate that the higher VOC anticipated for the PC71BM blends is counteracted by some non-radiative loss process, which may be engendered by the specific morphology of the fullerene-based active layers.17 In any case, it is clear that the two A-D-A molecules behave in an opposite fashion when blended with D or A materials. It is also clear that the aforementioned energetic driving forces are not solely responsible for the performance of the devices, although they could have separate implications for the photophysics of the blends.

Figure 2. Time-resolved PL decay dynamics of the BTCN-M (a) and BTCN-O (b) based materials. Excitation wavelength: 635 nm; detection wavelength: 800 nm. IRF: instrument response function. Multiexponential fits provide a guide to the eye. EL spectra for the BTCN-M (c) and BTCN-O (d) based devices. For comparison, the PL spectra for the pristine materials (excitation at 532 nm) are shown as dashed lines.

Figure 2 depicts the exciton quenching characteristics in the D:A blends. From the PL decay dynamics in Figure 2(a), is manifest that BTCN-M behaves oppositely in the presence of PBDB-T and PC71BM. Rapid quenching of the BTCN-M emission is achieved in the former blend, while the latter blend presents almost no observable effect on the kinetics of the PL decay. These findings are compatible with the relative performance of the devices in Table 1, and the steady-state PL spectra for these materials reported by Liu et al.12 The PL decay of the BTCN-O based blends behave in a similar way; more rapid quenching of the BTCN-O emission is observed when blended with PBDB-T than with PC71BM. The observation of relatively weak PL quenching in the highest-performing device (BTCN-O:PC71BM) demonstrates that exciton quenching, and the driving force for charge separation, is not the main efficiency-limiting factor in these material systems. In fact, the coincidence of poor PL quenching and high device performance has been reported in numerous OSC systems.18,19 Moreover, our recent work serves to demonstrate that low-offset blends can outperform OSCs incorporating large driving forces for charge separation by minimizing non-radiative voltage losses.19

A striking result is obtained when comparing the PL spectra of pristine components with the EL spectra from the blend devices. Figure 2(c) shows that the EL from the BTCN-M:PC71BM blend is identical to the PL from the isolated BTCN-M film. This could be explained by radiative recombination of the injected charges from within the bulk BTCN-M phase, or emission from a CT state which is degenerate with the singlet exciton. Although hybridization between the singlet exciton and CT state manifolds has been advocated for other OSCs with low energy offsets,19,20 the poor device performance, morphology (discussed later) and PL quenching in the blend strongly point towards the former scenario as the origin of the EL emission. On the other hand, the EL from the PBDB-T:BTCN-M blend is substantially redshifted (0.3 eV) from the BTCN-M PL, which is highly characteristic of emission from an interfacial D:A CT state.21 Indeed, the position of the EL peak (~1.33 eV) is in reasonable agreement with the energetic difference between the PBDB-T HOMO and the BTCN-M LUMO (1.28 eV).

Figure 2(d) shows that the EL from PBDB-T:BTCN-O overlaps substantially with the PL from pristine BTCN-O. The EL spectrum for this blend also contains an additional lower energy component, again centered at ~1.3 eV. This suggests that there is some, albeit weak, contribution to the EL signal from the PBDB-T:BTCN-O CT state. Prominent emission from the bulk BTCN-O phase, rather than the D:A interface may occur if the intermixing of the blend components is suboptimal.

The EL from the BTCN-O:PC71BM blend in Figure 2(d) is more difficult to assign on first appearance. Based on the relatively high device performance (PCE = 6.27%), one would expect a strong contribution from the CT state, although the aforementioned degeneracy between the singlet exciton and CT state may mask this effect. Even in the absence of this hybridization, one would still expect the blend EL spectra to resemble the emission from pristine BTCN-O, since charges ought to funnel from PC71BM into the lower-gap BTCN-O prior to recombination. It is possible that the interaction between BTCN-O and PC71BM triggers a more ordered microstructure of BTCN-O, thereby modifying the emission properties of the blend. However, it is worth pointing out that the EL spectrum for this blend strongly resembles the EL from pristine fullerene devices reported elsewhere.21,22 Fullerene aggregates in this blend, particularly those located at the active layer:electrode interface, are therefore attributed to the observed emission. Further evidence in support of these aggregates is shown later in the text.

To study the crystallinity of the different blend films, we employed grazing-incidence wide-angle X-ray scattering (GIWAXS). Figure 3(a-d) shows that for all blends, diffraction peaks can be found in both the in-plane (IP) and out-of-plane (OOP) directions. The corresponding line cuts in Figure 3(e) provide the specific molecular packing information. The fitted parameters for all the materials are tabulated in Table S1. In BTCN-O:PC71BM and PBDB-T:BTCN-O, the narrow (100) and (001) IP diffraction peaks (q ~ 0.28 and 1.77 Å-1, respectively) can be ascribed to the BTCN-O crystalline domains (see neat film data in Figure S2). The coherence length (CL) of the (100) and (001) IP diffraction peaks were calculated to be 18.9 and 4.9 nm for BTCN-O:PC71BM, and 19.5 and 2.3 nm for PBDB-T:BTCN-O. This suggests that the BTCN-O molecules exhibit highly ordered lamellar packing and π-π stacking in both blends. For the other blends, the IP diffraction peaks at 0.32 Å-1 for BTCN-M:PC71BM and 0.29 Å-1 for PBDB-T:BTCN-M originate from BTCN-M (see Figure S2). The CL values of these peaks (3.9 and 16.4 nm, respectively) are substantially smaller than that from the BTCN-O blend counterparts. This, along with the disappearance of the BTCN-M (001) IP diffraction peaks in both blends is indicative of poor π-π stacking and amorphous domains.

Figure 3. 2D GIWAXS patterns of the four D:A blends (a-d). The corresponding line cuts for the in-plane and out-of-plane directions are also shown (e).

To unravel the dynamics of charge separation in the D:A pairings, broadband femtosecond transient absorption (TA) spectroscopy was conducted on the pristine components and material blends. Figure 4 contains the TA spectra for these systems upon selective excitation of the (low gap) BTCN-M or BTCN-O component. A low pump fluence was implemented to minimize the effects of multiphoton absorption or annihilation processes between excitons and/or charges. Fluence dependent TA kinetics are presented in Figure S3. Figure 4(a-b) demonstrates that BTCN-M and BTCN-O have similar TA responses. At early times (before 0.1 ps), both signals are composed of a ground state bleach (GSB) spanning the visible region, and a photoinduced absorption (PIA) in the NIR, centered at ~1100 nm. In both cases, the GSB resembles the shape of the linear absorption spectra for these molecules in Figure 1(b), and decays over the course of 5 ns, but does not fall to zero. Likewise, the PIA features also decay to non-zero amplitude over the 5 ns time window, and the peak also gradually redshifts towards 1300 nm. Both observations point to the emergence of an additional long-lived excited state, which we attribute to free charge generation in the push-pull molecules.

Figure 4. Time-averaged broadband TA spectra for alongside the relevant GA-derived exciton and charge dynamics. The grey shaded area omits the scattering from the 710 nm pump. A pump fluence of 1.3 μJ cm-2 was used in all cases. The scaling factor for the normalization of the charge component, relative to the exciton component, is shown for clarity. Multiexponential fits provide a guide to the eye.

To address the interconversion dynamics of these spectrally-overlapping states, we employed a global analysis (GA) procedure based on a genetic algorithm. The details of this procedure can be found elsewhere,23 and are described in more detail in the SI. We performed this analysis on the spectra in the >875 nm region in order to avoid any complications from electroabsorption (EA) or thermal effects that might occur in the visible range. For all the systems under study, a two-component fit was applied. The first of these components was extracted from the early-time (<0.5 ps) TA spectra of the pristine BTCN-M or BTCN-O films, which can be solely assigned to the singlet exciton. The second component was allowed to evolve via the genetic algorithm. We note that extending this analysis to the visible range, or employing higher-order deconvolutions containing more than two spectral components did not significantly improve the convergence of the fitting. Both the spectral and kinetic components of these fits are shown in Figure S4. For the sake of comparison, the kinetic components are shown alongside the relevant time-resolved spectra in Figure 4.

Figure 4(c-d) depicts the interconversion of the initially generated BTCN-M excitons (red) into free charges (black). In the former case, 50% of the exciton population decays on the order of ~10 ps, and >90% of the excitons have fully decayed by 1 ns. Of the 50% that decay within the first ~10 ps, some excitons are converted into charges, as evidenced by the rise of the black curve in Figure 4(c). The relative amplitude of this component is almost one order of magnitude smaller than the exciton component, which suggests a low branching (~10%) of the initially generated excitons into free carriers, while the majority of excitons decay non-radiatively. The small number of free carriers that are generated eventually decay on the order of ~1 ns. The interconversion dynamics are somewhat slower for BTCN-O in Figure 4(d), which could be the result of slower exciton diffusion prior to recombination and/or quenching.24 Again, this may be connected to the enhanced aggregation effects in BTCN-O.

The time-resolved TA spectra for the PBDB-T:BTCN-M and PBDB-T:BTCN-O blends are shown in Figure 4(e-f). Both early-time responses are composed of a PIA in the NIR, centered at ~1100 nm, as well as a GSB peaking at ~700 nm. This response can be readily assigned to the BTCN-M and BTCN-O excitons. Unexpectedly, the early-time response also contains a strong, negative contribution spanning ~550-640 nm, which is not observed in the pristine BTCN-M or BTCN-O films. Control measurements on the pristine PBDB-T film under the same excitation conditions (710 nm, 1.3 μJ cm-2) in Figure S5 preclude the direct generation of PBDB-T excitons. Furthermore, the wide bandgap of PBDB-T, relative to BTCN-M, rules out the possibility of ultrafast energy transfer from BTCN-M to PBDB-T, as reinforced by the selective excitation of PBDB-T in the blends, also shown in Figure S5. We speculate that this contribution may be related to ultrafast hole transfer, or possibly direct excitation of the CT state. As mentioned previously however, we only consider the interconversion dynamics in the >875 nm region. Here, the broadband TA response dramatically changes shape as a function of time. Within 10 ps, the PIA peaks blueshift towards ~925 nm, and extend into the higher energy part of the spectra around 750 nm. This feature peaking at ~925 nm has previously been assigned to the hole polaron on PBDB-T,14 and is confirmed by our own control TA measurements on a PBDB-T:PC71BM blend, shown in Figure S6.

The GA results for PBDB-T:BTCN-M and PBDB-T:BTCN-O in Figure 4(g-h) show that 50% of the initial exciton population is quenched within ~1 and 5 ps, respectively. This observation of rapid exciton quenching qualitatively agrees with the PL quenching data in Figure 2. The slightly slower dynamics in the latter blend may be connected to a diffusion-limited exciton dissociation process,24 perhaps as a consequence of the BTCN-O aggregation. This might help to explain both the shape of the EL spectrum in Figure 2 (which, by comparison to PBDB-T:BTCN-M, is not purely associated with the interfacial CT state), and the relative performance of the devices. We note however, that the relative amplitude of the exciton and charge signatures provided by GA are almost identical for these two blends, which suggests that the branching between excitons and CT states is equally likely. A more substantive difference in the photophysics of these blends occurs after CT (>100 ps). Namely, the charge component decays faster in PBDB-T:BTCN-O (~500 ps) than in PBDB-T:BTCN-M (~700 ps). We remark upon this behavior later.

The TA data for the BTCN-M:PC71BM and BTCN-O:PC71BM blends is shown Figure 4(i-j). The time-resolved spectral responses are almost identical to that of the pristine BTCN-M and BTCN-O films. Exciton and polaron signatures are expected to heavily overlap in these low offset blends, and the lack of a clear PIA associated with the electron polaron on PC71BM is expected given the highly symmetric nature of the fullerene dipole.25 Consequently, the GA-derived dynamics in Figure 4(k-l) are not particularly informative for these blends, although the relative amplitude of the charge components do agree with the relative device performance of the BTCN-M:PC71BM and BTCN-O:PC71BM systems. In the former case, the amplitude of the charge component is almost identical to that of the pristine BTCN-M data in Figure 4(c), which points to the absence of additional charge generation in the blend. Furthermore, by comparison to the PBDB-T blends, exciton quenching in the PC71BM blends appears to be rather slow (50% decay on the order of 20 ps).

A particularly interesting aspect of the BTCN-O:PC71BM TA data in Figure 4(j) is the derivative-like feature observed at ~550 nm. This feature is reminiscent of a Stark signal generated via the EA of free carriers. The energetic position of this signal coincides with previous observations of EA in pristine fullerene films.26,27 This observation, on top of the EL data in Figure 2 support the notion of fullerene clusters in the studied blend. Such domains have been suggested to facilitate the separation and band-like transport of charges in OSCs.28–31 Indeed, the persistence of this signal, even after 5 ns, points toward the successful long-range separation of charges in BTCN-O:PC71BM, which would account for the comparatively high device performance.

Taken together, the TA data suggest that the dynamics of exciton-charge conversion in the BTCN-M and BTCN-O based blends are quite similar, but the extent of this interconversion has noticeable differences which align with the device performance. To further examine the loss pathways in the devices, we performed pump-push-photocurrent (PPPC) spectroscopy on operational devices containing each of the D:A blends. Here, excitons generated by a visible pump pulse branch into precursory states for charge generation, or other loss pathways. These precursor states are then optically perturbed by an IR “push” pulse, and the additional photocurrent created by this perturbation is measured (dJ).32 The relative amplitude of dJ corresponds to the number of precursor states that are separated by the push pulse which would otherwise be lost to recombination events.

Figure 5. PPPC transients for the D:A blend devices. (a) dJ: Push-induced photocurrent; J: pump-induced photocurrent. Pump: 400 nm, <2 μJ cm-2; push: 2000 nm, <80 μJ cm-2. PPPC transients superimposed onto the GA-derived charge kinetics for each of the D:A blend films (b-e). Solid lines are multiexponential fits for a guide to the eye.

Figure 5(a) shows that the effect of the push (dJ/J) follows the trend in the performance of the devices. The best performing device (BTCN-O:PC71BM) gives an order of magnitude lower signal than the worst performing device (BTCN-M:PC71BM). This indicates the presence of substantially more “bound” states that are unable to separate into free carriers in the latter system. In order to assign the nature of these bound states, we compare the dJ/J transients for each device with the relevant GA-derived charge dynamics. Figure 5(b-d) shows that in the majority of cases, the PPPC and TA data are in excellent agreement. This emphasizes that for these blends, the states studied in both experiments have the same nature, and that the behavior of these states is not significantly affected by external factors from the measurement procedure. In the case of the PBDB-T based blends, the moderately high dJ/J and the sub-ns decay of the signal resemble the geminate recombination (GR) of interfacial CT states that has been observed in numerous OSC systems.16,33–37 The larger dJ/J and faster decay of the TA and PPPC signals for PBDB-T:BTCN-O suggests more prominent GR losses in this blend than the PBDB-T:BTCN-M counterpart, which agrees with the relative device performance. The BTCN-O:PC71BM data in Figure 5(d) also displays a good correlation between the TA and PPPC data. Here, both signals are seen to rise on the ~10 ps timescale, and decay with a time constant beyond that which is measurable. The small amplitude of the PPPC transient, and the long-lived nature of these decay profiles suggests that the electronic states probed by the PPPC experiment are trapped charges that have already undergone separation.38 Intriguingly, the PPPC and TA data for the BTCN-M:PC71BM system in Figure 5(e) bear little resemblance. The TA data shows a growth of the signal on the ~10 ps timescale until a plateau is reached. Meanwhile, the PPPC transient exhibits a slow decay on the ~400 ps timescale. The PPPC trace likely contains some contribution from bound excitons and trap states that are reactivated by the push pulse.38,39

Figure 6. State energy diagram and morphological schematic to explain the photophysics of the four D:A blends.

The state diagram in Figure 6 outlines the general fate of the excited states in the D:A blends. Selective excitation of the A-D-A molecules yields singlet excitons, of which a very small number spontaneously dissociate in the bulk phase to form free charges. For all the blends except BTCN-M:PC71BM, this process is outcompeted by CT at the D:A interface. As demonstrated by the time-resolved PL and TA data, the rate of this process is primarily controlled by the energy offset between the D and A components; the larger offset in the PBDB-T based blends affords more rapid quenching of the excitons for both BTCN-O and BTCN-M. However, this process clearly does not dictate the efficiency of the devices. Instead, the efficiency of the devices is controlled by the fate of charges following CT at the interface of the D:A pairings. The TA data in Figure 4 shows that some of these CT states are able to separate into free charges on the order of ~10 ps. The remaining bound CT states undergo GR on the sub-ns timescale. These bound states can be reactivated by the NIR push to create free carriers, which are measured as an additional photocurrent by the PPPC measurement.

To discriminate the photophysics of the D:A blends further, we will remark upon their individual morphologies based on the observations herein, as well as the atomic force and scanning electron microscopy data for these blends reported by Li et al.12 A simplified picture for the morphology of each blend is given in Figure 6(b-e). Starting with the lowest efficiency blend in Figure 6(b), the strong phase separation in BTCN-M:PC71BM means that excitons are unable to reach the D:A interface prior to recombination. The low number of charges generated in this blend, represented by the low JSC, mostly originate from spontaneous exciton dissociation in the bulk BTCN-M phase. By contrast, the partial phase segregation in the BTCN-O:PC71BM (Figure 6(c)), affords relatively efficient separation of charges, thereby suppressing GR.40 We posit that fullerene aggregates, indirectly observed in the EL and TA data, help to facilitate this process, as well as the subsequent transport of charges.28–31 Consequently, the optical perturbation provided in the PPPC experiment impacts charges that have already escaped the D:A interface,38 rather than CT states that are bound at the interface, or excitons in the bulk phase of either component.39

According to the actual morphology data, the PBDB-T:BTCN-M blend also exhibits decent, albeit amorphous D-A intermixing,12 as presented in Figure 6(d). This, in tandem with the observation of efficient exciton dissociation can explain the moderate performance of this material system.40 The strong CT character of this blend, as pointed out by the EL data, coupled with the absence of developed charge transport channels mean that blend suffers from strong GR losses. According to the PPPC data, the PBDB-T:BTCN-O blend also suffers from these losses, but the extent and rate of these losses are even more prominent. This is puzzling considering the phase-separated and aggregate-prone morphology of the blend.12 Both factors should result in a lower D-A interfacial area and present less opportunity to form the bound CT states that undergo GR,36,41 in accordance with the EL data. We tentatively assign this discrepancy to the stronger electron-hole coupling at the D:A interface,42,43 perhaps stemming from the specific nature of orbital overlap between the planar BTCN-O molecule and PBDB-T. This may affect the hybridization between the excitonic and CT states which control charge separation in other non-fullerene acceptor based OSCs.19,20

In this work, we studied the opposing electron-donating and accepting characteristics of two A-D-A molecules, which differ only in the position of peripheral alkyl chains. BTCN-O and BTCN-M were each blended with a benchmark donor (PBDB-T) and acceptor (PC71BM). The relative device performance of these “D:A” pairings (BTCN-O:PC71BM>PBDB-T:BTCN-M>PBDB-T:BTCN-O>BTCN-M:PC71BM) show that BTCN-O works best as a donor, while BTCN-M behaves more as an acceptor. Although the primary event in charge generation (exciton splitting) is sensitive to the alignment of energy levels, as evidenced by the fast decay of the A-D-A emission in the high-offset PBDB-T blends, the trend in device performance can be explained by quantitative differences in the recombination and long-scale separation of charges, as observed in transient absorption and pump-push-photocurrent spectroscopy. We attribute these differences to the individual morphologies of the “D:A” blends. X-ray scattering data confirm that the morphologies are highly dependent on the different π-π stacking motifs of the two A-D-A molecules imposed by the extent of “twisting” in the A-D-A structure. The findings herein embody the complex structure-function relationship in small molecule OSCs, where shifting the position of a single molecular bond can drastically affect the optoelectronic performance of the material system. These changes can arise from a combination of many factors, including the modulation of inter/intramolecular coupling, shifts in the energies of excited states and differences in the interface:volume ratio of D:A blends. This is likely the reason for much of the trial-and-error approach in current material development, and emphasizes the need for careful spectroscopic characterization when screening new systems, since similar materials can behave so differently. Further challenges in material development may involve the successful implementation of molecular twisting to control the properties and performance of OSCs.44–49

Experimental section

Fabrication of films and devicesThe devices were fabricated with an ITO/ZnO/active layer/MoO3/Al structure. The pre-cleaned ITO substrates were coated with ZnO by spin coating a precursor solution, prepared by dissolving 0.2 g zinc acetate dihydrate and 0.055 mL ethanolamine in 2 mL of 2-methoxyethanol. Subsequently, BTCN-O:PC71BM and BTCN-M:PC71BM in a 10 mg mL-1 chloroform solution were spin-coated at 2000 rpm for 30 s to obtain a film thickness of approximately 100 nm. The blends of BTCN-O:PBDB-T and BTCN-M:PBDB-T at different D/A ratios were dissolved in a mixed solvent (DCB:CB = 1:1, v/v) at a donor weight concentration of 10 mg mL-1. Finally, the device fabrication was completed by thermally evaporating 10 nm thick MoO3 and 100 nm thick aluminum under vacuum at a pressure of 3×10-4 Pa. The blend films were prepared from the same solutions on glass substrates. The pristine films were coated with solutions of 10 mg mL-1 in chlorobenzene or chloroform on glass substrates.

Device characterizationAll current density-voltage (J-V) curves were obtained under an AAA solar simulator (XES-70S1, SAN-EI Electric Co., Ltd) calibrated with a standard photovoltaic cell equipped with a KG3 filter (certificated by the National Institute of Metrology) and a Keithley 2400source-measure unit. The EQE was measured by a Solar Cell Spectral Response Measurement System QE-R3-011 (Enli Technology Co., Ltd., Taiwan). The light intensity at each wavelength was calibrated with a standard single-crystal Si photovoltaic cell.

UV-Vis absorption spectroscopyRoom temperature absorption spectra of the thin films were obtained using a JACSO UV-Vis V670 spectrometer. The sampling interval was 1 nm.

Time-resolved photoluminescenceTime-correlated single photon counting (TCSPC) was performed on the thin films using a DeltaFlex TCSPC instrument (Jobin Yvon IBH, HORIBA) equipped with a NanoLED excitation source (wavelength: 635 nm; intensity: ~1 mW cm-2; pulse duration: <200 ps; repetition rate: 1 MHz) and air-cooled detector (detection wavelength: 800±6 nm). A 665 nm longpass filter was used to prevent scattering of the excitation light.

Steady-state electroluminescence and photoluminescenceEL and PL spectra were recorded with an Andor spectrometer (Shamrock sr-303i-B, coupled to a Newton EMCCD Si array detector cooled to -60 °C). The system was wavelength-calibrated by an argon lamp to a resolution better than 0.5 nm. For EL, a Keithley 2400 external current/voltage source meter was connected to prepared photovoltaic devices to support an external electric field. For PL, the pumping source was a 532 nm laser with a power of 3 mW.

Grazing-incidence wide-angle X-ray scattering (GIWAXS)GIWAXS measurements were performed at beamline 7.3.3 at the Advanced Light Source (Lawrence Berkeley National Laboratory, Berkeley, California). Samples were prepared on Si substrates using identical blend solutions as those used in the devices. The 10 keV X-ray beam was incident at a grazing angle of 0.12-0.16°, selected to maximize the intensity of scattering from the samples. The scattered X-rays were detected using a photon counting detector (Pilatus 2M, Dectris).

Ultrafast broadband transient absorption spectroscopyA Ti:sapphire regenerative amplifier (Spectra Physics Solstice, Newport Corporation) provided seed pulses (800 nm, <200 fs, 1 kHz repetition rate) to an optical parametric amplifier (TOPAS, Light Conversion) coupled to a frequency mixer (NIRUVis, Light Conversion) for the tunable visible pump. The seed pulses were also directed to a mechanical delay stage before entering a commercially available femtosecond transient absorption spectrometer setup (HELIOS, Ultrafast Systems). The pump (modulated at 500 Hz) and the broadband probe (~450-750 nm for visible; ~850-1400 nm for near-infrared) were focused onto a ~0.5 mm2 spot on the sample, which was placed in a nitrogen-purged cuvette during measurements. Data sets for the visible and near-infrared probe were chirp corrected and stitched together in the SurfaceXplorer software, and analyzed further in MATLAB.

Pump-push-photocurrent spectroscopyPPPC was performed using the same laser setup as for TA spectroscopy (see above). The 800 nm seed pulses were delayed on the mechanical stage, then used to generate the 400 nm pump via the second harmonic in a β-barium borate crystal. The 2000 nm push pulses were generated from the optical parametric amplifier/frequency mixer. The pump and push were focused onto a ~0.5 mm2 spot on the device. A lock in amplifier (SR-830, Stanford Research Systems) was used to measure the photocurrent induced by the pump and push beams, modulated at 1000 and 350 Hz, respectively.

ASSOCIATED CONTENT

Supporting Information. The following supporting information is available free of charge.J-V and EQE curves for the blend devices. GIWAXS data for the pristine materials. Fluence-dependent TA kinetics for the pristine BTCN-M and BTCN-O films. Spectral and kinetic TA components derived from global analysis. Broadband TA spectra for the pristine and blend films at 710 and 450 nm excitation. Time-resolved TA spectra and global analysis derived dynamics for the PBDB-T based blends and PBDB-T:PC71BM control blend. Details on the global analysis procedure used to fit the TA data.

AUTHOR INFORMATION

ORCID

Thomas R. Hopper: 0000-0001-5084-1914Deping Qian: 0000-0001-8637-9178Wei Ma: 0000-0002-7239-2010Change He: 0000-0002-9804-5455Jianhui Hou: 0000-0002-2105-6922Feng Gao: 0000-0002-2582-1740Artem A. Bakulin: 0000-0002-3998-2000

Notes

The authors declare no competing financial interests.

ACKNOWLEDGMENT

T.R.H. acknowledges Jiangbin Zhang for fruitful discussions. The authors acknowledge the Optoelectronics Group at The University of Cambridge for providing the global analysis software. A.A.B. is a Royal Society University Research Fellow. X-ray data was acquired at beamlines 7.3.3 at the Advanced Light Source, which is supported by the Director, Office of Science, Office of Basic Energy Sciences, of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231. The authors thank Chenhui Zhu at beamline 7.3.3 for assistance with data acquisition.

REFERENCES

(1) Duan, C.; Huang, F.; Cao, Y. Recent Development of Push-Pull Conjugated Polymers for Bulk-Heterojunction Photovoltaics: Rational Design and Fine Tailoring of Molecular Structures. J. Mater. Chem. 2012, 22 (21), 10416–10434.

(2) Hou, J.; Park, M.-H.; Zhang, S.; Yao, Y.; Chen, L.-M.; Li, J.-H.; Yang, Y. Bandgap and Molecular Energy Level Control of Conjugated Polymer Photovoltaic Materials Based on Benzo[1,2- B :4,5- b ′]Dithiophene. Macromolecules 2008, 41 (16), 6012–6018.

(3) Cui, C.; Fan, H.; Guo, X.; Zhang, M.; He, Y.; Zhan, X.; Li, Y. Synthesis and Photovoltaic Properties of D–A Copolymers of Benzodithiophene and Naphtho[2,3-c]Thiophene-4,9-Dione. Polym. Chem. 2012, 3 (1), 99–104.

(4) Giovannitti, A.; Thorley, K. J.; Nielsen, C. B.; Li, J.; Donahue, M. J.; Malliaras, G. G.; Rivnay, J.; McCulloch, I. Redox-Stability of Alkoxy-BDT Copolymers and Their Use for Organic Bioelectronic Devices. Adv. Funct. Mater. 2018, 28 (17), 1706325.

(5) Qian, D.; Ye, L.; Zhang, M.; Liang, Y.; Li, L.; Huang, Y.; Guo, X.; Zhang, S.; Tan, Z.; Hou, J. Design, Application, and Morphology Study of a New Photovoltaic Polymer with Strong Aggregation in Solution State. Macromolecules 2012, 45 (24), 9611–9617.

(6) Zhang, S.; Ye, L.; Hou, J. Breaking the 10% Efficiency Barrier in Organic Photovoltaics: Morphology and Device Optimization of Well-Known PBDTTT Polymers. Adv. Energy Mater. 2016, 6 (11), 1502529.

(7) Dou, L.; Gao, J.; Richard, E.; You, J.; Chen, C.-C.; Cha, K. C.; He, Y.; Li, G.; Yang, Y. Systematic Investigation of Benzodithiophene- and Diketopyrrolopyrrole-Based Low-Bandgap Polymers Designed for Single Junction and Tandem Polymer Solar Cells. J. Am. Chem. Soc. 2012, 134 (24), 10071–10079.

(8) Peng, Q.; Liu, X.; Su, D.; Fu, G.; Xu, J.; Dai, L. Novel Benzo[1,2- B :4,5- b ′]Dithiophene-Benzothiadiazole Derivatives with Variable Side Chains for High-Performance Solar Cells. Adv. Mater. 2011, 23 (39), 4554–4558.

(9) Yao, H.; Ye, L.; Zhang, H.; Li, S.; Zhang, S.; Hou, J. Molecular Design of Benzodithiophene-Based Organic Photovoltaic Materials. Chem. Rev. 2016, 116 (12), 7397–7457.

(10) Huo, L.; Zhang, S.; Guo, X.; Xu, F.; Li, Y.; Hou, J. Replacing Alkoxy Groups with Alkylthienyl Groups: A Feasible Approach To Improve the Properties of Photovoltaic Polymers. Angew. Chemie Int. Ed. 2011, 50 (41), 9697–9702.

(11) Li, M.; Ni, W.; Wan, X.; Zhang, Q.; Kan, B.; Chen, Y. Benzo[1,2-b:4,5-B′]Dithiophene (BDT)-Based Small Molecules for Solution Processed Organic Solar Cells. J. Mater. Chem. A 2015, 3 (9), 4765–4776.

(12) Liu, D.; Yang, L.; Wu, Y.; Wang, X.; Zeng, Y.; Han, G.; Yao, H.; Li, S.; Zhang, S.; Zhang, Y.; et al. Tunable Electron Donating and Accepting Properties Achieved by Modulating the Steric Hindrance of Side Chains in A-D-A Small-Molecule Photovoltaic Materials. Chem. Mater. 2018, 30 (3), 619–628.

(13) Kan, B.; Feng, H.; Wan, X.; Liu, F.; Ke, X.; Wang, Y. Y. Y.; Wang, Y. Y. Y.; Zhang, H.; Li, C.; Hou, J.; et al. Small-Molecule Acceptor Based on the Heptacyclic Benzodi(Cyclopentadithiophene) Unit for Highly Efficient Nonfullerene Organic Solar Cells. J. Am. Chem. Soc. 2017, 139 (13), 4929–4934.

(14) Kan, B.; Zhang, J.; Liu, F.; Wan, X.; Li, C.; Ke, X.; Wang, Y.; Feng, H.; Zhang, Y.; Long, G.; et al. Fine-Tuning the Energy Levels of a Nonfullerene Small-Molecule Acceptor to Achieve a High Short-Circuit Current and a Power Conversion Efficiency over 12% in Organic Solar Cells. Adv. Mater. 2018, 30 (3), 1–8.

(15) Clarke, T. M.; Durrant, J. R. Charge Photogeneration in Organic Solar Cells. Chem. Rev. 2010, 110 (11), 6736–6767.

(16) Bakulin, A. A.; Rao, A.; Pavelyev, V. G.; Van Loosdrecht, P. H. M.; Pshenichnikov, M. S.; Niedzialek, D.; Cornil, J.; Beljonne, D.; Friend, R. H. The Role of Driving Energy and Delocalized States for Charge Separation in Organic Semiconductors. Science 2012, 335 (6074), 1340–1344.

(17) Tang, Z.; Wang, J.; Melianas, A.; Wu, Y.; Kroon, R.; Li, W.; Ma, W.; Andersson, M. R.; Ma, Z.; Cai, W.; et al. Relating Open-Circuit Voltage Losses to the Active Layer Morphology and Contact Selectivity in Organic Solar Cells. J. Mater. Chem. A 2018, 6 (26), 12574–12581.

(18) Menke, S. M.; Ran, N. A.; Bazan, G. C.; Friend, R. H. Understanding Energy Loss in Organic Solar Cells: Toward a New Efficiency Regime. Joule 2018, 2 (1), 25–35.

(19) Qian, D.; Zheng, Z.; Yao, H.; Tress, W.; Hopper, T. R.; Chen, S.; Li, S.; Liu, J.; Chen, S.; Zhang, J.; et al. Design Rules for Minimizing Voltage Losses in High-Efficiency Organic Solar Cells. Nat. Mater. 2018, 17 (8), 703–709.

(20) Eisner, F. D.; Azzouzi, M.; Fei, Z.; Hou, X.; Anthopoulos, T. D.; Dennis, T. J. S.; Heeney, M. J.; Nelson, J. Hybridization of Local Exciton and Charge-Transfer States Reduces Nonradiative Voltage Losses in Organic Solar Cells. J. Am. Chem. Soc. 2019, 141 (15), 6362–6374.

(21) Tvingstedt, K.; Vandewal, K.; Gadisa, A.; Zhang, F.; Manca, J.; Inganäs, O. Electroluminescence from Charge Transfer States in Polymer Solar Cells. J. Am. Chem. Soc. 2009, 131 (33), 11819–11824.

(22) Kraus, H.; Heiber, M. C.; Väth, S.; Kern, J.; Deibel, C.; Sperlich, A.; Dyakonov, V. Analysis of Triplet Exciton Loss Pathways in PTB7:PC71BM Bulk Heterojunction Solar Cells. Sci. Rep. 2016, 6 (1), 29158.

(23) Gélinas, S.; Paré-Labrosse, O.; Brosseau, C.-N.; Albert-Seifried, S.; McNeill, C. R.; Kirov, K. R.; Howard, I. A.; Leonelli, R.; Friend, R. H.; Silva, C. The Binding Energy of Charge-Transfer Excitons Localized at Polymeric Semiconductor Heterojunctions. J. Phys. Chem. C 2011, 115 (14), 7114–7119.

(24) Markov, D. E.; Amsterdam, E.; Blom, P. W. M.; Sieval, A. B.; Hummelen, J. C. Accurate Measurement of the Exciton Diffusion Length in a Conjugated Polymer Using a Heterostructure with a Side-Chain Cross-Linked Fullerene Layer. J. Phys. Chem. A 2005, 109 (24), 5266–5274.

(25) Arbogast, J. W.; Foote, C. S. Photophysical Properties of C70. J. Am. Chem. Soc. 1991, 113 (23), 8886–8889.

(26) Causa’, M.; Ramirez, I.; Martinez Hardigree, J. F.; Riede, M.; Banerji, N. Femtosecond Dynamics of Photoexcited C 60 Films. J. Phys. Chem. Lett. 2018, 9 (8), 1885–1892.

(27) Devižis, A.; De Jonghe-Risse, J.; Hany, R.; Nüesch, F.; Jenatsch, S.; Gulbinas, V.; Moser, J.-E. Dissociation of Charge Transfer States and Carrier Separation in Bilayer Organic Solar Cells: A Time-Resolved Electroabsorption Spectroscopy Study. J. Am. Chem. Soc. 2015, 137 (25), 8192–8198.

(28) Gélinas, S.; Rao, A.; Kumar, A.; Smith, S. L.; Chin, A. W.; Clark, J.; van der Poll, T. S.; Bazan, G. C.; Friend, R. H. Ultrafast Long-Range Charge Separation in Organic Semiconductor Photovoltaic Diodes. Science 2014, 343 (6170), 512–516.

(29) Jakowetz, A. C.; Bohm, M. L.; Zhang, J.; Sadhanala, A.; Huettner, S.; Bakulin, A. A.; Rao, A.; Friend, R. H. What Controls the Rate of Ultrafast Charge Transfer and Charge Separation Efficiency in Organic Photovoltaic Blends. J. Am. Chem. Soc. 2016, 138 (36), 11672–11679.

(30) Melianas, A.; Pranculis, V.; Spoltore, D.; Benduhn, J.; Inganäs, O.; Gulbinas, V.; Vandewal, K.; Kemerink, M. Charge Transport in Pure and Mixed Phases in Organic Solar Cells. Adv. Energy Mater. 2017, 7 (20), 1700888.

(31) Jamieson, F. C.; Domingo, E. B.; McCarthy-Ward, T.; Heeney, M.; Stingelin, N.; Durrant, J. R. Fullerenecrystallisation as a Key Driver of Charge Separation in Polymer/Fullerene Bulk Heterojunction Solar Cells. Chem. Sci. 2012, 3 (2), 485–492.

(32) Bakulin, A. A.; Silva, C.; Vella, E. Ultrafast Spectroscopy with Photocurrent Detection: Watching Excitonic Optoelectronic Systems at Work. J. Phys. Chem. Lett. 2016, 7 (2), 250–258.

(33) Kurpiers, J.; Ferron, T.; Roland, S.; Jakoby, M.; Thiede, T.; Jaiser, F.; Albrecht, S.; Janietz, S.; Collins, B. A.; Howard, I. A.; et al. Probing the Pathways of Free Charge Generation in Organic Bulk Heterojunction Solar Cells. Nat. Commun. 2018, 9 (1), 2038.

(34) Hwang, I.-W.; Moses, D.; Heeger, A. J. Photoinduced Carrier Generation in P3HT/PCBM Bulk Heterojunction Materials. J. Phys. Chem. C 2008, 112 (11), 4350–4354.

(35) Müller, J. G.; Lupton, J. M.; Feldmann, J.; Lemmer, U.; Scharber, M. C.; Sariciftci, N. S.; Brabec, C. J.; Scherf, U. Ultrafast Dynamics of Charge Carrier Photogeneration and Geminate Recombination in Conjugated Polymer:Fullerene Solar Cells. Phys. Rev. B 2005, 72 (19).

(36) Mangold, H.; Bakulin, A. A.; Howard, I. A.; Kästner, C.; Egbe, D. A. M.; Hoppe, H.; Laquai, F. Control of Charge Generation and Recombination in Ternary Polymer/Polymer:Fullerene Photovoltaic Blends Using Amorphous and Semi-Crystalline Copolymers as Donors. Phys. Chem. Chem. Phys. 2014, 16 (38), 20329–20337.

(37) Dyer-Smith, C.; Howard, I. A.; Cabanetos, C.; El Labban, A.; Beaujuge, P. M.; Laquai, F. Interplay Between Side Chain Pattern, Polymer Aggregation, and Charge Carrier Dynamics in PBDTTPD:PCBM Bulk-Heterojunction Solar Cells. Adv. Energy Mater. 2015, 5 (9), 1401778.

(38) Dong, Y.; Cha, H.; Zhang, J.; Pastor, E.; Tuladhar, P. S.; McCulloch, I.; Durrant, J. R.; Bakulin, A. A. The Binding Energy and Dynamics of Charge-Transfer States in Organic Photovoltaics with Low Driving Force for Charge Separation. J. Chem. Phys. 2019, 150 (10), 104704.

(39) Weu, A.; Hopper, T. R.; Lami, V.; Kreß, J. A.; Bakulin, A. A.; Vaynzof, Y. Field-Assisted Exciton Dissociation in Highly Efficient PffBT4T-2OD:Fullerene Organic Solar Cells. Chem. Mater. 2018, 30 (8), 2660–2667.

(40) Dimitrov, S. D.; Azzouzi, M.; Wu, J.; Yao, J.; Dong, Y.; Tuladhar, P. S.; Schroeder, B. C.; Bittner, E. R.; McCulloch, I.; Nelson, J.; et al. Spectroscopic Investigation of the Effect of Microstructure and Energetic Offset on the Nature of Interfacial Charge Transfer States in Polymer: Fullerene Blends. J. Am. Chem. Soc. 2019, 141 (11), 4634–4643.

(41) Zhang, J.; Gu, Q.; Do, T. T.; Rundel, K.; Sonar, P.; Friend, R. H.; McNeill, C. R.; Bakulin, A. A. Control of Geminate Recombination by the Material Composition and Processing Conditions in Novel Polymer: Nonfullerene Acceptor Photovoltaic Devices. J. Phys. Chem. A 2018, 122 (5), 1253–1260.

(42) Onsager, L. Initial Recombination of Ions. Phys. Rev. 1938, 54 (8), 554–557.

(43) Nübling, F.; Hopper, T. R.; Kuei, B.; Komber, H.; Untilova, V.; Schmidt, S. B.; Brinkmann, M.; Gomez, E. D.; Bakulin, A. A.; Sommer, M. Block Junction-Functionalized All-Conjugated Donor–Acceptor Block Copolymers. ACS Appl. Mater. Interfaces 2019, 11 (1), 1143–1155.

(44) Ko, S.; Hoke, E. T.; Pandey, L.; Hong, S.; Mondal, R.; Risko, C.; Yi, Y.; Noriega, R.; McGehee, M. D.; Brédas, J.-L.; et al. Controlled Conjugated Backbone Twisting for an Increased Open-Circuit Voltage While Having a High Short-Circuit Current in Poly(Hexylthiophene) Derivatives. J. Am. Chem. Soc. 2012, 134 (11), 5222–5232.

(45) Hwang, Y.-J.; Li, H.; Courtright, B. A. E.; Subramaniyan, S.; Jenekhe, S. A. Nonfullerene Polymer Solar Cells with 8.5% Efficiency Enabled by a New Highly Twisted Electron Acceptor Dimer. Adv. Mater. 2016, 28 (1), 124–131.

(46) Wang, W.; Zhao, B.; Cong, Z.; Xie, Y.; Wu, H.; Liang, Q.; Liu, S.; Liu, F.; Gao, C.; Wu, H.; et al. Nonfullerene Polymer Solar Cells Based on a Main-Chain Twisted Low-Bandgap Acceptor with Power Conversion Efficiency of 13.2%. ACS Energy Lett. 2018, 3 (7), 1499–1507.

(47) Xu, S. jie; Zhou, Z.; Liu, W.; Zhang, Z.; Liu, F.; Yan, H.; Zhu, X. A Twisted Thieno[3,4- b ]Thiophene-Based Electron Acceptor Featuring a 14-π-Electron Indenoindene Core for High-Performance Organic Photovoltaics. Adv. Mater. 2017, 29 (43), 1704510.

(48) Zhan, C.; Yao, J. More than Conformational “Twisting” or “Coplanarity”: Molecular Strategies for Designing High-Efficiency Nonfullerene Organic Solar Cells. Chem. Mater. 2016, 28 (7), 1948–1964.

(49) Sasaki, S.; Drummen, G. P. C.; Konishi, G. Recent Advances in Twisted Intramolecular Charge Transfer (TICT) Fluorescence and Related Phenomena in Materials Chemistry. J. Mater. Chem. C 2016, 4 (14), 2731–2743.

1

21