Effects of side chain isomerism on the physical and photovoltaic properties of indacenodithieno[3,2-b]thiophene–quinoxaline copolymers: toward a side chain design for enhanced photovoltaic performance

Four new D–A polymers PIDTT-Q-p, PIDTT-Q-m, PIDTT-QF-p and PIDTT-QF-m, using indacenodithieno [3,2-b]thiophene (IDTT) as an electron-rich unit and quinoxaline (Q) as an electron-deficient unit, were synthesized via a Pd-catalyzed Stille polymerization. The side chains on the pendant phenyl rings of IDTT were varied from the parato the meta-position, and the effect of the inclusion of fluorine on the quinoxaline unit was simultaneously investigated. The influence on the optical and electrochemical properties, film topography and photovoltaic properties of the four copolymers were thoroughly examined via a range of techniques. The inductively electron-withdrawing properties of the fluorine atoms result in a decrease of the highest occupied molecular orbital (HOMO) energy levels. The effect of meta-substitution on the PIDTT-Q-m polymer leads to good solubility and in turn higher molecular weight. More importantly, it exhibits optimal morphological properties in the PIDTT-Q-m/PC71BM blends. As a result, the corresponding solar cells (ITO/PEDOT:PSS/polymer:PC71BM/LiF/Al) attain the best power conversion efficiency (PCE) of 6.8%. The structure–property correlations demonstrate that the meta-alkyl-phenyl substituted IDTT unit is a promising building block for efficient organic photovoltaic materials. This result also extends our strategy with regards to side chain isomerism of IDTT-based copolymers for enhanced photovoltaic performance.


Introduction
As a novel solar energy harvesting medium, polymer solar cells (PSCs) have been intensively investigated in recent years as they offer the potential to be light weight, exible and manufactured on a large-area scale at low cost. [1][2][3][4] So far, bulk-heterojunction polymer solar cells (BHJ PSCs), using a solution-processed active layer composed of an electron-donor and an electronacceptor, sandwiched between ITO and metallic electrodes, can attain promising power conversion efficiencies (PCEs) of 8-9%. 5 Conjugated donor-acceptor polymers, combining an electrondonating (D) and an electron-withdrawing (A) moiety, are particularly promising now, since judicious selection of D and A moieties can tailor the D-A interaction and p-electron delocalization, to achieve tunable band gaps, energy levels and charge transporting properties for ideal electron-donor materials. [6][7][8] In principle, an effective strategy is to combine a weak D and a strong A unit alternately. The weak D moiety can anchor a lowlying HOMO level, whilst, the strong A moiety can provide a favorable LUMO level and a suitable band gap. 9 In addition, further optimization of a given D-A framework via improvements in solubility, molecular weight and structural orientation can be achieved through side chain modulation. [10][11][12][13][14] To date, a variety of electron-rich arenes comprising of multiply fused aromatic systems have been reported. 15 Rigid backbones with horizontally or vertically extended p-conjugation can be formed by fastening or fusing adjacent building blocks to aromatic cores. One appealing electron-donating unit is the ladder-type indacenodithiophene (IDT) unit. [16][17][18][19][20][21] This structure of aromatic rings has enforced planarization which can easily suppress interannular rotation and enhance p-electron delocalization. The high degree of planarity is conducive to intermolecular charge-carrier hopping and intermolecular interactions between conjugated backbones, which thus results in high charge-carrier mobility. 22 In an effort to further extend the linear p-conjugation of the IDT unit, the central phenyl ring was covalently bonded to two thieno[3,2-b]thiophenes (TT) to form the indacenodithieno [3,2-b]thiophene (IDTT) arene, which enhances the conjugation of the system through increased planarization on changing the pentacyclic rings to heptacyclic rings. Recently, an IDTT-based polymer with a HOMO level of À5.3 eV and a medium band-gap (E g ) of 1.8 eV was reported by Jen et al., which can attain a PCE of 7%. 23 By copolymerizing IDTT with different acceptor units, tunable optoelectronic properties and charge-carrier mobilities were obtained with PCEs around 4-5%. 24,25 Several architectural designs of devices incorporating non-halogenated solvents and interfacial engineering have resulted in high PCE of 7%. 26,27 However, to expand the family of ladder-type arenes, most efforts have been focused on the designs of building blocks for polymer backbones, [28][29][30][31][32][33] whilst only a few reports have shed light on the inuence of side chain isomerization. [34][35][36] Recently, we studied two series of thiophene-quinoxaline 37-39 and IDTquinoxaline 40 copolymers and found that using meta-alkylphenyl instead of para-alkyl-phenyl side chains can enhance the solubility, molecular weight and photovoltaic performances of the copolymers. The corresponding uorinated IDT-based polymer offered a remarkably high V oc of 0.96 V with a PCE of 6.6%. Meanwhile, the non-uorinated analogue attained a PCE as high as 7.8%. The planarity and packing distance between the polymer backbones can be varied due to the steric hindrance between adjacent side chains, and then the solubility, molecular weights and packing properties can be changed on a macro level. For the structural optimization of conjugated materials, the proper anchoring of the side groups seems very useful for improving the device efficiency of PSCs. Therefore, here we attempt to extend this side-chain design strategy to the IDTT unit, which so far has been only used in IDT and quinoxaline-based copolymers.
To this end, we have designed and synthesized four copolymers based on the IDTT donor units, and quinoxaline acceptor units (Scheme 2). This is the rst direct comparison and evaluation of photovoltaic performances among the IDTT copolymers incorporating different side groups, each of which contains para-or meta-side chains on pendent phenyl rings. In addition, both non-uorinated (Q) and uorinated quinoxaline (QF) acceptor units were chosen for comparison, since the uorinated acceptor was expected to feature a lower-lying HOMO level due to the electron-withdrawing property of uorine atoms. [41][42][43][44][45][46][47] All polymers were prepared via the Stille coupling reaction of bis(trimethylstannyl)-substituted IDTT and dibromo-substituted quinoxaline monomers. The solubility, UV-Vis absorption and electrochemical properties were systematically investigated to understand the structure-property correlations of side chain isomerism and the inclusion of uorine. BHJ PSCs using PC 71 BM as the electron acceptor were fabricated for evaluating polymer performance. As anticipated, the devices based on uorinated copolymers featured higher V oc of close to 1.0 V. Compared to the para-substituted analogue PIDTT-Q-p, the meta-substituted polymer PIDTT-Q-m:PC 71 BM device attains a superior PCE of 6.8%, which is among the highest PCEs of IDTT copolymers recorded for the conversional BHJ PSC conguration. This nding agrees well with our previous study, where polymers with meta-substituted side chains show superior photovoltaic performances compared to para-substituted analogues. Characterizations of the photoresponse and lm morphology indicate that there is a direct correlation between lm morphology and device performance for these four copolymers.

Experimental section
Characterization 1 H NMR (400 MHz) and 13 C NMR (100 MHz) spectra were acquired using a Varian Inova 400 MHz NMR spectrometer. Tetramethylsilane was used as an internal reference with deuterated chloroform as the solvent. Size exclusion chromatography (SEC) was performed on an Agilent PL-GPC 220 Integrated High Temperature GPC/SEC System with refractive index and viscometer detectors. The columns are 3 PLgel 10 mm MIXED-B LS 300 Â 7.5 mm columns. The eluent was 1,2,4-trichlorobenzene. The working temperature was 150 C. The molecular weights were calculated according to relative calibration with polystyrene standards. UV-Vis absorption spectra were measured with a Perkin Elmer Lambda 900 UV-Vis-NIR absorption spectrometer. Square wave voltammetry (SWV) measurements were carried out on a CH-Instruments 650A Electrochemical Workstation. A three-electrode setup was used with platinum wires for both the working electrode and counter electrode, and Ag/Ag + was used for the reference electrode calibrated with a ferrocene/ferrocenyl couple (Fc/Fc + ). A 0.1 M nitrogen-saturated solution of tetrabutylammonium hexa-uorophosphate (Bu 4 NPF 6 ) in anhydrous acetonitrile was used as the supporting electrolyte. The polymer lms were deposited onto the working electrode from chloroform solution. Tappingmode atomic force microscopy (AFM) images were acquired with an Agilent-5400 scanning probe microscope using a Nanodrive controller with MikroMasch NSC-15 AFM tips and resonant frequencies of $300 kHz. Transmission electron microscopy (TEM) was performed with a FEI Tecnai T20 (LaB 6 , 200 kV). Without LiF/Al electrode deposition, the active layer was placed onto a copper grid using an aqueous dispersion of PEDOT:PSS. The sample was dried at room temperature before the TEM experiments were performed.
To a solution of thieno[3,2-b]thiophene (2.10 g, 15.0 mmol) in anhydrous THF (30 mL) at À30 C was added n-BuLi (6.3 mL, 2.5 M in hexane), the mixture was kept at À30 C for 0.5 h and warmed to 0 C. Then a solution of anhydrous ZnCl 2 (2.25 g, 16.5 mmol) in THF (10 mL) was added slowly. Aer the addition, the mixture was stirred at 0 C for 0.5 h. Finally, diethyl 2,5dibromoterephthalate (1) (2.38 g, 6.25 mmol) and Pd(PPh 3 ) 4 (0.14 g, 0.12 mmol) were added to the mixture. The reaction mixture was reuxed for 24 h. Aer cooling to room temperature, the excess solvent was distilled and the resulting solid was washed successively with ethanol. The residue was collected and puried by column chromatography with 1 : 2 (v/v) ethyl acetate-hexane as the eluent to give the nal compound as a yellow solid (2.19 g, 70.2%). 1

Compound 3
To a solution of compound 2 (1.50 g, 3.0 mmol) in anhydrous THF (50 mL) was added a solution of freshly prepared 3-hexylphenyl magnesium bromide from 1-bromo-3-hexylbenzene (3.62 g, 15.0 mmol) in anhydrous THF (30 mL). The solution was reuxed for 12 h. Aer cooling to room temperature, the organic layer was extracted with ethyl acetate (100 mL), washed successively with saturated brine and then dried over anhydrous MgSO 4 . The residue was puried by column chromatography with 1 : 15 (v/v) ethyl acetate-hexane as the eluent to afford a crude product as a dark red solid (2.04 g, 64.4%).

Compound 4
Compound 3 (2.0 g, 1.9 mmol) was dissolved in warm glacial acetic acid (100 mL) and conc. H 2 SO 4 (2 mL) was added slowly. The reaction mixture was reuxed for 30 min. Aer cooling to room temperature, the organic layer was extracted with dichloromethane (50 mL), washed successively with 1 M K 2 CO 3 aqueous solution and then dried over anhydrous MgSO 4 . The residue was puried by column chromatography with 1 : 5 (v/v) dichloromethane-hexane as the eluent to give the nal compound as a light yellow solid (1.78 g, 92.2%). 1

Monomer 2 (M2)
To a solution of compound 4 (1.02 g, 1.0 mmol) in anhydrous THF (10 mL) was added n-BuLi (1.0 mL, 2.5 M in hexane) at À30 C. The reaction mixture was stirred at À30 C for 0.5 h and then warmed to room temperature for another 2 h. Aer that, it was cooled to À30 C again and Me 3 SnCl (3.0 mL, 1 M in hexane) was added in one portion. The reaction mixture was stirred at room temperature overnight and then poured into water. The organic layer was extracted with diethyl ether (50 mL), washed successively with water, and dried over anhydrous MgSO 4  General procedure for polymerization 0.15 mmol of dibromo-substituted monomer, 0.15 mmol of bis(trimethylstannyl)-substituted monomer, tris(dibenzylideneacetone)dipalladium(0) (Pd 2 (dba) 3 ) (2.75 mg) and tri(o-tolyl) phosphine (P(o-Tol) 3 ) (3.65 mg) were dissolved in anhydrous toluene (12 mL) under a nitrogen atmosphere. The reaction mixture was reuxed with vigorous stirring for 24 h. Aer cooling to room temperature, the polymer was precipitated by pouring the solution into acetone and was collected by ltration through a 0.45 mm Teon lter. Then the polymer was washed in a Soxhlet extractor with acetone, diethyl ether and chloroform. The chloroform fraction was puried by passing it though a short silica gel column and then precipitated from acetone again. Finally, the polymer was obtained by ltration through 0.45 mm Teon lter and dried under vacuum at 40 C overnight.

PSC fabrication and characterization
The structure of polymer solar cells was Glass/ITO/PEDOT:PSS/ polymer:PC 71 BM/LiF/Al. As a buffer layer, PEDOT:PSS (Baytron P VP Al 4083) was spin-coated onto ITO-coated glass substrates, followed by annealing at 150 C for 15 minutes to remove water. The thickness of the PEDOT:PSS layer was around 40 nm, as determined by a Dektak 6 M surface prolometer. The active layer consisting of polymers and PC 71 BM was spin-coated from o-dichlorobenzene (oDCB) solution onto the PEDOT:PSS layer. The spin-coating was done in a glove box and they were directly transferred to a vapor deposition system mounted inside of the glove box. The thicknesses of active layers were in the range of 85-95 nm. LiF (0.6 nm) and Al (100 nm) were used as the top electrodes and were deposited via a mask under vacuum onto the active layer. The accurate area of every device (4.5 mm 2 ), dened by the overlap of the ITO and metal electrode, was measured carefully by microscope. The PCEs were calculated from the J-V characteristics recorded by a Keithley 2400 source meter under the illumination of an AM 1.5G solar simulator with an intensity of 100 mW cm À2 (Model SS-50A, Photo Emission Tech., Inc.). The light intensity was determined using a standard silicon photodiode. EQEs were calculated from the photocurrents under short-circuit conditions. The currents were recorded using a Keithley 485 picoammeter under monochromatic light (MS257) illumination through the ITO side of the devices.
Scheme 2 shows the synthetic routes for the polymers. The polymerizations were accomplished via Pd 2 (dba) 3 -catalyzed Stille coupling of the corresponding bis(trimethylstannyl)substituted monomers M1, M2 and dibromo-substituted monomers M3, M4, respectively. The same reaction time (24 h) was used for each polymer to study the inuences of the side chains and uorination. Aer polymerization, the crude polymer was washed in a Soxhlet extractor with acetone and diethyl ether for 24 h each. Aer that, the polymer was Soxhletextracted with chloroform. In this study, the properties of the polymer batches with the highest molecular weights were evaluated providing that they were readily soluble in chloroform, chlorobenzene, and oDCB. The molecular weights of the four polymers were measured by size exclusion chromatography (SEC) at 150 C with 1,2,4-trichlorobenzene as the eluent. The number-average molecular weights (M n ) of the meta-substituted polymer PIDTT-Q-m and PIDTT-QF-m are 41.8 and 17.2 kg mol À1 with polydispersity indexes (PDI) of 3.2 and 2.2, respectively. The para-substituted polymer PIDTT-Q-p and PIDTT-QF-p exhibit obviously lower M n of 33.2 and 14.1 kg mol À1 , with PDI of 2.9 and 2.2, respectively. The meta-substituted phenyl groups improve the solubility of the IDTT copolymers to provide higher molecular weight copolymers, which thus agree well with our previous study on IDT copolymers. 40 Similar results were also observed in thiophene-quinoxaline (TQ-m) copolymers. The kinked meta-side chains on quinoxaline moiety result in more twisted polymer backbone, which prevents aggregation in solution and decreases the enthalpy change DH diss during dissolution. From the equation: DG ¼ DH À TDS and T diss ¼ DH diss /DS diss , the dissolution temperature T diss is reached if the Gibbs free energy DG ¼ 0, Therefore, the lower DH diss would decrease T diss and improve the solubility of TQ-m polymers. 39 Compared to the non-uorinated analogues, both of the uorinated IDTT copolymers exhibit lower molecular weights, possibly owing to less coupling activity from increased steric hindrance between the IDTT and uorinated quinoxaline units.
Scheme 1 Synthetic routes of the monomers.

Absorption spectra
The normalized UV-Vis absorption spectra of the four polymers both in chloroform solution and in thin lm are shown in Fig. 1. All the polymers depict two distinct absorption bands in the wavelength range of 350-450 and 500-700 nm, corresponding to the p-p* transition and intramolecular charge transfer (ICT) between the IDTT and quinoxaline moieties, respectively. Compared to IDT-based polymers, horizontal extension of pconjugation by fastening outer TT units efficiently improve the ICT process, since the four IDTT-based polymers here exhibit stronger absorption in the long-wavelength regime. 19 The absorption maxima (l max ) of the polymers PIDTT-Q-p, PIDTT-Qm, PIDTT-QF-p and PIDTT-QF-m in dilute chloroform solutions are around 620 nm (Table 1). A well-resolved shoulder peak is recorded for each of the uorinated polymers. In thin lms, the absorption of the meta-substituted polymers PIDTT-Q-m and PIDTT-QF-m show bathochromic shis to 630 nm and 643 nm, respectively. A somewhat boarder absorption of the non-uorinated polymer PIDTT-Q-m is observed in comparison to the uorinated polymer PIDTT-QF-p. On the contrary, the parasubstituted polymers PIDTT-Q-p and PIDTT-QF-p exhibit little broader spectra in the solid state, l max values of which shi to 623 nm and 628 nm, respectively. The thin lm absorption spectra of the para-substituted polymers are comparable this time. These phenomena are related to the different aggregation and p-p stacking characteristics of the polymer backbones caused by the side chain and uorination. Similar absorption behaviors were also observed in our previous comparison between meta-and para-substituted IDT polymers. 40 The absorption edges of the lm spectra are located around 700 nm. Therefore, the optical band gaps extracted from absorption band edges are similar at around 1.78 eV (Table 1).

Electrochemical properties
The energy levels and band gaps of polymers are the key determinants of their photovoltaic performance, which can be estimated from their corresponding redox curves from electrochemical measurements. As shown in Fig. 2, square-wave voltammetry (SWV) was used to determine the oxidation (4 ox ) and reduction (4 red ) potentials of the four polymers. HOMO and LUMO levels were estimated from the peak potentials by setting the oxidative peak potential of Fc/Fc + vs. the normal hydrogen electrode (NHE) to 0.63 V, 48 and the NHE vs. the vacuum level to 4.5 V. 49 HOMO and LUMO levels were calculated according to the formula HOMO ¼ À(E ox + 5.13) eV and LUMO ¼ À(E red + 5.13) eV, where E ox and E red were determined from the oxidation and reduction peaks, respectively. 50 In comparison with the HOMO levels of IDT-based copolymers, those of IDTT-based copolymers are slightly higher in HOMO level due to the more electron-donating TT units on the backbone. The HOMO levels of polymers PIDTT-Q-p and PIDTT-Q-m are À5.64 eV and À5.68 eV, respectively. As expected, the two uorinated polymers PIDTT-QF-p and PIDTT-QF-m display low-lying HOMO levels of À5.76 eV and À5.82 eV, respectively, owing to the electronwithdrawing effect of the uorine atoms. 40,42,47 Compared to the corresponding para-substituted counterparts, the metasubstituted polymers PIDTT-Q-m and PIDTT-QF-m feature a little deeper HOMO levels, which can be attributed to the weaker electron-donating effect from the meta-alkyl-phenyl rings. 51 Since the V oc of BHJ PSCs is positively correlated to the energy difference between the HOMO level of electron donor and the LUMO level of electron acceptor, a low-lying HOMO level is a prerequisite for achieving a high V oc . The LUMO levels of the four polymers are À3.65 eV, À3.58 eV, À3.63 eV and À3.67 eV, respectively. The energy difference between the LUMO levels of the four polymers and that of PC 71 BM (À4.3 eV) are large enough for efficient exciton dissociation. According to the equation, E ec g ¼ e(E ox À E red ) (eV), the electrochemical band   This journal is © The Royal Society of Chemistry 2014 gaps of the four polymers are around 2.0 eV. The uorinated polymers PIDTT-QF-p and PIDTT-QF-m exhibit slightly boarder band gaps, possibly due to the fact that they are less planar and lower molecular weights.

Photovoltaic properties
To investigate the photovoltaic properties of the four polymers, bulk-heterojunction polymer solar cells (BHJ PSCs) with a device conguration of ITO/PEDOT:PSS/polymer:PC 71 BM/LiF/Al were fabricated. Phenyl-C 71 -butyric acid methyl ester (PC 71 BM) was used as the electron acceptor due to its good absorption properties across the visible spectrum. 52,53 The measurements of photovoltaic performances were carried out under an illumination of AM 1.5G simulated solar light at 100 mW cm À2 . The optimized results were obtained by varying polymer:PC 71 BM weight ratios, active layer thicknesses, post-annealing conditions and additives. The corresponding PSC parameters (shortcircuit current density J sc , open circuit voltage V oc , and ll factor FF) are summarized in Table 2. The J-V curves are shown in Fig. 3(a), for PIDTT-Q-m:PC 71 BM and PIDTT-Q-p:PC 71 BM based devices, with the corresponding V oc being 0.81 V and 0.83 V, respectively. As anticipated, both devices based on uorinated copolymers feature a higher V oc of 0.95 V and 0.92 V, respectively, which agrees with their deeper HOMO levels. The metasubstituted uorinated polymer PIDTT-QF-m exhibits a slightly higher V oc compared to the para-substituted analogue PIDTT-QF-p, which is also observed in our previous study of IDT based copolymers. On the other hand, the non-uorinated polymer PIDTT-Q-p and PIDTT-Q-m based devices display inverse results with regards to their corresponding V oc . In this case, one possible reason may come from the minor Fermi level shi as a result of the formation of polymer and PCBM aggregates, which can be affected by the domain sizes of D-A components. 54 Without any post-treatment, the PSC using the meta-substituted polymer PIDTT-Q-m display a superior J sc of 11.8 mA cm À2 in comparison with other three copolymers, which results in a PCE of 6.7%. Since these four copolymers show comparable absorption spectra in thin lms, the enhanced J sc of the PIDTT-Q-m based devices could be ascribed to its higher molecular weight and more favorable nanostructure of its D-A components according to subsequent morphology study. Using 2.5% of DIO as an additives, a slightly higher PCE of 6.8% is recorded, which is one of only a few high-performance results among IDTT based copolymers. This result demonstrates that our side chain isomerization strategy for enhanced photovoltaic performance can be extended from previously studied IDT copolymers to IDTT copolymers. As shown in Fig. 3(b), external quantum efficiencies (EQE) were measured to evaluate the photoresponse of the PSCs. The enhanced quantum efficiency in the region of 400-500 nm is attributable to the absorption of PC 71 BM in the visible region. 52 The photocurrents calculated via integrating the EQE with an AM 1.5G reference spectrum are listed in Table 2, which agrees  well with the corresponding J sc obtained from the J-V measurements. Among the four copolymers, the devices based on PIDTT-Q-m show higher photo conversion efficiencies over the whole visible region, implying more efficient charge collection and less charge recombination at the junction between the polymer PIDTT-Q-m and PC 71 BM. In addition, the corresponding PIDTT-Q-m device with 2.5% DIO demonstrates a slightly higher EQE, which is consistent with the J-V results.

Film morphology
To further understand the reasons for different photovoltaic performances of the four polymers, the morphology of the active layers were studied by atomic force microscopy (AFM) and transmission electron microscopy (TEM). AFM measurements were carried out to study the surface morphology of the blend layers. As shown in Fig. 4 (AFM), AFM images of the uorinated copolymers reveal very large polymer domains with a root mean square (RMS) roughness value of 6.9 nm for PIDTT-QFp:PC 71 BM and 10.1 nm for PIDTT-QF-m:PC 71 BM, respectively. Although the domain size of the non-uorinated PIDTT-Qp:PC 71 BM blend layer decreases, it also shows a rough surface with RMS roughness of 3.8 nm. For these three copolymers, we propose that the poor miscibility of the D-A components results in a non-optimal nanostructure, which in turn limits the photocurrent of the corresponding devices. The metasubstituted non-uorinated polymer PIDTT-Q-m:PC 71 BM blend layer however forms a continuous, ne phase-segregated morphology of its D-A components with a RMS roughness of 1.2 nm, which depicts a uniform and smooth surface. Aer mixing 2.5% DIO in the blend, the RMS roughness slightly increases to 1.9 nm, since the additive enables the development of short bril nanostructures with favorable grain boundaries. 55 To probe the morphology throughout the active layers, TEM was employed to investigate the real-space images in the polymer-fullerene blends. As shown in Fig. 4 (TEM), the PIDTT-Q-p/ PC 71 BM blend layer show large polymer fabric. Both of the PIDTT-QF-p/PC 71 BM and PIDTT-QF-m/PC 71 BM blend layers depict large phase separation, wherein 50-200 nm dark clusters are formed. Since PC 71 BM has a higher scattering density than the polymer, these dark clusters are ascribed to the aggregation of PC 71 BM. 56 For these three polymers, the dimensions of the phase separation and the discontinuous networks are much larger than the typical exciton diffusion length (10 nm), and thus the photogenerated excitons may recombine more easily during exciton diffusion. This may result in poor exciton separation, a low J sc and thereby limit the corresponding device performance. The PIDTT-QF-m/PC 71 BM blend layer has even bigger domain size compared to the PIDTT-QF-p/PC 71 BM blend, which is consistent with its lower PCE than the PIDTT-QF-p based device. On the contrary, a signicant reduction of the phase separation is observed in the PIDTT-Q-m:PC 71 BM blends. Both of the blend lms with and without DIO form continuous and tiny PC 71 BM aggregates. On the basis of the observations from AFM and TEM images, it is evident that the polymer PIDTT-Q-m/PC 71 BM has the most favorable morphology among these four blend layers. Continuous pathways have been formed properly, which subsequently enhance exciton diffusion as well as charge separation. As a result, more efficient exciton diffusion and dissociation in D-A phases correlate well with the superior J sc and PCE obtained for the PIDTT-Q-m based devices. Similar morphological properties were observed in our previous study of meta-substituted IDT-based copolymers. The different molecular conformation and solid state order of the IDTT copolymers may affect the lm morphology. From the previous density functional theory (DFT) calculations of the IDT copolymers, the meta-substituted side chains tend to wrap around the conjugated backbone, while the para-substituted side chains extend from the conjugated backbone in all directions. This extended side-chain conformation raises steric hindrance between adjacent chains, and thus a larger distance between polymer backbones were recorded via the grazing-incidence wide-angle X-ray scattering (GIWAXS). 40 Fig. 4 AFM topography (5 Â 5 mm 2 ) and TEM bright field images of optimized IDTT copolymer:PC 71 BM blended layers.

Conclusion
In summary, two pairs of IDTT-quinoxaline based copolymers with para-hexyl-phenyl or meta-hexyl-phenyl side chains on the IDTT units were synthesized and characterized to understand the effect of side chain isomerization. As anticipated, the BHJ PSCs based on the uorinated polymers PIDTT-QF-p and PIDTT-QF-m offer a high V oc of 0.92 V and 0.95 V, respectively. The meta-substituted polymers PIDTT-Q-m and PIDTT-QF-m offer better solubility and higher molecular weights. Although positioning the alkyl side chain in either the para-or metaposition of the phenyl ring has little inuence on the absorption and energy levels as well as the band gaps of corresponding copolymers, it plays an important role in forming a more appropriate and ne-grained nanostructure in the D-A blend. As a result, the PIDTT-Q-m:PC 71 BM device attains a superior photocurrent and a PCE as high as 6.8%. This result is among the highest efficiency achieved for IDTT copolymers used in conversional BHJ PSCs. It is also a comparatively high value for board band-gap polymers, with band gaps around 1.8 eV, which enables the PIDTT-Q-m polymer to be an appealing candidate for the front cell of tandem devices. Gratifyingly, we demonstrate that the meta-alkyl-phenyl substituted IDTT moiety is a promising building block for efficient organic photovoltaic materials. In our forthcoming work, through further structural optimization of the electron-withdrawing moieties and pendent side groups, it is entirely feasible to synthesize higher-performing IDTT-based conjugated molecules and polymers. Moreover, the here discussed structure-property correlations attest and extend our side-chain design strategy to IDTT-based copolymers, which is expected to be quite valuable for enhancing the photovoltaic performance of conjugated polymers.