Impact of dithienyl or thienothiophene units on the optoelectronic and photovoltaic properties of benzo[1,2,5]thiadiazole based donor–acceptor copolymers for organic solar cell devices

Department of Physics & Astronomy, U Hounseld Road, Sheffield, S3 7RH, UK. +44 (0)114 222 3555; Tel: +44 (0)114 222 3 Department of Chemistry, University of S a.iraqi@sheffield.ac.uk; Fax: +44 (0)114 22 Centre for Plastic Electronics and Departme South Kensington Campus, Exhibition Road Ossila Ltd., Kroto Innovation Centre, Sheffi † Electronic supplementary information ( data, GIWAXS measurements of pure po surface, PL spectra of neat and blend thi characteristics. See DOI: 10.1039/c4ra0718 ‡ Present address: Cavendish Laboratory 0HE, United Kingdom. § Present address: Ossila Ltd., Kroto Inno Cite this: RSC Adv., 2014, 4, 43142


Introduction
Solution-processable photovoltaic devices are a promising technology for low-cost solar energy conversion, with organic photovoltaic devices (OPVs) attracting signicant attention from a wide range of traditional research elds over the past decade. [1][2][3][4] OPV devices that employ blends of conjugated polymers and fullerene derivatives in the photoactive layer in a bulkheterojunction (BHJ) architecture have demonstrated a steady rise in efficiency and lifetime in recent years. 1,5 This rise has, in part, been due to the large number of novel electron donating semiconductors that have been synthesised. In particular, donor-acceptor copolymers with alternating electron-rich and electron-decient heterocycles along the conjugated backbone have proven to be promising semiconductor materials that have low optical band-gaps, and can thus efficiently harvest a relatively large portion of the visible and near-IR regions of the solar spectrum. [6][7][8] Benzothiadiazole (BT) is one example of an electron-decient moiety that can be included in a donor-acceptor copolymer. Many examples exist where this unit is anked by thiophene moieties and combined with a suitable electron donor such as 2,7-carbazole, 9 uorene, 10 cyclopentadithiophene 11 and benzodithiophene. 12,13 When blended with fullerene derivatives such as PCBM or PC 71 BM in an OPV device, power conversion efficiencies of up to 8% have been achieved under optimised processing conditions. 13 One strategy to further improve the light-harvesting efficiency of BT containing copolymers is to reduce their optical band-gaps by altering adjacent units with other heterocycles such as selenophene. [14][15][16][17] Fused heterocycles may also be used to modify the optical bandgap of the copolymer; an approach that forms the basis of this work. The synthesis of PFDT2BT-8 and PCDT2BT-8 have been reported previously, 18,19 with devices incorporating PFDT2BT-8 exhibiting maximum PCEs above 6% aer optimisation of the device architecture. 18 The copolymer PCDT2BT-8 is similar in structure to the relatively well-studied copolymer PCDTBT. 9,[20][21][22][23] Here, addition of octyloxy groups to the BT unit has been shown to increase the solubility of PCDT2BT-8 in common processing solvents relative to PCDTBT. Furthermore, a reduction in the HOMO level of PCDT2BT-8 relative to PCDTBT has been shown to afford an increase in device open-circuit voltage. 19 Substitution of the two thiophene units on PCDT2BT-8 and PFDT2BT-8 with thienothiophene moieties provides a further opportunity to modify the structural and optoelectronic properties of the copolymers. Synthesis of these structures affords two new polymers, namely PCDTTBT-8 and PFDTTBT-8. In this work a detailed physical study is undertaken to explore the optical and electronic properties of these four copolymers in order to understand their performance in OPV devices.
All polymers were puried before use, and separated via Soxhlet extraction methods, with reaction yields above 80%. Gel permeation chromatography measurements from the polymerization reactions (using polystyrene standards) have shown weight average molecular weight values M w ranging from 48 000 to 91 600 Da (Table 1) with polydispersity index values ranging from 1.47 to 2.98. All polymers displayed promising solubility in chlorinated solvents such as chloroform (CF), chlorobenzene (CB) and 1,2-dichlorobenzene (DCB).

Optical properties
Normalised UV-VIS absorption spectra of each copolymer in solution and thin-lm are presented in Fig. 2. It can be seen that the structure of the absorption spectra for each copolymer are qualitatively similar, with the onset of absorption for thin-lms being around 1.9 eV. These absorption spectra are in general characterised by two broad bands in the wavelength range 520-580 nm and 400-430 nm. At the wavelength of peak absorption, the molar extinction coefficients (3) of all polymers exceed 58 000 L mol À1 cm À1 , with the most absorptive polymer being PCDTTBT-8 which has 3 ¼ 67 200 L mol À1 cm À1 .
We nd a relative red-shi in optical absorption is observed for each copolymer when it is cast from solution into a thin-lm. Such effects are likely attributed to an increased aggregation of polymer chains. Comparison of the absorption spectra from thienothiophene-based copolymers show that for the 2,7carbazole derivatives, a relative red-shi in the onset of absorption is observed for PCDTTBT-8 compared to the polymer PCDT2BT-8. This difference is in qualitative agreement with several reports of band-gap reduction for thienothiophene containing copolymers relative to their thiophene counterparts. [24][25][26] However, no signicant difference between the optical band-gap is observed for the uorene-based copolymers PFDT2BT-8 and PFDTTBT-8. Note here that the absorption band located at approximately 420 nm for PFDT2BT-8 is apparently blue-shied to 400 nm in the polymer PFDTTBT-8. It is possible that this change originates from variations in the degree of electronic delocalization along the conjugated backbone of each copolymer, or differences in molecular weight.

Electrochemical properties
Cyclic-voltammetry (CV) measurements were conducted on dropcast polymer lms in acetonitrile with tetrabutylammonium   perchlorate as an electrolyte (Fig. 3). The redox potentials of the different polymers are presented in Table 1 along with the values of the HOMO and LUMO levels for each copolymer in this study as determined from their onset of oxidation and reduction respectively. It can be seen that all copolymers have very similar HOMO levels which are located between À5.2 eV and À5.3 eV relative to vacuum, while relatively larger variations in the positions of their LUMO levels (located between À3.5 eV and À3.2 eV) were observed.

Photovoltaic, structural & charge transport properties
Solar cell devices in a standard architecture were fabricated to determine the photovoltaic efficiency of each copolymer when blended with the fullerene derivative PC 71 BM. The device structure in each case was as follows: ITO/PEDOT:PSS (30 nm)/ active layer (50-100 nm)/Ca (5 nm)/Al (100 nm). To optimise the morphology of the photoactive semiconductor blend layer for photovoltaic performance, we have explored the inuence of the casting solvent, blend composition and the application of post-lm deposition annealing treatments. In Fig. 4 and Table 2, the photovoltaic characteristics and efficiency parameters derived from the current-voltage measurements are presented for solar cell devices aer active-layer optimisation. Device efficiency metrics for other solar cells prepared during the optimisation process, including the inuence of active layer blend ratio, are presented in the ESI. † As shown in Fig. 4, all optimised devices exhibit relatively efficient photovoltaic behaviour. We nd that the optimum weight composition of each photoactive layer is 1 : 4 wt% copolymer:PC 71 BM; a composition comparable to that used for other PCDTBT-derivatives. 14,18,19 The thickness of the photoactive semiconducting layer in each optimised device was found to be approximately 60 nm. Comparing the two carbazole containing copolymers PCDT2BT-8 and PCDTTBT-8, substitution of the thiophene heterocycles with thienothiophene permits the fabrication of photovoltaic devices having a PCE of 4%. For the best performing devices however, OPV devices utilising PCDT2BT-8 reach a maximum PCE of 4.1% whereas for OPV devices utilising PCDTTBT-8 a slightly higher PCE of 4.5% is determined. From Table 2 it is apparent that the improvement in PCE of the best PCDTTBT-8 devices relative to the PCDT2BT-8   devices results from increased short-circuit current density (J SC ) (9.8 mA cm À2 and 8.7 mA cm À2 respectively). This can be correlated with the relatively lower optical bandgap and higher molar extinction coefficient of PCDTTBT-8. Furthermore, comparison of the photovoltaic performance of PCDTTBT-8 (PCE of up to 4.5%) to a similar carbazole-based polymer described in the literature (PCTTBTC12) (PCE of 3.4%), 27 with dodecyloxy-substituents rather than octyloxysubstituents, also indicates a superior performance of PCDTTBT-8.
Comparing the photovoltaic efficiency of PFDT2BT-8:PC 71 BM blends with PFDTTBT-8:PC 71 BM blends, it can be seen that a relatively lower average PCE is obtained for the PFDTTBT-8 based OPV devices (4.1% against 5.8% for PFDT2BT-8:PC 71 BM). This difference arises from a relatively lower FF and J SC for the PFDTTBT-8 devices as the open-circuit voltage V OC in each system is similar (0.93 compared with 0.91 V for PFDT2BT-8:PC 71 BM). For OPV devices incorporating PFDT2BT-8 a maximum PCE of 6.2% is determined; the highest value in the material set studied.
A morphological study was undertaken to characterise the nanostructure of the blend lms. Specically, grazing-incidence wide-angle X-ray scattering (GIWAXS) was employed to characterise the degree of nanoscale crystallinity in the blend lms. These measurements were complimented by atomic force microscopy (AFM) imaging of the sample surface for each copolymer:PC 71 BM blend lm (see ESI †), with these scans potentially providing evidence for phase-separation at lengthscales greater than can be probed via GIWAXS. In Fig. 5, out-ofplane GIWAXS patterns are presented for each copolymer:PC 71 BM blend thin-lm having an identical lm composition to that of the best performing OPV devices (1 : 4 copolymer:PC71BM wt%). 2D GIWAXS images are provided in the ESI. † From these measurements it can be seen that X-ray scatter from each blend thin-lm is dominated by PC 71 BM, as evidenced by the relatively broad ring at q ¼ 1.34Å À1 which indicates an isotropic distribution of small fullerene clusters within each blend lm. 28 We note that complimentary measurements on as-cast pure copolymer thin-lms only evidence weak X-ray scatter in the out-of-plane direction (see ESI †) with signals attributed to p-p stacking of adjacent copolymers chains at q z ¼ 1.72Å À1 (d ¼ 3.65Å) for PFDT2BT-8 and q z ¼ 1.62Å À1 (d ¼ 3.88Å) for the remaining three copolymers.
Our GIWAXS measurements suggest that the blends studied here are characterised by a nely mixed distribution of copolymer and PC 71 BM domains with little long-range order present. This conclusion is supported by AFM scans of each blend thin-lm surface that are apparently smooth (RMS roughness values below 0.5 nm). Signicantly however, we nd that the intensity of the broad ring at q ¼ 1.34Å À1 (associated with PC 71 BM aggregation) is largest in the uorene copolymer PFDT2BT-8:PC 71 BM blend and is greater than that of X-ray scatter from PCDT2BT-8:PC 71 BM and the two fused-thiophene polymer blends PCDTTBT-8:PC 71 BM and PFDTTBT-8:PC 71 BM. This suggests that the degree of PC 71 BM crystallisation in the polymer blends is different. However, as the linewidth of the scatter peak at q ¼ 1.34Å À1 is similar in all blends, we conclude that the average size of the PC 71 BM aggregates is constant, although the fraction of molecules that nd themselves within such an aggregate is different. We propose that enhanced PC 71 BM aggregation in PFDT2BT-8:PC 71 BM facilitates the extraction of electrons through the creation of percolation pathways within the BHJ, hence suppressing non-geminate recombination. 29 At present, the origin of the enhanced PC 71 BM aggregation in PFDT2BT-8:PC 71 BM blends is not understood.
To determine whether any optical or electronic differences exist between each photovoltaic blend in order to rationalise the solar cell characteristics presented earlier, we have performed time-resolved optical spectroscopy to probe distinct stages in the photocurrent generation process. Specically, the relative yield of free polarons in the copolymers on the ms timescale was probed using transient absorption spectroscopy (TAS). To compliment these studies, organic eld-effect transistor (OFET) devices were fabricated to determine the charge-carrier mobility in each copolymer.  5 Out-of-plane GIWAXS profiles of X-ray scatter from 1 : 4 wt% copolymer:PC 71 BM blend thin-films. Here, the region of zero scattering intensity at low q corresponds to the position of the beam-stop.
To probe the relative yield of charge carriers in the blend, we have performed TAS measurements to monitor the dynamics of hole polarons generated aer photoexcitation of the copolymer and subsequent excited state dissociation at a copolymer:PC 71 BM interface. Note that steady-state photoluminescence (PL) measurements of each copolymer:fullerene blend lm indicate efficient (>99%) quenching of photoexcited states (see ESI †). Here, a lower blend concentration of PC 71 BM (20 wt%) than used in optimised devices was used to slow the recombination of dissociated polarons, as the decay dynamics occurred within a timescale commensurate with the response time of the instrument in blends with compositions corresponding to optimised devices. This data is presented in Fig. 6.
As shown, over the time-range 0.2-2 ms the transient absorption dynamics at 960 nm (corresponding to the peak absorption wavelength by the hole polarons) 30 from each blend thin-lm exhibit power-law decays, indicative of non-germinate recombination processes of dissociated polarons. 31,32 As shown, a similar DOD is determined for each copolymer:PC 71 BM blend. Previous studies have shown that the magnitude of DOD provides a reasonable indicator of the quantum yield of dissociated charges, with further reports of this signal correlating with photocurrent generation efficiency in a fabricated OPV device. 33 Here however, we do not nd sufficient evidence to suggest that variations in polaron yields between each copolymer:PC 71 BM sample correlate with OPV device efficiency as the different polymers display almost identical charge generation and recombination characteristics. For example, the blend system PFDT2BT-8:PC 71 BM exhibits the most efficient photovoltaic behaviour, however it displays similar DOD transient kinetics to all other blend lms studied.
Bottom-gate architecture OFET devices were fabricated to characterise the charge-carrier mobility of each of the copolymers. Transfer and output characteristics for typical copolymer OFETs are presented in the ESI † with extracted charge carrier mobilities detailed in Table 3.
Here it can be seen that the copolymer PFDT2BT-8 exhibits a higher charge carrier mobility than the thienothiophene derivative PFDTTBT-8 (values of 4.3 Â 10 À3 cm 2 V À1 s À1 and 7 Â 10 À4 cm 2 V À1 s À1 respectively). A relatively high on/off ratio of 10 4 is also determined for PFDT2BT-8. Furthermore, for the carbazole containing copolymers, mobility values of 2.1 Â 10 À3 cm 2 V À1 s À1 and 2.4 Â 10 À3 cm 2 V À1 s À1 are obtained for the PCDT2BT-8 and PCDTTBT-8 copolymers respectively. We therefore correlate the best performing solar cell devices that incorporate PFDT2BT-8 with the superior charge carrier mobility of this copolymer compared to the others studied. This characteristic also correlates well with the relatively higher device FF that suggests low charge recombination within the semiconductor blend layer. For the other copolymer blends studied, the relatively lower PCEs of solar cell devices are not found to be limited by the yield of hole polarons. Note that with the exception of PCDT2BT-8:PC 71 BM devices, all solar cells exhibit comparable photocurrent densities at relatively large reverse bias. This suggests that higher device efficiencies may yet be obtainable in these systems aer optimising other properties of the copolymer, for example its molecular weight or purity. 34,35 Experimental Materials PCDT2BT-8, 19 PFDT2BT-8, 18 9-(heptadecan-9-yl)-2,7-bis-(4,4,5,5tetramethyl-[1,3,2]dioxaborolan-2-yl)-9H-carbazole 9 and 4,7bis(5-bromothieno[3,2-b]thiophen-2-yl)-5,6-bis(octyloxy)-benzo-[c][1,2,5]-thiadiazole 36 were prepared according to literature procedures. 9,9-Dioctyluorene-2,7-diboronic acid bis(1,3-propanediol) ester was obtained commercially. Toluene was dried and distilled over sodium under an inert argon atmosphere. Acetonitrile (high performance liquid chromatography (HPLC) grade) was dried and distilled over phosphorus pentoxide under an inert argon atmosphere, then stored over molecular sieves (3Å).  To a ask containing 9,9-dioctyluorene-2,7-diboronic acid bis(1,3-propanediol) ester (0.5403 g, 0.9676 mmol), and 4,7-bis-(5-bromo-thieno[3,2-b]thiophen-2-yl)-5,6-bis-octyloxy-benzo-[1,2,5]thiadiazole (0.800 g, 0.9676 mmol) was added toluene (21 cm 3 ). The solution was degassed then a degassed solution of tetraethylammonium hydroxide (7.0 cm 3 , of a 20 wt% aqueous solution, 9.48 mmol) was added. Pd(OAc) 2 (8.0 mg, 0.0359 mmol) and tri(o-tolyl)phosphine (21.7 mg, 0.0713 mmol) were then added and the mixture was degassed and heated to 95 C for 48 hours. The mixture was cooled to room temperature and bromobenzene (0.1 cm 3 ) was added. The mixture was then degassed and heated to 90 C for 1 hour. The mixture was cooled to room temperature and phenylboronic acid (150 mg) was added. The mixture was then degassed and heated to 90 C for 3 hours. Aer cooling to room temperature, the mixture was dissolved in CHCl 3 (300 cm 3 ) and to this solution was added an ammonium hydroxide solution (28% in H 2 O, 200 cm 3 ). The mixture was heated to reux for 3 hours, cooled to room temperature and the organic phase was separated. Ethylenediaminetetraacetic acid disodium salt dihydrate (0.300 g, 0.80 mmol) was added to the organic phase and was le to stir overnight. The mixture was then washed with distilled water (2 Â 200 cm 3 ), the organic phase was separated and concentrated to about 50 cm 3 and poured into degassed methanol-water mixture (10 : 1, v/v, 500 cm 3 ). The resulting mixture was then stirred overnight and the resulting precipitate collected by ltration. The collected solid was cleaned using Soxhlet extraction with solvents in order methanol, acetone, hexane and toluene. The toluene fraction was concentrated to about 50 cm 3 and then poured into degassed methanol (500 cm 3 ). The resulting mixture was stirred overnight and the solid collected by ltration to afford the polymer as a red powder (0.82 g, yield 80%

Measurements
Polymer characterisation. Nuclear magnetic resonance (NMR) spectra were recorded on Bruker AMX400 400 MHz NMR spectrometer at 100 C in C 2 D 2 Cl 4 solutions with TMS as the internal standard. Polymer solutions in 1,2,4-trichlorobenzene at 100 C were used as samples for GPC analysis. The GPC curves were obtained by the RI-detection method, which was calibrated with a series of polystyrene narrow standards (Polymer Laboratories). Elemental analyses were carried out by the Perkin Elmer 2400 CHN Elemental Analyser for CHN analysis and by the Schöniger oxygen ask combustion method for anion analysis. UV-visible absorption spectra were measured by Hitachi U-2010 Double Beam UV/Visible Spectrophotometer. The absorbance of polymers was measured in solution in chloroform (spectrophotometric grade) at ambient temperature using rectangular quartz cuvettes (light path length ¼ 10 mm) purchased from Sigma-Aldrich. Samples of pristine polymer thin lms for UV-visible absorption spectra measurements were prepared by dip coating quartz plates into 1 mg cm À3 polymer solutions in chloroform (HPLC grade), with measurements carried out at ambient temperature. Cyclic voltammograms were recorded using a Princeton Applied Research Model 263A Potentiostat/Galvanostat. Measurements were carried out under argon at (25 AE 2) C. 10 ml of tetrabutylammonium hexa-uorophosphate solution in acetonitrile (0.1 mol dm À3 ) was used as the electrolyte solution. A three electrode system was used consisting of an Ag/Ag + reference electrode (silver wire in 0.01 mol dm À3 silver nitrate solution in the electrolyte solution), a platinum working electrode (2 mm-diameter smooth platinum disc, area ¼ 3.14 Â 10 À2 cm 2 ), and a platinum counter electrode (platinum wire). Polymer thin lms were formed by drop casting 1.0 mm 3 of polymer solutions in chloroform (HPLC grade) (1 mg cm À3 ) onto the working electrode, then dried in air. Ferrocene was employed as a reference redox system according to IUPAC's recommendation. 37 OPV device fabrication. Pre-patterned glass-ITO substrates (20 U , À1 ) were purchased from Ossila Limited. The substrates were cleaned by sonication in isopropyl alcohol and deionized water and dried with compressed nitrogen. PEDOT:PSS (HC Stark Clevios P VP AI4083, ltered with 0.45 mm PVDF) was spin cast on top of the cleaned substrates to form a 30 nm thick layer in air which was then annealed at 150 C prior to transferring to a nitrogen lled glovebox. A subsequent thermal anneal at 150 C was performed inside the glovebox for 1 h. The active layer (blend of copolymer:PC 71 BM) was dynamically deposited via spin casting on the anode buffer layer from a range of solvents. In all cases, the copolymer was dissolved at 4 mg ml À1 with a total blend concentration of 12-24 mg ml À1 depending on the blend ratio of copolymer to fullerene. The active layer thicknesses were varied by adjusting the spin-coating speed and measured using a surface proler. The cathode (a bilayer of calcium and aluminium -5 nm and 100 nm respectively) was thermally evaporated within a vacuum chamber held at a pressure < 10 À6 mbar. Devices were encapsulated using a UVtreated epoxy (Ossila Limited) before testing. Device performance was tested in air using a Newport 92251A-1000 solar simulator (AM1.5) with NREL certied silicon reference cell at 100 mW cm. À2 Here, an aperture mask (0.025 cm 2 ) was placed over each pixel to accurately dene the device area.
GIWAXS. Wide-angle X-ray diffraction patterns were obtained for each copolymer:PC 71 BM blend thin-lm in a grazing-incidence geometry at the I07 beamline at the Diamond Light Source (Didcot, UK). Blend thin-lms were deposited onto silicon/native oxide substrates for measurement. Samples were measured within a custom-built cell containing a slight overpressure of helium to minimise background X-ray scatter. For measurement, an 8 keV X-ray beam was incident on the sample surface at a grazing-incidence angle of 0.2 . Data was collected using a Pilatus 2M detector and analysed using the DAWN soware package (http://www.dawnsci.org). Silver Behanate powder was used as a calibrant.
Transient absorption spectroscopy. Micro-to milli-second transient absorption spectroscopy was performed on lms in a N 2 environment and all data shown is scaled for the fraction of photons absorbed at the excitation wavelength. The samples were excited by a dye laser (Photon Technology International Inc. GL-301) pumped by a nitrogen laser (Photon Technology International Inc. GL-3300) to give a pulse width of 0.6 ns at 4 Hz. Excitation was at 567 nm at an energy of 14 AE 2 mJ cm À2 . The samples were probed using a halogen lamp (Bentham, IL1) with a stabilised power supply (Bentham, 605). The probe light was detected using a silicon photodiode and the signal subsequently amplied and passed through electronic band-pass lters to improve the signal to noise ratio. The temporal resolution of the system was around 0.6 ms.
OFET device fabrication & characterisation. Silicon oxide substrates were purchased from Ossila Limited. These were cleaned in sequence by sonication in 1% Hellmanex III solution, hot DI water, isopropyl alcohol, cold DI water and compressed nitrogen. Gold electrodes (50 nm of Au on 1 nm of Cr as adhesion layer) were then thermally evaporated under vacuum onto the silicon oxide to dene ve transistor channels with width and length equal to 103 mm and 30 mm, respectively. The substrates were then cleaned with oxygen plasma for 30 seconds at P ¼ 100 W, treated in a 5 mMol solution of PFBT (pentauorobenzenethiol, Sigma-Aldrich), dried with compressed nitrogen and transferred to a dry glove box atmosphere. Substrates were then treated with HMDS (hexamethyldisilazane, Sigma-Aldrich) to prepare a hydrophobic surface and to minimise surface trap states prior to deposition of the organic semiconductors. The four copolymer solutions were prepared at a concentration of 2 mg ml À1 in chloroform and before use were heated at 45 C for two hours and ltered with a 0.45 mm PTFE lter. For semiconductor deposition, 25 ml of organic semiconductor solution was spin coated at 1000 rpm for 60 seconds, followed by a drying step at 2000 rpm from 5 seconds. The samples were subsequently le for twenty minutes under mild vacuum to facilitate complete evaporation of casting solvent. No further annealing treatments were applied. Complete OFETs were characterised using a PXI modular source measurement system (National Instruments PXI-4132) integrated with the FACT measurement system developed by Ossila. The gate electrode is provided by the boron-doped silicon wafer (r < 0.002 U cm), while the silicon oxide acts as gate dielectric with a capacitance per unit area of C Ox ¼ 1.09 Â 10 À8 F cm À2 . The electrical connections with the SMU are provided by thermally evaporated Al/Cr (80/5 nm) contact pads. For each copolymer studied, 15 bottom gate/bottom contact OFETs were fabricated.

Conclusions
In this study two new medium band-gap donor-acceptor copolymers comprising 2,7-carbazole or 2,7-uorene units positioned between thienothiophene and benzothiadiazole units have been synthesised. A photovoltaic study of copolymer:PC 71 BM blend thin-lms of these two new copolymers as well as those of analogous polymers where the thienothiophene units are replaced with dithienyl units has been conducted with the devices fabricated having power conversion efficiencies ranging between 4 and 6% and open-circuit voltages of approximately 0.9 V. A detailed study employing a range of complimentary techniques probing the structural, optical and electronic properties of the materials have been used to explore the underlying physical differences between the copolymers and provide insight into the device PCEs. We nd that thienothiophene substitution is an effective method to modify the optical absorption properties of the copolymers. Optical probes show no discernable differences in the photoexcited state quenching efficiency or charge generation yields between each copolymer:PC 71 BM blend. Probes of lm structure using X-ray scattering reveal that whilst the polymer component is largely amorphous in all cases, the degree of PC 71 BM crystallinity varies between the different polymer-blends studied. In photovoltaic devices, we nd that thienothiophene substitution does not result in improved device efficiency. However we nd that improved device efficiency (particularly deriving from improved ll-factors) is positively correlated with polymer blends in which there is higher hole mobility and more extensive PC 71 BM crystallisation. These results highlight the complexity in semiconductor design for OPV applications and suggest possible routes for polymer optimisation (for example M w ). For devices incorporating the copolymer PFDT2BT-8, relatively high photovoltaic efficiencies (maximum of 6.2%) have been obtained. This result suggests PFDT2BT-8 holds promise as a suitable semiconductor for single junction organic photovoltaic devices, or as a wide band-gap semiconductor for tandem solar cell devices.