Influence of the Electron Deficient Co‐Monomer on the Optoelectronic Properties and Photovoltaic Performance of Dithienogermole‐based Co‐Polymers

A series of donor–acceptor (D–A) conjugated polymers utilizing 4,4‐bis(2‐ethylhexyl)‐4H‐germolo[3,2‐b:4,5‐b′]dithiophene (DTG) as the electron rich unit and three electron withdrawing units of varying strength, namely 2‐octyl‐2H‐benzo[d][1,2,3]triazole (BTz), 5,6‐difluorobenzo[c][1,2,5]thiadiazole (DFBT) and [1,2,5]thiadiazolo[3,4‐c]pyridine (PT) are reported. It is demonstrated how the choice of the acceptor unit (BTz, DFBT, PT) influences the relative positions of the energy levels, the intramolecular transition energy (ICT), the optical band gap (Egopt), and the structural conformation of the DTG‐based co‐polymers. Moreover, the photovoltaic performance of poly[(4,4‐bis(2‐ethylhexyl)‐4H‐germolo[3,2‐b:4,5‐b′]dithiophen‐2‐yl)‐([1,2,5]thiadiazolo[3,4‐c]pyridine)] (PDTG‐PT), poly[(4,4‐bis(2‐ethylhexyl)‐4H‐germolo[3,2‐b:4,5‐b′]dithiophen‐2‐yl)‐(2‐octyl‐2H‐benzo[d][1,2,3]triazole)] (PDTG‐BTz), and poly[(4,4‐bis(2‐ethylhexyl)‐4H‐germolo[3,2‐b:4,5‐b′]dithiophen‐2‐yl)‐(5,6‐difluorobenzo[c][1,2,5]thiadiazole)] (PDTG‐DFBT) is studied in blends with [6,6]‐phenyl‐C70‐butyric acid methyl ester (PC70BM). The highest power conversion efficiency (PCE) is obtained by PDTG‐PT (5.2%) in normal architecture. The PCE of PDTG‐PT is further improved to 6.6% when the device architecture is modified from normal to inverted. Therefore, PDTG‐PT is an ideal candidate for application in tandem solar cells configuration due to its high efficiency at very low band gaps (Egopt = 1.32 eV). Finally, the 6.6% PCE is the highest reported for all the co‐polymers containing bridged bithiophenes with 5‐member fused rings in the central core and possessing an Egopt below 1.4 eV.

59% yield. All polymers were synthesized using microwave assisted Stille cross coupling polymerization reactions as shown in Scheme 1 and purifi ed by sequential Soxhlet extraction with methanol, acetone, and hexane to remove low molecular weight oligomers and catalyst residues. [ 27 ] In the case of PDTG-BTz and PDTG-PT , both polymers were readily soluble in chloroform but the crude polymers were both relatively low molecular weights (M n ≈ 7 kDa). Since it is known that OPV device performance usually improves using higher molecular weight polymers, [28][29][30][31] high molecular weight fractions with low polydispersity were isolated using preparative GPC in chlorobenzene. [ 32 ] For PDTG-DFBT , we found that the crude product was poorly soluble in both chloroform and chlorobenzene, precluding preparative GPC. The reduced solubility of polymer containing the DFBT co-monomer has previously been reported by other groups, and may be related to intra and intermolecular sulfur-fl uorine interactions. [ 33,34 ] Therefore extraction of the crude PDTG -DFBT with hot chlorobenzene resulted in isolation of a soluble fraction of low molecular weight. However, the majority of the yield was insoluble material. Molecular weights were measured by gel permeation chromatography (GPC) against polystyrene standards and are summarized in Table 1 . In the case of the PDTG -DFBT , the poor solubility of the polymer at higher molecular weights resulted in the use of the low molecular weight fraction in this study. The molecular lower device effi ciency. [ 14,15 ] Moreover, co-polymerization of DTG with different electron withdrawing groups, [ 16,17 ] such as 5-octyl-4 H -thieno [3,4c ]pyrrole-4,6(5 H )dione ( TPD ) provide polymers with PCE of 7.4% [ 17 ] with a lower J sc of 13.0 mA cm −2 but with an increased V oc of 0.87 V as compared to PDTG-BT . [ 13 ] Interestingly, the performance of some DTG based-polymers in OPV devices has shown an increase in the PCE compared to that of the analogous carbon or silicon bridged alternatives. [ 18,19 ] This effect has been attributed to the increase in carbon-germanium bond length compared to the carbon-carbon or carbon-silicon bonds in CPDT and DTS respectively. [ 19 ] The increase in bond length has been suggested to result in a reduction of steric hindrance between the solubilizing group on the bridging heteroatom and adjacent co-monomers, which facilitates crystallization and close ππ stacking of the polymers. A similar behavior was earlier reported for polymers containing silicon bridging atoms versus their carbon bridged analogues. [ 11,12 ] The encouraging results reported for DTG co-polymers have prompted this study into alternative electron-defi cient building blocks in DTG based-polymers, with the aim to investigate how the optoelectronic properties and device performance vary as a function of acceptor strength. For comparison purposes, we investigated three structural analogues of BT , namely 5,6-difl uorobenzo[ c ] [1,2,5] thiadiazole ( DFBT ), [1,2,5]thiadiazolo [3,4c ]pyridine ( PT ), and 2-octyl-2 H -benzo[ d ] [1,2,3]triazole ( BTz ). [20][21][22][23] Here, we report the synthesis, characterization and device properties of all three co-polymers ( PDTG-BTz , PDTG-DFBT , PDTG-PT ), and compare their performance to previously reported BT co-polymer. [ 13 ] We observe signifi cant differences in optoelectronic properties and device performance and relate these to the nature of the acceptor co-monomer. [ 19 ] [1,2,3]triazole, [ 24 ] 4,7-Dibromo- [1,2,5] thiadiazolo [3,4-c ] pyridine, [ 21,25 ] and 4,7-dibromo-5,6-difl uorobenzo[ c ][1,2,5] thiadiazole, [ 26 ] were all synthesized according to the previously reported literature. However, purifi cation of DTG-Sn 2 was complicated due to the fact that the monomer was non-crystalline. Attempted purifi cation over silica or alumina resulted in extensive de-stannylation as previously noted. [ 19 ] Therefore, preparative GPC was utilized to give high purity monomer in Scheme 1. Synthetic route to PDTG-BTz , PDTG-DFBT , and (Regiorandom) PDTG-PT . In the inset the chemical structure of PDTG-BT is presented.

. Synthesis
www.afm-journal.de www.MaterialsViews.com wileyonlinelibrary.com Table 1 . All polymers display qualitatively similar spectral shape in the long wavelength region, with pronounced double peaks apparent in both solution and solid state (Figure 1 a). Upon fi lm formation the longer wavelength peak increases in intensity relative to solution and slightly red-shifts indicative of enhanced interchain interactions. The similarity between the solution and thin fi lm absorption spectra at the higher wavelengths suggests that the polymers may be aggregating in solution at room temperature. Indeed as shown in Figure 1 b, heating the solutions results in a signifi cant decrease in intensity of the longer wavelength shoulder in all cases, as well as a slight blue shift of the absorption. Heating most likely increases torsional disorder in the backbone, disrupting the planarity of the polymers and decreasing their tendency to aggregate. Note the effect is less pronounced for PDTG-BTz than for other two co-polymers, in agreement with the solution state 1 H NMR data. Based on this observation, the peaks of PDTG-BTz , PDTG-DFBT , and PDTG-PT at 575 nm, 677 nm, and 773 nm, respectively in solution can be assigned to the intramolecular D-A charge transfer (ICT) interactions, [ 39,40 ] whereas the peaks at 616 nm, 735 nm, and 831 nm are attributed to the aggregated form of the co-polymers. Consequently, the peaks of the alternating dithienogermole-benzothiadiazole co-polymer ( PDTG-BT ), reported in our previous work, [ 13 ] at 700 nm and 756 nm in solution can safely be assigned to the D-A ICT interactions and aggregate form, respectively. Passing from solution to the solid state, a similar behavior is also presented by studying the absorption spectra of PDTG-BTz , PDTG-DFBT and PDTG-PT (Figure 1 a). In more detail, PDTG-BTz shows maxima at 585 nm with a shoulder at 626 nm, the PDTG-DFBT exhibits maxima at 739 nm with a shoulder at 675 nm and PDTG-PT reveals maxima at 848 nm weights of PDTG-PT and PDTG -BTz were quite similar and all three polymers have relatively low polydispersity (PDI) indexes.
The 1 H NMR of all three polymers (see Supporting Information) was consistent with their proposed structures. However, whilst PDTG-BTz exhibited a clear spectrum with sharp peaks, both PDTG-PT and PDTG-DFBT exhibited relatively broad featureless peaks in common with many other low band gap polymers. We attribute this broadening to the segmental aggregation of the polymer in solution and the presence of a variety of chain conformations which cannot freely rotate. The enhanced peak broadening for the lower band gap polymers may be due to the increased double bond character between the co-monomer units due to the lower lying LUMO energy levels and enhanced quinoidal character of the backbone. Enhanced quinoidal character would reduce free rotation in the backbone. In the case of PDTG-PT , we note that the synthesis route used may provide a mixture of regioisomers for the relative orientation of the [1,2,5] thiadiazolo [3,4-c ]pyridine. It has previously been shown that the 2-position of the [1,2,5]thiadiazolo [3,4-c ]pyridine (ortho to the pyridyl nitrogen) undergoes oxidative coupling more rapidly than the 5 position, which may lead to some enhancement in the percentage of regioregularity versus the purely random copolymer. [35][36][37][38] However in our case the broad 1 H NMR spectra prevented further analysis of their regioregularity.

. Optical and Electrochemical Properties
The normalized UV-Vis absorption spectra of co-polymers PDTG-BTz , PDTG-DFBT , and PDTG-PT in chlorobenzene solution and as thin fi lms are presented in Figure 1 , and the corresponding optoelectronic properties are summarized in  In comparing the effect of the germanium bridging group, it is useful to compare the properties of the present polymers to those of the previously reported analogous polymers incorporating bridging C or Si atoms with identical 2-ethylhexyl solubilizing groups. Thus, for example, the poorly soluble DFBT co-polymer with 4 H -cyclopenta [1,2- shows a similar dual peak spectra, with the longer wavelength aggregation peaks absorbing at 712 nm in chloroform solution and 761 nm as a thin fi lm, and with a band gap of 1.6 eV. [ 44 ] The BTz co-polymer with 4 H -silolo[3,2b :4,5b ′]dithiophene ( DTS ) exhibits a less pronounced dual peak absorption and has a maximum absorption of 567 nm in chloroform and 576 nm in thin fi lm, with an optical band gap of 1.8 eV. [ 45 ] Thus, replacement of the bridging Ge for Si appears to have a relatively minor infl uence of the optical properties in agreement with other studies, [ 19,46 ] whereas replacing the Ge with C appears to slightly red-shift the maximum absorption.
The ionization potential of PDTG-BTz , PDTG-DFBT , and PDTG-PT polymer thin fi lms was measured by photo-electron spectroscopy in air (PESA) to be -4.75 eV, -5.0 eV, and -4.9 eV respectively. From the direct measurements of the HOMO, we can approximate the LUMO levels by subtracting the optical band gap, to obtain values of -2.95 eV, -3.6 eV, and -3.6 eV for PDTG-BTz , PDTG-DFBT , and PDTG-PT , respectively as shown in Table 2 . For comparative purposes the ionization of PDTG-BT measured by the same technique is -4.8 eV, with a corresponding LUMO level of -3.3 eV. [ 13 ] From these results, it is evident that the replacement of the BT monomer with DFBT or PT shifts both the HOMO and LUMO levels to lower values due to the stronger electron accepting properties of both DFBT and PT . Comparing the HOMO and LUMO energy levels of PDTG-DFBT and PDTG-PT it can be seen that while the two co-polymers reveal similar LUMO levels, the HOMO of PDTG-DFBT is lying deeper than that of PDTG-PT in accordance with other examples in the literature [ 20,23 ] and verifying the assumption that the fl uorine atoms on the DFBT unit have a signifi cant electron withdrawing effect. Finally, the replacement of the BT with BTz results in an upshifted LUMO level for PDTG-BTz due to the with a shoulder at 784 nm. It should be noted that PDTG-BT display maxima at 778 nm with a shoulder at 714 nm. [ 13 ] Based on the assignment reported previously for the peaks of the co-polymers in solution, the peaks of PDTG-BTz , PDTG-BT , PDTG-DFBT , and PDTG-PT at 585 nm, 714 nm, 675 nm, and 784 nm, respectively are related to the ICT interactions and those at 626 nm, 778 nm, 739 nm, and 848 nm to the aggregated form.
Interestingly, the variations in the ICT transitions of our co-polymers do not strictly follow [ 3 ] the commonly observed trend reported in the literature that the stronger the electron accepting nature and quinoid forming ability of the electron defi cient unit, the stronger the intramolecular charge transfer and lower the E g opt . [ 21,41 ] The energy gap of the intramolecular charge transfer of the DTG-based co-polymers follows the order: PDTG-BTz > PDTG-DFBT > PDTG-BT > PDTG-PT whilst the electron accepting strength of the electron defi cient monomer follows the order: PT > DFBT > BT > BTz . [ 42 ] We rule out any differences in torsional disorder between BT and DFBT being the cause for this increase in band gap, since DFT calculations of the conjugated backbone (vide infra) suggest that the DFBT co-polymer is able to adopt a completely co-planar backbone despite the slightly larger size of fl uorine compared to hydrogen ( Figure 2 ). Similarly, we do not believe the lower molecular weight of PDTG-DFBT is responsible for the increase in band gap, since no differences in the optical absorption spectra of similarly low weight PDTG-BT versus higher molecular weight polymer were apparent. [ 14 ] In addition highly dilute solution spectra of the 'insoluble' higher weight fraction of PDTG-DFBT in tetrachloroethane also show no appreciable differences to the more soluble lower weight fraction, suggesting the effective conjugation length is reached in the soluble fraction. We note others have observed a similar increase in band gap upon the exchange of BT with DFBT in donoracceptor co-polymers, [ 7,26,34,43 ] and in agreement with these   and less intense in PDTG-DFBT but occurs in a similar region whilst it appears to be absent altogether for PDTG-BTz in agreement with its amorphous character.

. Photovoltaic Properties
Initial photovoltaic devices were prepared using a glass|ITO |PEDOT:PSS|Polymer:Fullerene-Blend|LiF|Al device structure, measured under 100 mW cm −2 AM 1.5 illumination and averaged over at least fi ve devices. Varying polymer:fullerene blends of ratios (1:2 wt%) prepared from 1,2-dichlorobenzene were screened. PC 70 BM was used to complement the polymers by capturing the low wavelength region. Their respective J-V curves and device properties can be seen in Figure 4 a and Table 3 . The highest PCE of 4.6% was achieved by PDTG-PT , while the PCE of PDTG-BTz and PDTG-DFBT were 2.4% and 2.1%, respectively. Previously, we have reported PDTG-BT to provide high J sc (18.4 mA cm −2 ) after device optimization but low V oc (0.56 V) and FF (0.42), limiting the overall PCE to 4.5%. [ 13 ] All three polymers investigated here show an increase in the V oc compared to PDTG-BT . For both PDTG-DFBT and PDTG-PT , the increment in V oc is in agreement with the experimentally observed increased ionization potentials relative to PDTG-BT , and can be related to the increase in energy gap between the polymer HOMO and the PC 70 BM LUMO. [49][50][51] Surprisingly, the highest recorded V oc was obtained by PDTG-BTz -based devices despite the fact that both experimental and computational results demonstrated that PDTG-BTz has the smallest ionization potential amongst the studied materials. The analogous Si bridged polymer, PDTS-BTz , also exhibits a similarly high voltage of 0.71 V, but lower overall effi ciencies. [ 45 ] We are unclear of the exact reasons for this, but believe it may be partly due to differences in the molecular packing of the pristine polymer fi lm and the blend fi lm. Although PDTG-BTz appears largely amorphous by WAXS, the optical spectra clearly indicate it has a marked tendency to aggregate in solution and the solid state. Incorporation of the fullerene may disrupt this weaker electron accepting properties of BTz as compared to BT , but the HOMO levels of the co-polymers are almost similar.
Quantum chemical (DFT) calculations were performed to predict the molecular energy levels and model the distribution of the frontier molecular orbitals of the co-polymers ( Figure 2 ). Trimer model compounds were modeled using Gaussian 09 at the B3LYP/6-31G* level. The side chains were modifi ed to methyl groups in order to simplify the calculations. The minimum-energy conformations of the trimers reveal that all have essentially a planar backbone conformation, with little differences in torsional disorder between them. The calculated HOMO-LUMO energy gaps of the trimers show reasonable correlation to the experimentally observed values (Table 2 ), with the same trend between materials observed. Moreover, even though the predicted HOMO levels are somewhat upshifted to that observed experimentally, they follow a broadly similar trend with the exception of PDTG-PT , which is predicted to have a similar HOMO level to PDTG-DFBT . These differences may be due to solid state packing effects for the actual polymer fi lms, which are not taken into account on the DFT calculations. From the molecular orbital diagrams in Figure 2 , it is possible to observe that in all three materials the HOMO is fully delocalized along the polymer chain axis. However, distinct differences are observed in the case of the LUMO, which we relate to the strength of the electron defi cient unit. For example both PDTG-PT and PDTG-DFBT show a signifi cant localization of the LUMO around the electron defi cient acceptor units compared to DTG in the backbone. However, PDTG-BTz shows comparatively less localization around the BTz unit and a LUMO which is more delocalized over the backbone of the trimer. The differences in the LUMO localization correlate well with the strength of the ICT interaction, with a stronger interaction observed for the more strongly localized LUMOs.

. Structural (X-Ray) Characterization
The ordering of thin fi lms of the polymers was investigated by wide angle X-ray scattering (WAXS). Films were prepared by drop casting from chlorobenzene at room temperature. As shown in Figure 3 , for both PDTG-PT and PDTG-DFBT there is clear diffraction peak present at 5.16° and 4.94° (2 θ ) corresponding to d-spacings of 17.11 nm and 17.91 nm respectively, which we attribute to the lamellar packing of the polymer backbones. [ 47,48 ] Both polymers are similar to the 17.6 nm spacing observed for PDTG-BT . [ 13 ] In contrast, there is no peak present for PDTG-BTz which indicates a more amorphous structure. We believe this may be due to the extra octyl side chain present on the BTz monomer which may hinder lamellar packing by preventing intercalation of the solubilizing chains. In addition, this octyl side chain initially protrudes in the same plane as the conjugated backbone, whereas the side chains on the germanium bridge protrude perpendicular to the conjugated backbone. This particular combination of in and out of plane solubilizing groups appears to frustrate the ordering of the polymer. There is a broad and weak diffraction peak around at 25.4° (2 θ ) for PDTG-PT corresponding to a d-space packing of 3.5 nm, which we attribute to ππ stacking distance, similar to that observed in PDTG-BT . This peak is considerably broader www.afm-journal.de www.MaterialsViews.com wileyonlinelibrary.com range and increased band gap. For PDTG-DFBT the optical band gap (1.53 eV) is similar to PDTG-BT , so we believe the reduced performance is principally due to the lower molecular weight of the polymer. We note in the case of PDTG-BT , low molecular weight polymers have shown signifi cantly lower photocurrents in blends with PC 70 BM . [ 14,15 ] We further note that during the preparation of this manuscript Ohshita and coworkers have reported a PDTG-PT with a M n of 6.2 kDa. [ 53 ] The UV-Vis spectra from this polymer showed a marked reduction in the long wavelength shoulder which is usually attributed to aggregation, compared to our data for the higher molecular weight polymer. Their OPV devices were fabricated in a similar manner and showed V oc of 0.57 V, J sc of 8.89 mA cm -2 , FF of 0.57 and overall PCEs of 2.9%. Again, the difference in molecular weight appears to have a large effect on the effi ciency, predominately though a reduction in the J sc . Examination of the EQE data (Figure 4 b) shows a broad photocurrent generation into the longer wavelengths for the PDTG-PT analogue with an approximate 50-60% EQE over the absorption window, whereas both the PDTG-BTz and the PDTG-DFBT give a much lower 20% and 35% over a smaller EQE wavelength range. The EQE spectra mirror the optical absorption spectra in all three cases. Particularly signifi cant is the charge generation out to 900 nm in the case of PDTG-PT due to its very low band gap of 1.32 eV allowing it to absorb more photons in the NIR region. Thus, PDTG-PT is one of the few co-polymers [ 30,54,55 ] with E g opt below 1.4 eV that exhibits such a high EQE at longer wavelengths. aggregation and increase disorder in the blend fi lm, resulting in a reduction of the effective conjugation length along the polymer backbone and therefore an increase in the effective ionization potential. The PDTG-BTz may be more prone to the effects of disorder than the other two polymers due to the higher lying LUMO level which corresponds to reduced quinoidal character of the conjugated backbone, in agreement with the solution 1 H NMR results. High quinoidal character is associated with increasing double bond character between the two co-monomer units, and therefore a more rigid co-planar backbone, which is less prone to torsional disorder. We note that an increase in polymer ionization potential has been observed for P3HT upon blending due to the effects of disorder. [ 52 ] In our case, PESA measurements were also performed upon the blended fi lms (see Supporting Information, Figure S10). For both PDTG-PT and PDTG-DFBT , no differences were observed in the measured ionization potential for the pristine and blend fi lms. However, for PDTG-BTz a small increase in ionization potential of 0.05 eV was consistently observed for the blend fi lm. Whilst not suffi cient to fully account the increase in open circuit voltage, in combination with other factors such as changes in charge recombination dynamics, they partly explain the anomalous open circuit voltage.
Amongst the three co-polymers, only PDTG-PT has a J sc and PCE comparable to that of PDTG-BT . PDTG-BTz , and PDTG-DFBT both show a large decrease in the J sc . For PDTG-BTz this could be attributed to the relatively narrow absorption  the nitrogen lone pair. [ 63,64 ] This coordination or complexation induces a reduction in both the HOMO and LUMO but it is more pronounced on the LUMO energy, thus lowering the band gap. As PEDOT:PSS is inherently acidic, the normal device structure may be limiting the full potential of the PDTG-PT : PC 70 BM system. Indeed, a recent study by Olsen, Bazan, and co-workers, upon a small molecule electron donor containing PT, demonstrated that removing the acidic hole injection layers and replace them with a solution processable nickel oxide injection layer resulted in a signifi cant enhancement in device performance. [ 65 ] Therefore, we were interested to investigate the potential of PDTG-PT in an inverted device architecture in which molybdenum oxide (MoO 3 ) was used as the interface layer. The inverted structure consisted of a glass/ITO/ZnO ( ≈ 30 nm)/Active layer/MoO 3 (10 nm)/Ag (100 nm) confi guration. Active layers were fabricated from the highest performing solutions containing 2.5 vol% DIO in 1,2-dichlorobenzene. The ZnO was produced by the sol gel method [ 66 ] and both the MoO 3 and silver were thermally evaporated. Comparisons of the device characteristics and EQE, for both the normal and inverted cells, are shown in Figure 6 and the data summarized in Table 4 . When comparing the normal and inverted device there is an increase in overall PCE from 5.2% to 6.6%. Both the normal and inverted devices exhibit identical V oc of 0.59 V. However the inverted device exhibits a signifi cantly enhanced photocurrent, from 17.6 to 19.6 mA cm −2 , and fi ll factor from 0.51 to 0.57 respectively. Observing the J-V curve, it is evident that there is a signifi cant increase in the shunt resistance of the inverted device. This increase in shunt resistance is typically associated with the reduction of leakage current and has been linked with: less leakage at the edge of the device, fewer pinholes in the fi lm and fewer trapping sites. [ 67,68 ] Similarly, the boost in the photocurrent for the inverted device may result from a number of effects, including changes in the vertical segregation of the active fi lm layer, improved optical interference effects as a result of changes in the optical fi eld density within the active layer or improved extraction of charges from the use of different hole and electron transporting layers. In addition the removal of PEDOT:PSS, which has a small but signifi cant absorption in the near-IR region, may also enhance the active layer absorption in the near-IR region. The EQE shows a general increase over the 650−900 nm range from approximately 55% to 60% resulting in an increase in overall current. This trend has also been observed by Reynolds and co-workers using the PDTG-TPD in normal and inverted device structure. [ 19,69 ] Overall, the use of an inverted device structure increased the effi ciency by 1.4%.

. Thin Film Morphology
As mentioned above, the PCEs of the PDTG-BTz / PC 70 BM and PDTG-DFBT / PC 70 BM devices were limited by the low J sc (6.17 and 5.74 mA cm −2 ). As J sc is partly determined by charge recombination, which can be correlated to nanoscale morphology, atomic force microscopy (AFM) topography was performed to probe the surface of the devices. Figure 5 shows the topography of all three polymer-fullerene blends spun into fi lms directly onto PEDOT:PSS. Large differences in surface morphology are clearly observed between the three blends. In the case of PDTG-BTz / PC 70 BM blend fi lms, rather large features on the order of 100-200 nm are apparent, with a root mean square (RMS) surface roughness of 6.70 nm. Such large domains are suggestive of poor intermixing between the polymer and fullerene and are generally considering nonoptimal for good device performance. [56][57][58][59] The high alkyl chain density of PDTG-BTz in comparison to the other co-polymers may lead to the poor intermixing of fullerene in this case, hampering intercalation of the fullerene between the conjugated polymer backbones. Blends of PDTG-DFBT also show relatively large features on the order of 20-40 nm, and fi lms with appreciable surface roughness (RMS of 6.77 nm). Here, the poor solubility of the polymer and its high propensity to aggregate in solution may be inhibiting effective mixing with PC 70 BM. Blends of PDTG-PT / PC 70 BM showed much fi ner mixing in addition to the lowest surface roughness (RMS of 1.78 nm). The good intermixing can help to explain the increased J sc simply by increasing the interfacial area for charge separation.

. PDTG-PT Optimizations
The promising initial performance of PDTG-PT prompted us to further investigate the use of processing additives to further enhance performance, with the particular aim of further improving the rather low fi ll factor of the as-spun fi lms. Three solvents additives were investigated, dimethylformamide (DMF), [ 60 ] 1-chloronaphthalene (CN) [ 61 ] and 1,8-diiodooctane (DIO) [ 62 ] in different volumetric ratios. The J-V characteristics of devices fabricated from 1:2 polymer: PC 70 BM solutions are shown in supporting information. While no signifi cant improvement in the PCE was recorded with the use of DMF and CN ( Table 3 ) the use of 2.5 vol% DIO afforded a modest increase in the overall PCE from 4.61% to 5.2% mainly as a result of an increase in fi ll factor, with J sc = 17.6 mA cm −2 , V oc = 0.58 V, and FF = 0.51.
Recent work has demonstrated the ability of the PT monomer to accept a Lewis acid through complexation of   [ 19,21,[24][25][26] 1 H and 13 C NMR spectra were recorded on a Bruker AV-400 (400 MHz for 1 H and 100 MHz for 13 C), using the residual solvent resonance of CDCl 3 as an internal reference. Number-average ( M n ) and weightaverage ( M w ) molecular weight were determined by Agilent Technologies 1200 series GPC running in chlorobenzene at 80 °C, using two PL mixed columns in series and calibrated against narrow polydispersity polystyrene standards. Electrospray mass spectrometry was performed with a Thermo Electron Corporation DSQII mass spectrometer. UV-Vis spectra were recorded on a UV-1601 Shimadzu UV-Vis spectrometer. Flash chromatography (FC) was performed on silica gel (Merck Kieselgel 60 F254 230-400 mesh). A customer build Shimadzu recSEC system was used to fractionate the polymers and purify the tin monomer. The system comprises a DGU-20A3 degasser, an LC-20A pump, a CTO-20A column oven, an Agilent PLgel 10 μ m MIXED-D column and a SPD-20A UV detector. Polymers were fractionated using chlorobenzene as eluent at 80 °C. Monomers were purifi ed on an identical system using running in hexane at 40 °C. Photo electron spectroscopy in air (PESA) measurements were recorded with a Riken Keiki AC-2 PESA spectrometer with a power setting of 5 nW and a power number of 0.5. Samples for PESA were prepared on glass substrates by spin coating. Data reported is the average of 3 separate thin fi lm measurements for pristine samples, and 2 fi lms for blend samples. X-ray diffraction (XRD) measurements were performed on the PANALYTICAL X′ PERT-PRO MRD diffractometer equipped with nickel-fi ltered Cu K α 1 beam and X′ CELERATOR detector, using current I = 40 mA and accelerating voltage V = 40 kV. Samples were prepared by drop casting.

. Conclusions
The development of three new DTG -based co-polymers utilizing Stille cross-coupling polymerization was demonstrated. In addition, an extensive structure-optoelectronic properties-device performance relationship studies has been performed using spectroscopic and morphological (structural) analysis techniques (UV-Vis, PESA, AFM, and XRD) as well as OPV device fabrication. Comparing the impact of the electron defi cient unit on the optoelectronic properties of the co-polymers, it is important to note the discrepancy between the strength of the acceptor and the ICT energy gap. Among the three co-polymers ( PDTG-BTz , PDTG-DFBT , PDTG-PT ), the PDTG-PT has shown similar device characteristics to that of PDTG-BT but with an improved V oc , increasing the overall PCE of the OPV device. PDTG-PT performed relatively well with or without additive which may be due to better mixing with PC 70 BM and the high EQE at long wavelengths. An inverted device structure was used to further enhance the performance of PDTG-PT to 6.6%. Thus, PDTG-PT is an ideal candidate for application in tandem solar cells confi guration due to its high effi ciency at very low band gaps ( E g opt = 1.32 eV), while the 6.6% PCE is the highest reported for all the co-polymers containing bridged bithiophenes with 5-member fused rings in the central core and possessing an E g opt below 1.4 eV.

. Experimental Section
All reactions were treated as air and light sensitive and performed under argon and in the dark. All glassware used were washed using