Abstract

Through manipulation of the solubilizing side chains, we were able to dramatically improve the molecular weight (𝑀𝑤) of 4,8-dialkoxybenzo[1,2-b:4,5-b′]dithiophene (BDT)/2,1,3-benzothiadiazole (BT) copolymers. When dodecyl side chains (P1) are employed at the 4- and 8-positions of the BDT unit, we obtain a chloroform-soluble copolymer fraction with 𝑀𝑤 of 6.3 kg/mol. Surprisingly, by moving to the commonly employed 2-ethylhexyl branch (P2), 𝑀𝑤 decreases to 3.4 kg/mol. This is despite numerous reports that this side chain increases solubility and 𝑀𝑤. By moving the ethyl branch in one position relative to the polymer backbone (1-ethylhexyl, P3), 𝑀𝑤 is dramatically increased to 68.8 kg/mol. As a result of this 𝑀𝑤 increase, the shape of the absorption profile is dramatically altered, with 𝜆max = 637 nm compared with 598 nm for P1 and 579 nm for P2. The hole mobility as determined by thin film transistor (TFT) measurements is improved from 1×106 cm2/Vs for P1 and P2 to 7×104 cm2/Vs for P3, while solar cell power conversion efficiency in increased to 2.91% for P3 relative to 0.31% and 0.19% for P1 and P2, respectively.

1. Introduction

Over the past few years considerable attention has been drawn to the field of organic photovoltaics (OPVs) [13]. These devices are lightweight and can be made on flexible substrates via low cost roll-to-roll processing offering a more affordable alternative to traditional PVs based on inorganic semiconductors [46]. However, a major stumbling block has been that OPVs are typically far less efficient than their inorganic counterparts. OPV power conversion efficiencies (PCEs) need to be dramatically improved in order to compete with inorganics and improved further still to achieve grid parity. Recently, researchers have made great strides in improving OPV PCEs from <4% in 2001 to the current record of 8.3% [7].

These devices employ a bulk heterojunction (BHJ) architecture consisting of an active (absorbing) layer, comprised of a phase-separated network of a polymer donor (p-type) phase and a fullerene acceptor (n-type) phase sandwiched between an anode and cathode. Ideally, the domain sizes within the active layer are less than 10 nm, ensuring that excitons generated within the bulk of a phase are able to diffuse to the donor-acceptor interface, where charge separation occurs. It is also necessary that the network is bicontinuous to ensure that the separated charges have unobstructed paths to their respective electrodes. Improving the active layer, both its components (donor and acceptor) and its morphology are the subject of the majority of OPV research and accounts for the recent upsurge in PCEs. Contributing to these improvements are (1) new processing methods that allow for more ideal phase separation and ordering within the absorbing layer [814], (2) new fullerene acceptor materials with elevated lowest unoccupied molecular orbitals (LUMOs) that allow for improved open circuit voltages (𝑉oc) and in-turn PCEs [15, 16] and (3) new polymer materials with highest occupied molecular orbital (HOMO) and LUMO energy levels designed specifically for OPV applications [17].

Hundreds of low bandgap polymers have now been reported for BHJ solar cells [17]. However, despite many new materials having attractive HOMO and LUMO levels for OPV applications, only a small fraction has been fabricated into efficient (PCE > 5%) organic solar cells. Frequently, this is a result of low molecular weight and poor solubility. It has been shown that higher molecular weights are generally required for good charge transport and OPV performance, while good solubility is required for the fabrication of uniform thin films and general ease of film deposition. Molecular weight and solubility often go hand-in-hand, since molecular weight growth can be terminated by precipitation of a poorly soluble polymer chain. Fortunately, 𝑀𝑛 and solubility can often be simultaneously improved by altering the side chains on the polymer.

With this in mind, we were interested in improving the molecular weight of a copolymer (H7) of 4,8-didodecoxybenzo[1,2-b:4,5-b′]dithiophene (C12-BDT) and 2,1,3-benzothiadiazole (BT) reported by Hou et al. [18]. When employed in a bulk heterojunction solar cell, this polymer exhibited a PCE of 0.9%, which seemed particularly low considering that both BDT and BT have been employed in copolymers that have been used to fabricate highly efficient OPVs [11, 1926]. In more recent reports of BDT copolymers, it has been suggested that linear alkoxy chains at the 4 and 8 positions of BDT result in poorly processable low molecular weight (𝑀𝑤) materials [20, 26]. We anticipated that by altering the side chains from linear (dodecoxy) to branched, we could enhance 𝑀𝑤 and solubility, and in turn improve upon the PCE of H7.

In this paper, we detail the syntheses, characterization, and photovoltaic properties of copolymers comprised of 4,8-dialkoxybenzo[1,2-b:4,5-b′]dithiophene (BDT) and 2,1,3-benzothiadiazole, where dodecoxy (P1, H7), 2-ethylhexyl (P2) and 1-ethylhexyl (P3) side chains are employed at the 4- and 8-positions (see Scheme 1). This contribution focuses on how simple alteration of the solubilizing side chains can lead to significant increases in polymer molecular weight, charge transport, and photovoltaic performance.

572329.sch.001

2. Experimental Section

2.1. Materials

3-Thiophenecarboxylic acid (1) was purchased from Matrix Scientific Co. Diethylamine, thionyl chloride, n-butyllithium (1.6 M in hexanes), trimethyltin chloride, CAUTION: trimethyltin chloride is a neurotoxin, and proper precaution should be taken when handling it, 2-ethylhexyl bromide, methanesulfonyl chloride and 3-octanol were purchased from Aldrich Chemical Co. Bromine was purchased from Fisher Chemical Co. Palladium tetrakistriphenyl-phosphine was purchased from Acros Chemical Co. 1-bromododecane and tetrabutylammonium bromide were purchased from Eastman Chemical Co. All other reagents were purchased from common commercial sources and used without further purification unless otherwise noted. THF was dried over Na/benzophenone ketal. 4,7-Dibromo-2,1,3-benzothiadiazole (11) was synthesized according to literature procedure [27]. The nonstannylated 4,8-didodecyloxybenzo[1,2-b:4,5-b′]dithiophene (5) and 4,8-bis(2-ethylhexyloxy)benzo[1,2-b:4,5-b′]dithiophene (6) were synthesized according to the published procedures [18, 21].

2.2. Instrumentation

Flash chromatography was performed on a Biotage Isolera Flash Purification System using Biotage SNAP Flash Purification Cartridges as the stationary phase. Microwave-assisted polymerizations were carried out using a CEM Discover Microwave reactor. 1H and 13C NMR spectra were recorded on a Bruker Avance DPX-300 NMR spectrometer. 119Sn NMR spectra were recorded on a Bruker Avance DRX-500 instrument. UV-vis absorption spectra were recorded on an Agilent 8453 diode-array spectrophotometer operating over a range of 190–1100 nm. GC-MS were recorded on an Agilent 6850 Series GC system coupled to an Agilent 5973 mass selective detector run in electron impact mode. Infrared spectra were recorded over the 450–4000 cm−1 region using on a Perkin Elmer Spectrum 100 spectrophotometer with an ATR sampling accessory equipped with a diamond anvil. Gel permeation chromatography (135°C in 1,2,4-trichlorobenzene) was performed by American Polymer Standards (Ohio).

2.3. Synthesis
2.3.1.  3-Thiophenecarbonyl Chloride (2)

3-Thiophenecarboxylic acid (20.0 g, 0.156 mol) (1) was added to a round bottom flask equipped with a reflux condenser and stopper. Thionyl chloride (100 mL) was then added, and the mixture was allowed to reflux for 4 hours under a blanket of argon. The excess thionyl chloride was distilled over and the solution was dried on a vacuum line to yield colorless crystals (22.7 g, 99%). Characterization was consistent with published data.

2.3.2. N,N-Diethylthiophene-3-carboxamide (3)

3-Thiophenecar-bonyl chloride (13.2 g, 0.09 mol) was dissolved in methylene chloride (45 mL) and added dropwise at 0°C to a solution of diethylamine (25 mL) in methylene chloride (25 mL). The solution was stirred at 0°C for 2 hours and then at RT overnight. Diethylamine (10 mL) was added and the mixture was stirred for another 2 hours. The diethylamine hydrochloride precipitate was filtered off and the organic phase was extracted with water. The solution was dried over magnesium sulfate and evaporated to give a brown oil (14.1 g, 85%). Characterization of the product was consistent with the published data. One pot procedure. 3-Thiophenecarboxylic acid (7.0 g, 0.055 mol) and thionyl chloride (5.0 mL, 0.068 mol) were added to a round bottom flask containing methylene chloride (200 mL) and cooled to 0°C. Triethylamine (15 mL) was then added dropwise over 15 minutes resulting in the formation of a white precipitate. The mixture was allowed to warm to room temperature. After 30 minutes, diethylamine (8 mL) was added dropwise over 10 minutes. The mixture was stirred for an additional 30 minutes at room temperature. The methylene chloride was evaporated, then ether was added, and the triethylamine hydrochloride precipitate was filtered and washed with a small amount of ether. The ether was then evaporated to leave a brown oil (8.6 g, 86%).

2.3.3.  4,8-Dihydrobenzo[1,2-b;4,5-b′]dithiophene-4,8-dione (4)

Compound 2 (14.1 g, 0.077 mol) was added to a 2-neck round bottom flask that was charged with dry THF (50 mL). n-Butyllithium (51 mL, 0.082 mmol, 1.6 M in hexanes) was added dropwise over 30 minutes at 0°C. The mixture was allowed to stir at 0°C for 30 minutes and then at RT for 2 hours. The mixture was poured over ice and allowed to sit for 4 hours. The precipitate was filtered, washed with water and methanol, and the olive green product was allowed to dry in air overnight. Typical yields range from 75%–92%. Characterization was consistent with the published data.

2.3.4.  4,8-Bis(1-ethylhexyloxy)benzo[1,2-b:4,5-b′]dithiophene (7)

Compound 3 (4.0 g, 18.1 mmol) was added to a flask containing Zn dust (2.6 g, 39.8 mmol) and NaOH (85.0 mL, 25%) solution. The green mixture was allowed to reflux for 1 hour. The color changed to orange and 3-octylmethanesulfonate (7.3 g, 35 mmol), and tetrabutylammonium bromide (600 mg, ~10 mol%) were added. After 2 hours, additional portions of 3-octylmesylate (1.34 g, 6.4 mmol) and Zn dust (1.2 g, 18.3 mmol) were added, and the mixture was refluxed overnight. After cooling, the mixture was quenched with H2O and separated with ether. The organic phase was dried over MgSO4 and evaporated to yield an orange oil (4.40 g, 54%). 1H-NMR (CDCl3), δ (ppm): 7.47 (d, 2H); 7.33 (d, 2H); 4.52 (m, 2H); 1.73 (m, 8H); 1.51 (m, 4H); 1.30 (m, 8H) 1.03 (m, 6H); 0.89 (m, 6H). 13C-NMR (CDCl3), δ (ppm): 143.1, 132.4, 130.5, 125.5, 120.7, 83.7, 33.7, 32.1, 26.9, 25.1, 22.6, 14.0, 9.6. MS (EI): calc’d 446.23, found (M+1)+, 447.3.

2.3.5. General Procedure for Stannylation of Compounds (5), (6), and (7)

 Compounds 57 in dry THF (20 equiv. by wt.) were cooled to −78°C in an acetone/dry ice bath. n-Butyllithium (2.2 equiv, 1.6 M) in hexanes was added dropwise. The solution was stirred at −78°C for 30 minutes and at RT for 2 hours. The solution was then cooled to −78°C and freshly prepared trimethyltin chloride solution (2.4 equiv, 20% by wt. in THF) was added dropwise, and the solution was allowed to stir at RT overnight. The solution was poured into cold water and extracted with ether. The solution was washed with water, and the organic phase was dried over anhydrous MgSO4 and evaporated to give a waxy yellow solid (>95% yield). White powders could be obtained by triturating with EtOH. No difference in purity was observed via NMR spectroscopy, nor were there differences observed in subsequent polymerization reactions. 2,6-Bis(trimethyltin)-4,8-didodecyloxybenzo[1,2-b:4,5-b′]dithiophene (8). 1H-NMR (CDCl3), δ (ppm): 7.52 (s, 2H); 4.30 (t, 4H); 1.89 (m, 4H); 1.59 (m, 4H); 1.23–1.43 (m, 32H); 0.89 (t, 6H); 0.46 (s, 18H). 13C-NMR (CDCl3), δ (ppm): 143.2, 140.5, 134.5, 133.0, 128.0, 73.6, 32.0, 30.6, 29.8, 29.7, 29.5, 29.4, 26.2, 22.7, 14.2, −8.3. 119Sn-NMR (CDCl3), δ (ppm) relative to SnMe3Cl (164), −33.99. 2,6-Bis(trimethyltin)-4,8-bis(2-ethylhexyloxy)benzo[1,2-b:4,5-b′]dithiophene (9). 1H-NMR (CDCl3), δ (ppm): 7.53 (s, 2H); 4.20 (d, 4H); 1.54–1.64 (m, 4H); 1.33–1.88 (m, 18H); 1.04 (t, 6H) 0.96 (t, 6H) 0.46 (s, 18H). 13C-NMR (CDCl3), δ (ppm): 143.3, 140.4, 133.9, 132.9, 128.0, 75.7, 40.7, 30.6, 29.3, 24.0, 23.2, 14.2, 11.4, −8.3. 119Sn-NMR (CDCl3), δ (ppm) relative to SnMe3Cl (164), −33.91. 2,6-Bis(trimethyltin)-4,8-bis(1-ethylhexyloxy)benzo[1,2-b:4,5- b′]dithiophene (10). 1H-NMR (CDCl3), δ (ppm): 7.53 (s, 2H); 4.58 (m, 2H); 1.74 (m, 8H); 1.56 (m, 6H); 1.33 (m, 8H); 1.05 (m, 6H) 0.90 (m, 6H) 0.44 (s, 18H). 13C-NMR (CDCl3), δ (ppm): 141.8, 139.8, 134.4, 133.8, 128.7, 83.2, 33.7, 32.2, 27.0, 25.1, 22.7, 14.1, 9.7, −8.3. 119Sn-NMR (CDCl3), δ (ppm) relative to SnMe3Cl (164), −34.18.

2.3.6. General Procedure for Polymer Synthesis of P1–P3

Under ambient atmosphere in a 5 mL microwave tube equipped with a stir bar was added the bis(trimethyltin) monomer (0.5 mmol) along with the dibromo monomer (0.485 mmol) and 2 mL of chlorobenzene. The mixture was stirred for 5 minutes and tetrakis(triphenylphosphine)palladium(0) (2.5–5 mol%) was added to the tube and the tube was capped and set in the microwave at 200°C for 10 minutes. The viscous gel was precipitated in methanol and filtered. The solid was then added to a Soxhlet thimble and subjected to extractions with hexanes (6 hrs), THF (24 hrs), and finally chloroform (24 hrs). The chloroform extract was evaporated almost to completion, and methanol was added to precipitate the polymer, which was filtered and dried under vacuum for 24 hours.

2.3.7. Poly[(4,8-Bis(1-dodecyloxy)benzo[1,2-b:4,5-b′]dithiophene-2,6-diyl-alt-2,1,3-benzothiadiazole-4,7-diyl] (P1)

(Yield from CHCl3 extract, 5–30%). FT-IR: 2917 (s), 2849 (s), 1577 (m), 1524 (m), 1489 (m), 1450 (s), 1400 (m), 1354 (s), 1273(m), 1177 (s), 1038 (br), 908 (m), 854 (m), 816 (s), 750 (w), 718 (m), 689 (m). GPC (TCB, 135°C): 𝑀𝑛=4,760 g/mol, 𝑀𝑤=6,310 g/mol, PDI=1.33. 𝜆max=598 nm (1,2-dichlorobenzene).

2.3.8. Poly[(4,8-bis(2-ethylhexyl)benzo[1,2-b:4,5-b′]dithiophene-2,6-diyl-alt-2,1,3-benzothiadiazole-4,7-diyl] (P2)

(Yield from CHCl3 extract, 5–20%). FT-IR: 2951 (m), 2916 (s), 2850 (s), 1573 (m), 1523 (m), 1487 (m), 1447 (s), 1397 (m), 1351 (s), 1257 (w), 1175 (s), 1030 (br), 906 (m), 852 (w), 817 (s), 715 (w), 688 (w). GPC (TCB, 135°C): 𝑀𝑛=2,880 g/mol, 𝑀𝑤=3,430 g/mol, 𝑃𝐷𝐼=1.19. 𝜆max=579 nm (1,2-dichlorobenzene).

2.3.9. Poly[(4,8-bis(1-ethylhexyl)benzo[1,2-b:4,5-b′]dithiophene-2,6-diyl-alt-2,1,3-benzothiadiazole-4,7-diyl] (P3)

(Yield from CHCl3 extract, 40–56%.) FT-IR: 2951 (m), 2916 (s), 2852 (s), 1575 (m), 1523 (m), 1486 (m), 1448 (s), 1396 (m), 1337 (s), 1257 (w), 1178 (s), 1019 (br), 906 (m), 853 (w), 818 (s), 715 (w), 688 (w). GPC (TCB, 135°C): 𝑀𝑛=27,100 g/mol, 𝑀𝑤=68,800 g/mol, 𝑃𝐷𝐼=2.54. 𝜆max=637 nm (1,2-dichlorobenzene).

2.4. Electrochemistry

Cyclic voltammetry (CV) was carried out using a computer-controlled Pine Model AFCBP 1 Bi-Potentiostat with PineChem software in a standard single-compartment, three electrode cell. The working electrode was glassy carbon, while the counter electrode was a platinum wire. The pseudoreference electrode was a silver wire and was calibrated against (Fc/Fc+). The polymer was drop-cast onto glassy carbon from a 2.5 mg/mL solution in chlorobenzene. All measurements were carried out in degassed solutions of acetonitrile with tetrabutylammonium hexafluorophosphate (0.1 M, electrochemical grade) as the supporting electrolyte. The scan rate used was 100 mVs−1. The electrochemical onsets were determined as the position at which the current differed by 2 μA from the baseline. The highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) energy levels were calculated from the oxidation (𝐸oxonset) and reduction (𝐸redonset) onsets, respectively, according to𝐸HOMO𝐸=oxonset𝐸+4.8(eV)(1)LUMO𝐸=redonset+4.8(eV)(2)

2.5. Powder X-Ray Diffraction

X-Ray powder diffraction data were collected at room temperature using a BRUKER P4 general-purpose four-circle X-ray diffractometer modified with a GADDS/Hi-Star detector at 40 kv and 30 mA for Cu Kα radiation (𝜆=1.5406 Å). The GADDS software suite was used to control the goniometer [28]. The samples were mounted on a loop and two frames were measured at 2𝜃=20 and 50° with exposure time of 180 seconds/frame. For the inset powder diffraction spectrum, one frame was measured at 2𝜃=25 with an exposure time of 180 seconds/frame for 10 frames. Area integration techniques were used to reproduce the single powder diffraction pattern for each frame. The powder patterns were each merged and analyzed using the EVA program to produce a single powder diffraction pattern [29].

2.6. Charge Transport Measurements

Bottom-contact thin-film transistors (TFTs) were fabricated by spin-coating (1000 rpm) polymer solutions (0.75 wt% in 1,2-dichlorobenzene) over the 1.2 cm × 1.2 cm substrate with prefabricated device structures containing a Si gate electrode, 300 nm SiO2 gate dielectric and 5 nm Ti/45 nm Au source-drain contacts with channel length 𝐿=5μm and channel width 𝑊=1000μm. Prior to spin coating, the substrates were cleaned by soaking in hot acetone, followed by isopropyl alcohol. The substrates were then cleaned in UV ozone for five minutes and rinsed with DI water. The surfaces were chemically treated by submerging the substrates for 30 min in a 17 mmol n-octyltrichlorosilane (OTS) solution in hexadecane at room temperature. The substrates were then removed and sonicated in isopropyl alcohol for 10 minutes, sonicated in chloroform for 10 minutes, and rinsed with isopropyl alcohol followed by DI water. The polymer solutions were stirred overnight prior to spin coating to ensure complete dissolution. After spin-coating, the substrates were placed in a vacuum oven overnight prior to measuring the current-voltage characteristics.

The transistors were characterized by measuring the drain current (𝐼D) against a sweeping gate voltage (𝑉GS) from −20 V to 60 V in the saturation regime, drain source voltage 𝑉DS=60 V. The measurements were made in ambient conditions using a commercially available probe station and an Agilent 4155C Semiconductor Parameter Analyzer. Five devices were measured for each material. The field-effect mobility (𝜇) was calculated from the slope of the (𝐼D)1/2 versus 𝑉GS graph, using the relationship in 𝐼D=12𝑊𝐿𝐶𝑖𝑉𝜇GS𝑉T2,(3) where 𝐶𝑖 is the capacitance per unit area for the SiO2 gate dielectric, 𝑊 is the channel width, 𝐿 is the channel length, and 𝑉T is the threshold voltage. A value for 𝜕𝐼1/2D/𝜕𝑉GS was determined from the slope of the linear portion of the (𝐼D)1/2 versus 𝑉GS plot.

2.7. Photovoltaic Characteristics

ITO-coated glass slides (Delta Technologies) were cleaned by sonication in acetone and isopropanol for 20 minutes each, followed by ozone cleaning for 30 minutes. A solution of PEDOT : PSS (Clevios P—H.C. Starck) at 1 : 1 PEDOT : PSS solution to deionized water was spin-coated onto the slides at 4000 RPM (resulting in a 40 nm layer). The films were then baked in a vacuum oven at 100°C for 30 minutes. Polymer solutions were then spin-coated onto the PEDOT : PSS-coated substrates. Solutions were prepared by mixing polymer and 1,2-dichlorobenzene (8 mg/mL for P2 and P3, 10 mg/mL for P1), then they were heated (80°C for P3, 100°C for P1 and P3) and stirred for 24 hours. The solution for P3 was then filtered while hot (0.45 μm PVDF (Millipore)). P1 and P2 were spin-coated from unfiltered, hot solutions, since filtration removed much of the polymer and resulted in poorer device performance and a high frequency of devices shorting. Unfiltered solutions of P3 resulted in devices with ~10% poorer power conversion efficiency. Optimal spin coating conditions were found to be 600 RPM for 30 s for P1 and P2, and 1500 RPM for 40 s for P3.

The films were set to dry for 30 minutes and then placed under vacuum for thermal deposition of electrodes (2 nm LiF/100 nm aluminum). The active area of the device was 38 mm2. All film thicknesses were determined via a JEOL JSPM-5200 operating in AFM tapping mode.

The current-voltage (𝐽-𝑉) characteristics were measured with a Keithley 236 digital source meter under simulated air mass (AM) 1.5 solar irradiation of 100 mW/cm2 (Oriel 150 W Xenon Light Source). The light intensity was calibrated with a NREL calibrated reference cell (Newport). The external quantum efficiency (EQE) curve was measured using a 300 W Oriel Xenon light source passed through an Oriel Cornerstone 260 monochromator, a Merlin lock-in amplifier, a calibrated Si UV detector, and an SR 570 low noise preamplifier.

3. Results and Discussion

3.1. Synthesis and Characterization

The synthetic routes to the monomers and polymers are shown in Scheme 2. Starting from 3-thiophenecarboxylic acid (1), we developed a new one-pot procedure for accessing the synthetic intermediate N,N-diethylthiophene-3-carboxamide (3). This involves the in situ preparation of thiophene-3-carbonyl chloride (2) via the reaction of 1 with thionyl chloride and excess triethylamine in dichloromethane. The triethylamine functions as a catalyst by generating the more reactive carboxylate and also reacts with the hydrochloric acid (HCl) byproduct generated by the chlorination reaction. After 30 minutes of stirring, a single equivalent of diethylamine was added to the reaction mixture containing 2 to form the desired product. The HCl generated by this reaction was reacted with the remaining triethylamine. The dione 4, the 4,8-dialkoxy-BDTs (57) and the bis-stannane monomers (810) are prepared by procedures similar to those in the literature [18, 21]. The purity of monomers 810 was confirmed by 119Sn-NMR spectroscopy (See Supplementary Materials available at http://dx.doi.org/10.1155/2011/572329) a very useful technique for determining whether there are any trace monostannane impurities in the monomers. These impurities are detrimental to molecular weight growth.

572329.sch.002

The polymers P1P3 were prepared by microwave-assisted Stille cross-coupling polymerization of the bis-stannane monomers 810 with 4,7-dibromo-2,1,3-benzothiadiazole (BT) in chlorobenzene using tetrakis(triphenylphosphine)palladium(0) ((Ph3P)4Pd) as the catalyst (see Scheme 1). Microwave heating was used because it is more efficient than conventional heating methods, and it has also been reported to give higher molecular weights [19, 22, 3032]. After reaction the polymers were precipitated into methanol, collected by filtration, and transferred to a Soxhlet extractor where the material was washed with hexanes and extracted with chloroform. The chloroform-soluble fraction was isolated to ensure the material was sufficiently soluble for the preparation OPV devices. In all cases, material remained in the extraction thimble after chloroform extraction; however, the amount of this material was highly variable. For P1 and P2, yields from the chloroform soluble fractions were often less than 10%, while for P3, yields as high as 55% were obtained. Assuming no crosslinking has occurred, the fact that material remained in the extraction thimble after chloroform extraction means that the chloroform fraction contained material with the highest molecular weight that is soluble in chloroform. The molecular weights of the these fractions were determined by gel permeation chromatography (GPC) at 135°C using 1,2,4-trichlorobenzene as the solvent, and the results are summarized in Table 1. P1, P2, and P3 were found to have weight average molecular weights (𝑀𝑤) of 6.3, 3.4, and 68.8 kg/mol, respectively. Our result for P1 differs dramatically from that of Hou et al. and may be due to differences in GPC protocols. Hou et al. ran elutions at RT using THF and found a 𝑀𝑤 of 31.4 kg/mol, while our GPC protocol was 135°C using 1,2,4-trichlorobenzene as eluent and found our 𝑀𝑤 to be 6.3 kg/mol. For P2, we find that despite several reports of the beneficial effect of the 2-ethylhexyl side chain on solubility and molecular weight, P2 has a lower 𝑀𝑤 than P1. Lastly we find that moving the ethyl branch in by one position (P2 compared with P3) results in an unexpectedly large increase in 𝑀𝑤.

Thermogravimetric analysis (TGA) was performed on polymers P1P3 under argon at 5°C/min as shown in Figure 1. The results indicate that the polymers have similar thermal stabilities with the onsets of degradation ranging from 268°C and 278°C. This degradation has been attributed to the loss of the side chains [18], which is supported by the fact that the P1 which has 12 carbon side chain loses a larger weight percentage compared with P2 and P3, which have 8 carbon side chains.

3.2. Optical and Electrochemical Properties

UV-visible absorption spectra (normalized by area) for polymers P1P3 in chloroform solution are shown in Figure 2. P2 is considerably blue-shifted of P1, with a 𝜆max of 579 nm compared to 598 nm for P1. The spectrum for P1 is similar in shape and slightly red-shifted of Hou et al.’s H7 (𝜆max=591 nm in THF) indicating comparable molecular weight. The shape of the absorption profile for P3 is very different from those of P1 and P2 with a very sharp low-energy feature dominating the spectrum—resulting in a 𝜆max of 637 nm for P3. This sharp low-energy feature has been seen in other low bandgap copolymers and is indicative of high molecular weight and often high solar cell performance [11, 19, 33, 34]. UV-Vis spectra of P1 and P3 films (dropcast from chloroform) are shown in the dashed spectra in Figure 2. A spectrum for P2 is not shown as dewetting of the chloroform solution prevented preparation of uniform film. The onset of absorption for P1 red shifts 45 nm transitioning from 685 nm in solution to 730 nm in the film. The shift in absorption onset for P3 is not as large, shifting 35 nm from 695 nm in solution to 730 nm in the film. The film spectrum for P1 shows the presence of low-energy shoulder that was not observed in solution, while the spectrum for P3 is broadened but similar in shape to the solution spectrum. The optical bandgaps for both P1 and P3 as determined by absorption onset is 1.69 eV.

Cyclic voltammetry measurements were made on polymer films of P1P3 prepared by dropcasting chloroform solutions of the polymers onto a glassy carbon electrode (see Figure 3). The onsets of the oxidation and reduction potentials for the three polymers were comparable—not surprising given that they share a common backbone. These results are summarized in Table 2. The oxidation onsets, determined as the point where the current differed from the baseline by 2 μA, were found to be 0.15, 0.13, and 0.12 V for P1, P2, and P3 giving rise to HOMO levels of −4.95 ([18] −5.10 eV), −4.93, and −4.92 eV, respectively. Similarly, the reduction onsets were found to be −1.72, −1.62 and −1.59 V for P1, P2, and P3 giving rise to LUMO levels of −3.08 ([18] −3.19 eV), −3.18, and −3.21 eV, respectively. These results give electrochemical bandgaps of 1.87 ([18] 1.91 eV), 1.75, and 1.71 eV for P1, P2, and P3.

3.3. Solid State Ordering and Charge Transport

To give insight into the internal packing structure of the polymers, powder X-ray diffraction (PXRD) was used. Samples were mounted on a loop and two frames were measured at 2𝜃=20 and 50° with exposure time of 180 seconds/frame. The integrated diffraction patterns are shown in Figure 4(a). For the inset powder diffraction pattern, 10 frames were measured at 2𝜃=25 with an exposure time of 180 seconds/frame, then they were added together. The intensities of the diffraction patterns decrease and diffraction peaks broaden from P1 to P2 to P3, indicating that structural order within the solid polymers decreases in the order P1 > P2 > P3. This order correlates with the increasing bulk of the side chain relative to the conjugated backbone. For P3, which has a much higher 𝑀𝑤, this lack of order is further exacerbated, since order has been shown to decrease as 𝑀𝑤 increases [35].

The peaks at low angle (2𝜃=4.15 for P1, 5.7° for P2 and 5.5° for P3) are assigned to the intermolecular distance between the polymer main chains separated by an alkoxy side chain (d1-spacing; see Figure 4(b)). The dodecyloxy chain (P1) gives the longest d1-spacing at 21.3 Å, while the shorter 2-ethylhexyloxy (P2) and 1-ethylhexyloxy (P3) side chains give small d1-spacings (15.5 and 16.1 Å, resp.). A plot of the low angle d-spacing versus no. of carbon atoms for the three polymers gave a slope of 0.98 Å/carbon atom, which is indicative of an interdigitation packing mode (see Figure 4(b)), not an end-to-end packing mode like that seen in polythiophenes [3638]. The peaks at 2𝜃=24.9 for P1, 24.5° for P2 and 23.8° for P3 correspond to the π-π stacking distances (d2-spacings) 3.58, 3.63, and 3.74 Å for P1, P2, and P3, respectively. Despite the increase in the π-π stacking distance due to the 1-ethylhexyl chain, it is worth noting that P3 still has a smaller π-π stacking distance than poly(3-hexylthiophene).

Thin-film transistors were fabricated from the three polymers to assess charge transport. Figure 5 shows the current-voltage characteristics of a representative device for each polymer with channel length 𝐿=5μm and channel width 𝑊=1000μm. Here, we plot the log(𝐼D) versus 𝑉GS in solid symbols (right axis) and the (𝐼D)1/2 versus 𝑉GS in open symbols (left axis) in the saturation regime (𝑉DS=60 V). The field-effect hole mobility (𝜇) of the devices shown here are 𝜇=1.36×106, 9.69×107 cm2/Vs and 8.17×104 cm2/Vs for P1, P2, and P3, respectively. These values represent an underestimation of the mobility of these materials, as optical inspection of the devices demonstrated that the film partially dewets the substrates, and the material covers only a fraction of the width of the device. Thus the real width in (3) is lower than the width defined by photolithography. The on/off current ratios for the three devices presented here are 102, 102, and 104, for P1, P2, and P3, respectively.

3.4. Photovoltaic Properties

The photovoltaic characteristics of the polymers were determined using a standard bulk heterojunction device architecture—ITO/PEDOT : PSS/P1P3 : PC61BM/LiF/Al. The devices (0.38 cm2) were fabricated under an inert atmosphere and tested in air. The active layers were deposited from 1,2-dichlorobenzene solutions (8 mg/mL for P2 and P3, 10 mg/mL for P1) and the optimal weight ratios of polymer : PC61BM was determined to be 1 : 2.5 for P1 and P3 and 1 : 2.2 for P2. The active layer thickness for the optimal devices as determined by atomic force microscopy (AFM) were 130, 50, and 100 nm for P1, P2, and P3, respectively. Due to the poor solubility of P2 and its low 𝑀𝑤, thicker films could not be achieved even at the lowest spin speeds. The current-voltage curves for the best devices for each polymer are shown in Figure 6, and the results are summarized in Table 3.

Consistent with the HOMO levels determined by cyclic voltammetry, the open circuit voltages (𝑉oc) are similar for the three polymers, near 0.7 V. The short circuit currents (𝐽sc) for P1 and P2 are similarly low (near 1 mA/cm2), while it is much higher for P3 at 8.93 mA/cm2. The fill factors (FFs) increase with increasing 𝑀𝑤 peaking at 0.46 for P3. Consequently, P3 has a much higher power conversion efficiency (PCE) peaking at 2.91% compared with 0.31% for P1 and 0.19% for P2.

The EQE spectra for the P1 and P2 devices (seen in Figure 7) show poor current generation across the spectrum with peak EQE less than 12%. P3, on the other hand, shows EQE above 30% from 330–670 nm, with a peak at 51% at 370 nm. Between 400 and 600 nm, the EQE dips, indicating that current production could be enhanced by using PC71BM instead of PC61BM as PC71BM absorbs in this range [26]. A comparison between these devices fabricated from P1 with Hou et al.’s H7 [18] reveals similar 𝑉oc’s (0.69 V for P1 versus 0.68 V for H7) and FFs (0.42 versus 0.44) though they obtain a higher 𝐽sc of 2.97 mA/cm2 resulting in a higher PCE of 0.90%. The discrepancy in results could be explained by the fact that Hou et al. used a Ca/Al cathode and tested the devices under nitrogen, whereas we used a LiF/Al cathode and tested in air. P3 outperforms H7 across the board, with higher 𝑉oc, 𝐽sc, FF, and PCE, highlighting the significance of our side chain variation. It is worth noting that the P3 solar cells were prepared from pristine DCB solution without the use of optimization techniques such as annealing or additive processing that could greatly enhance the PCE.

4. Conclusions

We prepared and characterized two new low bandgap copolymers based on BDT and BT that employ branched side chains at the 4 and 8 positions, poly[(4,8-bis(2-ethylhexyloxy)benzo[1,2-b:4,5-b′]dithiophene-2,6-diyl-alt-2,1,3-benzothiadiazole-4,7-diyl] (P2) and poly[(4,8-bis(1-ethylhexyloxy)benzo[1,2-b:4,5-b′]dithiophene-2,6-diyl-alt-2,1,3-benzothiadiazole-4,7-diyl] (P3) as well as the previously reported linear side chain analog poly[(4,8-didodecyloxy)benzo[1,2-b:4,5-b′]dithiophene-2,6-diyl-alt-2,1,3-benzothiadiazole-4,7-diyl] (P1). Surprisingly, transitioning from the linear dodecyl side chain (P1) to the branched 2-ethylhexyl side chain (P2) results in a decrease in 𝑀𝑤; however, by moving the ethyl branch in by one position relative to the polymer backbone (P3), we observe a dramatic increase in 𝑀𝑤 to 68.8 kg/mol compared with 6.3 kg/mol for P1 and 3.4 kg/mol for P2. This results in vastly different optical properties, with the appearance of a sharp low-energy feature in the UV-visible absorption spectrum of P3 that is not present in the spectra of P1 and P2. Despite the increase in the π-π stacking distance due to the 1-ethylhexyl branch in P3 relative to P1 and P2, the increased 𝑀𝑤 results in a factor of 100 increase in hole mobility, and a factor of 10 increase in the PCE of bulk heterojunction solar cells. We observe a peak PCE of 2.91% for devices based on P3 and PC61BM with minimal optimization indicating that this material has the potential to make highly efficient solar cells. The use of the 1-ethylhexyl side chain, and indeed, the general technique of moving side chain branches closer to the polymer backbone is a widely applicable means of increasing 𝑀𝑤 in conjugated polymers, many of which have not reached their full potential because of insufficient 𝑀𝑤.

Acknowledgments

This work was supported by grants from FiberCell Inc. and the Department of Energy (DOE DE-FG02-07ER46428).

Supplementary Materials

The contents of supporting information include 1H and 13C NMR spectra for 7 and 10, 119Sn NMR spectra for monomer 8--10, and FTIR spectra of P1--P3.

  1. Supplementary Material