Quinoxaline-Based Semiconducting Polymers: Effect of Fluorination

Aug 9, 2012 - J. Taylor , Hunan Yi , Ahmed Iraqi , Yiwei Zhang , Nicholas. W. Scarratt , Tao Wang , David. G. Lidzey. RSC Advances 2015 5 (123), 10160...
0 downloads 0 Views 4MB Size
Article pubs.acs.org/Macromolecules

Quinoxaline-Based Semiconducting Polymers: Effect of Fluorination on the Photophysical, Thermal, and Charge Transport Properties Akila Iyer, Josiah Bjorgaard, Trent Anderson, and Muhammet E. Köse* Department of Chemistry and Biochemistry, North Dakota State University, Fargo, North Dakota 58108, United States S Supporting Information *

ABSTRACT: A novel fluorinated quinoxaline-based acceptor has been synthesized and then copolymerized with an electron-rich benzodithiophene derivative to yield a low band gap polymer (PBDT-QxF). A non-fluorinated analogue of the same polymer (PBDT-Qx) has also been synthesized in order to disclose the effect of fluorination on polymer properties. PBDT-QxF exhibits better thermal and oxidative stability compared to non-fluorinated analogue. Fluorine atoms induce crystalline domains in solid statepossibly as a result of favorable C−F···H interactionswhereas such ordering is absent in PBDT-Qx. Principal component analysis on variable temperature absorption data collected in solution revealed a stabilization energy of ∼0.5 kcal mol−1 per repeat unit upon fluorination. Theoretical calculations predict higher oxidation potential for PBDT-QxF, which is confirmed by experimental data. Theoretical calculations also suggest inductive effect of fluorine atoms on electronic structure. The hole mobility of PBDTQxF is also higher than that of PBDT-Qx. Overall, the studies show promising photovoltaic properties of this novel monomer if used in low band gap polymers for organic solar cell applications.

1. INTRODUCTION Bulk heterojunction solar cells fabricated from low band gap polymers have shown impressive power conversion efficiencies (PCEs) in recent years.1−7 This is of great importance for developing flexible and large area devices with low-cost processing techniques.8,9 However, both PCEs and the longterm stability of the devices should be improved to increase the commercial potential of organic photovoltaic systems. There are promising reports on low band gap polymer:fullerene derivative blends with device PCEs exceeding 8%.10,11 The diversity of polymeric repeat units and the number of available structural modifications indicate that there is still much room for design and optimization of novel materials for improved photovoltaic activity. One of the most notable properties of some of the high performing polymers is the presence of fluorine atoms attached to the conjugated backbone. The fluorine atoms are believed to induce greater intra- and intermolecular interactions. The organic materials with fluorine atoms in their structure usually show better thermal and oxidative stability than their nonfluorinated analogues. When the electron-withdrawing F atom is introduced to the backbone of thieno[3,4-b]thiophene containing polymers, the HOMO level of the polymer decreased by ∼0.1 eV, yet the band gap remained the same.12 Zhou et al. reported a novel monomer with fluorine atoms attached to the 5,6-position of 2,1,3-benzothidiazole.13 The same group also reported another novel fluorinated monomer based on 2-alkylbenzo[d][1,2,3]triazole acceptor.14 The polymers made out of these fluorinated monomers yielded © 2012 American Chemical Society

solar cells with PCEs 7% or better. The hole mobility of the polymers have improved upon fluorination, though a reverse trend has also been observed by others.15 When designing a novel monomer to be used in low band gap material synthesis, it is important to know beforehand the expected frontier energy levels and optical band gap of resulting polymer.16 This is essential in order to focus on potentially interesting systems and decrease the amount of laborious synthesis work. Quinoxaline-based polymers have been reported to give high PCEs in solar cell applications.17,18 The most notable feature of quinoxaline incorporated polymers is their high open circuit voltages (0.9−1.0 V).3 To gain insight into the changes caused by fluorination, a novel fluorinated monomer based on quinoxaline acceptor has been synthesized and copolymerized with an electron-rich benzo[1,2-b:4,5b′]dithiophene unit. A non-fluorinated analogue of the same polymer was also synthesized in order to study the effect of fluorination on material properties. The photophysical and thermal properties of the polymers have been investigated. An extensive theoretical study has also been performed to explain the experimental data. We demonstrate that fluorination decrease the frontier energy levels of the resulting polymer, in agreement with the observations on other fluorinated polymers reported before.2,15 However, fluorination also causes a decrease in the band gap in solid state, though the mechanism Received: May 15, 2012 Revised: July 31, 2012 Published: August 9, 2012 6380

dx.doi.org/10.1021/ma3009788 | Macromolecules 2012, 45, 6380−6389

Macromolecules

Article

Scheme 1. Synthesis Scheme of Fluorinated Monomer and Polymers with Quinoxaline-Based Acceptors

dione, and the corresponding fluorinated quinoxaline was formed with 2-ethylhexyl side chains in appreciable yield under the same conditions. However, the final step failed, and this was inferred by the 1H NMR spectra by the presence of an aromatic proton which on integration showed a peak value as one. This suggested that we were able to incorporate only one bromine atom in the ring system. We also tried to do iodination of the former under vigorous condition, but the reaction was not successful. In our second route, we switched our attention toward the published route to synthesize 2,5-dibromo-3,4-difluoro-5,6benzenediamine by Zhang et al.19 Though the alternative route involved greater number of steps, each reaction had appreciable yield and could be accomplished with ease. Finally, the diamine 12 was ring cyclized to get QxF (Scheme 1 and Figure S1).20 Then, QxF acceptor unit was polymerized with benzo[1,2-b:4,5-b′]dithiophene as the donor by Stille coupling reaction.14 Microwave-assisted palladium-catalyzed cross-coupling polymerization was carried out (Scheme 1). The same donor unit was polymerized with the non-fluorinated analogue to get PBDT-Qx. The molecular weights determined by gel-

is believed to be improvement in planarization of conjugated backbone through favorable F atom induced intra- and/or intermolecular interactions.

2. RESULTS AND DISCUSSION Synthesis and Structural Characterization. The routes followed for the synthesis of novel acceptor molecule is shown in Scheme 1. The first route adopted for the synthesis QxF involved three simple steps. We started from the commercially available 4,5-difluoro-2-nitroaniline (1), which was then reduced by SnCl2·2H2O under acidic conditions to corresponding diamine (2). The latter was ring cyclized with hexane-3,4dione in good yield. The fluorinated quinoxaline (3) was then treated with a non-nucleophilic base and then quenched with trimethylsilyl chloride (TMSCl). This was followed by bromination in CHCl3. The presence of two fluorine atoms deactivates the aromatic ring system so much so that the incorporation of two bromine atoms in the ring becomes a challenge.14 The yield of final step was not quantitative. For solubility purposes, we tried the ring cyclization of 4,5difluorobenzene-1,2-diamine with 5,10-diethyltetradecan-7,86381

dx.doi.org/10.1021/ma3009788 | Macromolecules 2012, 45, 6380−6389

Macromolecules

Article

absorption spectra are slightly red-shifted with optical band gaps 1.88 and 1.66 eV for PBDT-Qx and PBDT-QxF, respectively. The photoluminescence spectra show similar changes to the absorption spectra (Figure S2), where PBDTQxF display ∼25 nm red-shifted emission spectrum when compared to that of PBDT-Qx at room temperature. If PBDTQxF solution is heated to 340 K, however, the color of the solution changes from dark blue to red and the absorption profile at that temperature exactly matches with that of PBDTQx at RT. This can be attributed to the aggregated domains present in the solution at room temperature. Thermal energy probably disintegrates the aggregates into free chains since lowenergy side of the PBDT-QxF spectrum show less scattered background. It is important to note that PBDT-Qx does not show any appreciable change in the absorption spectrum at other temperatures. Thus, it can be argued that fluorine atom induced interactions are overcome at 340 K. This is significant since a temperature-dependent analysis on the spectra can enable us to measure the strength of fluorine atom induced attractive interactions within the polymer. Within the view of above discussion, we have collected variable temperature absorption spectra of PBDT-QxF in chlorobenzene (Figure 2a). Above 340 K and below 304 K, the absorption profiles do not change significantly; therefore, we analyzed the data collected within this temperature range. As we pointed out above, the absorption spectrum show increased absorption at low energy of the spectrum is indicative of aggregate formation in the solution. We speculate that as the temperature is cooled down, there is an enhancement in backbone planarization due to attractive forces caused by fluorine atoms and hence a red-shift in the absorption spectrum. Planarized conjugated backbone results in improved π−π stacking of conjugated chains and leads to aggregation in the solution. The fact that PBDT-Qx does not show such redshift between solution and thin film spectra supports the above hypothesis. Principal component analysis on the data presented in Figure 2a revealed two factors contributing to the spectra. Nonorthogonal transformation matrix is used to convert abstract factors to chemically meaningful results.21 The coefficients in the transformation matrix have been determined by exploiting several conditions on the expected results. First, the deconvoluted spectra cannot be negative. Second, one of the principal components should have a similar absorption profile to the data collected at 340 K (Figure 2b). Third, the summation of eigenvector elements at each temperature should yield the same value (Figure 2c). On the basis of this prior knowledge on the nature of expected principal components, we have deconvoluted the spectra and computed loading values of each component in the original spectra. It is then possible to

permeation chromatography for PBDT-QxF and PBDT-Qx are 52.9 kDa (PDI = 2.89) and 51.2 kDa (PDI = 3.04), respectively. Optical and Electrochemical Properties. The UV−vis absorption characteristics of the polymers were measured in both chlorobenzene solution and thin films (Figure 1) with the

Figure 1. Normalized absorption spectra of polymers in chlorobenzene solutions (top) and in thin films (bottom).

relevant optical data collected in Table 1. Both polymers are soluble in common organic solvents, yet PBDT-QxF has limited solubility in some solvents such as chloroform. Solubility of PBDT-QxF, however, improves when the solution is heated. Room temperature (RT) absorption spectra of polymers show that PBDT-QxF has red-shifted absorption spectrum compared to non-fluorinated analogue. The film

Table 1. Optical and Electrochemical Properties of the Polymers along with the Corresponding Estimates from the Quantum Mechanical Calculations film

solution polymer

λmax (nm)

PL (nm)

λmax (nm)

PL (nm)

PBDT-Qx PBDT-QxF

(582) 551 (574) 616

666 693

(552) 581 (572) 624

671 723

a

a

theory Egopt

(eV)

1.88 1.66

b

c

HOMO (eV)

LUMO (eV)

−5.08 −5.32

−3.20 −3.66

Egcalc

(eV)

1.85 1.84

HOMOcalc (eV) −5.03 −5.19

a Absorption maxima were reported along with major vibronic sub-band in parentheses. bHOMO levels of the polymers were estimated in reference to the ferrocene-ferrocenium (Fc/Fc+) couple at a scan rate of 100 mV s−1 according to the empirical formula HOMO = −[Eonset − E(Fc/Fc+) + 4.8] eV. cLUMO levels were estimated according to formula LUMO = HOMO + Egopt.

6382

dx.doi.org/10.1021/ma3009788 | Macromolecules 2012, 45, 6380−6389

Macromolecules

Article

Figure 2. (a) Variable temperature absorption spectrum of PBDT-QxF in chlorobenzene solvent. (b) Principal component spectra of the data set shown in (a). Eigenvector results (loading of each component) are shown in (c). Enthalpy changes of the phase transition can be obtained from the slope according to van’t Hoff equation when natural logarithm of equilibrium constant is plotted against 1/T (R2 = 0.97).

find the equilibrium constant for the reaction free chain → aggregated chain (Figure 2d) using the van’t Hoff equation:22 ⎛ x ⎞ ΔSfree → agg ΔHfree → agg ln⎜⎜ free ⎟⎟ = − R RT ⎝ xagg ⎠

(1)

Here, x represents the mole fractions for the spectral components. ΔSfree→agg and ΔHfree→agg are the entropy (−114.5 cal K−1 mol−1) and enthalpy (−35.4 kcal mol−1) changes for the transition from aggregated form to free random chain in solution, respectively. The decrease in entropy is expected as the chain motion is restricted in aggregates. Aggregate formation is favored by enthalpy and the magnitude of enthalpy changes per repeat unit is −0.45 kcal mol−1. We believe most of the enthalpy changes occur due to intermolecular interactions involving fluorine atoms and the atoms in adjacent chains. The structural ordering in thin films of polymers has been investigated by X-ray diffraction (XRD) spectrometer (Figure 3). As expected, PBDT-QxF shows an intense peak at XRD spectrum, indicating a natural tendency to self-organize into

Figure 3. XRD patterns of polymer thin films used in this study.

6383

dx.doi.org/10.1021/ma3009788 | Macromolecules 2012, 45, 6380−6389

Macromolecules

Article

crystalline lamellae. The d-spacing for the peak at 4.8° corresponds to a plane separation of 18.6 Å. The fact that there are no higher order reflections and there is no discernible π−π stacking peak at higher angles hint relatively short-range order in the bulk. In contrast, the XRD spectrum of PBDT-Qx does not show any significant peaks, showing amorphous nature of this polymer in solid state. These results also in accordance with the changes in the absorption spectra of polymers when going from solution to solid state. To determine the HOMO and LUMO levels of polymers, cyclic voltammetry (CV) measurements were carried out under nitrogen in a three-electrode cell using 0.1 M Bu4NPF6 in anhydrous CH3CN as the supporting electrolyte. The recorded CV curves were referenced to an Ag quasi-reference electrode, which was calibrated using a ferrocene/ferrocenium (Fc/Fc+) redox couple as an external standard. The CV data and the curves of polymers are given in the Table 1 and Figure 4,

Figure 5. Thermogravimetric analysis of polymer decompositions in nitrogen atmosphere (a). Variations of PL intensity (em: 700 nm) under the similar excitation beam intensity at the same excitation wavelength (exc: 550 nm) are illustrated for both polymers in the lower panel (b).

Figure 4. Cyclic voltammograms of polymer thin films at a scan rate of 100 mV s−1.

respectively. The oxidation peaks are mostly irreversible, and the HOMO level of PBDT-Qx (−5.08 eV) is higher than that of PBDT-QxF (−5.32 eV). The LUMO level of PBDT-QxF is lower in energy than that of PBDT-Qx. Thus, fluorinated polymer exhibits deeper HOMO and LUMO levels due to highly electronegative fluorine atoms attached to the conjugated backbone. The effects of F atoms on energy levels are further discussed in Theoretical Calculations section. Thermal Degradation, Photostability, and Charge Transport Properties. Thermal stability of PBDT-QxF improves with the presence of highly electronegative fluorine atoms. Thermal gravimetric analysis (TGA) shows higher decomposition temperature for PBDT-QxF in comparison to PBDT-Qx (Figure 5a). In addition, there is still significant percent mass remaining at elevated temperatures whereas PBDT-Qx loses most of its weight. The weight loss percentages for sequential deattachment of alkoxy groups (20% weight loss per alkoxy group) can be roughly identified in Figure 5a for PBDT-QxF. Without ethyl groups, the percent mass remaining for PBDT-QxF should be around 53%, higher than the percent weight loss (45%) measured by TGA. Regardless, fluorination

improves the thermal stability of the PBDT-QxF. Differential scanning calorimetry experiments, however, did not show any glass transition temperature for both polymers (Figure S3) within the temperature range investigated. The thin films of polymers with similar optical density were excited at the same wavelength, and the time evolution of PL intensity was recorded at ambient conditions (Figure 4b). As expected, PBDT-QxF shows greater stability due to its high oxidation potential. Fluorinated polymers, in general, exhibit improved blocking properties for water penetration.23 Therefore, hindered diffusion of water within closely packed chains could also help enhanced photostability observed in PBDTQxF. The molecular weight and polydispersity index of both polymers are very similar to each other (Table 2). It is significant because the conjugated polymers show different carrier mobility regimes with differing molecular weights.24,25 The hole carrier mobilities of polymers were measured from the hole only devices (ITO/PEDOT:PSS/polymer/MoO3/Al) under the space charge limited current (SCLC) regime.26 The 6384

dx.doi.org/10.1021/ma3009788 | Macromolecules 2012, 45, 6380−6389

Macromolecules

Article

similar. There are only slight differences observed in the dihedral angles between comonomer units whereas bond length alternation is also slightly disturbed when F atoms are attached to the backbone (Figure S5). Frontier orbital images (Figure 6a) illustrate well-delocalized MOs for repeat units of both polymers. Nonetheless, the presence of high electron density on donor BDT units in HOMO plots and more MO coefficients on Qx(F) units in LUMO plots are also evident. Interestingly, there are fewer MO coefficients on F atoms. The presence of F atoms causes ∼0.15 eV stabilization of frontier levels in repeat units of BDT-QxF (dimer), hinting inductive effect of fluorination on electronic structure. This is also reflected in calculated oxidation potentials of both polymers (Figure 6b). Though, the calculated band gaps are essentially the same for these polymers. This is, however, not in contrast with the experimental findings. The absorption spectrum of PBDT-QxF collected at high temperatures exactly matches with the absorption spectrum of PBDT-Qx. The aggregation and F atom induced interactions play a major role in the observed differences in the absorption spectra at room temperature. We believe that these differences arise due to favorable C−F···H intra- and intermolecular interactions, which in turn improve

Table 2. Molecular Weight, Thermal Decomposition Temperature, and Hole Carrier Mobilities Measured under SCLC Regimea sample

Mn (kDa)

Mw (kDa)

PDI

Td (°C)b

Tg (°C)

μh (cm2/ (V·s))c

PBDT-Qx PBDT-QxF

51.2 52.9

155.6 152.7

3.04 2.89

294 302

− −

1.1 × 10−6 4.3 × 10−5

a

The molecular weights were determined by GPC with THF as eluent using polystyrene standards. bOnset decomposition temperature as measured by TGA. cDetermined by space-charge limited current (SCLC) technique.

hole mobilities of polymers (see Table 2) and the corresponding J−V curve are shown in Figure S4. Hole mobilities are quite low when compared to the hole mobility of a poly(3-hexylthiophene) (P3HT) device (1.1 × 10−4 cm2/(V s)) measured under identical conditions. Nevertheless, PBDTQxF has higher hole mobility than that of PBDT-Qx, which could be, in part, attributed to the ordered domains exhibited by this polymer in thin films. Theoretical Calculations. DFT optimized geometry of oligomers of BDT-QxF and BDT-Qx are structurally very

Figure 6. (a) Relative energies and orbital plots of frontier energy levels on two repeat units from B3LYP/6-31G* calculations. Variation of optical band gaps (b) and oxidation potentials (c) with the increasing number of repeat units are shown in the lower panels. 6385

dx.doi.org/10.1021/ma3009788 | Macromolecules 2012, 45, 6380−6389

Macromolecules

Article

Figure 7. Variation of potential energy with dihedral angle (ϕ) along with the corresponding excited state energy for model chromophores (BDTQx(F)-BDT). Electrostatic potential maps depict the areas of high electron density and hence reveal changes in charge distribution upon fluorination.

and −0.15, respectively. In the non-fluorinated analogue, the charges for C and H atoms are −0.20 and +0.14, respectively. This shows high electronegativity of fluorine atoms reverses the charge distribution on quinoxaline monomer. Actually, this can be best observed by analyzing electrostatic potential map as illustrated in Figure 7. Clearly, the presence of fluorine atoms decrease the electron density on QxF acceptor, thus making it electron poorer comonomer compared to Qx unit. The fact that the bond length parameters change only slightly and there are significant changes in charge distribution in electrostatic potential maps reveal inductive effect of fluorine atoms on quinoxaline comonomer. Fluorine atoms also increase the potential energy of bond rotation along the C−C bonds connecting monomer units (Figure 7). We believe this is due to relatively strong C−F···H interactions between comonomers as discussed above. The excited-state energies are also slightly smaller for BDT-QxFBDT chromophore, pointing out band gap reduction even fluorinated system adopts the same dihedral angle as nonfluorinated analogue. As a result, the theoretical results obtained on model chromophores are at least in qualitative agreement with the experimental observations.

backbone planarity for efficient chain aggregation. It is important to note that in the lowest energy conformation fluorine atoms prefer syn-orientation for the hydrogen atoms on BDT unit. The calculations with anti-orientation resulted in less favorable conformation (where C−F···S interactions are possible) with destabilization energy of ∼1.0 kcal mol−1 per repeat unit. Similar conformational studies on BDT-Qx oligomer showed syn-orientation as the lowest energy conformer again as in BDT-QxF oligomer, yet the energetic differences between two conformations is only ∼0.25 kcal mol−1 per repeat unit (much less than kT = 0.6 kcal mol−1). Therefore, PBDT-Qx chain structure could consist of both synand anti-orientations of comonomers with respect to each other at room temperature. As a result, the ability to form crystalline domains in solid state could be deterred by such chain structure in PBDT-Qx. This might explain the observed differences in the XRD data among the two polymers. We have performed calculations on model BDT-Qx(F)-BDT chromophores to further reveal the changes caused by F atoms on electronic and structural properties. In these calculations, a larger basis set was chosen 6-311++G* to account for possible long-range interactions. Yet, the dihedral angle calculated with B3LYP hybrid functional is ∼10° for BDT-QxF-BDT, very similar in magnitude to the calculated dihedral angles of BDTQxF oligomers (10−12°). The atomic charges were derived by fitting electrostatic potential calculated at HF level of theory on DFT optimized geometries. In fluorinated model chromophore, the atomic charges for C and F atoms in C−F bond are +0.17

3. CONCLUSIONS In summary, we have synthesized a novel electron-deficient fluorinated monomer based on the quinoxaline unit. Fluorination improves thermal and oxidative properties of the resulting polymer. We have also estimated the magnitude of entropy and 6386

dx.doi.org/10.1021/ma3009788 | Macromolecules 2012, 45, 6380−6389

Macromolecules

Article

(3.16 mL, 5.06 mmol, 2.25 equiv) was added dropwise over a period of 3 min under N2 at −78 °C. The solution was stirred for 20 min at the same temperature. Then to this solution, compound 2 (0.5 g, 2.25 mmol) in THF and trimethylsilyl chloride (0.78 mL, 8.19 mmol, 2.45 equiv) were added dropwise over a period of 10 min. The solution was then stirred at the same temperature for 3.5 h, and then it was quenched with 10 mL of NH4Cl. The reaction mixture was then warmed to RT and then poured onto a solution of saturated NH4Cl. The mixture was extracted with ethyl acetate, washed with water (3×), and dried over MgSO4. The solution was concentrated on a rotaevaporator. The residue was then dissolved in a CHCl3 solution, and bromine solution (0.92 mL, 18 mmol, 8 equiv) was added in one portion. The mixture was then stirred for 16 h in a flask shielded from light. The reaction mixture was then poured into a mixture of 10% NaOH and ice, then the organic phase was extracted with DCM, and the solution was concentrated to afford the crude compound. The compound was then purified by column chromatography by using hexane/DCM 4:1 as eluent to get the target compound as a colorless solid (72 mg, 8.5%). 1H NMR (CDCl3, 400 MHz, ppm): δ 3.11−3.05 (q, J = 8 Hz, 4H), 1.45 (t, J = 8 Hz, 6H). 13C NMR (CDCl3, 400 MHz, ppm): δ 158.8, 151.2 (d, J = 72 Hz, C−F), 148.5 (d, J = 72 Hz, C−F), 135.04, 109.5, 28.0, 11.8 GC-MS (m/z) Found: 379.90; calcd: 379.92. 5,8-Dibromo-2,3-diethyl-6,7-difluoroquinoxaline (QxF: Second Route). To a solution of 12 (23.4 mg, 77 μmol) in 6 mL of dried ethanol was added hexan-3,4-dione (0.01 mL, 91 μmol), and the solution was heated for 2 h under refluxing conditions. On cooling precipitate was formed which was filtered off. The compound was purified by column chromatography with hexane/DCM 1:1 as eluent to get the target compound as a white solid (14.8 mg, 51%). 1H NMR (CDCl3, 400 MHz, ppm): δ 3.11−3.05 (q, J = 8 Hz, 4H), 1.45 (t, J = 8 Hz, 6H). 19F NMR (CDCl3): δ −121.9. 13C NMR (CDCl3, 400 MHz, ppm): δ 158.8, 151.2 (d, J = 72 Hz, C−F), 148.5 (d, J = 72 Hz, C−F), 135.04, 109.5, 28.0, 11.8 GC-MS (m/z) Found: 379.90; calcd: 379.92. PBDT-QxF. In a clean microwave vial equipped with a stir bar, 4,8bis(octyloxy)benzo[1,2-b:4,5-b']dithiophene-2,6-diyl)bis(trimethylstannane) (120 mg, 1.0 equiv), QxF (59 mg, 1.0 equiv), Pd2dba3 (2.8 mg, 0.02 equiv), and tri(o-tolyl)phosphine (7.54 mg, 0.16 equiv) were taken. To this mixture, 2 mL of dry o-xylene was added inside a glovebox. The vial was sealed and the polymerization was done at 200 °C (at 300 W) for 20 min. The solution was then cooled and precipitated into 150 mL of methanol solution. The black solid was then filtered out into Soxhlet thimble. The solid was washed with methanol and hexane, and finally the polymer was extracted with odichlorobenzene. The polymer was then dried under vacuum for 48 h. Yield: 101 mg (97.9%). Deep blue solid. Mn = 52.9 kDa, Mw = 152.7 kDa, PDI = 2.88. PBDT-Qx. In a clean microwave vial equipped with a stir bar, (4,8bis(octyloxy)benzo[1,2-b:4,5-b′]dithiophene-2,6-diyl)bis(trimethylstannane) (49 mg, 1.0 equiv), 5,8-dibromo-2,3-diethylquinoxaline, 7 (22 mg, 1.0 equiv), Pd2dba3 (1.7 mg, 0.02 equiv), and tri(otolyl)phosphine (3.1 mg, 0.16 equiv) were taken. The rest of the polymerization procedure was similar to that of PBDT-QxF, except that the polymer was extracted using chlorobenzene as the solvent in the Soxhlet apparatus. Yield 30 mg (77%). Purple solid. Mn = 51.2 kDa, Mw = 155.6 kDa, PDI = 3.04. Device Fabrication and Characterization. Hole mobilities of the polymers were investigated by cell structure of ITO/PEDOT:PSS/ polymer/MoO3/Al. The devices were prepared on glass samples coated with patterned indium−tin oxide (ITO) purchased from Thin Film Devices, Inc., with a sheet resistance of 10−12 Ω/sq. The ITO glass samples were thoroughly cleaned in a sequence of sonication in detergent, deionized water, acetone, and isopropanol and were finally treated in a benchtop plasma cleaner (PE-50 benchtop cleaner, The Plasma Etch, Inc.) for 2 min. A solution of PEDOT:PSS (poly(3,4ethylenedioxythiophene):poly(styrenesulfonate)) (Clevious P VP AI 4083 H. C. Stark, Germany) was filtered through a 0.45 μm filter and then spin-coated at 4000 rpm for 60 s on the ITO electrode. Subsequently, the PEDOT:PSS film was baked at 120 °C for 40 min in the air to remove excess water. After transferring to a N2-filled

enthalpy changes when free chains of PBDT-QxF aggregates in the solution. Fluorine atoms help formation of solid state ordering, which is usually a desired property for the materials investigated for organic solar cells. The hole mobilities of devices are larger for PBDT-QxF along with a lower band gap measured in solid state suggest promising photovoltaic properties of this polymer. The synthesis of low band gap polymers with this novel comonomer is underway in our laboratories for applications toward efficient sunlight harvesting.

4. EXPERIMENTAL SECTION General Method and Materials. All the chemicals were purchased from Sigma-Aldrich and used as received. All compounds, except 2, 3, and QxF, were synthesized by following the previously reported literature.19,27,28 The reactions that were air-sensitive were performed in a glovebox in a nitrogen atmosphere unless otherwise stated. Freshly distilled THF from sodium with benzophenone as indicator was used. GC-MS was obtained on an Agilent Technologies 7890A GC system. 1H NMR and 13C NMR spectra were recorded on a Varian 400 and 500 MHz NMR instruments. The solvents for various NMR experiments are mentioned in the text. The chemical shifts are reported in parts per million (ppm) relative to internal standard TMS (0 ppm). A 19F NMR spectrum was obtained on a 400 MHz Varian instrument. The 19F NMR chemical shift is reported using TFA as an external reference standardized to −78.5 ppm. All UV−vis spectra were recorded using a Varian Cary 50 spectrometer. PL spectra were taken on a Jobin-Yvon Fluorolog 3 fluorescence spectrometer. The molecular weights were measured using gel permeation chromatography (GPC) with THF as an eluent, with a flow rate of 1 mL/min on a Waters modular system with 2410 refractive index detector and 515 pump. The system utilized a styragel 4E and 5E column and calibrated with polystyrene standards. Cyclic voltammetry (CV) measurements were carried out on an EDAQ Potentiostat 466 system. The conventional three-electrode setup consisting of a Pt working electrode, a Pt-wire counter electrode, and an Ag/AgCl reference electrode was used. [Bu4N]PF6 (0.1 M, CH3CN) was used as a supporting electrolyte. Ferrocene was used as an external standard. HOMO/LUMO levels of polymers were estimated in reference to the ferrocene−ferrocenium (Fc/Fc+) couple at a scan rate of 100 mV s−1. The TGA and DSC data were acquired using TA Instruments Q 500 T6A and Q 1000 DSC, respectively. The scan rate for TGA measurement was 20 °C/min while DSC measurements were performed with a scan rate of 10 °C/min. 4,5-Difluorobenzene-1,2-diamine (2). To a solution of compound 1 (3.1 g, 17.8 mmol) in 47 mL of concentrated HCl and 39 mL of dry ethanol, SnCl2·2H2O (27.52 g, 121.97 mmol) was added in several lots. After the complete addition, the solution was refluxed for 45 min and stirred at RT under ambient conditions. Aqueous KOH solution was added to adjust the pH of the solution. The alkaline solution was extracted with dichloromethane (3×), washed with water, and dried over MgSO4. The solution was concentrated to obtain the product as a crystalline white solid 2 (2.22 g, 86.5%). 1H NMR (DMSO-d6, 400 MHz, ppm): δ 6.44 (t, J = 12 Hz, 2H, Ar−H), 4.56 (br, 4H). 13C NMR (DMSO-d6, 400 MHz, ppm): δ 147.4, 139.7 (d, J = 72 Hz, C− F), 137.2 (d, J = 72 Hz, C−F), 124.6. GC-MS (m/z) Found: 144.0; calcd: 144.12. 2,3-Diethyl-6,7-difluoroquinoxaline (3). To a solution of 2 (0.5 g, 6.93 mmol) in 15 mL of benzene, hexane-3,4-dione (0.36 g, 6.93 mmol) was added, and the solution was refluxed in a Dean−Stark trap for 1 h. The mixture was then evaporated to dryness. The crude product was then purified by column chromatography using hexane:DCM (1:8) to get a solid product 3 (641 mg, 83.2%). 1H NMR (CDCl3, 400 MHz, ppm): δ 7.74 (t, J = 8 Hz, 2H, Ar−H), 3.02 (q, J = 8 Hz, 4H), 1.40 (t, J = 8 Hz, 6H). 5,8-Dibromo-2,3-diethyl-6,7-difluoroquinoxaline (QxF: First Route). To a 100 mL round-bottom flask containing diisopropylamine (0.79 mL, 5.63 mmol, 2.5 equiv) and THF (45 mL), 1.6 M of BuLi 6387

dx.doi.org/10.1021/ma3009788 | Macromolecules 2012, 45, 6380−6389

Macromolecules



glovebox, the polymer solution prepared in chlorobenzene (CB) (10 mg/mL) was spin-coated on top of the PEDOT:PSS layer at 600 rpm for 30 s. After an hour of solvent aging of the wet polymer films (∼80 nm), 20 nm of MoO3 and 80 nm of Al layer were thermally evaporated through a shadow mask at a base pressure of 2 × 10−6 mbar. The cells were encapsulated in the glovebox with a UV-curable epoxy under glass sheets and then taken out of glovebox for current−voltage (I−V) measurements. The device active area was ∼7.9 mm2 for all the devices discussed in this work. The I−V measurement of the devices was conducted on a computer-controlled Keithley 2400 source meter. Film thicknesses of polymer films were obtained with a Veeco (Dektak) step profiler for SCLC measurements. X-ray diffraction spectrometer (Philips X’Pert MPD) was used for structural analysis of thin films of polymers coated on aluminum substrates. The X-ray beam was generated by copper (Kα) target, using a tube voltage of 45 kV at electron beam current of 45 mA. The scanning angle was over the range of 2°−40°. Principal Component Analysis. The details of the methodology are given elsewhere.29 Principal component analysis allows one to reduce the data matrices into their lowest dimensionality by the use of factors and their corresponding loadings. The idea is to decompose the data matrix (D) into two matrices: fundamental factors (R) and loadings of each factor (C). The abstract factors are then transformed into chemically meaningful factors and eigenvectors by multiplying both R and C matrices with a transformation matrix (T).

D = (RT )(T −1C)

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by ND EPSCoR, NSF Grant #EPS0814442, and DOE under Award #DE-FG52-08NA28921. The technical assistance for determining the thermal properties of polymers from Heidi Docktor is greatly acknowledged. A.I. thanks Dr. Xuguang Liu for the useful discussions during the research.



REFERENCES

(1) Green, M. A.; Emery, K.; Hishikawa, Y.; Warta, W. Prog. Photovoltaics 2009, 17, 320. (2) Liang, Y.; Xu, Z.; Xia, J.; Tsai, S.-T.; Wu, Y.; Li, G.; Ray, C.; Yu, L. Adv. Mater. 2010, 22, E135. (3) Chen, J. W.; Cao, Y. Acc. Chem. Res. 2009, 42, 1709. (4) Li, Y. F. Acc. Chem. Res. 2012, 45, 723. (5) Zhou, H. X.; Yang, L. Q.; You, W. Macromolecules 2012, 45, 607. (6) Son, H. J.; He, F.; Carsten, B.; Yu, L. J. Mater. Chem. 2011, 21, 18934. (7) Boudreault, P.-L. T.; Najari, A.; Leclerc, M. Chem. Mater. 2010, 23, 456. (8) Spanggaard, H.; Krebs, F. C. Sol. Energy Mater. Sol. Cells 2004, 83, 125. (9) Coakley, K. M.; McGehee, M. D. Chem. Mater. 2004, 16, 4533. (10) He, Z. C.; Zhong, C. M.; Huang, X.; Wong, W. Y.; Wu, H. B.; Chen, L. W.; Su, S. J.; Cao, Y. Adv. Mater. 2011, 23, 4636. (11) http://www.konarka.com/index.php/site/pressreleasedetail, 2012. (12) Liang, Y. Y.; Feng, D. Q.; Wu, Y.; Tsai, S. T.; Li, G.; Ray, C.; Yu, L. P. J. Am. Chem. Soc. 2009, 131, 7792. (13) Zhou, H. X.; Yang, L. Q.; Stuart, A. C.; Price, S. C.; Liu, S. B.; You, W. Angew. Chem., Int. Ed. 2011, 50, 2995. (14) Price, S. C.; Stuart, A. C.; Yang, L. Q.; Zhou, H. X.; You, W. J. Am. Chem. Soc. 2011, 133, 4625. (15) Li, Z.; Lu, J. P.; Tse, S. C.; Zhou, J. Y.; Du, X. M.; Tao, Y.; Ding, J. F. J. Mater. Chem. 2011, 21, 3226. (16) Lin, Z. H.; Bjorgaard, J.; Yavuz, A. G.; Iyer, A.; Kose, M. E. RSC Adv. 2012, 2, 642. (17) Gadisa, A.; Mammo, W.; Andersson, L. M.; Admassie, S.; Zhang, F.; Andersson, M. R.; Inganas, O. Adv. Funct. Mater. 2007, 17, 3836. (18) Blouin, N.; Michaud, A.; Gendron, D.; Wakim, S.; Blair, E.; Neagu-Plesu, R.; Belletete, M.; Durocher, G.; Tao, Y.; Leclerc, M. J. Am. Chem. Soc. 2008, 130, 732. (19) Zhang, Y.; Chien, S. C.; Chen, K. S.; Yip, H. L.; Sun, Y.; Davies, J. A.; Chen, F. C.; Jen, A. K. Y. Chem. Commun. 2011, 47, 11026. (20) Heaton, A.; Hill, M.; Drakesmith, F. J. Fluorine Chem. 1997, 81, 133. (21) Kose, M. E.; Omar, A.; Virgin, C. A.; Carroll, B. F.; Schanze, K. S. Langmuir 2005, 21, 9110. (22) Tan, C. Y.; Pinto, M. R.; Kose, M. E.; Ghiviriga, I.; Schanze, K. S. Adv. Mater. 2004, 16, 1208. (23) Sianesi, D.; Marchionni, G.; De Pasquale, D. J. Organofluorine Chemistry: Principles and Commercial Applications; Plenum Press: New York, 1994. (24) Goh, C.; Kline, R. J.; McGehee, M. D.; Kadnikova, E. N.; Frechet, J. M. J. Appl. Phys. Lett. 2005, 86, 122110. (25) Kline, R. J.; McGehee, M. D.; Kadnikova, E. N.; Liu, J. S.; Frechet, J. M. J.; Toney, M. F. Macromolecules 2005, 38, 3312. (26) Liang, Y. Y.; Xiao, S. Q.; Feng, D. Q.; Yu, L. P. J. Phys. Chem. C 2008, 112, 7866. (27) Huo, L. J.; Hou, J. H.; Chen, H. Y.; Zhang, S. Q.; Jiang, Y.; Chen, T. L.; Yang, Y. Macromolecules 2009, 42, 6564.

(2)

(3)

Model oligomer calculations show a divergence of calculated optoelectronic properties with increasing oligomer length (Figure 6b,c). From fits to Meier’s equation, we approximate ef fective conjugation and ionization lengths when the changes in band gap and oxidation potential, respectively, become negligible (