Fine Structural Tuning of Cyanated Dithieno[3,2-b:2′,3′-d]silole

Article pubs.acs.org/Macromolecules

Fine Structural Tuning of Cyanated Dithieno[3,2‑b:2′,3′‑d]silole− Oligothiophene Copolymers: Synthesis, Characterization, and Photovoltaic Response Mirko Seri,*,† Margherita Bolognesi,‡ Zhihua Chen,§ Shaofeng Lu,§ Wouter Koopman,∥ Antonio Facchetti,*,§ and Michele Muccini*,∥ †

Istituto per la Sintesi Organica e la Fotoreattività (ISOF), Consiglio Nazionale delle Ricerche (CNR), Via P. Gobetti, 101, 40129 Bologna, Italy ‡ Laboratory MIST E-R, Via P. Gobetti, 101, 40129 Bologna, Italy § Polyera Corporation, 8045 Lamon Avenue, Skokie, Illinois 60077, United States ∥ Istituto per lo Studio dei Materiali Nanostrutturati (ISMN), Consiglio Nazionale delle Ricerche (CNR), Via P. Gobetti, 101, 40129 Bologna, Italy S Supporting Information *

ABSTRACT: We report here the synthesis and characterization of a new series of semiconducting polymers based on dithieno[3,2-b:2′,3′-d]silole (SiDT) copolymerized with cyanated oligothiophenes (-2T- or -4T-) units. The effect of the fine structural tuning of the thiophene-based spacer on optical, electronic, morphological, and photophysical properties of the resulting polymers is investigated and correlated with the organic photovoltaic (OPV) performance. Bulk heterojunction (BHJ) solar cells, using this class of copolymers as electron donor material, are fabricated, optimized, and fully characterized. As a result of rational structural modifications, PCEs of ∼5% and open-circuit voltages (VOC) greater than 0.8 V are achieved without the need of additional thermal annealing.

■

INTRODUCTION Solution-processed bulk-heterojunction (BHJ) organic solar cells represent the newest generation of technologies in solar power generation, offering benefits in terms of low manufacturing costs (i.e., high-throughput roll-to-roll processing), large area coverage, compatibility with flexible and lightweight substrates, earth-abundant constituents, and architectural tunability over multiple length scales.1 With concomitant research on material development and device structure optimization, BHJ solar cells have demonstrated a great potential in terms of power conversion efficiencies (PCEs).2 The major contribution to this result has been the development of new π-conjugated donor polymers with optimal chemical and physical properties. Knowledge on design and synthesis of π-conjugated polymers and deep understanding of the effect of the polymer’s chemical structures on optoelectrical properties are crucial to generate improved polymers for efficient BHJ solar cells. These devices typically consist of a photoactive layer, composed of an interpenetrating network of bicontinuous electron-donor polymer (D) and an electron-acceptor fullerene (A) domains, placed between a semitransparent anode, tindoped indium oxide (ITO), and a metal cathode. In order to effectively harvest the solar energy, optical absorption of the © XXXX American Chemical Society

active layer in polymer BHJ solar cells must be optimally matched with the region of maximum photon flux (i.e., 500− 800 nm).3 However, the absorption of sunlight is far from being the only parameter of importance. In fact, D:A self-organization and charge transport within the active layer are also key factors governing device performance.4 Through a rational materials development, a wide variety of p-type polymers were developed and employed as donor materials for BHJ solar cells.5 Among these, dithieno[3,2-b:2′,3′-d]silole (SiDT)-containing copolymers have attracted much attention due to their intrinsic advantages and potentialities in optoelectronic devices such as OLEDs,6 OFETs,7 and OPVs8 (Figure 1). Silicon-based monomers such as SiDT, analogous to their benzo-fused carbon-based counterparts, stiffen the polymer backbone, leading to a reduced conformational disorder and enhanced interchain π−π interactions as well as unique optical and charge transport properties in the solid state.9a Silole-based systems represent original systems in which the Si−C σ*-orbital effectively mixes with the π*-orbital of the butadiene fragment to afford a low-lying LUMO level and a relatively small Received: May 29, 2013 Revised: July 23, 2013

A

dx.doi.org/10.1021/ma4011186 | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules

Article

Figure 1. Chemical structures of some representative dithieno[3,2-b:2′,3′-d]silole (SiDT)-based copolymers.

bandgap.9b,c Additionally, silicon introduction also stabilizes the diene HOMO level compared to the carbon counterparts,10 which should, a priori, enhance the environmental stability and open-circuit voltages (VOC) of silole-polymer-derived BHJ OPVs. Furthermore, the tetravalence of the silicon atom, contained in the SiDT unit, offers two additional substitution sites for the incorporation of alkyl chains in order to enhance not only the solubility of the polymer but also the molecular packing and the interactions between the donor and acceptor materials.11 Recently, several groups have proven the potential of Sibased copolymers as donor materials for highly efficient BHJ solar cells. For instance, Bazan et al.12 obtained PCEs of 5.9%, with open-circuit voltage (VOC) of 0.57 V, short-circuit current density (JSC) of 17.3 mA/cm2, and fill factor (FF) of 61%, using new donor−acceptor (D−A) silole-containing copolymer (7, Figure 1) blended with PC71BM. A recent work of Tao et al.13, describing a new alternating copolymer (8, Figure 1) of dithienosilole and thienopyrrole-4,6-dione, afforded a remarkable PCE of 7.3%, with VOC = 0.88 V, JSC = 12.2 mA/cm2, and FF = 68%. These results prompted us to synthesize new silolecontaining conjugated polymers and to investigate their properties in BHJ solar cells. In this contribution, we report on the synthesis, characterization, and structure−property relationship of a series of novel p-type copolymers P(1)-2TSiDT, P(2)-2T-SiDT, P(3)-4T-SiDT, and P(4)-4T-SiDT (Scheme 1), which combine the dithienosilole (SiDT) moiety with dicyanated dithiophene (-2T-) and tetrathiophene (-4T-) comonomers. Functionalization of the oligothiophene core with strong electron-withdrawing cyano groups (H → CN) and core extension (2T → 4T) enable tuning of the copolymer HOMO and LUMO energy levels, π-electron density distribution, bandgap, carrier mobility, and free space generation while affecting film nanomorphology through polymer side-chain interdigitation.14,7a Particularly, the presence of the CN groups on the oligothiophene block where the

HOMO is localized should reduce HOMO energy, enabling greater open-circuit voltage compared to the unsubstituted systems. Solution-processed BHJ solar cells, using P(1)-2TSiDT, P(2)-2T-SiDT, P(3)-4T-SiDT, and P(4)-4T-SiDT as electron donor materials and PC61BM (or PC71BM) as electron acceptor counterpart, were fabricated, optimized, and fully characterized. The optical, electrical, morphological, and photophysical properties of the corresponding thin films are investigated to evaluate the relationship between molecular tuning, electronic structure, nanoscale morphology, kinetics of charge transfer processes, and OPVs performance for this class of polymers.

■

RESULTS AND DISCUSSION Synthesis and Thermal Properties. The present P(1)2T-SiDT, P(2)-2T-SiDT, P(3)-4T-SiDT, and P(4)-4T-SiDT copolymers (abbreviated as P(1), P(2), P(3), and P(4), respectively) were synthesized as shown in Scheme 1. The 4,4-bis(2-ethylhexyl)-2,6-bis(trimethyltin)dithieno[3,2-b:2′,3′d]silole (Sn-SiDT) monomer and dibrominated (unsubstituted or 3,3′-dicyano)oligothiophenes ((1)-2T-Br, (2)-2T-Br, (3)4T-Br, and (4)-4T-Br) comonomers were prepared and purified as described in the Supporting Information. Briefly, (1)-2T-Br and (4)-4T-Br were prepared according to known procedures (see Supporting Information). 3,3′-Dicyano-2,2′dithiophene was synthesized by dimerization of 3-cyanothiophene, which was then brominated in good yields to afford (2)2T-Br (Scheme 1B). The quaterthiophene (3)-4T-Br was prepared from the reaction of (2)-2T-Br with 2-trimethylstannyl-3-dodecylthiophene, followed by bromination with NBS. All copolymers were synthesized via Stille cross-coupling reaction in refluxing toluene using tris(dibenzylideneacetone)dipalladium(0) (Pd2(dba)3) and tri(o-tolyl)phosphine (Tol3P) as the catalyst. The resulting compounds were precipitated in methanol, collected by filtration, washed with methanol, and further purified by Soxhlet extraction to remove low molecular B

dx.doi.org/10.1021/ma4011186 | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules

Article

Scheme 1. Chemical Structure and Synthetic Route to (A) P(1)-2T-SiDT, P(2)-2T-SiDT, P(3)-4T-SiDT, and P(4)-4T-SiDT Polymers; (B) Cyanated Building Blocks

Table 1. Molecular Weights and Thermal, Optical, and Electrochemical Properties of P(1)−P(4) thin filmc

solutionb polymer

Mna (kDa)

Mwa (kDa)

PDI

Td (°C)

λmax (nm)

λonset (nm)

d Eopt gap (eV)

λmax (nm)

λonset (nm)

d Eopt gap (eV)

EHOMO (eV)

ELUMO (eV)

P(1)-2T-SiDT P(2)-2T-SiDT P(3)-4T-SiDT P(4)-4T-SiDT

8.6 4.6 6.7 6.4

15.9 10.8 15.0 12.5

1.8 2.3 2.3 2.0

>350 >380 >350 >360

514 550 542 529

610 685 708 636

2.03 1.81 1.75 1.95

530 586 588 599

655 717 723 726

1.89 1.73 1.71 1.71

−5.22 −5.63 −5.47 −5.52

−3.19 −3.82 −3.72 −3.57

a

Determined by GPC using polystyrene standards and 1,2,4-trichlorobenzene as eluent. bIn CHCl3. cSpin-coated from CHCl3 solutions on glass substrates. dEopt gap = 1240/λonset.

weight fractions. The remaining part was extracted with chloroform. The purified polymers P(1)−P(4) have average molecular weights (Mn), determined by high temperature GPC, ranging from 4.6 to 8.6 kDa with a polydispersity index (PDI) between 1.8 and 2.3, as summarized in Table 1. The resulting structures and purities are supported by 1H NMR spectra and elemental analysis (see the Supporting Information for details). Different batches of polymers have been prepared (and

characterized), and in all cases we obtained comparable and reproducible results. Polymers P(1) and P(2) are sufficiently soluble in chlorinated solvents for device fabrication (e.g., ∼10 mg/mL in chloroform) thanks to the presence of the two branched 2ethylhexyl alkyl chains linked to the silicon atom in the dithienosilole unit (Sn-SiDT). However, increasing the length of the thiophene-based comonomer, passing from P(1)−P(2) C

dx.doi.org/10.1021/ma4011186 | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules

Article

Figure 2. UV−vis absorption spectra of pristine polymers in (A) CHCl3 solutions and (B) thin films spin-coated from CHCl3. (C) Energy level diagram showing the HOMO and LUMO energies of polymers P(1)−P(4), estimated by cyclic voltammetry, and those of PC61BM. (D) Device structure of a conventional BHJ solar cell.

to P(3)−P(4), additional n-dodecyl chains have been introduced on the 3 or 4 thiophene spacer positions to improve the solubility even at room temperature. The thermal properties of P(1)−P(4) were investigated by both thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) at a temperature ramp rate of 10 °C/min under a nitrogen atmosphere (Figure S1). The TGA data (Table 1) indicate that all polymers are robust with decomposition temperatures (Td) above 350 °C, defined as the temperature at 5% mass loss. Notably, no significant thermal transitions were identified in the DSC scans up to ∼320 °C. Optical and Electrochemical Properties. The optical absorption spectra of the pristine polymers P(1)−P(4) in dilute chloroform solutions and as thin films are displayed in Figure 2. The detailed absorption data, including absorption maxima in solution and film, as well as the corresponding onset and bandgap values are summarized in Table 1. All compounds exhibit relatively broad absorption bands, which indicates that a significant part of the solar spectral flux is absorbed, contributing to photocurrent generation. The solution absorption spectra of polymers P(1)−P(4) (Figure 2A) exhibit a narrow variation of the λmax values, ranging from 514 to 550 nm. The λmax of P(2) is red-shifted by 36 nm versus that of the parent P(1). This effect is due to the presence of the −CN groups, which simultaneously lower both HOMO and LUMO energy levels and enhance T2 acceptor capabilities, resulting in a lower bandgap (Table 1). As expected, the electronic effect of the −CN substituents is attenuated in the larger T4 cores in P(3) and P(4), and the λmax red shift is reduced in these two polymers with respect to P(2). It is worth noting that despite the isomeric structures of P(3) and P(4), where the only difference is the position of the alkyl chains linked to the

thiophene spacers, the shapes of the corresponding absorption spectra are significantly different. In particular the broader spectrum of P(3) and the sharp shoulder peak at the longer wavelength region (∼650 nm) might suggest a more favorable backbone conformation, which could be responsible for a partial preaggregation of the polymer backbone in solution.15,6a The thin film spectra of P(1)−P(4) (Figure 2B) are significantly broadened (except P(3)) and red-shifted relative to the solution spectra. The resulting red shift magnitude film (Δλmax = λsol max − λmax) increases proceeding from P(1) to P(4) (Δλmax = 16, 36, 46, and 70 nm), probably due to the progressive enhancement of the interchain π−π interactions as a function of the substitution and length of the thiophene-based comonomers. With the exception of P(1), the thin film absorption spectra of the polymers P(2), P(3), and P(4) exhibit vibronic structures around the maximum of absorption, with a more pronounced shoulder at ∼650 nm, indicating a partial polymer chain ordering in the solid state. Note that P(3) exhibits the most intense shoulder relatively to the main absorption peak, suggesting stronger intermolecular interactions, in agreement with the polymer behavior in solution. The HOMO and LUMO energy levels of P(1)−P(4), estimated by cyclic voltammetry (CV, Figure S2) in combination with the optical absorption data,16 are reported in Table 1 and schematically depicted in Figure 2C. As shown by the cyclic voltammograms, the electrochemical oxidation onsets of P(1)−P(4) are located at +0.78, +1.19, +1.03, and +1.08 V, respectively. From the oxidation onsets, the HOMO energy levels of the present polymers were estimated using the relation EHOMO = −Eox(onset) − 4.44 (with the SCE energy level at −4.44 eV below the vacuum level).17 The estimated HOMO energies are −5.22, −5.63, −5.47, and −5.52 eV for P(1)− D

dx.doi.org/10.1021/ma4011186 | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules

Article

P(4), respectively. Note that these polymers are also expected to be more resistant to oxidation in the ground state in comparison to P3HT (HOMO: ∼−5.0 eV) because of their deeper HOMO levels.18 The reduction peaks of the cyclic voltammograms were not assigned since they were not clear; thus, the LUMO energy levels were calculated by the equation opt ELUMO = EHOMO + Eopt gap (Table 1, Egap were determined from the optical absorption spectra of pristine polymers in solution). As expected, the incorporation of electron-withdrawing −CN substituents on the thiophene rings effectively lowers both LUMO and HOMO energies. Indeed P(2), compared to P(1), exhibits a ΔELUMO (ELUMO P(1) − ELUMO P(2)) of 0.63 eV and ΔEHOMO (EHOMO P(1) − EHOMO P(2)) of 0.41 eV, resulting in a more stable and lower bandgap polymer (ΔEgap = Egap P(1) − Egap P(2) = 0.22 eV). A damping effect is expected for polymers P(3) and P(4) where the introduction of two electron-rich thiophene spacers should increase the π-electron density, thus attenuating the electronic effect of the −CN groups. Indeed, P(3) exhibits, compared to P(2), a ΔELUMO (ELUMO P(2) − ELUMO P(3)) of −0.10 eV, ΔEHOMO (EHOMO P(2) − EHOMO P(3)) of −0.16 eV, and a ΔEgap (Egap P(2) − Egap P(3)) of −0.06 eV. Similarly, P(4), compared to P(2), shows a ΔELUMO (ELUMO P(2) − ELUMO P(4)) of −0.25 eV, ΔEHOMO (EHOMO P(2) − EHOMO P(4)) of −0.11 eV, and a ΔEgap (Egap P(2) − Egap P(4)) of −0.14 eV. The LUMO levels for P(1)−P(4) were estimated to lie at −3.19, −3.82, −3.72, and −3.57 eV, respectively, yielding LUMO− LUMO offsets with the PC61BM acceptor of >0.3 eV, which should ensure efficient exciton dissociation.19 Furthermore, the electrochemically derived HOMO energies of P(1)−P(4) (Figure 2C) suggest that high open-circuit voltages (VOC)19 should be achievable when P(1)−P(4) are combined with PC61BM (ELUMO = −4.3 eV) or PC71BM in BHJ OPV devices. Photovoltaic Properties and Thin-Film Characterization. Bulk-Heterojunction Solar Cells. The potential of polymers P(1)−P(4) as donor (D) materials in OPV devices was investigated in bulk-heterojunction (BHJ) solar cells using PC61BM or PC71BM as acceptor (A) counterpart. The device structure employed, glass/ITO/PEDOT:PSS/active layer/LiF/ Al, is depicted in Figure 2D. The active blend was spin-coated from dry chloroform solutions (chloroform was found to be the optimum solvent), and the most efficient OPV devices were produced without additional thermal annealing (vide inf ra). Further details for the device fabrication and characterization are given in the Experimental Section. Table 2 and Table S1 summarize the photovoltaic response data including VOC, JSC, FF, and PCEs. OPV current density−voltage (J−V) plots, under standard illumination, for the solar cells based on P(1)− P(4):PC61BM (or PC71BM) active blends are shown in Figure 3 and Figure S3. A first optimization of the P(1)−P(4)-based BHJ solar cells was carried out by using PC61BM as acceptor material. The optimal D:A ratio was found to be 1:1 (w/w) for all polymers. Upon increasing (1.5:1) or decreasing (1:1.5) the donor content in the active blend, a significant drop in JSC, VOC, and FF was observed, resulting in lower PCEs. By comparing the performances of optimized OPV devices based on 1:1 (w/w) P(1)−P(4):PC61BM films (Figure 3A and Table 2), the main difference is represented by the short-circuit current density (JSC) values, which reflects the gap in the overall PCEs. Indeed, the most efficient solar cells based on P(1) and P(2) yield current densities of ∼2 mA/cm2 with a maximum PCEs of only 0.64% and 0.46%, respectively. BHJ OPVs based on P(3) and

Table 2. OPV Characteristics of the Most Representative Polymer:PCXXBM-Based BHJ Solar Cells polymer:PCXXBM ratio (w/w) P(1)-2T-SiDT:PC61BM P(1)-2T-SiDT:PC61BM P(1)-2T-SiDT:PC61BM P(1)-2T-SiDT:PC71BM P(2)-2T-SiDT:PC61BM P(2)-2T-SiDT:PC61BM P(2)-2T-SiDT:PC61BM P(2)-2T-SiDT:PC71BM P(3)-4T-SiDT:PC61BM P(3)-4T-SiDT:PC61BM P(3)-4T-SiDT:PC61BM P(3)-4T-SiDT:PC71BM P(3)-4T-SiDT:PC71BM P(3)-4T-SiDT:PC71BM P(3)-4T-SiDT:PC71BM P(4)-4T-SiDT:PC61BM P(4)-4T-SiDT:PC61BM P(4)-4T-SiDT:PC61BM P(4)-4T-SiDT:PC71BM P(4)-4T-SiDT:PC71BM P(4)-4T-SiDT:PC71BM P(4)-4T-SiDT:PC71BM

(1.5:1) (1:1) (1:1.5) (1:1) (1.5:1) (1:1) (1:1.5) (1:1) (1.5:1) (1:1) (1:1.5) (1:1) (1:1)a (1:1)a,b (1:1)a,b,c (1.5:1) (1:1) (1:1.5) (1:1) (1:1)a (1:1)a,b (1:1)a,b,c

VOC (mV)

JSC (mA/cm2)

FF (%)

PCE (%)

451 671 360 426 542 627 461 712 854 872 875 905 838 820 830 880 775 907 908 831 820 850

1.30 2.22 3.44 3.18 1.59 1.94 2.04 1.12 6.67 7.05 6.59 4.52 8.05 8.57 8.78 5.25 7.70 4.38 3.97 7.73 8.60 9.25

31 43 36 45 30 39 38 32 51 57 52 55 59 62 62 42 46 54 49 57 61 63

0.18 0.64 0.45 0.61 0.26 0.47 0.36 0.25 2.90 3.49 3.02 2.25 4.00 4.32 4.52 1.96 2.78 2.15 1.77 3.70 4.30 4.90

a

Additive: 2% (v/v) of 1,8-diiodooctane. b5 nm of MoO3 as anode buffer layer. cAlternative cathode: Ca (20 nm)/Al (80 nm).

P(4) afford JSC of ∼7 mA/cm2 with far greater PCEs of 3.49% and 2.78%, respectively. Since light harvesting (Figure 4A) and linear absorption coefficient spectra of the P(2)−P(4):PC61BM films (Figure S4) are similar, the higher JSC values observed for P(3)−P(4), relative to P(1)−P(2), can be mainly ascribed to the variations in terms of morphological (Figure 5), photophysical (Figure 6 and Table 3), and charge transport properties of the photoactive polymers (vide inf ra). The hole mobilities (μh) of the pristine polymer films (estimated by space charge limited current (SCLC)20 method) were found to be 1 order of magnitude higher for P(3) and P(4) (4 × 10−4 and 6 × 10−4 cm2 V−1 s−1, respectively) than P(1) and P(2) (∼10−5 cm2 V−1 s−1 for both polymers) as reported in Table S2 and Figure S5. Note that the charge mobilities calculated for P(3) and P(4) are in agreement with those of other highly efficient donor polymers.21,2b,6a For each polymer:PC61BM-based solar cell the average value av ), estimated from the open-circuit voltages of VOC (VOC observed at different D:A w/w ratios (Table 2), track the av av av progression Vav OC P(1) < VOC P(2) < VOC P(3) ≈ VOC P(4) av av av av rather than VOC P(1) < VOC P(3) ≈ VOC P(4) < VOC P(2) as expected from E(LUMOacceptor) − E(HOMOdonor) theoretical considerations19 (Table 1 and Figure 2C). The lower Vav OC of P(2), despite its low-lying HOMO level, could be ascribed to its poor film-forming properties, as confirmed by AFM images (Figure 5B), likely leading to poor exciton stabilization and dissociation, and low D:A phase segregation, all factors that are known to influence the VOC.22 P(3)- and P(4)-based solar cells, in agreement with the electrochemically derived E(LUMOacceptor) − E(HOMOdonor) offset and in combination with good film forming properties, show relatively high Vav OC (up to ∼850 mV). E

dx.doi.org/10.1021/ma4011186 | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules

Article

Figure 3. J−V plots, under illumination, of optimized as-cast BHJ solar cells based on (A) 1:1 (w/w) P(1)−P(4):PC61BM films; (B) 1:1 (w/w) P(1)−P(4):PC71BM films; (C, D) 1:1 (w/w) P(3):PC71BM and P(4):PC71BM films using progressively (i) 2% (v/v) of DIO, (ii) MoO3 as anode interfacial layer, and (iii) Ca/Al as cathode.

Figure 4. (A, B) UV−vis absorption spectra of (A) as-cast 1:1 (w/w) P(1)−P(4):PC61BM films spin-coated from CHCl3 solutions, (B) as-cast 1:1 (w/w) P(1)−P(4):PC71BM films spin-coated from CHCl3 solutions. (C, D) EQE plots of optimized as-cast BHJ solar cells based on (C) 1:1 (w/w) P(1)−P(4):PC61BM films and (D) 1:1 (w/w) P(3)−P(4):PC71BM (1:1) films using 2% (v/v) of DIO, MoO3 as anode interfacial layer, and Ca/Al as cathode.

Several unsuccessful attempts to optimize the thin-film nanomorphology of the 1:1 (w/w) P(1)−P(4):PC61BM device

were carried out (Table S1), for instance by using different solvents (e.g., chlorobenzene or o-dicholobenzene) and thermal F

dx.doi.org/10.1021/ma4011186 | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules

Article

active blends, in agreement with the observed low photocurrents. The incorporation of additives to the blend solvent, such as 1,8-diiodooctane (DIO), is a widely used method to promote an optimal self-organization of the BHJ components.25,24a Indeed, we found that the addition of 2% (v/v) DIO to P(3)− P(4):PC71BM blend solutions led to improved film morphologies, as discussed in the morphological section, and greatly enhanced PCEs (Table 2 and Figure 3C,D). P(3)-based BHJ OPVs show VOC, JSC, FF, and PCE of 838 mV, 8.05 mA/cm2, 59%, and 4.00%, respectively. Similarly, P(4)-based solar cells yield a PCE of 3.70% with VOC, JSC, and FF of 831 mV, 7.73 mA/cm2, and 57%, respectively. For both polymers, the mayor PCE improvement is the result of a doubled JSC, indicative of well-ordered nanoscale morphology of the active layer, which correlates well with the simultaneous increase of FF and charge mobilities. Indeed, the hole mobilities (μh) (estimated by SCLC model) of optimized (1:1 w/w) P(3)−P(4):PC71BM films processed with DIO (2% v/v) (Table S2 and Figure S6C,D) were found to be 1.1 × 10−4 and 2.0 × 10−4 cm2 V−1 s−1, respectively, very similar to those measured for the pristine polymers. In comparison, the μh of (1:1 w/w) P(3)− P(4):PC71BM-based devices processed without DIO (Table S2 and Figure S6A,B) were found to be 7.2 × 10−5 and 8.1 × 10−5 cm2 V−1 s−1, respectively, in agreement with abovementioned film morphological features. Further improvements in terms of PCEs, up to ∼5%, for (1:1 w/w) P(3)−P(4):PC71BM-based BHJ OPVs were obtained by replacing conventional anode and cathode interfacial layers (PEDOT:PSS and LiF) with thin films of molybdenum oxide (MoO3, 5 nm) and calcium (Ca, 20 nm), respectively. We found that the introduction of MoO3 (5 nm) mainly increases the FF and JSC values. Enhanced FF, passing from 59% to 62% and from 57% to 61% for P(3)- and P(4)-based devices, respectively, could be ascribed to the quasi-ohmic contact between the HOMO level of the polymer and the conduction band edge of MoO3 (∼5.4 eV),26 favoring hole extraction.27 Moreover, MoO3 interlayer seems to also promote a more favorable redistribution of the light intensity within the active layer, as a result of the refractive index matching between MoO3 and active blend,26 in agreement with the enhanced JSC measured for P(3)- and P(4)-based cells (from 8.02 to 8.57 mA/cm2 and from 7.73 to 8.60 mA/cm2, respectively). Additionally, the use of Ca/Al as cathode generates higher VOC and JSC values as a result of its lower work function, compared to LiF/Al, which primarily increases the built-in electric field of the cell, also responsible for optimal charge extraction and collection processes.28 As a result, the most efficient P(3)-based solar cell exhibits VOC = 830 mV, JSC = 8.78 mA/cm2, and FF = 62%, with PCE = 4.52%. Similarly, the P(4)-based solar cell shows VOC = 850 mV, JSC = 9.25 mA/cm2, and FF = 63%, with PCE = 4.90%. The spectral responses of these devices enhance our understanding of the different photocurrent generation efficiencies. Figures 4C and 4D show the external quantum efficiency (EQE) spectra of best devices based on 1:1 (w/w) P(1)−P(4):PC61BM and P(3)−P(4):PC71BM films. The EQE plots are consistent with the broad optical absorption spectra of the active blends (Figure 4A,B). EQE spectra of 1:1 (w/w) P(1)−P(4):PC61BM (Figure 5C) exhibit the highest generatedelectrons over incident-photons ratio, comprised within 10% and 35%, in the spectral range between 450 and 600 nm, corresponding to the lower energy absorption band of the

Figure 5. Tapping mode AFM images (size: 5 μm × 5 μm) of optimized as-cast 1:1 (w/w) P(1)−P(4):PCXXBM: (A) P(1):PC61BM (RMS = 0.7 nm); (B) P(2):PC61BM (RMS = 0.8 nm); (C) P(3):PC61BM (RMS = 1.9 nm); (D) P(4):PC61BM (RMS = 1.0 nm); (E) P(3):PC71BM (RMS = 1.2 nm); (F) P(4):PC71BM (RMS = 1.2 nm); (G) P(3):PC71BM processed with 2% (v/v) of DIO (RMS = 3.8 nm); (H) P(4):PC71BM processed with 2% (v/v) of DIO (RMS = 2.7 nm).

annealing (in the range 80−140 °C and for different times). Therefore, alternative approaches to enhance P(1)−P(4) solar cell performace included the use of (i) PC71BM as acceptor material, (ii) 1,8-diodooctane (DIO) as processing solvent additive, and (iii) molybdenum oxide (MoO3) and calcium (Ca) as alternative anode and cathode interfacial layers (IFLs). First, by replacing the acceptor PC61BM with PC71BM, a new set of P(1)−P(4):PC71BM-based devices, by using the best processing conditions found for PC61BM, were prepared in order to enhance the blend film absorption in the spectral range between 350 and 550 nm (Figure 4B).23 The resulting OPV performance are reported in Table 1 and Figure 3B. Despite the improved overall light harvesting of the composite blends, the resulting BHJ OPVs exhibit lower PCEs when compared to the analogous PC61BM-based devices. This result may reflect suboptimal film morphologies as observed by AFM investigation (Figure 5), likely due to the poorer miscibility of PC71BM with our polymer donors,24 thus limiting the D:A phase segregation, charge generation, and transport within the G

dx.doi.org/10.1021/ma4011186 | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules

Article

Figure 6. PL spectra of pristine polymer and blend films, spin-coated from CHCl3 on glass, of (A) P(1) and 1:1 (w/w) P(1):PC61BM; (B) P(2) and 1:1 (w/w) P(2):PC61BM; (C) P(3), 1:1 (w/w) P(3):PC61BM and 1:1 (w/w) P(3):PC71BM processed with 2% (v/v) of DIO; and (D) P(4), 1:1 (w/w) P(4):PC61BM and P(4):PC71BM processed with 2% (v/v) of DIO. Laser excitation: λmax = 450 nm. Insets: unquenched PL decays of pristine polymer films and quenched PL decays of blend films. Decays are all measured at the λ maximum of emission of the irradiated film.

Table 3. Time-Resolved Photoluminescence (TRPL) and PL Quenching Ratio Measurements of P(1)−P(4)-Based Films

a

polymera

PL lifetime (t1p, t2p) (ps)(b); weights (W1p, W2p) (%)

quenching ratio (Q)

P(1)-2T-SiDT P(2)-2T-SiDT P(3)-4T-SiDT P(4)-4T-SiDT P(1)-2T-SiDT:PC61BM P(2)-2T-SiDT:PC61BM P(3)-4T-SiDT:PC61BM P(4)-4T-SiDT:PC61BM P(3)-4T-SiDT:PC71BM + 2% DIO P(4)-4T-SiDT:PC71BM + 2% DIO

t1p = 59 ± 15 (6%), t2p = 12 ± 1 (94%) t1p = 304 ± 9 (29%), t2p = 44 ± 2 (71%) t1p = 374 ± 3 (68%), t2p = 42 ± 3 (32%) t1p = 324 ± 8 (63%), t2p = 39 ± 3 (37%) tb = 5.4 ± 0.2 (100%) t1b = 183 ± 31 (17%), t2b = 20 ± 1 (83%) t1b = 67 ± 3 (7%), t2b = 13 ± 1 (93%) t1b = 75 ± 1 (15%), t2b = 17 ± 1 (85%) t1b = 92 ± 7 (11%), t2b = 16 ± 1 (89%) t1b = 74 ± 5 (29%), t2b = 18 ± 1 (71%)

0.67 0.84 0.94 0.92 0.91 0.84

Films spin-coated on PEDOT:PSS from CHCl3 solutions.

a deeper understanding on how the molecular structure of polymers P(1)−P(4) affects the solar cell output parameters, we investigated the morphological differences of P(1)− P(4):PCXXBM blend films by tapping-mode AFM. Figure 5 shows topographic images of representative as-cast 1:1 (w/w) P(1)−P(4):PC61BM (or PC71BM) films processed with and without additive. The topographic images of P(1)- and P(2)based films (Figure 5A,B) exhibit a smooth and almost featureless surfaces with large and poorly defined domains, with unoptimal phase segregation of the BHJ components. Thus, poorly formed percolation pathways lead to limited charge transport and/or collection processes within the active blends, as confirmed by the low JSC, FF, and PCEs obtained from the corresponding BHJ OPVs. On the other hand, the surface morphology of P(3)−P(4):PC61BM films is significantly different (Figure 5C,D) and characterized by drastically reduced domains and finer nanostructures likely due to an enhanced polymer miscibility which promotes the intermixing,

polymers. Figure 4D shows the EQE responses of the most efficient P(3)−P(4):PC71BM-based devices, in which the profiles are greatly enhanced in the whole wavelength range (350−700 nm) with maxima of 46% and 51% (at ∼560 nm) respectively for P(3)- and P(4)-based OPVs, as expected from the high photocurrents measured for these devices. Convolution of these EQE spectra with the 1.5AM solar spectrum gave calculated short circuit current densities in good agreement, within the ∼5% experimental error, with those obtained from J−V measurements. Finally, because of the relatively large bandgap (∼1.7 eV) of P3 and P(4) the photocurrent is generated in the spectral range below 700 nm, as shown in the EQE spectra. Therefore, it is worth noting that this class of polymers can be also suitable for a bottom cell in a solution-processable tandem solar cell.29 Atomic Force Microscopy (AFM). It is known that the nanomorphology of the BHJ active layer plays an important role in determining the OPV performance.30,22 In order to gain H

dx.doi.org/10.1021/ma4011186 | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules

Article

significantly the PL lifetime. In particular, going from P(2) to P(3)−P(4), independently of the alkyl chain position, the longer component of the PL decay (at ∼300−350 ps) becomes predominant in the biexponential fitting expression. Since for some polymers longer PL decay lifetimes are attributed to higher crystallinity,34 an overall enhanced crystalline phase of P(3) and P(4) films compared to P(1) and P(2) can be expected. This would be in agreement with the higher charge transport properties (Table S2) and the increased nanostructuration of the surface morphology of the corresponding active blends, observed by AFM (Figure 5A,D), passing from P(1)−P(2)- to P(3)−P(4)-based films. For the polymer:fullerene blends in which PC61BM was used as the acceptor, most of the light is absorbed by the polymer and the contribution of PC61BM to absorption is negligible. Since also the emission quantum yield of fullerene derivatives is usually much lower (