High Voc Fullerene-Free Organic Photovoltaics Composed of PTB7 and Axially-Substituted Silicon Phthalocyanines

While the eciency of organic photovoltaics (OPV) has improved drastically in the past decade, such devices rely on exorbitantly expensive materials that are unfeasible for commercial applications. Moreover, examples of high voltage single-junction devices, which are necessary for several applications, particularly low-power electronics and rechargeable batteries, are lacking in literature. Alternatively, silicon phthalocyanines (R 2 -SiPc) are inexpensive, industrially scalable organic semiconductors, having a minimal synthetic complexity (SC) index, and are capable of producing high voltages when used as acceptors in OPVs. In the present work, we have developed high voltage OPVs composed of Poly({4,8-bis[(2-ethylhexyl)oxy]benzo[1,2-b:4,5-b′]dithiophene-2,6-diyl}{3-uoro-2-[(2-ethylhexyl)carbonyl] thieno [3,4 b]thiophenediyl}) (PTB7) and an SiPc derivative ((3BS) 2 -SiPc). Interestingly, while changes to the solvent system had a strong effect on performance, the PTB7:3BS-SiPc active layer were robust to spin speed, annealing and components ratio. This invariance is a desirable characteristic for industrial production. All PTB7:(3BS) 2 -SiPc devices produced high open circuit voltages between 1.0 and 1.07 V, while maintaining 80% of the overall eciency, when compared to their fullerene-based counterpart.


Introduction
Organic photovoltaics (OPVs) are a promising solar energy technology with the potential for low manufacturing cost and quick energy payback. 1,2 OPVs can be fabricated by solution-based techniques, such as spin-coating, blade-coating and a variety of printing techniques, facilitating their integration into continuous, high throughput processing, which is unachievable with traditional silicon technology. 1,3−5 Recently, single-junction OPVs with record power conversion e ciencies (PCE) of 15-19% have been reported, 6-9 approaching commercial solar cell performances. However, these high e ciencies are achieved using expensive small molecules and polymers, the production of which is not scalable as they require complex multi-step synthesis and puri cation methods. [10][11][12][13][14] Most of these state-of-the-art OPVs often excel in high current densities (Jsc) and ll factor (FF), while providing middling open-circuit voltage (Voc), between 0.7 and 0.8. [6][7][8][9] High-voltage OPVs are of great interest for certain applications, particularly for rechargeable batteries and low-power electronics, which always require a minimum voltage to operate. [15][16][17] While high Vocs can be obtained by tandem cells or cells in series, this requires all cells to produce similar currents, which is virtually impossible to achieve in applications where lighting is inhomogeneous. 15 The Voc of the majority of single-junction photovoltaics rarely surpasses 0.8 V, including silicon-based devices. Recently, a few research groups have used simple and low-cost small molecules as non-fullerene acceptors (NFAs) in OPVs, and achieved V OC ≥ 1.0 V. 12,18−21 Such architectures may hold the key for commercial viability as they can be manufactured on in an industrial scale, provided that the e ciency of the devices is su ciently high. 18 Axially substituted silicon phthalocyanines (R 2 -SiPc) are ideal candidates for low-cost, high-VOC acceptor materials. [21][22][23] While metal phthalocyanine (MPc) have been investigated in organic electronic applications for more than 50 year, R 2 -SiPcs are relatively understudied, having emerged in recent years 24 and successfully incorporated in multiple new application, including organic thin-lm transistors (OTFTs), [25][26][27][28][29][30] organic light-emitting diodes (OLEDs) [31][32][33] and in OPVs 21,31,34−37 . The synthetic complexity (SC) index 38 of R 2 -SiPcs have been calculated to be at least three times lower (SC = 12) 23 than that of several prominent OPV acceptors materials, such as PC 61 BM (SC = 36) 39 , Y6 (SC = 59) 40 and ITIC (SC = 67) 41 . The exceptionally low SC index of R 2 -SiPcs makes them exceedingly promising organic semiconductors for OPVs. Historically, R 2 -SiPcs have mainly been employed as ternary additives in OPVs. [34][35][36][37]42 However, in a recent study by Grant et. al 21 There are relatively few reports investigating the use of R 2 -SiPc as stand-alone acceptors in OPVs. [21][22][23] therefore it is vital to investigate pairing these cost-effective molecules with different donor polymers and optimizing these devices to exploit their full potential in OPVs.
In the present work we have fabricated OPVs using poly [[4,8-

Devices
Indium-tin-oxide (ITO) coated glass substrates (100 nm, 20 Ω/sq, 1 in by 1 in), purchased from Thin Film Devices Inc., were cleaned in an ultrasound bath sequentially with soapy water, DI water, acetone (99 %) and methanol (95 %) to remove any debris. The ITO slides were then dried with a N 2 jet and placed in an air plasma cleaner for 15 min to remove any residual organics. The zinc oxide (ZnO) electron-transport layer was deposited by spin coating 150 µl of an ethanolic solution of Zn(Ac) 2 • 2 H 2 O (3.3 %) and ethanolamine (0.9 %) at 2000 RPM, followed by annealing for 1 h in air at 180°C. The substrates were then move into a N 2 glovebox where the remainder of the procedure was carried out. Preparation of the active layer was achieved using a variety of conditions, details of which are provided in Table 1. The active layer components are illustrated in Fig. 1a. After deposition, the lms were dried in the glovebox, at room temperature, for 1h before being transferred to an evaporation chamber (Angstrom EvoVac), where MoO 3 (7 nm) and silver electrodes (70 nm) were deposited by physical vapor deposition at a pressure below 10 − 6 torr, to yield 5 individual 0.32 cm 2 devices per substrate, as de ned by shadow masks. Energy level diagram and device architecture are shown in Fig. 1b and 1c, respectively. Device performance was characterized using a custom push pin probe station connected to multiplexer and Keithley 2400 source meter. All OPVs were assessed under 1000 W m − 2 light intensity, provided by a solar simulator (Xenon lamp, AM1.5) and scanned between − 2.0 and 2.0 V. Light intensity was veri ed using an NREL certi ed silicon standard cell prior to every run. Series and shunt resistances have been calculated from the slope of the IV curves, at the Jsc and Voc, respectively. External quantum e ciency (EQE) plots were recorded using a Newport Quantx-300 instrument outside of the glovebox. Prior to EQE measurements, the devices were encapsulated in an epoxy resin (Norland NOA61) cured under UV-light. Atomic force microscopy (AFM) topography images of the active layer lms were obtained using a Bruker Dimension Icon instrument, with ScanAsyst-Air probes in tapping mode, at a frequency of 0.8 Hz; image processing was performed with NanoScope Analysis v1.8.

Results And Discussion
Bulk heterojunction (BHJ) OPV devices ( Figure 1c) were fabricated by combining PTB7 (donor) with (3BS) 2 -SiPc (acceptor, Figure 1a). The energy levels of PTB7, 3BS-SiPc and the other materials in the BHJ OPV devices are shown in Figure 1b. Extensive optimization of the active layer was performed, as shown in Table 1. The optimized parameters were spin-rate, annealing, donor:acceptor ratio and choice of solvent. Parameters including spin-rate, annealing time and temperature, donor:acceptor ratio, and choice of solvent were all investigated.   The best PTB7:(3BS) 2 -SiPc BHJ OPV device (12) was obtained with an excess of (3BS) 2 - SiPc (1:1.8 ratio) in CB, at a spin rate of 1500 RPM, resulting in a high V OC of 1.05 ± 0.01V, a modest J SC of 7.68 ± 0.07 mA·cm -2 , ll factor (FF) of 0.48 ± 0.01 and an overall PCE of 3.82 ± 0.04 ( annealing also showed a weak effect on device PCE (devices 6 to 9), slightly increasing FF while decreasing Jsc. Figure 2 shows AFM images before ( Figure 2a) and after annealing for 15 min at 100 °C ( Figure 2b). Comparison of the lms at various length scales shows that some larger amorphous features are formed during the annealing step, but in general overall surface morphology and height features are mostly retained. Longer annealing times were slightly detrimental to device performance, most likely due to disruption of the active layer through formation of large agglomerates, as suggested in Figure 2b. The (3BS) 2 -SiPc to PTB7 ratio (devices 10 to 13) only had a noticeable effect on the PCE when the ratio was reduced below 1.5, which is observed when comparing devices 10 and 11, with ratios of 1.0 and 1.5, respectively. Alternatively, devices 11, 12 and 13, with ratios of 1.5, 1.8 and 2.0, respectively, have remarkably similar performances. While PTB7:PC 61 BM devices have been optimized at a 1:1.5 ratio, 45 the PTB7:(3BS) 2 -SiPc BHJ OPV devices had the best performance when using a 1:1.8 ratio (Table 1). In summary, we found that the (3BS) 2 -SiPc:PTB7 blend is fairly invariant to spin rate, acceptor:donor ratio and annealing conditions. This low variability is a desirable property for high throughput manufacture, where minor variations are inevitable.
Alternatively, modi cation of the active layer solvent system had a signi cant impact on device performance. Solvent additives, namely DIO and DPE, have been previously reported to play a critical role in the lm morphology of the active layer in PTB7:PC 61 BM devices, resulting in signi cant improvements in performance. 46 Such high boiling point additives promote higher crystallinity and improved nanomorphology in PTB7:fullerene blends, 46 typically resulting in increases of more than 40% in PCE. We have observed these improvements as well in our baseline PTB7:PC 61 BM baselines (1 and 2, Table 1).
However, when incorporating these additives in the fabrication of PTB7:(3BS) 2 -SiPc BHJ OPV devices (device 18), we were unable to obtain functioning devices. This may be attributed to (3BS) 2 -SiPc, which has been reported to crystallize rapidly, [47][48][49] and the use of high boiling point additives thereby exacerbating this behaviour, creating micrometric or even submillimetric domains (patterns are visible to the naked eye) as opposed to the nanometric phase separation required for functioning OPVs. 50,51 In attempt to counter this crystallization we explored the use of low-boiling point solvent additives, such as DCM and CF (devices 15 and 16) but these solvent changes led to negligible improvements in performance. Using CF as a single low-boiling point solvent (device 17) resulted in thicker lms due to rapid evaporation, which approximately halved the PCE when compared to the best devices deposited with solutions in CB. We have also investigated if incorporating a low a low-M w PTB7 (16 kDa) instead of the conventional PTB7 (M w = 96 Kda) could improve the crystallinity of the P7B7 phase without additives. Functioning devices were obtained (19 and 20), but with lower e ciency compared to the optimized device achieved from low-M w PTB7 (12).
While the Jsc and FF of the PTB7:(3BS) 2 -SiPc BHJ OPV devices were modest, it is important to note the consistently high V OC between 1.00 -1.07 V, which is a very desirable and often rare characteristic in OPVs. Note that current state-of-the-art OPVs typically display Voc values around 0.8V. 7-9 The high voltages can be attributed to the large difference between the energy levels of the donor's highest occupied molecular orbital (HOMO) and the acceptor's lowest unoccupied molecular orbital (LUMO), as illustrated in Figure 1b. Table 2 compares the series and shunt resistances of our optimized device 12 with literature-based devices containing either PTB7 or P3HT as the donor and PC 61 BM or (3BS) 2 -SiPc as the acceptor.
Strikingly, the shunt resistance of the PTB7:(3BS) 2 -SiPc BHJ OPV device is signi cantly lower than the others, which indicates a high rate of charge recombination in the active layer lm and may offer a potential explanation for the relatively low Jsc compared to PTB7:PC 61 BM devices (Table 1 and 2). This may be ascribed to the small energy level offset only 0.2 eV between the donor and acceptor, which could be impairing the dissociation of excitons at the PTB7:(3BS) 2 -SiPc interface. 1 Moreover, AFM images ( Figure 2) show that domain sizes are in the hundreds of nm range, which is often too large for optimal BHJ OPV operation, given the average distance travelled by excitons before recombination is around 5 -15 nm. 1,51,52 Table 2. Thickness, series and shunt resistances of the optimized devices and relevant comparative devices.  SiPc provides some extra light absorption around 700 nm, the overall quantum e ciency is lower than that of the fullerene-based device. This trade off is often seen in SiPc-based devices. 21 Moreover, the EQE spectrum of device 12 is slightly blue-shifted and sharper in comparison to the baseline (1), which is associated with smaller crystallinity of the polymeric phase. 46,53 This suggests that (3BS) 2 -SiPc inhibits the crystallization of PTB7. We surmise that a new type of additive, that simultaneously promotes the crystallization of PTB7 while keeping the SiPc domain size small, could push the e ciency of this class of devices beyond that of the PTB7:PC 61 BM baseline, but the authors are not familiar with any additive that ts these requirements.

Conclusion
We paired PTB7, a high performing donor polymer, with a low cost and easy to synthesize acceptor (3BS) 2 -SiPc in OPVs and observed a signi cant improvement in V OC values. Device performance was robust to changes in the spin speed, acceptor:donor ratio and annealing; although this hinders a route towards device optimization, it is ultimately a desirable property for high throughput fabrication of OPVs. When replacing the fullerene acceptor with (3BS) 2 -SiPc, 80% of the overall device e ciency was retained, while a high average Voc of 1.05 V was otained. These ndings further establish SiPc-based acceptors as promising NFA candidates for high voltage OPV devices. Control of the crystallization of the SiPc will be key to yield the desired nanomorphology in the active layer will be key in the development of high   OPV characteristics a) I-V curves, where line thickness corresponds to the standard deviation of 4 devices, and b) EQE spectra of the optimized device (12) and the fullerene-containing baseline (1)