Fully Printed and Industrially Scalable Semitransparent Organic Photovoltaic Modules: Navigating through Material and Processing Constraints

While the power conversion efficiency (PCE) of organic photovoltaics (OPV) on small‐area lab cells has rapidly increased during the last few years, the performance on module level and the availability of OPV modules on the market is still limited, primarily due to specific constraints imposed by the industrial production process. This work deals with the upscaling process of latest‐generation OPV from small‐area lab cells to fully solution‐processed modules, which are compatible to industrial roll‐to‐roll (R2R) printing. This transfer is demonstrated step by step from material selection and process optimization for every single layer of the stack (photoactive layer, charge transporting layers, and solution‐processed top electrode)–including long‐term stability investigations (thermal and light)–to scaling up the device area by a factor of >100. Thus, a semitransparent OPV module with 10.8% PCE on 10.2 cm2 active area is achieved, which is among the highest performances for semitransparent, fully solution‐processed OPV modules. The individual developments all meet the requirements for industrial R2R printing (green solvents, processing in air, annealing ≤140 °C, etc.), which ensures that both the optimized layer stack and the fabrication process are fully scalable and easily transferable to large‐scale production.

While the power conversion efficiency (PCE) of organic photovoltaics (OPV) on small-area lab cells has rapidly increased during the last few years, the performance on module level and the availability of OPV modules on the market is still limited, primarily due to specific constraints imposed by the industrial production process.This work deals with the upscaling process of latest-generation OPV from small-area lab cells to fully solution-processed modules, which are compatible to industrial roll-to-roll (R2R) printing.This transfer is demonstrated step by step from material selection and process optimization for every single layer of the stack (photoactive layer, charge transporting layers, and solution-processed top electrode)-including long-term stability investigations (thermal and light)-to scaling up the device area by a factor of >100.Thus, a semitransparent OPV module with 10.8% PCE on 10.2 cm 2 active area is achieved, which is among the highest performances for semitransparent, fully solution-processed OPV modules.The individual developments all meet the requirements for industrial R2R printing (green solvents, processing in air, annealing ≤140 °C, etc.), which ensures that both the optimized layer stack and the fabrication process are fully scalable and easily transferable to large-scale production.
As the research focus is further directed toward scalable printed devices, ready for production, a new set of requirements emerges, defining new processing constraints.Among those requirements are the transfer to larger active device areas and the compatibility with continuous deposition processes as well as the connection of cells to form modules. [16,[19][20][21][22] Each of such processing constrains limits the range of parameters with which the device performance can be maximized.Thus, with the set of constraints involved in large-area roll-to-roll (R2R) printing, the performance of scalable devices can be significantly reduced compared to those achieved at laboratory scale.Many research efforts have already been made to address problems related to specific constraints and achieving promising performance.Among the most important requirements are processing with nonchlorinated solvents, film homogeneity, absence of defects, and processability in air from solution-including a printed top electrode.25][26] Furthermore, solution-processed hole-transport layers (HTLs) were recently reported, which can be coated on top of the AL within the inverted (n-i-p) device structure with similar performance compared to vacuum-processed molybdenum oxide. [27,28]The next step is to enhance the compatibility to a printed top electrode, which can vary depending on the application.29][30][31][32][33][34][35][36][37] However, in many cases these constraints and challenges have been studied and addressed individually.In this work, we combine all challenges to develop and present a new state-of-the-art that is close to industrial application by fabricating fully printed, industrially scalable, semitransparent devices with the latest generation of AL material.
With the aim of finding the best material system and layer stack that provide both high efficiency and good stability, in this work, we evaluate the most promising material candidates for electron-transport layer (ETL), AL, HTL, and top electrode on the basis of in-air doctor bladed cells and industrially feasible processes.Once the best system is found, we make in-depth lifetime studies regarding thermal and photostability.Finally, we transfer the system from cell to module level and increase the device area by two orders of magnitude to demonstrate a highly efficient semitransparent OPV module that can be easily scaled up industrially.

Development of a Scalable Multilayer Structure Processed from Solution
For this optimization, the inverted (n-i-p) device architecture (cf., Figure 1) is chosen, which is until now further developed toward industrial printing with higher stability being one of its current advantages. [16,29,38]

ETL: Reducing Thermal Treatment While Maintaining Long-Term Performance
[41] In the following, we evaluate nanoparticle inks of the two most prominent materials-zinc oxide and tin oxide (SnO 2 ).
Too rough bottom electrodes and the edges of P1 laser lines-with which the bottom electrode is structured into separate parts-can penetrate through the layer stack and induce shunts in the device.This can be circumvented, to a certain extent, by increasing the film thickness of the ETL.However, using a thicker layer can potentially limit the charge transport properties.For metal oxides, such as ZnO, a thermal treatment after layer deposition can significantly improve these properties and increase the stability. [42]However, taking an industrial scaling process into account, two constraints appear: first, polyethylene terephthalate (PET)-based substrates are the industrial state-ofthe-art for R2R-fabricated flexible OPV, and even for heatstabilized PET, processing temperatures are limited to 140 °C.Second, applying heat in a running production line implies a time limit, depending on the length of the heating unit and the targeted throughput speed, which allows only few minutes of thermal treatment.As such, we have studied the usability of materials for two cases of thermal treatment: a common lab process optimized for high performance (30 min at 200 °C) and the scalable process, which can be transferred to large-scale production (2 min at 120 °C).
Taking these processing constraints regarding temperature treatment into account, and considering additionally the effect of ETL thickness, we tested the long-term stability of PM7:BTP-4F-12-based cells under light exposure for layers with a thickness of 20 nm (thin layers) and 40 nm (thicker layers).The temporal evolution of the PCE is presented in Figure 2. The initial performance values and the evolution of the individual cell parameters can be found in Table S1 and Figure S2, Supporting Information, respectively.It can be observed that the light stability of the devices is high for both metal oxides (ZnO and SnO 2 ), if strong thermal treatment (200 °C, 30 min) is applied (Figure 2a).In this case, the ETL thickness also has no big influence on the device performance after 250 h of light degradation.However, if only 2 min at 120 °C of thermal treatment is applied to the ETL (see Figure 2b), only the ZnO layer shows a good light stability, whereas the SnO 2 -based devices lose performance very rapidly (mainly due to losses in J SC and V OC ), which is even more pronounced for thicker layers.This suggests that SnO 2 generally needs higher annealing temperatures to allow for stable devices, and that thicker ETLs further enhance the need for higher annealing times and/or temperatures.Consequently, a thin ZnO layer is chosen as the most suitable ETL for upscaling.Furthermore, the photocatalytic effect, which has been reported to cause instabilities in NFA-based devices using ZnO as ETL, [43][44][45][46][47] is circumvented by application of a UV filter (spectrum shown in Figure S1, Supporting Information), which absorbs sufficient light below 380 nm during light exposure to prevent the effect.

Photoactive Layer: Balancing Layer Thickness and Performance
When focusing on compatibility of the photoactive layer to large-scale manufacturing of semitransparent devices, apart from performance and stability, further aspects require consideration: processability with nonchlorinated ("green") solvents, lowest possible material cost, and film thickness dependence.
One of the most prominent new generation active layer systems is PM6:Y6.It is commonly dissolved in chloroform, which-being a chlorinated solvent-implies greater environmental concerns.A processing constraint of Y6 has been its preaggregation in certain solvents, including the nonchlorinated solvent o-xylene, leading to large Y6 domains in the film and thus to reduced performance. [48,49]However, this challenge has been overcome by modifying the side chains of Y6. [23][24][25][26] One of those tailored materials is Y6-C12 (BTP-4F-12), which enables favorable processability in o-xylene and thus provides excellent device performances. [23]58][59][60] Finally, R2R processing of OPV requires a certain minimum active layer thickness, first to prevent the formation of shuntscaused by spikes from layers underneath or from contact-based printing processes of the top layers (e.g., screen printing of the top electrode), and second to increase the amount of directly absorbed light in semitransparent devices, where no opaque back-reflecting layer can be used to allow the light to pass through the active layer a second time, which has a huge impact on the produced photocurrent and therefore greatly improves device performance.
As increasing the AL film thickness (in the range above 100 nm toward 300 nm) is often accompanied by a fill factor (FF) loss, an experiment is carried out in which three different donor materials (PM7, PM6, and PTQ10) are blended with the acceptor BTP-4F-12 and the effect of active layer thickness on the solar cell performance is investigated.As can be seen in Figure 3, PM6 shows a high PCE with a thin layer thickness of 115 nm, but features a linear decrease in efficiency with increasing film thickness, mainly due to a decreasing FF, which is also observed for PM7.][63][64] This underlines its suitability for upscaling-PTQ10 is able to form a high-performing device within a broad thickness range and is a cost-effective material-and was thus selected as active donor material of choice for fully solution-processed, semitransparent devices.

HTL: Energetic and Chemical Compatibility with AL and Top Electrode
A critical role for achieving scalable, solution-processed, semitransparent, highly efficient solar cells plays the HTL.The new, high performance active layer systems with their deep energy levels require an HTL with a high work function.So far, PEDOT:PSS is the industrial state-of-the art for inverted OPV modules; however, when coated on top of the AL, its incompatibility with novel nonfullerene acceptors (NFAs) and low-highest occupied molecular orbital (HOMO)-level polymers, which is mainly ascribed to its mismatching energy level, results in significant performance loss, especially in open-circuit voltage (V OC ). [33,65,66]69][70][71][72][73] Many attempts have been made to find energetically compatible HTLs for new NFA-based high-performance active layer systems, of which one common strategy has been the use of metal oxides in combination with PEDOT:PSS. [65,74,75]Recently, the problem could be solved by synthesizing a new PEDOT-based material system that uses a per-fluorinated ionomer (PFI) instead of PSS as counter ion. [27,28]We employ one of these materials, PEDOT:F (where "F" stands for the PFI), and test its potential for scalable devices.Additionally, we test a mixture of standard PEDOT:PSS and PFI with the aim of tuning the HTL work function by varying the PFI-to-PEDOT:PSS ratio.78][79][80][81] The effect that the work function increases with an increased amount of PFI added to PEDOT:PSS is demonstrated through Kelvin probe measurements of the corresponding films, applied by doctor blading (Figure 4).When the same films are integrated into PTQ10-based solar cells, it can be observed that, within a range of around 50 w/w% of PFI within the mixture, the V OC of the device increases in a similar manner as the work function of the HTL, which shows the high influence and necessity of high work function HTLs to energetically adapt to the active layer  (Figure 4).It can be further observed that above a certain ratio of PFI added to the mixture, the resistance of the film increases strongly (cf., Figure S3, Supporting Information), which, in addition to the V OC , has a high influence on the overall performance.Thus, an optimum PCE is reached for the blends with a ratio of 57.2 w/w% of PFI within the mixture (cf., Figure S3, Supporting Information) and therefore this blend is selected as HTL for the next experiment in comparison to PEDOT:F.
The resulting performance of the different material approaches is displayed in Figure 5. PEDOT:F fulfills the necessary requirements with its high work function (5.7 eV) and clearly outperforms PEDOT:PSS in V OC (0.86 V) and FF (66.4%).Furthermore, through blending PEDOT:PSS with 57.2 w/w% PFI the energy barrier can be overcome as well and a similar performance as with PEDOT:F can be achieved.

Solution-Processed Electrode: Semitransparency and Compatibility to All Other Layers
With the goal of making a fully solution-processed semitransparent stack, we replace the evaporated silver electrode (Ag) with silver nanowires (AgNWs).An AgNW ink based on isopropanol (IPA) is chosen, due to its good compatibility to adjacent layers as well as its good coatability, transparency, and conductivity.We test both PEDOT:F and the PEDOT:PSS:PFI blend with respect to their compatibility with the AgNW top electrode.It can be observed in Figure 6 that the PEDOT:PSS:PFI-even though electronically compatible to the adjacent active layer-loses its performance with AgNW coated directly on top.In contrast, PEDOT:F is compatible and consequently the selected HTL material to be employed with the AgNW electrode.

Stability Evaluation: Exposure to Different Degradation Sources
Semitransparent solar cells of the selected fully solutionprocessed architecture are tested against high temperature (2 h at 120 °C-representing the conditions of a typical industrial secondary encapsulation process, e.g., between glass panes) followed by 200 h of constant illumination with a metal halide lamp (representing device operation in the sun).As ternary active layer compositions have been shown to be able to improve microstructure and device performance, we test the selected active system PTQ10:BTP-4F-12 with and without PC 71 BM as  ternary component. [82,83]Figure 7a shows the relative PCE of the devices during the course of degradation (i.e., their PCE after degradation normalized to their PCE at t = 0; the performance parameters (PCE, V OC , J SC , and FF) at t = 0 are shown in Figure S4, Supporting Information).While the binary devices maintain >80% of their initial performance, the ternary devices maintain 90% of their initial PCE, which confirms the positive effect of PC 71 BM on the stability.Furthermore, a similar set of devices is exposed to 85 °C for 200 h to evaluate their behavior under prolonged elevated temperature.As shown in Figure 7b, the obtained results are similar to the previous degradation experiment.This time, devices with the binary active layer composition degrade by slightly more than 20% of their initial performance (cf. Figure S4, Supporting Information), while the ternary system again provides superior stability with relative PCEs of >90% after degradation.

Evaluation of Area Scalability: Semitransparent OPV Modules
To validate the scalability of the developed layer stack, solar cells and modules of different sizes are fabricated and investigated.Starting from individual solar cells with an active area of 9 mm 2 , we increase the device area by a factor of >100 to fabricate modules with eight cells connected in series and an active area of 10.2 cm 2 (see Figure S5, Supporting Information, for detailed information about the module layout).The performance of all devices is presented in Table 1.Upon scaling the device type Table 1.Photovoltaic device performance of cells and a respective module of >100 times larger active area.The performance of the best cell is presented together with mean values averaged over six devices (in brackets).and area from 0.1 cm 2 small cells to 10 cm 2 modules, no performance loss is observed, which underlines the excellent scalability of the selected materials and layer stack.
In Figure 8, the I-V curve of a 10 cm 2 semitransparent solar module (a) and a photograph (b) is shown.The module has an average visible transmittance (AVT) of 11.3%.The transmittance spectrum of the module is shown in Figure S6, Supporting Information.With an applied back-reflector, a higher I sc per cell area of 20.35 mA cm À2 is achieved, leading to a module PCE of 10.8% per active area and 10.2% with respect to the total module area with a geometrical FF (GFF) of 94.4%, which is among the highest performances of fully solution-processed semitransparent modules.This is especially notable because all process parts are undertaken in air, which underlines the industrial relevance of the system.

Conclusions
In this work, we presented a step-by-step development to achieve an OPV layer stack that fulfills the requirements for manufacturing stable, semitransparent, high-performance organic photovoltaic modules with industrially compatible fabrication processes.We pointed out challenges and considerations for each of the printed layers that arise when moving to scalable coating and selected suitable materials by evaluating and maximizing their performance and stability.Finally, the transition from small area solar cells to a 10 cm 2 module by using our developed environmental manufacturing process has been demonstrated and yielded a solar module with 10.8% efficiency, which is currently among the highest values for semitransparent, fully solutionprocessed OPV modules.This, in combination with a good stability, corroborates that OPV finally starts to become competitive on the market and brings the technology one step closer to a successful large-scale commercialization.
PC 71 BM was purchased from Nano-C.MoO 3 was acquired from Carl Roth and Ag from Evochem Advanced Materials, respectively.PEDOT:F was provided by the research group of Prof. Yinhua Zhou at the Wuhan National Laboratory for Optoelectronics, Huazhong University of Science and Technology (China).PEDOT:PSS (HTL Solar) was acquired from Heraeus and Nafion was acquired from Sigma-Aldrich (Merck).BM-HTL-1 was supplied by Brilliant Matters.Ink-Y AgNWs were acquired from Cambrios.ECOS AgNWs were obtained from HeiQ RAS.
Device Fabrication: Prior to application, the substrates for all devices were wiped with a microfibre tissue that was soaked with toluene.For the solar cells (Section 2.1 and 2.2), the ETLs, ZnO and SnO 2 , were sonicated for 5 min and filtered with a 0.45 μm polypropylene filter, respectively.Subsequently, the solutions were applied onto the glass/ITO substrates by doctor blading in air followed by a thermal treatment (140 °C, 3 min, in air) unless stated otherwise in the results and discussion section.
For the active layers with PM7 or PM6 (and PTQ-10 for Section 2.1.2) as donors and BTP-4F-12 as acceptor, the donor and acceptor were mixed with a ratio of 1:1.2, dissolved in o-xylene and stirred overnight at room temperature under nitrogen atmosphere.For the PTQ10-based binary (and ternary) solar cells, PTQ10, BTP-4F-12 (and PC 71 BM) were mixed and dissolved in o-xylene with a ratio of 1:1(:0.2) and stirred overnight at room temperature under nitrogen atmosphere.All active layers were applied by blade coating in air.For the active layer variation experiment (see Section 2.1.2),different coating speeds ranging from 5 to 90 mms at a substrate temperature of 65 °C were used to obtain different active layer thicknesses.After deposition, the films were thermally treated under nitrogen for 10 min at 100 °C.The HTLs were fabricated by 1) thermal evaporation of 10 nm MoO 3 or 2) blade coating PEDOT:F (or PEDOT: PSS:PFI).The PEDOT:F was deposited on the active layer at 45 °C substrate temperature without further pre-or posttreatment.The PEDOT:PSS:PFI mixture was diluted in 4 μL isopropanol per μL of the mixture and was coated at 70 °C without posttreatment.The top electrodes were fabricated by 1) a thermally evaporated 100 nm silver electrode (Ag) or 2) a blade-coated AgNW film (Ink-Y) at 65 °C substrate temperature and thermally treated at 120 °C for 3 min.
For the scalable, air-processed solar module and the respective cells (Section 2.3), substrate preparation and blade coating of ETL was done in a similar manner than the solar cells in Section 2.1 and 2.2.For the active layer PTQ10, BTP-4F-12 and PC 71 BM were mixed with a ratio of 1:1:0.2 and dissolved in o-xylene with a total concentration of 25 mg mL À1 (1:1:0.2) and stirred at 70 °C for 30 min and then cooled down and stirred at 40 °C overnight.The active layer was coated on a substrate heated with 50 °C temperature forming a dry film thickness of 185 AE 10 nm and subsequently thermally treated with 130 °C for 3 min in air.As HTL, BM-HTL-1 was doctor-bladed on top at 35 °C without any pre-or posttreatment.As top electrode, ECOS AgNWs were filtered with a 10 μm filter and coated at 40 °C substrate temperature without further treatment.The patterning lines for the module (P1, P2, P3) were done with a femto-second laser.
Characterization: Characterizations of the solar cells and modules were done with a solar simulator from LOT Quantum Design (LS0916 AAA).All devices (also the semitransparent ones) were irradiated from the glass/ITO side during measurement and light degradation.Currentvoltage characteristics and conductivity measurements were obtained from B2901A sourcemeter from Keysight Technologies Inc.Light degradation measurements were undertaken under nitrogen with applied UV filter through exposure with metal halide lamp light.Thermal degradation and thermal stress test were carried out under nitrogen atmosphere.Film thickness measurements were performed with a μsurf custom confocal microscope from NanoFocus AG.Work function measurements on ITO substrates were done with scanning Kelvin probe system SKP5050 from KP Technology using freshly peeled HOPG as reference.Transmission spectra were taken with a Lambda 950 UV/VIS Spectrometer from PerkinElmer including an integrating sphere.
AgNW ink.The authors acknowledge funding from the German Federal Ministry for Economic Affairs and Climate Action via the ZIM project "OPV4IoT" (FKZ: 16KN098724).The authors also acknowledge the "Solar Factory of the Future" as part of the Energy Campus Nuremberg (EnCN), which is supported by the Bavarian State Government (FKZ 20.2-3410.5-4-5).Y.Z.acknowledges funding from the National Natural Science Foundation of China (grant no.51973074).H.-J.E. and C.J.B. acknowledge funding from the European Union's Horizon 2020 INFRAIA program under grant agreement no.101008701 ("EMERGE").
Open Access funding enabled and organized by Projekt DEAL.

Figure 1 .
Figure 1.Targeted device architecture for scalable OPV devices with solution-processed ETL, AL, and HTL as well as solution-processed semitransparent top electrode.

Figure 4 .
Figure 4. Work function measurements (Kelvin probe, left axis, blue circles) on films of the PEDOT:PSS:PFI mixture coated on ITO, where the w/w% content of PFI within the mixture is varied.Additionally, the V OC of PTQ10-based devices with the respective PEDOT:PSS:PFI mixture as HTL is shown (right axis, red triangles).(Coating 100 w% PFI as HTL leads to nonfunctional devices).

Figure 8 .
Figure8.a) J-V curve and b) photograph of a fully solution-processed and fully air-processed, semitransparent module on glass/ITO with eight cells connected in series and a total active area of 10.17 cm 2 with back-reflector.