Fast Inline Roll-to-Roll Printing for Indium-Tin-Oxide-free Polymer Solar Cells Using Automatic Registration

ast inline roll-to-roll printing and coating on polyethylene terephthalate (PET) and barrier foil was demonstrated under ambient conditions at web speeds of 10 m/min for the manufacture of indium-tin-oxide-free (ITO-free) polymer solar cells comprising a 6-layer stack: silver-grid/PEDOT:PSS/ ZnO/P3HT:PCBM/PEDOT:PSS/silver-grid. The first and second layers were printed at the same time using inline processing at a web speed of 10 m/min where flexographic printing of a hexagonal silver grid comprises the first layer followed by rotary-screen printing of a PEDOT:PSS electrode as the second layer. The third and fourth layers were slot-die coated at the same time again using inline processing
at a web speed of 10 m/min of firstly zinc oxide as the electron transport layer followed by P3HT:PCBM as the active layer. The first three layers (silver-grid/PEDOT:PSS/ZnO) comprise a generally applicable ITO-free, semitransparent, electron-selective front electrode for inverted polymer solar cells. This electrode shows a low sheet resistance (-10 Ω/❑) and good optical transmission in the visible range (-60 %). 
The solar cell stack was completed by rotary-screen printing of a hole-collecting PEDOT:PSS layer at 2 m/min and a comb-patterned silver-grid back electrode at the same speed. The solar cells were post processed by using fast roll-to-roll switching to a functional state.

Introduction
Indium tin oxide (ITO) has for decades been the only widely available transparent electrode for a number of technologies, that is, organic photovoltaics (OPVs) and organic/polymer light emitting diode (OLED/PLED) devices. Research in these areas has been driven by the potential for low-cost manufacture at high throughput using roll-to-roll (R2R) techniques, for which the use of indium is unfavorable because of its scarcity and high cost. It is, therefore, paramount that alternatives to ITO are found, but just as important is their actual implementation in the research communities to convince researchers to adopt the use of these new electrode types instead of ITO, which is unlikely to be sustainable for large-scale technologies. Numerous attempts to substitute ITO, particularly in OPVs, have been reported, including PEDOT:PSS [poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate)], thin semitransparent metal layers, metal grid structures, metal nanowires, graphene, and carbon nanotubes with varying success. However, only the report in Ref. [11] has demonstrated the potential to be sufficiently scaleable for the final substrate to be produced at a speed and cost allowing for its consideration as a serious candidate for R2R printing.
Here we present a new flexible semitransparent substrate which, when used in the preparation of OPVs, provides similar performances to analogous modules prepared on ITOcovered substrates. This "flextrode" substrate is suitable for inverted-structure OPVs and is based on a high-conductivity PEDOT:PSS layer, which is coated with zinc oxide. For larger areas a flexo-printed silver grid reduces sheet resistance. We demonstrate how the ITO-free electrode material can be processed at high speed by printing several layers at 
the same time using inline printing and coating. We show that a length of 1000 m is easily manufactured within a few hours having full 2-dimensional registration of the printed pattern. We see this as the first real candidate for a mass producible replacement for ITO, and to promote the use of such substrates in academic research, this substrate is made freely available.

Results and Discussion
The flextrode substrate suitable for large-area OPVs was prepared by flexographic printing of a honeycomb-patterned silver grid structure (2 mm pitch) by using a water-based silver-nanoparticle ink (PFI-722 from PChem Associates with 60 wt % Ag content). This was followed by rotary-screen printing PEDOT:PSS on top of and in registry with the silver grid exposing one side of the silver for subsequent direct contacting and serial interconnection of modules by printing. The highly conductive PEDOT:PSS (Heraeus Clevios PH1000) was diluted with isopropanol at a concentration of 10:3 by weight to ensure proper printing. The inline processing speed was 10 m/min and is in reality only limited by the drying speed (in 2x2 m² ovens). Infrared dyers were used in conjunction with hot-air convection ovens at a temperature of 140C. ZnO ink as a stabilized nanoparticle suspension in acetone (55 mg/mL) was finally coated on top by slot-die coating (10 m/min) to smooth the surface and leave a small part of the bottom electrode exposed for later contacting. Critical to the success is the formulation and the application of the ZnO ing because it is reactive towards the atmosphere. The nanoparticulate nature of the ZnO and the acetone solvent employed serve to dissolve the PEDOT:PSS surface slightly where as the nanoparticles efficiently fill voids in the surface. The net result is a very smooth surface that is fully covered with ZnO. It should be emphasized that the ink formulation represents a 6-year effort of optimization and that the formulation of the ink and the details of the process depend quite closely on one another. The flextrode carrier substrate is either polyethylene terephthalate (PET, Dupont-Teijin, Melinex ST506) or barrier foil (Amcor, Ceramis). The three processing steps and a processing flowchart are shown in Figures 1 and 2. 

Figure 1. Processing of the flextrode substrate on PET: a) Flexo-printing of the silver grid; b) Rotary-screen printing of PEDOT:PSS; and c)slot-die coating of nanoparticle-based ZnO ink.

Figure 2. Flowchart visualizing the inline fabrication of the flextrode substrate and the solar cell module finalization. Currently, the silver grid together with highly conductive PEDOT:PSS layer, ZnO, and an active layer can be processed inline at 10 m/min. The illustrated processes are flexo-printing (F), rotary-screen printing (RSP), slot-die coating (SD), and drying (D).

When preparing small-area OPVs the silver grid is not necessary, and, therefore, the initial flexo-printing step of the silver grid was eliminated as demonstrated earlier. Using a silver grid lowers the sheet resistance of the front electrode but inherently blocks some of the surface from incoming light. Figure 3a and b show the transmittance characteristics for the substrate after each processing step, and the light blocking is clearly observed as the transmittance is reduced to around 50% after the processing of the silver grid.

 

Figure 3. a) Illustration of the optical transmittance after each processing step. b) Photos of the substrate after each processing step. c) Optical microscopy image of the silver grid lines with a PEDOT:PSS layer on top.

The grid structure was built up on hexagonal elements with a diamension of 2 mm between two parallel sides, whereas the nominal line width of 100 µm from the flexo-printing form was increased to an average line width of approximately 160 µm during printing with a grid peak height of up to 750 nm. he efficient consumption of silver was ensured by using a low volume anilox roller (1,5 mL/m-2, 480 L/cm-1) together with a sleeve-based elastomer printing form with a hardness of 65 Shore. An experimental run of ~1000 m flexo-printed grid on barrier foil led to a consumption of roughly 100 mL (200g) of ink. The cost of the grid when employing inline processing is essentially determined by costs of materials, drying and cleaning as machine time is already accounted for in the processing of the PEDOT:PSS layer.

The rotary-screen printing of the PEDOT:PSS on top of the silver grid leads to an average sheet resistance of 10.4 Ω/❑ for a typical ITO electrode on PET used in R2R processing of organic solar cells. The sheet resistance of the grid and grid/PEDOT:PSS was measured over a defined number of squares as described in the Supporting Information. Rotary-screen printing of the PEDOT:PSS film was carried out using a 215 mesh screen with an open area of 25% that leads to a theoretical wet-ink deposit of 20 µm. The ink covers the silver grid lines and follows its metrology as shown in Figure 3c. Rainbow-colored optical interference patterns of the thin PEDOT:PSS film on the silver grid lines are clearly visible. When preparing the flextrode substrade comprising silver-grid/PEDOT:PSS/ZnO, the process is finished by slot-die coating of the electron-transporting (hole blocking) ZnO layer on top of thye electrode stack (see process flow in Figure 2). Thus, the transparency is further reduced by a few percent to 60.3% at 550 nm and a maximum of 63.4% at 424 nm as summarized in Table 1.

In addition to making the flextrode substrate electron selective by enabling the hole-blocking characteristics, the ZnO layer also improves the surface roughness by an order of magnitude from approximately 80 nm down to below 7 nm on average. Similar to the PEDOT:PSS layer, the ZnO layer with a thickness of around 100 nm basically follows the metrology of the underlying layers but has a significantly better flattening effect. The >nO layer was dried passing through two ovens (length = 2 m) at 70°C and 140°C, respectively, and coated at a web speed of 5 or 10 m/min. Both layers were dried at 140°C in the case where the flextrode was prepared by coating both the ZnO and the active layer simultaneously using inline printing at 10 m/min.   The electrodes were manufactured in a 16-stripe design with a width of 13 mm and a gap of 2 mm between the stripes, but other electrode layouts are also possible depending upon the requirement of the final solar cell and module design. Here, the flexo-printing and register-controlled rotary-screen printing enabled a 2-dimensional patterning leading to electrode stripes (~300 mm length) along the web direction, whereas the ZnO layer was continously slot-die coated in 16 stripes of 13 mm width. The manufacturing speed of the 1 m² flextrode substrate (based on 305 mm web width, excluding setup times) is approximately 60 s (10 m/min, 3 individual runs). The complete process speed is in principle limited only by the dryer configuration and fully independent of the electrode design to this point. Exceptional process improvements can thus be achieved when the silver grid and the subsequent PEDOT:PSS layer are printed in an inline configuration. We are currently able to print both layers inline at 10 m/min with minimal waste and improved register control because of decreased foil shinkage. The calculated time for 1 m² of printed flextrode substrate is then reduced to 40 s with the advantage of less handling and rewinding processes.
   Two continous guidelines were simultaneously printed with the silver grid, which have been used for a camera-based real-time adjustment of the slot-die coating head for the final ZnO layer (and for the active layer in cases when the flextrode is preapred in situ). Printed marks were used to screen print the PEDOT:PSS layer in register on the underlying silver grid. Although these prints consume some silver ink and have no functionality for the electrode, they do help to decrease the human interaction for a fully automated process workflow. Finally, inkjet-printed barcodes printed during the first passage through the machine help to identify the electrodes and solar cells during later processing. A stribe camera system was used to monitor the layer quality and overprint accuracy even at higher speeds than the described ones. The process optimization tools are shown in Figure 4 with its positioning in the machine setup (Figure 5).<br>
   The flextrode substrates as shown in Figure 6a and b can either be used for large-scale processing of ITO-free OPVs on full R2R systems or smaller roll-coater setups such as described in Refs. [23] and [24]. After cutting individual sample pieces (Figure 6c) the substrate is suitable for spin coating or other small-scale in deposition processes such as rod coating or doctor blading.

OPV sample fabrication
Single ITO-free solar cells based on the flextrode/P3HT:PCBM/PEDOT:PSS/Ag Structure [P3HT=poly(3-hexylthiophene), PCBM=[6,6]-phenyl-C61-butyric acid methyl ester] have been successfully manufactured in a full R2R process and subsequently delivered an efficiency of up to 1.82% on an area of 6 cm². The fill factor exceeded 51% with a current density of ca. 7 mA/cm².
   Here, we verify the functional homogeneity of the flextrode substrate with a larger set of fully R2R-processed OPV modules with eleven serially connected cells. The modules (Figure 7b) with an active area of 66 cm² (module area ~ 100cm²) were manufactured by using the structure: flextrode/P3HT:PCBM/2x-PEDOT:PSS/Ag, for which the P3HT:PCBM active layer was slot-die coated, and the PEDOT:PSS (Agfa 5010) and the silver grid back electrode (Dupont 5025) were rotary-screen printed (see process flows 2, 3, and 4 in Figure 2). The modules were encapsulated from both sides by using a 72 µm-thick barrier foil (Amcor) using UV-curable glue (DELO LP655).

Figure 4. a) Camera-based guidelines detection for the cross-directional realtime adjustment of the slot-die head. b) Inkjet printing of individual barcodes for substrate and device identification. c) Section of the silver-grid flexo-printing form including the guidelines and register mark. d) Strobe camera for manipulating the printed layers at high speed. e) Camera output for monitoring the printed layers.

 

Figure 5. Process optimization tools, namely guideline detection (GL), strobe camera (C), and barcode inkjet printer (IJ) mounted on the R2R machine setup with flexo-printing (F), slot-die coating (SD), rotary-screen printing (RSP), and driers (D).

Light soaking for roughly 50 h at 1000 W/m² (AM 1.5G) was performed to stabilize the cell parameters before I-V measurements. The I-V curves of then randomly selected modules shown in Figure 7a have similar characteristics, with an average power conversion efficiency (PCE) of 1.5% (max. 1.60%), open-circuit voltage (Voc) of 5.54 V(max. 5.63 V), short-circuit current (Isc) of -32.09 mA (max -33.01 mA) and fill factor (FF) of 55.64% (max. 57.34%).

The average current density (Jsc) of approximately 5.3 mA/cm² per cell seems to be limited by the lower transmittance of the semitransparent electrode compated to ITO-based cells with the same active-layer composition. Future optimizations of different grid designs and thinner grid lines with less shadow loss are promising for further improvement of the flextrode substrade and the solar cell characteristics.

R2R switching
A final electrical-switching treatment is necessary to transform the whole solar cell stack based on the structure flextrode/active layer/PEDOT:PSS/electrode to a fully functional photovoltaic device by dedoping the PEDOT:PSS layer at the active-layer/PEDIT:PSS interface. Compared to grid-free PEDOT:PSS/ZnO substrates for small area devices (stripe width < 3 mm), the additional silver grid reduces the sheet resistance, enables large device areas, and improves the serial connection with the printed back electrode. Furthermore, and most importantly, it allows the distribution of a short pulse with high current density and high electric field over several square centimeters. All R2R-manufactured flextrode-based large-area devices were switched using a customized R2R switching setup as shown in Figure 8a. The switching head (Figure 8b and c) with its contact pin array is designed according to the specific module layout and connects to the bus bars of the silver comb back electrode. Proprietary software was used to control the whole switching procedure that includes: contact testing, applying the individual switching pulse per single cell, resistance measurement or dark I-V curve acquisition, and further switching cycles, if necessary in case of insufficient switching. All steps were performed in parallel for all cells and required approximately 15 s for a 16-cell module. Therefore, the switching of the 10 m flextrode substrate with 130 modules of roughly 6 cm width took about 32 min. The correct switching parameters for a roll of solar cell devices cannon be fully specified in advance and need some prior tests depending on the area, active layer thickness, and back electrode conductivity. Hereby, the software allows an individual adjustment of the electrical pulse length (typically 15 ms), current (up to 10 A, typically 0.2-0.5 A), voltage (up to 36 V, typically 29 V), treshold resistance (100-1000 Ω), and the number of switching cycles (optimum one cycle). After settings the individual parameters the machine runs automatically without further actions by the operator. Protocols of all switching parameters, especially the threshold resistance, allow for backtracking and identification of potentially unswitched cells using a log-file.
   It should be pointed out that all switching procedures can be easily performed by hand using a current-voltage source and then shorting the wires connected to the device. A video demonstrating the simple procedure is available.

Figure 6. a) Hundreds of meters of flextrode substrate (PET) ready for further R2R processing. b) Flextrode substrate with 16 individual electrode stripes. c) Small slextrode cut-out suitable for OPV manufacturing using spin coating.

 

Figure 7. a) I-V curves (dashed lines) of ten randomly selected OPV modules containing eleven serial connected cells. The red line illustrates the average I-V curve. b) Fully encapsulated OPV module fabricated on the flextrode substrate.

 

Figure 8. a) Photograph of the R2R switching setup with a mounted roll of fully R2R-processed solar cell modules. b) Switching of a solar cell module with the switching head in contact to the back-electrode bus bars. c) Lifted switching head showing the gold contact pin array. 

Conclusion
With our flextrode substrate we present an attractive ITO-free alternative for the fast manufacture of large-area OPVs with an inverted structure. The manufacturing is free of very scarce materials such as indium and allows fast, vacuum-free processing of individually designed electrodes The sheet resistance of the combined silver grid and PEDOT:PSS stack is almost six times lower than that of a typical flexible ITO-coated PET substrate. Fully R2R-processed solar cells and modules were successfully manufactured leading to PCEs of more than 1.8% on single cells and FFs of more than 60%. Modules could be manufactured with more than 1.6% PCE on the active area. A simple electrical-switching procedure is necessary to transform the solar cell to a functional device. This can be performed manually on small-scale devices or fully automated by using a customized R2R setup. To motivate the OPV scientific community performing research on ITO-free substrates, the flextrode comprising substrate/silver-grid/PEDOT:PSS/ZnO is available free of charge to all academics.

Acknowledgements
This work was supported by the Danish Ministry of Science, Innovation, and Higher Education through the EliteForsk initiative. We would like to thank Dr. Hanne Lauritzen for suggesting the name "flextrode".