Low-Temperature Graphene-Based Paste for Large-Area Carbon Perovskite Solar Cells

Carbon perovskite solar cells (C-PSCs), using carbon-based counter electrodes (C-CEs), promise to mitigate instability issues while providing solution-processed and low-cost device configurations. In this work, we report the fabrication and characterization of efficient paintable C-PSCs obtained by depositing a low-temperature-processed graphene-based carbon paste atop prototypical mesoscopic and planar n–i–p structures. Small-area (0.09 cm2) mesoscopic C-PSCs reach a power conversion efficiency (PCE) of 15.81% while showing an improved thermal stability under the ISOS-D-2 protocol compared to the reference devices based on Au CEs. The proposed graphene-based C-CEs are applied to large-area (1 cm2) mesoscopic devices and low-temperature-processed planar n–i–p devices, reaching PCEs of 13.85 and 14.06%, respectively. To the best of our knowledge, these PCE values are among the highest reported for large-area C-PSCs in the absence of back-contact metallization or additional stacked conductive components or a thermally evaporated barrier layer between the charge-transporting layer and the C-CE (strategies commonly used for the record-high efficiency C-PSCs). In addition, we report a proof-of-concept of metallized miniwafer-like area C-PSCs (substrate area = 6.76 cm2, aperture area = 4.00 cm2), reaching a PCE on active area of 13.86% and a record-high PCE on aperture area of 12.10%, proving the metallization compatibility with our C-PSCs. Monolithic wafer-like area C-PSCs can be feasible all-solution-processed configurations, more reliable than prototypical perovskite solar (mini)modules based on the serial connection of subcells, since they mitigate hysteresis-induced performance losses and hot-spot-induced irreversible material damage caused by reverse biases.

at the second disk, which is made of two holes with a diameter of 1 mm in diameter. Subsequently, the jet streams collide in a nozzle with a diameter of 0.3 mm between the second and the third disks.
During the passage of the sample through the nozzle, the turbulence of the solvent originates the shear force causing the exfoliation of graphite. 1,2 The as-produced dispersion was cooled down by a chiller and then collected in another container. The sample was re-processed three times by three consecutive WJM machines with a nozzle of 0.2, 0.15 and 0.1 mm diameters, respectively. The as-produced graphene dispersion was subsequently ultracentrifuged at 16 000 g for 30 min at 15 °C using a Beckman Coulter Optima™ XE-90 centrifuge with an SW32Ti rotor. Thus, the un-exfoliated graphite was removed through the sedimentation-based separation (SBS). 6 More in detail, 80% of the supernatant was collected by pipetting after the ultracentrifugation process. The pipetted sample was dried using a rotary evaporator (Heidolph, Hei-Vap Value) at 70 °C and 5 mbar. Subsequently, 500 mL of EtOH was added to the dried sample. The sample was then dispersed using a sonic bath for 10 min. Consecutively, the sample was centrifuged at 800 g using Beckman Coulter Optima™ XE-90 centrifuge with an SW32Ti rotor. The sediments were collected discarding the supernatant. This process of decantation was repeated twice to wash out the NMP residuals. Finally, the sediments were dispersed in 200 mL of EtOH, obtaining a concentration of graphene flakes of 0.9 mg mL -1 .
The graphene flakes used for the formulation of the pastes were produced in form of dispersions through the WJM method in NMP, as described above. The as-obtained graphene dispersion in NMP was dried in form of powder by means of a freeze dryer machine, to obtain the final freeze-dried graphene powder. Subsequently, the component of the pastes, including WJM-produced graphene and carbon black as the electrically conductive fillers and a thermoplastic mixture, were homogeneously mixed in IPA. Graphene-free carbon pastes were produced by following the same protocol used for graphene-based carbon paste, except for using a mixture of carbon black and graphite as the electrically conductive fillers with the same weight percentage relatively to the solid content. The weight content of the electrically conductive contents was optimized to achieve to best trade-off between electrical and mechanical performances, the most relevant ones are summarized in Devices' fabrication. The reference perovskite solar cells (PSCs) using Au as the counter electrode (CE) were fabricated by following to the protocols reported in ref. 7 for both small-area (active area = 0.09 cm 2 ) and large-area (active area = 1 cm 2 ) mesoscopic and planar n-i-p-structures, except for the use of graphene-doped compact TiO 2 (cTiO 2 +G) and graphene-doped mesoporous TiO 2 (mTiO 2 +G) in mesoscopic devices. The cTiO 2 +G and mTiO 2 +G layers were obtained by modifying the protocols reported for cTiO 2 and mTiO 2 , respectively, 7 as resulting from the addition of 1 vol% S4 graphene dispersion in EtOH in the cTiO 2 precursor solution or mTiO 2 paste. The carbon perovskite solar cells (C-PSCs) were fabricated by following the protocols used for the Au-based reference, except for the replacement of Au-based CE with carbon counter electrodes (C-CEs). The latter were obtained by depositing through spin coating the graphene-based carbon pastes, keeping the substrates at ambient temperature. For the small-area devices, four pixels were obtained on the same substrate.
Large-area C-PSCs were also produced by depositing C-CEs though doctor blading deposition of the graphene-based carbon paste or other commercially available carbon paste (DN-CP01, Dyenamo).
For the metallized miniwafer-like area (substrate area = 6.76 cm 2 ; aperture area = 4.00 cm 2 ) mesoscopic C-PSCs, both front and back electrodes (i.e., FTO and C-CE) were metallized by depositing through vacuum (10 -6 bar) thermal evaporation three 1 mm-width stripes of 100 nm-thick Au electrode, following the layout shown in the main text (Figure 5a).

Materials' and devices' characterization.
The carbon pastes were characterized by depositing ~40 µm-thick films through doctor blading on both glass and polyethylene terephthalate (PET) substrates.
The thickness of the films was measured using a contact surface profilometer (XP-200, Ambios Technology). The sheet resistance (R sheet ) measurements were performed using a four-probe system (Jandel RM3000 Test Unit) on films deposited on glass substrates. The resistivity (Ω cm) was calculated from the film R sheet (Ω sq -1 ) and thickness (cm) according to the equation: resistivity = R sheet × thickness. Scanning electron microscopy measurements were performed using a high- Incident Photon-to-current Conversion efficiency (IPCE) spectra were acquired by means of a homemade setup composed by a monochromator (Newport, mod. 74000) coupled with a Xe lamp (Oriel Apex, Newport) and a source meter (Keithley, mod. 2612). A home-made LabVIEW program controlled the spectra acquisition. The device data were analysed with OriginPro® 9.1 software.
Stability tests were carried out by following the ISOS-D-1 and ISOS-D-2 protocols 8 on unencapsulated devices, as reported in ref. 9

S5
S6 Figure S1. a.b) Photographs of representative small-area Au-based and C-CE-based mesoscopic PSCs using graphene doped ETLs (4 pixels on the same substrate), respectively. c,d) Photographs of representative largearea (1 cm 2 ) Au-based and C-CE-based mesoscopic PSCs using graphene-doped ETLs, respectively. All the photographs show the back-contact side of the devices.  Figure S4. J-V curve (reverse voltage scan) and the corresponding power density vs. voltage plot measured for the most efficient large-area (active area = 1 cm 2 ) mesoscopic C-PSC using graphene doped ETLs and produced by depositing the graphene-based C-CEs through doctor blading method. Table S1. Photovoltaic Figures of Merit measured in reverse voltage scan mode for the mesoscopic C-PSCs using graphene-doped ETLs and produced by depositing the graphene-based C-CEs through doctor blading method.  Figure S5. J-V curve (reverse voltage scan) and the corresponding power density vs. voltage plot measured for the most efficient large-area (active area = 1 cm 2 ) mesoscopic C-PSC using graphene-doped ETLs and C-CEs produced by depositing a commercially available carbon paste through doctor blading method. Figure S6. J-V curve (reverse voltage scan) and the corresponding power density vs. voltage plot measured for the most efficient large-area (active area = 1 cm 2 ) mesoscopic HTL-free C-PSC with C-CEs produced by depositing a commercial carbon pastes through doctor blading method. S14 Table S2. Summary of the PCEs of our best large-area C-PSCs and those reported in relevant literature, with the corresponding C-CE deposition methods, cell structure and cell area.