Slot-die coated bulk heterojunction vs layer-by-layer organic photovoltaics: Device architecture dependent degradation Open Access

Organic photovoltaic (OPV) devices have emerged as one of the most favorable green energy technologies since their introduction,¹ primarily due to the use of organic polymers or small molecules as semiconductors, which aligns with sustainability goals for environmentally friendly renewable energy sources. These devices are expected to reduce reliance on fossil fuels and mitigate global warming by lowering greenhouse gas emissions.^2–7 Their versatility allows for fabrication on various substrates such as plastics,^8–11 paper,^12,13 and steel,^14,15 making them lightweight and flexible, which facilitates their integration onto diverse surfaces. Potential applications include building-integrated photovoltaics (e.g., windows and façades),^16,17 indoor photovoltaics,^18–20 and agri-photovoltaics,^21,22 advancing efforts toward integrated sustainable energy solutions.

Over the last decade, the performance of state-of-the-art organic OPV devices has rapidly improved, achieving power conversion efficiencies (PCEs) of up to 20.8%²³ in laboratory settings at present, i.e., using spin-coating techniques in inert environments. This progress has been driven by advancements in device physics, device engineering, interfacial engineering, and, most notably, innovations in materials development.^24–30 The invention of non-fullerene acceptors (NFAs), such as Y-series acceptors, has been instrumental in surpassing the 15%³¹ efficiency barrier and reaching the 20%²³ mark. Unlike traditional fullerene acceptors, NFAs exhibit enhanced light absorption in the higher (near infrared) wavelength regions, enabling a broader absorption of the solar spectrum. When blended with donor polymers such as PM6³² and D18,³³ these NFAs demonstrate improved morphology, charge transport, and charge collection efficiencies, both in the bulk heterojunction (BHJ)^32,33 and layer-by-layer (LBL)^23,34 architectures, as recently demonstrated by our group.³⁴ In addition, the NFA energy levels can be optimized to align with donor polymers, reducing non-radiative losses to achieve higher V_oc. However, a notable performance gap persists when transitioning from laboratory techniques such as spin coating to industry-compatible fabrication methods such as slot-die coating in the air.

For OPVs to achieve commercial viability, it is essential to develop devices using industry-compatible fabrication techniques. While significant advancements in OPV performance have been achieved with spin-coating methods, recent efforts have shifted toward scalable and cost-effective techniques such as slot-die coating, roll-to-roll printing, flexography, gravure printing, blade coating, and inkjet printing.^35–43 These methods enhance scalability and throughput while reducing production costs and simultaneously offer precise control over film thickness and uniformity, which are crucial for optimizing device efficiency and reproducibility. Among these, slot-die coating has emerged as a prominent technique due to its compatibility with large substrates and its ability to produce continuous films at high speeds with minimal material waste. Reports have demonstrated efficiencies of ∼15%⁴⁴ for small devices fabricated using the slot-die method.^45–51 However, further efforts are required to scale up production for mini to large modules. Notably, a world record efficiency of 14.46%⁵² on an area exceeding 200 cm² was recently achieved using blade coating, as documented in the NREL chart. Inspired by such progress, we have developed efficient small and mini-module devices using slot-die coating, advancing the potential for scalable OPV production.

In addition to achieving high efficiencies and scalable devices, the long-term stability of OPVs under various stress conditions is a critical factor for commercialization. Recent advancements in material engineering, including the development of conjugated polymer donors and NFAs with high glass transition temperatures, have significantly improved both the thermal and light stability of OPVs.^53–55 Similarly, advancements in transport layer engineering, such as the use of oxide-based interlayers, 2D materials, and self-assembled layers, have further enhanced device stability.^56–60 A particular emphasis has been placed on improving light stability, as studies have revealed that exposure to UV light often causes severe degradation in devices due to photochemical reactions at the photoactive layer and interlayers. This issue has been mitigated through the incorporation of antioxidants,⁶¹ tuned interlayers,^59–62 and UV-blocking filters,^63–66 resulting in devices with extended light stability. Reports have demonstrated devices capable of maintaining stability for over 1000 h,⁶⁵ with some studies extrapolating a lifespan of up to 30 years.⁶⁷ However, achieving comparable advancements in thermal stability remains a challenge, and further demonstration of such long-term stability for slot-die air-processed OPV is still missing. In this work, we systematically tracked the light stability of slot-die air-processed devices and conducted a detailed study on their thermal stability following ISOS protocols.⁶⁸ These investigations aim to address existing bottlenecks and the still limited understanding of the degradation pathways of slot-die processed OPVs, paving the way for their broader adoption in commercial applications.

This study builds upon our previous work,³⁴ where we highlighted the enhanced stability of LBL processed (by spin-coating) conventional architecture OPV devices under thermal and light stress using various blend systems. Here, we report the fabrication and performance characterization of slot-die-coated OPV devices on transparent glass substrates, utilizing the conjugated polymer donor PM6 and non-fullerene acceptor Y7-12. The devices were processed in the air using the green solvent o-xylene, with both inverted and conventional configurations optimized for BHJ and LBL photoactive layer formations. Comprehensive J–V analysis and device statistics demonstrate the potential for fabricating high-efficiency small devices with excellent reproducibility using slot-die coating. Notably, the efficiencies achieved include a PCE of 15.24% with a fill factor (FF) exceeding 74% for BHJ devices. Detailed partial stack experiments revealed interfacial degradation, underscoring the critical need to focus on interlayers to improve thermal stability. Long-term light stability was successfully demonstrated for inverted BHJ devices with the incorporation of a long-pass filter, consistent with observations from other studies. Furthermore, the small devices were scaled to mini-modules, achieving a PCE of 13.06% with an active area of 13.8 cm². This work provides insights into the interplay between device architecture, material blending strategies, and scalable fabrication techniques, advancing the performance and stability of OPV devices. These findings offer important guidance for future efforts in the commercial implementation of organic photovoltaics.

For this study, both inverted and conventional device structures [Figs. 1(a) and 1(b)] were considered for fabrication. The conventional device structure is a widely adopted configuration for achieving high power conversion efficiencies (PCEs); indeed, nearly all state-of-the-art devices are reported using this architecture.²³ This is primarily due to recent advancements in electron transport layers (ETLs), such as PDINN, which offer enhanced charge selectivity and favorable energy level alignment with the lowest unoccupied molecular orbital (LUMO) of high-performance non-fullerene acceptors (NFAs). However, the long-term stability of these devices is often hindered by using PEDOT:PSS, which is both acidic and hygroscopic. In contrast, the inverted configuration utilizes various metal oxides (ZnO, MoO₃) as charge transport layers, generally resulting in more stable devices. In this study, both bulk heterojunction (BHJ) and layer-by-layer (LBL) methods were employed to form the photoactive layer in each configuration. Several previous studies have demonstrated that devices processed using the LBL method in a conventional structure via spin coating and employing the same solvent for both donor and acceptor materials outperform BHJ-processed devices in terms of efficiency and stability.^23,34,38 This enhancement is attributed to the more favorable molecular organization achieved through the sequential deposition, and from that, also improved morphological stability. LBL processing in inverted structures, however, is less commonly reported due to the inherent challenge of depositing the donor polymer on top of the NFA acceptor when using the same solvent system. In this work, we demonstrate the feasibility of employing slot-die coating to process all four device configurations using the same green solvent for both donor and acceptor materials.

In conventional architecture, PEDOT:PSS and ZnO were used as transport layers. ZnO was selected over other high-performing ETLs, such as PDINN, to maintain consistency in processing across both device types and to enable deposition from green solvent, water, and isopropyl alcohol, respectively. In the inverted structure,BM-HTL dispersion in ethanol was used as the hole transport layer in place of PEDOT:PSS, also due to its deeper-lying HOMO level. For the photoactive layer, PM6 was employed as the polymer donor and Y7-12 as the acceptor, owing to their good solubility in o-xylene. The full names of the materials are provided in the section titled Materials.

Figure 1(c) illustrates the schematic of the slot-die coating process used for the deposition of various layers. All layers except for the electrodes (ITO and Ag) were deposited using slot-die coating under ambient conditions. Additional details on the fabrication process are provided in the section titled Experimental section. Figure 1(d) presents the chemical structures of the donor material PM6 and the acceptor Y7-12. These materials were specifically chosen for their compatibility with benign solvents, enabling environmentally friendly device fabrication under ambient conditions.

The devices were systematically optimized using the slot-die coating technique to achieve higher efficiencies across all four configurations. The optimized coating parameters are provided in Table S1 (supplementary material). Device performance was analyzed through current density–voltage (J–V) measurements under one sun intensity (AM1.5G) light illumination. The performance parameters of 16 optimized devices for each configuration are summarized in Table I.

The J–V measurements revealed reasonably high device performance in all configurations. The highest efficiencies were achieved using the BHJ approach in both configurations: 15.24% for the conventional configuration and 14.50% for the inverted configuration. For the LBL configuration, the best PCEs were slightly lower, at 14.08% and 14.11% for the conventional and inverted configurations, respectively. This trend contrasts with those reported in a few other studies and may be attributed to small differences in the optimization routes for the LBL devices. In the conventional stack, the LBL devices exhibited a slightly reduced fill factor (FF) compared to their BHJ counterparts, attributed to minor recombination in the bulk. Conversely, in the inverted configuration, the LBL devices showed slightly lower J_sc values but comparable FF values to the BHJ devices. Across all configurations, the V_oc remained consistent with minimal variation. Overall, the FF exceeded 70% in all configurations, with the highest FF recorded at 74.87% in a conventional BHJ device.

Figures 2(a) and 2(b) present the J–V curves of the best-performing devices across the four configurations. Figures 2(c) and 2(d) display the external quantum efficiency (EQE) spectra of the devices. The calculated J_sc values from EQE measurements align well with the J_sc values obtained from J–V measurements, with an error margin of less than 5%. In all configurations, the devices show a broad spectral response from 400 to 850 nm. In the conventional configuration, the spectral response of the LBL device is marginally lower in the wavelength range of 650–800 nm, with suppressed acceptor contribution leading to slightly poorer light harvesting and charge collection. On the other hand, in the inverted LBL device, reduced EQE is observed in the 450–650 nm range. The histogram in Fig. 2(e) illustrates the statistical distribution of device performance. Conventional devices exhibited a narrower performance distribution compared to inverted devices. Overall, the low standard deviations highlight the excellent reproducibility of the devices regardless of configuration. These findings provide strong motivation for further exploration of the devices’ long-term stability and degradation pathways.

A systematic study was conducted to monitor the thermal degradation of all four configurations according to the ISOS protocols. The devices were placed on a hot plate at 85 °C inside a nitrogen atmosphere (glovebox), and the device J–V characteristics were measured at regular intervals to understand the degradation mechanism in accordance with ISOS-T1. The results are plotted against PCE, FF, J_sc, and V_oc vs time (hours), as shown in Fig. 3. Upon examining the PCE plot, the inverted devices were found to be significantly more stable than devices fabricated from the conventional configuration, both in BHJ and LBL. The inverted devices did not reach T80 even after 300 h of continuous thermal stress. In the inverted configuration, there was no significant difference between the LBL and BHJ devices, indicating a common degradation mechanism between them. One possible cause could be degradation at the interface of the device rather than in the bulk. This is supported by the fact that there was no change in the J_sc of the devices throughout the thermal stress period. The only observed drop was in the V_oc and FF of the devices. A similar trend was observed in the conventional configuration, where there was no change in J_sc, which remained flat over time. However, there was a slight difference in how V_oc was affected in LBL devices compared to BHJ devices in the conventional configuration. V_oc was slightly more affected in BHJ devices compared to LBL devices. This indicates other competing degradation mechanisms in addition to interface degradation.

To understand the degradation pathways, the light and dark J–V characteristics of the devices were tracked over time, as shown in Fig. S1 (supplementary material). As discussed earlier, the degradation is more severe in the conventional configuration, and the degradation pattern is similar in all four cases, where J_sc remains constant over time, and only V_oc and FF are affected. The dark J–V reveals that, over time, the series resistance of the devices increases in all cases. Although there are no significant changes in J_sc for the fresh and degraded devices, the EQE spectra reveal changes in the spectral response at certain wavelengths in the degraded devices compared to the fresh ones (Fig. S2 of the supplementary material). However, these changes in spectral response do not affect the total integrated current density of the devices. We attempt to investigate the underlying reasons below.

Light intensity-dependent current density and transient photocurrent (TPC) measurements were carried out to understand the carrier recombination and extraction processes in fresh and thermally stressed devices after 336 h. Figure S3(a) shows the TPC results for devices from all four configurations, and Fig. S3(b) shows the current density vs light intensity plot. The values are presented in Table S3.

In the case of conventional BHJ and LBL configurations, the charge carrier extraction time remains largely unaffected for fresh and degraded devices, suggesting that the degradation mechanisms are not associated with morphological changes in the photoactive layer.⁶⁹ Consequently, the observed V_oc drop likely originates from the degradation of one of the device interfaces, leading to recombination effects at that interface. This conclusion is further supported by the absence of significant changes in the α-values (Table S3), power-law exponent, and current density vs light intensity plot for fresh and degraded devices [Fig. S3(b)].

For inverted BHJ and LBL configurations, the charge carrier extraction time increases slightly in degraded devices, from 0.28 to 0.33 μs for BHJ and from 0.26 to 0.30 μs for LBL (Table S3). This indicates the presence of trap-assisted recombination. As the α-value for degraded devices shows minimal change, the conclusion that the degradation predominantly arises from one of the interfaces is supported, as was the case for the conventional device configuration.

Based on the premise that a common degradation mechanism affects all four device configurations, due to their similar lifetime plots, further studies were conducted exclusively on BHJ inverted devices, which perform slightly better in terms of stability. To understand the spectral response changes in EQE and thermal degradation between fresh and 120 h thermally stressed devices, reflectance measurements were taken and plotted alongside their corresponding measured EQE spectra [Fig. 4(a)]. The relative changes in reflectance correlated with changes in EQE, suggesting that the optical properties of one or more layers were altered in the degraded devices. To investigate whether the active layer blend exhibited any optical changes, UV–Vis spectroscopy was performed on both fresh and 120 h thermally stressed films (Fig. S4). The results showed no significant difference in absorbance, indicating that the active layer remained optically unchanged. Furthermore, 2D grazing incidence wide-angle x-ray scattering (GIWAXS) was performed on fresh and thermally stressed (72 h) PM6:Y7-12 blend coated on glass substrates. Measurements were taken [Fig. 4(c1)] at a grazing incident x-ray angle of 0.28°, with an x-ray wavelength of 1.5418 Å. The scattering signal is quite weak with respect to the strong amorphous background from the glass substrate (at Qz ∼ 1.5 Å⁻¹), but the lamellar stack of PM6 is visible at Q ∼ 0.3 Å⁻¹, showing a slightly more pronounced edge-on orientation, although with a significant contribution of randomly oriented crystallites. The 1D-integrated in-plane data in Fig. 4(c2) show no significant changes in the molecular packing structure between the fresh and thermally stressed samples. The observations from the UV–Vis absorbance spectrum and GIWAXS measurements are consistent with the finding that the short-circuit current density (J_sc) remained unaffected in the degraded devices. Subsequently, the top silver electrode was examined under an optical microscope [Fig. 4(b)], revealing that the silver underwent significant morphological changes during thermal stress. The initially smooth electrode surface became rough, with grain-like structures emerging. AFM analysis of the same devices [Fig. 4(d)] confirmed this, showing a marked increase in surface roughness from 7 to 33.5 nm and the formation of larger grains compared to the fresh devices. The morphological change at the silver electrode explains the change in EQE at a certain spectrum range. To further understand the degradation pathway, a partial device stack study was conducted to identify the specific layer(s) involved in the degradation process.

To investigate the degradation further, devices were subjected to the same thermal stress conditions described previously, 85 °C in a nitrogen atmosphere for 120 h, using two different partial device stacks and one complete stack, all compared against fresh devices. The device configurations used in this study are shown in Fig. 5. (a2) a partial stack coated up to PM6:Y7-12 (glass/ITO/ZnO/PM6:Y7-12), (a3) a partial stack coated up to BM-HTL (glass/ITO/ZnO/PM6:Y7-12/BM-HTL), and (a4) a complete device (glass/ITO/ZnO/PM6:Y7-12/BM-HTL/Ag). After 120 h of thermal stress, the device in Fig. 5(a2) was completed by coating a fresh BM-HTL and depositing a fresh silver electrode, while the device in Fig. 5(a3) was completed by depositing a fresh silver electrode. The J–V characteristics of these three thermally stressed devices were then measured and compared with those of a fresh device [Fig. 5(a1)]. Note that the J–V data for the fresh device were recorded before thermal stress and on the same devices used for thermal stress [Fig. 5(a4)].

The J–V characteristics [Fig. 5(b)] revealed that the devices [Fig. 5(a2)] coated up to the photoactive layer (PM6:Y7-12) showed minimal degradation, indicating that the active layer morphology and the interfaces beneath do not change when stressed for 120 h at 85 °C, supporting the above observations in UV–Vis, GIWAXS, TPC, and light-intensity studies. Furthermore, the devices [Fig. 5(a3)] coated up to BM-HTL experienced slight degradation, with a small drop in FF and V_oc. The complete device that was thermally stressed for 120 h degraded more severely compared to the device stressed up to BM-HTL, with the FF dropping significantly from 74.15% to 65.87%. The PV parameters of the devices used for the partial device stack studies are provided in Table S4. These findings support our earlier statement that degradation occurs predominantly at the top interface of the device. Minor instability effects in the BM-HTL may promote this interface degradation, which is amplified in the presence of the silver electrode. These results are in good agreement with previous observations obtained from optical microscopy and AFM characterization. Based on these findings, we emphasize that the PM6:Y7-12 blend remains stable when thermal stress is applied. Therefore, future efforts should focus on identifying alternative, more stable top transport and electrode layer combinations to improve overall device thermal stability. This is an interesting finding that could help address thermal degradation in these devices as the field progresses.

Several studies have reported improved light stability in inverted devices when UV or long-pass filters are used.^63–66 In this work, inverted BHJ devices were selected for the light stability study, as their thermal stability was found to be superior to that of conventional architectures. The devices were subjected to continuous 1 sun illumination under open-circuit conditions using a 450 nm long-pass filter and tested under two conditions: (i) without encapsulation in a nitrogen atmosphere (glovebox) and (ii) with encapsulation under ambient atmosphere (ISOS-L). The resulting photovoltaic parameters, PCE, FF, J_sc, and V_oc, are plotted as a function of time in Fig. 6. Remarkably, the unencapsulated devices under nitrogen showed no measurable degradation even after 800 h of continuous illumination, indicating excellent intrinsic stability when UV and a portion of blue light are excluded. These observations are consistent with previous reports suggesting that photochemical reactions at the interlayer–active layer interface, as well as photodegradation of the active layer itself, are significantly suppressed when UV and blue light exposure is filtered.^63–66 The device parameters before and after the 450 nm long-pass filter application are summarized in Table S5. In the case of encapsulated devices tested under ambient conditions, long-term stability was also observed, with minor decreases in V_oc and FF resulting in approximately a 12% drop in PCE after 900 h of illumination. These findings emphasize the need for improved barrier and encapsulation strategies to prevent moisture and oxygen ingress, which can contribute to the observed performance decline. Nonetheless, the results highlight the strong potential of slot-die-coated devices processed under ambient conditions for achieving long-term operational stability.

It is important to demonstrate the potential of producing large-area devices in the form of modules using the scalable slot-die coating technique. To produce mini-modules, we chose the inverted BHJ structure, which is stable under both thermal and light stress. The same optimized parameters used for small devices were applied to fabricate the mini-modules. For module integration, 6 sub-cells were connected using the mechanical scribing method, as described in Ref. 13. Figure 7(a) shows the I–V characteristics of the device with a photograph of a mini-module included. Figure 7(b) illustrates the schematic of monolithic integration, showing how the cells are connected in series. The device consists of six cells connected in series, each with an active area of 2.3 cm², resulting in a total active area of 13.8 cm².

When characterized under 1 sun illumination (AM1.5G), the mini-modules performed well, yielding strong photovoltaic performance as shown in Table II. Impressively, the best-performing module achieved a PCE of 13.06%, with an average efficiency of ∼12.85%. Compared to small-area devices, this represents an average relative PCE drop of about 9%. This reduction in efficiency is attributed primarily to a decrease in FF, while other photovoltaic parameters (V_oc and J_sc) remain largely unaffected. The loss in FF can be linked to increased series and reduced shunt resistance in the modules. In particular, the series resistance per cell (R_s/cell) in the modules was measured at 4.10 Ω cm² (Table I), whereas in the small-area devices, it was only 1.39 Ω cm² (Table II), representing nearly a threefold increase. This increase is mainly attributed to the sheet resistance of the ITO (∼17 Ω/□) and the 5 mm length of each cell used in module fabrication. In addition to the higher series resistance, the modules also exhibited a roughly twofold decrease in R_sh compared to the small-area devices. Light beam induced current (LBIC) measurements confirmed uniform charge carrier collection across all six individual cells, as the EQE of the cells matched well with the J_sc/cell values obtained from I–V measurements. Although charge collection remained uniform, a few localized defects were observed within the active regions of the module. These defects are believed to provide shunt paths that facilitate recombination, thereby lowering the overall R_sh of the module. Nevertheless, the PCE obtained for the mini-module is among the best recently reported [Fig. 7(e), Table S6] using slot-die methods. We emphasize the use of green solvents and air-coating in this work, which is important for future commercialization steps. This result highlights the promising potential of the developed device stack for OPV module fabrication using the scalable slot-die coating method under ambient conditions, which is crucial for industrial OPV development.

Two commonly used device architectures in organic solar cells, inverted and conventional stacks, with two different active layer blending methods, bulk heterojunction (BHJ) and layer-by-layer (LBL), were compared for their performance and processing capabilities using the scalable slot-die coating method under ambient conditions. The device stability and thermal degradation pathways were also investigated and compared. Irrespective of the device architecture, all devices showed impressive performance when coated by slot-die air-processing, with the best PCE of 15.24% achieved in the case of the conventional BHJ device. The inverted device demonstrated excellent stability under thermal stress. Investigation of the small performance drop revealed that it originates from the HTL and Ag interface of the devices, indicating the need for further studies to understand and stabilize this interface. The light soaking (1 Sun AM1.5G) stability was exceptionally good, with no signs of degradation after over 800 h for the devices tested under a nitrogen atmosphere and only a 12% relative PCE drop for the encapsulated devices light soaked under ambient condition when the devices were protected by a 450 nm long-pass filter. An equally impressive performance of PCE 13.06% was achieved with mini-modules fabricated using the same scalable technique under ambient conditions.

Poly[(2,6-(4,8-bis(5-(2-ethylhexyl-3-fluoro)thiophen-2-yl)-benzo[1,2-b:4,5-b′]dithiophene))-alt-(5,5-(1′,3′-di-2-thienyl-5′,7′-bis(2-ethylhexyl)benzo[1′,2′-c:4′,5′-c′]dithiophene-4,8-dione)] (PM6) and 2,2′-((2Z,2′Z)-((12,13-bis(2-butyloctyl)-3,9-diundecyl-12,13-dihydro-[1,2,5]thiadiazolo[3,4-e]thieno[2″,3″:4′,5']thieno[2′,3′:4,5]pyrrolo[3,2-g]thieno[2′,3′:4,5]thieno[3,2-b]indole-2,10-diyl)bis(methanylylidene))bis(5,6-dichloro-3-oxo-2,3-dihydro-1H-indene-2,1-diylidene))dimalononitrile (Y7-12, BO-4Cl, or Y7-BO) were used as the photoactive layer materials in the fabrication process and were procured from Brilliant Matters. The transport layers used ZnO nanoparticles (Avantama N-10) purchased from Avantama, PEDOT:PSS (Clevios^™ P VP AI 4083) purchased from Heraeus, and BM-HTL purchased from Brilliant Matters. All materials were used as received without further purification.

For the fabrication of all devices, the FOM alphaSC (Fig. S7) machine from FOM Technologies was used.

Devices were fabricated using four configurations: (1) glass/ITO/PEDOT:PSS/PM6:Y7-12 (BHJ)/ZnO/Ag, (2) glass/ITO/PEDOT:PSS/PM6:Y7-12 (LBL)/ZnO/Ag, (3) glass/ITO/ZnO/PM6:Y7-12 (BHJ)/BM-HTL/Ag, and (4) glass/ITO/ZnO/PM6:Y7-12 (LBL)/BM-HTL/Ag. Pre-patterned ITO substrates (Kintec) were used after thorough washing. For cleaning, the substrates were washed with soap solution, deionized water, acetone, and IPA in an ultrasonic bath for 10 min each, then blow-dried using a nitrogen gun. Prior to coating, the substrates were cleaned with ultraviolet ozone for better adhesion of layers. For all the layers, a shim with a width of 11 mm and a thickness of 0.1 mm was used, and a constant coating speed of 50 mm/min was maintained. In conventional configurations 1 and 2, the PEDOT:PSS layer was coated with a pump rate of 0.05 ml/min, where the bed temperature was maintained at 40 °C to achieve a thickness of around 25–30 nm, and then annealed at 130 °C for 20 min in the air. The PEDOT:PSS was diluted in IPA at a 2:3 volume ratio. The BHJ active layer (configuration 1), PM6:Y7-12, was prepared with 17.6 mg/ml (1:1.2 weight) in an o-xylene solvent. During the slot-die deposition of this layer, the coating bed, slot-die head, and solution syringe pump were maintained at 70 °C. The coating was performed at a pump rate of 0.07 ml/min to achieve a thickness of around ∼100 nm. After deposition, the layer was annealed at 110 °C for 5 min in a glovebox before the next layer was coated in the air. For the ZnO layer, a pump rate of 0.03 ml/min was used to achieve ∼25 nm, and it was annealed in the glovebox at 110 °C for 5 min. For the LBL active layer (configuration 2), the PM6 layer from o-xylene was deposited on top of PEDOT:PSS from an 8 mg/ml solution at a pump rate of 0.06 ml/min to achieve a 50 nm thick film, followed by the 50 nm Y7-12 layer, prepared in a 12 mg/ml solution, also deposited at a pump rate of 0.06 ml/min. Both layers were deposited using the same coating temperatures as in the BHJ process. Annealing of the LBL photoactive layer was performed after coating both PM6 and Y7-12 at 110 °C for 5 min inside a glovebox. For inverted devices (configurations 3 and 4), ZnO was deposited on top of the ITO using the same parameters as in the conventional configurations. The BHJ layer followed the same procedure as in the conventional setup. For the inverted LBL configuration, the Y7-12 layer was deposited on top of the ZnO, followed by the PM6 layer, using the same coating parameters as the conventional LBL process. The BM-HTL layer was coated at a pump rate of 0.8 ml/min and annealed at 110 °C in a glovebox to achieve ∼40 nm thickness. Finally, all the devices were completed by depositing a 100 nm silver electrode using thermal evaporation under a deposition pressure of less than 1 × 10⁻⁶ mbar. During all layer deposition, there was a waiting time of approximately one hour before the next layer deposition, primarily due to delays caused by die cleaning and preparation for the subsequent deposition step. During the waiting period, the devices were stored in the glovebox. The thicknesses obtained from the layers are provided in Table S1, along with the optimized coating parameters.

Mini-modules were fabricated in an inverted BHJ configuration: glass/ITO/ZnO/PM6:Y7-12 (BHJ)/BM-HTL/Ag. Pre-patterned 60 × 60 mm² ITO substrates (Kintec) with seven ITO strips were used after thorough washing, following the same cleaning procedure as for small device fabrication. For all layers, a 50 mm shim width was used, and a coating speed of 50 mm/min was maintained. The ZnO layer was deposited at a pump rate of 0.13 ml/min and annealed at 110 °C for 10 min. Next, the BHJ layer was deposited at a pump rate of 0.32 ml/min, followed by annealing at 110 °C for 5 min inside a glovebox. This was followed by the deposition of the BM-HTL layer at a pump rate of 0.36 ml/min, which was also annealed at 110 °C for 5 min. To achieve series connections between layers, mechanical scribing was performed. Finally, a 100 nm silver electrode was deposited using a thermal evaporator under a deposition pressure of less than 1 × 10⁻⁶ mbar.

Voltage sweep measurements were performed using a Keithley source meter unit (SMU) integrated with an Enlitech solar simulator. The instruments were controlled by software provided by Enlitech. The J–V scan was conducted in forward bias, from −1 to 1 V for small devices and −1 to 5 V for mini-modules, with a 20 mV interval and a 10 ms time delay. Measurements were carried out at 1 sun intensity (100 mW cm⁻²), calibrated using a reference silicon cell. A metal circular aperture mask with an area of 0.066 cm² was used for small devices to define their active area.

Quantum efficiency and UV–Vis absorption measurements were performed using the PVE300 Photovoltaic EQE system from Bentham. The scans were conducted over a wavelength range of 300–1000 nm with a 5 nm interval.

An LBIC instrument from InfinityPV, equipped with a 400 nm wavelength laser and a 40 μm beam width, was used to scan an area of 80 × 80 mm².

The instrument used is a Xenocs Xeuss 3.0 equipped with a microfocus Cu target x-ray source, focused and monochromatized with single-bounce multilayer reflective optics. The beam size at the sample was collimated to 0.5 × 0.5 mm² with scatterless silicon single-crystal slits. Measurements were taken at two incidence angles, 0.18° and 0.28°, i.e., just below and above the critical angle for total reflection for the glass substrate, with the latter yielding the strongest signal from the active layer film for the chosen beam size. The data were recorded with a DECTRIS Eiger 4M photon counting detector at 72 mm from the sample.

See the supplementary material for more information on slot-die coating parameters for small and mini-module development of OPVs, dark J–V comparison of fresh and thermally degraded devices, EQE comparison of fresh and thermally degraded devices, transient photocurrent (TPC) lifetime of fresh and thermally degraded devices, light intensity-dependent current density plots and corresponding tabulated values, UV–Vis spectra of fresh and thermally degraded devices, device performance before and after applying a long-pass filter, and a list of performance metrics for reference printed devices.

This project has received the funding from the European Union’s Horizon 2020 research and innovation program, Grant Agreement No. 101007084 (CITYSOLAR). M.M., J.W.A., and E.J. acknowledge the financial support from DFF FTP for the project EPIC-OPV, Grant No. 1032-00326B. M.M., C.Y.H., and K.W. acknowledge the financial support from the Independent Research Fund Denmark, Green Research (DFF Green), for the project LESOT (Grant No. 3164-00305B).

The authors have no conflicts to disclose.

All authors contributed equally to this work.

Eswaran Jayaraman: Conceptualization (lead); Data curation (lead); Formal analysis (lead); Investigation (lead); Methodology (lead); Writing – original draft (lead); Writing – review & editing (lead). Kun Wang: Data curation (supporting); Formal analysis (supporting); Investigation (supporting); Writing – original draft (supporting); Writing – review & editing (supporting). Jani Lamminaho: Data curation (supporting); Formal analysis (supporting); Investigation (supporting); Methodology (supporting); Writing – original draft (supporting); Writing – review & editing (supporting). Chun Yuen Ho: Conceptualization (supporting); Data curation (supporting); Formal analysis (supporting); Investigation (supporting); Methodology (supporting); Visualization (supporting); Writing – original draft (supporting); Writing – review & editing (supporting). Jens Wenzel Andreasen: Formal analysis (supporting); Investigation (supporting); Validation (supporting); Writing – original draft (supporting); Writing – review & editing (supporting). Morten Madsen: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Project administration (lead); Resources (lead); Supervision (lead); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal).

The data that support the findings of this study are openly available at https://doi.org/10.5281/zenodo.15009052.