Overcoming Moisture‐Induced Degradation in Organic Solar Cells

Unencapsulated organic solar cells are prone to severe performance losses in the presence of moisture. Accelerated damp heat (85 °C/85% RH) studies are presented and it is shown that the hygroscopic hole‐transporting PEDOT:PSS layer is the origin of device failure in the case of prototypical inverted solar cells. Complementary measurements unveil that under these conditions a decreased PEDOT:PSS work function along with areas of reduced electrical contact between active layer and hole‐transport layer are the main factors for device degradation rather than a chemical reaction of water with the active layer. Replacements for PEDOT:PSS are explored and it is found that tungsten oxide (WO3) or phosphomolybdic acid (PMA)—materials that can be processed from benign solvents at room temperature—yields comparable performance as PEDOT:PSS and enhances the resilience of solar cells under damp heat. The stability trend follows the order PEDOT:PSS << WO3 < PMA, with PEDOT:PSS‐based devices failing after few minutes, while PMA‐based devices remain nearly pristine over several hours. PMA is thus proposed as a robust, solution‐processable hole extraction layer that can act as a one to one replacement of PEDOT:PSS to achieve organic solar cells with significantly improved longevity.

Unencapsulated organic solar cells are prone to severe performance losses in the presence of moisture. Accelerated damp heat (85°C/85% RH) studies are presented and it is shown that the hygroscopic hole-transporting PEDOT:PSS layer is the origin of device failure in the case of prototypical inverted solar cells. Complementary measurements unveil that under these conditions a decreased PEDOT:PSS work function along with areas of reduced electrical contact between active layer and hole-transport layer are the main factors for device degradation rather than a chemical reaction of water with the active layer. Replacements for PEDOT:PSS are explored and it is found that tungsten oxide (WO 3 ) or phosphomolybdic acid (PMA)-materials that can be processed from benign solvents at room temperature-yields comparable performance as PEDOT:PSS and enhances the resilience of solar cells under damp heat. The stability trend follows the order PEDOT:PSS << WO 3 < PMA, with PEDOT:PSS-based devices failing after few minutes, while PMA-based devices remain nearly pristine over several hours. PMA is thus proposed as a robust, solution-processable hole extraction layer that can act as a one to one replacement of PEDOT:PSS to achieve organic solar cells with significantly improved longevity.
become a critical degradation factor in climates with high relative humidity. Often, the impact of water is twofold. It may lead to 1) a drastic loss in fill factor due to an S-shape (second diode) in the current-voltage ( J-V ) characteristics [13,[22][23][24] and 2) a reduction in short-circuit current linked to a loss in active area. [18] The resulting ill-shaped J-V curve originates from deleterious chemical and photochemical reactions at the interface between the absorber layer and the contacts. [23,25] The drop in fill factor of the J-V curve when exposing organic (or perovskite) solar cells to moisture is particularly prominent for devices using PEDOT: PSS (Poly(3,4-ethylenedioxythiophene)-poly(styrene sulfonate)) as hole extraction layer. [15,26,27] We have shown in ideally packaged devices that an important loss mechanism is derived from PEDOT:PSS featuring a large diffusion coefficient for water. [13] Rather than diffusing vertically, water can quickly diffuse from the edge of the device through the PEDOT:PSS layer and consequently inhibit charge extraction. [13,28,29] The resulting loss in active area due to water ingress can be visualized and tracked using electroluminescence or thermographic imaging. [13] In light of these considerations, it is clear that the technologically versatile PEDOT:PSS could be a source for severe depletion in device performance, particularly under damp heat conditions. As such, alternative hole extracting and transparent thin films-ideally processable from solution-are highly desirable. In fact, metal oxides such as molybdenum oxide (MoO 3 ) have been proposed as more moisture-resilient hole conductors when compared with PEDOT:PSS. [29] However, metal oxides with suitable electronic properties are usually processed from the gas phase using physical vapor deposition and may require elevated temperatures, falling short when compared to the excellent solution processability at low temperatures of PEDOT:PSS. There is a critical need in identifying solutionprocessable replacements for PEDOT:PSS, the current industrial state-of-the-art hole transport layer (HTL), to enable highthroughput manufacturing methodologies and fully explore the unique potential of organic photovoltaics. While failure modes in organic solar cells with PEDOT:PSS have been identified before, [15,24,30,31] reliable and valid alternatives for mass production are still not amply available.
Here, we present tungsten oxide (WO 3 ), processed from a nanoparticle dispersion, [32] and phosphomolybdic acid (PMA), processed from solution, [33,34] as hole-transporting layers, which require no postannealing at high temperatures and are robust under prolonged damp heat exposure and thus represent valid alternatives to PEDOT:PSS.

Results and Discussion
To set the focus of this study on the charge extraction layers, we prepared devices using P3HT:PCBM as the photoactive blend. We have previously shown that P3HT:PCBM is resilient against high levels of humidity and should not form a stability bottleneck here. [13] Exposing solar cells with the layer stack ITO/PEI/P3HT: PCBM/PEDOT:PSS/Ag to continuous damp heat conditions (85°C/85% RH) leads to the formation of an S-shape (second diode) in the current-voltage ( J-V ) characteristics of the device ( Figure 1). This typical evolution starts with a decrease in fill factor (FF) and later also affects the short-circuit current ( J sc ) before it eventually even lowers the open-circuit voltage (V oc ). This found behavior is the typical signature observed in j-V curves after degradation of organic solar cells of inverted architecture with PEDOT:PSS-based HTLs in damp heat, [31] an increased series resistance, which turns into an S-shaped J-V curve upon prolonged exposure to humidity, sometimes accompanied by a loss in J sc . [30] To find the origin of this degradation, a systematic study exposing partially finished devices to humidity (see Figure 2a) is conducted. Subsequent to damp heat exposure, the devices are finished as usual and measured. (Note: For all devices, we omitted the annealing step that is commonly performed to improve the P3HT:PCBM morphology in order to avoid any potential reversibility of the humidity effect on the glass/ITO/ PEI sample exposed to damp heat). The results can be seen in Figure 2b. The half-finished device that has been processed until the electron transport layer (ETL) before being exposed to damp heat shows the same photovoltaic performance as a reference device that has never been exposed to humidity. The half-finished device that contains additionally the AL even shows a slightly better performance (J sc ) than the reference, which could be explained by an annealing effect in the climate chamber (85°C) that may have slightly improved the AL morphology. Finally, the device that also contains the HTL during damp heat exposure shows a strong S-shape.
From the j-V curves in Figure 2b, we can conclude that neither PEI nor P3HT:PCBM are harmed by humidity. The results indicate a strong connection of the damp heat degradation to PEDOT:PSS, which also has been widely discussed in the literature. The following reasons for the degradation of PEDOT:PSS by humidity have been suggested.
There is wide agreement that the appearance of an S-shaped J-V curve is caused by the formation of a layer of reduced conductivity across the bulk heterojunction/PEDT:PSS interface. [15,35,36] Although the bulk conductivity of PEDOT:PSS decreases upon exposure to humidity, it is not sufficient to account for the formation of an S-shaped J-V curve. Kawano et al. suggested that an www.advancedsciencenews.com www.aem-journal.com increasing fraction of the PEDOT:PSS/blend interface becomes insulating, resulting from the reaction of acidic species in PSS with water. [15] Sharma et al. reported an interface dipole between PEDOT:PSS and the electrode, which they make responsible for S-shape formation. [24] Corazza et al. and Dupont al. invoked delamination induced by swelling of the polymer, which would cause a contact resistance between the AL and the HTL. [37,38] Other authors find morphological changes of PEDOT:PSS in the presence of water, such as phase separation between PEDOT and PSS, resulting in thin layers of nonconductive PSS in the HTL. [39] Additionally, changes in work function (WF) and ionization energy of PEDOT:PSS upon exposure to humidity have been observed, which were attributed to "dedoping" of the HTL. [24,40] While the observed role of PEDOT:PSS can be related to its known hygroscopic behavior, [19,41] the dominant effect of PEDOT:PSS on the device degradation might be linked to different aspects, such as 1) chemical reactions with the AL, 2) reduced mechanical contact to adjacent layers, and 3) altered electronic properties, three different aspects, which are further examined in this work.
First, devices are again fabricated only until the PEDOT:PSS layer and then placed under damp heat (85°C/85% RH) for 2 h. The PEDOT:PSS layer is then removed from half of the samples by applying a tape peel-off process, which results in clean delamination between the AL and the PEDOT:PSS. [19,42] Subsequently, a fresh PEDOT:PSS layer is coated onto the peeled samples and, finally, all devices are completed by evaporation of a silver top electrode.
The resulting J-V curves are shown in Figure 3a (one representative curve per variation, namely, a nondegraded reference cell, and two damp heat degraded cells, of which one received a fresh PEDOT:PSS replacement). As already shown in the previous experiment, the device that contains the 2 h-degraded PEDOT:PSS layer exhibits a strong S-shape. However, if this degraded PEDOT:PSS layer is peeled off and replaced by a new one before silver evaporation, the resulting device shows a performance very similar to an undegraded reference device, with only a slight loss in V oc . This confirms that no major chemical reaction with the AL has taken place.
Secondly, potential (partial) delamination of the PEDOT:PSS layer from the AL is investigated by laser beam-induced current (LBIC) measurements, where a helium-neon laser was used to illuminate the movable solar cell device, point by point, to extract a 2D image of the photocurrent of the solar cell active area. To this end, complete devices with PEDOT:PSS as HTL are exposed to damp heat conditions for 2 h. The LBIC image of such a device shows an inhomogeneous distribution of current (Figure 3b, right) on different microscopic levels, in contrast to an undegraded sample (left). On the one hand, there are regions of several mm 2 that possess significantly lower current values than other parts of the cell. On the other hand, the same behavior can be found on a (one order of magnitude) smaller length scale, that is, throughout the active cell area there are areas of %100 Â 100 μm 2 , which are surrounded by areas of the same size with significantly higher current. This behavior is especially pronounced in the "mm 2 areas" with lower currents. This observed inhomogeneity of photocurrent might be attributed to local water ingress through the top electrode, that leads to local swelling of the PEDOT:PSS, which in turn partially lowers the electrical contact between the hole transport layer and the AL. [41] Moreover, the heat-induced phase segregation of the AL leads to migration of PCBM toward the AL-HTL interface and the formation of PCBM crystallites at this position, which has been reported to enhance delamination processes. [41,43,44] However, despite a certain degree of current inhomogeneity over the active cell area, no real "dead spots" are observed, which indicates that the PEDOT:PSS is still everywhere in contact with the AL, that is, no complete delamination takes place. Additionally, the least degradation is observed at the top-most part of the cell, close to the edge that is defined by the laser line instead of the silver top electrode (i.e., there is also evaporated silver beyond this line, which is where the top electrode is contacted). On all other three edges, "crack-like" structures appear, which indicate an additional pathway for water to enter the cell laterally. Figure 2. a) Partial device structures being exposed to damp heat (DH, 85°C/85% RH) for 2 h, each one being at a different stage of the device fabrication process, before being fully finished. b) Current-voltage characteristics of all fully finished devices. The reference device has never been exposed to humidity.
www.advancedsciencenews.com www.aem-journal.com With LBIC being able to spatially resolve the reduced J sc of the device, which is caused by the S-shape of the J-V curve, this finding is also in line with the observation that under 1 sun illumination; still the complete photogenerated current of %9 mA cm À12 can be extracted from the 2 h-degraded solar cell, if a reverse bias of À1 V is applied (see brown J-V curve in Figure 3a).
Third, Kelvin probe measurements (Figure 3c) were performed in a dark environment on devices, that were coated until the PEDOT:PSS layer and placed under damp heat for 2 h, whereas freshly peeled highly oriented pyrolytic graphite (HOPG) with a WF of 4.65 eV was used to calibrate the reference tip. It can be observed that the WF of the PEDOT:PSS layer decreases when exposed to damp heat compared to the nonexposed reference layer. The average WF decreases by almost 0.2 eV from 4.68 to 4.49 eV and may thus create an extraction barrier for holes. Moreover, the statistical spread of WF values also increases, that is, there are measurement spots that show very severe WF losses of even %0.4 eV, which again suggests an inhomogeneous lateral distribution.
In conclusion, the reason for the decrease in device performance is a combination of locally reduced electrical contact between the AL and the hole-transport layer, as proposed by Dupont et al. and Voroshazi et al. [29,41] and the decrease in WF of the PEDOT:PSS layer, of which the first seems to have the bigger impact. We hypothesize that as long as delamination occurs on a scale smaller than the carrier diffusion length in PEDOT:PSS, delamination expresses itself as an increased contact resistance (leading to an S-shape, due to the formation of a space charge), rather than a reduced J sc (caused by the reduced contact area).
Due to the conclusion that damp heat degradation is mainly related to the PEDOT:PSS, two other HTL materials, namely, tungsten oxide (WO 3 ) and phosphomolybdic acid (PMA), are evaluated with respect to their behavior under damp heat. As displayed in Table 1, the respective devices show a similar initial performance for all three of these HTL materials.
Upon exposure to damp heat for four hours, WO 3 and PMA yield significantly more stable devices compared to PEDOT:PSS, as depicted in Figure 4a. It is to note that for all devices-despite not being encapsulated-water ingress is moderated to a certain extent by the 100 nm-thick evaporated silver top electrode, which results in a generally slower degradation in comparison to previous experiments on half-finished devices. PMA shows the highest stability with only a slight drop in FF but constant J sc and V oc , while WO 3 shows mainly a loss in J sc maintaining its V OC and high FF values (see Figure S1, Supporting Information). For both WO 3 and PMA, no S-shape (second diode) is formed upon degradation (see Figure 4b,c). Remarkably, there are no significant changes in the J-V characteristics of unencapsulated P3HT: PCBM cells using PMA as HTL during the course of 4 h in damp heat (85°C/85% RH), while PEDOT:PSS-based devices are completely nonfunctional after such treatment.
The reasons why PMA and WO 3 are superior to PEDOT:PSS in terms of resilience to humidity lie in their chemical structure. Metal oxides are particularly promising as these materials are at a high oxidative state and often relatively inert toward moisture. PMA has been shown to be rather insensitive to environmental conditions and even to baking at relatively high temperatures. [45] As PMA already contains 12 molecules of water in the crystal structure, it does not undergo structural or chemical changes in the presence of humidity, which would lead to morphological changes, with subsequent delamination, or changes in WF, respectively. device having been exposed to damp heat (DH, 85°C/85% RH) for 2 h before evaporating the top electrode, blue: in damp heat for 2 h and subsequently being overcoated with a fresh layer of PEDOT:PSS before electrode evaporation. b) LBIC scan of a P3HT:PCBM-based device with PEDOT:PSS HTL being exposed to damp heat for two hours (right), compared to a reference device, that was not exposed (left). c) Kelvin probe measurements performed on samples being exposed to damp heat for different times. In order to demonstrate that PMA also works with other, more efficient donor-acceptor material systems, P3HT:O-IDTBRbased devices are fabricated using PMA as hole-transport layer. This AL system is particular of high industrial relevance, since it can be deposited in ambient air and from green solvents, performing comparably with processing from halogenated solvents, [46] and also has been shown to be highly stable under constant illumination. [47] These P3HT:O-IDTBR-based devices, together with P3HT:PCBM-based devices, are exposed to damp heat conditions (85°C/85% RH) for an extended period of time (%100 h) to investigate potential long-term degradation effects. The results of this lifetime experiment are depicted in Figure 5 (see Figure S2, Supporting Information, for absolute values and individual performance parameters). During the first 6 h, the P3HT:PCBM cells lose 8% of their initial power conversion efficiency (PCE), while the P3HT:O-IDTBR cells gain 10% of their initial PCE (mainly due to an increased J sc , which may be an annealing effect that improves the AL morphology). After that, both types of devices exhibit a similar degradation rate, reaching 59% (P3HT:PCBM) and 82% (P3HT:O-IDTBR) of their initial PCE after 95 h in damp heat, which are excellent values given the fact that the devices are not encapsulated. These results corroborate the device's extraordinary resilience against humidity, if PMA is used as HTL. The reason for this is most likely the lower hygroscopicity of PMA in comparison with PEDOT:PSS.
To conclude, we find that the choice of HTL has a big influence on the moisture resilience of OPV devices. Therefore, in order to be suitable for OPV application, we propose that an optimal HTL material should be soluble in benign solvents, have suitable electronic properties, have a high level of transparency, be inert against oxidation, be hydrophobic, and form robust electronic interfaces. While WO 3 and PMA fulfill most of the properties above, PEDOT:PSS likely falls behind in the case of the last three points when compared with WO 3 and PMA.

Conclusion
We have observed a strong degradation of P3HT:PCBM-based organic solar cells upon exposure to damp heat conditions. Through examining the effect of high humidity and temperature on each layer individually, we were able to pinpoint PEDOT:PSS as the origin of degradation. Especially a partially reduced electrical contact between the AL and the PEDOT:PSS seems to be responsible for the photovoltaic device performance loss.
Moreover, we demonstrate that we can minimize this moisture-induced degradation by exchanging PEDOT:PSS with other hole transporting materials, such as WO 3 or PMA, the latter providing the highest stability. We therefore promote PMA as an interlayer material that shows a remarkably high resistance toward high temperature and humidity by preventing delamination between the AL and the HTL and thus represents a promising material candidate for the successful commercialization of stable organic solar cells.
Device Fabrication: ITO glasses by VisionTek (15 ohms/square) were treated with an ultrasonic bath for 15 min in acetone and for another 15 min in 2-propanol. UV/ozone treatment for 10 min was applied prior to layer deposition. To obtain the active material ink, P3HT and PCBM  www.advancedsciencenews.com www.aem-journal.com were blended at a ratio of 1:0.7 and dissolved with a total concentration of 17 mg mL À1 in o-xylene:1,2,3,4-tetrahydronaphtalene (7:1) and stirred under nitrogen atmosphere over night at 80°C. In the case of P3HT: O-IDTBR, the materials were blended at a ratio of 1:1 and total concentration of 30 mg mL À1 and dissolved in chlorobenzene with 5 vol% p-bromoanisole and stirred under nitrogen atmosphere at 80°C over night. The electron transporting layer (PEI, 0.1 wt% in n-butanol), the active layer (AL) (P3HT:PCBM), and the hole transport layers (PEDOT: PSS; WO 3 and PMA) were all applied by doctor blading in air. The PEI films were annealed for 10 min at 100°C in air. All devices were annealed for 5 min at 140°C under nitrogen atmosphere after application of the AL unless stated otherwise in the Results and Discussion. WO 3 (2.5 wt% in isopropanol) was ultrasonicated for 5 min prior to application and the films subsequently annealed at 80°C for 5 min inside the glovebox. PMA was dissolved in 2-propanol (1 mg mL À1 ) at room temperature and no postannealing step was applied after blade coating. As a last step, 100 nm of silver (Ag) were thermally evaporated to obtain devices with an active area of 26 mm 2 .
Damp Heat Experiments: The damp heat experiments were carried out according to the standardized test protocol ISOS-D-3 in a climate chamber by ESPEC at 85°C/85% RH. [48] Characterization: The solar cells were measured using a Keithley source measure unit and a class AAA solar simulator by LOT Quantum Design providing 1000 W cm À2 AM 1.5 G illumination.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.