Perovskite solar cells (PSCs) are among the most promising emerging photovoltaic technologies, due to their high efficiency, comparable to that of silicon solar cells. However, concerns about the stability of these devices remain, despite great progress achieved in recent years. To address these concerns, comprehensive investigations of their stability under realistic operating conditions are necessary. In this Perspective, we will discuss the outdoor testing of PSCs. We will first introduce degradation mechanisms relevant for intrinsic stability, as well as degradation mechanisms due to ambient exposure. Effective encapsulation of PSCs will then be discussed, followed by a summary of achieved progress and discussion of testing protocols and equipment to make outdoor testing more accessible. Finally, challenges and future outlook will be discussed.
Perovskite solar cells (PSCs) have been attracting increasing attention in recent years due to their rapid progress, with record efficiency of 25.7% for single-junction and 29.8% for tandem devices, respectively.1 Both efficiency and stability have been immensely improved since the first reports, but the progress in stability, in particular in tests relevant for real-life applications and commercialization, has been lagging behind the efforts to achieve record efficiencies. In fact, stability has been identified as a key obstacle toward PSC commercialization.1 Consequently, there has been increasing emphasis on the stability testing of PSCs. While the consensus on stability testing of the PSCs using updated International Summit on Organic Photovoltaic Stability (ISOS) protocols with the stated goal of improving comparability between test in different laboratories has been reported recently,2 further work is definitely needed to address the problem of stability. This includes not only further understanding of various degradation processes and the development of methods to counter them and extend device lifetime but also the implementation of more rigorous stability testing. ISOS protocols are primarily intended for testing of small, laboratory-based cells and they are not a replacement for industry testing standards.2 In fact, these tests are insufficient to satisfy the module stability testing standards according to International Electronic Commission (IEC) 61215 standard.1 One very significant difference is that the IEC standardized testing involves performing tests in sequence, which necessarily imposes considerably more strict demands on the device stability compared to passing single tests independently.1
While many stability testing protocols subject the devices to only one or at most two stressors, such as damp heat test that tests stability at elevated temperature and humidity but does not address stability under illumination, outdoor testing by its nature combines all the relevant stressors that can cause device degradation. Stability under combined stressors is recognized to be the harshest testing for PSC stability,3 which is highly relevant for stable outdoor operation. The outdoor stability testing is further complicated by the fact that operating temperature, ambient humidity, sunlight intensity and spectrum constantly vary. While there exist different accelerated aging protocols, both those established in IEC 61215 standard and ISOS protocols, the outcome of these tests cannot be considered as acceptable substitute for a prediction of actual lifetime under outdoor operation,1,4 despite claims from literature on lifetime predictions based on accelerated aging tests. For example, 20 years of operational stability was claimed on the basis of T90 value of 3260 h under damp heat test.5 Similar claims (operational stability exceeding 20 years outdoors) have also been made on the basis of observing 3.5% reduction after 1100 h of stability testing following ISOS-L-1 protocol.6
Such predictions would require accurate modeling of the perovskite degradation under different test scenarios, which is currently lacking as complex effects of multiple stressors are commonly not taken into account. For example, there have been models predicting lifetime and energy yields under outdoor conditions based on degradation kinetics at elevated temperature.7 However, since multiple factors contribute to device degradation, it is unlikely that a simple model would yield an accurate outdoor lifetime prediction. In addition, laboratory aging under maximum power point (MPP) was performed emulating weather data (illumination, temperature) to investigate device behavior under conditions resembling outdoor tests, and devices were found to exhibit both reversible and irreversible degradation.8 However, while illumination and temperature corresponded to actual weather data, the stability tests were performed on unencapsulated devices in nitrogen. In addition, as the processes involved in both reversible and irreversible degradation can be quite complex and different dynamic behaviors are observed in different degradation stages,9 it is expected that adequate modeling of the phenomena involved will likely be complicated. Finally, degradation acceleration factors have been reported for some types of PSCs.10 The acceleration factors have been derived from Arrhenius temperature dependence of the degradation following ISOS-L-3 testing protocol,10 but the devices used encapsulation protocol (polymer films + epoxy + cover glass) that is generally not compatible with damp heat and outdoor stability testing. It is thus critically important to determine acceleration factors for devices and more importantly modules that are properly encapsulated, and validate those acceleration factors. In addition, while alternative laboratory tests incorporating cycling of illumination and temperature resembling outdoor conditions have been proposed,4 these test scenarios have not been validated by extensive comparisons with actual outdoor tests. Furthermore, it should be noted that while such protocols could simulate outdoor conditions better compared to many common laboratory testing protocols, it is not clear how would such testing protocols serve the purpose of accelerated aging, i.e., testing protocols that could be performed within a significantly shorter period of time, which would then guarantee successful performance outdoors for a number of years.
Thus, comprehensive investigation of outdoor stability needs to be conducted to develop performance evolution patterns in different types of cells (planar vs mesoscopic, conventional n-i-p vs inverted p-i-n), since the degradation patterns are likely to differ between conventional architecture planar cells using doped 2,2′,7,7′-tetrakis[N,N-di(4-methoxyphenyl)amino]-9,9′-spirobifluorene (Spiro-OMeTAD) hole transport layer (HTL), inverted planar cells with inorganic HTLs, and conventional triple mesoporous cells without HTL and with carbon electrodes, as different degradation mechanisms would be present in different types of devices. The outdoor stability tests have been scarce until as recently as 2020,2 and the number of publications reporting outdoor stability tests is still smaller than 30. While there have been increased number of reports on outdoor stability of PSCs, further work is needed in this important area, to enable the development of acceleration factors for different accelerated aging tests to correlate the performance in those tests to that under real-life conditions.2 As there are multiple factors contributing to the device degradation, and their interaction can accelerate degradation, it is unlikely that any single accelerated aging test will be able to reliably predict outdoor lifetime. Most likely, a combination of tests on encapsulated devices involving multiple stressors will need to be conducted, such as damp heat (ISOS-D-3) combined with one of the light-soaking protocols (ISOS-L or ISOS-LC protocols) and thermal cycling (ISOS-T) protocols, with the last two possibly combined (ISOS-LT protocols) as a necessary minimum to address the degradation modes due to temperature, humidity, illumination, bias (positive bias as a result of illumination), and mechanical stress at the same time, since all of these would affect the results of outdoor testing. The relationships between the results of accelerated aging tests and outdoor tests then need to be carefully investigated on different types of solar cells. Obtaining more data from extensive outdoor testing of different perovskite devices under standardized conditions (ISOS protocols) is essential for developing comprehensive degradation models and deriving accurate acceleration factors. Therefore, in this Perspective, we will discuss relevant issues for the outdoor stability testing, including both intrinsic stability and encapsulation, and progress achieved to date for different types of PSCs. We will present the progress in outdoor testing for not only for single junction individual devices, but also modules and tandem devices. However, detailed discussion of specific issues relevant to tandems and modules is beyond the scope of this Perspective. Recent reviews and perspectives for modules11 and tandem cells12–14 are available for further details on stability issues specific to these types of devices. Finally, remaining challenges and future outlook will be discussed.
II. DEGRADATION MECHANISMS AND INTRINSIC STABILITY
To achieve stable performance and long lifetime outdoors, PSCs need to have excellent intrinsic stability, as well as to be well encapsulated to prevent ingress of moisture and oxygen from the ambient environment. The perovskite materials and PSCs are sensitive to the exposure to ambient (humidity, oxygen), illumination (both visible and UV illumination, but more seriously UV illumination), elevated temperature, and electrical bias.2 The degradation processes during operation and/or various stability testing protocols occur in the perovskite layer itself, at the interfaces, as well as in charge transport layers (CTLs) and electrodes. The decomposition of the perovskite upon exposure to different stressors generates volatile decomposition products, which can further accelerate device degradation.1 Degradation mechanisms in PSCs have been extensively studied. Detailed discussion of PSC stability, degradation mechanisms, and methods to improve stability is beyond the scope of this Perspective and can be found in review papers on the topic.3,15–18 While all the relevant phenomena are in some cases not yet fully understood, it is important to improve intrinsic stability as much as possible in order to maximize device lifetime under operating conditions. While the negative effects of exposure to humidity and oxygen can be countered by effective encapsulation, and thus they are not commonly considered for intrinsic stability testing, which is conducted under inert atmosphere,2 it is nevertheless important to understand how the degradation occurs upon exposure to moisture and oxygen, particularly since even small quantities of oxygen (of the order of 100 ppm) can result in rapid degradation of the perovskite material under illumination.19 Therefore, we will briefly discuss the degradation under exposure to each of the different stressors, namely, elevated temperature, bias, exposure to humidity and oxygen, and finally illumination.
A. Thermal stability
Perovskite materials, in particular, those based on volatile small organic cation methylammonium (MA), can decompose when exposed to thermal stress, while for other cations thermal stress can lead to undesirable phase transitions.15 The decomposition can be accelerated by water and oxygen and on some semiconductor–perovskite interfaces, such as ZnO/MAPbI3, where deprotonation of MA and decomposition of the perovskite occurs at elevated temperatures.15 Thermal stability can generally be improved by compositional engineering.1,15 Thermal stability is also affected by the device architecture and perovskite deposition method.15 For example, it was found that inverted solar cells using NiOx HTL exhibited improved thermal stability compared to devices with conventional architecture with TiO2 electron transport layer (ETL).20 Furthermore, thermal stability of conventional devices was affected by the ETL/perovskite interface, and the need for passivating surface defects in a metal oxide ETL was demonstrated.20 In general, thermal stress can also result in strain at interfaces, which can result in degradation of performance, and which can also be minimized by compositional engineering combined with device and encapsulation optimization.1 It should be noted that strain management can also have positive effects of improving the perovskite stability,1,18 and strain engineering has an important role in improving intrinsic stability of PSCs. Phase transition can also occur in some perovskite materials at temperatures relevant for solar cell operation, which would result in performance degradation.2 CTLs, in particular doped CTLs and those with low glass transition temperatures, also contribute to the degradation of PSCs at elevated temperature.1,21 In addition, diffusion is accelerated under elevated temperature, which involves perovskite ion migration, as well as electrode metal diffusion.1 Another important consideration concerning thermal stability is that the stability under constant temperature and under temperature cycling can have significant differences in terms of observed degradation patterns.22 Thus, it is important to include some form of thermal cycling in accelerated testing intended to provide relevant information for expected outdoor performance.
B. Stability under bias
Perovskite materials and devices also exhibit degradation under electrical bias, including both positive and negative bias, since electric bias can redistribute charged species in the devices.2 The degradation under reverse bias has been recently reviewed.23 Different mechanisms have been proposed to explain degradation under reverse bias, such as biased induced migration of halides into ETL, accumulation of ionic charges at contacts leading to reverse breakdown, and oxidation of iodide by influx of holes, and metal ion migration inducing the formation of shunts.24 Defects were found to play a role in degradation under reverse bias and illumination, and a negative effect of the deep iodine interstitial defects on the device stability due to defect-induced degradation of perovskite/CTL interfaces was found.25 Reverse bias also induced local heating and local shunts due to metal ion migration, and eventually metal electrode is eroded.26 The absence of metal electrode in carbon-based PSCs thus contributes to their increased stability as the degradation processes related to metal migration do not occur in these devices.26 At individual device level, proposed strategies include the optimization of perovskite composition, defect passivation, grain and grain boundary engineering to inhibit ion migration, device structure, and electrode optimization,23 similar to general strategies for stability improvement. Strategies to reduce thermal and reverse bias stresses in modules have also been outlined.23,24 Reverse bias/hotspot test is recognized as the most severe stressor for module stability, and it occurs due to failure of cells to work synchronously due to defects or shading.1 Nevertheless, there have been reports of mesoporous solar cells with carbon electrodes that are capable of passing this test,1 although this type of testing is the least commonly reported among various perovskite stability tests.
C. Ambient stability: Effects of moisture and oxygen
Halide perovskites generally have poor stability on exposure to moisture, or polar solvents, in general, since these lead to deprotonation of organic cation and the loss of volatile HI, which ultimately results in perovskite degradation.27 Degradation upon exposure to moisture involves the formation of hydrated phases at first, followed by perovskite degradation due to deprotonation of organic cation, and loss of volatile organic amines.15,28 Moisture stability can be improved by incorporation of more hydrophobic layers, including using 3D/2D perovskite materials.15 While the degradation in the presence of moisture occurs in the dark, exposure to oxygen in the dark is generally not as detrimental.15 However, the exposure to oxygen under illumination can lead to rapid degradation of the samples due to photooxidation.15,27 For example, devices under MPP testing exhibit dramatic difference in stability if tested in nitrogen and 0% RH air.29 Under illumination, oxygen serves as electron scavenger and superoxide ion is generated.30,31 This process leaves the excess of photogenerated holes in the perovskite, which leads to iodide oxidation, and consequently ion migration and/or photoinduced segregation of the perovskite.31 The photo-oxidation process is accelerated by the presence of moisture, as well as by elevated temperature.32 Thus, high quality encapsulation is essential for devices expected to operate outdoor, exposed to oxygen and humidity under sunlight, which can lead to increased temperature.
D. Stability under illumination
Degradation under illumination is absolutely critical in PSCs, as solar cells must be illuminated to operate. It is possible for devices to exhibit excellent stability in harsh tests in the dark, including damp heat and humidity-freeze cycle, and still exhibit significant degradation under illumination.33 As shown in Fig. 1, properly encapsulated conventional solar cells survived over 1800 h of damp heat and 75 cycles of humidity-freeze test, but devices exhibited degradation under maximum power point tracking (MPPT) testing.33 Thus, it is critical to understand the processes leading to degradation under illumination, in particular, when combined with additional stressors. Stability tests under illumination are often conducted in inert environments, which is indeed relevant to understand intrinsic stability of the devices. However, it is also highly relevant to conduct stability tests under illumination on encapsulated devices in ambient, since this would be more relevant for actual outdoor testing as it requires both intrinsically good stability and adequate encapsulation.
Illumination can lead to reversible changes, such as photoinduced halide segregation, and irreversible degradation, due to material decomposition, ion migration, irreversible chemical reactions.1 Decomposition under illumination occurs in excess lead halides present,1 as well as in some perovskite materials. For example, commonly used organic–inorganic lead halide perovskite, methylammonium lead iodide (MAPI), decomposes under illumination, which leads to the formation of PbI2, methylamine, HI, and I2 gases.34 Rapid degradation of the perovskite and consequent significant release of volatile decomposition products can result in fracture of the top layers (such as metal oxide conducting layer and metal electrode).34
In addition to perovskite degradation under illumination, interfaces and CTLs can also contribute to the degradation. Ultraviolet photocatalytic degradation has been identified as a significant problem in PSCs using metal oxide ETLs.16 Light soaking in the devices using TiO2 and/or SnO2 ETLs will result in the decomposition of perovskite near the interface.16,35 Efficient extraction of the electrons by the metal oxide results in the oxidation of iodide by excess photogenerated holes, resulting in the loss of I2 and decomposition of the perovskite.16 UV illumination can also result in the desorption of oxygen adsorbed on the surface vacancies, of the metal oxide, and generated free radicals and water will decompose the perovskite layer.16 Although the process is more pronounced on more photocatalytically active TiO2, the degradation still occurs in SnO2 based devices.16 Strategies to minimize this degradation include elimination of UV light component using filters or down-conversion materials,16,36 the modification of metal oxide/perovskite interface, optimization of the perovskite composition, and replacing the ETL material with UV-inactive alternative.16 Stability under illumination is also dependent on the choice of HTLs. For example, it was demonstrated that NiOx resulted in improved stability compared to an organic HTL poly(triaryl amine).34
As perovskite devices exhibit reversible and irreversible degradation under illumination, this needs to be taken into account in stability testing. The updated ISOS protocols have recognized that there are some issues that are specific to PSCs due to the nature of metal halide perovskite materials. This includes the importance of light–dark cycling (ISOS-LC protocols), as PSCs commonly exhibit recovery of the performance in the dark due to ion redistribution in the absence of illumination.2 As this more closely resembles actual outdoor conditions, such tests may be more relevant for obtaining a better estimate of lifetime during outdoor tests. The photodegradation and recovery in the dark were found to be affected by bias condition during aging.37 It should also be noted that stability under illumination is commonly tested at constant illumination levels, while in actual outdoor conditions illumination levels will vary. Since the PSCs exhibit unusually high open circuit voltage under low light conditions,2 this could also have some implications on stability under varying illumination levels. In addition, some PSCs exhibit recovery of performance in the dark, and some exhibit initial enhancement of performance during light soaking. This can result in complex diurnal variations of performance. Consequently, there have also been proposals that a drop in daily energy output by 20% should be defined as T80 instead of the drop in power conversion efficiency (PCE).38 It should also be noted that PSCs can exhibit different trends in efficiency vs time curves. While some PSCs exhibit “burn-in” (fast initial degradation followed by significantly slower degradation), this is not universal for this type of devices.2,29 Burn-in process was attributed to the redistribution of A-site cations in ABX3 perovskite.25 In general, increases or decreases in device efficiency under illumination, which saturate after a relatively short period of time, are likely caused by redistribution of the mobile ions in the device. The existence of various curve shapes, including non-monotonic behavior, indicates that extrapolations of lifetime, especially for more than an order of magnitude beyond the testing time, makes it inappropriate to extrapolate the lifetime.2 It should be noted that in actual outdoor tests it was reported that the shape of diurnal dependence of solar cell response would vary from day to day,39 which illustrates that the PSC stability is quite complex.
E. Ion migration
Ion migration occurs under illumination, electrical bias, and at elevated temperature. Since it is involved in both reversible phenomena, such as photosegregation, and irreversible degradation due to defect creation and irreversible chemical reactions, it is a significant contributor to the performance degradation under illumination.3 Different ions can migrate in halide perovskites, and main candidates have been identified to be hydrogen, methylammonium, and iodide.18,21 As the ion migration is understood to be mediated by defects,21 defect passivation, including bulk defects, grain boundaries, as well as interface defects, can reduce ion migration and improve stability under illumination.1,18,21 However, organic–inorganic metal halide perovskites are inherently unstable under illumination unless the following is simultaneously achieved: all defects are effectively passivated and photogenerated charges are rapidly extracted. Under illumination, photogenerated holes can oxidize the lattice iodide, which leads to the formation of interstitial iodine and iodine vacancy.27 Interstitial iodine can then readily migrate, which can lead to photoinduced segregation in mixed halide perovskites and ultimately to performance degradation in iodide-containing perovskite materials.27 Therefore, device architecture optimization to improve charge extraction (balanced extraction of electrons and holes to avoid accumulation of charge carriers within the devices) can also potentially contribute to improved stability,31 since excess charges (in particular, excess holes) reduce the stability of halide perovskite materials.31,40 The photoinduced degradation and ion migration are further exacerbated by the presence of oxygen, since oxygen acts as electron scavenger leaving excess photogenerated holes in the perovskite material.31
Ion migration eventually results in the reaction between the metal and the halide, which leads to irreversible performance degradation.15 This problem can be reduced by defect passivation,1 introducing electrode interlayers,1,15 or using carbon electrodes since most metals would have tendency to form metal halides.15 Alternatively, transparent conductive oxides can be used as top electrode to achieve excellent stability,15 but deposition conditions and device architecture (buffer layers if needed) need to be carefully optimized to minimize damage to the perovskite. Not only vertical ion migration, which ultimately results in deterioration of interfaces, CTLs, and electrode but also lateral ion migration, contribute to the perovskite device degradation.41 Lateral ion migration is observable at device edges under illumination, as well as forward and reverse bias, and even storage in the dark.41 The distribution of diffused ions is expected to depend on device polarity (inverted vs conventional), and experimental results were obtained on inverted devices.41 Electrode degradation can also be reduced by optimizing the CTL to improve perovskite/CTL interface and thus significantly reduce ion migration.42 As the degradation processes under illumination typically involve ion migration, and irreversible degradation of the performance will occur due to reactions between migrating halide and the metal electrode, carbon-based electrodes are of significant interest for PSCs, as they are non-reactive, and can also have high hydrophobicity providing further protection to the perovskite.43 Carbon based electrodes can be prepared by doctor-blading, screen printing, and transfer from another substrate, which can be achieved by pressing or lamination.43 These methods are of significant interest as they enable avoiding the use of vacuum evaporation and reducing the overall cost and the complexity of the manufacturing process.43
III. STRATEGIES FOR STABILITY IMPROVEMENT
In general, the intrinsic and ambient stability of the devices can be enhanced by optimizing the perovskite layer composition (using mixed halides, 3D/2D perovskites, and various passivation additives), interfaces (defect passivation at interfaces, interface modification layers), and device architecture in general (using more stable CTLs and electrodes, ensuring efficient charge extraction to avoid charge accumulation). Controlling the orientation of the perovskite, as different crystal facets have different stability, can also play an important role in achieving stable devices.44 The use of additives is a very common strategy to improve efficiency and stability of PSCs, as recently reviewed.45 Since a large variety of additives has been reported to date, detailed discussion of these is beyond the scope of this Perspective, and here we will highlight some relevant points without providing extensive discussion. Additives can improve the performance of all types of solar cells, including the more stable triple mesoscopic devices.46 In general, additives can affect the crystallization of the perovskite and its composition,45,47 or passivate defects in the perovskite.45,48 It is important to note that annealing temperature needs to be carefully optimized when using volatile additives,49 and that combination of additives to simultaneously passivate multiple defects can be very beneficial to the device stability.50 It is also worthwhile highlighting the use of additives to passivate grain boundaries as the main pathways for ion migration, which leads to the suppression of ion migration and consequent improvement in stability.51 Furthermore, while additives can potentially improve both efficiency and stability, passivation of deep defects affects the efficiency, while shallow defects affect the stability.48 Consequently, devices can have similar efficiency and dramatically different stability.48 Finally, additive engineering can also improve the stability of doped HTLs.51 Another common approach for improving stability of PSCs is the use of interface modification layers. The use of 3D/2D layers is particularly promising,42,52,53 as well as the use of interface modification layers for top perovskite/CTL interface to suppress ion migration and improve stability.42,54
In general, among different types of PSCs, the superior stability is commonly observed in triple-mesoscopic structure devices, since they have higher tolerance for defects, do not use organic HTLs, and use corrosion resistant carbon electrodes, which makes them less susceptible to degradation under various stressors (humidity, illumination) including reverse bias.1 For a review on this type of devices, see Ref. 55. They can survive outdoors for days even without encapsulation, and with encapsulation lifetime increases and the devices are even resistant to immersion in hot and cold water, as well as acidic and alkaline water.56 However, they also commonly exhibit lower efficiencies, and thus it is possible that a compromise between efficiency and stability will be necessary for PSC commercialization. Another possible alternative is all-inorganic solar cells. Their stability has been recently reviewed.21 Since these materials do not contain volatile organic cations, they are less susceptible to degradation under stresses, which lead to de-protonation of organic cations in hybrid organic–inorganic perovskites.21 However, these materials can still experience structural degradation due to phase transformations that result in performance degradation, and their record efficiencies lag behind that of organic–inorganic lead halide perovskites.21
The encapsulation of PSCs can serve multiple purposes, namely, the prevention of ingress of moisture and oxygen, loss of volatile decomposition products of the perovskite, and lead leakage. The encapsulation should thus exhibit low water vapor transmission ratio (WVTR), have thermal expansion coefficient matching that of the substrate, as well as exhibit high adhesion, chemical inertness, good mechanical strength, low oxygen transmission ratio (OTR), and compatible processing with perovskite materials.1,27 The latter requirement includes not only suitable processing temperature1 but also the lack of generation of volatile products during encapsulation process, which can react with the perovskite.57 High optical transmittance is relevant for encapsulations where the device is sandwiched between front cover glass and backsheet,27 but it is not relevant for packaging designs in which the glass substrate serves as the front glass cover. The encapsulant should also be electrically insulating to minimize the leakage current.27 Encapsulation strategies for PSCs have been reviewed recently.27,58–60
As the PSC research community commonly involves researchers who previously worked on dye-sensitized solar cells (DSSCs) or organic photovoltaics (OPV), encapsulation practices and materials from these fields have been transferred to PSCs. However, common practice of encapsulating organic optoelectronics with epoxy sealing involving a gap and desiccant is not the best practice for perovskite devices,60 and the use of Surlyn, commonly used in DSSC, is also not suitable due to its higher stiffness compared to encapsulants typically used in commercial photovoltaics, such as ethylene vinyl acetate (EVA).61 In general, encapsulation can follow three main approaches, namely blanket, edge, and combined blanket + edge encapsulation, as illustrated in Fig. 2. In addition, there are different methods of connecting the devices to the outside of the package, also illustrated in Fig. 2. Advantages and disadvantages of each type of encapsulation are briefly illustrated in the schematic diagram in Fig. 2, and they are discussed in detail in the following.
Different materials have been used for blanket encapsulation. Polyolefins are generally more suitable than commonly used EVA,39,62 as EVA can release acetic acid when exposed to moisture/damp heat conditions, which then reacts with the perovskite and causes degradation.62 The negative effects of acetic acid were reported to be minimized by primary encapsulation by Kapton, followed by lamination using EVA.63 Other reported encapsulants include poly(methyl methacrylate) (PMMA)/styrene-butadiene(SB), which was used for water immersion, HCl exposure, damp heat, and outdoor tests.64 However, considering the relatively short test times and degradation rates, encapsulation with poly(iso-butylene) (PIB) edge seal is likely to yield better performance. Polyurethane (PU) was reported to be another alternative encapsulant, with reduced lamination temperature listed as an advantage over polyolefin (POE) and EVA.65 PIB has also been used as blanket encapsulant.33
Different materials have also been reported for edge sealing, such as various epoxies and butyl rubber-based sealants. Butyl rubber-based sealants, such as PIB, and related materials are commonly used as edge sealants27,39,66–69 due to their low WVTR, but they have also been used as blanket encapsulant.33,42 In general, it is important to avoid packaging without blanket encapsulant due to the presence of voids in the package which allows the escape of volatile decomposition of the perovskite.33 In addition, edge-only sealing, in particular using epoxies, can result in delamination at elevated temperature even in the absence of the perovskite due to poor mechanical properties and insufficient adhesion.39,60 It should be noted that the width of the edge seal is important for achieved lifetime for a given edge sealant with a known WVTR. The insufficient width of the edge seal can occur for small laboratory-scale cells, especially when each substrate contains 4–6 devices. As a result, devices with excellent outdoor stability could be obtained, despite not being able to reach T80 of 1000 h in a damp heat test.42 Wide edge seals, 1 cm or wider, are needed to achieve long lifetimes based on reported moisture diffusion rates in PIB.60 It was also recently proposed based on modeling results that narrow width edge seals could be improved if they consisted of two bands, one serving as barrier and the other highly moisture absorbing band.70
It was reported that epoxy-based encapsulation as compared to laminated encapsulation is not appropriate for outdoor tests and damp heat tests.39 Thus, laminated encapsulation is preferable for PSC encapsulation. The laminated encapsulation can consist of polymer encapsulant, such as POE or others, and butyl rubber edge seal39,63,65,69,71 or butyl rubber-based material, such as PIB, can be used as both blanket and edge encapsulant.33,42 Although UV curable epoxy has been successfully used in damp heat test,72 this applied to flexible devices, while for rigid devices the use of UV curable epoxy was found to lead to delamination when exposed to 85°C,60 likely due to mismatch in thermal expansion of the epoxy and glass substrate. In addition to poor adhesion to the substrate, UV curable adhesives can also exhibit other undesirable properties, such as ageing and yellowing.58 In our experience, devices with epoxy-based encapsulation will commonly fail outdoor testing due to encapsulation failure (delamination), and laminated devices (PIB-based) are likely to exhibit significantly longer lifetimes during outdoor tests. As the epoxy-based encapsulation is also not suitable for other relevant tests, such as damp-heat, it should in principle not be used except in cases when details of degradation are investigated by various post-ageing measurements where disassembly is required, since the epoxy-based encapsulation is readily disassembled. Vacuum laminator use for encapsulation is recommended as it would result in better reproducibility, but it is not absolutely necessary when PIB is used as both blanket and edge encapsulant to obtain lifetimes exceeding 1500 h outdoors in a humid climate.42 It should be noted, however, that manual encapsulation can lead to observation of different behaviors under outdoor testing.73
In addition to overall packaging architecture, where blanket + edge encapsulation has a clear advantage, the issue of electrical contacts to the devices is also important. Moisture ingress around the electrical contacts can be a significant contributing factor to the performance degradation,74 and careful sealing of the contacts need to be performed to avoid problems. This is likely easier for module samples due to their larger surface area so that they can be compatible with processes and instrumentation for commercial photovoltaics (PV), compared to small area research cells on small area substrates. Encapsulation of devices between barrier films is relatively straightforward for flexible devices. Commercial silicon solar cells are also commonly encapsulated between front cover glass and back cover, with contacts made using conductive PV ribbons. Similar approach can be taken to the encapsulation of PSCs, but in this case, additional front cover glass is superfluous for devices already made on glass substrates. Thus, substrate serves the function of front cover glass, and electrical contacts are achieved via copper ribbons/PV ribbons can be glued to contacts.39 For small laboratory-scale devices, attachment of PV ribbons could also cause problems for some encapsulation strategies. For example, difficulties in using butyl rubber edge sealing were also reported, primarily related to the fragility of copper wires for contacts.75 Cover glass with a cavity and epoxy was thus proposed as alternative, and it was claimed that if encapsulation passes heating in air at 120°C for 5 min it would pass MPP testing in ambient.75 Such encapsulation could also pass submerging in water at 35 and 50°C for 30 min.75 However, epoxy edge seal can lead to complete degradation of devices immersed in water at 85°C within 16 h, while the use of PIB blanket encapsulation results in some devices surviving immersion in hot water up to 254 h.60 Silicon adhesive, commonly used as perimeter sealant outside butyl rubber edge seal,70 can likely be a better alternative to epoxy, due to inferior mechanical properties and adhesion of UV curable epoxies. To ensure good contact for encapsulated laboratory-scale devices, alternative approaches can be considered, as the use of copper wires or PV ribbons could in some cases result in poor contacts,75 and direct use of metal contacts which extend beyond the cover glass can lead to rapid contact degradation upon exposure to humidity for epoxy-based encapsulation.39 Alternatively, indium tin oxide (ITO)/fluorine doped tin oxide (FTO) contact pads outside the area covered by the cover glass can be used, as illustrated in Fig. 2, where there is a gap between the metal electrode and metal contact pad (with underlying ITO/FTO providing the connection) to ensure that no metal is in contact with the perovskite layer. A 2 mm gap was used between the evaporated metal and perovskite (underlying ITO served as connection) for good stability.62
It is also worth mentioning that thin films can play a role in device encapsulation, in particular, when combined with other encapsulation strategies. Thin film encapsulation can protect the devices from harsh conditions, but typically only for short periods of time. For example, damp heat testing for 1 h was reported for PMMA/vulcanized rubber encapsulation76 while atomic layer deposition of oxide films protected devices from immersion in water for 2 h.77 In comparison, the use of PIB blanket encapsulation (with or without epoxy outer edge seal) can result in devices surviving immersion in hot (85° C) water for more than 200 h.60 Thin films are of more interest for flexible devices, but it is expected that for both flexible and rigid devices they will need to be combined with a different encapsulation approach (encapsulant + cover + edge seal) to achieve sufficient stability over long testing time. The use of thin films in rigid devices combined with cover glass encapsulation was demonstrated to result in slower degradation compared to devices encapsulated without thin film.57 Thin films can offer protection based on both material used (for example ALD coating) or from texturing, as demonstrated for achieving superomniphobic behavior for PMMA films.78 Flexible devices in general are of significant interest for practical applications, but the sensitivity of the metal halide perovskite materials to exposure to oxygen and moisture creates difficulty in using common plastic substrates due to their relatively large WVTR and OTR. Consequently, stability tests of flexible perovskite PSCs rarely involve harsh tests at high humidity levels. Nevertheless, the encapsulation of flexible devices has been of interest to attempt to address this problem, and the topic has been discussed in a recent review article.79 Barrier films for encapsulation of flexible devices need to have low WVTR and OTR, and front encapsulant needs to be highly transparent.79 The barrier should also prevent lead leakage as well as outgassing of volatile degradation products, and block UV light.79 Among promising results for flexible PSC stability, the encapsulation of flexible devices, which enabled retaining over 85% of initial performance for damp heat test (1400 h), 50 thermal cycles and ten humidity-freeze cycles, has also been demonstrated.72 Satisfactory performance in damp heat test was obtained for both inverted devices with metal electrode and conventional devices with carbon electrode, while inverted devices with carbon electrode degraded to below 80% of initial efficiency in 1400 h. The encapsulation involved using hydrophobic buffer layer coated on top of the electrode, meltable polymer film, UV curable adhesive, barrier foils, and silicone edge sealant.72 Based on the demonstrated performance, it is possible that such devices may demonstrate good performance in outdoor testing, although the stability should first be verified using some testing protocol involving illumination. Under such scenario, it is likely that HTL-free and metal-free devices may exhibit better stability compared to devices with metal electrodes.
Finally, in addition to stability improvement, encapsulation could potentially serve another important function, and that is preventing lead leakage out of the damaged PSCs. Schematic diagrams of device packaging, including different locations of lead absorbing material, are shown in Fig. 3. Although other parts of the device are predicted to have higher environmental impact compared to lead halides,80 the presence of lead remains a concern since large scale deployment of PSCs could potentially result in significant lead contamination. Lead is a toxic metal,81,82 and it is generally present in PV due to lead-containing solder used for electrical connections in PV modules.81 In addition to general issue of lead-containing solder, a special concern applicable to PSCs, which contain small amount of lead in the perovskite material, is high water solubility of lead halides is particularly concerning as it increases bioavailability and uptake of lead from contaminated soil by the plants and it lowers the limit, which can be considered safe compared to other lead-containing electronics.82 While lead-free PSCs, such as tin-based PSCs, have been proposed as a safer alternative,82 there are also reports that demonstrated higher toxicity of SnI2 compared to PbI2.81 Furthermore, Sn-based cells are expected to have significantly higher environmental impact based on life-cycle analysis, and the majority of the environmental impact does not come from Pb in Pb-based devices.83 Consequently, there has been rising interest into the development of safe-by-design PSCs, using self-healing coatings, surface treatments, and/or lead adsorbing materials to prevent lead leakage.81,84–89 Different materials have been reported to date, and successful encapsulation has been demonstrated for both individual cells and mini-modules,85,86,88 A significant reduction of lead leakage has been demonstrated under different scenarios, such as water soaking or water dripping after damage,85,86,88,89 such as hail or even being rolled over by a car.85 The stability testing reports of devices using such encapsulations include thermal cycling and damp heat tests85 and illumination at MPP for 1000 h at 85°C.87 This demonstrates their potential utility for real-life applications, and it would be highly beneficial for further development to evaluate this type of encapsulation under actual outdoor conditions. It should also be noted that it has been recently proposed that iodide rather than lead is responsible for the observed toxicity of MAPI and PbI2 at low concentrations.90 Inhibition of plant growth was observed for methylammonium iodide as well, which does not contain lead, while lead containing chemicals (lead nitrate, lead bromide) required much higher concentrations to induce growth inhibition.90 Thus, it should be of interest to not only incorporate lead-adsorbing but also iodide-adsorbing materials into the packaging of solar cells. For the development of encapsulation it would be useful to have some consensus on accelerated tests for encapsulation only. For example, the use of accelerated testing for encapsulation using a pressure cooker was also proposed, and extreme conditions (120°C, 100% RH) enable screening of different packaging within one day.62
V. OUTDOOR TESTING RESULTS
In outdoor tests based on ISOS protocols, the devices can be kept at open circuit, kept at MPP using a fixed bias, or the test can be performed with MPPT.2 Outdoor stability testing has been reported for different types of PSC devices, including single-junction cells,35,39,42,56,57,64,71,91–97 modules,65,73,98,99 and tandem cells.66–69,100 The obtained results are summarized in Fig. 4. It can be observed that in some cases excellent stability has been demonstrated over long test durations, indicating that these devices are capable of good performance under realistic operating conditions.
While the majority of the cells subjected to outdoor tests have been encapsulated, there have also been reports of outdoor tests on cells without encapsulation, typically some form of mesoscopic devices, and the lifetime was as expected short.56,92 It should be noted, however, that planar devices if tested without encapsulation fail much faster, as indeed demonstrated on flexible planar devices that completely degraded in 10 h.93 One exception to the usual encapsulation strategies is a triple mesoscopic cell with coated ZrO2 hydrophobic layer, which exhibited improved performance compared to glass + epoxy encapsulation.94 While there exist a number of reports on film-based encapsulation of PSCs, this type of encapsulation is typically insufficient for more demanding stability tests, such as outdoor testing. It is likely that some of the more successful thin film encapsulations can be combined with laminated cover glass encapsulation to further extend the device lifetime.
Some of the outdoor testing results involve the use of encapsulation with epoxy and glass or plastic cover,56,57,66,91,98 while others used various butyl rubber-based encapsulation strategies.39,42,66,68,69,71 A variation in the latter method, with EVA as encapsulant and epoxy resin covering the edges for protection from EVA exposure was also reported.73 Another variation reported was PMMA/SB encapsulation with cover glass or Al foil, where PMMA served to protect the perovskite device and reduce surface roughness, while SB functioned as an adhesive/protective layer.64 In some cases, UV filter was part of the encapsulation package.95,98 Excluding UV part of the spectrum is expected to increase the lifetime, since perovskite is more sensitive to the short wavelength light exposure. Additional factor for consideration is the thermal conductivity of the encapsulation package. It has been demonstrated that using aluminum sheet instead of back cover glass leads to slower degradation, which was attributed to the prevention of heat accumulation.64 An application of luminescent hydrophobic polymer coating on both front and back side of PSCs was also reported to significantly enhance device stability.96 The majority of outdoor tests have been performed on rigid devices, while outdoor tests of flexible devices have been scarce. Nevertheless, some reports exist, including four-terminal tandem cells encapsulated between barrier sheets,100 and devices without encapsulation that failed in less than a day.93
In addition to demonstration of stability under realistic operation conditions, outdoor stability tests have also provided further insights into stability of single cells, modules, and tandem devices. For example, the device stability in outdoor testing was found to strongly depend on the HTL used, with commonly used Spiro-OMeTAD resulting in the worst stability.92 The solar cell response was reported to vary depending on the illumination intensity, with ionic component dominating at low illumination intensities and electronic component dominating at high illumination intensities.91 It has also been claimed that convex pattern of maximum power Pmax can be attributed to interface deterioration, while linear and convex patterns can occur due to humidity-induced degradation, and concave pattern can be attributed to light-induced degradation.73 All three patterns could be observed in the same mini-modules that were manually encapsulated.73 This illustrates the importance of reproducibility of device fabrication and complexities in understanding the degradation mechanisms. Finally, it was reported that PSCs exhibit low temperature coefficients over the range of 25–85°C.102 An energy yield model predicting output power based on temperature and illumination was developed, and the model enabled tracking degradation by monitoring the difference between measured and modeled power output.102 However, it should be noted that temperature coefficients obtained from indoor and outdoor measurements can vary considerably, illustrating the difficulties in fully understanding the ongoing processes during the outdoor testing.103 For example, it has been shown that the temperature coefficient in mini-modules was dependent on the perovskite composition, with mixed perovskite modules exhibiting different behavior (positive temperature coefficient) from those based on MAPI (negative temperature coefficient).98 Due to presence of transient processes on long timescales (several hours for daily variations to several hundred hours for seasonal changes) in general, it is difficult to separate the issues of stability and performance variations due to weather parameter variations (sunlight intensity, device/module temperature, sunlight spectrum, and incidence angle).103 Since different PSC devices (device architecture, perovskite composition) often exhibit different responses to prolonged illumination, it is not surprising that it would be challenging to fully understand the reasons for observed device behavior under outdoor testing. Nevertheless, efforts have been made to try to simplify the testing needed to predict device lifetime. For example, it was proposed that there was a linear dependence between T80 and time for ideality factor nID to reach a value of 2 in perovskite mini-modules consisting of inverted perovskite devices with MAPI active layer.73 This work is potentially interesting as it attempts to connect a practically relevant parameter (T80) to a parameter demonstrating a change in the recombination process (from Shockley–Reed–Hall recombination to multiple trap distribution).73 However, it should be noted that there are complications in the interpretation of the ideality factor, as it is possible to obtain ideality factor close to one or even lower than in situation where surface recombination is the dominant loss process.104 Thus, it is difficult to correctly assign measured ideality factor to any given recombination process, which makes it unlikely that the reported correlation between measured ideality factor and T80 would hold for other perovskite compositions and different device architectures, as well as devices encapsulated using different techniques (vacuum laminator, POE sealant, butyl rubber edge seal).
In tandem devices, outdoor testing of a four-terminal perovskite–perovskite tandem has revealed that a narrow bandgap cell was degrading faster compared to a wide bandgap cell.100 In contrast, in perovskite–organic tandems, the stability of the tandem device was significantly improved compared to both single junction perovskite and single junction organic solar cell.66 A detailed study of the stability performance of perovskite–Si tandem devices revealed that devices exhibit reversible degradation due to ion migration (reversed in the dark but resulting in increasing hysteresis over time) and irreversible degradation of the metal contacts, caused by illumination and exacerbated by temperature.68 Outdoor studies of tandem cells also revealed the need to take into account differences in bandgap thermal dependence of Si and perovskite so that bandgap optimized for operating conditions is different from that at RT.69
As the outdoor performance testing of PSCs is still in early stages, the comparisons between different laboratories/locations for outdoor testing have been scarce.95 While the outdoor testing has been conducted in a variety of locations and climates, there is a significant variation between device architectures and encapsulation methods for comparisons. It is also expected that cell performance in outdoor conditions can vary significantly depending on the location due to variations in incident spectrum.105 Thus, it is essential for further progress to conduct round-robin experiments on the devices fabricated in one location and tested in multiple locations. Round-robin comparison to date has been reported for maximum power measurements for three laboratories in Japan on metastable PSCs,106 and round-robin comparisons for outdoor performance would be very useful. Finally, while there is a definite need for further outdoor stability testing studies (less than 30 reports to date in total since the first development of PSC technology), there has been rapid progress in recent three years, not only in increased frequency of outdoor testing but also in encouraging demonstrations, such as integration of perovskite module with storage operating outdoors107 and even a stand-alone solar farm based on graphene–perovskite panels (nine panels with area 0.5 m2 each, and each panel consisting of 40 modules), which exhibited a T80 as high as 5832 h.63
VI. CONDUCTING OUTDOOR STABILITY TESTS
There are three different ISOS protocols for outdoor testing. ISOS-O-1, which is the simplest testing protocol and involves placing devices outdoors, under open circuit (OC) or MPP (biased at fixed voltage near MPP or active tracking).2 This straightforward protocol simply requires suitable mounting of the devices under testing outdoors, with monitoring of ambient conditions (temperature, humidity, irradiation levels). It should be noted that the degradation under OC condition is expected to be faster compared to MPP.108 ISOS-O-2 protocol involves testing of devices under OC or MPP, with I-V curves measured under natural sunlight.2 ISOS-O-3 is the most demanding testing protocol, which involves in situ MPP tracking, as well as measurement of I-V curves under both solar simulator and natural sunlight.2 Examples of outdoor stability tests reported in the literature using different stability protocols are shown in Fig. 5.
Another important issue in conducting stability tests in general is the number of devices tested. Unfortunately, many literature reports include tests on only one device.2 While including multiple devices on the same substrate for any test performed under OC condition, there are additional considerations for testing multiple devices under MPP (fixed voltage or active tracking). One important issue is the pattern used for etching the indium tin oxide (ITO) or fluorine-doped tin oxide (FTO) electrode. Many small laboratory cells use device architecture with common ITO/FTO electrode. For testing devices on the same substrate under MPP simultaneously, the devices must be electrically isolated from each other, i.e., each device must have its own ITO/FTO contact. Ideally, multiple devices from the same substrate, as well as devices from different substrates, should be tested to fully evaluate reproducibility. In an outdoor test, including multiple substrates should not be an issue as the illumination area is not limited, while it could be a problem for laboratory aging under solar simulator as many research groups use solar simulators with relatively small light spot sizes. In addition to using suitable electrode patterns/geometries for MPP testing of multiple devices simultaneously, additional multichannel instruments are needed for conducting the tests. In general, MPPT for PSCs is complex, due to the existence of hysteresis, which requires modification of conventional perturb and observe algorithms.2 When stability testing is performed under MPPT, hardware and MPPT algorithm used should be reported.2 Aging systems using MPPT testing109 and MPPT algorithms110,111 suitable for perovskite devices have been reported, which enable interested researchers to develop their own systems. It is also possible to obtain a commercial solution, although choices for instruments suitable for stability investigations of small laboratory cells are limited. Companies providing MPPT solutions, including multiple channels suitable for testing of small laboratory cells, include Infinity PV (options including weather channels are available), University of Ljubljana, Shenzhen Lancheng Technology, Shenzhen Puri Materials, and Greatcell Energy. For outdoor testing, any necessary electronics can be placed inside a suitable outdoor electronics box that can be purchased separately from a range of companies. Outdoor testing also requires weather monitoring, including temperature, humidity, and illumination level. Pyranometer can be used for measuring illumination level,91 and illumination level can be incorporated in some commercial weather stations. Weather stations are commonly used to monitor temperature and humidity.100 and there is a wide range of relatively low cost options available. In addition to weather station, additional temperature sensors located close to the device can be used,99 as the device temperature will differ from ambient air temperature and can be relevant for studying device degradation.
The most common testing setups for outdoor testing involve mounting the samples at a fixed angle. However, outdoor stability tests using tracking were also reported.91,95 It should be noted that the use of tracking should be encouraged, as it has been demonstrated that bifacial perovskite–silicon tandem cells can generate 55% more power when mounted on a horizontal single-axis tracker compared to fixed mounting.112 Both types of setups are relevant for roof-mounted installations. Building-integrated photovoltaics (BIPV) are another possible type of installation, but performance on vertical surfaces is not commonly investigated. It should also be noted that additional considerations may be needed for cells intended for BIPV installation compared to roof-mounted systems, and it is necessary to optimize the design of glazing system for such an application to minimize performance losses and optimize heat transfer.113
VII. CHALLENGES AND FUTURE OUTLOOK
As PSC stability is affected by multiple factors, with some of the degradation modes not fully understood at present, deducing relationships between different accelerated aging tests and the outdoor lifetime is expected to be difficult. While machine learning could be a possible solution to this problem,2 this would require obtaining more extensive data for algorithm training, and thus it is essential to obtain more data on outdoor performance of different types of devices. While progress has been made in recent years, due to variations in the device types and architectures further data are needed to conclusively establish trends and likely directions for improvements.
In addition, while the need for different encapsulation compared to simple epoxy + cover glass has also been recognized recently,39 wide scale adoption of these methods is needed for conducting successful outdoor testing. It is also likely that some further development in the area of encapsulation would be necessary, i.e., the development of encapsulants with improved adhesion, which are chemically inert with respect to PSC components so that PIB could be reserved for edge encapsulation only (as its adhesion is non-optimal). Extensive stability testing of encapsulation options involving lead sequestration is also needed, as these options typically involve evaluations of lead leakage but not necessarily extensive standardized stability testing. In these cases, it is essential to conduct tests on the performance of encapsulation on exposure to UV illumination, elevated humidity and temperature cycling, either independently or as a part of accelerated aging protocols. To further facilitate stability testing in general, adoption of harsh multi-stressor tests could be beneficial, since that would shorten the needed testing time and ensure that devices could perform well in an outdoor test. Example of such a harsh test would be damp heat test under 1 Sun illumination.3
A change in other common practices (use of common ITO/FTO electrode), the use of very small substrates, which do not leave sufficient space for encapsulation, is also needed, as well as availability of instrumentation for MPPT suitable for PSC testing. Changes in the publishing practices may also be needed to encourage and facilitate conducting long term stability studies in a field which is generally considered highly competitive and rapidly advancing may also be needed. For example, establishing a journal dedicated to stability and environmental impact of emerging PV, as well as other renewable/sustainable energy technologies, may be beneficial to further development of the field. Training workshops at conferences, establishment of winter/summer schools where researchers in the field could acquire practical skills and experience in device encapsulation and setting up outdoor testing would also likely be beneficial for achieving more widespread outdoor testing.
This work was supported by the Seed Funding for Strategic Interdisciplinary Research Scheme of the University of Hong Kong and RGC CRF Project No. 7018-20G.
Conflict of Interest
The authors have no conflicts to disclose.
Muhammad Umair Ali: Data curation (lead); Formal analysis (lead); Visualization (equal); Writing – review & editing (equal). Hongbo Mo: Visualization (supporting); Writing – review & editing (equal). Yin Li: Data curation (supporting); Writing – review & editing (equal). Aleksandra B. Djurišić: Conceptualization (lead); Writing – original draft (equal); Writing – review & editing (equal).
Data sharing is not applicable to this article as no new data were created or analyzed in this study.