Introduction

The exploitation of the solar energy, most typically the photovoltaic (PV) application, is a pivotal way to realize carbon neutrality1. PV installation has been growing, and is expected to reach 4500 GW globally by 20502,3. Present commercial PV production is dominated by crystalline silicon (c-Si) solar cells and significant progress has been made4. To make photovoltaics the most competitive energy choice for power generation, researchers are encouraged to explore alternative PV technologies with better cost performance.

Among emerging PV materials, metal halide perovskites have aroused enormous research interest owing to their excellent optical and electrical properties. The perovskite consists of the monovalent cation A (such as methylammonium (MA+), formamidinium (FA+) or Cs+), the divalent metal cation B (such as Pb+ or Sn+) and the monovalent anion X (such as I−, Br− or Cl−) to form an ABX3 formula. Recently, perovskite solar cells (PSCs) have achieved extraordinary progress in reaching high record power conversion efficiency (PCE) of 25.7%5 for single-junction ones, 28.0% for perovskite/perovskite tandem ones and 29.8% for perovskite/silicon tandem ones6. Moreover, the fabrication cost of PSCs has been estimated potentially to be half as low as that of c-Si solar cells7. These attractive achievements encourage the initial attempts of evolving PSCs from research items into commercial products8,9,10,11,12.

Besides the high PCE, high stability is another ticket to the PV market since the levelized cost of energy for PV power generation is comprehensively determined by the PCE, the material and fabrication cost, and the lifetime of the solar panels. A PV product is expected to be stable against diverse operational environments under global application scenarios, especially under harsh conditions like extremely high/low temperatures and light soaking. Some stability tests with intentionally harsh conditions to accelerate the module aging are established to extrapolate the potential long-term operational performance of solar modules. Among them, the International Electrotechnical Commission (IEC) series tests are the most widespread. The details of test conditions and passing requirements are gathered to be the IEC 61215 standard. Figure 1 and Table 1 show the simplified flowchart and protocols of module stability tests based on IEC 6121513,14. The test results are used to primarily judge the feasibility of the production. Overviewing from the literature, we can find that PSCs have the ability to pass parts of IEC 6121515,16, while the following problems still exist: (1) The test conditions are not strictly in accordance with the standard. For example, testing devices in N2 or other innert atomosphere instead of air. (2) Passing single test instead of a test sequence. For example, Fig. 1 shows that ultraviolet (UV) precondition test, thermal cycling test and humidity freeze test are carried out in sequence, which means the requirements for device stability are much higher than those for passing a single test. (3) The comprehensive degradation mechanisms of PSCs remain to be further clarified and specific standards for evaluating the stability of perovskite devices should be formulated as suggested by the consensus of the perovskite community17.

Fig. 1: Full test flow for design qualification and type approval of PV modules13,14.
figure 1

The tests that have already been reported for PSCs are highlighted in orange. STC, standard test condition; NMOT, nominal module operating temperature.

Full size image
Table 1 Summary of important tests of the IEC 61215 for PSCs13.
Full size table

From Fig. 1, we can find that light, heat, moisture and reverse bias are the main threats for solar cells to face under outdoor working conditions in addition to the mechanical stress. In this review, we retrospected the main degradation mechanisms of PSCs under those stimulations and summarized the improvement strategies with some remarkable work. All components of PSCs would bring stability issues and careful design of each layer is essential for stable PSCs. In addition to the materials themselves, related interfaces also contribute a lot to the stability. As for the halide perovskite, attention should also be paid to the grain boundaries (GBs). We proposed that specific standards for PSCs can assist in promoting research in the field. Accelerated aging tests cannot cover all the circumstances that a PV product would encounter during operation. Outdoor demonstration is a good opportunity to gain deeper understanding of operational stability. Success in it is quite meaningful to boost the confidence of investors and customers.

Stability of PSCs

Temperature and humidity stability

Solar cells in practical applications are supposed to cope with varied weather conditions, of which temperature and humidity are the crucial factors. In the IEC standard, three stability tests of thermal cycling, damp heat and humidity freeze correlate closely to the two factors. As seen in Fig. 1 and Table 1, requirements for these three tests resemble each other. The stability issues against temperature and humidity can be overviewed from the perovskite material, the functional layers, the interfaces and the back electrode.

Chemicals including water and oxygen molecules from the ambient condition could react with halide perovsktite and speed up the decomposition (Fig. 2a)18. MA+ and H2O form strong hydrogen bond, weakening the original bond between MA+ and the Pb-I octahedral. The deprotonation of the organic cation is accelerated and the proton could transfer to I− via H2O, producing volatile species of CH3NH2 and HI, while the reductive I− could be oxidized by the oxygen. When the ambient chemicals are well blocked outside, they will not damage the perovskite. Organic-inorganic hybrid halide perovskites including MA-based and FA-based ones also face the thermal stability issue19. From the thermogravimetric analysis, hybrid halide perovskites exhibited a high decomposition temperature over 250 °C (Fig. 2b)20,21. This temperature is much higher than the required 85 °C for terrestrial photovoltaic panels. However, decomposition of those organic-inorganic halide perovskites at 85 °C due to the volatilization and loss of various organic species in the open space was reported19. The loss of organic salts from hybrid perovskites could happen in different pathways which are determined by the components in perovskites and heating temperatures (Fig. 2c). The heat-induced degradation products of halide perovskite involve various gasses. As described, heat and moisture-induced decomposition pathways involve the escape of gaseous species from the device and the entry of species into the device. When the outgassing process is suppressed, the decomposition reaction tends to reach equilibrium before significant damage occurs to perovskites. With this aspect, the less volatile FA+ and inorganic Cs+ are also usually used as alternatives for MA+. However, non-photoactive yellow δ-phase of FAPbI3 and CsPbI3 rather than the black phase is thermodynamically stable at room temperature. Synthesis of low dimensional perovskite materials, compositional engineering and addition of external additives are three main approaches to improving the phase stability of FAPbI3 and CsPbI3 reported in the literature22. The stability of perovskite phase is the very premise of the long-term stability of PSCs.

Fig. 2: Thermal stability of halide perovskite components.
figure 2

a MAPbI3 reversibly forms monohydrate or dihydrate with moisture. Insertion of H2O weakens the bond between MA+ and Pb-I octahedral, speeding up thermal decomposition. Adapted with the permission from ref. 18 Copyright © 2015, American Chemical Society. b Thermogravimetric curves of MAPbI3 and FAPbI3, both of which show high decomposition threshold over 250 °C. Adapted with the permission from ref. 21 Copyright © 2019, American Chemical Society c Released gas species of several organic salts detected by chromatography–mass spectrometry upon heating at different temperatures. Thermal decomposition of perovskites could occur at the relatively low temperature of 85 °C. d Blanket and edge-sealing encapsulation for PSCs. Suitable blanket encapsulation provides a pressure-tight environment to suppress the outgassing process. c, d Adapted with the permission from ref. 23 Copyright © 2020, The American Association for the Advancement of Science.

Full size image

Encapsulation is effective for preventing those unwanted ingress and escape actions as all solar cells need to be well encapsulated before their practical applications. The encapsulation materials and methods are of vital importance to gain the satisfactory lifetime. A hermetic encapsulation scheme creates a pressure-tight environment and prevents the escape of gaseous decomposition products, retarding the degradation of PSCs (Fig. 2d)23. A suitable encapsulant should have low water vapor transmission rate to minimize the harm of moisture ingress. Thermal expansion coefficient (TEC) for the encapsulant should match that of the substrate to keep moisture blocking throughout the stability test. Otherwise, invading moisture freezes to cause further breaking of the encapsulation during the humidity freeze test. Other requirements for encapsulation materials include high adhesion, suitable processing temperature, chemical inertness, low oxygen transmission rate, good mechanical strength, etc24. PSCs that pass the IEC damp heat, humidity freeze or thermal cycling tests are dominantly based on a glass-to-glass encapsulation scheme, where PSCs are sandwiched between two glass sheets that are cohesive with the polymer encapsulant (Fig. 2d)23. With the encapsulation of glass/Ethylene vinyl acetate (EVA) or polyolefin (POE)/glass and the butyl rubber edge seal, PSCs achieved ~108% of its initial PCE after 1000 hours’ aging at 85 °C/85% RH and 200 thermal cycles25,26. Encapsulating perovskite solar cells with a piece of back cover glass and filling the space between the substrate and the back cover with the polymer of POE or polyisobutylene made MA-containing PSCs survive more than 1800 hours of damp heat test and 75 cycles of humidity freeze test, exceeding the requirement of IEC61215:2016 standard23. Inorganic compounds such as the indium-doped tin oxide (ITO) or MgF2 were also used to form the pressure-tight coverage and inhibit the thermal decomposition of halide perovskites in a similar manner. In addition, the escape of MAI mainly starts with the GBs and surfaces of halide perovskite films. Strengthening the GBs also benefit the thermal stability of halide perovskites. Preventing the temperature increase of PSCs under working conditions is also helpful.

Heat also challenges perovskite-related interfaces. Under thermal stress, the mismatch of TEC between the perovskite and the substrate results in strain in the perovskite film. The strain can reach as high as 50 MPa and become a driving force of film deformation and ion migration at the interface27. The mismatched TEC also brings about delamination of the layered structure (Fig. 3a)28,29, and fracture could even occur due to the low fracture energy of PSCs during the thermal cycling test and the humidity freeze test30. The strain can be alleviated with compositional engineering, low-temperature annealing, high TEC substrate and transporting layer selection27. An encapsulant with low elastic modulus and high mechanical strength enables mechanically stable PSC31. Strain is not always harmful since it can also be a tool to enhance the phase stability of FA or Cs perovskite. Masi et. al. proposed to stabilize FAPbI3 perovskite with embedded colloidal PbS quantum dots (QDs)32. PbS acted as a template, heading the growth of the perovskite due to its similar lattice parameter to FAPbI3. This template-grown FAPbI3 strained in both directions parallel to and perpendicular to the PbS-FAPbI3 interface, facilitating a stabilized black phase. Strain may be altered and relieved under thermal stress, inducing the loss of the stabilization effect. PbS provided deeper stabilization mechanisms of surface energies and chemical bonds beside strain. The PbS-FAPbI3 interface reduced the thermodynamic preference for the yellow phase by creating stronger chemical bonds with the black one. The increase in surface energies was small for the black phase but large for the yellow one, which selectively favored the black phase over the yellow one. For the counter interface, ions such as I− from halide perovskite could diffuse across the interface into the adjacent charge transport layer (CTL) and even to the back electrode and react with it. Toughening the surface of the perovskite film or preparing an ion diffusion barrier layer between the perovskite layer and the adjacent CTL makes sense for solving this concern33,34.

Fig. 3: Thermal stability concerns other than perovskite components.
figure 3

a. Cross-section SEM images showing delamination at the perovskite and substrate interface aged by strain. The strain can be produced under thermal stress due to the mismatched TEC. Adapted from ref. 29 b Formation of voids in t-BP/Li-TFSI co-doped Spiro-OMeTAD under thermal stress. Adapted with the permission from ref. 35 Copyright © 2017, American Chemical Society c Time-of-flight secondary-ion mass spectrometry of multiple ions in the fresh (dashed lines) and the 85 °C–1000 h-aged (solid lines) PSCs. Li+ from Spiro-OMeTAD could diffuse into the perovskite layer upon heating. Adapted with the permission from ref. 39 Copyright © 2021 Wiley-VCH GmbH d Reconstructed elemental 3D maps for the ions in 70 °C-aged devices. Au atoms could diffuse in PSCs. Adapted with the permission from ref. 43 Copyright © 2016, American Chemical Society e A compact-Bi layer beneath the metal electrode to block the diffusion of both the metal electrode and the external factors. Adapted from ref. 44.

Full size image

Organic HTMs have long been recognized as a contributor to the instability of PSCs. For the most widely used hole transport layer (HTL) of 2,2',7,7'-Tetrakis[N,N-di(4-methoxyphenyl)amino]-9,9'-spirobifluorene (Spiro-OMeTAD), it would undergo morphology change via outgassing or crystallization under heat, leading to the formation of voids and failure of separation (Fig. 3b)35,36. Additionally, migrated I− and gaseous iodine product from perovskite deteriorate its hole transport ability37,38. The commonly used Li-TFSI dopant could attract water due to its hygroscopic nature, which in turn causes the moisture-induced decomposition of the perovskite. Li+ could also migrate in the device and cause performance decay (Fig. 3c)39. Structural modification of existing hole transport materials (HTMs), alternative dopants and novel dopant-free HTMs are attempted to solve the HTL-induced instability, which have been summarized in previous review articles40,41. HTL-free device configurations are also good choice for solving this stability concern42.

Atoms from the metal back electrode such as composed of Au or other metals could diffuse across the HTL into the perovskite layer in PSCs at elevated temperatures under working conditions (Fig. 3d)43. The diffused metal atoms caused irreversible loss of VOC, FF, and JSC of the aged devices. Inserting an diffusion barrier layer such as composed of less-movable Cr43 or Bi44, amorphous (a)-TiO245, graphene and derivatives46,47, ITO48 or MoS249 underneath the metal electrode suppressed the detrimental effect of the metal electrode on device stability (Fig. 3e). Replacing those metal electrodes with chemically inert and non-movable back electrodes also works50. Conductive compounds including metal oxides such as ITO could work as stable back electrodes for PSCs51,52,53,54. Carbon materials including graphite, carbon black, carbon nanotubes and graphene are very promising electrode materials for PSCs55,56,57,58.

Light soaking stability

Light soaking stability is the most basic and most important indicator for solar cells. Currently, as the stability of perovskite has been gradually established, many reports used light soaking aging test for judging PSCs’ stability59,60,61,62, and some test protocols met the IEC standard63,64. To pass the light soaking aging test, it is necessary to understand the mechanisms of photo-induced degradation. Light soaking would cause the degradation of halide perovskites with a slow and complicated process, which has not been comprehensively revealed and is generally affected by the combination of light and light-induced heat and electric field in PSCs. The challenge brought by light soaking to PSCs is discussed as follows.

Different from ambient chemicals, light soaking is the basic requirement for PSCs to operate and cannot be sheltered. Therefore, halide perovskites must face the challenge of light soaking when all the other components in PSCs are made stable. Halide salts including PbI2, PbBr2, and PbCl2 are light sensitive and undergo photodissociation at elevated temperatures since photogenerated holes would be captured by halide ions at the surface or boundaries to form neutral iodine while electrons would be captured by lead cations to form neutral lead (Fig. 4b)66,68. Therefore, excess lead salts such as PbI2 in the perovskite films for PSCs would cause light soaking stability issue69,70,71,72.

Fig. 4: The challenge brought by light soaking to PSCs.
figure 4

a Possible pathways for light-induced ion migration, including Schottky defects, Frenkel defects, open space and wrong bonds at GBs, lattice distortions, nonuniform strain, and soften lattice. Adapted with the permission from ref. 65 Copyright © 2016, American Chemical Society b Schematic of the generation of iodide interstitial defects (Ii) in the PSC under illumination. The orange and green lines indicate the energetic and spatial distribution of degradation-generated Ii− and Ii+, respectively. The solid red and blue balls represent photogenerated holes and electrons, respectively. Photogenerated holes can be readily captured by Ii− due to its short trapping time. Adapted with the permission from ref. 66 Copyright © 2021, The Author(s), under exclusive licence to Springer Nature Limited c Schematic diagram of light-soaking upon films with different strains and photographs of the films after 500 h illumination. Adapted from ref. 67.

Full size image

Halide perovskites themselves are not so sensitive as lead halides since the photodissociation is highly correlated with the dynamic behavior of carriers while halide perovskites possess excellent photoluminescence properties and the photogenerated carriers can also be effectively extracted by selective contacts in PSCs. Elevated temperature induced by light soaking affects the carrier dynamics in PSCs. In addition, ions in halide perovskites are migratable and operating conditions affect the migration behavior of those ions. Ions mainly migrate through the point defects, GBs and surface or interfaces (Fig. 4a)65. Expanded or distorted lattice due to thermal expansion also promotes the migration of ions such as I− into the interstitials. The elevated temperature and light irradiation also promote the migration ability of the loosely bound ions such as MA+ and I− by reducing the ion migration activation energy (Ea)65. Ion migration could be driven by residual strain (Fig. 4c)67 and electric field caused by the accumulated charges in the device.

On one hand, migrated ions led to the phase segregation of the mixed halide and mixed cation perovskites. MAPbI(BrxI1-x)3 segregated into two I-rich and Br-rich crystalline domains under light soaking73,74. A photoinactive and current-blocking Cs-rich phase was reported to form in FA0.9Cs0.1PbI3 under illumination75. The illumination-induced phase segregation could be suppressed by proper composition design. On the other hand, the migrated ions could retard the transport of photogenerated carriers and lead to accumulated charges. The accumulated charges could then interact with the migrated ions to form defects such as I0 and Pb0, leading to gradual decay of PSCs during the long-term applications34. In fact, even the migrated ions do not interact with photogenerated carriers, it is still challenging for them to return to their original sites completely. Potential strategies for overcoming the above stability concerns include increasing Ea, removing driving force and blocking migration pathyways for ion migration.

Maintaining the device at a low temperature could increase Ea and prolong the lifetime of PSCs76. Immobilizing ions at GBs by introducing selected additives also contributed to the operational stability enhancement of PSCs77. Fluorine anions with enhanced electronegativity for a strong hydrogen bond with organic cations and a strong ionic bond with lead cations in halide perovskites suppressed ion migration at GBs and improved the device operational stability (Fig. 5a)78. In addition to common halogen ions, some pseudohalogen ions, such as SCN−, BF4− and PF6−, are expected to have a similar effect on suppressing ion migration79,80,81. Besides halogen ions, FA+ and Cs+ are more capable to suppress cation migration. Cs+ could increase the activation barrier for halide migration by inducing lattice compression82 while lager-sized cations such as acetamidinium, azetidinium and guanidinium could slow ion migration by the steric hindering effect83,84,85,86. Potassium was also reported to fill halide vacancies and immobilize halide ions at GBs of halide perovskites87,88. Various salt additives were applied to mitigate ion migration on the perovskite surface and grain boundarie for enhanced stability. The introduction of 5-ammonium valeric acid hydroiodide (5-AVAI) helped printable hole-conductor-free mesoscopic PSCs (p-MPSCs) with carbon electrodes pass the key IEC61215:2016 test and the fabricated device operated stably for >9000 h under light soaking test without obvious decay63. The ionic liquid additives including 1-butyl-3-methylimidazolium (BMIMBF4) and 1-butyl-1-methylpiperidinium tetrafluoroborate ([BMP]+[BF4]−) suppressed ion migration and formation of I2 at the GBs (Fig. 5b)89,90,91. The introduction of [BMP]+[BF4]− dramatically improved the operational stability of the inverted p-i-n PSCs under full sunlight at 85 °C and the T95 lifetime (time until the PCE drops to 95% of its initial value) for the device was estimated to be 1200 hours. Meanwhile, preparing perovskite films with reduced GBs could reduce migration pathways and deterioration space, contributing to solving related stability issues.

Fig. 5: Methods for inhibiting the ion migration in PSCs.
figure 5

a Incorporation of NaF into the perovskite units to form strong bonds with both cations and anions to immobilize ions at GBs. Adapted with the permission from ref. 78 Copyright © 2019, The Author(s), under exclusive licence to Springer Nature Limited b Photoluminescence images of control film and BMIMBF4-containing film under a constant applied bias (10 V). The bright areas represent photoluminescence emission from the perovskite films. The ionic liquid additive BMIMBF4 exhibits effectiveness to suppress the ion migration in halide perovskite. Adapted with the permission from ref. 90 Copyright © 2019, The Author(s), under exclusive licence to Springer Nature Limited.

Full size image

Trapping and insufficient charge extraction led to accumulation of photogenerated carriers. Preparing perovskite films with reduced deep trap states and passivating those trap states via proper additives reduced charge trapping. Slightly excess FAI was reported to compensate iodine vacancies and suppress the generation of defects during long-term operation of PSCs92. Charge extraction is determined by the interface contact between perovskite and CTLs, the energy level alignment in the device and the charge extraction ability of CTLs. Substrate-related interface voids could be formed due to the slow volatilization of residual dimethyl sulfoxide (DMSO) solvents at the interface during the preparation of perovskite films. Meanwhile, the interface between the perovskite film and the commonly used substrate is brittle. Under working conditions, stress induced by high and low temperature caused reconstruction, delamination, and formation of voids at the interface due to mismatched TEC and weak interface bonding as discussed in the last section. Voids and delamination prevented the injection of carriers and led to the accumulation of charges and formation of defects at the interface. The deteriorated interface further slowed down charge extraction and accelerated the deterioration. Partially replacing the liquid DMSO solvent with the solid-state additive of carbohydrazide reduced interfacial voids and stabilized the interface (Fig. 6a)93. The toughness of the interface can be enhanced by interface modification. 5-AVAI strengthened the TiO2/perovskite interface with its bifunctional groups of ammonium and carboxyl. Iodine-terminated self-assembled monolayer (I-SAM) between SnO2 electron transport layer (ETL) and the perovskite layer achieved a 50% increase of adhesion toughness and enhanced the mechanical reliability, which suppressed the formation of voids and delamination at the interface under working conditions (Fig. 6b)94. A coherent interface layer of FASnClx constructed by residual chloride on the surface of SnO2 colloids was also proposed and realized for the interface modification of PSCs with enhanced device stability5. The enhanced contact can also block ion migration pathways at the interface. In addition to the mechanical contact, the energy level and charge extraction ability of the selective contact materials should be carefully designed for sufficient charge extraction. Optimizing the structure of the substrate such as making it porous or arrayed to make the interface bulk contact could also improve the interface-related operational stability. More studies on developing stable and cost-effective charge transport materials with suitable lattice or appropriate specific surface energy and stable charge extraction ability to match the perovskite for tougher interface with excellent mechanical and electrical contact is also desired for fabricating stable PSCs.

Fig. 6: Stability issues at the interfaces.
figure 6

a Interfacial voids could be formed due to the slow release of residual DMSO. Adapted with the permission from ref. 93 Copyright © 2021, The American Association for the Advancement of Science b I-SAM increased the toughness of perovskite/SnO2 interface and suppressed the deterioration of the interface under long-term operating. Adapted with the permission from ref. 94 Copyright © 2021, The American Association for the Advancement of Science.

Full size image

When I0 and Pb0 are formed in PSCs, converting them back to Pb2+ and I− timely can also strengthen the device stability. A demonstrated method for this is introducing a redox shuttle into the film. A reported shuttle is the Eu3+-Eu2+ pair (Fig. 7a). It was found that although the reaction between I0 and Pb0 could happen thermodynamically, a kinetic energy barrier existed in the reaction which made the reaction incomplete. The introduced redox shuttle can effectively transfer electrons in a cyclical manner with the reduction of I0 by Eu2+ and the oxidization of Pb0 by Eu3+. As a result, the Pb0 and I0 were eliminated and the redox shuttle returned to its original state95. When aging perovskite films under 1 sun illumination for more than 1000 h, the Pb0/(Pb0+Pb2+) ratio in the film with the redox shuttle was about 2.5%, while that in the control film was 7.4%. The introduction of Eu3+-Eu2+ made PSCs retain 92% and 89% of the peak PCE under 1-sun continuous illumination or heating at 85 °C for 1500 h. Gd3+-Gd2+ could also work in a similar manner. In addition, the cheap metal element of iron was designed to realize this96. Fe3+ is oxidative while Fe2+ is reductive. The challenge is Fe3+ can not only oxidize Pb0, but also oxidize I− which leads to the formation of I0. Ferrocene (Fc) avoids such concern. Fc has been reported to react with I0 to form FcI, while FcI could react with Pb0 to form PbI2 and convert back to Fc for improved stability of PSCs (Fig. 7b)96.

Fig. 7: The redux shuttles for degenerate the defects in halide perovskite films to prolong the operational lifetime of PSCs.
figure 7

a Mechanism diagram of cyclically elimination of Pb0 and I0 defects and regeneration of Eu3+-Eu2+ metal ion pair. Adapted with the permission from ref. 95 Copyright © 2019, The American Association for the Advancement of Science b Sustainable recovery mechanism on comprehensive defects in MAPbI3 perovskite via a two-cyclical 4-step chain-reaction strategy. The red cycle represents redox potentials of Fc/Fc+ pair transferring charges from I0 to Pb0 defects, and blue cycle means intermediated FcPbI3 perovskite fixing MA defect back to MAPbI3. Adapted with the permission from ref. 96 Copyright © 2021 Wiley-VCH GmbH.

Full size image

Light soaking also interacts with functional layers such as ETL in PSCs and led to instability. UV could lead to the desorption of adsorbed oxygen molecules which passivated oxygen vacancies (OVs) on the surface of metal oxide ETLs (mainly TiO2) while the exposed OVs caused non-radiative recombination at the interface and performance decay of PSCs. UV shelters or down-conversion material could be used to block or convert the UV light to visible light for improved stability97,98. Surface modification of metal oxides or introducing an interlayer also suppressed the detrimental effect of those OVs61,99,100. Potential electron transport materials that have reduced intrinsic UV activated deep traps such as SnO2 or van der Waals tin oxides of Sn2O3 and Sn3O4 are available choices for stable PSCs101,102,103,104,105,106,107. SnO2 is the presently popular electron transport material and various surface modification strategies on SnO2 were developed for efficient and stable PSCs108,109.

Typical stability demonstration of PSCs

In Fig. 8, we exhibited typical stability results that were obtained by settling those above-mentioned degradation mechanisms. Figure 8a showed a p-i-n inverted planar PSC whose perovskite was incorporated with an ionic solid additive [BMP]+[BF4]−89. [BMP]+[BF4]− suppressed deep trap states in the perovskite and enhanced its stability under light soaking at 85 °C. The ability of suppressing deep trap states in perovskite is a key issue for the enhanced stability at 85 °C. The thermally more stable poly(4-butylphenyl-diphenylamine) (polyTPD) rather than Spiro-OMeTAD or NiO was selected to avoid HTL-related stability issues. Additionally, the Au electrode was modified with Cr to avoid the instability caused by the metal electrode. As a result, T95 lifetime of 1200 hours under full sunlight at 85 °C was achieved. The triple-mesoscopic p-MPSC is also a typical device design for inherent structural stability (Fig. 8b)110. In p-MPSCs, no additional HTL was used and holes transported in perovskite itself. The hydrophobic, chemically inert, and non-diffusible carbon electrode was adopted to replace the metal electrode. Meanwhile, the MAPbI3 crystal was localized and buried in the thick mesoscopic scaffolds of several micrometers and the interfaces or boundaries were strengthened by 5-AVAI. The decomposition or reconstruction was inhibited, and the irreversible ionic migration was suppressed. The prepared p-MPSCs successfully passed the main items of IEC61215:2016 qualification tests with over 9000-h operational tracking63. The n-i-p PSCs also demonstrated very excellent stability by retaining 99% of its peak PCE after 1450 h of continuous operation at 65 °C in N2 (Fig. 8c)111. In this work, the ETL was constructed by multilayer stacking of SnO2, [6,6]-phenyl-C61-butyric acid methyl ester (PCBM) and MnSO4. High throughput screening platform was designed to find the stable perovskite compositions such as MA0.1Cs0.05FA0.85Pb(I0.95Br0.05)3. A bilayer HTL composed of the stacking of dopant-free polymer of PDCBT (poly[2,2ʹʹʹʹ-bis[[(2-butyloctyl)oxy] carbonyl] [2,2ʹ:5ʹ,2ʹʹ:5ʹʹ,2ʹʹʹ-quaterthiophene]-5,5ʹʹʹ-diyl])) and acid-doped polymer of the conjugated poly(triarylamine) polymer (PTAA) doped by the stable Lewis acid tris(pentafluorophenyl)borane (BCF) was designed to form stable ohmic contact at elevated temperatures and buffer iodine vapors. In addition, a capping layer of MgF2 was prepared on top of Au electrode to form the close-contact encapsulation.

Fig. 8: Typical stability results for dissolving multiple degradation pathways.
figure 8

a A p-i-n inverted planar PSC utilizing the [BMP]+[BF4]− additive to suppress deep trap states, the thermally stable HTL polyTPD, and Cr-modified Au electrode to achieve excellent stability under full sunlight at 85 °C. Adapted with the permission from ref. 89 Copyright © 2020, The American Association for the Advancement of Science. b The p-MPSCs with the triple-mesoscopic structure, no HTL and carbon electrode as an inherently stable structural design. Stability of over 9000-h operational tracking was achieved due to localized effect of mesopores and interface strengthening effect of 5-AVAI. Adapted with the permission from ref. 110 Copyright © 2014, The American Association for the Advancement of Science. Adapted with the permission from ref. 63 Copyright © 2020 Elsevier Inc. c A modified n-i-p PSC structure with multiple layers of ETL, inherently stable perovskite composition, stable bilayered HTL, and MgF2 capping layer atop Au enables long-term operational stability at 65 °C under metal-halide lamps in N2. Adapted with the permission from ref. 111 Copyright © 2021, The Author(s), under exclusive licence to Springer Nature Limited.

Full size image

Stability of perovskite solar modules (PSMs)

Contrasting with the relatively abundant stability demonstrations of PSCs, there has been only a handful of reports on PSMs. This partially indicates the increasing complexity of the stability of PSMs. Fundamentally, the quality and uniformity of all functional layers in PSMs may be inferior to those of PSCs, which are potential threats to the long-term stability. Additionally, PSMs have their unique degradation mechanisms and pathways against PSCs.

Reverse bias stability

Hot-spot effect due to reverse bias is specifically related to solar modules and has been recognized as the severest damage mode for operational modules112. Hot-spot effect is caused by the reverse bias when the cells connected in series or in parallel fail to work synchronously because of either intrinsic fault or extrinsic shading or soiling. Such process mostly correlates with those modules that are regularly shaded by neighboring trees or buildings. In solar panels, those sub-cells or regions which fail to offer as high voltage as the parallel-connected ones or as high current density as the series-connected ones would serve as the load, which bears the reverse bias and wastes the electricity. The reverse bias has been reported to cause overheating. Decomposition of perovskites could be triggered then by the voltage and the heat113. This overheating may degrade the PSM as well as destroy the encapsulating materials. The reverse bias could also drive the migration of heat-activated mobile ionic species to accumulate at the interface of the perovskite and CTL, mediating a tunneling process of either electrons or holes. With a short duration of the bias, the reverse bias degradation is partially recoverable via redistribution of the migrated ions. When the bias is prolonged or repeated, the degradation gradually turns irreversible, involving the oxidation of I− by injected holes66. It was reported that placing an interfacial ion blocking layer and decreasing the number of mobile ionic defects could alleviate reverse bias degradation. Installing bypass diodes is adopted in well-established photovoltaic technologies like silicon cells to avoid the reverse bias damage. However, the breakdown voltages for PSCs are reported to be considerably lower, implying lower bias tolerance. This means that more bypass diodes would be needed, raising the fabrication cost. It is significant to identify PSCs with better voltage tolerance. Recently, Bogachuk et. al. represented superior resilience against reverse bias of p-MPSCs, which did not degrade until the applied reverse-bias exceeds −9 V114. They fabricated p-MPSMs with 56.8 cm2 aperture area and subjected them to the hot-spot test with the requirement of IEC 61215–2:2016, no efficiency loss was found, indicating that the PSMs passed the hot-spot accelerated test. This result has again shown unique structural advantages of p-MPSCs for stability, underlining the greatest potential for industrialization.

Laser scribe issues

The interconnection of sub-cells in a solar module is typically achieved via laser scribe. Three scribes called P1, P2 and P3 are designed to selectively ablate the front electrode, the perovskite layer and CTLs, and the back electrode, respectively. These scribes mainly result in four degradation pathways in PSMs115. Firstly, laser could induce light and thermal degradation of the perovskite following the pathways discussed in previous sections. Degradation products like PbI2 will trigger further degradtion of PSMs. Secondly, incomplete removal of the CTLs during scribes creates extra shunt resistance, possibly increasing the risk of potential-induced degradation. Thirdly, P2 scribe leads to the direct contact between the perovskite and the back electrode. Irreversible damage would occur due to the reaction between perovskite and metal. Introducing interlayers or using unreactive back electrode materials are effective solutions, as is clarified above. Lastly, unideal P3 scribe leaves grooves for moisture and oxygen ingress. Encapsulation solves this problem while elaborate encapsulation designs for PSMs are required.

Figure 9 summarized some reported stability results of PSCs and PSMs according to the IEC light soaking test, UV precondition test, damp heat test and thermal cycling test. The present progresses are very encouraging. In those accelerated aging test, PSCs have exhibited positive results towards industrialization. One remaining challenge is that most test conditions are diverse and not very strict. Such as for the key light soaking stability, the temperature for the test ranged from room temperature to 85 °C. As discussed in previous sections, the light soaking stability of PSCs dramatically relied on the temperature. It seems light soaking stability under higher temperature is more persuasive. Light soaking stability of PSCs at 85 °C is not widely reported at present. In addition, as discussed in the introduction section, some stability results should be collected for cells aged under various conditions in a sequence. Related studies for this have not been well reported. In addition, more studies on modules towards the IEC standard and outdoor demonstrations are also desired in the next stage of PSCs’ development.

Fig. 9: Several promising stability results of PSCs and PSMs according to the IEC standard.
figure 9

a Light soaking test63,116,117,118,119,120. b UV precondition test64,100,121. c Damp heat test23,63,116,119,122,123. d Thermal cycling test31,63,123,124.

Full size image

Conclusion and outlook

Several pioneering companies are actively pushing the industrialization of PSCs. The IEC standard on terrestrial PV modules, with the latest version of IEC 61215:2021, sets a series of strict stability tests for the evaluation of PV modules’ performance against known degradation and failure modes. As the most widely accepted protocol for PV technologies, it is deemed a minimum threshold for PSCs to the market. In this review, we introduced intrinsic and extrinsic factors that would cause failure to pass the tests, summarized common methods for solving these issues, and exhibited some results of passing the tests. It is important to note that current reports of passing single or several tests with separate PSCs are far from being satisfactory. Meeting the IEC 61215 standards requires passing the full tests. Achieving this demands comprehensive and profound understanding of perovskite components, stability properties of device structures, and encapsulation. More results of PSMs’ stability are also welcome. A merit of PSCs against Si-based and thin film solar cells is the solution-processable fabrication. However, solution-processed films tend to be rough and have defective regions, such as burs. These imperfections may lead to seams between the perovskite and the CTL. This is unfavorable to device performance and leaves channels for the destructive factors. Traditional layer-by-layer PSC structures are rather prone to the roughness especially when each functional layer is quite thin. For p-MPSCs, the perovskite grows inside the as-fabricated scaffold, the contact problem can be ignored. The undulated scaffold consisting of disordered nanoparticles endows p-MPSCs with high tolerance of imperfect perovskite morphology. The triple mesoscopic device structure is advantageous on large-area fabrication via screen-printing technique. The inherent stability of this structure lies in the hydrophobicity and inertness of carbon electrodes, the elimination of susceptible HTLs, and the protective mesoporous scaffold.

The IEC standard was initially set up in 1993 based on research experience on crystalline silicon solar cells and modules. It was modified and supplemented in accordance with deeper understanding of known failure modes and newly discovered modes in both experimental and practical investigations. Testing requirements of amorphous silicon, CdTe, and Cu(In,GA)(S,Se)2 based modules had been included as its subsets. Inheriting current IEC standard for the qualification of PSCs may not be appropriate, considering the largely distinct material properties and device architectures over those PV technologies. The ionic nature of perovskites accounts for the biggest difference. Some unique phenomena including hysteresis, efficiency fluctuation during day/night cycles, bias-induced degradation and others have been attributed to ion migration. We believe a suitable stability qualification protocol for PSCs should place emphasis on ionic behaviors. A feasible scheme is modifying test details and adding new tests for existing protocols (e.g., the International Summit on Organic Photovoltaic Stability (ISOS) protocols125), considering the relatively short development time of PSCs. A proprietary protocol for PSCs is expected to be established with adequate knowledge of degradation mechanisms from research and primary commercialization trials. Additionally, we think it is important to make the protocol openly-accessible, which is not the case for the IEC standard. This will trigger the improvement in PSCs’ stability via the convenience for comparisons and communications of stability results from different research groups.

Finally, test results from any qualification protocol are not construed as a quantitative prediction of module lifetime. The artificial aging conditions are not equivalent to real operational environment. There may be unexpected failure modes observed in operation that are not covered in the protocol126. Hence, we suggest that outdoor stability tracking is highly required beyond accelerated aging tests in lab. Velilla et. al. carried out a high-throughput outdoor performance analysis of MAPbI3 minimodules and identified different performance evolution patterns. The ideal factor (nID) was calculated taking advantage of the different illumination conditions encountered during day–night cycles. A linear relationship was observed between the time until modules reached 80% of its initial PCE (T80) and until nID reached 2 (TnID2). Since nID values reflect recombination processes in the devices, this relationship made it possible to link recombination pathways to the degradation profile. This methodology provided insight to enable improved understanding of the device degradation processes under outdoor conditions that are relevant to the commercial application of PSCs127. Several researchers have conducted outdoor measurements and gathered meaningful data on real-world operation of PSCs and PSMs128,129,130,131. WonderSolar put up demonstration power systems with areas of 110 m2 based on 3600 cm2 p-MPSMs. We believe that this would further gain confidence from investors and customers.