Stability of Perovskite Solar Cells: Degradation Mechanisms and Remedies

Inorganic–organic metal halide perovskite light harvester-based perovskite solar cells (PSCs) have come to the limelight of solar cell research due to their rapid growth in efficiency. At present, stability and reliability are challenging aspects concerning the Si-based or thin film-based commercial devices. Commercialization of perovskite solar cells remains elusive due to the lack of stability of these devices under real operational conditions, especially for longer duration use. A large number of researchers have been engaged in an ardent effort to improve the stability of perovskite solar cells. Understanding the degradation mechanisms has been the primary importance before exploring the remedies for degradation. In this review, a methodical understanding of various degradation mechanisms of perovskites and perovskite solar cells is presented followed by a discussion on different steps taken to overcome the stability issues. Recent insights on degradation mechanisms are discussed. Various approaches of stability enhancement are reviewed with an emphasis on reports that complied with the operational standard for practical application in a commercial solar module. The operational stability standard enacted by the International Electrotechnical Commission is especially discussed with reports that met the requirements or showed excellent results, which is the most important criterion to evaluate a device’s actual prospect to be utilized for practical applications in commercial solar modules. An overall understanding of degradation pathways in perovskites and perovskite solar cells and steps taken to overcome those with references including state-of-the-art devices with promising operational stability can be gained from this review.


INTRODUCTION
Organic-inorganic metal halide perovskite-based perovskite solar cells (PSCs) are at the epitome of attention to the solar cell research community due to their rapid growth in efficiency over a short period of time. It was first reported as a sensitized solar cell in 2009 with an efficiency of 3.81% (Kojima et al., 2009). Today, the efficiency of the laboratory-scale devices has exceeded 25% (Kim et al., 2020a). The outstanding optoelectronic property of the perovskite facilitates the growth in efficiency (Yin et al., 2016a;Wu et al., 2019a;Jena et al., 2019). Rapid growth of efficiency bears a promise to replace commercial Si-based and thin film solar cells. The major obstacle in the way of commercialization is easy degradation of perovskites and PSCs (Boyd et al., 2019;Jena et al., 2019;Shi et al., 2020a;Wang et al., 2020a;Wang et al., 2020b;Yang et al., 2021). Both are vulnerable to external factors of ambient conditions. Various approaches are taken to overcome degradation, but commercial PSCs are still not realized. The rapid growth in efficiency of laboratory-scale devices always goes ahead of progress in stability in PSCs. Understanding the stability factors of perovskites and PSCs and studying the degradation mechanisms are extremely important (Wu et al., 2019a;Dunfield et al., 2020;Kumar et al., 2020). Knowledge of the degradation pathways has utmost importance in developing remedies for the factors responsible for degradation. In this review, almost all possible degradation mechanisms including the structural specialty of perovskite and PSCs are discussed in the first part. That will be helpful for the reader toward a systematic understanding of the degradation pathways in perovskites and PSCs. In the second part, various approaches to enhance the stability are discussed. We attempted to provide an updated and systematic overview on the remedies to overcome the stability issues in PSCs. As PSCs are considered for commercial applications, standard methods must be available to test the stability and reliability of the device. The International Electrotechnical Commission (IEC) sets up criteria that can be treated as a standard to trial the stability and reliability of PSCs. Those are exclusively discussed along with reports and attempts to achieve that Wang et al., 2020a). In all the sections discussing the strategies to improve the stability, the relevant reports with high operational stability (e.g., negligible loss of PV performance after several hundreds of hours' operation under harsh conditions such as 85°C, 85% relative humidity) are reviewed.

STABILITY ISSUES IN LIGHT-HARVESTER PEROVSKITE AND PEROVSKITE SOLAR CELLS
The long-time stability of perovskite and perovskite solar cells (PSCs) is very important for commercialization of PSCs Yang et al., 2021). But, various factors are responsible for the degradation of perovskite and other constituents of PSCs. In this section, the various stability issues are discussed. Studying various decomposition mechanisms is very important to adopt strategies to prevent decomposition. The degradation mechanisms can be understood by reconciling various experiments with theoretical calculations. The stability issues can be divided into: 1) Intrinsic stability-stability issues solely caused by the molecular and crystallographic structure of the perovskite (ideally MAPbI 3 is discussed) and 2) Extrinsic stability-stability issues of mainly perovskite and the other components of PSCs upon external factors that are unavoidable in case of practical applications. One component of the PSCs can also cause degradation of the other component such as the decomposition of perovskite by a metallic electrode. A classification of various stability issues is depicted in Scheme 1.

Structural Stability
Lev Perovski discovered and determined the crystallographic structure of the mineral CaTiO 3 in 1839. That particular structure is called perovskite and it refers to a set of compounds with a certain ABX 3 crystal structure. A refers to a larger monovalent cation, B is a smaller bivalent cation, and X is a monovalent anion that bonds with both A and B. The stability of a certain perovskite structure can be predicted by the Goldschimdt tolerance factor t that can be determined by the following simple equation (Li et al., 2016a): where R A , R B , and R X are the ionic radii of the A, B, and X ions, respectively. Ideally, t > 1 results in hexagonal or tetragonal structures. 1 > t > 0.9 produces a cubic structure, 0.9 > t > 0.71 gives an orthorhombic or rhombohedral structure, and t < 0.71 does not produce a perovskite crystal structure ( Figures 1A,B) (Yin et al., 2014;Ma et al., 2020a). The Goldschimdt tolerance factor t is a zeroth-order approximation that facilitates ease of calculation in most of the cases, but anomaly has been found in approximately 26% of cases, as mentioned in some literatures (Bartel et al., 2019). Bartel et al proposed another tolerance factor that can be presented by the following equation (Bartel et al., 2019): SCHEME 1 | Various stability issues in perovskites and perovskite solar cells.
where R A , R B , and R X are the ionic radii of the respective species and n A is the oxidation state of the A cation. The possibility of a perovskite structure formation arises when t < 4.18 and the lesser value increases the probability of the formation of perovskite. Equation 2 cannot predict the exact crystallographic structures such as Eq. 1, but it can predict the probability of perovskite formation more accurately than that of Eq. 1 and an anomaly has been observed in mere 8% of cases (Poglitsch and Weber, 1987). So, it is highly prescribed to use both the equations on a certain species to accurately predict the probability of formation of a perovskite structure and the exact crystallographic structure (Dunfield et al., 2020). In case of a monovalent A and divalent B moiety, they donate one and two electrons respectively to balance the tri-negative charges of the X anion. Thus, a bandgap is formed between the unoccupied B cation's p orbital and the occupied X anion's p orbital (Yin et al., 2014). The bandgap often lies in the visible range in case of metal inorganic-organic halide perovskites, thus making them suitable for solar photon harvesting. Those perovskites are the subject of this review.
Photon-harvesting properties of these species were first reported by Kojima et al. (2009). Various formulations are possible including the alloyed ones. In general, the A-site is represented by organic species methylammonium (MA + ), formamidinium (FA + ), or metallic Cs + . The B-site is generally occupied by Pb 2+ or Sn 2+ . The X-site is composed by halide anions such as chloride (Cl − ), bromide (Br − ), or iodide (I − ) (Stoumpos et al., 2013a;Sutton et al., 2018). These different components must be tuned wisely to achieve suitable tolerance factors (t and τ) while maximizing the solar photonharvesting ability of the resultant species (Dunfield et al., 2020).
The size of the respective moieties is very important. For example, if the ionic radius of A is too small, it cannot separate two B moieties effectively which leads to an edge-sharing octahedron while a much bigger A would lead to the formation of lower dimensional face-sharing octahedral structures, which may not be conducive to effective solar photon harvesting due to reasons such as wider band gap. A suitable A-site species should be able to perfectly separate two B-site cations to achieve a symmetrical 3D cubic lattice (Quarti et al., 2016). However, the phases are changed with temperature and the properties of the perovskite are also changed as per. This subject is discussed in the following section.

Thermodynamic Phase Stability
Thermodynamic phase stability as determined by the tolerance factor (t) for a particular species and also of the alloyed species are discussed in this section. As mentioned in 2.1.1, 1 > t > 0.9 leads to an ideal cubic perovskite structure (Yin et al., 2014). But, when departed from that value, the BX 6 octahedron starts to distort (Filip et al., 2014). For the ideal MAPbI 3 molecule, t 1.02 and it forms a tetragonal perovskite structure at room temperature. The ionic radius of FA + is 2.79 Ǻ , which is larger than that of MA + (2.7 Ǻ ) (Amat et al., 2014). The marginally oversized FA + leads to t 1.04 for FAPbI 3 and it crystallizes into a non-perovskite hexagonal phase at room temperature and it must be converted into a cubic perovskite structure by annealing at more than 150°C temperature Lee et al., 2014a;Lee et al., 2015). On the other hand, Cs + has an ionic radius of 1.81 Ǻ that is much smaller than that of FA + and MA + . Thus, t equals 0.81 in case of CsPbI 3 . So, it crystallizes to an orthorhombic phase at room temperature and annealing at 250°C is needed for the formation of a cubic phase (Amat et al., 2014;Chen et al., 2017). Despite having other beneficial properties, pure FAPbI 3 and CsPbI 3 often lag behind due to the thermodynamic phase stability issues, as annealing at a higher temperature leads to defects in perovskite films (Boyd et al., 2019;Ma et al., 2020a). The t values presented in Table 1 (Kieslich et al., 2014;Nagabhushana et al., 2016). However, many reports have calculated t as per the crystal ionic radii of the species. In most of the efficient devices, alloys are used. Understanding the thermodynamics of these alloys is very crucial and an alloyed species can be treated as a linear combination of two pristine perovskites (Saliba et al., 2016a). Thus, the suitability of an alloyed perovskite can be determined by: 1) To select proper B-site and X-site moieties' ratios to obtain a suitable optical band gap and 2) to select a proper ratio of the A-site moieties to achieve a suitable tolerance factor (Saliba et al., 2016a;Saliba et al., 2016b;Ma et al., 2020a). The alloying at the A-site is governed by the Gibbs free energy of mixing (ΔG mix ) at constant pressure and volume (Schelhas et al., 2019): where ΔG mix is the enthalpy of mixing and TΔS mix is the entropy of mixing. ΔG mix < 0 is energetically favored to mix while ΔG mix > 0 may lead to phase segregation. These thermodynamic factors must be kept in mind while designing an alloy (Schelhas et al., 2019). Phase stability of various perovskites is listed in Table 1 with the respective temperatures.

Electronic Band Structure and Stability Toward Defects
Understanding the electronic band structure of the perovskite is crucial to determine the degree of defect tolerances. Bulk MAPbI 3 is known as defect-tolerant, even the films are tolerant to defects/ vacancies compared with other light-harvesting semiconductor counterparts with respect to photovoltaic efficiency Dhar et al., 2018). Density functional calculations are used to elucidate the effect of defect on MAPbI 3 and are represented in . Different growth conditions of the perovskite significantly influence the defect formation energies and thus the concentration of defects. In the I-rich/Pb-poor case, the scenario is dominated by V Pb which leads the Fermi level to stay near the valence band. When the ratio of Pb is increased, the required energy to form a defect also increases and these cause V Pb and MA i to gain almost equal formation energies. So, the Fermi level resides near the midgap. A further increment of Pb with respect to I would cause the Pb-rich/I-poor condition and MA i and MA I would determine the defect structure. The semiconductor would become n-type. Detrimental defects such as I MA , I Pb , Pb I , and Pb i have very high formation energies. So, their formation is very rare. Hence, the formation of deep trap states is also rare. Defects with lower formation energies such as V MA , V Pb , MA i , I i , Pb MA , and MA Pb may occur and form trap states ∼0.05 eV in the bandgap. So, the formation of center of nonradiative recombination is less probable. This theoretical prediction of vacancy tolerance has also been experimentally verified (Steirer et al., 2016;Dhar et al., 2018). The degree of defect tolerance in perovskites such as MAPbI 3 is quite attractive compared with its other photovoltaic counterparts such as Si. A very low defect in Si can destroy its photovoltaic prospects. This should be credited to the exceptional band structure of MAPbI 3 (Yin et al., 2014). This kind of study is extended to other related perovskites. A strong correlation between defect tolerance and structural stability (t) has been observed ( Figure 1E). Perovskites such as MAPbBr 3 , CsPbBr 3 , CsSnI 3 , CsSnBr 3 , and CsSnCl 3 have t 0.82 to 0.85 where t of FAPbI 3 , FASnI 3 , MASnI 3 , and CsPbI 3 do not fall in this range (Xu et al., 2014;Steirer et al., 2016;Kang and Wang, 2017). So, favorable t can be achieved by tuning the alloying at the A-site and it also can be an effective way to mitigate defects. It can be assumed that a perovskite with a high structural stability is also a defect-tolerant species. The defect-tolerances mentioned above are referred to bulk. The properties of the surfaces of these materials can be very different from the bulk. The chemical potentials can be highly varied due to the perturbation of the electronic structure on the surface than that of the bulk (Schulz et al., 2019). These defects are very crucial in extrinsic stability of perovskites because the surface is exposed first to the environment or to another interface. These are discussed in the following sections.

Extrinsic Stability
In the previous section, intrinsic stability factors of perovskites, particularly of MAPbI 3 have been discussed. It is found that the perovskites have satisfactory intrinsic stability compared to their counterparts. However, the perovskites used in devices are subjected to external factors such as heat, moisture, light, and interface materials (Schulz et al., 2019). Perovskites and PSCs demonstrate instability to these factors. So, understanding degradation of perovskites and PSCs by these extrinsic factors is very crucial to take proper steps to mitigate these problems. In this section, important external factors affecting perovskites and PSCs are discussed.

Interaction With Water Molecule and Pathways of Degradation
Water alone as well as in the presence of oxygen is one of the most pervasive factors affecting perovskites. Water and any other polar solvent can affect perovskite due to their ability to form solvated phases. Polar solvents can be avoided by tuning the deposition strategies but, exposure to environmental moisture is unavoidable. Water molecule can hydrate the perovskite and form a monohydrate phase (Leguy et al., 2015;Askar et al., 2017).
Both the reactions are reversible. So, if the hydrated products would be kept in an inert or moisture-free environment, it can be dehydrated to produce back perovskite with some minuscule irreversibility due to phase segregation. The hydration of the perovskite causes structural deformation of (PbI 6 ) 4− octahedra which converts the 3D network of the (PbI 6 ) 4− octahedra into a 1D chain of octahedra for the monohydrate and 0D for dihydrates. These weaken the chemical bonds between the A-site moiety and Frontiers in Electronics | www.frontiersin.org August 2021 | Volume 2 | Article 712785 the (PbI 6 ) 4− , thus making the perovskite more vulnerable to other extrinsic factors such as heat, electric bias. Once the perovskite will be saturated by the moisture, the hydration will become irreversible and it irreversibly will decompose to produce CH 3 NH 3 I and PbI 2 (Christians et al., 2015). The effect of moisture is depicted in Leguy et al., 2015;Wang et al., 2017a).

Thermochemical Stability of Perovskite
MAPbI 3 can be degraded at 85°C temperature while being kept in an inert atmosphere, i.e. not under oxygen or moisture. A direct proof of this has been observed by comparing the nano-electrical properties of the pristine perovskite and after heating that ( Figure 2F) (Conings et al., 2015). The proposed reversible and irreversible pathways of degradation of perovskite by heating are mentioned later (Conings et al., 2015;Juarez-Perez et al., 2016;Kim et al., 2017). CH 3 NH 3 PbI 3 (s)#CH 3 NH 2 g + HI g + PbI 2 (s) CH 3 NH 3 PbI 3 (s) → CH 3 I g + NH 3 g + PbI 2 (s) Device Thermal Stability Reversible and irreversible decomposition of light harvester perovskites are discussed in the previous section. Other components are also needed in PSCs. Those also can be degraded upon heating at a marginally higher temperature. Spiro-OMeTAD is a widely used hole-transporting material in n-i-p PSCs. It can be crystallized at 100°C, thus degrading the solar cell efficiency. Additives used with hole-transporting layers (HTLs) can be evaporated at an even lower temperature of 85°C (Bailie et al., 2014;Jena et al., 2017). Oxides are attempted to replace organic charge-transporting layers, but electron-transporting oxides such as ZnO also can degrade the perovskite upon heating at 100°C (Liu et al., 2018a).

Light Stability
Stability of the light harvester material and the device are both crucial in determining the overall stability. Factors such as oxygen, moisture, and heating can be reduced or nullified by adopting certain strategies, but illumination by solar light is obvious. Many reports confirm that MAPbI 3 does not degrade upon illumination even for several hundreds of hours, as confirmed by absorption spectra and other characterizations (Mei et al., 2014;Seo et al., 2018). The photoinduced changes in the perovskite even can be beneficial for PV properties, e.g., the ionic conductivity is increased by several orders of magnitude upon illumination of visible light . Lightinduced lattice expansion of the perovskite improves PEC of the device (Tsai et al., 2018). Stability of different constituents of the PSC under solar illumination is discussed in this section.

Photostability of a Charge-Transporting Layer
Organic charge-transporting layers such as fullerene-based materials are prone to degradation upon illumination (Abate et al., 2013;Mateker and McGehee, 2017). Though TiO 2 -based PSCs give many highest PV efficiencies, TiO 2 layers are highly prone to degrade by UV light, even in inert condition. Rapid trapinduced recombination takes place on the titania surface due to  (Leguy et al., 2015). Copyright 2015, American Chemical Society. (E) Schematic of moisture-induced degradation mechanism of the perovskite film which starts from grain boundaries (Wang et al., 2017a). Reproduced with permission. Copyright 2017, Royal Society of Chemistry. (F) Conductive atomic force microscopy (c-AFM) images of pristine MAPbI 3 film (top) and MAPbI 3 film after heating at 85°C for 24 h under N 2 (bottom) (Conings et al., 2015). Reproduced with permission. Copyright 2015, Wiley-VCH.
Frontiers in Electronics | www.frontiersin.org August 2021 | Volume 2 | Article 712785 6 photoinduced oxygen desorption (Dulub et al., 2007;Mazumdar et al., 2015a;Li et al., 2016b). This degradation induces several recombination pathways and results in rapid degradation of device performance. Many oxides such as SnO 2 , ZnTiO 3 , and BaSnO 3 are UV-inert and do not obviously degrade upon long illumination. Hole transport oxides such as NiO x generally have wide bandgaps; hence, they are photostable. So, those may be suitable as stable charge-transporting layer. Many small organic molecule-based charge-transporting layers face degradation upon illumination due to UV (Wei et al., 2021).

Light-Induced Ion Redistribution in Perovskites
Light-induced redistribution of halide as well as metal ions is an important phenomenon observed in MAPbI 3 and other perovskites. The photoluminescence of MAPbI 3 film increases rapidly upon illumination over a short time (DeQuilettes et al., 2016). This is proven to be caused by the migration of I − species away from the light or the illuminated area of the film. The exact mechanism has not yet been reported, but evidence indicates that it is related to the slow diffusion of the ionic species along the perovskite layer, so it is also connected to the point defects such as halide vacancies. The increment in photoluminescence is more pronounced in the presence of oxygen and moisture due to an additional passivation effect of superoxide molecule (Brenes et al., 2017). Utilizing this phenomenon, 89% of internal photoluminescence quantum efficiency could be achieved with 32 s carrier lifetime and 19.2% stable PCE in a SnO 2 -based n-i-p device. Hoke et al reported that APb(Br y I 1-y ) 3 perovskites undergo phase segregation to Br − and I − -rich perovskite phases upon illumination (Hoke et al., 2015). The phase segregation is reversible and the perovskite gains the initial structure and composition in dark. Increased photoluminescence of all these phase-segregated materials comes from lower energy states as carrier-trapping domains are created due to the presence of iodine-rich phases with smaller bandgaps. This indicates that the Br-rich and I-rich phases are effective in carrier trapping. This is labeled as "Hoke effect" and this phenomenon is only observed when the Br content is greater than 20% of the X-site. The Hoke effect is not observed when the Br content is less than 20%. It causes a problem for PSCs in a tandem device where the desirable bandgap is generally 1.7-1.8 eV, so the perovskite contains >20% Br and the PSCs suffer structural instability. Light-induced cation segregation is also observed in alloyed perovskites . In a triple alloyed perovskite at the A-site (FA 0.79 MA 0.16 Cs 0.05 ) 0.97 Pb(I 0.84 Br 0.16 ) 2.97 is studied by time-of-flight secondary ion mass spectrometry (TOF-SIMS) (Christians et al., 2018). Upon illumination, it is observed that the halide and lead distribution remain unchanged through the depth of the device, but all the A-site cations are redistributed. The redistribution of Cs + is more than that of MA + and FA + . This phenomenon is only observed with TiO 2 electron-transporting layers (ETLs), but not observed with other ETLs such as SnO 2 .

Photochemical Degradation
As a semiconductor, MAPbI 3 forms an excited state upon illumination. In the excited state, the ionic mobility increases due to iodide vacancies which is generated by oxidation of iodide atoms by photogenerated holes. Transformation to iodine from iodide causes reduction of ionic/atomic size and causes it to leave the lattice and form an interstitial vacancy Tang et al., 2016;Cappel et al., 2017;Kim et al., 2018).
This process leads to irreversible decomposition products such as lead, iodine, and MAI. The iodine can be removed by sublimation, and thus increases the rate of this reaction. This process is depicted in Figures 3A,B. The presence of metallic lead has been detected by XPS even after illumination of short time (Tang et al., 2016;Cappel et al., 2017). Evidence of photoinduced degradation due to operating voltage has also been reported. High-performance PSCs with n-i-p structure are generally stable at the maximum power point. However, some devices demonstrate slow decomposition at both open circuit voltage and maximum power point (Wang et al., 2017b).

Oxidation and Photooxidation of Charge-Transporting Layers and Perovskites
Several oxides TiO 2 , SnO 2 , and NiO x are used in PSCs as chargetransporting layers. TiO 2 is prone to oxidation in the presence of UV light and oxygen. It can catalyze oxidative decomposition of other materials in contact with them. These can limit the stability of the device. However, replacing TiO 2 by SnO 2 can prevent this (Christians et al., 2018). Organic molecules used as chargetransporting layers such as fullerene-based molecules are prone to photooxidation. The degradation of a chargetransporting layer is detrimental to the carrier mobility, thus affecting the long-term efficiency of PSCs. HTLs such as Spiro-OMeTAD require some degree of oxidation to generate free holes in the HOMO, which enhances the p-type conductivity and facilitates the hole transportation to the metal electrode Christians et al., 2015;.
Metal halide perovskites are highly prone to photooxidation. Oxygen starts diffusing perovskite crystals almost instantaneously upon illumination and the decomposition is completed within an hour (Aristidou et al., 2017). Oxygen absorbs and diffuses through iodide vacancies in this process. Iodide vacancies preexist and increase rapidly upon illumination and thus facilitate the diffusion of oxygen (Abdelmageed et al., 2016;Wang et al., 2017b;Dhar et al., 2017).

Interaction With Electrodes and Degradation at Interfaces
Reaction of various species of the perovskite with the metal electrode is very crucial because this is the one way of decomposition which cannot be prevented by encapsulation. Certain metals are able to directly make redox couple with the perovskite itself or react with PbI 2 (Zhao et al., 2016). Au does not react with the perovskite but it reacts with reactive polyiodide melts formed by perovskite decomposition (Abate et al., 2013). Most of the metals react with the perovskite decomposition products. Three of the reaction pathways of the metal electrode are mentioned below and depicted in Figure 3C. These are found by the TOF-SIMS studies. The corresponding depth profiles and the simulated elemental maps are presented in Figures 3D,E respectively (Kato et al., 2015;Domanski et al., 2016;Zhao et al., 2016).
1) X-site species are formed by perovskite degradation; those diffuse to metal electrode. This reaction leads to corrosion of the metal electrode and also enhances decomposition of perovskite. 2) Metal electrode forms a redox couple with Pb 2+ . The perovskite decomposes quickly forming metallic Pb. 3) Metal from electrode may diffuse to the perovskite and react to form metal halides, thus decomposing the perovskite and degrading the metal electrode.

STRATEGIES TO RESIST DEGRADATION AND TO IMPROVE STABILITY
In the previous section various degradation mechanisms of perovskite and PSCs are discussed. Various strategies have been adopted to mitigate the cause of degradation and to improve the stability while keeping up the efficiency. Studying the increment of efficiency needs much patience than that of research solely focused on increasing the efficiency because the stability must be studied over long time. Some strategies are targeting one particular degradation path while multipronged approaches are also reported where one particular strategy can help in resisting multiple ways of degradations. The stabilityenhancing steps must be decided after deliberation so that the strategy to resist one type of degradation should not introduce new ways of other degradations. The authors propose different ways of classification of these strategies (Wu et al., 2019a;Boyd et al., 2019;Wang et al., 2020a;Kumar et al., 2020;Yang et al., 2021). Our approach is summarized in Scheme 2.

Stability Induced by Charge-Transporting Layer
Charge-transporting materials play a key role in determining the efficiency and stability of PSCs. The electronic band structure of the charge-transporting layer is very important in this regard. The energy levels (valence band maximum-conduction band minimum in case of semiconductor species such as metal oxides, HOMO-LUMO in case of molecular species) should be well matched to facilitate the transfer of photogenerated charge followed by efficient transportation of charges to the respective electrodes. Well-matched energy levels can facilitate charge separation, and thus can reduce the probability of recombination. These can help in improving the short-circuit current and the fill factor. Mobility of charge (electron or hole) is also an important factor in determining the competence of a charge-transporting material. High mobility of the chargetransporting material helps in faster charge transport. If two charge-transporting materials with different mobility are at the same carrier concentration, the material with higher mobility will have higher conductivity. Efficient transportation leads to lesser energy/carrier loss and that leads to higher open circuit voltage.
The charge-transporting material should be stable at the interface of the perovskite. Compact charge-transporting layers further prevent external moisture or oxygen to reach the perovskite. Diffusion engineering of the charge-transporting layer can highly boost the stability as the outward diffusion of iodine from the perovskite layer and inward diffusion of external elements can be controlled simultaneously (Bera et al., 2015;Hou et al., 2015;Wang et al., 2020c).

Electron-Transporting Layers
Various metal oxides (binary and ternary) have been utilized as the electron-transporting layer of PSCs. The conduction and valence band of MAPbI 3 lie in −3.9 and −5.4 eV, respectively. The ETL material's conduction band edge must be equal to or lower than that of −3.9 eV to be able to receive photogenerated electron from the perovskite, while the valence band edge should lie well below −5.4 eV to block any photogenerated holes from the perovskite ( Figure 4) Mazumdar et al., 2019;Wang et al., 2020c). The development of PSCs was started as a sensitized solar cell and the mesoporous TiO 2 layer was used as the ETL (Kojima et al., 2009) which was the best candidate for a sensitized solar cell (Mazumdar and Bhattacharyya, 2012;Mazumdar and Bhattacharyya, 2013;Mazumdar et al., 2015b). Several nanostructures of TiO 2 have been employed as ETL of PSCs (Qiu et al., 2013;Yella et al., 2014;Zheng et al., 2016;Wang et al., 2020c). Some of them are proven to be advantageous like branches of a nanodendrite structure can shorten the electron diffusion length and thus enhance the light trapping. Snaith et al replace PSCs with a mesoporous TiO 2 layer by a much thinner layer, effectively a planar device . The efficiency boosts to 8.6% from 2.6% due to reduced recombination pathways between ETL and HTL. However, the biggest problem with TiO 2 is that it can be degraded by the UV part of the solar radiation which is discussed in Photostability of Charge Transporting Layer.
ZnO is proposed as an excellent candidate as ETL due to its high electron mobility compared to TiO 2 (Zuo et al., 2015;You et al., 2016). However, ZnO ETL faces several issues such as improper wetting by the perovskite, degradation of the perovskite (Son et al., 2014). The degradation of perovskite occurs at the ZnO interface due to the presence of radicals such as the hydroxyl (-OH) group (Dkhissi et al., 2016). Two approaches have been taken to solve this problem: 1) To introduce an additional layer with a proper energy alignment between ZnO and the perovskite. A ZnO-CdS core-shell structure is used as the ETL . The CdS lies between ZnO and the perovskite. Proper band alignment of CdS allows photogenerated electron flow from the perovskite to ZnO, while it debars the perovskite to come in direct contact with ZnO which may cause degradation of perovskite. This device is proven to be stable up to at least 60 days. 2) Physical deposition of ZnO by high vacuum methods such as sputtering and pulse laser deposition (PLD) is proven to be highly effective to achieve efficient and stable devices Mazumdar et al., 2020). The surface of physically deposited ZnO films is devoid of chemical moieties such as hydroxyl group which causes degradation of perovskite. So those can assure high efficiency and good stability.
SnO 2 is proposed to be a suitable alternative to TiO 2 Christians et al., 2018;Wang et al., 2019a;Wang et al., 2019b;Wang et al., 2019c;Wang et al., 2020c). Initial efficiencies were not satisfactory due to high rate of recombination at the interface. The problem was attempted to be solved by surface treatment (Huang et al., 2018). Poor crystallinity of SnO 2 is another obstacle to achieve good efficiencies. Tuning the synthesis of SnO 2 nanostructures leads to crystalline SnO 2 which produces efficient and stable PSCs Wang et al., 2019a). WO 3 is another candidate with good potential and various efforts are being engaged to acquire high efficiency (Huang et al., 2010;Mahmood et al., 2015;Wang et al., 2019a).
Ternary metal oxides such as SrTiO 3 , BaSnO 3 , and Zn 2 SnO 4 are also investigated as ETLs. SrTiO 3 can be a suitable candidate with high electron mobility (5-8 cm 2 V −1 s −1 ) (Shin et al., 2017;Neophytou et al., 2019). Its conduction band edge is slightly higher than that of TiO 2 , so it can produce a higher open circuit voltage. The charge recombination on the surface can be suppressed due to its higher dielectric constant . For the same reason, quaternary IGZO has been tried as  . Reproduced with permission. Copyright 2018, Wiley-VCH. (B-F) Bilayer of ETL (TiO 2 and SnO 2 ) optimized as per free energy difference (ΔG) between the ETL and the perovskite and the electron mobility (μ e ) of the ETL. (B) Schematic illustration of electron transportation. Electron extraction occurs from the perovskite to the FTO via injection, transportation, and collection. (C) Illustration of ETLs with different free energies (ΔG) and electron mobilities (μ e ). Case 1: SnO 2 (larger ΔG with higher μ e ) and TiO 2 (smaller ΔG with lower μ e ). Case 2: SnO 2 @TiO 2 (larger ΔG with similar μ e ) and TiO 2 @SnO 2 (smaller ΔG with similar μ e ). (D) Schematic illustration of the bilayered SnO 2 @TiO 2 ETL. The dotted red circles indicate examples of physical defects on the SnO 2 ETL. (E) Illustration of bilayered ETLs with large ΔG and moderate μ e . (F) Histograms of devices fabricated by TiO 2 , SnO 2 , TiO 2 @SnO 2 , and SnO 2 @TiO 2 ETLs (Song et al., 2017 ETL (Rao et al., 2019). However, perovskite crystallization on the surface of these oxides and surface area are issues of concern. Organic ETLs are also widely reported in the PSCs literature. Those can be subjected to low-temperature processing and can easily be used in flexible devices. Those can be divided into two types, namely fullerene-based and non-fullerene ETLs. Organic ETLs form compact layer for efficient electron extraction. Organic ETLs can diffuse to grain boundaries of perovskite film and can passivate the charge traps, thus stabilizing the device. The high cost or the difficulty in synthesis of some organic ETLs are major hurdles for commercialization Yin et al., 2016a).
Composition or interface engineering is proven to be very useful to overcome deficiencies of a particular ETL with good potential (Li et al., 2019a;Ma et al., 2020a). Another layer of a suitable material is used in several reports to increase the charge transport across the ETL . A layer of graphene is introduced with SrTiO 3 to significantly increase the current conduction and the stability of the device . In composite layers, understanding and controlling the interfacial charge transfer at the heterojunction has the paramount importance. Song et al identified that two factors namely the free energy difference (ΔG) between the ETL and the perovskite and the electron mobility (μe) of the ETL are the key factors in regulating the charge transfer and transportation (Song et al., 2017). Two metal oxides, spin-coated SnO 2 and TiO 2 (nanoparticle and anodized, two types) are chosen. Their combinations are tuned as per the understanding of ΔG and μe. The best combination of these two ETLs yields in PCE of 21.1%, in a planar device (Figure 4).
Our approach is gradient energy alignment in case of ETL . Two oxides In 2 O 3 and SnO 2 are used as the ETL one upon another layer (Wang et al., 2020d). It was carried out by a low temperature process. The presence of In 2 O 3 assists in the formation of a low-trap-density perovskite film. The conduction band of In 2 O 3 is shallower than that of the ITO electrode. It highly enhances the charge transfer from the perovskite to ETL and minimizes the loss of V OC . A planar PSC achieves a power conversion efficiency of 23.24% (a certified efficiency of 22.54%) which is higher than that of the control device with only SnO 2 (21.42%). This device retained 97.5% of its efficiency over 80 days, in a N 2 atmosphere without encapsulation. Several methods have been attempted in interface engineering. Doping of oxides by a suitable metal is a widely used strategy Akin, 2019;Bang et al., 2020;Wang et al., 2021). Mobility or conductivity of the oxide can be greatly enhanced upon doping by suitable moiety and it also can enhance the stability by suppressing the carrier recombination in ETL. A few examples are available where an organic layer is used along with the oxide ETL Zhang et al., 2019a;Duan et al., 2019;Wang et al., 2020b). Both the species can enhance the electron transportation in synergy where one of them can be effective in blocking holes. Understanding the diffusion of ions from perovskite to the ETL is very important in designing novel ETLs which can provide stability to the PSCs. The intelligent designing strategy of the ETL has also been proven to be effective. A sparse array of TiO 2 nanorods are nanopatterned and the ETL film is passivated by a ultrathin layer of PMMA:PCBM. The TiO 2 nanorods remain incompletely covered and the resistance would be low in the uncovered regions. This device achieved 23.17% PCE and encapsulated devices retain ∼91.7% of the PCE after 1,000 h of damp-heat test (Peng et al., 2021). A few examples of ETLs with good PV performance and excellent stability are listed in Table 2 along with the respective stability test conditions.

Hole-Transporting Layers
A HTL's valence band maximum must lie between the bandgaps of the light harvester perovskite to receive photogenerated holes. Additionally, the conduction band should be high enough to stop any electron . Various inorganic materials are reported such as NiOx, C-based structures such as graphene oxide (GO), reduced graphene oxide (RGO), C-nanostructures, and various Cu-based materials ( Figure 5A).
NiO or NiO x are highly reported as HTL in both n-i-p and p-in structures . Pure NiO often suffers from low conductivity, which can be improved by proper doping like Li, Cu, Mg (Jung et al., 2015;Yin et al., 2016b). Some NiO films need a high processing temperature. Various efforts have been engaged to lower that temperature. Optical transparency is an another issue needed to be handled to achieve a suitable NiO film for PSCs. High chemical stability of NiO films promotes the stability of PSC Xie et al., 2017;Abzieher et al., 2019). Cu-based several materials CuI, CuSCN, CuO, CuS, CuInS 2 , CuFeO 2 , CuGaO 2 , and CuCrO 2 have suitable band alignments to be a HTL of PSCs (Premalal et al., 2012;Subbiah et al., 2014;Chatterjee and Pal, 2016;Arora et al., 2017;Zhang et al., 2018a;Tamilselvan and Bhattacharyya, 2018;Hou et al., 2019a;Akin et al., 2019;Wu et al., 2019a). The conductivity is generally higher than their organic counterparts such as Spiro-OMeTAD. But, they may suffer from high recombination rate, as investigated by ac-impedance spectroscopy . Thanks to the high hydrophobicity and environment-stability, materials such as CuI can achieve high stability. 93% of the initial efficiency has been retained after 24 days (Sun et al., 2016a). CuSCN is another attractive candidate owing to its high hole mobility of 0.01-0.1 cm 2 V −1 s −1 compared to its organic counterparts such as spiro-OMeTAD with a hole mobility of 4 × 10 -5 cm 2 V −1 s −1 . High interface contact resistance is an obstacle to achieve satisfactory efficiency. Deposition strategies have been tuned to achieve a compact thin film of CuSCN which leads to less contact resistance and more than 20% efficiency is achieved (Arora et al., 2017). Degradation at the CuSCN interface with the Au electrode is an important stability issue that has been solved by introducing an additional material such as RGO in between (Arora et al., 2017). We employed an organic molecule 2,3,5,6-tetrafluoro-7,7,8,8-tetracyanoquinodimethane to modify the CuSCN interface (Hou et al., 2020a). It helps in avoiding the voltage loss and producing an efficient and stable device. Solutionprocessed CuO and Cu 2 O show remarkable properties (Sun et al., 2016b;Chatterjee and Pal, 2016). CuFeO 2 , CuGaO 2 , and CuCrO 2 have also been proven to be stable hole-transporting materials (Zhang et al., 2018a;Hou et al., 2019a;Akin et al., 2019). High efficiencies of 17.1% could be achieved and 90% of the PCE is retained after 200 h. CuSeCN HTL is employed in an inverted planar device which needed solvents much cheaper and less volatile than solvents needed in CuSCN. The device yields 15.97% PCE and retains 83% of its efficiency after 20 days in ambient condition .
Carbon-based materials can produce stable PSCs as HTL because of their good chemical and thermal stability. Their electronic band structure, suitable Fermi levels, and good conductivity make them suitable candidates to be utilized as the HTL and electrode materials simultaneously (Habisreutinger et al., 2014;Ihly et al., 2016). The carbon-based materials are cost effective, flexible, and with high work function. Those made them suitable for broader applications. Some C-based materials such as single-walled Carbon nanotube can be mixed with other HTLs to enhance their properties (Habisreutinger et al., 2014).
Organic HTLs are widely reported in PSCs literature which can be divided into two parts: 1) Small molecules and 2) Polymers. Spiro-OMeTAD and several of its derivatives are most common of them. The hole mobility values are low, so Li-salts are often added to increase the mobility (Kim et al., 2012).
But, a Li-salt may be prone to absorb water, and thus can be detrimental to stability (Snaith and Grätzel, 2006). A very high price of Spiro-OMeTAD is a major obstacle in the path of commercialization. Fluorinated analogs of Spiro-OMeTAD are found to be hydrophobic and highly effective in hole-transporting owing to favorable shifting of the electronic states. PCE of 24.8% is achieved with a V OC of only 0.3 V lesser than the theoretical limit (Jeong et al., 2020). Other small molecules are often added with Spiro-OMeTAD to enhance its hole-transporting properties and to increase the stability.
Jen's group utilized a small conjugated molecule, DRSBDT-2OT. DRSBDT-2OT, with an acceptor-donor-acceptor structure . This molecule has a high molecular planarity due to noncovalent intramolecular interactions between S and O which leads to low reorganization energy and high hole mobility. When mixed with spiro-OMeTAD, this molecule manifests higher hole mobility due to surface passivation effects in comparison with the spiro-OMeTAD, the mixed HTL-based device exhibits a significantly enhanced efficiency of 21.31%. It also enhances the stability. The mixed hole device retains 82% of the initial PCE in ambient condition for a light exposure of 240 h, while the PCE of the control device with only spiro-OMeTAD as HTL reduces to 51% of the initial efficiency ( Figures 5E-I).
In between polymers, PEDOT:PSS is the most commonly used HTL not only for inverted type PSCs, but also for organic solar cells (Jeng et al., 2013). It has several issues, its lower work function than that of perovskites causes a significant amount of potential energy loss. Its acidic nature tends to corrode ITO. So, other polymer derivatives are synthesized to replace it and those could significantly boost the stability (Lim et al., 2014).
Interface engineering is also an intelligent method to boost the efficiency of the device while increasing the stability. We followed this strategy, picked up two most common HTL NiO x and Spiro-OMeTAD, and used their combination as the HTL . Deliberate combination of these two HTL results in 21.6% PCE in a n-i-p planar device which is higher than that of the only spiro-based device's efficiency of 19.1%. It also enhances the stability (Figures 5B-D). 90% of the initial PCE could be retained after exposure of 1,200 h. Currently, a significant amount of effort is engaged to enhance the efficiency and stability of the PSCs by intelligent HTL engineering. The representative examples of HTL with good PV performance and excellent stability are listed in Table 3 along with respective stability test information.

Composition Engineering
Composition engineering of the perovskite is proven to be a successful approach to enhance the efficiency and stability. In many cases, one step taken to change the composition leads to improvement of various stability issues. In perovskite materials such as MAPbI 3 , the conduction band minimum is generally directed by the p orbital of Pb, and the valence band maximum by the s and p orbitals of Pb and the p orbital of I (Yin et al., 2014). The A site cation contributes to the deep energy levels but not to the band edges. The change in A-site does not change the optical or optoelectronic properties. So, the A-site can easily be engineered to tune other physical properties such as thermal stability without affecting the light-harvesting efficiency, which depends on the bandgap. FA can make stronger bonds with the PbX 6 octahedra than that of MA because it has more number of H, which make stronger H-bonding with PbX 6 (Amat et al., 2014). So, the thermal stability increases Pering et al., 2017). Replacing the A-site by an inorganic species such as Cs + can further improve the thermal stability because it can make primary bonding with PbX 6 which is much stronger than H-bonds (Lee et al., 2014a;Ma et al., 2020a). Thermodynamic instability of FAPbI 3 and CsPbI 3 is a barrier to use them in PSCs as non-perovskite phases often occur (Li et al., 2016a;Lin et al., 2018). The tolerance factor of FAPbI 3 has a high value and it is more stable in the hexagonal δ H phase (yellow phase), and the δ H to α phase-transition temperature is higher than room temperature. CsPbI 3 is stabilized to an orthorhombic structure (δ O phase) at room temperature because its tolerance factor is low (Li et al., 2016a). Using alloyed perovskite is an effective strategy to increase the thermal stability while keeping up the structural stability of the perovskite (Hou et al., 2019a). Thin film of alloyed perovskite FA 0.85 Cs 0.15 PbI 3 demonstrated improved stability in a highly humid environment (Li et al., 2016a). In alloyed perovskites, more than one species are used in the A-site. The calculation of tolerance factor of the alloyed perovskite is discussed in Intrinsic Stability. Increased thermal stability of alloyed perovskites contributes to the stability of the PSC. 5-20% of Cs content can increase the thermal stability . The strategy of alloying is also important. We adopted a method of posttreatment where FAI solution in isopropanol is used to treat the perovskite film (Xin et al., 2020). It is proven to be highly effective in passivating the defects. The posttreatment particularly targets the defective grain boundaries of the crystalline perovskite film. 20.62% PCE could be achieved compared to the 19.26% of the control device. The FAI post-treated device retains 95% of the initial PCE after 60 days of storage under N 2 and 78% of the PCE after 30 days of storage in ambient condition. Achieving stable perovskite phase for FAPbI 3 is a big challenge due to thermodynamic factors, as discussed in Structural Stability and Thermodynamic Phase Stability. It is predicted that FAPbI 3 can be more efficient in solar energy conversion than MAPbI 3 . Graetzel et al used a vapor-assisted method, MASCN or FASCN is used to convert yellow δ-phase FAPbI 3 into black α-phase FAPbI 3 , which is also stable . This phase transition should take high temperature, but it can be carried out at 150°C in the presence of SCN − ion. The SCN − attached on the surface and started to form the α-phase, then it continued to the bulk. Once the α-phase is formed, it cannot convert back into the δ-phase due to high energy barrier. The α-phase black FAPbI 3 results in PCE > 23% and good stability.
Bulkier cation than that of FA like guanidinium (GA) has been used to alloy with MA (Jodlowski et al., 2017;Ding et al., 2021). The formation energy is lower than that of the FA alloy. So, the thermodynamic stability of the perovskite increases. The device manifests good environmental stability even at 60°C temperature and continuous solar light illumination. Other N-containing larger groups are used to alloy with MA. The devices show enhanced operational stability in ambient conditions. Ferdani et al proposed that bigger A-site cations can resist the iodide migration which is also large in size, and thus enhance the stability (Ferdani et al., 2019). Methylenediammonium (MDA) and Cs + cations are used simultaneously to alloy with FAPbI 3 , these two cations relieve the strain on FAPbI 3 crystal, and thus facilitate the formation of α-phase FAPbI 3 . More than 25% PCE is achieved with satisfactory stability (Kim et al., 2020a). Partial substitution of the X-site iodide by other halide is also a strategy of composition engineering of the perovskite. But, it must be done with great care because it may cause change of optoelectronic properties of perovskite (Juarez-Perez et al., 2016). Substitution of I − by Br − and Cl − increases the bandgap, thus decreasing the photocurrent. So, a limited amount of substitution is done which increases the interaction of the halide with the PbI 6 octahedra, thus increasing the stability. Multi-cation and multi-halide systems are also reported with enhanced stability (Saliba et al., 2016a;Saliba et al., 2016b). We reported a quasi-heteroface PSCs by a double-layered perovskite film where perovskites with two different bandgaps are used which are achieved by tuning the content of halides (Ren et al., 2020). An additional built-in electric field has been detected which helps in the enhancement of PCE and stability. A Cs 0.15 FA 0.85 PbI 3 /Cs x FA 1-x PbI 3 core-shell heterostructure with a Cs-rich Cs x FA 1-x PbI 3 quasi-shell structure is proven to be effective in reducing defect density and suppressing ion migration. 20.7% PCE is achieved and the encapsulated device retains 95% of its initial PCE after 1,000 h of continuous operation, a projected lifetime of ∼2 years (Peng et al., 2020). McGehee et al studied device performance and photostability across a compositional space of FA and Cs at the A-site at various halide compositions. It is observed that increasing Cs at the A-site rather than more Br at the X-site to raise perovskite band gap is more useful to improve both V OC and photostability . Intelligent tuning of halide content has been proven to be beneficial. Kang et al employed a double-halide passivation strategy by varying the concentration of precursors MACl and MABr to engineer the morphology, intrinsic defects, and optical band gap of the perovskite. Better-aligned energy levels, suppressed grain boundary recombination losses, and cascade charge transfer yielded a device with PCE of ∼20.3% with excellent air, photo, and thermal stabilities (Tyagi et al., 2020). All-inorganic PSCs are highly stable and improving their efficiency can be a key to achieve commercial PSCs Zhou et al., 2020;Du et al., 2021).
Although many authors treat dimensional engineering of perovskite as "additive engineering," dimensional engineering arises from change in certain component of the perovskite. So, this is discussed in this section. In the previous part, we discussed how substitution of A-site by bulkier cation can increase the stability. But, with increasing size of A-site moiety, the distance between the BX 6 octahedra also increases and after a certain point, 3D corner-sharing geometry becomes impossible. Hence, the structure would become 2D, layered BX 6 octahedra intercalated with bulky A-site anions. The evolution of 3D perovskite from 2D also can be represented in terms of the thickness of the layer n. The 2D perovskite can be represented by a formula (A′) m (A) n-1 B n X 3n+1 , where A′ can be divalent (m 1) or monovalent (m 2) cations that form a bilayer or monolayer connecting the inorganic (A) n-1 B n X 3n+1 2D sheets; the layer thickness of metal halide sheets is denoted by n and that can be adjusted by tuning precursor composition (Yuan et al., 2016;Zhou et al., 2018). A large, high aspect ratio organic cation can take the A′ site. The ideal 2D perovskite is where n 1. When 1 < n ≤ 5, the species is labeled as quasi-2D. A mixture of 3D perovskite and lower n-phase (n ≤ 3) can form in case of high n-values (n 30-60); these are called quasi-3D perovskite. When the layer is continuous, i.e., the layer thickness tends to infinity (n ∞), it is a 3D perovskite (Mei et al., 2014). The evolution of perovskite from 2D to 3D with respect to n is described by a schematic in Figure 6A. 2D halide perovskite layers can hypothetically be obtained by cutting the corresponding 3D perovskite along <100>, <110>, <111> crystallographic planes, which would orient the 2D perovskite along the plane it has been cut through. The <111> oriented 2D perovskites have a formula A′ 2 A q-1 B q X 3q+2 (q > 1) and can only be constructed from group 15 B 3+ ions (e.g., Bi, Sb, As) (Vargas et al., 2018). They are often efficient in solar photon harvesting owing to their p-type like character and smaller effective masses for electron and hole. In 2D or quasi-2D perovskites, the bandgaps are higher than that of their 3D counterpart and may not be suitable to be an efficient light harvester. But, those have superior moisture and thermal stability than that of their 3D counterpart because the hydrophobic big A-site cation minimizes surface energy of the crystal. 2D and 3D perovskites are often mixed and that forms quasi-2D perovskite where 2D and 3D crystals are periodically repeated. Addition of 2D perovskite can greatly enhance the stability of the PSCs due to its superior stability under heat, moisture, and light (Tsai et al., 2016). Also, ion migration is suppressed in quasi-2D perovskite due to larger activation energies for the formation of point defect. Liu's group has reported highly efficient 2D/3D PSCs (Huang et al., 2019a). 2D perovskites can be further classified into Ruddlesden-Popper (RP), Dion-Jacobson (DJ) phases and phases with alternating cations in the interlayer space (ACI) (Li et al., 2019b;Paritmongkol et al., 2019;Zhang et al., 2019c). These three phases are explained in Figure 6B. A RP phase 2D perovskite is generally represented by formula A′ 2 A n-1 B n X 3n+1 , where A′ is an aryl ammonium or an alkyl cation. Comparatively weak van der Waals gap forms between a bilayer of monovalent cations and two adjacent lead halide sheets (Paritmongkol et al., 2019). DJ phase perovskite can be represented by A′A n-1 B n X 3n+1 , where A′ is a diamine. A diamine compound is employed and the two amino groups from the same moiety make H-bonding with two ends of two adjacent metal halide sheets. This makes the DJ phase more stable (Li et al., 2019b). The ACI phase perovskite has formula A′A n B n X 3n+1 , where the large A′ cation fills the interlayer and the small A cation resides in the metal halide sheets (Zhang et al., 2019c). The H-bonding characters in the RP and DJ phases with the metal-halide are described in Figure 6C.
A large spacer cation 2-thiophenemethylammonium is incorporated into the 3D perovskite to form a 2D/3D hybrid structure Zhou et al., 2019b). It induces crystalline growth and orientation, passivation of trap states, and hindering the motion of ions. This helps in improved carrier lifetime and reduced recombination losses. It yields 21.49% PCE and the encapsulated device retains 99% of PCE even after 1,680 h of operation under ambient condition. The suppressed ion migration results in operational stability of 1 year. A novel 2D RP perovskite (AA) 2 MA 3 Pb 4 I 13 yields 18.42% efficiency (Wu et al., 2020a). Phenylethylammonium is used to facilitate formation as well as control of 2D perovskite, which is also able to suppress charge recombination in CsPbI 3 . The device achieved 17% PCE and 94% of it could be retained over 2000 h of storage under low humidity (Ye et al., 2020). A molten salt spacer n-butylamine acetate is proven to be made a strong coordination bond with the perovskite and a phase-pure quantum well film is formed with vertically aligned micro-grains. The device yields 16.25% efficiency, 1.31 V of V OC . It retains ∼90% efficiency under 65 ± 10% humidity for 4,680 h, under operation at 85°C for 558 h, or continuous light illumination for 1,100 h (Liang et al., 2021). Fluoroarene-based RP 2D perovskite can highly stabilize the device owing to the ultra-hydrophobicity of pentafluorophenylethylammonium spacer cation . Large area PSC module could be fabricated with excellent ambient-and photo-stability when 4-fluorophenethylammonium iodide has been employed (Lee et al., 2021). Aromatic formamidiniums are proven to be effective spacer due to its ability to make multiple NH. . .I hydrogen bonds between itself and the (PbI 6 ) 4− . Chen's group employed benzamidine hydrochloride as a spacer and a highly efficient device is achieved with PCE 23.36% and long-term thermal stability under 80°C, over 1,400 h operation .
Presence of 2D perovskite in a light harvester 3D perovskite may affect the transportation of the photogenerated charges. To avoid this, 2D-3D heterostructures are introduced (Wang et al., 2017c;Lee et al., 2018a). The 2D perovskite spontaneously segregated at the grain boundary with the 3D perovskite. So, it can reduce the loss due to recombination, but it does not affect the charge transport. These help in the improvement of both performance and stability. A stable and highly crystalline 2D (C 4 H 9 NH 3 ) 2 PbI 4 film is grown on the top of a 3D film using a solvent-free solid-phase in-plane growth method. A high built-in potential is observed at the 2D/3D heterojunction which results in high photovoltage. A steady-state efficiency of 24.35% is achieved and the encapsulated device could retain 94% of the initial PCE after 1,056 h under the damp heat test (85°C/85% relative humidity) and 98% after 1,620 h under full-sun illumination (Jang et al., 2021a).
Though DJ phase 2D perovskites are more stable than those of their RP counterparts, the devices lack in efficiency. A device based on DJ phase (PDA)(MA) n-1 Pb n I 3n+1 2D perovskite yielded 13.3% PCE but manifested exceptional stability. More than 95% of PCE has been retained after 168 h of damp-heat test (85°C, RH 85%) and 4,000 h operation in ambient . 1,4-butanediamine iodide-based DJ perovskite yields in a PSC with 17.91% efficiency and good stability . Another stable device is reported with 2D perovskite ODAPbI 4 , with octyldiammonium ion . The PSC shows PCE of 21.6% and high stability under a constant relative humidity of 85% and ambient temperature. Insightful reviews are available on 2D perovskite-based PSCs and even only focused on RP or DJ-based 2D perovskite. Those can be helpful for readers to study a wide range of examples (Gao et al., 2018;Liang et al., 2018;Zhang et al., 2020;Li et al., 2021a;Liu et al., 2021b). The linear thermal expansion coefficient of perovskite is higher (∼45 × 10 -6°C−1 ) than that of the glass substrate (10 × 10 -6°C−1 ). Most of the perovskite films for PSCs are deposited at elevated temperature and when the films are cooled to room temperature, the perovskite contracts more than that of the substrate. This causes tensile stress in perovskite film which leads to instability (Ramirez et al., 2018). Various steps have been taken to improve stability under thermally induced mechanical stress and to resist fracture of the perovskite film, like X-site engineering. This topic is discussed in detail in some recent reviews (Boyd et al., 2019;Li et al., 2020b).

Additive Engineering
Other than the necessary or basic components to make a working PSC, other materials can be added to a certain part to improve the efficiency and stability (Ghosh and Singh, 2019;Wang et al., 2020b). A wide array of materials has been utilized and many of them have been proven to be beneficial. These are called additives which is kind of an umbrella term encompassing a huge matrix of materials with different functionalities. The additives can be categorized into two parts as per their time of addition during the fabrication of the PSC: 1) Additives used with precursors. 2) Posttreatment additives. The category 1, additives that are added during the perovskite formation, those can be added with different precursors like, 1) additive added with the perovskite precursor of the one-step method, 2) the additive added with the organic solvent of the two-step perovskite formation, 3) the additive is added with the antisolvent. The additive engineering is a huge field of research and insightful reviews solely on PSC additives are available (Peng et al., 2020;Ren et al., 2020). Some examples of additives that substantially influence PSCs are discussed in this section. There are some additives that are beyond classification or multifunctional. Liu et al used a molecule 2,2-difluoropropanediamide (DFPDA), as predicted from theoretical calculations for passivation of defects (Cai et al., 2021). It is noticed that the carbonyl group can tightly bind with the Pb 2+ , thus restricting the movement of ions. This additive is found to be multifunctional and can ''ameliorate" all the instability issues. 22.21% of PCE is reported in a n-i-p device with considerable stability. Representative additive species are summarized in Table 4 including the device configuration, stability, and PCE.

Lewis Acids
A Lewis acid can accept a lone pair of nonbonding electron, and thus can passivate electron-rich defects . The added Lewis acid species can form a Lewis adduct with any under-coordinated halides and PbX 3 − via ionic or coordinate bonding. Restricting these species via strong bonding can eliminate the corresponding defects or traps that are responsible for nonradiative recombination. Positively charged metal cations are proven to be effective in passivation. Two groups have demonstrated that a little excess of PbI 2 can effectively passivate the perovskite film Yang et al., 2020a). That leads to longer carrier lifetime and reduced halide vacancies for high-efficiency perovskite solar cells. Randomly distributed PbI 2 may have detrimental effects (Tumen-Ulzii et al., 2020). Ordering can be achieved by ligand modulation. 22% PCE could be achieved over 20% PCE of the control device and 72% of the initial efficiency is retained after 360 h constant illumination at maximum power point (Yang et al., 2020b). CdI 2 treatment significantly reduces interfacial charge recombination loss via strong Cd-I bonds which reduces I-vacancy at the surface. 21.9% of PCE was achieved with voltage deficit of only 0.31 V than the theoretical value. 92% of initial efficiency is retained after constant illumination at 1 Sun intensity for 1,000 h . Group 1 cations, Li + , Na + , K + are proven to be effective in passivation which can change the lattice constants of the perovskite crystal, while bigger size Cs + and Rb + cannot change the lattice constants (Li et al., 2020c). Li + and Na + can be diffused from the HTL and the source respectively (Bi et al., 2017;Abdi-Jalebi et al., 2018a).
Several reports are available on the improvement of PCE and stability upon the addition of KI (Kubicki et al., 2017;Abdi-Jalebi et al., 2018a;Cao et al., 2018). K + can passivate following the way mentioned before while I − can fill up the iodide vacancies of perovskite. NaF is also found to be effective in the same way.
Other metal cations such as Zn 2+ , Mg 2+ , Ni 2+ , and Eu 3+ -Eu 2+ are also proven to be effective and provided PSCs with high stabilities (Saliba et al., 2016b;Poindexter et al., 2017;Abdi-Jalebi et al., 2018b;Gong et al., 2018;Wang et al., 2019b;Kooijman et al., 2019). Several organic molecules are also found to be useful as additive Lewis acids for PSCs (Yang et al., 2020b;Luo et al., 2020;Peng et al., 2020). Fullerene derivatives are very suitable for this due to their high electron affinity. Many of them have multiple utilities. They can passivate the trap states by receiving electrons on one hand and also they can cover the grain boundaries on the other hand, thus restricting the movement of mobile ions along grain boundaries by creating steric hindrance (Shao et al., 2014;Xu et al., 2015;Saliba et al., 2016b;Chiang and Wu, 2016;Poindexter et al., 2017;Abdi-Jalebi et al., 2018b;Gong et al., 2018;Wang et al., 2019b;Kooijman et al., 2019). The C 60 and C 60 derivatives have shown limited performance in enhancing the PCEs and 80-85% of efficiency could be retained over long exposure, as found in early studies. Phenyl-C 61 -butyric acid methyl ester (PCBM) and its derivatives are also proven to be beneficial. α-bis PCBM used as an additive and 20.8% efficiency could be achieved in a mesoporous TiO 2based device . 96% of the initial efficiency is retained after 600 h of continuous exposure.

Lewis Bases
A Lewis base contains an electron-rich moiety with a lone pair of electron which can easily be donated to a suitable species . In case of perovskites, a Lewis base can bond with the Pb 2+ cation and thus passivate defects (Yang et al., 2018b). As an electron-rich moiety is required, the Lewis base is usually an organic molecule with N, O, or S as the electron donor site.
N-donor Lewis bases generally contain pyridine or primary amino group which have a lone pair of electron on N, which can be readily donated. Addition of pyridine improves the film quality, passivates the surface defects, and the PCE is increased to 19% from 16.9% . The improvement of film quality is caused by the pyridine that made the film formation slower. Long-chain polymers with pyridine or primary amino group can passivate the defects effectively and deliver an efficient and stable device. (-NH 2 ) tails can bond with Pb 2+ to facilitate defect passivation and they also can fill in A-site vacancies (Zhou et al., 2019c;Wu et al., 2020b). Amine-based Lewis base additive results in highly stable PSCs that can retain 95% of the initial efficiency under humid condition at room temperature . Small molecules such as urea, caffeine have proven to be effective. Those can retard the perovskite film formation and high efficiency and stability can be achieved by successful defect passivation. Particularly, caffeine can make a strong bonding with PbI 2 , 20.25% of efficiency can be achieved with thermal stability at 85°C, 1,300 h (Wang et al., 2019c;Wang et al., 2019e). N-based Lewis bases theophylline and theobromine have also been proven to enhance efficiency and stability (Wang et al., 2019e).
Commonly used solvents such as DMF, DMSO can interact with the perovskite by the lone pair of oxygen on O (Lee et al., 2014b;Jeon et al., 2014). Poly (methyl methacrylate) PMMA can enable heterogeneous nucleation when added with the antisolvent . It promotes heterogeneous nucleation by making an adduct with PbI 2 , which causes random formation of nuclei. This minimizes the Gibbs free energy of formation and the crystal can grow in thermodynamically preferred orientation. These pave the way for the formation of smooth perovskite film with fewer defects and a satisfactory efficiency of 21% is achieved. Polymeric species with C O group has proven to be effective in passivating the defects in the perovskite, hence stabilizing the devices (Chang and Wang, 2020;Jiang et al., 2020;Ren et al., 2020).
S-donor Lewis bases are more effective in electron donation than their O-counterparts. Several S-based organic molecules such as thiourea, thiophene, and their derivatives are proven to be able to make strong bonds with under-coordinated Pb 2+ Guo et al., 2019). Organic molecule with both N and S donor sites, a thiadiazole derivative, is proven to be very effective in stabilizing the perovskite film. 19.04% of efficiency was achieved with good stability under heat and humidity (Zhu et al., 2018b). Zhang et al reported simultaneous usage of S-donor Lewis base and a PCBM derivative Lewis acid (Zhang et al., 2018b). The Lewis acid and Lewis base synergistically passivate PbX 3 and Pb 2+ antisite defects and the charge-carrier separation is improved. 21.7% efficiency was achieved over 19.3% efficiency of the control device. Significant stability is achieved, 93% of PCE is retained under ambient condition over 3,600 h. Multi-ligandbased pentaerythritol tetrakis (3-mercaptopropionate) is effective in suppressing Pb 2+ and Pb 0 defects (Wu et al., 2020c). S-moiety in WS 2 has also proven to be effective . Poly (thioctic acid) is highly hydrophobic and presence of both C S and C O can effectively passivate the defects and yield a stable device (Xin et al., 2020).
Sargent et al have experimented Lewis bases with different donor moiety such as C O, S O, As O, P O, end groups (Na Quan et al., 2020). The dative bond of P O: Pb has the strongest binding energy. So, P-based Lewis bases are proposed to be very promising in passivating defects in perovskite. Several P-based species are reported with satisfactory efficiency and stability (Yang et al., 2020b;Wang et al., 2020d;Ji et al., 2020).
Zwitterionic species consisting of two functional groups can passivate both positively charged and negatively charged defects simultaneously. A few reports of successful implementations of Zwitterions to enhance PSC efficiency and stability are available (Zheng et al., 2019a;Zhou et al., 2019d). Graetzel et al recently employed a crown ether compound to modify the perovskite film (Su et al., 2020). It acts as a host in host-guest complexes on the perovskite surface, which significantly reduces surface electronic defects. Non-radiative recombination is suppressed by 40% and moisture permeation is also greatly reduced. More than 23% of PCE is achieved with significant stability in ambient and operational conditions.

Ionic Liquids
Ionic liquids (IL) are salts consisting of a large organic cation and small or large, inorganic or organic anion. Owing to the larger sizes, the interionic space is more and the electrostatic forces between them are lower. So, the salts become liquids below 100°C, often at room temperature. ILs are not volatile and not flammable. ILs have high ionic conductivity, electrochemical stability and can dissolve various ionic and covalent compounds (Ghosh and Singh, 2019;MacFarlane et al., 2016;Rogers and Seddon, 2003;Welton, 1999). Certain ILs are hydrophobic in nature. ILs can improve the PSC by various ways. Imidazolium-based ILs are proven to be effective in forming high-quality perovskite film by uniform nucleation and preventing rapid crystallization (Shahiduzzaman et al., 2015) ( Figure 7B). The film quality can be tuned by varying the viscosity of the IL (Shahiduzzaman et al., 2017). Imidazoliumbased ILs have also shown excellent hydrophobic property (Salado et al., 2016). The PCE has been improved up to 20 days of storage due to continuing evaporation of the ILs. These devices are stable up to 90 days (Saliba et al., 2016a;Singh and Miyasaka, 2018). A fluorous-functionalized imidazolium-based ionic salt (FIm) and an ionic salt 1-(4-ethenylbenzyl)-3- (3,3,4,4,5,5,6,6,7,7,8,8,8-tridecafluorooctylimidazolium iodide (ETI) have been proven to be useful in enhancing thermal and atmospheric stability of the PSCs Xia et al., 2019). Triazolium-based ILs have strong hydrophobicity due to the formation of self-assembly (Wang et al., 2019f). The amine group bonded with Pb 2+ to passivate the defects, thus an efficient and stable device is formed. Zhou et al reported an amino group functionalized imidazolium-based IL with strong hydrophobicity (Zhou et al., 2019c). It makes H-bonding with the iodide. The film formation is retarded and a high-quality crystalline film is formed. The H-bonding restricts the iodide ion, which boosts the stability. Moreover, the hydrophobic nature highly helped in stability. An efficient device is formed which can retain 94% of the initial PCE after 840 h of device. Imidazolium-based ILs are widely used and demonstrated good moisture and thermal stability. They are proven to be also capable of restricting the movement of A-site cation (Wang et al., 2019g). ILs are not only capable of modifying the perovskite film but they are also useful in interface modification. The oxide charge-transporting layer often suffers from defects, nonuniformity, and high roughness. An IL can modify the interface between the oxide film and the perovskite (Wojciechowski et al., 2014). Improved crystallinity of the perovskite film is observed. Yang et al successfully modify a mesoporous titania surface with an imidazolinium-based IL . The roughness of the oxide film reduces significantly and 19.62% PCE is achieved with good stability. The IL also increases the electron mobility of the ETL and reduces trap-state density by restricting ion migration in perovskite. Piperazinium iodide containing both R 2 NH and R 2 NH 2 + groups on the same six-membered ring acts as both an electron donor and an electron acceptor to react with different surface-terminating ends on perovskite films. This is highly effective in suppressing non-radiative recombination losses and in passivating defects, results in PCE of 23.37%, one of the highest in inverted devices . Snaith et al reported efficient and highly stable PSCs with p-i-n architecture. The exceptional stabilities are achieved by an imidazolium-based IL and a piperidinium-based salt. ILs with bigger anionic groups are also reported and it is observed that the cationic part of the IL directs the modification property to the perovskite (Bai et al., 2019). A series of water-soluble triazolium ILs are used in MAPbI 3 -based PSCs, more than 20% of efficiency is reported with good stability (Wang et al., 2019f). Hagfeldt et al used a polymeric molten salt poly (1-vinyl-3-ethyl-acetate) imidazole tetrafluoroborate as an additive (Wang et al., 2020e). It has various anchoring sites that can bond with Pb 2+ defects at grain boundaries and the interface of the perovskite film by forming a coordination bond. That could efficiently passivate the defects and boost the photo-, thermal-, and moisture stability of perovskite films. 21.4% PCE is achieved in the p-i-n device and more than 92% of the initial efficiency is retained after 1,200 h under continuous full Sun illumination at 70-75°C ( Figures  7C-J). Protic amine carboxylic acid ILs are environmentfriendly and bifunctional; they can bond strongly with the perovskite by forming C O. . .Pb chelate bonds and N-H. . .I hydrogen bonds and high-efficiency, high-stability devices could be achieved (Chao et al., 2020). Various surface defects in perovskites and how different additive engineering strategies can be used to passivate them are described by a schematic in Figure 7A.

Surface Modifier
In the previous section, we have discussed that ILs can be used to modify the interfaces. Other compounds also can be used to modify the surface of charge-transporting layer and the perovskite. TiO 2 ETL are prone to UV degradation. Various fullerene-based molecules such as C 60 -SAM (self-assembled monolayer), PCBM, and their derivatives are used to improve the UV-stability of titania films (Abrusci et al., 2013;Wojciechowski et al., 2014;Chen et al., 2016). Many fullerene derivatives are successful to increase the open circuit voltage by surface modification. Various ILs are proven to be effective surface modifier of ETL oxides. Several approaches are also taken to modify the HTL surface. V 2 O 5 is used to replace LiTFSI with Spiro HTL and an efficiency of 20.1% is achieved (Wang et al., 2019h). V 2 O 5 is also proven to be beneficial for efficiency and stability while being used with PEDOT:PSS. Perovskite films are also modified by several species and dopant. Several Cs-based perovskite quantum dots (QDs) are used to modify the perovskite (Liu et al., 2018a;Zheng et al., 2019b). The QDs can improve the optoelectronic properties and charge separation. Other inorganic salts PbSO 4 , Pb 3 (PO 4 ) 2 are used to enhance water resistance . Luminescent perovskite nanoparticles are stabilized by crosslinking and high photoluminescence quantum yield could be retained even after one and half years of storage (Jang et al., 2021b). A novel molecule: p-phenyl dimethylammonium iodide (PDMAI) with ammonium group on both terminals as the modifier of the perovskite layer is also proven to be an effective passivation material . The hydrophobicity induced by alkyl and aryl tail gives environmental stability. In another report, we introduced a thin layer of poly (triaryl amine) on the top of the perovskite film before annealing (Hou et al., 2019b). It causes a smooth and compact perovskite film with passivated surface defects and grain boundaries. The modified film shows faster charge transfer capability and longer lifetime. The hydrophobic polymer on the top of the perovskite induces stability. The PSC with modified film have 17.77% PCE compared to 16.15% of the control device. The modified device retained 83% of initial PCE under ambient condition over one month, while the control PSC only can retain 56% of the initial PCE. A high-pressure nitrogenextraction strategy is used during a Slot-die coating of perovskite on a large area substrate. 19.4% of PCE is achieved for a 40 × 40 mm 2 module, which is highest on a large-area module . We introduced a cross-linking agent aluminum acetylacetonate [Al (acac) 3 ] as an interface layer between ETL and the perovskite. Well-matched energy levels and improved grain size and crystallinity of the perovskite helped in achieving 20.87% PCE in a flexible PSC. The devices have manifested outstanding moisture stability and bending resistance .

To Prevent Electrode-Induced Degradation
Degradation mechanisms induced by metal electrodes at the dark side are discussed in detail in Interaction With Electrodes and Degradation at Interfaces. Some strategies to avoid electrodeinduced degradation are discussed here. Either an alternative electrode can be employed or an additional barrier layer can be introduced to prevent degradation due to metallic electrodes . Carbon-based materials are presented to be most suitable candidate to replace the metal electrodes (Habisreutinger et al., 2014;Gholipour et al., 2016;Aitola et al., 2017;Grancini et al., 2017;Wang et al., 2020a). Some C-structures also can be simultaneously used as the charge-transporting layer (Wang et al., 2020a). Carbon nanostructures with mesoporous oxides like titania have been reported to give stable devices (Habisreutinger et al., 2014;Gholipour et al., 2016;Aitola et al., 2017;Grancini et al., 2017;Huang et al., 2018;Tian et al., 2018). Another alternative is transparent conductive oxides such as ITO or MoO 3 (Bush et al., 2016). Graphene-based electrodes have also been utilized to stabilize organic photovoltaic devices (Walsh et al., 2021). Those can produce stable devices but, whether conducting oxides or Carbon-based materials, the conductivities are much lower than that of metals. So, adding a metal contact on the top of the electrode can enhance the efficiency.
An effective barrier layer in between the metal electrode and the charge-transporting layer can prevent the metal and the halogen species from the perovskite to come in contact with each other. C-based materials and ITO have been used as the interface layer. Other than that, various oxides and organic molecules have been utilized such as Cu  Frontiers in Electronics | www.frontiersin.org August 2021 | Volume 2 | Article 712785 phthalocyanine, MoO 3 /SnO 2 , MoO 3 /Al, SnO 2 , cross-linked polymers (Zhao et al., 2014;Beal et al., 2016;Sanehira et al., 2016). Morphology of the barrier is very important in this regard, a rough morphology can introduce channels for diffusion of halides from the perovskite and metal from electrodes, but a planar morphology can make the barrier impermeable, thus enabling a highly stable device . Excellent reviews are available where measures taken to prevent electrodeinduced degradation like interface modification, C-electrode are discussed in detail and various examples are listed (Li et al., 2020b;Hadadian et al., 2020).

Encapsulation
One of the most effective tools to improve the stability of PSCs is encapsulation i.e. packaging of the device with suitable materials. Encapsulation can resist external factors causing degradation of the PSCs. An effective encapsulation should resist external moisture, oxygen to reach the perovskite layer. It should stop the irreversible loss of volatile organic solvents or halides from the device and should be able to withstand mechanical stress during thermal fluctuation of the device. It should withstand external mechanical factors such as rain, snow, and hail (Wang et al., 2020a). If the encapsulation is also on the top-side, i.e., toward the light, it should be easily cleanable and should be enough transparent to allow ample amount of light to reach the actual device (Kempe, 2017). The ability of an encapsulation material can be measured by water vapor transmission rate (WVTR) (Lee et al., 2018b). Moisture instability of perovskite is detected at early stage and multilayered inorganic/ organic film with Al 2 O 3 is used for encapsulation (Chang et al., 2015). Epoxy resins are widely used as an edge-sealant that is used to stick a glass on the metallic electrode (Dong et al., 2016). Surlyn encapsulant is also reported to produce a stable device . Various aspects of epoxy resin encapsulation are studied. Cross-linked grain encapsulation facilitates simultaneous enhancement of moisture tolerance and defect passivation of perovskite. 22.7% of PCE is achieved for 1.55 eV bandgap perovskite and >90% of the initial PCE can be retained for over 10,000 h, in shelf storage under ambient condition (Xiao et al., 2021). The representative encapsulation results are listed in Table 5.
Various encapsulation methods are depicted in Figure 8. 2) Humidity-freeze test: This is performed to determine the ability of the module to withstand the effects of high temperature and humidity followed by low temperature. First, it is kept at 85°C, RH of 85% for 20 h followed by rapid freezing to 0°C (100°C h −1 maximum) and then to −40°C for 4 h. These can be repeated for maximum 10 cycles. 3) Damp heat test: It is carried out to test the performance of the module over long time under humidity. The module is tested under 85 ± 2°C temperature, 85 ± 5% RH for 1,000 h.
Holzhey and Saliba penned an excellent overview on this including a useful flowchart that can be a guide for researchers, who want to test devices for IEC 61215 (Holzhey and Saliba, 2018). In case of all the three tests, the requirement is that the degradation of PCE must not exceed 5% of the value measured before the test. The perovskite solar cells are highly prone to degradation and the technology is also very new compared to Si-based or thin film-based solar cells. So, only a few reports are available on attempts to succeed the IEC criteria ( Table 6). McGehee et al concentrated on encapsulation of PSCs and employed an effective packaging by taking in account multiple designing parameters . Their device succeeded the IEC 61215:2016 damp heat test. More than 90% of the PCE is retained after 1,000 h of operation at 85°C, 85% humidity ( Figure 8B). The same encapsulation also helps the PSCs to succeed IEC 200 thermal cycles. Razera et al combined optical, electrical, and microstructural characterization to elucidate degradation paths and attempted to fabricate a PSC that can pass the IEC criteria (Razera et al., 2020). Mendes et al developed a laser-assisted hermetic encapsulation method, which is highly effective to protect the device from external factors and the devices are able to comply with the IEC 61646 standard (Emami et al., 2020;Emami et al., 2018). Lamination of the device using a conducting polymer doped by non-hygroscopic borane induces high stability and only 5.2% degradation of the PCE is observed after 1,000 h of damp-heat test at 85°C, 85% humidity (Heo et al., 2019). Ho-Baillie et al used gas chromatography-mass spectrometry to identify volatile products from PSC designed by them (Shi et al., 2020b). A cost-effective encapsulation is deliberated using polymer/glass stack where the polymer is injected after fixing the glass. A device with multiple A-site cation and multiple X-site anion perovskite achieves and exceeds IEC 61215:2016 standard by surviving more than 1800 h of the damp heat test and 75 cycles of the humidity freeze test ( Figures 8C,D). Han et al reported printable PSCs to qualify IEC61215:2016 criteria with 9,000 h of operational tracking. 5-ammoniumvaleric acid at grain boundaries inhibits decomposition, reconstruction, and irreversible ionic migration in the MAPbI 3 perovskite (Mei et al., 2020). A few reports are available where the device is tested under damp-heat conditions at 85°C temperature and 85% relative humidity, but the degradation of PCE is far more than 5% (Table 7). Those results are also important because those can guide to adopt strategies to achieve highly stable PSCs. Inverted devices are often proven to be highly stable and moreover they are more suitable for tandem devices (Li et al., 2021b;Chen et al., 2021). Compositional engineering of the perovskite has primary importance (Kim et al., 2019a;Matsui et al., 2019). Several reports are available where highly stable devices are achieved by the modification of the charge-transporting layer or by modifying the interface by a stabilizing or hydrophobic material (Kim et al., 2019b;Lee et al., 2019;Agresti et al., 2020;Kim et al., 2020b;Kumar et al., 2020). Metal electrodes have been replaced by a carbon electrode that is water resistant and can improve stability (Wu et al., 2019c). It promises a cost-effective alternative to the noble metal-based electrodes. As discussed before, several encapsulation strategies are proven to be effective (Lv et al., 2020). Reports on IEC 61215 and IEC 61646 test results on PSCs are listed in Tables 6, 7. In Table 6, the devices that passed (or nearly passed) the test are tabulated where Table 7 listed other excellent and representative results. The determining factors that enable exceptional stability in the mentioned devices are mentioned in the "Remarks" column.

CONCLUSION AND OUTLOOK
The PSCs' efficiency rapidly crossed 25% for laboratory scale devices. But stability and reliability are needed to be developed quickly for commercialization. In fact, the commercialization needs simultaneous development of high efficiency devices with large area along with long-term stability. To prevent the degradation, understanding the mechanisms for degradation is highly valued. Promising development of power conversion efficiency of PSCs triggers a race to achieve higher efficiencies, but all the degradation pathways are not well understood until now. More research should be engaged for better understanding of degradation mechanisms. That knowledge will greatly help in adopting strategies to further improve the efficiency and stability of PSCs. Various characterization methods of spectroscopy, electrochemistry, electron microscopy are employed to understand the degradation mechanisms in PSCs. Combination of these techniques with proper theoretical support can be helpful for better understanding of degradation. Various methods are employed to overcome the factors of instabilities. Many approaches are proven to be multitasking, i.e., one strategy or one added material can improve multiple factors. In between the recent trends, additive engineering is looking very promising. More research efforts should be engaged on various novel additives such as Lewis acid/base, IL, surface modifier. Those are proven to greatly influence the perovskite film and also can greatly help in stability from moisture and heat. Encapsulation is another important factor for commercialization. Most of the PSCs fully or partially passed the IEC criteria, all employed novel encapsulation methods. The causes of instability may lie on different factors of a particular device, so multipronged approaches can be effective to tackle those issues simultaneously and to produce a stable device with high efficiency. As the PSCs have already achieved satisfactory efficiency and the highest efficiencies are being updated quite swiftly, the focus should be concentrated on operational stability. Standard protocol for testing the operational stability has already been established. Any new device should be examined with the recommended standard of stability testing. Preparing a large-scale device is very important. These two can lead PSCs to solar module for practical application. Tandem devices where generally a Si-solar cell is attached with PSCs to harvest the lower energy solar photons not absorbed by perovskite are highly promising because the output can be enhanced while the stability of the PSC can be increased. The worldwide PSCs research community is already large and the volume of published literature is increasing every day. Establishing a database, customization and accepting standard testing protocols are the need of the hour.

AUTHOR CONTRIBUTIONS
SM has written the main draft of the manuscript. YZ and XZ have provided project motivation, guidance and corrected the manuscript. All authors contributed to the article and approved the submitted version.