1 Introduction

As one of the sustainable clean energy sources, photovoltaic technology has been developed vigorously in recent decades. Among them, lead halide perovskite solar cells (PSCs) stand out due to their rapidly increasing power conversion efficiency (PCE), and are currently considered as the most encouraging and promising candidate for the next generation photovoltaic technology [1,2,3,4,5,6,7]. In a PSC, the active light-harvesting materials are generally metal halide perovskites with a structural formula of ABX3 (A: monovalent cation, CH3NH3+, HC(NH2)2+, Cs+; B: divalent metal cation, Pb2+, Sn2+; X: halide anion, I−, Br−, Cl−) [8,9,10,11], which are responsible for converting the incident sunlight into free carriers in the devices, therefore playing a crucial role in the conversion of energy. The extraction of free carriers depends on the electron/hole transport layers (ETL/HTL) that sandwich the perovskite layer and transport charge carriers to corresponding electrodes. Therefore, typical configurations of PSCs involve conductive glass/ETL/perovskite/HTL/electrode (n-i-p) or conductive glass/ HTL/perovskite/ ETL/electrode (p-i-n), as shown in Fig. 1 [12,13,14,15,16].

Fig. 1
figure 1

a Basic structures of PSCs: (i) mesoporous structure with cathode/compact layer (TiO2)/mesoporous layer (TiO2 or Al2O3)/perovskite/HTL/anode, (ii) conventional structure with cathode/ETL/perovskite/HTL/anode, and (iii) inverted structure with anode/HTL/ perovskite/ETL/cathode. b Schematic representation of the interfaces in a planar-structured PSC. ➀, ➁, ➂, and ➃ represent ETL/perovskite interface, perovsktite/HTL interface, cathode/ETL interface, and HTL/anode interface, respectively. Reproduced with permission from Ref. [12]. Copyright 2018, Wiley-VCH

Full size image

PSCs show advantages over commercially available solar cells in terms of low cost and little energy consumption in device fabrication. As the dominating share of the photovoltaic industry, monocrystalline silicon solar cells suffer from high fabrication cost, high-temperature preparation, bulky active layer and long payback time. In contrast, PSCs can be fabricated by solution methods at low temperature with very low fabrication cost and little material consumption [17,18,19,20,21,22,23]. GaAs solar cells, which have realized the highest PCE of single-junction solar cells, are only used in small-area markets such as space stations [24], while large-area PSCs (over 60 cm2) with PCE over 20.5% have been achieved [25, 26], promising their large-area applications. Other thin film solar cells, such as copper indium diselenide (GIGS), cadmium telluride (CdTe), and quantum dot solar cells, are subject to the slow growth of efficiency [27]. Notably, the certified PCE of PSCs has exceeded 25.7% after one decade study, owing to the excellent photophysical properties of perovskites [5]. Considering the theoretical Shockley–Queisser limit of ~ 31% [28,29,30,31], there is still a relatively large space for PCE improvement for PSCs. Despite these advantages, device stability of PSCs is a critical issue to their commercial applications [32,33,34,35,36,37]. It has been recognized that device stability is closely related to a series of chemical reactions between perovskite and ETL/HTL [5, 38,39,40], electrodes or the environment issues (e.g., moisture, oxygen, light) [41,42,43,44,45,46].

The chemical reactions of lead halide perovskite have significant impacts on interfacial defects [47, 48], charge transport/extraction [49], and thus photovoltaic performance and device stability of PSCs [50,51,52,53,54,55]. Because of the relatively active chemical properties, perovskites may react with oxygen [56,57,58,59], water [60,61,62,63], Lewis acids and bases [64, 65] and some metals [66,67,68,69,70], at perovskite/charge transport layer interfaces or grain boundaries (GBs). In general, these reactions fall into two categories: 1) the reactions with species in ambient environment, including water, oxygen, and light; 2) the reactions with other materials in the device, including HTL/ETL, metal electrodes, and alternative modifiers. In order to suppress the degradation reactions of perovskites, it’s necessary to reduce the reactivity of perovskites through doping additives or compositional modifications [65, 71,72,73,74,75,76,77]. Despite significant efforts in additive engineering, the stability of PSCs still can’t meet the requirements for commercialization [78,79,80,81]. Physical isolation, another effective strategy for improving device stability, also attracts enormous attention simultaneously [82,83,84,85,86,87,88,89]. Some specific materials can serve as buffer layers to reduce the detrimental interfacial reactions. For example, the modified ZnO isolation layer with good electrical conductivity between the Ag electrode and ETL can effectively suppress the reaction between perovskite and Ag [90]. The hydrophobic diketopyrrolopyrrole-based polymers are deposited on perovskites surfaces to prevent the perovskites from being invaded by water [91]. In conclusion, understanding the chemical reactions at the interfaces can help researchers find feasible approaches to prevent the degradation of lead halide perovskites and improve the stability of PSCs.

In this review, we describe a variety of interfacial chemical reactions under different conditions, and the reactions mechanisms are also summarized. We retrospectively examined the established understanding of how the interfacial reactions affect the defects/traps formation, non-radiative recombination, ion migrations, and eventually device stability. A relatively clear relationship between device stability and interfacial chemical reactions is revealed in this part. According to different chemical reactions mechanism, we also review some available strategies for improving the chemical stability of perovskites. Finally, potential suggestions on reducing or avoiding the detrimental interfacial reactions in device fabrication and commercialization are proposed.

2 Interfacial Chemical Reactions

2.1 Environmental Factors

Although encapsulated lead halide PSCs can insulate air, the challenge of achieving an ideal encapsulation effect makes it necessary to explore how environmental factors affect the chemical stability of PSCs. Both pure MAPbI3 and FAPbI3 are very sensitive to air due to their intrinsic instability, which can be demonstrated by their color changes from black to yellow in a few minutes, especially in high-humidity environments. Studies have shown that water and oxygen in the air can react with perovskites through different pathways. In addition, ambient light can induce the decomposition of perovskites, which will be discussed in this part.

2.1.1 Water-Induced Reactions

Many studies have revealed that the reactions of lead halide perovskites with H2O can accelerate the degradation process and deteriorate the chemical stability of PSCs [60, 61, 92]. In order to express the reaction mechanism more concisely, all chemical reactions of perovskite in this review are based on the archetypal MAPbI3 unless otherwise stated. Walsh et al. proposed the simple acid–base reversible reactions between MAPbI3 and H2O as follows [48], and a plausible decomposition pathway for MAPbI3 is shown in Fig. 2a.

Fig. 2
figure 2

a Possible decomposition pathway of hybrid halide perovskites in the presence of water. A water molecule, 1, is required to initiate the process with the decomposition being driven by the phase changes of both hydrogen iodide, (2, soluble in water) and the methylammonia (3, volatile and soluble in water). This pathway results in the formation of a yellow solid, which corresponds to the experimentally observed PbI2, 4. Reproduced with permission from Ref. [48]. Copyright 2014, American Chemical Society. b Photographs of CH3NH3PbI3 films deposited on FTO and stored under ambient conditions for several days. c Evolutions of photoelectron spectra of I 3d5/2, Pb 4f, C 1s and O 1s. b, c Reproduced with permission from Ref. [97]. Copyright 2016, American Chemical Society. (Color figure online)

Full size image
$$ \left[ {(CH_{3} NH_{3} )PbI_{3} } \right]_{n} + H_{2} O{ \leftrightharpoons }\left[ {(CH_{3} NH_{3} )_{{(n - 1)}} (PbI_{3} )_{n} } \right][H_{3} O^{ + } ] + CH_{3} NH_{2} \uparrow $$
(2.1-1)
$$ \left[ {(CH_{3} NH_{3} )_{{(n - 1)}} (PbI_{3} )_{n} } \right][H_{3} O^{ + } ] \rightleftharpoons HI \uparrow + PbI_{2} + \left[ {(CH_{3} NH_{3} )PbI_{3} } \right]_{{(n - 1)}} + H_{2} O $$
(2.1-2)

By combining Eqs. 2.1-1 and 2.1-2, the whole degradation process of MAPbI3 in the presence of H2O can be generalized as follows:

$$ \left( {CH_{3} NH_{3} } \right)PbI_{3} \xrightarrow{{H_{2} O}}HI \uparrow + PbI_{2} + CH_{3} NH_{2} \uparrow $$
(2.1-3)

CH3NH2 and HI both exist in gas phases at room temperature, and thus the continuous release of gas products promotes the reaction to proceed in the forward direction. In consequence, MAPbI3 will completely degrade into PbI2 once the open system contains water [93].

Wang et al. further proposed that the degradation product HI can continue to decompose into H2 and I2 under the stimulation of ultraviolet (UV) light [94]. The photoreaction process of HI under UV irradiation is illustrated below:

$$ 2HI\xrightarrow{{UV}}H_{2} \uparrow + I_{2} $$
(2.1-4)

The whole degradation reaction of MAPbI3 when H2O and UV light coexist is generalized in Eq. 2.1-5 with the combination of Eqs. 2.1-3 and 2.1-4. Therefore, the final solid products of MAPbI3 decomposition are PbI2 and I2.

$$ \left( {CH_{3} NH_{3} } \right)PbI_{3} \xrightarrow{{H_{2} O + UV}}\frac{1}{2}H_{2} \uparrow + PbI_{2} + CH_{3} NH_{2} \uparrow + \frac{1}{2}I_{2} $$
(2.1-5)

Considering that lead halide perovskites are prone to absorb H2O molecule in the air to form a hydrated complex [50, 51, 63], the hydration processes of MAPbI3 are displayed in Eqs. 2.1-6 and 2.1-7, respectively. Notably, the decoloring process of black perovskites induced by the generation of monohydrated phase CH3NH3PbI3·H2O is reversible [95], and that can be reconverted into MAPbI3 again through dehydration process [63]. However, the degradation reactions become irreversible once the dihydrate (CH3NH3)4PbI6·2H2O forms (Eq. 2.1-7). Here, in the hydration process, the [PbI6]4− in the 3D network of MAPbI3 decays to a 0D framework of isolated octahedral. Ptasinska et al. pointed out that a transient phase PbI2+xx− (0 ≤ x < 1) forms during the evolution of a hydrated complex under ambient conditions [96]. The resulting PbI2+xx− is reactive when exposed to air and decomposes into lead-containing compounds inducing amorphous PbO, Pb(OH)2, and PbCO3 (Eqs. 2.1-9, 2.1-10, 2.1-11), which is evidenced by X-ray photoelectron spectra (XPS) characterizations in Fig. 2c.

$$\left({CH}_{3}{NH}_{3}\right){PbI}_{3}+{H}_{2}O\rightleftharpoons \left({CH}_{3}{NH}_{3}\right){PbI}_{3}\cdot {H}_{2}O$$
(2.1-6)
$$4\left({CH}_{3}{NH}_{3}\right){PbI}_{3}+{8H}_{2}O\to {{\left({CH}_{3}{NH}_{3}\right)}_{4}PbI}_{6}\cdot {2H}_{2}O+3Pb{\left(OH\right)}_{2}+6HI\uparrow $$
(2.1-7)
$${{\left({CH}_{3}{NH}_{3}\right)}_{4}PbI}_{6}\cdot {2H}_{2}O\to {{\left({CH}_{3}{NH}_{3}\right)}_{x}PbI}_{2+x}+\left(4-x\right){CH}_{3}{NH}_{2}\uparrow +\left(4-x\right)HI\uparrow +{2H}_{2}O$$
(2.1-8)
$$ 2\left( {CH_{3} NH_{3} } \right)_{x} PbI_{{2 + x}} + 2CO_{2} + O_{2} \to 2PbCO_{3} + 2I_{2} + 2xCH_{3} NH_{2} \uparrow + 2xHI \uparrow $$
(2.1-9)
$$ \left( {CH_{3} NH_{3} } \right)_{x} PbI_{{2 + x}} + H_{2} O + \frac{1}{2}O_{2} \to Pb\left( {OH} \right)_{2} + I_{2} + xCH_{3} NH_{2} \uparrow + xHI \uparrow $$
(2.1-10)
$$Pb{\left(OH\right)}_{2}\to PbO+{H}_{2}O$$
(2.1-11)

It’s reported that trapped charges can facilitate the degradation reactions of MAPbI3 in humid conditions. Ahn et al. deposited different polarity ions on the perovskite surfaces in humidified nitrogen and discovered that MAPbI3 is irreversibly decomposed to yellow PbI2 only when moisture and charges coexist (Fig. 3a–d) [97]. Figure 3e–l demonstrate that the degradation of perovskite starts from the GBs and the striking resemblance in kelvin probe force microscopy (KPFM) measurements (Fig. 3m and n) indicates that charges are preferentially trapped along GBs, which demonstrates that trapped charges can give rise to irreversible degradation. The degradation mechanism can be summarized: first, the perovskite material undergoes a hydration reaction in humid environment. Next, the organic cations, like MA+, will be deprotonated with the help of the local electric field induced by the charges trapped at the defect sites. The deprotonation process in the presence of water is shown below:

Fig. 3
figure 3

a–d Schematic illustration of perovskite degradation processes (left), e–h top-view (middle) and i–l cross-sectional (right) SEM images of perovskite layers (a, e, i) before, (b, f, j) after 6 h, (c, g, k) 12 h and (d, h, l) 18 h by ion deposition in humidified nitrogen. The color change from dark brown to yellow in a-d represents the gradual degradation process. Black lines and their widths in a–d represent grain boundaries and degradation extent, respectively. Scale bars, 200 nm. m Topography and n surface potential profile of MA0.6FA0.4PbI2.9Br0.1 film obtained from KPFM measurements after deposition of N2-positive ions. Reproduced with permission from Ref. [98]. Copyright 2016, Springer Nature. (Color figure online)

Full size image
$$ CH_{3} NH_{3}^{ + } + H_{2} O\xrightarrow{{TC}}CH_{3} NH_{2} \uparrow + H_{3} O^{ + } $$
(2.1-12)

Here TC represents trapped charges. The release of gas phase CH3NH2 will shift the following hydration equilibrium reaction to the right side, causing the perovskite to start irreversible decomposition:

$$ \left( {CH_{3} NH_{3} } \right)PbI_{3} + H_{2} O\xrightarrow{{TC}}PbI_{2} + CH_{3} NH_{2} \uparrow + H_{3} O^{ + } + I^{ - } $$
(2.1-13)

In addition, oxygen can also accelerate the aging of perovskite in humid conditions, which is attributed to the scavenging action of O2 on the H3O+ proton formed in the aforementioned deprotonation process (Eq. 2.1-12). The overall degradation reaction of MAPbI3 with the participation of O2 is shown as follows:

$$ \left( {CH_{3} NH_{3} } \right)PbI_{3} + \frac{1}{4}O_{2} \xrightarrow{{H_{2} O + TC}}PbI_{2} + CH_{3} NH_{2} \uparrow + \frac{1}{2}H_{2} O + \frac{1}{2}I_{2} $$
(2.1-14)

In summary, lead halide perovskite can easily decompose into PbI2, CH3NH2, HI, and other products in humid conditions with/without other factors (e.g. UV light, trapped charges, oxygen). The degradation reaction of MAPbI3 is usually irreversible due to the formed gas products (e.g. CH3NH2, HI) will release into the air. Therefore, protecting perovskites from water is crucial to improve the stability of PSCs.

2.1.2 Oxygen-Induced Reactions

Oxygen can induce the degradation of lead halide perovskites under certain conditions [98]. A study found that O2 molecules are only physically attached to the perovskites surfaces without chemical reactions. Once the adsorbed O2 captures an excess electron to form superoxide (O2·−), the degradation reactions of perovskites will proceed [99]. Haque et al. reported that O2 could permeate into the grain surface and interior of MAPbI3, which could also be reduced to highly reactive O2·− with the help of photo-excited electrons [57, 100]. It’s observed that the Al2O3/MAPbI3 system can produce more O2·− than TiO2/MAPbI3 system. In Fig. 4a, the TiO2 film can accept an electron from the photo-excited MAPbI3 because of the favorable energy offset at the heterojunction. Therefore, the fewer electrons transferring to oxygen leads to a lower yield of O2·− for the TiO2/MAPbI3 system. Simultaneously, ab initio simulations demonstrate the O2·− prefer energetically to occupy the iodide vacancies sites [101]. The photo-induced O2·− formation is the key factor for the degradation reactions of MAPbI3:

Fig. 4
figure 4

a Schematic model showing the electron transfer of the photoexcited electrons in the MAPbI3 layers to oxygen resulting in the formation of superoxide. Reproduced with permission from Ref. [101]. Copyright 2015, Wiley-VCH. b Photovoltaic performance characteristics of p-i-n PSC treated by H2O2 via the gas-phase deposition method with urea hydrogen peroxide for 40 s, compared to a control device, measured under AM1.5 100 mW/cm2 simulated sunlight. Reproduced with permission from Ref. [103]. Copyright 2019, Elsevier. c Schematic representation of the photo-oxidative degradation process of the MAPbI3 (001) surface. Reproduced with permission from Ref. [108]. Copyright 2019, Royal Society of Chemistry

Full size image
$$ \left( {CH_{3} NH_{3} } \right)PbI_{3} \xrightarrow{{light}}\left( {CH_{3} NH_{3} } \right)PbI_{3}^{*} $$
(2.1-15)
$$ O_{2} \xrightarrow{{\left( {CH_{3} NH_{3} } \right)PbI_{3}^{*} }}O_{2}^{{ \cdot - }} $$
(2.1-16)
$$ 4(CH_{3} NH_{3} )PbI_{3}^{*} + O_{2}^{{( \cdot \, - )}} \to 4PbI_{2} + 2I_{2} + 2H_{2} O + 4CH_{3} NH_{2} \uparrow $$
(2.1-17)

CH3NH3PbI3* carries both photo-induced electrons and holes (Eq. 2.1-15), and O2 captures an electron from CH3NH3PbI3* to form O2·− (Eq. 2.1-16). The study found that MAPbI3 films with large grains degrade more slowly, which is attributed to large crystallites containing the fewer surface reaction sites. Based on the studies above, Sultana et al. discovered that the degradation product PbI2 (Eq. 2.1-17) could further react with O2 to produce lead oxyiodide under UV irradiation [38], which can be expressed as follows:

$$ PbI_{2} \xrightarrow{{UV}}e_{{cb}}^{ - } \left( {PbI_{2} } \right) + h_{{vb}}^{ + } \left( {PbI_{2} } \right) $$
(2.1-18)
$${e}_{cb}^{-}\left(Pb{I}_{2}\right)+{O}_{2}\to {O}_{2}^{\bullet -}+Pb{I}_{2}$$
(2.1-19)
$$\left(x+y\right)Pb{I}_{2}+\frac{y}{2}{O}_{2}^{\bullet -}\to {(Pb{I}_{2})}_{x}{(PbO)}_{y}+y{I}_{2}$$
(2.1-20)

The electron is excited from the valence band of PbI2 to its conduction band under UV irradiation (Eq. 2.1-18), which then transfers to O2 molecular to generate O2·− (Eq. 2.1-19). Finally, the highly reactive O2.− further reacts with PbI2 generating lead oxyiodide.

It’s reported that O2.− will capture an acid proton of the CH3NH3+ to generate hydroperoxyl radical (HO2·) once approaching the ammonium group (Eq. 2.1-21) [102, 103]. Subsequently, two different reaction mechanisms are proposed to express next degradation process of perovskite. The first one is that two HO2· interact to produce hydrogen peroxide (H2O2) under lights (Eq. 2.1-22). Ultimately, H2O2 reacts with Pb0 on the perovskite surface generating PbO or Pb(OH)2 (Eqs. 2.1-23, 2.1-24). It is worth noting that the formed PbO has an unexpected passivation effect, preventing the accumulation of anion vacancies and formation of Pb-Pb dimers, which can increase the open-circuit voltage (Voc) for inverted PSCs as exhibited in Fig. 4b. The second one demonstrates that HO2· radicals can dissociate into H2 and O2, and the remaining O2 can serve as a reactant again for the continuous degradation process (Eq. 2.1-25).

$$ CH_{3} NH_{3}^{ + } + O_{2}^{{ \bullet \, - }} \leftrightarrow CH_{3} NH_{2} \uparrow + HO_{2} \cdot $$
(2.1-21)
$$ 2HO_{2} \cdot \leftrightarrow O_{2} \uparrow + H_{2} O_{2} $$
(2.1-22)
$${P{b}^{0}+H}_{2}{O}_{2}\to Pb{(OH)}_{2}$$
(2.1-23)
$${P{b}^{0}+H}_{2}{O}_{2}\to PbO+{H}_{2}O$$
(2.1-24)
$$ 2HO_{2} \cdot \leftrightarrow 2O_{2} \uparrow + H_{2} \uparrow $$
(2.1-25)

Hillhouse et al. reported that H2O can accelerate the photooxidation reaction of MAPbI3 and proposed one plausible degradation pathway [104]. H2O rapidly undergoes the deprotonation reaction to form HO2· and HO− in the presence of O2·− (Eq. 2.1-26). As mentioned above, the newly formed HO2· will interact to release H2 and O2 (Eq. 2.1-25). Unfortunately, HO− triggers the decomposition of MAPbI3 into PbI2 (Eq. 2.1-27), which subsequently reacts with H2O to generate PbIOH and HI (Eq. 2.1-28) in the following steps:

$$ O_{2}^{{( \bullet - )}} + H_{2} O \leftrightarrow HO^{ - } + HO_{2} \cdot $$
(2.1-26)
$$\left({CH}_{3}{NH}_{3}\right){PbI}_{3}+H{O}^{-}\to {PbI}_{2}+{H}_{2}O+{CH}_{3}{NH}_{2}\uparrow +{I}^{-}$$
(2.1-27)
$$Pb{I}_{2}+{H}_{2}O\to PbIOH+HI\uparrow $$
(2.1-28)

It’s worth noting that O2·− more rapidly reacts with H2O (Eq. 2.1-26) than CH3NH3+ (Eq. 2.1-21) due to a lower activation energy for the whole reaction pathway, which demonstrates that H2O can accelerate photooxidative degradation of MAPbI3.

Except for the degradation products mentioned above like PbI2 and lead oxyiodide, Snaith et al. detected metal lead (Pb0) on the perovskite surface [105] and proposed a plausible formation mechanism of Pb0 [102]. An iodide ion (I−) abstracts a photogenerated hole (h+) to produce an iodine atom (I·) (Eq. 2.1-29), which is accompanied by rapid site exchange of iodide from a regular to interstitial lattice site. Two iodine atoms combine with each other to generate I2, simultaneously leaving two iodine vacancies (VI·) (Eq. 2.1-30). Subsequently, VI· captures an electron to generate Farbe center (VI·e’) (Eq. 2.1-31), which then reduces the Pb2+ in the adjacent site to Pb+ (Eq. 2.1-32). Finally, Pb+ undergoes a disproportionation reaction to produce Pb0 (Eq. 2.1-33).

$${I}^{-}+{h}^{+}\to {I}^{\bullet }$$
(2.1-29)
$$2{I}^{\bullet }\to {I}_{2}+2{V}_{{I}^{\bullet }}$$
(2.1-30)
$${V}_{{I}^{\bullet }}+{e}^{^{\prime}}\to {V}_{{I}^{\bullet }}{e}^{^{\prime}}$$
(2.1-31)
$${{Pb}^{2+}+V}_{{I}^{\bullet }}{e}^{^{\prime}}\to {V}_{{I}^{\bullet }}{Pb}^{+}$$
(2.1-32)
$$2{Pb}^{+}\to {Pb}^{2+}+{Pb}^{0}$$
(2.1-33)

Based on the above research, a photo-oxidative degradation mechanism of MAPbI3 is proposed for rapid surface oxidation and slow inner hydration [101, 106]. Wang et al. reported the three-step degradation process of MAPbI3 as displayed in Fig. 4c [107]. Step I: O2 on the MAPbI3 surface capture a photogenerated electron from the conduction band of MAPbI3 to form O2− (Eq. 2.1-16). Step II: O2− rapidly oxidizes the PbI2-terminated surface to produce PbO, resulting in a breakage of the Pb–I frameworks and the exposure of the MAI-terminated surface. Subsequently, the underlying MAI-terminated surface is further oxidized to generate H2O and PbO or the unstable Pb(OH)2. Step III: The oxidation products PbO and Pb(OH)2 can act as protective layers to inhibit the further oxidation of the internal MAPbI3. The newly produced H2O molecules from the surface oxidation and decomposition of Pb(OH)2 can slowly hydrate with the internal MAPbI3, finally leading to the structural collapse of MAPbI3.

In conclusion, the oxygen-induced reactions of perovskites need the participation of light or water. Firstly, O2 captures photogenerated electrons generating O2−, which subsequently introduces a series of decomposition pathways under different conditions. Finally, the generating degradation products mainly include CH3NH2, I2, PbI2, and PbO (or Pb(OH)2).

2.1.3 Light-Induced Reactions

Unlike oxygen, light can induce the perovskites degradation without any presence of other environmental factors (e.g., water, oxygen). Study demonstrated that UV light can induce severe degradation of lead halide perovskites [108, 109]. As shown in Fig. 5a, Gao et al. detected Pb0 on the MAPbI3 surface after about 120 min of UV irradiation with a wavelength of 408 nm [110]. The proposed degradation reactions of MAPbI3 under UV irradiation can be generalized in Eqs. 2.1-34, 2.1-35. Figure 5b shows that the ratio of Pb0 remains nearly constant after 480 min of UV irradiation, which indicates that the decomposition of MAPbI3 has already saturated.

Fig. 5
figure 5

a Pb 4f7/2 decomposition and b metallic Pb fraction during laser irradiation. a, b Reproduced with permission from Ref. [111]. Copyright 2017, American Chemical Society. c Schematic representation of photodegradation mechanism. 1) Hot carrier generation, 2) deprotonation and release of MA gas, and 3) formation of HI gas and VI. Reproduced with permission from Ref. [112]. Copyright 2018, Royal Society of Chemistry. d, e, Photoluminescence enhancement and quenching mechanisms. Ion dynamics in MAPbI3 thin film promoting PLIE in d, when the probability of I0 species encounters is small and Frenkel pair annihilation is boosted by electron trapping, and PLID in e, when the probability of I0 species encounters is high, boosting I2 molecule formation. d, e Reproduced with permission from Ref. [113]. Copyright 2019, Springer Nature

Full size image
$$ \left( {CH_{3} NH_{3} } \right)PbI_{3} \xrightarrow{{UV}}PbI_{2} + HI \uparrow + CH_{3} NH_{2} \uparrow $$
(2.1-34)
$$ PbI_{2} \xrightarrow{{UV}}Pb^{0} + I_{2} $$
(2.1-35)

Yan et al. proposed a possible photodegradation mechanism of MAPbI3 concerning hot carriers [111]. As shown in Fig. 5c, the first step is that MAPbI3 is excited by light (hv > 3 eV) to generate long-lived hot carriers (Eq. 2.1-36). Then a hot electron reacts with the CH3NH3+ via columbic coupling to generate a free proton and release CH3NH2 near surface or GB regions (Eq. 2.1-37). Finally, the free protons interact with undercoordinated I− on the surface, generating volatile HI and iodine vacancies (Eq. 2.1-38). The release of HI can induce the destruction of Pb-I framework from the corner-shared to the face-shared Pb-I octahedral, causing the degradation of MAPbI3 into PbI2.

$$hv\left(>3\mathrm{eV}\right)\to {e}^{-}\left(hot\right)+{h}^{+}(hot)$$
(2.1-36)
$$ CH_{3} NH_{3}^{ + } + e^{ - } (hot) \to CH_{3} NH_{2} + H^{ + } + e^{ - } $$
(2.1-37)
$${I}^{-}+{H}^{+}\to HI\uparrow +{V}_{I}$$
(2.1-38)

It’s reported that halide defects have an influence on the photostability of MAPbI3 [112]. Petrozza et al. found that long-living carrier traps associated with halide defects could trigger photoconversion, which drives both photoluminescence intensity enhancement (PLIE) and photoluminescence intensity decrease (PLID). In Fig. 5d and e, PLIE and PLID processes are proposed to explain the observations. The timescales involved in PLID and PLIE are consistent with the reported ionic activities, such as ion/defect annihilation and migration rates [113, 114]. Generally, vacancies (e.g., VPb) and interstitial halogen (e.g., Ii) defects dominate in the perovskites. VPb is only moderately active as a trap, while Ii can remarkably trap both electrons and holes via (+ /0) and (0/ −) transitions. When the probability of encountering I0 species is very low, the light-induced PLIE is associated with annihilation of the Ii−VI+ Frenkel pair:

$$ I_{i}^{ + } /I_{i}^{ - } ....V_{I}^{ + } + e^{ - } \xrightarrow{{light}}I_{i}^{0} /I_{i}^{ - } ....V_{I}^{ + } \xrightarrow{{light}}I_{i}^{0} + pristine\;material $$
(2.1-39)

The PLID mechanism is proposed to be a bimolecular reaction boosted by increasing the encountering probability of I0 species, occurring near the film surface filled with long-lived traps:

$$2{I}_{i}^{0}\stackrel{light}{\to }{I}_{2}$$
(2.1-40)

The two processes may coexist and compete in the perovskite material. If traps densities are adequately low, photo-induced PLIE will be a dominant effect. However, photo-induced PLID will play a key role when high-density traps appear near the film surface that may act as a reservoir for photogenerated species, eventually causing perovskites degradation. Therefore, passivating under-coordinated surface sites can prohibit defect formation and hence enhance the photostability of perovskites.

2.2 Charge Transport Layers

In addition to the aforementioned degradation reactions induced by environmental factors, the chemical stability of lead halide PSCs can be affected by charge transport layers, including ETL and HTL. Generally, the interfacial chemical reactions are not desirable, which could lead to destroyed structure and reduced charge transport/extraction. However, for certain charge transport layers such as Cl-containing SnO2, interfacial reactions have been demonstrated to be beneficial for charge transport and chemical stability of PSCs. Considering the diversity of charge transport materials, a variety of interfacial reactions and their effects on device stability have been reported in literature. Therefore, the chemical reactions in PSCs induced by charge transport layers are elucidated in this session.

2.2.1 ETL-Induced Reactions

TiO2 has been widely used as an electron transport material in the early study of PSCs [36, 115,116,117,118,119,120,121]. Belmonte et al. reported that TiO2 interacts with MAPbI3 mainly through binding I− in MAPbI3 to undercoordinated Ti4+ in TiO2 [49, 122]. Notably, this Ti-I-Pb bond is not strong due to the little hybridization between Pbs–p state and Tid orbital, hence the adsorption/desorption of I− at the interface may occur easily. The interactions between TiO2 and perovskite are reversible under both positive and negative biases, as shown in Fig. 6a–c. At a positive bias, weakly bonded I− migrate towards the hole transport layer contact and hence I vacancies remain at MAPbI3/TiO2 interface, and the positive charges are compensated by electrons injection and accumulations at TiO2. In contrast, excess I− may accumulate at MAPbI3/TiO2 interface under a negative bias. The Ti-I-Pb bonds easily accommodate excessive defect or ion charges in a highly reversible manner to generate capacitive currents. However, it's worth noting that this reversible interaction has a negligible effect on the photovoltaic performance and chemical stability of PSCs.

Fig. 6
figure 6

Diagram representing iodide migration and chemical species present at the interfaces. a At positive-bias iodine ions are forced to migrate toward the hole selective contact where the reaction with spiro-OMeTAD+ occurs. The iodide defective layer is formed at the TiO2/MAPbI3 interface. b At zero-bias the neutral case appears. c At negative-bias spiro-OMeTAD only partially returns to its oxidized, conductive state. Iodide ions accumulate at the TiO2/perovskite interface. a-c Reproduced with permission from Ref. [49]. Copyright 2016, Wiley-VCH. d Scheme showing the reversible chemistry reactions process occurring at the solid-state TiO2/CH3NH3PbI3 interface. Reproduced with permission from Ref. [124]. Copyright 2017, American Chemical Society. e Degradation scheme of CH3NH3PbI3 perovskite solar cells during light exposure test: TiO2/CH3NH3PbI3. Reproduced with permission from Ref. [125]. Copyright 2014, American Chemical Society

Full size image

Studies on film’s current response to I2 vapor have revealed that reversible chemical reactions occur at TiO2/I2 and MAPbI3/I2 interfaces [123]. Rand et al. discovered the pathways for I2 passivation of MAPbI3 surface defects from the photoconductivity data of MAPbI3 films. According to the current response of TiO2 film to I2 vapor pressures, they proposed a plausible reaction mechanism between TiO2 and I2. As shown in Eqs. 2.2-1, 2.2-2, and 2.2-3, TiO2 and MAPbI3 surfaces undergo strong reversible reactions with I2 gas, respectively. Combining these three reactions, the chemical interaction between TiO2 and MAPbI3 can be generalized in Eq. 2.2-4.

$${e}_{MAPb{I}_{3}}^{-}+{trap}_{MAPb{I}_{3}}^{0}\rightleftharpoons {trap}^{-}$$
(2.2-1)
$${trap}^{-}+\frac{1}{2}{I}_{2}\rightleftharpoons {I}_{adsorbed(MAPb{I}_{3})}^{-}$$
(2.2-2)
$${e}_{Ti{O}_{2}}^{-}+\frac{1}{2}{I}_{2}\rightleftharpoons {I}_{adsorbed(Ti{O}_{2})}^{-}$$
(2.2-3)
$${e}_{Ti{O}_{2}}^{-}+{I}_{adsorbed(MAPb{I}_{3})}^{-}\rightleftharpoons {I}_{adsorbed\left(Ti{O}_{2}\right)}^{-}+{e}_{MAPb{I}_{3}}^{-}+{trap}_{MAPb{I}_{3}}^{0}$$
(2.2-4)

I− in MAPbI3 is adsorbed on the TiO2 surface and deplete an electron on TiO2, leaving a free electron as well as a trap on the MAPbI3 surface (Fig. 6d). The energy level of the trap would determine whether the electron is captured or freely transmitted. Therefore, MAPbI3 may be chemically reduced by TiO2 due to the differences in work function. In addition, the newly formed traps on the MAPbI3 surface will make a difference to carrier density and recombination rate, which could deteriorate the device stability.

TiO2 was also reported to catalyze MAPbI3 decomposition in the presence of light [124]. In Fig. 6e, TiO2 extracts an electron from I− under light irradiation generating I2 (Eq. 2.2-5), which deconstructs perovskites crystal and reduces their chemical stability. As mentioned in 2.2.1 part, CH3NH3+ can be deprotonated to generate CH3NH2 and H+ in humid conditions. The presence of I2 and H+ will accelerate the reaction (Eq. 2.2-6) to proceed forward, further destroying the perovskite structure.

$$ 2I^{ - } \xrightarrow{{TiO_{2} + light}}I_{2} + 2e_{{TiO_{2} }}^{ - } $$
(2.2-5)
$${{I}^{-}+I}_{2}+3{H}^{+}+2{e}_{Ti{O}_{2}}^{-}\rightleftharpoons 3HI\uparrow $$
(2.2-6)

TiO2 acts as a catalyzer rather than a reagent in the reactions with perovskites, which accelerates the degradation process of perovskites to some extent. Therefore, reducing the catalytic performance of TiO2 is an effective way to improve the photovoltaic performance and chemical stability of TiO2-based PSCs.

It’s reported that ZnO ETL can also interact chemically with perovskites [125,126,127,128,129]. Anta et al. proposed that an acid–base reaction could occur at the ZnO/perovskite interface [130], triggering the decomposition of perovskite into PbI2 in humid conditions. In addition, this interaction could cause redissolution of the ZnO substrate, whose morphology changes from spherical nanoparticles to aciculate particles (Fig. 7a).

Fig. 7
figure 7

a SEM of hZnO and Al2O3 substrates before (left) and after (right) deposition of perovskite layer. Reproduced with permission from Ref. [131]. Copyright 2016, Royal Society of Chemistry. b FTIR spectra of ZnO films on glass annealed at various temperatures. c Photographs of CH3NH3PbI3 films deposited on thermally pretreated ZnO layers and heated to 100 °C for the indicated time: top row (left to right): pretreated at 25, 100, 200 °C. Bottom row (left to right): pretreated at 300 °C, 400 °C, and SiO2/CH3NH3PbI3. b, c Reproduced with permission from Ref. [132]. Copyright 2015, American Chemical Society. d High-resolution of O 1s XPS core level spectra of ZnO with annealing at 200 °C for 1 h in air. e XRD patterns and photographs of perovskite film on annealed ZnO without thermal annealing and annealing at 100 °C for 30 min. d, e Reproduced with permission from Ref. [133]. Copyright 2015, American Chemical Society. (Color figure online)

Full size image

Additionally, the deprotonation reaction between ZnO and perovskite can cause the thermal degradation of perovskite. Kelly et al. found that the basic hydroxyl groups and residual acetate ligands exist on the ZnO surface through the Fourier transform infrared spectroscopy (FTIR) spectra analysis (Fig. 7b), which can be eliminated or reduced by high-temperature calcination of ZnO films (Fig. 7c) [131]. Otherwise, the hydroxide or residual acetate ligands will react with CH3NH3+ destroying the crystal structure of perovskite, which is evidenced in Fig. 7d and e [132]. And this decomposition process of MAPbI3 can be expressed as follows:

$${OH}^{-}{+CH}_{3}{NH}_{3}^{+}\to {CH}_{3}{NH}_{3}OH$$
(2.2-7)
$$ CH_{3} NH_{3} OH\xrightarrow{{heat}}CH_{3} NH_{2} \uparrow + H_{2} O $$
(2.2-8)

As shown in Eqs. 2.2-7 and 2.2-8, the whole degradation reactions are the deprotonation process of CH3NH3+, and CH3NH3OH easily decomposes into CH3NH2 gas and H2O under heat. The decomposition of CH3NH3OH can promote the reaction (Eq. 2.2-7) to proceed forward continuously, which accelerates MAPbI3 degradation. It’s reported that MAPbI3 decomposes easily to form HI in humid conditions [94], which subsequently reacts with ZnO resulting in reduced electron mobility of ZnO and deteriorative chemical stability of PSCs. This neutralization reaction can be presented in the following reaction [127]:

$$2HI+ZnO\to Zn{I}_{2}+{H}_{2}O$$
(49)

SnO2 is widely used in PSCs owing to its good electron transport ability and chemical stability. However, the interactions between the Cl-containing SnO2 and perovskite still exist. Recently, Seok et al. reported that a FASnClx interlayer at a SnO2/perovskite interface could be formed by an interfacial reaction between Cl-bonded SnO2 and Cl-containing FAPbI3 perovskite (Fig. 8a and b) [5], which is related to the easy formation of Sn-based perovskites in the presence of Cl−, FA+ and Sn2+ [133, 134]. The coherent interlayer reduces the interfacial charge recombination and enhances charge transport/extraction, achieving stable PSCs with a high PCE of 25.8% (Fig. 8c and d). However, Sn-Cl bonds did not form when the Cl-containing FAPbI3 solution was applied on a pure SnO2 surface. In comparison, Cl-bonded SnO2 can interact with Cl−-free FAPbI3 precursor to form Sn-I bonds by Cl−–I− exchange. In addition, Pang et al. also discovered the spontaneous ion-exchange reaction between Cl− and I− at the SnOx-Cl/MAPbI3 interface (Fig. 8e and f), which could effectively passivate the physical contact defects. The diffusion of Cl− in the MAPbI3 films promoted the grain longitudinal growth and decreased the GB density [135]. It is worth noting that the reactions between SnOx–Cl and perovskite effectively passivate the interface defects, thereby improving the photovoltaic performance and chemical stability of the PSCs.

Fig. 8
figure 8

a ToF-SIMS depth profiles for the perovskite and Cl-bonded SnO2 on FTO. b Simulation of the formation of the FASnClx interlayer between perovskite and SnO2. c J-V curves of the best-performing device, measured in reverse (red solid line) and forward (blue dashed line) modes. d Maximum power point tracking measured for the PSC fabricated using Cl-bonded SnO2 and Cl-containing FAPbI3. a-d Reproduced with permission from Ref. [5]. Copyright 2021, Springer Nature. e UV-vis spectra of SnOx-Cl, MAI, SnOx-Cl+MAI, and SnI4, respectively. f XRD pattern of SnOx-Cl + MAI and reference MA2SnI6 powder samples, providing the feasibility of the ion-exchange reaction. e, f Reproduced with permission from Ref. [136]. Copyright 2019, Wiley-VCH. (Color figure online)

Full size image

Fullerene derivatives, like phenyl-C61-butyric acid methyl ester (PCBM) and [6, 6]-Phenyl-C71-butyric acid methyl ester (PC71BM), are another type of popularly used electron transport materials particularly in inverted PSCs. It’s reported that PCBM undergoes an electron transfer reaction with halogens in perovskites to produce PCBM-nX radicals [74, 136]. Additionally, halogens favorably interact with C70 face rather than O face of PC71BM through covalent bonds [137]. Such strong interactions can suppress Pb-I antisite defects and ions migration at perovskite/PCBM interfaces, which can lead to improved stability and reduced hysteresis of PSCs [138]. However, C60, another commonly used ETL in the inverted PSCs, barely undergoes interfacial chemical reactions with perovskites because of its chemical inertness.

2.2.2 HTL-Induced Reactions

In general, the widely used HTLs in inverted PSCs (i.e., p-i-n structure) are mainly poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) [139, 140], poly[bis(4-phenyl)(2,4,6-trimethylphenyl)amine (PTAA) [141,142,143] and NiOx [144, 145]. As an inorganic HTL, NiOx is regarded as a promising HTL candidate with the advantages of good optical transparency, excellent stability and low cost [146, 147]. The interaction between NiOx and perovskite has been studied by many groups. McGehee et al. proposed a surface-assisted electron transfer-proton transfer (ET-PT) reaction mechanism at the NiOx/perovskite interface [148]. They deposited various perovskite precursors (MAI, MABr, MACl, CsI, and PbI2 in acetonitrile solvent, respectively) onto NiOx surface and observed that only MAI could bleach NiOx film, which demonstrates that Ni≥3+ sites can oxidize I− only in the presence of a proton donor but neither Br− or Cl− because of their much higher oxidation potentials. In conclusion, Ni≥3+ defect sites can act not only as Lewis acid to oxidize I− but also as Brønsted base to deprotonate CH3NH3+. The perovskite solution was deposited on the various HTLs (PTAA, poly-TPD, and NiOx) to detect the differences in the perovskite films through XRD analysis. It’s observed that PbI2 accumulates at the NiOx/perovskite interface or scatters throughout PTAA-based perovskite. However, only the interfacial PbI2 can block the extraction of holes resulting in the Voc loss of PSCs. The whole reaction can be shown in Fig. 9a.

Fig. 9
figure 9

a Schematic of the formation process of PbI2-xBrx at the interface. b Dark and light J-V curves of Cs0.25FA0.75Pb(Br0.2I0.8)3 perovskite solar cells with 0–3 mol% excess A-site. a, b Reproduced with permission from Ref. [150]. Copyright 2020, Elsevier. c Diagram of the passivation mechanism. d Fabrication procedure for perovskite solar cells by R2R process. e J-V curves of flexible devices (1.01 cm2) with and without HI treatment. Inset presents the photograph of flexible perovskite solar cells. f Normalized PCE of encapsulated PSCs based on NiOx and NiOx (HI) hole-transport layers under ambient condition (25–55% relative humidity and 25–35 ℃) for 30 days. c-f Reproduced with permission from Ref. [146]. Copyright 2021, Royal Society of Chemistry

Full size image
$$(C{H}_{3}{NH}_{3})Pb{I}_{3}+{Ni}^{\ge 3+}{O}_{x}\leftrightarrow {Ni}^{\ge 2+}{O}_{x}H+{PbI}_{2}+{CH}_{3}{NH}_{2}\uparrow +\frac{1}{2}{I}_{2}$$
(2.2-10)

According to reaction Eq. 2.2-10, the resulting A-site deficient in this region increases interface recombination and reduces chemical stability of PSCs. The study found that 1–5 mol% excess A-site cations were incorporated into the perovskite precursor solution to prevent the reaction above. Therefore, the formation of PbI2 layer at the NiOx/perovskite interface is inhabited, and the Voc was improved by > 200 mV as shown in Fig. 9b.

Wang et al. reported that trivalent nickel compound (NiOOH) on NiOx HTL surfaces can be reduced to nickel iodide (NiI2) by soaking the NiOx HTLs in hydroiodic acid (HI) during roll-to-roll printing of flexible PSCs [144], as shown in Fig. 9c and d, which can enhance NiOx/perovskite interface contact and ameliorate the work function of NiOx film. The reaction process is shown below:

$$NiOOH+3HI\to Ni{I}_{2}+2{H}_{2}O+{\frac{1}{2}I}_{2}$$
(2.2-11)
$$N{i}_{2}{O}_{3}+6HI\to 2Ni{I}_{2}+3{H}_{2}O+{I}_{2}$$
(2.2-12)

Subsequently, I− in NiI2 can coordinate with Pb in perovskite to form a Pb-I bond, inducing an orderly growth of perovskite lattice and enhancing the crystallinity of the perovskite film. Consequently, flexible PSCs with improved PCE as well as remarkable chemical stability were achieved (Fig. 9e and f).

In addition to NiOx, the commonly used HTLs are PEDOT:PSS and PTAA in inverted PSCs. We note that PEDOT:PSS and PTAA are rarely reported to react with perovskites. However, PEDOT:PSS has been shown to corrode ITO due to its acidic composition, which can react with In2O3 [13, 139, 149, 150]. As a consequence, the dissociated indium ions can diffuse into the perovskite layer and then deteriorate device performance and stability [151, 152].

The normal PSCs (i.e. n-i-p structure) using 2,2',7,7'-Tetrakis[N,N-di(4-methoxyphenyl)amino]-9,9'-spirobifluorene (spiro-OMeTAD) as HTL have achieved efficiency records, however, the instability of spiro-OMeTAD itself and its additives limits their commercial applications [153]. Therefore, the interfacial reactions between perovskite and spiro-OMeTAD under various conditions have been investigated to improve the device stability. Sultana et al. detected the signal of [spiro-OMeTAD(PbI2)nPbI]+ adducts in the mass spectra and found that the interaction between spiro-OMeTAD and perovskite was beneficial for PSCs operations [38]. In addition, Belmonte et al. reported that I− is driven towards the spiro-OMeTAD HTL and react with the oxidized spiro-OMeTAD+ under a positive bias (Eq. 2.2-13) [49]. This irreversible reaction turns the spiro-OMeTAD+ into neutral molecule, which prevents the oxidation of spiro-OMeTAD [154] and consequently decreases the conductivity of HTL. Hence, this process has a negative effect on the photovoltaic performance and chemical stability of PSCs.

$$spiro-OMeTA{D}^{+}+{I}^{-}\to spiro-OMeTAD-I$$
(2.2-13)

Besides, several common additives such as Lithium bis(trifluoromethanesulfonyl)-imide (Li-TFSI) and 4-tert-butylpyridine (TBP) in spiro-OMeTAD can influence the stability of perovskite active layers. Many studies have shown that the hygroscopic nature of Li-TFSI can accelerate the degradation of perovskite in humid conditions [155, 156]. As shown in Fig. 10a and b, TBP can react with PbI2 to form new complexes [PbI2 \(\bullet \) xTBP], which disintegrates the perovskite structure and deteriorates the device performance [157, 158]. The reaction can be expressed as follows:

Fig. 10
figure 10

a The UV-vis absorption of TBP liquid (dark line) and PbI2 solution in TBP (red line). b Pb 4f7/2 XPS spectra consuming of PbI2. a, b Reproduced with permission from Ref. [159]. Copyright 2014, Royal Society of Chemistry. c Crystal structural evolution from CH3NH3PbI3, CH3NH3PbI3·H2O and PbI2 to PbIOH. d FE-SEM images of top views and the cross-section views of relevant phases CH3NH3PbI3, PbI2, and PbIOH formed in lead halide perovskite solar cell at different stages. c, d Reproduced with permission from Ref. [161]. Copyright 2017, American Chemical Society

Full size image
$$ PbI_{2} + xTBP \to PbI_{2} \cdot xTBP $$
(2.2-14)

Chen et al. explored the decomposition of PSCs under operating conditions in the presence of light and H2O [159]. The crystal evolution processes are displayed in Fig. 10c. As mentioned in Sect. 2.1.1, MAPbI3 can react with H2O generating intermediate CH3NH3PbI3·H2O in low-humidity conditions [63, 95, 96], which has little effect on the device performance. However, CH3NH3PbI3·H2O can further decompose into PbI2 once exposed to high humidity, destroying the perovskite structure and reducing the chemical stability of PSCs. H2O would further react with axial I of [PbI6]4− to form [PbI4O2]6− under light irradiation or in the presence of spiro-MeOTAD. Given that the ionic radius of I− (2.06 Å) is significantly bigger than that of O2− (1.26 Å) [136], [PbI4O2]6− is prone to deform. These deformed [PbI4O2]6− octahedrons further connect with each other to become PbIOH with the coordination structure of [PbI5O3]9−. Figure 10d presents the scanning electron microscopy (SEM) images of PSCs at different degradation stages. It’s worth noting that in addition to H2O and light, spiro-MeOTAD is another key factor to stimulate the decomposition of MAPbI3 into PbIOH, indicating that spiro-OMeTAD has a crucial influence on the chemical stability of PSCs.

Because of the intrinsic thermal instability of organic HTLs in normal PSCs, a lot of research efforts are devoted to inorganic HTLs [160,161,162], such as CuSCN, which is a promising candidate due to its low cost and durability [163,164,165]. It’s reported that the CuSCN can catalyze the thermal degradation process of perovskite films even in the absence of moisture and oxygen [82, 166,167,168], although the thermal stability of CuSCN itself is excellent. CH3NH3I and CuSCN can react to form CH3NH3SCN and CuI as follows:

$$ \left( {CH_{3} NH_{3} } \right)PbI_{3} + CuSCN\xrightarrow{{heat}}CuI + PbI_{2} + CH_{3} NH_{3} SCN $$
(2.2-15)

2.3 Metal Electrodes

Lead halide perovskites not only react with the adjacent layers like charges transport layers, but also experience chemical interactions with the top metal electrodes through ions migrations. The common metal electrodes, such as aluminum (Al), silver (Ag) and gold (Au), can be corroded when in contact with hybrid perovskites or air. These metal ions can diffuse through charge transport layers into perovskites and meanwhile the halide species like I− can migrate to the metal electrode, which result in reactions between metal and halide ions. It has been reported that almost all reactions between metal electrode and perovskites cause severe performance and stability deterioration of PSCs. Therefore, exploring the reactions of the metal electrodes is helpful to have a better understanding of the chemical stability issues of PSCs.

2.3.1 Ag Electrode-Induced Reactions

Ag is prone to suffer from corrosion or contamination in contact with lead halide perovskite films. Studies found that I− in perovskites can react with spiro-OMeTAD+ at the interface, resulting in a large amount of I− accumulating in the spiro-OMeTAD layer [49, 169]. Strikingly, I− passes through the Spiro-OMeTAD layer and migrates to the Ag contact, meanwhile, Ag ions also diffuse into the perovskite layer from the top electrode, generating AgI in the PSCs [93, 170, 171].

Kato et al. reported that moisture can promote the production of AgI and a five-steps mechanism is proposed to understand the observations (Fig. 11a) [171], including (1) diffusion of H2O into perovskite through the pinholes in spiro-MeOTAD layer; (2) H2O-induced decomposition of MAPbI3 and production of volatile species containing I− (e.g., HI); (3) Migration of these volatile species to the bottom or top Ag layer; (4) Surface diffusion of the volatile species containing I−; (5) AgI formation. The chemical reaction of AgI formation can be expressed as follows:

Fig. 11
figure 11

a Schematic illustration of a proposed mechanism of AgI formation. Reproduced with permission from Ref. [174]. Copyright 2015, Wiley-VCH. ToF-SIMS elemental depth profiles b before and c after a thermal treatment at 85 °C for 24 h. d The depth profiles of Ag-, I- and CN- after different temperature of thermal treatment. b-d Reproduced with permission from Ref. [175]. Copyright 2017, Wiley-VCH. e The possible reaction mechanism of the PH1000-involved electrochemical corrosion of Ag electrode. Reproduced with permission from Ref. [41]. Copyright 2018, Wiley-VCH

Full size image
$$2HI+2Ag\to 2AgI+{H}_{2}\uparrow $$
(2.3-1)

In addition, thermal treatment can also accelerate the formation of AgI in inverted PSCs. In Fig. 11b–d, time-of-flight secondary ions mass spectroscopy (ToF–SIMS) tests revealed that thermal treatment triggered a significant accumulation of I−, I2−, and CN− ions at the PCBM/Ag interface, indicating two crucial facts: the decomposition of MAPbI3 and the diffusion of both I− and MA+ ions [172]. The formation of AgI accelerates the release of MA+ and I− ions from the GBs of the perovskite layer and reconstructs the grain domains, leading to more defects both in the perovskite films and at the interface. Consequently, the PSCs suffer from degraded performance due to the formation of AgI.

Ma et al. reported electrochemical corrosion of Ag grid electrodes by PEDOT:PSS in flexible PSCs, which is a major reason for the low performance and poor chemical stability of the devices [41]. This redox reaction involves the reduction of the highly conductive PEDOT:PSS layer and the oxidation of the Ag electrode, which is proposed to be three steps including (Fig. 11e): (1) Ag electrode easily loses an electron to form Ag+ with the aid of acidic PEDOT:PSS layer; (2) PEDOT:PSS layer receives an electron and is reduced; (3) Ag+ enters into the perovskite precursor solution and reacts with I− to generate AgI.

Therefore, regardless of the structure of PSCs, the corrosion of Ag electrode is usually ascribed to the chemical reactions between Ag and the components of perovskites, which is achieved through Ag+ migration or halide ions diffusion [173,174,175]. Eventually the formed AgI impedes the charge transport and reduces the photovoltaic performance and chemical stability of PSCs.

2.3.2 Au Electrode-Induced Reactions

Au, an inert metal and a common electrode material [5, 176], has been found to react with lead hybrid perovskite in previous studies [177,178,179,180,181]. Researches show that Au diffuses from the electrode across the HTL into the perovskite layer under certain conditions such as light soaking [182] or heat aging [183], resulting in an irreversible loss in photovoltaic performance and chemical stability of PSCs. Tarasov et al. proposed the corrosion mechanism of Au electrode based on its reaction with iodine-based perovskites [184]. Generally, the intensive UV irradiation of perovskite would induce the release of volatile I2 and MAI, forming highly reactive polyiodide melts with a general formula of MAI-nI2. This MAI-nI2 could react strongly with Au at room temperature and form [AuI2]− and [AuI4]− complexes. Consequently, a new (MA)2Au2I6 phase is observed on the Au interface, which is detrimental to the performance of PSCs.

$$ 2MAI - nI_{2} + 2Au\xrightarrow{{light}}\left( {MA} \right)_{2} Au_{2} I_{6} $$
(2.3-2)

In addition, MAPbI3/Au interface can catalyze the perovskite degradation reaction. A type of reduction/oxidation reaction termed underpotential deposition (UPD) involving lead, iodine, and hydrogen can occur at the MAPbI3/Au interface [185]. UPD is a surface adsorption reaction that changes the oxidation state of ions, and occurs spontaneously at a lower voltage than the bulk reaction potential. XPS measurements demonstrate that Pb0UPD and I0UPD form at MAPbI3/Au surface, and the whole degradation pathway of MAPbI3/Au can be described with the following steps (Fig. 12a): (1) I− loses an electron to form I0UPD that absorbs on the Au surface (Eq. 2.3-3), which induces the decomposition of MAPbI3 into PbI2 and CH3NH3+. (2) CH3NH3+ captures a free electron to form methylamine gas and H0UPD simultaneously (Eq. 2.3-4). (3) The adsorbed I0UPD reacts with H0UPD to generate HI, I2, or H2 gases (Eq. 2.3-5, 2.3-6, 2.2-7). (4) The byproduct CH3NH2 reacts with PbI2 via PbI (CH3NH2) interphase to form more HI, imines, and Pb0UPD (Eq. 2.3-8). Eventually, the degradation reaction of MAPbI3 will stop when the surface of the Au catalyst is completely covered with Pb0UPD. Hence, the detrimental interfacial chemical reactions provide a source for defects and reduce the chemical stability of PSCs.

Fig. 12
figure 12

a Proposed noble metal catalyzed degradation at MAPbI3/Au heterojunctions where the byproducts methylamine and PbI2 (in red) lead to the eventual formation of Pb0. Reproduced with permission from Ref. [188]. Copyright 2019, American Institute of Physics. b Proposed chemical corrosion of Al electrode of p-i-n type PSCs by the diffusion of CH3NH3I. Reproduced with permission from Ref. [192]. Copyright 2020, Wiley-VCH. In situ XRD analysis of the degradation of the LED stack with moist air in the dark: c In situ XRD spectra versus time. d Integrated XRD peak intensities of key materials versus time. e In situ SEM analysis on the LED stack in an environmental chamber at various time points from 0 to 180 min; the last one, SEM image near the boundary of the Al area after 180 min exposure showing that only the area covered with Al undergoes H2O-mediated decomposition. c-e Reproduced with permission from Ref. [193]. Copyright 2016, American Chemical Society

Full size image
$$ I^{ - } \xrightarrow{{Au}}I_{{UPD}}^{0} + e^{ - } $$
(2.3-3)
$$ CH_{3} NH_{3}^{ + } + e^{ - } \xrightarrow{{Au}}CH_{3} NH_{2} \uparrow + H_{{UPD}}^{0} $$
(2.3-4)
$${I}_{UPD}^{0}+{H}_{UPD}^{0}\to HI\uparrow $$
(2.3-5)
$${2I}_{UPD}^{0}\to {I}_{2}$$
(2.3-6)
$${2H}_{UPD}^{0}\to {H}_{2}\uparrow $$
(2.3-7)
$$C{H}_{3}N{H}_{2}+Pb{I}_{2}\stackrel{Au}{\to }C{H}_{2}=NH\uparrow +{Pb}_{UPD}^{0}+2HI\uparrow $$
(2.3-8)

Notably, Au not only reacts with perovskite through ions migrations but also can be corroded by interaction with the inorganic holes transport material CuSCN. It has been reported that the reactive thiocyanate anions are prone to react with Au electrode under an electrical bias, forming an undesired potential barrier for charge carriers [153, 165, 186, 187], which may seriously affect the device performance and chemical stability.

2.3.3 Al Electrode-Induced Reactions

Although Al is commonly employed as the top electrode in inverted PSCs, Al is chemically unstable due to its easy oxidation by air and reactions with lead halide perovskites [171, 188]. For example, Ma et al. found that both CH3NH3+ and I− can migrate towards the Al electrode through a PC61BM layer, resulting in the chemical corrosion of the Al electrode and the degradation of the perovskite film [189]. In particular, CH3NH3I is highly acidic and can react with Al as shown below:

$$ 6CH_{3} NH_{3} I + 2Al \to 6CH_{3} NH_{2} \uparrow + 2AlI_{3} + 3H_{2} \uparrow $$
(2.3-9)

For the overall degradation of PSC, the chemical corrosion of Al electrode can be described in Eq. 2.3-10. The reaction process involves the following steps (Fig. 12b) including: (1) formation of CH3NH3I from perovskite decomposition, (2) diffusion of CH3NH3I through the thin PC61BM region, (3) reaction between CH3NH3I and Al, yielding AlI3, CH3NH2 and H2, (4) formation of bubble and volcano-like surface defects owing to the release of H2 gas, (5) further decomposition of perovskite due to the Al3+ diffusion into perovskite layer.

$$6(C{H}_{3}{NH}_{3})Pb{I}_{3}+2Al\to 2Al{I}_{3}+{6PbI}_{2}+{6CH}_{3}{NH}_{2}\uparrow +{3H}_{2}\uparrow $$
(2.3-10)

To further investigate the Al-perovskite reactions, Zhao et al. employed in situ X-ray diffraction (XRD) analyses (Fig. 12c and d) and SEM measurements (Fig. 12e) to study the redox reactions between the Al electrode and perovskite [190]. The chemical degradation process can be divided into three stages. Stage 1: Al reacts with a perovskite in the presence of moisture, resulting in the reduction of Pb2+ to Pb0 and the crystal nucleation of MA4PbI6·2H2O, which can be expressed as follows:

$$ 2Al + 4\left( {CH_{3} NH_{3} } \right)PbI_{3} + 2H_{2} O \to \left( {CH_{3} NH_{3} } \right)_{4} PbI_{6} \cdot 2H_{2} O + 3Pb^{0} + 2Al^{{3 + }} + 6I^{ - } $$
(2.3-11)

Stage 2: with the continuous redox reaction between Al and Pb2+, the further loss of Pb2+ in MA4PbI6·2H2O will lead to the formation of MAI as shown below:

$$ 2Al + 3\left( {CH_{3} NH_{3} } \right)_{4} PbI_{6} \cdot 2H_{2} O \to 12CH_{3} NH_{3} I + 3Pb^{0} + 2Al^{{3 + }} + 6I^{ - } + 2H_{2} O $$
(2.3-12)

Stage 3: Once metal Al is completely oxidized to Al3+, further exposure to moisture could cause Pb0 to reoxidize and hydrate forming PbO·xH2O. In addition, volatile MAI gradually disappears due to prolonged exposure to air. Here, moisture not only promotes ions diffusion but also keeps the reactions between Al and MAPbI3 forward.

3 Strategies for Improving the Chemical Stability

Interfacial chemical reaction is a key factor affecting the chemical stability of lead halide PSCs. According to the above discussion, not all the interfacial reactions in PSCs are detrimental to device performance and chemical stability, such as the reactions at SnOx-Cl/perovskite interface. Therefore, in order to improve the chemical stability of PSCs, it’s necessary to inhibit unfavorable interfacial reactions. Two effective strategies have been proposed to reduce adverse interfacial reactions in the past few years. One is to insert a buffer layer at the contact interface to block ions migration or the invasion of water and oxygen. At present, many materials, such as PMMA [167], Al2O3 [82], and Cr2O3 [181], are introduced to optimize contact interfaces as a buffer layer. Another approach is to employ additives, such as 2-amylpyridine [158] and ionic liquids [191], in perovskites to reduce the chemical activity of the reactants.

3.1 Buffer Layers

Various buffer layers have been developed to modify the perovskite surface to prohibit the chemical reactions with various materials, including air, NiOx, spiro-OMeTAD, CuSCN, TiO2, ZnO, metal electrode and so on, based on different reaction mechanisms. The effective approaches for these interface modifications are detailed as follows.

Air/perovskite interface: The buffer layer, especially polymer materials, can isolate moisture or oxygen and inhibit the degradation of lead halide perovskite films. For example, hygroscopic polymer poly(ethylene oxide) (PEO) can chemically interact with undercoordinated Pb ions on a perovskite surface, passivating defect sites and reducing charge recombination loss [192]. In Fig. 13a and b, PEO prevents surface hydration reaction of perovskite by absorbing H2O before perovskites do, leading to greatly improved chemical stability of PSCs. A number of hydrophobic polymers, such as polystyrene [193], PTzDPPBTz [91] and carbon-based materials [194], are also used to protect perovskites from H2O. In Fig. 13c, Trichloro(3,3,3-trifl uoropropyl)silane was introduced between a perovskite layer and C60 ETL, which reacts with a tiny amount of H2O existing in the perovskite film to form silanols [193]. Then these silanols can automatically cross-link through forming Si–O–Si (siloxane) bonds to make the insulating layer protect the underlying perovskite film from water erosion.

Fig. 13
figure 13

a Schematic illustration of polymer PEO thin film assembled on the perovskite structure. b Operational stability test for evolution of PCE of FA0.87Cs0.13Pb(I0.87Br0.13)3 devices without and with PEO, continuously measured under xenon-lamp simulated fullspectrum AM 1.5, 100 mW cm-2 equivalent irradiance in air (humidity ~50 RH%) without any ultraviolet filter, held at maximum power point during ageing, and tested at 0.03-s time intervals. a, b Reproduced with permission from Ref. [195]. Copyright 2018, Royal Society of Chemistry. c Schematic diagram showing the cross-link process of fluoro-silane layer on the perovskite film. Reproduced with permission from Ref. [196]. Copyright 2016, Wiley-VCH. d The corresponding redox potential of I-/I3-and the homo levels of P3HT, PDCBT, and spiro-MeOTAD. e The unencapsulated device photostability tests under continuous one sun illumination in a home-built chamber filled with N2. d, e Reproduced with permission from Ref. [199]. Copyright 2017, AAAS

Full size image

It has been reported that low-dimensional materials covering perovskite surfaces can effectively insulate O2 and improve the perovskite film quality [58]. For example, Kanatzidis et al. inserted a 1D thiazole ammonium lead iodide (TAPbI3) interfacial layer between the perovskite and HTL, which serves to passivate the three-dimensional (3D) perovskite film and prevent oxygen invasion [195], leading to an improvement of device efficiency as well as enhancement of chemical stability.

NiOx/perovskite interface: The interfacial redox reaction between Ni3+ in NiOx and MAI salt will generate PbI2-rich holes extraction barriers, which can lower Voc and accelerate perovskite degradation by deprotonating precursor amine and oxidizing iodide to interstitial iodine [148]. A modifier layer, trimethylolpropane tris(2-methyl-1-aziridinepropionate) (SaC-100), was used to modify the NiOx/MAPbI3 contact interface [83]. Results reveal that N and O atoms in SaC-100 can passivate the uncoordinated Ni3+ and Pb2+ defects in NiOx and MAPbI3 through producing Lewis adducts, respectively. In addition, the reaction between NiOx and MAPbI3 is suppressed in the presence of SaC-100, which is helpful for improving the conductivity of NiOx and inhibiting the degradation of MAPbI3 films.

Spiro-OMeTAD/perovskite interface: Buffer layer materials can not only resist the invasion of air, but also block the undesired contacts between adjacent layers. For example, Wang et al. reported an effective buffer layer montmorillonite (MMT), which was inserted between perovskite and spiro-OMeTAD to prevent TBP from reacting with PbI2, because MMT can interact with TBP and build its intercalated structure by hydrogen bond. Consequently, the efficiency and chemical stability of PSCs are largely improved simultaneously [157]. I2 deriving from perovskite tends to react with HTLs, such as Spiro-OMeTAD, whose HOMO level is close to the oxidation potential of I−/I3− (~ − 5.0 eV). In Fig. 13d, depositing a polythiophene derivative with a lower HOMO level approaching − 5.3 eV on the perovskite surface can effectively suppress I−/I3− reaction, resulting in PSCs with enhanced efficiency and stability (Fig. 13e) [196].

CuSCN/perovskite interface: In order to promote the chemical stability of PSCs with CuSCN HTLs, it’s necessary to insert a buffer layer between perovskite and CuSCN. Snaith et al. not only introduced mesoporous Al2O3 at the perovskite/CuSCN interface to reduce the contact area of CuSCN with perovskite, but also encapsulated the complete cell with PMMA to prevent degradation products from release [82]. It has been found that if an electrical bias is applied on the PSC, an electrical potential-induced reaction between Au and SCN− would occur, leading to poor operational stability of PSCs. Therefore, a thin reduced graphene oxide (rGO) interlayer was introduced between CuSCN and Au electrode to migrate the degradation reaction [165]. In addition, Cu2O [197] and Cs:NiOx [198] interfacial layers were also used to reduce these detrimental interfacial chemical reaction for PSCs.

TiO2/perovskite interface: Ito et al. performed light irradiation tests of PSCs with/without a Sb2S3 inserted layer between TiO2 and MAPbI3 [124]. For the device without Sb2S3, the black MAPbI3 layers completely decompose to yellow PbI2 by losing HI and CH3NH2. On the other hand, the device with Sb2S3 can deactivate the reaction of I−/I2 at the surface of TiO2, so the MAPbI3 layer is stable and durable against light exposure as shown in Fig. 14a.

Fig. 14
figure 14

a Variation of photoenergy conversion efficiencies of solar cells during light exposure (AM1.5, 100 mW cm-2) without encapsulation in air for 12 h. Reproduced with permission from Ref. [125]. Copyright 2014, American Chemical Society. Composition depth profiles of the perovskite devices by SIMS: b with only PCBM as ETL; c with PCBM/ZnO bilayer as ETL. b, c Reproduced with permission from Ref. [204]. Copyright 2015, Elsevier. d Schematic energy level diagrams of cells with PCBM/Ag and PCBM/AgAl. Reproduced with permission from Ref. [206]. Copyright 2016, Elsevier. (Color figure online)

Full size image

ZnO/perovskite interface: A thin layer of MgO and protonated ethanolamine were in sequence deposited on the ZnO surface to resolve the poor chemical stability of ZnO-based device [199]. Moreover, 1,2-ethanedithiol [200], poly(ethylenimine) [132] and graphene oxide quantum dots [126] can also play the same role.

Metal/perovskite interface: The chemical instability of PSCs induced by metal electrode corrosion can be suppressed by interfacial modifications. I− in the perovskite can migrate to an Al electrode through PCBM layer to react with Al, forming aluminum iodide at the Al/perovskite interface. A bathocuproine (BCP) buffer layer has been reported to insert the interface between the Al electrode and PC61BM [189]. Test results support that this buffer layer can block the CH3NH3I migration and hence prevent the Al electrode from chemical corrosion by CH3NH3I, improving the chemical stability of the PSCs. Qiu et al. introduced a ZnO layer between PCBM and an Al electrode to prevent perovskite from reacting with Al, ToF–SIMS characterizations demonstrated that ZnO layer can effectively inhibit the diffusion of I− ions (Fig. 14b and c) [201].

Cr/Cr2O3 interlayer provides a buffer to shield top Ag electrodes from chemical corrosion caused by HI liberated from the MAPbI3 degradation [181, 202]. Compared with pure Ag, an AgAl alloy electrode shows much higher stability in aging tests, which is related to AlOx formed at the PCBM/AgAl interface during thermal evaporation and aging [203]. In Fig. 14d, this AlOx interlayer can suppress the migration of Ag atoms to the active layer, strengthen the metal contact with PCBM and prevent moisture encroachment. In addition, nanostructured carbon layers, including N-doped graphene, PCBM and carbon quantum dots, were inserted between a perovskite layer and an Ag electrode to block I− and Ag diffusion [204]. The degradation reaction of perovskite by ions migration was ultimately inhibited, and thus the chemical stability of PSCs was greatly enhanced. Moreover, a graphene barrier between CuSCN and Au can inhibit I− migration and perfectly block Au diffusion, as shown in Fig. 15a–c [205], successfully restraining undesired chemical reactions between I− and the Au electrode. Many groups introduced stable buffer layers through interfacial reactions to reduce ions migration [206]. For example, hexamethyldisilathiane is deposited on a perovskite surface to react with Pb2+ and form stable PbS buffer layer, which can effectively suppress I− diffusion and prevent corrosion of metal electrodes [207].

Fig. 15
figure 15

The iodide ion (I-) and Au diffusion in CuSCN/Au and CuSCN/GRP (3)/Au PSCs. a Schematic of the diffusion process during the light illumination under an applied electrical bias and dark state without the bias. Cross-sectional STEM images of CuSCN/Au b and CuSCN/GRP (3)/Au PCSs c after 3 cycles of continuous light illumination (12 h) and dark state (12 h). The scale bar is 200 nm. The EDX scan (I and Au) along the vertical lines (a´, b´, and c´) are also provided. a-c Reproduced with permission from Ref. [208]. Copyright 2020, Elsevier. d Chemical structure and moisture induced polymerization of E2CA with marked functional groups: C=C (red), C≡N (olive) and C=O (cyan). e Schematic illustrations of spontaneous grain polymerization in MAPbI3-E2CA films. E2CA chemically anchors to GBs with C=O and C≡N groups and spontaneously polymerizes to a polymer at GBs when exposed in moisture air. d, e Reproduced with permission from Ref. [214]. Copyright 2019, Elsevier. f XRD patterns of pristine and aged samples of control film and film containing BMIMBF4 (0.3 mol%) on NiO/FTOcoated glass substrates. The stars represent the decomposition product of PbI2 in the films. The insets show images of the aged samples (around 2.8 cm × 2.8 cm) after 72 h of light-soaking at 60-65 °C. Reproduced with permission from Ref. [217]. Copyright 2019, Springer Nature. (Color figure online)

Full size image

3.2 Additives Engineering

Water-induced reactions: Various additives have been demonstrated to change the physical or/and chemical properties of metal halide perovskites and thus improve the moisture-/oxygen-/light- stability of the materials [208]. For example, a fluorinated additive named 1,1,1-trifluoro-ethylammonium iodide (FEAI) was introduced into MAPbI3 to reduce the moisture-induced degradation reactions, because the hydrophobic CF3-terminal group on the perovskite surface resist the invasion of moisture [209]. Zhao et al. employed hygroscopic polyethylene glycol (PEG) scaffold to fabricate moisture-stable perovskite films because the omnipresent PEG molecules can absorb H2O efficiently. The resulting compact moisture barrier around perovskite crystal grains can prevent water from penetrating into the films [210]. The additive ethyl 2-cyanoacrylate (E2CA) can spontaneously polymerize into a compact polymer once exposed to moisture as shown in Fig. 15d and e, blocking the penetration channels of moisture due to its hydrophobic nature [211]. Moreover, poly(4-vinylpyridine) (PVP) [212], poly(ethylene oxide) (PEO) [192], and trimethylolpropane triacrylate (TMTA) [213] can also be used as additives to reduce the reactions between perovskite and H2O.

Oxygen-induced reactions: Snaith et al. introduced the ionic liquid 1-butyl-3-methylimidazolium tetrafluoroborate (BMIMBF4) into perovskite precursor and found that [BMIM]+ cations were bound to the surface sites, hence suppressing the surface degradation reactions induced by oxygen or moisture adsorption (Fig. 15f) [214]. The thiourea has been demonstrated to form Pb–S bonds on the outermost layer of perovskite, which efficiently suppressed the absorption of oxygen and prevented the formation of superoxide [215].

Light-induced reactions: The solid ionic additive 1-butyl-1-methylpiperidinium tetrafluoroborate ([BMP]+[BF4]−) was adopted to enhance the photostability of PSCs under full-spectrum sunlight [191]. It’s reported that the generation of I2 under illumination is one of the main factors causing the photo-degradation of perovskites [216]. The interstitial I− ions capture the holes to become neutral interstitial iodine atoms, and these neutral atoms need to diffuse and combine to produce I2 [112]. [BMP]+[BF4]− therefore suppressed the photochemical reactions of perovskite by reducing the surface defects, such as interstitial pairs and iodide vacancies, and retarding diffusion of neutral iodine interstitials.

CTL-induced reactions: The additives can also inhibit undesirable reactions occurring at charges transport layer/perovskite interfaces. Han et al. used a new additive 2-amylpyridine (2-Py) to replace TBP in Spiro-OMeTAD and found that the corrosion of perovskite was suppressed [158]. The existence of the amyl chain at the ortho-position of pyridine in 2-Py is vital to restrain the perovskite degradation reaction, which is attributed to the following aspects: 1) the amyl chain near the nitrogen atom can reduce the coordination ability of pyridine; 2) the steric hindrance from the o-substitution of pyridine prevents it from penetrating the perovskite. In addition, 2-Py can increase the dissolution and dispersion of Li-TFSI compared with TBP, which has a positive effect on the conductivity of the Spiro-OMeTAD HTL. Other additives, like TiO2 nanoparticles (NPs)-modified CNT (CNT:TiO2) [217], graphene oxide (GO) [218], and 1,6-diazidohexane (N3) [219] can also be introduced into spiro-OMeTAD to minimize the TBP-PbI2 complex formation and hence improve the chemical stability of the perovskite/HTL interface.

Metal electrode-induced reactions: It has been reported that the ions migration can initiate the degradation of devices [113, 220], which can be efficiently inhibited by some additives with special functional groups. For example, caffeine with two conjugated carboxyl groups interacts strongly with Pb2+ ions to slow down the perovskite crystal growth and thus produce high-quality films [221]. In Fig. 16, the Energy-dispersive X-ray spectra (EDX) mapping shows that Ag can diffuse into the whole perovskite region and I− ions also migrate through the PTAA layer in the control device, while there is no obvious indication of such similar ions migrations in the caffeine-incorporated device. The caffeine significantly suppresses ions migration, and hence the chemical stability of PSCs is effectively enhanced. Chen et al. employed the methimazole (MMI) to form the MMI-PbI2 complex in situ at the GBs, and these surface patches can also effectively suppress Ag diffusion and simultaneously retard I− migration [222].

Fig. 16
figure 16

a-f Energy-dispersive X-ray spectra (EDX) mapping of the aged pure PVSK device a Ag, b I, and c Pb, and the aged caffeine-containing PVSK device d Ag, e I, and f Pb. g, h EDX line scans of g aged pure PVSK device and h aged caffeine-containing PVSK device. Reproduced with permission from Ref. [224]. Copyright 2019, Elsevier. (Color figure online)

Full size image

The multifunctional additive of dipentaerythritol pentaacrylate (DPPA) can impede I− movements due to synergistic effects of its functional groups [223]. The CH2=CH groups in DPPA crosslink at GBs and block the channels of ions migrations. And the -OH groups in DPPA can bond with I− in the perovskite through hydrogen bond interaction, chemically immobilizing these migrated I−. In addition, the polystyrene (PS) [224] and PCBM [74] can also obstruct ions migrations across GBs and dissociated ions can only move around their original sites. Therefore, the problem of chemical stability induced by ions migration can be well solved.

4 Conclusions and Outlook

Although lead halide PSCs have already achieved a high certified PCE up to 25.7% through a low-cost and simple fabrication process, the long-term stability still lags far behind the commercial application standard. Therefore, the top priority is to explore the degradation mechanism of lead halide PSCs in more depth, which is helpful for achieving high-efficiency and stable devices. As one of the most important factors influencing the photovoltaic performance and chemical stability of PSCs, interfacial chemical reactions due to environment conditions, interface contacts and metal electrode are systematically elucidated in this paper. Their reaction mechanisms as well as influence on interface defects, non-radiative recombination and degradation process are summarized. Up to now, there is no unified theory that generalizes all interface reactions. For example, lead halide perovskites degrade easily to form PbI2 and other lead-containing compounds once exposed to humid environment. O2 capture photo-induced electrons to generate superoxide O2·− with the aid of light, which can induce quick degradation of lead halide perovskite films. In addition, some chemical reactions at charge transport layers/perovskite interface can produce undesirable interfacial defects and reduce charge transport capacity of ETL or HTL, eventually deteriorate the chemical stability and photovoltaic performance of PSCs. It’s noteworthy that Cl-containing SnO2 ETL can react friendly with lead halide perovskites to passivate the interface defects and improve the device efficiency. The interfacial reactions are induced by ions migrations, for example, I− can diffuse into a metal electrode and simultaneously metal ions can also migrate through the perovskite, which corrode the metal electrode and decrease the operational lifetime of PSCs.

At present, various buffer layers or additives have been employed to suppress the detrimental interfacial reactions and optimize the device performance as well as long-term stability. Therefore, it’s necessary to investigate the chemical properties of additives or buffer layer materials in depth, because their different functional groups may play completely different roles in the photovoltaic performance. The fluorine-containing additives can improve the hydrophobic properties of lead halide perovskite films, and effectively reduce the humidity-induced degradation of PSCs. The detrimental I2 in the lead halide perovskite easily induces the shallow charge traps and accelerates the light-decomposition of perovskite. The hydrazine functional group can reduce these I2 in perovskite precursor solutions to I−, which significantly improve the light stability of perovskites. In addition, more attentions should be paid to benign chemical reaction in PSCs and thus we can make use of these reactions to further improve the efficiency and stability of PSCs. For example, the organic long chain amine (e.g. phenylethylamine, butylamine) can react with PbI2 to form stable 2D perovskite on the bulk perovskite surface, contributing to the decreased non-radiative recombination and enhanced water resistance of 3D perovskite films. Investigating novel and stable charge transport layer and electrode materials is also of significance for the chemical stability of PSCs. Future molecular design of charge transport materials should consider the energy level alignment, high carrier mobility and stability. The cheap and chemically stable carbon electrode might be an alternative to reduce the cost and deterioration for lead halide PSCs. It’s expected that rational management of interfacial chemical reactions in the whole PSCs will lead to substantial performance enhancement, paying the way for the commercialization of PSCs in the future.