Introducing hydrogen bonding is also an effective method to passivate defects of perovskite films^[18−21]. Li et al. demonstrated that (4-aminobutyl) phosphonic acid hydrochloride (4-ABPACl) can heal the grain boundaries through strong hydrogen bondings between –PO(OH)₂ and –NH₃⁺ terminals and the perovskite (O–H···I and N–H···I) (Fig. 2(c))^[18]. The modified perovskite gave higher PCEs. In addtion to these molecules with hydroxyl and amino groups, fluorides are also used as passivators by forming hydrogen bonding between fluorine atoms and MA⁺ or FA^+[21]. These interactions can not only modulate crystallization to enhance crystallinity but also improve structural stability of perovskite films.

In short, different defects in perovskite layer are discussed, and typical passivation strategies are highlighted. More investigation on the generation, location and types of defects would be helpful for making highly efficient and stable PSCs.

Considering the structural feature and ionic nature of perovskite materials, most trap states are point defects and inevitably charged. The positively charged defects include uncoordinated Pb²⁺ ions, halogen vacancies, interstitial cations, A_X, B_X and B_A anti-sites, while negatively charged defects are uncoordinated I⁻ ions, cation vacancies, interstitial halogen ions, X_A, X_B and A_B anti-sites (Fig. 1(b))^[4]. For anti-site defects, A_X means A cation occupies X site. In addtition to charged defects, metal Pb clusters can also form in the perovskite films due to the reduction of uncoordinated Pb²⁺. The generation of a defect is directly determined by its formation energy. Once a defect forms, it will introduce additional transition energy level (Fig. 1(c))^[5]. The transition levels near the conduction band minimum (CBM) or valence band maximum (VBM) edge are called shallow-level trap states, like vacancies and interstitial ions, whose formation energy is relatively low. The trapped carriers in shallow-level defects have high de-trapping probability to fall back into the CBM or VBM, thereby contributing little to nonradiative recombination. But defect migration can lead to undesirable local band bending, phase segregation, and current–voltage (J–V) hysteresis, as well as degradation of perovskite films due to interfacial reactions^[6]. Transition energy levels located within the middle third of forbidden band are identified as deep-level trap states, like uncoordinated ions, anti-sites and Pb clusters, which have higher formation energy. Different from shallow-level trap state, once a deep-level trap state is formed, it will serve as charge recombination centre and induce nonradiative recombination, thus decreasing the PCE. In order to diminish negative impact of various defects, the passivators should have following functions: (1) providing chemical bonding to deactivate the defects; (2) forming wide-bandgap phases or type-I heterojunction at grain boundaries and interfaces to suppress nonradiative recombination; (3) optimizing crystal orientation to reduce structural defects.

Lewis acid/base coordination chemistry is widely employed to passivate uncoordinated defects. Lewis acid and Lewis base can accept and donate electrons, which means that they can passivate electron-rich and hole-rich defects via forming Lewis adducts, respectively. Fullerene (C₆₀) and its derivatives PCBM are the most popular Lewis acids due to their excellent electron-accepting ability, making them powerful to passivate Lewis base type defects. Combining theoretical calculation and experimental characterizations, Xu et al. revealed that PCBM can effectively passivate I_Pb antisite and uncoordinated I⁻ to suppress nonradiative recombination and promote electron extraction (Fig. 2(a))^[7]. In addition, a PCBM passivation layer can also inhibit anion migration that may result in J–V hysteresis^[8]. Compared with Lewis acid, Lewis base type passivators are more widely employed. Most Lewis bases base on nitrogen, oxygen, and sulphur atoms containing lone pair of electrons. Noel et al. first reported that orgainc molecules like thiophene and pyridine can serve as Lewis base to passivate uncoordinated Pb²⁺ (Fig. 2(b))^[9]. Thiophene and pyridine can donate electrons to Pb²⁺ and form a coordinate covalent bond, neutralizing the excess positive charge of uncoordinated Pb²⁺. As a result, the treated perovskite films demonstrated suppressed nonradiative recombination and enhanced carrier lifetimes. Later, Lewis bases bearing amines, hydroxyl and carbonyl groups show effective passivation on uncoordinated Pb²⁺ defects^[10−13]. Moreover, some conjugated small molecules or polymers also show excellent passivation on Lewis acid traps. These molecules usually contain functional groups like thiophene, pyridine, amine, thiocyanato and so on, thus providing multiple passivation sites^[14−17].

Organic-inorganic hybrid perovskite materials are ionic crystals with typical ABX₃ perovskite structure^[3], where A is monovalent cations like CH₃NH₃⁺ (MA⁺), NH₂CHNH₂⁺ (FA⁺) and Cs⁺, B is divalent cations (Pb²⁺ or Sn²⁺), and X is halogen anions (I⁻, Br⁻ or Cl⁻) (Fig. 1(a)). Owing to the diversity of compositions, various defects like vacancies, interstitial ions, and anti-sites can be generated in the solution-processed polycrystalline films. These defects mainly exist at grain boundaries and interfaces, which significantly influence the efficiency and stability of PSCs. Therefore, most works have focused on passivating defects to improve the device performance.

Recently, halogen bonding has been used to enhance the efficiency of PSCs^[22−25]. Snaith et al. identified that the uncoordinated I⁻ at perovskite surface are responsible for the charge accumulation and consequent recombination losses. They modified perovskite surface by using iodopentafluorobenzene (IPFB), which can form halogen bonding with uncoordinated I⁻ to passivate these defects^[22]. In IPFB, the fluorine atoms withdraw electrons out of the benzene ring, then the benzene ring in turn withdraws electrons from I⁻, thereby leaving a partially positive charge around iodine atom (Fig. 2(d)). This electropositive region would interact with halogen anions to form strong halogen bonding. Similar results were achieved by using 4-iodo-2,3,5,6-tetrafluorobenzoic acid (I-TFBA)^[25].

Perovskite solar cell (PSC) is now a shining star in photovoltaics field^[1]. Benefiting from excellent optoelectronic properties of perovskite materials, the certified power conversion efficiency (PCE) of PSCs has reached 26.0%^[2], showing great potential for commercialization. In essence, the efficiency of solar cells is determined by the radiative and nonradiative recombination of photogenerated charge carriers. The unfavorable nonradiative recombination mainly assisted by the trap states leads to severe charge carrier loss and thus unsatisfactory efficiency.

Ionic passivation bases on electrostatic interaction between passivators and the charged defects in perovskite films. In earlier studies, Na⁺ and K⁺ cations were selected to passivate defects like MA vacancies and uncoordinated halogen ions via ionic bonding^[26]. Owing to small ionic radius, Na⁺ and K⁺ cations are unfavorable to occupy A-sites, but they can occupy the interstitial sites in perovskite lattice to suppress ion migration and eliminate hysteresis^[27]. Some research suggested that Na⁺ and K⁺ cations can also assemble at grain boundaries and surface, forming KI or KBr to passivate defects and improve carrier lifetimes (Fig. 2(e))^[28]. Since You et al. reported highly efficient PSCs through phenylethylammonium iodide (PEAI) post-treatment (Fig. 2(f))^[29], organic ammonium halides have been intensively used as effective passivators. The halogen ions can fill in halogen vacancies, while −NH3+ terminal can interact with negatively charged defects in perovskite films, suppressing nonradiative recombination^[30]. Beyond these materials, zwitterions can simultaneously passivate positively and negatively charged defects (Fig. 2(g))^[31]. Usually, zwitterions contain positively charged ammonium group (−NH3+) and negatively charged carboxylate (−COO−), phosphate (−PO42−) or sulfonic (−SO3−) groups^[31−33].

Introducing wide-bandgap phases to form a type-I heterojunction at grain boundaries and interfaces can effectively suppress nonradiative recombination. The most common approach is to introduce excess PbI₂ to perovskite films (Fig. 3(a))^[34]. The type-I energy alignment (Fig. 3(b)) between PbI₂ and perovskite can hold back charge transfer at grain boundaries (Fig. 3(c)), thereby suppressing nonradiative recombination^[35]. Excess PbI₂ might also induce film degradation. In view of this, converting excess PbI₂ to two-dimensional (2D) perovskites by using long-chain ammoniums becomes a frontier technique. The 2D perovskites can establish type-I heterojunction to suppress nonradiative recombination (Fig. 3(d))^[36]. Moreover, the strong hydrophobicity of 2D perovskites would protect 3D perovskite from water erosion. Based on this technique, scientists have made highly efficient and stable PSCs. It should be noted that the issues for 2D or quasi-2D perovskites are crystal orientation and layer-layer distance. The undesirable crystal orientation and long layer-layer distance would definitely block charge transport in the film and at interfaces, which might sacrifice J_sc and FF^[37]. Therefore, optimizing crystal orientation and layer-layer distance of 2D or quasi-2D perovskites is critical.

Though using passivators is promising to deactivate various defects, the initial generation of defects should not be ignored. Perovskite layers made from solution are polycrystalline films that inevitably present grain boundaries. Since grain boundaries contain various defects, increasing grain size to reduce grain boundaries is vital for high-performance PSCs. Moreover, the optoelectronic anisotropy of perovskite films is quite important. Different crystal facets might present different trap states^{[38, 39]}. Single-crystal perovskites show less defects. Zheng et al. investigated the dependence of defect density on crystallographic facets via density functional theory (DFT) calculations based on a cubic FAPbI₃ model^[40]. They revealed that there are no defects on (100) facet, while the (110), (111) and (211) facets present defects (Fig. 3(e)). By incorporating a trace amount of surface-anchoring alkylamine ligands (AALs) into perovskite precursor, they achieved (100)-oriented crystal growth with lower defect density and enhanced carrier mobility. Similar phenomenon was observed in Pb-Sn mixed perovskites^{[41, 42]}. Optimizing crystal orientation and exploring facet-dependent defect variation have attracted interests^[43−45].