Lead Halide Perovskite Nanocrystals in the Research Spotlight: Stability and Defect Tolerance.

This Perspective outlines basic structural and optical properties of lead halide perovskite colloidal nanocrystals, highlighting differences and similarities between them and conventional II-VI and III-V semiconductor quantum dots. A detailed insight into two important issues inherent to lead halide perovskite nanocrystals then follows, namely, the advantages of defect tolerance and the necessity to improve their stability in environmental conditions. The defect tolerance of lead halide perovskites offers an impetus to search for similar attributes in other related heavy metal-free compounds. We discuss the origins of the significantly blue-shifted emission from CsPbBr3 nanocrystals and the synthetic strategies toward fabrication of stable perovskite nanocrystal materials with emission in the red and infrared parts of the optical spectrum, which are related to fabrication of mixed cation compounds guided by Goldschmidt tolerance factor considerations. We conclude with the view on perspectives of use of the colloidal perovskite nanocrystals for applications in backlighting of liquid-crystal TV displays.

In the past few years, lead halide perovskites (LHPs) in the form of colloidal nanocrystals (NCs), such as organic–inorganic CH3NH3PbX3 LHPs (often denoted as MAPbX3, with MA standing for the methylammonium cation) and all-inorganic CsPbX3 LHPs (X = Cl, Br, I), have been intensively investigated for various applications such as light-emitting devices (LEDs) and photodetectors due to their color-tunable and narrow-band emission as well as easy synthesis, convenient solution-based processing, and low fabrication cost. We refer interested readers to some recent reviews for comprehensive treatment of these topics.[1−6] Most striking has been the impact of thin-film perovskites in photovoltaics,[7] with extremely high power conversion efficiencies of more than 22%,[8] and the reports of light-emitting diodes with external quantum efficiencies over 10%.[9,10] The literature underpinning the development in bulk and thin-film perovskites is very extensive and is not covered herein; instead, we focus on nanoscale perovskite NCs and their emerging applications. In this Perspective, we first provide some short historical remarks on LHPs and then outline their basic structural and optical properties, highlighting differences and similarities between the LHP NCs and conventional II–VI and III–V semiconductor quantum dots (QDs). We then proceed with more detailed insight into two important issues inherent to LHP NCs, namely, the innate advantage of so-called defect tolerance and the necessary steps required to improve their stability under the environmental conditions found in devices. We then discuss several issues that need to be addressed in the burgeoning field of LHP NCs, such as the origin of the significantly blue-shifted photoluminescence (PL) from CsPbBr3 NCs and the synthetic strategies toward fabrication of stable mixed-cation LHP NC materials with an optimum Goldschmidt tolerance factor (TF) that emit in the red and near-infrared part of the optical spectrum. The defect tolerance of LHP NCs offers strong inspiration to search for similar attributes in other related compounds, especially those that do not contain toxic lead or other heavy metals. We also provide our view on the perspectives on the use of the colloidal LHP NCs for applications in backlighting of liquid-crystal displays for television (LCD TV displays) and other related color conversion and color enhancement applications.

Basic Properties of Lead Halide Perovskite Nanocrystals. The synthesis of bulk CsPbX3 compounds was reported as early as 1893,[11] whereas their perovskite crystal structure and photoconductive, and hence semiconductive, nature was only discerned later, in the 1950s.[12] Since then and until the late 1990s, CsPbX3 compounds were rather thoroughly characterized as to the details of their crystallography and phase diagrams, including direct structural characterization by X-ray and neutron scattering and by nuclear magnetic resonance.[13−28] Also, a number of lead-free halide perovskites were studied at that time; CsSnX3 phase transitions were characterized,[6,29−31] and specific electrical conductivity was observed.[32] CsGeCl3 was reported to have dielectric constants comparable to BaTiO3 while exhibiting ferroelectric characteristics as well.[33] In 1978, Weber et al. synthesized and determined the crystal structure of MAPbX3 for the first time.[34]

The motivation to investigate LHPs in the form of colloidal NCs has its roots in prior successes of colloidal QDs of conventional semiconductors (CdSe, CdTe, PbSe, InP, and the like).[36] LHP NCs have spurred intense research efforts owing to, on one hand, their extremely facile synthesis (Figure , upper part) and, on the other hand, their very bright PL covering the entire visible spectral range (Figure a). These highly crystalline, cubic-shaped NCs (Figure b) reflect the intrinsic near-cubic symmetry of the crystal lattice (Figure c). Just like their perovskite oxide ancestors (i.e., CaTiO3), LHPs crystallize into an ABX3-like lattice that comprises three-dimensional (3D) corner-shared [PbX6] octahedra (X being Cl, Br, I). There are commonly three cations, namely, cesium (Cs+), methylammonium (MA, CH3NH3+), and formamidinium (FA, CH(NH2)2+), which fit into a 12-coordinate A-site formed in between [PbX6] octahedra. According to the Goldschmidt TF,[37] any substantially larger or smaller, for example, by 10% or more, A-site ion would destabilize the lattice and induce conversion into lower-dimensional lead halide compounds, with much larger bandgaps, as was observed experimentally.[38]

Figure 1

Colloidal synthesis of CsPbX3 NCs and (a) photographs of resulting colloidal solutions of CsPbX3 NCs; (b) HAADF-STEM image of a single CsPbBr3 NC; (c) idealized perovskite crystal structure with 3D interconnection of PbX6 octahedra. In reality, all APbX3 perovskites not only adopt the cubic polymorph but are commonly observed in lattices with lower symmetry, for example, orthorhombic and tetragonal, due to distortions along one or more Pb–X–Pb axes. Adapted from refs (35) and (112).

In contrast to other semiconductor materials (Si, GaAs, Cd chalcogenides, In pnictides), LHPs are highly ionic compounds. Hence, it is not surprising that they readily and easily form highly crystalline NCs even at room temperature. Colloidal synthesis of CsPbX3 NCs, depicted in Figure ,[35] represents just one out of numerous variations of the ionic coprecipitation method, optimized to obtain narrow size dispersions. Size control and colloidal stability are imparted by the capping ligands, typically a mixture of a carboxylic acid (such as oleic acid, OA) and alkylamines (such as oleylamine, OLA).[39−42]

The first colloidal synthesis of organic–inorganic MAPbBr3 NCs was reported in 2014 by Galian and Perez-Prieto, who used an alkyl ammonium bromide with a medium-sized chain to stabilize small-sized crystallites in a suspension;[43] the same group further enhanced their PL quantum yield (QY) to 100%.[44] Soon after that first publication, Zhong’s group introduced a ligand-assisted reprecipitation (LARP) technique (as shown in Figure a) in a mixture of a good and a poor solvent to produce MAPbX3 (X = Cl, Br, I) NCs with a tunable bandgap by varying halide elements; the same group also reported improved LARP and in situ fabrication later.[39,45−47] Later in Huang’s related report,[40] bandgap tunability of MAPbBr3 NCs while controlling the LARP process by modifying the poor solvent’s temperature was demonstrated (Figure b), and NCs with high PL QYs of up to 93% and high crystallinity (Figure c) were obtained. 1D and 2D perovskite NCs have also been explored, and quantum confinement has been completely verified and quantified in the 2D case.[41,42,48] Simple top-down fabrication of MAPbBr3 and MAPbI3 NCs by employing a mixture of OA and OLA ligands as coordinating solvents under ultrasonication was also demonstrated by Huang et al.[49] The ultrasonication approach was also demonstrated by Hintermay et al. and Tong et al.[50,51]

Figure 2

Colloidal synthesis of MAPbBr3 NCs and (a) schematic diagram of the LARP reaction system; (b) colloidal solutions of MAPbBr3 NCs obtained by using different temperatures of toluene solvent; (c) HRTEM image of a single MAPbBr3 NC. Adapted from ref (40). Copyright 2015, Wiley Online Library.

Combinations of LHP NCs can provide wide color gamuts covering the whole visible spectral range (400–700 nm with CsPbX3 and MAPbX3 NCs) and the emission can even be extended into the infrared (up to 800 nm with FAPbI3 NCs). In the visible, emission line widths are narrow, typically less than 100 meV, corresponding to a full width at half-maximum (fwhm) of 12–50 nm. The lower widths are seen at shorter wavelengths, in the blue, and a meaningful means of comparison is to take the fwhm divided by the central wavelength (i.e., the fractional bandwidth)[52] as this would bear some relationship to the size distribution and highlight major differences in the latter where the fractional bandwidths differ significantly. PL QYs are high, even without the benefits of core–shell passivation, and can reach peak values of up to 95–100%.[35,53] Such high PL QYs are a direct consequence of the defect-tolerant nature of LHP’s electronic structure, which we will consider in detail below.

These attractive optical characteristics of LHP NCs are counterbalanced by several major issues related to the stability of these materials. The key difficulty from the viewpoint of chemical stability concerns MAPbX3 NCs.[43] Due to the low energy of formation, MA-based LHPs can eventually decompose into PbX2 and volatile byproducts (i.e., CH3NH2, HI, I2, etc.). This decomposition is greatly accelerated by the high surface area of LHP NCs and by moisture, oxygen, heat, light, and their combined effects.[54,55] Often, MA-based LHP NCs decompose during isolation and purification procedures. Higher durability has been observed with FA- and Cs-based LHP NCs.[35,53] Owing to the considerable ionicity of the bonding, yet another challenge specific to all LHP NCs is their instability in essentially all polar solvents. In addition, LHP NCs exhibit rather moderate thermal stability due to either low melting points of 400–500 °C (CsPbX3) or thermal decomposition (MAPbI3 at ca. 150–200 °C; FAPbI3 at ca. 290–300 °C). In addition, a great challenge originates from the rather labile and dynamic nature of the ligand binding in these materials,[56] causing a loss of colloidal stability during the purification of LHP NC colloids. These challenges have led to intense research into alternative ligand chemistries[57,58] and developing coatings with protective polymeric or inorganic layers,[59,60] which we will consider in detail below.

Another form of structural instability comes from the polymorphism, which is especially pronounced for iodide-based LHPs (CsPbI3 and FAPbI3). 3D polymorphs of CsPbI3 and FAPbI3 are thermodynamically metastable and undergo transitions to wide-bandgap 1D polymorphs.[13−16,61−65] Thin films and NCs of CsPbI3 and FAPbI3 exhibit extended but finite stability in their 3D polymorphs (days to several months), primarily due to surface effects.[35,53,66−69] Thermodynamic instability is caused by the Cs and FA ions being, respectively, slightly too small and too large for the A-site, as determined by the Goldschmidt TF and by the octahedral factor for the required dense packing in 3D perovskites.[37,70−74] Combined with the chemical instability of the MAPbI3 NC system, a “red wall” exists for LHP NCs—a difficulty to obtain stable NCs with PL in red and near-infrared spectral regions.

APbX3 perovskites that feature 3D interconnection of PbX6 octahedra are of primary interest. These octahedra form either an ideal cubic lattice (typical for FAPbBr3 and FAPbI3) or a similar 3D orthorhombic one (CsPbX3). In the case of iodide LHPs (FAPbI3 and CsPbI3), 3D -phases are metastable at room temperature, and the instability decreases upon reduction of crystallite size from bulk to NCs.[53,75] Although FAPbI3 NCs are stable for at least several months, CsPbI3 NCs are highly unstable and, at best, retain their red PL for several weeks, only. Poor chemical stability of MAPbI3 and poor phase stability of its FA and Cs cousins had been previously termed by us as the “perovskite red wall”.[53] To illustrate the mitigation strategy on this issue, which can be based on employment of mixed-cation perovskites, we briefly review the underlying reasons for the phase transformation illustrated in Figure . Perovskite structures can be viewed as a close-packing of ions, and hence, the Goldschmidt TF concept, commonly used for metal-oxide perovskites[37] can be also extended to LHPs.[71,72] For ideal 3D cubic close-packing, the Goldschmidt TF is calculated aswhere rA, rPb, and rX are the ionic radii of each ion. In an ideal close-packing case, TF = 1. Although for more ionic oxides TF = 0.8–1 is known as an empirical stability range, higher covalency in LHPs and nonsphericity of their A-cations (both MA and FA) lead to the observation of stable 3D perovskites only for TF ≥ 0.9. The data from Travis et al.[71] for APbX3 compounds provides TF values for CsPbBr3 of 0.9 and for CsPbI3 of 0.89, indicating that CsPbI3 is likely a borderline case. In FAPbI3, the nonsphericity of the cation complicates the analysis. Travis et al. estimated a radius of 2.53 Å for the FA cation, and the corresponding TF value for FAPbI3 is 1.03. Cs–FA mixed-cation perovskites with composition-averaged TF values fall within the stability window, as demonstrated recently.[53,76] Similar strategies of the use of mixed-cation perovskite thin films have also led to major recent advances in photovoltaics, as demonstrated on Cs0.17FA0.83(PbI1–Br)3 (x = 0–1), (FAPbI3)1–(MAPbBr3) (x = 0–0.3), and (Cs/MA/FA/Rb)(PbI1–Br)3.[77−81]

Figure 3

Illustration of the “perovskite red wall”: 3D phases of CsPbI3 (orthorhombic) and FAPbI3 (pseudocubic) materials, with bandgap energies at ca. 710 and 840 nm, respectively, easily convert into wide-bandgap 1D polymorphs. The transition is typically observed as a change of the color from dark-red/black to yellow.

The composition control of LHP NCs is more flexible and convenient than that for many conventional semiconductor QDs. The tunability of perovskite NC compositions can be achieved after synthesis through subsequent anion exchange, which is more facile than that for many conventional II–VI and III–V QDs. In chalcogenide NCs, cation exchange is quite common and easy to certain degrees;[82] however, the anion exchange is rarely reported in such materials. Anion sublattice bonding is rather stronger than that of the cation sublattice, while the anions themselves are often bulkier than the cations, making anion exchange difficult without using extreme conditions, and usually any exchange that is observed is not topotaxial.[83,84]

Another outstanding feature of perovskite NCs is that they can have high PL QYs, which have even reached 100%,[44] by virtue of their fortuitous band structures, as discussed further in the next section. The PL fwhm of perovskite NCs is narrower than that for most of the other types of QDs.[1,35,36,39] Narrower line width emission is said to be more saturated, placing the fluorescence color coordinates more toward the curved edge of the CIE chromaticity space (e.g., CIE 1931 standard).[39,85] Combinations of three emitters (red, green, and blue), which lie close to the fully saturated boundary curve, can then create the widest range of perceived colors, termed the color gamut, by display and lighting manufacturers. The cost of production of perovskite NCs is regarded as low because of their solvent processing and relatively low temperature synthesis.

The relatively low or nonoccurrence of fluorescence blinking[86,87] of LHP NCs is an attractive prospect for hot carrier/multiexciton effects as it is probably a marker for relatively weak Auger recombination effects. However, the photothermal stability of the materials under high photon energy and at high fluences is a factor that needs to be addressed in order to fully realize the benefits of such effects.

Defect Tolerance of Lead Halide Perovskite Nanocrystals. One of the most striking features of LHPs is their high tolerance toward defects. The term “defect tolerance” here means that, though the optical and electronic properties of perovskites often appear as though there are no electronic traps or excessive doping present, structural and other characterization methods do point to a large density of various structural defects. From the electronic point of view, such behavior suggests preservation of a clean bandgap upon creation of typical defects such as vacancies or surface-related sites because their defect energy levels reside entirely within either the valence band (VB) or the conduction band (CB) manifolds but not within the bandgap itself. In this regard, perovskite NCs are highly unusual;[88] they are highly luminescent without recourse to any electronic surface passivation, whereas such passivation is mandatory to achieve a high PL QY from conventional QDs derived from metal chalcogenides (i.e., CdSe) or metal pnictides (i.e., InP).

The defect tolerance had been rationalized theoretically for a variety of perovskite compounds. For CsPbBr3, for instance, the surfaces of NCs, point defects in the bulk material,[89] as well as grain boundaries[90] were all shown to either form shallow trap states or to be resonant with VB and CB states. The defect tolerance is partly attributed to the high ionicity of bonding in LHPs. Furthermore, mixing of a Pb lone pair s orbital and an iodine p orbital results in antibonding coupling in the perovskite lattice, with the bandgap opening up between two antibonding bands. Because of this band structure, structural defects that may arise from the halide and MA or other A+-type vacancies tend to have energy levels that fall within the CB and VB, respectively, rather than lying within the bandgap itself. On the contrary, in conventional, defect-intolerant semiconductors such as Si, CdSe, or GaAs, the bandgap is formed between bonding and antibonding orbitals, leading to enclosure of all defect states either as shallow or as midgap states as bonding is locally weakened at all defect sites (point defects, dislocations, planar defects, surfaces, etc.). The comparison is schematically depicted in Figure .

Figure 4

(a) Schematics of two limiting cases of a band-structure in semiconductors: defect-intolerant (conventional, left) and ideal hypothetical defect-tolerant (right). Bonding and antibonding orbitals are denoted as σ and σ*, respectively. (b) Simplified depiction of the bonding in APbI3 (adapted from ref (96)). The VB exhibits the desired antibonding character at its maximum, as in the ideal defect-tolerant case in (a). Copyright 2015, Materials Research Society.

A second reason for having clean bandgaps relates to the energy of defect formation in LHPs. Halide and A-site vacancies (VX and VA) are easily formed as a pair of Schottky vacancies, thus maintaining overall charge neutrality of the lattice. Fortunately, in the LHPs, other point defects, such as interstitially or antisite misplaced atoms, have much higher energies of formation,[91] often even above the formation energy of the parent compound. This scenario is illustrated in Figure for MAPbI3. On the basis of thermodynamic calculations, ionic compensation of point defects in MAPbI3 has been suggested as a charge carrier concentration self-compensation mechanism.[92] Defect tolerance is similarly expected to be of high relevance also in 2D perovskites.[93] NCs can be robust light emitters, even when a large number of ligands are displaced from the surface, and yet the influence of consequent surface defects in trapping charge carriers is negligible.[94,95]

Figure 5

Energy levels associated with the defect states corresponding to neutral and charged vacancies (VPb, VI, VMA), neutral and charged interstitials (Pbi, Ii, MAi), and neutral and charged states associated with antisites (PbI and IPb) in MAPbI3. Adapted from ref (91). Copyright 2015, American Chemical Society.

Synthetic Strategies toward Improving the Stability of Lead Halide Perovskite Nanocrystals. Employing different ligands to improve or change the properties of as-prepared materials is a very common strategy in the colloidal QD field, and this is particularly relevant to increasing the stability of LHP NCs given their innate sensitivity to water and other polar solvents. Figure a shows an attempt to use different ligands other than the commonly used OA or OLA by Luo et al.[97] By using branched capping ligands, (3-aminopropyl) triethoxysilane (APTES) and polyhedral oligomeric silsesquioxane (POSS) PSS-[3-(2-aminoethyl)amino]propylheptaisobutyl substituted (NH2-POSS), the authors showed that APTES and POSS could be used as passivators and stabilizers of MAPbBr3 NCs. PL monitoring under exposure to ethanol under UV lamp irradiation indicated enhanced stability when using those two ligands. Galian and Pérez-Prieto et al. reported MAPbBr3 perovskite NCs with a PL QY of ∼100% by using 2-adamantylammonium bromide (ADBr) as the only capping ligand.[44] The photodarkening of these nanoparticles under prolonged irradiation, attributed to moisture, can be avoided by the formation of cucurbit[7]-uril-adamantyl ammonium host–guest complexes (AD@CB) on the NC surface. Figure b demonstrates the higher photostability of MAPbBr3 NCs with the latter coating in toluene dispersions even under water with UV photoirradiation.

Figure 6

(a) Molecular structure of APTES and NH2-POSS and photograph of MAPbBr3 NCs (different capping ligands) dispersed in ethanol under UV light. Adapted from ref (97). (b) PL of MAPbBr3 NCs PAD–CB (in black) and PAD (in green) dispersed in toluene and in contact with water as a function of the irradiation time. (Right side) Images of colloidal dispersions immediately after addition (left) of water and 120 min afterward (right); the inset shows the molecular structures of cucurbit[7]uril (CB) and 2-adamantylammonium (AD) ligands. Adapted from ref (44). Copyright 2016 and 2016, Wiley Online Library. (c) Crystal structures of the CsPbI3 (i) cubic and (ii) orthorhombic perovskite. PL spectra for CsPbI3–OA (iii) and CsPbI3–TMPPA (iv). Insets of (iii and iv): Solutions of the respective washed NCs under UV light at different times following synthesis. Adapted from ref (98). (d) Schematics showing the core–shell type of octylammonium lead bromide nanomaterials over MAPbBr3 NPs. Adapted from ref (99). Copyright 2016 and 2016, The Royal Society of Chemistry. (e) Optical stability of CsPbBr3–I and CsPbBr3–I/ZnS. Adapted from ref (100). Copyright 2017, Wiley Online Library. (f) PL intensity as a function of time after storing the self-passivation layer formation on mixed-halide perovskite NCs after acetone etching CsPb(BrI1–)3 NCs in cyclohexane under ambient conditions. Adapted from ref (101). Copyright 2017, The Royal Society of Chemistry.

Besides the issue of stability in contact with moisture and under irradiation with light, it is well-known that CsPbI3 NCs suffer from a facile cubic perovskite to orthorhombic phase transformation (as demonstrated in Figure c(i,ii)), which may be a limiting factor for their optoelectronic applications. By replacing the conventionally used OA with an alkyl phosphinic acid, Wang et al. obtained phase-stable cubic perovskite CsPbI3 NCs (Figure c(iii,iv)).[98] By changing the ligands, the as-prepared sample remained luminescent for over 20 days while the OA comparison sample showed no emission to the naked eye.

Producing core–shell structures to increase stability is yet another widely used strategy in colloidal semiconductor QDs. Similar treatments have also been used in perovskite syntheses. Bhaumik et al.[99] reported a putative mixed MA–octylammonium lead bromide perovskite core–shell-type structure (Figure d). With a thin shell and little to contrast for the core from the shell in TEM images, it was difficult for the authors to show direct evidence of the formation of a shell; however, indirect evidence from elemental analyses and improved PL stability was taken as tentative evidence of successful shell formation. The emission color was tunable in the blue to green range by using different MA–octylammonium ratios (438–521 nm), while the PL QY was as high as 92%. Their solution-processed material was reported to be stable at least for 2 months under ambient conditions.

Chen et al. reported a NC architecture made of CsPbX3/ZnS heterodimers synthesized via a facile solution-phase process (Figure e).[100]Figure e compares the PL stabilities for pure CsPbBr3–I and CsPbBr3–I/ZnS heterodimers. The CsPbBr3–I /ZnS heterodimer could keep for about 12 days without any protection in air, while pure CsPbBr3–I QDs became unstable and blue-shifted within 1 day under the same conditions.

Jing et al. found that the stability of mixed-halide CsPb(BrI1–)3 NCs could be dramatically enhanced by using a selective acetone etching method.[101] This formed a passivation layer on iodine-rich perovskite NCs by partial iodine etching to instead leave a bromine-rich surface passivation layer (Figure f). After the treatment, the PL 50% decay constant was around 17500 h compared with 20 h for the untreated NCs. In other words, the PL stability was increased almost 1000-fold.

In terms of postsynthetic treatments, the employment of silica or silicone derivative coatings on LHP NCs has been proven useful. Huang et al. fabricated SiO2-encapsulated MAPbBr3 QDs by using a small amount of water in analytical-grade toluene to hydrolyze tetramethyl orthosilicate.[102] Photostability tests were carried out at a relative humidity of 60%, and after 7 h, the PL of the encapsulated powders remained at 94% of the initial value, higher than that for the unencapsulated sample, where the PL had declined to 38% of the original level (Figure a).

Figure 7

(a) Evaluation of photostability of MAPbBr3(MAPB)-QD and MAPB-QDs/SiO2 powders in relative humidities (RHs) of 60 and 80% under 470 nm LED light irradiation. Optical images of the as-prepared colloidal MAPB-QD solutions with TEOS and TMOS and storage after 4 days. Adapted from ref (102). Copyright 2016, American Chemical Society. (b) Thiol-functionalized POSS structure: a schematic diagram illustrating the POSS coating process to obtain perovskite NC powders. Adapted from ref (103). Copyright 2016, The Royal Society of Chemistry. (c) Photostability test of MP-CsPbBr3 and CsPbBr3. Adapted from ref (104). Copyright 2016, Wiley Online Library. (d) Schematic of the silicone resin (SR) coating process for the preparation of SR/PVP-CsPbX3. Thermal stability test of SR/PVP-CsPbBr3 and CsPbBr3 QDs. Photostability test of SR/PVP-CsPbBr3 and CsPbBr3 QDs under continuous UV light irradiation. Adapted from ref (105). Copyright 2017, The Royal Society of Chemistry.

The first successful water-resistant coating of solid-state perovskite powders was demonstrated by Huang et al.[103] through surface passivation of CsPbX3 (X = Br or I) with POSS molecules, as shown in Figure b. In the form of aqueous suspensions, CsPbX3/POSS composites retained their emission unchanged for several months. The POSS coating was also useful when two-color emitters were formed by mixing different composition perovskite NCs as it prevented undesirable anion exchange reactions between the different constituents from occurring in the powder state. The benefits of this passivation strategy were demonstrated when green-emitting POSS-CsPbBr3 and red-emitting POSS-CsPb(Br/I)3 NC powder mixtures were used to fabricate all-perovskite solid-state luminophore down-conversion white light LEDs.

Wang et al. used commercially available mesoporous silica mixed with green CsPbBr3 NCs[104] to similarly bestow water resistance and prevent ion exchange in their mixtures of different composition. The photostability comparison is shown in Figure c. By infiltrating perovskite precursors into mesoporous silica after drying, Dirin et al. showed the formation of perovskite NCs entrapped within the pores.[88] Sun et al. used a similar hydrolysis approach[102] with another silica source APTES.[106]

Hai et al. reported a simple fabrication method for emissive flexible films composed of polyvinylpyrrolidone (PVP) as a matrix polymer and codoping blue, green, and red CsPbX3 (X = I, Br, and Cl) as guest fluorophores at various ratios.[105] A schematic of their hydrophobic silicone resin (SR)/PVP NC composite film, SR/PVP-CsPbX3, is presented in Figure d. PVP-coated NCs (as single or multiple component mixtures) were electrospun to form nanofiber films using single- or multinozzle electrospinning. To provide further protection from humidity and facilitate handling, SR was deposited onto the surface of the composite electrospun nanofibers to obtain water-stable nanofibrous membranes.

Apart from silica coating, polymer coating has also proved useful in LHP NC passivation. Meyns et al. demonstrated the addition of poly(maleicanhydride-alt-1-octadecene) (PMA) into the precursor mixture during the synthesis of perovskite NCs.[107] The normalized integrals of the emission peaks between 460 and 600 nm over 12 h of constant irradiation showed higher emission signals for samples with PMA compared with untreated NCs (Figure a). By reducing the ligand surface exchange rate, the ligand binding was tightened in the presence of the PMA, reducing the scope for the NC surface to interact with the surrounding medium, thereby improving the NC stability.

Figure 8

(a) Normalized integrals of the emission peaks between 460 and 600 nm of CsPbBr3 with/without PMA over 12 h of constant irradiation. The inset shows the structure of PMA. Adapted from ref (107). (b) Stability of CsPbBr3 perovskite colloid and film. Perovskite colloids mixed with different solvents (ethanol, IPA). Time evolution of fluorescence intensity after adding 1:1 v:v solvents into perovskite colloids. Time evolution of fluorescence intensity after immersing perovskite/polymer hybrid films in different solvents. Adapted from ref (108). (c) (i) Relative PL QYs of as-synthesized CsPbBr3 NCs in water after different times. Inset pictures show as-synthesized samples before and after 60 min of water soaking under a UV lamp. (ii) Relative and absolute PL QYs of 150 μm thick nanocube polymer composite films after over 4 months of water-soaking. Thin 3 μm composite films also showed enhanced water stability (see inset pictures). Adapted from ref (109). Copyright 2016, 2017, and 2016, respectively, American Chemical Society.

Zhang et al. formed water-resistant polystyrene microhemispheres (MHSs) embedded with CsPbX3 (X = Cl, Br, I) NCs (denoted as NCs@MHSs) as hybrid multicolor and multiplexed optical coding agents.[110] PVP acted as the capping ligand and was adsorbed onto the perovskite NC surface and in doing so formed a protective layer. The PVP surface thus formed also served as an interface layer for further addition of an additional polystyrene matrix allowing the CsPbX3 NCs to be embedded in polymer MHSs. The well-passivated CsPbX3 NCs@MHSs were incorporated into live cells showing high stability and noncytotoxicity and functioned as useful multicolor luminescent probes.

Hou et al. demonstrated stable core–shell colloidal LHP NCs using a copolymer templated synthesis approach.[108] The block copolymer served as a confined nanoreactor during perovskite crystallization and passivated the perovskite surface by forming a multidentate capping shell. The polymer nanoshell provided an additional layer for further surface modifications, useful for self-assembly and so forth and also served to passivate and improve the photostability of the NCs. Figure b compares the PL stability of CsPbBr3 NCs with the multidentate copolymer ligand and with small-molecule ligands (OA and OLA) upon exposure to ethanol and propan-2-ol. While OA/OLA-capped NCs quenched immediately after mixing the colloids with both solvents and the PL totally disappeared within 3 h, the multidentate polymer/perovskite NC samples exhibited stable fluorescence after more than 25 h in ethanol and for up to 50 days after adding IPA.

Raja et al. reported enhanced water and light stability by encapsulation of CsPbBr3 NCs into matched presynthesized hydrophobic macroscale polymeric matrixes.[109] Their CsPbBr3 QDs lost all emission after 60 min of contact with water (Figure c(i)), while the NC/polymer composite films functioned even after more than 4 months of continuous immersion in water (Figure c(ii)). The author also claimed no detectable lead leaching into the water that was in contact with the encapsulated perovskites.

Summary and Future Outlook. There are a number of research avenues related to LHP NCs that will require attention in the forthcoming years. One of the puzzling questions concerns the origin of the significantly blue-shifted PL from CsPbBr3 NCs. Interestingly, both the PL peak and absorption edge from CsPbBr3 NCs never exceed 520 nm, even at NC sizes far beyond the quantum-confinement regime (>20 nm). In fact, bulk CsPbBr3 has an optical band gap at 2.25 eV (551 nm), both in our experiments and in the literature.[111] Our experience shows that the PL peak for NCs larger than 11 nm is always at exactly 520 nm, fully ruling out the quantum size effects at these large sizes as the origin of the blue shift. At present, the atomistic origin of this effect remains unclear. Rather broad X-ray diffraction reflections of CsPbBr3 NCs make it difficult to differentiate between the orthorhombic (nearly cubic) lattice of the bulk material and other possible distortions of the ideal cubic lattice. A recent study suggested significant and dynamic structural disorder that involves formation and re-formation of twin planes between orthorhombic perovskite subdomains in CsPbBr3.[112]

It has been not easy to push the emission of LHP NCs toward the red and near-infrared spectral range while maintaining reasonable material stability. An effective strategy to overcome this so-called “red wall” is mixing larger FA+ and smaller Cs+ in one lattice, thereby compensating for the poor individual fits of these ions separately. An additional stabilizing factor in this case is provided by the high entropy of mixing.[80] Formation of mixed-cation compositions in iodide-based LHPs has become a major strategy in thin-film solar cell research, yielding the highest power conversion efficiencies of up to 22%: FA/MA,[79,113−115] Cs/MA,[116] Cs/FA,[70,77,78,80] Cs/MA/FA,[117] or even Rb/Cs/MA/FA.[81] Recently, this approach has been extended to LHP NCs, namely, for a (Cs/FA)PbI3.[53] Similarly, other mixed-cation formulations have been investigated as well, including Au–CsPbBr3, Cs1–RbPbBr3, and so on.[118−120] Further work will establish the synthesis procedures and elucidate structures for corresponding multinary LHP NCs.

The defect tolerance of LHP NCs offers strong inspiration to search for similar attributes in other related compounds, especially those that do not contain toxic lead or other heavy metals.[6,121] Similar electronic structures and defect-tolerant behavior are to be expected from the main-group metals, which offer both s and p electrons for the formation of the VB and CB. A first example is through the replacement of Pb2+ with Bi3+, an ion of similar size. Yet, the resulting compounds of composition Cs3M2X9 (M = Sb, Bi) have vastly different crystal structures, dominated by 0D or 2D networks of Bi–X polyhedra, and exhibit no significant PL at ambient conditions.[122] A full structural analogue of 3D perovskites can be constructed by replacing Pb2+ with a 1:1 mixture of M+ and one M3+, forming so-called double perovskites, A2M+M3+X6, such as such as Cs2BiAgCl6 and Cs2AgInCl6.[123,124] The electronic band structure of thallium halides also shows a strong resemblance to LHPs.[125] Finally, the most obvious strategy—replacement of Pb2+ with Sn2+ and Ge2+—has thus far failed due to oxidative instability, even with respect to trace quantities of oxygen. Even trace amounts of Sn4+ and Ge4+ degenerately dope such semiconductors. In this regard, a somewhat surprising finding is the bright and air-stable emission, albeit with broad fwhm in excess of 100 nm, from (C4N2H14Br)4SnX6 (X = Br, I),[126] a compound comprising isolated SnX64– octahedra in a land of large organic cations. One can assume that oxidative stability is enabled by these cations that prevent diffusion of oxygen to the Sn2+ sites. This observation might open an avenue to other stable hybrid organic–inorganic lead-free perovskites.

Many strategies discussed in this Perspective for perovskite NC stability enhancement would leave the NCs inaccessible in terms of injection of charges, which could be detrimental for a number of optoelectronic applications. They still have a vast possibility of applications such as color-conversion and color-enhancing layers. If the stability of LHP NCs can successfully be improved, with the narrow PL fwhm of just 18–20 nm in the green at 530 nm (CsPbBr3, FAPbBr3 NCs) and 35 at 630 nm (CsPb(Br/I)3) and high PL QYs of up to 95–100%, LHP NCs may become a strong competitor to traditional colloidal QDs for applications in backlit TV displays and in related color-conversion and color-enhancing applications. At present, two principal types of QD emitters in the red and green have been successfully commercialized in LCD TVs: CdSe-based QDs by Sony in 2014 and InP-based QDs by Samsung in 2015 (under the brand name SUHD TV). Perovskite NCs could be used to replace CdSe or InP QDs in those commercialized LCD TVs, potentially exceeding their performance in terms of color saturation and brightness in the longer term. Under the pressure of increasingly stringent legislation for the use of heavy metals in consumer electronics, Cd use is being limited in such applications. Lead, on the other hand, is exempted for several applications, such as in lead-acid batteries, produced globally on the millions of tons scale. For comparison, one TV display of typical 40–60 in. dimension requires only several mg of QDs,[127] summing up to at most several kilograms at substantial TV display market penetration. LHP NCs could offer strong competition with regard to InP-based QDs arising from the inherently much narrower, size-independent emission, being at 530 nm twice as narrow as the equivalent III–V-based NCs (fwhm ≈ 40 at 530 nm for InP-based QDs).