Colloidal semiconductor nanocrystals (NCs), also known as colloidal quantum dots, have inspired intensive study for decades due to their unique optical and electrical properties, which are closely related to their size, dimension and morphology^[1-4]. These outstanding properties, such as narrow emission bandwidth, size-dependent emission and high luminescent efficiency, combined with solution-processed ability, contribute to their huge potentials for applications in various optoelectronic devices^[5-8]. The past decade has witnessed the impressive development of an ancient but novel type of materials, metal halide perovskite NCs^[9-11]. Generally, named after Russian mineralogist Lev Perovski, perovskites refer to a large family of materials adopting a similar crystal structure to calcium titanate (CaTiO₃) with a chemical formula of ABX₃. As for the halide perovskite NCs, “A” usually represents methylammonium (MA), formamidinium (FA), cesium (Cs), rubidium (Rb) or the mixture; “B” can be lead (Pb), tin (Sn), bismuth (Bi) or the mixture; “X” refers to halide anions including chlorine (Cl), bromine (Br), iodine (I) or the mixture. The halide perovskite NCs can be divided into hybrid organic–inorganic perovskite NCs, where A is an organic group, and all-inorganic NCs when A is an inorganic cation. Since the first application of CH₃NH₃PbX₃ NCs as sensitizers for dye-sensitized solar cells in 2009^[12], enormous attention has been attracted to make further exploration into this field^[13-15]. Currently, with dramatically fast development speed, the highest power conversion efficiencies (PCEs) of solar cells based on hybrid perovskite NCs has already reached 23%^[16], mainly benefiting from their high light absorption coefficients (> 1 × 10⁴ cm⁻¹) and long balanced charge-carrier diffusion lengths (> 100 nm). Endowed with outstanding optical properties, the hybrid perovskite NCs are also regarded as promising cost-effective materials for light-emitting diodes (LEDs) and other optical devices^[17-19]. However, the hybrid perovskite NCs are extremely sensitive to oxygen, moisture, heat and light due to the existence of organic groups in the structure^[20]. In contrast, the all-inorganic perovskite NCs counterparts are ideal candidates as inherently equipped with higher thermal decomposition temperature and more stable crystal structures^{[21, 22]}.

In order to efficiently adjust the optical properties of perovskite NCs, it is crucial to understand the relationships between structures and characteristics. As is well known, the metal halide perovskites have three main crystalline structures including cubic, orthorhombic, and tetragonal phases. In these different phases, it is generally believed that the cubic phase appears at relatively high temperature^[23], but CsPbX₃ NCs can exhibit the cubic phase above room temperature due to size reduction^[24]. Commonly, the CsPbCl₃ and CsPbBr₃ NCs are proven to be relatively stable in composition while the unstable cubic CsPbI₃ NCs enjoy the large band-gap for photovoltaic applications (E_g = 1.73 eV)^[25]. Taking the crystal structure of cubic CsPbBr₃ as an example, as is shown in Fig. 1(a), Cs ions occupy the vertexes of cubes while six Br ions assemble an octahedron with a Pb ion in the center of the framework. Previous reports demonstrate that Cs ions have little effect on the electronic structure of band edge, thus the excitation and recombination of electrons and holes are confined within the [PbBr₆]^4– octahedra^[24].

Since the first synthesis of colloidal CsPbX₃ NCs was reported by Kovalenko and co-workers in 2015^[26], tremendous attention has been paid to developing different synthetic strategies for preparing high-quality of colloidal CsPbX₃ NCs^{[24, 27-30]}. These studies not only shed light on the growth mechanism but also lay a concrete foundation for further practical applications of CsPbX₃ NCs. In this work, we will thoroughly review the existing protocols for the controllable synthesis of CsPbX₃ NCs and their various analogues by focusing on several key parameters in the reaction process. Besides, strategies for stability improvement will also be generally summarized. At last, a brief introduction to several promising applications of CsPbX₃ NCs in sundry fields is presented, with some current challenges and future development perspectives being proposed.

The first synthesis of all-inorganic metal halide perovskite crystals CsPbX₃ (X = Cl, Br or I), and their analogue Cs₄PbX₆ (X = Cl or Br) in solution phase was reported as early as 1893 by Wells and his collaborators^[31]. The systematic studies in the structural and physical properties of all-inorganic metal halide perovskites can be traced back to more than sixty years ago, in which Moller studied the crystal structure and photoconductivity of CsPbX₃ bulk materials^[32]. However, the research failed to attract intensive attention due to the poor optical properties and complex fabrication strategy until the more facile and scalable method to synthesize CsPbX₃ NCs was reported by Protesescu and co-workers in 2015^[26]. In this work, the colloidal CsPbX₃ NCs with uniform cubic shape and outstanding optical properties were successfully synthesized through a hot-injection method. In a typical synthesis, specific amount of Pb(II) halide (PbX₂) powder is dissolved into a degassed mixture of the surfactants (usually organic amine and organic acid, OAm and OA) and the solvent (usually octadecene), followed by quick-injection of the heated pre-synthesized Cs precursor (usually Cs-oleate) into the reactant at a certain temperature (usually 120–200 °C). The very fast nucleation and growth of the CsPbX₃ NCs can finish within several seconds owing to the intrinsic ionic crystal feature. Consequently, the size of cubic CsPbX₃ NCs can be tuned effectively by changing the reaction temperature or the precursor concentration rather than the growth time. The schematic diagram of the hot-injection method is presented in Fig. 2(a). The all-inorganic perovskite NCs with mixed halide composition, such as CsPb(Cl/Br)₃ and CsPb(Br/I)₃, can be easily achieved through changing the ratios of PbX₂ ingredients. Figs. 1(b)–1(d) show the uniform cubic shape and outstanding optical properties of the as-synthesized CsPbX₃ NCs. The facile and efficient hot-injection method of CsPbX₃ NCs lays a concrete foundation for the following intensive study of the unique properties and wide applications.

In 2017, Yerama et al. applied an under the pressure-assisted hot-injection approach to realize a linearly upscale synthesis of CsPbBr₃ NCs^[33]. In a typical synthesis, the reactor is maintained vacuum to heat the PbX₂ precursor, followed by the injection of warm Cs-oleate under a mild vacuum with the pressure reaching 1−10 mbar. The under pressure-governed hot-injection method could yield multigram-scale CsPbBr₃ NCs, 200 times more compared with the traditional method, with a narrow emission width (~22 nm) and a high photoluminescence quantum yield (PLQY) of 76%.

In addition, Chen et al.^[34] reported a one-pot strategy based on the conventional synthesis method, in which all reagents, including PbX₂ (X = Cl, Br, and I) salts solution and Cs-oleate solution, were loaded into a single reaction pot and subsequently heated to the desired reaction temperature in air. The as-synthesized CsPbX₃ exhibited a shape of quantum dots instead of nanocubes. This non-injection method can yield a large amount of high-quality NCs on the gram scale with a one-batch reaction. It also can realize size control, compositional modulation and band-gap tuning simultaneously, which pave the way for further applications.

Although the hot-injection method conducted at high temperatures has been widely applied, several problems still remain to be solved. For example, the scalable synthesis may not be achievable using this hot-injection method due to the requirement of conventional heating, inert atmosphere and injection conditions. Besides, the modification scope of morphology and optical properties is very limited at high temperatures. To compensate for the disadvantages of the hot-injection method, partially inspired by a ligand-assisted re-crystallization strategy to fabricate hybrid CH₃NH₃PbX₃ NCs^[35], Zeng group developed a complementary re-crystallization method through creating a supersaturated condition when inorganic ions including Cs⁺, Pb²⁺ and X^– transformed from good solvents into poor solvents at room temperature^[36]. The schematic illustration of the synthesis route is shown in Fig. 2(b). Firstly, CsX and PbX₂ (X = Cl, Br, I or their mixtures) of specific amounts under their solubility are dissolved completely in a good solvent (dimethyl formamide (DMF) or dimethyl sulfoxide (DMSO) commonly), with surfactants (OAm and OA). Subsequently, a small amount of the above solution is added into a bad solvent (usually toluene). The solubility of inorganic ions in toluene is much smaller (within 10⁻⁷ g/mL) while DMF or DMSO and toluene are mutually soluble. Given that metal ions are insoluble in toluene (within 10^–7 g/mL), a highly supersaturated state is produced immediately inducing a rapid re-crystallization of CsPbX₃ NCs in the mutually mixed solution. In this case, surfactants can help to disperse all precursors in various nonpolar solvents and to control the size, shape and optical properties as well. Apart from the re-crystallization method, ultrasonication is another method for synthesis of CsPbX₃ NCs at room temperature. Tong et al.^[37] developed a versatile, polar-solvent-free and single-step approach for scalable synthesis of highly luminescent CsPbX₃ NCs with tunable-thickness through ultrasonication. As shown in Fig. 2(c), the precursors consisting of Cs₂CO₃, PbX₂ and surface ligands (OAm and OA) are dissolved in a nonpolar solvent (mineral oil or octadecene). The sonication firstly induces the formation of the Cs-oleate and then helps to yield the colloidal CsPbX₃ NCs, which exhibit enhanced air stability. Although the CsPbX₃ NCs are obtained at room temperature without inert atmosphere protection, they still exhibit comparable optical properties to that of NCs synthesized at high temperatures using hot-injection method.

Traditional colloidal CsPbX₃ NCs with organic ligands capped on the surface are easily aggregated, resulting from the strong dispersive interactions of the surfactants^[38]. Moreover, for the tiny CsPbX₃ NCs, there are usually difficulties in the isolation and purification process^[39]. To solve this problem, Kovalenko et al. reported another facile, non-colloidal and ligand-free strategy to prepare a large variety of perovskite APbX₃ NCs using mesoporous silica (meso-SiO₂) as a growth template^[40] (Fig. 3(a)). This method benefits a lot from the unique defect-tolerant character of perovskite NCs, getting rid of the strict demanding of surface-passivation^[9]. In the synthesis, various commercially available mesoporous silica templates (preheated to 50 °C) are firstly immersed into highly concentrated solutions of the precursor salts (AX and PbX₂), which exceeds the pore volume of the template. Then, the mixture is stirred in an inert atmosphere at room temperature to achieve the full impregnation. Afterwards, excess solution is removed through filtration to prevent the formation of perovskite crystals outside the pores. Finally, the as-obtained samples are dried under vacuum at a relatively high temperature. The as-prepared perovskite NCs exhibit pretty high PLQY exceeding 50%. Furthermore, as the size of the NCs can be controlled facilely by choosing different silica templates, such as hexagonally ordered one-dimensional channels (2.5–7 nm pore widths), interconnected cubically ordered cavities (3 nm), or disordered pore networks (15–50 nm pore widths) of a specific characteristic size, this protocol can also open the possibility of a systematic way to study the size-dependent properties at small sizes.

The simple, efficient and microwave-assisted protocol was developed by the Zhang group^[41], and different-shaped colloidal CsPbX₃ NCs are successfully synthesized, which exhibit and the highest PLQY of 75%. During the synthesis process, all precursors, including cesium acetate (CsOAc), lead halide, trioctylphosphine oxide (TOPO), OA, OAm, and octadecene (ODE), are simply mixed in a microwave quartz tube without any pre-synthesis of Cs-oleate. By simple microwave irradiation, stable colloidal CsPbX₃ NCs in the shapes of nanocubes, nanorods, NWs, quadrate nanoplates and hexagonal nanoplates can be achieved.

Microfluidic reaction systems have been widely used in the mechanism study or the fine-tuning of reaction factors^[42]. As far as the formation mechanism of CsPbX₃ NCs is concerned, the accurate and systematic understanding of related parameters is very limited because of the extremely fast kinetics and intricate involved variables^[26]. Consequently, Kovalenko and coworkers adopted a droplet-based microfluidic platform to in-situ gain the insights into the formation of CsPbX₃ NCs over a millisecond to several seconds time span^[43]. Meanwhile, the online photoluminescence and absorption measurements allow the rapid depiction of various governing factors, such as the molar ratios of precursors, compositions and temperatures (Fig. 3(b)).

Very recently, Liao and co-workers reported a general one-step electrospinning strategy for in-situ growth of all-inorganic perovskite NCs in polymer fibers (Fig. 3(c))^[44]. The optical properties of the as-prepared perovskite NCs@polymer fibers can be readily tuned by tailoring the compositions of lead halide and cesium halide precursors within the electrospinning solution, allowing their PL emission covering the whole visible range. What’s more, the stability against water and thermal of the composites was fundamentally enhanced, with their PLQYs keeping 70% of its original value after immersion in water for 192 h and maintaining over 50% at 80 °C for 2 h.

Currently, the most synthesis methods of CsPbX₃ NCs fail to avoid using toxic solvents. In order to realize the environmental-friendly production of CsPbX₃ NCs, Zhu et al. developed a new mechano-synthesis strategy which is free of solvent and high reaction temperature^[45]. The simplified synthesis process is depicted in Fig. 3(d). Taking the synthesis of CsPbBr₃ NCs as an example, the metal salts CsBr and PbBr₂ are directly mixed through high-speed mechanically milling or grinding. The color of the precursors would change into yellow and CsPbBr₃ NCs or partial bulk products appeared within a few minutes. Subsequently, further surface treatment with OAm was necessary to improve the luminescence of the CsPbBr₃ NCs^[46]. The rationale may be attributed to the reactivity of the mobile phase on the exposed solid reactants’ surfaces^[47]. This mechano-synthesis method provides the greenest avenue in the preparation CsPbX₃ NCs.

Non-solvent preparation of CsPbX₃ NCs can be realized through the gas-phase synthesis as well. For instance, Jin group^[48] explored a facile and general chemical vapor deposition (CVD) method to fabricate CsPbX₃ nanowires (NWs) and microwires (MWs), in which a mixture of CsX and PbX₂ powders as precursors heated in the middle of the furnace at about 300 to 350 °C in the flow of argon gas. The as-prepared CsPbX₃ NCs were horizontally oriented with high density in mica, and a variety of CsPbBr₃ nanostructures, including Y-shaped branches or interconnected NW or MW networks can be achieved via simple tuning the deposition time. The as-synthesized high-quality ligand-free CsPbX₃ networks grown on mica substrates could serve as outstanding substrates for photoelectronic devices and other applications.

The optical properties can be conveniently controlled by tuning the halide compositions of the precursors during the synthesis of CsPbX₃ NCs. However, the synthesis of highly bright pure CsPbCl₃ NCs and CsPbI₃ NCs is usually not as easy as that of CsPbBr₃ NCs. Fortunately, similar to their organic-inorganic hybrid counterparts, a more facile post-synthetic compositional modulation method was developed by Kovalenko group^[49] and Manna group^[50] independently almost at the same time. Thanks to their ionic features, the bandgap and PL emission are readily tunable over the entire visible spectral region through the fast halide anion-exchange reaction. CsPbBr₃ NCs are commonly chosen as the initial materials due to their good stability and strong fluorescence. Several types of halide sources have been investigated, which include the simple PbX₂ salts, organometallic Grignard reagents (MeMgX), oleylammonium halides (OAmX), octadecylammonium halides (ODAX), tetrabutylammonium halides (TBAX). The anion-exchange reaction quickly happens within seconds upon the mix of CsPbBr₃ NCs and halide precursors together, leading to a blue shift (Br to Cl or I to Br routes) or a redshift (Br to I or Cl to Br routes) of the PL emission wavelength. Besides, it is worth to be noted that this anion exchange reaction can occur by directly mixing CsPbX₃ NCs of different halide compositions as well, even without any additional halide sources. The shape, crystal structure and even the PL properties of the as-obtained NCs are almost identical to their parent NCs, except a little size change after the anion-exchange. As is mentioned above, the synthesis of CsPb(Cl/I)₃ NCs is not available due to the significant difference in the ionic radii. Accordingly, the anion-exchange works only for the Cl–Br and Br–I couples in both directions, but never from CsPbCl₃ to CsPbI₃ nor in the reverse direction. Treating CsPbCl₃ NCs with a large excess of iodide source would give rise to transforming the PL color from blue directly to red, lacking any intermediate colors, and vice versa. Fig. 4 shows the overview of different routes and precursors for the anion-exchange reactions on CsPbX₃.

In a broad sense, the general formula of the all-inorganic cesium lead halide perovskites can be written as Cs_aPb_bX_a+2b, where the most common structure is CsPbX₃ (a = 1, b = 1). Although the elemental compositions of these analogues are similar, their crystal structures are totally different. For the CsPbX₃ NCs, as is mentioned above, the [PbX₆]⁴⁻ octahedrons are corner-shared in the unit cells (Fig. 1(a)), which is considered as a three-dimensional (3D) structure. As regards the CsPb₂X₅ NCs and Cs₂PbX₄ NCs, both named two-dimensional (2D) structures, the [PbX₆]⁴⁻ octahedrons tend to form planes with Cs⁺ inserted between them. The [PbX₆]⁴⁻ octahedrons in the Cs₄PbX₆ NCs are separated from each other while Cs⁺ cations fill the centers of four adjacent [PbX₆]⁴⁻ octahedrons, forming the zero-dimensional (0D) structure consequently.

Some reports have demonstrated remarkable improvement in the optical properties of perovskite materials with a reduction of dimensionality^[51-53]. As for the 2D structures, the CsPb₂Br₅ nanoplates obtained by the precipitation reaction and fast anion exchange was reported by Yu group^[54], whose PLQY in solution was claimed to reach 87% and PL spectra were able to cover the entire visible spectral region. Besides, the CsPb₂X₅ NWs were also successfully prepared by Deng group through a ligand-mediated method with enhanced stability^[55]. In addition, a dual-phase composite CsPbBr₃-CsPb₂Br₅ NCs were synthesized by Sun group^[56]. The parasite of CsPb₂Br₅ phase on the cubic CsPbBr₃ NCs could reduce diffusion length of excitons and decrease the trap density. The CsPb₂Br₅ microplates were also successfully prepared by Tang etal.^[57]. Besides, very recently, another all-inorganic perovskite analogue enjoying two-dimensional phase was reported for the first time with the hybrid composition of Cs₂PbI₂Cl₂^[58]. This compound exhibited a direct bandgap of 3.04 eV and a large in-plane resistivity. A strong UV-light response and an effective in-plane α-particle counting ability make this 2D perovskite polymorph a very promising optoelectronic material.

Some work on the 0D Cs₄PbX₆ NCs indicates that the NCs exhibit outstanding luminescent properties as well. For example, novel perovskite-related Cs₄PbBr₆ semiconductors were fabricated by an easy interface reaction strategy in a liquid–liquid immiscible two-phase system^[59]. Although the particle sizes were on the micrometer scale, the PLQY could reach 40%–45%. Mohammed and co-workers^[60] reported a facile room-temperature synthesis of pure Cs₄PbBr₆ NCs in solid and claimed the high luminescence properties with a 45% PLQY of this 0D perovskite derivative. The exciton binding energy was estimated to be 353 meV.

However, there are still some controversies lying in the accurate luminescence mechanism of these perovskite polymorphs. Some researchers argue that the bright luminescence probably arises from the inevitable presence of highly luminescent CsPbX₃ NCs even at quite low concentrations^{[56, 61]}. Meanwhile, recent related studies have firmly confirmed the discontinuous electronic structures and ultra-large bandgap of Cs₄PbX₆ due to its decentralized crystal structure, which is scarcely possible to be highly luminescent^{[62, 63]}. Moreover, the bandgap of CsPb₂Br₅ was measured to be 2.98 eV, thus revealing the theoretic weak PL emission^[64].

An interesting phenomenon is the easy phase transformation of these perovskite analogues between each other under various conditions, such as the extra addition^[62] or extraction^{[65, 66]} of precursors, ligand-mediation^{[63, 67, 68]}, light induction^[67] and so on. As to the transformation from Cs₄PbX₆ to CsPbX₃, the process can be both considered as the addition of PbX₂ or the extraction of CsX, thereby the Cs₄PbX₆ should be regarded as a PbX₂-deficient structure or a CsX-rich structure accordingly. Manna group tried to convert the non-luminescent Cs₄PbX₆ NCs to the luminescent CsPbX₃ NCs by adding extra PbX₂ into the pure Cs₄PbX₆ NCs synthesized in a Cs-rich environment^[62]. As is shown in Fig. 5(a), the distinctions of optical properties and morphologies after injection of PbX₂ can be clearly observed. However, Zhang group^[65] realized this phase transformation from the other perspective via an ingeniously designed experiment. In detail, a two-phase reaction system was established and the non-luminescent Cs₄PbX₆ NCs were dispersed in a nonpolar organic phase while the other phase was dispersed in water. Since the solubility of CsX salts in water was very high, the CsX composition in the Cs₄PbX₆ NCs was quickly stripped across the oil/water interface into the water phase (Fig. 5(b)). Consequently, the highly luminescent CsPbX₃ NCs layer gradually formed and expanded with the reaction time prolonging. More importantly, the as-obtained NCs were proved to be more stable than those CsPbBr₃ NCs prepared through the traditional hot-injection method, which could be attributed to the possible passivation during the water treatment.

Also by the Manna group^[66], the transformation from Cs₄PbX₆ to CsPbX₃ through the extraction of CsBr could also be achieved either by thermal annealing (physical approach) or by chemical reaction with Prussian Blue (chemical approach) (Fig. 5(c)). According to their results, the physical approach could be facilely carried out on a dried film without addition of any chemicals, while the reaction with Prussian Blue in solution could generate a full transformation into the CsPbX₃ phase.

Surface ligands also play a significant role in the phase tuning. For instance, the Alivisatos group^[63] invented a ligand-mediated method for phase transformation from CsPbBr₃ NCs to the Cs₄PbBr₆ derivative, which was governed by a two-step dissolution-recrystallization mechanism (Fig. 5(d)). In this process, the OAm was added to trigger the transformation reaction, and alkyl-thiol ligands acted as the modulator to improve the size uniformity and chemical stability of the Cs₄PbBr₆. Similarly, a reversible structural and compositional transformation between cubic CsPbX₃ NCs and rhombohedral Cs₄PbX₆ NCs was induced through the precisely controlling the ratio of OAm to OA ligands^[69] (Fig. 5(e)). As was revealed by the surface analysis, the ligand shell of CsPbBr₃ NCs consisted of bonded ammonium ligands and OA, while the ligand shell of Cs₄PbBr₆ was composed of both OAm and OA both in a bonded state. As a result, the reversible transformation could be tuned via the acid-base equilibrium. In contrast, the study on the transformation between CsPbX₃ NCs and CsPb₂X₅ NCs is relatively limited. Deng group^[70] reported a strategy for using alkyl-thiols to induce the transformation of CsPbBr₃ to CsPb₂Br₅ with controlled morphology by rational tuning the ratios of the alkyl-thiol ligands to alkyl-amines or to alkyl-acids (Fig. 6(a)). A novel reversible light-mediated compositional and structural transition process from orthorhombic CsPbBr₃ to tetragonal CsPb₂Br₅ nanosheets or vice versa was demonstrated in their following work (Fig. 6(b))^[67].

Pb is known as a highly toxic element which is one of the major barriers to hinder the commercialization of CsPbX₃ NCs. To solve this problem, several replacements, such as tin (Sn), manganese (Mn), germanium (Ge), bismuth (Bi) and antimony (Sb), have attracted great attention^{[68, 71-76]}. Cation exchange reaction can be an effective method to introduce these divalent elements^[77]. Among these environment-friendly substitutions, CsSnX₃ perovskites and their derivatives are regarded as the most promising materials. The lead-free CsSnX₃ NCs are very sensitive to moisture and air, because the Sn²⁺ ions are very easy to be oxidized to Sn⁴⁺. The Cs₂SnI₆ NCs were successfully synthesized by a modified hot-injection method^[78]. Prepared from SnX₄ precursors, Cs₂SnI₆ NCs were found to be more stable under ambient conditions than CsSnX₃ NCs, and perhaps potential materials for utilization in solar cells. So far, several strategies have been mainly reported to improve the stability of CsSnBr₃ NCs. Firstly, excess fluoric additives, such as SnF₂^[79] and perfluorooctanoic acid^[80], can effectively suppress the oxidation of Sn²⁺ because the F^- acting as a strong electron-withdrawing group has a stronger interaction with Sn²⁺. Besides, strong steric hindrance of branched ligands can also prevent the structure of the CsSnX₃ NCs from O₂ and H₂O attack^[81].

Briefly, various parameters such as reaction temperature, reaction time and surface ligands have significant effects on the shape, size and surface properties of all inorganic halide perovskite NCs, which should be taken into thorough consideration in the controllable synthesis. A series of investigations have been conducted in order to systematically clarify the sundry effects of reaction parameters on the shape control of CsPbX₃ NCs.

The NCs often have a high specific surface area, and the surface chemistry can remarkably affect the physical and chemical properties of NCs, which is often used as a fundamentally important modulator in control of their surface modification, shape, luminescent efficiency and stability. For the NCs capped with organic surfactants, the role of amino and carboxylic functional groups is indispensable for the successful controllable synthesis. When it comes to the effects of the organic surface ligands on the shape control or other oriented behavior of the CsPbX₃ NCs, the amount and carbon chain length as well as the molecular structure of the organic surfactants are commonly taken into consideration.

To get a thorough understanding of the effects of carbon chain lengths on the shape of the perovskite NCs, Pan et al.^[82] carried out a comprehensive study to probe the roles of different pairs of carboxylic acids and organic amines with varying chain lengths in the ligand-mediated synthesis of the all-inorganic perovskite NCs. According to their results, as is shown in Fig. 7(a), the organic amines played a more effective role in the shape control of the NCs, leading to the growth of anisotropic platelet structures, while the ammonium ions bound more weakly to the surface than the carboxylate ligands. The morphology of CsPbBr₃ NCs tended to become thinner nanoplatelets when shorter chain amines were used, while the larger nanocubes were achieved at high temperature reactions when shorter chain carboxylic acids were involved. This phenomenon probably resulted from the faster diffusion of short amines, leading to the faster exchange rates and anisotropic growth of the CsPbX₃ NCs. Additionally, the binding between the organic ligands and the surfaces may be weakened under high temperature, thus giving rise to larger CsPbX₃ NCs.

Another systematic study focusing on the ligand-mediated approach to control the shape of CsPbX₃ NCs at room temperature was conducted by the Deng group^[83]. They proposed a facile re-precipitation strategy to synthesize 0D spherical CsPbX₃ quantum dots, nanocubes, 1D nanorods, and 2D nanoplatelets, and these well-defined morphologies could be tuned by changing the chain lengths of organic acids and amines. The corresponding effects of different amino and carboxylic functional groups in the dimensionality of CsPbX₃ NCs are presented in Fig. 7(b). Thus, it can be seen that varying chain lengths and structures of organic surfactants can realize the anisotropic growth of CsPbX₃ NCs. Manna group^[84] created a facile method to tune the width of CsPbBr₃ NWs, from the non-confined regime (10 nm) to strong quantum-confinement regime (3.4 nm), only by introducing carboxylic acids, octanoic acid or hexanoic acid (Fig. 7(c)). These acid ligands with short aliphatic chains were believed to inhibit the growth of some crystallographic facets, while enhancing some other sets of them. Not only the chain length of organic acid, but also that of organic amines could affect the morphology of CsPbX₃. Yuan et al.^[85] selected several organic amines with different chain lengths, oleylamine, cetylamine and octylamine together with oleic acid respectively as surfactants to synthesize a series of CsPbX₃ NCs at different temperatures. It was found that the CsPbBr₃ NCs synthesized in the presence of short carbon-chain amine had a tendency of low-dimensional growth, leading to the formation of nanoplates and nanosheets at low temperature.

Apart from varying chain lengths and structures, the tuning of the ratio of organic acids to organic amines also plays an important role in the controllable synthesis of CsPbX₃ NCs. Deng group succeeded in producing low-dimensional nanosheets and NWs with high yields by simply tuning the ratios of organic amines to acids in the reaction process^[86]. The schematic illustration for the formation of perovskite NWs and nanosheets is given in Fig. 7(d). Meanwhile, it was also found that different morphologies could be better distinguished and selected while the bis(2-ethylhexyl)-amine (BEHA) participated in the reaction, and the morphology control may be attributed to different binding kinetics of the two capping agents on each crystal facet.

A similar report on the effects of different OAm to OA ratios on the morphology control was reported by Liang and co-workers as well^[87]. Accordingly, the shapes of the as-obtained CsPbBr₃ NCs with bright blue emission could be systematically engineered into single and lamellar-structured 0D quantum dots, as well as face-to-face stacking 2D nanoplatelets and flat-lying 2D nanosheets via tuning the ratios of OA to OAm. Consistent with other studies, the claimed internal mechanism was the growth rate of different crystal facets of CsPbBr₃ NCs due to the preferential crystal facets under the existence of different surfactants.

Reaction time is a crucial and effective parameter to regulate the growth and shape of NCs. As to the colloidal synthesis of CsPbX₃ NCs, the growth and formation usually take a quite short time even within several seconds^[24]. To study the evolution of the CsPbX₃ NCs with reaction time prolonging, the Yang group conducted a series of pioneering work^[88]. They carefully investigated the formation dynamics of CsPbX₃ NCs by reacting cesium oleate with PbX₂ in a hot solvent of ODE with OAm and OA at 150–250 °C and analyzed the intermediates at different reaction times. The formation sketch at 150 °C is illustrated in Fig. 8(a). They found that the initial products (when t < 10 min) were mainly composed of the CsPbX ₃ nanocubes with size ranging from 3 to 7 nm. After 10 min, a few thin NWs with diameters around 9 nm were found in the products. With the reaction time increasing, more NWs were formed but the amount of the nanocubes was decreased, while some large square-shaped nanosheeet appeared. At a later formation stage (40–60 min) of the reaction, the nanosheets dissolved, and the products were dominated by NWs with diameters uniformly below 12 nm and lengths up to 5 μm. The uniform NWs stably existed until the reaction time exceeding 90 min and the main products of irregular large particles were finally found. Noticeably, these intermediate morphologies coexisted with each other instead of independent presence. This observation distinctly revealed the impacts of reaction time on the morphology evolution of CsPbX₃ NCs. The similar growth process was also observed in the lead-free inorganic perovskite analogue NCs. For the first time, Deng group^[78] successfully synthesized different-shaped Cs₂SnI₆ NCs by a modified traditional hot-injection protocol under argon atmosphere. After the injection of Cs-precursor within 1 min, most of the samples were found to be spherical quantum dots with an average diameter of 2.5 nm. Further increasing the reaction time led to the formation of nanorods, which were dominated over the products. Then the nanorods progressively evolved into longer NWs. With reaction time reaching 30 min, the NWs assembled side-by-side to form nanobelts, which were finally transformed into nanoplates with a thickness of around 8 nm. The aforementioned results indicated that low-dimensional nanostructures such as quantum dot, nanocubes were the preferable products in the shorter reaction time. With the extension of the reaction time, the dominant morphologies tended to be 1D NWs or nanorods and 2D nanoplates and nanosheets.

For the traditional hot-injection method, the reaction temperature is one of the key parameters in controllable synthesis of CsPbX₃ NCs^[89]. Alivisatos group reported a controllable synthesis of CsPbX₃ NCs, and the shape could be tuned from nanoplatelets to nanocubes via changing the reaction temperature from 90 to 150 °C (Fig. 8(b))^[90]. In this work, green-emitting CsPbBr₃ nanocubes were produced under a higher reaction temperature of 150 °C. In contrast, cyan-emitting nanoplatelets and blue-emitting nanoplatelets were obtained at lower temperatures of 130 and 90 °C, respectively. This result revealed that the lower temperature had a favorable impact on the formation of anisotropic nanostructures. Similar results are also observed in other reports^{[85, 88]}.

Generally speaking, the general morphologies of inorganic halide perovskites NCs can be divided into 0D nanospheres (quantum dots) and nanocubes, 1D NWs and nanorods, plus 2D nanoplates and nanosheets. Since the shapes of NCs are in close association with the further utilization in different optoelectronic fields, the dimension-controllable synthesis of inorganic perovskite NCs has been a hot spot.

The 0D nanocubes and nanospheres are the most commonly achieved morphologies through a traditional hot-injection method with longer organic ligands at a relatively high temperature for short reaction time. The size-control of nanocubes can be easily realized by tuning the OAm/OA ratios^{[86, 87]}, reaction temperatures^{[85, 88, 90]} and the concentration of precursors^[26]. As for nanospheres, similar synthetic methods can also be employed. CsPbBr₃ nanospheres with an average diameter of 2.4 nm were firstly reported through the hot-injection method with a OA to OAm ratio of 0.6 : 0.3 at 90 °C^[87]. The ultra-small CsPbX₃ nanospheres with a nearly amorphous structure can also be obtained at very low reaction temperatures (0 °C)^[36]. Organic surfactants with shorter chain lengths have been successfully used to produce CsPbX₃ nanospheres with an average diameter of 4.3 nm as well^[83].

1D NWs and nanorods are widely used in some specific optoelectronic applications, such as photodetectors and lasers. In contrast to the 0D nanospheres and nanocubes, the formation of 1D NWs usually takes a longer reaction time. The pioneering work of 1D CsPbX₃ NWs was reported by the Yang group^[88], and the 1D NWs were obtained by prolonging reaction time up to 1 h, in which the morphology of the NCs changed from nanocubes to large nanoparticles. Fine-tuning capping ligands could also assist the formation of anisotropic NWs by selective binding to certain crystal facets^[84]. Single-crystalline CsPbBr₃ NWs were able to directly form precursor powders with controllable length and width by regulating the reaction time^[91]. Additionally, extra adding of hydrohalic acids (HX, X = Cl, Br, I) could affect the length of CsPbX₃ NWs^[92]. On the other hand, however, studies on CsPbX₃ nanorods are still very limited. The CsPbX₃ nanorods with the length of 200 nm were firstly synthesized via a ligand-mediated process using acetate acid and dodecylamine^[83]. Another example of successful production of the uniform rod-shaped structure was hybrid halide CsPb(Br/I)₃ nanorods through a modified hot-injection method^[93].

The 2D CsPbX₃ nanoplates and nanosheets have attracted much attention due to their unique structures and outstanding properties as future promising candidates for optoelectronic devices. In general, lower reaction temperatures and hybrid short and long ligands tend to generate the formation of nanoplates and nanosheets. This discovery has been firmly ensured by many groups^{[82, 85, 90, 94, 95]}. Besides, the polarity of solvents is found to play a significant role in the morphology evolution from nanocubes to nanoplates, and further to nanobars^{[96, 97]}. What’s more, the introduction of additional metal halides as models to induce the self-assembly of nanorods into nanoplates has been reported as well^[98].

One of the most challenging obstacles to the practical application of perovskite NCs lies in their inherent poor photo- and thermo-stability, and they have been proven to easily degrade in the presence of oxygen and moisture. Compared with hybrid organic-inorganic halide perovskites, all-inorganic CsPbX₃ NCs are endowed with relatively better stability. However, the intrinsic ionic character still hinders their further utilization in many fields. Hence, to overcome these stability problems, many strategies have been developed to passivate the unstable surfaces, mainly including surface modification by organic ligands or inorganic chemicals and encapsulation in some stable shells or water-proof materials.

The traditionally used organic amines and organic acids can not only control over the morphology, but also protect the surface and enhance the stability. Except for the most used OAm and OA, sorts of branched capping ligands have been employed due to their strong steric effect. For example, trioctylphosphine (TOP)^[99], trioctylphosphine oxide (TOPO)^[100], bis-(2,2,4-trimethylpentyl) phosphinic acid (TMPPA)^[101], di-dodecyl dimethyl ammonium bromide (DDAB)^[102] along with hybrid S^2- and DDA⁺ ion pair^[103] and so on, have been introduced to passivate the surface of CsPbX₃ NCs either by in-situ capping or post-synthetic ligand exchange methods, which facilitate the stability in air and polar solvents, as well as enhancing the optical properties. Moreover, the organic surfactants can be in-situ carbonized on the surface via X-ray treatment, resulting in intensive protection against easy removal and dissolution^[104]. The improvement of stability can be realized by the introduction of inorganic chemicals coating on the surface. According to previous reports, the PbBr₂-rich surface favored the increase of PLQY and improvement of stability, thus the post-treatment process was conducted by using PbBr₂^{[105, 106]}, ZnBr₂^{[107, 108]} and HBr^[109] etc. as resources. In addition, inorganic anions K^+[110] and Ag^+[111] were also introduced to passivate surfaces.

Apart from the surface treatment with organic or inorganic surfactants, the encapsulation of CsPbX₃ NCs into more stable materials is another feasible strategy to improve the stability. Silicon is commonly used as protective coating shells: mesoporous silica was used as a template to assist the formation of CsPbX₃ NCs, and the as-synthesized CsPbX₃ powders exhibited high PL efficiencies along with improved stability^[40]. Effective CsPbX₃ NCs/SiO₂ nanocomposites were obtained by directly mixing the CsPbX₃ NCs with mesoporous SiO₂ with pore sizes of 12–15 nm^[112]. The SiO₂ coated CsPbX₃ composites were also achieved through in-situ decoration of (3-aminopropyl)triethoxysilane (APTES)^[113] and NH₂-polyhedral oligomeric silsesquioxane (NH₂-POSS)^[114]. Moreover, another facile sol-gel method was also utilized to prepare the monodisperse CsPbX₃/SiO₂ core/shell nanostructures. What’s more, other oxides or sulfides, including Al₂O₃^[115], TiO₂^[116], MoS₂^[117] and ZnS^[118], were also utilized in the encapsulation process. Furthermore, the versatile carbon materials, such as graphene^[119] and nanotubes^[120] were both promising modified candidates in further application. Polymer coating shells were also regarded as efficient stabilization way due to their outstanding hydrophobic features. Polystyrene (PS) beads^[121] and fibers^[122] were applied as capsule shells against exposure to water and thermal treatment. Besides, poly(methyl methacrylate) (PMMA)^[123] and polyvinyl pyrrolidone (PVP)^[124] encapsulation were also utilized in LEDs devices, and poly(maleic anhydride-alt-1-octadecene) (PMA) capping agent was also proved to efficiently prevent CsPbX₃ from aggregation by tightening the superficial ligand bindings^{[125, 126]}.

The all-inorganic metal halide perovskite is no doubt one of the most promising materials emerging in recent years, and it can be clearly foreseen that this booming material is still equipped with great development potentials. The highly ionic features endow CsPbX₃ NCs with defect-tolerance, which consequently results in high PLQYs and narrow full width at half maximum (FWHM)^{[127, 128]}. Additionally, their PL emission can be tuned simply by changing the halide composition and their bandgap is insensitive to the temperature, quite different from traditional metal chalcogenide and pnictide NCs^{[26, 129]}. Thanks to all these excellent optoelectronic properties, CsPbX₃ NCs are very perspective in various applications including solar cells^[130-134], light-emitting diodes (LEDs)^{[119, 135-137]}, photodetectors (PDs)^[138-141], lasers^{[29, 142-144]}, photocatalysis^{[119, 145, 146]} and field-effect transistors (FETs)^[147-149].

Perovskite solar cells, especially organic-inorganic hybrid ones are widely investigated and the power conversion efficiency (PCE) has reached 23.3% in 2019^[150]. Compared with hybrid solar cells, which suffer from the problem of instability, CsPbX₃ NCs are more thermally stable and temperature insensitive. The cubic phase of CsPbX₃ NCs is crucial to acquire suitable band gap for achieving high PCE^{[133, 134]}. Researchers have discovered that introducing bromide into crystal lattice or introducing some surfactants to the surface by post-treatment will help stabilize the cubic phase of CsPbI₃^{[151, 152]}. The optimal PCEs obtained through the aforementioned two methods could reach 9.8% and 13.4%, respectively. Besides, to form a thin and compact inorganic lead oxysalt layer on the perovskite surface is also an efficient general strategy to enhance the PCE efficiency^[153]. A higher PCE of 14.5% for CsPbX₃ was obtained via compressing CsPbI₃/CsPbI₂Br bilayers together to form a tandem solar cell^[154], as is shown in Fig. 9. Very recently, a tetragonal β-phase of CsPbI₃ was reported, which exhibited higher stability and a more favorable bandgap^[155]. This material allowed for PCEs of 18% after the passivation of the surface trap state with choline iodide. Although PCEs of CsPbX₃ are still far away from that of hybrid solar cells, the rapid growing efficiency and environmental stability clarify the bright prospects of CsPbX₃ NCs as the next-generation solar cells materials.

Taking advantage of their wide and tunable emission wavelength, CsPbX₃ NCs have drawn great attention in the applications of lightening and display as backlight LEDs. Many efforts have been made from three aspects including carrier injection efficiency, radiative recombination efficiency and injection balance to improve performance and stability of LEDs^{[56, 102, 156]}. The first electrically driven CsPbX₃ NCs based LED was reported by the Zeng group (Fig. 10)^[157]. The external quantum efficiency (EQE) increases from less than one 1% to 16.48% within three years^[147], exhibiting an ultrafast progressing speed. To the best of our knowledge, currently, the highest EQE of inorganic perovskite quantum-dot LEDs involved of CsPbX₃ NCs has exceeded 20% by mixing presynthesized CsPbBr₃ perovskite NCs with MABr additives as the passivation layer^[148]. In spite of remarkable development achieved, the practical application of the CsPbX₃ NCs-based LEDs is still challenging as EQE of blue LED is just around 0.1%, which is far away from the prerequisite to produce white LEDs^{[94, 109]}.

In a reverse process, the PDs should extract photo-generated carriers quickly and efficiently to ensure ultrafast response and high turn/off ratio. The high absorption efficiency (2 × 10⁵ cm^–1)^[46] and large carrier mobility^[109] enable CsPbX₃ NCs a potential candidate for light detection. Aligned films of CsPbI₃ NCs displayed excellent photosensitivity (on/off ratio of 105) and fast response speed (rise time of 24 ms and decay time of 29 ms)^[139]. Besides, the 2D CsPbBr₃ NCs are preferable for fabricating large-area and crack-free planar films^[149]. The ultrathin and flexible PDs showed a detectable limit of 21.5 pW/cm² and a fast response time of 20 ns, implying its promising application in flexible devices (Fig. 11(a))^[140].

Very recently, the CsPbX₃ NCs were found to be efficient catalysts for photocatalytic CO₂ reduction. CsPbBr₃ was reported to be utilized in the CO₂ reduction to produce the mono-carbon product with nearly one-unit selectivity (Fig. 11(b))^{[119, 145]}. It is undoubtedly that high absorption coefficient, high charge carrier mobility and long charge diffusion length of CsPbX₃ are very beneficial for light-harvesting and charge migration to the active center. Problems left are the instability of CsPbX₃ in aqueous solution and the use of highly toxic element lead in CsPbX₃ NCs. Further efforts should be devoted to developing environmental benignly IHP NCs for photocatalytic applications^[146].

The unique optical features of high PLQYs, tunable emission and temperature insensitive also render CsPbX₃ NCs as the ideal gain media for constructing laser devices via coupling with another optical cavity^[29]. The best CsPbX₃ NCs-base lasing devices displayed a threshold as low as 2.1 μJ/cm² with multicolor (400–700 nm) (Fig. 11(c))^[144]. Low threshold, high-quality factor and adjustable multicolor make CsPbX₃ NCs hopeful gain media in lasing devices.

Except for the above optoelectronic applications, the CsPbX₃ NCs can also be utilized in field-effect transistors (FETs) due to their large charge carrier mobility^[158] and ambipolar conductivity^[159]. Moreover, quite different from the traditional silicon semiconductors, perovskite thin films are flexible, endowing it with great potential in stretchable and wearable electronics. Besides, combining CsPbBr₃ NCs and 2D semiconductor materials, such as MoS₂, to form heterojunctions is also a potential excellent strategy to design high-performance phototransistors (Fig. 11(d))^[160]. An obstacle that prohibits wide investigation of perovskite-based FETs lies in the gate-field screening effect resulting from ion migration^[161]. Various methods including optimizing perovskite microstructures, using less polar cations in perovskites, or using external light excitation for ion migration compensation have been employed to solve the problems^[162].

In this review, recent advances on representative fabrication methods of all-inorganic halide perovskite NCs and their related applications have been summarized, especially focusing on the controllable synthesis methods and parameter regulation. Despite remarkable achievements obtained in the past four years, there still exist many issues to be addressed and plenty of challenges to be defeated before realizing the practical application of CsPbX₃ NCs.

The most intractable problem is their poor stability against moisture, oxygen, heat and persistent light irradiation. The ionic nature of CsPbX₃ NCs decides that they are very sensitive to the polar solvents such as water, consequently restrict their applications in many areas. Continuous heat or light irradiation would cause detachment of surface ligands at elevated temperatures, which are crucial for protecting CsPbX₃ NCs from aggregation and the formation of carrier traps. Common methods used to address these two issues are surface coating, ligand treatment, and the introduction of a host matrix. Inert shells such as aluminum oxide and silica have been found to effectively improve resistance of CsPbX₃ NCs to moisture and light irradiation^[115]. Carbonized surface ligands created by low-flux X-ray irradiation are also reported to effectively protect the CsPbX₃ from water and air^[104]. The implant of CsPbX₃ onto silica or into polymer will prohibit the NCs from aggregation thus improving their stability^[121–124]. Although the aforementioned techniques have been developed and great progress has been made, it is still challenging to stabilize the CsPbX₃ NCs without affecting their optoelectronic properties. Moreover, much attention has been made to stabilizing the green-emitting CsPbBr₃ NCs. Other CsPbX₃ NCs are waiting to be investigated and protected.

It is well known that lead in CsPbX₃ NCs is toxic to health, which prevents CsPbX₃ NCs from practical application and commercialization. Some eco-friendly elements including tin (Sn), bismuth (Bi), germanium (Ge), indium (In), and antimony (Sb) have been investigated as alternatives to lead^{[163, 164]}. However, these lead-free perovskite NCs display much lower PLQY (less than 1%) compared with CsPbX₃ NCs, which make them unfavorable for optoelectronic applications. Furthermore, these substitutions are less stable and can be easily oxidized in the air^[78]. More efforts should be devoted to developing lead-free all-inorganic perovskite NCs with satisfactory optoelectronic properties.

It is worth noting that the commonly used core-shell structure in metal-chalcogenide quantum dots is not feasible for all-inorganic perovskite NCs because of the spontaneous anion-exchange reactions. However, coating the all-inorganic perovskite NCs with traditional metal-chalcogenides to form a core-shell heterostructure not only can serve as protection layer against air and moisture, but also can tune their optoelectronic properties for various applications. On the other hand, the adoption of this core-shell heterostructure owns the potential to radically circumvent fast Auger recombination, which will undoubtedly improve the efficiency of all-inorganic perovskite NCs in high flux applications^[165].

In conclusion, more efforts should be contributed to explore effective methods for all-inorganic perovskite NCs stabilization while preserving their optoelectronic property. The exploitation of lead-free all-inorganic halide perovskite NCs with comparable efficiency is also highly urgent for their commercialization. The creation of some core-shell heterostructures coupled with traditional metal chalcogenides is a promising approach for developing high flux all-inorganic halide perovskite NCs. We believe that the all-inorganic halide perovskite NCs will emerge as a new class of fascinating optoelectronic materials and hold a bright future for various applications.

This work was supported by the Fundamental Research Funds for the Central Universities (No.2019RC020).