Metamorphoses of Cesium Lead Halide Nanocrystals

Conspectus Following the impressive development of bulk lead-based perovskite photovoltaics, the “perovskite fever” did not spare nanochemistry. In just a few years, colloidal cesium lead halide perovskite nanocrystals have conquered researchers worldwide with their easy synthesis and color-pure photoluminescence. These nanomaterials promise cheap solution-processed lasers, scintillators, and light-emitting diodes of record brightness and efficiency. However, that promise is threatened by poor stability and unwanted reactivity issues, throwing down the gauntlet to chemists. More generally, Cs–Pb–X nanocrystals have opened an exciting chapter in the chemistry of colloidal nanocrystals, because their ionic nature and broad diversity have challenged many paradigms established by nanocrystals of long-studied metal chalcogenides, pnictides, and oxides. The chemistry of colloidal Cs–Pb–X nanocrystals is synonymous with change: these materials demonstrate an intricate pattern of shapes and compositions and readily transform under physical stimuli or the action of chemical agents. In this Account, we walk through four types of Cs–Pb–X nanocrystal metamorphoses: change of structure, color, shape, and surface. These transformations are often interconnected; for example, a change in shape may also entail a change of color. The ionic bonding, high anion mobility due to vacancies, and preservation of cationic substructure in the Cs–Pb–X compounds enable fast anion exchange reactions, allowing the precise control of the halide composition of nanocrystals of perovskites and related compounds (e.g., CsPbCl3 ⇄ CsPbBr3 ⇄ CsPbI3 and Cs4PbCl6 ⇄ Cs4PbBr6 ⇄ Cs4PbI6) and tuning of their absorption edge and bright photoluminescence across the visible spectrum. Ion exchanges, however, are just one aspect of a richer chemistry. Cs–Pb–X nanocrystals are able to capture or release (in short, trade) ions or even neutral species from or to the surrounding environment, causing major changes to their structure and properties. The trade of neutral PbX2 units allows Cs–Pb–X nanocrystals to cross the boundaries among four different types of compounds: 4CsX + PbX2 ⇄ Cs4PbBr6 + 3PbX2 ⇄ 4CsPbBr3 + PbX2 ⇄ 4CsPb2X5. These reactions do not occur at random, because the reactant and product nanocrystals are connected by the Cs+ cation substructure preservation principle, stating that ion trade reactions can transform one compound into another by means of distorting, expanding, or contracting their shared Cs+ cation substructure. The nanocrystal surface is a boundary between the core and the surrounding environment of Cs–Pb–X nanocrystals. The surface influences nanocrystal stability, optical properties, and shape. For these reasons, the dynamic surface of Cs–Pb–X nanocrystals has been studied in detail, especially in CsPbX3 perovskites. Two takeaways have emerged from these studies. First, the competition between primary alkylammonium and cesium cations for the surface sites during the CsPbX3 nanocrystal nucleation and growth governs the cube/plate shape equilibrium. Short-chain acids and branched amines influence that equilibrium and enable shape-shifting synthesis of pure CsPbX3 cubes, nanoplatelets, nanosheets, or nanowires. Second, quaternary ammonium halides are emerging as superior ligands that extend the shelf life of Cs–Pb–X colloidal nanomaterials, boost their photoluminescence quantum yield, and prevent foreign ions from escaping the nanocrystals. That is accomplished by combining reduced ligand solubility, due to the branched organic ammonium cation, with the surface-healing capabilities of the halide counterions, which are small Lewis bases.

Change of Surface: This work shows how a single surfactant quaternary ammonium salt has all the ingredients for a tight surface passivation that boosts the photoluminescence quantum yield (PLQY) to near unity and makes perovskite nanocrystals stable in dispersion even at elevated temperatures.

■ INTRODUCTION
Cesium lead halide nanocrystals, especially the perovskite ones, have been intensively investigated in the last years thanks to their simple synthesis and appealing optical properties, above all their efficient and spectrally narrow photoluminescence (PL). 5,6 Such properties make this class of materials promising as low-cost optoelectronics components. However, it appeared from the beginning that these materials, particularly in the form of colloidal nanocrystals, suffer from poor stability as they are very reactive toward their surroundings. While this aspect is detrimental for many practical applications, from a chemist's viewpoint it offers the opportunity to investigate, master, and exploit their various possible transformations: these can be structural 1 or compositional, 2 and they can affect the surface 4 Figure 1. Nanocrystals of different compounds within the Cs−Pb−Br system (a). Side-view of the Cs + cation substructure shared among the same compounds (b). Thanks to this common structural feature, nanocrystals can be converted one into another by ion trade reactions exchanging PbX 2 .
Since ion trade reactions preserve the nanocrystal backbone, their intermediates are epitaxial heterostructures (c). Transformations between Cs− Pb−Br compounds cause instability issues but provide opportunities for applications as well. One example is the templated conversion of a film of CsPbBr 3 emissive nanocrystals into nonemissive Cs 4 PbBr 6 nanocrystals upon exposure to butylamine vapors, which can be reverted by mild heating (d ■ CHANGING STRUCTURE BY ION TRADE REACTIONS At first sight, the reactivity of Cs−Pb−X (X = Cl, Br, I, Figure  1a) nanocrystals appears similar to that of binary metal chalcogenide, ME (M = Cd, Pb; E = S, Se, Te), nanocrystals: both classes of materials can exchange ions with their surroundings and modify their composition, with some of the ions migrating in and out of the nanocrystals, while others provide a sturdy backbone to the structure during the process. 7,8 In chalcogenide nanocrystals, anions are bigger than cations and constitute the stable network inside which the small cations migrate. 7 Conversely, the research into lead halide nanostructures revealed that halide vacancies, deformability of the perovskite crystal structure, and lower free energy barriers for vacancy-mediated ionic diffusion underlie the higher mobility of halide anions as compared to Cs + cations, despite their similar sizes. 9,10 The reactivity landscape of Cs−Pb−X nanocrystals is much wider than that of II−VI or IV−VI chalcogenide semiconductor nanocrystals. First, the mobility of ions in Cs−Pb− X compounds is higher than in conventional semiconductors. This stems from the higher ionicity of the metal−halide bonds as compared to metal−chalcogenide ones. For example, the ionicity (difference in Pauling electronegativity) of Cs−X and Pb−X bonds falls in the range of ∼1.9−2.4 and ∼0.8−1.3, respectively, while for (In/Pb/Cd/Zn/Ag/Cu)−(S/Se/Te) pairs the range is ∼0.2−0.9. The higher ionicity of bonds lowers the activation energy for ion migration within the structure, leading to higher reactivity. 11,12 Furthermore, those in the Cs−Pb−X systems are ternary compounds, and this additional level of structural complexity is game-changing. While binary chalcogenide semiconductors are limited to exchanging anions or cations, Cs−Pb−X nanocrystals can also capture or release nominally neutral formula units like PbX 2 and CsX, thus undergoing major changes in their structure and stoichiometry. This makes the concept of ion exchange too limited to adequately capture their reactivity. Instead, Cs−Pb− X nanocrystals are capable of what we will call ion trade reactions, that is, reactions in which a nanocrystal releases or captures species from the environment with a net flux of atoms, while retaining structural relationships between reactant and product nanocrystals. Consistently with this vision, chemical transformations among many Cs−Pb−X compounds have been rationalized with a Cs + cubic substructure capturing or releasing PbX 2 units (Figure 1b). 10 The Cs + substructure undergoes distortions as it adapts to accept or dispatch ions, without suffering any major changes. A proof is the observation of epitaxial heterostructures between reagent and product compounds, where the Cs + sublattice went uninterrupted across the junction. Reports for the CsX → γ-CsPbX 3 , 13 the Cs 4 PbX 6 → γ-CsPbX 3 , 1 and the γ-CsPbBr 3 → CsPb 2 Br 5 transformations 14 encompassed all the ternary stoichiometries within the Cs−Pb−X system (Figure 1c). It is worth noting that the stoichiometry changes following these ion trade reactions heavily affect the electronic properties of nanocryst-als: mild reaction conditions are enough to turn insulators (CsX and Cs 4 PbX 6 ) into direct-bandgap (CsPbX 3 ) or indirectbandgap (CsPb 2 X 5 ) semiconductors.
The concepts of ion trade reaction and Cs + substructure preservation give us hints on how a transformation takes place, but they leave aside the reasons for why it takes place. Reactions causing a stoichiometry change necessarily involve the trade of neutral species (CsX or PbX 2 ); thus they are driven by unbalances in the partition equilibrium of those species between the nanocrystal and its surrounding environment. The simplest case is when the species are directly added or removed from the chemical environment. Examples are reactions driven by the addition of lead-rich compounds such as PbX 2 or Pb-oleate, 13,15 but also the Cs 4 PbBr 6 → CsPbBr 3 transformation triggered by the sequestration of Cs + by Prussian Blue 16 and the concomitant release of Br − to maintain charge neutrality (= CsBr subtraction). Another driving force is the solubility of neutral species, which can interfere with partition equilibria as well. Almeida et al. rationalized the solubility of PbX 2 in nonpolar solvents as being dependent on the concentrations of [R-NH 3 ] + and [R-COO] − ions, which in turn depend on ligands concentration and temperature. 17 Higher temperatures shift the acid−base equilibrium R-NH 2 + R-COOH ⇄ [R-NH 3 ] + + [R-COO] − toward the reagents, justifying the inverse solubility of PbX 2 in the reaction. They also demonstrated that conditions favoring high solubility of PbX 2 , namely, high ligand concentrations and low temperatures, promote the synthesis of the lead-deficient phase Cs 4 PbBr 6 . Opposite conditions favor instead the formation of CsPbBr 3 . Although this study was conducted on direct syntheses, the same principles apply to postsynthetic transformations. One example is the CsPbBr 3 → Cs 4 PbBr 6 transformation caused by the addition of amines, which increase the solubility of PbBr 2 and extract it from the nanocrystals. 18 This transformation also affects solid films when they are exposed to butylamine vapors, and can be used to prepare patterned films of both emissive CsPbBr 3 and nonemissive Cs 4 PbBr 6 nanocrystals ( Figure 1d): this was achieved by irradiating regions of the film with X-rays, which partly cross-linked the organic ligands coating the nanocrystals, creating a barrier against the diffusion of butylamine.
Thermal annealing is another way of triggering ion trade reactions. Palazon et al. demonstrated that heating films of Cs 4 PbBr 6 or CsPbBr 3 nanocrystals leads to the in situ formation of traces of more lead-rich compounds, CsPbBr 3 or CsPb 2 Br 5 , respectively. 16,19 Both transformations were rationalized by a ligand-mediated extraction of CsBr, that etches the surface of nanocrystals with the concomitant release of PbBr 2 . This excess of PbBr 2 then intercalates inside the surrounding grains through a solid-state ion trade reaction, driving their transformation to a more lead-rich stoichiometry: Cs 4 PbBr 6 turns into CsPbBr 3 , while CsPbBr 3 turns into CsPb 2 Br 5 . These lead-rich phases disappeared in both samples upon annealing above 300°C. A common explanation can be found in the temperature-driven expulsion of PbBr 2 from the crystal lattice, again a solid-state ion trade reaction, which might be related to PbBr 2 approaching its melting temperature (373°C, but lowered in the presence of other species).

■ CHANGING COLOR BY ION EXCHANGE
So far, we focused on ion trade reactions that induce a change in the nanocrystal stoichiometry. The more familiar ion exchange reactions are a special case of the ion trade class, Accounts of Chemical Research pubs.acs.org/accounts Article where the net flux of ions between the nanocrystal and its surroundings is null and the stoichiometry and structure remain unchanged. This is a common option for traditional binary chalcogenide semiconductors, and it is available for compounds within the Cs−Pb−X system as well. Reports on anion exchange reactions date back to the first colloidal syntheses of CsPbX 3 perovskite nanocrystals by hot injection. 2,21,22 For example, Akkerman et al. achieved fast and complete anion exchange of CsPbBr 3 to CsPbCl 3 and CsPbI 3 nanocrystals (Figure 2a−d) and also provided an early evidence of Cs + to methylammonium cation exchange when treating the nanocrystals with methylammonium halides. 2 The anion exchange reactions were found to proceed with both homogeneous (ammonium halides dissolved in toluene) and heterogeneous precursors (powdered PbX 2 ) and even between pairs of nanocrystals, that is, CsPbBr 3(NC) + CsPb(Cl or I) 3(NC) → CsPb(Br:Cl or Br:I) 3(NC) . Due to the high anion mobility, these reactions easily reached completion and, thanks to the cation substructure preservation, maintained the nanocrystal shape, size, and size distribution. Different from the more drastic reactions involving PbX 2 trade, the preservation of crystal structure and general stoichiometry resulted in a continuous fine-tuning of the optoelectronic properties: on CsPbX 3 nanocrystals, ion exchange reactions allowed tuning of the spectral position of the PL anywhere from 3.18 eV (CsPbCl 3 ) to 1.87 eV (CsPbI 3 ) without significant increase in the spectral width as compared to the starting CsPbBr 3 nanocrystals (Figure 2a). Moreover, Akkerman et al. noted that, starting from CsPbBr 3 nanocrystals, the PLQY after exchange to CsPbCl 3 or CsPbI 3 dropped to values that were in line with those of as-prepared and CsPbCl 3 or CsPbI 3 nanocrystals, which are generally less bright than CsPbBr 3 nanocrystals. 2 In addition, Mishra et al., when performing a CsPbBr 3 → CsPbCl 3 anion exchange on nanocrystals and then the inverse CsPbCl 3 → CsPbBr 3 reaction, 23 noted that the PLQY of the final CsPbBr 3 nanocrystals was even higher than that of the initial sample. That result suggests that the creation of additional defects is unlikely. The increase in PLQY at the Accounts of Chemical Research pubs.acs.org/accounts Article end of the cycle can be ascribed to a more efficient saturation of the Br − vacancies on the surface of the final CsPbBr 3 nanocrystals compared to the starting ones. Overall, anion exchange reactions do not appear to entail a significant formation of new defects, although further investigations on the topic are needed. The ability to tune the optoelectronic properties by anion exchange reactions without compromising the morphology and stability of nanocrystals has found many applications to date. For example, Palazon et al. 24 demonstrated a photolithographic approach to produce patterned CsPbBr 3 + CsPb-(Cl:Br) 3 nanocrystal films by masked exposure to HCl vapors. The templating was achieved by cross-linking ligands under Xray exposure and enabled the preparation of films with regions emitting in different colors. Brennan et al. 25 exploited the morphology-preserving features of anion exchange to prepare mixed-halide CsPb(Br x I 1−x ) 3 nanocrystals with narrow size distribution for self-assembly, starting from monodisperse CsPbBr 3 nanocrystals. A series of CsPb(Br x I 1−x ) 3 nanocrystal superlattices with PL tunable from green to red was thus prepared (Figure 2e−h) and their stability under UV illumination was tested. The photoannealing of mixed-halide NC superlattices with a 385 or 470 nm LED light at an incident fluence of I exc ≈ 100 mW/cm 2 resulted in iodide expulsion and reconversion to CsPbBr 3 nanocrystals (albeit with a photobrightening), a process enforced by the iodide photooxidation and I 2 sublimation. Throughout the photoinduced transformation back to CsPbBr 3 , the nanocrystals preserved their sizes and shapes, both in solution and in superlattices. That observation agrees with the cationic substructure preservation principle.
Anion exchange is not limited to perovskite CsPbX 3 nanocrystals. A later work by Akkerman et al. 15 reported anion exchange in the wide-gap Cs 4 PbX 6 nanocrystals. Albeit less colorful, the tunability absorption spectrum of Cs 4 PbX 6 nanocrystals was demonstrated by tuning the sharp absorption peak from 284 nm (pure Cs 4 PbCl 6 ) to 367 nm (pure Cs 4 PbI 6 ). 15 Here, however, the absorption spectrum for the mixed halide compositions was broader than that for the pure halide compositions. This depends on the molecular-like nature of transitions within individual octahedra. As these octahedra are disconnected in the Cs 4 PbX 6 structure, for any mixed composition the absorption spectrum is the convolution of the several optical transitions in the populations of octahedra, differing from one another by both the nature of halide ions surrounding the central Pb 2+ ion and their mutual spatial arrangement in the coordination environment.
Cation exchange on CsPbX 3 nanocrystals has been reported as well, both on the Cs + sites and on the Pb 2+ sites. Cesium can be exchanged with organic cations (typically methylammonium and formamidinium). These exchanges are mostly studied to stabilize the otherwise unstable black CsPbI 3 perovskite phase, which shows remarkable photovoltaic performance, 26 and to push their PL further into nearinfrared. 27 The case of Pb 2+ → M 2+/3+ exchange is yet another possibility: postsynthetic exchanges with divalent cations (M = Mn 2+ , Zn 2+ , Cd 2+ , Sn 2+ ), as well as trivalent ones (Bi 3+ , Ce 3+ , in these cases at the doping level), have been explored. 28 However, in most cases the process was incomplete, and was generally much slower than anion exchange reactions. Van der Stam et al. 29 rationalized those limitations by a combination of weak driving forces leading to self-limited reactivity and cation diffusion limited kinetics. To overcome this last limitation, they proposed metal halides as precursors, because they dissolve as undissociated MX 2 molecules in nonpolar solvents. In the proposed mechanism, X binds to superficial halide vacancies, securing the cation to the nanocrystal. In addition, the binding energy is enough to break a Pb−X bond, allowing fast Pb 2+ → M 2+ replacement. This anion-assisted cation exchange has been exploited on other occasions, 30 one remarkable example being the complete CsPbBr 3 → CsSnI 3 transformation. 31 This interesting case cannot be described by a simple ion exchange and instead fits into the more general category of ion trade reactions. Reacting CsPbBr 3 with SnBr 2 did not result in any cation exchange: 31 instead, the synergetic replacement of Br for I provided the driving force needed to fully swap lead for tin in the structure. This is a three-player process, where SnI 2 reacted as a neutral species and caused the expulsion of PbBr 2 , while the Cs + cations provided a stable backbone for the structure.

■ CHANGING COLOR BY SHAPE
Modifying quantum confinement by changing the shape and size of the CsPbX 3 nanocrystals allows tuning of their optical properties without changing the nanocrystal composition. This is usually achieved by directly synthesizing cubes with tunable edge lengths, 3,32 nanoplatelets and nanosheets of discrete thicknesses, 33,34 or nanowires. 35 Shape control of CsPbBr 3 nanocrystals, and in particular the balance between cube and platelet shapes, again depends on the concentration-and temperature-dependent acid−base equilibrium determined by the balance between [R-NH 3 ] + and [R-COO] − species, 17 similarly to the PbX 2 solubility discussed earlier. High acid concentration or low temperature shift the equilibrium toward the oleylamine protonation. In turn, oleylammonium ions, now in high concentrations, start successfully competing with Cs + ions on the surface of the growing nanocrystals, a condition that appears to promote the formation of platelets. Lower acid concentrations or higher temperatures instead shift the equilibrium toward unprotonated amines, promoting the formation of cubes. However, the exact reason behind the breaking of the growth symmetry under oleylammonium-rich conditions remains unclear. On the one hand, these conditions might promote the initial formation of monolayer [oleylammonium] 2 PbBr 4 sheets, consisting of a layer of edge-sharing octahedra sandwiched between two layers of close packed oleylammonium ligands. These sheets then grow thicker as the reaction proceeds. On the other hand, subtle kinetically driven mechanisms might be at work, as that invoked by Riedinger et al. in the formation of CdSe nanoplatelets. 36 Recent advances in the lead halide perovskite nanocrystal synthesis include the introduction of benzoyl halides as halide source separate from metal ions. 37 That advancement enabled broader experimentation with synthetic conditions such as metal/halide ratio and choice of surfactants used in the synthesis. For example, when oleylamine, which is a primary amine, is replaced with a secondary amine, the shape of the obtained CsPbBr 3 nanocrystals is always cubic regardless of temperature and growth conditions. 3 This was explained by considering that secondary ammonium ions bind more weakly to the surface of nanocrystals than primary ammonium ions and hence they cannot act as growth templates. This initial hypothesis was supported by a large body of experimental and computational data that demonstrated a predominance of oleate molecule in the ligand passivation shell of nanocrystals synthesized in the presence of secondary ammonium species. The cube edge length could be controlled by changing the A stronger quantum confinement is desirable to push the PL outside the green spectral region toward higher energy and can be achieved by growing ultrathin CsPbBr 3 nanoplatelets or nanosheets. 33,39 For example, PL shifts to ∼2.70, 2.76, and 2.83 eV for 5, 4, and 3 monolayer-thick CsPbBr 3 nanoplatelets (Figure 3f−i), respectively. 33 Nanowires with a controlled thickness from ∼3 to 20 nm provide another alternative for tuning the quantum confinement in CsPbBr 3 nanostructures (Figure 3j−p). 35,40,41 CsPbBr 3 nanowires are an interesting case of materials with mixed quantum-confined and bulk-like characteristics, both in dispersions and in films. Thin, blueemitting CsPbBr 3 nanowires (3.5 ± 0.5 nm in diameter) display reversible, concentration-dependent PL shift of up to Figure 3. Shape is another dimension of control over optical properties of CsPbBr 3 nanocrystals. (a−d) TEM images of nanocubes synthesized with didodecylamine at various temperatures, from ∼6.2 nm at 50°C to ∼19 nm at 140°C. 3 Shape-pure nanocubes synthesized with secondary amines display narrow photoluminescence and multiple absorption features in toluene dispersion, indicating the resolution of various electronic transitions (e). 38 (f) Scanning transmission electron microscopy (STEM) image of CsPbBr 3 nanoplatelets along with (g, h) high-resolution transmission electron microscopy (HRTEM) images in face and side views, respectively. 33 (i) Comparison of absorption and PL spectra for CsPbBr 3 nanoplatelets (NPLs) of various thicknesses and nanocubes. 33 (j−o) TEM and STEM images and thickness histograms for CsPbBr 3 nanowires, along with corresponding absorption and PL spectra (p). 35 Using alkylphosphonic acids produces CsPbBr 3 nanocrystals with a truncated octahedral shape, as illustrated in (q) HRTEM images and corresponding models. 44 The size of truncated octahedra can be tuned from ∼5 nm to ∼9.2 nm (r−t) by changing reaction time, with corresponding changes in quantum confinement, as tracked by optical absorption and PL (u). 45   Accounts of Chemical Research pubs.acs.org/accounts Article More elaborate shape tuning is made possible by resorting to other types of surfactants. Zhang et al. devised a synthesis in which the only surfactants were alkyl phosphonic acids. 44 During the heat-up procedures necessary to dissolve all the reactants, these acids partially underwent condensation reactions and formed phosphonic anhydrides. Indeed, the surface of CsPbBr 3 nanocrystals was found to be coated by both hydrogen phosphonates (i.e., deprotonated phosphonic acids) and alkyl phosphonic anhydrides. These ligands bind strongly to facets that are rich in Pb, and these are not only the (010), (101) and (101̅ ) ones of the orthorhombic perovskite phase but also additional higher index facets. As a result, the nanocrystals had a cuboctahedral shape (Figure 3q). A similar result in terms of shape control was achieved in a more recent work, 45 in which custom-synthesized oleylphosphonic acid was used. 46 The main advantage of using oleylphosphonic acid in lieu of alkyl phosphonic acids is that the former are much more soluble in the nonpolar or moderately polar solvents used to prepare colloidally stable suspensions of nanocrystals, and this has remarkable consequences over the stability of nanocrystals under air: when colloidal suspensions of nanocrystals prepared using alkyl phosphonic acids are exposed to air, the protonation of the hydrogen phosphonates due to moisture should transform them into charge-neutral phosphonic acids, which get detached from the surface of the nanocrystals and, being insoluble, precipitate. This process slowly destabilizes the nanocrystals, which aggregate over time. When instead nanocrystals prepared using oleyl phosphonic acids are exposed to air, moisture may again protonate the surface bound hydrogen phosphonates, which again are transformed into charge-neutral phosphonic acids that detach from the surface. This time, however, these acids are soluble in the solvent used to disperse nanocrystals and can bind back to their surface by losing a proton or even by hydrogen bonding interactions. 45 The shape transformation of CsPbBr 3 nanocrystals can be also achieved by external stimuli. For example, photoannealing of blue-emitting quantum-confined CsPbBr 3 nanoplatelets transformed them into green-emitting CsPbBr 3 nanobelts with PLQY as high as 65% and amplified spontaneous emission thresholds as low as ∼0.25−1 mJ/cm 2 in the solid state. 47 The stages of the photoinduced transformation were captured by TEM, which evidenced an evolution from nanoplatelets, self-assembled face-to-face into stacks, into thicker nanocrystals and belts 30−70 nm wide, over the course of 5 min exposure to a 365 nm LED source. The transformation took advantage of several factors: the strained crystal structure of thin nanoplatelets, their labile surface passivation, and the influence of moisture. The product of photoannealing was a sturdy film of CsPbBr 3 nanobelts that did not lose their PL nor did they dissolve upon exposure to toluene or to polar solvents (methanol, ethanol, isopropanol). The increased brightness and stability of the CsPbBr 3 nanobelts was exploited to fabricate green-emitting LEDs.

AND PERFORMANCE
The surface chemistry of CsPbX 3 nanocrystals is a key to their stability and improvement of optical properties. 48 (Figure 4b,c). Also, the DDDMAB treatment boosted the PLQY from around 50% to over 90% and made the nanocrystals colloidally stable up to 80°C in toluene (Figure 4d,e). 4 The good solvation of Cs-oleate in toluene as compared to the poor solvation of DDDMAB is the reason behind the enhanced properties of DDDMAB-capped nanocrystals.
Proper surface passivation was also found to have a key role in stabilizing blue-emitting CsPb 1−x Cd x Br 3 alloyed nanocrystals: 53 when part of the Pb 2+ ions in CsPbBr 3 is replaced by Cd 2+ ions, the structure of the nanocrystals changes from orthorhombic to cubic, and the band gap widens (Figure 4f). As-synthesized, these nanocrystals expel Cd 2+ ions (and Br − ions, to maintain charge neutrality) over time, and their emission color shifts to green in a few days. Replacing their native surface ligands with quaternary ammonium bromide ligand pairs prevented the loss of Cd 2+ ions (and boosted the PLQY). Here, again, DDDMAB was particularly effective. This behavior was rationalized by assuming that surface bromide vacancies are likely facilitating the expulsion of Cd 2+ ions. When such vacancies are saturated, by coating the nanocrystal surface with ammonium bromide ligand pairs, the loss of Cd 2+ ions is prevented and the nanocrystals preserve their blue emission over time (Figure 4g).

■ CONCLUDING REMARKS
In this Account, we have discussed various transformations of cesium lead halide nanocrystals. The diversity of structures, stoichiometries, morphologies, and surfaces in these materials and their inherently fast reactivity produce a broad spectrum of dynamic properties. Given the highly tunable chemistry of cesium lead halide nanocrystals, a common strategy is to obtain a well-defined material (i.e., a nanocrystal with specific structure, size, shape, and passivation) with optimized properties and then fight against their tendency toward reactivity, in order to preserve such properties. This, however, requires a good comprehension of the transformations themselves. From our experience, there are three lessons that come to aid for that. First, the preservation of the Cs + cation substructure limits the range of structures among which nanocrystals can transform. Second, the reactivity pathways in Accounts of Chemical Research pubs.acs.org/accounts Article the Cs−Pb−X system can be described based on ion trade reactions, which unify stoichiometry-changing and stoichiometry-preserving transformations (i.e., ion exchanges) and account for peculiar behaviors such as anion-assisted cation exchange reactions. Third, the competition between nanocrystal core and environment for the affinity with passivating ligands and the competition between ligands for the surface sites determine the nanocrystal shape during synthesis and the sample stability. As our understanding of these principles improves, we expect that the reactivity of cesium lead halide nanocrystals will increasingly turn from a challenge to an opportunity. For example, mixing of all three halides (Cl, Br, and I) inside individual Cs−Pb−X nanocrystals leads to the formation of Ruddlesden−Popper defect planes that effectively slice the crystal into separate crystallographic domains. 54 Compositional tuning of those domains may result in a series of strongly quantum-confined wells, potentially enabling us to engineer electron relaxation cascades in individual nanocrystals. Furthermore, a slight doping of mixed CsPb(I 1−x :Br x ) 3 nanocrystals with chloride may lead to their stabilization against halide segregation under illumination, as was demonstrated in bulk. 55 That coupled together with polymer-enhanced surface chemistry will likely deliver colortunable and photostable single photon emitters. 56 On a broader perspective, knowing how nanocrystals transform can help prevent or exploit such transformations. For example, halide migration can probably be halted in compounds with mixed covalent/ionic bonding such as chalcohalides 57 or metal−halide compounds in which the formation energy of halide vacancies is high. Growing a shell even a few layers thick of such materials on halide perovskite nanocrystals might prevent them from undergoing anion exchange and other unwanted reactions.