Halide‐Exchange Arrest Enables Reabsorption‐Free CsPbCl₃/CsPbI₃ Perovskite Core/Shell Nanocrystals

Abstract

Expanding the Stokes shift of lead‐halide perovskite nanocrystals (NCs) without compromising their sharp, fast excitonic emission has remained elusive, as high halide mobility erases the compositional gradients required for stable core/shell architectures. Here, it is shown that introducing a CdCl₂ passivation step prior to halide exchange provides a simple solution. Treating CsPbCl₃ NCs with CdCl₂ eliminates halide‐vacancy traps, enhances emission yield, and crucially blocks inward diffusion of I⁻, arresting the Cl⁻ → I⁻ exchange after just a few monolayers. This produces CsPbCl₃/CsPbI₃ core/shell NCs that absorb at 3.14 eV from the core and emit at 1.91 eV from the shell, achieving an apparent Stokes shift of ≈1.2 eV. The heterostructures exhibit ≈70% photoluminescence quantum yield, fast emission lifetime (≈10 ns) and complete suppression of reabsorption losses, as confirmed by liquid‐waveguiding experiments. Transient absorption spectroscopy and DFT modeling reveal an inverted type‐I band alignment with ultrafast (≈60 ps) core‐to‐shell exciton transfer. This fully solution‐processed chemistry enables heterostructuring‐based wavefunction engineering – long employed to expand the capabilities of conventional quantum dots – now realized in perovskite NCs, which provides a practical route to reabsorption‐free perovskite emitters for advanced photonic and quantum technologies.

Achieving large Stokes shifts in perovskite nanocrystals is challenging due to halide mobility that disrupts stable core–shell structures. Using CdCl₂ passivation, this work stabilizes CsPbCl₃/CsPbI₃ heterostructures, yielding a ≈1.2 eV Stokes shift, high quantum yield, and fast exciton transfer, enabling reabsorption‐free emitters for advanced photonic and quantum technologies.

1. Introduction

Over the past few years, lead halide perovskite nanocrystals (LHP NCs) of general formula APbX₃, where A is a monovalent cation (typically Cs⁺ or organic cations like methylammonium or formamidinium) and X is Cl, Br or I, have attracted significant interest due to their potential in a wide range of optoelectronic and photonic applications,^{[ 1 ]} spanning energy conversion, ^{[ 2 ]} artificial lighting,^{[ 3} , ⁴ , ^{5 ]} sensing,^{[ 6} , ^{7 ]} photonics,^{[ 3} , ⁴ , ⁵ , ⁸ , ^{9 ]} and radiation detection.^{[ 10} , ^{11 ]} These materials offer compelling advantages, including low‐temperature, solution‐processable, and tunable syntheses,^{[ 3} , ⁴ , ⁸ , ¹² , ^{13 ]} high photoluminescence (PL) quantum yields (Φ_PL ) owing to their inherent defect tolerance,^{[ 9 ]} fast luminescence lifetimes,^{[ 12 ]} and ease of integration into functional devices.^{[ 3} , ⁴ , ⁸ , ^{14 ]}

Despite these remarkable properties and the rapid progress in understanding their synthetic chemistry and physical properties, key challenges in tailoring their photophysics remain unresolved, which limits their deployment in technologically relevant applications. A major obstacle is the typical small Stokes shift of excitonic quantum dots – only a few tens of meV in LHP‐NCs ^{[ 1} , ¹² , ^{15 ]} – which results in a substantial spectral overlap between excitonic absorption and emission. This leads to self‐absorption of the emitted luminescence, particularly problematic for applications involving light propagation through the device, such as luminescent solar concentrators,^{[ 15} , ¹⁶ , ^{17 ]} waveguides and photonic fibers, and scintillators for radiation detection.^{[ 18 ]} In traditional semiconductor quantum dots, such as II‐VI or III‐V materials, the issue of small Stokes shift has been successfully addressed via wavefunction engineering strategies,^{[ 19} , ^{20 ]} most notably through the design of heterostructures with type‐I or quasi‐type II band alignment in core/shell,^{[ 21} , ²² , ²³ , ²⁴ , ^{25 ]} dot‐in‐rod,^{[ 26} , ²⁷ , ²⁸ , ²⁹ , ^{30 ]} or core/crown geometries.^{[ 31} , ³² , ^{33 ]} In these systems, a narrow‐gap domain confines the exciton and serves as the emission center, while the wide‐gap portion dominates absorption. This configuration results in an “apparent” Stokes shift, as the PL‐resonant (typically core) absorption becomes negligible relative to the off‐resonant (typically shell‐related) one, effectively suppressing reabsorption of the emitted light. Another effective approach involves doping with emissive impurities, such as transition metal ions (e.g., Mn²⁺, Ag⁺, Cu^+/++)^{[ 34} , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ^{41 ]} or engineered point defects ^{[ 42} , ^{43 ]} that act as mid‐gap color centers or introduce mid‐gap states that trap one of the charge carriers, forming intragap bound excitons. These systems exhibit large apparent Stokes shifts and negligible resonant absorption, though at the cost of broader (≥60‐80 nm) and slower (≥100‐200 ns) PL due to strong crystal field effects and the forbidden nature of the emissive transitions (spin or parity) and/or largely reduced oscillator strength due to small electron‐hole spatial overlap.^{[ 34} , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ^{41 ]}

However, in LHP NCs, only doping with color centers has been successfully achieved to date.^{[ 1} , ⁴⁴ , ⁴⁵ , ⁴⁶ , ⁴⁷ , ⁴⁸ , ^{49 ]} This limitation stems from two intrinsic factors: (i) the high ionic mobility of the perovskite lattice, which causes rapid halide interdiffusion during synthesis or halide exchange reactions.^{[ 12} , ⁵⁰ , ⁵¹ , ⁵² , ^{53 ]} As a result, efforts to create LHP‐based heterostructures with distinct halide compositions, such as CsPbCl₃/CsPbBr₃ or CsPbBr₃/CsPbI₃, resembling traditional CdSe/CdS designs where the core and shell differ only for the anion composition, have mostly failed, yielding mixed‐halide alloys rather than well‐defined heterojunctions.^{[ 54 ]} Examples of CsPbBr₃/Cs₄PbBr₆ ^{[ 55} , ^{56 ]} or CsPbCl₃/Pb₃S₂Cl₂ ^{[ 57 ]} heterostructures have been reported, but their domains appear to be photophysically disconnected, and interfacial defects often dominate their physics. Exciton funneling mimicking core/shell localization has been obtained in 2D layered perovskite nanosheets but the thick emissive domain retained significant resonant absorption, leading to partial suppression of reabsorption losses.^{[ 58 ]} (ii) The antibonding character of the band‐edge states, which underlies their defect tolerance^{[ 9 ]} but also impedes the formation of engineered intragap states, preventing conventional electronic doping motifs. As a result, the only LHP NC systems with engineered Stokes shifts to date are those doped with luminescent centers like Mn²⁺ or rare‐earth ions.^{[ 44} , ⁴⁵ , ⁴⁷ , ⁴⁹ , ⁵⁹ , ⁶⁰ , ^{61 ]} Although rare‐earth doping is facilitated by the coordination environment of the perovskite lattice, which represents a valuable advantage with respect to conventional NCs in which lanthanide doping is challenging, these systems typically exhibit long PL lifetimes due to spin‐ or parity‐forbidden transitions. Self‐trapped excitons have also been reported in mixed halides,^{[ 62} , ^{63 ]} but these, too, suffer from long, non‐tunable emission dynamics.

Developing LHP‐NCs that combine excitonic photophysics with suppressed reabsorption remains a major challenge due to the difficulty of stabilizing heterostructures for wavefunction engineering. Here, we attempt to contribute to this endeavor by reporting the synthesis and photophysical characterization of core/shell CsPbCl₃/CsPbI₃ NCs obtained via controlled anion exchange. Inspired by a recent work by Manna and co‐workers, who observed a transient CsPbCl₃/CsPbI₃ intermediate phase during in situ Cl⁻ → I⁻ exchange with optical absorption attributable to CsPbCl₃ and PL typical of CsPbI₃, and the literature on the role of ligands ^{[ 64} , ^{65 ]} and surface vacancy controlled ion exchange in halides,^{[ 52 ]} we hypothesized that it is possible to isolate heterostructured CsPbCl₃/CsPbI₃ NCs by arresting the anion exchange process before complete halide mixing via passivation of surface defects of the parent NCs. We therefore employed surface passivation of CsPbCl₃ NCs with CdCl₂, which mitigates surface defects, enhances emission yield,^{[ 11 ]} and suppresses I⁻ diffusion into the NC core (hereafter referred to as CdCl₂:CsPbCl₃ NCs). Optical spectroscopy experiments corroborated by density functional theory (DFT) calculations demonstrate the formation of an inverted type‐I junction where CdCl₂:CsPbCl₃ core excitons are transferred to the CsPbI₃ shell in ≈60 ps, resulting in UV absorption from the core and fully decoupled shell emission in the red spectral region. Under high‐fluence excitation, resulting in multiexciton population of the particle core, the relatively slow core‐to‐shell exciton transfer rate is largely outpaced by nonradiative Auger decay of core biexcitons (τ _XX ≈5–10 ps), which prevents the formation of shell biexcitons. This is similar to the behavior of ultra‐thick shell CdSe/CdS NCs that feature high purity single exciton emission due to Coulomb‐blocked hole transfer in the particle core.^{[ 66 ]} Finally, the formation of CsPbCl₃/CsPbI₃ heterostructured NCs enables the complete suppression of reabsorption losses, as confirmed by optical waveguiding experiments over long optical distances (up to 20 cm), which suggest a possible route for future progress in wavefunction engineering of perovskite‐based heterostructures for photonic applications.

2. Results and Discussion

2.1. Synthesis, Structure and Optical Properties of CsPbCl₃/CsPbI₃ Hetero‐NCs

The two key reagents required for synthesizing the parent CsPbCl₃ NCs and the PbI₂ precursor were prepared according to established protocols,^{[ 12 ]} as detailed in the Methods section. The synthesis scheme is sketched in Scheme S1 (Supporting Information). The as‐synthesized CsPbCl₃ NCs display the expected optical characteristics, exhibiting a pronounced band‐edge absorption at 3.14 eV accompanied by an excitonic PL peak at 3.05 eV (Figure  1a,b). The PL quantum yield (Φ_PL ) is relatively low, ≈1%–2%, and the time‐resolved PL traces (Figure 1e) reveal heavily quenched multiexponential decay kinetics, a typical behavior that reflects dominant nonradiative losses in Cl‐based LHPs. These losses are primarily attributed to hole trapping in deep intragap states, which are generally associated with undercoordinated surface chloride ions bound to undercoordinated lead ions.^{[ 11 ]} As expected, attempts to directly form a stable CsPbCl₃/CsPbI₃ phase by adding PbI₂ to the as‐synthesized CsPbCl₃ NCs under ambient conditions were unsuccessful. Similar to previous reports, the addition of PbI₂ resulted in complete and uncontrolled halide exchange, fully replacing Cl⁻ with I⁻. This rapid transformation is evidenced by in situ absorption and PL spectra in Figure 1a,b and by the corresponding photographs under ambient and UV light: upon the addition of 100 µL of a 50 mM PbI₂ parent solution to CsPbCl₃ NCs in toluene (absorbance = 1.1 OD at 3.14 eV), the band edge PL of CsPbCl₃ suddenly disappeared and both spectra evolved within a few minutes to resemble those of CsPbI₃. As quantified in Figure 1f, the progress of the halide exchange reaction was accompanied by a non‐monotonic evolution of the emission intensity: initially, the band edge luminescence of CsPbCl₃ suddenly dropped, which was ascribed to the formation of defective Cs‐I surface layers, followed by the gradual increase of Φ_PL reaching ≈20% in the fully exchanged particles, which is consistent with the generally observed higher efficiency of CsPbI₃ NCs with respect to the Cl‐based analogue.^{[ 12 ]} Readers are referred to ref.[50] for a detailed structural analysis of the distinct stages involved in the CsPbCl₃ → CsPbI₃ transformation. The progress of the anion exchange reaction is reflected also in the luminescence energy. As seen in Figure 1g, the PL peak energy moves toward the red until it matches the emission of pure CsPbI₃ NCs produced by hot injection, indicating that Cl⁻ ions have negligible residual solubility in fully exchanged NCs. The close match between the PL of fully exchanged and directly synthesized NCs is further highlighted in the inset of Figure 1b. Lowering the PbI₂ concentration in an attempt to slow the exchange also proved ineffective since smaller amounts failed to initiate the halide exchange, consistent with the findings of Manna and co‐workers.

Figure 1.

In situ a) optical absorption and b) PL spectra recorded during the exchange reaction of pristine CsPbCl₃ NCs upon PbI₂ addition, resulting in complete CsPbCl₃‐to‐CsPbI₃ conversion. Inset of (b): CsPbI₃ band‐edge emission from fully exchanged NCs (as in panel b, red line) or heterostructured NCs (as in panel d, dashed red line), compared to CsPbI₃ NCs obtained via hot‐injection synthesis (grey shaded area). In situ c) optical absorption and d) PL spectra of CdCl₂‐treated CsPbCl₃ NCs upon PbI₂ addition, leading to the formation of CsPbCl₃/CsPbI₃ core/shell heterostructures. Photographs in (a,c) and (b,d) show the final NC solutions under ambient or UV illumination, respectively. e) PL decay curves of pristine CsPbCl₃ (dark blue), CdCl₂:CsPbCl₃ (blue), and CsPbCl₃/CsPbI₃ heterostructured NCs (red) with the shaded area showing the instrument response function. f) Evolution of the PL quantum yield (Φ_PL ) and g) PL peak energy (E_PL) as a function of exchange reaction time for pristine (circles) and CdCl₂‐treated (triangles) CsPbCl₃ NCs. The horizontal line in (g) corresponds to the peak energy of the hot‐injection synthesized CsPbI₃.

The simultaneous occurrence of a rapid, uncontrolled anion exchange and a low Φ_PL suggests that surface defects play a critical role in both halide mobility and nonradiative losses. Because halide exchange initiates at the surface and propagates inward,^{[ 52 ]} we hypothesized that effective surface stabilization could confine anion exchange to the outer shell. Concurrently, passivation of surface traps would suppress nonradiative recombination, thereby improving Φ_PL and enabling efficient transfer of band edge excitons from the CsPbCl₃ core to the red‐emitting CsPbI₃ shell in the target heterostructure. Recent studies have demonstrated that post‐synthetic treatment with excess CdCl₂, which partially substitutes surface Pb with Cd, effectively suppresses carrier trapping.^{[ 11 ]} The introduced Cd, a cation that favors tetrahedral coordination, reorients the surface Cl ions into a more favorable, highly coordinated configuration. This structural rearrangement inhibits the formation of trap states, leading to enhanced Φ_PL . Building on this strategy, we applied CdCl₂ treatment to the CsPbCl₃ NCs. Briefly, as‐synthesized NCs were dissolved in 8 ml toluene and ≈300 mg of CdCl₂ was added to the stirring CsPbCl₃ solution at ambient temperature (≈30 °C). Two hours later, the excess CdCl₂ was discarded by centrifugation and the clear supernatant, containing CdCl₂‐treated CsPbCl₃ NCs, was preserved for further experimentation. Consistent with previous studies, this led to a substantial increase of Φ_PL to 20% (Figure S1, Supporting Information) and turned the PL decay profile to single exponential with lifetime τ = 2.4 ns^{[ 67 ]} (Figure 1e). The washed CdCl₂‐treated particles were mapped using Energy‐dispersive X‐ray (EDX) spectroscopy, which highlighted the presence of ≈10±1% of Cd with respect to Pb (Figure S2, Supporting Information).

Following surface treatment, halide exchange was performed on CdCl₂:CsPbCl₃ NCs in toluene solution prepared under identical conditions to the untreated sample, with matched NC concentration, as confirmed by optical absorption measurements. As shown in Figure 1c, the absorption spectrum of the treated NCs remained largely unchanged over time upon the addition of 100 µL of a 50 mM PbI₂ solution. This optical stability was also visually apparent: while the pristine NCs turned deep red under ambient light due to red‐shifted absorption from extensive iodide incorporation, the treated NCs remained transparent (see photographs in Figure 1a,c). In situ PL measurements (Figure 1d) confirmed the formation of the CsPbI₃ phase on CdCl₂‐treated NCs, as evidenced by the appearance and progressive – albeit slower compared to standard CsPbCl₃ NCs – intensification of red PL that gradually red‐shifted over time before stabilizing at ≈1.91 eV (650 nm). This value is ≈100 meV higher in energy than that of fully exchanged analogues. The observed energy offset likely arises from partial Cl solubility in the iodine‐based shell facilitated by the proximity of the particle surfaces that alleviate the otherwise significant lattice mismatch^{[ 68 ]} between Cl‐ and I‐based lattices. On the other hand, this effect likely prevents the formation of a mixed CsPb(Cl_1‐xI_x)₃ phase in fully exchanged particles. We note that strong quantum confinement of excitons in the CsPbI₃ shell could, in principle, also contribute to the observed PL shift. As in pristine CsPbCl₃ NCs, surface‐limited halide exchange in CdCl₂‐treated particles resulted in an enhanced Φ_PL of ≈70% (Figure 1f), attributable to exciton recombination within the CsPbI₃ shell, which hosts fewer nonradiative surface defects than the CdCl₂‐treated CsPbCl₃ surface. Consistent with the curing effect of CdCl₂, the heterostructured NCs exhibited a ≈3‐fold increase in efficiency (20% → 70%) relative to CdCl₂:CsPbCl₃ NCs, in contrast to the >15‐fold brightening observed for untreated, defect‐rich particles. Notably, the ability to achieve strongly Stokes‐shifted shell emission with relatively high efficiency represents a distinctive advantage of inverted type‐I lead halide perovskite NC heterostructures, arising from their defect‐tolerant nature and not readily attainable in conventional semiconductor heterostructures without additional wide‐bandgap overshelling due to severe surface trapping losses. After completion of the exchange reaction, the sample remained stable over time, confirming the formation of heterostructured NCs with absorption dominated by the CsPbCl₃ core and emission from the iodine‐rich shell.

High resolution transmission electron microscopy (HRTEM) experiments and Energy Filtered TEM (EF‐TEM) chemical analysis of the heterostructured NCs, shown in Figure  2 , confirm the formation of a iodine‐rich shell and provide deeper insights into the particle structure following controlled anion exchange. Specifically, the morphology of the CdCl₂:CsPbCl₃ NCs remains essentially unchanged after the Cl⁻/I⁻ exchange, as shown by the comparison of the HRTEM images in Figure 2a,b, which represent the NCs before and after the exchange, respectively. The particle size slightly increases from 7.19 ± 0.06 to 7.57 ± 0.07 nm following iodine exchange, as evidenced by the size distribution histograms in respective panels right to Figure 2a,b. Fast Fourier transformed (FFT) images (Figure 2c,d) confirm that the particles are crystalline and adopt the cubic Pm3̅m structure, with a slight increase in lattice parameter resulting from Cl/I exchange (evidenced by the subtle elongation of the FFT spots in the heterostructured NCs, highlighted in the yellow squares). To determine the elemental distribution, iodine and chlorine signals were mapped by EF‐TEM across numerous NCs at both 200 kV and 80 kV, with representative results shown in Figure 2e–i. The iodine maps reveal that iodine is distributed over all the NC, with enhanced intensity along the edges, indicating that all particles in the ensemble have undergone heterostructuring, consistent with the complete disappearance of the CsPbCl₃ emission in Figure 1d. EDX analysis (Figure S3, Supporting Information) further quantifies the presence of both Cl and I. Importantly, no particles were found to contain only Cl or only I, indicating the absence of unexchanged parent particles as well as fully exchanged CsPbI₃ NCs within the ensemble. This observation is consistent with the synthetic approach, which does not employ a secondary shell‐growth precursor and therefore suppresses heteronucleation of pure CsPbI₃ NCs. A more detailed analysis of the heterointerfaces lies beyond the scope of this work and will be the subject of future investigations.

Figure 2.

Representative TEM image of a) CdCl₂‐treated CsPbCl₃ NCs, and b) CsPbCl₃/CsPbI₃ heterostructures with the corresponding size histogram analysis on the top right. FFT images showing interplanar distances of c) CdCl₂‐treated CsPbCl₃ (blue outline), and d) CsPbCl₃/CsPbI₃ heterostructures (red outline). Elemental mapping performed at 200 kV by energy filtered TEM of e) iodine (N edge), and f) chlorine (L edge), showing the presence of both elements in all particles. Yellow outline in (e) corresponds to the same particle outlined in (b). g) Magnified HRTEM of CsPbCl₃/CsPbI₃ showing lattice fringes. h) Energy filtered map of the iodine N edge from the same region, acquired at 80 kV. (i) Overlay of (g) and (h) highlighting iodine distribution.

2.2. Density Functional Theory Calculations

To qualitatively evaluate the electronic structure and the energy level alignment between CsPbCl₃/CsPbI₃ core/shell NCs, we employed DFT calculations using the Perdew‐Burke‐Ernzerhof (PBE) exchange‐correlation functional (refer to the computational details below). As a starting point, we modeled a cubic, charge‐balanced CsPbCl₃ NC with a size of 3 nm and stoichiometry Cs₂₀₀Pb₁₂₅Cl₄₅₀, exhibiting a wide bandgap and delocalized one‐electron wavefunctions for electrons and holes (Figure  3a). To avoid unnecessary complexity in the NC model, we indeed assumed a defect‐free CsPbCl₃ system, effectively emulating the experimental CdCl₂‐passivated NCs, which exhibit the highest Φ_PL .

Figure 3.

Octahedral‐style atomistic models and electronic structure analysis of CsPbCl₃ NCs with different CsPbI₃ shell configurations. (Top panels): a) Initial 3 nm cubic CsPbCl₃ model, along with representative PbI₆ single‐octahedron substitutions at b) facet, c) corner, d) edge and e) mixed surface sites. In these models, PbCl₆ octahedra are shown in green. PbI₆ octahedra are distinguished by location: orange (facets), purple (corners), and magenta (edges). Cs atoms are omitted for clarity. (Middle panels): Partial density of states (pDOS) plots showing valence band (VB) and conduction band (CB) contributions from PbCl₆ and PbI₆ octahedra for the models in (a‐e). The same color scheme as in the top panels is used, with Cs contributions also included in cyan. Horizontal color‐coded line segments represent the contribution of Cl‐ or I‐based, as well as Cs, to a given MO. Horizontal dotted lines indicate the CB edge (CBE, red) and VB edge (VBE, blue). Inset values indicate the HOMO – LUMO gaps. The energy zero is referenced to the VBE of the core‐only CsPbCl₃ NC model. (Bottom panels): Wavefunction isosurfaces for the frontier MOs at the VBE and CBE for the models in (a‐e). PbCl₆‐related octahedra are omitted for clarity, while PbI₆ octahedra are shown in black. A hollow octahedron is drawn in cases where the wavefunction isosurface fully obscures the actual octahedron. Note that for the CsPbCl₃ LUMO isosurface, we selected a higher‐energy MO wavefunction to ensure full delocalization, as DFT is well known to artificially localize LUMOs near the conduction band edge.

Building on this model, we then performed a halide exchange reaction at different representative locations on the surface of the CsPbCl₃ NC model, namely at its facets, corners, and edges (Figure 3b–d, top panels). In all these cases, the halide exchange involves the substitution of single octahedron configurations of PbCl₆ with PbI₆ units. In addition, we also analyzed octahedral dimer configurations (Figure S4, Supporting Information, top panel), focusing on the same surface regions, in which two connected PbCl₆ octahedra (typically via corner‐sharing Cl atoms) form a structural unit of Pb₂Cl₁₁ that is substituted with Pb₂I₁₁. For the single octahedron configurations, we observed that replacing PbCl₆ with PbI₆ at the corners and edges leads to a dominant contribution of the I‐based octahedra to the VB MOs, extending more than 1 eV below the VBE. In the CB, the PbI₆ units share the electronic density near the band edge with PbCl₆ units (Figure 3c,d, middle panel). Both these configurations result in a redshifted bandgap relative to the pristine CsPbCl₃ model. On the other hand, when the halide exchange occurs on the facets, no such significant contribution to the VB frontier MOs is observed; instead, PbCl₆ remains the dominant contributor at the VBE – a trend even more pronounced in the CB (Figure 3b, middle panel). Surprisingly, this configuration leads to a slight blueshift compared to the initial CsPbCl₃, with the LUMOs being pushed upwards in energy.

Regarding the spatial distribution of the electron and hole charge densities at the band edge, the corner configuration clearly localizes the hole in the PbI₆ domains, while the electrons appear delocalized between the corner PbI₆ and adjacent edge PbCl₆ octahedra (Figure 3c, bottom panel). The edge configuration also localizes the hole predominantly in PbI₆ octahedra, but the electron is slightly more delocalized within the PbCl₆ framework, extending nearly halfway across a facet (Figure 3d, bottom panel). Unlike the corner and edge configurations, the facet configuration exhibits both electron and hole delocalized across PbCl₆ and PbI₆ domains (Figure 3b, bottom panel). Importantly, we noted that the same trends are obtained when substitutions are introduced in equal stoichiometric amounts at all surface sites (Figure S5, Supporting Information). Thus, the corner and edge single octahedron configurations are consistent with the formation of an inverted type‐I band alignment, as inferred from experimental Φ_PL measurements.

We have also simulated the substitution of two adjacent octahedra, with dimeric configurations. In this case, the bandgap decreases (Figure S4, Supporting Information) with little effect on the electronic structure. However, we have observed a significant distortion of the perovskite lattice near the substitution site, which tentatively suggests that CsPbCl₃ can only sustain localized substitutions with PbI₂. At higher concentrations, such substitutions may lead to destruction of the NC or complete formation of CsPbI₃. Overall, the most effective Cl⁻ → I⁻ exchange for producing the inverted type‐I band alignment occurs at the corners, where PbI₆ octahedra contribute most significantly to both the VB and CB compared to other surface regions, followed by edge and facets.

Finally, we modeled a CsPbCl₃ NC in which all relevant surface sites (edges, corners and facets) were functionalized simultaneously with PbI₆‐based octahedra, aiming to more realistically represent the CsPbCl₃/CsPbI₃ core/shell heterostructure observed in TEM images (Figure 2h,i), where iodine appears scattered across the NC surface. Here, we included only replacements of single octahedra configurations (Figure 3e, top panel), thus avoiding dimers. When all possible surface configurations are included, the inverted type‐I band alignment becomes even more evident, as confirmed by the pDOS plot and charge density isosurfaces (Figure 3e, middle and bottom panels). In this scenario, PbI₆ octahedra at the edges contribute most prominently to the VB and CB frontier MOs, followed by the faceted PbI₆ octahedra, which dominate at deeper energies.

In a previous work, we noticed that direct Cl⁻ → I⁻ replacement can lead to the formation of local Ruddlesden‐Popper phases (RPPs) domains.^{[ 69 ]} When RPP domains are included in our NC models, DFT calculations consistently predict an inverted type‐II band alignment, regardless of the specific shape of the RPP domains (Figure S6, Supporting Information). Overall, our DFT results, supported by experimental evidence, suggest that iodine incorporation in CsPbCl₃ NCs occurs at surface sites but in scattered positions, minimizing lattice distortion. This distribution prevents RPP phases formation and enables an optimal inverted type‐I band alignment that favors emission from these octahedra at longer wavelengths.

2.3. Exciton Transfer and Multiexciton Photophysics in Core/Shell CsPbCl₃/CsPbI₃ NCs

Next, we investigated the photophysical processes leading to the formation of excitons in the CsPbI₃ shell and the competing nonradiative pathways in greater detail. As illustrated in Figure 1c,d, the red shell emission at 1.91 eV lacks a clear counterpart in the absorption spectrum. This is consistent with the fact that CsPbCl₃ largely dominates the volume of the heterostructure NCs, as designed, and implies that the red emission is mediated by absorption in the CsPbCl₃ core, followed by energy localization in the iodide‐based shell. To investigate this phenomenon and quantify the exciton localization dynamics, we performed transient absorption (TA) measurements using 3.2 eV excitation, above the CsPbCl₃ bandgap, on the parent NCs after CdCl₂ surface reconstruction and the heterostructures. The TA spectra collected under low irradiance conditions (corresponding to an average exciton occupancy of 〈N〉 ≈ 0.05 per NC) are shown in Figure S7 (Supporting Information). Both samples exhibit the characteristic bleach signals of the 1S transition of CsPbCl₃ ≈3.17 eV, which are consistent with their linear absorption spectra. Due to the low shell thickness of the heterostructures, no bleach signal was detected in the red spectral region. Despite their similar spectral features, the bleach dynamics differ markedly between the core and core/shell NCs. At very low excitation fluence (f), where the average number of excitons per NC is 〈N〉 < 0.1, the CdCl₂:CsPbCl₃ NCs display a nearly single‐exponential decay with a time constant consistent with their PL lifetime of ≈2.1 ns (Figure  4a), suggesting that either residual PL efficiency losses occur on a faster timescale than the time resolution of our TA setup or they affect hot excitons, as in B‐type blinking.^{[ 70 ]} In contrast, in the single exciton regime, the heterostructured NCs exhibit a pronounced initial decay with τ _ET ≈60 ps lifetime (Figure 4a), which is ascribed to the depopulation of excited states in the CsPbCl₃ core in favor of the CsPbI₃ shell; after which this converges into a similar ns long decay (τ _X ∼ 2 ns) observed for the parent CdCl₂:CsPbCl₃ NCs and ascribed to the decay of core excitons that do not localize in the shell domain. Similar behavior has been reported for exciton transfer in Mn²⁺, Yb³⁺‐doped CsPbX₃ (X: Cl, Br) NCs.^{[ 45} , ⁴⁷ , ^{59 ]} Based on the bleach kinetics of the parent and heterostructure NCs, we estimate an exciton transfer efficiency ${\varphi }_{ET}=(1-\frac{{\tau }_{ET}}{{\tau }_X})\times 100=97\%$, which is consistent with the complete disappearance of the core emission observed in Figure 1d. For the CdCl₂:CsPbCl₃ NCs, increasing f results in the development of a fast initial contribution in the TA kinetics due to the decay of biexcitons (Figure 4b, corresponding to 〈N〉 ≈1.6). Consistent with the literature,^{[ 71} , ^{72 ]} the biexciton lifetime, τ _XX = 6 ps is much shorter than the expected radiative lifetime, defined as τ _XX,Rad = τ _X,Rad /4 ≈ 500 ps, resulting in the dominant contribution of nonradiative Auger recombination, with efficiency Φ_AR ≈99%. Similarly, at high f, the heterostructured NCs exhibited an additional, faster bleach decay component with a lifetime of ≈7 ps, matching the biexciton lifetime of the parent CdCl₂:CsPbCl₃ NCs (inset of Figure 4b). Meanwhile, the contributions by core‐to‐shell exciton transfer and radiative decay remained largely unchanged. The observation of faster Auger‐driven biexciton decay than core‐to‐shell exciton localization lifetime in the heterostructure NCs (${\tau }_{XX}\sim \frac{{\tau }_{ET}}{10}$) also suggests that shell multiexciton formation upon core pumping is kinetically hindered. To test this hypothesis, we conducted fluence‐dependent, time‐resolved PL measurements on the heterostructures and pristine CsPbI₃ NCs. As shown in Figure 4c, at low‐f values, the shell PL follows single‐exponential kinetics with a lifetime of ≈10 ns, which is slightly faster than the emission lifetime of pure CsPbI₃ NCs emitting at 680 nm (20 ns, Φ_PL ≈40%, Figure 4d), consistent with the partial alloying with residual Cl. Upon increasing f, the CsPbI₃ NCs exhibited the typical emergence of fast multiexciton contribution to the PL kinetics with lifetime τ _XX ≈360 ps.^{[ 73} , ⁷⁴ , ^{75 ]} Conversely, the PL kinetics of the CsPbCl₃/CsPbI₃ NCs remained unchanged for over 4 orders of magnitude of excitation fluence, thus confirming the scenario of hindered multiexciton localization in the shell region. This peculiarity of these heterostructures could potentially be interesting for designing high‐purity single‐photon emitters with no spurious biexciton contributions, even under intense optical pumping, similar to the behavior of ultrathick shell CdSe/CdS NCs.^{[ 66 ]} Further confirmation of this mechanism comes from the identical fluence dependence of the integrated PL intensity measured for the parent CsPbCl₃ NCs and the heterostructures, indicating that saturation of the shell emission is determined by state filling and Auger decay in the particle core (Figure 4f). We further notice that the fluence‐dependent PL measurements on the heterostructures reveal linear, unsaturated growth of a broad 2.5 eV PL (Figure 4e), suggesting the presence of deep surface defects that do not affect the core‐to‐shell carrier dynamics. In contrast, fluence‐dependent PL measurements of the CdCl₂:CsPbCl₃ NCs did not exhibit the emergence of such an emission (Figure S8, Supporting Information), indicating that it originates from the Cl⁻/I⁻ exchange as further confirmed below (vide infra). The nature of this defect emission is currently under active investigation and is beyond the scope of this work. Nevertheless, based on literature,^{[ 76 ]} a possible origin could be trapping of band edge excitons in undetectable Cs₂Pb₂Cl₂I₂ phase at the core/shell interface deep trap states associated to residual undercoordinated halide ions.^{[ 77 ]} A dedicated study will be conducted to optimize the emission yield through specific surface passivation strategies.

Figure 4.

Ground state bleach decay kinetics of CdCl₂:CsPbCl₃ NCs (blue circles) and the heterostructured NCs (red circles) at a) low excitation fluence (〈N〉 < 0.1) and b) high excitation fluence (〈N〉  ≈ 1.6). The black lines in (a) are fits to a single exponential decay for the parent NCs (τ  =  2.2 ns) or double exponential for the CsPbCl₃/CsPbI₃ NCs (τ₁= 60 ps, τ₂ =  2.1 ns). Insets: a): schematic depiction of exciton localization from the CsPbCl₃ core to the CsPbI₃ shell. Inset of b): Biexciton (XX) bleach kinetics obtained by subtracting the low fluence bleach dynamics from the high fluence one. c) Normalized PL decay curves of CsPbCl₃/CsPbI₃ NCs at increasing excitation fluence showing nearly single exponential decay and no sign of XX contributions. Curves are vertically shifted for clarity. d) PL decay at low and high exciton occupancy for CsPbI₃ NCs obtained by hot injection synthesis, highlighting fast initial XX decay component at high fluence. e) PL spectra of heterostructured NCs a function of excitation fluence. The corresponding iodide band edge PL intensity is reported in “f” (red) together with the parent CdCl₂:CsPbCl₃ NCs (blue) showing identical trend of the respective band edge emission intensities and the linear growth of defect emission at 2.5 eV (black).

To further clarify the role of surface chemistry in the partial exchange reaction, we performed control experiments using CsPbCl₃ NCs synthesized using Pradhan and coworker's method, which includes oleyl ammonium chloride as an additional precursor.^{[ 78 ]} This approach yields ammonium‐capped NCs with Φ_PL ≈20%, which is comparable to that of our CdCl₂:CsPbCl₃ NCs. The formation of the CsPbI₃ phase in these ammonium‐capped NCs required substantially more PbI₂ (400 µL) than in conventional CsPbCl₃ NCs, but the resulting red emission was weak and nearly matched the intensity of the defect‐related emission (Figure S9a, Supporting Information). Increasing the amount of PbI₂ (to 490 µL) had a negligible effect and eventually led to complete conversion to CsPbI₃. In contrast, red emission was readily achieved on the same concentration of CdCl₂:CsPbCl₃ NCs with only 210 µL of PbI₂ required (Figure S9b, Supporting Information). Moreover, precise tuning of the PbI₂ concentration enabled systematic control of the intermediate halide composition. As the PbI₂ amount increased, the red emission became more intense and red‐shifted (Figure S9b, Supporting Information), reflecting progressive iodide incorporation. Notably, the intensity of the defect‐related emission remained nearly constant across different iodide concentrations. This implies that this feature is intrinsic to the mixed‐halide NCs and is not influenced by iodide content or CdCl₂ treatment. These control studies corroborated our hypothesis that the nature of the parent CsPbCl₃ NC surface and the density of surface ligands critically affect the formation of the desired CsPbI₃ domain. Specifically, CsPbCl₃ NCs synthesized using 500 µL each of oleic acid and oleyl amine yielded optimal results, whereas doubling the ligand volumes to 1 mL each resulted in less controlled halide exchange. This sensitivity is consistent with the surface‐initiated nature of halide exchange and highlights the importance of restricting iodide penetration into the core. In this configuration, a thin CsPbI₃ shell efficiently harvests excitons generated in the CsPbCl₃ core and emits red light, thus fulfilling the intended design of the heterostructure.

2.4. Demonstration of Suppressed Reabsorption in Heterostructured NCs

Finally, to demonstrate the practical advantage of a large Stokes shift for photon management applications, we characterized the waveguiding performance of the CsPbCl₃/CsPbI₃ NCs over long optical paths, up to tens of centimeters. For comparison, we conducted light propagation measurements on liquid waveguides containing toluene solutions (with the same band‐edge absorbance of 0.1 over 1 mm) of CsPbCl₃/CsPbI₃ NCs against standard CsPbI₃ NCs. The emission spectra were collected at one end of the waveguide while exciting the solutions at various distances, d, along its length with a 375 nm laser source. Figure  5a,b report the absorption spectra measured orthogonally to the waveguide axis and the PL spectra normalized to their low energy tail to focus exclusively on the reabsorption losses without additional geometrical contributions to waveguiding. Figure 5c shows the corresponding spectrally integrated PL intensity values. Consistent with the small Stokes shift of 82 meV, the CsPbI₃ NCs showed a dramatic drop in the PL intensity, with nearly 70% loss in 7 cm and resulting in a ≈12 nm redshift of the emission maximum, as the excitation source was moved further from the detector due to the progressive reabsorption of the high‐energy portion of the PL spectrum. In contrast, the CsPbCl₃/CsPbI₃ NCs showed constant emission intensity and spectral profile across up to 21 cm of propagation, indicating negligible reabsorption due to shell transitions. The photograph in Figure 5d captures the effect showing excellent waveguiding capability of the heterostructured NCs solution with intense red PL being emitted from the opposite side of the liquid waveguide.

Figure 5.

Optical absorption and normalized PL spectra as a function of increasing propagation distance “d” in toluene solutions with comparable band edge absorbance (0.1 OD) of a) CsPbI₃ and b) CsPbCl₃/CsPbI₃ NCs. Inset: sketch of experimental geometry. The black spectra correspond to the largest distance, d = 21 cm. c) Normalized integrated PL intensity extracted from panels (a) and (b) as a function of d. d) Photograph of the liquid waveguide containing CsPbCl₃/CsPbI₃ NCs, showing intense PL emission at the opposite end of the guide from the UV excitation spot (λ_exc = 375 nm).

3. Conclusion

In summary, we have demonstrated a viable strategy for engineering perovskite heterostructures with large apparent Stokes shifts and suppressed reabsorption losses by stabilizing CsPbCl₃ NCs with CdCl₂ and partially exchanging surface halides to form CsPbCl₃/CsPbI₃ core/shell architectures. Surface passivation with CdCl₂ proved essential for confining iodide incorporation to a thin shell while simultaneously suppressing nonradiative losses, thereby enabling efficient exciton transfer from the Cl‐rich core to the I‐rich shell. Time‐resolved spectroscopy and DFT modeling confirmed the formation of an inverted type‐I band alignment, featuring ultrafast (≈60 ps) core‐to‐shell exciton transfer and kinetically hindered biexciton localization in the shell – an advantageous combination for high‐purity single‐photon emission. Most notably, waveguiding experiments revealed complete suppression of reabsorption, in stark contrast to the severe optical losses observed in CsPbI₃. These results provide proof‐of‐principle that wavefunction engineering, long considered inaccessible in halide perovskites due to rapid halide interdiffusion, can be achieved through surface‐mediated exchange chemistry. Beyond advancing the fundamental understanding of exciton management in perovskite heterostructures, this work outlines a promising pathway toward reabsorption‐free perovskite NCs for photonic devices, including optical fibers, scintillators, quantum light sources, and luminescent concentrators (we point out that for practical solar technologies further research is needed to extend solar coverage beyond the UV). Overall, this study demonstrates the feasibility of controlled heterostructuring and opens promising avenues for future developments in perovskite nanophotonics.

4. Experimental Section

Chemicals

Cadmium chloride (CdCl₂, ≥99.99%), cesium carbonate (Cs₂CO₃, 99.95%), hydrochloric acid (HCl, 37%), lead acetate trihydrate (≥99%), lead chloride (PbCl₂, 98%), lead Iodide (PbI₂, 99%), octadecene (ODE, 90%), oleic acid (OA, 90%), and oleyl amine (OlAm, 70%), were purchased from Sigma–Aldrich. Trioctylphosphine (TOP, 97%) was purchased from abcr‐GmbH. Toluene (≥98%) was purchased from VWR. All chemicals and solvents were used as purchased without any further purification.

Synthesis of Cesium‐Oleate (Cs‐Oleate)

Cs₂CO₃ (405 mg, 1.24 mmol) was mixed with OA (1.25 mL) and ODE (10 mL) in a 50 mL three‐neck round‐bottom flask. The mixture was degassed at 120 °C for 1 h and subsequently the temperature was raised to 150 °C under nitrogen atmosphere and was maintained for 30 min. Once a clear solution was obtained, it was allowed to cool down and transferred to an air‐tight vial capped with a rubber septum. Since the solution solidifies at room temperature, it was preheated to 100 °C for obtaining a transparent solution before being used in the synthesis.

Synthesis of CsPbX₃ (X: Cl, I) Nanocrystals (NCs) Using Cs‐Oleate

CsPbX₃ NCs were synthesized by the method previously reported by Protesescu et al.^{[ 12 ]} Briefly, 0.188 mmol of PbX₂ (52 mg for PbCl₂, 87 mg for PbI₂) was combined with 500 µL each of OA and OlAm in 5 mL ODE in a 50 ml three‐neck round bottom flask and this reaction mixture was degassed at 110 °C for 45 min. 1 mL of TOP was injected for the complete solubilization of PbCl₂ salt at 100 °C. For PbI₂ salt, this step was avoided. After degassing, temperature was increased to 180 °C under nitrogen. At 180 °C, 400 µL of Cs‐oleate (prepared as described above) was swiftly injected followed by rapid quenching by immersing the reaction flask in ice‐water bath. Once the reaction mixture was cooled to 50 °C, it was taken out, and centrifugation at the relative centrifugal force (rcf) of 3823×g (6000 rpm) for 10 min. The supernatant was discarded and the precipitate was dissolved in 8 mL toluene.

To check the effect of surface ligand concentration, the above synthesis was repeated by doubling the amount of both OA and OlAm to 1 mL, and by keeping all the other reagents at the same amount.

Synthesis of Oleylammonium Chloride

1 mL of HCl (37%) was mixed with 10 mL of OlAm in a 50 mL three‐neck round‐bottom flask. The mixture was heated to 120 °C and maintained at this temperature for 2 h. Then the temperature was raised to 150 °C, and the solution was stirred for an additional 30 min. After cooling, the solution was transferred to a nitrogen‐purged glass vial capped. Since the mixture solidifies at room temperature, it was preheated to obtain a clear solution prior to use in the synthesis.

Synthesis of CsPbCl₃ NCs Using Oleylammonium Chloride

This synthesis was carried out following the method previously reported by Pradhan and coworkers.^{[ 78 ]} Briefly, Cs₂CO₃ (32 mg, 0.1 mmol), lead acetate trihydrate (76 mg, 0.2 mmol), 1 mL OA, and 10 mL ODE were loaded to a 50 mL three‐neck flask and was degassed at 110 °C for 45 min. Then the temperature was raised to 250 °C under nitrogen atmosphere. At 250 °C, 1.2 mL of oleylammonium chloride (prepared as described above) was swiftly injected and the reaction was annealed for 1 min. Then the heating was removed and the mixture was allowed to cool down naturally. At 40 °C, the solution was extracted out and centrifuged at rcf of 3823×g (6000 rpm) for 10 min. The supernatant was discarded and the precipitated NCs were redispersed in toluene.

Synthesis of Lead Iodide Precursor

1 mmol of PbI₂ powder was mixed with OA (2.6 mL), and OlAm (2.6 mL) in 15 mL of ODE as the solvent in a 50 mL three‐neck round‐bottom flask. The solution was degassed at 110 °C for 30 min, followed by increasing the temperature to 150 °C under nitrogen atmosphere. The solution was maintained at this temperature for ≈5 min to ensure complete dissolution of the PbI₂ salt. Then the solution was allowed to cool to room temperature and stored in an air‐tight vial previously purged with nitrogen.

Optical Spectroscopy Studies

Optical absorption measurements were carried out for colloidal dispersion in toluene using an Agilent Cary 50 UV–vis spectrophotometer. PL measurements were performed in reflection mode using a Varian Cary Eclipse fluorescence spectrometer, exciting the samples at 3.26 eV (380 nm). In‐situ absorption and PL measurements were carried out during anion exchange with the same instruments, collecting spectra each ≈20 s over ≈7 min till the spectral evolution ended. Relative PL quantum yield measurements were carried out by comparing samples with freshly made, and iso‐absorbing at the excitation wavelength (380 nm), acidic solutions (0.5 M H₂SO₄) of quinine sulfate used as a reference. In all cases, the PL spectra were corrected for detector efficiency. Fluence‐dependent steady‐state PL measurements were carried out by exciting the sample at 3.26 eV using the output from APOLLO‐Y optical parametric amplifier (OPA) coupled with a 10 W Hyperion amplified Ytterbium laser operated at 15 kHz, which generated 260 fs fundamental pulses at 1030 nm. The emitted light was fiber‐coupled into a TM‐C10083CA Hamamatsu Mini Spectrometer. Excitation laser fluence was varied across the 2 × 10⁻¹⁰–2 × 10⁻⁵ mJ cm² range resulting in an average excitonic population per NC, 〈N〉 raging from ≈0.002 to 232, having extracted the optical cross section (6.3 × 10⁻¹⁵ cm² at 370 nm) from Supporting Figure XX using methods described by Makarov et al.^{[ 74 ]} Light propagation measurements were carried out filling an EPR glass tube (≈25 cm optical path) with a colloidal solution of NC with 0.1 absorbance at the 1S exciton edge. Time resolved (trPL) measurements were performed – also as a function of excitation fluence – in a single‐photon counting configuration using a VIS‐photo‐multiplier tube coupled to a Cornerstone 260 1/4 m VIS‐NIR Monochromator (ORIEL) and a time‐correlated single‐photon counting unit (time resolution ∼400 ps). PL propagation, trPL, and fluence dependent PL share the same pulsed excitation source at 3.26 eV operated at 20 kHz.

Transmission Electron Microscopy

Transmission Electron Microscopy experiments were performed in a JEOL JEM‐2200FS microscope equipped with an in‐column Omega filter and in a JEOL JEM‐ARM 200F NEOARM operated at 80 or 200 kV. Given that the synthesis and treatment of the nanocrystals require organic compounds, carbon contamination occurs when the sample is exposed to the electron beam. The use of commonly employed methods, such as in‐situ electron beam shower or glow discharge, was found to be ineffective to remove the problem, thus preventing focused‐beam techniques (STEM, EDX mapping). Even lowering the acceleration voltage to 80 kV did not mitigate the issue. Parallel‐beam TEM was therefore used, where contamination remained tolerable even at atomic resolution. Energy‐filtered TEM performed with Gatan GIF Continuum system enabled compositional analysis, but contamination prevented Cl/I mapping on the same area. Therefore, iodine and chlorine maps were acquired independently on fresh regions, across multiple nanocrystals. The presence of both elements in each single particle was confirmed, even if a variable distribution was observed, which reflects the variability of the Cl–I exchange at the nanoscale.

Femtosecond Transient Absorption Measurement

Measurements were performed using a Helios TA unit from Ultrafast Systems. The Hyperion module was operated at 1.87 kHz as seed to generate a white light supercontinuum (probe beam), and a 3.8 eV pump beam by mean of the APOLLO‐Y OPA described above. The pump beam was synchronously chopped and phase‐locked at 0.938 kHz, blocking every other pulse. For fluence‐dependent TA analysis, the excitation pump fluence was adjusted in the 4.2 × 10⁻⁶ – 2.7 × 10⁻⁴ J cm⁻² range.

Computational Details

Atomistic simulations at the DFT level were performed using the PBE exchange‐correlation functional,^{[ 79 ]} a double‐ζ basis set (DZVP), and the Gaussian and Plane Waves (GPW) method, as implemented in the CP2K 2024.3 quantum chemical software package.^{[ 80 ]} Relativistic effects were included for all atoms using Goedecker‐Teter‐Hutter (GTH) effective core potentials. All structures were optimized in the gas phase within cubic boxes of 4 × 4 × 4 nm to ensure that the box boundaries did not introduce spurious interactions. The grid cutoff was set to 400 Ry for all calculations. The isosurface value used for the HOMO and LUMO plots is |0.003| (e⁻ /bohr³)^1/2.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.