Quantitative imaging of anion exchange kinetics in halide perovskites

Significance

Fundamental understanding of chemical transformation mechanism in solid-state semiconductors can offer guidelines for engineering these materials with various functionalities. Microscopically resolving such transformation process on the single-particle level can provide a direct picture of the dynamics, but has always been challenging due to the deficiencies of characterization techniques and material dimensions. This work presents a unique design on realizing the quantitative imaging of anion exchange reaction kinetics in halide perovskites by taking advantage of their high and tunable photoluminescence and characteristic lattice dynamics, coupled with controllable chemical treatments and optical microscopic techniques. Our study on kinetically visualizing ion exchange behavior in solid-state materials could also inspire mechanistic insights into other phenomena in a broad context of physical science.

Abstract

Ion exchange, as a postsynthetic transformation strategy, offers more flexibilities in controlling material compositions and structures beyond direct synthetic methodology. Observation of such transformation kinetics on the single-particle level with rich spatial and spectroscopic information has never been achieved. We report the quantitative imaging of anion exchange kinetics in individual single-crystalline halide perovskite nanoplates using confocal photoluminescence microscopy. We have systematically observed a symmetrical anion exchange pathway on the nanoplates with dependence on reaction time and plate thickness, which is governed by the crystal structure and the diffusion-limited transformation mechanism. Based on a reaction–diffusion model, the halide diffusion coefficient was estimated to be on the order of ${10}^{-14}{\text{cm}}^2\cdot {\text{s}}^{-1}$. This diffusion-controlled mechanism leads to the formation of 2D perovskite heterostructures with spatially resolved coherent interface through the precisely controlled anion exchange reaction, offering a design protocol for tailoring functionalities of semiconductors at the nano-/microscale.

The potential of chemical transformations has been extensively exploited in modifying the structures, compositions, and morphologies of materials to discover novel properties for various useful applications (1). Ion exchange is one such transformation that provides an additional pathway to achieve controllability in materials engineering which is potentially limited by direct synthetic routes (2–4). Significant efforts on utilizing partial cation exchange on metal chalcogenides in colloidal nanoscience as well as in solid-state chemistry have enabled the design and production of a wide range of crystalline nanomaterials with various structural complexities, such as segmented superlattice (5) or core–shell heterostructure (6) for efficient light-emitting or photovoltaic applications. Recently, a rapid exchange behavior on anion sites of halide perovskites has been reported (7–11). These studies have demonstrated the versatility of anion exchange in realizing the bandgap tunability and accessing otherwise metastable phases (12).

Along with these phenomenological investigations on ion exchange for expanding the library of materials, mechanistic insight into this transformation is also of great interest and fundamental importance. Work on studying the ion exchange kinetics has been predominantly carried out for cation exchange in metal chalcogenides nanocrystals (NCs) (4, 13, 14). By contrast, there have been limited efforts on studying the analogous anion exchange before the recent emergence of halide perovskites as promising optoelectronic materials (15–19). It is thus of our interest to understand the anion exchange mechanism which would provide guidelines for tailoring the material functionalities. On the other hand, kinetical analysis based on a spatially resolved transformation process in single particle has never been achieved due to the limitations imposed by imaging techniques and material dimensions. Conventional methods for probing cation exchange in those NCs typically rely on transmission electron microscopy (TEM) and TEM-based energy-dispersive X-ray spectroscopy (5, 20–22). These techniques, although helpful for resolving the structural or compositional changes in single NCs of different exchange extents (23, 24), pose inconvenience due to the operation under vacuum. Moreover, the electron beam would be destructive to exposed materials (25) and even triggers additional changes to the as-formed NCs structures (26), which is especially true for the halide perovskites. In contrast to that, the emerging halide perovskites will enable fast and noninvasive probing of anion exchange mechanism by optical characterizations thanks to their high photoluminescence efficiency and composition-dependent bandgap tunability. Initial investigations have developed a microkinetic description of nanoscale anion exchange for halide perovskites NCs (27). However, the rapid reaction rate in NCs and limitations in temporal and spatial resolution make it elusive to directly observe the anion exchange kinetics on single-particle level for demonstrating the plausible mechanism.

We observed anion exchange kinetics in cesium lead halide perovskites (CsPbX₃, X = Cl, Br, I) by spatially mapping the emission evolution of individual single-crystalline nanoplates, through confocal photoluminescence microscopy. Our improved chemical vapor transport (CVT) growth method (28) provided dry, solid perovskite nanoplates with low defect densities and symmetrical features for better resolving intrinsic material properties, and also enabled the tunability in lateral size and thickness for systematic studies. An experimental design of the vapor-phase anion exchange procedure additionally contributed to the rapid reaction quenching for immediate characterizations of their optical behaviors. Moreover, the micrometer-scale lateral size of these nanoplates and corresponding minute-long reaction time made the experimental system compatible with both the spatial and temporal resolutions needed for microscopic studies.

As the starting material for subsequent anion exchange studies, CsPbBr₃ nanoplates were heteroepitaxially grown on mica substrates by a one-step atmosphere–pressure CVT method (Methods). The optical bright-field image of the as-grown CsPbBr₃ nanoplates on mica substrates shows a well-defined square shape with typical lateral width of 10–20 μm (SI Appendix, Fig. S1A). High-resolution TEM (HRTEM) image of a thin nanoplate (SI Appendix, Fig. S1B) and the corresponding fast Fourier transform (FFT) pattern (SI Appendix, Fig. S1C) indicate its single-crystalline structure which can be indexed to the crystal symmetry of the [001] zone plane of CsPbBr₃ lattice. Further analysis of the FFT pattern implies the highly symmetrical nature of our nanoplates (see details in SI Appendix, Fig. S1C). Fig. 1A shows a wide-field photoluminescence spectrum of CsPbBr₃ nanoplate as well as its optical image (Bottom Inset) under laser excitation. The CsPbBr₃ nanoplates are highly emissive with an emission peak centering at around 526 nm, in agreement with the reported bandgap (29). A corresponding scanning electron microscopy (SEM) image (Fig. 1 A, Top Inset) of the nanoplate further confirms its morphology.

Fig. 1.

Anion exchange process in single-crystalline halide perovskite nanoplates and its characterizations. (A) Photoluminescence (PL) spectrum and image (Bottom Inset), and SEM image (Top Inset) for one starting CsPbBr₃ nanoplate. (Scale bars, 5 μm.) (B) Schematic of the experimental approach for performing vapor-phase anion exchange reaction on CsPbBr₃ nanoplates (heating temperature of the hotplate was calibrated as constant within 175–180 °C). (C–F) The different anion exchange status of individual nanoplates (thickness: 352, 374, 352, and 360 nm) with reaction time of 10, 15, 20, and 25 min, respectively. (Left Column) Confocal PL mapping of the nanoplates exhibiting different anion distributions. (Scale bars, 5 μm.) (Center Column) Local emission spectra collected (collection spot <500 nm in diameter) for each nanoplate at the face site (position 1), the edge site (position 2), and the corner site (position 3). (Right Column) Profiles of the iodine ratio (converted from the emission wavelength) along the dashed lines in each of the nanoplates.

A facile approach was developed for controllable vapor-phase anion exchange reaction on the as-grown CsPbBr₃ nanoplates, as schematically illustrated in Fig. 1B (see details in Methods); the iodide vapor transformed CsPbBr₃ to CsPb(Br_1−xI_x)₃ and eventually to CsPbI₃ (12) due to the rapid anion exchange process (7–9, 11). The whole exchange reaction proceeds under a constant temperature quantified as around 170 $°$C for the nanoplates sample itself (see details in SI Appendix). Rapid quenching of the reaction enabled us to capture different anion exchange stages as examined by photoluminescence measurement. As shown in SI Appendix, Fig. S2, the emission color and peak wavelength evolve with anion exchange reaction time. And, the coexistence of green and red emission colors and the emergence of two distinctive photoluminescence peaks indicate the formation of a heterostructure with Br- and I-rich regions on a single nanoplate, implying a diffusion-limited pathway of the anion exchange process. Additional characterizations including SEM and atomic force microscopy (AFM) (SI Appendix, Fig. S3) demonstrate the well-preserved morphology of the nanoplates after anion exchange reaction.

To elucidate the anion diffusion pathway, time evolution studies were performed by confocal microscopic photoluminescence imaging. Confocal laser scanning microscopy in lambda (wavelength) scan mode allows the emission from individual nanoplates to be spatially resolved with each pixel ($\le$500 nm) of the scanned image correlated to a specific emission wavelength (30). The anion exchange at four different intermediate reaction stages was visualized through the evolution of emission color on individual nanoplates (Fig. 1 C–F, Left Column). The emission color evolution indicates decrease in bandgap with increase in I/Br ratio. Based on this observation, we identified a step-by-step pathway of the anion exchange reaction, which initially starts from the peripheral sites of the nanoplates before propagating symmetrically into the center, along with certain crystalline direction of CsPbX₃ structure (suggested by HRTEM and FFT pattern shown in SI Appendix, Fig. S1). To analyze the reaction pathway quantitatively, local emission spectra of three representative positions (face, edge, and corner) were acquired for the nanoplates at each stage to further demonstrate the diffusion pathway (Fig. 1 C–F, Center Column). Composition profiles along each horizontal dashed line across the nanoplates (Fig. 1 C–F, Right Column) were obtained by converting the I ratio at each pixel from emission wavelength following the bowing relation with the bandgaps of CsPb(Br_1−xI_x)₃ (SI Appendix, Fig. S4). These line profiles prove the formation of heterojunction on single nanoplates and further suggest a possible diffusion-determining anion exchange rate.

These observations led to a systematic investigation for a comprehensive picture of the anion exchange kinetics. Based on our further-modified CVT method, we were able to obtain thickness-controllable perovskite nanoplates (SI Appendix, Fig. S5) to examine the effect of thickness on the anion exchange kinetics; we found anion exchange rates varied with plate thickness. A matrix of confocal photoluminescence images in Fig. 2 A–E reveals the time evolution of the anion exchange process for five groups of nanoplates with different thickness ranges (20–40 nm; 50–80 nm; 100–130 nm; 220–260 nm; 350–380 nm). Statistical experiments verified the comparable reaction rates for nanoplates in the same thickness range, thus individual nanoplates can be chosen as the representatives for our time evolution studies. The qualitative observation clearly identified that the anion exchange on nanoplates among different thickness groups follows the similar trend of a symmetrical propagation pathway but with different rates. To quantitatively study the thickness effect on anion exchange kinetics, the time evolution of I ratio at three sites (face, edge, corner) of nanoplates was plotted for each thickness group (Fig. 2F). We found that the reaction time for a complete anion exchange increases with plate thickness; for example, a 20-nm-thick nanoplate requires only 10 min to complete the anion exchange reaction whereas 25 min is needed for a nanoplate thicker than 120 nm. This feature can be attributed to the simple scaling of the size in a diffusion-controlled reaction scheme, providing further support for the proposed anion exchange mechanism.

Fig. 2.

Illustration of the thickness- and time dependence of anion exchange. Confocal PL mapping on individual nanoplates of different thicknesses and reaction times. (A) Thickness range: 20–40 nm (38, 30, 27, 28, and 30 nm, respectively). Reaction times are 0, 3, 4, 7, and 10 min from left to right, respectively. (B) Thickness range: 50–80 nm (65, 73, 64, 55, and 50 nm, respectively). Reaction times are 0, 6, 8, 10, and 15 min from left to right, respectively. (C) Thickness range: 100–130 nm (110, 114, 116, 112, and 127 nm, respectively). Reaction times are 0, 10, 15, 18, and 25 min from left to right, respectively. (D) Thickness range: 220–260 nm (260, 240, 253, 220, and 236 nm, respectively). Reaction times are 0, 10, 15, 20, and 25 min from left to right, respectively. (E) Thickness range: 350–380 nm (362, 352, 374, 352, and 360 nm, respectively). Reaction times are 0, 10, 15, 20, and 25 min from left to right, respectively. (F) Plots of iodine ratio changing with reaction time for the three different sites (face, edge, and corner denoted in Fig. 1) of the nanoplates in identified thickness ranges from top to bottom, corresponding to those in images from A to E, respectively. (All scale bars, 5 μm.)

Motivated by these experimental observations, we performed theoretical analysis and modeling to simulate the anion exchange process. A detailed hypothesis was put forward that the anion exchange in halide perovskites contains two steps: (i) initial surface reaction; and (ii) subsequent solid-state diffusion (Fig. 3A). To evaluate the rates of the exchange reaction and anion diffusion in perovskite structures, a model of one-dimensional (1D) diffusion coupled with surface reaction was constructed to mimic the vertical diffusion channel extracted from the central column of the nanoplates (Fig. 3B). Correspondingly, the experimental information of local emission spectrum from this central column can also be obtained through confocal laser probe. This can be further converted to I ratio and taken as the average value over the entire vertical length, given that the vertical resolution of confocal microscope (∼500 nm) exceeds the plate thickness (Fig. 3B). Analysis of the anion diffusion pathway in this channel is also based on the assumption that the I ratio change is only due to diffusion in the z direction from the top surface in a short time period (not longer than 10 min in our case). We collected the experimental I ratio of the central surface spot for a series of nanoplates with different thickness and same reaction time, as represented by some of those nanoplates shown in SI Appendix, Fig. S6. Theoretically, a modified finite-difference method was utilized for simulation, combined with the first-order reaction mechanism and Fick’s second law in our 1D model, thus giving the theoretical average I ratio with dependence on plate thickness (see details in SI Appendix). From the simulation, a reaction constant k of $\left(1.5\sim 3.5\right)\times {10}^{-3}{\text{s}}^{-1}$ and diffusion coefficient D of $\left(1\sim 5\right)\times {10}^{-14}{\text{cm}}^2\cdot {\text{s}}^{-1}$ can be well fitted with the experiment data (Fig. 3C).

Fig. 3.

Schematics and simulations of the anion exchange process. (A) Schematic of the anion exchange process which contains: (i) surface reaction for anion exchange between n-C₄H₉NH₃I vapor and CsPbBr₃ nanoplates; and (ii) solid-state diffusion of ${\text{I}}^{-}$ in CsPbBr₃ nanoplates. (B) Schematics illustrating the realization of obtaining both the experimental (left side, from confocal laser probe) and theoretical (right side, from 1D diffusion model) value of I ratio at the center of nanoplates surface. (C) Simulation of the relation between I ratio (average) at the center of plates surface versus plates thickness. The dashed area exhibits simulation result with the reaction constant k given as $\left(1.5\sim 3.5\right)\times {10}^{-3}{\text{s}}^{-1}$ and the diffusion coefficient D as $\left(1\sim 5\right)\times {10}^{-14}{\text{cm}}^2·{\text{s}}^{-1}$, well fitted with the experimental data points shown as the hollow dots.

To elucidate the mechanism of anion exchange reaction on perovskite nanoplates, we further analyzed our simulation results alongside experimental observations. The halide diffusion coefficient D on the order of ${10}^{-14}{\text{cm}}^2\cdot {\text{s}}^{-1}$ estimated from simulation is comparable with the values from previous reports and in agreement with the halide vacancy-assisted diffusion mechanism (31, 32). By contrast, the surface reaction constant k is not dependent on spatial parameters, and its high value suggests the surface reaction can be very rapid. We thus concluded the anion diffusion is the rate-determining step that gives rise to the gradual transformation to heterogeneous Br/I mixed perovskites and ultimately to CsPbI₃, enabling the anion exchange propagation to be visualized with the temporal resolution at a minute level. Furthermore, the diffusion length L is approximately proportional to $\sqrt{Dt}$ (t means time) matching the experimental results. For example, a reaction time of 10 min yields an L value of around 40–50 nm in our vertical diffusion channel (z direction) (SI Appendix, Fig. S6). As extended to the in-plane (x, y) direction, such diffusion length (below or around 100 nm) is much shorter than the large lateral size of the nanoplates (typically around 10 μm). Therefore, such inconsiderable diffusion length scale of ions in perovskite nanoplates with high aspect ratio (i.e., width is much greater than height) contributes to the observable anisotropic characteristic that lateral anion exchange pathway outcompetes the vertical way. This leads to the formation of 2D heterostructures with Br- and I-rich separated regions on a single-crystalline nanoplate at intermediate anion exchange stages.

As a proof-of-concept study for the as-formed perovskite heterostructures, we further investigated the evolution of carrier behavior and recombination pathway associated with the anion exchange process. Photoluminescence lifetime imaging microscopy was used as another quantitative imaging technique to probe the carrier dynamics of the perovskite nanoplates throughout the anion exchange process. The photoluminescence lifetime mappings for individual nanoplates at different anion exchange stages were shown in Fig. 4 A–D. No significant changes of lifetime were observed during this transformation, demonstrating that the optical quality of perovskite is retained. Within each of individual nanoplates, the lifetime mappings show homogeneous distribution for both of CsPbBr₃ (Fig. 4A) and CsPbI₃ (Fig. 4D), whereas the heterostructures (Fig. 4 B and C) exhibit strong contrast at different sites. Quantitative fitting to the photoluminescence decay curves with biexponential function at three denoted sites in Fig. 4B identifies the lifetime as 0.524 ns for face, 1.726 ns for edge, and 10.11 ns for corner, respectively (Fig. 4E). The lifetime increase from central to peripheral region of nanoplates is associated with the increasing I/Br ratio (Fig. 4F), which is consistent with our previous report on the photoluminescence lifetime in mixed Br and I perovskites (9). Such increases in lifetime could also be attributed to the carrier flow from the Br-rich to I-rich region driven by the bandgap energy funneling effect (33, 34), due to the formation of coherent heterojunction with built-in halide gradient and bandgap decrease from CsPbBr₃ to CsPbI₃. The carrier dynamics at such heterointerface resulting from the diffusion-limited mechanism of anion exchange may imply areas of interest in photophysics and inspire further investigations.

Fig. 4.

Time-resolved PL imaging for different anion exchange stages. A–D (Top Row) PL lifetime imaging of nanoplates. A–D (Bottom Row) Corresponding confocal PL mapping of nanoplates. The thickness values are 110, 374, 250, and 30 nm, respectively. (A) The starting CsPbBr₃ nanoplate. (B) Heterostructure formed at early reaction stage. (C) Heterostructure formed at later reaction stage. (D) CsPbI₃ nanoplate after conversion from CsPbBr₃. (E) Time-resolved PL decay curves in square dots and their fitting curves in solid lines for the nanoplate in B at the face, edge, and corner sites, respectively. (F) Steady-state PL spectra for the nanoplate in B at the face, edge, and corner sites, respectively. (All scale bars, 5 μm.)

In conclusion, we showed that the anion exchange kinetics in individual halide perovskite nanoplates was visually and quantitatively resolved by microscopic photoluminescence imaging, revealing a diffusion-controlled mechanism that was also validated by theoretical simulations. We further demonstrated that the 2D perovskite heterostructures can be created through the precisely controlled anion exchange approach that we developed here, which would generate fundamental research interest in photophysics, such as the photoinduced carrier dynamics at the coherent heterointerface. Overall, the presented study not only provides a quantitative mechanistic insight into understanding ion exchange behavior in solid-state semiconductors, but also stands out as a creative design by combining the strengths of materials properties and dimension with chemical experiment control and optical microscopic techniques for quantitatively imaging a chemical reaction and the process of 2D heterostructure formation. Such concept and strategy can be further generalized to the study of similar properties in other material systems and various phenomena in the context of physical science.

Methods

Additional details regarding the materials and methods may be found in SI Appendix.

Growth and Vapor-Phase Anion Exchange of Perovskites.

The CsPbBr₃ perovskite nanoplates were grown by a CVT method (28, 35–37). The vapor-phase anion exchange reaction of the as-grown CsPbBr₃ nanoplates was carried out in a capped 20-mL glass vial as shown in Fig. 1B. The iodine source of 10 mg n-C₄H₉NH₃I (n-Butylammonium iodide) powder was first placed inside the vial and was evenly distributed at the bottom. The mica substrate (with CsPbBr₃ nanoplates grown on the top surface) was then placed on top of the evenly distributed tiny powder layer at the bottom of the vial. The thickness of mica substrate is between 50 and 75 μm. Then, the reaction vial was placed on a hotplate (with constant temperature calibrated as within 175–180 °C) for heating; the exact temperature of the sample on substrate has been measured by an infrared pyrometer as maintaining 170 $\pm$ 2 °C during the experiment (see quantification details in SI Appendix). The anion exchange reaction was allowed to proceed between the CsPbBr3 nanoplates and continuously evaporated n-C₄H₉NH₃I vapor for different periods of reaction time to study the time evolution of the reaction. The reaction was effectively quenched by removing the vials from the hotplate and taking the sample out immediately. This entire procedure was conducted in an argon-filled glovebox.

Confocal Laser Scanning Photoluminescence Microscopy.

Confocal laser scanning photoluminescence microscopy was performed using a Carl Zeiss 710 LSM confocal microscope with a 50× 0.6 numerical aperture (N.A.) objective and analyzed by Zen software. A laser excitation at 405 nm was used with the laser power and the electric gain optimized for the dynamic range of the emission intensity from different samples. All images were 512×512 pixels collected at 10 ms per line with a pinhole size of 1 Airy unit. Lambda scans were performed by collecting a series of photoluminescence images while scanning the wavelength range within 470–710 nm with a 5-nm spectral window.

Photoluminescence Lifetime Imaging Microscopy.

The carrier recombination rates and lifetime distributions over the perovskite nanoplates at different anion exchange stages were measured by a time-resolved microscopic imaging system (Zeiss 510 NLO AxioVert 200M). A Ti:sapphire laser system (Spectra-Physics Mai Tai) with a pulse width of less than 100 fs and a repetition rate of 80 MHz was applied to excite the samples through a 50$\times$ objective with 0.6 N.A. at an excitation wavelength of 405 nm (second-harmonic laser wavelength). The illuminating power was controlled by the neutral density filter. The photoluminescence signal was collected by the same objective and filtered by a long-pass filter (488 nm) and a bandpass filter (530 nm/30 nm) before entering a controllable pinhole with a diameter of 50 μm in front of a Hamamatsu photomultiplier tube. The lifetime images were taken with an acquisition time of 60 s.