Mapping emission heterogeneity in layered halide perovskites using cathodoluminescence

Recent advancements in the fabrication of layered halide perovskites and their subsequent modification for optoelectronic applications have ushered in a need for innovative characterisation techniques. In particular, heterostructures containing multiple phases and consequently featuring spatially defined optoelectronic properties are very challenging to study. Here, we adopt an approach centered on cathodoluminescence, complemented by scanning electron microscopy coupled with energy-dispersive x-ray spectroscopy analysis. Cathodoluminescence enables assessment of local emission variations by injecting charges with a nanometer-scale electron probe, which we use to investigate emission changes in three different systems: PEA2PbBr4, PEA2PbI4 and lateral heterostructures of the two, fabricated via halide substitution. We identify and map different emission bands that can be correlated with local chemical composition and geometry. One emission band is characteristic of bromine-based halide perovskite, while the other originates from iodine-based perovskite. The coexistence of these emissions bands in the halide-substituted sample confirms the formation of lateral heterostructures. To improve the signal quality of the acquired data, we employed multivariate analysis, specifically the non-negative matrix factorization algorithm, on both cathodoluminescence and compositional datasets. The resulting understanding of the halide replacement process and identification of potential synergies in the optical properties will lead to optimised architectures for optoelectronic applications.


Introduction
Over the past two decades, lead halide perovskites have attracted attention owing to their exceptional optoelectronic properties [1][2][3].Recent research endeavors have been Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence.Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
primarily dedicated to advancing the synthesis techniques for perovskite materials with reduced dimensionality-i.e. with different connectivity of the basic perovskite octahedral unit.This led from the synthesis of the traditional three-dimensional perovskite structure to two-dimensional crystalline structures, such as nanosheets, and even to so-called 0D perovskites [4], through a variety of synthetic approaches.
The emergence of hybrid organic/inorganic two-dimensional (2D) perovskites, also known as 'layered perovskites', has, in particular, ignited significant interest within the scientific community.This interest stems from their remarkable optoelectric features, including enhanced fluorescent emission compared to their 3D counterparts [5,6], which renders them particularly attractive for a wide range of technological applications.
2D perovskite structures extend the compositional versatility inherent to their three-dimensional (3D) counterparts by the additional choice of the organic cations [7].However, they exhibit a distinctive structural arrangement, characterised by the sequential alignment of two types of layers: one consisting of corner-sharing [BX 6 ] 4− octahedra, and one composed of organic molecules.These organic molecules have a lower dielectric constant and higher exciton binding energies compared to the octahedral layer [7,8].As a result, carefully choosing the organic components and optimising the stoichiometry of lead halide perovskites becomes critical: such choices and finetuning play a pivotal role in enhancing the optoelectronic characteristics and efficiency of these materials [9][10][11][12][13].
The selection of the halide anion (X) plays a pivotal role in defining the crystalline lattice of the material, consequently defining the electronic properties [14].Remarkably, it offers the opportunity to fine-tune the band gap of a lead halide perovskite through various approaches either during-or postsynthesis.These include the use of different stoichiometric ratios between two precursor halides, as well as atomic substitution of the halide in a second reaction step [15].This capability to modulate the band gap [16] showcases the versatility of lead halide perovskite materials [17].
In this work, we studied the emission properties of established pure PEA 2 PbBr 4 and PEA 2 PbI 4 microsheets, but also performed a halide exchange process on PEA 2 PbBr 4 , involving the substitution of bromide anions with their iodide counterparts.This exchange was undertaken with the aim of generating a PEA 2 PbBr 4 -PEA 2 PbI 4 core-frame heterostructure.The resulting optical properties are expected to correlate with the chemical distribution, and therefore we employed hyperspectral cathodoluminescence (CL) to probe them with a nanosized electron beam.In our investigation, we collected the CL signal generated in a scanning electron microscope (SEM-CL), measuring the photons emitted by the different perovskite phases.While photoluminescence (PL) is commonly used to study the local optical properties of nanomaterials and specifically layered perovskites, CL has not been extensively employed in this field yet and, being based on a fundamentally different mechanism to excite the sample, can provide complementary information to optical techniques.
To mitigate the potential damage inflicted by an electron beam (knock-on damage, inelastic scattering, and localised heating), which is known to be a common issue for 2D materials and halide perovskites, we employed a pulsed electron beam.This mode allowed us to obtain sufficient CL signal using currents of tens to hundreds of picoamperes [18,19].To maximise the information output from hyperspectral maps and to generate high-resolution emission maps we applied MultiVariate Analysis (MVA), which can be used to significantly improve the signal-to-noise ratio (SNR) as well as highlighting correlations in the local spectral features [20].This systematic approach facilitated the identification and spatial localisation of the different components within the heterostructure and provided evidence of the successful halide exchange process.It is worth noting that the MVA approach demostrated here can also be extended to other hyperspectral imaging techniques, for example based on PL.

Material fabrication
Lead(II)bromide (PbBr 2 , 98%), lead(II)iodide (PbI 2 , 99%), hydrobromic acid (HBr, 48%), hydroiodic acid (HI, 57%, distilled, 99.999% trace metal basis), hypophosphorous acid (H 3 PO 2 , 50%), phenethylamine (PEA, 99%), phenethylammonium iodide (PEAI, 98%), acetone (99.5%), ethyl acetate (99.5%),N, N-dimethylformamide (DMF, 99.8%, anhydrous), toluene (99.7%), 1-octanol (99%), were purchased from Sigma-Aldrich without any further purification.PEA 2 PbBr 4 microcrystals were obtained using a previously published protocol with minor modifications [21].Specifically, 92 mg (0.25 mmol) of PbBr 2 was dissolved in 200 μl of HBr, followed by dilution with 2 ml of acetone.Subsequently, 75 μl (0.6 mmol) of PEA was injected into the solution, resulting in the immediate formation of PEA 2 PbBr 4 microcrystals with thicknesses of few microns and lateral sizes of some tens of microns.The reaction mixture was continuously agitated for a minimum of 3 h to ensure a complete reaction.The microcrystals were then separated from the solution by centrifugation at 6000 rpm for 2 min, followed by redispersion in 2 ml of acetone.This washing procedure was repeated at least two more times, and the purified microcrystals were dried under vacuum for 1 h.For the synthesis of PEA 2 PbI 4 , the same procedure was applied, with the substitution of PbBr 2 and HBr for PbI 2 (115 mg, 0.25 mmol) and a mixture of HI (200 μl) and H 3 PO 2 (108 μl), respectively.Furthermore, to account for the higher solubility of PEA 2 PbI 4 in acetone, ethyl acetate was used as the solvent.For CL, SEM, Energy-Dispersive x-ray (EDX) spectroscopy analysis and height profile measurements the as-prepared PEA 2 PbBr 4 and PEA 2 PbI 4 microcrystals were exfoliated by one-sided 3M scotch tape on a precleaned silicon substrate.The formation of the PEA 2 PbBr 4 -PEA 2 PbI 4 lateral heterostructures was initiated by exposing the silicon substrate with the exfoliated PEA 2 PbBr 4 flakes to a 0.1 mol l −1 solution of PEAI in octanol.After 5 min the substrate was washed thoroughly with toluene to remove any excess of the PEAI solution.

Electron microscopy
SEM imaging and compositional mapping via EDX was performed in a Zeiss Gemini SEM 560 (Zeiss, Oberkochen, Germany) equipped with a field-emission gun, operating at 30 kV acceleration voltage.EDX data (Oxford Instruments, X-Max, 80 mm 2 ) were acquired operating at 30 kV.

CL hypermapping was performed in an Attolight Allalin 4027
Chronos SEM-CL.The spectra were acquired with an iHR320 spectrometer (focal length of 320 mm, 150 gratings per mm blazed at 500 nm, 700 μm entrance slit) and an Andor 1024 pixel charge-coupled device (readout rate of 50 kHz MHz and 100 ms acquisition time per pixel).All the measurements were performed at room temperature under high vacuum (<10 −7 mbar) with an acceleration voltage of 10 kV and a pulsed beam with a frequency of 80 MHz, a pulse duration of ∼7 ps and a current of approximately 100 pA as measured by a Faraday cup.

Profilometry
The height profile of individual exfoliated PEA 2 PbBr 4 microsheets were measured by an optical profilometer (Zeta-20 by ZETA).

Photoluminescence spectroscopy
The PL spectra of the microcrystalline powders were acquired using an Edinburgh Instruments fluorescence spectrometer (FLS920).The PL spectra of the microcrystalline powders were collected by exciting the samples at 350 nm.

Data processing
Hyperspy [22] and Lumispy [23], both open source Python libraries, were used for the management and processing of data.

Scanning electron microscopy and optical profilometry
We carried out an initial SEM characterisation in order to understand the general morphology of the microsheets and the local elemental distribution.The pure phases (figures 1(a)-(b) comprise fairly flat, homogeneous flakes with a regular profile with tapered edges, and some visible surface imperfection due to layering.In the secondary electron (SE) image (figure 1(c)) a PEA 2 PbBr 4 -PEA 2 PbI 4 lateral heterostructure exfoliated microsheet is shown to feature numerous thin («um) plates that are stacked on top of each other and have irregular borders.The roughness at the edges, while occasionally present in the pure phases, appears enhanced after the anion replacement process.The top surface also presents a texture suggesting incomplete layers terminating the crystal.
To optimise SEM-CL parameters and understand the interaction volume within the sample, we used an optical profilometer to measure the thickness of a set of representative PEA 2 PbBr 4 crystals.We found an average height of 2.4 μm, with values varying between 0.4 and 4.1 μm (figure S1).While the edge is sharp on the μm scale, the morphology is rougher on the nanoscale, with the edges of the single flakes presenting a jagged geometry.The structure of the flakes and their stacking is expected to play a role in the atomic exchange process, as local defects can provide channels for the infiltration of halides.
We carried out SEM-EDX measurements to locally investigate the halide replacement (figure 1).To optimise the EDX signal, we denoised the acquired hyperspectral map using non-negative matrix factorization (NMF).The NMF algorithm plays a crucial role in extracting physically significant components from any type of dataset, yielding results with high contrast and specific spatial distributions, and can decompose the spectra into factors that can be related to the chemical composition [24].As can be clearly seen from figure 1(c), the halide exchange between bromide and iodide is occurring almost exclusively along the edge of the crystals, as can be expected given the 2D nature of the material.It is important to note, however, that the dominant EDX signal originates in depth in the sample, and the sensitivity of this approach to replacement on a thin surface layer is low.We first studied the CL emission in the pristine PEA 2 PbBr 4 and PEA 2 PbI 4 samples.As heterogeneity in emission of other lead halide perovskite crystals [16,25,26] has been reported, this provides insight into the effect of edges and morphological features on the optical emission.We used an acceleration voltage of 10 kV in order to avoid surface effects [27], and an average integrated pulsed beam current of approximately 100 pA.The hyperspectral maps clearly reveal a distinct spatial distribution of the emission from both the PEA 2 PbBr 4 and PEA 2 PbI 4 crystals.The edges of crystals predominantly exhibit strong radiative emission, while the central region displays a lower and spectrally different emission (figure 2).In the following we will refer to the component centered at 407 nm (bromide) or 528 nm (iodide) as 'primary emission' for each sample.We will refer to the second component (at longer wavelength) for each sample as 'secondary emission'.
Similar spectral line shapes have also been found in integrated photoluminescence (figure S3), where the spectral information was integrated over a large area.In the case of the iodide pure phase a splitting of the emission peaks is clearly visible and matches with the CL observation, although with a slight shift in absolute wavelength which is commonly observed in comparisons between the two techniques.The PEA 2 PbBr 4 PL emission also appears broad as in CL, and without the clear presence of two distinct peaks.
The complexity of these CL and PL signals is challenging to model effectively.A single Gaussian fit is unable to correctly fit the spectral emission profile over an entire crystal for each material due to the combination of these two emission components.
In order to identify physical components without operator bias, we applied the NMF algorithm to the CL datasets [24,28].This algorithm for both PEA 2 PbI 4 and PEA 2 PbBr 4 decomposes the emission profiles (figure S4) into two separate components (figures 2(g), (h): one is characterised by a sharp profile, while the other factor is asymmetric.The second component, which we refer to as secondary emission, in both cases contributes to the enlargement and apparent shift of the emission maxima towards longer wavelengths (a red shift).
Indeed, as depicted in figure 2, one factor represents the primary emission of PEA 2 PbBr 4 and PEA 2 PbI 4 , corresponding to wavelengths of 407 nm (3.04 eV) and 528 nm (2.34 eV) respectively (value estimate with a correct fitting in figure S5), revealing a common behaviour for the two halides.In both cases, these emissions are prominently concentrated along the edges of crystals, consistently with existing literature [29].
To unravel the complexity of this effect, resulting in two characteristic signals localised at the edges and in the bulk respectively, it is important to note that the CL process includes several steps, all of which affect the measured signal: (1) interaction of the primary electron with the sample, (2) carrier diffusion, (3) carrier radiative recombination and (4) extraction of the generated photon.There are therefore several potential phenomena that can result in the higher emission along the edges, depending on both the chemical composition and the morphology of the microsheets under examination.Furthermore, perovskite microsheets are particularly susceptible to the high-energy electron beam, which can induce damage through chemical alterations, sample heating, and the introduction of defects [18].The effects of beam damage can lead to a variation in optical emission, together with intrinsic factors [18].For example an inhomogeneous distribution of halide species (from synthesis, ageing or beam damage) can have an effect on the spatial heterogeneity of the CL signal.The local radiative recombination efficiency can also depend on the presence of defects or morphological features, such as the rough exposed edges of sheets that constitute the edges of crystals.It should be noted that in 2D perovskites the lateral diffusion length for carriers typically reaches micrometers [30,31], and this diffusion is anisotropic (longer diffusion length in-plane compared to out-of-plane), so that the carriers can easily reach recombination sites along the edges of the sample [29,30,32].This, combined with the complex morphology of the microsheets, can be directly responsible for some of the spatial heterogeneity in the emission, leading to the difference in primary and secondary emission signal.The re-absorption of photons within the perovskite should also be taken into account: radiative recombination of charges in depth within the crystal is unlikely to result in photons that are then collected.The self-absorption can affect the spectral profile of the measured CL signal, with the high-energy photons being filtered out; this phenomenon could potentially explain why the secondary emission, originating from the bulk, has a red-shifted profile compared to the primary emission.
To understand the volume that is sampled in the CL analysis, we simulated the interaction of the electron beam with the samples using CASINO [33].As shown in figure S6, for both bromide and iodide microsheets, the interaction volume is smaller than the microsheet thickness mesured by profilometry.Emission mostly originates from a depth of ∼100-500 nm within the microsheets and therefore the properties being probed are those of the bulk of the crystals.
To delve into the origins of the secondary emission, we conducted a comprehensive investigation in areas where the nanoplates exhibited increased thickness and irregular morphology, as observed in the SEM images (figure 3).Applying the NMF algorithm to these regions, it became apparent that the secondary emission was typically localised along thicker regions and areas with thickness variations.This emission can thus be ascribed to the effects described above, including a set of several phenomena (such as presence of defects and different geometries for photon extraction) linked to the irregularity of the flakes resulting from the exfoliation.It is also possible that the increased intensity of the CL signal in regions that are thicker, and have strong thickness variations, is caused by the carriers excited in-depth which are able to move laterally and recombine in areas that are shallower (closer to the top surface), from which photons are more likely to be collected.Electrons injected in thick areas can thus cause luminescence in neighbouring, thinner areas which are at the same depth in the sample as the volume where carriers are generated.As the CL signal intensity is assigned to the pixel being scanned, while the photons are collected from the entirety of the sample, this sequence of events would assign a strong luminescence to areas that are higher (thicker) then their surroundings.We employed a similar approach to analyse samples after the halide exchange process, using the same experimental parameters.As for the pristine case, a general qualitative agreement was found between the CL data (figure 4) and the PL data (figure S3).We performed NMF analysis on CL hyperspectral maps to discern the emission from each phase, as well as any variation internal to a single phase that has been observed in PEA 2 PbBr 4 and PEA 2 PbI 4 .Upon applying the NMF algorithm to this dataset, we identified three principal components.
The first two components are the emissions of perovskite containing bromine and iodine, closely resembling those observed in pure samples.The third component modeled the secondary emissions, which visibly altered the shape and wavelength of the emission band associated to the PEA 2 PbBr 4 phase.The PEA 2 PbI 4 phase does not show a secondary emission when in the heterostructure, as it is localised along the edges, where the secondary component is weaker.Compared to the pristine phases, we observed a slight redshift in the wavelength of PEA 2 PbBr 4 emission (from 407 to 410 nm) and a corresponding blue shift in the emission of PEA 2 PbI 4 (from 528 to 522 nm).
Figure 5 clearly illustrates that the PEA 2 PbBr 4 emission originates from the bulk of the material as well as the edges, and is homogeneous in the absence of morphological features.Despite the presence of a secondary emission component (orange curve in figure 4(a)), the strongest emission still predominantly occurs along the crystal's edge.Notably, such secondary emission is attributed to the bromide component, as PEA 2 PbBr 4 still constitutes most of the structure, with the PEA 2 PbI 4 phase being confined along the edges of the microsheets.In order to accurately evaluate the presence of PEA 2 PbI 4 emission across the crystal border and the co-presence of bromide signal, we extract a line profile by integrating over a region of interest (ROI).As reported in figure 5, strong CL features from both phases are visible in a region of ∼500 nm from the surface.Further inside the crystal, the iodide emission ceases, while the bromide phase is still emissive.
In accordance with SEM-EDX, this demonstrates the copresence of bromine and iodine in a region next to the edges; it also proves that both the remaining bromide phase and the new iodide phase are optically active, thus demonstrating the potential of CL to monitor the halide exchange processes.Further evidence of the formation of a PEA 2 PbI 4 phase is provided by x-ray diffraction (figure S8).
The localisation of this exchange process along the edges is a consequence of the crystal structure of these perovskites.The octahedra on which the halogen exchange is to take place are passivated by the large organic cations along the vertical direction.As a result, the exchange rate is considerably faster along the lateral direction where the octahedra do not undergo passivation by organic molecules.
This type of behavior is in line with in-gas-phase exchange methods using HBr and PEA 2 PbI 4 vapour as the starting material [34].

Conclusions
In summary, the integration of cathodoluminescence spectroscopy with scanning electron microscopy has unveiled heterogeneous optical emission from pure-phase layered halide perovskites, as well as demonstrating the effects of partial anion exchange between bromide and iodide within perovskite microsheets, resulting in a lateral heterostructure.This study demonstrates not only the efficacy but also the robustness of cathodoluminescence as a characterisation tool.As technologically-relevant materials, perovskites often present design challenges in order to control phenomena such as phase segregation and ion mobility.The fine sensitivity of CL to local optoelectronic properties and an easy integration with SEM can help in understanding fundamental material properties and engineer device properties.Furthermore, due to its high spatial resolution, CL can typically access single-grain information more easily than optical techniques, and pulsed sources enable the study of charge dynamics on ultrafast timescales.Finally, this study highlights the potential for harnessing machine learning techniques to derive valuable insights from hyperspectral data, thus opening up exciting avenues for future research and analysis for complex multiphase optoelectronics systems.Skłodowska-Curie Funding Program (Project Together, Grant agreement No. 101067869).The authors would like to thank Jacopo Stefano Pelli Cresi for discussions on data processing.

Figure 1 .
Figure 1.(a)-(b) SEM images of (a) PEA 2 PbBr 4 and (b) PEA 2 PbI 4 microsheets.(c) SEM image of a microsheet after exposure to the PEAI solution.A faceted, rough morphology is visible at the edges.EDX analysis evidences the formation of PEA 2 PbBr 4 -PEA 2 PbI 4 lateral heterostructures with the iodide phase at the edges and the bromide phase in the bulk.

Figure 2 .
Figure 2. (a)-(b) Secondary electron images of microsheets.(c)-(f) Spatial distribution of the CL emission in the sample for the two NMF decomposition factors.(g)-(h) NMF decomposition components from CL hypermaps.

Figure 3 .
Figure 3. (a) Secondary electron image of the investigated PEA 2 PbI 4 crystal.(b) NMF factors within the ROI (dashed yellow rectangle).(c)-(e) Spatial distribution of the CL emission intensity.

Figure 5 .
Figure 5. (a) Spatial distribution of emission for PEA 2 PbBr 4 (purple) and PEA 2 PbI 4 (green), with the region of interest (ROI) in yellow.(b) Emission intensity in the ROI across a line profile through the edge (y axis: distance along line profile, x axis: wavelength).(c) Plot of emission profile as a function of depth in the microsheet.