Structural and chemical properties of anion exchanged CsPb(Br(1−x)Cl x )3 heterostructured perovskite nanowires imaged by nanofocused x-rays

Over the last years metal halide perovskites have demonstrated remarkable potential for integration in light emitting devices. Heterostructures allow for tunable bandgap depending on the local anion composition, crucial for optoelectronic devices, but local structural effects of anion exchange in single crystals is not fully understood. Here, we investigate how the anion exchange of CsPbBr3 nanowires fully and locally exposed to HCl vapor affects the local crystal structure, using nanofocused x-rays. We study the nanoscale composition and crystal structure as function of HCl exposure time and demonstrate the correlation of anion exchange with changes in the lattice parameter. The local composition was measured by x-ray fluorescence and x-ray diffraction, with general agreement of both methods but with much less variation using latter. The heterostructured nanowires exhibit unintentional gradients in composition, both axially and radially. Ferroelastic domains are observed for all HCl exposure times, and the magnitude of the lattice tilt at the domain walls scales with the Cl concentration.


Introduction
Metal halide perovskites (MHPs) have been intensively investigated by the scientific community in the recent years, owing their impressive potential on applications ranging from solar cells, light-emitting diodes, and photodetectors [1][2][3][4][5].The additional degrees of freedom achieved in MHP nanocrystals result in improved brightness and narrow-band photoluminescence (PL) quantum yield [6][7][8].A key role integrating optoelectronic devices is reserved for MHP nanowires [9,10], which present directed emission with tunable wavelength, depending on their size and local composition [11].CsPbBr 3 is one MHP compound that exhibits high stability, making it especially promising for industrial use [12,13].
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.
Intense effort has been employed on the search of new processing methods for devices based on CsPbBr 3 nanowires [14].There is an extra interest of the scientific community on heterostructures, which opens a new degree of freedom for controllable emission wavelength.It is possible to fully or locally replace the halide component with anion exchange, which leads to a controllable band gap depending on the local ionic ratio [15][16][17].There are different methods for anion exchange, ranging from the use of organic halide salts solutions [18] to pyrophoric halide silanes [19] and solid-solid diffusion [20], but difficulties with synthesizing, handling and the timescale of these processes usually make them challenging for industrial use.Alternatively, gas-based processes, in particular HCl and Cl 2 vapour for CsPbBr 3 chlorination, represent a fast and controllable room-temperature way of anion exchange [21][22][23].In particular, HCl offers a low cost and straight forward process, making it promising for large scale production.By setting the exposure time, one can control the anion composition and control the band gap to cover a large portion of the visible spectrum [23].
Heterostructured CsPb(Br (1−x) Cl x ) 3 nanowires can be produced in a lithographic process, by using a polymer to protect parts of the nanostructure while the rest is exposed to HCl vapor [14].A complication is that MHPs are known to present ferroelastic properties due to their low symmetry [24][25][26].The presence of ferroelastic domains has been recently reported on CsPbBr 3 nanowires, which can affect their potential photoelectronic applications [26,27].Effects of local anion exchange on the crystal structure of the nanowires, including the possibility of ferroelastic domains induced by the chemical inhomogeneity, are so far unknown.
Studying domains in nanostructures is challenging, since it requires a technique able to probe the crystal structure with good spatial resolution, but also high strain sensitivity.Recent advances in x-ray focusing optics, combined with the improved brilliance of new fourth generation synchrotron sources, turned scanning nanofocused x-ray diffraction (nano-XRD) into a powerful tool to study objects in the nanoscale, being able to image strain and lattice tilt with resolution in the order of tens of nanometers [27].With the concomitant analysis of the x-ray fluorescence, one can combine nanofocused scanning x-ray fluorescence (nano-XRF) with nano-XRD to correlate the local chemical composition with changes in the crystal structure [28,29].Here, we use this approach to study the effects of total and local chlorination on the structure of CsPb(Br (1−x) Cl x ) 3 nanowires.The properties of nanowires exposed at times ranging from 0 to 45 s, as well as when locally covered by a polymer during anion exchange, were systematically investigated.

Methods
An anodized aluminum oxide (AAO) matrix with cylindrical pores of 200 nm diameter was used as a template for growing CsPbBr 3 nanowires as previously described [30,31].The AAO was placed on top of a liquid precursor solution of CsBr and PbBr 2 dissolved in dimethylsulfoxide (DMSO) until CsPbBr 3 nucleated inside the pores.When the pores were filled, free-standing nanowires grew outside of the template, as shown in figure 1(a).The free-standing nanowires are almost square in their cross-section despite the cylindrical pores, as evidenced in figure 1(b).They present lengths ranging from 1 to 10 μm and typically have hemispherical tips.Lab source XRD and transmission electron microscopy (TEM) have shown that they grow with 001 planes orthogonal to the long axis [30,31].
To prepare for synchrotron experiments, a cleanroom tissue was used to scrape the surface and transfer nanowires from the AAO matrix to fresh Si 3 N 4 membranes.The nanowires were then exposed to HCl vapor and Cl atoms diffused into the structure, replacing Br and forming CsPb(Br (1−x) Cl x ) 3 .To create heterostructures, some of the nanowires were first covered with PMMA which was locally uncovered using electron-beam lithography (EBL) in a process based on non-polar solvents [14].A schematic representation of the nanowires processing can be seen in figure 1(c).In a diffusion model of the anion exchange process, the diffusion can occur from the surface of the nanowire to the center, resulting in a radial gradient.After processing, there can be further ion migration from chlorinated segments to unchlorinated segments [23].
We performed two synchrotron nanobeam experiments.The first set of measurements was performed on a series of nanowires fully exposed (FE) to HCl vapor for different times ranging from 0 to 45 s (FE-00, FE-15, FE-30, and FE-45).This experiment was performed at the I14 beamline of the Diamond Light Source (United Kingdom), with the energy of the incident beam fixed at 14 keV, beam flux of 1 × 10 9 photons s −1 and an area Excalibur 3M x-ray detector positioned 280 mm from the sample [32].The x-ray beam was focused to near 50 nm × 50 nm (vertical x horizontal) spot by using a set of KB mirrors.A schematic representation of the experimental setup can be seen in figure 2(a).The nanowire axis was set along z, as defined in the inset, and the x-ray detector was positioned at the 2θ Bragg condition of the 004 reflection of the CsPbBr 3 orthorhombic (Pbnm) phase.The nanowire sample was scanned in the focus using a piezo stage.A fixed step size of 50 nm was set on both x and y directions for map acquisitions, while counting time per pixel was set at 25 ms.By scanning the sample at different incidence θ angles around the Bragg condition, i.e. a rocking curve, we could reconstruct the three-dimensional reciprocal space associated with each position along the nanowires, thereby creating high spatial resolution images of the lattice parameter, as well as lattice tilts both around the optical axis and along the rocking angle [27,33].The angle of incidence was scanned in steps of 0.05°for the reciprocal space mappings.The real and reciprocal space coordinates, as well as lattice tilts α (rotation around q y ) and β (rotation around q z ), are defined in the inset of figure 2(a).A four-element silicon drift detector (SDD) was used to acquire the XRF spectra at the same time, allowing for real space maps of the nanowire with chemical contrast.
The second set of x-ray measurements were performed on the heterostructured nanowires at the NanoMAX beamline, MAX IV synchrotron, Sweden.In this case the energy of the x-ray beam was set at 15 keV, with beam flux at 1.7 × 10 9 photons s -1 and a Merlin x-ray detector positioned 300 mm from the sample [34].A set of KB mirrors were used to focus the beam down to 60 × 60 nm 2 (vertical x horizontal) spot size.Maps were acquired using fixed steps of 60 nm both in x and y, and 0.05°in θ.For each pixel, the acquisition time was set as 10 ms.Spatially resolved compositional maps could be retrieved via XRF acquisition.Data treatment of the XRF made use of the peak fitting function of the multiplatform Python-based code PyMca [35].Scans acquired for all incident angles were re-aligned and XRF was summed up to improve statistics.

Results and discussions
The maps of lattice spacing and both lattice tilts for the FE-00 (reference) nanowire are depicted in figure 2(b).One can see that, besides a small region in the right tip, the lattice constant along the probed direction is nearly unchanged for the whole structure, presenting values around 11.74 Å, as expected for CsPbBr 3 .Although some features are visible on both α and β maps, there is no clear pattern.This behaviour has been reported before for pure CsPbBr 3 nanowires at room temperature [27].Complete lattice spacing and tilt maps of FE-00, FE-15, FE-30 and FE-45 nanowires are shown in figure S1 of the Supplementary Material (SM).
It is expected that nanowires exposed for longer time present higher Cl/Br ratio, and therefore the Bragg peak retrieved for each nanowire can be shifted with respect to the others.Figure 2(c) plots the overall rocking curve of the nanowires, sacrificing spatial resolution by summing up the diffracted intensity from the structures.One can notice that the center of mass position is different for each rocking curve, and structures with longer exposure time present lower lattice spacing, i.e. higher Q, as expected.The green and blue dashed lines in figure 2(c) mark theoretical lattice spacings for pure CsPbBr 3 and CsPbCl 3 , respectively.Although FE-00 rocking curve aligns with pure CsPbBr 3 , as expected, even the most chlorinated nanowire, FE-45, does not reach pure CsPbCl 3 .Thus, complete exchange requires more than 45 s exposure.
Figure 2(d) plots the corresponding lattice parameter of all four nanowires alongside the average Br concentration, as calculated via XRF peak fitting.Concentrations are obtained by normalizing the mass fraction using Pb as reference, and assuming Br/Pb = 3 for the pure CsPbBr 3 nanowire.Both ordinate axes in figure 2(d) were set to scale according to theoretical CsPb(Br (1−x) Cl x ) 3 lattice parameter values retrieved from Vegard's law [36].The correlation between lower Br, and therefore higher Cl, concentration with lower lattice spacing is clear.Although we also collected the Cl K α1 XRF data, this signal was too weak to be useful due to the lower emission energy (2.622 keV), and therefore lower XRF yield, compared to Br K α1 (11.924 keV).In general, the XRD and XRF curves match well.Note that the time dependence is not linear, which is consistent with a process that is limited by solid diffusion as previously reported [23].The partially exchanged nanowires at 15 and 30 s show significantly broader XRD peaks than the 0 and 45 s ones, presumably because there is a larger radial variation of composition.
A locally exchanged nanowire was selected for nano-XRD and nano-XRF studies in a second experiment.Figure 3(a) shows the lattice parameter (upper panel), α (middle panel) and β (bottom panel) maps of the heterostructure.A lattice spacing gradient can be seen from the chlorinated (left) side to the nominally pristine (right) side of the nanowire.The α map shows smooth variations along the structure, with striped domains of different sizes rotated around the optical axis.The very left tip presents a highly tilted domain.The striped pattern can be also seen in the β map, although the spatial position of domains does not completely coincide on both tilt maps.
Nano-XRF measurements performed concomitantly to nano-XRD also allows for chemically resolved maps of the nanowires.Figure 3(b) shows the Br concentration map of locally exchanged nanowire.A Br concentration gradient is clear from the chlorinated to the non-chlorinated part.The local lattice parameter (black curve) and Br concentration (green curve) along the nanowire axis, averaged along the nanowire radius, are plotted together in figure 3(c).The two methods are in good agreement, but XRD shows much less variation.A similar analysis can be done to retrieve the average lattice spacing and Br concentration along the radius of the nanowire.The lattice parameter map in figure 3(a) indicates that diffusion may not occur homogeneously in the cross-section, but instead a core-shell like profile appears near the junction.Curves showing the lattice spacing and Br average along x, plotted along y, can be seen in figure 3(d).We have used data from the inside the black dashed box as marked in figure 3(a). .By scanning the beam with the sample, maps were acquired for each angular position.The scattering vector components q x , q y and q z , as well as lattice tilts α (rotation around q y ), β (rotation around q x ) and γ (rotation around q z ) are defined in the inset.The x-ray fluorescence detector position is in accordance with NanoMAX setup, while it is placed in backscatter geometry at I14. X-ray beam spot and maps step size were set at 50 nm for the experiment performed at I14 and 60 nm at NanoMAX.The heterostructured nanowire does not show a sharp junction, but rather axial and radial gradients.A second heterostructured nanowire gave a similar result, as one can see in the SM (figure S2).Core-shell heterostructures have previously been reported in nanowires exposed to HCl [21] and Cl 2 [37], and we have previously reported that the rate limiting step in the exchange process is solid diffusion [23].The core-shell structure presents higher bandgap on the outside, meaning that electrons and holes will collect in the core.Intentional core-shell nanowire structures, which can present enhanced emission efficiency, have been reported in other material systems [38].We see three possible unintentional paths for the Cl: (1) through the PMMA, (2) through a gap between the PMMA and the nanowire, and (3) through the nanowire after processing.The first option should not lead to an axial gradient in the nominally protected segment, while the third option should not lead to a radial gradient.Therefore, we tentatively assign this to Cl diffusion in an unintentional air gap between the PMMA and the nanowire after the EBL step.This could appear due to locally enhanced electron beam dose to back scattering from the high-Z nanowire, in which case it should be possible to overcome with a slightly lower dose.We have previously demonstrated significantly sharper heterojunctions [23,37].Nanofocused x-rays can help guide the development of this sensitive process.
The composition change in the nanowire could lead to changes in the domain pattern along the nanowire.It is notable in figure 3(a) that both α and β maps present vertical striped domains along the nanowire axis, with sizes in the order of a few hundreds of nanometers.This is a signature of the ferroelastic nature of this material, which has been previously described by our group and others [26,39,40].Subsequent domains in the β show stepped tilts of about 0.20°.Although domains are visibly vertical, the tilt switching does not immediately take place but instead a blurry region can be seen in all domain interfaces.This is an indication that domain wall planes are rotated around z, as defined in figure 2(a).This is consistent with (112)-type domain walls seen in similar pure CsPbBr 3 systems [26,27,40,41].For this kind of domains, a rotation of the orthorhombic unit cell around x leads to lattice mismatch on both directions perpendicular to the nanowire axis, where a lattice tilt is expected from the length difference between a and b cell directions of this crystalline phase.for longer exposure times.This is expected for homogeneously exchanged nanowires following Vegard's law, as the ratio between a and b would lower as the nanowire approaches to pure CsPbCl 3 (a = 7.9136 Å, b = 7.9145 Å, c = 11.1861Å) [36].The average domain lattice tilt for nanowires FE-00, FE-15, FE-30 and FE-45 are plotted in figure 4. The non-chlorinated reference nanowire shows randomly oriented domains, and a lattice tilt in the order of 0.35°is noted near its right tip.For this nanowire, a large part of the tilt takes place around the optical axis, and therefore is seen in the α map.For the chlorinated ones a striped domain pattern arises in the β maps, and a clear trend can be seen, with highest average lattice tilt on FE-15 and lowest value on FE-45.The gap between a and b in the CsPb(Br (1−x) Cl x ) 3 nanowires decreases with x, and in the limit of a fully chlorinated object domain tilts would be expected to reach ω = 0.01°, in contrast to the the pure CsPbBr 3 one.Theoretical values of the expected domain tilt for the time series nanowires, calculated using the chemical composition extracted from the XRF data, are also plotted in figure 4 (green squares).The corresponding error bars have been carried from the uncertainties in composition, as seen in figure 2(d).Experimental and theoretical data follow the same trend and values match in the error range, indicating smooth diffusion of Cl during the exchange process.Thus, we find that the ferroelastic tilts scale with the composition.

Conclusion
In summary, we have studied structural properties of CsPbBr 3 nanowires fully and locally exposed to HCl vapor.Nano-XRD and nano-XRF techniques were used to retrieve information regarding the crystal lattice spacing and tilts, as well as the local chemical composition of the nanowires.The composition retrieved from XRD and XRF show agreement, but with lower variation for XRD.The Cl content increases in the HCl exposure time series, as expected, and with a sublinear dependence that fits with a diffusion-limited process.The nanowire exposed for the longest time (45 s) still had about 20% Br.We observe ferroelastic domains in all the exposed nanowires, with a magnitude of crystal tilt that follows Vegard's law.One locally exchanged heterostructured nanowire, half protected from the HCl exposure by lithography, was also investigated.Rather than the desired sharp heterojunction, the images show unintentional gradients in both the axial and radial directions.These results show that nanofocused x-ray methods can be used to image the local composition and crystal structure perovskite nanowires, guiding the development of the anion exchange processes.Z Zhang https:/ /orcid.org/0000-0003-0678-1699H Chen https:/ /orcid.org/0000-0002-5122-486XD Dzhigaev https:/ /orcid.org/0000-0001-8398-9480M A Gomez-Gonzalez https:/ /orcid.org/0000-0003-2725-4820J E Parker https://orcid.org/0000-0002-2514-5762A Björling https://orcid.org/0000-0001-5681-2292A Mikkelsen https:/ /orcid.org/0000-0002-9761-0440J Wallentin https:/ /orcid.org/0000-0001-5909-0483

Figure 1 .
Figure 1.(a, b) SEM images of as-grown free-standing CsPbBr 3 nanowires.Some of the crystals stick out of the pores, with uncovered lengths of up to 20 μm.(c) Schematic representation of the chlorination process.The nanowires are initially transferred from the AAO matrix to a Si 3 N 4 substrate.Selected nanowires can be fully (upper route) or locally (bottom route) exposed to HCl vapor, resulting in axially homogeneous or heterostructured nanowires, respectively.In the second case, part of the crystal is protected using PMMA, which avoids anion exchange during exposure.

Figure 2 .
Figure 2.(a) Schematic representation of the nano-XRD/nano-XRF experimental setup.The selected nanowire was positioned with the c axis (Pbnm) along the horizontal plane and rotated to the Bragg angle θ, while the diffraction detector was positioned at 2θ.By scanning the beam with the sample, maps were acquired for each angular position.The scattering vector components q x , q y and q z , as well as lattice tilts α (rotation around q y ), β (rotation around q x ) and γ (rotation around q z ) are defined in the inset.The x-ray fluorescence detector position is in accordance with NanoMAX setup, while it is placed in backscatter geometry at I14. X-ray beam spot and maps step size were set at 50 nm for the experiment performed at I14 and 60 nm at NanoMAX.(b) Plot of the lattice parameter, α and β maps for the reference, non-exchanged, nanowire, which were acquired by analysing the reconstructed three-dimensional reciprocal space associated to each pixel in the map.(c) Intensity as a function of the lattice spacing (rocking curve) for the reference (green), as well as the nanowires exposed to HCl for 15 s (yellow), 30 s (light blue) and 45 s (dark blue).Green and blue vertical dashed lines indicate theoretical lattice spacing for pure CsPbBr 3 and CsPbCl 3 , respectively.(d) Average lattice parameter (black dots) and Br concentration ratio as measured by XRF (green squares) for the time series nanowires.Both ordinate axes were set to scale according to theoretical CsPb(Br (1−x) Cl x ) 3 lattice parameter values calculated using Vegard's law.All plotted lattice parameter refers to the long orthorhombic axis, set as c in the Pbnm group space for CsPbBr 3 .Uncertainties in lattice parameter are propagated from deviations calculated by gaussian fits in (c), while error bars in the concentration come from the peak fitting analysis.
Figure 2.(a) Schematic representation of the nano-XRD/nano-XRF experimental setup.The selected nanowire was positioned with the c axis (Pbnm) along the horizontal plane and rotated to the Bragg angle θ, while the diffraction detector was positioned at 2θ.By scanning the beam with the sample, maps were acquired for each angular position.The scattering vector components q x , q y and q z , as well as lattice tilts α (rotation around q y ), β (rotation around q x ) and γ (rotation around q z ) are defined in the inset.The x-ray fluorescence detector position is in accordance with NanoMAX setup, while it is placed in backscatter geometry at I14. X-ray beam spot and maps step size were set at 50 nm for the experiment performed at I14 and 60 nm at NanoMAX.(b) Plot of the lattice parameter, α and β maps for the reference, non-exchanged, nanowire, which were acquired by analysing the reconstructed three-dimensional reciprocal space associated to each pixel in the map.(c) Intensity as a function of the lattice spacing (rocking curve) for the reference (green), as well as the nanowires exposed to HCl for 15 s (yellow), 30 s (light blue) and 45 s (dark blue).Green and blue vertical dashed lines indicate theoretical lattice spacing for pure CsPbBr 3 and CsPbCl 3 , respectively.(d) Average lattice parameter (black dots) and Br concentration ratio as measured by XRF (green squares) for the time series nanowires.Both ordinate axes were set to scale according to theoretical CsPb(Br (1−x) Cl x ) 3 lattice parameter values calculated using Vegard's law.All plotted lattice parameter refers to the long orthorhombic axis, set as c in the Pbnm group space for CsPbBr 3 .Uncertainties in lattice parameter are propagated from deviations calculated by gaussian fits in (c), while error bars in the concentration come from the peak fitting analysis.
For pure CsPbBr 3 (a = 8.207 Å, b = 8.255 Å, c = 11.759Å) one can calculate the expected tilt to be1 presents a Cl gradient along its axis, values can slightly diverge.A second locally exposed nanowire was studied using the same techniques, and results shown in figureS2of the SM point for similar behaviors regarding both the lattice spacing gradient and the ferroelastic domains.It not only corroborates our hypothesis, but also shows the reproducibility of the chlorination process using HCl.Ferroelastic domains are also seen in the time series, as one can see in figure S1.The lattice tilts, however, decrease

Figure 3 .
Figure 3. (a) Lattice parameter (upper panel), α (middle panel) and β (bottom panel) maps of the nanowire.(b) XRF map showing the spatially resolved Br concentration, evidencing chemical gradient along the nanowire axis above the noise level.(c) Plot of the lattice parameter (black) and Br concentration (green) as a function of the x position along the nanowire axis.Both variables were obtained by averaging data from (a) and (b) along the y axis.(d) Plot of the lattice spacing (black dots) and Br concentration (green squares) as a function of the y position along the nanowire radius.Variables were averaged along x inside the black dashed box marked in (a) and (b).Pixel size in (a) and (b) is 60 nm.All plotted lattice parameter refers to the long orthorhombic axis, set as c in the Pbnm group space for CsPbBr 3 .