Surface Versus Bulk State Transitions in Inkjet-Printed All-Inorganic Perovskite Quantum Dot Films

The anion exchange of the halides, Br and I, is demonstrated through the direct mixing of two pure perovskite quantum dot solutions, CsPbBr3 and CsPbI3, and is shown to be both facile and result in a completely alloyed single phase mixed halide perovskite. Anion exchange is also observed in an interlayer printing method utilizing the pure, unalloyed perovskite solutions and a commercial inkjet printer. The halide exchange was confirmed by optical absorption spectroscopy, photoluminescent spectroscopy, X-ray diffraction, and X-ray photoemission spectroscopy characterization and indicates that alloying is thermodynamically favorable, while the formation of a clustered alloy is not favored. Additionally, a surface-to-bulk photoemission core level transition is observed for the Cs 4d photoemission feature, which indicates that the electronic structure of the surface is different from the bulk. Time resolved photoluminescence spectroscopy indicates the presence of multiple excitonic decay features, which is argued to originate from states residing at surface and bulk environments.

There have been some major issues with the halide-based perovskites, specifically, optimizing film quality [39], reducing the material instability in laboratory and extreme

Materials and Methods
For a detailed procedure on the quantum dot ink synthesis, the reader should refer to earlier works [9,37]. After synthesizing the quantum dot powders, they were converted to inks for printing by dissolving, separately, 280 mg of CsPbBr 3 and CsPbI 3 powders in 25 mL of cyclohexane (Sigma-Aldrich, St. Louis, MO, USA). The CsPbBr 3−x I x ink solution was prepared by combining a 1:1 mass ratio of CsPbBr 3 and CsPbI 3 ink solutions, known as direct mixing. Mixed inks (CsPbBr 3−x I x ) and pure inks (CsPbBr 3 and CsPbI 3 ) were printed on indium tin oxide (ITO)-coated polyethylene terephthalate (PET) flexible substrates (Sigma-Aldrich, St. Louis, MO, USA) or amorphous glass substrates. In a study by Hu et al. [49], annealing was found to reduce vacancy defects, and the inkjet printing process intrinsically is an annealing deposition method, as there is an abrupt temperature increase in an inkjet printer head.
X-ray diffraction (XRD) was employed to investigate the crystal structure for all printed films using a Rigaku Smartlab diffractometer (Rigaku Corporation, Akishimashi, Tokyo, Japan) equipped with a Cu-Kα X-ray source with a wavelength of 1.54 Å. XRD is complemented by X-ray photoemission spectroscopy (XPS) to further establish the composition of the photovoltaic printed films. XPS was performed with a SPECS Phoibos 150 hemispherical analyzer (SPECS Surface Nano Analysis GmbH, Berlin, Germany) using a non-monochromatized Al-K α X-ray radiation and a pass energy of 15 eV. All XPS was performed at room temperature in an ultra-high vacuum chamber with a chamber pressure better than 5.0 × 10 −10 mbar. Transmission electron microscopy (TEM) and select area electron diffraction (SAED) were performed on a JEOL JEM-2100 scanning The optical absorption and photoluminescent properties of the quantum dot inks, single layer, and inter-layer printed films were measured using an Ocean Optics HR4000CG-UV-NIR high resolution spectrometer, combined with an Ocean Optics DH-2000-BAL Deuterium-Halogen light source (Ocean Insight, Orlando, FL, USA). Time resolved photoluminescence (TRPL) measurements were carried out at room temperature, on increasing thicknesses of CsPbBr3 single-layer printed films, using an ultrafast laser equipped with a Ti:Sa amplifier system that generates 4 mJ pulses at 800 nm with 100 fs pulse width and a repetition rate of 1 kHz. A portion of the 800 nm pulse is sent to an optical parametric amplifier to generate a 310 nm, 100 fs laser pulse. The sample PL was collected in a backscattering orientation with a high numerical aperture 90 • off-axis parabolic mirror. The PL spectra were measured with an Avantes CCD spectrometer (Avantes USA, Lafayette, CO, USA) using a 325 nm long-pass filter to block the pump signal from entering the spectrometer. The measured PL spectra were corrected using a tungsten halogen lamp as a calibration source.

Structural and Optical Properties of Inks and Printed Films
The direct mixed perovskite (CsPbBr 3−x I x ) inks obtained from mixing the two pure CsPbBr 3 and CsPbI 3 quantum dot solutions are shown in Figure 1a. This alloying of the perovskite nanoparticles occurred in a solution at room temperature without any temperature treatment or other further efforts to aid the process (like stirring). Figure 1a shows how the color of the quantum dot ink mixture changes from brown to orange with time, a signature of the material resolving into an alloy of intermediate band gap. The color change is believed to be due to an anion exchange, whereby the Br and I ions migrate within the pure perovskite solutions to form a new mixed perovskite [50]. The alloying process takes approximately 30 min to 1 h; however, for larger nanoparticles the alloying process may be on a different time scale and requires further consideration.
complemented by X-ray photoemission spectroscopy (XPS) to further establish the composition of the photovoltaic printed films. XPS was performed with a SPECS Phoibos 150 hemispherical analyzer (SPECS Surface Nano Analysis GmbH, Berlin, Germany) using a non-monochromatized Al-Kα X-ray radiation and a pass energy of 15 eV. All XPS was performed at room temperature in an ultra-high vacuum chamber with a chamber pressure better than 5.0 × 10 −10 mbar. Transmission electron microscopy (TEM) and select area electron diffraction (SAED) were performed on a JEOL JEM-2100 scanning transmission electron microscope (JEOL USA, Inc., Peabody, MA, USA) at an accelerating voltage of 200 kV.
The optical absorption and photoluminescent properties of the quantum dot inks, single layer, and inter-layer printed films were measured using an Ocean Optics HR4000CG-UV-NIR high resolution spectrometer, combined with an Ocean Optics DH-2000-BAL Deuterium-Halogen light source (Ocean Insight, Orlando, FL, USA). Time resolved photoluminescence (TRPL) measurements were carried out at room temperature, on increasing thicknesses of CsPbBr3 single-layer printed films, using an ultrafast laser equipped with a Ti:Sa amplifier system that generates 4 mJ pulses at 800 nm with 100 fs pulse width and a repetition rate of 1 kHz. A portion of the 800 nm pulse is sent to an optical parametric amplifier to generate a 310 nm, 100 fs laser pulse. The sample PL was collected in a backscattering orientation with a high numerical aperture 90° off-axis parabolic mirror. The PL spectra were measured with an Avantes CCD spectrometer (Avantes USA, Lafayette, CO, USA) using a 325 nm long-pass filter to block the pump signal from entering the spectrometer. The measured PL spectra were corrected using a tungsten halogen lamp as a calibration source.

Structural and Optical Properties of Inks and Printed Films
The direct mixed perovskite (CsPbBr3−xIx) inks obtained from mixing the two pure CsPbBr3 and CsPbI3 quantum dot solutions are shown in Figure 1a. This alloying of the perovskite nanoparticles occurred in a solution at room temperature without any temperature treatment or other further efforts to aid the process (like stirring). Figure 1a shows how the color of the quantum dot ink mixture changes from brown to orange with time, a signature of the material resolving into an alloy of intermediate band gap. The color change is believed to be due to an anion exchange, whereby the Br and I ions migrate within the pure perovskite solutions to form a new mixed perovskite [50]. The alloying process takes approximately 30 min to 1 h; however, for larger nanoparticles the alloying process may be on a different time scale and requires further consideration. CsPbX 3 quantum dots have been shown to be more ionic in nature and in their interactions with ligands, which suggests an ionic migration mechanism [51]. Indeed, it has been shown that inorganic halide perovskites have high ionic conductivities due to anion migration and the exchange/replacement of the halide ions in these perovskites is facile [50,52,53]. Thus, when mixing two pure, unalloyed perovskite quantum dot solutions, the nanoparticle alloy formation is likely facilitated by anionic migration and to Nanomaterials 2022, 12, 3956 4 of 17 some extent by the free ligands present in the solution [50,51]. The ease and rapidity of the halide exchange/alloying indicates this is thermodynamically favorable.
In this experiment, three different printed layers are investigated, as shown in Figure 1b-d: (1) printing single layer films of pure (CsPbBr 3 or CsPbI 3 ) and/or mixed perovskite CsPbBr 3−x I x inks, where the mixed perovskites have directly alloyed in solution from different anion nanoparticles, (single-layer printing), (2) printing successive layers of unalloyed CsPbBr 3 and CsPbI 3 one on top of the other (inter-layer printing), and (3) printing a thick layer (several layers) of CsPbI 3 onto a substrate followed by printing a thick layer of CsPbBr 3 on top of the CsPbI 3 layer (bi-layer printing).
The XRD patterns for the single-layer printed films of pure CsPbX 3 quantum dots and the direct mixed CsPbBr 3−x I x quantum dots are shown in Figure 2a nanoparticle alloy formation is likely facilitated by anionic migration and to som by the free ligands present in the solution [50,51]. The ease and rapidity of the ha change/alloying indicates this is thermodynamically favorable.
In this experiment, three different printed layers are investigated, as shown in 1b-d: (1) printing single layer films of pure (CsPbBr3 or CsPbI3) and/or mixed pe CsPbBr3−xIx inks, where the mixed perovskites have directly alloyed in solution fr ferent anion nanoparticles, (single-layer printing), (2) printing successive layers loyed CsPbBr3 and CsPbI3 one on top of the other (inter-layer printing), and (3) pr thick layer (several layers) of CsPbI3 onto a substrate followed by printing a thick CsPbBr3 on top of the CsPbI3 layer (bi-layer printing).
The XRD patterns for the single-layer printed films of pure CsPbX3 quantu and the direct mixed CsPbBr3−xIx quantum dots are shown in Figure 2a, indicating a structure in the orthorhombic phase at room temperature in agreement with the li [54][55][56] and reference card ICSD 28312. The (110)   The structure changes, accompanying the nanoparticle mixed anion alloy formation, as characterized by X-ray diffraction (XRD), are consistent with a single phase material. Thus, by mixing two pure perovskites inks we can obtain a completely alloyed solution with a lattice constant, which falls between the pure (unalloyed) perovskites. The absence of Nanomaterials 2022, 12, 3956 5 of 17 peaks representative of CsPbBr 3 and CsPbI 3 in the XRD data of CsPbBr 3−x I x also suggests the absence of CsPbBr 3 and CsPbI 3 clusters within the mixed perovskite. Not only does the lattice constant in perovskites change with the changing anion, as is the case here with the alloying of a bromide with an iodide to form a CsPbBr 3−x I x , but we note that in other perovskites, the lattice parameters can change with the nature of the cationic A-site rare earth ion [57].
The crystallite sizes of the printed films for the pure and mixed perovskites were extracted from the XRD profiles using the Scherrer equation [58] and a standard Lorentzian distribution fitting function of the (220) Bragg peak. The crystallite sizes for all CsPbX 3 films were determined to be roughly 7.1 ± 1.0 nm. This indicates the halide exchange in direct mixing of pure perovskite solutions does not affect the crystallite size significantly. This surprising retention of crystallite size agrees with a study performed by Akkerman et al. [50]. As a comparison to the printed films, the TEM images of CsPbBr 3 nanoparticle inks were obtained and indicate the particles possess square-like profiles, Figure 2c. The TEM images were used to obtain the size distribution, as shown in Figure 2d, indicating that the average diameter of the nanoparticles is 6.74 ± 0.1 nm, in good agreement with the crystallite size in the films determined with XRD. Interestingly, the XRD peaks of the directly mixed perovskite from a mixed nanoparticle solution, appear closer to the CsPbBr 3 diffraction peak placement than CsPbI 3 . This indicates that the direct mixed perovskite is heavily weighted towards a large Br anion concentration The structural changes of the mixed anion alloy nanoparticle is consistent with the changes in the optical properties. The CsPbBr 3−x I x inks and the printed films are highly luminescent, displaying vivid colors, as shown in Figure 3a, suggesting the optical properties have been tuned to span a broad color spectrum. The optical absorption and background subtracted photoluminescence spectra of the pure CsPbBr3 and CsPbI3 and the direct mixed CsPbBr 3−x I x quantum dot single-layer printed films are shown in Figures 3b and 3c, respectively. The absorption spectra, Figure 3b, for the pure CsPbBr 3 film indicates absorption onsets around 499 ± 2 nm and CsPbI 3 quantum dot printed film indicates strong absorption all over the visible region in agreement with the literature [16,50]. The optical absorption onset, which appears around 542 ± 3 nm for the printed films from alloy CsPbBr 3−x I x quantum dots (directly alloyed in solution from different anion (Br and I) nanoparticles). This optical absorption for the printed films, from alloy CsPbBr 3−x I x quantum dots, is located between the absorption onsets of the pure (unalloyed) CsPbBr 3 and CsPbI 3 quantum dot films. The photoluminescent spectra, Figure 3c, of pure CsPbBr 3 and CsPbI 3 quantum dot printed films shows peaks around 504 ± 1 nm and 691 ± 2 nm, respectively, in agreement with the literature [50]. The photoluminescent peak for the alloyed CsPbBr 3−x I x quantum dot printed films, following formation from the mixed nanoparticle solutions, appears at 546 ± 4 nm, which again falls between the pure (unalloyed) quantum dot printed films. The absorption and photoluminescent data provide no indication of the clustering of either CsPbBr 3 or CsPbI 3 in the direct mixed CsPbBr 3−x I x single-layer printed films. As the direct mixed CsPbBr 3−x I x quantum dot printed films exhibit absorption onsets and photoluminescent peaks falling between those of the pure perovskite films and somewhat closer to the CsPbBr 3 films, in agreement with XRD, these data suggest that anion exchange produces an alloyed single phase perovskite while again supporting the idea that the direct mixed perovskite is heavily weighted towards a higher Br anion concentration. Nanomaterials 2022, 12, x FOR PEER REVIEW 6 of 18

Nanoparticle Alloying by Anion Exchange across an Inter-Layer Printed Film Interface
In the interest of ascertaining and confirming that alloying is favored across the interface between different anions in the printed films, an inter-layer printing technique was employed by printing alternating layers of pure CsPbBr3 and CsPbI3, beginning with a layer of CsPbI3, as seen in Figure 1. Figure 3d shows a comparison of the XRD data for printed films of the CsPbBr3, CsPbI3 and an inter-layer printed film prepared by the sequential printing of layers from the pure CsPbBr3 and CsPbI3 nanoparticle inks. There is a clear indication of alloying across the interlayer. This inter-layer printing method results in interdiffusion across the interface of the CsPbBr3 and CsPbI3 heterolayers. Interestingly, the (220) and (110) Bragg peaks of the interlayer printed film, as with the direct mixed CsPbBr3−xIx single-layer printed films, again fall between the (220) and (110) peaks of the pure (unalloyed) CsPbBr3 and CsPbI3 films indicating that there is a lattice constant shift due to anion exchange between the inter-diffused layers of CsPbBr3 and CsPbI3. The optical absorption and photoluminescent spectra of the inter-layer printed films (the denoted inter-layers in Figure 3e,f are compared to those of the direct mixed CsPbBr3−xIx single-layer printed quantum dot film, indicating close agreement).
There exists the possibility of limitations to alloying of the heterolayer Br and I anion nanoparticle perovskite inter-layer printed films, which could be consistent with either a higher barrier to interdiffusion or limits to diffusion lengths and rates once printed. These limitations to alloying once the perovskite films are printed were tested using the bi-layer printing method, where a thick layer of CsPbBr3 is printed and then a thick layer of CsPbI3 is printed on top of the CsPbBr3 films. The photoluminescent spectrum, see Supplementary Figure S1, suggests the thicker bi-layer printing method results in unmixed, phaseseparated components rather than a uniform alloyed CsPbBr3−xIx perovskite film as seen in the thinner inter-layer printed films, where inter-layer mixing is more uniform (

Nanoparticle Alloying by Anion Exchange across an Inter-Layer Printed Film Interface
In the interest of ascertaining and confirming that alloying is favored across the interface between different anions in the printed films, an inter-layer printing technique was employed by printing alternating layers of pure CsPbBr 3 and CsPbI 3 , beginning with a layer of CsPbI 3 , as seen in Figure 1. Figure 3d shows a comparison of the XRD data for printed films of the CsPbBr 3 , CsPbI 3 and an inter-layer printed film prepared by the sequential printing of layers from the pure CsPbBr 3 and CsPbI 3 nanoparticle inks. There is a clear indication of alloying across the interlayer. This inter-layer printing method results in interdiffusion across the interface of the CsPbBr 3 and CsPbI 3 heterolayers. Interestingly, the (220) and (110) Bragg peaks of the interlayer printed film, as with the direct mixed CsPbBr 3−x I x single-layer printed films, again fall between the (220) and (110) peaks of the pure (unalloyed) CsPbBr 3 and CsPbI 3 films indicating that there is a lattice constant shift due to anion exchange between the inter-diffused layers of CsPbBr 3 and CsPbI 3 . The optical absorption and photoluminescent spectra of the inter-layer printed films (the denoted inter-layers in Figure 3e,f are compared to those of the direct mixed CsPbBr 3−x I x single-layer printed quantum dot film, indicating close agreement).
There exists the possibility of limitations to alloying of the heterolayer Br and I anion nanoparticle perovskite inter-layer printed films, which could be consistent with either a higher barrier to interdiffusion or limits to diffusion lengths and rates once printed. These limitations to alloying once the perovskite films are printed were tested using the bi-layer printing method, where a thick layer of CsPbBr 3 is printed and then a thick layer of CsPbI 3 is printed on top of the CsPbBr 3 films. The photoluminescent spectrum, see Supplementary Figure S1, suggests the thicker bi-layer printing method results in unmixed, phase-separated components rather than a uniform alloyed CsPbBr 3−x I x perovskite film as seen in the thinner inter-layer printed films, where inter-layer mixing is more uniform (Figure 3d-f) and single-layer printed films made from mixing pure nanoparticle solutions (Figure 3a-c).
In both the single-layer printed alloyed quantum dots film, from mixing pure nanoparticle solutions, and the inter-layer printed film, with mixing occurring at the interfaces, the end result was a material where a final mixed halide state is seen to be strongly preferred. In other words, when mixed together, either in solution or as a heterolayer system, the CsPbBr 3−x I x perovskites prefer to exist as a single phase mixed alloy rather than remain phase separated, i.e., strongly clustering. The data from single-layer printed film, from mixing pure nanoparticle solutions prior to printing, and the inter-layer printed films, with mixing occurring at the interfaces, suggest that the mixed state acts as a thermodynamic sink and provides a minimum in the energy landscape during the reaction phase of the halide exchange. The thicker bi-layer system is an exception, as seen in supporting information, as this retains some of the separation of the stacked CsPbBr 3 and CsPbI 3 films, which could be due to the increased layer thickness and/or rapid solvent evaporation acting as limiting factors in the halide exchange reaction and hetero-layer interdiffusion process. Not all solution based deposition or printing methods lead to the formation of a lead perovskite QD alloy thin film. The cations Cs and formamidinium are not reported to diffuse across the interface in Cs 0.25 FA 0.75 PbI 3 to CsPbI 3 heterostructures [59].
From these data it can be inferred that the interlayer printing method has a few advantages compared to the single-layer and bi-layer method. The interlayer printing method results in a completely alloyed film from two distinct parent products, whereas the bi-layer method results in a segregated non-alloyed film with multi-component optical properties. In other words, the interlayer method is capable of producing homogeneous films from multiple source inks that maintain the attractive optical properties of the alloyed materials. Additionally, the ability to print from multiple sources to produce new alloys implies that the interlayer printing method is more versatile and possibly more costeffective than the single-layer printing method. The interlayer method uses a few parent (CsPbX 3 ) solutions (inks), while allowing for various mixed halide films to be produced, thus reducing the total number of perovskite solutions (inks) compared to the single-layer method, which requires a separate ink for each type of film, whether it is a mixed halide or pure halide.

Surface-to-Bulk Core Level Shift
One can imagine, upon the completion of anion (halide) exchange forming a new alloy (CsPbBr 3−x I x ), the possibility of electronic surface state and bulk states, which are very different from those in non-alloyed (CsPbBr 3 or CsPbI 3 ) materials, is likely. Additionally, the surface could exhibit an electronic structure different from the bulk, which may be observable in angular-dependent photoemission spectroscopy.
Interestingly, a surface-to-bulk core level shift [60][61][62] was observed for the Cs 4d core level XPS feature for both the CsPbBr 3 and the alloy CsPbBr 3−x I x single-layer printed films, the latter of which forms from the direct mixing of the different anion nanoparticles in solution prior to printing, as shown in Figures 4a and 4b, respectively. In order to increase the surface sensitivity of XPS, the photoemitted electron take-off angle was varied from 0 • to 40 • . By increasing the take-off angle of the emitted electrons from 0 • to 40 • , the intensity of a surface contribution will appear to increase in intensity relative to the bulk contribution [60,61]. In Figure 4a,b, curve A (red dots) corresponds to a photoemitted electron takeoff angle of 40 • and curve B (black triangles) corresponds to a takeoff angle of 0 • , with respect to the surface normal. Beneath each curve, the surface (dashed lines) and bulk (solid lines) contribution fittings are displayed. The intensity difference, curve C, between curve A and curve B, is plotted as a solid green line.
As shown in Figure 4, both films (CsPbBr 3 and CsPbBr 3−x I x ) exhibit increases in their respective photoemission intensity, i.e., surface states, which emerge with increasing takeoff angles. If one considers the intensity difference curve, curve C (solid green line), then for CsPbBr 3 curve C indicates the surface contribution is a broad peak at lower binding energy and the bulk contribution is a dip at higher binding energy. In comparison, curve C for CsPbBr 3−x I x films indicates the surface contribution is a broad peak at higher binding energy than the bulk contribution (dip) at lower binding energy. The appearance of a surface state at low binding energy in CsPbBr 3 and at high binding energy in CsPbBr 3−x I 3 suggests that the surface terminations are different for the quantum dot films. As shown in Figure 4, both films (CsPbBr3 and CsPbBr3−xIx) exhibit increases in their respective photoemission intensity, i.e., surface states, which emerge with increasing takeoff angles. If one considers the intensity difference curve, curve C (solid green line), then for CsPbBr3 curve C indicates the surface contribution is a broad peak at lower binding energy and the bulk contribution is a dip at higher binding energy. In comparison, curve C for CsPbBr3−xIx films indicates the surface contribution is a broad peak at higher binding energy than the bulk contribution (dip) at lower binding energy. The appearance of a surface state at low binding energy in CsPbBr3 and at high binding energy in CsPbBr3−xI3 suggests that the surface terminations are different for the quantum dot films.
The X-ray photoemission (XPS), shown in Supplementary Figure S2, was utilized in order to investigate the composition of the CsPbBr3−xIx quantum dot printed films, formed after mixing the different anion nanoparticle solutions. The Cs 3d5/2 and 3d3/2 core level photoemission peaks are located at 723.8 ± 0.2 eV and 737.88 ± 0.2 eV, respectively. The Pb 4f7/2 and 4f3/2 core level peaks are located at 138.1 ± 0.1 eV and 142.9 ± 0.2 eV, respectively. While the surface segregation of metallic Pb in lead halide perovskites is a well-established phenomenon [63][64][65][66][67][68], no evidence of Pb 0 clusters at the film surface was evident in core level photoemission spectra at the Pb core level photoemission features. The XPS spectra of Br shows up as a single broad photoemission peak centered around 68 eV, with an additional feature at higher binding energies due to the Cs 4d core level feature. A modified Voigt function was used to fit the broad Br core level peak and peak positions for Br 3d5/2, and 3d3/2 were extracted and determined to be 68.1 ± 0.2 eV and 69.6 ± 0.2 eV, respectively. The core level photoemission features of I 3d5/2 and I 3d3/2 are located at 618.4 ± 0.2 eV and 629.9 ± 0.2 eV, respectively. All of the core level photoemission features for Cs, Pb, I, and Br agree well with values reported in the literature for CsPbBr3, CsPbI3 and mixed perovskites [34,56,[69][70][71]. The photoemission intensities were used to calculate a Br:I atomic ratio of roughly 2.4:0.6, suggesting a Br weighted direct mixed perovskite, CsPbBr2.4I0.6, in agreement with the XRD, optical absorption spectroscopy, and photoluminescent spectroscopy data. The XPS of the Cs 4d and Pb 4f7/2 and 4f3/2 core level photoemission features provide further evidence that halide clustering, in the alloy, does not occur, as there are no satellite features that are common in Cs-halide clustered materials The X-ray photoemission (XPS), shown in Supplementary Figure S2, was utilized in order to investigate the composition of the CsPbBr 3−x I x quantum dot printed films, formed after mixing the different anion nanoparticle solutions. The Cs 3d 5/2 and 3d 2/3 core level photoemission peaks are located at 723.8 ± 0.2 eV and 737.88 ± 0.2 eV, respectively. The Pb 4f 7/2 and 4f 3/2 core level peaks are located at 138.1 ± 0.1 eV and 142.9 ± 0.2 eV, respectively. While the surface segregation of metallic Pb in lead halide perovskites is a well-established phenomenon [63][64][65][66][67][68], no evidence of Pb 0 clusters at the film surface was evident in core level photoemission spectra at the Pb core level photoemission features.
The XPS spectra of Br shows up as a single broad photoemission peak centered around 68 eV, with an additional feature at higher binding energies due to the Cs 4d core level feature. A modified Voigt function was used to fit the broad Br core level peak and peak positions for Br 3d 5/2 , and 3d 3/2 were extracted and determined to be 68.1 ± 0.2 eV and 69.6 ± 0.2 eV, respectively. The core level photoemission features of I 3d 5/2 and I 3d 3/2 are located at 618.4 ± 0.2 eV and 629.9 ± 0.2 eV, respectively. All of the core level photoemission features for Cs, Pb, I, and Br agree well with values reported in the literature for CsPbBr 3 , CsPbI 3 and mixed perovskites [34,56,[69][70][71]. The photoemission intensities were used to calculate a Br:I atomic ratio of roughly 2.4:0.6, suggesting a Br weighted direct mixed perovskite, CsPbBr 2.4 I 0.6 , in agreement with the XRD, optical absorption spectroscopy, and photoluminescent spectroscopy data. The XPS of the Cs 4d and Pb 4f 7/2 and 4f 3/2 core level photoemission features provide further evidence that halide clustering, in the alloy, does not occur, as there are no satellite features that are common in Cs-halide clustered materials [72,73]. Cs-halide free clusters are also very unlikely. The Cs lattice atom can exist in only one oxidation state, in a lattice with large electron affinity anions, suggesting that the surface-to-bulk core level shift is attributable to a large difference in the electronic structure of the surface versus the bulk of the thin films. Furthermore, there is a large separation in the binding energies of the surface versus bulk peaks, confirming the idea that the surface electronic structure is vastly different from the bulk [62].

Transient Recombination Measurements
The presence of defects and trap states in these materials is well known and as there is a surface-to-bulk state transition one could postulate, with confidence, that defects/trap states located near the surface may be different, with respect to the excitonic recombination time, than defect/trap states located within the bulk of the material/film. Indeed, in previous work on these materials the transport data and drift carrier lifetimes indicated regions of fast and slow recombination [9], which suggests the presence of at least two recombination pathways and possibly more. Multiple excitonic recombination pathways and energy loss mechanisms would cause the TRPL profile to broaden along the time axis, especially if these states have a strong influence on the exciton recombination [74]. The origins of multiple recombination lifetimes regarding these materials is somewhat ambiguous; moreover, there seems to be little consensus in the literature on established lifetime ranges.
A list of recent TRPL studies on CsPbBr 3 and CsPbBr 3−x I x along with the decay times and postulated origins for those lifetimes are contained in Table 1. Something to note before discussing the postulated origins is that not all of the studies have listed multiple decay event lifetimes, some have one lifetime and others only two or three. Limiting the fitting equations to one or two term exponential equations results in the data fits being too high in the data and/or too low in the data along the vertical axis. This poses a problem for the correct interpretation of the TRPL data and whether or not those fits (single and two term exponential equations) represent the system correctly. Looking carefully at Table 1, the reader can see that the majority of the postulated origins of transient events can be broken down into the following types: surface states, bulk states, traps states, radiative recombination, and non-radiative recombination. Multiple origins for recombination events across similar systems lead credence to the idea that the fitting equations need to account for more than two events in any given system, especially if said system is capable of multiple energy transitions from the excited state, as has been suggested for inorganic lead halides [75]. Another cause for concern is related to the time-ranges for the various postulated origins, which all seem to overlap, at least in the cases presented in Table 1. It may be more instructive to focus on the surface and bulk state times associated only with CsPbBr 3 . The time constant range for surface states is τ surface = 1.22-23 ns and for bulk states the range is τ bulk = 3.55-233 ns, clearly there is some overlap again. If one considers the average of the lifetimes, rather than the time-range, then recombination due to surface states is roughly 12 ns and bulk states is roughly 118 ns. We then can say with some confidence that, typically, τ surface < τ bulk . As trap states can exist at the surface and in the bulk, it is proposed that there should be four possible decay events that would correspond to surface-trap-, surface-, bulk-trap-and bulk-like states, such that τ surface < τ surfacetrap < τ bulk < τ bulktrap . In this scenario, it is assumed that trap states (shallow) are long-lived states that do not promote recombination, although it is acknowledged that deep trap states may exist and would promote fast recombination. Table 1. List of some recent TRPL measurements for CsPbBr 3 and CsPbBr 3−x I x materials. Column 1, labeled 'Material', indicates the type of material investigated along with reference number. Columns 2, 4 and 6, labeled 'τ 1 , 'τ 2 and 'τ 3 , respectively, are the TRPL lifetimes measured in the associated studies. Columns 3, 5 and 7, labeled 'origin of τ 1 , 'origin of τ 2 and 'origin of τ 3 , respectively, are the physical origin of the lifetimes in columns 2, 4 and 6, respectively, according to the associated reference. TRPL measurements for the CsPbBr 3 films of increasing thickness are shown in Figure 5a-c. The TRPL profiles exhibit a broad exponential decay profile, which suggests that the regular exponential decay functions may not describe the events properly. Indeed single, double and triple regular and stretched exponential functions cannot be used to confidently fit the data. A least squares fitting method was employed with the following quadruple stretched exponential Equation: where A 1 -A 4 are the weighted contribution terms, which describe the relative importance of the exponential terms and thus the prevalence of a recombination event. The lifetimes, τ 1 -τ 4 , correspond to the recombination lifetimes of the exciton recombination events. The power terms, β 1 -β 4 , influence the horizontal broadening of the exponential terms, representing the vertical shift of Equation (1), rather than the noise floor of the TRPL signal.
The associated recombination lifetimes for the CsPbBr 3 films where the single-layer printing method is used, are extracted from the fittings using Equation (1) and plotted versus increasing film thickness in Figure 5d. All lifetime curves are distinct, with respect to the error bars, for all sample thicknesses, meaning none of the curves intersect or overlap.
The weighted contributions to the TRPL spectra are described by the A i terms in Equation (1).The A i for each exponential subpeak term versus film thickness are plotted in Figure 5e. The A i indicates, for each τ i , the overall contribution to the fit function, contained within the decay profiles shown in Figure 5a-c. Furthermore, the A i term mathematically indicates the extent of vertical stretching or vertical compressing along the intensity axis for each exponential subpeak of Equation (1), such that if A i > 0.0, a vertical stretching occurs, and if A i < 0.0, a vertical compression results. All weighting term curves in Figure 5c are greater than 0.0 for all thicknesses, meaning they all contribute to a vertical stretching of the fitting function, Equation (1). For the most part, the weighting term curves do not vary appreciably with increasing film thickness. A 2 (black line-square) and A 3 (blue line, triangle) are well separated from A 1 (red line, circle) and A 4 (green line, hourglass). The weighted term averages are as follows: A avg1 = 0.24, A avg2 = 0.46, A avg3 = 0.05 and A avg4 = 0.25. This would make the average A i weighting term inequality appear as A avg3 < A avg1 < A avg4 < A avg2 . Due to the dominate nature of the surfaces state contributions to the photoluminescence decay profiles [90][91][92][93], it is expected that surface related recombination events should have the largest A i terms on average. Additionally, it is expected that trap related states will have less density compared to their counterpart states (surface and bulk) and should have smaller weighting term contributions accordingly. Applying these guidelines to the weighting terms leads to the following model-based inequality, A bulktrap <, A bulk < A surfacetrap < A surface , which is similar to the model-based lifetime inequality, τ surface < τ surfacetrap < τ bulk < τ bulktrap , except the subscripts are in reverse order. If we use the model-based weighting term inequality as a guide, then assignments of the average weighting terms are as follows A surface = A avg2 ; A surfacetrap = A avg1 ; A bulk = A avg4 ; A bulktrap = A avg3 .
The extended broadening along the time axis, as is apparent in the TRPL data of Figure 5a-c, is a major source of errors in most TRPL fittings if left unaccounted for in the fitting function. It is also challenging to establish a single source or physical process responsible for the broadening due to the complexities inherent in the exciton formation and subsequent recombination process. Once an exciton is formed radiative and nonradiative interactions exist and thus associated recombination and charge transfer rates [74]. Non-radiative interactions that may or may not induce recombination can delay or expedite the eventual recombination of the exciton. The temporal delay leads to a broadening or horizontal stretching along the time axis in TRPL spectra. The broadening terms for each exponential subpeak from Equation (1) versus increasing film thickness are plotted in Figure 5f. In this study, the broadening term is defined as follows: where τ i is the lifetime term from the TRPL fitting equation, Equation (1), see Figure 5d. β i is the power of the exponential power in Equation (1). If BT i is > 1.0 then a horizontal stretching (broadening) occurs, whereas if BT i < 1.0, then horizontal compression ensues. For all film thicknesses, the BT 1 , BT 3 , and BT 4 terms are greater than 1.0, suggesting they add to the horizontal broadening, whereas the BT 2 is less than 1.0 for all thicknesses, which means this acts to compress the fitting curve, similar to a fast recombination process with negligible delay before recombination. If we assume that short delay or fast recombination processes contribute less broadening than the slow recombination processes, then a model-based broadening term inequality can be established, so that BT surface < BT surfacetrap < BT bulk < BT bulktrap . This is similar to the model-based lifetime inequality, τ surface < τ surfacetrap < τ bulk < τ bulktrap . On average the broadening terms are BT avg1 = 1.97, BT avg2 = 0.51, BT avg3 = 4.38, BT avg4 = 2.48, such that BT avg2 < BT avg1 < BT avg4 < BT avg3 . If we use the model-based broadening term inequality as a guide, then the assignments of the average broadening terms are as follows: BT surface = BT avg2 ; BT surfacetrap = BT avg1 ; BT bulk = BT avg4 ; BT bulktrap = BT avg3 . The associated recombination lifetimes for the CsPbBr3 films where the single-layer printing method is used, are extracted from the fittings using Equation (1) and plotted versus increasing film thickness in Figure 5d. All lifetime curves are distinct, with respect to the error bars, for all sample thicknesses, meaning none of the curves intersect or overlap.
Interestingly, the lifetime curves seem to group into two regions, namely fast and slow recombination time regions, where for fast recombination τ < 4.0 ns (includes τ1, τ2, and τ4) and for slow recombination τ > 4.0 ns (includes τ3). The lifetime curves in the fast All of the fitting constants are consistent in their identification of the proposed process. These findings are extremely important in the context of the debate regarding the existence and role of surface and bulk states in the charge transport properties of CsPbX 3 photovoltaic cells. More measurements are necessary to explore the role of phonon coupling and the chemical nature of the trap states in these materials.

Conclusions
This study demonstrates the anion exchange of CsPbBr 3 and CsPbI 3 solutions/films results in high quality alloyed CsPbBr 3−x I x nanoparticle textured thin films, through both the direct mixing of nanoparticles in solution and heterolayer printed films, with mixing occurring at the interlayer. The halide exchange in both methods results in a single phase completely alloyed mixed perovskite as confirmed by XRD, optical absorption spectroscopy, photoluminescent spectroscopy, and X-ray photoemission spectroscopy. The mixed halide state acts as a thermodynamic sink, providing a minimum in the energy landscape of the halide anion exchange reaction. Additionally, the electronic structure of the film surface is shown to be very different compared to the bulk through the observation of a surface-tobulk core level shift, which provides insight on the origins of the excitonic decay process. The success of this work indicates that the rapid prototyping of various perovskite inks and multilayers is realizable. This fundamental fabrication method will lead to a deviceby-design routine for promoting next generation solar cell and optical materials/devices, specifically custom lasing devices or LED display screens.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/nano12223956/s1, Figure S1: Photoluminescence of the bi-layer CsPbBr 3 /CsPbI 3 quantum dot printed thin film, the presence of two separate peaks located at roughly 490 nm and 650 nm suggests the presence of segregated CsPbBr 3 (green line) and CsPbI 3 (red line), thus confirming the bi-layer printing method results in unmixed layers of CsPbBr 3 /CsPbI 3 ; Figure S2: X-ray photoelectron spectroscopy of (a) the Cs 3d core level peaks, (b) Pb 4f core level peaks and (c) the Br 3d core level peaks and Cs 4d core level peaks and (d) the I 3d core level peaks for the direct mixed perovskite CsPbBr 2.4 I 0.6 quantum dot printed thin films; Figure S3: Optical absorption and photoluminescence profiles for CsPbBr 3 films, printed with single-layer printing method, of (a) thickness 1, (b) thickness 2 and (c) thickness 3.