Impact of light illumination on the surface structure of two-dimensional Ruddlesden–Popper perovskite in the fabrication process

Two-dimensional Ruddlesden–Popper (2DRP) perovskites are promising owing to their excellent environmental stability and competitive efficiency. During the fabrication process, 2DRP perovskites were often unintentionally exposed to light in the laboratory. However, the influence of light illumination on the surface structure of 2DRP during fabrication is unclear. Herein, the photodegradation of 2DRP perovskite (phenethylammonium lead iodide, PEA2PbI4) is comprehensively investigated using x-ray photoelectron spectroscopy, scanning electron microscopy, atomic force microscopy, and photoluminescence spectroscopy. We observed that only high-energy light, including that from a blue laser, air mass 1.5G, and notably, the daily used fluorescent lamp (FL) in the laboratory, significantly degraded PEA2PbI4. However, the red laser and ultraviolet-filtered FL, which had low energies, did not cause photodegradation. From this systematic study, we can explain the discrepancies in the surface morphologies previously studied. For instance, randomly oriented nanorod or rough surface of PEA2PbI4 mostly stems from photodegradation. We observed that photodegradation occurred more strongly when the films were illuminated during annealing than when they were illuminated after all fabrication processes were completed. We suggest that this difference stems from the completeness of the structure and the photodegraded PbI2 passivation effect. Our study provides two key guidelines for the fabrication of PEA2PbI4 films. The daily-use FL in the laboratory must be avoided for high-quality samples, and dark conditions are highly recommended, at least during the annealing process.


Introduction
Recently, two-dimensional Ruddlesden-Popper (2DRP) perovskites have been intensively studied to overcome the environmental instability of their three-dimensional (3D) counterparts [1][2][3][4][5]. They have exhibited outstanding potential for use in optoelectronic devices, such as solar cells and light-emitting diodes, because they offer beneficial properties that extend beyond excellent stability, such as high photoluminescence (PL) quantum yields [6][7][8], structural versatilities [9,10], and tunable band gaps [11][12][13]. The 2DRP perovskite has a unique crystal structure, in which bulky organic cation spacers are intercalated between each octahedron sheet consisting of regular 3D perovskite. Bulky organic cations are known to be the primary factors that enhance material stability. It protects the octahedral layer from oxygen and moisture owing to its high hydrophobicity [14][15][16] and inhibits ion migration [17].
To take advantage of this characteristic, a multidimensional perovskite adopting the 2DRP perovskite was introduced, and it delivers a high-power conversion efficiency of 21% with relatively long-term stability compared to conventional 3D perovskite [18,19]. To optimize that, the optoelectrical properties of 2DRP perovskites, such as energy funneling, charge-transfer excitons, and carrier transport, have been intensively studied [13,[20][21][22][23][24][25][26][27][28][29][30][31]. Notably, these studies reported different and even conflicting surface conditions for 2DRP perovskites, even though their fabrication methods were almost identical. For example, Chirvony et al showed a random orientation of an approximately 200 nm rod on the surface, and they claimed it to be polycrystalline [32]. Zhang et al showed many small bumps at the grain boundaries in scanning electron microscopy (SEM) images [28]. In addition, Yang et al observed a rough surface morphology in the atomic force microscopy (AFM) image, which is in contrast to previous results [33].
During the fabrication process, 2DRP perovskites were often unintentionally exposed to light, such as a daily-use fluorescent lamp (FL) or sunlight from the window, in the laboratory. In previous works [34,35], they observed that light illumination leads to the loss of iodine in 2DRP perovskites, which results in the generation of lead iodide (PbI 2 ) and eventually, metallic lead (Pb 0 ), which has been reported as the most detrimental defect [36][37][38][39]. In this reason, we assume that above discrepancies may be, at least partially, originated from an unintended light illumination in the fabrication processes. However, it remains unclear how photodegradation occurs, and its impact has not been determined because there are no systematic studies on this topic.
This study makes a unique contribution to the literature by investigating the effect of light illumination on all sample-fabrication processes. Various light sources, such as blue lasers, red lasers, a laboratory FL without/without an ultraviolet (UV) filter, and air mass (AM) 1.5, will be used to understand the photodegradation process of 2DRP perovskite. A phenethylammonium lead iodide (PEA 2 PbI 4 ) film was selected as the model structure to represent the 2DRP perovskite. We monitored the morphological evolution resulting from photodegradation via AFM and SEM. The degree of degradation and surface chemical structure were examined using PL and x-ray photoelectron spectroscopy (XPS). From this systematic study, we will address (1) which energy (wavelength) of light induces photodegradation, (2) how strongly it impacts the optoelectronic properties of 2DRP perovskite, and (3) which process of the film fabrication is most severely affected.

Impact of light illumination on optical image
Fabrication of the PEA 2 PbI 4 films consisted of three steps. They were first spin coated on top of the indium tin oxide substrates, then annealed on a hot plate, and left to stand by for the next process, as schematically shown in figure 1(a) (see SI section 1). The thickness of the pristine PEA 2 PbI 4 films was measured to be approximately 250 nm using surface profilometer, (DektakXT Bruker). This thickness is thick enough the surface morphology can be limitedly affected by substrate interaction. The PEA 2 PbI 4 films were illuminated using light sources with different spectra, which included a blue laser (405 nm), red laser (638 nm), AM 1.5G, FL, and UV-filtered FL for each process. (See SI section 2) For simplicity, the films investigated in this study will be notated using the representations given in table 1.
We measured the optical images of the samples illuminated during the above three fabrication processes. For DS samples, there are no changes in the optical images, regardless of which kind of light illumination (not shown here). Figure 1(b) shows the optical properties of DA blue , DA AM1.5G , and DA FL . The films were illuminated up to the target exposure time and then kept in the dark until the annealing process was complete. The spot on the film illuminated by the blue laser (spot size is 5 mm × 5 mm) and AM 1.5G (spot size covers the whole film) gradually became transparent as time increased (highlighted by the blue and orange dashed boxes). In contrast, we can not observe any changes in DA FL . Changes in films illuminated by the UV-filtered FL, and red laser were not observed in the optical image. Similarly, the AF samples exhibited undetectable changes, except for AF blue (see SI section 3). These results demonstrate that the light illumination effects during the annealing process are stronger than those after all fabrication processes are completed. In addition, at first glance, all lights except the blue laser and AM 1.5G do not seem to affect the surface condition of the PEA 2 PbI 4 films.

Ensuring the quality of PEA 2 PbI 4 film
The quality of fabricated pristine PEA 2 PbI 4 film was confirmed using AFM, SEM, and XRD. The pristine film fabricated under dark conditions has a very-flat morphology that is characterized by a sharp layered planar structure and does not exhibit bumps and pinholes as shown in figures 2(a) and (b). XRD spectra  demonstrated well-defined vertical layered structure. (figure 2(c)). In figure 2(d), this was further confirmed by the line profile, where the distinct monolayer height of pristine PEA 2 PbI 4 was in good agreement with those of previous studies [34,40]. Figure 3 shows the SEM and AFM images that were obtained to investigate the morphology of the PEA 2 PbI 4 films illuminated by different light sources at different points in the fabrication process. Line profiles (red lines below) were obtained from the solid green lines in the corresponding images. As shown in figure 3(a), AF blue was heavily photodegraded. In the SEM image, the surface exhibits many pinholes with cracks. The AFM image provided more detailed information that showed this surface was completely covered by small bumps. It is no longer possible to identify the layered structure owing to such a broken surface morphology. As shown in figure 3(b), the morphology shown in the AF AM1.5G SEM image is difficult to distinguish from that of the pristine film, although it is magnified 50 000 times. However, the AFM image shows that the surface morphology is similar to that of AF blue . It was fully covered by tiny bumps and pinholes, which were too small to be detected by SEM. Unexpectedly, AF FL showed signs of photodegradation, unlike the optical image ( figure 1(c)). As shown in figure 3(c), sparsely located small bumps and shallow pinholes are observed in both the SEM and AFM images. It is very similar to the previously reported morphologies in many papers, which can be interpreted as the resulting photodegradation [32]. (See SI section 4 for the large image) This evidence that even the daily use of an FL in the laboratory can induce significant photodegradation on the surface of PEA 2 PbI 4 within several minutes. In contrast, photodegradation did not occur for AF UV-F FL and AF red (figures 3(d) and (e)). The morphologies of both films are almost identical to those of the pristine film, and they maintain distinguishable layered structures. Similar changes in morphology are also observed for DA samples in figures 3(f)-(j). In this case, the surface morphologies of the films were more severely damaged. In comparison with figures 3(a)-(c), the bumps and pinholes are much larger and denser (figures 3(f)-(h)), and the line profiles clearly show this trend. Unlike AF AM1.5G , the bumps were only detected via AFM, and the bumps in DA AM1.5G were clearly observed in the SEM images. Similar to AF UV-F FL and AF red , we observed no morphological changes in DA UV-F FL and DA red . From this result, we can conclude that PEA 2 PbI 4 was more severely photodegraded when illuminated during the annealing process, as expected from figure 1. This means that low-energy light does not induce photodegradation regardless of when the film is illuminated.

Investigation on the stoichiometry change by XPS analysis
XPS measurements were performed to study the effects of changes in the chemical structure of the samples.
To avoid x-ray damage, the exposure time of the x-rays was minimized and properly controlled during the measurement [35]. The samples were carefully treated (under near-dark conditions) to avoid exposure to unintended external light throughout the entire process. Figure 4(a) shows the atomic ratio between I and Pb (I/Pb ratio) of the AF samples up to 1000 s. The atomic ratio was calculated from the XPS peak areas under I 3d 5/2 and Pb 4f, considering each relative sensitivity factor. All PEA 2 PbI 4 films before illumination had an I/Pb ratio that was very close to 4.0, which is the stoichiometric ideal. As the illumination time increased, AF blue (blue line) and AF AM1.5G (yellowish-brown line) gradually decomposed, thereby losing their constituent iodine ions by the following equation [34]. Therefore, the I/Pb ratio decreased to 2.05 and 2.59, respectively: As observed in the SEM and AFM images (figure 3(c)), the I/Pb ratio of the AF FL (cyan line) decreased to 3.75. Moreover, AF red (red line) and AF UV-F FL (green line) maintained their I/Pb ratio throughout the entire illumination period (see SI section 5 for all the data). Figure 4(b) shows the N/Pb ratio for the AF samples up to 1000 s. All films before illumination had N/Pb ratios between 1.92 and 2.05, where the stoichiometric ideal value was 2.0. The N/Pb ratio was sharply reduced in AF blue , AF AM1.5G , and AF FL . It can be inferred that the organic spacer cation disappeared because N 1s is a direct indicator of the organic spacer cation PEA + ([C 6 H 5 C 2 H 4 NH 3 ] + ). In contrast, AF red and AF UV-F FL exhibited no signs of decomposition.
From previous literature, the bandgap of PEA 2 PbI 4 and PEA are 2.65 and 9.79 eV, respectively [23,41]. Thus, in the case of high-energy lights, such as blue laser and AM 1.5G, with energies higher than 2.65 eV and smaller than 9.79 eV, it is expected that the dominant interaction will be with PEA 2 PbI 4 , rather than  PEA. Based on the given fact, it can be inferred that the observed disappearance of the organic spacer is not a direct result of high-energy light interacting with PEA, but a result of an interaction between PEA and the photogenerated free electron present in the conduction band of PEA 2 PbI 4 . In figures 4(a) and (b), we observe that the reduction rate of N/Pb is faster than that of I/Pb, and the increase in the absolute intensity of Pb 4f is owing to the decrease in the C-N bond peak in the C 1s spectra (see SI section 5). From these two observations, we suggest that (1) the organic spacer cation is easily detached from the surface by high-energy light (blue laser, AM 1.5G, FL) owing to its high volatility, and (2) the segregation of the [PbI 6 ] − octahedra layers occurs after the loss of organic spacer cations because the I − ion migration is no longer suppressed.
The films illuminated during the annealing process were investigated in the same manner. As expected from the morphological analysis, the I/Pb and N/Pb ratios of DA were considerably faster and more strongly reduced than those of AF (see figures 4(c) and (d)). As shown in figure 4(d), the N/Pb ratios of DA blue , DA AM1.5G , and DA FL decreased significantly to 0.00, 0.09, and 1.14, respectively (see SI section 6 for all the data). This observation provides evidence that a blue laser or AM1.5G can completely detach the surficial organic cations within 1000 s.
For the AF samples depicted in figure 4(a), the rate of decrease of the I/Pb ratio slowed down gradually, and the ratio reached 2.0, which implies that the surface of PEA 2 PbI 4 is dominantly decomposed into PbI 2 . In contrast, for DA ( figure 4(c)), the rate of decrease of the I/Pb ratio did not slow down as it approached 2.0, instead, the sample continued to decompose. This indicates that the mechanisms of the DA and AF samples are fundamentally different.

Investigation of metallic Pb by photodegradation
We analyzed the Pb 4f core-level spectra to understand the underlying degradation mechanisms in detail. Figure 5 displays the Pb 4f spectra of the PEA 2 PbI 4 films illuminated by a blue laser (figures 5(a) and (b)), FL (figures 5(c) and (d)), and UV-filtered FL (figures 5(e) and (f)). The Pb 4f core levels located at approximately 138.5 eV and 143.5 eV (blue area) correspond to the Pb 2+ state that originated from well-structured PEA 2 PbI 4 . The new peak occurred at a lower binding energy (red area), corresponding to the Pb 0 state [36][37][38][39]42]. The atomic concentrations of the Pb 0 state calculated using the area ratio ([Pb 0 /(Pb 2+ +Pb 0 )]) are also denoted in each figure, where the values are AF blue (6.5%), DA blue (8.3%), AF FL (5.1%), and DA FL (7.5%). AF UV-F FL and DA UV-F FL did not exhibit any Pb 0 in the Pb 4f spectra. Figures 5(a)-(d) shows that more Pb 0 was created when the film was illuminated during the annealing process. Based on our observations, we suggest the following photodegradation model for PEA 2 PbI 4 . For the DA samples, we observed that the I/Pb ratio did not saturate at 2.0, and it continuously reduced until most of the I − ions on the surface were depleted. In contrast, for the AF-sample case, the I/Pb ratio slows down as it approaches 2.0. We suggest that this phenomenon occurred owing to the completeness of the structure. In the AF case, 2DRP perovskites have a vertically well-ordered structure. Thus, photodegraded PbI 2 can be formed on the top layer of the 2DRP perovskites, which can act as a passivation layer. Therefore, the photodegradation process was suppressed. In contrast, the DA samples do not have an orderly structure yet. Therefore, it is difficult for these samples to form a passivation layer of PbI 2 , and consequently, the photodegradation process is not suppressed until most of the I − ions on the surface are depleted. Figure 6 shows the normalized PL intensity of the films illuminated by different light sources during different parts of the fabrication process. (See SI section 7 for the spectra) The upper row (figure 6(a)) and lower row (figure 6(b)) represent AF and DA, respectively. The PL spectra were obtained statistically with at least ten samples in each batch to minimize the experimental error that may occur from sample-to-sample variation. The average PL intensity of the pristine films was plotted as a reference line of 1.0. Moreover, each film was measured only once to exclude the unwanted degradation effect that can be derived from the excitation light used in the PL measurement. As expected, a reduction in the PL intensity was clearly observed for films illuminated by the blue laser, AM 1.5G, and FL for both rows, whereas the films illuminated by the red laser and UV-filtered FL maintained their initial PL intensity. Similar to other measurement results, DA samples exhibited a faster reduction in PL than that of AF samples. This result implies that non-radiative recombination is enhanced by photodegradation [34,36,43,44]. Many studies have reported the generation of trap states over the surface and grain boundaries of perovskites after photodegradation. Among the trap states, it is well known that Pb 0 significantly reduces the PL intensity. Therefore, the PL spectra also support the faster photodegradation and increased creation of Pb 0 in PEA 2 PbI 4 for the DA cases.

Conclusion
We investigated the influence of light illumination on PEA 2 PbI 4 films during different parts of its fabrication process using various light sources. It was observed that only light with high energy, including that from a blue laser, AM 1.5G, and notably, the daily used FL at the laboratory, could significantly degrade the PEA 2 PbI 4 . In contrast, the red laser and UV-filtered FL did not cause photodegradation. From this study, we can explain the discrepancies in the surface morphologies previously studied. For instance, randomly oriented nanorod or rough surface of PEA 2 PbI 4 mostly stems from photodegradation. We observed that photodegradation occurred more actively when the films were illuminated during the annealing process, which was confirmed by the more severely broken morphologies and faster reduction of the relative atomic ratios. This was also confirmed by the strong quenching of the PL intensity. We suggest that the difference between the DA and AF samples is owing to the completeness of the structure. Because AF samples have a well-ordered structure, photodegraded PbI 2 can be formed on the top layer of 2DRP, which can act as a passivation layer. Consequently, additional photodegradation processes were suppressed, unlike in the unordered DA case. Our observations provide two key guidelines for the fabrication of PEA 2 PbI 4 films: (1) the daily use of an FL may be detrimental to the fabrication of high-quality samples. Therefore, UV filters must be used in laboratories; (2) dark conditions are highly recommended, at least during the annealing process, to avoid severe surface degradation. We believe that our observations provide useful guidelines and new insights to understand the photodegradation of 2DRP perovskite films.

Data availability statement
The data that support the findings of this study are available upon reasonable request from the authors.