X-ray dose effects and strategies to mitigate beam damage in metal halide perovskites under high brilliance X-ray photon sources

INTRODUCTION

Metal halide perovskite (MHP) materials have received enormous attention due to their photovoltaic and optoelectronic properties suitable for solar cells^[1,2], photonic devices^[3,4], and X-ray detectors^[5]. However, many aspects of their properties must be uncovered to move their technologies from lab to market^[6]. The MHP instabilities under operation and the correlated photo-structural-chemical transformations require advanced characterization tools coupled with in situ and operando experiments for a comprehensive nanoscale view^[7]. X-ray fluorescence (XRF)^[8], X-ray diffraction (XRD)^[9,10], X-ray beam-induced current (XBIC)^[11], X-ray excited optical luminescence (XEOL)^[12] and coherent diffraction imaging (CDI)^[13] with focused beams are increasingly sophisticated and accessible at fourth-generation synchrotron sources, becoming powerful tools to investigate MHPs, or even complete perovskite solar cells (PSCs). Nevertheless, in most cases, those tools strongly interact with the materials, leading to undesirable transformations. Although promising, focused X-ray-based techniques face high radiation doses deposited on the samples, which may cause damage to the materials and devices, altering their properties. Stuckelberger et al. showed that the electronic properties of the perovskite are the most sensitive, leading to a drastic reduction of the XBIC signal under X-ray irradiation but with no evidence of the element’s composition changes^[14]. X-ray photoemission (XPS) experiments also demonstrated the sensibility of the electronic properties, suggesting that the organic part of the perovskite [formamidinium (FA) and methylammonium (MA)] suffer radiolysis, producing volatile species^[15,16]. Ferrer Orri et al. revealed that the degradation mechanism for electrons and X-rays is similar and is probably caused by the interaction of electrons, also generated by the X-rays, with the organic part of the MHP. In defective grain boundaries, even low doses of electrons change grain orientations^[17]. Also, the degradation seems to be facet-dependent, indicating that the sample crystalline quality is crucial to the resilience to X-ray damage^[18]. Interestingly, the cryogenic conditions do not avoid degradation but favor sample amorphization^[19,20]. Conversely, temperatures around 90 °C do not induce additional severe perovskite degradation^[19]. Besides the critical insights already uncovered, a comprehensive investigation of several X-ray dose aspects on the morphological, structural, and chemical properties of the MHP films extensively applied in PSC devices remains missing.

This work uses a correlative multi-technique approach to investigate the effect of the focused X-ray beam on the element’s distribution, morphology, structure, and optoelectronic properties of the [(FAPbI₃)_0.87(MAPbBr₃)_0.13]_0.92(CsPbI₃)_0.08 (CsFAMAPb(Br,I)₃) compound. Such composition is relevant because it is widely used in fundamental studies and devices and presents a mixture of the main elements/molecules used in MHP. We irradiated the samples with controlled X-ray doses under different atmospheres and temperatures and evaluated the damage using correlative multiple micro and nanoscale techniques. The correlative data showed that the X-ray dose caused damage on the exposed surface, mainly affecting the iodine and organic elemental components, leading to topographical changes by ejecting material and forming an excavated area. The local optical reflectivity within this area indicates the formation of a metallic layer coating the surface. Surprisingly, the bulk compound underneath the excavated area preserved its optoelectronic properties. As expected, decreasing the X-ray dose and using an inert atmosphere reduced the damage. However, selective ejection seems to be independent of the temperature conditions. We demonstrate that combining an inert atmosphere with a controlled dose reduction at room temperature is the way to investigate hybrid organic-inorganic MPHs with state-of-the-art X-ray nanoprobes. Finally, we applied the optimized conditions below the critical dose to demonstrate the feasibility of studying MHPs with the X-ray ptychography (CDI) technique to highlight the grain morphology and sample details with a 15-nm spatial resolution. Such a technique had never been reported before in this class of compounds.

EXPERIMENTAL

RESULTS AND DISCUSSION

Figure 1A shows the nano-XRF map of the iodine emission of a 5 × 5 µm² irradiated area acquired at the RT-Air condition. To irradiate and map the sample, we used the fly-scan snake mode with the scan starting on the left-bottom corner of the image and the beam spot going up and down while sliding to the right related to the sample. The X-ray dose after irradiation corresponds to 2.9 GGy deposited in the scanned region, where 1 Gy = 1 J/kg is a measure of the total energy deposited per quantity of material irradiated^[33,34]. The estimated dose considers the total absorption of the compound at the irradiation energy so that doses deposited using distinct X-ray energies can be appropriately compared. The total amount of absorbed 10-keV photons by the 330-nm-thick perovskite film corresponds to 2.53 × 10²¹ ph/m². We present the dose calculation and its Python code in the Supplementary Information (SI).

Figure 1. X-ray dose effect of raster scan in a hybrid MHP sample by an X-ray focused beam (2.9 GGy) in the air at room temperature by multiple techniques. (A) Iodine nano-XRF map collected during the high X-ray irradiation. (B-D) Iodine, lead, and bromide, respectively, nano-XRF maps of the irradiated area. (E) µ-FTIR map and (F) FTIR spectra at two different locations of the C-N antisymmetric stretching of FA (1,700 cm^-1). (G) optical reflectivity image of the irradiated sample. (H) AFM topography image; (I) photoluminescence spectra.

The decreasing iodine XRF emission in the irradiated region is observed in Figure 1B. To inspect this irradiated region, we performed a larger scan of 10 × 10 µm² centered at the same point as the previous 5 × 5 µm² scan but with a much lower dose. The dose for the measurement corresponds to about 0.2 GGy in the scanned region, representing about 7% of the dose used for the irradiation. The iodine map reveals a reduction of the number of iodine atoms in a square with the exact dimension of the previously irradiated region. The average fluorescence counts correspond to about a 28% decrease. Interestingly, the Pb and Br maps do not show any evidence of composition changes [Figure 1C and D]. A noticeable additional effect [Figure 1A] is the decreasing iodine XRF emission from left to right in the scan direction. This gradient is not fortuitous but is systematically observed in various samples under high doses [Supplementary Figure 2]. As the horizontal nanoprobe size (500 nm) is larger than the fly scan step (200 nm), the initial irradiation (first vertical lines) hits a fresh sample. In contrast, the following vertical lines have already seen the beam and lost material.

As the hybrid organic-inorganic perovskite comprises a mixture of MA and FA, which are not observable by XRF, the effect on the organic part was evaluated using µ-FTIR. Figure 1E shows a µ-FTIR map obtained by the integration of the C-N antisymmetric stretching vibration (1,700 cm^-1) of the FA molecules^[35]. The decreased intensity in the X-ray irradiated region indicates that the organic molecules were also removed or degraded in the same area as for iodine. Comparing the complete FTIR spectra in the inner damaged region with the outer part (reference) reveals a change in the background for higher wavenumber [Figure 1F], indicating a compound with a more metallic characteristic and different refraction indices on the sample surface. This hypothesis is also qualitatively corroborated by a brighter color observed in the optical microscopy image [Figure 1G]. In addition, a scanning electron microscopy (SEM) measurement performed in a similar sample shows [Supplementary Figure 3] the damaged area. One can observe that the grain morphology is unaffected and a layer is formed on top of the MHP. The AFM measurements revealed a remarkable effect of the beam on the film topography [Figure 1H]. We observed an excavated depth in the irradiated area, corroborating the material consumption for iodine and organic molecules. We expected that the decrease in iodine and organic cations content would lead to changes in the PL peak position. Surprisingly, however, the peak in the irradiated region remained stable (in shape and position) with just a decreased intensity corresponding to a decrease in sample thickness. Figure 1I, showing the PL spectra acquired using a µ-PL instrument with a spot size of 1 µm, indicates that the remaining perovskite compound underneath the excavated area preserves the bandgap properties. Supplementary Figure 4 shows the decrease in PL intensity through a line scan in the damaged area.

Figure 2 shows the X-ray dose effects investigated RT-Air conditions with a fixed photon flux yielding an absorbed dose rate of 4.7 KGy/s. The total irradiation time on the area of 5 × 5 µm² was adapted so that the samples absorbed three different dose levels: 2.9 (as in Figure 1), 1.9, and 0.7 GGy. As mentioned above, the 10 × 10 µm² inspection corresponds to an additional 0.2 GGy. The equivalent of 10-keV photons is given in Supplementary Table 1. The data set in Figure 2A is adapted from Figure 1 and plotted along with the two new sets with medium (1.9 GGy) and low (0.7 GGy) doses [Figure 2B and C] for comparison.

Figure 2. Absorbed X-ray dose effect for three different doses in air atmosphere. From left to right, nano-XRF irradiation, nano-XRF inspection, and C-N antisymmetric stretching µ-FTIR maps for doses of (A) 2.9 GGy, (B) 1.9 GGy, and (C) 0.7 GGy. (D) The relative decrease in the iodine content, and (E) µ-FTIR intensity.

The iodine consumption attenuates as the total deposited dose reduces to 1.9 and 0.7 GGy (Figure 2B and C, respectively). Comparing the ratio between the iodine fluorescence in a 1 × 1 µm² area in the inner and the non-irradiated regions, we note that the highest dose leads to an iodine consumption of about 28%, attenuated to 16% at 1.9 GGy and 5% at 0.7 GGy [Figure 2D]. The last measurement is within the error bar and is barely seen on the XRF map.

Regarding the organic cations, the µ-FTIR map (Figure 2A-C, FTIR column) shows the expected tendency of diminishing the beam damage with a smaller amount of energy deposited on the sample. The dose effect is mitigated with an attenuation down to 0.7 GGy dose. The tendency is clear by evaluating the integrated profile at the 1,700 cm^-1 resonance (C-N antisymmetric stretching) in the irradiated region, which gives about a 30% decrease for the two highest doses and is absent for the smallest one [Figure 2E]. The profile shows a reduction broader than in the irradiated area (pink column), which should come partially from the limited spatial resolution at 1,700 cm^-1 due to the diffraction limit at this wavelength (5,882 nm). This broadening reduces at shorter wavelengths, as shown by the µ-FTIR images at the region of N-H stretching vibrations (3,100-3,400 cm^-1) [Supplementary Figure 5], with a sharper profile more consistent with the irradiated area.

The decrease in iodine and organic cation FA corroborates with the excavation in the irradiated area. The total thickness of the perovskite was previously measured by a profilometer to be around 330 nm. Taking the depth observed in the AFM topographic images [Supplementary Figure 6], the ratio of the excavation to the film thickness is estimated to be 27% (90 nm) and 21% (70 nm) from the highest to the intermediated dose, respectively. These ratios are comparable to the iodine attenuation shown in Figure 2D. We should point out that error bars for these measurements are within 15%, as measured by the statistical fluctuations outside the irradiated area.

These experiments were performed using the same dose rate and changing the exposition time. One can inquire which parameter is relevant, dose rate or total dose. To verify if the damage depends on the dose rate, we measured the iodine XRF signal as a function of time at a fixed sample position for two rates (1.2 and 0.23 GGy/s) at the energy of 10 keV. Figure 3A shows that the higher dose rate causes the expected faster decay in the iodine content. However, when we present the iodine consumption not as a function of time but as a function of the total absorbed dose [Figure 3B], we note that the iodine decay for both dose rates is superimposed. The decaying signal fits with two exponential components for both dose rates and gives similar decay coefficients: 972 and 28 GGy^-1 for 1.2 GGy/s and 916 and 36 GGy^-1 for 0.23 GGy/s, suggesting that the process occurs independently of the dose rate. A reasonable hypothesis could be that the free surface excavation is faster initially. However, as more iodine and organic molecules are knocked out, the remaining atoms at the surface, Pb and Br, form a thin coating metallic layer (evidence from FTIR broad band formed and optical images), slowing the process. Nonetheless, additional experiments are necessary to confirm this hypothesis.

Figure 3. Effect of the absorbed dose rate on iodine signal with the X-ray beam in a fixed position for two different absorbed dose rates. (A) Normalized iodine XRF signal in function of time and (B) the dose.

This experiment demonstrates that reducing the dose rate is ineffective in mitigating beam damage in the air. What matters is the integrated absorbed dose needed to extract reliable information from the X-ray techniques. A similar experiment was performed at two photon energies, 10 and 14 keV, below and above the Pb and Br absorption edges for completeness. The X-ray fluxes were controlled to achieve the same absorbed dose rate at both energies, accounting for absorption and photon energy variation. Supplementary Figure 7 shows that the variation in iodine content is the same for both X-ray energies provided the absorbed dose is the same, corroborating the observation that the total dose absorbed by the material is the determining factor to the iodine consumption in air and at room temperature.

The atmosphere and temperature conditions were investigated as another strategy to mitigate the beam damage. The same dose conditions (2.9, 1.9, and 0.7 GGy) were applied in three environmental situations: RT-Air, RT-N₂, and Cryo-N₂. The RT-Air condition corresponds to previously discussed measurements [Figures 1 and 2]. At the RT-N₂ condition and the highest dose of 2.9 GGy, a faint irradiated area was still observed (Figure 4A, XRF columns), yielding iodine reduced by about 16%, less than the same dose at the RT-Air condition [Figure 2]. The Cryo-N₂ condition (Figure 4B, XRF columns) seems to mitigate the effect in the XRF maps, with no observable iodine loss in the scanned region, as observed in Figure 4C (within the error bar of ±15%). The Cryo-N₂ condition seems to freeze the iodine mobility. Surprisingly, the damage persists in the organic part for both RT-N₂ and Cryo-N₂ (Figure 4A and B, µ-FTIR column), with similar intensity [Figure 4D]. The AFM topography (Figure 4A and B, AFM column, Figure 4E) confirms that the inert atmosphere does not entirely mitigate the beam excavation. An additional effect observed in the Cryo-N₂ condition is the formation of cracks in the irradiated region.

Figure 4. Effect temperature and atmosphere on damage from an X-ray dose of 2.9 GGy. From left to right, nano-XRF damage scan, nano-XRF probe scan, µ-FTIR, AFM and optical images for samples measured at (A) room temperature and nitrogen obtained by a nitrogen flow and (B) cryogenic temperature and nitrogen obtained using a cold nitrogen flow. (C) Attenuation in the iodine content, (D) µ-FTIR intensity and (E) AFM depth profiles for the different conditions.

Figure 5 shows the nano-XRF and µ-FTIR of the regions measured with the smaller doses, 1.9 and 0.7 GGy, at RT-N₂. Reducing the absorbed dose and combining it with an inert atmosphere effectively mitigate the damage. The inorganic elemental distribution is stable, even if the organic component still suffers a small reduction. The topographic look-up by AFM (not shown) did not present any observable excavation. Interestingly, the PL properties are almost unaltered at the RT-N₂ condition, with a slight blueshift emission compared to the pristine sample (Figure 5C and D, column µ-PL).

Figure 5. Perovskite damage with reduced dose and inert atmosphere. Sample irradiated with (A) 1.9 GGy and (B) 0.7 GGy in RT-N₂. From left to right, nano-XRF damage scan, nano-XRF probe scan and µ-FTIR (1,700 cm^-1). (C and D) Respective µ-PL at RT-N₂.

The high X-ray dose (2.9 GGy) on hybrid organic-inorganic MHP produces an ion-selective ejection at the film surface, consuming iodine and organic cations but leaving Pb and Br composition unaltered, causing an excavation and altering the visible optical surface reflectivity. As expected, the X-ray dose reduction attenuates the damage, with almost complete mitigation of the consumption of iodine and organic molecules at the lowest dose (0.7 GGy) even in the air atmosphere. Under a similar dose but N₂ atmosphere, damage effects are attenuated, suggesting that the inert atmosphere is relevant to mitigating iodine loss from the sample and that the air (O₂ and H₂O) is somehow active in the iodine remotion process. The effectiveness of the N₂ atmosphere in mitigating damage has already been reported in these compounds, even though the mechanism was unclear^[12]. Nevertheless, even in an inert atmosphere, there is still damage under the highest dose of 2.9 GGy. Such a critical dose should be avoided when investigating these compounds. The situation seems to be worse under cryogenic conditions. The formation of cracks in the irradiated region is probably caused by the decreased ionic mobility that helps to homogenize defects and relaxes stress. Combining the intermediate dose (1.9 GGy) with N₂ atmosphere seems safe to mitigate effects and keep photovoltaic properties unaltered [Figure 5].

Figure 6 presents a schematic model for the damage caused by the X-ray beam in the organic-inorganic MHP. The X-ray absorption cross-section of the light elements, such as carbon and nitrogen from the organic part of the perovskite, is very weak, and the organic molecules are essentially transparent to the X-ray beam. On the contrary, the inorganic cage of heavy atoms absorbs the X-ray photons more efficiently, decaying through varying pathways, including the emission of a cascade of electrons that interact more effectively with the organic molecules, eventually creating ions or radicals. Both are reactive and can be neutralized or undergo reactions with another species^[36,37]. Several reactions have been reported to produce new species that can be trapped in the structure or released as a gas, such as PbBr₂ crystallite or I₂ molecules^[17,38]. This process will generate new punctual defects in the compound, eventually modifying their properties. However, if several defects are formed near each other, the 3D perovskite cage will be destroyed. Therefore, a small or intermediate dose could produce a defective material compared to a destroyed one by a severe dose.

Figure 6. Schematic representation of the X-ray beam damage on the MHP.

The conservation of energy and momentum dictates that the electron must transfer energy to an atom when they interact in elastic scattering. If the transferred energy exceeds the displacement energy (E_d), the atom moves to an interstitial position, creating a defect. The E_d depends on the bond strength, crystal lattice, and atomic weight of an atom^[39]. As the perovskite is composed of medium to heavy elements, an intense displacement in the bulk is not expected. However, if an atom is at the surface, the E_d is lower, and the atom can leave the specimen in a sputtering-like process^[39]. One expects a high sputtering contribution from the organic molecules, independently of the atmosphere and temperature, because they are more volatiles. However, those from the inorganic elements (Pb, I, and Br) should be smaller unless they have high mobility and are more reactive, as is the case of the iodine ions exposed to the atmosphere of the air. However, even in this condition, the contribution of sputtering from inorganic elements should be smaller compared with organic molecules.

Hence, we hypothesize that the MHP damage mechanism combines radiolysis and sputtering, with the first making a larger contribution. Radiolysis produces reactive species that react with other elements, forming volatile species that leave the sample. In the air, the O₂ and H₂O molecules can react with perovskite in various ways, creating solid species such as PbO, Pb(OH)₂, PbCO₃, PbIOH, Pb, PbI₂, and PbBr₂^[17,40,41]. Ferrer Orri et al. reported forming the PbBr₂ crystalline phase as a product after electron or X-ray irradiation^[17]. Also, volatile species, such as I₂, HI, and deprotonated or degraded FA/MA molecules (CH₃I, NH₃, HCN, C₃H₃N₃), can form and leave the sample^[42-44]. The iodine departure is corroborated by the absence of iodine accumulation at the boundaries of the 5 × 5 µm² irradiated area. These reactions occur close to the surface where the sputtering is more intense, and volatile molecules face a short path to escape. The N₂ atmosphere inhibits the O₂ and H₂O degradation paths. Consequently, the degradation is less severe, and inhibiting the formation of the iodine volatile species limits the iodine loss. However, the organic molecules still suffer from radiolysis and sputtering, independent of the environmental conditions.

To highlight the relevance of controlling the X-ray dose and environmental conditions when applying X-ray nanoscopy for perovskite investigation, we report an X-ray ptychography study in an MHP compound. X-ray ptychography^[13] is a phase contrast technique that reveals morphological aspects not seen in traditional scanning X-ray microscopy in transmission (STXM) or XRF modes. As a phase contrast technique, ptychography is more sensitive to low-Z elements than absorption contrast techniques. We prepared and intentionally irradiated a Cs_0.05FA_0.95PbI₃ perovskite film, depositing a severe dose of 350 GGy on a single spot of 200 × 500 nm², corresponding to about 3.1 × 10²³ 10-keV-photons/m². After irradiation, we performed a ptychography map on a 10 × 10 µm² surface with steps of 100 nm, depositing a smaller dose of 0.0015 GGy. With such a dose, observing the XRF map is possible even if the signal/noise is limited. Figure 7A shows the ptychography phase analysis, revealing the heterogeneous morphology of the sample. The calculated resolution is about 15 nm. The XRF map (integrated over iodine and cesium fluorescence emissions) collected simultaneously [Figure 7B] reveals less resolved details owing to the poorer statistics and worse spatial resolution. The inorganic damage around the damaged region can be barely seen. With X-ray ptychography, one observes a larger damaged region extending over a region bigger than 2 µm. This result means that the irradiation probably affects the organic light elements over a larger area around the irradiation position. Additionally, one can observe features such as grain morphology and some cracking that are not visible with fluorescence.

Figure 7. Simultaneous X-ray ptychography and fluorescence images of a perovskite. (A) Ptychography and (B) iodine and cesium nano-XRF images of Cs_0.05FA_0.95PbI₃ perovskite compound.

The intentional irradiation removed some iodine over an area of about 1 µm² but depleted the organic molecules (seen by ptychography) over a larger area than the spot size. The possibility of measuring in fly-scan mode and under a controlled dose and N₂ atmosphere was essential to mitigate eventual sample damages. The phase contrast using very low doses clearly brings new insights into the morphology and local heterogeneities of MHP compounds. Hu et al. reported an X-ray ptychography measurement of an inorganic perovskite CsPb_0.6Sn_0.4I₃ but did not discuss the resolution^[45]. To our knowledge, this is the first X-ray ptychography demonstration in MHP with organic components. Ptychography is a high-sensitivity technique accessing subtle details and contrasts in sample morphology related to the material's electronic density. We foresee that our results pave the way to new correlative in situ and operando experiments with MHP. Specifically, this method is a prospect as a new probe for the early stages of degradation in MHP, including hybrid, inorganic, and Pb-free families, which were, up to now, inaccessible through other techniques.

CONCLUSIONS

Our results bring fundamental insights into the damage caused by X-ray beams on MHP compounds. We effectively applied a correlative multi-technique approach to investigate the damage caused by highly intense focused X-ray beams on hybrid organic-inorganic MHPs. By irradiating the compounds with high X-ray doses, we observed several effects that helped us understand the damage process.

Using focused X-ray beams from newly available synchrotron radiation sources for perovskite characterization will be challenging as the absorbed dose determines the damage, and each technique requires different doses to produce enough signal-to-noise ratio to extract information. Then, evaluating how the beam damage affects the desired properties as the dose increases is highly recommended before any material investigation. Our results show that combining an inert atmosphere with a controlled intermediate dose is compulsory to mitigate beam damage and provide conditions to understand the photo-structural-chemical properties of MHPs. Contrary to the expected common sense, cryogenic conditions are not the best for performing X-ray measurements under high doses because they reduce mobility and cause morphological changes. Thoroughly reporting the X-ray deposited power on samples, i.e., the absorbed dose in Gy, and environmental conditions will help the community to find the best experimental conditions to work safely.

As a clear demonstration of how powerful the use of sophisticated new techniques in synchrotron radiation facilities can be, we applied the optimized conditions below the harmful dose to demonstrate the feasibility of investigating MHPs with the X-ray ptychography technique, uncovering morphological details never seen before in MHP. Our multispectral images are a pivotal step in developing and applying the X-ray ptychography technique in beam-sensitive samples for high-resolution imaging, especially in the case of heterogeneous and hierarchical functional materials in which the multiscale properties determine final device performances.