Probing elemental diffusion and radiation tolerance of perovskite solar cells via non-destructive Rutherford backscattering spectrometry

Over the last decade, there has been a surge of interest in organic–inorganic halide perovskites (OIHPs) due to their potential as a promising optoelectronic material for applications such as solar cells,¹ indoor photovoltaic cells,² light-emitting diodes (LEDs),³ lasers,⁴ and photodetectors.⁵ OIHPs are of particular interest for their light-harvesting capability because of their wide optical absorption range,⁶ long electron and hole diffusion lengths (>1 µm),^7,8 long carrier lifetimes,⁹ high charge carrier mobilities,¹⁰ and tunable bandgaps via cation or halide substitutions,^2,11 which have led to extraordinary advances in the field of photovoltaics (PVs). Moreover, solution-processed deposition techniques for PSCs are appealing because of their high defect tolerance, which helps to reduce production costs and boost volume production. For terrestrial applications, PSCs are being looked at as a possible complement to traditional PV technologies due to their outstanding optoelectronic properties and impressive power conversion efficiency (PCE) of up to 26.1%, which is on par with commercially available crystalline-Si solar cells.¹² Recent studies have shown that unlike other inorganic materials, such as Si and GaAs, perovskites have remarkable tolerance to much higher radiation damage, making them an ideal candidate for space PV applications [e.g., in low earth orbit (LEO) satellites and deep space].^13–16 OIHPs exhibit electronic as well as ionic motion;¹⁷ ionic motion results in defect states and is one of the possible sources of the anomalous hysteresis behavior in PSCs detrimental to device performance and long-term stability.¹⁸ 

In the ABX₃-type crystal structure of perovskites, there can be numerous types of defects, such as vacancies (V_A⁺, V_B²⁺, and V_X⁻), interstitial (i_A⁺, i_B²⁺, and i_X⁻), anti-site substitutions (A⁺ ⇔ X⁻, and X⁻ ⇔ B²⁺), and cation substitutions (A⁺ ⇔ B²⁺);^19,20 each of these defects has different defect formation energies, which determines their location in the bandgap, i.e., defects with low defect formation energies tend to be shallow and cluster around the valence band maximum or the conduction band minimum, while defects with high defect formation energies tend to be found at deeper energy levels near the middle of the bandgap.²¹ Typical perovskite films are polycrystalline, and their multiple-grain boundaries are regarded to be one of the key defect formation sites; as a result, their defect density is much higher than that of a single-crystal perovskite.²² For the next stage of the development of PSCs, a better knowledge of defect formation and the variables that induce defect formation is essential. Normally, vacancies and interstitial defects are produced in significant quantities because of ion migration, which further combines with environmental stimulants, such as moisture and thermal stress to accelerate the perovskite’s deterioration process.²³ In general, OIHPs are an intriguing material family that offers a variety of desirable optoelectronic properties. Despite their ease of fabrication through solution-processing, comprehending their underlying nature remains complex due to various external factors (e.g., light and heat) that may influence their device performance. Depending on their intensity, these external factors can directly or indirectly cause ionic motion across layers. For example, it is highly likely that the ionic species in OIHP’s photoactive layer (e.g., I⁻, Br⁻, Cl⁻, Cs⁺, and Pb²⁺) and even other elements from different layers (e.g., Au, Sn) start moving across the bulk of the device, which starts and accelerates the process of degradation. Previous reports have indicated that Au⁺ interstitials in PSCs share a chemical resemblance with iodine interstitials. As a consequence, Au⁺ can readily generate and move through the perovskite using the same pathways as iodine interstitials, thereby playing a role in the ion migration.^24,25 To observe and understand such problems, it is important to look at the elemental depth profile of the device over time.

Regarding the exploration of ion movement in PSCs, the PV community has employed a diverse array of methods, such as electrochemical impedance spectroscopy (EIS), x-ray photoelectron spectroscopy (XPS), conductive atomic force microscopy (c-AFM), and thermal admittance spectroscopy (TAS).^26–29 Measuring the movement of various elemental species through these techniques proves challenging, given that most of these methods examine ion migration without elemental sensitivity and are indirect in nature. Recently, time-of-flight secondary ion mass spectrometry (ToF-SIMS) has been used to study PSCs, which is a direct elemental technique providing detailed elemental and molecular distributions of the various ionic and elemental species as a function of depth.³⁰ However, ToF-SIMS is a destructive technique that suffers from low spatial resolution and a time-consuming data analysis process. In a ToF-SIMS setup, secondary ions are sputtered from the sample using a primary focused ion beam (usually Cs⁺ or Ga⁺), and it requires the sample under study to be cleaved. Given that OIHPs have an organic component, they are susceptible to preferential sputtering and matrix effects, which can introduce inaccuracies in quantification and produce unintended artifacts when analyzed via ToF-SIMS.³¹ This also alters the physical state of the device rendering it to be non-operational for any post-characterization measurements on the same device, which is a major disadvantage of this technique. Rutherford backscattering spectrometry (RBS) presents a solution to these limitations as it is a non-destructive technique, meaning it does not alter the physical state of the sample (although it will create radiation-induced defects in the sample). This advantage is due to the fact that no sample preparation is required before conducting RBS on the sample, and the data acquisition process in RBS is much simpler and faster. Moreover, it can provide quantification of the elemental species present in a sample as a function of depth in terms of their areal density. This makes it a suitable technique for investigating how material loss in the sample varies over time. It is important to acknowledge here that RBS has a limitation compared to ToF-SIMS. RBS cannot specifically detect exact ion species; instead, it can identify the elemental compositions present within the sample as a function of depth. Despite this, RBS could serve as a complementary or even a preferable technique over ToF-SIMS in certain scenarios where the reusing of the studied sample is needed for repeated analyses. At present, there are only a limited number of investigations concerning a somewhat connected subject, involving the application of RBS to analyze perovskite films and PSCs.^32,33 Lang et al. conducted RBS on an MAPbI₃ perovskite film to gain a quantitative depth profiling of Pb and I.³² This research was conducted on a stack comprising just a few layers of a solar cell device, rather than on the complete assembly of a functional device, which differs from real-world conditions. Recently, Hussain et al. have investigated ion migration in an aged non-operational PSC having a device architecture of glass/ITO/NiO_x/CsPbI₂Br/C₆₀/Cu. This investigated device architecture can be seen as relatively straightforward for RBS fitting because the elemental diffusion of the various inorganic species can be easily distinguished.³³ This is because most part of the device in this structure consists of inorganic species with sufficient differences between their atomic masses to provide good RBS sensitivity and resolution. Additionally, there is minimal interference from the lighter organic components in this structure. However, the commonly used device architecture and perovskite compositions, such as triple cation mixed halide compositions, involve a substantial presence of organic species within the device.³⁴ This results in the introduction of additional variables when comparing fresh and aged devices. Consequently, this makes RBS fitting comparatively more challenging due to the complex interplay of these components. Therefore, in order to fully harness the capabilities of RBS for investigating ion migration in PSCs, it is essential to apply this technique to the commonly adopted device models.

The triple cation mixed halide OIHP photo-absorber [Cs_0.05(MA_0.17FA_0.83)_0.95Pb(I_0.83Br_0.17)₃] has been one of the widely used compositions in PV applications due to its ability to offer a highly durable device design and maintain consistent PCE values across repeated device fabrications.¹⁵ In the present study, an operational PSC having a device architecture of glass/ITO/SnO₂/Cs_0.05(MA_0.17FA_0.83)_0.95Pb(I_0.83Br_0.17)₃/spiro-OMeTAD/MoO₃/Au has been studied. The findings obtained through this research could offer valuable insights to the PV community, by introducing a non-destructive characterization method, that addresses contentious topics in photovoltaics—hysteresis, defect effects, radiation tolerance, and ion migration.

PbI₂ (99.99%) and PbBr₂ (>98.0%) were purchased from TCI America. Formamidinium iodide and methylammonium bromide were purchased from GreatCell Solar Materials. Li-TFSI [bis(trifluoromethane)sulfonimide lithium salt] was purchased from Alfa Aesar. CsI (99.999%) and 4-tert-butylpyridine (tBP, 96%) were purchased from Sigma-Aldrich. SPIRO [2,2′,7,7′-tetrakis(N,N-di-p-methoxyphenylamino)−9,9′-spirobifluorene] (≥99.5%) was purchased from Lumtec. The SnO₂ colloidal dispersion was obtained from Alfa Aesar and diluted to 1.5% in deionized water before use. All the solvents [N,N-Dimethylformamide (DMF), dimethylsulfoxide (DMSO), chlorobenzene, toluene, methanol, acetone, isopropanol, and methyl acetate] were obtained from Sigma-Aldrich.

Ce-borosilicate substrates (Martin Materials Solutions) with dimensions of 25.4 mm × 25.4 mm × 0.25 mm with ITO deposited in-house at NREL (sheet resistance ∼15 Ω □⁻¹) were cleaned by sequential sonication with acetone (15 min) and isopropanol (15 min). Substrates were blow-dried with nitrogen followed by 10 min of UV-ozone treatment. 150 µl SnO₂ colloidal dispersion was dropped on each substrate followed by spin-coating at 3000 r.p.m for 15 s. The coated substrates were placed at a hotplate set at 150 °C for 30 min. This was followed by a further 10 minUV-ozone after which the substrates were transferred to a N₂ glove box where perovskite active layer fabrication was completed as detailed below. PbI₂ (507 mg), PbBr₂ (73.4 mg), methylammonium bromide (22.4 mg), formamidinium iodide (172 mg), and CsI (15.6 mg) precursors were mixed in 1 ml of a DMF/DMSO solvent mixture (4:1 v/v) and vortexed to form a 1.26 M ink. The ink was filtered using a 0.45 mm nylon filter. The ink (50 µl) was dropped on the ITO/glass substrate and spun at 1000 r.p.m. for 10 s followed by 6000 r.p.m. for 20 s. Chlorobenzene (150 µl) was dropped in a continuous stream at the spinning substrates with 5 s remaining at the end of the spin cycle. This antisolvent rinse step changed the appearance of the spinning film from transparent to mild orange. After completion of the spin cycle, the substrate was immediately placed on a hotplate set at 100 °C for 60 min. Within seconds of contact with the hotplate, the film converted into the black perovskite phase. Spiro-OMeTAD was deposited next by dynamically spinning 10 µl of the spiro-OMeTAD solution at 5000 r.p.m for 15 s. This solution was prepared just before deposition by dissolving 36.1 mg spiro-OMeTAD, 14.4 µl tBP, and 8.8 µl Li-TFSI (520 mg in 1 ml acetonitrile) in 0.5 ml CB. After that, the MoO₃ layer was thermally evaporated on top of the spiro-OMeTAD layer, and then finally, 80 nm of gold was thermally evaporated by using a pre-designed shadow mask to make the top contacts on the PSC.

RBS on the PSC has been performed using the NEC 9SH 3 MV Pelletron accelerator in the ion beam laboratory (IBL) at the University of North Texas (UNT).³⁵ All the experiments were performed using a 2 MeV He⁺ beam in the microprobe beamline connected to the 9SH accelerator. The IBM experimental geometry has been used where the incident beam, the scattered beam, and the sample normal are all in the same horizontal plane such that the sum of the incident beam angle α (α = 0°), beam exit angle β (β = 35°), and the backscattered angle θ (θ = 145°) is 180°. The backscattered particles are detected by using a passivated implanted planar silicon (PIPS) particle detector from Mirion Technologies (Canberra), model No. PD25-11-300 AM, having a solid angle of 34 mili-steradian, and the operating voltage for the detector was 40 V.

The RBS data were analyzed using the SIMNRA software package,³⁶ where a simulated sample was created according to the thickness values of each layer in the PSC stack. Each of the layer’s concentrations was modified until a reasonable fit was obtained. SIMNRA accepts layer thickness in the form of areal density (atoms cm⁻²), which is not generally known. To convert the thickness into areal density, the SRIM/TRIM software package was used,³⁷ and detailed information on the process is given in the supplemental information. All the details of each layer in the PSC stack are shown in Table S1. An example of SIMNRA simulated spectrum along with the experimental RBS spectrum obtained with 2 MeV He⁺ beam on a spot of the active area of the PSC is shown in Fig. S1. To further extract the individual contribution from the perovskite layer and its elemental species, e.g., Cs, I, Pb, and Br, the MultiSIMNRA software package was used, which can act as a plugin for the existing SIMNRA software.³⁸ To simplify the discussion, the layers are referred to in the PSC by their assigned numbers unless it is necessary to specifically mention element(s) from those layers. The assigned numbers for the layers are as follows: layer-1 (Au), layer-2 (MoO₃), layer-3 (spiro-OMeTAD), layer-4 (perovskite), layer-5 (SnO₂), and layer-6 (ITO).

A freshly obtained N–I–P PSC with a moderate PCE of 16.09% was used in this study [schematic depiction of the PSC architecture and its current-density vs voltage (J–V) characteristic curve with its PV parameters are shown in Figs. 1(a) and 1(b), respectively], and a chosen spot on the active area underwent raster scanning via a 2 MeV He⁺ beam spanning an area of 469 μm × 469 μm. Since this marked the initial RBS attempt on PSCs, the degree of charge collection/fluence was deliberately restricted to minimize potential beam-induced damage. In the first scan, the active area was exposed to a radiation fluence of 1.76 × 10¹⁴ He⁺/cm². Figure 1(c) shows the RBS spectrum from scan-1, specifically highlighting the individual contribution originating solely from the perovskite and the top-contact (Au) layer. The contribution of the perovskite layer was further divided into individual contributions originating from elemental species, including Br, I, Cs, and Pb. Within the fresh PSC, the signals corresponding to Br, I, Cs, Pb, and Au were detected within energy ranges of 1264–1548 keV, 1362–1676 keV, 1398–1660 keV, 1440–1754 keV, and 1744–1870 keV, respectively. Subsequent to this, three additional successive scans were performed on the identical active region, exposing the sample to radiation fluences of 3.41 × 10¹⁴_, 3.95 × 10¹⁴, and 7.10 × 10¹⁴ He⁺/cm², in each respective run. At the end of the RBS experiment, the designated region of the PSC had experienced a cumulative radiation fluence of ∼1.62 × 10¹⁵ He⁺/cm² (2 MeV He⁺ is roughly equivalent to 500 keV H⁺ if the mass per nucleon is considered). This amount of radiation fluence surpasses the annual proton fluence by a few orders of magnitude in various outer space trajectories, as shown in Fig. S2. To assess the impact of damage induced by the beam on the PSC and to identify any indications of beam-induced ion migration, all spectra were compared collectively, as illustrated in Fig. 1(d).

Many studies have already confirmed the high tolerance of the PSCs to H⁺ irradiation with energies and fluences ranging from 50 keV to 68 MeV and 10¹¹–10¹⁵ H⁺/cm², respectively.^39–43 However, there is a limited body of research on the impact of He⁺ radiation on PSCs. Recently, Kirmani et al. have conducted the He⁺ radiation on one of the P–I–N-type triple cation mixed halide PSCs, subjecting it to a fluence of 2 × 10¹² He⁺/cm². This resulted in an ∼23% reduction in the initial PCE.⁴⁴ A plausible explanation for this decrease in PCE could be attributed to radiation damage caused by the 2 MeV He⁺ beam on the PSC. However, it was not previously established whether this damage could be responsible for ion migration within the PSC. Any movement in the mobile species Cs, I, Br, and Pb would cause shifts in the RBS spectrum. Yet, based on the observation presented in Fig. 1(d)_, no significant energy shifts are evident in any of the scans performed on the same spot of the PSC. This implies that the positions of the elemental species remain unchanged, thus ruling out the possibility of beam-induced ion migration during the RBS experiment. While it is clear that no elemental displacement is detected, PSCs are known to degrade under this high 2 MeV He⁺ fluence.⁴⁴ This points toward the possibility of alternate damage modes beyond displacement damage, given the unique soft lattice nature of perovskites, and needs to be further explored.⁴⁵ In terms of radiation tolerance, these results indicated that, from a structural perspective, the studied device architecture exhibited resilience against radiation. This level of radiation tolerance is a vital requirement for space PV applications, and the OIHPs demonstrate a notable edge compared to the Si and GaAs semiconductors, as the latter are prone to deterioration when subjected to a fluence of 10¹⁵ H⁺/cm.^40,46

Another noteworthy insight drawn from the RBS findings was the substantial difference between layer thicknesses calculated from RBS spectra on the active region of the PSC and the thickness values derived through conventional methodologies (such as cross-sectional scanning electron microscopy (SEM), and profilometer]. The contrast between these values is presented in Table I, and the energy-to-thickness conversion of each layer is shown in Fig. S3. Given that RBS data fitting necessitates an initial estimation of the layer composition used in the experiment, having accurate knowledge of the actual layer thicknesses is of paramount importance when conducting RBS analysis.

Following the scan of the active region of the device, the 2 MeV He⁺ beam was employed to raster scan a specific spot within the non-active area of the device, covering an area of 469 μm × 469 μm. In the initial scan, the non-active area was exposed to a total radiation fluence amounting to 7.56 × 10¹³ He⁺/cm². Figure 2(a) illustrates the data obtained from one of the RBS scans on the non-active region, showing a distinct representation of the contribution exclusively from the perovskite layer. This component was subsequently divided to reveal the individual contributions originating from elemental species, including Br, I, Cs, and Pb. Within the fresh device, signals corresponding to Br, I, Cs, and Pb were detected in the non-active region within energy ranges of 1350–1639, 1454–1765, 1482–1757, and 1533–1843 keV, respectively. Subsequently, two more scans were conducted on the identical location within the non-active area, with radiation fluences of 1.28 × 10¹⁴ and 2.81 × 10¹⁴ He⁺/cm² administered in each run. To investigate the influence of beam-induced damage on the non-active area of the PSC, along with any possible indications of beam-induced ion migration, a comparison of all spectra is represented in Fig. 2(b). Upon the completion of the three scans on the non-active area, the cumulative radiation fluence accumulated by this specific spot reached ∼4.85 × 10¹⁴ He⁺/cm², still notably surpassing numerous outer space trajectories (as illustrated in Fig. S2). As evident from Fig. 2(b), there were no noticeable shifts in the RBS spectra among any of the scans conducted on the non-active region of the PSC. These outcomes were similar to the findings of the RBS examination performed on the active area of the PSC, suggesting that beam-induced ion migration has negligible effects on ion migration, regardless of the scanning area within the device.

In order to investigate the impact of environmental degradation on elemental migration within the PSC, the device was subjected to ambient atmospheric conditions for five months, leading to the subsequent round of experiments. After this five-month period, another RBS scan was performed on the same active area of the PSC, utilizing a 2 MeV He⁺ beam (with a fluence of 2.77 × 10¹⁵ He⁺/cm² and a raster scanned area measuring 234 μm × 234 μm), as depicted in Fig. 3(a). At first glance, the spectrum appears quite similar to that of a fresh PSC. However, upon a more thorough comparison between the RBS spectra of the fresh device and the five-month-old device, it becomes evident that certain changes have taken place within the device, as illustrated in Fig. 3(b). From Fig. 3(b), it is noticeable that the higher energy edge of the top Au electrode remains unchanged in its position. However, there appears to be a shift toward the higher energy side of the spectrum in the contributions from the spiro-OMeTAD, perovskite, SnO₂, and ITO layers. This suggests that the device has undergone a reduction in the elemental concentrations of various layers, implying that the layers situated below the Au have experienced some material loss. This occurrence might be attributed to degradation, as a significant proportion of the organic material within the device could have been depleted or leached out over time. A similar assessment was carried out on a non-active area of the five-month-old PSC (with a fluence of 2.28 × 10¹⁵ He⁺/cm² and a raster scanned area measuring 234 μm × 234 μm), albeit the subsequent scan was conducted on a slightly different, close by non-active area after the five month duration. The RBS scan performed on this particular region is depicted in Fig. 3(c). The comparison between the RBS scans acquired from a non-active area of the fresh and the five-month-old PSC is illustrated in Fig. 3(d).

A similar pattern was observed within the non-active region, resembling the findings observed in the active area of the PSC. Although there was no shift in the high-energy edge of the spectra, the phenomenon of size reduction persisted. The outcomes from Fig. 3(d) closely resemble those observed in Fig. 3(b), with the exception that Au and MoO₃ layers are absent in the participating layers for the RBS. In this scenario, the spiro-OMeTAD and perovskite layers maintain their high energy edge, yet their thickness has diminished, likely attributed to degradation. This reduction in thickness possibly arises from the loss of a substantial amount of organic material from the PSC over time. An additional feature, represented by a peak within the energy range of ∼1380–1500 keV in Fig. 3(c), stems from the contribution of the ITO layer. This layer was not initially present at the spot chosen for the RBS spectra of the fresh device [depicted in Fig. 2(b)], where only the glass/SnO₂/Cs_0.05(MA_0.17FA_0.83)_0.95Pb(I_0.83Br_0.17)₃/spiro-OMeTAD layers existed. In contrast, the scan represented in Fig. 3(c) was performed on a non-active area where the signals from the ITO layer were inadvertently detected during the raster scanning of the He⁺ beam near the ITO boundary of the substrate. This situation was not observed in the earlier scan of the fresh non-active area, i.e., during the second round of the RBS at a non-active area, glass/ITO/SnO₂/Cs_0.05(MA_0.17FA_0.83)_0.95Pb(I_0.83Br_0.17)₃/spiro-OMeTAD layers were present.

Figure 4 illustrates a comparison between the fitted spectrum of the fresh PSC and the five-month-old PSC after degradation, aiming to demonstrate the contrast among the different elemental species present within the device. For the RBS fitting process in SIMNRA, after initially fitting the spectrum for the fresh PSC at the active area [Fig. 4(a)], adjustments were made to the concentrations of various elemental species within the device layers in order to achieve the most accurate fit for the five-month-old PSC [as depicted in Fig. 4(b)]. Interestingly, the quantity of Au within the uppermost layer-1 was noted to have decreased, with traces of C, O, and I being present at the top. Additionally, a significant amount of Au was detected in layer-3, layer-4, and layer-5. This observation holds significance due to the usual notion that Au migration is hindered in devices equipped with a MoO₃ layer positioned between the top contact and the hole transport layer.⁴⁷ However, contrasting outcomes were observed here, suggesting that layer-2 was ineffective in preventing Au migration within the aged device. The potential explanations for this could be attributed to the non-uniformity of layer-2 during the device fabrication process, which might have created specific regions within the film where Au atoms could migrate downward throughout the device. Alternatively, the presence of surface defects, such as oxygen vacancies in the MoO₃ itself, could be a factor contributing to this phenomenon. Given that MoO₃ acts as an oxidizing agent and halide ions can be potentially reduced, this opens possibilities for redox reactions between the metal oxide and the metal-halide adsorption/dissociation products. Consequently, this process can reduce Mo, leading to an increasingly defective and reactive MoO₃ surface with time.⁴⁸ This potential degradation of layer-2 is supported in Fig. 4(b) and Table S2, where faint traces of I and Br species were detected within layer-2 of the aged device. As for the primary source of elemental migration, the movement of I and Br was observed toward the upper portion of the device within layer-2 and layer-3, while Pb was identified within layer-3, layer-5, and layer-6.

The existence of these species in varying layers of the device had been previously validated through ToF-SIMS in prior studies.⁴⁹ Nonetheless, in the current study, the application of RBS has facilitated the simultaneous quantification of these species (as indicated in Table S2) without causing any modifications to the physical state of the device (i.e., no materials removal via sputtering unlike ToF-SIMS). A reduction in the presence of organic species, such as C, N, H, and O, was also noted within layer-3 and layer-4. This finding accounts for the observed contraction in the spectra of the five-month-old PSC, in comparison to the fresh one depicted in Fig. 3(b). This reduction is attributed to the release of degradation by-products, including volatile organic compounds, which might have dissipated from the device over time due to the ongoing degradation within the ambient atmosphere.^50–52 The migration of Sn and In species from layer-6 to layer-5 was also detected. A few studies corroborate this finding, as ITO tends to degrade under acidic conditions, potentially causing the migration of In and Sn species.^53,54 To elucidate the source of this acidic environment within the current device structure, one plausible scenario involves the degradation by-products of the perovskite layers. These by-products could include volatile acid species, such as HI and HBr, which may create the acidic conditions necessary to initiate ITO layer degradation.^52,53 However, further work is needed to verify these hypotheses.

Drawing insights from Figs. 4(c) and 4(d), it was inferred that elemental migration had also taken place within the non-active area of the PSCs. However, the extent of material loss from the non-active region was considerably greater compared to the situation in the active area. This is supported by the fact that only minute quantities of Cs, I, and Pb were detected in other layers. A small quantity of iodine was identified in the top spiro-OMeTAD layer. This observation implies that iodine migrated to the top surface within the non-active area, but it may have escaped the device leaving only a faint trail behind. Interestingly, a faint trace of Cs was unexpectedly identified on the upper surface of the spiro-OMeTAD layer within the non-active area, which contrasts with the behavior of Cs atoms in the active area. However, when comparing the concentrations of fresh and aged devices using Tables S2 and S3, it becomes apparent that the aged device’s non-active area contains notably lower quantities of Pb, I, and Br. This decline in elemental species within the non-active area can be attributed to the absence of the top Au contact. Considering the lack of Au in the non-active region, it is reasonable to suggest that volatile degradation by-products (which include Br and I) and aqueous compounds (containing Pb) could have escaped more extensively from the device in comparison to what was observed in the active area.^50,52 For a clearer understanding and improved visual illustration of the significant elemental movements within the active area of the device, a schematic is represented in Fig. 4(e), showing the movements of Au, I, Br, Pb, Sn, and In species from their original layer locations.

In summary, the utilization of RBS technique to non-destructively investigate the elemental depth profile of multi-layered PSCs and to explore the inter-diffusion of diverse elemental species within OIHPs across different interfaces with nanometer-scale spatial resolution has been effectively demonstrated within a high-performing PSC. The influence of beam-induced damage on ion migration and preferential sputtering appears negligible, as RBS would have identified any movement of heavy ion species, such as Cs, I, Br, and Pb, within the fresh PSC. This absence of movement was consistently observed even after subjecting the same PSC spot to multiple scans, accumulating a total radiation fluence of ∼1.62 × 10¹⁵ He⁺/cm². This remarkable radiation tolerance positions it as an ideal candidate for outer space missions, consistent with earlier works in the literature. Within the active area of the deteriorated device, RBS revealed the migration of elemental species, including Au, In, Sn, Pb, I, and Br. Further analysis is required to address uncertainties stemming from the fitting of lighter elements, such as C, N, H, and O. Conversely, within the non-active portion of the degraded device, the changes identified in the RBS spectra were mainly ascribed to material loss from the device resulting from the degradation. It is possible that a significant quantity of various elemental species might have escaped to a greater extent from the device within the non-active area, in contrast to what was observed in the active region.

As the characterization of the device holds significant importance in studies involving PSCs, the comprehensive mapping of the device’s elemental depth profile becomes essential to establish a genuine correlation between the device’s performance and its internal processes. Moreover, it is essential to explore various device architectures, as the components other than OIHP within the device may exhibit different behaviors under radiation when the device structure is changed. The utilization of the RBS technique holds the potential to significantly benefit these aforementioned studies. By enabling rapid device analysis, it facilitates the correlation of device characteristics observed in pre-irradiation, post-irradiation, and unirradiated states, all within the same device set, providing more reliable results.

Included in the SI are descriptions of the layer thickness conversion from nm to atoms/cm2 for all the layers of the PSC. Also included is a snapshot from the SIMNRA software, showing the fitting of the experimentally obtained RBS spectrum with the simulated spectrum obtained on an active area of the PSC. Simulated annual fluences from the Space Environment Information System (SPENVIS) as a function of proton are also provided. Additionally, a procedure to convert the mass density to the areal density or physical thickness of the sample is provided. Utilizing this procedure, the physical thickness of various layers of the devices is provided. A comparison of concentrations between the fresh and the aged device on the active area and non-active area of the device is provided.

UNT acknowledges partial support from NSF under Grant Nos. HBCU-EiR-2101181 and ECCS-2210722. This work was authored, in part, by the National Renewable Energy Laboratory, operated by the Alliance for Sustainable Energy, LLC, for the U.S. Department of Energy (DOE) under Contract No. DE-AC36-08GO28308. NREL authors were supported by the operational energy capability improvement fund (OECIF) of the Department of Defense. The views expressed in this article do not necessarily represent the views of the DOE or the U.S. Government.

The authors have no conflicts to disclose.

Mritunjaya Parashar: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Writing – original draft (equal); Writing – review & editing (equal). Mohin Sharma: Conceptualization (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Writing – review & editing (equal). Darshpreet Kaur Saini: Investigation (equal); Validation (equal); Writing – review & editing (equal). Todd A. Byers: Formal analysis (equal); Investigation (equal); Methodology (equal); Writing – review & editing (equal). Joseph M. Luther: Funding acquisition (equal); Investigation (equal); Methodology (equal); Resources (equal); Visualization (equal); Writing – review & editing (equal). Ian R. Sellers: Conceptualization (equal); Investigation (equal); Methodology (equal); Supervision (equal); Validation (equal); Writing – review & editing (equal). Ahmad R. Kirmani: Conceptualization (equal); Investigation (equal); Methodology (equal); Visualization (equal); Writing – review & editing (equal). Bibhudutta Rout: Conceptualization (equal); Funding acquisition (equal); Project administration (equal); Resources (equal); Supervision (equal); Validation (equal); Writing – original draft (equal); Writing – review & editing (equal).

The data that support the findings of this study are available within the article and its supplementary material. Any specific experimental and simulation fitted data can be supplied upon reasonable request.