Short-pulsed proton irradiation stability in slot-die-coated bifacial perovskite photodetectors Authors: Hryhorii P Parkhomenko, Yerassyl Yerlanuly, Aidana K Azamat, Chang-Qi Ma, Marat Kaikanov, Annie Ng, Askhat N Jumabekov Published: 2025 ABSTRACT AbstractFlexible perovskite-based photodetectors are emerging as promising candidates for next-generation optoelectronics, particularly in aerospace and radiation-intense environments. However, their stability under high-energy particle exposure remains insufficiently understood, especially for devices fabricated using scalable deposition methods. In this work, we investigate bifacial flexible perovskite photodetectors fabricated via slot-die-coating and subjected to short-pulsed (150 ns) 170 keV proton irradiation at fluences of 1012 and 5·1012 protons/cm2. The external quantum efficiency (EQE) decreased by ~ 30–40% after low-dose irradiation and by ~ 50% at higher fluence, with stronger losses observed for illumination through the MoO3/Au/MoO3 (dielectric-metal-dielectric, DMD) side. Spectral responsivity declined from 0.39 A/W (PET-side) and 0.30 A/W (DMD-side) in reference devices to 0.20 A/W and 0.16 A/W, respectively, after high-dose exposure. Similarly, peak detectivity dropped from 5.9·1011 Jones to ~ 3·1011 Jones, reflecting the combined effects of reduced charge collection and increased noise. These results demonstrate that flexible slot-die-coated bifacial perovskite photodetectors possess high radiation tolerance, highlighting the robustness of scalable device architectures for use in space, aerospace, and high-energy physics applications.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-33200-6.Keywords: Proton irradiation, Slot-die coating, Flexible, Bifacial, Perovskite, PhotodetectorSubject terms: Materials science, Optics and photonics, Physics Introduction The rapid advancement of hybrid organic-inorganic perovskite materials has revolutionised the field of optoelectronics over the past decade, offering remarkable potential for a wide range of applications, including photovoltaics, light-emitting diodes, lasers, and photodetectors1–5. Among these, perovskite-based photodetectors have gained increasing attention due to their superior optoelectronic properties, such as high absorption coefficients, tunable band gap, long carrier diffusion lengths, and low defect densities6–8. These attributes enable high responsivity, fast response times, and low detection limits, making perovskite photodetectors attractive candidates for next-generation sensing platforms9–11. Recent efforts have expanded the design of perovskite photodetectors to flexible and bifacial configurations, enabling conformable and lightweight devices that can collect incident light from both sides12–14. Such capabilities are especially valuable in applications like wearable health monitoring, robotic vision, curved imaging surfaces, and particularly in spaceborne and aerospace systems, where weight, mechanical flexibility, and omnidirectional sensitivity are essential15–20. Realising such architectures at scale requires fabrication methods beyond laboratory-scale spin coating. Slot-die coating, compatible with large-area and roll-to-roll processing, is a promising route for manufacturing flexible perovskite devices21–23. Proton irradiation, depending on energy and fluence, induces damage through both ionisation and displacement effects. These interactions can create deep-level traps, disrupt lattice structures, and cause phase segregation in perovskite films, leading to increased recombination losses and reduced photoconversion efficiency24–27. While a lot of studies have addressed the radiation hardness of perovskite solar cells, particularly under continuous ion fluxes or gamma irradiation24–29, relatively few have focused on short-pulsed proton irradiation24,30. Pulsed proton irradiation is particularly relevant because the entire fluence is delivered within a nanosecond-scale interval, producing an extremely high instantaneous dose rate. Under these conditions, the material experiences rapid ionisation spikes, transient local heating, and a dense burst of displacement events before any significant thermal or structural relaxation can occur31–33. As a result, the defect formation dynamics may differ from those generated under continuous irradiation, potentially leading to distinct trap distributions or defect clustering. Moreover, short, intense particle bursts are characteristic of a condition more representative of real-time particle events in certain harsh environments such as space, particle accelerators, and nuclear facilities18. Studying the pulsed regime provides a more realistic representation of the operating conditions that flexible perovskite photodetectors may encounter in practical radiation-exposed applications. Furthermore, most existing radiation studies have been limited to rigid, spin-coated perovskite devices, which do not represent the practical geometries or fabrication strategies needed for scalable deployment. Therefore, there remains a lack of data on how slot-die-coated, flexible perovskite devices under high-energy irradiation, which can significantly degrade the performance and stability of semiconductor layers and interfaces28,29,34. This gap is especially pronounced for bifacial architectures, where additional optical and structural factors further influence radiation response. In this work, we conduct a systematic investigation of flexible, slot-die-coated bifacial perovskite photodetectors exposed to short-pulsed (150 ns, 170 keV) proton irradiation. By systematically analysing photodetector parameters of the reference (as-prepared) and irradiated (with fluences up to 5·1012 protons/cm2) devices. The obtained results demonstrate the viability of flexible bifacial perovskite photodetectors for operation in radiation-intense environments and advance the broader development of radiation-hardened, large-area optoelectronic technologies. Results and discussion A schematic layout of the flexible bifacial perovskite-based photodetector on polyethylene terephthalate (PET) substrate is illustrated in Fig. 1a. In the first experiment, devices were subjected to a single proton pulse of 150 ns duration and 170 keV energy, with a total fluence of 1012 protons/cm2. A second batch of photodetectors underwent five successive pulses under the same conditions, leading to an accumulated fluence of 5·1012 protons/cm2. A beam of protons was directed from the MoO3/Au/MoO3 electrode towards the PET substrate. The total proton dose was derived by integrating the current density profile of the beam, obtained from its oscillogram, over the exposure time (see Fig. S1 in the Supplementary Information). Figure 1b presents the band alignment diagram of the device, where the SnO2 layer serves as the electron transport layer (ETL) and Spiro-OMeTAD as the hole transport layer (HTL), indium tin oxide (ITO) and MoO3/Au/MoO3 (dielectric-metal-dielectric, DMD) as transparent electrodes. The band energies presented in Fig. 1b are adopted from literature sources15,24. The full device configuration is PET (125 μm)/ITO (150 nm)/SnO2 (30 nm)/Perovskite (550 nm)/Spiro-OMeTAD (200 nm)/MoO3 (10 nm)/Au (10 nm)/MoO3 (35 nm). (a) Structural schematic of the perovskite photodetector; (b) band alignment of the functional layers; (c) SRIM-simulated trajectories and stopping profile of 170 keV protons within the device stack. To evaluate the interaction of protons with the multilayer device stack, a Stopping and Range of Ions in Matter (SRIM)35,36 simulation was carried out. The model replicated the experimental conditions, with 170 keV protons directed perpendicularly through the MoO3/Au/MoO3 DMD electrode. The experimentally measured thicknesses by profilometer of each functional layer were used, while mass densities were adopted from both published sources37–39 and SRIM’s built-in database. As shown in Fig. 1c, the SRIM simulation reveals pronounced scattering and energy loss of the incoming protons across the entire device stack. The corresponding displacement estimate, presented in Fig. S2 of the Supplementary Information, confirms that all layers experience displacement events. This outcome highlights the suitability of the chosen irradiation setup for probing radiation-induced degradation18. The external quantum efficiency (EQE) is one of the key parameters describing the spectral response of a photodetector. It quantifies the ratio between the number of charge carriers collected in the external circuit and the number of incident photons at a specific wavelength. Mathematically, it is expressed as40: where Jph denotes the photocurrent density, I(λ) represents the optical intensity at wavelength λ, q is the elementary charge, and hν corresponds to the photon energy. Figure 2a shows the EQE spectra measured under short-circuit conditions for illumination from both the PET-side and the DMD-side. In the reference (non-irradiated) devices, broad photoresponse windows were observed. For PET-side illumination, the detectors exhibited sensitivity across 350–750 nm, with a full width at half maximum (FWHM) of about 400 nm and peak EQE values exceeding 75%. For DMD-side illumination, the EQE extended from 420 to 750 nm, with a narrower FWHM of ~ 330 nm and a maximum EQE surpassing 55%. (a) EQE, (b) responsivity, and (c) detectivity spectra of the perovskite photodiodes, respectively. Proton irradiation significantly affected device performance. At a low fluence of 1012 protons/cm2, EQE dropped by approximately 30% for PET-side illumination and 40% for DMD-side illumination. At a higher dose (5·1012 protons/cm2), both illumination configurations showed a ~ 50% decline. These findings contrast with earlier results on CH3NH3PbI3-based photodetectors, which lost ~ 50% of their EQE at only 1012 protons/cm2 of 150 keV irradiation under continuous low-intensity beams41. Importantly, the optical absorption of the perovskite active layer, ITO, and DMD layers remained nearly unchanged at fluences up to 5·1012 protons/cm2 (see Fig. S3b, S4a, b in the Supplementary Information). Similar stability of optical properties has been reported for perovskite, Spiro-OMeTAD, and SnO2 layers under irradiation levels as high as 1013 protons/cm[224,30,42,43. These observations suggest that the decline in the EQE originates not from a reduction in photogeneration, but rather from hindered charge extraction and an increase in recombination processes within the device24,30. The enhanced carrier recombination is further supported by the reduction in photoluminescence (PL) intensity, while the peak position remains unchanged (see Fig. S3c in the Supplementary Information). Since PL intensity reflects the intrinsic quality of perovskite semiconductors44, its suppression implies the formation of deep trap states in the perovskite bulk, which promote non-radiative recombination and diminish radiative emission42,45. Nevertheless, X-ray diffraction (XRD) analysis confirmed that proton exposure did not produce detectable modifications in the crystal structure of the perovskite films (see Fig. S3a in the Supplementary Information). To further assess the impact of irradiation on recombination processes, we analysed the dependence of Voc on light intensity (I) (see Fig. S5 in the Supplementary Information). The slope of Voc plotted against ln(I) increases when the devices are exposed to proton irradiation, indicating a shift toward stronger trap-assisted Shockley–Read–Hall (SRH) recombination in the bulk of the active layer46–48. This trend is consistent with the formation of shallow radiation-induced defect states within the perovskite, which enhance the density of recombination-active traps47,49. Additionally, interfacial defects at the perovskite/electrode interfaces may constitute a contribution to performance degradation. These traps can modify local band bending, hinder charge extraction, and manifest as shifts in the dark current–voltage (J–V) characteristics50,51 (see Fig. S6a in the Supplementary Information). Such shifts are primarily caused by dynamic phenomena, including ion migration and charge trapping/de-trapping at defect sites and interfaces, which alter the internal electric field and recombination dynamics, resulting in observable changes in curve shape and position50,51. An additional contributing factor for the reduction in EQE is the increase in series resistance (Rs), as shown in Fig. S6b of the Supplementary Information. This originates from proton-induced damage to the transparent ITO and DMD electrodes, which raises their sheet resistance (see Fig. S4c, d in the Supplementary Information). The degraded conductivity of these contacts diminishes the short-circuit current and further lowers the EQE49,52,53. Another critical parameter for evaluating photodetector performance is the spectral responsivity (R), which indicates the photocurrent generated per unit of incident monochromatic optical power. It can be calculated directly from the EQE spectrum54,55: . Figure 2b presents the responsivity of the studied perovskite photodetectors. For reference samples, the maximum responsivity reached 0.39 A/W for illumination from the PET-side and 0.30 A/W for illumination from the DMD-side, both peaking at a wavelength of 730 nm. Responsivity peak is consistent with reports, where metal-based electrodes are evaporated on top of the device, which also show responsivity maxima around 720–740 nm30,56–59 . Proton irradiation led to a noticeable reduction in responsivity. At the lower fluence (1012 protons/cm2), R values dropped to 0.26 A/W (PET-side) and 0.18 A/W (DMD-side). Under the higher dose (5·1012 protons/cm2), the responsivity further declined to 0.20 A/W and 0.16 A/W for PET- and DMD-side illumination, respectively. This degradation trend is consistent with the reductions observed in EQE. The dark current is another key factor in assessing photodetector performance, since lower leakage currents enable the detection of weak optical signals60. Figure S6a presents the dark J–V characteristics plotted on a semi-logarithmic scale. After proton irradiation, the dark J–V curves exhibit a shift of the minimum current point toward positive bias along with a reduced dark current, which can be attributed to irradiation-induced defect formation and ion redistribution. These effects modify the built-in electric field and interfacial energetics while introducing carrier trapping and higher injection barriers, thereby suppressing leakage conduction. From the J–V curves, the shunt resistance (Rsh) and Rs were extracted using the voltage-dependent differential resistance method61,62, as shown in Fig. S6b of the Supplementary Information. Interestingly, proton irradiation led to an increase in Rsh at both low and high fluence levels, which in turn enhances the suppression of leakage currents and slightly benefits device performance. In contrast, the Rs remained largely unaffected at the lower irradiation dose but rose noticeably after exposure to the higher proton fluence. This increase suggests irradiation-induced deterioration of the charge transport layers and/or electrode contacts. Detectivity is another essential figure of merit, as it defines the ability of a photodiode to sense weak optical signals. The theoretical upper limit of detectivity for a self-powered device is given by the Johnson noise-limited detectivity (Dth*), which can be expressed as30,60: , where A represents the active area of the photodetector, is thermal noise or Johnson noise, kB is the Boltzmann constant, T is the absolute temperature, and Rsh corresponds to the shunt resistance. Figure 2c depicts the spectral profiles of Dth*. The reference photodetectors demonstrated strong sensitivity across a broad spectral window. For PET-side illumination, the response extended in the range 350–750 nm, with a peak detectivity of 5.9·1011 Jones at 730 nm. Under DMD-side illumination, the active range spanned 420–750 nm, with a maximum of 4.6·1011 Jones at the same wavelength. Low-dose proton irradiation did not lead to significant changes in detectivity for both illumination conditions. However, under the higher irradiation fluence, Dth* declined markedly, consistent with the degradation of responsivity, thereby reducing the device’s ability to detect weak optical signals. The noise spectral density is another one of the most important experimentally determined noise parameters in photodetectors, as it reflects the total current contributions from all underlying noise sources. It was evaluated using the following expression30,60: where RL=50 Ω corresponds to the input impedance of the oscilloscope, Pout(f) is the difference between the measured power spectrum of the device and the background noise of the setup, BW denotes the measurement bandwidth, and G is the transimpedance gain of the amplifier (in A/V). The measured noise of the photodetectors was corrected by subtracting the noise of the preamplifier. The resulting noise spectral density values are displayed in Fig. 3. At 100 Hz, the reference devices exhibited a noise level of 5.4·10− 9 mA/Hz1/2. Following low-fluence irradiation, the noise rose slightly to 5.5·10− 9 mA/Hz1/2. After high-dose exposure, the noise increased substantially, reaching 8.3·10− 8 mA/Hz1/2, which can be attributed to defect generation within the functional layers of the device. However, it is worth noting that, apart from the pronounced increase observed at certain frequencies under high-dose irradiation, the noise levels across most of the spectrum remain largely unchanged. Noise spectral density of the flexible bifacial perovskite photodetectors. Figure 4 summarises the specific detectivity55,60 () as a function of frequency and wavelength for both PET- and DMD-side illumination. The calculation of D*, based on the measured inoise at 100 Hz and incorporating the combined influence of all noise mechanisms63, is shown in Fig. S7 of the Supplementary Information. Both Fig. 4 and Fig. S7 demonstrate that detectivity decreases consistently across all measured frequencies and wavelengths after irradiation, with stronger degradation observed at the higher proton fluence. Specific detectivity as a function of frequency and wavelength for both (a, c, e) PET- and (b, d,f) DMD-side illumination: (a, b) reference; (c, d) irradiated 1012 protons/cm2; (e, f) irradiated 5·1012 protons/cm2. The noise equivalent power (NEP) represents the minimum optical power required for a photodiode to achieve a signal-to-noise ratio (S/N) of unity within a 1 Hz bandwidth. It effectively sets the lower detection limit of the device and is expressed as65: , where inoise is the experimentally measured noise current spectral density under self-powered operation (see Fig. 3), and R is the spectral responsivity. Figure 5 provides contour plots of NEP as a function of both wavelength and noise frequency. For unirradiated devices, NEP values remained below 10− 11 W/Hz1/2 for both PET- and DMD-side illumination, demonstrating excellent sensitivity to weak light signals. After exposure to a low proton fluence, NEP values showed only a modest increase. However, at the higher fluence, NEP degraded significantly, reaching approximately 10− 10 W/Hz1/2, reflecting the impaired noise performance and responsivity of the photodiodes. NEP as a function of frequency and wavelength for both (a, c, e) PET- and (b, d, f) DMD-side illumination: (a, b) reference; (c, d) irradiated 1012 protons/cm2; (e, f) irradiated 5·1012 protons/cm2. In addition to steady-state measurements, we examined the transient response of the perovskite photodetectors, focusing on the rise and fall times, which are important indicators of photodetector speed. These correspond to the intervals required for the photocurrent to increase from 10% to 90% of its maximum value and to decrease from 90% to 10%, respectively65 (see Fig. S8 in the Supplementary Information). Prior to irradiation, the rise and fall times were determined to be 8.4 s and 0.2 ms for the PET-side illumination and 9.9 s and 0.3 ms for the DMD-side illumination. The relatively long rise times observed in our devices are larger than those typically reported for spin-coated perovskite photodetectors, which often exhibit millisecond- to sub-second-scale response times30,66,67. In our case, the flexible, slot-die-coated architecture introduces a higher density of bulk and/or interfacial trap states compared with spin-coated devices, which leads to slow carrier de-trapping and a gradual establishment of steady-state photocurrent. Following proton exposure, both parameters exhibited a modest increase. This behaviour can be attributed to shallow defect states additionally generated within the perovskite layer and at the electrode interfaces by proton bombardment9,68. Such defect states are not strongly involved in recombination via the Shockley–Read–Hall (SRH) mechanism but instead act primarily as charge-trapping sites, thereby reducing carrier mobility and increasing the rise and fall times. A comprehensive overview of the changes in key photodetector parameters before and after irradiation for both the PET- and DMD-sides of illumination is provided in Table 1. For comparison, Table S1 in the Supplementary Information summarises the responsivity and detectivity of the photodetectors based on perovskite and compound semiconductor before and after proton irradiation. Table 1Summarised characteristics of perovskite photodetectors under different conditions.Illumination sideRmax, A/WDth*max ·1011, Jonesinoise ·10− 9 at 100 Hz, mA/Hz1/2trise, stfall, msReferencePET0.395.95.58.40.2DMD0.264.69.90.31012 proton/cm2PET0.35.95.410.30.7DMD0.184.212.10.55·1012 proton/cm2PET0.24.58321.90.6DMD0.163.612.50.6 Summarised characteristics of perovskite photodetectors under different conditions. Beyond the intrinsic radiation tolerance demonstrated in this work, several approaches may further improve the stability of flexible perovskite photodetectors under proton irradiation. A robust encapsulation or thin oxide protective layer can not only protect devices from proton irradiation but also limit the ingress of moisture, which tends to accelerate radiation-induced chemical degradation69,70. Employing radiation-tolerant electrode materials, such as conductive oxides or carbon-based contacts, may help mitigate interface damage that contributes to carrier recombination29,71. Additionally, passivation layers can reduce the presence of mobile or reactive ionic species that amplify defect formation under energetic particle exposure70. Integrating these strategies with scalable fabrication routes offers a promising pathway toward more radiation-resilient perovskite optoelectronic devices. Conclusion This study presented a detailed investigation of flexible, slot-die-coated bifacial perovskite photodetectors subjected to short-pulsed proton irradiation (150 ns, 170 keV), a condition of critical relevance for applications in space exploration, aerospace instrumentation, and nuclear environments. The integration of structural, optical, and electrical characterisation with controlled irradiation experiments allowed us to disentangle the main mechanisms responsible for device degradation, while also highlighting the resilience of the slot-die-coated perovskite absorber layer and the scalability of the fabrication approach. The fundamental optical absorption and crystalline structure of the perovskite films remained intact even at fluences up to 5·1012 protons/cm2. This observation indicates that irradiation did not cause extensive phase decomposition or lattice collapse, but rather introduced trap states and electrode-related damage that hindered charge transport and extraction. As a result, the degradation in performance was primarily reflected in the optoelectronic response. Specifically, the external quantum efficiency decreased by ~ 30% (PET-side illumination) and ~ 40% (DMD-side illumination) at a fluence of 1·1012 protons/cm² and by nearly 50% (PET- and DMD-side illuminations) at 5·1012 protons/cm2. Responsivity followed the same trend, declining from initial values of 0.39 A/W (PET-side) and 0.30 A/W (DMD-side) to 0.20 A/W and 0.16 A/W after high-dose irradiation. Likewise, peak detectivity dropped from 5.9·1011 Jones (PET-side) and 4.6·1011 Jones (DMD-side) to 4.5·1011 Jones and 3.6·1011 Jones after high-dose irradiation, driven by the effects of reduced responsivity. These consistent patterns across all figures of merit indicate the dominant role of radiation-induced trap formation and electrode conductivity losses in limiting device performance. Despite these changes, the photodetectors retained low dark current levels and effective leakage suppression throughout the irradiation campaign. The increase in shunt resistance after exposure mitigated some of the negative effects of trap-assisted recombination, preserving signal-to-noise characteristics and extending operational viability even under degraded responsivity. This balance between vulnerability and resilience highlights the potential of bifacial perovskite photodetectors to remain functional under harsh radiation conditions, provided that targeted strategies are implemented to reinforce the most susceptible components. These insights pave the way for the design of next-generation perovskite photodetectors capable of reliable operation in radiation-intense environments, bridging the gap between laboratory-scale demonstrations and real-world deployment in advanced aerospace and high-energy physics systems. Experimental part Materials All the chemicals were used as received without any further purification. PET/ITO substrates (thickness: 125 µm, sheet resistance: 45 Ω/sq) were obtained from Mekoprint. Tin(IV) oxide colloidal nanoparticle solution (SnO2, 15 wt.% in H2O colloidal dispersion) was purchased from Alfa Aesar. Lead iodide (PbI2, 99%), lead bromide (PbBr2, 99.999%), bis(trifluoromethane)sulfonimide lithium salt (Li-TFSI, 99%), 4-tert-Butylpyridine (4-tBP, 98%), cesium iodide (CsI, 99.999%), chlorobenzene (CB, 99.80%), acetone (99.80%), and dimethyl sulfoxide (DMSO) were obtained from Merck. Methylammonium iodide (MAI, 99.995%), Formamidinium iodide (FAI, > 99.99%), and Methylammonium chloride (MACl > 99.99%) were obtained from GreatCell Energy. 2-Propanol (IPA, 99.80%) was obtained from a local supplier. 2,2′,7,7′-Tetrakis-(N,N-di-4-methoxyphenylamino)-9,9′-spirobifluorene (Spiro-MeOTAD, 99.50%) was obtained from Lumtec. Gold (Au, 99.99%) and molybdenum oxide (MoO3, 99.95%) were obtained from Kurt J. Lesker. Device fabrication To fabricate flexible bifacial perovskite photodetectors, PET/ITO substrates were sequentially cleaned in an ultrasonic bath for 10 min each in deionized (DI) water with detergent, DI water, acetone, and isopropanol (IPA), then dried using a nitrogen gun. The cleaned substrates were subsequently treated with UV-ozone for 15 min to enhance surface wettability. Slot-die-coater (VectorSC, FOM technologies, Denmark) was used to deposit all the functional layers on clean substrates. The tin oxide ETL was prepared from a commercial colloidal SnO2 dispersion solution, which was diluted with DI water at a volume ratio of 1:3. The SnO2- layer was printed on cleaned PET/ITO substrates at 150 µL/min pump rate and 80 cm/min chuck speed at temperature 50 °C, followed by annealing on a hot plate at 130 °C for 30 min. After annealing, the substrates underwent an additional UV-ozone treatment for 30 min to further improve film quality and interfacial properties. The hybrid triple-cation perovskite ink, 1.4 M Cs0.175FA0.75MA0.075Pb(I0.88Br0.12)3 with 5 mol% MACl in 1 ml DMSO was prepared according to the procedure described by S.S. Sangale et al. 74. The perovskite layer was coated at a pump rate of 70 µL/min and a chuck speed of 70 cm/min at a temperature of 130 °C. The films were then annealed on a hot plate at 100 °C for 10 min and allowed to cool to room temperature. The ink of the hole transport layer was prepared by stirring 45 mg of Spiro-OMeTAD, 28.8 µL of 4-tBp, and 17.5 µL of Li-TFSI solution (520 mg/mL in ACN) in 1 mL of CB for an hour. The Spiro-OMeTAD was coated at a pump rate of 150 µL/min and a chuck speed of 80 cm/min at a temperature of 25 °C. All three device functional layers were printed in air at a temperature and relative humidity of 18–21 °C and ∼27–33%, respectively. Finally, the MoO3/Au/MoO3 (dielectric-metal-dielectric, DMD) top contact was deposited via thermal evaporation (Angstrom Engineering Inc., Kitchener, Canada) under a base pressure of 2 ·10− 6 Torr. Deposition rates were set to 0.5 Å/s for MoO3 and 0.1 Å/s for Au. The optimal thicknesses for the MoO3/Au/MoO3 stack were 10/10/35 nm, respectively. Irradiation The irradiation experiments were performed at the INURA high-current pulsed proton accelerator located at Nazarbayev University31. INURA operates with a high-voltage pulse generator that delivers ~ 100 ns pulses to the accelerator diode, with a voltage rise rate exceeding 1012 V/s. During exposure, the photodetector samples were mounted inside a vacuum chamber on a rotary stage, positioned 35 cm from the diode, with the chamber evacuated to ~ 5·10–2 mTorr. All irradiations were carried out in a single vacuum cycle to prevent sample disturbance between exposures. The proton beam was incident from the MoO3/Au/MoO3 electrode side and directed toward the PET substrate. Beam parameters were monitored using a collimated Faraday cup equipped with a magnetic cut-off, and their stability was verified by applying at least ten test pulses prior to irradiating the devices. The beam spot size was approximately 5 cm in diameter, defined by the shape of the ion diode, while our samples measured 2.5 × 2.5 cm2, fitting well within the beam. Additionally, the beam current density distribution was monitored using a Faraday Cup with a magnetic cut-off, which was moved across the beam to confirm a uniform irradiation area. Materials characterisations and device measurements An ultraviolet–visible spectrometer (UV–Vis, Lambda 1050, PerkinElmer, USA) was employed to analyse the optical properties of thin films. X-ray diffraction (XRD, Rigaku Smartlab, Japan) analysis was used to study the structural properties of perovskite layers. The photoluminescence (PL) spectra of samples were obtained using a FLS 1000 spectrometer (Edinburgh Instruments, UK). The sheet resistance of electrodes was measured using a four-point system (RM3000, Jandel, UK). A semiconductor device analyser (B1500A, Keysight, USA) coupled with a 3 A + solar simulator (Oriel Sol3A, Newport, USA) providing an AM 1.5G illumination intensity of 100 mW/cm2 was used to measure the J–V characteristics of the devices. An external quantum efficiency (EQE) measurement system (ORIEL IQE 200, Newport, USA) was used to acquire the EQE spectra of the photodetectors. For the noise current measurements, a battery-powered low-noise current preamplifier (SR570, Stanford Research Systems, USA) combined with an oscilloscope (SDS5032X, Siglent, China) operating in Fast Fourier Transform (FFT) mode was employed inside a shielded Faraday cage to minimise external interference. Chronoamperometry measurements were performed using a potentiostat (Autolab PGSTAT302N, Metrohm, Switzerland) to evaluate the photoresponse time of the devices. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements A.N.J.and YY acknowledge financial support from the Science Committee of the Ministry of Science and Higher Education of the Republic of Kazakhstan (grant numbers AP26194151, AP23483937, and BR28712419). A.N.J. acknowledges financial support from the Nazarbayev University Collaborative Research Program (grant number 211123CRP1613). A.N. acknowledges financial support from the Science Committee of the Ministry of Science and Higher Education of the Republic of Kazakhstan (grant number AP19576154) and from the Nazarbayev University Collaborative Research Program (grant number 211123CRP1613). H.P. gratefully acknowledges support from the Polish National Agency for Academic Exchange under the Ulam NAWA Program (grant number BNI/ULM/2024/1/00019). Author contributions Y.Y. and A.K.A. performed experimental work, including fabrication, characterisation, measurements, and testing, with contributions from H.P.P. The proton irradiation was performed by M.K. The interpretation of the results was performed by H.P.P. The manuscript was written by H.P.P., with contributions from the other authors. C.M., A.N., and A.N.J. supervised the project and helped with writing the manuscript. Funding A.N.J. and Y.Y. acknowledge financial support from the Science Committee of the Ministry of Science and Higher Education of the Republic of Kazakhstan (grant numbers AP26194151, AP23483937, and BR28712419). A.N.J. acknowledges financial support from the Nazarbayev University Collaborative Research Program (grant number 211123CRP1613). A.N. acknowledges financial support from the Science Committee of the Ministry of Science and Higher Education of the Republic of Kazakhstan (grant number AP19576154) and from the Nazarbayev University Collaborative Research Program (grant number 211123CRP1613). H.P. gratefully acknowledges support from the Polish National Agency for Academic Exchange under the Ulam NAWA Program (grant number BNI/ULM/2024/1/00019). Data availability The data presented in this study are available on request from the corresponding author. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Hryhorii P. Parkhomenko and Yerassyl Yerlanuly contributed equally to this work. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement The data presented in this study are available on request from the corresponding author.