Introduction

The advent of the mobile Internet era has promoted the emergence of “Internet + mobile healthcare”, which is a state-of-the-art medical system and an innovative technology for monitoring human well-being. Direct X-ray detection has become an increasingly important tool in health monitoring, offering simplified system configuration and high spatial resolution capabilities. However, to meet the demands of real-time health inspection on mobile devices, the performance of the detection material must be comprehensively evaluated in terms of semiconductor properties, wearable electronics, toxicity, and lightweight characteristics.

Recently, a new generation of metal halide semiconductors has gained prominence in the field of radiation detection. It has been reported that metal halide single crystal X-ray detectors exhibit sensitivity over 1000 times greater than commercial α-Se detectors, with 120 times lower detection limits1. However, these metal halides pose inherent challenges: (1) Pb2+ and Bi3+ are extensively utilized in metal halides. Pb2+ is known to be biotoxic and, therefore, incompatible with wearable devices. Bi3+-based metal halides typically exhibit indirect band gaps with large electron and hole effective masses2; (2) the coupling of metal halides to flexible substrates is not always perfect due to the uncontrolled coordination between the uncoordinated bonds and the substrate component. Therefore, none of these materials can be ideally combined to achieve uniformly beneficial characteristics for fabricating compatible X-ray detectors3.

The exploration of metal-free perovskite (MFPs) alternatives is a burgeoning field. These materials possess an ABX3 structure, constructed by an octahedral network of organic molecules and NH4+ (or N2H5+)-halide/pseudohalide, primarily interconnected through hydrogen bonding4,5,6,7,8. The diverse designs for organic molecular enable MFPs to exhibit semiconductor properties comparable to those of state-of-the-art metal-based perovskites (for example, the trap density of DABCO–NH4Br3 crystals is as low as 2.6 × 1010 cm−3, and the hole mobility and electron are 2.00 and 0.67 cm2 V−1 s−1, respectively. DABCO = N-N-diazabicyclo[2.2.2]octonium)9. Meanwhile, replacing all the metal cations with molecules significantly reduces the challenges associated with its preparation and processing, rendering it more suitable for lightweight, flexible, cost-effective, and other related material requirements10,11,12.

In this framework, the MFPs have demonstrated remarkable performance in X-ray detection. Zhao et al. prepared high-quality DABCO-NH4I3 single crystals that yield a sensitivity of up to 1997 μC Gyair−1 cm−2 13. Li et al. introduced polar units into MFPs to increase hydrogen bond formation sites, and synergistically achieved ion migration inhibition. A flexible degradable X-ray detector with a sensitivity of 740.8 μC Gyair−1 cm−2 was successfully prepared6. Apparently, these accomplishments are insufficient, and it is imperative to conduct further research on the potential of MFPs materials. From the perspective of component structure, the participation of halides may be the main reason for limiting the amplification of its performance. It is considered that halogen elements (Cl, Br, and I) are the primary instigators of ion migration in perovskite (Fig. 1a). This detrimental phenomenon not only alters the local chemical composition but also profoundly influences their electrical properties and even the operational mechanism of corresponding devices14. To address this issue comprehensively, it is advisable to substitute halogen elements with pseudohalides (such as BF4−, PF6−, and ClO4−) to form pseudohalide–MFPs15,16,17. Firstly, the migration of pseudohalides ions involves a transition in motion mode, such as anion rotation, which significantly elevates the ion migration energy. Additionally, substitution with pseudohalides has been demonstrated to result in a significant enhancement of intermolecular hydrogen bonding18,19.

Fig. 1: Pseudohalide–MFPs structure calculation.
figure 1

a Schematic of MFPs ion migration and defect formation. b MFPs bandgap statistics, where organic cations represent the corresponding MFPs, for example: ODABCO denotes (ODABCO)-NH4X3 (X = Cl, Br and I). c Design of organic cations in pseudohalide–MFPs (top). Electrostatic potential of organic cations and tolerance factors of corresponding structures (bottom). d Diagram of the crystal structures for the three pseudohalide–MFPs. e The calculated band gap. f Isosurface map of (MDABCO)–NH4(BF4)3 and (MDABCOBr)-NH4(BF4)3. g Electrostatic potential of pseudohalide–MFPs, and the Mulliken charges of H atoms participating in hydrogen bonds. qA-H represents the H charge of the A-site cation, qN-H represents the H charge of the NH4+. h DFT calculated energy profile along the ionic migration path (top). Schematic of simulated BF4− migration (bottom).

Full size image

This substitution, however, falls short of the requirements for X-ray detection applications. This is because: (1) the component lacks heavy elements that can effectively cut off X-rays (as X-ray absorption exponentially increases with the atomic number), resulting in poor detection sensitivity. (2) Pseudohalide–MFPs often possess ultra-wide band gap (Eg > 6 eV) (Fig. 1b), according to Devanathan et al. empirical model, which determines that the electron–hole pair production energy is large at X-ray excitation. The larger electron–hole pair creation energy will lead to fewer charge carriers upon X-ray excitation under the same condition20,21,22. (3) The biotoxicity of pseudohalide–MFPs has not been thoroughly and reliably assessed previously, leading to uncertainty in the usage and manufacturing process of wearable X-ray detectors. Therefore, it is crucial to promote the X-ray absorption of pseudohalide–MFPs crystals and further refine their semiconductor properties in order to advance novel radiation detection materials. Meanwhile, confirming it’s non-toxic and environmentally friendly is also particularly important for further wearable X-ray detectors.

Herein, we carefully addressed aforesaid challenge by constructing heavy atom covalent bonds (C–Br/Cl) into pseudohalide–MFPs ((MDABCOCl/Br–NH4(BF4)3, MDABCO = N-methyl-N’-diazabicyclo[2.2.2]octonium) to significantly increase their X-ray absorption cross-section, crystalline density, and synergistically realizes the band nature modulation, ionic migration inhibition. Theoretical calculations demonstrate that the incorporation of heavy atoms additionally enhances hydrogen bonding in the pseudohalide–MFPs system. Interestingly, this also increases crystal stiffness, rendering it more radiation-resistant. On this basis, the single crystal (MDABCOBr)–NH4(BF4)3 device yields a record sensitivity of up to 2377 μC Gyair−1 cm−2, while simultaneously exhibiting remarkable radiation stability. In addition, we have successfully validated the biotoxic safety of (MDABCOBr)–NH4(BF4)3 at the cellular level and subsequently engineered a wearable and flexible X-ray detector that maintains its performance even under repeated mechanical bending.

Results and discussion

Structure and properties of pseudohalide–MFPs

Here, we incorporated the heavy elements (Cl, Br) into MDABCO2+ and investigated the band nature evolution and radiation detection performance of emerging pseudohalide–MFPs (MDABCO–Cl/Br)–NH4(BF4)3 due to the MDABCO–NH4(BF4)3 system is easy to manipulate at the molecular level (Fig. 1c). The electrostatic potential mapping reflects the organic molecule’s characteristics post-design, wherein halogenation induces a redistribution of electron density, resulting in a positive charge on the remaining part of the molecule and a negative charge on the halogen side. Correspondingly, the structural evolution information is evaluated by the tolerance factor τ, which ensures that the crystal structure is stable with the appropriate dimensions (Fig. 1c)23,24.

The (MDABCO–Cl/Br)–NH4(BF4)3 single crystals were prepared at room temperature (~25 °C). Transparent crystals can be obtained by carefully controlling the evaporation of solvents from aqueous solutions containing HBF4 (Supplementary Fig. 1). The scanning electron microscope (SEM) analyses confirmed the presence of flat crystal cutoff surface (Supplementary Fig. 2). Firstly, the crystal structure information is revealed by single-crystal X-ray diffraction (XRD). Crystal structure analyses disclose that the space group P21 with monoclinic system changes to P1 with triclinic after inserting the C–Cl/Br covalent bond. The density also gradually increases (from 1.728 g cm−3 for (MDABCO)–NH4(BF4)3 to 1.950 g cm−3 for (MDABCOBr)–NH4(BF4)3). In detail, the three pseudohalide–MFPs all exhibit three-dimensional structures. (MDABCO–Cl/Br)2+ cation is centrally located within a cage-like NH4(BF4)6 polyhedra structure. The hydrogen bond between C–H···F, C–H/Cl···F and C–H/Br…F is responsible for the stabilization of the (MDABCO)–NH4(BF4)3, (MDABCOCl)–NH4(BF4)3, and (MDABCOBr)–NH4(BF4)3 components, respectively (Fig. 1d and Supplementary Fig. 3–5)25,26.

The properties of pseudohalide–MFPs semiconductors are intricately linked to the band gap; therefore, we employ density functional theory (DFT) to compute the band structure. As shown in Fig. 1e, the band gap of (MDABCO)–NH4(BF4)3 is 7.4 eV, which closely aligns with the reported for other analogous compounds18. Interestingly, there was a reduction in the band gap to 6.6 eV for (MDABCOCl)–NH4(BF4)3 and 5.5 eV for (MDABCOBr)–NH4(BF4)3, respectively, after heavy atom modification. The reported pseudohalide–MFPs band gap presented here is believed to be the smallest known to date. The dominance of the change in the band gap is undoubtedly attributed to the fine-tuning of the structure. The Hirshfeld partition of molecular density (IGMH) is employed herein to analyze diverse chemical bonds and interactions in pseudohalide–MFPs27. Fig. 1f and Supplementary Fig. 6 illustrate the presence of covalent bonds (C–C, B–F), hydrogen bonds (C–H…F), and steric hindrance effect within spherical molecules in pseudohalide–MFPs. Meanwhile, the presence of highly electronegative Br and Cl atoms leads to a more refined van der Waals intermolecular interaction in (MDABCOBr)–NH4(BF4)3 and (MDABCOCl)–NH4(BF4)3.

The interactions between these components are expected to influence the structure coupling. The heightened electrostatic potential difference between (MDABCO–Cl/Br) and NH4(BF4)6 enhances the strength of hydrogen bonding compared to MDABCO2+, and the additional van der Waals forces acting upon also facilitate the coupling between (MDABCO–Cl/Br) and NH4(BF4)6 pseudohalide–MFPs. It can also be intuitively seen from the Mulliken charge population diagram that MDABCOBr2+ and MDABCOCl2+ have higher inductive effects on anion groups, especially the former (Fig. 1g)28. The charge density distribution of pseudohalide–MFPs in Supplementary Figs. 7–9 provides further evidence, revealing a higher occurrence of charge deletions in the conduction band minimum (CBM) and valence band maximum (VBM) for (MDABCO)–NH4(BF4)3 compared to (MDABCOCl)–NH4(BF4)3 and (MDABCOBr)–NH4(BF4)3 crystals, suggesting its stronger electronegativity (yellow indicates charge deletions while blue represents charge aggregation).

Considering these, the band gap reduction in (MDABCO–Cl/Br)–NH4(BF4)3 can be easily comprehended. Specifically, the intermolecular interactions in pseudohalide–MFPs dominate the frontier orbital separation of organic molecules, as well as orbital coupling with NH4(BF4)6 octahedra29. As shown in Supplementary Fig. 10, the pseudohalide–MFPs conduction band (CB) is mainly composed of organic molecules, including C-s, N-p, and H-s orbital hybrids. By contrast, the Cl-s orbital is observed in (MDABCOCl)–NH4(BF4)3, and there is a clear overlap with the N-p orbital, which tailed CBM to a lower energy state10. Based on the aforementioned analysis, it can be attributed to the electron delocalization resulting from the significant electronegativity difference in DABCOCl2+ and the stronger interaction within the (MDABCOCl)–NH4(BF4)3. Interestingly, the CB of (MDABCOBr)–NH4(BF4)3 is composed of two main parts, and the situation of the latter part is similar to that of (MDABCOCl)–NH4(BF4)3. In contrast, the former part emphasizes the hybridization between C-s and F-p orbitals, thereby elucidating that the modulation of band gap primarily arises from intricate intermolecular interactions.

Ion migration driven by the high electric field in radiation detection can significantly deteriorate electrical properties. This situation is particularly exacerbated in halide–MFPs, primarily due to the considerably lower ion migration barrier18. However, the migration process of BF4− reveals that including a rotational process in its movement dramatically increases the ion migration barrier in pseudohalide–MFPs (Fig. 1h). The results also show that structural adjustment can effectively improve the ion migration barrier of pseudohalide–MFPs. Defect formation in pseudohalide–MFPs was further carried out. The DFT calculation reveals that H atom on the –CH3 in MDABCO2+ are readily formed point defects VH at CB and VB, as well as charge defects VH+1 at VB (formation energy ~2.7 eV). Once the H atom is replaced by Br/Cl, it leads to an increase in the formation energy of VBr and VCl, reaching 3 eV. Additionally, the charge defect VBr−1 and VCl−1 experience a rise up to 6 eV due to the aforementioned intermolecular van der Waals interaction (Fig. 1f). Unlike halide–MFPs, X-site defects in pseudohalide–MFPs are difficult to form because they involve overcoming both covalent B–F bonds and intermolecular hydrogen bonds. The simulation results indicate that the formation energy (VB and VB+) of all three compounds >10 eV, particularly for (MDABCOCl)–NH4(BF4)3 and (MDABCOBr)–NH4(BF4)3, rendering their formation nearly impossible. Still, this increasing trend also signifies a gradual reinforcement of hydrogen bonding (Fig. 2a).

Fig. 2: Defect, binding energy and hardness testing of pseudohalide–MFPs.
figure 2

a Defect formation energy of pseudohalide–MFPs. b Visualization map of the distribution of the interactions and the 2D fingerprint plots of (MDABCOBr)-NH4(BF4)3. c Calculated binding energies between MDABCOBr2+-NH4(BF4)6, MDABCOCl2+-NH4(BF4)6, and MDABCO2+-NH4(BF4)6. d The Coulomb interactions of pseudohalide–MFPs. e 1H NMR spectra of pseudohalide–MFPs. f Load-force-dependent indentation depth curve of different pseudohalide–MFPs. g Hardness and Modulus comparison of different pseudohalide–MFPs.

Full size image

The subsequent investigation examines the impact of structural modulation interaction on the stability of pseudohalide–MFPs. We present a visualization of the distribution of interactions using Hirshfeld surface analysis30,31. The prevalence of blue and white regions in all three pseudohalide–MFPs suggests that weak interactions between cations and anions dominate the system (Fig. 2b, Supplementary Figure 11). The map simultaneously illustrates that (MDABCOCl)–NH4(BF4)3 and (MDABCOBr)–NH4(BF4)3 exhibit a larger surface distribution area, attributable to their more interaction. The ultimate result is that the components in (MDABCOCl)–NH4(BF4)3 and (MDABCOBr)–NH4(BF4)3 exhibit heightened binding energy (lower binding energy means easier dissociation), and the coulomb forces between the anions-cations, thereby attaining optimal stability under same conditions, especially the (MDABCOBr)–NH4(BF4)3 (Fig. 2c, d). Also, the observation of H(-NH4+) signals toward lower chemical shifts in the 1H NMR spectra for (MDABCOBr)–NH4(BF4)3 also serves as evidence for enhanced hydrogen bond strength associated with binding energy (Fig. 2e). As shown in Supplementary Fig. 12, the decomposition temperature measured by thermos-gravimetric analysis also confirms this trend. The thermal decomposition of (MDABCOBr)–NH4(BF4)3 mainly occurred at ~220 °C, higher than that of the other two analogs.

Crystal stability is also intricately linked to the mechanical indices related to lattice stiffness. The mechanical properties of the (001) plane were evaluated through a nanoindentation measurement. The load-force-dependent indentation depth curve of pseudohalide–MFPs crystal in Fig. 2f provides elastic and plastic deformation information of the sample under the nanotip. It indicates that the indentation depth varies across different single crystals when subjected to identical loads32,33. The indentation hardness and Young’s modulus of the single crystals are presented in Fig. 2g. The order of (MDABCO)–NH4(BF4)3 < (MDABCOCl)–NH4(BF4)3 < (MDABCOBr)–NH4(BF4)3 can be attributed to stronger hydrogen bonds and other interactions in (MDABCOBr)–NH4(BF4)3. Further DFT calculations confirmed the mechanical properties, including hardness, Young’s modulus, Poisson’s ratio, and shear modulus. The data extracted using Elastic-post software are listed in Supplementary Table 134. The representative three-dimensional and two-dimensional plots are shown in Supplementary Figs. 13 and 14, and it can be seen that the trend of contour lines is consistent with the experimental measurement.

X-ray detection and imaging

Then, the electrical and X-ray detection properties of the pseudohalide–MFPs single crystals were evaluated. Due to the incorporation of large electronegative elements, the dielectric constant (εr) of (MDABCOCl)–NH4(BF4)3 and (MDABCOBr)-NH4(BF4)3 surpasses that of (MDABCO)-NH4(BF4)3 (Supplementary Figure 15). A larger εr implies a weakened shielding effect on charges, facilitating their separation35. As shown in Supplementary Figure 16a, we calculated the X-ray absorption coefficients of pseudohalide–MFPs and commercial Si semiconductors at photon energies ranging from 1 keV to 10 MeV. The introduction of the Br element in (MDABCOBr)–NH4(BF4)3 significantly increases its absorption coefficient, particularly surpassing that of Si within the energy range of 15–200 keV. Supplementary Figure 16b displays the thickness of the MFPs single crystals, enabling it to absorb 100% of X-ray energy at 20 keV. On this basis, an Au/pseudohalide–MFPs crystal/Au planar X-ray detector is constructed. Fig. 3a shows the current–voltage (I–V) curve for the (MDABCOBr)–NH4(BF4)3 device at different dose rates. The charge collection efficiency is related to the carrier mobility µ and the carrier lifetime τ. High µτ products are preferred for outstanding-performance electronics. We derive the µτ product by fitting the dark conductivity shown in Supplementary Fig. 17 using a modified Hecht equation36. The resulting µτ value were 4.25 × 10−5 cm2V−1 for (MDABCOBr)–NH4(BF4)3, 8.36 × 10−6 cm2V−1 for (MDABCOCl)–NH4(BF4)3 and 1.43 × 10−6 cm2V−1 for (MDABCO)–NH4(BF4)3, respectively. A large µτ product typically results in a large device SNR, allowing the detector to resolve low dose-rate X-rays.

Fig. 3: X-ray detector performance of pseudohalide–MFPs single crystals.
figure 3

a The irradiated J–V curves of based (MDABCOBr)-NH4(BF4)3 X-ray detector under different dose rates. b Time response of (MDABCOBr)-NH4(BF4)3 and (MDABCO)–NH4(BF4)3 devices. c Noise current spectra of three pseudohalide–MFPs devices. d Photocurrent stability of (MDABCOBr)–NH4(BF4)3 device under different dose rates. e X-ray response of current density of (MDABCOBr)–NH4(BF4)3 device with various dose rates at different electric field. f The sensitivity of the (MDABCOBr)–NH4(BF4)3 device at different electric fields is obtained from (e). g The dose rate dependent SNR of the different pseudohalide–MFPs devices. h Dark current drift of (MDABCOBr)–NH4(BF4)3 device under the electric field of 200 V mm−1. i On/off and long-time operation stability of the (MDABCOBr)-NH4(BF4)3 device. j The X-ray image of (MDABCOBr)–NH4(BF4)3 device.

Full size image

The response time of the X-ray detector plays a crucial role in practical applications as it directly impacts the detection speed. The response time was estimated using the transient current acquisition method. As shown in Fig. 3b and Supplementary Figure 18, the impulse response curves show a rise and fall time (from 10% to 90% of the saturated photocurrent) of 3.01 ms and 5.05 ms, respectively, for the (MDABCOBr)–NH4(BF4)3 device, which is faster than the other two analogs. In addition, the decay time information is obtained by fitting the impulse response curve, and the target device has a minimum decay time of 2.35 ms.The short response time is due to low defect density, fewer grain boundaries, and uniform lattice orientation. This advantage allows the device to obtain a more convenient charge transport path. For X-ray detection applications, noise is also an important figure of merit, which was evaluated with a fast Fourier transform signal analyzer37. Figure 3c demonstrates a significant reduction in noise for (MDABCOBr)–NH4(BF4)3 device compared to both (MDABCO)–NH4(BF4)3 and (MDABCOCl)–NH4(BF4)3. As mentioned above, this indicates an enhancement in lattice stiffness, effectively suppressing structural perturbations during detection.

The evaluation of X-ray detector performance relies heavily on the determination of sensitivity and detection limit. To assess sensitivity, we quantified the photocurrent response in terms of the on/off ratio under various dose rates and applied electric fields (Fig. 3d, Supplementary Figs. 19–21). The generated current density in each determined electric field exhibits a linear correlation with the X-ray dose rate, as shown in Fig. 3e. The (MDABCOBr)–NH4(BF4)3 device yield a sensitivity of 2377 μC Gyair−1 cm−2 at 200 V mm−1, which is 118 times higher than that of the commercial ɑ-Se detector (20 μC Gyair−1 cm−2@10,000 V mm−1) and represents a record value ever reported for MFPs devices (Supplementary Table 2). The sensitivity for (MDABCO)–NH4(BF4)3 and (MDABCOCl)–NH4(BF4)3 devices are 617 μC Gyair−1 cm−2 and 1522 μC Gyair−1 cm−2, respectively (Fig. 3f, Supplementary Figs. 19 and 20). While this is a modest gap to the performance of state-of-the-art metal halide X-ray detectors, our design principles establish that heavy atom design and structural optimization are key to driving the performance of pseudohalide–MFPs. According to the given definition, the detection limit fixed bias is defined as a dose rate that yields a signal-to-noise ratio of 333. Figure 3g illustrates that the lowest detection limit for (MDABCOBr)–NH4(BF4)3 device is 50.1 nGyair s−1, better than the other two pseudohalide–MFPs devices.

The deterioration of device performance caused by ion migration is regarded as a persistent ailment in metal halides. The current hysteresis serves as visual evidence for ion migration. Supplementary Figure 22 presents the dark current cycle curves of (MDABCO)–NH4I3 and (MDABCOBr)–NH4(BF4)3 devices in the −200 V and 200 V ranges. The (MDABCO)–NH4I3 exhibited a more pronounced hysteresis behavior attributed to severe halogen ion migration. Regarding device operation, migrating ions at the electrode-crystal interface should be cautioned, as it can undergo chemical reactions with the electrode, ultimately leading to its gradual failure38. We designed an immersion experiment to compare the corrosion of electrodes in pseudohalide–MFPs and halide–MFPs devices (see details in Supplementary Fig. 23). It can be seen that after immersing in (MDABCO)–NH4I3 dispersion solution for 72 h, the Cu film undergoes a noticeable darkening accompanied by a deteriorated coverage, and nearly disappears after soaking for 120 h, indicating its susceptibility to corrosion within this I-rich environment. Conversely, when immersed in the (MDABCOBr)–NH4(BF4)3 dispersion solution for 120 h, no discernible changes occur for the Cu film, indicating that electrode degradation in the pseudohalide–MFPs system is not related to the active material (Supplementary Fig. 24). This experimental conclusion is consistent with the results of previous simulations (Fig. 1h), which show that the disgusting ion migration in pseudohalide–MFPs is largely inhibited.

Further, we obtained the ion migration energy activation (Ea) of MFPs by fitting the temperature-dependent conductivity curve with the Nernst–Einstein equation39. The results also indicate that the activation energy (Ea) of pseudohalide–MFPs is greater than that of halide–MFPs devices, which aligns with the results from the simulation experiments. Meanwhile, the careful structural design of (MDABCOBr)–NH4(BF4)3 devices enables the highest ion migration energy in the pseudohalide–MFPs device (Supplementary Fig. 25 and Fig. 1h).

Then, the operating stability of (MDABCOBr)–NH4(BF4)3 devices is evaluated. As shown in Fig. 3h, the device operates over 4000 s at an electric field of 200 V mm−1 with a dark current drift only 4.0 × 10−8 nA cm−1 s−1 V−1, which is lower than reported value for analog single crystals device ((MDABCO)–NH4(PF6)3, 3.35 × 10−6 nA cm−1 s−1 V−1)18. Impressively, the detector exhibited a stable X-ray on/off response at different dose rates even after 10,000 s, and maintained its photocurrent signal without decay under continuous X-ray irradiation for an extended period of time (Fig. 3i). The stability of (MDABCO)–NH4I3 was compared under the same operating conditions. The device exhibited significant dark current fluctuation during long-term operation, with a drift value of 2.66 × 10−8 nA cm−1 s−1 V−1, which is two orders of magnitude higher than the (MDABCOBr)–NH4(BF4)3 devices. Additionally, the device also showed significant instability under X-ray irradiation, which differs from the target device. It has been concluded that the instabilities are primarily caused by the migration of halogen ions at high pressures, as previously confirmed (Supplementary Fig. 26).

Therefore, (MDABCOBr)–NH4(BF4)3 single crystal detectors are expected for high-performance X-ray imaging owing to their remarkable sensitivity, stability, and low detection limit. The X-ray imaging process involves the movement of the image object using an X–Y shift table positioned between the X-ray source and the single-pixel detector. The acquisition of X-ray images is achieved by capturing the transmitted X-ray photons versus location (x, y)40. Finally, as shown in Fig. 3j, the imaging of the metal radiation logo is clearly displayed.

Biotoxicity and flexibility

Metal halide perovskites are popular in the field of photoelectronics; however, ensuring their biosafety remains challenging due to the presence of heavy metals. For example, Pb2+ can enter the bloodstream through ingestion, breathing, and skin contact, distribute to soft tissues such as lungs, heart, liver, kidneys, and brain, and accumulate in bones. The accumulation of Pb2+ can regulate fundamental physiological functions in the human body, as well as interfere with heme activity, which is essential for oxygen binding in the blood. Moreover, these Pb2+ have the potential to disrupt enzyme and receptor function within soft tissues (Fig. 4a)41,42. Unfortunately, there is currently insufficient evidence to support the biocompatibility of pseudohalide–MFPs. Therefore, it is imperative to conduct a systematic investigation into the biocompatibility of pseudohalide–MFPs. This study investigates the cytocompatibility of (MDABCOBr)–NH4(BF4)3 crystals. The extracts of varying concentrations of this material were placed in 96-well plates (Supplementary Fig. 27). The L929 fibroblasts were selected for the cell viability test43. Fig. 4b, c presents the cells viability data and fluorescence images at different incubation times. It can be seen that the growth ability of L929 cells can be enhanced in the content range of 12.5–100 µg ml−1, which is consistent with the results reported previously (Fig. 4d, e)44,45. The cells viability also remained better with the extension of the incubation time. In contrast, the study conducted by Benmessaoud et al. revealed that even at concentrations as low as 50 µg ml−1, MAPbI3 still triggers a significant level of cell death (from 10% to 30%). Furthermore, it was observed that concentrations up to 200 µg ml−1 could result in the death of over half of the cell population46.

Fig. 4: Biocompatibility in vitro of (MDABCOBr)-NH4(BF4)3.
figure 4

a Illustration of the lead toxification in the human body. b Schematic of replacing Pb2+ with NH4+ as the B-site cation. c Flow chart of CCK8 experiment. d Cell viability of (MDABCOBr)-NH4(BF4)3 over 24 and 48 h calculated as the fraction of total living cells. e (MDABCOBr)-NH4(BF4)3 images of the live/dead staining of planktonic S. aureus with different treatments.

Full size image

The utilization of heavy and bulky equipment can lead to significant discomfort, while prolonged usage exacerbates the psychological and physical burden on the user. Hence, employing lightweight and flexible imaging devices proves to be an effective means of alleviating this load. Furthermore, flexible devices have a greater propensity for bending and conforming to biological surfaces than conventional solid devices. Leveraging the exceptional X-ray detection performance and biocompatibility of (MDABCOBr)–NH4(BF4)3 single crystals, we endeavored to fabricate a wearable X-ray detector (Fig. 5a). Poly(vinylidene fluoride) (PVDF) was selected as a flexible substrate because of its suitable physicochemical properties and mechanical properties. As shown in the Supplementary Fig. 28, we measured the stability of PVDF at high dose rates and high humidity conditions, confirming its excellent resistance to radiation and the environment, which ensures that our equipment will not suffer from matrix failure.

Fig. 5: Mechanical flexible and X-ray imaging evaluation.
figure 5

a Schematic diagram of the flexible X-ray device laminated to the human body (Inset: the flexibility schematic). b Current response and c sensitivity of the device at different bending angles. d Dark current drift of the device before and after 1600 bends. e Sensitivity of the device under different electric fields and bending times. f Gain factor of the device at different dose rate under an electric field of 200 V mm−1. g The operating stability of flexible devices under different doses and long-term switching. h Imaging of flexible devices.

Full size image

The fabricated device has good homogeneity, and the X-ray excitation current is highly consistent in different regions (Supplementary Figs. 29 and 30). As shown in Fig. 5b, c, the corresponding response current, as well as sensitivity, remains almost constant for different radii of curvature, showing excellent device robustness. In addition, we test the flexible device in continuous bending cycles, where the bending radius of each time is defined as 10 mm. Fig. 5d shows that there is only a negligible change (from 7.0 × 10−8 nA cm−1 s−1 V−1 to 8.0 × 10−8 nA cm−1 s−1 V−1) in the dark current drift of the (MDABCOBr)–NH4(BF4)3 device after 1600 bends. As shown in Fig. 5e, the sensitivity remained essentially constant even after 1600 bends at different electric fields.

In addition, the gain factor of the flexible (MDABCOBr)–NH4(BF4)3 device reaches 9.1 at 32.67 uGyair s−1 under an electric field of 200 V mm−1, which indicates that the flexible pseudohalide–MFPs has great potential as a sensitive X-ray detection device. Interestingly, the gain factor gradually decreases to 2.7 as the dose rate increases to 77.44 uGyair s−1, which can be attributed to the progressive filling of charge carriers into shallower traps, leading to smaller gains under higher irradiation intensities (Fig. 5f)2. In particular, the continuous and fast switching current under the electric field of 200 V mm−1 is not attenuated, indicating the device has extraordinary operation stability (Fig. 5g). Simultaneously, we analyzed the storage stability of (MDABCOBr)–NH4(BF4)3 materials and devices in air. Supplementary Figure 31a shows that the material maintained its structure after 10 months in air. The corresponding device is also retested unaffected by the attack of oxygen and water molecules in the air, and the change from the initial performance is almost negligible (Supplementary Fig. 31b, c). The ultra-stability against humidity and oxygen is due to the protection of molecular perovskite octahedra by polar enhanced DABCOBr2+ ions and the enhanced interaction between the various components of the optimized structure. To demonstrate the imaging capability of the flexible pseudohalide–MFPs X-ray detector at a low dose rate, we successfully captured an image of a “gasolene station” logo constructed with a metal base using an X-ray dose rate of 1.2 uGyair s−1 (this process is consistent with that described in Fig. 3f). It is evident that the clarity of the logo and the high signal-to-noise ratio are readily apparent (Fig. 5h). These achievements indicate that flexible pseudohalide–MFPs with high stability, outstanding bending performance, and imaging capability, which are promising in wearable devices for mobile health.

Discussion

We have successfully demonstrated the feasibility of employing structural modulation in pseudohalide–MFPs to achieve enhanced sensitivity, lower detection limit, and prolonged stability for X-ray detection. In detail, the heavy atom Br was designed and introduced. The results showed that not only the stronger interaction, but also the X-ray absorption ability in (MDABCOBr)–NH4(BF4)3 was improved. As a result, the corresponding device achieves a recorded sensitivity of 2377 μC Gyair−1 cm−2. Further, the cytotoxicity test was reported and confirmed the biocompatibility of (MDABCOBr)–NH4(BF4)3 material. In view of the safety and high performance of pseudohalide–MFPs, a flexible wearable X-ray detector was prepared. The device can maintain smooth operation without attenuation under large-angle bending and thousands of bending times. The as-fabricated devices also demonstrated superior X-ray imaging. In summary, this study offers solutions from material design perspective to accommodate the challenges in imaging and detection for rapidly evolving healthcare diagnostics.

Methods

Materials

N-N′-diazabicyclo[2.2.2] octane (DABCO), methyl iodide (CH3I), hydroiodic acid (HI, 45%), Tetrafluoroboric acid (HBF4, 50%), NH4BF4, methyl acetate and polyvinylidene difluoride (PVDF) were purchased from Aladdin Reagent Ltd. All the chemicals were used as received without further purification.

(DABCO)–NH4(BF4)3 Crystal Growth

First, (DABCO)–NH4(BF4)3 polycrystalline powders are prepared by antisolvent precipitation. Typically, NH4BF4 (0.452 g, 4.31 mmol) was dissolved in 2 mL H2O followed by the addition of 0.40 mL H3PO4 (85%) and DABCO (0.30 g, 1.36 mmol) with constant stirring. The prepared powder is filtered and dried, then redissolved in deionized water and subjected to slow evaporation at room temperature in order to obtain the required crystals.

(MDABCO)–NH4(BF4)3 crystal growth

The preparation of MDABCO–NH4(BF4)3 crystals is similar to that of DABCO–NH4(BF4)3, the only difference being that the DABCO is replaced by MDABCO-I.

(MDABCOCl)–NH4(BF4)3 crystal growth

NH4BF4 (4.31 mmol) was dissolved in 2 mL H2O followed by the addition of DABCOCl (2.86 mmol) with constant stirring. The key step is to add a small amount of HBF4 to adjust the PH of the solution and provide BF4−. Finally, the solution is placed at room temperature and slowly evaporates. After about 2 weeks, cubic grains begin to precipitate.

(MDABCOBr)–NH4(BF4)3 crystal growth

The growth method is the same as MDABCOCl–NH4 (BF4)3, except that MDABCOCl is replaced by MDABCOBr.

PVDF @(MDABCOBr)–NH4(BF4)3 composite flexible film preparation

In total, 100 mg dried MDABCOBr–NH4(BF4)3 powder was mixed with 150 mg PVDF (Mw ~400,000) in the mortar. 1.5 mL methyl acetate was added to adequately dissolve the mixture in a glass vial. Then 30 min-ultrasonic treatment was made to facilitate the dispersion of (MDABCOBr)–NH4(BF4)3 particles. Then the mixed suspension was evenly poured onto the Teflon template, the thickness of the composite membrane was controlled to be 1 mm, and placed on a high temperature platform at 120 °C for 30 min. After removing the solvent, the film is easily removed from the template with tweezers. No further any post-deposition treatments except for evaporation of the electrodes.

Device fabrication preparation

For the detectors, a thickness of 100 nm Au electrodes was deposited on the single crystals and flexible film by the vacuum evaporation method. The effective illumination area of the device was 0.04 cm2.

Biocompatibility

We used fluorescent staining to observe L929 cell morphology (Rhodamine Phalloidin, Solarbio), and used cell counting kits (CCK-8; Beyotime, China) to assess the cytotoxicity of pseudohalide–MFP. Different concentrations (12.5, 25, 50, 75, and 100 μg mL−1) of (MDABCOBr)–NH4(BF4)3 were examined on L929 cells, respectively. The effect of cell activity was calculated by measuring the absorbance at 450 nm.

First-principles calculations

In the density functional theory (DFT) calculations, the Vienna Ab initio Simulation Package (VASP) is employed to investigate crystal and electron properties based on the projector augmented wave (PAW) method. The exchange-correlation potential is described by using the Perdew–Burke–Ernzerhof functional under the generalized gradient approximation (GGA). The van der Waals interaction is considered via the zero damping DFT-D3 Grimme method. In the electronic minimizations, the plane-wave cutoff energy of 500 eV and a 4×3×3 k-point grid are used. The high symmetry path is determined by using the automatic program SeeK-path. For other calculations, a higher cutoff energy of 600 eV is used. The convergence criterion of forces acting on each atom is set to 0.03 Ev Å−1 in the full relaxation of the cell volume, shape, and atom positions in the defect formation energy calculations and in obtaining initial and finial structure of that in ion diffusion energy calculations. In the latter calculations, 8 equidistant climb images and a 3 × 3 × 2 k-point grid are used in the climbing image nudged elastic band (CI-NEB) method.

Characterization section

TGA and NMR

TGA spectra were obtained by the thermogravimetric analyzer (PT, 1600). NMR was performed by using AVANCE III 400 with a frequency of 400 MHz, and deuterated DMSO was used as a solvent.

SEM and XRD

The surface morphology was measured by SEM using Hitachi S-4800 scanning electron microscope. XRD patterns were characterized by the Bruker D2 PHASER Diffractometer with the Cu Kα line.

Nanoindentation measurement

The bottom surface of the crystal was attached to the slide with adhesive. Place the fixed sample and slide on the platform with the test surface (001) facing upwards. The slide was fixed by vacuum adsorption and the sample test surface was observed with the microscope provided with the equipment. Indentation experiments were performed on Bruker Hysitron TI 980 with a Berkovich indenter (three-sided pyramid-shaped tip). Five points were tested for each crystal.