Figure 1a and Figure S1 show the X-ray absorption coefficient and thickness-dependent attenuation efficiency of various scintillators. The Cs5Cu3Cl6I2 presents a similar X-ray absorption capability with CsI:Tl and LYSO:Ce, and much higher than Si, indicating its great application potential of Cs5Cu3Cl6I2 in X-ray detection. The grinding method is a facile approach for the preparation of Cs5Cu3Cl6I2.22 However, it is anticipated that the grinding method may introduce numerous anion vacancy defects, leading to a higher prevalence of lattice defects in the samples, consequently introducing unavoidable structural perturbations (lattice strain) in the material. To mitigate the potential adverse effects of lattice strain, we employed a solution-assisted crystallization (SAC) method for material synthesis, and observed a significant enhancement in both luminescent and scintillation performance. The obtained Cs5Cu3Cl6I2 powder was subsequently blended with polydimethylsiloxane (PDMS) to fabricate a flexible polymer film (Figure S2). The scintillator films demonstrated outstanding blue-emitting properties when exposed to X-rays with a photon energy up to 45 keV (Figure S3). The radioluminescence (RL) spectra of the films, based on SAC-Cs5Cu3Cl6I2 under steady-state X-ray irradiation, exhibited broad blue light emission peaks, surpassing those of ground Cs5Cu3Cl6I2 and commercialized LYSO:Ce scintillator (Fig. 1b). To facilitate a comprehensive comparison with reported results in the literature, we adopted the methodology utilized in previous studies, using commercial LYSO as a reference for evaluating the light yield of our Cs5Cu3Cl6I2.36 The light yield of the SAC-Cs5Cu3Cl6I2@PDMS film reached a surprisingly high value of 103,102 ph MeV− 1, which is three times higher than that of LYSO:Ce, twice as high as the ground Cs5Cu3Cl6I2@PDMS film in this study, and surpasses previously reported copper-based scintillators (Table S1).
To investigate the underlying mechanisms of the enhanced scintillation properties, a comprehensive series of characterizations were conducted. As shown in Fig. 1c and Figure S4, the SEM images of Cs5Cu3Cl6I2 synthesized by SAC presented uniform and regular grain with average grain size of 2.55 µm, while the Cs5Cu3Cl6I2 prepared by grinding was irregular and the grain size was larger (~ 3.05 µm). Furthermore, SEM images at a larger scale revealed that SAC-Cs5Cu3Cl6I2 demonstrated a more homogeneous and well-defined micromorphology compared to grinded-Cs5Cu3Cl6I2, which exhibited noticeable agglomeration phenomena with large clusters-an observation absent in SAC-Cs5Cu3Cl6I2.
X-ray diffraction (XRD) is a powerful characterization technique for the information of crystallographic structure and strain of materials. The crystal structure of Cs5Cu3Cl6I2 belongs to orthorhombic space group Cmcm, and consists of 1D zigzag [Cu3Cl6I2]n5n− chains and charge-balancing Cs+ cations.20 The XRD pattern of Cs5Cu3Cl6I2 synthesized by these two methods were consistent well with the simulated data, and no impurity phase was found (Fig. 1d). The intensity enhancement and sharpening of the diffraction peaks demonstrate better crystal quality of the SAC-Cs5Cu3Cl6I2, compared with the grinded one. We further analyzed the internal lattice strain in the two materials. Figure 1e shows the schematic diagram of the relationship between strain and XRD diffraction peak changes. A perfect lattice arrangement will result in an XRD pattern with sharp and strong diffraction peaks. In the presence of strain within the crystal, however, the XRD diffraction peaks will exhibit shifts and broadening.37 A consistent tensile strain induces uniform lattice expansion, causing a shift of the XRD diffraction peak to lower 2θ values. Conversely, uniform compressive stress produces the opposite effect.32 In most materials, strain is nonuniform and complex to simply classify as tensile or compressive. In this case, an indicator of strain is the broadening of XRD peaks in comparison to that of the strain-free material.37,38
The broadening of XRD peaks could reflect the information of lattice strain, which leads to local distortion of lattice planes and different interplanar distance across the material. In principle, the XRD peak broadening is originated from instrument response, grain size and strain in the lattice, which can be mathematically expressed by:38
$$\varDelta {d}_{obs}^{2}=\varDelta {d}_{inst}^{2}+\varDelta {d}_{size}^{2}+\varDelta {d}_{\epsilon }^{2}$$
1
where, Δdobs is the difference in d-spacing at full width at half-maximum (FWHM) for a given (hkl) plane; Δdinst, Δdsize, and Δdε are the instrument, grain size, and lattice strain contributions to the observed peak width, respectively. The grain size of Cs5Cu3Cl6I2 was on the order of micrometer, thus the Δdsize was small enough to be neglected. Therefore, the Eq. (1) can be further evolved as:
$$\varDelta {d}_{\epsilon }=\sqrt{\varDelta {d}_{obs}^{2}-\varDelta {d}_{inst}^{2}}=\epsilon d$$
2
where ε is the strain in the lattice and d is the interplanar spacing for different (hkl) planes. As usual, the contribution of instrument (Δdinst), can be obtained by the XRD pattern of a calibration standard such as LaB6 under the same conditions.28 As shown in Figure S5, a linear fitting is obtained from the XRD results of LaB6. Then Δdε versus d of these two kinds of Cs5Cu3Cl6I2 are presented in Fig. 1f. Clearly, for each lattice plane, the Δdε of grinded Cs5Cu3Cl6I2 has a larger value than that of SAC-Cs5Cu3Cl6I2, indicating a higher lattice strain of grinded Cs5Cu3Cl6I2. Besides, liner fitting of those data points shows a smaller slope of SAC-Cs5Cu3Cl6I2 than that of grinded one, meaning that the lattice strain is released during the SAC conditions. Based on the above analysis, we achieved morphology and lattice strain management of Cs5Cu3Cl6I2 powder through the effective and simple SAC strategy.
Figure 2a presents the photoluminescence excitation (PLE) and photoluminescence (PL) spectrum of SAC-Cs5Cu3Cl6I2, showing excitation peak at 300 nm and blue emission centered at 470 nm with FWHM of ~ 88 nm. A large Stokes shift of 170 nm can be observed, demonstrating that the emission of Cs5Cu3Cl6I2 shows negligible self-absorption. The steady PL spectra in Fig. 2b show that the PL intensity of SAC-Cs5Cu3Cl6I2 is larger than that of grinded one, indicating that the nonradiative recombination within SAC-Cs5Cu3Cl6I2 is decreased, originated from the released lattice stain and reduced defects due to the SAC method.
It is noteworthy that the difference in the PLQY values between the two samples shows a significantly smaller variation compared to the differences observed in RL shown in Fig. 1b. This is attributed to differences in the working mechanisms of perovskites under visible light excitation and X-ray excitation conditions.39 The time-resolved PL (TRPL) and corresponding exponential fitting results are presented in Fig. 2c and Table S2, and the lifetime of SAC and grinded Cs5Cu3Cl6I2 are calculated to be 47.56 µs and 41.31 µs, respectively. Consequently, the PLQY of SAC and grinded Cs5Cu3Cl6I2 are 99.5% and 86.6%, respectively (Fig. 2d). Near-unity PLQY of 99.5% indicates that the SAC-Cs5Cu3Cl6I2 exhibits excellent photoluminescence properties, in consistent with the outstanding quality of SAC-Cs5Cu3Cl6I2.
To dig deep into the photoluminescence properties of SAC and grinded Cs5Cu3Cl6I2, we performed the temperature-dependent PL tests from 293 K to 453 K. As shown in Fig. 2e-f and Figure S6, the PL intensity decreases with the increase of temperature, because of thermal quenching effect.40 While the PL intensity of grinded Cs5Cu3Cl6I2 shows a rapidly decrease than that of SAC-Cs5Cu3Cl6I2. After heated to 453 K, the PL intensity of SAC-Cs5Cu3Cl6I2 retains 82.5% of the intensity at 293 K, much higher than the grinded Cs5Cu3Cl6I2 (46.3%), demonstrating better thermal stability of SAC-Cs5Cu3Cl6I2. Furthermore, the exciton binding energy can be obtained by:41
$$I\left(T\right)=\frac{{I}_{0}}{1+A\text{e}\text{x}\text{p}(-\frac{{E}_{b}}{{k}_{B}T})}$$
3
where, I(T) and I0 are the integrated PL intensities at temperatures T and 0 K, respectively, A is the proportionality parameter of the nonradiative process, Eb is the exciton binding energy, and kB is the Boltzmann constant. Figure S7 displays the fitting results of integrated PL intensity versus 1/T using Eq. (3), from which we can conclude that the Eb of SAC-Cs5Cu3Cl6I2 (497.4 meV) is larger than that of grinded one (330.3 meV). A larger Eb means that the exciton dissociation process requires more energy, beneficial for radiative recombination and thermal stability.23 In conclusion, the Cs5Cu3Cl6I2 prepared by SAC exhibits better photoluminescence properties, including negligible self-absorption, high PLQY of 99.5%, and robust thermal stability.
Next, we evaluated the scintillation properties and X-ray detection performance of the scintillators. Figure 3a shows the integrated RL intensity of various scintillators, which was collected using a CMOS detector under different X-ray dose rates, indicating a good liner response. And the largest slope of SAC-Cs5Cu3Cl6I2 scintillator manifests its highest sensitivity to X-rays than that of grinded-Cs5Cu3Cl6I2 and LYSO:Ce scintillators. The detection limit of SAC-Cs5Cu3Cl6I2 is estimated to be 34.9 nGyair s− 1 through linear fitting when the signal-to-noise ratio (SNR) is 3, which is significantly lower than that of grinded-Cs5Cu3Cl6I2 (82.1 nGyair s− 1), LYSO:Ce (84.0 nGyair s− 1) and the medical diagnostics requirement (5.5 µGyair s− 1).42 To better assess its comprehensive stability, we compared the scintillation properties of our SAC-Cs5Cu3Cl6I2 films against humidity, heat, and X-ray irradiation. As shown in Fig. 3b and Figure S8a, the RL intensity of SAC-Cs5Cu3Cl6I2 film could retain impressive 91.2% of its initial scintillation properties after immersing in water for 55 min, demonstrating an excellent humidity stability. For its scintillation properties against heat, SAC-Cs5Cu3Cl6I2 film was heated from 293 K to 453 K, the RL intensity decreased slightly with the temperature rising because of the thermal quenching effect (Fig. 3c and Figure S8b). At 453 K, the SAC-Cs5Cu3Cl6I2 film still exhibited excellent scintillation properties (82.3% of RL intensity at 293 K), indicating the outstanding thermal stability of SAC-Cs5Cu3Cl6I2 film (Figure S8c). Furthermore, under continuous X-ray irradiation with a dose rate of 45 mGyair s− 1, no obvious change in the scintillation properties of SAC-Cs5Cu3Cl6I2 film was found after a total X-ray dose of 648 Gyair (Fig. 3d and Figure S8d). The total dose of 648 Gyair equals ∼6,480,000 times of the X-ray dose required for a typical X-ray chest radiograph, which needs ∼0.1 mGyair per time for commercial instruments.43 Thus, the remarkable stability of our SAC-Cs5Cu3Cl6I2 scintillator under harsh working conditions makes it a competitive candidate in medical imaging and nondestructive testing.
We next integrated the Cs5Cu3Cl6I2@PDMS film with the CMOS substrate by blade coating to obtain monolithic imagers. Figure 4a shows schematic diagram of the monolithic imager, and Fig. 4b presents the photograph of our device. The CMOS substrate utilized in this study comprises approximately 1,400,000 pixels with a 1260 × 1172 matrix configuration. The size of each pixel is 49.5 µm. For comparison, LYSO:Ce and CsI:Tl imagers were used, in which the freestanding scintillators were directly placed on the CMOS substrate (Figure S9b-c). This later modular device structure is the most common in commercial devices.44,45 Fig. 4c shows images of a roller pen with our monolithic Cs5Cu3Cl6I2, modular LYSO:Ce, and Cs:Tl based X-ray imagers under an X-ray dose rate of 200 µGyair s− 1, respectively. The experimental data conspicuously demonstrates that the images of the spring in the roller pen obtained by the LYSO:Ce and CsI:Tl devices appear blurred in comparison to that acquired using our Cs5Cu3Cl6I2 device. This observation underscores the effectiveness of the monolithic device structure in substantially reducing optical crosstalk in X-ray imagers.
To assess the high-resolution imaging capabilities of our detector, we conducted imaging tests with X-ray magnification by positioning the object at a distance from the detector. Figure 4d illustrates the schematic diagram of X-ray path amplification. Under the amplification condition, the spatial resolution was evaluated using standard line pair X-ray test patterns (Fig. 4e and Figure S10). The results demonstrate that the integrated imaging system we designed exhibits an outstanding spatial resolution exceeding 30 lp mm− 1. Subsequently, we measured the Modulation Transfer Function (MTF) of the imager using the slanted-edge method. When the MTF reaches 0.2, the spatial resolution is calculated to be approximately 39.9 lp mm− 1, which means the device can distinguish two adjacent objects that are only ~ 12.5 µm apart. (Fig. 4f, g). It is worth noting that the aforementioned data were obtained under actual X-ray irradiation conditions. We also note that literature often employs cameras or CCDs to complete the imaging process.24,46Their experimental setups typically involve both X-ray path magnification and visible light path magnification. The latter makes a significant contribution to signal magnification as it can be conveniently accomplished through optical lenses.
The above discussion highlights the capabilities of our monolithic device in real working conditions, particularly in terms of resolving small objects. To provide a comprehensive evaluation of the device, and for a horizontal comparison, we conducted spatial resolution tests on the sample without any path magnification - that is, with the object placed directly against the detector (Figure S11). The X-ray image of line pair test patterns shows the spatial resolution is ~ 9 lp mm− 1 (Fig. 4h and Figure S12), which is even higher than that of 7.1 lp mm− 1 of the commercialized direct conversion a-Se detector. This high-resolution capability is attributed to the in-situ integration of the scintillator with CMOS. The monolithic device structure effectively eliminates the optical crosstalk that typically arises from the air gap between the scintillator and CMOS.
Commercial X-ray instruments typically require high radiation doses for imaging, resulting in health risks to patients. To address this concern, we investigated the low-dose imaging capabilities of our monolithic Cs5Cu3Cl6I2 imager. As illustrated in Fig. 5a, the imaging results of a spring within a roller pen were remarkably clear at X-ray doses greater than 10 µGyair. Impressively, even at a reduced X-ray dose of 0.8 µGyair (Figure S13), the intricate details of the spring remained distinguishable. It's worth noting that obtaining a high-quality image using conventional X-ray instruments usually necessitates a dose of around 100 µGyair. These results underscore the potential of our monolithic Cs5Cu3Cl6I2 imager for achieving high-resolution X-ray imaging with significantly lower radiation doses. Furthermore, afterglow is also a significant index to evaluate the imaging performance. Cs5Cu3Cl6I2 has been reported to exhibit negligible afterglow (0.1% at 10 ms).24 As shown in Video S1, a roller pen moved at constant speed on a conveyor belt under X-ray. The images were recorded by our Cs5Cu3Cl6I2 imager at a frame rate of 18 frames per second (FPS). As the roller pen moved, the internal structure of roller pen presented clearly and without any residual shadow, indicating our Cs5Cu3Cl6I2 imager is applicable for high-resolution and fast dynamic X-ray imaging. To further demonstrating the capabilities of our device, Fig. 5b-f show X-ray imaging outcomes on various objects, including peanuts, a bank card, a metal bookmark inscribed with Chinese characters ("wei lai ke qi"), the Solar Divine Bird (a symbol of Chengdu), and a steel tape. These images not only demonstrate exceptional clarity but also enable the identification of intricate details within the objects. This further highlights the significant potential value of this technology across various application scenario, spanning from medical diagnostics to industrial quality control.