Crystal Growth and First-Principle Calculation. Figure 1a presents the schematic diagram of the custom vertical Bridgman growth furnace with three zones. The temperature gradient can be adjusted from 10°C to 25°C for a length of 40 mm and the descending speed can be adjusted up to 0.001 mm/h. The pure Cs3Cu2I5 and 2.5 at% Li+-doped Cs3Cu2I5:Li single crystals are grown by the self-seeding Bridgman technique with a capillary tube. The as-grown Cs3Cu2I5:Li single crystal ingots with a diameter of 12 mm and transparent machined sample without cracks and visible inclusions are exhibited in Fig. 1a. The actual Li+ doped concentration was confirmed by ICP-OES measurement. The actual Li+ concentration is about 0.92 at% for 2.5 at% Li+-doped Cs3Cu2I5 crystal, indicating that the Li+ was introduced successfully. The powder XRD pattern of the polycrystalline Cs3Cu2I5:Li can be indexed to the pure Cs3Cu2I5 PDF card #45–0077 revealing the absence of impurity phases like CsI or CuI, as shown in Fig. 1b. Cs3Cu2I5 crystallizes into the orthorhombic space group of Pnma at room temperature. Consider for the radius of Li+ (0.59 Å) is closer to that of Cu+ (0.60 Å), the Li+ exhibits a tendency to substitute for the Cu+ site.
To gain mechanistic insights into the optical properties of the Li+ doped effect on Cs3Cu2I5 single crystals, the electronic band structures are derived by density-functional theory (DFT) calculation with the PBE hybrid functional. Both the valence band edges in Figs. 1c and d are completely flat between the F and Z direction, which differ from the large bandwidth and dispersion band edges of conventional 3D structures. The direct band-gap of Cs3Cu2I5 is smaller than that of Li+-doped Cs3Cu2I5, indicating a broken symmetry after introducing Li+ to enlarge the band-gap. According to the DOS and PDOS map, the valence band maximum (VBM) is mainly contributed by the I-p orbitals and some Cu-d orbitals. Whereas, the conduction band minimum (CBM) density states are mainly composed of Cu-4s and I-5p orbitals. The flat VBM energy profiles of pure and Li+-doped Cs3Cu2I5 indicate localized charge distributions corresponding to the charge density maps of VBM and CBM (Figs. 1e and f). In particular, most of the charges of Li+-doped Cs3Cu2I5 transfer from the [LiI6] octahedrons to [CuI6] octahedrons while some charges transfer from the metal to halide upon photoexcitation from VBM to CBM as shown in Fig. 1f marked with red circle. Compared to the pure Cs3Cu2I5, more prominent charge transfer in Li+-doped Cs3Cu2I5 enhances the conductivity and doping with Li extends the carrier lifetime and improves the carrier mobility.
Optical properties of pure and Li + -doped Cs 3 Cu 2 I 5 single crystal. The survey XPS spectrum of Cs3Cu2I5:Li crystal in Fig. 2a reveals the C 1s, Cs 3d, Cu 2p, and I 3d components and the high-resolution spectra are conducted to validate the chemical states of matrix ions. The binding energies of Cu 2p1/2 and Cu 2p3/2 are consistent with Cu-I bond and demonstrate the presence of Cu+ in the host. The satellite peak for the Cu2+ oxidation state cannot be observed. The binding energies of the 3d orbital correspond to the Cs and I + 1 and − 1 states, respectively. Compared to the XPS result of Cs3Cu2I5 crystal, presenting in Figure S1, the Li+ doped has no influence on chemical states of the crystal composition, except that the peaks position has a slight shift of 0.4 eV. The PLE and PL spectra of pure Cs3Cu2I5 and Cs3Cu2I5:Li single crystals was measured and shown in Fig. 2b while the time resolved PL decay time curves were shown in Fig. 2c. After Li+ doping, the excitation (at 310 nm) and emission (at 445 nm) bands do not show obvious changes, but the decay time becomes slightly longer from 1002 ns to 1015 ns, indicating that the Li+-doped Cs3Cu2I5 crystal has a longer electron-hole recombination time. Hence, the mobility-lifetime (µτ) product increases with Li+ doping boding well for direct X-ray detection. Figures 2d and e present the ultraviolet-visible absorption spectra of the pure and Li+-doped Cs3Cu2I5 sample, respectively. The inset is the Tauc plot curve converted by the Kubelka-Munk equation for the band gap calculation26. Both the pure and Li+-doped Cs3Cu2I5 exhibit a sharp absorption edge indicative of a direct band gap. The corresponding band gap energy (Eg) of Cs3Cu2I5:Li is estimated to be 3.62 eV, which is bigger than that of the pure Cs3Cu2I5 crystal (3.57 eV) and slightly exceeds that of the conventional semiconductor X-ray detector 27. This trend is in good agreement with the DFT calculation shown in Figs. 1c and d.
Electronic properties of the Cs 3 Cu 2 I 5 single crystal. The carrier mobility is an important factor for X-ray detection, and the Hall-effect measurement was used to investigate the electronic properties of pure and Cs3Cu2I5:Li single crystal. Two samples exhibit n-type conductivity. The Cs3Cu2I5:Li single crystal has higher carrier mobility of 9.52 cm2 V− 1 s− 1 than the pure Cs3Cu2I5 single crystal (6.49 cm2 V− 1 s− 1), and it is larger than that of Cs3Bi2Br9 (1.54 cm2 V− 1 s− 1)28 and two times that of Cs2AgBiBr6 (3.17 cm2 V− 1 s− 1)29 and MAPbBr3 (4.16 cm2 V− 1 s− 1)30. It means that the Li dopant enhances the conductivity of the Cs3Cu2I5 single crystal, and the positive effect on the conductivity is confirmed by experiments and DFT calculation.
The pure Cs3Cu2I5 and Cs3Cu2I5:Li single crystal ingots are machined into squares and integrated into X-ray detectors with the stacked structure of Au/Cs3Cu2I5 or Cs3Cu2I5:Li/PCBM/Au and level alignment as shown in Fig. 3a. Figure 3b compares the absorption coefficients of typical and potential materials for X-ray detection, which were calculated according to the photon cross-section database31. Since Cs3Cu2I5 consisting of heave elements of Cs, Cu, and I have a large high density (4.52 g cm− 3), it has a bigger X-ray absorption coefficient than Cs2AgBiBr6, α-Se, and commercial silicon, and comparable with CdTe. The photoconductivity measurements are carried out to determine the optoelectronic properties of the Cs3Cu2I5 and Cs3Cu2I5:Li single crystal, as shown in Fig. 3c. The photocurrent response is monitored by applying a voltage ranging from 0 to 60 V and the Hecht formula is employed to fit the photocurrent-voltage curve:
I = I0µτV/L2(1-exp(-L2/µτV))
where I0 is the saturated current, L is the thickness of the Cs3Cu2I5 and Cs3Cu2I5:Li single crystal, and V is the applied voltage. The µτ products of the pure Cs3Cu2I5 and Cs3Cu2I5:Li single crystal are determined to be 1.4×10− 4 cm2 V− 1 and 2.9×10− 4 cm2 V− 1, respectively, and the bigger µτ product indicates enhanced transport of photogenerated charges. Moreover, the Cs3Cu2I5:Li single crystal has a smaller resistivity of 2.7×1010 Ω cm compared to 4.3×1011 Ω cm of the Cs3Cu2I5 single crystal, as shown in Fig. 3(d). Figure 3(e) shows the current density-voltage (J-V) characteristics of the X-ray detectors made of the Cs3Cu2I5 and Cs3Cu2I5:Li single crystals. The Cs3Cu2I5 X-ray detector exhibits obviously lower dark currents than Cs3Cu2I5:Li due to the high resistivity, and the dark current mainly comes from the injected current form the electrodes at the applied electric field. Meanwhile, the Cs3Cu2I5:Li X-ray detector shows a significantly enhanced light current of 2.0×10− 8 A cm− 2 at an electric field of 45 V/mm, which is more than twice that of the Cs3Cu2I5 X-ray detector of 9.3×10− 9 A cm− 2. The light current originates from the X-ray excited photocurrent, and the Li-doped Cs3Cu2I5:Li single crystal possesses higher conductivity due to enhanced charge transport thus facilitating high-sensitivity X-ray detection.
Direct X-ray detection. To obtain dynamic information from the Cs3Cu2I5 and Cs3Cu2I5:Li X-ray detectors, the time-resolved light current response for various X-ray dose rates from 1579 µGyair s− 1 to 82 µGyair s− 1 are presented in Fig. 4a, the inset is the picture of detector. Both detectors show decent response signals upon on/off X-ray switching and good linear relationship with the X-ray dose rates. The Cs3Cu2I5:Li X-ray detector exhibits obviously higher dark currents and light currents with larger X-ray on-off than the pure Cs3Cu2I5 detector at an electric field of 45 V/mm, indicating the Cs3Cu2I5:Li X-ray detector has more sensitive X-ray detection properties. Figure S2 presents the current response of the Cs3Cu2I5 and Cs3Cu2I5:Li detectors during X-ray on/off at different electric fields and X-ray dose rates.
To investigate the sensitivity of the Cs3Cu2I5 and Cs3Cu2I5:Li X-ray detectors, the X-ray generated photocurrent density are plotted as a function of X-ray dose rates as shown in Figs. 4b and 4c. The sensitivity of the X-ray detector can be calculated from the slope of the photocurrents versus X-ray dose rates plots acquired at different electric fields. The pure Cs3Cu2I5 detector has X-ray sensitivities of 441.5 µC Gyair−1 cm− 2, 363.3 µC Gyair−1 cm− 2, 288.5 µC Gyair−1 cm− 2, 233.1 µC Gyair−1 cm− 2 and 168.9 µC Gyair−1 cm− 2 at 45, 37, 30, 28 and 15 V/mm, respectively. In comparison, the Cs3Cu2I5:Li X-ray detector shows enhanced sensitivities of 831.1 µC Gyair−1 cm− 2, 661.8 µC Gyair−1 cm− 2, 422.0 µC Gyair−1 cm− 2, 311.5 µC Gyair−1 cm− 2 and 95.6 µC Gyair−1 cm− 2 at electric fields of 45, 37, 30, 28 and 15 V/mm, respectively.
The signal-to-noise ratio (SNR) is another important parameter for X-ray detection and can be calculated by the following equation:
SNR = Isignal / Inoise = (Iphoto – Idark) / Inoise,
where Iphoto is the average current during X-ray irradiation, Idark is the average dark current derived from parallel experiments at each bias, and Inoise is the noise current calculated from the standard deviation of the photocurrents. The SNRs of the X-ray detectors at an electric field of 45 V/mm are plotted as a function of dose rates in Fig. 4d, and a linear dependence with X-ray dose rates is observed. According to the International Union of Pure and Applied Chemistry standard, the detector signal should maintain an SNR of over 3. The Cs3Cu2I5:Li X-ray detector has a smaller limit-of-detection (LoD) of 34.8 nGyair s− 1 than the pure Cs3Cu2I5 X-ray detector of 335.1 nGyair s− 1, which is consistent with the higher sensitivity of the Cs3Cu2I5:Li X-ray detector. Then, we highlight the performance of perovskite based X-ray direct conversion detectors and summarize the important parameters including the µτ product, sensitivity, detection limit and resistivity in Table S1. The Cs3Cu2I5:Li crystal presents good comprehensive performance.
To evaluate the detection stability under continuous operating conditions, X-ray on/off experiments are carried out at the 45 V/mm external electric field and 1579 µGyair s− 1 X-ray dose for 3600 s as shown in Figs. 4e and S3. Both the Cs3Cu2I5 and Cs3Cu2I5:Li X-ray detectors exhibit reproducible and stable response with small X-ray on/off changes, indicating excellent materials and working stability under continuous X-ray exposure. The temporal baseline tracking of the X-ray detectors shown in Figure S4, and the current drift (Idrift) can be calculated by the following equation:
Idrift = (It – I0) / (E×S×t),
where It is the current at time t, I0 is the current immediately after stabilization, E is the electric field, and S is the area of the device. Both the Cs3Cu2I5 and Cs3Cu2I5:Li X-ray detectors are able to maintain stable dark and light currents in 3600 s, and both of the Cs3Cu2I5 and Cs3Cu2I5:Li X-ray detectors show smaller light current drifts of 1.17×10− 4 A cm–1 s–1 V–1 and 1.35×10− 4 A cm–1 s–1 V–1, indicating excellent working stability.
The X-ray imaging capability of the Cs3Cu2I5:Li X-ray detector is assessed by x-y scanning experiments with a key containing a plastic handle and metal bit as the imaging target. As shown in Fig. 4f, the X-ray image clearly reveals a clear outline of the key consistent with the optical image in the scanning mode.