Highly stable and spectrum-selective ultraviolet photodetectors based on lead-free copper-based perovskites

Ying Lia, Zhifeng Shi*a, Wenqing Lianga, Lintao Wanga, Sen Lia, Fei Zhanga, Zhuangzhuang Maa, Yue Wanga, Yongzhi Tiana, Di Wua, Xinjian Li*a, Yuantao Zhangb, Chongxin Shanac and Xiaosheng Fang*d
aKey Laboratory of Materials Physics of Ministry of Education, Department of Physics and Engineering, Zhengzhou University, Daxue Road 75, Zhengzhou 450052, China. E-mail: shizf@zzu.edu.cn; lixj@zzu.edu.cn
bState Key Laboratory on Integrated Optoelectronics, College of Electronic Science and Engineering, Jilin University, Qianjin Street 2699, Changchun 130012, China
cState Key Laboratory of Luminescence and Applications, Changchun Institute of Optics Fine Mechanics and Physics, Chinese Academy of Sciences, Changchun 130033, China
dDepartment of Materials Science, Fudan University, Shanghai 200433, China. E-mail: xshfang@fudan.edu.cn

Received 29th August 2019 , Accepted 14th October 2019

First published on 17th October 2019

Ultraviolet (UV) photodetectors with high spectrum selectivity are highly desired for certain special applications, like selected-wavelength imaging and UV-phototherapy. Here, for the first time, a highly stable and spectrum-selective self-powered UV photodetector based on lead-free Cs3Cu2I5 films was demonstrated. By designing the Cs3Cu2I5/GaN heterojunction, a narrow spectral response “window” of 300–370 nm was realized, and the working mechanism was investigated through the dependence of the spectral response on the absorber thickness and bias voltage. Because of the high material integrity of the Cs3Cu2I5 films and competent interfacial charge transfer from Cs3Cu2I5 to GaN, the proposed device exhibits a remarkable photodetection capability. Moreover, the proposed photodetector without encapsulation exhibits excellent working stability in ambient air, and typically it can operate continuously for more than 12 h under a high operation temperature of 373 K, indicating a high temperature tolerance and desired performance for practical applications under harsh environments. In addition, the device was used as the sensing pixels in a UV imaging system, and high-resolution imaging patterns were achieved. The results highlight the great potential of such lead-free Cs3Cu2I5 as environment-friendly alternatives for stable and spectrum-selective UV photodetection without band-pass filters that can be employed in imaging.

New concepts

An intrinsically large light absorption coefficient, and high and well-balanced carrier transport ability endow metal-halide perovskites with great promise for high-performance photodetectors. However, compared to the intensive efforts on broadband photodetectors, narrowband photodetectors based on perovskites, especially targeting the ultraviolet (UV) region, have not been reported. Here, for the first time solution-processed lead-free Cs3Cu2I5 films were used as the light absorber to fabricate spectrum-selective UV photodetectors by using the charge collection narrowing concept, overcoming the shortcomings of conventional strategies for making narrowband photodetectors (high cost, complex configuration, and difficult integration), and a narrow response “window” of 300–370 nm was achieved. Unlike traditional perovskites, lead-free Cs3Cu2I5 is very stable against heat, UV light, and environmental oxygen/moisture. Typically, the unencapsulated device can endure a high working temperature (373 K) for 12 h continuous operation in ambient air. Further, UV imaging applications were demonstrated by deploying the photodetectors as sensing pixels. The results suggest that lead-free Cs3Cu2I5 is potentially an attractive candidate for stable and narrowband UV photodetectors, rendering it potentially useful for the assembly of optoelectronic systems in the future.


Ultraviolet (UV) photodetectors with high spectrum selectivity and small full-width at half-maximum (FWHM) are highly desired for some special applications, where detecting a specific wavelength while suppressing the background emission is required, like flame detection to explore the existence of a specific kind of chemical element, selected-wavelength imaging, fluorescence detection, defense and UV-phototherapy.1–5 In general, narrowband detection can be achieved by combining broadband detectors with band-pass filters,6–8 using narrowband absorption of photoactive materials,9 or deliberately enhancing the absorption in a specific wavelength range through the plasmon effect.10 For the first method, the design complexity of the photodetector systems is increased and the device is difficult to integrate, and the inherent limitations of filters including sensitivity to contamination, high cost, and surface scratches have to be considered. The second method is infeasible for a narrowband spectral response over a long-wavelength range. For the third method, it has a few problems, including high cost and complex configuration. Moreover, plasmon enhanced absorption cannot be adjusted over a large spectral range. Thus, there is a great need for low-cost narrowband photodetectors with architectural simplicity. In this regard, filter-free spectrum-selective photodetectors have recently attracted intense attention, for example, charge collection narrowing (CCN) with thick absorbing layers has been investigated for narrowband detection over a wide wavelength range, involving different kinds of materials.11–15

Recently, metal-halide perovskites have attracted huge attention due to their excellent optoelectronic features, including high light absorption coefficient, tunable emission wavelength, direct bandgap, and high and well-balanced carrier transport ability.16–25 These properties make the materials an ideal candidate for optoelectronic devices. By developing such materials as the light absorber, much progress in broadband photodetectors has been witnessed in recent years.26–30 In contrast, narrowband detectors based on perovskites materials have rarely been reported. In this connection, there are a few representative studies that deserve comments herein. For example, Huang et al. and Lin et al. demonstrated high-performance narrowband photodetectors with a tunable spectral response by using methylammonium lead trihalide perovskite single crystals as the light absorber layer.10,31 Similar photodetection behaviors were also reported in Kuang's and Zhang's studies.32,33 In their cases, the success of bulk single crystal narrowband photodetection depends on the sharp change of the light absorption coefficient close to the band edge and surface-charge recombination, in other words, manipulation of the external quantum efficiency via CCN. In addition, by using two-dimensional (2D) perovskite single crystals, Li and coworkers successfully fabricated tunable narrowband photodetectors, and they attributed it to the self-trapped states in 2D perovskites and large conductivity in the out-of-plane direction.34 Very recently, they reported a spectrum-selective dual-band photodetector based on centimeter-sized 2D layered perovskites, and the dual-band spectral response can be adjusted over the visible range by composition engineering.35 Besides, by using all-inorganic CsPbX3 (X = Cl/Br/I) polycrystalline thick films as the photoactive layer, Zeng and co-workers realized narrowband detection in the visual band and demonstrated the potential application in simulating the human visual system.36 Despite the inspiring progress that has been made on perovskite narrowband photodetectors, the environmental instability and toxicity of conventional lead-halide perovskites has become a stumbling block that limits their further applications. Thus, exploring lead-free and environmentally stable perovskites to solve the above obstacles is urgent. Moreover, an indisputable fact is that the previously reported perovskite narrowband photodetectors always operate in the visible range, and there has been no relevant report targeted at the UV region as far as we know. Besides, in previous research on perovskite narrowband devices, the photodetector performance was mainly related to the responsivity, specific detectivity, external quantum efficiency, and other related parameters, but stability issues of the unencapsulated device were rarely investigated.

Following this line of thought, for the first time, a highly stable and spectrum-selective UV photodetector based on lead-free Cs3Cu2I5 films was demonstrated. By using the CCN concept, the surface-charge recombination can be controlled through the thickness of the absorber and the device only responds to photons with an energy in a narrow “window” of 300–370 nm. The photodetectors demonstrate a high responsivity of 0.28 A W−1, a specific detectivity of 1.4 × 1012 Jones, an on/off photocurrent ratio of 1.2 × 105, and fast response speeds of 95/130 μs (800 Hz) under UV light excitation (320 nm). Additionally, the device without encapsulation exhibits excellent working stability against thermal degradation in the open air, indicating a remarkable temperature tolerance and desired compatibility for applications under harsh conditions. Furthermore, we demonstrated high-resolution UV imaging applications based on the fabricated narrowband photodetectors. The results suggest that the present lead-free Cs3Cu2I5 can serve as a good candidate for stable and spectrum-selective UV photodetection without band-pass filters.

Results and discussion

The morphology of the Cs3Cu2I5 films on an n-GaN template was examined by scanning electron microscopy (SEM). As shown in Fig. 1a, the Cs3Cu2I5 films were characterized by a dense morphology with a high surface coverage, and the mean grain size was about 330 nm. The energy dispersive X-ray spectroscopy (EDS) elemental mapping results (Fig. 1b) show that the Cs, Cu, and I elements are uniformly distributed over the selected square areas. The EDS spectrum quantitative analysis shown in Fig. S1 (ESI) further confirmed that the atomic ratio of Cs[thin space (1/6-em)]:[thin space (1/6-em)]Cu[thin space (1/6-em)]:[thin space (1/6-em)]I is 3[thin space (1/6-em)]:[thin space (1/6-em)]1.98[thin space (1/6-em)]:[thin space (1/6-em)]5.22, close to the stoichiometry of the Cs3Cu2I5 material. Fig. 1c shows typical X-ray diffraction (XRD) patterns of the as-grown samples in comparison with the XRD patterns of the standard Cs3Cu2I5. A set of diffraction peaks at 24.11°, 24.86°, 25.68°, 26.40°, and 30.75° can be assigned to the (122), (400), (320), (222), and (004) planes of the orthorhombic Cs3Cu2I5 (space group Pnma, a = 10.1743 Å, b = 11.6532 Å, c = 14.3621 Å, α = β = γ = 90°). Moreover, X-ray photoelectron spectroscopy (XPS) measurements were performed to identify the valence state of Cu ions. As shown in Fig. S2 (ESI), the binding energy at 932.2 and 951.9 eV can be assigned to Cu 2p3/2 and Cu 2p1/2, respectively, corresponding to the Cu (+1) state as reported in other studies.37–40 In addition, the characteristic satellite peak of Cu2+ is reported at ∼943 eV,41 which is absent in our results. Therefore, the valence state of Cu in the products is +1, not divalent copper. Fig. 1d shows the schematic crystal structure of orthorhombic Cs3Cu2I5, in which two types of Cu+ sites were incorporated, a trigonal site and a tetrahedral site. Both of them form [Cu2I5]3− and each [Cu2I5]3− is separated by Cs+ ions.42 Preliminary optical studies (Fig. 1e) show that the Cs3Cu2I5 films exhibit a strong and broad photoluminescence (PL) emission peak at 442 nm, and the FWHM is 70 nm. The corresponding absorption edge is at 326 nm. In addition, the Cs3Cu2I5 films have high light transmission (>95%) in the visible spectral region, as shown in the measured transmission spectrum (Fig. S3, ESI). Fig. S4 (ESI) shows the absorption spectrum of the Cs3Cu2I5 films grown on the GaN template, and two clear and steep absorption edges occur at ∼326 and 365 nm, respectively, corresponding to the bandgap absorption of the upper Cs3Cu2I5 and underlying GaN template. Note that a relatively large Stokes shift of ∼116 nm implies that excited-state structural reorganization or exciton self-trapping is the possible transition mechanism of Cs3Cu2I5.42–44 In order to gain more information on the carrier recombination dynamics, we measured the PL decay of the Cs3Cu2I5 films using transient-state PL measurements. As shown in Fig. 1f, the PL decay trace strictly fitted to a biexponential mode, producing an average lifetime of ∼794.2 ns, and the detailed fitting parameters are listed in Table S1 (ESI). Such a large carrier recombination lifetime allows longer time for collection and transport of photo-generated carriers, and is also an indicator of good photovoltaic performance for such materials. As deduced from previous research about copper-halide compounds, broad emission with a large lifetime is a feature of Cu(I)-complex phosphorescent emission involving Jahn–Teller distortion.42,45,46 In detail, a Cu(I) halide with a d10 closed-shell favors a tetrahedral geometry. In the photoexcited state, Cu(I) with tetrahedral geometries forms a Jahn–Teller distorted Cu(II)-d9 center.42 In other words, the electronic configuration of Cu(I) 3d10 changes to Cu(II) 3d9 after the absorption of the bandgap photon energy, causing Jahn–Teller distortion and subsequent reorganization of the excited-state structure.
image file: c9mh01371g-f1.tif
Fig. 1 (a) Top-view SEM images of the Cs3Cu2I5 films. (b) EDS elemental mapping results for a selected region in (a). (c) XRD patterns (upper) of the as-grown Cs3Cu2I5 films and the standard XRD patterns (bottom) for orthorhombic Cs3Cu2I5. (d) Crystal structure of Cs3Cu2I5. (e) Absorption and PL spectra of the Cs3Cu2I5 films. (f) Time-resolved PL decay and fitting curve of the Cs3Cu2I5 films.

As we all know, the instability of conventional lead-halide perovskites is the main obstacle for their further applications in optoelectronic devices. Typically, lead-halide perovskites in moist air would decompose due to the hygroscopic amine salts, and even if the resulting materials and devices are dried again, the chemical reaction process is irreversible. In this study, the environmental stability of the Cs3Cu2I5 films was tested by studying the effects of heat, UV light irradiation, and storage period on their optical and structural properties. Firstly, we performed thermal cycling PL tests at 293–373 K to evaluate the thermal stability of the Cs3Cu2I5 films, and the results are shown in Fig. 2a–d. It can be observed that the relative PL intensity of the Cs3Cu2I5 films gradually decreases with the increase of temperature (heating process) due to the added nonradiative recombination caused by the heating effect. To get a better understanding of this, PL decay measurements at different temperatures were performed. As shown in Fig. S5a (ESI), the PL decay curve at 293 K can be fitted with a biexponential decay, and an average PL decay lifetime of 794.2 ns was obtained. Also, we observed that the PL lifetime depends strongly on the temperature. As summarized in Fig. S5b (ESI), with the increase of temperature, the PL lifetime decreases gradually. As deduced from previous research, the decrease of the PL lifetime can be explained by the increase of charge-carrier trapping states as the relaxation pathway.47,48 After the first cooling process was administered (Fig. 2a), a small decay of about 6.6% was observed, much better than for conventional lead-halide perovskites.49,50 The unrecoverable behavior may result from the added structural defects in the Cs3Cu2I5 films due to the heating effect. Correspondingly, the radiative recombination probability of photo-generated carriers was reduced. More excitingly, over the subsequent two heating/cooling cycles, almost no significant emission decay occurs, as plotted in Fig. 2b and c. The insets of Fig. 2d display three representative photographs of the sample captured at cycle 1 (293 K), cycle 6 (373 K), and cycle 10 (293 K), from which the recoverable emission feature can be more intuitively observed. It should be mentioned that, in the ten heating/cooling cycles, the spectral shape and emission position did not change except for the change in peak intensity. Fig. S6 (ESI) shows the corresponding PL spectra over the third cycle, and other cycles demonstrate the same phenomenon. We think that there are two main reasons for the better stability of Cs3Cu2I5 films than that of conventional perovskites. On the one hand, Cs3Cu2I5 has a 0D light-sensitive site of [Cu2I5]3−, and low-dimensional perovskites generally exhibit better structural stability than the three-dimensional perovskite phase.51 On the other hand, for conventional organic–inorganic hybrid perovskites (CH3NH3PbX3), the CH3NH3+ cation was extremely sensitive to oxygen and moisture. It has been reported that replacing methylammonium with inorganic cesium offers the perovskite extra thermal stability up to its melting at ∼500 °C,52 but the decomposition enthalpy of CsPbX3 is relatively low. Experimentally, thermogravimetric analysis of the Cs3Cu2I5 films was conducted to measure the resistance of the material to high temperature. As shown in Fig. S7 (ESI), the decomposition of the Cs3Cu2I5 product starts at a temperature as high as 925 K, which is much higher than the decomposition temperature of 523 K for CH3NH3PbBr3 and 853 K for CsPbBr3.53 Secondly, the photostability of the Cs3Cu2I5 films was further tested under continuous UV light irradiation (265 nm, 3.0 mW cm−2), and the PL spectra were extracted with a time interval of 0.5 h. As shown in Fig. 2e and Fig. S8 (ESI), after continuous UV light irradiation for 8 h, a tiny decay of about 2% was generated with the FWHM and peak position of the PL spectra unchanged. The excellent stability of Cs3Cu2I5 films against UV light degradation shows them to be a reliable light absorber for photodetection applications. Moreover, the environmental stability of the Cs3Cu2I5 films, without any encapsulation and protection, was investigated by intermittently recording their XRD patterns and PL spectra after different storage periods. As shown in Fig. 2f, the as-grown Cs3Cu2I5 films maintain the structural integrity without decomposition after 28-day storage in ambient conditions (20 °C, 35–50% humidity). In addition to a slight change of the peak intensity, no other diffraction signals appear, proving higher stability against oxygen and water degradation than conventional lead-halide perovskites. Also, the original emission performance can be almost retained after 28-day storage with the spectral shape unchanged, as shown in Fig. 2g. The results obtained demonstrate good stability of the Cs3Cu2I5 films against heat, UV light, and environment oxygen/moisture, and are also evidence of a reliable Cs3Cu2I5 photoactive material compatible for practical applications.

image file: c9mh01371g-f2.tif
Fig. 2 (a–c) Three heating/cooling cycling PL measurements of the Cs3Cu2I5 films. (d) The statistically integrated PL intensity of the Cs3Cu2I5 films at two representative temperature points (293 and 373 K) over ten cycling measurements. The insets show the corresponding photographs of the films after different heating/cooling cycles under 265 nm light illumination. (e) Photostability test of the as-grown Cs3Cu2I5 films under continuous UV light irradiation (265 nm, 3.0 mW cm−2). (f) XRD, and (g) PL intensity evolution of the Cs3Cu2I5 films after different storage periods in ambient air.

Fig. 3a shows the schematic structure of the Cs3Cu2I5/GaN heterojunction photodetector, and the detailed preparation procedures can be found in the Experimental section. Based on the UV-visible absorption spectra and the UV photoelectron spectroscopy (UPS) data (Fig. S9, ESI), the energy band alignment of the Cs3Cu2I5/GaN heterojunction was drawn. Considering the fact that the Cs3Cu2I5 films exhibit weak p-type conductivity properties with a hole concentration of about 4.3 × 1015 cm−3 (Table S2, ESI), a p–n heterojunction can be established once the Cs3Cu2I5 films and GaN are in contact, thus a built-in electric field would be formed at the hetero-interface because of the typical type-II band alignment, as seen in Fig. 3b. Once the photo-generated electron–hole pairs diffuse into the hetero-interface, they can be separated in opposite directions by the built-in electric field without an external power supply. The holes will transport to the perovskite side, whereas the electrons will move in the opposite direction, generating photocurrent in the external circuit. Moreover, the built-in electric field is capable of reducing the electron–hole recombination and preventing the electron transport from GaN to Cs3Cu2I5, favoring a low reverse dark current and thus improving the detectivity. To support the above statement of charge carrier transfer at the hetero-interface, steady-state and transient-state PL spectra of Cs3Cu2I5 and the Cs3Cu2I5/GaN hybrid structure were therefore measured and put together for comparison. As shown in Fig. 3c, compared to the bare Cs3Cu2I5 films, the integrated PL intensity of the Cs3Cu2I5/GaN hybrid structure was reduced by ∼39%, reflecting transfer of photo-generated carriers from Cs3Cu2I5 to GaN. Meanwhile, the carrier lifetime of the Cs3Cu2I5/GaN hybrid structure is reduced, indicating that the PL decay channel for the heterojunction device is faster, and it is also direct evidence of electron transfer behavior. Therefore, the GaN layer can be regarded as the electron-quenching layer. To study the performance of the photodetector, we firstly examined the dependence of the photoresponse on the incident light wavelength at zero bias. An interesting phenomenon was observed in which the spectral response of the photodetectors depends strongly on the thickness of the Cs3Cu2I5 films. As shown in Fig. 3d, for the 0.2 μm-thickness Cs3Cu2I5 films (canary-yellow line), a broad spectral response ranging from 200 to 370 nm was obtained, coinciding with the absorption of the Cs3Cu2I5/GaN hybrid structure (Fig. S4, ESI). By contrast, the responsivity in the short-wavelength range (200–300 nm) is relatively weaker than that in the long-wavelength range (300–370 nm), which may be due to the severe surface charge recombination (SR). For a better understanding of the proposed mechanisms, added experiments were carried out by increasing the thickness of the Cs3Cu2I5 films to different values. As seen in Fig. 3d, with the increase of the film thickness, the responsivity of the photodetectors in the short wavelength range starts to decrease, and finally disappeared into the baseline at a thickness of 1.2 μm, producing a desirable narrowband spectral response ranging from 300 to 370 nm. Note that this is the first report on a spectrum-selective UV photodetector based on lead-free perovskites as far as we know. The working mechanisms of the highly selective responsivity for the Cs3Cu2I5/GaN heterojunction can be explained by Fig. 3g. In detail, the photoactive layer (Cs3Cu2I5) of the photodetector can be divided into two regions: a depletion region (DR) near the GaN layer and a neutral region (NR) in the outer-layer near the Au electrode. As we all know, the absorption coefficient of the material increases as the incident wavelength decreases, and correspondingly the average light penetration depth is reduced. Therefore, for deep UV light excitation with a short penetration depth, a large amount of photo-generated carriers is generated in the NR close to the Cs3Cu2I5 top surface and far away from the DR. They would be extinguished by the severe SR rather than being collected by the electrodes because of the limited electron diffusion length (EDL). As a result, the photo-generated carriers under deep UV light excitation cannot reach the DR, and are not capable of contributing to the photocurrent, making the photodetector highly spectrum-selective. For UV light excitation, the photons can penetrate deeper into the Cs3Cu2I5 films because of the relatively smaller absorption coefficient and thus the photo-generated carriers are most likely to diffuse into the DR. With the aid of the built-in electric field (E) in the DR, the electron–hole pairs can be separated and collected rapidly. In this way, narrowband detection with the photon energy near the bandgap of Cs3Cu2I5 can be achieved. The above observations indicate the “filter” effect of the outer-layer of the Cs3Cu2I5 films. To depict the spectrum-selective feature of such photodetectors more clearly, bias-dependent responsivity spectra of the Cs3Cu2I5/GaN heterojunction photodetector were measured. As shown in Fig. 3e, one can see that a negative bias can induce an obvious increase of the responsivity because of the enhanced charge extraction efficiency. An additional observation is that, with the increase of the reverse bias voltage, the responsivity in the short-wavelength range (200–300 nm) increases much more rapidly than that in the long-wavelength range (300–370 nm). Therefore, one conclusion can be made that the heterojunction photodetector will lose the narrowband detection ability under reverse bias (above −5 V). The above observation can be well understood by the fact that the applied negative bias shares the same direction as the E of the Cs3Cu2I5/GaN heterojunction, giving rise to a higher electric field and a larger width of the DR. Thus, it is still possible that the photo-generated carriers under deep UV light excitation, even with a small light penetration depth, were quickly separated and collected in the time scale of their lifetime, contributing to the photocurrent. The above discussions on bias-dependent responsivity can be better unveiled by the energy band diagrams of the Cs3Cu2I5/GaN heterojunction under zero and negative bias, as shown Fig. 3f.

image file: c9mh01371g-f3.tif
Fig. 3 (a) Schematic illustration of the Cs3Cu2I5/GaN heterojunction device. (b) Energy band alignment of the Cs3Cu2I5/GaN heterojunction. (c) PL spectra and PL decay curves of Cs3Cu2I5 and the Cs3Cu2I5/GaN hybrid with a thickness of 0.2 μm. Response spectra of the photodetector with (d) different absorber thicknesses, and (e) different bias voltages of the photodetector. (f) Energy band schematics of the Cs3Cu2I5/GaN heterojunction under zero and negative bias. (g) The proposed mechanism for the surface-charge recombination assisted photodetector.

Based on the rationale and working mechanisms outlined above, we selected 1.2 μm-thickness Cs3Cu2I5 films to deliver a narrowband photodetector with a response between 300 and 370 nm. Fig. 4a shows the current–voltage (IV) curves of the heterojunction photodetector under 320 nm UV illumination with varying light intensities (1 μW cm−2–4.2 mW cm−2). From the curves, it can be observed that there is a remarkable rectification feature with a rectification ratio over 106 (±3 V) in the dark. Here, the rectification behavior was considered to result from the Cs3Cu2I5/GaN heterojunction because of the established Ohmic contacts of Au/Cs3Cu2I5/Au and In/GaN/In (Fig. S10, ESI). In addition, one can see that both the photovoltage and photocurrent increased gradually with increasing light intensity because of the increased population of photo-generated carriers. Typically, the current of the device at zero bias greatly increases from 9.03 × 10−13 A in the dark to 1.86 × 10−7 A under UV light (4.2 mW cm−2), and a short-circuit current (Isc) of 0.18 μA and an open-circuit voltage (Voc) of 1.07 V were obtained (Fig. 4b), manifesting a significant photovoltaic behavior and enabling the device to operate as a self-powered photodetector. Furthermore, the dependence of the photocurrent on the light intensity is shown in Fig. 4c. The data can be fitted well by the power law IP = α·Pβ, in which P is the light excitation power, α represents a proportional constant, and β is an empirical value that reflects the recombination of photoexcited carriers. The value of β is 0.93, close to that of an ideal photodetector (β = 1), which suggests a small recombination loss and the high quality of the Cs3Cu2I5/GaN heterojunction. Fig. 4d shows the temporal photocurrent response of the device with different light intensity in self-powered mode. Apparently, the photocurrent increases gradually with the excitation intensity, and the heterojunction photodetector can be switched reversibly between the on and off states, yielding a maximum current on/off ratio of 1.2 × 105 (4.2 mW cm−2). Even after 2000 cycles of operation, the photodetector can work well with high reproducibility and stability (Fig. S11, ESI). Besides, it is worth noting that the on/off ratio of the photodetector can still reach 50 even under low light irradiation of 1.2 μW cm−2, suggesting the capability to detect a weak signal. Based on the above data, the linear dynamic range (LDR) or photosensitivity linearity was estimated by using the expression54–56

image file: c9mh01371g-t1.tif(1)
in which Ip* represents the photocurrent obtained under 1.0 mW cm−2 light excitation, and Id is the dark current. At zero bias, the LDR value was determined to be ∼85 dB, better than the state-of-the-art commercial photodetectors.25 Also, other device parameters, including the responsivity (R) and specific detectivity (D*), of the heterojunction photodetector were calculated based on the following formulas
R = (IpId)/Popt (2)
image file: c9mh01371g-t2.tif(3)
image file: c9mh01371g-t3.tif(4)
in which Ip is the photocurrent, A represents the active area (0.008 cm2), and Popt is the intensity of incident light, f is the bandwidth (3500 Hz, as discussed later), NEP represents the noise equivalent power, and image file: c9mh01371g-t4.tif is the noise current obtained by employing a lock-in amplifier at varying frequencies. Fig. S12 (ESI) shows the curve of noise current versus frequency, and the noise level per unit bandwidth (1 Hz) was measured to be 0.53 pA Hz−1/2. Consequently, R and D* were calculated to be 0.28 A W−1 and 1.4 × 1012 Jones (1 Jones = 1 cm Hz1/2 W−1), respectively, under 1 μW cm−2 light illumination at zero bias, which are comparable to many previously reported perovskite-based heterojunction photodetectors, as summarized in Table 1. Fig. 4e plots the calculated values of R and D* versus light intensity. Due to the increased carrier recombination probability at higher light intensity, both values present a declining trend with increasing light intensity, consistent with other reports.57–59

image file: c9mh01371g-f4.tif
Fig. 4 (a) IV curves of the photodetector tested in the dark and under different light irradiation intensity (320 nm). (b) Enlarged IV curves showing the photovoltaic effect of the Cs3Cu2I5/GaN heterojunction. (c) Logarithmic plot of the photocurrent versus light irradiation power at a bias of 0 V. (d) Photocurrent response measured under 320 nm light excitation with varying intensity at zero bias. (e) Responsivity and specific detectivity of the photodetector versus light intensity. (f) Rising and falling edges for estimating the rise time (tr) and fall time (tf) of the photodetector.
Table 1 Summary of the performance of heterojunction photodetectors
Device structure Response range (nm) Lead-free (yes/no) Responsivity (A W−1) D* (×1012 Jones) On/off ratio Ref.
FTO/MA3Bi2I9/Ag 380–900 Yes 0.002 1.3 60
TiO2/MAPbI3/spiro-OMeTAD 300–800 No 0.55 1 104 61
ITO/CsPbBr3–ZnO NPs/Ag 400–530 No 0.0115 12.86 62
Graphene/(PEA)2SnI4/MoS2/graphene 400–700 Yes 0.121 0.008 63
In/GaN/Cs2AgBiBr6/Au 200–550 Yes 1.46 9.4 104 64
ITO/SnO2/Cs2AgBiBr6/Au 300–520 Yes 0.11 0.024 65
Au/spiro-OMeTAD/CsPbBr3/SnO2/ITO 300–550 No 0.172 4.8 1.3 × 105 66
CsPbX3/α-Si radial junction 300–700 No 0.054 2.1 × 103 67
Au/MoS2–CsPbBr3/Au 350–550 No 4.4 0.025 104 68
In/GaN–Cs3Cu2I5/Au 300–370 Yes 0.28 1.4 1.2 × 105 This work

Further, the response speed of the Cs3Cu2I5/GaN photodetector was measured to evaluate its capability to follow a varied optical signal. The schematic illustration is shown in Fig. S13a (ESI), in which a waveform generator was used to produce the pulsed light (1 Hz to 8 kHz), and an oscilloscope was used to record the change of photocurrent with time. Fig. S13b–d (ESI) display the photoresponse to pulsed light with different modulation frequencies, and the obtained data were normalized relative to the photoresponse tested under 1 Hz pulsed light. It can be seen that the photodetector can work with remarkable reproducibility in a wide frequency range. From the relative balance (ImaxImin)/Imax shown in Fig. S13e (ESI), a high 3dB frequency of ∼3500 Hz was obtained, indicating that the Cs3Cu2I5/GaN photodetector is able to monitor ultrafast optical signals. Further, the response speed of the device was assessed by analyzing the rising and falling edges of a single response cycle, as shown in Fig. 4f and Fig. S13f (ESI). The response speeds were calculated to be 95/130 μs at 800 Hz and 32/43 μs at 5 kHz. It should be mentioned that the speed achieved is comparable to or faster than many studies on perovskite photodetectors.61,62,67,68

As mentioned above, such lead-free Cs3Cu2I5 films demonstrated remarkable material stability against heat, UV light, and environment moisture/oxygen degradation (Fig. 2). For real application in the photodetection field, research on the working stability of such heterojunction devices is highly necessary and desired to be conducted. In the present case, we firstly examined the long-term operation stability of the unencapsulated device at four temperature points (293, 333, 353, and 373 K) in ambient air (35–50% humidity), in which a zero bias and an excitation intensity of 0.90 mW cm−2 were fixed and the photocurrent was recorded in real time. It can be seen in Fig. 5a that the photocurrent at 293 K shows an increasing trend over the whole period (12 h). This may be due to the “photoactivation” behavior as observed in previous studies.69 Generally, after continuous running for a long time, the temperature of the working area would increase. On the one hand, the raised temperature favors the separation of electrons and holes. On the other hand, the increased charge carrier density at high temperature could reduce the effective energy barrier height. Both factors favor easier carrier transport and contribute to an increase in photoconductivity. The same situation also happens at 333 K. As presented in Fig. 5b, the photocurrent displays an increasing trend in the first 7 h and then shows a reducing trend in the last 5 h. Therefore, we speculate that the negative role of temperature also exists at such a temperature level, and the undesired heating effect has to be considered at this moment. As the temperature was further increased to 353 and 373 K, obvious photocurrent decay of ∼8.9% and 10% occurred over 12 h continuous running, as shown in Fig. 5c and d. Even so, the device stability in our work is still better than that in other studies.70,71 Note that the photocurrent decay can be attributed to the added trap states induced by the heating effect, by which the photo-generated carriers could be trapped with a certain probability and cause the decay of the photocurrent. Further, we evaluated the thermal stability of the Cs3Cu2I5/GaN heterojunction photodetector over a longer running time, and the responsivity of the device was intermittently collected at 373 K in ambient air. As shown in Fig. 5e, a very slow declining rate of responsivity can be distinguished and about a 10% decay occurs after a 300 h aging test. The above results suggest that the photocurrent decay or photo-generated carrier trapping does not increase in proportion with the running time, or the heat resistance of such a heterojunction device enables one to balance the accumulation of the heating effect with time. Most likely, such a good heat resistance ability of the Cs3Cu2I5/GaN heterojunction benefits from a rational device design with an all-inorganic structure. In addition, it is worth noting that the photocurrent decay at high temperature is recoverable when the device is cooled down to room temperature (RT). Fig. 5f shows the temporal photoresponse curves of the device after running for 12 h at 293 and 373 K, respectively, and the reduced photocurrent at 373 K is consistent with the above results. Meanwhile, after the photodetector was cooled down to RT, the photocurrent could recover to 97% of the original level, demonstrating the recoverability and excellent temperature tolerance of the photodetector, and making the practical applications of such devices a real possibility. Since such a temperature level is not sufficient to cause a chemical reaction or phase transition in the Cs3Cu2I5 films, the structural and optical properties of the materials can be well preserved, as seen in Fig. S14 (ESI). Therefore, once the working device was naturally cooled down to RT, the trap states in the Cs3Cu2I5 films will reduce remarkably, but could not recover to the initial level. Moreover, we observed that the recoverability of the photocurrent depends strongly on the working temperature. As shown in Fig. S15a (ESI), for the aging test at 353 K, the photocurrent of the device can recover to 100% of the original level, while, for the aging test at 393 K (Fig. S15b, ESI), the photocurrent of the device can recover to 90% of the original level. The results suggest that the cooling down process (back to 293 K) could not remove the trap states absolutely, especially at higher temperature. According to the temperature-dependent photocurrent–time curves, the on/off photocurrent ratios at four temperature points were calculated and are summarized in Fig. 5g. Obviously, the on/off photocurrent ratio decreases with increasing temperature under the same test conditions and at 373 K the on/off photocurrent ratio is reduced by 16%, while, after the working device is naturally cooled down to 293 K, the on/off ratio could almost recover to its original value. More excitingly, the present device, without any encapsulation and protection, can work efficiently after one-month storage in the open air, and maintain almost the same photocurrent under the same test conditions (320 nm, 0.9 mW cm−2), as presented in Fig. 5h. The excellent storage stability achieved above is superior to the performance of organic–inorganic hybrid perovskite-based devices.70,72 The above observations suggest that the studied heterojunction photodetector has stable and reproducible characteristics, as well as good temperature resistance, rendering it potentially useful for the assembly of optoelectronic systems in the future.

image file: c9mh01371g-f5.tif
Fig. 5 Normalized photocurrent of the unencapsulated heterojunction photodetector versus running time at (a) 293 K, (b) 333 K, (c) 353 K, and (d) 373 K. The light intensity is fixed at 0.9 mW cm−2. (e) Responsivity evolution of the heterojunction photodetector monitored at 373 K in ambient air. (f) Temporal photoresponse curves of the device after aging for 12 h at 293 and 373 K, and after naturally cooling to 293 K. (g) One thermal cycling measurement of the device revealing its variation of the on/off photocurrent ratio at different temperatures. (h) Long-term storage stability of the photodetector in ambient air. The last eight cycles are the photoresponse after storage for 1 month.

To assess the capability of the photodetector for recording image information, which is very important in optical fields such as fax machines and cameras, a UV imaging system was constructed by using the heterojunction photodetector as sensing pixels. Fig. 6a displays the schematic illustration of the measurement setup, in which a photo-mask with a hollow pattern was put between the laser illumination and the device, and the photo-mask can be moved in horizontal and vertical direction (xy direction) by a 2D translation stage controlled by a software-programmed computer. In the working process, the laser beam passed through the photo-mask and irradiated the photodetector; the current signals from the heterojunction device and the corresponding position coordinates of the photo-mask were recorded in real time by a lock-in amplifier and a computer simultaneously, which could be transformed to a current mapping pattern. Note that the measured dark current and photocurrent respectively represent the background noise level and light intensity level at each pixel, inducing the dark and bright contrast in the imaging patterns. Fig. 6b shows the high-resolution imaging results under 325 nm light illumination, including the abbreviation of the hollow ultraviolet logo “UV” and a solid symbol of a deer. It can be seen that the two images were clearly resolved and well consistent with the photo-mask in shape, implying the high fidelity features of the imaging system and the reliable imaging function of the device.

image file: c9mh01371g-f6.tif
Fig. 6 (a) Schematic illustration of the experimental setup for the image sensing application of the studied heterojunction photodetector. (b) The corresponding current mapping of deer and UV patterns.


In conclusion, we have demonstrated a highly-stable and spectrum-selective self-powered UV photodetector based on solution-processed lead-free Cs3Cu2I5 films by using the CCN concept. The device only responds to photons with an energy in a narrow “window” of 300–370 nm, and we investigated its working mechanism through the dependence of the spectral response on the absorber thickness and bias voltage. In addition to excellent photodetection performance, the proposed photodetector without encapsulation also demonstrates good operation stability in ambient air. Typically, it can endure a high operation temperature of 373 K for 12 h continuous working; even after a month storage in the open air, the photodetector can function properly and its photodetection capability can almost be retained, which indicates a good temperature tolerance and desired compatibility for practical applications under harsh conditions. Further, with the help of an optically-assisted displacement platform, the hybrid heterojunction detector performed as the sensing pixels in a UV imaging system. The results obtained above may promise stable and spectrum-selective UV photodetection based on such lead-free Cs3Cu2I5 materials, and the simple device structure will show great potential for hybrid optoelectronic systems.

Experimental section

In this work, commercially available n-GaN/Al2O3 templates were used as substrates, which were ultrasonically cleaned with ethanol, acetone, and deionized water for 10 min, respectively. Next, the template was treated with Ar gas plasma for 15 min. The GaN layer was ∼3.0 μm in thickness and showed n-type conductivity with a carrier concentration of 5.9 × 1018 cm−3 and a carrier mobility of 352 cm2 V−1 s−1. The Cs3Cu2I5 films are prepared using solution processing via a one-step spin-coating method. In detail, the precursor solution mixed with 0.5 M CsI and 0.5 M CuI in dimethyl sulfoxide (completely dissolved at 75 °C) was spin-coated on the prepared substrate at 500 rpm for 10 s and then 2000 rpm for 30 s. The thickness of the films can be controlled by adjusting the speed and time of the spin-coating process. Following that, the sample was annealed at 100 °C for 1 h under N2 conditions to evaporate the solvent and enhance the crystallization.

For device preparation, an In electrode (∼100 nm) on GaN and a Au electrode (∼20 nm) on the Cs3Cu2I5 films were prepared by thermal evaporation by using a shadow mask and the heterojunction was established at the area where Cs3Cu2I5 and GaN overlapped.

The morphology of the prepared Cs3Cu2I5 films was tested by field-emission SEM (Jeol-7500F, 15 keV). XRD was conducted using a Panalytical X’Pert Pro. The PL spectra were recorded using a spectrofluorometer (Horiba; Fluorolog-3), and the transient-state PL measurement was conducted using a pulsed NanoLED (Horiba; 280 nm). The absorption spectra were measured using a Shimadzu UV-3150 spectrophotometer. The electrical properties of the Cs3Cu2I5 films and GaN layer were measured through Hall effect measurements (ACCENT HL5500PC, UK). The optoelectrical and electrical measurements were conducted by using a testing system that includes a monochromator, a digital SourceMeter (Keithley 2636B), an optical chopper (SRS, SR540), light sources, and an oscilloscope (Tektronix, DPO2012B) in air.

Conflicts of interest

There are no conflicts to declare.


This work was supported by the National Natural Science Foundation of China (No. 11774318 and 11604302), the Key Scientific Research Projects of Higher Education in Henan Province (18A140007), and the Support Program for Scientific and Technological Innovation Talents of Higher Education in Henan Province (19HASTIT017).

Notes and references

  1. B. Aufiero, H. Talwar, C. Young, M. Krishnan, J. Hatfield, H. Lee, H. Wong, I. Hamzavi and G. Murakawa, J. Photochem. Photobiol., B, 2006, 82, 132 CrossRef CAS PubMed.
  2. G. Wilson and R. DeFreez, Proc. SPIE, 2004, 5416, 157 CrossRef.
  3. F. Guo, B. Yang, Y. Yuan, Z. Xiao, Q. Dong, Y. Bi and J. Huang, Nat. Nanotechnol., 2012, 7, 798 CrossRef CAS PubMed.
  4. S. Nishiwaki, T. Nakamura, M. Hiramoto, T. Fujii and M. Suzuki, Nat. Photonics, 2013, 7, 240 CrossRef.
  5. H. Chen, K. Liu, L. Hu, A. Al-Ghamdi and X. Fang, Mater. Today, 2015, 18, 493 CrossRef CAS.
  6. T. Xu, Y.-K. Wu, X. Luo and L. Guo, Nat. Commun., 2010, 1, 1 Search PubMed.
  7. S. Yokogawa, S. Burgos and H. Atwater, Nano Lett., 2012, 12, 4349 CrossRef CAS PubMed.
  8. Z. Li, S. Butun and K. Aydin, ACS Photonics, 2015, 2, 183 CrossRef CAS.
  9. E. Cicek, R. McClintock, C. Cho and B. Rahnema, Appl. Phys. Lett., 2013, 103, 191108 CrossRef.
  10. Y. Fang, Q. Dong, Y. Shao, Y. Yuan and J. Huang, Nat. Photonics, 2015, 9, 679 CrossRef CAS.
  11. A. Armin, R. Jansen-van Vuuren, N. Kopidakis and P. Burn, Nat. Commun., 2015, 6, 6343 CrossRef CAS PubMed.
  12. P. Ni, C. Shan, S. Wang, X. Liu and D. Shen, J. Mater. Chem. C, 2013, 1, 4445 RSC.
  13. Y. Zhong, T. Sisto, B. Zhang, K. Miyata, X. Zhu, M. Steigerwald and F. Ng, J. Am. Chem. Soc., 2017, 139, 5644 CrossRef CAS PubMed.
  14. A. Solanki, S. Li, H. Park and K. Crozier, ACS Photonics, 2018, 5, 520 CrossRef CAS.
  15. Q. Xiong, F. Chowdhury and X. Wang, IEEE J. Sel. Top. Quantum Electron., 2018, 24, 1900406 Search PubMed.
  16. Z. Shi, Y. Li, Y. Zhang, Y. Chen, X. Li, D. Wu, T. Xu, C. Shan and G. Du, Nano Lett., 2017, 17, 313 CrossRef CAS PubMed.
  17. N. Jeon, H. Na, E. Jung, T. Yang, Y. Lee, G. Kim, H. Shin, S. Seok, J. Lee and J. Seo, Nat. Energy, 2018, 3, 682 CrossRef CAS.
  18. Q. Chen, H. Zhou, Z. Hong, S. Luo, H. Duan, H. Wang, Y. Liu, G. Li and Y. Yang, J. Am. Chem. Soc., 2014, 136, 622 CrossRef CAS PubMed.
  19. K. Ding, Y. Fang, S. Dong, H. Chen, B. Luo, K. Jiang, H. Gu, L. Fan, S. Liu, B. Hu and L. Wang, Adv. Opt. Mater., 2018, 6, 1800347 CrossRef.
  20. T. Chiba, Y. Hayashi, H. Ebe, K. Hoshi, J. Sato, S. Sato, Y.-J. Pu, S. Ohisa and J. Kido, Nat. Photonics, 2018, 12, 681 CrossRef CAS.
  21. C. Sun, Y. Zhang, C. Ruan, C. Yin, X. Wang, Y. Wang and W. Yu, Adv. Mater., 2016, 28, 10088 CrossRef CAS PubMed.
  22. F. Zhang, Z. Shi, Z. Ma, Y. Li, S. Li, D. Wu, T. Xu, X. Li, C. Shan and G. Du, Nanoscale, 2018, 10, 20131 RSC.
  23. S. Stranks, G. Eperon, G. Grancini, C. Menelaou, M. Alcocer, T. Leijtens, A. Herz, L. Petrozza and H. Snaith, Science, 2013, 342, 341 CrossRef CAS PubMed.
  24. Y. Zhang, W. X. Xu, X. J. Xu, J. Cai, W. Yang and X. S. Fang, J. Phys. Chem. Lett., 2019, 10, 836 CrossRef CAS PubMed.
  25. Z. Shi, Y. Li, S. Li, X. Li, D. Wu, T. Xu, Y. Tian, Y. Chen, Y. Zhang, B. Zhang, C. Shan and G. Du, Adv. Funct. Mater., 2018, 28, 1707031 CrossRef.
  26. B. Yang, F. Zhang, J. Chen, S. Yang, X. Xia, T. Pullerits, W. Deng and K. Han, Adv. Mater., 2017, 29, 1703758 CrossRef PubMed.
  27. X. Li, D. Yu, J. Chen, Y. Wang, F. Cao, Y. Wei, Y. Wu, L. Wang, Y. Zhu, Z. Sun, J. Ji, Y. Shen, H. Sun and H. Zeng, ACS Nano, 2017, 11, 2015 CrossRef CAS PubMed.
  28. Z. Zheng, F. Zhuge, Y. Wang, J. Zhang, L. Gan, X. Zhou, H. Li and T. Zhai, Adv. Funct. Mater., 2017, 27, 1703115 CrossRef.
  29. W. Deng, L. Huang, X. Xu, X. Zhang, X. Jin, S. Lee and J. Jie, Nano Lett., 2017, 17, 2482 CrossRef CAS PubMed.
  30. Y. Li, Z. Shi, L. Lei, Z. Ma, F. Zhang, S. Li, D. Wu, T. T. Xu, X. Li, C. Shan and G. Du, ACS Photonics, 2018, 5, 2524 CrossRef CAS.
  31. Q. Lin, A. Armin, P. Burn and P. Meredith, Nat. Photonics, 2015, 9, 687 CrossRef CAS.
  32. H. Rao, W. Li, B. Chen, D. Kuang and C. Su, Adv. Mater., 2017, 29, 1602639 CrossRef PubMed.
  33. L. Li, Y. Deng, C. Bao, Y. Fang, H. Wei, S. Tang, F. Zhang and J. Huang, Adv. Opt. Mater., 2017, 5, 1700672 CrossRef.
  34. J. Li, J. Wang, J. Ma, H. Shen, L. Li, X. Duan and D. Li, Nat. Commun., 2019, 10, 806 CrossRef PubMed.
  35. J. Wang, J. Li, S. Lan, C. Fang, H. Shen, Q. Xiong and D. Li, ACS Nano, 2019, 13, 5473 CrossRef CAS PubMed.
  36. J. Xue, Z. Zhu, X. Xu, Y. Gu, S. Wang, L. Xu, Y. Zou, J. Song, H. Zeng and Q. Chen, Nano Lett., 2018, 18, 7628 CrossRef CAS PubMed.
  37. K. Lu, Y. Lei, R. Qi, J. Liu, X. Yang, Z. Jia, R. Liu, Y. Xiang and Z. Zheng, J. Mater. Chem. A, 2017, 5, 25211 RSC.
  38. C. Bi, S. Wang, Q. Li, S. V. Kershaw, J. Tian and A. L. Rogach, J. Phys. Chem. Lett., 2019, 10, 943 CrossRef CAS PubMed.
  39. R. Roccanova, A. Yangui, H. Nhalil, H. Shi, M. Du and B. Saparov, ACS Appl. Electron. Mater., 2019, 1, 269 CrossRef CAS.
  40. T. Li, X. Mo, C. Peng, Q. Lu, C. Qi, X. Tao, Y. Ouyang and Y. Zhou, Chem. Commun., 2019, 55, 4554 RSC.
  41. R. P. Vasquez, Surf. Sci. Spectra, 1993, 2, 149 CrossRef CAS.
  42. T. Jun, K. Sim, S. Iimura, M. Sasase, H. Kamioka, J. Kim and H. Hosono, Adv. Mater., 2018, 30, 1804547 CrossRef PubMed.
  43. Z. Yuan, C. Zhou, Y. Tian, Y. Shu, J. Messier, J. Wang, L. Van de Burgt, K. Kountouriotis, Y. Xin, E. Holt, K. Schanze, R. Clark, T. Siegrist and B. Ma, Nat. Commun., 2017, 8, 14051 CrossRef CAS PubMed.
  44. C. Zhou, Y. Tian, M. Wang, A. Rose, T. Besara, N. Doyle, Z. Yuan, J. Wang, R. Clark, Y. Hu, T. Siegrist, S. Lin and B. Ma, Angew. Chem., 2017, 129, 9146 CrossRef.
  45. M. Halcrow, Chem. Soc. Rev., 2013, 42, 1784 RSC.
  46. Y. Kim, J. Kim, J. Park, K. Park and Y. Lee, Chem. Commun., 2017, 3, 2858 RSC.
  47. M. Jones, S. S. Lo and G. D. Scholes, J. Phys. Chem. C, 2009, 113, 18632 CrossRef CAS.
  48. X. Wu, M. T. Trinh, D. Niesner, H. M. Zhu, Z. Norman, J. S. Owen, O. Yaffe, B. J. Kudisch and X. Y. Zhu, J. Am. Chem. Soc., 2015, 137, 2089 CrossRef CAS PubMed.
  49. Z. Shi, S. Li, Y. Li, H. Ji, X. Li, D. Wu, T. Xu, Y. Chen, Y. Tian, Y. Zhang, C. Shan and G. Du, ACS Nano, 2018, 12, 1462 CrossRef CAS PubMed.
  50. S. Wei, Y. Yang, X. Kang, L. Wang, L. Huang and D. Pan, Chem. Commun., 2016, 52, 7265 RSC.
  51. H. Tsai, W. Nie, J. Blancon, C. Stoumpos, C. Soe, J. Yoo, J. Crochet, S. Tretiak, J. Even and A. Sadhanala, Adv. Mater., 2018, 30, 1704217 CrossRef PubMed.
  52. C. C. Stoumpos, C. D. Malliakas, J. A. Peters, Z. F. Liu, M. Sebastian, J. Im, T. Chasapis and A. C. Wibowo, Cryst. Growth Des., 2013, 13, 2722 CrossRef CAS.
  53. H. Cho, Y. Kim, C. Wolf, H. Lee and T. Lee, Adv. Mater., 2018, 30, 1704587 CrossRef PubMed.
  54. X. Gong, Z. Tong, Y. Xia, W. Cai, J. Moon, Y. Cao, G. Yu, C. Shieh, B. Nilsson and A. Heeger, Science, 2009, 325, 1665 CrossRef CAS PubMed.
  55. Y. Liu, Y. Zhang, K. Zhao, Z. Yang, J. Feng, X. Zhang, K. Wang, L. Meng, H. Ye, M. Li and S. Liu, Adv. Mater., 2018, 30, 1707314 CrossRef PubMed.
  56. Y. Li, Z. Shi, X. Li and C. Shan, Chin. Phys. B, 2019, 28, 017803 CrossRef CAS.
  57. C. Wu, Z. Wang, L. Liang, T. Gui, W. Zhong, R. Du, C. Xie, L. Wang and L. Luo, Small, 2019, 15, 1900730 CrossRef PubMed.
  58. B. Yang, F. Zhang, J. Chen, S. Yang, X. Xia, T. Pullerits, W. Deng and K. Han, Adv. Mater., 2017, 29, 1703758 CrossRef PubMed.
  59. Y. Guo, C. Liu, H. Tanaka and E. Nakamura, J. Phys. Chem. Lett., 2015, 6, 535 CrossRef CAS PubMed.
  60. A. Hussain, A. Rana and M. Ranjan, Nanoscale, 2019, 11, 1217 RSC.
  61. G. Tong, X. Geng, Y. Yu, L. Yu, J. Xu, Y. Jiang, Y. Sheng, Y. Shi and K. Chen, RSC Adv., 2017, 7, 18224 RSC.
  62. C. Li, C. Han, Y. Zhang, Z. Zang, M. Wang, X. Tang and J. Du, Sol. Energy Mater. Sol. Cells, 2017, 172, 341 CrossRef CAS.
  63. C. Fang, H. Wang, Z. Shen, H. Shen, S. Wang, J. Ma, J. Wang, H. Luo and D. Li, ACS Appl. Mater. Interfaces, 2019, 11, 8419 CrossRef CAS PubMed.
  64. Y. Li, Z. Shi, L. Lei, S. Li, D. Yang, D. Wu, T. Xu, Y. Zhi, Y. Lu, Y. Wang, L. Zhang, X. Li, Y. Zhang, G. Du and C. Shan, Adv. Mater. Interfaces, 2019, 6, 1900188 CrossRef.
  65. C. Wu, B. Du, W. Luo, Y. Liu, T. Li, D. Wang, X. Guo, H. Ting, Z. Fang, S. Wang, Z. Chen, Y. Chen and L. Xiao, Adv. Opt. Mater., 2018, 6, 1800811 CrossRef.
  66. H. Zhou, J. Zeng, Z. Song, C. R. Grice, C. Chen, Z. Song, D. Zhao, H. Wang and Y. Yan, J. Phys. Chem. Lett., 2018, 9, 2043 CrossRef CAS PubMed.
  67. J. Lu, X. Sheng, G. Tong, Z. Yu, X. Sun, L. Yu, X. Xu, J. Wang, J. Xu, Y. Shi and K. Chen, Adv. Mater., 2017, 29, 1700400 CrossRef PubMed.
  68. X. Song, X. Liu, D. Yu, C. Huo, J. Ji, X. Li, S. Zhang, Y. Zou, G. Zhu, Y. Wang, M. Wu, A. Xie and H. Zeng, ACS Appl. Mater. Interfaces, 2018, 10, 2801 CrossRef CAS PubMed.
  69. M. Leng, Y. Yang, K. Zeng, Z. Chen, Z. Tan, S. Li, J. Li, B. Xu, D. Li, M. Hautzinger, Y. Fu, T. Zhai, L. Xu, G. Niu and J. Tang, Adv. Funct. Mater., 2018, 28, 1704446 CrossRef.
  70. K. Aitola, K. Domanski, J. Baena, K. Sveinbjörnsson, M. Saliba, A. Abate, M. Grätzel, E. Kauppinen, E. Johansson, W. Tress, A. Hagfeldt and G. Boschloo, Adv. Mater., 2017, 29, 1606398 CrossRef PubMed.
  71. Y. Li, Z. Shi, S. Li, L. Lei, H. Ji, D. Wu, T. Xu, Y. Tian and X. Li, J. Mater. Chem. C, 2017, 5, 8355 RSC.
  72. J. Yang, Q. Hong, Z. Yuan, R. Xu, X. Guo, S. Xiong, X. Liu, S. Braun, Y. Li, J. Tang, C. Duan, M. Fahlman and Q. Bao, Adv. Opt. Mater., 2018, 6, 1800262 CrossRef.


Electronic supplementary information (ESI) available: EDS, XPS, and transmission spectra of the Cs3Cu2I5 films, absorption of the Cs3Cu2I5 films, GaN layer, and the Cs3Cu2I5/GaN hybrid, temperature-dependent PL decay curves, heating/cooling cycling PL measurements, thermogravimetric data, photostability, and UPS of the Cs3Cu2I5 films, confirmation of the Ohmic contacts between the semiconductors and metal, stability and noise current of the photodetector, high-frequency test of the device at zero bias, morphology, structural, and optical properties of the Cs3Cu2I5 films before and after heating treatment, confirmation of the recoverability of the photocurrent after the aging test, Table S1 fitting parameters of the PL decay curves, Table S2 electrical properties of the Cs3Cu2I5 films and GaN. See DOI: 10.1039/c9mh01371g

This journal is © The Royal Society of Chemistry 2019