Enhanced Spectral Response of ZnO-Nanorod-Array-Based Ultraviolet Photodetectors by Alloying Non-Isovalent Cu–O with CuAlO2 P-Type Layer

CuAlO2 was synthesized by a hydrothermal method, in which the Cu–O dimers were incorporated by simply altering the ratio of the reactants and the temperature. The incorporation process increases the grain size in CuAlO2, and modulates the work function and binding energies for CuAlO2 due to the partial substitution of Cu+ 3d10 with Cu2+ 3d9 orbitals in the valence band maximum by alloying non-isovalent Cu–O with a CuAlO2 host. Based on the ZnO nanorod arrays (NRs) ultraviolet photodetector, CuAlO2/Cu–O fabricated by the low-cost drop-coating method was used as the p-type hole transport layer. The incorporation of the Cu–O clusters into CuAlO2 lattice to enhance the conductivity of CuAlO2 is an effective way for improving ZnO NRs/CuAlO2 device performance. The photodetectors exhibit significant diode behavior, with a rectification ratio approaching 30 at ±1 V, and a dark saturation current density 0.81 mA cm−2. The responsivity of the ZnO-NRs-based UV photodetector increases from 13.2 to 91.3 mA/W at 0 V bias, with an increase in the detectivity from 2.35 × 1010 to 1.71 × 1011 Jones. Furthermore, the ZnO NRs/[CuAlO2/Cu–O] photodetector exhibits a maximum responsivity of 5002 mA/W at 1.5 V bias under 375 nm UV illumination.


Introduction
Ultraviolet (UV) photodetectors have come to be used effectively in numerous fields, such as detection of missiles, solar UV monitoring, and ozone hole detection [1]. In the process of detector development, one-dimensional (1D) nanomaterials have attracted considerable attention, owing to the excellent photoelectric properties of their ultrahigh intrinsic photoelectric, quantum confinement effect, multiple light confinement, and subwavelength size effects [2]. Due to the large surface-to-volume ratio and 1D nanostructure, the 1D nanomaterial can serve as a carrier transport channel. Thus, the transport time can be shortened and the lifetime of photogenerated holes and electrons can be extended, which leads to high response and internal photoconductive gain [2,3]. For example, SnO 2 -and ZnO-nanowire-based devices exhibit higher transport rate than their nanoparticle-based counterparts [4]. Therefore, some 1D nanomaterials, such as SiC, NiO, ZnSe, AlGaN, and ZnO, have become commercially available [5]. Among them, ZnO is one of the essential wide-bandgap semiconductors, with a direct bandgap of 3.37 eV and excitation binding energy of 60 meV at room temperature, as well as a non-centrosymmetric crystal structure [6]. On account of its unique properties such as environmental friendliness, abundant raw materials, and chemical stability, ZnO has been considered as a promising photoelectronic material [7]. Until now, it has been synthesized by a wide variety of chemical and physical techniques, such as chemical vapor deposition [8], pulsed laser deposition [9], molecular beam epitaxy [10], and the hydrothermal method [11]. Due to the low cost, controlled shape, and large-area compatibility, the hydrothermal method is extremely suitable for the fabrication of high-quality ZnO NRs [12,13]. In terms of applications, ZnO NRs have, until characteristics. Also, this work provides a new pathway towards high performance UV photodetectors with the delafossite CuMO 2 materials.

Fabrication of ZnO/CuAlO 2 Heterojunction
A hydrothermal method was used to prepare the ZnO NRs on fluorine-doped tin dioxide (FTO) glass substrate. In detail, the FTO glass substrate was sonicated with deionized water. ZnO precursor solution was prepared by dissolving appropriate zinc acetate dihydrate (98.0-101.0%, Alfa Aesar (China) Chemicals Co., Ltd., Shanghai, China) and 0.035 mol diethanolamine (99%, Alfa Aesar (China) Chemicals Co., Ltd., Shanghai, China) in 50 mL isopropanol (≥99.7%, Shanghai Titan Scientific Co., Ltd., Shanghai, China). After stirring evenly, the transparent precursor solution was deposited by spin-coating onto the FTO glass substrate and annealed in air atmosphere at 300 • C for an hour. Then the substrates glass covered with a seed layer was dipped into a 150 mL Teflon-lined steel autoclave, consisting of equimolar solution hexamethylenetetramine (≥99.0%, Chengdu Kelong Chemical Co., Ltd., Chengdu, China) and zinc nitrate hexahydrate (≥99.5%, Chengdu Kelong Chemical Co., Ltd., Chengdu, China). The autoclave was heated at 85 • C for four hours. Finally, the grown nanorods were annealed to obtain better crystallization properties.
The CuAlO 2 was also fabricated via a hydrothermal process. Firstly, a 3:1 mole ratio of cuprous chloride (≥97.0%, Chengdu Kelong Chemical Co., Ltd., Chengdu, China) and sodium meta-aluminate (≥41.0%, Shanghai Titan Scientific Co., Ltd., Shanghai, China) was dissolved in 50 mL deionized water, followed by 0.6 g sodium hydroxide (≥98.0%, Chengdu Kelong Chemical Co., Ltd., Chengdu, China). After stirring for more than an hour at room temperature, the well-mixed precursor solution was sealed in a 100 mL Hastelloy autoclave. The autoclave was first heated in an oven at 400 • C for four hours, and then cooled to room temperature naturally. The precipitate was washed with diluted hydrochloric acid (36.0-38.0%, Chengdu Kelong Chemical Co., Ltd., Chengdu, China) solution (1 mol/L), diluted ammonia (25.0-28.0%, Chengdu Kelong Chemical Co., Ltd., Chengdu, China) solution (1 mol/L), deionized water, and absolute alcohol in sequence several times. After drying in air at 60 • C for five hours, the samples were stored in a low-temperature and dry environment. Meanwhile, 1.8:1 mole ratio of cuprous chloride and sodium meta-aluminate was dissolved in 50 mL deionized water, followed by 0.6 g sodium hydroxide in order to obtain alloying non-isovalent Cu-O with CuAlO 2 . Using the same synthesis steps, the precursor solution was stirred evenly and loaded into a Hastelloy autoclave. For another, the autoclave was placed into the oven at 360 • C for four hours. Also, the powder was washed in the same way. CuAlO 2 and CuAlO 2 /Cu-O crystals were dissolved in ethanol, which was placed in an ultrasonic for more than four hours to obtain stable and uniform dispersion. This dispersion was dropped on the preheated ZnO NRs and annealed in air at 300 • C for 20 min. Au was evaporated as an anode via electron beam evaporation. Finally, the device was scribed to an area of 0.09 cm 2 to form a vertical structure of FTO/ZnO NRs/CuAlO 2 (or [CuAlO 2 /Cu-O])/Au.

Characterization
The structure of prepared samples was measured using X-ray diffraction ( XRD-6100, Shimazu, Tokyo, Japan) using Cu Ka radiation (λ = 1.5406 Å, 30 kV and 10 mA). The material morphologies were characterized by the scanning electron microscopy (SEM, ZEISS Gemini 300, ZEISS, Oberkochen, Germany) equipped with an X-ray energy-dispersive spectrometer (EDS). (The acceleration voltage is 3 kV when the topography is photographed, and the acceleration voltage is 15 kV when the energy spectrum mapping is photographed. The probe beam is 3 pA-20 nA, and the stability is better than 0.2%/h.) X-ray photoelectron spectroscopy (XPS, Thermo Scientific K-Alpha, Thermo Fisher Scientific Inc., Waltham, MA, USA) was used to investigate the chemical state of Cu element, which were calibrated based on C 1s peak 284.8 eV (The spot size is 400 µm, the working voltage is 12 kV, and the filament current is 6 mA. The full-spectrum scanning energy is 150 eV, and the step size is 1 eV. The narrow-band scanning pass energy is 50 eV, and the step size is 0.1 eV). The work functions of the samples were measured by the ultraviolet photoelectron spectroscopy (UPS, Thermo ESCALAB XI+, Thermo Fisher Scientific Inc., Waltham, MA, USA). The optical properties of the sample were conducted in the wavelength range from 250 to 800 nm with PerkinElmer Lambda 950 UV-visible spectrometer (PE 950, PerkinElmer Inc., Waltham, MA, USA). From the manufacturer's data analysis, the measurement error associated with the transmission value is estimated to be about 0.25%. Electrochemical impedance spectra (EIS) were carried out on electrochemical workstation (CHI660E, CH Instruments, Inc., Shanghai, China) under dark conditions in the frequency range from 10 6 to 10 −1 Hz. The external and apparent quantum efficiency (EQE and AQE, respectively) in the range of 320-440 nm was measured by QEX10 (PV Measurements, Inc, Washington, USA) at room temperature. It was calibrated with a standard Si solar cell before measuring. The light source was a xenon lamp with a power of 65 mW cm −2 and the frequency of the chopper was 120 Hz. The light was incident from the glass surface, and the spot diameter on the sample was 2 mm. The current-voltage (I-V) and time-resolved current curves were collected through Keithley 2400 digital source meter. The effective illumination area of the device was 3.14 mm 2 .  Figure 1b). This optical trapping structure of nanorods can improve the absorption of ultraviolet light [42]. Figure 1c shows the SEM and EDS results of the pure CuAlO 2 sample, which is uniform, compact, and smooth with particle sizes of 200-400 nm. This polygonal grain shape suggests that the treatment process, such as sonication, does not alter its crystallinity [39]. The illustration in Figure 1c reveals that the elemental percentages of Cu (~24.60 at.%), Al (~26.85 at.%), and O (~48.55 at.%) match well with the stoichiometric proportion of pure CuAlO 2 . In Figure 1d, Cu-O incorporation into CuAlO 2 obviously enhances the grain size (400-600 nm). The CuAlO 2 /Cu-O is rough and flaky with randomly grown minor crystallites. This flaky particle is beneficial for the improvement of the conductivity of CuAlO 2 , as reported by other researchers [43]. It can be found that the atomic ratio of Al decreases with the incorporation of Cu-O clusters. Elemental mapping of CuAlO 2 /Cu-O (Figure 1e-g) reveals that the film is mainly composed of Cu, Al, and O, which distribute homogeneously in the sample. Figure 2a shows the XRD pattern of ZnO nanorods, which is consistent with the hexagonal wurtzite ZnO (JCPDS card 36-1451) [44]. The (002) peak with the strongest intensity is located at 34.6 • , indicating that the ZnO NRs have high c-axis growth orientation. It is observed that other weak peaks at 43.7 • and 60.1 • have the orientations along (102) and (103), respectively. There are no impurity peaks observed in the ZnO NRs. Also in Figure     The optical transmittance for ZnO NRs and CuAlO2 films was examined by using UV-vis spectroscopy from 250 to 800 nm, as shown in Figure 2b,c. The illustration of Figure 2b,c shows the (ahν) 2 versus hν of ZnO NRs and CuAlO2 films. CuAlO2 has more than 70% transmittance in the visible region and a significant absorption at about 340 nm. The energy band gap of CuAlO2 films (~3.11 eV) can be determined using the Tauc's The optical transmittance for ZnO NRs and CuAlO 2 films was examined by using UVvis spectroscopy from 250 to 800 nm, as shown in Figure 2b,c. The illustration of Figure 2b,c shows the (ahν) 2 versus hν of ZnO NRs and CuAlO 2 films. CuAlO 2 has more than 70% transmittance in the visible region and a significant absorption at about 340 nm. The energy band gap of CuAlO 2 films (~3.11 eV) can be determined using the Tauc's formular [46]. The organic residues or high grain boundary scattering can lead to significant optical absorption below the bandgap energy [47]. ZnO has more than 60% transmission in the visible region, and ZnO NRs can mainly absorb UV light. The calculated band gap for ZnO NRs is about 3.25 eV, which is lower than that for bulk ZnO (~3.37 eV) at room temperature. It is because the defect density increases with the increase in specific surface area, which leads to the formation of defect levels in the band gap of ZnO [6].

Results and Discussion
To verify the hybridization of Cu-O dimers with CuAlO 2 host, XPS was used to investigate the chemical state of the compound, which witnesses the variation in Cu valence states (Figure 2d-f). The spectrum of CuAlO 2 (Figure 2d) manifests that the Cu 2p 3/2 and Cu 2p 1/2 related peaks are found at 932.0 and 951.7 eV, respectively, which is consistent with prior work [48][49][50]. The monovalent state of copper (Cu + ) can be confirmed through this phenomenon, as there is no shake-up line of Cu 2p 3/2 in the spectrum. After the hybridization of Cu-O dimers (Figure 2e,f), the binding energies of Cu 2p 3/2 and Cu 2p 1/2 of CuAlO 2 slightly shift to 932.5 eV and 952.3 eV, respectively [28]. This is because the strong two-dimensional confinement of Cu-O bonds restrains the electronic structure [51]. The two weak peaks appear at 934.2 eV and 954.1 eV, corresponding to the binding energies of Cu 2p 3/2 and Cu 2p 1/2 , respectively, for CuO [52]. Furthermore, the difference between these two energy levels is close to 20 eV and the core level spectrum of Cu 2+ compounds has a strong satellite feature (shake-up satellite) caused by 3d vacancy states in the electronic configurations of atoms [53,54]. This feature has been widely used to distinguish Cu + from Cu 2+ in various copper compounds. Quantitative analysis of the XPS spectra provides Cu 2+ /Cu + atomic ratios of 1.2 for CuAlO 2 /Cu-O. XPS and XRD measurements indicate that Cu-O is hybridized with the CuAlO 2 lattice in the forms of crystalline CuO clusters, which is also in agreement with previously reported results [30]. From the XPS valence band spectra in Figure 2g,h, the valence band positions of CuAlO 2 and CuAlO 2 /Cu-O are 0.11 eV and 0.54 eV lower than Fermi level, respectively.
Next, the work function of CuAlO 2 and CuAlO 2 /Cu-O was investigated from the UPS measurements (Figure 2i). The work function can be estimated from this equation [55]: Through a linear extrapolation of the secondary electron cut off edge, the E cutoff of CuAlO 2 is about 15.97 eV in the high binding energy. Combined with the energy of the incident photon, the work function for CuAlO 2 is calculated to be about 5.25 eV, while that for CuAlO 2 /Cu-O is about 5.31 eV. In addition, the ionization potential increases from 5.36 to 5.85 eV. These results manifest that the valence band maximum (VBM) keeps away from the vacuum level and the binding energy increases through the hybridization of Cu-O with the CuAlO 2 host. CuAlO 2 /Cu-O is not the physical mixture of CuO clusters in the CuAlO 2 matrix. CuO clusters can induce the formation of Cu 2+ 3d 9 orbitals, which are easier to activate in the VBM of CuAlO 2 [56,57], thus, enhancing the p-type conduction of CuAlO 2 [30]. Hence, the energy band level and electronic properties of CuAlO 2 can be adjusted by alloying Cu-O with a CuAlO 2 host. Other Cu + -based ternary p-type transparent oxides also experience a similar valence band shift effect after optimizing their electrical properties [24].
To demonstrate the application of the delafossite materials, ZnO-NRs-based photodetectors were fabricated in this work (Figure 3a). Figure 3b,c reveal the external quantum efficiency (EQE) of the as-fabricated ZnO/CuAlO 2 and ZnO/[CuAlO 2 /Cu-O] UV detectors at corresponding wavelengths of the UV light at zero bias. The performance of the UV photodetector is typically characterized by the responsivity (R), which can be calculated by using the following equation [58,59]: where λ is the radiation wavelength, c is the speed of light, and h is the Planck constant. The response wavelength range is from 320 to 440 nm and the peak responsivity is found at 370-380 nm, which correlates well with the high optical absorptance lower than 400 nm wavelength. The device is visible-blind ultraviolet detection, which is consistent with the results of transmittance spectra. The peak responsivity of the ZnO/[CuAlO 2 /Cu-O] photodetector is about 91.3 mA/W, higher than the 13.2 mA/W of ZnO/CuAlO 2 photodetector, indicating that the incorporation of Cu 2+ significantly improves the performance of the device. It is ascribed to the increase in the photocurrent, as more carriers are collected under illumination after the incorporation of Cu 2+ in the CuAlO 2 thin films. The maximum responsivity reveals the high capability to detect UV signals and the presence of internal gain in the device. The related UV-to-visible rejection ratio (375-440 nm) of the devices is about 75, which confirms the UV spectrum selectivity. This is attributed to the wide bandgap of the ZnO (~3.25 eV) and the CuAlO 2 (~3.11 eV) semiconductors, which leads to a dramatic increase in UV light absorption in the active material of the device. There is no significant change in the quantum efficiency for wavelengths higher than 400 nm, which may be related to the reduction in photogenerated carrier due to the decrease in the optical absorptance of ZnO [47]. After applied bias voltage, the photoresponsivity for the photodetector increases up to 5002 mA/W at −1.5 V under 375 nm wavelength (Figure 3d,e). This enhancement is attributed to the fact that applying a negative bias can enhance the width of built-in electric field, which leads to separate photogenerated carriers more effectively. In short, the ZnO/[CuAlO 2 /Cu-O] detectors deliver the high performance in the UV-A region. presence of internal gain in the device. The related UV-to-visible rejection ratio (375-440 nm) of the devices is about 75, which confirms the UV spectrum selectivity. This is attributed to the wide bandgap of the ZnO (~3.25 eV) and the CuAlO2 (~3.11 eV) semiconductors, which leads to a dramatic increase in UV light absorption in the active material of the device. There is no significant change in the quantum efficiency for wavelengths higher than 400 nm, which may be related to the reduction in photogenerated carrier due to the decrease in the optical absorptance of ZnO [47]. After applied bias voltage, the photoresponsivity for the photodetector increases up to 5002 mA/W at −1.5 V under 375 nm wavelength (Figure 3d,e). This enhancement is attributed to the fact that applying a negative bias can enhance the width of built-in electric field, which leads to separate photogenerated carriers more effectively. In short, the ZnO/[CuAlO2/Cu-O] detectors deliver the high performance in the UV-A region. Furthermore, the specific detectivity is used to characterize the sensitivity of the photodetector. The higher the specific detectivity, the better the detection ability of the device for weak signals. It can be calculated using the following formula [47]: where D* is the specific detectivity and Jdark is the dark current density at bias voltage. The detectivity of ZnO/CuAlO2 is calculated to be 2.35 × 10 10 cm Hz 1/2 W −1 . The incorporation of Cu-O clusters enhances the hole transport capacity and increases the photocurrent Furthermore, the specific detectivity is used to characterize the sensitivity of the photodetector. The higher the specific detectivity, the better the detection ability of the device for weak signals. It can be calculated using the following formula [47]: where D * is the specific detectivity and J dark is the dark current density at bias voltage. The detectivity of ZnO/CuAlO 2 is calculated to be 2.35 × 10 10 cm Hz 1/2 W −1 . The incorporation of Cu-O clusters enhances the hole transport capacity and increases the photocurrent density, thereby improving the responsivity. Therefore, the ZnO/[CuAlO 2 /Cu-O] detector shows a higher detectivity of 1.71 × 10 11 cm Hz 1/2 W −1 for weak signal. Figure 3f shows the stability and the time response characteristics of the devices measured at −1 µV bias under illumination of 17.7 mW/cm 2 at 365 nm. All the devices demonstrate excellent stability and repeatability in five cycles. The photogenerated current of the ZnO/[CuAlO 2 /Cu-O] device is about 6.99 µA, significantly higher than the ZnO/CuAlO 2 device (~0.78 µA). The photocurrent enhancement is mainly due to the reduction in carrier recombination by incorporating Cu-O clusters, which is in agreement with the previous study [39]. Under UV illumination, ZnO NRs absorb photons to generate electron-hole pairs, and holes migrate toward the ZnO surface, discharging the negatively charged adsorbed oxygen ions (O 2(ad) − + h + → O 2(g) ). After the UV light is off, the photocurrent decay is governed by the slow surface-related process. Oxygen molecules adsorbed physically on the ZnO surface capture electrons from the conduction band to become chemisorbed oxygen (O 2(g) + e − → O 2(ad) − ) [12,60,61]. Therefore, the response processes can be fitted by the following biexponential decay model [62]: where τ 1 and τ 2 are two relaxation time constants. The rise time is defined as the time to rise from 10% to 90% of the highest photocurrent value after UV illumination, and decay time is defined as the time to fall from 90% to 10% of the highest photocurrent value after switching off UV illumination. More technically, R 2 is a measure of goodness of fit, and its value ranges from 0 to 1. When the R-squared value of the trend line is closer to 1, the degree of fitting is better and the trend line is much more reliable. The rise and decay times of the ZnO/CuAlO 2 detector are 1.97 s and 2.81 s in Figure 3g To help understand the carrier transport process, the schematic energy band diagram of the UV photodetector based on the ZnO/CuAlO 2 heterojunction is shown in Figure 4a. The bandgap of ZnO is about 3.25 eV, while the electron affinity and work function have been reported to be about 4.35 and 4.45 eV, respectively [63]. The values of the band gap, the work function, and valence band position for CuAlO 2 are 3.11, 5.25, and 5.36 eV, respectively, in this work. The type II band alignment between ZnO and CuAlO 2 facilitates carrier separation and transportation. When UV light is irradiated from FTO glass, ZnO absorbs photons to generate electron-hole pairs predominantly. Under the action of the built-in electric field, the electron-hole pairs are separated and collected into the electrode, resulting in the formation of the photocurrent. The CuAlO 2 , as the p-type layer, is beneficial for carrier separation. The reason is that CuAlO 2 as an energy barrier layer reduces the recombination rate, and improves the electron injection into the ZnO layer [29,64].   Figure 4b. The photocurrent curve coincides well with the dark current curve, indicating that the FTO and Au electrode are ohmic contacts. As for the ZnO NRs/Au device, it shows a relatively weak photoresponse. However, the obvious rectifying diode characteristic for ZnO/CuAlO2 confirms the formation of a p-n junction rather than a Schottky junction. The ZnO/CuAlO2 heterojunction with the forward current density to the reverse current density ratio is 18 at ±1 V, and is lower than the ratio of 30 at ±1 V for the diode made of ZnO/[CuAlO2/Cu-O] (Figure 4b). It demonstrates the ability of ZnO/CuAlO2 heterojunctions to separate photogenerated carriers. Moreover, the incorporation of Cu 2+ can also improve the electrical conductivity. The current increases greatly when the ZnO/[CuAlO2/Cu-O] heterostructure photodetector is illuminated by UV light compared with dark current at −1 V, indicating a highly UV-sensitive photoconduction and photocurrent. The strength of the space electric field is increased with the increase in reverse bias, resulting in a more efficient separation of the photogenerated carriers. Therefore, the ZnO/[CuAlO2/Cu-O] heterojunction can further enhance the electrical properties.
In addition, the diode ideality factor (n) and dark saturation current density (Jo) can be extracted from semi-log plot of the I-V curve using the following Shockley ideal diode equation [65,66]:   Figure 4b. The photocurrent curve coincides well with the dark current curve, indicating that the FTO and Au electrode are ohmic contacts. As for the ZnO NRs/Au device, it shows a relatively weak photoresponse. However, the obvious rectifying diode characteristic for ZnO/CuAlO 2 confirms the formation of a p-n junction rather than a Schottky junction. The ZnO/CuAlO 2 heterojunction with the forward current density to the reverse current density ratio is 18 at ±1 V, and is lower than the ratio of 30 at ±1 V for the diode made of ZnO/[CuAlO 2 /Cu-O] (Figure 4b).
It demonstrates the ability of ZnO/CuAlO 2 heterojunctions to separate photogenerated carriers. Moreover, the incorporation of Cu 2+ can also improve the electrical conductivity. The current increases greatly when the ZnO/[CuAlO 2 /Cu-O] heterostructure photodetector is illuminated by UV light compared with dark current at −1 V, indicating a highly UV-sensitive photoconduction and photocurrent. The strength of the space electric field is increased with the increase in reverse bias, resulting in a more efficient separation of the photogenerated carriers. Therefore, the ZnO/[CuAlO 2 /Cu-O] heterojunction can further enhance the electrical properties.
In addition, the diode ideality factor (n) and dark saturation current density (J 0 ) can be extracted from semi-log plot of the I-V curve using the following Shockley ideal diode equation [65,66]: where J is the dark current density, J 0 is the reverse current density, v is the applied voltage, n is the ideality factor, T is the absolute temperature, and k is the Boltzmann constant. In order to compare the p-n junction characteristics of the two devices, we selected the same linear region (the blue line) to fit the LnJ-V curves. The shape of I-V curves depends on the height of the Schottky barriers at the interface of semiconductors. As for the UV photodetectors, the n drops from 2.3 to 1.9 and J 0 decreases from 0.81 to 0.24 mA cm −2 .
The relatively large J 0 can be attributed to the presence of structure imperfections and high density of interface traps in the ZnO/CuAlO 2 heterojunction, causing a current tunnel path and increasing the reverse current [67]. Thus, through the incorporation of Cu 2+ in CuAlO 2 , high-quality CuAlO 2 /Cu-O films and a high-performance heterojunction were achieved.
To gain more insights into the reason for device performance improvement, the EIS measurements were used to investigate interfacial carrier separation and transfer ability in the frequency range from 0.1 Hz to 1 MHz in the dark (Figure 4d). From Figure 4d, one can see that the resistances consist of charge transfer resistance (R ct ) obtained in the high-frequency region, recombination resistance (R r ) obtained in the low-frequency region, and series resistance (R s ) originating from a combination of contact resistance in the overall circuit and electrolyte [68]. The constant phase element (CPE) is the non-ideal frequencydependent capacitor. By the analysis of Zview, the ZnO/[CuAlO 2 /Cu-O] detector has a much lower charge transfer resistance (~27.84 Ω) than that of the ZnO/CuAlO 2 (~121.2 Ω). The results suggest that the incorporation of Cu 2+ leads to a more effective charge separation and a faster interfacial charge transfer. Table 1 lists several parameters of self-powered ZnO NRs/[CuAlO 2 /Cu-O] photodetectors and other ZnO-based p-n junction UV detectors. We can see that the responsivity of ZnO/[CuAlO 2 /Cu-O] is lower than the n-ZnO/CdS/p-GaN structure under zero bias. The excellent responsivity of the n-ZnO/CdS/p-GaN structure is due to the CdS insert layer, which is used as a carrier transition layer to effectively reduce the interfacial charge recombination, resulting in better photoresponse characteristics. ZnO NRs/[CuAlO 2 /Cu-O] has a simple structure, which is through hybridizing Cu-O clusters into the CuAlO 2 lattice to adjust the energy band and enhance the conductivity of the materials, thereby improving the performance of photodetector. It has a higher responsivity of 5.002 A/W at 1.5 V. Compared with other ZnO-NRs-based photodetectors, ZnO NRs/[CuAlO 2 /Cu-O] shows higher response and excellent detectivity, indicating that CuAlO 2 /Cu-O is a promising p-type material for optoelectronic community.

Conclusions
In summary, a novel ZnO-NRs-based UV photodetector was prepared by all-wet synthesis methods at relatively low temperatures. The photodetector has a good self-powered performance due to good type II band alignment. Further, hybridizing Cu-O clusters into the CuAlO 2 lattice can adjust the energy band and enhance the conductivity of the materials, thereby improving performance of photodetector. The ZnO/[CuAlO 2 /Cu-O] photodetector delivers a strong responsivity of 91.3 mA/W at 0 V bias, which is almost seven times higher than the ZnO/CuAlO 2 photodetector (~13.2 mA/W) under UV radiation of 65 mW cm −2 . The detectivity also increases from 2.35 × 10 10 to 1.71 × 10 11 Jones. Particularly, at 1.5 V bias, the responsivity for the ZnO/[CuAlO 2 /Cu-O] photodetector is about 5002 mA/W under 375 nm UV illumination. Therefore, hybridizing Cu-O clusters provides a new idea to improve the electrical properties of the delafossite CuMO 2 (M = Al, Ga, In, Fe) materials for high-performance optoelectronic devices.