Cuprous oxide (Cu₂O) is an important p-type semiconductor and has attracted significant attention for photocatalysis due to its non-toxicity, low cost, and facile controllable synthesis [1-4]. However, the relatively poor photocatalytic performance of Cu₂O has severely restricted practical applications in environment remediation and energy conversion. It has been reported that Cu₂O cubes show minimal photocatalytic activity because of the ultrathin surface layer and energy barrier which prevents charge transfer [2, 5, 6]. Heterostructures (HCs) have been used to improve the photocatalytic activity of Cu₂O cubes [7, 8], with notable examples including metals/Cu₂O [9-14], semiconductors/Cu₂O [15-19], carbon/Cu₂O HCs [20-24]. However, the formed HCs did not show good photocatalytic stability even when their photocatalytic activity was significantly improved [16, 25].

In a previous study [16], as-obtained CuO/Cu₂O HCs formed by mild etching of Cu₂O greatly improved the photocatalytic stability of Cu₂O cubes, though the photocatalytic activity was unimpressive. In another study [21], we found that Au/Cu₂O cubes prepared by photo-deposition effectively enhanced the photocatalytic activity of Cu₂O cubes. Herein, we hypothesize that combining CuO with Au nanodomains on Cu₂O cubes would synergistically improve both photocatalytic activity and stability. To test this hypothesis, we prepared dual surface HCs, i.e. CuO/Cu₂O and Au/Cu₂O on Cu₂O cubes, using mild surface oxidative etching and deposition of CuO and Au nanoparticles. The morphologies and photoelectric currents of the prepared materials were characterized and the photocatalytic activity and stability were determined and compared to those of Cu₂O, Au/Cu₂O, and CuO/Cu₂O. In this way, the synergistic effects of CuO and Au surface HCs on the photocatalytic activity and stability were thoroughly investigated herein.

All chemical reagents were of analytical grade and used as received. Cu₂O cubes were fabricated according to a previously reported method [26]. In a typical procedure, 10 mL of 0.2 mol/L NaOH aqueous solution was added dropwise to 100 mL of 10 mmol/L CuCl₂ aqueous solution at 55 ℃. After stirring for 0.5 h, 10 mL of ascorbic acid aqueous solution (0.6 mol/L) was added dropwise to the above reaction solution. The mixed reaction solution was maintained at 55 ℃ under stirring for another 5 h. The reaction solution gradually became brick-red, suggesting the formation of Cu₂O. The obtained samples were washed with distilled (DI) water and absolute ethanol, then dried in a vacuum oven at 50 ℃ for 6 h.

The CuO/Cu₂O samples were prepared as follows. A sample of Cu₂O cubes (43.3 mg) was added to 30 mL of 50 mmol/L ethylenediamine (EDA) aqueous solution under ultrasonication. Subsequently, the suspension solution was sealed in a PTFE vessel and heated at 180 ℃ for 5 h. After the reaction, the product was collected via centrifugation and washed with DI water and absolute ethanol three times. Finally, the product was dried in a vacuum oven at 50 ℃ for 6 h.

CuO/Cu₂O HCs (20 mg) was dispersed into 3 mL of DI water under ultrasonication. Subsequently, the suspension was photo-irradiated using a 300-W Xe lamp. A HAuCl₄ (2.0 mL, 0.125 g/L) aqueous solution was then injected into the above suspension under vigorous stirring, and the suspension was further irradiated for 1 h. Afterwards, the product was collected via centrifugation, washed as described above, and dried in a vacuum oven at 50 ℃ for 6 h.

The X-ray diffraction (XRD) patterns of the samples were recorded using a Bruker D8 Discover diffractometer with Cu-K_α radiation (λ = 0.15406 nm) in a 2θ range from 20° to 80°. Transmission electron microscopy (TEM) images were obtained using a FEI Tecnai G2 F20 S-TWIN microscope operating at an accelerating voltage of 200 kV. Elemental analysis of the samples was conducted using an energy-dispersive X-ray (EDX) analysis system attached to the TEM instrument. For TEM, the dried samples were sonicated in ethanol and subsequently deposited onto a carbon grid. Scanning electron microscopy (SEM) images were recorded using a Hitachi S4800 scanning electron microscope. UV-visible diffuse reflection absorption spectra were obtained using a PE Lambda 650s UV-vis spectrometer. X-ray photoelectron spectroscopy (XPS) measurements was performed using an Escalab 250 Xi spectrometer. Photoluminescence (PL) spectra were measured using a PE-LS55 fluorescence spectrophotometer. Nitrogen adsorption-desorption measurements were conducted using a Quantachrome NOVA4200e analyzer. Before measurements, all samples were degassed at 200 ℃ for 6 h under vacuum. Electron spin resonance (ESR) spectra of radicals in spin-trapped 5, 5-dimethyl-1-pyrroline-N-oxide (DMPO) were measured using a Bruker A300 electron paramagnetic resonance spectrometer.

First, 15 mg of sample was added to a 50 mL of 2×10^–5 mol/L methyl orange (MO) aqueous solution. The suspended solution was then stirred for 0.5 h to reach an adsorption-desorption equilibrium in the dark. Subsequently, the solution was photo-irradiated from a distance of ca. 15 cm using a 300-W Xe lamp (PLS-SXE300, 150 mw/cm²) equipped with a 420-nm cut-off filter to remove UV light. 1 mL aliquot of the reaction suspension was removed and centrifuged at a given time intervals. Photocatalytic activity was evaluated according to the absorption spectra of the dyes recorded using a PE Lambda 650s UV-vis spectrophotometer.

Photocurrent was measured using a computer-controlled electrochemical work station (Autolab PGSTAT204) in a standard three electrode system. The working electrodes were prepared by mixing the as-obtained samples (10 mg) with 0.5 mL of Nafion aqueous solution (1%) under ultrasonicated for 15 min to obtain a homogeneous slurry. The slurry (0.1 mL) was subsequently dipped onto a 1 × 2 cm ITO slice and dried in air. A 300-W Xe lamp equipped with a UV cutoff filter was used as a light source. Current-time curves were collected at 1.0 and –0.05 V vs. SCE.

Fig. 1 shows SEM, TEM, and HRTEM images of the as-obtained Au/CuO/Cu₂O HCs samples. Randomly but well-dispersed CuO and Au nanoparticles were observed on the surfaces of the Cu₂O cubes. The CuO nanodomains exhibited polyhedral shapes while the Au nanodomains showed almost spherical shapes. In Fig. 1(B) and (C), lattice spacings of 2.32 and 2.53 Å were assigned to the (111) and (002) planes of CuO, respectively, indicated both nanodomains were intimately contacting Cu₂O. The corresponding Fast Fourier Transform (FFT) diffraction patterns in Fig. 1(F) and the EDX spectrum (Fig. 1(H)) further indicate the presence of Au nanoparticles. The XRD patterns in Fig. 1(J) of the as-prepared samples show strong Cu₂O peaks, but no peaks corresponding to Au or CuO were observed. These results suggest minimal loading amounts of Au and CuO compared to that of Cu₂O. Moreover, as shown in Fig. 1(G), the mean size of the Au nanoparticles is ca. 35 nm, much larger than that of Au/Cu₂O [25]. This strongly suggests that etching made the Cu₂O cube surface more negative, facilitating the rapid growth of Au nuclei [21]. Therefore, the surface barriers of Cu₂O were significantly reduced and CuO/Cu₂O HCs with intimate contacts with the substrate were formed.

To explore surface states of the Cu₂O etched by EDA, XPS spectra of Au/CuO/Cu₂O were compared with those of pristine Cu₂O and CuO/Cu₂O, as shown in Fig. 2(A) and Table 1. The peaks at 932.2 and 934.1 eV were attributed to Cu(Ⅰ) and Cu(Ⅱ), respectively [27]. Compared to pristine Cu₂O, the Cu(Ⅰ) peak shifted slightly to a lower binding energy. As shown in Table 1, the surface area ratio of Cu(Ⅱ) to Cu(Ⅰ) of the as-obtained Au/CuO/Cu₂O was compared to those of pristine Cu₂O and CuO/Cu₂O. The ratio of Cu(Ⅱ) to Cu(Ⅰ) of the Au/CuO/Cu₂O material was only slightly higher than that of pristine Cu₂O. Combined with the SEM results, these results indicate that most Cu(Ⅱ) came from the CuO polyhedron rather than the CuO film formed under ambient conditions. It also suggests that most of the naturally-grown CuO film was removed. It should be noted that naturally-grown CuO thin films on Cu₂O cubes act as a barrier layer preventing charge transfer in the Cu₂O cubes [2, 4, 5].

To explore the deposition mechanism of CuO nanoparticles, XPS of the intermediates during surface etching of Cu₂O was performed and the results are shown in Fig. 2(B). The XPS results show N 1s peaks at 399.9 and 399.4 eV in the Au/CuO/Cu₂O and CuO/Cu₂O samples, respectively, indicating the formation of EDA complexes [28]. We performed control experiments with EDA etching at room temperature under ambient conditions and observed many etched pits on the Cu₂O surface, as shown in Fig. 3(A), indicating EDA was responsible for etching. When a lower concentration of EDA was used (5 mmol/L), CuO nanoplates appeared on the Cu₂O surface, as shown in Fig. 3(B), indicating etching is a kinetically controlled process [28]. These control experiments clearly show that CuO HCs formed on the Cu₂O cubes. We hypothesized that Cu(Ⅰ) and EDA rapidly formed complexes, Cu(en)_x, upon dissolution and subsequent deposition of the CuO nanodomains on the substrate [16, 28]. Moreover, the SEM and TEM images in Fig. 1 show that Au surface HCs also formed. To summarize, dual HCs, i.e. CuO/Cu₂O and Au/Cu₂O, were successfully formed on the Cu₂O surface.

Fig. 4(A) and (B) show photoelectric current measurements with positive and negative biases for each sample. All the results showed that the Au/CuO/Cu₂O had a much higher photoelectric current than pristine Cu₂O, Au/Cu₂O, and CuO/Cu₂O. It can be concluded that the enhanced photoelectric current of Au/CuO/Cu₂O arises from the synergistic effect of the dual surface heterostructures. The photocatalytic activities of the as-prepared Au/CuO/Cu₂O HCs were evaluated by photo-degradation of methyl orange (MO) and compared to those of the pristine Cu₂O, Au/Cu₂O, and CuO/Cu₂O, as shown in Fig. 4(C). The Au/CuO/Cu₂O HCs photo-degraded ˃95% of the MO within 30 min while the pristine Cu₂O photo-degraded ˂10% within 120 min and Au/Cu₂O only 90% within 90 min. Fig. 4(D) shows the intrinsic photocatalytic reaction rates of each photocatalyst. The photo-degradation rate of the Au/CuO/Cu₂O HCs was 230 times that of pristine Cu₂O, 7 times that of CuO/Cu₂O, and 4 times that of Au/Cu₂O. To clarify the underlying mechanism of the enhanced photocatalytic activity, apparent quantum efficiencies of the samples were calculated and are listed in Table 2. The apparent quantum efficiency of Au/CuO/Cu₂O was 123 times that of pristine Cu₂O and 5.4 times that of Au/Cu₂O. It should be noted the CuO and Au nanoparticles showed almost no photocatalytic ability towards the photodegradation of MO [16, 25]. Hence, dual the HCs, i.e. CuO/Cu₂O and Au/Cu₂O, significantly improved the photocatalytic activity.

The results of the photocatalytic stability tests of the as-obtained samples are shown in Fig. 5(A). The Au/CuO/Cu₂O HCs maintained nearly 80% of their already high activity after eight cycles, in contrast to five cycles for the Au/Cu₂O (Fig. 5(B)) [25]. After the cycling test, Cu₂O maintained very good morphology, as shown in Fig. 5C, though the Cu₂O surface was somewhat roughened. The above results indicate that the dual HCs of CuO and Au on Cu₂O cubes significantly enhanced the photocatalytic stability of Cu₂O. Compared to CuO and Au alone, Au/CuO/Cu₂O exhibited a synergistic effect with regards to photocatalytic stability. It should be noted that the pristine Cu₂O showed very poor photocatalytic activity and was not considered for stability comparison.

To confirm the synergistic effect, a series of control experiments were performed. Long duration dark adsorption tests in Fig. 6(A) showed negligible adsorption of MO dyes for all samples. To further rule out the effect of surface area, BET specific surface areas of the Au/CuO/Cu₂O and pristine Cu₂O cubes were measured and the results are shown in Fig. 6(B). The calculated photo-degradation rate per surface area unit of Au/CuO/Cu₂O was determined to be much higher than that of the pristine Cu₂O cubes. Therefore, we excluded the effect of the slightly increased surface area on the overall photocatalytic activity. Considering the negligible amount of Au and CuO nanoparticle loading, it is unlikely that the light absorbance increase would be significant (Fig. 7), as supported by the ratios of apparent quantum efficiencies (Table 2). Based on the XPS valance band spectrum (Fig. 7(C)), the valence band maximum (VBM) values of the samples were similar (approximately 1.9 eV). Thus, the interaction between Cu₂O and CuO/Au did not result in significant changes in the VBM.

The PL spectra in Fig. 8(A) show distinct emission bands at approximately 485 nm, which can be attributed to the excitonic transitions from the different sub-levels of the conduction band (CB) to the Cu d-levels of the valence band (VB) [31]. The decreased PL intensity of Au/CuO/Cu₂O indicated better charge separation compared to those of Cu₂O and CuO/Cu₂O. To further confirm charge separation and identify possible radicals, ESR was performed on the as-obtained samples (Fig. 8(B)). The ESR spectra indicate that hydroxyl radicals (•OH) were predominantly generated with minor production of superoxide radicals (O₂•−), in good agreement with the literature [10]. This suggests that the hydroxyl radicals played a dominant role in the photo-degradation of MO by the Au/CuO/Cu₂O HCs. The hydroxyl radical signal of the Au/CuO/Cu₂O was significantly stronger than that of either the pristine Cu₂O cubes or CuO/Cu₂O, providing further evidence for the synergistic effect of Au/CuO/Cu₂O for improving charge separation.

Based on the above experimental results, we proposed a photocatalytic mechanism, as shown in Fig. 9. Since the CB and VB of CuO are lower than the corresponding bands of Cu₂O [32, 33], CuO/Cu₂O forms a type Ⅱ staggered band structure. The formation of CuO/Cu₂O HCs facilitated charge transfer and acted as a protecting layer to improve photocatalytic stability [2, 6]. CuO/Cu₂O maintained better photocatalytic stability than Au/Cu₂O, indicating that the CuO nanodomains imparted increased stability. This can be explained by: (1) the increased stability of CuO compared to that of Cu₂O; (2) the crystalline CuO improved charge separation leading to significantly reduced accumulation of photogenerated holes at the interface, in good agreement with the literature [34]. In a control experiment, CuO/Cu₂O was prepared using NaCl as a mild oxidative etchant (Fig. 10(A)). The control sample also exhibited improved photocatalytic activity (Fig. 10(B)). It should be noted that surface residual EDA may also act as an electron donor to enhance photocatalytic activity when fabricating CuO/Cu₂O HCs. The formed Au/Cu₂O HCs facilitated charge transfer by building up an internal electric field, which also improved photocatalytic activity. According to the band position alignment in Fig. 9, photo-excited electrons from Cu₂O would be injected into CuO and Au while holes from CuO would transfer to Cu₂O. Localized surface plasmon resonance (LSPR) of the Au nanoparticles would facilitate charge separation, but only at approximately λ = 520 nm, as shown in Fig. 7(A) [25]. In earlier studies [16, 25], pristine CuO cubes and Au nanoparticles alone did not show considerable catalytic activities. Thus, it can be concluded that the synergistic effect of Au/CuO/Cu₂O greatly improved both photocatalytic activity and stability of the prepared catalysts.

In conclusion, CuO and Au nanodomains were easily prepared on Cu₂O cubic surfaces. The as-obtained Au/CuO/Cu₂O HCs exhibited greatly enhanced photocatalytic activity and stability. The photodegradation rate of Au/CuO/Cu₂O was 230 times that of the pristine Cu₂O, 7 times that of CuO/Cu₂O, and 4 times that of Au/Cu₂O. The Au/CuO/Cu₂O HCs retained nearly 80% of its original activity after eight cycles in contrast to five cycles for the Au/Cu₂O. The improved photocatalytic activities and stabilities can be mainly attributed to the synergistic effect of CuO/Cu₂O and Au/Cu₂O HCs on the Cu₂O cubes.