Modification of large area Cu2O/CuO photocathode with CuS non-noble catalyst for improved photocurrent and stability

In this work, a three-layered heterostructure Cu2O/CuO/CuS was obtained through a low-cost and large-area fabrication route comprising electrodeposition, thermal oxidation, and reactive annealing in a sulfur atmosphere. Morphological, microstructural, and compositional analysis (AFM, SEM, XRD, EDS, XPS) were carried out to highlight the surface modification of cuprous oxide film after oxidation and subsequent sulfurization. Impedance, voltammetric, and amperometric photoelectrochemical tests were performed on Cu2O, Cu2O/CuO, and Cu2O/CuO/CuS photocathodes in a sodium sulfate solution (pH 5), under 100 mW cm−2 AM 1.5 G illumination. A progressive improvement in terms of photocurrent and stability was observed after oxidation and sulfurization treatments, reaching a maximum of − 1.38 mA cm−2 at 0 V versus RHE for the CuS-modified Cu2O/CuO electrode, corresponding to a ~ 30% improvement. The feasibility of the proposed method was demonstrated through the fabrication of a large area photoelectrode of 10 cm2, showing no significant differences in characteristics if compared to a small area photoelectrode of 1 cm2.

Cuprous oxide (Cu 2 O) is one of the most investigated p-type semiconductor material in the framework of photoelectrochemical hydrogen production 1-3 , mainly because of its optical properties and potentially lowcost synthesis 4 . It has indeed a direct bandgap of ~ 2.1 eV, capable to absorb a large part of the solar spectrum, resulting in a theoretical maximum photocurrent density of − 14.7 mA cm −2 at AM 1.5 G condition 4 , and has a favorable conduction band position with respect to the hydrogen evolution reaction potential. Moreover, the simple chemical composition, based on earth-abundant elements, allows its synthesis through many different routes (hydrothermal synthesis 5,6 , thermal oxidation 7 , sputtering 8,9 , electrodeposition 10 ), and to achieve a large variety of architectures (micro and nano-crystals 5,6,11 , thin-films [8][9][10] , nanowires 12,13 ). For thin-film based devices, electrodeposition is the major candidate towards the development of low-cost and large-area production and it has been already exploited to achieved efficiency records during the last decade for Cu 2 O-based photoelectrodes 4,14,15 . However, the main drawback affecting this material is the poor stability against photodegradation. In fact, upon illumination, electrons reach the solid-liquid interface reducing the semiconductor to the metallic state (Cu 2 O → Cu) progressively decreasing the photocurrent generated 16 . To minimize this phenomenon, the formation of heterostructures, together with the implementation of a catalyst, is a common and one of the most effective strategy. However, the most performing overlayers often rely on poorly scalable or expensive fabrication techniques such as atomic layer deposition (ALD) 4,14,15 , allowing the deposition of a fewnanometer thick layer, highly conformal and optically transparent. A different approach relies on the exploitation of simple and affordable routes such as thermal or chemical treatments that would result in the formation of the different heterostructures (Cu 2 O/C 12 10 . While the majority of the studies report the growth of cuprous oxide on Au-sputtered substrate at relatively high pH values (pH ~ 12-12.5) 29 , on FTO the plating conditions vary more among different literature works 30 . Preliminary tests were carried out at varying pH (9)(10)(11)(12), keeping constant the bath formulation, the deposition temperature (40 °C), and the current density (− 0.1 mA cm −2 ). The desired microstructure was found growing the film onto the FTO substrate, at pH 11 ( Fig. 1). The intensity of the different peaks revealed the film to be preferentially oriented along the (111) direction, whose diffraction peak was predominant. Theoretically, (111) is the most interesting orientation from the charge transfer viewpoint since Cu 2 O conductivity is attributed to the presence of point defects such as O vacancies 30 , resulting in a higher photoelectrochemical activity 31 .
In agreement with XRD patterns (Fig. 1), SEM micrographs of Cu 2 O films showed a truncated pyramid shape morphology typical of (111) plane of the cubic cell (Fig. S1, Fig. 2a). The oxide characteristic morphology became clearer and more distinct as the deposition charge density increases (≥ 0.75 C cm −2 ) (Fig. S1) while lower deposition time resulted in a more compact surface, with a finer structure and lower surface roughness. A www.nature.com/scientificreports/ progressive increase in the dimension of the crystallites size was observed for longer deposition times, as a result of the growth of the Cu 2 O nuclei going from the nanometer to the micrometer scale (Fig. S1). Theoretically, large-grained material is highly desirable from a semiconducting viewpoint, a more ordered and less-defected structure would result in more efficient charge separation and extraction since recombination is more likely to occur at defects. After air annealing, the XRD pattern showed well-defined peaks associated with cuprous and cupric oxide 32 , in agreement with the formation of the heterojunction (Fig. 1). The preferential orientation of the Cu 2 O underlayer was maintained, as shown by the major diffraction peak Cu 2 O (111), while the CuO phase appeared to be randomly oriented as indicated by the relative intensity of (002), (111), (113) diffraction peaks ( Fig. 1, Fig. S2). As expected, by increasing the annealing time a larger portion of the material was oxidized to cupric oxide ( Fig. S3), maintaining the layered structure. EDS results were in agreement with XRD data, detecting the formation of a phase richer in oxygen, as evinced by the relative peak heights respectively for Cu and O (Fig. 2a,b). Surface micrographs of Cu 2 O/CuO samples showed a significant change in the morphology, characterized by a fine structure and a lower surface roughness, as confirmed by AFM analysis (Fig. 2a,b). It is worth to remark that the quality of the Cu 2 O/CuO heterojunction was dependent on the underneath Cu 2 O, not only in terms of the structure at the microscale (roughness) but especially at the macroscopic one. The sample homogeneity at the macroscale was found to be greatly affected by the electrodeposition charge density, the formation of macroscopic defects (holes) in the layer was indeed directly correlated to the thickness of the cuprous oxide, the thicker the layer the lower the surface homogeneity after oxidation. On the other hand, for charge densities ≤ 1 C cm −2 no macroscopic defects were detected. This was hypothesized to be related to the stresses involved during the oxidation process and the formation of the heterostructures, small delaminated powdery material, coming from the sample, could be found in the quartz vials in case of a thick cuprous oxide film.

Cu 2 O/CuO/CuS heterostructure. Copper sulfide (CuS) overlayer was grown on Cu 2 O/CuO by reactive
annealing in a sulfur-containing atmosphere, using nitrogen as a gas carrier. The sulfurization of the oxidized samples was investigated in a relatively narrow temperature interval (300-400 °C) after preliminary investigation. Specifically, an appreciable reactivity was found for temperatures higher than 250 °C while an upper limit has been set due to the low boiling point of sulfur (~ 444 °C). Above such a temperature, sulfur would evaporate quickly providing an irregular and unpredictable supply of gaseous sulfur species on the sample surface during the test. At 400 °C, the reactivity was found to be too high since a short amount of time (~ 3 min) min was enough to convert both the oxide layers into CuS, nonetheless, the optimal reaction time was found to be 2 min. However, to better control the growth of the top sulfide layer, the temperature was reduced to 300 °C and the www.nature.com/scientificreports/ reaction time extent to 5 min. It is important to notice that the reactivity of Cu 2 O is much greater than CuO; if exposed, the underneath Cu 2 O would quickly react having no control over the process. The homogeneity and coverage of the upper CuO layer are thus crucial for the formation of a three-layered heterostructure. Because of that, particular attention was paid to the optimization of the furnace setup, allowing to carry out surface treatments homogeneously. For the sulfurization step, the samples were placed at the far end of a one-hand closed quartz tube (3 cm diameter) placing elemental sulfur upstream at low nitrogen flux (3 Nl h −1 ). The formation of Cu 2 O/CuO/CuS came along with a slight color change into a golden-brownish appearance, while for an overreacted sample the color would result in a deep blue, typical of bulk CuS. XRD analysis detected only small peaks belonging to the covellite phase, suggesting the formation of a thin overlayer (Fig. 3a,b) along with the slight color change. The GDOES compositional profile along the z-axis confirmed the formation of the three-layered heterostructure showing three well-separated layers (Fig. 3c). The sulfur signal was only detected corresponding to the surface, indicating that the sulfurization involved only the cupric oxide layer. On the other hand, the copper and oxygen signals varied along with the thickness in agreement with the Cu 2 O/CuO structure where the oxygen-poor phase was found in the proximity of the substrate (Fig. 3c). EDS elemental mapping (Fig. 3d,  Fig. S4) of the fractured surface of the sample further confirmed the compositional stratification, in agreement with XRD and GDOES results. The sulfurized surface was similar to the oxidized sample, showing to be compact and with a small-grained morphology characterized by oval clusters, a further reduction in the average surface roughness was observed (Fig. 4, Fig. S5).
To assess with more accuracy the surface elemental composition, samples were analyzed by XPS. The survey spectra of the sample Cu 2 O/CuO/CuS has been reported in Fig. 5a, where copper, sulfur, and oxygen characterized the sample composition. Being introduced from outside the measurement chamber and considering www.nature.com/scientificreports/ adventitious hydrocarbon from XPS, also a small carbon feature was detected. The intensity of both Cu and S peaks were significantly higher compared to the O one, confirming the presence of CuS in the uppermost layers of the sample, while Cu 2 O and CuO were buried. The main peak of the Cu Auger LMM lines (dotted box in the figure) had a BE = 336 eV, which is compatible with an oxidation state equal to + 2. From the analysis of the survey spectrum, we deduced the formation of the CuS compound. However, the chemical state of these elements was better studied from a high-resolution XPS analysis (Fig. 5b,c). The binding energy position of Cu 2p 3/2 and Cu 2p 1/2 (932.1 eV and 952.1 eV, respectively) S 2p 3/2 and S 2p 1/2 (162.1 eV and 163.2 eV, respectively) were in good agreement with the literature 21,33 . The symmetrical behavior of copper peaks and the measured sulfur binding energy confirmed the formation of CuS. Regarding the copper oxide buried layers, we noted that the very weak shoulder at around 943 eV, where the Cu shake-up feature is generally observed, also suggested that the oxide layers were below the CuS one 21 .
Photoelectrochemical characterization. Photoelectrochemical voltammetry tests were carried out to evaluate the semiconductor behavior at increasing cathodic bias. All the tests were performed in a 0.5 M Na 2 SO 4 solution at pH 5 and the potential values recorded were converted to the reversible hydrogen electrode (RHE)   24 , improving the charge separation (Fig. 6). As a consequence, the photocurrents recorded at low cathodic bias were much higher than the bare cuprous oxide, reaching − 1.06 mA cm −2 at 0 V versus RHE after annealing in the air for 2 h. Carrying out a shorter oxidation treatment (1 h) showed no variation in the immersion potential, indicating that the cupric oxide layer grew homogeneously all over the surface only after 1 h of air annealing; nevertheless, the photocurrent recorded was lower (− 0.75 mA cm −2 at 0 V vs RHE) (Fig. S7), in agreement with previous reports 18,21 . It is also worth to notice that, in the case of Cu 2 O, the majority of the current recorded was correlated to the reduction of the semiconductor material to the metallic state (Cu 2 O → Cu), sustaining the photocurrent densities at a high cathodic bias (< 0 V vs RHE). The energy level of the self-reduction reaction lies indeed between the conduction band and the hydrogen evolution energy level, favoring the degradation reaction rather than the proton reduction 3 . However, the dark current of the Cu 2 O/ CuO photoelectrode was different from zero, especially approaching 0 V versus RHE, suggesting either a sign of catalytic activity of the overlayer or the presence of a secondary reaction such as self-reduction. The overall photoelectrode performances were further improved with the addition of the copper sulfide overlayer (Fig. 7a,b). The immersion potential under illumination was shifted towards more positive values at + 0.74 V versus RHE and the maximum photocurrent recorded was increased while no significant changes were observed in the curve profile. At 0 V versus RHE, the Cu 2 O/CuO/CuS heterostructures showed a photocurrent of − 1.38 mA cm −2 , corresponding to ~ 100% and ~ 30% enhancement with respect to Cu 2 O and Cu 2 O/CuO respectively. Although no huge variations were observed in the immersion and onset potentials, a significant increase in the photocurrent was observed at moderate biases, from + 0.3 to 0 V versus RHE (Fig. 7b), when compared to the oxidized sample. Copper sulfide is thus expected to act as a catalyst, improving the charge transfer from the semiconductor surface to the electrolyte, as reported by a previous literature work 21 . To highlight the role of CuS overlayer, the charge transfer of the photogenerated electrons was studied by electrochemical impedance spectroscopy (EIS) at + 0.3 V versus RHE, under illumination (Fig. 7c). The typical semicircular feature of the Nyquist plot was found for all the three photoelectrodes, showing a progressive reduction in the semicircle diameter, associated with the charge transfer resistance (R ct ), passing from Cu 2 O to Cu 2 O/CuO and Cu 2 O/CuO/ CuS heterostructures, confirming the staggered band alignment between Cu 2 O and CuO 17,18,24 and the catalytic activity of CuS 21,26,33 . The results are in agreement with the photoelectrochemical polarization tests where higher photocurrents were recorded for the CuS-modified surface. www.nature.com/scientificreports/ To observe the differences in photostability, the surface potential was maintained constant at 0 V versus RHE while intermittently illuminating the sample. Figure 8 showed the difference in photocurrent stability over time for the Cu 2 O, Cu 2 O/CuO, and Cu 2 O/CuO/CuS samples. Cuprous oxide photoelectrode exhibited a quick quenching of the photocurrent density, in agreement with the previous considerations made on its stability. During the test, the semiconductor was indeed reduced to the metallic state, progressively diminishing the amount of photoactive material available, resulting in a photocurrent density of − 0.025 mA cm −2 after 1800 s of chopped illumination (Fig. 8b). The double and triple heterostructures showed higher photostability than bare cuprous oxide as expected from the LSV investigation. In particular, the sulfurized sample showed photocurrent of − 0.51 mA cm −2 at 500 s and − 0.31 mA cm −2 at 1800 s, corresponding respectively to 82% and 15% higher than the oxidized sample. The majority of the photocurrent was lost after about 10 min and subsequently stabilized at values higher than those of Cu 2 O/CuO heterostructure. After the polarization test, the sample surface showed a similar morphology to the as-synthesized one (Fig. S8) while the surface composition was found to be richer in copper, as shown by EDS (Fig. S9), suggesting that a part of the photocurrent was associated to self-reduction rather than hydrogen evolution reaction. Although the performance improvement was visible, photodegradation occurred and the results were in contrast with a previous report on Cu 2 O/CuO/CuS heterostructure 21 . In this regard, the employment of a substrate (Cu) could result in a more favorable energy level alignment due to the higher work function value than FTO.
To demonstrate the feasibility of the approach proposed, a large area photoelectrode of about 10 cm 2 was fabricated. The samples showed to be macroscopically homogeneous at each synthesis step showing a clear color change, distinguishable at naked-eye (Fig. 9). More importantly, to evaluate the scalability of the fabrication  www.nature.com/scientificreports/ proposed, the photoelectrochemical performances of the large area photoelectrode were compared to a smaller area device of 1 cm 2 . The amperometric tests (Fig. 10) showed no differences between the two samples, indicating that the modification of the cupric oxide layer through reactive annealing was reliable as demonstrated to homogeneously modify the overall surface oxide surface.

Conclusions
Sulfurization treatment has been proved to be an effective way to modify the surface of Cu 2 O/CuO heterostructure. The reactive annealing resulted in the formation of a three-layered heterostructure having a CuS overlayer, homogeneously covering the CuO surface. Thermal oxidation and reactive annealing showed the formation of CuO and CuS layers respectively, having no preferential orientation contrary to the (111) preferentially oriented Cu 2 O underneath layer. A progressive reduction of surface roughness occurred after the treatments and the characteristic pyramidal texture of Cu 2 O was flattened. Photoelectrochemical data showed that such modification improved the stability, reducing photodegradation, and increased the maximum photocurrent delivered (− 1.38 mA cm −2 at 0 V vs RHE under 100 mW cm −2 AM 1.5 G illumination). The fabrication of a 10 cm 2 sample demonstrated that the proposed method is scalable and promising for potentially large-area and low-cost devices.