Abstract
Cu2O has been deposited on quartz substrates by reactive ion beam sputter deposition. Experimental results show that by controlling argon/oxygen flow rates, both n-type and p-type Cu2O samples can be achieved. The bandgap of n-type and p-type Cu2O were found to be 2.3 and 2.5 eV, respectively. The variable temperature photoluminescence study shows that the n-type conductivity is due to the presence of oxygen vacancy defects. Both samples show stable photocurrent response that photocurrent change of both samples after 1,000 seconds of operation is less than 5%. Carrier densities were found to be 1.90 × 1018 and 2.24 × 1016 cm−3 for n-type and p-type Cu2O, respectively. Fermi energies have been calculated, and simplified band structures are constructed. Our results show that Cu2O is a plausible candidate for both photoanodic and photocathodic electrode materials in photoelectrochemical application.
1. Introduction
Recently the development of environmental friendly, green, and sustainable sources of energy has become a global issue. The use of fossil fuel as a source of energy will not be possible in the near future due to its limited amount and nonrenewability. In order to overcome this issue, use of low-cost semiconductor materials for the design of efficient solar convertors is imperative. One possible approach is the use of transition metal oxide (TMO) semiconductor materials in photoelectrochemical (PEC) water splitting to convert solar energy to storable hydrogen energy. Cu2O is a promising semiconductor material for solar energy conversion with a direct bandgap of 2.0–2.5 eV [1–3] and high-absorption coefficient. Especially, since the conduction and valence bands of Cu2O are located close to the reduction and oxidation potentials of water, Cu2O is also a feasible material for the production of hydrogen and oxygen through solar PEC water splitting. Development of simple and economical techniques to deposit photo-active n-type and p-type Cu2O electrodes is essential for the fabrication of PEC cells or solar cells based on n-Cu2O/p-Cu2O homojunction which is expected to achieve devices with high efficiency. Though Cu2O is a promising material, the development of Cu2O-based applications is still delayed due to lack of n-type Cu2O. Recently, both photoanodes and photocathodes are formed from Cu2O thin films, and the results indicate that Cu2O is potentially applicable for PEC water splitting [4, 5]. Cu2O prepared by thermal oxidation or by physical vapor deposition usually contains copper vacancy () defects with a carrier density as high as 1018 cm−3 that results in its p-type conductivity [6]. On the contrary, as our previous report [7] indicates, high-quality n-type Cu2O material can be deposited by controlling the oxygen flow rates using reactive ion beam sputter deposition (IBSD) in addition to other studies [8–10] done by sputtering and electrodeposition without doping. In this article, we report the demonstration of both n-type and p-type Cu2O photoelectrodes fabricated by reactive IBSD. Results on conductivity type, photocurrent stability, and carrier densities are presented.
2. Experimental
Cu2O samples were deposited on quartz substrates in an IBSD chamber [2, 7, 11]. The substrates were cleaned ultrasonically in ethyl alcohol for 10 minutes and then subsequently in deionized water. An air gun with dry nitrogen spray is used to remove all the water droplets and moisture from the surface of the substrates. A metallic copper target (99.99%) was positioned 35 mm downstream of the ion source, and the prepared substrates were placed at 65 mm upstream of the copper target. Two sets of samples with Ar : O2 flow rates of 4.5 sccm : 0.5 sccm and 4.2 sccm : 0.7 sccm were prepared, where sccm stands for standard cubic centimeter per minute. The samples deposited with Ar : O2 = 4.5 : 0.5 were labeled as Sample 1, while the samples deposited with Ar : O2 = 4.2 : 0.7 were labeled as Sample 2. Both samples were deposited at 450°C for 1.5 hours. The surface morphology of both samples was studied by a field-emission scanning electron microscope (FE-SEM, JEOL JSM-6500F, 15 keV). X-ray photoelectron spectroscopy (XPS) investigation was achieved employing a VG ESCA® Scientific Sigma Probe with Al Kα source at 1486.4 eV. Temperature dependent photoluminescence (PL) properties were characterized from 10 to 300 K using a blue laser at 405 nm with a power of 5 mW as the excitation light source. The spectra were dispersed by a Triax 550 spectrometer and detected by a Symphony CCD detector cooled to −70°C. PEC measurements were carried out utilizing a Gamry G 300 potentiostat with an Ag/AgCl electrode as the reference electrode, Cu2O samples as the working electrode, and Pt as the counter electrode. During PEC measurement, the Cu2O samples were attached to a customized electrochemical cell, and a xenon lamp with an intensity of 100 mW/cm2 was used for illumination. The electrochemical cell was filled with argon purged 0.5 M K2SO4 as the electrolyte.
3. Results and Discussion
Figures 1(a) and 1(b) show FE-SEM micrographs of Sample 1 and Sample 2, respectively. From Figure 1(a), Sample 1 exhibits a morphology containing Cu2O nanorods and small Cu2O nanostructures. The Cu2O nanorods exhibit a length of approximately 1 μm and a diameter of ∼0.2 μm. As the oxygen flow rate increases, Sample 2 exhibits a smooth thin film surface morphology with tiny nanoparticles on the surface, as shown in Figure 1(b).
(a)
(b)
Figure 2(a) shows Cu2p core level spectra of both samples. From Figure 2(a), Cu2p3/2 and Cu2p1/2 binding energy peaks of both samples are located at 932.4 eV and 952.4 eV [12], respectively. The absence of shake-up satellite peaks in the Cu2p spectral around 943 eV and 960 eV indicates the absence of CuO from both Sample 1 and Sample 2. From Figure 2(b), the O1s core level spectra of both samples were deconvoluted into two peaks p1 and p2 located at ∼530.5 and ∼532.3 eV [12], respectively. The first peak p1 is attributed to oxygen located in ideal metal oxide matrix, while the second peak p2 is attributed to surface adsorbed oxygen or water. The area percentages of p2 in Sample 1 and Sample 2 are 7.8% and 40%, respectively. This indicates that Sample 1 has lower oxygen content that comes from surface adsorbed oxygen or water, indicating a better surface quality. The atomic percentage ratios of Cu : O were calculated and were found to be 64.5 : 35.5 and 62.6 : 37.4 for Sample 1 and Sample 2, respectively. XPS results show that both samples are close to stoichiometric Cu2O, while higher oxygen flow rate results in higher oxygen atomic percentage.
(a)
(b)
Figure 3 shows baseline corrected PL spectra of Sample 1 and Sample 2 taken at 10 K. From Figure 3(a), Sample 1 shows a strong PL emission centered at ∼1.72 eV, while Sample 2 shows no PL emission peaks at all. As temperature increases, the intensity of 1.72 eV emission decreases while its peak position shifts to lower energy level as temperature is raised to 100 K (Figure 3(b)). The slope of the 1.72 eV peak position emission was found to be –5.2 × 10−4 eV/K, which indicates that this emission is due to the presence of oxygen vacancy defects [13]. The PL study shows that Sample 1 contains defects only, while no -related emissions were found, which suggests that Sample 1 is likely to be of n-type conductivity.
(a)
(b)
Figure 4 demonstrates Tauc’s plots which indicate the direct bandgap nature of both samples. From Figure 4, Sample 1 can be characterized with the bandgap at E1 = 2.3 eV [14] and Sample 2 can be characterized with the bandgap at E2 = 2.5 eV [2]. As clearly seen from Figure 4, Sample 1 shows a strong band tail indicating the presence of subbandgap states (defect states). These defect states are likely due to oxygen vacancy which can be formed ∼0.38 eV below the conduction band minimum [15]. From Figure 4, it is clearly seen that the bandgap becomes narrower when the oxygen flow rate decreases.
Figures 5(a) and 5(b) show transient photocurrent properties of Sample 1 under −0.3 V (Figure 5(a)) and +0.3 V (Figure 5(b)) bias relative to Ag/AgCl reference electrode. The result evidently shows that, under negative bias, the sample shows negligible photocurrent. Conversely, when the sample is biased positively, stable anodic photocurrent is observed, demonstrating that Sample 1 is of n-type. Figures 5(5) and 5(d) show transient photocurrent properties of Sample 2 under −0.3 and +0.3 V biases, respectively. When Sample 2 is biased negatively, stable cathodic photocurrent is observed, while under positive bias, it shows almost zero anodic photocurrent. The observation of stable and dominant cathodic photocurrents indicates that Sample 2 is of p-type. The transformation of the carrier type from n-type to p-type when the oxygen flow rate increases indicates that oxygen vacancy defects are the sources of n-type conductivity in Cu2O as observed from PL data.
(a)
(b)
(c)
(d)
Figure 6 shows the linear sweep voltammetry measurement of Sample 1 and Sample 2. Figure 6(a) clearly shows that, in a broad window of measurement voltage, Sample 1 exhibits anodic photocurrent across the whole voltage range and the value increases as the potential becomes more positive. On the contrary, Figure 6(b) shows in the same broad window of measurement voltage, Sample 2 shows cathodic photocurrent across the whole voltage range and the value increases as the potential becomes more negative. This observation of n-type conductivity form Sample 1 and p-type conductivity from Sample 2 is consistent with transient photocurrent characterization results.
(a)
(b)
From Figure 7(a), the Mott–Schottky plot of Sample 1 shows the n-type nature of the photoelectrode since the value of the slope is positive. However, the Mott–Schottky plot of Sample 2 shows typical behavior of p-type electrode since the value of the slope of the slope is negative. The result of the slopes of the Mott–Schottky plots of Cu2O electrodes is an additional confirmation of carrier types investigated by using both the photocurrent and linear sweep voltammetry measurement. The acceptor and donor densities can be calculated employing the Mott–Schottky equations [16]:where the negative sign is for p-type and positive sign for n-type conductivity, is the charge carrier density, donor density () for n-type or acceptor density () for p-type semiconductors, represents the capacitance of the space charge region, ε is the relative dielectric constant of Cu2O which equals to 6.3 [17], and V is the applied electrode potential (volts). The carrier densities can thus be estimated using the following equation:
(a)
(b)
and calculated using data from Figures 7(a) and 7(b) were found to be 1.90 × 1018 cm−3 and 2.24 × 1016 cm−3 for Sample 1 and Sample 2, respectively. The higher carrier concentration of Sample 1 implies a faster charge transfer and then an enhanced PEC performance [18]. Furthermore, from the intercept of the Mott–Schottky plots, the flat band potentials () for Sample 1 and Sample 2 are −0.73 and 0.42 V versus the reference electrode, respectively. The calculated flat band potentials of both samples are almost consistent with the previous reports [5, 10, 19, 20]. The more negativity and positivity of the flat band potentials of the photoanode and photocathode are essential since the higher the flat band potential results higher band bending and larger space charge region potentials in the semiconductor/electrolyte interfaces.
To explain the use of the two electrodes for water reduction and oxidation, simplified band diagrams of both samples are constructed. Since Sample 1 is n-type, the Fermi level position of Sample 1 is calculated usingwhere is the conduction band minimum of Sample 1. In Equation (3), is the density of donor which can be approximated as the density of electrons which equals to 1.90 × 1018 cm−3. is the effective density of state in the conduction band which can be found aswhere is the effective mass of electrons which equals to 0.99 m0 [21]. The calculation result shows that the Fermi energy level of Sample 1 is at 0.3 eV below conduction band minimum. Similar calculation is carried out for Sample 2 using the following equation:where is the valance band maximum and is the effective density of state in the the valance band of Sample 2 which can be found as
By taking the approximation that equals to hole density and that the effective mass of the hole in Cu2O is 0.58 m0 [21], the Fermi energy position of Sample 2 was found to be 0.9 eV above the valance band maximum.
Figure 8 shows simplified band diagrams of Sample 1 and Sample 2 with energy scales relative to both vacuum and normal hydrogen electrode (). From Figure 8, it shows that the band structures of both samples remain approximately unchanged, while the only difference is the location of the Fermi energy level. One important issue in PEC cell technology is to find semiconductor photoelectrodes that can absorb the visible light and drive hydrogen (oxygen) evolution reaction using the photogenerated electrons (holes). This needs the semiconductor to have proper band alignment relative to the water redox potentials. From Figure 8, it can be found that of Sample 1 (n-type) is at 0.1 eV above , indicating that Sample 1 is suitable to act as a photoanode material. Also from Figure 8, of Sample 2 (p-type) is at 1.1 eV below , indicating that Sample 2 is suitable to act as photocathode. This demonstrates that, by controlling argon/oxygen flow rates, Cu2O deposited by reactive IBSD is suitable to act as both photoanode and photocathode in PEC water splitting applications.
4. Conclusion
Cu2O with novel optical properties and PEC activities has been deposited successfully by reactive IBSD by controlling the argon/oxygen flow rates. PL study shows that n-type Cu2O is due to the presence of oxygen vacancy defects. We demonstrated the bandgap position alignments of the two photoelectrodes with respect to the reduction and oxidation of water, and the result shows that both samples are promising for PEC water splitting applications.
Data Availability
Data supporting this research article are available from the corresponding author upon reasonable.
Conflicts of Interest
The authors declare that they have no conflicts of interest.
Acknowledgments
This research work was supported by the Ministry of Science and Technology of Republic of China under contract number MOST 107-2221-E-011-084.