Photocatalytic and Adsorption Performances of Faceted Cuprous Oxide (Cu2O) Particles for the Removal of Methyl Orange (MO) from Aqueous Media

Particles of sub-micron size possess significant capacity to adsorb organic molecules from aqueous media. Semiconductor photocatalysts in particle form could potentially be utilized for dye removal through either physical adsorption or photo-induced chemical process. The photocatalytic and adsorption capabilities of Cu2O particles with various exposed crystal facets have been studied through separate adsorption capacity test and photocatalytic degradation test. These crystals display unique cubic, octahedral, rhombic dodecahedral, and truncated polyhedral shapes due to specifically exposed crystal facet(s). For comparison, Cu2O particles with no clear exposed facets were also prepared. The current work confirms that the surface charge critically affects the adsorption performance of the synthesized Cu2O particles. The octahedral shaped Cu2O particles, with exposed {111} facets, possess the best adsorption capability of methyl orange (MO) dye due to the strongest positive surface charge among the different types of particles. In addition, we also found that the adsorption of MO follows the Langmuir monolayer mechanism. The octahedral particles also performed the best in photocatalytic dye degradation of MO under visible light irradiation because of the assistance from dye absorption. On top of the photocatalytic study, the stability of these Cu2O particles during the photocatalytic processes was also investigated. Cu(OH)2 and CuO are the likely corrosion products found on the particle surface after the photocorrosion in MO solution. By adding hole scavengers in the solution, the photocorrosion of Cu2O was greatly reduced. This observation confirms that the photocatalytically generated holes were responsible for the photocorrosion of Cu2O.


Introduction
One of the main issues that human society faces in the 21st century is pollution of water and air caused by the increasing industrial activities and human consumption of natural resources including minerals, coal, petroleum and their derivatives. Pollution poses serious threat to not only human health, but also marine life, which in turns affects livelihood of human being through the food chains.
The average size of the Cu 2 O particles with standard deviation is summarized in Table 1. When 0.01 M of NaOH was used, porous Cu 2 O particles of about 0.8 µm in size were obtained (Figure 1a). The particles consisted of agglomerated (non-separable) smaller nanoparticles with mean size around 40 nm. The small nanoparticles have no specific exposed facets, and the agglomerated cluster has an equal axial shape. For convenience, this morphology is denoted as spherical Cu 2 O particles. When NaOH concentration increased from 0.01 to 0.03 M, the particle was transformed into cubic shape with six {100} facets exposed (Figure 1b). To form the cubic shape, the growth rate in the <100> direction must be the slowest among all other growth directions. This was achieved by an increased NaOH concentration together with the addition of ascorbic acid. The obtained cubic Cu 2 O particles show uniform size and sharp edges.

Morphology and Crystal Structure of the Synthesized Cu2O and Surface Area
The average size of the Cu2O particles with standard deviation is summarized in Table 1. When 0.01 M of NaOH was used, porous Cu2O particles of about 0.8 μm in size were obtained (Figure 1a). The particles consisted of agglomerated (non-separable) smaller nanoparticles with mean size around 40 nm. The small nanoparticles have no specific exposed facets, and the agglomerated cluster has an equal axial shape. For convenience, this morphology is denoted as spherical Cu2O particles. When NaOH concentration increased from 0.01 to 0.03 M, the particle was transformed into cubic shape with six {100} facets exposed (Figure 1b). To form the cubic shape, the growth rate in the <100> direction must be the slowest among all other growth directions. This was achieved by an increased NaOH concentration together with the addition of ascorbic acid. The obtained cubic Cu2O particles show uniform size and sharp edges.  of agglomerated spherical shape with no clear exposed facets; and (b) Cu2O of cubic shape with exposed {100} facets. Figure 2 shows the Field Emission Scanning Electron Microscopy (FESEM) image of welldefined, smooth and uniform octahedral shaped Cu2O particles. Each particle has eight exposed {111} facets. The synthesis method was similar to the formation of the cubic shape, except that ascorbic acid was replaced by a stronger reducing agent, hydrazine hydrate.   Figure 2 shows the Field Emission Scanning Electron Microscopy (FESEM) image of well-defined, smooth and uniform octahedral shaped Cu 2 O particles. Each particle has eight exposed {111} facets. The synthesis method was similar to the formation of the cubic shape, except that ascorbic acid was replaced by a stronger reducing agent, hydrazine hydrate.

Morphology and Crystal Structure of the Synthesized Cu2O and Surface Area
The average size of the Cu2O particles with standard deviation is summarized in Table 1. When 0.01 M of NaOH was used, porous Cu2O particles of about 0.8 μm in size were obtained (Figure 1a). The particles consisted of agglomerated (non-separable) smaller nanoparticles with mean size around 40 nm. The small nanoparticles have no specific exposed facets, and the agglomerated cluster has an equal axial shape. For convenience, this morphology is denoted as spherical Cu2O particles. When NaOH concentration increased from 0.01 to 0.03 M, the particle was transformed into cubic shape with six {100} facets exposed (Figure 1b). To form the cubic shape, the growth rate in the <100> direction must be the slowest among all other growth directions. This was achieved by an increased NaOH concentration together with the addition of ascorbic acid. The obtained cubic Cu2O particles show uniform size and sharp edges.  of agglomerated spherical shape with no clear exposed facets; and (b) Cu2O of cubic shape with exposed {100} facets. Figure 2 shows the Field Emission Scanning Electron Microscopy (FESEM) image of welldefined, smooth and uniform octahedral shaped Cu2O particles. Each particle has eight exposed {111} facets. The synthesis method was similar to the formation of the cubic shape, except that ascorbic acid was replaced by a stronger reducing agent, hydrazine hydrate.   Both oleic acid and glucose were used in the synthesis. Apparently, they are able to reduce the growth along the <110> direction, resulting the exposed facets.  Figure 3 shows flat and smooth surfaces of rhombic dodecahedral Cu2O with 12 {110} facets. Both oleic acid and glucose were used in the synthesis. Apparently, they are able to reduce the growth along the <110> direction, resulting the exposed facets.   5 h), the size of the smooth and uniform truncated polyhedral Cu2O particle was seen to be relatively large (5 μm) compared to the other particles. It was observed that by lowering the concentration of the precursors and shortening the cooling time, some irregular cubic shaped Cu2O particles were obtained with an average size of 2 μm (not shown here). Therefore, both factors, the concentration and the cooling rate, are important to synthesize the truncated polyhedral Cu2O particles.   Table 1. Not surprisingly, the spherical shaped Cu2O yields the highest surface area. The one for the truncated polyhedral Cu2O could not be measured, due to the large size of the particles exceeding the detection limit of the equipment (~2 μm).
The crystal structure of various shaped Cu2O synthesized by the wet chemical method is confirmed by the X-ray diffraction (XRD) pattern shown in Figure 6. The diffraction peaks labeled with "★" matches well with the standard data of Cu2O (JCPDS No. 05-0667), confirming the   5 h), the size of the smooth and uniform truncated polyhedral Cu 2 O particle was seen to be relatively large (5 µm) compared to the other particles. It was observed that by lowering the concentration of the precursors and shortening the cooling time, some irregular cubic shaped Cu 2 O particles were obtained with an average size of 2 µm (not shown here). Therefore, both factors, the concentration and the cooling rate, are important to synthesize the truncated polyhedral Cu 2 O particles.  Figure 3 shows flat and smooth surfaces of rhombic dodecahedral Cu2O with 12 {110} facets. Both oleic acid and glucose were used in the synthesis. Apparently, they are able to reduce the growth along the <110> direction, resulting the exposed facets.   5 h), the size of the smooth and uniform truncated polyhedral Cu2O particle was seen to be relatively large (5 μm) compared to the other particles. It was observed that by lowering the concentration of the precursors and shortening the cooling time, some irregular cubic shaped Cu2O particles were obtained with an average size of 2 μm (not shown here). Therefore, both factors, the concentration and the cooling rate, are important to synthesize the truncated polyhedral Cu2O particles.   Table 1. Not surprisingly, the spherical shaped Cu2O yields the highest surface area. The one for the truncated polyhedral Cu2O could not be measured, due to the large size of the particles exceeding the detection limit of the equipment (~2 μm).
The crystal structure of various shaped Cu2O synthesized by the wet chemical method is confirmed by the X-ray diffraction (XRD) pattern shown in Figure 6. The diffraction peaks labeled with "★" matches well with the standard data of Cu2O (JCPDS No. 05-0667), confirming the   Table 1. Not surprisingly, the spherical shaped Cu 2 O yields the highest surface area. The one for the truncated polyhedral Cu 2 O could not be measured, due to the large size of the particles exceeding the detection limit of the equipment (~2 µm).
The crystal structure of various shaped Cu 2 O synthesized by the wet chemical method is confirmed by the X-ray diffraction (XRD) pattern shown in Figure 6. The diffraction peaks labeled with " " matches well with the standard data of Cu 2 O (JCPDS No. 05-0667), confirming the formation of a single cubic phase structure. All the peaks were perfectly indexed to crystalline Cu 2 O not only in their peak positions (2θ values of 29.60 • , 36.52 • , 42.44 • , 61.40 • and 73.55 • ), but also in their relative intensity. It was observed that the XRD peaks for all the faceted samples were relatively sharp, due to the good crystallinity of the crystals. However, the crystallinity for the spherical shaped Cu 2 O particles was poor, as indicated by the weak and broad peaks. formation of a single cubic phase structure. All the peaks were perfectly indexed to crystalline Cu2O not only in their peak positions (2θ values of 29.60°, 36.52°, 42.44°, 61.40° and 73.55°), but also in their relative intensity. It was observed that the XRD peaks for all the faceted samples were relatively sharp, due to the good crystallinity of the crystals. However, the crystallinity for the spherical shaped Cu2O particles was poor, as indicated by the weak and broad peaks.   formation of a single cubic phase structure. All the peaks were perfectly indexed to crystalline Cu2O not only in their peak positions (2θ values of 29.60°, 36.52°, 42.44°, 61.40° and 73.55°), but also in their relative intensity. It was observed that the XRD peaks for all the faceted samples were relatively sharp, due to the good crystallinity of the crystals. However, the crystallinity for the spherical shaped Cu2O particles was poor, as indicated by the weak and broad peaks.

FTIR and Zeta Potential Analyses
The Fourier Transform Infrared (FTIR) spectra for all the five samples synthesized are shown in Figure 7. Though the spectrum was run from 400 to 4000 cm −1 , the bands from metal oxide are generally below 1000 cm −1 [20]. For Cu 2 O, the bands that lie at 1130, 798 and 621 cm −1 were attributed to the stretching vibration of Cu-O in Cu 2 O. Thus, the results obtained have proven the existence of Cu 2 O. No bands related to CuO, which would appear at 588, 534 and 480 cm −1 [21], were detected. The use of oleic acid may lead to carboxylate (C=O stretch) on the surface of the particles which would show in the FTIR spectra at 1710 cm −1 . Based on Figure 7, the peak for the carboxylate could not be detected across all the synthesized Cu 2 O particles. This clearly proves that the surfaces of the Cu 2 O particles are ligand free.
The zeta potentials for all the Cu 2 O samples were measured and the results are tabulated in Table 2. The surface charge of the spherical, cubic and octahedral shaped Cu 2 O was positively charged (indicated by the "+" sign) while the surfaces of the rhombic dodecahedral and truncated polyhedral Cu 2 O were negatively charged (indicated by the "−" sign). The surface charge is related to the surface bonding state on the exposed facets. Our observation that the octahedral particles with exposed {111} planes possess the most positive surface charge indicates that there are probably more un-coordinated Cu bonds on the particle surfaces. On the other hand, the surface charge of the rhombic dodecahedral Cu 2 O particles consisting of 12 {110} planes was found to be negative, suggesting the O ions are likely to be dominant on the exposed surface. This result is contrary to some previous reports [16,18] that claim the rhombic dodecahedral particles are positively charged based on visualizing the atomic models of Cu 2 O. Strictly speaking, this approach (cutting and visualizing a particular crystal model) cannot deduce the information on particle surface charge.

FTIR and Zeta Potential Analyses
The Fourier Transform Infrared (FTIR) spectra for all the five samples synthesized are shown in Figure 7. Though the spectrum was run from 400 to 4000 cm −1 , the bands from metal oxide are generally below 1000 cm −1 [20]. For Cu2O, the bands that lie at 1130, 798 and 621 cm −1 were attributed to the stretching vibration of Cu-O in Cu2O. Thus, the results obtained have proven the existence of Cu2O. No bands related to CuO, which would appear at 588, 534 and 480 cm −1 [21], were detected. The use of oleic acid may lead to carboxylate (C=O stretch) on the surface of the particles which would show in the FTIR spectra at 1710 cm −1 . Based on Figure 7, the peak for the carboxylate could not be detected across all the synthesized Cu2O particles. This clearly proves that the surfaces of the Cu2O particles are ligand free.
The zeta potentials for all the Cu2O samples were measured and the results are tabulated in Table 2. The surface charge of the spherical, cubic and octahedral shaped Cu2O was positively charged (indicated by the "+" sign) while the surfaces of the rhombic dodecahedral and truncated polyhedral Cu2O were negatively charged (indicated by the "−" sign). The surface charge is related to the surface bonding state on the exposed facets. Our observation that the octahedral particles with exposed {111} planes possess the most positive surface charge indicates that there are probably more un-coordinated Cu bonds on the particle surfaces. On the other hand, the surface charge of the rhombic dodecahedral Cu2O particles consisting of 12 {110} planes was found to be negative, suggesting the O ions are likely to be dominant on the exposed surface. This result is contrary to some previous reports [16,18] that claim the rhombic dodecahedral particles are positively charged based on visualizing the atomic models of Cu2O. Strictly speaking, this approach (cutting and visualizing a particular crystal model) cannot deduce the information on particle surface charge.

The Adsorption Performance of Cu 2 O Particles
The adsorption performance of the Cu 2 O particles was evaluated using solution of MO, a negatively charged dye ( Figure 8). It was observed that Cu 2 O particles with positive surface charges (spherical, cubic and octahedral) were able to adsorb MO due to the electrostatic force. The adsorption was almost complete after 300 min for the octahedral Cu 2 O. Both the rhombic dodecahedral and truncated Cu 2 O particles possess negative charges, and there was nearly no adsorption due to electrostatic repulsion. Based on the zeta potential results shown earlier (Table 2), it was observed that the octahedral sample was more positively charged, followed by the cubic and then the spherical structures. The sequence agrees well with the adsorption performance despite the difference in the specific area: although the spherical sample possessed a higher surface area, the adsorption was observed to be slower and no further adsorption could be seen after 360 min. Hence, it is proven that the surface charge critically affects the adsorption performance of the synthesized Cu 2 O particles.
As the adsorption performance of octahedral sample was evidently the best, it is of interest to further understand the adsorption mechanism and capacity of this material. Figure 9 presents the adsorption isotherm of the Cu 2 O octahedral shaped sample.

The Adsorption Performance of Cu2O Particles
The adsorption performance of the Cu2O particles was evaluated using solution of MO, a negatively charged dye (Figure 8). It was observed that Cu2O particles with positive surface charges (spherical, cubic and octahedral) were able to adsorb MO due to the electrostatic force. The adsorption was almost complete after 300 min for the octahedral Cu2O. Both the rhombic dodecahedral and truncated Cu2O particles possess negative charges, and there was nearly no adsorption due to electrostatic repulsion. Based on the zeta potential results shown earlier (Table 2), it was observed that the octahedral sample was more positively charged, followed by the cubic and then the spherical structures. The sequence agrees well with the adsorption performance despite the difference in the specific area: although the spherical sample possessed a higher surface area, the adsorption was observed to be slower and no further adsorption could be seen after 360 min. Hence, it is proven that the surface charge critically affects the adsorption performance of the synthesized Cu2O particles.
As the adsorption performance of octahedral sample was evidently the best, it is of interest to further understand the adsorption mechanism and capacity of this material. Figure 9 presents the adsorption isotherm of the Cu2O octahedral shaped sample.

The Adsorption Performance of Cu2O Particles
The adsorption performance of the Cu2O particles was evaluated using solution of MO, a negatively charged dye ( Figure 8). It was observed that Cu2O particles with positive surface charges (spherical, cubic and octahedral) were able to adsorb MO due to the electrostatic force. The adsorption was almost complete after 300 min for the octahedral Cu2O. Both the rhombic dodecahedral and truncated Cu2O particles possess negative charges, and there was nearly no adsorption due to electrostatic repulsion. Based on the zeta potential results shown earlier (Table 2), it was observed that the octahedral sample was more positively charged, followed by the cubic and then the spherical structures. The sequence agrees well with the adsorption performance despite the difference in the specific area: although the spherical sample possessed a higher surface area, the adsorption was observed to be slower and no further adsorption could be seen after 360 min. Hence, it is proven that the surface charge critically affects the adsorption performance of the synthesized Cu2O particles.
As the adsorption performance of octahedral sample was evidently the best, it is of interest to further understand the adsorption mechanism and capacity of this material. Figure 9 presents the adsorption isotherm of the Cu2O octahedral shaped sample.   The curve is fitted into the well-known Freundlich (Equation (1)) and Langmuir (Equation (2)) models: Freundlich : q = K f C Langmuir : q = q m KC 1 + KC (2) where q (mg/g) is the amount of adsorbed MO, C the concentration of MO at equilibrium, q m (mg/g) the maximum adsorption capacity, K f and n are Freundlich constants and K is the Langmuir constant. The amount of MO adsorbed onto the photocatalyst, q, is given by: where C o and C f represent the initial and final concentration of MO in the solution, V is the volume of MO solution (L) and M is the mass of photocatalyst added (g). The Freundlich and Langmuir models can be linearized as: Langmuir : If the adsorption isotherm exhibits Langmuir behavior, it indicates monolayer adsorption. In contrast, a good fit into the Freundlich model indicates a heterogeneous surface binding. The results of fitting the isotherm curves to Freundlich and Langmuir models are summarized in Table 3. The fitting is clearly better for the Langmuir model. On the other hand, the R 2 value is less than 0.8 when the data is fitted with the Freundlich model. This illustrates that the adsorption of MO by the octahedral Cu 2 O is governed by the monolayer adsorption. The MO adsorbed (q m ) by the octahedral shaped Cu 2 O sample was found to be 96.42 mg/g (experimentally) and 66.0 mg/g (theoretically). The values are comparable with some reported adsorbents such as rice husk (40.58 mg/g) [22] and raw date pits (80.29 mg/g) [23].

The Photocatalytic Performance of Cu 2 O Particles
The photocatalytic activity of the as-prepared Cu 2 O samples was evaluated by degradation of MO under visible light irradiation. Photolysis test was also carried out as a reference. Water loss due to evaporation was calibrated when reporting the dye concentration. Figure 10 shows the photocatalytic performance. When assessing the photocatalytic activity, reference has to be made with the adsorption experiment as shown in Figure 8. Through comparison, it is evident that the octahedral and cubic shaped Cu 2 O particles demonstrated substantial photocatalytic degradation of MO under visible light irradiation. The demonstrated good activity is attributed to the adsorption of the dye to the surface before the photo degradation occurs.
The rhombic dodecahedral and truncated Cu 2 O showed limited activity despite the fact that their negative surface charge repels the dye molecules. Huang et al. [24] have attributed the poor photocatalytic activity of the rhombic dodecahedral shaped Cu 2 O to the surface residues from the oleic acid used during the synthesis. However, as discussed earlier, in the current study, no FTIR peaks related to oleic acid could be found. Therefore, our current work confirms that the rhombic dodecahedral Cu 2 O indeed possesses some degree of photocatalytic activity. peaks related to oleic acid could be found. Therefore, our current work confirms that the rhombic dodecahedral Cu2O indeed possesses some degree of photocatalytic activity. No obvious dye degradation was observed for spherical particles once the adsorption curve is plotted together with the photocatalytic one, as shown in Figure 11. The poor crystallinity could be the main reason, which has resulted in the fast recombination of the photo-generated electron-hole pairs.

Band Gap, Band Edge Positions and Proposed Photodegradation Mechanism
To determine the flatband potential, Mott-Schottky graph was obtained by measuring the apparent capacitance as a function of potential under depletion condition at the semiconductorelectrolyte junction based on: No obvious dye degradation was observed for spherical particles once the adsorption curve is plotted together with the photocatalytic one, as shown in Figure 11. The poor crystallinity could be the main reason, which has resulted in the fast recombination of the photo-generated electron-hole pairs. peaks related to oleic acid could be found. Therefore, our current work confirms that the rhombic dodecahedral Cu2O indeed possesses some degree of photocatalytic activity. No obvious dye degradation was observed for spherical particles once the adsorption curve is plotted together with the photocatalytic one, as shown in Figure 11. The poor crystallinity could be the main reason, which has resulted in the fast recombination of the photo-generated electron-hole pairs.

Band Gap, Band Edge Positions and Proposed Photodegradation Mechanism
To determine the flatband potential, Mott-Schottky graph was obtained by measuring the apparent capacitance as a function of potential under depletion condition at the semiconductorelectrolyte junction based on:

Band Gap, Band Edge Positions and Proposed Photodegradation Mechanism
To determine the flatband potential, Mott-Schottky graph was obtained by measuring the apparent capacitance as a function of potential under depletion condition at the semiconductor-electrolyte junction based on: where C sc is the capacitance of the space charged region, e the electron charge (1.602 × 10 −19 C), ε the diaelectric constant of the semiconductor, ε 0 the permittivity of free space (8.85 × 10 −14 F cm −1 ), N the donor density, E the applied potential, E fb the flatband potential, k the Boltzmann constant (1.38 × 10 −23 J K −1 ), and T the absolute temperature. Extrapolation of the linear plot to the applied potential axis leads to the value for E fb . Table 4 summaries the valence band potential position, the type of semiconductor, and the optical bandgap values that were determined by the optical absorption measurement.  Figure 12 shows the band edge positions (with reference to NHE at pH~7) and band gap energy of different Cu 2 O samples. The measured bandgap value only slightly varies from 1.9 to 2.0 eV, and the largest difference in the valence band position is about 0.19 eV. The difference in the band gap energy and band potential positions could be due to the size and the specific exposed facet(s) [25]. It was reported that the main oxidation species for MO degradation are • OH radicals under our experimental pH condition [26]. In general, there are two routes that the • OH radicals are generated, viz., through the photo-generated holes in the valence band or the photo-generated electrons in the conduction band. However, none of the particles are able to produce • OH radicals directly from the photo-generated holes as the required semiconductor valence band potential is +1.58 eV for pH = 7 (+1.99 eV vs. NHE [27]), far more anodic than valence band potentials of all these Cu 2 O particles. Therefore, the mechanism of • OH radical generation should be through the photogenerated electrons from the conduction band. The electrons first react with adsorbed oxygen to generate O 2 −• and H 2 O 2 , which then react to form • OH radicals [25]. In addition to the photocatalytic action discussed above, there is another possible dye degradation mechanism through the so-called photo-assisted degradation. Under such mechanism, the electrons generated by the dyes after light adsorption are injected to the conduction band of the photocatalyst. The electrons generated by the dye itself, rather than by the photocatalyst in a typical photocatalytic reaction, will cause chemical destruction of the dye molecules [28]. The LUMO (lowest unoccupied molecular orbital) energy of MO is around −1.636 eV with reference to NHE at~pH 7 [29], so it is possible for the electrons to be injected from MO to the conduction band of Cu 2 O, leading to the photo-assisted degradation of MO itself. Because the energy gap between the LUMO and HOMO (highest occupied molecular orbital) of MO is about 2.35 eV [29], which is larger than the band gap of the Cu 2 O samples (1.9-2.0 eV), it becomes impossible to choose a light illumination that only excites the dye but not the photocatalyst-such experiment would be able to differentiate the photocatalytic action vs. the photo-assisted degradation. However, based on the fact that the spherical samples did not degrade MO, it is reasonable to deduce that the photo-assisted degradation is not likely to be a main contributing mechanism since it would not be affected by the charge recombination in the photocatalyst. In other words, if the photo-assisted degradation does exist, it should be manifested through the spherical particles.

Stability of the Synthesized Cu2O Particles
The stability of Cu2O is determined via the solar light illumination of intensity 100 mW/cm 2 for 12 h when the particles were dispersed in MO solution. The samples were collected and analyzed using scanning electron microcopy (SEM), X-ray photoelectron microcopy (XPS) and FTIR. Comparison was made with 25% methanol (v/v) added to the solution to understand the source of the photocorrosion. Figure 13 compares the morphology of the faceted Cu2O particles at the as-synthesized state (column 1), after light illumination in the (MO + methanol) solution (column 2), and after light illumination in the MO solution (column 3). Comparing the first and the third columns, it is observed that the spherical shaped (Figure 14a) Cu2O particles do not seem to have suffered much photocorrosion. In relation to nearly zero photocatalytic activity as reported earlier, it strongly suggests that corrosion of the Cu2O is related to photo-generated species.
In the case of the cubic and octahedral shaped Cu2O particles (Figure 13b,c), small amount of precipitates could be observed on the surface. For the rhombic dodecahedral Cu2O particles (Figure  13d), the precipitates could be seen on the surface of the particles. On the other hand, the truncated polyhedral Cu2O (Figure 13e) particles suffered from minimum photocorrosion as only a very small amount of precipitates could be found on the surface.
The addition of methanol, a hole scavenger, in the MO solution has clearly alleviated the severity of photocorrosion (comparing columns 2 and 3). This suggests that the photocatalytically generated holes are responsible for the photocorrosion of Cu2O.
To determine the corrosion mechanism, we took the octahedral sample as a representative and exposed it under the solar light irradiation for an extended period of 120 h. After that, XPS and FTIR were employed in the analysis of the corrosion products on the particle surface. XRD was also explored but the intensity of the corrosion products was not strong enough for a firm identification.
Possible photocorrosion products of Cu2O include Cu, Cu(OH)2 and CuO. As the binding energies for Cu 0 and Cu 1+ are 932.67 eV and 932.6 eV, respectively [30], it is difficult to determine the phase of the photocorrosion products through Cu 2p3/2 scan. As such, XPS scan for the Auger peak of Cu LMM is needed to differentiate between Cu 0 and Cu 1+ . The Auger LMM energy for Cu 0 and Cu 1+ lie in 334.95 eV and 336.80 eV respectively [31]. Figure 14 shows the Auger LMM spectra of octahedral shaped Cu2O particles which have Auger LMM energy of 341 eV. This Auger energy generated corresponds to the Auger energy of Cu 1+ . Figure 15 shows the Cu 2p3/2 XPS spectra of the Cu2O particles. The main peaks at 932.7 and 935.4 eV correspond to the binding energies of Cu2O and Cu(OH)2 respectively [31,32]. Lastly, the satellite peaks on the higher binding energy, 943.6 eV, indicates the presence of an unfilled Cu 3d shell and thus confirms the existence of Cu 2+ on the sample surface.

Stability of the Synthesized Cu 2 O Particles
The stability of Cu 2 O is determined via the solar light illumination of intensity 100 mW/cm 2 for 12 h when the particles were dispersed in MO solution. The samples were collected and analyzed using scanning electron microcopy (SEM), X-ray photoelectron microcopy (XPS) and FTIR. Comparison was made with 25% methanol (v/v) added to the solution to understand the source of the photocorrosion. Figure 13 compares the morphology of the faceted Cu 2 O particles at the as-synthesized state (column 1), after light illumination in the (MO + methanol) solution (column 2), and after light illumination in the MO solution (column 3). Comparing the first and the third columns, it is observed that the spherical shaped (Figure 14a) Cu 2 O particles do not seem to have suffered much photocorrosion. In relation to nearly zero photocatalytic activity as reported earlier, it strongly suggests that corrosion of the Cu 2 O is related to photo-generated species.
In the case of the cubic and octahedral shaped Cu 2 O particles (Figure 13b,c), small amount of precipitates could be observed on the surface. For the rhombic dodecahedral Cu 2 O particles (Figure 13d), the precipitates could be seen on the surface of the particles. On the other hand, the truncated polyhedral Cu 2 O (Figure 13e) particles suffered from minimum photocorrosion as only a very small amount of precipitates could be found on the surface.
The addition of methanol, a hole scavenger, in the MO solution has clearly alleviated the severity of photocorrosion (comparing columns 2 and 3). This suggests that the photocatalytically generated holes are responsible for the photocorrosion of Cu 2 O.
To determine the corrosion mechanism, we took the octahedral sample as a representative and exposed it under the solar light irradiation for an extended period of 120 h. After that, XPS and FTIR were employed in the analysis of the corrosion products on the particle surface. XRD was also explored but the intensity of the corrosion products was not strong enough for a firm identification.
Possible photocorrosion products of Cu 2 O include Cu, Cu(OH) 2 and CuO. As the binding energies for Cu 0 and Cu 1+ are 932.67 eV and 932.6 eV, respectively [30], it is difficult to determine the phase of the photocorrosion products through Cu 2p 3/2 scan. As such, XPS scan for the Auger peak of Cu LMM is needed to differentiate between Cu 0 and Cu 1+ . The Auger LMM energy for Cu 0 and Cu 1+ lie in 334.95 eV and 336.80 eV respectively [31]. Figure 14 shows the Auger LMM spectra of octahedral shaped Cu 2 O particles which have Auger LMM energy of 341 eV. This Auger energy generated corresponds to the Auger energy of Cu 1+ . Figure 15 shows the Cu 2p 3/2 XPS spectra of the Cu 2 O particles. The main peaks at 932.7 and 935.4 eV correspond to the binding energies of Cu 2 O and Cu(OH) 2 respectively [31,32]. Lastly, the satellite peaks on the higher binding energy, 943.6 eV, indicates the presence of an unfilled Cu 3d shell and thus confirms the existence of Cu 2+ on the sample surface.    The XPS confirms that the top surface (4-8 nm) of the Cu2O particles contain Cu(OH)2 precipitates. To explore possible photocorrosion products beneath the top layer, FTIR measurement was used. As shown in Figure 16a, the broad bands centered at 3436 and 1638 cm −1 are attributed to the O-H stretching and bending modes of water [16,33]. The peaks located at 631, 809 and 1156 cm −1 are attributed to the stretching vibration of Cu-O in Cu2O.
When zooming into the details in the range of 400 to 800 cm −1 , the adsorption band at 530 cm −1 associated with CuO is revealed (Figure 16b). The low intensity suggests that the amount of CuO is relatively small.
Based on the above analyses, the photocorrosion precipitates formed on the surface of Cu2O particles are mainly Cu(OH)2 and a small amount of CuO. This finding agrees with our early analysis that the photo-generated electrons are responsible for the MO degradation, while the holes are left to oxidize Cu + to Cu 2+ state. The presence of the photocorrosion products will hinder the photocatalytic activity of Cu2O as they block the light from reaching the Cu2O surface. Adding hole scavenger has alleviated the photocorrosion, and this observation also confirms that the holes are responsible for the photocorrosion of Cu2O.  The XPS confirms that the top surface (4-8 nm) of the Cu2O particles contain Cu(OH)2 precipitates. To explore possible photocorrosion products beneath the top layer, FTIR measurement was used. As shown in Figure 16a, the broad bands centered at 3436 and 1638 cm −1 are attributed to the O-H stretching and bending modes of water [16,33]. The peaks located at 631, 809 and 1156 cm −1 are attributed to the stretching vibration of Cu-O in Cu2O.
When zooming into the details in the range of 400 to 800 cm −1 , the adsorption band at 530 cm −1 associated with CuO is revealed (Figure 16b). The low intensity suggests that the amount of CuO is relatively small. Based on the above analyses, the photocorrosion precipitates formed on the surface of Cu2O particles are mainly Cu(OH)2 and a small amount of CuO. This finding agrees with our early analysis that the photo-generated electrons are responsible for the MO degradation, while the holes are left to oxidize Cu + to Cu 2+ state. The presence of the photocorrosion products will hinder the photocatalytic activity of Cu2O as they block the light from reaching the Cu2O surface. Adding hole scavenger has alleviated the photocorrosion, and this observation also confirms that the holes are responsible for the photocorrosion of Cu2O. The XPS confirms that the top surface (4-8 nm) of the Cu 2 O particles contain Cu(OH) 2 precipitates. To explore possible photocorrosion products beneath the top layer, FTIR measurement was used. As shown in Figure 16a, the broad bands centered at 3436 and 1638 cm −1 are attributed to the O-H stretching and bending modes of water [16,33]. The peaks located at 631, 809 and 1156 cm −1 are attributed to the stretching vibration of Cu-O in Cu 2 O.
When zooming into the details in the range of 400 to 800 cm −1 , the adsorption band at 530 cm −1 associated with CuO is revealed (Figure 16b). The low intensity suggests that the amount of CuO is relatively small. Based on the above analyses, the photocorrosion precipitates formed on the surface of Cu 2 O particles are mainly Cu(OH) 2 and a small amount of CuO. This finding agrees with our early analysis that the photo-generated electrons are responsible for the MO degradation, while the holes are left to oxidize Cu + to Cu 2+ state. The presence of the photocorrosion products will hinder the photocatalytic activity of Cu 2 O as they block the light from reaching the Cu 2 O surface. Adding hole scavenger has alleviated the photocorrosion, and this observation also confirms that the holes are responsible for the photocorrosion of Cu 2 O.

Spherical Cu 2 O Particles
Copper(II) acetate (Cu(OAc) 2 ) (0.01 mol) was dissolved into 20 mL of deionized water under constant stirring at 500 rpm. Forty milliliters of 0.01 M NaOH was then added to the solution followed by 0.01 M of ascorbic acid (20 mL). After stirring for 30 min, the solution was centrifuged at 6000 rpm for 3 min and then washed with deionized water and ethanol solution. The powder collected was then dried in a vacuum oven at 60 • C for 6 h.

Cubic Cu 2 O Particles
Copper(II) acetate (Cu(OAc) 2 ) (0.01 mol) was dissolved in 20 mL of deionized water under constant stirring at 500 rpm. Forty milliliters NaOH solution (0.03 M) was then added, followed by 20 mL of 0.01 M ascorbic acid. After stirring for 30 min, the solution was centrifuged at 6000 rpm for 3 min followed by washing with deionized water and ethanol solution. The powder collected was then dried in a vacuum oven at 60 • C for 6 h.

Octahedral Cu 2 O Particles
Copper(II) acetate (Cu(OAc) 2 ) (0.01 mol) was dissolved in 20 mL of deionized water under constant stirring at 500 rpm. Forty milliliters NaOH solution (0.03 M) was then added to the solution followed by addition of 40 mL hydrazine hydrate solution (0.05 M). After 45 min of stirring, the solution was centrifuged at 6000 rpm for 3 min and washed with deionized water and ethanol solution. The powder collected was dried in a vacuum oven at 60 • C for 6 h.

Rhombic Dodecahedral Cu 2 O Particles
Copper(II) sulfate (CuSO 4 ·5H 2 O) (0.25 g) was dissolved in 40 mL deionized water. When the powder was fully dissolved, 5 mL oleic acid and 20 mL absolute ethanol were added to the solution. The solution was vigorously stirred at 700 rpm for 30 min before being heated at 90 • C. Twenty milliliters NaOH solution (0.08 M) was then added to the CuSO 4 solution under constant stirring for 5 min at 90 • C. Afterwards, 3.42 g of glucose, which was dissolved in 30 mL of heated deionized water, was added to the solution. The final solution was left to stir at 90 • C for 45 min. After that, the solution was centrifuged and washed with hexane (10 times) and ethanol (3 times). The powder collected was then dried in a vacuum oven at 60 • C for 6 h.

Truncated Polyhedral Cu 2 O Particles
Copper(II) acetate (Cu(OAc) 2 ) (0.015 mol) was dissolved in 40 mL deionized water at 70 • C. After continuous stirring at 500 rpm for 5 min, 10 mL of 9.0 M NaOH and 0.6 g glucose were added into the solution. The solution continued to be stirred at 70 • C for about 60 min before being cooled down to room temperature. The powder was washed and centrifuged with ethanol (6 times) and deionized water (3 times). Finally, the powder was dried in a vacuum oven at 60 • C for 6 h.

Materials Characterization
Crystal structure was identified by X-ray diffraction (XRD) using a Shimadzu LabX-6000 diffractometer (Shimadzu Corporation, Tokyo, Japan) with Cu Kα radiation (λ = 1.54178 Å). A step size of 0.02 • over 2θ ranging from 10 • to 80 • was used with a scanning rate at 2.33 • per minute. The accelerating voltage and emission current were 40 kV and 30 mA, respectively.
The morphology of the samples was examined by field emission scanning electron microscopy (FESEM, JEOL JSM-7600F, JEOL Ltd., Tokyo, Japan) and transmission electron microscopy (TEM, JEOL JEM-2010, JEOL Ltd., Tokyo, Japan). The specific surface areas were evaluated using a Micromeritics ASAP 2020 (Micromeritics Instrument Corporation, Norcross, GA, USA) surface analyzer based on the BET theory. The samples were outgassed under vacuum and heated to 100 • C before the test.
The surface chemical analysis was carried out by X-ray photoelectron spectroscopy (XPS) using a VG ESCALab 220i-XL system (Thermo Scientific, Waltham, MA, USA). Mg Kα X-ray (hν = 1253.6 eV) from twin anode X-ray gun was employed using a large area lens mode for analysis with photoelectron takeoff angle of 90 • with respect to surface plane. The maximum analysis depth is in the range of 4-8 nm. Survey spectra were acquired for elemental identification while high-resolution spectra were acquired for chemical state identification and surface composition calculation. For chemical state analysis, a spectral deconvolution was performed by a curve-fitting procedure based on Lorentzians broadened by Gaussian using the manufacturer's standard software. The error of binding energy is estimated to be within 0.2 eV.
FTIR was carried out in a Perkin Elmer Instruments Spectrum GX FTIR spectrometer (PerkinElmer, Waltham, MA, USA). Synthesized particles were mixed with standard KBr particles and pressed into thin pellets. The spectral range was 400-4000 cm −1 and a total of 40 scans were recorded at a resolution of 4 cm −1 averaging each spectrum.
Zeta potential was measured by a Mavern Nanosizer system (Malvern Instruments Ltd, Malvern, UK). Particles (5-10 mg) were dispersed in deionized water (~pH 7) and sonicated for 5 min. The equilibrium time was set at 120 s and each sample was run for 10 times to obtain the average value of the surface charge.
Optical absorption of bulk powders was measured on a Perkin Elmer Lambda 900 UV-Visible spectrometer (PerkinElmer, Waltham, MA, USA) in the diffuse reflectance spectroscopy mode over the spectra ranging from 250 to 800 nm. The optical diffuse reflectance spectrum of Cu 2 O was measured on a Shimadzu 2550 UV-vis-NIR spectrometer (Shimadzu Corporation, Tokyo, Japan) using BaSO 4 as a reference standard. The bandgap of Cu 2 O was calculated using the Kubelka-Munk function.
The flat band potentials were measured by impedance spectroscopy based on the Mott-Schottky plots. To prepare for the test samples, 15 mg of the as-synthesized Cu 2 O powder was sonicated in 1 mL of ethanol to obtain a homogeneous mixture. The Cu 2 O suspension was then drop-casted on a conductive fluorine-tin oxide (FTO) glass substrate with adhesive tapes acting as spacers attached on the four edges. The substrate was then dried at 80 • C, and the adhesive tape attached on the top side of the substrate was removed. Electrical contact was formed by first applying silver paint on the top uncoated area of FTO, and then sticking a conductive copper tape onto the dried silver paint. Three electrodes were used for the impedance measurements which include the working electrode (the Cu 2 O film), a Pt counter electrode, and a reference electrode (Ag/AgCl, saturated in KCl). A 0.1 M Na 2 SO 4 solution was used as the electrolyte. The measurements were carried out by a Gamry electrochemical impedance spectrometer, and the potential was systemically varied between +0 V and +2.0 V with frequency of 50 Hz.

Photocatalytic and Adsorption Experiments
Methyl orange (MO) was chosen as a representative dye to test the photocatalytic degradation activity and the adsorption capacity of the prepared Cu 2 O samples. Solution of the dye was prepared by dissolving the dye in deionized water at 20 ppm concentration, and the solution pH was around 6.7. One hundred mg of the powder samples were dispersed in 50 mL MO solution for the dye degradation test under visible light irradiation. A 100 mL glass beaker, wrapped with aluminum foil on its side wall, was used as the reactor with light shone from the top. The irradiation source comes from a solar simulator equipped with a 300 W Xe-lamp (HAL-320, Asahi Spectra Co., Ltd., Kita-ku, Japan). Super cold filter (YSC0750) was used to provide visible light ranging from 420 to 700 nm. The light intensity was around 50 mW/cm 2 . The amount of photocatalyst used was chosen through a few trial runs based on their degradation speed.
An adsorption isotherm test was carried out in the dark to prevent the potential photocatalytic degradation of MO under light. The equilibrium adsorption isotherm was determined using various concentrations of MO (10, 20, 30, 50 and 80 ppm). For each test, 100 mg of adsorbent was added to 50 mL MO solution. After 48 h, the equilibrium concentration was measured.
To determine the adsorption capacity, 100 mg of synthesized powder was added to 50 mL MO solution of various concentrations (10, 20, 30, 50 and 80 ppm). Stirring was applied throughout the duration of the test (1440 min) at a speed of 400 rpm in dark. At different intervals, the sample was collected, centrifuged and measured using the UV-visible spectrometer to determine the dye concentration.

Photocatalytic Stability Study
Two separate tests were carried out to measure the photocatalytic stability of Cu 2 O particles and to determine the possible root cause of the instability. In the first experiment, 50 mg powder was dispersed in 50 mL MO solution with 10 ppm concentration. For the second experiment, 50 mg of the powder was dispersed in 40 mL of MO solution (10 ppm) mixed with 10 mL of absolute methanol (purity > 99%). Both were constantly stirred under a solar simulator equipped with a 300 W Xe-lamp (HAL-320, Asahi Spectra Co., Ltd., Kita-ku, Japan) for 12 h under the intensity of 100 mW/cm 2 . An AM 1.5 G filter (400 to 1100 nm) was used.

Conclusions
In this paper, we have studied the adsorption and photocatalytic performances of the various faceted Cu 2 O samples for MO removal from its solution. The adsorption capability of Cu 2 O particles was found to be mainly determined by the surface charge. The octahedral shaped Cu 2 O particles with exposed {111} facets performed the best due to its most positive surface charges. The adsorption of the octahedral shaped Cu 2 O was found to follow the Langmuir monolayer mechanism.
The spherical shaped Cu 2 O without clearly defined facets did not display photocatalytic activity. All the faceted samples showed different degree of photocatalytic activities. The octahedral shaped Cu 2 O particles with exposed {111} facets performed the best in photocatalytic degradation of MO under visible light. The photo-generated electrons are responsible for the degradation of MO solution, while the photo-generated holes attack Cu 2 O, causing photocorrosion. It was observed that the corrosion precipitates are mainly Cu(OH) 2 , together with a small amount of CuO. These photocorrosion products hinder the photocatalytic activity of Cu 2 O and thus shorten the service life of Cu 2 O particles. The addition of hole scavengers such as methanol has shown to have alleviated the corrosion attack on Cu 2 O.