Photocatalytic Recovery of Gold from a Non-Cyanide Gold Plating Solution as Au Nanoparticle-Decorated Semiconductors.

In this work, a photocatalytic process was carried out to recover gold (Au) from the simulated non-cyanide plating bath solution. Effects of semiconductor types (TiO2, WO3, Nb2O3, CeO2, and Bi2O3), initial pH of the solution (3-10), and type of complexing agents (Na2S2O3 and Na2SO3) and their concentrations (1-4 mM each) on Au recovery were explored. Among all employed semiconductors, TiO2 exhibited the highest photocatalytic activity to recover Au from the simulated spent plating bath solution both in the absence and presence of complexing agents, in which Au was completely recovered within 15 min at a pH of 6.5. The presence of complexing agents remarkably affected the size of deposited Au on the TiO2 surface, the localized surface plasmon effect (LSPR) behavior, and the valence band (VB) edge position of the obtained Au/TiO2, without a significant change in the textural properties or the band gap energy. The photocatalytic activity of the obtained Au/TiO2 tested via two photocatalytic processes depended on the common reduction mechanism rather than the textural or optical properties. As a result, the Au/TiO2 NPs obtained from the proposed recovery process are recommended for use as a photocatalyst for the reactions occurring at the conduction band rather than at the valence band. Notably, they exhibited good stability after the fifth photocatalytic cycle for Au recovery from the actual cyanide plating bath solution.

Introduction

Gold (Au) plating is an essential process used to prepare the surface of some electronic parts in the printed circuit boards (PCBs) owing to its high electrical conductivity, high reliability, excellent planarity, good solderability, and high corrosion resistance.[1,2] A high growth rate of the electronics industry is predicted to be observed in our country during 2021–2023 due to a number of various technological developments, such as the rollout of the 5G technology, the increase of smart car markets, and consumer demand for portable computers (PCs), notebooks, tablets, and smartphones, resulting from work from home (WFH) and distance learning of students during the COVID-19 situation.[3] Typically, Au plating can be carried out via both cyanide and non-cyanide Au salts by electroplating or electroless plating. The electrolytic Au cyanide plating can be operated in a wide pH range from acidic to alkaline pH, while non-cyanide plating is effective at neutral or alkaline pH.[1] To achieve a stable plating system, various kinds of complexing agents and additives are employed to obtain a good quality of plated specimens.[1,4] Thus, the spent plating bath solution generated from the process contains various kinds of chemicals and metal ions such as chromium (Cr), copper (Cu), iron (Fe), or nickel (Ni) ions, as well as the residual Au complex. The presence of precious metals such as Au in the effluent is an economic loss. Thus, various processes have been developed to recover Au from the discharged solution, such as precipitation, electrolysis, adsorptive reduction, ion exchange, solvent extraction, and bioproduction and biorecovery.[5,6] In addition to the mentioned processes, another promising process that can be used to recover Au from the plating bath solution is the photocatalytic process because it possesses the viability to recover Au from aqueous solutions of both chloride species[7] and cyanide species.[8] In addition, it is environmentally friendly, involves low operating costs, and can operate at ambient conditions.[9,10]

The photocatalytic process is a chemical process that occurs under the joint action between light and a photocatalyst.[11,12] The transition-metal oxides or semiconductors such as TiO2, ZnO, WO3, etc., are generally used as a photocatalyst because they have no energy levels to promote the recombination of the photogenerated electron (e–) and hole (h+) in their structure. The process efficiency depends upon various parameters, including the redox potential of dissolved metals relative to the utilized semiconductor, the presence of electron and/or hole scavengers, irradiated light intensity, solution pH, and type and loading of utilized semiconductors.[13] Recently, it was reported that TiO2/SiO2 exhibited a higher photocatalytic removal of Au(I) from gold–cyanide complexes than bare TiO2. The presence of hole scavengers such as methanol can promote the deposition of metallic Au on the surface of the utilized photocatalyst.[8] Using an amphiphilic photocatalyst based on a polyoxometallate (W10O324–)/surfactant (dimethyldioctadecyl ammonium) hybrid, the photocatalytic recovery of Au can be carried out under the pH value of 3–6 and depended on the catalyst loading and concentration of sacrificial agents.[14] The photoreduction of Au(III) via TiO2 increased with the increase of reaction times and photocatalyst loadings.[15] However, excessive loading of TiO2 increased the solution turbidity, which decreased the exposure of TiO2 with the light irradiation. The photoreduction decreased with the increase of the initial concentration of Au(III). The presence of Ag(I) and Cu(II) decreased the photoreduction of Au(III) due to the competition in adsorption and reduction, while the presence of Fe(III) can enhance the photoreduction due to the reduction and quick oxidation of Fe(II). The tannin (TA)-TiO2 heterostructure induced a high selectivity of Au(III) adsorption, which can prolong the lifetimes of excited electrons and then allow the adsorption/reduction of Au(III) from the solution.[16]

As mentioned above, Au plating can be carried out via both cyanide and non-cyanide baths. The traditional cyanide bath has been successful for a long time in comparison with the non-cyanide bath because it exhibits more outstanding strengths, including high bath stability and superior performance of obtaining gold films with excellent properties.[1,17] However, the development of cyanide baths is currently restricted due to their toxicity and material compatibility problems.[1,18] Thus, a safe and environmentally friendly process or a non-cyanide bath is currently gaining attention. With the increase of the industrial demand for electroless plating via non-cyanide baths, a large quantity of wastewater is produced. Based on our knowledge, very limited work related to Au recovery from a non-cyanide bath effluent by the photocatalytic process is published. For example, Chen et al.[19] attempted to recover Au(I) from a thiosulfate solution using the defect-rich MoS2 (DR-MoS2) nanoflowers. It was found that the efficiency of the utilized photocatalyst for the Au(I) reduction can be improved with the aid of ultrasonic treatment. The selective deposition of Au nanoparticles (NPs) on the surface of MoS2 was promoted by the edge sites and S defects of DR-MoS2 nanoflowers.

In this work, to aid the generation of Au-containing effluent from non-cyanide plating baths, the photocatalytic recovery of Au from the simulated spent plating bath solution was explored via various semiconductors. The effect of initial solution pH, irradiation time, and types and concentrations of complexing agents was explored. Finally, the photocatalytic activity of the obtained Au NP-decorated semiconductor that exhibited the best photocatalytic recovery of Au from the non-cyanide plating bath was tested via two photocatalytic processes, including the methylene blue (MB) degradation and the Au recovery from actual spent cyanide plating bath solution.

Results and Discussion

Morphology of Utilized Commercial Semiconductors

The crystal structure of all utilized semiconductors was first evaluated by XRD analysis. As demonstrated in Figure a, the XRD pattern of the parent TiO2 displayed diffraction peaks at 2θ values of 25.3, 37.8, 48.0, 53.90, 55.10, and 62.70°, corresponding to the crystal planes of (101), (004), (200), (105), (221), and (204), respectively, (JCPDS file no. 21-1272) for the anatase phase and displayed 2θ values of 27.41, 36.06, and 41.21°, corresponding to the crystal planes of (110), (101), and (111), respectively, (JCPDS file no. 21-1276) for the rutile phase. For WO3, its XRD pattern showed the diffraction peaks of the monoclinic phase at 2θ values of 23.07, 23.56, 24.31, 26.54, 28.87, 33.21, 34.12, 35.37, 41.84, 47.21, 48.32, 49.88, and 55.90°, corresponding to the crystal planes of (002), (020), (200), (120), (1̅12), (022), (202), (1̅22), (222), (004), (040), (140), and (420), respectively, (JCPDS file no. 43-1035). The CeO2 revealed XRD peaks at 2θ values of 28.54, 33.07, 47.47, 56.31, and 59.08°, corresponding to the crystal planes of (111), (200), (220), (311), and (222), respectively, (JCPDS file no. 340394). For Nb2O5, its XRD pattern exhibited the diffraction peak of both orthorhombic and monoclinic phases. The characteristic peaks of the orthorhombic phase appeared at 2θ values of 22.55, 28.30, 28.91, 36.49, 46.09, and 47.47°, corresponding to the crystal planes of (001), (180), (200), (181), (002), and (110), respectively, (JCPDS file no. 71-0336) and the monoclinic structure exhibited a peak at a 2θ of 23.72°, indicating the crystal plane of (4̅05) (JCPDS file no. 37-1468). The XRD pattern of Bi2O3 showed the main characteristic peaks of the α-monoclinic phase at 2θ values of 24.51, 25.71, 26.88, 27.35, 27.96, 33.21, 35.01, 37.58, 46.28, and 52.35°, corresponding to the crystal planes of (1̅02), (002), (111), (120), (012), (121), (022), (112), (3̅11), and (321), respectively, (JCPDS file no. 71-2274). The crystallite sizes of all employed semiconductors estimated using Scherrer’s equation are listed in Table and can be ranked in the order Bi2O3 > WO3 > Nb2O5 > CeO2 > TiO2.

Figure 1

Representative XRD patterns of (a) commercial semiconductors and (b) Au/TiO2 obtained from the photocatalytic recovery of Au from the simulated spent plating bath solution in the absence and presence of complexing agents at a semiconductor loading of 4 g/L and a light intensity of 4.1 mW/cm2.

Table 1

Properties of the Commercial Semiconductors and Obtained Au/TiO2

			textural property
photocatalysts	crystallite size of the semiconductora (nm)	anatase content (%)	BET area (m2/g)	average pore size (nm)	total pore volume (cm3/g)	band gap energy (eV)	decorated Aub (wt %)	particle size of Auc (nm)
TiO2	21.26	85.84	52.38	5.43	0.0629	3.35
WO3	77.01		0.595	4.60	0.0005	2.65
CeO2	54.66		5.432	5.46	0.0065	3.00
Nb2O5	62.33		2.487	4.96	0.0021	3.14
Bi2O3	116.82		0.141	1.93	0.0004	2.75
Au/TiO2-NC	22.94	84.89	55.42	6.83	0.0875	3.20	1.37 ± 0.02	10.3 ± 3.07
Au/TiO2-C	21.43	84.28	56.05	7.02	0.0896	3.19	1.02 ± 0.03	4.23 ± 1.84

Estimated from XRD analysis using Scherrer’s equation at crystal planes of (101), (002), (111), (001), and (111) for TiO2, WO3, CeO2, Nb2O5, and Bi2O3, respectively.

Estimated from SEM-EDX analysis.

Estimated from HRTEM analysis.

In the heterogeneous photocatalytic process, the adsorption of active species on the surface of semiconductors must occur first, followed by the photocatalytic reaction. The presence of a high specific surface area can induce effective charge adsorption as well as the photocatalytic activity of the utilized photocatalysts.[20] The N2 adsorption/desorption isotherms of all employed commercial semiconductors are shown in Figure a. All explored semiconductors exhibited the type IV isotherm according to the IUPAC classification with H4-shaped hysteresis loops, indicating the presence of slitlike pores.[21] The quantitative values of the BET surface area, average pore size, and total pore volume are summarized in Table . The TiO2 exhibited the highest BET surface area of 52.38 m2/g and a total pore volume of 0.0629 cm3/g, followed by CeO2 (5.432 m2/g), Nb2O5 (2.487 m2/g), WO3 (0.595 m2/g), and Bi2O3 (0.141 m2/g), respectively.

Figure 2

Representative N2 adsorption/desorption isotherms of (a) commercial semiconductors and (b) Au/TiO2 obtained from the photocatalytic recovery of Au from the simulated spent plating bath solution in the absence and presence of complexing agents at a semiconductor loading of 4 g/L and a light intensity of 4.1 mW/cm2.

The optical absorption spectra of all utilized commercial semiconductors were also examined in the wavelength of 300–800 nm at room temperature. As displayed in Figure a, both TiO2 and Nb2O5 displayed high absorbance at wavelengths lower than 380 nm and exhibited a very low intensity spectrum at wavelengths higher than 380 nm, indicating their visible light inertness. At an identical wavelength, CeO2 showed a higher absorbance spectrum than TiO2 and Nb2O5; however, it was still active in the UV region. Either WO3 or Bi2O3 clearly exhibited a larger red shift in the absorbance spectra than other utilized semiconductors up to a wavelength of 480 nm, suggesting their ability to absorb both UV light and the short-wavelength visible light. The optical band gap energy (Eg) of all employed semiconductors was quantitatively determined via the Tauc plot according to eq .[22]where α is the optical absorption coefficient, hν is the photon energy, and A is the proportionality constant. In this equation, n is related to the type of optical transition process in semiconductors (n = 1/2 and 2 for the direct and indirect band gap semiconductors, respectively).

Figure 3

Representative UV–vis spectra of (a) commercial semiconductors and (b) Au/TiO2 obtained from the photocatalytic recovery of Au from the simulated spent plating bath solution in the absence and presence of complexing agents at a semiconductor loading of 4 g/L and a light intensity of 4.1 mW/cm2 and their Tauc plots (inset figure).

The band gap values of all utilized semiconductors were obtained from the x-intercept of the plot of (αhν)2 against the photon energy (inset of Figure a) and are listed in Table . The values of the band gap energy of all utilized semiconductors can be ranked in the order TiO2 > Nb2O5 > CeO2 > Bi2O3 > WO3, indicating the reverse trend of the visible light absorption ability.

Photocatalytic Recovery of Au from the Simulated Spent Non-Cyanide Plating Bath Solution

Effect of Semiconductor Types

The photocatalytic recovery of Au from the simulated spent plating bath solution at a pH of 3.2 by commercial semiconductors at a loading of 4 g/L and a light intensity of 4.1 mW/cm2 is shown in Figure . Under the dark condition, approximately 72 and 33% of Au ions were adsorbed in the presence of TiO2 and CeO2, while only 2–4% of Au ions were adsorbed in the presence of WO3, Nb2O5, and Bi2O3. This indicated the highest adsorption capacity of Au ions via TiO2. Typically, the adsorption of some active substances on the surface of the adsorbent occurs via electrostatic interactions, which usually depend on the available surface area and surface charges of adsorbents.[23] A high adsorption of Au ions in the presence of TiO2, followed by CeO2, might be due to their high BET surface area and total pore volume available to adsorb some AuCl4– species compared to other utilized semiconductors. In addition, the value of surface charges or point of zero charges (PZCs) of semiconductors is dependent on the solution pH. That is, when the solution pH is higher than the PZC value, the adsorbent usually exhibits negative surface charges and strongly prefers to adsorb cationic molecules. However, when the solution pH is lower than the PZC value, the adsorbent demonstrates positive surface charges and favors the adsorption of anionic molecules.[24] In this case, the experimental PZC values of TiO2, WO3, CeO2, Nb2O5, and Bi2O3 were ∼6.6, ∼5.6, ∼6.4, ∼6.2, and ∼10.4, respectively. Thus, all utilized semiconductors displayed positive surface charges, which are able to adsorb the anionic charge of AuCl4– in an acidic solution.[25]

Figure 4

Photocatalytic recovery of Au from the simulated spent plating bath solution at a pH of 3.2 by commercial semiconductors at a loading of 4 g/L and a light intensity of 4.1 mW/cm2.

Under the light irradiation, the Au ions were completely recovered within 30 min in the presence of either TiO2 or WO3 and within 120, 480, and 480 min in the presence of CeO2, Nb2O5, and Bi2O3, respectively. TiO2 displayed the highest rate of Au recovery even though it has the highest band gap energy (∼3.35 eV), probably due to its highest adsorption capacity to the Au ions, which readily proceed to the reaction with the photogenerated e– under the irradiated light. However, this is not the case with WO3. Although it exhibited very low Au ion adsorption capacity, it displayed a very fast rate of Au recovery, probably attributed to its short band gap energy, which can harvest a high quantity of incident light to proceed to the photocatalytic reaction. This suggested that both textural and optical properties of semiconductors played an important role in the photocatalytic recovery of Au from the simulated spent solution.

Effect of Solution pH

Typically, the Au ions can exist in different forms in a chloride solution depending on the solution pH. That is, at a pH below 6.2, the Au ions are dominantly found in the form of tetrachloroaurate (AuCl4–), while both AuCl4– and hydroxylated gold chloride species [AuCl3(OH)−] are mainly found at a pH of 6.2–8.4. At a higher pH value (>8.4), the large hydroxylated gold chloride species are still dominant, starting from AuCl2(OH)2– to the others.[25]Figure shows the effect of the initial pH on the photocatalytic recovery of Au from the simulated spent plating bath solution using the TiO2 and WO3 at a loading of 4 g/L and a light intensity of 4.1 mW/cm2. In the absence of light, approximately 72 and 84% of Au ions were adsorbed by TiO2 at pH values of 3.2 and 6.5, respectively, but around 27% of Au ions were adsorbed at a pH of 10.0. However, approximately 3–7% of Au ions were adsorbed in the presence of WO3 for all pH values. This is attributed to the synergetic effect of the PZC value and textural properties of both semiconductors. That is, the TiO2 exhibited a high adsorption capacity under acidic conditions because it theoretically exhibits the positive surface charges as TiOH2+ species,[26] which can interact well with the negative charge of hydroxylated gold chloride species. However, in a basic solution, TiO2 exhibited negative surface charges (as TiO–),[26,27] which do not favor electrostatic interactions with the large hydroxylated gold chloride species such as AuCl2(OH)2–. Therefore, the discrimination of Au ions in the basic environment is possibly caused by the preferential adsorption of Au at the pore surfaces of TiO2 via the intermolecular cohesion forces. Low adsorption of Au ions in the presence of WO3 might be due to its low surface area. In the presence of irradiated light, a fast reduction of Au ions was observed in the presence of TiO2 at all pH values, likely due to the immediately photocatalytic reduction of adsorbed Au ions. In WO3, although almost similar quantities of Au ions were adsorbed in a dark environment for all pH values, a fast decrease of Au ions was observed via the solution with a pH of 6.5. This might be due to an appropriate aggregation of WO3 in the solution that has a pH value close to the PZC value, which can induce an effective transfer of e– and h+ to the adjacent particles under a thorough stirring condition, resulting in a prolonged e––h+ lifetime as well as an enhanced photocatalytic activity.[28] The sluggish reduction rate of Au ions in a strong basic solution (pH of 10.0) might be due to the steric hindrance of the large hydroxylated gold chloride molecules that competitively combined with the WO3 to proceed with the photocatalytic reaction.

Figure 5

Effect of initial pH on the photocatalytic recovery of Au from the simulated spent plating bath solution via (a) TiO2 and (b) WO3 at a loading of 4 g/L and a light intensity of 4.1 mW/cm2.

Effect of Complexing Agents

To produce smooth and ductile gold deposits with low stress and hardness as well as a good dimensional tolerance from a typical non-cyanide gold plating bath, two types of complexing agents, including sulfite (SO32–) and thiosulfate (S2O32–), are usually added to promote the bath stability at a near-neutral pH.[29] Thus, the spent solution discharged from the plating unit usually contains a trace quantity of a residual sulfite–thiosulfate mixed agent. To investigate the effect of complexing agents on the photocatalytic recovery of Au from the simulated spent non-cyanide plating bath, two types of complexing agents were employed, including Na2SO3 and Na2S2O3. Figure shows the photocatalytic recovery of Au from the simulated spent plating bath solution at pH of 6.5 using TiO2 and WO3 at a loading of 4 g/L and a light intensity of 4.1 mW/cm2 in the presence of complexing agents. It clearly demonstrated that the presence of both complexing agents affected the overall efficiency of the photocatalytic process. Under the dark condition, the presence of both complexing agents decreased the quantity of adsorbed Au ions on the surface of both semiconductors compared with that in the absence of complexing agents. A slightly high quantity of Au ions was adsorbed via TiO2 in the presence of Na2SO3 compared with Na2S2O3, while they did not adsorb via WO3 in the presence of both complexing agents. This might be attributed to the effect of the available surface area of the employed semiconductors and the different forms of Au complex species in the presence of complexing agents. That is, the dominant Au species in the presence of Na2SO3 and Na2S2O3 are Au(SO3)23– and Au(S2O3)23–, respectively.[29] A low adsorption of Au(S2O3)23– on the surface of TiO2 compared with Au(SO3)23– was probably due to its large molecular size, which can hamper the adsorption by steric hindrance. In addition, the byproduct S2O32– that can be generated during the adsorption of Au(S2O3)23– according to reaction (1) can obstruct further adsorption of Au(S2O3)23– on the adsorbent surface. A small quantity of both Au complex species was adsorbed on the surface of WO3, presumably due to a very low surface area of WO3.Under irradiated light, the presence of 2 mM Na2SO3 did not hinder the photocatalytic recovery of Au using TiO2, and all Au ions were completely recovered within 5 min. However, it remarkably decreased the photocatalytic recovery of Au using WO3, in which approximately 80% of Au ions were recovered within 90 min. The presence of 2 mM Na2S2O3 slightly slowed down the photocatalytic recovery of Au using TiO2, in which all Au ions were completely reduced within 10 min. However, it completely interrupted the photocatalytic recovery of Au using WO3. This might be due to the synergetic effect of the low surface area of WO3 and the molecular steric hindrance of Au(S2O3)23–.

Figure 6

Effect of complexing agents on the photocatalytic recovery of Au from the simulated spent plating bath solution at pH of 6.5 via (a) TiO2 and (b) WO3 at a loading of 4 g/L and a light intensity of 4.1 mW/cm2.

In the presence of both sulfite and thiosulfate at a pH value of 6.0–7.5, the dominant species of Au ions are reported in the form of sulfite and thiosulfate Au complexes as (Au2S2O3)(SO3)2.[1−5][1−5] As also demonstrated in Figure , neither adsorption nor photocatalytic reduction of Au was observed via WO3 in the presence of mixed complexing agents for all utilized concentrations. Different concentrations of complexing agents exhibited different behaviors of Au ion adsorption in the TiO2 system. That is, the adsorption of Au ions on the surface of TiO2 was extremely low in the presence of low concentrations of mixed complexing agents (1 and 2 mM of each agent). Slightly higher adsorption can be achieved in the presence of high concentrations of mixed complexing agents (3 and 4 mM of each agent). This is probably because a high concentration of (Au2S2O3)(SO3)25– in the bulk liquid can initiate a high concentration gradient between the bulk phase and the surface of TiO2, resulting in an increased mass transfer as well as an increased adsorption capacity.[30,31] In the presence of light, increasing the concentration of mixed complexing agents slightly decreased the photocatalytic efficiency for Au recovery compared with that in the absence of complexing agents, probably attributed to the shielding behavior of densely surrounded species in the presence of a high concentration of complexing agents. Based on the obtained results, among all employed commercial semiconductors, TiO2 exhibited the highest photocatalytic recovery of Au from the simulated spent non-cyanide plating bath both in the absence and presence of complexing agents at all explored concentrations.

Morphology and Photocatalytic Activity of the Obtained Au/TiO2

As reported elsewhere, the Au NP-decorated semiconductor exhibited an effective photocatalytic activity for various applications such as wastewater remediation,[32−36] hydrogen production from water splitting,[37,38] precious metal recovery,[19,39] glycerol conversion to value-added compounds,[40−42] etc. Thus, the morphology and photocatalytic activity of the Au/TiO2 obtained from the simulated spent plating bath solution in the absence and presence of mixed complexing agents, denoted Au/TiO2-NC and Au/TiO2-C, respectively, were then explored.

As qualitatively shown in Figure b, both Au/TiO2-NC and Au/TiO2-C exhibited the main characteristic peaks of the TiO2-based material with comparable anatase content (Table ), calculated according to the Spurr and Myers equations (eq ).[43] No characteristic peaks of decorated Au were observed, probably because of their presence in good dispersion of small quantities or even in amorphous form. Thus, the SEM-EDX analysis was carried out to identify the presence of decorated Au on the surface of TiO2. As displayed in Figure a-1,b-1, a good dispersion of Au was clearly observed with the entire loading of 1.37 ± 0.02 and 1.02 ± 0.03 wt % in Au/TiO2-NC and Au/TiO2-C, respectively (Table ). On HRTEM analysis, the average particle size of Au in Au/TiO2-C was significantly smaller than that in the Au/TiO2-NC (Figure a-2,b-2). This is probably due to the effect of different forms of Au complex in the absence and presence of complexing agents during the photocatalytic process. That is, in the presence of sulfite–thiosulfate complexing agents, a large Au complex form such as (Au2S2O3)(SO3)25– generally forms.[1] This large Au complex might hinder the competitive adsorption and photocatalytic reduction of each other, higher than those in the absence of complexing agents, which in turn reduced the agglomeration of deposited Au NPs on the TiO2 surface. As regards the textural property, both types of Au/TiO2 still displayed H4-shaped hysteresis loops of the type IV isotherm with comparable textural properties (Figure b and Table ), which were slightly better than those of pristine TiO2. This is due to the good dispersion of Au NPs on the TiO2 surface.[44]where xA is the anatase weight fraction, IA is the relative reflection intensities of anatase, and IR is the relative reflection intensities of rutile.

Figure 7

Representative SEM (1) and HRTEM (2) images of Au/TiO2 obtained from the photocatalytic recovery of Au ions from the simulated spent plating bath solution in the (a) absence and (b) presence of complexing agents.

The absorption spectra of the Au/TiO2 obtained from the photocatalytic recovery of Au from the simulated spent plating bath solution in the absence and presence of complexing agents were also explored in the range of 300–800 nm (Figure b). It can be observed that the Au/TiO2 obtained from both conditions exhibited a red shift in absorption toward the wavelength longer than 400 nm, indicating the ability to absorb more incident visible light. The Au/TiO2-NC revealed an intense broad band centered at a wavelength of 550 nm due to the oscillation and polarization of the excited electrons in the decorated Au structure called LSPR, which can effectively promote visible light absorption.[45] The Au/TiO2-C exhibited unclear LSPR behavior in comparison with the Au/TiO2-NC, probably due to the presence of too small particle sizes of Au NPs in its structure,[46] causing a low intensity of light absorption and scattering cross sections as well as the local electromagnetic field enhancement.[47] By applying Tauc’s equation, the band gap energies of both Au/TiO2 were comparable at 3.19–3.20 eV, which were lower than that of the TiO2-based material.

The chemical state and bonding of Ti, O, and decorated Au were then examined by XPS analysis. The survey spectra clearly revealed the spectrum of O 1s, Ti 2p, and C 1s peaks of the O, Ti, and C (from carbon tape) (Figure ). Using high-resolution XPS, the spectra of the pristine TiO2 showed two symmetric peaks at binding energies of 464.1 and 458.4 eV (Figure a-1), corresponding to the Ti 2p1/2 and Ti 2p3/2 components of Ti4+.[48−50] For the Au/TiO2, the characteristic peaks of both Ti 2p components still appeared at the same binding energy compared with TiO2. No satellite peaks at the low binding energy of both doublets were observed in both Au/TiO2 (Figure b-1,c-1), indicating the absence of the Ti3+ state or the defective structure. The XPS spectra of asymmetric O 1s contained two symmetric peaks of the O2– and OH– of the crystallite network at binding energies of ∼530.2 and ∼531.7 eV, respectively[51−53] (Figure a-2,b-2,c-2). The XPS spectra of Au revealed two principal peaks of Au 4f5/2 and Au 4f7/2 at binding energies of 86.7 and 83.1 eV with a spin–orbit splitting of 3.7 eV (Figure b-3,c-3), which is in good agreement with the values of the monometallic Au.[54] The presence of shoulder peaks with a higher binding energy for each Au 4f component was not observed, indicating the presence of monometallic Au on the TiO2 surface.[41,54] At low binding energy, the XPS spectra of TiO2, Au/TiO2-NC, and Au/TiO2-C, respectively, displayed maximum binding energies of around 2.74, 1.70, and 1.40 eV (Figure a), indicating a negative shift of the VB edge of TiO2 in the presence of decorated Au NPs. According to the obtained VB edges of all samples, the minimum conduction band (CB) of the respective samples would occur at −0.61, −1.50, and −1.79 eV. Therefore, the band positions of all samples are roughly sketched in Figure b.

Figure 8

Representative survey XPS spectra of the TiO2 and Au/TiO2 obtained from the photocatalytic recovery of Au from the simulated spent plating bath solution in the absence and presence of complexing agents.

Figure 9

(a) Representative HR-XPS spectra of the (1) Ti 2p, (2) O 1s, and (3) Au 4f of (a) TiO2, (b) Au/TiO2-NC, and (c) Au/TiO2-C.

Figure 10

(a) Representative XPS spectra of (1) TiO2, (2) Au/TiO2-NC, and (3) Au/TiO2-C photocatalysts at low binding energy and (b) scheme of their band position.

Figure illustrates the photocatalytic activity of the obtained Au/TiO2 samples through two photocatalytic processes under UV–vis light. It can be seen that both Au/TiO2-NC and Au/TiO2-C exhibited comparable photocatalytic activity for MB degradation under UV–vis light. Approximately 80% of MB was degraded within 180 min, which was lower than that of TiO2, in which around 88% of MB was degraded at the same reaction time (Figure a). According to the nature of the photocatalytic process, when the semiconductor or photocatalyst (S) absorbs the irradiated light that has the photon energy higher than its band gap energy, the electron (e–) will be excited from the VB to the CB, leaving the hole (h+) at the VB, as shown in reaction .[55,56] The photogenerated h+ can react with the adsorbed water (H2O) to form hydroxyl radicals (HO•) and proton (H+) (reaction ), while the photoexcited e– at the CB can react with the dissolved oxygen (O2) to form the superoxide radicals (O2•–) (reaction ). Both HO• and O2•– generated are the reactive oxidizing species, which can attach and degrade the MB molecules according to reaction .[56]where S* is the S in the excited state.

Figure 11

Photocatalytic activity of the obtained Au/TiO2 via (a) MB degradation and (b) Au recovery from the actual spent cyanide plating bath solution.

The low photocatalytic activity of MB degradation in the presence of Au/TiO2 in comparison with the TiO2 might be attributed to their less-positive VB edge to produce HO• (Figure b). Thus, the degradation of MB in the presence of Au/TiO2 is mainly dominated by the h+ or O2•– and not HO• as usual.[57] The photocatalytic degradation of MB via all employed photocatalysts was then fitted with the Langmuir–Hinshelwood model, as expressed in eq .[33]where C0 is the initial concentration of the substance, C is the concentration of the substance at a particular time t, and k is the first-order reaction rate constant.

As tabulated in Table , the apparent rate constants for MB degradation in the presence of Au/TiO2-NC and Au/TiO2-C were found to be 0.0080 and 0.0085 min–1, which were lower than that of the pristine TiO2 for 1.38- and 1.30-fold, respectively. Compared with the literature, the Au/TiO2 prepared from the proposed method gave a rate constant comparable to that of Au/TiO2 prepared by the sol–gel method (k ∼ 0.0067 min–1)[58] but lower than that of the TiAu1 prepared by the incipient wetness method (k ∼ 0.0687 min–1)[57] or TiO2–1%Co–2.5%Fe prepared by the solvothermal method (k ∼ 0.0186 min–1).[59]

Table 2

Apparent Reaction Rate Constants for the Photocatalytic Activity of TiO2 and Au/TiO2

	MB degradation		Au recovery
photocatalyst	k (min–1)	R2	k (min–1)	R2
TiO2	0.0110	0.9274	0.0005	0.9516
Au/TiO2-NC	0.0080	0.9760	0.0234	0.9982
Au/TiO2-C	0.0085	0.9865	0.0035	0.9988

Regardless of the photocatalytic reduction of Au from the actual spent cyanide plating bath solution, approximately 3% of Au(CN)2– was reduced during the reaction time of 2 h using pristine TiO2 as the photocatalyst (Figure b). However, approximately 77 and 24% of Au(CN)2– were reduced using Au/TiO2-NC and Au/TiO2-C as photocatalysts, respectively. Typically, in the oxygenated system, the reduction reaction of Au(CN)2– to Au via the photogenerated e– occurs at the VB (reaction ) at a reduction potential of around −0.57 V/NHE.[60] Moreover, the dissolved O2 can competitively react with the photogenerated e– to form H2O2 at a slightly low reaction potential (−0.146 V/NHE) according to reaction .[61] The generated H2O2 can dissociate in the presence of the incident light or react with the photogenerated e– to form HO• radicals (reactions ). Meanwhile, the formed CN– ions are readily oxidized by the photogenerated h+ and/or HO• at VB to form CNO– according to reactions and R10.[8,61]According to the sketched band position (Figure b), both TiO2 and Au/TiO2 samples exhibited a more negative CB level than the reduction potential of Au(CN)2–, indicating their ability to reduce Au(CN)2– to Au from the spent cyanide plating bath solution. A lack of photocatalytic reduction of Au(CN)2– via TiO2 during the reaction time of 2 h was probably due to a competitive reaction of dissolved O2 with the photoexcited e–, resulting in a low quantity of remaining e– to react with the Au(CN)2–. As the reaction proceeded, a massive amount of dissolved O2 was gradually consumed, and the deposition of Au over the surface of TiO2 could be started. These Au clusters can act as an electron sink to trap the migrated e–, which can prolong the lifetime of e––h+ pairs as well as increase the photocatalytic efficiency.[62−64] The photocatalytic reduction of Au(CN)2– via TiO2 was previously observed at around 75% when the reaction time was extended to 8 h (data not shown). This is not for the case of Au/TiO2 samples in which a faster photocatalytic reduction of Au(CN)2– was observed. This might be due to a combined effect between the presence of Au NPs that acted as the electron sink during the reaction, which can improve the charge transfer rate and decrease the band gap energy of both Au/TiO2 samples compared with TiO2, effectively harvesting the incident light to drive the excitation of e– from the VB to the CB to perform the photocatalytic reduction. The Au/TiO2-C exhibited slightly lower activity to reduce Au(CN)2– than the Au/TiO2-NC, due to its lower LSPR behavior. Using the Langmuir–Hinshelwood model, the Au/TiO2-NC exhibited apparent rate constants for Au(CN)2– recovery of 0.0234 min–1, which was higher than those for TiO2 and Au/TiO2-C of around 46.8- and 7.0-fold, respectively (Table ).

Based on the obtained results, although the Au/TiO2-C exhibited a photocatalytic recovery of Au from the actual spent cyanide plating bath solution lower than that of Au/TiO2-NC, it displayed the photocatalytic activity higher than that of pristine TiO2 of around 25% at 2 h. The repetitive photocatalytic recovery of Au from the actual spent cyanide plating bath solution was then studied using the obtained Au/TiO2-C as the original photocatalyst. Briefly, after the first experiment, the solid NPs were separated from the plating bath solution by centrifugation at 11 000 rpm, washed thoroughly with deionized water, dried at 80 °C for 3 h, and then subjected to the second experiment, and this was repeated. As displayed in Figure a, the quantity of decorated Au on the TiO2-based material increased with increasing photocatalytic cycles. Moreover, the Au/TiO2 obtained at a particular cycle can enhance the fast dwindling of Au(CN)2– from the cyanide plating bath solution in the next cycle, probably due to a high LSPR behavior in the presence of high Au content in the Au/TiO2 NPs.[45] After the fifth cycle, the size of decorated Au NPs was distributed in the range of 6–19 nm with an average value of 10.39 ± 3.28 nm (Figure b). The deposited Au NPs can simply separate from the TiO2-based material using selective dissolution via a commercial chemical such as Aqua regia.[8]

Figure 12

(a) Repetitive photocatalytic recovery of Au from the actual spent cyanide plating bath solution by the obtained Au/TiO2-C at 120 min and (b) the particle size distribution after the fifth cycle.

According to the nature of the proposed application of the obtained Au/TiO2-C for the Au recovery from the actual cyanide plating bath solution, the properties of the obtained Au/TiO2 at each photocatalytic cycle changed accordingly due to the accumulation of decorated Au. Thus, the stability of the employed photocatalyst was examined from the phase change of the TiO2-based material. As shown in Figure , the XRD pattern of the Au/TiO2 obtained at the fifth cycle (Au/TiO2-C(5)) still maintained the main characteristic peaks of the TiO2-based material. The Au/TiO2-C(5) NPs were found to exhibit almost similar anatase contents (∼84.86%) compared to the Au/TiO2-C NPs (∼84.28%), suggesting their stability after the fifth use.

Figure 13

Representative XRD patterns of the Au/TiO2-C before and after the fifth use (Au/TiO2-C(5)) for Au recovery from the cyanide plating bath solution.

Conclusions

Among all employed semiconductors, the photocatalytic recovery of Au from the simulated non-cyanide plating bath solution was accomplished well using the commercial TiO2 as the photocatalyst at a loading of 4 g/L and pH of 6.5 because of its high surface area. The presence of either Na2S2O3 or Na2SO3 at the concentrations of 1–4 mM insignificantly affected the photocatalytic recovery of Au. The Au was completely recovered within 5 and 15 min in the absence and presence of mixed complexing agents for all investigated concentrations. However, the presence of complexing agents remarkably affected the morphology of obtained Au/TiO2 samples as well as the LSPR behavior. That is, the Au/TiO2-C exhibited LSPR behavior lower than the Au/TiO2-NC with a negative shift of the VB edge position. Regardless of the photocatalytic activity, both Au/TiO2 samples exhibited poor photocatalytic activity for MB degradation in comparison with TiO2 due to their less-positive VB position to produce the HO• radicals. However, both Au/TiO2 samples exhibited a higher photocatalytic activity to recover Au from the spent cyanide plating bath solution than TiO2. This is because the deposited Au clusters promote the red shift of light absorption of TiO2 as well as LSPR behavior. The Au/TiO2-C exhibited an anatase content comparable to that of the TiO2-based material after the fifth photocatalytic cycle of Au recovery from the actual cyanide plating bath solution.

Materials and Methods

Chemicals

All chemicals employed in this work were analytical grade, including chloroauric acid (99.9% HAuCl4·3H2O; Sigma-Aldrich), sodium hydroxide (NaOH; Sigma-Aldrich), sodium thiosulfate (99.5% Na2S2O3·5H2O; Kemaus), and sodium sulfite (98% Na2SO3; Kemaus). Five commercial semiconductors were utilized as photocatalysts, including titanium dioxide (TiO2; 99.5%, Sigma-Aldrich), tungsten(VI) oxide (WO3; 99.9%, Sigma-Aldrich), niobium(V) oxide (Nb2O5; 99.99%, Sigma-Aldrich), cerium(IV) oxide (CeO2; 99.995%, Sigma-Aldrich), and bismuth(III) oxide (Bi2O3; 99.999%, Sigma-Aldrich).

Photocatalytic Recovery of Au from the Simulated Spent Non-Cyanide Plating Bath Solution

The photocatalytic recovery of Au from the simulated spent non-cyanide plating bath solution was tested via several types of commercial semiconductors at an identical loading under UV–vis light irradiation. In each experiment, approximately 1.2 g of the respective semiconductors was dispersed in 300 mL of the simulated spent plating bath solution and filled in a 400 mL double-wall cylindrical glass reactor, which was then placed in the middle of the UV-protected box at the right position to receive the irradiated light. Afterward, the solution was agitated continuously with a magnetic stirrer at a constant rate of 400 rpm in the absence of light for 30 min to provide a good dispersion and equilibrium adsorption of Au ions on the surface of the semiconductor. After the equilibration period, the system was irradiated with the UV–vis light (100–600 nm) generated by a high-pressure mercury lamp (120 W; RUV 533 BC, Holland) at a light intensity of 4.1 mW/cm2. The temperature of the photoreactor was precisely controlled at around 32 °C using the water circulation system as the reactor jacket. As the time proceeded, approximately 5 mL of the solution was sampled and centrifuged at 11 000 rpm (5804R, Eppendorf) to separate the solid portion from the aqueous mixture, and then the remaining Au ions were analyzed by flame atomic absorption spectrometry (Flame-AAS, AAnalyst 200+ flas 400; PerkinElmer). The amount of recovered Au from the simulated spent plating bath solution could be obtained by subtracting the initial concentration of Au ions with the unrecovered one. The results reported here are the average value of three repetitive experiments, and the acceptable error in this work was not greater than 3%.

Characterizations

The light absorption capacity and the diffuse reflectance spectra over the wavelength range of 300–800 nm were measured by ultraviolet–visible–near-infrared spectrophotometry (UV–vis; Lambda 950, PerkinElmer). The crystallite structure of all semiconductors was analyzed by X-ray diffractometry (XRD; D2 Phaser, Bruker). The surface morphology and chemical composition of the fresh semiconductors and Au-decorated semiconductors were primarily observed via scanning electron microscopy (SEM; JSM-IT-500HR, JEOL) with energy-dispersive X-ray spectrometry (EDX; JED-2300, JEOL) and high-resolution transmission electron microscopy (HRTEM; JEM-3100F, JEOL) with an accelerating voltage of 300 kV. The surface area, average pore size, and total pore volume were examined using a Multipoint Surface Area Analyzer (Tristar II3020, Micromeritics) via the Brunauer–Emmett–Teller (BET) method. The bonding and chemical state of the elements were determined with an X-ray photoelectron spectrometer (XPS; Axis Supra, Kratos, U.K.) with a Delay Line detector (DLD) and a monochromatic Al Kα (hν = 1486.6 eV) source. Accurate binding energies (±0.1 eV) were established with respect to the position of the adventitious carbon C 1s peak at 284.8 eV.

Photocatalytic Activity of the Obtained Au NP-Decorated Semiconductors

The photocatalytic activity of the Au NP-decorated semiconductors was examined with two photocatalytic processes, including the MB degradation and the Au recovery from the actual spent cyanide plating bath solution. The experimental setup of the MB degradation was similar to that in Section . In each experiment, approximately 200 mL of MB solution (50 mg/L, pH 7.58) with a photocatalyst loading of 0.25 g/L was first stirred at 300 rpm in the dark condition for 30 min and then irradiated by UV–vis light (100–600 nm) at a light intensity of 4.1 mW/cm2. As the reaction proceeded, approximately 2.5 mL of MB solution was collected at a particular time and centrifuged at 11 000 rpm (5804R, Eppendorf) to separate the solid particles from the solution. Finally, the remaining MB concentration was quantitatively determined by measuring the absorbance at 664 nm using a UV–vis spectrophotometer (UV-1800, Shimadzu).

The Au recovery from the actual spent cyanide plating bath solution was determined using a similar experimental setup with the KAu(CN)2 as the Au precursor at a photocatalyst loading of 4 g/L. The spent solution had a pH of 8.8–9.0 and contained Au ions of around 6.55 ± 0.12 mg/L and a trace quantity of iron (Fe), nickel (Ni), zinc (Zn), and copper (Cu) at concentrations less than 0.14, 0.99, 0.40, and 0.18 mg/L, respectively. The remaining concentration of Au ions in the spent solution was measured along the reaction times by flame atomic absorption spectrometry (Flame-AAS, AAnalyst 200+ flas 400; PerkinElmer).