Preparation of Cu-Doped TiO 2 Photocatalyst with Thermal Plasma Torch for Low-Concentration Mercury Removal

A high-quality, Cu-doped TiO2 photocatalyst was prepared via a single-step process using an atmospheric-pressure plasma torch system. Degussa P-25 was thermally doped with Cu at Cu/(Cu + TiO2) ratios of 0–5 wt.%. The raw and resulting nanoparticles were characterized using TEM, XRD, UV-Vis, and XPS. TEM showed that the particle size of plasma-treated TiO2 was generally < 50 nm. 67–75% of the resulting particles, by number were between 10 and 20 nm. The remaining particles were < 10 nm (~10%) and between 20 and 30 nm (~10%). The XRD results showed that Cu doping decreased the anatase/rutile crystalline ratio compared to untreated P-25. Nevertheless, the greater the amount of Cu added, the greater the anatase/rutile ratio was for the Cu-doped TiO2. The UV-Vis results showed that the absorption wavelength for plasma-treated TiO2 extended to the visible light range, especially for TiO2 doped with 5 wt.% Cu. The XPS results revealed that the form of Ti was Ti and Ti, O was O, and Cu was Cu and Cu. The Hg breakthrough tests indicated that the Cu-doped TiO2 underwent appreciable Hg removal under visible-light irradiation. Doped Cu effectively suppressed Hg reemission from the TiO2 surface.


INTRODUCTION
Mercury (Hg) emissions are a major concern to human health and the environment due to their toxicity, bioaccumulative properties, and low concentrations that make them difficult to control (Bittrich et al., 2011a, b).Coal-fired power plants are known to be the largest source of Hg in most countries with high Hg emissions (Fang et al., 2012;Wu et al., 2012).Nevertheless, low-concentration Hg, within 1-10 ppb v in the coal-combustion flue gases, is difficult to control due to mass transfer limitations.Hg species including elemental (Hg 0 ), oxidized (Hg 2+ ), and particlebound Hg (Hg p ) are generated during coal combustion.Hg p can be easily removed by electrostatic precipitators or bag filters.Hg 2+ is soluble in water and readily captured by a wet scrubber.Hg 0 , however, is volatile and insoluble in water, so it is difficult to remove with conventional control techniques.Chemically modified activated carbon has been extensively studied for removing low-concentration Hg, mainly via adsorption.Cu treatment, such as CuCl 2 impregnation, has also been found to markedly enhance Hg 0 removal by activated carbons via both adsorption and catalytic oxidation (Lee et al., 2009a, b).The use of titanium dioxide (TiO 2 ) nanoparticles in adsorption and catalytic oxidation of Hg 0 has also been reported in recent studies (Wu et al., 1998;Pitoniak et al., 2003;Biswas and Wu, 2005;Pitoniak et al., 2005;Li andWu, 2006, 2007;Tsai et al., 2011;Hsi and Tsai, 2012a, b).Gaseous Hg 0 molecules are adsorbed and oxidized on the active surface of TiO 2 to form a sorbent-mercury complex (HgO-TiO 2(complex) ) (Suriyawong et al., 2009).Kim et al. (2010) impregnated anatase-type TiO 2 (MC90, Ishihara Corp.) with CuCl 2 to prepare catalyst supports for oxidizing gaseous Hg 0 during selective catalytic reduction of NO x .They showed that the catalyst decomposed to release Cl by high-temperature calcination, but could be reversibly restored to its original form by exposure to HCl (g) .
The authors reported an increase in Hg oxidation with an increase in CuCl 2 loading and HCl concentration.
Branching electron-hole pairs and exterior surface area are key variables that impact the effectiveness of TiO 2 nanoparticles for Hg control.Choi et al. (1994) reported that metal ion dopants in the TiO 2 crystalline matrix influence photoreactivity, charge carrier recombination rates, and interfacial electron-transfer rates.Other studies doped TiO 2 with metals such as Fe (Oh et al., 2003), Al (Choi et al., 2007), and Cu (Chiang et al., 2002;Zhang et al., 2004;Slamet et al., 2005;Colón et al., 2006;Park et al., 2006;Araña et al., 2008;Xin et al., 2008;Yoong et al., 2009) to improve the reactivity and kinetics of the catalyst.These results support that only select transition metals, including Cu 2+ , can virtually inhibit electron-hole recombination to improve the catalyst (Butler and Davis, 1993;Araña et al., 2008).Preventing the recombination of charge carriers enhances photocatalyst activity.
Several works have shown that Cu-doped TiO 2 nanoparticles can be fabricated via liquid-phase syntheses (e.g., impregnation and sol-gel) (Chiang et al., 2002;Slamet et al., 2005;Colón et al., 2006;Araña et al., 2008;Xin et al., 2008;Yoong et al., 2009) or vapor-phase syntheses (e.g., sputtering and evaporation condensation) (Zhang et al., 2004;Park et al., 2006).The fabrication process affects the purity and surface properties of the resulting TiO 2 nanoparticles, which subsequently influence the photocatalytic properties.Generally, liquid synthesis methods require multiple steps to prepare high-quality nanoparticles.Evaporation condensation using thermal plasma is advantageous because it develops nanoparticles with clean surfaces while modifying the catalyst surface in a single step.
In our previous study, Al-doped TiO 2 nanoparticles were formed in a single step with a non-transferred DC thermal plasma system (Tsai et al., 2012).For the present study, it was anticipated that thermal plasma would provide sufficient energy to dope Cu into the crystal structure of TiO 2 , resulting in substitution of Cu for Ti instead of only physical deposition of Cu onto the TiO 2 surface.The resulting physical and chemical characteristics of Cu-doped TiO 2 were examined.The Hg 0 removal effectiveness of Cu-doped TiO 2 in the presence of O 2 , H 2 O, and ultraviolet (UV) or visible light (VL) is discussed.

Preparation of Cu-Doped TiO 2 Nanoparticles
Cu-doped TiO 2 nanoparticles were prepared by an atmospheric-pressure thermal plasma system using pure Cu powder (purity 99.5%) and commercial Degussa P-25 as precursors.The thermal plasma system (Fig. 1) mainly consists of a DC non-transferred plasma torch connected to a power supply (Model PHS-15C, Taiwan Plasma Corp., Taiwan), a stainless steel chamber (stainless steel 310 with i.d.= 30 cm and length = 100 cm), and a stainless steel powder feeder.The synthesized TiO 2 nanoparticles were collected by a stainless steel powder filter.A vacuum pump (GVD-050A, ULVAC) and a buffer tank were used for capturing any particles that bypassed the filter.The plasma system was operated at 30 A and 200 V. Ultrahighpurity (UHP) Ar and O 2 were mixed as the plasma gas at Ar:O 2 = 3:1 by volume, with a total flow rate of 60 L/min.The Cu and P-25 powder feedstocks were vertically injected into the plasma jet using UHP Ar as the carrier gas with a flow rate of 2 L/min.The Cu/(Cu + P-25) mass ratio was controlled at 0, 1, 3, and 5 wt.%.The total powder feeding rate was 0.2 g/min.Cu-doped TiO 2 with 0 wt.%Cu means that the P-25 particles passed through the plasma system without feeding Cu powder.

Characterization of Cu-Doped TiO 2 Nanoparticle
The morphology of synthesized Cu-doped TiO 2 nanoparticles was examined using a transmission electron microscope (TEM, Philips CM-200).The crystalline structure of the nanoparticles was identified with powder X-ray diffraction (XRD, Rigaku Rinet 200).The X-ray source was Cu κα radiation (λ = 1.5405Å).The accelerating voltage and applied current were 40 kV and 30 mA, respectively.The scan range (2θ) was 20° to 80° with a scanning rate of 0.02° s -1 .The mass fraction of anatase in synthesized Cudoped TiO 2 nanoparticles was calculated from XRD data using the following equation (Spurr and Myers, 1957): where f A is the mass fraction of crystalline anatase in the TiO 2 nanoparticles, I R is the intensity of the (110) rutile reflection, and I A is the intensity of the (101) anatase reflection.Diffuse reflectance spectra from 300 to 800 nm were determined with a UV-visible scanning spectrophotometer (Hitachi U-3010).The Ti2p, O1s, and Cu2p bonding of nanoparticles was examined using X-ray photoelectron spectroscopy (XPS, ULVAC-PHI 1600).All binding energies (BE) were referenced to the C1s peak at 285 eV.

Hg Capture Experiments
Cu-doped TiO 2 synthesized at the Cu/(Cu + P-25) mass ratio of 0 wt.% and 5 wt.% were evaluated for control of low-concentration gaseous Hg 0 .A certified Hg 0 permeation tube (VICI Metronics) heated at 70 ± 0.1°C generated Hg 0containing gas with a concentration of 10-15 µg Hg 0 Nm -3 .Hg 0 with a known concentration was mixed with N 2 , O 2 , and water vapor, which was generated by passing N 2 through a water bubbler.The gases were introduced into a temperature-controlled mixing chamber to ensure steady gas concentrations, then carried (1.5 L/min) to the reactor through heated lines/tubes to prevent water condensation.The photochemical reactor containing 30 mg Cu-doped TiO 2 was irradiated with UV or VL light.The UV lamp (F10T8BLB, Sankyo Denki) has wavelength between 325 and 400 nm with a sharp peak at 364.5 nm.The VL lamp (ESS-DE-500W, Philips) is a household halogen lamp with a broad peak between 450 and 800 nm.The spectra of the light sources were obtained using a spectrophotometer (Ocean Optics, USB2000).Two UV lamps were used ~5 cm from the photochemical reactor; the UV intensity was measured (Lutron YK35-UV light meter) to be 3.5 ± 0.5 mW cm -2 .Because the VL lamp is more intense than the UV lamp, it was located 40 cm from the photochemical reactor to maintain a similar light intensity of 3.8-4.5 mW cm -2 (Lutron LX-103 light meter).UV light emitted from the VL lamp was negligible, approximately 0.01-0.03mW cm -2 measured at the reactor.A thin polycarbonate plate in front of the VL lamp was used to filter out this UV and prevent interference.Hg 0 removal experiments were performed at 25°C under atmospheric pressure.The exhaust gas from the photochemical reactor flowed through a moisture trap (i.e., a nefion tube) to remove water vapor and minimize interference in Hg detection.The effluent gas then flowed through a gold amalgamation column held by a heating coil (Brooks Rand model AC-01) where Hg 0 is adsorbed.The concentrated Hg 0 on the gold was thermally desorbed and sent to a cold-vapor atomic fluorescence spectrophotometer (Model III, Brooks Rand Lab).To allow for sufficient Hg concentration, six minutes were needed per test run.Ten runs were performed for each set of conditions.Finally, the exhaust from the detector passed through a carbon trap before release into a fume hood.

Characteristics of Cu-Doped TiO 2 Nanoparticles
TEM images of nanoparticles developed at various Cu/ (Cu + P-25) mass ratios are shown in Figs.2(a)-(d).The feedstock Cu powder was non-spherical and between 0.5 and 1.5 µm (Fig. 2(e)).However, TEM results showed that the materials prepared at various Cu/(Cu + P-25) mass ratios were < 50 nm, indicating that the Cu powder was successfully vaporized and doped into TiO 2 .TEM micrographs also demonstrated that the formed nanoparticles were homogeneous, with no significant phase separations or coatings on the surface.The particle size-distributions for Cu-doped TiO 2 were estimated from TEM images (Fig. 3).The particle size of plasma-treated TiO 2 was generally < 50 nm: 67-75%, by number, were 10-20 nm, ~10% were < 10 nm, and ~10% were 20-30 nm.Interestingly, doping with Cu that decreased the particle size of TiO 2 as an increase in the number of particles < 10 nm was noted.The increase in the number of smaller particles (< 10 nm) may be attributed to the suppression of particle growth by the introduction of Cu atoms into the TiO 2 crystal structure.De Villeneuve et al. (2005) reported that impurities could influence the nucleation, growth, and structure of crystals.Kubota (2001) also addressed that crystal growth is markedly affected by the impurities presented in the system.Our observational results are also in agreement with our previous study (Tsai et al., 2012).On the contrary, the number of 10-20 nm particles increased slightly as the Cu/(Cu + P-25) mass ratio increased from 1 to 5 wt.%.These increases may be attributed to sintering caused by heating of nanoparticles when the nanoparticles travelled through the plasma reactor.
Fig. 4 shows XRD patterns for the Cu-doped TiO 2 nanoparticles synthesized at various Cu/(Cu + P-25) mass ratios.All peak intensities were normalized to the intensity of the anatase (101) peak.All samples showed anatase and rutile crystalline phases (Fig. 4).The f A value for samples synthesized at Cu/(Cu + P-25) mass ratios of 0, 1, 3, and 5 wt.% was 0.74, 0.59, 0.65, and 0.69, respectively.The f A value of the precursor, P-25, is 0.74, so Cu doping causes a decrease in the anatase fraction and/or an increase in the rutile fraction of the materials.Previous research has shown rutile content to increase with increasing reaction temperature (Karvinen, 2003).This is consistent with our results; the high temperatures associated with plasma treatment are believed to transform anatase into rutile.It is also worth noting that for Cu-doped TiO 2 prepared with increasing Cu/(Cu + P-25) mass ratios (from 1 to 5 wt.%), the f A value increased from 0.59 to 0.69.This observation is consistent with our previous study for Al-doped TiO 2 (Tsai et al., 2012).Two routes are proposed to explain the observed phenomena.First, amorphous or short-range copper oxide phases in TiO 2 that cannot be identified with XRD may cause a reduction in rutile content.Zhang et al. (2004) reported that < 65.97 wt.% Cu in TiO 2 had no significant cooper oxide peaks in the XRD powder diffraction patterns.Second, under the high-temperature plasma environment, the injected Cu powder vaporized and subsequently doped into the TiO 2 lattice.Therefore, the extent of structure defection increased with increasing Cu concentration.Park et al. (2006) reported a similar observation.
Fig. 5 shows the diffuse reflectance spectra of Cu-doped TiO 2 from 300 to 800 nm.As expected, the 0 wt.%Cudoped TiO 2 lacked significant absorption beyond the basic absorption sharp edge at 390 nm.This absorption edge corresponds to the band gap of TiO 2 at 3.18 eV.Compared to the undoped TiO 2 , the absorption sharp edge of Cu-doped TiO 2 had a 15 nm blue shift.Anpo et al. (1987) and Zhang et al. (2000) reported that the blue shift of the absorption edge was mainly due to quantum size effects (particle size approximately 1-10 nm).TEM (Fig. 3) showed that the number of particles < 10 nm increased after doping with Cu.It is reasonable to suspect, therefore, that quantum size  effects also cause the observed blue shift in the materials prepared for this study.Additionally, the 1-5 wt.% Cudoped TiO 2 photocatalyst absorbed visible light.Several studies showed that the band gaps of CuO and Cu 2 O are 1.7 eV and 2.1 eV, respectively (Mishima, 1996;Nie et al., 2002).Absorption of visible light, therefore, is attributed to the presence of nanosized CuO and Cu 2 O in the TiO 2 .As mentioned earlier, however, CuO and Cu 2 O were not detected with XRD (Fig. 4), again suggesting that Cu species are highly distributed throughout TiO 2 nanoparticles in the high-temperature plasma environment.Moreover, absorption in the UV region for the 1, 3, and 5 wt.%Cudoped TiO 2 occurred at approximately 375, 380, and 385 nm, respectively.These results suggest that the extent of the red shift and broadening is dependent on the amount of Cu in TiO 2 .Consequently, Cu doped into TiO 2 in the plasma environment appears to generate a new Fermi level, which may be caused by strain fields accommodating the lattice match between Cu and TiO 2 (110) (Diebold, 2003).The results presented here confirm that a portion of Cu ion doped into the TiO 2 lattice alters the energy gap; the Cu doping causes an extension of TiO 2 absorption into the visible region.XPS analyses were carried out to determine the chemical and electronic structure of formed nanoparticles and the valence states of select elements.Fig. 6 shows Ti2p, O1s, and Cu2p XPS spectra corresponding to TiO 2 prepared with various Cu ratios.The O1s peak of TiO 2 was located at 529.9 eV (Wagner et al., 1980).The Cu2p spectra with peaks at 932.4 and 933.6 eV confirmed the presence of Cu + and Cu 2+ in the 5 wt.%Cu-doped TiO 2 crystal lattice (Poulston et al., 1996;Yano et al., 2003;Faungnawakij et al., 2009).Ti 4+ was successfully deconvoluted from the Ti2p binding energy for 0-5 wt.% Cu-doped TiO 2 .Ti 3+ peaks can also be found in the spectrum for the 5 wt.%Cudoped TiO 2 .These peaks are indicative of TiO 2 and Ti 2 O 3 being present (Wang and Ro, 2007).The reduction of Ti 4+ into Ti 3+ may be attributed to the presence of Cu species in the formed TiO 2 .Additionally, the formal charge generated from the substitution of Cu 2+ by Cu + can also be compensated via the transformation of Ti 4+ into Ti 3+ .Oxygen vacancies in the Cu-doped TiO 2 structure may also cause the reduction of Ti 4+ into Ti 3+ .It is well known that thermal plasma provides a highly reactive chemical environment, including high radical density and temperature that can induce the scission of Ti-O bonds.Therefore, oxygen deficiencies are expected on the surface of TiO 2 nanoparticles after plasma treatment.

Hg Removal with Cu-Doped TiO 2 Nanoparticles
Fig. 7 shows the Hg breakthrough curves for the 0 and 5 wt.%Cu-doped TiO 2 nanoparticles with changing experimental parameters, including O 2 concentration, humidity, and the type of light sources.These variables were evaluated individually by increasing the O 2 concentration from 0 to 12%, introducing H 2 O (20% relative humidity), and selecting dark, UV, or VL light.The experimental results showed that the Hg 0 removal efficiency of 0 wt.%Cu-doped TiO 2 was greater than that of 5 wt.%Cu-doped TiO 2 for O 2 concentrations ≤ 3% (Runs 0-120 in Fig. 7).The Hg 0 removal efficiency of 5 wt.%Cu-doped TiO 2 nanoparticles, however, was greater than that of 0 wt.%Cu-doped TiO 2 for O 2 concentrations from 6% to 12% (Run 121-240 in Fig. 7).These results suggest that in oxygen-rich conditions, O 2 adsorbed on the Cu-doped TiO 2 surface can form O• radicals, which enhance catalytic oxidation of Hg 0 .Furthermore, oxygen vacancies in the 5 wt.%Cu-doped TiO 2 , as described earlier, may enhance conductivity and be capable of trapping electrons that enhance Hg 0 oxidation and increase Hg removal.These results support that catalytic oxidation of Hg 0 with subsequent adsorption of Hg 2+ is the dominant mechanism for Hg control by TiO 2 nanoparticles.
A potential downfall of TiO 2 photocatalysts is that their surface hydrophilicity can be enhanced due to light irradiation (Fujishima et al., 2000).This increase in surface hydrophilicity may lead to competitive adsorption of H 2 O and subsequently prevent sufficient adsorption of Hg (Runs 41-60 in Fig. 7).Moisture caused a decrease in the Hg capacity of TiO 2 photocatalysts, causing desorption of Hg species from the 0 wt.%Cu-doped TiO 2 surface, especially when exposed to UV light.These results are consistent with the literature (Li andWu, 2006, 2007;Tsai et al., 2011).Li and Wu (2006) reported that physically adsorbed Hg 0 is desorbed from the surface of a SiO 2 -TiO 2 composite when exposed to high concentrations of H 2 O, suggesting that Hg 0 is weakly adsorbed on the sorbent surface.Interestingly, unlike the 0 wt.%Cu-doped TiO 2 , minimal desorption of Hg was observed from the 5 wt.%Cu-doped TiO 2 material under 0% O 2 and humid conditions, with UV or VL light (Runs 41-60 in Fig. 7).This observation may be attributed to the binding effects provided by Cu 2+ in the 5 wt.%Cudoped TiO 2 , based on the aforementioned characterizations.
In copper, the outer level electron orbital is full (4s 1 3d 10 ).Therefore, we speculate that Cu 2+ and Hg may have stronger covalent bonding than the bonds formed between TiO 2 and Hg that rely on the partially filled outer electron orbital of Ti (3d 2 ).
Based on the results of this study and those studies available in the literature, the following mechanism for Hg capture with Cu-doped TiO 2 nanoparticles is proposed (Fujishima et al., 2000;Jin and Shiraishi, 2004;Slamet et al., 2005;Li andWu, 2006, 2007): In addition to its covalent bonding characteristics, Cu ions also function as electron traps in this mechanism (Eqs.( 13), ( 14), and ( 16)).Chiang et al. (2002) reported that because Cu 2+ has an unfilled 3d electron shell and Cu 2+ reduction is thermodynamically feasible, it could be assumed that electron trapping by Cu II O occurs on the TiO 2 surface.As a result, the rate of the electron-hole recombination reaction decreases and, therefore, more holes are available for the Hg reduction/oxidation reactions.Cu doping may thus decrease the reduction of HgO to Hg 0 and subsequently decrease Hg reemission from the TiO 2 surface.
It is imperative to note that this study used simple gas streams consisting of only Hg 0 in a moisture-O 2 -N 2 mix.If the prepared Cu-doped TiO 2 were used to control Hg 0 emissions from stationary sources such as utility boilers, they would be exposed to more complex gas streams that also contain fly ash, CO, and several acid gases.A typical untreated flue gas derived from the combustion of a United States' low sulfur eastern bituminous coal can contain: 5-7% H 2 O, 3-4% O 2 , 15-16% CO 2 , 1 ppb total Hg, 20 ppm CO, 10 ppm hydrocarbons, 100 ppm HCl, 800 ppm SO 2 , 10 ppm SO 3 , 500 ppm NO x , and balance N 2 .(Granite et al., 1999;Granite and Pennline, 2002;Granite et al., 2007).Additional studies on the impacts of these flue gas components on Hg removal with metal-doped TiO 2 are needed.

CONCLUSIONS
Cu-doped TiO 2 nanoparticles were successfully prepared via a single-step process using Cu powder and Degussa P-25 nanoparticles with a non-transferred plasma torch system.Cu-doped TiO 2 nanoparticles formed at Cu/(Cu + P-25) mass ratios from 0 to 5 wt.% were < 50 nm.The crystal structure of the formed nanoparticles consisted primarily of anatase and rutile, and the mass fraction of anatase decreased with increasing Cu content.Characteristic peaks for copper and copper oxides were not present in the XRD patterns for Cu-doped TiO 2 nanoparticles, indicating that the doped copper was well distributed and amorphous.UV absorption for the 1, 3, and 5 wt.%Cu-doped TiO 2 was observed at 375, 380, and 385 nm, respectively.Cu doping and oxygen vacancies in the TiO 2 crystal structure may also cause phase transformations from TiO 2 to Ti 2 O 3 .Hg breakthrough testing showed that Hg capture is enhanced with increasing O 2 concentrations, but the addition of moisture results in competitive H 2 O adsorption and decreases Hg removal.Doping TiO 2 with Cu, however, minimizes desorption of Hg under high moisture conditions.
Hg breakthrough ratio of Cu-doped TiO 2 in the continuous adsorption experiment under various test conditions.