Ternary mixed-oxide synergy effects of nano TiO2-FexOy-MOk (M = Mn, Ce, Co) on α-pinene catalytic oxidation process assisted by nonthermal plasma

Biomass fuel from plant waste is an important substitute for fossil fuels. The odorous volatile organic compounds (VOCs) is released during the drying process of biomass, which causes harm to the ecological environment and human health. The α-pinene is a typical odorous VOC generating during plant waste drying. Ternary nanocatalyst composed of TiO2, Fe x O y and MO k (M = Mn, Co, Ce) was manufactured by ultrasonic assisted hydrothermal process. The α-pinene catalytic oxidation property of ternary nanocatalysts were investigate in a dielectric barrier discharge reactor assisted by nonthermal plasma. The pore structure parameters of ternary nanocatalysts were observed qualitatively and analyzed quantitatively by transmission electron microscopy (TEM) and N2 adsorption test, respectively. The phase composition and active element valence of these three kinds nanocatalysts were analyzed and compared by X-ray diffraction (XRD) and X-ray Photoelectron spectroscopy (XPS). The test data showed TiO2-Fe x O y -CoO k ternary nanocatalyst had more complete microporous and mesoporous pore structure, better element dispersion and stronger redox performance. Meanwhile, TiO2-Fe x O y -CoO k sample had the best performance with α-pinene catalytic conversion achieving 83.3% and CO2 selectivity higher than 90% at specific energy density (SED) of 620 J·l−1, which was almost half of the energy consumption of single non-thermal plasma catalytic reaction during the nonthermal plasma-catalyst (NTP-C) synergistic catalytic activity experiments. At the same time, the ternary nanocatalysts could obviously reduce the generation concentration of O3 and NO x in the process of α-pinene catalytic oxidation. The synergy effect between TiO2, Fe x O y and CoO k was better than that of TiO2, Fe x O y and CeO k or MnO k . It could be expected as an effective method to improve the redox performance of ternary nanocatalysts by optimizing the microstructure and elemental composition, which would also be a promising way to enhance odorous VOCs catalytic oxidation efficiency and reduce the energy consumption in the NTP-C synergistic catalytic system.


Introduction
Terpenes are kinds of natural hydrocarbon widely existing in vegetation, which can be released by many kinds of plants, vegetables and fruits. According to statistics, the total number of known terpenoids is more than 22000 [1]. Terpenes are typical Volatile organic compounds (VOCs) generating during biomass fuel drying process, and are considered as one of the main sources of odorous VOCs [2]. As important components of air pollutants, VOCs have become the focus of the atmospheric environment research all over the world [3][4][5]. Many VOCs In this study, in consideration of the economic feasibility, we prepared ternary mixed-oxide nanocatalyst with TiO 2 , Fe x O y and MO k (M=Mn, Co, Ce) by ultrasonic assisted hydrothermal method, and established a nonthermal plasma-catalyst (NTP-C) synergistic catalytic system in PPC method to remove α-pinene, a kind of typical VOCs species releasing during biomass fuel drying process. By comparing the removal efficiency of αpinene, the catalytic selectivity of CO 2 and the by-product generation concentrations of O 3 and NO x in the NTP-C synergistic catalytic system, the VOCs catalytic properties of ternary mixed-oxide nanocatalysts under different specific energy density (SED) were evaluated.

Catalysts preparation
The ternary mixed-oxide nanocatalysts were manufactured by ultrasonic assisted hydrothermal process. Fe(NO 3 ) 3 ·9H 2 O (99.9%, analytically pure, Macklin, Shanghai, China), Co(NO 3 ) 2 ·6H 2 O (99.99%, analytically pure, Macklin, Shanghai, China), Ce(NO 3 ) 3 ·6H 2 O (99.9%, analytically pure, Macklin, Shanghai, China) and Mn(NO 3 ) 3 ·6H 2 O (98%, analytically pure, Macklin, Shanghai, China) were used as precursors of iron oxides, cobalt oxides, cerium oxides and manganese oxides, respectively. The nano-TiO 2 particles was generated from the precursor of Tetra-butyl ortho-titanate (TBOT). Fe(NO 3 ) 3 ·9H 2 O and Co(NO 3 ) 2 ·6H 2 O were dissolved into deionized (DI) water orderly at ambient temperature. Under the condition of ultrasonic wave and continuous magnetic stirring, appropriate amount of ethanol was added to the mixture. The stainless steel autoclave with poly tetra fluoroethylene (PTFE) lining was heated at 175°C for 12 h. Subsequently, the autoclave was cooled to the environment temperature, and TBOT was added to the above mixed solution drop by drop and stirred continuously under the action of ultrasonic wave, until the precipitation was formed. Aging again in autoclave at 175°C for 4 h. Finally, the mixture was collected after repeated centrifugation and washing. The precipitates were desiccated at 135°C for 10 h, and calcined at 450°C for 3.5 h under air flow. The prepared nanocatalysts were ground and sieved for VOCs catalytic reaction test and physicochemical characteristics analysis. The nanocatalyst sample with molar ratio of Fe:Co:Ti=1.5:1.0:17.5 was expressed as TiO 2 -Fe x O y -CoO k . The TiO 2 -Fe x O y -CeO k or TiO 2 -Fe x O y -MnO k nanocatalyst was prepared by using Ce(NO 3 respectively, as shown in table 1. As comparison samples, binary mixed-oxide catalysts of Fe x O y -CoO k , Fe x O y -CeO k , Fe x O y -MnO k were also prepared in the manner described above but not included TBOT.

Experimental setup
The schematic diagram of NTP-C synergistic catalytic system for α-pinene degradation was shown in Scheme 1, including α-pinene gas generator, NTP catalytic reactor and exhaust gas detection. Nitrogen as the carrier gas was supplied by standard gas from cylinder. All gas flows were adjusted by mass flow controller. The α-pinene vapor was produced from an evaporator by pumping specified volume of liquid α-pinene regularly, corresponding to mass flow rate about 0.219 g·h −1 . And the α-pinene vapor was directly blown away by the carrier gas. The concentration of α-pinene in the reaction gas mixture was adjusted by amending the standard gas flow rate of N 2 . In the acrylic gas mixing box (10 l internal volume), a certain amount of water vapor was pumped into the gas mixture to control the relative humidity of the reaction gas. The relative humidity of the mixed gas was monitored by the temperature and humidity meter. The experiment was carried out by adjusting initial carrier gas and keeping α-pinene evaporation continuously. Before each concentration test, the gas concentration in the gas mixing box had been stabilized for 30 min.
In this experiment, the plasma reactor was designed in dielectric barrier discharge (DBD) method with a quartz tube as the discharge barrier. The length and outer diameter of the quartz tube was 300 mm and 12 mm, respectively. And the wall thickness was 0.6 mm. The high-voltage electrode of DBD plasma reactor was a copper rod placed in the center of quartz tube in the size of diameter at 3.2 mm and length at 300 mm. While, the lowvoltage electrode of DBD was a copper mesh wrapped on the outer wall of quartz tube. The width of copper mesh was 25 mm, which was the length of discharge area exactly. The CTP-2000 (Suman, Nanjing, China) AC power converter was introduced as the discharge energy generator with voltage, current and frequency controlled accurately. The output voltage waveform was sinusoidal, the frequency was adjusted from 6 kHz to 11 kHz and the peak voltage was arranged from 4 kV to 18 kV. The 0.2 g ternary mixed-oxide nanocatalysts of 40-80 mesh, were loaded into the catalyst bed downstream of the plasma reactor with GHSV at 300000 ml·(g·h) −1 .
The inlet and outlet concentrations of α-pinene in the reactor were supervised by a VOCs gas analyzer Ultra RAE 3000 (RAE, California, USA), and the conversion rates of CO x (CO 2 and CO), and the concentrations of NO x (NO 2 , NO and N 2 O) were measured by a portable special gas analyzer TD500-sh (Shouhe, Beijing, China). The concentration of O 3 was measured by O 3 monitor TYBX31C (Tangyi, Shanghai, China). The varying concentrations of α-pinene and CO x , and by-product generation concentrations of O 3 and NO x were tested and evaluated. The α-pinene conversion and CO 2 selectivity were calculated as follows:

Catalyst characterization
The surface areas and pore size distributions of the three kinds of ternary mixed-oxide nanocatalysts were measured by Tristar II(3020) micro pore analyzer (Maxon, Illinois, USA). The adsorption isotherms of the prepared nanocatalysts for N 2 were measured at 77 K after degasification in vacuum at 623 K for 10 h. Brunauer-Emmett-Teller (BET) and Barrett-Joyner-Halenda (BJH) were used to calculate the specific surface areas and the pore size distributions of the nanocatalysts, respectively. The characteristic peaks of x-ray diffraction (XRD) was tested Bruker D 8 advance analyzer (Bruker, Frankfurt, Germany), under the radiation of Mo Kα, with the diffraction intensity from 10°to 90°, the step size of 0.02°, and the point counting time at 1 s. The element phases in the nanocatalysts were distinguished by comparing the characteristic peaks of XRD patterns with international center for diffraction data (ICDD). The transmission electron microscope (TEM) of JEM 2010 (HR) (JEOL, Tokyo, Japan) was used to capture the advanced microstructure picture of nanocatalysts. Thermal Escalab 250XI (Thermo Fisher, Boston, USA) was adopted as the analytical equipment for x-ray Photoelectron spectroscopy (XPS). The x-ray source was 150 W, the reference line was C 1 s at 284.6 eV, the radiation of Al Kα was 1486.6 eV, the pass energy was 46.95 eV, and the accuracy of binding energy was±0.3 eV.

Results and discussion
3.1. Catalytic activity of mixed-oxide nanocatalysts on α-pinene removal efficiency and CO 2 selectivity 3.1.1. The α-pinene removal activity of mixed-oxide nanocatalysts The variation of α-pinene removal efficiency with different SED was studied under the same test conditions. The ternary mixed-oxide nanocatalysts could further promote the degradation of α-pinene with plasma compared with the binary mixed-oxide catalysts. As shown in figure 1(a), the catalytic conversion of α-pinene was obtained 80% at about SED 1200 J·l −1 when α-pinene was oxidized in a single plasma catalytic reaction. After adding mixed-oxide nanocatalysts into the plasma catalytic system, the removal efficiency of α-pinene was significantly improved, with the order of Among these nanocatalysts, FeCoO x /TiO 2 sample had the best performance with the α-pinene catalytic conversion reaching 83.3% at SED 620 J·l −1 , which was almost half of the energy consumption of single plasma catalytic reaction. In addition, the combination of NTP and mixed-oxide nanocatalysts also significantly improves the CO 2 selectivity compared with the single plasma catalytic reaction, as shown in figure 1(b). At about 620 J·l −1 , the selectivity of CO 2 was only 45.6% under the single plasma catalytic reaction and no more than 60% under NTP-C catalytic system with binary mixed-oxide nanocatalysts. In contrast, when ternary mixed-oxide nanocatalysts were set in NTP-C system, the CO 2 selectivity increased obviously, and the maximum reached 90.6% with TiO 2 -Fe x O y -CoO k nanocatalyst under SED 600 J·l −1 . Therefore, it could be proposed that α-pinene was more easily degraded and completely oxidized to carbon dioxide under the catalytic reaction of plasma combining with ternary mixed-oxide. These results also indicated that mixed-oxide nanocatalysts played a significant part in VOCs degradation and oxidation products formation, which could improve the energy utilization efficiency and VOCs catalytic effects. 3.1.2. The by-product generation of mixed-oxide nanocatalysts O 3 produced during the plasma catalytic reaction process was an undesirable by-product because it could lead to the emission of secondary pollutants. Figure 2(a) showed O 3 production as a function of SED during α-pinene catalytic oxidation. The results showed that the concentration of O 3 in the reaction products catalyzed by NTP without catalysts was significantly higher than that with mixed-oxide nanocatalysts. Meanwhile, the ternary mixed-oxide nanocatalysts exhibited lower O 3 generation concentration compared to the binary ones.
The formation concentration of O 3 decreased with the increase of SED, which was due to the active electrons collided with N 2 and O 2 to generating N and O atoms during the plasma catalytic reaction in the gas mixture. The chemical bond dissociation energies of N 2 and O 2 were 945 kJ mol −1 and 498 kJ mol −1 , respectively [21]. Therefore, under the relatively low SED, O 2 was more likely to react with high-energy electrons, resulting in more oxygen radicals and ozone formation during the plasma catalytic reaction. When the SED increased high enough, oxygen free radicals and ozone could react with N free radicals to form NO or NO 2 . For this reason, the concentration of O 3 began to decrease with the further increase of SED. However, compared with the single NTP catalytic reaction, the plasma combined with mixed-oxide nanocatalysts could effectively reduce the emission of O 3 . When SED was 600 J·l −1 , the export volume of O 3 decreases from 109 ppm to 81 ppm under the catalytic reactions with TiO 2 -Fe x O y -CoO k nanocatalyst. It was inferred that the oxygen atoms decomposed from O 3 on the surface of the nanocatalyst might participate in the catalytic oxidation of α-pinene, or it might convert into chemically adsorbed oxygen to be fixed [6]. But which reaction path would this kind of oxygen atom take still needed a deep investigation during the process of α-pinene catalyzed by NTP combined with ternary mixedoxide nanocatalysts.
In addition, the NO x generation caused by NTP-C process for α-pinene was also studied. The concentrations of NO x generation on mixed-oxide nanocatalysts and blank tube were measured and compared as shown in figure 2(b). In all test conditions, the NO x generation concentrations improved with the SED increasing. Both binary and ternary mixed-oxide nanocatalysts were beneficial to restrain NO x generation. In order to further understand the catalytic reaction process, the NO x composition was analyzed at SED of 200 J·l −1 , 600 J·l −1 and 1200 J·l −1 as exhibited in figure 2(c). It could be seen that the formation of NO 2 was higher than NO and N 2 O in either the single plasma catalytic reaction or the plasma coupled with mixed-oxide nanocatalysts. In consideration of the results of O 3 concentration showed in figure 2(a), it could be inferred that the reaction efficiency of NO oxidation to NO 2 was related to the consumption of O 3 . Furthermore, the concentrations of NO x generation on ternary mixed-oxide nanocatalysts were significantly lower than that on the binary ones in the process of α-pinene catalytic oxidation. Further compared the test data of ternary mixedoxide nanocatalysts, it could be found that, the gain effect of Co, Mn and Ce doping into TiO 2 -Fe x O y on inhibiting the by-product formation of NO x followed the order of TiO 2

The long-term tests with ternary mixed-oxide nanocatalysts
The long-term catalytic oxidation of α-pinene under the plasma catalytic reaction on ternary mixed-oxide nanocatalysts were carried out at about SED of 600 J·l −1 . The catalytic stability of these nanocatalysts were shown in figure 3. In the long-term test, the removal rate of α-pinene on TiO 2 -Fe x O y -CoO k , TiO 2 -Fe x O y -CeO k and TiO 2 -Fe x O y -MnO k nanocatalyst were almost kept at about 83.3%, 68.7% and 60.6%, respectively, which were basically the same as the α-pinene removal rate shown in figure 1. Therefore, the ternary mixed-oxide nanocatalysts with cobalt, manganese or cerium as the doping element had good stability for NTP-C catalytic oxidation of α-pinene.

Characterization of the catalysts 3.2.1. BET analysis
The test data of N 2 adsorption and desorption process was profiled in figure 4, and the pore structure characteristics of these ternary mixed-oxide nanocatalysts in specific surface area (S BET ), total pore volume (V t ) and average pore size (D a ) were summarized in table 2. According to S BET , V t and D a results, the pore structure characteristics of TiO 2 -Fe x O y -CoO k nanocatalyst were better than those of TiO 2 -Fe x O y -CeO k and TiO 2 -Fe x O y -MnO k nanocatalysts. The S BET of TiO 2 -Fe x O y -CoO k nanocatalyst was increased by 36.8% compared with TiO 2 -Fe x O y -CeO k nanocatalyst and 73.8% higher than that of TiO 2 -Fe x O y -MnO k nanocatalyst. In the comparison of specific space volume, the V t of TiO 2 -Fe x O y -CoO k nanocatalyst was 21.5% and 25.5% larger than that of TiO 2 -Fe x O y -CeO k and TiO 2 -Fe x O y -MnO k nanocatalyst, respectively. While, the D a of TiO 2 -Fe x O y -CoO k nanocatalyst was reduced by 39.6% compared with TiO 2 -Fe x O y -CeO k nanocatalyst and 54.9% lower than that of TiO 2 -Fe x O y -MnO k nanocatalyst. It was supposed that, Fe-Co, Fe-Ce and Fe-Mn oxides could be well dispersed on the surface of nano-TiO 2 , and Ti-Fe-Co oxide species were better dispersed on the surface of nanocatalyst in comparison. Thus, the effects of Ti-Fe-Co oxide species on promoting the formation of micropores, increasing S BET and V t , and reducing D a , were much more remarkable. Although the Ti-Fe-Co, Ti-Fe-Ce and Ti-Fe-Mn oxides led to some changes in S BET , V t and D a of the nanocatalysts, the data of pore structure characteristics still stayed in the same order of magnitude. This was due to the mesopores in nano-TiO 2 occupied a very important proportion in these nanocatalysts [37]. Although the formation of micropore could improve the pore structure characteristics, to a certain extent, it would be inhibited by mesoporous existing in the nanocatalysts [59,60].

TEM measurements
The microstructure of these three different kinds of ternary mixed-oxide nanocatalysts was observed by TEM. Figure 5(a) exhibited the TEM image of TiO 2 -Fe x O y -CoO k nanocatalyst. The TiO 2 -Fe x O y -CoO k nanocatalyst was composed of uniform and fine ellipsoidal nanoparticles, and the particle size tended to be consistent. There was no obvious agglomeration between the particles. The surface of the nanoparticles was smooth and the shape was regular. The complete pore structure with both micropore and mesoporous was formed in , it could be observed that the micropore and mesoporous structure of TiO 2 -Fe x O y -MnO k nanocatalyst was significantly reduced. These results were consistent with the data of BET test. In order to have a clearer understanding of the crystalline phase composition, high resolution TEM was carried out on TiO 2 -Fe x O y -CoO k nanoparticle to observe and detect the species structure. As shown in figure 5(d), the collaborative doping of Fe x O y -CoO k species had little effects on basic morphology of anatase TiO 2 with the lattice spacing of 0.352 nm corresponding to (101) plane and 0.235 nm matching to (004) plane exhibiting obviously, which was consistent with XRD results discussed following. Figure 6 showed the XRD results of ternary mixed-oxide nanocatalysts.  (211), 26.72°(220), 32.87°(222) and 56.89°(433), accordingly [61,62]. Fe x O y or CoO x in TiO 2 -Fe x O y -CoO k nanocatalyst had no obvious characteristic peak reflection, which indicated that the active species had high dispersion on the surface of nanocatalyst. It might also be due to the particle size formed was too small to be recognized, or the active species of Fe x O y or CoO x entered the anatase TiO 2 lattice [63,64]. The same situation of low or absent Fe x O y and CeO x peaks intensity occurred in TiO 2 -Fe x O y -CeO k nanocatalyst, it could be considered that Fe x O y and CeO x active species also had high dispersion.

XPS and EDX analysis
For obtaining a better understanding of the oxidation state and surface composition of metals in ternary mixedoxide nanocatalysts, XPS analysis was introduced. Figure 7 exhibited the XPS spectra of Fe 2p, O 1 s, Ti 2P, Co 2p, Ce 3d and Mn 2p in these three kinds of nanocatalysts. According to Gaussian fitting, the valence states of each element were calculated. Table 3 summarized the specific binding energies and individual element concentrations at different valence states.
In TiO 2 -Fe x O y -CoO k , TiO 2 -Fe x O y -CeO k and TiO 2 -Fe x O y -MnO k , the XPS spectra of Fe 2p were shown in figure 7(a), in which two separate peaks according with Fe 2p 3/2 (appearing at 709.9 eV approximately) and Fe2p 1/2 (appearing at about 723.8 eV) [65]. Meanwhile, the Fe 3+ peak appeared at about 717.8 eV was regarded as the satellite peak of Fe 2 O 3 . The wide Fe 2p 3/2 peak could be divided into two overlapping characteristic peaks. A characteristic peak was located at 709.8 eV, which belonged to Fe 2+ . Another characteristic peak at about 711.6 eV belonged to Fe 3+ [66]. These two characteristic peaks confirmed that iron coexisted in the valence states of +2 and +3 in TiO 2 -Fe x O y -CoO k , TiO 2 -Fe x O y -CeO k and TiO 2 -Fe x O y -MnO k , and their proportions were shown in table 3. For these three kinds of ternary mixed-oxide nanocatalysts, there were different electron transfer processes between the doping elements during the catalytic oxidation reaction of VOCs. The electron transfer process in TiO 2 -Fe x O y -CoO k nanocatalyst was Fe 2+ +Co 3+ ↔Fe 3+ +Co 2+ [66]. In TiO 2 -Fe x O y -CeO k nanocatalyst, the electron transfer process was Fe 2+ +Ce 4+ ↔Fe 3+ +Ce 3+ [67]. While the electron transfer process in TiO 2 -Fe x O y -MnO k nanocatalyst was Fe 2+ +Mn 4+ ↔Fe 3+ +Mn 3+ [65]. Figure 7(b) showed the O 1 s spectra of these three kinds of ternary mixed-oxide nanocatalysts. According to the results of curve-fitting calculation, the spectrum of O 1 s could be split into two overlapping peaks, one was the characteristic peak of lattice oxygen (O L ), which appeared between 530.2 eV and 530.3 eV, and the other was the characteristic peak of chemisorbed oxygen (O ad ), with the center at 531.2-531.6 eV [68]. On the surface of TiO 2 -Fe x O y -CoO k sample, the chemisorbed oxygen component reached 43.6%, which was much higher than that of TiO 2 -Fe x O y -CeO k nanocatalyst (36.1%) and TiO 2 -Fe x O y -MnO k nanocatalyst (24.8%), as shown in table 3. Because of the high fluidity, the chemisorbed oxygen was regarded as the most energetic type of oxygen [69]. At the same time, based on the O 1 s spectra of TiO 2 -Fe x O y -MnO k nanocatalyst, it could be realized that the chemisorbed oxygen in TiO 2 -Fe x O y -CeO k and TiO 2 -Fe x O y -CoO k nanocatalysts shifted to higher binding energies slightly, from 531.7 eV of TiO 2 -Fe x O y -MnO k to 531.8 eV of TiO 2 -Fe x O y -CeO k and 531.9ev of TiO 2 -Fe x O y -CoO k , respectively. The binding energy of lattice oxygen also showed a similar change trend in the three nanocatalysts. Figure 7(c) showed the XPS spectra of Ti 2p as the nanocatalyst supports, which included Ti 2p 1/2 characteristic peak located at around 463.7 eV and Ti 2p 3/2 characteristic peak at about 457.6 eV [70]. The results showed that Ti in the valence state of +4 was stable and dominant on the nanocatalyst surface. By comparing the XPS spectra of the characteristic peaks of Ti in the three kinds of nanocatalysts, it could be found that the characteristic peaks of Ti on the catalyst support did not change significantly, although the dopings in ternary mixed-oxide nanocatalysts were different, such as cobalt oxides, cerium oxides or manganese oxides.
For TiO 2 -Fe x O y -CoO k sample, the Co 2p 1/2 characteristic peak at 796.7 eV approximately and Co 2p 3/2 characteristic peak at about 780.7 eV were both observed in the XPS spectrum of Co 2p. And there were two  satellite peaks appearing at 787.1 eV and 803.6 eV [12], which were closed to the two main peaks of Co 2p 1/2 and 2p 3/2 , respectively, as shown in figure 7(d). In the relatively high binding energy regions, the two wider and milder satellite structures were produced by metal-to-ligand charge transfer, also referred to as the shakeup course of high spin cobalt [71]. The Co 2p 3/2 characteristic peak was composed of Co 3+ and Co 2+ spectrum with binding energy of 780.3 eV and 782.2 eV accordingly. The results showed that Co 2+ and Co 3+ coexisted on TiO 2 -Fe x O y -CoO k catalyst surface, and the atomic composition of Co 3+ achieved 61.3% approximately. The cobalt species in +3 valence state were existed in relatively high activity and produced more anion defects, which could enhance the adsorption and oxidation reaction process of VOCs. In TiO 2 -Fe x O y -CeO k nanocatalyst, the XPS spectrum of Ce 3d results were shown in figure 7(e). The Ce 3d pattern matched the core holes of spin-orbit splitting of 3d 5/2 and 3d 3/2 , which could be further divided into multiple peaks of u and v based on the binding energies, named as u, u′, u″, u‴ and v, v′, v″, v‴, respectively [72]. The u′ and v′ peaks exhibited the presence of Ce 3+ ion in TiO 2 -Fe x O y -CeO k nanocatalyst, and the u, u″, u‴ and v, v″, v‴ peaks demonstrated the existence of Ce 4+ ion [11]. The whole Ce 3d pattern confirmed that Ce 3+ and Ce 4+ species coexisted on the surface of TiO 2 -Fe x O y -CeO k nanocatalyst. The Ce 3+ ion was an important inducement to form charge balance and unsaturated chemical bond. Between the oxides of Fe and Ce, the negative charge shifted from Fe 2+ to Ce 4+ enhanced the collaboration between Fe and Ce. TiO 2 -Fe x O y -CeO k nanocatalyst with Ce 3+ and Ce 4+ redox couple was more likely to construct oxygen vacancy on catalyst surface which was conducive to the adsorption and chemisorption of oxygen.
As shown in figure 7(f), Mn 2p characteristic peaks of TiO 2 -Fe x O y -MnO k nanocatalyst in the XPS spectrum were at around 653.3 eV and 642.4 eV [73]. The asymmetric curve of Mn 2p 3/2 peak could be divided into three multi overlapping peaks based on Gaussian fitting calculation, which further confirmed the coexistence of complex Mn ion with different valence states in TiO 2 -Fe x O y -MnO k nanocatalyst. The Mn 2+ peak appeared at about 641.5 eV, the Mn 3+ peak arose at around 642.6 eV, and the Mn 4+ peak occurred at 644.3 eV approximately. It was difficult to distinguish the complex MnO x with trivalent state, due to the binding energy difference less than 2.8 eV. In order to accurately identify the concentration and atomic composition of Mn ions on TiO 2 -Fe x O y -MnO k nanocatalyst surface, the quantitative analysis was carried out according to the separation peak coverage area calculation, and the results were shown in table 3. The oxidation capacity of MnO x mixed species follows MnO 2 >Mn 2 O 3 >Mn 3 O 4 [74]. In TiO 2 -Fe x O y -MnO k nanocatalyst, manganese was mainly dispersed on the surface of catalyst in +3 valence state, as shown in table 3. These results indicated that TiO 2 -Fe x O y -MnO k nanocatalyst had the potential of further improvement of oxidation capacity.
As shown in figure 8, the perform of EDX analysis for the confirmation of elemental composition in ternary mixed-oxide nanocatalysts was carried out. According to EDX, the atomic composition was in accord with the data exhibited in table 1, with deviation less than 1%.
3.3. Mechanism analysis on NTP-C synergistic effect with ternary mixed-oxide nanocatalysts Figure 9 proposed the NTP-C synergistic effect with combination of TiO 2 -Fe x O y -MO k and NTP process based on the test results obtained in this research and other achievements reported previously [75,76]. Considering the similar structure-activity relationship with VOCs degradation, the reaction mechanism in the NTP-C synergistic process can be simplified into three aspects [21]. (1) VOCs are easily adsorbed on the surface of the catalyst, resulting in a strong collision reaction with the active components [4]. (2) The surface active sites on the catalyst participate in the degradation of pollutants under the discharge energy [77]. (3) The reaction activation energy of O 3 with VOCs is also low, which easily leads to the formation of CO and CO 2 [78]. During the NTP-C catalytic α-pinene process with ternary mixed-oxide nanocatalysts, three reaction paths accelerated the catalytic oxidation. First, the high-energy particles and active species formed by NTP, such as energetic electron and oxygen radical, could decompose α-pinene in collisions. Second, the mixed transition metal oxides could offer strong redox couples, such as Fe 3+ ↔Fe 2+ [79], Co 3+ ↔Co 2+ [80], Ce 4+ ↔Ce 3+ [81] and Mn 4+ ↔Mn 3+ (Mn 3+ ↔Mn 2+ ) [33,82], which obviously enhanced the electron transfer during the α-pinene catalytic oxidation reaction. Third, the nano-TiO 2 had the dual function of catalyst support and catalytic component [83]. On one hand, nano-TiO 2 improved the nanocatalysts surface areas and enhanced the transition metal oxides dispersibility, and on the other hand, the reactive species, such as * O, * C 10 H 16 and * O 3 , could also generated on the surface of nano-TiO 2 and further transformed into the final products. Therefore, the NTP-C synergistic effect with ternary mixed-oxide nanocatalysts exhibited higher catalytic oxidation activity compared to that with binary mixed-oxide nanocatalysts or without catalysts.

Conclusions
The α-pinene catalytic performance of ternary mixed-oxide nanocatalysts combined with NTP by was systematically researched based on the comparision of catalytic reaction efficiencies and the analysis of nanocatalysts properties. The catalytic performance of mixed-oxide nanocatalysts combined with NTP were compared in terms of α-pinene removal rate, CO 2 selectivity, by-product generation concentrations of O 3 and NO x . The results showed the introduction of ternary mixed-oxide nanocatalysts into plasma catalytic reaction could significantly improve the catalytic conversion of α-pinene and reduce the by-products generation. The removal efficiency of α-pinene on TiO 2 -Fe x O y -CoO k nanocatalyst reached 83.3% at about 620 J·l −1 , which was almost half of the energy consumption of plasma catalytic reaction. At the same time, TiO 2 -Fe x O y -CoO k nanocatalyst also showed the best CO 2 selectivity, which reached more than 90%. Compared with TiO 2 -Fe x O y -CeO k and TiO 2 -Fe x O y -MnO k nanocatalysts, the excellent performance of TiO 2 -Fe x O y -CoO k nanocatalyst was consistent with the results of microstructure characterization. Considering the side reaction productions of O 3 and NO x , the results showed the reaction efficiency of NO oxidation to NO 2 was related to the consumption of O 3 . Compared with Mn doping, Co or Ce doping was more beneficial to inhibit the formation of by-product of NO x .
According to nanocatalysts properties analysis, the N 2 adsorption-desorption experiments showed that mixed-oxides of Fe-Co, Fe-Ce and Fe-Mn in these three kinds of nanocatalysts changed the S BET , V t and D a , but the variations belonged to the same order of magnitudes. The TEM results showed the surface of TiO 2 -Fe x O y -CoO k nanocatalyst was smoother, and the shape of nanocatalyst particle was more regular with forming complete micropores or mesopores structures, compared with TiO 2 -Fe x O y -CeO k and TiO 2 -Fe x O y -MnO k nanocatalysts. According to XRD results, there were strong and significant diffraction peaks of anatase TiO 2 in all these three ternary mixed-oxide nanocatalysts. While CoO x and CeO x active species had higher dispersion than MnO x . Iron coexisted in +2 and +3 valence states in TiO 2 -Fe x O y -MO k (M=Co, Ce, Mn) nanocatalysts. The mixed valence of Co, Ce and Mn had an important influence on the catalytic oxidation Figure 9. Plausible reaction mechanism for the α-pinene removal by NTP-C synergistic effects with ternary mixed-oxide nanocatalysts.
performance of ternary mixed-oxide nanocatalysts. Finally, we believed that a clearer understanding of the performance of ternary mixed-oxide nanocatalysts coupled with NTP for the catalytic oxidation of α-pinene was profiled, and a new direction of optimizing the composition of nanocatalysts for improving the catalytic efficiency and selectivity of VOCs was proposed.