Toluene, a typical volatile organic compound (VOC), is emitted from architectural coatings, motor vehicles and various industrial processes [1, 2], and is an important precursor for forming PM2.5, O₃ and photo-chemical smog, which are harmful to the environment and human health [3-5]. Over the decades, numerous technologies have been developed to limit the emission of VOCs, using techniques such as adsorption, photocatalysis, catalytic combustion, bio-decomposition and membrane separation[6-9]. In recent years, non-thermal plasma (NTP) for the degradation of VOCs has attracted significant attention because of its high efficiency in the removal of VOCs at ambient temperature and pressure. This technology functions through the generation of hard electrons, active oxygen species, hydroxyl radicals and energetic nitrogen species to oxidize the VOCs [10-12]. However, high energy consumption by the plasma and large amounts of by-products have significantly limited its industrial applications [13, 14]. A promising strategy to overcome these drawbacks is the combination of NTP and catalysis, which has the potential to enhance the reaction rate, product selectivity and energy utilization ratio [15-17]. The catalysts used thus far in combined plasma-catalytic systems include various metallic oxides, such as those of Mn [13, 15-18], Co [19, 20], Ni [18], Ti [13, 19], Cu [18, 20], Ag [16, 17], Fe [11, 18]and Ce [18-20]. The active oxygen species generated from the decomposition of O₃, such as atomic oxides and superoxide radicals, are reported to play important roles in the oxidation of VOCs in plasma-catalytic processes [10, 21], while among the transition metal oxides, MnO₂ is among the most active for the decomposition of O₃ [22, 23]. MnO₂ shows a wide diversity of phase structures, crystal morphologies and tunnel structures, which significantly influence its catalytic performance [24-27]. As such, the effect of MnO₂ phase structures on the plasma-catalytic degradation of VOCs merits detailed investigation.

In this work, α-, β-, γ-and δ-MnO₂ catalysts were synthesized by a one-step hydrothermal method. The relationship between the plasma-catalytic performance of toluene degradation and the different phase structures of MnO₂ was investigated. A surface dielectric barrier discharge (SDBD) reactor was utilized to produce the plasma. The catalysts and plasma were then combined in an in-plasma catalytic system by placing the catalysts in the SDBD reactor. Powder X-ray diffraction (XRD), Raman spectroscopy (Raman), scanning electron microscopy (SEM), transmission electron microscopy (TEM), the Brunauer-Emmett-Teller (BET) method, H₂ temperature-programmed reduction (H₂-TPR) and X-ray photoelectron spectroscopy (XPS) were used to evaluate the relationship between the phase structure and plasma-catalytic performance. The by-products of toluene degradation in the plasma-catalytic system were analyzed to deduce the reaction mechanism of toluene over the MnO₂catalysts.

Four different phase structures of MnO₂ were synthesized by the hydrothermal process as described in previous reports [25, 27, 28]. All reagents were of analytical grade and used without further treatment. For the preparation of α-MnO₂ [29], 2.37 g KMnO₄was dissolved in 0.4 mol/L CH₃COOH solution (150 mL). After stirring magnetically for about 0.5 h to form a homogeneous solution, it was transferred to a Teflon-lined stainless steel autoclave (200 mL), sealed and maintained at 140 ℃ for 12 h in an oven. Thereafter, the autoclave was naturally cooled to room temperature and the sediment in the solution was washed with distilled water and ethanol several times, and dried at 80 ℃ for about 12 h. Similar to the α-MnO₂, for the preparation of β-MnO₂, 6.42 g (NH₄)₂S₂O₈ and 3.17 g MnSO₄·H₂O were mixed in 150 mL distilled water at 150 ℃ for 16 h. To obtain γ-MnO₂[30], 6.32 g MnSO₄·H₂O was reacted with 8.578 g (NH₄)₂S₂O₈ at 90 ℃ for 24 h. The δ-MnO₂ was obtained from the reaction of 0.52 g MnSO₄·H₂O and 2.81 g KMnO₄ at 200 ℃ for 24 h [27].

Nitrogen adsorption apparatus (JW-BK132F, Beijing JWGB Sci. & Tech. Co., Ltd, Beijing, China) was used to measure the specific surface area and pore characteristics by the multipoint BET method. The samples were preprocessed at 200 ℃ prior to the measurements.

The microstructures and morphologies of the prepared samples were investigated by SEM (Ultra 55, Carl Zeiss AG, USA) and TEM (JEM-2010, Japan). Samples were prepared by coating the powder onto a conductive tape for the SEM measurements. For the TEM measurements, samples were prepared by dispersing the powder in ethanol solvent and dropping onto the membrane.

Powder XRD was conducted using a powder diffractometer with Cu K_α radiation (Model D/max RA, Rigaku Co., Japan). The data were collected at scattering angles (2θ) ranging from 10° to 80° with a step size of 0.02°.

H₂-TPR measurements were conducted on a Chemisorption Analyzer equipped with a custom-made thermal conductivity detector (TCD). Prior to the measurement, a precisely weighed 50 mg sample was purged with He at 200 ℃ for 1 h and then naturally cooled to room temperature, then purged with N₂ containing 6% H₂ (30 mL/min) and heated from 100 to 600 ℃ at a linear heating rate of 10 ℃/min.

XPS was performed using a Thermo ESCALAB 250 instrument, with Al K_α X-ray radiation (hν = 1486.6 eV) at 150 W as the excitation source. The data were corrected by setting the binding energy of adventitious carbon (C 1s) at 285.0 eV.

Raman spectroscopy was performed at room temperature with a resolution of approximately 0.6 cm⁻¹using an SPEX-1403 laser. A back-scattering configuration was used to excite the crystals with an Ar-ion laser at a wavelength of 514.5 nm (Raman: Lab RAM-HR, SPEX-1403, France).

Fig. 1 shows a schematic diagram of the experimental setup, which consisted of three parts: a tabular plasma-catalytic reactor (volume = 140 mm × 50 mm × 30 mm), a reaction gas supply system and an analytical instrument. Two planar SDBD generators were installed on the top and bottom plates of the tabular plasma-catalytic reactor, respectively. The SDBD generator (Fig. 2) consisted of three parts: a quartz glass plate (170 mm × 80 mm × 1 mm) as the dielectric, a high-voltage electrode (copper plate, 140 mm × 50 mm × 1 mm) and a grounded electrode (stainless steel net, 160 mm × 70 mm × 1 mm). The SDBD plasma was generated using a pulsed power supply system created in-house. Fig. 3 shows a simplified circuit diagram of the system, which was based on fast MOSFETs, and could produce high-voltage pulses with a peak voltage up to 6 kV and a width of 1-4 μs at a repetition rate of up to 30 kHz. The pulse energy was up to 30 mJ/pulse, and the average power was up to 1 kW. Twelve frosted-glass slides (160 mm × 28 mm × 1.25 mm) for the loading of catalysts were vertically placed in the cavity of the tabular plasma-catalytic reactor. The catalysts were coated onto the glass slides using a dip-coating method, with 0.3 g catalyst used for each glass slide in each test. The gas stream was composed of (120 ± 5) × 10^-4% toluene, 20 vol% O₂ and 80 vol% N₂ (total flow rate = 2 L/min), and was mixed in a mixing chamber (0.5 L) before entering the reactor. The experiments were carried out at atmospheric pressure and ambient temperature, and the decomposition experiments using the plasma were initiated when the concentration of toluene at the outlet reached a steady state (i.e., the concentration at the outlet was approximately the same as at the inlet). The results of blank experiments indicated that the four catalysts had no capacity for the catalytic oxidation of toluene when used alone.

The concentration of toluene and production of CO/CO₂ were evaluated on-line using a chromatograph (GC, PuxiG5, China) equipped with two flame ionization detectors and a nickel converting equipment for the determination of CO/CO₂. The O₃ and NO_x concentrations were determined by an O₃detector (UV-100, Eco Sensors, USA) and NO_x detector (Photon Ⅱ + PGD100, Madur, Austria).

The off-gas contained various gaseous products, which were captured using an adsorption tube (Tenax TA/Tenax GR) at the specific input energy (SIE) of 160 J/L for 30 min. The adsorbed compounds were released and injected into a GC/MS (Agilent 7890A GC equipped with an Agilent 5975C MS) using a thermal desorption instrument (TDI, PERSEE-TP7, China).

The SIE in the tabular plasma-catalytic reactor, toluene removal efficiency (η, %), carbon dioxide yield and carbon monoxide yield (CO_x yield, %) were calculated as follows. The waveform, current and voltage were obtained using a digital oscilloscope (TDS 2012C, Tektronix), current probe (CP8030A, Zhiyong) and high-voltage probe (P6015A, Tektronix), respectively.

(1)

(2)

(3)

(4)

(5)

The toluene removal efficiencies of the four MnO₂ polymorphs via plasma-catalytic degradation are depicted in Fig. 4 as a function of the SIE. In this figure, we can see that the removal efficiencies improved as the SIE was increased, regardless of the catalyst. When the NTP was used alone, the toluene removal efficiency was 32.5 % at the SIE of 160 J/L. In comparison, when the four MnO₂ polymorphs were introduced, the removal efficiencies over α-MnO₂, β-MnO₂, γ-MnO₂ and δ-MnO₂increased to ca. 78.1%, 47.4%, 66.1% and 50.0%, respectively, at the SIE of 160 J/L. Thus, the introduction of catalysts into the plasma significantly improved the toluene removal efficiency, with α-MnO₂ affording the greatest improvement.

The CO_x yield, the amount of NO_x generated and the concentration of O₃generated by plasma alone and by the plasma-catalytic processes with different catalysts are illustrated in Fig. 5. Like the toluene removal efficiencies, the CO_x yields also increased with increasing SIE, regardless of the catalyst (Fig. 5(a)). With the introduction of the MnO₂ catalysts, the CO_x yields of the plasma-catalytic systems were greatly improved, in particular for α-MnO₂. The generation of NO_x and O₃ with and without the catalysts are shown in Figs. 5(b) and (c), respectively. NO_x was generated by the reaction between N₂ and O₂ in the discharge zone of the NTP reactor [31, 32], and the amount increased with increasing SIE. The concentration of NO_x generated in our experiments was similar to previously reported values [33, 34]. Fig. 5(c) presents the O₃production in the plasma system and the amount of residual O₃ in the plasma-catalytic systems as a function of the SIE, showing positive correlations between the O₃ production and the SIE values. The O₃ production increased from 200 × 10^-4% to 1000 × 10^-4% when the SIE was increased from 50 to 350 J/L. O₃ is reported [10] to play a significant role in the degradation of toluene. The in-situ decomposition of O₃ leads to the formation of atomic oxygen, which can react with pollutants. However, the residual O₃ in the exhaust gas is itself a source of secondary air pollution, so its amount must be carefully controlled. As shown in Fig. 5(c), the introduction of the MnO₂catalysts heavily reduced the concentration of residual O₃, except in the case of β-MnO₂.Comparison of the toluene removal efficiencies in Fig. 4 with the residual O₃ concentrations in Fig. 5c indicates that the adsorption of O₃ at the catalyst surface played a key role in the degradation of toluene. The decomposition of O₃ proceeds through two kinetically irreversible steps, i.e., the adsorption of O₃ on the catalyst surface and the desorption of O₂. The intermediate products of this reaction are peroxide species, which in turn react with toluene [22].

The catalytic performances of the MnO₂ catalysts were closely related to their physical and chemical properties. As reported [26, 35], all of the different MnO₂ polymorphs are built from units of edge-sharing MnO₆ octahedra. Depending on the different ways in which the MnO₆octahedra are interlinked, the resulting crystal phase structures are composed of different proportions of tunnels or interlayers [26, 35]. The crystal phase structure of α-MnO₂ consists of one-dimensional (2 × 2) and (1 × 1) tunnels, which are composed of double chains of edge-sharing MnO₆octahedra, and stabilized by K⁺ or H₃O⁺ [36, 37]. The pyrolusite form, β-MnO₂(P4₂/mnm), is the densest and most stable polymorph of MnO₂, having a rutile-type structure with an infinite chain of MnO₆octahedra sharing opposite edges. Each chain is corner-linked with four like chains, as a consequence of the formation of (1 × 1) channels in the β-MnO₂ structure [37, 38]. The structure of γ-MnO₂is characterized by the random intergrowth of pyrolusite layers (1 × 1 channels) and a ramsdellite matrix (2 × 1 channels), in which the basic building blocks of MnO₆octahedra share the edges and corners. As a result, the structure of γ-MnO₂ is the most complex [30, 39]. Finally, δ-MnO₂is constructed from sheets of edge-sharing MnO₆octahedra forming a 2D layer structure, separated by layers of OH⁻, K⁺ or H₂O [40].

Chen et al. [24] have suggested that the channel structure of MnO₂ catalysts influences their capacity to adsorb oxygen at the surface, which is directly related to the oxidation of NO. Liang et al. [25] found that CO molecules were chemisorbed onto MnO₂ nanorods via direct contact with the tunnel structure of the rods, thus greatly influencing the catalytic performance for CO oxidation. Zhang et al. [27] concluded that the interlayer structure of MnO₂was beneficial for the diffusion and adsorption of HCHO molecules, and promoted their oxidation more strongly than the (2 × 2) tunnel structure. Indeed, our experimental results were in agreement with these reports, the tunnel structure will influence the performance of the catalysts. The toluene removal efficiencies of the different MnO₂catalysts were linked to the tunnel structure, as discussed below.

The crystallographic structures of the various MnO₂ catalysts were investigated by XRD. The XRD patterns of the used MnO₂ catalysts were also recorded. As shown in Fig. 6, the four MnO₂ catalysts showed different reflections, all of which could be indexed to the respective pure crystal phase (i.e., α-MnO₂(JCPDS Card No. 44-0141), β-MnO₂(JCPDS Card No. 24-0735), γ-MnO₂(JCPDS Card No. 14-0644) and δ-MnO₂(JCPDS Card No. 80-1098)). The diffraction peaks for α-MnO₂ and β-MnO₂ were much greater in intensity and narrower in width compared with those for γ-MnO₂ and δ-MnO₂. The poor crystallinity of γ-MnO₂ was mainly attributed to the incorporation of defects during synthesis, which involved a random intergrowth of pyrolusite layers within the ramsdellite matrix [39]. As for δ-MnO₂, the large set of diffraction peaks indicated a high degree of structural disorder in certain crystallographic directions [41]. The characteristic diffraction peaks remained present in the XRD profiles of all the used catalysts. This verified that the crystal phases of MnO₂ were retained after the plasma-catalytic reaction. However, the positions of the diffraction peaks for δ-MnO₂were shifted to higher diffraction angles by about 0.2°, revealing a decrease in the d spacing.

Raman spectroscopy was also performed to investigate in more detail the crystal structures of α-, β-, γ-and δ-MnO₂before and after the reaction. As shown in Fig. 7, the peaks between 500 and 700 cm^-1 were assigned to the stretching modes of MnO₆octahedra. The Raman scattering spectrum of α-MnO₂, containing K⁺, contained bands at 183, 192, 384, 572, 628 and 748 cm^-1[36, 42]. The bands at 572 and 628 cm^-1originated from the breathing vibrations of MnO₆ octahedra within a tetragonal hollandite-type framework. For β-MnO₂, one strong band at 643 cm^-1 and two weak bands at 529 and 750 cm^-1were identified as the stretching modes of MnO₆ octahedra [36]. Since the synthetic process for γ-MnO₂ results in a random intergrowth of pyrolusite layers (β-MnO₂) within a ramsdellite (R-MnO₂) matrix, the spectra of γ-MnO₂ consisted of a superposition of the peaks of these two crystal forms of MnO₂. The sharp band at 630 cm^-1 and the weak band at 575 cm^-1 corresponded to R-MnO₂, and the strong band at 656 cm^-1 along with the weak band at 530 cm^-1corresponded to β-MnO₂[38]. The Raman spectra of δ-MnO₂ displayed two main bands at around 570 and 638 cm^-1. The band at 638 cm^-1 may have originated from the symmetric stretching vibration (Mn-O) of the MnO₆ groups, and the band at 570 cm^-1 was identified as the Mn-O stretching of the baseline plane of the MnO₆ sheet [40]. Comparing the Raman spectra of the MnO₂ catalysts before and after the plasma-catalytic reaction, it was found that the peak at around 570 cm^-1, which was considered as the Mn-O lattice vibration in the MnO₂ octahedral lattice, decreased in intensity for δ-MnO₂and increased for γ-MnO₂. Additionally, the band at around 637 cm^-1, which can be assigned to the Mn₃O₄structure, increased greatly in intensity for β-, γ-and δ-MnO₂[43]. In contrast, no significant structural change was observed for α-MnO₂ before and after the plasma-catalytic reaction.

The BET specific surface areas (A_BET), total pore volumes (V_pore) and average pore sizes (D_pore) of the different MnO₂catalysts are summarized in Table 1. It can be seen that δ-MnO₂ had the highest specific surface area and total pore volume among the investigatedcatalysts, while γ-MnO₂ had the highest average pore size.

The surface structure and morphology of the MnO₂ catalysts were studied by SEM and TEM analyses. As shown in Figs. 8 and Fig. 9, α-MnO₂, which was composed of uniform nanorods acting as dispersive nanowires, displayed a dendritic nanostructure. The lengths of the nanowires ranged from 0.4 to 1.6 µm, with the diameters ranging from 20 to 40 nm. The panoramic morphology of the β-MnO₂ powder consisted of rod-like crystals, which were 40-200 nm in diameter and 0.4-2 µm in length. As for γ-MnO₂, it consisted of microparticles with a morphology resembling sea urchins, covered with numerous crystalline and well-ordered nanowires. The nanowires had lengths ranging from 0.9 to 2.0 μm and diameters of around 10-20 nm. δ-MnO₂ displayed a hierarchical architecture with microspherical cores and nanosheet coronas, forming a globular morphology with diameters of 1-2 μm and widths of 10-20 nm.

The H₂-TPR profiles of the four MnO₂ catalysts were measured to explore the redox potentials of the catalysts. As shown in Fig. 10(a), the β-MnO₂contained two narrow peaks at 314 and 330 ℃ and a broad peak at 438 ℃. The hydrogen consumption calculated from the lower-temperature peak was double that of the higher-temperature peak. According to previous reports [44, 45], the low-temperature peaks originate from the reduction of MnO₂ to Mn₂O₃and of Mn₂O₃to Mn₃O₄, whereas the high-temperature peak was attributed to the reduction of Mn₃O₄ to MnO. The profile of γ-MnO₂contained three reduction peaks, which were all shifted to slightly to lower temperatures, i.e., 280, 288 and 403 ℃, but was otherwise qualitatively similar to that of β-MnO₂. The areal ratio of the higher-temperature peak to the lower-temperature peak was measured to be around 1:1. The low-temperature peaks and high-temperature peak in this catalyst can be attributed to the reduction of MnO₂ to Mn₂O₃and of Mn₂O₃ to MnO, respectively. In contrast, the TPR profiles of α-and δ-MnO₂ were very different from those of β-and γ-MnO₂. α-MnO₂showed two reduction peaks at 316 and 326 ℃. Similarly, δ-MnO₂exhibited two overlapping reduction peaks at 280 and 300 ℃. The detailed reduction process giving rise to these peaks is not clear, but probably involved the reduction of MnO₂ to MnO with the simultaneous reductions of Mn₂O₃and Mn₃O₄. For all four MnO₂ polymorphs, the main reduction product was MnO. This was deduced from the observation of green powders after the H₂-TPR experiments [25]. Based on the H₂-TPR analyses, it was concluded that the initial reduction temperatures of the four MnO₂catalysts decreased in the order of γ-≈ δ- > α-≈ β-. However, the initial H₂ consumption rate per mole of Mn is considered a better indicator for comparing the reducibility of the catalysts [46]. Fig. 10(b) shows the initial H₂ consumption rates per mole of Mn before reduction reached 20% (i.e., before any phase transformation) versus inverse temperature. The results indicate that the initial H₂ consumption rates decreased in the order of α-MnO₂ > β-MnO₂ > δ-MnO₂ > γ-MnO₂. Thus, even though γ-MnO₂and δ-MnO₂ had shown the lowest initial reduction temperatures, α-MnO₂achieved the highest initial H₂ consumption rate. This might imply that α-MnO₂ possessed better oxygen mobility, allowing more oxygen to be produced and adsorbed on the surface of the catalyst and become available to the plasma-catalytic reaction.

XPS analyses were conducted to identify the superficial elemental species for each catalyst, specifically the surface content of Mn and O. As shown in Fig. 11(a), each catalyst showed two dissymmetric peaks located at 642.2 (Mn 2p_3/2) and 653.9 eV (Mn 2p_1/2). The Mn 2p_3/2and Mn 2p_1/2 peaks could each be decomposed into two sub-peaks, corresponding to different Mn oxidation states. The sub-peaks located at 641.9 (or 642.1) and at 653.5 (or 653.7) eV were assigned to Mn³⁺(2p_3/2) and Mn³⁺(2p_1/2) ions, respectively, while the sub-peaks situated at 642.8 and 654.2 eV were assigned to Mn⁴⁺(2p_3/2) and Mn⁴⁺(2p_1/2) ions, respectively [27, 47]. By quantitative analysis of the Mn 2p spectra, the molar ratios of Mn⁴⁺/Mn³⁺ on the surface were calculated, as presented in Table 2. The different crystal phase structures of the MnO₂ catalysts exhibited clearly different surface Mn⁴⁺/Mn³⁺molar ratios, which followed the sequence δ-MnO₂ > α-MnO₂ > γ-MnO₂ > β-MnO₂.

The XPS spectra of O 1s are shown in Fig. 11(b). The O 1s spectra could be divided into three regions, wherein the binding energy (BE) in the range of 529.2-530.8 eV was denoted O_lat(Mn-O-Mn bond), corresponding to lattice oxygen species; the BE at 531.0-532.6 eV was assigned to oxide defects or surface-chemisorbed oxygen species (O_sur, Mn-OH bond), which originated from surface oxygen vacancies; and the BE at 532.6-534.0 eV corresponded to adsorbed molecular water (O_ads, H-O-H bond) [48-50]. From Fig. 11(b), it can be seen that O_lat, O_surand O_ads were located at 529.8 ± 0.2, 531.6 ± 0.2 and 533.1 ± 0.3 eV, respectively. Relative to the other three catalysts, the O 1s peak of β-MnO₂ was shifted by 0.4 eV toward lower binding energies [24, 25, 27]. It is well known that surface-chemisorbed oxygen species are more active than lattice-bound oxygen species, and play critical roles in oxidation reactions because of their higher mobility [48, 51]. Therefore, the molar ratios of surface-bound oxygen, i.e., O_sur/O_lat, were also measured and are given in Table 3. The ratios O_sur/O_lat decreased in the order γ-MnO₂ > α-MnO₂ > β-MnO₂ > δ-MnO₂, which is consistent with the sequence of catalytic activities for toluene oxidation for the four MnO₂ catalysts. In addition, some hydroxyl species were also present on the surface of α-MnO₂.

From the aforementioned results, it can be concluded that the catalytic performance of the different MnO₂catalysts in the plasma-catalytic reaction was related to the variations in their tunnel structure, crystal phase and other physicochemical properties. The XPS results indicated that the molar ratios of O_sur/O_lat on the catalyst surface decreased in the sequence γ-MnO₂ > α-MnO₂ > β-MnO₂ > δ-MnO₂. The H₂-TPR experiments revealed that α-MnO₂ possessed the highest oxygen mobility. The tunnel structure analyses showed that α-MnO₂ was composed of one-dimensional channels with (2 × 2) and (1 × 1) tunnels, which was beneficial for the adsorption and diffusion of toluene to active sites. The XRD and Raman spectra of fresh and used MnO₂ catalysts suggested that all of the polymorphs except α-MnO₂ underwent structural changes during the plasma-catalytic reaction. All of these results support the designation of α-MnO₂as the best-performing catalyst in the plasma-catalytic degradation of toluene.

Moreover, it has been reported that the Mn-O bond strength of MnO₂ can also affect the catalytic performance of MnO₂catalysts [25, 37, 39]. An increase of the Mn-O bond length indicates a decrease of the bond strength [25]. The average Mn-O bond lengths of the α-, β-, γ-and δ-MnO₂ catalysts [25, 37, 39] were calculated as 1.98, 1.88, 1.91 and 1.94 Å , respectively. This implies that the Mn-O bond strengths increase in the order: α- < δ- < γ- < β-MnO₂. Thus, the Mn-O bond is most easily broken in the reaction of α-MnO₂, which hence showed the best catalytic performance among the tested MnO₂ oxides.

To identify the by-products generated in the off-gases from the toluene degradation process, GC-MS measurements were conducted during the reactions with sole NTP or the combined plasma-catalysts with α-, β-, γ-and δ-MnO₂. The reaction SIE was set at 160 J/L to ensure the conversion of toluene and the generation of sufficient detectable by-products. In each experiment, an adsorption column was used to collect the outlet gas for 30 min, and the gas was then desorbed in a thermal analyzer equipped with a GC-MS for subsequent analyses. As shown in Fig. 12, the main products included carbon dioxide (CO₂), 2-methylpropene (C₄H₈), ethane (C₂H₆), furan (C₄H₄O), water (H₂O), benzene (C₆H₆), methyl formate (HCOOCH₃), acetaldehyde (CH₃CHO), toluene (C₇H₈) and acetone (CH₃COCH₃). Previous studies have identified the by-products formed in the NTP process [12, 52]. In the present study, both the number of species detected and the total amounts of the organic by-products were heavily reduced with the introduction of the α-, β-, γ-and δ-MnO₂catalysts (Fig. 12). The lower levels of organic by-products in the combined plasma-catalytic process implied that the toluene had been deeply oxidized to CO₂and CO in this process.

In the plasma reactor (i.e., NTP alone), the major reaction pathways of toluene degradation were initiated by high-energy electrons, hydroxyl radicals, excited nitrogen species and active oxygen species (i.e., •OH, N, N₂^*, O•, O₂^*, O₂^-). The degradation of toluene proceeded via two processes: direct degradation and secondary oxidation. Direct degradation was induced by the collision of toluene molecules with electrons and by the reaction between toluene and gas-phase radicals, as illustrated in Fig. 13(a). In direct degradation, the collision between toluene molecules and high-energy electrons triggered the excitation of toluene into an activated state, accompanied by opening of the aromatic ring. Then, the reaction proceeded by a series of oxidation steps by O•, •OH, etc., leading to the generation of CO₂and H₂O. In the secondary oxidation process, radicals such as •OH, N, N₂^*, O•, O₂^* and O₂^- were generated in the plasma process and reacted with toluene molecules in the activated state. This resulted in opening of the aromatic ring, followed by a suite of hydroxylation steps, which eventually led to the formation of CO₂and H₂O [1].

In the plasma-catalytic process, the major pathways of toluene degradation were likewise initiated by high-energy electrons and radical species. As shown in Fig. 13(b), the proposed reaction mechanism could be divided into two parts [11]: reactions in the gas phase and reactions at the catalyst surface. The former involved the direct degradation of toluene through collision with electrons followed by oxidation driven by the active free radicals (•OH, N, N₂^*, O•, O₂^*, O₂^-). The latter involved the oxidation of adsorbed toluene and its intermediate by-products (2-methylpropene, ethane, furan, benzene, methyl formate, acetaldehyde and acetone) by active species (such as O• and •OH). The reactions at the catalyst surface depended on the chemisorption of toluene, the Mn-O bond strength of MnO₂, the amounts of chemisorbed oxygen species and the transformations among MnO₂, Mn₂O₃, Mn₃O₄ and MnO. In the plasma-catalytic process, the O₃, O₂ and electrons in the gas stream could be adsorbed and transported to the toluene or intermediate by-products at the catalyst surface via a facile interconversion among Mn⁴⁺, Mn³⁺ and Mn²⁺ states, which then promoted the deep oxidation of toluene to produce greater amounts of CO₂. Furthermore, MnO₂is the most active metal oxide for the decomposition of O₃ [22, 23], thereby producing O₂ and O•, which can also oxidize toluene and intermediate organic by-products.

α-, β-, γ-and δ-MnO₂catalysts were synthesized by a hydrothermal method, and their performance in the plasma-catalytic oxidation of toluene was evaluated. When NTP was used alone, the toluene removal efficiency was 32.5% with the SIE of 160 J/L. Among the investigated catalysts, α-MnO₂showed the best activity, increasing the toluene conversion from 32.5% to 78.1% at the SIE of 160 J/L. For β-MnO₂, γ-MnO₂and δ-MnO₂at the same SIE, their toluene conversion efficiencies were measured at 47.4%, 66.1% and 50.0%. Among the four MnO₂ polymorphs, α-MnO₂possessed a double-tunneled structure, the most stable crystal phase in the plasma and the weakest Mn-O bond strength. α-MnO₂contained a larger content of surface-adsorbed oxygen than β-and δ-MnO₂, but slightly lower than γ-MnO₂. Thus, α-MnO₂achieved by far the most impressive catalytic performance for the degradation of toluene among the four tested polymorphs. In addition, the degradation routes of toluene in the plasma-only and plasma-catalytic processes were investigated and a mechanism was proposed. The introduction of MnO₂ catalysts promoted the adsorption of O₃, O₂, electrons and radical species in the gas stream, which were then transported to molecules of toluene (or its intermediate by-products) at the catalyst surface through facile interconversion among Mn⁴⁺, Mn³⁺and Mn²⁺states. This promoted the deep oxidation of toluene to produce greater amounts of CO₂.