Synergistic photocatalytic activity of a combination of carbon nanotubes-graphene-nickel foam nanocomposites enhanced by dielectric barrier discharge plasma technology for water purification

Degradation activity of plasma catalysis between dielectric barrier discharge (DBD) and carbon nanotubes-graphene-nickel foam (CNTs-G-Nif) has been studied in treatment of dye wastewater. CNTs-G-Nif was prepared through a two-step chemical vapor deposition (CVD) approach. The composite has been characterized by different techniques such as X-ray diffraction (XRD), scanning electron microscope (SEM) and Raman spectroscopy. SEM results showed that the Nif as the growth substrate was evenly wrapped by G and then CNTs were successfully grown on G as the support. The growth mechanism of composite was proposed. The possible coupled catalytic mechanism between DBD and CNTs-G-Nif were addressed. In addition, the modification on G-Nif was found by SEM during the discharge process in liquid phase. And the modification mechanism of DBDP acting on composites was discussed. Finally, by means of analyses of ultraviolet-visible (UV-Vis) spectroscopy, Fourier transform infrared (FT-IR) spectroscopy, gas chromatography-mass spectrometry (GC-MS) and liquid chromatography-mass spectrometry (LC-MS), the general degradation pathway and stepwise degradation pathways of alizarin green (AG) were proposed in detail.


INTRODUCTION
With high chroma, large displacement, high content of electrolyte and complex composition, the printing and dyeing wastewater was considered as the most pernicious industrial wastewater. Therefore, it must be thoroughly mineralized and degraded to eliminate environmental impacts (Wang et al. ; Ben et al. ). AG, a kind of typical anthraquinone dyes was selected as a main research object in this study. The current treatment methods for wastewater pollution mainly included physical methods, biological methods (Farabegoli et al. ), physical and chemical methods (Hayeeye et al. ; Liu et al. ), and advanced oxidation processes (AOPs). AOPs are most suitable for removing persistent and recalcitrant organic compounds. AOPs do not form harmful byproducts and have a short reaction time because of the use of hydroxyl radicals formed by a combination of ozone (O 3 ), hydrogen peroxide, and ultraviolet (UV) (Colindres et al. ), Fenton's reagent (Kumar et al. ), ultrasonic (Shen et al. a, b), electrochemical oxidation (Rubi-Juarez et al. ), photocatalytic oxidation (Liu et al. ; Hou et al. ), plasma process (Jiang et al. ; Magureanu et al. ), combined technologies, etc.
It has been proved that the low temperature plasma is capable of producing a variety of highly oxidizing substances such as hydroxyl radical (·OH), O 3 , hydrogen peroxide (H 2 O 2 ) and superoxide radical (·O 2 À ) (Jiang et al. ). In addition, low temperature plasma also contained UV radiation, shock wave, strong electric field and pyrolysis. Therefore, low temperature plasma could be made use of treating pollutants (Magureanu et al. ), especially for the refractory aromatic organics in printing and dyeing wastewater. The discharge forms of low temperature plasma mainly included pulse discharge, direct current corona discharge, glow discharge, sliding arc discharge and DBD. Among these discharge forms above, low temperature plasma technology based on DBD was easy to operate and able to obtain high treatment effects. In recent years, it has received great attention in the area of water treatment (Wang et  Compared with other AOPs, plasma technology alone showed certain superiority in degradation. However, there were still some block blanks in this field, especially in the reduction of the energy consumption and further enhancing utilization effect of the active substances in plasma system (Chen et al. ). Therefore, the combined systems of plasma have been extensively studied to solve the problems above recently, mainly including the combination with inorganic ions and various kinds of catalytic materials. In recent years, electrode catalytic materials of carbon class (Cc) composite have been studied. A wide range of C materials included activated carbon, CNTs, carbon nanofibers, G and various composite materials. Mousset et al. () used different forms of G as cathode material in electro-Fenton (EF), implying the possibility of using pristine G in EF for the treatment of wastewater. Ji et al. () demonstrated that flow-through electrode of polyaniline supported on lamellar-structured G could be used for electrocatalytic reduction of Cr 6þ . Huang et al. () provided a simple hydrothermal method to synthesize wall-like layered metal oxide MMoO 4 (M ¼ Ni, Co) nanosheet arrays, which are directly grown on flexible carbon cloth and used as a non-removal for advanced supercapacitors. Adhesive electrode. Zhang et al. () have used a simple two-step method to fabricate Ni 1/3 Co 2/3 Se 2 nanowires and nanosheets on carbon cloth. They are interconnected to form a layered three-dimensional network structure, showing high electrochemical HER active. Vecitis et al. () reported that active MWCNT filter was effective toward the adsorptive removal and electrochemical oxidation of the aqueous dyes, Methylene Blue and Methyl Orange, and the oxidation of the aqueous anions, chloride and iodide. Shen et al. () confirmed that G materials could enhance the removal effect of dye, oil, organic solvents, heavy metal ions and gas pollutants. Zhang et al. () reported that the CNTspillared RGO composite materials could exhibit an excellent visible light photocatalytic performance in degrading dye Rhodamine B because of the unique porous structure and the exceptional electron transfer property of G. Based on the studies above, there were reasons to consider that Cc electrode catalytic materials could also be combined with DBDP to treat wastewater on account of its unique discharge structure. Recently, (Sun et al. ) used the solid carbon source at 1,073 K to produce large, high quality and controllable G and doped G. Latorre et al. () used phenomenological model to describe the growth process of CNTs prepared in CVD method. Lv et al. () suggested that CNTs could be combined with G to form three-dimensional (3D) structure. However, the CNTs-G-Ni f was prepared through a two-step CVD approach with methane (CH 4 ) functioned as the carbon source which was distinguished from the ethanol functioned likewise (Zhu et al. ). 2D materials can be water treated through photothermal evaporation, antibacterial and photocatalysis. Therefore, research on this application is also explosive growth (Xie et al. a, b). Due to its large absorption surface, strong absorption in the broad solar spectrum and effective photothermal conversion, two-dimensional nanomaterials have attracted great attention in water production induced by photothermal evaporation. And two-dimensional photothermal materials surpassed graphene in terms of clean water production (Xie et al. a, b). However, since the band gap of the most commonly used 2D materials is not as good as 3D carriers in terms of photocatalysis (Zhang et al. ), there are certain limitations.
As a kind of fixed phase of 3D material, CNTs-G-Ni f should be more suitable for the DBDP reaction system, compared with powder catalysts. Such CNTs-G-Ni f was convenient for subsequent separation and cleaning. At the same time, it could be used as a pseudo electrode material due to its special electrochemical properties. Meanwhile, DBDP could solve the problem of activation on CNTs which should be activated before used. In addition, DBDP also had modification on G and CNTs in wastewater, which was interesting to be studied as well. In summary, a novel degradation system, where CNTs-G-Ni f was combined with DBDP, would be first proposed and investigated in this study. The degradation effects AG by DBDP and DBD/ CNTs-G-Ni f were evaluated with kinetics model, respectively. The growth mechanism of prepared composites, coupled with the catalytic mechanism between DBD and CNTs-G-Ni f , and degradation mechanism of AG were explored one by one.

Preparation
The Ni f (2 cm × 2 cm, thickness 1.0 mm, 99.99% purity) was washed for several times by an ultrasonic cleaner using acetone, ethanol and distilled water, respectively. Then, the Ni f was taken out and dried slowly. A Ni f was placed in quartz boat (20 mm). The quartz boat was pushed into the center of a horizontal quartz tube (100 mm diameter) in a tube furnace (Anhui BEQ equipment technology Co., Ltd China, Figure 1(a)) and heated to 1,273 K under mixtures gas of 30 standard m 2 /min H 2 and 200 standard m 3 /min (sccm) Ar. Then, reaction gas of CH 4 was fed to the system at flow rates of 15 sccm, keeping 30 min at 1,273 K for G growth. Subsequently, the sample was rapidly cooled down to room temperature under H 2 and Ar flow, resulting in high quality G layers grown on both sides of the Ni f . The dip-drawing method was used to coat G with a catalyst solution. The G-Ni f substrate was dipped into 10 mL ethanol/ polyethylene glycol (PEG) (10%, w/w) solution containing 0.1 mol·L À1 Ni(NO 3 ) 2 for about 5 min, and then drawn out and dried in air slowly. Prior to the growth of CNTs, the G-Ni f substrate was heated to 773 K for 10 min under Argon (Ar) flow (170 sccm), followed by H 2 annealing (50 sccm).
Then, the furnace temperature was continued to increase to 1,173 K under Ar and H 2 flow. The CH 4 (30 sccm) mixed with Ar and H 2 was added into the tube furnace at the reaction temperature of 1,173 K for 40 min. Subsequently, the sample was cooled down to room temperature under protection of Ar flow. And the CNTs-G-Ni f could be obtained.

Characterization
The phase and composition of the as-prepared samples were detected by XRD studies using an X-ray diffractometer with Cu Kα radiation under 40 kV and 40 mA with ranging from 10 to 80 at a scanning rate of 10 ·min À1 (D/MAX-IIIA, Japan). Raman spectroscopy studies were carried out on a Dispersive MicroRaman spectrometer (Invia, Renishaw) with an argon-ion laser at an excitation wavelength of 514 nm in the range 500-3,000 cm À1 . The morphology of the prepared samples was examined using a SEM (Hitachi S-4700, Japan).  NaOH (0.1 M) were applied to adjust the pH of solution (pH 2.0-11.0) determined by a pH meter (pHS-25C, Shanghai Yidian Scientific Instrument Co., Ltd China). The conductivity of solution (105-2,000 μS·cm À1 ) was adjusted by KCl and determined using an electrical conductivity meter (DDS-307, Shanghai Yueping Scientific Instrument Co., Ltd China). The corresponding concentrations of AG (50-200 mg·L À1 ) were configured and the corresponding input powers (56-112 W) were adjusted by controlling currents. Coupling system of DBDP/CNTs-G-Ni f were operated with CNTs-G-Ni f 2 cm × 2 cm. The absorbance of sample was determined by at regular intervals during degradation process.

Analysis of samples
The samples were taken from the degradation process of AG under DBDP/CNTs-G-Ni f . The UV-Vis spectrum was determined by using Shimadzu UV-1601PC UV-Vis Scanning Spectrophotometer with the wavelength ranging from 200 to 800 nm. FT-IR spectra of the samples were recorded by Tensor 27, Bruker spectrometer (Germany) over the wavelength range of 4,000-400 cm À1 . The GC-MS analysis was performed to identify the intermediates produced through the AG degradation using a method described in our previous study 8 . Meanwhile, the samples were also analyzed by LC-MS spectrometry (Water Alliance 2795 LC) to assist the identification of degradation products.

Macroscopic feature
It could be seen from Figure 2 that they were Ni f , G-Ni f and CNTs-G-Ni f from left to right in order. The clean Ni f presented silvery white and had metallic luster. After the growth of G on Ni f , the color of composite was changed to be grey and the texture became hardened. After the growth of CNTs on G-Ni f , the color of composite was jet black. Inverting the colors of digital photo, some changes on the surface of Ni f could be found after compositing G and CNTs. Black dots could be found on the surface of G-Ni f . It could be explained that Ni may be separated out from sample during the cooling process. But few black dots could be found on the surface of CNTs-G-Ni f , which illustrated that G-Ni f was completely covered with CNTs.

Morphology and structure analysis
Figure 3 showed images of XRD and Raman in regard to CNTs-G-Ni f . XRD spectrum depicted in Figure 3(a) showed a series of peaks. The peak (002) stood for the existence of element C while the peaks (111), (200) and (220) represented the existence of element Ni. Besides, there were no existences of other elements or oxides. This was consistent with result that the material composition could be seen from SEM. From Figure 3(b), the existences of D-peak and 2D-peak indicated that the material had a typical graphite structure (He et al. ). 2D-peak was relatively weak, indicating the G was multilayer (Zhu et al. ). G-peak was mainly caused by the stretching vibration of C ¼ C (sp 2 ) on the sheet of G. D-peak was referred to as the defect peak, which was caused by the stretching vibration of disordered or sp 3 hybridized C atoms in the carbon material, corresponding to the defect level of carbon materials (Georgakilas et al. ). The literature proved that the integrity of the composite's structure could be maintained during the growing process of G and CNTs on Ni f (He et al. ). Figure 4 showed the images of SEM in different growth stages of composite materials. As illustrated in Figure 4(a), the surface of Ni f was slightly rough, but the whole was relatively smooth within 100 μm. It could be seen clearly that the surface of Ni f substrate was encased by squamous G layers after first step CVD growth from Figure 4(b). The relative datum manifested that there were 2-5 layers G when it was grown based on Ni (Zhang et al. ). It was obvious that there were many wrinkles on G through the illustration of Figure 4(b). This was due to the inconsistent coefficient of thermal expansion between Ni f and G (Dong et al. a, b). It could also be seen that there were some white matters on the surface of G, which may be the Ni separated out from the skeleton of Ni f during the cooling process (Dong et al. a, b). The micro display was exactly echoed with the macro display of colorinvert photos in Figure 2. As shown in Figure 4(c), it could be found that there were thickly-dotted tiny crystals clustered on the G layer randomly. This was because that Ni salt on the surface of G layer was turned into Ni nanoparticles after high temperature with 773 K in the tube furnace, which could be as the subsequent catalyst for growing CNTs (Dong et al. a, b). As shown in Figure 4(d), rough and curved CNTs were randomly grown on the surface of G after the second step of CVD growth. As can be seen from Figure 4, the whole composite maintained the honeycomb-shaped structure after repeated growth. The overall morphology of each stage was similar to that reported by the literatures (Dong et al. a, b; Zhu et al. ). This fully demonstrated that it was feasible to prepare similar products in different carbon source conditions by replacing ethanol with CH 4 .

Effect of operational parameters
In terms of wastewater treatment with DBDP, there were many influence factors. Precisely, four common influence factors would be studied in this study, including concentration of dye, pH of solution, conductivity of solution and input power of DBDP. Figure S1 showed the changes between decolorization rate of AG and values of four factors. The decolorization rate was decreased when values of concentration, pH and conductivity were gradually increased. But the decolorization rate was increased when values of input power were gradually increased. The amount of active material and high energy electron produced in plasma discharge was constant at same time when the input power kept unchanged. When the concentration of AG was higher, the relative decolorization rate was lower. It could be found that acidic condition was of benefit to degradation of AG. The data showed that the oxidation potential of ·OH was 2.74 eV when pH was 3 but the oxidation potential of ·OH was only 2.34 eV when pH was 9. It illustrated that the oxidation ability of ·OH was stronger under acidic condition (Dong et al. a, b). DBD was non-equilibrium discharge and the energy was transferred to other molecules through the collision so that the target could be excited or dissociated. With the increase in conductivity, the process of micro-discharge was hindered and the formation of strong reactive free radicals was slowed down, thus the degradation reaction was reduced. The increase of input power could promote the increase of energy in the reaction, which was conducive to produce more active substances and high energy electrons. Thus, the increase in input power was beneficial to the degradation of AG.

Establishment of the kinetics model equation
The reaction order showed the relationship between the concentration of the remaining reactants and the reaction rate. Through the analysis of the experimental data, the reaction order could be determined. It was demonstrated that the degradation of AG could be regarded as the pseudo-firstorder reaction kinetics in DBDP alone by plotting of ln(c/ c 0 )-t. It can also be obtained from Figure S2 that ln(c/c 0 )-t presented better linear fitting before 60 seconds. As can be seen in Table 1, the R 2 values of ln(c/c 0 )-t which were greater than 0.99, 0.97 and 0.93 accounted for 45, 85 and 100% of the total R 2 values, respectively, which illustrated the degradation process of AG could be described in a first-order kinetic model in 60 seconds. When the reaction time exceeded 60 seconds, the data points started to fluctuate, and then the degradation process was no longer accurate with the firstorder reaction kinetics model. It also reflected that the degradation reaction process became complicated and the large number of intermediates would be produced. Thus, the relationship between concentration of residual AG and reaction rate was affected. The degradation process of AG was an imprecise first-order reaction kinetics model when the reaction time was more than 60 seconds.
The proportionality constant of the apparent chemical reaction rate equation could be called the apparent chemical reaction rate constant. In the DBDP reaction system, the apparent chemical equation was determined as follows: where c is the concentration of target pollutant, k DBDP is the kinetic constant of DBDP system, k obs is the apparent chemical reaction rate constant, and t is time.     () showed that the kinetic constant was affected by the target concentration and ozone consumption when the light intensity was constant. The similar kinetics model equation was used to describe the degradation process by other AOPs except DBDP in our previous research (Shen et al. a, b).
For the study of dye wastewater treatment using plasma, concentration of AG, pH, conductivity, and the input power were selected as the factors. The following empirical model was used to describe the apparent reaction rate constant in this study: where k obs is the apparent reaction rate constant, A refers to the former factor, c 0 is the concentration of AG (mg·L À1 ), [OH À ] is pH value, K con is the solution conductivity (μS·cm À1 ), P is input power (W), and α, β, γ and ε are corresponding mathematical constants. Equation (3) could be got by logarithm of the above equation.
ln k obs ¼ ln A þ α ln c 0 þ β ln [OH À ] þ γ ln K con þ ε ln P (3) Since the voltage in this study was set to 70 V, Equation (3) becomes: Then, linear fitting of the ln k obs served as function of ln c 0 , ln [OH À ], ln K con and ln P.

Application of combined systems
As shown in Figure 5, the kinetic constant of DBDP alone was 0.0179. The kinetic constant of DBDP/CNTs-G-Ni f was namely k (DBDP/CNTs-G-Nif) ¼ 0.0257. The kinetic constant of CNTs-G-Ni f alone k (CNTs-G-Nif) could be approximated as 0 in 150 s. Therefore, cofactor E, which was commonly used to evaluate synergistic effect of the combined system before, would be got through the Equation (8): Data presentation of synergistic effect was that 1 þ 1 > 2. The cofactor E ¼ 1.44> 1 illustrated that there was synergistic effect DBDP/CNTs-G-Ni f . This result was consistent with coupling effect.

Exploration of reaction mechanisms
The preparation mechanism on composite In order to understand the growth of composite at various stages better, a feasible preparation mechanism was proposed by probing into the process of preparation in this study. As Figure 6 illustrated that C atoms would infiltrate into metallic Ni and the whole temporarily formed solid solution state influenced by high temperature when CH 4 and foam nickel were used as the carbon source and the base of growth, respectively. Relevant data study indicated that the solubility of C in Ni was about 0.9% when the reaction temperature reached to 1,173 K (Li et al. ). Because of the cooling effect, the oversaturated C atoms were separated out of Ni surface to form grapheme (Kim et al. ; Guermoune et al. ), namely G-Ni f . The G-Ni f dipped in Ni salt/ethanol/PEG solution was slowly pulled out to form coating state on its surface (Kim et al. ). Then, Ni nanoparticles, the growth catalysts of CNTs, were randomly formed on the surface of G due to the thermal expansion and H 2 etching during the process of heat treatment (Nessim et al. ). On account of subsequent high temperature, the rapid spread of C atoms piled and accumulated from the bottom of nano-Ni growth catalysts to form directional CNTs (Czekaj et al. ; Kim et al. ) which was confirmed as multi-walled CNTs (Dong et al. a, b), namely CNTs-G-Ni f . In addition, to make a vision, the nanocatalysts with smaller size could be loaded on the surface of CNTs to form nanocatalysts-CNTs-G-Ni f in order to further enhance the properties of composite materials.

The coupled catalytic mechanism between DBDP and CNTs-G-Ni f
An explicably coupled catalytic mechanism was expounded in Figure 7 relying on the process of degradation treatment in this study. CNTs-G-Ni f was located in the bottom of water phase and was separated by a layer of quartz medium with low voltage pole of DBDP. The current flowed from high voltage pole would firstly flow through CNTs-G-Ni f then flowed to low voltage pole. Therefore, the CNTs-G-Ni f could be regarded as the pseudo-cathode materials of DBDP, enhancing the stability of discharge and the performance of mass transfer, to improve the effect of plasma catalysis. In addition, the composite CNTs-G-Ni f which presented the honeycomb structure on the whole could target contaminants in a relatively wide range. The CNTs grown under the support of G could   lower the overpotential of the substrate in the reaction of DBDP. At the same time, CNTs as the channel of electron transport could promote electron transfer and enlarge current response (Sun et al. ). Thus, the special plasma catalysis effect could be produced for the mineralization of substances such as dye of AG and its degradation products.
Since CNTs had special one-dimensional hollow tubular structures and high specific surface area (Lv et al. ), the surface of inside and outside layer of CNTs had a lot of active points which could be regarded as host of active material generated by discharge (Sánchez-Polo et al. ; Valdman et al. ) and adsorption points which could be utilized to capture and adsorb organic contaminants in aqueous solution (Liu et al. ).
On account of the effect in the atmosphere of DBDP, dye molecules adsorbed on the surface of CNTs were in an excited state to originate photoinduced electron which would firstly transfer to carbon tube, and then transfer to Ni substrate step by step following energy level. O 2 in the solution or on the surface of composite material would be captured by photo-induced electron, then reduced into ·O 2 À (Yim & Liu ). Thus, the efficiency of electron transfer was accelerated in reduction reaction of O 2 , which could improve the productivity of H 2 O 2 . The positive hole reacted with H 2 O or OH À on the surface of the composite material, leading to generate a large amount of ·OH. Meanwhile, H 2 O around the active points on the CNTs would also more easily influenced by shock waves or strong electric fields of plasma to form ·OH and ·H. The O 2 in the water medium could have the reaction with ·H to form ·O 2 À which would be converted into ·OH and other strong oxidation free radical through chemical reactions (Zhang et al. ; Gong et al. ). In a word, the cyclic degradation model 'adsorption capture/catalytic oxidation' was realized in DBDP/CNTs-G-Ni f .

Degradation mechanism on intermediates and pathway of AG
Analysis of UV-Vis spectrum. Among the wave length of 200-800 nm, there were five main characteristic peaks of AG which were at 254 nm, 284 nm, 412, 606 and 644 nm, respectively. Peak at 254 nm corresponded to the absorption peak of benzene ring. Peaks at 284 and 412 nm corresponded to the absorption peaks of anthraquinone structure. Peaks at 606 and 644 nm meant that the absorption of the conjugate structure was in visible part (Xiong et al. ). It could be seen from Figure 8(a) that the structure of AG was very symmetrical while the conjugate effect made the color of AG green. With time, the absorption peaks gradually decreased in DBDP/CNTs-G-Ni f system, indicating that the molecular structure was constantly opened. When the treatment time was 60 seconds, the absorption peaks of the visible part disappeared completely, indicating that large molecular structures of the solution were almost completely destroyed. When the treatment time reached to 90 seconds, there were weak absorption peaks only in the ultraviolet region, indicating that organic matters with aromatic ring were almost completely oxidized. At 120 seconds, all characteristic peaks disappeared completely, indicating that the decolorization was complete without the presence of the chromophoric groups, and the macromolecule organics with the aromatic ring structures were mineralized into micro-molecule acids and other small molecule products.
Analysis of FT-IR spectrum. As could be seen from Figure 8(b), peaks near 1,500 cm À1 represented the C ¼ C vibration absorption peak of the typical benzene ring skeleton when the sample was in absence of degradation. Usually, there were four peaks, but the four peaks sometimes did not appear at the same time. Peak at 1,029.89 cm À1 was the vibration absorption peak of -SO 3 Na. Peaks at 1,261.02 cm À1 and 1,177.95 cm À1 represented the stretching vibration peak of C ¼ O in anthraquinone. Peak at 3,444.21 cm À1 represented the vibration absorption peak caused by -NH. The peak position of substituent group in benzene ring was at 617.14 cm À1 . After 30 seconds of treatment, C ¼ C vibration absorption peak of benzene ring skeleton turned into 1,633.49 cm À1 , moving into takanami, which meant that original molecules with big conjugate system had become a relatively small system of benzene ring. In addition, the characteristic peak of -SO 3 Na disappeared. The peak of -NH was transferred to 3,440.54 cm À1 . Although the peak type still existed, the vibration amplitude of the peak had decreased. The peak of C ¼ O disappeared, indicating that C ¼ O had been oxidized by active substances. In terms of the dye with anthraquinone structure, the adjacent benzene rings could be oxidized and broken easily once C ¼ O was destroyed (Salazar et al. ). When the degradation time was 60 seconds, most peaks were disappeared and the remained peaks were weakened in some degree, indicating that the large molecular structure was disappeared, and the intermediate products were continuously mineralized to form small molecular products in the DBDP/CNTs-G-Ni f system. This was consistent with the analysis of UV-Vis spectrum at 60 seconds.
Analysis of GC-MS and LC-MS spectrum. According to the analysis results of GC-MS, it could be doped out from Figure S4 that the main intermediate products were phthalic acid, hydroquinone, mesoxalicacid and various small molecular acids like acetic acid during the degradation process of AG by DBDP/CNTs-G-Ni f . From the analysis results of LC-MS, it could be conjectured from Figure S5 that the relative molecular weights of the main intermediates existed in the degradation process of AG by DBDP/ CNTs-G-Ni f were 240, 124, 116 and 108, respectively. Through the analysis, possible intermediates were 1, 4-dihydroxyanthraquinone, 4-methyl catechol, butenedioic acid as well as p-benzoquinone, etc.
Analysis of degradation pathway. The analysis of the spectrums above, including UV-Vis, FT-IR, GC-MS and LC-MS allowed proposing a possible degradation pathway of Figure 9 for AG mineralization by DBDP/CNTs-G-Ni f . Table 2 showed the major identified intermediates. Macromolecule of AG was broken at C-N and 1,4dihydroxyanthraquinone (A 1 ) and 4-methyl catechol (C 2 ) were formed after oxidation and nucleophilic reactions. The ring of A 1 would be opened through the bombardment of high-energy free radicals to form hydroquinone (A 2 ) and phthalic acid (A 3 ). The p-benzoquinone (A 4 ) was turned into maleic acid (A 5 ). Meanwhile, A 1 could be also oxidized to 1,4,9,10-anthraquinone (B 1 ) which was the transition state, and then A 3 and A 5 or catechol (B 4 ) and B 3 were formed by ring open. A 3 , B 4 , and C 2 were translated into a long chain of acids such as hexadiendioic acid and keto acid by further oxidation. Long-chain acids continued to turn into small molecular carboxylic acids such as oxalic acid (C 4 ), acetic acid and formic acid, which were generally reported as the end-products before totally mineralized into CO 2 and H 2 O (He et al. ; Salazar et al. ). Figure 10 explained the general sub-step pathways of degradation. As shown in S 1 , the products of two different states would be formed during the first moments of oxidation when the bonds were broken at site 1 or 2. But -NH 2 could turn into -OH under the function of ·OH (He et al. ) so that the final transformations presented as A 1 and C 1 . As shown in S 2 , the phenol on the benzene ring of C 2 would be further oxidized into the ketone. Then, on the one hand, C 3 and C 4 could be formed through progressive oxidization of benzene ring. On the other hand, the methyl groups on the benzene ring could be replaced to form alcohols. The alcohols were gradually oxidized to form aldehydes and acids (Song et al. ), and then were finally converted to C 3 and C 4 by ring-open reaction.
As shown in S 3 , A 1 was transformed into other intermediates by breaking bonds or further oxidation under the action of active substances. The specific transition processes of intermediate products were illustrated in Figure 11. From S 4 -S 8 , the oxidation transformation paths of the intermediate product in S 3 were explained in detail. In S 4 , double bond was broken, ·H was transferred, and the structure was rearranged to form A 5 . In S 5 , 1,4-naphthoquinone through protonation and transferring of double bond was turned into 1,3-indenadione (Fanchiang & Tseng ; Yuan et al. ). In S 6 , the ketone was decomposed into A 3 after cleavage of ring, and B 3 was formed by the further oxidation until it was mineralized to CO 2 and H 2 O completely. In S 7 , phenol on the benzene ring of B 4 would be further oxidized to ketone and then oxidized to form B 5 by opening ring (Mvula & Sonntag ; Shen et al. , a, b). In S 8 , A 2 could be oxidized to A 4 (Khataee et al. ), and A 4 was finally oxidized to form A 5 due to the breakage of ring (Li et al. ).

CONCLUSIONS
Degradation process of AG could be described in a firstorder kinetic model by DBDP alone within 60 seconds. It was feasible to prepare similar product CNTs-G-Ni f using different carbon sources by replacing ethanol with methane through a two-step CVD approach. The cofactor E ¼ 1.44 > 1 illustrated that there was synergistic effect DBDP/CNTs-G-Ni f . The cyclic degradation model 'adsorption capture/ catalytic oxidation' was realized in DBDP/CNTs-G-Ni f . AG could be mineralized into CO 2 and H 2 O by DBDP/CNTs-G-Ni f . The results above illustrated that such CNTs-G-Ni f composite represented the possibility of a new family of innovative carbon materials for plasma catalysis.