Promotional effect of N-doped CeO2 supported CoOx catalysts with enhanced catalytic activity on NO oxidation

Journal of Molecular Catalysis A: Chemical 398 (2015) 344–352

Contents lists available at ScienceDirect

Journal of Molecular Catalysis A: Chemical journal homepage: www.elsevier.com/locate/molcata

Promotional effect of N-doped CeO2 supported CoOx catalysts with enhanced catalytic activity on NO oxidation Yang Yu a,b , Qin Zhong a,b,∗ , Wei Cai a,b , Jie Ding a,b a b

School of Chemical Engineering, Nanjing University of Science and Technology, Nanjing 210094, PR China Nanjing AIREP Environmental Protection Technology Co., Ltd Nanjing, Jiangsu 210091, PR China

a r t i c l e

i n f o

Article history: Received 23 August 2014 Received in revised form 24 December 2014 Accepted 1 January 2015 Available online 3 January 2015 Keywords: N-doped CeO2 CoOx –CeO2 Selective catalytic oxidation NO

a b s t r a c t A series of neoteric CoOx /N-doped CeO2 catalysts were synthesized by partly substituting the lattice oxygen of CeO2 with nitrogen by a simple g-C3 N4 -modified sol–gel method and comprehensively characterized by XRD, H2 -TPR, XPS, BET, TEM, TG, UV–vis DRS, PL, EIS, NO(O2 )-TPD and EPR. The results demonstrated that: (1) The N-doped catalysts showed larger surface areas and pore volumes, which were favorable for the adsorption of reactant gas; (2) Replacing O with N could promote the reduction of resultant catalysts and assist cobalt oxide in changing the valence and the support in supplying the oxygen; (3) By a sol–gel method, the CoOx crystallites in these catalysts were encapsulated by CeO2 with only a small fraction of Co ions on the surface and strongly interacting with CeO2 . Such structure maximized the interaction between CoOx and CeO2 in three dimensions, resulting in unique redox properties. Moreover, the prepared materials were evaluated in the selective catalytic oxidation of NO. The results showed that the N-doped materials exhibited higher catalytic activity than the un-doped one due to those physicochemical changes. An enhanced mechanism on the improvement of catalytic performance was proposed, and this could pave the way for the designed and synthesis of new highly catalytic activity catalysts. © 2015 Elsevier B.V. All rights reserved.

1. Introduction The emission of nitrogen oxides (NOx ) produced from stationary and mobile combustion sources is one of the major contributors in the atmospheric contamination [1]. Selective catalytic oxidation (SCO) method plays a critical role in those main technologies of NOx removal, such as NOx storage, reduction (NSR) [2–4], continuously regenerating trap (CRT) [5] and selective catalytic reduction (SCR) [6–8]. The NO/NO2 feed ratio to the SCR catalyst has an important effect on the overall conversion of NOx , and conditions that lead to “fast” SCR have been reported for reaction mixtures with an equimolar NO/NO2 ratio as opposed to those with only NO or NO2 [9]. Typical NOx emissions from combustion processes contain significantly more NO than NO2 , and the inclusion of an NO oxidation step prior to the SCR process is useful for increasing SCR rates of reaction [10]. In order to enhance the activity of the catalyst, many efforts have been paid to modify the material, for example, by anion or cation

∗ Corresponding author. Tel.: +86 25 84315517; fax: +86 25 84315517. E-mail address: [email protected] (Q. Zhong). http://dx.doi.org/10.1016/j.molcata.2015.01.002 1381-1169/© 2015 Elsevier B.V. All rights reserved.

doping. Doping involves typically substitution of anions (e.g., N, F and S) at O sites and transition metal cations (e.g., Fe, V and Mo) at Ti sites [11–13]. The effect of ion doping on the activity of catalyst strongly depends on many factors such as the dopant concentration, the distribution of the dopant, the configuration of doping ions and so on. Chen et al. prepared Ce-doped V2 O5 –WO3 /TiO2 catalysts for the NH3 –SCR reaction and found the increase of chemisorbed oxygen on the surface could be caused by the Ce addition [14]. Shen et al. reported that Fe-doped CeO2 could increase the oxygen vacancy concentration on its surface and proved that oxygen vacancy was an important role to influence the catalytic performance [15]. Su et al. investigated the effect of the enhancement of halogen ions-doping on the catalytic activities of Al2 O3 , ZrO2 , and TiO2 catalysts in SCR [16]. Yu et al. [17] reported that S-doped nanocrystalline titania exhibited a strong visible-lightinduced antibacterial effect, which was related to the formation of active oxygen species. Wei et al. [18] found N–S-codoped TiO2 had high activity for decompositions of methyl orange, due to the creation of surface oxygen vacancies and the enhancement of surface acidity by N and S codoping. CeO2 has a variety of applications in solid oxide fuel cells, oxygen storage capacitors and fast ion conductors owing to its unique crystal structure and redox properties [19].

Y. Yu et al. / Journal of Molecular Catalysis A: Chemical 398 (2015) 344–352

From after-mentioned reports, it could be seen that the effects by Ndoping were also beneficial to improve the catalytic activity of the catalyst. Following on from TiO2 , nitrogen doping of ceria should be of interest, but there have been only a few reports along these lines. Nitrogen doping of ceria holds great potential in enhancing the performance of this material as catalyst. Incorporation of non-metal elements into CeO2 is still a great challenge. The major difficulty is the lack of a reliable method to incorporate the appropriate amount of nitrogen into the metal oxide matrix. Jorge et al. have doped nitrogen into ceria powder by sintering CeO2 in NH3 flow at very high temperature, but further studies were not reported [20,21]. However, incorporation of non-metal elements into CeO2 by this method was complicated and time-consuming. Therefore, a simple and convenient method to prepare the N-doped CeO2 was urgent need. As the most stable phase of carbon nitride compounds under ambient conditions, polymeric graphitic carbon nitride (g-C3 N4 ) has received much attention as a promising “metal-free” photocatalyst [22]. In this study, the g-C3 N4 was used as the nitrogen sources to synthesize the N-doped CeO2 by a facile sol–gel method for the first time. The as-prepared N-doped CeO2 supported CoOx catalysts showed better catalytic activity than the un-doped one. The physicochemical properties of the catalysts were presented to understand the effect of adding N on the catalytic performance of the catalysts in detail. This paper proposed a new strategy to improve the activity of catalysts used in NO oxidation. 2. Experimental 2.1. Preparation of the catalysts All starting reagents were purchased from Sinopharm Chemical Reagent Company (Shanghai, China) and were used without further purification. Deionized water was used throughout. g-C3 N4 power was synthesized via heating melamine in a tube furnace. A certain amount of melamine was put into an alumina crucible which was first heated in air at 500 ◦ C for 2 h and was further heated in air at 520 ◦ C for 2 h with a temperature rise rate of 4 ◦ C min−1 . The product was collected and ground into powder. CoOx /N-doped CeO2 were synthesized by a citrate sol–gel method. Take CC-N15 as an example, first of all, 3.689 g Ce(NO3 )3 ·6H2 O, 0.4365 g Co(NO3 )2 ·6H2 O and 3.150 g citric acid were dissolved into deionized water under stirring for 0.5 h, then 0.15 g of the as-prepared g-C3 N4 were added into the abovementioned solution to continue stirring for another 1 h. After that the mixture was heated at 90 ◦ C under stirring until it became a viscous gel and dried at 120 ◦ C for 12 h. The obtained solid was calcined in air at 500 ◦ C for 4 h with a heating rate of 4 ◦ C min−1 . The catalyst was denoted as CC-N15, CC-N25 and CC-N35 were referred to the weight of g-C3 N4 was 0.25 and 0.35 g, respectively. 2.2. Samples characterization The powder XRD patterns were recorded on a Beijing Purkinjie general instrument XD-3 X-ray diffraction using Cu-K␣ radiation at 36 kV and 20 mA (2 from 5◦ to 80◦ ). The scanning speed is 8◦ min−1 and the step value is 0.04◦ . Specific surface areas of the different catalysts were determined by N2 adsorption–desorption measurements at −196 ◦ C by employing the Brunauer–Emmet–Teller (BET) method (Gold App V-sorb 2800p), and the pore volume and pore size of the samples were calculated by Barrett–Joyner–Halenda (BJH) method. The systematic error calculated for BET surface area, pore volume and pore size was given with the accuracy at ±1 m2 g−1 , ±1 mm3 g−1 and ±0.1 nm. Prior to N2 adsorption, the sample was outgassed at 200 ◦ C

345

for 12 h to desorb moisture adsorbed on the surface and inside the porous network. X-ray photoelectron spectra (XPS) was performed on a Thermo ESCALAB 250 (USA) apparatus with Al K␣ X-rays (hv = 1486.6 eV) radiation operated at 150 W to investigate the surface atomic concentrations and the oxidation state distribution of the elements in the samples. The samples were compensate for charging with low-energy electron beam, and the peak of C 1s (Binding Energy = 284.4 eV) was used to correct for sample charging. This reference gave BE values with an accuracy ±0.1 eV. And the atomic surface ratios of the corresponding species were given with the accuracy at ±0.1%. The penetration depth of the XPS probe is 10 nm. The micromorphology of the catalysts was examined on a JEOL JEM-2100 transmission electron microscope (TEM), and the sample was deposited on a copper mesh by means of dipcoating. The acceleration voltage is 200 kV. IR or FTIR measurements were performed on a Nicolet-iZ10 (iS10) spectrometer with a resolution of 4 cm−1 with 32 co-added scans from 4000 to 650 (400) cm−1 in transmission mode at room temperature. Diffuse reflectance spectroscopy (DRS) was carried out on a Shimadzu UV-2550 UV–vis spectrophotometer. BaSO4 was the reference sample and the spectra were recorded in the range of 200–800 nm. The thermogravimetric (TG) analysis of samples were performed from 50 to 800 ◦ C with a Netzsch STA 449C Jupiter instrument under the flow of N2 at a heating rate of 20 ◦ C min−1 . Electron paramagnetic resonance (EPR) spectra were carried out on a Bruker EMX-10/12 X-band (∼9.7 GHz) spectrometer at room temperature. Temperature-programmed desorption (TPD) was carried out on automated chemisorption analyzer (Quantachrome Instruments). About 200 mg sample was used. After NO (or O2 ) saturation in 1 h, the gas was switched to He for 0.5 h. Subsequently, TPD was performed by ramping the temperature at 10 ◦ C min−1 to 700 (or 800) ◦ C in He (70 ml min−1 ). Desorption of NO (or O2 ) was detected by a thermal conductivity detector (TCD). Hydrogen temperature programmed reduction (H2 -TPR) was performed in a quartz U-tube reactor on an automated chemisorption analyzer (Quantachrome Instruments) by the GC method. About 100 mg sample was pretreated in N2 stream at 600 ◦ C for 0.5 h. As the sample was cooled downed to 50 ◦ C, switched N2 to H2 –N2 mixture gas (10% H2 , v/v) at a flow rate of 70 ml min−1 . H2 TPR was performed by heating the sample from 50 to 700 ◦ C, at the same time, the consumption of H2 was detected by a thermal conductivity detector (TCD). EIS measurements were performed on an electrochemical workstation (CHI 660B Chenhua Instrument Company, Shanghai, China) based on a conventional three-electrode system comprised of glassy carbon electrode as the working electrode, platinum wire as the counter electrode, and Ag/AgCl (saturated KCl solution) as the reference electrode. The EIS was performed in 0.05 M PBS solution with a frequency range from 0.01 Hz to 100 kHz at 0.308 V, and the amplitude of the applied sine wave potential in each case was 5 mV. 2.3. Catalytic testing The catalytic oxidation of NO was performed in a fixed-bed flow microreactor under atmospheric pressure. Typically, 300 mg sample (sieve fraction of 40–60 mesh) was placed in a quartz reactor (6.8 mm i.d.); the reactant gas mixture (390 ppm NO, 8% O2 , N2 balance) was fed to the reactor with a total flow rate of 100 mL min−1 , corresponding to a gas hour space velocity (GHSV) of 35400 h−1 . The steady-state tests were conducted isothermally every 25 ◦ C from 200 to 400 ◦ C and the gas products (after 90 min reaction)

346

Y. Yu et al. / Journal of Molecular Catalysis A: Chemical 398 (2015) 344–352

were analyzed by a Ecom-JZKN flue gas analyzer (Germany). The NO conversion is defined as: NO

conversion

to

NO2 =

(N O2i n − NO2out ) × 100% N Oin

(1)

3. Result and discussion 3.1. characterization of catalysts 3.1.1. XRD analysis Fig. 1 shows the XRD patterns of g-C3 N4 and CC-Nx. The XRD pattern of the synthesized g-C3 N4 showed two broad peaks at 1 3.1 and 2 7.3◦ , which could be indexed as (0 0 2) and (1 0 0) diffraction planes [23]. For CC-Nx samples, all catalysts presented the reflection peaks at 2 8.5, 3 3.1, 4 7.5 and 5 6.3◦ , corresponding to the (1 1 1), (2 0 0), (2 2 0) and (3 1 1) plane of the fluorite structure of CeO2 , without additional phases such as CeN [24]. Moreover, among the CC-Nx composites, no cobalt oxides diffraction peaks were detected, suggesting a better dispersion of Co species or Co species were doped into the lattice of CeO2 . Furthermore, no g-C3 N4 phases were detected. The same consequence was also observed by Han et al. [23] and Katsumata et al. [24]. The FTIR spectra of gC3 N4 and CC-Nx are showed in Fig. S1. For the pure g-C3 N4 , three main absorption regions can be observed clearly. The broad peak at 3000–3500 cm−1 was ascribed to the stretching vibration of N H and the stretching vibration of O H of the physically adsorbed water [25–27]. The strong band of 1200–1700 cm−1 , with the characteristic peaks at 1 2 4 0, 1 3 2 0, 1 4 0 7, 1 5 6 7 and 1 6 4 0 cm−1 , was attributed to the typical stretching vibration of CN heterocycles [25]. In addition, the peak at 8 0 8 cm−1 could correspond to the breathing mode of triazine units [27]. It could be found that the CC-Nx composites showed no FTIR absorption of these peaks, indicating the structure of g-C3 N4 was destroyed by high temperature calcination with the Co and Ce precursors. 3.1.2. Redox properties (H2 -TPR) The reducibility of the samples were investigated using H2 -TPR measurements. As showed in Fig. 2, four well-defined reduction regions were observed at low temperature region (peak ␣), medium temperature region (peak ␤ and ␥) and high temperature region (peak ␴). At the low-temperature range below 250 ◦ C, peak ␣ was attributed to the reduction of the adsorbed oxygen, peak ␤ and ␥ at medium-temperature were attributed to the reduction of Co3+ to Co2+ and Co2+ to Co, respectively [28]. The peak ␴ at high-

Fig. 2. H2 -TPR profiles of CC-Nx.

temperature region could be attributed to the reduction of surface Ce4+ [29,30]. For the N-doped samples, the reduction peaks of ␤, ␥ and ␴ shifted to lower temperature. This shift indicated that the N-doped samples could promote the reducibility of neighboring Co and Ce ions and possess better reducibility or oxygen mobility. 3.1.3. X-ray photoelectron spectroscopy (XPS) results XPS was performed in order to further illuminate the surface composition and the chemical state of the elements existing in the catalysts. The deconvoluted photoelectron spectra of the Ce 3d, Co 2p, O 1s and N 1s are shown in Fig. 3. Ce 3d core level spectra (Fig. 3(a)) for all the Ce-contained samples were deconvoluted into eight contributions. The spin–orbit splitting of Ce 3d5/2 and Ce 3d3/2 was 18.5 eV for all the samples, which was in good agreement with literature [31]. The u’” satellite peak at about 9 1 6.5 eV was the fingerprint of Ce4+ state, and its high intensity and area suggested that the main part of the ceria was in Ce4+ oxidation state. After deconvolution the appearance of bands labeled u’ and v’ were typical for Ce3+ ions, which suggested that both oxidation states Ce4+ and Ce3+ coexist in the surfaces of ceria-containing samples [31]. It is well known that the interpretation of the core level Ce 3d spectra is not straightforward due to the hybridization between the Ce 4f levels and the O 2p states [32]. The Co 2p spectra are displayed in Fig. 3(b). The Co 2p3/2 peak position was in accordance with the presence of Co3 O4 , while the weak shakeup satellite at around 786 eV from the main spin–orbit component was attributed to the surface hydroxyl species (i.e., Co OH) [33]. The 2p3/2 binding energy of Co2+ was close to that of Co3+ , while the two oxidation states of cobalt could be distinguished by a distinct shake up satellite of Co2+ [34]. The surface atomic ratio of (Co3+ )xps to (Co3+ + Co2+ )xps was complied in Table 1. It could be seen that the atomic ratio of Co3+ /(Co3+ + Co2+ ) increased by N-doping. Different valence state of cobalt in the catalyst was apt to promote the oxidation–reduction reaction.

Table 1 Surface characterization results from XPS. Sample

Fig. 1. XRD patterns of g-C3 N4 and catalysts with various nitrogen dopings.

CC-N0 CC-N15 CC-N25 CC-N35

Co3+ /(Co3+ + Co2+ )

Co/(Co + Ce) Nominal

XPS

0.15 0.15 0.15 0.15

0.11 0.098 0.10 0.099

34.4 77.2 59.4 63.8

Y. Yu et al. / Journal of Molecular Catalysis A: Chemical 398 (2015) 344–352

347

Fig. 3. XPS spectra of (Ce 3d, Co 2p, O 1s) of CC-Nx catalysts and N 1 s spectra of CC-N25.

Table 1 lists the surface Co/(Co + Ce) atomic ratios derived from XPS analysis. The results shows the surface Co/(Co + Ce) ratios were about 2/3 of the nominal one, indicative of a surface Ce enrichment and Co deficiency. While for the Co3 O4 /CeO2 catalysts prepared by co-precipitation, cobalt enrichment on the surface was observed by Liotta et al. [35]. Hence, the preparation process may have played an important role for surface Ce enrichment. During the previous preparation of Co3 O4 –CeO2 catalysts, it was found that the

surfactant-template method can generate precursors with a CeO2 covered encapsulation structure [36]. The encapsulation structure could occur in high temperature calcination process [37]. Similarly, for Pt/CeO2 prepared by modified microemulsion using CTAB as a cationic surfactant, a total encapsulation structure has also been proposed [38]. However, different from the totally encapsulated Pt/CeO2 catalysts, a small fraction of XPS-detectable Co atoms still existed on the surface for the catalysts studied here which could

Fig. 4. N2 adsorption–desorption isotherms (a) and pore size distributions (b) of CC-Nx.

348

Y. Yu et al. / Journal of Molecular Catalysis A: Chemical 398 (2015) 344–352

Fig. 5. TEM and HRTEM imagines of CC-Nx. (A) CC-N0; (B) CC-N15; (C) CC-N25; (D) CC-N35.

be ascribed to the surface migration of Co ions during calcination. It was reported that the calcination process could drive the cobalt from bulk to surface [39]. In this work, since the calcination temperature (500 ◦ C) was not high enough, no bulk Co species were formed on the surface (confirmed by XRD), and only a small fraction of Co existed as surface species, which was believed to be entirely interacting with CeO2 . The enrichment of ceria species were favorable

for the activation of O2 , which was a rate-determining process in oxidation reaction [40]. Hence such encapsulation structure maximized the interaction between the cobalt oxide and ceria in three dimensions and provided sites for O2 activation. O 1s spectra were mainly composed of two components. The peak of O 1s named as O␤, with a binding energy of 528–530 eV, contributed to the lattice oxygen in the metal oxides. The high bind-

Y. Yu et al. / Journal of Molecular Catalysis A: Chemical 398 (2015) 344–352

ing energy (531–532 eV) was attributed to the surface chemisorbed oxygen (O␣). Fig. 3(c) shows the O 1s spectra of CC-Nx samples, it could be seen that the binding energy position of crystal lattice oxygen in all the N-doped samples was lower than CC-N0 (529.2 eV). The reason was the lower electronegativity of nitrogen compared with oxygen, which could result in the increased outer electron density of O and consequently decreased the binding energy. This phenomenon indicated that the N replace the position of O, forming the N-doped CeO2 . Moreover, the nitrogen 1 s core level of CC-N25 is displayed in Fig. 3(d). It revealed weak and diffuse N 1s peaks due to its low N content. Consequently, it was difficult to reliably deconvolute the nitrogen peak. However, we observed a peak at about 3 9 9 eV, which provided experimental evidence that nitrogen incorporated into CeO2 . In general, the nitrogen 1s core level from CeN should appear at 396.2 eV and molecularly chemisorbed nitrogen or NO2 type species appear above 400 eV [21]. Nitrogen doping into the CeO2 lattice reduced the electron density around the nitrogen compared with the highly electronegative oxygen. Consequently, it was reasonable that the binding energy of nitrogen in nitrogen-doped CeO2 was higher than that in CeN. The N 1s spectra of CC-N35 was also given in Fig. S2. The results were consistent with the analysis of O 1s and further confirmed the forming of N-doped CeO2 . Taken together, the H2 -TPR and XPS results indicated the incorporation of N into the CeO2 lattice and the tight interaction between cobalt and ceria. In order to investigate the effect of N-doping on the CoOx /CeO2 catalysts, many methods were used to analyze the physical and chemical properties of the catalysts. 3.1.4. BET analysis The N2 adsorption–desorption isothermal plots and the corresponding BJH pore size distribution curves of these samples were displayed in Fig. 4. It could be seen from Fig. 4(a) that all these samples were of classical type IV as defined by IUPAC [41] and exhibited hysteresis loops of type H3, which indicated that these samples contained mesopores (2–50 nm) with narrow slit-like shapes [41]. Furthermore, the corresponding pore size distributions of these samples determined by the BJH method from adsorption branch of the corresponding isotherms exhibited narrow peak centered at about 2–19 nm, indicating that these samples possessed uniform mesopore size distribution (Fig. 4(b)). The BET surface area, pore volume and pore size of these catalysts are summarized in Table 2. From Table 2, it could be seen that the surface areas of the N-doped samples were much larger than that of un-doped one, which could be favorable for the adsorp-

349

Table 2 The surface area and pore structure parameters of CC-Nx. Sample

CC-N0

CC-N15

CC-CN25

CC-CN35

BET (m2 g-1 ) Pore size (nm) Pore volume (mm3 g-1 )

21 6.9 66

33 18.6 203

44 13.9 189

45 16.2 238

tion of reactant gas. The surface area apparently increased as the increase of N content from 0 to 25 and remained almost constant when the content continued increasing to 35. It could be deduced that the larger surface area was due to the decomposition of laminar g-C3 N4 . In addition, both the pore size and pore volume increased, which was also beneficial for the adsorption of the reactant gas and further improved the catalytic performance. The results were confirmed by NO-TPD as the NO adsorption amounts of these Ndoped samples were much larger than that of blank one. The TG profiles of CC-N0 and CC-N25 are displayed in Fig. S3, the CC-N0 sample showed a significant weight loss after 500 ◦ C, which could be attributed to the sintering effect of CeO2 at high temperature, in contrast, the CC-N25 showed excellent thermal stability, indicating the N-doping could prevent the CeO2 from sintering. The IR and BET surface area in the Fig. S4 and Table S1 also demonstrated our after-mentioned results. 3.1.5. TEM analysis TEM was used to investigate the morphology and microstructure of the sample. Fig. 5 shows the TEM and HR-TEM imagines of CC-Nx. It was obvious that all the samples were agglomerated seriously due to the small particle size, which was about 12 nm in CC-N0 or CC-N35, and about 28 nm in others. The results were consistent with XRD analysis. Furthermore, the absence of Co-related particles’ lattice in HRTEM images suggested a better dispersion of CoOx crystallites or Co species were doped into the CeO2 matrix, which was also confirmed by XRD results. The HRTEM showed the fringes of all the catalysts were observed as about 0.312 nm, which matched the (1 1 1) crystallographic plane of CeO2 [42]. The results indicated that the catalysts tended to grow along the (1 1 1) plane preferentially, due to the stability of the crystallographic plane [43]. 3.1.6. UV–vis diffuse reflectance measurements analysis The information on the surface of electronic states can be obtained from UV–vis diffuse reflectance measurements. As showed in Fig. 6(a), all the samples showed one broad peak from ca. 200 to 350 nm. Three absorption maximum centered at ca. 255, 285 and 340 nm were observed in CC-Nx. The latter two absorp-

Fig. 6. UV–vis DRS spectra (a) and the energy of band gap (b) of CC-Nx catalysts with different nitrogen dopings.

350

Y. Yu et al. / Journal of Molecular Catalysis A: Chemical 398 (2015) 344–352

Fig. 7. PL spectra (a) and EIS profiles (b) of CC-N0 and CC-N25.

tion maximum were ascribed to O2− → Ce4+ charge transfer and interband transitions, respectively. The poorly resolved former maximum corresponding O2− → Ce3+ [44,45]. The absorption band observed at ca. 420 nm can be attributed to the overlapped d–d transitions of Co in its octahedral sites [46]. The peak observed at ca. 700 nm corresponded to Co2+ ions in its tetrahedral geometry [46]. It was observed that the optical absorption edges of nitrogen-doped CeO2 samples shifted slightly toward to longer wavelength (red shift) and the absorbance in the visible light region strengthened compare with that of un-doped CeO2 . It was previously reported that the nitrogen doping can lead to a mixing of N 2p orbital with O 2p orbital to form intermediate energy levels and shift the absorption edge toward visible light region [47]. The energy band gaps of these catalysts are shown in the insert of Fig. 6(b). The bandgap energy of CC-Nx (x = 0, 15, 25, 35) were 1.75, 1.73, 1.61 and 1.66 eV. Obviously, the band gaps became small in comparison to CC-N0, which indicated the N-doped samples could reduce the Eg . The decrease of the band gap was beneficial to boost the generation of electrons and holes. These results confirmed that N-doping was effective for narrowing the band gap, due to the fact that N-doping could form a narrow band above the valence band and produce more oxygen vacancies on the surface of the catalysts [47,48]. Generally, the optical absorption properties of the catalysts became stronger with increasing amount of oxygen vacancies [56]. Thus the amount of O2 − increased with N-doping, which was also detected in EPR patterns (Fig. S5). The vacancies and active oxygen

over the catalyst surface played an important role in the process of NO oxidation [49]. In other words, the interaction between Co and Ce became stronger over the nitrogen-adding sample.

3.1.7. PL and EIS analysis The oxygen vacancies of catalysts can be determined by using PL spectra. According to the reports, when the content of surface oxygen vacancies increased to a high level, the diversity of the surface state might slow the radiative recombination process of generating electrons and holes, and thus decreased the intensity of the PL spectrum [50–52]. Fig. 7(a) shows the PL spectra of CC-N0 and CC-N25, the PL intensity decreased obviously with the N-doped sample. Therefore, it indicated that the number of surface oxygen vacancies increased. Many studies have revealed that nitro and nitrate groups are the intermediates in the process of NO oxidation [49]. Oxygen vacancies could absorb O2 in gas phase to form superoxide ions [56], which was beneficial for prompting the catalytic activity. EIS measurement was employed to investigate the charge transfer resistance and the separation efficiency between the electrons and holes. From Fig. 7(b), it could be seen that the diameter of the Nyquist circle of CC-N25 composite was smaller than that of CCN0. It indicated that composite CC-N25 had lower resistance than that of CC-N0. The result demonstrated that the introduction of N into CeO2 could enhance the separation and transfer efficiency of electron–hole pairs, which was consistent with the PL results [22].

Fig. 8. NO-TPD (a) and O2 -TPD (b) of CC-Nx.

Y. Yu et al. / Journal of Molecular Catalysis A: Chemical 398 (2015) 344–352

3.1.8. Adsorption properties (NO-TPD, O2 -TPD) TPD experiments were conducted to test the reversibility of NO (or O2 ) adsorption. The TPD profiles of NO on the CC-Nx catalysts are shown in Fig. 8(a). As the temperature increased from 50 to 700 ◦ C, NO desorption was observed on these samples. All the samples showed three main temperature ranges (␣, ␤ and ␥), suggesting that NO adsorbed on three different sites [53]. NO desorption was completed over all the samples at below 700 ◦ C. The low temperature peak ␣ was ascribed to the desorption of NO from weakly bonded sites and the peak ␤ and ␥ in high temperature range was due to the strongly adsorbed species, i.e., nitrates [54]. The NO desorption amount was consistent with NO capacities of these sorbents. The weakly adsorbed NO (peak ␣) increased with BET surface area, indicating the larger surface areas were favorable for NO adsorption and desorption, thus improving the catalytic activities. In addition, the adsorbed NO was easy to be desorbed in the temperature range from 50 ◦ C to 250 ◦ C, which indicated that the adsorption NO was weak. Generally, there are three kinds of active oxygen species, ␴ O species had a low desorption temperature (<350 ◦ C), which could be attributed to molecular O2 , O2 − and O2 2 adsorbed on oxygen vacancies [55]; the high desorption temperature (>750 ◦ C) of O species (not shown) was assigned to the oxygen in crystal lattice. The desorption temperature of ␦ O species was between these two desorption temperatures, which was related with oxygen defect and was considered as partial crystal oxygen [56]. The O2 -TPD profiles are shown in Fig. 8(b). For all catalysts, desorption peak was observed in the temperature range of 50–800 ◦ C. Obviously, all the catalysts showed two desorption peaks of ␴ and ␦, the CC-N25 showed the highest adsorption capacity of ␴ and ␦. Since the O-substitution with N produced oxygen vacancies in the fluorite-type lattice, the increased low temperature desorption was associated with the oxygen vacancy [55]. O2 in gas phase could be absorbed by oxygen vacancies to form O2 − , which was beneficial for prompting the catalytic activity. In order to further confirm the above conclusion, EPR test of CC-N0 and CC-N25 were used to detect the O2 − , showed in Fig. S5. Based on the literature EPR data [57], the g-value (g = 2.0074) was characteristic of paramagnetic materials containing superoxide ions. It could be seen in Fig. S5 that CC-N0 has almost no signal of superoxide ions, whereas the EPR intensity of g = 2.0074 significantly increased after N doping. It indicated that N doping could improve the formation of superoxide ions and further enhance the catalytical performance.

4. Catalytic activity of NO oxidation In order to investigate the effect of N-doping on the activities of the CoOx /CeO2 catalysts, the performance of the catalysts with different amounts of nitrogen doping contents at a temperature range of 200–400 ◦ C was tested (Fig. 9). For all catalysts, NO conversion increased with increasing temperature before 325 ◦ C and began to decrease after that. Under identical operating conditions, CC-N0 shows the lowest activity of only 28.2% NO conversion at 325 ◦ C. After the N element was introduced, the catalytic activities enhanced significantly over the whole range of temperature investigated. The highest NO conversion (53.2%) was obtained at 325 ◦ C. However, further doping N into the CoOx /CeO2 decreased the activity. The results suggested that an appropriate amount of N was conductive to improve the activity of the CoOx /CeO2 catalyst. Based on the results of catalyst characterization and activity test, the reasons for the enhancement in the catalytic activity of CoOx /Ndoped CeO2 catalysts could be explained as follows: The study indicated that N-doping could enhance the BET surface area and pore volume, which facilitated the dispersion of active component and the spread of reactant molecules. Meanwhile, the

351

Fig. 9. Catalytic performance of CC-Nx. (Reaction conditions: 390 ppm NO, 8% O2 , N2 as balance gas, GHSV = 35400 h−1 .)

addition of N could enhance the redox activity and improve the adsorption ability of reactant gas. PL, EIS and UV–vis identified that the N-doping could inhibit the recombination of electrons and holes, resulting in the increase of oxygen vacancies. Using the sol–gel method, CoOx crystallites were considered to be encapsulated by CeO2 , with only a small fraction of Co ions exposing on the surface and strongly interacting with CeO2 . Such a structure maximizes the interaction between CoOx and CeO2 in three dimensions, as confirmed by XPS and H2 -TPR. What is more, more active oxygen (O2 − ) was created on the surface of CoOx /N-doped CeO2 catalysts, which was confirmed by EPR experiments. Active oxygen (O2 − ) was an important intermediate species in the process of NO oxidation. In summary, these physicochemical changes enhanced the catalytic performance of N-doped catalysts. 5. Conclusions In the present work, CoOx /N-doped CeO2 catalysts were prepared by the g-C3 N4 -decorated sol–gel method to apply in the oxidation of NO. Furthermore, CoOx /CeO2 catalyst was prepared for comparison. Especially, the structure, reduction, dispersion, surface composition, adsorption and catalytic performance of these synthesized catalysts have been investigated systematically. Combining with the above-mentioned characterization results, several conclusions can be drawn as follows: (1) CoOx /N-doped CeO2 catalysts showed better catalytic performance than CoOx /CeO2 , which may be related to its excellent reduction property and suitable adsorption/desorption behavior. (2) By the sol–gel method, more ceria were exposed on the surface, this kind of structure was favorable for the O2 activation. (3) There was a strong interaction between cobalt and the support via charge transfer, the addition of nitrogen could produce oxygen vacancies and enhance the activity. Acknowledgments This work was financially supported by the Assembly Foundation of The Industry and Information Ministry of the People’s Republic of China 2012 (543), the National Natural Science Foundation of China (U1162119), Science and Technology Support Program of Jiangsu Province (BE2014713), Natural Science Foundation of Jiangsu Province (BK20140777), Scientific Research Project of Environmental Protection Department of Jiangsu Province

352

Y. Yu et al. / Journal of Molecular Catalysis A: Chemical 398 (2015) 344–352

(2013003), Research Fund for the Doctoral Program of Higher Education of China (20113219110009), Industry-Academia Cooperation Innovation Fund Projects of Jiangsu Province (BY2012025) and (BY2014004-10), Science and technology project of Nanjing (201306012). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.molcata. 2015.01.002. References [1] W. Klose, S. Rincón, Fuel 86 (2007) 203–209. [2] C.M.L. Scholz, V.R. Gangwal, M.H.J.M. de Croon, J.C. Schouten, Appl. Catal. B: Environ. 71 (2007) 143–150. [3] N. Takahashi, K. Yamazaki, H. Sobukawa, H. Shinjoh, Appl. Catal. B: Environ. 70 (2007) 198–204. [4] Z.Q. Liu, W.S. Epling, J.A. Anderson, J. Phys. Chem. C 115 (2011) 952–960. [5] Z.H. Liu, Y.S. Ge, J.W. Tan, C. He, J. Environ. Sci. 24 (2012) 624–631. [6] S.L. Zhang, H.Y. Li, Q. Zhong, Appl. Catal. A: Gen. 435–436 (2012) 156–162. [7] Q. Zhong, T.J. Zhang, Y.T. Li, W.H. Ma, H.X. Qu, Chem. Eng. J. 174 (2011) 390–395. [8] S.L. Zhang, X.X. Liu, Q. Zhong, Y. Yao, Catal. Commun. 25 (2012) 7–11. [9] I. Malpartida, O. Marie, P. Bazin, M. Daturi, X. Jeandel, Appl. Catal. B: Environ. 113–114 (2012) 52–60. [10] M.F. Irfan, J.H. Goo, S.D. Kim, Appl. Catal. B: Environ. 78 (2008) 267–274. [11] G.S. Qi, R.T. Yang, Appl. Catal. B: Environ. 44 (2003) 217–225. [12] K.S. Yang, Y. Dai, B.B. Huang, J. Phys. Chem. C 111 (2007) 18985–18994. [13] W.Y. Choi, A. Termin, M.R. Hoffmann, J. Phys. Chem. 98 (1994) 13669–13679. [14] L. Chen, J.H. Li, M.F. Ge, J. Phys. Chem. C 113 (2009) 21177–21184. [15] Q. Shen, G.Z. Lu, C.H. Du, Y. Guo, Y.Q. Wang, Y.L. Guo, X.Q. Gong, Chem. Eng. J. 218 (2013) 164–172. [16] C.L. Su, F.Y. Notoya, E.J. Sasaoka, Ind. Eng. Chem. Res. 42 (2003) 5770–5774. [17] J.C. Yu, W.K. Ho, J.G. Yu, H.Y. Yip, P.K. Wong, J.C. Zhao, Environ. Sci. Technol. 39 (2005) 1175–1179. [18] F.Y. Wei, L.S. Ni, P. Cui, J. Hazard. Mater. 156 (2008) 135–140. [19] Y. Zhang, F. Hou, Y.W. Tan, Chem. Commun. 48 (2012) 2391–2393. [20] A.B. Jorge, J. Fraxedas, A. Cantarero, A.J. Williams, J. Rodgers, J.P. Attfield, A. Fuertes, Chem. Mater. 20 (2008) 1682–1684. [21] A.B. Jorge, Y. Sakatani, C. Boissière, C. Laberty-Roberts, G. Sauthier, J. Fraxdas, C. Sanchez, A. Fuertes, J. Mater. Chem. 22 (2012) 3220–3226. [22] H.Y. Huang, R.T. Yang, Langmuir 17 (2001) 4997–5003. [23] C.C. Han, L. Ge, C.F. Chen, Y.J. Li, X.L. Xiao, Y.N. Zhang, L.L. Guo, Appl. Catal. B: Environ. 147 (2014) 546–553. [24] K.I. Katsumata, R. Motoyoshi, N. Matsushita, K. Okada, J. Hazard. Mater. 260 (2013) 475–482.

[25] S.C. Yan, Z.S. Li, Z.G. Zou, Langmuir 26 (2010) 3894–3901. [26] J. Fu, Y.L. Tian, B.B. Chang, F.N. Xi, X.P. Dong, J. Mater. Chem. 22 (2012) 21159–21166. [27] M.J. Bojdys, J.O. Müller, M. Antonietti, A. Thomas, Chem. Eur. J. 14 (2008) 8177–8182. [28] L. Xue, C.B. Zhang, H. He, Y. Teraoka, Appl. Catal. B: Environ. 75 (2007) 167–174. [29] G. Picasso, M. Gutiérrez, M.P. Pina, J. Herguido, Chem. Eng. J. 126 (2007) 119–130. [30] F.J. Perez-Alonso, I. Melián-Cabrera, M.L. Granados, F. Kapteijn, J.L.G. Fierro, J. Catal. 239 (2006) 340–346. [31] E. Béche, P. Charvin, D. peranau, S. Abanades, G. Flamant, Surf. Interface Anal. 40 (2008) 264–267. [32] A.E. Nelson, K.H. Schulz, Appl. Surf. Sci. 210 (2003) 206–221. [33] S.C. Petitto, M.A. Langell, J. Vac. Sci. Technol. A 22 (2004) 1690–1696. [34] M.M. Natile, A. Glisenti, Chem. Mater. 17 (2005) 3403–3414. [35] L.F. Liotta, G. Di Carlo, G. Pantaleo, A.M. Venezia, G. Deganello, Appl. Catal. B: Environ. 66 (2006) 217–227. [36] J.Y. Luo, M. Meng, X. Li, X.G. Li, Y.Q. Zha, T.D. Hu, Y.N. Xie, J. Zhang, J. Catal. 254 (2008) 310–324. [37] H. Chen, A. Sayari, A. Adnot, F. Larachi, Appl. Catal. B: Environ. 32 (2001) 195–204. [38] C.M.Y. Yeung, K.M.K. Yu, Q.J. Fu, D. Thompsett, M.I. Petch, S.C. Tsang, J. Am. Chem. Soc. 127 (2005) 18010–18011. [39] M. Kang, M.W. Song, C.H. Lee, Appl. Catal. A: Gen. 251 (2003) 143–156. [40] W. Cai, Q. Zhong, S.L. Zhang, W. Zhao, Chem. Eng. J. 236 (2014) 223–232. [41] J.Y. Luo, M. Meng, X. Li, X.G. Li, Y.Q. Zha, T.D. Hu, Y.N. Xie, J. Zhang, J. Catal. 254 (2008) 310–324. [42] L.J. Liu, Y. Cao, W.J. Sun, Z.J. Yao, B. Liu, F. Gao, L. Dong, Catal. Today 175 (2011) 48–54. [43] F. Esch, S. Fabris, L. Zhou, T. Montini, C. Africh, Science 309 (2005) 752–755. [44] R. Si, Y.W. Zhang, S.J. Li, B.X. Lin, C.H. Yan, J. Phys. Chem. B 108 (2004) 12481–12488. [45] B.M. Reddy, P. Bharali, P. Saikia, J. Phys. Chem. C 112 (2008) 11729–11737. [46] L. Poul, N. Jouini, F. Fiévet, Chem. Mater. 12 (2000) 3123–3132. [47] A. Ryoji, M. Takeshi, O. Takeshi, A. Koyu, T. Yasunori, Science 293 (2001) 269–271. [48] S.X. Liu, X.Y. Chen, J. Hazard. Mater. 152 (2008) 48–55. [49] W. Zhao, Q. Zhong, Y.X. Pan, R. Zhang, Chem. Eng. J. 228 (2013) 815–823. [50] J.J. Xu, Y.H. Ao, D.G. Fu, C.W. Yuan, Colloid Surf. A-Physicochem. Eng. Asp. 334 (2009) 107–111. [51] D.H. Kim, D.K. Choi, S.J. Kim, K.S. Lee, Catal. Commun. 9 (2008) 654–657. [52] J.G. Yu, H.G. Yu, B. Cheng, X.J. Zhao, J.C. Yu, W.K. Ho, J. Phys. Chem. B 107 (2003) 13871–13879. [53] J. Xu, Y.J. Wang, Y.F. Zhu, Langmuir 29 (2013) 10566–10572. [54] B. Azambre, L. Zenboury, A. Koch, J.V. Weber, J. Phys. Chem. C 113 (2009) 13287–13299. [55] Q. Liang, X.D. Wu, D. Weng, H.B. Xu, Catal. Today 139 (2008) 113–118. [56] Z. Zhao, X.G. Yang, Y. Wu, Appl. Catal. B: Environ. 8 (1996) 281–297. [57] E. Giamello, Catal. Today 41 (1998) 239–249.