Low-temperature combustion of methane over graphene templated Co3O4 defective-nanoplates

Transition metal oxides are the potential catalysts to replace noble-metal based catalyst for the catalytic combustion of methane due to the tolerable reactivity and low cost. However, these catalysts are challenged by the low temperature reactivity. Herein, the surface defective Co3O4 nanoplates are realized through a facile co-precipitation and thermal reduction method with the association of GO. The resultant catalysts (CoGO50) demonstrate a superior low-temperature reactivity for the methane oxidation to CO2 and H2O in comparison with the common Co3O4 catalyst. The reliable stability of CoGO50 catalyst was proved by 80 h testing with intermittent feeding of water vapor. The experimental analysis demonstrates that the presence of a small amount of GO significantly affects the catalysts in surface valence state, active oxygen species and surface oxygen vacancies through reacting with the cobalt oxide as a reductant. Moreover, GO plays as 2D confine template to form smaller and thinner nanoplates. This work provides a facile method to control the surface properties of catalyst not only for Co3O4 based catalysts but also for wider solid catalysts.

www.nature.com/scientificreports/ In this work, we developed graphene oxide involved Co 3 O 4 catalyst (CoGO) through a facile co-precipitation method. The low temperature activity of the resultant catalyst was highly improved, which exhibits 100 °C decrease for the complete conversion of methane in compared with the common Co 3 O 4 . Effects of GO addition on the structure and surface chemistry of the catalyst as well as the reaction stability were thoroughly investigated.

Results
Effects of GO on the structure and texture of catalysts. The scanning electron microscope (SEM) images show that all samples of Co 3 O 4 , CoGO50 and CoGO100 have the hexagonal plate-like morphology (Fig. 1A,C,E). Noted that the precursor of Co 3 O 4 , i.e., CoOOH, also possesses the hexagonal nanoplate morphology, suggesting that the shape of Co 3 O 4 was well maintained upon the calcination ( Figure S1). The average sizes of CoGO20, CoGO50 and CoGO100 nanoplates (~ 150-250 nm) are slightly smaller than that of Co 3 O 4 (~ 300 nm) (Figs. 1 and S2). This suggests that GO addition not only protects the plate shape of Co 3 O 4 but also confines the growth of Co 3 O 4 crystals. The transmission electron microscope (TEM) images of single nanoplate of Co 3 O 4 , CoGO50 and CoGO100 display that the thickness of CoGO50 is thinner than others, which may contribute to the confinement effect of GO flakes in the preparation (Fig. 1B,D,F). In addition, no graphene layers were found from both CoGO50 and CoGO100, suggesting that GO has been consumed in the calcination of catalysts.
The high-resolution TEM (HRTEM) image of CoGO50 shows an interplanar spacing of 0.287 nm, corresponding to the cubic Co 3 O 4 (220), and an interplanar spacing of 0.467 nm, assigned to the cubic Co 3 O 4 (111) plane (Fig. 1G). The microstructure of CoGO50 is well consistent with that of Co 3 O 4 ( Figure S3). Consistently, no lattice fringe of graphic carbon was observed from the HRTEM image of CoGO50 (Fig. 1G). The corresponding selected-area electron diffraction (SAED) pattern reveals the typical diffraction spots of hexagonal Co 3 O 4 , suggesting a high crystallinity of CoGO50 (Fig. 1H).
The crystalline structure of the catalysts was measured by X-ray diffraction (XRD), as shown in Fig. 2A. Before calcination, CoGO50 precursor exhibits the crystal structure of CoOOH (JCPDS PDF#07-0169). In addition, no signals of GO were detected from CoGO50 precursor, suggesting that GO is highly dispersed on the CoOOH surface with ultrathin thickness.  23 . Similarly, no peaks of GO (~ 12°) or graphitic carbon (~ 25°) were observed from CoGO50 and CoGO100, indicating that GO was decomposed in the calcination or the amount of residual GO/GO-derivatives is beyond the detection limit 24 .
The structure of samples was further investigated by Raman spectroscopy (Fig. 2B). The spectrum of GO contains a G band at 1595 cm −1 , arising from the first-order scattering of sp 2 carbon atoms in a 2D hexagonal lattice, and a D band at 1343 cm −1 , ascribed to the vibrations of carbon atoms in plane terminations of disordered graphite 25,26 . For the CoGO50 precursor, the characteristic D and G bands of GO with weak intensities were also observed ( Fig. 2B inset), indicating that the surface covered GO is ultrathin. This is consistent with the XRD results that GO is highly dispersed on the CoOOH surface. The spectra of CoGO50 and CoGO100 post peaks at 467 cm −1 (E g ), 515 cm −1 (F 1g 1 ), 613 cm −1 (F 2g 1 ) and 679 cm −1 (A 1g 1 ) 27 , which are the same as that of Co 3 O 4 spinel structure 28 . This reveals that the bulk structure of Co 3 O 4 was well maintained for the samples of CoGO50 and CoGO100. Moreover, no signals of graphtic D and G bands were observed from both CoGO50 and CoGO100, suggesting the amount of GO or rGO is below the detection limit.
The inner structure and surface area of the samples were measured by nitrogen adsorption isotherms (Fig. 2C). All the samples showed the IV-typed sorption isotherm with H3-typed hysteresis loop in the relative pressure range of P/P 0 = 0.5-1.0, suggesting a mesoporous structure for these samples 29,30 . The pore sizes calculated by the Barrett-Joyner-Halenda (BJH) method show a unimodal distribution centred at ~ 12.7 nm for these samples (Fig. 2D), which is consistent with the TEM observations. For CoGO50 and CoGO100, the Brunauer-Emmett-Teller (BET) specific surface areas are 39.7 and 43.2 m 2 g −1 , respectively, which are higher than that of Co 3 O 4 (31.5 m 2 g −1 , Table S1). This indicates that the addition of GO increases the BET surface area through forming smaller and thinner nanoplates. Moreover, the samples of CoGO50 and CoGO100 also exhibit larger pore volumes (0.31 and 0.28 cm 3 g −1 ) than that of Co 3 O 4 (0.23 cm 3 g −1 ), suggesting that the existence of GO layer promotes the formation of pore during the calcination.
The thermal behaviours of the catalysts were investigated by a thermogravimetric-mass analyser (TG-MS) in the temperature range of 60-1000 °C. As shown in Fig. 3A, the CoGO50 precursor (i.e., GO covered CoOOH) exhibits weight loss mainly at ~ 110 °C and ~ 320 °C. Combined with the mass spectrometry signals, the weight loss at low temperature of 80-120 °C is ascribed to evaporation of adsorption water while the weight loss centred at 286 °C is contributed to the dehydration reaction of phase transition from CoOOH to Co 3 O 4 as well as the thermal reduction of GO to reduced GO (rGO) 17 . Furthermore, CO 2 signal was detected at 319 °C, which reflects the deep oxidation of GO/rGO to CO 2 and H 2 O. It confirms that the surface covered GO was decomposed in the calcination of catalyst, in line with the observations of TEM and Raman measurements. The decomposition of rGO layer would impact the surface chemistry of Co 3 O 4 if the surface oxygen of Co 3 O 4 is involved in the oxidation of rGO. The thermal stability of resultant catalysts was tested by TG (Fig. 3B). Except the desorption of water at low temperature, all the samples kept stable in weight at the temperature of < 890 °C, suggesting a reliable thermal stability for these catalysts. On the other hand, TG curves show weight loss in the temperature ranges of 890-930 °C, which can be attributed to the phase transition from Co 3 O 4 to CoO with O 2 generation ( Figure S4). The weight loss rate of this step was 6.4%, which is close to the theoretical value of O 2 release (6.6%).   www.nature.com/scientificreports/ For the twin peaks of Co 2p, the binding energies (BE) at 781.5 eV and 796.5 eV are attributed to Co 2+ while BEs at 794.5 eV and 779.0 eV are assigned to Co 3+ (Fig. 4B) 32 . As shown in Table 1, the surface Co 2+ /Co 3+ ratio of CoGO50 is 0.86, which is significantly higher than that of Co 3 O 4 (0.58) and CoGO100 (0.65). This indicates that the surface of Co 3 O 4 is reduced by GO and its derivatives in the preparation. In the calcination of catalyst, the tightly contacted GO would be thermal-reduced firstly by losing oxygen-containing groups under the heating conditions. With higher temperature, the rGO tends to be further oxidized to CO 2 and H 2 O, where oxygen molecules and surface oxygen of Co 3 O 4 are involved in the oxidation reaction ( Figure S5). Therefore, partial of Co 3+ atoms on the catalyst surface would be reduced to Co 2+ and/or oxygen vacancies would be formed on the catalyst surface 33 .

Scientific Reports
As shown in Fig. 4C, the O1s spectra of catalysts were composed by two peaks centred at 529.7 ± 0.1 eV (lattice oxygen, O latt ) and 531.2 ± 0.2 eV (surface adsorbed oxygen, O ads ) 29   www.nature.com/scientificreports/ 0.75, which is higher than that of Co 3 O 4 and CoGO100 (Table 1). As indicated by the Co 2p results, the partial surface oxygen was consumed in the calcination, which may lead to the formation of oxygen vacancies 33 . Furthermore, the oxygen vacancies would enhance the adsorption and activation of molecular oxygen, resulting in high O ads /O latt ratio for CoGO50. It is well known that high O ads /O latt reflects high catalytic activity for the oxidation of hydrocarbons at low temperature [35][36][37] . Therefore, this suggests that CoGO50 possesses the highest reactivity for methane oxidation, which is well consistent with the following reaction test. In addition, the XPS analysis suggests that the spent CoGO50 have slightly decline in the O ads /O latt (0.69) and Co 2+ /Co 3+ (0.73), which are still higher than those of CoGO100 and Co 3 O 4 ( Figure S6). The redox properties of catalysts were investigated by hydrogen temperature-programmed reduction (H 2 -TPR) (Fig. 4D). All catalysts show two reduction peaks at 200-300 °C, attributable to the reduction of Co 3+ to Co 2+ , and at 350-400 °C, assigned to the reduction of CoO to metallic cobalt (Co 0 ) 32,38 . The reduction temperature downshifted from 276 °C of Co 3 O 4 to 214-246 °C of CoGO50. This reveals that CoGO50 contains more active oxygen species, which are much easier to be reduced. Therefore, CoGO50 would take high oxidative activity in the catalytic oxidation reaction.

Catalytic combustion of methane over CoGO catalysts. Methane catalytic combustion on Co 3 O 4
and CoGOx (x = 20, 50 and 100) catalysts were tested in a micro-fixed-bed reactor at 200 to 500 °C. The reaction was carried out with a feeding conditions of 1 vol% of CH 4 , 10 vol% of O 2 and 89 vol% of N 2 and GHSV = 30,000 mL g −1 h −1 . With temperature increasing, as shown in Fig. 5A, the methane conversion on these catalysts increased from 0 to 100% in the range of 200 to 500 °C. However, CoGOx catalysts exhibit higher activities at low temperature than Co 3 O 4 . In detail, the temperature for 100% conversion of methane (T 100 ) is 500 °C for Co 3 O 4 , which significantly decreases to 425, 400 and 450 °C for CoGO20, CoGO50 and CoGO100 catalysts, respectively. Similarly, the temperatures for methane conversion of 10% (T 10 ), 50% (T 50 ) and 90% (T 90 ) of CoGOx are much lower than that of Co 3 O 4 ( Table 2). CoGO50 has the highest activity for methane conversion with the lowest T 100 , T 90 , T 50 and T 10 and T 100 , which are 100 °C, 105 °C, 103 °C and 105 °C lower than the corresponding values of Co 3 O 4 . It demonstrates that GO involved preparation is significantly improve the reactivity of CoGO50 through forming surface defects.
The apparent activation energy (E a ) of CH 4 oxidation over the catalysts was also calculated via the Arrhenius plots (Fig. 5B). The order trend of E a for CH 4 oxidation on these catalysts is followed as Co 3 O 4 (86.5 kJ mol −1 ) > CoGO100 (61.4 kJ mol −1 ) > CoGO20 (59.9 kJ mol −1 ) > CoGO50 (53.4 kJ mol −1 ), which means CH 4 was most easily activated and oxidized by CoGO50. It is well consistent with the reactivities of these catalysts. The reactivity of CoGO50 for methane combustion was compared with the reported results over various catalysts  Table 2. Reaction temperature for obtaining 10%, 50%, 90% and 100% methane conversion (1 vol% CH 4 , 10 vol% O 2 and N 2 balance, P = 1 bar, GHSV = 30,000 mL g −1 h −1 ). www.nature.com/scientificreports/ (Table S2). CoGO50 is more active than the cobalt oxide catalyst and even comparable to the noble metal-based catalysts under the similar conditions. To investigate the temperature sensibility of methane conversion, both Co 3 O 4 and CoGO50 were comparatively measured by using in-situ diffuse reflectance infrared Fourier transform spectra (DRIFTS) with simulated feeding at different temperature (Fig. 6). The band at 2880 cm −1 represents the C-H stretching, which is ascribed to the adsorbed and/or the activated CH 4 on the catalyst 23,39 . The bands at 1560 cm −1 and 1420 cm −1 can be assigned to the asymmetric and symmetric stretching vibration of intermediate carbonates (ν as CO 3 2+ and ν s CO 3 2+ ), respectively 40 . The band at 3380 cm −1 could be assigned to the stretching vibration of hydroxyl, which may reflect the end product of adsorbed water 23,39 . The band at 2200 cm −1 is ascribed to the end product CO 2 39 . For both samples, the intensities of all signals from CH 4 adsorption/activation, intermediates and end products increase with the reaction temperature, in line with the real reaction results. However, the DRIFTS spectra of CoGO50 show stronger signals of intermediates and products at low temperature in compared with that of Co 3 O 4 (Fig. 5A,B). The signal of CO 2 can be observed from CoGO50 even at 200 °C, while the temperature with visible CO 2 peak is 325 °C for Co 3 O 4 . This is well consistent with the real reaction tests that CoGO50 have the same reactivity for methane at the temperature more than 100 °C lower than Co 3 O 4 .

Catalysts T 10 ( o C) T 50 ( o C) T 90 ( o C)
The effects of impurities of feed gas on the performance of CoGO50 were tested by using water vapor and SO 2 , respectively. Figure 7A shows the stability of CoGO50 for methane combustion at 375 and 400 °C. At 375 °C, the  www.nature.com/scientificreports/ CH 4 conversion remain stable around 92%. With 10% water vapor feeding, the CH 4 conversion decreased rapidly and kept stable on 75%. This indicates that vapor environment has a suppression effect on the CH 4 oxidation. On the one hand, water vapor may adsorb and cover partial of active sites of catalyst to form Co(OH) 2 41 . On the other hand, the oxidation of CH 4 to CO 2 and H 2 O may be thermodynamically limited as the feeding water is one of the products. The conversion can be 100% recovered after switching the vapor off. At 400 °C, the CH 4 conversion presents the similar change tendency in the conditions of introduce/shut-down10% vapor, which first declined from 100 to ~ 85% and then returned to 100%.
The poisoning effects of sulphur dioxide on the catalyst was also tested over CoGO50, in which 50 ppm SO 2 was introduced into the reaction system at 350 °C (Fig. 7B). Upon SO 2 , the methane conversion rapidly decreased from 91 to ~ 65% in 3 h. It was then slowly increased to ~ 70% in the following 5 h. After switching SO 2 off and purging with N 2 , the CH 4 conversion slightly increased to ~ 85% in 4 h and then kept stable, suggesting that more than 93% of original reactivity was recovered. This indicates that SO 2 tends to impact the activity of CoGO50, irreversibly. SO 2 can strongly adsorb on the catalyst via the reactions of Co 3 O 4 + SO 2 → Co 3 O 4 ·SO 2 , which may further form sulphates through the reactions of 2Co 3 O 4 ·SO 2 + O 2 → 2Co 3 O 4 ·SO 3 or Co 3 O 4 ·SO 3 → Co 3 (SO 4 ) 4 42 .

Discussion
In summary, the 2D Co 3 O 4 nanoplates containing surface defects have been demonstrated by using graphene oxide flakes not only as shaping templates but also as sacrifice agent for defect formation. The resultant CoGO50 exhibited a superior low-temperature reactivity for the methane combustion to CO 2  Methane catalytic combustion. The catalytic combustion of methane was tested in a fixed bed quartz tube reactor (ID = 6 mm, L = 500 mm) at atmospheric pressure 43 . 100 mg catalyst (40-60 mesh) was packed in the middle of reactor with quartz wool layers on both ends of reactor. Mixture gas containing 1 vol% CH 4 , 10 vol% O 2 and 89 vol% N 2 was supplied by Pujiang Gas Co. The above feed gas was introduced to reactor with a flow rate of 50 mL min −1 , which corresponds to a gas hourly space velocity (GHSV) of 30,000 mL g −1 h −1 . The temperature was increased stepwise from 200 to 550 °C at a ramp of 5 °C min −1 . The temperature dependence of CH 4 conversion was tested with an interval of 25 °C in the range of 200-550 °C. The reaction products were analysed by a micro gas chromatograph (INFICON 3000) equipped with MS5A and Plot Q columns and TCD detector. Before analysis, the reaction temperature at each step was stabilized at least 30 min. To investigate the effects of impurities on the catalytic performance, 10 vol% of water vapor was introduced into the reactor together with feed gas mixture. For the tests of sulphur tolerance, a mixed gas of 50 ppm SO 2 , 1 vol% CH 4 , 10 vol% O 2 balanced with high purity N 2 was used with a flow rate of 50 mL min −1 .
Characterization. The morphology of the catalyst samples was measured by a transmission electron microscope (TEM, JEM-2100, 200 kV) and a scanning electron microscope (SEM, Zeiss SUPRA 55 SAPPHIRE, 2-20 kV). The structure of Co 3 O 4 based catalysts was recorded by X-ray diffraction (XRD, Rigaku Ultima IV) with Cu Kα radiation (λ = 0.15406 nm, 40 kV, 40 mA). The specific surface area and pore size distribution of catalysts were measured by N 2 -sorptions on an automatic micropore physisorption analyser (TriStar II 3020). Specific surface areas of the products were calculated by the Brunauer-Emmett-Teller (BET) method and pore sizes were calculated using the Barrett-Joyner-Halenda (BJH) method. Raman spectroscopy measurements were performed using a Renishaw Raman spectrometer using a 12.5 mW laser source at an excitation wavelength of 532 nm. The near-surface chemical information of catalysts was analysed by X-ray photoelectron spectroscopy (XPS, Thermo Scientific K-Alpha) using Al Kα (hν = 1486.6 eV) as the excitation source. XPS peak positions were corrected with the help of the C 1s peak at 284.8 eV. The thermal behaviours of catalysts and the catalyst precursor were analysed by a thermogravimetric analyser with a mass spectrometer (TGA-MS, NETZSCH, STA 449F3) operated from 60 to 1000 °C with heating rate of 5 °C min −1 under an air flow (20 mL min −1 ). Hydrogen temperature-programmed reduction (H 2 -TPR) was carried out on a chemical adsorption analyser (Micromeritics AutoChem II 2920) equipped with a thermal conductivity detector (TCD). In situ diffuse reflectance infrared