Graphite (99.9%) was purchased from LiuMao Graphite Mining Company. All the other chemicals were commercially available in at least analytical reagent grade and were used without further purification. Acetonitrile and aqueous ammonia (25%) were purchased from Beijing Chemicals Works.

Graphene oxide was prepared by the modified Hummers’ methods previously reported ^{[5, 15]}. HCl water solution (5%) was used to remove residual Mn elements. After the graphene oxide powder was dried in a 60 ^oC oven for more than 24 h, the powder was heated to 250 ^oC at 20 ^oC/min. Then the powder was pyrolyzed to pyrolytic graphene oxide (PGO). These PGO materials were used as the carbon precursor and acetonitrile or aqueous ammonia was used as the nitrogen precursor respectively. N₂ (> 99.999%) was used as the carrier gas to feed the saturated vapor of acetonitrile or aqueous ammonia to the PGO (shown in Scheme 1). For the annealing experiment, 50 mg PGO was packed into a quartz tube and the oven was heated to 250, 450, 650, or 850 °C and maintained for 6 h. These N-doped graphene materials were labelled as Ac-T and Am-T (Ac and Am for acetonitrile and aqueous ammonia, respectively, and T indicated the annealing temperature). For example, Ac-850 means the samples was prepared from PGO and acetonitrile at 850 °C. The weight of PGO before and after the treatment was recorded.

Scheme 1. Preparation of N-doped graphene.

For the oxidation of ethylbenzene, 10 mg catalyst, 1 mmol ethylbenzene, 3 mmol tert-butyl hydroperoxide (TBHP), and 3 mL H₂O were put into a 50 mL sealed pressure glass vessel with magnetic stirring. After reaction at 80 ^oC for 24 h, 0.05g dodecane and 6 mL CH₂Cl₂ were added into the tube as internal standard and extraction liquid, respectively. Upon removal of the catalyst, the mixture was analyzed by an Agilent GC 7820A equipped with a HP-5 column and an FID detector.

The decomposition ratio of TBHP in the different catalysts was determined by the gasometric method given in our previous paper ^[22]. For each experiment, 30 mg catalyst, 30 mmol TBHP, and 30 mL H₂O were used with magnetic stirring.

TEM images were obtained with a Tecnai G220 S-Twin microscope operating at an accelerating voltage of 200 kV. SEM images were collected with a KYKY-AMRAY-1000B microscope. Raman measurements were performed under ambient conditions using a 532nm (2.33 eV) laser in the back-scattering configuration on a Jobin-Yvon HR800 spectrometer. XPS spectra were obtained using an Axis Ultra spectrometer (Kratos, UK). A monochromatic Al-K_α (1486.6 eV) X-ray source was used at a power of 225 W (15 kV, 15 mA). To compensate for surface charging effects, the binding energies were calibrated using the C 1s carbon peak at 284.8 eV.

The morphology of these N-PGO materials was analyzed by TEM, SEM, and their mass change before and after the doping process. The mass changes of these PGO materials are listed in Table 1. For the acetonitrile treated samples at 250 °C, where the decomposition ratio of acetonitrile was low, the weight loss resulted from the removal of oxygen groups, while at higher temperatures (450-850 °C), their weight increases and particularly, at 850 °C, the weight increased several fold. At high temperatures, the decomposition ratio of acetonitrile increased much, which resulted in much carbon deposition and weight gain processes. The PGO materials treated with aqueous ammonia show increasing amounts of weight loss with annealing temperatures. Two reasons can explain this phenomenon: the removal of oxygen groups in the inert atmosphere and the carbon atom is oxidized by steam at high temperatures. SEM images of the PGO materials before and after the N-doping treatment are shown in Fig. 1. The PGO materials (Fig. 1(a)) are formed by separated graphene sheets because of the delamination of graphite oxide in the pyrolytic process. After the doping process, their layered structure remained indicating the uniformity of the N-doping and carbon deposition processes. Moreover, the TEM images of the Am-850 and Ac-850 materials (Fig. 2) showed that the PGO materials after the treatment with acetonitrile have gone through a heavy carbon deposition process. These TEM images are consistent with the weight changes shown in Table 1. It is clear that all the materials shown were typical layered shapes of graphene, which means that the annealing treatment did not change the original structure of the PGO.

Table 1 Mass change of PGO after the doping process. Temperature (°C) Nitrogen resource (%) Acetonitrile Ammonia water 250 -14 -5.3 450 0.30 -8.2 650 20 -22 850 330 -82

Table 1 Mass change of PGO after the doping process.

Fig. 1. SEM images of (a) PGO, (b,c) Ac-850, and (d) Am-850.

Fig. 2. TEM images of (a) Am-850 and (b) Ac-850.

Raman spectroscopy was also used to further understand the doping process. All these carbon materials showed an obvious D band at 1350 cm^-1 and a G band at 1580 cm^{-1 [15]} in Fig. 3. For the aqueous ammonia treated samples, the I_D/I_G ratio increased with annealing temperatures because of more corrosion action of steam at higher temperatures. Of these N-doped graphene materials, the Ac-850 sample had the largest I_D/I_G ratio indicating that this sample has more defects than the other samples. Because of carbon deposition, the N-doped graphene with acetonitrile as nitrogen source showed a gradual overlap of the D and G bands ^[23]. This overlap was obvious for the Ac-850 sample as the most carbon deposition occurred at 850 °C. The changing trends of the Raman spectra agreed with the weight change data in Table 1 and the TEM images. Furthermore, at the same treatment temperature, the aqueous ammonia treated samples showed a higherI_D/I_G ratio than the acetonitrile treated ones because more oxygen remained in these examples.

Fig. 3. Raman spectra of the N-doped graphene materials.

The XPS spectra provided more information on the growth mechanism of these N-doped graphene materials. The O and N contents from XPS are listed in Table 2. First, the XPS results showed that the N content of all these materials increased with annealing temperatures except for the Ac-650 sample. A general trend that the O content decreased with the doping of N occurred. The loss of O was partially due to the reductive substitution by N in the doping process ^[24]. Besides, a high temperature treatment under an inert atmosphere also improved the crystallinity of the PGO and derived O ^[25]. Am-850 has a higher O content than Am-650 due to the strong steam corrosion at a higher temperature. Hence, aqueous ammonia in this annealing process not only decreased the oxygen amount of the PGO (effect of ammonia gas) but also corroded carbon atoms at high temperatures (effect of steam). Although the Am-850 sample did not have the highest N content, but because of the extraction of oxygen, it still possessed the largest I_D/I_G ratio in the Raman spectra.

Table 2 Ocontent and N content of the N-doped samples determined by XPS. Sample O content N content Ac-250 11 0.59 Ac-450 7.9 4.8 Ac-650 4.4 3.8 Ac-850 3.0 6.1 Am-250 9.5 2.0 Am-450 7.9 2.6 Am-650 3.9 4.6 Am-850 4.6 5.2

Table 2 Ocontent and N content of the N-doped samples determined by XPS.

In order to further understand the growth mechanism of the N species in these N-graphene materials, a deconvolution process was used to get the detailed composition of the N species (Fig. 4). The N elements were fitted into five species: pyridinic nitrogen species at 398.4 eV, amino nitrogen species at 399.3 eV, pyrrollic nitrogen species at 400.2 eV, and the species at 401.1 and 403.5 eV assigned to graphitic and N⁺-O^- nitrogen, respectively ^{[15, 24, 26]}. At low annealing temperatures (250 and 450 °C), most of the nitrogen species were the pyridinic and amino nitrogen species. At high temperatures (650 and 850 °C), the graphitic and N⁺-O^- nitrogen species dominates.

Fig. 4. N 1s XPS spectra of the N-doped graphene materials.

The trend of changing nitrogen species with temperature is shown in Fig. 5 to show the influence of annealing temperature. For the nitrogen species of the aqueous ammonia treated samples, the pyridinic, amino, and pyrrollic nitrogen species first increased with temperature and then decreased. In contrast, the graphitic and N⁺-O^- nitrogen species only increased with temperature. The different behavior of these nitrogen species with temperature resulted from their distinct chemical natures ^[27]. For example, the amino nitrogen species are only attached to the graphene sheets by a single C-N bond, which is a result of the oxygen groups being reduced by ammonia. In contrary, the breaking of three C-C bond and formation of C-N bonds are needed to form the graphitic nitrogen. Clearly, a higher temperature is needed to accomplish this doping reaction. Moreover, amino nitrogen species are not stable at high temperatures since only an unstable single C-N bond attached them to the graphene sheets. The nitrogen species with low binding energies (pyridinic amino and pyrrollic nitrogen) in the XPS spectra are easily formed and at the same time, easily broken off at high temperatures. However, the formation of graphitic and N⁺-O^- species needs high temperatures to overcome the high activation energy. The higher binding energies of the graphitic and N⁺-O^- nitrogen species than the other nitrogen species is strong support for our deduction. The N-doped samples prepared from acetonitrile and PGO followed the same trend because of the same chemical nature of the different nitrogen species. Our deduction also explained the phenomenon that Ac-650 has less nitrogen than the Ac-450 sample. At 650 °C, the amino nitrogen species decreased a lot and the other nitrogen species did not increase significantly. The different chemical nature of the different nitrogen species led to a best temperature to dope nitrogen atoms into the graphene sheets ^{[26, 27]}. In contrast, the aqueous ammonia treatment process oxidized carbon atoms at high temperatures to provide more oxygen groups to be reduced by ammonia. Therefore, the N content of the aqueous ammonia treated samples did not show a volcano shape variation with temperature, but only an increase.

Fig. 5. Content of different N species as a function of annealing temperatures for the N-doped graphene materials.

The oxidation of ethylbenzene was chosen as the probe reaction to test the catalytic activity of these N-graphene materials. All the reactions gave acetophenone as the main products and some other oxidation products like benzaldehyde and benzoic acid as side products. Generally, the N-doped graphene materials treated at high temperatures exhibited better catalytic activity than those of those treated at low temperatures. However, Am-850 had a lower catalytic activity than the other materials. As we have demonstrated in a previous paper that the graphitic nitrogen is responsible for the activation of TBHP ^[15], here we also correlated the contents of total nitrogen and graphitic nitrogen species with the yields of acetophenone in Table 3.

Table 3 Catalytic activity of N-doped graphene materials in the selective oxidation of ethylbenzene. Entry Sample Yield of acetophenone (%) Nitrogen element (%) Oxygen element (%) Graphitic nitrogen (%) 1 Ac-250 55.9 0.59 11 0.21 2 Ac-450 84.1 4.8 7.9 1.2 3 Ac-650 87.9 3.8 4.0 1.4 4 Ac-850 81.6 6.1 3.0 3.4 5* Ac-850 49.9 6.1 3.0 3.4 6 Am-250 55.8 2.0 9.5 0.18 7 Am-450 86.5 2.6 7.9 0.35 8 Am-650 93.2 4.6 3.9 0.77 9 Am-850 57.9 5.2 4.6 1.1 10* Am-850 81.9 5.2 4.6 1.1 Reaction conditions: 1 mmol ethylbenzene, 3mmol TBHP, 10 mg catalyst and 3 mL H2O were put into a 50 mL sealed pressure glass vessel with magnetic stirring, 80 °C (* 65 °C), 24 h.

Table 3 Catalytic activity of N-doped graphene materials in the selective oxidation of ethylbenzene.

From Table 3 we can see that it is not total nitrogen but only the graphitic nitrogen species that are active in this catalytic reaction. First, for the instance of Ac-450 and Ac-650, the Ac-650 sample has a total nitrogen content of 3.8% which is lower than the content of Ac-450 at 4.8%. Nevertheless, the similarity of the contents of graphitic nitrogen species (1.5% and 1.7%) causes them to show similar yields of acetophenone. However, it is of note that the Am-850 sample which has the most abundant defects and the second highest graphitic nitrogen species content was less active than all the other materials. We proposed that this sample was too active to promote the decomposition of TBHP because too much oxidant was wasted. This is shown by the TBHP decomposition curves shown in Fig. 6 where Ac-850 showed a significantly faster decomposition rate of TBHP than that of any other samples. The most active Ac-650 showed a high and modest decomposition rate. Our similar observation that the kinetic control of the decomposition of H₂O₂ was essential in the benzene oxidation reaction is strong support to our proposal in this paper ^[22]. Second, the ability of the two precursors in activating the PGO materials was also compared. Obviously, Ac-650 prepared from aqueous ammonia showed a higher yield than the other catalysts. In order to further compare the activities of these N-doped graphene materials, we also used the Ac-850 and Am-850 samples to catalyze this reaction at a lower temperature (65 °C). Compared with the reactions at 80 °C, Am-850 showed a better catalytic activity, while the Ac-850 sample was less active. This phenomenon agreed with Fig. 6 that Am-850 was much more active than the Ac-850 sample in catalyzing the decomposition of TBHP. Therefore at a lower temperature, Am-850 would show abetter catalytic activity due to its better TBHP decomposition ability. Therefore, aqueous ammonia is a more effective doping precursor than acetonitrile in activating carbon materials in this ethylbenzene oxidation reaction.

Fig. 6. TBHP decomposition by the N-doped graphene materials.