*Article* **Enhanced Catalytic Hydrogen Peroxide Production from Hydroxylamine Oxidation on Modified Activated Carbon Fibers: The Role of Surface Chemistry**

**Wei Song 1,2,\* , Ran Zhao <sup>1</sup> , Lin Yu <sup>1</sup> , Xiaowei Xie <sup>1</sup> , Ming Sun <sup>1</sup> and Yongfeng Li <sup>1</sup>**


**Abstract:** Herein, direct production of hydrogen peroxide (H2O<sup>2</sup> ) through hydroxylamine (NH2OH) oxidation by molecular oxygen was greatly enhanced over modified activated carbon fiber (ACF) catalysts. We revealed that the higher content of pyrrolic/pyridone nitrogen (N5) and carboxylanhydride oxygen could effectively promote the higher selectivity and yield of H2O<sup>2</sup> . By changing the volume ratio of the concentrated H2SO<sup>4</sup> and HNO<sup>3</sup> , the content of N5 and surface oxygen containing groups on ACF were selectively tuned. The ACF catalyst with the highest N5 content and abundant carboxyl-anhydride oxygen containing groups was demonstrated to have the highest activity toward catalytic H2O<sup>2</sup> production, enabling the selectivity of H2O<sup>2</sup> over 99.3% and the concentration of H2O<sup>2</sup> reaching 123 mmol/L. The crucial effects of nitrogen species were expounded by the correlation of the selectivity of H2O<sup>2</sup> with the content of N5 from X-ray photoelectron spectroscopy (XPS). The possible reaction pathway over ACF catalysts promoted by N5 was also shown.

**Keywords:** activated carbon fiber; surface modification; hydrogen peroxide; hydroxylamine; pyrrolic/ pyridone nitrogen

## **1. Introduction**

Hydrogen peroxide (H2O2), as a green chemical, attracts research attention in both energy and environmental related fields because it has the highest content of reactive oxygen among the common oxidants and the green by-product [1]. It has been widely used as a bleach in the paper and textile industry, an energy carrier in fuel cells, an oxidant in chemical production, wastewater treatment, hydrometallurgy and electronics industry [2]. Notwithstanding, the current industrial production of H2O<sup>2</sup> is mainly through the anthraquinone oxidation process [3], which involves multistep reactions, massive energy consumption and waste generation. Furthermore, the cost and safety problems are also raised ineluctably by the handle, transport and storage of high concentration H2O2. Nevertheless, in many practical applications, H2O<sup>2</sup> with only a low concentration could satisfy the demand in the reactions, such as selective oxidation, on-site degradation of dye, sewage treatment and disinfection (<30 mM) [4–6]. In this context, research on alternative production methods of H2O<sup>2</sup> and it's in situ use has been the research focus [5,7–15]. The direct generation of H2O<sup>2</sup> by the reaction of molecular hydrogen (H2) and oxygen (O2) is considered the most promising method [16–19], but the industrial application is obscured by the dangers of the explosive reaction mixture and the insufficiency of catalysts with high selectivity without considering the reaction systems of O<sup>2</sup> and H<sup>2</sup> at high pressure [20–25]. In recent years, both photo- and electro-catalytic H2O<sup>2</sup> production techniques are in the process of research, but the former is still suffered from a low selectivity and yield of H2O<sup>2</sup>

**Citation:** Song, W.; Zhao, R.; Yu, L.; Xie, X.; Sun, M.; Li, Y. Enhanced Catalytic Hydrogen Peroxide Production from Hydroxylamine Oxidation on Modified Activated Carbon Fibers: The Role of Surface Chemistry. *Catalysts* **2021**, *11*, 1515. https://doi.org/10.3390/ catal11121515

Academic Editors: Florica Papa, Anca Vasile and Gianina Dobrescu

Received: 30 October 2021 Accepted: 11 December 2021 Published: 13 December 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

with affordable raw materials while the latter faces the problems of low efficiency and complicated devices accompanied with high cost [2,26,27].

Therefore, in order to avoid the risk of explosion, it would be an effective way to choose an appropriate hydrogen source replacing H<sup>2</sup> in the direct synthesis process of H2O2. Thus, hydroxylamine (NH2OH) was viewed as an available alternative to H<sup>2</sup> since it could be transformed into H2O<sup>2</sup> by O<sup>2</sup> in the conditions of room temperature and normal pressure in an aqueous solution (2NH2OH + O<sup>2</sup> = N<sup>2</sup> + 2H2O + H2O2) [28]. This reaction is a simple step and easy to handle procedure with two major kinds of catalytic systems. Homogeneous manganese complexes catalysts with high TOF values were firstly studied in this reaction [29–31], but they suffered from separating and recycling problems. Afterward, noble metal particles (Au and Pd) dispersed on different supports were reported to catalyze this system effectively [32–34], but the low concentration of H2O<sup>2</sup> (0.05–0.1 wt.%) generation and the high cost of noble metal remain as the major impediment for the industrial applications.

Based on the research above, activated carbon (AC) was found to be an effective catalyst used in the direct H2O<sup>2</sup> production process through NH2OH oxidation by O<sup>2</sup> in our earlier research [35,36]. We found that the catalytic properties of AC were closely related to the surface oxygen-containing groups. In order to fulfill wider and higher demands in practical application, the selectivity and activity towards H2O<sup>2</sup> formation still need to be further improved. Moreover, the reactivity usually originated from the structure of carbon materials. Considering this, nitrogen doping on carbon materials has been found to be one of the most effective ways to improve the selectivity of H2O<sup>2</sup> in metal-free catalytic systems, although the selectivity usually depends on both oxygen- and nitrogen-doped atoms [13]. It is noteworthy that N-doping played a beneficial role for H2O<sup>2</sup> selectivity only at a low active surface site density while becoming detrimental at higher contents in some reactions [37,38]. Meanwhile, the microporous volume content on carbon materials was also found to have proportionality with selectivity in many catalytic systems [37,39].

Considering the above factors, the easily available polyacrylonitrile-based (PANbased) activated carbon fiber (ACF) would be an ideal candidate catalyst material. On the one hand, compared to the AC material with a large number of mesopores, PAN-based ACF has only microporous structures and the pore channels are directly open on the surface. More importantly, PAN-based ACF has an intrinsic nitrogen content without any further nitrogen-doped steps and the relatively simple surface modification process would be cost-effective [40]. In the present work, for the sake of comparative study of the generation and modification on the surface doped species, the improvements on the content and type of surface functional groups were intended by treating the PAN-based ACF with concentrated mixed acid in different volume ratios. The selectivity and yield of H2O<sup>2</sup> in the reaction on modified ACF catalysts were well interconnected with the amounts of pyrrolic/pyridone nitrogen (N5) and desorbed carboxyl-anhydride groups from the ACF surface. The possible reaction pathway over ACF catalysts promoted by N5 was also shown.

#### **2. Results and Discussion**

#### *2.1. Material Structure*

Figure 1 shows the scanning electron microscopy (SEM) images of ACF-0 sample and the corresponding elemental mapping images, along with the SEM images of ACF samples before and after surface modification by different volume ratios of concentrated H2SO<sup>4</sup> and HNO<sup>3</sup> (H2SO4/HNO<sup>3</sup> (*v*/*v*)). From Figure 1a–d, three main elements (C, N, O) were found on the ACF-0 sample which exhibited a long fiber feature with a diameter of about 15 µm. The surface roughness of the ACF samples gradually increased with the increase of H2SO4/HNO<sup>3</sup> (*v*/*v*) from 0.5 to 4, as shown from Figure 1e–h. For the ACF-Rw sample, the surface still kept smooth similar to the ACF-0 sample, while slight surface roughness was observed on the surface of ACF-R1. By increasing the value of

H2SO4/HNO<sup>3</sup> (*v*/*v*) to 2 and 4, some auricular-like sheet protrusions were both found on the surface of ACF-R2 and ACF-R4 samples. Consequently, the surface modification caused differently morphological changes on the ACF samples through the erosion of carbon surface by different H2SO4/HNO<sup>3</sup> (*v*/*v*), and some microporous structures on the ACF samples might be destroyed through mixed acid treatment with higher content of concentrated H2SO4. roughness was observed on the surface of ACF-R1. By increasing the value of H2SO4/HNO3 (*v*/*v*) to 2 and 4, some auricular-like sheet protrusions were both found on the surface of ACF-R2 and ACF-R4 samples. Consequently, the surface modification caused differently morphological changes on the ACF samples through the erosion of carbon surface by different H2SO4/HNO3 (*v*/*v*), and some microporous structures on the ACF samples might be destroyed through mixed acid treatment with higher content of concentrated H2SO4.

**Figure 1.** SEM images of the ACF-0 sample and the corresponding elemental mapping images for (**a**) overlap, (**b**) C, (**c**) N, (**d**) O. The SEM images of ACF samples after surface modification (**e**) ACF-Rw, (**f**) ACF-R1, (**g**) ACF-R2, (**h**) ACF-R4. **Figure 1.** SEM images of the ACF-0 sample and the corresponding elemental mapping images for (**a**) overlap, (**b**) C, (**c**) N, (**d**) O. The SEM images of ACF samples after surface modification (**e**) ACF-Rw, (**f**) ACF-R1, (**g**) ACF-R2, (**h**) ACF-R4.

The pore size distribution of ACF-0 and ACF-R4 samples, and nitrogen adsorptiondesorption isotherms of the ACF-0 sample without surface modification is shown in Figure 2. The detailed texture parameters for the ACF samples were shown in Table 1. Moreover, the ACF-0 sample displayed the features of type I isotherm, indicating the existence of micropores. There were two types of pores on the ACF-0 sample: micropore (1.41 nm and 1.13 nm) and supermicropore (0.78 and 0.57 nm) [41]. As for the ACF-R4 sample, the micropore content (1.14 nm) was greatly enhanced and some micropores enlarged to 1.69 nm while the supermicropore content was decreased compared with the ACF-0 sample. As displayed in Table 1, minor changes in the pore size and surface area were found between the ACF-0 and ACF-Rw samples but obviously decrease in micropore volume and surface area were observed on the ACF-R2 and ACF-R4 samples. As for the ACF-Rw sample, the surface area of micropores (*S*mic.), the total surface area (*S*BET) and the surface area of mesoporous (*S*mes.) was 895, 934 and 39 m2/g, respectively. Correspondingly, the micropore volume (*V*mic.), the total pore volume (*V*total) and the micropore width (*d*pore.) of ACF-Rw were calculated to be 0.361, 0.395 cm3/g and 0.85 nm. These results were similar to those of the ACF-0 sample. With the increase of H2SO4/HNO3 (*v*/*v*), the values of *V*Total and *S*BET initially reduced to 0.323 cm3/g and 686 m2/g on the ACF-R1 sample, then declined to 0.229 cm3/g and 481 m2/g on the ACF-R2 sample. Moreover, the *d*pore for ACF samples by mixed acid oxidation increased slightly from 0.87 to 0.95 nm, with increasing H2SO4/HNO3 (*v*/*v*) from 0.5 to 4. Notably, the mesopores of the ACF samples were not altered much through the modification by mixed acid, whereas the micropores decreased remarkably by higher values of H2SO4/HNO3 (*v*/*v*), especially in the ACF-R2 and ACF-R4 samples. This suggests that the higher content of H2SO4 caused severe destruction of the The pore size distribution of ACF-0 and ACF-R4 samples, and nitrogen adsorptiondesorption isotherms of the ACF-0 sample without surface modification is shown in Figure 2. The detailed texture parameters for the ACF samples were shown in Table 1. Moreover, the ACF-0 sample displayed the features of type I isotherm, indicating the existence of micropores. There were two types of pores on the ACF-0 sample: micropore (1.41 nm and 1.13 nm) and supermicropore (0.78 nm and 0.57 nm) [41]. As for the ACF-R4 sample, the micropore content (1.14 nm) was greatly enhanced and some micropores enlarged to 1.69 nm while the supermicropore content was decreased compared with the ACF-0 sample. As displayed in Table 1, minor changes in the pore size and surface area were found between the ACF-0 and ACF-Rw samples but obviously decrease in micropore volume and surface area were observed on the ACF-R2 and ACF-R4 samples. As for the ACF-Rw sample, the surface area of micropores (*S*mic.), the total surface area (*S*BET) and the surface area of mesoporous (*S*mes.) was 895, 934 and 39 m2/g, respectively. Correspondingly, the micropore volume (*V*mic.), the total pore volume (*V*total) and the micropore width (*d*pore.) of ACF-Rw were calculated to be 0.361, 0.395 cm3/g and 0.85 nm. These results were similar to those of the ACF-0 sample. With the increase of H2SO4/HNO<sup>3</sup> (*v*/*v*), the values of *V*Total and *S*BET initially reduced to 0.323 cm3/g and 686 m2/g on the ACF-R1 sample, then declined to 0.229 cm3/g and 481 m2/g on the ACF-R2 sample. Moreover, the *d*pore for ACF samples by mixed acid oxidation increased slightly from 0.87 to 0.95 nm, with increasing H2SO4/HNO<sup>3</sup> (*v*/*v*) from 0.5 to 4. Notably, the mesopores of the ACF samples were not altered much through the modification by mixed acid, whereas the micropores decreased remarkably by higher values of H2SO4/HNO<sup>3</sup> (*v*/*v*), especially in the ACF-R2 and ACF-R4 samples. This suggests that the higher content of H2SO<sup>4</sup> caused severe destruction of the microporous structures while the higher content of HNO<sup>3</sup> preserved the textual characteristics of the ACF sample.

istics of the ACF sample.

**Figure 2.** The pore size distribution (PSD) of ACF-0 and ACF-R4 samples obtained from the density functional theory method and N2 adsorption-desorption isotherms of ACF-0 sample. **Figure 2.** The pore size distribution (PSD) of ACF-0 and ACF-R4 samples obtained from the density functional theory method and N<sup>2</sup> adsorption-desorption isotherms of ACF-0 sample.

microporous structures while the higher content of HNO3 preserved the textual character-

**Table 1.** Textural parameters of the ACF samples before and after surface modification. **Table 1.** Textural parameters of the ACF samples before and after surface modification.


[a] Multipoint Braunauer-Emmett-Teller (BET). [b] Calculated by the t-plot method. [c] Estimated from the amounts of gas adsorbed at a relative pressure of 0.994. [d] Average pore diameter calcu-[a] Multipoint Braunauer-Emmett-Teller (BET). [b] Calculated by the t-plot method. [c] Estimated from the amounts of gas adsorbed at a relative pressure of 0.994. [d] Average pore diameter calculated from 2*V*Total/*S*BET for slit pore.

#### lated from 2*V*Total/*S*BET for slit pore. *2.2. Surface Properties*

*2.2. Surface Properties*  The Fourier transformation infrared (FTIR) spectra of ACF samples is shown in Figure 3a. According to the references, the peak at 1225 cm−1 is originated from the stretching mold of C–N and C–O in carboxylic anhydrides, ethers, lactones and phenols [42,43]. The peak at 1405 cm−1 and 1580 cm−1 is respectively related to the nitrogen groups and the double bond of C=C in quinoid structure. Meanwhile, the peak at 1730 cm−1 is owing to the stretching vibration of the C=O band in carboxyl and lactones groups attached to the aromatic rings, and the peak at 910 cm−1 is related to the anhydride groups [41,43,44]. After modification by different values of H2SO4/HNO3 (*v*/*v*), the intensities of the above peaks were wholly enhanced to different extents, suggesting the formation of large quantities of oxygen-containing species on ACF surface. For ACF-R1, ACF-R2 and ACF-R4 samples, the intensities of the peaks at 910 cm−1, 1225 cm−1, 1580 cm−1 and 1730 cm−1 were all greatly increased, indicating the enlargement in phenols, quinones, lactones, carboxyls and anhydrides. The highest peak intensity was found on ACF-R2 and ACF-R4 samples, especially at the position of 1730 cm−1, which confirms the further enrichment of anhydride and car-The Fourier transformation infrared (FTIR) spectra of ACF samples is shown in Figure 3a. According to the references, the peak at 1225 cm−<sup>1</sup> is originated from the stretching mold of C–N and C–O in carboxylic anhydrides, ethers, lactones and phenols [42,43]. The peak at 1405 cm−<sup>1</sup> and 1580 cm−<sup>1</sup> is respectively related to the nitrogen groups and the double bond of C=C in quinoid structure. Meanwhile, the peak at 1730 cm−<sup>1</sup> is owing to the stretching vibration of the C=O band in carboxyl and lactones groups attached to the aromatic rings, and the peak at 910 cm−<sup>1</sup> is related to the anhydride groups [41,43,44]. After modification by different values of H2SO4/HNO<sup>3</sup> (*v*/*v*), the intensities of the above peaks were wholly enhanced to different extents, suggesting the formation of large quantities of oxygen-containing species on ACF surface. For ACF-R1, ACF-R2 and ACF-R4 samples, the intensities of the peaks at 910 cm−<sup>1</sup> , 1225 cm−<sup>1</sup> , 1580 cm−<sup>1</sup> and 1730 cm−<sup>1</sup> were all greatly increased, indicating the enlargement in phenols, quinones, lactones, carboxyls and anhydrides. The highest peak intensity was found on ACF-R2 and ACF-R4 samples, especially at the position of 1730 cm−<sup>1</sup> , which confirms the further enrichment of anhydride and carboxylic groups by a higher content of H2SO<sup>4</sup> in mixed acid.

boxylic groups by a higher content of H2SO4 in mixed acid.

**Figure 3.** The (**a**) FTIR and (**b**) Raman spectra of ACF samples. **Figure 3.** The (**a**) FTIR and (**b**) Raman spectra of ACF samples.

As shown in Figure 3b, Raman spectroscopy detection was conducted to investigate the defects on carbon structure of ACF samples with surface modification. Usually, carbon fiber mainly has two characteristic peaks, one of which is the D peak at the position of 1350–1375 cm−1, and the other is the G peak at the position of 1580–1603 cm−1 [45]. The D peak is related to amorphous and defects of carbon structure while the G peak is related to graphite crystal structure. Generally, the calculation of *ID*/*IG* ratio from integral areas values of D and G peak was used to measure the structural defects of carbon materials [46]. It is widely known that *ID*/*IG* value increases with more structural defects generated on the carbon material. Obviously, the intensity of Raman spectra on ACF samples gradually increased, meanwhile the *ID*/*IG* values of all the ACF samples increased from 0.95 to 1.07 with increasing H2SO4/HNO3 (*v*/*v*) from 0.5 to 2. Whereas the *ID*/*IG* values of ACF-R4 decreased to 1.06 with increasing H2SO4/HNO3 (*v*/*v*) from 2 to 4. Therefore, it is believed that there were more surface defects and structural changes on the ACF carbon framework according to the surface modification with more H2SO4 contents in mixed acid. These results were well matched with the textural characteristics in ACF samples shown in Table 1. As shown in Figure 3b, Raman spectroscopy detection was conducted to investigate the defects on carbon structure of ACF samples with surface modification. Usually, carbon fiber mainly has two characteristic peaks, one of which is the D peak at the position of 1350–1375 cm−<sup>1</sup> , and the other is the G peak at the position of 1580–1603 cm−<sup>1</sup> [45]. The D peak is related to amorphous and defects of carbon structure while the G peak is related to graphite crystal structure. Generally, the calculation of *ID*/*I<sup>G</sup>* ratio from integral areas values of D and G peak was used to measure the structural defects of carbon materials [46]. It is widely known that *ID*/*I<sup>G</sup>* value increases with more structural defects generated on the carbon material. Obviously, the intensity of Raman spectra on ACF samples gradually increased, meanwhile the *ID*/*I<sup>G</sup>* values of all the ACF samples increased from 0.95 to 1.07 with increasing H2SO4/HNO<sup>3</sup> (*v*/*v*) from 0.5 to 2. Whereas the *ID*/*I<sup>G</sup>* values of ACF-R4 decreased to 1.06 with increasing H2SO4/HNO<sup>3</sup> (*v*/*v*) from 2 to 4. Therefore, it is believed that there were more surface defects and structural changes on the ACF carbon framework according to the surface modification with more H2SO<sup>4</sup> contents in mixed acid. These results were well matched with the textural characteristics in ACF samples shown in Table 1.

The narrow scan of C 1s regions in X-ray photoelectron spectroscopy (XPS) of ACF samples is exhibited in Figure 4a. Moreover, the deconvolution results of the C 1s spectrum are given in Table 2. For modified carbon materials, the C 1s spectra usually involved graphitic carbon (C–graphite, Peak I), ether, alcohol or phenolic groups (C–O, Peak II), carbonyl or quinone groups (C=O, Peak III), carboxylic groups (–COO–, Peak IV) and Peak V for the satellite peak from the π–π\* electron shake-up [47–49]. The intensities of peak I was decreased by the oxidation of mixed acid, whereas the intensities for the peaks attributed by C–O groups were increased [49]. However, the areas of peak II of all ACF samples by acid oxidation increased not so obviously compared to that of peak III or peak IV, which indicated phenolic groups may not be tailored much by adjusting different values of H2SO4/HNO3 (*v*/*v*). Similarly, the integral area of peak IV for both ACF-Rw (4.5%) and ACF-R1 sample (5.4%) was more than two times larger than that of the ACF-0 (2.0%). Notably, the area of peak IV increased to 7.9% and 8.7% on the ACF-R2 and ACF-R4 samples respectively, clearly confirming the generation of a large amount of surface carboxylic The narrow scan of C 1s regions in X-ray photoelectron spectroscopy (XPS) of ACF samples is exhibited in Figure 4a. Moreover, the deconvolution results of the C 1s spectrum are given in Table 2. For modified carbon materials, the C 1s spectra usually involved graphitic carbon (C–graphite, Peak I), ether, alcohol or phenolic groups (C–O, Peak II), carbonyl or quinone groups (C=O, Peak III), carboxylic groups (–COO–, Peak IV) and Peak V for the satellite peak from the π–π\* electron shake-up [47–49]. The intensities of peak I was decreased by the oxidation of mixed acid, whereas the intensities for the peaks attributed by C–O groups were increased [49]. However, the areas of peak II of all ACF samples by acid oxidation increased not so obviously compared to that of peak III or peak IV, which indicated phenolic groups may not be tailored much by adjusting different values of H2SO4/HNO<sup>3</sup> (*v*/*v*). Similarly, the integral area of peak IV for both ACF-Rw (4.5%) and ACF-R1 sample (5.4%) was more than two times larger than that of the ACF-0 (2.0%). Notably, the area of peak IV increased to 7.9% and 8.7% on the ACF-R2 and ACF-R4 samples respectively, clearly confirming the generation of a large amount of surface carboxylic groups by the higher content of H2SO4.

groups by the higher content of H2SO4.

**Figure 4.** High resolution of XPS spectra in (**a**) C 1s (**b**) O 1s regions of ACF samples. **Figure 4.** High resolution of XPS spectra in (**a**) C 1s (**b**) O 1s regions of ACF samples.

**Table 2.** Deconvolution results of the C 1s XPS spectra for the ACF samples, values given in % of total intensity. **Table 2.** Deconvolution results of the C 1s XPS spectra for the ACF samples, values given in % of total intensity.


Figure 4b exhibits the narrow scan of XPS spectra in O 1s regions of the ACF samples. Moreover, the deconvolution results of the O 1s spectrum are displayed in Table 3. As shown in Figure 4b, the O 1s XPS spectra can be deconvoluted into three main peaks, namely Peak I, Peak II and Peak III, which are associated with the C=O group, C―O group and adsorbed H2O or O2, respectively [42]. The adsorbed CO or CO2 in the ACF surface can be attributed to the minor Peak IV, the binding energy of which was at 536.9–537.0 eV. Obviously, the intensities of both Peak III and Peak IV decreased by surface modification, whereas the peaks corresponding to C=O groups increased evidently. As for Peak I, the intensities increased from 25.5 to 30.5% by surface modification with increasing the Figure 4b exhibits the narrow scan of XPS spectra in O 1s regions of the ACF samples. Moreover, the deconvolution results of the O 1s spectrum are displayed in Table 3. As shown in Figure 4b, the O 1s XPS spectra can be deconvoluted into three main peaks, namely Peak I, Peak II and Peak III, which are associated with the C=O group, C–O group and adsorbed H2O or O2, respectively [42]. The adsorbed CO or CO<sup>2</sup> in the ACF surface can be attributed to the minor Peak IV, the binding energy of which was at 536.9–537.0 eV. Obviously, the intensities of both Peak III and Peak IV decreased by surface modification, whereas the peaks corresponding to C=O groups increased evidently. As for Peak I, the intensities increased from 25.5 to 30.5% by surface modification with increasing the content of H2SO4, and similar results were obtained on ACF-R2 and ACF-R4 samples.

Additionally, the atomic ratio of surface O/C in the ACF samples by acid oxidation was enhanced significantly from 21.3 to 32.9% with increasing the H2SO4/HNO<sup>3</sup> (*v*/*v*) from 0.5 to 4. The above results suggested that more carboxylic species were generated by mixed acid oxidation with higher content of H2SO4, being consistent with the results of FTIR measurement.

**Table 3.** Deconvolution results of the O 1s XPS spectra for the ACF samples, values given in % of total intensity.


For the sake of examining the crucial role of the intrinsic nitrogen doped in ACF samples, the deconvolution results of N 1s XPS profiles of ACF samples are exhibited in Figure 5. Moreover, the corresponding results of the deconvolution are displayed in detail in Table 4. According to the curve fitting results and references, five distinct types of nitrogen contained species were deconvoluted from the N 1s spectra: NX (-NO2), N4 (pyridine-N oxide), NQ (quaternary N), N5 (pyrrolic/pyridone) and N6 (pyridine) [50–53]. It was evident that the content of N6 in ACF-0 was highest among all ACFs. Moreover, the content of N5 significantly increased to 33.2%, 43.4%, 50.5% and 46.3% for the ACF-Rw, ACF-R1, ACF-R2 and ACF-R4, respectively. As shown in Table 4, the content of N6 on the ACF-0 decreased from 15.6 to 5.2% corresponding to the ACF-R2. Moreover, the content of both NQ and N4 on the ACF sample decreased nearly one half by surface modification. It was reported that −NO<sup>2</sup> and pyridine were the main forms of the nitrogen introduced from HNO<sup>3</sup> oxidation and different forms of nitrogen can be transformed to each other [42,54]. The content of NX initially reached the maximum (32.3%) on ACF-Rw, then decreased to 22.3%, 15.3% and 19.0% on ACF-R1, ACF-R2 and ACF-R4, respectively. No content of NX can be observed on the ACF-0 sample without surface modification. Accordingly, when the nitrogen form predominated in the ACF sample was quaternary N, the mixed acid modification transformed them to −NO<sup>2</sup> with the higher content of HNO3. Meanwhile, more pyrrolic nitrogen species were generated by a higher content of H2SO4. In addition, the atomic ratios of surface N/C on ACF samples were gradually enhanced from 1.5 to 2.4 with the increase of H2SO4/HNO<sup>3</sup> (*v*/*v*) from 0.5 to 2, whereas that on ACF-R4 sample decreased to 2.0 by increasing the value of H2SO4/HNO<sup>3</sup> (*v*/*v*) to 4. These results suggest that the surface N-containing groups could be effectively tuned by mixed acid oxidation with different volume ratios of concentrated H2SO<sup>4</sup> and HNO3.

**Figure 5.** High resolution of XPS spectra in N 1s regions of ACF samples. **Figure 5.** High resolution of XPS spectra in N 1s regions of ACF samples.



ACF-R4 4.7 46.3 22.9 7.1 19.0 2.0

The temperature-programmed desorption (TPD) results of the ACF samples were shown in Figure 6. After being heated, carbon oxides were the main decomposition products of surface oxygen-containing functional groups [55–58]. As shown in Figure 7, the anhydrides and carboxylic acids usually decomposed into CO2 at relatively lower temperatures while the lactones decomposed into CO2 at higher temperatures. Meanwhile, the carboxylic anhydrides, ethers, phenols, carbonyl-quinones generally decomposed into CO [58]. Only little quantities of COx were obtained on the ACF-0 sample while significant quantities of COx were obtained on the other three ACF samples. For the ACF samples modified by mixed acid, the data of COx gradually rose with increasing the H2SO4/HNO3 (*v*/*v*) from 0.5 to 4. Especially, the desorption quantity of CO from the ACF-R1 sample was almost five-fold larger than that of the ACF-0 sample, illustrating the formation of large quantities of phenol and carbonyl-quinone groups. On the flip side, the desorption The temperature-programmed desorption (TPD) results of the ACF samples were shown in Figure 6. After being heated, carbon oxides were the main decomposition products of surface oxygen-containing functional groups [55–58]. As shown in Figure 7, the anhydrides and carboxylic acids usually decomposed into CO<sup>2</sup> at relatively lower temperatures while the lactones decomposed into CO<sup>2</sup> at higher temperatures. Meanwhile, the carboxylic anhydrides, ethers, phenols, carbonyl-quinones generally decomposed into CO [58]. Only little quantities of CO<sup>x</sup> were obtained on the ACF-0 sample while significant quantities of CO<sup>x</sup> were obtained on the other three ACF samples. For the ACF samples modified by mixed acid, the data of CO<sup>x</sup> gradually rose with increasing the H2SO4/HNO<sup>3</sup> (*v*/*v*) from 0.5 to 4. Especially, the desorption quantity of CO from the ACF-R1 sample was almost five-fold larger than that of the ACF-0 sample, illustrating the formation of large quantities of phenol and carbonyl-quinone groups. On the flip side, the desorption amount of CO<sup>2</sup> from the ACF-R2 sample was almost more than 15 times greater than the

amount of CO2 from the ACF-R2 sample was almost more than 15 times greater than the

ACF-0 sample, primarily owing to the remarkable generation of lactones, anhydrides and carboxylic acids. The quantities of CO<sup>x</sup> obtained from ACF-R4 were very similar to the ACF-R2 sample. ACF-0 sample, primarily owing to the remarkable generation of lactones, anhydrides and carboxylic acids. The quantities of COx obtained from ACF-R4 were very similar to the ACF-R2 sample. ACF-0 sample, primarily owing to the remarkable generation of lactones, anhydrides and carboxylic acids. The quantities of COx obtained from ACF-R4 were very similar to the ACF-R2 sample.

*Catalysts* **2021**, *11*, x FOR PEER REVIEW 9 of 18

**Figure 6.** TPD profiles of ACF samples before and after surface modification. **Figure 6.** TPD profiles of ACF samples before and after surface modification. **Figure 6.** TPD profiles of ACF samples before and after surface modification.

**Figure 7.** Deconvolution of the TPD profiles for ACF samples, in which peak a, peak b, peak c from the CO2 desorption of the carboxyl, anhydride, lactone groups while peak d, peak e, peak f from the CO desorption of the anhydride, phenol, carbonyl groups on ACF samples. **Figure 7.** Deconvolution of the TPD profiles for ACF samples, in which peak a, peak b, peak c from the CO2 desorption of the carboxyl, anhydride, lactone groups while peak d, peak e, peak f from the CO desorption of the anhydride, phenol, carbonyl groups on ACF samples. **Figure 7.** Deconvolution of the TPD profiles for ACF samples, in which peak a, peak b, peak c from the CO<sup>2</sup> desorption of the carboxyl, anhydride, lactone groups while peak d, peak e, peak f from the CO desorption of the anhydride, phenol, carbonyl groups on ACF samples.

Tables 5 and 6 show the detailed data of CO and CO<sup>2</sup> desorbed from specific surface groups on ACF samples. The desorption quantities of CO and CO<sup>2</sup> on ACF-0 sample were 244 µmol/g and 65 µmol/g, severally, and they were raised to 1008 µmol/g and 400 µmol/g on ACF-Rw sample. Upon increasing the H2SO4/HNO<sup>3</sup> (*v*/*v*) from 0.5 to 1, the desorption quantities of CO and CO<sup>2</sup> on the ACF-R1 sample remarkably raised to 1207 µmol/g and 678 µmol/g, respectively. Nevertheless, the amounts of CO desorbed from carbonyl-quinone groups on the ACF-R2 sample decreased to 125 µmol/g. While compared with ACF-Rw, more than two times larger amounts of CO<sup>2</sup> desorbed from carboxyl and anhydride groups were also found on the ACF-R2 sample. The desorption quantities of CO<sup>2</sup> and CO on the ACF-R4 sample were very similar to the ACF-R2 sample. Considering all these examinations, it could be deduced that the largest amounts of carboxyl (407 µmol/g) and anhydride (425 µmol/g) were obtained on the ACF-R2 and ACF-R4 samples while the most enrichment of phenol groups was detected on the ACF-R1 sample. This means that the moderate content of H2SO<sup>4</sup> produced more phenol groups while the higher content of H2SO<sup>4</sup> in the mixed acid created more carboxylic and anhydride groups. These results were consistent with the FTIR and XPS results of ACF samples.

**Table 5.** The desorption quantities of CO<sup>2</sup> from the ACF samples by the deconvolution of the TPD profiles.


Desorption temperatures: [a] 255–275 ◦C, [b] 430–451 ◦C, [c] 611–623 ◦C.

**Table 6.** The desorption quantities of CO from the ACF samples by the deconvolution of the TPD profiles.


Desorption temperatures: [d] 458–491 ◦C, [e] 630–660 ◦C, [f] 785–812 ◦C.
