Mechanothermal Approach for N-, S-, P-, and B-Doping of Carbon Nanotubes: Methodology and Catalytic Performance in Wet Air Oxidation

The texture and the surface chemistry of carbon nanotubes (CNTs) were modified using a solvent-free methodology involving a ball-milling mechanical treatment and thermal treatment under nitrogen in the presence of adequate precursors (melamine, sodium thiosulfate, sodium dihydrogen phosphate, and boric acid) of different heteroatoms (N, S, P, and B, respectively). The incorporation of these heteroatoms promotes significant changes in the pristine textural and chemical properties. This easy post-doping method allows the introduction of large concentrations of heteroatoms. Their effect on the catalytic activity of the materials was evaluated in the oxidation of oxalic acid by catalytic wet air oxidation (CWAO), as an alternative to the noble metal and rare earth oxide catalysts traditionally used in this process. Improved catalytic activities were obtained using the N-, P-, and B-doped CNTs in oxalic acid oxidation, while the S-doped CNT sample underperformed in comparison to the pristine material.


Introduction
Wet air oxidation (WAO) is an interesting process to treat effluents with an organic concentration too high for biological treatments and simultaneously too low for incineration [1,2]. The non-catalytic process requires high temperatures (200-320 • C) and pressures (20-200 bar) to warrant solubility of oxygen and to improve mineralization and degradation rates [3][4][5]; however, in the last decades, several efforts have been made to decrease these severe operating conditions by introduction of homogeneous or heterogeneous catalysts, mostly based on the use of noble metals and metal oxides [6][7][8]. Several metal oxides (e.g., Cu, Zn, Mn, Fe, Co, and Bi) and noble metals (e.g., Ru, Pt, Pd, and Rh) have been investigated for CWAO [9][10][11][12][13][14][15][16][17]; however, deactivation phenomena are frequent, in particular the leaching of active metals to the liquid phase, which severely limits the efficiency of these heterogeneous catalytic processes. In addition, noble metals and rare earth oxides are facing an increasing price due to their scarcity, and they also cause harmful impacts on the environment. In this context, the use of metal-free carbon materials as catalysts has been recently preferred for many CWAO applications [6,[18][19][20][21][22][23][24][25]. The use of metal-free carbon materials can solve the problems associated with metal leaching to the liquid phase [22,24,26]. In fact, carbon materials are assuming an increasing role as catalysts in the fields of energy conversion/storage and environmental technologies. Wide

Materials
Commercial multiwalled carbon nanotubes NC3100 (Nanocyl TM , Sambreville, Belgium) were used as raw material (CNT-O). Pristine CNT-O were ball milled with different solid precursors in grinding jars using zirconium oxide balls in a Retsch MM200 equipment (Retsch GmbH, Haan, Germany) without any gas flow using the milling conditions determined in a previous work [46]. A constant vibration frequency of 15 vibrations/s during 4 h was applied for all ball milled materials. Sodium thiosulfate was evaluated in the S-doping and sodium dihydrogen phosphate and boric acid were tested as P and B precursors, respectively. N-doping was performed using melamine, repeating a previously reported methodology [38]. The doped samples were obtained by mixing 0.6 g of CNT with 0.26 g of N, S, P, or B using the suitable precursor and labelled as indicated in Table 1. After mechanical mixing, the samples were subjected to a thermal treatment to promote the decomposition of the heteroatom precursor at the temperature indicated in Table 1.

Characterization Techniques
Pristine and modified CNTs were characterized to assess the main changes in their textural and surface properties promoted by the mechanothermal treatment. The textural characterization of the materials was based on the N 2 adsorption-desorption isotherms, determined at −196 • C in a Quantachrome NOVA 4200e multi-station apparatus (Quantachrome Instruments, Boynton Beach, FL, USA). The C/H/N/S and O composition of the modified samples was determined by Elemental Analysis (EA) carried out on, respectively, vario MICRO cube and rapid OXY cube analyzers from Elemental GmbH, Langenselbold, Germany. X-ray Photoelectron Spectroscopy (XPS) has been used to assess the carbon surface composition. Analyses were performed using a Kratos AXIS Ultra HSA (Manchester, UK), with VISION software for data acquisition, CASAXPS software for data analysis. Thermogravimetric analysis (TGA) was carried out using an STA 409 PC/4/H Luxx Netzsch thermal analyzer (Selb, Germany). The morphology was observed by SEM/EDXS analysis using a FEI Quanta 400FEG ESEM/EDAX Genesis X4M instrument (Hillsboro, OR, USA). Additional details can be found elsewhere [47].

Catalytic Wet Air Oxidation (CWAO) Experimental Procedure
Experiments were carried out in a 160 mL 316-stainless steel high-pressure batch reactor Mod. 4564 (Parr Instruments, Moline, IL, USA), equipped with a stirring and temperature controller PARR 4842. Detailed reactor description may be found elsewhere [48]. Briefly, 75 mL of an oxalic acid aqueous solution (1000 mg L −1 ) and 0.2 g of catalyst were placed into the reactor. The reactor was flushed with pure nitrogen until the complete removal of oxygen, and then pressurized with 5 bar of nitrogen and pre-heated up to 140 • C under continuous stirring (500 rpm). When the desired temperature was reached, pure air was injected to obtain a total pressure of 40 bar inside the reactor (corresponding to 7 bar of O 2 partial pressure), this being considered time zero for the reaction (t = 0 min). A non-catalytic oxidation run was carried out with air in the absence of catalyst, experiment denoted as WAO experiment. Reproducibility tests showed relative errors lower than ±5%.

Catalytic Wet Air Oxidation (CWAO) Analytical Techniques
Liquid aliquots (around 1 mL) were withdrawn from the reactor during reaction and centrifuged for further analysis. The samples were analysed by high-performance liquid chromatography (HPLC) with a Hitachi Elite LaChrom system equipped with a Diode Array Detector (L-2450). An Alltech OA-1000 column (300 mm × 6.5 mm) working at room temperature in isocratic elution mode with a H 2 SO 4 solution (5 mmol L −1 ) at a flow rate of 0.6 cm 3 min −1 (as mobile phase) was used for oxalic acid identification and quantification (performed at λ = 210 nm). Analyses were made using an injection volume of 15 µL. Calibration curves (5-1200 mg L −1 ) were made and linear responses were obtained in this range (R 2 of calibration lines higher than 0.999). All solutions and standard compounds were prepared with ultrapure water with a resistivity of 18.2 mΩ at room temperature, obtained from a Milllipore Mili-Q system.

Materials Characterization
Textural properties of the ball milled CNT samples were evaluated by the N 2 adsorption-desorption isotherms performed at −196 • C, in order to evaluate the changes promoted by the mechanothermal treatment with the different heteroatoms. Figure 1 shows that N 2 adsorption isotherms of all samples are of type II, accordingly to IUPAC classification [49]. For all samples the adsorption and desorption branches are not coincident, with a hysteresis loop being observed. The enlargement of the hysteresis is due to two contributions in the mesoporous character: the entanglement of nanotubes and the access to the inner cavity of the tubes [50]. The N 2 adsorption isotherms at low relative pressure are very similar for the CNT-O, CNT-N, and CNT-S samples, in agreement with the small differences in the surface areas (Table 2); however, some differences were observed in the amount of N 2 adsorbed at higher relative pressures ( Figure 1). Liquid aliquots (around 1 mL) were withdrawn from the reactor during reaction and centrifuged for further analysis. The samples were analysed by high-performance liquid chromatography (HPLC) with a Hitachi Elite LaChrom system equipped with a Diode Array Detector (L-2450). An Alltech OA-1000 column (300 mm × 6.5 mm) working at room temperature in isocratic elution mode with a H2SO4 solution (5 mmol L −1 ) at a flow rate of 0.6 cm 3 min −1 (as mobile phase) was used for oxalic acid identification and quantification (performed at λ = 210 nm). Analyses were made using an injection volume of 15 μL. Calibration curves (5-1200 mg L −1 ) were made and linear responses were obtained in this range (R 2 of calibration lines higher than 0.999). All solutions and standard compounds were prepared with ultrapure water with a resistivity of 18.2 mΩ at room temperature, obtained from a Milllipore Mili-Q system.

Materials Characterization
Textural properties of the ball milled CNT samples were evaluated by the N2 adsorptiondesorption isotherms performed at −196 °C, in order to evaluate the changes promoted by the mechanothermal treatment with the different heteroatoms. Figure 1 shows that N2 adsorption isotherms of all samples are of type II, accordingly to IUPAC classification [49]. For all samples the adsorption and desorption branches are not coincident, with a hysteresis loop being observed. The enlargement of the hysteresis is due to two contributions in the mesoporous character: the entanglement of nanotubes and the access to the inner cavity of the tubes [50]. The N2 adsorption isotherms at low relative pressure are very similar for the CNT-O, CNT-N, and CNT-S samples, in agreement with the small differences in the surface areas (Table 2); however, some differences were observed in the amount of N2 adsorbed at higher relative pressures ( Figure 1). The introduction of the precursors during the mechanical treatment promoted significant changes in the nanotube structures leading to the increase/decrease of the accessible surface area for N2 adsorption and decreasing the free-space between the tubes (evaluated by the total pore volume ( Table 2)). The mechanical step reduced the entanglement of the CNTs, also leading to shorter CNTs by breaking up the tubes without affecting the tube diameters [46], therefore increasing the surface area of the carbon nanotubes. On the other hand, the introduction of N, S, P, and B-groups on the defects of the CNTs may block the access of N2 to the inner cavities and the presence of surface groups may also play an attractive effect between the tubes, leading to higher agglomeration of the material The introduction of the precursors during the mechanical treatment promoted significant changes in the nanotube structures leading to the increase/decrease of the accessible surface area for N 2 adsorption and decreasing the free-space between the tubes (evaluated by the total pore volume ( Table 2)). The mechanical step reduced the entanglement of the CNTs, also leading to shorter CNTs by breaking up the tubes without affecting the tube diameters [46], therefore increasing the surface area of the carbon nanotubes. On the other hand, the introduction of N, S, P, and B-groups on the defects of the CNTs may block the access of N 2 to the inner cavities and the presence of surface groups may also play an attractive effect between the tubes, leading to higher agglomeration of the material [38]. The addition of the nitrogen precursor leads to a small increase of the surface area comparing to the original sample (192 m 2 /g); however, the addition of all the other precursors leads to a decrease of the surface area, this effect being more noticeable for samples CNT-B and CNT-P (45 and 93 m 2 /g, respectively). This decrease is in line with the reduction in the carbon amount measured by elemental analysis (Table 3). The pore size distributions of all the samples, obtained by Non-Linear Density Function Theory (NLDFT) calculations, are presented in Figure 2. The ball milling treatment significantly modifies the pore size distribution. The pristine CNTs present a high number of large pores with a pore width larger than 16 nm, whereas for the ball milled samples the number of larger pores is low. This change is mainly related to the ball milling treatment, since a significant reduction of the number of larger pores occurs when the pristine CNTs are ball milled without the presence of heteroatoms precursors [38]. However, sample CNT-B practically does not present pores with widths lower than 14 nm, but there is a significant number of larger pores, in agreement with the small surface area observed.
C 2019, 5, x 5 of 15 [38]. The addition of the nitrogen precursor leads to a small increase of the surface area comparing to the original sample (192 m 2 /g); however, the addition of all the other precursors leads to a decrease of the surface area, this effect being more noticeable for samples CNT-B and CNT-P (45 and 93 m 2 /g, respectively). This decrease is in line with the reduction in the carbon amount measured by elemental analysis ( Table 3). The pore size distributions of all the samples, obtained by Non-Linear Density Function Theory (NLDFT) calculations, are presented in Figure 2. The ball milling treatment significantly modifies the pore size distribution. The pristine CNTs present a high number of large pores with a pore width larger than 16 nm, whereas for the ball milled samples the number of larger pores is low. This change is mainly related to the ball milling treatment, since a significant reduction of the number of larger pores occurs when the pristine CNTs are ball milled without the presence of heteroatoms precursors [38]. However, sample CNT-B practically does not present pores with widths lower than 14 nm, but there is a significant number of larger pores, in agreement with the small surface area observed.   Table 3. Bulk characterization by elemental analysis (EA) and surface chemistry characterization by X-ray photoelectron spectroscopy (XPS) and thermogravimetric (TGA) analyses of pristine and mechanothermally treated carbon nanotubes (CNT) samples. TGA analyses were performed in order to evaluate the thermal decomposition of the heteroatom precursors, to define the final thermal treatment temperature and to assess the functionalization degree of the samples. Figure 3 shows the TGA profiles of the samples mechanically treated with the  TGA analyses were performed in order to evaluate the thermal decomposition of the heteroatom precursors, to define the final thermal treatment temperature and to assess the functionalization degree of the samples. Figure 3 shows the TGA profiles of the samples mechanically treated with the selected precursors, before and after the thermal treatment. A detailed study with sample CNT-N was reported in a previous work [38]. Briefly, the thermal decomposition of melamine generates a significant weight loss (34%) between 250 and 350 • C, mainly due to the release of N-volatile compounds, followed by decomposition of the main chain and additional release of nitrogen volatiles up to 900 • C (less than 7%), the total weight loss being close to the amount of melamine mixed with the CNTs [51]. After the thermal treatment at 600 • C, the total weight loss in TGA until 900 • C is lower than 9%, suggesting that the precursor has been previously decomposed. Regarding the TGA profile of sample CNT-S before thermal treatment, it is observed that sodium thiosulfate is poorly decomposed until 700 • C; however, almost 50% of degradation is observed until 900 • C, which corresponds to the weight percentage on the physical mixture CNT/precursor. For this reason, 900 • C was chosen as the final temperature for the thermal treatment. Although the CNT-S sample had been treated at 900 • C, the TGA profile after the thermal treatment still shows around 15% of weight loss between 700 and 900 • C. The thermal degradation of sodium thiosulfate can form Na 2 SO 3 and S, which in the presence of oxygen may form Na 2 SO 4 and SO 2 [52,53]. Therefore, it is expected that this sample would present some reactivity to oxygen after the thermal treatment and, if exposed to room conditions, it could explain its weight loss in the TGA at high temperatures. Sample CNT-P before the thermal treatment only presents a weight loss near 10% in the TGA temperature range studied. Two steps of decomposition can be observed around 200 and 250 • C, probably due to the dehydration of sodium dihydrogen phosphate [54]. The sample remains stable after thermal treatment at 300 • C, without additional weight loss in TGA. The thermal stability of the resulting residue suggests that phosphate could have been dehydrated to polyphosphates or metaphosphates [55]. In the TGA profiles of the CNT-B sample before and after thermal treatment the largest weight loss occurred until 200 • C, which is the reason why this temperature was chosen as the final temperature during the thermal treatment. Dehydration of boric acid is expected to occur above 150 • C, leading to the formation of metaboric acid that can be further transformed to boron oxide (B 2 O 3 ) [56,57].
The mechanical treatment does not allow any functionalization of the carbon surface, as already reported in previous works [46,47]. It is during the thermal treatments that the decomposition of the precursors occurs, promoting the enrichment of the carbon surface with heteroatoms. Table 3 shows both bulk and surface characterization of the CNT samples after thermal treatments. Regarding the bulk composition in terms of CNHS, pristine CNTs revealed high purity with a carbon content of 97.5% and an oxygen content of 1.2%. With the addition of the heteroatoms, the raw composition suffers from small to large changes, depending on the precursor used. The melamine treated sample only revealed the incorporation of 3.5% of nitrogen onto the CNT surface, as determined by EA. A strong decrease of the carbon content is observed in all other samples (around 50%, being even higher in the case of the CNT-B sample). However, these decreases are in line with the introduction of other elements as S/P/B/O and Na in the cases where sodium thiosulfate and sodium dihydrogen phosphate were used as precursors. EA revealed the incorporation of 9.2% of sulphur in the sample CNT-S, along with a significant increase of oxygen (26.5%). The remaining composition (around 10%) can be attributed to sodium from the precursor. These results are in agreement with the surface composition determined by XPS, where 10.4% of sodium has been detected. Considering that the initial weight percentage of S on the physical mixture was 20.9%, around 50% of the sulphur was kept on the carbon nanostructure by the applied approach. In the case of the CNT-P sample, bulk composition revealed the incorporation of a large amount of oxygen, accompanied by the introduction of 4.5% of P and also the retention of sodium. The most drastic change occurred with the sample CNT-B, where a large presence of B (18.7%) accompanied by near 50% of oxygen (XPS) was observed. without additional weight loss in TGA. The thermal stability of the resulting residue suggests that phosphate could have been dehydrated to polyphosphates or metaphosphates [55]. In the TGA profiles of the CNT-B sample before and after thermal treatment the largest weight loss occurred until 200 °C, which is the reason why this temperature was chosen as the final temperature during the thermal treatment. Dehydration of boric acid is expected to occur above 150 °C, leading to the formation of metaboric acid that can be further transformed to boron oxide (B2O3) [56,57].  The nature of the N, S, P, and B-functionalities incorporated with the mechanothermal treatments was investigated using the corresponding spectra obtained by XPS (Figure 4). The asymmetric N1s spectrum of the sample CNT-N was deconvoluted in three peaks: quaternary nitrogen (NQ, B.E. = 401.4 ± 0.3 eV; 20%), pyrrole (N5, B.E. = 400.1 ± 0.2 eV, 35%), and pyridinic (N6, B.E. = 398.6 ± 0.3 eV; 45%) structures, which are typically formed in carbon samples prepared above 550 • C [58]. Assumptions regarding the N1s spectrum fitting may be found elsewhere [38]. The S2p spectrum recorded for sample CNT-S was deconvoluted assuming that each of the 4 components has a S 2p 3/2 and 2p 1/2 doublet separated by 1.18 eV and with an intensity ratio of 2:1. The fitted S 2p 3/2 peaks were ascribed to Na 2 S 2 O 3 (S1, B.E. = 162.4 eV), elemental S (S2, B.E. = 164.1 eV), Na 2 SO 3 (S3, B.E. = 167.1 eV), and Na 2 SO 4 (S4, B.E. = 169.1 eV) [59]. The weak S1 component suggests that a small amount of precursor was not decomposed; however, the presence of the other three components suggests that the thermal degradation of sodium thiosulfate led to the formation of Na 2 SO 3 and S. In addition to the TGA results, the presence of the Na 2 SO 4 species suggested that, even in the absence of oxygen during the thermal treatment, the sample became very reactive to oxygen, showing that some Na 2 SO 3 may be converted into Na 2 SO 4 [52,53]. The P 2p spectrum of sample CNT-P has also been deconvoluted considering the P 2p 3/2 and 2p 1/2 doublet separated by 0.84 eV, with an intensity ratio of 2:1. A single peak was found at 134.6 eV, which is commonly attributed to phosphorous in the form of metaphosphates (PO 3 − ) [59]. The absence of a peak corresponding to the binding energy at 133.2 eV (typical of phosphate, (PO 4 ) 3− confirms that phosphate has suffered a thermal dehydration to metaphosphates, in agreement with the TGA results [54,55]. The surface morphology of the doped samples was assessed by SEM/EDS and compared with the pristine materials to confirm the enrichment of the carbon surface with heteroatoms. As can be seen in Figure 5, sample CNT-O displays highly entangled and twisted tubes. This entanglement was noticeably reduced after the ball-milling treatment, for all doped samples. Indeed, in previous work [46], it was observed (by TEM) that the ball-milling treatment is highly effective in disentangling and shortening the CNTs. Figure 5 shows that sample CNT-N presents a morphology similar to that of CNT-O, the main difference being the reduction in the entanglement. EDS analysis confirms the presence of N-groups on the surface of the CNTs ( Figure 5). The other doped samples exhibit a heterogeneous morphology, with regions where the heteroatom compounds have been deposited on the surface, identified by the different zones (Zi). Sample CNT-P presents a more homogenous dispersion of the heteroatoms than CNT-S, but in both it is evident the existence of particles with planar faces suggest the presence of P and S in the form of metaphosphates (PO3 − ) and Na2SO3, respectively, since the EDS analysis reveals that the increase of each heteroatom is always accompanied by a corresponding increase in the amount of oxygen. These findings confirm the XPS and TGA results. Sample CNT-B presents a distinct morphology, in which the tubes seem to be in the form of dreadlocks. This observation could explain the low surface area of this sample (Table 1) due to the agglomeration of the tubes. The presence of boron in the form of boron oxide is confirmed in the surface of the nanotubes by the EDS analysis since oxygen is present in high amount when boron is identified. The surface morphology of the doped samples was assessed by SEM/EDS and compared with the pristine materials to confirm the enrichment of the carbon surface with heteroatoms. As can be seen in Figure 5, sample CNT-O displays highly entangled and twisted tubes. This entanglement was noticeably reduced after the ball-milling treatment, for all doped samples. Indeed, in previous work [46], it was observed (by TEM) that the ball-milling treatment is highly effective in disentangling and shortening the CNTs. Figure 5 shows that sample CNT-N presents a morphology similar to that of CNT-O, the main difference being the reduction in the entanglement. EDS analysis confirms the presence of N-groups on the surface of the CNTs ( Figure 5). The other doped samples exhibit a heterogeneous morphology, with regions where the heteroatom compounds have been deposited on the surface, identified by the different zones (Z i ). Sample CNT-P presents a more homogenous dispersion of the heteroatoms than CNT-S, but in both it is evident the existence of particles with planar faces suggest the presence of P and S in the form of metaphosphates (PO 3 − ) and Na 2 SO 3 , respectively, since the EDS analysis reveals that the increase of each heteroatom is always accompanied by a corresponding increase in the amount of oxygen. These findings confirm the XPS and TGA results. Sample CNT-B presents a distinct morphology, in which the tubes seem to be in the form of dreadlocks. This observation could explain the low surface area of this sample (Table 1) due to the agglomeration of the tubes. The presence of boron in the form of boron oxide is confirmed in the surface of the nanotubes by the EDS analysis since oxygen is present in high amount when boron is identified.

Catalytic Activity in Catalytic Wet Air Oxidation (CWAO)
The carbon materials doped with different heteroatoms were selected for catalytic tests in the oxidation of oxalic acid by CWAO. The materials were chosen with the aim to clarify their potential use as metal-free catalysts in the degradation of organic compounds by CWAO. Moreover, catalyst evaluation was outlined with the ambition to clarify the more adequate properties of carbon materials in the selected application. Oxalic acid was selected as a model compound to evaluate the process efficiency. It is representative of short chain carboxylic acids, which are typical refractory compounds appearing as end-products in the advanced oxidation processes. Figure 5 shows that oxalic acid is poorly oxidized under temperature and pressure employed in the absence of a catalyst, while CNT materials show high catalytic performance, with almost complete degradation of the model compound after 120 min of reaction, regardless of the sample tested (pristine or doped).

Catalytic Activity in Catalytic Wet Air Oxidation (CWAO)
The carbon materials doped with different heteroatoms were selected for catalytic tests in the oxidation of oxalic acid by CWAO. The materials were chosen with the aim to clarify their potential use as metal-free catalysts in the degradation of organic compounds by CWAO. Moreover, catalyst evaluation was outlined with the ambition to clarify the more adequate properties of carbon materials in the selected application. Oxalic acid was selected as a model compound to evaluate the process efficiency. It is representative of short chain carboxylic acids, which are typical refractory compounds appearing as end-products in the advanced oxidation processes. Figure 6 shows that oxalic acid is poorly oxidized under temperature and pressure employed in the absence of a catalyst, while CNT materials show high catalytic performance, with almost complete degradation of the model compound after 120 min of reaction, regardless of the sample tested (pristine or doped). Among the materials tested, the best performance was obtained with the sample CNT-N, which oxidized more than 85% of the initial oxalic acid concentration in 30 min. For this same period, the CNT-B and CNT-P samples achieved 62% and 51% of oxalic acid degradation, respectively, while only 27% and 22% were obtained with CNT-O and CNT-S, respectively. These results indicate that incorporation of N, P, and B heteroatoms onto the commercial CNTs improved the intrinsic catalytic activity of the pristine materials, leading to a faster degradation of oxalic acid. In contrast, CNT-S exhibited the worst performance among the CNTs tested.
It is well accepted in the literature that the catalytic activity of carbon materials in oxidation reactions can be enhanced by adjustment of their textural and chemical properties [31]. Regarding the textural properties of the tested samples, no correlation was found between the catalytic activity and the surface area of these materials ( Table 2). In fact, the doped samples with the highest surface areas and pore volume (CNT-N and CNT-S), presented the most opposite performances in the oxalic acid oxidation, while the samples with the lowest SBET (CNT-B and CNT-P) showed considerable catalytic activity. Hereupon, the heteroatoms present in the carbon sample must play an important role in the oxidation of oxalic acid. Despite all the reported studies regarding the use of carbon materials as catalysts in CWAO, the role of the surface chemistry in this reaction is still under debate. While the role of O-, N-, and S-containing groups have been extensively reported as alternative to the noble metal and rare earth oxide catalysts traditionally used in this process [6,28,31,[60][61][62][63][64][65][66][67][68][69][70][71], little attention has been given to the study of P and B in CWAO studies, contrary to what has happened in the field of electrochemistry [40,[72][73][74]. The catalytic activity enhancement in the presence of Nfunctionalities on CNTs was previously explained by the improvement of the electron interaction between carbon surface and oxygen [47]. Oxalic acid degradation pathway by CWAO is accepted to involve the adsorption of oxygen and oxalic acid on the carbon surface, oxygen being activated into reactive species that react with the adsorbed oxalic acid to produce carbon dioxide and water [48]. In the presence of N-groups with unpaired electrons, such as the pyridinic groups, there is an electron donation effect which may enhance the adsorption of oxygen, improving the surface oxidation mechanism [48]. Independently of the larger (nitrogen) or smaller (boron, phosphorous) electronegativity than carbon, it is believed that these heteroatoms break the electroneutrality of the graphitic structure [75]. Either the high electron-donating ability of P [75] and the intrinsic electron accepting nature of B, may be translated into higher affinity to oxygen as compared to the pristine CNTs, improving the adsorption of oxygen [75] and the formation of the oxygen reactive species, which is expected to enhance the oxidation of oxalic acid [76]. A positive role of P-, B-, and Nheteroatoms in carbon blacks was also found by Diaz de Tuesta et al. [77] in a recent work on the catalytic wet peroxide oxidation of phenol. The doped-carbons yielded values of the kinetic rate Among the materials tested, the best performance was obtained with the sample CNT-N, which oxidized more than 85% of the initial oxalic acid concentration in 30 min. For this same period, the CNT-B and CNT-P samples achieved 62% and 51% of oxalic acid degradation, respectively, while only 27% and 22% were obtained with CNT-O and CNT-S, respectively. These results indicate that incorporation of N, P, and B heteroatoms onto the commercial CNTs improved the intrinsic catalytic activity of the pristine materials, leading to a faster degradation of oxalic acid. In contrast, CNT-S exhibited the worst performance among the CNTs tested.
It is well accepted in the literature that the catalytic activity of carbon materials in oxidation reactions can be enhanced by adjustment of their textural and chemical properties [31]. Regarding the textural properties of the tested samples, no correlation was found between the catalytic activity and the surface area of these materials (Table 2). In fact, the doped samples with the highest surface areas and pore volume (CNT-N and CNT-S), presented the most opposite performances in the oxalic acid oxidation, while the samples with the lowest S BET (CNT-B and CNT-P) showed considerable catalytic activity. Hereupon, the heteroatoms present in the carbon sample must play an important role in the oxidation of oxalic acid. Despite all the reported studies regarding the use of carbon materials as catalysts in CWAO, the role of the surface chemistry in this reaction is still under debate. While the role of O-, N-, and S-containing groups have been extensively reported as alternative to the noble metal and rare earth oxide catalysts traditionally used in this process [6,28,31,[60][61][62][63][64][65][66][67][68][69][70][71], little attention has been given to the study of P and B in CWAO studies, contrary to what has happened in the field of electrochemistry [40,[72][73][74]. The catalytic activity enhancement in the presence of N-functionalities on CNTs was previously explained by the improvement of the electron interaction between carbon surface and oxygen [47]. Oxalic acid degradation pathway by CWAO is accepted to involve the adsorption of oxygen and oxalic acid on the carbon surface, oxygen being activated into reactive species that react with the adsorbed oxalic acid to produce carbon dioxide and water [48]. In the presence of N-groups with unpaired electrons, such as the pyridinic groups, there is an electron donation effect which may enhance the adsorption of oxygen, improving the surface oxidation mechanism [48]. Independently of the larger (nitrogen) or smaller (boron, phosphorous) electronegativity than carbon, it is believed that these heteroatoms break the electroneutrality of the graphitic structure [75]. Either the high electron-donating ability of P [75] and the intrinsic electron accepting nature of B, may be translated into higher affinity to oxygen as compared to the pristine CNTs, improving the adsorption of oxygen [75] and the formation of the oxygen reactive species, which is expected to enhance the oxidation of oxalic acid [76]. A positive role of P-, B-, and N-heteroatoms in carbon blacks was also found by Diaz de Tuesta et al. [77] in a recent work on the catalytic wet peroxide oxidation of phenol. The doped-carbons yielded values of the kinetic rate constants of hydrogen peroxide degradation up to eight times larger than that obtained with the original carbon.

Conclusions
In the search for metal-free catalysts as cost-effective and eco-sustainable materials for replacing metal-based catalysts usually used in the CWAO process, the surface chemistry of carbon nanotubes was modified using a solvent-free methodology involving mechanical and thermal treatments in the presence of precursors of different heteroatoms (N, S, B, P). Although the methodology allows to incorporate different functional groups and large amounts of each heteroatom, in some cases it also promotes significant changes in the textural properties. An outstanding extent of oxidation of oxalic acid was achieved in the presence of the N-doped sample, followed by B-and P-doped samples, improving the intrinsic catalytic activity of the pristine material, whereas S-doped sample exhibited the worst performance. No correlation between the catalytic activity and the surface area of these materials was found, suggesting that the heteroatom incorporated into the carbon sample plays an important role in the degradation of oxalic acid, probably due to the high electron-donating ability of P and the intrinsic electron accepting nature of B.