Effect of benzophenone on the physicochemical properties of N-CNTs synthesized from 1-ferrocenylmethyl (2-methylimidazole) catalyst

Vertically-aligned nitrogen-doped carbon nanotubes (v-N-CNTs) were synthesized via the chemical vapour deposition (CVD) technique. 1ferrocenylmethyl(2-methylimidazole) was employed as the source of the Fe catalyst and was dissolved in different ratios of acetonitrile/benzophenone feedstock which served as both the carbon, nitrogen, and oxygen sources. The morphological difference in N-CNTs was as a result of increased oxygen concentration in the reaction mix and not due to water vapour formation as observed in the oxygen-free experiment, indicating specifically, the impact of oxygen. Raman and X-ray photoelectron spectroscopy (XPS) revealed surface defects and grafting of oxygen functional groups on the sidewall of N-CNTs. The FTIR data showed little or no effect as oxygen concentration increases. XPS analysis detected the type of nitrogen species (i.e. pyridinic, pyrrolic, graphitic, or molecular nitrogen forms) incorporated in the N-CNT samples. Pyrrolic nitrogen was dominant and increased (from 8.6 to 11.8 at.%) as oxygen concentration increases in the reaction precursor. An increase in N content was observed with the introduction of a lower concentration of oxygen, followed by a gradual decrease at higher oxygen concentration. Our result suggested that effective control of the reactant mixtures can manipulate the morphology of N-CNTs. DOI:10.46481/jnsps.2020.105


Introduction
Vertically-aligned carbon nanotubes (v-CNTs) have been found to be fascinating for various range of applications, such as catalysis (as catalyst support) [1,2,3], electronics [4,5,7] with heteroatoms, such as B, P, S and N, into the sp 2 carbon framework has been reported [13,14,15,16]. These electronrich atoms help fine-tune the electronic properties of CNTs [2]. Also, nitrogen incorporation into CNTs alters the wall thickness, crystallinity and diameter of CNTs [17]. The nitrogen embedded into CNTs can take various forms. The most common are graphitic-nitrogen, pyrrolic-nitrogen, pyridinic-nitrogen and molecular N 2 stuck in the interior of CNT structures [18,19,20]. The nitrogen composition largely depends on the solubility of nitrogen in the catalyst nanoparticle during the reaction at a specified temperature. It has been shown that the type and concentration of nitrogen obtained depend on the nature of the catalyst employed (i.e. ferrocene or ferrocenyl derivatives), synthetic temperature, gas flow rate and type of nitrogencontaining precursors [21,22,23].
Several methods have been employed in the synthesis of N-CNTs; namely arc-discharge [24], laser deposition [25] and chemical vapour deposition (CVD) [26]. Of these, the CVD technique has been the method commonly used for large-scale N-CNT synthesis [27]. However, the control of the reaction conditions in CVD technique is somewhat tricky, as catalyst poisoning due to limiting carbon diffusion rate and formation of amorphous carbon on Fe substrate surface is common [28,29]. Many researchers have reported the introduction of oxygen [30], water [31,32] and CO 2 [33], ethyl benzoate [34] among other reaction gases to improve N-CNTs quality and catalyst activity [35]. However, an excess level of oxygen-containing species could lead to N-CNTs etching [29,32,36]. Recently, Sakurai et al. [37] reported that the introduction of the oxygencontaining molecule (e.g. H 2 O) during CVD synthesis enhanced the growth of CNTs and prolong catalyst lifetime at temperatures above 750 • C. This resulted in the removal of amorphous carbon through water vapour etching to give a graphitic nanostructured carbon network [26]. Fatuba et al. [38] also reported that the addition of oxygen-containing aromatic compounds (i.e. growth enhancer), such as methyl benzoate and benzaldehyde into the reaction mixture tailored the size and controlled the N-CNTs wall numbers and alignment [38]. The essential role of the growth enhancer compared to previously reported approach (such as H 2 O), is to control the wall numbers, diameters and to reactivate catalyst particles [39,40,41].
In this study, we report for the first time, the use of benzophenone in the reactant mixture to modify N-CNTs growth and morphology. Benzophenone was also employed to improve the solubility of the ferrocenyl imidazolium catalyst in acetonitrile. We elucidate the effect of oxygen on the type of nitrogen incorporated in N-CNTs. The morphology, surface area and stability of N-CNTs were studied at varying oxygen concentration levels.

Materials and characterization
Ferrocene (≥ 97%), ferrocenemethanol (98%), 2-methylimidazole (≥ 98.2%) and sodium borohydride (95%), potassium hydrogen phthalate (≥ 99.5%) were obtained from Sigma Aldrich Ltd. South Africa. Acetonitrile (HPLC grade, 99.9%), toluene (≥99.5%) and ethanol (98%) were purchased from Merck Chemicals South Africa. Nitric acid (55%) and sulphuric acid (98%) were purchased from Saarchem, South Africa. 10% Hydrogen in argon (purchased from AFROX gases, South Africa) was used as a carrier gas for the synthesis of N-CNTs. Images of the synthesized N-CNTs were obtained by using scanning electron microscopy (SEM) (JOEL JEM 1010) and transmission electron microscopy (TEM) (JOEL JSM 6100). Higher magnification images of N-CNTs were obtained from high-resolution transmission electron microscope (HRTEM). Elemental analysis was conducted on a LECO CHNS elemental analyser. The crystallinity of the N-CNTs was determined with a Rigaku/D max RB powder X-ray diffractometer using graphite monochromatized high-density Cu Kα radiation (λ= 0.15406). The thermal stabilities of N-CNTs were determined using a Q Series T M Thermal Analyzer TGA/DSC (Q600). The Fourier transform infrared (FTIR) spectra of N-CNTs were recorded on a PerkinElmer Spectrum RX1 FTIR spectrometer by embedding the samples into KBr pellets. The adsorption-desorption isotherms and surface area of N-CNTs were determined on a Micrometrics Tristar II surface area analyser. The graphitic nature of the N-CNTs was determined by a Raman spectrometer (DeltaNu Advantage 532 TM ). Four accumulated spectra were collected to access the homogeneity of the samples. The synthesized N-CNTs were purified in nitric acid under microwave irradiation using a CEM Discover SP microwave instrument. The surface chemical composition of N-CNTs was analysed using X-ray photoelectron spectroscopy (XPS). XPS analysis was conducted on a Quantum 2000 instrument using a monochromated Al Kα source and charge neutralizer, with a pass energy of 117.4 eV. The peaks were deconvoluted using CasaXPS programme. The surface charge on the N-CNTs in ultra-pure water was determined with a Malvern Zetasizer (NS500). Boehm titration was conducted to quantify the acidic functional groups on the N-CNT surfaces. Potassium hydrogen phthalate (KHP) was used as the primary standard for the standardization of NaOH solutions using phenolphthalein as the indicator [42]. 0.20 g of each N-CNTs samples were placed in separate bottles. 25 mL NaOH (0.05 M), Na 2 CO 3 (0.025M) and NaHCO 3 (0.05 M) were added to the bottle, sealed and shaken for 24 h. The solutions were then filtered and titrated against standardized HCl or NaOH [43]. Different functional groups (i.e. phenolic, carboxylic, lactonic and hydroxyl groups) were calculated based on the amount of acid or base consumed. The general procedure described by Pan et al. [44] was used to synthesize FcMeCH 3 . Briefly, ferrocenemethanol (1 mM) and 2-methyl-1H-imidazole (1.1 mM) was refluxed in acetic acid for 9 h at 60 • C. The product formation was monitored using preparative TLC plates with a solvent system of CH 2 Cl 2 /MeOH (4:1). The product was neutralized with 50% KOH in distilled water to remove the acetic acid and then puried by column chromatography. The final product was washed in Na 2 SO 4 and finally dried under vacuum to obtain yellow crystals. Detailed characterization of FcMeCH 3 has been reported in our previous work [45].

Synthesis of N-CNTs
N-CNTs were synthesized by pyrolyzing FcMeCH 3 catalysts in acetonitrile at 850 o C using the CVD method. In the experiment, different concentrations of benzophenone were added to the reaction mixture to study the effect of oxygen on the growth of N-CNTs. The CVD procedure and set-up described by Oosthuizen et al. [46] was followed. Briefly, 0.25 g of the catalyst was added to 0.5, 1.0, 1.5 and 2.0 g of benzophenone to produce 1, 2, 3 and 4 wt.% oxygen, respectively. The mixture was dissolved in acetonitrile (as carbon and nitrogen source) to make a 10 g solution . The reactant mixture was injected using a syringe at 0.8 mL min −1 through the quartz tube placed in a muffle furnace. The mixture was swept through the tube by 10% hydrogen in argon carrier gas for 100 mL min −1 . After 30 min of reaction, the furnace was allowed to cool to room temperature, and the product was collected from the hot region of the furnace. N-CNTs from 1-4 wt.% oxygen is denoted as N-CNTs-1%, N-CNTs-2%, N-CNTs-3% and N-CNTs-4%, respectively. N-CNTs-0% was synthesised by dissolving FcMeCH 3 catalyst in acetonitrile. For comparison, N-CNTs-Fe was synthesized using ferrocene and toluene as catalyst and solvent, respectively.

Purification procedure for N-CNTs
N-CNTs were purified using microwave digestion. Briefly, N-CNTs (0.8 g) were dispersed in nitric acid (6 M) by ultrasonic agitation for 45 min. After sonication, each sample was purified by a microwave assisted irradiation. This was done by placing 50 mL of the dispersed sample in a thermal resistant Teflon (Milestone (TFM)) vessel on a sample rotor available for 4 vessels. The microwave was set at 100 W power and ramped from room temperature to 100 o C for 30 min. After digestion, the obtained suspension was filtered on 0.1 µm PTFE membrane. The collected solid samples were washed with deionized water until a neutral pH was obtained. Afterwards, the N-CNTs were washed with alcohol and dried in an oven at 100 • C for 24 h.

TEM analysis
The morphology of N-CNTs was studied by TEM. The obtained images are shown in Figure 1. The incorporation of nitrogen correlated with the bamboo-like structure typical of N-CNTs [47] (Figure 1a-f). The use of FcMeCH 3 as a catalyst in acetonitrile and benzophenone gave mainly clean N-CNTs ( Figure 1) and in good yield (Table 1). This could be attributed to the cleaning effect of oxygen as it reacts with amorphous carbon to form CO 2 . N-CNTs and carbon sphere (CS) are obtained in toluene solvent.
The N-CNTs yields decrease as the concentration of oxygen increases due to the formation of CO 2 from unreacted carbon and oxygen. The TEM images of N-CNTs-1% and N-CNTs-2% (Figure 1a and b) showed a curly tubular structure. This could be as a result of Fe catalyst left inside the N-CNTs with smaller diameters [48]. The bamboo compartment of N-CNTs-1%, N-CNTs-2%, N-CNTs-3% and N-CNTs-4% decreased as the concentration of oxygen increased (Table reftab2). All N-CNTs obtained are opened at the tips, while some region along the tube gave stacked cup-like cones. This suggests that the bamboo structures were obtained by tip growth mechanism [49]. The cup-like cones appear to be more prominent as the oxygen concentrations increased (Figure 1c and e). N-CNTs-0% and N-CNTs-Fe exhibit relatively straight tubes and a wall thickness of ∼15 nm. The wall thickness decreases as the concentration of oxygen from benzophenone increases ( Table 2). This is due to a reduction in the number of corrugated carbon layers and the closure of tubes which resulted in reduced compartment distances [50]. Table 2 shows the effect of oxygen on the inner diameter (ID) and outer diameter (OD) of the synthesized N-CNTs. From the results, it is believed that oxygen plays a vital role in modulating the morphology and diameters of N-CNT [39]. The OD decrease as the oxygen content in the reaction mixtures increases. This is due to the effect of oxygen on the catalyst leading to a decrease in Fe particle size as a result of catalyst migration, sintering, and precipitation processes [51,52]. It was suggested that oxygen enhances the catalyst activity by removing amorphous carbon which prevents N-CNTs surface 207  [41]. N-CNTs-1%, N-CNTs-2% and N-CNTs-3% gave smaller ID (14±7 nm to 16±5 nm). However, larger ID N-CNTs was obtained for N-CNTs-4% (i.e. 33±8 nm). This is due to excess oxygen content, leading to etching of the outer walls which largely affects N-CNTs quality. Figure 2 shows the HRTEM images of N-CNTs with varying oxygen contents. An increase in the d 002 interlayer spacing of the graphitic carbon was observed as the OD decreases. The interlayer d-spacing increased from 0.339 nm (N-CNTs-1%) (Figure 2b) to 0.344 nm (N-CNTs-3%) (Figure 2d). The increase in the d 002 spacing is due to the curvature of smaller diameter N-CNTs and higher strain caused by the structural defect on the nanotube walls [53]. Also, the regular bamboo compartment for N-CNTs-4% (Figure 2f) was destroyed. This is attributed to supersaturation of molten Fe catalyst particles with carbon [54]. It could also be as a result of highly reactive oxygen at the surface or within the molten Fe nanoparticles which form FeO (i.e. Fe + O 2 →FeO + O), leading to etching of the graphitic carbon.

SEM analysis
The morphology of N-CNTs was analysed using SEM. The obtained images are shown in Figure 3 (a-f). Figure 3 a-e manifested the effect of oxygen on N-CNTs growth and alignment. This was as a result of the reaction of oxygen with very reactive hydrogen radical involved in the hydrocarbon-based growth of nanotubes [20]. This helps to scavenge unreactive hydrogen which inhibits the growth of sp 2 like graphitic sheets [30]. For example, the vertical alignment was observed for N-CNTs-1%, N-CNTs-2% and N-CNTs-3% (Figure 3 b-d) compared to N-CNTs-Fe (Figure 3f) The alignment was depleted at higher oxygen concentration (as observed in N-CNTs-4%). This could be attributed to the partial oxidization Fe-catalyst which reduced catalyst density, leading to reduced N-CNTs nucleation [55].  (Table 1).

Thermal studies
The thermal stabilities of N-CNTs with different oxygen wt.% loading was studied as shown in Figure 4. TGA analysis was measured in air at 25-1000 • C to give an idea of the oxygen content and the purity of the samples. The first mass loss due to loss of water appears before 100 • C. N-CNTs-1% shows a significant weight loss at 386 • C while N-CNTs with 2-4 wt.% oxygen showed a weight loss between 390-530 • C. N-CNTs-0% is the most thermally stable with the decomposition temperature at 589 • C. The oxygen treated N-CNTs started to decompose at the on-set point between 334 and 430 • C (Table 3). All N-CNTs showed weight loss after decomposition above 87% with a residual mass between 9.6-0.5%. From DTG curves, the maximum mass loss temperature for 1-4% oxygentreated N-CNTs is between 392 and 514 • C. Further investigation by Raman, XPS and FTIR analysis was done. Figure 5 shows the Raman spectra of N-CNTs-0%, N-CNTs-1%, N-CNTs-2%, N-CNTs-3%, N-CNTs-4% and N-CNTs-Fe. The two prominent peaks observed at ∼1330 and ∼1573 cm −1 are assigned to the D-and G-bands, respectively. The intensity ratios of the D-and G-bands (I D /I G ) shows the defect level of graphitic carbon materials [56,57,58]. The I D /I G ratio of N-CNTs-0% and N-CNT-Fe is 0.74 and 0.66, respectively (Table  4). After introduction of oxygen from benzophenone in the reactant precursor, the I D /I G ratio increased to 0.97, 0.93, 0.85 and 0.79 for N-CNTs-1%, N-CNTs-2%, N-CNTs-3% and N-CNTs-4%, respectively. This is as a result of incorporation of surface 208  oxygen functionalities and N atoms which produces more defects and disorders on the graphitic structure of the N-CNTs. The lower I D /I G ratio in N-CNT-Fe and N-CNTs-0% indicates that fewer defects are introduced in the carbon lattices due to less nitrogen atom intrusion into the graphitic carbon network compared to N-CNTs-1%, N-CNTs-2% and N-CNTs-3%, respectively. The width of the G-band peak also indicates the level of doping in N-CNTs [59,60]. Table 4 shows that the Gband width of N-CNTs with varying concentrations of oxygen follows the order of N-CNTs-1% > N-CNTs-2% > N-CNTs-3% > N-CNTs-4% > N-CNTs-0% > N-CNTs-Fe. This result suggested a possible increase in N-doping at lower oxygen concentration.  Figure 6 shows the FTIR spectra of N-CNTs from 0-4% of oxygen and N-CNTs-Fe. Peaks at around 2927 and 2625 cm −1 are assigned to the O-H and CH 3 stretching vibrations [61], respectively. The prominent band at 2381 cm −1 is assigned to the characteristic absorbance of CO 2 groups [62], while peaks at 1763, 1567 and 1030 cm −1 are assigned to stretching vibrations of C=O, C=N and C-O functional groups, respectively [63]. The peaks at 1375 cm −1 are assigned to stretching vibrations of C-NH 3 [64]. The presence of C=N and C-N functional group on the purified N-CNTs indicates the substitution of graphitic sp 2 carbon with nitrogen, leading to the bamboo configuration observed in TEM images [65]. For N-CNTs-0%, the intensity of the C=O band peak at 1763 cm −1 was weaker than that from 1-4% oxygen, which becomes broader as the concentration of  Table 5. According to this method, NaHCO 3 , Na 2 CO 3 and NaOH, neutralize carboxyl groups, carboxyl groups and lactones; and carboxyl groups, lac-  tones and phenols, respectively. Therefore, different functional groups can be calculated from the volume of acid and bases used. The acid functional groups on N-CNTs-1%, N-CNTs-2%, N-CNTs-3% increases a little as the oxygen concentration increases compared to N-CNTs-0% and N-CNTs-Fe, while the amount of basic functional groups significantly increases. This indicates that the oxygen functionalities on the surface of N-CNTs synthesized in the presence of oxygen are more basic than N-CNTs synthesized in acetonitrile only [66]. From the results, N-CNTs-1%, N-CNTs-2%, N-CNTs-3% contains high concentration of basic group (≥ 1.025 mmol/g) ( Table 5). Additionally, N-CNTs-2% has the highest concentration of the phenolic group.

Surface chemistry of N-CNTs
Zeta potential (ζ ) measurement provides information on the adsorption of ions (H + and OH − ) from aqueous suspension and dispersibility which lead to the formation of net charge on  [67]. These net charges lead to the formation of the electrical double layer which stabilizes the suspension and prevents particle aggregation. The properties of nanoparticles are largely affected by their colloidal stability. Nanoparticles with zeta potential less than -25 mV or above +25 mV are said to have a high degree of stability [67]. Table 6 shows the variation in the zeta potential of N-CNTs-0%, N-CNTs-1%, N-CNTs-2%, N-CNTs-3%, N-CNTs-4% and N-CNTs-Fe nanofluids. Our result showed that the zeta potential follows the   order N-CNTs-2% > N-CNTs-1% > N-CNTs-3% > N-CNTs-Fe > N-CNTs-0% > N-CNTs-4%. The zeta potential increases as the concentration of oxygen increases but drops sharply at higher oxygen concentration. According to this measurement, the oxidized N-CNTs are negatively charged in the aqueous phase as a result of oxygen-containing functional group ionization [68]. The effect of oxygen on the porosity of N-CNTs-0%, N-CNTs-1%, N-CNTs-2%, N-CNTs-3% and N-CNTs-4% was characterized by BET analysis. The nitrogen-adsorption isotherms of all N-CNTs are of type IV with different hysteresis loops in the high-pressure regions (P/P o = 0.7-1), suggesting the presence of mesoporous structure [69]. As shown in Table 7, the surface areas of N-CNTs follows the order: N-CNTs-2% > N-CNTs-1% > N-CNTs-3% > N-CNTs-Fe > N-CNTs-0% > N-  CNTs-4%. This indicates that the surface area of the N-CNTs can be modified by the introduction of oxygen into the reactant mixture.

Elemental analysis
The elemental composition and the bonding environment of the C, O and N species were determined by XPS analysis, and the result is presented in Table 8. Figure 7 shows the highresolution N 1s energy region of selected N-CNTs (N-CNTs-0%, N-CNTs-3%, and N-CNTs-Fe). The deconvolution of the spectra gave three distinct N 1s peaks centred at 398.50, 400.18 and 401.20 eV assigned to pyridinic, pyrrolic and graphitic nitrogen, respectively [70]. A steady increase in the level of nitrogen-doping was observed during the CVD synthesis, followed by a gradual decrease due to an increase in oxygen con-212  (Table 8); a result consistent with Raman and TGA data. Compared with N-CNTs-0% and N-CNTs-Fe, the N-CNTs-3% gave higher pyrrolic-nitrogen species which could be attributed to the presence of active site caused by the lower amount of oxygenated species on the graphitic carbon framework [71]. At a low oxygen concentration in benzophenone, pyridinic-N species was formed (Table 8). At a high oxygen content, pyrrolic-N was obtained [72]. This may be due to the change in the elemental ratio (C: N: O) in the precursor mixture. The amount of nitrogen incorporated into N-CNTs obtained in our study is higher compared to other studies [34,73]. This was attributed to the higher amount of nitrogen contained in the ferrocenyl imidazolium catalyst. Additionally, the decrease in nitrogen content could be ascribed to the presence of O in benzophenone which we believe inhibits nitrogen incorporation into N-CNTs. Deconvolution of O 1s spectra of N-CNTs-0%, N-CNTs-3% and N-CNTs-Fe peaks gave two bands centred at 531.26 and 533.40 eV assigned to C=O and C-O [74], respectively. The elemental analysis results (Table 8) corroborate XPS result with increased nitrogen-doping triggered by addition of varying amount of oxygen. Table 9 shows the detailed analysis of C 1s peaks of N-CNTs-0%, N-CNTs-3% and N-CNTs-Fe. The deconvoluted C 1s peaks produced five components at 284.3, 285.8, 287.0, 287.9 and 289.4 eV, assigned to C=C, C-C, hydroxyl, carbonyl and carboxyl functional groups, respectively [75,76]. From the XPS analysis, the carboxyl and carbonyl functional groups increase as the sp 2 carbon decreases. For example, the atomic percentage of C=O increased from 3.8 (N-CNTs-0%) to 8.8% (N-CNTs-3%). The oxidation of C=C is confirmed by an increase in C-C components, which led to the formation of new functional groups on N-CNT surfaces.

Powder XRD pattern studies
The XRD profiles of N-CNTs (i.e. 0-4 wt.% oxygen) and N-CNTs-Fe showed the crystalline nature of N-CNTs ( Figure  8). All diffraction patterns showed the formation of (002) crystalline carbon plane (i.e. 26 • ), indicative of CNTs formation [77]. Other peaks at 44.5 • , 49.1 • and 77.6 • correspond to (100), (221) and (401) reflections of the graphite structure of N-CNTs, respectively. The weak peaks at 37.6 • and 43.5 • are assigned to Fe 3 C and Fe 2 O 3 , respectively, which are stuck inside the core of N-CNTs [78,79]. The XRD diffraction pattern for N-CNTs showed a decrease in the intensities of (002) peaks as the alignment increases, particularly, from N-CNTs-1% to N-CNTs-3%. Also, the diffraction peak intensities of the (002) plane for N-CNTs-1%, N-CNTs-2% and N-CNTs-3% are weaker than those of N-CNTs-0% and N-CNTs-Fe. This shows that N-CNTs-Fe and N-CNTs-0% have fewer structural defects since N-doping create faults in the graphitic layers. This result agrees with the Raman results ( Table 4). The interlayer d-spacing increases from 0.339 to 0.352 nm as oxygen concentration increases (Table 10). The increase in the d 002 spacing is due to curvature of smaller diameter N-CNTs and higher strain caused by the structural defect on the nanotube walls. This result is consistent with d 002 spacing obtained from HRTEM analysis.

Conclusion
This study presented the role of oxygen and nitrogen-doping as a promising method to improve the physicochemical properties of N-CNTs. This has critical implications for reproducibility in N-CNT synthesis, particularly on the effect of oxygen in diameter and wall thickness control. It can be concluded that the introduction of an appropriate amount of oxygen promotes N-CNTs growth with clean walls and reduced diameters. From XPS analysis, pyrrolic-N was predominantly incorporated into the crystalline CNT structure at a high oxygen concentration. Nitrogen-doping was further confirmed by TGA analysis and Raman spectroscopy. Lastly, the understanding of the effect of oxygen species on the morphology and surface area of N-CNTs during synthesis is critical in vast numbers of industrially promising supported metal nanoparticles catalyst design.