Size-Dependent Chlorinated Nitrogen-Doped Carbon Nanotubes: Their Use as Electrochemical Detectors for Catechol and Resorcinol

,


Introduction
Phenol and its derivatives are widely used in many felds, such as tanning, cosmetic, dye, chemical, and pharmaceutical industries [1,2]. Tese compounds contain aromatic rings with one or more hydroxyl groups functionalized on the carbon atom, ranging from simple phenols to high molecular weight polymers. Terefore, the development of methods that can be used for sensitive and selective monitoring of phenolic compounds is very important for environmental protection [3,4]. Catechol (CC) and resorcinol (RS), together with hydroquinone (HQ), are a class of phenolic compounds called dihydroxybenzene isomers and are extensively used as chemical intermediates in the manufacture of many products, such as pesticides, agrochemicals, dyes, cosmetics, food additives, medicines, insecticides, and explosives [5][6][7]. Tese phenolic compounds are known as one of the signifcant threats for environmental pollutants because of their poor degradability and strong toxicity in nature [8,9]. Traces of CC and RS are widely distributed in our ecosystem, for example, in our water sources, and they usually coexist as pollutants [10][11][12]. It is known that CC can produce renal tube destruction, liver function reduction, and strong central nervous system suppression when adsorbed in the gastrointestinal tract at high doses [13]. On the other hand, prolonged exposure to RS resulted in suppression of thyroid hormone synthesis in humans, haematological abnormalities, carcinogenesis, and fatal cases of human fetus poisoning [14]. As such, catechol is considered more toxic than resorcinol, and both are considered more toxic than phenol [15,16]. Catechol is fatally toxic to fsh at concentrations of 5-25 mg/L [17], and it has been detected in wastewater from coal conversion processes (its concentration may be as high as 5300 mg/L at low temperature wastewater) [18], crude wood tar and drainage water from bituminous shale, and water from coal carbonization and gasifcation [19].
Phenols and associated compounds have been listed as priority pollutants by the Ministry of Environment and Forests (MoEF), the Government of India, and the United States Environmental Protection Agency (USEPA). As a result, Environmental Protection Agency (EPA) regulations call for controlling the level of these phenolic compounds to <1 ppm in wastewater [20]. In a quest to alleviate these hazardous compounds in water, efective and real-time monitoring of phenolic compounds using methods such as capillary electrophoresis [21], spectrophotometry [22], chromatography analysis [23], and electrochemical analysis has been explored.
Electrochemical techniques have recently drawn more attention as the detectors of choice due to their characteristic of being low cost, fast, simple, less time-consuming, and sensitive [24,25]. However, since CC and RS are isomers, which may result in the overlap of their oxidation peaks, there are still challenges in simultaneously distinguishing and detecting CC and RS using bare electrodes. To solve this problem, researchers have developed new efcient and sensitive materials for simultaneously detecting dihydroxybenzenes by coating the surface of working electrodes with various carbon-based nanomaterials with excellent electronic transfer rate and good stability [26]. Among the developed carbon-based nanomaterials, multiwalled carbon nanotubes (MWCNTs) have aroused widespread attention [27]. Reports have surfaced on the use of surface-modifed and heteroatom-doped MWCNTs coated on working electrode materials as electrochemical sensors for catechol with fewer studies on the detection of catechol in the presence of resorcinol [11,[28][29][30][31][32][33][34][35][36][37].
Te oxygen-plasma-treated CNTs which were modifed onto a GCE surface showed excellent electrochemical behaviour for the detection of catechol with the detection limit of 0.60 μM and exhibited long-term stability [28]. An electrochemical sensor based on fuorine and nitrogen codoped carbon dots decorated laccase was fabricated and used for the detection of catechol, with a low detection limit of 0.014 μM and sensitivity of 219.17 μA·cm −2 ·mM −1 obtained [29]. Oxygen-plasma-treated CNT electrodes were used for the detection of catechol, and it was found that this electrode had better electrochemical performance for the analysis of catechol than that of as-synthesized CNT electrode and exhibited long-term stability [30]. Samarium oxide decorated functionalized multiwall CNTs were used as electrode material for the detection of catechol with a low detection limit of 0.03 μM obtained and recovery of 94.5 to 99% in water samples [31]. A carbon paste electrode modifed with the poly(sulfosalicylic acid) and multiwalled carbon nanotubes (MWCNTs) composite was constructed and used as an electrochemical sensor for catechol, and low detection limits of 0.16 μM were obtained [32]. An electrochemical sensor based on modifcation of GCE by the composite containing temperature-responsive polymer polystyrene-poly-N, N-diethyl acrylamide-polystyrene (PS-PDEA-PS), and fullerenes carboxylate multiwalled carbon nanotubes (C 60 -MWCNTs) was fabricated and used for the detection of catechol [33]. Te detection limit obtained for catechol was 1.45 μM, and the sensor possessed good stability and reproducibility [33]. A sensitive electrochemical sensor for catechol based on multiwalled carbon nanotubes poly(1,5-diaminonaphthalene) composite modifed GCE which showed a low detection limit of 0.01 μM for catechol has been reported [34]. A novel catechol sensor fabricated from double layers of functionalized multiwalled carbon nanotubes coated with zinc oxide on a glassy carbon electrode showed a low detection limit of 0.027 μM for catechol [35]. An electrochemical sensor for catechol and nitrite based on gold nanoparticles (AuNPs) deposited on chito-san@N,S codoped multiwalled carbon nanotubes composite modifed GCE was fabricated and exhibited low detection limits of 0.2 μM for both catechol and nitrite [36]. A surfactant-modifed graphene paste electrode designed for the determination of catechol in the presence of resorcinol was developed, and the detection limit for catechol was found to be 0.106 μM [37]. A fexible composite paper Fe-Cu-based metal-organic framework/reduced graphene oxide electrode was prepared for the simultaneous detection of catechol and resorcinol, with detection limits of 0.016 and 0.020 μM, respectively [11].
In our previous studies, we have evaluated the role of chlorine addition during carbon nanomaterials (CNMs) synthesis in the presence and absence of nitrogen sources using catalytic vapour deposition (CVD) injection and bubbling methods on the morphology of the CNMs [38,39]. From the reported literature, we expected the addition of chlorine to (i) result in an increase in the growth rate of the CNTs, (ii) aid in the purifcation of the carbon nanomaterials, and (iii) increase the amount of nitrogen doped into the carbon nanostructures. In our previous study, we also found that chlorine increased the growth rate of the CNTs, and that varying the amount of chlorine in the feed containing a nitrogen source using an injection CVD method, resulted in the formation of open-ended and closed-ended CNTs of variable morphologies [39]. Materials synthesized with the addition of large concentrations of chlorine in the feed also appeared more cleaner and were highly graphitic [39] while the synthesis of CNTs by adding only chlorine vapour to the CVD reactor containing the catalyst resulted in the creation of CNTs that had defects on the outer walls which in turn facilitated the growth of secondary carbon nanofbers at their walls [38]. In the present study, we needed to clarify the role of chlorine concentration further on the growth rate, morphology, purifying ability, and electrochemical response of nitrogendoped carbon nanotubes using a bubbling CVD method.
To the best of our knowledge, there are no reports on the electrochemical detection of dihydroxybenzene isomers using a GCE modifed with chlorinated N-doped CNTs, specifcally the role of the size of the carbon nanomaterials on the sensing capability.

Fabrication of Chlorinated N-Doped CNTs by
Bubbling CVD Method. About 1.0 g of a 10% Fe-Co/CaCO 3 catalyst, which was prepared using a method by Mhlanga et al. [40], was weighed and added to the quartz boat (purchased from Protea Laboratory Solutions), which was then placed in the centre of a quartz tube. Nitrogen (N 2 , purchased from Afrox) was used as a carrier gas at a fow rate of 240 mL/min, while acetylene (C 2 H 2 , purchased from Afrox) was used as a source of carbon at a fow rate of 90 mL/min. Te fow rates used were adopted from a method by Tetana et al. [41]. In the beginning, the furnace (which was custom-made at the University of the Witwatersrand) was gradually heated to 800°C at a heating rate of 10°C/min under fowing N 2 (50 mL/min). After the temperature of the furnace reached 800°C, the N 2 fow was adjusted to 240 mL/min, and the C 2 H 2 gas was also opened (fowing at a fow rate of 90 mL/ min), and both gases were bubbled into a mixture of acetonitrile and dichlorobenzene of various volume ratios, and the vapours created were fown into the CVD reactor. After 60 min of reaction, the system was left to cool to room temperature under a continuous fow of N 2 (50 mL/min). Te quartz boat was then removed from the reactor, and the carbon deposit that formed was weighed. Te carbon deposit was purifed in 30% nitric acid (HNO 3 ), by stirring under refux at 100°C for 1 hour, and this was followed by fltration of the material until the pH of the fltrates reached about 7. Te resultant purifed material was then dried in an oven overnight at 100°C.

Modifcation of GCE with Chlorinated N-Doped CNTs and Electrochemical
Tests. Te glassy carbon electrode (GCE) was cleaned before modifcation with chlorinated N-doped CNTs. Te GCE was frst polished with Al 2 O 3 (0.3 μm) powder on a polishing cloth. Tis was followed by rinsing the GC electrode with deionized water, which was followed by sonication for 10 minutes in the following solvents: HNO 3 (1 : 1) acid solution, ethanol, and fnally distilled water. Te above cleaning procedure was repeated several times until a shiny surface was obtained on the GC electrode.
Chlorinated N-doped CNTs suspension was prepared by mixing and dispersing 6 mg of chlorinated N-doped CNTs and 1 mg of PVDF in 400 μL dimethyl formamide (DMF) with the acid of ultrasonic agitation. About 50 μL of the chlorinated N-doped CNTs suspension was dropped onto the surface of a polished GCE, and the solvent was evaporated in an oven at 40°C. Te procedure was repeated two times, meaning the electrode was coated two times. Cyclic voltammetry (CV) and diferential pulse voltammetric (DPV) experiments were carried out in an electrochemical cell containing 10 mL of a 0.1 M phosphate bufer solution (PBS, pH 7.2) prepared at room temperature as the supporting electrolyte, and the scan rate used for CV was 50 mV·s −1 . A 0.020 mol/L stock solution of analytes (CC, HQ, and RS) was prepared in distilled water using 10 mL volumetric fasks. For DPV, the parameters were as follows: amplitude, P H (50 mV); pulse width, P W (60 mV); potential step, S H (10 mV); limit potential, E V (0.6 V); and step time, S T (200 ms). Te detailed electrode preparation procedure and electrochemical testing are presented in Scheme 1.

Characterization.
Transmission electron microscopy (TEM) (T12 FEI TECNAI G 2 SPIRIT) operating at 120 kV and scanning electron microscopy (SEM) (FEI Quanta 200) were utilized for the determination of the acid-purifed and unpurifed nanostructures morphology and their size distribution. For TEM, the nanostructure materials were sonicated in ethanol and deposited onto a holey carboncoated TEM Cu grid once the homogeneous mixture was obtained. For SEM, the nanostructured materials were poured onto a tape that was attached to an SEM stub. Te nanostructured materials on the stub were frst coated with carbon, followed by gold-palladium to prevent them from charging. Raman spectroscopy (Jobin-Yvon T6400 micro-Raman spectrometer) was used to deduce the crystallinity of the nanostructured materials, where excitation was provided by the 514 nm green laser and a spectral resolution of 3-5 cm −1 . Te amount of residues and other impurities present before and after acid purifcation of the nanostructured materials was determined by thermogravimetric analysis (TGA) (Perkin Elmer TGA 7). Tis was achieved by loading the nanostructured material sample onto a platinum pan, and the sample was heated to 900°C at a heating rate of 5°C/min, in a fowing air stream at 20 mL/min. Te binding energy and nitrogen environments and their quantities in the nanostructured materials were evaluated using X-ray photoelectron spectroscopy (XPS) (Kratos Analytical (UK) provided by Rhodes University) with an AXIS Ultra DLD with an Al (monochromatic) anode equipped with a charge neutralizer. Electrochemical detection of analytes and the response behaviour of GC-coated electrodes were performed using cyclic voltammetry and diferential pulse voltammetry (Biologic SP-240, Potentiostat (Bruno Steiner Laboratory Consultancy), operated by EcLab software). A glassy carbon working electrode (A � 0.07 cm 2 ), platinum wire which acted as a counter electrode, and Ag/AgCl (3 M NaCl) which was used as a reference electrode were used in the electrochemical analysis.
International Journal of Electrochemistry

Structural and Morphological Studies.
Te morphological properties of the nanostructured materials prepared by varying the concentration of chlorine in the acetonitrile feed were evaluated using TEM and SEM. Figure 1(a) represents a TEM image taken from our previous study prepared by the addition of chlorine vapours into the CVD reactor that contained an iron-cobalt catalyst as substrate while using dichlorobenzene as a source of chlorine [38].
As it can be clearly seen from the fgure, secondary carbon nanofbers (CNFs) were grown on the surface of the principal CNTs ( Figure 1(a), circled parts), and the growth was thought to have emanated from defected tubes, and that the defects were created by chlorine [38]. In this study, the efect of chlorine concentration on the morphology and the growth of nitrogen-doped CNTs were therefore evaluated. Figures 1(b) and 1(c) present TEM images obtained when about 33.3 vol.% of dichlorobenzene was mixed with 66.7 vol.% of acetonitrile making the volume ratio to be 2 : 1 (CH 3 CN : DCB). Te average outer diameters (obtained using ImageJ) of the chlorinated nitrogen-doped CNTs have increased tremendously (∼115 nm, Supplementary Figure S1b, Table 1) when compared to the diameter of CNTs prepared from pure DCB which was about 33 nm [38]. Some of the CNTs appeared compartmented which is evidence of nitrogen doping or could also have emanated from the morphology of the metal catalyst that was used as a substrate [42] (Figure 1(b)), and some intratubular segmentations can also be observed which appeared to also be responsible for their increased diameters. Metal nanoparticles encapsulated inside the CNTs were also observed in some CNTs (Figure 1(b) circled). Te corresponding SEM images are presented in Figure 1(d) and Supplementary Figure S1a, and they showed that the CNTs were entangled.
Nanostructures generated from the solution mixtures containing an increased concentration of DCB (66.7 vol.%) and reduced concentration of CH 3 CN (33.3 vol.%) appeared much thinner, with notable bamboo-compartmented structures, and had rough surfaces, whereas some were hollow and others bent (Figure 2(a)). All these structural features showed that these materials were highly defected. Teir average outer diameters (∼57 nm) were greatly reduced in comparison to those generated at low concentrations of DCB (33.3 vol.%) ( Table 1, Supplementary Figures S1c and S1d). Nitrogen-doped CNTs of various sizes and morphologies were obtained for nanostructures generated at high concentrations of DCB (66.7 vol.%), which showed that the selectivity was lost at this concentration ( Figure 2(b)). Te observed increase in the diameter of the CNTs with the addition of chlorine during the synthesis of chlorinated N-doped CNTs may imply that the catalyst particle increased in size during the CVD process.
Note that the CNT diameter increased tremendously at low chlorine concentrations, due to an interaction between iron and Cl 2 , which modulates the iron-cobalt catalyst and promotes the aggregation of catalyst particles by thermal difusion [43]. Terefore, we can conclude that at low chlorine concentration, the size of the catalyst and the diameter of the CNTs increased because of chlorine addition. However, when the concentration of chlorine was doubled, the diameter of the NCNTs decreased tremendously to almost half of that produced at low concentrations. Tese phenomena can be explained to arise because of the counteracting of the thermal difusion of iron by iron etching with chlorine [43]. In our previous study [39], the addition of about 80% of chlorine to the CH 3 CN feed resulted in the production of mixtures of open-ended CNTs, closed-ended entangled CNTs, and carbon nanospheres, and the yield only increased by 0.18 g from the original catalyst mass. A yellowish-brownish liquid was also observed inside   International Journal of Electrochemistry the quartz tube downstream, confrming that iron was etched by chlorine at high chlorine concentrations. In conclusion, we can say it is important to control the concentration of chlorine added to the acetonitrile feed to avoid etching of the iron nanoparticles because catalyst etching advances as the chlorine concentration increases. Termal gravimetric studies were conducted for NCNTs synthesized in the presence of various concentrations of chlorine to study their decomposition and oxidation resistance. TGA curves are presented in Figure 3. We observed a sharp weight decrease at ∼500°C in the TGA spectra of Ndoped samples produced at high concentrations of chlorine (1 : 2 ratio), which indicates that the chlorinated NCNTs are likely to react and also contain uniformly distributed defects along their entire surface (Figures 4(a) and 4(b)). For chlorinated NCNTs generated from a 2 : 1 volume ratio feed, the initial degradation was observed at higher temperatures of 550°C to 850°C which showed that the materials were less disordered (Figure 4(a)). Te higher initial degradation temperatures in these materials could be because of the larger diameter CNTs produced in this feed (Figure 4(a)). Loss of water was also observed by the weight loss at around 80°C for NCNTs (1 : 2). About 10% weight loss was also observed at temperatures below 500°C from NCNTs (1 : 2), and this was attributed to oxidation of amorphous carbon [44]. Te observed DTGA curve was also broad, which could indicate the oxidation of various shaped nanomaterials at the 2 : 1 volume ratio (Figure 4(b)). TGA data also depicted that the weight loss was around 100% and 92.3% for chlorinated NCNTs generated from 1 : 2 and 2 : 1 solvent mixture, respectively (Supplementary Table S1). Te remaining material observed from chlorinated NCNTs generated from 2 : 1 solvent mixture could be the catalyst residues. Tese catalyst particles were observed from the TEM images trapped inside the chlorinated NCNTs tubes (Figure 1(b), circled parts) and could not be removed during acid treatment due to the thick walls of the chlorinated NCNTs whereas with the chlorinated NCNTs generated from 1 : 2 solvent mixture, all the catalyst particles were removed by acid treatment, which was made easy by their thin walls, and also, most of the catalyst nanoparticles were etched by large amounts of chlorine that were added during synthesis. Supplementary Figure S2 presents TEM images of pristine NCNTs (1 : 2) and NCNTs (2 : 1) before purifcation with acid and catalyst particles can be seen on top of the CNTs. Te absence of the catalyst residual mass in the NCNTs (1 : 2) could also be attributed to the purifying property of chlorine as literature has shown that chlorine can act as a purifying agent [45]. It was also observed from the TGA curves that the oxidation resistance of the nanotubes decreased as the chlorine concentration in the feed increased as a result of increased density of defects and wall reactivity due to the incorporation of N within the carbon crystal (Figure 3(a)).
Te structural disorder of the two chlorinated NCNT materials was compared using Raman spectroscopy. Figure 4 presents representative spectra generated from 2 : 1 and 1 : 2 CH 3 CN : DCB volume ratios. Evaluation of the I D /I G ratio revealed that chlorinated N-doped CNTs generated from 1 : 2 CH 3 CN : DCB feed ratio were highly defected with ratios of ∼1.0, showing that an increase in chlorine concentration signifcantly increased the level of disorder in the chlorinated NCNTs, in comparison to I D /I G ratio of ∼0.63 obtained for chlorinated N-doped CNTs generated from 2 : 1 volume ratios. Similar observations were made from the TEM analysis, as the chlorinated NCNTs walls had rough surfaces or appeared corrugated with an increase in DCB concentration. Te absence of the 2D peak on NCNTs (1 : 2) materials also supports the observation that these materials were less graphitic. Tis could again be correlated to the etching of iron by chlorine which occurred at high chlorine concentrations. It appears that highly crystalline materials were generated at low levels of chlorine than at high levels of chlorine. At low levels of chlorine, there is increased diffusivity of carbon in the iron catalyst, which enables carbon to bond to more favourable sites in the CNT crystal, thus improving their crystallinity [43].
XPS analysis of materials generated in the presence of pure DCB as reported in our previous study did not show any nitrogen content, and the amount of chlorine was also very minimal [46]. Tis was expected since CNTs were synthesized by bubbling nitrogen gas and acetylene gas through dichlorobenzene solvent at room temperature, and the chlorine vapours generated could not have been sufcient to functionalize on the surface of the CNTs. However, the Cl2p peak could still be deconvoluted into two peaks at 199 and 200.1 eV, and these were associated with chloride ion and C-Cl bonds [46]. Te Cl 2p 1/2 /2p 3/2 ratios were below ca.1.6, which indicates that only ionic chloride was present [46]. EDS analysis showed that about 0.12 wt.% of chlorine was present in this sample. Te survey XPS spectrum of both chlorinated NCNTs revealed the presence of C1s, O1s, and N1s peaks (Supplementary Figure S3). A Cl2p peak was not observed in the chlorinated N-doped CNTs. Te lack of incorporation of chlorine into this material could be a result of the high synthesis temperature of 800°C utilized during the synthesis, as opposed to the synthesis temperature of 700°C that was used for the synthesis of chlorinated CNTs from pure DCB. In our previous study, the efect of temperature on the morphology of chlorinated CNTs using DCB as a source of chlorine was investigated [38]. CNTs generated at a reaction temperature of 800°C did not present any secondary CNF growth on the surface of the CNTs, and the materials were highly graphitic with very low I D /I G values of less than 1 [47]. Te highly graphitic nature of the material was attributed to the purifcation property of chlorine. It was suggested that at high synthesis temperatures, chlorine molecules were generated at a quick rate which resulted in them being at a very close proximity to each other, after which they reacted to form Cl 2 molecules which acted as purifying agents and were hence not involved in defect creation, but they simply escaped through the trap after purifying the material. Both materials revealed the main peak due to C1s, with a peak at ∼283 eV representing the C-C sp 2 , which was a typical graphitic carbon peak (Figures 5(a) and 5(b)). Tree additional peaks were deconvoluted at 284, 286, and 289.0 eV assigned to the C-C sp 3 , C-Cl/C-CN, and O�C-O bonding environments, respectively.      International Journal of Electrochemistry the graphitic structure [48]. We also evaluated the dominating nitrogen species in chlorinated N-doped CNTs produced from the two feeds since it is known that the physicochemical and electrocatalytic properties of N-doped CNTs rely on the type of nitrogen bonding formed during synthesis and on the nitrogen content [49]. Te deconvoluted N1s XPS spectra of chlorinated N-doped CNTs (2 : 1) generated from low concentrations of chlorine revealed that the quaternary nitrogen species was dominating in this material ( Figure 5(c)). Tese results suggest a more homogeneous distribution of nitrogen species in the CNT crystal. Graphitic N forms by merging N atoms into the defect region of graphene, resulting in a graphene with fewer defects [50]. Te deconvoluted N1s XPS spectra of chlorinated N-doped CNTs (1 : 2) generated from high concentrations of chlorine showed that the molecular and pyrrolic N species were dominating ( Figure 5(d)). Te pyrrolic nitrogen is known as a substitutional nitrogen, and it is part of a fve-membered ring and is sp 3 hybridized. Pyrrolic arrangements can induce capped and disordered structures creating bamboo-shaped compartments, and they can also be located at the edges [51,52]. XPS corroborated with Raman data where higher nitrogen content in chlorinated N-doped CNTs (1 : 2) was attributed to the higher I D /I G ratio (1.0), showing a less graphitic structure with more defects and more prone to nitrogen heteroatoms [53].
We also analysed the O1s peak, and three peaks appearing at 530.9, 531.7, and 532.9 eV were deconvoluted for chlorinated N-doped CNTs (2 : 1) generated at low chlorine concentration, which was attributed to residual metal oxide, C�O and C-O oxygen functional groups, respectively ( Figure 5(e)). However, only the C�O and C-O oxygen functional group peaks were observed for chlorinated N-doped CNTs (1 : 2) generated at high chlorine concentration ( Figure 5(f )). Te presence of a residual metal oxide peak in materials generated from low chlorine concentrations confrms metal encapsulation in these chlorinated N-doped CNTs (2 : 1) as observed from TEM and TGA studies. (2 : 1)/GCE. Te electrochemical behaviour of CC and RS was frst studied at diferent electrodes in a 0.1 M PBS of pH � 7.2. Te corresponding cyclic voltammetry (CV) curves are presented in Figure 6. Catechol exhibited a quasi-reversible redox reaction at a bare GCE and GCE modifed with NCNTs (1 : 2) with ∆E p of ∼200 mV (Figure 6(a)). Te oxidation peak obtained at bare GCE had an exceptionally low resolution and intensity, indicating that CC cannot be quantifed on a bare GCE. Te oxidation and reduction peaks were less defned, but their intensities increased when using GCE modifed with NCNTs (1 : 2), and this was attributed to the instability of their oxidizing and reducing intermediates [49]. Quasi-reversibility was maintained when using a GCE modifed with NCNTs (2 : 1), even though ∆E p slightly decreased to 184 mV. Te obtained CC oxidation and reduction peaks were well-defned and highly intense when detected using GCE modifed with NCNTs (2 : 1). Tese results indicated that NCNT flms generated from a mixture containing 2 : 1 (CH 3 CN : DCB) ratio can accelerate electron transfer. Similar electrochemical activity was observed for solutions containing RS as an analyte. However, RS exhibited an irreversible oxidation peak at potentials of ∼0.75 V (Figure 6(b)). Te RS peak was more defned and enhanced when using the GCE modifed with NCNTs (2 : 1). Enhanced anodic peak currents were observed during the detection of acetaminophen using DPV when using GCE modifed with larger diameter chlorinated N-doped CNTs produced at high temperature (850°C) as compared to modifcation with smaller diameter CNTs generated at 750°C [54]. Te increased electrochemical performance of NCNTs (2 : 1) modifed electrode was also attributed to the greatest graphitization observed in this material as evidenced by XPS and a low I D /I G Raman value compared to high values obtained with NCNTs (1 : 2).

Efect of pH.
We then evaluated the efect of pH on the oxidation peak of CC and RS in 0.1 M PBS using a GCE modifed with larger diameter NCNTs (2 : 1) material. Te electrochemical response of the phenolic compounds was seen to be infuenced by the pH since it afected the involvement of proton transference in the redox process [55]. Figures 7(a) and 7(b) show the efect of pH on anodic peak current and peak potential of CC (0.2 mM) and RS (0.8 mM) analysed using NCNTs (2 : 1)/GCE. Te anodic peak current of CC decreased with an increase in pH from pH 2-4 and then increased to almost the same current as that measured from pH 2 to 5, and another large decrease was observed at pH 6-8. pH of 2 was, however, taken as an optimum based on the well-defned anodic DPV peak that was obtained compared to an anodic peak obtained at pH 5 (Figure 7(c)). In the case of RS, a decrease in pH was also observed at pH 2-6, with a slight increase observed at pH 7 but still lower than that obtained at pH 2. As a result, pH 2 was also taken as the optimum pH condition to be used for the simultaneous detection of CC and RS. Tis behaviour is acceptable because of the pKa of each molecule (CC � 9.4 and RS � 10.0), which means that with increasing pH, the molecules acquire a negative charge as they approach the pH in which they are protonated [56]. In our study, we have stipulated that chlorine-containing functional groups and nitrogen atoms on chlorinated N-doped CNTs surface may become deprotonated and possess negative charges at higher pH values [57]. As such, dihydroxybenzene isomers also become readily available to be deprotonated to form anions with increasing pH values [49]. Te potential of the oxidation peak E pa at CC and RS shifted negatively with an increase in pH, indicating that the protons were directly involved in the redox process [58]. However, a deviation is observed when the pH changed from 4 to 5 where a shift in the peak potential to more positive potential was observed. A positive shift in potential from pH 4 to 5 could imply that the quasi-reversible electrochemical process is shifted towards the irreversible process [59]. And the fact that the current intensity of the peak potential at pH 5 is equal to the current International Journal of Electrochemistry  intensity at pH 2 can also clarify that the processes taking place at these two pHs must be similar; hence, it is expected for them to be oxidized at almost the same potential.
CV curves obtained for solutions containing 0.2 mM CC and RS in 0.1 M PBS at pH 2 were then used to evaluate their behaviour at this pH. Te obtained CV curves measured in the presence of both CC and RS in PBS bufer pH 2 over NCNTs (2 : 1)/GCE revealed two well-separated oxidation peaks at ∼0.58 V for CC and ∼0.86 V for RS (Figure 7(d)). As a result, a PBS bufer solution of pH 2 was used for the simultaneous detection of CC and RS to obtain the optimal sensitivity and selectivity.

Efect of Scan Rate.
A NCNTs (2 : 1)/GCE sensor was used for the simultaneous detection of CC and RS in 0.1 M PBS of pH 2. DPV curves showed well-separated peaks at voltages of +0.58 V for CC and +0.86 V for RS (Figure 8(a)). Te electrooxidation process of CC and RS was then studied by CV varying the scan rates from 10 to 55 mVs −1 using 0.2 mM CC and RS in 0.1 M PBS (pH 2) (Figure 8(b)). Te redox peak currents of CC (r � 0.9958 (I pa ) and 0.9754 (I pc )) and RS (r � 0.9914) increased with an increase in scan rate (Figure 8(b)). Tere was a linear relationship between the square root of the scan rate and the oxidation peak current, which confrmed that the electrooxidation process of CC and RS on NCNTs (2 : 1)/GCE is regulated by the difusion-controlled electrochemical process (Figures 8(c) and 8(d)) [60]. A shift in the oxidation peak potentials to higher potentials with increasing scan rate was attributed to a tendency of the redox process to shift from quasireversible to irreversible [61]. Tis is also because the concentration of the analyte at the surface of the electrode does not vary at high scan rates since the electron transfer rate constant becomes slow with fast scans since more extreme potentials are needed to induce electron transfer [62].

Simultaneous Determinations of CC and RS.
Te simultaneous and quantitative determination of the two dihydroxybenzene isomers was carried out by DPV of NCNTs (2 : 1)/GCE when the concentration of one isomer changed and of one remained constant (Figure 9). Figures 9(a) and 9(b) show DPV curves of 0.2 mM RS with diferent concentrations of CC and 0.2 mM CC with different concentrations of RS. Te peak current of CC increased linearly with an increase in CC concentration from 0.1 to 1.2 mM (R 2 = 0.9930), and the peak current of RS also increased linearly with an increase in RS concentration from 0.05 to 1.2 mM (R 2 = 0.9986) (Figure 9(c)). Te limits of detection for CC and RS were estimated to be 0.059 μM and 0.034 μM (3S/M), respectively. In addition to the current limit of detection and wider linear range, the proposed material possesses the virtue of a simple electrode preparation process and avoids the time-consuming electrode modifcation process utilized in other materials. Tese results suggest that CC and RS can be selectively and simultaneously determined in NCNTs (2 : 1)/GCE in their binary mixtures. Te excellent analytical performance may be attributed to the morphology of NCNTs (2 : 1) and the synergistic efect of chlorine on the N-doping efect and its enhanced defect creation which greatly increased the electrocatalytic active area and active sites, consequently promoting the electron transfer reaction of dihydroxybenzene isomers on the electrode surface. Te limits of detection obtained in the current work were also compared to those obtained for detection of catechol and resorcinol, using  other types of modifed CNTs, and they were quite comparable proving that the current sensor can be used for detection of this analytes in water (Table 2).
To evaluate the repeatability of GCE modifed with NCNTs (2 : 1), the peak currents of ten successive measurements of DPV in the binary solution mixture of 0.2 mM RS and 0.3 mM CC were determined. Te repeatability of the modifed sensor was determined in the same electroanalytical solution. Tere was no diference in the peak current signals for both CC and RS across all measurements (Supplementary Figure S4). Tis is an indication that the modifed sensor has excellent repeatability. Te relative standard deviations (RSDs) of 0.72% and 3.14% were obtained for RS and CC, respectively.
Te stability of the NCNTs (2 : 1)/GCE was also tested. In this case, the NCNTs (2 : 1)/GCE was covered and stored at room temperature for 2 weeks. Te stability was calculated by comparing the percentage (%) of current retention to the initial response. Te percentage retention of CC and RS reached 87% and 94% of the initial value, respectively, indicating high stability.

Interference Studies.
To evaluate the selectivity of the NCNTs (2 : 1)/GCE, the infuence of some possible interferents on the determination of 0.4 mM CC and RS was studied ( Figure 10). It is considered to have no interference when the current variation caused by interferents is less than 5%. It was found that a 200-fold concentration of Cl − , NO 3 − , K + , Ca 2+ , Mg 2+ , Zn 2+ , and Fe 3+ had no obvious interference to the determination of CC and RS, with the maximum signal change of 5% observed ( Figure 10). When Cu 2+ was added as an interference ion, an additional peak was observed at around 0.19 V, which had almost the same intensity as that of CC and RS (Supplementary Figure S5). Te CC peak and RS peak in the Cu 2+ solution were also reduced, but Cu 2+ did not interfere much because it was observed at a diferent potential from our studied analytes. Te observed reduction was 3 times for CC and by almost 2 times for RS. We can conclude not only that Cu 2+ will cause interference if present at low concentrations but also that the electrode is capable of simultaneous detection of CC, RS, and Cu 2+ ions in solution.
We also investigated the interference of hydroquinone (HQ) which is mostly present in water solutions together with CC and RS. DPV curve of HQ, CC, and RS alone in solution revealed that their oxidation peaks occurred at   Figure 11). However, when HQ was analysed together with CC in the same solution, the CC peak shifted to less positive potentials toward the potential where HQ is oxidized which resulted in their peaks overlapping, and a shoulder peak due to oxidation of CC could be observed at around 0.60 V (Figure 11). When all three dihydroxybenzene (HQ, CC and RS) analytes were mixed together in a ternary mixture, the HQ oxidation peak shifted to more positive potentials, resulting in it overlapping with the CC peak, which resulted in an observation of one broad peak at 0.60 V. However, an HQ peak at these concentrations could still be observed as a shoulder peak appearing at around 0.52 V (Figure 11).
Te prepared sensor was also tested for its practical application as a detector for CC and RS in local tap water samples using the standard addition method. In this method, tap water was diluted with a 0.1 M PBS (pH 2) in the ratio of 1 : 4. Te recovery was performed by spiking the electrolyte solution with 0.100 mM CC and RS in both tap water samples. Te percentage recoveries presented in Table 3 showed that the proposed sensor can successfully detect CC and RS in real water samples.

Conclusions
Te role of chlorine on the morphology of nitrogen-doped carbon nanotubes was evaluated by the addition of various amounts of dichlorobenzene, and the resultant materials were coated onto the surface of a GCE for use as electrochemical detectors. Large diameter NCNTs were produced when the amount of chlorine in the feed was lower, and small diameter NCNTs were produced when the amount of chlorine in the feed was larger than that of the nitrogen source. Large diameter NCNTs coated electrode showed a wide linear range for studied concentrations and a low detection limit for the simultaneous detection of CC and RS which was attributed to them being highly graphitic. It was suggested that the addition of a low concentration of chlorine increased the carbon difusivity in the catalyst, which resulted in the formation of compartmented larger diameter NCNTs, whereas when a large concentration of chlorine was added in the feed, etching of iron by chlorine occurred resulting in the formation of materials with smaller average diameters and was highly disordered. Te low detection limits obtained that were comparable to those obtained for heteroatom-doped CNTs in the literature showed   that the chlorinated NCNTs (2 : 1) composite could be employed as an electrode material for the development of electrochemical sensors. Te practical applicability of the detector was also shown by catechol and resorcinol detection in real water samples.

Data Availability
All data that support the fndings of this study are included in the article (and additional data are provided in the supplementary fles).

Conflicts of Interest
Te authors declare that there are no conficts of interest.