Fabrication of chlorine nitrogen co-doped carbon nanomaterials by an injection catalytic vapor deposition method

The synthesis of both covalently bonded chlorine and nitrogen-doped carbon materials (Cl-N-CNMs) has been little studied. In this paper we report on the investigation of the synthesis of Cl-N-CNMs using a feedstock containing a mixture of dichlorobenzene (DCB), acetylene and acetonitrile over a Fe-Co/CaCO3 catalyst using an injection CVD method at 800 °C. By varying the acetonitrile:DCB concentration ratio (66.7:33.3; 33.3:66.7 and 20:80), the morphology and physicochemical properties of the CNMs was varied. The products contained varying amounts of Cl (0.5%–1.2%) and N (0.88%–1.47%) and the total amount of Cl and N increased with the Cl content in the feed, as determined by XPS. A graphitic N environment dominated in feeds containing 33.3 and 66.7 vol.% DCB, whilst pyrrolic N dominated in feeds containing pure acetonitrile and 80 vol.% DCB. The chlorine in the feed promoted the formation of CNMs with various shapes namely horn-shaped, spaghetti-like, and pencil-like shapes, some with open-ends and others with closed-ends as determined by TEM and SEM studies. Although no direct correlation with the amounts of the reactants used and the morphology of the products was established, trends in the product shapes were noted with highly defected products produced from 66.7 vol.% DCB, and feeds containing 33.3 and 80 vol.% had tubes with similar open-ended horn-shaped morphology and less defects.


Introduction
Carbon nanotubes (CNTs) with controlled morphology are well known and the CNT morphology has been shown to impact on the physicochemical properties of the CNTs. Further, foreign atoms (dopants) can be introduced into the CNT structure. The addition of foreign atoms into the CNTs can lead to either functionalization of the CNT walls or substitution of a carbon atom in the CNT wall.
Doping carbon nanomaterials (CNMs) with nitrogen atoms has been shown to enhance the application of the carbons in various fields. These include, their use as polymer fillers in composite materials to enhance their strength or conductivity [1], supports in catalysis [2,3], adsorbents in water purification [4], adsorbents for CO 2 [5], and as storage materials for energy (lithium-ion batteries and supercapacitors) [6]. Indeed, much work has appeared on the use of N-doped carbon nanomaterials (N-CNMs) as electrode materials in supercapacitors [6][7][8][9][10][11]. Use of N-doped CNMs as supercapacitors has been encouraged by the high surface area, excellent electronic conductivity, high reversibility and eco-friendliness of the doped carbons [12]. The beneficial effect of nitrogen has been related to the specific bond types formed with carbon [13]. The have been reported the magnetic properties of the N-doped CNMs [14].
Incorporation of halogens and other atoms into the carbon structures have been found to enhance their electrical and thermal conductivity. Decoration of CNT fibers with iodine was found to significantly raise the inter-tube interfacial electrical and thermal transport and thus boosting the overall electrical and thermal transport performance of the fibers [15]. Electrical and thermal properties of CNT films were greatly enhanced after doping them with mild and safe iodonium salts [16]. The interfacial electrical and thermal transport within the inter-tube interface was greatly improved when using carbon nanotubes fibers doped with gold (Au) nanoparticles [17]. Chemical doping of reduced graphene oxide (rGO) sheets with chlorine introduced extra charge carriers into the rGO and these were found to enhance the rGO electrical conductivity [18]. Chlorinated rGO sheets were found to be highly stable in solvents like N,N-dimethylformamide, possibly because of the enhanced repulsive forces between the graphene oxide sheets arising from the absorbed Cl ions [19]. Yeon et al found that etching a polymer derived SiCN with chlorine resulted in the formation of a carbon material that exhibited meso-and micro-porosity and possessed a high specific surface area of 800-2400 m 2 g -1 [20]. Zera et al synthesized a nitrogen-doped carbide derived carbon aerogel with high capacitance by chlorine etching of a SiCN aerogel [13]. In a previous study by our group, the effect of chlorine on the morphology of CNMs prepared by a catalytic pyrolysis of a range of chlorinated organic hydrocarbons using C 2 H 2 over an Fe-Co/CaCO 3 catalyst was studied [21]. Here it was shown that the type of chlorinated organic reagent used influenced the morphology of the CNMs [21]. Chlorinated holey double-walled carbon nanotubes (DWCNTs) which showed high repeatable response to humid environment and a good reversible behavior after sensor purging by dry air were synthesized by etching the walls of DWCNTs by hot concentrated sulphuric acid, followed by saturating the edge carbon sites with chlorine via CCl 4 vapor [22].
Interestingly, the role of chlorine (and chloride salts) on the morphology of N-doped CNTs have only been the subject of a few studies. For example, 30% of N-doped horn-like CNTs with dumbbell-shaped open-ends were prepared by reducing pentachloropyridine with metallic sodium in a stainless steel autoclave; the reaction also produced hollow carbon nanospheres [23]. Various N-doped carbon nanostructures including particles, whiskers, square frameworks, lamellar layers, hollow spheres and tubular structures have been synthesized by designed chemical reactions of carbon halides (such as CCl 4 , C 2 Cl 6 ) and nitridation reagents such as NaN 3. This reaction occurred in the absence of any templates and catalysts in an autoclave at various temperatures [24]. The rupture of thick CNTs with arms or branches was observed when using a modified pyrolysis CVD method, where sodium chloride was added into a by-product liquid trap [25]. The nitrogen content was increased in N-doped CNTs synthesized using halogenated ferrocenyl catalysts [26]. A fluorine substituted ferrocene catalyst produced a higher nitrogen-doping level in N-CNTs as compared to a chlorine substituted ferrocene catalyst, but the chlorine substituted catalyst also yielded iron-filled N-CNTs [26]. Boron chlorine CNTs (BClCNTs), which presented excellent oxygen reduction reactions (ORR) performance, were successively prepared from metal-free substrates via chemical tailoring of two-dimensional boron carbide (B 4 C) with Cl 2 [27]. Changes in electronic and magnetic properties of nitrogen doped CNTs were observed after functionalizing them with chlorine in chlorine plasma atmosphere and oxygen. Experimental results and theoretical calculations showed that chlorine bonded compounds (i.e. C-Cl, C-N-Cl) formed on the surface of nitrogenated CNTs [28]. Functionalization of the N doped CNTs with chlorine and oxygen (N-CNTs:Cl and N-CNTs:O), revealed that the density and the length of the nanotubes decreased on chlorination but increased on oxidation whilst the field emission properties were enhanced in N-CNTs:Cl but reduced in N-CNTs:O [29]. From the above reports it can be observed that simple routes to make both N and Cl doped CNTs have only resurfaced recently, from the two reports by Ray et al [28,29], In this study, the effect of chlorine on the morphology of N-doped CNTs made with 1,2-dichlorobenzene (DCB) using an injection CVD method was investigated. The goal was to evaluate the modification of N-doped CNTs with chlorine. To the best of our knowledge, the role of chlorine on the morphology of N-CNTs using a Fe-Co/CaCO 3 catalyst and a simple injection CVD method has not been studied previously. The concentration of chlorine was varied to explore its effect on the morphology of the N-doped CNTs. It was observed that addition of both CH 3 CN and DCB (together with C 2 H 2 ) gave products that contained both covalent Cl and N in the new carbons. The results were compared with a previous report that showed that CNTs produced from the pyrolysis of DCB without N showed secondary CNT growth structures [30].

Experimental section
2.1. Synthesis of the catalyst [31] Fe(NO 3 ) 3 ·9H 2 O and Co(NO 3 ) 2 ·6H 2 O were used to prepare the catalyst. Calculated amounts of the Fe and Co nitrates were weighed and mixed in separate beakers. The salts were then dissolved in 30 ml of distilled water to make 0.3 mol l −1 Fe and 0.3 mol l −1 Co precursor solutions. The metal solutions were mixed, transferred to a burette, and added dropwise to a 10 g CaCO 3 support that was placed in a beaker with constant stirring. The mixture was left stirring for 30 min. The beaker containing the metal-support mixture was then dried in an oven at 120°C for 12 h. The solid was then cooled to room temperature, ground using a mortar and pestle, and followed by screening through a 150 μm molecular sieve. The catalyst powder was then calcined at 400°C for 16 h in a static air oven. This catalyst was characterized as reported in an earlier article [31].

Synthesis of chlorinated CNMs and N-doped CNMs using pure DCB and CH 3 CN
Synthesis of chlorinated CNMs and N-doped CNMs was carried out at 800°C in a CVD furnace in the presence of a mixture of N 2 and C 2 H 2 gases with flow rates of 240 and 90 ml min −1 for N 2 and C 2 H 2, respectively. Both N 2 and C 2 H 2 were used as carrier gases with C 2 H 2 also as the carbon-containing reactant. The dichlorobenzene (DCB) or acetonitrile (CH 3 CN) solvent, or solvent mixture (20 ml) was placed into a 20 ml syringe driven by a SAGE syringe pump. About 1 g of the synthesized catalyst was placed in a quartz boat that was then inserted into the middle of the quartz tube reactor (32 cm diam×1 m length). The quartz tube was then placed inside a furnace with the boat positioned at the center of the furnace at room temperature and then set to start heating. After the temperature of the furnace had reached 800°C, the solvent was injected at a rate of 0.24 ml min −1 into the quartz tube reactor using a peristaltic pump. The reaction time was 1 h. At the end of the reaction, the furnace was cooled down to room temperature under a N 2 atmosphere (40 ml min −1 ). The formed carbon soot was removed from the tube and weighed. The carbon product produced from CH 3 CN was purified by refluxing in 30% HNO 3 at 110°C for 4 h. The carbon product produced from DCB was purified using mild conditions (stirring in 30% HNO 3 acid at room temperature) since harsher temperatures destroyed the product. Both carbon products were then filtered and washed with distilled water until the pH of the filtrates reached ∼7. The carbon products were then dried in an oven at 120°C overnight. The schematic representation of the experimental setup is presented in supplementary figure S1 (available online at stacks.iop.org//0/000000/ mmedia).

Synthesis of chlorinated N-CNMs using DCB/CH 3 CN mixtures
A similar experiment to that used in section 2.2 was followed, but in this case a 20 ml mixture of CH 3 CN and DCB of various volume ratios were placed in a 20 ml syringe driven by a SAGE syringe pump. The carbon soot produced was removed from the tube, weighed and some of the product purified by refluxing in 30% HNO 3 at 110°C for 4 h. The carbon product was then filtered and washed with distilled water until the pH of the filtrates reached ∼7. The carbon product was then dried in an oven at 120°C overnight.

Characterization of the carbon nanomaterials
The morphology and size distribution of the produced CNMs before and after treatment with HNO 3 acid were analyzed by transmission electron microscopy (TEM) using a T12 FEI TECNAI G 2 SPIRIT operating at 120 kV. The samples for TEM analysis were prepared by sonication in ethanol and thereafter deposited on a holey carbon coated TEM Cu grid. The morphology and size distribution of the CNMs were also determined by scanning electron microscopy (SEM) using a FEI Nova Nanolab instrument. The powdered samples were placed on a tape that was attached to a stub. The samples were coated with carbon and palladium to prevent them from charging. The graphitic nature of the CNMs was characterized by Raman spectroscopy using a Jobin-Yvon T6400 micro-Raman spectrometer. Excitation was provided by the 532 nm green laser with spectral resolution of 3-5 cm -1 . X-ray photoelectron spectroscopy (XPS) analysis was done using an AXIS Ultra DLD with an Al (monochromatic) anode equipped with a charge neutralizer, supplied by Kratos Analytical (UK) at Rhodes University.

Results and discussion
The morphology of the CNMs was first evaluated using pure CH 3 CN and pure DCB (in C 2 H 2 /N 2 ), to obtain base-line data for the study.
TEM images of purified CNMs obtained from pure CH 3 CN revealed the presence of mixtures of CNMs, hollow thin-walled CNTs (figure 1(a)), bamboo compartmented CNTs (supplementary figure S2(a)) and carbon nanospheres (CNSs) ( figure 1(b)). A rod-shaped carbon nanofiber (CNF) with a cylinder-like tip or base (supplementary figure S2(b) circled part) was also observed from the TEM images.
SEM images of the nanomaterials generated from CH 3 CN revealed that the majority of the CNMs were 'rodshaped' (figure 1(c) and supplementary figure S2(c)). A diameter distribution curve showed that a majority of the rod-shaped CNMs had an average outer diameter of ∼53 nm (figure 1(d) and table 1). The rod-shaped CNMs grew from or were attached to some large, agglomerated particles (figure 1(c)). The agglomerates were previously observed in other studies e.g., during pyrolysis of acetonitrile using a Mg-Co-Al layered hydroxide as a catalyst material wherein carbon plate-like particles were observed [32]. The length of the rod-shaped CNMs ranged from ∼60 to 160 nm. A few spherical CNMs (supplementary figure S2(d) arrow) and a few long CNTs were also observed from SEM images (supplementary figure S2(d); circled part). No secondary growth of carbon on the rod-like materials was observed when pure acetonitrile was used.
TEM and SEM images of purified CNMs prepared from pure dichlorobenzene (DCB) revealed mixtures of entangled thin-and thick-walled CNTs (figures 2(a)-(c)). The tube tips and bases were not clearly observed from the TEM images due to entanglement and the great length of the tubes. The density and length of the CNTs increased when DCB was used as a carbon source instead of CH 3 CN. This must be due to the catalyst used or how it was used in this study. Previous studies have shown growth of large density of bamboo-compartmented CNTs, when a mixture of acetonitrile and ferrocene were used as a source nitrogen, carbon and catalyst by an injection-vertical CVD method [33]. Secondary CNF growth was not observed from the TEM images, as were previously observed during pyrolysis of DCB using a bubbling CVD method [21,30]. A diameter distribution curve revealed that the CNTs had an average outer diameter of ∼45 nm (figure 2(d) and table 1). Growth of a large quantity of highly entangled CNTs in the form of a CNT sponge was also reported in the literature from spray pyrolysis of a mixture of ferrocene and DCB [34].
The effect of using DCB/CH 3 CN solvent mixtures of various ratios on the morphology of N-doped CNMs was also studied by TEM and SEM. The concentrations, based on the volume percent of nitrogen and chlorine sources used were: CH 3 (b)). Previous studies have shown that acid treatment opens the tube ends [35], but in our study the tube-ends were open even before acid treatment. This could be one of the roles of chlorine; to act as an oxidizing agent leading to opening of the tube ends. Doubling the DCB concentration in the feed to 66.7 vol.% resulted in formation of CNTs and CNFs with various sizes (supplementary figure S3(b)). A similarity was again observed from CNMs generated from both unpurified (supplementary figures S4(c) and (d)) and purified samples. Rod-shaped CNTs, some with closed-ends and others with open-ends were obtained. Large-sized CNFs with pencil-shapes, some with open-ends and others with closed-ends were also obtained (supplementary figure S3(b) circled parts). A high magnification TEM image of the large, circled tube in supplementary figure S3(b) shows that the materials are fibrous, and a catalyst particle can be seen embedded at the pencil-like tip of the CNF (supplementary figure S3(c)). Previous studies have explained this phenomenon as growth termination of CNT due to the full coverage of the catalyst by the carbon layers [36]. Open-ended CNTs with no catalyst particle at their tip were also observed from the materials obtained at DCB concentrations of 33.3 and 66.7 vol.% (supplementary figures S3(a) and (b) circled). The type of growth termination that took place here can be explained as out-migration of the catalyst from the growing nanotube induced by weakened adhesion strength between catalyst and CNT [36]. Another scenario for growth termination due to Ostwald ripening is observed from the TEM images of the un-purified CNTs generated from feeds containing 33.3 vol.% DCB (supplementary figure S4(b) circled) [36]. A further increase in DCB concentration to 80 vol.% DCB resulted in formation of a large quantity of entangled uniform CNTs ( figure S3(d)), which could be attributed to the presence of a large quantity of chlorine in the feed. Enhanced CNT growth, as was observed when pure DCB was used as a feed (figures 2(a)-(c)), was noted.
SEM images were also recorded to obtain more information about the morphology and the dominating carbon nanostructures formed from the different feeds. Images obtained from un-purified samples containing 33.3 vol.% DCB showed the presence of rod-shaped CNMs, covered by droplets of metal catalyst particles (supplementary figures S5(a) and (b)). SEM images of the purified samples revealed that the majority of the  figure S6 and table 1). A measurable quantity of the rod-shaped CNMs was also observed with this feed, with an average diameter distribution of ∼78 +/− 53 nm (supplementary figures S7(c) and (d)). The open-ended CNTs were 'horn-shaped' (supplementary figure S7(c)) and the rod-shaped CNMs appeared to have increased in length in comparison to those obtained from pure CH 3 CN (supplementary figure S2(c)), with the measured lengths of ∼350 to 540 nm; this is about three times the value obtained in pure CH 3 CN. This suggests that the presence of chlorine enhanced the growth of the CNMs. This must be an effect of the synthesis method and the type of catalyst used because Ray observed a decrease in CNT density and length after chlorination of N-doped CNTs using an inductively plasma-coupled reactor in flowing Cl gas [29]. The diffusion and precipitation of the reactive carbon species was faster when chlorine was added resulting in faster growth rate and increase in nanotubes length and density [29]. A measurable quantity of the carbon nanospheres was also obtained from this feed solution (supplementary figure S7(d)).
SEM images obtained from un-purified samples containing 66.7 vol.% DCB also showed the presence of rod-shaped CNMs, covered by droplets of metal catalyst particles (supplementary figures S5(c) and (d)). SEM images of the purified samples revealed that the majority of the CNMs in this feed were 'pencil-shaped' CNFs with most of them showing small un-defined tip-ends (figures 3(c) and (d)). The average outer diameter of the observed CNMs was about 154 nm (supplementary figure S6(b) and table 1). Diameters of the nanotubes were also slightly increased in another study after chlorination of N doped CNTs [29]. Some of the CNFs were very large with 'pencil-shaped' morphologies, some had closed-ends whilst others had open-ends (some appear to have broken tips) (supplementary figure S8). An almost full CNM can also be seen in supplementary figure S8(c) and it has both ends narrowed in a form of a 'pencil' shape. The increase in the amount of chlorine in the acetonitrile feed resulted in structural changes from the 'rod-shaped' CNMs to the 'pencil-shaped' CNFs. There was also an increase in the thickness of the CNFs and the increased narrowing of the thick tubes with an increase in the amount of chlorine in the feed (figure 3(c); circled part and supplementary figures S8(b) and (c)). Tubeend narrowing was also observed from the secondary grown CNFs in our previous studies, during pyrolysis of DCB using a bubbling CVD method [21,30]. CNT tube narrowing at their tips was also observed by other authors and they attributed it to tip growth termination of MWCNTs [37]. The reaction mechanism to form the 'pencil-shaped' CNTs is still unknown.  sp 3 -hybridized C-C bonds which may arise from grain boundaries, vacancies, pentagons, heptagons and graphene edges [38]. An amorphous carbon D2 band was observed from the spectra of CNTs generated from feeds containing 33.3 and 66.7 vol.% DCB. The assignment of all the observed Raman bands is given in supplementary table S1.
The intensity ratio of the D and G (I D /I G ) Raman bands, using a Lorentzian fit, were 0.68, 0.72, 1.8 and 0.63 for purified CNMs generated from solutions containing pure CH 3 CN, and 33.3, 66.7 and 80 vol.% DCB in CH 3 CN respectively. The degree of disorder increased with an increase in the amount of chlorine in the feed from 33.3 to 66.7 vol.% but decreased for materials prepared from an 80 vol.% DCB feed solution. This shows that the graphitic nature of the carbon materials was affected by the amount of chlorine, and that more than one factor is at play in establishing the morphology of the N-doped carbon. (Note-the effect is not linear as one factor plays a role at low concentration, another at 80%). Only a small decrease in the peak ratio from 0.68 (pure CH 3 CN) to 0.63 (80 vol.% DCB) was observed indicating that both these materials have similar graphic nature. It is envisaged that at DCB concentration of 80 vol.% there is an increase in the supply of chlorine vapors into the feed. This situation results in the formation of Cl 2 molecules which then functionalize the CNTs and also acts as purifying agents when they interact with the CNTs, resulting in the formation of cleaner, highly graphitic CNTs [39]. Formation of Cl 2 molecules from two individual Cl atoms is energetically favorable when the two Cl atoms are at close proximity to each other [40].
The I D /I D′ ratio was also evaluated to provide information about the type of defects present in the materials. Large I D /I D′ values were obtained for the purified CNTs which decreased in a consistent manner with DCB feed content (I D /I D′ =16, 9.3 and 3.7 for 33.3, 66.7 and 80 vol.% DCB). These values are related to the sp 3 content (16), a mixture of vacancy-like and hoping defects (9.3) and boundary-like (3.7) defects, respectively [41][42][43].
Raman bands obtained from CNMs generated from pure DCB also showed a highly intense D band that indicates the existence of disorder ( figure 6). A large I D /I G value of 1.5 also suggested that the materials had disordered carbon. The most plausible explanation is that the presence ofchlorine created disorder due to functionalization of the outer wall of the graphene sheet, but the phenomena depends on the amount of chlorine in the feed.
Second order bands were also observed from purified CNMs generated from solutions containing 33.3 and 80 vol.% DCB. A G * band at ∼2400 and a 2D band split into two bands (figures 6(b) and (e)) were also noted. A defect induced G * band was very weak in both samples. A weak 2D band for CNMs generated from feeds containing 33.3 vol.% DCB was very intense for CNMs generated from feeds containing 80 vol.% DCB CNMs. The data suggest that the latter materials were highly graphitic, which is in agreement with the I D /I G peak ratio. Another weak defect induced band (D+G band) at ∼2900 cm -1 was observed from purified CNMs generated from solutions containing 80 vol.% DCB ( figure 6(e)).
The degree of disorder was also related to the concentration of DCB in the feed and was found to increase tremendously with an increase in DCB concentration from 33.3 (I D /I G =0.72) to 66.7 (I D /I G =1.8) vol.%, but decreased when the DCB concentration was increased further to 80 vol.% (I D /I G =0.63) ( figure 7(a)). Among all studied products, chlorinated nitrogen doped CNTs generated in feeds containing 66.7 vol.% DCB had the largest amount of defects that are conducive to electrocatalytic processess [44]. This data agrees with the TEM analysis where CNMs of various shapes, and sizes were obtained from feeds containing 66.7 vol.% DCB due to increased disorders in the CNMs. The data also agrees with the results obtained by Kou et al for their BClCNTs where the degree of defects was highest in feeds containing 6.6% Cl (studied feeds were 3.3, 6.6 and 9.9 Cl ratios) [27]. To further explain the formation of highly graphitic materials for acetonitrile feeds that contained a large concentration of chlorine, we compared the wavelength of the D and G peaks. The D and G bands blue shifted after addition of 80 vol.% DCB as compared to those obtained in feeds containing 33.3 vol.% DCB ( figure 7(b)). According to literature the blue shift is due to decreased doping of an atom in quantity, which in our case will be decreased doping of nitrogen [27]. The data is consistent with our ealier conclusions that at high DCB concentrations of 80 vol.% most of the chlorine enters the reactor as Cl 2 molecules which act as purifying agents. However, the D and G bands red shifted for feeds containing 66.7 vol.% DCB, which suggests p-type doping and increased doping, this suggests doping or functionalization of material with both chlorine and nitrogen.
XPS data were collected on all the samples generated from CH 3 CN, DCB and CH 3 CN:DCB with various ratios (supplementray figures S10 and S11). All spectra revealed peaks for C, O, N and Cl. The spectra for the Clcontaining samples are shown in figure 8. The Cl spectra contained two peaks that are associated with the Cl2p 3/2 and Cl2p 1/2 binding energies, due to the spin-orbit splitting of the Cl2p core level (0.5 ratio and 1.6 eV separation) [45][46][47]. The two peaks centred at 200.1 eV and 201.7 eV are associated with covalent C-Cl (and/or   Cl-C=O bonds) [45][46][47], while the two peaks at lower energies (199.6 eV and 198.1 eV) are associated with ionic Cl. It is thus clear that the synthesis method produced covalent C-Cl bonds in all cases. The total amount of Cl present in the samples and the distribution of covalent and ionic chlorine on the CNMs are shown in table 2 and in supplementary table S2. The total amount of chlorine in the samples increased with an increase in the amount of DCB in the CH 3 CN feeds. The C1s XPS data also confirmed the functionalization of carbon with chlorine as evidenced by the presence of C-Cl peak at ∼287 to 289 eV (supplementary figure S12).
The data shows that C-Cl bonds were formed on the CNT surface. The bonds could be formed by attack of Cl radicals or by attack from Cl 2 [48]. The C-Cl bonds are expected to be formed by reaction of carbon defects or dangling bonds with chlorine to form covalent bonds with the carbon.
The presence of the ionic chlorine could be due to incomplete removal of the catalyst during acid treatment which resulted in formation of metal chlorides due to the reaction of the Cl with the Fe/Co. The percentage of metal residues present in the purified samples after TGA analysis are presented in supplementary table S3. The residual mass left after purification was 0, 5.0, 7.0 and 1.8 for pure DCB, 33.3, 66.7 and 80 vol.% DCB respectively (supplementary table S3) which paralleled the ionic chloride content. Also a shift to higher to higher temperatures for the first peak due to oxidation of CNTs was observed for materials generated from feeds containing 80 vol.% DCB which suggest removal of amorphous carbon and metal content (supplementary table  S3). The data agrees with the low I D /I G value obtained from Raman spectroscopy.
Analysis of the N XPS data for pure CH 3 CN and mixtures containing 33.3, 66.7 and 80 vol.% DCB, respectively gave N incorporation amounts as 1.31, 1.57, 0.88 and 1.47 at.% (table 3). For all samples, the N 1s peak was deconvoluted into four peaks at 399.3, 401, 402.7 and 405.8 eV attributed to pyrrolic (N pyr ), quaternary (N Q ), oxidized (N Ox ) and molecular (N Mo ) nitrogen respectively (figure 9). C1s spectra was also deconvoluted and it confirmed the presence of C-N groups (supplementary figure S13).
An additional peak appearing at 397.6 eV assigned to pyridinic nitrogen (N P ) appeared in the N 1 s spectra of CNTs generated from feeds containing 33.3 vol.% DCB (figure 9). In our earlier studies chlorine was found to introduce on-site defects on the walls of the CNTs [30], and the presence of these defects can possibly provide an easy route for nitrogen to be added onto the walls of the CNTs substituting some of the carbon atoms.
A pyridic nitrogen peak was only observed in feeds that contained DCB and can be ascribed to chlorine induced edge defects (figure 9). A decrease in the pyrrolic and molecular nitrogen peak ratios was observed after addition of 33.3 and 66.7 vol.% DCB, which suggests rearrangement of the carbon and nitrogen atoms within the carbon lattice, which is enhanced in the presence of chlorine (figure 10). A quaternary or graphitic nitrogen peak concentration increased tremendously (thrice the amount obtained in pure CH 3 CN) after addition of 33.3 vol.% DCB suggesting increased substitution of carbon atoms with nitrogen into the CNT walls (table 2, figures 9 and 10). The concentration of the quaternary N peak after addition of 66.7 vol.% DCB decreased and was the same intensity as of that obtained in pure CH 3 CN (table 3, figures 9 and 10). This could be due to competition between chlorine functionalization of carbon and nitrogen doping, which was affirmed by Raman data where a high I D /I G of ∼1.8 and p-type bonding was obtained from feeds containing 66.7 vol.% DCB as compared to I D /I G values of 0.63 and 0.72 obtained from feeds containing pure acetonitrile and 33.3 vol.% DCB. An increase in the concentration of quaternary N (twice the amount obtained in pure CH 3 CN) was also observed from feeds containing 80 vol.% DCB (table 3, figures 9 and 10). Pyrrolic N also increased greatly in feeds containing 80 vol.%, authors have suggested that pyrrolic N arises due to N substitution in a Stone-Wales defect [49,50] or due to asymmetric local bonding [51]. Highly defected CNMs contained lower amounts of graphitic nitrogen incorporated into the CNTs (figure 10).

Conclusion
The synthesis of chlorine-functionalized nitrogen-doped and un-doped CNMs using an injection CVD method and a Fe-Co/CaCO 3 catalyst was successful. DCB was found to be a good chlorine source and that impacted on the nitrogen doping into the CNTs. Chlorine influenced the morphology, length, and outer diameters of the N-doped CNMs, and enhanced their growth. A limited quantity of short rod-shaped CNMs were obtained when pure CH 3 CN was used as a feed solution, whilst highly entangled CNTs were obtained when pure DCB was used as a feed. Variation of the amount of DCB in the CH 3 CN feed led to horn-, straw-, spaghetti-like and pencilshaped open-ended and closed-ended N-doped CNTs and CNFs. Various scenarios of for the growth termination of CNTs were observed i.e., (i) growth termination of CNTs due to ful coverage of catalyst by carbon  layers, (ii) due to out-migration of the catalyst from the growing CNT induced by Ostwald ripening and weakened adhesion strength between the catalysts and the CNT. Defects associated with sp 3 hybridization, and a combination of vacancy-like, hopping defects and boundary like defects were obtained from Raman analysis of materials generated from feed solutions containing 33.3, 66.7 and 80 vol.% DCB, respectively. Graphitic nitrogen species dominated from the XPS data of CNMs generated from feeds containing 33.3 and 80 vol.% DCB, which correlates with the Raman data where low I D /I G values below 1 were obtained. The highests amount of chlorine that is covalently bonded to carbon was observed in acetonitrile feeds that contained the largest amount of DCB. It appears that the presence of nitrogen created defects in the graphene sheet due to nitrogen doping, while chlorine created disorder due to functionalization of the outer wall of the graphene sheet. Chlorine also acted as a purifying agent for nitrogen feeds that contained the largest amount of chlorine. This work has shown that it is possible to modify the morphology of N-doped CNMs by adding chlorine to the N source. The open-ended CNTs generated from chlorine-nitrogen feed mixtures could potentially be applied in field emission devices, as supercapacitor electrodes, for an electrochemical storage of energy, and as electro catalyst for oxygen reduction reactions.