Boron and nitrogen doping in graphene: an experimental and density functional theory (DFT) study

Boron (B) and Nitrogen (N) doped few layer graphene (BNG) is directly synthesized via electric arc discharge (EAD) method. NH3 and BCl3 gas mixtures are used in the synthesis atmosphere. Raman spectroscopy is used to determine graphene’s purity and number of layers. The investigation of structure and morphology of pristine graphene and BNG are carried out via Transmission Electron Microscopy (TEM). The presence of B and N in the structure of graphene is detected by Energy Dispersive X-ray Spectroscopy (EDS) analysis. Elemental mapping show that N and B are distributed homogeneously in the graphene structure. It is observed that doping process did not affect the positions of the D, G and 2D bands in the Raman spectroscopy. The effect of doping on the number of layers of graphene is found negligible. TEM results exhibit that pristine graphene and BNG have 5 to 6 layers. Besides, the theoretical calculations based on Density Functional Theory (DFT) are employed to support experimental studies. Theoretical results based on DFT showed that bonding of B and N is favorable.

Synthesis of graphene can be classified into two different approaches: bottom-up and top-down. For top down, graphene is obtained directly from graphite. Arc discharge, solution-based exfoliation, micromechanical cleavage, electrochemical exfoliation, exfoliation of graphite intercalation compounds, exfoliation of graphite oxide, evaporation with laser and unzipping carbon nanotubes are some of the top down synthesis methods. On the other hand, in the bottom-up approach, small organic molecules are used to form graphene layers via catalytic processes. Chemical vapour deposition (CVD), epitaxial growth on SiC and SiC degradation on high temperature are the most commonly used bottom-up synthesis methods. The arc discharge method have some advantages comparing to other methods due to its low cost, high efficiency and no need for any catalyst. Also, graphene which is synthesized by arc discharge has good dispersibility in organic solvent and high thermal stability. In the EAD method, many parameters such as carbon sources, buffer gases, pressure,catalysts current/ voltage, electrode morphologies /diameters affect the purity and number of layers of synthesized graphene [6][7][8][9][10].
Chemical doping is a commonly used method for modifying the properties of materials. The elemental composition, surface and electronic properties of the material can be altered by chemical doping process. For carbon based materials, chemical doping is aiming to increase electrical and thermal conductivity [11]. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Generally, there are two techniques to dope graphene: (1) the adsorption of the selected materials to the graphene surface, (2) substitutional doping of specific atoms into the lattice structure of graphene [4].
Various elements such as Nitrogen (N), Sulfur (S), Boron (B), Phospor (P), Oxygen (O) and Fluorine (F) can be doped into graphene. B and N are mostly preferred chemical dopants for graphene since they have similar atomic dimensions with that of C and moreover they can form strong bonds with C. B and N doped graphene can be used in several applications area such as fuel cells, sensors, spin filter devices and lithium ion batteries [2,3,12].
There are many process used for the direct synthesis of B and N doped graphene such as EAD, CVD, solvothermal method, segregation growth. According to other methods, arc discharge is suitable for mass production. Also this method provides in situ substitution of atoms such as N and B. during graphene synthesis process. The disadvantages of the method are the high current/voltage requirement, low level of doping and mostly the multi-layered graphene production [13][14][15].
Zou et al synthesized N doped graphene nano flakes by arc discharge. Doping process is increasing the static magnetization property while decreasing the electrical conductivity. Also graphene lattices substantially reduce crystallization degree. Dey et al synthesized B and N doped graphene quantum dots by two-step chemical processes consisting of arc discharge and chemical scissoring. Li et al reported that synthesis of large scale N doped multilayer (2-6 layers) graphene via arc discharge in the He and NH 3 mixing atmosphere. Sedelnikova et al were synthesized carbon products by arc discharge method with the addition of B and/or N. Morphological and electrical properties of synthesized carbon products has been investigated. They observed that boronnitrogen doped carbon products showed the highest electrical conductivity. Wu et al synthesized monolayer graphene doped with N and B via CVD technique by using solid precursors such as urea and boric acid. Synthesized graphene has better electrical and optical properties than plasma treated N and B doped graphene. Bepete et al reported basic methodology for growing large area graphene doped with small BN domains on copper foils using a one-step CVD technique with methane, boric acid powder and nitrogen gas as the C, B and N source. Wang et al developed a simple low-cost approach for mass production of B and N doped graphene with various N and B doping levels by thermal annealing graphene oxide with boric acid under ammonia atmosphere. The resultant samples were showed that good oxygen reduction reaction (ORR) for electrocatalytic activities. Guan et al synthesized few layer N doped graphene nanosheets (GNSs) by arc discharge method under N atmosphere at high temperature. The N doping graphene has been performed at 800°C. They claimed that this method was cheaper and safer than other methods [3,[16][17][18][19][20][21][22].
In this study, BNG is synthesized directly by EAD. NH 3 and BCl 3 gas mixtures are used in the synthesis atmosphere. Raman spectroscopy results of pristine graphene indicates successful synthesis of high purity, 5 to 6 layers of graphene. According to Raman spectroscopy results of BNG, B and N atoms are entered to graphene lattice. The presence of B and N in the structure of graphene are also detected by EDS analysis. Furthermore the crumpled structure of graphene is observed by TEM. Moreover, DFT calculations with B3LYP method have been utilized on doping of B and N into graphene structure.

Material and methods
2.1. Experimental BNG are synthesized using homemade stainless steel arc discharge reactor. High purity graphite rods (99,9%) are used as anode and cathode. Arc discharge reactor is kept under vacuum for 5 min. After the vacuum process, the reactor is filled with NH 3 -BCl 3 gas mixtures and the arc current is kept constant at 150 A. According to the literature available, the key difference is the buffer gases we used in the synthesis of BNG. Formation of BNG occurs at the inner surface of the reactor ( figure 1(b)). All the experimental parameters are given in table 1. BNG synthesis diagram are shown in figure 1(a). The synthesis of high purity and few layers graphene by EAD method are described in a previous study [6].
Raman spectroscopy (Renishaw in via reflex) are used to determine graphene's purity and number of layers. The Raman spectra is measured from 500 to 3000 cm −1 via an excitation wavelength of 532 nm. Structure and morphology investigations of the graphene and Selected Area Diffraction (SAED) are carried out via TEM (JEOL JEM 2100 UHR) at an accelerating voltage of 200 kV. The relative amount of N and B in the structure of synthesized graphene is determined by EDS (Bruker XFlash 6-100) systems. Elemental mapping is performed to determine the distribution of N and B in graphene.

Theoretical work
The theoretical calculations employed in this study have been based on DFT [23]. The software of Gaussian 09 [24] has been used with the B3LYP-Hybrid method [25,26]. The basis set of 6-31G (d,p) has been utilized in calculations for all atoms.
The graphene cluster used for calculations has been modeled as two layers including 60 C atoms (30 C atoms in each layer). Dangling bonds of the C atoms have been saturated with H atoms in order to obtain neutral cluster. In this work, all atoms of the structures except H atoms were relaxed during all DFT calculations. The terminating H atoms have been kept fixed during the calculations. B and N atoms have been doped into each layer of the graphene structure.
Equilibrium geometry (EG) calculations are used to optimize geometries and obtain energies. In present study, energy difference values include zero point energy (ZPE) corrections. These energies are calculated using the frequency keyword (freq=raman) in Single Point Energy (SPE) calculations. In addition, Raman frequency and thermal energy are calculated by SPE calculations at 298 K in Gaussian software [27]. This energy value is computed as follows: where E is the sum of the electronic, zero point and thermal energies. Additionally, the convergence criteria are 12×10 −4 for gradients of root-mean-square (rms) displacement, 18×10 -4 for max displacement, 3×10 −4 for rms force and 45×10 −5 for max force for theoretical calculations utilized in this study.

Raman analysis
We employ Raman spectroscopy technique in order to detect the characteristic bands of synthesized pristine and BNG. D, G and 2D bands are located at 1335 cm −1 , 1573 cm −1 and 2670 cm −1 , respectively for pristine graphene as shown in figure 2(a). D band indicates defects in the structure of carbon based materials due to the lattice distortion introduced by impurity elements. The position and shape of the G band shows doping effect, defects, number of graphene layers and stress. The 2D band is the characteristic band of graphene and its shape is related to the number of layers. L G /L 2D ratio is calculated with Raman spectroscopy results in order to determine the total graphene layers while, L D /L G ratio is used to obtain information about graphene's purity. According to the literature, L D /L G ratio should be between 0 and 1. In a previous work [28], for a single graphene L G /L 2D ratio was reported as 0.25.   Table 2 shows L D /L G and L G /L 2D ratios of pristine graphene and BNG. As seen from figure 2(b), doping process did not affect the positions of the D,G and 2D bands which is similar with previous findings [29].
Graphene layers have a more irregular structure than pristine graphene due to the doping process [30]. Besides, the number of layers of graphene has also increased after doping process. Comparative Raman spectroscopy results for pristine graphene and BNG are shown in figure 3.
An increase in the intensity of the D band of BNG is observed compared to that of pristine graphene ( figure 3). This result indicates that B and N atoms are introduced into the lattice of graphene. On the other hand, B and N doping resulted with a downshift in the relative intensity of the G band. The 2D band is sensitive to doping process and defects. As shown in figure 3a decrease in the relative intensity and a broadening are observed in the 2D band of BNG.  On the other hand, the relative ratio of L D /L G is increased from 0.70 to 1.25. According to the literature available, distortions occur in the graphene structure because of different bond distance in the BNG structure (C-C, C-B, C-N bonds). The L G /L 2D ratio varies between 1.34 and 1.61 for pristine graphene and BNG, respectively. According to L G /L 2D ratio, pristine graphene and BNG have 5 to 6 layers. The effect of doping on the number of layers of graphene is found negligible [30].

TEM
The morphological observations of pristine and doped graphene are performed via TEM. Figure 4 shows 2D planar-like structure for pristine graphene and BNG. The arrow marks in figure 2(c) also indicate the number of layers, which is in good agreement with our Raman results. Graphene has the 0.213∼0.215 nm spacing for the hexagonal diffraction pattern [31]. BNG has ∼0,248 nm spacing due to structures distortion caused by the intercalation of N and B atoms. Also, SAED pattern of BNG shows a ring like diffraction pattern with dispersed bright spots [32]. SAED pattern are consistent with previous studies [12,33]. Elemental mapping is also performed in order to prove the existence of N and B as shown in figure 5. It is observed that both N and B are distributed homogenously on graphene layers [11,34].
Elemental mapping results show that N and B are distributed homogeneously in the graphene structure. As a result of EDS analysis, 69,76% C, 10,14% N and 20,10% B are determined in the structure.

Theoretical calculations
Firstly pristine graphene cluster is optimized by EG calculation. The optimized structure of two layers graphene cluster is represented in figure 6. B and N atoms have been doped into the layers of the graphene structure. Two B atoms and two N atoms have been used for this purpose. In order to simplify the modelling studies have been performed based on the two layers graphene and two B and N atoms. B and N atoms have been doped on different seven positions in two layers graphene structure. These seven positions have been optimized by EG calculations. The optimized geometries for these positions have been presented in figure 7. Table 3 shows energy values including ZPE and thermal corrections for these positions. As can be seen from the table 3, total energy values (including corrections) for position 1 have been calculated to be −2310.3090268 a.u., which designates that doping of B and N atoms into the graphene structure as position 1 (B-N) is most favorable. Moreover, for the position 1, total energies have been computed as 1.71 and 2.08, respectively based on the Raman frequency calculations (by SPE). On the other hand, L D /L G and L G /L 2D ratios for synthesized BNG samples are calculated as 1.25 and 1.61, respectively. The theoretical calculations are performed based on two BNG layers however, the experimentally obtained BNG are 5 to 6 layers. The variation between theoretically