A first-principles study of lithium and sodium storage in two-dimensional graphitic carbon nitride

WANG Meng-yao, LI Jia

(Guangdong Provincial Key Laboratory of Thermal Management Engineering & Materials, Graduate School at Shenzhen, Tsinghua University, Shenzhen518055, China)

Abstract: Two-dimensional carbon nitride is considered a very good battery electrode material owing to its uniform-size pores and the presence of nitrogen atoms. First-principles calculations were used to investigate the adsorption and storage of lithium and sodium on monolayer g-C2N. The capacities of lithium and sodium ion batteries for monolayer g-C2N are 596 (LiC2N) and 276 (NaC4N2) mAh/g, respectively. The average Li binding energy reaches 2.39 eV relative to isolated Li atoms, which suggests that the lithium capacity achieved on g-C2N might not be sustained during cycling. By varying the ratio of C to N atoms, it is found that the average Li binding energy is reduced to only 1.69 eV for C∶N ～ 5∶1, indicating a significant improvement in cycling performance while maintaining the reversible capacity. The mobility barrier energies to Li ion diffusion between two layers in bulk structures with AA and AB stacking sequences are 0.25 and 1.23 eV, respectively, indicating that high Li ion conductivity could be achieved in bulk g-C2N with AA stacking. These calculations demonstrate that graphitic carbon nitride with uniform-size pores can be used as an electrode material with high capacity and high lithium mobility.

Key words: First-principles calculation; Graphitic carbon nitride; Lithium and sodium storage

1 Introduction

Lithium ions batteries (LIBs) have been widely studied over the last decades owing to their excellent battery performance, such as high energy density, light weight, no memory effects and limited self-discharge[1-4]. LIBs are considered as one of the most suitable rechargeable batteries for electric vehicles, portable electronic equipment and other industrial applications. To support the continuous upgrading of these devices, some performance of LIBs needs further improvement, such as energy density, cost, rate performance and durability[5,6]. Natural graphite is the most commonly used as anode material in commercial LIBs because of its low cost, safety and good cycling stability. The theoretical capacity of the graphite anode is 372 mAh/g, which becomes unable to meet the high-capacity requirement for modern electronic devices[7]. With the increasing demand for large-capacity and high-rate LIBs, development of cost-effective new electrode materials remains a grand challenge.

Graphene has attracted a lot of attention for potential use as possible anode materials for LIBs. Graphene nanosheets have been successfully synthesized with controlled interlayer distance and used as the electrode materials for large-capacity LIBs[8]. The capacity is up to 540 mAh/g. In addition, the incorporation of carbon nanotubes and fullerene into the graphene nanosheet can increase the capacity up to 730 and 784 mAh/g, respectively. Moreover, graphene can accommodate a large number of porous defects that can store more lithium ions, thus increasing the capacity. Meanwhile, porous defects may act as transmission channels where lithium ions can migrate among different graphene layers, leading to the fast mobility of lithium ions. In 2010, graphdiyne, a new carbon-based material, has been synthesized with a high yield on Cu foil[9]. Compared with graphite, in which carbon atoms bond through sp2hybridization orbitals, graphdiyne is composed of sp2and sp hybridized carbon atoms with a high void density. Accordingly, extensive efforts have been devoted to the investigation of the potential application of graphdiyne as electrode materials in LIBs. Sun et al.[10]have carried out first-principles calculations and found that the capacity may increase up to 744 mAh/g (LiC3). Zhang et al.[11]found that Li ions can easily pass through voids of the graphdiyne with a low energy barrier of 0.35-0.50 eV, which suggests an excellent mobility of lithium ions. These results obtained by theoretical calculations have been confirmed experimentally recently. Huang et al.[12]reported the application of graphdiyne as electrode materials with the reversible capacities of about 520 mAh/g after 400 cycles at a current density of 500 mA/g. At a higher current density of 2 A/g, the capacities can reach up to 420 mAh/g even after 1 000 cycles. Zhang et al.[13]have also demonstrated a high specific capacity of 552 mAh/g after 200 cycles using graphdiyne powder as LIB anodes.

Besides porosity, heteroatom doping is another strategy to improve the electrochemical performance of the graphene-like membrane as electrode materials. For carbon-based electrode materials, nitrogen doping is considered to be one of the most effective ways to increase the specific capacity with suitable binding energies for ions[14-17]. Among various nitrogen-doped structures, pyridine N is shown to improve both specific capacity and rate performance of LIBs[18-22]. Therefore, graphitic carbon nitride, as an analog of graphite, is regarded as a promising electrode of LIBs owing to its high content of pyridine N and uniform pores[23]. Marlies et al.[24]studied lithium adsorption onto 2D graphitic carbon nitride membranes, C3N4, C6N8and bulk C3N4, and found that lithium ions mainly interacted with the pyridinic N to give rise to a high capacity of Li. However, a high content of pyridine N would also cause the irreversible capacity of Li and the structural instability of 2D graphitic carbon nitride membranes. Recently, 2D graphitic carbon nitride membranes (g-C2N) with less nitrogen content has been synthesized via a simple wet-chemical reaction[25]. Its low content of pyridine N and uniform distribution of pores would be beneficial for its application in the LIBs with a high stability. In this work, employing first-principles calculations, we have investigated the electrochemical performance of monolayer C2N as LIB electrode materials. The energy density of monolayer g-C2N can reach up to 596 mAh/g, with a high average adsorption energy of 2.39 eV. The high adsorption energy could result in the irreversible capacity of Li. By reducing the concentration of the N to 16.7 at.%, the average adsorption energy of Li ions would decrease to 1.69 eV, leading to a favorable cycle performance of battery. Also, the low barrier of Li diffusion through the pores of bulk C2N indicates the good rate performance of battery.

2 Computational details

All calculations were performed using density functional theory implemented in the Viennaabinitiosimulation package (VASP)[26]. The projected augmented wave potential[27]and generalized gradient approximation of the Perdew-Burke-Eznerhof functional[28]were used to describe the electron-ion interaction and exchange-correlation energy, respectively. The DFT-D3 empirical correction method was employed to accurately describe the van der Waals interactions[29]. The cutoff energy was set to 520 eV. The bulk and monolayer g-C2N were simulated by using a 2×2 supercell structure. The Brillouin zones for the bulk and monolayer g-C2N were sampled by 3×3×9 and 3×3×1Γ-centeredk-points.For monolayer of g-C2N, a vacuum layer larger than 2 nm was chosen to eliminate the spurious interaction between the periodically repeated images. For structural relaxation, the convergence for energy and the residual force on each atom were set to 10-5eV and 0.1 eV/nm, respectively. To validate the stability of Li adsorption on g-C2N, the average adsorption energy of Li is defined asEads= (nE(Li)+E(g-C2N)-E(Lin@g-C2N ))/n,wherenis the number of adsorbed Li atoms, andE(Li),E(g-C2N) andE(Lin@g-C2N) are the total energies of a single Li atom, pure monolayer or bulk g-C2N, and monolayer or bulk g-C2N withnLi atom adsorption, respectively.

Fig.1 (a) Geometrical structure of 2×2 supercell for monolayer g-C2N. Geometrical structures of bulk g-C2N with (b) AA and (c) AB stacking. The gray and blue spheres represent carbon and nitrogen atoms, respectively. In (b) and (c), larger black and red spheres represent carbon and nitrogen atoms in the bottom.

3 Results and discussion

The computational lattice constant of g-C2N is a=0.832 9 nm, which is in good agreement with the experimental result (a=0.830 nm)[25]. In order to investigate the lithium storage capacity of g-C2N, we first consider an isolated Li atom adsorption on the monolayer of g-C2N. There are three different stable sites for isolated Li adsorption on the monolayer of g-C2N (see Fig. 1a). The most stable site (site 1) for Li adsorption is in the pore of g-C2N with the adsorption energy ofEads= 4.45 eV, and the strong interaction between Li and surrounding N atoms leads to a shift of Li from the center of the pore, as shown in Fig. 2. The distance between Li and its closest N is about 0.217 nm, and Li is located in the plane of g-C2N without introducing any observable distortion to the g-C2N monolayer structure. For two less stable adsorption sites (site2 and site3), the adsorption energies of Li are 1.38 eV and 1.58 eV, respectively. The larger adsorption energy of Li at site3 originates from the strong interaction between Li and N atoms. And the Li is located above the g-C2N monolayer with the height of 0.181 nm.

Fig. 2 (a) Top (left) and lateral (right) views of one Li atom absorbed in the pore of the monolayer g-C2N, position 1; (b) Top (left) and lateral (right) views of Li1/3C2N; (c) Top (left) and lateral (right) views of twenty-four Li atoms absorbed in the pore of the monolayer g-C2N, six in each pore. Geometrical structures of different carbon nitride monolayer of (d) C52N20H4, (e) C56N16H8, (f) C60N12H12. The white, blue, red, purple, and orange spheres represent carbon, nitrogen, lithium over the membrane, lithium under the membrane and hydrogen, respectively.

When increasing the capacity of Li storage to Li1/6C2N, Li atoms tend to occupy each of the big pores, as shown in Fig. 2a. In this case, the calculated average adsorption energy is 4.59 eV. The little increase in adsorption energy is due to the interaction of dipole moments (Li-N dipole) in neighboring pores. Moreover, there is no distortion in the g-C2N monolayer structure. In consideration of the large steric space, the pores may accommodate more than one Li atom. When one additional Li is introduced to Li1/6C2N, it can stay in one of the pores with the average adsorption energy of 4.19 eV. When the concentration of Li reaches to Li1/3C2N, each pore contains two Li atoms, and Li atoms are located at each side of the pore and interacted with the neighboring three nitrogen atoms of the pore, as shown in Fig. 2b. It can be seen that the pores are large enough to accommodate two lithium ions with no obvious distortion. With the introduction of more Li atoms one by one, we find each pore can hold six Li atoms to reach a loading of 24 Li per supercell, which corresponds to the composite of LiC2N with the specific capacity of 596 mAh/g. It is shown in Fig. 2c that the g-C2N monolayer maintains structural integrity, with a slight deformation. Li atoms are distributed uniformly on both sides of the g-C2N monolayer, with three located at one side of the g-C2N monolayer and the others at another side. All Li atoms are around about 0.1 to 0.14 nm away from the g-C2N monolayer, which is less than half of interlayer space in bulk g-C2N, indicating that a similar capacity could be obtained in bulk g-C2N.

Fig. 3 shows the overall average adsorption energy of the Li atoms plotted against the corresponding amounts of lithium adsorbed on the monolayer g-C2N. When the number of Li atoms adsorbed on the monolayer g-C2N increases from one to four, the average adsorption energy increases from 4.45 eV to 4.59 eV. However, as the number of lithium atoms increases, the average adsorption energies will decrease greatly. When the uptake of Li on g-C2N monolayer reaches to 24 Li per supercell, the average adsorption energy of Li ions is 2.39 eV, still larger than the cohesive energy of lithium (1.63 eV)[30]. In this case, the corresponding cycle performance of LIBs may not be very good. To increase the cycle performance of LIBs, one possible route is to control the carbon-nitrogen ratio of graphitic carbon nitride. By replacing the N atoms in the g-C2N monolayer with isovalent CH groups, we can obtain a series of new structures with different ratios of C and N. Fig. 2(d)-(f) show the structural configurations of C52N20H4, C56N16H8and C60N12H12, which are obtained by substituting one, two and three nitrogen atoms at the large hole sites with CH groups, respectively. All three graphitic carbon nitride composites are found to be able to adsorb up to 24 Li ions. It is shown in Fig. 3 that the average adsorption energies of Li ions in these structures firstly increase a little, then decrease greatly with the increasing number of Li adsorption. More importantly, for the same number of Li, the average adsorption energy decreases gradually with the decrease of nitrogen doping concentration. When the nitrogen concentration is reduced from 33.3 at.% (C2N) to 16.7 at.% (C60N12H12), the average adsorption energy of 24 Li ions is decreased from 2.39 eV to 1.69 eV, which is close to the cohesive energy of Li (1.63 eV). These results suggest that the cycle performance of the battery will be greatly improved when the ratio of C to N reduces to 5∶1. Phonon calculations also show that there is no imaginary frequency for the structure of C60N12H12, indicating the stability of the proposed structure of C60N12H12.

Fig. 3 The average adsorption energy of lithium ions in different carbon nitrogen monolayer with respect to the number of lithium ion adsorption.

To evaluate the rate performance of carbon nitride as anode materials, it is necessary to investigate the mobility of Li ions. The lithium ion migration barrier was calculated by the climbing-image nudged elastic band methods[31]. The different bulk g-C2N structures, AA or AB stacking sequences as shown in Fig. 1b and c, are adopted to study the migration of Li ions. For the AA stacking sequence, the barrier of Li diffusion between two g-C2N layers is 0.25 eV, as shown in Fig. 4a. This barrier is similar as that of Li ions diffusion on the pristine graphene, showing a good ion conductivity of g-C2N with AA stacking sequence. It is shown that the barrier of Li diffusion between two layers of C60N12H12(AA stacking sequences) is 0.29 eV. This value is similar to that in g-C2N (0.25 eV, AA stacking sequences), indicating that the mobility of Li in C60N12H12is as good as that in g-C2N. However, the barrier of Li ion diffusion between two layers in the bulk structure of AB stacking sequence increases to 1.23 eV, which is five times larger than that in AA stacking sequence. Based on the above results, the Li ions have a good mobility when the g-C2N with AA stacking sequences is used for the electrode materials for the large-capacity LIBs.

Fig. 4 Energy profiles for lithium diffusion between the two layers of g-C2N with (a) AA and (b) AB stacking sequences.

Likewise, we also considered the possibility of the monolayer g-C2N material as the anode material of sodium ion battery. Fig. 5 shows the overall average adsorption energies of the sodium atoms as a function of the number of adsorbed sodium. It is shown that the average adsorption energies of sodium ions increase a little first, then decrease greatly with the increasing number of Na adsorption. When the uptake of sodium on g-C2N monolayer reaches to 12 Na per supercell, the average adsorption energy of Na ions is 2.27 eV, still larger than the cohesive energy of sodium (2.18 eV).

Fig. 5 The average adsorption energy of sodium ions in g-C2N with respect to the number of sodium ion adsorption.

While a sodium ion was adsorbed again, the average adsorption energy of sodium ions was changed to 2.12 eV, less than the cohesive energy of sodium (2.18 eV). Thus, the maximum number of sodium ion adsorption in this case can only reach 12, and the corresponding sodium ion battery capacity was 276 mAh/g (NaC4N2).

4 Conclusions

The adsorption of Li atoms on monolayer and the diffusion of Li atoms in bulk g-C2N have been investigated. With the presence of the nitrogen atoms and homogeneous pores, the adsorption energy of lithium atoms is much higher than that of primitive graphene. When 24 lithium atoms are adsorbed in each 2×2 super cell, the average adsorption energy of Li is still much higher than the cohesive energy of lithium. In that case, the irreversible battery energy density of monolayer g-C2N can reach up to 596 mAh/g(LiC2N). The corresponding sodium ion capacity of this material can only achieve 276 mAh/g(NaC4N2). We also found that the mobility barrier for the lithium atoms in AA stacking structure was 0.25 eV, which indicated the good Li ion conductivity in bulk g-C2N with AA stacking. With the concentration of N reduced to C60N12H12, the cycle performance of the battery will be greatly improved. This indicates that the g-C2N materials with suitable modification could be used as the excellent anode materials with both the high capacity and good Li mobility.