Density Functional Theory Study of B, N, and Si Doped Penta-Graphene as the Potential Gas Sensors for NH3 Detection

Designing a high-performance gas sensor to efficiently detect the hazardous NH3 molecule is beneficial to air monitoring and pollution control. In this work, the first-principles calculations were employed to investigate the adsorption structures, electronic characteristics, and gas sensing properties of the pristine and B-, N-, P-, Al-, and Si-doped penta-graphene (PG) toward the NH3, H2S, and SO2 molecules. The results indicate that the pristine PG is insensitive to those toxic gases due to the weak adsorption strength and long adsorption distance. Nevertheless, the doping of B, N, Al, and Si (B and Al) results in the transition of NH3 (H2S and SO2) adsorption from physisorption to chemisorption, which is primarily ascribed to the large charge transfer and strong orbital hybridizations between gas molecules and doping atoms. In addition, NH3 adsorption leads to the remarkable variation of electrical conductivity for the B-, N-, and Si-doped PG, and the adsorption strength of NH3 on the B-, N-, and Si-doped PG is larger than that of H2S and SO2. Moreover, the chemically adsorbed NH3 molecule on the N-, B-, and Si-doped PG can be effectively desorbed by injecting electrons into the systems. Those results shed light on the potential application of PG-based nanosheets as reusable gas sensors for NH3 detection.


Introduction
In recent years, the excessive emission of poisonous ammonia (NH 3 ) has brought a serious threat to the ecological environment and human health [1]. Long-term exposure to NH 3 can cause bronchitis, lung swelling, and even death [2]. In this regard, the highly efficient detection of trace NH 3 gas is regarded as an effective method to restrain its negative effects. Nowadays, many researchers have focused on the fields of seeking the high-performance sensing materials for NH 3 detection [3][4][5]. Among them, the twodimensional (2D) nanomaterials have great potential application in gas senors due to the good stability, large surface-to-volume ratio, and massive active sites [6][7][8][9]. For instance, C-vacancy defected Ti 2 CO 2 monolayer was predicted by the first-principles method to be a reusable gas sensor for NH 3 detection [10]. In addition, NH 3 adsorption based on borophene [11], C 2 N monolayer [12], phosphorus carbide monolayer [13], and Nidoped InN monolayer [14] has been theoretically studied to evaluate the possibility of those materials as an NH 3 sensor. However, some intrinsic shortcomings, including the insufficient sensitivity, excessively large adsorption strength, and long recovery time, restrict their industrial utilization. Thus, it is necessary to explore the high-performance sensor material for NH 3

detection.
Penta-graphene (PG), a new 2D allotrope of graphene, was recently proposed and predicted to have excellent dynamical and mechanical stability up to 1000K [15]. Moreover, many theoretical investigations showed that PG has a great potential application in gas capture and sensing [16][17][18], hydrogen storage [19], and lithium-ion batteries [20] due to its

Calculation Method and Details
All the gas sensing simulations on pristine penta-graphene (PG), doped-PG, and charged PG were performed using the Dmol 3 module [28] of Materials studio software (Accelrys, California, USA). The generalized gradient approximation in the form of the Perdew-Burke-Ernzerhof [29] and the double numerical plus polarization (DNP) basis set [30] were chosen for all the spin calculations. The Grimme approach [31] and DFT semi-core pseudopotential (DSSP) method [32] were used to describe the van der Waals interactions and core treatment, respectively. The 3 × 3 ×1 supercell of PG (54 atoms) with 20 Å along the z direction was adopted during the simulations, and the Monkhorst−Pack meshes of 4 × 4 ×1 (10 × 10 ×1) were employed to conduct the geometry optimizations (electronic properties calculations), and the Hirshfeld charge was calculated to analyze the charge transfer between gas molecules and substrates. In addition, the global orbital cutoff radius was set as 4.6 Å, and the convergence tolerances were 1.0 × 10 −5 Ha, 0.002 Ha/Å, and 0.005 Å for total energy, maximum force, and maximum displacement, respectively. In doped PG systems, the formation energy (E for ) of B, N, P, Al, or Si atoms could be determined by the following equation [17]: (1) where E PG and E doped-PG are the total energies of pristine and doped PG, and E C and E doped-atom are the chemical potential of C and doped atoms, respectively. The chemical potentials of C (B, P, Al, and, Si) were obtained from the average energy for every C (B, P, Al, and Si) atom in bulk diamond (B, P, Al, and Si), while the chemical potential of the N atom was calculated by the N 2 molecule. The adsorption characteristics of NH 3 , H 2 S, and SO 2 over pristine and doped PG were studied in this work, and the adsorption energy (E ads ) could be obtained by [33,34]  where E gas+substrate and E substrate are the total energies of pristine PG (doped PG) with and without gas adsorptions, and E gas is the energy of a single gas molecule. As known, a more negative E ads means a stronger gas-substrate interaction.

Electronic Properties and Stability of Doped PG
The optimized lattice constant of PG with P-421M symmetry is 3.63 Å, and the calculated band gap is 2.43 eV (Figure 1c), both of which match well with previous reported results [35]. There are two types of carbon atoms, i.e., the sp 2 hybridized carbon atom (C1 site) and sp 3 hybridized carbon atom (C2 site), as shown in Figure 1a. Thus, the C1 and C2 sites substituted by B, N, P, Al, or Si atoms are both taken into account, and the obtained formation energy of different doped PG is listed Table 1. The difference between our results and other studies may be ascribed to the different supercell size and dopant concentration. One can see that N and P substitutions at the C1 site are the exothermic process, while the other doping cases are the endothermic process. Additionally, all the dopant atoms prefer to substitute the C1 atom due to the smaller formation energy, which is consistent with previous calculated results [17,21]. Therefore, the cases of B, N, P, Al, and Si substituting the C1 site are discussed in the following sections.  (2) where E gas+substrate and E substrate are the total energies of pristine PG (doped PG) with and without gas adsorptions, and E gas is the energy of a single gas molecule. As known, a more negative Eads means a stronger gas-substrate interaction.

Electronic Properties and Stability of Doped PG
The optimized lattice constant of PG with P-421M symmetry is 3.63 Å, and the calculated band gap is 2.43 eV (Figure 1c), both of which match well with previous reported results [35]. There are two types of carbon atoms, i.e., the sp 2 hybridized carbon atom (C1 site) and sp 3 hybridized carbon atom (C2 site), as shown in Figure 1a. Thus, the C1 and C2 sites substituted by B, N, P, Al, or Si atoms are both taken into account, and the obtained formation energy of different doped PG is listed Table 1. The difference between our results and other studies may be ascribed to the different supercell size and dopant concentration. One can see that N and P substitutions at the C1 site are the exothermic process, while the other doping cases are the endothermic process. Additionally, all the dopant atoms prefer to substitute the C1 atom due to the smaller formation energy, which is consistent with previous calculated results [17,21]. Therefore, the cases of B, N, P, Al, and Si substituting the C1 site are discussed in the following sections.   [17] +1.00 (+0.74) [17] N-doped PG −0.10 (−0.01) [17] +1.80 (+1.90) [

Adsorption of NH3, H2S and SO2 on Pristine PG
In order to evaluate the detection ability of PG, the adsorption behaviors of NH3, H2S, and SO2 on the pristine PG were investigated in this section. Considering the structure  In order to evaluate the detection ability of PG, the adsorption behaviors of NH 3 , H 2 S, and SO 2 on the pristine PG were investigated in this section. Considering the structure symmetry of PG, six possible adsorption sites, including two top sites (labeled as A and B), three bridge sites (labeled as C, D, and E), and one hollow site (labeled as F) were taken into account, as displayed in Figure 1a. For convenience, one gas molecule adsorption on the PG is defined as "gas@PG", for instance, NH 3 @PG means the NH 3 adsorption on PG. The calculated adsorption energy, adsorption distance, charge transfer, and band gap of three toxic gases adsorbed on PG are presented in Table 2. It is found that the adsorption energies of the NH 3 , H 2 S, and SO 2 molecules are −0.298 eV∼−0.382 eV, and the adsorption distances are in the range of 2.661 Å to 3.169 Å (Figure 2). This indicates that all the toxic gases belong to the representative physisorption, which can be also demonstrated by the small Hirshfeld charge transferred from PG to gases. Moreover, compared with the pristine PG, the band gap of PG after gas molecule adsorptions shows little change. Therefore, the pristine PG exhibits poor detection ability toward the NH 3 , H 2 S, and SO 2 molecules due to weak interactions and negligible resistance variation. symmetry of PG, six possible adsorption sites, including two top sites (labeled as A and B), three bridge sites (labeled as C, D, and E), and one hollow site (labeled as F) were taken into account, as displayed in Figure 1a. For convenience, one gas molecule adsorption on the PG is defined as "gas@PG", for instance, NH3@PG means the NH3 adsorption on PG. The calculated adsorption energy, adsorption distance, charge transfer, and band gap of three toxic gases adsorbed on PG are presented in Table 2. It is found that the adsorption energies of the NH3, H2S, and SO2 molecules are −0.298 eV ∼ −0.382 eV, and the adsorption distances are in the range of 2.661 Å to 3.169 Å ( Figure 2). This indicates that all the toxic gases belong to the representative physisorption, which can be also demonstrated by the small Hirshfeld charge transferred from PG to gases. Moreover, compared with the pristine PG, the band gap of PG after gas molecule adsorptions shows little change. Therefore, the pristine PG exhibits poor detection ability toward the NH3, H2S, and SO2 molecules due to weak interactions and negligible resistance variation.  To further comprehend the micro-interaction mechanism between gas molecules and PG, the density of states (DOSs) and charge density difference (CDD) of various adsorption systems were calculated, as shown in Figure 3 and Figure 4. For the NH3@PG, H2S@PG, and SO2@PG, the gas-substrate interactions are mainly contributed by the weak hybridizations between PG and p orbitals of gas molecules. By contrast, NH3-s, H2S-s, and SO2-s orbitals show the weaker interactions with the penta-graphene due to the small peak value of DOSs (see Figure 3). For the NH3@PG and H2S@PG, the partial electrons of NH3 or H2S are transferred to PG (Figure 4a,b), indicating the NH3 and H2S molecules act as electron donors, which agrees well with the results of the Hirshfeld charge (Table 1). In addition, compared with the case of H2S, NH3 loses more electrons and thus has the lower adsorption energy. For the SO2@PG, a part of the charge of PG is transferred to the SO2 molecule, as shown in Fig. 4c, which suggests the SO2 molecule acts as the electron acceptor. However, the small transfer charge between gas molecules and PG still results in weak interactions, and thus the pristine PG shows poor detection ability toward these toxic gases. To further comprehend the micro-interaction mechanism between gas molecules and PG, the density of states (DOSs) and charge density difference (CDD) of various adsorption systems were calculated, as shown in Figures 3 and 4. For the NH 3 @PG, H 2 S@PG, and SO 2 @PG, the gas-substrate interactions are mainly contributed by the weak hybridizations between PG and p orbitals of gas molecules. By contrast, NH 3 -s, H 2 S-s, and SO 2 -s orbitals show the weaker interactions with the penta-graphene due to the small peak value of DOSs (see Figure 3). For the NH 3 @PG and H 2 S@PG, the partial electrons of NH 3 or H 2 S are transferred to PG (Figure 4a,b), indicating the NH 3 and H 2 S molecules act as electron donors, which agrees well with the results of the Hirshfeld charge (Table 1). In addition, compared with the case of H 2 S, NH 3 loses more electrons and thus has the lower adsorption energy. For the SO 2 @PG, a part of the charge of PG is transferred to the SO 2 molecule, as shown in Fig. 4c, which suggests the SO 2 molecule acts as the electron acceptor. However, the small transfer charge between gas molecules and PG still results in weak interactions, and thus the pristine PG shows poor detection ability toward these toxic gases. Membranes 2021, 11, x FOR PEER REVIEW 5 of 11

Adsorption of NH3, H2S, and SO2 on Doped PG
The adsorption behaviors of NH3, H2S, and SO2 on M-doped penta-graphene (M-PG, M= B, N, Al, and Si) were also investigated, and the obtained calculation results and corresponding lowest energy configurations are displayed in Table 2 and Figure 5. Compared with the NH3 on pristine PG, doping B, N, Al, or Si atoms into PG nanosheets can obviously enhance the NH3 adsorption strength with the Eads of −1.06 eV ∼ −2.46 eV, resulting in the transition of NH3 adsorption from physisorption to chemisorption. In addition, the adsorption distances are significantly shortened (Figure 5a-d), and the charge number transferred from NH3 to the doped PG is remarkably increased in comparison with the case of NH3 adsorption on the pristine PG. By contrast, only the doping of Al and B atoms can strengthen the adsorption of PG toward H2S and SO2, the rest of the dopants show a slight change (see Table 2). In particular, the H2S is dissociation-chemisorbed on the Al-PG surface with one H bonding to the nearest neighbor C atom, as shown in Figure 5f. Therefore, the doping of B, N, P, Al, and Si atoms may be able to improve the detection ability of the NH3, H2S, and SO2 molecules.
The variation of electric conductivity (σ) could be utilized to evaluate the sensitivity of a material before and after molecular adsorption, which is defined as σ ∝ exp(−Eg/2KBT), where Eg, KB, and T are the band gap, Boltzmann constant, and temperature, respectively. As mentioned above, NH3 and H2S adsorptions lead to a little charge in the band gap of the pristine PG, which demonstrates that PG has poor sensitivity to NH3 and H2S. However, the band gap of B-, N-, and Si-PG undergo great variation after NH3 adsorption, as shown in Figure 6 (e, f, h), implying that NH3 adsorption has a huge impact

Adsorption of NH3, H2S, and SO2 on Doped PG
The adsorption behaviors of NH3, H2S, and SO2 on M-doped penta-graphene (M-PG, M= B, N, Al, and Si) were also investigated, and the obtained calculation results and corresponding lowest energy configurations are displayed in Table 2 and Figure 5. Compared with the NH3 on pristine PG, doping B, N, Al, or Si atoms into PG nanosheets can obviously enhance the NH3 adsorption strength with the Eads of −1.06 eV ∼ −2.46 eV, resulting in the transition of NH3 adsorption from physisorption to chemisorption. In addition, the adsorption distances are significantly shortened (Figure 5a-d), and the charge number transferred from NH3 to the doped PG is remarkably increased in comparison with the case of NH3 adsorption on the pristine PG. By contrast, only the doping of Al and B atoms can strengthen the adsorption of PG toward H2S and SO2, the rest of the dopants show a slight change (see Table 2). In particular, the H2S is dissociation-chemisorbed on the Al-PG surface with one H bonding to the nearest neighbor C atom, as shown in Figure 5f. Therefore, the doping of B, N, P, Al, and Si atoms may be able to improve the detection ability of the NH3, H2S, and SO2 molecules.
The variation of electric conductivity (σ) could be utilized to evaluate the sensitivity of a material before and after molecular adsorption, which is defined as σ ∝ exp(−Eg/2KBT), where Eg, KB, and T are the band gap, Boltzmann constant, and temperature, respectively. As mentioned above, NH3 and H2S adsorptions lead to a little charge in the band gap of the pristine PG, which demonstrates that PG has poor sensitivity to NH3 and H2S. However, the band gap of B-, N-, and Si-PG undergo great variation after NH3 adsorption, as shown in Figure 6 (e, f, h), implying that NH3 adsorption has a huge impact

Adsorption of NH 3 , H 2 S, and SO 2 on Doped PG
The adsorption behaviors of NH 3 , H 2 S, and SO 2 on M-doped penta-graphene (M-PG, M=B, N, Al, and Si) were also investigated, and the obtained calculation results and corresponding lowest energy configurations are displayed in Table 2 and Figure 5. Compared with the NH 3 on pristine PG, doping B, N, Al, or Si atoms into PG nanosheets can obviously enhance the NH 3 adsorption strength with the E ads of −1.06 eV∼−2.46 eV, resulting in the transition of NH 3 adsorption from physisorption to chemisorption. In addition, the adsorption distances are significantly shortened (Figure 5a-d), and the charge number transferred from NH 3 to the doped PG is remarkably increased in comparison with the case of NH 3 adsorption on the pristine PG. By contrast, only the doping of Al and B atoms can strengthen the adsorption of PG toward H 2 S and SO 2 , the rest of the dopants show a slight change (see Table 2). In particular, the H 2 S is dissociation-chemisorbed on the Al-PG surface with one H bonding to the nearest neighbor C atom, as shown in Figure 5f. Therefore, the doping of B, N, P, Al, and Si atoms may be able to improve the detection ability of the NH 3 , H 2 S, and SO 2 molecules.  The variation of electric conductivity (σ) could be utilized to evaluate the sensitivity of a material before and after molecular adsorption, which is defined as σ ∝ exp(−E g /2K B T), where E g , K B , and T are the band gap, Boltzmann constant, and temperature, respectively. As mentioned above, NH 3 and H 2 S adsorptions lead to a little charge in the band gap of the pristine PG, which demonstrates that PG has poor sensitivity to NH 3 and H 2 S. However, the band gap of B-, N-, and Si-PG undergo great variation after NH 3 adsorption, as shown in Figure 6e,f,h, implying that NH 3 adsorption has a huge impact on the resistivity of B-, N-, and Si-doped PG, thus these doped PG exhibits high NH 3 sensitivity. Furthermore, the adsorption strength of NH 3 on the B-, N-, and Si-PG surface is much larger than that of H 2 S and SO 2 on the same substrates (see Table 3). Those results indicate that B-, N-, and Si-PG show excellent NH 3 selectivity and sensitivity and could be a promising gas sensor for NH 3 detection. In addition, compared with the Al-PG, the B-PG also has good H 2 S sensitivity due to the obvious change of band gap, as displayed in Figure 6i. Nonetheless, the band gap of Al-PG and B-PG remains unchanged in spite of SO 2 chemisorption (Figure 6k,l), which suggests that Al-and B-PG could not be the SO 2 sensing materials. Table 3. Adsorption energy (E ads ) of various adsorption systems, Hirshfeld charge transferred from gases to doped PG (Q T ), and band gap of the system before adsorption (E g ) and after adsorption (E g '). The positive (negative) Q T means that gases lose (gain) charge.    Table 3). Considering the large adsorption strength and charge transfer, the SO2 adsorbed on the B-and Al-doped PG also belongs to chemisorption.

Structures
In order to further understand the micro-interactions of the gas-substrate, we calculated the DOSs of the N (S) atom of NH3 (H2S/SO2) and its nearest neighbor dopant atoms for various adsorption systems, as displayed in Figure 8. For the NH3@N-PG system, the NH3 is adsorbed on top of its nearest neighbor C atom, and the obvious orbital hybridizations between C-2p and N-2p of NH3 can be observed in the range of −12.5 eV to −10 eV, while the relatively weak interactions between C-sp and N-2s are found at −21.05 eV as   Figure 7e. However, for the SO 2 @B-PG and SO 2 @Al-PG systems, the charges are mainly accumulated around the SO 2 molecule, while the charges are depleted near the B and Al dopants, thus the SO 2 acts as an electron acceptor and obtains the charge from B-PG and Al-PG by 0.105 eand 0.203 e − (see Table 3). Considering the large adsorption strength and charge transfer, the SO 2 adsorbed on the Band Al-doped PG also belongs to chemisorption.
In order to further understand the micro-interactions of the gas-substrate, we calculated the DOSs of the N (S) atom of NH 3 (H 2 S/SO 2 ) and its nearest neighbor dopant atoms for various adsorption systems, as displayed in Figure 8. For the NH 3 @N-PG system, the NH 3 is adsorbed on top of its nearest neighbor C atom, and the obvious orbital hybridizations between C-2p and N-2p of NH 3 can be observed in the range of −12.5 eV to −10 eV, while the relatively weak interactions between C-sp and N-2s are found at −21.05 eV as shown in Figure 8a. For the NH 3 @B-PG, NH 3 @Al-PG, and NH 3 @Si-PG systems (Figure 8b-d), there are significant hybridizations between N-2p of NH 3 and B-, Al-, and Si-sp orbitals ranging from −10.0 eV to −5.0 eV. Moreover, there are also weak hybridizations of N-2s and B-, Al-, and Si-2s around −20.0 eV. In summary, the strong interactions between the N atom of NH 3 and its nearest neighbor atoms result in NH 3 chemisorption. As for the H 2 S@B-PG and H 2 S@Al-PG, the strong adsorption strength of the H 2 S molecule is primarily contributed by the orbital interactions between S-2p of H 2 S and B-and Al-sp in the range of −10.0 eV to 0.0 eV (Figure 8e-f). By contrast, the strong orbital hybridizations between S-2p of SO 2 and B-and Al-sp occur in the larger range of −10.0 eV to 2.5 eV for the SO 2 @B-PG and SO 2 @Al-PG systems, which leads to the SO 2 molecule being chemically adsorbed over the B-and Al-doped PG.

Recovery of Doped PG after Sensing Toxic Gases
The short recovery time at room temperature is also of great importance for a good sensing material. Consequently, the feasibility of the toxic gases (NH 3 , H 2 S, and SO 2 ) desorbed from the N-, B-, Si-, and Al-doped PG was explored in this section. As displayed in Table 3, the adsorption energies of NH 3 , H 2 S, and SO 2 on the doped PG with chemisorption are all lower than −0.97 eV. As known, the gas molecule can be released from a solid material when the corresponding E ads is higher than −0.50 eV [36]. Accordingly, those toxic gases cannot be desorbed immediately after adsorption, and the recovery time becomes very long at room temperature. Recently, the introduction of a negative charge could effectively modulate the adsorption strength of a gas molecule on the substrate [37][38][39]. In this regard, the adsorption energies of NH 3 , H 2 S, and SO 2 on the doped PG with a different negative charge were calculated, and the results are shown in Figure 9. With the increase in injected negative charge, the adsorption strength of NH 3 on B-, Al-, and Si-doped PG is gradually decreased, while that of H 2 S and SO 2 on B-and Al-doped PG is gradually increased. This indicates that the H 2 S or SO 2 removal from the B-PG and Al-PG surface becomes more difficult after the introduction of electrons due to the stronger adsorption strength. Nevertheless, the adsorption energies of NH 3 @B-PG and NH 3 @Si-PG are increased from −2.46 eV to −0.38 eV and from −1.11 eV to −0.29 eV after adding 2.0 e − into the systems. For the NH 3 @N-PG system, the adsorption energy is also remarkably changed from −1.06 eV to −0.24 eV (−0.35 eV) after injecting 0.5 e − (1.0 e − ) into the N-PG substrate. Thus, we can conclude that the chemically adsorbed NH 3 molecule on the N-, B-, and Si-doped PG could be easily desorbed by controlling the number of injected electrons into the systems. Considering the N-, B-, and Si-doped PG also exhibit good selectivity and sensitivity toward the NH 3 molecule as discussed above, we can therefore say that the N-, B-, and Si-doped PG could be a promising sensing material for NH 3 detection. The short recovery time at room temperature is also of great importance for a good sensing material. Consequently, the feasibility of the toxic gases (NH3, H2S, and SO2) desorbed from the N-, B-, Si-, and Al-doped PG was explored in this section. As displayed in Table 3, the adsorption energies of NH3, H2S, and SO2 on the doped PG with chemisorption are all lower than −0.97 eV. As known, the gas molecule can be released from a solid material when the corresponding Eads is higher than −0.50 eV [36]. Accordingly, those toxic gases cannot be desorbed immediately after adsorption, and the recovery time becomes very long at room temperature. Recently, the introduction of a negative charge could effectively modulate the adsorption strength of a gas molecule on the substrate [37][38][39]. In this regard, the adsorption energies of NH3, H2S, and SO2 on the doped PG with a different negative charge were calculated, and the results are shown in Figure 9. With the increase in injected negative charge, the adsorption strength of NH3 on B-, Al-, and Sidoped PG is gradually decreased, while that of H2S and SO2 on B-and Al-doped PG is gradually increased. This indicates that the H2S or SO2 removal from the B-PG and Al-PG surface becomes more difficult after the introduction of electrons due to the stronger adsorption strength. Nevertheless, the adsorption energies of NH3@B-PG and NH3@Si-PG are increased from −2.46 eV to −0.38 eV and from −1.11 eV to −0.29 eV after adding 2.0 e − into the systems. For the NH3@N-PG system, the adsorption energy is also remarkably changed from −1.06 eV to −0.24 eV (−0.35 eV) after injecting 0.5 e − (1.0 e − ) into the N-PG substrate. Thus, we can conclude that the chemically adsorbed NH3 molecule on the N-, B-, and Si-doped PG could be easily desorbed by controlling the number of injected electrons into the systems. Considering the N-, B-, and Si-doped PG also exhibit good selectivity and sensitivity toward the NH3 molecule as discussed above, we can therefore say that the N-, B-, and Si-doped PG could be a promising sensing material for NH3 detection.

Conclusions
In summary, the adsorption behaviors and gas sensing properties of the pristine and N-, B-, P-, Al-, and Si-doped PG toward three toxics gases (NH3, H2S, and SO2) were studied by DFT calculations in this work. In addition, the adsorption characteristics, electronic properties, and the sensing mechanism were also discussed. The results of formation energy show that the substitution of the C1 atom by N, B, P, Al, or Si is energetically favorable. The pristine PG exhibits poor sensing ability of the NH3, H2S, and SO2 molecules due to weak adsorption strength. In contrast, doping B, N, Al, or Si atoms into PG nanosheets can obviously improve the NH3 adsorption strength, while only the doping of the B and

Conclusions
In summary, the adsorption behaviors and gas sensing properties of the pristine and N-, B-, P-, Al-, and Si-doped PG toward three toxics gases (NH 3 , H 2 S, and SO 2 ) were studied by DFT calculations in this work. In addition, the adsorption characteristics, electronic properties, and the sensing mechanism were also discussed. The results of formation energy show that the substitution of the C1 atom by N, B, P, Al, or Si is energetically favorable. The pristine PG exhibits poor sensing ability of the NH 3 , H 2 S, and SO 2 molecules due to weak adsorption strength. In contrast, doping B, N, Al, or Si atoms into PG nanosheets can obviously improve the NH 3 adsorption strength, while only the doping of the B and Al atoms enhance the interactions between the H 2 S and SO 2 molecules and PG, which was further verified by the analysis of DOSs, CDD, and the charge transfer. Additionally, the adsorption strength of NH 3 on B-, N-, and Si-doped PG is much larger than that of H 2 S and SO 2 on the same substrates, demonstrating that the B-, N-, and Si-doped PG have good NH 3 selectivity from the mixtures of H 2 S and SO 2 . In addition, doped PG exhibits high NH 3 sensitivity due to its great variation of resistivity after NH 3 adsorption. More interestingly, the chemically adsorbed NH 3 molecule on the N-, B-, and Si-doped PG could be effectively desorbed by controlling the number of injected electrons into the systems. Therefore, the N-, B-, and Si-doped PG can be the potential materials to develop NH 3 sensors with high sensitivity and selectivity.
Author Contributions: G.C.: Conceptualization, validation, formal analysis, and visualization; H.Z.: methodology, writing-review and editing, and project administration; H.X.: software, investigation, supervision, writing-review and editing, and project administration; L.G.: data curation, writingoriginal draft preparation, and funding acquisition. All authors have read and agreed to the published version of the manuscript.