A theoretical study of gas adsorption on silicene nanoribbons and its application in a highly sensitive molecule sensor

Abstract

Inspired by recent successes in the development of two-dimensional based gas sensors capable of single gas molecule detection, we investigate the adsorption of gas molecules (N₂, NO, NO₂, NH₃, CO, CO₂, CH₄, SO₂, and H₂S) on silicene nanoribbons (SiNRs) using density functional theory (DFT) and nonequilibrium Green's function (NEGF) methods. The most stable adsorption configurations, adsorption sites, adsorption energies, charge transfer, quantum conductance modulation, and electronic properties of gas molecules on SiNRs are studied. Our results indicate that NO, NO₂, and SO₂ are chemisorbed on SiNRs via strong covalent bonds, suggesting its potential application for disposable gas sensors. In addition, CO, NH₃, and H₂S are chemisorbed on SiNRs with moderate adsorption energy, alluding to its suitability as a highly sensitive gas sensor. The quantum conductance is detectably modulated by chemisorption of gas molecules which can be attributed to the charge transfer from the gas molecule to the SiNR. Other studied gases are physisorbed on SiNRs via van der Waals interactions. It is also found that the adsorption energies are enhanced by doping SiNRs with either B or N atoms. Our results suggest that SiNRs show promise in gas molecule sensing applications.

---

I Introduction

Although the global development of chemical activities in the past century is a consequence of human demands, it also affects human health and life quality from the associated environmental release of poisonous substances in the forms of solids, liquids and vapours. Therefore, gas detection is vital for managing chemical processes to prevent health hazards such as air pollution, device contamination, and for medical diagnosis. Today, in developing novel gas sensing materials with high sensitivity, even down to the single molecule level, high selectivity, high stability, quick response and recovery, and low power consumption are of considerable importance. Graphene, the first discovered two-dimensional (2D) material,^1,2 renders outstanding properties such as high surface-volume ratio, low electronic temperature noise, remarkably high carrier mobility, high chemical and thermal stability, and fast response time, which make it promising in the development of ultrasensitive sensors with high packing density, higher sensitivity, better selectivity, faster recoverability, and lower power consumption.^3–5 The potential application of graphene in gas sensors has been widely studied both experimentally^6,7 and theoretically.^8–10 However, the physisorption of common gas molecules such as CO₂, CO, CH₄, H₂O, H₂, N₂, NO₂, and NO on pristine graphene⁸ restrict its potential for single molecule detection.^11,12 It is found that the sensing ability of graphene can be enhanced by introducing dopants or defects.^9,10,13

Inspired by the alluring properties of graphene, silicene, the silicon analogue of graphene, having a buckled honeycomb structure,^14–16 has garnered considerable interest because of its remarkable properties including ferromagnetism,¹⁷ half-metallicity,¹⁸ quantum Hall effect,¹⁹ giant magnetoresistance,²⁰ and superconductivity.²¹ Takeda and Shiraishi¹⁴ reported silicene for the first time in 1994 and Guzmán-Verri and Voon¹⁵ coined its name in 2007. Although free-standing silicene is not stable, it was experimentally fabricated on Ag,^21–24 Ir,²⁵ ZrB₂,²⁶ and ZrC.²⁷ Using first-principles calculations, various substrates such as h-BN,²⁸ SiC,²⁹ GaS,³⁰ graphene³¹ and ZnS³² which have weak van der Waals (vdW) interactions with silicene have been also studied to improve its stability. The extraordinary properties of silicene along with its compatibility with silicon based nanoelectronics may give the edge to silicene rather than graphene. Numerous potential applications of silicene in spintronics,³³ field-effect transistors (FETs),^34,35 and sensing devices^36,37 have been proposed. Unlike a flat graphene sheet, the silicene honeycomb structure is buckled^15,16 due to the tendency of silicon atoms to adopt sp³ and sp² hybridization rather than only sp² hybridization.²² This buckled structure makes it possible for the band gap of silicene to be tuned more intensively with an external electric field³⁸ and with binding adsorbates³⁹ as compared to graphene.⁴⁰ Furthermore, silicene shows a considerably higher chemical reactivity for atoms^41–46 and molecules^47–53 adsorption than graphene due its buckled formation with a great deal of potential applications for silicene based nanoelectronic devices,⁵⁴ hydrogen storage,⁴⁵ thin film solar cell absorbers,⁴⁶ hydrogen separation membranes,⁴⁷ and molecule sensors.^49–51,55

Similar to graphene, silicene is a zero band gap semimetal.¹⁵ It is expected that the effects of gas molecule adsorption on the electronic properties of a material without a band gap is much less than that of a semiconductor with an intrinsic band gap. A number of methods have been proposed to induce a band gap in a silicene sheet including doping,⁵⁶ substrate effects,⁵⁷ chemically functionalization,^58,59 electric field,³³ nanomesh and nanoholes.^60–62 One of the possible methods to introduce a band gap in silicene is achieved by cutting the sheet into silicene nanoribbons (SiNRs).^16,34,63 SiNRs have been grown on Ag (100), Ag (110), Au (110) substrates.^64–67 Aufray et al. have synthesized an array of highly uniform parallel SiNRs with width of ∼1.7 nm on Ag (110) substrate.⁶⁴ The intrinsic band gap and small dimension along with free reactive edges make nanoribbons more attractive than sheets for gas nanosensor applications. Several experimental and theoretical studies of graphene nanoribbons (GNRs) sensing properties have been already reported.^68–71 The CO, NO, NO₂, and O₂ molecules are chemisorbed on armchair GNRs (AGNRs), while the adsorption of NH₃ and CO₂ on AGNRs is between weak chemisorption and strong physisorption.⁷⁰ It has been found that CO, CO₂, NO, NO₂, and O₂ molecules draw electrons from AGNRs, while NH₃ donates its electrons into AGNRs. Little attention has been focused on molecule adsorption on SiNRs. A recent theoretical work by Osborn and Farajian has proven that ASiNRs can be used for detection of CO molecule due to its weak chemisorption on ASiNRs.⁷² They also found that H₂O and O₂ are strongly chemisorbed on ASiNRs, while CO₂ and N₂ barely affect ASiNR's conductance.

It remains an open question as to what is the adsorption behavior of nitrogen-, sulfur-, and other carbon-based gas molecules, including NO, NO₂, NH₃, SO₂, H₂S, and CH₄ which are all of great practical interest for environmental, medical, and industrial applications, on SiNRs. In this paper, we employed first-principles methods based on density functional theory (DFT) to investigate the adsorption behavior of CO, CO₂, CH₄, N₂, NO, NO₂, NH₃, SO₂, and H₂S molecules on SiNRs. Our results promulgate the promising future of SiNR in the development of ultrahigh sensitive sensor platforms.

II Computational methods

All the calculations are carried out based on first-principles DFT combined with nonequilibrium Green's function (NEGF) implemented in Atomistix ToolKit (ATK) package.^73–75 The Generalized Gradient Approximation of Perdew–Burke–Ernzerhof (GGA-PBE) with a double-ζ polarized basis set is adopted to solve Kohn–Sham equations and to expand electronic density. The density mesh cut off is set to be 150 Rydberg. The Grimme vdW correction (DFT-D2)⁷⁶ is also employed to describe long-range vdW interactions.⁷⁷ In order to take into account the vdW interactions, an additional term (E_vdW) is added to the DFT total energy (E_DFT):

E_DFT-D2 = E_DFT + E_vdW

Besides vdW interactions, since our systems have two subsystems: the SiNR (A) and the gas molecule (B), so-called basis set superposition errors (BSSE) are expected due to the incompleteness of the Linear Combination of Atomic Orbitals (LCAO) basis set. In an isolated A system, only the basis orbitals in the A system describe it. While, A and B are coupled, the basis orbitals in the system B will also be used to describe system A, resulting in a larger available basis set for system A. Consequently, there will be an artificial interaction which decreases the total energy. To eradicate BSSE, a counterpoise (cp) correction should be added to the total energy.⁷⁸ Therefore, the total energy of the system considering the long-range vdW interactions and artificial attractions between two subsystems is:

E_total = E_DFT-D2 + E^cp

To avoid the mirroring interactions, a vacuum space of 25 Å is considered in x and y directions in which the structures are not periodic. The electronic temperature is kept constant at 300 K. All the structures are completely relaxed, prior to the calculations, up until the force and stress are less than 0.01 eV Å⁻¹ and 0.005 eV Å⁻³, respectively. 1 × 1 × 21 k-points in the Brillouin zone are sampled for geometry optimization and 1 × 1 × 121 k-points for total energy, electronic properties, charge transfer, and electron transport calculations. To investigate the charge transfer and transport properties, the gas sensing system is divided into three regions: two electrode regions (left and right) and a scattering region (the central region), as illustrated in Fig. 1. The transmission coefficient T(ε) can be obtained from the retarded Green's function G(ε) using

T(ε) = G(ε)Γ^L(ε)G^†(ε)Γ^R(ε)

Fig. 1 Schematic structural model of SiNR-based gas sensor with two electrodes (black boxes). The cyan and green balls represent Si and H atoms, respectively. Possible adsorption sites of gas molecules on the SiNR are: I (hill), II (valley), III (bridge), and IV (hollow).

Here Γ^L(R) is the broadening function of the left (right) electrode which defined as

where Σ^L(R), the left (right) electrode self-energy. The conductance is determined by the transmission coefficient at the Fermi level,

C(ε) = G₀T(ε)

where G₀ = 2e²/h is the quantum conductance, in which e is the electron charge and h is Planck's constant. Further details on computational methods can be found in ESI material.†

III Results and discussions

In this study, all the nanostructures are made of long and non-periodic 7-ASiNRs (L = 35 Å). Their edges are passivated by H while their surfaces are kept pristine because the surfaces are less reactive than the edges⁷⁹ and silicene lacks complete sp³ hybridization.⁸⁰ It has been reported that 7-ASiNR is a paramagnetic semiconductor with a band gap of ∼0.56 eV.^62,64 The semiconducting ASiNR is chosen because it is expected that the adsorption of gas molecules has a much smaller effect on the electronic properties of zigzag SiNRs (ZSiNRs) which are metallic. To evaluate chemical sensing of SiNR-based nanosensors, single gas molecules (N₂, NO, NO₂, NH₃, CO, CO₂, CH₄, SO₂, and H₂S) are initially placed about 3 Å from the ASiNRs surface. The adsorption behaviors of molecules on SiNR are investigated after full relaxation. The structural stability of the molecule's adsorption on SiNR is addressed using adsorption energy (E_ad) which is

E_ad = E_{SiNR+molecule} − E_SiNR − E_molecule

where E_{SiNR+molecule}, E_SiNR, E_molecule denote the total energies of the SiNR–molecule system, pristine SiNR, and the isolated gas molecule, respectively. Based on the definition, the more negative E_ad is, the stronger adsorption of gas molecules on SiNR would be. By considering different adsorption configurations of gas molecules on SiNR and calculating each configuration's adsorption energy, it is discovered that different gas molecules prefer to be adsorbed with different geometries. To commence the relaxation, the molecules can be placed at four different positions including valley, hill, bridge sites, and hollow, as shown in Fig. 1. At these particular positions, different molecular orientations are considered. The structural stabilities of diatomic molecules (N₂, NO, and CO) are examined when their molecular axis is aligned perpendicular and parallel with respect to the SiNR's surface. Furthermore, for NO and CO molecules, the O atom can point up and down. For the triatomic molecule CO₂ with 180° bond angles, two orientations, parallel and perpendicular to the surface, are tested. For other triatomic (NO₂, SO₂, and H₂S), tetratomic (NH₃), and pentatonic (CH₄) molecules, two orientations are considered. In the first orientation, the N atom (in NO₂ and NH₃), S atom (in SO₂ and H₂S), and C atom (in CH₄) point to the SiNR's surface, while, in the second orientation, the H atom (in NH₃, H₂S, and CH₄), and O atom (in SO₂ and NO₂) point away from the SiNR's surface. In order to find the most favorable location for gas molecules to be adsorbed, the adsorption energies of gas molecules on the edge site and the interior site are also compared. As an example, the calculated adsorption energy of a CO gas molecule on an edge atom site is 0.35 eV less than that of an interior (center) atom site, indicating that the gas molecules prefer to be adsorbed on the edge sites. This is in agreement with those previously reported for CO adsorption on SiNR.⁷² Fig. 2 presents the most stable adsorption configurations of gas molecules on pristine ASiNRs.

Fig. 2 The most stable adsorption configurations (top and side view) for N₂, NO, NO₂, NH₃, CO, CO₂, CH₄, SO₂, and H₂S on pristine ASiNR. The cyan, green, blue, red, grey, and yellow balls represent Si, H, N, O, C, and S atoms, respectively. The bond lengths (in unit of Å) of molecule and binding distances between the molecule and the SiNR are also given.

As previously mentioned in Section II, the long range vdW interactions which come from non-local dispersion and artificial interactions are important factors in studying adsorption behaviour of gas molecules on SiNR. In order to examine the impact of vdW and counterpoise corrections on our calculations, the adsorption energy of gas molecules on SiNR and binding distances between gas molecules and SiNR are obtained by DFT-D2+cp (considering vdW dispersion and counterpoise correction) and PBE-GGA approximation and shown in Fig. 3. The results show that the absolute values of adsorption energy calculated by DFT-D2+cp are significantly more than that of GGA-PBE approximation. In addition, the optimum binding distances between molecules and SiNR obtained by DFT-D2+cp are a little less than calculated distances via GGA-PBE approximation.

Fig. 3 (a) Adsorption energies of gas molecules on SiNR obtained by DFT-D2+cp (considering vdW dispersion and counterpoise correction) and PBE-GGA approximation (b) binding distances between the ribbon and molecules obtained by DFT-D2+cp (considering vdW dispersion and counterpoise correction) and PBE-GGA approximation.

The calculated adsorption energies and binding distances between the gases and ASiNR which are obtained by DFT-D2+cp method are also specified in Table 1. Our results show that CO is chemically adsorbed on SiNR with an adsorption energy of −0.92 eV. In the most stable structure, the CO molecule prefers to be adsorbed in a vertical orientation when the C atom is connected to a Si-edge atom with a covalent bond at a hill position with a binding distance of 1.89 Å. The bond length of the adsorbed CO is 1.16 Å which is little longer than that of an isolated CO molecule (1.14 Å). For CO₂ and CH₄, the results indicate that they are physisorbed on SiNR with a small adsorption energy of −0.31 and −0.18 eV, respectively. Both molecules prefer to be placed at hollow site after relaxation. The N₂ molecule is energetically stable when it is horizontally situated at the center of a hexagonal silicon cell near the edge. The low adsorption energy of −0.31 eV shows that N₂ is physically adsorbed on SiNRs. However, SiNR is highly reactive to other N-based gas molecules such as NH₃, NO₂, and NO, whose adsorption energies are −1.12, −2.53, −2.68 eV, respectively. NH₃ is chemically adsorbed on top of Si-edge atom at a hill position with the formation of a N–Si covalent bond (2 Å). The adsorption energy of NO₂ and NO is even less than −2 eV showing a very strong chemisorption. For the NO₂ molecule, the N–O bond interacts with the Si–Si bond of the SiNR at the edge, where the N–Si and O–Si covalent bonds' length are 1.91 and 1.77 Å, respectively. The chemisorption of the NO molecule on SiNR is even stronger than that of NO₂ molecule. The N atom is bonded to Si–Si atoms on the edge of the SiNR to form a N–Si–Si triangle with N–Si and Si–Si bond lengths of 1.81 and 2.32 Å, respectively. The O atom is also connected to another Si atom on top of the Si–hexagon with a bond length of 1.73 Å. The adsorption energy of the H₂S molecule on top of the SiNR is −0.64 eV, showing that the adsorption of the molecule is a weak chemisorption. Strong chemisorption is also observed for the SO₂ molecule, whose adsorption energy is −2.63 eV. In this case, the O atoms of SO₂ molecule are bonded to Si–Si bond at the SiNR edge, where the covalent bond lengths are 1.70 Å.

Table 1 The calculated adsorption energy (E_ad), binding distance which is the shortest atom to atom distance between molecule and device (D), and the charge transfer from ribbon to molecule (ρ). P = pristine, DB = dangling bond defect, N = nitrogen doped, and B = boron doped

Device	Gas	Ead (eV)	D (Å)	ρ (e)	Device	Gas	Ead (eV)	D (Å)	ρ (e)	Device	Gas	Ead (eV)	D (Å)	ρ (e)
a From ref. 39, 48–51 and 81.
P-SiNR	N2	−0.30	3.47	−0.110	P-SiNR	2N2	−0.57	3.36	−0.097	DB-SiNR	N2	−0.32	3.28	−0.090
	NO	−2.68	1.73	−0.892		2NO	−5.32	1.72	−0.888		NO	−3.23	1.74	−0.908
	NO2	−2.53	1.77	−0.851		2NO2	−5.08	1.78	−0.849		NO2	−2.94	1.92	−0.848
	NH3	−1.12	2.00	−0.535		2NH3	−2.18	2.00	−0.493		NH3	−1.24	1.98	−0.520
	CO	−0.92	1.89	−0.406		2CO	−1.78	1.89	−0.386		CO	−1.52	1.83	−0.395
	CO2	−0.31	3.34	−0.101		2CO2	−0.58	3.38	−0.093		CO2	−0.35	3.27	−0.100
	CH4	−0.18	3.40	+0.025		2CH4	−0.32	3.41	+0.021		CH4	−0.22	3.33	+0.020
	SO2	−2.64	1.74	+0.626		2SO2	−5.11	1.73	+0.620		SO2	−2.72	1.80	−0.615
	H2S	−0.64	2.45	−0.342		2H2S	−1.25	2.45	−0.375		H2S	−0.79	2.45	−0.323
N-SiNR	N2	−0.35	2.59	−0.151	B-SiNR	N2	−0.55	1.60	−0.352	P-Silicene^a	N2	−0.08	—	+0.005
	NO	−3.74	1.72	+0.138		NO	−3.82	1.38	−0.233		NO	−0.57	1.99	+0.46
	NO2	−3.60	1.75	−0.041		NO2	−3.53	1.55	−0.162		NO2	−1.53	1.75	+0.82
	NH3	−1.35	1.98	−0.559		NH3	−1.14	1.63	−0.663		NH3	−0.60	2.04	+0.10
	CO	−1.89	2.00	−0.274		CO	−1.59	1.50	−0.452		CO	−0.10	3.24	+0.03
	CO2	−0.35	2.70	−0.140		CO2	−0.32	3.28	−0.110		CO2	−0.04	—	∼0.00
	CH4	−0.27	3.57	+0.016		CH4	−0.25	3.39	+0.014		CH4	−0.08	—	+0.004
	SO2	−3.48	1.75	−0.210		SO2	−2.95	1.57	−0.204		SO2	−1.08	2.17	+0.17
	H2S	−1.86	2.40	−0.305		H2S	−0.90	1.98	−0.407		H2S	−0.212	2.28	—

---

To sum up, the interaction of N₂, CO₂, and CH₄ gas molecules with a SiNR is mostly vdW type adsorbing via physisorption. A SiNR cannot be an appropriate sensor for detection of NO, NO₂, and SO₂ gas molecules because of their strong chemisorption on silicene. However, SiNRs can be considered as a disposable molecule sensor for detection of NO, NO₂, and SO₂. Besides, the strong covalent bonding of these molecules to SiNR makes it possible to tune the electronic properties of SiNR for nanoelectronics applications, as will be discussed later. Finally, the moderate adsorption energies of NH₃, CO, and H₂S gas molecules make SiNRs a promising material for sensitive gas sensors or gas filters because they can be easily desorbed from the SiNR by heating. It should also be noted that environmental gas molecules (N₂, CO₂, O₂, and H₂O) can change the NH₃ and CO sensing capability of SiNRs. Unlike physisorption of N₂ and CO₂ gas molecules on SiNR, the water and oxygen molecules are strongly chemisorbed on SiNR.⁷² Therefore, it is critical that water and oxygen molecules be removed from the environment to preserve the proper sensing capability of SiNRs.

To investigate the effects of gas molecules on the conductance of SiNR, the quantum conductances of the pristine SiNR before and after adsorption are calculated, as shown in Fig. 4. Upon adsorption of N-based gas molecules, the band gap of pristine SiNR (0.56 eV) is decreased slightly for N₂ (0.52 eV) and NH₃ (0.48 eV) and is somewhat increased for NO₂ (0.68 eV) and NO (0.68 eV). The conductances of SiNR before and after adsorption of N₂ molecule are similar due to the physisorption of N₂ gas molecule on SiNR, showing insensitivity of SiNR sensor to N₂ gas molecule. However, the overall conductances of SiNRs are detectably dropped after NH₃, NO₂, and NO, confirming their chemisorption on SiNRs. The conductance reduction for a NO molecule is more vivid than other N-based gas molecules which are consistent with the calculated adsorption energies. For C-based gas molecules, the band gap of pristine SiNR is preserved for CO₂ (0.56 eV) and slightly decreased for CH₄ (0.52 eV) and CO (0.48 eV). The overall conductance of pristine SiNR is visibly reduced after CO adsorption, while barely changed after CO₂ adsorption, and almost unchanged after CH₄ adsorption which confirms the calculated adsorption energies for C-based gas molecules where the CO adsorption energy is more negative than others. Upon adsorption of S-based gas molecules, the band gap of pristine SiNR is unaltered for SO₂ (0.56 eV) and decreased for H₂S (0.48 eV) gas molecule. A reduction in the conductance of the SiNR from SO₂ and H₂S adsorption is also observed, confirming the sensing capability of SiNR for SO₂ and H₂S gas molecules.

Fig. 4 Quantum conductance of pristine SiNR before and after (a) N₂, NO, NO₂, NH₃, (b) CO, CO₂, CH₄, (c) SO₂ and H₂S gas molecules adsorption.

The electron charge transfer between gas molecules and SiNR is also investigated using Mulliken population analysis, as shown in Table 1. The positive value of charge indicates a charge transfer from SiNR surface to molecule. N₂, NO, NO₂, NH₃, CO, CO₂, and H₂S are electron-donating gas molecules, while CH₄ and SO₂ have an electron-withdrawing capability. A large amount of charge transfer is observed for NO (−0.892 e), NO₂ (−0.851 e), NH₃ (−0.535 e) and CO (−0.406 e) due to their strong electron-donating characteristics. These large charge transfer molecules to SiNR are also correlated to the strong binding energies of the gas molecules chemisorption on SiNR. The calculated adsorption energies of all studied gas molecules on SiNR are distinctly larger than those of GNR,⁷⁰ showing higher sensitivity of SiNR toward the gas molecules. Furthermore, comparing these findings with gas molecules adsorption on pristine silicene,^{39,48–50,81} see Table 1, the value of adsorption energy and charge transfer are much larger in favor of SiNR, pointing to the great potential of SiNRs for application as a highly sensitive molecule sensor.

The origin of the chemisorption of NO, NO₂, NH₃, CO, SO₂ and H₂S and physisorption of N₂, CO₂, and CH₄ gas molecules on SiNR can be revealed by studying their density of states (DOS). Fig. 5(a) presents the total DOS of pristine SiNR and the gas molecules. The frontier molecular orbitals, the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO), of NO, NO₂, and SO₂ are closer to the Fermi level of SiNR than that of NH₃, CO, and H₂S, representing that SiNRs show higher sensitivity for the former molecules. Furthermore, these molecules have higher reactivity to SiNRs as compared to N₂, CO₂, and CH₄ which are physisorbed on the SiNR. These results show reasonable agreement with the calculated adsorption energies and are consistent with earlier research on gas molecule adsorption on silicene sheets.^48–51 As previously mentioned, silicene is more reactive than graphene because of its tendency to adopt sp² + sp³ hybridization over sp² hybridization.²² The difference between the reactivity of silicene and graphene can be understood from Fig. 5(b), where PDOS of silicon/carbon atom located at the edge of SiNR/GNR is calculated. While both silicon (Ne 3s²3p²) and carbon (1s²2s²2p²) contribute four valence electrons in the ribbon, the contributions of s orbitals are negligible near the Fermi level. Therefore, only p orbitals could be considered. Comparing the p orbital contributions of Si and C atoms near the Fermi level, it is clear that the corresponding electronic edge states of 7-ASiNR are closer than that of 7-AGNR to the Fermi level, causing stronger reactivity of SiNR for molecule adsorption.

Fig. 5 (a) DOS of N₂, NO, NO₂, NH₃, (b) CO, CO₂, CH₄, SO₂ and H₂S gas molecules and SiNR. The positive and negative values represent spin-up and -down states, respectively. (b) PDOS of SiNR (Si-3s and Si-3p) and GNR (C-2s and C-2p). The inset shows the schematic of a SiNR/GNR, in which a Si/C edge atom is distinguished by a black circle. The cyan and red balls represent Si/C and H atoms, respectively. The Fermi level of SiNR and GNR are set to zero.

To gain more insight into the chemisorption and physisorption of gas molecules, their electronic total charge densities are calculated, as depicted in Fig. 6. It can be seen that there is no electron orbital overlap between N₂, CO₂, and CH₄ gas molecules and SiNR, confirming this fact that only weak physisorption takes place in these cases. In contrast, a strong orbital overlap is observed between NO, NO₂, and SO₂ gas molecules and SiNR, resulting in more orbital mixing and a larger charge transfer. It is clear that this strong orbital overlap results in significant changes to the electronic properties of SiNR. Finally, a moderate orbital overlap between NH₃, CO, and H₂S gas molecules and SiNR suggest the potential application of SiNR as a reusable gas sensor because the electronic properties of SiNR which are moderately changed can be retrieved by desorbing gas molecules from SiNR. The results correspond with calculated adsorption energies.

Fig. 6 Electronic total charge densities for the adsorption of N₂, NO, NO₂, NH₃, CO, CO₂, CH₄, SO₂, and H₂S gas molecules on pristine ASiNR.

In order to find the impact of nanoribbon width on the adsorption energy, we also considered the CO molecule adsorption on 5-ASiNR, 6-ASiNR, and 8-ASiNR. The values of adsorption energy of CO molecule on 5-ASiNR, 6-ASiNR, 7-ASiNR, and 8-ASiNR are −0.98, −0.95, −0.93, and −0.91 eV, respectively. This overall small decrease in absolute value of adsorption energy with decreasing surface coverage of the SiNR is in agreement with those previously reported for CO molecule adsorbed on AGNR.⁸²

Since upon adsorption of individual gas molecules a conduction reduction is observed in the properties of SiNRs, significant alterations in the electronic properties of SiNRs are expected at higher concentrations of gas molecules. To evaluate this premise, a higher surface coverage with two gas molecules is considered and the impacts of gas molecules adsorption on the conduction of SiNR are investigated. Fig. 7 presents the most stable configurations of two adsorbed gas molecules on opposite edges of the SiNR. Based on the calculations, an individual gas molecule can be adsorbed anywhere at the SiNR's edges and its position will not change the quantum conductance. The second gas molecule can also be adsorbed on the same side or the opposite side of the first gas molecule. The results show that the conductance changes are almost the same for two cases, in agreement with the findings of Osborn et al.⁷² By calculating adsorption energies, it is found that adding an additional gas molecule to the system doubles the value of adsorption energies of gas molecules on the SiNR, see Table 1. It is clear that the NO, NO₂, NH₃, CO, SO₂, and H₂S sensing capability of SiNR, which is exposed by two gas molecules, is increased due to the significant reduction in the conductance of pristine SiNR and the strong chemisorption of gas molecules to the SiNR (their adsorption energies are less than −1 eV), as shown in Fig. 8. By performing Mulliken population analysis of the SiNR with one adsorbed molecule and comparing to the SiNR with two adsorbed molecules, it is found that there are no appreciable differences between the amounts of charge transfer, see Table 1.

Fig. 7 The most stable adsorption configurations for two N₂, NO, NO₂, NH₃, CO, CO₂, CH₄, SO₂, and H₂S on pristine ASiNR.

Fig. 8 Quantum conductance of pristine SiNR before and after (a) two-N₂, -NO, -NO₂, -NH₃, (b) -CO, -CO₂, -CH₄, (c) -SO₂ and -H₂S gas molecules adsorption.

It is predicted that the edges of SiNRs are not well controlled similar to GNRs.^83,84 As a result, it seems difficult to achieve a fully edge hydrogenated SiNR without any dangling bond (DB) defects. It is found that edge DB defects are much more probable than surface DB defects in silicon nanowires.^85–88 Therefore, it is necessary to investigate the sensing capability of a SiNR with an edge DB defect. To this end, a hydrogen atom is removed from one silicone edge atom, and only the gas adsorption above the DB defect is tested. The most stable adsorption configurations of gas molecules on SiNR with an edge DB defect are illustrated in Fig. 9.

Fig. 9 The most stable adsorption configurations for N₂, NO, NO₂, NH₃, CO, CO₂, CH₄, SO₂, and H₂S on ASiNR with an edge DB defect. The bond lengths (in unit of Å) of molecule and binding distances between the molecule and the SiNR are also given.

Fig. 10 shows the conductance modulation of ASiNR with an edge DB defect upon adsorption of gas molecules. As one can see the edge DB defects do not limit the sensing capability of the SiNR. The absolute value of adsorption energies of all gas molecules on SiNR with an edge DB defect, see Table 1, are increased, leading to detectable modifications in SiNR's conductance. Based on their adsorption energies, it can be understood that NO (−3.23 eV), NO₂ (−2.94 eV), NH₃ (−1.24 eV), CO (−1.52 eV), and SO₂ (−2.72 eV) are strongly chemisorbed on the SiNR with a DB defect, while H₂S (−0.79 eV) is weakly chemisorbed on the SiNR. However, the N₂ (−0.32 eV) and CH₄ (−0.22 eV) molecules are still physisorbed on the SiNR with small adsorption energy. The Mulliken population analysis of the SiNR with an edge DB defect and adsorbed molecule shows that a DB defect slightly decreases the charge transfer between SiNR and gas molecules, see Table 1.

Fig. 10 Quantum conductance of SiNR with an edge DB defect before and after (a) N₂, NO, NO₂, NH₃, (b) CO, CO₂, CH₄, (c) SO₂ and H₂S gas molecules adsorption.

Here, we study the effects of doping on the gas molecules sensing capability of SiNR. To this end, doping effects of SiNRs with N and B impurities have been considered. It has been found that a single N or B dopant prefers to be substituted with Si edge atoms because the formation energy of N or B impurity at the edge is lower than other positions of the nanoribbon.⁵⁹ To focus on the effects of doping on the sensing capability of the SiNR, the gas molecules adsorption above the B and N impurities and their nearest neighbours are studied. It is discovered that the gas molecules energetically tend to be chemisorbed on B atom in B-doped SiNR, and chemisorbed on the Si atoms nearest to the N atom in N-doped SiNR. The adsorption energies of gas molecules on the doped SiNR are larger than that of pristine SiNR, proving that doping improves sensing capabilities of SiNR, see Table 1. These findings concur with previous studies on gas detection based on doped silicene.⁵⁰

The most stable configuration of adsorbed gas molecules on N-doped and B-doped ASiNR are depicted in Fig. 11. For CO gas molecule on N-doped SiNR, two covalent bonds are formed between C and the two nearest Si atoms to the N atom, while, for B-doped SiNR, C atom and B atom are covalently bonded. The adsorption energies of CO molecule on N- and B-doped SiNR are 2.04 and 1.72, greater than that of pristine SiNR. The NO gas molecule is also strongly chemisorbed on N- and B-doped SiNR. Although three covalent bonds are formed between NO and nearest Si atoms to the N atom in N-doped SiNR, the adsorption of NO on B-doped SiNR totally destroys the configuration of atoms on the edge. The most stable configurations for NO₂, NH₃, and SO₂ gas molecules adsorption on N- and B-doped SiNR are quite similar to the pristine SiNR. The only difference is that they prefer to be adsorbed on top of a B atom and the nearest Si atom to the N atom. The CO₂ and CH₄ molecules on the N- and B-doped SiNR and N₂ on the N-doped SiNR are still physisorbed. However, the N₂ molecule is chemically adsorbed on B-doped SiNR with adsorption energy of −0.55 eV which is 1.79 times greater than that of pristine SiNR. Similar to gas adsorption on pristine SiNR, the conductances of N-doped and B-doped SiNR are detectably changed upon adsorption of NO, NO₂, NH₃, CO, SO₂, and H₂S, see Fig. 12. Furthermore, doping a B atom into the SiNR can improve the N₂ gas sensing capability of SiNR, compared to that of pristine SiNR. Charge transfer analysis shows that N₂, NO₂, NH₃, CO, CO₂, H₂S, and SO₂ act as donors, while NO and CH₄ act as acceptors for N-doped SiNR. However, for B-doped SiNR, all gases have electron donating capability except CH₄, see Table 1.

Fig. 11 The most stable adsorption configurations for N₂, NO, NO₂, NH₃, CO, CO₂, CH₄, SO₂, and H₂S on (a) N-doped and (b) B-doped ASiNR. The cyan, green, blue, red, grey, yellow, and pink balls represent Si, H, N, O, C, S, and B atoms, respectively. The bond lengths (in unit of Å) of molecule and binding distances between the molecule and the SiNR are also given.

Fig. 12 Quantum conductance of N-doped SiNR before and after (a) N₂, NO, NO₂, NH₃, (b) CO, CO₂, CH₄, (c) SO₂ and H₂S gas molecules adsorption. Quantum conductance of B-doped SiNR before and after (d) N₂, NO, NO₂, NH₃, (e) CO, CO₂, CH₄, (f) SO₂ and H₂S gas molecules adsorption.

Finally, a design of SiNR-based sensor to detect gas molecules is proposed, as shown in the inset of Fig. 13(a). Two semi-infinite 4-ZSiNRs as the leads and a 98 Å long 7-ASiNR gas sensing zone are the main elements of the device. We have assumed that 5 gas molecules with an average distance of 18 Å are adsorbed on the edge sites of the detection zone. By applying a bias voltage across the leads, conductance can be measured before and after adsorption. To assess the sensing performance of the proposed device for the gas molecules the current versus bias voltage is calculated in Fig. 13(a). For clarity and brevity, the I–V_bias curves for 7-ASiNR before and after adsorption of NO₂ and NO molecules are considered because the currents induced by other molecules are much smaller than that of NO₂ and NO molecules. As can be seen, the current is always almost zero for pristine 7-ASiNR before adsorption of molecules since 7-ASiNR is a semiconductor with a direct band gap of 0.56 eV, as shown in Fig. 13(b). However, upon NO₂ adsorption, the current remarkably increases as a function of increasing voltage bias, the curve is almost linear indicating a metallic behavior. The current can be also increased by NO adsorption; however, the current value is much smaller than that of NO₂. These results can be confirmed by analyzing the band structures of 7-ASiNR after NO and NO₂ adsorption, as shown in Fig. 13(b). As mentioned before, the gas molecules are able to tune the electronic properties of SiNRs. The adsorption of NO₂ and NO on 7-ASiNR can transform the system to n-type and p-type semiconductors, respectively, due to the deep defect states induced by gas molecules, as shown in Fig. 13(b). These n- and p-type semiconducting SiNRs can be applied in nanoscale electronic devices such as a p–n junction diode and n- or p-type FETs. It is also expected that NO₂ gas molecule adsorption by SiNR can give rise to a conductance enhancement because of the metallic behavior of SiNR after NO₂ molecule adsorption. However, for NO molecule adsorption, the conductance may be enhanced by some amount since the NO adsorption makes the SiNR a p-type semiconductor.

Fig. 13 (a) I–V_bias curves for the SiNR sensor before and after NO and NO₂ adsorption. The inset shows the schematic of a SiNR-based sensor which consists of a semiconducting 7-ASiNR (detection region) and two metallic 4-ZSiNRs (electrodes). (b) Band structures of pristine 7-ASiNR before and after NO and NO₂ adsorption.

IV Conclusions

In summary, we employed first-principle calculations to explore the adsorption geometry, adsorption energy, charge transfer, electronic band structure, and quantum conductance modulation of SiNRs with gas molecules (N₂, NO, NO₂, NH₃, CO, CO₂, CH₄, SO₂, and H₂S) adsorption. Our results reveal that SiNRs are capable of detecting CO, NH₃, and H₂S with high sensitivity because they are chemically adsorbed on the SiNR and transfer a few electrons to the SiNR. Quantum conductance modulation of SiNRs is clearly detectable upon chemisorption of gas molecules. Furthermore, SiNRs are not appropriate to serve as a sensor for NO, NO₂, and SO₂ due to the fact that they are strongly chemisorbed with covalent bonds to SiNRs and transfer a large amount of electrons to the SiNR. However, the strong bonding of NO and NO₂ molecules is an effective way to tune the electronic properties of SiNRs to fabricate p-type and n-type transistors, respectively. The other gas molecules are physisorbed. We also found that increasing the density of gas molecules will result in more significant changes in quantum conductance. In addition, dangling bond defects which are unavoidable through fabrication process will not hinder the sensing capability of SiNRs. The sensing capability of a SiNR-based sensor can also be increased by either N or B doping. It is found that doping a SiNR with a B atom can enhance the detection capability of N₂ gas molecules. On the basis of our results, SiNRs can be considered as a promising material to detect individual gas molecules.

Acknowledgements

This work was supported in part by the Florida Education Fund's McKnight Junior Faculty Fellowship.

References

1. K. S. Novoselov, A. K. Geim, S. Morozov, D. Jiang, Y. Zhang, S. Dubonos, I. Grigorieva and A. Firsov, Science, 2004, 306, 666–669.
2. K. Novoselov, A. K. Geim, S. Morozov, D. Jiang, M. Katsnelson, I. Grigorieva, S. Dubonos and A. Firsov, Nature, 2005, 438, 197–200.
3. S. Basu and P. Bhattacharyya, Sens. Actuators, B, 2012, 173, 1–21.
4. F. Schedin, A. K. Geim, S. V. Morozov, E. W. Hill, P. Blake, M. I. Katsnelson and K. S. Novoselov, Nat. Mater., 2007, 6, 652–655.
5. Q. He, S. Wu, Z. Yin and H. Zhang, Chem. Sci., 2012, 3, 1764–1772.
6. J. D. Fowler, M. J. Allen, V. C. Tung, Y. Yang, R. B. Kaner and B. H. Weiller, ACS Nano, 2009, 3, 301–306.
7. S. Some, Y. Xu, Y. Kim, Y. Yoon, H. Qin, A. Kulkarni, T. Kim and H. Lee, Sci. Rep., 2013, 3, 1868.
8. O. Leenaerts, B. Partoens and F. M. Peeters, Phys. Rev. B: Condens. Matter Mater. Phys., 2008, 77, 125416.
9. Y. Zhang, Y. Chen, K. Zhou, C. Liu, J. Zeng, H. Zhang and Y. Peng, Nanotechnology, 2009, 20, 185504.
10. J. Dai, J. Yuan and P. Giannozzi, Appl. Phys. Lett., 2009, 95, 232105.
11. J. T. Robinson, F. K. Perkins, E. S. Snow, Z. Q. Wei and P. E. Sheehan, Nano Lett., 2008, 8, 3137.
12. G. Lu, L. E. Ocola and J. Chen, Nanotechnology, 2009, 20, 445502.
13. J. Dai and J. Yuan, Phys. Rev. B: Condens. Matter Mater. Phys., 2010, 81, 165414.
14. K. Takeda and K. Shiraishi, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 14916.
15. G. G. Guzmán-Verri and L. L. Y. Voon, Phys. Rev. B: Condens. Matter Mater. Phys., 2007, 76, 075131.
16. S. Cahangirov, M. Topsakal, E. Aktürk, H. Şahin and S. Ciraci, Phys. Rev. Lett., 2009, 102, 236804.
17. X. Q. Wang, H. D. Li and J. T. Wang, Phys. Chem. Chem. Phys., 2012, 14, 3031–3036.
18. F. Zheng and C. Zhang, Nanoscale Res. Lett., 2012, 7, 422.
19. C. C. Liu, W. X. Feng and Y. G. Yao, Phys. Rev. Lett., 2011, 107, 076802.
20. C. Xu, G. Luo, Q. Liu, J. Zheng, Z. Zhang, S. Nagase, Z. Gao and J. Lu, Nanoscale, 2012, 4, 3111–3317.
21. L. Chen, B. Feng and K. Wu, Appl. Phys. Lett., 2013, 102, 081602.
22. P. Vogt, P. De Padova, C. Quaresima, J. Avila, E. Frantzeskakis, M. C. Asensio, A. Resta, B. Ealet and G. Le Lay, Phys. Rev. Lett., 2012, 108, 155501.
23. B. Feng, Z. Ding, S. Meng, Y. Yao, X. He, P. Cheng, L. Chen and K. Wu, Nano Lett., 2012, 12, 3507–3511.
24. J. Mannix, B. Kiraly, B. L. Fisher, M. C. Hersam and N. P. Guisinger, ACS Nano, 2014, 8, 7538–7547.
25. L. Meng, Y. Wang, L. Zhang, S. Du, R. Wu, L. Li, Y. Zhang, G. Li, H. Zhou and W. A. Hofer, et al., Nano Lett., 2013, 13, 685–690.
26. A. Fleurence, R. Friedlein, T. Ozaki, H. Kawai, Y. Wang and Y. Yamada- Takamura, Phys. Rev. Lett., 2012, 108, 245501.
27. T. Aizawa, S. Suehara and S. Otani, J. Phys. Chem. C, 2014, 118, 23049–23057.
28. T. P. Kaloni, M. Tahir and U. Schwingenschlögl, Sci. Rep., 2013, 3, 3192.
29. H. Liu, J. Gao and J. Zhao, J. Phys. Chem. C, 2013, 117, 10353.
30. Y. Ding and Y. Wang, Appl. Phys. Lett., 2013, 103, 043114.
31. Y. Cai, C.-P. Chuu, C. M. Wei and M. Y. Chou, Phys. Rev. B: Condens. Matter Mater. Phys., 2013, 88, 245408.
32. M. Houssa, B. van den Broek, E. Scalise, G. Pourtois, V. V. Afanas'eva and A. Stesmansa, Phys. Chem. Chem. Phys., 2013, 15, 3702.
33. W.-F. Tsai, C.-Y. Huang, T.-R. Chang, H. Lin, H.-T. Jeng and A. Bansil, Nat. Commun., 2013, 4, 1500.
34. Z. Ni, Q. Liu, K. Tang, J. Zheng, J. Zhou, R. Qin, Z. Gao, D. Yu and J. Lu, Nano Lett., 2012, 12, 113–118.
35. L. Tao, E. Cinquanta, D. Chiappe, C. Grazianetti, M. Fanciulli, M. Dubey, A. Molle and D. Akinwande, Nat. Nanotechnol., 2015, 10, 227–231.
36. H. Sadeghi, S. Bailey and C. J. Lambert, Appl. Phys. Lett., 2014, 104, 103104.
37. R. G. Amorim and R. H. Scheicher, Nanotechnology, 2015, 26, 154002.
38. N. D. Drummond, V. Zólyomi and V. I. Fal'ko, Phys. Rev. B: Condens. Matter Mater. Phys., 2012, 85, 075423.
39. T. P. Kaloni, G. Schreckenbach and M. S. Freund, J. Phys. Chem. C, 2014, 118, 23361–23367.
40. P. Lazar, F. Karlický, P. Jurečka, M. Kocman, E. Otyepková, K. Šafářová and M. Otyepka, J. Am. Chem. Soc., 2013, 135, 6372–6377.
41. X. Lin and J. Ni, Phys. Rev. B: Condens. Matter Mater. Phys., 2012, 86, 075440.
42. J. Sivek, H. Sahin, B. Partoens and F. M. Peeters, Phys. Rev. B: Condens. Matter Mater. Phys., 2013, 87, 085444.
43. H. Sahin and F. M. Peeters, Phys. Rev. B: Condens. Matter Mater. Phys., 2013, 87, 085423.
44. G. A. Tritsaris, E. Kaxiras, S. Meng and E. Wang, Nano Lett., 2013, 13, 2258–2263.
45. J. Wang, J. Li, S.-S. Li and Y. Liu, J. Appl. Phys., 2013, 114, 124309.
46. B. Huang, H. J. Xiang and S.-H. Wei, Phys. Rev. Lett., 2013, 111, 145502.
47. W. Hu, X. Wu, Z. Li and J. Yang, Phys. Chem. Chem. Phys., 2013, 15, 5753.
48. J.-W. Feng, Y.-J. Liu, H.-X. Wang, J.-X. Zhao, Q.-H. Cai and X.-Z. Wang, Comput. Mater. Sci., 2014, 87, 218.
49. W. Hu, N. Xia, X. Wu, Z. Li and J. Yang, Phys. Chem. Chem. Phys., 2014, 16, 6957.
50. J. Prasongkit, R. G. Amorim, S. Chakraborty, R. Ahuja, R. H. Scheicher and V. Amornkitbamrung, J. Phys. Chem. C, 2015, 119, 16934–16940.
51. R. Chandiramoulia, A. Srivastavab and V. Nagarajan, Appl. Surf. Sci., 2015, 351, 662–672.
52. G. García, M. Atilhan and S. Aparicio, Phys. Chem. Chem. Phys., 2015, 17, 16315–16326.
53. X. Wang, H. Liu and S. T. Tu, RSC Adv., 2015, 5, 65255–65263.
54. A. Kara, H. Enriquez, A. P. Seitsonen, L. C. Lew Yan Voon, S. Vizzini, B. Aufray and H. Oughaddou, Surf. Sci. Rep., 2012, 67, 1–18.
55. E. Akbari, Z. Buntat, A. Afroozeh, S. E. Pourmand, Y. Farhang and P. Sanati, RSC Adv., 2016, 6, 81647–81653.
56. A. Lopez-Bezanilla, J. Phys. Chem. C, 2014, 118, 18788–18792.
57. N. Gao, J. Li and Q. Jiang, Chem. Phys. Lett., 2014, 592, 222–226.
58. N. Gao, W. T. Zheng and Q. Jiang, Phys. Chem. Chem. Phys., 2012, 14, 257–261.
59. S. M. Aghaei, M. M. Monshi, I. Torres and I. Calizo, RSC Adv., 2016, 6, 17046.
60. F. Pan, Y. Wang, K. Jiang, Z. Ni, J. Ma, J. Zheng, R. Quhe, J. Shi, J. Yang, C. Chen and J. Lu, Sci. Rep., 2015, 5, 9075.
61. S. M. Aghaei and I. Calizo, J. Appl. Phys., 2015, 118, 104304.
62. S. M. Aghaei and I. Calizo, in Proceeding of IEEE SoutheastCon (SECon-2015), Fort Lauderdale, April 9–12, 2015, pp. 1–6.
63. Y. Ding and J. Ni, Appl. Phys. Lett., 2009, 95, 083115.
64. B. Aufray, A. Kara, S. Vizzini, H. Oughaddou, C. Leandri, B. Ealet and G. Le Lay, Appl. Phys. Lett., 2010, 96, 183102.
65. P. De Padova, C. Quaresima, C. Ottaviani, P. M. Sheverdyaeva, P. Moras, C. Carbone, D. Topwal, B. Olivieri, A. Kara, H. Oughaddou, B. Aufray and G. Le Lay, Appl. Phys. Lett., 2010, 96, 261905.
66. P. De Padova, O. Kubo, B. Olivieri, C. Quaresima, T. Nakayama, M. Aono and G. Le Lay, Nano Lett., 2012, 12, 5500–5503.
67. M. R. Tchalala, H. Enriquez, A. J. Mayne, A. Kara, S. Roth, M. G. Silly, A. Bendounan, F. Sirotti, T. Greber, B. Aufray, G. Dujardin, M. A. Ali and H. Oughaddou, Appl. Phys. Lett., 2013, 102, 083107.
68. A. N. Abbas, G. Liu, B. Liu, L. Zhang, H. Liu, D. Ohlberg, W. Wu and C. Zhou, ACS Nano, 2014, 8, 1538–1546.
69. A. N. Abbas, G. Liu, A. Narita, M. Orosco, X. Feng, K. Müllen and C. Zhou, J. Am. Chem. Soc., 2014, 136, 7555.
70. B. Huang, Z. Y. Li, Z. R. Liu, G. Zhou, S. G. Hao, J. Wu, B. L. Gu and W. H. Duan, J. Phys. Chem. C, 2008, 112, 13442–13446.
71. A. Saffarzadeh, J. Appl. Phys., 2010, 107, 114309.
72. T. H. Osborn and A. A. Farajian, Nano Res., 2014, 7, 945–952.
73. J. Taylor, H. Guo and J. Wang, Phys. Rev. B: Condens. Matter Mater. Phys., 2001, 63, 245407.
74. M. Brandbyge, J.-L. Mozos, P. Ordejon, J. Taylor and K. Stokbro, Phys. Rev. B: Condens. Matter Mater. Phys., 2002, 65, 165401.
75. Atomistix Toolkit version 2015.0, QuantumWise, Copenhagen, Denmark, see http://www.quantumwise.com.
76. S. Grimme, J. Comput. Chem., 2006, 27, 1787.
77. S. Grimme, C. Mück-Lichtenfeld and J. Antony, J. Phys. Chem., 2007, 111, 11199–11207.
78. F. Boys and F. Bernardi, Mol. Phys., 1970, 19, 553.
79. M. E. Dávila, A. Marele, P. De Padova, I. Montero, F. Hennies, A. Pietzsch, M. N. Shariati, J. M. Gómez-Rodríguez and G. LeLay, Nanotechnology, 2012, 23, 385703.
80. T. H. Osborn, A. A. Farajian, O. V. Pupyesheva, R. S. Aga and L. L. Y. Voon, Chem. Phys. Lett., 2011, 511, 101–105.
81. S. Zhong, F. Ning, F. Rao, X. Lei, M. Wu and L. Zhou, Int. J. Mod. Phys. B, 2016, 1650176.
82. K. K. Paulla and A. A. Farajian, J. Phys. Chem. C, 2013, 117, 12815–12825.
83. M. Y. Han, B. Özyilmaz, Y. B. Zhang and P. Kim, Phys. Rev. Lett., 2007, 98, 206805.
84. Z. Chen, Y.-M. Lin, M. J. Rooks and P. Avouris, Phys. E, 2007, 40, 228–232.
85. L. B. Luo, X. B. Yang, F. X. Liang, J. S. Jie, C. Y. Wu, L. Wang, Y. Q. Yu and Z. F. Zhu, J. Phys. Chem. C, 2011, 115, 24293–24299.
86. J. S. Park, B. Ryu, C. Y. Moon and K. J. Chang, Nano Lett., 2010, 10, 116–121.
87. L. B. Luo, X. B. Yang, F. X. Liang, H. Xu, Y. Zhao, X. Xie, W. F. Zhang and S. T. Lee, J. Phys. Chem. C, 2011, 115, 18453–18458.
88. K.-H. Hong, J. Kim, J. H. Lee, J. Shin and U. I. Chung, Nano Lett., 2010, 10, 1671–1676.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra21293j

---