CO₂ adsorption on single-walled boron nitride nanotubes containing vacancy defects

Abstract

The adsorption of a CO₂ molecule on the vacancy defect type of armchair (5,5) and zigzag (10,0) single-walled boron nitride nanotubes was studied based on Density Functional Theory (DFT). Vacancy defects were studied and the geometrical modifications implemented on the original hexagonal lattice yielded a considerable level of changes in the electronic properties. These changes are reflected in a greater level of CO₂ reactivity in relation to the adsorption over a pristine structure. For all types of studied CO₂ molecule interaction, we have found a chemical adsorption process based on binding energy. Furthermore, the CO₂ adsorption takes place on the top of the vacancy region. A decomposition state was observed when the CO₂ molecule interacted with the armchair nanotube with a vacancy on the nitrogen site. By comparing the values of the adsorption energies with those from other defect approaches present in the literature, we conclude that the proposed protocol presents a possible tool to develop stable and sensible carbon dioxide sensors.

---

I. Introduction

Boron nitride nanotubes (BNNTs) were first predicted in the mid 90’s through tight binding calculations, and synthesized shortly after that by means of an arc-discharge method.^1,2 Since the advent of these studies, several theoretical and experimental investigations concerning the properties of these novel materials have been carried out.^3,4 These efforts, together with the many advantages these materials present, have turned BNNTs into fundamental tools for the nanotechnology field. Among these advantages is the fact that BNNTs tend to present an indirect band gap greater than 2 eV, regardless of the geometrical parameters considered.¹ Also, these materials present great chemical stability,⁵ excellent mechanical resistance,⁶ high thermal conductivity⁷ and considerable resistance to both heat and electric currents.⁸ Indeed, these features have turned BNNTs into even more interesting materials than their carbon counterparts,⁹ as far as gas sensor technology and other kinds of nanotechnology applications are concerned.

All the aforementioned advantages are known to be valid for ideal nanotubes, i.e., structures which are defect free. In the actual process of fabrication, however, defects that are either spontaneously or artificially produced may alter the physical and electronic properties of BNNTs. These changes in properties might lead to effects that can alternatively limit or enhance the applicability of these structures. In this sense, one of the early works in this field has shown that mechanically-induced defects lead to the formation and dislocation of dipoles on BNNT walls, provided certain conditions of temperature and pressure are considered.¹⁰ Similarly, another work, which considered transition electronic microscopy (TEM) together with theoretical modeling, suggested that defects created through electronic irradiation are mainly of a divacancy nature. This study has also shown that a set of several vacancies induces the creation of extended defects, which locally alter the nanotube diameter and chirality, thus affecting their electronic and optical properties.¹¹ TEM is an appropriate experimental method for imaging defective structures and also for inducing in situ the appearance of defects. By using energetic electrons (100 keV), this technique is capable of removing single atoms from the lattice through direct knock-on collisions.¹¹ An interesting signatures of point defects (such as single vacancies) can be seen in the high-resolution electron microscopy images of a bundle of BNNTs taken after being irradiated by an energetic electron beam.¹¹ Activated boron nitride, as an effective adsorbent for metal ions and organic pollutants, has been also studied in the literature.¹²

Some relevant theoretical works also propose that the presence of vacancies raises the concentration of electrons or holes depending on if the defect is produced in the boron or in the nitrogen atom sites.¹³ Likewise, a recent work by Li et al.¹⁴ shows that a charge redistribution takes place near the defect site on the wall of an armchair BNNT, such that this effect might increase the chemical reactivity of BNNTs regarding the capture of molecular species. Moreover, there is an urgent need for new carbon dioxide capture strategies. Therefore, a systematic investigation on the role played by vacancy-type defects on favoring the capture of CO₂ is highly desired. The importance of this particular molecule lies in fact that CO₂ is one of the main anthropic greenhouse gases, thus dangerously inducing severe climate changes.¹⁵

Recently, a theoretical study based on Density Functional Theory (DFT) calculations has shown that a single graphene-like hexagonal BN sheet endowed with a vacancy on the boron site could efficiently capture a carbon dioxide molecule.¹⁶ It was reported that in one of the steps of the adsorption process, the CO₂ was broken down into an oxygen molecule and a carbon atom captured by the lattice. This step is observed to take place through an intermediate chemisorption state on the defective BN sheet. Therefore, we make use of single-walled armchair (henceforward referred to as BNNT(5,5)) and zigzag (BNNT(10,0)) boron nitride nanotubes to investigate the influence of nitrogen or boron sites on the reactivity of a CO₂ molecule. Specifically, the main question is regarding the chiral angle effects as well as the influence on the CO₂ adsorption process due to the defect site.

The remainder of this work is organized as follows: Section II presents the computational settings, followed by our results and their discussions in Section III; we summarize the main features of this work in Section IV.

II. Computational details

Density functional theory calculations of CO₂ adsorption were carried out on infinitely large boron nitride nanotubes by means of the DMol³ code^17,18 to compute equilibrium geometries, total energies, charge analysis and electronic densities. The electronic exchange correlation was treated within the generalized gradient approximation (GGA)¹⁹ with the PW91 functional.²⁰ The PW91 functional was recently used to study the interaction of BNNT with gas molecules.²¹

The positions of all the atoms were fully relaxed until the following convergence criteria were met: 0.001 Ha Å⁻¹ for the force constant and 0.003 Å for displacement. The self-consistent field computations criterion was chosen to be 10⁻⁶ Ha. The electronic wave functions were expanded in a 4.4 version double numerical plus polarization basis set (DNP) truncated at a real space cut-off of 4.1 Å. The DNP basis set was found to lead to a small basis set superposition error (BSSE) (about 1.62 kJ mol⁻¹).²² Therefore, we have used BSSE correction. Due to the presence of boron and nitrogen atoms in the model, all calculations were spin unrestricted. A 0.005 Ha smearing²³ and 6 Pulay direct inversion of the iterative subspace (DIIS)²⁴ was applied to the system to facilitate convergence of the electronic structures.

It must be noted that BNNTs used in nanotechnology applications are known to be of very large length. However, the explicit consideration of such a large system tends to be computationally expensive. In order to surpass this difficulty, while treating the system with a suitable approach, the 3D periodic boundary conditions were applied to the whole system, accomplishing simulation of the infinitely large BNNTs. The size of the vacuum space in-between two neighbor image tubes was set to be 20 Å to prevent the interaction between the atoms of each unit cell. The Brillouin zone for a single cell was sampled by 1 × 1 × 11 special k-points.

In order to compute the adsorption energy,²⁵ we make use of the total energy of the isolated BNNT (E_BNNT) and the CO₂ molecule energy (E_CO2), as well as of the energy of the complex system composed of the gas molecule interacting with the nanotube (E_BNNT+CO2). The equation below employs these terms in the calculation of the adsorption energy:

E_ads = E_BNNT+CO2 − E_BNNT − E_CO2.

It is important to remark that negative and positive values of E_ads denote exothermic and endothermic adsorption, respectively. The E_ads values and other properties are calculated based only on the lowest-energy structural configuration for the CO₂–BNNT systems. In this work, the charge distribution on the system was analyzed using the Mulliken population method,²⁶ despite its known limitations, since our interest was to analyze the trend of charge distribution across the same models.

Our work concerns the influence of the vacancy defects of BNNTs and of their chiralities during the CO₂ adsorption process. In other words, we compare the adsorption of the gas molecule with and without structural defects for two different symmetries.

III. Results and discussions

In order to better present our simulation results, we divided the present section in three subsections. The first one (III A) considers the interaction of pristine nanotubes interacting with a CO₂ molecule. Subsection III B regards the effects of vacancy type on the properties of the nanotube itself. Finally, subsection III C considers the adsorption process of the CO₂ molecule on vacancy endowed nanotubes.

A. Pristine BNNTs interacting with CO₂

Throughout the work, we adopted single-walled boron nitride nanotubes of the zigzag (5,5) and armchair (10,0) chiralities. We first take into account the interaction of a CO₂ molecule on a defect free nanotube for means of future comparison. Consequently, we can observe the effect of vacancy defects on the gas adsorption. We begin by separately performing the geometry optimization of a nanotube of each chirality. This was accomplished through the use of the DMol³ package by considering a total of 60 atoms for the BNNT(5,5), and 80 atoms for the BNNT(10,0), together with periodic boundary conditions in order to mitigate border effects. A B–N length of approximately 1.450 Å was obtained for the pristine nanotubes, regardless their chirality. The free CO₂ molecule achieved a typical linear structure with a C–O bond length of 1.166 Å.

As a next step, we placed the CO₂ molecule in the neighborhood of pristine BNNTs, and carried out a geometry optimization. Fig. 1 presents the frontal and side views of the equilibrium configuration for the two different nanotube chiralities considered. One can observe that, regardless of the chirality, both the BNNTs and the CO₂ molecule preserve their original geometrical characteristics. For the case of BNNT(5,5), the CO₂ molecule configuration was parallel to the tube axis and located outside the BNNT wall, with its carbon atom directly in front of the nitrogen atom of the BNNT, and its oxygen atoms close to the boron atoms of the BNNT.

Fig. 1 Frontal and side views of the equilibrium configurations for the CO₂ molecule related to BNNTs: (a) and (b) BNNT(5,5); (c) and (d) BNNT(10,0). The blue atoms represent nitrogen, the pink atoms represent boron, the red atoms are for oxygen and carbon is the gray atom.

This behavior can be understood in terms of the polar character of the B–N bonding, which creates a higher charge concentration over the nitrogen atoms, due to the greater electronegativity of nitrogen when compared to boron. Thus, it is expected that the CO₂ atoms are organized by means of electronic affinity: the carbon atom of the carbon dioxide is positively polarized and interacts with a nitrogen atom whereas the oxygen atoms are attracted by the boron of the nanotubes.

As for BNNT(10,0), it was calculated that the CO₂ molecule was positioned above the center of a nanotube hexagon. This configuration was achieved after the molecule performed a 38° rotation in relation to its original position around its axis, with its oxygen atoms directly above the nanotube boron atoms. In this case, as in previous reported works,²⁷ the zigzag bonding pattern of the B–N plays an important role in the molecular orientation related to the tube axis.

Table 1 presents the adsorption energies (E_ads), equilibrium distances (D), charge transfers (Q_T) and energy gaps (E_g) for the BNNT–CO₂ for both considered chiralities. As it is shown, the optimized configurations present small values of adsorption energies and large equilibrium distances, which indicate that the interaction between the carbon dioxide molecule and the pristine boron nitride nanotube is of a physical adsorption nature. In this case, the interaction is known to be mediated by van der Waals type interactions. A comparison of the values obtained for E_ads between the (5,5) and (10,0) tubes highlights a small difference of 0.02 eV. This fact suggests a slight tendency of the molecule to interact better with the zigzag rather than with the armchair pristine nanotube.

Table 1 Results for the CO₂ adsorption on the pristine (5,5) and (10,0) BNNT optimized geometries

System	Eads (eV)	D (Å)	Eg (eV)	QT (e)
BNNT(5,5)–CO2	−0.16	3.00	4.35	−0.003
BNNT(10,0)–CO2	−0.18	2.95	4.03	−0.005

---

The charge analysis, carried out through the Mulliken method, suggests that about 0.61e and 0.54e are transferred from the boron atom to the adjacent nitrogen atom in the (5,5) and (10,0) BNNTs, respectively. This fact endorses the partially ionic character of the B–N bond in BNNTs. By using a Density of States (DOS) analysis, our results also indicate the presence of band gaps of 4.35 eV and 4.03 eV for the (5,5) and (10,0) nanotubes, respectively. These results are in agreement with previous data present in the literature.^13,28 From Table 1, we also note that no appreciable energy gap difference is observed for the BN nanotubes. This is an indication that the physical adsorption does not affect the electronic properties of pristine BNNTs. It is important to remark that these results are consistent with the fact that pure BNNTs are quasi-inert structures. Indeed, the low values observed for the interaction energies are a measure of the high stability of these materials without defects.²

B. BN nanotubes with vacancy-type defect

As a weak interaction was obtained (between the pristine nanotube and the carbon dioxide molecule), we intent to verify whether or not a vacancy-type defect can favor the adsorption phenomena. In order to do so, first a prior analysis of this kind of defect on the nanotube itself must be performed. This subsection is devoted to understanding the effects of a vacancy defect on the geometrical and electronic properties of BNNTs. The vacancy defect was introduced by removing a boron or nitrogen atom from the nanotube lattice. We also compare the behavior of a boron vacancy, which is denoted by BNNT(5,5)V_B and BNNT(10,0)V_B, with a nitrogen vacancy, BNNT(5,5)V_N and BNNT(10,0)V_N, in terms of the geometrical and electronic properties of the nanotubes.

We present, in Fig. 2, the optimized geometries for the vacancy defect BNNTs. The geometry deformation due to the vacant atom site is shown for all structures. Fig. 2a suggests a stretching in the radial direction for the BNNT(5,5)V_B in the region of the removed boron atom. One can note that the relaxation induces the displacement of a nitrogen atom (N₂₁) toward the tube and of two other nitrogen atoms, N₂₃ and N₄, outside the tube. In this case, we observed the formation of a 12-fold ring, without new bonding types involving nitrogen close to the vacancy due to the dangling bond. The equilibrium distances between the nitrogen atoms were found to be 2.67 Å for N₂₃–N₄, 2.38 Å for N₂₃–N₂₁ and 2.38 Å for N₂₁–N₄. These values indicate that, indeed, no new bonding was established between the nitrogen atoms.

Fig. 2 Frontal and side views for the optimized geometries of the vacancy defect BNNTs. (a) BNNT(5,5)V_B, (b) BNNT(5,5)V_N, (c) BNNT(10,0)V_B, (d) BNNT(10,0)V_N. The blue atoms represent nitrogen and the pink atoms represent boron.

Fig. 2b concerns the effect of a nitrogen vacancy instead of the boron vacancy of Fig. 2a. In this case, one can see that deformations appeared on a smaller scale than in the previous case. The two-coordinated boron atom B₂₀ relaxes outside the tube whereas the opposite happens for both B₂₂ and B₃₉. As can be seen, the formation of new bonding between atoms B₂₂ and B₃₉ takes place, thus altering the original hexagonal lattice around the vacancy site. The distance between the aforementioned boron atoms (B₃₉–B₂₂) is 1.82 Å, which corresponds to the formation of a five-fold and nine-fold adjacent rings arrangement. It is important to remark that these geometrical results for the vacancy defect of BNNT(5,5) are in agreement with other previous results from the literature.¹⁴

Switching to the zigzag chirality, Fig. 2c represents the BNNT(10,0)V_B case. A pronounced radial deformation is observed around the vacancy site. Atom N₃₉ is displaced outside the nanotube, whereas atoms N₇₇ and N₄₁ move inside the nanotube. Also, a local rebuilding of the bonding between nitrogen atoms N₇₇ and N₄₁ takes place, thus promoting changes in the original hexagonal lattice. The bond distance between nitrogen atoms N₇₇–N₄₁ is 1.46 Å, thus being responsible for the formation of a five-fold and nine-fold adjacent rings arrangement.

A smaller level of geometrical deformation was observed to take place for the BNNT(10,0)V_N, as can be seen in Fig. 2d. A weak bonding is observed between boron atoms B₂ and B₃₈. The bonding length was 1.78 Å, which also results in the formation of five-fold and nine-fold adjacent rings. It can be seen that both the binding boron atoms are dislocated inside the nanotube, whereas the two-coordinated boron atom B₇₉ is pushed outward from the nanotube. We must, again, stress that these results obtained for the vacancy defect of BNNT(10,0) are in agreement with previously published work.²⁹ We thus conclude, by analyzing Fig. 2, that for both BNNT(5,5) and BNNT(10,0), the vacancy formation of boron atom causes greater structural deformation than the nitrogen vacancy.

These geometrical distortions result in important modifications of the electronic properties of the nanotubes. Fig. 3 represents the DOS due to the vacancy defect type. We emphasize that this kind of defect induces impurity states near to the Fermi level in all cases.

Fig. 3 DOS of BNNT(5,5) on the left and BNNT(10,0) on the right, for pristine and vacancy defect BNNTs with boron and nitrogen vacancies.

For BNNT(5,5)V_B, a final state in the valence band reduced the original band gap to 2.85 eV. For BNNT(5,5)V_N, the appearance of an isolated state between the end of the valence band and the beginning of the conduction band due to the vacancy formation creates a state of 1.37 eV near the valance band. The zigzag BNNT(10,0) presents the same trend in the electronic behavior changes to its armchair counterpart when vacancy defects are introduced. A final state inside the original gap of BNNT(10,0)V_B at 2.11 eV occurs near the valence band. BNNT(10,0)V_N, on the other hand, presents an isolated state of 0.97 eV between the end of the valence band and the beginning of the conduction band, close to the valence band.

The charge density analysis also reveals important information related to vacant BNNTs. Fig. 4a shows that in the case of BNNT(5,5)V_B, the charge is concentrated over the aromatic ring, particularly above the nitrogens with dangling bonds (N₂₃, N₄ and N₂₁). For BNNT(5,5)V_N (Fig. 4b), one can observe a higher charge concentration over the B₃₉ and B₂₂ bonds.

Fig. 4 Charge density of vacancy defect BNNTs. (a) BNNT(5,5)V_B, (b) BNNT(5,5)V_N, (c) BNNT(10,0)V_B, (d) BNNT(10,0)V_N.

For the BNNT(10,0)V_B (Fig. 4c), the charge is also mainly concentrated over the two-coordinated N₃₉ nitrogen atom. It is important to note that, despite the presence of the bonding between atoms N₄₁ and N₇₇, Fig. 4c does not reveal a significant charge concentration between these nitrogen species that would favor strong interaction in bound systems. BNNT(10,0)V_N (Fig. 4d), similarly to BNNT(5,5)V_N, presents a high concentration of charge over the B₃₈–B₂ bond. These results show that for both chiralities, the variation in charge distribution occurs on the vacancy region. It also demonstrates that the vacancy of a nitrogen atom induces a region with an excess of electrons, whereas that of boron causes a deficiency of electrons. Therefore, the electrostatic behavior of these defects can be useful for the capture of molecular species.

C. Vacant BN nanotubes interacting with a CO₂ molecule

After the optimization and analysis of vacant BNNTs, the goal was to investigate the interaction of these species with a CO₂ molecule. The equilibrium configurations are shown in Fig. 5, in which we can note a symmetry breaking of the carbon dioxide molecule in relation to its original state.

Fig. 5 Frontal and side views for the optimized CO₂ configurations on the vacancy defect BNNTs. (a) BNNT(5,5)V_B, (b) BNNT(5,5)V_N, (c) BNNT(10,0)V_B, (d) BNNT(10,0)V_N. The blue atoms represent nitrogen, the pink atoms represent boron, the red atoms are for oxygen and carbon is the gray atom.

Table 2 presents the comparison between E_ads, D, E_g and Q_T for the different vacancy defect BNNTs interacting with the CO₂ molecule. Considering BNNT(5,5)V_B–CO₂ (Fig. 5a), CO₂ is interacting with the nanotube through a C–N bond, whose length is 1.40 Å. In this case, the CO₂ molecule presents an inner angle of 118.4°, with the C–O distances being altered to 1.20 Å and 1.36 Å. The binding energy between the carbon dioxide molecule and BNNT(5,5)V_B is −4.42 eV (Table 2), suggesting that the interaction is of a chemical adsorption nature. The adsorption energies obtained for CO₂ in (8,0) BNNT with a boron antisite (BN) using PAW-PBE³⁰ are smaller than that obtained with vacancy formation. The charge transfer is 0.6e and occurs from the carbon to the nitrogen atom due to electronegativity differences. In the interesting case of BNNT(5,5)V_N–CO₂, we observe that a complete dissociation of the CO₂ molecule takes place. The final product is an oxygen atom and a carbon monoxide (CO) molecule. In this case, the binding energy was not included due to the carbon dioxide dissociation. Additionally, these investigations of CO₂–BNNT interactions may provide insight into the mechanism and surface characterization on nanotube planes, for the application of CO₂ storage and the screening and design of materials for CO₂ capture, and how the dissociation may affect the partial charge distribution of the vacancy sites.^31–33

Table 2 Binding energies and equilibrium distances for CO₂ adsorption on vacancy defect BNNT-CO₂ complexes

System	Eads (eV)	D (Å)	Eg (eV)	QT (e)
BNNT(5,5)VB–CO2	−4.42	1.40	3.41	−0.50
BNNT(5,5)VN–CO2	—	—	2.65	—
BNNT(10,0)VB–CO2	−4.26	1.40	2.14	−0.45
BNNT(10,0)VN–CO2	−6.53	1.30	1.31	0.82

---

As can be seen in Fig. 5b, the dissociated oxygen atom ends up binding to the three boron atoms in the vacancy region, thus yieling an average B–O bond length of approximately 1.50 Å. The substitutional energy of an oxygen atom in the (5,5) BN nanotube was calculated from the chemical potentials (E_sub = E_BNNT+O − E_BNNT − νO + νN); the covalent bond between the oxygen and the boron atoms is of about −13.46 eV, thus being of the same order of magnitude as that of the B–N bond in the nanotube. Interestingly, the CO molecule, rather than being separated from the nanotube lattice, turns out to interact with BNNT(5,5)V_N through a relatively weak bond estimated to be of about 0.8 eV, which characterizes a physisorption process. An evidence of the weak interaction between the CO molecule and the BN nanotube is that the C–O distance preserves the same 1.15 Å value of the free species. It is important to remark that experimental studies present in the literature¹⁶ have dealt with a similar situation regarding the dissociation, albeit in the scope of a graphene-type sheet with a boron vacancy interacting with a CO₂ molecule. It was reported that the CO₂ molecule suffered decomposition in such a way that the carbon atom was bound to the nitrogen of the sheet, whereas a formed oxygen molecule maintained a weak interaction with the BN sheet.¹⁶ Although in our work this process took place for a nitrogen vacancy system, these theoretical results confirm that dissociation phenomena can indeed be observed experimentally.

For BNNT(10,0)V_B–CO₂, Fig. 5c shows that the interaction with CO₂ results in the breaking of the weak bond between N₄₁ and N₇₇. In this case, the CO₂ molecule was linked to the tube by means of a bond between the carbon atom and N₄₁ as well as by another bond between the oxygen and N₃₉. The C–N₄₁ presents a bond length of 1.40 Å whereas that of O–N₃₉ is about 1.38 Å. The inner angle of CO₂ in this configuration was found to be 116.4°, and the distances are 1.20 Å and 1.36 Å. The obtained binding energy between the CO₂ molecule and BNNT(10,0)V_B was −4.26 eV, thus suggesting that this interaction is also of a chemisorption nature. For BNNT(10,0)V_N–CO₂ (Fig. 5d), a bond breaking between sites B₂–B₃₈ and the formation of two bonds between the CO₂ molecule and BNNT(10,0)V_N were observed. An interaction between the carbon atom and B₃₈, and another between the oxygen atom and B₇₉, were observed. In this configuration, the C–B₃₈ distance was found to be 1.64 Å and that of O–B₇₉ was 1.37 Å. The inner angle of the molecule changed to 120.2° and the C–O distances were altered to 1.20 Å and 1.35 Å. In this case, it is possible to note that the oxygen binding to the boron atom in the nanotube is closer than that of the carbon. The binding energy was observed to be −6.53 eV, which also suggests a chemisorption interaction between the molecule and BNNT(10,0)V_N.

These results suggest that BNNT(5,5)V_B is more suitable for capture of CO₂ than BNNT(5,5)V_N, since while the former interacts with the molecule by means of a chemical adsorption, the interaction of the latter yields the decomposition of the molecule to a carbon monoxide molecule, which is highly toxic. Regarding the chirality effects, we note that both zigzag nanotubes (BNNT(10,0)V_B and BNNT(10,0)V_N) capture the CO₂ molecule through a chemisorption process. In this case the adsorption energy of the BNNT(10,0)V_N–CO₂ complex is observed to be 2.27 eV larger than that of BNNT(10,0)V_B–CO₂. Thus, a system derived from the boron vacancy is shown to be more efficient as far as carbon dioxide adsorption is concerned.

By comparing the binding energies of the zigzag and armchair nanotubes, one concludes that the kind of defect plays an important role in the adsorption energy. Although BNNT(5,5)–CO₂ presents only a slight difference of 0.16 eV from BNNT(10,0)–CO₂ in the complex binding energy, when the structures contain a vacancy defect this difference greatly increases. Specifically, the adsorption energy is 2.11 eV greater for BNNT(5,5)V_B–CO₂ than for BNNT(10,0)V_N–CO₂.

Table 2 also shows the charge transfer between the tubes and the CO₂ molecule. We can note that in the case of the vacancy on the boron site, the nanotube presents a charge deficient region, in such a way that the carbon dioxide molecule behaves like an n-type dopant. On the other hand, when the vacancy occurs on a nitrogen site, the BNNT presents an excess of charge, in which case the carbon dioxide molecule behaves like a p-type dopant.

In order to exploit the adsorption effects of the CO₂ molecule on the electronic properties of vacancy defect BNNTs, we plotted the density of states due to the molecular adsorption in Fig. 6. It can be observed that, in the case of BNNT(5,5)V_B, the absence of an electronic state at the end of the valence band occurs after the adsorption of the CO₂ molecule. Also, the associated gap is observed to be 3.41 eV. In the case of BNNT(5,5)V_N, despite the molecular decomposition, a state near the conduction band remains present, which yields an associated gap of 2.65 eV.

Fig. 6 DOS of BNNT(5,5) on the left and BNNT(10,0) on the right, for pristine and vacancy defect BNNTs with an adsorbed CO₂ molecule.

For both BNNT(10,0)V_B–CO₂ and BNNT(5,5)V_B–CO₂, the vanishing of a state at the end of the valence band occurs, which yields a gap of 2.14 eV. A state between the valence and conduction bands of BNNT(10,0)V_N suffers a small shift of 1.31 eV. Table 2 presents an expected gap for all the analyzed structures. Fig. 7 plots the charge density in order to show the charge concentration in the interaction between BNNTs and the CO₂ molecule. We observe that CO₂ was bound exactly on the vacancy region with those atoms that present dangling bonds.

Fig. 7 Charge density of vacancy defect BNNTs with an adsorbed CO₂ molecule. (a) BNNT(5,5)V_B, (b) BNNT(5,5)V_N, (c) BNNT(10,0)V_B, (d) BNNT(10,0)V_N.

IV. Conclusions

In this work, we performed first-principles calculations to study the interaction of a CO₂ molecule with the external surfaces of pristine and vacancy defect armchair (5,5) and zigzag (10,0) boron nitride nanotubes. We observe that the carbon dioxide molecule interacts weakly with pure BNNTs through van der Waals type interactions. Considering the vacancy defects of BNNTs, we observe that conformational changes in the hexagonal lattice result in changes in the electronic properties that, in turn, result in a greater reactivity with the CO₂ molecule. We observed excellent adsorption energies that point toward a chemisorption state between BNNTs and the CO₂ molecule in most cases. A decomposition reaction of the CO₂ molecule when interacting with BNNT(5,5)V_N was also observed. In this case, as an oxygen atom remains trapped on the lattice, a carbon monoxide molecule is formed. By analyzing the adsorption energies and comparing them with other values present in the literature, we conclude that the methodology presented in this paper provides an efficient route to obtain reliable CO₂ sensors.

Acknowledgements

The authors gratefully acknowledge the financial support from the Brazilian Research Councils CNPq, CAPES, FINATEC and FAPDF.

References

1. A. Rubio, J. L. Corkill and M. L. Cohen, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 49, 5081.
2. N. G. Chopra, R. J. Luyken, K. Cherrey, V. H. Crespi, M. L. Cohen, S. G. Louie and A. Zettl, Science, 1995, 269, 966.
3. E. C. Anota and G. H. Cocoletzi, J. Mol. Model., 2013, 19, 2335.
4. R. Arenal, O. Stephan, M. Kociak, D. Taverna, A. Loiseau and C. Colliex, Phys. Rev. Lett., 2005, 95, 127601.
5. E. Bengu and L. D. Marks, Phys. Rev. Lett., 2001, 86, 2385.
6. A. P. Suryavanshi, M. Yu, J. Wen, C. Tang and Y. Bando, Appl. Phys. Lett., 2004, 84, 2527.
7. C. Zhi, Y. Bando, C. Tang and D. Golberg, Mater. Sci. Eng., R, 2010, 70, 92.
8. J. Cumings and A. Zettl, Solid State Commun., 2004, 129, 661.
9. S. Iijima, Nature, 1991, 354, 56.
10. H. F. Bettinger, T. Dumitrica, G. E. Scuseria and B. I. Yakobson, Phys. Rev. B: Condens. Matter Mater. Phys., 2002, 65, 41406.
11. A. Zobelli, C. P. Ewels, A. Gloter, G. Seifert, O. Stephan, S. Csillag and C. Colliex, Nano Lett., 2006, 6, 1955.
12. J. Li, X. Xiao, X. Xu, J. Lin, Y. Huang, Y. Xue, P. Jin, J. Zou and C. Tang, Sci. Rep., 2013, 3, 3208.
13. P. Piquini, R. J. Baierle, T. M. Schmidt and A. Fazzio, Nanotechnology, 2005, 16, 827.
14. X. M. Li, W. Q. Tian, X. R. Huang, C. C. Sun and L. Jiang, J. Nanopart. Res., 2009, 11, 395.
15. S. A. Montzka, E. J. Dlugokencky and J. H. Butler, Nature, 2011, 476, 43.
16. Y. Jiao, A. Du, Z. Zhu, V. Rudolph, G. Q. Lu and S. C. Smith, Catal. Today, 2011, 175, 271.
17. Accelrys Software Inc., Discovery Studio Modeling Environment, Release 3.5, Accelrys Software Inc., San Diego, 2012.
18. B. Delley, J. Chem. Phys., 1990, 92, 508.
19. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865.
20. J. P. Perdew, J. A. Chevary, S. H. Vosko, K. A. Jackson, M. R. Pederson, D. J. Singh and C. Fiolhais, Phys. Rev. B: Condens. Matter Mater. Phys., 1992, 46, 6671.
21. J.-X. Zhao and Y.-H. Ding, J. Phys. Chem. C, 2008, 112, 20206.
22. Y. Inada and H. Orita, J. Comput. Chem., 2008, 29, 225.
23. M. Weinert and J. W. Davenport, Phys. Rev. B: Condens. Matter Mater. Phys., 1992, 45, 13709.
24. P. Pulay, J. Comput. Chem., 1982, 3, 556.
25. E. N. C. Paura, W. F. Cunha, P. H. O. Neto, G. M. Silva, J. B. L. Martins and R. Gargano, J. Phys. Chem. A, 2013, 117, 2854.
26. F. L. Hirshfeld, Theor. Chem. Acc., 1977, 4, 129.
27. Y. Li, Z. Zhou and J. Zhao, Nanotechnology, 2008, 19, 15202.
28. S. F. Wang, Y. Zhang, J. M. Zhang, K. W. Xu and J. Vicent, Appl. Phys. A, 2012, 109, 601.
29. H. S. Kang, J. Phys. Chem. B, 2006, 110, 4621.
30. H. Choi, Y. C. Park, Y.-H. Kim and Y. S. Lee, J. Am. Chem. Soc., 2011, 133, 2084.
31. K.-J. Lee and S.-J. Kim, Bull. Korean Chem. Soc., 2013, 34(10), 3022.
32. Y. G. Liu and J. Wilcox, Environ. Sci. Technol., 2011, 45, 809.
33. A. B. Silva-Tapia, X. Garcia-Carmona and L. R. Radovic, Carbon, 2012, 50, 1152.

---