The influence of Cu-doping on aluminum nitride, silicon carbide andboron nitride nanotubes’ ability to detect carbon dioxide; DFT study

Zabiollah Mahdavifar n, Nasibeh AbbasiComputational Chemistry Group, Department of Chemistry, Faculty of Science, Shahid Chamran University, Ahvaz, Iran

H I G H L I G H T S

� Cu strongly bound to the outer sur-faces of AlNNT, BNNT and SiCNT.

� Adsorption of CO2 onto the Cu-functionalized AlNNT, BNNT andSiCNT is studied.

� CO2 tends to be strongly physisorbedonto Cu-decorated AlNNT, BNNTand SiCNT.

� A Cu-doped nanotube is a promisingcandidate for monitoring CO2 mole-cules.

� A considerable charge transferreduced the energy gap of the nano-tube/Cu–CO2 system.

G R A P H I C A L A B S T R A C T

The potential usage of Cu-functionalized armchair [4,4] silicon carbide (SiC), aluminum nitride (AlN) andboron nitride (BN) single-walled nanotubes as nanodevices for CO2 monitoring is investigated

.

a r t i c l e i n f o

Article history:Received 26 August 2013Received in revised form13 September 2013Accepted 16 September 2013Available online 21 September 2013

Keywords:Inorganic nanotubesCu-functionalizedCO2 adsorptionDFT

a b s t r a c t

In this research, the potential use of Cu-functionalized [4,4] silicon carbide (SiC), aluminum nitride (AlN)and boron nitride (BN) single-walled nanotubes as nanodevices for CO2 monitoring is investigated. It isfound that Cu-doping the different sites of the considered nanotubes and combining these nanotubeswith CO2 gas molecules are both exothermic processes, and the relaxed geometries are stable. Our resultsreveal that the CO2 gas molecules can be strongly physisorbed on the Cu-doped nanotubes, accompaniedby large adsorption energy. Compared with the weak adsorption of CO2 molecule onto pristine BNNT andSiCNT, the CO2 molecule tends to be strongly physisorbed onto Cu-decorated BNNT and SiCNT with anappreciable adsorption energy. Furthermore, the results indicate that Cu-functionalized SiCNT is morefavorable than Cu-doped BNNT and AlNNT structures for CO2 adsorption. Natural bond orbital analysisindicates that the adsorption of a CO2 molecule onto Cu-doped nanotubes is influenced by the electronicconductance and mechanical properties of the nanotube, which could serve as a signal for a gas sensor. Itappears that the considerable charge transfer from the Cu-doped nanotubes to a CO2 molecule reducesthe energy gap. These observations suggest that the Cu-doped-SiCNT, -BNNT and -AlNNT can beintroduced as promising candidates for gas sensor devices that detect CO2 molecules.

& 2013 Elsevier B.V. All rights reserved.

1. Introduction

Carbon dioxide, CO2, is the focus of much current researchactivity [1,2]. Detection of CO2 molecules is very important in

environmental, biological, and industrial processes [3–5] becauseCO2 molecules are the most readily available renewable, non-toxic,non-flammable, and highly functional carbon resource [6]. On theother hand, CO2 is the main greenhouse gas that contributes toclimate change, which is increasing drastically due to the combus-tion of fossil fuels and chemical processing. The need for CO2

sensors for medical applications, atmospheric concentration con-trol, and indoor climate monitoring [7–9] is one of the most

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/physe

Physica E

1386-9477/$ - see front matter & 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.physe.2013.09.008

n Corresponding author. Tel.: þ98 9163015227; fax: þ98 611 3331042.E-mail addresses: z_mahdavifar@scu.ac.ir, zb_nojini@scu.ac.ir (Z. Mahdavifar).

Physica E 56 (2014) 268–276

exciting challenges and important priorities of the scientificcommunity. However, carbon nanotubes (CNTs) and even pristineboron nitride (BNNT) and silicon carbide (SiCNT) nanotubescannot detect these important molecules because they cannot beeffectively adsorbed on the nanotube surface. Thus, gas sensors forCO2-detection with high sensitivity and fast response times arehighly desirable.

Sensing gas molecules is critical to environmental monitoring,control of chemical processes, space missions, and agricultural andmedical applications [10]. In the context of sensor developmentwork, single-walled carbon nanotubes are potentially ideal materi-als for chemical sensing due to their high aspect ratios, largesurface areas, and unique thermal and electronic properties, as wellas high their sensitivity, fast response and changing electricalproperties at room temperature in the presence of gaseousmolecule [11–13]. Thus far, a few molecular gases, such as NH3

[14], O2 [15], NO2 [16] and SO2 [17], can be sensed by pristine CNT.In recent years, significant experimental and theoretical attemptshave been made to investigate new inorganic nanotubes as sensors.For this purpose, some semiconducting nanotubes, such as alumi-num nitride (AlN), silicon carbide (SiC) and boron nitride (BNNT)nanotubes, as well as metal functionalized versions of thesenanotubes, were tested to improve the sensitivity of gas detecting.

Being mainly semiconductor materials with wide band gaps,BNNTs were first theoretically predicted in 1994 [18] and thenexperimentally synthesized in 1995 [19]. Compared to CNTs, boronnitride nanotubes exhibit different desirable physical and chemicalproperties. For instance, the electronic properties of BNNTs areindependent of tube diameter, helicity, number of tube walls andhigh temperature environments [20]. In addition, they resistoxidation up to 800 1C [21] and show high thermal conductivityand excellent piezoelectricity [22,23]. Because of these unusualmechanical properties and excellent chemical and thermal stabi-lities, BNNTs are promising for potential applications in nanoelec-tronic devices such as sensors and hydrogen storage media [24].

Silicon carbide (SiC) nanotubes, which were first synthesized in2001, are analogous to carbon nanotubes in many respects and exhibitone-dimensional tubular forms [25,26]. The structure and stability ofsingle-walled SiC nanotubes have been investigated in detail by abinitio theory [27–30]. It was found that the SiCNTs are semiconductingmaterials with a large band gap and that their stability is diameter-independent [31] though weakly dependent on helicity [32], unlikecarbon nanotubes. Additionally, SiCNT is of technological interest fordevices operated at high temperatures, at high-frequencies and inharsh environments [33]. An attractive feature is that SiCNTs havehigher reactivity than carbon nanotubes (CNTs) or BNNTs due to theirgreat polarity [34]. For example, theoretical investigations have shownthat a fewmolecules, such as O2 [35], H2 [36] CO and HCN [37] and NOand N2O [38], could be chemisorbed on the exterior surface of SiCNTswith large binding energy. In addition, SiCNTs functionalized withtransition metals (TMs) have shown that the adsorption of anindividual TM on SiCNTs is significantly stronger than that on CNTs[39]. However, Group III nitride semiconductor nanostructures, such asboron nitride nanotubes (BNNTs), have attracted more attentionbecause of their wide band gaps. Aluminum nitride nanotubes(AlNNTs) are an inorganic type of quasi-one-dimensional nanotubes.They are isoelectronic with CNTs and have been successfully synthe-sized by Tondare et al. and other research groups [40–42]. Aluminumnitride nanotubes (AlNNTs), like BNNTs, show unique properties, suchas large band gaps (�6.2 eV), high thermal conductivity, and others[43]. The interaction of AlNNTs with gases, except with some gasmolecules such as H2O, N2 and O2 [44], CH4 [45], and ammonia, [46]has seldom been investigated and remains largely an unexplored area.

In the present study, an attempt has been made to scrutinizethe properties of CO2 adsorption onto Cu decorated AlNNTs, BNNTsand SiCNTs using DFT calculation. Our previous research work [47]

indicated that CO2 gas molecule could be adsorbed onto pristineBN and SiC nanotubes through weak Van der Waals interactions,whereas in the case of AlNNTs, the CO2 molecule tends to bestrongly chemisorbed on the AlNNTs with considerable adsorptionenergy. Hence, this study had two aims: 1) a theoretical investiga-tion of Cu-doping on AlN, SiC and BN nanotubes; and 2) improve-ment of the weak ability of pristine SiC and BN nanotubes toadsorb CO2 molecule using Cu-functionalized AlN, BN and SiCnanotubes. Therefore, this study may provide new insight to CO2

gas sensing and monitoring nanotechnology.

2. Computational details

The adsorption of CO2 gas molecules onto Cu-functionalized[4,4] SiC, [4,4] AlN and [5,5] BN single walled armchair nanotubesis investigated using density functional theory. All calculations arecarried out using the G03 [48] package. The exchange correlationpotential is approximated by the spin-polarized generalized gra-dient approximation (GGA) using the Perdew–Burke–Ernzerhof(PBE–PBE) corrections [49] plus an extra basis set CEP-121G [50]for the Cu atom and the conventional 6–31G basis set for all otheratoms. A one-dimensional periodic boundary condition (PBC) withsix k points sampling in the Brillouin zone is applied along thetube axis to manufacture pristine AlN, BN and SiC nanotubes. Inthis case, the length of these cells is found to be approximately6.41Å (see Fig. 1). The details of the results, such as the geometryparameters and electronic structures of pristine AlN, SiC and BNnanotubes using the same level of theory (PEB-PEB/6-31G), aredescribed in detail in a previous research work by our group [47].One Cu metal atom per unit cell is placed along the tube axisdirection with a vertical distance of approximately 5 Å from thetube walls with two possible adsorption sites: 1) the atomic-adsorption-site (the top of the Al and N atoms in AlNNTs, Si and Catoms in SiCNTs and B and N atoms in BNNTs) and 2) the H-adsorption-site (top of the center of hexagonal ring). The con-sidered absorption sites are shown in Fig. 1.

Based on our calculations, the adsorption energy curves, as wellas the intermolecular nanotube-CO2 distance is also investigatedfor each system (see Fig. 2). The binding energy Ebin is calculatedusing the below expression:

Ebin ¼ Etube=M�ðEtubeþEMÞ ð1Þ

where Etube/M is the total energy of the Cu-decorated AlN, SiC andBN nanotubes; and Etube and EM are the total energies of theisolated nanotubes and Cu metal atom, respectively. It should benoted that the tube/Cu complex with more negative bindingenergy (Ebin) has more stability and could be selected as a tube/metal system with the most energetically favorable configuration.

To examine the adsorption of a CO2 gas molecule onto Cu-doped AlNNTs, SiCNTs and BNNTs, the most energetically favorabletube/Cu systems (the complexes in the minima of adsorptionenergy curves) are fully optimized without any restriction. TheCO2 gas molecule is located on top of the Cu metal atom decoratednanotubes in each unit cell with a lateral distance between gasmolecules of at least 6.41Å (the length of unit cell) to eliminateinteraction between neighboring gas molecules. In addition, tobetter understand the nanotube/Cu–CO2 interactions, the adsorp-tion energy (Eads) of gas onto the metal-decorated nanotubes isdefined as follows:

Eads ¼ Etube=M�gas�ðEtube=MþEgasÞ ð2Þ

where Etube/M-gas denotes the total energy of adduct Cu-decoratedAlNNTs, BNNTs and SiCNTs with the corresponding gas molecule;and Etube/M and Egas are the total energies of the Cu-decoratednanotubes and isolated gas molecule, respectively. According to

Z. Mahdavifar, N. Abbasi / Physica E 56 (2014) 268–276 269

Eq. (2), a negative adsorption energy implies the formation of astable complex, and a positive adsorption energy corresponds to alocal minimum in which the adsorption the gas molecule on to thenanotubes is prevented by a barrier. It is noteworthy that theadsorption energy consists of both interaction (Eint) and deforma-tion (Edef ) energy contributions, which both occur during theadsorption process. Therefore, the following equations are appliedto calculate these contributions:

Eads ¼ Edef þEint ð3Þ

Eint ¼ Etube=M�gas�ðEtube=MincomplexþEgasincomplexÞ ð4Þ

Edef ¼ Edef gasþEdef tube ð5Þ

where Etube/M in complex/Egas in complex denotes the total energy of thetube/Cu and gas in complexes and Edef tub/Me/Edef gas is thedeformation energy of the tube and gas in its relaxed geometry.

The natural bond orbital (NBO) calculations, such as partialcharge transfers, HOMO and LUMO energies and natural atomicorbital occupancies, are investigated. The details of the NBOcalculations are reported in our most recent our research [47].The Wiberg bond index (WBI), which comes from manipulation ofthe density matrix in the orthogonal natural atomic orbital basedon the NBO analysis [51,52], is also considered. The WBI demon-strates the strength of the covalent character (a larger WBI impliesstronger covalent character) and closely relates to the bond order.WBI is expressed as the following mathematical definition:

WBI ¼∑kp2jk ¼ 2pjj�p2jj ð6Þ

where pjk and pjj denote the density matrix elements and chargedensity in the atomic orbital, respectively.

In the past few decades, using density-functional-based theory[53], several attempts have been made to introduce new conceptsthat can be used to understand molecular reactivity. Several ofthese functions, such as chemical potential (μ) and hardness (η),have become more important within the context of densityfunctional theory. Chemical potential and chemical hardness aredefined as the first- and second-order partial derivatives of thetotal electronic energy (E) with respect to the number of electrons(N) at a fixed external potential (υ(r)), respectively [54]. Accordingto Janak's theorem [55] these functions are approximated asfollows:

μ¼ ∂ E∂ N

� �υð r!Þ; T

ffi ðεLþεHÞ2

ð7Þ

η¼ 12

∂ 2E

∂ N2

� �υð r!Þ; T

ffi ðεL�εHÞ2

ð8Þ

where εH and εL denote the energies of the highest occupiedmolecular orbital (HOMO) and the lowest unoccupied molecularorbital (LUMO) from density functional theory calculations,respectively.

Parr et al. [56] have proposed an electrophilicity index (ω) interms of the chemical potential and chemical hardness as ameasure of the energy lowering of a chemical species as a resultof its maximum electron flow between donor and acceptor. Theyhave defined electrophilicity index as follows:

ω¼ μ2

2ηð9Þ

Such an index is intended to be a measure of the energy loweringof the chemical species as a result of its maximum electron flowfrom the environment and is therefore a measure of the capacityof species to accept an arbitrary number of electrons.

B site H site

N site

C site

Si site

H site

Al site H site

N site

Fig. 1. Optimized unit cell of (a) AlNNT, (b) BNNT and (c) SiCNT nanotubes; different possible AlNNT, BNNT and SiCNT sites for CO2 adsorption.

Z. Mahdavifar, N. Abbasi / Physica E 56 (2014) 268–276270

3. Result and discussion

3.1. Cu functionalized AlNNTs, SiCNTs and BNNTs

The geometry optimization of pristine armchair [4,4] SiC, AlNand BN single walled nanotubes is performed in the framework ofdensity functional theory using the PEB-PEB/6-31G functionalmethod. The calculated average Al�N, B�N and Si–C bondlengths are approximately 1.84, 1.46 and 1.83 Å, respectively.Moreover, the obtained energy gaps are nearly 2.41, 4.39 and1.77 eV for AlNN, BNNT and SiCNT, respectively, suggesting semi-conductor character. In addition, our results are compared withprevious research work [45,57], and it is found that the structuralparameters and electronic properties observed here are in agree-ment with those previously reported. The details of the NBOcalculations, including partial charges (esu), bond order andelectronic properties, are summarized in Table S1 in the support-ing information section.

The results obtained in our previous research [47] suggest theneed for some modifications to activate the surface of BNNT andSiCNT for efficient adsorption of CO2 molecules. Therefore, theinfluence of Cu metal atom doping on AlN, BN and SiC nanotubeson the adsorption of CO2 molecules is considered in this section.

First, the adsorption of a Cu metal atom onto the outer surfacesof AlN, BN and SiC single walled nanotubes with various adsorp-tion sites, including the top of the Al and N atoms in AlNT, Si and Catoms in SiCNT and B and N atoms in BNNTs (atomic-adsorption-sites) and the top of the center of hexagon (H-adsorption-site),were studied. To obtain the best position for the Cu atom on theouter surface of the considered nanotubes, the binding energycurves were obtained for all configurations using Eq. (1). Theseresults, depicted in Fig. 2, show the typical potential energysurfaces of the Cu metal atom on the walls of the studiednanotubes. Analysis of these potential energy surfaces indicatesthat the most energetically favorable positions for doping Cu atom

onto the AlN, SiC and BN nanotube surfaces are top of the N-, C-and B-sites, respectively. In other words, when a Cu metal atomapproaches the nitrogen atom in AlNNTs, carbon atom in SiCNTand boron atom in BNNT, the most negative adsorption energiesare obtained.

The final relaxed structures of the nanotube/Cu systems, whichare obtained from optimization of the most energetically favorablenanotube/Cu configurations, are depicted in Fig. 3. Furthermore,the calculated binding energies and intermolecular distancesbetween the Cu metal atom and AlNNT, SiCNT and BNNNT surfacesare listed in Table 1. As shown in Table 1, all of the binding energyvalues are negative, which implies that all relaxed nanotube/Custructures are favorable. The data summarized in Table 1 havesome interesting features. First, for both AlNNTs and SiCNTs, theseresults indicate that the N- and C-sites of AlNNTs and SiCNTs,respectively, are the best sites for the adsorption of a Cu atom ontothe outer surface of these nanotubes. These results may arise fromthe increased electronegativity of the nitrogen or carbon atomrelative to the aluminum or silicon in the AlNNTs or SiCNTs, whichis in good agreement with previous research [39,58]. For example,in the context of AlN nanotubes, the Pauling electronegativity ofthe nitrogen atom is 3.04, which is much larger than that of Al(1.61) and leads to a significant charge transfer from aluminum tonitrogen. Therefore, this structure could become a potential Cu-selective binding material. In addition, the nearest intermoleculardistances are 1.98 Å for the N (1)–Cu bond and 2.09 Å for the C–Cubond in the AlN/Cu and SiC/Cu systems, respectively. In the case ofBNNT, the most stable structure of the nanotube/Cu complexindicates that the position of the Cu atom is changed from theB-adsorption-site to the center of B�N bond with distances of2.26 Å and 2.24 Å from the B and N atoms, respectively (see Fig. 3).Note that there is significant electron density in the center of B�Nbond because of the very slight difference in electronegativitybetween boron and nitrogen. This result agrees well with researchconducted by Wu and Zeng [59].

Fig. 2. Typical potential energy surfaces for CO2 molecule adsorption as function of tube-CO2 distance on (a) the N-site of AlNNT, (b) the B-site of BNNT and (c) the C-site ofSiCNT using the PBEPBE/6-31G method.

Z. Mahdavifar, N. Abbasi / Physica E 56 (2014) 268–276 271

As observed in Table 1, it is found that the interactionbetween a Cu atom and AlNNTs and SiCNTs are very strong. Thebinding energies for Cu adduct onto the AlNNT and SiCNT wallsare approximately �134.37 and �129.73 kJ/mol, respectively,

meaning that a chemisorption process occurred. For BNNT, thebinding energy of the relaxed BNNT/Cu geometry is calculated tobe �37.87 kJ/mol. This result indicates that the interactionsbetween Cu and BNNT are slightly weak and that Cu-doping ontothe BNNT is a physisorption process. Furthermore, the deformationenergy values for the relaxed structures of Cu-doped AlN, BN andSiC nanotubes are 14.28, 9.89 and 16.33 kJ/mol, respectively (seeTable 2S in Supporting information), which indicates that nosignificant curvature in the geometry of these nanotubes occurswhen a Cu atom adsorbs onto the surface of the nanotubes.

Natural bond orbital (NBO) calculations were performed toinvestigate the properties of the nanotube/Cu systems. The partialcharges of atoms derived from NBO calculations are summarizedin Table 2. It is clear from this table that the atomic charges of theCu atoms were increased from zero in free Cu atom to �0.3 esu inCu-decorated AlNNTs and SiCNTs. In addition, the charges of the Al(1) and Si (1) atoms (atom numbering is depicted in Fig. 3) of theAlN and SiC nanotubes are decreased from 1.84 and 1.72 esu to1.71 and 1.53, respectively (compare data listed in Tables 1S and 2).Therefore, the electron population analysis reveals that consider-able charge transfers take place in AlNNTs and SiCNTs when a Cumetal atom is introduced onto the exterior surface of thesenanotubes. In the case of BNNTs, the obtained charge of Cu atomis only 0.14 esu, which indicates that no significant charge transferoccurs during Cu atom doping of the BNNT. This result is inagreement with the low binding energies of this consideredsystem.

Al 1Al2

N1

Cu

Al3B1N1

Cu

B2B3

C1Si1Si2

Cu

Fig. 3. Fully optimized geometry of Cu adsorption onto the (a) AlNNT, (b) BNNT and (c) SiCNT surface.

Table 1Binding energy (Eb (kJ/mol)) and nearest intermolecular distance, r (Å), of AlNNT–Cu, BNNT–Cu and SiCNT–Cu.

Length re (Å) Eb (kJ mol�1)

AlNNT–Cu Al–Cu 2.70 �134.37N1–Cu 1.98

BNNT–Cu B–Cu 2.26 �37.87N–Cu 2.24

SiCNT–Cu C–Cu 2.09 �129.73Si–Cu 2.51

Table 2Calculated NBO partial charges for nanotube–Cu systems using NBO analysis.

Natural charge(esu)

AlNNT–Cu N1 Al1 Al2 Al3 Cu�1.85 1.71 1.71 1.85 0.30

BNNT–Cu B1 N1 Cu – –

0.95 �1.17 0.14 – –

SiCNT–Cu C Si1 Si2 Cu –

�1.78 1.53 1.53 0.29 –

Z. Mahdavifar, N. Abbasi / Physica E 56 (2014) 268–276272

The WBI bond order calculations (see Table 3) indicate that theCu atom strongly bound to the outer surface of the AlNNT (the Cuadsorbed onto the N-site) with a total bond order of approximately0.77. Additionally, the WBI analysis showed that the Cu atombound to the C and Si atoms of SiCNT with 0.72 and 0.41 bondorders, respectively. In the case of the BNNT, the results indicatethat the Cu atom weakly bound to the outer surface of the BNNT.The WBI analyses are presented in Table 3. Overall, the bindingenergy values and the NBO analysis show that the AlNNT/Cu andSiCNT/Cu systems exhibit chemisorption processes that are morefavorable than the those of the BNNT/Cu system, which undergoesa physisorption process.

A comparison between the HOMO/LUMO energy gap for pris-tine (Table 1S) and Cu-doped nanotubes (Table 4) shows that theenergy gap of Cu-doped nanotubes is decreased. Thus, the reac-tivity of functionalized nanotubes is increased, and the systembehaves like a metal system. According to the global reactivityindices of pristine and Cu-doped nanotubes, such as hardness andelectrophilicity, it can be concluded that when the Cu metal atomadsorbed onto the outer surface of the considered nanotubes, thehardness values were slightly decreased and electrophilicityvalues were slightly increased. These changes caused the reactivityof the system to increase. To summarize, the strong adsorptionenergies of Cu-doped nanotubes, the results obtained from theglobal reactivity indices, and the NBO calculations confirm that Cu-doped nanotubes are more reactive than pristine AlNNTs, BNNTsand SiCNTs. Furthermore, it seems that they are more sensitivethan pure titled nanotubes for gas monitoring in nano devices. Totest this sensitivity, the adsorption of CO2 molecule onto Cu-dopedAlNNT, BNNT and SiCNT are studied.

3.2. Adsorption of a CO2 molecule onto Cu-doped AlNNTs, SiCNTs andBNNTs

To investigate CO2 adsorption onto the Cu-doped AlN, SiC andBN nanotubes, the CO2 gas molecule is horizontally positionedabove the Cu metal atom in the relaxed geometry of the nanotube/Cu systems. Based on the results of research performed by Inntamet al. [60], the initial distance between CO2 molecule and Cu atomis chosen to be approximately 2 Å. The new geometries of the

nanotube/Cu–CO2 systems were fully optimized using PEB-PEB/6-31G level of theory in conjugation with the CEP-121G basis set forthe Cu atom, as well as the PBC method. The final optimizedgeometries of the nanotube/Cu–CO2 systems are depicted in Fig. 4.It is obvious from Fig. 4 that while the Cu atom bound to the outersurface of the tubes, the CO2 molecule bound directly to the Cuatom without any dissociation process. The structural parametersof the relaxed nanotube/Cu–CO2 geometries, including the nearestintermolecular distance between Cu atom and nanotubes as wellas the equilibrium distance between the Cu atom and the CO2 gasmolecule, are listed in Table 5. It can be concluded from this tablethat the Cu metal directly bound to carbon atoms with bondlengths of approximately 1.98, 1.98 and 2.03 Å for AlNNTs, BNNTsand SiCNTs, respectively (see Fig. 4). In other words, it appears thatthe CO2 molecule strongly adsorbed onto the Cu-doped nanotubes.

To better understand the adsorption properties of CO2 adsorbedonto Cu decorated nanotubes, the adsorption energies are calcu-lated using Eq. 2, and the obtained data are listed in Table 6.According to Table 6, the adsorption energies of the SiCNT/Cu–CO2

and BNNT/Cu–CO2 structures are approximately �88.04 kJ/moland �64.98 kJ/mol, respectively, which suggests that a strongphysisorption process occurred. In our previous research [47], weshowed that the interaction of CO2 molecules onto pristine SiC andBN nanotubes is very weak and that this interaction is dominatedby weak Van der Waals interactions. In addition, the adsorption ofCO2 onto pristine SiC and BN nanotubes could not considerablychange the mechanical and electrical properties of these nano-tubes. Therefore, it can be concluded from this research that Cu-functionalized SiC and BN nanotubes are more favorable thanpristine SiCNT and BNNT for CO2 adsorption. On the other hand,the adsorption energy gained by CO2 adsorption onto the AlNNT/Cu system is approximately �71.38 kJ/mol. It can be concludedthat Cu-doped AlNNTs, BNNTs and SiCNTs could act as gas sensordevices for detecting CO2 gas molecules because of their appreci-able adsorption energies and the high sensitivity of their surfaces.In our previous research [50], we showed that the adsorption ofCO2 molecules onto pristine AlN nanotubes is a chemisorptionprocess with an adsorption energy of approximately �117.1 kJ/mol. In comparison, the calculated adsorption energies of anAlNNT/CO2 with an AlNNT/Cu–CO2 system (compared with Ref[50]) reveal that pristine aluminum nitride nanotube is morefavorable than Cu-doped AlN nanotubes for detecting CO2 mole-cules. In conclusion, it appears that Cu-functionalized SiC and BNnanotubes significantly improve the ability of titled nanotubes toadsorb CO2 molecules, whereas Cu-functionalized AlNNTs havedecreased sensitivity for CO2 detection.

The mechanism of gas sensing is related to two major factors:changes in the relaxed geometry due to the deformation energy,and changes in the electronic band structure due to the partialelectron charge transfer between the gas molecule and nanotube.Therefore, the deformation energy and natural bond orbital (NBO)analysis of the considered systems are investigated. Because theCO2 molecule adsorbed onto the outer surface of Cu-dopedAlNNTs, SiCNTs and BNNTs, the interaction (Eint) and deformation(Edef) energies resulting from contributions of the adsorptionenergy are calculated using Eqs. (4) and (5). The calculated dataare listed in Table 6. It is clear from Table 6 that considerablecurvature in the structure of the nanotubes occurred when the CO2

molecule adsorbed onto the Cu-doped nanotubes. The obtaineddeformation energy values for the relaxed geometry of nanotube/Cu–CO2 systems are 84.76, 107.34 and 112.8 kJ/mol for SiCNTs,AlNNTs and BNNTs, respectively (see Table 6). In conclusion, Cudecorated AlN, BN and SiC nanotubes could act as sensors for CO2

molecules because significant perturbations are observed in thetheir relaxed geometries during the CO2 adsorption process. Theseperturbations fulfill the mechanism of the sensing condition.

Table 3Wiberg bond indices of the nanotube–Cu systems calculated by PBEPBE/6-31Gmethod (atom numbering corresponds to Fig. 3).

Bond order

AlNNT–Cu N1–Al1 N1–Al2 N1–Al3 N1–Cu Al1–Cu0.40 0.40 0.66 0.77 0.08

BNNT–Cu B1–N1 B1–Cu N1–Cu0.74 0.32 0.26

SiCNT–Cu C–Si1 C–Si2 C–Cu Si1–Cu Si2–Cu0.67 0.66 0.59 0.35 0.34

Table 4Highest occupied molecular orbital (HOMO), lowest unoccupied molecular orbital(LUMO), HOMO–LUMO gap (Δ) energy, electronic chemical potential (m), hardness(η), softness (S) and electrophilicity (ω) of nanotube–Cu systems.

HOMO(eV)

LUMO(eV)

HOMO–LUMOgap (eV)

m(eV)

η

(eV)S(eV�1)

ω

(eV)

Cu �4.74 �0.88 3.86 �2.81 3.85 0.26 1.02AlN–Cu �3.81 �2.86 0.95 �3.34 0.95 1.05 5.86BN–Cu �3.68 �1.56 2.12 �2.62 2.12 0.47 1.62SiC–Cu �3.76 �3.34 0.42 �3.87 0.42 0.79 5.92

Z. Mahdavifar, N. Abbasi / Physica E 56 (2014) 268–276 273

To investigate the electronic properties of nanotube/Cu–CO2

systems, NBO calculations were also performed. The partial atomiccharges of CO2 gas molecules, nanotubes and Cu atoms in nanotube/Cu-gas geometries are summarized in Table 7. It is found fromTable 7 that considerable charge transfers occur from Cu atoms to

CO2 molecules when CO2 molecules are adsorbed onto Cu-dopedAlN, SiC and BN tubes. A consequence of this phenomenon is that theatomic charge of the Cu atoms became more positive in thenanotube/Cu-gas systems than in the tube/Cu systems. For example,during CO2 adsorption onto the BNNT/Cu system, the atomic chargeof the C atom decreased from 0.88 esu in free CO2 gas molecule to0.52 esu in the BNNT/Cu–CO2 system (compare Table 1s and Table 7).Additionally, the atomic charge of the Cu atom increased from 0.14esu in the BNNT/Cu (see Table 2) to 0.53 esu in the BNNT/Cu–CO2,which indicates that there is a considerable charge transfer from theCu metal atom to the carbon atom of the CO2 molecule. This trend ofcharge transfer is also observed for AlNNT and SiCNT nanotubes.Comparison of the partial atomic charge of nanotubes in thenanotube/Cu-gas systems with nanotube/Cu systems reveals thatno significant charge transfers occurred from metal to nanotubeduring CO2 adsorption. As an example, for BNNTs, The there is only aslight charge variation in B(1) and N(1), from 0.95 and �1.17 inBNNT/Cu systems to 0.97 and �1.18 in BNNT/Cu/CO2 systems,respectively. However, the considerable charge transfers from theCu to CO2 molecule is in accordance with strong adsorption energiesof the interaction between CO2 and Cu in the nanotube/Cu-gassystems (see Table 7).

The Wiberg bond index (WBI), which reflects the strength of thebond order character, is also considered. The results of the WBIanalysis are presented in Table 8. Analysis of WBI data shows that theCO2 gas molecule strongly bound to the Cu-doped nanotubescompared to pristine nanotubes. Typically, upon adsorption of aCO2 molecule onto BNNT/Cu, the bond order of the N(1)–Cu bond

Table 5Adsorption energy (Eads, kJ/mol) and nearest intermolecular distance, r (Å), of CO2

adsorption on AlNNT/Cu, BNNT/Cu and SiCNT/Cu systems.

Configuration Length r (Å) Eads (kJ/mol)

AlNNT N1–Cu 1.94 �71.38Cu–C 1.98N1–Cu 2.04

BNNT B1–Cu 2.27 �64.98Cu–C 1.98C1–Cu 1.99

SiCNT Si1–Cu 2.38 �88.04Cu–Cg 2.03

Table 6Interaction energy (Eint), deformation energy (Edeform) and adsorption energy (Eads)of CO2 adsorbed onto Cu-doped nanotubes.

Eint (kJ mol�1) Edeform (kJ mol�1) Eads (kJ mol�1)

AlNNT �178.71 107.34 �71.38BNNT �177.80 112.80 �64.98SiCNT �172.79 84.74 �88.04

O1

O2C

Cu

N1Al1

Al3Al2N2 N4

Cu

N 1B1

B2B3

CO1 O2

N2

C1 Si1C3

C5

Cg

C4

O1O2

Cu

Si2

Si3 Si4

Si5

Fig. 4. Fully optimized geometry of CO2 adsorption onto the (a) AlNNT/Cu, (b) BNNT/Cu and (c) SiCNT/Cu systems.

Z. Mahdavifar, N. Abbasi / Physica E 56 (2014) 268–276274

increased from 0.26 to 0.44 (atom numbering is depicted inFigs. 3 and 4). On the other hand, the WBI bond order of C–O inthe CO2 molecule decreased as follows: the bond orders of C–O(1) and C–O(2) changed from 2 in a pristine CO2 molecule to 1.22 and1.76 in a CO2 molecule adsorbed onto the Cu-AlNNT systems,respectively. Furthermore, it is clear from these results that the CO2

molecule bound to Cu-doped nanotubes with C and O pointing to theCu atom. Based on the NBO analysis and structural parameters, theCu–C bonds are stronger than Cu–O bonds (see Table 8).

Consequently, the electronic properties of the nanotube/Cu-CO2

systems, such as the HOMO/LUMO gap energies and the globalreactivity indices, are considered (see Table 9). It is found that when aCO2 molecule adsorbed onto the Cu-doped nanotubes, the gapenergies of Cu-doped AlN, SiC and BN nanotubes increased. Theseresults suggest that the conductivity and reactivity of these nano-tubes is significantly decreased during the adsorption process. Thecontour plots of the HOMO level upon adsorption of a CO2 moleculeonto the functionalized nanotubes is presented in Fig. 5. According tothe obtained adsorption energy results, global reactivity indices, andNBO calculations, the Cu decorated AlN, BN and SiC single wallednanotubes allow the fabrication of a single-chip CO2 sensor devicebecause of their appreciable adsorption energies, significant chargetransfer, perturbation in the relaxed geometries as well as consider-able changes in the structural parameters and electronic properties ofAlN, SiC and BN nanotubes.

4. Conclusion

The adsorption of a CO2 molecule onto Cu functionalized AlN, SiCand BN nanotubes is investigated using density functional theory. Thegeometrical structures, electronic properties and natural bond orbital

(NBO) are analyzed. Based on our calculations, it is determined thatthe most energetically favorable configurations are obtained when theCu atom is adsorbed into the N-, C- and B-sites of AlNNTs, SiCNTs andBNNTs, respectively. Furthermore, an analysis of the relaxed nanotube/Cu geometries indicates that the Cu atom strongly bonded into theouter surface of AlN and SiC nanotubes with appreciable bindingenergies through a chemisorption process. The adsorption energy andNBO analysis results indicate that CO2 molecules tend to undergostrong physisorption onto the Cu-doped AlN, BN and SiC nanotubes. Incomparison with our previous research, it is determined that the Cufunctionalized SiC and BN nanotubes exhibit considerably improvedability to monitor CO2. On the other hand, the sensitivity of AlNNT for

Table 7Calculated NBO partial charges for CO2 adsorption onto nanotube/Cu systems.

Partial charges (esu)

AlNNT N1 Al1 Al2 Al3 Cu C O1 O2

�1.83 1.72 1.82 1.86 0.44 0.55 �0.52 �0.45BNNT N1 B1 Cu C O1 O2 – –

�1.18 0.97 0.53 0.52 �0.54 �0.46 – –

SiCNT C1 Si1 Cu Cg O1 O2 – –

�1.80 1.50 0.50 0.59 �0.59 �0.52 – –

Table 8Wiberg bond indices of the nanotube–Cu systems calculated by the PBEPBE/6-31Gmethod (atom numbering corresponds to Fig. 4).

Bond order

AlNNT N1–Al1 N1–Al2 N1–Al3 N1–Cu Cu–C Cu–O C–O1 C–O2

0.30 0.48 0.61 0.82 0.53 0.42 1.21 1.75BNNT N1–B1 N1–Cu B1–Cu Cu–C Cu–O1 C–O1 C–O2 –

0.61 0.44 0.23 0.49 0.40 1.22 1.76 –

SiCNT C1–Si1 C1–Cu Si1–Cu Cu–Cg Cu–O1 Cg–O1 Cg–O2 –

0.53 0.72 0.41 0.41 0.45 1.19 1.78 –

Table 9Highest occupied molecular orbital (HOMO), lowest unoccupied molecular orbital(LUMO), HOMO–LUMO gap (Δ) energy, electronic chemical potential (m), hardness(η), softness (S) and electrophilicity (ω) of nanotube–Cu–CO2 systems.

Configuration HOMO(eV)

LUMO(eV)

HOMO–LUMOgap(eV)

m(eV)

η

(eV)S(eV�1)

ω

(eV)

AlNNT �4.19 �3.03 1.16 �3.61 1.16 0.86 5.61BNNT �4.05 �1.66 2.39 �2.85 2.39 0.42 1.70SiCNT �4.32 �3.14 1.18 �3.73 1.18 0.84 5.87

Fig. 5. Typical contour plots of the HOMO of the CO2 adsorption on (a) AlNNT/Cu,(b) BNNT/Cu and (c) SiCNT/CU systems.

Z. Mahdavifar, N. Abbasi / Physica E 56 (2014) 268–276 275

CO2 detection is decreased when Cu atom is doped onto the outersurface of this AlNNT. Overall, we demonstrate that Cu-dopedAlNNTs, SiCNTs, and BNNTs, as well as pristine AlNNTs, are promisingcandidates for sensing CO2 molecules.

Appendix A. Supplementary material

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.physe.2013.09.008.

References

[1] J. Zhao, Y. Ding, Journal of Chemical Theory and Computation. 5 (2009) 1099.[2] Y. Jiao, A. Du, Z. Zhu, V. Rudolph, S.C. Smith, Journal of Materials Chemistry 20

(2010) 10426.[3] B. Muller, P.C. Hauser, Analyst 121 (1996) 339.[4] L.C. Brousseau, D.J. Aurentz, A.J. Benesi, T.E Mallouk, Analytical Chemistry 69

(1997) 688.[5] P. Rowinski, R. Bilewicz, M.J. Stebe, E. Rogalska, Analytical Chemistry 74 (2002)

1554.[6] T. Sakakura, J.C. Choi, H. Yasuda, Chemical Reviews 107 (2007) 2365.[7] A. Chakradhar, U. Burghaus, Surface Science 616 (2013) 171.[8] D. Diamond, S. Coyle, S. Scarmagnani, J. Hayes, Chemical Reviews 108 (2008)

652.[9] P.C. Jurs, G.A. Bakken, H.E. McClelland, Chemical Reviews 100 (2000) 2649.[10] R. Joseph., S.J. Li, Chemical Reviews 108 (2008) 352.[11] J.W. Mintmire, B.I. Dunlap, C.T. White, Physical Review Letters 68 (1992) 631.[12] N. Hamada, S.I. Sawada, A. Oshiyama, Physical Review Letters 68 (1992) 1579.[13] Z.B. Nojini, S. Samiee, Journal of Physical Chemistry C 115 (2011) 12054.[14] J. Kong, N. Franklin, C. Zhou, M. Chapline, S. Peng, K. Cho, H. Dai, Science 287

(2000) 622.[15] P.G. Collins, K. Bradley, M. Ishigami, A. Zettl, Science 287 (2000) 1801.[16] J. Li, Y.J. Lu, Q. Ye, M. Cinke, J. Han, M. Meyyappan, Nano Letters 3 (2003) 929.[17] A. Goldoni, R. Larciprete, L. Petaccia, S. Lizzit, Journal of the American Chemical

Society 125 (2003) 11329.[18] A. Rubio, J.L. Corkill, M.L. Cohen, Physical Review B 49 (1994) 5081.[19] N.G. Chopra, R.J. Luyken, K. Cherrey, V.H. Crespi, M.L. Cohen, S.G. Louie,

A. Zettl, Science 269 (1995) 966.[20] C. Zhi, Y. Bando, C. Tang, D. Golberg, Materials Science and Engineering

Reports 70 (2010) 92.[21] Y. Chen, J. Zou, S.J. Campbell, G.L. Caer, Applied Physics Letters 84 (2004) 2430.[22] J. Wang, V.K. Kayastha, Y.K. Yap, Z. Fan, J.G. Lu, Z. Pan, I.N. Ivanov, A.A. Puret-

zky, D.B. Geohegan, Nano Letters 5 (2005) 2528.[23] S.M. Nakhmanson, A. Calzolari, V. Meunier, J. Bernholc, M.B. Nardelli, Physical

Review B 67 (2003) 235406.[24] S.H. Jhi, Y.K. Kwon, Physical Review B 69 (2004) 245407.[25] X.H. Sun, C.P. Li, W.K. Wong, N.B. Wong, C.S. Lee, S.T. Lee, B.K. Teo, Journal of

the American Chemical Society 124 (2002) 14464.[26] H.C. Pham, N. Keller, G. Ehret, M.J. Ledouxi, Journal of Catalysis 200 (2001)

400.

[27] Y. Miyamoto, B.D. Yu, Applied Physics Letters 80 (2002) 586.[28] M. Menon, E. Richter, A. Mavrandonakis, G. Froudakis, A.N. Andriotis, Physical

Review B 69 (2004) 115332.[29] S. Jiuxu, Y. Yintang, L. Hongxia, G. Lixin, Z. Zhiyong, Journal of Semiconductors

31 (2010) 114003.[30] A. Mavrandonakis, G. Froudakis, A.N. Andriotis, Nano Letters 3 (2003) 1481.[31] Y.F. Li, Z. Zhou, Y.S. Chen, Z.F. Chen, Journal of Chemical Physics 130 (2009)

204706.[32] M.W. Zhao, Y. Xia, F. Li, R.Q. Zhao, S.T. Lee, Physical Review B 71 (2005) 085312.[33] A. Fissel, B. Schroter, W. Richer, Applied Physics Letters 66 (1995) 3182.[34] R.Q Wu, M. Yang, Y.H. Lu, Y.P. Feng, Z.G. Huang, Q.Y. Wu, Journal of Physical

Chemistry C 112 (2008) 15985.[35] F. Cao, X. Xu, W. Ren, C. Zhao, Journal of Physical Chemistry C 114 (2010) 70.[36] G. Mopurmpakis, G.E. Froudakis, G.P. Lithoxoos, J. Samios, Nano Letters 6

(2006) 1581.[37] R.Q. Wu, M. Yang, H.Y. Lu, Y.P. Feng, Z.G. Huang, Q.Y. Wu, Journal of Physical

Chemistry C 112 (2008) 15985.[38] G.H. Gao, H.S. Kang, Journal of Chemical Theory and Computation 4 (2008)

1690.[39] J.X. Zhao, Y.H. Ding, Journal of Physical Chemistry C 112 (2008) 2558.[40] V.N. Tondare, C. Balasubramanian, S.V. Shende, D.S. Joag, V.P. Godbole, S.

V Bhoraskar, M. Bhadbhade, Applied Physics Letters 80 (2002) 4813.[41] C. Balasubramanian, S. Bellucci, P. Castrucci, M. De Crescenzi, S.V. Bhoraskar,

Chemical Physics Letters 383 (2004) 188.[42] G. Stan, C. Ciobanu, T. Thayer, G. Wang, J. Creighton, K. Purushotham,

L. Bender-sky, R. Cook, Nanotechnology 20 (2009) 35706.[43] W.G. Liu, G.H. Chen, X.C. Huang, D. Wu, Y.P. Yu, Journal of Physical Chemistry C

116 (2012) 4957.[44] K. Rezouali, M.A. Belkhir, A. Houari, J. Bai, Computational Materials Science 45

(2009) 305.[45] Z. Mahdavifar, M. Haghbayan, Applied Surface Science 263 (2012) 553.[46] Z. Zhou, J. Zhao, Y. Chen, P.R. Schleyer, Z. Chen, Nanotechnology 18 (2007)

424023.[47] Z. Mahdavifar, N. Abbasi, E. Shakerzadeh, Sensors and Actuators B 185 (2013)

512.[48] M.J. Frisch, et al., Gaussian03, Revision B, Pittsburgh, PA, 2003.[49] J.P. Perdew, K. Burke, M. Ernzerhof, Physical Review B 77 (1996) 3865.[50] W. Stevens, H. Basch, J. Krauss, Journal of Chemical Physics 81 (1984) 6026.[51] A.W. Ehlers, E.J. Baerends, K. Lammertsma, Journal of the American Chemical

Society 124 (2002) 2831.[52] K.K. Pandey, G. Frenking, J. Euro, Inorganic Chemistry 2004 (2004) 4388.[53] P. Geerlings, F. De Proft, W. Langenaeker, Chemical Reviews 103 (2003) 1793.[54] R.G. Parr, R.A. Donelly, M. Levy, W.E. Palke, Journal of Chemical Physics 68

(1978) 3801.[55] J.F. Janak, Physical Review B 18 (1978) 7165.[56] R.G. Parr, L. Szentpaly, S. Liu, Journal of the American Chemical Society 121

(1999) 1922.[57] M. Zhao, Y. Xia, X. Liu, Z. Tan, B. Huang, C. Song, L. Mei, Journal of Physical

Chemistry B 110 (2006) 8764.[58] J.M. Zhang, H.H. Li, Y. Zhang, K.W. Xu, Physica E 43 (2011) 1249.[59] X. Wu, X.C. Zeng, Journal of Chemical Physics125 (2006) 044711.[60] C. Inntam, J. Limtrakul, Journal of Physical Chemistry C 114 (2010) 21327.

Z. Mahdavifar, N. Abbasi / Physica E 56 (2014) 268–276276