Thin Solid Films 534 (2013) 650–654

Contents lists available at SciVerse ScienceDirect

Thin Solid Films

j ourna l homepage: www.e lsev ie r .com/ locate / ts f

Ammonia monitoring by carbon nitride nanotubes: A density functional study

Javad Beheshtian a, Maziar Noei b, Hammed Soleymanabadi c, Ali Ahmadi Peyghan c,⁎a Department of Chemistry, Shahid Rajaee Teacher Training University, P.O. Box: 16875-163, Tehran, Iranb Department of Chemistry, Mahshahr Branch, Islamic Azad University, Mahshahr, Iranc Young Researchers and Elite Club, Central Tehran Branch, Islamic Azad University, Tehran, Iran

⁎ Corresponding author. Tel.: +98 912 5061827.E-mail address: ahmadi.iau@gmail.com (A.A. Peygha

0040-6090/$ – see front matter © 2013 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.tsf.2013.03.033

a b s t r a c t

a r t i c l e i n f o

Article history:Received 17 August 2012Received in revised form 14 March 2013Accepted 15 March 2013Available online 27 March 2013

Keywords:Carbon nitrideNanotubesDensity functional theoryElectronic propertiesDensity of statesBand gap

The adsorption of one to three NH3 molecule(s) on the exterior surface of a carbon nitride nanotube (CNNT)has been investigated using density functional theory. It has been found that a NH3 molecule is physicallyadsorbed on the tube surface with adsorption energies in the range of 15.47 to 25.08 kJ/mol. The most stableobtained configuration is that in which the NH3 molecule has been adsorbed above center of a porous site ofthe tube surface via forming hydrogen bonds. The adsorption of the second and third NH3 molecules at theporous sites is independent of each other. Density of states analysis reveals that the intrinsic semiconductorCNNT may transform to an extrinsic n-type semiconductor with higher electrical conductance upon the ad-sorption of NH3 molecules. The obtained results suggest that the CNNTs may be potentially used in NH3 sen-sor applications.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Detection of toxic gaseous molecules is of critical importance in in-dustrial, environmental, and medical issues. An adsorbed moleculecan change the electronic and transport properties of the semiconduc-tors, resulting in a detectable signal related to the concentration of thetarget gas. Nanostructured materials for gas sensing have been ofgreat interest due to their unique and interesting properties includinghigh surface-to-volume ratio and sensitive electronic structures [1–6].Since the discovery of the carbon nanotubes (CNTs) [7], extensive stud-ies have been focused on their application in sensors [8]. SynthesizedCNTs may show metallic or semiconductive properties depending ontheir radius and chirality, which limits their application in gas sensors.Therefore, several studied have been focused on the other types ofnanotubes as gas sensors including BN, AlN, SiC, etc. [9–11]. For exam-ple, we have recently shown that AlN nanotubes can be used as poten-tial efficient gas sensors for formaldehyde detection [12]. This is whilethat they are not able to detect CO and H2O molecules [12,13]. Wehave also demonstrated that MgO nanotubes selectively act againstCO and NO gaseous molecules [14].

Carbon nitride has attracted increasing interest due to its interestingelectronic, chemical, andmechanical properties aswell as large numberofmethodswhich are available to synthesize carbon nitride compounds[15,16]. Theoretical studies have predicted that carbon nitridenanotubes (CNNTs) with CN or C3N4 stoichiometry may exist [17]. Caoet al. have synthesized porous CNNTs with the CN stoichiometry

n).

rights reserved.

through the reaction of cyanuric chloride (C3N3Cl3) with sodium at230 °C using NiCl2 as catalyst [18]. Recently, Koh et al. [19] have useddensity functional theory (DFT) to show that CNNTs could be attractivehydrogen sorbent materials.

It is known that the NH3 is a low boiling point compound and vol-atile. Therefore, it is very important to develop sensitive sensors todetect the gaseous NH3 molecules. Hence, we have been interestedin finding whether there is a potential possibility for pristine CNNTsserving as a chemical sensor for NH3 molecule. To this end, we havestudied the adsorption of one to three NH3 molecule(s) on the exteri-or surface of a single-walled CNNT comprising of triazine units(which have been synthesized by Cao et al. [18]) using DFT calcula-tions. Particular attention has been paid to understand the changesof electronic properties of the tube by ammonia molecule adsorption.

2. Computational methods

We have considered adsorption of NH3 molecule on the surface of aCNNT which has been rolled up from corresponding graphitic-CN. Thestructure of the graphitic-CN is constructed by C and N atoms, and theC\N and C\C bonds form a porous structure as shown in Fig. 1. LikeCNTs, CNNTs can be zigzag or armchair depending on how thegraphitic-CN is rolled up, and they are labeled by the chiral indexes(m, n) according to the notation given by White et al. [20]. Herein, wehave selected a (9, 0) zigzag CNNT consisting of 72 C and 72 N atoms,in which the end atoms have been saturated by hydrogen atoms toavoid the boundary effects. The DFT through B3LYP/6-31G* level of the-ory has been carried out in GAMESS suit of program codes [21], for all ofthe calculations. It should be noted that the B3LYP is a commonly used

Fig. 1. Top and side views of an optimized (9, 0) zigzag carbon nitride nanotube (CNNT) and its density of states (DOS) plot.

651J. Beheshtian et al. / Thin Solid Films 534 (2013) 650–654

functional in the investigation of nanostructured materials [22–25].Studying the chemisorption of ammonia on BN nanotubes, we have re-cently shown that 6-31G* basis set is accurate enough comparing withthe larger ones such as 6-31G** and 6-31++G** [26]. We have definedadsorption energy in the usual way as:

Ead ¼ E CNNTð Þ þ E NH3ð Þ–E CNNT–NH3ð Þ–EBSSE ð1Þ

where E (CNNT) is the energy of the isolated tube, E (NH3) is the energyof a single NH3molecule, E (CNNT − NH3) corresponds to the energy ofthe CNNT in which the NH3 has been adsorbed on the outer wall, andEBSSE is the energy of the basis set superposition error.

3. Results and discussion

3.1. NH3 adsorption on the CNNT

Fig. 1 shows top and side views of the optimized structure of the(9, 0) zigzag CNNT. Also, a picture of the planar sheet, where thetube originates from it, was provided. The CNNT converges tounsmooth tubular surface due to the N\N lone pair repulsions andthe porous structure. Calculated length and diameter of the optimizedCNNT are approximately 23.34 and 7.35 Å, respectively. There arethree types of C\N bonds and two types of C\C bonds in the (9, 0)CNNT. The parallel bonds to the tube axis have been denoted as ver-tical (V) bonds and the other bonds as diagonal (D) ones. The DC\N

bonds can be in two different positions as shown in Fig. 1, includingthe bonds near to DC\C and VC\C with equilibrium lengths of 1.33and 1.34 Å, respectively. The bond length of both DC\C and VC\C is

about 1.52 Å. The Mulliken population charge analysis shows a netcharge of +0.41 and −0.41 e on the C and N atoms, respectively, in-dicating partially ionic nature.

In order to find energetically favorable adsorption configurations(local minima), NH3 molecule was initially placed at different posi-tions on the tube surface with different orientations including indi-vidual N or H atom being close to the surface. For example, we haveput the NH3 molecule above the center of a hexagonal ring so thatthe N or three H atoms of NH3 were close to the ring. In this case,after full relax optimization, we have only obtained one stable config-uration (C, Fig. 2) in which the NH3 is located above the center ofhexagon in such a way that the distances of its N atom from three Catoms of the hexagon are about 3.35, 3.36, and 3.42 Å. Calculated ad-sorption energy for this configuration is about 15.47 kJ/mol, indicat-ing physical nature of the process and a charge about 0.11 e istransferred from the NH3 to the tube.

We have also initially put the NH3 molecule above C\N and C\Cbonds and then performed relax optimization. However, two stableconfigurations were obtained when the NH3 molecule was located onthe VC\C or DC\C bonds from its N head as shown in Fig. 2 (configura-tion B). It should be noted that as all properties of the obtained config-urations for both C\C bonds are nearly alike, only one of themhas beenreported here. In this configuration, the distances of the N atom in NH3

and C atoms of C\C bonds are about 2.97 Å and calculated adsorptionenergy is about 17.81 kJ/mol. The large distances and small adsorptionenergy along with a small charge transfer (Table 1) of NH3 to the tubeindicate that this process is a weak physisorption in nature. However,the interaction of NH3 with the tube through configuration B is some-what stronger than that through the C. It can be explained by the lone

Fig. 2. Partial structure of NH3 adsorbed carbon nitride nanotubes (CNNTs) and their density of states (DOS) plots.

Table 1Adsorption energy per NH3 molecule (Ead, kJ/mol) on the CNNT, average charge on theone adsorbed NH3, HOMO and LUMO energies, and HOMO/LUMO energy gap (Eg) ofthe studied systems (in eV).

System Ead QT (e) EHOMO ELUMO Eg aΔEg(%)

CNNT – – −7.49 −3.47 4.02 –

A 25.08 −0.27 −6.11 −3.52 2.59 35.6B 17.81 0.13 −7.36 −3.37 3.07 0.7C 15.47 0.11 −7.31 −3.37 3.19 2.02NH3 25.12 −0.27 −6.13 −3.58 2.55 36.63NH3 25.16 −0.28 −6.16 −3.63 2.53 37.1

a The change of Eg of CNNT upon the NH3 adsorption.

652 J. Beheshtian et al. / Thin Solid Films 534 (2013) 650–654

pair repulsion forces between the N atom of NH3 and N atoms of hexag-onal ring in the configuration C which result in a smaller adsorptionenergy.

The other initial configuration is that in which we have put theNH3 molecule above the center of a porous site of the tube surfaceso that N or three H atoms of NH3 are close to the surface. As shownin Fig. 1 the edge of a porous site is constructed by six N atoms, there-fore, it is expected that these atoms should repel the NH3 molecule ifit gets close to the porous site from its N head. Our calculations con-firm this phenomenon so that only one stable NH3/CNNT complex(configuration A, Fig. 2) has been obtained in which the hydrogenatoms of NH3 are close to the surface. Calculated adsorption energy

653J. Beheshtian et al. / Thin Solid Films 534 (2013) 650–654

(25.08 kJ/mol) shows that this configuration is the most stableamong all of the considered configurations. The result is readily un-derstood by the fact that in this configuration a number of hydrogenbonds are formed between the H atoms of NH3 and N atoms of the po-rous site with bond distances in the range of 2.68 to 2.97 Å. However,it is well known that the hydrogen bond is stronger than a van derWaals interaction. A charge of 0.27 e is also transferred from electronrich N atoms of the porous site to the NH3 molecule.

Subsequently, selecting configuration A as the most stable complexwe have investigated the co-adsorption of two and threeNH3moleculeson the porous sites in themiddle area of the tube. Optimized structuresof 2NH3– and 3NH3–CNNT complexes are shown in Fig. 3 and their cor-responding data are listed in Table 1. The results of adsorption energy,charge transfer, and structural analysis show that the NH3 adsorptionsin these configurations are independent of each other i.e. the firstadsorbed NH3 does not influence the adsorption of the second oneand so on. The adsorption energies in the case of co-adsorption of twoand three NH3 molecules are about 25.12 and 25.16 kJ/mol per mole-cule, respectively which fairly are equal to that of one NH3 adsorption(25.08 kJ/mol, configuration A). This trend can be observed for chargetransfer so that the average charge transfer from the tube to a NH3mol-ecule in both cases of 2NH3 and 3NH3 co-adsorptions is about 0.27 ewhich is as equal as that of 1NH3 adsorption.

3.2. Electronic properties

In order to consider the sensitivity of the CNNT to the NH3molecule,density of states (DOS) plots of the pristine CNNT and different NH3/CNNT complexes were drawn. The calculated DOS plot in Fig. 1 shows

Fig. 3. Optimized structures of complexes of two and three NH3 molecules co-adso

that the CNNT is a semiconductor with HOMO (the highest occupiedmolecular orbital)/LUMO (the lowest unoccupied molecular orbital)gap (Eg) of 4.02 eV. Considering the DOS plots of configurations B andC demonstrates that the NH3 adsorption process cannot change theelectronic properties of the CNNT and therefore its Eg is not significantlychanged. While Fig. 3 shows that compared to the DOS of pristine tube,the DOS of configuration A shows a new peak appearing just beneaththe Fermi level at the energy level of −6.11 eV. Partial DOS analysisdemonstrates that this state has a dominant N 2p orbital characterwhich mainly comes from NH3 molecule. The contribution of NH3 andthe tube in this state was calculated to be about 79 and 21%, respective-ly, based on the partial DOS analysis. The peak indicates that after theadsorption of NH3 molecule, the system becomes an n-type semicon-ductor (with smaller Eg), and thus, a significant increase in electricalconductivity is expected, compared to the pristine tube. It is noteworthyto mention that the Eg (or band gap in bulk materials) is a major factordetermining the electrical conductivity of a material [27]. Generally,smaller Eg at a given temperature leads to larger electrical conductivity.

However, in configuration A, the Eg decreases from 4.02 eV in thefree CNNT to 2.59 eV upon the adsorption of one NH3 molecule. Theconsiderable change about 1.43 eV in the Eg value demonstrates thehigh sensitivity of the electronic properties of CNNTs to the NH3 mol-ecule. It suggests that CNNT can transform the presence of NH3 mol-ecule into an electrical signal, and therefore it may be potentiallyused in NH3 sensors. However, the conventional sensing mechanismis such that upon exposure to the target gas molecules charge transferbetween the adsorbed gas species and the surface of semiconductoroccurs. This charge transfer will either increase or decrease the con-centration (or mobility) of the major carriers in the semiconductor,

rbed on the carbon nitride nanotubes and their density of states (DOS) plots.

654 J. Beheshtian et al. / Thin Solid Films 534 (2013) 650–654

resulting in an increase or decrease of the electrical conductivity ofthe sensor under operating conditions.

We think that the proposed sensor may benefit from some advan-tages including high sensitivity: the Eg of CNNT is appreciably sensitivetowards the presence of NH3 so that it decreases by about 35.6% uponthe adsorption of NH3 molecule, pristine application: this nanotubecan detect the NH3 molecule in its pristine type without manipulatingits structure through doping, chemical functionalization, making de-fect, etc., and short recovery time: the adsorption energy of NH3 mol-ecule is not so large to hinder the recovery of CNNTs and thereforethe sensor will possess short recovery times since based on the con-ventional transition state theory, the recovery time can be expressedas:

τ ¼ ν−10 exp Ead=kTð Þ

where T is the temperature, k is the Boltzmann's constant, and ν0 isthe attempt frequency. According to this equation, great Ead valueswill prolong the recovery time in an exponential manner.

Calculated DOS plots of 2NH3– and 3NH3–CNNT complexes con-firm that the adsorptions of NH3 molecules on the porous sites ofCNNT are independent of each other. As shown in Fig. 3, only the in-tensity of newly appeared impurity state is increased by increasingthe number of adsorbed NH3 molecule and only a slight change ofEg can be observed. Calculated Eg values for the complexes of one,two, and three NH3 adsorbed on the CNNT are about 2.59, 2.55, and2.53 eV, respectively, indicating that the Eg of the complexes is insig-nificantly decreased by increasing the number of the adsorbed NH3

molecules.Next, we investigated the possibility effect of the tube length on

the Ead and electronic properties of the nanotube. To this end, aCNNT was selected with optimized length of about 40.2 Å, being16.9 Å longer than the former studied tube. This tube constituted of126 C and 126 N atoms, in which the end atoms saturated with hy-drogen atoms. The results indicate that HOMO (−7.50 eV), LUMO(−5.53 eV), and Eg (3.97 eV) of this tube are nearly equal to thoseof the shorter one. Adsorption energy of a NH3 on the longer tubewithin configuration A (about 25.21 kJ/mol) and the change of Eg(about −35.1%) were calculated to be somewhat as same as thoseon the former tube. Thus, it can be concluded that the effects of in-creasing the length of the tube may be negligible on the adsorptionprocess and electronic properties of the tube.

We have repeated all of the calculations using TPSS andB3PW91DFTfunctionals on the CNNT and the most stable complex (configuration A,Fig. 2). Xiao et al. [28] have shown that for several studied semiconduc-tors, B3PW91 predicts gaps substantially better than all modern hybridfunctionals. Our results (Table 2) indicate that the adsorption energyvalues depend on the kind of used functional so that the values obtainedusing TPSS (functional of Tao, Perdew, Staroverov, and Scuseria) andB3PW91 are somewhat smaller than the values of B3LYP. The Egamounts of TPSS are significantly smaller than those of B3LYP whilethose of B3PW91 slightly differ. However, the results of both methodsindicate that the Eg of CNNT significantly reduces upon the adsorption

Table 2Adsorption energy of NH3 on the CNNT (Ead, kJ/mol), average charge on the adsorbedNH3, HOMO and LUMO energies, and HOMO/LUMO energy gap (Eg) of the configura-tion A (Fig. 2, in eV).

Functional System Ead QT (e) EHOMO ELUMO Eg aΔEg (%)

B3PW91 CNNT – – −7.59 −3.62 3.97 –

A 20.98 0.17 −6.18 −3.67 2.51 −21.5TPSS CNNT – – −6.98 −3.78 3.20 –

A 18.22 0.12 −5.59 −3.83 1.76 −45.0

a The change of Eg of CNNT upon the NH3 adsorption.

process within configuration A which is in agreement with the B3LYPresults.

4. Conclusions

The adsorption of one to three NH3 molecule(s) on the exteriorsurface of CNNT was investigated using DFT calculations. It wasfound that the NH3 molecule prefers to be physically adsorbed at po-rous site of the CNNT through forming hydrogen bonds releasing en-ergy of about 25.08 kJ/mol. The results show that the adsorptions oftwo and three NH3 molecules at porous sites are independent ofeach other. DOS reveals that upon the adsorption of NH3 moleculesat the porous sits of the CNNT, the intrinsic semiconductor tubemay be transformed to an extrinsic n-type semiconductor with higherelectrical conductance. Thus, CNNTs may be a promising candidate fordetection of toxic NH3 gaseous molecules and may cover some advan-tageous including short recovery time, high sensitivity, and also pris-tine usability.

References

[1] N. Kılınç, E. Şennik, Z.Z. Öztürk, Thin Solid Films 520 (2011) 953.[2] J. Beheshtian, M. Kamfiroozi, Z. Bagheri, A.A. Peyghan, Chin. J. Chem. Phys. 25

(2012) 60.[3] A.L. Verma, S. Saxena, G.S.S. Saini, V. Gaur, V.K. Jain, Thin Solid Films 519 (2011)

8144.[4] A. Ahmadi Peyghan, A. Omidvar, N.L. Hadipour, Z. Bagheri, M. Kamfiroozi, Physica

E 44 (2012) 1357.[5] M. Penza, F. Antolini, M. Vittori-Antisari, Thin Solid Films 472 (2005) 246.[6] Y.-f. Hu, Z.-h. Zhang, H.-b. Zhang, L.-j. Luo, S.-z. Yao, Thin Solid Films 520 (2012)

5314.[7] S. Iijima, Nature 354 (1991) 56.[8] D.W.H. Fam, A. Palaniappan, A.I.Y. Tok, B. Liedberg, S.M. Moochhala, Sens. Actua-

tors B 157 (2011) 1.[9] G. Gao, S.H. Park, H.S. Kang, Chem. Phys. 355 (2009) 50.

[10] J. Beheshtian, M.T. Baei, Z. Bagheri, A.A. Peyghan, Microelectron. J. 43 (2012) 452.[11] J. Beheshtian, H. Soleymanabadi, M. Kamfiroozi, A. Ahmadi, J. Mol. Model. 18

(2012) 2343.[12] A. Ahmadi, N.L. Hadipour, M. Kamfiroozi, Z. Bagheri, Sens. Actuators B 161 (2012)

1025.[13] J. Beheshtian, Z. Bagheri, M. Kamfiroozi, A. Ahmadi, Struct. Chem. 23 (2012) 653.[14] J. Beheshtian, M. Kamfiroozi, Z. Bagheri, A. Ahmadi, Physica E 44 (2011) 546.[15] A. Liu, M. Cohen, Science 245 (1989) 841.[16] D. Teter, R.J. Hemley, Science 271 (1996) 53.[17] Y. Miyamoto, M. Cohen, S. Louie, Solid State Commun. 102 (1997) 605.[18] C. Cao, F. Huang, C. Cao, J. Li, H. Zhm, Chem. Mater. 16 (2004) 5213.[19] G. Koh, Y.-W. Zhang, H. Pan, Int. J. Hydrogen Energy 37 (2012) 4170.[20] C. White, D. Robertson, J.W. Mintmire, Phys. Rev. B 47 (1993) 5485.[21] M.W. Schmidt, K.K. Baldridge, J.A. Boatz, S.T. Elbert, M.S. Gordon, J.H. Jensen, S.

Koseki, N. Matsunaga, K.A. Nguyen, S. Su, T.L. Windus, M. Dupuis, J.A. Mont-gomery, J. Comput. Chem. 14 (1993) 1347.

[22] L. Chen, C. Xu, X.-F. Zhang, T. Zhou, Physica E 41 (2009) 852.[23] J. Beheshtian, M. Kamfiroozi, Z. Bagheri, A. Ahmadi, Comput. Mater. Sci. 54 (2012)

115.[24] F. Trani, M. Causà, S. Lettieri, A. Setaro, D. Ninno, V. Barone, P. Maddalena,

Microelectron. J. 40 (2009) 236.[25] T.C. Dinadayalane, J.S. Murray, M.C. Concha, P. Politzer, J. Leszczynski, J. Chem.

Theory Comput. 6 (2010) 1351.[26] A. Ahmadi, J. Beheshtian, N.L. Hadipour, Struct. Chem. 22 (2011) 183.[27] S.S. Li, Semiconductor Physical Electronics, Second ed. Springer, USA, 2006.[28] H. Xiao, J. Tahir-Kheli, W.A. Goddard, J. Phys. Chem. Lett. 2 (2011) 212.