Simulation of Young's Modulus of Single-walled Carbon Nanotubes by Molecular Dynamics

Embed Size (px)

Citation preview

  • 5/22/2018 Simulation of Young's Modulus of Single-walled Carbon Nanotubes by Molecular Dynamics

    1/8

    Physica B 352 (2004) 156163

    Simulation of Youngs modulus of single-walled carbon

    nanotubes by molecular dynamics

    Bao WenXinga,b,, Zhu ChangChuna, Cui WanZhaoa

    aSchool of Electronics & Information Engineering, Xian Jiaotong University, Xian 710049, PR ChinabThe Second North-west Institute for Ethnic Minorities, Yinchuan 750021, PR China

    Received 15 April 2004; accepted 13 July 2004

    Abstract

    Based on molecular dynamics (MD) simulation, the Youngs moduli of carbon nanotubes are studied. The inter-

    atomic short-range interaction and long-range interaction of carbon nanotubes are represented by a second generation

    reactive empirical bond order (REBO) potential and LennardJones (LJ) potential, respectively. The obtained potential

    expression is used to calculate the total potential energies of carbon nanotubes. Three types of single-walled carbon

    nanotubes (SWCNTs), armchair, zigzag and chiral tubules, are calculated, respectively. The computational results show

    that the Youngs moduli of SWCNTs are in the range of 929.8711.5 GPa. From the simulation, the Youngs moduli of

    SWCNTs are weakly affected by the tube chirality and tube radius. The numeric results are in good agreement with theexisting experimental results.

    r 2004 Elsevier B.V. All rights reserved.

    PACS: 61.48.+c; 62.20.Dc; 31.15.Qg

    Keywords:Carbon nanotubes; Molecular dynamics; Youngs modulus; Potential function

    1. Introduction

    Since the first report of carbon nanotubes

    (CNTs) by Iijima [1], CNTs have got a lot ofattention because of their unique electronic and

    mechanical properties. As a one-dimensional

    structure, CNTs can be thought of as one sheet

    or multiple sheets of graphene rolled into a

    cylinder. They have single or multiple layers ofcarbon atoms in the tube thickness direction,

    called single-walled carbon nanotubes (SWCNTs)

    or multi-walled carbon nanotubes (MWCNTs),

    respectively. According to different chiral angles,

    SWCNTs can be classified into zigzag (y=01),

    armchair (y=301) and chiral tubule (01oyo

    301) [2].

    ARTICLE IN PRESS

    www.elsevier.com/locate/physb

    0921-4526/$ - see front matterr 2004 Elsevier B.V. All rights reserved.

    doi:10.1016/j.physb.2004.07.005

    Corresponding author. Institute of Vacuum Microelectro-

    nics & Microelectromechanical System, School of Electronics &

    Information Engineering, Xian Jiaotong University, Xian

    710049, PR China. Tel.: +86-29-82668644; fax: +86-29-

    82663957.

    E-mail address: [email protected] (B. WenXing).

    http://www.elsevier.com/locate/physbhttp://www.elsevier.com/locate/physb
  • 5/22/2018 Simulation of Young's Modulus of Single-walled Carbon Nanotubes by Molecular Dynamics

    2/8

    There are some experimental studies of the

    Youngs modulus of CNTs. Liu et al. [3]reported

    the Youngs modulus of CNTs is 10.1 TPa with

    the diameter increasing from 8 to 40nm bymeasuring resonance frequency of carbon nano-

    tubes. Krishnan et al.[4] used TEM to observe the

    vibration of an SWCNT at room temperature and

    reported Youngs modulus of SWCNTs in the

    range from 0.90 to 1.70 TPa, with an average of

    1.25 TPa. Tombler et al. [5]used AFM to bend an

    SWCNT and reported the Youngs modulus of

    SWCNTs around 1.2 TPa. Yu et al. [6]conducted

    nanoscale tensile tests of SWCNT ropes pulled by

    AFM tips under a scanning electron microscope

    and reported the Youngs modulus of SWCNT

    ropes ranged from 0.32 to 1.47 TPa (mean

    1002GPa). Demczyk et al. [7] reported the

    Youngs modulus of MWCNTs range from 0.8

    to 0.9 TPa used TEM to bend an individual tubes.

    There are also some theoretical studies for

    predicting the Youngs modulus of CNTs. Lu [8]

    used an empirical force-constant model to deter-

    mine several elastic moduli of SWCNTs and

    MWCNTs and obtained the Youngs modulus of

    about 1 TPa. Li et al.[9] used molecular structural

    mechanics method and obtained the Youngs

    modulus of 1050750 GPa for MWCNTs. Jin etal.[10]used MD and force-constant approach and

    reported the Youngs modulus of SWCNTs to be

    about 123677 GPa. Cornwell and Wille [11] used

    the MD with the TersoffBrenner potential[12]to

    obtain the Youngs modulus of SWCNTs about

    0.8 TPa. Zhou et al. [13] used first principles and

    reported the Youngs modulus of SWCNTs to be

    0.76 TPa. Vodenitcharova et al. [14] used con-

    tinuum mechanics model to obtain the Youngs

    modulus of SWCNTs about 4.88 TPa.

    The objectives of this paper are to compute theYoungs modulus of SWCNTs in detail using MD

    simulation. MD simulations are the powerful tools

    for investigation of the mechanical properties and

    structural formation of carbon nanotubes. The key

    to the MD simulation is to choose an exactly

    potential energy model. A modified empirical

    potential energy model is presented to describe

    the bonding and nonbonding interaction. The

    effects of carbon nanotubes structure, such as

    tube radius and chirality on Youngs modulus of

    SWCNTs, are examined. The accuracy and

    stability of this method have been verified through

    its application to several single layer and multi-

    layer graphene sheets with different model sizes.The computational results of the Youngs modulus

    of SWCNTs are in good agreement with existing

    experimental values. The organization of this

    paper is as follows. In the second section, MD

    simulation method and Verlet algorithm [15] are

    described. The total potential energies and inter-

    atomic forces are described in Section 3. The

    expression of Youngs moduli of CNTs and

    graphite are presented in Section 4. Results and

    discussion are given in Section 5. The final section

    is the conclusions.

    2. Molecular dynamics simulation method

    In order to compare theoretical results with

    experimental data, the ultimately investigated

    model structures should represent stable or meta-

    stable minimal energy configurations, which have

    to be obtained by finite-temperature structure

    optimization. The MD method has to rely on a

    mathematical description of the total energy of thesystem as a function of all atomic coordinates.

    Focusing on the MD method, the Newtons

    equation of motions for all atoms in the structure

    has to be solved,

    ma md2ri

    dt2

    Xjai

    frij; 1

    where a is the acceleration of particle i, ri is the

    position of particle i, rij is the distance between

    particleiand particlej,fis the total force acting onthe particleiand determined by the gradient of the

    total potential energy of the system,f=rEtot. In

    order to achieve control of the energy conserva-

    tion, it would be desirable to determine the atomic

    positions and velocities at each simulation time

    step. Standard Verlet algorithm has the obvious

    disadvantage that the positions and velocities are

    not synchronized. Swope presented a method of

    directly getting the positions and velocities at each

    simulation time step , and provided an easy way of

    ARTICLE IN PRESS

    B. WenXing et al. / Physica B 352 (2004) 156163 157

  • 5/22/2018 Simulation of Young's Modulus of Single-walled Carbon Nanotubes by Molecular Dynamics

    3/8

    calculating the velocities into Verlet algorithm[16],

    rth rt hvt h2at=2; 2

    vth vt h ath at =2; 3

    where r is the position, v is the velocity, a is the

    acceleration at time t and time t+h, respectively,

    andh is the time step. As can be seen from Eq. (3),

    to calculate the new velocities requires the accel-

    erations at time t and time t+h, respectively.

    Before going into any MD simulation, the basic

    principles of the calculation of the total potential

    energies and inter-atomic forces have to be described.

    3. Total potential energies and inter-atomic forces

    The reliability of MD simulations techniques

    depend on the use of appropriate inter-atomic

    energies and forces. Inter-atomic energies and forces

    were calculated generally using an empirical bond-

    order potential that describes the covalent bonding

    within both small hydrocarbon molecules and

    CNTs. It has been successfully used to study the

    structure and mechanical properties of CNTs

    [12,1719]. However, the empirical bond-order

    potential is short-ranged and does not include

    long-range nonbond interactions. For instance,

    long-range nonbond interactions can be either

    van-der-Waals or Coulomb interactions, or both.

    Van-der-Waals interactions normally have a weak

    influence on the overall mechanical behavior among

    the atomic interactions of the CNTs [20]. A long-

    range LJ potential is added to model the van-der-

    Waals interactions that exist in the SWCNTs. It

    becomes activated only after the short-ranged

    REBO potential goes to zero. The total potential

    energy of the system may be described by summingbonding energy and nonbonding energy, Etot=E-

    bonding+Enonbonding. For CNTs and graphite, the

    bonding term can be described precisely by REBO

    potential, and the nonbonding term is mainly the

    energy of van-der-Waals force.

    The van-der-Waals force is most often modeled

    using the LJ 612 potential function [21]:

    ELJr 4 s

    r

    6

    s

    r

    12 ; 4

    whereris the distance between interacting atoms,and s are the LJ parameters, for graphite and

    CNTs, 0:00239 eV and s 0:341 nm [22],

    respectively. The cutoff distance of LJ potentialis 2:5s; no interactions are evaluated beyond thisdistance.

    Brenner (2002) [23] presented a second-genera-

    tion reactive empirical bond order (REBO) poten-

    tial energy expression for hydrocarbons. In the

    REBO potential, the total potential energy of the

    system is given by

    EREBO X

    i

    Xji1

    ERrij bijEArij

    ; 5

    where rij

    is the distance between pairs of nearest-

    nearby atoms iandj, bijis a many-bond empirical

    bond-order term between atoms i and j that is

    derivable from Huckel electronic structure theory.

    ER and EA are the repulsive and attractive pair

    terms, respectively,

    ERr fcr1Q

    rA exp ar; 6

    EAr fcrX3n1

    Bnebnr; 7

    where the parameters Q, A, a; B, b are used to fitthe pairs terms, the values for all the parameters

    are given in Ref [23]. The function fc is a cutoff

    function, and is given by

    fcr

    1; roRmin;

    1cos prRmin

    RmaxRmin

    h in o=2; RminoroRmax;

    0; r4Rmax

    8>:

    8

    where RmaxRmin defines the distance over the

    function goes from one to zero, for graphiteand CNTs, Rmax=0.26 nm and Rmin=0.17 nm,

    respectively.

    4. Expression of Youngs modulus

    Youngs modulus is one of the important

    characterizations of mechanical properties of

    materials. In classical mechanics the Youngs

    ARTICLE IN PRESS

    B. WenXing et al. / Physica B 352 (2004) 156163158

  • 5/22/2018 Simulation of Young's Modulus of Single-walled Carbon Nanotubes by Molecular Dynamics

    4/8

    modulus is defined as

    Y s

    F=A

    DL=L0; 9

    where s is the axial stress, is the strain, F is theaxial tensile force acted on the object, A is the

    cross-sectional area of object, L0 is the initial

    length and DL is the elongation under the force F.

    The load conditions of CNTs and graphite are

    illustrated inFig. 1,respectively.After putting the tensile force on CNTs or

    graphite in axial direction and calculating the

    variation of the total potential and length, we can

    get the Youngs moduli of CNTs or graphite.

    5. Results and discussion

    In this study, Youngs moduli of graphite and

    SWCNTs are simulated using our molecular

    dynamics simulation program. The MD simula-

    tion program can run on FreeBSD 4.6 and

    Windows 2000 platform. The simulation time step

    is 1 fs. The Langevin thermostats are applied tosimulate and maintain temperature at 300 K. MD

    simulation was performing approximately 5000

    steps when the energy attained a constant mini-

    mum value. A periodic boundary condition has

    been used along the axial direction to mode the

    tubes under axial loading.

    5.1. Youngs modulus of graphite

    The Youngs modulus of graphite is calculated

    to verify the feasibility of the present method.

    These calculations can provide useful information

    for CNTs. In Eq. (9) the crossing-sectional area of

    graphite is expressed as A NWdR; Nis numberof graphene layers, W is width of graphite, the

    interspacing between graphene layers dR are

    assumed to be 0.34 nm.

    The computational results of Youngs moduli of

    graphite are showed in Tables 1 and 2, respec-

    tively. There is a negligible difference of Youngs

    modulus in Table 1 between the single layer of

    graphite and multi-layers of graphite, and the

    results inTable 2are weakly affected by model sizeof graphite. The average of Youngs modulus of

    graphite is 1026.176 GPa, it is very close to the

    experimental value of 1025.0 GPa [24,25]. Fig. 2

    shows the variation of Youngs modulus of

    graphite with number of graphene layers.

    5.2. Youngs modulus of SWCNTs

    Based on this method, the Youngs moduli of

    SWCNTs are studied. Eq. (9) is still used for

    calculating the Youngs modulus, the cross-sec-tional area A of SWCNTs is 2pRdR; where Rstands for the tube radius and dR stands for the

    wall thickness with a value of 0.34 nm[5,8]. Three

    types of SWCNTs, armchair, zigzag and chirality,

    are considered. By scaling atomic positions along

    the axial direction, the variation of the total energy

    and variation of tube length are calculated.

    The stressstrain curve of armchair (8, 8)

    SWCNTs is obtained and given in Fig. 3. It is

    seen in the figure that the elastic limit is at the

    ARTICLE IN PRESS

    Fig. 1. Load conditions of graphite and SWCNTs.

    B. WenXing et al. / Physica B 352 (2004) 156163 159

  • 5/22/2018 Simulation of Young's Modulus of Single-walled Carbon Nanotubes by Molecular Dynamics

    5/8

    strain value of 0.08. Beyond the elastic limit, the

    stress-strain curve deviate from a straight line.

    Computational results of Youngs modulus of

    SWCNTs for different tube radius are shown in

    Fig. 4. There is a slight difference between the

    trends of variation in Youngs modulus of arm-

    chair and zigzag SWCNTs inFig. 4.The Youngsmodulus of chiral SWCNTs is slightly lower than

    those values of armchair and zigzag. The differ-

    ence between maximum and minmum Youngs

    modulus values is about 30 GPa. The effect of the

    chiral angle on the Youngs modulus of SWCNTs

    is shown in Fig. 5. With the variation of chiral

    angles, the Youngs modulus value is slightly

    fluctuated from 904 to 935 GPa. The effect of tube

    radius and tube chirality on Youngs modulus of

    SWCNTs is very weak.

    ARTICLE IN PRESS

    Table 1

    The variation of Youngs modulus of Graphite with number of

    graphene layers

    Number ofgraphene layers

    Y (GPa) (presentprediction)

    Y (GPa)(experimental

    value)

    1 1030.530

    2 1032.033

    3 1033.771 1025.000

    4 1022.156

    5 1012.390

    Average 1026.176

    Model size: height is 10.508 nm, width is 1.107 nm.

    Table 2The variation of Youngs modulus of graphite with model size

    Height (nm) Width (nm) Y (GPa)

    16.898 2.336 1012.230

    8.378 0.861 1061.025

    33.938 1.107 1036.707

    16.898 1.107 1033.246

    10.508 1.107 1030.530

    Fig. 2. Youngs modulus of graphite.

    Fig. 3. The curve of variation of axial stresss with the axialstrain (armchair (8,8)).

    Fig. 4. Youngs modulus of SWCNTs for different tube radius.

    B. WenXing et al. / Physica B 352 (2004) 156163160

  • 5/22/2018 Simulation of Young's Modulus of Single-walled Carbon Nanotubes by Molecular Dynamics

    6/8

    In order to compare numerical results with

    other predictions of Youngs modulus, all the

    computational results in this study are given in

    Tables 3 and 4, respectively. Table 3 lists thevalues of Youngs modulus of armchair, zigzag,

    chiral SWCNTs with different tube radius. Table 4

    lists the values of Youngs modulus for different

    chiral angers. The present approach gives slightly

    lower values of Youngs modulus than those of

    reported in [810,26]. The average of Youngs

    modulus of SWCNTs is 929.8711.5 GPa and

    agrees very well with the existing experiment

    values [35]. Comparisons between numerical

    results and experimental results and other theore-

    tical prediction of Youngs modulus of SWCNTs

    listed inTable 5.

    ARTICLE IN PRESS

    Fig. 5. The effect of the chiral angle on the Youngs modulus ofSWCNTs.

    Table 3

    Youngs modulus of SWCNTs for different tube radius

    (n, n) Number of atoms Radius (nm) Length (nm) Y (GPa)

    Armchair

    (8,8) 1168 0.542 8.854 934.960

    (10,10) 1460 0.678 8.854 935.470

    (12,12) 1752 0.814 8.854 935.462

    (14,14) 2324 0.949 10.084 935.454

    (16,16) 3040 1.085 11.560 939.515

    (18,18) 3924 1.220 13.281 934.727

    (20,20) 5000 1.356 15.250 935.048

    Average 935.80570.618

    Zigzag

    (14,0) 840 0.548 6.230 939.032

    (17,0) 1360 0.665 8.362 938.553

    (21,0) 1890 0.822 9.428 936.936

    (24,0) 2400 0.939 10.500 934.201

    (28,0) 3080 1.096 11.563 932.626

    (31,0) 3720 1.213 12.621 932.598

    (35,0) 4900 1.370 14.757 933.061

    Average 935.28772.887

    Chirality

    (12,6) 1344 0.525 9.023 927.671

    (14,6) 1896 0.696 11.367 921.616

    (16,8) 2240 0.828 11.279 928.013

    (18,9) 2520 0.932 11.279 927.113

    (20,12) 3920 1.096 14.921 904.353

    (24,11) 3844 1.213 13.215 910.605

    (30,8) 4816 1.358 14.792 908.792

    Average 918.309710.392

    Average 929.8711.5

    The length-to-radius ratio is greater than 10.

    B. WenXing et al. / Physica B 352 (2004) 156163 161

  • 5/22/2018 Simulation of Young's Modulus of Single-walled Carbon Nanotubes by Molecular Dynamics

    7/8

    As can be seen inTable 5, the values of Youngs

    modulus obtained by different researchers is

    different. The reasons and effect of geometric

    structure of SWCNTS of elastic moduli can begiven in the following:

    (i) Different wall thickness values were used to

    calculate the Youngs moduli: in Refs.

    [810,26,27] and this present prediction, dR

    0:34 nm was used, and obtained the values ofYoungs modulus ranging from 0.9 to

    1.26 TPa. But in Ref. [14] dR 0:0617 nmwas used, and the obtained values of Youngs

    modulus was 4.88 TPa. Compared with the

    CC bond length of CNTs, 0.0617 nm the

    magnitude of wall thickness seems too small.

    To use dR 0:34 nm calculated the values ofYoungs modulus were closed to experimental

    values than Ref. [14] inTable 5.

    (ii) Difference in the tube model sizes: in Refs.

    [2730], the length-to-radius ratio is smaller than

    10. Therefore, the edge effects may have effect

    on the results. In this present case and Ref. [10],

    the length-to-radius ratio is greater than 10 to

    ensure that edge effects can be ignored.

    (iii) Different accuracy of theoretical approach: ab

    initio, DFT and tight-binding are quantum-

    mechanical-based methods, the results ob-

    tained are reliable than other methods, but

    the computational model is only composed of

    several ten to several hundred atoms [2630].

    MD methods can be used to model several

    thousand to several million atoms and the

    results are more close to real CNT. Used

    accurate potential model, the simulation

    results of MD method are very closer to

    experimental data.

    (iv) The effect of endcap on Youngs modulus of

    SWCNTs: There are two different models,

    open-ended SWCNTs [810,26,28,29] andclose-capped SWCNTs[27], respectively. From

    Table 5 we see that the Youngs modulus of

    ARTICLE IN PRESS

    Table 4

    Youngs modulus of SWCNTs for different chiral angle

    (n,n) Angle (deg) Radius (nm) Y (GPa)

    (35,0) 0.0 1.370 933.061

    (20,2) 4.715 0.825 918.019

    (12,2) 7.589 0.514 920.674

    (8,2) 10.893 0.359 913.530

    (12,4) 14.920 0.565 919.823

    (10,4) 16.100 0.489 912.852

    (14,6) 17.000 0.696 921.616

    (18,9) 19.106 0.933 927.113

    (20,12) 21.780 1.097 904.353(24,18) 25.284 1.430 907.028

    (20,20) 30.0 1.356 935.048

    Table 5

    Comparisons between numerical results and experiment results and other theoretical prediction of Youngs modulus of SWCNTs

    Methods Wall thickness value (nm) Y (GPa) Ref.

    Armchair Zigzag Chirality

    MD 0.34 935.8 935.3 918.3 This present

    Empirical force-constant model 0.34 972.0 975.0 973.0 [8]

    Molecular structural mechanics model 0.34 1050750 1080720 [9]MD and Force approach 0.34 123677 [10]

    Continuum mechanics model 0.0617 4880 [14]

    Tight-binding 0.34 12201250 12201260 12401250 [26]

    Tight-binding 0.28 950 [29]

    Ab initio 0.34 1060 11401180 1030 [27]

    Ab initio 1000 [28]

    DF-TB 0.34 11002800 [30]

    Measuring resonance frequencya 1001000 [3]

    TEMa 9001700 [4]

    AFMa 1200 [5]

    aExperimental methods.

    B. WenXing et al. / Physica B 352 (2004) 156163162

  • 5/22/2018 Simulation of Young's Modulus of Single-walled Carbon Nanotubes by Molecular Dynamics

    8/8

    open-ended SWCNTs in this present prediction

    and other prediction is similar to close-capped

    SWCNTs[27]. From Ref.[27],a tube with long

    structures are expected to avoid the influenceof endcap effects.

    (v) The effect of tube radius and tube chirality on

    Youngs modulus of SWCNTs: most studies

    [810,14,2629]show that the values of Youngs

    moduli of SWCNTs are independent of tube

    radius and tube chirality, except for Ref. [30].

    This is also validated by this present work.

    6. Conclusions

    Using MD method, we study the Youngs moduli

    of graphite and SWCNTs in detail. The Verlet

    algorithm, which was improved by Swope in MD

    method, was used to obtain atomic positions and

    velocities at each simulation time step. Based on the

    modified empirical potential function models, the

    van-der-Waals forces in SWCNTs are taken into

    account. The results show that the average of

    Youngs modulus of graphite is 1026.176 GPa,

    which is very well closed to the experiment value.

    The average of Youngs modulus of SWCNTs is

    929.8711.5 GPa, which is slightly lower than theYoungs modulus of graphite. It is also in good

    agreement with the existing experimental ones. The

    results show that the Youngs moduli of SWCNTs

    are independent of the tube radius and tube

    chirality. All these illustrate the simulation of

    Youngs modulus of SWCNTs with MD and

    modified potential energy model is quite workable

    and reliable. It is expected that the present model

    will be further developed to model other covalent-

    bonded materials and analyze the mechanical

    properties of nanotube-based composites.

    Acknowledgements

    The research is supported by the National Natural

    Science foundation of China (Grant No. 60036010)

    and China Doctoral foundation of National Educa-

    tion Bureau (Grant No. 2000069823).

    References

    [1] S. Iijima, Nature 354 (1991) 56.

    [2] M.S. Dresselhaus, G. Dresselhaus, R. Saito, Carbon 33

    (1995) 883.

    [3] J.Z. Liu, Q.S. Zheng, Q. Jiang, Phys. Rev. Lett. 86 (2001)

    4843.

    [4] A. Krishnan, E. Dujardin, T.W. Ebbesen, et al., Phys. Rev.

    B 58 (1998) 14013.

    [5] T.W. Tombler, C.W. Zhou, L. Alexseyev, et al., Nature

    405 (2000) 769.

    [6] M.F. Yu, B.S. Files, S. Arepalli, R.S. Ruoff, Phys. Rev.

    Lett. 84 (2000) 5552.

    [7] B.G. Demczyk, Y.M. Wang, J. Cumings, et al., Mat. Sci.

    Eng. A-Struct. 334 (2002) 173.

    [8] J.P. Lu, Phys. Rev. Lett. 79 (1997) 1297.

    [9] C.Y. Li, T.W. Chou, Compos. Sci. Technol. 63 (2003)

    1517.[10] Y. Jin, F.G. Yuan, Compos. Sci. Technol. 63 (2003) 1507.

    [11] C.F. Cornwell, L.T. Wille, Solid. State. Commun. 101

    (1997) 555.

    [12] D.W. Brenner, Phys. Rev. B 42 (1990) 9458.

    [13] X. Zhou, J. Zhou, Z.C. Ou-Yang, Phys. Rev. B 62 (2000)

    13692.

    [14] T. Vodenitcharova, L.C. Zhang, Phys. Rev. B 68 (2003)

    165401.

    [15] L. Verlet, Phys. Rev. 159 (1967) 98.

    [16] C.S. William, C.A. Hans, J. Chem. Phys. 76 (1982) 637.

    [17] D.H. Robertson, D.W. Brenner, J.W. Mintmire, Phys.

    Rev. B 45 (1992) 12592.

    [18] B.I. Yakobson, C.J. Brabec, J. Bernholc, Phys. Rev. Lett.

    76 (1996) 2511.[19] M. Huhtala, A. Kuronen, K. Kaski, Comput. Phys.

    Commun. 146 (2002) 30.

    [20] T. Hertel, R.E. Walkup, P. Avouris, Phys. Rev. B 58

    (1998) 13870.

    [21] J.E. Lennard-Jones, Proc. Roy. Soc. 106A (1924) 441.

    [22] L.A. Girifalco, M. Hodak, R.S. Lee, Phys. Rev. B 62

    (2000) 13104.

    [23] D.W. Brenner, O.A. Shenderova, J.A. Harrison, et al.,

    J. Phys.: Condens. Mater 14 (2002) 783.

    [24] B.T. Kelly, Physics of Graphite, Applied Science Press,

    London, 1981.

    [25] O.L. Blakslee, D.G. Proctor, E.J. Selden, et al., Appl.

    Phys. 41 (1970) 8.[26] E. Herna ndez, C. Goze, P. Bernier, A. Rubio, Phys. Rev.

    Lett. 80 (1998) 4502.

    [27] G.V. Lier, C.V. Alsenoy, V.V. Doren, P. Geerlings, Chem.

    Phys. Lett. 326 (2000) 181.

    [28] D. Sa nchez-Portal, E. Artacho, J.M. Soler, Phys. Rev. B

    59 (1999) 12678.

    [29] J. Cai, R.F. Bie, X.M. Tan, C. Lu, Physica B 344 (2004) 99.

    [30] Z. Peralta-Inga, S. Boyd, J.S. Murray, C.J. OConnor,

    P. Politzer, Struct. Chem. 14 (2003) 431.

    ARTICLE IN PRESS

    B. WenXing et al. / Physica B 352 (2004) 156163 163