Вы находитесь на странице: 1из 4

PHYSICAL REVIEW B 66, 153407 共2002兲

Nonlinear elastic properties of carbon nanotubes subjected to large axial deformations


T. Xiao* and K. Liao†
School of Mechanical and Production Engineering, Nanyang Technological University, Singapore 639798
共Received 25 July 2002; published 24 October 2002兲
Nonlinear elastic properties of single-wall carbon nanotubes subjected to large-scale tensile deformation
were investigated using the second-generation Brenner potential. The energy change of the nanotubes was
found to be a cubic function of tensile strains. In contrast to values of in-plane stiffness C based on small
strains reported previously in the literature, the present study showed that C is linearly dependent on axial
tensile deformation. C is also affected by the chirality of the tubes to some extent when axial strains are large.
Consequently, these tubes should be considered in the context of anisotropic elastic shell models in the event
of employing continuum mechanics in the analysis of nanotubes.

DOI: 10.1103/PhysRevB.66.153407 PACS number共s兲: 61.48.⫹c, 62.20.Dc

Since the discovery of carbon nanotubes 共CNTs兲 by Iijima ventionally by engineers, are equivalently utilizable for ana-
and Ichihashi1,2 in 1991, the prospect of their potential ap- lyzing CNTs. The larger the number of atoms in a CNT, the
plications has initiated intensive research. Being the stron- more analogous it is to a continuum object. In this sense, the
gest and stiffest among man made fibers, these nanowires are advantages of continuum mechanics may compensate the
anticipated to play a key role in future high-performance disadvantages of ab initio and other MD methods. Con-
nanocomposites. As such, better understanding and accurate tinuum mechanics has been employed in many occasions to
characterization of their mechanical properties are needed. study the behavior of CNTs. Yakobson, Brabec, and
To study CNTs and their relatives, quantum mechanics cal- Bernholc6 proposed a continuum shell model with properly
culations are most reliable and accurate. However, at present, chosen tubular parameters to explained the pattern transfor-
ab initio method is merely available for relatively simple mations of SWNTs subjected to large axial compression,
atomic structures because of enormous amount of computa- such that the critical deformation at which buckling occurs
tions required. In most cases when CNTs with a large num- can be estimated. Tersoff10 introduced an approach of calcu-
ber of carbon atoms are concerned, molecular dynamics lating energies of fullerene balls by viewing fullerenes as
共MD兲 simulations using empirical force fields are frequently distorted graphene sheets and applied an elastic cone shell
employed. For instance, Robertson, Brenner, and Mintmire3 model to fullerene balls to approximate their energies. Ru,11
have examined the energetics of a large number of single- Govindjee and Sackman12 presented shell models for multi-
wall nanotubes 共SWNTs兲 using the Tersoff-Brenner walled carbon nanotubes to study deformation and buckling.
potential,4,5 and found that the strain energy per atom of a It is demonstrated by all these works that continuum me-
SWNT is sensitive to its radius when the radius is small, but chanics is particularly useful for the study of large-sized
it is insensitive to other aspects of the lattice structure. Ya- atomic systems.
kobson, Brabec, and Bernholc6 investigated morphological In this paper, we consider the situation of SWNTs sub-
pattern transformations of SWNTs also by means of the jected to large-scale tensile deformation. Nonlinear elastic
Tersoff-Brenner potential, and showed that the shape change properties of SWNTs that can be used in continuum descrip-
of nanotubes corresponds to an abrupt release of energy. tion of their elastic response, were studied using the second-
Zhang, Lammert, and Crespi7 investigated the plastic defor- generation Brenner potential.
mation of SWNTs using tight binding method, and found that In continuum mechanics, the constitutive relations, i.e.,
the elastic limit of SWNTs depends on chirality due to bond relations of load and deformation, have to been established
rotation in the graphitic network, although their elastic prop- prior to solving a specific problem. If the material is homo-
erties are nearly independent of it. Srivastava, Menon, and geneous and isotropic, it becomes the problem of defining
Cho8 calculated the strain energy of a SWNT under uniaxial two basic parameters,13 Young’s modulus Y and Poisson’s
compression using both tight binding method9 and the ratio ␯. Suppose a material undergoes a uniaxial tension test,
Tersoff-Brenner potential. Their study showed that at low Y is conventionally defined as
strain before any change of morphological pattern happens,
the two sets of results are in good agreement with each other. 1 ⳵ 2E
Y⫽ , 共1兲
At high-strain buckling occurs in the simulation with the V ⳵␧2
Tersoff-Brenner potential, and a deviation between them was
found, as had been analyzed in detail by Yakobson, Brabec, where V is the volume of the material, E the total strain
and Bernholc.6 energy under tensile deformation, and ␧ the tensile strain.
Although the aforementioned molecular simulations are However, the definition in Eq. 共1兲 encounters difficulty in
quite reliable in the investigations of nanostructures, inten- CNTs because there is no translational invariance in the ra-
sive processing time is required and large input and output dial direction.14 For these hollow CNTs, different thicknesses
that result are hard to handle in engineering design. On the have been adopted,15 and the volume V is not properly de-
other hand, continuum mechanics, which has been used con- fined. Therefore, in-plane stiffness C which is independent of

0163-1829/2002/66共15兲/153407共4兲/$20.00 66 153407-1 ©2002 The American Physical Society


BRIEF REPORTS PHYSICAL REVIEW B 66, 153407 共2002兲

any shell thickness, is defined as an alternative measurement


of the mechanical characteristic of nanotubes,

1 ⳵ 2E
C⫽ , 共2兲
S ⳵␧2
where S is the surface area of the nanotube.
A graphene sheet, which possesses a two-dimensional
hexagonal structure, is isotropic in the case of small
deformation.13 Since a single CNT can be viewed as a
graphene sheet rolled up into a cylindrical tube with a con-
stant radius,16 CNTs are isotropic as well when tension de-
formation is small, and they can be characterized by a uni-
form cylindrical shell model with only two elastic
parameters. This argument has been confirmed by the results FIG. 1. Energy change per atom of nanotube 共10, 10兲 under
axial tension. Crosses are simulation results using the second-
of Sanchez-Portal, Artacho, and Soler,17 which showed that
generation Brenner potential, and solid line is the regression curve
stiffness C does not vary significantly with the radii and
using Eq. 共4兲.
chirality of SWNTs. Other results18 –20 also showed that the
strain energy per atom is insensitive to the lattice structures
bon and hydrocarbon molecules, which provides a better de-
of CNTs. If a representative thickness h is defined for a
scription of bond energies and length for carbon nanotubes.
graphene sheet or a single CNT,6,19,21 then their volume is
Here this second-generation Brenner potential is adapted to
V⫽Sh, and the representative Young’s modulus is
compute the strain energy E of CNTs in Eq. 共2兲. The ener-
getics of two groups of SWNTs subjected to axial tension
C
Y⫽ . 共3兲 were examined: arm-chair group from 共4, 4兲 to 共15, 15兲,32
h and zigzag group from 共7, 0兲 to 共20, 0兲, plus 共22, 0兲 and 共25,
0兲. The values of the in-plane stiffness C of SWNTs were
The lattice constants of a perfect graphite were experi- determined by Eq. 共2兲. Analyzing the computational results
mentally determined,22,23 with a 0 ⫽0.2462 nm and c 0 of each SWNT, it is found that for every SWNT the relation
⫽0.6707 nm, so that the specific volume per atom V 0 between ⌬E, the energy changes per unit area, and the ap-
⫽0.0088 nm3 and the specific area per atom S 0 plied strain ␧ can be nearly perfectly fitted into a cubic poly-
⫽0.02624 nm2 . The interplanar separation h 0 ⫽c 0 /2 nomial function with two coefficients, a and b, as follows:
⫽0.335 nm, is defined as the representative thickness of the
graphite. The unit of energy per atom for graphite, 1 eV/ ⌬E⫽a␧ 2 ⫺b␧ 3 , 共4兲
atom, is equal to 6.1 J/m2.
where values of a, b are determined by regression analysis.
Divergence exists in the values of the in-plane stiffness C
or representative modulus Y.3,8,17–20,24,25 Among them the The energy change per atom ⌬E of a 共10, 10兲 SWNT with
respect to applied strain ␧ subjected to axial tension, is illus-
maximum value3 is C⫽59 eV/atom⫽360 J/m2 , the mini-
trated in Fig. 1. Regression analysis showed that a
mum value24 is C⫽220 J/m2 . Typically C⫽342 J/m2 is ex-
pected for SWNTs.14 The differences, possibly attributed to ⫽25.6 eV/atom, b⫽48.2 eV/atom, with a multiple correla-
the simulation methods employed or different ways of load tion coefficient R⬎0.999, and the estimated standard error
applications, are in general within an acceptable range of less than 0.005. Similar high-R values and low-standard er-
same order of magnitude. rors were also obtained for other SWNTs from simulations of
Although different values of stiffness are reported, all of tensile response.
them are assumed to be constants. It will be shown later that The force per unit circumference acting on the ends of a
the in-plane stiffness C may not be treated as a constant in SWNT, T, is the first derivative of Eq. 共4兲:
general, it is dependent on the strain ␧ of the CNTs, and can ⳵E
only be approximated as a constant when ␧ is very small. It T⫽ ⫽2a␧⫺3b␧ 2 共5兲
has been known that CNTs are remarkably resilient,6 able to ⳵␧
retain their elastic behavior when strained to 15% when the and the in-plane stiffness C is the second derivative of Eq.
material yields. Under large-scale deformation, nonlinear 共4兲. According to Eq. 共2兲, we have
elastic response of CNTs becomes more obvious, and there-
fore it must be taken into account. 1 ⳵ 2 E ⳵ 2 ⌬E
In many cases3,6,24,26 –30 the Tersoff-Brenner potential4,5 is C⫽ ⫽ ⫽ ␣ ⫺ ␤ ␧, 共6兲
S ⳵␧2 ⳵␧2
used in molecular mechanics simulations to study the behav-
ior of carbon-based structures. This potential function accu- where ␣ ⫽2a, ␤ ⫽6b. The coefficient ␣ reflects the initial
rately describes the bonding energies of graphite phase of in-plane stiffness of the tube, and ␤ the influence of ␧ on the
carbon atoms, and the results are in good agreement with in-plane stiffness C. It is seen from Eq. 共4兲 that coefficients a
those of ab initio calculations.3 Recently Brenner presented a and b reflect the linear and nonlinear properties of the tube,
second-generation potential energy function31 for solid car- respectively. At small deformation when ␧ approaches 0, C

153407-2
BRIEF REPORTS PHYSICAL REVIEW B 66, 153407 共2002兲

FIG. 2. Tensile force of nanotube 共10, 10兲 versus axial tensile FIG. 3. Variation of coefficients a and b with the increase of
strain. The solid line is the curve of Eq. 共5兲, crosses on the curve tube diameter, of 12 armchair-type tubes from 共4, 4兲 to 共15, 15兲, and
correspond to those in Fig. 1. The dash line is the tangent line of the 16 zigzag-type tubes from 共7, 0兲 to 共20, 0兲, plus tube 共22, 0兲 and
curve at the origin, T⫽2a␧, depicting the specific condition of Eq. tube 共25, 0兲. The solid lines are used as guides to the eye.
共5兲 when the material is considered linear at small deformation.
ment with our result of T max⫽27 N/m). That value is nearly
approximates to ␣. The in-plane stiffness C reported in the
twice that of arm-chair tubes, suggesting that the onset of
literatures to date conceptually corresponds to the coefficient
plastic response depends on chirality of the tubes. Here we
␣, a constant.
focus on defect-free CNTs under completely elastic deforma-
Variation of tensile force T with the strain ␧ of a 共10, 10兲
tion, with no structural rearrangement during loading, so that
SWNT according to Eq. 共5兲 is illustrated in Fig. 2, where the
the elastic strain limit and the force limit are insensitive to
in-plane stiffness C is the initial slope of curve. In Fig. 2 it is
chirality.
seen that when ␧ is small 共less than 0.02兲, the curve coin-
Coefficients a and b versus different tube diameters from
cides with its tangent at the origin 共dash line兲, suggesting that
simulations of two groups of SWNTs under axial tension are
the nonlinear constitutive relation may be simplified into
depicted in Fig. 3. It is found that coefficient a of the two
Hooke’s law. The nonlinear response becomes apparent with
groups of CNTs are close to each other, indicating that at
increasing ␧. The force-strain curve tends to flatten out, im-
small deformation the in-plane stiffness C 共at the moment
plying that the material is gradually softening, until the point
C⫽2a⫽ ␣ ) is insensitive to chirality. That coefficient b of
when the tangent of the curve is leveled, that is, ⳵ T/ ⳵ ␧⫽0 or
the two groups of CNTs are distinct, indicates that at large
C⫽0, at which point the tensile force reaches its maximum
deformation, C 共now C⫽ ␣ ⫺ ␤ ␧) is not only being affected
T max :
by ␧, but also sensitive to chirality. Moreover, the values of a
T max⫽2a␧ max⫺3b␧ max
2
, 共7兲 and b seem to approach their limits at large tube radii. This is
expected because CNTs are constructed conceptually by roll-
where ␧ max is the elastic strain limit for CNTs under tensile ing up a single sheet of graphene,16 in the limit of infinite
deformation, ␧ max⫽␤/␣. radius, coefficients a and b become that of a graphene sheet
From our simulation results, ␧ max is 15% and 17% for in specific directions. As shown in Fig. 3 limiting values a 2
zigzag and arm-chair group, respectively. And both groups ⫽28.3 eV/atom, b 2 ⫽60 eV/atom. for zigzag-type 共n, 0兲
have identical maximum force (T max) around 27 N/m, be- tubes, and limiting value a 1 ⫽25.5 eV/atom, b 1
yond which CNTs will collapse. Previous MD ⫽47.2 eV/atom for armchair-type 共n, n兲 tubes. Using the
simulations6,8,33 performed for similar tubes showed only lin- limiting values a 2 , b 2 , and Eq. 共6兲, we get the in-plane
ear elastic deformations to ␧ max . Although at present the ten- stiffness C 2 ⫽56.6– 360␧ eV/atom along the direction per-
sile failure strain for CNTs is still unsettled, various experi- pendicular to vector 共n, 0兲 in graphene. Similarly, using the
mental results,34,35 suggest that it rarely exceeds 10%, which limiting values a 1 , b 1 , and Eq. 共6兲, we get the in-plane
is smaller than our simulation results. stiffness C 1 ⫽51– 283␧ eV/atom along the direction perpen-
One possibility is that while the CNTs simulated here are dicular to vector 共n, n兲 in graphene. Because a 2 ⬎a 1 , 共n, 0兲
defect-free and under ideal loading condition and therefore tubes have higher initial stiffness, but since b 2 ⬎b 1 the stiff-
deformed elastically until failure, defects exist in real CNTs ness of the 共n, 0兲 tubes decreases faster than that of the 共n, n兲
that could substantially reduce their failure strain. Moreover, tubes. Consequently the elastic strain limit of zigzag 共n, 0兲
it is demonstrated by several theoretical studies7,33,36 that at tubes, are about 2% lower than that of the arm-chair 共n, n兲
high strain structural defects of CNTs due to bond rotation or tubes. When ␧ is small, the second term in the expressions of
bond reconstruction8 could occur, result in plastic deforma- C 1 and C 2 can be neglected, the graphene is now treated as
tion. Taking the bond rotation effect into account, Zhang, an isotropic material with an average in-plane stiffness C
Lammert, and Crespi7 reported a maximum elastic limit of ⫽(C 1 ⫹C 2 )/2⫽53.8 eV/atom⫽328 J/m2 . This value is in
about 173 nN/nm per tube radius for group of zigzag tubes good agreement with C⫽342 J/m2 for graphene, reported
共which equals 27.5 N/m per tube circumference, in agree- recently.14

153407-3
BRIEF REPORTS PHYSICAL REVIEW B 66, 153407 共2002兲

Our results suggest that the in-plane stiffness of CNTs is In sum, we have simulated the tensile elastic response of
dependent on diameters of the tubes when they are small. two groups of SWNTs under large-scale deformations by
Yao and Lordi37 predicted that Young’s modulus increases means of the second-generation Brenner potential. It is found
significantly with decreasing tube diameter for smaller diam- that the energy change of the tubes per unit area is a two-
eter, whereas the calculation of Hernandez et al.18 predicted coefficient cubic function of applied tensile strain. Therefore,
that the in-plane stiffness decreases slightly with decreasing the in-plane stiffness C of the tubes, which is linearly related
tube diameter for smaller diameter, and the results of Lu19 to the applied strain, characterizes the nonlinear elastic re-
showed that stiffness is almost completely independent of the sponse of CNTs. C is also dependent on the chirality of the
tube size. These divergences are believed to attributed to tubes when the tensile strain is large. This anisotropy is due
different effects of the curvature on the bonding of tubes to the fact that the molecular structure of CNTs lost their
from various MD methods. In Fig. 3, coefficient b increases
symmetry at large deformations. When strain is less than
slightly with a decrease in tube diameter while coefficient a
0.02, the values of C reduce to a constant, and it is insensi-
is almost unchanged, suggesting that the initial stiffness of
tive to chirality. The diameters of the tubes also affect the
CNTs are less sensitive to the diameter of the tubes, but the
value of C to some extent when radii are small while at large
nonlinear properties depend somewhat on tubes with smaller
diameters. diameters, C approaches that of the planar graphene sheet.

20
*Email address: mtxiao@ntu.edu.sg X. Zhou, J. J. Zhou, and Z. C. Ou-Yang, Phys. Rev. B 62, 13 692

Corresponding author. Email address: askliao@ntu.edu.sg 共2000兲.
1
S. Iijima, Nature 共London兲 354, 56 共1991兲.
21
C. Q. Ru, Phys. Rev. B 62, 9973 共2000兲.
2
S. Iijima and T. Ichihashi, Nature 共London兲 363, 603 共1993兲.
22
B. T. Kelly, Physics of Graphite 共Applied Science, London,
3 1981兲, p. 3.
D. H. Robertson, D. W. Brenner, and J. W. Mintmire, Phys. Rev.
23
B 45, 12 592 共1992兲. M. S. Dresselhaus, G. Dresselhaus, K. Sugihara, I. L. Spain, and
4
J. Tersoff, Phys. Rev. Lett. 61, 2879 共1988兲. H. A. Goldberg, Graphite Fibers and Filaments 共Springer-
5
D. W. Brenner, Phys. Rev. B 42, 9458 共1990兲. Verlag, Berlin, 1988兲, p. 35.
24
6
B. I. Yakobson, C. J. Brabec, and J. Bernholc, Phys. Rev. Lett. 76, C. F. Cornwell and L. T. Wille, Solid State Commun. 101, 555
共1997兲.
2511 共1996兲. 25
7 J. M. Molina, S. S. Savinsky, and N. V. Khokhriakov, J. Chem.
P. H. Zhang, P. E. Lammert, and V. H. Crespi, Phys. Rev. Lett. 81,
Phys. 104, 4652 共1996兲.
5346 共1998兲.
8
26
S. Sawada and N. Hamada, Solid State Commun. 83, 917 共1992兲.
D. Srivastava, M. Menon, and K. Cho, Phys. Rev. Lett. 83, 2973 27
A. Maiti, C. J. Brabec, C. M. Roland, and J. Bernholc, Phys. Rev.
共1999兲. Lett. 73, 2468 共1994兲.
9
M. Menon, E. Richter, and K. R. Subbaswamy, J. Chem. Phys. 28
A. Maiti, C. J. Brabec, and J. Bernholc, Phys. Rev. Lett. 70, 3023
104, 5875 共1996兲. 共1993兲.
10
J. Tersoff, Phys. Rev. B 46, 15 546 共1992兲. 29
C. J. Brabec, A. Maiti, and J. Bernholc, Chem. Phys. Lett. 219,
11
C. Q. Ru, J. Appl. Phys. 87, 7227 共2000兲; Phys. Rev. B 62, 16 962 473 共1994兲; C. J. Brabec, A. Maiti, C. Roland, and J. Bernholc,
共2000兲; J. Mech. Phys. Solids 49, 1265 共2001兲; J. Appl. Phys. ibid. 236, 150 共1995兲.
89, 3426 共2001兲. 30
S. Iijima, C. Brabec, A. Maiti, and J. Bernholc, J. Chem. Phys.
12
S. Govindjee and J. L. Sackman, Solid State Commun. 110, 227 104, 2089 共1996兲.
共1999兲. 31
D. W. Brenner, O. A. Shenderova, J. A. Harrison, S. J. Stuart, B.
13
L. D. Landau and E. M. Lifshitz, Theory of Elasticity 共Pergamon, Ni, and S. B. Sinnott, J. Phys.: Condens. Matter 14, 783 共2002兲.
Oxford, 1986兲, p. 35. 32
The structure of a SWNT, specified by the vector 共n, m兲, that
14
B. I. Yakobson and Ph. Avouris, Top. Appl. Phys. 80, 287 共2001兲. corresponds to a section of the nanotube perpendicular to its
15
Using the relations of flexural rigidity D⫽Y h 3 /12(1⫺ ␯ 2 ) and axis, is denoted as tube 共n, m兲. Details refer to Ref. 16, p. 37.
C⫽Y h, Yakobson 共Ref. 6兲 finds Y ⫽5.5 TPa and representative 33
J. Bernholc, C. Brabec, M. B. Nardelli, A. Maiti, C. Roland, and
thickness h⫽0.066 nm, while Lu 共Ref. 19兲 adopts the interlayer B. I. Yakobson, Appl. Phys. A: Mater. Sci. Process. A67, 39
separation in graphite as representative thickness, h⫽0.34 nm. 共1998兲.
16 34
R. Saito, G. Dresselhaus, and M. S. Dresselhaus, Physical Prop- D. A. Walters, L. M. Ericson, M. J. Casavant, J. Liu, D. T. Col-
erties of Carbon Nanotubes 共Imperial College Press, London, bert, K. A. Smith, and R. E. Smalley, Appl. Phys. Lett. 74, 3803
1998兲, p. 35. 共1999兲.
17 35
D. Sanchez-Portal, E. Artacho, and J. M. Soler, Phys. Rev. B 59, M. F. Yu, B. S. Files, S. Arepalli, and R. S. Ruoff, Phys. Rev. Lett.
12 678 共1999兲. 84, 5552 共2000兲.
18 36
E. Hernandez, C. Goze, P. Bernier, and A. Rubio, Phys. Rev. Lett. M. B. Nardelli, B. I. Yakobson, and J. Bernholc, Phys. Rev. B 57,
80, 4502 共1998兲. R4277 共1998兲.
19
J. P. Lu, Phys. Rev. Lett. 79, 1297 共1997兲. 37
N. Yao and V. Lordi, J. Appl. Phys. 84, 1939 共1998兲.

153407-4

Вам также может понравиться