Physica E 43 (2011) 14191426

Contents lists available at ScienceDirect

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

Flow-induced instability of double-walled carbon nanotubes based on

nonlocal elasticity theory
T.-P. Chang a,n, M.-F. Liu b
a
Department of Construction Engineering, National Kaohsiung First University of Science and Technology, 1 University Road, Yenchao 824, Kaohsiung, Taiwan, ROC
b
Department of Applied Mathematics, I-Shou University, Kaohsiung, Taiwan, ROC

a r t i c l e i n f o abstract

Article history: Instability occurs in double-walled carbon nanotubes when a uid ows through them. This is inves-
Received 21 October 2010 tigated using an elastic shell model based on Donnells shell theory. The dynamic governing equations
Received in revised form of double-walled carbon nanotubes are derived on the basis of nonlocal elasticity theory, and the van
17 March 2011
der Waals interaction between the inner and outer walls is considered. Instability induced by a
Accepted 21 March 2011
pressure-driven steady ow is studied. The numerical computations reveal that as the ow velocity
Available online 29 March 2011
increases, double-walled carbon nanotubes have a destabilizing style to get through multi-bifurcations
of the rst (pitchfork) and second (Hamiltonian Hopf) bifurcations in turn. It can be concluded that the
critical ow velocity of the ow-induced instability is closely correlated to the ratio of the length to the
radius of double-walled carbon nanotubes, the pressure of the uid and the small size effects.
& 2011 Elsevier B.V. All rights reserved.

1. Introduction
Since the landmark paper by Iijima [1], carbon nanotubes
(CNTs) have attracted worldwide attention due to their potential
use in the elds of chemistry, physics, nano-engineering, elec-
trical engineering and materials science. CNTs can be used as
strong, light and high toughness bers for composite structures,
parts of nano-devices and for hydrogen storage [25]. Although
classical or local continuum models, such as beam and shell
models, are practical in analyzing CNTs for large systems,
however, size effects often become remarkable at nanometer scales,
therefore, the modeling of size-dependent phenomena has become a
topic of interest [6,7]. Based on the theory of nonlocal elasticity [8],
the scale effect was claried in elasticity by assuming the stress to
be a functional of the strain eld at every point in the body. In this
sense, the internal size scale could be considered simply as a
material parameter in the constitutive equations. The application
of nonlocal elasticity models in nanomaterials was proposed by
Peddieson et al. [9]. They applied the nonlocal elasticity to formulate
a nonlocal version of EulerBernoulli beam model, and concluded
that nonlocal continuum mechanics could potentially play an
important role in nanotechnology applications. Further applications
of the nonlocal continuum mechanics have been utilized in inves-
tigating the mechanical behavior of CNTs. Sudak [10] investigated
the innitesimal column buckling of multi-walled carbon nanotubes
n
Corresponding author. Tel.: 886 7 6011000x2111; fax: 886 7 6011017.
E-mail address: tpchang@ccms.nkfust.edu.tw (T.-P. Chang).
(MWCNTs) combining not only van der Waals forces but also the
effects of small length scales. His results demonstrated that as the
small length scale gets larger in magnitude the critical axial strain
gets smaller compared to the results with classical continuum
mechanics. Zhang et al. [11] investigated a nonlocal multi-shell
model for the axial buckling of MWCNTs under axial compression.
Their results indicated that both the buckling mode and the length
of tubes have contributions to the inuence of the small scale on the
axial buckling strain. Zhang et al. [12,13] adopted the theory of
nonlocal elasticity to investigate free transverse vibrations of dou-
ble-walled carbon nanotubes. Wang and Hu [14] investigated
exural wave propagation in single-walled carbon nanotubes, their
study focuses on the wave dispersion by considering a model of
traditional Timoshenko beam in conjunction with the theory of
nonlocal elasticity. Wang [15] studied wave propagation in carbon
nanotubes by nonlocal continuum mechanics. They investigated
wave propagation in CNTs with both EulerBernoulli and
Timoshenko beam models by considering the nonlocal elasticity.
Recently, uid ow inside CNTs has become an interesting
subject. The properties of uidity, diffusivity, and viscosity, and
the dynamics of uid in a ne pore have been investigated
[16,17]. The dynamic properties of hydrogen bonding [18], the
effects of walluid interaction [19], the dependence of uid
behavior on the spatial size of CNTs [20], and other issues have
been extensively investigated in the nanoow and microow
elds. The instability problems of uid-lled CNTs are of central
interest in the eld. Wang et al. [21] investigated the elastic
buckling of multi-walled carbon nanotubes (MWCNTs) under
external radial pressure using a multi-walled shell model, and

1386-9477/$ - see front matter & 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.physe.2011.03.015
1420 T.-P. Chang, M.-F. Liu / Physica E 43 (2011) 14191426

the results demonstrated that the multi-walled shell model is in

good agreement with the experiment. Recently, the ow-induced
instability of single-walled carbon nanotubes (SWCNTs) has been
investigated by modeling carbon nanotubes with the Eulerian
beam model [22]. Besides, Donnells shell model for uid-
conveying MWCNTs with the consideration of the van der Waals
interaction has been presented in Ref. [23]. Wang [24] developed
a nonlocal double-elastic beam model to perform the vibration
analysis of double-walled carbon nanotubes conveying uid. Lee
and Chang [25] investigated the vibrations of uid-conveying
double-walled carbon nanotubes by adopting a nonlocal double-
elastic beam model. Wang [26] initiated a theoretical analysis of Fig. 1. The shell model of double-walled carbon nanotubes with moving uid
wave propagation of uid-conveying single-walled carbon nano- inside the inner tube.
tubes based on strain gradient elasticity theory with considera-
tion of both inertia and strain gradients, in which two small-scale
parameters were accounted for. Wang [27] developed a new small scale effect is considered by using the nonlocal elasticity
theoretical model, based on the modied couple stress theory, theory stated before.
for the vibration analysis of uid-conveying microtubes by The governing equations of motion of the model are written as
introducing one internal material length scale parameter. follows [30]:
The purpose of the present paper is to investigate the small
scale effects on the ow-induced instability of double-walled @2
Dr4 w1 rt h 1e0 a2 r2 w1 1e0 a2 r2 c12 w1 w2
carbon nanotubes (DWCNTs) using Donnells shell model. @t 2
2 1 @2 F1
1e0 a2 r p , 4
R1 @x2
2. Nonlocal elasticity theory
4 @2 2
Based on Eringen nonlocal elasticity model [28], the stress at a Dr w2 rt h 1e0 a2 r w2
@t 2
reference point x in a body is considered as a function of strains of all 1 @2 F2
2
the points in the near region. The above assumption is in agreement 1e0 a2 r c21 w2 w1 , 5
R2 @x2
with the atomic theory of lattice dynamics and experimental obser-
vations on phonon dispersion. Eh @2 w1 Eh @2 w2
Consider a homogeneous and isotropic elastic solid, the con- r4 F1  , r4 F2  , 6
R1 @x2 R2 @x2
stitutive equation is
Z  
L In mpR1 =L @ @ 2
rx C0 : a9x0 x9, tex0 dVx0 , 1 p rf 0 U w1 , 7
V mp In mpR1 =L @t @x
where symbols : is the inner product with double contraction, C0 where
is the elastic stiffness matrix of classical isotropic elasticity, r(x) " ! !#
denotes the nonlocal stress tensor at x, and e(x0 ) is the strain @2 1 @2
r2 , r4 r2 r2 , 8
tensor at any point x0 in the body. The kernel function a(9x0 x9,t) @x2 R2i @y
2

is the nonlocal modulus, 9x0 x9 is the Euclidean distance, and

where x is the axial coordinate along the tube length direction, t is
t e0a/l, where e0 is a constant appropriate to each material, a is
the time, Ri is the radius, wi(x,t) represents the transverse
an internal characteristic size (e.g. size of CC bond, lattice
displacements, where i1 for the inner tube, and 2 for the outer
spacing, granular distance), and l is an external characteristic size
tube. E is Youngs modulus, h is the tube thickness, D is the
(crack size, wave length, etc.). The volume integral in Eq. (1) is
exural rigidity, rt is the mass density of the tube, rf is the mass
over the region V occupied by the body. The kernel function is
density of the uid, p is the ow pressure, In is the modied Bessel
given as [28]
p function of order n, prime is the derivative with respect to the
a9x9, t 2pl2 1 K0 xUx=lt, 2 spatial variable, and cij is the van der Waals interaction coefcient
where K0 is the modied Bessel function. that is expressed as [31]
 
By incorporating Eqs. (1) and (2), the following equation can 1001 pes12 13 1120 pes6 7
cij  Eij  Eij Rj , 9
be derived: 3a 4 9a4

1e20 a2 r2 r C0 : e: 3 where e, s are parameters and

Z p=2
It is noted that the scale e0a in Eq. (3) will lead to small scale dy
Elij Rj Ri l , 10
effect on the response of structures in nano-size, furthermore, the 0 1Kij cos2 yl=2
classical (local) elasticity theory is recovered if the scale coef-
cient e0a is set to be zero. where l is an integer, and
4Rj Ri
Kij : 11
3. Small scale effect on coupled shell model for DWCNT Rj Ri 2
conveying uid
The forces per unit length in the axial and circumferential
In Fig. 1, DWCNTs with van der Waals interaction between the directions are denoted as
inner and outer walls are modeled using Donnells cylindrical
shell theory [29]. The uid inside the inner tube is assumed to be 1 @2 Fi @2 Fi 1 @2 Fi
Nxi 2 2
, Nyi 2
, Nxyi  , 12
an ideal incompressible, and the ow is driven by pressure. The Ri @y @x Ri @x@y
 T.-P. Chang, M.-F. Liu / Physica E 43 (2011) 14191426 1421

where Fi(x,y) is the stress function. Denote the axial and circum- where Mxi is the bending moment per unit length. Substituting
ferential displacements of ui(x,y) and ni(x,y), and the stressstrain Eqs. (15) and (16) into the right-hand side of Eq. (6), the
relationship is obtained as differential equations for the stress function Fi are obtained as
  , !2
Eh @ui u @vi mwi px
Nxi  , 13 Ehp2 A1,n t p2 n2
1m2 @x Ri @y Ri F1 cosnysin
L2 R1 L2 R21 L
  , !2  
Eh wi @ui 1 @vi 4Ehp2 A2,n t 4p2 n2 2px
Nyi  m : 14 cosn ysin , 18
1m2 Ri @x Ri @y L2 R1 L2 R21 L

Consider DWCNTs of length L as shown in Fig. 1, assume that the , !2

Ehp2 A3,n t p2 n2 px
inner and outer tubes have the same boundary conditions, which
F2 cosnysin
are simply supported at the two ends, and the displacement L2 R2 L2 R22 L
modes are determined by , !2  
2 2 2
4Ehp A4,n t 4p n 2px
X
2 2 cosnysin : 19
L2 R2 L2 R2 L
w1 Am,n tsinmpx=Lcosny, 15
m1
Furthermore, based on Eqs. (12)(14), (18) and (19), it can be
easily proved that the circumferential displacement ni satises
X
2
w2 Am 2,n tsinmpx=Lcosny: 16 the following equation:
m1 Z 2p
@vi
dy 0, 20
where Am,n(t) is the time-dependent function. m is the mth 0 @y
transverse displacement mode in the axial direction, and n is
the nth circumferential mode. Needless to say, Eqs. (15) and (16)
satisfy the following boundary conditions:
( ! )
@2 wi @2 wi
wi 0 and Mxi D m 0 at x 0,L, 17
@x2 R2 @y
2
i

x 1010
7
Ref. [24]
6 Present study
Natural frequency (Hz)

0
0 500 1000 1500 2000
U (m/s)

Fig. 2. The rst natural frequency as a function of uid ow velocity with

L/(2R2) 10 n 1 and e0a/L 0.8.

25
Ref. [25]
Dimensionless first frequency

Present study
20

15

10

0
0 1 2 3 4 5 6 7
Dimensionless flow velocity vn
Fig. 4. Evolution of the imaginary and real part of the eigenvalues with velocity U
Fig. 3. First dimensionless frequency as a function of dimensionless ow velocity for L/R2 10 and n 2. Solid line: e0a0.0 nm, dashed line: e0a 1.0 nm, dotted
for a clampedclamped DWCNT with e0a/L 0.1. line: e0a 2.0 nm.
1422 T.-P. Chang, M.-F. Liu / Physica E 43 (2011) 14191426

and Eventually, a set of the linear ordinary differential equations for

Nxi 0 at x 0,L: 21 the unknown Am,n(t) is achieved as Eqs. (A.1)(A.4) in Appendix.
Furthermore, Eqs. (A.1)(A.4) are converted into the state space
Based on Galerkins method, the following equation can be with eight rst order differential equations as Eqs. (A.1)(A.8) in
obtained: Appendix.
Z L Z 2p Finally, the characteristic equation of the system of Eq. (A.13)
Xi UWj x, ydydx 0: 22 is described as
0 0

From Eqs. (4) and (5), letting Xi and Wj be detlIC 0, 26

@2 where (I) is an identity matrix and (C) is the coefcient matrix of

4 2 2
X1 Dr w1 rt h 1e0 a2 r w1 1e0 a2 r c12 w1 w2 the right-hand side of Eq. (A.13), and l is the eigenvalue of the
@t 2
system. In general, the eigenvalue is a complex number. Its real
1 @2 F1
1e0 a2 r2 p , 23 part implies an attenuation/amplication factor of the vibration
R1 @x2 amplitude due to the dissipation of the energy from the ow, and
the imaginary part represents the modal frequency of the system.
4 @2 2 2 1 @2 F2
X2 Dr w2 rt h 1e0 a2 r w2 1e0 a2 r c21 w2 w1  , Obviously, the eigenvalue is a function of the ow velocity. As the
@t2 R2 @x2
ow velocity increases, the effective stiffness of the system
24
decreases due to the effects of centrifugal force and/or Coriolis
(
force. When the ow velocity inside the inner tube increases to
cosnysin pLx , j 1,3
Wj x, y
: 25 certain value, the effective stiffness of the system diminishes,
cosnysin 2pL x , j 2,4
namely, the imaginary part goes to zero. At this moment, the

Fig. 5. Evolution of the imaginary and real part of the eigenvalues with velocity U for L/R2 50 and n 2. Solid line: e0a 0.0 nm, dashed line: e0a 1.0 nm, dotted line:
e0a 2.0 nm.
 T.-P. Chang, M.-F. Liu / Physica E 43 (2011) 14191426 1423

instability occurs and the corresponding ow velocity is referred the gap between the two walls is 0.34 nm. The parameters that are
to as the critical velocity. used to calculate the van der Waals interaction coefcient are taken
as e 2.96 mev, a1.42 A, and s 3.407 A [31,32]. The other
modeling parameters of the tubes are Eh360 J/m2, the thickness
4. Numerical computations and discussions h0.34 nm, the mass density of the tube rt 2300 kg/m3, the mass
density of the uid rf 1000 kg/m3, and D0.85 eV[33]. The inner
Two quantitative comparisons are performed to prove the radius and outer radius are taken as R1 11.9 and R2 12.2 nm,
validity of the present approach. In quantitative comparisons, respectively.
rst of all, the numerical results based on the present study are In the numerical examples, two cases, n2 and n 5, are
compared with those obtained from Ref. [24]. All the input data considered. The evolution of the real and imaginary parts of the
must be exactly the same for comparison, therefore, the gap eigenvalues of the system (for n2) as the ow velocity U are
between the two walls is taken as h 0.34 nm, Youngs modulus of presented in Figs. 4 and 5, respectively, for the ratios L/R2 10 and
the DWCNT is E1 TPa, and the mass density is rt 2300 kg/m3 for L/R2 50. The solid lines, dashed lines and dotted lines denote the
CNTs and pf 1000 kg/m3 for the uid. The inner radius and outer results without the scale effect (e0a0.0 nm) and with the scale
radius are taken as R1 3.5 nm and R2 R1 h, separately, and the effects (e0a1.0 nm and e0a2.0 nm) separately. As it is detected
van der Waals interaction coefcients are selected from the from Fig. 4(a) that for L/R2 10, the imaginary parts decreases
expression written in Eq. (3) of Ref. [24]. The boundary conditions as the ow velocity increases, when the ow velocity is up to
of DWCNT are simply supported at both ends. In Fig. 2, the rst Ud 1942 m/s (for e0a0.0 nm) or Ud 1901 m/s (for e0a1.0 nm )
natural frequency is plotted as a function of uid ow velocity or Ud 1837 m/s (for e0a2.0 nm), a pitchfork bifurcation (diver-
with L/(2R2)10, n 1 and e0a/L0.8. As it can be detected from gence instability) occurs, after that the tube loses stability due to
Fig. 2, the numerical results based on the present study are in having a positive real part as shown in Fig. 4(b). This implies that the
good agreements with those obtained from Ref. [24], besides, the
natural frequencies of the nonlocal shell model are less than those
of the nonlocal beam model, and the critical ow velocity of the
nonlocal shell model is 1636 m/s, which is slightly less than
1653 m/s obtained from the nonlocal beam model. The differ-
ences occur because the nonlocal shell model is adopted in the
present study, while the nonlocal beam model is used in Ref. [24];
in addition, the small scale effects are considered for both the
uid and the van der Waals force as stated in Eqs. (4)(7) in the
present nonlocal shell model, however, the small scale effect was
not taken into account for the uid in the nonlocal beam model in
Ref. [24]. Secondly, the numerical results based on the present
study are compared with those obtained from Ref. [25]. The
DWCNT has an inner diameter di of 0.35 nm and an outer
diameter do of 1.75 nm. The gap between the two walls is taken
as h 0.35 nm, the mass density is rt 2300 kg/m3 for CNTs and
rf 1000 kg/m3 for the uid respectively. Youngs modulus of the
DWCNT is E 1 TPa, the van der Waals interaction coefcient is
c 0.069 TPa, and the other geometric and material parameters
used in the calculations are as follows: e0a/L0.10 and L/d0 10.
It should be noted that the boundary conditions of DWCNT are
assumed as clamped at both ends in Ref. [25], therefore, the
transverse displacement mode shapes of Eqs. (15) and (16) must
be changed to satisfy the clampedclamped boundary conditions
for the purpose of comparisons, furthermore, all the rest of the
derivations in the present study must be reformulated in accor-
dance with the change of the transverse displacement mode
shapes for clampedclamped boundary conditions. In Fig. 3,
the rst dimensionless frequency is plotted as a function of
dimensionless ow velocity for a clampedclamped DWCNT with
e0a/L0.1. Once again, the numerical results based on the present
study are in good agreements with those obtained from Ref. [25],
both the natural frequency and the critical ow velocity of the
nonlocal shell model are less than those of the nonlocal beam
model. The discrepancies occur because the nonlocal shell model
is adopted in the present study, while the nonlocal beam model is
used in Ref. [25]; besides, the small scale effects are considered
for both the uid and the van der Waals force in the proposed
nonlocal shell model, on the other hand, the small scale effects are
neglected for both the uid and the van der Waals force in the
nonlocal beam model in Ref. [25].
In the present study, the DWCNTs under consideration are
simply supported at both ends, as shown in Fig. 1, and water that Fig. 6. Evolution of the imaginary and real part of the eigenvalues with velocity U
is driven by pressure goes through the inner tube at a steady for L/R2 10 and n 5. Solid line: e0a0.0 nm, dashed line: e0a 1.0 nm, dotted
velocity. The inner and outer tubes have the same length L and line: e0a 2.0 nm.
1424 T.-P. Chang, M.-F. Liu / Physica E 43 (2011) 14191426

tube absorbs energy from the ow to amplify the vibration. However,
when the ow velocity is continually up to Ur 2626 m/s (for
e0a0.0 nm) or Ur 2586 m/s (for e0a1.0 nm ) or Ur 2476 m/s
(for e0a2.0 nm), the stability is restored, and then the second
instability (utter instability) on a Hamiltonian Hopf bifurcation
comes rapidly at Uf 3514 m/s (for e0a0.0 nm) or Uf 3454 m/s
(for e0a1.0 nm) or Uf3308 m/s (for e0a 2.0 nm). In the case of
L/R2 50, the bifurcating characteristics become more complicated
than in the case of L/R2 10 since the comparative stiffness of the
system is smaller.
For instance, before the utter instability comes into effect, rst
divergence instability and restabilization occur very closely at Ud
1457 m/s and Ur 1462 m/s (for e0a0.0 nm ) or Ud 1436 m/s and
Ur 1442 m/s (for e0a1.0 nm) or Ud 1380 m/s and Ud 1385 m/s
(for e0a2.0 nm). For the case of n5, similar bifurcation of the
system are shown in Figs. 6 and 7, respectively. The details of the
comparisons of the critical ow velocity are listed in Table 1. Based
on the numerical computations presented in Table 1 and Figs. 47, it
can be concluded that the natural frequency of DWCNTs and the
critical ow velocity of the ow-induced instability are closely
correlated to the ratio of the length to the radius of DWCNTs, the
pressure of the uid and the small scale effects. Furthermore, it is
interesting to notice that as the small scale effects are considered,
the natural frequencies and the rst critical ow velocities of
DWCNTs decrease as compared to the results with the classical
(local) continuum mechanics, these phenomena are in good agree-
ments with the results from the nonlocal double-beam model in
Refs. [24,25]. Basically these two papers dealt with the frequencies
and the critical ow velocity of double-walled carbon nanotubes
conveying uid by adopting nonlocal double-elastic beam model.
However, we used nonlocal shell model to cope with the ow-
induced instability of double-walled carbon nanotubes, in fact, that
is the principal goal and originality of the present study.

5. Conclusions

Fluid-conveying double-walled carbon nanotubes (DWCNTs)

Fig. 7. Evolution of the imaginary and real part of the eigenvalues with velocity U
for L/R2 and n 5. Solid line: e0a 0.0 nm, dashed line: e0a 1.0 nm, dotted line:
e0a 2.0 nm.
are modeled based on Donnells shell theory by taking into
account the van der Waals interaction between the inner and
outer walls, in conjunction with the consideration of the small
scale effects by nonlocal elasticity theory. The instability that is
induced by a pressure-driven steady ow is studied. The numer-
ical computations indicate that as the ow velocity increases,
DWNCTs have a way to get through multi-bifurcations of the rst
(pitchfork) and second (Hamiltonian Hopf) bifurcations in turn. It
can be concluded that the natural frequency of DWCNTs and the
critical ow velocity of the ow-induced instability are closely
correlated to the ratio of the length to the radius of DWCNTs, the
pressure of the uid and the small size effects. Furthermore, it is
interesting to notice that as the small scale effects are considered,
the natural frequencies and the rst critical ow velocities of
DWCNTs decrease as compared to the results with the classical
(local) continuum mechanics, therefore, the small scale effects
play an important role on performing the instability analysis in
the uid-conveying DWCNTs. Finally, it should be noted that only
small deformation of the shell is concerned in the present study
since the nonlocal shell model is derived on the basis of the linear
shell theory. In order to deal with the large deformation of the
shell, the nonlinear shell theory should be adopted.

Table 1
Comparisons of critical ow velocity U (m/s). D divergence, R restabilization, F utter.

n U L/R2 10e0 L/R2 10e0 L/R2 10e0 L/R2 50e0 L/R2 50e0 L/R2 50e0
a 0.0 nm a 1.0 nm a 2.0 nm a0.0 nm a 1.0 nm a 2.0 nm

2 D 1942 1901 1837 1457 1436 1380

2 R 2626 2586 2476 1462 1442 1385
2 F 3514 3454 3308 2514 2478 2386
5 D 1425 1312 1091 6842 6198 5187
5 R 1942 1788 1482 8938 8240 6842
5 F 3607 3329 2758 17,870 16,420 13,700
 T.-P. Chang, M.-F. Liu / Physica E 43 (2011) 14191426 1425

Acknowledgment A matrix form of Eqs. (A.5A.12) is written as

2_ 3 2 3
A 1,n t 0 1 0 0 0 0 0 0
This research was partially supported by the National Science 6 B_ t 7 6 C c0 m0
07
2
A t
3
6 1 7 6 21 0 0 C21  r21h m11 0 0 76 1,n 7
6_ 7 6 t
7 B1 t 7
Council in Taiwan through Grant NSC-96-2221-E-327-018-MY2. 6 A 2,n t 7 6 0 0 0 1 0 0 0 0 76
6_ 7 6 0 7 A2,n t 7
6
0 76 7
00 m0
The authors are grateful for this support. 6 B 2 t 7 6 C42 C43 0 0 0  cr 21h m22
6 7 6 t 76
6 B2 t 7
7
6 0
6 A_ 3,n t 7 6 0 0 0 0 1 0 0776 A t 7
  6 3,n 7
6 7 6 7
6 B_ t 7 6 7
c m 0 0 m0 c 0 m0
6r hm 21 3
0 0 0  g23n m33  r21h m33 0 0 0 76 B t 7
6 3 7
6 3 7 6 t 3 t
74 A4,n t 5
6_ 7 6 17
4 A 4,n t 5 4 0 0 0
00 m0
0 0 0 
m0
0
00 m0
 5 B4 t
0 0  cr 21h m44 0 0 0  g24n m44  cr 21h m44 0
Appendix B_ 4 t t t

A:13
A set of the linear ordinary differential equations for the
unknown Am,n(t) is written as follows: where
!
! U 2 rf p2 In pR1 =L c012 m01
U 2 rf p2 In pR1 =L c012 m01 C21  g21n  1 a1 b1  ,
A 1,n t g21n  1 a1 b  A t 2m1 I0n pR1 =L rt h m1
2m1 I0n pR1 =L 1
rt h m1 1,n
4U rf LIn 2pR1 =L c012 m01 4U rf LIn 2pR1 =L
 1 a1 b1 A_ 2,n t A t 0, A:1 C24 1 a2 b1 ,
3m1 I0n pR1 =L rt h m1 3,n 3m1 I0n pR1 =L
! 8U rf LIn 2pR1 =L
2
U 2 rf p2 In 2pR1 =L c00 12 m02 C42 1 a1 b1 ,
A 2,n t g2n  1 a2 b1  A t 3m2 I0n pR1 =L
m2 I0n 2pR1 =L rt h m2 2,n
!
8U rf LIn 2pR1 =L c00 12 m02 U 2 rf p2 In 2pR1 =L c00 12 m02
 1 a2 b1 A_ 1,n t A t 0, C43  g 2
1 a2 b1  ,
3m2 I0n pR1 =L rt h m2 4,n 2n 
m2 I0n 2pR1 =L rt h m2
A:2
 0  rt hpL rf L2 In pR1 =L
m3 2 c0 m0 c0 m0 m01 ,
A 3,n t g3n  21 3 A3,n t 21 3 A1,n t 0, A:3 2 2I0n pR1 =L
m3 rt h m3 rt h m3
 0  rt hpL rf L2 In 2pR1 =L
m4 2 c00 21 m04 c00 21 m04 m02 ,
A 4,n t g4n  A4,n t A t 0, A:4 2 4I0n 2pR1 =L
m4 rt h m4 rt h m4 2,n
rt hpL
where m03 ,
2
p2 2
 2
n 2p n2
a1 e0 a2 , b1 e0 a2 2 , a2 e0 a2 , b2 e0 a2 2 , rt hpL
L R1 L R2 m04 ,
2
c012 c12 1 a1 b1 , c00 12 c12 1 a2 b1 , rt hpL rf L2 In pR1 =L
m1 1 a1 b1 1 a1 b1 ,
c012 c21 1 a1 b2 , c 00
21 c21 1 a2 b2 : 2 2I0n pR1 =L

Eqs. (A.1A.4) can be converted into the state space with eight rt hpL rf L2 In 2pR1 =L
rst order differential equations as follows: m2 1 a2 b1 1 a2 b1 ,
2 4I0n 2pR1 =L
A_ 1,n t B1 t, A:5
rt hpL
! m3 1 a1 b2 ,
2 2 2
U rf p In pR1 =L c012 m01
B_ 1 t  g21n  1 a1 b1  A1,n t
2m1 I0n pR1 =L rt h m1 rt hpL
m4 1 a2 b2 ,
2
4U rf LIn 2pR1 =L c012 m01
1 a2 b1 q2 t A3,n t, A:6 2 !2 , !2 3
3m1 I0n 2pR1 =L rt h m1
2 pL 4 p2 n2 Ehp4 p2 n2 5,
g
1n D 2 2 2 4 2
2
2m1 L R1 R1 L L R1
A_ 2,n t B2 t, A:7
2 !2 , !2 3
!
U 2 rf p2 In 2pR1 =L c00 12 m02 2 pL 4 4p2 n2 16Ehp4 4p2 n2 5,
B_ 2 t  g22n  1 a2 b1  A t g
2n D 2
2 2
m2 I0n 2pR1 =L rt h m2 2,n 2m2 L R1 R21 L4 L2 R1

8U rf LIn pR1 =L c00 12 m02 2 !2 , !2 3

 1 a1 b1 q1 t A t, A:8 2 2 4 2 2
3m2 I0n pR1 =L rt h m2 4,n g23n
1 4D p n Eh
2 2 4
p p n
2 5,
rt h L2 R2 R2 L L2 R2
A_ 3,n t B3 t, A:9
2 !2 , !2 3
 0  2 1 4 4p2 n2 16Ehp4 4p2 n2 5
m3 2 c0 m0 c0 m0 g D 2 2 :
B_ 3 t  g3n  21 3 A3,n t 21 3 A1,n t 0, A:10 4n
rt h L2 R2 R22 L4 L2 R2
m3 rt h m3 rt h m3

A_ 4,n t B4 t, A:11
References
 0 
m4 2 c00 21 m04 c00 21 m04
B_ 4 t  g4n  A t A t: A:12 [1] S. Iijima, Nature 354 (1991) 56.
m4 rt h m4 4,n rt h m4 2,n [2] A. Bachtold, P. Hadley, T. Nakaniski, C. Dekker, Science 294 (2001) 1317.
1426 T.-P. Chang, M.-F. Liu / Physica E 43 (2011) 14191426

[3] M.S. Dresselhaus, P. Avouris, Top. Appl. Phys. 80 (2001) 1. [19] V.P. Sokhan, D. Nicholson, N. Quirke, J. Chem. Phys. 115 (2001) 3878.
[4] Z.C. Tu, Z.C. Ou-Yang, Phys. Rev. B 65 (2002) 233407. [20] R.E. Tuzun, D.W. Noid, B.G. Sumpter, R.C. Merkle, Nanotechnol. 7 (1996)
[5] K.T. Lau, D. Hui, Composites B 33 (2003) 263. 241.
[6] P.E. Sheehan, C.M. Lieber, Science 272 (1996) 1158. [21] C.Y. Wang, C.Q. Ru, A. Mioduchowski, Int. J. Solids Struct. 40 (1996) 3893.
[7] B.I. Yakobson, M.P. Campbell, C.J. Brabec, J. Bernholc, Comput. Mater. Sci. 8 [22] J. Yoon, C.Q. Ru, A. Mioduchowski, Int. J. Solids Struct. 43 (2006) 3337.
(1997) 241. [23] Y. Yan, X.Q. He, L.X. Zhang, J. Appl. Phys. 102 (2007) 044307-1.
[8] A.C. Eringen, Nonlocal Polar Field Models, Academic, New York, 1976. [24] L. Wang, Comput. Mater. Sci. 45 (2009) 584.
[9] J. Peddieson, G.R. Buchanan, R.P. McNitt., Int. J. Eng. Sci. 41 (2003) 305. [25] H.L. Lee, W.J. Chang, J. Phys.: Condens. Matter 21 (2009) 115302.
[10] L.J. Sudak, J. Appl. Phys. 94 (2003) 7281. [26] L. Wang, Comput. Mater. Sci. 49 (2010) 761.
[11] Y.Q. Zhang, G.R. Liu, J.S. Wang, Phys. Rev. B. 70 (2004) 205430. [27] L. Wang, J. Fluids Struct. 26 (2010) 675.
[12] Y.Q. Zhang, G.R. Liu, X. Han, Phys. Lett. A 340 (2005) 258. [28] A.C. Erignen, J. Appl. Phys. 54 (1983) 4703.
[13] Y.Q. Zhang, G.R. Liu, X.Y. Xie, Phys. Rev. B 71 (2005) 195404-1. [29] M. Amabili, F. Pellicano, M.P. Paidoussis, J. Sound Vib. 225 (1999) 655.
[14] L. Wang, H. Hu, Phys. Rev. B 71 (2005) 195412. [30] H.S. Shen, C.L. Zhang, Composite Struct. 92 (2010) 1073.
[15] Q. Wang, J. Appl. Phys. 98 (2005) 124301. [31] X.Q. He, S. Kitipornchai, K.M. Liew, J. Mech. Phys. Solids 53 (2005) 303.
[16] I. Bitsanis, J.J. Magda, M. Tirrell, J. Chem. Phys. 87 (1987) 1733. [32] R. Saito, R. Matsuo, T. Kimura, G. Dresselhaus, M.S. Dresselhaus, Chem. Phys.
[17] S. Paul, A. Chandra, Chem. Phys. Lett. 373 (2003) 87. Lett. 348 (2001) 187.
[18] G. Hummer, J.C. Rasaiah, J.P. Noworyta, Nature 414 (2001) 188. [33] C.Q. Ru, J. Appl. Phys. 87 (2000) 7227.