Professional Documents
Culture Documents
A Collocation Mixed Finite Element Method For The Analysis of Flexoelectric Solids
A Collocation Mixed Finite Element Method For The Analysis of Flexoelectric Solids
net/publication/349164616
CITATIONS READS
14 328
5 authors, including:
Qian Deng
Huazhong University of Science and Technology
63 PUBLICATIONS 1,401 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Xinpeng Tian on 22 March 2021.
a r t i c l e i n f o a b s t r a c t
Article history: A collocation mixed finite element method (MFEM) for direct and converse flexoelectricity in piezoelec-
Received 23 September 2020 tric materials is developed for 2D problems. The size-effect phenomenon in micro/nano structures is con-
Received in revised form 14 December 2020 sidered by the strain- and electric intensity vector-gradient effects. C0 continuous finite element method
Accepted 27 January 2021
is inadequate to treat flexoelectricity problems involving the size-effect. To this end, the MFEM with
Available online 09 February 2021
Lagrangian multipliers to treat these solids has been reported recently. With existing MFEM, the compu-
tational efficiency is low due to the additional nodal degrees of freedom (DOFs) for the Lagrangian mul-
Keywords:
tipliers. In this study, a new collocation MFEM is proposed, in which the number of the DOFs, when
Flexoelectricity
collocation MFEM
compared to the traditional Lagrangian approach, can be reduced. At the same time, the kinematic con-
Gradients of strain and electric intensity straints between the displacement and strain are guaranteed. These kinematic constraints are satisfied by
vector the collocation method at some internal points in the finite elements. The present collocation MFEM can
Higher efficiency be used to solve flexoelectricity problems with higher efficiency. Its accuracy is verified by comparing the
Geometric dependence numerical results with available analytical solutions for the bending of a cantilever beam and the com-
pression of a truncated pyramid, respectively. The results indicate that flexoelctricity is strongly related
to the geometry of the physical problem. It is shown that flexoelectricity increases significantly with the
decrease of the sample size. The same occurs when, for the beam problem, the ratio of the length to depth
dimensions increases; similarly, for the truncated pyramid problem, when the ratio of the width of the
bottom and top surfaces increases.
Ó 2021 Elsevier Ltd. All rights reserved.
1. Introduction tricity was only performed a few decades later in the 2000s, see,
e.g. Ma and Cross (2001), Ma and Cross (2006). Perspectives on
Recent progress in microelectronics is, in large part, driven by the future directions for research on flexoelectricity are given in
continuous miniaturization of devices and the use of nanotech- some review papers (Krichen and Sharma, 2016; Wang et al.,
nologies in which size-dependent effects cannot be ignored. The 2019; Zhuang et al., 2020; Deng et al., 2020).
direct flexoelectric effect describes the coupling between the strain To utilize the flexoelectric effect, the strain gradients have to be
gradients and the electric polarization (Kogan, 1964; Meyer, 1969; relatively large; they are therefore more easily generated in nano-
Sharma et al., 2006). A non-uniform strain, i.e. the presence of scale structures. The dimensions of such solids are of the same
strain gradients, may potentially break the inversion symmetry order of the material length scale parameter used in generalized
thereby inducing electric polarization even in centrosymmetric theories of continua. As has been shown experimentally, the stiff-
crystals (Tagantsev, 1986; Tagantsev et al., 2009; Maranganti ness of such a structure increases with a decrease in its size. The
et al., 2006). Flexoelectricity is found to be a universal electrome- presence of strain gradients can also be realized by differences in
chanical coupling that exists in all dielectric materials even with material properties at the interfaces of these materials even under
a centrosymmetric crystal structure (Yudin and Tagantsev, 2013; a uniform stress (Deng et al., 2014a, 2014b).
Deng et al., 2014a, 2014b, 2020). The existence of flexoelectric In order to perform good design of devices with flexoelectric
effect in solids has been observed experimentally as first reported properties, it is necessary to analyze general boundary value prob-
by Harris (1965). However, a systematic measurement of flexoelec- lems (BVPs) of these components. It is well-known that classical
continuum mechanics neglect the influence of the material
⇑ Corresponding author. microstructure and the results are size-independent. To overcome
E-mail address: jan.sladek@savba.sk (J. Sladek). intrinsic limitations of classical elasticity, atomistic models have
https://doi.org/10.1016/j.ijsolstr.2021.01.031
0020-7683/Ó 2021 Elsevier Ltd. All rights reserved.
X. Tian, J. Sladek, V. Sladek et al. International Journal of Solids and Structures 217-218 (2021) 27–39
been developed to describe the micro-scale phenomena in materi- dient is enforced by Lagrangian multipliers. In the formulation, the
als. Extremely high requirements on computer memory in these displacement gradient and Lagrangian multipliers are set as addi-
models have led to the development of multiscale approaches tional nodal DOFs to displacements, electric potential and polariza-
where atomistic and continuum subdomains are bridged. Due to tion (Amanatidou and Aravas, 2002; Mao et al., 2016). Therefore,
intrinsic heterogeneities in the atomistic-continuum coupling, extra DOFs are introduced thereby leading to lower computational
some physically unrealistic phenomena have been observed espe- efficiency. For example, there are 87 DOFs for a 2D quadrilateral
cially for time-dependent problems. Another approach to treat this element in formulation by Mao et al. (2016); and the correspond-
problem is based on the phenomenological theory of flexoelectric- ing number is 47 in Deng et al’s formulation. This is a significant
ity within the generalized thermodynamics of a dielectric. This drawback for analysing large scale models. A new finite element
involves introducing a flexoelectric contribution into the total free method with greater efficiency would therefore be highly desir-
energy (Tagantsev, 1986). The flexoelectricity can be realized as able; this is the subject of the present paper.
the direct flexoelectric effect (Sharma et al., 2006; Gharbi et al., In this study, a collocation MFEM for direct and converse flex-
2009), and converse flexoelectricity. The former refers to the linear oelectricity in piezoelectric materials is developed. The size-effect
coupling of electric fields and strain gradients, and it is frequently is considered by including the strain gradients, electric field gra-
investigated in the literature. The converse flexoelectricity is the dients, and their coupling in the constitutive equations of the
coupling between the stress and applied electric intensity vector piezoelectric materials. The C0 continuous approximation is
gradients (Yang et al., 2004). Maugin (1980) showed the duality applied independently for displacements and strains. The kine-
between the theory of electric field gradient and the theory of flex- matic constraints between strains and displacements are satisfied
oelectricity. Hu and Shen (2009) and Shen and Hu (2010) have by collocation method at judiciously chosen internal points of the
extended the general flexoelectric theory by the surface effects elements (Dong and Atluri, 2011; Bishay et al., 2012). In contrast,
for nano-sized elastic dielectrics. They have developed the varia- the Lagrange multipliers are used to enforce these kinematic con-
tional principle for these problems. Recently, flexoelectricity in straints in the traditional MFEM (Amanatidou and Aravas, 2002;
biological membranes and soft electrets has attracted the attention Mao et al., 2016). The electric potential and electric intensity vec-
of several researchers. The corresponding flexoelectric membrane tor are approximated by C0 continuity in the same manner. The
theory and nonlinear flexoelectric theory in electret soft materials corresponding constraint between the electric intensity vector
have been proposed (Mohammadi et al., 2014; Deng et al., 2014a, and the electric potential are also satisfied by the same colloca-
2014b; Rahmati et al., 2019). tion method. No extra DOFs are introduced in this collocation
For nano-sized flexoelectric structures it is necessary to apply MFEM. Thus, there are just 12 DOFs for a 2D quadrilateral ele-
gradient theory in which the governing equations are partial differ- ment (eight displacements and four potentials). In the case of iso-
ential equations (PDE) of the fourth order. To solve flexoelectric geometric analysis, the non-uniform rational B-spline functions
problems with strain- and electric intensity vector-gradient effects, with higher order continuity can be employed; no extra DOFs
several numerical methods have been developed, namely, moving are therefore needed (Thai et al., 2018), like in the present
least square (MLS) (Sladek et al., 2013), meshfree formuation approach.
method (Abdollahi et al., 2014, 2015), isogeometric analysis (IGA) A MFEM code based on this formulation is developed in this
(Thai et al., 2018; Nguyen et al., 2018, 2019; Liu et al., 2019), hier- study and its veracity is demonstrated with two example prob-
archical B-spline method (Codony et al., 2019), and finite element lems, The first is a simple cantilever beam problem for which the
method (FEM) (Yvonnet and Liu, 2017; Sladek et al., 2018; analytical solution is available. The second example is a truncated
Amanatidou and Aravas, 2002; Mao et al., 2016; Deng et al., pyramid under a compressive load. Before discussing these numer-
2017, 2018). ical examples, a review of the governing equations in flexoelectric-
Among the numerical methods mentioned above, the FEM has ity is perhaps in order. This will be followed by a presentation of
been well established as a powerful computational tool for analys- the formulation of the proposed collocation MFEM.
ing general BVPs with complex geometries. However, the conven-
tional FEM cannot be used to study flexoelectricity due to the 2. Direct and converse flexoelectricity
strain- and electric intensity vector-gradient effects, where second
derivatives of the primary fields (displacement and electric poten- The electric enthalphy density for piezoelectric solids can be
tial) are required. Two modified FEMs to resolve this problem have written as (Maranganti et al., 2006; Hu and Shen, 2009)
been reported in the literature. The first approach involves the use
of C1 continuity elements. For example, Yvonnet and Liu (2017)
1 1 1
H¼ cijkl eij ekl aij Ei Ej ekji eij Ek þ g jklmni gjkl gmni f ijkl Ei gjkl
have applied C1 Argyris triangular elements for soft flexoelectric 2 2 2
solids at finite strains. Sladek et al. (2018) have developed conform- 1
bklij eij Ek;l hijkl Ei;j Ek;l ð1Þ
ing elements with C1 continuity, where each node has 9 degrees of 2
freedom (6 mechanical quantities, electric potential, and two where symbols a and c are used for the second-order permittiv-
potential gradients) for 2D flexoelectric problems. The C1 continu- ity and the fourth-order elastic constant tensors, respectively. The
ous element is established by using higher order shape functions. piezoelectric coefficient is denoted by e and f is the direct flexo-
It is, however, difficult to develop C1 elements for 3D problems. electric coefficient. The tensor g is used for higher order elastic
Another way to resolve the gradient problem is to use mixed finite coefficients representing the strain-gradient elasticity. The sym-
element methods (MFEMs) which are relatively more convenient to bols b and h are used for the converse flexoelectric coefficients
develop. Following the works of Amanatidou and Aravas (2002) for and higher-order electric parameters, respectively. The strain ten-
the MFEM in gradient theory of elasticity, Mao et al. (2016) con- sor eij and the electric field vector Ej are defined as (Parton and
structed a MFEM formulation for flexoelectricity with extra nodal Kudryavtev, 1988)
degrees-of-freedom (DOFs) for polarizations and developed a 2D
1
element to solve general BVPs. Deng et al. (2017), Deng et al. eij ¼ ui;j þ uj;i ; Ej ¼ /;j ð2Þ
(2018) have also developed a MFEM with strain gradient and flex- 2
oelectricity and extended it to 3D flexoelectricity problems. where ui and / are displacements and electric potential,
In traditional MFEM (Mao et al., 2016; Deng et al., 2017, 2018), respectively.
the kinematic relationship between displacement field and its gra- The strain-gradient tensor g is given by
28
X. Tian, J. Sladek, V. Sladek et al. International Journal of Solids and Structures 217-218 (2021) 27–39
29
X. Tian, J. Sladek, V. Sladek et al. International Journal of Solids and Structures 217-218 (2021) 27–39
2 3 2 3
Q 11 b1 b1 0 2 3 set as independent variables with C0 continuity. However, the
6Q 7 6 0 e11
6 31 7 6 0 b3 7
76 strains should also satisfy the geometric relationship with the
6 7¼6 74 e33 7
5 displacements. To this end, the strain values from these two con-
4 Q 13 5 4 0 0 b2 5
2e13 siderations are made to be equal at the Gauss points in the ele-
Q 33 b1 b1 0
2 32 3 2 3 ð14Þ ments (Dong and Atluri, 2011; Bishay et al., 2012). The
a1 0 0 0 E1;1 2 3 E1;1 kinematic constraints between strains and displacements are
60 e11
6 a2 0 0 7 6E 7
76 3;1 7 6
6E 7
e33 7 6 3;1 7 thus satisfied by the collocation method at Gauss points. This
þ q2 6 76 7 ¼ UT 4 5 þ q2 H 6 7
40 0 a1 0 54 E1;3 5 4 E1;3 5 is similarly applied to the electric potential and electric intensity
2e13 vector. By using this collocation scheme, each node has only
0 0 0 a2 E3;3 E3;3
three DOFs (two displacements and one electric potential) which
Recently, the authors have derived the governing equations for is significantly smaller in number as compared to the traditional
piezoelectric solid with direct and converse flexoelectric effects MFEM via Lagrangian multipliers (Mao et al., 2016; Deng et al.,
(Sladek et al., 2018) 2017).
rij;j ðxÞ sijk;jk ðxÞ ¼ 0 The variational formulation of the FEM in gradient theory can
be derived from the principle of virtual work as (Hu and Shen
Di;i ðxÞ Q ij;ji ðxÞ ¼ 0 ð15Þ 2009)
Z
The essential and natural boundary conditions (b.c.) can be
prescribed:
rij deij þ sijk dgijk þ Dk dEk þ Q ij dEi;j dX
V
Z
Z
Z
Z
¼ t i dui dC þ Ri dsi dC þ S d/dC þ Z dpdC ð23Þ
1) Essential b.c.: ui ðxÞ ¼ ui ðxÞon Cu , Cu C Ct CR CS CZ
si ðxÞ ¼ s i on Cs ; Cs C ð16Þ
Where t i , Ri , S, and Z are prescribed values corresponding to the
external work on the right hand side of (23).
/ðxÞ ¼ /ðxÞ on C/ ; C/ C The mechnanical displacements and electric potential in each
element as shown in Fig. 1 are expressed in terms of nodal values
@/ and shape functions
pðxÞ ¼ ¼ pðxÞ on Cp ; Cp C
@n !
ðeaÞ
u1 ðxÞ P4 u1
¼ N a ðn1 ; n2 Þ
u ðxÞ
3 V e
a¼1
u3
ðeaÞ
!
2) Natural b.c.:t i ðxÞ ¼ t i ðxÞ on Ct Ct [ Cu ¼ C; Ct \ Cu ¼ £ P4 u1
ðeaÞ
ð24Þ
ðeaÞ a ðeaÞ
¼ a¼1 q N ðn1 ; n2 Þ; q :¼ ðeaÞ
u3
Ri ðxÞ ¼ Ri ðxÞ on CR ; CR [ Cs ¼ C; CR \ Cs ¼ £ ð17Þ P4 ðeaÞ a
/ðxÞjV e ¼ a¼1 / N ðn1 ; n2 Þ
SðxÞ ¼ SðxÞ on CS ; CS [ C/ ¼ C; CS \ C/ ¼ £
where qðeaÞ and /ðeaÞ are nodal displacements and electric
potential, respectively.
ZðxÞ ¼ Z ðxÞ on CZ ; CZ [ Cp ¼ C; CZ \ Cp ¼ £ The gradient in global coordinates can be expressed within the
Where finite element V e in terms of deivatives in the local (intrinsic) coor-
dinates as
@ui @/
si :¼ ; p :¼ ; Ri :¼ nk nj sijk ; Z :¼ ni nj Q ij ð18Þ
@n @n @=@x1 @=@n1 1
¼ ½Y e ; ½Y e ¼ ½J e
and the traction vector, and the electric charge are defined as @=@x 3 Ve @=@n2
@ qi X
t i :¼ nj rij sijk;k þ k qi ðxc Þ kdðx xc Þ ð19Þ !
@p c @x1 =@n1 @x3 =@n1 X ðecÞ
x1 Nc;1
ðecÞ
x3 Nc;1
½J e ¼ ¼ ð25Þ
@x1 =@n2 @x =@n Ve
ðecÞ
x1 Nc;2
ðecÞ
x3 Nc;2
@a X
3 2
c
S :¼ nk Dk Q kj;j þ k aðxc Þ kdðx xc Þ ð20Þ
@p c Hence
P ðeaÞ ea
with qi :¼ nk pj sijk ; a :¼ ni pj Q ij , d(x) being the Dirac delta func- @f
@xi e
¼ a f bi ðn1 ; n2 Þ;
V
tion and pi is the Cartesian component of the unit tangent vector ea
b1 ðnÞ ¼ Y e11 ðnÞNa;1 ðnÞ þ Y e12 ðnÞNa;2 ðnÞ; ð26Þ
on C.
ea
The jump at a corner (xc) on the oriented boundary contour C is b3 ðnÞ ¼ Y e21 ðnÞNa;1 ðnÞ þ Y e22 ðnÞNa;2 ðnÞ
defined as
The expression for the approximation of the electric intensity
k qi ðxc Þ k :¼ qi ðxc þ 0Þ qi ðxc 0Þ ð21Þ vector within V e , from the Maxwell equation, is given by
!
k aðxc Þ k :¼ aðxc þ 0Þ aðxc 0Þ ð22Þ E1 /;1
fEgjV e ¼ ¼
E3 V e /;3
Ve
! ð27Þ
3. The mixed finite element using collocation method X n ea o ðeaÞ n ea o ea
b1 ðnÞ
¼ B/ ðnÞ / B/ ðnÞ ¼ ea
a b3 ðnÞ
To solve the strain- and electric intensity vector-gradient
effects in flexoelectric materials, a collocation MFEM is devel- The electric intensity vector is treated as independent variable;
oped. In the present scheme, the displacements and strains are thus,
30
X. Tian, J. Sladek, V. Sladek et al. International Journal of Solids and Structures 217-218 (2021) 27–39
Fig. 1. Global Cartesian coordinates x1 x2 x3 , curvilinear coordinates n1 n2 , for the 4-node quadrilateral finite element.
0 1 0 1
b1i b1i Substituting Eqs. (31) into (28), one can obtain
B 2C B 2C X
B bi C B C
EIn ¼ ð 1 n1 n2 n1 n2 Þ B C ¼ fPðnÞgT B bi C EIn
i ðxÞ ¼ fLðnÞgT xea /ðeaÞ ; fLðnÞgT :¼ fPðnÞgT ½A1 ð32Þ
i ðxÞ ð28Þ
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} B 3C B 3C i
Ve
Ve @ bi A @ bi A a
fPðnÞgT
b4i b4i or
! !
or n o EIn X fLðnÞgT xea
1 ðxÞ
/ðeaÞ
1
n oT EIn ðxÞ ¼ ¼
Ve E ðxÞ
In T
fLðnÞg xea
EIn ðxÞ ¼ EIn a
e
1 ðxÞ EIn
3 ðxÞ V e
3 V 3
V e X mea
1 ðnÞ
0 1 1 ¼ /ðeaÞ
b1 b13 a 3 ðnÞ
mea
B 2 C ð29Þ
B b1 b23 C X
¼ ð 1 n1 n2 n1 n2 ÞB
B 3
C
3C
T ea
@ b1 b3 A i ðnÞ :¼ fLðnÞg
mea xea
i ¼ fLðnÞgc bi ðnc Þ ð33Þ
c
b41 b43
and
with the coefficients bci being determined from equating the n oT X
two approximations at collocation points xc , selected as the Gauss
EIn ðxÞ ¼ EIn
1 ðxÞ EIn ¼
3 ðxÞ V e 1 ðnÞ
ð mea 3 ðnÞ Þ/
mea ðeaÞ
EIn ð34Þ
i ðx Þ ¼ Ei ðx Þ. Thus,
c c
0 1 0 10 1 1
EIn
i ðx Þ
1
1 n11 n12 n11 n12 bi Since @x@1 ¼ Y e11 @n@1 þ Y e12 @n@2 ; @x@3 ¼ Y e21 @n@1 þ Y e22 @n@2 , the electric
Ve Ve
B In 2 C B CB C
B Ei ðx Þ C B 1 n21 n22 n21 n22 C B b2i C intensity vector gradient can be derived as
B C B CB C
B In 3 C ¼ B 3 3 3 3 CB 3 C
@ Ei ðx Þ A @ 1 n1 n2 n1 n2 A @ bi A
@EIn
e i ðxÞ @ @
EIn
i ðx Þ
4
1 n41 n41 n41 n42 b4i ¼ Y e11 þ Y e12 fPðnÞgT fbi g
V
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl
ffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} @x1 @n1 @n2
½A ð30Þ Ve |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
0 1 fP 1 ðnÞgT
0 1 Bea 1
Ei ðx1 Þ /i ðn Þ X
B ea 2 C ¼ fS1 ðnÞgT xea /ðeaÞ
B E ðx2 Þ C XB C i
B i C B B/i ðn Þ C ðeaÞ ea c ea c a
¼ B C ¼ B ea 3 C/ ; B/i ðn Þ ¼ bi ðn Þ;
@ Ei ðx3 Þ A B C
a @ B/i ðn Þ A
E ðx4 Þ Bea 4 @EIn
i ðxÞ @ @
Ve /i ðn Þ
i
¼ Y e21 þ Y e22 fPðnÞgT fbi g
@x3 @n1 @n2
Hence, Ve |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
0 1 0 1 fP 3 ðnÞgT
b1i
ea
bi ðn1 Þ X
B 2C ¼ fS3 ðnÞg T
xea /ðeaÞ
B bi C XBB bea 2 C
C
i
0 1 oT 1
0n
EIn 01
0n o 1 1;1 ðxÞ 11 ðnÞ
lea Bea 1
B In C 0 1B eij ðn Þ C
EIn ðxÞ B E ðxÞ C X B l31 ðnÞ C
B a 1
Bn C
B ea 2 oT C
ea
B ;1 C B 3;1 C C ðeaÞ ij
@ n o A ¼ B In C ¼ B ea C/ ð37Þ B 2C X B B ðn Þ C
B E ðxÞ C @ l ðnÞ A B ij C
a B eij C
EIn ðxÞ @ 1;3 A aij :¼ B C 1
B a3 C ¼ ½A Bn C
a 13
;3 e B o T C
V
EIn ðxÞ lea
33 ðnÞ @ ij A a B B ðn Þea 3
C
3;3 Ve B eij C
4
aij @ n oT A
A similar derivation can be repeated for the approximation of ea 4
Beij ðn Þ
the strain tensor. The expression for the strain tensor obtained 0n oT n o1 ð44Þ
ðeaÞ
from the geometric relationship within V e is given by Bea ðn1 Þ qu
0 1 B eij C
Bn C
ea
b1 ðnÞ 0 B ea 2 oT n ðeaÞ o C
X B C XB Beij ðn Þ qu C Xn o
fegjV e ¼ Bea ðeaÞ
; Bea
ea B C
e ðnÞ qu e ðnÞ ¼ @ b3 ðnÞ A; 1 1
0 qðeaÞ ¼ ½ A Bn o n o C ¼ ½ A cea
1 ea ea
u
B
a B Bea ðn3 Þ
T
ðeaÞ C
ij
a
b ðnÞ 1
b ðnÞ ð38Þ q C a
! 2 3 2 1 B eij u
C
ðeaÞ @n oT n oA
u1 ðeaÞ
qðeaÞ
u ¼ Bea 4
eij ðn Þ qu
ðeaÞ
u3
in which
The strain tensor is also set as independent variable; thus, 0n oT n o1
ðeaÞ
0 Bea ðn1 Þ qu
11 0 11 B e ij C
aij aij Bn C
B 2C B 2C B ea 2 oT n ðeaÞ o C
B aij C Ba C n o B Beij ðn Þ qu C
B C
^eIn
ij ðxÞ ¼ ð 1 n1 n1 n2 ÞB C ¼ fPðnÞgT B ij C; cea :¼ B n C ð45Þ
B ea 3 oT n ðeaÞ o C
n2 B a3 C B a3 C ij
Ve @ ij A @ ij A B Beij ðn Þ qu C
ð39Þ B C
a4ij a4ij @n oT n oA
( ea 4
Beij ðn Þ
ðeaÞ
qu
eInij ; if i ¼ j
^eIn
ij :¼
ij ; if i–j
2eIn Then,
0 1
a1ij
or B 2C
Ba C Xn o Xn o
T B ij C
eIn11 ðxÞ eIn33 ðxÞ 2eIn13 ðxÞ V e ^eIn
ij ðxÞ e ¼ fPðnÞg B 3 C ¼ fPðnÞg ½A
T 1
cea ¼ fLðnÞgT cea
@ aij A |fflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflffl}
ij ij
V
0 1 1 a a
a11 a133 a113 a4ij
fLðnÞgT
B 2 C
B a11 a233 a213 C ð40Þ
¼ ð 1 n1 n2 n1 n2 ÞB B 3
C ð46Þ
@ a11 a333 a313 C
A or
a411 a433 a413 0 1 0 1
eIn11 ðxÞ
T
fLðnÞg cea
n o XB 11
C
where the coefficents acij (c ¼ 1; 2; 3; 4) are determined from the ^eIn ðxÞ B In C
¼ @ e33 ðxÞ A ¼ B fLðnÞgT cea C
@ 33 A
ð47Þ
Ve
collocation of ^eij ðxc Þ ¼ ^eij ðxc Þ at the Gauss points xc in the finite ele-
In e a
2eIn 13 ðxÞ
T
V fLðnÞg cea
13
ment V e with Since
X n ea c oT 0n oT 1
eij ; if i ¼ j
^eij :¼ ; ^eij ðxc ÞV e ¼ Beij ðn Þ quðeaÞ ð41Þ Bea ðn1 Þ
B eij C
2eij ; if i–j Bn C
B ea 2 oT C
a
n o n o n o B Beij ðn Þ C
where B C
fLðnÞg
T
cea ¼ fLðnÞgT bea qðeaÞ ea
; bij :¼ B n C
ij ij u
B ea 3 oT C
Bea c T ea B Beij ðn Þ C
e11 ðn Þ ¼ 0 ;
b1 ðnc Þ B C
T T ð42Þ @n oT A
Bea c
e33 ðn Þ ¼ 0 bea c
3 ðn Þ
; Bea c
e13 ðn Þ ¼ bea c
3 ðn Þ
ea
b1 ðnc Þ ea 4
Beij ðn Þ
Thus, ð48Þ
0 1 0 10
^In 11 The following is finally obtained,
ij ðx Þ
1
e 1 n11 n12 n11 n2 aij 1
B C B CB C 0 In 1 0 1
B In 2 C
B ^eij ðx Þ C B 2 2 CB 2 C ^e11 ðxÞ fLðnÞgT bea
B 1 n1 n2 n1 n2 C B aij C
2 2
B C B C XB 11
C
B C ¼ BB
CB C eIn ðxÞ V e ¼ B In C
@ ^e33 ðxÞ A ¼
B fLðnÞgT bea C
B ^In 3 C 3 3 CB 3 C @ 33 A
e
B ij ðx Þ C B 1 n 3
n 3
n n CBa C
@ A @ 1 2 1 2 A @ ij A
2^e13 ðxÞ V e
In a T
fLðnÞg bea
0 1
13
ð49Þ
^eIn 4
1 n1 n2 n1 n2 4 4 4
a4ij
ij ðx Þ
4 T
Ve |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl
ffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} Mea 11 ðnÞ
½A XB C X ea
qðeaÞ ¼ B Mea ðnÞ T C qðeaÞ ¼ M quðeaÞ
0n oT 1 ð43Þ u @ 33 A u
a a
Bea ðn1 Þ Mea
T
B eij C 0 1 13 ðnÞ
B C ^eij ðx1 Þ
Bn oT C
B Bea ðn2 Þ C B C where
XB eij C B ^eij ðx2 Þ C
B C ðeaÞ B C ea
¼ Bn C q ¼B C Mea
T
:¼ fLðnÞgT bea
B ea 3 oT C u B ^eij ðx3 Þ C 11 ðnÞ 11 ¼ k1 ðnÞ 0
a B
B ðn Þ C @ A
B eij C
B C ^e ðx4
Þ e
@n oT A Mea
33 ðnÞ
T
:¼ fLðnÞgT bea
33 ¼ 0 kea
3 ðnÞ
ij V
ea 4
Beij ðn Þ
T ea
Mea
13 ðnÞ :¼ fLðnÞgT bea
13 ¼ k3 ðnÞ kea
1 ðnÞ
ð50Þ
Hence
32
X. Tian, J. Sladek, V. Sladek et al. International Journal of Solids and Structures 217-218 (2021) 27–39
0 1 Ceg CeR
ð ea1 ðnÞ
v 0Þ
0 n In o 1 B C
ð0 3 ðnÞ Þ
v
ea
B C P P R 1 R 1 n eg cg T
B
^e ðxÞ
;1 C
XBBðv 3 ðnÞ
ea
1 ðnÞ Þ C
ea
v
C
C eg a 1 1 fm ðnÞg ½K Meg a ðnÞ þ ½F Weg a ðnÞ
gIn ðxÞ V e ¼ @n o A ¼ B
B C
n o
^eIn ðxÞ a B ð eaj ðnÞ 0 Þ C þfleg cg ðnÞgT ½UT Meg a ðnÞ
ðe aÞ
qu g
B
1
C ð57Þ o
;3 Ve @ ð0 ea
ðnÞ
j Þ A
3 þ fmeg cg ðnÞgT ½PT fmeg a ðnÞg þ q2 fleg cg ðnÞgT ½Hfleg a ðnÞg /ðeg aÞ det½J eg dn1 dn2
ðj ea
ðnÞ ea
ðnÞ
j Þ P R1
X 3 1
¼ e
1
e
S ðnÞNcg ðnÞh g ðnÞdnv
g
qðeaÞ
u ¼ W ea
qðeaÞ
u P
Ceg CeS
R1
Z ðnÞnj ðnÞbj g g ðnÞh g ðnÞdnv
e c e
a þ eg
1
Ceg CeZ
The above approximations further results in the following ð61Þ
expression being obtained,
X T The system of algebraic equations (60) - (61) can be applied for
rij deij V e ¼
T
dqðecÞ
u ½Mec ðnÞ ½C M ea ðnÞ quðeaÞ ð½Kfmea ðnÞg þ ½Uflea ðnÞgÞ/ðeaÞ a general BVP described by the gradient theory with direct and
c;a
converse flexoelectricity.
X
sijk dgijk e ¼
T T
½FT fmea ðnÞg/ðeaÞ þ l ½G Wea ðnÞ qðeaÞ
2
dqðecÞ
u Wec ðnÞ u
V
c;a
4. Numerical examples
X
ðecÞ ðeaÞ
qðeaÞ
T
Dk dEk jV e ¼ d/ fm ðnÞg
ec
½P fm ðnÞg/
T ea
þ ½K M ðnÞ þ ½F W ðnÞ
ea ea
c;a
u A computer code for collocation MFEM based on the formula-
tions presented in the previous sections has been developed in this
X ðecÞ study. To demonstrate its veracity, two example problems are pre-
Q ij dEi;j V e ¼ d/ flec ðnÞg ½U Mea ðnÞ qðeaÞ þ q2 ½Hflea ðnÞg/ðeaÞ
T T
u sented here, namely, (a) a simple cantilever beam with a transverse
c;a
end load, and (b) a truncated pyramid under compression.
ð58Þ
33
X. Tian, J. Sladek, V. Sladek et al. International Journal of Solids and Structures 217-218 (2021) 27–39
Fig. 3. Relative errors and convergence rates for the present collocation MFEM. Fig. 5. Deflections of the cantilever beam with different flexoelectric coefficients.
Fig. 7. Electric field of the cantilever beam with different flexoelectric coefficients. Fig. 9. Electric field distributions of the cantilever beam with five various
thicknesses.
Fig. 12. Distributions of mechanical fields and flexoelectric responses under uniform load on the top surface and rigid support on the bottom surface, the bottom surface is
grounded and an electrode is attached on the top surface. (a) longitudinal displacement u2 (m), (b) longitudinal strain e22, (c) the electric potential /(V), (d) the electric field E2
(V/m).
37
X. Tian, J. Sladek, V. Sladek et al. International Journal of Solids and Structures 217-218 (2021) 27–39
Acknowledgements
References
Abdollahi, A., Millan, D., Peco, C., Arroyo, M., Arias, I., 2015. Revisiting pyramid
compression to quantify flexoelectricity: A three-dimensional simulation study.
Phys. Rev. B 91, 104103.
Abdollahi, A., Peco, C., Millan, D., Arroyo, M., Arias, I., 2014. Computational
Fig. 14. Effective electric field Eeff of the truncated pyramid with different aspect evaluation of the flexoelectric effect in dielectric solids. J. Appl. Phys. 116,
ratios, where s2 : h : s1 = 1 : 0.5 : R. 093502.
Amanatidou, E., Aravas, N., 2002. Mixed finite element formulations of strain-
gradient elasticity problems. Comput. Method. Appl. M. 191, 1723–1751.
Bishay, P.L., Sladek, J., Sladek, V., Atluri, S.N., 2012. Analysis of functionally graded
a given heigh h, the bottom length s2 is confirmed. Changing R multiferroic composites Using Hybrid/Mixed Finite Elements and Node-Wise
would lead to the change of the top length s1. As shown in Material Properties. CMC 29, 213–262.
Fig. 14, three values of h = 20 lm, 40 lm, and 60 lm, respectively Codony, D., Marco, O., Fernández-Méndez, S., Arias, I., 2019. An immersed boundary
hierarchical B-spline method for flexoelectricity. Comput. Method. Appl. M.
are chosen to study the shape effects on the effective electric field 354, 750–782.
Eeff of the truncated pyramid with a uniform load 10 MPa on the Cross, L.E., 2006. Flexoelectric effects: Charge separation in insulating solids
top surface. The effective electric field Eeff is found to increase with subjected to elastic strain gradients. J. Mater. Sci. 41, 53–63.
Deng, F., Deng, Q., Yu, W., Shen, S., 2017. Mixed finite elements for flexoelectric
decreasing value of R for all these three cases. As a smaller R value
solids. J. Appl. Mech. 84, 081004.
result in a larger difference between s1 and s2; this leads to larger Deng, F., Deng, Q., Shen, S.P., 2018. A three-dimensional mixed finite element for
strain gradients which produce a stronger flexoelectric effect flexoelectricity. J. Appl. Mech. 85, 031009–31011.
inside the truncated pyramid. Deng, Q., 2017. Size-dependent flexoelectric response of a truncated cone and the
consequent experimental measurement of flexoelectric properties. J. Appl.
Mech. 84, 101007–101011.
Deng, Q., Liu, L., Sharma, P., 2014a. Flexoelectricity in soft materials and biological
membranes. J. Mech. Phys. Solids 62, 209–227.
5. Conclusions Deng, Q., Liu, L., Sharma, P., 2014b. Electrets in soft materials: Nonlinearity, size
effects, and giant electromechanical coupling. Phy. Rev. E. 90, 012603.
In this paper, the governing equations with direct and con- Deng, Q., Lv, S., Li, Z., Tan, K., Shen, S., 2020. The impact of flexoelectricity on
materials, devices, and physics. J. Appl. Phys. 128, 080902.
verse flexoelectricity have been derived. A collocation MFEM for- Dong, L., Atluri, S.N., 2011. A simple procedure to develop efficient & stable hybrid/
mulation was presented and a 2D element also developed to mixed elements, and Voronoi cell finite elements for macro- & micromechanics.
solve standard BVPs in flexoelectricity. In this collocation MFEM CMC 24, 61–104.
Gharbi, M., Sun, Z.H., Sharma, P., White, K., 2009. The origins of electromechanical
formulation, C0 continuous approximation was applied for inde- indentation size effect in ferroelectrics. Appl. Phys. Lett. 95, 142901.
pendent displacement, strain, electric potential and electric inten- Gitman, I., Askes, H., Kuhl, E., Aifantis, E., 2010. Stress concentrations in fractured
sity vector variables. The kinematic constraints between primary compact bone simulated with a special class of anisotropic gradient elasticity.
Int. J. Solids Struct. 47, 1099–1107.
fields (displacements, potential) and secondary fields (strain, elec-
Harris, P., 1965. Mechanism for the shock polarization of dielectrics. J. Appl. Phys.
tric intensity) are satisfied by a collocation method at judiciously 36, 739–741.
chosen internal points in the elements. This scheme significantly Hu, S.L., Shen, S.P., 2009. Electric field gradient theory with surface effect for nano-
reduces the number of the DOFs when compared to the tradi- dielectrics. CMC 13, 63–87.
Krichen, S., Sharma, P., 2016. Flexoelectricity: A perspective on an unusual
tional Lagrangian approach, and at the same time, the kinematic electromechanical coupling. J. Appl. Mech. 83, 030801–30811.
constraints between the displacement and strain are guaranteed. Kogan, S.M., 1964. Piezoelectric effect during inhomogeneous deformation and
The corresponding 2D collocation MFEM code has also been acoustic scattering of carriers in crystals. Sov. Phys. Solid State 5, 2069–2070.
Lekhnitskii, S.G., 1963. Theory of Elasticity of an Anisotrophic Elastic Body. Holden-
developed. Two example problems have also been presented to Day.
demonstrate its veracity for obatining accurate and convergent Liang, X., Shen, S., 2013. Size-dependent piezoelectricity and elasticity due to the
solutions. In the first example, namely, a cantilever beam subject electric field-strain gradient coupling and strain gradient elasticity. Int. J. Appl.
Mech. 5, 1350015.
to an end load, both the deflection and the electric response Liu, C., Wang, J., Xu, G., Kamlah, M., Zhang, T.Y., 2019. An isogeometric approach to
obtained were in in good agreement with the results from theo- flexoelectric effect in ferroelectric materials. Int. J. Solids Struct. 162, 198–210.
retical solutions. In the second example of a truncated pyramid, Ma, W., Cross, L.E., 2001. Large flexoelectric polarization in ceramic lead magnesium
niobate. Appl. Phys. Lett. 79, 4420–4422.
good agreement of the effective piezoelectric constant with avail- Ma, W., Cross, L.E., 2006. Flexoelectricity of barium titanate. Appl. Phys. Lett. 88,
able solution in the literature was also obtained. The size- 2004–2007.
dependence of flexoelectricity was successfully predicted for both Mao, S., Purohit, P.K., Aravas, N., 2016. Mixed finite-element formulations in
piezoelectricity and flexoelectricity. P. R. Soc. A Math. Phy. 472, 20150879.
these two example problems using the collocation MFEM devel-
Maranganti, R., Sharma, N.D., Sharma, P., 2006. Electromechanical coupling in
oped in this study. Larger aspect ratios of the physical dimensions nonpiezoelectric materials due to nanoscale nonlocal size effects: Green’s
of the solid body lead to larger strain gradients and flexoelectric- function solutions and embedded inclusions. Phys. Rev. B 74, 014110.
ity. The collocation MFEM formulation developed in this paper Maugin, G.A., 1980. The method of virtual power in continuum mechanics:
Applications to coupled fields. Acta Mech. 35, 1–80.
can be applied to, for example, structural health monitoring of Meyer, R.B., 1969. Piezoelectric effects in liquid crystals. Phys. Rev. Lett. 22, 918–
flexoelectric sensors or actuators. 921.
38
X. Tian, J. Sladek, V. Sladek et al. International Journal of Solids and Structures 217-218 (2021) 27–39
Mohammadi, P., Liu, L., Sharma, P., 2014. A theory of flexoelectric membranes and Sladek, J., Stanak, P., Han, Z.D., Sladek, V., Atluri, S.N., 2013. Applications of the MLPG
effective properties of heterogeneous membranes. J. Appl. Mech. 81, 011007– method in engineering & Sciences: A review. CMES-Comp. Model. Eng. 92, 423–
11012. 475.
Nguyen, B., Zhuang, X., Rabczuk, T., 2018. Numerical model for the characterization Tagantsev, A.K., 1986. Piezoelectricity and flexoelectricity in crystalline dielectrics.
of Maxwell-Wagner relaxation in piezoelectric and flexoelectric composite Phys. Rev. B 34, 5883–5889.
material. Comput. Struct. 208, 75–91. Tagantsev, A.K., Meunier, V., Sharma, P., 2009. Novel electromechanical phenomena
Nguyen, B., Zhuang, X., Rabczuk, T., 2019. NURBS-based formulation for nonlinear at the nanoscale: phenomenological theory and atomistic modelling. MRS
electro-gradient elasticity in semiconductors. Comput. Method. Appl. M. 346, Bullet. 34, 643–647.
1074–1095. Thai, T.Q., Rabczuk, T., Zhuang, X., 2018. A large deformation isogeometric approach
Parton, V.Z., Kudryavtsev, B,A., 1988. Electromagnetoelasticity: piezoelectrics and for flexoelectricity and soft materials. Comput. Method. Appl. M. 341, 718–739.
electrically conductive solids, Taylor & Francis. Wang, B., Gu, Y., Zhang, S., Chen, L.Q., 2019. Flexoelectricity in solids: Progress,
Rahmati, A.H., Yang, S., Bauer, S., Sharma, P., 2019. Nonlinear bending deformation challenges, and perspectives. Prog. Mater. Sci. 106, 100570.
of soft electrets and prospects for engineering flexoelectricity and transverse Yaghoubi, S.T., Mousavi, S.M., Paavola, J., 2017. Buckling of centrosymmetric
(d31) piezoelectricity. Soft. Matter. 15, 127–148. anisotropic beam structures within strain gradient elasticity. Int. J. Solids Struct.
Sharma, P., Maranganti, R., Sharma, N.D., 2006. Electromechanical coupling in 109, 84–92.
nanopiezoelectric materials due to nanoscale nonlocal size effects: Green Yang, X.M., Hu, Y.T., Yang, J.S., 2004. Electric field gradient effects in anti-plane
function solution and embedded inclustions. Phys. Rev. B 74, 014110. problems of polarized ceramics. Int. J. Solids Struct. 41, 6801–6811.
Shen, S.P., Hu, S.L., 2010. A theory of flexoelectricity with surface effect for elastic Yudin, P., Tagantsev, A., 2013. Fundamentals of flexoelectricity in solids.
dielectrics. J. Mech. Phys. Solids 58, 665–677. Nanotechnology 24, 432001.
Sladek, J., Sladek, V., Stanak, P., Zhang, Ch., Tan, C.L., 2017. Fracture mechanics Yvonnet, J., Liu, L., 2017. A numerical framework for modeling flexoelectricity and
analysis of size-dependent piezoelectric solids. Int. J. Solids Struct. 113, 1–9. maxwell stress in soft dielectrics at finite strains. Comput. Method. Appl. M.
Sladek, J., Sladek, V., Wunsche, M., Zhang, C., 2018. Effects of electric field and 313, 450–482.
strain gradients on cracks in piezoelectric solids. Eur. J. Mech. A-Solid. 71, Zhuang, X., Nguyen, B.H., Nanthakumar, S.S., Thai, T.Q., Alajlan, N., Rabczuk, T., 2020.
187–198. Computational modeling of flexoelectricity – A review. Energies 13, 1326.
39