Professional Documents
Culture Documents
Mathematical Theory of
Elasticity of Quasicrystals
and Its Applications
With 82 figures
Author
Tianyou Fan
Department of Physics, School of Science
Beijing Institue of Technology
Beijing 100081, China
ISBN 978-7-03-025669-0
Science Press Beijing
Tianyou Fan
January 2010, Beijing
Contents
Preface
Chapter 1 Crystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Periodicity of crystal structure, crystal cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Three-dimensional lattice types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3 Symmetry and point groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 Reciprocal lattice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.5 Appendix of Chapter 1: Some basic concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Chapter 2 Framework of the classical theory of elasticity . . . . . . . . . . . . . 13
2.1 Review on some basic concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 Basic assumptions of theory of elasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3 Displacement and deformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.4 Stress analysis and equations of motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.5 Generalized Hooke’s law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.6 Elastodynamics, wave motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Chapter 3 Quasicrystal and its properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.1 Discovery of quasicrystal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.2 Structure and symmetry of quasicrystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.3 A brief introduction on physical properties of quasicrystals . . . . . . . . . . . . . 29
3.4 One-, two- and three-dimensional quasicrystals . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.5 Two-dimensional quasicrystals and planar quasicrystals . . . . . . . . . . . . . . . . 30
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Chapter 4 The physical basis of elasticity of quasicrystals . . . . . . . . . . . . . 35
4.1 Physical basis of elasticity of quasicrystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.2 Deformation tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.3 Stress tensors and the equations of motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.4 Free energy and elastic constants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.5 Generalized Hooke’s law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.6 Boundary conditions and initial conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.7 A brief introduction on relevant material constants of quasicrystals . . . . . 43
4.8 Summary and mathematical solvability of boundary value or initial-
viii Contents
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
Chapter 10 Dynamics of elasticity and defects of quasicrystals . . . . . . 191
10.1 Elastodynamics of quasicrystals followed the Bak’s argument . . . . . . . . 192
10.2 Elastodynamics of anti-plane elasticity for some quasicrystals . . . . . . . . 192
10.3 Moving screw dislocation in anti-plane elasticity . . . . . . . . . . . . . . . . . . . . . 194
10.4 Mode III moving Griffith crack in anti-plane elasticity . . . . . . . . . . . . . . . 197
10.5 Elasto-/hydro-dynamics of quasicrystals and approximate analytic
solution for moving screw dislocation in anti-plane elasticity . . . . . . . . . 199
10.6 Elasto-/hydro-dynamics and solutions of two-dimensional decagonal
quasicrystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
10.7 Elasto-/hydro-dynamics and applications to fracture dynamics of
icosahedral quasicrystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
10.8 Appendix of Chapter 10: The detail of finite difference scheme . . . . . . . 221
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
Chapter 11 Complex variable function method for elasticity of
quasicrystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
11.1 Harmonic and quasi-biharmonic equations in anti-plane elasticity of
one-dimensional quasicrystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
11.2 Biharmonic equations in plane elasticity of point group 12mm two-
dimensional quasicrystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
11.3 The complex variable function method of quadruple harmonic
equations and applications in two-dimensional quasicrystals . . . . . . . . . . 231
11.4 Complex variable function method for sextuple harmonic equation
and applications to icosahedral quasicrystals . . . . . . . . . . . . . . . . . . . . . . . . . 243
11.5 Complex analysis and solution of quadruple quasiharmonic
equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
11.6 Conclusion and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
Chapter 12 Variational principle of elasticity of quasicrystals,
numerical analysis and applications . . . . . . . . . . . . . . . . . . . . . . . 257
12.1 Basic relations of plane elasticity of two-dimensional quasicrystals . . . . 258
12.2 Generalized variational principle for static elasticity of quasicrystals . . . . 259
12.3 Finite element method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
12.4 Numerical examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
Chapter 13 Some mathematical principles on solutions of elasticity
of quasicrystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
13.1 Uniqueness of solution of elasticity of quasicrystals . . . . . . . . . . . . . . . . . . . 273
Contents xi
A.I.9 Detail of complex analysis of generalized cohesive force model for plane
elasticity of two-dimensional point groups 5m, 10mm and 10, 10
quasicrystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 345
A.I.10 On the calculation of integral (9.2-14) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
Appendix II Dual integral equations and some additional calculations . . . . . 348
A.II.1 Dual integral equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348
A.II.2 Additional derivation on the solution of dual integral equations
(8.3-8) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
A.II.3 Additional derivation on the solution of dual integral equations
(9.8-8) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357
.
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
Chapter 1
Crystals
This book discusses mainly elasticity and defects of quasicrystals, however qua-
sicrystals have inherent connection with crystals. This chapter provides the basic
knowledge on crystals which may be beneficial to study quasicrystals and relevant
topics.
Table 1.2-1 Crystals and the relationship of the length of sides and angles
Crystal system Characters of cell
Triclinic a = b = c, α = β = γ
Monoclinic a = b = c, α = γ = 90◦ = β
Orthorhombic a = b = c, α = β = γ = 90◦
Rhombohedral a = b = c, α = β = γ = 90◦
Tetragonal a = b = c, α = β = γ = 90◦
Hexagonal a = b = c, α = β = 90◦ , γ = 120◦
Cubic a = b = c, α = β = γ = 90◦
Among each crystal system there are certain classes of crystals that are classified
based on the configuration such that whether the face centre or body centre contains
lattice point. For example, the cubic system can be classified as three classes: the
simple cubic, body centre cubic and face centre cubic. According to this classifica-
tion, the seven crystal systems contain 14 different lattice types, called Bravais cells
as shown in Fig. 1.2-1.
Apart from the above mentioned 14 Bravais cells with three-dimensional lattices,
there are 5 Bravais cells of two-dimensional lattice, we do not go any further.
1.3 Symmetry and point groups 3
1, 2, 3, 4, 6, ¯
I = (I), m = 2̄, 4̄, (1.3-2)
Note: T = C3 D2 means the composition between operations C3 and D2 , where suffixes 3
denotes a 3-fold axis.
O = C3 C4 C2 means the composition between operations C3 , C4 and C2 , where 3 repre-
sents a 3-fold axis, 2 represents a 2-fold axis.
1.4 Reciprocal lattice 5
The concept and sign of point groups will be used in the subsequent chapters.
such that the lattice with the base vectors {b1 , b2 , b3 } is the reciprocal lattice LR of
crystal lattice L with the base vectors {a1 , a2 , a3 }. Between bi and aj there exist
the relationship
a2 × a3 a3 × a1 a1 × a2
b1 = , b2 = , b3 = , (1.4-2)
Ω Ω Ω
then
1
Ω∗ = .
Ω
The position of any point in the reciprocal lattice can be expressed by
G = h1 b1 + h2 b2 + h3 b3 (1.4-3)
h1 , h2 , h3 = ±1, ±2, · · ·
In general, the course of crystallography does not contain the contents given in this
section. Because the discussion here is dependent on quasicrystals, especially with
the elasticity of quasicrystals, we have to introduce some of the simplest relevant
arguments.
In 1900, Planck put forward the quantum theory. Soon after Einstein developed
the theory and explained the photo-electric effect, which leads to the photon concept.
Einstein also studied the specific heat cv of crystals by using the Planck quantum
theory. There are some unsatisfactory points in the work of Einstein although his
formula explained the phenomenon of cv = 0 at T = 0, where T denotes the absolute
temperature. To improve Einstein s work, Debye[3] and Born et al. [4,5] applied the
quantum theory to study the specific heat arising from lattice vibration in 1912 and
1913 respectively, and got a great success. The theoretical prediction is in excellent
agreement to the experimental results, at least for the atom crystals.
The propagation of the lattice vibration is called the lattice wave. Under the
long-wavelength approximation, the lattice vibration can be seen as continuum elas-
tic vibration, i.e., the lattice wave can be approximately seen as the continuum
elastic wave. The motion is a mechanical motion, but Debye and Born assumed
that the energy can be quantized based on Planck’s hypothesis. With the elastic
wave approximation and quantization, Debye and Born successfully explained the
specific heat of crystals at low temperature, and the theoretical prediction is con-
sistent with the experimental results in all range of temperature, at least for the
atomic crystals. The quanta of the elastic vibration, or the smallest unit of energy
of the elastic wave is named phonon, because the elastic wave is one of acoustic
waves. Unlike photon, the phonon is not an elementary particle, but in the sense of
quantization, the phonon presents natural similarity to that of photon and other ele-
mentary particles, thus can be named quasi-particle. The concept created by Debye
and Born opened the study on lattice dynamics, an important branch of solid state
physics. Yet according to the view point at present, the Debye and Born theory on
solid belongs to a phenomenological theory, though they used the classical quantum
theory.
Landau[6] further developed the phenomenological theory and put forward the
concept of elementary excitation. According to his concept, photon and phonon etc.
1.5 Appendix of Chapter 1: Some basic concepts 7
F = E − TS (1.5-1)
be minimum, in which E is the internal energy, S is the entropy and T is the absolute
temperature.
8 Chapter 1 Crystals
Landau proposed the so-called second order phase transition theory by introduc-
ing a macroscopic order parameter η to describe (order-disorder) phase transition,
i.e., assuming that the free energy can be expanded as a power series of η,
1
F = α(|G|)(T − TC (G))η 2 + higher-order terms, (1.5-4)
2
where the constant α is related to the reciprocal vector G (the concepts on the re-
ciprocal vector and the reciprocal lattice, refer to Section 1.4). Further, Anderson[7]
showed that for crystals if the density of periodic crystals can be expressed by
Fourier series (the expansion exists due to the periodicity of the structure in three-
dimensional lattice or the reciprocal lattice)
ρ(r) = ρG exp{iG · r} = |ρG | exp{−iΦG + iG · r}, (1.5-5)
G∈LR G∈LR
with the amplitude |ρG | and the phase angle ΦG , due to ρ(r) being real, |ρG | = |ρ−G |
and ΦG = −ΦG , the order parameter is
η = |ρG | . (1.5-7)
1.5 Appendix of Chapter 1: Some basic concepts 9
Anderson pointed out that for crystals the phase angle ΦG contains the phonon
u for crystal, i.e.,
ΦG = G · u, (1.5-8)
In this book we do not discuss incommensurate phases, but quasicrystals are related
with so-called incommensurate structure, we mention it in brief.
Since 1960’s, incommensurate crystals have been studied by many physicists,
see e.g. [8]. The incommensurate phase means that it is plus an additional in-
commensurate modulate at the basic lattice, in which the modulated ones may be
displacements or compounds of atoms or arrangement of spin etc. As an example,
if a modulated displacement λ is plus to a lattice with period a, and if λ/a is an
rational number, the crystal becomes a super-structure with long period (which is
the integer times of a); and if λ/a is an irrational number, the crystal becomes an
incommensurate structure. In this case along the modulate direction the periodicity
is lost. The modulation can be one-dimensional, e.g. Na2 CO3 , NaNO2 , etc, or
two-dimensional, e.g. TaSe2 , quartz, etc, or three-dimensional, e.g. Fe1−x O, etc.
In the incommensurate phases, the modulation is only a “perturbation” of another
period of the basic lattice, the diffraction pattern of the basic lattice holds, i.e., the
crystallography symmetry holds, so one calls the structure being the incommensu-
rate crystals. It is noticed that there are new degrees of freedom in the phases, be
named the phasons. Here the phason modes present long-wavelength propagation
similar to that of the phonon modes. In Chapter 4 we shall study that the phason
modes in the quasicrystals present quite different nature, i.e., the motion of atoms
exhibits discontinuous jumps rather than the long-wavelength propagation. In addi-
tion, in the quasicrystals there is non-crystallographic orientational symmetry, which
is essentially different from that of incommensurate crystals.
The crystals are solid with long-range order due to regular atom arrangement. In
contrast, there is a solid without order, but it has short-range order in scale within
the atom size. This material is the aphomus, as a branch of the condensed matter
physics.
References
[1] Kittel C.Introduction to the Solid State Physics. New York: John Wiley & Sons,
Inc, 1976
[2] Wybourne B G. Classical Group Theory for Physicists. New York: John Wiley &
Sons, Inc, 1974
[3] Debye P. Die Eigentuemlichkeit der spezifischen Waermen bei tiefen Temperaturen.
Arch de Genéve, 1912, 33(4): 256–258
References 11
[4] Born M, von Kármán Th. Zur Theorie der spezifischen Waermen, Physikalische
Zeitschrift. 1913, 14(1): 15–19
[5] Born M, Huang K. Dynamic Theory of Crystal Lattices, Oxford: Clarendon Press,
1954
[6] Landau L D, Lifshitz M E. Theoretical Physics V: Statistical Physics. 3rd ed. Oxford:
Pergamon Press, 1980
[7] Anderson P W. Basic Notions of Condensed Matter Physics. Menlo Park: Benjamin-
Cummings, 1984
[8] Blinc B and Lavanyuk AP. Imcommensurate Phases in Dielectrics, I. II. Amsterdan:
North Holand, 1986
Chapter 2
Framework of the classical theory of
elasticity
A quantity with both magnitude and direction is named vector, denoted by a, and
a = |a| represents its magnitude. The scalar product is defined as a · b = ab cos(a, b)
of two vectors a and b. The vector product is defined as a × b = n ab sin (a, b),
in which n is the unit vector perpendicular to both a and b, so |n| = 1. A more
general definition on vector is given later.
To describe the vector and the tensor, it is convenient to introduce the coordinate
frame. We will consider the orthogonal frame. Assume that e1 , e2 and e3 are three
unit vectors and mutually perpendicular, i.e., e1 · e2 = 0, e2 · e3 = 0, e3 · e1 = 0
a = a1 e1 + a2 e2 + a3 e3
or (2.1-1)
a = (a1 , a2 , a3 ).
where c11 , c12 , · · · , c33 are some scalar constants. The relation (2.1-2) is the coordi-
nate transformation, which can be expressed in the matrix form
⎡ ⎤ ⎡ ⎤
e1 e1
⎢ ⎥ ⎢ ⎥
⎣ e2 ⎦ = [C] ⎣ e2 ⎦ , (2.1-3)
e3 e3
where ⎡ ⎤
c11 c12 c13
⎢ c c22 c23 ⎥
[C] = ⎣ 21 ⎦,
c31 c32 c33
which is an orthogonal matrix, consequently
here notation “T” marks transpose operation, and “−1” the inversion operation. It
follows, ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
e1 e1 e1
⎢ e ⎥ T ⎢ e ⎥ −1 ⎢ e ⎥
⎣ 2 ⎦ = [C] ⎣ 2 ⎦ = [C] ⎣ 2 ⎦ . (2.1-5)
e3 e3 e3
The summation sign in the right-handside of (2.1-9) is omitted, when the repeated
indexes in cij aj represent summing. Henceforth, the summation convention will be
used throughout.
A set of number (a1 , a2 , a3 ) satisfying the relation (2.1-9) under the linear trans-
formation (2.1-2) is a vector regardless its physical meaning. This is an algebraic
definition of vector; it is more general than saying that the vector has both magni-
tude and direction.
2.1.4 Tensor
Let us define nine numbers in the orthogonal frame e1 , e2 , e3 as A:
⎡ ⎤
A11 A12 A13
A = ⎣ A21 A22 A23 ⎦ , (2.1-10)
A31 A32 A33
under the linear transformation, then A is a tensor of rank 2, where cij are given
by (2.1-3), and the summation sign is omitted in the right-handside of (2.1-11). It
is evident that the concept of tensor is an extension of that of vector. According to
the definition Aij represents a tensor where i=1, 2, 3, j=1, 2, 3. It is understood
that it represents a component with the indexes i and j of the tensor.
16 Chapter 2 Framework of the classical theory of elasticity
1) Unit tensor
0, i = j,
I = δij = (2.1-12)
1, i = j,
which is named the Kronecker delta conventionally.
2) Transpose of tensor
⎡ ⎤
A11 A21 A31
AT = ⎣ A12 A22 A32 ⎦ . (2.1-13)
A13 A23 A33
mA = mAij . (2.1-15)
5) Product of tensors
AB = Aij Bkl . (2.1-16)
r = r + u (2.3-1)
or
u = r − r = xi − xi . (2.3-1 )
In Fig. 2.3-1, e1 , e2 , e3 depicts any orthogonal coordinate system, especially we
use the rectilinear coordinate system (x1 , x2 , x3 ) or (x, y, z). Assume that O1 in R
is a point near the point O, the radius vector connecting them is dr = dxi . The
point O1 becomes point O1 in R after deformation. The radius vector connecting
points O1 and point O1 is dr = dxi = dxi + dui . The displacement of point O1 is
u , thus
u = u + du, (2.3-2)
i.e.,
dui = ui − ui (2.3-3)
and ∂ui
dui = dxj . (2.3-4)
∂xj
(2.3-4) expresses the Taylor expansion at point O and takes the first order term
only. Under the small deformation assumption, this reaches a very high accuracy.
It denotes
18 Chapter 2 Framework of the classical theory of elasticity
∂ui
= εij + ωij , (2.3-5)
∂xj
in which
1 ∂ui ∂uj
εij = + , (2.3-6)
2 ∂xj ∂xi
1 ∂ui ∂uj
ωij = − , (2.3-7)
2 ∂xj ∂xi
here εij is a symmetric tensor
εij = εji (2.3-8)
and called the strain tensor, while ωij an asymmetric tensor, which has only three
independent components
1 ∂uy ∂uz
Ωx = ωyz = − ,
2 ∂z ∂y
1 ∂uz ∂ux
Ωy = ωzx = − , (2.3-9)
2 ∂x ∂z
1 ∂ux ∂uy
Ωz = ωxy = − .
2 ∂y ∂x
The physical meaning of εij describes the volume and shape change of a cell, and
that of ωij the rigid-body rotation, which is independent of deformation. Henceforth,
it suffices to consider εij .
The components ε11 , ε22 and ε33 (if denote x = x1 , y = x2 , z = x3 , then we have
εxx , εyy and εzz ) represent normal strains, describing the volume change of a cell,
while ε32 = ε23 , ε13 = ε31 and ε12 = ε21 (or εyz = εzy , εzx = εxz and εxy = εyx )
represent shear strains, describing the shape change of a cell.
of the surface elements, the former are called normal stresses, and the latter shear
stresses.
According to the angular momentum conservation, one finds that
this means the stress tensor is a symmetric tensor, and (2.4-2) is named the shear
stress mutual equal law.
The external surface forces density (tractions) Ti subjected to the surface of a
body should be balanced with the internal stresses, this leads to
σij nj = Ti , (2.4-3)
where nj is the unit vector along the outward normal to the surface element. People
also call Ti area force density.
Equation (2.4-3) describes the stress boundary conditions which play a very
important role for elasticity.
This operation is the reflection or mapping. In addition, we know that between σij
in e1 , e2 , e3 and σij in e1 , e2 , e3 , there are (refer to Section 2.1)
σkl = αkj αli σji , (2.5-6)
Therefore, under the transformation, for Cijkl in (2.5-1) (or Bijkl in (2.5-4)) whose
suffixes containing 3 with an odd number of times (1 or 3) will change sign, while
the others will remain invariant. Considering the symmetry of the crystal, however,
the physical properties including Cijkl (or Bijkl in (2.5-4)) should remain unchanged
under symmetric operation (including the reflection). So it is obvious that all com-
ponents with an odd number of suffixes 3 must vanish, i.e.,
with
ξ = x1 + ix2 , η = x1 − ix2 .
22 Chapter 2 Framework of the classical theory of elasticity
∂2 ∂2
where φ is the scalar potential, and ψ the vector potential, and ∇2 = 2
+ 2+
∂x ∂y
∂2
, (2.6-5) are typical wave equations of mathematical physics. To solve the prob-
∂z 2
lem apart from the boundary conditions one needs initial conditions, i.e.,
ui (xi , 0) = ui0 (xi ),
xi ∈ Ω .
u̇i (xi , 0) = u̇i0 (xi ),
2.7 Summary
The classical theory of elasticity is concluded to solve the following initial-boundary
value problem
1 ∂ui ∂uj
εij = + ,
2 ∂xj ∂xi
∂σij ∂ 2 uj
= ρ 2 − fj , t > 0, xi ∈ Ω ,
∂xi ∂t
σij = Cijkl εkl ,
ui (xj , 0) = ui0 (xj )
xj ∈ Ω (Initial condition)
u̇i (xj , 0) = u̇i0 (xj )
24 Chapter 2 Framework of the classical theory of elasticity
σij nj = Ti , t > 0, xj ∈ Γt
(Boundary condition)
ui = ūi , t > 0, xj ∈ Γu
where ui0 (xj ), u̇i0 (xj ), Ti and ūi are known functions, Ω denotes the region of ma-
terials we studied, Γt and Γu are parts of boundary Γ on which the tractions and
∂ 2 uj
displacements are prescribed respectively and Γ = Γt + Γu . If = 0, the prob-
∂t2
lem reduces to a static problem as pure boundary value problem, there are no initial
conditions at all.
References
[1] Landau L D, Lifshitz E M. Theoretical Physics V: Theory of Elasticity. Oxford:
Pergamon Press, 1986
[2] Born M, Huang K. Dynamic Theory of Crystal Lattices. Oxford: Clarendon Press,
1954
[3] Anderson P W. Basic Notions of Condensed Matter Physics. Menlo Park: Benjamin-
Cummings, 1984
Chapter 3
Quasicrystal and its properties
duce ordered structure with quasiperiodicity and called the novel alloy as “quasicrys-
tal” formally, in which their theoretical (computed) diffraction pattern in excellent
agreement with that of the experimental observation. Soon after, other groups, for
example, Ye et al[3] , Zhang et al[4] etc. also found similar structure of five-fold
symmetry and icosahedral quasicrystal in Ni-V and Ni-Ti alloys.
The icosahedral quasicrystals are one of three-dimensional quasicrystals, in which
the atomic arrangement is quasiperiodic in three directions. Another three-dimensio-
nal quasicrystal is cubic quasicrystal observed by Fung et al[5] later.
mentioned that the number of base vectors for crystal N is identical to the number
of the dimensions d of the crystal, i.e., N = d. However, because quasicrystals have
quasiperiodic symmetry (including both or either quasiperiodic translational and ori-
entational symmetries disallowed by the rule of crystallography), the Miller indices
cannot be used and instead we need to employ six indices (n1 , n2 , n3 , n4 , n5 , n6 ).
This feature implies, it is necessary to introduce higher dimensional (four-, or five-
or six-dimensional) spaces to characterize the symmetry of quasicrystals. This idea
is identical to that of group theory, i.e., the quasiperiodic structure is periodic in
higher dimensional space (four-, or five- or six-dimensional space). Quasicrystals in
the real three-dimensional space (physical space) may be seen as a projection of a
periodic lattice in the higher dimensional space (mathematical space). The projec-
tion of the periodic lattice at four-, five-, and six-dimensional space to the physical
space generates one-, two- and three-dimensional quasicrystals, respectively. The
six-dimensional space is denoted by E 6 , which consists of two sub-spaces, one is the
physical space, called the parallel space and denoted by E3 , another is the comple-
3
mentary space also called the vertical space and denoted by E⊥ , so that
E 6 = E3 ⊕ E⊥
3
, (3.2-1)
References
[1] Shechtman D, Blech I, Gratias D et al. Metallic phase with long-range orientational
order and no translational symmetry. Phys Rev Lett, 1984, 53(20): 1951–1953
References 31
[2] Levine D, Steinhardt P J.Quasicrystals: A new class of ordered structure. Phys Rev
Lett, 1984, 53(26): 2477–2450
[3] Ye H Q, Wang D, Kuo K H. Five-fold symmetry in real and reciprocal space. Ultra-
microssopy, 1985, 16(2): 273–277
[4] Zhang Z, Ye H Q, Kuo K H. A new icosahedral phase with m35 symmetry. Phil Mag
A, 1985, 52(6): L49–L52
[5] Feng Y C, Lu G, Witers R I. An incommensurate structure with cubic point group
symmetry in rapidly solidified V-Vi-Si alloy. J Phys Condens Matter, 1989, 1(23):
3695–3700
[6] Bendersky L. Quasicrystal with one-dimensional translational symmetry and a ten-
fold rotation axis. Phys Rev Lett, 1985, 55(14): 1461–1463
[7] Chattopadhyay K, Lele S, Thangarai N et al. Vacancy ordered phases and one-
dimensional quasiperiodicity. Acta Metall, 1987, 35(3): 727–733
[8] Fung K K, Yang C Y, Zhou Y Q et al. Icosahedrally related decagonal quasicrystal
in rapidly cooled Al-14-at.%-Fe alloy. Phys Rev Lett, 1986, 56(19): 2060–2063
[9] Urban K, Mayer J, Rapp M et al. Studies on aperiodic crystals in Al-Mn and Al-V
alloys by means of transmission electron microscopy. Journal de Physique Colloque,
1986, 47C(3): 465–475
[10] Wang N, Chen H, Kuo K H. Two-dimensional quasicrystal with eightfold rotational
symmetry. Phys Rev Lett, 1987, 59(9): 1010–1013
[11] Li X Z, Guo K H. Decagonal quasicrystals with different periodicities along the
tenfold axis in rapidly solidified Al-Ni alloys. Phil Mag Lett, 1988, 58(3): 167–171
[12] Merdin R, Bajema K, Clarke R et al. Quasiperiodic GaAs-AlAs heterostructures.
Phys Rev Lett, 1985, 55(17): 1768–1770
[13] Hu A, Tien C, Li X et al. X-ray diffraction pattern of quasiperiodic (Fibonacci)
Nb-Cu superlattices. Phys Lett A, 1986, 119(6): 313–314
[14] Feng D, Hu A, Chen K J et al. Research on quasiperiodic superlattice. Mater Sci
Forum, 1987, 22(24): 489–498
[15] Terauchi H, Noda Y, Kamigami K et al. X-Ray diffraction patterns of configurational
Fibonacci lattices. J Phys Jpn, 1988, 57(7): 2416–2424
[16] Chen K J, Mao G M, Fend D et al. Quasiperiodic a-Si: H/a-SiNx: H multilayer
structures. J Non-cryst Solids, 1987, 97(1): 341–344
[17] Yang W G, Wang R H, Gui J. Some new stable one-dimensional quasicrystals in
Al65 Cu20 Fe10 Mn5 alloy. Phil Mag Lett, 1996, 74(5): 357–366
[18] He L X, Li X Z, Zhang Z et al. One-dimensional quasicrystal in rapid solidified
alloys. Phys Rev Lett, 1988, 61(9): 1116–1118
[19] Biham O, Mukamel D, Shtrikman. Symmetry and stability of icosahedral and other
quasicrystalline phases. Phys Rev Lett, 1986, 56(20): 2191–2194
32 Chapter 3 Quasicrystal and its properties
[38] Mayou D. Generalized Drude formula for the optical conductivity of quasicrystals.
Phys Rev Lett, 2000, 85(6), 1290–1293
[39] Notomi M, Suzuk H, Tamamura T et al. Document lasing action due to the two-
dimensional quasiperiodicity of photonic quasicrystals with a Penrose lattice. Phys
Rev Lett, 2004, 92(12): 123906
[40] Feng Z F, Zhang X D, Wang Y Q. Negative refraction and imaging using 12-fold-
symmetry quasicrystals. Phys Rev Lett, 2005, 94(24): 247402
[41] Lifshitz R, Arie A, Bahabad A. Photonic quasicrystals for nonlinear optical frequency
conversion. Phys Rev Lett, 2005, 95(13): 13390
[42] Marn W, Megens M. Experimental measurement of the photonic properties of icosa-
hedral quasicrystals. Nature, 2005, 436(7053): 993–996
[43] Kohmoto M, Sutherland B. Electronic states on a Penrose lattice. Phys Rev Lett,
1986, 56(25): 2740–2743
[44] Sutherland B. Simple system with quasiperiodic dynamics: a spin in a magnetic field.
Phys Rev B, 1986, 34(8): 5208–5211
[45] Fujiwara T, Yokokawa T. Universal pseudogap at Fermi energy in quasicrystals. Phys
Rev Lett, 1991, 66(3): 333–336
Chapter 4
The physical basis of elasticity of
quasicrystals
where ⊕ denotes the direct sum; u is in the physical space, or the parallel space E3 ;
3
w is in the complement space, or the perpendicular space E⊥ , which is an internal
space
Furthermore, the two displacement vectors depend on the coordinate vector r
For simplicity, the superscript of r will be removed hereafter. From the angle of
mathematical theory of elasticity of quasicrystals and its technological applications
the formulas (4.1-1) and (4.1-2) are sufficient for comprehending the following con-
tents of the book. If readers are interested in more of the physical background on
the phonon and phason fields in quasicrystals, we suggest that they could read the
Appendix of this chapter (i.e., the Section 4.9).
With basic formulas (4.1-1) and (4.1-2) and some fundamental conservation laws
well known in physics, the macroscopic basis of the continuous medium model of
elasticity of quasicrystals can be set up, in some extent, the discussion is an extension
to that in Chapter 2, which will be done in the following sections.
⎡ ⎤
∂ux 1 ∂ux ∂uy 1 ∂ux ∂uz
⎢ + + ⎥
⎢ ∂x 2 ∂y ∂x 2 ∂z ∂x
⎥
⎢ ⎥
⎢ 1 ∂u ∂uy ∂uy 1 ∂uy ∂uz ⎥
⎢ x ⎥
=⎢
⎢ 2 ∂y
+
∂x ∂y 2 ∂z
+
∂y
⎥
⎥
⎢ ⎥
⎢ ⎥
⎢ 1 ∂ux ∂uz 1 ∂uy ∂uz ∂uz ⎥
⎣ + + ⎦
2 ∂z ∂x 2 ∂z ∂y ∂z
⎡ ⎤
1 ∂uy ∂ux 1 ∂uz ∂ux
⎢ 0 − − − − ⎥
⎢ 2 ∂x ∂y 2 ∂x ∂z ⎥
⎢ ⎥
⎢ 1 ∂u ∂uy
1 ∂uz ∂uy
⎥
⎢ − x
− − − ⎥
+⎢
⎢ 2 ∂y ∂x
0
2 ∂y ∂z
⎥
⎥
⎢ ⎥
⎢ ⎥
⎢ 1 ∂ux ∂uz 1 ∂uy ∂uz ⎥
⎣ − − − − 0 ⎦
2 ∂z ∂x 2 ∂z ∂y
1 ∂ui ∂uj 1 ∂uj ∂ui
= + − − = εij + ωij ,
2 ∂xj ∂xi 2 ∂xi ∂xj
1 ∂ui ∂uj
εij = + , (4.2-3)
2 ∂xj ∂xi
1 ∂ui ∂uj
ωij = − , (4.2-4)
2 ∂xj ∂xi
which means the gradient of the phonon vector u can be decomposed into two
parts εij and ωij , in which εij has contribution to the deformation energy, and ωij
represents a kind of rigid-body rotations. We consider only εij , which is the phonon
deformation tensor, or the strain tensor, it is a symmetric tensor: εij = εji .
Similarly for the phason field we have
∂wi
dwi = dxj (4.2-5)
∂xj
and ⎡ ⎤
∂wx ∂wx ∂wx
⎢ ∂x ∂y ∂z ⎥
⎢ ⎥
⎢ ⎥
∂wi ⎢ ∂wy ∂wy ∂wy ⎥
∇w = =⎢
⎢ ∂x
⎥, (4.2-6)
∂xj ⎢ ∂y ∂z ⎥
⎥
⎢ ⎥
⎣ ∂wz ∂wz ∂wz ⎦
∂x ∂y ∂z
though it can be decomposed into symmetric and asymmetric parts, all components
∂wi
of contribute to the deformation of the quasicrystals, the phason deformation
∂xj
38 Chapter 4 The physical basis of elasticity of quasicrystals
which is asymmetric tensor wij = wji , and describes the local rearrangement of
atoms in a cell.
The difference between εij and wij given by (4.2-3) and (4.2-7) are originated
from the physical properties of the phonon modes and the phason modes. This
can also be explained by the group theory, in that, they follow different irreducible
representations for some symmetry transformations for most quasicrystal systems,
except the three-dimensional cubic quasicrystal system. The detail about this is
omitted here.
For the three-dimensional cubic quasicrystals, the phason modes exibit the same
behaviour as that of the phonon modes, which will be particularly discussed in
Chapter 9.
x, y, z ∈ Ω (4.3-1)
⎪
⎪
⎪ ∂Hij + gi = 0,
⎩
∂xj
based on the momentum conservation law.
Apply the angular momentum conservation law to the phonon field
d
r × ρu̇dΩ = r × f dΩ + r × T dΓ (4.3-2)
dt Ω Ω Ω
the linear case. Lubensky et al gave their formulation based on the hydrodynamics
principle, so equation (4.3-6) may be seen as elasto-/hydro-dynamic equation of
quasicrystals. In particular, the second equation of (4.3-6) presents the dissipation
feature of motion of the phason degrees in the dynamic process, it is irreversible
thermodynamically.
The elastodynamics of quasicrystals poses a great challenge, see e.g. [32, 33], we
discuss it only in a quite narrow scope (only in Chapter 10), the influence of those
distinguishing arguments in the circle of the quasicrystal study is limited to the
range in Chapter 10. The content therein does not affect the results and conclusions
presented in other chapters.
1 1 1 1
= Cijkl εij εkl + Rijkl εij wkl + Kijkl wij wkl + Rijkl wij εkl
2 2 2 2
= Fu + Fw + Fuw , (4.4-1)
where Fu , Fw and Fuw denote the parts contributed by the phonon, the phason and
the phonon-phason coupling, respectively, and
∂2F
Cijkl = (4.4-2)
∂εij ∂εkl 0
is the phonon elastic constant tensor, discussed in Chapter 2 already, and
[C]9×9 .
[K]9×9 .
In addition,
∂2F
Rijkl = , (4.4-6)
∂εij ∂wkl 0
∂2F
Rijkl = (4.4-7)
∂wij ∂εkl 0
are the elastic constants of the phonon-phason coupling. To be noted that, the
suffixes i, j, l belong to space E3 and k belongs to space E⊥
3
, and
Rijkl = Rjikl , Rijkl = Rklij , Rklij = Rijkl , (4.4-8)
but
Rijkl = Rklij , Rijkl = Rklij , (4.4-9)
of which all components can be expressed by the symmetric matrices
and
[R]T = [R ] , (4.4-10)
where T denotes the transpose operator. The composition of four matrices [C] , [K] ,
[R] and [R ] forms a 18 × 18 matrix
⎡ ⎤
[C] [R] [C] [R]
[C, K, R] = =⎣ ⎦. (4.4-11)
T
[R ] [K] [R] [K]
where
σij
= [σij , Hij ]T ,
Hij
(4.5-3)
εij T
= [εij , wij ] .
wij
σij nj = Ti ,
(x, y, z) ∈ Γt , (4.6-1)
Hij nj = hi ,
where nj represents the unit outward normal vector at any point at Γ , Ti and hi the
traction and generalized traction vectors, which are given functions at the boundary.
Formula (4.6-1) is called the stress boundary conditions. And for the latter case,
ui = ūi ,
(x, y, z) ∈ Γu , (4.6-2)
wi = w̄i ,
4.7 A brief introduction on relevant material constants of quasicrystals 43
where ūi and w̄i are known functions at the boundary. Formula (4.6-2) is named
the displacement boundary conditions.
If Γ = Γt (i.e., Γu = 0), the problem for solving equations (4.2-3), (4.2-7),
(4.3-1) and (4.4-14) under boundary conditions (4.6-1) is a traction boundary value
problem. While Γ = Γu (i.e., Γt = 0) , the problem for solving equations (4.2-3),
(4.2-7), (4.3-1) and (4.4-14) under boundary conditions (4.6.2) is a displacement
boundary value problem.
If Γ = Γu + Γt and both Γt = 0, Γu = 0, the problem for solving equations
(4.2-3), (4.2-7), (4.3-1) and (4.4-14) under boundary conditions (4.6-1) and (4.6.2)
is a mixed boundary value problem.
For the dynamic problem, if taking the wave equations (4.3-5) together with equa-
tion (4.2-3), (4.2-7) and (4.4-14), besides boundary conditions (4.6-1) and
(4.6-2), we must prescribe initial value conditions:
in which ui0 (x, y, z, 0), u̇i0 (x, y, z, 0),wi0 (x, y, z, 0) and ẇi0 (x, y, z, 0) are given func-
∂ui
tions and u̇i = etc. In this case the problem is called initial-boundary value
∂t
problem.
If taking the wave equations coupling diffusion equations (4.3-6) together with
(4.2-3) and (4.4-14), the initial value conditions are
C13 = 66.63(GPa),
R1 = −1.1, |R2 | < 0.2(GPa)
and there are no measured data for K1 , K2 (but we can use those obtained by the
Monte-Carlo simulation), and Γw can approximately be taken the value of the icosa-
hedral quasicrystal. In addition, the tensile strength σc = 450MPa for decagonal
Al-Cu-Co quasicrystals before annealing, and σc = 550MPa after annealing. The
hardness for decagonal Al-Cu-Co quasicrystals is 4.10GPa[30,31] , the fracture tough-
√
ness is 1.0 ∼ 1.2MPa m [30] .
With these basic data, the computation for stress analysis for statics and dy-
namics can be undertaken.
For a dynamic problem, based on the Bak’s argument, the mathematical formu-
lation is
∂σij ∂ 2 ui ∂Hij ∂ 2 wi
+ fi = ρ 2 , + gi = ρ 2 , xi ∈ Ω , t > 0, (4.8-6)
∂xj ∂t ∂xj ∂t
1 ∂ui ∂uj ∂wi
εij = + , wij = , xi ∈ Ω , t > 0, (4.8-7)
2 ∂xj ∂xi ∂xj
ui |t=0 = ui0 , u̇i |t=0 = u̇i0 , wi |t=0 = wi0 , ẇi |t=0 = ẇi0 , xi ∈ Ω . (4.8-11)
For a dynamic problem, based on the argument of Lubensky et al, the mathe-
matical formulation is
The solution satisfying all equations and corresponding initial conditions and
boundary conditions is just the realistic solution of elasticity of quasicrystals math-
ematically and has physical meaning.
The existence and uniqueness of solutions of elasticity of quasicrystals will further
be discussed in Chapter 13.
46 Chapter 4 The physical basis of elasticity of quasicrystals
where G is a reciprocal vector, and LR the reciprocal lattice (the concepts on the
reciprocal vector and reciprocal lattice, referring to Chapter 1), ρG is a complex
number
ρG = |ρG | e−iΦG (4.9-2)
with an amplitude |ρG | and phase angle ΦG , due to ρ(r) being real, |ρG | = |ρ−G |
and ΦG = −Φ−G .
There exists a set of N base vectors, {Gn }, so that each G ∈ LR can be written
as mn Gn for integers mn . Furthermore N = kd, where k is the number of
the mutually incommensurate vectors in the d-dimensional quasicrystal. In general
k = 2. A convenient parametrization of the phase angle is given by
Φn = Gn · u + G⊥
n · w, (4.9-3)
in which u can be understood similar to the phonon like that in conventional crystals,
while w can be understood the phason degrees of freedom in quasicrystals, which
describe the local rearrangement of unit cell description based on the Penrose tiling.
Both are functions of the position vector in the physical space only, where Gn is the
reciprocal vector in the physical space E3 just mentioned and G⊥ n is the conjugate
3
vector in the perpendicular space E⊥ . We can realize that the above mentioned Bak’s
hypothesis is a natural development of Anderson’s theory introduced in Section 1.5.
Almost in the same time, Levine et al[4] , Lubensky et al [5∼8] , Kalugin et al[9] ,
Torian and Mermin[10] , Jaric[11] , Duneau and Katz[12] , Socolar et al[13] , Gahler
and Phyner[14] carried out the study on elasticity of quasicrystals. Though the
researchers studied the elasticity from different descriptions, e.g. the unit-cell de-
scription based on the Penrose tiling is adopted too, but the density-wave description
based on the Laudau phenomenological theory on symmetry-breaking of condensed
matter has played the central role and been widely acknowledged. This means there
are two elementary excitations of low-energy, phonon u and phason w for quasicrys-
tals, in which vector u is in the parallel space E3 and vector w is in the perpendicular
3
space E⊥ , respectively. So the total displacement field for quasicrystals is
ū = u ⊕ u⊥ = u ⊕ w,
u = u(r ), w = w(r ),
i.e., u and w depend upon special radius vector r in parallel space E3 only, this is
formula (4.1-2). For simplicity the superscript of r is removed in the presentation
in Sections 4.2∼4.8.
48 Chapter 4 The physical basis of elasticity of quasicrystals
and
F0 : S,
F1 : L,
F2 : LS,
F3 : LSL,
F4 : LSLLS.
The Fibonacci sequence is ordered but not periodic. The geometric expression of
the sequence can be shown in the axis E1 i.e., E in Fig. 4.9-1 (a), and E is the so-
called parallel space, and that perpendicular to which is the so-called vertical space
E2 , i.e., E⊥ . The Fibonacci sequence is a useful tool to describe the geometry of
one-dimensional quasiperiodic structure, like that the Penrose tiling to describe the
geometry of two- and three-dimensional quasicrystals. The Fig. 4.9-1 may help us
to understand the internal-space E⊥ . For one-dimensional quasicrystals, the figure
can give an explicit description, while for two- and three-dimensional quasicrystals
there is no such an explicit graph.
Since quasicrystals belong to one of incommensurate phases, and there are pha-
son modes in the incommensurate crystals, denoted by w(r ), which may be un-
derstood as the corresponding new displacement field, if people have knowledge on
incommensurate phases, then they may easily understand the origin of phason modes
in quasicrystal, though conventional incommensurate crystals are not certainly the
actual quasicrystals.
4.9 Appendix of Chapter 4: Description on physical basis of elasticity... 49
The phonon variables u(r ) appearing in the physical space E3 , vector u rep-
resents the displacement of lattice point deviated from its equilibrium position due
to the vibration of the lattice. The propagation of this vibration is sound waves in
solids. Though vibration is a mechnical motion, which can be quantized, the quan-
tum of this motion is named phonon. So the u field is called phonon field in physical
terminology. The gradient of the u field characterizes the changes in volume and in
shape of cells—this is identical to that in the classical elasticity (see, e.g. Chapter
2, and previous sections of this chapter).
As mentioned before, the phason variables are substantively related with struc-
tural transitions of alloys, some of them can be observed from the characteristics of
diffraction patterns. Lubensky et al[5,7] and Horn et al[15] discussed the connection
between the phenomena and phason strain. These profound observations could not
be discussed here, reader can refer to the review given by Hu et al[16] . This makes us
know that the phason modes exist. The physical meaning of phason variables can be
explained as a quantity to describe the local rearrangement of atoms in a cell. We
know that the phase transition in crystalline materials is just induced by the atomic
local rearrangement. The unit-cell description on quasicrystals mentioned above
predicts that w describes the local arrangement of Penrose titling. These findings
50 Chapter 4 The physical basis of elasticity of quasicrystals
may help us to understand the meaning of the unusual field variables. Afterward ex-
perimental investigations by neutron scattering[17∼20] , Moessbauer spectroscopy[21] ,
NMR (Nuclear Magnetic Resonance)[22,23] and specific heat measurements[24,25] , the
concept of thermal-induced phason flips has been suggested, this is identical to the
diffusive essentiality of phasons. Note that the so-called diffusion here is quite differ-
ent from that in metallic periodic crystals (which mainly results from the presence
of vacancies in the lattice, and the vacancies are not necessary for atomic motion in
the quasicrystal structures). We will discuss this aspect in Chapter 10.
It should be noted that vector u and vector w are different in nature under
certain symmetry operations. This can be explained by the group theory. The
discussion is omitted here.
References
[1] Bak P. Phenomenological theory of icosahedral incommensurate (quaisiperiodic) order,
in Mn-Al alloys. Phys Rev Lett, 1985, 54(8): 1517–1519
[2] Bak P. Symmetry, stability and elastic properties of icosahedral incommensurate crys-
tals. Phys Rev B, 1985, 32(9): 5764–5772
[3] Landau L D, Lifshitz E M. Theoretical Physics V: Statistical Physics. 3rd ed. New
York: Pregamen Press, 1980
[4] Levine D, Lubensky T C, Ostlund S et al. Elasticity and dislocations in pentagonal
and icosahedral quasicrystals. Phys Rev Lett, 1985, 54(8): 1520–1523
[5] Lubensky T C, Ramaswamy S, Nad Toner J. Hydrodynamics of icosahedral quasicrys-
tals. Phys Rev B, 1985, 32(11): 7444–7452
[6] Lubensky T C, Ramaswamy S, Toner J. Dislocation motion in quasicrystals and im-
plications for macroscopic properties. Phys Rev B, 1986, 33(11): 7715–7719
[7] Lubensky T C, Socolar J E S, Steinhardt P J et al. Distortion and peak broadening
in quasicrystal diffraction patterns. Phys Rev Lett, 1986, 57(12): 1440–1443
[8] Lubensky T C. Introduction to Quasicrystals. Jaric M V. Boston: Academic Press,
1988
[9] Kalugin P A, Kitaev A, Levitov L S. 6-dimensional properties of Al0.86 Mn0.14 alloy.
J Phys Lett, 1985, 46(13): 601–607
[10] Torian S M, Mermin D. Mean-field theory of quasicrystalline order. Phys Rev Lett,
1985, 54(14): 1524–1527
[11] Jaric M V. Long-range icosahedral orientational order and quasicrystals. Phys Rev
Lett, 1985, 55(6): 607–610
[12] Duneau M, Katz A. Quasiperiodic patterns, Phys Rev Lett, 1985, 54(25): 2688–2691
[13] Socolar J E S, Lubensky T C, Steinhardt P J. Phonons, phasons, and dislocations in
quasicrystals. Phys Rev B, 1986, 34(5): 3345–3360
References 51
[14] Gahler F, Rhyner J. Equivalence of the generalised grid and projection methods for
the construction of quasiperiodic tilings. J Phys A: Math Gen, 1986, 19(2): 267–277
[15] Horn P M, Melzfeldt W, Di Vincenzo D P et al. Systematics of disorder in quasiperi-
odic material. Phys Rev Lett, 1986, 57(12): 1444–1447
[16] Hu C Z, Wang R H, Ding D H. Symmetry groups, physical property tensors, elasticity
and dislocations in quasicrystals. Rep Prog Phys, 2000, 63(1): 1–39
[17] Coddens G, Bellissent R, Calvayrac Y et al. Evidence for phason hopping in icosahe-
dral Al-Fe-Cu quasicrystals. Europhys Lett, 1991, 16(3): 271–276
[18] Coddens G, Sturer W. Time-of-flight neutron-scattering study of phason hopping in
decagonal Al-Co-Ni quasicrystals. 1999, Phys Rev B, 1999, 60(1): 270–276
[19] Coddens G, Lyonnard S, Hennion B et al. Triple-axis neutron-scattering study of
phason dynamics in Al-Mn-Pd quasicrystals. Phys Rev B 2000, 62(10): 6268–6295
[20] Coddens G, Lyonnard S, Calvayrac Y et al. Atomic (phason) hopping in perfect icosa-
hedral quasicrystals Al70.3 Pd21.4 Mn8.3 by time-of-flight quasielastic neutron scatter-
ing. Phys Rev B, 1996, 53(6): 3150–3160
[21] Coddens G, Lyonnard S, Sepilo B et al. Evidence for atomic hopping of Fe in perfectly
icosahedral Al-Fe-Cu quasicrystals by57 Fe Moessbauer spectroscopy. J Phys, 1995,
5(7): 771–776
[22] Dolisek J, Ambrosini B, Vonlanthen P et al. Atomic motion in quasicrystalline
Al70 Re8.6 Pd21.4 : A two-dimensional exchange NMR study. Phys Rev Lett, 1998, 81(17):
3671–3674
[23] Dolisek J, Apih T, Simsic M et al. Self-diffusion in icosahedral Al72.4 Pd20.5 Mn7.1 and
27
phason percolation at low temperatures studied by Al NMR. Phys Rev Lett, 1999,
82(3): 572–575
[24] Edagawa K, Kajiyama K. High temperature specific heat of Al-Pd-Mn and Al-Cu-Co
quasicrystals, Mater Sci and Eng A, 2000, 294∼296(5): 646–649
[25] Edagawa K, Kajiyama K, Tamura R et al. High-temperature specific heat of qua-
sicrystals and a crystal approximant. Mater Sci and Eng A, 2001, 312(1∼2): 293–298
[26] Ding D H, Yang W G, Hu C Z et al. Generalized elasticity theory of quasicrystals.
Phys Rev B, 1993, 48(10): 7003–7010
[27] Fan T Y, Wang X F, Li W et al. Elasto-/hydro-dynamics of quasicrystals. Phil Mag,
2009, 89(6), 501–512
[28] Edagawa K, Takeuchi S. Elasticity, dislocations and their motion in quasicrystals,
Dislocation in Solids. 2007, Chapter 76. ed. by Nabarro E R N, Hirth J P, 367–417
[29] Edagawa K, Giso Y. Experimental evaluation of phonon-phason coupling in icosahe-
dral quasicrystals. Phil Mag, 2007, 87(1): 77–95
[30] Meng X M, Tong B Y, Wu Y K. Mechanical properties of quasicrystal Al65 Cu20 Co15 .
Acta Metallurgica Sinica, 1994, 30(2): 61–64 (in Chinese)
52 Chapter 4 The physical basis of elasticity of quasicrystals
We here briefly list the crystal systems and Laue classes of one-dimensional qua-
sicrystals, in which the concept of point group must be concerned, and we do not
concern with the concept of space group.
or
[εxx , εyy , εzz , 2εyz , 2εzx , 2εxy , wzz , wzx , wzy ] , (5.1-4)
[σxx , σyy , σzz , σyz , σzx , σxy , Hzz , Hzx , Hzy ] (5.1-5)
where a short notation for the phonon elastic constant tensor is used, i.e., index
11→1, 22→2, 33→3, 23→4, 31→5, 12→6 and Cijkl is denoted as Cpq accordingly,
C11 = C1111 = C2222 , C12 = C1122 , C33 = C3333 , C44 = C2323 = C3131 ,
∂
= 0, (5.2-1)
∂z
Then
∂ui ∂wz
= 0, i = 1, 2, 3, = 0. (5.2-2)
∂z ∂z
Hence,
1 ∂uz 1 ∂uz
εzz = wzz = 0, εyz = εzy = , εzx = εxz = , (5.2-3)
2 ∂y 2 ∂x
∂σij ∂Hij
= 0, = 0. (5.2-4)
∂z ∂z
The generalized Hooke’s law is simplified as
⎧
⎪
⎪ σxx = C11 εxx + C12 εyy ,
⎪
⎪
⎪
⎪
⎪
⎪ σyy = C12 εxx + C11 εyy ,
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ σxy = σyx = 2C66 εxy ,
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ σzz = C13 (εxx + εyy ),
⎪
⎪
⎪
⎨
σyz = σzy = 2C44 εyz + R3 wzy ,
(5.2-5)
⎪
⎪
⎪
⎪
⎪
⎪ σzx = σxz = 2C44 εzx + R3 wzx ,
⎪
⎪
⎪
⎪
⎪
⎪ Hzz = R1 (εxx + εyy ),
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ H = 2R3 εzx + K2 wzx ,
⎪ zx
⎪
⎪
⎪
⎪
⎪
⎪
⎩ Hzy = 2R3 εyz + K2 wzy .
In the absence of the body force and generalized body force the equilibrium equations
are
∂σxx ∂σxy ∂σyx ∂σyy ∂σzx ∂σzy
+ = 0, + = 0, + = 0, (5.2-6)
∂x ∂y ∂x ∂y ∂x ∂y
∂Hzx ∂Hzy
+ = 0. (5.2-7)
∂x ∂y
5.2 Decomposition of the problem into plane and anti-plane problems 57
this is the classical plane elasticity of conventional hexagonal crystals. The second
one is ⎧
⎪
⎪ σyz = σzy = 2C44 εyz + R3 wzy ,
⎪
⎪
⎪
⎪
⎪
⎪ σzx = σxz = 2C44 εzx + R3 wzx ,
⎪
⎪
⎪
⎪
⎪
⎪ Hzx = 2R3 εzx + K2 wzx ,
⎪
⎪
⎪
⎪
⎪
⎪ Hzy = 2R3 εyz + K2 wzy ,
⎨
∂σzx ∂σzy ∂Hzx ∂σzy (5.2-9)
⎪
⎪ + = 0, + = 0,
⎪
⎪
⎪
⎪
∂x ∂y ∂x ∂y
⎪
⎪
⎪
⎪ 1 ∂uz 1 ∂uz
⎪
⎪ εzx = = εxz , εzy = = εyz ,
⎪
⎪ 2 ∂x 2 ∂y
⎪
⎪
⎪
⎪
⎪
⎪ ∂wz ∂wz
⎩ wzx = , wzy = ,
∂x ∂y
which is a phonon-phason coupling elasticity problem, involving only the displace-
ments uz and wz . It is an anti-plane elasticity problem.
The plane elasticity described by (5.2-8) has been studied, extensively using the
stress function approach, e.g. it introduces
∇2 ∇2 U = 0.
∇2 uz = 0, ∇2 wz = 0, (5.2-11)
∂ ∂
where ∇2 = 2
+ 2 , so uz and wz are harmonic functions.
∂x ∂y
It is well known that the two-dimensional harmonic functions uz and wz can be
a real part or an imaginary part of any analytic functions φ(t) and ψ(t) of complex
√
variable t = x + iy, i = −1, respectively, i.e.,
uz (x, y) = Re φ(t),
(5.2-12)
wz (x, y) = Re ψ(t).
⎧
⎪ σxx = C11 εxx + C12 εyy + C13 εzz + 2C16 εxy + R1 wzz ,
⎪
⎪
⎪
⎪
⎪
⎪ σyy = C12 εxx + C22 εyy + C23 εzz + 2C26 εxy + R2 wzz ,
⎪
⎪
⎪
⎪
⎪
⎪ σzz = C13 εxx + C23 εyy + C33 εzz + 2C36 εxy + R3 wzz ,
⎪
⎪
⎪
⎪
⎪
⎪ σyz = σzy = 2C44 εyz + 2C45 εzx + R4 wzx + R5 wzy ,
⎨
σzx = σxz = 2C45 εyz + 2C55 εzx + R6 wzx + R7 wzy , (5.3-1)
⎪
⎪
⎪
⎪
⎪
⎪ σxy = σyx = C16 εxx + C26 εyy + C36 εzz + 2C66 εxy + R8 wzz ,
⎪
⎪
⎪
⎪
⎪
⎪ Hzx = 2R4 εyz + 2R6 εzx + K1 wzx + K4 wzy ,
⎪
⎪
⎪
⎪
⎪
⎪ Hzy = 2R5 εyz + 2R7 εzx + K4 wzx + K2 wzy ,
⎪
⎩
Hzz = R1 εxx + R2 εyy + R3 εzz + 2R8 εxy + K3 wzz ,
where the short notation is used for the phonon elastic constant tensor, such that
the index 11→1,22→2,33→3, 23→4,31→5,12→6 and Cijkl is denoted as Cpq ; for the
phason elastic constants, K3131 = K1 , K3232 = K2 , K3333 = K3 ,K3132 = K4 and for
the phonon-phason coupling elastic constants, R1133 = R1 , R2233 = R2 , R3333 = R3 ,
R2331 = R4 , R2332 = R5 , R3131 = R6 , R3132 = R7 , R1233 = R8 .
Under the assumption (5.2-1), the problem can be decomposed into two separate
problems as follows:
⎧
⎪
⎪ σxx = C11 εxx + C12 εyy + 2C16 εxy ,
⎪
⎪
⎨ σyy = C12 εxx + C22 εyy + 2C26 εxy ,
σxy = σyx = C16 εxx + C26 εyy + 2C66 εxy , (5.3-2)
⎪
⎪
⎪
⎪ σzz = C 13 ε xx + C 23 εyy + 2C36 ε xy ,
⎩
Hzz = R1 εxx + R2 εyy + R3 εzz + 2R8 εxy ,
and ⎧
⎪
⎪ σyz = σzy = 2C44 εyz + 2C45 εzx + R4 wzx + R5 wzy ,
⎨
σzx = σxz = 2C45 εyz + 2C55 εzx + R6 wzx + R7 wzy ,
(5.3-3)
⎪
⎪ Hzx = 2R4 εyz + 2R6 εzx + K1 wzx + K4 wzy ,
⎩
Hzy = 2R5 εyz + 2R7 εzx + K4 wzx + K2 wzy
in which the problem described by equations (5.3-2) is plane elasticity of monocline
crystals. By introducing the displacement potential G(x, y),
∂2 ∂2 ∂2
ux = C16 2 + C26 2 + (C12 + C66 ) G,
∂x ∂y ∂x∂y
∂2 ∂2 ∂2
uy = − C11 2 + C66 2 + 2C16 G,
∂x ∂y ∂x∂y
with constants
2
c1 = C16 − C11 C66 , c2 = 2(C16 C12 − C11 C26 ),
2
c3 = C12 − 2C16 C26 + 2C12 C66 − C11 C22 ,
2
c4 = 2(C26 C12 − C16 C22 ), c5 = C26 − C22 C66 .
Since this is classical elasticity and does not have a direct connection to phason
elasticity of one-dimensional quasicrystal, we do not consider it further.
We are interested in the problem described by equation (5.3-3), which is a
phonon-phason coupling problem. Substituting the equations of deformation ge-
ometry into the stress-strain relations and then into the equilibrium equations, we
obtain the governing equation:
∂4 ∂4 ∂4 ∂4 ∂4
a1 4 + a2 3 + a3 2 2 + a4 + a5 4 F = 0, (5.3-4)
∂x ∂x ∂y ∂x ∂y ∂x∂y 3 ∂y
where ⎧
⎪ ∂2 ∂2 ∂2
⎪
⎪ u = R 6 + R 5 + (R 4 + R 7 ) F,
⎨ z ∂x2 ∂y 2 ∂x∂y
(5.3-5)
⎪
⎪ ∂2 ∂2 ∂2
⎪
⎩ wz = − C55 + C + 2C F,
44 45
∂x2 ∂y 2 ∂x∂y
and
⎧
⎨ a1 = R62 − K1 C56 , a2 = 2(R6 (R4 + R7 ) − K1 C45 − K4 C55 ),
a3 = 2R5 R6 + (R4 + R7 )2 − K1 C44 − K2 C55 − 4K4 C45 , (5.3-6)
⎩
a4 = 2[R5 (R4 + R7 ) − K2 C45 − K4 C44 ], a5 = R52 − K2 C44 .
μk = αk + iβk (5.3-8)
a5 μ4 + a4 μ3 + a3 μ2 + a2 μ + a1 = 0 (5.3-9)
and μ1 = μ2 .
If the roots of equation (5.3-9) are double roots, i.e., μ1 = μ2 , then
Substitution of formula (5.3-7) into (5.3-5) and then into (5.3-3) we arrive at the
complex representation of the displacements and stresses as follows:
⎧
⎪ 2
⎪
⎪
⎪
⎪ u = 2Re [R6 + (R4 + R7 )μk + R5 μ2k ]fk (zk ),
⎪
⎪
z
⎪
⎪ k=1
⎪
⎪
⎪
⎪ 2
⎪
⎪
⎪
⎪ w z = −2Re (C55 + 2C45 μk + C44 μ2k )fk (zk ),
⎪
⎪
⎪
⎪ k=1
⎪
⎪
⎪
⎪ σ = σ
⎪
⎪
zy yz
⎪
⎪ 2
⎪
⎪
⎪
⎪ = 2Re [R6 C45 − R4 C55 + (R6 C44 − R4 C45 + R7 C45 − R5 C55 )μk
⎪
⎪
⎪
⎪
⎪
⎪
k=1
⎪
⎪ +(R7 C44 − R5 C45 )μ2k ]fk (zk ),
⎪
⎪
⎪
⎪
⎨ σzx = σxz
⎪
⎪ 2
⎪
⎪
⎪
⎪ = 2Re [R4 C55 − R6 C45 + (R5 C55 + R4 C45 − R6 C44 − R5 C55 )μk
⎪
⎪
⎪
⎪ k=1
⎪
⎪
⎪
⎪ +(R3 C45 − R7 C44 )μ2k ]μk fk (zk ),
⎪
⎪
⎪
⎪ 2
⎪
⎪
⎪
⎪ H = 2Re [(R7 + R5 μk )(R6 + R4 μk + R7 μk + R5 μ2k )
⎪
⎪ zx
⎪
⎪
⎪
⎪
k=1
⎪
⎪ −(K4 + K2 μk )(C55 + 2C45 μk + C44 μ2k )]fk (zk ),
⎪
⎪
⎪
⎪
⎪
⎪ 2
⎪
⎪
⎪
⎪ H = 2Re [(R6 + R4 μk )(R6 + R4 μk + R7 μk + R5 μ2k )
⎪
⎪
zy
⎪
⎪ k=1
⎪
⎩ −(K + K μ )(C + 2C μ + C μ2 )]f (z ),
4 2 k 55 45 k 44 k k k
(5.3-11)
2
where fk (zk ) ≡ ∂ Fk (zk ) ∂zk2
= Fk (zk ).
Determination of analytic functions Fk (zk ) must be supplemented with the bound-
ary conditions of specific problems, which will be given in Chapters 7 and 8.
∇2 ∇2 F = 0. (5.5-2)
5.6 The space elasticity of hexagonal quasicrystals 63
∂2 ∂2 2 ∂
2
∇2i = + + γ , i = 1, 2, 3, 4 (5.6-4)
∂x2 ∂y 2 i
∂z 2
with mi , li and γi defined by
∂2
Hzx = R3 [(m1 + 1)F1 + (m2 + 1)F2 + (m3 + 1)F3 ]
∂x∂z
∂ 2 F4 ∂2
− R3 + K2 (l1 F1 + l2 F2 + l3 F3 ),
∂y∂z ∂x∂z
∂2
Hzy = R3 [(m1 + 1)F1 + (m2 + 1)F2 + (m3 + 1)F3 ]
∂y∂z
∂ 2 F4 ∂2
+ R3 + K2 (l1 F1 + l2 F2 + l3 F3 ).
∂y∂z ∂y∂z
Harmonic equations (5.6-3) can be solved under appropriate boundary condi-
tions, which will be discussed in Chapter 8.
References
[1] Wang R H, Yang W G, Hu C Z et al. Point and space groups and elastic behaviour
of one-dimensional quasicrystals. J Phys Condens Matter, 1997, 9(11): 2411–2422
[2] Fan T Y. Mathematical theory of elasticity and defects of quasicrystals. Advances
in Mechanics, 2000, 30(2): 161–174 (in Chinese)
[3] Fan T Y, Mai Y W. Elasticity theory, fracture mechanics and some relevant thermal
properties of quasicrystalline materials. Appl Mech Rev, 2004, 57(5): 325–344
[4] Liu G T, Fan T Y, Guo R P. Governing equations and general solutions of plane
elasticity of one-dimensional quasicrystals. Int J Solid and Structures, 2004, 41(14):
3949–3959
[5] Liu G T. The complex variable function method of the elastic theory of quasicrys-
tals and defects and auxiliary equation method for solving some nonlinear evalution
equations, Dissertation. Beijing Institute of Technology, 2004 (in Chinese)
[6] Peng Y Z, Fan T Y. Elastic theory of 1-D quasiperiodic stacking 2-D crystals. J.
Phys: Condens Matter, 2000, 12(45): 9381–9387
[7] Peng Y Z. Study on elastic three-dimensional problems of cracks for quasicrystals,
Dissertation. Beijing Institute of Technology, 2002 (in Chinese)
66 Chapter 5 Elasticity theory of one-dimensional quasicrystals and simplification
the quasiperiodic plane, and axis z (or x3 ) as the periodic axis, then xy- plane is
the elasticity isotropic plane, within which the elastic constants are
C1111 = C2222 = C11 ,
C1122 = C12 ,
C1212 = C1111 − C1122 = C11 − C12 = 2C66 .
This shows that C66 is not independent. Other independent elastic constants are
out of the plane, i.e.,
C2323 = C3131 = C44 ,
C1133 = C2233 = C13 ,
C3333 = C33 .
which are listed in Table 6.0-2.
The relevant phason elastic constants and phonon-phason coupling elastic con-
stants are listed in Tables 6.0-3∼6.0-6.
Table 6.0-5 The phason elastic constants for Laue class 15 (Kijkl )
11 22 23 12 13 21
11 K1 K2 0 K5 0 K5
22 K2 K1 0 −K5 0 −K5
23 0 0 K4 0 0 0
12 K5 −K5 0 K 0 K3
13 0 0 0 0 K4 0
21 K5 −K5 0 K3 0 K
K = K1 + K2 + K3 .
in which B is the bulk modulus, G the shear modulus, and ν the Poisson’s ratio,
respectively.
The phason elastic constants, for a decagonal Al-Ni-Co quasicrystal, anisotropic
diffuse scattering has been observed in synchrotron X-ray diffraction measurements[2] .
It has been shown that the measurement can attributed the phason elastic constants,
although no quantitative evaluation on K1 and K2 . The Monto-Carlo simulation
was used to evaluate the phason elastic constants, e.g. given by Table 6.0-8[3] .
2
where 1GPa = 1010 dyn/cm . It should be noted that the values given by the Monte-
Carlo simulation are subjected to verification.
Recently the experimental measurement for phonon-phason coupling elastic con-
stants for decagonal quasicrystals have been achieved, the results are listed in Table
6.0-9.
Table 6.0-9 Coupling elastic constants for point group 10, 10, Al-Ni-Co
decagonal quasicrystals[4]
Alloy R1 /GPa |R2 |/GPa
Al-Ni-Co −1.1 <0.2
Tables 6.0-2∼6.0-6 listed all the elastic constants. Consider a plane in two-
dimensional quasicrystals, and assume that it is perpendicular to the periodic sym-
metrical axis (say, axis z). In this case,
so the strains are wzz = wzx = wxz = wzy = wyz = 0. Assumption (6.0-1) leads to
εzz = εxz = εyz = 0, and Table 6.0-2 is simplified to the following Table 6.1-1 for
the phonon elastic constants for the plane elasticity.
The phonon elastic constants listed in Table 6.1-1 can be expressed in the fourth-
order tensor
Table 6.1-2 Phason elastic constants for plane elasticity for point group
10mm quasicrystals
11 22 12 21
11 K1 K2 0 0
22 K2 K1 0 0
12 0 0 K1 −K2
21 0 0 −K2 K1
This means
K1111 = K2222 = K2121 = K1 ,
(6.1-3)
K1122 = K2211 = −K2112 = −K1221 = K2
74 Chapter 6 Elasticity of two-dimensional quasicrystals and simplification
and other Kijkl = 0, and the expression of them by tensor of four rank is
Table 6.0-4 in the present case is simplified into Table 6.1-3 as follows.
One can find that for plane elasticity point group 5m has the same elastic constants
with point group 10mm, so they can be discussed in the same line.
The definition of strain tensor are as given in Chapter 4, i.e. ,
1 ∂ui ∂uj ∂wi
εij = + , wij = . (6.1-7)
2 ∂xj ∂xi ∂xj
In Chapter 4 it was seen that the stress, strain and elastic constant tensors can be
expressed by matrices. The above mentioned elastic constants may be denoted by
matrix[CKR]. For the present case, the strain vector defined by (4.4-12) is simplified
into
[εij wij ] = [ε11 ε22 ε12 ε21 w11 w22 w12 w21 ] (6.1-8)
and [CKR] is
⎡ ⎤
L + 2M L 0 0 R R 0 0
⎢ L L + 2M 0 0 −R −R 0 0 ⎥
⎢ ⎥
⎢ 0 0 M M 0 0 −R R ⎥
⎢ ⎥
⎢ 0 0 M M 0 0 −R R ⎥
[CKR] = ⎢
⎢
⎥
⎥
⎢ R −R 0 0 K 1 K2 0 0 ⎥
⎢ R −R 0 0 K2 K1 0 0 ⎥
⎢ ⎥
⎣ 0 0 −R −R 0 0 K1 −K2 ⎦
0 0 R R 0 0 −K2 K1
6.1 Basic equations of plane elasticity of two-dimensional quasicrystals: point . . . 75
⎡ ⎤
L + 2M L 0 0 R R 0 0
⎢ L + 2M 0 0 −R −R 0 0 ⎥
⎢ ⎥
⎢ M M 0 0 −R R ⎥
⎢ ⎥
⎢ M 0 0 −R R ⎥
=⎢ ⎥ . (6.1-9)
⎢ K1 K2 0 0 ⎥
⎢ ⎥
⎢ (symmetry) K1 0 0 ⎥
⎢ ⎥
⎣ K1 −K2 ⎦
K1
The free energy (or strain energy density) is
1 1
F = Lεii εii + M εij εij + K1 wij wij + K2 (wxx wyy − wxy wxy )
2 2
+ R [(εxx − εyy )(wxx + wyy ) + 2εxy (wxy − wyx )] (6.1-10)
By (6.1-9) and (4.5-3) or by (6.1-10) and (4.5-1), the generalized Hooke’s law for
plane elasticity of point group 10mm quasicrystals of ten-fold symmetry is
⎧
⎪ σxx = L(εxx + εyy ) + 2M εxx + R(wxx + wyy ),
⎪
⎪
⎪
⎪
⎪
⎪ σyy = L(εxx + εyy ) + 2M εyy − R(wxx + wyy ),
⎪
⎪
⎪
⎪
⎨ σxy = σyx = 2M εxy + R(wyx − wxy ),
⎪
Hxx = K1 wxx + K2 wyy + R(εxx − εyy ), (6.1-11)
⎪
⎪
⎪
⎪ Hyy = K1 wyy + K2 wxx + R(εxx − εyy ),
⎪
⎪
⎪
⎪
⎪
⎪ Hxy = K1 wxy − K2 wyx − 2Rεxy ,
⎪
⎪
⎩
Hyx = K1 wyx − K2 wxy + 2Rεxy .
σxz = 2M εxz ,
(6.1-12)
σyz = 2M εyz .
Monograph [9] pointed first out that (6.1-9)∼(6.1-11) hold for the plane elasticity
for both point group 5m and point group 10mm, this is identical to the argument of
Ref. [6].
Equations (6.1-11) are the physical basis of elasticity of point groups 5m and
10mm quasicrystals. The geometry (or kinetics) basis of the subject is equation
(6.1-7). Another necessary basis comes from statics, i.e.,
⎧
⎪ ∂σxx ∂σxy ∂σyx ∂σyy
⎪
⎨ ∂x + ∂y = 0, + = 0,
∂x ∂y
(6.1-13)
⎪
⎪ ∂Hxx ∂Hxy ∂Hyx ∂Hyy
⎩ + = 0, + = 0.
∂x ∂y ∂x ∂y
76 Chapter 6 Elasticity of two-dimensional quasicrystals and simplification
in which
∂2 ∂2 ∂ux ∂uy
∇2 = 2
+ 2, ∇·u= + .
∂x ∂y ∂x ∂y
then the first two equations are automatically satisfied already. Substituting (6.2-2)
78 Chapter 6 Elasticity of two-dimensional quasicrystals and simplification
into the last two equations in (6.2-1), there follows the equations
⎧
⎪
⎪ ∂2ϕ ∂2 ∂2
⎪
⎨ (αΠ 1 + βΠ 2 ) + αΠ1 − βΠ2 ψ = 0,
∂x∂y ∂y2 ∂x2
(6.2-3)
⎪
⎪ ∂2 ∂2 ∂2ψ
⎪
⎩ αΠ2 2 − βΠ1 2 ϕ + (αΠ2 + βΠ1 ) = 0.
∂x ∂y ∂x∂y
where
∂2 ∂2 ∂2 ∂2
Π1 = 3 2
− 2, Π2 = 3 2
− 2, (6.2-4)
∂x ∂y ∂y ∂x
⎧
⎨ α = R(L + 2M ) − ωK1 , β = RM − ωK1 ,
(6.2-5)
⎩ ω = M (L + 2M ) .
R
Equation (6.2-3) is much simpler compared with (6.2-1), still it can be simplified by
letting
∂2 ∂2 ∂2F
ϕ = βΠ2 2 − αΠ1 2 F, ψ = (αΠ1 + βΠ2 ) . (6.2-6)
∂x ∂y ∂x∂y
in which F (x, y) is any function, then the first equation of (6.2-3) is satisfied. Substi-
tuting (6.2-6) into the second one of equations (6.2-3), after manipulation, it reduces
to
∇2 ∇2 ∇2 ∇2 F = 0. (6.2-7)
This is the final governing equation of the plane elasticity of point groups 5m and
10mm quasicrystals. We call F (x, y) as a displacement potential function, or simply
displacement potential. The equation (6.2-7) is a quadruple harmonic equation, the
order of which is much higher than that in the classical elasticity, where that is
biharmonic equation.
All of displacement and stress components can be expressed by potential function
F (x, y) as follows:
⎧
⎪
⎪ ∂2 2
⎨ ux = [M αΠ1 + (L + 2M )βΠ2 ] ∂x∂y ∇ F,
⎪
(6.2-8)
⎪
⎪ ∂2 ∂2
⎪
⎩ uy = M αΠ1 2 − (L + 2M )βΠ2 2 ∇2 F,
∂y ∂x
⎧
⎪
⎪ ∂2
⎨ wx = ω(α − β)Π1 Π2 ∂x∂y F,
⎪
(6.2-9)
⎪
⎪ 2 ∂
2
2 ∂
2
⎪
⎩ wy = −ω αΠ1 2 + βΠ2 2 F,
∂y ∂x
6.3 Simplification of the basic equations set: stress potential function method 79
⎧
⎪ ∂3 2
⎪
⎪ σ = 2M (L + M )αΠ 1 ∇ F,
⎪
⎪
xx
∂y 3
⎪
⎪
⎪
⎪
⎨
∂3
σyy = 2M (L + M )αΠ1 2 ∇2 F, (6.2-10)
⎪
⎪ ∂x ∂y
⎪
⎪
⎪
⎪
⎪
⎪ 3
⎪
⎩ σxy = σyx = −2M (L + M )αΠ1 ∂ ∇2 F,
∂x∂y 2
⎧
⎪ ∂ 2 2 2 ∂ 2 ∂2 2 ∂2
⎪
⎪ Hxx = αβ ∇ ∇ ∇ F + ω(K1 − K2 ) αΠ1 2 + βΠ2 2 F,
⎪
⎪ ∂y ∂y ∂y ∂x
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ ∂ ∂3
⎪
⎪ Hyy = αβ ∇2 ∇2 ∇2 F − ω(K1 − K2 )(α − β)Π1 Π2 2 F,
⎨ ∂y ∂x ∂y
⎪
⎪
⎪ ∂ 2 2 2 ∂ 2 2
⎪ 2 ∂ 2 ∂
⎪
⎪ H xy = −αβ ∇ ∇ ∇ F − ω(K 1 − K 2 ) αΠ1 + βΠ 2 F,
⎪
⎪ ∂x ∂x ∂y 2 ∂x2
⎪
⎪
⎪
⎪
⎪
⎪ 3
⎪
⎩ Hyx = αβ ∂ ∇2 ∇2 ∇2 F + ω(K1 − K2 )(α − β)Π1 Π2 ∂ F.
∂x ∂x∂y 2
(6.2-11)
[12]
Li and Fan have suggested the approach, and the practice shows that it is very
effective. In the next two chapters many applications of the approach will be given.
In addition (6.1-12) and (6.1-14) yield the final governing equation for anti-plane
elasticity
∇2 uz = 0. (6.2-12)
It is obvious that the anti-plane problem is decoupled with the plane one.
The strain components εij and wij can be expressed by the stress components σij
80 Chapter 6 Elasticity of two-dimensional quasicrystals and simplification
yields
⎧
⎪
⎪ 1 2 2 K1 + K 2 2 2 R ∂ ∂
⎪
⎪ ∇ ∇ ϕ+ ∇ ∇ ϕ+ Π1 ψ1 − Π2 ψ2 = 0,
⎪
⎪ 2C(L + M ) 2C C ∂y ∂x
⎪
⎪
⎨ C
∂
+ M ∇2 ψ1 + R Π1 ϕ = 0,
⎪ K1 − K2
⎪ ∂y
⎪
⎪
⎪
⎪
⎪
⎪ C ∂
⎩ + M ∇2 ψ2 − R Π2 ϕ = 0.
K1 − K2 ∂x
(6.3-6)
in which Π1 and Π2 are defined by (6.2-4), C is given by (6.3-3). By now, the
numbers of equations and unknown functions have been reduced to 3.
Now introducing a new unknown function G(x, y), such as
⎧
⎪
⎪ ϕ = D ∂ Π1 ∇2 G,
⎪
⎪
⎪
⎪ ∂y
⎪
⎪
⎪
⎪ 1
⎪ 2 2 2 2
⎨ ψ1 = − R (M K1 − R2 )[(L + 2M )(K1 − K2 ) − 2R ]∇ ∇ ∇ G
⎪
(6.3-7)
⎪
⎪ ∂2
⎪
⎪ + (L + M )(K1 − K2 )R Π1 Π2 G,
⎪
⎪ ∂x∂y
⎪
⎪
⎪
⎪ 2
⎪
⎪ ∂
⎩ ψ2 = (L + M )(K1 − K2 )R Π1 Π2 G.
∂x∂y
If
∇2 ∇2 ∇2 ∇2 G = 0, (6.3-8)
At the same time the equation (6.2-12) holds for anti-plane elasticity too.
Apart from this, the phonon and phason elastic constants of point group 5,5 and
point group 10, 10 quasicrystals are the same as those of point groups 5m and 10mm.
The corresponding elastic constant matrix is
⎡ ⎤
L + 2M L 0 0 R1 R1 R2 −R2
⎢ ⎥
⎢ L L + 2M 0 0 −R1 −R1 −R2 R2 ⎥
⎢ ⎥
⎢ ⎥
⎢ 0 0 M M R2 R2 −R1 R1 ⎥
⎢ ⎥
⎢ ⎥
⎢ 0 0 M M R2 R2 −R1 R1 ⎥
[CKR] = ⎢ ⎢ ⎥
⎥
⎢ R1 −R1 R2 R2 K1 K2 0 0 ⎥
⎢ ⎥
⎢
⎢ R1 −R1 R2 R2 K2 K1 0 0 ⎥ ⎥
⎢ ⎥
⎢
⎣ R2 −R2 −R1 −R1 0 0 K1 −K2 ⎥ ⎦
−R2 R2 R1 R1 0 0 −K2 K1
⎡ ⎤
L + 2M L 0 0 R1 R1 R2 −R2
⎢ ⎥
⎢ L + 2M 0 0 −R1 −R1 −R2 R2 ⎥
⎢ ⎥
⎢ ⎥
⎢ M M R2 R2 −R1 R1 ⎥
⎢ ⎥
⎢ ⎥
⎢ M R2 R2 −R1 R1 ⎥
=⎢
⎢
⎥.
⎥
⎢ (symmetry) K1 K2 0 0 ⎥
⎢ ⎥
⎢ 0 ⎥
⎢ K1 0 ⎥
⎢ ⎥
⎢ −K2 ⎥
⎣ K1 ⎦
K1
(6.4-2)
By this elastic constant matrix, the stress-strain relation can be written as
⎧
⎪ σxx = L(εxx + εyy ) + 2M εxx + R1 (wxx + wyy ) + R2 (wxy − wyx ),
⎪
⎪
⎪
⎪
⎪
⎪ σyy = L(εxx + εyy ) + 2M εyy − R1 (wxx + wyy ) − R2 (wxy − wyx ),
⎪
⎪
⎪
⎪
⎪
⎪ σ = σyx = 2M εxy + R1 (wyx − wxy ) + R2 (wxx + wyy ),
⎨ xy
Hxx = K1 wxx + K2 wyy + R1 (εxx − εyy ) + 2R2 εxy , (6.4-3)
⎪
⎪
⎪
⎪
⎪
⎪ Hyy = K1 wyy + K2 wxx + R1 (εxx − εyy ) + 2R2 εxy ,
⎪
⎪
⎪
⎪ Hxy = K1 wxy − K2 wyx − 2R1 εxy + R2 (εxx − εyy ),
⎪
⎪
⎪
⎩
Hyx = K1 wyx − K2 wxy + 2R1 εxy − R2 (εxx − εyy ).
In addition, the stresses σij and Hij satisfy the same equilibrium equations as
(6.1-13).
Substituting (6.1-7) into (6.4-3), then into (6.1-12) leads to the equilibrium equa-
6.4 Plane elasticity of point group 5, 5 pentagonal and point group 10, 10 . . . 83
in which
M (L + 2M )
ω= , R2 = R12 + R22 . (6.4-6)
R2
The functions ϕ(x, y) and ψ(x, y) defined by (6.4-4) automatically satisfy the first
two of equations (6.4-4), and the substitution of formulas (6.4-6) into the second two
84 Chapter 6 Elasticity of two-dimensional quasicrystals and simplification
Then the first one of equations (6.4-7) is identically satisfied, and from the second
one we find that
∇2 ∇2 ∇2 ∇2 F = 0. (6.4-11)
∂ 2 2 2 ∂
Hxx = −c1 c2 R ∇ ∇ ∇ F + R−1 K0 [c2 (L + 2M )Λ21 + c1 M Λ22 ]F, (6.4-12h)
∂y ∂y
∂ 2 2 2 ∂
Hxy = c1 c2 R ∇ ∇ ∇ F − R−1 K0 [c2 (L + 2M )Λ21 + c1 M Λ22 ]F, (6.4-12i)
∂x ∂x
∂ 2 2 2 ∂
Hyx = −c1 c2 R ∇ ∇ ∇ F − c0 K0 Λ1 Λ2 F, (6.4-12j)
∂x ∂y
∂ 2 2 2 ∂
Hyy = −c1 c2 R ∇ ∇ ∇ F + c0 K0 Λ1 Λ2 F (6.4-12k)
∂y ∂x
with
c0 = R(L + M ), K0 = K1 − K2 . (6.4-13)
The results were given by Li and Fan[15] and Li et al[16] . Recently, Li and Fan[14] have
derived the final governing equation of elasticity of the same point groups through the
stress potential method, the resulting equation is also quadruple harmonic equation,
of course the unknown function is the stress potential. Application will be shown
in Chapter 8 for solving notch problem of point group 5, 5 and point group 10, 10
two-dimensional quasicrystals.
The frequent appearance of the quadruple harmonic equations in (6.2-7), (6.3-8)
and (6.4-11), suggests that this kind of equations is very important in theory and
practice.
C11 − C12
L = C12 , M= = C66 . (6.5-1)
2
86 Chapter 6 Elasticity of two-dimensional quasicrystals and simplification
+ K3 δil δjk + 2(K2 + K3 )(δi1 δj2 δk1 δl2 + δi2 δj1 δk2 δl1 ),
i, j, k, l = 1, 2. (6.5-3)
Rijkl = 0. (6.5-4)
and there are the equilibrium equations in the absence of the body force
⎧
⎪
⎪ ∂σxx ∂σxy ∂σyx ∂σyy
⎪
⎨ ∂x + ∂y = 0, + = 0,
∂x ∂y
(6.5-9)
⎪
⎪ ∂Hxx ∂Hxy ∂Hyx ∂Hyy
⎪
⎩ + = 0, + = 0.
∂x ∂y ∂x ∂y
Eliminating the stress and strain components from (6.5-7)∼(6.5-9), we obtain the
equilibrium equations expressed by displacement components as below:
⎧
⎪ ∂
⎪
⎪ M ∇2 ux + (L + M ) ∇ · u = 0,
⎪
⎪ ∂x
⎪
⎪
⎪
⎪
⎪
⎪ ∂
⎪
⎪ M ∇2 uy + (L + M ) ∇ · u = 0,
⎨ ∂y
(6.5-10)
⎪
⎪ ∂ ∂wx ∂wy
⎪
⎪ K 1 ∇ 2
w + (K 2 + K3 ) + = 0,
⎪
⎪
x
∂y ∂y ∂x
⎪
⎪
⎪
⎪
⎪
⎪ ∂ ∂wx ∂wy
⎪
⎩ K1 ∇2 wy + (K2 + K3 ) + = 0.
∂x ∂y ∂x
∇2 ∇2 F = 0, ∇2 ∇2 G = 0, (6.5-12)
√
z = x + iy(i = −1), then
⎧
⎪
⎪
⎪ F (x, y) = Re z̄ϕ1 (z) + ψ1 (z)dz ,
⎨
(6.5-13)
⎪
⎪
⎪
⎩ G(x, y) = Re z̄π1 (z) + χ1 (z)dz .
phason fields there is coupling, the corresponding elastic constants Rijkl are the
same given by (6.1-6), i.e.,
Rijkl = R(δi1 − δi2 )(δij δkl − δik δjl + δil δjk ), i, j, k, l = 1, 2. (6.6-1)
⎧ 2
⎪ ∂ ∂ wx ∂ 2 wy ∂ 2 wx
⎪
⎪
2
M ∇ ux + (L + M ) ∇ · u + R +2 − = 0,
⎪
⎪ ∂x ∂x2 ∂x∂y ∂y 2
⎪
⎪
⎪
⎪ 2
⎪
⎪ ∂ ∂ wy ∂ 2 wx ∂ 2 wy
⎪
⎪ 2
⎨ M ∇ uy + (L + M ) ∂y ∇ · u + R ∂x2 − 2 ∂x∂y − ∂y 2
⎪ = 0,
⎪ 2 2
⎪
⎪ ∂ wx ∂ 2 wy ∂ ux ∂ 2 uy ∂ 2 ux
⎪ K1 ∇ wx + (K2 + K3 )
⎪
2
+ +R −2 − = 0,
⎪
⎪ ∂y 2 ∂x∂y ∂x2 ∂x∂y ∂y2
⎪
⎪
⎪
⎪ 2 2
⎪
⎪ ∂ wx ∂ 2 wy ∂ uy ∂ 2 ux ∂ 2 uy
⎪
⎩ K1 ∇ wy + (K2 + K3 )
2
+ +R +2 − = 0.
∂x∂y ∂x2 ∂x2 ∂x∂y ∂y 2
(6.6-5)
It is evident that, if K2 + K3 = 0, the equations will be reduced to (6.2-1). In
fact, K2 + K3 = 0, so the equations are more complex than those given in the
previous sections. But the final governing equation for the present case presents
more interesting in mathematical physics, we can see immediately.
At first we introduce two auxiliary function ϕ(x, y) and ψ(x, y) in such way
⎧
⎪ ∂2ϕ ∂2ψ ∂2ψ
⎪
⎪ u = (L + M ) + M + (L + 2M ) ,
⎪
⎪
x
∂x∂y ∂x2 ∂y 2
⎪
⎪
⎪
⎪
⎪
⎪ ∂2ϕ ∂2ϕ ∂2ψ
⎪
⎪ u = − (L + 2M ) 2 + M 2 + (L + M )
⎪
⎨ y ∂x ∂y ∂x∂y
,
⎪ 2 (6.6-6)
⎪
⎪ ∂ ϕ ∂2ψ ∂2ψ
⎪ wx = −ω 2
⎪ + − ,
⎪
⎪ ∂x∂y ∂x2 ∂y 2
⎪
⎪
⎪
⎪ 2
⎪
⎪ ∂ ϕ ∂2ϕ ∂2ψ
⎪
⎩ wy = ω − 2 −2 .
∂x2 ∂y ∂x∂y
⎧
⎪
⎪ ∂2ϕ ∂2 ∂2
⎨ (γΠ1 + δΠ2 ) ∂x∂y + αΠ1 ∂y 2 − βΠ2 ∂x2 ψ = 0,
⎪
(6.6-7)
⎪
⎪ ∂2ψ ∂2 ∂2
⎪
⎩ (γΠ2 + δΠ1 ) + αΠ2 2 − βΠ1 2 ϕ = 0,
∂x∂y ∂y ∂x
in which
92 Chapter 6 Elasticity of two-dimensional quasicrystals and simplification
⎧
⎪ ∂2 ∂2 ∂2 ∂2
⎪
⎪ Π 1 = 3 − , Π 2 = 3 − ,
⎪
⎪ ∂x2 ∂y 2 ∂y2 ∂x2
⎪
⎨
α = R(L + 2M ) − ω(K1 + K2 + K3 ), (6.6-8)
⎪
⎪
⎪
⎪ β = RM − ωK1 , δ = RM − ω(K1 + K2 + K3 ),
⎪
⎪
⎩
γ = R(L + 2M ) − ωK1 .
and ω is given above. At last the displacement potential F (x, y) is introduced
⎧
⎪
⎪ ∂2 ∂2
⎪
⎨ ϕ = βΠ 2 − αΠ1 F,
∂x2 ∂y2
(6.6-9)
⎪
⎪ ∂2F
⎪
⎩ ψ = (γΠ1 + δΠ2 )
∂x∂y
where
⎧
⎪
⎪ 2 ∂2 ∂2 2 ∂2 ∂2
⎪
⎨ ∇ = + , Λ = − ,
∂x2 ∂y 2 ∂x2 ∂y 2
(6.6-11)
⎪
⎪ R2 (L + M )(K2 + K3 )
⎪
⎩ ε= .
[M (K1 + K2 + K3 ) − R2 ][(L + 2M )K1 − R2 ]
A part of above results was reported by Refs. [12], [15] and [17].
If introducing stress potential functions φ(x, y), ψ1 (x, y) and ψ2 (x, y), such as
∂2φ ∂2φ ∂2φ
σxx = 2
, σyy = 2
, σxy = σyx = − ,
∂y ∂x ∂x∂y
∂ψ1 ∂ψ1 ∂ψ2 ∂ψ2
Hxx = , Hxy = − , Hyx = − , Hyy = , (6.7-2)
∂y ∂x ∂y ∂x
then equilibrium equations ∂σij /∂xj = 0 and ∂Hij /∂xj = 0 will be automatically
satisfied. Based on the generalized Hooke’s law (6.4-3), all strain components can
be expressed by relevant stress components:
⎧
⎪ 1 1
⎪
⎪ ε = (σ + σyy ) + [(K1 + K2 )(σxx − σyy )
⎪ xx 4(L + M ) xx
⎪ 4c
⎪
⎪
⎪
⎪ − 2R1 (Hxx + Hyy ) − 2R2 (Hxy − Hyx )],
⎪
⎪
⎪
⎪
⎪
⎪
1 1
(σxx + σyy ) − [(K1 + K2 )(σxx − σyy )
⎪
⎪ εyy =
⎪
⎪ 4(L + M ) 4c
⎪
⎪
⎪
⎪ − 2R 1 (H + H ) − 2R 2 (Hxy − H)],
⎪
⎪
xx yy
⎪
⎪ 1
⎪
⎪ ε = εyx = [(K1 + K2 )σxy − R2 (Hxx + Hyy ) + R1 (Hxy − Hyx )],
⎪
⎨ xy 2c
1 1
⎪
⎪ wxx = (Hxx − Hyy ) + [M (Hxx + Hyy ) − R1 (σxx − σyy ) − 2R2 σxy ],
⎪
⎪ 2(K 1 − K 2 ) 2c
⎪
⎪
⎪
⎪ 1 1
⎪
⎪ wyy = − (Hxx − Hyy ) + [M (Hxx + Hyy )
⎪
⎪ 2(K1 − K2 ) 2c
⎪
⎪
⎪
⎪ − R (σ − σ ) − 2R σ ],
⎪
⎪ 1 xx yy 2 xy
⎪
⎪
⎪
⎪ 1
wxy = [−R2 (σxx −σyy )+2R1 σxy ]+
1 M
(Hxy +Hyx )+ (Hxy −Hyx ),
⎪
⎪
⎪
⎪ 2c 2(K1 −K2 ) 2c
⎪
⎪
⎪
⎪ 1 1 M
⎩wyx = [R2 (σxx −σyy)−2R1 σxy]+ (Hxy +Hyx )− (Hxy −Hyx ),
2c 2(K1 −K2 ) 2c
(6.7-3)
6.8 Stress potential of point group 8mm octagonal quasicrystals 95
in which
c = M (K1 + K2 ) − 2(R12 + R22 ). (6.7-4)
where
∂2 ∂2 ∂2 ∂2 ∂2 ∂2
∇2 = 2
+ 2, Π1 = 3 2
− 2, Π2 = 3 2
− 2.
∂x ∂y ∂x ∂y ∂y ∂x
The equations (6.7-5) will be satisfied when we choose a new function G, which
is called the stress function, such that
φ = c1 ∇2 ∇2 G,
∂ ∂
ψ1 = − R1 Π1 + R2 Π2 ∇2 G,
∂y ∂x
∂ ∂
ψ2 = R1 Π2 − R2 Π1 ∇2 G, (6.7-6)
∂x ∂y
in which
c
c1 = +M (6.7-7)
K1 − K 2
and
∇2 ∇2 ∇2 ∇2 G = 0. (6.7-8)
So the final governing equation based on the stress potential is the same as that
based on the displacement potential.
The strain compatibility equations and the definition on stress potential are the
same as those given by (6.7-1) and (6.7-2), and the strain-stress relations are as
follows:
1 1
εxx = (K1 + K2 )(L + 2M ) − 2R2 σxx
(L + M )c 4
1 1
− (K1 + K2 )L + 2R2 σyy − R(L + M )(Hxx + Hyy ) , (6.8-1a)
4 2
1 1 1
εyy = − (K1 + K2 )L + 2R2 σxx + [(K1 + K2 )(L + 2M )
(L + M )c 4 4
1
−2R2 ]σyy + R(L + M )(Hxx + Hyy ) , (6.8-1b)
2
1
εxy = εyx = [(K1 + K2 )σxy + R(Hxy − Hyx )], (6.8-1c)
2c
1 1
wxx = R(K1 − K2 )(σyy − σxx )
(K1 − K2 )c 2
+ (K1 M − R2 )Hxx − (K2 M − R2 )Hyy , (6.8-1d)
1 1
wyy = R(K1 − K2 )(σyy − σxx )
(K1 − K2 )c 2
− (K2 M − R2 )Hxx + (K1 M − R2 )Hyy , (6.8-1e)
1
wxy = {R(K1 + K2 + 2K3 )σxy
(K1 + K2 + 2K3 )c
+ (K1 + K2 + K3 )M − R2 Hxy − (K3 M + R2 )Hyx , (6.8-1f)
1
wyx = {−R(K1 + K2 + 2K3 )σxy
(K1 + K2 + 2K3 )c
−(K3 M + R2 )Hxy + (K1 + K2 + K3 )M − R2 Hyx , (6.8-1g)
in which
c = M (K1 + K2 ) − 2R2 .
If introducing stress potential functions φ(x, y), ψ1 (x, y) and ψ2 (x, y) such as
∂2φ ∂ 2φ ∂2φ
σxx = , σyy = , σ xy = σ yx = − ,
∂y2 ∂x2 ∂x∂y
∂ψ1 ∂ψ1 ∂ψ2 ∂ψ2
Hxx = , Hxy =− , Hyx = − , Hyy = . (6.8-2)
∂y ∂x ∂y ∂x
then equilibrium equations (6.5-9) (or ∂σij /∂xj = 0, ∂Hij /∂xj = 0) will be auto-
matically satisfied.
Substituting the stress formulas (6.8-2) into the strain-stress relation (6.8-1) then
6.8 Stress potential of point group 8mm octagonal quasicrystals 97
φ = −c3 c4 ∇2 ∇2 G,
1
A= RG. (6.8-7)
2
In the derivation the relation c2 = c4 − c3 has been used. The stress potentials
φ, ψ1 , ψ2 can be expressed by the new function G(x, y), i.e.,
98 Chapter 6 Elasticity of two-dimensional quasicrystals and simplification
⎧
⎪
⎪ φ = −c3 c4 ∇2 ∇2 G,
⎪
⎪
⎪
⎪ ∂3 ∂3 ∂3
⎨ 1
ψ1 = R c4 2 Π1 + c3 3 Π1 − c2 2 Π2 G,
2 ∂x ∂y ∂y ∂x ∂y (6.8-8)
⎪
⎪
⎪
⎪ 3 3
∂3
⎪
⎪ 1 ∂ ∂
⎩ ψ2 = R c2 Π1 − c3 3 Π2 − c4 Π2 G.
2 ∂x∂y 2 ∂x ∂x∂y 2
∂4 1 !
= ∇2 ∇2 − Λ2 Λ2 , Π1 Π2 = ∇2 ∇2 − 4Λ2 Λ2 , c2 = c4 − c3 ,
∂x2 ∂y 2 4
∂2 ∂2 ∂2 ∂2
∇2 = 2
+ 2 , Λ2 = 2
− 2,
∂x ∂y ∂x ∂y
with
R2 (L + M )(c3 − c4 ) R2 (L + M )(K2 + K3 )
ε= = .
−c1 c3 + R2 (L + M )c3 [(K1 + K2 + K3 )M − R2 ][K1 (L + 2M ) − R2 ]
(6.8-11)
The final governing equation in this case is exactly in agreement to that given by
the displacement potential formulation, discussed in Section 6.
solutions may be constructed with relative ease. For this purpose we consider an
example.
From Section 6.3, we have deformation compatibility equations
⎧ 2
⎪ ∂ εxx ∂ 2 εyy ∂ 2 εxy
⎪
⎨ ∂y 2 + ∂x2 = 2 ∂x∂y ,
(6.9-1)
⎪
⎪
⎩ ∂wxy = ∂wxx , ∂wyx = ∂wyy .
∂x ∂y ∂y ∂x
Then the substitution of (6.3-2) into (6.9-1) yields the compatibility equations ex-
pressed by σij and Hij
2 L+M ∂2 ∂2
∇ (σxx + σyy ) − −
C ∂x2 ∂y2
× [(K1 + K2 )(σxx − σyy ) − 2R(Hxx + Hyy )]
L + M ∂2
=8 [(K1 + K2 )σxy − R(Hxy + Hyx )] (6.9-2a)
C ∂x∂y
1 ∂ (R2 − M K1 )Hxy + (R2 − M K2 )Hyx
Rσxy −
C ∂x K1 − K 2
1 ∂ 1 ∂
= (Hxx −Hyy )+ [M (Hxx +Hyy )−R(σxx −σyy )] (6.9-2b)
2(K1 −K2 ) ∂y 2C ∂y
1 ∂ (R2 − M K1 )Hxy + (R2 − M K2 )Hyx
−Rσxy −
C ∂y K1 − K 2
−1 ∂ 1 ∂
= (Hxx −Hyy )+ [M (Hxx +Hyy )−R(σxx −σyy )], (6.9-2c)
2(K1 −K2 ) ∂x 2C ∂x
in addition there are so-called St. Venant boundary conditions at the end sections
⎧ h/2 h/2
⎪
⎪
⎪
⎪ σxx dy = 0, yσxx dy = Mz ,
⎪
⎪
⎪ −h/2
⎪
−h/2
⎪
⎨ h/2 h/2
Hxx dy = 0, yHxx dy = Lz , (6.9-5)
⎪
⎪ −h/2 −h/2
⎪
⎪
⎪
⎪ h/2 h/2
⎪
⎪
⎪
⎩ σxy dy = 0, Hyx dy = 0.
−h/2 −h/2
where Mz and Lz represent the resultant moments of σxx and Hxx . The direction
of vector of the moments is z. Boundary conditions (6.9-5) are relaxation boundary
conditions, this gives some flexibility for solution.
At first we assume that the value of Lz momentarily be undetermined, and
assume further
and A1 and f (y) are to be determined. Substituting σxx = A1 y into (6.9-5), we find
that
2Mz
A1 = 3 (6.9-6)
h
so that
Mz
σxx = y, (6.9-7)
I
where I = 1 · h3 /12 represents the inertia moment of the transverse section with
height h and width 1.
It can be verified that the above results have satisfied equations (6.7-3) and
(6.7-2).
Substituting (6.9-7) into (6.9-2b) yields
Rσxx 2RMz y
Hxx = = , (6.9-8)
1 M 1 M
+ I +
2(K1 − K2 ) 2C K1 − K2 C
References 101
References
[1] Chernikov M A, Ott H R, Bianchi A et al. Elastic moduli of a single quasicrystal of
decagonal Al-Ni-Co: evidence for transverse elastic isotropy. Phys Rev Lett, 1998,
80(2): 321–324
[2] Abe H, Taruma N, Le Bolloc’h D et al. Anomalous-X-ray scattering associated with
short-range order in an Al70 Ni15 Co15 decagonal quasicrystal. Mater Sci and Eng A,
2000, 294∼296(12): 299–302
[3] Jeong H C, Steinhardt P J. Finite-temperature elasticity phase transition in decago-
nal quasicrystals. Phys Rev B, 1993, 48(13): 9394–9403
[4] Edagawa K. Phonon-phason coupling in decagonal quasicrystals. Phil Mag, 2007,
87(18∼21): 2789–2798
[5] Fan T Y, Mai Y W. Elasticity theory, fracture mechanics and relevant thermal prop-
erties of quasicrystalline materials. Appl Mech Rev, 2004, 57(5), 325–344
[6] Hu C Z, Yang W G, Wang R H et al. Symmetry and physical properties of qua-
sicrystals. Prog Phys, 1997, 17(4): 345–374
102 Chapter 6 Elasticity of two-dimensional quasicrystals and simplification
In Chapters 5 and 6, with the physical basis of quasicrystal elasticity based on the
density wave model, we have performed some mathematical operations, by proper
simplification, to reduce the original problems to the boundary value problems of
high-order partial differential equations, and to establish the standard solving proce-
dure and the fundamental solutions. This work is the development of the boundary
value problems in classical elasticity. Here, we need to pose a question: Do these
mathematical operations contribute to solving the quasicrystal elasticity problems?
This is answered only by practice. The following two chapters will provide the ap-
plications of these theories, including the solutions to some dislocation, crack and
interface problems in one- and two-dimensional quasicrystals. The calculation re-
sults indicate that the mathematical operations discussed in Chapters 5 and 6 are
powerful in solving these problems.
As we know, almost all monographs of classical elasticity do not place their main
focus upon dislocation and crack problems. These problems have been investigated
in monographs relating to dislocation theory and fracture mechanics, respectively.
As a monograph on theories of elasticity of quasicrystals, the current work is not
going to focus entirely on dislocations and cracks in quasicrystals. Therefore, as
an attempt to examine the theories developed in Chapters 5 and 6 and to show
their applications in elasticity of quasicrystals, we present some calculation exam-
ples of certain realistic dislocation and crack problems in elasticity of quasicrystals.
The methods developed in this work can be used for studying other problems in
quasicrystals.
Historically, soon after the discovery of quasicrystals, scientists proposed the
possibility of existing dislocations in quasicrystals. De and Pelcovits[1,2] first studied
elastic field around dislocations and disclinations in quasicrystals. Furthermore,
Ding et al[3,4] conducted systematic investigation of this topic by using the Green
function method.
Now people have recognized that quasicrystal is a kind of ordered phase of
quasiperiodic long-range. Similar to crystals with ordered phase of long-range, the
breaking of long-range regularity takes place usually through topological defects,
i.e., dislocations lead to the breaking of the long-range symmetry. As mentioned
in Chapter 3, quasicrystals also possess the orientational symmetries incompatible
with those permitted in crystal theory. In quasicrystals, another type of defect,
disclinations, also exists simultaneously, which lead to breaking of the orientational
symmetry in quasicrystals. In some cases, the crystalline phases often coexist with
the quasicrystalline phases. So there is another kind of defects, the interface be-
tween quasicrystal and crystal. The existence of these defects including the stacking
fault dramatically affects the mechanical properties of quasicrystals. Therefore, it is
important to study the elastic properties of quasicrystals with dislocations, discli-
nations, interfaces and stacking faults.
As aforementioned, during the study of dislocations in quasicrystals, physicists
have developed some mathematical methods, such as the Green function method.
These important methods are available in relevant literature. Hereafter, by using the
elementary methods developed in Chapters 5 and 6, complex variable function and
Fourier transform, we study the displacement and stress fields around dislocations
in one- and two-dimensional quasicrystals quantitatively. The three-dimensional
dislocation and dynamic dislocation problems will be discussed in Chapters 9 and
10 respectively.
in the physical space E|| , while b⊥ in the supplementary space or the vertical space
|| || ||
E⊥ . For one-dimensional quasicrystals, b|| =(b1 , b2 , b3 ) and b⊥ =(0,0, b⊥ 3 ) since
⊥ ⊥ || ⊥ || || || ⊥
b1 = b2 =0. Therefore, b ⊕ b =(b1 , b2 , b3 , b3 ) here, which can be dealt with as
|| || || || ||
the superposition of (b1 , b2 ,0,0) and (0,0, b3 , b⊥ 3 ). The (b1 ,b2 ,0,0) corresponds to the
blade dislocation in regular hexagonal crystals, whose elastic solution is available in
common metal physics or dislocation monographs, (e.g. Refs.[7],[8]). The one corre-
||
spondence to the component (0,0,b3 , b⊥ 3 ) is the screw dislocation in one-dimensional
quasicrystal. We solve the elastic field induced by this dislocation in the following.
In Section 5.2, we have obtained the governing equations for anti-plane strain
7.1 Dislocations in one-dimensional hexagonal quasicrystals 105
∇2 uz = 0, ∇2 wz = 0. (7.1-1)
The boundary conditions corresponding to the screw dislocation with the Burgers
||
vector (0, 0, b3 , b⊥
3 ) are
⎧ " "
⎨ uz ""y=0+ − uz ""y=0− = b ,
3
" " (7.1-2a)
⎩
wz "y=0+ − wz "y=0− = b⊥ 3
or
duz = b3 , dwz = b⊥
3, (7.1-2b)
Γ Γ
where Γ indicates an arbitrary contour surrounding the dislocation core. The solu-
tion to the boundary value problem (7.1-1) and (7.1-2) is
b3 y b⊥
3 y
uz = arctan , wz = arctan . (7.1-3)
2π x 2π x
The stress components can be extracted according to the stress-strain relation (5.2-8)
in Chapter 5 as
⎧
⎪
⎪ C44 b3 y R3 b⊥
3 y
⎪
⎪ σ = σ = − − ,
⎪
⎪
xz zx
2π x2 + y 2 2π x2 + y 2
⎪
⎪
⎪
⎪
⎪
⎪ C44 b3 x R3 b⊥ x
⎪
⎪ σ = σ = − + 3
,
⎨ yz zy
2π x2 + y 2 2π x2 + y 2
(7.1-4)
⎪
⎪
⎪ K 2 b⊥
y R 3 b y
⎪
⎪ Hzx = − 3
− 3
,
⎪
⎪ 2π x2 + y 2 2π x2 + y 2
⎪
⎪
⎪
⎪
⎪
⎪ K b⊥ R3 b3
⎪
⎩ Hzy = − 2 3
x
+
x
.
2π x2 + y 2 2π x2 + y 2
||
The displacement and stress fields corresponding to the Burgers vector b|| ⊕ b⊥ =(b1 ,
|| ||
b2 , b3 , b⊥
3 ) can be obtained by superposing the above elastic field on that of the
regular hexagonal crystals.
The elastic strain energy induced by the screw dislocation is
1 ∂uz ∂wz
W = σzj + Hzj dx1 dx2
2 Ω ∂xj ∂xj
1 R0 2π ∂uz ∂wz
= r σzj + Hzj drdθ
2 r0 0 ∂xj ∂xj
2 1 R0
= (C44 b3 + K2 b⊥2 ⊥
3 + 2R3 b3 b3 ) ln , (7.1-5)
4π r0
106 Chapter 7 Application I—Some dislocation and interface problems and . . .
where r0 is the size of the dislocation core, R0 is the size of dislocation network or
inclusion, which are available in theory of dislocations in regular crystals.
where ξ is the Fourier transform parameter, the above equation can be reduced into
4
d2
− ξ2 F̂ = 0. (7.2-3)
dy 2
This is an ordinary differential equation with constant coefficients, and its general
solution is
where the symbol T indicates the matrix transpose. Therefore, (7.2-5) can be ex-
pressed as
F̂ (ξ, y) = XY e−|ξ|y . (7.2-5 )
ûx (ξ, y) = iξ −1 X[2nξ 2 Y + (m − 5n) |ξ| Y − (2m − 5n)Y ]e−|ξ|y , (7.2-8a)
ûy (ξ, y) = |ξ|−1 X[2nξ 2 Y − (m + 5n) |ξ| Y + (2m + 5n)Y ]e−|ξ|y , (7.2-8b)
ŵx (ξ, y) = −iω(α − β)ξ −1 X[2n |ξ|3 Y − 12ξ 2 Y
+ 15 |ξ| Y − 10Y ]e−|ξ|y , (7.2-9a)
ŵy (ξ, y) = −ω(α − β) |ξ|−1 X[4 |ξ|3 Y − 12ξ 2 Y
+ 15 |ξ| Y − (10 + e0 − e1 )Y ]e−|ξ|y , (7.2-9b)
σ̂xx (ξ, y) = 2M α(L + M )X(−2ξ 2 Y + 8 |ξ| Y − 13Y )e−|ξ|y , (7.2-10a)
σ̂yy (ξ, y) = 2M α(L + M )X(2ξ 2 Y − 4 |ξ| Y + 3Y )e−|ξ|y , (7.2-10b)
−1 2
σ̂xy (ξ, y) = σ̂yx (ξ, y) = i2M α(L + M )ξ |ξ| X(2ξ Y − 6 |ξ| Y
+ δY )e−|ξ|y , (7.2-10c)
−1 3
Ĥxx (ξ, y) = −ω(α − β)(K1 − K2 ) |ξ| X[4 |ξ| Y
− 16ξ 2 Y + 27 |ξ| Y − (25 + e2 )Y ]e−|ξ|y (7.2-11a)
3
Ĥyy (ξ, y) = −ω(α − β)(K1 − K2 )X[−4 |ξ| Y + 12ξ 2 Y − 15 |ξ| Y
+ (10 − e1 )Y ]e−|ξ|y , (7.2-11b)
3
Ĥxy (ξ, y) = iω(α − β)(K1 − K2 )ξ −1 |ξ| X[−4 |ξ| Y + 12ξ 2 Y
− 15 |ξ| Y + (10 + e2 )Y ]e−|ξ|y , (7.2-11c)
3
Ĥyx (ξ, y) = −iω(α − β)(K1 − K2 )ξ −1 |ξ| X[−4 |ξ| Y + 16ξ 2 Y
− 27 |ξ| Y + (25 − e1 )Y ]e−|ξ|y , (7.2-11d)
where
⎧
⎨ m = M α + (L + 2M )β, n = M α − (L + 2M )β,
2αβ 2αβ α+β (7.2-12)
⎩ e1 = , e2 = +
ω(α − β)(K1 − K2 ) ω(α − β)(K1 − K2 ) α − β
and α, β, and ω are given in (6.2-5).
The inverse of Fourier transform reads
∞
1
F (x, y) = F̂ (ξ, y)e−iξx dξ, (7.2-13)
2π −∞
7.2 Dislocations in quasicrystals with point groups 5m and 10mm symmetries 109
||
by dislocation with Burgers vector (b1 ,0,0,0) in quasicrystals of point group 10mm
ten-fold symmetry as below
b1 y (L + M )K1 xy
ux = arctan + , (7.2-24a)
2π x (L + M )K1 + (M K1 − R2 ) r2
r
b1 (M K1 − R2 )
uy = − ln
2π (L + M )K1 + (M K1 − R2 ) a
2
(L + M )K1 y
+ , (7.2-24b)
(L + M )K1 + (M K1 − R2 ) r2
$ % 3
b1 (L + M )K1 2x y
wx = , (7.2-25a)
2π (L + M )K1 + (M K1 − R2 ) r4
$ % 2 2
b1 (L + M )K1 2x y
wy = , (7.2-25b)
2π (L + M )K1 + (M K1 − R2 ) r4
y(3x2 + y 2 )
σxx = −A , (7.2-26a)
r4
2 2
y(x − y )
σyy =A , (7.2-26b)
r4
x(x2 − y 2 )
σxy = σyx = A , (7.2-26c)
r4
R(K1 − K2 ) x2 y(3x2 − y 2 )
Hxx = −A , (7.2-27a)
M K1 − R 2 r6
2
R(K1 − K2 ) x y(3y2 − x2 )
Hyy = −A , (7.2-27b)
M K1 − R 2 r6
2 2
R(K1 − K2 ) xy (3x − y 2 )
Hxy =A , (7.2-27c)
M K1 − R 2 r6
3 2
R(K1 − K2 ) x (3y − x2 )
Hyx = −A , (7.2-27d)
M K1 − R 2 r6
#
where r = x2 + y 2 , a is the size of dislocation core and
$ %
b1 (L + M )(M K1 − R2 )
A= . (7.2-28)
π (L + M )K1 + (M K1 − R2 )
k 0 b⊥
1x 2(x2 − y 2 ) (x2 − y 2 )(3x2 − y 2 )(3y 2 − x2 )
Hyx = − + , (7.2-33d)
2πe1 r4 r8
where
2c1 c2 c1 c2 c1 c
e1 = , e2 = + 2 ,
c0 k0 c0 k0 c1 c2
c1 = (L + 2M )K2 − R2 , c2 = M K2 − R2
and c0 , c1 ,c2 and k0 are given in Chapter 6, i.e.,
c0 = (L + 2M )R, c1 = (L + 2M )K1 − R2 , c2 = M K1 − R2 ,
k0 = R(K1 − K2 ).
||
Superposition of the above two solutions yields the solution of dislocation (b1 , 0,
||
b1 , 0). The solution of (0, b2 , 0, b⊥
⊥
2 ) can be determined similarly. Their superposition
|| ||
yields the solution of (b1 , b2 , b⊥ ⊥
1 b2 ). Part of this work can be found in paper given
,
by Li and Fan[5] . Readers may examine that the present solutions are identical to
those given by Ding et al[4] using the method of Green functions (Note that L = C12
and M = (C11 − C12 )/2=C66 ).
∇2 ∇2 ∇2 ∇2 F = 0, (7.3-1)
where X and Y have the same meanings as those in Section 7.2, ξ is the parameter
of Fourier transform and
R2 = R12 + R22 . (7.3-3)
After some algebraic manipulation with (6.2-8)∼(6.2-11), the displacement com-
ponents in the Fourier transform domain are
ûx (ξ, y) = iξ −1 R̄0 X[2nξ 2 Y + (m − 5n) |ξ| Y − (2m − 5n)Y ]e−|ξ|y , (7.3-4a)
−1 2 −|ξ|y
ûy (ξ, y) = |ξ| R̄0 X[2nξ Y − (m + 5n) |ξ| Y + (2m + 5n)Y ]e , (7.3-4b)
3
ŵx (ξ, y) = ic0 ξ R̄02 X[4 |ξ| Y − 12ξ 2 Y + 15 |ξ| Y − 10Y ]e−|ξ|y ,
−1
(7.3-4c)
−1 2 3 2 2 −|ξ|y
ŵy (ξ, y) = c0 |ξ| R̄0 X[4 |ξ| Y − 12ξ Y + 15 |ξ| Y − (10 + e0 R0 )Y ]e ,
(7.3-4d)
where
⎧
⎪
⎪ −[C66 c1 + C11 c2 ]
⎨ m = c2 + c1 , n = c2 − c1 , e0 = ,
Rc0
(7.3-5)
⎪
⎪ R1 − iR2 sgn ξ
⎩ R0 = R1 + iR2 sgn ξ , R̄0 = .
R R
Similarly, the stress components in the Fourier transform domain are
σ̂xx (ξ, y) = 2c0 c2 R−1 R̄0 X(−2ξ 2 Y + 8 |ξ| Y − 13Y )e−|ξ|y , (7.3-6a)
−1 2 −|ξ|y
σ̂yy (ξ, y) = 2c0 c2 R R̄0 X(2ξ Y − 4 |ξ| Y + 3Y )e , (7.3-6b)
−1
σ̂xy (ξ, y) = σ̂yx (ξ, y) = i2c0 c2 R R̄0 (sgnξ)
· X(2ξ Y − 6 |ξ| Y + 7Y )e−|ξ|y ,
2
(7.3-6c)
3
Ĥxx (ξ, y) = c0 k0 R̄02 X[4 |ξ| Y − 16ξ 2 Y + 27 |ξ| Y
− (25 + e2 R02 )Y ]e−|ξ|y , (7.3-6d)
3
Ĥyy (ξ, y) = c0 k0 R̄02 X[−4 |ξ| Y + 12ξ 2 Y − 15 |ξ| Y
+ (10 − e1 R02 )Y ]e−|ξ|y , (7.3-6e)
Ĥxy (ξ, y) = ic0 k0 R̄02 (sgnξ)X[−4 |ξ|3 Y + 12ξ 2 Y − 15 |ξ| Y
+ (10 + e2 R02 )Y ]e−|ξ|y , (7.3-6f)
3
Ĥyx (ξ, y) = ic0 k0 R̄02 (sgnξ)X[−4 |ξ| Y
+ 16ξ 2 Y − 27 |ξ| Y + (25 − e1 R02 )Y ]e−|ξ|y , (7.3-6g)
⎧
⎪
⎨ e1 = 2c1 c2 , e2 = c1 c2 c1 c2
+ ,
c0 k0 R c0 k0 R c1 c2 (7.3-7)
⎪
⎩
c1 = (L + 2M )K2 − R2 , c2 = M K2 − R2 ,
where c0 and K0 are given in (6.4-13).
114 Chapter 7 Application I—Some dislocation and interface problems and . . .
In the Fourier transform domain, solutions (7.3-4), (7.3-6) and the first three
expressions in boundary conditions (7.3-8) yield
9C(ξ)
A(ξ) = , B(ξ) = 2C(ξ), D(ξ) = 0, (7.3-9)
4
and the dislocation condition leads to
⎧
⎪
⎪ dux = −4c1 R−1 (R1 ReC + R2 sgnξImC),
⎨
Γ
(7.3-10)
⎪
⎪
⎩ duy = 4c1 R−1 (R2 ReC − R1 sgnξImC),
Γ
where the indicator Re and Im stand for the real and imaginary parts of a complex,
respectively. Finally, expressions (7.3-10) and (7.3-8) define
(R1 + iR2 sgnξ)b1 R0 b1
C(ξ) = ReC + iImC = − =− , (7.3-11)
4πRc1 4πc1
where R0 is given by (7.3-5), and c1 is given by (6.4-9), i.e.,
Therefore,
b1 y c − c xy
1 2
ux = arctan + , (7.3-13a)
2π x c1 r2
b1 r c1 − c2 r y2
uy = − ln + ln + 2 , (7.3-13b)
2π a c1 a r
7.3 Dislocations in quasicrystals with point groups 5, 5̄ five-fold and 10, 10 . . . 115
c 0 b1 R1 2x3 yy 2 (3x2 + y 2 )
R2
wx = + , (7.3.13c)
2πc1 R r4 r4R
3
c 0 b1 R1 y 2 (3x2 + y 2 ) R2 2x y
wy = + , (7.3-13d)
2πc1 R r4 R r4
where c0 is given by (6.4-13), and a indicates the radius of dislocation core.
The corresponding stress components can be determined by (7.3-13) and gener-
alized Hooke’s law (6.4-3):
$ %
c0 c2 b1 y(3x2 + y 2 )
σxx = − , (7.3-14a)
πc1 R r4
$ %
c0 c2 b1 y(x2 − y 2 )
σyy = , (7.3-14b)
πc1 R r4
$ %
c0 c2 b1 x(x2 − y 2 )
σxy = σyx = , (7.3-14c)
πc1 R r4
3 2
c0 k0 b1 R1 x2 y(3x2 − y 2 ) R2 x (3y − x2 )
Hxx = − + , (7.3-14d)
πc1 R r6 R r6
2 2
c0 k0 b1 R1 x2 y(3x2 − y 2 ) R2 xy (3x − y 2 )
Hyy = − − , (7.3-14e)
πc1 R r6 R r6
2
c0 k0 b1 R1 xy 2 (3x2 − y 2 ) R2 x y(3y 2 − x2 )
Hxy = − + , (7.3-14f)
πc1 R r6 R r6
2
c0 k0 b1 R1 x3 (3y 2 − x2 ) R2 x y(3x2 − y 2 )
Hyx = − − + . (7.3-14g)
πc1 R r6 R r6
||
Now let us consider the case of b1 = 0 and b⊥ 0. In this case, the corresponding
1 =
boundary conditions are
⎧ #
⎪
⎪ σij (x, y) → 0, Hij (x, y) → 0, x2 + y 2 → ∞,
⎪
⎪
⎪
⎪ H (x, 0) = 0,
⎨ yy
ux (x, 0) = uy (x, 0) = 0, (7.3-15)
⎪
⎪
⎪
⎪
⎪
⎪
⎩ dwx = b⊥1, dwy = 0.
Γ Γ
By using the similar analysis, the corresponding displacement and stress fields can
be determined as the following formulas (7.3-16) and (7.3-17),
c1 b⊥
1 R1 xy c1 − c2 2xy 3
ux = −
πc0 e1 R r2 c1 r4
R2 y 2 c1 − c2 y 2 (x2 − y 2 )
+ + , (7.3-16a)
R r2 c1 r4
116 Chapter 7 Application I—Some dislocation and interface problems and . . .
c1 b⊥
1 R1 y 2 c1 − c2 y 2 (x2 − y 2 )
uy = − −
πc0 e1 R r2 c1 r4
R2 xy c1 − c2 2xy 3
+ + , (7.3-16b)
R r2 c1 r4
y R2 − R2 xy(3x2 − y 2 )(3y 2 − x2 )
b⊥
1 1 2
wx = arctan +
2π x e1 R2 3r 6
2R1 R2 y 2 (3x2 − y 2 )2
+ , (7.3-16c)
e1 R2 3r 6
2
b⊥
1 r R1 − R22 y 2 (3x2 − y 2 )2
wy = e2 ln +
2πe1 a R2 3r 6
2R1 R2 xy(3x2 − y 2 )(3y 2 − x2 )
− , (7.3-16d)
e1 R2 3r 6
2c2 b⊥1 R1 x2 y(3x2 − y 2 ) R2 x3 (3y 2 − x2 )
σxx =− + , (7.3-17a)
πe1 R R r6 R r6
2c2 b⊥1 R1 y 3 (3x2 − y 2 ) R2 xy 2 (3y 2 − x2 )
σyy =− + , (7.3-17b)
πe1 R R r6 R r6
2c2 b⊥
1 R1 xy 2 (3x2 − y 2 ) R2 x2 (3y 2 − x2 )
σxy = σyx = − + , (7.3-17c)
πe1 R R r6 R r6
2
K0 b ⊥
1 y R1 −R22 2R1 R2
Hxx = −(e1 +e2 ) 2 +x h21 (x, y)− h22 (x, y) , (7.3-17d)
2πe1 r R2 R2
2
K0 b ⊥1 y R1 − R2
2
2R1 R2
Hyy =− h22 (x, y) + h 21 (x, y) , (7.3-17e)
2πe1 R2 R2
2
K0 b ⊥
1 x R1 −R22 2R1 R2
Hxy = (e1 +e2 ) 2 +y h21 (x, y)− h22 (x, y) , (7.3-17f)
2πe1 r R2 R2
2
K0 b ⊥1 x R 1 − R2
2
2R1 R2
Hyx =− h 22 (x, y) + h 21 (x, y) , (7.3-17g)
2πe1 R2 R2
in which
2xy(3x2 − y 2 )(3y 2 − x2 )
h21 (x, y) = ,
r8
(7.3-18)
2(x2 − y 2 ) (x2 − y 2 )(3x2 − y 2 )(3y 2 − x2 )
h22 (x, y) = + .
r4 r8
||
The solution to dislocation problem (b1 ,0,b⊥
1 ,0) can be determined by superpos-
7.4 Dislocations in quasicrystals with eight-fold symmetry 117
ing (7.3-13) and (7.3-14) onto (7.3-16) and (7.3-17). The solution to dislocation
||
problem (0,b2 ,0,b⊥
2 ) can be determined in the similar way. As a result, the solu-
|| ||
tion to dislocation problem with Burgers vector (b1 ,b2 ,b⊥ ⊥
1 ,b2 ) can be determined
completely.
The above work can be found in literature [6].
in which ⎧
⎪
⎪ 2 ∂2 ∂2 2 ∂2 ∂2
⎪
⎨ ∇ = ∂x2 + ∂y 2 , Λ = ∂x2 − ∂y 2 ,
(7.4-2)
⎪
⎪ R2 (L + M )(K2 + K3 )
⎪
⎩ ε=
[M (K1 + K2 + K3 ) − R2 ][(L + 2M )K1 − R2 ]
(see (6.6-10),(6.6-11) for detail). Equation (7.4-1) is more complicated than those
of (6.2-7) and (7.3-1), so the solution of which is also more complex than those
discussed in the previous sections. Due to the space limitation we cannot list whole
procedure of the solution, only give some main results of them in the following, in
which the Fourier transform and complex variable function methods are respectively
used.
The eigen roots of equation (7.4-4) depend on the value of parameter ε, Zhou[9]
gave a detailed discussion for the solutions corresponding to case (1): 0 < ε < 1 and
118 Chapter 7 Application I—Some dislocation and interface problems and . . .
case (2): ε < 0, but calculation is tremendous complex and lengthy which cannot
be included here. For the case (1) the solution is
⎧ 2 2
⎪
⎪ 1 b1 λ1 +λ22 y λ2 λ1 +λ22 y λ2
⎪
⎪ x
u (x, y) = arctan + +arctan −
⎪
⎪ 2π 2 λ x λ1 λ1 x λ1
⎪
⎪ 1
⎪
⎪
⎪
⎪ 2λ 3 xy 2λ1 xy
⎪
⎪ +(F 3 C + F4 D) arctan − arctan
⎪
⎪ x2 − (λ23 + λ24 )y 2 x2 − (λ21 + λ22 )y 2
⎪
⎪
⎪
⎪ x2 +2λ2 xy+(λ21 +λ22 )y 2 x2 +2λ4 xy+(λ23 +λ24 )y 2
⎪
⎪ 1
⎪
⎪ + F 5 ln 2 +F6 ln 2
⎪
⎪ 4π x −2λ2 xy+(λ21 +λ22 )y 2 x −2λ4 xy+(λ23 +λ24 )y 2
⎪
⎪
⎪
⎪ 2λ1 λ2 y 2
⎪
⎪ 1 λ2
⎪
⎪ uy = H1 arctan 2 − 2 arctan
⎪
⎪ 2π x + (λ21 − λ22 )y 2 λ1
⎪
⎪
⎪
⎪ 2
⎪
⎪ 2λ3 λ4 y λ4
⎪
⎪
⎪
+H2 arctan 2
x + (λ 2 − λ2 )y 2 − 2 arctan λ
⎪
⎪ 3 4 3
⎪
⎪
⎪
⎪ 1 4 2
x + 2(λ1 − λ2 )x y 2 2 2
⎪
⎪ + H3 ln 1 +
⎪
⎪ 4π (λ21 + λ22 )2 y 4
⎪
⎪
⎪
⎪
⎪
⎪ x4 + 2(λ23 − λ24 )x2 y 2
⎪
⎪ +H ln 1 + ,
⎪
⎪ 4
(λ23 + λ24 )2 y 4
⎪
⎪
⎪
⎪
2 2 2
⎨ 1 b⊥ 1 λ1 +λ2 y λ2 λ1 +λ22 y λ2
wx (x, y) = arctan + +arctan −
⎪
⎪
2π 2 λ1 x λ1 λ1 x λ1
⎪
⎪
⎪
⎪ 2λ3 xy
⎪
⎪ +(G C + G D) × arctan 2
⎪
⎪ 3 4
x − (λ23 + λ24 )y 2
⎪
⎪
⎪
⎪
⎪
⎪ 2λ1 xy
⎪
⎪ − arctan 2
⎪
⎪ x − (λ 2 + λ2 )y 2
⎪
⎪ 1 2
⎪
⎪
⎪
⎪ 1 x 2
+2λ xy+(λ 2 2 2
1 +λ2 )y x2 + 2λ4 xy+(λ23 +λ24 )y 2
⎪
⎪ + G5 ln 2
2
+G6 ln 2 ,
⎪
⎪ 4π x −2λ2 xy+(λ21 +λ22 )y 2 x −2λ4 xy+(λ23 + λ24 )y 2
⎪
⎪
⎪
⎪
⎪
⎪ 1 2λ1 λ2 y 2 λ2
⎪
⎪ w = I arctan − 2 arctan
⎪
⎪ y
2π
1
x2 + (λ21 − λ22 )y 2 λ1
⎪
⎪
⎪
⎪
⎪
⎪ 2λ3 λ4 y 2
λ4
⎪
⎪
⎪
⎪
+I2 arctan 2
x + (λ 2 − λ2 )y 2 − 2 arctan λ
⎪
⎪ 3 4 3
⎪
⎪
⎪
⎪ 1 4 2
x + 2(λ1 − λ2 )x y 2 2 2
⎪
⎪ + I3 ln 1 +
⎪
⎪ (λ21 + λ22 )2 y 4
⎪
⎪
4π
⎪
⎪
⎪
⎪ x4 + 2(λ23 − λ24 )x2 y 2
⎪
⎪ +I4 ln 1 + ,
⎪
⎩ (λ23 + λ24 )2 y 4
(7.4-5)
in which F1 , · · · , F6 , G1 , · · · , G6 , H1 , · · · , H4 and I1 , · · · , I4 are some functions of
λ1 , λ2 , λ3 and λ4 which are constants constituted from the original material constants
M, L, K1 , K2 , K3 and R, the expressions are very complicated and lengthy, we have
7.4 Dislocations in quasicrystals with eight-fold symmetry 119
to omit them.
By using the similar procedure, the solution for the case (2) can also be obtained.
But the solving procedure is very tedious due to the complexity of the final governing
equation (7.4-1). We omit them for simplicity.
The equation (7.4-1) can also be solved by complex variable function method. For
this purpose the equation can be rewritten as
8
∂ ∂8 ∂8
+ 4(1 − 4ε) + 2(3 + 16ε)
∂x8 ∂x6 ∂y 2 ∂x4 ∂y 4
∂8 ∂8
+4(1 − 4ε) 2 6 + 8 F = 0. (7.4-6)
∂x ∂y ∂y
The solution of equation (7.4-6) can be expressed in terms of four analytic functions
Fk (zk ) of complex variable zk (k = 1, 2, 3, 4), i.e.,
4
F (x, y) = 2Re Fk (zk ), zk = x + μk y (7.4-7)
k=1
where
∂ 6 Fk (zk )
fk (zk ) = . (7.4-10)
∂zk6
Some detail of this work can be found in Ref. [10].
120 Chapter 7 Application I—Some dislocation and interface problems and . . .
which is continuous transition. The transition process gives rise to interface be-
tween crystal and quasicrystal. Consequently, analysis of the interface problem of
quasicrystals is of significance. In this section we give a phenomenological study on
elastic behaviour of the interface for one-dimensional quasicrystal-isotropic crystal.
Refs.[15∼17] pointed out that the phase transition is induced by the phason strains.
Such a problem is formidable. Here we focus only on the determination of the
strains, and further studies will be given in Chapter 9 for icosahedral quasicrystal-
cubic crystal interface.
Consider an orthorhombic quasicrystal at the upper half-plane (i.e., y > 0), the
phonon-phason coupling problem is governed by (5.4-3), i.e.,
∂4 ∂4 ∂4
a1 4 + a3 2 2 + a5 4 F = 0, (7.6-1)
∂x ∂x ∂y ∂y
in which F (x, y) denotes the displacement potential, a1 , a3 , a5 the material constants
composed of Cij , Ki and Ri defined by (5.3-6). We assume that the crystal coexisting
with the quasicrystal is located at the lower half-plane with thickness h(i.e., −h <
y < 0). The plane y = 0 is the interface between the quasicrystal and crystal,
as shown in Fig.7.6-1. For simplicity, suppose the crystal be an isotropic material,
(c)
characterizing by elastic constants Cij (E (c) , μ(c) ). At the interface there are the
following boundary conditions:
μ(c)
k= (7.6-3)
h
and where μ(c) and h the shear modulus and the thickness of the crystal. We further
assume that the outer boundaries are stress free.
Taking the Fourier transform
∞
F̂ (ξ, y) = F (x, y)eiξx dx (7.6-4)
−∞
Substituting (7.6-8) into the second one of (7.6-10) then into the second one of (7.6-2)
yields
B = αA, (7.6-11)
where
−λ1 c2 − λ31 c4 + i(c1 + λ21 c3 )
α= ,
λ2 c2 + λ32 c4 + i(−c1 − λ22 c3 )
The Fourier transform of the phonon stress and displacement component at the
interface are
σ̂zy (ξ, 0) = A(ξ) |ξ| (λ1 e−λ1 |ξ|y + αλ2 e−λ2 |ξ|y ),
ûz (ξ, 0) = A(ξ)ξ 2 [(−R6 + λ21 R5 )e−λ1 |ξ|y + α(R6 − λ22 R5 )e−λ2 |ξ|y ]. (7.6-13)
τ fˆ(ξ)
A(ξ) = . (7.6-15)
|ξ| (λ1 + αλ2 ) − kξ 2 [−R6 + λ21 R5 + α(R6 − λ21 R5 )]
Thus, the stress and displacement components for phonon and phason fields can be
evaluated, e.g.
⎧ ∞ ∞
⎪
⎪ 1 −iξx 1
⎪ σzy =
⎪ σ̂zy e dξ = A(ξ) |ξ|(λ1 e−λ1 |ξ|y + αλ2 e−λ2 |ξ|y )e−iξx dξ,
⎪
⎪ 2π −∞ 2π −∞
⎪
⎪ ∞
⎪
⎪
⎪
⎪ 1
⎪
⎪ u z = ûz e−iξx dξ
⎪
⎪ 2π −∞
⎪
⎪ ∞
⎨ 1
= A(ξ)ξ 2 [(−R6 + λ21 R5 )e−λ1 |ξ|y + α(R6 − λ22 R5 )e−λ2 |ξ|y ]e−iξx dξ,
⎪
⎪ 2π −∞
⎪
⎪ ∞
⎪
⎪
⎪
⎪ 1
ŵz e−iξx dξ
⎪
⎪ wz =
⎪
⎪ 2π −∞
⎪
⎪ ∞
⎪
⎪
⎪
⎪ =−
1
A(ξ)ξ 2 [(−C55 +λ21 C66 )e−λ1 |ξ|y +α(C66 −λ22 C55 )e−λ2 |ξ|y ]e−iξx dξ.
⎩
2π −∞
(7.6-16)
The phason strain field presents important effect in the phase transition of
crystal-quasicrystal, which is determined by the above solution such as
124 Chapter 7 Application I—Some dislocation and interface problems and . . .
⎧ ∞
⎪
⎪ ∂wz 1
⎪
⎪ w zx = = −i A(ξ)ξ 3 [(−C55 + λ21 C66 )e−λ1 |ξ|y
⎪
⎪ ∂x 2π −∞
⎪
⎪
⎪
⎪ +α(C66 − λ22 C55 )e−λ2 |ξ|y ]e−iξx dξ,
⎪
⎪
⎨ ∂wz
wzy = (7.6-17)
⎪
⎪ ∂y
⎪
⎪ ∞
⎪
⎪ 1
⎪
⎪
⎪ = A(ξ)ξ 3 [λ1 |ξ| (−C55 + λ21 C66 )e−λ1 |ξ|y
⎪
⎪ 2π −∞
⎪
⎩ +λ |ξ| α(C − λ2 C )e−λ2 |ξ|y ]e−iξx dξ.
2 66 2 55
The solution varies from material constants Cij , Ki , Ri of quasicrystals, and the
material constant μ(c) of crystals, applied stress τ and the size h of the crystals, so
the results are interesting.
Further discussion on interface will be given in Section 9.2 of Chapter 9.
References
[1] De P, Pelcovits R A. Linear elasticity theory of pentagonal quasicrystals. Phys Rev
B, 1987, 35(16): 8609–8620
[2] De P, Pelcovits R A. Disclination in pentagonal quasicrystals. Phys Rev B, 1987,
36(17): 9304–9307
References 125
[3] Ding D H, Wang R H, Yang W G et al. General expressions for the elastic displace-
ment fields induced by dislocation in quasicrystals. J Phys Condens Matter, 1995,
7(28): 5423–5436
[4] Ding D H, Wang R H, Yang W G et al. Elasticity theory of straight dislocation in
quasicrystals. Phil Mag Lett, 1995, 72(5): 353–359
[5] Li X F, Fan T Y. New method for solving elasticity problems of some planar qua-
sicrystals and solutions. Chin Phys Lett, 1998, 15(4): 278–280
[6] Li X F, Duan X Y, Fan T Y et al. Elastic field for a straight dislocation in a decagonal
quasicrystal. J Phys: Condens Matter, 1999, 11(3): 703–711
[7] Yang S H, Ding D H. Fundamentals to Theory of Crystal Dislocations. Vol II. Beijing:
Science Press, 1998 (in Chinese)
[8] Firth J P, Lothe J. Theory of Dislocations. John Wiley, Sons. New York, 1982
[9] Zhou W M. Dislocation, crack and contact problems in two- and three-dimensional
quasicrystals. Dissertation Beijing Institute of Technology, 2000 (in Chinese)
[10] Li L H. Study on complex variable function method and exact analytic solutions of
elasticity of quasicrystals. Dissertation. Beijing Institute of Technology, 2008 (in
Chinese)
[11] Fan T Y, Li X F, Sun Y F. A moving screw dislocation in a one-dimensional hexagonal
quasicrystals. Acta Physica Sinica (Oversea Edition), 1999, 8(3): 288–295
[12] Li X F, Fan T Y. A straight dislocation in one-dimensional hexagonal quasicrystals.
Phy Stat Sol (b), 1999, 212(1): 19–26
[13] Edagawa K. Dislocations in quasicrystals, Mater Sci Eng A, 2001, 309∼310(2):
528–538
[14] Fan T Y, Trebin H R, Messeschmidt U et al. Plastic flow coupled with a crack
in some one- and two-dimensional quasicrystals. J Phys: Condens Matter, 2004,
16(37): 5229–5240
[15] Hu C Z, Wang R H, Ding D H. Symmetry groups, physical property tensors, elasticity
and dislocations in quasicrystals. Rep Prog Phys, 2000, 63(1): 1–39
[16] Li F H. In: Jacaman M J, Torres M. Crystal-Quasicrystal Transitions. Elsevier Sci
Publ, 1988, 13–47
[17] Li F H, Teng C M, Huang Z R et al. In between crystalline and quasicrystalline
states. Phil Mag Lett, 1988, 57 (1): 113–118
[18] Fan T Y, Xie L Y, Fan L et al. Study on interface of quasicrystal-crystal. Chin.
Phys. B, submitted 2009
[19] Kordak M, Fluckider T, Kortan A R et al. Crystal-quasicrystal interface in Al-Pd-
Mn. Prog Surface Sci, 2004, 75(3∼8): 161–175
Chapter 8
Application II: Solutions of notch and
crack problems of one-and
two-dimensional quasicrystals
Quasicrystals are potential materials to be developed for structural use, and their str-
ength and toughness attract attention of researchers. Experimental observations[1,2]
have shown that quasicrystals are brittle. With common experience of conventional
structural materials, we know that failure of brittle materials is mainly related to
the existence and growth of cracks. Chapter 7 indicated that dislocations have been
observed in quasicrystals, and the accumulation of dislocations will eventually lead
to cracking of the material. Now let us study crack problems in quasicrystals that
have both theoretical and practical value in the view of applications in future.
Chapters 5∼7 have discussed some elasticity and dislocation problems in one-
and two-dimensional quasicrystals. It has shown that when the quasicrystal config-
uration is independent of one coordinate, e.g. variable z, its elasticity problem can
be decoupled into a plane problem and an anti-plane problem. In the case of one-
dimensional quasicrystals, if the z-direction accords with the quasiperiodic axis, the
above plane problem belongs to classic elasticity problem, and the anti-plane problem
is a coupling problem of phonon and phason fields. In the case of two-dimensional
quasicrystals, if z-direction represents the periodic axis, the above plane problem
is a coupling problem of phonon and phason fields, and the anti-plane problem be-
longs to a classical elasticity problem. Due to use of decomposition procedure, the
resulting problem can be dramatically simplified. Chapters 5 and 6 have given their
corresponding fundamental solutions, and Chapter 7 conducted the solutions of dis-
locations in detail. The present chapter is going to focus on crack problems, to
continue using the above schemes, such as the fundamental solutions developed in
Chapters 5 and 6, and the Fourier transform and complex variable function meth-
ods used in Chapter 7. But it emphasizes the complex function method, which
will be developed in Sections 8.1, 8.2 and 8.4, this approach is powerful. Problems
displayed in Sections 8.1 and 8.2 are relatively simpler, the detailed introduction
may help reader understand and further handle the principle and technique of the
complex potential method, though the representation does not beyond the classi-
cal Muskhelishvili[18] method, this is helpful to understand the solutions for more
complicated problems displayed in Section 8.4 and in the next chapter. The further
summary of the method will be introduced in Chapter 11, because the contents in
Section 8.4 and Chapter 9 bring some new insights into the study and go beyond
the Muskhelishvili method, the farther general discussion is necessary and may be
beneficial.
Based on the common nature of exact solutions of different static and dynamic
cracks in different quasicrystal systems for linear and nonlinear deformation (which
are discussed in this chapter and Chapters 9, 10 and 14), the fracture theory of qua-
sicrystalline material is suggested in Chapter 15, which can be seen as a development
of fracture mechanics of conventional structural materials.
can be decomposed into a plane elasticity problem of regular crystal and an anti-
plane elasticity problem of the phonon-phason coupling field. Plane elasticity prob-
lems of regular crystal have been studied extensively in classical elasticity, and its
crack problems have also been studied in the classical fracture theory, see e.g. Ref.
[3]. Therefore, we skip the discussion of crack problems in regular crystal. The anti-
plane elasticity problem of the phason-phonon coupling field is described by using
the following basic equations:
⎧
⎪
⎪ σyz = σzy = 2C44 εyz + R3 wzy
⎪
⎪
⎪
⎨ σzx = σxz = 2C44 εzx + R3 wzx
(8.1-1)
⎪
⎪ Hzy = K2 wzy + 2R3 εzy
⎪
⎪
⎪
⎩ H = K w + 2R ε
zx 2 zx 3 zx
⎧ 1 ∂uz
⎪
⎪ εyz = εzy = ,
⎪
⎪ 2 ∂y
⎪
⎪
⎪
⎪
⎪
⎪ 1 ∂uz
⎪
⎨ εzx = εxz = 2 ∂x
(8.1-2)
⎪
⎪ ∂wz
⎪
⎪ wzy = ,
⎪
⎪ ∂y
⎪
⎪
⎪
⎪
⎪
⎩ w = ∂wz
zx
∂x
∂σzx ∂σzy ∂Hzx ∂Hzy
+ = 0, + =0 (8.1-3)
∂x ∂y ∂x ∂y
The derivation in Chapter 5 shows that the above equations can be reduced to
∇2 uz = 0, ∇ 2 wz = 0 (8.1-4)
Note that the phason stress τ2 at the crack surface is assumed from point of view
physically, though its measurement result has not been reported yet. For simplicity,
we can assume τ2 =0, sometimes.
In the following we are going to solve the boundary value problem of (8.1-4)
and (8.1-5) first. The complex variable function method will be used. To do so, we
introduce the complex variable
√
t = x + iy = reiθ , i = −1 (8.1-7)
From equations (8.1-4), it is known that both uz (x, y) and wz (x, y) are harmonic
functions that can be expressed in terms of the real part or the imaginary part of
two arbitrary analytic functions φ1 (t) and ψ1 (t) of complex variable t in a region
occupied by the quasicrystal. For simplicity we can call φ1 (t) and ψ1 (t) as complex
potentials. Here, assume
uz (x, y) = Reφ1 (t)
(8.1-8)
wz (x, y) = Reψ1 (t)
in which the symbol Re indicates the real part of a complex number.
It is well known that if a function F (t) is analytic, then
∂F dF ∂F idF
= , = . (a)
∂x dt ∂y dt
Furthermore, assume
where symbol Im denotes the imaginary part of a complex number, and P (x, y) and
Q(x, y) represent the real and imaginary parts of F (t), respectively. Therefore, the
Cauchy-Riemann relations of an analytic function is
∂P ∂Q ∂P ∂Q
= , =− . (c)
∂x ∂y ∂y ∂x
With the aid of relation (a), formula (8.1-8) and equations (8.1-1) and (8.1-2)
lead to ⎧
⎪ ∂ ∂
⎪
⎪ σyz = σzy = C44 Reφ1 + R3 Reψ1
⎪
⎪ ∂y ∂y
⎪
⎪
⎪
⎪
⎪
⎪ ∂ ∂
⎪
⎨ σzx = σxz = C44 ∂x Reφ1 + R3 ∂x Reψ1
(8.1-9)
⎪
⎪ ∂ ∂
⎪
⎪ Hzx = K2 Reψ1 + R3 Reφ1
⎪
⎪ ∂x ∂x
⎪
⎪
⎪
⎪
⎪
⎪ ∂ ∂
⎩ Hzy = K2 Reψ1 + R3 Reφ1
∂y ∂y
8.1 Crack problem and solution of one-dimensional quasicrystals 131
Based on the Cauchy-Riemann relation (c), the above equation can be rewritten
as
σzx − iσzy = C44 φ1 + R3 ψ1
(8.1-10)
Hzx − iHzy = K2 ψ1 + R3 φ1
where φ1 = dφ1 /dt, ψ1 = dψ1 /dt.
According to formula (8.1-10), we have
σyz = σzy = −Im(C44 φ1 + R3 ψ1 )
(d)
Hzy = −Im(K2 ψ1 + R3 φ1 )
whose derivation can be obtained in terms of the strict complex variable function
method given in Section 8.7—Appendix of Chapter 8.
From (8.1-12)
⎧
⎪
⎪ i(K2 τ1 − R3 τ2 ) t
⎪
⎨ φ1 (t) = C K − R2 1− √
44 2 3 t2 − a2
(8.1-13)
⎪
⎪ i(C 44 τ 2 − R 3 τ 1 ) t
⎪
⎩ ψ1 (t) = 1− √
C44 K2 − R32 t2 − a2
and substitution of the above expressions into the first expression in (8.1-11) yields
t
σzx − iσzy = iτ1 − √ (8.1-14)
t2 − a2
132 Chapter 8 Application II: Solutions of notch and crack problems of one-and...
⎪ (8.1-15)
⎪
⎪ τ1 r 1 1
⎩ σyz = σzy = cos θ − θ1 − θ2
(r1 r2 )1.2 2 2
where
t = reiθ , t − a = r1 eiθ1 , t + a = r2 eiθ2 (8.1-16)
or
⎧ # # #
⎪
⎪ r = x2 + y 2 , r1 = (x − a)2 + y 2 , r2 = (x + a)2 + y 2
⎨
y (8.1-16 )
⎪
⎪ y y
⎩ θ = arctan , θ1 = arctan , θ2 = arctan
x x−a x+a
which can be shown in Fig. 8.1-2.
Similarly
t
Hzx − iHzy = iτ2 − √ (8.1-17)
t2 − a2
For these stress components there are expressions similar to (8.1-15).
As a result, ⎧
⎨ √ τ1 x , |x| > a
σzy (x, 0) = x2 − a2 (8.1-18)
⎩
0, |x| < a
⎧
⎨ √ τ2 x , |x| > a
Hzy (x, 0) = x2 − a2 (8.1-19)
⎩
0, |x| < a
The above two formulas indicate that at y = 0 and |x| < a : σzy = 0, Hzy = 0.
Therefore, the solution given above also satisfies the boundary conditions at the
crack surfaces.
8.1 Crack problem and solution of one-dimensional quasicrystals 133
#
Formulas (8.1-14) and (8.1-17) also show that when x2 + y 2 → ∞, σyz = τ1 ,
σxz = 0 and Hzy = τ2 , Hzx = 0, namely, the solution given above satisfies boundary
conditions at infinity.
The above formulas show that stresses have singular characteristics near crack tips,
for example
⎧ τ1 x +
⎨ σzy (x, 0) = √x2 − a2 → ∞, x → a
⎪
(8.1-20)
⎪
⎩ Hzy (x, 0) = √ τ2 x → ∞, x → a+
x2 − a2
Define the mode III stress intensity factors of the phonon and phason fields such
that:
||
#
KIII = lim
+
2π(x − a)σzy (x, 0),
x→a
⊥
#
KIII = lim 2π(x − a)Hzy (x, 0),
x→a+
then
|| √ ⊥
√
KIII = πaτ1 , KIII = πaτ2 , (8.1-21)
where suffix “III” stands for model III (longitudinal shearing mode) [3].
Now let us calculate the crack strain energy:
a
WIII = 2 (σzy ⊕ Hzy ) (uz ⊕ wz )dx
0 a
=2 [σzy (x, 0)uz (x, 0) + Hzy (x, 0)wz (x, 0)]dx (8.1-22)
0
From (8.1-24), we can determine the crack strain energy release rate (crack growth
force) such that
Clearly, the crack energy and energy release rate are related not only to the phonon
but also to the phason and the phonon-phason coupling fields.
If τ2 = 0, then
K2 (KIII )2
GIII = (8.1-26)
C44 K2 − R32
Furthermore if R3 = 0, we have
||
πaτ12 (KIII )2
GIII = , or GIII = . (8.1-27)
C44 C44
Since GIII comprehensively describes the coupling effect of the phonon and phason
fields and the stress states near crack tip, we recommend
as the fracture criterion of quasicrystals under mode III deformation, where GIIIC
is the critical (threshold) value of GIII , the mode III fracture toughness to be deter-
mined by testing.
The stress intensity factor and the strain energy release rate are fundamental
physical parameters, and constitute the basis of brittle fracture theory for both the
conventional crystalline as well as the quasicrystalline materials.
Formulas (8.1-1)∼(8.1-10) in the preceding section still hold in this section, and
other notation and symbol are similar. Therefore, formula (8.1-11) should be modi-
fied as
C44 (φ1 − φ1 ) + R3 (ψ1 − ψ1 ) = 2iτ1 f (x),
(8.2-2)
K2 (ψ1 − ψ1 ) + R3 (φ1 − φ1 ) = 0,
where
0, x < −a,
f (x) = (8.2-3)
1, −a < x < 0.
The conformal mapping
2
H 1+ζ
t = ω(ζ) = ln 1 + , (8.2-4)
π 1−ζ
maps the domain in the t-plane onto the interior of the unit circle γ in the ζ-plane,
ζ = ξ + iη. Therefore, crack tip t = 0 corresponds to ζ = −1, and t = −a accords to
two points on the unit circle in the ζ-plane such that
136 Chapter 8 Application II: Solutions of notch and crack problems of one-and...
⎧ √
⎪
⎪ −e−πa/H + 2i 1 − e−πa/H
⎪ σ
⎨ −a = ,
2 − e−πa/H
√ (8.2-5)
⎪
⎪ −πa/H
⎩ σ−a = −e
⎪ − 2i 1 − e−πa/H
,
2 − e−πa/H
where σ = eiϕ = ζ||ζ|=1 denotes the value of ζ on γ.
Equations in Section 8.1 and the Appendix of this chapter (i.e., Section 8.7) are
useful. Nevertheless, the first 2iω (σ)τ in (8.7-5) should be changed to 2if ω (σ)τ ,
and the first term in (8.7-5)
2iτ 1 ω (σ)
dσ
C44 2πi γ σ−ζ
should be modified as
2iτ 1 ω (σ)
f dσ.
C44 2πi γ σ−ζ
Thus, equation (8.7-5) for the present problem becomes
⎧
⎪ R3 2iτ1 1 ω (σ)
⎪ φ (ζ) +
⎨ ψ (ζ) = f dσ,
C44 C44 2πi γ σ − ζ
(8.2-6)
⎪
⎪ R3
⎩ ψ (ζ) + φ (ζ) = 0,
K2
where f is the function given by formula (8.2-3), which takes values between σ−a
and σ−a in the ζ-plane.
Integration of the right-handside of equation (8.2-6) leads to
1 ω (σ) 1 1 1+ζ
f dσ = ln(σ − 1) − ln(σ − ζ)
2πi γ σ − ζ 2πτ 1 − ζ (1 − ζ)(1 + ζ 2 )
σ=σ−a
ζ 2 1 σ−i
− ln(1 + σ ) + ln
2(1 + ζ 2 ) 2(1 − ζ 2 ) σ + i σ=σa
≡F (ζ) (8.2-7)
where σ−a and σ−a are given by formula (8.2-5). Based on the above relation, (8.2-6)
is further reduced to
K2 τ1 R3 τ1
φ (ζ) = 2iF (ζ), ψ = − 2iF (ζ). (8.2-8)
C44 K2 − R32 C44 K2 − R32
Now let us calculate the stresses. During this process, the following relations will
be used:
φ (ζ) 2iK2 τ1 F (ζ)
φ1 (t) = = · ,
ω (ζ) C44 K2 − R32 ω (ζ)
8.2 Crack problem in finite-sized one-dimensional quasicrystals 137
which shows that the stress distribution is independent of the material constants.
Substitution of (8.2-7) into (8.2-9) leads to the explicit expressions of stresses.
The forms of F (ζ) and F (ζ) are relatively complex, while the expressions of stress
in terms of variable ζ(= ξ + iη) are concise. In an attempt to invert them to the
t-plane, inverse of transform (8.2-4) must be used,
√
−1 −(eπt/H − 2) ± 2i eπt/H − 2
ζ = ω (t) = . (8.2-10)
eπt/H − 2
Substitution of (8.2-10) into (8.2-9) yields to the final expressions of σzx , σzy , Hzx
and Hzx in the t-plane, which are very complex. Here we skip this procedure.
Now let us calculate the stress intensity factors. According to (8.1-15) and
(8.1-16), it is known that at the region such that r1 /a 1,
|| ||
K θ1 K θ1
σzx = − √ III sin , σzy = √ III cos ,
2πr1 2 2πr1 2
therefore
||
K θ1 θ1 K
σzx − iσzy = − √ III sin − i cos = − √ III , (8.2-11)
2πr1 2 2 2πt1
t1 = ω(ζ1 ), (8.2-12)
√ F (ζ) ⊥
KIII = lim πτ1 # , KIII =0 (8.2-13)
ζ→−1 ω (ζ)
√ √
2Hτ1 2eπa/H − 1 + 2eπa/H 1 − e−πa/H
KIII = ln √ , (8.2-14)
2π 2eπa/H − 1 − 2eπa/H 1 − e−πa/H
⊥
If we do not assume τ2 = 0 in (8.2-1), then the stress intensity factor KIII can be
evaluated, and the expression is similar to (8.2-14), this is the extension of work
given by Ref. [4] for classical elasticity to quasicrystal elasticity.
The configuration is shown in Fig. 8.2-2, there are the following boundary conditions
⎧
⎪ y = ±H, −∞ < x < ∞ : σzy = 0, Hzy = 0
⎪
⎪
⎪
⎪
⎪
⎪ x = ±∞, −H < y < H : σzx = 0, Hzx = 0
⎪
⎪
⎨
y = ±0, −∞ < x < −a : σzy = 0, Hzy = 0; (8.2-15)
⎪
⎪
⎪
⎪
⎪
⎪ −a < x < 0 : σzy = −τ1 , Hzy = −τ2 ;
⎪
⎪
⎪
⎩
L < x < ∞ : σzy = 0, Hzy = 0
transforms the region at z-plane onto the interior of the unit circle γ at ζ-plane, in
which
1 − e−πa/H
α= , β = e−πL/H . (8.2-17)
1 − e−π(a+L)/H
Then substitute (8.2-17) into (8.2-6), we can find the solution φ (ζ) so the stress
intensity factors such as
8.3 Griffith crack problems in point groups 5m and 10mm quasicrystals based... 139
⎧
⎪ (0,0) # F (ζ)
⎪
⎪ KIII = lim −2 2πω(ζ)τ1
⎪
⎪ ω (ζ)
⎪
⎪
ζ→1+0
⎪
⎪ √ √ √
⎪
⎪ 2Hτ1 1+ α √ 1 + αβ
⎪
⎪ = √ ln √ − β ln √ ,
⎨ π 1−β 1− α 1 − αβ
(8.2-18)
⎪
⎪ (L,0) # F (ζ)
⎪
⎪ KIII = lim −2 2π(L − ω(ζ))τ1
⎪
⎪ ω (ζ)
⎪
⎪
ζ→−1−0
⎪
⎪ √ √ √
⎪
⎪ 2Hτ1 √ 1+ α 1 + αβ
⎪
⎩ = √ β ln √ − ln √ ,
π 1−β 1− α 1 − αβ
in which
K2 τ1 R3 τ1
φ (ζ) = 2iF (ζ), ψ (ζ) = − 2iF (ζ),
C44 K2 − R32 C44 K2 − R32
√ √
2H αA αβM
F (ζ) = 2 −
π (1 + ζ)2 + α(1 − ζ)2 (1 + ζ)2 + αβ(1 − ζ)2
2H iα(1 − β)(1 − ζ 2 ) i−ζ
+ 2 2 2 2 2
ln , (8.2-19)
π [(1 + ζ) + α(1 − ζ) ][(1 + ζ) + αβ(1 − ζ) ] 1 − iζ
√ √
(1 + α) 1 + αβ
A = ln √ , M = ln √ .
1− α 1 − αβ
⊥
If we do not assume τ2 = 0 in (8.2-15), then the stress intensity factor KIII can
be similarly evaluated, and the expression is similar to (8.2-18), this extends the
study for the classical elasticity. The detail can be found in Refs. [5] and [6], some
calculations on the function F (ζ) refer to Major Appendix of this book.
section, we are going to solve this problem using method of stress functions. On
the one hand, this demonstrates the problem-solving procedure based on method of
stress functions; on the other hand this is targeted to examine the results obtained
by the method of displacement functions. For a correct solution, which can be
examined using any available method.
(∞)
Consider a Griffith crack under the action of external traction, i.e., σyy = p, and
the crack is assumed to penetrate the periodic axis(z-direction) of the quasicrystals,
as shown in Fig. 8.3-1. Similar to the analysis in the preceding section, within
the framework of Griffith’s theory, this problem can be replaced by an equivalent
crack problem shown in Fig. 8.3-2. Furthermore, assume the external traction being
independent of z, therefore the deformation of the quasicrystal is also independent
of z, namely
∂ui ∂wi
= 0, = 0, i = 1, 2, 3. (8.3-1)
∂z ∂z
According to the analysis performed in Chapter 6, under this case, the two-
dimensional quasicrystal elasticity problem can be decoupled into a plane elasticity
problem of phonon-phason coupling and an anti-plane pure elasticity problem. In
this case, the latter only has a trivial solution under mode I external traction, which
can be neglected. The plane elasticity problem of phonon-phason coupling with
point groups 5m and 10mm has been studied in Section 6.2, and its final governing
equation is
∇2 ∇2 ∇2 ∇2 F = 0. (8.3-2)
Here F (x, y) is the displacement potential function introduced in Section 6.2.
Fig. 8.3-1 Griffith crack along the periodic axis of quasicrystal and
subjected to a tension
8.3 Griffith crack problems in point groups 5m and 10mm quasicrystals based... 141
Fig. 8.3-2 The same Griffith crack as Figure 8.3-1 with external traction
acting on crack surfaces
As shown in Fig. 8.3-2, the Griffith crack is under the action of uniform traction
at crack surfaces and without far-field traction, i.e., σyy (x, 0) = −p, |x| < a. We
modify this problem into the semi-plane problem, i.e., only study the case in the
upper half-plane or the lower half-plane under the following conditions:
⎧ #
⎪
⎪ x2 + y 2 → ∞ : σij = Hij = 0,
⎪
⎪
⎪
⎪
⎪
⎪ y = 0, |x| < a : σyy = −p, σyx = 0,
⎨
Hyy = 0, Hyx = 0, (8.3-3)
⎪
⎪
⎪
⎪
⎪
⎪ y = 0, |x| > a : σyx = 0, Hyx = 0,
⎪
⎪
⎩
uy = 0, wy = 0.
By performing the Fourier transform on equation (8.3-2)
∞
F̂ (ξ, y) = F (x, y)eiξx dx. (8.3-4)
−∞
stands for the transpose of a matrix. Fourier transforms of the displacement and
stress components can be expressed in term of F̂ (ξ, y), i.e., X and Y as discussed in
Section 7.2.
Solution (8.2-6) has satisfied the boundary condition (8.2-3) at infinity, and the
left boundary condition in (8.2-3) results in
⎧
⎪ 21C(ξ)|ξ| − 3(32 − e2 )D(ξ)
⎪
⎨ A(ξ) = 3 ,
2 |ξ|
(8.3-7)
⎪
⎩ B(ξ) = 6C(ξ)|ξ| − 21D(ξ)
⎪
ξ2
and the following set of dual integral equations:
⎧ ∞
⎪ 2
⎪
⎪ [C(ξ)ξ − 6D(ξ)] cos(ξx)dξ = −p, 0 < x < a,
⎪
⎪
⎪ d11 0
⎪
⎪
⎪ ∞
⎪
⎪
⎪
⎨ ξ −1 [C(ξ)ξ − 6D(ξ)] cos(ξx)dξ = 0, x > a,
0
∞ (8.3-8)
⎪
⎪ 2
⎪
⎪ D(ξ) cos(ξx)dξ = 0, 0 < x < a,
⎪
⎪ d12 0
⎪
⎪
⎪
⎪ ∞
⎪
⎪
⎩ ξ −1 D(ξ) cos(ξx)dξ = 0, x > a.
0
2αβ α−β
e2 = + ,
ω(α − β)(K1 − K2 ) α + β
M (L + 2M )
α = R(L + 2M ) − ωK1 , β = RM − ωK1 , ω= ,
R
d11 and d22 are given as
⎧ nR
⎪
⎪ d11 = ,
⎪
⎪ (4M/L + M )(L + 2M )(M K1 − R2 )
⎪
⎪
⎪
⎪
⎨ nR2
d12 = , (8.3-9)
⎪ d0 M (L + 2M )
⎪
⎪
⎪
⎪ d0 = −{(M K1 − R2 )[(L + 2M )(K1 + K2 ) − 2R2 ]
⎪
⎪
⎪
⎩
−[(L + 2M )K1 − R2 ][M (K1 + K2 ) − 2R2 ]}
n = M α − (L + 2M )β.
8.3 Griffith crack problems in point groups 5m and 10mm quasicrystals based... 143
The theory of dual integral equations is offered in the Major Appendix of this mono-
graph. Accordingly, solution to the set of equations (8.3-8) is
where J1 (aξ) is the first order Bessel function of the first kind.
So far, the unknown functions A(ξ), B(ξ), C(ξ), and D(ξ) have been determined
completely. In the view of mathematics, this problem has been solved in the Fourier
transform space. However, in the view of physics, we need to perform the Fourier
inverse ∞
1
F (x, y) = F̂ (ξ, y)e−iξx dξ (8.3-11)
2π −∞
in order to express the field variables in the physical space.
Obviously, once F (x, y) is determined from the integral (8.3-11), uj , σjk , and
Hjk can be determined by substituting F (x, y) into (8.3-8)∼(8.3-11). Alternatively,
ûj (ξ, y), ŵj (ξ, y), σ̂jk (ξ, y), and Ĥjk (ξ, y) can be determined by substituting X(ξ)
and Y (ξ) into (8.3-8)∼(8.3-11), and then their Fourier inverses finally lead to uj , σjk ,
and Hjk . Luckily, the above integrals with Bessel function can be expressed explicitly
using elementary functions. Nevertheless, their final expressions in terms of variables
x and y are extremely complex. However, the final expressions appear more concise
if using (r, θ), (r1 , θ1 ), and (r2 , θ2 ) to stand for the three polar coordinate systems
with the crack center, left crack tip, and right crack tip as origins, respectively, as
shown in Fig. 8.1-2, similar to (8.1-16), i.e.,
x = r cos θ = a + r1 cos θ1 = −a + r2 cos θ2 ,
(8.3-12)
y = r sin θ = r1 sin θ1 = r2 sin θ2 .
The following infinite integrals involving Bessel functions are used for stress cal-
culation: ⎧
⎪
⎪
∞
1 z
⎪
⎪ −ξz
J1 (aξ)e dξ = 1− ,
⎪
⎪ 0 1/2
⎪
⎪
a (a2 − z 2 )
⎪
⎪ ∞ ∞
⎪
⎪
⎪
⎪ ξJ1 (aξ)e−ξz dξ =
a
⎪
⎨ 0 ξJ 1 ,
0 (a2 − z 2 )3/2
(8.3-13)
⎪ ∞
⎪ 3az
⎪
⎪ 2 −ξz
ξ J1 (aξ)e dξ = ,
⎪
⎪
⎪
⎪ 0 (a2 − z 2 )5/2
⎪
⎪ ∞
⎪
⎪
⎪
⎪ 3 −ξz 3a(4z 2 − a2 )
⎪
⎩ ξ J 1 (aξ)e dξ = 7/2
,
0 (a2 − z 2 )
where z = x + iy.
144 Chapter 8 Application II: Solutions of notch and crack problems of one-and...
From Section 6.7, we find that, based on stress potential method the final governing
equation of plane elasticity of point group 10, 10 decagonal quasicrystals is
∇2 ∇2 ∇2 ∇2 G = 0. (8.4-1)
Substituting expression (8.4-2) into equation (6.7-6) then into equations (6.7-2) leads
to
⎧
⎪ σxx = −32c1 Re[Ω (z) − 2g4 (z)],
⎪
⎪
⎪
⎪
⎪
⎪ σyy = 32c1 Re[Ω (z) + 2g4 (z)],
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ σxy = σyx = 32c1 ImΩ (z),
⎪
⎨
Hxx = 32R1 Re[Θ (z) − Ω (z)) − 32R2 Im(Θ (z) − Ω (z)], (8.4-3)
⎪
⎪
⎪
⎪
⎪
⎪ Hxy = −32R1 Im[Θ (z) + Ω (z)) − 32R2 Re(Θ (z) + Ω (z)],
⎪
⎪
⎪
⎪
⎪
⎪ Hyx = −32R1 Im[Θ (z) − Ω (z)) − 32R2 Re(Θ (z) − Ω (z)],
⎪
⎪
⎪
⎪
⎩
Hyy = −32R1 Re[Θ (z) + Ω (z)) + 32R2 Im(Θ (z) + Ω ],
146 Chapter 8 Application II: Solutions of notch and crack problems of one-and...
where ⎧
⎨ Θ(z) = g (IV) (z) + z̄g (IV) (z) + 1 z̄ 2 g (IV) (z)
2 3 4
2 (8.4-4)
⎩ (IV) (IV)
Ω (z) = g3 (z) + z̄g4 (z)
in which the prime, two prime, three prime and superscript (IV) denote the first
to fourth order derivatives of gj (z)(j = 1, · · · , 4) to variable z, in addition Θ (z) =
dΘ(z)/dz.
We further derive the complex representations of displacement components of
phonon and phason fields. The first two equations of (6.7-3) can be rewritten as
⎧
⎪ K1 + K2 1
⎪
⎨ εxx = c2 (σxx + σyy ) − σyy − [R1 (Hxx + Hyy ) + R2 (Hxy − Hyx )],
2c 2c
⎪
⎪ K + K 1
⎩ εyy = c2 (σxx + σyy ) − 1 2
σxx + [R1 (Hxx + Hyy ) + R2 (Hxy − Hyx )],
2c 2c
(8.4-5)
where
c + (L + M )(K1 + K2 )
c2 = . (8.4-6)
4(L + M )c
Substituting equations (8.4-3) into (8.4-5) and by integration yield
K1 + K2 ∂
ux =128c1 c2 Reg4 (z) − φ
2c ∂x
32(R12 + R22 )
+ Re[g3 (z) + z̄g4 (z) − g4 (z)] + f1 (y),
c
K1 + K2 ∂
uy =128c1 c2 Img4 (z) − φ
2c ∂y
32(R12 + R22 )
− Im[g3 (z) + z̄g4 (z) + g4 (z)] + f2 (x).
c
With these results and other equations of equations (6.7-3) one finds that
df1 (y) df2 (x)
− = .
dy dx
This means these two functions must be constants which only give rigid-body dis-
placements. Omitting the trial functions f1 (y), f2 (x), one obtains
ux + iuy = 32(4c1 c2 − c3 − c1 c4 )g4 (z) − 32(c1 c4 − c3 )[g3 (z) + zg4 (z)], (8.4-7)
where
R12 + R22 K1 + K 2
c3 = , c4 = . (8.4-8)
c c
Similarly, the complex representations of displacement components of phason
fields can be expressed as follows:
32(R1 − iR2 )
wx + iwy = Θ(z). (8.4-9)
K1 − K2
8.4 Stress potential function formulation and complex variable function method... 147
Fig. 8.4-1 (a) The elliptic notch in a decagonal quasicrystal subjected to a uniform
pressure and traction tree at infinity; (b) An infinite decagonal quasicrystal with an
elliptic notch subjected to a tension
148 Chapter 8 Application II: Solutions of notch and crack problems of one-and...
For the problem shown by Fig. 8.4-1 (a) the boundary conditions can be expressed
as follows:
1
g4 (z) + g3 (z) + z̄g4 (z) = − pz̄, z ∈ L. (8.4-13)
32c1
R1 ImΘ(z) + R2 ReΘ(z) = 0,
z ∈ L. (8.4-14)
−R1 ReΘ(z) + R2 ImΘ(z) = 0,
Multiplying the second formula of (8.4-14) by i and adding it to the first, one obtains
Θ(z) = 0, z ∈ L. (8.4-15)
Because the function g1 (z) does not appear in the displacement and stress formu-
las, boundary equations (8.4-12), (8.4-13) and (8.4-15) are enough for determining
the unknown functions g2 (z), g3 (z) and g4 (z). However the calculation cannot be
completed at the z-plane due to the complicity of the evaluation, we must use the
conformal mapping
1
z = ω(ζ) = R0 + mζ (8.4-16)
ζ
to transform the region with ellipse at the z-plane onto the interior of the unit circle γ
a+b a−b
at the ζ-plane, refer to Fig.8.4-2, where ζ = ξ+iη = ρeiϕ and R0 = ,m= .
2 a+b
8.4 Stress potential function formulation and complex variable function method... 149
Fig. 8.4-2 Conformal mapping from the region at z-plane onto the interior of the unit
circle γ at ζ-plane
And we have
Φj (ζ)
Fj (z) = Fj (ω(ζ)) = Φj (ζ), Fj (z) = , (j = 1, · · · , 4). (8.4-18)
ω (ζ)
Substituting (8.4-17) into (8.4-12), (8.4-13) and (8.4-15), then multiplying both sides
1 dσ
of equations by , and integrating along the unit circle, we have
2πi σ − ζ
1 Φ4 (σ)dσ 1 Φ3 (σ)dσ 1 ω(σ) Φ4 (σ)dσ
+ +
2πi γ σ−ζ 2πi γ σ − ζ 2πi γ ω(σ) σ − ζ
p 1 ω(σ)dσ
=− ,
32c1 2πi γ σ − ζ
1 Φ4 (σ)dσ 1 Φ3 (σ)dσ 1 ω(σ) Φ4 (σ)dσ
+ +
2πi γ σ − ζ 2πi γ σ − ζ 2πi γ ω(σ) σ − ζ
p 1 ω(σ)dσ
=− ,
32c1 2πi γ σ − ζ
1 Φ2 (σ)dσ 1 ω(σ) Φ3 (σ)dσ
+
2πi γ σ − ζ 2πi γ ω (σ) σ − ζ
2 2
1 ω(σ) Φ4 (σ)dσ ω(σ) ω (σ) Φ4 (σ)dσ
+ − = 0, (8.4-19)
2πi γ [ω (σ)]2 σ − ζ γ [ω (σ)]3 σ−ζ
pR0 (1 + m2 )ζ
Φ3 (ζ) = ,
32c1 mζ 2 − 1
pR0
Φ4 (ζ) = − mζ. (8.4-20)
32c1
Substitution of
2
ω(σ) σ2 + m ω(σ) ω (σ) 2σ(σ 2 + m)2
= σ , = −
ω (σ) mσ 2 − 1 ω (σ)3 (mσ 2 − 1)3
1 σ 2 + m Φ3 (σ)dσ ζ2 + m
σ = ζ Φ (ζ),
2πi γ mσ 2 − 1 σ − ζ mζ 2 − 1 3
1 σ(σ 2 + m)2 Φ4 (σ)dσ ζ(ζ 2 + m)2
= Φ (ζ).
2πi γ (mσ 2 − 1)3 σ − ζ (mζ 2 − 1)3 4
Utilizing the above mentioned results, the phonon and phason stresses can be de-
termined at the ζ-plane. We here only give a simple example, i.e., along the edge of
notch (ρ = 1), there are phonon stress components such as
1 − 3m2 + 2m cos 2ϕ
σϕϕ = p , σρρ = −p, σρϕ = σϕρ = 0,
1 + m2 − 2m cos 2ϕ
which are identical to the well-known results of the classical elasticity theory.
8.4 Stress potential function formulation and complex variable function method... 151
The detail for some further principle of the complex function method will be
discussed in-depth in Chapter 11. For the further implications and applications of
the results to fracture mechanics of quasicrystals, refer to Chapter 15.
4
σyy = −2c3 c4 Re (1 + 2μ2k + μ4k )gk (zk ), (8.5-4b)
k=1
4
σxy = σyx = 2c3 c4 Re (μk + 2μ3k + μ5k )gk (zk ), (8.5-4c)
k=1
8.6 Other solutions of crack problems in one- and two-dimensional quasicrystals 153
4
Hxx = RRe [(4c4 − c3 )μ2k + 2(3c3 − 2c4 )μ4k − c3 μ6k )]gk (zk ), (8.5-4d)
k=1
4
Hxy = −RRe [(4c4 − c3 )μk + 2(3c3 − 2c4 )μ3k − c3 μ5k )]gk (zk ), (8.5-4e)
k=1
4
Hyx = −RRe [c3 μk + 2(c4 − 2c3 )μ3k − (4c4 − c3 )μ5k )]gk (zk ), (8.5-4f)
k=1
4
Hyy = RRe [c3 + 2(c4 − 2c3 )μ2k − (4c4 − c3 )μ4k )]gk (zk ), (8.5-4g)
k=1
in which
∂ 6 Gk (zk ) dgk (zk )
gk (zk ) = , gk (zk ) = ,
∂zk6 dzk
(K1 + K2 + K3 )M − R2 K1 M − R 2
c3 = , c4 = .
K1 + K2 + 2K3 K1 − K 2
We now consider an elliptic hole L : x2 /a2 + y 2 /b2 = 1, at which there are the
boundary conditions
σxx cos(n, x)+σxy cos(n, y) = Tx , σxy cos(n, x)+σyy cos(n, y) = Ty , (x, y) ∈ L,
Hxx cos(n, x)+Hxy cos(n, y) = hx , Hyx cos(n, x)+Hyy cos(n, y) = hy , (x, y) ∈ L.
(8.5-5)
The complex variable zk can be rewritten as
zk = xk + iyk ,
(8.5-6)
xk = x + αk y, yk = βk y.
The second formula of equations (8.5-6) represents a coordinate transformation.
Further taking conformal mapping, the complex potentials satisfying the bound-
ary conditions can be determined. The detail is omitted here.
The characteristics of crack solutions mentioned here and those given in the
previous sections provide a basis of fracture theory of quasicrystalline materials
summarized in Chpater 15.
which maps the domain in the t-plane with a Griffith crack onto the interior of the
unit circle γ in the ζ-plane (ζ = ξ + iη = ρeiϕ ), and the elliptic hole L onto a unit
circle γ (similar to Fig. 8.4-2).
On the unit circle γ, ζ = σ ≡ eiϕ , where ρ = 1. Furthermore, with the conformal
mapping (8.7-3), the unknown functions φ1 (t) and ψ1 (t) and their corresponding
derivatives can be expressed as
φ1 (t) = φ1 [ω(ζ)] = φ(ζ), ψ1 (t) = ψ1 [ω(ζ)] = ψ(ζ)
(8.7-4)
φ1 (t) = φ (ζ)/ω (ζ), ψ1 (t) = ψ (ζ)/ω (ζ)
With the similar procedure that adopted in Section 8.4, we can find
⎧
⎪
⎪ 1 φ (σ) 1 ω (σ) dσ
⎪
⎪ dσ = φ (ζ), φ (σ) =0
⎪
⎪ 2πi γ σ − ζ
2πi γ ω (σ) σ−ζ
⎪
⎪
⎨ 1 ψ (σ) 1 ω (σ) dσ
dσ = ψ (ζ), − ψ (σ) =0 (a)
⎪
⎪ 2πi γ σ − ζ
2πi γ ω (σ) σ−ζ
⎪
⎪
⎪
⎪ 1 ω (σ) a
⎪
⎪
⎩ 2πi σ−ζ
dσ =
2
γ
⎛ & ⎞
2
C44 τ2 − R3 τ1 ⎝ t t
= ia − − 1⎠
C44 K2 − R32 a a
References
[1] Hu C Z, Yang W Z, Wang R H et al. Symmetry and physical properties of quasicrys-
tals. Adv Phys, 1997, 17(4): 345–376 (in Chinese)
[2] Meng X M, Dong B Y, Wu Y K. Mechanical property of quasi-crystal Al65 Cu20 Co15 .
Acta Metal Sinica, 1994, 30(2): 61–64 (in Chinese)
[3] Kanninen M F, Popelar C H, Advanced Fracture Mechanics. Cambridge: Cambridge
University Press, 1985; Fan T Y. Foundation of Fracture Theory. Beijing: Science
Press, 2003 (in Chinese)
[4] Fan T Y. Exact analytic solutions of stationary and fast propagating cracks in a
strip. Science in China, A, 1991, 34(5): 560–569
[5] Fan T Y. Mathematical Theory of Elasticity of Quasicrystals and Its Application.
Beijing Institute of Technology Press. Beijing, 1999 (in Chinese)
[6] Li L H, Fan T Y. Exact solutions of two semi-infinite collinear cracks in a strip of
one-dimensional hexagonal quasicrystal. Applied Mathematics and Computation,
2008, 196(1): 1–5
[7] Shen D W, Fan T Y. Exact solutions of two semi-infinite collinear cracks in a strip.
Eng Fracture Mech, 2003, 70(8): 813–822
[8] Li X F, Fan T Y, Sun Y F. A decagonal quasicrystal with a Griffith crack. Phil Mag
A, 1999, 79(8): 1943–1952
[9] Li L H, Fan T Y. Complex function method for solving notch problem of point group
10, 10 two-dimensional quasicrystals based on the stress potential function. J Phys:
Condens Matter, 2007, 18(47): 10631–10641
[10] Li X F. Defect problems and their analytic solutions of the theory of elasticity of
quasicrystals. Dissertation. Beijing Institute of Technology, 1999 (in Chinese)
[11] Zhou W M, Fan T Y. Plane elasticity problem of two-dimensional octagonal qua-
sicrystal and crack problem. Chin Phys, 2001, 10(8): 743–747
[12] Zhou W M. Mathematical analysis of elasticity and defects of two- and three-dimens-
ional quasicrystals . Dissertation. Beijing: Beijing Institute of Technology, 2000 (in
Chinese)
[13] Li L H. Study on complex variable function method and analytic solutions of elasticity
of quasicrystals. Dissertation. Beijing: Beijing Institute of Technology, 2008 (in
Chinese)
References 157
mathematical theory of the elasticity and the analytic solutions. Because of the large
number of field variables and field equations involving elasticity of these two kinds
of three-dimensional quasicrystals, the solution presents tremendous difficulty. We
continue to develop the decomposition procedure adopted in the previous chapters,
this can reduce the number of the field variables and field equations, and three-
dimensional elasticity can be simplified to two-dimensional elasticity to solve for
some cases with important practical applications. The introducing of displacement
potentials or stress potentials[4,5] can further simplify the problems. In the work
some systematic and direct methods of mathematical physics and function theory
have been developed, and a series of analytic solutions are constructed, which will
be included in the chapter. Because the calculations are very complex, we would like
to introduce them in detail as much as possible in order to facilitate comprehension
of the text.
which are similar in form to those given in previous chapters, but here ui and wi
have 6 components, and εij and wij have 15 components in total.
The equilibrium equations are as follows:
∂σij ∂Hij
= 0, = 0, (9.1-2)
∂xj ∂xj
which are also similar in form to those listed in previous chapters, however here
adding σij and Hij gives 15 stress components.
Between the stresses and strains there is the generalized Hooke’s law such as
σij = Cijkl εkl + Rijkl wkl , Hij = Rklij εkl + Kijkl wkl , (9.1-3)
[εij , wij ] = [ε11 ε22 ε33 ε23 ε31 ε12 w11 w22 w33 w23 w32 w12 w32 w13 w21 ] (9.1-5 )
9.1 Basic equations of elasticity of icosahedral quasicrystals 161
and the stress components are also arranged according to the same order, i.e.,
[σij , Hij ] = [σ11 σ22 σ33 σ23 σ31 σ12 H11 H22 H33 H23 H12 H32 H13 H21 ] (9.1-5 )
⎡ ⎤
1 1 1 0 0 0 0 1 0
⎢
⎢ −1 −1 1 0 0 0 0 −1 0 ⎥
⎥
⎢ ⎥
⎢ 0 0 −2 0 0 0 0 0 0 ⎥
⎢ ⎥
⎢ ⎥
⎢ 0 0 0 0 0 −1 1 0 −1 ⎥
⎢ ⎥
[R] = R ⎢
⎢ 1 −1 0 0 1 0 0 0 0 ⎥
⎥. (9.1-6)
⎢ ⎥
⎢ 0 0 0 −1 0 −1 0 0 1 ⎥
⎢ ⎥
⎢ 0 0 0 0 0 −1 1 0 −1 ⎥
⎢ ⎥
⎢ ⎥
⎣ 1 −1 0 0 1 0 0 0 0 ⎦
0 0 0 −1 0 −1 0 0 1
⎧
⎪
⎪ σxx = λθ + 2μεxx + R(wxx + wyy + wzz + wxz ),
⎪
⎪
⎪
⎪ σyy = λθ + 2μεyy − R(wxx + wyy − wzz + wxz ),
⎪
⎪
⎪
⎪ σzz = λθ + 2μεyy − 2Rwzz ,
⎪
⎪
⎪
⎪ σyz = 2μεyz + R(wzy − wxy − wyx ) = σzy ,
⎪
⎪
⎪
⎪ σzx = 2μεzx + R(wxx − wyy − wzx ) = σxz ,
⎪
⎪
⎪
⎪ σxy = 2μεxy + R(wyx − wyz − wxy ) = σyx ,
⎪
⎪
⎨ Hxx = R(εxx − εyy + 2εzx ) + K1 wxx + K2 (wzx + wxz ),
Hyy = R(εxx − εyy − 2εzx ) + K1 wyy + K2 (wxz − wzx ), (9.1-7)
⎪
⎪
⎪
⎪ Hzz = R(εxx + εyy − 2εzz ) + (K1 + K2 )wzz ,
⎪
⎪
⎪
⎪ Hyz = −2Rεxy + (K1 − K2 )wyz + K2 (wxy − wyx ),
⎪
⎪
⎪
⎪ Hzx = 2Rεzx + (K1 − K2 )wzx + K2 (wxx − wyy ),
⎪
⎪
⎪
⎪ Hxy = −2R(εyz + εxy ) + K1 wxy + K2 (wyz − wzy ),
⎪
⎪
⎪
⎪ Hzy = 2Rεyz + (K1 − K2 )wzy − K2 (wxy + wyx ),
⎪
⎪
⎪
⎪ Hxz = R(εxx − εyy ) + K2 (wxx + wyy ) + (K1 − K2 )wxz ,
⎩
Hyx = 2R(εxy − εyx ) + K1 wyx − K2 (wyz + wzy ),
where θ = εxx + εyy + εzz denotes the volume strain and εij and wij are defined by
(9.1-1). This explicit expression was first given by Ding et al[1] .
Substituting (9.1-7) into (9.1-2) yields one of forms of the final governing equations—
the
⎧ equilibrium equations in terms of displacements as follows:
⎪
⎪ ∂
⎪
⎪ μ∇2 ux + (λ + μ) ∇ · u
⎪
⎪ 2 ∂x 2
⎪
⎪
⎪
⎪ ∂ wx ∂ wx ∂ 2 wx ∂ 2 wy ∂ 2 wy ∂ 2 wz
⎪
⎪ + R + − + 2 − 2 + 2 = 0,
⎪
⎪ ∂x2 ∂x∂z ∂y 2 ∂x∂y ∂y∂z ∂x∂z
⎪
⎪
⎪
⎪ ∂
⎪
⎪ μ∇2 uy + (λ + μ) ∇ · u
⎪
⎪ ∂y
⎪
⎪ 2
⎪
⎪ ∂ wy ∂ 2 wx ∂ 2 wy ∂ 2 wx ∂ 2 wy ∂ 2 wz
⎪
⎪ +R − − −2 −2 +2 = 0,
⎪
⎪ ∂x2 ∂x∂y ∂y 2 ∂z∂y ∂x∂z ∂y∂z
⎪
⎪
⎪
⎪ ∂
⎪
⎪ μ∇2 uz + (λ + μ) ∇ · u
⎪
⎪
⎪
⎪ 2 ∂z
⎪
⎪ ∂ 2 wx ∂ 2 wx ∂ 2 wz ∂ 2 wz ∂ 2 wz
⎪
⎪
∂ wx
− − −
⎪
⎨ + R 2
2 + + 2 = 0,
∂x ∂x∂y ∂y 2 ∂x2 ∂y 2 ∂z2
⎪ ∂ 2 wx ∂ 2 wx ∂ 2 wy ∂ 2 wz ∂ 2 wz (9.1-8)
⎪
⎪ K1 ∇2 wx + K2 2 − +2 + −
⎪
⎪ ∂x∂z ∂z 2 ∂y∂z ∂x 2 ∂y 2
⎪
⎪ 2
⎪
⎪ ∂ u ∂ 2
u ∂ 2
u ∂ 2
u ∂ 2
u ∂ 2
uz ∂ 2 uz
⎪
⎪ +R
x
+2
x
−
x
−2
y
−2
y
+ − = 0,
⎪
⎪ ∂x2 ∂x∂z ∂y 2 ∂x∂y ∂y∂z ∂x2 ∂y 2
⎪
⎪
⎪
⎪ ∂ 2 wx ∂ 2 wy ∂ 2 wz ∂ 2 wy
⎪
⎪ K1 ∇2 wy + K2 2 −2 −2 −
⎪
⎪
⎪
⎪ ∂y∂z ∂x∂z ∂x∂y ∂z 2
⎪
⎪ 2 2 2 2
⎪
⎪ ∂ uy ∂ ux ∂ uy ∂ ux ∂ 2 uy ∂ 2 uz
⎪
⎪ +R −2 + +2 −2 −2 = 0,
⎪
⎪ ∂x2 ∂y∂z ∂y 2 ∂x∂y ∂x∂z ∂x∂y
⎪
⎪
⎪
⎪ ∂ 2 wz ∂ 2 wy ∂ 2 wx ∂ 2 wx
⎪
⎪ (K1 − K2 )∇2 wz + K2 2 −2 + −
⎪
⎪ 2 ∂x2 ∂y 2
⎪
⎪
∂z ∂y∂x
⎪
⎪ 2 2 2 2
⎪
⎪ ∂ uz ∂ ux ∂ uz ∂ uz
⎩ +R +2 + − = 0,
∂x2 ∂x∂z ∂y 2 ∂z 2
9.1 Basic equations of elasticity of icosahedral quasicrystals 163
where
∂2 ∂2 ∂2 ∂ux ∂uy ∂uz
∇2 = + + , ∇·u= + + .
∂x2 ∂y2 ∂z 2 ∂x ∂y ∂z
Equations (9.1-8) are 6 partial differential equations of second order on displace-
ments ui and wi . So the number of the field variables and field equations is reduced
already. But obtaining solution is still very difficult, one of reasons is the boundary
conditions for quasicrystals being much more complicated than those of the classical
theory of elasticity. In the subsequent sections we will make a great effort to solve
some complex boundary value problems through different approaches.
It is obvious that the material constants of λ, μ, K1 , K2 and R are very important
for stress analysis for different icosahedral quasicrystals, which are experimentally
measured through various methods (e.g. X-ray diffraction, neutron scattering etc)
and listed by Tables 9.1-1∼9.1-3 respectively as follows:
In this table the measurement unit of λ, μ and B is GPa, and B = (3λ + 2μ)/3
represents the bulk modulus, and ν = λ/2(λ + μ) the Poisson’s ratio, respectively.
It is needed to point out that equations (9.1-8) are not the only form of final
governing equation of elasticity of icosahedral quasicrystals, there are other forms
which will be discussed in Section 9.5.
164 Chapter 9 Theory of elasticity of three-dimensional quasicrystals and its...
Because there are only two components uz and wz , and others have vanished, the
corresponding strains are only
1 ∂uz 1 ∂uz ∂wz ∂wz
εyz = εzy = , εxz = εzx = , wzy = , wzx = . (9.2-2)
2 ∂y 2 ∂x ∂y ∂x
From the formulas listed in Section 9.1, the non-zero stress components are
⎧
⎪
⎪ σxz = σzx = 2μεxz + Rwzx ,
⎪
⎪
⎪
⎪ σ yz = σzy = 2μεyz + Rwzy ,
⎪
⎪
⎪ zx = (K1 − K2 )wzx + 2Rεxz ,
⎪ H
⎨
Hzy = (K1 − K2 )wzy + 2Rεyz ,
(9.2-3)
⎪
⎪ H xx = 2Rεxz + K2 wzx ,
⎪
⎪
⎪
⎪ Hyy = −2Rεxz − K2 wzx ,
⎪
⎪
⎪
⎪ Hxy = −2Rεyz − K2 wzy ,
⎩
Hyx = −2Rεyz − K2 wzy
9.2 Anti-plane elasticity of icosahedral quasicrystals and problem of interface... 165
∂2 ∂2
where ∇21 = + .
∂x2 ∂y 2
One can see that equations (9.2-5) are similar to (5.2-11), which can be solved
using a procedure similar to that adopted in Chapters 5, 7 and 8.
As an example of solution of anti-plane elasticity of icosahedral quasicrystals we
discuss the interface problem between centre-body cubic crystals and icosahedral
quasicrystals.
The physical model is similar to that proposed in Section 7.6, i.e., the icosahedral
quasicrystal is located in upper half-space y > 0, whose governing equations are
listed above, while the centre-body cubic crystal lies in lower space y < 0 with finite
thickness h (refer to Fig. 7.6-1) and governed by the following equation:
∇2 u(c)
z = 0, (9.2-6)
in which y > 0 and A(ξ) and B(ξ) are arbitrary functions to be determined.
166 Chapter 9 Theory of elasticity of three-dimensional quasicrystals and its...
where
μ(c)
k= , (9.2-10)
h
then the relation between the two unknown functions is obtained
R
B(ξ) = − A(ξ). (9.2-11)
K1 − K 2
and from the first one of conditions (9.2-9), one determines the unknown function
τ fˆ(ξ)
A(ξ) = − 2
(9.2-12)
R
+ μ |ξ| + k
K1 − K 2
so that
Rτ fˆ(ξ)
B(ξ) = , (9.2-13)
R2
(K1 − K2 ) + μ |ξ| + k
K1 − K 2
in which τ and k is defined by (9.2-9) and (9.2-10) respectively. Thus the problem
is solved. The phason strain field can be determined as
⎧ ∞
⎪
⎪ 1 |ξ| fˆ(ξ)
⎪ zy
⎨ w (x, y) = −Rτ e−|ξ|y−iξx dξ,
2π −∞ (R − μ(K1 − K2 )) |ξ| + k(K1 − K2 )
2
⎪ ∞
⎪
⎪ 1 ξ fˆ(ξ)
⎩ wzx (x, y) = iRτ 2
e−|ξ|y−iξx dξ.
2π −∞ (R − μ(K1 − K2 )) |ξ| + k(K1 − K2 )
(9.2-14)
Note that y > 0. The integrals in (9.2-14) can be evaluated by the residue
theorem introduced in the Major Appendix of this book.
For illustration, consider the first example. Let that f (x) = 1, as −a/2 < x <
2 a
a/2, and f (x) = 0, as x < −a/2 and x > a/2, so fˆ(ξ) = sin ξ , then from
ξ 2
solution (9.2-14), we obtain
9.2 Anti-plane elasticity of icosahedral quasicrystals and problem of interface... 167
⎧
⎪
⎪ a μ(c) R(K1 − K2 )τ 1 μ(c) (K1 − K2 ) a
⎪
⎪ w zy (x, y) = sin
⎪
⎪ h [μ(K1 − K2 ) − R2 ]2 2 μ(K1 − K2 ) − R2 h
⎪
⎪ (c)
⎪
⎪ μ(c) (K1 − K2 ) y μ (K1 − K2 ) x
⎪
⎪
⎪
⎨ × exp − cos
μ(K1 − K2 ) − R2 h μ(K1 − K2 ) − R2 h
⎪
⎪
⎪ a μ(c) R(K1 − K2 )τ 1 μ(c) (K1 − K2 ) a
⎪
⎪ w (x, y) = sin
⎪
⎪
zx
h [μ(K1 − K2 ) − R2 ]2 2 μ(K1 − K2 ) − R2 h
⎪
⎪ (c)
⎪
⎪
⎪
⎪ μ(c) (K1 − K2 ) y μ (K1 − K2 ) x
⎩ × exp − sin
μ(K1 − K2 ) − R2 h μ(K1 − K2 ) − R2 h
(9.2-15)
(c)
in which k = μ /h and the normalized expression has been used, i.e., x/h, y/h.
Then consider the second example f (x) = δ(x), then the integrals (9.2-14) will be
⎧
⎪
⎪ μ(c) R(K1 − K2 )τ
⎪
⎪ w (x, y) =
⎪
⎪
zy
[μ(K1 − K2 ) − R2 ]2
⎪
⎪
⎪
⎪ (c)
⎪
⎪ μ(c) (K1 − K2 ) y μ (K1 − K2 ) x
⎪
⎪ × −
⎪
⎨ exp sin
μ(K1 − K2 ) − R2 h μ(K1 − K2 ) − R2 h
(9.2-16)
⎪
⎪ (c)
R(K1 − K2 )τ
⎪
⎪ μ
⎪
⎪ wzx (x, y) =
⎪
⎪ [μ(K1 − K2 ) − R2 ]2
⎪
⎪
⎪
⎪ (c)
⎪
⎪ μ(c) (K1 − K2 ) y μ (K1 − K2 ) x
⎪
⎩ × exp − cos
μ(K1 − K2 ) − R2 h μ(K1 − K2 ) − R2 h
The detail of the evaluation is given in the Major Appendix of the book.
The results are quite interesting. In the first example, the phason strain field is
dominated by the elastic constants μ, K1 , K2 , R and μ(c) of quasicrystal and crystal,
applied stress τ and geometry parameters a and h, while in the second example
the geometry parameter is only h. For different τ /μ, μ(c) /μ,a/h and given values of
μ, K1 , K2 and R, one can find a rich set of numerical results. The computation used
the measured values of these quantities for Al-Pd-Mn icosahedral quasicrystals are
provided by Tables 9.1-1, 9.1-2 and 9.1-3:
The numerical results show that the influence of the ratio μ(c) /μ of shear modulus
of crystal and quasicrystal is very evident. In addition the influence of the applied
stress τ /μ is also very important. While the influence of a/h is not evident for the
first example.
Another feature of solution here is quite different from that in Section 7.6 due
to the difference of quasicrytalline systems.
This work is given in the Ref. [24].
168 Chapter 9 Theory of elasticity of three-dimensional quasicrystals and its...
∂
=0 (9.3-1)
∂z
and
R=0 (9.3-2)
The conditions (9.3-1) and (9.3-2) result in a phonon-phason decoupled plane elas-
ticity, in which there are
Based on conditions (9.3-1) and (9.3-2), the final governing equations (9.1-8)
reduce to
⎧
⎪ ∂
⎪
⎪ μ∇21 ux + (λ + μ) ∇1 · u1 = 0,
⎪
⎪ ∂x
⎪
⎪
⎪
⎪
⎪
⎪ ∂
⎪
⎪ μ∇21 uy + (λ + μ) ∇1 · u1 = 0,
⎪
⎪ ∂y
⎪
⎪
⎪
⎪ μ∇ 2
u = 0,
⎨ 1 z 2
2 ∂ wz ∂ 2 wz (9.3-4)
⎪ K1 ∇1 wx + K2 ∂x2 − ∂y 2 = 0,
⎪
⎪
⎪
⎪
⎪
⎪
⎪ ∂ 2 wz
⎪
⎪ K1 ∇21 wy − 2K2 = 0,
⎪
⎪
⎪
⎪
∂x∂y
⎪
⎪ 2
⎪
⎪ ∂ wx ∂ 2 wy ∂ 2 wy
⎪ 2
⎩ (K1 − K2 )∇1 wz + K2 −2 − = 0,
∂x2 ∂x∂y ∂y 2
where
∂2 ∂2 ∂ux ∂uy
∇21 = 2
+ 2, u1 = (ux , uy ), ∇1 · u1 = + .
∂x ∂y ∂x ∂y
Because the phonons and phasons are decoupled, the first three equations of (9.3-
4) are pure phonon equilibrium equations, in addition uz is independent of ux and
uy , and the second three equations in (9.3-4) are pure phason equilibrium equations.
Yang et al[3] solved the equations under the dislocation conditions
dui = bi , dwi = b⊥i , (9.3-5)
Γ Γ
9.4 Phonon-phason coupled plane elasticity of icosahedral quasicrystals... 169
where Γ represents a path enclosing the dislocation core. The authors used the
Green function method to calculate. The results are
⎧
⎪
⎪ b1 y λ + μ xy b2 μ r λ + μ x2
⎪
⎪ ux = arctan + + ln + ,
⎪
⎪ 2π x λ + 2μ r2 2π λ + 2μ r0 λ + 2μ r2
⎪
⎪
⎪
⎪
⎪
⎪ b μ r λ + μ y2 b2 y λ + μ xy
⎪ uy = − 1
⎪ ln + + arctan − ,
⎪
⎪ 2π λ + 2μ r0 λ + 2μ r 2 2π x λ + 2μ r2
⎪
⎪
⎪
⎪
⎪ uz = b3 arctan y ,
⎪
⎪
⎪
⎪
⎨
2π x
b⊥
1 y K 2 2xy3 xy
wx = arctan + 2 −
⎪
⎪ 2π x 2K5 r4 r2
⎪
⎪
⎪
⎪ b⊥ K 2 r 2x2 y 2 b⊥ K2 xy
⎪
⎪ − 2 2 ln + 3
⎪
⎪
+ 4 2π K1 r2
,
⎪
⎪ 4π K5 r0 r
⎪
⎪ ⊥ ⊥
⎪
⎪ b⊥ K 2 r 2x2 y 2 b2 y K22 xy 3 − x3 y b3 K2 y 2
⎪ wy = 1 2 ln −
⎪ + arctan + − ,
⎪
⎪ 4π K5 r0 r4 2π x 2K5 r4 2π K1 r2
⎪
⎪
⎪
⎪ b⊥ K xy b⊥ 2
b⊥
⎪
⎩ wz = 1 K1 2 − 2 K1 K2 y
+ 3
arctan ,
y
2π K5 r2 2π K5 r2 2π x
(9.3-6)
in which
K5 = K12 − K1 K2 − K22 . (9.3-7)
The first three of (9.3-6) are well known solution of pure phonon field in the
classical theory of dislocation, and the second three of (9.3-6) are new results for
pure phason field. Because Yang et al[3] ignored the coupling terms, the interaction
between phonons and phasons could not be revealed.
R = 0
and obtained the complete theory for the plane elasticity of quasicrystals. In the
study the assumption (9.2-1) or (9.3-1) still maintains, i.e.,
∂
= 0. (9.4-1)
∂z
In this case the three-dimensional elasticity can be reduced into a plane elasticity
problem. From condition (9.3-1) directly we have
170 Chapter 9 Theory of elasticity of three-dimensional quasicrystals and its...
Thus the number of the field variables and field equations are reduced from 36 to
32. Though the reduction of the total number is not so much, the resulting equation
system has been greatly simplified and with the following form:
⎧ 2
⎪ ∂ ∂ wx ∂ 2 wy ∂ 2 wy
⎪
⎪ μ∇ 2
u + (λ + μ) ∇ 1 · u 1 + R + 2 − = 0,
⎪
⎪
1 x
∂x ∂x2 ∂x∂y ∂y 2
⎪
⎪
⎪
⎪ ∂ ∂ 2 wy ∂ 2 wx ∂ 2 wy
⎪
⎪ μ∇21 uy + (λ + μ) ∇1 · u1 + R − −
⎪
⎪ 2 = 0,
⎪
⎪ ∂y ∂x2 ∂x∂y ∂y 2
⎪
⎪
⎪
⎪ ∂ 2 wx ∂ 2 wy ∂ 2 wx
⎪ 2
⎨ μ∇1 uz + R − 2 − + ∇21 wz = 0,
∂x2 ∂x∂y ∂y 2
2 2
⎪
⎪ ∂ wz ∂ 2 wz ∂ ux ∂ 2 uy ∂ 2 ux ∂ 2 uz ∂ 2 uz
⎪
⎪ K ∇ 2
1 1 x w +K 2 − +R −2 − + − = 0,
⎪
⎪ ∂x2 ∂y 2 ∂x2 ∂x∂y ∂y 2 ∂x2 ∂y 2
⎪
⎪
⎪
⎪ ∂ 2 wz ∂ 2 uy ∂ 2 ux ∂ 2 uy ∂ 2 uz
⎪
⎪ ∇ 2
− − −
⎪
⎪ K 1 1 w y 2K 2 + R + 2 2 = 0,
⎪
⎪ ∂x∂y ∂x2 ∂x∂y ∂y 2 ∂x∂y
⎪
⎪
⎪
⎪ ∂ 2 wx ∂ 2 wy ∂ 2 wy
⎩ (K1 − K2 )∇21 wz + K2 − 2 − + R∇21 uz = 0,
∂x2 ∂x∂y ∂y 2
(9.4-3)
where ∇21 and ∇1 · u1 are the same of those in Section 9.3, but the suffix 1 of the
two-dimensional Laplace operator will be omitted in the following for simplicity.
The equation set is much simpler than that of (9.1-8) but still quite complicated.
If we introduce a displacement potential F (x, y) such as
⎧
⎪ ∂2
⎪
⎪ u = R ∇2 ∇2 [μαΠ1 + β(λ + 2μ)Π2 ]F
⎪
⎪
x
∂x∂y
⎪
⎪
⎪
⎪ ∂2 ∂4 ∂4 ∂4
⎪
⎪ − −
⎪
⎪ +c 0 R Λ (3μ λ)
∂x4
+ 10(λ + μ)
∂x2 ∂y 2
(5λ + 9μ)
∂y 4
F
⎪
⎪ ∂x∂y
⎪
⎪
⎪
⎪ ∂2 ∂2
⎪
⎪ u = R∇ 2 2
∇ μα Π1 − β(λ + 2μ) Π 2 F
⎪
⎪
y
∂y 2 ∂x2
⎪
⎪
⎪
⎪ ∂6 ∂6 ∂6 ∂6
⎪
⎪ 2
+c0 RΛ (λ + 2μ) 6 − 5(2λ + 3μ) 4 2 + 5λ 2 4 + μ 6 F,
⎨ ∂x ∂x ∂y ∂x ∂y ∂y
2
2 2
⎪
⎪ ∂ ∂ ∂
⎪
⎪ uz = c1 (α − β)Λ2 Π1 Π2 + α 2 Π12 + β 2 Π22 F,
⎪
⎪ ∂x∂y ∂y ∂x
⎪
⎪
⎪
⎪ ∂2
⎪
⎪ −ω ∇2 2c0 Λ2 ∇2 − (α − β)Π1 Π2 F
⎪
⎪ w x =
⎪
⎪ ∂x∂y
⎪
⎪
⎪
⎪ ∂2 2 ∂2 2
⎪
⎪
2 2 2 2
wy = −ω∇ c0 Λ Λ ∇ + α 2 Π1 + β 2 Π2 F,
⎪
⎪ ∂y ∂x
⎪
⎪
⎪
⎪ ∂ 2
∂ 2
∂2
⎪
⎩ wz = c2 (α − β)Λ2 Π1 Π2 + α 2 Π12 + β 2 Π22 F,
∂x∂y ∂y ∂x
(9.4-4)
9.4 Phonon-phason coupled plane elasticity of icosahedral quasicrystals... 171
where
⎧
⎪
⎪ α = (λ + 2μ)R2 − ωK1 , β = μR2 − ωK1 , ω = μ(λ + 2μ),
⎪
⎪
⎪
⎪
⎨ μK22 +(K1 −3K2 )R2 (K1 −2K2 )Rω (K2 μ−R2 )ω
c0 = ω , c1 = , c2 = ,
μ(K1 −K2 )−R 2 μ(K1 −K2 )−R 2 μ(K1 −K2 )−R2
⎪
⎪
⎪
⎪
⎪
⎪ ∂2 ∂2 ∂2 ∂2 ∂2 ∂2 ∂2 ∂2
⎩ Π1 = 3 2 − 2 , Π2 = 3 2 − 2 , ∇2 = 2 + 2 , Λ2 = 2 − 2 ,
∂x ∂y ∂y ∂x ∂x ∂y ∂x ∂y
(9.4-6)
in which the suffix 1 of the two-dimensional Laplace operator is omitted and
c0 ∂ 10 α ∂ 10 α ∂ 10
L= − 10 + 5 4 − 5 8 2
− 10 11 − 10
β ∂x β ∂x ∂y β ∂x6 ∂y4
α ∂ 10 α ∂ 10 α ∂ 10
+ 10 10 − 11 − 5 5 − 4 − . (9.4-7)
β ∂x4 ∂y6 β ∂x2 ∂y8 β ∂y 10
Assuming
R2
1 (9.4-8)
μK1
(this is understandable, because the coupling effect is weaker than that of phonon),
then from equations (9.4-6) and (9.4-7),
β c0 2 2 2 2 2 2
→ 1, ∇2 L = ∇ ∇ ∇ ∇ ∇ ∇ , (9.4-9)
α β
substituting (9.4-9) into (9.4-5), we find that
∇2 ∇2 ∇2 ∇2 ∇2 ∇2 F (x, y) = 0. (9.4-10)
This sextuple harmonic equation is the governing equation for plane elasticity
of quasicrystals based on the displacement potential formulation. With the aid of
the generalized Hooke’s law, the phonon and phason stress components can also be
expressed in terms of potential function F (x, y) and these expressions are omitted
here due to the limitation of the space.
In other words, equation set (9.4-4) gives a fundamental solution in terms of
F (x, y) for the plane elasticity problem of an icosahedral quasicrystal. Once the
function F (x, y) satisfying equation (9.4-4) is determined for prescribed boundary
conditions, the entire elastic field of an icosahedral quasicrystal can be found from
(9.4-4). This formulation has been reported briefly by Fan and Guo[4] . In some
extent it is a development of Li and Fan[18] for elasticity of two-dimensional qua-
sicrystals. The application of the formulation and relevant solution will be given in
Section 6.
172 Chapter 9 Theory of elasticity of three-dimensional quasicrystals and its...
such as
⎧
⎪ ∂ 2 ϕ1 ∂ 2 ϕ1 ∂ 2 ϕ1
⎪
⎪ σxx = , σ = − , σ = ,
⎪
⎪ ∂y 2
xy
∂x∂y
yy
∂x2
⎪
⎪
⎪
⎪
⎪
⎪ ∂ϕ2 ∂ϕ2
⎪
⎨ σzx = ∂y , σzy = − ∂x ,
⎪
(9.5-2)
⎪
⎪ ∂ψ1 ∂ψ1 ∂ψ2
⎪
⎪
⎪ Hxx = ∂y , Hxy = − ∂x , Hyx = ∂y ,
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎩ Hyy = − ∂ψ2 , Hzx = ∂ψ3 , Hzy = − ∂ψ3 ,
∂x ∂y ∂x
with
9.6 A straight dislocation in an icosahedral quasicrystal 173
⎧ 2
⎪
⎪ ∂ ∂ 2
⎪
⎪ ϕ 1 = c 2 c3 R 2 Π 2 − Λ Π 1 ∇2 ∇2 G,
⎪
⎪ ∂y ∂x2
⎪
⎪
⎪
⎪ ϕ2 = −c3 c4 ∇2 ∇2 ∇2 ∇2 ∇2 G
⎪
⎪
⎪
⎪ 2
⎪
⎨ ∂2 ∂
ψ1 = c1 c2 R 2 2 2 Π1 Π2 − Λ Π1 ∇2 G + c2 c4 Λ2 ∇2 ∇2 ∇2 ∇2 G,
2 2
⎪ ∂y ∂x
⎪
⎪ 2
⎪
⎪ 2
∂2
⎪
⎪
∂ ∂
Π 2
− 2
Π 2
∇2 ∇2 ∇2 ∇2 G,
⎪ 2
⎪ ψ = c1 2c R 2 2 2 Λ 1 2 ∇ G − 2c2 c4
Π
⎪
⎪ ∂x∂y ∂x ∂x∂y
⎪
⎪
⎪
⎪ 1
⎩ ψ3 = − K2 c3 c4 ∇2 ∇2 ∇2 ∇2 ∇2 G,
R
(9.5-3)
then the equilibrium equations and the deformation compatibility equations will be
identically satisfied if
∇2 ∇2 ∇2 ∇2 ∇2 ∇2 G = 0 (9.5-4)
under the approximation R2 /K1 μ 1, which is the final governing equation of plane
elasticity of icosahedral quasicrystals, function G(x, y) is named the stress potential
in which ⎧
⎪ R(2K2 − K1 )(μK1 + μK2 − 3R2 )
⎪
⎪ c 1 = ,
⎪
⎪ 2(μK1 − 2R2 )
⎪
⎪
⎪
⎪
⎪
⎪ 1
⎪
⎪ c2 = K2 (μK2 − R2 ) − R(2K2 − K1 ),
⎪
⎪ R
⎪
⎪
⎪
⎪ (μK2 − R2 )2
⎪
⎪ 2
⎪
⎨ c 3 = μ(K 1 − K 2 ) − R − ,
μK1 − 2R2
(9.5-5)
⎪
⎪ 1 μK 1 − 2R 2
⎪
⎪ c4 = c1 R + c3 K1 + ,
⎪
⎪ 2 λ+μ
⎪
⎪
⎪
⎪
⎪
⎪ ∂2 ∂2 ∂2 ∂2
⎪
⎪ Π1 = 3 2 − 2 , Π2 = 3 2 − 2 ,
⎪
⎪ ∂x ∂y ∂y ∂x
⎪
⎪
⎪
⎪
⎪
⎪
2 2 2 2
⎩ ∇2 = ∂ + ∂ , Λ2 = ∂ − ∂
∂x2 ∂y2 ∂x2 ∂y2
In derivation of (9.5-4) the approximation (9.4-8) is used at the last step.
This work is given by Ref. [5], which may be seen as a development of the study
for two-dimensional quasicrystals given by Guo and Fan (see e.g. Fan[19] or Guo and
Fan[20] ).
In the following we use the formulation of Section 9.4 to solve the above boundary
value problem. Performing the Fourier transform to equation (9.4-10) and the above
boundary conditions, we obtain the solution at the transformed domain, then taking
9.6 A straight dislocation in an icosahedral quasicrystal 175
in which r 2 = x2 + y2 , r0 is the radius of the dislocation core and cij are constants
shown as follows:
⎧
⎪ 2 2 2 2 ⊥
⎪
⎪ c12 = 2c0 (μ(2R +c0 μ)(λ +3λμ+2μ )b1 +R(−e(λ+μ)+2μc0 (λ+2μ) )b1 ) ,
⎪
⎪
⎪
⎪ −e(2e+μc0 (λ+2μ))+μc0 (λ+2μ)(e+2μc0 (λ + 2μ))
⎪
⎪
⎪
⎪
⎪
⎪
2c0 R(λ + μ)(2Rμ(λ + μ)b1 + 2μc0 (λ + 2μ)b⊥
⎪
⎪ 1)
⎪
⎪
c13 =
−e(2e
,
⎪
⎪ + μc 0 (λ + 2μ)) + μc0 (λ + 2μ)(e + 2μc0 (λ + 2μ))
⎪
⎪
⎪
⎪
⎪
⎪ (2c20 μ3 (λ + 2μ) − 2e2 )b1 + 2c0 R(λ + 3μ)eb⊥
⎪
⎪ c = 1
,
⎪
⎪ 21
−e(2e + μc (λ + 2μ)) + μc (λ + 2μ)(e + 2μc (λ + 2μ))
⎪
⎪ 0 0 0
⎪
⎪
⎪
⎪
⎪
⎪ 2c0 (−μ2 (λ + μ)(−2R2 + c0 (λ + 2μ))b1 + R(−(λ + μ)e + 2c0 μ2 )b⊥ 1)
⎪
⎪ c 22 = ,
⎪
⎪ −e(2e + μc0 (λ + 2μ)) + μc0 (λ + 2μ)(e + 2μc0 (λ + 2μ))
⎪
⎨
⎪ 2c0 R(λ + μ)(2Rμ(λ + μ)b1 + 2c0 μ2 b⊥ 1)
⎪
⎪ c23 = ,
⎪
⎪ −e(2e + μc 0 (λ + 2μ)) + μc0 (λ + 2μ)(e + 2μc0 (λ + 2μ))
⎪
⎪
⎪
⎪
⎪
⎪
3c1 e{2(c0 μ + 7e)μc0 (λ + 2μ)b1 + R(54c20 (λ2 + 3λμ + μ2 )
⎪
⎪
⎪
⎪
⎪
⎪ −2(α − β)(e + μc0 (λ + 2μ)))b⊥
⎪
⎪ 1}
⎪
⎪ c 31 = ,
⎪
⎪ 4c0 R(−e(2e + μc 0 (λ + 2μ)) + μc 0 (λ + 2μ)(e + 2μc0 (λ + 2μ))
⎪
⎪
⎪
⎪
⎪
⎪
3c1 e(2μ(−e + μc0 (λ + 2μ))b1 + R(−2e + 2μc0 (λ + 2μ))b⊥
⎪
⎪ 1)
⎪ 32
⎪ c =
−e(2e
,
⎪
⎪ + μc 0 (λ + 2μ)) + μc 0 (λ + 2μ)(e + 2μc 0 (λ + 2μ))
⎪
⎪
⎪
⎪
⎪
⎪ −3ec1 (2Rμ(λ + μ)b1 + 2μc0 (λ + 2μ)b⊥ 1)
⎪
⎩ c33 = ,
−e(2e + μc0 (λ + 2μ)) + μc0 (λ + 2μ)(e + 2μc0 (λ + 2μ))
176 Chapter 9 Theory of elasticity of three-dimensional quasicrystals and its...
⎧
⎪
⎪ −2e(2Rμ(λ + μ)b1 + 2μc0 (λ + 2μ)b⊥ 1)
⎪
⎪ c42 = ,
⎪
⎪ −e(2e + μc0 (λ + 2μ)) + μc0 (λ + 2μ)(e + 2μc0 (λ + 2μ))
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
c43 = 0,
⎪
⎪
⎪
⎪
⎪
⎪ {−4eμ2 c0 (λ + 2μ)b1 + R(2(λ + 2μ)(e + 0.5μc0 )
⎪
⎪
⎪
⎪
⎪
⎪ +μ(2β 2 μ + 2c20 (λ + 2μ)2 + c0 (λ + 2μ)(−βμ + R2 (λ + μ)))b⊥
1}
⎪
⎪ c51 =− ,
⎪
⎪ R(−e(2e + μc0 (λ + 2μ)) + μc0 (λ + 2μ)(e + 2μc0 (λ + 2μ)))
⎪
⎪
⎪
⎪
⎪
⎪
2e(2Rμ(λ + μ)b1 + 2μc0 (λ + 2μ)b⊥
⎪
⎪ 1)
⎪
⎨ c52 =− ,
−e(2e + μc0 (λ + 2μ)) + μc0 (λ + 2μ)(e + 2μc0 (λ + 2μ))
⎪
⎪ c53 =0
⎪
⎪
⎪
⎪
⎪
⎪
3c2 e{(2(c0 μ + 7e)μc0 (λ + 2μ)b1 + R(54c20 (λ2 + 3λμ + μ2 )
⎪
⎪
⎪
⎪
⎪
⎪ −2(α − β)(e + μc0 (λ + 2μ)))b⊥
⎪
⎪ 1 )}
⎪
⎪ c61 =− ,
⎪
⎪ 4c0 R(−e(2e + μc0 (λ + 2μ))(e + 2μc0 (λ + 2μ)))
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ 3ec2 (2μ(−e + μc0 (λ + 2μ))b1 + R(−2e + 2μc0 (λ + 2μ))b⊥1)
⎪
⎪ c62 = ,
⎪
⎪ −e(2e + μc0 (λ + 2μ)) + μc0 (λ + 2μ)(e + 2μc0 (λ + 2μ))
⎪
⎪
⎪
⎪
⎪
⎪ −3ec2 (2Rμ(λ + μ)b1 + 2μc0 (λ + 2μ)b⊥1)
⎪
⎩ c63 = .
−e(2e + μc0 (λ + 2μ)) + μc0 (λ + 2μ)(e + 2μc0 (λ + 2μ))
(9.6-5)
2
with e = −(λ + μ)R .
For the other two typical problems, in which the Burgers vector of the dislocation
is denoted by (0, b2 , 0, 0, b⊥ ⊥
2 , 0) and (0, 0, b3 , 0, 0, b3 ) respectively, a complete similar
consideration will yield similar results, which are omitted here. Alternatively, the
(2) (2) (3) (3)
expressions are denoted as uj , wj , uj , wj .
Analytic expressions for elastic field for a dislocation (b1 , b2 , b3 , b⊥ ⊥ ⊥
1 , b2 , b3 ) in the
icosahedral quasicrystal can be obtained by superposition of the corresponding ex-
pressions for the elastic fields for (b1 , 0, 0, b⊥ ⊥
1 , 0, 0), (0, b2 , 0, 0, b2 , 0) and (0, 0, b3 , 0, 0,
b⊥
3 ), namely,
We can see that the interaction among phonon-phonon, phason-phason and phonon-
phason is very evident, so the solution (9.6-4) is quite different from the solution
given by Yang et al[3] (whose solution for phonon displacement field is given by the
first three formulas of equations (9.3-6), and will be quoted again in the following, see
formula (9.6-7)), where they took R = 0, i.e., they assumed the phonon and phason
are decoupled, so the solution for phonon is the same as the classical solution for
crystals. It is obvious that our solution given by (9.6-4) explores that the realistic
9.6 A straight dislocation in an icosahedral quasicrystal 177
case for quasicrystals is quite different from that of crystal. To illustrate the coupling
effect we give some numerical results in Fig. 9.6-1 and Fig. 9.6-2 for the normalized
Fig. 9.6-1 The displacement u1 /b1 versus x for different coupling elastic constants
Fig. 9.6-2 The displacement u1 /b1 versus y for different coupling constants
displacement u1 /b1 versus x and y respectively, in which the results exhibit the
influence of the coupling constant R; the coupling effect is quite remarkable. In the
calculation we take the data of elastic moduli as
and the phonon-phason coupling elastic constant for three different cases: i.e., R/μ =
0, R/μ = 0.04 and R/μ = 0.06, in which the first one corresponds to decoupled case.
178 Chapter 9 Theory of elasticity of three-dimensional quasicrystals and its...
The figures show that the coupling effect is very important, and the displacement
is increasing with the growth of value of R.
For icosahedral quasicrystals with the presence of a dislocation, there are five
independent elastic constants. If R = 0, wi = 0, our solution is exactly reduced to
the solution of dislocation of crystals, i.e.,
⎧
⎪
⎪ b1 y λ + μ xy b2 μ r λ + μ x2
⎪
⎪ u = arctan + + ln + ,
⎪
⎪
x
2π x λ + 2μ r2 2π λ + 2μ r0 λ + 2μ r2
⎪
⎪
⎪
⎨
b1 μ r λ + μ x2 b2 y λ + μ xy
⎪ uy = − ln + + arctan − ,
⎪
⎪ 2π λ + 2μ r0 λ + 2μ r2 2π x λ + 2μ r2
⎪
⎪
⎪
⎪
⎪
⎪
⎩ u = b3 arctan y .
z
2π x
(9.6-7)
The displacement potential function formulation establishes the basis for solving
defects problem in icosahedral quasicrystals. The formulation is simplified the so-
lution process. In the subsequent steps a systematic Fourier analysis is developed,
which provides a constructive procedure to find the analytic solution, it is effective
not only for dislocation problem, but also for more complicated mixed boundary
value problems (e.g. crack problems refer to Ref. [25]).
The present solution can be used as a fundamental solution for a dislocation in
an icosahedral quasicrystal. Therefore, many elasticity problems in an icosahedral
quasicrystal can be directly solved with the aid of this fundamental solution by
superposition.
This work has been published in Ref. [17].
G(x, y) = Re[g1 (z) + z̄g2 (z) + z̄ 2 g3 (z) + z̄ 3 g4 (z) + z̄ 4 g5 (z) + z̄ 5 g6 (z)], (9.7-1)
where gi (z) are arbitrary analytic functions of z = x + iy, and the bar over complex
variable or complex function denotes the complex conjugate.
By equations (9.5-2)∼(9.5-4) and (9.7-1), the stresses can be expressed as follows:
⎧
⎪
⎪ σxx + σyy = 48c2 c3 RImΓ (z),
⎪
⎪
⎪
⎪
⎪
⎪ σyy − σxx + 2iσxy = 8ic2 c3 R(12Ψ (z) − Ω (z)),
⎪
⎪
⎪
⎪
⎪
⎪ σzy − iσzx = −960c3 c4 f6 (z),
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ 24λR
⎪
⎪ σzz = c2 c3 ImΓ (z),
⎪
⎪ (μ + λ)
⎪
⎨
Hxy − Hyx − i(Hxx + Hyy ) = −96c2 c5 Ψ (z) − 8c1 c2 RΩ (z), (9.7-2)
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ Hyx − Hxy + i(Hxx − Hyy ) = −480c2 c5 f6 (z) − 4c1 c2 RΘ (z),
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ Hyz + iHxz = 48c2 c6 Γ (z) − 4c2 R2 (2K2 − K1 )Ω (z),
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ 24R2
⎪
⎩ Hzz = c2 c3 ImΓ (z),
(μ + λ)
where
⎧
⎪
⎪ R(2K2 − K1 )(μK1 + μK2 − 3R2 )
⎪
⎪ c 1 = ,
⎪
⎪ 2(μK1 − 2R2 )
⎪
⎪
⎪
⎪
⎪
⎪ 1
⎪
⎪ c = K (μK2 − R2 ) − R(2K2 − K1 ),
⎨ 2 R 2
⎪
⎪ (μK2 − R2 )2
⎪
⎪ c3 = μ(K1 − K2 ) − R2 − ,
⎪
⎪ μK1 − 2R2
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ 1 μK1 − 2R2
⎩ c4 = c1 R + c3 K1 + ,
2 λ+μ
180 Chapter 9 Theory of elasticity of three-dimensional quasicrystals and its...
and
⎧
⎪
⎪ Ψ (z) = f5 (z) + 5z̄f6 (z),
⎪
⎪
⎪
⎪ Γ (z) = f4 (z) + 4z̄f5 (z) + 10z̄ 2 f6 (z),
⎪
⎪
⎪
⎪
⎨ Ω (z) = f (z) + 3z̄f (z) + 6z̄ 2 f (z) + 10z̄ 3 f (z),
3 4 5 6
(9.7-3)
⎪
⎪ Θ(z) = f2 (z) + 2z̄f (z) + 3z̄ 2 f (z) + 4z̄ 3 f (z) + 5z̄ 4 f (IV ) (z),
⎪
⎪ 3 4 5 6
⎪
⎪
⎪
⎪ 2
⎪ c5 = 2c4 − c1 R, c6 = (2K2 − K1 )R2 − 4c4 μK2 − R .
⎪
⎩
μK1 − 2R2
In the above expressions, the function g1 (z) does not appear, this implies that for
stress boundary value problem in this formalism only five complex potentials g2 (z),
g3 (z), g4 (z), g5 (z) and g6 (z) are needed, and one can take g1 (z) = 0. For simplicity,
we have introduced the following new symbols:
(9) (8) (7)
g2 (z) = f2 (z), g3 (z) = f3 (z), g4 (z) = f4 (z),
(9.7-4)
(6) (5)
g5 (z) = f5 (z), g6 (z) = f6 (z),
(n)
where gi denotes the n-th derivative with the argument z, accordingly f1 (z) = 0.
Similar to formulation in Chapter 8, the complex representations of displacement
components can be written as follows (here we have omitted the rigid-body displace-
ments)
⎧
⎪
⎪ uy + iux = −6c2 R
2c3
+ c7 − 2c2 c7 RΩ (z),
⎪
⎪
⎪
⎪ μ+λ
⎪
⎪
⎪
⎪ 4
⎪
⎪
⎪ uz = μ(K + K ) − 3R2 (240c10 Imf6 (z))
⎪
⎪
⎪ 1 2
⎪
⎪
⎨ 2
+ c1 c2 R Im(Θ(z) − 2Ω (z) + 6Γ (z) − 24Ψ (z))),
(9.7-5)
⎪
⎪ R
⎪
⎪ w + iw = − (24c 9 Ψ (z) − c8 Θ(z)),
⎪
⎪
y x
c1 (μK1 − 2R2 )
⎪
⎪
⎪
⎪
⎪
⎪ w = 4(μK2 − R2 )
⎪
⎪ (240c10 Imf6 (z))
⎪
⎪
z
(K1 − 2K2 )R(μ(K1 + K2 ) − 3R2 )
⎪
⎪
⎩
+ c1 c2 R2 Im(Θ(z) − 2Ω (z) + 6Γ (z) − 24Ψ (z))),
in which
⎧ c3 K1 + 2c1 R
⎪
⎪ c7 = ,
⎪
⎪ μK1 − 2R2
⎪
⎪
⎪
⎪
⎨ c8 = c1 c2 R(μ(K1 − K2 ) − R2 ),
(9.7-6)
⎪
⎪ (μK2 − R2 )2
⎪
⎪ c = c + 2c c c − ,
⎪
⎪
9 8 2 4 3
μK1 − 2R2
⎪
⎪
⎩
c10 = c1 c2 R2 − c4 (c2 R − c3 K1 ).
9.7 An elliptic notch/Griffith crack in an icosahedral quasicrystal 181
hx = 0, hy = 0. (9.7-8 )
Multiplying the second formula of (9.1-12) by -i and adding it to the first, we obtain
f6 (z) + f6 (z) = 0,
4c11 Re[f5 (z) + 5z̄f6 (z)] + (2K2 − K1 )RRe[f4 (z) + 4z̄f5 (z)
+ 10z̄ 2 f6 (z) + 20f6 (z)] = 0, (9.7-14)
in which
4c4 (μK2 − R2 )
c11 = (2K2 − K1 )R − . (9.7-15)
(μK1 − 2R2 )R
However further calculation will be very difficult on the z-plane owing to the com-
plexity of the manipulation, we must employ conformal mapping
1
z = ω(ζ) = R0 + mζ (9.7-16)
ζ
to transform the region with the ellipse at the z-plane onto the interior of the unit
circle γ at ζ-plane, in which
a+b a−b
R0 = , m= .
2 a+b
Let
fi (z) = fi [ω(ζ)] = Φi (ζ), i = 2, 3, · · · 6. (9.7-17)
Substituting (9.7-16) into (9.7-10), (9.7-11), (9.7-13) and (9.7-14), then multiplying
both sides of equations by dσ/[2πi(σ − ζ)](σ represents the value of ζ at the unit
9.7 An elliptic notch/Griffith crack in an icosahedral quasicrystal 183
circle) and integrating around the unit circle γ, by means of Cauchy’s integral for-
mula and analytic extension of the complex variable function theory, we obtain (see
the Major Appendix of this book for details)
⎧
⎪ R0 ipζ(ζ 2 + m)(m3 ζ 2 + 1) (2K2 − K1 )R0
⎪
⎪ Φ2 (ζ) = + ×
⎪
⎪ 2c2 c3 R (mζ 2 − 1)3 2c2 c3 c11
⎪
⎪
⎪
⎪
⎪
⎪ pmζ 3 (ζ 2 + m)[m2 ζ 6 − (m3 + 4m)ζ 4 + (2m4 + 4m2 + 5)ζ 2 + m]
⎪
⎪ ,
⎪
⎪ (mζ 2 − 1)5
⎪
⎪
⎨
R0 ipζ(m2 + 1) (2K2 − K1 )R0 pmζ 3 (ζ 2 + m)(mζ 2 − m2 − 2)
Φ3 (ζ) = − ,
⎪
⎪ 4c2 c3 R (mζ 2 − 1) 12c2 c3 c11 (mζ 2 − 1)3
⎪
⎪
⎪
⎪
⎪
⎪ R0 (2K2 − K1 )R0 pmζ(ζ 2 + m)
⎪
⎪ Φ4 (ζ) = − ipmζ − ,
⎪
⎪ 12c2 c3 R 2c2 c3 c11 (mζ 2 − 1)
⎪
⎪
⎪
⎪
⎩ Φ5 (ζ) = − (2K2 − K1 )R0 pmζ, Φ6 (ζ) = 0.
⎪
48c2 c3 c11
(9.7-18)
The elliptic notch problem has been solved.
The solution of the Griffith crack subjected to a uniform pressure can be obtained
if put m = 1, R0 = a/2 in the above notch solution. The solution of crack can be
expressed explicitly in the z-plane, for example,
$ %
z ia2 y 3(2K2 − K1 )R ipa2 y
σyy =Im ip √ +# −1 + #
z 2 − a2 (z 2 − a2 )3 2c11 (z 2 − a2 )3
(2K2 − K1 )R ipy(2a4 − 3z z̄) (2K2 − K1 )R a2 pz(z z̄ − a2 )
+ # − #
2c11 (z 2 − a2 )5 4c11 (z 2 − a2 )5
(2K2 − K1 )R a2 pz̄
+ # , (9.7-19)
4c11 (z 2 − a2 )3
#
2c3 ip
uy = − 6c2 R + c7 Re (z − z 2 − a2 )
μ+λ 24c2 c3 R
#
2K2 − K1 z z̄ a2
+ p √ −√ − z 2 − a2
24c2 c3 c11 z 2 − a2 z 2 − a2
ip z z̄ a2 2K2 − K1
− 2c2 c7 Re √ −√ − z̄ − ipy
8c2 c3 R z 2 − a2 z 2 − a2 4c2 c3 c11
$ %
2K2 − K1 a2 [(z z̄ − a2 ) + 2iyz̄]
+ p #
16c2 c3 c11 (z 2 − a2 )3
#
2K2 − K1 a2 2z z̄
+ p √ −√ +2 z −a2 2 . (9.7-20)
16c2 c3 c11 z 2 − a2 z 2 − a2
From equations (9.7-19) and (9.7-20), the stress intensity factor and energy release
rate can be evaluated as follows:
184 Chapter 9 Theory of elasticity of three-dimensional quasicrystals and its...
√
K1 = πap,
a
1 ∂
GI = 2 (σyy (x, 0) ⊕ H(x, 0))(uy (x, 0) ⊕ wy (x, 0))dx
2 ∂a −a
1 1 c7
= + (KI )2 , (9.7-21)
2 λ + μ c3
Fig. 9.7-2 Influence of phason and phonon-phason coupling to the crack opening
displacement
9.8 Elasticity of cubic quasicrystals—the anti-plane and axisymmetric deformation 185
Fig. 9.7-3 Influence of phason and phonon-phason coupling to the energy release rate
This work developed the previous work for elasticity of two dimensional qua-
sicrystals of Fan and co-workers. The work is helpful to understand quantitatively
the influence of elliptic notch and crack on the mechanical behaviour of icosahedral
quasicrystals. The stress intensity factor and energy release rate are also obtained
as the direct results of the solution, which are useful for fracture mechanics.
The rigorous theory on the complex potential method will be summarized in
Chapter 11, which can also be referred to Ref [26].
∇2 u3 = 0, ∇2 w3 = 0.
The solution can be derived from the relevant discussion in Chapters 5,7,8 and
Section 9.2, so it need not be mentioned again.
For the axisymmetric case Zhou and Fan[23] developed a displacement potential
theory to reduce basic equations to a single partial differential equation with higher
order in circular cylindrical coordinate system (r, θ, z), i.e., we assume
∂
=0 (9.8-1)
∂θ
and by the generalized Hooke’s law,
⎧
⎪ σrr = C11 εrr + C12 (εθθ + εzz ) + R11 wrr + R12 (wθθ + wzz ),
⎪
⎪
⎪
⎪
⎪
⎪ σθθ = C11 εθθ + C12 (εrr + εzz ) + R11 wθθ + R12 (wrr + wzz ),
⎪
⎪
⎪
⎪
⎪
⎪ σzz = C11 εzz + C12 (εrr + εθθ ) + R11 wzz + R12 (wθθ + wrr ),
⎪
⎪
⎪
⎨ σzr = σrz = 2C44 εrz + 2R44 wrz ,
(9.8-2)
⎪
⎪ Hzz = R11 εzz + R12 (εθθ + εrr ) + K11 wzz + K12 (wθθ + wrr ),
⎪
⎪
⎪
⎪
⎪
⎪ Hrr = R11 εrr + R12 (εθθ + εzz ) + K11 wrr + K12 (wθθ + wzz ),
⎪
⎪
⎪
⎪
⎪
⎪ Hθθ = R11 εθθ + R12 (εrr + εzz ) + K11 wθθ + K12 (wrr + wzz ),
⎪
⎪
⎩
Hzr = Hrz = 2R44 εrz + 2K44 wrz ,
which are ⎧
⎪
⎪
∂ur ur ∂uz
⎪
⎪ εrr = , εθθ = , εzz = ,
⎪
⎪ ∂r r ∂z
⎪
⎪
⎪
⎪ 1 ∂ur ∂uz
⎪
⎨ εrz = εzr = + ,
2 ∂z ∂r
(9.8-3)
⎪
⎪ ∂wr wr ∂wz
⎪
⎪ wrr = , wθθ = , wzz = ,
⎪
⎪ ∂r r ∂z
⎪
⎪
⎪
⎪ 1 ∂wr ∂wz
⎪
⎩ wrz = wzr = +
2 ∂z ∂r
If all the displacements and stresses can be expressed by a potential F (r, z) (the
detail on definition of F (r, z) is given by [23]), which satisfies
2 6 2 2 4
∂8 ∂ 1 ∂ ∂ ∂ 1 ∂ ∂
8
−b 2
+ 6
+c 2
+
∂z ∂r r ∂r ∂z ∂r r ∂r ∂z 4
2 3 2 2 4
∂ 1 ∂ ∂ ∂ 1 ∂
−d + + e + F = 0, (9.8-5)
∂r2 r ∂r ∂z 2 ∂r2 r ∂r
⎧ √
⎪
⎪ r2 + z 2 → ∞ : σzz = p0 Hzz = 0, σrz = 0, Hrz = 0,
⎨
z = 0, 0 r a, σzz = σrz = 0; Hzz = Hrz = 0, (9.8-6)
⎪
⎪
⎩
z = 0, r > a : σrz = 0, uz = 0; Hrz = 0, wz = 0.
⎧ ∞
⎪
⎪ ξAi (ξ)J0 (ξr)dξ = Mi p0 , 0 < r < a,
⎨
0
∞ (9.8-8)
⎪
⎪
⎩ Ai (ξ)J0 (ξr)dξ = 0, r > a
0
in which J3/2 (aξ) is the first kind Bessel function of 3/2-order (refer to the Major
Appendix for the detailed calculation).
After some calculation, the stress intensity factor KI , strain energy WI and strain
energy release rate GI can be obtained as follows:
2√ 1 ∂WI 3M p2 a
KI = πap, WI = M p2 a3 , GI = = , (9.8-10)
π 2πa ∂a 2π
where M is the constant composed of the elastic constants which is quite lengthy so
has not been included here.
References
[1] Ding D H, Yang W G, Hu C Z et al. Generalized theory of elasticity of quasicrystals.
Phys Rev B, 1993, 48(10): 7003–7010
[2] Hu C Z, Wang R H, Ding D H et al. Point groups and elastic properties of two-
dimensional quasicrystals. Acta Crystallog A, 1996, 52(2): 251–256
[3] Yang W G, Ding D H et al. Atomtic model of dislocation in icosahedral quasicrystals.
Phil Mag A, 1998, 78(6): 1481–1497
[4] Fan T Y, Guo L H, Final governing equation of plane elasticity of icosahedral qua-
sicrystals. Phys Lett A, 2005, 341(5): 235–239
[5] Li L H, Fan T Y. Final governing equation of plane elasticity of icosahedral quasicrystals–
stress potential method. Chin Phys Lett, 2006, 24(9): 2519–2521
[6] Reynolds G A M, Golding B, Kortan A R et al. Isotropic elasticity of the Al-Cu-Li
quasicrystal. Phys Rev B, 1990, 41(2): 1194–1195
[7] Spoor P S, Maynard J D, Kortan A R. Elastic isotropy and anisotropy in quasicrys-
talline and cubic Al-Cu-Li. Phys Rev Lett, 1995, 75(19): 3462–3465
[8] Tanaka K, Mitarai, Koiwa M. Elastic constants of Al-based icosahedral quasicrystals.
Phil Mag A, 1996, 76(10): 1715–1723
[9] Duquesne J-Y, Perrin B. Elastic wave interaction in icosahedral AlPdMn. Physica
B, 2002, 316–317: 317–320
190 Chapter 9 Theory of elasticity of three-dimensional quasicrystals and its...
their consideration and comparison. Though researchers believe that the hydrody-
namics based on the argument of Lubensky et al is more fundamentally sound, the
major shortcoming so far is lack of proper experimental data confirmation. Recently
the research interest in the respect is growing up[13∼16] , but the most important ac-
complishment shall still be the quantitative results.
They claimed that law of the momentum conservation holds for both phonons and
phasons, namely for linear and small deformation case the equations of motion are
∂σij ∂ 2 ui ∂Hij ∂ 2 wi
=ρ 2 , =ρ 2 , (10.1-3)
∂xj ∂t ∂xj ∂t
⎧
⎪ ∂uz ∂wz
⎪
⎪ σzy = σyz = μ +R ,
⎪
⎪ ∂y ∂y
⎪
⎪
⎪ ∂u ∂w
⎨ σxz = σzx = μ z + R z ,
∂x ∂x (10.2-1)
⎪
⎪ ∂wz ∂uz
⎪
⎪ Hzy = (K1 − K2 ) +R ,
⎪
⎪ ∂y ∂y
⎪
⎪
⎩ H = (K − K ) ∂wz + R ∂uz .
zx 1 2
∂x ∂x
Substituting expressions (10.2-1) into equations of motion of (10.1-3) yields
∂ 2 uz
μ∇2 uz + R∇2 wz = ρ ,
∂t2
∂ 2 wz
R∇2 uz + (K1 − K2 )∇2 wz = ρ . (10.2-2)
∂t2
If define displacement functions φ and ψ as[7]
1 ∂2φ 1 ∂ 2ψ
∇2 φ = , ∇2 ψ = , (10.2-4)
s21 ∂t2 s22 ∂t2
where
1 #
α= [μ − (K1 − K2 ) + (μ − (K1 − K2 ))2 + 4R2 ] (10.2-5)
2
and +
εj
sj = , j = 1, 2, (10.2-6)
ρ
1 #
ε1,2 = [μ + (K1 − K2 ) ± (μ − (K1 − K2 ))2 + 4R2 ],
2
sj can be understood as the speeds of wave propagation in anti-plane deformation
of the material. It is obvious that the wave speeds result from the phonon-phason
coupling. If there is no coupling, i.e., R → 0, then
+ &
μ (K1 − K2 )
s1 → , s2 → , (10.2-7)
ρ ρ
+ +
μ K1 − K2
where represents the speed of transverse wave of the phonon field and
ρ ρ
represents the speed of the pure phason elastic wave, requiring K1 − K2 > 0.
Substituting (10.2-3) into (10.2-1) the stresses can be expressed by φ and ψ
194 Chapter 10 Dynamics of elasticity and defects of quasicrystals
⎧
⎪ ∂φ ∂ψ
⎪
⎪ σyz = σzy = (αμ + R2 ) + R(α − μ) ,
⎪
⎪ ∂y ∂y
⎪
⎪
⎪
⎪ ∂φ ∂ψ
⎨ σxz = σzx = (αμ + R2 ) + R(α − μ) ,
∂x ∂x
(10.2-8)
⎪
⎪ Hzy = R(α + (K1 − K2 )) ∂φ 2 ∂ψ
⎪
⎪ + (α(K 1 − K 2 ) − R ) ,
⎪
⎪ ∂y ∂y
⎪
⎪
⎪
⎩ Hzx = R3 (α + (K1 − K2 )) ∂φ ∂ψ
+ (α(K1 − K2 ) − R2 ) .
∂x ∂x
Formulas (10.2-3) and (10.2-8) give the expressions for displacements and stresses in
terms of displacement functions φ and ψ, which satisfy the standard wave equations
(10.2-4) for elastodynamics of anti-plane elasticity of three-dimensional icosahedral
quasicrystals.
The above discussion is valid for anti-plane elasticity of three-dimensional cu-
bic quasicrystals or one-dimensional quasicrystals too. The difference between these
quasicrystals is only the material constants. If μ, K1 − K2 and R are replaced by
C44 , K44 and R44 (see Section 9.8) for cubic quasicrystals, or by C44 , K2 and R3 for
one-dimensional hexagonal quasicrystals with the Laue classes 6/mh and 6/mh mm
(see Sections 7.1 or 8.1) one can find the similar equations.
The solution of (10.2-4) can be done by using method for solving pure wave
equations in classical mathematical physics.
where
∂2 ∂2 ∂2 ∂2
∇21 = 2
+ 2 , ∇21 = 2
+ 2, (10.3-4a)
∂x ∂y1 ∂x ∂y2
,
yj = βj y, βj = 1 − V 2 /s2j , j = 1, 2. (10.3-4b)
Let complex variable zj be
√
zj = x + iyj , i= −1, (10.3-5)
the solution of equations (10.3-3) is
where F1 (z1 ) and F2 (z2 ) are analytic functions of z1 and z2 respectively, and notation
Im marks the imaginary part of a complex function.
The boundary condition (10.3-1) determines the analytic functions as[7]
A1 y1 A2 y2
φ(x, y1 ) = arctan , ψ(x, y1 ) = arctan (10.3-7a)
2π x 2π x
with constants
|| ||
αb3 + Rb⊥ 3 αb3 − Rb⊥ 3
A1 = , A 2 = , (10.3-7b)
α 2 + R2 α2 + R2
the displacement field is determined in the fixed coordinate system as follows
1 2 β1 y 2 β2 y ||
uz (x, y, t) = α arctan + R arctan b
2π(α2 + R2 ) x−Vt x−Vt 3
(10.3-8a)
β1 y β2 y
+ arctan − arctan αRb⊥ 3 ,
x−Vt x−Vt
1 2 β1 y 2 β2 y
wz (x, y, t) = R arctan + α 3 arctan b⊥
2π(α2 + R2 ) x−Vt x−Vt 3
β1 y β2 y ||
+ arctan − arctan αRb3 . (10.3-8b)
x−Vt x−Vt
The expressions for strains and stresses are omitted here due to limitation of space.
We give the evaluation on the energy of the moving dislocation. Denote energy
W per unit length on the moving dislocation which consists of the kinetic energy
Wk and potential energy Wp defined by the integrals
2 2
1 ∂uz ∂wz
Wk = ρ + dx1 dx2 ,
2 ∂t ∂t
Ω
(10.3-9)
1 ∂uz ∂wz
Wp = σij + Hij dx1 dx2 ,
2 ∂t ∂t
Ω
respectively, where the integration should be taken over a ring r0 < r < R0 , r0
denotes the size of the dislocation core, and R0 the size of so-called dislocation net
196 Chapter 10 Dynamics of elasticity and defects of quasicrystals
1 1 R0
W ≈ μb + ρV 2 b2
2
ln ,
2 4π r0
ρb2 R0
m0 = ln ,
4π r0
this is exactly identical to the well known Eshellby solution for crystals[17] .
The above discussion holds for anti-plane elasticity of three-dimensional cubic
or one-dimensional hexagonal quasicrystals, only the material constants μ, K1 − K2
and R should be replaced by C44 , K44 and R44 , or by C44 , K2 and R3 , respectively.
It is easy to prove that (10.4-9) satisfies the relevant boundary conditions, and
is the exact solution.
Similarly, σxz = σzx , Hzx and Hzy can also be expressed explicitly.
From (10.4-8), as y = 0, it yields
xτ
√ − τ, |x| > a,
σyz (x, 0) = x − a2
2 (10.4-10)
−τ, |x| < a
The stress presents singularity of order (x − a)−1/2 as x → a.
The stress intensity factor for Mode III for phonon field is
||
# √
KI = lim+ π(x − a)σyz (x, 0) = πaτ. (10.4-11)
x→a
This is identical to the classical Yoffe solution[18] , there the stress intensity factor is
also independent of crack moving speed V .
Now we calculate energy of the moving crack, which is defined by
a
W=2 [σzy (x, 0) ⊕ Hzy (x, 0)][uz (x, 0) ⊕ wz (x, 0)]dx
0
1 1
= (Δ1 α−Δ2 R)τ πa = [αβ2 (α(K1 −K2 ) − R2 )−β1 R2 (α + (K1 − K2 ))]πa2 τ.
Δ Δ
(10.4-12)
The crack energy release rate is
1 ∂W 1
G= = [αβ2 (α(K1 −K2 )−R2 ) − β1 R2 (α + (K1 −K2 )](KI )2 . (10.4-13)
2 ∂a 2Δ
The above discussion, results and conclusions hold too for anti-plane elasticity
of three-dimensional cubic or one-dimensional hexagonal quasicrystals, only the ma-
terial constants μ, K1 − K2 and R need to be replaced by C44 , K44 and R44 or by
C44 , K2 and R3 , respectively. The results are given by the author in the Chinese
edition of this book in 1999.
which are different from those based on the argument of Lubensky et al. Meanwhile
Rochal and Lorman[15] did not completely agree to the above equations even though
they followed the Bak’s argument. They suggested the density ρ in the second equa-
tion listed above should be replaced by ρef , where ρef represents the generalized
effective phason density, but the meaning of the quantity is not very clear and its
measurement is difficult. Fan et al[13] , Rochal and Lorman[16] have respectively sug-
gested a new version on elasto-/hydro-dynamics of quasicrystals, this compromises
different models, e.g. those proposed by Ref. [1] and Ref. [5]. At present the experi-
mental data are not so sufficient to verify each model individually. It is the anthor’s
opinion that it shall be highly desirable that all researchers collaborate the common
features of different models and find a simplest version for elasto-/hydro-dynamics
of quasicrystals, then carry out systematic, theoretical, numerical and experimental
work; that would be beneficial for promoting the study in the field.
Fan et al[13] presented results incorporating the arguments of both Bak and
Lubensky et al. They suggested the equations of motion under linear and small
deformation be written as ⎧
⎪
⎪ ∂ 2 ui ∂σij
⎨ ρ 2 = ,
∂t ∂xj (10.5-1)
⎪
⎪ ∂wi ∂Hij
⎩ κ = ,
∂t ∂xj
It is obvious that the first equation of (10.5-1) is equation of conventional elastody-
namics, in which ρ is the mass density as mentioned in the previous section, while
the second one is diffusion equation, where κ = 1/Γw , in which Γw the kinematic
coefficient of phason field of the material defined by Lubensky et al[1] . It can be seen
that the second equation of (10.5-1) is a linearized result of hydrodynamics of qua-
sicrystals of Lubensky et al. The dynamic equations (10.5-1) should be named as the
elasto-/hydro-dynamic equations for quasicrystals which are identical to equations
(6) of Ref. [16]. This treatment is believed to be physically sound and reconciles the
contradiction between the arguments of Bak and Lubensky et al; it also reconciles
the contradiction between Refs. [5 ∼ 12] and Ref. [15], though they all follow the
framework of Bak’s argument.
As an example we discuss herein an approximate analytic solution of moving
dislocation in anti-plane problem of three-dimensional icosahedral or cubic or one-
dimensional hexagonal quasicrystal, see Fan et al[13] . The stress-strain relations in
this case are the same as (9.2-3), the difference lies in only the equations of motion
which are given by (10.5.1), such that for anti-plane which become
⎧ 2
⎪
⎨ μ∇2 uz + R∇2 wz = ρ ∂ uz ,
∂t2 (10.5-2)
⎪
⎩ R∇ uz + (K1 − K2 )∇2 wz = κ ∂wz .
2
∂t
10.5 Elasto-/hydro-dynamics of quasicrystals and approximate analytic solution... 201
Equations (10.5-2) are wave-diffusion mixed type equations which are different from
(10.2-2), so the solution will be difficult to obtain.
The following boundary conditions for screw dislocation moving along direction
Ox with speed V are considered as
⎧ #
2 2
⎨ x + y → ∞ : σij (x, y, t) → 0,
⎪ Hij (x, y, t) → 0,
(10.5-3)
⎪
⎩
duz = b3 , dwz = b⊥3.
Γ Γ
Solving boundary value problem (10.5-2), (10.5-3) is more complicated than solv-
ing either a pure wave propagation problem or a pure diffusion problem. We take
the perturbation method along with variational method. Based on the physical
consideration R/μ should be small and it can be taken as the perturbation param-
eter. Firstly, we calculate the zeroth-order perturbation solution. The governing
equations (10.5-2) reduce to
⎧
⎪
⎪ ∂ 2 u0
⎨ ρ 2 − μ∇2 u0 = 0,
∂t
(10.5-4)
⎪
⎪ ∂w 0
⎩ κ − (K1 − K2 )∇2 w 0 = 0,
∂t
is used, here (x1 , y, t) denotes fixed coordinate system and (x, y) the moving coor-
dinate system, and
∂ ∂ ∂2 ∂2
= −V , 2
= V2 2. (10.5-6)
∂t ∂x ∂t ∂x
where
V2 Vκ V
β12 = 1 − , β22 = = , (10.5-8)
c2 K1 − K2 (K1 − K2 )Γw
202 Chapter 10 Dynamics of elasticity and defects of quasicrystals
#
in which c = μ/ρ for icosahedral quasicrystals is the velocity of transverse wave
of phonon field.
Because the equations in (10.5-7) are decoupled, and considering dislocation
conditions in (10.5-3) we have the following two separate dislocation problems such
as
⎧ 2 0
⎪
⎪ ∂ u 1 ∂ 2 u0
⎨ 2
+ 2 = 0,
∂x β1 ∂y2
(10.5-9a)
⎪
⎪ 0
⎩ du = b3 ,
Γ
⎧ 2 0
⎪
⎪ ∂ w ∂w0 ∂ 2 w0
⎨ 2
+ β22 + = 0,
∂x ∂x ∂y 2
(10.5-9b)
⎪
⎪
⎩ dw0 = b⊥3.
Γ
b3 β1 y
u0 = arctan . (10.5-10)
2π x
w0 = w1 + w2 , (10.5-11)
in which w1 satisfies
⎧ 2
⎪
⎪ ∂ w1 ∂ 2 w1
⎨ + = 0,
∂x2 ∂y2
(10.5-12)
⎪
⎪
⎩ dw1 = b⊥
3,
Γ
The equation in problem (10.5-14) is more complex than that in (10.5-9a) and
(10.5-12), so its analytic solution is very difficult to obtain. We can do only ap-
proximate solution by variational principle (refer to Section 13.2 in Chapter 13).
Omitting the detail the approximate solution is
b⊥
3 β22 x2 y
w2 = − . (10.5-15)
2π (x2 + y 2 )(1 + β22 x)
b⊥
3 R 2β22 x2 y(3x2 − y 2 + 6β22 x3 + 2β24 x4 )
u1 = − ,
2π (ρV 2 − μ) (x2 + y 2 )2 (1 + β22 x)3
(10.5-18)
1 b3 R 2(β13 − β1 )x3 y
w =− .
2π(K1 − K2 ) (x2 + β12 y 2 )2 (1 + β22 x)
⎧
⎪
⎪ 1 b3 β1 x b⊥ β22 R
⎪
⎪ε = ε = + 3
⎪
⎪
yz zy 2 + β 2 y2 π ρV 2 − μ
⎪
⎪
2 2π x 1
⎪
⎪ 6 − 12x4 y 2 + y 6 + 6(x7 − 3x5 y 2 )β 2 + 2(x8 − 3x6 y 2 )β 4
⎪
⎪ 3x
⎪
⎪ × 2 2
,
⎪
⎪ (x2 + y 2 )3 (1 + β22 x)3
⎪
⎪
⎪
⎪
⎪
⎪
b⊥ β22 R
⎪
⎪ 1 b β1 y
⎪
⎪εzx = εxz = − 3 2 + 3
⎪
⎪ 2 2π x + β1 y 2 2 π ρV 2 − μ
⎪
⎪
⎨ 14x y −2xy +(−3x y+38x4 y 3+x2 y 5 )β22 −8(x7 y−3x5 y 3 )β24−2(x8 y−3x6 y3 )β26
3 3 5 6
× ,
⎪
⎪
2 2 3
(x +y ) (1+β2 x) 2 4
⎪
⎪
⎪
⎪
⎪
⎪w b⊥ 2 2
3 x(x + y + 2β2 xy )
2 2 b3 R x3 (x2 − 3β12 y 2 )(β13 − β1 )
⎪
⎪ = − ,
⎪
⎪
zy
2π (x 2 + y 2 )2 (1 + β 2 x) π K − K 2 2
2 (1 + β2 x)(x + β1 y )
2 2 3
⎪
⎪ 2 1
⎪
⎪
⎪
⎪ b⊥ y(2β22 x3 + y 2 + 4β22 xy 2 + x2 (1 + 2β24 y 2 ))
⎪
⎪wzx = − 3
⎪
⎪ (x2 + y 2 )2 (1 + β22 x)2
⎪
⎪
2π
⎪
⎪
⎪
⎪
x2 y(x2 + 2β22 x3 − 2β12 β22 xy 2 − 3β12 y 2 )(β13 − β1 )
⎪
⎪
b R
⎩ + 3
π K1 − K2 (1 + β22 x)2 (x2 + β12 y 2 )3
(10.5-20)
and the stresses can be expressed by
⎧
⎪
⎪ σyz = σzy
⎪
⎪
⎪
⎪
⎪
⎪ b⊥ 2β22 Rμ 3x6 −12x4 y 2 +y 6 +6(x7 −3x5 y 2 )β22 +2(x8 −3x6 y 2 )β24
⎪
⎪ 3
⎪
⎪ =
⎪
⎪ 2π ρV 2 −μ (x2 +y 2 )3 (1+β22 x)3
⎪
⎪
⎪
⎪
⎪
⎪
2 2
Rx(x +y +2β2 xy ) 2 2 b μβ1 x 2R2 x3 (x2 −3β12 y 2 )(β13 −β1 )
⎪
⎪ + + 3 − ,
⎪
⎪ 2 2 2
(x +y ) (1+β2 x) 2 2
2π x +β1 y 2 2 K1−K2 (1+β2 x)(x +β1 6y )
2 2 2 2 3
⎪
⎪
⎪
⎪
⎪
⎪
⊥
b R 2β2 μ 2
⎪
⎪ σzx = σxz = 3
⎪
⎪ 2π ρV 2 −μ
⎪
⎪
⎪
⎪
⎪
⎪ 14x3 y 3−2xy 5+(−3x6 y+38x4 y 3 +x2 y 5 )β22−8(x7 y−3x5 y 3 )β24−2(x8 y−3x6 y 3 )β26 )
⎪
⎪
⎪
⎪ ×
⎪
⎪ (x2 +y 2 )3 (1+β22 x)4
⎪
⎪
⎪
⎪ y(2β2 x +y +4β2 xy +x (1+2β24 y 2 ))
2 3 2 2 2 2
⎪
⎪
⎪
⎪ −
⎪
⎪ (x2 +y 2 )2 (1+β22 x)2
⎪
⎪
⎪
⎨
b 2R2 x2 y(x2 +2β22 x3 −2β12 β22 xy 2 −3β12 y 2 )(β13 −β1 ) μβ1 y
+ 3 − ,
⎪
⎪ 2π K1 −K2 2 2
(1+β2 x)2 (x2 +β1 y 2 )3 2
x2 +β1 y 2
⎪
⎪ ⊥ 2β 2 R2 3x6 −12x4 y 2 +y 6 +6(x7 −3x5 y 2 )β 2 +2(x8 −3x6 y 2 )β 4
⎪
⎪ b
⎪
⎪ 3 2 2 2
⎪Hzy = 2π ρV 2 −μ
⎪ (x2 +y 2 )3 (1+β22 x)3
⎪
⎪
⎪
⎪
⎪
⎪ x(x2 +y 2 +2β22 xy2 ) b3 R 2x3 (x2−3β12 y 2 )(β13 −β1 )
⎪
⎪ β1 x
⎪
⎪ +(K1 −K2 ) 2 + − .
⎪
⎪
2 2
(x +y ) (1+β2 x) 2 2
2π x +β1 y 2 2 2 2
(1+β2 x)(x +β1 y ) 2 2 3
⎪
⎪
⎪
⎪ b⊥ 2β22 R
⎪
⎪
⎪
⎪Hzx = 3
⎪
⎪ 2π ρV 2 −μ
⎪
⎪
⎪
⎪ 14x3 y 3−2xy 5+(−3x6 y+38x4 y 3+x2 y 5 )β22−8(x7 y−3x5 y 3 )β24−2(x8 y−3x6 y 3 )β26
⎪
⎪
⎪
⎪ ×
⎪
⎪ (x2+y 2 )3 (1+β22 x)4
⎪
⎪
⎪
⎪ y(2β2 x +y +4β2 xy +x2 (1+2β24 y 2 ))
2 3 2 2 2
⎪
⎪
⎪
⎪ −(K 1−K 2 )
⎪
⎪ (x2 +y 2 )2 (1+β22 x)2
⎪
⎪
⎪
⎪ 2
⎪
⎪ b R 2x y(x2 +2β22 x3 −2β12 β22 xy2 −3β12 y 2 )(β13 −β1 ) β1 y
⎪
⎩ + 3 − .
2π 2 2
(1+β2 x)2 (x2 +β1 y 2 )3 2
x2 +β1 y2
(10.5-21)
For cubic or hexagonal quasicrystals, the solutions are exactly similar, with
10.5 Elasto-/hydro-dynamics of quasicrystals and approximate analytic solution... 205
μ, (K1 − K2 ) and R replaced by C44 , K44 and R44 or by C44 , K2 and R3 respec-
tively in the above equations.
It is easy to observe that the solution contains the contributions coming from
three parts of wave propagation, diffusion and their interaction. The solution on
phonon field is dominated by wave propagation, and the solution on phason field is
dominated by diffusion, so the result is quite different from the dynamic solution of
the dislocation based on the Bak’s argument, which was discussed in Section 10.3
already. The present solution can reduce to that in static dislocation as V = 0,
which can also reduce to the solution of crystals, this demonstrates the correctness
of the work. The solution reveals that the phason field resists motion of dislocation,
this is the reason of difficulty of plastic deformation of quasicrystalline materials.
The illustrations of the solution and comparison with other solutions for the case
#
V = 0.1c, c = μ/ρ are depicted in Figs.10.5-1∼10.5-4, in which the solution given
by Ref. [7] has been introduced in Section 10.3. In the numerical computation we
take the data of material constants as
b⊥
3 = 0.8b3
Fig.10.5-1 Variation of uz /b3 at x = 0.01 mm versus y
Fig.10.5-2 Variation of uz /b3 at y = 0.01 mm versus x
206 Chapter 10 Dynamics of elasticity and defects of quasicrystals
Fig.10.5-3 Variation wz /b3 at x = 0.01 mm versus y
Fig.10.5-4 Variation wz /b3 at y = 0.01 mm versus x
R
ρ = 4.1kg/cm3 , μ = 70GPa, K1 = 74.2, K2 = −37MPa, = 0.004,
μ
1
Γw = = 4.8 × 10−19 m3 · s/kg = 4.8 × 10−10 cm3 · μs/g
κ
#
given in Tables 9.1-1∼9.1-3. We recall x = x1 − V t, c = μ/ρ.
onal quasicrystals
Among 200 quasicrystals observed to date, there are 70 two-dimensional decagonal
quasicrystals; so this kind of solid phases plays an important role in the material.
For simplicity, herein only point group 10mm two-dimensional decagonal quasicrystal
will be considered. We denote the periodic direction as the z-axis and the quasiperi-
odic plane as the xy-plane. Assume that a crack in the solid along the periodic
direction, i.e., the z-axis. It is obvious that elastic field induced by a uniform tensile
stress at upper and lower surfaces of the specimen is independent of z, so ∂/∂z = 0.
In this case, the stress-strain relations are reduced to
⎧
⎪
⎪ σxx = L(εxx + εyy ) + 2M εxx + R(wxx + wyy ),
⎪
⎪
⎪
⎪ σ = L(εxx + εyy ) + 2M εyy − R(wxx + wyy ),
⎪
⎪ yy
⎪
⎪
⎨ σxy = σyx = 2M εxy + R(wyx − wxy ),
⎪
Hxx = K1 wxx + K2 wyy + R(εxx − εyy ), (10.6-1)
⎪
⎪
⎪ Hyy = K1 wyy + K2 wxx + R(εxx − εyy ),
⎪
⎪
⎪
⎪
⎪
⎪
⎪ Hxy = K1 wxy − K2 wyx − 2Rεxy ,
⎪
⎩ H = K w − K w + 2Rε ,
yx 1 yx 2 xy xy
where L = C12 , M = (C11 − C12 )/2 are the phonon elastic constants, K1 and K2
the phason elastic constants, R phonon-phason coupling elastic constant.
Substituting (10.6-1) into (10.5-1), we obtain the equations of motion of decago-
nal quasicrystals as follows:
⎧ 2 2 2 2
2
⎪ ∂ ux 2 ∂ ux 2 ∂ uy 2 ∂ ux ∂ wx ∂ 2 wy ∂ 2 wx
⎪
⎪ = c1 2
+ (c1 − c2 ) + c + c3 2
+2 − ,
⎪
⎪ ∂t2 ∂x2 ∂x∂y 2
∂y 2 ∂x2 ∂x∂y ∂y 2
⎪
⎪ 2
⎪ ∂ 2 uy
⎪ 2 2 2
∂ 2 wx ∂ 2 wy
⎪
⎪ 2 ∂ uy 2 2 ∂ ux 2 ∂ uy 2 ∂ wy
⎨ = c 2 + (c1 − c 2 ) + c 1 + c 3 − 2 − ,
∂t2 ∂x2 ∂x∂y ∂y 2 ∂x2 ∂x∂y ∂y 2
2
⎪
⎪ ∂wx ∂ wx ∂ 2 wx ∂ 2 ux ∂ 2 uy ∂ 2 ux
⎪
⎪ = d 2
+ + d 2
− 2 − ,
⎪
⎪ ∂t 1
∂x2 ∂y2 2
∂x2 ∂x∂y ∂y 2
⎪
⎪
⎪
⎪ ∂wy ∂ 2 wy ∂ 2 wy ∂ 2 uy ∂ 2 ux ∂ 2 uy
⎪
⎩ = d21 + + d2
+ 2 − ,
2
∂t ∂x2 ∂y 2 ∂x2 ∂x∂y ∂y 2
(10.6-2)
in which
& & & + + +
L + 2M M R K1 R K2
c1 = , c2 = , c3 = , d1 = , d2 = , d3 = ,
ρ ρ ρ κ κ κ
(10.6-3)
208 Chapter 10 Dynamics of elasticity and defects of quasicrystals
note that constants c1 , c2 and c3 have the meaning of elastic wave speeds, while
d21 , d22 and d23 do not represent wave speeds, they are diffusive coefficients.
The decagonal quasicrystal with a crack is shown in Fig.10.6-1. It is a rectangu-
lar specimen with a central crack of length 2a(t) subjected to a dynamic or static
tensile stress at its ends ED and F C, in which a(t)represents the crack length as
a function of time, and for dynamic initiation of crack growth, the crack is stable,
so a(t) = a0 = constant, for fast crack propagation, a(t) varies with time. At first
we consider dynamic initiation of crack growth, then study crack fast propagation.
Due to the symmetry of the specimen only the upper right quarter is considered.
Referring to the upper right part and considering a fix grips case, the following
boundary conditions should be satisfied:
⎧
⎪
⎪ ux = 0, σyx = 0, wx = 0, Hyx = 0, on x = 0 for 0 y H,
⎪
⎪
⎪
⎨ σxx = 0, σyx = 0, Hxx = 0, Hyx = 0, on x = L for 0 y H,
⎪
σyy = p(t), σxy = 0, Hyy = 0, Hxy = 0, on y = H for 0 x L, (10.6-4)
⎪
⎪
⎪
⎪ σyy = 0, σxy = 0, Hyy = 0, Hxy = 0, on y = 0 for 0 x a(t),
⎪
⎪
⎩
uy = 0, σxy = 0, wy = 0, Hxy = 0, on y = 0 for a(t) x L,
where p(t) = p0 f (t) is a dynamic load if f (t) varies with time, otherwise it is a static
load (i.e., if f (t) = const), and p0 = const with the stress dimension.
The initial conditions
⎧ are
⎪
⎪
ux (x, y, t)|t=0 = 0, uy (x, y, t)|t=0 = 0,
⎨
wx (x, y, t)|t=0 = 0, wy (x, y, t)|t=0 = 0, (10.6-5)
⎪
⎪ ∂ux (x, y, t) ∂uy (x, y, t)
⎩ |t=0 = 0, |t=0 = 0.
∂t ∂t
10.6 Elasto-/hydro-dynamics and solutions of two-dimensional decagonal... 209
For implementation of finite difference all field variables in both governing equa-
tions (10.6-2) and boundary-initial conditions (10.6-4), (10.6-5) must be expressed
by displacements and their derivatives. This can be done through the constitutive
equations (10.6-1). The detail of the finite difference scheme is given in Appendix
of this chapter.
For the related parameters in this section, the experimentally determined mass
density for decagonal Al-Ni-Co quasicrystal: ρ = 4.186 × 10−3 g · mm−3 is used and
elastic phonon moduli are taken as C11 = 2.3433, C12 = 0.5741(1012 dyn/cm2 =
102 GPa) which are obtained by resonant ultrasound spectroscopy[19] , we have also
chosen phason elastic constants K1 = 1.22 and K2 = 0.24(1012 dyn/cm2 = 102 GPa)
estimated by Monto-Carlo simulation[20] and Γw = 1/κ = 4.8 × 10−19 m3 · s/kg =
4.8×10−10 cm3 ·μs/g[21] .The coupling constant R has been measured for some special
cases recently, see Chapter 6 and Chapter 9 respectively. In computation we take
R/M = 0.01 for coupling case corresponding to quasicrystals, and R/M = 0 for
decoupled case which corresponds to crystals.
The stability of the scheme is the core problem of finite difference method which
depends upon the choice of parameter α = c1 τ /h, the ratio between time step and
space step substantively. The choice is related to the ratio c1 /c2 , i.e., the ratio
between speeds of elastic longitudinal and transverse waves of the phonon field. To
determine the upper bound for the ratio to guarantee the stability, according to our
computational practice and experiences of computations for conventional materials,
α = 0.8 is chosen in all cases. The computational results are stable.
special short time interval our specimen can be seen as an “infinite specimen”.
The comparison given by Fig.10.6-3 shows the numerical results are in excellent
agreement with those of Maue’s solution within the short interval in which the
solution is valid.
Our solution for the case of wx = wy = 0 is also compared with numerical
solutions of conventional crystals, e.g. Murti’s solution[23] and Chen’s solutions[24] ,
which are also shown in Fig.10.6-3, it is evident, our solution is in good agreement
with available solutions.
The mesh size or the space step of the algorithm can influence the computational
accuracy too. To check the accuracy of the algorithm we take different space steps
shown in Table 10.6-1, which indicates if h = a0 /40 the accuracy is good enough.
The check is carried out through static solution, because the static crack problem
in infinite body of decagonal quasicrystals has exact solution, see Chapter 8, the
normalized static intensity factor is equal to unit. In the static case, there is no
wave propagation effect, L/a0 3, H/a0 3 the effect of boundary to solution is
very weak, and for our present specimen L/a0 4, H/a0 8, which may be seen as
an infinite specimen, so the normalized static stress intensity factor is approximately
but with highly precise equal to unit. The table shows that the algorithm is with a
quite highly accuracy when h = a0 /40.
Table 10.6-1 The normalized static S.I.F. of quasicrystals for different space
steps
h a0 /10 a0 /15 a0 /20 a0 /30 a0 /40
K̄ 0.9259 0.94829 0.96229 0.97723 0.99516
Errors 7.410% 5.171% 3.771% 2.277% 0.484%
10.6 Elasto-/hydro-dynamics and solutions of two-dimensional decagonal... 213
There are many oscillations in the figure, especially the stress intensity factor.
These oscillations characterize the reflection and diffraction between waves coming
from the crack surface and the specimen boundary surfaces. The oscillations are
influenced by the material constants and specimen geometry including the shape
and size.
In this section, we focus on the discussion for the “phase” of fast crack propagation.
To explore the inertia effect caused by the fast crack propagation, the specimen
is designed under the action of constant load P (t) = p rather than time-varying
load, but the crack grows with high speed in this case. The problem for fast crack
propagation is a nonlinear problem, because one part of the boundaries—crack is
with unknown length beforehand. For this moving boundary problem, we must give
additional condition for determining solution. That is, we must give a criterion
checking crack propagation or crack arrest at the growing crack tip. This criterion
can be imposed in different ways, e.g., the critical stress criterion or critical energy
criterion. The stress criterion is used in this paper:σyy < σc represents crack arrest,
σyy = σc represents critical state and σyy > σc represents crack propagation. Here
we take σc = 450M P a for decagonal Al-Ni-Co quasicrystals, which was obtained by
referring measured value by Meng et al for decagonal Al-Cu-Co quasicrystals, refer
to Ref.[2] in Chapter 8, the modification by referring the hardness of alloys Al-Ni-Co
and Al-Cu-Co, and the hardness on decagonal Al-Ni-Co can be found in paper given
by Takeuchi et al[30] . The simulation of a fracturing process runs as follows:
Given the specimen geometry and its material constants we first solve the initial
dynamic problem in the way previously described. When the stress σyy reaches a
prescribed critical value σc the crack is extended by one grid interval. The crack
now continues to grow, by one grid interval at a time, as long as the σyy stress level
ahead of the propagating crack tip reaches the value of σc . During the propagation
stage the time that elapses between two sequential extensions is recorded and the
corresponding velocity is evaluated.
The crack velocity for quasicrystals and periodic crystals is constructed in Fig.10.6-
5, from the figure, we observe that the velocity in quasicrystals is lower than that
of the periodic crystals; the phonon-phason coupling makes the quasicrystals being
different from periodic crystals. The reason for this is not so clear. Though Mikulla
et al[25] gave an explanation for crack slowly propagation (i.e., the quasi-static crack
growth) in decagonal quasi crystals, the case is quite different from the present crack
fast propagation.
We find that the fast crack propagating velocity is obviously different in qua-
10.6 Elasto-/hydro-dynamics and solutions of two-dimensional decagonal... 215
Fig.10.6-6 Variation of crack velocity versus crack growth size for different load levels
scanning tunneling microscopy for quasi-static crack growth. Because the fast propa-
gation and quasi-static crack growth belong to two different regimes, the comparison
cannot be easily made.
1 ∂ui ∂uj ∂wi
εij = + , wij = ,
2 ∂xj ∂xi ∂xj
σij = Cijkl εkl + Rijkl wkl , Hij = Rklij εkl + Kijkl wkl .
In particular the elastic constants are quite different from those discussed in the
previous sections, in which the phonon elastic constants can be expressed such as
and the phason elastic constant matrix [K] and phonon-phason coupling elastic one
[R] are defined by formula (9.1-6) in Chapter 9, which are not listed here again.
Substituting these non-zero stress components into the equations of motion
and through the generalized Hooke’s law and strain-displacement relation we obtain
the final dynamic equations as follows:
2
∂ 2 ux ∂ux 2
2 ∂ ux 2
2
2 ∂ uy
2
2 ∂ ux 2 ∂ wx ∂ 2 wy ∂ 2 wx
+θ = c1 +(c1 −c2 ) +c +c3 +2 − ,
∂t2 ∂t ∂x2 ∂x∂y 2 ∂y2 ∂x2 ∂x∂y ∂y 2
∂ 2 uy ∂uy ∂ 2 uy ∂ 2 ux ∂ 2 uy ∂ 2 wy ∂ 2 wx ∂ 2 wy
2
+θ = c22 2
+(c21 −c22 ) +c21 2
+c23 2
−2 − ,
∂t ∂t ∂x ∂x∂y ∂y ∂x ∂x∂y ∂y 2
2 2
∂ 2 uz ∂uz 2 ∂ ∂2 2 ∂ wx ∂ 2 wx ∂ 2 wy ∂ 2 wz ∂ 2 wz
+θ = c2 + uz +c 3 − −2 + + ,
∂t2 ∂t ∂x2 ∂y 2 ∂x2 ∂y2 ∂x∂y ∂x2 ∂y2
2
∂wx ∂ ∂2 ∂2 ∂2
+θwx = d1 + w x +d 2 − wz
∂t ∂x2 ∂y 2 ∂x2 ∂y2
2
∂ ux ∂ 2 uy ∂ 2 ux ∂ 2 uz ∂ 2 uz
+d3 −2 − + −
∂x2 ∂x∂y ∂y 2 ∂x2 ∂y2
2 2
2
2
∂wy ∂ ∂ ∂ wz ∂ uy ∂ 2 ux ∂ 2 uy ∂ 2 uz
+θwy = d1 + w y −d 2 +d 3 +2 − −2 ,
∂t ∂x2 ∂y 2 ∂x∂y ∂x2 ∂x∂y ∂y2 ∂x∂y
2
∂wz ∂ ∂2
+θwz = (d1 −d2 ) + wz
∂t ∂x2 ∂y2
2 2
∂ wx ∂ 2 wx ∂ 2 wy ∂ ∂2
+d2 − −2 +d 3 + uz ,
∂x2 ∂y 2 ∂x∂y ∂x2 ∂y 2
(10.7-4)
218 Chapter 10 Dynamics of elasticity and defects of quasicrystals
in which
& + &
λ + 2μ μ R K1 K2 R
c1 = , c2 = , c3 = , d1 = , d2 = , d3 = ,
ρ ρ ρ κ κ κ
(10.7-5)
note that constants c1 , c2 and c3 have the meaning of elastic wave speeds, while d1 , d2
and d3 do not represent wave speed, but are diffusive coefficients and parameter θ
may be understood as a manmade damping coefficient as in the previous section.
Consider an icosahedral quasicrystal specimen with a central crack shown in
Fig.10.6-1, all parameters of geometry and loading are the same with those given
in the previous section, but in the boundary conditions there are some different
points, which are given as below if only the upper right quarter of the specimen is
considered:
⎧
⎪
⎪ u = 0, σyx = 0, σzx = 0, wx = 0, Hyx = 0, Hzx = 0, on x = 0 for 0 y H,
⎪ x
⎪
⎪
⎪
⎨ σxx = 0, σyx = 0, σzx = 0, Hxx = 0, Hyx = 0, Hzx = 0, on x = L for 0 y H,
⎪
σyy = p(t), σxy = 0, σzy = 0, Hyy = 0, Hxy = 0, Hzy = 0, on y = H for 0 x L,
⎪
⎪
⎪
⎪ σyy = 0, σxy = 0, σzy = 0, Hyy = 0, Hxy = 0, Hzy = 0, on y = 0 for 0 x a(t),
⎪
⎪
⎪
⎩ u = 0, σ = 0, σ = 0, w = 0, H = 0, H = 0, on y = 0 for a(t) < x L.
y xy zy y xy zy
(10.7-6)
The initial conditions are
The numerical results for dynamic initiation of crack growth problem, the phonon
and phason displacements are shown in Fig.10.7-1.
√
and the normalized dynamics stress intensity factor (S.I.F.) K̃I (t) = KI (t)/ πa0 p0
is used, the results are illustrated in Fig.10.7-2, in which the comparison with those
of crystals are shown, one can see the effects of phason and phonon-phason coupling
are evident very much.
For the fast crack propagation problem the primary results of the dynamic stress
intensity factor versus time are given in Fig.10.7-3.
Fig. 10.7-3 Normalized stress intensity factor of propagating crack with constant crack
speed versus time
stress presents singularity with order r −1/2 , in which r denotes the distance measured
from the crack tip. For the fast crack propagation, which is a nonlinear problem—
moving boundary problem, one must provide additional condition for determining
solution. For this purpose we give a criterion for checking crack propagation/crack
arrest based on the critical stress criterion. Applications of this additional condition
for determining solution enable us to conduct numerical simulation of the moving
boundary value problem and examine the crack length-time evolution. However,
much important and difficult problems are left open for further study[31] .
convenience, the mesh size h is taken to be the same in both x and y directions. The
grid is extended beyond the half step by adding four special grid lines x = −h/2, x =
L + h/2, y = −h/2, y = H + h/2, which form the grid boundaries.
Denoting the time step by τ and using central difference approximations, the
finite difference formulation of equations (10.6-2), valid at the inner part of the grids
is
222 Chapter 10 Dynamics of elasticity and defects of quasicrystals
⎧
⎪
⎪ ux (x, y, t + τ ) = 2ux (x, y, t) − ux (x, y, t − τ )
⎪
⎪ τ 2
⎪
⎪
⎪
⎪ + c1 [ux (x + h, y, t) − 2ux (x, y, t) + ux (x − h, y, t)]
⎪
⎪ h 2
⎪
⎪ τ
⎪
⎪ + (c21 − c22 )[uy (x + h, y + h, t)
⎪
⎪
⎪
⎪ h
⎪
⎪ −uy (x + h, y − h, t) − uy (x − h, y + h, t) + uy (x − h, y − h, t)]
⎪
⎪
⎪
⎪ τ 2
⎪
⎪ c2 [ux (x, y + h, t) − 2ux (x, y, t) + ux (x, y − h, t)]
⎪
⎪ +
⎪
⎪ h 2
⎪
⎪ τ
⎪
⎪ + c3 [wx (x + h, y, t) − 2wx (x, y, t) + wx (x − h, y, t)]
⎪
⎪ h 2
⎪
⎪ τ
⎪
⎪ +2 c23 [wy (x + h, y + h, t) − wy (x + h, y − h, t)
⎪
⎪ h
⎪
⎪
⎪
⎪ −wy (x − h, y + h, t) + wy (x − h, y − h, t)]
⎪
⎪ τ 2
⎪
⎪
⎪
⎪ − c3 [wx (x, y + h, t) − 2wx (x, y, t) + wx (x, y − h, t)],
⎪
⎪ h
⎪
⎪
⎪
⎪ uy (x, y, t + τ ) = 2uy (x, y, t) − uy (x, y, t − τ )
⎪
⎪
⎪
⎪ τ 2
⎪
⎪ c2 [uy (x + h, y, t) − 2uy (x, y, t) + uy (x − h, y, t)]
⎪
⎪ +
⎪
⎪ hτ 2
⎪
⎪
⎪
⎪ + (c21 − c22 )[ux (x + h, y + h, t) − ux (x + h, y − h, t)
⎪
⎪ 2h
⎪
⎪
⎪
⎪ −ux (x − h, y + h, t) + ux (x − h, y − h, t)
⎪
⎪
⎪
⎪ τ 2
⎪
⎪
⎪
⎪ + c1 [uy (x, y + h, t) − 2uy (x, y, t) + uy (x, y − h, t)]
⎪
⎪ h 2
⎪
⎪ τ
⎪
⎪ + c3 [wy (x + h, y, t) − 2wy (x, y, t) + wy (x − h, y, t)]
⎪
⎪
⎪
⎪ h τ 2
⎨
−2 c23 [wx (x + h, y + h, t) − wx (x + h, y − h, t)
⎪ 2h
⎪
⎪ −wx (x − h, y + h, t) + wx (x − h, y − h, t)]
⎪
⎪
⎪
⎪ τ 2
⎪
⎪
⎪
⎪ − c3 [wy (x, y + h, t) − 2wy (x, y, t) + wy (x, y − h, t)],
⎪
⎪ h
⎪
⎪ τ
⎪
⎪ wx (x, y, t + τ ) = wx (x, y, t) + d22 2 [ux (x + h, y, t) − 2ux (x, y, t) + ux (x − h, y, t)]
⎪
⎪ h
⎪
⎪
⎪
⎪ τ
+d21 2 wx (x + h, y, t) + wx (x − h, y, t) − 4wx (x, y, t)
⎪
⎪
⎪
⎪ h
⎪
⎪
⎪
⎪ −
⎪
⎪ +w x (x, y + h, t) + w x (x, y h, t)
⎪
⎪
⎪
⎪
⎪
⎪ −2d 2 τ
uy (x + h, y + h, t) − uy (x + h, y − h, t)
⎪
⎪ 2
⎪
⎪ (2h)2
⎪
⎪
⎪
⎪ −u − − −
⎪
⎪ y (x h, y + h, t) + uy (x h, y h, t)
⎪
⎪
⎪
⎪ τ
⎪
⎪ −d22 2 [ux (x, y + h, t) − 2ux (x, y, t) + ux (x, y − h, t)] ,
⎪
⎪ h
⎪
⎪ 2 τ
⎪
⎪ wy (x, y, t + τ ) = wy (x, y, t) + d2 h2 [uy (x + h, y, t) − 2uy (x, y, t) + uy (x − h, y, t)]
⎪
⎪
⎪
⎪ τ
⎪
⎪ +d21 2 wy (x + h, y, t) + wy (x − h, y, t) − 4wy (x, y, t)
⎪
⎪
⎪
⎪ h
⎪
⎪
⎪
⎪ −
⎪
⎪
+w y (x, y + h, t) + w y (x, y h, t)
⎪
⎪
⎪
⎪ τ
⎪
⎪ 2
ux (x + h, y + h, t) − ux (x + h, y − h, t)
⎪
⎪ +2d 2
⎪
⎪ (2h)2
⎪
⎪
⎪
⎪ −u
⎪ x (x − h, y + h, t) + ux (x − h, y − h, t)
⎪
⎪
⎪
⎪ τ (10.8-1)
⎩ −d22 2 [uy (x, y + h, t) − 2uy (x, y, t) + uy (x, y − h, t)] .
h
10.8 Appendix of Chapter 10: The detail of finite difference scheme 223
The displacements at mesh points located at the special lines are determined
by satisfying the boundary conditions, we obtain respectively for points on the grid
lines x = −h/2 and x = L + h/2.
h h
− −
ux L+2 h , y, t =ux L−2 h , y, t
2 2
h (10.8-2a)
−h
2 −
wx L+ h
, y, t=wx L−2 h , y, t
2 2
d2 (c2 − 2c2 ) h2 h
±2 22 21 2 22 uy L− h , y + h, t − u y
2
L− h
, y − h, t
c3 d2 − c1 d1 2 2
2 2 2 2
1 c3 d2 − c1 d3 h h
± 2 2 w y
2
, y + h, t − w y
2
, y − h, t ,
2 c3 d2 − c21 d21
h h
L− 2 L− 2
(10.8-2b)
h h 1 h h
−2 −2
uy L+ h , y, t = uy L− h , y, t ± ux L− h , y + h, t − ux L− h , y − h, t
2 2
2 2 2 2 2
(10.8-2c)
−h
2
h
2
wy L+ h
, y, t = wy L− h
, y, t
2 2
(10.8-2d)
where the equations (10.8-2a) and (10.8-2b) related to x = −h/2 is not valid. From
the first condition of (10.6-5), at x = 0, ux = 0 and wx = 0. To satisfy the condition
the displacements ux and wx at x = −h/2 is approximated by
⎧
⎪
⎪ h h
⎨ ux x, − , t = −ux x, , t ,
2 2 (10.8-3)
⎪
⎪ h h
⎩ wx x, − , t = −wx x, , t .
2 2
(10.8-4a)
224 Chapter 10 Dynamics of elasticity and defects of quasicrystals
−h
2 −h
2
wx x,H+ h ,t = wx x,H− h ,t
2 2
1 c2 d2 − c22 d23 h
h
± 32 22 wy x + h, 2
, t − w y x − h, 2
, t ,
2 c2 d1 − c23 d22
h h
H− 2 H− 2
(10.8-4b)
h
−h
2 2
uy x,H+ h , t = uy x,H− h ,t
2 2
(10.8-4c)
h
−h
2 2
wy x,H+ h , t = wy x,H− h ,t
2 2
d2 (c2 − c2 ) h
h
± 22 2 1 2 2 2 ux x + h,H−2
h ,t − ux x − h,H−
2
h ,t
c1 d1 − c3 d2 2 2
2 2 2 2
1 c d − c3 d2 h
h
± 12 32 2 2
2
wx x + h,H− h ,t − wx x − h,H−
2
h ,t ,
2 c1 d1 − c3 d2 2 2
(10.8-4d)
in which, equations (10.8-4c) and (10.8-4d) related to y = −h/2 is valid only along
the crack surface, namely, only for x a − h/2 at y = 0, in which the crack
terminates. From the last condition of (10.6-5), at y = 0 and the ahead of the crack,
uy = 0, wy = 0. To satisfy this condition the displacements uy and wy at y = −h/2
is approximated by
⎧
⎪ h h
⎪
⎨ uy x, − , t = −uy x, , t ,
2 2
(10.8-5)
⎪
⎪ h h
⎩ wy x, − , t = −wy x, , t .
2 2
ux h h ux h h ux h h
− ,− ,t = ,− ,t + − , ,t
uy 2 2 uy uy
2 2 2 2
ux 3h h u h 3h
−0.5 ,− ,t + x − , ,t ,
uy 2 2 uy 2 2
wx h h w h h wx h h
− ,− ,t = x ,− ,t + − , ,t
wy 2 2 wy wy
2 2 2 2 (10.8-6)
wx 3h h wx h 3h
−0.5 ,− ,t + − , ,t .
wy 2 2 wy 2 2
Similar expressions are used for deriving the displacement components at (−h/2, H +
h/2), (L + h/2, H + h/2)and(L + h/2, −h/2).
By following relevant stability criterion of the scheme the computation is always
stable and achieves high precision. Discussions on this aspect are omitted here due
to space limitation.
References
[1] Lubensky T C , Ramaswamy S and Joner J. Hydrodynamics of icosahedral quasicrys-
tals. Phys Rev B, 1985, 32(11): 7444–7452
[2] Socolar J E S, Lubensky T C and Steinhardt P J. Phonons, phasons and dislocations
in quasicrystals. Phys Rev B, 1986, 34(5): 3345–3360
[3] Bak P. Phenomenological theory of icosahedral in commensurate (quasiperiodic)order
in Mn-Al alloys. Phys Rev Lett, 1985, 54(14): 1517–1519
[4] Bak P. Symmetry, stability and elastic properties of icosahedral in commensurate
crystals. Phys Rev B, 1985, 32(9): 5764–5772
[5] Ding D H, Yang W G, Hu C Z et al. Generalized elasticity theory of quasicrystals.
Phys Rev B, 1993, 48(10): 7003–7010
[6] Hu C Z, Wang R H, Ding D H. Symmetry groups, physical property tensors, elasticity
and dislocations in quasicrystals. Reports on Progress in Physics, 2000, 63(1): 1–39
[7] Fan T Y, Li X F, Sun Y F. A moving screw dislocation in one-dimensional hexagonal
quasicrystal. Acta Physica Sinica(Overseas Edition), 1999, 8(3): 288–295
[8] Fan T Y. A study on special heat of one-dimensional hexagonal quasicrystals. J
Phys: Condens Matter, 1999, 11(45): L 513–L 517
[9] Fan T Y, Mai Y W. Partition function and state equation of point group 12mm
two-dimensional quasicrystals. Euro Phys J B, 2003, 31(1): 25–27
[10] Fan T Y, Mai Y W. Elasticity theory, fracture mechanics and some relevant thermal
properties of quasicrystalline materials. Appl Mech Rev, 2004, 57(5): 325–344
[11] Li C L, Liu Y Y. Phason-strain influences on low-temperature specific heat of the
decagonal Al-Ni-Co quasicrystal. Chin Phys Lett, 2001, 18(4): 570–572
226 Chapter 10 Dynamics of elasticity and defects of quasicrystals
In Chapters 7∼9 we frequently used the complex variable function method to solve
problems of elasticity of quasicrystals, and many exact analytic solutions were ob-
tained by the method. In those chapters we only provided the results, and the
underlying principle and details of the method could not be discussed. Considering
the relative new feature and particular effect of the method, it is helpful to attempt a
further discussion in depth. Of course this may lead to a slight repetition of relevant
content of Chapters 7∼9.
It is well-known that the so-called complex potential method in the classical
elasticity is effective, in general, mainly for solving harmonic and biharmonic partial
differential equations, for these equations the solutions can be expressed by analytic
√
functions of single complex variable z = x + iy, i = −1. In addition, the quasi-
biharmonic partial differential equations can be solved by analytic functions of two
complex variables such as z1 = x + α1 y, z2 = x + α2 y, in which α1 , α2 are com-
plex constants. The study of elasticity of quasicrystals has led to discovery of some
multi-harmonic and multi-quasiharmonic equations, which cover a quite wide range
of partial differential equations appearing in the field to date and have been intro-
duced in Chapters 5∼9. The discussion on the complex analysis for these equations
is significant. We know that the Muskhelishvili complex potentials method for the
classical isotropic plane elasticity[1] , which solves mainly the biharmonic equation,
made great contributions for a quite wide range of fields in science and engineer-
ing. And the complex potential method developed by Lekhnitzkii[2] for the classical
anisotropic plane elasticity which solves mainly the quasi-biharmonic equation. The
present formulation and solutions of the complex analysis for, e.g. quadruple and
sextuple harmonic equations and quadruple-quasiharmonic equation, are at least a
new development of the complex potential method of the classical elasticity. Though
the new method is used to solve elasticity problems of quasicrystals at present, it
may be extended into other disciplines of science and engineering in future.
At first we simply review the complex potential method for harmonic and bi-
harmonic equations, which does not belong to a new innovation from point of view
of methodology, then focus on those for quadruple and sextuple harmonic equa-
tions and quadruple quasiharmonic equation, and with discussions in detail pre-
senting their new features from the angle of elasticity as well as complex potential
method.
∇2 ∇2 F = 0, ∇2 ∇2 G = 0. (11.2-1)
where φ1 (z), ψ1 (z), π1 (z) and χ1 (z) are any analytic functions of complex variable
√
z = x + iy, (i = −1). For these kinds of biharmonic equations Muskhelishvili[1]
11.3 The complex variable function method of quadruple harmonic equations and... 231
developed systematical complex variable function method, reader can find some
details in the well known monograph, we need not discuss them any more.
Because it is relatively simpler for the case of point groups 5m and 10mm, which
belong to the special case of point group 5, 5 and point groups 10, 10, we here discuss
only the final governing equation of plane elasticity of pentagonal of point groups 5,
5 and decagonal quasicrystals of point groups 10, 10
∇2 ∇2 ∇2 ∇2 G = 0, (11.3-1)
where G(x, y) is the stress potential function. The solution of equation (11.3-1) is
1 2 1 3
G = 2Re g1 (z) + z̄g2 (z) + z̄ g3 (z) + z̄ g4 (z) , (11.3-2)
2 6
The Section 8.4 shows, from fundamental solution (11.3-2) one can find the complex
representation of the stresses as below:
232 Chapter 11 Complex variable function method for elasticity of quasicrystals
⎧
⎪
⎪ σxx = −32c1 Re(Ω (z) − 2g4 (z)),
⎪
⎪
⎪
⎪ σyy = 32c1 Re(Ω (z) + 2g4 (z)),
⎪
⎪
⎪
⎪
⎪
⎪
⎨ σxy = σyx = 32c1 ImΩ (z),
Hxx = 32R1 Re(Θ (z) − Ω (z)) − 32R2 Im(Θ (z) − Ω (z)), (11.3-3)
⎪
⎪
⎪ H = −32R Im(Θ (z) + Ω (z)) − 32R Re(Θ (z) + Ω (z)),
⎪
⎪
⎪ xy 1 2
⎪
⎪
⎪
⎪ Hyx = −32R1 Im(Θ (z) − Ω (z)) − 32R2 Re(Θ (z) − Ω (z)),
⎪
⎪
⎩
Hyy = −32R1 Re(Θ (z) + Ω (z)) + 32R2 Im(Θ (z) + Ω (z),
where ⎧
⎨ (IV ) (IV ) 1 (IV )
Θ(z) = g2 (z) + z̄g3 (z) + z̄ 2 g4 (z),
2 (11.3-4)
⎩ Ω (z) = g (IV ) (z) + z̄g (IV ) (z),
3 4
in which the prime, two prime, three prime and superscript (IV ) denote the first to
fourth order derivatives of gi (z) to variable z, in addition Θ (z) = dΘ(z)/dz, it is
evident that Θ(z) and Ω (z) are not analytic functions.
By some derivation from (11.3-3) we have the complex representation of the
displacements such as
ux + iuy = 32(4c1 c2 − c3 − c1 c4 )g4 (z) − 32(c1 c4 − c3 )(g3 (z) + zg4 (z)), (11.3-5)
32(R1 − iR2 )
wx + iwy = Θ(z) (11.3-6)
K1 − K 2
with constants
c
c = M (K1 + K2 ) − 2(R12 + R22 ), c1 = + M,
K1 − K2
c + (L + M )(K1 + K2 ) 2
R + R22
K1 + K 2 (11.3-7)
c2 = , c3 = 1 , c4 = .
4(L + M )c c c
In the following we consider only the stress boundary value problem, i.e., at the
boundary curve Lt the tractions (Tx , Ty ) and generalized tractions (hx , hy ) are given,
and there are the stress boundary conditions such as
From (11.3-8) and after some derivation the phonon stress boundary condition
can be reduced to the equivalent form
i
g4 (z) + g3 (z) + zg4 (z) = (Tx + iTy )ds, z ∈ Lt . (11.3-10)
32c1
σyy − σxx + 2iσxy = 64c1 Ω (z) = 64c1 [h3 (z) + z̄h4 (z)], (11.3-14)
By inspection, the state of phonon and phason stresses is not altered if replacing
Consider now the case when the region S, occupied by the quasicrystal, is multiply
connected. In general the region is bounded by several simple closed contours s1 ,
s2 , · · · , sm , sm+1 , the last of these contours is to contain all the others, depicted in
Fig. 11.3-1, i.e., a plate with holes. We assume that the contours do not intersect
themselves and have no points in common. Sometimes we call s1 , s2 , · · · , sm as inner
boundaries and sm+1 as outer boundary of the region. It is evident that the points
z1 , z2 , · · · , zm are fixed points in the holes, but located out of the material.
Similar to the discussion of the classical elasticity theory (refer to [1]), we can obtain
m
h4 (z) = Ak ln(z − zk ) + h4∗ (z), (11.3-22)
k=1
m
m
h4 (z) = Ak z ln(z − zk ) + γk ln(z − zk ) + h4∗ (z), (11.3-23)
k=1 k=1
m
h3 (z) = γk ln(z − zk ) + h3∗ (z), (11.3-24)
k=1
recalling zk denote fixed points outside the region S, h3∗ (z), h4∗ (z) are holomorphic
(analytic and single-valued, refer to the Major Appendix of this book) in S, Ak real
constants, γk , γk complex constants.
11.3 The complex variable function method of quadruple harmonic equations and... 235
ux + iuy = 32(4c1 c2 − c3 − c1 c4 )h4 (z) − 32(c1 c4 − c3 )(h3 (z) + zh4 (z)). (11.3-26)
[ux+iuy ]sk = 2πi{[32(4c1 c2−c3−c1 c4 )+32(c1 c4−c3 )]Ak z+32(4c1 c2−c3−c1 c4 )γk+γk (z)},
(11.3-27)
in which [ ]Sk denotes the increase undergone by the expression in brackets for one
anti-clockwise circuit of the contour sk . Hence it is necessary and sufficient for the
single-valuedness of phonon displacements that in the formulas (11.3-22)∼(11.3-25)
32(R1 − iR2 )
[wx + iwy ]sk = (−2πi)γk . (11.3-29)
K1 − K 2
Hence it is necessary and sufficient for the single-valuedness of phason displacements
if
γk = 0. (11.3-30)
It will now be shown shat the quantities γk , γk may be very simply expressed in
terms of Xk , Yk , where (Xk , Yk ) denotes the resultant vector of the external stresses,
exerted on the contour sk . By (11.3-10), applying it to the contour sk , one has
with
Xk = Tx ds, Yk = Ty ds.
sk sk
In the present case the normal vector n must be directed outwards with respect
to the region Sk . Consequently, the contour sk must be traversed in the clockwise
direction. Taking this fact into consideration, one obtains
i
−2πi(γk − γk ) = (Xk + iYk ). (11.3-32)
32c1
236 Chapter 11 Complex variable function method for elasticity of quasicrystals
Ak = 0,
where
1 4c1 c2 − c3 − c1 c4
d1 = , d2 = −
64c1 π[32(4c1 c2 − c3 − c1 c4 ) + 1] 2c1 π[32(4c1 c2 − c3 − c1 c4 ) + 1]
(11.3-34)
and which are independent of the suffix k. So that
⎧
⎪
⎪ m
⎪
⎪ h (z) = d (Xk + iYk ) ln(z − zk ) + h4∗ (z),
⎪
⎪
4 1
⎨ k=1
m
(11.3-35)
⎪
⎪ h3 (z) = d 2 (Xk − iYk ) ln(z − zk ) + h3∗ (z),
⎪
⎪
⎪
⎪
⎩ h (z) = h k=1 (z).
2 2∗
We can conclude that the complex functions h2 (z), h3 (z), h4 (z) must be expressed
by formulas (11.3-35) to assure the single-valuedness of stresses and displacements,
where h2∗ (z), h3∗ (z), h4∗ (z) are holomorphic in S.
From the point of view of application in Chapter 9, this chapter and Chapters 14,
15 the consideration of infinite regions is likewise of major interest. We assume that
the contour sm+1 goes to infinity.
Since equations (11.3-13) and (11.3-14) are similar to the classical elasticity the-
ory, we have
m
m
X= Xk , Y = Yk ,
k=1 k=1
h03 (z), h04 (z) are functions, holomorphic in region S, including the point at infinity,i.e,
for sufficiently large |z| they may be expanded into series of the form
a1 a2 a1 a2
h04 (z) = a0 + + 2 + ··· , h03 (z) = a0 + + 2 + ··· . (11.3-37)
z z z z
11.3 The complex variable function method of quadruple harmonic equations and... 237
On the basis of (11.3-2) the state of stresses will not be altered by assuming
a0 = a0 = 0.
By the theorem of Laurent expansion, the function h2∗ (z) may be represented in
region S including point at infinity by the series
+∞
h2∗ (z) = cn z n . (11.3-38)
−∞
Substituting equations (11.3-36) and (11.3-38) into equation (11.3-16), one has
(Hxx − Hyy ) − i(H xy + Hyx )
+∞
d 2 1 2d1
= 2 × 32(R1 + R2 cn nz n−1 + z̄ − 2 + h03 (z) + z̄ 2 + h04 (z) ,
−∞
z 2 z3
(11.3-39)
hence it follows that for the stresses to remain finite as |z| → ∞, one must have
cn = 0, n 2.
It is obvious that the phonon and phason stresses will be bounded, if these conditions
are satisfied. Hence one has finally
⎧
⎨ h4 (z) = d1 (X + iY ) ln z + (B + iC)z + h04 (z),
h3 (z) = d2 (X − iY ) ln z + (B + iC )z + h03 (z), (11.3-40)
⎩
h2 (z) = (B + iC )z + h02 (z),
where B , C are unknown real constants to be determined, h02 (z) is function, holo-
morphic in region S, including the point at infinity, thus it has the form similar to
that of (11.3-37),
a a
h02 (z) = a0 + 1 + 22 + · · · . (11.3-41)
z z
We have assumed that a0 = a0 = 0 already, now further assume a0 = 0, i.e.,
If we constrain our discussion only for case of stress boundary value problems, then
the problems will be solved under boundary conditions (11.3-10) and (11.3-11). For
some complicated regions solutions of the problems cannot be directly obtained in
the physical plane (i.e., the z-plane). We must use a conformal mapping
z = ω(ζ) (11.3-43)
to transform the region studied in the plane onto interior of the unit circle γ in the
mapping plane (say e.g. ζ-plane).
Substituting (11.3-43) into (11.3-40) we have
⎧
⎨ h4 (z) = Φ4 (ζ) = d1 (X + iY ) ln ω(ζ) + Bω(ζ) + Φ40 (ζ),
h3 (z) = Φ3 (ζ) = d2 (X − iY ) ln ω(ζ) + (B + iC )ω(ζ) + Φ30 (ζ), (11.3-44)
⎩
h2 (z) = Φ2 (ζ) = (B + iC )ω(ζ) + Φ20 (ζ),
where
Φj (ζ) = hj [ω(ζ)], Φj0 (ζ) = h0j [ω(ζ)], j = 1, · · · , 4.
In addition
Φi (ζ)
hj (z) = .
ω (ζ)
At the mapping plane the boundary conditions (11.3-10) and (11.3-11) stand for
Φ (σ) i
Φ4 (σ) + Φ3 (σ) + ω(σ) 4 = (Tx + iTy )ds, (11.3-10 )
ω (σ) 32c1
(R2 − iR1 )Θ(σ) = i (hx + ihy )ds, (11.3-11 )
where σ = eiϕ represents the value of ζ at the unit circle (i.e., ρ = 1). From these
boundary value equations we can determine the unknown functions Φi (ζ)(i = 2, 3, 4).
Due to Φ1 (ζ) = 0, we now have three unknown functions Φj (ζ)(j = 2, 3, 4). Taking
conjugate of (11.3-10 ) yields
Φ (σ) i
Φ4 (σ) + Φ3 (σ) + ω(σ) 4 =− (Tx − iTy )ds. (11.3-10 )
ω (σ) 32c1
Substituting the first one of equations (11.3-4) into (11.3-11 ), then multiplying
dσ/2πi(σ − ζ) to both sides of (11.3-10 ) and (11.3-11 ) and integrating along the
unit circle γ leads to
11.3 The complex variable function method of quadruple harmonic equations and... 239
⎧
⎪
⎪ 1 Φ4 (σ)dσ 1 Φ3 (σ)dσ 1 ω(σ) Φ 4 (σ)dσ 1 1 tdσ
⎪
⎪ + + =
⎪
⎪ 2πi σ−ζ 2πi σ−ζ 2πi ω (σ) σ − ζ 32c1 2πi σ−ζ
⎪
⎪
γ γ γ γ
⎪
⎪
⎪
⎪ 1 Φ4 (σ)dσ 1 Φ3 (σ)dσ 1 ω(σ) Φ4 (σ)dσ 1 1 tdσ
⎪
⎪ + + =
⎪
⎪ 2πi σ−ζ 2πi σ−ζ 2πi
ω (σ) σ − ζ 32c1 2πi σ−ζ
⎪
⎪ γ γ γ γ
⎪
⎨
1 Φ2 (σ)dσ 1 ω(σ) Φ 3 (σ)dσ
+
⎪
⎪ 2πi γ σ − ζ 2πi γ ω (σ) σ − ζ
⎪
⎪
⎪
⎪ 2 2
⎪
⎪ 1 ω(σ) Φ4 (σ)dσ ω(σ) ω (σ) Φ4 (σ)dσ
⎪
⎪ + −
⎪
⎪ 2πi γ [ω (σ)]2 σ − ζ [ω (σ)]3 σ−ζ
⎪
⎪ γ
⎪
⎪
⎪
⎪ 1 1 hdσ
⎪
⎩ = ,
R1 − iR2 2πi γ σ − ζ
(11.3-45)
- - -
where t = i (Tx + iTy )ds, t = −i (Tx − iTy )ds, h = i (h1 + ih2 )ds in equations
(11.3-45), which are the function equations to determine complex potentials Φj (ζ)
which are analytic in the interior of the unit circle γ, and satisfy boundary value
conditions (11.3-45) at the unit circle.
The boundary conditions can be expressed by equations (11.3-10) and (11.3-11), for
simplicity, we assume hx = hy = 0. Thus
⎧
⎪
⎨ i (Tx + iTy )ds = i (−p cos(n, x) − ip cos(n, y))ds = −pz = −pω(σ),
⎪
⎩ i (hx + ihy )ds = 0.
(11.3-47)
In addition in this case in formulas (11.3-44)
X = Y = 0,
B = 0, B = C = 0, B = C = 0, (11.3-48)
so Φi (ζ) = Φi0 (ζ), but in the following we omit the superscript of the functions Φi0 (ζ)
for simplicity.
The conformal mapping is
1
z = ω(ζ) = R0 + mζ , (11.3-49)
ζ
to transform the region containing ellipse at the z-plane onto the interior of the
unit circle at the ζ-plane, refer to Fig.11.3-3, where ζ = ξ + iη = ρeiϕ and R0 =
a+b a−b
,m= .
2 a+b
Substituting (11.3-48) and (11.3-49) into the first two equations of function equation
(11.3-46), one obtains
11.3 The complex variable function method of quadruple harmonic equations and... 241
Fig.11.3-3 Conformal mapping from the exterior of the elliptic hole at z-plane onto the
interior of the unit circle at ζ-plane
⎧
⎪
⎪ pR0 (1 + m2 )ζ
⎨ Φ3 (ζ) = , (11.3-50a)
32c1 mζ 2 − 1
⎪
⎪ pR0 (11.3-50b)
⎩ Φ4 (ζ) = − mζ.
32c1
Substitution of
2
ω(σ) σ2 + m ω(σ) ω (σ) 2σ(σ 2 + m)2
= σ , = −
ω (σ) mσ 2 − 1 ω (σ)3 (mσ 2 − 1)3
Utilizing the above mentioned results, the phonon and phason stresses can be deter-
mined at the ζ-plane, so the displacements.
242 Chapter 11 Complex variable function method for elasticity of quasicrystals
If take m = 1, from (11.3-50) we can obtain solution of the Griffith crack, in particu-
lar the explicit solution at z-plane can be explored by taking inversion ζ = ω −1 (z) =
#
z/a − z 2 /a2 − 1 (as m = 1) into the relevant formulas. The concrete results are
given in Section 8.4, which are omitted here.
In this case
p
X = Y = 0, Tx = Ty = 0, B= , B = C = 0,
64c1 (11.3-51)
B = C = 0, t = t = h = 0,
The problem is shown in Fig.11.3-4, but the conformal mapping is different to that
of example 1 and example 2.
Fig. 11.3-4 Infinite plane with an elliptic hole subjected to a distributed pressure at a
part of surface of the hole and its conformal mapping at ζ-plane, in which the region at
z-plane is mapped onto the exterior of the unit circle at ζ-plane.
11.4 Complex variable function method for sextuple harmonic equation and... 243
⎪
⎪ 1 p (mζ 2
+ 1) σ2 − σ1
⎪
⎪ · · 2 × 2Re z2 ·
⎪
⎪
+
− 2 − ζ)(σ1 − ζ)
⎪
⎪ 32c
1
2πi
(ζ m)
(σ 2
⎪
⎪
⎪
⎪ m (σ2 − ζ)(σ1 − ζ) + (σ2 + σ1 − 2ζ)(σ2 − σ1 )
⎪
⎪ + z 2 − R 0 ζ − ·
⎪
⎪ ζ (σ2 − ζ)(σ1 − ζ)
⎪
⎪
⎪
⎪ (mζ 2
+ 1)(ζ 2
+ m) 1
⎪
⎪ ip d1 (z1 − z2 − z1 + z2 )
⎪
⎪ (ζ − m)
2 3 ζ − σ1
⎪
⎪
⎪
⎪
⎪
⎪ 1 1 1
⎩ + (d2 − d1 )(z1 − z2 ) 2 + +
ζ ζ (ζ − σ1 )2
(11.3-53)
where
m m
z1 = R σ1 + , z2 = R σ2 + .
σ1 σ2
⎪
⎪ Hyx + Hxy + i(Hxx − Hyy ) = −480c2 c5 f6 (z) − 4c1 c2 RΘ (z),
⎪
⎪
⎪
⎪
⎪
⎪ Hyz + iHxz = 48c2 c6 Γ (z) − 4c2 R2 (2K2 − K1 )Ω (z),
⎪
⎪
⎪
⎪ 24R2
⎪
⎪
⎩ Hzz = c2 c3 ImΓ (z),
(μ + λ)
(11.4-3)
where
⎧
⎪ Ψ (z) = f5 (z) + 5z̄ f6 (z),
⎪
⎪
⎪
⎪
⎪
⎪ Γ (z) = f4 (z) + 4z̄ f5 (z) + 10z̄ 2 f6 (z),
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ Ω (z) = f3 (z) + 3z̄ f4 (z) + 6z̄ 2 f5 (z) + 10z̄ 3 f6 (z),
⎪
⎪
⎪
⎪
⎪
⎪ (IV)
Θ(z) = f2 (z) + 2z̄ f3 (z) + 3z̄ 2 f4 (z) + 4z̄ 3 f5 (z) + 5z̄ 4 f6
⎪
⎪
(z),
⎪
⎪
⎪
⎪
⎪
⎪ R(2K2 − K1 )(μK1 + μK2 − 3R2 )
⎨ c1 = ,
2(μK1 − 2R2 ) (11.4-4)
⎪
⎪ 1
⎪
⎪ c3 = K2 (μK2 − R2 ) − R(2K2 − K1 ),
⎪
⎪ R
⎪
⎪
⎪
⎪ (μK2 − R2 )2
⎪
⎪
⎪
⎪ c2 = μ(K1 − K2 ) − R2 − ,
⎪
⎪ μK1 − 2R2
⎪
⎪
⎪
⎪ 1 μK1 − 2R2
⎪
⎪ c4 = c1 R + c3 K1 +
⎪
⎪ 2 λ+μ
⎪
⎪
⎪
⎪ μK2 − R2
⎪
⎩ c5 = 2c4 − c1 R, c6 = (2K2 − K1 )R2 − 4c4 .
μK1 − 2R2
11.4 Complex variable function method for sextuple harmonic equation and... 245
In the above expressions,the function g1 (z)is not used and to be assumed g1 (z) = 0
so f1 (z) = 0 for simplicity, we have introduced the following new symbols:
(9) (8) (7)
g2 (z) = f2 (z), g3 (z) = f3 (z), g4 (z) = f4 (z),
(6) (5)
(11.4-5)
g5 (z) = f5 (z), g6 (z) = f6 (z),
(n)
where gi denotes n-th derivative with the argument z. Similar to manipulation in
the previous section, the complex representations of displacement components can
be written as follows (here we have omitted the rigid body displacements):
⎧
⎪
⎪ 2c2
⎪
⎪ u y + iu x = −6c 3 R + c 7 Γ (z) − 2c3 c7 RΩ (z),
⎪
⎪ μ+λ
⎪
⎪
⎪
⎪ 4
⎪
⎪ uz = (240c10 Imf6 (z)
⎪
⎪ μ(K + K 2 ) − 3R
2
⎪
⎪
1
⎪
⎨ +c1 c2 R2 Im(Θ(z) − 2Ω (z) + 6Γ (z) − 24Ψ (z))),
(11.4-6)
⎪
⎪ R
⎪
⎪ wy + iwx = − c1 (μK1 − 2R2 ) (24c9 Ψ (z) − c8 Θ(z)),
⎪
⎪
⎪
⎪
⎪
⎪ 4(μK2 − R2 )
⎪
⎪ wz =
⎪
⎪ (240c10 Imf6 (z)
⎪
⎪ (K1 − 2K2 )R(μ(K1 + K2 ) − 3R2 )
⎪
⎩
+c1 c2 R2 Im(Θ(z) − 2Ω (z) + 6Γ (z) − 24Ψ (z))),
in which
c2 K1 + 2c1 R
c7 = , c8 = c1 c3 R(μ(K1 − K2 ) − R2 ),
μK1 − 2R2
(μK2 − R2 )2
c9 = c8 + 2c3 c4 c2 − , c10 = c1 c3 R2 − c4 (c3 R − c2 K1 ).
μK1 − 2R2
(11.4-7)
Utilizing equation (11.4-3) and the first two formulas of equations (11.4-8) ,one has
−4c2 c3 R[3(f4 (z)+4z̄f5 (z)+10z̄ 2 f6 (z))−(f3 (z)+3zf4 (z)+6z 2 f5 (z)+10z 3 f6 (z))]
-
= i (Tx + iTy )ds, z ∈ L.
(11.4-10)
Taking conjugate on both sides of equation (11.4-10) yields
−4c2 c3 R[3(f4 (z)+4zf5 (z)+10z 2 f6 (z))−(f3 (z)+3z̄f4 (z)+6z 2 f5 (z)+10z̄ 3 f6 (z)
-
= −i (Tx − iTy )ds, z ∈ L.
(11.4-11)
Similarly, from equations (11.4-3) and the first two formulas of (11.4-9) ,one obtains
48c2 (2c4 − c1 R)Ψ (z) + 2c1 c2 RΘ(z) = i (hx + ihy )ds, z ∈ L. (11.4-12)
Furthermore we assume
Tz = hz = 0, (11.4-13)
for simplicity and by the third equations in (11.4-8) and (11.4-9) and the formulas
of (11.4-3) and (11.4-13),one has
f6 (z)+f6 (z) = 0,
4c11 Re[f5 (z)+5z̄f6 (z)]+(2K2 −K1 )RRe[f4 (z)+4z̄f5 (z)+10z̄ 2 f6 (z)+20f6 (z)] = 0,
z∈L (11.4-14)
in which
4c4 (μK2 − R2 )
c11 = (2K2 − K1 )R − . (11.4-15)
(μK1 − 2R2 )R
As we have shown in the previous section, complex analytic functions (i.e., the
complex potentials) must be determined by boundary value equations, and the dis-
cussion is as below.
Consider now the case when the region S, occupied by the body, is multiply con-
nected,see Fig 11.3-1.
Since the stress must be single-valued and equation (11.4-16)
we know that f6 (z)is holomorphic and hence single-valued in the region with contour
sm+1 , so the complex function can be expressed as
z
f6 (z) = f6 (z)dz + constant, (11.4-20)
z0
c1 (σyy − σxx − 2iσxy ) + ic2 [Hxy − Hyx + i(Hxx + Hyy )] = −192ic2 c3 c4 Ψ (z)
shows that f5 (z) is holomorphic in the region enclosed by contour sm+1 , so one has
fi∗∗ (z)(i = 2, · · · , 6) are functions, holomorphic outside sm+1 , not including the
point at infinity. By the theorem of Laurent expansion, the function h2∗ (z) may be
represented outside sm+1 by the series
+∞
fi∗∗ (z) = ain z n (i = 2, · · · , 6.) (11.4-27)
n=−∞
Substituting the first equation of (11.4-26) and (11.4-27) into equation (11.4-16),one
has $m ∞
%
1
σzy − iσzx = −960c3 c4 bk + na6n z n−1 , (11.4-28)
z −∞
k=1
hence it follows that for the stress to remain finite as |z| → ∞, one must have
a6n = 0, n 2. (11.4-29)
Similarly, from equations (11.4-15)∼(11.4-18), to make the stress be bounded, the
following conditions are also be satisfied
So we can obtain the expressions of the complex function fi (z)(i = 2, · · · , 6) for the
stress to remain finite as |z| → ∞,for example
m
f6 (z) = bk ln z + (B + iC)z + f60 (z), (11.4-31)
k=1
where B, C are unknown real constants to be determined, f60 (z) is function, holomor-
phic outside sm+1 , including the point at infinity. The determination of unknown
contants B, C etc is similar to that given in subsection 11.3.4, but the details are
omitted here due to limitation of the space.
z = ω(ζ) (11.4-32)
is used to transform the region at z- plane onto the interior of the unit circle γ at
ζ- plane. Under the mapping the unknown functions fi (z) become
Substituting (11.4-32) and (11.4-33) into the first relation of boundary condition
(11.4-14) and multiplying dσ/2π(σ−ζ) to both sides of the equation, then integrating
along the unit circle γ and |ζ| < 1 yields
1 Φ6 (σ) 1 Φ6 (σ)
dσ + dσ = 0.
2πi σ−ζ 2πi σ−ζ
γ γ
This shows
Φ6 (ζ) = 0 (11.4-34)
according to the Cauchy integral formula.
Substitution of (11.4-32)∼(11.4-34) into boundary conditions (11.4-10)∼
(11.4-12) and the second one of conditions (11.4-14) and taking manipulation sim-
ilarly to above leads to boundary value equations for determining the unknown
functions Φi (ζ)(i = 2, · · · , 5) at ζ - plane, i.e.,
3 Φ4 (σ) 4 ω(σ) Φ5 (σ) 1 Φ3 (σ)
dσ + (σ) σ − ζ
dσ − dσ
2πi σ − ζ 2πi ω 2πi γ σ−ζ
γ
γ
1 ω(σ) Φ4 (σ) 1 [ω(σ)]2 Φ5 (σ) [ω(σ)]2 ω (σ) dσ
−3 dσ − 6 2 − 3 Φ5 (σ)
2πi γ ω (σ) σ − ζ 2πi γ ω (σ) ω (σ) σ −ζ
1 1 t
= dσ,
4c2 c3 2πi γ σ − ζ
(11.4-35)
3 Φ4 (σ) 4 ω(σ) Φ4 (σ) 1 Φ3 (σ)
dσ + dσ − dσ
2πi γ σ − ζ 2πi γ ω (σ) σ − ζ 2πi γ σ − ζ
2 2
1 ω(σ) Φ3 (σ) 1 ω(σ) Φ5 (σ) ω(σ) ω (σ)Φ5 (σ) dσ
−3 dσ − 6 i −
2πi γ ω (σ) σ − ζ 2π γ [ω (σ)]2 [ω (σ)]3 σ−ζ
1 1 t
= dσ,
4c2 c3 R 2πi γ σ − ζ
(11.4-36)
1 Φ2 (σ) 1 ω(σ) Φ3 (σ)
dσ + 2 dσ+
2πi γ σ − ζ 2πi γ ω (σ) σ − ζ
2 2 2
1 ω(σ) Φ4 (σ) ω(σ) ω (σ)Φ4 (σ) dσ 1 ω(σ) Φ 5 (σ)
3 − + 4
2πi γ [ω (σ)]2 [ω (σ)]3 σ−ζ 2πi γ [ω (σ)]3
3 3 3
ω(σ) ω (σ)Φ5 (σ) ω(σ) ω (σ)Φ5 (σ) ω(σ) ω (σ)Φ5 (σ) dσ
−3 4
+ 3 −
[ω (σ)] [ω (σ)]5 [ω (σ)]4 σ−ζ
1 h
= dσ,
2πi σ−ζ
(11.4-37)
4c11 Φ5 (σ) (2K2 − K1 )R Φ4 (σ) ω(σ) Φ5 (σ)
dσ + +4 dσ = 0, (11.4-38)
2πi γ σ − ζ 2πi γ σ−ζ ω (σ) σ − ζ
11.4 Complex variable function method for sextuple harmonic equation and... 251
in which t = i (Tx + iTy )ds, t = −i (Tx − iTy )ds, h = i (h1 + ih2 )ds. For
given configuration and applied stresses we can obtain the solution by solving these
function equations.
to transform the exterior of the ellipse at the z-plane onto the interior of the unit
circle γ at the ζ− plane, in which
ω(σ) σ2 + m
= σ
ω (σ) mσ 2 − 1
and
ζ2 + m ζ2 + m
ζ Φ 5 (ζ) = ζ (α1 + 2α2 ζ + 3α3 ζ 2 + · · ·)
mζ 2 − 1 mζ 2 − 1
1 Φ4 (σ)
dσ = Φ4 (ζ)
2πi σ−ς
γ
are analytic in |ζ| < 1 and continuous in the unit circle γ, by means of Cauchy
integral formula, from equation(11.4-34), we have
1 σ 2 + m Φ5 (σ) ζ2 + m
σ dσ = ζ Φ (ζ)
2πi mσ 2 − 1 σ − ς mζ 2 − 1 5
γ
Substituting
2ζ(mζ 2 + 1)2 2ζ(mζ 2 + 1)2 α2 α3
Φ 5 (ζ) = α1 + 2 +3 2 +···
(ζ 2 − m)3 (ζ 2 − m)3 ζ ζ
are analytic in |ζ| > 1 and continuous in the unit circle γ, by means of Cauchy
integral formula and analytic extension of the complex variable function theory,
from equation(11.4-34), we obtain
1 Φ3 (σ) 1 ω(σ) Φ4 (σ)
dσ = 0, dσ = 0
2πi σ−ς 2πi ω (σ) σ − ς
γ γ
1 ω(σ) Φ5 (σ) ω(σ)2 ω (σ)
2
dσ
2 − 3 Φ5 (σ) =0
2πi ω (σ) ω (σ) σ −ς
γ
Substituting the above results into equation(11.4-34), with the help of equation
(11.4-38),one has
R0 pζ(ζ 2 + m)(m3 ζ 2 + 1)
Φ2 (ζ) = −
2c2 c3 R (mζ 2 − 1)3
(2K2 −K1 )R0 pmζ 3 (ζ 2 +m)[m2 ζ 6 −(m3 +4m)ζ 4 +(2m4 +4m2 +5)ζ 2 +m]
+
2c2 c3 C 11 (mζ 2 −1)5
R0 pζ(m2 + 1) (2K2 − K1 )R0 pmζ 3 (ζ 2 + m)(mζ 2 − m2 − 2)
Φ3 (ζ) =− −
4c2 c3 R (mζ 2 − 1) 12c2 c3 C 11 (mζ 2 − 1)3
(11.4-41)
The elliptic notch problem is solved. The solution of the Griffith crack subjected
to a uniform pressure can be obtained corresponding to the case m = 1, R0 = a/2 of
the above solution. The solution of crack can be expressed explicitly in the z−plane,
the concrete results refer to Section 9.7 in Chapter 9.
11.5 Complex analysis and solution of quadruple quasiharmonic equation 253
In some sense, the complex potential approach above mentioned is a new devel-
opment of Muskhelishvili approach of the classical plane elasticity.
Apart from the development to the complex potential theory and method, the
conformal mapping technique has also been extended. According to the monograph
[1], application of the conformal mapping is limited within the rational function
class. But it can be extended into the transcendental function class, and some exact
analytic solutions for more complicated cracked configurations are achieved, see e.g.
Chapter 8.
The method is effective not only for solving elasticity problems but also for
solving some simple plasticity problems, see e.g. Fan and Fan[10] , Li and Fan[11,12]
and Fan and Tang[13] .
References
[1] Muskhelishvili N I. Some Basic Problems of the Mathematical Theory of Elasticity.
Groningen: P Noordhoff., 1956
[2] Lekhnitskii S G. Theory of Elasticity of an Anisotropic Body. Holden-Day. San
Francisco, 1963
[3] Liu G T, Fan T Y, Guo R P. Governing equations and general solutions of plane
elasticity of one-dimensional quasicrystals. Int J Solid and Structures, 2004, 41(14):
3949−3959
[4] Liu G T. The complex variable function method of the elastic theory of quasicrys-
tals and defects and auxiliary equation method for solving some nonlinear evolution
equations. Dissertation. Beijing Institute of Technology, 2004 (in Chinese)
[5] Liu G T, Fan T Y. The complex method of the plane elasticity in 2D quasicrystals
point group 10mm ten-fold rotation symmetry notch problems. Science in China E,
2003, 46(3): 326−336
[6] Li L H, Fan T Y. Final governing equation of plane elasticity of icosahedral quasicrystals–
stress potential method. Chin Phys Lett, 2006, 24(9): 2519−2521
[7] Li L H, Fan T Y. Complex function method for solving notch problem of point group
10 and 10 two-dimensional quasicrystal based on the stress potential function, J.
Phys.: Condens. Matter, 2006, 18(47): 10631−10641
[8] Li L H, Fan T Y. Complex function method for notch problem of plane elasticity of
icosahedral quasicrystals, Science in China, G, 2008, 51(6): 773−780
[9] Li W, Fan T Y. Study on elastic analysis of crack problem of two-dimensional decago-
nal quasicrystals of point group 10,10. Mod Phys Lett, B, 2009, 23(16), 1989–1999
[10] Fan T Y, Fan L. Plastic fracture of quasicrystals. Phil Mag, 2008, 88(4): 321−335
[11] Li W, Fan T Y. Study on plastic analysis of crack in decagonal Al-Ni-Co quasicrystals.
Mod. Phys. Lett, B, in press, 2009
References 255
From Chapter 5 to Chapter 11, we developed mainly analytic theories and meth-
ods. The elasticity problems of quasicrystals were reduced to boundary value or
initial-boundary value problems of some partial differential equations to solve, in
which some analytic methods were used. For some boundary value problems, these
methods are extremely powerful, even capable of obtaining exact analytic solutions.
In Chapter 14 and Major Appendix we will further develop the analytic method in
studying some problems such as nonlinear deformation etc.
However there are limitations themselves for these analytic methods. In general,
they can only treat some problems with simple configurations and simple boundary
conditions; while for more complicated problems, the methods cannot display their
power, one must use numerical methods.
In Chapter 10 we have applied the finite difference method to determining the
numerical solutions for some elasto-/hydro-dynamics problems of quasicrystals. In
this chapter we derive a variational principle of elasticity of quasicrystals, which
is the foundation of the subsequent development of the finite element method for
problems of quasicrystals. Discretization is the main feature of finite difference
method and finite element method. It has been shown that solutions obtained by
these two schemes can approach to the analytic solutions as the discreted mesh
(or element) tends to infinitesimal. In the other hand, in contrast to the analytic
solutions (or classical solutions), finite element solutions belong to one kind of so-
called weak solutions (or generalized solutions) according to the modern theory
of partial differential equations. The further mathematical principle on the weak
solutions of elasticity of quasicrystals will be developed in Chapter 13, which will
help us to understand that the finite element method is one of important tools to
implement weak solutions from other angles.
where
Cijkl = Lδij δkl + M (δik δjl + δil δjk ), (12.1-4)
C11 − C12
L = C12 , M= , (12.1-5)
2
Kijkl = (K1 − K2 − K3 )δik δjl + K2 δij δkl + K3 δil δ
(12.1-6)
+2(K2 + K3 )(δi1 δj2 δk1 δl2 + δi2 δj1 δk2 δl1 ),
Rijkl = R1 (δi1 − δ)(δij δkl − δik δjl + δil δjk )
+R2 [(1 − δij )δkl + δij (δi1 − δi2 )(δk1 δl2 − δk2 δl1 )] i, j, k, l = 1, 2.
(12.1-7)
and the definitions about C12 , C12 (orL, M ), K1 , K2 , K3 , R1 and R2 have been dis-
cussed in detail in Chapters 6∼8.
For the points group 10mm, R1 > 0, R2 = 0, K1 > 0, K2 > 0, K2 + K3 = 0; for
the point group 8mm, R1 > 0, R2 = 0, K1 > 0, K2 > 0, K3 > 0; for the point group
10,10 R1 > 0, R2 > 0, K1 > 0, K2 > 0, K2 + K3 = 0; for the point group 12mm,
R1 = 0, R2 = 0,K1 > 0,K2 > 0,K3 > 0 (but some measured results show K2 < 0,
this does not influence the extreme value of energy functional, only influences the
minimum value of the functional in the proof for theorem of variational principle of
elasticity of quasicrystals, in Section 12.2).
The strains in above relevant formulas are denoted by
12.2 Generalized variational principle for static elasticity of quasicrystals 259
1 ∂ui ∂uj ∂wi
εij = + , wij = (12.1-8)
2 ∂xj ∂xi ∂xj
∂σij ∂Hij
+ fi = 0, + gi = 0. (12.1-9)
∂xj ∂xj
The above formulas hold in any interior point of region Ω , and at boundary Γt , the
stresses satisfy the boundary conditions
σij nj = Ti ,
(x1 , x2 ) ∈ Γt , (12.1-10)
Hij nj = hi
ui = ūi ,
(x1 , x2 ) ∈ Γu , (12.1-11)
wi = w̄i ,
where Ti is the traction vector, hi the generalized traction vector at boundary Γt , ūi
and w̄i the given displacements at boundary Γu , ni the unit outward normal vector
at any point of the boundary and Γ = Γu + Γt .
In the following only the static problems are studied, and the initial value condi-
tions will not be concerned. For the dynamic problems, the initial value conditions
must be used, which have been discussed in Chapter 10.
and displacement boundary conditions (12.1-11), let the energy functional of qua-
sicrystals
Π = F dΩ + (fi ui + gi wi )dΩ + (Ti ui + hi wi )dΓ (12.2-1)
Ω Ω Γt
to take a minimum value, then they will be the solution satisfying the equilibrium
equations (12.1-9) and the stress boundary conditions (12.1-10), in which F is defined
by
⎧ εij wij
⎪
⎪ F =
⎪
⎪ σ ij dε ij + Hij dwij = Fu + Fw + Fuw ,
⎪
⎪ 0 0
⎪
⎪
⎪
⎨ 1
Fu = Cijkl εkl ,
2 (12.2-2)
⎪
⎪
⎪
⎪ 1
⎪
⎪ Fw = 2 Kijkl wkl ,
⎪
⎪
⎪
⎩
Fuw = Rijkl εij wkl
where
∂F ∂F
σij = , Hij = , (12.2-4)
∂εij ∂wij
∂F ∂ ∂F ∂ ∂F
δεij dΩ = δui dΩ − δui dΩ
Ω ∂εij ∂xj ∂εij ∂xj ∂εij
Ω Ω
∂F ∂ ∂F
= nj δui dΓ − δui dΩ .
Γu +Γt ∂εij Ω ∂x j ∂ε ij
Because the displacements are given at the boundary Γu , at which δui = δ ūi = 0,
the above formula has been reduced to
∂F ∂F ∂ ∂F
δεij dΩ = nj δui dΓ − δui dΩ . (12.2-5)
Ω ∂εij Γt ∂εij Ω ∂xj ∂εij
Due to wij = ∂wi /∂xj , a similar analysis to what adopted just above gives rise to
∂F ∂F ∂ ∂F
δwij dΩ = nj δwi dΓ − δui dΩ . (12.2-6)
Ω ∂wij Γt ∂wij Ω ∂xj ∂wij
Since δui and δwi are of arbitrary and independent variation at region Ω and bound-
ary Γ there follows
∂ ∂F ∂ ∂F
+ fi = 0, + gi = 0, (x1 , x2 ) ∈ Ω ,
∂xj ∂εij ∂xj ∂wij
∂F ∂F
nj − hi = 0, nj − Ti = 0, (x1 , x2 ) ∈ Γt .
∂wij ∂εij
Substituting (12.2-4) into the above formulas yields the equilibrium equations and
stress boundary conditions. This shows that ui and wi satisfying equations of defor-
mation geometry, stress-strain relations and displacement boundary conditions and
making energy functional to have minimum value, should be the solution satisfying
the equilibrium equations and stress boundary conditions.
(2) Sufficiency
The sufficiency of the conditions given by the theorem means that, if among
all sets of the displacements ui and wi satisfy relations of deformation geometry,
displacement boundary, the one that satisfies the equilibrium equations and stress
boundary conditions makes in (12.2-1) an extremum. In what follows, we show that
Π is minimum.
262 Chapter 12 Variational principle of elasticity of quasicrystals, numerical analysis...
Suppose that quantities ui , εij , wi and wij obey the stress-strain relations
(12.1-3) and satisfy the displacement boundary conditions (12.1-13). Let
∗
εij = εij + δεij , u∗i = ui + δui ,
∗ (12.2-8)
wij = wij + δwij , wi∗ = wi + δwi
in which
∂F ∂F
δΠ = δεij + δwij − fi δui − gi δwi dΩ − (Ti δui + hi δwi )dΓ ,
Ω ∂εij ∂wij Γt
(12.2-12)
2 1 ∂2F 1 ∂2F ∂2F
δ Π = δεij δεkl + δwij δwkl + δεij δwkl dΩ .
Ω 2 ∂εij ∂εkl 2 ∂wij ∂wkl ∂εij ∂wkl
(12.2-13)
Applying the Green formula to (12.2-12) leads to
δ Π = 0.
Because ui and wi (through the corresponding σij and Hij ) satisfy the equilibrium
equations and stress boundary conditions, this means that the energy functional
takes extreme value.
According to the discussion in Ref.[2], we do some extension, i.e., L + M >
0, M > 0, K1 > 0, K2 > 0, K3 > 0, M K1 > R2 ,then elasticity of two-dimensional
quasicrystals presents stability, in the case the stress-strain elastic matrix should be
positive definite, so
12.3 Finite element method 263
δ 2 Π > 0.
This ensures that δ Π = 0 takes not only an extreme value, but also the minimum
value of the energy functional.
But recent experimental results and some simulation show that there may be
K2 < 0, this does not influence the energy taking extreme value, but the extreme
value may not be minimum value.
Collaborating variational principle with theory of functional analysis, we can
prove that existence, uniqueness and stability of solution of the boundary value
problem (12.1-3),(12.1-8)∼(12.1-11).This is concerned not only with the numerical
implementation, but also with other topics, the detailed discussion of which will be
given in Chapter 13 or see Guo and Fan[4] .
The variational principle can be extended to dynamic case, in which it is needed
only to extend the energy functional (12.2-1) to be as follows:
..
Π = F dΩ + [(fi − ρ ui )ui + (gi − κẇi )wi )dΩ + (Ti ui + hi wi )dΓ , (12.2-14)
Ω Ω Γt
where the meaning of ρ and κ can be found in Chapter 10. From (12.2-14) we
can obtain the corresponding variational equation similar to (12.2-3), but which is
equivalent to the elasto-/hydro-dynamic equations and related boundary and initial
conditions of quasicrystals. The further discussion is omitted here.
(m) (m )
(x1 , x2 ) ∈ Γ (mm ) , (12.3-1)
wi = ui ,
where Γ (mm ) represents the interface of element m and element m ;
(3) At the displacement boundary Γu ,the displacements should satisfy the con-
ditions such as
ui = ūi ,
(x1 , x2 ) ∈ Γu . (12.3-2)
wi = w̄i ,
264 Chapter 12 Variational principle of elasticity of quasicrystals, numerical analysis...
By satisfying these conditions, the discrete form of the energy functional Π is given
by
N
(m) (m) (m) (m) (m) (m) (m) (m)
Π∗ = (F (m) −fi ui −gi wi )dΩ − (Ti ui +hi wi )dΓ .
(m)
m=1 Ω (m) Γt
(12.3-3)
Taking variation to (12.3-3) and letting it to be zero, by using the method similar
to that adopted in the previous section, we can show that within an element the
equilibrium equations are satisfied by
⎧ (m)
⎪
⎪ ∂ ∂F (m)
⎪
⎨ + fi = 0,
∂xj ∂εij
(m) (x1 , x2 ) ∈ Ωm , (12.3-4)
⎪
⎪ ∂ ∂F (m)
⎪
⎩ + gi = 0,
∂xj ∂wij
N
∗ 1 (m) T (m) (m) (m) T (m)
Π = {u } {K }{u } − {u } {R } , (12.3-6)
m=1
2
where column vector {u(m) } represents the element displacement (including phonons
and phasons) vector, {u(m) }T is its transpose, the row vector, and {K (m) } the
element stiffness matrix , i.e.,
T
[K (m) ] = [B] [D][B]Δdx1 dx2 , (12.3-7)
Ωm
in which, Δ denotes the thickness of the element and Δ=1.0 for plane strain case
(i.e., the thick body, and we do not consider plane stress case, i.e., the thin body),
is the element strain matrix , n is the number of nodes with differential operator
12.3 Finite element method 265
matrix, ⎡ ⎤
∂
⎢ 0 0 0 ⎥
⎢ ∂x ⎥
⎢ ∂
0 ⎥
⎢ 0 0 ⎥
⎢ ∂y ⎥
⎢ ⎥
⎢ ∂ ∂ ⎥
⎢ 0 0 ⎥
⎢ ∂y ∂x ⎥
⎢ ⎥
[L] = ⎢
⎢ 0 0
∂
0 ⎥
⎥ (12.3-9)
⎢ ∂x ⎥
⎢ ∂ ⎥
⎢ 0 0 0 ⎥
⎢ ∂y ⎥
⎢ ⎥
⎢ ∂ ⎥
⎢ 0 0 0 ⎥
⎢ ∂y ⎥
⎣ ∂ ⎦
0 0 0
∂x
[Bi ] = [L][Ii ] (12.3-9 )
and [I] = [I1 , I2 , · · · , In ] is the element interpolation matrix, n is the number of
nodes as pointed out just before, here n = 4, i.e.,
⎡ ⎤
Ii 0 0 0
⎢ 0 Ii 0 0 ⎥
[Ii ] = ⎢ ⎥
⎣ 0 0 Ii 0 ⎦ , i = 1, 2, 3, 4. (12.3-10)
0 0 0 Ii
where Ii is the i-th node interpolation function. In (12.3-7) matrix [D] denotes the
plane elastic constants matrix of two-dimensional quasicrystals, namely
⎡ ⎤
L + 2M L 0 R1 R1 R2 −R2
⎢ L L + 2M 0 −R1 −R1 −R2 R2 ⎥
⎢ ⎥
⎢ 0 0 2M 2R2 2R2 −2R1 2R1 ⎥
⎢ ⎥
⎢
[D] = ⎢ 1R R 2R K K 0 0 ⎥.
1 2 1 2 ⎥
⎢ R1 −R 2R K K 0 0 ⎥
⎢ 1 2 2 1 ⎥
⎣ R2 −R2 2R1 0 0 K1 + K2 + K3 K3 ⎦
−R2 R2 2R1 0 0 K3 K1 + K 2 + K 3
(12.3-11)
The strain vector and stress vector corresponding to (12.3-11) are arranged by
in which
266 Chapter 12 Variational principle of elasticity of quasicrystals, numerical analysis...
T
f
= [f1x , f1y , g2x , g2y , · · · , f4x , f4y , g4x , g4y ],
g
T
T
= [T1x , T1y , h1x , h1y , · · · , T4x , T4y , h4x , h4y ]
h
f T
and , are the transposes of the row vectors, x = x1 , y = x2 .
g h
After assemblage of the element matrix, (12.3-6) may be rewritten in terms of
matrix form
1
Π ∗ = {u}T [K]{u} − {u}T {R}. (12.3-13)
2
Carrying out the operation
δΠ ∗ = 0, (12.3-14)
where [K], {u} and {R} are the global stiffness matrix, displacement and external
force vectors generated by assemblage of element stiffness matrixes, element dis-
placement and external force vectors, and {u} is solution vector unknown to be
determined. By solving algebraic equation set (12.3-15) the displacement vector can
be determined, and so do the strains and stresses. The stresses of quasicrystals at the
Gauss integration points can be evaluated through the element node displacements
in accordance with the following version
(m)
(m)
(m)
(m)
σ u u u
= [S] = [D][B] = [D][L][I] ,
H m m m
(12.3-16)
where
[S] = [D][B] = [D][L][I]
The phonon and phason stresses, displacements and free energy (strain energy den-
sity) are ploted versus radial distance in Figs.12.4-2∼12.4-6 for different material
constants, respectively.
In the Fig.12.4-3, it seems that the phason stress is one order of magnitude less
than the phonon stress, but this by no means shows that the phason stress are less
than phonon stress in general. One of the reasons resulting in the lower phason
stresses compared with the phonon stresses lies in the boundary conditions (12.4-1).
Instead of (12.4-1), if we take
then the computation shows that the phason stresses are greater than the phonon
stresses. This suggests that the boundary conditions play an important role in the
analysis. Due to the lack of the knowledge especially lack of the measuring data
268 Chapter 12 Variational principle of elasticity of quasicrystals, numerical analysis...
on the generalized tractions hi , one often supposes that the tractions to be zero,
this influences the correctness of the computational results, especially to those of
the phason field. Even so, the computaion reveals the effects of the phason and
phonon-phason coupling.
Fig.12.4-6 shows that the variation of the free energy (or the strain energy den-
sity) in the cylinder as function of coordinates, versus radial distance for different
material constants.
The mesh configuration of the phason field presents some changes before and
after deformation, which are shown in Fig. 12.4-7, this explores the anisotropic
270 Chapter 12 Variational principle of elasticity of quasicrystals, numerical analysis...
Fig. 12.4-7 The mesh of phason field before and after deformation:
K1 /M = 0.8, K2 /M = 0.6, K3 /M = 0.4, R1 /M = 0.1[9]
Fig. 12.4-8 FEM numerical results of normalized phonon stress near the crack tip versus
normalized distance
1. K1 /M = 0.8, K2 /M = 0.6, K3 /M = 0.4, R/M = 0.0
2. K1 /M = 0.7, K2 /M = 0.6, K3 /M = 0.4, R/M = 0.1
272 Chapter 12 Variational principle of elasticity of quasicrystals, numerical analysis...
References
[1] Hu H C. Variational Principles of Elasticity, Beijing. Science Press, 1981 (in Chinese)
[2] De P, Pelcorits R A. Linear elasticity theory of pentagonal quasicrystals. Phys Rev
B, 1987 35(28): 8609−8620
[3] Ding D H, Yang W G, Hu C Z et al. Generalized elasticity theory of quasicrystals.
Phy Rev B, 1993 48(16). 7003−7010
[4] Li X F, Fan T Y, Sun Y F. A decagonal quasicrystal with a Griffith crack. Phil Mag
A, 1999, 79(8): 1943−1952
[5] Guo L H, Fan T Y. Solvability of boundary value problems of elasticity of three-
dimensional quasicrystal. Appl Math Mech, English Edition 2007, 28(8): 1061−1070
[6] Guo Y C, Fan T Y. Mode II Griffith crack in decagonal quasicrystals. Appl Math
Mech, English Edition, 2001, 22(10), 1311−1317
[7] Fan T Y and Mai Y W. Elasticity theory, fracture mechanics and some relevant
thermal properties of quasicrystalline materials, Appl. Mech. Rev, 2004, 57(5),
325−344.
[8] Zhou W M, Fan T Y, Plane elasticity and crack problem of octagonal quasicrystals.
Chin. Phys. 2001, 10 (6), 277−284
[9] Wu X F, Numerical analysis of power-law hardening material and quasicrystalline
material with defects, Dissertation, Beijing Institute of Technology, 1998 (in Chinese)
Chapter 13
Some mathematical principles on solutions
of elasticity of quasicrystals
Starting from Chapter 4 we studied the elasticity of quasicrystals and gave many
solutions. A further discussion on some common features of the solutions will be
offered in this chapter.
should be the solution of the problem too. The “differences” should satisfy zero
boundary conditions. The work done by the external forces is
0= (Δf ⊕ Δg) · (Δu ⊕ Δw)dΩ
Ω
+ (ΔT ⊕ Δh) · (Δu ⊕ Δw)dΓ = 2 ΔU dΩ , (13.1-5)
Γ Ω
ΔU 0, (13.1-6)
ΔU = 0. (13.1-7)
Based on the positive definite property of ΔU , Δεij and Δwij must be zero,
(1) (2) (1) (2)
apart from only rigid displacements, so that εij = εij , wij = wij , etc.
At the same time, at boundary Γu , one has
Δu = 0, Δw = 0.
⎪ 2 2 2 (13.2-4)
⎪
⎪ ∂w ∂w ∂w
⎪ 2
⎪ wi 1,Ω =
i
+
i
+
i
dΩ .
⎩ ∂x ∂y ∂z
Ω
This norm is suitable only for case of homogeneous boundary conditions (13.2-2),
otherwise the norm will not be in this form.
With the norm in (13.2-4), denote the space defining ui (x, y, z) and wi (x, y, z) is
by H (Ω ). If introduce inner product
⎧ $ (1) (2) %
⎪ (1) (2) (1) (2)
⎪
⎪ (1) (2) ∂ui ∂ui ∂ui ∂ui ∂ui ∂ui
⎪
⎪ (u , ui ) = + + dΩ ,
⎨ i Ω ∂x ∂x ∂y ∂y ∂z ∂z
⎪ $ (1) (2) (1) (2) (1) (2)
% (13.2-5)
⎪
⎪ ∂wi ∂wi ∂wi ∂wi ∂wi ∂wi
⎪
⎪
(1) (2)
⎩ (wi , wi ) = ∂x ∂x
+
∂y ∂y
+
∂z ∂z
dΩ ,
Ω
276 Chapter 13 Some mathematical principles on solutions of elasticity of quasicrystals
and the inner product at space V (13.2-5), then V is also a Hilbert space and
V ⊂ H (Ω ). (13.2-7)
σij (X) = σji (X) = Cijkl εkl (X) + Rijkl wkl (X),
Hij = Rklij εkl (X) + Kijkl wkl (X),
Theorem 1 The variational problem associated with the boundary value prob-
lem (13.2-1)∼(13.2-3) is to obtain X ∈ H so that X
in which
(1) Exists solution, and the number of solutions does not exceed one;
(2) If there is a solution of problem (13.2-16), which must be the solution of
problem (13.2-12), and vise versa;
278 Chapter 13 Some mathematical principles on solutions of elasticity of quasicrystals
σij = Cijkl εkl + Rijkl wkl , Hij = Kijkl wkl + Rklij εkl , (13.3-3)
⎡ ⎤
∂1 0 0
⎡ ⎤ ⎢ 0 ∂2 0 ⎥
∂1 0 0 ⎢ ⎥
⎢ ⎥
⎢ ⎥ ⎢ 0 0 ∂3 ⎥
⎡ ⎤ ⎢ 0 ∂2 0 ⎥ ⎢ ⎥
⎢ ⎥ ⎢ ∂2 0 0 ⎥
˜ (1)
∂ 0 ⎢
⎢ 0 0 ∂3 ⎥
⎥
⎢
⎢
⎥
⎥
˜ =⎣
∂ ⎦, ˜ (1)
∂ =⎢ ⎥, ˜ (2)
∂ =⎢ 0 ∂3 0 ⎥,
˜ (2)
∂ ⎢ ∂2 ∂1 0 ⎥ ⎢ ⎥
0 ⎢ ⎥ ⎢ 0 0 ∂1 ⎥
⎢ 0 ∂3 ∂2 ⎥ ⎢ ⎥
⎣ ⎦ ⎢ ⎥
⎢ ∂3 0 0 ⎥
∂3 0 ∂1 ⎢ ⎥
⎣ 0 ∂1 0 ⎦
0 0 ∂2
∂
∂i = ,
∂xi
T T
Ũ , F̃ , ∂ T , σ̃ T , ε̃T denote the transpose of Ũ , F̃ , ∂, σ̃, ε̃, σ i = (σi1 , σi2 , σi3 ) the
i-th row of matrix (σij )3×3 , H i = (Hi1 , Hi2 , Hi3 ) the i-th row of matrix (Hij )3×3 .
So that, (13.3-2) can be rewritten as the matrix form
ε̃ = ∂ Ũ . (13.3-2 )
T
∂σ1j ∂σ2j ∂σ3j ∂H1j ∂H2j ∂H3j
Note that = ∂ T σ̃, then (13.3.1)
∂xj ∂xj ∂xj ∂xj ∂xj ∂xj
can be rewritten by
. = ρU
∂ T σ̃ + F .̈ . (13.3-1 )
Putting
⎡ ⎤ ⎡ ⎤
··· C11ij ··· C1111 C1122 C1133 C1112 C1123 C1131
⎢ ··· ··· ⎥ ⎢ ⎥
⎢ C22ij ⎥ ⎢ C2211 C2222 C2233 C2212 C2223 C2231 ⎥
⎢ ⎥ ⎢ ⎥
⎢ ··· C33ij ··· ⎥ ⎢ C3311 C3322 C3333 C3312 C3323 C3331 ⎥
C=⎢
⎢
⎥
⎥ =⎢
⎢
⎥
⎥ ,
⎢ ··· C12ij ··· ⎥ ⎢ C1211 C1222 C1233 C1212 C1223 C1231 ⎥
⎢ ⎥ ⎢ ⎥
⎣ ··· C23ij ··· ⎦ ⎣ C2311 C2322 C2333 C2312 C2323 C2331 ⎦
··· C31ij ··· 6×6
C3111 C3122 C3133 C3112 C3123 C3131 6×6
280 Chapter 13 Some mathematical principles on solutions of elasticity of quasicrystals
⎡ ⎤
··· K11ij ···
⎢ ··· K22ij ··· ⎥
⎢ ⎥
⎢ ··· K33ij ··· ⎥
⎢ ⎥
⎢ ··· K12ij ··· ⎥
⎢ ⎥
K =⎢
⎢ ··· K23ij ··· ⎥
⎥
⎢ ··· K31ij ··· ⎥
⎢ ⎥
⎢ ··· K13ij ··· ⎥
⎢ ⎥
⎣ ··· K21ij ··· ⎦
··· K32ij ···
⎡ ⎤
K1111 K1122 K1133 K1112 K1123 K1131 K1113 K1121 K1132
⎢ K2211 K2222 K2233 K2212 K2223 K2231 K2213 K2221 K2232 ⎥
⎢ ⎥
⎢ K3311 K3322 K3333 K3312 K3323 K3331 K3313 K3321 K3332 ⎥
⎢ ⎥
⎢ K1211 K1222 K1233 K1212 K1223 K1231 K1213 K1221 K1232 ⎥
⎢ ⎥
=⎢
⎢ K2311 K2322 K2333 K2312 K2323 K2331 K2313 K2321 K2332 ⎥
⎥ ,
⎢ K3111 K3122 K3133 K3112 K3123 K3131 K3113 K3121 K3132 ⎥
⎢ ⎥
⎢ K1311 K1322 K1333 K1312 K1323 K1331 K1313 K1321 K1332 ⎥
⎢ ⎥
⎣ K2111 K2122 K2133 K2112 K2123 K2131 K2113 K2121 K2132 ⎦
K3211 K3222 K3233 K3212 K3223 K3231 K3213 K3221 K3232 9×9
⎡ ⎤
··· R11ij ···
⎢ ··· R22ij ··· ⎥
⎢ ⎥
⎢ ··· R33ij ··· ⎥
R =⎢ ⎥
⎢ ··· R12ij ··· ⎥
⎢ ⎥
⎣ ··· R23ij ··· ⎦
··· R31ij ··· 6×9
⎡ ⎤
R1111 R1122 R1133 R1112 R1123 R1131 R1113 R1121 R1132
⎢ R2211 R2222 R2233 R2212 R2223 R2231 R2213 R2221 R2232 ⎥
⎢ ⎥
⎢ R3311 R3322 R3333 R3312 R3323 R3331 R3313 R3321 R2232 ⎥
=⎢ ⎥ ,
⎢ R1211 R1222 R1233 R1212 R1223 R1231 R1213 R1221 R1232 ⎥
⎢ ⎥
⎣ R2311 R2322 R2333 R2312 R2323 R2331 R2313 R2321 R2332 ⎦
R3111 R3122 R3133 R3112 R3123 R3131 R3113 R3121 R3132 6×9
then
C R
D = (dij )15×15 = .
RT K
Here the order of index i, j of C is the same with those of the phonon strain tensor,
the order of index i, j of K, R is the same with those of phason strain tensor, and
RT is the transpose of R. From the above expressions one can find that due to the
symmetry of C and K (see e.g. (4.4-3) and (4.4-5)), the matrix D is symmetric.
The generalized Hooke’s law (13.3-3) can be rewritten as
σ̃ = Dε̃, (13.3-3 )
13.3 Matrix expression of elasticity of three-dimensional quasicrystals 281
Put
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
a1 (x) c1 (x) u01
⎢ a2 (x) ⎥ ⎢ c2 (x) ⎥ ⎢ u02 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢ a3 (x) ⎥ ⎢ c3 (x) ⎥ 0 ⎢ u03 ⎥
A(x) = ⎢ ⎥ , B(x) = ⎢ ⎥ , Ũ = ⎢ ⎥ ,
⎢ b1 (x) ⎥ ⎢ d1 (x) ⎥ ⎢ w10 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎣ b2 (x) ⎦ ⎣ d2 (x) ⎦ ⎣ w20 ⎦
b3 (x) 6×1
d3 (x) 6×1
w30 6×1
⎡ ⎤
T1
⎢ T2 ⎥ ⎡ ⎤
⎢ ⎥ ˜ (1)
⎢ T3 ⎥ ∂ 0
σ̃ = ⎢
0 ⎥ , ˜n = ⎣
∂
n ⎦,
⎢ ⎥
⎢ h1 ⎥ 0 ˜ (2)
∂
⎣ h2 ⎦ n
h3 6×1
⎡ ⎤
cos(n, x1 ) 0 0
⎢ 0 cos(n, x2 ) 0 ⎥
⎢ ⎥
⎢ ⎥
˜ (1)
∂ =⎢
0 0 cos(n, x3 ) ⎥,
n ⎢ cos(n, x2 ) cos(n, x1 ) 0 ⎥
⎢ ⎥
⎣ 0 cos(n, x3 ) cos(n, x2 ) ⎦
cos(n, x3 ) 0 cos(n, x1 )
⎡ ⎤
cos(n, x1 ) 0 0
⎢ 0 cos(n, x2 ) 0 ⎥
⎢ ⎥
⎢ 0 0 cos(n, x3 ) ⎥
⎢ ⎥
⎢ cos(n, x2 ) 0 0 ⎥
⎢ ⎥
˜ (2)
∂ =⎢ 0 cos(n, x3 ) 0 ⎥,
n ⎢ ⎥
⎢ 0 0 cos(n, x1 ) ⎥
⎢ ⎥
⎢ cos(n, x3 ) 0 0 ⎥
⎢ ⎥
⎣ 0 cos(n, x1 ) 0 ⎦
0 0 cos(n, x2 )
˜ (1) , ∂
˜n , ∂
where ∂ ˜ (2) are obtained from the differential operator matrices ∂ ˜ (1) , ∂
˜ ,∂ ˜ (2)
n n
through a replacement ∂ i by cos(n, xi ).
Equation (13.3-4) can be rewritten as
0
Ũ (x, t) = Ũ , x ∈ (∂Ω )u . (13.3-4 )
Considering the similarity of the left-handside of (13.3-5) with the first term of (13.3-
1), then (13.3-5) can be rewritten as ∂ ˜ T σ̃ = σ̃ 0 , x ∈ (∂Ω )σ , In addition by using
n
(13.3-2 ) and (13.3-3 ) there is
˜ T D∂
∂ ˜ Ũ = σ̃ 0 , x ∈ (∂Ω )σ . (13.3-5 )
n
282 Chapter 13 Some mathematical principles on solutions of elasticity of quasicrystals
where
∂Ω = (∂Ω )u + (∂Ω )σ .
From (13.3-3) and Chapter 4 known σij = σji , in terms of Gauss formula, (13.3-2 )
and (13.3-3 ), there exists
˜ T D∂ ˜ Ũ · ηdx = − ∂σij 1 ∂Hij 2
−∂ ηi + ηi dx
∂xj ∂xj
Ω Ω
∂ 1 2 ∂ηi1 ∂ηi2
= − (σij ηi + Hij ηi ) − σij + hij dx
∂xj ∂xj ∂xj
Ω
∂η 1 ∂η 2
= − (σij ηi1 +Hij ηi2 )nj dS + σij i +Hij i dx
∂Ω ∂xj ∂xj
Ω
$ %
1 ∂ηi1 1 ∂ηj1 ∂η 2
= σij + σij + Hij i dx
2 ∂xj 2 ∂xi ∂xj
Ω
= [(σij (Ũ )εij (η 1 ) + Hij (Ũ )wij (η 2 )]dx
Ω
T
= σ̃(Ũ ) · ε̃(η)dx = ˜
(∂η) ˜ Ũ dx,
D∂ (13.4-2)
Ω Ω
13.5 The uniqueness of weak solution 283
Because C0∞ is dense in H01 (Ω ),(13.4-3) holds for ∀η(x) ∈ (H01 (Ω ))6 .
In contrast, if Ũ (x) ∈ (C (Ω̄ ))6 , and (13.4-3) is valid ∀η(x) ∈ (H01 (Ω ))6 , we can
2
do derivation in counter order of the above procedure, and find (13.4-1) through the
fundamental lemma of variational method[2] . So that we have
Definition Assume F̃ ∈ (L2 (Ω ))6 , if Ũ (x) ∈ (H01 (Ω ))6 , and (13.4-3) holds
∀η(x) ∈ (H01 (Ω ))6 , then say Ũ (x) being the weak solution (or generalized solution)
of the boundary value problem
−∂ ˜ T D∂˜ Ũ (x) = F̃ (x), x ∈ Ω , t > 0 (13.3-8)
Ũ (x) |∂Ω = 0. (13.3-9 )
3
n 2
∂vi
|v|21 = dx,
i=1 k=1 Ω ∂xk
n 3
n 2
2 2 ∂vi ∂v ∂v1 ∂v2 ∂vn
where |v| = vi2 , |vx | = = , , ,··· , . Obvi-
i=1 i=1 k=1
∂xk ∂xk ∂xk ∂xk ∂xk
ously, the Note 1 holds for vector function v, too.
Lemma(Korn inequality[4,5] ) Assume Ω is a bounded region with boundary
∂Ω of sufficient smooth in Rn , and ∀v = (v1 , v2 , · · · , vn ) ∈ H01 (Ω ), there is
n 2
∂vi ∂vk 2
+ dx c1 v1 ,
Ω ∂x k ∂x i
i,k=1
284 Chapter 13 Some mathematical principles on solutions of elasticity of quasicrystals
where λ1 , λ2 are positive constants, then for any F̃ ∈ (L2 (Ω ))6 , displacement bound-
ary value problem (13.3-8), (13.3-9 ) exists unique weak solution (or generalized
solution).
T
Proof Put Ũ , η = (∂η) D∂ Ũ dx, then (13.4-3) can be rewritten as
Ω
At first we prove ·, · is a new inner product at (H01 (Ω ))6 . For this purpose it
needs to prove: Ũ , Ũ 0, and Ũ , Ũ = 0 ⇔ Ũ = 0, ∀Ũ ∈ (H01 (Ω ))6 .
In the following we give only an outline of the proof, the detail is omitted. In
addition Ũ , η in (13.5-1) is positive definite bilinear functional at H01 (Ω ), the proof
can be done from the Lax-Milgram theorem (see Section 13.2).
Due to the assumption, matrix D = (dij )15×15 being positive definite, matrix
D = (dij )15×15 and unit matrix I are in contract, i.e., there exists a reversible matrix
C such that
D = CTC
(note that here C is not the phonon elastic constant matrix). Then
T
Ũ , Ũ = ˜ ˜
(∂ Ũ ) D ∂ Ũ dx = ˜ Ũ )T (C T C)∂
(∂ ˜ Ũ dx
Ω Ω
T
= ˜ Ũ ) (C ∂
(C ∂ ˜ Ũ )dx 0,
Ω
.,U
U . = 0 ⇔ .U
(C ∂ . )T (C ∂
.U .U
. )dx = 0 ⇔ C ∂ . = 0.
Ω
.U
Because C is reversible, ∂ . = 0, i.e.,
∂ui ∂ui ∂uj ∂wi
= 0, + = 0 (i = j), = 0, i, j = 1, 2, 3.
∂xi ∂xj ∂xi ∂xj
∂wi
It follows = 0 that wi should be constant, besides Ũ |∂Ω = 0, and wi = 0 at
∂xj
the boundary. In similar analysis we find that ui = 0 at boundary. Thus Ũ = 0 at
boundary.
13.5 The uniqueness of weak solution 285
In this way we have proved ·, · is a new inner product at (H01 (Ω ))6 , the corre-
sponding norm is U . (1) = U .,U . 12 .
Secondly, at (H01 (Ω ))6 the new norm ·(1) is equivalent to the initial norm ·1 .
We are going to give the proof about this.
From the Cauchy’s inequality, the assumption of the theorem and Note 1, there
is
/ /2 15
/ / ˜ Ũ )i dij (∂
˜ Ũ )j dx λ2 ˜ Ũ )T (∂
˜ Ũ )dx
/Ũ / = (∂ (∂
(1)
Ω i,j=1 Ω
⎡ ⎤
3 2 3 2 3 2
⎢ ∂ui ∂ui ∂uj ∂wi ⎥
= λ2 ⎣ + + + ⎦ dx
i=1
∂xi i,j=1
∂xj ∂xi i,j=1
∂xj
Ω i<j
⎧ ⎫
⎪ 3 2 2 2 3 2 ⎪
⎨ ∂u
3
∂ui ∂uj ∂wi ⎬
i
λ2 +2 + + dx
⎪
⎩ i=1 ∂xi i,j=1
∂xj ∂xi ∂xj ⎪ ⎭
Ω i,j=1
i<j
3
2 2
∂ui ∂wi
2λ2 + dx
i,j=1
∂xj ∂xj
Ω
" "2 / /2
" " / /
= 2λ2 "Ũ " 2c /Ũ / .
1 1
In other hand, from the assumption of the theorem, the Korn inequality and
Note 1 there is
/ /2 15
/ / ˜ Ũ )T (∂
/Ũ / = (∂˜Ũ )i dij (∂
˜ Ũ )j dx λ1 (∂ ˜ Ũ )dx
(1)
Ω i,j=1 Ω
⎡ ⎤
3 2 3 2 3 2
⎢ ∂ui ∂ui ∂uj ∂wi ⎥
= λ1 ⎣ + + + ⎦ dx
i=1
∂xi i,j=1
∂xj ∂xi i,j=1
∂xj
Ω i<j
⎡ ⎤
3
2 3 2
⎣1 ∂ui ∂uj ∂wi ⎦
= λ1 + + dx
4 i,j=1 ∂xj ∂xi i,j=1
∂xj
Ω
2 3 2
1 ∂ui ∂wi
λ1 c2 dx + λ1 dx
4 ∂xj i,j=1 Ω
∂xj
Ω
" " / /2
1 " "2 / /
min λ1 c2 , λ1 "Ũ " c /Ũ / .
4 1 1
Consequently, we have proved the equivalency between the new norm ·(1) and the
initial norm ·1 .
286 Chapter 13 Some mathematical principles on solutions of elasticity of quasicrystals
Finally, for F̃ ∈ (L2 (Ω ))6 , by using Schwarz inequality and the fact that the
embedding H01 (Ω ) ⊂ → L2 (Ω ) is a compact embedding, we have
" "
" " / / / / / /
" " / /
" F̃ · ηdx " /F̃ / · η M /
/ F̃
/
/ · η M
/ /
1 /F̃ / · η1 , ∀η ∈ (H01 (Ω ))6 ,
" " 1
" "
Ω
i.e., η → F̃ · ηdx (∀η ∈ (H01 (Ω ))6 ) is a unique continuous linear functional
Ω
at (H01 (Ω ))6 .
Therefore, from Riesz theorem, there must be a unique y F̃ ∈ (H01 (Ω ))6 , such
that
F̃ · ηdx = y F̃ , η , ∀η ∈ (H01 (Ω ))6 .
Ω
Thus (13.5-1) is rewritten as
3 4
Ũ , η = y F̃ , η , ∀η ∈ (H01 (Ω ))6 .
This shows Ũ = y F̃ is the unique weak solution (generalized solution) of displace-
ment value problem (13.3-8), (13.3-9 ).
In the above proof, using Korn inequality is crucial (the second Korn inequality
will be used for the stress boundary value problem, but this discussion is not included
here).
References
[1] Fan T Y, Mai Y W. Elasticity theory, fracture mechanics and some relevant thermal
properties of quasicrystalline materials. Appl Mech Rev, 2004, 57(5): 325–344
References 287
From Chapter 4 to Chapter 13 we mainly discussed the elasticity and relevant prop-
erties of quasicrystals, which belong to linear regime both physically and mathemat-
ically. Their mathematical treatment is relatively easy though the calculations are
quite complex.
The current chapter is to give a simple description on deformation and fracture
of quasicrystals with nonlinear behaviour, considering the great difficulty in this
topic. For the conventional engineering materials including crystalline material, the
nonlinear behaviour means mainly plasticity. In the study on the classical plasticity
there are two different theories, one is the macroscopic plasticity theory, which is
based on some assumptions concluded from certain experimental data, and the other
is so-called crystal plasticity theory, which is based on the mechanism of motion of
dislocation, and in some extent can be seen as a “microscopic” theory. The difficulty
for quasicrystal plasticity lies in lack of both enough macro- and micro-data. At
present the macroscopic experiments have not, as yet, been properly undertaken.
Though there is some work on the mechanism in microscopy of the plasticity, the
data are very limited. This leads to the constitutive law of quasicrystals being
essentially unknown. Due to the reason the systematic mathematical analysis on
deformation and fracture for the material is not available so far.
In spite of these difficulties, study on plasticity of quasicrystals has aroused a
great deal of attention of researchers[1∼8] . But the analytic quantitative work may
be at an infant stage. Considering the readers’ interest and the development level, it
is beneficial to give a brief discussion on some simple problems of nonlinear behaviour
of the material with some simple models and by extending results in the study of
linear regime. Of course, these discussions are not complete, which may provide
some hints for further development of the area.
This chapter is arranged as follows. First, we discuss some experimental results
on the nonlinear deformation behaviour of quasicrystals, then describe a possible
plastic constitutive equation of the material. In view of the difficulty for setting
up the equations, we turn to introduce nonlinear elastic constitutive equations of
quasicrystals which are available at present though not equivalent to the plastic
constitutive equations. The Sections 14.4 and 14.5 may be seen as applications of
the macro-constitutive law in which some nonlinear solutions on quasicrystals are
presented. In Section 14.6 another version of the study based on the dislocation
model or “microscopic model” is exhibited, which achieves the same results given in
Sections 14.4 and 14.5.
variable that can be regarded as a reference stress representing the current mi-
crostructural state of the material and is used to accommodate the model to the
description of different materials or hardening mechanism[12] . Combining relevant
information, formula (14.1-1) can be used to well predict experimental curves, e.g.
recorded by Fig. 14.1-1. It should be pointed out, though there are some similar
forms to (14.1-1) in the classical plasticity, they are quite different substantively.
For example, the current parameters B, m and σ̂ are different from those appearing
in relevant formulas in the classical plasticity. Those parameters in the classical
plasticity were measured from pure macroscopic approach rather than dislocation
model, some detail about the latter can be found in Refs. [9-12], which, from the
angle of methodology, are different from those adopted in the classical plasticity. Un-
fortunately there is lack of the comprehensive macroscopic experimental data (e.g.
the data arising from multiaxial loading condition) for quasicrystal plasticity so far.
Fig. 14.1-2 (a) Stress-strain curves of decagonal b-Co decagonal Al-Ni-Co single
quasicrystals for different specimen orientations at strain rate of 10−5 /s, (b) Definition of
the specimen orientations A⊥ , A and A45 [9]
in which σY is the uniaxial yield limit of the material, then (14.2-1) represents the
initial yield surface. On the other hand if
Y = Y (h), (14.2-3)
in which
εkk = εxx + εyy + εzz , wkk = wxx + wyy + wzz , σkk = σxx + σyy + σzz ,
where σ0 , A and n are the material constants of the quasicrystals, which can be
(e)
measured through a uniaxial test, εeff represents the quantity at elastic deformation
stage and σ0 the uniaxial tensile yield stress.
In contrast to (14.2-4), the equations (14.2-5) can not describe deformation his-
tory, as they are substantively nonlinear elastic constitutive equations rather than
plastic constitutive equations related to quasicrystalline materials. However they
can describe plastic deformation in the case of proportional loading and no unload-
ing. It is evident either (14.2-4) or (14.2-5) belong to a supposed incremental plastic
constitutive law or total plastic constitutive law for quasicrystals. One cannot say
whether they are correct or not due to lack of enough experimental data.
If one has the constitutive equations (14.2-4) or (14.2-5) then by coupling the
equations of deformation geometry
1 ∂ui ∂uj ∂wi
εij = + , wij = (14.2-7)
2 ∂xj ∂xi ∂xj
and the equilibrium equations
∂σij ∂Hij
= 0, = 0, (14.2-8)
∂xj ∂xj
the basic framework of the theory of macro-plasticity of quasicrystals, in the sense
of incremental or total deformation, of quasicrystals can be set up.
At present there is lack of such data, so the equations (14.2-1), (14.2-2) and
equations (14.2-4) have not been established yet. With the same reason the equations
(14.2-5) have not been set up either. This is the major difficulty of macro-plasticity
theory currently. It is evident the possible theory is nonlinear, because the material
parameters are not constants any more, and the mechanical behaviour is dependent
with the history of deformation process in general. The solution is of course more
difficult than that for elastic deformation.
Due to relative simplicity of the equations (14.2-5), for some simple configu-
rations, e.g. anti-plane elasticity of one-dimensional hexagonal, three-dimensional
cubic and three-dimensional icosahedral quasicrystals, one can probe into plastic
analysis by using the proposed constitutive equations.
the following nonlinear elasticity and the total plasticity introduced by equations
(14.2-5). We here do not constrain the concrete form of relationship between stresses
and strains. The results obtained in the following may provide some useful hints for
further plastic analysis.
Consider the following nonlinear elastic constitutive relations i.e., define the free
energy (or strain energy density)
εij wij
F (εij , wij ) = σij dεij + Hij dwij , (14.3-1)
0 0
then there is
∂F ∂F
σij = , Hij = . (14.3-2)
∂εij ∂wij
So the formulas (14.3-1) and (14.3-2) can be seen as constitutive law for linear as well
as nonlinear elasticity of quasicrystals, which, in general, cannot describe the plastic
deformation. If there is proportional loading and without unloading the relationship
can give appropriate description of the plastic deformation.
In above formulas εij and wij are phonon and phason strain tensors given by
1 ∂ui ∂uj ∂wi
εij = + , wij = (14.3-3)
2 ∂xj ∂xi ∂xj
where ui and wi denote the phonon and phason displacement vectors, σij and Hij
the phonon and phason stress tensors, respectively, which satisfy the equilibrium
equations (if the body forces and generalized body forces are omitted)
∂σij ∂Hij
= 0, = 0. (14.3-4)
∂xj ∂xj
Fig. 14.4-1 Atomic cohesive force zone near the crack tip for anti-plane elasticity for
some quasicrystals
14.4 Nonlinear solutions based on simple models 297
in which
R3
F1 (ζ) = φ (ζ) + ψ (ζ),
C44
R3
F2 (ζ) = ψ (ζ) + φ (ζ)
K2
and
φ(ζ) = φ1 (t) = φ1 (ω(ζ)), ψ(ζ) = ψ1 (t) = ψ1 (ω(ζ)),
where
a+d
t = ω(ζ) = (ζ + ζ −1 )
2
represents the conformal mapping from t-plane (xy-plane) onto ζ-plane, under the
mapping the region of xy-plane is transformed onto the interior of the unit circle
at ζ-plane, and φ (ζ) and ψ (ζ) are derivatives of the functions to the new complex
variable ζ, and θ1 is the angle of point at the unit circle corresponding to the crack tip
(i.e., y = 0, x = a, and there is relation cos θ1 = a/(a + d) between the corresponding
points at t-plane and ζ-plane).
From the solution, the size of plastic zone d is determined as
πτ1
d = a sec −1 , (14.4-2)
2τc
πτ1
i.e., θ1 = and the crack tip tearing displacement is
2τc
4K2 τc a πτ1
δIII = ln sec . (14.4-3)
π(C44 K2 − R32 ) 2τc
It is obvious that the results are very simple and explicit, and the effects of phason to
plastic deformation and fracture are explored. If taking the tearing displacement as a
nonlinear fracture parameter, then a fracture criterion for one-dimensional hexagonal
quasicrystal is
δIII = δIIIC (14.4-4)
is suggested in which δIIIC is the critical value the crack tip tearing displacement,
measured by experiment, a material constant.
In the low stress level case, i.e., τ1 /τc 1, the deformation is linear elastic, in
which 2
πτ1 1 πτ1
sec =1+ + ···
2τc 2 2τc
if we remain the first two terms then
2 4
πτ1 1 πτ1 1 πτ1
ln sec = + + ···
2τc 2 2τc 12 2τc
298 Chapter 14 Nonlinear behaviour of quasicrystals
Fig. 14.4-3 Normalized crack tip opening displacement versus normalized stress and
comparison[27]
Based on the parameter δI given by (14.4-7) we can suggest the nonlinear fracture
criterion for pentalgonal and/or decagonal quasicrystals for the Mode I loading as
δI = δIC , (14.4-8)
in which δIC denotes critical value of the crack opening displacement, measured by
experiment, a material constant.
In linear elastic case, i.e., p/σc 1, through a similar analysis like in the previous
subsection the crack tip opening displacement reduces to
GI
δI = , (14.4-9)
σc
where GI is defined by (8.3-18), this gives the simple connection between these two
parameters, so the criterion (14.4-8) reduces to energy release rate criterion for linear
elastic case discussed in Chapter 8.
in which
(μK2 − R2 )2
2 1 μK1 − 2R2
c2 = μ(K1 − K2 ) − R − , c4 = c1 R + c2 K1 +
μK1 − 2R2 2 λ+μ
(14.4-11)
R(2K2 − K1 )(μK1 + μK2 − 3R2 )
with c1 = .
2(μK1 − 2R2 )
The variation of the normalized crack tip opening displacement versus the normalized
applied stress is shown in Fig. 14.4-4.
Icosahedral quasicrystal Al-Mn-Pd
Icosahedral quasicrystal Al-Cu-Li
Conventional crystal aluminum
δI/a
p/σc
where F is defined by (14.3-1) and ni the unit vector of outward normal at any point
of an arc in a quasicrystal, and at which there are
σij nj = Ti , Hij nj = hi ,
where Ti denotes the traction vector and hi the generalized traction vector, shown
in Fig. 14.5-1, respectively.
14.5.2 Relation between crack tip opening displavement and the gener-
alized Eshelby integral
By using the path-independency of E-integral (14.5-2), we take the integration path
Γ shown in Fig. 14.5-2 and let the path close to
the surface of the plastic zone as
close as possible. Now the integration path is , and along segment AB and
ACB
BC, i.e.,
dy = 0, T1 = 0, T 2 = σc , h2 = Hc
so that
B
∂u ∂wy
E= GdΓ = GdΓ = dx + Hc
σc dx
Γ ACB A ∂x ∂x
= σc [(uy )B − (uy )A ] + Hc [(wy )B − (wy )A ] ≈ σc [(uy )B − (uy )A ] = σc δI . (14.5-3)
Fig. 14.5-2 The integration path for evaluating crack tip opening displacement
This proves the relation between the generalized Eshelby integral and crack tip
opening displacement, it shows the E-integral presents an equivalency to the crack
tip opening displacement, where σc is the atomic cohesive force (or the plastic yield
strength in macroscopic sense) and the “generalized atomic cohesive force” of the
quasicrystal material. In microscopy the quantity is meaningful, but it has not
been measured macroscopically at present, the effect of which is omitted in formula
(14.5-3) by this reason. For Mode II and Mode III crack the proof is similar.
Since by the generalized BCS model and generalized DB model one can obtain
the same result on crack tip opening displacement, which is also as a direct result
of the generalized Eshelby integral for quasicrystals, we realize that the generalized
Eshelby energy-momentum tensor theory is the uniform physical basis of the two
models.
304 Chapter 14 Nonlinear behaviour of quasicrystals
E = Ec (14.5-4)
as fracture criterion, where Ec is the critical value of the integral, a material constant,
which can be measured through some conventional specimens, the discussion will be
given in Section 14.8 (the Appendix of Chapter 14).
Though the measurement of the critical value of the generalized Eshelby integral
may be easier, but the evaluation of values of E-integral for plastic deformation of
quasicrystals is very difficult. Therefore the implementation of the criterion (14.5-
4) is not so convenient in practice. Instead, people can use fracture criterion of
crack tip opening displacement considering the equivalency between the crack tip
opening displacement δ(δI , δII , δIII ) and E-integral, we suggest taking the crack tip
opening displacement as a fracture parameter. Elasto-plastic crack solutions for
some one- and two-and three-dimensional quasicrystals have been found based on
the generalized Dugdale-Barenblatt model, refer to Refs. [14, 18, 27], in which the
size of the plastic zone and crack tip opening displacement are determined, with
these data an equivalent plastic fracture criterion is suggested as
δ = δc , (14.5-5)
in which δc represents the critical value of the crack tip opening displacement, a
material constant. The evaluation of crack tip opening displacement has been exactly
completed for large plates with central crack and narrow plastic zone and introduced
in Section 14.4 or refer to Refs. [14, 18, 27], and some approximate solution for other
configuration can be obtained by complex variable function method (an outline of
the method may be interpreted in Refs. [14, 18, 27]) and other methods (mainly the
approximate methods and numerical methods). The shortcoming of this criterion
lies in some difficulties of the determination of δc , but which can be obtained through
relation
Ec
δc = . (14.5-6)
σc
and the Section 14.8 shows the measurement of Ec is easier.
Thus, the collaboration of criterions (14.5-4) and (14.5-5) makes the nonlinear
fracture analysis for quasicrystals possible. Some details will be given in the Ap-
pendix of this chapter (i.e., Section 14.8).
14.6 Nonlinear analysis based on the dislocation model 305
Fig. 14.6-1 Schematic picture of a crew dislocation pile-up coupled with a crack in
anti-plane elasticity of quasicrystals
306 Chapter 14 Nonlinear behaviour of quasicrystals
for icosahedral quasicrystals. The origin of the model was given by Bilby, Cottrell
and Swinden[20] and Bilby, Cottrell, Smith et al[21] for crystals, we developed model
to the study for some one- and two-dimensional quasicrystals[8] . The model can
be formulated by the following boundary conditions for one-dimensional hexagonal
quasicrystals:
(x2 + y 2 )1/2 → ∞ : σij = 0, Hij = 0,
y = 0, |x| < l : σyz = −τ (∞) , Hyz = 0, (14.6-1)
y = 0, l < |x| < l + d : σyz = −τ (∞) + τc = 0, Hyz = 0.
Some detailed statement for the formulation can refer to Ref. [18].
The boundary conditions (14.6-1) has been shown that the nonlinear (plastic
deformation) problem is linearized mathematically. So it has been transformed into
a linear problem (or an “equivalent” elasticity problem) described by the final gov-
erning equations
∇2 uz = 0, ∇2 wz = 0. (14.6-2)
and boundary conditions (14.6-1), detail about this can be found in the text of
Chapters 5, 7 and 8. The nonlinear fracture problem can be solved if we can obtain
the solution of equation (14.6-2) under boundary conditions (14.6-1). It is obvious
the problem is complicated. Though we can solve the problem in the methodology
developed in Chapter 8, but there is simpler procedure to solve it if we are interested
in only some plastic deformation parameters around crack tip and if we have a
dislocation solution, the latter is given in Section 7.1. For this purpose we first
introduce a dislocation density function f (ξ), then problem (14.6-1) and (14.6-2)
can transformed into the following singular integral equation problem such as
f (ξ)dξ τ (x)
= , (14.6-3)
L ξ−x A
in which ξ denotes the dislocation source point coordinate, x denotes the field point
coordinate at the real axis and L represents interval (l, l + d). From the dislocation
solution given in Section 7.1 the constantA is defined by
&
1 x + (l + d) l (∞)
=− 2 i 2τc arccos −τ π
π A x − (l + d) l+d
" " " " (14.6-5)
τc " (l + d)2 − lx " " (l + d)2 + lx "
+ 2 "
arc cosh " " "
− arc cosh " " .
π A (l + d)(l − x) " (l + d)(l + x) "
Because the function f (x) should be a real number, the factor multiplying the
imaginary number i must be zero in the first term of right-handside of (14.6-5), this
leads to
l
2τc arccos − τ (∞) π = 0,
l+d
i.e., (∞)
πτ
d = l sec −1 . (14.6-6)
2τ c
This is the same as (14.4-2) if l is replaced by a.
From solution (14.6-5) we evaluate amount of dislocations N (x) such as
x
N (x) = f (ξ)dξ. (14.6-7)
0
Substituting (16.6-5) (coupled with (14.6-6)) into (14.6-7), we can get N (l + d) and
N (l), so the amount of dislocation motion is
2b lτc l+d 4K2 τc l πτ (∞)
δIII = b3 [N (l + d) − N (l)] = 32 ln = ln sec .
π A l π(C44 K2 − R32 ) 2τc
(14.6-8)
This is the same as (14.4-3) apart from difference of notations.
Fig. 14.6-2 Schematic picture of an “edge” dislocation pile-up coupled with a crack in a
plane elasticity of quasicrystals
In the singular integral equation the shear stress distribution function, τ (x) is
the same as before, but here A is replaced by:
b1 (L + M )(M K1 − R2 )
A= . (14.6-11)
π (L + M )K1 + (M K1 − R2 )
which has been found in Section 7.2 for point groups 5m and 10mm quasicrystals.
According to the condition concerning factor multiplying the imaginary i be zero
in (14.6-5), the plastic zone size is determined by (14.6-6), this is similar to (14.4-6),
but the physical meaning is somewhat different.
By a similar calculation on amount of dislocation motion we can obtain the slip
of the crack tip such as
(∞) (∞)
2b l τc (L+M ) 2τ 2τc l 1 K1 2τ
δII = 21 ln sec = + 2
ln sec .
π A M +K1 πτc π M +K1 M K1 −R πτc
(14.6-12)
It is evident that the result of (14.6-12) is similar to that of (14.4-7), but the phys-
ical meaning is different. According to the macroscopic fracture mechanics, the
crack tip slip corresponding to the Mode II crack tip displacement (i.e., the crack
tip sliding displacement) δII . But for the Mode I crack tip opening displacement
(14.4-7) there is no physical basis of dislocation model, though the similar math-
ematical treatment can be easily given, which is omitted here. Even if this, the
dislocation model provides a powerful support to the nonlinear analysis of previous
sections.
So that
∂F ∂ ∂ui ∂wi
GdΓ = − σij + Hij dx1 dx2 . (14.8-1)
C ∂x1 ∂xj ∂x1 ∂x1
D
curves intersect the upper face of the crack at points B and B , and the lower face
at points A and A , respectively. Due to
dx2 = dy = 0, Ti = 0, hi = 0
mathematical proof about this is given here. The proof can be done for one-, two-
and three-dimensional quasicrystals and for Mode I, Mode II and Mode III cracks.
For simplicity here we only discuss Mode III crack for one-dimensional hexagonal
quasicrystals.
At the linear elastic case, the generalized Hooke’s law
is reduced to
σyz = σyz = 2C44 εyz + R3 wzy ,
for the phonon-phason coupling anti-plane shearing (or longitudinal shearing) state
in elasticity of one-dimensional hexagonal quasicrystals, one finds the free energy
(or strain energy density) is
1
F = C44 (ε2xz + ε2yz ) + K2 (wzx
2 2
+ wzy ) + R3 (εzx wzx + εyz wzy ). (14.8-3)
2
Substituting the crack solution given in Chapter 8 into (14.8-3) we have
1 ⊥ 2 ⊥
F = {C44 [K2 KIII − R3 KIII ] + 2K2 [C44 KIII − R3 KIII ]2
4(C44 K2 − R32 )
⊥ ⊥
+ 2R3 [K2 KIII − R3 KIII ][C44 KIII − R3 KIII ]}, (14.8-4)
√ ⊥
√
where KIII = πaτ1 and KIII = πaτ2 are the stress intensity factors associated
with phonon and phason fields respectively.
Integrating the quantity given by (14.8-4) around a path enclosing the crack
tip, then one can obtain the first term of the E-integral. Because of the path-
independency of the integral, we can take a half-circle with the crack tip as its origin
and with radius r, such that
−π −π
F dy = F r cos θdθ = 0, (14.8-5)
π π
According to the definition of Section 14.3, it is known that between the traction,
generalized traction and the phonon, phason stresses there are
In addition,
∂ ∂ ∂r ∂ ∂θ ∂ ∂
= + = cos θ − sin θ .
∂x ∂r ∂x ∂θ ∂x ∂r ∂θ
Substituting these results and relations into the second and third terms of the
integral we obtain
−π
∂ui ∂wi K2 (KIII )2 + C44 (KIII
⊥ 2 ⊥
) − 2R3 KIII KIII
− σij nj + Hij nj dΓ = 2 .
π ∂x1 ∂x1 C44 K2 − R3
(14.8-8)
From (14.8-5) and (14.8-8), we find that
K2 (KIII )2 + C44 (KIII
⊥ 2 ⊥
) − 2R3 KIII KIII
EIII = 2 = GIII . (14.8-9)
C44 K2 − R3
and (14.3-2) do not remain one to one correspondence, the physical background of
the E-integral does not hold. But if we define the potential energy per unit thickness
for a plane (i.e., a two-dimensional) region Ω occupied by a quasicryslal, and denote
the boundary of the region by Γ
V = F dxdy − (Ti ui + hi wi )dΓ , (14.8-12)
Ω Γ
Π = BV, (14.8-13)
2U
E= , (14.8-15)
B(W − a)
where W represents the width, B the thickness of the specimen and U the area
under P -Δ curve (shown by Fig. 14.8-2), i.e.,
U = P dΔ, (14.8-16)
in which P is the load (force per unit thickness) at loading point, and Δ the dis-
placement of the same point. When the initiation of crack growth is observed, then
the value of E-integral is marked as the fracture toughness of the quasicrystal.
If the phason field is absent, the material is
degenerated to conventional structural material.
In this case, the E-integral is reduced to conven-
tional Eshelby integral or J-integral, the latter
was introduced by Rice[22] and Cherepanov[23] .
Begley and Landes[24,25] put forward the exper-
imental study on J-integral, further promoting
the development of nonlinear fracture theory and
its applications for conventional engineering ma-
terials. These experiences would be helpful for
Fig. 14.8-2 The depiction of the the experimental study of nonlinear fracture of
specimen deformation energy quasicrystalline materials.
References
[1] Calliard D. Dislocation mechanism and plasticity of quasicrystals: TEM observations
in icosahedral Al-Pd-Mn. Materials Sci Forum, 2006, 509(1): 49–56
[2] Geyer B, Bartsch M, Wollgarten M et al. Plastic deformation of icosahedral Al-Pd-
Mn single quasicrystals. Experimental results. Phil Mag A, 2000, 80(7): 1151–1164
References 315
[20] Bilby B A, Cottrell A H, Swinden K H. The spread of plastic yield from a notch.
Proc R Soc A, 1963, 272(2): 304–314
[21] Bilby B A, Cottrell A H, Smith E et al. Plastic yealding from sharp notches, ibid,
1964, 279(1): 1–9
[22] Rice J R. A path independent integral and approximate analysis of strain concentra-
tion by notches and cracks. J Appl Mech, 1968, 35(4): 379–386
[23] Cherepanov G P. On crack propagation in solids. Int J Solids and Structures, 1969,
5(8): 863–871
[24] Begley G T, Landes J D. 1972, The J-integral as a fracture criterion, Fracture Tough-
ness, ASTM STP 514, American Society for Testing and Materials, Philadelphia,
1–20
[25] Landes J D, Begley G T. 1972, The effect of specimen geometry on JIC , Fracture
Toughness, ASTM STP 514, American Society for Testing and Materials, Philadel-
phia, 24–29
[26] Muskhelishvili N I. Singular Integral Equations. Groningen, Noordhoff, 1956
[27] Li W, Fan T Y. Study on plastic analysis of crack problem in icosahedral quasicrys-
tals. Phil. Mag., 2009, 89(31): 2823–2831
Chapter 15
Fracture theory of quasicrystals
−1/2
stresses for phonons as well as for phasons appear singular in order of r1 (r1 →0),
and other terms can be ignored when compared to this term. Although stress sin-
gularity is implausible, it is the result of idealized mathematical model. Quite
a few researchers indicated its severe weakness in theory and the paradox of its
methodology[1,2] . However, prior to the actual establishment of more reasonable
fracture theory, we still continue using this theory in expectable near future.
If temporarily accepting this theory, we focus on the field variables near crack
tip. If in (8.3-14) only keeping the term in order of (r1 /a)−1/2 , we have
⎧
⎪
⎪
⎪ KI 1 1 3
⎪ xx
⎪ σ = √ cos θ 1 1 − sin θ 1 sin θ 1 ,
⎪
⎪ 2πr1 2 2 2
⎪
⎪
⎪
⎪
⎪
⎪ KI 1 1 3
⎪
⎪ σ = √ cos θ 1 + sin θ sin θ 1 ,
⎪
⎪
yy
2πr1 2
1
2
1
2
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ √
KI 1 3
⎪
⎪ σ xy = σ yx = cos θ1 cos θ1 ,
⎪
⎪ 2πr 1 2 2
⎪
⎪
⎨
d21 K 3 3 5
Hxx = − √ I sin θ1 2 sin θ1 + sin θ1 cos θ1 , (15.1-2)
⎪
⎪ 2πr1 2 2 2
⎪
⎪
⎪
⎪
⎪
⎪
d21 KI 3 5
⎪ Hyy = √
⎪ sin2 θ1 cos θ1 ,
⎪
⎪
⎪
⎪ 2πr 1 2 2
⎪
⎪
⎪
⎪
⎪
⎪ d21 KI 3 5
⎪ xy
⎪ H = − √ sin2 θ1 sin θ1 ,
⎪
⎪ 2πr 1 2 2
⎪
⎪
⎪
⎪
⎪
⎪ d21 KI 3 3 5
⎪
⎩ yx
H = √ sin θ 1 2 cos θ1 − sin θ 1 sin θ 1 ,
2πr1 2 2 2
in which ⎧
⎨ x
p √ − 1 , |x| > a,
σyy (x, 0) =
⎩ x2 − a2 (15.1-4)
−p, |x| < a,
It is one of results of Sections 8.3 and 8.4, and (15.1-3) represents a physical param-
eter describing fracture behaviour of quasicrystals under Mode I loading condition.
The physical meaning of the generalized surface tractions hi = Hij nj is clear,
but have not been measured so far, we do not considered hi at the physical boundary
||
(simply assume zero). Therefore, we only obtain the KI , but the stress intensity
factor for phason field still exits if we do not assume the generalized tractions hi to be
||
zero. Can we use the stress intensity factor KI in the parallel space (physical space)
15.1 Linear fracture theory of quasicrystals 319
which is called crack strain energy with the suffix “I” to indicate the Mode I crack.
It can be found from solution in Sections 8.3 and 8.4 that under the assumption
of generalized surface traction hi = Hij nj being free, the crack strain energy is still
relevant to both the elastic constant K1 of the phason field and the phonon-phason
coupling elasticity constant R apart from relevant to the phonon elastic constants
L = C12 , M = (C11 − C12 )/2.
In Section 8.3, we define the strain energy release rate (crack growth force) for
point groups 5m and 10mm,
1 ∂W1 πap2 1 K1
GI = = +
2 ∂a 4 L+M M K1 − R 2
1 1 K1
= + (KI )2 , (15.1-7)
4 L+M M K1 − R
320 Chapter 15 Fracture theory of quasicrystals
GI = GIC (15.1-10)
as the crack initiation criterion, where GIC is the critical value, a material constant
determined experimentally. With the availability of explicit expression GI , the mea-
surement of GIC is convenient, to be discussed in the next section. The above results
have been documented in Chapters 8, 9 and the relevant references.
With these common features of cracks in quasicrystalline materials, the funda-
mental of fracture theory for the material can be set up.
for point groups 5m and 10mm, where S is the sample span, B the sample thickness,
W the sample width, a the crack length plus the size of the machined notch, and P
is the external force (per unit length). Finally, the GIC value can be determined by
measuring the critical external force PC . For other quasicrystal systems there are
similar results.
It can be found in fracture mechanics that the stress intensity factor of the compact
tensile sample as shown in Fig. 15.2-1 is
a 1/2 a 3/2
PS
KI = 3/2
29.6 − 185.5
BW W W
a 5/2 a 7/2 a 9/2
+ 655.7 − 1017.0 + 638.9 . (15.2-3)
W W W
where B, W , a, and P have the same meanings above. The GIC value can be deter-
mined by measuring the critical external force PC . For other quasicrystal systems
there are similar results.
For one-dimensional hexagonal quasicrystals we have obtained the crack tip slid-
ing displacement for Mode III crack as
4K2 τc a πτ1
δIII = ln sec (15.3-1)
π(C44 K2 − R32 ) 2τc
and for two-dimensional quasicrystals with point groups 5m and 10mm the crack tip
opening displacement for Mode I crack is
2σc a 1 K1 πp
δI = + ln sec . (15.3-2)
π L+M M K1 − R2 2σc
For other quasicrystal systems the results are similar.
The plastic zone size around the crack tip is
πτ1
d = a sec −1
2τc
for one-dimensional hexagonal quasicrystals and
πp
d = a sec −1
2σc
for two-dimensional quasicrystals with point groups 5m and 10mm.
And we have fracture criterion for mode I crack
δI = δIC . (15.3-3)
For Mode II and Mode III crack there are similar criterion, which have been discussed
in Chapter 14.
As pointed out in Chapter 14 the Eshelby integral can also be a fracture param-
eter, and based on which one can set up a fracture criterion, the full discussion can
be found there.
The experimental measurement of nonlinear fracture toughness of quasicrystals
has been introduced in Section 14.8 of Chapter 14, it does not mention any more.
Fig. 15.4-1 Normalized dynamic stress intensity factor of central stationary crack
specimen under Heaviside impact loading (for icosahedral Al-Pd-Mn quasicrystals)
With the results we can propose the fracture criterion for dynamic crack initiation
in which KI (t) is the dynamic stress intensity factor evaluated by different ap-
proaches, KId (σ̇) represents the dynamic fracture toughness for initiation of crack
growth of the material and is measured by test, a material constant, but is function
of loading rate σ̇. While for fast crack propagation/crack arrest problems we have
results, e.g. shown in Fig. 15.4-2 for the central crack specimen there is fracture
criterion such as
Fig. 15.4-2 Normalized dynamic stress intensity factor of propagating crack with
constant crack speed of central crack specimen (for icosahedral Al-Pd-Mn quasicrystals)
15.5 Measurement of fracture toughness and relevant mechanical parameters of ... 325
in which HV = 4.10GPa.
The values of fracture toughness for general alloys for black metals measured by
Ma et al[13] are much greater than the above data, those for aluminum alloys and
√
other colour metals are also, e.g., for aluminum 33MPa m refer to Fan[14] . So one
finds that quasicrystals are very brittle.
The author of the book minds the indirect measurement for fracture tough-
ness of quasicrystals through indentation perhaps is not so exact, because formula
(15.5-1) is empirical, the exact measurement should use the stress intensity factor
formula. Due to the high brittleness of the material, maybe it is easy by taking the
indentation approach.
326 Chapter 15 Fracture theory of quasicrystals
they obtained
σc = 450MPa (15.5-5)
before annealing and
σc = 550MPa (15.5-6)
after annealing for decagonal Al-Ni-Co quasicrystal.
The Fig. 15.5-1 shows the SEM morphology of grain interior containing large
hole, Fig. 15.5-2 shows diagram of indentation crack under applied load 100g before
annealing and after annealing, Fig. 15.5-3 shows the SEM fractograph and fracture
feature for decagonal Al65 Cu20 Co15 quasicrystal.
References
[1] Li X F, The defect problems and their analytic solutions in the theory of elasticity
of quasicrystals, Dissertation, Beijing Institute of Technology, Beijing, 1999 (in Chi-
nese); Li X F, Fan T Y, Sun Y F. A decagonal quasicrystal with a Griffith crack.
Phil Mag A, 1999, 79(8): 1943–1952
[2] Fan T Y. Mathematical theory of elasticity and defects of quasicrystals. Advances
of Mechanics, 2000, 30(2): 161–174 (in Chinese)
[3] Fan T Y, Mai Y W. Theory of elasticity, fracture mechanics and some relevant
thermal properties of quasicrystalline materials. Appl Mech Rev, 2004, 57(5): 235–
244
[4] Rudhart C, Gumbsch P, Trebin H R. Crack propagation in quasicrystals. Quasicrys-
tals, Trebin H R. Berlin: Wiely Press, 2003
[5] Fan T Y, Guo L H. The final governing equation of plane elasticity of icosahedral
quasicrystals. Phys Lett A, 2005, 341(5): 235–239
[6] Zhu A Y, Fan T Y. Elastic analysis of Mode II Griffith crack in an icosahedral
quasicrystal. Chinese Physics, 2007, 16(4): 1111–1118
[7] Li L H, Fan T Y. Complex variable method for plane elasticity of icosahedral qua-
sicrystals and elliptic notch problem. Science in China, G, 2008, 51(6): 1–8
[8] Fan T Y, Fan L. Plastic fracture of quasicrystals. Phil Mag, 2008, 88(4): 323–335
[9] Fan T Y, Fan L. Relation between generalized Eshelby integral and generalized BCS
model and generalized DB model. Chin. Phys. B, in press, 2010
[10] Zhu A Y, Fan T Y. Dynamic crack propagation of decagonal Al-Ni-Co decagonal
quasicrystals. J Phys: Condens Matter, 2008, 20(29): 295217
[11] Wang X F, Fan T Y and Zhu A Y. Dynamic behaviour or the icosahedral Al-Pd-Mn
quasicrystal with a Griffith crack, Chin. Phys. B, 2009, 18(2): 709–714. or Zhu A
328 Chapter 15 Fracture theory of quasicrystals
References
[1] Sommerfeld A. Vorlesungen ueber theoretische Physik. Vol. V: Elastische Theorie.
Wiesbaden: Diederich Verlag, 1952
[2] Landau L D, Lifshitz E M. Theoretical Physics. Vol. VII: Theory of Elasticity.
Oxford: Pergamon Press, 1986
[3] Bak P. Symmetry, stability and elastic properties of icosahedral incommensurate
crystals. Phys Rev B, 1985, 32(9): 5764–5772
[4] Horn P M, Malzfeldt W, DiVincenzo D P et al. Systematics of disorder in quasiperi-
odic Material. Phys Rev Lett, 1986, 57(12): 1444–1447
[5] Lubensky T C. Introduction to Quasicrystals. Jaric M V. Boston: Academic Press,
1988.
[6] Hu C Z, Wang R H, Ding D H. Symmetry groups, physical property tensors, elasticity
and dislocations in quasicrystals. Rep Prog Phys, 2000, 63(1): 1–39
[7] Fan T Y, Mai Y W. Elasticity theory, fracture mechanic and some thermal properties
of quasicrystalline materials. Appl Mech Rev, 2004, 57(5), 325–344
[8] Feuerbacher M, Urban K. Platic behaviour of quasicrystalline materials. in: Qua-
sicrystals. Trebin H R, Wiely Press, Berlin, 2003
[9] Fan T Y, Trebin H R, Messerschmidt U et al. Plastic flow coupled with a crack
in some one- and two-dimensional quasicrystals. J Phys: Condens Matter, 2004,
16(47): 5229–5240
[10] Fan T Y, Fan L. Plastic fracture of quasicrystals. Phil Mag, 2008, 88(4): 523–535
[11] Fan T Y, Wang X F, Li W et al. Elasto-hydrodynamics of quasicrystals. Phil Mag,
2009, 89(6), 501–512
[12] Wang X F, Fan T Y. Dynamic behaviour of the icosahedral AL-Pd-Mn quasicrystals
with a Griffith crack. Chin Phys, 2009, 18(2), 709–714
[13] Audier M, Guyot P. Al4 Mn quasicrystal atomic structure, diffraction data and Pen-
rose tiling. Phil Mag B, 1986, 53(1): L43–L51
[14] Jaric M V, Gratias D. Extended Icosahedral Structures. Boston: Academic, 1989
[15] Jaric M V, Lundquist S. Proc Adiatico Research Conf Qusicrystals, Singapore: World
Scientific, 1990
[16] Elser V, Henley C L. Crystal and quasicrystal structures in Al-Mn-Si alloys. Phys
Rev Lett, 1985, 55(26): 2883–2886
[17] Henley C L, Elser V. Quasicrystal structure of (Al, Zn)49 Mg32 . Phil Mag B, 1986,
53(3): L59–L61
[18] Lubensky T C, Ramaswamy S, Toner J. Hydrodynamics of icosahedral quasicrystals.
Phys Rev B, 1985, 32(11): 7444–7452
[19] Lubensky T C, Socolar J E S, Steinhardt P J et al. Distortion and peak broadening
in quasicrystal diffraction patterns. Phys Rev Lett, 1986, 57(12): 1440–1443
References 331
in which both P (x, y) and Q(x, y) are functions with real variables, and called the
real and imaginary parts, respectively, and marked by
There is a sort of complex functions called analytic functions (or regular func-
tions, and single-valued analytic functions are called holomorphic functions) which
have important applications in many branches of mathematics, physics and engi-
neering. The concepts related with this are discussed as follows.
The complex function f (z) is analytic in a given region, this means that it can
be expanded in the neighbourhood of any point z0 of the region into a non-negative
integer power series (i.e., the Taylor series) of the form
∞
f (z) = an (z − z0 )n , (A.I-2)
n=0
in which an is a constant (in general, a complex number). The concept will be used
frequently in later calculation.
Another definition of an analytic function is that if the complex function f (z)
given in the region, whose real part P (x, y) and imaginary part Q(x, y) are single
valued, have continuous partial derivatives of the first order, and satisfy Cauchy-
Riemann conditions such as
∂P ∂Q ∂P ∂Q
= , =− (A.I-3)
∂x ∂y ∂y ∂x
in the region.
This kind of function, P and Q are named mutually conjugate harmonic ones.
From (A.I-3) it follows that
2 2
∂ ∂2 ∂ ∂2
∇2 P = + P = 0, ∇2
Q = + Q = 0.
∂x2 ∂y2 ∂x2 ∂y 2
closed curve (sometime called simple curve for simplicity), we can obtain that f (z)
is analytic in the region if
f (z)dz = 0. (A.I-4)
Γ
The result is known as the Cauchy’s integral theorem (or simply called the Cauchy’s
theorem) which is frequently used in some chapters in the text and in subsequent
description here.
The theory of complex functions proves that the above definitions are mutually
equivalent.
An important result of the Cauchy’s theorem is so-called the Cauchy’s formula, i.e.,
if f (z) analytic in a single-connected region D + bounded by a closed curve Γ and
continuous in D+ + Γ (Fig. A.I-1), then
1 f (t)
dt = f (z), (A.I-5)
2πi Γ t−z
Just as mentioned in the beginning that f (z) is analytic in D + , the value of the
integral
f (z)
dt
γ −z
t
will not be changed when ρ is reducing. Therefore the limit mark in the left-handside
of (A.I-7) can be removed. In addition
2π iθ
f (z) dt ρe
dt = f (z) = f (z) dθ = 2πif (z).
γ t − z γ t − z 0 ρeiθ
1 dt 1, k = 0,
=
2πi Γ tk (t − z) 0, k > 0,
Appendix I Outline of complex variable functions and some additional calculations 337
A.I.3 Poles
Suppose a finite point in z-plane (i.e., z is not a point at infinity), and in the
neighbourhood of the point, the function presents the form as follows:
A0 A1 Am
G(z) = + + ··· + , (A.I-13)
z − a (z − a)2 (z − a)m
where A1 , A2 , · · · , Am are constants, such that f (z) is called having a pole with
order m, z = a is the pole.
If a is a point at infinity, f0 (z) in (A.I-12) is regular at point at infinity (i.e.,
f (t) = c0 + c1 z −1 + c2 z −2 + · · · ), while at z = ∞,
G(z) = A0 + A1 z + · · · + Am z m , (A.I-14)
What is the meaning of the residue? Suppose f (z) analytic in the neighbourhood
of point z = a, but except z = a, and infinite at z = a. In this case the point z = a
is named isolated singular point. The residue of the function f (z) at point z = a is
the value of the integral
1
f (z)dz,
2πi Γ
in which Γ represents any closed contour enclosing point z = a. For a residue we
will use the resignation as Res f (a).
If z = a is a m-order pole of f (z), its residue may be evaluated from the following
formula and
1 dm−1
Resf (a) = lim m−1 {(z − a)m f (z)} , (A.I-15)
(m − 1)! z→a dz
obviously the integral is
f (z)dz = 2πiResf (a).
Γ
# #
of the integral has two poles (− k/m, 0) and ( k/m,0), the value of the integral
is equal to
∞
1 1 −iωt
e dω = I1 = lim + + + + + ,
2π −∞ −mω 2 + k R→∞,r→0 CR 1 2 3 C1 C2
(A.I-19)
where the first integral in the right-handside of (A.I-19) is carried out on path of
the grand half-circle, the second and fourth ones are on the path along the real axis
# # # #
except intervals (−r − k/m, − k/m + r) and (−r + k/m, k/m + r), the fifth
#
and sixth ones are on two small half-circle arcs C1 and C2 with origins (− k/m, 0)
#
and ( k/m, 0) and radius r respectively. Because the integrand in the interior
enclosing by the integration path in (A.I-19) is analytic, according to the Cauchy’s
theorem (referring to formula (A.I-3))
I1 = 0. (A.I-20)
Based on the bebaviour of the integrand and the Jordan lemma, the first one in the
right-handside of (A.I-19) must be zero. So that
lim + + + + =0
R→∞,r→0 1 2 3 C1 C2
and
lim + + = I = − lim + .
R→∞,r→0 1 2 3 r→0 C1 C2
# #
At arc C1 : ω + k/m = reiθ1 , dω = ireiθ1 dθ1 , and at C2 : ω − k/m = reiθ2 , dω =
ireiθ2 dθ2 . Substituting these into the above integrals and after some simple calcula-
tions, we obtain
+
π k
I= # sin t (A.I-21)
m k/m m
340 Major Appendix: On Some mathematical materials
This result will be used in the evaluation of some integrals in Chapter 9, which
can be seen the Subsection AI.10 of this appendix.
f1 (z) = f2 (z), z ∈ Γ,
we can say that f1 (z) and f2 (z) are analytic extension to each other, we can also
f1 (z) as z ∈ D1
say that function F (z) =
f2 (z) as z ∈ D2
analytic at D = D1 + D2 is an analytic extension of f1 (z) as well as f2 (z).
z = ω(ζ), (A.I-22)
(2) If ω(ζ) is analytic and single-valued in region Ω and transforms the region
into region D, then the inversion ζ = ω −1 (z) is analytic and single-valued in region
D and maps D onto Ω .
(3) If D is a region and c is a simple closed curve in it, and its interior belongs to
D, and if ω −1 (z) is analytic, and maps c onto a closed curve γ at Ω region bilaterally
single-valued, then ω(ζ) is analytic and single-valued in the region and maps D onto
the interior of Ω .
In the text we mainly used the following two kinds of conformal mapping, i.e.,
(1) Rational function conformal mapping, e.g.
c
ω(ζ) = + a0 + a1 ζ + · · · + an ζ n (A.I-23)
ζ
or
1 1
ω(ζ) = Rζ + b0 + b1 + · · · + bn n , (A.I-24)
ζ ζ
in which, c, a0 , a1 , · · · , an , R, b0 , b1 , · · · , bn are constants. These mappings can be
used in studying infinite region with a crack at physical plane onto the interior of
unit circle at mapping plane. In the monograph of Muskhelishvili[3] , he postulated
that his method is only suitable for this kind of mapping functions. Fan[4] extended
it to transcendental mapping functions and achieved exact analytic solutions for
crack problems for complicated configuration.
(2) Transcendental functions, e.g.
H (1 + ζ)2
ω(ζ) = ln 1 + (A.I-25)
π (1 − ζ)2
and 5# πa 6
2W
ω(ζ) = arctan 1 − ζ 2 tan − a, (A.I-26)
π 2W
which can be used to transform a finite specimen with a crack onto the interior of
unit circle at mapping plane, where H, W and a represent sample sizes and crack
size.
where
4Hα(1 − β) 1 − σ (1 + σ)2
ω (σ) = − .
π 1 + σ [(1 + σ) + α(1 − σ 2 )2 ][(1 + σ)2 + βα(1 − σ)2 ]
2
1−σ 2i 4
= −2x, dσ = dx, (1 + σ)2 = ,
1+σ (1 − ix)2 (1 − ix)2
1 1 − ix (1 − ix)[1 − ζ − ix(1 + ζ)] 1 − ζ − x2 (1 + ζ) − 2ix
= = − .
σ−ζ 1 − ζ + ix(1 + ζ) (1 − ζ)2 + x2 (1 + ζ)2 (1 − ζ)2 + x2 (1 + ζ)2
So that
1
pHα(1 − β) ix[1 − ζ − x2 (1 + ζ) − 2ix]
F (ζ) = − dx
π2 −1 (1 − αx2 )(1− βαx2 )[(1 − ζ)2 + x2 (1 + ζ)2 ]
1
4pHα(1 − β) x2
=− dx
π2 0 (1 − αx2 )(1 − βαx2 )[(1 − ζ)2 + x2 (1 + ζ)2 ]
2pH α(1 − β)(1 − ζ 2 )
=
π [α(1 − ζ) + (1 + ζ)2 ][βα(1 − ζ)2 + (1 + ζ)2 ]
2 2
1+ζ 1+ζ
× arctan − arctan
1−ζ −1 − ζ
√ √ √ √
4pH α arctanh α βα arctanh α
− 2 − .
π α(1 − ζ)2 + (1 + ζ)2 γα(1 − ζ)2 + (1 + ζ)2
(A.I-27)
In the last step the evaluation is used of the Mathematica 3.0[19] .
By considering
√
1+ α √
A = ln √ = 2 arctanh α,
1− α
√
1 + βα √
M = ln √ = 2 arctanh βα,
1 − βα
1+ζ 1+ζ i i−ζ
arctan = − arctan = ln ,
−1 + ζ 1−ζ 2 1 − iζ
then (A.I-27) is just the formula (8.2-19).
in which the first two terms are known referring to the text, and the single-valued
analytic function Φ4∗ (ζ) satisfies the following boundary condition:
ω(σ)
Φ4∗ (σ) + Φ3∗ (σ) +
· Φ4∗ (σ) = f0 ,
ω (σ)
where
i ω(σ)
f0 = (Tx + iTy )ds − (d1 − d2 )(X + iY ) ln σ − · d1 (X − iY ) · σ
32c1 ω (σ)
−2Bω(σ) − (B − iC )ω(σ)
(A.I-28)
and (referring to Fig. 11.3-4)
X + iY = (Tx + iTy )ds = ip(z1 − z2 ).
1 dσ
Multiplying both sides of the above equation by , and integrating around
2πi σ − ζ
the unit circle, we have
Φ4∗ (σ)
1 1 ω(σ) Φ4∗ (σ) 1 Φ3∗ (σ) 1 f0
dσ + dσ + dσ = dσ,
2πi γ σ − ζ 2πi γ ω (σ) σ − ζ
2πi γ σ − ζ 2πi γ σ − ζ
(A.I-30)
in which according to the Cauchy’s integral formula (referring to formula (A.I-5))
there is
1 Φ4∗ (σ)
dσ =Φ4∗ (ζ)
2πi γ σ − ζ
and in terms of analytic extension principle and the Cauchy’s theorem (referring to
Sections A.I-5 and A.I-1)
1 ω(σ) Φ4∗ (σ)
=0
2πi γ ω (σ) σ − ζ
in which
σ2 m dσ m σ2 m σ2 − ζ
σ+ = σ2 − σ1 − ln + ζ+ ln ,
σ1 σ σ−ζ ζ σ1 ζ σ1 − ζ
σ2
dσ σ1 − ζ
= ln .
σ1 σ−ζ σ2 − ζ
because the integrand is single-valued analytic function in the region outside the
unit circle γ. The remaining term is
1 ln σ
I(ζ) = dσ.
2πi γ σ − ζ
2πi 1
= − .
σ1 − ζ ζ
So that
I(ζ) = ln(σ1 − ζ) − ln ζ + const.
Appendix I Outline of complex variable functions and some additional calculations 345
Hence function Φ4∗ (ζ) is determined so the function Φ4 (ζ) in which the constant
term is omitted:
1 p mR0 σ2 σ2 − ζ
Φ4 (ζ) = · · − ln + z ln + z1 ln(σ1 − ζ) − z2 ln(σ2 − ζ)
32c1 2πi ζ σ1 σ1 − ζ
+ip(d1 − d2 )(z1 − z2 ) ln ζ,
which is just the first formula of (11.3-53), others can be similarly derived. In the
derivation the classical work of Muskhelishvili[1] is referred.
∇2 ∇2 ∇2 ∇2 G = 0 (A.I-31)
respectively from (11.3-53), one can find other corresponding three solutions, by
superposing which one can obtain solution
⎧
⎪ (1) 1 σc (a + d)ϕ2 1 1 σc
⎪
⎪ Φ4 (ζ) = · · − ·
⎪
⎪ 32c 1 π ζ 32c 1 2πi
⎪
⎪
⎨ σ −ζ σ2 + ζ
(ζ − σ2 )(ζ + σ2 )
2
z ln + ln − a ln
⎪
⎪ σ2 − ζ σ2 + ζ (ζ + σ2 )(ζ − σ2 )
⎪
⎪
⎪
⎪
⎪ Φ (1) (ζ) = 1 · σc (a + d)ϕ2 · 2ζ − 1 · σc a ln (ζ − σ2 )(ζ + σ2 ) ,
⎩ 3
32c1 π ζ 2 − 1 32c1 2πi (ζ + σ2 )(ζ − σ2 )
(A.I-35)
iϕ
where σ = e represents the value of ζ at the unit circle in the mapping plane and
σ2 = eiϕ2 , a = (a + d) cos ϕ2 .
And the solution of problem (A.I-31), (A.I-34) is known, i.e.,
⎧ 1 p 1
⎪ (2)
⎨ Φ4 (ζ) = − 32c1 2 (a + d) ζ ,
⎪
(A.I-36)
⎪
⎪ p ζ
⎩ Φ3(2) (ζ) = − (a + d) .
32c1 (ζ 2 − 1)
(1)
The superposition of (A.I-35) and (A.I-36) gives the total solution Φ4 (ζ) = Φ4 (ζ)+
(2) (1) (2)
Φ4 (ζ), Φ3 (ζ) = Φ3 (ζ) + Φ3 (ζ), for example the first term of Φ4 (ζ) is
1 p 1 1 σc (a + d)ϕ2 1
− (a + d) + · · (A.I-37)
32c1 2 ζ 32c1 π ζ
and Φ2 (ζ) has not been listed here because it is too lengthy. So the stresses and
displacements are determined already. In addition, we know that
Φ (ζ)
σij , Hij ∼ (A.I-38)
ω (ζ)
in which the constants c1 , c2 , c3 are defined in Section 11.3, so the solution holds for
point group 10, 10 decagonal quasicrystals as well. When we assume R1 = R, R2 = 0
in the equation (A.I-42), δI will be the corresponding solution of point group 10mm
decagonal quasicrystals, i.e.,
2σc a 1 K1 π σ (∞)
δI = CT OD = + ln sec ,
π L+M M K1 − R2 2 σc
at real axis ξ1 , and by a generalized Jordan lemma the integral along the big half-
circle is zero.
# (1) # (2)
If in Fig. A.I-2 we put ω = ξ, ω1 = ξ1 , ω2 = ξ2 , k/m = ξ1 , − k/m = ξ1 ,
according to the additional integration path at complex ξ-plane, through the similar
manner for evaluating integral (A.I-18) then we can obtain results (9.2-15) and (9.2-
(1) (2)
16) respectively where ξ1 , ξ1 are defined by (A.I-43).
348 Major Appendix: On Some mathematical materials
or
⎧ ∞ n
⎪
⎪
⎪
⎪ y αj
ajk fj (y)Jνj (xy)dy = gj (x), 0 < x < 1,
⎨ 0
k=1
∞ j = 1, 2, · · · , n (A.II-2)
⎪
⎪
n
⎪
⎪
⎩ ajk fj (y)Jνj (xy)dy = 0, x > 1,
0 k=1
or
⎧ ∞ ∞
⎪
⎪ g1 (ξ1 , ξ2 , s, x1 , x2 )f (ξ1 , ξ2 , s)Jα (ξ1 x1 )Jβ (ξ2 x2 )dξ1 dξ2 = h(x1 , x2 , s),
⎪
⎪
⎪
⎪ 0 0
⎪
⎪
⎨ (x1 , x2 ) ∈ Ω1 ,
∞ ∞
⎪
⎪
⎪
⎪ g2 (ξ1 , ξ2 , s, x1 , x2 )f (ξ1 , ξ2 , s)Jα (ξ1 x1 )Jβ (ξ2 x2 )dξ1 dξ2 = 0,
⎪
⎪
⎪
⎪ 0 0
⎩
(x1 , x2 ) ∈ Ω2 .
(A.II-3)
Appendix II Dual integral equations and some additional calculations 349
Among them the equations (A.II-1) are the simplest ones, which will be discussed in
the following only. The equations (A.II-2) are dealt with multi-unknown functions,
and (A.II-3) are the two-dimensional dual integral equations, these two kinds of dual
integral equations are more complicated.
In equations (A.II-1), f (x) is unknown function to be determined, g(x) is known
one, α and ν are constants, Jν (xy) the first kind Bessel function of ν order. Titchm-
arsh[6] and Busbridge[7] gave the analytic solution of the equations. Various au-
thors[8∼16] discussed the solutions with different methods. Here only the procedure
of Refs. [6,7] is introduced. Titchmarsh[6] gave formal solution for the case α >
0. Busbridge extended the discussion to the case α > −2 and gave proof for the
existence of the solution. The solution is given through a complex integral as follows
k+i∞
1 Γ(1/2 + ν/2 + s/2)
f (x) = 2s−α ψ(s)x−s ds, (A.II-4)
2πi k−i∞ Γ(1/2 + ν/2 + α/2 − s/2)
in which s = σ + iτ and
C+i∞
1 Γ(1/2 + ν/2 − α/2 + w/2) g(α + 1 − w)
ψ(s) = · dw, (A.II-5)
2πi C−i∞ Γ(1/2 + ν/2 + w/2) w−s
in above formulas Γ(x)represent the Euler gamma function. The solution (A.II-4)
holds for both α > 0 and α > −2.
For α > 0, the solution can be expressed by real integral as
1
(2x)1−α/2 1 1+α/2
f (x) = μ Jν+α/2 (μx)dμ g(ρμ)ρν+1 (1 − ρ2 )α/2−1 dρ
Γ(α/2) 0 0
(A.II-4 )
and for α > −2, which is in form
1
2−α/2 x−α
f (x) = μ1+α/2 Jν+α/2 (x) y ν+1 (1 − y2 )α/2 g(y)dy
Γ(1 + α/2) 0
1 1 (A.II-4 )
+ y α+1 (1 − y 2 )α/2 dy (xu)2+α/2 g(yu)Jν+1+α/2 (xu)du .
0 0
1
Theorem If α > −2, − ν − 1 < α − < ν + 1, the Mellin transforms of g(x)
2
and f (x) exist, the latter is analytic in the strip region −ν < Re s = σ < α and
350 Major Appendix: On Some mathematical materials
σ−α+ε
has the order O(|t| ) (ε > 0, t → ∞), where s = σ + iτ is the Mellin transform
parameter, then equations (A.II-1) have one and only one solution (A.II-4).
Proof Because the strict proof given by Busbridge [7] is very lengthy, we
cannot quote its all details here, instead, only a rough outline of the proof is figured
out in the following. One can find that in the proof a quite lot of complex variable
function knowledge are used, this seems that the theory on dual integral equations
presents inherent connection with complex analysis. So the Section I of this appendix
is helpful for the present discussion too.
1
At first assume that 0 < α < 2, −ν − 1 < α − < ν + 1 and the Mellin transform
2
of f (x) ∞
f (s) = f (x)xs−1 dx, s = σ + iτ
0
is analytic in region −ν < σ < α, and assume as ε > 0, as t → ∞ has order
−α+ε
O(|t| ).
According to the definition, the Mellin transform of y α Jν (xy) is
∞
2α+s−1 Γ(α/2 + ν/2 + s/2)
J α (s) ≡ [yα Jν (xy)]y s−1 dy = α+s . (A.II-6)
0 x Γ(1 − α/2 + ν/2 − s/2)
Recall that s = σ + iτ . By using the notations of relevant Mellin transform the
left-handside of the first and second equations (A.II-1) become
∞ C+i∞
1
α
y f (y)Jν (xy)dy = f (s)J α (1 − s)ds,
0 2πi C−i∞
∞ C+i∞
1
f (y)Jν (xy)dy = f (s)J 0 (1 − s)ds
0 2πi C−i∞
and substituting (A.II-6) into the above formulas yields
C+i∞ α−s
1 2 Γ(1/2 + α/2 + ν/2 − s/2)
f (s)ds = g(x), 0 < x < 1,
2πi C−i∞ x1−s Γ(1/2 − α/2 + ν/2 + s/2)
C+i∞ α−s
1 2 Γ(1/2 + ν/2 − s/2)
f (s)ds = 0, x > 1.
2πi C−i∞ Γ(1/2 + ν/2 + s/2)
Put
2α−s Γ(1/2 + ν/2 + s/2)
f (s) = ψ(s). (A.II-7)
Γ(1/2 + α/2 + ν/2 − s/2)
Then the above equations reduce to
⎧ C+i∞
⎪
⎪ 1 Γ(1/2 + ν/2 + s/2)
⎪
⎨ ψ(s)xs−1−α ds = g(x), 0 < x < 1,
2πi C−i∞ Γ(1/2 + ν/2 − α/2 + s/2)
⎪ C+i∞
⎪ 1
⎪ Γ(1/2 + ν/2 − s/2)
⎩ ψ(s)xs−1 ds = 0, x > 1.
2πi C−i∞ Γ(1/2 + ν/2 + α/2 − s/2)
(A.II-8)
Appendix II Dual integral equations and some additional calculations 351
Multiplying xα−w to the first one of (A.II-8), where w = u + iv and σ − u > 0, then
integrating over (0, 1) to x and
C+i∞
1 Γ(1/2 + ν/2 + s/2) ds
ψ(s) = g(α − w + 1), (A.II-9)
2πi C−i∞ Γ(1/2 + ν/2 − α/2 + s/2) s−w
where u < C and 1
g(α − w + 1) = g(x)xα−w dx.
0
The left-handside of equation (A.II-9) is analytic everywhere in the strip zone
−ν < σ < α
−α+ε
except the simple pole s = w, and behaves order O(|t| ). If we move the inte-
gration path from σ = C to σ = C < u, see Fig. A.II-1, based on the Cauchy’s
integral formula (referring to formula (A.I-5)),
C +i∞
1 Γ(1/2 + ν/2 + s/2) ds
ψ(s)
2πi C −i∞ Γ(1/2 + ν/2 − α/2 + s/2) s−w
Γ(1/2 + ν/2 + w/2)
= g(α − w + 1) − ψ(w).
Γ(1/2 + ν/2 − α/2 + w/2)
This translation of the integration line corresponds to form a closed region, the value
of the integral around the closed region is just equal to the second term of the above
formula including the sign of the term. Because the left-handside is analytic as
u > C , so the right-handside. In addition,
Γ(1/2 + ν/2 − α/2 + w/2)
ψ(w) − g(α − w + 1) (A.II-10)
Γ(1/2 + ν/2 + α/2)
is analytic for the case
1 1 1 1
+ ν − α + w = 0, −1, −2, · · · .
2 2 2 2
So that
1 1
2 1+ν
ψ(w) = g(ξ)ξ dξ η −w−ν (η 2 − ξ 2 )α/2−1 dη.
Γ(α/2) 0 ξ
By exchanging the integral order (this means the integration zone at ξη-plane is
changed), we may find that
1 1
2
ψ(w) = η −w−ν dη g(ξ)ξ 1+ν (η 2 − ξ 2 )α/2−1 dξ
Γ(α/2) 0 0
1 1
2
= η α−ω
dη g(ξ)ξ 1+ν (1 − ξ 2 )α/2−1 dξ.
Γ(α/2) 0 0
It is evident that the second pair of dual integral equations (A.II-6) has the zero
solution, i.e., D(ξ) = 0, and we only consider the first pair in (A.II-13), which is
⎧ ∞
⎪ 2
⎪
⎨ C(ξ) |ξ| cos(ξx)dξ = −p, 0 < x < a,
d11 0
∞ (A.II-14)
⎪
⎪
⎩ −1
ξ C(ξ) |ξ| cos(ξx)dξ = 0, x > a.
0
Because 1/2
πξx
cos(ξx) = J−1/2 (ξx)
2
and denoting
1/2
1/2 x πad11
ξ C(ξ) = f (ξ), η = aξ, ρ= , g(ρ) = a p,
a 2ρ
then (A.II-14) is reduced to
⎧ ∞
⎪ 2
⎪
⎨ d ηf (η)J−1/2 (ηρ)dη = g(ρ), 0 < ρ < 1,
11 0
∞ (A.II-14 )
⎪
⎪
⎩ f (η)J−1/2 (ηρ)dη = 0, ρ > 1,
0
which becomes one of the standard dual integral equations shown by (A.II-1) with
1
α = 1, ν=− , g(ρ) = g0 ρ−1/2 , g0 = const = a(πad11 )1/2 p.
2
In this case it is very easy to calculate the solution of dual integral equations
(A.II-14) (or (A.II-14 )) by formulas (A.II-4) and (A.II-5), but the key step is choos-
ing the integration path. In previous introduction on Titchmarsh-Busbridge solu-
tion, we mentioned that it must require that −ν > k > α, −ν < C < α and k < C.
At present case ν = −1/2, α = 1 such that 1/2 < k < 1 and 1/2 < C < 1. The
concrete calculation is as below. Due to
1 1
g0
g(α + 1 − t) = g(ρ)ρα−t dρ = g0 ρ−1/2 ρ1−t dρ = ,
0 0 3/2 − t
where t = t1 + it2 represents a complex variable, and requires that t1 < 3/2. Sub-
stituting the relevant data and the above result into (A.II-5), we have
C+i∞
1 Γ(−1/4 + t/2) 1 1
ψ(s) = g0 dt,
2πi C−i∞ Γ(1/4 + t/2) t − s 3/2 − t
the integration path is shown in Fig. A.II-2. In this case the integrand has only one
pole at point t = 3/2 of first order. According to the formula for evaluating residue
(A.I-15) the above integral is very easy obtained as
Appendix II Dual integral equations and some additional calculations 355
√
Γ(1/2) 1 π
ψ(s) = g0 = g0 , (A.II-15)
Γ(1) 3/2 − s 3/2 − s
1 √
whereas Γ(1) = 1, Γ = π. Substituting the result into formula (AII.4) leads
2
to k+i∞ s−α
√ 1 2 Γ(1/4 + s/2) −s
f (η) = g0 π η ds. (A.II-16)
2πi k−i∞ Γ(1/4 − s/2)
In terms of the inversion of the Mellin transform[26,27] ,
k+i∞
1 Γ(1/2 + μ/2 + s/2)
2s−λ (βη)−s ds
2πi k−i∞ Γ(1/2 + μ/2 + λ/2 − s/2) (A.II-17)
−λ/2 1−λ/2
=2 (βη) Jμ+λ/2 (βη).
and
πad11 −1
C(ξ) = ξ −1/2 f (ξ) = ξ J1 (aξ).
2
This is just the result given by (8.3-10).
The calculation through (A.II-4 ) and (A.II-4 ) yields the same result, so the
correctness of the result is demonstrated.
∞
ξAi (ξ)J0 (ξr)dξ = Mi p0 , 0 < r < a,
0
∞ (A.II-18)
Ai (ξ)J0 (ξr)dξ = 0, r>a
0
is solved and obtained the solution (9.8-8). We here give the detail for the derivation.
According to the standard type of the equations here α = 1, ν = 0, g(ρ) = g0 =
const, so
1
g0
g(α + 1 − t) = g(ρ)ρα−t dρ = , Ret = t1 < 2,
0 2−t
C+i∞
1 Γ(1/2) 1 1 2 1
ψ(s) = g0 dt = g0 √ ,
2πi C−i∞ Γ(1/2 + t/2) t − s 2 − t π2−s
which is just the solution (9.8-8), a little bit difference with that lies in we here used
the normalized coordinate ρ = r/a. In the last step of the calculation the inversion
of the Mellin transform (A.II-17) was used.
The above two subsections demonstrate the effect and simplicity of complex vari-
able function method in evaluating solutions of Titchmarsh-Busbridge dual integral
equations.
References 357
The system of dual integral equations (AII-2) and its applications are discussed
by Fan[17] , the two-dimensional dual integral equations (AII-3) are solved approxi-
mately by Fan and Sun[18] , in which some applications are also given.
References
[1] Muskhelishvili N I. Some Basic Problems of the Mathematical Theory of Elasticity,
Groningen: Noordhoff Ltd, 1953
[2] Privalov I I. Introduction to Complex Variable Functions Theory. Moscow: Science,
1984 (in Russian)
[3] Lavrentjev M A, Schabat B V. Methods of Complex Functions Theory. 6th ed.
Moscow, Science, 1986 (in Russian)
[4] Fan T Y. Semi-infinite crack in a strip. Chin Phys Lett, 1990, 8(9): 401–404
[5] Shen T W, Fan T Y. Two collinear semi-infinite cracks in a strip. Eng Fract Mech,
2003, 70(8): 813–822
[6] Titchmarsh B C. Introduction to the Fourier Integrals. Oxford: Clarenden Press,
1937
[7] Busbridge I W. Dual integral equations. Math Soc Proc, 1938, 44(2): 115–129
[8] Weber H. Ueber die Besselschen Functionen und ihre Anwendung auf die Theorie der
electrischen Stroeme. J fuer reihe und angewandte Mathematik, 1873, 75(1): 5–105
[9] McDonald H M. The electrical distribution induced on a disk plased in any fluid of
force. Phil Mag, 1895, 26(1): 257–260
[10] King L V. On the acoustic radiation pressure on circular disk. Roy Soc London Proc
Ser A, 1935, 153(1): 1–16
[11] Copson E T. On the problem of the electric disk. Edinburg Math Soc Proc, 1947,
8(1): 5–14
[12] Tranter C J. On some dual integral equations. Quar J Math Ser 2, 1951, 2(1): 60–66
[13] Sneddon I N. Fourier Transforms. New York: McGraw-Hill, 1951
[14] Gordon A N. Dual integral equations. London Math Soc J, 1954, 29(5): 360–369
[15] Noble B. On some dual integral equations. Quar J Math Ser 2, 1955, 6(2): 61–67
[16] Noble B. The solutions of Bessel function dual integral equations by a multiplying
factor method. Cambridge Phil Soc Proceeding, 1963, 59(4): 351–362
[17] Fan T Y. Dual integral equations and system of dual integral equations and their
applications in solid mechanics and fluid mechanics. Mathematica Applicata Sinica,
1979, 2(3): 212–230 (in Chinese)
[18] Fan T Y, Sun Z F. A class of two-dimensional dual integral equations and applica-
tions. Appl Math Mech, 2007, 28(2): 247–252
[19] Wolfran St. The Mathematica Book. 3 rd . Wolfran St. Cambridge: Cambridge
University Press, 1996
358 Major Appendix: On Some mathematical materials
[20] Li X F. The defect problems and their analytic solutions in the theory of elasticity
of quasicrystals. Dissertation. Beijing Institute of Technology, 1999 (in Chinese)
[21] Zhou W M, Fan T Y. Plane elasticity problem of two-dimensional octagonal qua-
sicrystal and crack problem. Chin Phys, 2001, 10(8): 743–747
[22] Zhou W M. Mathematical analysis of elasticity and defects of two- and three-dimen-
sional quasicrystals. Dissertation. Beijing: Beijing Institute of Technology, 2000 (in
Chinese)
[23] Zhu A Y, Fan T Y, 2007, Elastic analysis of a mode II crack in an icosahedral
quasicrystal, Chin. Phys., 16(5), 1111–1119
[24] Zhu A Y, Fan T Y, 2009, Elastic analysis of a Griffith crack in an icosahedral Al-
Pd-Mn quasicrystal, Int. J. Modern Phys. B, 23(10), 1–16
[25] Li W, Fan T Y, Study on plastic analysis of crack problem in icosahedral quasicrys-
tals. Phil. Mag., 2009, 89(31): 2823–2831.
[26] Watson G N. A Treatise on the Theory of Bessel Functions. New York: Amazon
Com, 1955
[27] Erdelyi A. Higher Transcendental Functions. Vol.2. New York: McGraw-Hill, 1953
Index
U Vertical mapping, 3
Uniqueness of solution, 267 X
V X-ray diffraction patterns, 1
Vertical space, 27 Y
Variational principle, 252 Yield surface, 285
Variational method Yield strength, 285