Kosevich A.M. The crystal lattice

Подождите немного. Документ загружается.

12.4 Frequency of Local Vibrations in the Presence of a Two-Dimensional (Planar) Defect

293

If ∆ρ < 0 then there is a gap at κ = 0 between the local frequency ω

and the band

of the continuous spectrum:

− ω



∆ρ



. (12.4.16)

It is interesting to note that this gap is proportional to the perturbation squared (

∆ρ

)

To continue our analysis at κ = 0 we assume W

αβ

= W

αβ

forthesakeofsim-

plicity; then

(κ)=−

∆ρ

−



∆ρ

+ W



. (12.4.17)

One can see that the behavior of ω

as a function of κ depends on the signs of the

perturbation parameters ∆ρ and W

.For∆ρ < 0

|∆ρ|



|∆ρ|

−W



. (12.4.18)

If W

< 0 a dispersion relation f or the localized wave is characterized by curve (1)

in Fig. 12.4 for all κ.AtW

> 0, localized states can exist only for wave numbers

0 < κ < κ

(curve (2) in Fig. 12.4). The value of κ

for which the dispersion curve

corresponding to the localized wave touches the top boundary of the band can be found

from the equation

(κ

)=0.

Fig. 12.4 Dispersion lines of localized waves: (1) vibrations exist at

all wave vectors; (2) the dispersion line ends on the boundary of the

frequency band; (3) the dispersion line for localized vibrations of the

type of a surface wave.

2. V

< 0. Now (12.4.10) necessarily has a solution for ω < s

= s

−



bω



= s

−







∆ρ

+ W

αβ



. (12.4.19)

294

12 Localization of Vibrations Near Extended Defects

The speciﬁc feature of this spectrum is that at ﬁxed κ the frequency of the corre-

sponding vibrations is separated from the spectrum of bulk crystal vibrations by a g ap

vanishing as κ → 0 as a power law. It follows from (12.4.11) that as κ → 0 not only

the gap width but also its relative value tends to zero (curve 3 in Fig. 14.4). Using the

simpliﬁcation (12.4.17) we can obtain

κ −ω

∼







∆ρ

+ W



. (12.4.20)

It is easily seen that these v ibrations are localized near the planar defect. Since small

wave vectors k play the main role in the integral in (12.4.10) we can take into account

the approximation bk  1 and sim plify the integral:

χ(z)=

2π

χ(0 )

∞



−∞

cos kz dk

− s

, (12.4.21)

where s is the sound velocity along the z-axis (2s = bω

Perform the integration in (12.4.21) and use (12.4.19):

(z)=χ

(0) exp



−



− ω

|z|



= χ

(0) exp



−



(κ)





= χ

(0) exp



(κ)

a |z|



(12.4.22)

One should remember that the case V

(κ) < 0 is considered. As we see, th e amp litude

χ(z) exhibits an exponential decay typical for surface waves. The rate of decay of the

amplitude depends on the d irection of the vector κ.IfV

(κ) is not negative at all κ,the

localized wave with the d ispersion relation (12.4.19) and the coordinate dependence

(12.4.22) exist only for the directions κ for which V

(κ) < 0.

The scheme presented allows one in principle to describe the vibrations of the free

surface of a crystal. Generally, the atomic interaction in a crystal decreases with in-

creasing distance between the atoms. Thus, in analyzing the equ a tions for the lattice

vibrations, only the elements of the matrix α(n) with comparatively small n are taken

into account. In a model where a small number of elements α(n) is nonzero, the free

crystal surface is equivalent to a certain planar defect. Indeed, a crystal with a free

surface can be obtained from an unbounded ideal crystal by cutting along a certain

plane between two atomic layers and replacing the force matrix of an ideal crystal

by another matrix of atomic force constants that results in the interaction between the

atomic planes vanishing.

A scalar model is convenient for describing the bulk vibrations, but is unsatisfactory

for analyzing the surface vibrations. To construct a consistent theory of long-wave

12.4 Frequency of Local Vibrations in the Presence of a Two-Dimensional (Planar) Defect

295

surface vibrations it is natural to turn to elasticity theory making some assumptions on

the properties of a planar defect that models the free crystal surface.

The vibrations localized near the planar defect are described as follows. The equa-

tions of elasticity theory for the bulk vibrations that occur with frequency ω have the

form

ρω

+ ∇

kl mn

∇

= 0, (12.4.23)

and the elasticity modulus tensor of a crystal λ

iklm

in an isotropic approximation is

reduced to the two Lamé coefﬁcients λ and G

iklm

= λδ

+ G(δ

+ δ

If a solid has a defect coin cident with the plane z = 0, the “perturbed” Lamé

coefﬁcients and the mass density are given by



= λ + Lhδ(z), G



= G + Mhδ(z) ,



= ρ + ∆ρhδ(z)=ρ [1 −(h − a)δ(z)] ,

(12.4.24)

where h is the “thickness” of the planar defect; L and M are characteristics of its

elastic properties; a is the interatomic distance along the z-axis in an ideal crystal.

Substituting (12.4.24) into (12.4.23), we get a system of equations generalizing the

scalar equation (12.4.7). It is clear that such a description of the planar defect has

a literal meaning only fo r hk  1. Therefore, the singular functions describing the

space perturbation localization in (12.4.24) must be treated with caution. Analyzing

the vibrations near the linear defect, we have seen that using the δ-like perturbation in

the long-wave approximation leads to introducing a ﬁnite upper limit of the integration

over the wave vectors. Similarly, in the case of a perturbation of the type (12.4.24)

caused by the planar defect, when we use the method of Fourier transformations along

the z-axis, it is necessary to restrict the possible values of k

by the limiting value

∼ 1/h. In the case we are interested in it is determined by

= π . (12.4.25)

The solution o f a similar elasticity theory problem has shown that the frequency of

waves localized near the defect splits off from the edge of the lowest-frequency band

of vibrations (in the isotropic approximation, from the boundary of the transverse

vibration frequency). This frequency is at a distance of δω ∼ κ

from the band

boundary, in agreement with (12.4.20).

This approach may also be used to study surface waves in an elastic half-space

(Rayleigh waves), assuming that they are localized vibrations near the planar defect.

If this planar defect is a free crystal surface, than as a result of the perturbation of

elastic moduli the connection between elastic half-spaces z > 0 and z < 0 vanishes.

The latter can be provid ed in a straigh tfo rward way by setting

L = −λ, M = −G , h = a . (12.4.26)

296

12 Localization of Vibrations Near Extended Defects

Substituting (12.4.26) into (12.4.24), taking (12.4.25) into account, and using the

described method of ﬁnding the possible frequencies of surface vibrations, one can

show that these waves h ave the dispersion relation ω = s

κ,wheres

coincides with

the known velocity of Rayleigh waves (Kosevich and Khokhlov, 1970). We do not

prove this statement here because the calculations are cumbersome.

Elastic Field of Dislocations in a Crystal

13.1

Equilibrium Equation for an Elastic Medium Containing Dislocations

Our initial deﬁnition of a dislocation



∇

= −b

(13.1.1)

may be rewritten in a somewhat different form using the notation of a distortion tensor

= ∇



= −b

. (13.1.2)

In dislocation theory the elastic distortion tensor may be r egarded as an independent

quantity describin g the crystal deformation. Similarly to the strain tensor ε

and the

stress tensor σ

it is a single-valued function of the coordinates even in the presence

of a dislocation.

Let us write down the condition (13.1 .2) in a differential form. For this purpose

we transform the integral around the contour L into an integral over any surface S

spanning this contour:





ilm

∇

The constant vector b is represented in the form of an integral over the same surface

by means of the 2D δ-function introduced in (10.3.4)



δ(ξ) dS

. (13.1.3)

The Crystal Lattice: Phonons, Solitons, Dislocations, Superlattices, Second Edition. Arnold M. Kosevich

 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

ISBN: 3-527-40508-9

298

13 Elastic Field of Dislocations in a Crystal

As the contour L is arbitrary, the equality of th ese in tegrals means the equality of the

integrand expressions:

ilm

∇

= −τ

δ(ξ). (13.1.4)

This is just the desired differential form of (13.1.1). It is clear that on the disloca-

tion line (ξ = 0), which is a line of singular points, a representation in the form of

derivatives is meaningless.

If there are simultaneously many dislocations in a crystal with relatively small sep-

arations (but large compared to the lattice constant), their averaged consideration be-

comes important. This consideration is useful in problems where an exact ﬁeld distri-

bution between separate dislocations is of no interest and in which a theory operates

with physical quantities averaged over small volume elements. It is clear that many

dislocation lines should run through such “physically inﬁnitesimal” volume elements.

The equation that expresses the main property of dislocation deformations is for-

mulated by generalizing (13.1.4). We introduce the dislocation density tensor α

requiring its integral on the surface spanned on any contour L to be equal to the sum

of Burgers vectors b of all dislocation lines enveloped by this contour:



= b

. (13.1.5)

The tensor α

replaces the expression on the r.h.s. of (13.1.4)

ilm

∇

= −α

, (13.1.6)

and describes a continuous dislocation distribution in a crystal.

It follows from (13.1.4), (13.1.6) that in the case o f a discrete dislocation

= τ

δ(ξ). (13.1.7)

As is seen from (13.1.6), the tensor α

should satisfy the condition

∇

= 0, (13.1.8)

which in the case of a single dislocation shows that the Burgers vector is constant

along the dislocation line.

With such an interpretation of dislocations, the tensor u

becomes the primary

quantity that describes the crystal deformation. The displacement vector u cannot

then be introduced. Indeed, at u

= ∇

the l.h.s. of (13.1.4) would identically

become zero over the whole crystal volume.

Equations (13.1.6) or (13.1.4), together with

∇

+ f

= 0, (13.1.9)

and Hooke’s law co nstitute a complete system of equilibrium equations of an elastic

medium with dislocations.

13.2 Stress Field Action on Dislocation

299

In a scalar elastic ﬁeld the role of a distortion tensor is played by the vector h that

in a medium without vortices is determined by

h = grad u. (13.1.10)

If vortex dislocatio ns are present in the medium, the vector h obeys the equation

curl h = −α, (13.1.11)

where α is the density of vortices. For a separate vortex, α = τbδ(ξ).

An analog of the stress tensor is the vector σ = Gh that is determined by the elastic

scalar ﬁeld equilib rium equation:

div σ = −f , (13.1.12)

where f is the analog of the force density.

13.2

Stress Field Action on Dislocation

Let us consider a dislocation loop D in a ﬁeld of external (with respect to a dislocation)

elastic stresses σ

and ﬁnd a force acting on the dislocation. We calculate the work

done by external forces δR for an inﬁnitely small movement of the loop. If this work

is given as

δR =



FδX dl , (13.2.1)

where δX is the displacement element of a d islocation line, F will determine the force

with which an elastic stress acts on a unit length of the dislocation.

Let the dislocation displacement generate a certain change in the displacement vec-

tor δu. The work of external stresses performed in a volume V with a dislocation loop

δR =



δu

∞

, (13.2.2)

where S

∞

is the surface enveloping the volume V.

In the given case it is convenient to use transformations of the integral (13.2.2)

based on Gauss’s theorem. Therefore the displacement u near a dislocation should

be considered as a single-valued function of the coordinates. Then, the vector u will

have a d iscontinuity along the surface S

spanned on the dislocation loop. To exclude

discontinuity points of the vector u, we surround the loop D with a certain closed

surface S

outside which (in the volume V



) the function u = u(r) is continuous.

300

13 Elastic Field of Dislocations in a Crystal

Then,

δR =



∇

(σ

δu

) dV



−



δu



∇

δu



δε





δu

(13.2.3)

where the unit normal vector n

is directed outwards with respect to the surface S

Since the dislocation is not associated with additional volume force in a crystal,

then ∇

= 0 and the ﬁrst integral on the r.h.s. of (13.2.3) vanishes. To calculate

the last two integrals, we choose S

as the surface going along the upper and lower

banks of the cut S

(with a “clearance” h) and connected by an inﬁnitely thin tube S

of radius ρ that envelopes the line D (Fig. 13.1).

Fig. 13.1 Scheme of closed surface surrounding the cut S

Fig. 13.2 Dilatation ﬁeld of a screw dislocation with an axis along [111]

in a α-Fe crystal calculated by Chou (1965).

13.2 Stress Field Action on Dislocation

301

If we disregard the inhomogeneity of material generated by a dislocation and that is

unimpor tant in evaluating the interactio n between the dislocation and an elastic ﬁeld ,

the volume integral remaining on the r.h.s. of (13.2.3) is simpliﬁed. Indeed, when the

cut banks come closer (h → 0) and the tube radius decreases (ρ → 0), the integral

over the volume V



, because of continuity of σ

and ε

, transforms into an integral

over the whole volume



δε

dV →



δε

dV . (13.2.4)

The surface integral in (13.2.3) can also be simpliﬁed by a corresponding limiting

transition. We note that the integral over the surface of the tube S

vanishes as ρ → 0,

since the dislocation ﬁeld has the following obvious property:

lim

ρ→0

ρu(r)=0.

On the cut banks still present in the integrals, the values of the continuous functions

in the limit are the same and the limiting values of u are different in a constant

value b. Thus, instead of (13.2.3), we g et

δR =



δε



dV +



δε

dV + b



, (13.2.5)

where σ



is the stress deviator and ε



is the strain deviator [(ε



= ε

−(1/3)δ

)].

The ﬁrst two terms of the integral (13.2.3) due to th e obvious identity in (13.2.5) follow

from



−



= σ



In the last term on the r.h.s. in (13.2.5) the symbol of an inﬁnitely small variation

δ is taken outside the integral sign since the stress distribution σ

is assumed to be

given.

If the element of the dislocation line dl is displaced by δX theareaofthesurface

element S

changes by

δS

= e

imn

δX

dl. (13.2.6)

The relation (13.2.6) should be used in transforming the last integral in (13.2.5).

If the displacement δX is in the dislocation slip plane, (13.2.6) characterizes the

changes in the crystal completely. If the displacement is perpendicular to the slip

plane, additional conditions have to be taken into account that follow from the medium

continuity. When the dislocation moves without breaking the medium continuity the

local relative variation of the volume given by (10.3.4) exists. Using this formula, a

formal replacement can be made to transform the second integral in (13.2.5)

δε

dV → e

imn

δX

dl. (13.2.7)

302

13 Elastic Field of Dislocations in a Crystal

Taking into account (13.2.6), (13.2.7), we single out in the work δR the part con-

nected with inelastic medium deformation that accompanies the dislocation migration:

δR =



δε

dV +



imn



−



δX

dl. (13.2.8)

The ﬁrst integral in (13.2.8) equals the increase in elastic energy of a medium in a

solid volume. The linear integral over the dislocation loop

δR



ilm



δX

dl (13.2.9)

will determine the work done for a dislocation displacement. Comparing (13.2.1),

(13.2.9) it is clear that the force acting per unit length is

= e

ilm



. (13.2.10)

A formula such as (13.2.10) where the stress deviator σ



is replaced by the stress

tensor σ

was ﬁrst obtained by Peach and Koehler (1950). The necessity of taking

into account an inelastic change in the medium volume in climbing of a dislocation,

i. e., the necessity to replace σ

by σ

− (1/3)δ

, was indicated by Weertman

(1965).

As simple examples of using (13.2.10), we consider the forces acting on screw and

edge dislocations. Let the Oz-axis be parallel to the dislocation line (τ

= −1). In the

case of a screw dislocation, b

= b and

scr

= bσ

, F

scr

= −bσ

. (13.2.11)

In the case of an edge dislocation, we direct the Oz-axis along its Burgers vector

= b). Then,

edg

= bσ

, F

edg

= −bσ



b(2σ

−σ

). (13.2.12)

It is interestin g to determine the projection of the force (13.2.10) on the slip plane

of the corresponding dislocation element. Let κ be a vector that is perpendicular to

the dislocation line in the slip plane. Th en, this projection (we denote it as f ) is equal

f = κF = e

ikl

f = n

, (13.2.13)

where n =[kτ] is the vector normal to the slip plane.

Since the vectors n and b are mutually perpendicular, by choosing two of the co-

ordinate axes along these vectors, we see that the force f is determined by one of the

components σ

only. If the dislocation is a plane curve lying in its slip plane and is in

a homogeneous elastic stress ﬁeld then the force f is the same for all dislocation line

elements, independent of their position on the slip plane.