|
The elastic distortion around a
straight screw dislocation of infinite length can be represented in terms of a
cylinder of elastic material deformed as defined by
Volterra. The following
illustration shows the basic geometry. |
|
|
|
|
|
|
|
|
|
|
A screw dislocation
produces the deformation shown in the left hand picture. This can be modeled by
the Volterra deformation mode as shown in the right hand picture - except for
the core region of the dislocation, the deformation is the same. A radial slit
was cut in the cylinder parallel to the z-axis, and the free surfaces
displaced rigidly with respect to each other by the distance b,
the magnitude of the Burgers vector of the screw dislocation, in the
z-direction. |
|
|
In the core region the strain is very
large - atoms are displaced by about a lattice constant. Linear elasticity theory thus is not a valid
approximation there, and we must exclude the core region. We then have no
problem in using the Volterra approach; we just have to consider the core
region separately and add it to the solutions from linear elasticity theory.
|
|
The elastic field in the dislocated
cylinder can be found by direct inspection.
First, it is noted that there are no
displacements in the
x and y directions, i.e. ux =
uy = 0 . |
|
In the z-direction, the
displacement varies smoothly from 0 to b as the angle
q goes from 0 to 2p. This can be expressed as |
|
|
|
|
|
uz |
= |
b · θ
2π |
= |
b
2π |
· tan1(y/x) |
= |
b
2π |
· arctan (y/x) |
|
|
|
|
|
|
Using the
equations for the strain
we obtain the strain
field of a screw dislocation: |
|
|
|
|
|
εxx |
= |
yy = εzz
= εxy = εyx = 0 |
εxz |
= |
εzx = |
b
4π |
· |
y
x2 + y2 |
= |
b
4π |
· |
sin θ
r |
εyz |
= |
εzy = |
b
4π |
· |
x
x2 + y2 |
= |
b
4π |
· |
cos θ
r |
|
|
|
|
|
The corresponding
stress field is also
easily obtained from the relevant
equations: |
|
|
|
|
σxx |
= |
σyy = σzz = sxy = σyx = 0 |
σxz |
= |
σzx = |
G · b
2π |
· |
y
x2 + y2 |
= |
G · b
2π |
· |
sin θ
r |
σyz |
= |
σzy = |
G · b
2π |
· |
x
x2 + y2 |
= |
G · b
2π |
· |
cos θ
r |
|
|
|
|
|
In
cylindrical
coordinates, which are clearly better matched to the situation, the
stress can be expressed via the following relations: |
|
|
|
|
|
σrz |
= |
σxy
cosθ + σyz sinθ |
|
|
|
σθ z |
= |
σxz sinθ +
σyz cosθ |
|
|
|
|
|
|
|
Similar relations hold for the
strain. We obtain the simple equations: |
|
|
|
|
|
εθ z |
= |
εzθ
= |
b
4πr |
|
|
|
|
σθ z |
= |
σzθ
= |
G · b
2πr |
|
|
|
|
|
|
The elastic distortion contains no
tensile or compressive components and consists of pure shear. szq
acts parallel to the z axis in radial planes of constant q and sqz acts in the fashion of a
torque on planes normal to the axis. The field exhibits complete radial
symmetry and the cut thus can be made on any radial plane q = constant. For a dislocation of
opposite sign, i.e. a left-handed screw,
the signs of all the field components are reversed. |
|
There is, however, a serious problem
with these equations: |
|
|
|
|
The stresses and strains are proportional to
1/r and therefore diverge to infinity
as r → 0 as shown in the
schematic picture on the left. |
|
|
|
This makes no sense and therefore the cylinder
used for the calculations must be hollow to avoid r - values that are
too small, i.e. smaller than the core radius r0. |
|
|
|
Real crystals,
of course, do (usually)
not contain hollow dislocation cores. If we
want to include the dislocation core, we must do this with a more advanced
theory of deformation, which means a non-linear atomistic theory. There are,
however, ways to avoid this, provided one is willing to accept a bit of
empirical science. |
|
|
|
The picture simply illustrates that strain and
stress are, of course, smooth functions of r. The fact that
linear elasticity theory can not cope with the core, does not mean that there
is a real problem. |
|
|
|
|
How large is radius
r0 or the extension of the dislocation core? Since the theory used is
only valid for small strains, we may equate the core region with the region
were the strain is larger than, say, 10%. From the equations
above it is seen that the strain exceeds about
0,1 or 10% whenever r »
b. A reasonable value for the dislocation core radius r0
therefore lies in the range b to 4b, i.e.
r0 ³ 1 nm in most
cases. |
|
|
|
|
The stress field of an edge
dislocation is somewhat more complex than that of a screw dislocation, but can
also be represented in an isotropic cylinder by the
Volterra construction.
|
|
|
Using the same methodology as in the
case of a screw dislocation, we replace the edge dislocation by the appropriate
cut in a cylinder. The displacement and strains in the
z-direction are zero and the deformation is basically a
"plane strain". |
|
|
It is not as easy as in the case of the screw
dislocation to write down the strain field, but the
reasoning follows the same line of arguments. We simply look at the
results: |
|
|
|
|
|
|
σxx |
= |
D · y |
3x2 + y2
(x2 + y2)2 |
σyy |
= |
D · y |
x2 y2
(x2 + y2)2 |
σxy |
= |
σyx = |
D · x |
x2 y2
(x2 +
y2)2 |
σzz |
= |
σzx = σyz = σzy = 0 |
|
|
|
|
|
|
We used the abbreviation D = Gb
/2π (1 ν)
. |
|
The stress field has, therefore, both
dilational and shear components. The largest normal stress is sxx which acts parallel to the Burgers
vector. Since the slip plane can be defined as y = 0, the maximum
compressive stress (sxx is
negative) acts immediately above the slip plane and the maximum tensile stress
(sxx is positive) acts immediately
below the slip plane. |
|
|
The effective pressure (given by the
sum over the normal
components of the stress) is |
|
|
|
|
|
p |
= |
2 · (1 + ν) · D
3 |
· |
y
x2 + y2 |
|
|
|
|
|
|
|
We thus have compressive stress above the slip
plane and tensile stresses below - just as deduced from the
qualitative picture of an
edge dislocation; graphical representation of the
stress field of an edge
dislocation is shown in the link. |
|
For edge dislocations (and screw
dislocations too), the sign of the stress- and strain components reverses
if the sign of the Burgers vector is reversed. |
|
|
Again, we have to leave out the
dislocation core; the core radius again can be taken to be about 1b -
4b |
|
We are left with the case of a
mixed dislocation.
This is not a problem anymore. Since we have a linear isotropic theory, we can
just take the solutions for the edge- and screw component of the mixed
dislocation and superimpose, i.e. add them. |
|
|
As far as "simple" elasticity theory
goes, we now have everything we can obtain. If better descriptions are needed,
the matter becomes extremely complicated! But thankfully, this simple
description is sufficient for most applications. |
© H. Föll