STATIC: 

Strain Types
 
Version 1.25
Creation Date: 30 Aug 2002
Last Modified: 30 Aug 2002
 

Starting with Version 1.05, static, allows the user to select from a pre-defined set of lattice
types, or to define an arbitary lattice. This choice is made in the subroutine setlat.f.

In many cases, however, you may want to start with a basic lattice type, say a face-centered
cubic (fcc) lattice, and then distort the lattice. For example, you might want to calculate
the elastic constants of a crystal based on the tight-binding parametrization.

setlat.f allows you to generate an arbitrary strain, or to select one of several defined strain
types. What we mean by strain is defined in many solid-state theory textbooks, including
Kittel and Ashcroft & Mermin. In general, a strain is represented by a strain tensor, which
transforms a set of primitive lattice vectors.

To see this, let's start with an arbitrary primitive lattice:

a1 a1x a1y a1z X

( )(a2
a3
= a2x a2y a2z
a3x a3y a3z
)( )
. Y
Z
     (1)

where X, Y, and Z are unit vectors pointing along the Cartesian axis and the parenthesis
surround matrices.

A general strain will bend and perhaps change the length of the unit vectors. Since a rotation
of the coordinate axis does not change the total energy of the system, we can restrict ourselves
to non-rotating strains. These strains can be represented by a symmetric strain matrix, which
has the effect of changing the relationship between X, Y, and Z. This matrix is usually
expressed in terms of ei, the strain coefficients:

X' 1 + e1 ½ e6 ½ e5 X

( )( Y'
Z'
= ½ e6 1 + e2 ½ e4
½ e5 ½ e4 1 + e3
)( ). Y
Z
    .      (2)

The strain matrix changes the unit vectors in (1) to the primed vectors of (2). Thus the new
lattice can be described as:

a1' a1x a1y a1z 1 + e1 ½ e6 ½ e5 X

( )(
a2'
a3'
= a2x a2y a2z
a3x a3y a3z
)( . ½ e6 1 + e2 ½ e4
½ e5 ½ e4 1 + e3
)( ) . Y
Z
    .      (3)

By Hook's law, if the strains ei are small, the change in the energy of the lattice is proportional
to the square of the strains:
 + O[e3]     .      (4)

The linear terms vanish when we are at equilibrium, or if the strain causes no change in the
volume of the crystal. Otherwise, the "deltas" are related to the strain on the crystal, and the
Cij are the elastic constants. Since the strain matrix (2) is symmetric, we can assume the
Cij tensor to be symmetric, too:

Cij = Cji   .     (5)

There are thus 21 independent elastic constants. This number can be reduced by symmetry.
For example, in a cubic crystal, such as copper, niobium, or diamond, symmetry demands
that, e.g. C11 = C22. In fact, only three independent elastic constants remain, so that (4) reduces
to

E = E0 + ½ C11 [ e11 + e22 + e33] + C12 [ e1e2 + e2e3 + e3e1] + ½ C44 [ e44 + e55 + e66] + O[e3]     .    

(6)

(Equation (6) assumes that we start from equilibrium, i.e., there are no stress terms.)

It is obvious the several linear combinations of elastic constants and strains will occur
frequently. They occur so frequently, in fact, that the static program has several pre-defined
strains included in the code. These strains are defined in the setlat.f subroutine. The table
below describes the strains already defined. Any user can of course add additional strains by
editing setlat.f. In addition, as can be seen in option zero, a general strain can be entered.

In each entry of the table, only the listed ei are non-zero. The column labeled "Parameters
Entered" indicates the actual input expected by the code, in the order it should be given. The
remaining column labels should be self-explanatory, except for column 5. There, "QE/V"
means "the quadratic part of the energy (4) divided by the volume."

 
Parameters Volume
Index Strain QE/V Notes
Entered Conserved?
No strain: no additional
0 None   Yes 0 strains are imposed on the
lattice
Not in See
1 All ei All ei General arbitrary strain
general Equation(4)
e1=[(1+x)/(1-x)]½-
2 ½ (C11 + C22 - Use one of these forms when
1 ,e2=[(1-x)/(1+x)]½-1 2
x Yes 2 C12) x2 + it is known that E(-x) = E(x).
e1 = x , e2 = - O[x4] See, e.g., Example IV.
-2
x ,e3 = x2/(1-x2)
e1=[(1+x)/(1-x)]½-
21 ½ (C11 + C22 - Same as 2 or -2, except for
1 ,e2=[(1-x)/(1+x)]½-1
x Yes 2 C12) x2 + use in the case E(x)<>E(-x)
e1 = x , e2 = - O[x3] SeeExample X (C11-C22).
-21
x ,e3 = x2/(1-x2)
 3 e1=e2=[(1+x2/4] ½-1 ,e6=x Use one of these forms when
½ C66 x2 +
x2 Yes it is known that E(-x) = E(x).
-3 e3 = x2/(4-x2) , e6 = x O[x4]
See, e.g., Example V.
``Bain Path''
1/18 (C11 + for fcc and bcc Crystals.
 -1/3
e1 = e2 = (1+x) e3 = (1+ C22 + 4 C33 + 2 See Example VI. Also see
4 x yes
x)2/3 C12 - 4 C13 - 4 the E(c/a) at fixed volume in
C23) x2 + O(x3) the hcp elastic constants
tutorial
e1,2,3 = (1+x)/(1+3x)1/3 -1  2 (C44 + C55 + Analog of the ``Bain Path''
5 x yes
e4,5,6 = 2x/(1+3x)1/3 C66) x2 + O(x3) along the <111> direction.
½ C44x2 + E(-x)=E(x) in a hexagonal
6 e3 = ¼ x2 e4 = x x2 yes
O(x4) lattice. See Example X (C44).

Suggestions on how to use these strains can be found in the examples section of this manual.