Chapter 8 – Symmetry in Crystal Physics – p. 1 - 9. Symmetry in ...
Transcript of Chapter 8 – Symmetry in Crystal Physics – p. 1 - 9. Symmetry in ...
Chapter 8 – Symmetry in Crystal Physics – p. 1 -
9. Symmetry in Crystal Physics
9.1. Description Physical Properties of Crystals by Tensors
Isotropic material: EχPvr
= (el. polarization, el. susceptibility, el. field); EPvr
||
physical properties, which describe the relation between vectors are generally described by scalars, implying that
both vectors are parallel.
Anisotropic materials: EP;r
are not necessarily parallel. Suitable description?
Example:
Electric susceptibility χ
vvvv EχPvr
= and vhvh EχPvr
=
In this 2D example, two scalars are required to describe the effect of the vertical field strength, same for the
horizontal field strength.
General case in three dimensional space:
Chapter 8 – Symmetry in Crystal Physics – p. 2 -
Eχχχχχχχχχ
Pvr
=
333231
232221
131211
or
3332321313
3232221212
3132121111
EχEχEχPEχEχEχPEχEχEχP
++=++=++=
Abbreviated version:
3,2,1 ; 3
1=∑=
=iEχP j
jiji
Einstein’s Notation:
3,2,1, ; == jiEχP jiji
Ultimate abbreviation for people, who don’t like summation-symbols: Summation over all
indices, which occur twice in the same term.
The dielectric susceptibility is said to be a second rank tensor.
General definition of a tensor:
A tensor is defined as a set of 3r components which describes a physical quantity. In addition
a tensor has to have special transformation properties upon a change of the basis of the
coordinate system. These transformation properties will be discussed in section 9.4.
r is called the rank of a tensor:
• Scalar: 1 = 30 component, zero-rank tensor
• Vector: 3 = 31 component, 1st-rank tensor
• Matrix: 9 = 32 components, 2nd-rank tensor
• general: 3r components, r-rank tensor
General relation between two physical properties A and B:
3,2,1,...,,,,,...,,, ; ............ == urqpnkjiAaB upqrunpqrijknijk with
• A: f-rank tensor
• B: g-rank tensor
Chapter 8 – Symmetry in Crystal Physics – p. 3 -
• a: (f+g)-rank tensor
Another example: stress, strain, elastic modulus and Hooke’s law
The stress tensor:
Body acted on by external forces: state of stress.
We consider a simple case: (1) case of homogeneous stress: stress is independent of position
in the body, (2) static equilibrium, (3) no body forces or torques.
Stress is described by 9 stress components dAdFσ /= (stress, area of body element, force
acting on area), i.e. second rank tensor ijσ :
The fact that the stress components represent a tensor is to be proved. Proves can be fount in the textbooks on
crystal physics (see literature) .
Obviously, the stress has 3 normal components 332211 ,, σσσ and 6 sheer components
323121231312 ,,,,, σσσσσσ . Condition (2) (static equilibrium) immediately requires, that there are
certain relationships between the sheer components (otherwise there would be a torque acting
on the body): 133132232112 ,, σσσσσσ === , the stress tensor is a symmetric second rank tensor:
jiij σσ = .
(Latter relationship still holds in the case of inhomogeneous stress or body forces, see textbooks).
Chapter 8 – Symmetry in Crystal Physics – p. 4 -
The strain tensor:
Strain in 1 dimension:
Strain: dldxε =
In a body, along one direction, of the material in all three directions might change. Therefore:
Extension: i
iij l
xe∂∂
=
The extension contains a symmetrical part and a antisymmetrical part. The antisymmetrical
part describes a rotation of the body, the symmetrical part describes the actual deformation:
Strain tensor ijε (symmetrical part of the extension):
++
++
++
=
3332233113
3223222112
3113211211
332313
232212
131211
)(21)(
21
)(21)(
21
)(21)(
21
eeeee
eeeee
eeeee
εεεεεεεεε
Thus, the strain tensor is also a symmetric second rank tensor.
With the previous definitions we can reformulate Hook’s law for a general solid:
klijklij σsε =
The elastic modulus ijkls is a 4th rank tensor, in principle containing 81 components.
Chapter 8 – Symmetry in Crystal Physics – p. 5 -
9.2 Examples of Tensors Representing Physical Properties
Similar as in the case of the discussed examples many physical properties can be described in
tern of tensors of different rank. Here some examples (from E. Hartmann, Introduction to
Crystal Physics):
9.3. Polar and Axial Vectors and Tensors
We consider a basis transformation from an old coordinate system ie to a new coordinate
system 'ie , described by a transformation matrix jia . Both sets of basis vectors are chosen
orthonormal, yielding ijji aa =−1 .
Chapter 8 – Symmetry in Crystal Physics – p. 6 -
Normally, a vector ip in the old coordinate system would be described in the new coordinate
system as:
jiji pap =' (note the summation over j according to the Einstein notation).
Such a (normal) vector is also called a polar vector.
In physics, there are some vectors, however, which have slightly different transformation
properties. These vectors are usually connected to a definition involving the vector product:
321
321
321
bbbaaaeee
bac
rrrrrr
=×=
(the vector product describes a vector of length ababc ∠= sin , is perpendicular to a ar and br
,
and which forms a right handed system cba rrr ,, ). We consider a basis transformation, which
changes the coordinate system from right-handed to left handed or vice versa, such as the
inversion i:
Chapter 8 – Symmetry in Crystal Physics – p. 7 -
In general, a basis transformation of a polar vector generates a change of sign, if the
transformation changes the hand. A change of hands is deduced from the determinant ija of
the transformation matrix
• 1=ija : Transformation leaves hand of the axes unchanged (rotation)
• 1−=ija : Transformation changes hand of the axes (inversion, reflection)
Transformation of a axial vector:
jijiji paap =' (note the summation over j according to the Einstein notation).
Physical examples of axial vector (1st rank axial tensor) are the angular momentum
( vrmL rrr×= ) or the magnetic flux density ( BvQF
rrr×= ).
In general a tensor a connecting two properties A and B via ............ upqrunpqrijknijk AaB = is axial if
either A or B are axial. In every other case it is polar.
9.4. Transformation Properties of Tensors
Again, we consider a basis transformation to a new coordinate system. A scalar (0-rank
tensor) does not change upon this operation:
TT =' (0-rank tensor)
A (polar) vector ip is described in the new coordinate system as:
jiji pap =' (or jiji TaT =' 1st-rank tensor)
For a (polar) 2nd-rank tensor
lklk qTp =
we obtain with kiki pap =' and jljl qaq =' or jjll qaq '= :
jijjjlkliki qTqaTap '''' == or
Chapter 8 – Symmetry in Crystal Physics – p. 8 -
kljlikij TaaT =' (2nd-rank tensor)
General transformation laws for tensors upon basis transformation:
Tensor Rank Polar Tensor Axial Tensor
0 klij TT ='
1 jiji TaT ='
jijiji TaaT ='
2 kljlikij TaaT ='
kljlikijij TaaaT ='
3 lmnknjmilijk TaaaT ='
lmnknjmilijijk TaaaaT ='
n upqrnukrjqipnijk TaaaaT ...... ...' =
upqrnukrjqipijnijk TaaaaaT ...... ...' =
9.5. Intrinsic Symmetry of Physical Properties
A k-rank tensor has up to 3n different components. However, the number of independent
components is much smaller in most cases, either due to intrinsic symmetries of the physical
property described (this section) or due to the crystal symmetry (section 9.6).
9.5.1 Symmetry by Definition
Some properties are defined such that the corresponding tensors exhibit an inner symmetry.
Examples:
Strain tensor ijε : 2nd-rank, symmetric second rank tensor, 6 independent components
Stress tensor ijσ : 2nd-rank, symmetric second rank tensor, 6 independent components
Elastic modulus ijkls : 4th rank tensor with jilkijlkjiklijkl ssss === , 36 independent components.
9.5.2 Equilibrium Properties and Thermodynamic Arguments
Chapter 8 – Symmetry in Crystal Physics – p. 9 -
For tensors describing equilibrium properties, thermodynamic relations significantly reduce
the number of independent components.
Example:
We consider elastic, electric and magnetic work plus heat exchange in a crystal (stress, strain,
el. field, el. polarization, magn. field, magn. polarization, temperature, entropy):
dTTSdH
HSdE
ESdσ
σSdS
dTTJdH
HJdE
EJdσ
σJdJ
dTTPdH
HPdE
EPdσ
σPdP
dTTε
dHHε
dEEε
dσσε
dε
kk
kk
klkl
ik
k
ik
k
ikl
kl
ii
ik
k
ik
k
ikl
kl
ii
ijk
k
ijk
k
ijkl
kl
ijij
∂∂
+
∂∂
+
∂∂
+
∂∂
=
∂∂
+
∂∂
+
∂∂
+
∂∂
=
∂∂
+
∂∂
+
∂∂
+
∂∂
=
∂∂
+
∂∂
+
∂∂
+
∂∂
=
According to the 1st and 2nd law of thermodynamics we obtain (reversible process):
TdSdJHdPEdεσdqdwdU llkkijij +++=+=
We change the set of independent variables by introducing a Gibb’s free energy:
TSJHPEεσUdG llkkijij −−−−=
yielding
SdTdHJdEPdσεdG llkkijij ++−−=
From comparison with the total differential of
dTTGdH
HGdE
EGdσ
σGdG l
lk
kij
ij
∂∂
+
∂∂
+
∂∂
+
∂∂
=
we obtain
STGJ
HGP
EGε
σG
kk
kk
ijij
−=
∂∂
−=
∂∂
−=
∂∂
−=
∂∂ ; ; ;
From the commutability of the second derivatives (Schwartz theorem) it follows for the
dielectric susceptibility jkχ :
Chapter 8 – Symmetry in Crystal Physics – p. 10 -
jkk
j
jkkjj
kkj χ
EP
EEG
EEG
EPχ =
∂∂
=
∂∂∂
−=
∂∂∂
−=
∂∂
=22
Therefore, the dielectric susceptibility tensor is symmetric. Similar arguments hold for the
magnetic susceptibility )/( kllk HJψ ∂∂= and the elastic modulus )/( klijijkl σεs ∂∂= .
Moreover it follows that the tensors describing direct and reciprocal effects are identical. A an
example we consider the piezoelectrical (stress -> el. polarization) and the reverse
piezoelectrical effect (el. field -> strain):
kijk
ij
ijkkijij
k dEε
σEG
EσG
σP
=
∂∂
=
∂∂∂
−=
∂∂∂
−=
∂∂ 22
Similar relations are found for the piezomagnetic effect )/()/( lijijllij HεσJq ∂∂=∂∂= , the
magneto-electrical polarization )/()/( lkkllk HPEJλ ∂∂=∂∂= , thermal dilatation and
piezocaloric-effect )/()/( ijijlk σSTεα ∂∂=∂∂= , pyroelectric and electrocaloric effect
)/()/( kkk ESTPp ∂∂=∂∂= and pyromagnetic and magneto-caloric effect
)/()/( lll HSTJm ∂∂=∂∂= . Thus the above set of equations can be simplified significantly:
TcddHmdEpdσdSdTmdHdEdσqdJdTpdHdEdσddPdTdHqdEddσsdε
klkkijij
lklmklkijlijl
kklkkklijkijk
ijllijkkijklijklij
ln+++=+++=+++=+++=
αψλλχ
α
9.5.3. Transport Properties and Onsager’s Principle
In irreversible thermodynamics, transport processes are described by sets of corresponding
thermodynamic forces iX and fluxes ij , chosen such that ii Xjσ =& corresponds to the rate
of entropy production. For the corresponding linear system of transport equations
jiji XLj = ,
Onsager’s reciprocity relation states that
Chapter 8 – Symmetry in Crystal Physics – p. 11 -
jiij LL = .
Example: Electric and heat transport
• Electric transport: force k
kq
xTX
∂Φ∂
=1 (el. potential); flux: electrical flux density k
qj
• Linear transport equations: k
ikq
iq
xTLj
∂Φ∂
=1 (classical definition:
kiki
q
xj
∂Φ∂
−= σ )
è Onsager relation: kiik LL = or kiik σσ = (conductivity tensor is symmetric)
• Heat transport: force k
kQ
xTX
∂∂
=)/1( ; flux: heat flux density k
Qj
• Definition of Peltier and Seebeck effect:
kq
ikk
ikiQ jπ
dxdTkj +−= with
lklk
q
dxdσj Φ
=
kikkiki
q
dxdσEσj Φ
+= β withl
klkβ
dxdTβE −=
with dxdTT
dxdT
dTTd
dxTd 2)/1()/1( −== it follows:
lklik
kiki
Q
dxd
TTσπ
dxTdTkj Φ
+−−=1)/1()( 2
kik
lkliki
q
dxd
TTσ
dxTdTβσj Φ
+−−=1)/1()( 2
è Onsager relation: klkl πTβ = (relation between Seebeck and Peltier effect)
9.6 Neumann’s Principle: Crystal Symmetry and Tensor Symmetry
CT GG ⊇ with ( ⊇ : subgroup or equal)
TG : symmetry group of tensor T
CG : symmetry group of crystal C
Chapter 8 – Symmetry in Crystal Physics – p. 12 -
The group of symmetry elements of a physical property includes all symmetry elements of the
crystal, i.e.
• Polar tensor: upqrnukrjqipnijk TaaaaT ...... ...=
• Axial tensor: upqrnukrjqipijnijk TaaaaaT ...... ...=
Must be fulfilled for all transformation matrices ija corresponding to symmetry operations of
the point group of the crystal (crystal class) (remark: in general, it is not necessary to test all
symmetry operations, but only the set of generating operations of the point group).
Neumann’s principle may reduce the number of independent components of a tensor or may
require that some tensor elements must vanish.
Example 1:
We consider the el. susceptibility tensor ijχ defined via jiji EχP = (polar symmetrical tensor
of rank 2). We consider a crystal which belongs to crystal class 4 (crystallographic) or C4
(Schoenfliess).
The generating element is C4.
−=
100001010
4Ca .
We obtain:
Chapter 8 – Symmetry in Crystal Physics – p. 13 -
−−
−−=
−−−
−=
−
−=
331323
131112
231222
331323
231222
131112
332313
232212
131211
332313
232212
131211
100001010
100001010
100001010
χχχχχχχχχ
χχχχχχχχχ
χχχχχχχχχ
χχχχχχχχχ
Thus:
(a) 2211 χχ =
(b) 0122112 =⇒−= χχχ
(c) 02313232313 ==⇒−== χχχχχ
Therefore, the electric polarizability tensor of a crystal belonging to class 4 has only two
independent components:
=
33
11
11
332313
232212
131211
000000
χχ
χ
χχχχχχχχχ
Example 1:
We consider the process of second harmonic / sum frequency generation, which is described
by the second order hyperpolarizability ijkχ , a polar tensor of rank 3
( ...++= kjijkjiji EEχEχP ; for a fields ωE , 'ωE oscillating with frequency ω , 'ω , ijkχ
generates frequency components with frequency 2ω , 2 'ω , 'ωω + , 'ωω − ):
Chapter 8 – Symmetry in Crystal Physics – p. 14 -
We consider a crystal class which contains the inversion i described by the transformation
matrix
−−
−=
100010001
ia or
≠=−
=−=jiji
a ijiji
;0;1
δ
Neumann’s principle requires that
lmn
lmn
lmnlmn
knjmil
lmnkni
jmi
ili
ijk aaa
χχ
χδδδ
χχ
−=−−−=
−−−=
=
∑)1)(1)(1(
))()((
Therefore, 0=ijkχ , i.e. second harmonic / sum frequency generation is forbidden in crystals /
media with inversion symmetry. Note: In media with inversion symmetry, all processes
described by polar tensors of odd order and axial tensors of even order are forbidden!
9.8 Contracted Matrix Notation
In literature, a special matrix notation is often used to simplify the representation of higher
rank tensors.
Example:
Hook’s law: klijklij σsε =
The elastic modulus is a 4th rank tensor with a maximum of 81 components, but symmetry
with respect to suffixes i,j and k,l reduces the number of free components to 36.
We contract the index pairs i,j to a single index m and k,l to n according to the following rule:
6)2,1( ;5)1,3( ;4)3,2( ;3)3,3( ;2)2,2( ;1)1,1( →→→→→→
Using the definition
==
=6,5,4 ;23,2,1 ;
mεmε
εij
ijm
Chapter 8 – Symmetry in Crystal Physics – p. 15 -
6,5,4,3,2,1 ; == nijn σσ
=====
=
=
6,5,4, ;46,5,4;3,2,1 ;26,5,4;3,2,1 ;2
3,2,1, ;
nmsnmsnms
nms
s
ij
ij
ij
ij
mn
Hook’s law is expressed as:
=
6
5
4
3
2
1
666564636261
565554535251
464544434241
363534333231
262524232221
161514131211
6
5
4
3
2
1
σσσσσσ
εεεεεε
cccccccccccccccccccccccccccccccccccc
or mnmn s σε =
Remarks:
(1) The coefficients in the definition of the contracted notations are necessary to take into
account the reduced number of terms in the summations.
(2) This is only a notation trick! The transformation properties remain unchanged.
9.9 Value of a Physical Property in a Given Direction
The value T of a physical property described by a 2nd rank tensor via jiji qTp = in the
direction of qr is defined as qpT /||= , where ||p is the component of pr parallel to qr .
Example:
Chapter 8 – Symmetry in Crystal Physics – p. 16 -
Electrical conductivity σ , electrical field: nEE rr= with nr : unit vector in field direction.
EEE
EE
EEj
EEjnjj ij
iji
i σ==⋅=⋅=r
rrr||
With Ejn /||=rσ it follows:
jiijn nnσσ =r
This equation can be used to derive tensor components from a physical measurement or vice
versa.
9.10 Geometrical Representation: The Representation Quadratic
For the important group of symmetric 2nd-rank tensors (as an example we again consider the
conductivity jiji Ej σ= ), there is a simple geometrical way of representation, the so called
representation quadratic defined as:
1=jiij xxσ or 1222 2112133132232333
2222
2111 =+++++ xxxxxxxxx σσσσσσ
This is a second degree surface, in most cases it corresponds to an ellipsoid (tensor ellipsoid):
It can be shown that upon basis transformation the representation quadratic behaves like a
symmetrical 2nd-rank tensor. Thus the transformation properties of the tensor can be derived
from a (graphical) inspection of the transformation properties of the representation quadratic.
Chapter 8 – Symmetry in Crystal Physics – p. 17 -
We can choose a basis transformation to a coordinate system, in which ijσ is diagonal:
=
3
2
1
'000'000'
'σ
σσ
σ
The directions of this special set of basis vectors are referred to as the principle axes of the
tensor. In the new basis, the tensor ellipsoid points along the coordinate axes:
the representation quadratic takes a simple form
1'''''' 2333
2222
2111 =++ xxx σσσ
and physical equations involving the tensor become particularly simple:
111 ''' Ej σ= , 222 ''' Ej σ= , 333 ''' Ej σ= .
The representation quadratic has two important geometrical properties (example: el.
conductivity jiji Ej σ= ):
(a) The radius r in a given direction nr is related to the physical property nrσ in this
direction via 2/1 rn =rσ . The resulting component of jr
parallel to nEE rr= is
Ej nrσ=|| .
(Proof: From 1=jiij xxσ with ii rnx = , we obtain 12 =jiij nnr σ . With the result from
section 9.9 ( jiijn nnσσ =r ), we obtain 2/1 rn =rσ .)
Chapter 8 – Symmetry in Crystal Physics – p. 18 -
(b) The direction of jr
is along the normal of the representation quadratic at the endpoint
of the radius (without proof, see textbooks).
Example 2: Optical properties of crystals
Optical properties of an isotropic medium:
• EDrr
εε0= with εε0 dielectrical permittivity or
EDrr
=ηη0 with ηη0 dielectrical impermeability
• Maxwell equations ( 1=µ ):
velocity of electromagnetic wave ηε
ccv ==
refractive index: η
ε1
===vcn
Optical properties of an anisotropic medium:
• jiji ED εε0= with ijεε 0 dielectrical permittivity tensor or
ijij ED =ηη0 with ijηη0 dielectrical impermeability tensor (both 2nd-rank symmetric)
• Maxwell equations ( 1=µ ): (for proof see textbooks, e.g. Nye) In general, two plane
polarized waves with different velocity may be propagated along one direction
(double refraction).
Chapter 8 – Symmetry in Crystal Physics – p. 19 -
Graphical representation: We consider the representation quadratic of the relative dielectric
impermeability tensor ijη , the so called indicatrix (note: (a) principal axes are chosen; (b) in
are called the principal refractive indices, but the refractive index is not a tensor!):
1233
222
211 =++ xxx ηηη or 12
3
23
22
22
21
21 =++
nx
nx
nx
The indicatrix has the following important property (for lengthy proofs see textbooks):
We consider wave propagation along 0P. The central section through the indicatrix,
perpendicular to the propagation direction is an ellipse. The axes of this ellipse represent the
two polarisation of Dr
and the semi-axes 0A and 0B are identical to the refractive indices
An and An for the two waves.
From these properties of the indicatrix and Neumann’s principle we can immediately classify
all crystal classes with respect to their optical properties:
(a) Optical anaxial crystals: cubic (classes 23 , 3m , 432 , 324 , mm3 ):
• indicatrix is a sphere (several Cn / Sn axes with n>2)
Chapter 8 – Symmetry in Crystal Physics – p. 20 -
⇒ no double refraction in any direction
(b) Optical uniaxial crystals: teragonal ( 4 , 4v
, m4 , 422 , mm4 , m24 , mmm4 ), trigonal
( 3 , 3 , 32 , m3 , m3 ), hexagonal ( 6 , 6 , m6 , 622 , mm6 , 26m , mmm6 )
• indicatrix is an ellipsoid of revolution along principal symmetry axes (one Cn / Sn axes
with n>2)
⇒ no double refraction along principal symmetry axis (one optical axis)
(a) Optical biaxial crystals: triclinic (1 , 1 ), monoclinic ( 2 , m , m2 ), orthorhombic ( 222 ,
2mm , mmm ):
• indicatrix is a triaxial ellipsoid (no Cn / Sn axes with n>2)
Chapter 8 – Symmetry in Crystal Physics – p. 21 -
⇒ no double refraction along two axes (optical axes)
9.11 Curie’s Principle
Often, crystal properties are considered under some external influence (electrical field, strain,
etc.). Here, Curie’s principle states:
ECC GGG ∩=~ with ( ∩ greatest common subgroup)
CG ~ : symmetry group of crystal C under external influence of E
CG : symmetry group of crystal C
EG : symmetry group of external influence E
Example: electro-optical and photoelastic effects
We consider the change of the relative dielectric impermeability tensor under the influence of
an electric or stress and expand it in term of a power series:
...0 ++++= klijkllkijklkijkijij EEkEr σπηη
• :ijkr linear electro-optical tensor → linear electro-optical effect (Pockels effect)
• :ijklk quadratic electro-optical tensor → quadratic electro-optical effect (Kerr effect)
• :ijklπ piezoptical tensor → photoelastic effect
As an example, we consider the Pockels effect in ADP (ammonium-dihydrogen-phosphate).
The crystal class of ADP is m24 , belonging to the tetragonal system. The crystal is optical
Chapter 8 – Symmetry in Crystal Physics – p. 22 -
uniaxial. For wave propagation along the principal symmetry axis, no double diffraction
occurs.
We apply an electric field along the principal symmetry axis. The group of the electric field
vector contains an ∞–fold rotation axis and ∞ mirror planes containing the axis (group )m∞ .
As a set of common symmetry elements two mirror planes and a C2 axis survive. We identify
the common subgroup 2mm , which belongs to the orthorhombic crystal system, i.e. an optical
biaxial group.
→ The optical uniaxial ADP crystal becomes biaxial if an electric field along z
(principal symmetry axis) is applied and double refraction occurs along z.
Electro-optical and photoelastic effects are usually employed in optical elements which
change the direction of polarization or modulate the intensity.