See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/301881139

Simple model for Crystal Field Theory

Article · March 2016

CITATION READS
1 7,223

1 author:

Francisco Marcelo Fernández

Universidad Nacional de La Plata
746 PUBLICATIONS 5,982 CITATIONS

SEE PROFILE

All content following this page was uploaded by Francisco Marcelo Fernández on 14 June 2016.

The user has requested enhancement of the downloaded file.

arXiv:1604.00340v1 [cond-mat.mtrl-sci] 30 Mar 2016

Simple model for Crystal Field Theory

Francisco M. Fernández
INIFTA (UNLP, CCT La Plata-CONICET), División Quı́mica Teórica
Blvd. 113 S/N, Sucursal 4, Casilla de Correo 16, 1900 La Plata, Argentina

April 4, 2016

Abstract

We investigate a simple model for the prediction of the splitting of the

3d orbitals of a metal ion in the environment of ligands. The electrons
are considered to be independent and their interaction with the ligands is
represented by the Dirac delta function in three dimensions. We discuss
several cases where the model is successful and a few ones where it is not.

1 Introduction
The crystal field theory (CFT) developed by Bethe [1] long ago was utilized by
physicists to explain magnetic properties and absorption spectra of transition
metals and other compounds. This theory is commonly discussed in specialized
textbooks [2, 3] as well as in standard books on inorganic chemistry [4].
CFT treats the interaction between the metal ion and the ligands as a purely
electrostatic problem in which the ligand atoms or molecules are considered to
be mere point charges (or point dipoles). In spite of this oversimplification CFT
provides a very simple and easy way of treating numerically many aspects of
the electronic structure of complexes.
The purpose of this paper is to investigate an even simpler CFT model were
the electrostatic repulsion between the 3d electrons (assumed to be noninteract-

1
ing) and the ligands is represented by a Dirac delta function in three dimensions
centered at the ligand positions. Although this extremely short-range interac-
tion may appear to be rather too unrealistic at first sight some of the results
are surprisingly reasonable.
In section 2 we develop the model that we later apply to some examples in
section 3. We consider cases where the model is successful and others in which it
fails. Finally we summarize the main results and draw conclusions in section 4.

2 The model
The model takes into account only 3d electrons that are considered to be inde-
pendent. It is further assumed that in the absence of ligands each electron can
be described by an effective hydrogen-like Hamiltonian

h̄2 2 Ze2
H0 = − ∇ − , (1)
2m 4πǫ0 r

where Z is an effective core charge. The effect of M negative ligands around the
central metal ion is taken into account by means of a perturbation Hamiltonian
of the form
M
X
H ′ = V0 δ(r − rk ), (2)
k=1

where V0 > 0 is the strength of the electrostatic interaction between a d electron

and the ligands and δ(r − rk ) is the well known Dirac delta function in three
dimensions. In the absence of ligands the 3d hydrogen-like orbitals
n o
{ϕj , j = 1, 2, . . . 5} = ϕdz2 , ϕdx2 −y2 , ϕdxz , ϕdyz , ϕdxy , (3)

are degenerate.
In order to calculate the splitting of the 5-fold degenerate 3d level we resort
to perturbation theory. The perturbation corrections of first order to the orbital
energies are given by the roots ǫ′n , n = 1, 2, 3, 4, 5 of the secular determinant

|H′ − ǫ′ I| = 0, (4)

2
where the elements of the perturbation matrix H′ are
M
X
′
Hij = hϕi | H ′ |ϕj i = V0 ϕi (rk )ϕj (rk ). (5)
k=1

Note that the calculation reduces to obtaining the values of each 3d orbital at
the locations of the ligands.
The calculation is greatly simplified if we write the 3d real orbitals as [5]

ϕj (x, y, z) = f (ρ)uj (x, y, z)

√  3/2
2 Z Zr
f (ρ) = √ e−ρ/3 , ρ = , (6)
81a20 π a0 a0
where a0 is the Bohr radius and the form factors uj (x, y, z) read

Orbital dz2 dx2 −y2 dxz dyz dxy

. (7)
√1 1



u 12
2z 2 − x2 − y 2
2 x2 − y 2
xz yz xy

In this way, the matrix elements (5) can be written in the simpler form
M
′ λ X
Hij = ui (xk , yk , zk ) uj (xk , yk , zk ) ,
L4
k=1
λ = V0 f (ZL/a0 )2 L4 , (8)

where L is the distance between each one of the ligands and the metal nucleus.
For the time being we assume that all such distances are equal in order to
simplify the presentation.
It follows from the Hellmann-Feynman theorem
M
dǫ X
= V0−1 hψ| H ′ |ψi = ψ (rk )2 ≥ 0, (9)
dV0
k=1

that ǫ (V0 ) ≥ ǫ(0). Therefore, the matrices constructed in the way just indicated
do not satisfy the so called center of gravity rule [3, 4]. Although this rule is
irrelevant for our purposes we can build matrices that satisfy it within present
simple model. For example, the modified perturbation matrix
1
H′cg = H′ − tr (H′ ) I, (10)
5
where tr denotes the trace of a matrix and I is the 5 × 5 identity matrix, already
satisfies the center of gravity rule.

3
3 Examples
In what follows we test the predictive power of the model outlined in the pre-
ceding section on several complexes of various geometries.

3.1 Tetrahedral complex

In the case of four ligands at the vertexes of a tetrahedron we have

 
1 1 1
 
L  −1 −1 1 
 
rs = √  , (11)
3 −1 1 −1 

 
1 −1 −1

where each row of the structure matrix rs is one of the ligand position vectors
rk with modulus |rk | = L. A straightforward calculation leads to the following
diagonal perturbation matrix:
 
0 0 0 0 0
 
 0 0 0 0 0
 

 
′
H = λ 0 0 4/9 0 0 . (12)
 
 
 0 0 0 4/9 0
 

 
0 0 0 0 4/9

We appreciate that the model predicts the widely accepted splitting of the en-


ergy levels ǫ (dz2 ) = ǫ dx2 −y2 < ǫ (dxz ) = ǫ (dyz ) = ǫ (dxy ) [2–4] with the
energy interval ǫ(T2 ) − ǫ(E) = ∆T = 4λ/9.

4
3.2 Octahedral complex

The octahedral structure matrix

 
1 0 0
 
0 −1 0 
 

 
−1 0 0 
 

rs = L  , (13)
0 1 0 
 

 
0 0 1 
 

 
0 0 −1

describes four ligands located on the xy plane at the vertexes of a square and
two more at opposite sides on the z axis. It leads to the diagonal perturbation
matrix  
1 0 0 0 0
 
 0 1 0 0 0 
 
 
′
H = λ 0 0 0 0 0 , (14)
 
 
 0 0 0 0 0 
 
 
0 0 0 0 0
that exhibits the well known splitting of the energy levels ǫ (dxz ) = ǫ (dyz ) =


ǫ (dxy ) < ǫ (dz2 ) = ǫ dx2 −y2 . In this case the energy interval is ǫ(T2g )−ǫ(Eg ) =
∆O = λ. The most interesting feature of present simple model is that it correctly
predicts the ratio between the tetrahedral and octahedral splittings ∆T = 94 ∆O
that is commonly obtained by more elaborate methods [2]. This ratio is valid
provided that all the model parameters in λ (Z, L, V0 ) are exactly the same for
both complexes.

5
3.3 Cubic complex

From the structure matrix that represents eight ligands on the vertexes of a
cube  
1 1 1
 
−1 1 1
 
 
 
−1 −1 1
 
 
 
1 −1 1
 
L  
rs = √  , (15)
3
 1 −1 −1


 
1 1 −1
 
 
 
−1 1 −1
 
 
 
−1 −1 −1
we easily obtain the diagonal perturbation matrix
 
0 0 0 0 0
 
 0 0 0 0 0
 

 
′
H = λ  0 0 8/9 0 0 , (16)
 
 
 0 0 0 8/9 0
 

 
0 0 0 0 8/9


that predicts the well known splitting ǫ (dz2 ) = ǫ dx2 −y2 < ǫ (dxz ) = ǫ (dyz ) =
ǫ (dxy ) and the energy difference ǫ(T2g ) − ǫ(Eg ) = ∆C = 89 λ. Besides, the model
predicts the correct ratio between the magnitudes of the cubic and octahedral
splittings ∆C = 98 ∆O obtained by means of more elaborate methods [2]. Note
9 9
that the ratios 4 ∆T = ∆O = 8 ∆C are mentioned in most textbooks on the
field [2–4] but explicitly calculated in a few of them [2]. Undoubtedly, present
calculation is far simpler.

3.4 Square antiprismatic complex

If we rotate the four ligands on one of the cube faces perpendicular to the z axis
by an angle π/4 about that axis we obtain the structure matrix for a square

6
antiprismatic complex
 √ 
0 2 1
 √ 
 − 2 0 1 
 
 √ 
 0 − 2 1 
 
 √ 
2 0 1 
 
L 
rs = √  . (17)
3 1

−1 −1 

 
 1 1 −1 
 
 
 −1 1 −1 
 
 
−1 −1 −1

A straightforward calculation yields

 
0 0 0 0 0
 
 0 4/9 0 0 0
 

 
′
H = λ  0 0 8/9 0 0 , (18)
 
 
 0 0 0 8/9 0
 

 
0 0 0 0 4/9


that predicts the correct splitting ǫ (dz2 ) < ǫ dx2 −y2 = ǫ (dxy ) < ǫ (dxz ) =
ǫ (dyz ).

3.5 Trigonal bipyramidal complex

The structure matrix describing three ligands on the vertexes of an equilateral

triangle on the xy plane and two more on the z axis is given by
 
1 0 0
 √ 
 −1/2 3/2 0 
 
 √ 
rs = L  −1/2 − 3/2 0  . (19)
 
 
 0 0 1 
 
 
0 0 −1

7
From it we obtain the diagonal perturbation matrix
 
11/12 0 0 0 0
 
 0 3/8 0 0 0
 

 
′
H = λ 0 0 0 0 0 , (20)
 
 
 0 0 0 0 0
 

 
0 0 0 0 3/8


that predicts the well known orbital splitting ǫ (dxz ) = ǫ (dyz ) < ǫ dx2 −y2 =
ǫ (dxy ) < ǫ (dz2 ).

3.6 Pentagonal bipyramidal complex

If 5 ligands are located at the vertexes of a pentagon on the xy plane and two
more on the z axis the resulting structure matrix
 
1 0 0
 √ q√ 
 5−1 5+5
0

 4
q√ 8

 √ 
 − 5−1 5−5
0
 
 √4 8

q√ 
rs = L  − 5+1 − 5−5 , (21)
 
 √ 4 8 0 
 5−1 q √
5+5


 4 − 8 0 

 
 0 0 1 
 
0 0 −1

leads to the diagonal perturbation one

 
13/12 0 0 0 0
 
 0 5/8 0 0 0 
 
 
′
H = λ 0 0 0 0 0 , (22)
 
 
 0 0 0 0 0 
 
 
0 0 0 0 5/8


that predicts the correct orbital splitting ǫ (dxz ) = ǫ (dyz ) < ǫ dx2 −y2 =
ǫ (dxy ) < ǫ (dz2 ).

8
3.7 Square complex

If there are four ligands at the vertexes of a square on the xy plane the structure
matrix reads
 
1 0 0
 
 0 −1 0 
 
rs = L  . (23)
 −1 0 0 
 
 
0 1 0
The symmetry elements of this complex are those of the point group D4h and
its character table predicts that the degeneracy of the d orbitals should be

{dz2 }, dx2 −y2 , {dxy }, {dxz , dyz } [6]. Present oversimplified model leads to a
somewhat greater degeneracy as shown by the diagonal perturbation matrix
 
1/3 0 0 0 0
 
 0 1 0 0 0 
 
 
′
H = λ 0 0 0 0 0 . (24)
 
 
 0 0 0 0 0 
 
 
0 0 0 0 0

Note that the dxy orbital remains degenerate with {dxz , dyz } in spite of the fact
that the former and latter are bases for the B2g and Eg irreducible represen-
tations, respectively [6]. Present model also fails if we distort the octahedral
complex along the z axis leading to either D4h or C4v symmetry [6]. The reason
is that the three orbitals dxy , dxz and dyz vanish at all ligand locations and
are not affected by the distortion. We conclude that the Dirac delta interaction
appears to be unsuitable for the description of these complexes

4 Conclusions
The model developed in this paper for the splitting of the 3d orbitals of a metal
ion in a ligand environment is extremely simple and easy to apply because it only
requires the calculation of the values of the d orbitals at the ligand positions in

9
 space. It provides the well known order of energy levels in most of the cases and
even yields the relative magnitudes of the splittings. Although it fails to predict
the splitting of the orbitals in distorted octahedral complexes we think that it is
an interesting model for the qualitative (and even semi quantitative) discussion
of such molecules, particularly because the standard approach is considerably
more difficult to apply [2].

References
[1] H. A. Bethe, Termaufspaltung in kristallen, Ann. Physik 3 (1929) 133-208.

[2] C. J. Ballhausen, Ligand Field Theory, McGraw-Hill Book Company, New

York, 1962).

[3] R. G. Burns, Mineralogical Applications of Crystal Field Theory, Second

(Cambridge University Press, Cambridge, 1993).

[4] F. A. Cotton and G. Wilkinson, Advanced Inorganic Chemistry, Interscience

Publishers, New York, 1972).

[5] H. Eyring, J. Walter, and G. E. Kimball, Quantum Chemistry, John Wiley

& Sons, New York, 1944).

[6] F. A. Cotton, Chemical Applications of Group Theory, 3rd ed. (John Wiley
& Sons, New York, 1990).

10

View publication stats