Quantum Chemical Methods

it is not possible to solve the Schrodinger equation exactly for a many-body

system
N
1 2 M N Z I N 1 N 1
H elec   i    
ˆ electron-electron terms
are the problem
i 1 2 I 1 i 1 rIi i 1 j i rij

Strategies used in quantum chemistry:

1. treat electron-electron interactions as a minor perturbation
• called perturbation theory
• more on this later

2. „guess‟ a suitable form of  and optimize it

• called variational method
• basis for most of quantum chemistry
• we‟ll look at this now

21
Variational Method
Example for the Hydrogen atom:
1/ 2
1
0.8000

real 1s:     e r real wavefunction

 
0.7000

0.6000

0.5000

R(r)
0.4000

0.3000

0.2000

0.1000

0.0000
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

r [bohr]

22
Variational Method
Example for the Hydrogen atom:
1/ 2
1
0.8000

real 1s:     e r real wavefunction

 
0.7000
trial, alpha = 0.5
trial, alpha = 0.75
0.6000
trial, alpha = 1.0

0.5000

R(r)
0.4000

 2 
3/ 4

trial 1s:    e r

2
0.3000


   0.2000

0.1000
we want to pick  to give the best
0.0000
agreement with the real solution 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

r [bohr]

Obvious questions:
1. can we really do this?
2. how do we „optimize‟  (i.e. how do we pick )?
3. what form of  should we use?

23
Variational Principle – Some Background
there exists an infinite set of wavefunctions, {}, such that:
Hˆ   E    0,1, 2,...

where the energies E are the eigenvalues for a particular state:

E0  E1  E2 ...  E ...

 i  j d  ij
 *
since the wavefunctions are orthonormal:

* ˆ
 i j  i E j  j d
     *
H d

 i H  j d  E j ij
* ˆ
 for i  j  ij  1

E j   *j Hˆ  j d

24
Variational Principle – Some Background
E j   *j Hˆ  j d energy of a specific state

• we can‟t solve this because we don‟t know what j is

variational approach (basic idea):

• let‟s take a guess at the form of j and try to get as close as
possible to the real wavefunction
• trial wavefunction must meet certain constraints to be valid
• more on this later

4th postulate of quantum mechanics:

• an arbitrary wavefunction can be expressed as a linear
combination of the set {i}

   ci  i
i

25
Variational Principle – Some Background
trial wavefunction:    ci  i
i
• choose  to be orthonormal
• don‟t know exact form of {i}

      i  c j  j d

2 * *
d 1 ci
i j

  ci*c j  *i  j d
i j

  ci*c j ij
i j

  ci for i  j  ij  1
2

  ci  1
2

26
Variational Principle – Some Background
trial wavefunction:
Let‟s consider evaluating the energy associated with the trial wavefunction :

   ci  i   d   ci  1
2 2

i i
• choose  to be orthonormal
• don‟t know exact form of {i}

 * * ˆ  
  H d    i ci i H  j c j  j  d
ˆ
*

  ci*  c j  *i Hˆ  j d
i j

  ci*  c j E j ij
i j

  ci E i for i  j  ij  1
2

i
Thus, the energy associated with  is determinable from all the coefficients ci
27
and their associated energies Ei
Variational Principle – Some Background
what we know:

   ci  i  d   ci  1  H d   ci Ei
ˆ
2 2 2

i i i

compare the trial energy with the ground state energy:

 * ˆ
   0     i i  i E0

2 2 2
H d E d c E c
i i

28
Variational Principle – Some Background
what we know:

   ci  i  d   ci  1  H d   ci Ei
ˆ
2 2 2

i i i

compare the trial energy with the ground state:

 * ˆ
   0  d   ci  Ei  E0 
2 2
H d E
i

29
Variational Principle – Some Background
what we know:

   ci  i  d   ci  1  H d   ci Ei
ˆ
2 2 2

i i i
compare the trial energy with the ground state:

 * ˆ
   0  d   ci  Ei  E0 
2 2
H d E
i

Both ci2 and Ei ‒E0 must be equal to or greater than zero:

 ci  Ei  E0   0
2

* ˆ
     0  d  0
2
Thus, we have H d E
and with some rearrangement:

 Ĥ d
 *
• this is called the variational
 E0 integral
 d
2

30
Variational Principle

 Ĥ d
energy of trial
function, 
 *
true ground state energy of
 E0
 d
2 the system
normalization
factor

Variational Theorem
The expectation value of the energy is always greater than or equal to the
true energy

for any suitable trial wavefunction:

• the energy of the trial function is guaranteed to be higher than the real
ground state energy
• if we happen upon the real wavefunction we get E0
• otherwise, we get an upper bound on E0
• basic principle: trial wavefunction with lower energy is a
„better‟ trial wavefunction
31
Construction and Optimization of Trial Wavefunctions

 Ĥ d
 *

 E0 nice concept, but how do we use it?

 d
2

Linear variation:
1. pick a trial function that is a linear combination of basis functions
N
   aii well-defined functions
i 1

coefficients in linear expansion

2. modify coefficients {ai} to minimize the energy

the „best‟ trial wavefunction gives the lowest energy

32
Construction and Optimization of Trial Wavefunctions
trial function is a linear combination of basis functions
N
   aii
Back to the Hydrogen atom: i 1

a single gaussian function is not a good trial wavefunction!!!

1/ 2
1
real 1s:     e r 0.8000

  0.7000
real wavefunction
trial, alpha = 0.5

0.6000

 2 
1 3/ 4

trial 1s:    ai  i  ei r

2 0.5000

  
R(r)

i 1
0.4000

0.3000

a1=1.0, 1=0.5 0.2000

0.1000

0.0000
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

r [bohr]

33
Construction and Optimization of Trial Wavefunctions
trial function is a linear combination of basis functions
N
   aii
i 1
Back to the Hydrogen atom:
a linear combination of gaussian functions is a
1/ 2 good trial wavefunction!!!
1
real 1s:     e r 0.6000

  0.5000
real wavefunction
trial wavefunction

 2 
3 3/ 4

trial 1s:    ai  i 
0.4000
i r 2
e
  
R(r)

i 1 0.3000

a1=0.1, 1=3.5
0.2000

a2=0.5, 2=0.6 0.1000

a3=0.4, 3=0.2 0.0000

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

r [bohr]
optimized to fixed to define the basis function
minimize energy 34
Linear Variation
the best trial wavefunction minimizes the energy

to „optimize‟ , we need a relationship between E and {ai}

 N * * ˆ  N  N N
* ˆ    a   H   a   d   a *
a * ˆ
i j  i H  j d

    i i j j
H d  i 1   j 1   i 1 j 1
E 
       
* N N


N N


d
   i i  j j 
     j d
* *
a * *
a d a i a j i
i 1   j 1  i 1 j 1

in compact notation:
N N

 ai*a j H ij
i 1 j 1
H ij   i* Hˆ  j d
E N N

 i a j Sij
a *

i 1 j 1
Sij   i* j d

35
Linear Variation
N N

 a a H *
i j ij
• energy of the trial wavefunction

E
i 1 j 1 • we want to minimize this energy for all possible
N N

 a a S * linear combinations of th basis function

i 1 j 1
i j ij
 lower energy = better wavefunction
• Hij and Sij are constants
• ai‟s are variables
Minimization:
From calculus, a necessary condition for a function (i.e., energy) to be at its
minimum is that Its derivatives w. r. t. all of its free variables are zero.
E
 0 for all k
ak
Performing this partial differentiation gives rise to N secular equations:
N

a  H
i 1
i ki  ES ki   0 for all k

36
Secular Equations
N

a  H
i 1
i ki  ES ki   0 for all k

This set of N equations (running over k) involves N unknowns (ai‟s)

From linear algebra, a set of N equations in N unknowns has a non-trivial
solution if and only if the determinant formed from the coefficients of the
unknowns (Hki‒Eki) is equal to zero, that is

H 11  ES 11 H 12  ES 12 H 1N  ES 1N
H 21  ES 21 H 22  ES 22 H 2 N  ES 2 N
0 Secular determinant

H N 1  ES N 1 H N 2  ES N 2 H NN  ES NN

• to solve, we need E first

• get E from secular determinant
• the Ej‟s can be used to get a set of ai‟s for each state by solving the
secular equations
37
Secular Equations
H 11  ES 11 H 12  ES 12 H 1N  ES 1N
H 21  ES 21 H 22  ES 22 H 2 N  ES 2 N we can get N trial wavefunctions
0 from N basis functions
H N 1  ES N 1 H N 2  ES N 2 H NN  ES NN

E0, E1, E2, …,EN-1

N
 0   ai 0i
i 1
N

a  H
i 1
i 1i  ES 1i   0 N
1   ai1i
N i 1
 ai  H 2i  ES 2i   0
i 1

N
 N 1   ai N 1i
N

a  H
i 1
i Ni  ES Ni   0
i 1
38
Linear Variation - Summary
Procedure: N

1. define a trial function by selecting a set of N basis functions :   a

i 1
i i

2. For that set of basis functions, evaluate Hij and Sij

3. construct the secular determinant , and determine the N roots Ej of the

secular equation.
N 
4. solve  ai  H1i  ES1i   0 for each of the N values of Ej to determine aij
 i 1  for that MO.
Results:
set of coefficients {ai} for each state j that give lowest energy for a
specific trial wavefunction

Note:
• more basis functions, i, gives more coefficients to „tweak‟
• leads to a lower energy, and better wavefunction

39
Variational principle
All the properties of a molecule can be found if the wavefunction is known: e.g., the
electron density distribution or the partial atomic charge. The wavefunction can be
found by solving the Schrödinger equation…. But the latter is impossible to solve
analytically!! We use the variational principle to go around that problem and find
approximate wavefunction.

The variation principle is the basic principle to determine wavefunction of complicated

molecular systems. The idea is to optimize the coefficient cA and cB of the
wavefunction,  =cA A + cB B, such that the system is the most stable, i.e. the energy
is minimal.

Example: The Hydrogen Molecule H-H

 we‟ll try to find the best coefficients ci to give to the trial wavefunction  in order
to approach the true energy.
 assume the wavefunctions for hydrogens A and B are known.
 we need a wavefunction  for hydrogen molecule. How to build it?
LCAO principle
 = cAA + cB B H=E E >Etrue

Aim: Try to find the (cA, cB) that minimize the calculated E
 Energy in the LCAO approach

(1) E
  * H  d

 * d
 Numerator:
 * H  d   c AA  cBB H c AA  cBB  d
 c 2A   AH A d  cB2  B HB d  c AcB   AHB d  c AcB  B H A d
 A   A HAd  B   B HBd    A HBd   B HAd
A is a Coulomb integral: it is related to the energy of the e- when it occupies A. ( < 0)
 is a Resonance integral: it is zero if the orbital don‟t overlap. (at Re, <0)

         B  2c AcB 
2 2
* H d c A A c B (1)

 Denominator: c 2A A  cB2  B  2c AcB 

E 2
  2
d   c A A  c BB 2
d c AS AA  cB2 S BB  2c AcB S AB
 c 2A   A2 d  cB2  B2 d  2c AcB   AB d
(1)
 c 2AS AA  cB2 S BB  2c AcB S AB
How to get the secular equations?
c A2 A  cB2 B  2c AcB 
E 2
c AS AA  cB2 SBB  2c AcB S AB
Let‟s find the coefficients cA and cB to stabilize the energy of the system w.r.t. both cA
and cB
Let‟s start with E c A  0
cA2 A  cB2B  2cAcB   E (cA2 S AA  cB2 SBB  2cAcB S AB )
cA2 ( A  ES AA )  cB2 (B  ES BB )  2cAcB (   ES AB )  0

2cA A  2cAES AA  2cB   2cB ES AB  0

cA ( A  ES AA )  cB (   ES AB )  0

Sij  0,  1, i j
i j
Hückel approximation

 c A ( A  E )  cB   0

Let‟s continue with E cB  0

we arrived at a system with two simultaneous equations:
 c A ( A  E )  cB   0
Secular equations
c A  cB ( B  E )  0

In order to have a solution, other than the trivial solution cA = cB = 0, we must

have:
A  E 
0 Secular determinant should be zero
 B  E
The 2 roots give the energies of the molecular
 A  E   B  E    2
0 orbitals formed from the AOs

E  E
1 x
  x 1
x 2  1  x  1, x  1
 0  0 
E 1 x
1


In this example the secular equation has two solutions and the H2 molecule has
two MOs optimized by energy:

c1  1, c2  1  1   A  B , E  1
c1  1, c2  1  2   A  B , E  1
Homonuclear diatomic molecules
1) =cAA + cB B : with A = B → A= B= 

E  (1)
  E 2    SE 2  0 1 S
 
E  (2)
1 S
1 cB  c A
cA 
 
c A   E  cB   SE   0 (1) 21  S 1/ 2 bonding

 
c A   SE   cB   E  0 (2) cA 
1
antibonding
21  S 1/ 2 cB  cA

0
 
E 
 1 S 
antibonding= {2(1-S)}-1/2(A - B)


E 
1 S

bonding= {2(1+S)}-1/2(A + B)

 0
 
E 
1 S 

Eantibonding= - E-

E 
1 S
Ebonding = E+-

  S
Eantibonding    E 
1 S Since: 0 < S < 1  Eantibonding >
  S Ebonding
Ebonding  E   
1 S

Note 1: He2 has 4 electrons  ground-state configuration: 12 2*2  He2 is not
stable!

Note 2: If we neglect the overlap integral (S=0), Eantibonding = Ebonding = 

 The resonance integral  is an indicator of the strength of covalent bonds
Molecular orbitals of small conjugated molecules
The Hückel approximation
Here, we investigate conjugated molecules in which there is an alternation of
single and double bonds along a chain of carbon atoms.
In the Hückel approach, the  orbitals are treated separately from the  orbitals,
the latter form a rigid framework that determine the general shape of the
molecule.
All C are considered similar → only one type of coulomb integral  for the C2p
atomic orbitals involved in the  molecular orbitals spread over the molecule.
The secular determinant
The  molecular orbitals are expressed as linear combinations of C2pz atomic
orbitals (LCAO), which are perpendicular to the molecular plane.

Ethene, CH2=CH2: =cA A+cBB, where A and B are the C2pz orbitals of each
carbon atoms.
 Butadiene, CH2=CH-CH=CH2: =cAA + cB B +cc C + cD D
The coefficients can be optimized by the same procedure described before:
express the total energy E as a function of the ci and then minimize the E with
respect to those coefficients ci. Inject the energy solutions in the secular equations
and extract the coefficients minimizing E.
 Following these methods and since A= B= ,
we obtain those secular determinants:

  E    SE 
 Ethene, CH2=CH2: 0
  SE    E 
  E   AB  S AB E   AC  S AC E   AD  S AD E 
 Butadiene:  BA  S BA E    E   BC  S BC E   BD  S BD E 
0
 CA  SCA E   CB  SCB E    E   CD  SCD E 
 DA  S DA E   DB  S DB E   DC  S DC E    E 

Hückel approximation:
1) All overlap integrals Sij= 0 (i≠j).
2) All resonance integrals between non-neighbors, i,i+n=0 with n> 2
3) All resonance integrals between neighbors are equal, i,i+1= i+1,i+2 =
 Severe approximation, but it allows us to calculate the general picture of the
molecular orbital energy levels.
Ethene and frontier orbitals
Within the Hückel approximation, the secular determinant becomes:
  E  
   E    2  0
2

   E 
E- =  -  energy of the Lowest Unoccupied
Molecular Orbital (LUMO)
E+ =  +  energy of the Highest Occupied Molecular Orbital
(HOMO)
LUMO= 2*

2||

HOMO= 1

 HOMO and LUMO are the frontier orbitals of a molecule.

 those are important orbitals because they are largely responsible for many
chemical and optical properties of the molecule. Note: The energy needed to
excite electronically the molecule, from the ground state 12 to the first excited
state 11 2*1 is provided roughly by 2|| ( is often around -0.8 eV)
 Butadiene and delocalization energy

  E   0 0
   E   0  4th order polynomial  4 roots E
0
0    E  
0 0    E  There
is 1e- in each 2pz orbital of the four carbon
atoms  4 electrons to accommodate in the 4 -
type molecular orbitals  the ground state
3 nodes
configuration is 12 22
= E4
 The greater the number of internuclear nodes,
the higher the energy of the orbital
2 nodes
 Butadiene C4H6: total -electron binding energy,
= E3
E is E = 2E1+2E2= 4 + 4.48 with two -bonds
LUMO= 3*
 Ethene C2H4:E = 2 + 2 with one -bond
= E2  Two ethene molecules give: E = 4 + 4 for two
1 node separated -bonds.
HOMO= 1  The energy of the butadiene molecule with two -
bonds lies lower by 0.48 (-36kJ/mol) than the sum
= E1 of two individual -bonds: this extra-stabilization of a
0 node conjugated system is called the “delocalization
energy”
Top view of the MOs
Variational Method
Example for the Hydrogen atom:
1/ 2
1
0.8000

real 1s:     e r real wavefunction

 
0.7000
trial, alpha = 0.5
trial, alpha = 0.75
0.6000
trial, alpha = 1.0

0.5000

R(r)
0.4000

 2 
3/ 4

trial 1s:    e r

2
0.3000


   0.2000

0.1000
we want to pick  to give the best
0.0000
agreement with the real solution 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

r [bohr]

Obvious questions:
1. can we really do this?  variational principle
2. how do we „optimize‟  (i.e. how do we pick )?  linear variation method
3. what form of  should we use?

50
Trial Wavefunctions
how do we select a trial wavefunction for molecules?
1. Assumptions
• each electron occupies a molecular orbital (analogous to atomic orbitals)
• molecular orbitals are orthonormal

2. Constraints
• trial wavefunction must satisfy Pauli exclusion principle
• trial wavefunction must be antisymmetric
• trial wavefunction must not distinguish between electrons

51
Molecular Orbitals
in atoms: • each electron occupies 1 atomic orbital
e.g. 1s, 2s, 2pz, …
• atomic orbitals are centered on nucleus
• atomic orbitals are orthonormal

in molecules: • each electron occupies 1 molecular orbital

• molecular orbitals can be delocalized across all atoms
• molecular orbitals are orthonormal

C-C -bond in ethene C-C -bond in ethene

52
Molecular Orbitals
molecular orbitals depend on spatial coordinates and spin
Spatial orbital, i(r): • function of the position vector, r
• describes the spatial distribution of the electron:
 i (r) dr  probability of finding electron i in dr
2

• form an orthonormal set:

  (r) j (r)   ij
*
dr i

Spin functions:
• Helec does not include spin  would need a relativistic treatment to get spin
• artificially introduce orthonormal spin functions: g() = () or ()

 d  *
         d  *
       1
 d  *
         d  *
     0
•  represent spin coordinates

53
Molecular Orbitals
molecular orbitals depend on spatial coordinates and spin
Spin orbital, i(x):
  r    

  x   or
  r    


• spin orbitals are used in quantum chemical calculations

• treatment of spin increases the number of degrees of freedom

(r)  (r, )  (x)

spatial coordinates spin coordinates

• for a system with N electrons, there are 4N degrees of freedom

54
Constraints on the Trial Wavefunction
for variational principle to apply, the trial wavefunction must satisfy the same
constraints as the real wavefunction
1. Pauli exclusion principle:
• for atoms  no two electrons can have same set of quantum numbers
• for molecules  no two electrons can be in same state
 no two electrons can have same xi
2. Antisymmetry:
•  must be antisymmetric with respect to interchange of coordinates x between
two electrons
  x1 ,..., xi , x j ,..., x N     x1 ,..., x j , xi ,..., x N 
• if  is antisymmetric, the Pauli exclusion principle will be satisfied

3. Indistinguishability:
• uncertainty principle  we can‟t know exact positions and velocities of electrons
•  must not distinguish between particles

  1s (1)2s (2) versus   21/2 1s (1)2s (2)  1s (2)2s (1) 

invalid because electrons 1 valid because electrons 1 and 2 can
and 2 are in specific orbitals be in either orbital
55
Molecular Orbitals  Trial Wavefunction
we want the form of the trial wavefunction to closely resemble the true
wavefunction
N
1 2 M N Z I N 1 N 1
  i    
cannot treat electron-
Hˆ elec electron terms exactly
i 1 2 I 1 i 1 rIi i 1 j i rij

let‟s forget about e-e interactions for now

N
1 2 M N ZI
Hˆ eff   i  
i 1 2 I 1 i 1 rIi

solve Schrödinger equation

Hˆ eff   E 

solution is product of molecular orbitals

(x1, x 2 ,..., xi , x j ,..., x N )  1(x1 ) 2 (x 2 )... i (xi )  j (x j )...  N ( x N )

56
Hartree Product Wave Functions
When the only terms in Hˆ are the one-electron kinetic energy and nuclear
attraction terms, the operator is separable:
N
Hˆ   hˆi
i 1

Where hˆi is the one-electron Hamiltonian:

M
1 Zk
hi   i  
ˆ 2

2 k 1 rik

Eigenfunctions of hˆi must satisfy the one-electron Schrödinger equation:

hˆi i   i i

Because Hˆ is separable, its many-electron eigenfunctions can be constructed

As products of one-electron eigenfunctions:

 HP (x1, x2 ,..., xi , x j ,..., x N )  1(x1 ) 2 (x 2 )... i ( xi )  j ( x j )...  N ( x N )

57
Hartree Product Wave Functions
 HP (x1, x2 ,..., xi , x j ,..., x N )  1(x1 ) 2 (x 2 )... i ( xi )  j ( x j )...  N ( x N )

The eigenvalue of  HP can be found as follows:

Hˆ  HP  Hˆ 1  2 ...  N
N
  hi 1  2 ...  N
i 1

 ( h1 1 )  2 ...  N  1 ( h2  2 )...  N  ...  1  2 ...( hN  N )

 (1 1 )  2 ...  N  1 ( 2  2 )...  N  ...  1  2 ...( N  N )
N
   i 1  2 ...  N
i 1

N 
    i   HP
 i 1 
This proves that the energy eigenvalue of the many-electron wavefunction
is the sum of the one-electron energy eigenvalues. 58
 Hartree Product Wave Functions
 HP (x1, x2 ,..., xi , x j ,..., x N )  1(x1 ) 2 (x 2 )... i ( xi )  j ( x j )...  N ( x N )
• called a Hartree product
• appealing because it is simple from a mathematical standpoint

Hartree product is an independent-electron wavefunction:

 HP  x1,..., x N  dx1...dx N  1  x1  dx1 2  x 2  dx 2 ...  N  x N 

2 2 2 2
dx N
• probability that electron 1 is in dx1 is independent of probability that
electron 2 is in dx2, etc.
• in reality the motion of electron 1 should depend on the motion of all the
other electrons  electron correlation (more later)

implications:
• Hartree product neglects instantaneous electron-electron interactions
• for an N electron system, electron 1 „feels‟ the average Coulomb repulsion of
the other N-1 electrons

59
 Hartree Product Wave Functions
consider flipping 2 coins:

P( H1 , H 2 )  P( H1 ) P( H 2 )

probability of both coins probability of coin probability of coin

landing heads up 1 being heads 2 being heads

if a probability can be written as a product of probabilities of individual events, those

events are independent of each other

consider a 2 electron example:

 HP  x1 , x 2  dx1dx 2  1  x1  dx1 2  x 2  dx 2
2 2 2

probability electron 1 is in dx1 at probability probability

x1 and electron 2 is in dx2 at x2 electron 1 is in electron 2 is in
dx1 at x1 dx2 at x2

60
Hartree Product Wave Functions
 HP (x1, x2 ,..., xi , x j ,..., x N )  1(x1 ) 2 (x 2 )... i ( xi )  j ( x j )...  N ( x N )
let’s test HP against the criteria for a valid wavefunction

1. Indistinguishability:

12
HP
(x1 , x2 )  1 (x1 ) 2 (x2 )
• distinguishes between electrons
 21
HP
(x1 , x2 )  1 (x2 ) 2 (x1 ) • invalid form of wavefunction

2. Antisymmetry:

 (x1 , x2 )  1 (x1 ) 2 (x2 )

HP
12
12
HP
 x1 , x2   21HP  x1 , x2 
• not antisymmetric
 21
HP
(x1 , x2 )  1 (x2 ) 2 (x1 ) • invalid form of wavefunction

61
Slater Determinant Wavefunction
deficiencies of HP can be overcome by taking linear combinations of product
wavefunctions
12  x1 , x2   21/ 2  1  x1  2  x2   2  x1  1  x2 

1. Indistinguishability:
• does not distinguish between
• electron 1 can be found in either orbital electrons
• electron 2 can be found in either orbital • valid form of wavefunction

2. Antisymmetry:
12  x1 , x2   21/ 2  1  x1  2  x2   2  x1  1  x2 
• antisymmetric
 21  x1 , x 2   2  1  x 2   2  x1    2  x2  1  x1  
1/ 2
• valid form of
 21/ 2  1  x1   2  x 2    2  x1  1  x 2   wavefunction

 12  x1 , x 2 

62
Slater Determinant Wavefunction
12 is a Slater determinant
12  x1 , x2   21/ 2  1  x1  2  x2   2  x1  1  x2 

normalization constant determinant

Determinants
• square array of elements
• “lines” on left and right (don‟t confuse with matrices that use brackets)
• functions as a shorthand for writing the sum of products:

a11 a12
 a11a22  a12 a21
a21 a22

a11 a12 a13

a22 a23 a21 a23 a21 a22
a21 a22 a23  a11  a12  a11
a32 a33 a31 a33 a31 a32
a31 a32 a33
63
Slater Determinant Wavefunction
Properties of Determinants (D)

1. if all elements in a column or row are 0, D = 0

2. multiplying a row or column by k multiplies D by k

3. switching two columns or two rows changes the sign of D
4. if two columns or two rows are identical, D = 0
5. if two columns or two rows are multiples of each other, D = 0

6. multiplying a column (or row) by k and adding it to another column

(or row) leaves D unchanged

64
Slater Determinant Wavefunction
virtually all quantum chemical methods use Slater determinants
For an N electron system:
1  x1   2  x1   N  x1 
1  x 2  2  x2  N  x2 
SD  x1 , x 2 ,..., x i , x j ,..., x N    N !
1/2

1  x N  2  xN  N xN 
• columns labeled by molecular orbitals
• rows labeled by electrons
 
• shorthand notation:  SD x1 , x2 ,..., xi , x j ,..., x N  1 2 ...i  j ... N

Based on properties of determinants:

• switching two rows changes sign of SD  antisymmetry
• if two columns (molecular orbitals) are identical SD = 0  two
electrons can‟t occupy same orbital (Pauli exclusion principle)

65
Slater Determinant Wavefunction
Hartree product was an independent electron wavefunction
what about Slater determinants?
consider two electron system with electrons of opposite spin:
  x1 , x2   12 1  x1    1  r1   1  2  x2    2  r2   2 
square to get probability:
 dx1dx2  1 2  1  r1   1  2  r2   2   1  r2  2  2  r1   1 
2 2

define P(r1,r2)dr1dr2 = probability of finding electron 1 in dr1 and electron 2 in dr2:

P(r1 , r2 )dr1dr2   d1d2  dr1dr2

2

 1   1  r1   2  r2    1  r2   2  r1   dr1dr2
2 2 2 2

2 

• probability is just products  electrons of opposite spin are independent with Slater determinant
• in fact, P(r1,r1)  0  with Slater determinant two electrons of opposite spin can occupy
the same point in space

66
Slater Determinant Wavefunction
Hartree product was an independent electron wavefunction
what about Slater determinants?
consider two electron system with electrons of same spin:
  x1 , x2   12 1  x1    1  r1   1  2  x2    2  r2   2 
using analysis on preceding slide

P(r1 , r2 )  1 
2 1  1
 r  2  r2    1  r2   2  r1 
2 2 2 2


  1*  r1  2  r1  2*  r2  1  r2   1  r1  2*  r1  2  r2  1*  r2  

• probability is not just products  electrons of same spin are correlated with Slater determinant
• called exchange correlation because last two terms exchange electron coordinates between
different spatial orbitals

• consequence of antisymmetrizing the wavefunction, and related to Pauli exclusion principle

• P(r1,r1) = 0  with Slater determinant two electrons of same spin cannot occupy the same
point in space

67
Slater Determinant Wavefunction
For an N electron system:
1  x1   2  x1   N  x1 
1  x 2  2  x2  N  x2 
SD  x1 , x 2 ,..., x i , x j ,..., x N    N !
1/2

1  x N  2  xN  N xN 
• columns labeled by molecular orbitals
• rows labeled by electrons
 
• shorthand notation:  SD x1 , x2 ,..., xi , x j ,..., x N  1 2 ...i  j ... N

Physics captured:
• electrons of opposite spin are treated independently
 neglects instantaneous Coulomb interactions
• electrons of same spin are correlated through exchange interactions
 accounts for Pauli exclusion principle
 significant improvement over Hartree product

68