Helium Atom, Approximate Methods

22nd April 2008

I. The Helium Atom and Variational Principle: Approximation

Methods for Complex Atomic Systems

The hydrogen atom wavefunctions and energies, we have seen, are deter-
mined as a combination of the various quantum ”dynamical” analogues of
classical motions (translation, vibration, rotation) and a central-force inter-
action (i.e, the Coulomb interaction between an electron and a nucleus).

Now, we consider the Helium atom and will see that due to the attendant
3-body problem for which we cannot determine a close-for, first-principles
analytic solution, we will have to find recourse in approximate methods.

The Helium atom has 2 electrons with coordinates r 1 and r2 as well as a

single nucleus with coordinate R. The nuclues carries a Z = +2e charge.

The Schrodinger equation is:

!
h̄2 2 h̄2 2 h̄2 2
− ∇ − ∇1 − ∇ ψ(R, r1 , r2 ) +
2M 2me 2me 2
!
2e2 2e2 e2
− − + ψ(R, r1 , r2 ) = Eψ(R, r1 , r2 )
4πo |R − r1 | 4πo |R − r2 | 4πo |r1 − r2 |

where the symbol ”nabla”, when squared, is given by:

∂2 ∂2 ∂2
∇2 = + +
∂x2 ∂y 2 ∂z 2
Keep in mind that the R, r, and r represent the Cartesian coordinates of
each paticle. This is a 3-body problem and such problems are not solved
exactly. Thus, the problem will be reformulated in terms of 2 variables. The
first approximations: M me , fix the nucleous at the origin (R) = 0. Thus,
the Schrodinger equation in relative variables is:

1
 h̄2  2e2 1 1 e2
  
2 2
−∇1 − ∇2 ψ(r1 , r2 )− + + ψ(r1 , r2 ) = Eψ(r1 , r2 )
2me 4πo r1 r2 4πo |r2 − r1 |

Recall that the ∇2 , representing the kinetic energy operator, in spherical

polar coordinates is:

1 ∂ ∂ 1 ∂ ∂ 1 ∂2
   
r12 + sinθ1 +
r12 ∂r1 ∂r1 r12 sinθ1 ∂θ1 ∂θ1 r12 sin2 θ1 ∂φ21

The Independent Electron Approximation to Solving the Helium

Atom Schrodinger Equation
If we neglect electron-electron repulsion in the Helium atom problem,
we can simplify and solve the effective 2-body problem.

• Solve the relative motion problem (separate out the center of mass
motion as we have seen earlier)

• Center of mass is assumed to be the nucleus; good approximation for

heavier nuclei

The Hamiltonian is now:

Ĥ = KEe1 + KEe2 + Vne1 + Vne2

h̄2  2e2 1 1
  
2 2
= −∇1 − ∇2 − +
2me 4πo r1 r2
!  !
−h̄2 2 2e2 1 −h̄2 2 2e2 1
 
= ∇1 − + ∇2 −
2me 4πo r1 2me 4πo r2

Under the independent electron approximation, if we take the total He atom

wavefunction as a product of the individual electron wavefunctions (here ap-
proxmiated as hydrogen-like wavefunctions):

Ψ(r1 , r2 ) = ψ(r1 )ψ(r2 )

ĤΨ(r1 , r2 ) = Ĥ1 ψ(r1 ) + Ĥ2 ψ(r2 )

2
 This yields E = E1 + E2 . Recal the hydrogen-like energies and wavefunc-
tions are:

• Ψnlm (r) = Rnl (Zr/ao )Ylm (θ, φ) (Z=1 for hydrogen, Z=2 for helium)

2
• Energies are: En = − Z 2Eh n12

Thus, approximate wavefunctions and energies for helium:

Ψn1 , l1 , m1 , n2 , l2 , m2 (r1 , r2 )ψspin = Ψn1 , l1 , m1 (r1 )Ψn2 , l2 , m2 (r2 )ψspin

Z 2 Eh 1 1
 
En1 ,l1 ,m1 ,n2 ,l2 ,m2 = − +
2 n21 n22

How sound is the independent electron approximation (no electron-electron

repulsion model)? We can compare the predicted ionization potentials
with the exerimental values. The ionization potential is the energy required
to extract an electron from an atom. For the Helium atom, this can be
represented as an equation such as:

He → He+ + e−

The energy change associated is:

∆E = EHe,1s − EHe,(1s)2
= −54.4eV − (−108.8eV ) = 54.4eV

The ground state energy for Helium can also be contrasted as:

22 Eh 1 1
 
0
E = E1s,1s =− 2
+ 2
2 1 1
= −4Eh
= −108.8eV signif icantlytoonegative

3
The experimental value for the ionization potential (from mass spectroscopic
measurements) is 24.6 eV; so the independent electron approach entails sig-
nifican error. Because the electrons are not allowed Coulombic repulsion,
the energy required to remove a particular electron is higher than the ex-
perimental value. This is also validated by the much lower (and thus more
attractive/favorable) ground state energy predicted by the independent elec-
tron model compared to experiment.
Moving Beyond the Independent Electron Model: Perturba-
tive and Variational Methods
noindent Perturbation Theory

The Helium atom Hamiltonian, we recall, is:

h̄2  2e2 1 1 e2
  
Ĥ = −∇21 − ∇22 ψ(r1 , r2 ) − + +
2me 4πo r1 r2 4πo |r2 − r1 |
= Ĥ 0 + Ĥ 1 Ĥ 0 ψ 0 = E 0 ψ 0

We see that the electron-electron repulsion term can be treated as a ”per-

turbation” to the independent electron Hamiltonian. in this sense, we can
choose to define, develop, and include various orders of perturbative cor-
rections to the energies and wavefunctions. WThough we are currently
applying this to the problem of electron-electron repulsion, we will see later
that modern advanced methods for electronic strucuture calculations em-
ploy perturbation methods to account for important electron correlation.
For now, we concern ourselves with the development of perturbation theory
and application to correct for two-body Coulomb repulsion in the Helium
atom.

First Order Perturbation Theory

First, expand the total wavefunction up to first order contributions:

ψn = ψn0 + ψn1
En = En0 + En1
     
Ĥ 0 + Ĥ 1 ψn0 + ψn1 = En0 + En1 ψn0 + ψn1

Expanding the expression:

Ĥ 0 ψn0 + Ĥ 1 ψn0 + Ĥ 0 ψn1 + Ĥ 1 ψn1 = En0 ψn0 + En1 ψn0 + En0 ψn1 + En1 ψn1

4
The first terms on each side are equivalent. Last terms are approxi-
mated/assumed to be neglibly small–this is a perturbation energy
and wavefunction. This leaves:

Expanding the expression:

Ĥ 1 ψn0 + Ĥ 0 ψn1 = En1 ψ 0 + En0 ψn1

Solve for ψn1 and En1 . Left multiply by ψn0∗ and integrate:

Expanding the expression:

Z Z Z Z
dτ ψn0∗ Ĥ 1 ψn0 + dτ ψn0∗ Ĥ 0 ψn1 = dτ ψn0∗ En1 ψn0 + dτ ψn0∗ En0 ψn1

 ∗
Since Ĥ 0 is Hermitian, dτ ψn0∗ Ĥ 0 ψn1 = dτ Ĥ 0 ψn0 ψn1 .
R R

Z Z Z
dτ ψn0∗ Ĥ 1 ψn0 + dτ ψn1 En0 ψn0∗ = En0 dτ ψn0∗ ψn1 + En1

Solving for En1 , the fist order perturbative correction to the inde-
pendent electron energy level for helium gives:

Z
En1 = dτ ψn0∗ Ĥ 1 ψn0

Expilcitly, this means, En = En0 + En1 ; we have added a ”small” perturbative

correction to the reference independent electron energy for a given energy
level, n!
Now, what about the correction to the wavefunction? Let’s recall:

Ĥ 1 ψn0 + Ĥ 0 ψn1 = En1 ψ 0 + En0 ψn1

We will solve for ψn1 by expanding ψn as a linear combination of unper-

turbed wavefunctions as follows:

5
 ψn = ψn0 + anj ψj0
X

j6=n

Left multiply by ψk0∗ and integrate:

Z Z Z Z
dτ ψk0∗ Ĥ 1 ψn0 + dτ ψk0∗ Ĥ 0 ψn1 = dτ ψk0∗ En1 ψn0 + dτ ψk0∗ En0 ψn1

Z Z Z Z
dτ ψk0∗ Ĥ 1 ψn0 dτ ψk0∗ Ĥ 0 anj ψj0 dτ ψk0∗ En1 ψn0 dτ ψk0∗ En0 anj ψj0
X X
+ = +
j6=n j6=n

Z Z Z Z
dτ ψk0∗ Ĥ 1 ψn0 + anj Ej0 dτ ψk0∗ ψj0 = En1 dτ ψk0∗ ψn0 + En0 dτ ψk0∗ ψj0
X X
anj
j6=n j6=n

  Z Z
ank Ek0 − En0 = En1 dτ ψk0∗ ψn0 − dτ ψk0∗ Ĥ 1 ψn0

Now, we consider two cases:

Z
k=n → En1 = dτ ψn0∗ Ĥ 1 ψn0

dτ ψk0∗ Ĥ 1 ψn0 1
R
Ĥkn
k 6= n → ank = ≡
En0 − Ek0 En0 − Ek0

Thus, the first order correction to the wavefuntion is:

ψn = ψn0 + ψn1 = ψn0 + anj ψj0

X

j6=n

1
! !
Ĥjn dτ ψk0∗ Ĥ 1 ψn0
R
= ψn0 + ψj0 = ψn0 + ψj0
X X

j6=n
En0 − Ej0 j6=n
En0 − Ek0

Higher Order Corrections

For higher order corrections, energies and wavefunctions are expanded in

6
like manner:

ψn = ψn0 + ψn1 + ψn2 En = En0 + En1 + En2

The second-order correction to the energy is thus determined to be (see

other sources for derivation):

Z Z 1
Ĥjn
En2 = dτ ψn0∗ Ĥ 1 ψn1 = dτ ψn0∗ Ĥ 1 ψ0
X
E 0 − E0 j
j6=n n j
1 Ĥ 1
X Ĥnj jn
=
j6=n
En0 − Ej0

Helium Atom: First Order Perturbation Correction to Account

for Electron-Electron Repulsion
From the above discussion, the first order correction to the ground state
energy of Helium is:

Z
En1 = dτ ψn0∗ Ĥ 1 ψn0

If we assume an unperturbed wavefunction as the product of the 2 Helium

electrons in 1s orbitals:

 !1/2 2
Z3
ψ 0 (r1 , r2 ) = φ1s (r1 )phi1s (r2 ) =   e−Zr1 e−Zr2
π
1
Ĥ 1 =
r12

Z6 1
Z Z  
En1 = dτ ψn0∗ Ĥ 1 ψn0 = dτ e−Zr1 e−Zr2 e−Zr1 e−Zr2
π2 r12
Z6 e−2Zr1 e−2Zr2 5
Z
= dτ = ZEh
π2 r12 8

7
Thus, the ground state energy with first order correction is :

5 −11
E0 = E00 + E01 = −4Eh + ZEh = Eh = −74.8eV
8 4

• Neglect of electron-electron repulsion: E 0 = -108.8 eV

• First order correction: E0 = -78.8 eV
• 13th order correction: E0 = -79.01 eV
• Experiment: E = -70.0 eV
• Perturbation theory results may be greater or less than the true energy,
unlike variational results
• Perturbation, unlike variational theory, can be used to calculate any
energy level, not just the ground state.

Some comments on the Many-Electron Problem: Coordinate de-

pendence and correlation
Solving the Schrodinger equation for an N-electron atom means solving for
a function of 3N coordinates. Such problems are treated numerically as
we will see further below. Moreover, if we acknowledge that the individual
electrons present differing environemnts (core versus valence eletrons), we
can approxmiate the N-electron wavefunction (eigenfunction of Schrodinger
equation) in terms of individual electron orbitals, each dependent on
the coordinates of a single electron. This is the orbital approxima-
tion. The wavefunctions are written:

ψ(r1 , r2 , ...., rn ) = φ1 (r1 )φ2 (r2 )φ3 (r3 )...φn (rn )

Each of the functions, φn , is associated with an orbital energy,  n . It is
important to note that this form does not imply fully indepdent electrons,
since, as will be evident, each electron’s dynamics (and hance wavefunction)
is governed by the effective field/potential of all the other electrons in the
atom. The one-electron orbitals turn out to be well-approximated by the
hydrogen-atom wavefunctions.

NOTE:
• Due to the electron-electron repulsion term in the He atom Hamilto-
nian not being spherically symmetric, the Schrodinger equation cannot
be solved analytically. Numerical methods are applied along with ap-
proximations.

8
 • One such approxmiation is the neglect of full electron-electron cor-
relation (motion of electrons is correlated; electrons stay out of each
other’s way; correlation opposes repulsion, and thus contributes an
energy-lowering influence).

• Advanced electron correlation methods are well-developed and rou-

tinely applied.

The Hartree Model

If we neglect, for the moment, any electron-electron correlation, then we can
say that electron 1 can interact with the nucleus and a spatially averaged
charge distribution due to elecron 2. The ”physical” picture here is
of electron 2 ”smeared” out sphercially around the nucleus; thus, its charge
density is also diffuse around the nucleus. The charge in a given volume
element dτ at a point around the nucleus is given by:

−eφ∗ (r2 )φ(r2 )dτ

The effective potential that electron 1 interacts with now is determined by

the electron-nucleus interaction and the interaction of electron 1 with the
spherically symmetric (smeared out) charge distribution arising from a
diffuse electron 2:

−2e2 e2
Z
V1ef f (r1 ) = + φ∗2 (r2 ) φ2 (r2 )dr2
4πo |r1 | rπo |r1 − r2 |

This model is the Hartree model and serves as a starting point for treating
many-electron atoms.

Thus, for many-electron atoms, the approach to solving the Schrodinger

Equation is:
Note:

• Approxmiate the total wavefunction as a product of individual or-

bitals, each orbital depending on the coordinates of a single electron
(reduce N-body problem to N one-body problems)

• The set of N equations is solved self-consistenly to obtain the one-

electron energies, i and orbitals φi .

• The approach is approximate due to neglect of electron correlation.

9
Electron Spin: Extension to Intrinsic Angular Momentum
Discussion of angular momentum: spin does not arise naturally (using a
non-relativistic formalism). Experimentally, spin is required to explain de-
flection of Ag atoms in a magnetic field (recall Stern-Gerlach experiment
from chapter 17). Silver atoms are deflected in either one of two direc-
tions in a magnetic field oriented along the z-diretion of the lab frame of
reference; the direction is either up or down. Consider:

• For an Ag atom to be deflected, it requires non-zero total magnetic

moment and an associated non-zero total angular momentum

• Current in a loop of wire generates a magnetic field; the loop has a

magnetic moment

• Ag atoms have a closed inner shell (no net angular momentum for
closed shells; symmetry); outer valence electron, 5s must contain all
magnetic moment.

• s-orbital has no angular momentum — NO ORBITAL ANGULAR

MOMENTUM

• Thus, the 5s electron in Ag contains an intrinsic spin angular mo-

mentum; we associate this with spin. Spin is intrinsic since it is
independent of environment.

• Note that this idea and picture of intrinsic particle spin is a convenient
classical way of accounting for experiment.

Introduce spin angular momentum.

– Spin operators: ŝ2 and ŝz Spinwavef unctions:α and β

– Spin quantum number: s = 1/2, ms (takes on two values, ms =
1/2 and ms = −1/2.

Spin angular momentum follows general angular momentum behavior

as we discussed for orbital angular momentum.

h̄2 1
 
ŝ2 α = h̄2 s(s + 1)α = +1 α
2 2
h̄2 1
 
2 2
ŝ β = h̄ s(s + 1)β = +1 β
2 2

10
 h̄
ŝz α = ms h̄α = α
2
h̄
ŝz β = ms h̄β = − β
2

The orthonormality conditions are:

Z Z
α∗ βdσ = β ∗ αdσ = 0
Z Z
∗
α αdσ = β ∗ βdσ = 1

Indistinguishability of Electrons and Pauli Exclusion

Variational Method for Ground State

We can make use of the quantum mechanical postulate relating the
expectation value of an observable, in this case the ground state en-
ergy, E0 , to arrive at a bounding value for the ground state energy
and associated wavefunction.

Recall the relations:

Ĥψo = Eo ψo

Z Z
ψo∗ Ĥψo dr = ψo∗ Eo ψo dr = Eo

ψo not known, so assume a trial funtion, χ. Determine the expectation

of the Hamiltonian, Ĥ and define it to be the Variational energy,
Evar .

χ∗ Ĥχdr
R Z
Evar = R ∗ = χ∗ Ĥχdr
χ χdr

The last equality assumes normaalization of the guess wavefunction, χ.

The Variational Theorem states:

11
 Evar ≥ Eo f or any choice of wavef unction χ

short ”proof” of variational theorem. Begin by expanding the test

wavefunction as a linear combination of orthonormal wavefunctions
(recall idea of expansions of functions using orthonormal, complete
basis functions):

X
χ= an ψn
n

Z
|an |2 = 1
X
χ∗ χdr = normalization
n

The variational energy is:

Z
χ∗ Ĥχdr = |an |2 En
X
Evar =
n

Now, subtract the ground state energy, whatever that may be, from
the approximate variational energy:

|an |2 En − |an |2 Eo = |an |2 (En − Eo ) ≥ 0

X X X
Evar − Eo =
n n n

since all the individual energies must be equal to or greater than the
ground state energy! Thus, as an approximation, one can posit sev-
eral trial wavefunctions, determine the variational energy, and then
optimize in the space of wavefunctions. No doubt, this is not the
most efficient approach to determine a valid functional form, but such
information is generally obtained from chemical intuition.
The variational method also requires that variational parameters ap-
pear in the formulation so as to allow optimization, as the following
example shows.

The one-dimensional harmonic oscillator:

h̄2 d kx2
Ĥ = − +
2µ dx2 2

12
Let’s consider a trial wavefunction:

h̄2 d kx2
Ĥ = − +
2µ dx2 2

Trial wavefunction:

 1/4
γ 2 /2
χ= e−γx ; γ variable (variational parameter)
π

h̄2 d
 1/2 Z !
γ kx2
Z
−γx2 /2 2 /2
Evar = χ∗ Ĥχdr = e − + e−γx dx
π 2µ dx2 2

h̄2 h̄2 γ k
 1/2 Z ∞ ∞  1/2 Z ∞
γ k γ
Z 
−γx2 2 −γx2 2
=− γe dx + (γx) e dx + x2 e−γx dx = + ≥ Eo
2µ π −∞ −∞ 2 π −∞ 4µ 4γ

Now, allow γ to vary and optimize Evar :

dEvar h̄2 k
= − 2 =0
dγ 4µ 4γ

(kµ)1/2
γ= ≡α
h̄

This leads, with no surprise, to the exact solutions to the 1-D H.O.

 1/4
α 2 /2 h̄ω
χ = ψo (x) = e−αx → Evar = = Eo
π 2

Now, returning to the Helium atom, we can attempt to use the varia-
tional principle to determine appropriate wavefunctions and energies.
Recall from earlier the Helium atom Hamiltonian operator:

h̄2
!
2e2 1 1 e2
 
(−∇21 − ∇22 ψ(r1 , r2 )− + + ψ(r1 , r2 ) = Eψ(r1 , r2 )
2me 4πo r1 r2 4πo |r2 − r1 |

This can be rewritten in atomic units to arrive at:

13
 1 1 1 1
  
ĤHe = − ∇21 + ∇22 − 2 + +
2 r1 r2 |r2 − r1 |
Z
Evar = χ∗ Ĥχdr

We can try a wavefunction that is a product of two hydrogenic 1s

orbitals, each one included to represent the relative motion of the
electron-nucleus system:
χ = φZ̃ Z̃
1s (r1 )φ1s (r2 )

Here, we roll a variational parameter, Z̃ into our definition of the trial

wavefunction:

!3/2
1 Z̃
φZ̃
1s =√ e−Z̃r/ao (1 − electron H atom orbitals)
π ao

Z̃ is an effective charge representing a variational parameter. Our trial

wavefunction now becomes:

!1/2 !1/2
Z̃ 3 −Z̃r1 Z̃ 3 −Z̃r2
χ(r1 , r2 ) = e ao e ao
π π

The variational energy is:

Z̃ 6 −1  2 1 1 1
Z −Z̃r1 −Z̃r2
     −Z̃r1 −Z̃r2
Evar = e ao e ao ∇1 + ∇22 − 2 + + e ao e ao dτ
π 2 r1 r2 r12
27
= Evar (Z̃) = Z̃ 2 − Z̃
8

dEvar (Z̃) 27 27
= 2Z̃ − =0 Z̃ = <2
dZ̃ 8 16
= −77.5eV

14
The variational approximation gives reasonable results, but keep in
mind that the effective nuclear charge is no longer 2, but less.

The Hartree-Fock Self-Consistent Field Method

Let’s now consider the Hartree-Fock approach to solving the many-
electron atom Schrodinger wave equation. We discuss this approach as
it is the starting point for modern computational electronic structure
methods. The method, along with many of the approximations and
mathematical/physical insights and observations required to employ
them, made possible what are today considered ”off-the-shelf” canned
computational chemistry software packages. Work continues to refine,
innovate, and bring advanced theoretical approaches to these methods.
A general approach to Hartree-Fock Methods Based on the Helium
Atom

To solve the Schrodinger equation, we must first assume some func-

tional form of the wavefunction. As discussed above, the orbital
approximation within which the total wavefunction is written as
a product of individual atomic orbitals (each orbital depends on
the coordinates of a single electron and each orbital is asso-
ciated with a unique orbital energy). This effectively transforms
the problem into an independent electron problem, with the caveat
that the ”independence” is not strict; the individal electrons feel an
effective potential from the other electrons as discussed above.

χ(r1 , r2 ) = φ(r1 )φ(r2 )

Since we don’t know yet what these orbitals may look like (though
we can guess that they will be much like the hydrogenic orbital wave-
functions) we can label these individual orbitals as φ HF (r1 ), φHF (r2 ).
These are the Hartree-Fock orbitals. We can also solve for the orbital
energies in a variational manner; this suggests that the orbitals we use
will incorporate a variational parameter.

If we interpret the charge density of electron 2 as:

eφ∗ (r2 )φ(r2 ) = e |φ(r2 )|2

15
we can write an effective potential for electron 1 by integrating the
Coulomb interaction over the charge density of electron 2. This is
the Coulomb potential felt by electron 1 averaged over all positions of
electron 2, weighted by the charge density:

1
Z
U1ef f (r1 ) = e2 φ∗ (r2 ) φ(r2 )dr2
r12

Now we can write an effective Hamiltonian and single-electron Schrodinger

equation for electron 1 as:

1 Z
Ĥ1ef f (r1 ) = − ∇21 − + U1ef f (r1 )
2 r1
Ĥ1ef f (r1 )φHF (r1 ) = 1 φHF (r1 )

Note that these are written in terms of coordinates of a single elec-

tron!. It is evident, however, that we are still not completely absolved
from knowing anything about electron 2, since we need the distrubition
function (probability density) of electron 2 to construct the effective
potential for electron 1. Likewise, for a many-electron problem, we
would need information on all other electrons to solve this one-electron
problem. The approach thus requires a self-consistent approach as
outlined below.

– Start with trial wavefunction, φ(r 2 )

– Use φ(r2 ) to calculate U1ef f (r1 )
∗ U1ef f (r1 ) = e2 φ∗ (r2 ) r112 φ(r2 )dr2
R

– Solve Schrodinger Eqn. for electron 1

∗ Ĥ1ef f (r1 ) = − 12 ∇21 − Z
r1 + U1ef f (r1 )
∗ Ĥ1ef f (r1 )φHF (r1 ) = 1 φHF (r1 )
– Use φ(r1 ) to calculate U1ef f (r2 )
∗ U2ef f (r2 ) = e2 φ∗ (r1 ) r112 φ(r1 )dr1
R

– Solve Schrodinger Eqn. for electron 2

16
 ∗ Ĥ2ef f (r2 ) = − 21 ∇22 − Z
r2 + U1ef f (r2 )
∗ Ĥ2ef f (r2 )φHF (r2 ) = 2 φHF (r2 )
– Continue iteration until SELF-CONSISTENT results for φ(r 1 )
and φ(r2 ) are obtained, i.e., wavefunctions don’t change from one
iteration to the next.

We now have the solution to the full Schrodinger equation as:

χ(r1 , r2 ) = φ(r1 )φ(r2 )

We can calculate the full Hartree-Fock energy using the full Hamilto-
nian as:

– Koopman’s theorem (ionization energies related to orbital ener-

gies)
– Filling of many-elecron atom orbitals (corrections to orbital en-
ergy perspective)

Pauli Exclusion and Anti-Symmetry of Wavefunctions

Good Quantum Numbers, Terms, Levels, and States

17