1 Two and Three dimensional problems

We now turn to the study of the Schrodinger equation in two and three dimensions. Let us start
by considering a central potential in two dimensions. A central potential will depend only on |�x|,
which in polar coordinates is r; thus V (�x) = V (r).
In general, the TISE will be written as
� �
h̄2 � 2
− ∇ + V (�x) ψ(�x) = Eψ(�x)
2m x

� x is the gradient with respect to �x.

Here ∇
In this course, we shall only consider central potentials, i.e. where the potential is only a function
of |�x|. A standard approach is the separation of variables.
In some special cases, the above equation can be separated by x, y, z. An example is the 3d
harmonic oscillator, for which we have
1 1
V (|�x|) = mω 2 |�x|2 = mω 2 (x2 + y 2 + z 2 )
2 2

The gradient operator is

2 ∂2 ∂2 ∂2
∇ =
� + +
∂x2 ∂y 2 ∂z 2
from which it follows that the Hamiltonian splits into

H = Hx + Hy + Hz

where
h̄2 ∂ 2 1
Hx = − + mω 2 x2
2m ∂x 2 2
is the Hamiltonian for a 1d SHO in the x direction, and similarly for Hy and Hz .
Now we simply assume that the stationary states also factorise as

ψ(�x) = X(x)Y (y)Z(z)

and each of the X, Y, Z are the stationary states of the 1d problem for that coordinate. Thus the
complete basis of stationary states for the 3d harmonic oscillator is

ψn1 n2 n3 = ψn1 (x)ψn2 (y)ψn3 (z)

where ψn1 (x) is the n1 th excited state of the 1d harmonic oscillator.
The energy of the above state is given by
� �
3
E n1 n2 n3 = n1 + n 2 + n 3 + h̄ω
2

1
Thus the ground state energy of the 3d harmonic oscillator is E000 = 32 h̄ω.
For the excited states, we encounter the phenomenon of degeneracy. The next higher energy value,
the first excited value is 52 h̄ω, but this can be achieved in three different ways, by taking (n1 , n2 , n3 )
to be (1, 0, 0) or (0, 1, 0) or (0, 0, 1). The three corresponding wavefunctions ψ100 , ψ010 , ψ001
are completely different functions having the same energy eigenvalue. We say that there is a
“threefold” degeneracy.

1.1 Plane polar coordinates

Let us begin by studying plane polar coordinates in 2d as an exercise. These results can be used
for cases where the particle is confined to a plane, such as in surface physics. To switch to plane
polar coordinates, we use the following relations

x = r cos φ , y = r sin φ
where
r≥0 , 0 ≤ φ < 2π
The inverse transformation is given by
� y
r= x2 + y 2 , φ = tan−1
x
(note that we must be careful in defining the branch for tan− 1 such that (x, y) and (−x, −y) are
mapped to different values of φ). Using the chain rule, we can calculate
2 2
�2= ∂ + ∂
∇
∂x2 ∂y 2
as follows.
∂ ∂r ∂ ∂φ ∂
= +
∂x ∂x ∂r ∂x ∂φ
∂ sin φ ∂
= cos φ −
∂r r ∂φ
Similarly we find
∂ ∂r ∂ ∂φ ∂
= +
∂y ∂y ∂r ∂y ∂φ
∂ cos φ ∂
= sin φ +
∂r r ∂φ
Summing and squaring we get

�2= ∂ +1 ∂ + 1 ∂
2 2
∇
∂r2 r ∂r r2 ∂φ2

This is almost separated among the radial and angular variables, and it is now useful to consider
the TISE for a central force.
� �
h̄2 � 2
− ∇ + V (r) ψ(r, φ) = Eψ(r, φ)
2m

2
Let us propose that the wavefunction can be factorised into a radial part R and an angular part Φ

ψ(r, φ) = R(r)Φ(φ)

Inserting the gradient operator and the factorised wavefunction into the TISE, we get
� �
h̄2 ∂2 1 ∂ 1 ∂2
− + + + V (r) R(r)Φ(φ) = ERΦ
2m ∂r2 r ∂r r2 ∂φ2
This is simplified as
� �
h̄2 ∂2 1 ∂ h̄2 R ∂ 2
− Φ + R − Φ + V (r)R(r)Φ(φ) = ERΦ
2m ∂r2 r ∂r 2m r2 ∂φ2

Dividing both sides by R(r)Φ(φ) and using the simplified notation R� (r) = ∂R
∂r
and Φ� (φ) = ∂Φ
∂φ

� �
h̄2 1 1 h̄2 1 ��
− R�� + R� − Φ + V (r) = E
2m R r 2mr2 Φ

Here now the first and third term depends on R and the second on Φ, while the right hand side is
just a number.
Now let us see if we can solve the angular part

Φ�� (φ) = λΦ(φ)

This will allow us to insert the solution into the above equation and obtain a purely radial equation.
Now solving this angular part is very straightforward, we have

Φ(φ) = einφ

for some unknown number n, and thus λ = −n2 . Now since φ is a periodic variable, the solution
to this equation must come back to itself as φ → φ + 2π. Thus

einφ → ein(φ+2π) = e2πin einφ

For the right hand side to be same as the left hand side, we see that n must be a real integer;
n = . . . , −2, −1, 0, 1, 2, . . .
Inserting this solution into the TISE above, we find an ordinary differential equation
� � � �
h̄2 1 � h̄2 n2
− R + R + V (r) +
��
R = ER
2m r 2mr2
Thus we have now reduced the problem to a one dimensional problem. There are however three
differences: (i) the differential operator is changed, having a first derivative together with the
second derivative, (ii) the effective potential has picked up an extra term

h̄2 n2
Ṽ (r) = V (r) +
2mr2
and (iii) the range of r is of course r ≥ 0.

3
Anticipating our need to normalise the wave function, we can choose to normalise the radial and
angular parts separately. The normalisation of Φ is given by
� �
dφ|einφ |2 = dφ = 2π

and the correctly normalised angular wave function is

1
Φ(φ) = √ einφ
2π

The extra term that appears in the potential can be associated orbital angular momentum. This we
can anticipate of course from classical physics. In quantum mechanics, the angular momentum
will become an operator. Since L � = �x × p�, and both �x and p� lie in a plane, we can have angular
momentum only along the z-axis
L̂z = x̂p̂y − ŷ p̂x
. In the position basis,
� � � �
∂ ∂
L̂z → x −ih̄ − y −ih̄ = −ih̄(x∂y − y∂x )
∂y ∂x

Changing to polar coordinates we find that

∂
L̂z → −ih̄
∂φ
which was the operator for the which the angular wavefunction was an eigenfunction. The angular
wave functions are eigenfunctions of the angular momentum

L̂z einφ = nh̄einφ

and we see that orbital angular momentum is quantised in integer units of h̄. This is also true in 3
dimensions, where the details are more complicated.
The effective potential in the Schrodinger equation can be written as

L2z
Ṽ (r) = V (r) +
2mr2

where by L2z we really mean the eigenvalue of the operator L̂2z .

So far, all we know about the potential is that it is central. For a central force, angular momentum
is conserved. In such cases, it is useful to reformulate the problem in terms of angular momentum
(which is completely solved) leaving only the radial problem. The radial problem, of course,
depends on the details of the potential and can only be solved only in special cases.