4.

1 Schrödinger Equation in
Spherical Coordinates

∂Ψ p2
i~ ∂t = HΨ, where H = 2m + V

~ 2
p → (~/i)∇ implies i~ ∂t = − 2m ∇2Ψ + V Ψ
∂Ψ

R
normalization: d3 r | Ψ | 2 = 1

If V is independent of t, ∃ a complete set of

stationary states 3 Ψn(r, t) = ψn(r)e−iEnt/~,
where the spatial wavefunction satisfies the
time-independent Schrödinger equation:
~2 ∇ 2 ψ + V ψ = E ψ .
− 2m n n n n

An arbitrary state can then be written as a sum

over these Ψn(r, t).
Spherical symmetry
If the potential energy and the boundary
conditions are spherically symmetric, it is
useful to transform H into spherical
coordinates and seek solutions to
Schrödinger’s equation which can be written
as the product of a radial portion and an
angular portion: ψ(r, θ, φ) = R(r)Y (θ, φ), or
even R(r)Θ(θ)Φ(φ).

This type of solution is known as ‘separation of

variables’.

Figure 4.1 - Spherical coordinates.

In spherical coordinates, the Laplacian takes
the form:
   
1 ∂ ∂f 1 ∂ ∂f
∇2 f = 2 r2 + 2 sin θ
r ∂r ∂r r sin θ ∂θ ∂θ
!
1 ∂ 2f
+ .
r 2 sin2 θ ∂φ2

After some manipulation, the equations for the

factors become:
d2 Φ 2Φ,
= −m
dφ2
 
d dΘ
sin θ sin θ + l(l + 1) sin2 θ Θ = m2Θ, and
dθ dθ
  2
d dR 2mr
r2 − [V (r) − E]R = l(l + 1)R,
dr dr ~2

where m2 and l(l + 1) are constants of

separation.
The solutions to the angular equations with
spherically symmetric boundary conditions
are: Φm = (2π)−1/2eimφ
and Θml ∝ P l
m (cos θ),

where m is restricted to the range −l, ..., l,

 |m|
m 2
Pl (x) ≡ (1 − x ) |m|/2 d Pl (x) is the
dx
‘associated Legendre function,’ and Pl (x) is
the lth Legendre polynomial.

The product of Θ and Φ occurs so frequently

in quantum mechanics that it is known as a
spherical harmonic:
" #1/2
(2l + 1) (l − |m|)!
Ylm(θ, φ) =  eimφPlm(cos θ),
4π (l + |m|)!
where  = (−1)m for m ≥ 0 and  = 1 for
m ≤ 0, and the spherical harmonics are
orthonormal:
Z π Z 2π
0
dθ sin θ dφ [Ylm(θ, φ)]∗[Ylm
0 (θ, φ)] = δll0 δmm0 .
0 0
While the angular part of the wavefunction is
Ylm(θ, φ) for all spherically symmetric
situations, the radial part varies.

The equation for R can be simplified in form by

substituting u(r) = rR(r):
" #
2 2
~ d u 2
~ l(l + 1)
− 2
+ V + 2
u = Eu,
2m dr 2m r
R 2
with normalization dr |u| = 1.

This is now referred to as the radial wave

equation, and would be identical to the
one-dimensional Schrödinger equation were it
not for the term ∝ r −2 added to V , which
pushes the particle away from the origin and
is therefore often called ‘the centrifugal
potential.’

Let’s consider some specific examples.

Infinite spherical well
(
0, r < a
V (r) =
∞, r > a.

The wavefunction = 0 for r > a;

for r < a, the differential equation is
" # √
d2 u l(l + 1) 2 u, where k ≡ 2mE .
= − k
dr 2 r2 ~

The ’stationary’ eigenfunctions of this potential

are all bound states, confined to the region
r < a.

The solutions to this equation are Bessel

functions, specifically the spherical Bessel and
spherical Neumann functions of order l:
u(r) = Arjl (kr) + Brnl (kr),
 l
1 d sin x
where jl (x) ≡ (−x)l ,
x dx x
 l
l 1 d cos x
and nl (x) ≡ −(−x) .
x dx x
The requirement that the wavefunctions be
‘regular’ at the origin eliminates the Neumann
function from any region including the origin.
The Bessel function is similarly eliminated
from any region including ∞.

Figure 4.2 - First four spherical Bessel functions.

The remaining constants, k (substituting for E)
and A, are satisfied by requiring that the
solution vanish at r = a and normalizing,
respectively: jl (ka) = 0 ⇒ ka = βnl , where βnl
is the nth zero of the lth spherical Bessel
function.

Adding the angular portion, the complete

time-independent wavefunctions are

ψnlm(r, θ, φ) = Anl jl (βnl r/a)Ylm(θ, φ),

~2 2
where Enl = β .
2ma2 nl
4.2 Hydrogen Atom

The hydrogen atom consists of an electron

orbiting a proton, bound together by the
Coulomb force. While the correct dynamics
would involve both particles orbiting about a
center of mass position, the mass differential
is such that it is a very good approximation
to treat the proton as fixed at the origin.

The Coulomb potential, V ∝ 1r , results in a

Schrödinger equation which has both
continuum states (E > 0) and bound states
(E < 0), both of which are well-studied sets
of functions. We shall neglect the former, the
confluent hypergeometric functions, for now,
and concentrate on the latter.
Including constants, the potential is
e2 1 , leading to the following
V = − 4π
0r
differential equation for u:
" #
2 2
~ d u 2
e 1 2
~ l(l + 1)
− 2
+ − + 2
u = Eu.
2m dr 4π0 r 2m r

This equation can be simplified with two

√
substitutions: since E < 0, both κ ≡ −2mE
~
and ρ ≡ κr are non-negative real variables;
furthermore, ρ is dimensionless.

With these substitutions, u(ρ) satisfies:

" #
d2 u ρ0 l(l + 1) me2
2
= 1− + 2
u; ρ0 ≡ 2
.
dρ ρ ρ 2π0~ κ

Having simplified this equation more or less as

much as possible, let us now look at the
asymptotic behavior:

As ρ → ∞, the constant term in brackets

d 2u
dominates, or dρ2 → u, which is satisfied by
u = Ae−ρ + Beρ. The second term is irregular
as ρ → ∞, so B = 0 ⇒ u → Ae−ρ as ρ → ∞.
Similarly, as ρ → 0, the term in brackets ∝ ρ−2
2u
dominates, leading to dρ2 → l(l+1)
d
ρ2
u, which is
satisfied by u = Cρl+1 + Dρ−l . The second
term is irregular at ρ → 0, so
D = 0 ⇒ u → Cρl+1 as ρ → 0.

We might hope that we can now solve the

differential equation by assuming a new
functional form for u which explicitly includes
both kinds of asymptotic behavior: let
u(ρ) ≡ ρl+1e−ρv(ρ). The resulting differential
equation for v is
d2 v dv
ρ 2 + 2(l + 1 − ρ) + [ρ0 − 2(l + 1)]v = 0.
dρ dρ

Furthermore, let us assume that v(ρ) can be

expressed as a power series in ρ:
∞
X
v(ρ) = aj ρj .
j=0
The problem then becomes one of solving for
{aj }.
If we had expanded in a series of orthonormal
functions, it would now be possible to
substitute that series into the differential
equation and set the coefficients of each term
equal to zero.

Powers of ρ are not orthonormal, however, so

we must use a more difficult argument based
on the equation holding for all values of ρ to
group separately the coefficients for each
power of ρ:
" #
2(j + l + 1) − ρ0
aj+1 = aj .
(j + 1)(j + 2l + 2)

Let’s examine the implications of this recursion

relation for the solutions to the Schrödinger
equation.
As j → ∞, aj+1 → j+1 2 a , which is the same
j
term-to-term ratio as e2ρ. Thus, u → Aρl+1eρ,
a solution we rejected. The only way to
continue to reject this solution is for the
infinite series implied by the recursion relation
to terminate due to a zero factor: i.e.,
2(jmax + l + 1) = ρ0. If n ≡ jmax + l + 1, then
n is the familiar principal quantum number.
~2 κ 2 me4 −13.6 eV
ρ0 = 2n ⇒ E = − =− 2 2 2 2≡ 2
.
2m 8π 0~ ρ0 n

1 4π0~2 −10 m.
Also, κ = ⇒a≡ = 0.529 × 10
an me2

The truncated series formed in this way is,

apart from normalization, the ‘well-known’
associated Laguerre polynomials:
2l+1
v(ρ) ∝ Ln−l−1 (2ρ), where
 p
p p d
Lq−p(x) ≡ (−1) dx Lq (x) and
 q
d
Lq (x) ≡ ex dx (e−xxq ) is the q th Laguerre
polynomial.
With these definitions, the orthonormalized
solutions to the Schrödinger equation for
hydrogen can now be written as
l  
2r 2l+1 2r
ψnlm = Anl e−r/na Ln−l−1 Ylm(θ, φ),
na na
v
u
u 2 3 (n − l − 1)!
where Anl ≡ t 3
na 2n[(n + l)!]
Z
and dr dθ dφ r 2 sin θ ψnlm
∗ ψ
n0l0 m0 = δnn0 δll0 δmm0 .

Figure 4.4 - First few hydrogen radial wave

functions, Rnl (r).
Figure 4.6 - Surfaces of constant |ψ|2 for the
first few hydrogen wave functions.
The eigenvalues of these states are
En = − me4
2 2 2 2 , which correspond very
32π 0 ~ n
closely to the measured absorption and
emission spectra of hydrogen.

Figure 4.7 - Energy levels and transitions in the

spectrum of hydrogen.
4.3 Angular Momentum
The classical mechanics quantity L = r × p
becomes the quantum mechanical operator
L = r × (~/i)∇. Interestingly, Lx and Ly do
not commute: [Lx, Ly ] = i~Lz , .... On the
other hand, [L2, L] = 0. Thus, it is sensible to
look for states which are simultaneously
eigenfunctions of both L2 and one
component of L.

Choosing Lz as that component, let’s define a

‘ladder operator’ such as is used for the
harmonic oscillator problem: L± ≡ Lx ± iLy .
With this definition, [L2, L±] = 0 and
[Lz , L±] = ±~L±.

If f is an eigenfunction of both L2 and Lz , it

can be shown that L±f is also an
eigenfunction of those same operators.
Furthermore, its eigenvalue of L2 is
unchanged, while its eigenvalue of Lz is raised
(lowered) by ~.
But it cannot be that the eigenvalue of Lz
exceeds the magnitude of L. Therefore, there
must exist a ‘top rung’ of the ‘ladder’ ft such
that L+ft = 0. For this state, let the
eigenvalues of Lz and L2 be ~l and λ,
respectively. Since L±L∓ = L2 − L2 z ± ~Lz , it
can be shown that λ = ~2l(l + 1).

Similarly, there must exist a bottom rung fb

such that L−fb = 0. For this state, let the
eigenvalue of Lz be ~l. It can be shown that
λ = ~2l(l − 1), so l = −l.

There must be some number of integer steps

between l and −l, so l must be either an
integer or a half-integer. It is sometimes
called the azimuthal quantum number.

The joint eigenstates of L2 and Lz are

characterized by eigenvalues ~2l(l + 1) and
~m, respectively, where l = 0, 1/2, 1, 3/2, ...
and m = −l, −l + 1, ..., l − 1, l.
The eigenfunctions of L2 and Lz can be
identified by expressing all of the above
operators (Lx, Ly , Lz , L±, L2) in spherical
coordinates. These are just the operators of
which the Ylm(θ, φ) are the eigenfunctions.
Thus, when we solved for the eigenfunctions
of the hydrogen atom, we inadvertently found
those functions which are simultaneously
eigenfunctions of H, L2, and Lz .

Note also that we have discovered that the

azimuthal quantum number, l, in addition to
taking on integer values, may also take on
half-integer values, leading to a discussion of
the property of ‘spin’.
4.4 Spin

In classical systems, two different words are

used to describe two rather similar types of
rigid body rotation: ‘spin’ for rotation about
its center of mass; ‘orbital’ for rotation of its
center of mass about another axis.

The same two words are used in quantum

mechanical systems, but they do not refer to
similar types of motion. Experiments have
shown that the behavior of electrons in
magnetic fields, for example, cannot be
explained without invoking the existence of a
constant of motion in addition to the energy
and momentum. It apparently must be
characterized by an intrinsic angular
momentum S, or spin, in addition to whatever
extrinsic angular momentum L it might carry.
Spin quantities are defined in analogy to
comparable orbital angular momentum
[Sx, Sy ] = i~Sz , ...; S± ≡ Sx ± iSy ;
quantities: q
S±|s mi = ~ s(s + 1) − m(m ± 1) |s (m ± 1)i;
S 2|s mi = ~2s(s + 1)|s mi; Sz |s mi = ~m|s mi,
where s = 0, 1/2, 1, 3/2, ... and
m = −s, −s + 1, ..., s − 1, s.

This time, however, the eigenfunctions are not

expressible as a function of any spatial
coordinates.

Every elementary particle has a specific and

immutable value of s, which we call its spin:
e.g., π mesons have spin 0; electrons have
spin 1/2; photons have spin 1; deltas have
spin 3/2; and so on.

In contrast, the orbital angular momentum l for

each particle can take on any integer value,
and can change whenever the system is
perturbed.
An electron has spin s = 1/2, leading to two
eigenstates which we can call spin up (↑) and
spin down (↓) referring to the ‘projection of
the spin on the z axis’. We will express the
eigenfunction as a two-element
! column !
1 0
matrix, or spinor : χ+ ≡ ; χ− ≡ .
0 1

In the same notation, the operators are 2 × 2

matrices which we will express in terms of the
Pauli spin matrices:
! ! !
0 1 0 −i 1 0
σx ≡ ; σy ≡ ; σz ≡ .
1 0 i 0 0 −1
Therefore, S = (~/2)σ.

The σz matrix is already diagonal, so the

eigenspinors of Sz are simply χ+ and χ− with
eigenvalues +~/2 and −~/2, respectively.

Since these eigenstates span the space, we can

express any general spinor as a sum of these:
χ = aχ+ + bχ−.
If you measure Sz on a particle in this state χ,
you will measure +~/2 with probability |a|2
and −~/2 with probability |b|2.

Suppose now that you want to measure Sx on

this same state. The e.f. of Sx are
(x)
χ± = √1 (χ+ ± χ−) expressed in the basis of
2
the e.f. of Sz , with e.v. ±~/2. Thus hSxi =
(1/2)|a + b|2(+~/2) + (1/2)|a − b|2(−~/2).

NB, even if the particle is in a pure ‘up’ state

relative to the z axis (i.e., a, b = 1, 0), it can
be in either of two different states when
projected on the x axis. This uncertainty is
an expected result since [Sz , Sx] 6= 0 (i.e.,
incompatible observables).
Experiments on electrons imply that the
electrons possess a magnetic moment
unrelated to any orbital motion, as though
the electron were a macroscopic charged
body ‘spinning’ on an axis through its center
of mass. This magnetic moment is related to
this presumed ‘spin’ through µ = γ S, leading
to a Hamiltonian H = −µ · B = −γ B · S.

This term in the Hamiltonian can be used to

explain on a purely quantum mechanical basis
the observation of Larmor precession and the
Stern-Gerlach experiment.
Larmor precession

Consider a particle of spin 1

2 at rest in a
uniform magnetic field: B = B0k̂, so that
H = −γB0Sz . The eigenstates of H are the
same as those of Sz , χ+ and χ− with
eigenvalues ∓ γB20~ , respectively.

Since H is time independent, the general

solution to the time-independent Schrödinger
equation, i~ ∂χ
∂t = Hχ, can be expressed in
terms of the stationary states:
χ(t) = aχ+e−iE+t/~ + bχ−e−iE−t/~.
Since χ is normalized, we can substitute
a = cos α
2 and b = sin α . Calculating hSi:
2
hSxi = 2~ sin α cos(γB0t/2),
hSy i = − 2~ sin α sin(γB0t/2),
and hSz i = 2~ cos α.
These equations describe a spin vector which
has a constant component in the field
direction, but which precesses in the xy-plane
with frequency ω = γB0.

Figure 4.7 - Precession of hSi in a magnetic field.

Stern-Gerlach experiment

Suppose a beam of spin- 12 particles moving in

the y direction passes through a region of
inhomogeneous B = (B0 + αz)k̂, where the
constant α denotes a small deviation from
homogeneity.

Figure 4.11 - The Stern-Gerlach configuration.

In addition to the effect of the Lorentz force,
the inhomogeneity gives rise to a spatially
dependent energy which affects the
wavefunctions as follows:
χ(t) = aχ+e−iE+T /~ + bχ−e−iE−T /~
= (aeiγB0T /2χ+)ei(αγT /2)z
+(be−iγB0T /2χ−)e−i(αγT /2)z ,
where T is the amount of time spent in the
inhomogeneous field.

The z-dependent portion of the wavefunction is

~
equivalent to pz = ± αγT
2 , and leads to the
splitting of the beam into 2s + 1 individual
beams, demonstrating the quantization of Sz .
Addition of angular momenta

Suppose we have a system with two spin-1/2

particles. Since each can be ‘up’ or ‘down’,
there are four possible combinations:
↑↑, ↑↓, ↓↑, ↓↓.

Define the total angular momentum:

S ≡ S(1) + S(2).
(1) (2)
Then Sz χ1χ2 = (Sz + Sz )χ1χ2
(1) (2)
= (Sz χ1)χ2 + χ1(Sz χ2) = ~(m1 + m2)χ1χ2.

∴ m↑↑ = 1; m↑↓ = 0; m↓↑ = 0; m↓↓ = −1.

There are two e.s. degenerate in Sz , but not in

S 2. Sort them out by applying S− to the ↑↑
(1) (2)
state: S−(↑↑) = (S− ↑) ↑ + ↑ (S− ↑)
= (~ ↓) ↑ + ↑ (~ ↓) = ~(↓↑ + ↑↓).
Thus, there are three e.s. in one group and one
in another: |smi = |1 1i (↑↑);
|1 0i [ √1 (↑↓ + ↓↑)]; |1 − 1i (↓↓);
2
|0 0i [ √1 (↑↓ − ↓↑)].
2

The first three e.s. constitute a triplet set, and

the last a singlet. You can show that they are
all e.s. of S 2 with e.v. = ~2s(s + 1), as
anticipated.

Finally, to generalize, if you combine s1 with s2,

the possible resulting total spin of the system
ranges over every value between (s1 + s2) and
|s1 − s2| in integer steps. This relationship is
summarized in the following equation
involving the Clebsch-Gordan coefficients:
X s s s
|s1m1i |s2m2i = Cm11m
2
2 m |smi
s
and the reciprocal relation, where
m = m1 + m2 and s ranges as noted above.

These relationships hold for both orbital and

spin angular momentum, and mixtures, and
can be used to express products of spherical
harmonics as a sum of spherical harmonics.