Notes 20: The Aharonov-Bohm effect

Forces play a central role in Newtonian mechanics. The force on an electric charge is due to electric and
magnetic fields. Forces and fields do not appear directly in the quantum mechanical Hamiltonian.
Instead, the corresponding potentials are central. The question then arises as to what are more
fundamental – fields or potentials? The Aharonov-Bohm effect provides an answer to this question.

The Aharonov – Bohm effect

∂A
E = −∇φ − , B = ∇ × A. (20.1)
∂t

The potentials are not unique. The same fields are obtained from two sets of potentials that are related by
a gauge transformation

∂χ
φ → φ ′= φ − , A → A′= A + ∇χ . (20.2)
∂t

In quantum mechanics, the Hamiltonian for a charged particle is

1
=
H ( −i∇ − qA ) ⋅ ( −i∇ − qA ) + qφ . (20.3)
2m

Let a solution of the TDSE for this Hamiltonian be Ψ.

If we apply a gauge transformation, the Hamiltonian becomes

1
H′
= ( −i∇ − qA′ ) ⋅ ( −i∇ − qA′ ) + qφ ′. (20.4)
2m

If this Hamiltonian is to represent the same system as that in equation (20.3), the solutions to the TDSE
can only differ by a phase factor, i.e.

Ψ ′ =ΨeiΛ . (20.5)

Then we have after much algebra

1
 ∂Ψ ′
i = H ′Ψ ′
∂t
∂Ψ  i
( q∇χ − ∇Λ + 2qA ) Ψ 
1
⇒ i = H Ψ + ( q∇χ − ∇Λ ) ⋅  ∇Ψ + (20.6)
∂t m 2m 
i  ∂Λ ∂χ 
+
2m
( q∇ 2 χ − ∇ 2 Λ ) Ψ +  
 ∂t
−q  Ψ.
∂t 

We have the TDSE for Ψ provided Λ − q χ is a constant, which we can take to be zero to recover the
original wave function in the absence of the gauge transformation. Then

Ψ ′ =Ψeiqχ  . (20.7)

The gauge transformation simply leads to a change in the phase of the wave function.

In classical mechanics, a charged particle is not influenced by the electromagnetic fields in regions where
the fields are zero. For a long time it was assumed the same was true in quantum mechanics. But in 1959
Aharonov and Bohm showed that the vector potential can affect the particle’s quantum behavior even in
regions where the field is zero. This remarkable result indicates that the potentials that are physically
more fundamental than the fields.

Consider an infinitely long straight uniform solenoid of radius a. A steady current produces a magnetic
field that is uniform inside the solenoid but is zero outside the solenoid. However, the vector potential is
not zero outside the solenoid. Using the magneto-static Lorenz gauge ∇ ⋅ A = 0, the vector potential in
cylindrical polar coordinates outside the solenoid is

Φ ˆ
A= φ. (20.8)
2π r

Here Φ =π a 2 B is the magnetic flux through the solenoid. Since there is no charge on the solenoid, the
scalar potential is zero and the Hamiltonian is

2 2 q 2 iq
H=− ∇ + A⋅A + A ⋅∇ (20.9)
2m 2m m

Now consider a charge q that is constrained to move in a circle of radius b, where b > a. The wave
function will depend only on the azimuthal angle, φ , and the Schrödinger equation is

 2 1 d 2ψ iq Φ 1 dψ q 2  Φ 
2

− + +  ψ =
Eψ . (20.10)
2m b 2 dφ 2 m 2π b b dφ 2m  2π b 

We look for solutions of form eiλφ , which gives

2
 qΦ  2mb 2
2

 λ −  = E. (20.11)
 2π   2

For the wave function to be single-valued, λ must be an integer, l say, and the energy eigenvalues are

qΦ 
2
2 
=El 2 
l−  . (20.12)
2mb  2π  

Clearly, the energies depend on the strength of the field on the inside of the solenoid, even though the
particle moves in a field free region.

Now consider the more general case in which the charge particle moves through a field free region where
the vector potential is A. We can make a gauge transformation

A′= A + ∇χ (20.13)

to make the vector potential zero in the field free region. This requires
r
χ (r ) =
− ∫ A ⋅ dr, (20.14)
r0

where r0 is some convenient reference point. Since ∇ × A = 0 in the field free region, χ ( r ) is
independent of the path taken from r0 to r, (provided a continuous deformation from one path to the other
does not cross into regions where A is not curl-free). Then the transformation

Ψ = Ψ ′e − iqχ  , (20.15)

[see equation (20.7)] will give a Schrödinger equation that is independent of the vector potential

∂Ψ ′ 2 2
i =− ∇ Ψ ′ + V Ψ ′. (20.16)
∂t 2m

If we can solve equation (20.16), it is then straightforward to find the wave function when the field-free
vector potential is non-zero by adding a phase factor as in equation (20.15).

In the experiment proposed by Aharonov and Bohm a beam of electrons is split in two, and passed either
side of a long solenoid before being brought back together. The beams pass through magnetic field free
regions. The non-zero magnetic potential, given by equation (20.8), leads to the beams having different
phases just before they recombine

qΦ  1  qΦ
q
g =∫ A ⋅ dr = ∫  φˆ  ⋅ r φˆ =
h 2π   r 
±
2
, ( ) (20.17)

3
where the plus sign corresponds to the beam in which the electrons move in the same directions as A, i.e.
in the same direction as the current. The phase difference, qΦ  , leads to measurable interference.

Berry pointed out that this experiment could be interpreted in terms of geometric phase. To see why,
suppose a charged particle is confined to a box outside the solenoid, which is centered at R, by a potential
V ( r − R ) . The eigenfunctions of the Hamiltonian satisfy

1
 −i∇ − qA ( r )  ⋅  −i∇ − qA ( r ) ψ n + V ( r − R )ψ n = Enψ n .
2m 
(20.18)

We saw earlier how to solve this equation. Let

ψ n = ψ n′ e −iqχ  , (20.19)

where
r
− ∫ A ( r′ ) ⋅ dr′.
χ (r ) = (20.20)
R

Then

2 2
− ∇ ψ n′ + V ( r − R )ψ n′ =
Enψ n′ . (20.21)
2m

Note that ψ n′ is a function of r - R. To find Berry’s phase we carry the box around the solenoid. Note

∫ψ ′ ( r − R ) e ∇ R ψ n′ ( r − R ) e − iqχ   d 3r
iq χ 
ψ n ∇ Rψ n = n
∗

  − iq χ  iq 
∫ψ ′ ( r − R ) e ∇ Rψ n′ ( r − R ) − ψ n′ ( r − R ) e − iqχ ∇ R χ d 3r
∗ iq χ
= e 
n

(20.22)
 iq 
∫ψ n′ ( r − R ) ∇ Rψ n′ ( r − R ) − ψ n′ ( r − R ) A ( R ) d 3r
∗
=
  
iq
− ∫ψ n′∗ ( r − R ) ∇ψ n′ ( r − R )d 3r − A ( R ) .
=


In the last step, the gradient is with respect to r. This operator is proportional to the momentum operator.
We know that in a stationary state, the expectation value of the momentum operator is zero (because the
expectation value of the position is constant). Hence

iq
− A ( R ).
ψ n ∇ Rψ n = (20.23)


Using this in the Berry’s phase formula, we get

4
 q q
γ n = i ∫ ψ n ∇ Rψ n ⋅ dR = ∫
 A ( R ) ⋅ dR = ∫ ∇ × A ⋅ dS. (20.24)
 S

In the last integral, S is the surface bounded by the circular path taken by the center of the box. This
surface includes the solenoid. Since B = ∇ × A, we get

q qΦ
γ n=
S∫ B ⋅ dS=

. (20.25)

This is the same as the Aharonov-Bohm result. We have shown that there can be electromagnetic effects
in regions where the fields are zero. But note Berry’s phase depends only on the magnetic flux and not
directly on the vector potential. This is required for electromagnetic theory to be gauge invariant.