Professional Documents
Culture Documents
Chapter 18: Scattering in One Dimension
Chapter 18: Scattering in One Dimension
B. Zwiebach
April 26, 2016
Contents
1 Scattering in One Dimension 1
1.1 Time Delay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2 An Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Figure 1: A potential of range R. The potential vanishes for x > R and is infinite for x ≤ 0.
We consider (elastic) scattering in the general potential shown in Fig. 3. This potential is given by
V(x) 0 < x < R ,
V (x) = 0 x > R, (1.1)
∞ x < 0.
We call this a finite range potential, because the nontrivial part V(x) of the potential does not extend
beyond a distance R from the origin. Moreover, we have an infinite potential wall at x = 0. Thus
all the physics happens in x > 0 and incoming waves from x → ∞ will be eventually reflected back,
giving the physicist information about the potential. The restriction to x > 0 will also be useful when
1
we later study the more physical case of scattering in three dimensions. In that case, when using
spherical coordinates, we have r > 0 and most of the concepts to be learned here will apply.
Consider first the case where V(x) = 0, so that
(
0 x > 0,
V (x) = (1.2)
∞ x < 0,
as shown in figure 2. We have a free particle, except for the wall at x = 0. The solutions can be
constructed using linear combination of momentum eigenstates e±ikx where
2mE
k2 = . (1.3)
~2
Figure 2: The zero potential and its energy eigenstates φ(x) = sin kx are needed to compare with the V(x) 6= 0
problem.
The solution has an incoming wave e−ikx and an outgoing wave eikx combined in such a way that
φ(0) = 0 as required by the presence of the wall:
e−ikx eikx
φ(x) = − + = sin kx . (1.5)
2i 2i
The incoming wave is the first term to the right of the first equal sign and the reflected wave is the
second term. Both carry the same amount of probability flux, but in opposite directions.
Now consider the case V(x) 6= 0. This potential always acts over the finite range 0 < x < R and
we will eventually be interested in computing the energy eigenstate wavefunction ψ(x) in this region.
For the moment, however, let us consider ψ(x) in the region x > R. We will take the incoming wave
to be the same one we had for the zero-potential solution φ(x):
e−ikx
Incoming wave: − (1.6)
2i
The outgoing wave to be added to the above requires an eikx , to have the same energy as the incoming
wave solution. We now claim that the most general solution includes a phase factor, so we have
eikx
Outgoing wave: e2iδ , δ ∈ R. (1.7)
2i
2
We note that the phase δ cannot be a function of x: the free Schrödinger equation for x > R only
allows phases linear in x, but that would change the value of the momentum associated with the
outgoing wave, which we have already argued must be k. Furthermore, δ cannot be complex because
then we would fail to have equality between incident and reflected probability fluxes. This condition
requires that the norm-squared of the numbers multiplying the exponentials e±ikx be the same. Thus,
δ is a real function that depends on the energy E of the eigenstate and, of course, of the potential
V (x). A natural range of δ is from zero to 2π, but it will be easier to let it range through all of the
reals R in order to achieve a δ that is a continuous function of the energy. Assembling together the
incident and reflected components of the x > R solution we get
1
−e−ikx + eikx+2iδ = eiδ sin(kx + δ) ,
ψ(x) = 2i for x > R. (1.8)
This is called the canonical solution for x > R. For any particular feature of the solution φ(x) that
we might find at kx = a0 , giving x = a0 /k, we would find the same feature in ψ(x) at kx̃ + δ = a0 ,
giving x̃ = a0 /k − δ/k. For small δ > 0, the wave is pulled in by an amount δ/k, and the potential is
exerting attraction. For small δ < 0, the wave is pushed out by an amount |δ|/k, and the potential
is exerting repulsion. Note also that δ and δ ± π give exactly the same ψ(x). This is simplest to see
from the first form in (1.8).
Figure 3: The solution φ(x) for zero potential is shown with a dashed line. For x > R, the solution ψ(x)is
shown as a continuous line. As compared to φ(x), it is spatially shifted towards the origin by a distance δ/k.
We define the scattered wave ψs (x) as the extra wave in the solution ψ(x) that would vanish for
the case of zero potential, i.e.
ψ(x) = φ(x) + ψs (x). (1.9)
Note that both φ and ψ have the same incident wave, so ψs must be an outgoing wave. We find,
indeed
e−ikx eikx+2iδ e−ikx eikx eikx+2iδ eikx
ψs (x) = ψ(x) − φ(x) = − + + − = − , (1.10)
2i 2i 2i 2i 2i 2i
and therefore we get
ψs (x) = eiδ sin δ eikx = As eikx , with As ≡ eiδ sin δ . (1.11)
As is called the scattering amplitude, being the amplitude for the scattered wave. While we can’t
normalize our states yet (for that we need wave packets) the probability to scatter is captured by
|As |2 = sin2 δ. (1.12)
3
1.1 Time Delay
The quantity δ(E) determines the time delay of a reflected wave packet, as compared to a wave packet
that would encounter a zero potential. Indeed we claim that the delay is given by
∆t = 2~ δ′ (E) , (1.13)
where prime denotes derivative with respect to the argument, and the evaluation is done for the
central energy of the superposition that builds the wave packet. If ∆t < 0, then the particle spends
less time near x = 0, either because the potential is attractive and the particle speeds up, or because
the potential is repulsive and the particle bounces before reaching x = 0. If ∆t > 0, then the particle
spends more time near x = 0, typically because it slows down or gets temporarily trapped in the
potential.
We write the incoming wave in the form
Z ∞
ψinc (x, t) = dk g(k)e−ikx e−iE(k)t/~ , x > R , (1.14)
0
where g(k) is a real function peaked around k = k0 . We write the associated reflected wave by noticing
that the above is a superposition of waves as in (1.6) and the reflected wave must be the associated
superposition of waves as in (1.7). We must therefore change the sign of the momentum, change the
overall sign and multiply by the phase e2iδ :
Z ∞
ψrefl (x, t) = − dk g(k)eikx e2iδ(k) e−iE(k)t/~ , x > R . (1.15)
0
We now use the stationary phase approximation to figure out the motion of the wave packet peak. We
must have a stationary phase when k = k0 :
d E(k)t
0= kx + 2δ(k) −
dk ~
k0
dδ dE t
=x+2 −
dk k0 dk k0 ~
dE dδ dE t
=x+2 − (1.16)
dk k0 dE E(k0 ) dk k0 ~
1 dE dδ
=x+ 2~ − t
~ dk k=k0dE E(k0 )
~k0 dδ
=x+ 2~ − t ,
m dE E(k0 )
giving
~k0 dδ
x = t − 2~ . (1.17)
m dE E(k0 )
Here ~k0 /m is the familiar group velocity v0 of the wavepacket. If there had been no phase shift δ,
say because V(x) = 0, there would be no time delay and the peak of the reflected wavepacket would
follow the line x = v0 t. Therefore the delay, as claimed, is given by
∆t = 2~ δ′ (E(k0 )) . (1.18)
4
We can compare the time delay to a natural time in the problem. We first rewrite
dk dδ 2 dδ
∆t = 2~ = 1 dE
. (1.19)
dE dk dk
~ dk
We know that
1 dE dω
= = v0 , (1.20)
~ dk dk
so we can then write
dδ ∆t
= v0 . (1.21)
dk 2
1
Multiplying this by R, we have
1 dδ ∆t delay
= 2R = . (1.22)
R dk v
free transit time
0
The left-hand side is unit free, and the right hand side is the ratio of the time delay to the time the
free particle would take to travel in and out of the range R.
1.2 An Example
Consider the following example in which we have an attractive potential:
0 , for 0 < x < a ,
−V
V (x) = 0, for x > a , (1.23)
∞ , for x < 0 .
The potential is shown in Figure 4 where the energy E > 0 of the energy eigenstate is indicated with
a dashed line. The solution is of the form
(
eiδ sin(kx + δ) x > a,
ψ(x) = ′
(1.24)
A sin(k x) 0 < x < a.
Figure 4: An attractive potential. We are interested in calculating the energy eigenstates for all E > 0.
5
The solution in the region x > a is simply the canonical solution, and the solution in the region x < a
follows because we must have ψ(0) = 0. The constants k and k ′ are given by
2mE 2m(E + V0 )
k2 = , k ′2 = . (1.25)
~2 ~2
Matching ψ and ψ ′ at x = a we find the conditions that will eventually give us the unknown phase
shift δ:
k′ ′
tan ka + k cot k a
cot δ = k′
. (1.31)
1− ′
k cot k a tan ka
While this is a complicated formula to analyze directly, we can always plot it with a computer for
different values of V0 . As usual we characterize the well by the constant z0 , defined by
2mV0 a2
z02 = . (1.32)
~2
Note also that q
′
ka = z02 + (ka)2 . (1.33)
1 dδ
Figure 5 shows the phase factor δ, the quantity sin2 δ, the time delay a dk , and the amplitude |A|
inside the well as functions of ka, for z02 = 3.40.
Note that the phase δ begins at 0 for zero energy and reaches −π for infinite energy. The excursion
of the phase is thus π and as we will see, this happens because for this value of z0 the potential would
have one bound state.
The value of |As |2 = sin2 δ represents the probability of scattering and it peaks for the value of ka
for which the phase δ has absolute value π/2.
Next in the plot is the unit-free delay a1 dkdδ
. Notice that the delay is negative. This is easily
explained: as the particle moves over the well, its kinetic energy is increased by V0 , the particle speeds
up as it reaches and bounces off the wall.
6
The last plot shows the magnitude |A| of the constant that gives the amplitude of the wavefunction
in the region 0 < x < a.
At very large energy E ≫ V0 , the particle hardly notices the potential. Indeed we see that
δ approaches −π which, as noted below (1.8), is equivalent to δ = 0 and means no phase shift.
Accordingly sin2 δ → 0, meaning no scattering. We also have a1 dk
dδ
→ 0, meaning no time delay and,
finally, |A| → 1 as in the free solution.
Figure 5: Various quantities plotted as functions of ka, for z02 = 3.40. Top: the phase shift, going from zero to
−π. Second: the scattering amplitude |As |2 = sin2 δ. Third: the delay relative to the free transit time. Last:
the norm |A| of the amplitude of the wave inside the well.
Andrew Turner transcribed Zwiebach’s handwritten notes to create the first LaTeX version of this
document.
7
MIT OpenCourseWare
https://ocw.mit.edu
For information about citing these materials or our Terms of Use, visit: https://ocw.mit.edu/terms.