Connection Formulae
Connection Formulae
D.Rubin
February 18, 2008
Connection Formulae
The WKB approximation falls apart near a turning point. Then E − V → 0
so √1p → ∞. And because the momentum goes to zero the wavelength gets
very long and the approximation is only valid if the wavelength is short
compared to the distance over which the potential changes. But if want to
determine bound state energies, we need to be able to match wave functions
at the turning points. The strategy to overcome this limitation of the WKB
wave functions at the turning points is to
1. Linearize the potential at the turning point (x = 0). V (x) = V (0)+x dV
dx
2. Solve Schrodinger’s equation exactly near the turning point for the
linear potential There will be two cases to consider. One where dVdx
>0
and the particle has positive kinetic energy to the left of the turning
point (x < 0), and negative kinetic energy to the right (x > 0). The
other case is when dV
dx
< 0 and then there is negative kinetic energy to
the left and positive to the right. We consider first dV
dx
> 0.
h̄2 d2 ψ
− 2
+ (V (0) + xV 0 )ψ = Eψ
2m dx
d2 ψ 2m 0 p2
= − (E − V (0) − xV )ψ = − ψ
dx2 h̄2 h̄2
The turning point is at x = 0 and that is where V (x) = E. So V (0) = E
and we have
d2 ψ
= α3 xψ
dx2
where α3 = 2m h̄2
V 0 . Then if we define the dimensionless parameter
z = αx we have
d2 ψ
= zψ
dz 2
which is Airy’s equation and the general solution is
ψp = aAi(x) + bBi(x)
1
ψp is referred to as the patching wave function since its sole purpose is
to patch together the WKB wave functions on each side of the turning
point.
3. Determine the WKB wave function in the region of the linear potential.
It will be different on each side of the turning point. To the left of the
turning point for case one, the WKB wave function is
A B
ψl (x) = √ eiφ(x) + √ e−iφ(x)
p p
and
1Zx 0 0
Z x
3 1 2 3
φ(x) = p(x )dx = α 2 (−x) 2 dx0 = − (−αx) 2 (1)
h̄ 0 0 3
where
q q 3 1
p(x) = 2m(E − V ) ∼ 2m(−xV 0 ) = h̄α 2 (−x) 2
(Since we are to the left of the turning point x < 0 and the argument
of the square root is positive.) The WKB wave function to the left of
the turning point where the slope of the potential is positive is
A 2 3
B 2 3
ψl (x) = 1 1
e−i 3 (−αx) 2 + 1 1
ei 3 (−αx) 2 (2)
h̄ 2 (−α3 x) 4 h̄ 2 (−α3 x) 4
or in terms of the parameter z = αx
A 2 3
B 2 3
ψl (x) = 1 1 e−i 3 (−z) 2 + 1 1 ei 3 (−z) 2 (3)
(h̄α) (−z)
2 4 (h̄α) (−z) 2 4
To the right of the turning point, (x > 0), and the WKB wave function
is
C D
ψr (x) = q eφ(x) + q e−φ(x)
|p| |p|
and
1Zx 2 3
φ(x) =|p(x0 )|dx0 = (αx) 2 (4)
h̄ 0 3
and the WKB wave function to the right of the crossing point is
C 2 3
D 2 3
ψr (x) = 1 1
e 3 (αx) 2 + 1 1
e− 3 (αx) 2 (5)
h̄ (α3 x)
2 4 h̄ (α3 x)
2 4
2
or in terms of the parameter z
C z2 2 3
D −3z2 2 3
ψr (x) = 1 1 e
3 + 1 1 e (6)
(h̄α) 2 z 4 (h̄α) 2 z 4
4. Now we want to use the patching wave function to connect the WKB
wave to the left of the turning point with the WKB wave function to the
right. We use the large negatve z asymtotic form of the Airy functions
h i 3 3
3
1 2 π 1 1 i 23 (−z) 2 i π4 −i 32 (−z) 2 −i π4
Ai(z) ∼ √ 1 sin 3
(−z) 2 + 4
= √ 1 2i
e e −e e
π(−z) 4 π(−z) 4
z 0
h 3
i 3 3
i 23 (−z) 2 i π4 −i 32 (−z) 2 −i π4
1 2 π 1 1
Bi(z) ∼ cos (−z) + = e e +e e
2
√ 1 3 4 √ 1
π(−z) 4 π(−z) 4 2
Then setting
ψl (z) = aAi(z) + bBi(z)
we have
2 3 2 3
ψl (x) = 1
A
1 e−i 3 (−z) 2 + 1
B
1 ei 3 (−z) 2
(h̄α) 2 (−z) 4 (h̄α) 2 (−z) 4
3 3
1 i 32 (−z) 2 i π4 −i 32 (−z) 2 −i π4
= √ 1 (b − ia)e e + (b + ia)e e
2 π(−z) 4
Evidently
B b − ia π A b + ia π
1/2
= √ ei 4 and 1/2
= √ e−i 4
(h̄α) 2 π (h̄α) 2 π
Solving for a and b in terms of A and B we have that
π π
r r
b= (Aeiπ/4 + Be−iπ/4 ) and a = (Ae−iπ/4 + Beiπ/4 ) (7)
h̄α h̄α
and that takes care of matching the WKB wave function to the left of
the turing point to the patching wave function.
Now let’s do the same thing to the right of the turning point. Here we
use the large positive z asymtotic form of the Airy functions. Namely
2 3
Ai(z) ∼ √ 1
2 π(z)1/4
e− 3 z 2
z0
2 3
√ 1 1/4 e 3 z 2
Bi(z) ∼
π(z)
3
Setting
ψr (z) = aAi(z) + bBi(z)
we have
C 2 3
D 2 3 1 2 3 1 2 3
1 1 e3z2 + e− 3 z 2 = a √
1 1 1/4
e− 3 z 2 + b √ 1/4
e3z2
(h̄α) z
2 4 (h̄α) z 2 4 2 π(z) π(z)
from which we see that
s
4π π
r
a= D and b = C (8)
h̄α h̄α
Finally we equate equations 7 and 8 to eliminate a and b yielding the
connection formulae
s
4π π
r
D= (Ae−iπ/4 + Beiπ/4 ) (9)
h̄α h̄α
and
π π
r r
C= (Aeiπ/4 + Be−iπ/4 ) (10)
h̄α h̄α
In summary, for a barrier to the right, the connection formulae are
D = 21 (Ae−iπ/4 + Beiπ/4 ) C = Aeiπ/4 + Be−iπ/4
barrier to right
A = Deiπ/4 + 21 Ce−iπ/4 B = De−iπ/4 + 21 Ceiπ/4
4
and equation 4 will become
1Zx 2 3
φ(x) = |p(x0 )|dx0 = − (αx) 2 (12)
h̄ 0 3
The result is that we interchange A for B and C for D. The connection
formulae for a barrier to the left are given here for reference.
D = Ae−iπ/4 + Beiπ/4 C = 21 (Aeiπ/4 + Be−iπ/4 )
barrier to left
A = 21 Deiπ/4 + Ce−iπ/4 B = 12 De−iπ/4 + Ceiπ/4
5. Now that we know how to match WKB wave functions at the turning
points we can derive the quantization condition. Suppose that we are
trying to determine the energy of a bound state. Imagine something
like the harmonic oscillator potential with V 0 < 0 at the left turning
point (x = x1 ) and V 0 > 0 at the right turning point (x = x2 ). Then
the region x < x1 is classically forbidden, (we refer to it as region I)
and
C 1 x |p(x0 )|dx0 D − 1 x |p(x0 )|dx0
R R
ψI (x) = q e h̄ x1 + q e h̄ x1 (13)
|p| |p|
In region II the WKB wave function for x > x1 is
A i B −i
Rx Rx
p(x0 )dx0 p(x0 )dx0
ψII (x) = √ e h̄ x1
+ √ e h̄ x1
(14)
p p
In order that ψI (x) be finite for x 0, it must be that D = 0. Then
connection formulae for a barrier to the left yield
0 = Ae−iπ/4 + Beiπ/4 , → A = −iB, A = Ce−iπ/4 , B = Ceiπ/4 (15)
And using Equation 15 to substitute for A and B, Equation 14 be-
comes
C h̄i xx p(x0 )dx0 −iπ/4 C −i
R Rx
p(x0 )dx0 +iπ/4
ψII (x) = √ e 1 + √ e x1
h̄
(16)
p p
Next we have to connect WKB wave functions for region II and III
at x = x2 . In order to apply the connection formulae for a barrier to
the right we need to have the wave function in the form
A0 i B 0 −i
Rx Rx
p(x0 )dx0 p(x0 )dx0
ψ = √ e h̄ x2
+ √ e h̄ x2
(17)
p p
5
We can write Equation 16 in the same form as follows
C i(
R x2 Rx
p(x0 )dx0 + p(x0 )dx0 −iπ/4)
ψII (x) = √ e h̄ x1 x2
p
C −i (
R x2 Rx
p(x0 )dx0 + p(x0 )dx0 −iπ/4)
+ √ e h̄ x1 x2
p
Cei(θ−π/4) = A0 , Ce−i(θ−π/4) = B 0
where
1 Z x2
θ= p(x0 )dx0
h̄ x1
Now we can use the connection formulae for a barrier to the right to
connect ψII with ψIII .
C0 D0
1
Rx Rx
|p(x0 )|dx0 − h̄1 |p(x0 )|dx0
ψIII (x) = q e h̄ x1
+q e x1
(18)
|p| |p|
But C 0 = 0 so
1
cos θ = 0, ⇒ θ = (n + )π (19)
2
and Z x2
1
p(x)dx = h̄π(n + ) (20)
x1 2
Equation 20 determines the allowed energies of the system.