The WKB Approximation

Griffiths problem 8.2: Alternative derivation of WKB

(a)

ψ(x) = eif (x)/h̄

dψ i 0 if /h̄
= fe
dx h̄
2
d ψ i 00 if /h̄ 1
= f e − 2 f 02 eif /h̄
dx2 h̄ h̄
So the Schrödinger equation is
i 00 if /h̄ 1 p2 (x)
f e − 2 f 02 eif /h̄ = − 2 eif /h̄
h̄ h̄ h̄
or
ih̄f 00 − f 02 + p2 (x) = 0. (1)

(b)

f (x) = f0 (x) + h̄f1 (x) + h̄2 f2 (x) + · · ·

f0 = f00 + h̄f10 + h̄2 f20 + · · ·
(f 0 )2 = f002 + h̄(2f00 f10 ) + h̄2 (f102 + 2f00 f20 ) + · · ·

Plug these into equation (1) to find

ih̄[f000 + h̄f100 + · · ·] − [f002 + h̄(2f00 f10 ) + h̄2 (f102 + 2f00 f20 ) + · · ·] + p2 (x) = 0.

Whence, collecting like powers of h̄ (dimensional analysis!)

f002 = p2 (x) (2)

if000 = 2f00 f10 (3)
if100 = f102 + 2f00 f20 (4)

(c) From eqn. (2), we obtain

f00 (x) = ±p(x) (5)

Z
f0 (x) = ± p(x) dx. (6)

Meanwhile, take the derivative of (5) to find f000 = ±p0 (x). Plug this into the left-hand side of (3) to obtain

±ip0 (x) 2(±p(x))f10

=
ip0 (x)
f10 (x) =
2p(x)
Z 0 Z
i p (x) i dp i
f1 (x) = dx = = log p(x) (7)
2 p(x) 2 p 2

1
Meanwhile, Griffiths [8.10] is
C
R
ψ(x) = eif /h̄ ≈ p e±(i/h̄) p(x) dx
p(x)
Take the log of each side Z
i i 1
f (x) = ± p(x) dx + log C − log p(x)
h̄ h̄ 2
and incorporate the constant “log C” into the constant of integration to find
Z
h̄
f (x) = ± p(x) dx + i log p(x)
2
| {z } | {z }
same as (6) same as (7)

Griffiths problem 8.5: The quantum bouncer

(a)

V(x)

slope = mg
E

0 x
0 y

(b) Apply the Schrödinger equation

h̄2 d2 ψ
− + V (x)ψ(x) = Eψ(x)
2m dx2
to the quantum bouncer to find that

h̄2 d2 ψ
− = (E − mgx)ψ(x) for x > 0
2m dx2
with the boundary condition ψ(0) = 0.

Rewrite as
d2 ψ 2m2 g
 
E
= x− ψ(x),
dx2 h̄2 mg

2
change variable (shift of origin) to y = x − E/mg,

d2 ψ 2m2 g
= yψ(y),
dy 2 h̄2
and then change to the dimensionless variable
1/3
2m2 g

z= y
h̄2
to find
d2 ψ
= zψ(z).
dz 2
This O.D.E. has the solution
ψ(z) = aAi(z).

[It also has the solution Bi(z), but that solution is obviously unnormalizable.]

(c) In addition, the solution must satisfy the boundary condition

1/3 
2m2 g
 
E
ψ = 0 at x = 0, i.e. at y = −E/mg, i.e. at z = − .
h̄2 mg

In other words, the eigenenergies are related to the zeros z0 of Ai(z) through
1/3
h̄2

E = −mg z0
2m2 g
= −( 12 h̄2 mg 2 )1/3 z0
= −(3.766 × 10−23 Joule) z0

The zeros of Ai(z) are tabulated in Abramowitz and Stegun, Handbook of Mathematical Functions with
Formulas, Graphs, and Mathematical Tables. (And also in the Digital Library of Mathematical Functions,
release date 2011-08-29, National Institute of Standards and Technology, http://dlmf.nist.gov/9.9#T1,
table 9.9.1.)

zero energy (Joules)

−2.338 8.805 × 10−23
−4.088 15.395 × 10−23
−5.521 20.791 × 10−23
−6.787 25.559 × 10−23
−7.944 29.916 × 10−23
.. ..
. .

3
Griffiths problem 8.6: The quantum bouncer in the WKB approximation

This all hinges on Griffiths [8.47]:

Z x2
(n − 14 )πh̄ = p(x) dx
Z0 x2 p
= 2m(E − V (x)) dx
0
Z E/mg p
= 2m(E − mgx) dx [[use u = (mg/E)x]]
0
Z 1√
E √
= 2mE 1 − u du
mg 0
s
2E 3 1 √
Z
= 1 − u du [[use y = 1 − u]]
mg 2 0
s
2E 3 1 √
Z
= y dy
mg 2 0
s 1
2E 3

1 3/2
= y
mg 2 3/2 0
s
2E 3 2
 
=
mg 2 3

Solve for E:
E = ( 89 π 2 h̄2 mg 2 )1/3 (n − 41 )2/3

Plug in to build the table of values:

n energy (Joules)
1 8.738 × 10−23
2 15.371 × 10−23
3 20.777 × 10−23
4 25.549 × 10−23
5 29.910 × 10−23
.. ..
. .

The last value is accurate to 2 parts in 10,000!

Griffiths problem 8.14: WKB for the Coulomb problem

The effective potential is

e2 1 h̄2 `(` + 1)
 
Veff (r) = − +
4π0 r 2m r2

4
or, in atomic units,  
1 1 `(` + 1)
Veff (r) = − + .
r 2 r2
Thus
p
p(r) = 2[E − Veff (r)]
s  
1 `(` + 1)
= 2 E+ −
r 2r2
p s
2(−E) r `(` + 1)
= −r2 + − .
r (−E) 2(−E)
I prefer to use the constant (−E), which is positive, rather than E.

Veff(r)

p(r)

r
r1 = a r2 = b

The turning points are when p(r) = 0, that is,

2 rtp `(` + 1)
−rtp + − = 0,
(−E) 2(−E)
with solutions
p
1/(−E) ± 1/(−E)2 − 4`(` + 1)/2(−E)
rtp =
p 2
1 ± 1 − 2`(` + 1)(−E)
=
2(−E)

5
or
p
1− 1 − 2`(` + 1)(−E)
r1 = a =
2(−E)
p
1 + 1 − 2`(` + 1)(−E)
r2 = b = .
2(−E)
This means that p
2(−E) p
p(r) = (r − a)(b − r).
r
Note that for a < r < b, the quantity under the square root sign is positive, namely
r `(` + 1)
(r − a)(b − r) = −r2 + (a + b)r − ab = −r2 + − .
(−E) 2(−E)

Now we’re ready to invoke the WKB condition (in atomic units with h̄ = 1)
Z b
(n − 12 )π = p(r) dr
a
Z b p
p 1
= 2(−E) (r − a)(b − r) dr
a r
p π √ √
= 2(−E) ( b − a)2
2
p π √
= 2(−E) [(a + b) − 2 ab]
2" s #
p π 1 `(` + 1)
= 2(−E) −2
2 (−E) 2(−E)
"s #
π 2 p
= − 2 `(` + 1) .
2 (−E)

Thus
s
p 2
2(n − 21 ) + 2 `(` + 1) =
(−E)
p 2
4[(n − 12 ) + `(` + 1)]2 =
(−E)
1
(−E) = p .
2[(n − 21 ) + `(` + 1)]2

To convert from atomic units back to regular units, remember that the symbol E is shorthand for Ẽ =
E/(2 Ry) so
Ry
E=− 1
p .
[n − 2 + `(` + 1)]2