Beruflich Dokumente
Kultur Dokumente
ℏ 𝜕 ℏ 𝜕 ℏ 𝜕
𝑝𝑥 → , 𝑝𝑦 → , 𝑝𝑧 → , [4.2]
𝑖 𝜕𝑥 𝑖 𝜕𝑦 𝑖 𝜕𝑧
or
ℏ
𝑝𝑦 → ∇, [4.3]
𝑖
for short. Thus
𝜕Ψ ℏ2 2
𝑖ℏ =− ∇ Ψ + +𝑉Ψ, [4.4]
𝜕𝑡 2𝑚
where
2
𝜕2 𝜕2 𝜕2
∇ ≡ 2+ 2+ 2 [4.5]
𝜕𝑥 𝜕𝑦 𝜕𝑧
is the Laplacian, in Cartesian coordinates.
The potential energy V and the wave function Ψ are now functions of 𝒓 = (𝒙, 𝒚, 𝒛) and t.
The probability of finding the particle in the infinitesimal volume 𝑑 3 𝑟 = 𝑑𝑥𝑑𝑦𝑑𝑧 is |Ψ(𝑟, 𝑡)|2 𝑑3 𝑟,
and the normalization condition reads
∫ |Ψ|2 𝑑 3 𝑟 = 1, [4.6]
with the integral taken over all space. If the potential is the independent of time, there will be a
complete set of stationary states,
ℏ2 2
− ∇ ψn + 𝑉𝜓𝑛 = 𝐸𝑛 𝜓𝑛 , [4.8]
2𝑚
With the constants 𝑐𝑛 determined by the initial wave function, Ψ(𝑟, 0), in the usual way. (If the
potential admits continuum states, then the sum in Equation 4.9 becomes an integral).
*Problem 4.1
(a) Work out all of the canonical commutation relations for components of the operators
r and p: [𝑥, 𝑦], [𝑥, 𝑝𝑦 ], [𝑥, 𝑝𝑥 ], [𝑝𝑦 , 𝑝𝑧 ], and so on. Answer:
[𝑟𝑖 , 𝑝𝑗 ] = −[𝑝𝑖 , 𝑟𝑗 ] = 𝑖ℏ𝛿𝑖𝑗 , [𝑟𝑖 , 𝑟𝑗 ] = [𝑝𝑖 , 𝑝𝑗 ] = 0, [4.10]
(b) Show that
𝑑 1 𝑑
〈𝑟〉 = 〈𝑝〉, 𝑎𝑛𝑑 〈𝑝〉 = 〈−∇𝑉〉 [4.11]
𝑑𝑡 𝑚 𝑑𝑡
(Each of these, of course, stands for three equations—one for each component.)
(c) Formulate Heisenberg’s uncertainty principle in three dimensions. Answer:
𝜎𝑥 𝜎𝑝𝑥 ≥ ℏ/2, 𝜎𝑦 𝜎𝑝𝑦 ≥ ℏ/2, 𝜎𝑧 𝜎𝑝𝑧 ≥ ℏ/2 [4.12]
But there is no restriction on, say, 𝜎𝑥 𝜎𝑝𝑦.
Typically, the potential is a function only of the distance from the origin. In that case it is natural
to adopt spherical coordinates, (𝒓, 𝜽, 𝝓) (see Figure 4.1). In spherical coordinates the Laplacian
takes the form2
1 𝜕 2 𝜕 1 𝜕 𝜕 1 𝜕2
∇2 = (𝑟 ) + (𝑠𝑖𝑛𝜃 ) + ( ) [4.13]
𝑟 2 𝜕𝑟 𝜕𝑟 𝑟 2 𝑠𝑖𝑛𝜃 𝜕𝜃 𝜕𝜃 𝑟 2 sin2 𝜃 𝜕𝜙 2
ℏ2 1 𝜕 2 𝜕𝜓 1 𝜕 𝜕𝜓 1 𝜕 2𝜓
− [ 2 (𝑟 )+ 2 (𝑠𝑖𝑛𝜃 ) + 2 2 ( 2 )] + 𝑉𝜓 = 𝐸𝜓 [4.14]
2𝑚 𝑟 𝜕𝑟 𝜕𝑟 𝑟 𝑠𝑖𝑛𝜃 𝜕𝜃 𝜕𝜃 𝑟 sin 𝜃 𝜕𝜙
We begin by looking for solutions that are separable into products:
ℏ2 𝑌 𝑑 2 𝑑𝑅 𝑅 𝜕 𝜕𝑌 𝑅 𝜕 2𝑌
− [ (𝑟 )+ 2 (𝑠𝑖𝑛𝜃 ) + 2 2 ( 2 )] + 𝑉𝑅𝑌 = 𝐸𝑅𝑌
2𝑚 𝑟 2 𝑑𝑟 𝑑𝑟 𝑟 𝑠𝑖𝑛𝜃 𝜕𝜃 𝜕𝜃 𝑟 sin 𝜃 𝜕𝜙
Figure 4.1: Spherical coordinates: radius r, polar angle 𝜃, and azimuthal angle 𝜙.
Dividing by Y R and multiplying by −2𝑚𝑟 2 /ℏ2 :
1 𝑑 2 𝑑𝑅 2𝑚𝑟 2 1 1 𝜕 𝜕𝑌 1 𝜕 2𝑌
{ (𝑟 ) − 2 [𝑉(𝑟) − 𝐸]} + { (𝑠𝑖𝑛𝜃 ) + 2 }=0
𝑅 𝑑𝑟 𝑑𝑟 ℏ 𝑌 sin 𝜃 𝜕𝜃 𝜕𝜃 sin 𝜃 𝜕𝜙 2
The term in the first curly bracket depends only on r, whereas the remainder depends only on 𝜃
and 𝜙; accordingly, each must be a constant. For reasons that will appear in due course, I will write
this “separation constant” in the form 𝑙(𝑙 + 1):3
1 𝑑 𝑑𝑅 2𝑚𝑟 2
(𝑟 2 ) − 2 [𝑉(𝑟) − 𝐸] = 𝑙(𝑙 + 1); [4.16]
𝑅 𝑑𝑟 𝑑𝑟 ℏ
1 1 𝜕 𝜕𝑌 1 𝜕 2𝑌
{ (𝑠𝑖𝑛𝜃 ) + 2 } = −𝑙(𝑙 + 1) [4.17]
𝑌 sin 𝜃 𝜕𝜃 𝜕𝜃 sin 𝜃 𝜕𝜙 2
*Problem 4.2 Use separation of variables in Cartesian coordinates to solve the infinite cubical
well (or “particle in a box”):
𝜕 𝜕𝑌 𝜕 2𝑌
𝑠𝑖𝑛𝜃 (𝑠𝑖𝑛𝜃 ) + 2
= −𝑙(𝑙 + 1) sin2 𝜃 𝑌 [4.18]
𝜕𝜃 𝜕𝜃 𝜕𝜙
You may have encountered this equation already – it occurs in the solution to Laplace’s equation
in classical electrodynamics. As always, we try separation of variables:
1 𝑑 𝑑Θ 1 𝑑2Φ
{ [𝑠𝑖𝑛𝜃 (𝑠𝑖𝑛𝜃 )] + 𝑙(𝑙 + 1) sin2 𝜃} + =0
Θ 𝑑𝜃 𝑑𝜃 Φ 𝑑𝜙 2
The first term is a function only of 𝜃, and the second is a function only of 𝜙, so each must be
constant. This time I’ll call the separation constant 𝑚2 :4
1 𝑑 𝑑Θ
[𝑠𝑖𝑛𝜃 (𝑠𝑖𝑛𝜃 )] + 𝑙(𝑙 + 1) sin2 𝜃 = 𝑚2 [4.20]
Θ 𝑑𝜃 𝑑𝜃
1 𝑑2 Φ
= −𝑚2 [4.21]
Φ 𝑑𝜙 2
The 𝜙 equation is easy:
𝑑2 Φ
= −𝑚2 Φ => Φ(ϕ) = eimϕ
𝑑𝜙 2 [4.22]
[Actually, there are two solutions: exp(𝑖𝑚𝜙) and exp(−𝑖𝑚𝜙), but we’ll cover the latter by
allowing m to run negative. There could also be a constant factor in front, but we might as well
absorb that into Θ. Incidentally, in electrodynamics we would write the azimuthal function (Φ) in
terms of sines and cosines, instead of exponentials, because electric potentials must be real. In
quantum mechanics there is no such constraint, and the exponentials are a lot easier to work with.]
[4.23]
ow, when 𝜙 advances by 2𝜋, we return to the same point in space (see Figure 4.1), so it is natural
to require that5
In other words, exp[𝑖𝑚(𝜙 + 2𝜋)] = exp(𝑖𝑚𝜙) , 𝑜𝑟 exp(2𝜋𝑖𝑚) = 1. From this it follows that m
must be integer:
The 𝜃 equation,
𝑑 𝑑Θ
𝑠𝑖𝑛𝜃 (𝑠𝑖𝑛𝜃 ) + [𝑙(𝑙 + 1) sin2 𝜃 − 𝑚2 ]Θ = 0
𝑑𝜃 𝑑𝜃 [4.25]
May not be familiar. The solution is
|𝑚|
|𝑚| 𝑑
𝑃𝑙𝑚 (𝑥) ≡ (1 − 𝑥2) 2 ( ) 𝑃𝑙 (𝑥) [4.27]
𝑑𝑥
And 𝑃𝑙 (𝑥) is the lth Legendre polynomial. We encountered the latter (Equation 3.91) as orthogonal
polynomials on the interval (-1, +1); for our present purposes it is more convenient to define them
by the Rodrigues formula:
1 𝑑 𝑙 2
𝑃𝑙 (𝑥) ≡ 𝑙 ( ) (𝑥 − 1)𝑙
2 𝑙! 𝑑𝑥
[4.28]
For example,
1 𝑑 2
𝑃0 (𝑥) = 1, 𝑃1 (𝑥) = (𝑥 − 1) = 𝑥
2 𝑑𝑥
1 𝑑 2 2 1
𝑃2 (𝑥) = ( ) (𝑥 − 1)2 = (3𝑥 2 − 1)
4.2 𝑑𝑥 2
And so on. The first few Legendre polynomials were listed in Table 3.1. As the name suggests,
𝑃𝑙 (𝑥) is a polynomial (of degree l) in x, and is even or odd according to the parity of l. But 𝑃𝑙𝑚 (𝑥)
is not, in general, a polynomial – if m is odd it carries a factor of √1 − 𝑥 2 :
1 𝑑 1
𝑃20 (𝑥) = (3𝑥 2 − 1), 𝑃21 (𝑥) = (1 − 𝑥 2 )1/2 [ (3𝑥 2 − 1)] = 3𝑥 √1 − 𝑥 2 ,
2 𝑑𝑥 2
𝑑 1
𝑃22 (𝑥) = (1 − 𝑥 2 ) ( ) [ (3𝑥 2 − 1)] = 3(1 − 𝑥 2 ),
𝑑𝑥 2
etc. [On the other hand, what we need is 𝑃𝑙𝑚 (cos 𝜃), and √1 − cos2 𝜃 = 𝑠𝑖𝑛𝜃, so 𝑃𝑙𝑚 (cos 𝜃) is
always a polynomial in cos 𝜃, multiplied – if m is odd – by sin 𝜃. Some associated Legendre
functions of cos 𝜃 are listed in Table 4.1.]
Notice that l must be nonnegative integer for the Rodrigues formula to make any sense;
moreover, if |𝑚| > 𝑙, then Equation 4.27 says 𝑃𝑙𝑚 = 0. For any given l, then, there are (2𝑙 + 1)
possible values of m:
But wait! Equation 4.25 is a second – order differential equation: It should have two linearly
independent solutions, for any old values of l and m. Where are all the other solutions? Answer:
They exist, of course, as mathematical solutions to the equation, but they are physically
unacceptable because they blow up at 𝜃 = 0 and/or 𝜃 = 𝜋, and do not yield normalizable wave
functions (see Problem 4.4).
Now, the volume element in spherical coordinates7 is
𝑑 3 𝒓 = 𝑟 2 𝑠𝑖𝑛𝜃𝑑𝑟𝑑𝜃𝑑𝜙
[4.30]
So the normalization condition (Equation 4.6) becomes