Chapter 2

Simple Model Systems: Solving the Schrödinger Equation

2.1Introduction

Chapter 1 set out the axioms. This chapter applies them to the smallest set of problems that can be solved exactly, in one dimension, and it is worth being explicit about why these particular problems and no others.

Each one is a permanent tool rather than an exercise:

All five of these reappear later. The three-dimensional atoms of Chapters 4 and 5 reduce, after separating the angles, to a one-dimensional radial equation of exactly the form solved here.

Remark 2.1.1First-pass route

Read stationary states, the infinite well, and the harmonic oscillator. Work the three self-checks before the Summary. This route shows twice how quantization enters: walls restrict the allowed wavelengths, while the oscillator ladder must have a lowest state. The finite well, tunnelling, and numerical project are optional extensions on a first reading.

RemarkPrerequisite recall: second-order equations

For ψ ″ + k 2 ψ = 0 , the general solution is A sin ⁡ k x + B cos ⁡ k x ; for ψ ″ − κ 2 ψ = 0 , it is A e κ x + B e − κ x . Boundary conditions select the allowed linear combinations and sometimes the allowed energies. The companion Mathematical Methods chapter on ordinary differential equations develops these facts; the two displayed solutions suffice here.

2.2Stationary States

The time-dependent Schrödinger equation (TDSE) of Chapter 1 is

i ℏ ∂ ∂ t | Ψ ( t ) ⟩ = H ˆ | Ψ ( t ) ⟩ . (2.1)
Theorem 2.2.1Separation of the TDSE

If H ˆ does not depend on time, then Eq. (2.1) admits solutions of the form

| Ψ ( t ) ⟩ = e − i E t / ℏ | ψ ⟩ , (2.2)

where | ψ ⟩ satisfies the time-independent Schrödinger equation (TISE)

H ˆ | ψ ⟩ = E | ψ ⟩ . (2.3)

Try | Ψ ( t ) ⟩ = f ( t ) | ψ ⟩ with | ψ ⟩ time-independent. Substituting into Eq. (2.1),

i ℏ f ˙ ( t ) | ψ ⟩ = f ( t ) H ˆ | ψ ⟩ .

Taking the inner product with ⟨ ψ | and dividing by f ⟨ ψ | ψ ⟩ ,

i ℏ f ˙ f = ⟨ ψ | H ˆ | ψ ⟩ ⟨ ψ | ψ ⟩ ≡ E ,

a constant, since the left side depends only on t and the right side not at all. Integrating gives f = e − i E t / ℏ , and substituting back gives Eq. (2.3).

Remark 2.2.1Why “stationary”

The state (2.2) does change in time — its phase rotates — but the phase cancels in every probability:

| Ψ ( x , t ) | 2 = | e − i E t / ℏ | 2 | ψ ( x ) | 2 = | ψ ( x ) | 2 ,

and likewise for the expectation value of any time-independent operator. Nothing observable evolves. Motion requires a superposition of stationary states with different energies, whose relative phases advance at different rates — the mechanism behind wave packets (Sec. 2.3.1) and quantum beats (Chapter 11).

In one dimension, with H ˆ = p ˆ 2 / 2 m + V ( x ) and p ˆ = − i ℏ d / d x , Eq. (2.3) reads

− ℏ 2 2 m d 2 ψ d x 2 + V ( x ) ψ ( x ) = E ψ ( x ) (2.4)

This is the equation solved five times below. Two general facts about it save work every time.

Theorem 2.2.2Continuity conditions

Wherever V ( x ) is finite, ψ and d ψ / d x are both continuous. Where V has an infinite jump, ψ remains continuous but d ψ / d x need not.

Rearranging Eq. (2.4), ψ ″ = 2 m ℏ 2 ( V − E ) ψ . If V is finite then ψ ″ is finite, so ψ ′ is continuous, and hence ψ is too. If V is infinite the argument fails for ψ ′ , but ψ must still be continuous for ψ ″ to be integrable.

Theorem 2.2.3Node counting

The bound states of Eq. (2.4) in one dimension are non-degenerate, can be chosen real, and may be ordered so that the n -th state ( n = 1 , 2 , 3 , … ) has exactly n − 1 nodes.

Theorem 2.2.3 is quoted rather than proved, but it is used constantly: it is how one confirms that a numerical solution has not skipped a level, and it reappears as the rule n − l − 1 for hydrogen in Chapter 4.

2.3The Free Particle

Take V ( x ) = 0 everywhere. Equation (2.4) becomes

d 2 ψ d x 2 = − k 2 ψ , k = 2 m E ℏ , (2.5)

with general solution

ψ k ( x ) = A e i k x + B e − i k x . (2.6)

Taking the two terms separately, e i k x moves to the right and e − i k x to the left, each with

E = ℏ 2 k 2 2 m , p = ℏ k . (2.7)

There is no boundary condition and hence no quantization: every E > 0 is allowed. This is the continuum, and it is the first appearance of the object that Chapters 9 to 11 depend on.

Remark 2.3.1Normalization of continuum states

A plane wave is not square-integrable: ∫ − ∞ ∞ | e i k x | 2 d x = ∫ d x diverges. Two standard repairs exist. One is to confine the particle to a large box of length L and normalize to 1 / L , taking L → ∞ at the end. The other, used throughout the research literature and adopted in Chapter 10, is to normalize to a Dirac delta,

⟨ k | k ′ ⟩ = δ ( k − k ′ ) or ⟨ ε | ε ′ ⟩ = δ ( ε − ε ′ ) , (2.8)

the second being energy normalization. Which convention is in use changes factors of 2 π and densities of states in every formula, so it must always be stated.

2.3.1Wave Packets and Group Velocity

A single plane wave is spread over all space and describes nothing localized. A particle is built as a superposition:

Ψ ( x , t ) = 1 2 π ∫ − ∞ ∞ A ( k ) e i ( k x − ω ( k ) t ) d k , ω ( k ) = E ( k ) ℏ = ℏ k 2 2 m . (2.9)

Expanding ω ( k ) about the central wavenumber k 0 , ω ≃ ω 0 + ( k − k 0 ) d ω / d k , the packet moves without changing shape at the group velocity

v g = d ω d k | k 0 = ℏ k 0 m = p 0 m , (2.10)

which is the classical velocity — a first instance of the correspondence principle. Keeping the next term in the expansion makes the packet spread, because different k components travel at different speeds.

2.4The Infinite Square Well

Let

V ( x ) = { 0 , 0 < x < a , ∞ , otherwise. (2.11)

Since V = ∞ outside, ψ must vanish there, and by Theorem 2.2.2 ψ is continuous, so

ψ ( 0 ) = ψ ( a ) = 0 . (2.12)

Inside, Eq. (2.4) is Eq. (2.5) again, with general solution most conveniently written

ψ ( x ) = A sin ⁡ k x + B cos ⁡ k x . (2.13)

Apply the boundary conditions in turn. At x = 0 , ψ ( 0 ) = B = 0 , leaving ψ = A sin ⁡ k x . At x = a ,

A sin ⁡ k a = 0 ⟹ k a = n π , n = 1 , 2 , 3 , … (2.14)

( n = 0 gives ψ ≡ 0 , which is not a state; negative n merely changes the sign.) Hence k n = n π / a and, from Eq. (2.5),

E n = ℏ 2 k n 2 2 m = n 2 π 2 ℏ 2 2 m a 2 , n = 1 , 2 , 3 , … (2.15)

Normalization fixes A :

1 = ∫ 0 a | A | 2 sin 2 ( n π x a ) d x = | A | 2 a 2 ⟹ A = 2 a , (2.16)

using ∫ 0 a sin 2 ⁡ ( n π x / a ) d x = a / 2 . Therefore

ψ n ( x ) = 2 a sin ( n π x a ) , 0 < x < a (2.17)

These are orthonormal, ∫ 0 a ψ m ψ n d x = δ m n , as Chapter 1's general theorem for Hermitian operators requires.

Remark 2.4.1Three lessons

First, confinement quantizes: the discrete spectrum came entirely from the two boundary conditions, not from any postulate. Second, the ground-state energy is not zero, E 1 = π 2 ℏ 2 / 2 m a 2 ; a confined particle cannot be at rest, which is the uncertainty principle in energy form. Third, ψ n has n − 1 nodes inside the well, confirming Theorem 2.2.3.

Example 2.4.1Numbers for an electron and for a molecule

Evaluate E 1 for an electron confined to a = 0.1 nm (an atom) and for a nitrogen molecule confined to a = 1 cm (a laboratory box).

Solution.

Use E 1 = π 2 ℏ 2 / 2 m a 2 . For the electron, with ℏ = 1.055 × 10 − 34 J s and m = 9.11 × 10 − 31 kg,

E 1 = π 2 ( 1.055 × 10 − 34 ) 2 2 ( 9.11 × 10 − 31 ) ( 1.0 × 10 − 10 ) 2 = 6.0 × 10 − 18   J = 38   eV .

Comparable to atomic binding energies — so quantization matters at this scale, as it must.

For N 2 , m = 4.65 × 10 − 26 kg and a = 10 − 2 m:

E 1 = π 2 ( 1.055 × 10 − 34 ) 2 2 ( 4.65 × 10 − 26 ) ( 10 − 2 ) 2 = 1.2 × 10 − 38   J = 7.5 × 10 − 20   eV .

Compare k B T ≈ 0.025 eV at room temperature: the level spacing is smaller by eighteen orders of magnitude, so the levels are utterly unresolvable and the motion is classical. The same formula covers both cases; only m a 2 differs.

2.5The Finite Square Well

The infinite well, and the two facts worth carrying forward: the levels rise as n^2, and n has n-1 interior nodes. Counting nodes is the check applied to every solution in this book.
Figure 2.1. The infinite well, and the two facts worth carrying forward: the levels rise as n 2 , and ψ n has n − 1 interior nodes. Counting nodes is the check applied to every solution in this book.

Let

V ( x ) = { − V 0 , | x | < a , 0 , | x | > a , V 0 > 0 , (2.18)

and seek bound states, − V 0 < E < 0 . Define

k = 2 m ( E + V 0 ) ℏ (inside) , κ = − 2 m E ℏ (outside) , (2.19)

both real and positive.

The solutions are oscillatory inside and exponential outside. Because V ( − x ) = V ( x ) , the Hamiltonian commutes with parity, so the eigenstates can be chosen even or odd — which halves the work.

Even solutions

Take

ψ ( x ) = { C cos ⁡ k x , | x | < a , D e − κ | x | , | x | > a . (2.20)

Matching ψ and ψ ′ at x = a (Theorem 2.2.2):

C cos ⁡ k a = D e − κ a , − C k sin ⁡ k a = − D κ e − κ a . (2.21)

Dividing the second by the first eliminates both constants:

k tan ⁡ k a = κ (2.22)

Odd solutions

Take ψ = C sin ⁡ k x inside and ψ = ± D e − κ | x | outside. The same division gives

− k cot ⁡ k a = κ (2.23)

Solving the conditions

Equations (2.22) and (2.23) are transcendental: they have no closed-form solution. The standard route is graphical. Introduce the dimensionless variables

z = k a , z 0 = a 2 m V 0 ℏ , (2.24)

and note from Eq. (2.19) that k 2 + κ 2 = 2 m V 0 / ℏ 2 , so κ a = z 0 2 − z 2 . The conditions become

tan ⁡ z = z 0 2 − z 2 z (even) , − cot ⁡ z = z 0 2 − z 2 z (odd) , (2.25)

to be solved for 0 < z < z 0 . Plotting both sides against z and reading off the intersections gives the bound-state energies.

The intersections are the bound states. Each crossing of z0^2-z^2/z with z is an even state and with - z an odd one, so they alternate; the quarter-circle always meets the first branch of z, however small z0 is, which is Theorem 2.5.1.
Figure 2.2. The intersections are the bound states. Each crossing of z 0 2 − z 2 / z with tan ⁡ z is an even state and with − cot ⁡ z an odd one, so they alternate; the quarter-circle always meets the first branch of tan ⁡ z , however small z 0 is, which is Theorem 2.5.1.
Theorem 2.5.1Number of bound states

The finite well of Eq. (2.18) has

N = ⌈ 2 z 0 π ⌉ (2.26)

bound states, alternating even, odd, even, . In particular N ≥ 1 : a one-dimensional well always binds at least one state, however shallow.

The last statement is worth remembering, and it is special to one dimension — in three dimensions a shallow well may bind nothing at all.

Remark 2.5.1Penetration into the forbidden region

Outside the well E < V , which is classically forbidden, yet Eq. (2.20) gives ψ ∝ e − κ | x | , nonzero for all finite x . The particle has a nonzero probability of being found where a classical particle could never be, over a characteristic depth 1 / κ . This is the same exponential tail that becomes tunnelling in the next section, and — in Chapter 4 — the reason a low- l electron can reach a nucleus that a centrifugal barrier appears to exclude it from.

The wavefunction does not stop at the wall. It decays into the classically forbidden region over the depth 1/, which is the quantity Remark 2.5.1 names and the next section turns into tunnelling.
Figure 2.3. The wavefunction does not stop at the wall. It decays into the classically forbidden region over the depth 1 / κ , which is the quantity Remark 2.5.1 names and the next section turns into tunnelling.

2.6The Potential Barrier and Tunnelling

Now invert the well. Let

V ( x ) = { V 0 , 0 < x < L , 0 , otherwise, (2.27)

and send a particle in from the left with 0 < E < V 0 — classically unable to pass.

Write the solution in three regions:

ψ ( x ) = { A e i k x + B e − i k x , x < 0 , F e κ x + G e − κ x , 0 < x < L , C e i k x , x > L , (2.28)

with k = 2 m E / ℏ and κ = 2 m ( V 0 − E ) / ℏ . The absence of a left-moving wave for x > L encodes the physical setup: nothing comes back from the far side.

Inside the barrier the wave decays rather than oscillates, and it decays without reaching zero. Whatever amplitude survives at x=L oscillates onward, and that surviving fraction is T. The three pieces are joined by imposing Theorem 2.2.2 — and &#x27; continuous at both walls — so the curve has no kink and no step anywhere; solving those four conditions is what produces Eq. (2.29).
Figure 2.4. Inside the barrier the wave decays rather than oscillates, and it decays without reaching zero. Whatever amplitude survives at x = L oscillates onward, and that surviving fraction is T . The three pieces are joined by imposing Theorem 2.2.2 — ψ and ψ ′ continuous at both walls — so the curve has no kink and no step anywhere; solving those four conditions is what produces Eq. (2.29).

Matching ψ and ψ ′ at x = 0 and at x = L gives four linear equations in A , B , F , G , C ; eliminating B , F and G leaves the ratio C / A , and the transmission coefficient T = | C / A | 2 is

T = [ 1 + V 0 2 sinh 2 ⁡ ( κ L ) 4 E ( V 0 − E ) ] − 1 (2.29)

For a thick or high barrier, κ L ≫ 1 , we may use sinh ⁡ x ≃ 1 2 e x , giving the form that is actually used:

T ≃ 16 E ( V 0 − E ) V 0 2 e − 2 κ L (2.30)
Remark 2.6.1Exponential sensitivity

The dominant factor is e − 2 κ L : the transmission falls exponentially with barrier width and with the square root of the barrier height. That extreme sensitivity is what makes tunnelling both a precise tool and a violent process. It underlies the scanning tunnelling microscope, radioactive α decay, and — the case relevant to this course — strong-field ionization, in which a laser field bends the atomic potential into a barrier of finite width so that the bound electron can escape without absorbing a photon.

Example 2.6.1Tunnelling through a thin barrier

An electron of energy E = 1 eV meets a barrier of height V 0 = 2 eV and width L = 0.5 nm. Find T , and then find it again for L = 1.0 nm.

Solution.

First κ . With V 0 − E = 1 eV = 1.602 × 10 − 19 J,

κ = 2 ( 9.11 × 10 − 31 ) ( 1.602 × 10 − 19 ) 1.055 × 10 − 34 = 5.12 × 10 9   m − 1 .

For L = 0.5 nm, κ L = 2.56 , so e − 2 κ L = e − 5.12 = 6.0 × 10 − 3 . The prefactor is 16 E ( V 0 − E ) / V 0 2 = 16 ( 1 ) ( 1 ) / 4 = 4 , so

T ≃ 4 × 6.0 × 10 − 3 = 2.4 × 10 − 2 .

About one electron in forty gets through.

Doubling the width to 1.0 nm gives κ L = 5.12 and e − 2 κ L = e − 10.24 = 3.6 × 10 − 5 , so

T ≃ 1.4 × 10 − 4 .

Doubling the width cut the transmission by a factor of 170 . That is Remark 2.6.1 in numbers, and it is why a tunnelling current measures distance to a fraction of an atomic diameter.

Left: across the whole range of energies the thick-barrier form of Eq. (2.30) is indistinguishable from the exact Eq. (2.29), which is why it is the one used. Right: the two numbers of Example 2.6.1, and the slope between them.
Figure 2.5. Left: across the whole range of energies the thick-barrier form of Eq. (2.30) is indistinguishable from the exact Eq. (2.29), which is why it is the one used. Right: the two numbers of Example 2.6.1, and the slope between them.

2.7The Harmonic Oscillator

Let

V ( x ) = 1 2 m ω 2 x 2 , so H ˆ = p ˆ 2 2 m + 1 2 m ω 2 x ˆ 2 . (2.31)

This system earns its place three times over: it is exactly solvable; any smooth potential near a stable minimum is approximately harmonic, which is why it describes every molecular vibration in Chapter 7; and its solution introduces the ladder-operator method that Chapter 3 reuses for angular momentum.

2.7.1Ladder Operators

Definition 2.7.1Annihilation and creation operators
a ˆ = m ω 2 ℏ ( x ˆ + i m ω p ˆ ) , a ˆ † = m ω 2 ℏ ( x ˆ − i m ω p ˆ ) . (2.32)
Theorem 2.7.1Commutator and Hamiltonian
[ a ˆ , a ˆ † ] = 1 , (2.33)

and

H ˆ = ℏ ω ( a ˆ † a ˆ + 1 2 ) = ℏ ω ( N ˆ + 1 2 ) , N ˆ ≡ a ˆ † a ˆ . (2.34)

For the commutator, expand using [ x ˆ , p ˆ ] = i ℏ :

[ a ˆ , a ˆ † ] = m ω 2 ℏ [ x ˆ + i p ˆ m ω , x ˆ − i p ˆ m ω ] = m ω 2 ℏ ( − i m ω [ x ˆ , p ˆ ] + i m ω [ p ˆ , x ˆ ] ) .

Both terms equal ℏ / m ω , so the bracket is 2 ℏ / m ω and the whole expression is 1 .

For the Hamiltonian, multiply out:

a ˆ † a ˆ = m ω 2 ℏ ( x ˆ 2 + p ˆ 2 m 2 ω 2 + i m ω [ x ˆ , p ˆ ] ) = m ω x ˆ 2 2 ℏ + p ˆ 2 2 ℏ m ω − 1 2 .

Multiplying by ℏ ω ,

ℏ ω a ˆ † a ˆ = m ω 2 x ˆ 2 2 + p ˆ 2 2 m − ℏ ω 2 = H ˆ − ℏ ω 2 ,

which rearranges to Eq. (2.34).

2.7.2The Spectrum

Why the spectrum is what it is. a^ climbs and a descends in steps of ; the ladder would run down forever and give unbounded negative energies were it not for a0=0, which is the single fact that fixes the ground state and with it every level above it.
Figure 2.6. Why the spectrum is what it is. a ˆ † climbs and a ˆ descends in steps of ℏ ω ; the ladder would run down forever and give unbounded negative energies were it not for a ˆ | 0 ⟩ = 0 , which is the single fact that fixes the ground state and with it every level above it.

Let | n ⟩ be an eigenket of N ˆ with N ˆ | n ⟩ = n | n ⟩ . From Eq. (2.33),

[ N ˆ , a ˆ ] = − a ˆ , [ N ˆ , a ˆ † ] = + a ˆ † , (2.35)

so a ˆ | n ⟩ has eigenvalue n − 1 and a ˆ † | n ⟩ has n + 1 : the operators step down and up the ladder.

The ladder must have a bottom, because

n = ⟨ n | a ˆ † a ˆ | n ⟩ = ‖ a ˆ | n ⟩ ‖ 2 ≥ 0 . (2.36)

If n were not a non-negative integer, repeated application of a ˆ would eventually produce a state with negative eigenvalue, contradicting Eq. (2.36). The only escape is that the chain terminates on a state with a ˆ | 0 ⟩ = 0 , whence n = 0 , 1 , 2 , …

Theorem 2.7.2Harmonic oscillator spectrum
E n = ℏ ω ( n + 1 2 ) , n = 0 , 1 , 2 , … (2.37)

with

a ˆ | n ⟩ = n | n − 1 ⟩ , a ˆ † | n ⟩ = n + 1 | n + 1 ⟩ , | n ⟩ = ( a ˆ † ) n n ! | 0 ⟩ . (2.38)

The normalization constants follow from Eq. (2.36): ∥ a ˆ | n ⟩ ∥ 2 = n gives n , and ∥ a ˆ † | n ⟩ ∥ 2 = ⟨ n | a ˆ a ˆ † | n ⟩ = ⟨ n | ( N ˆ + 1 ) | n ⟩ = n + 1 gives n + 1 .

Remark 2.7.1Two features to carry forward

The levels are equally spaced by ℏ ω — unique among the systems in this chapter, and the reason a vibrational spectrum shows one dominant line rather than a converging series (Chapter 8). And the ground-state energy is 1 2 ℏ ω , not zero: the zero-point energy. A chemical bond is never still, even at absolute zero.

2.7.3Position Representation

The levels are equally spaced, which no other potential in this chapter manages: the infinite well gave n^2. Each n still carries its n nodes, and each still leaks past the classical turning points where the level meets the parabola.
Figure 2.7. The levels are equally spaced, which no other potential in this chapter manages: the infinite well gave n 2 . Each ψ n still carries its n nodes, and each still leaks past the classical turning points where the level meets the parabola.

To obtain the wave functions, write a ˆ | 0 ⟩ = 0 in position space. With p ˆ = − i ℏ d / d x , Def. 2.7.1 gives

( x + ℏ m ω d d x ) ψ 0 ( x ) = 0 ⟹ d ψ 0 ψ 0 = − m ω ℏ x d x , (2.39)

which integrates to a Gaussian. Introducing the natural length scale and the dimensionless coordinate

ξ = x m ω ℏ , (2.40)

and normalizing with ∫ e − ξ 2 d ξ = π ,

ψ 0 ( x ) = ( m ω π ℏ ) 1 / 4 e − ξ 2 / 2 . (2.41)

Applying a ˆ † repeatedly, as in Eq. (2.38), generates the rest. The result is

ψ n ( x ) = ( m ω π ℏ ) 1 / 4 1 2 n n ! H n ( ξ ) e − ξ 2 / 2 (2.42)

where H n are the Hermite polynomials, given by

H n ( ξ ) = ( − 1 ) n e ξ 2 ( d d ξ ) n e − ξ 2 , (2.43)

the first few being

H 0 = 1 , H 1 = 2 ξ , H 2 = 4 ξ 2 − 2 , H 3 = 8 ξ 3 − 12 ξ , H 4 = 16 ξ 4 − 48 ξ 2 + 12 . (2.44)

H n has n real zeros, so ψ n has n nodes — consistent with Theorem 2.2.3 once the counting is shifted for the label starting at n = 0 . The parity is ( − 1 ) n , since H n contains only even or only odd powers.

Example 2.7.1Zero-point energy of a real bond

The H 2 molecule has a vibrational wavenumber ν ˜ = 4401 cm − 1 . Find ℏ ω in eV and the zero-point energy, and compare with the bond dissociation energy of 4.48 eV.

Solution.

Convert using 1 cm − 1 = 1.2398 × 10 − 4 eV:

ℏ ω = 4401 × 1.2398 × 10 − 4 = 0.546   eV ,

so the zero-point energy is

E 0 = 1 2 ℏ ω = 0.273   eV .

This is 6 % of the dissociation energy — small but not negligible, and it must be included whenever bond strengths are compared. It also explains why the measured dissociation energy of H 2 differs from that of D 2 : the heavier isotope has smaller ω = k / μ and hence a lower zero-point energy, so it sits deeper in the same potential well and is harder to break. The potential curve is identical; only the nuclear mass differs.

2.7.4First-Pass Self-Checks: Boundaries and the Oscillator

The same free particle in the same length, counted under two boundary conditions. Hard walls give n=1,2, and one state per level; periodicity gives m=0,1,2,, so every level except m=0 is doubly degenerate and the staircase climbs in steps of two. The two curves are enumerations of the allowed quantum numbers, not fits: at any energy they differ, yet their slopes agree in the limit of many states, which is why either box gives the same density of states. Self-check 2 does the counting that this figure draws, and the companion course Condensed Matter Physics starts from the periodic staircase.
Figure 2.8. The same free particle in the same length, counted under two boundary conditions. Hard walls give n = 1 , 2 , … and one state per level; periodicity gives m = 0 , ± 1 , ± 2 , … , so every level except m = 0 is doubly degenerate and the staircase climbs in steps of two. The two curves are enumerations of the allowed quantum numbers, not fits: at any energy they differ, yet their slopes agree in the limit of many states, which is why either box gives the same density of states. Self-check 2 does the counting that this figure draws, and the companion course Condensed Matter Physics starts from the periodic staircase.
Example 2.7.2Self-check 1: a probability in the infinite well

For the ground state in a well on 0 < x < a , find the probability of finding the particle in the left half.

Solution.

Using Eq. (2.17) with n = 1 ,

P ( 0 < x < a / 2 ) = 2 a ∫ 0 a / 2 sin 2 ( π x a ) d x .

Since ∫ sin 2 ⁡ ( b x ) d x = x / 2 − sin ⁡ ( 2 b x ) / ( 4 b ) ,

P = 2 a [ x 2 − a 4 π sin ( 2 π x a ) ] 0 a / 2 = 2 a a 4 = 1 2 .

The answer also follows from the symmetry | ψ 1 ( x ) | 2 = | ψ 1 ( a − x ) | 2 ; the integral verifies it.

Example 2.7.3Self-check 2: periodic boundaries and counting k states

Put a free particle in a one-dimensional box of length L with the periodic condition ψ ( x + L ) = ψ ( x ) . Find the allowed k values and the number of states in an interval [ k , k + d k ] .

Solution.

For ψ k ( x ) = L − 1 / 2 e i k x , periodicity requires

e i k ( x + L ) = e i k x ⟹ e i k L = 1 ⟹ k n = 2 π n L , n ∈ ℤ .

Adjacent values are separated by Δ k = 2 π / L . Therefore the number of one-particle states in d k is

d N = d k Δ k = L 2 π d k

for one direction and one internal state. Counting both signs over a positive interval of magnitudes doubles this result; counting spin also doubles it. Say which of those choices has been made, every time: a density of states is meaningless until it is stated whether it counts one spin or both, and whether it is per sample, per volume or per cell.

This short bridge from a box to k -space counting is used twice in the companion courses. Condensed Matter Physics Chapter 1 turns it into the free-electron density of states; Statistical Physics Chapter 11 turns the same count into the phase-space density n λ T 3 that decides whether a gas needs quantum statistics at all. The counting is identical in both; only the question asked of it differs.

Example 2.7.4Self-check 3: the oscillator ground state by two routes

Derive the normalized oscillator ground state first from the differential Schrödinger equation and then from the ladder condition. Show that the two routes give the same function and energy.

Solution.

Try a normalizable Gaussian ψ 0 ( x ) = A e − α x 2 / 2 in

− ℏ 2 2 m ψ 0 ″ + 1 2 m ω 2 x 2 ψ 0 = E 0 ψ 0 .

Because ψ 0 ″ = ( α 2 x 2 − α ) ψ 0 , substitution gives

[ ℏ 2 α 2 m + 1 2 ( m ω 2 − ℏ 2 α 2 m ) x 2 ] ψ 0 = E 0 ψ 0 .

This must hold for every x , so the x 2 coefficient vanishes: α = m ω / ℏ . The constant part then gives E 0 = ℏ ω / 2 . Normalization yields

1 = | A | 2 ∫ − ∞ ∞ e − α x 2 d x = | A | 2 π α , A = ( m ω π ℏ ) 1 / 4 .

The ladder route starts from a ˆ | 0 ⟩ = 0 , which becomes Eq. (2.39) and integrates to the same Gaussian with the same normalization, Eq. (2.41). Finally, Eq. (2.34) gives E 0 = ℏ ω ( 0 + 1 / 2 ) . Differential and algebraic methods have found the same normalized state, not merely the same energy.

2.8Summary

2.9Exercises

  1. The infinite well, start to finish. Derive Eqs. (2.15) and (2.17) yourself, applying both boundary conditions and carrying out the normalization integral. Then verify ∫ 0 a ψ 1 ψ 2 d x = 0 explicitly.

  2. Matching at a step. Derive Eq. (2.22) from Eq. (2.21), and Eq. (2.23) by the same route for the odd solutions. Explain why dividing the two matching equations is legitimate and what it accomplishes.

  3. Counting bound states. A well has V 0 = 5 eV and half-width a = 0.2 nm, for an electron.

    1. Compute z 0 from Eq. (2.24).

    2. Use Theorem 2.5.1 to predict the number of bound states, and say which are even and which odd.

    3. Solve Eq. (2.25) graphically or numerically for the lowest z , and convert it to an energy in eV.

  4. Tunnelling. Repeat Example 2.6.1 for a proton instead of an electron, at the same energy, height and width L = 0.5 nm. By what factor does T change, and which quantity in κ is responsible? Comment on why tunnelling is an electronic phenomenon far more often than a nuclear one.

  5. Oscillator algebra.

    1. Reproduce the proof of Theorem 2.7.1.

    2. Derive Eq. (2.35) from Eq. (2.33).

    3. Using Eq. (2.38), compute ⟨ n | x ˆ 2 | n ⟩ by writing x ˆ in terms of a ˆ and a ˆ † , and hence verify ⟨ V ⟩ = 1 2 E n — the virial theorem for a quadratic potential.

  6. Hermite polynomials. Use Eq. (2.43) to generate H 1 and H 2 and check them against Eq. (2.44). Then verify that ψ 1 from Eq. (2.42) is normalized, using ∫ ξ 2 e − ξ 2 d ξ = π / 2 .

2.10Project: Solving One-Dimensional Problems Numerically

The problem. Build a general numerical solver for Eq. (2.4) and validate it against every exact result in this chapter. This solver is reused in Chapters 3, 4, 5 and 7, so build it to be trusted.

Work in atomic units ( ℏ = m = 1 ), where Eq. (2.4) is − 1 2 ψ ″ + V ψ = E ψ .

  1. On paper. Discretize on a uniform grid x i = x 0 + i h using ψ ″ ( x i ) ≃ ( ψ i + 1 − 2 ψ i + ψ i − 1 ) / h 2 . Write down the matrix 𝐇 explicitly, giving its diagonal and off-diagonal entries.

  2. On the computer. Validate against the infinite well: take V = 0 on ( 0 , a ) with ψ = 0 at both ends, and compare the lowest five eigenvalues with Eq. (2.15). Report the relative error and how it scales when you halve h — you should see it fall by about four, since the finite difference is second-order accurate.

  3. On the computer. Validate against the harmonic oscillator: take V = 1 2 x 2 on a domain wide enough that ψ has decayed, and confirm E n = n + 1 2 for n = 0 … 5 . Plot your ψ 0 and ψ 2 against Eqs. (2.42) and (2.44).

  4. On the computer. Now the finite well. Take V = − V 0 on | x | < a and zero outside, choose V 0 and a giving z 0 ≈ 5 , and find all bound states. Check the count against Theorem 2.5.1, check the parity alternation, and compare each energy with a numerical solution of Eq. (2.25).

  5. On the computer. Plot one bound state on a logarithmic vertical axis and confirm that the tail outside the well is a straight line of slope − κ , with κ from Eq. (2.19). This is Remark 2.5.1 made quantitative.

  6. On the computer. Tunnelling. Rather than eigenvalues, compute T ( E ) from Eq. (2.29) for a barrier of your choice and plot it against E from 0 to 2 V 0 on a logarithmic scale. Mark E = V 0 . Comment on what happens above the barrier, where classically T would be exactly 1 .

  7. What has to move. One animation: a Gaussian wave packet built as in Eq. (2.9) approaching your barrier, with | Ψ ( x , t ) | 2 shown as it splits into a reflected and a transmitted part. Caption it with the fraction that got through, and compare that fraction with Eq. (2.29) evaluated at the packet's central energy.

The check. Part (b) is the one that certifies the solver. If the error does not fall by a factor of four when h is halved, the discretization is wrong — most often a missing factor of 1 2 or h 2 in the matrix. Fix it before going further; every later chapter depends on this code.

Be ready to answer. In part (f), transmission above the barrier is not 1 , and oscillates. Explain the oscillation physically — what is interfering with what — and identify the energies at which T returns exactly to 1 .