Quantum Mechanics

Dynamics of matter waves

• We learned that the discoveries such as blackbody radiation and photoelectric eect led to a general
conclusion that energy of electromagnetic waves is transferred and transmitted in discrete quanta.
The quantization of electromagnetic radiation could be interpreted as a manifestation of particle-like
properties of radiation, which was decisively demonstrated by the observation of Compton scattering.
A generalization of particle-wave duality to ordinary matter resulted in the de Broglie's proposal that
a special kind of wave is associated to ordinary particles, such as atoms or electrons. This proposal
found a dramatic conrmation in the observation of electron and neutron diraction on crystal lattices.

• We described de Broglie's waves by a complex function of coordinates, the wavefunction ψ(r, t). In
order to give it a physical interpretation we confronted the somewhat strange reality of quantum mea-
surements which produce random outcomes in order to accommodate quantization rules. We adopted
a probabilistic interpretation of the wavefunction by making an analogy with electromagnetism. This
led to the Born's postulate, that |ψ(r, t)|2 is the probability (density) that the particle would be found
in a measurement at location r at time t.
• If ψ(r, t) provides the true microscopic description of a particle, we need a wave equation which
species dynamics, that is how the particle moves in the presence of external forces. Newton's laws
specify dynamics in classical physics, but they need to govern only the time dependence of position
and momentum of a particle. Now we need an equation of motion which species how the entire
wavefunction evolves as a function of coordinates and time. Such an equation was formulated by
Schrodinger and it gave birth to modern quantum mechanics.

Schrodinger equation
• There are a few physical requirements we must satisfy in constructing the quantum equation of motion
(wave equation):

1. The fundamental requirement of matter wave dynamics is that it be compatible with classical
physics. We know that macroscopic objects are governed by Newton's laws, so any quantum
mechanical law must have a mathematical limit in which it reduces to classical laws. We will later
identify this limit with cases in which one can neglect the Planck's constant, h ≈ 0.
2. Another requirement from the quantum mechanical equation of motion is that it have a plane-
wave solution ψ(r, t) = exp(i(kr − ωt)) in the absence of external forces. Such a solution describes
a particle with well dened constant momentum p = ~k and energy E = ~ω , according to the de
Broglie's and Planck's postulates.

3. The equation must be linear in the wavefunction ψ(r, t) in order to allow superpositions of waves.:
if ψ1 (r, t) and ψ2 (r, t) are solutions, then ψ(r, t) = a1 ψ1 (r, t) + a2 ψ2 (r, t) must also be a solution
for arbitrary a1 and a2 . The experimentally observed diraction of matter waves on crystals is
completely consistent with such linear superposition. Also, the analogous wave superpositions are
familiar in electrodynamics to which matter wave dynamics is very similar.

• We start from the requirement 2 and consider the simplest free motion of a particle in the absence of
external forces. The wavefunction ψ(r, t) = exp(i(kr − ωt)) contains information about the particle's
momentum and energy which we can extract by taking appropriate partial derivatives:

 
∂ ∂ ∂
p = ~k ... − i~ x̂ + ŷ + ẑ ei(kx x+ky y+kz z−ωt) ≡ −i~∇ei(kr−ωt) = pei(kr−ωt)
∂x ∂y ∂z

∂ i(kr−ωt)
E = ~ω ... i~ e = Eei(kr−ωt)
∂t

37
 Now we use the requirement 1 and observe that classical mechanics relates the energy of a (non-
relativistic) particle to its momentum:
p2
E=
2m
If we multiply this equation by ψ(r, t) = ei(kr−ωt) and use the formulas above to extract p2 = |p|2 and
E from the plane-wave function, we obtain the following equation of motion:

~2 ∇2 ∂
− ψ(r, t) = i~ ψ(r, t)
2m ∂t

• This wave equation correctly describes de Broglie waves, but it is applicable only to the uniform
motion of free particles. How can we describe acceleration due to external forces? Since the above
wave equation naturally deals with energy of the particle, we can use potential energy V to describe
forces. Classical physics tells us that the particle's potential energy is converted to kinetic energy when
the particle accelerates, but its total energy E is constant:

p2
E= + V (r)
2m
Just as above, we now convert this into a wave equation:

~2 ∇2 ∂
− ψ(r, t) + V (r)ψ(r, t) = i~ ψ(r, t)
2m ∂t
The added term is constructed by requiring that the equation be linear in ψ(r, t) and as similar as
possible to its classical counterpart. This is a complete quantum-mechanical equation of motion which
Schrodinger wrote in 1925.

• We can demonstrate the compatibility of Schrodinger's equation with classical mechanics by considering
a wave-packet. A wave-packet is the wave with as small uncertainty of position and momentum as
allowed by the Heisenberg's principle, ∆px ∆x ∼ ~. Such a wave is localized in space but occupies
some small nite volume in order to have a reasonably well dened momentum. We must assume
that the potential V (r) varies slowly, so that it can be regarded constant in the volume occupied by
the wave-packet. The Schrodinger's equation describes in great detail how the wave-packet evolves
and propagates through space. However, if we are not interested in the internal changes of the wave-
packet, which anyway occur inside a very small volume, we can apply the Schrodinger's equation to
the carrier plane wave at the current location of the wave-packet. This gives us precisely the classical
equation of motion which governs the evolution of the (most probable) instantaneous momentum p
and position r of the wave-packet:
p2
E= + V (r)
2m
The assumption that the wave-packet is very small would be exact if the Planck's constant were zero
(∆px ∆x ∼ ~ ≈ 0). For all practical purposes, the Planck's constant is indeed extremely small in our
macroscopic world. However, inside atoms, the potential V (r) varies extremely rapidly over extremely
short distances, which becomes signicant despite the smallness of h. This is why we need to apply
the full quantum-mechanical theory to study the subatomic world.

Time-independent Schrodinger equation

• The general Schrodinger equation is time-dependent and predicts the full time evolution of a wave-
function. However, its full power is needed only in non-equilibrium situations. Most situations that
we shall consider occur in equilibrium and can be handled by a simplied time-independent equation.
For example, the equilibrium state of a Hydrogen atom should not change in time.

• We seek a stationary solution of the Schrodinger equation in the form:

ψ(r, t) = ψ(r)e−iωt

38
 Substituting this in the Schrodinger equation yields:

~2 ∇2
− ψ(r) + V (r)ψ(r) = Eψ(r)
2m
where E = ~ω and the factors of e−iωt on both sides were canceled out. This is the time-independent
(or stationary) Schrodinger equation. Its solution is a wavefunction ψ(r) which depends only on
coordinates. The wavefunction depends not only on the potential V (r), but also on the energy E.
• We distinguish two kinds of solutions ψ(r). Bound states are the solutions which can be normalized in
a nite volume V as: ˆ
d3 r |ψ(r)|2 = 1
V

Such states are localized. |ψ(r, t)|2 = |ψ(r)e−iωt |2 = |ψ(r)|2 as the probability
Since we interpret
(density) that the particle be detected in a measurement at location r, the normalization tells us that
the particle can be found inside the nite volume V with complete certainty. It turns out that such
solutions exist only if the energy E belongs to a discrete set of values, called energy levels. We say
that the energy spectrum of bound states is discrete.

• Extended states are the solutions ψ(r) which cannot be normalized in a nite volume because ψ(r) is
nite everywhere in space. A dierent normalization condition has to be used. Extended states are
typically found at high energies, when kinetic energy is so large that the potential V (r) cannot localize
the particle. The spectrum of extended states is continuous.

• The stationary Schrodinger equation allows us to calculate the energy spectrum of any quantum system.
Substituting a simple Coulomb potential V (r) ∝ |r|−1 for example provides a description of a Hydrogen
atom, that is an electron moving in the electric eld of a heavy proton. The electron is bound to
the proton in a neutral atom, so that its energy spectrum is discrete. The Schrodinger equation
predicts the spectrum which is completely consistent with all observations of Hydrogen radiation spectra
and hence provides a complete and successful microscopic theory of the Hydrogen atom (Bohr's and
Sommerfeld's descriptions were phenomenological). The extended states, on the other hand, describe
electron scattering from a Hydrogen ion (bare proton).

Some important properties

• The solutions of the stationary Schrodinger equation are labeled by so called quantum numbers. These
are either the physical measurable quantities that characterize the stationary quantum states of the
system, or their representatives (which are demonstrated in the examples below). Dierent systems
have dierent quantum numbers.

• Energy E is always a quantum number. In many situations multiple solutions of the stationary
Schrodinger equation can be found for the same energy. These solutions are said to be degenerate.
We must label degenerate solutions by additional quantum numbers. In general, conserved physical
quantities provide quantum numbers. For example, the potential V (r) ∝ |r|−1 of a Hydrogen atom
is spherically symmetric, so that angular momentum is conserved and represents a good quantum
number.

• The Schrodinger equation always has a trivial solution ψ(r) = 0, which describes the absence of a
particle.

• If ψ(r, t) is a solution of the time-dependent Schrodinger equation, so is Aψ(r, t) for any complex
constant A. However, these two solutions are physically equivalent. First, when we normalize the
wavefunction Aψ(r, t) we x the modulus |A|. Only the phase of A remains undetermined. However,
this phase does not aect any observable properties as we shall discuss later, and hence is unphysical.
Only relative phases matter in quantum mechanics and give rise to various quantum interference
phenomena.
 • If ψ1 (r, t) and ψ2 (r, t) are solutions of the time-dependent Schrodinger equation, so is A1 ψ1 (r, t) +
A2 ψ2 (r, t). On the other hand, A1 ψ1 (r)+A2 ψ2 (r) is the solution of the stationary Schrodinger equation
only is ψ1 (r) and ψ2 (r) are degenerate solutions at the same energy E .

• If ψ1 (r) and ψ2 (r) are the non-degenerate solutions of the stationary Schrodinger equation, with ener-
gies E1 and E2 respectively, then they are orthogonal:
ˆ
d3 r ψ1∗ (r)ψ2 (r) = 0

Quantum-mechanical eects
• This section illustrates a couple of important quantum phenomena which do not have classical ana-
logues. The examples are solved in one-dimensional space to keep things simple and demonstrate some
important calculation techniques.

Particle in an innitely deep quantum well

• Here we solve the one-dimensional Schrodinger equation:

~2 00
− ψ (x) + V (x)ψ(x) = Eψ(x)
2m
in the potential:  
 ∞ , x<0 
V (x) = 0 , 0<x<a
∞ , a<x
 

This potential is an innitely deep quantum well: a particle can be located only at 0<x<a since it
costs innite potential energy to place it anywhere else.

• We seek solutions at nite values for energy E. Also, ψ(x) must be nite in order to be normalizable.
Only such solutions are physically observable.

• Clearly, ψ(x) must be zero outside the potential well, because otherwise we get a nite value Eψ(x)
on the right-hand side and innite value V (x)ψ(x) on the left-hand side for x<0 or x > a.
• ψ(x) must vary smoothly with x. Otherwise, the second derivative on the left-hand side blows up and
we cannot satisfy the equation inside the well where V (x), ψ(x) and E are nite.

• Consequently, we must require ψ(0) = 0 and ψ(a) = 0. These are boundary conditions.
• Inside the well V (x) = 0 so the Schrodinger equation is very simple:

~2 00
− ψ (x) = Eψ(x) , 0<x<a
2m
The most general solution of this dierential equation is:

ψ(x) = A sin(kx) + B cos(kx)

where A and B are complex constants to be determined, and k can be thought of as a quantum number.
• We can immediately use the boundary conditions:

x = 0: ψ(0) = B = 0
x = a: ψ(a) = A sin(ka) = 0

40
 The rst condition eliminates B. The second condition could be satised by choosing A = 0, but this
would give us the trivial solution ψ(x) = 0. Therefore, we must choose a nite value for A, which still
cannot be determined, and select such k that sin(ka) = 0. This is readily accomplished by

π
k= n
a
where n is a positive integer. Note that choosing n = 0 would also give us the trivial solution ψ(x) = 0,
so we don't accept it. Furthermore, choosing n < 0 is equivalent to choosing a positive value for n
and reversing the sign of A. This does not generate a physically dierent solution: it is merely a
non-physical global phase factor eiπ = −1. All physically non-equivalent and non-trivial solutions are
generated for positive integers n.
• Now, substitute ψ(x) = A sin(kx) in the Schrodinger equation:

~2 k 2
A sin(kx) = E × A sin(kx) , 0<x<a
2m
We see that energy takes discrete values:

~2 k 2 π 2 ~2 2
E= = n , n ∈ {1, 2, 3 . . .}
2m 2ma2
• We can determine A only from the normalization condition:

ˆa ˆa  
a sin(2ka)
2
dx |ψ(x)| = |A| 2
dx sin2 (kx) = |A|2 1− =1
2 2ka
0 0

However, since ka = nπ , we nd sin(2ka) = sin(2πn) = 0. Therefore:

r
2
A= × eiθ
a
The value of θ is arbitrary, so set θ = 0.
• The normalized solutions of the Schrodinger equation are:
r
2  nπ 
ψn (x) = sin x , 0<x<a
a a
ψn (x) = 0 , x<0 or x>a
with energy levels
π 2 ~2 2
En = n
2ma2
obtained for n ∈ {1, 2, 3 . . .}.
• It is customary to call n a quantum number in a problem of this kind. All states are localized, so the
energy spectrum is fully discrete and easily labeled by one dimensionless (integer) number. Physically,
n simply represents energy.

Quantum tunneling through a potential barrier

• Here we solve the one-dimensional stationary Schrodinger equation of the potential:
 
 0 , x<0 
V (x) = V , 0<x<a
0 , a<x
 

where V is nite and positive. This potential describes a barrier. Classically, a particle has larger
energy if it sits within 0<x<a than anywhere else. A quantum particle is free on both the left and
right side of the barrier, and there is no localization. Therefore, all quantum states will be extended,
and the spectrum will be continuous.

41
• Outside the barrier the Schrodinger equation is simple:

~2 00
− ψ (x) = Eψ(x) , x<0 or x>a
2m
and the solutions are plane waves. This time we write the solution in the form

ψ(x) = AR eikx + AL e−ikx , x<0

ψ(x) = CR eikx + CL e−ikx , x>a

with unknown coecients AL , AR , CL , CR . This choice is motivated by the view of a right-moving
plane de Broglie wave AR eikx approaching the barrier from x → −∞ and partially reecting back
−ikx
as AL e . The subscripts L, R refer to the direction of wave propagation, left or right. The wave
2 2
scattering on the barrier is elastic: energy must be conserved, and since E = ~ k /2m the incident
and reected wave must have the same wavenumber k . The wave beyond the barrier must have the
same form, with possibly dierent coecients CL and CR . However, we shall seek solutions with
CL = 0 and ask only what fraction of the incident wave passes through the barrier and continues to
propagate rightwards (CR ). We could have considered a more complicated problem with CL 6= 0, but
then we would have two incident waves coming from opposite directions which both partially reect
and partially go through the barrier.

• Since all states are extended, energy spectrum is continuous. In fact, energy E can be arbitrarily small
because the barrier occupies only a nite fraction of space and hence has no physical impact in far-away
regions. The energy spectrum is simply the same as for free particles.

• In classical physics, a particle can cross the barrier only if its energy is larger than the barrier potential,
E >V. In that case, a classical particle always crosses the barrier, while for E<V the particle must
bounce back. In quantum mechanics even particles (waves) with E<V can partially pass through the
barrier and be detected on the other side. This is known as quantum tunneling.
• We wish to calculate reection and transmission probabilities for the incident wave. We need to solve
the Schrodinger equation inside the barrier:

~2 00
− ψ (x) = (E − V )ψ(x) , 0<x<a
2m
We have to distinguish two cases. For E>V we can nd solutions in the form:

0 0 ~2 k 02
ψ(x) = BR eik x + BL e−ik x , 0<x<a , E−V = >0
2m
but for E<V the solutions are:

~2 κ2
ψ(x) = βR eκx + βL e−κx , 0<x<a , E−V =− <0
2m
This can be easily veried by substituting in the Schrodinger equation only one of the two components
of ψ(x) and noting that k0 and −k 0 (or κ and −κ) give the same energy and can be used to form
superpositions. Here we assume k 0 > 0 or κ > 0.
• Now we need to relate the coecients by writing boundary conditions. The wavefunction must be
nite, and have a nite second derivative. This implies that ψ(x) and ψ 0 (x) must vary smoothly with
x, specically across the points x=0 and x = a. We are interested in:

|AL |2
R = ··· reection probability (the probability of bouncing back)
|AR |2
|CR |2
T = ··· transmission probability (the probability of passing through the barrier)
|AR |2

42
 • For E >V:
ψ(0) = AR + AL = BR + BL
ψ (0) = ik(AR − AL ) = ik 0 (BR − BL )
0

0 0
ψ(a) = BR eik a + BL e−ik a = CR eika
0 0
ψ 0 (a) = ik 0 (BR eik a − BL e−ik a ) = ikCR eika
These boundary conditions are four linear equations with ve unknowns. We can express four coe-
cients in terms of one coecient which can be ultimately determined only by normalization. Therefore,
we choose to express all coecients in terms of AR which we need not determine if we are interested
only in the reection and transmission probabilities. Solving these equations we nd:

−1
4k 2 k 02

R= 1+
(k 2 − k 02 )2 sin2 (k 0 a)
−1
(k 2 − k 02 )2 sin2 (k 0 a)

T = 1+
4k 2 k 02

• For E<V:
ψ(0) = AR + AL = βR + βL
ψ 0 (0) = ik(AR − AL ) = κ(βR − βL )
ψ(a) = βR eκa + βL e−κa = CR eika
ψ 0 (a) = κ(βR eκa − βL e−κa ) = ikCR eika
Solving these equations we nd:

−1
4k 2 κ2

R= 1+
(k 2 + κ2 )2 sinh2 (κa)
−1
(k 2 + κ2 )2 sinh2 (κa)

T = 1+
4k 2 κ2
which, of course, can be obtained from the E >V solution by substituting k 0 = −iκ. Note that in
all cases R+T =1 as it should be: the incoming particle must be either reected or transmitted. To
verify this, note that R and T have the forms R = 1/(1 + a) and T = 1/(1 + a−1 ), which always yields
R + T = 1.

Born's interpretation of the wavefunction, expectation values and quantum mea-

surements
• According to the Born's interpretation, P (r, t) = |ψ(r, t)|2 is the probability density to detect the
quantum particle at position r in a measurement at time t, provided that the wavefunction ψ(r, t) is
properly normalized. We can apply this interpretation to stationary solutions as well, in which case
the time dependence drops out: P (r, t) = |ψ(r)e−iωt |2 = |ψ(r)2 | = P (r).
• Since P (r) is the probability density of nding the particle at position r, we can calculate the average
position of the particle: ˆ ˆ
hri = d3 r rP (r) = d3 r ψ ∗ (r)rψ(r)

The result makes sense only if the particle is localized inside a nite volume.

43
 • We can generalize this to calculate expectation values for other physical quantities, for instance momen-
tum. Physical quantities, or observables, are represented by operators in quantum mechanics which act
on the wavefunction. For example, we constructed the Schrodinger equation by extracting momentum
and energy from the wavefunction and tting to the energy conservation law. We took a time derivative
to extract energy and spatial derivatives to extract the momentum vector from the wavefunction. We
can regard these derivatives as operators which convert the function they act on into another function.
Then, we associate the following operators to energy and momentum observables:

∂
i~ −→ energy: E
∂t
∂
−i~ −→ momentum in x direction: px
∂x
The average momentum of a particle in a stationary state is:
ˆ
hpi = d3 r ψ ∗ (r) (−i~∇) ψ(r)

where
∂ ∂ ∂
∇ = x̂ + ŷ + ẑ
∂x ∂y ∂z

• In general, the expectation value of an observable O = hÔi dened by the operator Ô can be time-
dependent: ˆ
O(t) = d3 r ψ ∗ (r, t)Ôψ(r, t)

• Hamiltonian Ĥ is the operator associated with energy. The Schrodinger equation tells us that

~2 ∇2
Ĥ = − + V (r)
2m
If we calculate its expectation value in a stationary state, we nd hĤi = E = const independent of
time.

Quantum measurement
• The Schrodinger equation can be written in the compact operator form:

Ĥψ = Eψ

This is known in mathematics as an eigenproblem. Think of ψ(r) as a vector with a complex element
for every point r in space. Clearly, such a vector has innitely many elements (uncountable innity),
but the ordering of elements labeled by r can be xed (for example, sort the element labels r by x,
then y, then z coordinates).Ĥ is a matrix dened in the same vector space (called Hilbert
Then,
space). If we nd a vector ψ such that Ĥψ is a constant E times the same vector ψ , then E is called an
eigenvalue and ψ an eigenvector. The normalization and orthogonality of wavefunctions we discussed
above translates into the normalization and orthogonality of vectors in this vector space.

• If the system is in a stationary state with energy E , then the measurement of energy will produce the
result E with complete certainty. Let ψE (r) be the eigenvector of Ĥ corresponding to the eigenvalue E .
−iEt/~
The stationary state of energy E is given by the full wavefunction ψ(r, t) = ψE (r)e . However,
any superposition of solutions to the time-dependent Schrodinger equation is also a non-stationary
solution: X
ψ(r, t) = AE ψE (r)e−iEt/~
E

where AE are arbitrary energy-dependent complex numbers, and the sum over energies includes both
the discrete portions of the spectrum and continuous segments (it becomes an integral in continuous
 segments). This time-dependent wavefunction is not an eigenstate of the Hamiltonian. Outcomes
of energy measurements in this state are random in accord with many experimental observations.
However,
ˆ 2
2 ∗
P (E) = |AE | = d3 r ψE (r, t)ψ(r, t)

is the probability (density) that E will be the outcome of a measurement (we use the orthogonality
property of wavefunctions to relate AE with the integral of wavefunctions). If a normalized stationary
wavefunction does not exist at some energy E0, E = E0 can never be an outcome of a measurement.
Therefore, only Hamiltonian eigenvalues can be obtained by measuring energy (but the outcomes are
random).

• What about other observables? For any observable represented by the operator Ô we dene an eigen-
problem:
Ôψ = Oψ
and require by analogy that the measurements can yield only random outcomes from the set of eigen-
values O of the operator Ô. The probability of measuring O is given by:

ˆ 2
∗
P (O, t) = d3 r ψO (r, t)ψ(r, t)

where ψO (r, t) is the eigenfunction of O corresponding to its eigenvalue O, and ψ(r, t) is the current
wavefunction of the system. This is the postulate about quantum measurement in quantum mechanics.

45