1 THE PRINCIPLES OF QUANTUM THEORY

Table 1.2 The postulates of quantum mechanics

2
|!| = 1

I. The state of the system is described as fully as possible by the

wavefunction (r1, r2, . . . ) where (r1, r2, . . . ) are the spatial coordinates of
the particles (1, 2, . . . ).
II. For a system described by the normalized wavefunction (r1, r2, . . . ),
the probability of finding particle 1 in the volume element d1 at r1, particle
2 in the volume element d2 at r2, etc. is equal to | |2d1d2 . . . .
III. For each observable property of a system there is a corresponding
hermitian operator ) built from the following position and linear
momentum operators:
X=x

Yx =

41

$ d
i dx

ikx

Re e = cos kx

ikx

Im e = sin kx

Fig. 1.15 The square modulus of a wavefunction corresponding

to a definite state of linear momentum is a constant, so it
corresponds to a uniform probability of finding the particle
anywhere.

IV. Suppose the system is described by a wavefunction that is an

eigenfunction of ) with eigenvalue :
) =
Then the outcome of a measurement of the observable property will
be .
V. Suppose the system is described by a normalized wavefunction that is
a linear combination of eigenfunctions of ):

= c1 1 + c 2 2 + $ =

c
k

with )k = kk

Then when the value of an observable is measured, each measurement

gives one of the eigenvalues k with a probability equal to |ck |2. The mean
(that is, average) value of the measurements is equal to the expectation
value .

= *) d

We have reached an important point in our study of quantum

mechanics. Having described the postulates, which are summarized in Table 1.2, we can now use them to understand atomic
and molecular structure and chemical change. This task will
occupy our attention for the remainder of the text.

1.9 The Heisenberg uncertainty principle

The conclusion that, if the momentum is specified precisely,
then it is impossible to predict the location of the particle is one
conclusion that we can draw from the Heisenberg uncertainty
principle proposed by Werner Heisenberg in 1927:
It is impossible to specify simultaneously, with arbitrary precision, both the momentum and the position of a particle.

Complementary observables
The fact that particles are described by wavefunctions and that
the outcome of observations depends on the properties of operators and eigenvalues leads to profound dierences between
quantum mechanics and classical mechanics. We can begin to
appreciate these dierences by considering the wavefunction for
a particle travelling in one dimension towards positive x with
linear momentum k$. As we saw in Section 1.7, the wavefunction is Neikx, where N is the (real) normalization factor.
Where is the particle? To answer this question, we use
Postulate II to calculate the probability density:
| |2 = (Neikx)*(Neikx) = N2(eikx)(eikx) = N2

particle (Fig. 1.15). In other words, if the wavefunction of the

particle is given by Neikx, then we cannot predict where we will
find the particle. The extraordinary conclusion is that, if we
know the linear momentum precisely, then we can say nothing
about the position. This complementarity of two observables, in
this case linear momentum and position, pervades the whole of
quantum mechanics.
How do we recognize complementary observables, and what
are their consequences? Can we specify the energy of a molecule
at the same time, for instance, as its dipole moment or are they
complementary too? First, we consider linear momentum and
position in more detail, then generalize to other properties.

(1.18)

This probability density is independent of x, so, wherever we

look on the x-axis, there is an equal probability of finding the

Before discussing the principle further, we must establish its

other half: that if we know the position of a particle exactly then
we can say nothing about its momentum. The argument draws
on the idea of regarding a wavefunction as a superposition of
eigenfunctions, and runs as follows.
If we know that the particle is at a definite location, its wavefunction must be large there and zero everywhere else (Fig. 1.16).
Such a wavefunction can be created by superimposing a large
number of harmonic (sine and cosine) functions, or, equivalently, a number of eikx functions. In other words, we can create
a sharply localized wavefunction, called a wavepacket, by forming a linear combination of wavefunctions that correspond to
many dierent linear momenta. The superposition of a few harmonic functions gives a wavefunction that spreads over a range

42

1 THE PRINCIPLES OF QUANTUM THEORY

!
Location
of particle

x
Fig. 1.16 The wavefunction for a particle at a well-defined
location is a sharply spiked function which has zero amplitude
everywhere except at the particles position.

of locations (Fig. 1.17). However, as the number of wavefunctions in the superposition increases, the wavepacket becomes
sharper on account of the more complete interference between
the positive and negative regions of the individual waves. When
an infinite number of components is used, the wavepacket is
a sharp, infinitely narrow spike, which corresponds to perfect
localization of the particle. Now the particle is perfectly localized.

!
x

5
21

Fig. 1.17 The wavefunction for a particle with an ill-defined

location can be regarded as the superposition of several
wavefunctions of definite wavelength that interfere
constructively in one place but destructively elsewhere.
As more waves are used in the superposition (as given by
the numbers attached to the curves), the location becomes
more precise at the expense of uncertainty in the particles
momentum. An infinite number of waves is needed to
construct the wavefunction of a perfectly localized particle.

However, we have lost all information about its momentum

because, as we saw above, a measurement of the momentum will
give a result corresponding to any one of the infinite number of
waves in the superposition and which one it will give is unpredictable. Hence, if we know the location of the particle precisely
(implying that its wavefunction is a superposition of an infinite
number of momentum eigenfunctions), then its momentum is
completely unpredictable.
The quantitative version of the Heisenberg uncertainty principle is
pq 12 $

In this expression p is the uncertainty in the linear momentum parallel to the axis q, and q is the uncertainty in position
along that axis. These uncertainties are precisely defined, for
they are the root mean square deviations of the properties from
their mean values:
p = {p2 p2}1/2

[1.19b]

Table 1.3* Constraints of the uncertainty principle

Variable 1
Variable 2

px

py

pz

x
y
z

spreadsheet to construct superpositions of cosine

px

py

k=1

pz

introduced to keep the superpositions with the same overall

magnitude. Explore how the probability density 2 (x) changes
with the value of N.

q = {q2 q2}1/2

If there is complete certainty about the position of the particle

(q = 0), then the only way that eqn 1.19a can be satisfied is
for p = , which implies complete uncertainty about the
momentum. Conversely, if the momentum parallel to an axis is
known exactly (p = 0), then the position along that axis must
be completely uncertain (q = ).
The p and q that appear in eqn 1.19 refer to the same direction
in space. Therefore, whereas simultaneous specification of the
position on the x-axis and momentum parallel to the x-axis
is restricted by the uncertainty relation, simultaneous location
of position on x and motion parallel to y or z is not restricted.
Table 1.3 summarizes the restrictions that the uncertainty principle implies.

interActivity Use mathematical software or an electronic

functions as (x) = (1/N )cos(kx), where the constant 1/N is

(1.19a)

* Pairs of observables that cannot be determined simultaneously with arbitrary

precision are marked with a blue rectangle; all others are unrestricted.