QM 4.

Measurables, May 31, 2005 1

Chapter 4. What Do We Measure When We Study

Quantum Systems?

4.1. Generalities

x 1. Introduction. Understanding fully how the properties of a quantum sys-

tem are measured requires a detailed analysis of each specic measurement.

We do not have the time and you do not have the knowledge required for

this analysis. There are, however, features that are common to many mea-

surements and we study them here by looking at a few examples.

I will discuss a class of measurements that detect energy change. They

begin by \preparing" the molecules on which measurements are performed

in an initial state i in which their energy has a well-dened value Ei . Af-

ter this preparation, the measurement exposes the system to an \external"

agent, such as light or a stream of electrons of known energy or a stream of

molecules, etc. The interaction between the system and the external agent

causes both of them to change state. For example, a molecule in the state i

that is exposed to light may absorb a photon and change its state to f . We

say that photon absorption caused a transition from i to f . During this

change of state the energy changes from Ei to Ef . Energy conservation re-

QM 4. Measurables, May 31, 2005 2

quires that the energy h

( is the frequency of the light) of the lost photon

matches the energy increase Ef Ei in the molecule. When we determine

the frequencies of the absorbed photons we measure the magnitude of the

energy dierence Ef Ei .

To connect theory to such measurements, quantum mechanics postulates

that the values of the energy that can be observed in an experiment must be

the eigenvalues of the energy operator. This rule is valid in general: the result

of a measurement of a dynamical variable must be one of the eigenvalues

of the operator representing that dynamical variable. This postulate places

the eigenvalue problem at the center of quantum mechanics. If theory can

calculate the eigenvalues of various operators, it can predict the results of

various measurements.

In this chapter I discuss the \preparation" of the initial state, give three

examples of energy eigenvalues and use them to discuss electron energy-loss

spectroscopy and emission spectroscopy as two examples of measurements of

energy dierences. The purpose of the chapter is to exemplify how quantum

mechanics connects measurements to operator eigenvalues.

People performing measurements on quantum systems are very ingenious

and imaginative and have developed an astonishing number of clever methods

QM 4. Measurables, May 31, 2005 3

for obtaining information about molecules and for testing quantum theory.

The ones discussed here are the simplest.

4.2. The preparation of the initial state

x 2. Not all quantum systems have \well-dened" energy. I will discuss here

only measurements in which the system was prepared in an initial state i

that has a well-dened energy Ei . Here `well-dened' means that we are

certain that the energy of the molecule is equal to a specic eigenstate Ei of

the energy operator.

Preparing a system in a state of well-dened energy is not always easy.

There are practical situations in which we know only the probability P1 that

the system is in the state 1 having the energy E1 , the probability P2 that

the system is in the state 2 having the energy E2 , . . . . Here E1 , E2 , etc.,

are eigenvalues of the energy operator. This situation occurs whenever our

system consists of many molecules interacting with each other (i.e. they form

a gas or a liquid).

Why is there such an uncertainty? The molecules in a gas or a liquid

collide with each other and the energy of any given molecule is changed

by these collisions. At a given moment a given molecule might be in the

QM 4. Measurables, May 31, 2005 4

lowest energy state E1 . But, a short time later it might collide with a very

energetic molecule and its energy may increase to E2 . Then it may lose energy

in a subsequent collision and return to E1 . Thus, the energy of any given

molecule will change in time in a very chaotic manner. We call the lowest

energy state the ground state and we say that the molecule is excited, or that

it is in an excited state, when its energy is greater than that of the ground

state. While a given molecule will change its state in a random manner, the

number of excited molecules, at any given time, is constant. The higher the

temperature, the higher the mean energy of collision between the molecules

and the higher the number of excited molecules in the ensemble.

Statistical Mechanics provides equations that allow us to calculate the

probability Pn that we nd, in the ensemble, a molecule having the energy

En . I will not give this formula here, but inform you of a simple criterion. You

know, from your physics studies, that the average translational energy of a

molecule or an atom in an ensemble (gas, liquid, solid, or a polymer) is equal

to 3kB T =2. Here kB = 1:38071023 J/K and T is the temperature in kelvin.

This is the mean energy of two colliding molecules. Energy conservation tells

us that it is unlikely that a molecule with energy E1 will be excited by a

QM 4. Measurables, May 31, 2005 5

collision to an energy En , if

En E1 kB T (4.1)

The word `unlikely' is not accidental: the energy of a few molecules exceeds

this mean value (a student's grade can be higher than the average) and they

can excite their collision partner to a state of higher energy than the one

allowed by Eq. 4.1. But highly energetic molecules are as rare as perfect

students, and few molecules reach the high-energy states that violate the

rule given above.

It so happens that the energy required to excite the electrons in a molecule

is much larger than kB T , which is of order of 0:025 eV at room temperature.

This means that the electrons in molecules are in the ground state, at most

of the temperatures we are likely to use in the laboratory.

The energy to excite the vibrational motion is not as high. Nevertheless,

at room temperature most molecules are in the ground vibrational state, and

only a small fraction is excited.

The energy needed to excite the rotational motion is much smaller than

kB T and many molecules are rotationally excited at room temperature. It is

in such situations that we say that the energy is not well dened even though

we know from Statistical Mechanics how to calculate the probabilities P1 , P2 ,

QM 4. Measurables, May 31, 2005 6

. . . that the molecules have the rotational energies E1 , E2 , . . . . When we say

that the energy is not well dened, we mean that we do not know what the

energy of a specic molecule in the ensemble is.

A similar (but not identical) situation is encountered when the molecules

are exposed to an extremely short pulse of light, before a measurement. Such

pulses contain photons of a variety of energies and we do not know which

one is absorbed by the molecule. Because of this, the molecule can have one

of several energies and we have no way of knowing for sure which one it is.

We do not discuss in this chapter measurements on systems in which

the quantity being measured does not have a well-dened initial value. We

analyze such a situation when we discuss the \hot bands" in the absorption

and emission spectra of diatomic molecules (see Chapter 17).

4.3. Three examples of energy eigenstates

x 3. Introduction. To make the discussion of measurements in quantum me-

chanics as concrete as possible (at this stage), I will give three examples of

energy eigenstates, for three simple systems.

In general, the energy eigenstates are discrete, as long as the system

is conned to a nite volume. Sometimes the volume is so large that the

QM 4. Measurables, May 31, 2005 7

dierence between two consecutive energy eigenvalues is so small that we

cannot detect it experimentally. In such cases it appears that the energy of

the system takes continuous values.

We call the discrete energy eigenvalues En ; n = 1; 2; : : : the energy spec-

trum of the system. Often we represent the energy spectrum by an energy

level diagram (see Fig. 4.1). This gure implies the existence of a vertical

energy axis (not shown) and indicates the energy eigenvalues by horizontal

lines. The arrows between these lines show possible transitions in the mole-

cule when it interacts with an external agent (e.g. light, a passing electron).

For example, A indicates a transition from an initial state of energy E1 to

the state of energy E2 . Since the molecule gains the energy E2 E1 , the ex-

ternal agent must lose the same amount of energy. If we measure the energy

lost by the external agent we know the energy gained by the molecule. C

shows a transition in which the molecule gives the external agent the energy

E3 E1 (for example, the molecule emits a photon, and the emitted light

is the external agent). We can measure the energy gain by the agent and

determine the magnitude of E3 E1 .

Below I give energy spectra for three systems and discuss in more detail

how such a measurement can be performed.

QM 4. Measurables, May 31, 2005 8

E5

E4

E3

E2

C
A

E1

Figure 4.1: In this graph the vertical axis (which is not shown explicitly)

represents the energy. The horizontal lines indicate the energy eigenvalues of

the system (e.g. a molecule). The vertical arrows show possible transitions

in the system, as a result of its interaction with an external agent (light, a

passing electron, etc.)

QM 4. Measurables, May 31, 2005 9

x 4. A particle in a one-dimensional box. The energy eigenvalues of a particle

(e.g. an electron) in an one-dimensional box,

2 2 n2
h
En = ; n = 1; 2 : : : ; (4.2)
2mL2

were derived in Chapter 3 (see x20, Eq.3.41). Here m is the mass of the

particle, L is the length of the one-dimensional box in which the particle is

trapped, and h
is Planck's constant.

Quantum mechanics postulates the it is impossible for such a particle to

have an energy that satises

2 2 22
h 2 2 32
h
< < (4.3)
2mL2 2mL2

This is in stark contrast to classical mechanics, which has no principle that

imposes such a restriction. Classical theory allows the energy (or any other

dynamic variable) to change continuously, as we act on the system.

Let us try to reach a better understanding of these statements. To change

the energy of the particle I have to act on it with a force. By using the laws

of classical mechanics I can show that if I change the force innitesimally

and act with it for a short time, the energy of the particle changes by an

innitesimal amount. We don't need Newton to tell us that this is true: if I

change the force with which I hit a tennis ball by a small amount, its velocity
QM 4. Measurables, May 31, 2005 10

(hence energy) will change very little. I can thus give the ball any energy I

want, by manipulating the force. This is what we mean when we say that,

in classical mechanics, the energy can be changed continously.

This statement is not correct in a system that obeys Quantum Mechanics.

If the system is in the state E1 and the force is too small to place the particle

in the state E2 or E3 , etc., it would have no eect on the energy of the particle.

x 5. Isn't this a contradiction? At rst sight it seems that classical and

quantum mechanics are mutually exclusive: a measurement can either satisfy

the restrictions imposed by Eq. 4.3 or not. This puts us in a tight spot: we

know that both theories agree very well with a very large number of dierent

experiments. Why then are we in a situation that seems to say that if one

theory is correct, the other must be wrong?

This apparent mystery can be easily solved by examining the quantum-

mechanical condition Eq. 4.3. If mL is suciently large, the energy dierence

between two successive energy levels is smaller than the sensitivity of our

energy measurements. Because of this it appears to us that the energy can

be changed continuously.

Exercise 4.1 Use Eq. 4.2 to calculate the energy dierence En+1 En for

n = 1 or 10, for two cases: (a) L = 1 cm and m = 1 gr, and (b) L = 20

A
QM 4. Measurables, May 31, 2005 11

and m equal to the mass of the electron.

This kind of behavior is general. Whenever the predictions of the quan-

tum and classical theory are in conict, one nds that under certain con-

ditions (e.g. large mass) our instruments are not ne enough to detect the

restrictions imposed by quantum mechanics and the world behaves as if the

classical theory is correct. Classical mechanics is correct when our measure-

ments are not sensitive enough to reveal its inadequacy.

x 6. The vibrational energy of a diatomic molecule. The properties of a

diatomic molecule will be studied in detail in Chapters 16 and 17. Here I

give a minimum of information needed for understanding how the vibrational

energy is connected to the observations made by certain experiments.

The motion of a diatomic molecule is a composite of its translational,

rotational, and vibrational motions. To avoid being overwhelmed by details,

let us assume that somehow we can create an ensemble of diatomic molecules

that do not translate and rotate, but are allowed to vibrate. We can come

close to this by trapping the molecules in an argon ice.

The energy eigenvalues of a molecule that vibrates, but does not rotate

or translate, are given by

2
1 1
Ev = v + ! v +
h xe h
!; v = 0; 1; 2; : : : (4.4)
2 2
QM 4. Measurables, May 31, 2005 12

is Planck's constant, and ! and xe are numbers that change from

Here h

molecule to molecule. They are determined experimentally or are calculated

by using Quantum Mechanics. ! is the frequency of the vibration and has

units of sec1 . Because h

has units of energytime, h
! has units of energy

and so does xe h
!. v is an integer called the vibrational quantum number.

Spectroscopists have a habit of inventing strange energy units and they

use cm1 as a unit for vibrational energy. Don't ask why, the story is too long

and this is not the place to tell it. You can use the table given in Appendix

2 to convert cm1 to other energy units.

For the CO molecule, the constants have the values:

! = 2170:21 cm1
h (4.5)

! = 13:46 cm1
xe h (4.6)

These values allow us to calculate the vibrational energies given by Eq. 4.4

(see WorkBookQM.4.1).

The diagram of the vibrational energy levels of CO is shown in Fig. 4.2.

The energies were calculated in WorkBookQM.4.1, from Eq. 4.4 and the

parameter values given by Eqs. 4.5 and 4.6.

x 7. The energy eigenstates of the hydrogen atom. I will also use the hydrogen

atom | along with the vibrational energy of a diatomic molecule | to

QM 4. Measurables, May 31, 2005 13

v=3 E3 = 7760.62 cm-1

Vibrational energy

v=2 E2 = 5509.65 cm-1

v=1 E1 = 3285.6 cm-1

v=0 E0 = 1088.47 cm-1

Figure 4.2: Energy level diagram for the vibrational energies of CO. v is the

vibrational quantum number and Ev is the corresponding vibrational energy

eigenvalue.
QM 4. Measurables, May 31, 2005 14

explain how energy measurements are made. The energy eigenvalues of the

electron in the hydrogen atom are

e4
En = ; n = 1; 2; : : : (4.7)
2 n2
2(4"0 )2 h

Here e = 1:6022 1019 C is the electron charge in coulomb, h

= 1:0546

1034 is Planck's constant in joule seconds, "0 = 8:854 1012 C2 N1 m2

(N = newton, m = meter) is the permittivity of vacuum, and = (me

mp )(me + mp )1 kg is the reduced mass of the atom. In this expression,

me = 9:109 1031 kg is the electron mass and mp = 1:673 1027 kg is the

proton mass. Eq. 4.7 gives the energies in joules. To transform joule into

a more convenient unit, such as eV, we use the table given in Appendix 2.

Eq. 4.7 ignores the translational energy of the atom; the errors introduced

by this are very small.

There are an innite number of energy eigenvalues, and the lowest are

given in the energy level diagram in Fig. 4.3. The values of En were calculated

in WorkBookQM4.2.

4.4. Energy measurements by electron scattering

x 8. Introduction. We can now get back to our main purpose: to test the

statement that the energy of a system can only be one of the eigenvalues of
QM 4. Measurables, May 31, 2005 15

En
n=4 -0.850 eV
n=3 -1.511 eV

n=2 -3.400 eV

n=1 -13.597 eV

Figure 4.3: Energy level diagram for the hydrogen atom. The energies were

calculated from Eq. 4.7, in WorkBookQM.4.2.

QM 4. Measurables, May 31, 2005 16

^ (the Hamiltonian). To do this, we have to gure

the total energy operator H

out how to measure the energy. It will turn out that we can only measure

energy dierences.

A standard method for measuring energy in quantum physics is to bom-

bard the system with a stream of simple particles (e.g. electrons, or neutrons,

or photons), which I'll call `projectiles'. The molecules whose energy I want

to study are the `targets'. I will make the sample containing the molecules so

thin that a projectile either collides with one target molecule, or goes through

the ensemble without a collision. I will send towards the target projectiles

having a known energy ei and measure their energy ef after the collision.

The collision must conserve energy, so if Ei is the initial energy of the

molecule and Ef is its nal energy (after a collision) then

Ei + ei = Ef + ef (4.8)

The total energy Ei +ei before the collision is equal to the total energy Ef +ef

after the collision.

Since in this experiment I measure ei and ef I can calculate Ef Ei from

the energy conservation condition Eq. 4.8. In most practical situations, the

target (the diatomic molecule or the hydrogen atom) is in the ground state

(the state of lowest possible energy). To keep things simple I will analyze
QM 4. Measurables, May 31, 2005 17

here experiments in which this is the case.

Classical mechanics says that in such collisions the energy of the target

(and therefore the energy of the projectile) can take any value compatible

with energy conservation. In particular in an experiment in which Ei and

ei are xed, the energy of the projectile after the collision can vary continu-

ously between ei (the projectile gave no energy to the target) and zero (the

projectile gave all its energy to the target).

However, Quantum Mechanics says that this is not true: Ei and Ef cannot

^
take any good old value; they must be eigenvalues of the energy operator H.

Since the energy eigenvalues take discrete values, the dierence

ei ef = Ef Ei (4.9)

(which we measure in this experiment) must also take discrete values. This

equation says that if we perform the experiment at a xed incident projectile

energy ei , the energy ef of the projectile after the collision must take discrete

values.

The predictions made by the classical and the quantum mechanics are dra-

matically dierent, and they can be tested by scattering experiments. Such

experiments were rst performed by Franck and Hertz, who used electron

scattering by hydrogen atoms, and showed that the energy lost by electrons
QM 4. Measurables, May 31, 2005 18

was discrete and could be calculated from Eq. 4.7. This equation was derived

by Bohr, before the experiments were performed.

x 9. Electron scattering from a gas of diatomic molecules. Let us look now

at the results of an experiment in which the projectiles are electrons and the

target is a gas of diatomic molecules that can vibrate but cannot rotate or

translate. I will tell you later what happens in reality when rotation and

translation also take place.

According to Eq. 4.9, the dierence ef ei must be equal to the dierence

between two energy eigenvalues of the vibrating molecule. These eigenvalues

are given by Eq. 4.4. Combining the two equations leads to

ei ef = Evf Evi
" # " 2 #
1 1 2 1 1
= vf + ! vf +
h ! vi +
xe h ! vi +
h xe h
!
2 2 2 2
" 2 2 #
1 1
= (vf vi ) h
! vf + vi + xe h
! (4.10)
2 2

Let us apply this formula to a gas of CO molecules. When you learn

Statistical Mechanics, you will nd out that at a suciently low temperature,

most CO molecules in the gas have the lowest possible vibrational energy

! xe h
h !
E0 =
2 4
QM 4. Measurables, May 31, 2005 19

which corresponds to vi = 0. This simplies Eq. 4.10 to

" 2 #
1 1
ei ef = h
!vf xe h
! vf +
2 4
h i
= h ! vf2 + vf
!vf xe h (4.11)

Since vf can be 0, 1, 2, . . . , Quantum Mechanics makes a very peculiar

prediction: the dierence ef ei can only take the values

ef ei = 0 if vf = 0

ef ei = h
! 2xe h
! if vf = 1

ef ei = 2h! 6xeh
! if vf = 2
..
.

! = 2170:21 cm1 and xeh

For a CO molecule, h ! = 13:46 cm1 , and the

allowed energy losses for the electrons are given in Table 4.1 for the rst few

values of vf (see WorkBookQM.4.3).

This is a remarkable prediction. Classical physics proclaims that the

projectile may lose any amount of energy. The discreteness of the energy

lost by the electron | which reects the discreteness of the values of the

vibrational energy | is a purely quantum phenomenon.

Countless such experiments have been performed, and the energy lost by

electrons is that predicted by Quantum Mechanics. These and many other

QM 4. Measurables, May 31, 2005 20

vf ef ei

1 2197.131 cm1

2 4421.18 cm1

3 6672.15 cm1

4 8950.04 cm1

5 11254.9 cm1

Table 4.1: The energies that an electron can lose when colliding with a

vibrating CO molecule

experiments are a direct test of the statement that the energy of a system

(the molecule) must be an eigenvalue of its Hamiltonian.

Exercise 4.2 Calculate the possible energy losses in an electron-scattering

experiment in which the target is a gas of hydrogen atoms. Give the results

in eV.

Exercise 4.3 In a gas of CO molecules, a small minority have vibrational

energy corresponding to v = 1. Calculate the possible energies of the elec-

trons colliding with these molecules, if the initial energy of the electrons is

ei = 1:5 eV.
QM 4. Measurables, May 31, 2005 21

Exercise 4.4 What do you think: is the analysis performed for electrons

also valid if the projectiles are neutrons? What kind of dierences between

electron and neutron scattering do you expect? Consider both theoretical

and practical dierences.

x 10. A practical application of this experiment. The observation that the

change in energy of the electron beam is given by Eq. 4.11 can be used

to measure the vibrational frequency ! and the constant xe for a diatomic

molecule. Normally we use optical spectroscopy (photon absorption) for such

measurements, but there is one situation where electron scattering is advan-

tageous. Many people are interested in the behavior of molecules adsorbed

(this means stuck) on a solid surface. If I expose a clean metal surface (e.g.

Ni) to a gas (e.g. CO) and the surface is not very hot, the surface will be

covered with a layer of CO molecules. By estimating the size of a molecule,

I can calculate that there will be about 1014 molecules on a cm2 of surface.

This is an awfully small amount of material (calculate how many moles). To

make the situation worse, these molecules are spread out in a very thin layer

(about 3 108 cm thick). If I send a photon to bounce o this surface, the

chance that the photon is absorbed is very small.

However, the interaction of an electron with a molecule is much stronger

QM 4. Measurables, May 31, 2005 22

then the interaction of the photon with the molecule. Moreover, electron

detection is more ecient than photon detection. If I send a beam of electrons

to collide with the surface, a sucient number of them will lose energy to

the diatomic molecule. This number is big enough to allow me to detect

how much energy was lost. By measuring how much energy the electrons

lost, I measure the frequency of the diatomics stuck on the surface. Since

dierent diatomic molecules have dierent vibrational frequencies, I can tell

(by measuring the frequency) what kind of molecules are present on the

surface. In this way, I can perform chemical analysis of very small numbers

of molecules adsorbed on a surface.

I can also use this tool to study chemistry. Let us assume that I suspect

that CO dissociates when adsorbed on Ni. How can I tell if this happens?

If the molecule dissociates, the CO frequency is absent in the measurement

and I will observe the frequency of the O and C atoms vibrating in the

direction perpendicular to the surface. If CO does not dissociate, I can

do some interesting experiments. I can heat the surface and monitor the

departure of CO from the surface by observing that the number of electrons

losing energy goes down with temperature (less CO on the surface, fewer

electrons lose energy). Or I can put H2 on the surface and monitor if it

QM 4. Measurables, May 31, 2005 23

reacts with CO. If it does, the signal from CO decreases in time. These

procedures can be used not only for diatomics but for any kind of molecule;

they all have specic vibrational frequencies which can be used to identify

them.

This technique, called electron energy-loss spectroscopy, is routinely used

in surface science laboratories all over the world.1 It is more sensitive than

photon absorption experiments, but it has lower resolution (it does not mea-

sure frequency as accurately as does photon absorption). Other disadvan-

tages are: the experiment can be performed only at very low pressure (to

avoid collisions of the electrons with gas molecules) and sometimes the elec-

trons destroy the molecule we want to study.

4.5. Energy measurements by photon emission

x 11. Photon emission. Now that you know the story of electron energy-

loss experiments, you should wonder what happens to the system (e.g. a

vibrating molecule) after it takes energy from the electron. What is the

excited molecule going to do next? It turns out that quantum systems \do
1
H. Ibach and D. L. Mills, Electron Energy Loss Spectroscopy, Academic Press, New

York, 1982; G. A. Somorjai, Chemistry in Two Dimensions: Surfaces, Cornell University

Press, Ithaca NY, 1981

QM 4. Measurables, May 31, 2005 24

not like" to be in an excited state. In most cases (but not always), they will

lower their energy by emitting photons.

In this emission process, energy must be conserved. If the energy of the

system (e.g. a molecule) after photon emission is Ef , then

Ef Ei = h
(4.12)

where is the frequency of the emitted photon and h

is its energy.

Since must be positive, Ef must be smaller than Ei : the energy of the

nal state of the molecule must be lower than the energy of its initial state.

This is not surprising, since to create the photon (which has energy) the

molecule must lower its own energy.

Because the energy of the molecule is discrete, a molecule that is initially

in a state having the energy E3 can only emit photons of energy E3 E2 , or

E3 E1 . I assume here, as is usual, that E1 < E2 < E3 < : : :.

The emission process is often described by using an energy-level diagram

with the transitions indicated by arrows between the initial and the nal

state of the molecule. For the vibrating CO molecule, such a diagram is

shown in Fig. 4.4. The arrow pointing down indicates that a molecule in

the state with v = 3 and energy E3 = 7760:62 cm1 emitted a photon of

= 7760:62 3285:6 cm1 , leaving the molecule in the state

frequency h
QM 4. Measurables, May 31, 2005 25

v=3 E3 = 7760.62 cm-1

Vibrational energy

v=2 E2 = 5509.65 cm-1

v=1 E1 = 3285.6 cm-1

v=0 E0 = 1088.47 cm-1

Figure 4.4: The energy level diagram for the vibrational motion of CO. The

downward-pointing arrow indicates that the molecule in a state with v = 3

emitted a photon and ended in the state with v = 1. The photon energy is

E3 E1 . The upward-pointing arrow indicates an excitation of the molecule

(by photon absorption or by collision with an electron) from a state with

v = 0 to one with v = 1.
QM 4. Measurables, May 31, 2005 26

with v = 1 and the energy e1 = 3285.6 cm1 .

The arrow pointing up indicates that the molecule was excited from the

state with v = 0 to the state with v = 1. If the excitation was caused

by collision with an electron or a neutron, the projectile lost the energy

E1 E0 = 3285:6 1088:47 cm1 .

Because the energies of the molecule take discrete values, the energy of

the emitted photons must also be discrete.

x 12. Applications. Photon emission by excited molecules has many inter-

esting implications. Each excited atom or molecule emits light of a specic

frequency and this can be used, like a nger print, in analytical chemistry.

If your doctor wants to know if you have mercury poisoning, he could take

a bit of hair (Hg accumulates in hair) and zap it with a powerful laser. This

heats the hair and breaks it up into atoms, some of which are excited. The

excited atoms will emit light. If the emitted light has a frequency typical to

the mercury atoms, then you have some degree of mercury poisoning. The

doctor can tell, from the number of photons of emitted with that frequency,

how bad the poisoning is. This method, when pushed to extremes, can detect

a single atom (under favorable conditions).

Another obvious application is to lighting. If you take a neon gas at low

QM 4. Measurables, May 31, 2005 27

pressure and you pass an electric discharge through it, you excite the atoms

and they will emit light. Since they will emit light typical to neon, you have

yourself a source of colored light. If you want `white' light, you surround the

tube containing the gas with a ceramic material that converts light of one

color into light of many colors. Neon lights, white lights, halogen lights, all

function by using the same principle: use a current to produce fast charged

particles, which hit the atoms, excite them, and make them emit light.

Lasers are based on the same idea but use additional tricks that make the

emitted light have special properties. The scanner at the supermarket, your

CD player, and other laser-containing devices work because excited systems

(in this case electrons in a solid) feel an urge to lose energy by emitting light.

While this urge to be dull (unexcited) is widespread in the physical world

it is not always satised by photon emission. Many molecules nd relief

from excitation by doing chemistry. A whole eld of physical chemistry,

namely photochemistry, is based on the fact that some excited molecules get

rid of the excess energy by breaking into fragments. We benet from that.

Your eye contains molecules that absorb light and perform chemical reactions

that send electric signals to your brain. The brain turns these signals into

an image. The food chain is based on the photochemistry taking place in

QM 4. Measurables, May 31, 2005 28

the leaves of plants, which absorb photons and use their energy for chemical

synthesis. The chemical reaction caused by photon absorption can take place

only if it is faster than photon emission. There would hardly be any life on

earth if the molecules in the leaves lose energy by re-emitting the absorbed

photons.

Eorts to turn light into electricity in photoelectric cells are all trying

to get as many excited electrons in a solid as possible and to force them to

provide electricity rather than emit photons.

x 13. I conclude this chapter by warning you that our analysis of the mea-

surements performed in quantum mechanics is incomplete. As far as energy

conservation is concerned, the projectile can lose the energy (Ef Ei )=h if

this quantity is smaller than its initial energy. However, Nature is subtle

and there are other rules in this game, besides energy conservation. When a

collision with a projectile changes the energy of a quantum system from Ei

to Ef , we say that a transition from Ei to Ef has taken place. A transition

allowed by energy conservation takes place with a certain probability: some

transitions are very probable, others take place rarely, and still others don't

take place at all. To see how this works, let us assume that we perform an

energy-loss experiment in which the projectile (e.g. an electron) has enough

QM 4. Measurables, May 31, 2005 29

energy to excite the system from E0 to E1 or to E2 or to E3 . It so hap-

pens that the transition E0 ! E1 is very probable, that E0 ! E2 has low

probability, and that the probability of the transition E0 ! E3 is so low

that we might not be able to detect it. For some transitions the emission

probability is equal to zero: when this happens we say that the transition is

forbidden. Given these facts, what will we observe in an electron energy-loss

experiment?

The energy of most electrons after the collision will be the same as the

initial energy (before the collision) since there is a fair probability that the

electrons pass through the sample without losing energy. A large number

will have the nal energy ef = ei (E1 E0 )=h, a few will have energy

ef = ei (E2 E0 )=h, and the number of those emerging with the energy

ei (E3 E0 )=h is so low that it is below our detection limit. These results

could cause some embarrassment. If calculations predict that the eigenvalue

E3 exists and you do not see it in the experiment, you might conclude that the

theory is wrong. Not so. The same theory, when used properly, will predict

that the transition E0 ! E3 is very improbable and your experiment is unable

to detect the existence of the state with energy E3 . Other experiments, using

other projectiles will obey dierent rules and are able to detect the transition
QM 4. Measurables, May 31, 2005 30

E0 ! E3 . Electron scattering will detect the transition E3 ! E2 , conrming

the existence of the state of energy E3 .

I will tell you more about the probability of observing a given transition

in future chapters.