Вы находитесь на странице: 1из 189

The University of Hong Kong

Department of Physics

PHYS2627 Introductory Quantum Physics


Teaching notes

January 2009
Contents

Table of Contents i

1 Electromagnetic Radiation Behaving as Particles 1

1.1 Classical Particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.2 The Electromagnetic Waves . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.3 Blackbody Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.4 The Photoelectric Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

1.5 The Production of X-ray . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

1.6 The Compton Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

1.7 Pair Production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

1.8 Dual Nature of Electromagnetic Radiation . . . . . . . . . . . . . . . . . . 20

2 Matter Behaving as Waves 21

2.1 Classical Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

2.2 Matter Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

2.3 Particle Diffraction and Interference . . . . . . . . . . . . . . . . . . . . . . 24

2.4 The Heisenberg Uncertainty Principle . . . . . . . . . . . . . . . . . . . . . 30

2.5 Is it a Wave or a Particle? . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

2.6 A Two-Slit Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

3 The Schrödinger Equation 39

i
3.1 The Free-Particle Schrödinger Equation . . . . . . . . . . . . . . . . . . . . 39

3.2 The Schrödinger Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

3.3 Stationary States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

3.4 Physical Conditions: Well-Behaved Functions . . . . . . . . . . . . . . . . 44

3.5 Expectation Values and Uncertainties . . . . . . . . . . . . . . . . . . . . . 45

4 Solutions of Time-Independent Schrödinger Equation I 49

4.1 Classical and Quantum-mechanical Bound States . . . . . . . . . . . . . . 49

4.2 Solving the TISE for a Piecewise-constant Potential . . . . . . . . . . . . . 51

4.3 Particle in a Box - The Infinite Well . . . . . . . . . . . . . . . . . . . . . . 52

4.4 The Finite Well . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

4.5 The Simple Harmonic Oscillator . . . . . . . . . . . . . . . . . . . . . . . . 62

5 Solutions of Time-Independent Schrödinger Equation II 65

5.1 The Potential Step . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

5.2 Barriers and Tunneling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

5.3 The 3D Infinite Well . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

6 Structure of the Atom 79

6.1 The Atomic Nature of Matter . . . . . . . . . . . . . . . . . . . . . . . . . 79

6.2 The Rutherford Model of Atom . . . . . . . . . . . . . . . . . . . . . . . . 80

6.3 Line Spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

6.4 The Bohr Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91

6.5 The Frank-Hertz Experiment . . . . . . . . . . . . . . . . . . . . . . . . . 97

7 The Hydrogen Atom 99

7.1 The Schrödinger Equation for the Hydrogen Atom . . . . . . . . . . . . . . 99

7.2 Solving the TISE for the Hydrogen Atom . . . . . . . . . . . . . . . . . . . 101

7.3 Quantization of Energy and Angular Momentum . . . . . . . . . . . . . . . 104

ii
7.4 Probability Densities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

7.5 Spectroscopic Notation and Selection Rule . . . . . . . . . . . . . . . . . . 114

7.6 Orbital Magnetism and the Normal Zeeman Effect . . . . . . . . . . . . . . 115

8 Many-Electron Atoms 121

8.1 Electron Spin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121

8.2 Identical Particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126

8.3 The Pauli Exclusion Principle . . . . . . . . . . . . . . . . . . . . . . . . . 131

8.4 Electronic States in Many-Electron Atoms . . . . . . . . . . . . . . . . . . 133

8.5 The Periodic Table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135

8.6 Properties of the Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . 137

8.7 X-ray Spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142

9 Statistical Mechanics 146

9.1 A Simple Thermodynamic System . . . . . . . . . . . . . . . . . . . . . . . 147

9.2 Boltzmann Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

9.3 Quantum Distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157

9.4 The Quantum Gas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163

9.5 Massless Bosons: The Photon Gas . . . . . . . . . . . . . . . . . . . . . . . 169

9.6 The Laser . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171

A X-ray Diffraction 177

B Differential Equations 179

C Separation of Variables of the Time-independent Schrödinger Equation


for the Hydrogen Atom 182

D Magnetic Dipole 184

iii
Chapter 1

Electromagnetic Radiation Behaving


as Particles

Until the end of 19th century, classical physics appeared to be able to explain all physical
phenomena. The physical universe was regarded as containing matter that can be classi-
fied into two categories — particles and waves. The properties of particles and waves are
very different, and we have no problem to distinguish a particle from a wave. In Fig. 1.1,
we show a summary of classical concepts on particles and waves.


 

 

 

 

 

 

 

 

 

 

 

 

 

 

 
  
  
 
  
 
  
 
  
 
  
 
  
 
  
 
  
 
  
 
  
 
  
 
  
 
  
 
 

   
   
   
   
   
   
   
   
   
   
   
   
   
   


 
 
 
 
 
 
 
 
 

Physical

 
 
 
 
 
 
 

Observables

 
 
 
 
 
 
 
 
 
 
 









 
 





 


 




   




(mass, momentum, position, ...)   (wavelength, amplitude, phase
PARTICLES WAVES








  





 
  
velocity...)








satisfy: Newton’s equation

  satisfy: wave equation





 
   solutions: How do waves diffract?







solutions: How do
How do particles respond 
trajectories look?

 
 





 
 


 


after collisions?



  
How do waves propagate



 

  
 
 
 
 
 
 
 
                              
in media?
 
 
 
 
 
 
 
 
 
 
 
 
Figure 1.1: A summary of classical concepts on particles and waves.

At the turn of the 20th century, a series of remarkable experiments were conducted

1
Chapter 1 Electromagnetic Radiation Behaving as Particles 2

that required new concepts radically different from those of classical physics. One of the
new concepts is that there is no clear distinction between particles and waves. In this
chapter, we consider some early experiments indicating electromagnetic radiation (EM
radiation) — which is regarded as waves in classical physics — has properties that we
normally associate with particles. Let’s start our discussion from a brief review on the
properties of classical particles.

1.1 Classical Particles

When classical physics attempts to explain the behavior of particles, it is answering ques-
tions such as:

• What is the trajectory of a spaceship moving around the earth?


• If a billiard ball hits a wall at a certain angle, what will be the outcome?

Answers of all these questions are given by the Newtonian mechanics which describes the
motion about particles.

The key properties of a classical particle are:

mass m
charge q
position x
momentum p
1 2 p2
kinetic energy mv =
2 2m
potential energy V (x)
d2 x
equation of motion m =F
dt2

where F is the applied force. If the particle is charged, the force can arise from an electric
field and/or a magnetic field. It could also experience mechanical force. Once the force
is known, one can completely describe the particle in the physical world. There are some
interesting points that can be made about particles in classical physics:

• There is no quantization of particle energy. What we mean by this is that the


total energy of the particle is a continuous variable; i. e. the energy can change
Chapter 1 Electromagnetic Radiation Behaving as Particles 3

in infinitesimal steps. For example, the energy of a pendulum can start from zero
(amplitude of vibration is zero) and increased continuously as the amplitude is
increased.

• Physical observables such as momentum, position, energy, etc., can all be defined
or measured with complete certainty. Of course, it is possible that the measuring
devices may have some error which may not allow an accurate measurement. But in
principle, we can make completely precise measurements on any physical observable.

These points seem rather trivial and obvious; however, we now know that they are not
satisfied by particles under all conditions.

1.2 The Electromagnetic Waves

The wave theory of electromagnetic radiation was formalized in the 1800s by James Clerk
Maxwell. He explained electromagnetism with a great success by a set of equations
known as Maxwell equations. His equations predicted that electric and magnetic fields
may exist in vacuum regions where no electric charge or current is present. In particular,
if the fields at one point of space vary with time, then some variation of the fields must
occur at other point of space at some other time. Thus changes in magnetic and electric
fields should propagate through space. The propagation of such disturbance is called
an electromagnetic wave (EM wave), as illustrated in Fig. 1.2. Maxwell’s work also
predicted that the waves should propagate in vacuum at a speed of 3 × 10 8 m/s; this agrees
closely with the speed measured for the propagation of light. Thus Maxwell concluded
that light itself is an EM wave.

Figure 1.2: An electromagnetic plane wave travelling in the z direction.


Chapter 1 Electromagnetic Radiation Behaving as Particles 4

The first experimental proof for the existence of electromagnetic waves was given
by the German physicist Heinrich R. Hertz. He used the electric current pulse in a
discharge across a spark gap to produce electromagnetic waves which propagated into the
surrounding space. Hertz found that the generated waves had both electric and magnetic
fields, travelled at the same speed as light but had a much longer wavelength. He also
demonstrated that these waves have the same properties as light waves — reflection,
refraction, interference and polarization. Hertz’s work convinced other physicists that
electromagnetic waves and light are the same thing.

Note that electromagnetic radiation refers to the energy carried by EM waves, although
this term is often used interchangeably with “electromagnetic waves”. Light is not the
only example of electromagnetic wave. Although all electromagnetic waves have the same
fundamental nature, many features of their interaction with matter depend upon their
frequencies. Figure 1.3 shows the electromagnetic spectrum — the distribution of all EM
waves classified according to their frequencies or wavelengths. The visible light spans only
a narrow frequency interval.

Figure 1.3: The electromagnetic spectrum.

A characteristic property of all waves (including EM waves) is that they obey the
principle of superposition:

When two or more waves of the same nature travel past a point at the same time, the
instantaneous amplitude there is the sum of the instantaneous amplitudes of the
individual waves.

This characteristic waves property leads to many wave phenomena such as the interference
of light. When two or more trains of light waves meet in a region, they interfere to produce
a new wave whose instantaneous amplitude is the sum of those of the original waves. The
interference of light was first demonstrated in 1801 by Thomas Young. In this experiment,
Chapter 1 Electromagnetic Radiation Behaving as Particles 5

monochromatic light from a single source is incident on a pair of slits (see Fig. 1.4). From
each slit, secondary waves spread out as if they are originated from that slit. This is an
example of diffraction which is a characteristic wave phenomenon. Owing to interference,
the screen is not evenly lit but shows a pattern of alternate bright and dark lines.

Figure 1.4: (a) Young’s double-slit experiment. (b) The interference fringes observed on
the screen.

Interference and diffraction are found only in waves — the particles we are familiar
with do not behave in those ways. Thus Young’s experiment is a proof that light is a kind
of wave. Maxwell’s theory further tells us what kind of wave it is: electromagnetic. At
the end of nineteenth century, the nature of light seemed to be settled forever.

1.3 Blackbody Radiation

The first indication that the classical wave picture of electromagnetic radiation does not
work well followed from the failure of wave theory to explain the observed spectrum of
thermal radiation — the type of electromagnetic radiation emitted by an object merely
because of their temperature.

When we put our hand near a lightbulb, we feel warm. It is due to the electromagnetic
energy emitted from the lightbulb. The emitted radiation is distributed continuously over
all frequencies. In fact, this emission occurs for any object at temperature greater than
absolute zero. If a body is in thermal equilibrium with its surroundings, and therefore is
at constant temperature, it must emit and absorb energy at the same rate. The radiation
emitted or absorbed under such circumstance is known as thermal radiation.
Chapter 1 Electromagnetic Radiation Behaving as Particles 6

The spectrum of the thermal radiation emitted from an object depends strongly on its
temperature. For example, a hot piece of metal gives off visible light whose color varies
with the temperature of the metal, going from red to yellow to white as it becomes hotter
and hotter. In fact, our eyes only respond to the color corresponding to the most intense
emission although other frequencies present in the spectrum as well.

Figure 1.5: Measurement of the spectrum of thermal radiation. A device such as a prism
is used to separate the wavelengths emitted by the object.

Fig. 1.5 shows a typical experimental setup for measuring the spectrum of the thermal
radiation emitted by an object maintained at a constant temperature. In this experiment,
the intensity dI of the emitted electromagnetic radiation in a narrow frequency range
between f and f + df is measured. Thus we can determine the spectral energy density
dU/df , the electromagnetic energy dU per frequency range df in volume V , at any desired
frequency f since
c 1 dU
dI = × .
4 V df
The plot of the observed spectral energy density versus the frequency shows the spectrum
of the thermal radiation emitted by the object.

In general, the detailed form of the spectrum of the thermal radiation emitted by
a hot body depends not only on the temperature but also on other factors such as the
Chapter 1 Electromagnetic Radiation Behaving as Particles 7

composition of the body and its surface roughness. However, experiments show that
there is one class of hot bodies emits thermal spectra with a universal character. These
are called blackbodies which are ideal objects that absorb all radiation incident upon
them regardless of frequency. In the laboratory, we can approximate a blackbody by a
cavity with a small hole cutting in one wall kept at a constant temperature (see Fig. 1.6).
It is the hole, and not the box itself, that acts like a blackbody.

Figure 1.6: A cavity filled with electromagnetic radiation. The hole in the wall of the
cavity represents an ideal blackbody.

Thermal radiation absorbed or emitted by a blackbody is called blackbody radia-


tion. The universal properties of blackbody radiation makes it a particular interest for
physicists. Fig. 1.7 shows the observed spectrum of the blackbody radiation for various
temperatures.

Figure 1.7: Spectrum of blackbody radiation. Note that the peak occurs at different
positions for different temperatures.
Chapter 1 Electromagnetic Radiation Behaving as Particles 8

From the spectrum, we can observe two interesting properties:

1. The total intensity radiated over all wavelengths, which is proportional to the area
under the curve of the spectral energy density dU/df , increases as the temperature is
increased. From careful measurement, it was found that the total intensity increases
as the fourth power of the temperature:
Z
c 1 dU
I= × df = σT 4 (1.1)
4 V df
where the Stefan-Boltzmann constant

σ = 5.671 × 10−8 W/(m2 · K4 )

The above equation is called the Stefan-Boltzmann law.

2. The frequency at which the spectral energy density dU/df reaches its maximum
value increases as the temperature is increased. Experimental results showed that
the wavelength λmax at which the thermal emission of electromagnetic energy from a
body of temperature T is maximum obeys:

λmax T = 2.898 × 10−3 m · K (1.2)

This results is known as Wien’s displacement law.

Classical wave theory predicted that the spectrum of blackbody radiation is described
by the so-called Rayleigh-Jeans formula1 :
dU 8πV
= kB T × 3 f 2 (1.3)
df c
where kB is the Boltzmann’s constant whose value equals 1.381 × 10−23 J/K. As
shown in Fig. 1.8, the spectral energy density calculated with this formula approaches
the experimental data at small frequencies but fails miserably at high frequencies. The
failure of the Rayleigh-Jeans formula at high frequencies is known as the ultraviolet
catastrophe and represents a serious problem for classical physics.

The new physics giving the correct interpretation of thermal radiation was proposed
by the German physicist Max Planck in 1900. He suggested that an oscillating atom can
absorb or re-emit energy only in discrete bundles called quanta. It implies that the total
energy E of an oscillating atom must be an integer multiples of a certain quantity , i. e.

E = n n = 1, 2, 3, . . . (1.4)
1
For the derivation of the Rayleigh-Jeans formula, read R. Harris, Modern Physics, Appendix C
Chapter 1 Electromagnetic Radiation Behaving as Particles 9

Figure 1.8: The classical Rayleigh-Jeans formula is failed to fit the observed spectral
energy density of the blackbody radiation.

where n is the number of quanta. Moreover, the energy of each quanta  is related to the
frequency f of the radiation by
 = hf (1.5)
where h is a proportionality constant called the Planck’s constant. Based on this as-
sumption, Planck found that the spectral energy density is given by the Planck’s formula2 :
dU hf 8πV
= hf /k T × 3 f2 (1.6)
df e B −1 c
The Planck’s formula fits the experimental data very well for both small and high fre-
quencies.

As we know that the Rayleigh-Jeans law works well at small frequencies, we expect
(1.6) reduces to (1.3) in the limit of f → 0. Then we can obtain the relationship between
the Stefan-Boltzmann constant and the Planck’s constant:
2π 5 kB
4
σ= (1.7)
15c2 h3
Using the value of the Stefan-Boltzmann constant obtained from the data of the spectral
energy density available at 1900, Planck determined the value of the Planck’s constant h
to be:
h = 6.56 × 10−34 J · s
And the presently accepted value of h is

h = 6.626 × 10−34 J · s
2
For the derivation of the Planck’s formula, read R. Harris, Modern Physics, Appendix C
Chapter 1 Electromagnetic Radiation Behaving as Particles 10

1.4 The Photoelectric Effect

When a metal surface is illuminated by light, electrons may be emitted from the surface.
This phenomenon is known as the photoelectric effect and the emitted electrons are
called photoelectrons. It was discovered by Heinrich Hertz in 1887 during his research
of electromagnetic radiation.

Figure 1.9: Apparatus for observing the photoelectric effect.

A sample experimental setupt for observing the photoelectric effect is illustrated in


Fig. 1.9. The experiment must be done in an evacuated tube, so that the electrons do
not lose energy in collisions with the air molecules. In the evacuated tube, electrons can
be released from a metal electrode by shining it with light. If the released electrons have
enough energy to overcome the opposing potential difference between the two electrodes,
they can reach the other electrode. Then the rate of the electron emission is measured as
an electric current i by an ammeter in the external circuit. And the stopping potential
Vs is determined by increasing the potential difference between the two electrodes until
the ammeter reading drops to zero. The maximum kinetic energy of the electrons is then
given by:
KEmax = eVs (1.8)
where e is the magnitude of the electric charge of an electron.

In the classical picture, an electron absorbs energy from the light wave incident on
it until the binding energy of the electron to the metal is exceeded, at which point the
electron is released. The minimum amount of energy required to remove an electron from
Chapter 1 Electromagnetic Radiation Behaving as Particles 11

the metal is called the work function φ. Table 1.1 lists the values of the work function
of various metals.

Material Work function φ (eV)


Cesium 1.9
Potassium 2.2
Sodium 2.3
Magnesium 3.7
Zinc 4.3
Chormium 4.4
Tungsten 4.5

Table 1.1: The work function of various metals

Figure 1.10: The photoelectric current i as a function of the potential difference V for
two different light intensities.

Figure 1.10 shows the experimental results of the photoelectric effect. How are the
predictions of the classical theory compared with the experimental results?

1. The maximum kinetic energy (determined from the stopping potential) is totally inde-
pendent of the intensity of the light source. Experimental results show that doubling
the intensity of the source leaves the stopping potential unchanged. It disagrees with
the wave theory which predicts that the more intense the light, the greater the kinetic
energy of the released electrons.
Chapter 1 Electromagnetic Radiation Behaving as Particles 12

2. The photoelectric effect does not occur at all if the frequency of the light source is
below a certain value. This value, which is characteristic of the kind of metal used
in the experiment, is called the cutoff frequency fc . According to the wave theory,
as long as the light is intense enough to release the electrons, the photoelectric effect
should occur no matter what the frequency or wavelength is.

3. The first photoelectron are emitted virtually instantaneously (within 10 −9 s) after the
light source is turned on. In the wave theory, the energy of the wave is uniformly
distributed over the wavefront. A period of time should elapse before an individual
electron absorb enough energy from the light wave to leave the metal, which is not
consistent with experimental results.

All these experimental results suggest the failure of the wave theory to account for the
photoelectric effect.

A successful explanation of the photoelectric effect was developed in 1905 by Albert


Einstein. Based on Planck’s ideas, Einstein proposed that the energy of light is concen-
trated in localized bundles called photons rather than distributed continuously over the
wavefront. The energy of a photon associated with light of frequency f is equal to

E = hf (1.9)

where h is the Planck’s constant.

In Einstein interpretation, a photoelectron is released as a result of an encounter with


a single photon. Each photon at frequency f has the same energy hf , thus changing
the intensity of the light will only change the number of photon as well as the number
of released electron per second. When a photon falls on the metal, the entire energy of
the photon is delivered instantaneously to a single electron. If the frequency f is large
enough for the photon energy to be greater than the work function φ of the metal, the
photoelectron will be released instantaneously. However, if the frequency is so small that
the photon energy is smaller than the work function, the photoelectric effect will not occur.
The photon theory appears to explain all the observed features of the photoelectric effect.

According to Einstein, the photoelectric effect in a given metal should obey the equa-
tion:
KEmax = hf − φ (1.10)
The maximum kinetic energy corresponds to the release of the least tightly bound electron.
Some electrons may emerge with a smaller kinetic energies as they may lose energy through
Chapter 1 Electromagnetic Radiation Behaving as Particles 13

interactions with other electrons in the metal. On the other hand, a photon that supplies
an energy equal to φ corresponds to light with frequency equal to the cutoff frequency f c .
At this frequency, there is no excess energy converting to kinetic energy of the electron.
So the cutoff frequency is given by
φ
fc = (1.11)
h
and the corresponding cutoff wavelength is
c hc
λc = = (1.12)
fc φ

Figure 1.11: Millikan’s results for the photoelectric effect in sodium.

Robert Millikan carried out detailed test of the photon theory in 1915. From the slope
of the line of the stopping potential Vs against the frequency f (see Fig. 1.11), Millikan
obtained a value of the Planck’s constant:

h = 6.57 × 10−34 J · s

which is quite close to the presently accepted value.

1.5 The Production of X-ray

X-rays were discovered in 1895 by the German physicist Wilhelm Roentgen (1845-1923).
He found that a highly penetrating radiation of unknown nature (X-rays) is produced
when fast moving electrons strike on matter. Not long after this discovery, it was revealed
that X-rays are electromagnetic radiation of very short wavelengths.
Chapter 1 Electromagnetic Radiation Behaving as Particles 14

Fig. 1.12 shows an X-ray tube in which X-rays are produced by bombarding a metal
target with high-speed electrons. In the X-ray tube, the electrons undergo accelera-
tion when they are suddenly brought to rest in the collision with the target. So they
are expected to emit electromagnetic waves according to classical electromagnetic the-
ory. The radiation produced by the braking electrons under such circumstances is called
the bremsstrahlung radiation, and the process producing such radiation is called the
bremsstrahlung process.

Figure 1.12: An X-ray tube.

Fig. 1.13 shows the spectrum of the X-rays produced by the bremsstrahlung process
when electrons of kinetic energy 25 keV strike a molybdenum target. From the spectrum,
we can see that the X-rays produced vary in wavelength, but none has a wavelength
shorter than a certain value λc . The existence of this minimum wavelength is difficult to
explain classically but is easily understood in terms of the photon theory.

Figure 1.13: The spectrum of the X-rays produced by the bremsstrahlung process when
25 keV electrons strike a molybdenum target.

According to the photon theory, X-rays consists of photons as they are electromag-
Chapter 1 Electromagnetic Radiation Behaving as Particles 15

netic wave. In the bremsstrahlung process, an electron may suffer numerous collisions
with various atoms in the target, producing a photon each time (see Fig. 1.14). Each
photon represents the conversion of part of the electron’s kinetic energy into radiation.
Thus the photon energy hf is equal to the decrease in kinetic energy KE − KE0 of the
electron. Clearly, the highest frequency photon (i.e. the lowest wavelength photon) that
can be produced corresponds to the complete conversion of the electron’s kinetic energy
in a single process. When the tube is operated at potential difference V , the electrons
strike the target with kinetic energy eV . (Here we have neglected a small correction for
the work functions of the cathode and target because they are much smaller than eV .)
Therefore,
hc
eV =
λc
hc 1240 eV · nm
⇒ λc = = (1.13)
eV V
which agrees with the experimental result. Hence, we may regard the bremsstrahlung
process as an inverse photoelectric effect in which electron KE is transformed to photon
energy. Note that this process will occur only if there is an atom nearby to take away the
recoil momentum of the collision.

Figure 1.14: A schematic diagram of the bremsstrahlung process.

1.6 The Compton Effect

When electromagnetic radiation passes through matter, it interacts with atoms by means
of scattering from loosely bound, nearly free electrons. Parts of the energy of the radia-
tion is given to the electron, which is released from the atom; the remainding energy is
re-radiated as electromagnetic radiation. According to the wave picture, the scattered ra-
diation is less energetic than the incident radiation but has the same wavelength. However,
experiments showed that X-ray increases in wavelength after the scattering, contradicting
Chapter 1 Electromagnetic Radiation Behaving as Particles 16

with the prediction of classical theory. This phenomenon is known as the Compton
effect which was first observed by Arthur H. Compton (1892-1962) in the early 1920s.

Compton explained the Compton effect using the photon concept. He treated the
X-rays as a collection of photons with the following properties:

1. According to special relativity, an object with zero mass should have momentum p
related to its energy E by E = pc

2. According to classical EM wave theory, EM waves do carry momentum.

According to photon theory, the interaction of X-rays and electrons is simply a collec-
tion of separate two-particle collisions between a single photon and a stationary electron.
Fig. 1.15 shows the change in momentum and energy during the collision. Initially, the
photon has energy E given by
E = hf = hc/λ (1.14)

and linear momentum p where


p = E/c = h/λ (1.15)

since it is a massless particle.

Figure 1.15: Momentum and energy in the Compton scattering.


Chapter 1 Electromagnetic Radiation Behaving as Particles 17

The electron is initially at rest and has rest energy me c2 . After the scattering, the
photon has energy E 0 and momentum p0 (= E 0 /c), and moves in a direction at an angle
θ with respect to its original direction. The electron with total energy Ee (= γu me u) and
momentum pe (= γu me c2 ) moves in a direction at an angle φ with respect to the initial
photon. The usual conditions of conservation of energy and momentum are then applied:

Einitial = Efinal ⇒ E + me c2 = E 0 + Ee (1.16)


(px )initial = (px )final ⇒ p = p0 cos θ + pe cos φ (1.17)
(py )initial = (py )final ⇒ 0 = p0 sin θ − pe sin φ (1.18)

Combining the momentum equations, we obtain

p2e = p2 − 2pp0 cos θ + p02 (1.19)

The relativistic relationship between the energy and momentum of the electron is

Ee2 = p2e c2 + m2e c4 (1.20)

Substituting previous expressions of pe and Ee into the above equation, we have

(E + me c2 − E 0 )2 = c2 (p2 − 2pp0 cos θ + p02 ) + m2e c4


1 1 (1 − cos θ)
⇒ 0
− = (1.21)
E E me c 2
This equation can also be written as
h
λ0 − λ = (1 − cos θ) (1.22)
me c
where λ is the wavelength of the incident photon and λ0 is the wavelength of the scattered
photon. This equation was derived by Compton in about 1923, and the quantity h/me c =
0.002426 nm is known as the Compton wavelength of the electron.

Equations (1.21) and (1.22) gives us the change in energy or wavelength of the photon
as a function of the scattering angle θ. Since the quantity on the right-hand side of these
equations is never negative, thus λ0 is always greater than λ for θ 6= 0◦ — the scattered
photon has a longer wavelength (i. e. a smaller energy) than the incident photon. The
change in wavelength ranges from 0 at θ = 0◦ to twice the Compton wavelength at
θ = 180◦ .

Using apparatus as shown in Fig. 1.16, Compton carefully measured the wavelength
of the scattered X-ray. His experimental results confirmed the dependence of λ0 on the
Chapter 1 Electromagnetic Radiation Behaving as Particles 18

Figure 1.16: Schematic diagram of Compton’s apparatus.

scattering angle θ (see Fig. 1.17). He also found that the wavelength shift agrees with
the prediction of (1.22), proving the validity of his formula. Compton’s study on the
X-ray scattering gives us a direct experimental confirmation that X-ray photons behave
like particles with momentum p = hf /c.

Figure 1.17: Compton’s original results for X-ray scattering.

1.7 Pair Production

Pair production is a process that may occur when photons encounter atoms. In this
process, the photon loses all its energy and then two particles are created: an electron
Chapter 1 Electromagnetic Radiation Behaving as Particles 19

and a positron, as shown in Fig. 1.18. (A positron is a positively charged particle with
the same mass and same magnitude of charge as electron.) It is an example of creation
of rest energy in which the photon energy is converted into the relativistic total energies
E+ and E− of the positron and electron.

hf = E+ + E− = (me c2 + K+ ) + (me c2 + K− ) (1.23)

Figure 1.18: In pair production, a gamma-ray photon becomes an electron and a positron
which curve in different ways in a magnetic field.

Since the kinetic energies K+ and K− are always positive, the photon must have
an energy of at least 2me c2 = 1.02 MeV for this process to occur. This process, like
bremsstrahlung, will not occur unless there is an atom nearby to carry the necessary
recoil momentum.

The reverse process,


electron + positron → photon

also occurs; this process is known as electron-positron annihilation and can occur for
free electrons and positrons as long as at least two photons are created. In this process,
the electrons and positrons disappear and are replaced by two photons. Conservation of
energy requires that, if E1 and E2 are the photon energies,

(me c2 + K+ ) + (me c2 + K− ) = E1 + E2 (1.24)


Chapter 1 Electromagnetic Radiation Behaving as Particles 20

Usually K+ and K− are negligibly small, we can assume the positron and electron to be
essentially at rest. Momentum conservation then requires the two photons to have equal
energies of me c2 and to move in exactly opposite directions.

1.8 Dual Nature of Electromagnetic Radiation

We have established firm experimental basis for the particle nature of electromagnetic
radiation. Experiments revealed that electromagnetic radiation behaving as a collection
of particles known as photons which have the following properties:

• like an electromagnetic wave, photons move at the speed of light in vacuum;

• they have zero mass and zero rest energy

• they carry energy and momentum, which are related to the frequency and wave-
length of the electromagnetic wave by E = hf and p = E/c = h/λ;

• they can be created or destroyed when radiation is emitted or absorbed;

• they can have particle-like collisions with other particles such as electrons.

This list of properties shows us a particle picture for electromagnetic radiation. It is


very different from the wave description of electromagnetic radiation although they both
have very strong experimental support. Obviously, the wave and particle explanations of
electromagnetic radiation are not consistent with each other.

This dilemma of wave-particle duality cannot be resolved with a simple explanation.


The best we can do is to say that neither the wave or particle picture is correct all the time,
and both are needed for a complete description of physical phenomena. In other words, the
two pictures are complementary with each other. The “true nature” of electromagnetic
radiation consists of both wave and particle characters, even though nothing in everyday
life to help us visualize that.
Chapter 2

Matter Behaving as Waves

In the last chapter, we have discussed some early experiments revealing the particle-like
behavior of electromagnetic radiation, which is regarded as waves in classical physics.
Here, we study the experimental evidence indicating that material particles, such as elec-
trons, possess wave-like properties. This wave-particle duality evolved to become part
of a new theory called quantum physics. Quantum physics is one of the major branch of
modern physics that provides a consistent description of matter on the microscopic scale.
Note that quantum physics is correct for both small and large things, and converges to
the special case of classical physics in the limit of large things. Before moving on, let’s
take a brief review on the properties of classical waves first.

2.1 Classical Waves

A wave is a disturbance that propagates in a periodically repeating fashion, often trans-


ferring energy. In some kind of waves, particle motion may be involved. Examples are
waves on the ocean surface, where water molecules move up and down to create the wave.
Some waves are created in a different way. For example, oscillating electric and magnetic
fields produce electromagnetic waves.

Properties of a wave are described by its amplitude and wavelength (or frequency).
We can use an appropriate differential equation to describe a wave. For one-dimensional
case, a simple equation describing a wave is:

∂ 2 y(x, t) 1 ∂ 2 y(x, t)
= (2.1)
∂x2 v 2 ∂t2

21
Chapter 2 Matter Behaving as Waves 22

where v is the travelling velocity of the wave. A basic solution of this equation is a simple
plane waves travelling along +x direction with the form:
ω
y(x, t) = A sin(kx − ωt) where =v (2.2)
k

Classical waves have some unique features that classical particles do not have:

• The wavelength of individual wave is discrete. The wave equation discussed above
is a second-order differential equation. To find the solution, we need to impose
boundary conditions on the equation. The solutions for the allowed wavelengths
(frequencies) come from solving the wave equation with appropriate boundary con-
ditions. Not all possible frequencies may be allowed for a given problem.

• We cannot precisely define the location of a wave and its wavelength simultaneously
at any time. We know from experience and classical physics that it is impossible
to precisely determine these two quantities at the same time. If we try to create a
“wave-packet” highly localized in space, we lose the knowledge of the wave’s wave-
length. However, if we try to create a “plane wave” with a well-defined wavelength,
we lose the knowledge regarding the spatial position of the wave.

2.2 Matter Waves

In the last chapter, we have studied the dual particle-wave nature of electromagnetic
radiation. Do material objects also have such dual particle-wave nature? In a bold
and daring hypothesis in his doctoral dissertation, Louis de Broglie (1892-1987) chose
the latter alternative. Lacking any experimental evidence to support his hypothesis, de
Broglie suggested that associated with any material particle moving with momentum p,
there is a wave of wavelength λ, related to p according to

λ = h/p (2.3)

Such wave was called a matter wave by de Broglie. The wavelength λ computed ac-
cording to the above equation is called the de Broglie wavelength.

Just like any other wave, a matter wave should have a wavelength, a frequency, a
speed as well as an amplitude that varies with position and time. The generic term for
the function giving the amplitude of a wave is wave function. Because what actually
Chapter 2 Matter Behaving as Waves 23

oscillates depends on the kind of wave, we use different symbols for the wave functions
of different kinds of waves. For a matter wave, the wave function is usually denoted as
Ψ(x, t).

What does the amplitude of matter waves measure? The square of the amplitude
of the matter wave associated with a particle in a given region is proportional to the
probability for finding the particle in that region, just like the case that the amplitude of
the electric field determines the probability of observing a photon. That is to say,
!2
probability of finding amplitude of the associated

the particle in a region matter wave in that region

The de Broglie wavelength of a particle shows itself when a wave-type experiment (such
as diffraction) is performed on it. Of course, the outcome of a wave-type experiment
depends on the wavelength. Why was the de Broglie wavelength not directly observed
before de Broglie’s time? The de Broglie wavelengths of ordinary objects are very small
since Planck’s constant h is so small. There is no experiment that can be done to reveal
the wave nature of macroscopic objects. And the wave behavior will be observable only for
particles of atomic or nuclear size. The difficulties for doing wave-type experiments with
atomic or nuclear size particles were not solved before the time of de Broglie’s hypothesis.

Example 2-1

Compute the de Broglie wavelength of the followings:

(a) A 100-kg automobile travelling at 100 m/s.


(b) A smoke particle of mass 10−9 g moving at 1 cm/s.
(c) An electron with a kinetic energy of 1 eV.

Solution:

(a) Using the classical relation between velocity and momentum,

h h 6.626 × 10−34 J · s
λ= = = 2
= 6.63 × 10−38 m
p mv (10 kg)(100 m/s)

(b) As in part (a),

h h 6.626 × 10−34 J · s
λ= = = = 6.63 × 10−20 m
p mv (10−12 kg)(10−2 m/s)
Chapter 2 Matter Behaving as Waves 24

(c) The rest energy (mc2 ) of an electron is 5.11 × 105 eV. Since the kinetic energy (1 eV)
is much less than the rest energy, we can use non-relativistic kinematics.
√ p
p = 2mK = (2)(9.109 × 10−31 kg)(1 eV)(1.602 × 10−19 J/eV)
= 5.40 × 10−25 kg · m/s

Then,
h 6.626 × 10−34 J · s
λ= = = 1.23 × 10−9 m
p 5.40 × 10−25 kg · m/s

2.3 Particle Diffraction and Interference

To show the wave nature of material particles, we must use particles in the atomic realm.
The indications of wave behavior come from interference and diffraction experiments of
these tiny particles.

In electron diffraction, an accelerated electrons beam strikes a crystal in exactly


the same way as the X-ray beam in X-ray diffraction3 . The scattered electron beam from
the crystal atoms is then photographed. As depicted in Fig. 2.1, the similarity between
the electron diffraction and X-ray diffraction pattern strongly suggests that the electrons
are behaving as waves.

Figure 2.1: (Left) The electron diffraction pattern for aluminum manganese alloy. (Right)
The X-ray diffraction pattern for NaCl crystal.

A direct comparison of the “rings” produced in scattering by polycrystalline materials


is shown in Fig. 2.2. The similarity between the electron scattering and X-ray scattering
3
Read appendix A for the details of X-ray diffraction
Chapter 2 Matter Behaving as Waves 25

is striking, and it is also a strong evidence for the similarity in the wave behavior of
electrons and X-rays. Such experiment were first done in 1927 by George P. Thomson
(1892-1975), who shared the 1937 Nobel prize for this work.

Figure 2.2: Comparison of X-ray diffraction and electron diffraction. The upper half of
the figure shows the result of scattering of 0.071 nm X-rays by an aluminum foil, and the
lower half shows the result of scattering of 600 eV electrons by aluminum.

The first experimental confirmation of the wave nature of electrons was given by
Clinton Davisson (1881-1958) and Lester Germer (1896-1971) when they investigated the
diffraction of electron beams from the surface of nickel crystals in 1926. Figure 2.3 shows
the schematic view of their apparatus. Electrons leave the heated filament F and are
accelerated by the voltage V . The beam strikes a crystal and the scattered beam is
detected at an angle φ relative to the incident beam by a detector.

Fig. 2.3 also shows the results of one of the experiments performed by Davisson and
Germer. When the accelerating voltage is set at 54 V, there is an intense reflection of the
beam at the angle of φ = 50◦ . Let us see how these results give quantitative confirmation
of the de Broglie wavelength.
Chapter 2 Matter Behaving as Waves 26

Figure 2.3: (Left) Apparatus used by Davisson and Germer to study electron diffraction.
(Right) Their results for V = 54 V. Each point on the plot represents the relative intensity
when the detector is located at the angle φ measured from the vertical axis.

If we assume that each of the atoms of the crystal can act as a scatterer, then the
scattered electron waves can interfere, and we have a crystal diffraction grating for the
electrons as illustrated in Fig. 2.4. Because the electrons were of low energy, they did not
penetrate far into the crystal. And it is sufficient to consider the diffraction take place on
the plane of atoms at the surface. The situation is entirely similar to the diffraction of
light using a diffraction grating with spacing equal to the atomic spacing d of the crystal.
So the maxima of the diffraction occur at angle φ such that

d sin φ = nλ (2.4)

where n(= 1, 2, 3, . . .) is the order number of the maximum.

Figure 2.4: The crystal surface acts like a diffraction grating with spacing d.
Chapter 2 Matter Behaving as Waves 27

The peak at φ = 50◦ must be the first-order peak (n = 1), because no peak was
observed at smaller angle. If this is indeed an interference maximum, the corresponding
wavelength is
λ = d sin φ = (0.215 nm)(sin 50◦ ) = 0.165 nm

as d = 0.215 nm for the nickel crystal. We can compare this value with the de Broglie
wavelength of the electrons. An electron accelerated through a potential difference of
54 V has a kinetic energy of 54 eV and therefore a momentum of
p 1p 1
p= 2m(KE) = 2mc2 (KE) = (7430 eV)
c c
So the de Broglie wavelength is
h hc 1240 eV · nm
λ= = = = 0.167 nm
p pc 7430 eV
using hc = 1240 eV · nm. This is in excellent agreement with the value found for the
diffraction maximum, and provides strong evidence in favor of de Broglie theory. For this
experimental work, Davisson shared the 1937 Nobel prize with G. P. Thomson.

The wave nature of particles is not exclusive to electrons; any particle with momentum
p has de Broglie wavelength h/p. For example, diffraction of neutrons by a salt crystal
produces the same characteristic patterns as the diffraction of electrons or X-rays (see
Fig. 2.5).

Figure 2.5: Diffraction of neutrons by a NaCl crystal.

The definitive evidence for the wave nature of light was deduced from the double-slit
experiment performed by Thomas Young in 1801. In principle, it should be possible to do
Chapter 2 Matter Behaving as Waves 28

double-slit experiments with particles and thereby directly observe their wavelike behavior.
However, the difficulties of producing double slits for particles were not solved until recent
years. The first double-slit experiment with electrons was done in 1961. Figure 2.6 shows
the diagram of the apparatus with a photograph of the resulting intensity pattern. The
similarity with the double-slit pattern of light (Figure 1.4) is striking.

Figure 2.6: (Left) Double-slit apparatus for electrons. Electrons from the filament are ac-
celerated through 50 kV and pass through the double slit. (Right) Double-slit interference
pattern produced by electrons.

Figure 2.7: Double-slit apparatus for neutrons.

A similar experiment can be done for neutrons using apparatus as shown in Fig. 2.7.
Thermal neutrons with average kinetic energy ≈ kT ≈ 0.025 eV from a nuclear reactor are
incident on a crystal. By scattering with the crystal through a particular angle, neutrons
of specific wavelength are selected. After passing through the double slit, the neutrons
are counted by the scanning slit assembly, which moves laterally. Figure 2.8 shows the
resulting pattern of intensity maxima and minima, which indicates that interference has
occurred and neutrons have wave nature. We can calculate the wavelength of the neutrons
Chapter 2 Matter Behaving as Waves 29

by estimating the spacing between adjacent maxima from the observed pattern. The result
was found to be agree very well with the de Broglie wavelength of the neutron beam.

Figure 2.8: Intensity pattern observed for double-slit interference of neutrons (with KE
= 0.00024 eV). [Source: R. Gahler and A. Zeiling, American Journal of Physics 59, 316
(1991) ]

It is also possible to do a similar experiment with atoms. Figure 2.9 shows the result-
ing intensity pattern for double slit interference of helium atoms. Although the results
are not as dramatic as those for electrons and neutrons, there is a clear evidence of inter-
ference maxima and minima, and the separation of the maxima gives a wavelength that
is consistent with the de Broglie wavelength.

Figure 2.9: Intensity pattern observed for double-slit interference of helium atoms.
[Source: O. Carnal and J. Mlynek, Physical Review Letters 66, 2689 (1991) ]
Chapter 2 Matter Behaving as Waves 30

2.4 The Heisenberg Uncertainty Principle

The wave nature of material particles such as electrons implies inherent uncertainties in
their particle properties. For example, suppose an electron beam strikes on a single slit
as shown in Fig. 2.10. The electron wave would spread out after passing through the
slit, resulting in an uncertainty in the x-component of momentum of electrons detected
beyond the slit, ∆px . That is to say, if we repeat the experiment many times with identical
settings, the momentum detected after passing through the slit still varies over a range
of values. On the other hand, there would be an uncertainty in the position of electrons
existing the slit, ∆x.

Figure 2.10: Single slit diffraction of an electron plane wave.

The inherent uncertainties are the inescapable consequence of matter’s wave nature
that increased precision in the knowledge of position implies decreased precision in the
knowledge of wavelength and vice versa. In fact, such property is true for all kinds of
waves. Let’s explain this in details. First, we consider a pure sine wave of wavelength λ0
with the form:
f (x) = A sin k0 x
where the wave number k0 = 2π/λ0 . It is a wave that repeats itself endlessly from −∞ to
+∞, as shown in Fig. 2.11(a). So we don’t know the “position” of the wave. If we want
to use a wave to represent a particle, the wave must be localized, i. e. able to be confined
to a very small region of space. Because of its infinite extent, the pure sine wave is of no
use in localizing particles.
Chapter 2 Matter Behaving as Waves 31

Figure 2.11: Building a single isolated pulse from pure sine waves requires a continuum
of wave numbers.

Now consider what happens when we add two other waves of slightly different wave-
lengths (i. e. different wave numbers) and properly chosen amplitudes to our original wave.
As shown in Fig. 2.11(b), the resulted pattern still repeats itself endlessly from −∞ to
+∞, but now we know a bit more about the “position” of the wave. However, the wave-
length is no longer precisely determined since we have added three different wavelengths
together. That is to say, our knowledge of the “position” of the wave has improved at the
expense of our knowledge of its wavelength.

If we continued to add waves of different wavelengths with properly chosen amplitudes,


Chapter 2 Matter Behaving as Waves 32

we could eventually reach a situation similar to that illustrated in Fig. 2.11(e). It has no
amplitude outside a rather narrow region of space ∆x. To achieve this situation, we must
add together a large number of waves of slightly different wave numbers k. The wave thus
represents a range of wave numbers (i. e. range of wavelengths) that we denote as ∆k.
When we have a single sine wave, ∆k is zero and ∆x is infinite. As we increase ∆k, ∆x
would decrease. Fourier analysis shows that the mathematical relationship between ∆x
and ∆k is
1
∆x∆k ≥ (2.5)
2
Since we have not yet defined ∆x and ∆k precisely, they should be regarded as estimates.
Equation (2.5) implies that the position can be determined only at the expense of our
knowledge of its wavelength for any type of wave.

Because of inherent uncertainties in its particle properties, particles are approximately,


but not completely, localized in a certain region of space. Therefore, we represent a moving
particle by a wave pulse (also known as wave packet), like that shown in Fig. 2.12.
Its amplitude is negligibly small except for a region of space of dimension ∆x. This
corresponds to a particle localized in the region of dimension ∆x. Thus a wave pulse
is formed by the superposition of a large number of waves with varying wavelengths
(i. e. varying wave numbers). It results in a range of momenta ∆px since each momentum
corresponds to a unique de Broglie wavelength. Mathematically, we can show that the
wave pulse and particle move together — whenever the particle goes, its de Broglie wave
pulse move along with it like a shadow. This agrees with our intuition that the matter
wave pulse should move at the same speed as the particle.

Figure 2.12: A wave pulse.

Of course, the uncertainty relationships governing any type of wave also apply to
matter wave pulses. Using the de Broglie wave relationship px = h/λ along with the
expression k = 2π/λ, we find the momentum of the particle
hk
px = = ~k (2.6)

Chapter 2 Matter Behaving as Waves 33

where ~ = h/(2π) = 1.055 × 10−34 J · s. It relates the momentum of a particle to the wave
number of the corresponding matter wave. So the range of the wave numbers contained
in the matter wave pulse is given by

∆k = ∆px /~

Combining with (2.5), we then obtain

~
∆x∆px ≥ (2.7)
2
Similar and independent arguments can be applied in the other directions as necessary.
Then we have
~ ~
∆y∆py ≥ and ∆z∆pz ≥ .
2 2
The energy-frequency relationship for photons is E = hf = ~ω and so ∆E = ~∆ω.
We assume this relationship can be carried over to particles for calculating the uncertainty
in energy. Notice that a width in time ∆t is related to a width in the temporal frequency
∆ω by a mathematical relationship similar to that for ∆x and ∆k, i. e.
1
∆ω∆t ≥ (2.8)
2
Substituting ∆E = ~∆ω into the above equation, we obtain
~
∆E∆t ≥ (2.9)
2
Equation (2.7) and (2.9) are the Heisenberg uncertainty relationships. They are the
mathematical representations of the Heisenberg uncertainty principle, which states:

“It is not possible to make a simultaneous determination of the position and the
momentum of a particle with unlimited precision.”,

and

“It is not possible to make a simultaneous determination of the energy and the time
coordinate of a particle with unlimited precision.”

These relationships give an estimate of the minimum uncertainty that can result from
any experiment. We may, for other reasons, do more worse than (2.7) and (2.9), but we
cannot do better no matter how well designed our measuring apparatus might be.
Chapter 2 Matter Behaving as Waves 34

These relationships have a profound impact on our view of nature. Equation (2.7) and
(2.9) say that nature imposes a limit on the accuracy with which we can do experiments.
The uncertainty principle represents a sharp break with the ideas of classical physics
that, with enough skill and ingenuity, it is possible to simultaneously measure a particle’s
position and momentum to any desired degree of accuracy.

Example 2-2

An electron moves in the x direction with a speed of 3.6 × 106 m/s. We can measure its
speed to a precision of 1 %.

(a) With what precision can we measure its position?


(b) What can we say about its motion in the y direction?

Solution:

(a) The electron’s momentum is

px = mvx = (9.109 × 10−31 kg)(3.6 × 106 m/s) = 3.28 × 10−24 kg · m/s

The uncertainty ∆px is 1 % of this value, i. e. 3.28 × 10−26 kg · m/s. The uncertainty
in position is then
~ 1.055 × 10−34 J · s
∆x = = = 1.61 nm
2∆px 3.28 × 10−26 kg · m/s
which is roughly 10 times of atomic diameter.
(b) If the electron is moving in the x direction, then we know its speed in the y direction
precisely; thus ∆py = 0. The uncertainty relationship ∆y∆py ≥ ~/2 then requires
∆y to be infinite. As a result, we know nothing at all about the y coordinate of the
electron.

2.5 Is it a Wave or a Particle?

This may be the most perplexing question for the student studying quantum mechanics.
The simplest answer is that “it”, which is either an electromagnetic radiation or a material
particle, has no predetermined nature. The observation itself — whether the experimenter
Chapter 2 Matter Behaving as Waves 35

bounces light off it, places something in its path, or interacts with it in any way —
determines whether “it” will exhibit a wave or a particle nature.

Whether an object will exhibit wave or particle behavior depends on how its wave-
length λ compares to the “relevant dimension” of the experimental apparatus D. If
λ  D, the object will exhibit a particle behavior; and if λ & D, it will exhibit a wave
behavior, as illustrated in Fig. 2.13.

Figure 2.13: An “experiment” in which a disturbance behaves as a wave or a particle,


depending on the relative size of the wavelength and the relevant dimensions D of the
apparatus.

Can we observe the wave and particle nature of an object simultaneously? Let us
again consider the double-slit experiment of electrons. We can determine which slit that
the electron passed through by surrounding each slit with an electromagnetic loop (Fig-
ure 2.14). In trying to determine which slit the electron went through, we are examining
the particle-like behavior of electron. When we examine the interference pattern, we
are studying the wave-like behavior of the electron. If we perform this experiment, we
would no longer record an interference pattern on the screen. Instead, we would observe
a pattern similar to that as shown in Fig. 2.14, with “hits” in front of each slit but no
interference fringes. The interference pattern would be destroyed no matter what sort of
device is used to determine through which slit the particle has passed.
Chapter 2 Matter Behaving as Waves 36

Figure 2.14: Apparatus to record passage of electrons through slits. No interference


fringes are seen on the screen.

Bohr resolved this dilemma by pointing out that the particle-like and wave-like as-
pects of nature are complementary. Both are needed — they just can’t be observed
simultaneously. Bohr’s principle of complementarity states that:

It is not possible to simultaneously describe physical observables in terms of both


particles and waves.

Physical observables are those quantities such as position, velocity, momentum, and en-
ergy that can be experimentally measured. At any given instance, we can only observe
either the particle properties or wave properties of an object.

2.6 A Two-Slit Experiment

There is no more direct way to see the link between the wave and particle natures of an
object than the double-slit experiment. If the double-slit experiment is performed with
very low intensity levels of light, individual flashes can be seen on the observing screen, as
shown in Fig. 2.15. We can see that the observed pattern looks quite random for the case
of only a small number of flashes detected. In fact, a different pattern may appear on the
Chapter 2 Matter Behaving as Waves 37

Figure 2.15: The double slit interference pattern for light in which the photons are de-
tected one by one. The four pictures show the observed pattern as increasing number of
counts are detected.

screen if we repeat this experiment in identical ways. We cannot make any prediction to
the location where the next photon will be found. But we still know that the probability
of observing a flash (a photon) is proportional to the square of the electric field amplitude.
Although the light is being detected as one particle at a time, its wave nature is still
apparent.

In fact, similar results would be obtained if we repeat the double-slit experiment


for a weak electron source. We would observe the build up of an electron interference
pattern as increasing number of electrons are detected. What determines the probability
of observing an electron? How can we interpret the probability of finding an electron in
the wave picture?

As mentioned before, the amplitude of a matter wave tells us the probability of finding
Chapter 2 Matter Behaving as Waves 38

the associated particle. In fact, the value of the wave function Ψ(x, t) for the matter wave
associated with a moving body at the point x in space at the time t is related to the
likelihood of finding the body there at that moment. The wave function Ψ(x, t) itself
has no direct physical significance, and it can have a complex value. It’s the square of
the modulus of the wave function |Ψ(x, t)|2 , which is known as the probability density,
that determines the probability of finding the particle. If Ψ(x, t) is the wave function
describing a particle, the probability of observing the particle between the positions x
and x + dx at the time t is given by

P (x)dx = Ψ∗ (x, t)Ψ(x, t)dx = |Ψ(x, t)|2 dx (2.10)

where Ψ∗ (x, t) denotes the complex conjugate of Ψ(x, t).

We can link the wave and particle natures of an object together using the concept of
~
probability. For radiation, its propagation is governed by an electric field E(x, t) and the
probability of observing a photon is proportional to hE(x, t)2 i. Similarly, the propaga-
tion of a material particle is governed by a wave function Ψ(x, t) and the probability of
observing the particle is proportional to |Ψ(x, t)|2 .

The probability interpretation of the wave function was first proposed in 1926 by Max
Born (1882-1970) who is one of the founders of the quantum theory. The determination
of the wave function Ψ(x, t) will be discussed in more detail in the next chapter.
Chapter 3

The Schrödinger Equation

In non-relativistic quantum physics, we can learn about the future behavior of a particle
by solving a second-order differential equation4 known as the Schrödinger equation.
It is the fundamental equation of quantum mechanics introduced by Erwin Schrödinger
(1887-1961) in 1926. Like Newton’s law, the Schrödinger equation must be written down
for a certain force acting on the particle. Unlike Newton’s laws, it does not give the
trajectory of the particle. Instead, the solution of the Schrödinger equation gives the wave
function of the particle, which carries information about the particle’s wave-like behavior.
In this chapter, we discuss the foundation of the Schrödinger equation in details.

3.1 The Free-Particle Schrödinger Equation

How do we determine the wave function of a matter wave? In one sense, all types of
waves are the same. For each, there is an underlying wave equation, of which the wave
function must be a solution.

The wave equation obeyed by matter waves is the Schrödinger equation. Here, we
consider only the special case of free-particles. In the absence of external forces, the
Schrödinger equation is:
~2 ∂ 2 Ψ(x, t) ∂Ψ(x, t)
− 2
= i~ (3.1)
2m ∂x ∂t
It is a partial differential equation involving both time and position.

We might hope to derive Equation (3.1) from first principles. However, there is simply
4
Read appendix B for a brief review on differential equations

39
Chapter 3 The Schrödinger Equation 40

no more basic physical principle on which it is built. Its acceptance as a “law” rests on
its ability to give correct prediction, e. g. probabilities of finding particles. Although we
can’t derive it, we will soon argue that at least it has a plausible foundation.

Another problem with the Schrödinger equation is that there is a complex quantity

i = −1 in the equation. It might be incorrectly concluded that a matter wave is not
“real” although it seem to fit perfectly with the fact that the wave function Ψ(x, t) is not
directly observable. In fact, the reason for the i is not that matter waves are unreal, but
that they can’t be represented by a single real function. Like electromgnetic waves, they
have two parts by nature, and a complex function enables us to handle them together. It’s
simply a matter of convenience to represent a matter wave by a single complex function.

However, can we physically interpret a complex wave function? Actually there is no


need to do so as the wave function itself Ψ(x, t) cannot be physically detected. What can
be measured directly is the wave function’s “square” — the probability density |Ψ(x, t)|2
which is a real, non-negative quantity. Experiments verify that the probability of detecting
a particle is proportional to the sum of the squares of the real and imaginary parts of the
wave function:

[ReΨ(x, t)]2 + [ImΨ(x, t)]2 = Ψ∗ (x, t)Ψ(x, t) = |Ψ(x, t)|2

where Ψ(x, t) = ReΨ(x, t) + iImΨ(x, t). Fortunately, we will usually work with just a
single real wave function; and when we say “the square of the wave function” |Ψ(x, t)|2 ,
it is a simple square for such case. But when we use the full complex function, this
expression is understood to mean the product of the function and its complex conjugate.

Let us now reveal the foundation of the free-particle Schrödinger equation by consid-
ering its most basic solution. A free particle is one that is not under the influence of any
force. Thus we postulate that the wave function of a free-particle has the basic form:

Ψ(x, t) = Aei(kx−ωt) ≡ A cos(kx − ωt) + iA sin(kx − ωt) (3.2)

which represents a wave of amplitude A travelling in the positive x direction. To verify


that this is a solution of Equation (3.1), the question is:
~2 ∂ 2 Aei(kx−ωt) ? ∂Aei(kx−ωt)
− = i~
2m ∂x2 ∂t
Taking partial derivatives on both sides, we have
~2
− (ik)2 Aei(kx−ωt) = i~(−iω)Aei(kx−ωt)
2m
~2 k 2
⇒ = ~ω (3.3)
2m
Chapter 3 The Schrödinger Equation 41

The cancellation of the functional dependence on position and time means that function
(3.2) obeys the Schrödinger equation for all values of x and t, provided only that k and ω
are related as in Equation (3.3). We can use a similar approach to show that A cos(kx−ωt)
and A sin(kx − ωt) are not solutions of (3.1) although Aei(kx−ωt) ≡ A cos(kx − ωt) +
iA sin(kx − ωt).

The requirement that (3.3) must hold is the key to see how the Schrödinger wave
equation relates to the classical physics of particles. We hope that it isn’t at odds with
the fundamental wave-particle relationships, p = ~k and E = ~ω. Inserting them in (3.3)
gives:
p2 1
E= = mv 2 (3.4)
2m 2
This merely says that the particle’s kinetic energy must equal its total energy, which is
the classical truth, because a free particle has no potential energy. Thus, the Schrödinger
equation is related to a classical accounting of energy.

Figure 3.1 shows the plot of Aei(kx−ωt) at t = 0, giving some insight into the mathe-
matical nature of a plane wave. The real part Ψ is a cosine and the imaginary part is a
sine. The two parts are out of phase in such a way that their sum is constant — it varies
neither in position or time. This agrees with the calculation of the probability density:

|Ψ(x, t)|2 = Ψ∗ (x, t)Ψ(x, t) = [Ae−i(kx−ωt) ][Ae+i(kx−ωt) ] = A2

That the probability per unit length is constant means that a particle represented by a
plane wave would be equally likely to be found anywhere.

Figure 3.1: A plane matter wave Aei(kx−ωt) plotted at t = 0.


Chapter 3 The Schrödinger Equation 42

3.2 The Schrödinger Equation

In the last section, we discussed the Schrödinger equation in the absence of external
forces (Eqn. (3.2)). Using p = ~k and E = ~ω and noting that p2 /(2m) = (1/2)mv 2 , the
free-particle Schrödinger equation becomes
p2
Ψ(x, t) = EΨ(x, t) ⇒ KEΨ(x, t) = EΨ(x, t)
2m
We concluded that the Schrödinger equation is based on energy accounting — without
external interaction, the energy E is kinetic.

A plausible way to generalize the Schrödinger equation to include the effects of an


external force is merely to tack the potential energy U (x) associated with that force.
Classical physics tells us that it is impossible to define a potential energy for nonconser-
vative forces. So our generalization is good only for conservative forces. Adding potential
energy, the Schrödinger equation should have the form:

(KE + U (x))Ψ(x, t) = EΨ(x, t)

So we might guess the appropriate partial differential equation replacing (3.2) to be:
~2 ∂ 2 Ψ(x, t) ∂Ψ(x, t)
− 2
+ U (x)Ψ(x, t) = i~ (3.5)
2m ∂x ∂t
This time-dependent Schrödinger equation (TDSE) has passed numerous experi-
mental tests, so we accept it as the correct law obeyed by matter wave. To determine
the behavior of a particle in classical mechanics, we solve F~ = md2~r/dt for ~r(t), given
knowledge of the net external force on the particle F~ . The analogous task in quantum
mechanics is to solve the TDSE for Ψ(x, t), given knowledge of the potential energy U (x).

3.3 Stationary States

The first step in solving (3.5) is to use a standard mathematical technique called sep-
aration of variables5 . In this approach, we assume that the wave function may be
expressed as a product of a spatial part and a temporal part:

Ψ(x, t) = ψ(x)φ(t) (3.6)


5
On separating variables, a partial differential equation in, say, N variables is reduced to N ordinary
differential equations, each involving only a single variable.
Chapter 3 The Schrödinger Equation 43

Inserting (3.6) into (3.5) and then factoring out those terms constant with respect to the
partial derivatives, we have:

~2 d2 ψ(x) dφ(t)
− φ(t) 2
+ U (x)ψ(x)φ(t) = i~ψ(x)
2m dx dt
The next step in separation of variables is to divide both sides by the product ψ(x)φ(t).
We then obtain:
~2 1 d2 ψ(x) 1 dφ(t)
− 2
+ U (x) = i~ (3.7)
2m ψ(x) dx φ(t) dt
The two sides of this equation depend on entirely independent variables, the left side on
position and the right side on time. The only way to make this equation hold for all time
and positions is for each side to equal to a time- and space-independent constant, the
same for both side. Denoting this separation constant as C gives:

~2 1 d2 ψ(x) 1 dφ(t)
− 2
+ U (x) = i~ =C (3.8)
2m ψ(x) dx φ(t) dt

In general, the potential energy could be a function of time, U (x, t), and the technique
we have used here would fail. Here, we will consider only the cases in which the potential
energy is time independent. For such cases, we have two separate, ordinary differential
equations, linked together by the separation constant C. Let’s analyze them one by one.

First, we consider the “temporal half” of (3.8) which can be rewritten as:

dφ(t) iC
= − φ(t)
dt ~
The solution of this equation is:

φ(t) = e−i(C/~)t = cos[(C/~)t] − i sin[(C/~)t] (3.9)

Each part of this solution is a pure sinusoidal function whose angular frequency ω = C/~.
A pure frequency implies a pure, well-defined energy through the fundamental relationship
E = ~ω. Therefore,
C = ~ω = E

In separating variables, we focus on those solutions whose energy is well defined, and the
separation constant is that energy.

According to assumption (3.6), the total wave function is now:

Ψ(x, t) = ψ(x)e−i(E/~)t (3.10)


Chapter 3 The Schrödinger Equation 44

Since we haven’t considered any particular potential energy U (x), ψ(x) is still general.
Nonetheless, we calculate the probability density:

|Ψ(x, t)|2 = Ψ∗ (x, t)Ψ(x, t) = [ψ ∗ (x)e+i(E~)t ][ψ(x)e−i(E~)t ] = |ψ(x)|2 (3.11)

Its time dependence disappear! So the solutions of form (3.10) are called stationary
states. For stationary states, all probabilities are static and can be calculated from the
spatial part ψ(x).

Let’s now turn to the “spatial half” of (3.8). Replacing C by E and multiplying both
sides by ψ(x), we obtain the time-independent Schrödinger equation (TISE):
~2 d2 ψ(x)
− + U (x)ψ(x) = Eψ(x) (3.12)
2m dx2
We can see that the potential energy U (x) distinguishes one application from another,
just as the net force does classically. In contrast, the temporal part of the wave function
φ(t) is always e−i(E/~)t for any potential energy U (x). Unfortunately, the functions U (x)
for which (3.12) has a simple solution are few, and this dictates what we regard as a
simple case. In the next two chapters, we are going to study the spatial wave functions
ψ(x) that follow from different “simple” potential energies.

3.4 Physical Conditions: Well-Behaved Functions

For each potential energy, we seek the wave function ψ(x) that describes the particle. But
certain physical conditions must be met in all cases. We have the following requirements
for a physical wave function ψ(x).

1. To be physically acceptable, a wave function must be single-valued. There should be


only one probability for a particle to be at a specific location at a specific time. As
the probability density is proportional to the amplitude of the matter wave, the wave
function must be single-valued.

2. To be physically acceptable, a wave function must be smooth. There are two aspects for
smoothness: continuity of the wave function and continuity of its derivative. Never
may the wave function itself be discontinuous. Not only this would be contrary
to what we expect of wavelike phenomena, but it would also represent a particle
of infinite kinetic energy. It’s because the kinetic energy term in the Schrödinger
equation (3.12) has a second order derivative of ψ(x) with respect to position.
Chapter 3 The Schrödinger Equation 45

With one exception, the first derivative or slope of the wave function must also be
continuous. If the first derivative is continuous, the second derivative and thus the
kinetic energy is finite, which must hold if the potential energy U (x) and total energy
E are both finite, for the Schrödinger equation is based on these three energies adding
up. The exception arises if we do allow the potential energy to jump to infinity at
any point along the x-axis for then the restriction against the kinetic energy being
infinite at that point fails. In this idealized but useful case of walls of infinite potential
energy, the wave function’s slope need not be continuous at the walls.

3. To be physically acceptable, a wave function must be normalizable. In other words,


we require a 100 % probability of finding the particle somewhere in the space. The
probability per unit length is |Ψ(x, t)|2 , so the normalizable condition implies:
Z
|Ψ(x, t)|2 dx = 1 (3.13)
all space

A wave function is said to be normalized if it obeys this equation, which is known as


the normalization condition. One consequence of this condition is that the wave
function itself can’t become infinite. Another consequence is that the wave function
must approach zero as x → ±∞.

These constraints on physically acceptable solutions of the Schrödinger equation will lead
to some specific conditions at the boundaries known as the boundary conditions.

3.5 Expectation Values and Uncertainties

By the Heisenberg uncertainty principle, we cannot speak with certainty about the po-
sition of a particle unless we know nothing about its momentum. Thus we might no
longer guarantee the outcome of a single measurement of any quantity that depends on
the particle’s position. However, we can still determine the expectation value which is
the average value for a large number of measurements taken under identical conditions.

Suppose ψ(x) is the spatial wave function for the stationary state of a time independent
potential U (x). In this case, the probability per unit length of detecting the particle is
|ψ(x)|2 . As illustrated in Fig. 3.2, the probability of finding the particle in the interval
dx around x is equal to:
P (x)dx = |ψ(x)|2 dx, (3.14)
Chapter 3 The Schrödinger Equation 46

Figure 3.2: The probabilities of finding the particle described by ψ(x) between any two
positions x1 and x2 is determined by summing up the probabilities in each interval dx.

If we can calculate this probability at different positions, then we can find the average
outcome for a large number of measurements, i. e. the expectation value. For example,
suppose we obtain the value x1 for n1 times, x2 for n2 times and so on by repeating an
experiment many times. The average value of x is given by
P
n 1 x1 + n 2 x2 + . . . n i xi
xav = = P
n1 + n 2 + . . . ni

The number of times ni that we measure each xi is proportional to the probability P (xi )dx
to find the particle in the interval dx at xi . Thus the expectation value of x is:
R
xP (x)dx
hxi = Rall space (3.15)
all space
P (x)dx

(Note that the expectation value hxi is also denoted as x̄.) For normalized wave function,
Z
hxi = x|ψ(x)|2 dx (3.16)
all space

By analogy, the expectation value of any spatial function f (x) is:


Z
hf (x)i = f (x)|ψ(x)|2 dx (3.17)
all space

In quantum mechanics, we define the uncertainty as standard deviation. In the case


of position, the uncertainty is:
sZ
∆x = (x − hxi)2 |ψ(x)|2 dx (3.18)
all space
Chapter 3 The Schrödinger Equation 47

We can rewrite this equation as:


p
∆x = hx2 i − hxi2 (3.19)

So we can find the uncertainty of x once we obtain the expectation values hxi and hx2 i.

Example 3-1

The initial wave function of a particle is given as

Ψ(x, 0) = C exp(−|x|/x0 )

where C and x0 are constant.

(a) Sketch Ψ(x, 0). Find C in terms of x0 such that Ψ(x, 0) is normalized.
(b) Calculate the probability that the particle will be found between −x0 ≤ x ≤ +x0 .

Solution:

(a) Ψ(x, 0) is symmetric, decaying exponentially from the origin in either direction, as
shown in the below figure. The decaying length x0 represents the distance over
which the wave amplitude is diminished by the factor 1/e from its maximum value
Ψ(0, 0) = C.
Chapter 3 The Schrödinger Equation 48

The normalization requirement is


Z +∞ Z +∞
2 2
|Ψ(x, 0)| dx = C exp(−2|x|/x0 )dx = 1
−∞ −∞
Z +∞
2
⇒ 2C exp(−2x/x0 )dx = 2C 2 (x0 /2) = 1
0

⇒ C = 1/ x0

(b) The probability P is the area under the curve of |Ψ(x, 0)|2 from −x0 to +x0 . So it is
given by:
Z +x0 Z +x0
2
P = |Ψ(x, 0)| dx = 2 |Ψ(x, 0)|2 dx
−x0 0
Z +x0
= 2C 2 exp(−2x/x0 )dx = 2C 2 (x0 /2)(1 − e−2 ) = 0.8647
0
Chapter 4

Solutions of Time-Independent
Schrödinger Equation I

In this chapter, we study the applications of time-independent Schrödinger equation to


bound states. A particle is in a bound state when it is confined by an external force
to a finite region of space. For instance, a mass on a spring is in a bound state. Here,
we concentrate on the bound states of one-dimensional systems in which the potential is
piecewise-constant, i. e. the potential is constant for all positions except at a finite number
of discontinuities. Of course, real potentials are continuous, finite functions. However, if a
real potential changes rapidly only in a few very small regions of space, it can be modeled
accurately by a piecewise-constant potential for which the solution of the TISE can be
found easily.

4.1 Classical and Quantum-mechanical Bound States

Classical and quantum bound states have some common properties. To study quantum
bound particles, we must first understand how classical bound particles behave.

Let’s consider a simple example of a mass attached to a spring with spring constant
κ. Figure 4.1 shows how the energy of the mass varies with its position. The total
energy E = KE + U is constant no matter what initial value of E is given to the system.
Increasing E would increase the amplitude A of the motion. But at any finite E, the mass
is bound within the region −A ≤ x ≤ +A, i. e. the region “inside” the potential energy
curve as shown in Fig. 4.1.

49
Chapter 4 Solutions of Time-Independent Schrödinger Equation I 50

Figure 4.1: Energy versus position for a mass attached to a spring.

Suppose the mass is moving to the right starting from an arbitrary position x. As it
moves, its kinetic energy decreases while its potential energy increases. When it reaches
the point x = +A, its kinetic energy becomes zero and the mass stops momentarily.
The mass then return to the origin due to the force pulled by the spring. So the point
x = +A is a turning point. Similarly, the other turning point is located at x = −A.
No matter how much energy E it has, the mass cannot proceed to the regions x > +A
and x < −A where its kinetic energy would have to be negative. So these regions are
known as classically forbidden region while the region −A ≤ x ≤ +A is known as
classically allowed region.

Next, we consider a different bound system in which bound states are possible. Figure
4.2 shows a system composed of a large atom fixed at the origin and a small atom moving
freely in response to their shared electrostatic interaction. If the total energy is negative,
labeled Ebound , the small atom will be bound, oscillating between the turning points xa
and xb . However, if the total energy is positive, labeled Eunbound , the small atom has
only a single turning point. Were the small atom moving toward the large atom, it would
momentarily stop at this point, then move away, never to return. This situation differs
from the case of a mass on a spring, for which there is no total energy large enough for
the mass to be unbound.
Chapter 4 Solutions of Time-Independent Schrödinger Equation I 51

Figure 4.2: Energy versus position for the interatomic force between a large atom fixed
at the origin and a small one free to move.

In common with classical bound states, quantum bound states have constant total
energy energy, shared between kinetic and potential. But in quantum mechanics, we
don’t speak of a particle’s position at a given time. Instead, we seek the wave function
Ψ(x, t). We will soon find that quantum bound states are standing waves. Accordingly,
there are only certain discrete states possible for a matter wave bound by a potential
energy U (x). Moreover, these states exist only for particular values of the total energy
E, which are known as the energy eigenvalues. Unlike classical bound states, quantum
bound states are restricted to only certain energies, i. e. the energy is quantized. There
is another interesting difference between quantum and classical bound states. A classical
particle is restricted to classcially allowed region. However, wave functions for quantum
mechanical bound states often extend into classically forbidden region.

4.2 Solving the TISE for a Piecewise-constant Poten-


tial

The solution of the TISE is completely trivial for a piecewise-constant potential. What
makes finding the solution so easy is the simple analytic form of the “pieces” of the solution
Chapter 4 Solutions of Time-Independent Schrödinger Equation I 52

of the TISE in each region of space in which the potential is constant. The particular
functional form of these pieces depends on whether the energy of the state is less than
or greater than the potential energy in that region. But in either case, the pieces of the
solution are easy to write down. Indeed, the only non-trivial part of solving the problem
is hooking the pieces together.

Although the details of how to solve bound-state and unbound-state problems differ,
certain characteristics are shared by both kinds of states. Below are the procedures to
obtain the bound and unbound states for a piecewise-constant potential:

1. Divide the whole space (−∞ < x < ∞) into regions of constant potential according to
the values of U (x). Classify each region as classically allowed or classically forbidden
according to the energy of the state E.
2. Write down the general solution to the TISE in each region. (For more complicated
U (x), we may need to use some mathmetical techniques to find the general solution.)
3. Impose physical conditions on the pieces of ψ(x) in the first and last regions. These
conditions will depend on whether the state is bound or unbound: for bound states,
they ensure the wave function goes to zero as x → ±∞; for unbound states, they
correspond to circumstances in the laboratory.
4. Use continuity conditions to hook the pieces of the wave function together.
5. Finish the problem as follows:

Bound States: Impose boundary conditions (usually in the limit x → ±∞) to


obtain quantized energies. Then normalize the wave function ψ(x).
Unbound States: Evaluate the probability current densities in the first and
last regions. From the probability current density, calculate the probabilities for
transmission and reflection.

4.3 Particle in a Box - The Infinite Well

In this problem, we consider a particle of mass m moving freely in a one-dimensional


“box” of length L. Suppose the walls of the box, which are assumed to be perfectly rigid,
are located at x = 0 and x = L. The particle is completely trapped within the “box” and
they must have the following properties:

1. The regions x < 0 and x > L are forbidden. The particle cannot leave the box.
Chapter 4 Solutions of Time-Independent Schrödinger Equation I 53

2. The particle can move freely between 0 and L at constant speed and thus with
constant kinetic energy.
3. No matter how much kinetic energy the particle has, its turning points are at x = 0
and x = L.

A potential-energy function describing a particle in such situation is:


(
0 0≤x≤L
U (x) = (4.1)
∞ x < 0 or x > L

The potential energy is sketched in Fig. 4.3. It is known as the infinite well potential
energy, for which we want to solve the Schrödinger equation.

Figure 4.3: Infinite well potential energy.

What are the boundary conditions that the solution must satisfy? Because it is phys-
ically impossible for the particle to be outside the box, we require

ψ(x) = 0 for x < 0 or x > L (4.2)

In other words, there is zero probability of finding the particle outside the box. Because
the solution is zero everywhere outside the box, continuity condition requires that

ψ(x = 0) = 0 and ψ(x = L) = 0 (4.3)

At all points inside the box, the potential energy is U (x) = 0. Plugging this potential
energy into the TISE (3.12) gives

~2 d 2 ψ
− = Eψ(x) (4.4)
2m dx2
Chapter 4 Solutions of Time-Independent Schrödinger Equation I 54

That means we need to solve a differential equation of the form


d2 ψ
= −k 2 ψ(x)
dx2

where k = 2mE/~ is a constant. Its solution must be a function whose second derivative
equal to a negative constant times the function itself. Two such functions are

ψ1 (x) = sin kx and ψ2 (x) = cos kx

Both are solutions of Equation (4.4). Moreover, they are linearly independent because
ψ2 (x) is not a multiple or a rearrangement of ψ1 (x). Therefore, the general solution of
the TISE for a particle in a rigid box is:

ψ(x) = A sin kx + B cos kx (4.5)

The constant A and B can be determined by applying the boundary conditions (4.3).
First, the wave function must go to zero at x = 0, i. e.

ψ(x = 0) = A · 0 + B · 1 = 0

It can be satisfied only if B = 0. Thus the physically meaningful solution is

ψ(x) = A sin kx (4.6)

The wave function must also go to zero at x = L, i. e.

ψ(x = L) = A sin kL = 0

This condition could be satisfied by A = 0, but we would reject it to avoid the unphysical
solution of ψ = 0 at all points. We would instead consider the boundary condition is
satisfied by

kL = nπ ⇒ k= n = 1, 2, 3, . . . (4.7)
L
Notice that the integer n starts from 1, not 0. If n = 0, then k = 0 and ψ = 0 at all
points, a physically meaningless solution. Thus the solutions to the TISE for a particle
in a rigid box are
 nπx 
ψn (x) = A sin n = 1, 2, 3, . . . (4.8)
L
We have found a whole family of solutions, each corresponding to a different value of the
integer n. The wave functions, which are of typical standing-wave form, represents the
Chapter 4 Solutions of Time-Independent Schrödinger Equation I 55

sustained states of motion of a particle in a box. Quantum mechanically, bound states


are standing waves.

For each value of the quantum number n, there is a specific wave function ψ n (x)

describing the stationary state of the particle with energy En . Remembering k = 2mE/~,
we can solve for En :
n 2 π 2 ~2
En = n = 1, 2, 3, . . . (4.9)
2mL2
The particle’s energy must be one of these discrete values. The only allowed energies
for the particle are E1 , 4E1 , 9E1 , 16E1 , . . . . All intermediate values are forbidden. In
other words, the particle’s energy is quantized! Thus the interger n, which determines the
energy quantization, is called the quantum number.

We can determine the constant A by considering the normalization condition. Inserting


our newly found wave function into (3.13), we have
Z Z +∞ Z L  nπx 
2 2 2
|Ψ(x, t)| dx = |ψn (x)| dx = A sin2 dx = 1
all space −∞ 0 L
r
2
⇒ A= for n = 1, 2, 3, . . . (4.10)
L

We now have a complete solution to this problem. The normalized wave functions
representing the allowed states for a particle in an infinite well are
 r
 2  nπx 

sin 0≤x≤L
ψn (x) = L L (4.11)

 0 x < 0 or x > L

Figure 4.4 shows the wave functions ψ(x), probability densities |ψ|2 and their correspond-
ing energies for the four lowest energy states. The lowest energy state, for which n = 1,
is known as the ground state and its energy is called the ground state energy. The
states with higher energies (n > 1) are known as the excited states.

We have found that there is no quantum state having E = 0 for a particle in a box.
Unlike a classical particle, a quantum bound particle cannot be stationary! Otherwise, it
would violate the Heisenberg uncertainty principle. Because the particle is somewhere
between the box, its position uncertainty is ∆x = L. If the particle were at rest in the
box, we would know that its velocity and momentum are exactly zero with no uncertainty,
i. e. ∆px = 0. But then ∆x∆px = 0 would violate the Heisenberg uncertainty principle.
Thus it follows from the uncertainty principle that a confined particle cannot be at rest.
Chapter 4 Solutions of Time-Independent Schrödinger Equation I 56

Figure 4.4: The wave functions ψ(x), probability densities |ψ|2 and their corresponding
energies of the four lowest energy states for a particle in an infinite well.

Example 4-1

Show that the expectation value of x is L/2 for a particle in an infinite potential well
located at 0 ≤ x ≤ L, which is independent of the quantum number n.

Solution:
Since ψn = 0 except for 0 ≤ x ≤ L, the expectation value of x for the particle in the
quantum state n is
Z L
hxi = x|ψ|2 dx
0
Z
2 L  
2 πx
= x sin dx
L 0 L
Z   
1 L 2πx
= x 1 − cos dx
L 0 L
1
=
2
This result is independent of n as required.
Chapter 4 Solutions of Time-Independent Schrödinger Equation I 57

Sketching wave functions


We can draw wave functions for quantum states of different energies as follows:

1. Draw a graph of the potential energy U (x). Show the allowed energy E as a
horizontal line. Locate the classical turning points.
2. Draw the wave function as a continuous, oscillatory function between the turning
points. The wave function for quantum state n has n anti-nodes and n − 1 nodes
(excluding the ends).
3. Make the wavelength longer (larger node spacing) and the amplitude higher in
regions where the kinetic energy is smaller. Make the wavelength shorter and
the amplitude lower in regions where the kinetic energy is larger.
4. Bring the wave function to zero at the edge of an infinitely high potential-energy
“wall”.
5. Draw the wave function to be decaying exponentially inside a classically forbid-
den region where E < U . The penetration distance h increases as E gets closer
to the top of the potential-energy well.

As an example, below figure shows how to sketch the wave functions for the n = 1
and n = 4 states of a infinite potential energy well by following the above steps.
Chapter 4 Solutions of Time-Independent Schrödinger Equation I 58

Example 4-2

An electron is trapped in a one-dimensional region of length 1.0 × 10−10 m (a typical


atomic diameter).

(a) How much energy must be supplied to excite the electron from the ground state to
the first excited state?
(b) In the ground state, what is the probability of finding the electron in the region from
x = 0.090 × 10−10 m to 0.110 × 10−10 m?

Solution:

(a) The ground state energy is

~2 π 2 (1.055 × 10−34 J · s)2 π 2


E1 = 2
= −31 −10 2
= 6.03 × 10−18 J = 37.6 eV
2mL 2(9.109 × 10 kg)(1.0 × 10 m)

With E2 = 4E1 , the energy difference ∆E is

∆E = E2 − E1 = 3E1 = 3(37.6 eV) = 112.8 eV

(b) The probability P is the area under the curve of |ψ(x)|2 from x1 to x2 , i. e.
Z x2 Z   x2
2 2 x2 2  πx  x 1 2πx
P = |ψ| dx = sin dx = − sin = 0.383%
x1 L x1 L L 2π L x1

4.4 The Finite Well

The finite well involves a more realistic potential energy function than the infinite po-
tential well, so it’s a more accurate model in many circumstances. Since the “height” of
the potential energy walls is no longer infinite, we cannot find the analytical solution of
the TISE for a finite well.

As depicted in Fig. 4.5, the potential energy function that defines a finite well is given
by: (
0 0≤x≤L
U (x) = (4.12)
U0 x < 0 or x > L
Chapter 4 Solutions of Time-Independent Schrödinger Equation I 59

Because it is finite, a particle could be given enough total energy (E > U0 ) to escape, but
we restrict our attention to bound states where E < U0 . Note that there is still an abrupt
jump in potential energy at the walls. It means that the force is infinite there and so a
classical particle bouncing off the wall would still do so instantly.

Figure 4.5: Finite well potential energy.

For convenience, we rewrite the time-independent Schrödinger equation (3.12) as:


d2 ψ(x) 2m[U (x) − E]
= ψ(x) (4.13)
dx2 ~2
In the region 0 ≤ x ≤ L, the proportionality constant between ψ(x) and its second order
derivatives is negative. That means ψ(x) and d2 ψ(x)/dx2 are of the opposite sign which
describes an oscillatory function like sine. On the other hand, the proportionality constant
is positive in the regions outside the well. So ψ(x) and d2 ψ(x)/dx2 must be of the same
sign. Oscillatory functions do not meet this criterion, but exponential functions do. For
U (x) = U0 , equation (4.13) is satisfied by a function of the form:
r
2m(U0 − E)
ψ(x) = Ae±αx where α ≡ (4.14)
~2
In spite of the troubling notion of negative kinetic energy (which we return to later), we
have no reason to exclude the wave function outside the well. The wave function would
be decaying exponentials on both sides outside the well so long as we assume ψ(x) = e−αx
for x > 0 and ψ(x) = e+αx for x < 0. Thus, they fit our physical conditions outside the
well.

The wave function depicted in Fig. 4.6 clearly shows the difference in character between
the region inside the well (U < E), where ψ(x) is oscillatory, and those outside (U >
E), where it exponentially decays. It is physically acceptable because it satisfies the
Schrödinger equation, does not diverge, and is smooth everywhere.
Chapter 4 Solutions of Time-Independent Schrödinger Equation I 60

Figure 4.6: For a finite well, the behavior of the wave function depends on which is larger
— E or U .

Figure 4.7: Unless the energy is just right, a solution of the Schrödinger equation would
diverge at infinity.

So where does the energy quantization come in? Figure 4.7 shows an acceptable finite
well standing wave with its “exponential tails” in the classically forbidden region, while
Chapter 4 Solutions of Time-Independent Schrödinger Equation I 61

above and below it are mathematical solutions of the Schrödinger equation for slighly
p
higher and lower energies. The wave number k = 2π/λ is equal to 2mE/~, so a higher
E implies a shorter wavelength λ. From this figure, we can see that only wave functions
of certain energies can satisfy the smoothness condition without diverging at one end or
both. So energy is also quantized for a particle in a finite well.

Figure 4.8 shows allowed wave functions and probability densities for both the finite
and infinite wells, where the horizontal axis is at a height proportional to that state’s
energy. The finite and infinite wells have some common properties such as the quan-
tized energies, a ground state in which the total energy is not zero. But how about the
difference?

Figure 4.8: Wave functions, probability densities, and energies in the finite and infinite
wells.

Perhaps the most remarkable difference is that finite well wave functions penetrate
into classically forbidden region. Although exponentially decaying, there is a non-zero
probability of finding the particle outside the well where the kinetic energy is negative
as E < U0 . We measure how far ψ(x) extends into a classically forbidden region by a
parameter called the penetration depth δ. It is defined to be the distance over which the
wave function ψ(x) drops by a fraction of 1/e. From Equation (4.14), the exponentially
decaying wave function outside the well has the form:

ψ(x) ∝ e−α|x|
Chapter 4 Solutions of Time-Independent Schrödinger Equation I 62

So the exponential decays by 1/e at the point where α|x| = 1. Hence, the penetration
depth is given by:
1 ~
δ≡ =p (4.15)
α 2m(U0 − E)
This says that the penetration should become deeper as the energy E nears the value of
the confining potential U0 . We can show that the penetration distance is consistent with
the uncertainty in the position of a particle that entered into the forbidden region.

Actually, it’s problematic to detect a particle with negative kinetic energy. To de-
tect the particle in the classically forbidden region, the experiment would need a position
uncertainty no larger than ∼ δ, the interval in which the probability of detection is signif-
icant. This would introduce a momentum uncertainty no smaller than ∼ ~/δ, implying
kinetic energy uncertainty no smaller than ∼ ~2 /(2mδ 2 ). Such uncertainty in the kinetic
energy is approximately equal to the magnitude of the (negative) kinetic energy. Thus,
no experiment could be certain that the detected particle has negative kinetic energy.

Related to penetration of classically forbidden region is another important distinction


between finite and infinite wells. As Figure 4.8 shows, the energy levels in a finite well
are lower than those of an infinite well of equal width. This is true simply because the
wave functions in a finite well do not stop abruptly at the walls.

4.5 The Simple Harmonic Oscillator

The one-dimensional simple harmonic oscillator is a more complicated bound state system.
An example of a classical oscillator is a mass m subjected to a restoring force F = −κx,
where x is the displacement from its equilibrium position. The oscillator has maximum
kinetic energy at x = 0, and its kinetic energy vanishes at the turning points x = ±A.

Why we analyze the motion of such a system using quantum mechanics? Although we
never find in nature an example of a one dimensional particle oscillator, there are systems
that behave approximately as one, for example, a diatomic molecule vibrating with small
oscillations. In fact, any system in a potential energy minimum behaves approximately
like a simple harmonic oscillator.

A force F = −κx has the associated potential energy U = 21 κx2 . Inserting the potential
energy, the time independent Schrödinger equation becomes:
~2 d 2 ψ 1 2
− + κx ψ(x) = Eψ(x) (4.16)
2m dx2 2
Chapter 4 Solutions of Time-Independent Schrödinger Equation I 63

This differential equation is difficult to solve directly, and so we guess its solutions. The
solution must approach zero as x → ±∞, and in the limit x → ±∞ the solutions of
2
equation (4.16) behave like exponentials of −x2 . We therefore try ψ(x) = Ae−ax , where
A and a are constant that are determined by evaluating (4.16) for this choice of ψ(x). We
plug ψ(x) and d2 ψ/dx2 into the equation to see if there is a solution:
~2 2 2 1 2 2
− (−2aAe−ax + 4a2 x2 Ae−ax ) + κx2 (Ae−ax ) = EAe−ax
2m 2
~2 a 2a2 ~2 2 κ 2
⇒ − x + x =E (4.17)
m m 2
Note that we are looking for a solution which is valid for any x, not just for one specific
value. In order for this to hold, it must be true that

2a2 ~2 κ κm
− + =0 ⇒ a= (4.18)
m 2 2~
r
~2 a ~ κ 1
−E =0 ⇒ E = = ~ω0 (4.19)
m 2 m 2
p
where the classical frequency ω0 = κ/m. From the normalization condition, the coeffi-
cient A can be shown to be given by:
 mκ 1/8
A=
π 2 ~2
Figure 4.9 illustrate the solution we have found. We again see the penetration of quantum
wave into classically forbidden region.

Figure 4.9: The ground state of a one-dimensional harmonic oscillator. The kinetic energy
K equals the difference between the total energy E and the potential energy U = 12 κx2 .

The solution we have found corresponds to the ground state of the harmonic oscillator.
2
The general solution is of the form ψn (x) = Afn (x)e−ax , where fn (x) is a polynomial in
Chapter 4 Solutions of Time-Independent Schrödinger Equation I 64

which the highest power of x is xn . The corresponding energies are:


  r
1 κ
En = n + ~ω0 n = 0, 1, 2, 3, . . . where ω0 = (4.20)
2 m

Figure 4.10: Wave functions, probability densities, and energies of a harmonic oscillator.

The energy levels of a harmonic oscillator are illustrated along with their corresponding
wave functions and probability densities in Fig. 4.10. These energy levels are uniformly
spaced, in contrast to the cases of finite and infinite wells. All these solutions have the
property of penetration of probability density into the forbidden region beyond the classi-
cal turning points. The probability density oscillates, somewhat like a sine wave, between
2
the turning points, and decreases like e−2ax to zero beyond the turning points. Moreover,
an oscillator is most likely to be found at the center where its speed is smallest. It agrees
with the predictions of classical mechanics that an oscillator is most likely to be found
where its speed is smallest.
Chapter 5

Solutions of Time-Independent
Schrödinger Equation II

In this chapter, we apply the time-independent Schrödinger equation to investigate some


more realistic systems. We begin from the study of unbound states in one dimension.
Because an unbound particle isn’t confined to a region of space, standing waves do not
form and energy is not quantized. However, we find that the particle still exhibits sur-
prising behavior due to its wave nature. For example, a particle may be turned back
by a force when classical mechanics say it should proceed, and it may prevail against a
force where classical physics say it should be turned back. Next, we study the simplest
three-dimensional bound system — the 3D infinite well. We shall see that, for bound
systems in three dimensions, boundary conditions imposed on the wave function lead to
three-dimensional standing waves and three quantum numbers governing quantization of
energy and other physical properties.

5.1 The Potential Step

We now consider a free particle encountering a simple force — an abrupt change in


potential energy known as a potential step. A free particle moves to the right in a region
(x < 0) of zero potential until it encounters a very narrow region where it experiences a
a backward “kick”. Passing through, its potential energy jumps abruptly to U0 and stays
there. The narrow region near x = 0 is the only place where the electron experiences a

65
Chapter 5 Solutions of Time-Independent Schrödinger Equation II 66

force. For this case, the potential energy is given by:


(
0 x<0
U (x) = (5.1)
U0 x≥0

Just as a light wave striking a glass plate divides — some transmitted, some reflected
— so does a matter wave divide as it encounters an abrupt change in conditions. But
because the matter wave is related to the probability of finding the particle, the division
of the wave governs the probability of a given incident particle being transmitted or
reflected. To say anything about this, we must be able to distinguish between particles
moving one way and those moving the other, and to do this we must be careful about the
wave functions we use. In Section 3.1, we discussed the plane-wave solution of the free
particle Schrödinger equation:

Ψ(x, t) = Ae+ikx−iωt

This represents a particle with a well-defined momentum moving in the positive x-


direction. For the cases we considered here, we will use it to represent a steady beam of
particles all moving in the positive x-direction, with A related to the number per unit dis-
tance. By replacing k by −k, we represent a beam with well-defined momentum moving
in the negative x-direction as:
Ψ(x, t) = Ae−ikx−iωt

Because we will concentrate on stationary states, the spatial parts are our main concern:
ψ(x) = e+ikx moves right and ψ(x) = e−ikx moves left.

Notice that the time-independent Schrödinger equation (3.12) for U = 0 is


r
d2 ψ(x) 2mE 2mE
2
= − 2 ψ(x) = −k 2 ψ(x) where k = (5.2)
dx ~ ~2
All four functions sin kx, cos kx, e+ikx , e−ikx satisfy the equation. Why here we should use
e+ikx and e−ikx instead of sin kx and cos kx? It’s because we need to keep directions of
motion separate in an unbound system. So e+ikx and e−ikx are the correct choice.

Potential energy step, E > U0


Let’ us start with the case in which an incident particle has kinetic energy E greater
than the potential energy jump as shown in Fig. 5.1. Classically, it should merely slows
down abruptly at x = 0, then proceed at reduced speed. There is no classical turning
point and the particle would not rebound.
Chapter 5 Solutions of Time-Independent Schrödinger Equation II 67

Figure 5.1: A free particle encounters a potential step that is classically surmountable.

But quantum mechanics tells us another story. In the region x < 0, we will have a
right-moving incident wave, and we allow for a left-moving reflected wave. Each solution
of (5.2) is still a solution if it includes an arbitrary multiplicative constant. So we use
Ae+ikx for the incident wave and Be−ikx for the reflected wave.

In the region x > 0 where U = U0 , the time-independent Schrödinger equation is:


r
d2 ψ(x) 2m(E − U0 ) 02 0 2m(E − U0 )
2
=− 2
ψ(x) = −k ψ(x) where k = (5.3)
dx ~ ~2
The mathematical solutions here are the same as those of (5.2) except that k is replaced
by k 0 . To again distinguish directions, we choose the pair of functions e+ikx and e−ikx .
Beyond x = 0, there is no change in potential energy — no force to reflect anything
moving to the right. So there should be nothing moving to the left in this region, and
e−ikx is thus physically inapplicable. In summary, the solutions to the TISE in the two
regions are:
ψx<0 (x) = Ae+ikx + Be−ikx
Incident Reflected
wave wave
0
(5.4)
ψx>0 (x) = Ce+ik x
Transmitted
wave

Figure 5.2 represents these various parts schematically.

Note that the complex square of each part in (5.4) is proportional to the number of
particles per unit distance. So the densities of the incident particles, the reflected particles
and the transmitted particles are given by:

|ψincident |2 = A∗ A, |ψreflected |2 = B ∗ B, |ψtransmitted |2 = C ∗ C (5.5)


Chapter 5 Solutions of Time-Independent Schrödinger Equation II 68

Figure 5.2: Incident particles may be transmitted or reflected when they encounter a
potential step whose height U0 is smaller than the particle energy.

We find that a matter wave reflects on encountering the potential step, which is classically
impossible. In fact, without allowing for reflection in our matter wave, we could not satisfy
the smoothness requirement:

dψx<0 dψx>0
ψx<0 (0) = ψx>0 (0) and = (5.6)
dx x=0 dx x=0
Applying these conditions on the wave function in (5.4) gives:

A+B =C and k(A − B) = k 0 C (5.7)

Since no restriction on the value of k arises from conditions (5.7), there is no quantization
of energy and so an incident particle may have any energy.

What conditions (5.7) give us are reflection and transmission probabilities. We calcu-
late these from ratios of numbers of particles per unit time in the corresponding beams:
number number distance
= × ∝ |ψ|2 v ∝ |ψ|2 k (5.8)
time distance time
since v = p/m = ~k/m ∝ k. The transmission probability T is thus:
 ∗  0
number transmitted/time |ψtransmitted |2 kx>0 C C k
T = = 2
= (5.9)
number incident/time |ψincident | kx<0 A∗ A k
And the reflection probability R is:
number reflected/time |ψreflected |2 kx>0 B∗B
R= = = (5.10)
number incident/time |ψincident |2 kx<0 A∗ A
It can be shown that T and R can be expressed as:
p √ √
4kk 0 E(E − U0 ) (k − k 0 )2 ( E − E − U 0 )2
T = =4 √ √ , R= = √ √
(k + k 0 )2 ( E + E − U 0 )2 (k + k 0 )2 ( E + E − U 0 )2
Chapter 5 Solutions of Time-Independent Schrödinger Equation II 69

Since the total number of particles should be conserved, the transmission and reflection
probabilities must add to 1. We can verify that this is true for our transmission and
reflection probabilities.

Potential energy step, E < U0


Now we consider what happens when particles encounter a potential step whose height
U0 is greater than the particle energy E as shown in Fig. 5.3. A classical particle would
slow to a stop somewhere in the tiny region around the origin, then reverse its direction.
It certainly should reflect.

Figure 5.3: A free particle encounters a potential step that is classically insurmountable.

The analysis here differs very little from the case that E > U0 . For x < 0, we will
still have two solutions representing incident and reflected waves. But the situation has
changed for x > 0. The time-independent Schrödinger equation in this region becomes:
r
d2 ψ(x) 2m(U0 − E) 2 2m(U0 − E)
2
= 2
ψ(x) = α ψ(x) where α = (5.11)
dx ~ ~2
Note that ψ(x) and its second derivative are now of the same sign. Oscillatory functions
don’t satisfy this condition, but exponentials do. The solutions of (5.11) are:

ψ(x) = e+αx and ψ(x) = e−αx (5.12)

However, the region to the right of the step extends forever, we must throw out e+αx , for
it diverges as x → +∞. Thus, in place of (5.4), we now have

ψx<0 (x) = Ae+ikx + Be−ikx


Incident Reflected
wave wave (5.13)
ψx>0 (x) = Ce−αx
Chapter 5 Solutions of Time-Independent Schrödinger Equation II 70

Applying the smoothness conditions (5.6) on our wave function gives:

A+B =C and ik(A − B) = −αC (5.14)

A closer look at the magnitude of B tell us that the magnitudes of B and A are equal.
It follows that the reflection probability:
B∗B
R= =1 (5.15)
A∗ A
Because R and T must add to 1, it implies that the transmission probability T is zero.
The wave function falls exponentially to zero beyond x = 0, and nothing survives to be
found infinitely far away. Although the reflection does not occur at x = 0, the incoming
wave does reflect completely. Inferring T from R is the best way to determine it. We
should not calculate T by equation (5.8) since the speed v is not “real” for a negative KE.

On the other hand, it’s justified to calculate the number per unit distance since |ψ(x)|2
is real. Because the wave penetrates the step, it is possible to find the particle at the right
side of the step. However, we should note that as long as no attempt is made to find the
particle, we merely have an undisturbed wave that happens to penetrate the classically
forbidden step. The penetration depth δ is the same as in Section 4.4:
1 ~
δ= =p (5.16)
α 2m(U0 − E)
The closer E is to U0 , the greater is δ, and the slower the decay of the wave function. In
any case, the wave function is very small for x  δ.

Figure 5.4: Reflection and transmission probabilities for a potential step.

Figure 5.4 shows the reflection and transmission probabilities versus E/U0 for a po-
tential step. If E < U0 , the wave is completely reflected. When E > U0 , the reflection
probability falls rapidly with increasing E.
Chapter 5 Solutions of Time-Independent Schrödinger Equation II 71

5.2 Barriers and Tunneling

Tunneling is one of the most important and startling ideas in quantum mechanics. The
simplest situation is a potential barrier, a potential energy jump that is only temporary.
Let us apply the Schrödinger equation to study how tunneling occurs in a barrier.

Figure 5.5: Particles of energy E incident on a potential barrier whose height U0 > E.

Suppose particles of energy E incident from the left on a square potential barrier as
shown in Figure 5.5. The barrier is represented by the potential energy function:
(
0 x < 0 or x > L
U (x) = (5.17)
U0 0≤x≤L

Figure 5.6: The wave function for particles of different energies incident from the left.
Our experience then leads us to expect solution as shown in Figure 5.6. The intensity
of the wave can be found by proper application of the continuity conditions, which we do
Chapter 5 Solutions of Time-Independent Schrödinger Equation II 72

not discuss here. The intensity is found to depend on the energy of the particles as well
as the height and thickness of the barrier. Classically, the particles should never appear
at x > L, since they don’t have sufficient energy to overcome the barrier. This situation
is an example of quantum mechanical tunneling or barrier penetration. Note that the
particles can never be observed in the classically forbidden region 0 ≤ x ≤ L, but they
can tunnel through that region and be observed in the region x > L.

Although the potential energy barrier of Figure 5.5 is quite schematic and hypothetical,
there are many examples of tunneling in nature. Here, we consider the following examples.

1. Alpha decay. In this form of radioactive decay, an atomic nucleus emits an aggregate
of two protons and two neutrons known as an alpha particle. Classically, alpha
particles with kinetic energy less than 35 MeV cannot escape from the parent nucleus
due to the nuclear attraction. However, experiment reveal that alpha particles with
kinetic energy less than 35 MeV can escape from the nucleus. It is resulted from the
tunneling of the particles through a Coulombic-shape potential barrier as shown in
Fig. 5.7. Most alpha particles have energies E in the range of 4-9 MeV, yet their
decay rates differ by more than 20 orders of magnitude. The probability of the alpha
particle tunneling in a single encounter with the nuclear surface is very small. The
transmission probability in the tunneling is extremely sensitive to the value of E.

Figure 5.7: To escape from the parent nucleus, an alpha particle must tunnel.

2. The Scanning Tunneling Microscope. Images of individual atoms on the surface


of materials can be made with the scanning tunneling microscope (STM). Electrons
are trapped on a surface by a potential barrier. When a needle-like probe is placed
Chapter 5 Solutions of Time-Independent Schrödinger Equation II 73

about 1 nm from the surface, electrons can tunnel through the barrier between the
surface and the probe, producing a current that can be recorded in an external circuit
(see Figure 5.8). The current is very sensitive to the width of the barrier. In practice,
a feedback mechanism keeps the current constant by moving the tip up an down.
The motion of the tip gives a detail map of the surface.

Figure 5.8: Schematic diagram of an scanning tunneling microscope in operation.

3. Ammonia Inversion. Figure 5.9 is a representation of the ammonia molecule NH3 .


The Coulomb repulsion of the three hydrogen atoms establishes a barrier against the
nitrogen atom moving to a symmetric position on the opposite side of the plane of
the hydrogens shown as dashed lines in the figure. In fact, the nitrogen atom tunnels
back and forth with a high frequency.

Figure 5.9: (Left) A schematic diagram of an ammonia molecule. (Right) The potential
energy seen by the nitrogen atom in an ammonia molecule.
Chapter 5 Solutions of Time-Independent Schrödinger Equation II 74

4. The Tunnel Diode. A tunnel diode is an electronic device that makes use of the
tunneling phenomenon. The potential barrier “seen” by an electron in a tunnel diode
can be represented schematically by Figure 5.10. With no applied voltage, tunneling
occurs equally in both direction — no net flow of current. When a potential difference
is applied, the situation become asymmetric. Right and left tunneling rates differ,
and a net current flows. The response of a tunnel diode to applied voltages is unusual
and very fast. The current doesn’t steadily increase as applied voltage is increased;
at some points, it decreases. Moreover, changing the applied voltage changes the
transmission rates almost instantly.

Figure 5.10: The potential energy barrier seen by an electron in a tunnel diode.

5.3 The 3D Infinite Well

The simplest three-dimensional bound system is the 3D infinite well. The principle fea-
tures of the solution remain the same as in 1D case, but an important new feature known
as degeneracy is introduced. Here, we show how this occurs.

As always, the starting point is the potential energy. Fig. 5.11 depicts a 3D infinite
well in which the particle is bound in a box-shaped region by infinitely “high” walls at
Chapter 5 Solutions of Time-Independent Schrödinger Equation II 75

x = 0 and Lx , y = 0 and Ly , and z = 0 and Lz . The potential energy is therefore


(
0 0 ≤ x ≤ Lx , 0 ≤ y ≤ Ly , 0 ≤ z ≤ Lz
U (x, y, z) = (5.18)
∞ otherwise
As in one dimension, the wave function cannot penetrate outside infinite walls. So we
concentrate on the region inside where U = 0.

Figure 5.11: A three-dimensional infinite well.

To begin with, we need a Schrödinger equation that is valid in three dimensions.


As mentioned before, the Schrödinger equation in one dimension is based on an energy
accounting of the form KE + U = E. In three dimensions, the potential energy U is
a function of all spatial coordinates: U = U (x, y, z). So the wave function Ψ should
be a function of x, y, z and t. And only the kinetic energy term on the left need to
be generalized. The logical generalization is to replace the derivatives with respect to x
by derivatives with respect to both x, y and z. Thus the Schrödinger equation in three
dimensions is:
 2 
~2 ∂ ∂2 ∂2 ∂
− 2
+ 2 + 2 Ψ(x, y, z, t) + U (x, y, z)Ψ(x, y, z, t) = i~ Ψ(x, y, z, t) (5.19)
2m ∂x ∂y ∂z ∂t

As in one dimension, we consider stationary states in which the energy is well defined
and keeps constant. This comes from assuming the wave function to be a product of
spatial and temporal parts, i. e.

Ψ(x, y, z, t) = ψ(x, y, z)φ(t)


Chapter 5 Solutions of Time-Independent Schrödinger Equation II 76

By separating the variables as in Section 3.3, we find that the temporal part is again
φ(t) = e−i(E/~)t . And it again follows that |Ψ(x, y, z, t)|2 = |ψ(x, y, z)|2 . In addition, the
spatial part satisfies the time-independent Schrödinger equation:
 2 
~2 ∂ ∂2 ∂2
− + + ψ(x, y, z) + U (x, y, z)ψ(x, y, z) = Eψ(x, y, z) (5.20)
2m ∂x2 ∂y 2 ∂z 2

We can solve the partial differential equation (5.20) by the separation of variables. In
this approach, we assume that the spatial wave function is a product of three functions,
each of an independent variable:

ψ(x, y, z) = F (x)G(y)H(z)

Next, we insert this product into (5.20), then divide by the product. Noting that U = 0
inside the well, we thus obtain:
1 d2 F (x) 1 d2 G(y) 1 d2 H(z) 2mE
2
+ 2
+ 2
=− 2
F (x) dx G(y) dy H(z) dz ~
Each of the spatial variables is now in a separate term, so the usual argument applies: If
the equation holds for all x, y, z, each term must be a constant. Therefore,
d2 F (x) d2 G(y) d2 H(z)
= Cx F (x) = Cy G(y) = Cz H(z)
dx2 dy 2 dz 2
Just as in one dimension, each function has a second derivatives proportional to itself.
Because the wave function is zero outside infinite walls, continuity demands that it go
to zero at all walls. These conditions give us the solutions identical to those of the 1D
infinite well:
nx πx ny πy nz πz
F (x) = Ax sin G(y) = Ay sin H(z) = Az sin (5.21)
Lx Ly Lz

where nx , ny , nz are integers. Inserting (5.21) back into (5.20) yields:

n2x π 2 n2y π 2 n2z π 2 2mE


− 2 − 2 − 2 =− 2
Lx Ly Lz ~
 2 2 
nx ny n2z π 2 ~2
⇒ Enx ,ny ,nz = + + (5.22)
L2x L2y L2z 2m

The overall wave function is the product of the individual ones, and using A for the prodct
Ax Ay Az , we have
nx πx ny πy nz πz
ψnx ,ny ,nz (x, y, z) = A sin sin sin (5.23)
Lx Ly Lz
Chapter 5 Solutions of Time-Independent Schrödinger Equation II 77

Each set of the three quantum numbers (nx , ny , nz ) yields a unique wave function — a
unique three-dimensional standing wave.

Perhaps the most important lesson here is that whereas boundary conditions in one
dimension leads to energy quantization according to a single quantum number, imposing
boundary conditions in three dimensions naturally leads to three quantum numbers. In
general, there is an independent quantum number for each dimension.

Degeneracy
An ideal crucial to many applications of Schrödinger equation in three dimensions is
that multiple states can have the same energy and the number of equal-energy states
increases with the symmetry of the system.

Suppose the box is a cube which is as symmetric as possible, i. e. Lx = Ly = Lz ≡ L.


Equation (5.22) becomes

π 2 ~2
Enx ,ny ,nz = (n2x + n2y + n2z ) (5.24)
2mL2

2 2
π ~
(nx , ny , ny ) Enx ,ny ,nz / 2mL 2

(1, 1, 1) 3
(2, 1, 1) 6
(1, 2, 1) 6
(1, 1, 2) 6
(1, 2, 2) 9
(2, 1, 2) 9
(2, 2, 1) 9
(3, 1, 1) 11
(1, 3, 1) 11
(1, 1, 3) 11
2, 2, 2) 12

Table 5.1: States in a cubic 3D infinite well

Table 5.1 shows sets of quantum numbers for many allowed energies of a 3D infinite
well. We can see that different sets of quantum numbers may have the same energy.
However, each set of quantum numbers always corresponds to a different wave function.
Chapter 5 Solutions of Time-Independent Schrödinger Equation II 78

The coincidence — different wave functions having the same energy — is called degeneracy,
as illustrated in Fig. 5.12. Energy levels corresponding to multiple wave functions are thus
said to be degenerate. For example, the energy 27[π 2 ~2 /(2mL2 )] is said to be 4-fold
degenrate. On the other hand, energy levels corresponds to a single wave function are
said to be nondegenerate.

Figure 5.12: Degeneracy in a cubic infinite well: Equal energy but different states.

Degeneracy results from symmetry. However, if instead of a cubic well, we had a


well of unequal lengths. Much of the symmetry would be lost, and the results would be
splitting of formerly degenerate levels. For example, the energy levels E2,1,1 , E1,2,1 , and
E1,1,2 , which are degenerate for the cubic well, would no longer have the same energy for
a non-cubic well.
Chapter 6

Structure of the Atom

Our goal in this chapter is to investigate the atomic structure in details based on experi-
mental studies of atoms. We study how physicists made use of the information obtained
from experiments to develop atomic models which help us to understand and explain
the properties of atoms. In particular, we discuss the experiments that led to the Bohr
model, which is based on the concepts of matter waves. Although this model is not
strictly correct from the viewpoint of quantum mechanics, it still remains a convenient
mental picture of the atom since it can help us to understand many atomic properties.

6.1 The Atomic Nature of Matter

The idea that matter consists of small, indivisible particles is traceable to Greek philoso-
phers Leucippus and his pupil Democritus. They call these particles atoms. However,
due to the lack of experimental evidence, the existence of atoms was not widely accepted
until the early 1900s. The existence of atoms was confirmed in 1911 by Jean Perrin
(1870-1942) through experimental validation of Einstein’s theory of Brownian motion.

John Dalton (1766 - 1844) is the founder of the modern atomic theory. In 1808, he
proposed the atomic theory of matter which consists of the following statements:

(i) All matter is composed of extremely small particles called atoms which retain their
identity in chemical reaction.
(ii) The atoms of any pure substance (element) are alike (on the average, at least) in
mass and other physical properties.

79
Chapter 6 Structure of the Atom 80

(iii) Compounds are formed by the combination of atoms of the constituent elements, in
simple numerical proportions.

The atomic theory of matter has been fully confirmed by countless experiments. So we
have no doubt that all matter is made up of atoms.

Now we know the relations between atoms and matter, but what are the properties of
atoms? The basic properties of atoms are summarized as follows.

1. Atoms are very small, about 0.1 nm (0.1 × 10−9 m) in radius. We cannot “see” an
atom even using a high resolution optical microscope.

2. Atoms are stable — they do not spontaneously break apart into smaller pieces or
collapse; thus the internal forces holding the atom together must be in equilibrium.

3. Atoms contain negatively charged electrons, but are electrically neutral. We can easily
observe that bulk matter is electrically neutral, and we assume that this is likewise
a property of the atoms. Experiments with beams of individual atoms support this
assumption. From these experimental facts we deduce that an atom with Z negatively
charged electrons must also contain a net positive charge of Ze.

4. Atoms emit and absorb electromagnetic radiation. Systematic studies showed that
each element emits and absorbs its own characteristic line spectrum consisting of
a unique set of discrete frequencies. Any successful theory of atomic structure must
be able to account for these atomic spectral lines.

6.2 The Rutherford Model of Atom

An early model of the structure of atom was proposed by J. J. Thomson (1856-1940) in


1898. In his model, atoms are just positively charged lumps of matter with negatively
charged electrons embedded in them, like raisin in a fruitcake (see Fig. 6.1). The positively
charged matter is assumed to be uniformly charged, and its mass is essentially the mass
of an atom. Because Thomson had played an important role in the discovery of electron,
his idea was taken seriously. But the real atom turned out to be quite different.

To study the structure of atom, Hans Geiger (1882-1945) and Ernest Marsden (1889-
1970) performed scattering experiments of alpha particles in the laboratory of Ernest
Rutherford (1871-1937) in 1910. As shown in Fig. 6.2, a narrow beam of alpha particles
Chapter 6 Structure of the Atom 81

Electron

Positively charged matter

Figure 6.1: The Thomson model of the atom.

emitted from a radioactive sample is directed to strike at a thin gold foil. A screen was
set on the other side to observe the flashes produced by the alpha particles struck on it.

Figure 6.2: The Rutherford scattering experiment.

According to the Thomson model, the probability for the alpha particles scattering
at large angles should be extremely small. When a positively charged particle passes
through an atom, it would be deflected from its original trajectory by the electrical forces
exerted by the atom. These forces are (1) a repulsive force due to the positive charge
of the atom, and (2) an attractive force due to the negatively charged electrons. An
alpha particle (with mass ≈ 8000 electron masses) is much heavier than an electron and
yet much lighter than the gold atom. We thus only need to consider the positive charge
of the gold atom as the cause of the deflection of the alpha particle that pass through
the gold foil. In the Thomson model, the positive charge of an atom is assumed to be
Chapter 6 Structure of the Atom 82

uniformly spread over its volume. It implies that only a weak electrical force exerted on
the alpha particles when they pass through the gold atoms. So the alpha particles ought
to be deflected only slightly, 1◦ or less.

However, Geiger and Marsden found that although most of the alpha particles were not
deflected much, a few were scattered through very large angles. Some were even scattered
in the backward direction. Their experimental result has a remarkable difference from the
prediction of the Thomson model. Thus the Rutherford scattering experiment showed
the Thomson model to be incorrect.

Figure 6.3: The Rutherford model of the atom.

To explain the results of the scattering experiment, Rutherford proposed that the
positive charge and mass of the atom were concentrated at its center in a region called
the nucleus, with the electrons some distance away (see Fig. 6.3). With an atom being
largely empty space, we can see why most alpha particles go right through a thin foil.
However, when an alpha particle come near a nucleus, the intense electric field there
scatters it through a large angle as illustrated in Figure 6.4.

Figure 6.4: Scattering by a nuclear atom.

During the scattering, the nucleus of charge +Ze exerts a repulsive force:
(ze)(Ze)
F = (6.1)
4π0 r 2
Chapter 6 Structure of the Atom 83

on the projectile of charge +ze. (We assume the projectile is always outside the nucleus
and feels the full nuclear charge.) The atomic electrons, with their small mass, do not
appreciably affect the path of the projectile and so we neglect their effect on the scattering.
We also assume that the nucleus is much more massive than the projectile that it does
not move during the scattering process; since no recoil motion is given to the nucleus, the
initial and final kinetic energies K of the projectile are the same.

Figure 6.5: Scattering by a nuclear atom for different impact parameter b.

As Figure 6.5 shows, for each impact parameter b, there is a corresponding scattering
angle θ, and we want to find how b varies with θ. The projectile can be shown to follow
a hyperbolic path. In polar coordinates r and φ, the equation of the hyperbola is:

1 1 zZe2
= sin φ + (cos φ − 1) (6.2)
r b 8π0 b2 K
Since the final position is φ = π − θ as r → ∞, the above expression reduces to

zZe2 θ
b= cot (6.3)
8π0 K 2
This formula implies that projectiles approaching with smaller impact parameter will be
scattered through larger angles, as shown in Fig. 6.5.

The Fraction of Projectiles Scattered at Angles Greater than θ


From Figure 6.4, we can see that every projectile with impact parameters less than
a given value of b will be scattered at angles greater than its corresponding angle θ.
Suppose the foil were one atom thick — a single layer of atoms packed tightly together,
as in Figure 6.6. Each atom is represented by a circular disc of area πR 2 . For scattering
at angles greater than θ, the projectile must approach the atom within a circular disc
Chapter 6 Structure of the Atom 84

of area πb2 . If the projectiles are spread uniformly over the area of the disc, then the
fraction of projectiles that fall within that area is just πb2 /πR2 .

Figure 6.6: Scattering geometry for many atoms.

A real scattering foil may be of atoms thousands or ten of thousands thick. Let t and
A be the thickness and area of the foil; and let ρ and M be the density and molar mass
of the material of which the foil is made. Then the number of atoms or nuclei per unit
volume is: 
 
ρAt 1 ρNA
n = NA = (6.4)
M At M
where NA is the Avogadro’s number (the number of mole per unit volume). As seen by
an incident projectile, each nucleus contributes an area
 −1
1 ρNA t
=
nt M
to the field of view of the projectile. Assume that the incident particles are spread
uniformly over the area of the foil. The fraction of projectiles scattered at angles greater
than θ is just the fraction that approaches an atom within the area πb2 :

f<b = f>θ = ntπb2 (6.5)

The Rutherford Scattering Formula and Its Experimental Verifi-


cation
For a particle being scattered into a small angular range dθ at angle θ, we require that
the impact parameter lie within a small range of values db at b. The fraction of particle
being scattered in such angular range is then

df = nt(2πb db)
Chapter 6 Structure of the Atom 85

from Equation (6.5). By differentiating Equation (6.3), we find db in terms of dθ:


  
zZ e2 2 θ 1
db = − csc dθ (6.6)
2K 4π0 2 2
 2  2 2
zZ e θ θ
∴ |df | = πnt csc2 cot dθ (6.7)
2K 4π0 2 2

Figure 6.7: Particles entering with impact parameter between b and b + db are distributed
uniformly along a ring of width dθ.
Suppose we place a detector for the scattered projectiles at the angle θ and a distance
r from the nucleus as shown in Fig. 6.7. In order to calculate the rate at which the
projectiles are scattered into the detector, we must know the probability per unit area
for them to be scattered into the ring. We call this N (θ), and it equals to |df |/dA where
dA = (2πr sin θ)rdθ is the area of the ring. After some manipulation, we find:
 2  2 2
nt zZ e 1
N (θ) = 2 4 (6.8)
4r 2K 4π0 sin (θ/2)
This is the Rutherford scattering formula.

In Rutherford’s laboratory, Geiger and Marsden tested the predictions of this formula
in a series of experiments. They used alpha particles to scatter from nuclei in a variety
of thin metal foils. Four predictions of the Rutherford scattering formula were tested:

(a) N (θ) ∝ K −2 . To test this prediction, Geiger and Marsden varied the speed of the
alpha particles while keeping the thickness of the scattering foil fixed. They do so
by slowing down the alpha particles emitted from the radioactive source using a thin
sheet of mica. From independent measurements, they knew the effect of different
thickness of mica on the speed of the alpha particles. Their experimental results are
once again in excellent agreement with the expected relationship (see Fig. 6.8).
Chapter 6 Structure of the Atom 86

Figure 6.8: The dependence of scattering rate on the kinetic energy of the incident alpha
particles for scattering by a given foil.

(b) N (θ) ∝ Z 2 . In this experiment, Geiger and Marsden performed scattering for a
variety of different scattering materials with approximately the same thickness. As
shown in Fig. 6.9, the results are consistent with the proportionality of N (θ) to Z 2 .

Figure 6.9: The dependence of scattering rate on the nuclear charge Z for foils made of
different materials.

(c) N (θ) ∝ t. With a source of 8 MeV alpha particles from radioactive decay, Geiger and
Marsden used scattering foils of varying thickness t while keeping the scattering angle
Chapter 6 Structure of the Atom 87

fixed at about 25◦ . The linear dependence is apparent in their results as shown in
Fig. 6.10. It also shows that, even at this moderate scattering angle, single scattering
is much more important than multiple scattering.

Figure 6.10: The dependence of scattering rate on foil thickness for different scattering
foils.

(d) N (θ) ∝ sin−4 (θ/2). Using a gold foil, Geiger and Marsden varied θ from 5◦ to 150◦
to obtain the relationship between N and θ plotted in Figure 6.11. The agreement
with the Rutherford formula is again very good.

Figure 6.11: The dependence of scattering rate on the scattering angle for gold foil.
Chapter 6 Structure of the Atom 88

Thus all predictions of the Rutherford scattering formula were confirmed by experi-
ment, and the “nuclear atom” was verified. Rutherford is credited with the “discovery”
of the nucleus.

The Closest Approach of a Projectile to the Nucleus


In the derivation of Equation (6.8), Rutherford assumed that the size of the target
nucleus is small compared with the minimum distance d to which incident alpha particles
approach the nucleus before being deflected away. Rutherford scattering therefore gives
us a way to find an upper limit to nuclear dimensions.

Let us calculate the distance of closest approach d for the scattering of alpha
particles as follows. An alpha particle will get closest to the atom when it approaches a
nucleus head on, which will be followed by a 180◦ scattering. At the instant of closest
approach, the initial kinetic energy of the particle is entirely converted to electric potential
energy. So at that instant the kinetic energy is equal to

1 zZe2
K= (6.9)
4π0 d
The distance of closest approach is thus given by

e2 zZ
d= (6.10)
4π0 K

The maximum kinetic energy found in alpha particles of natural origin is about
8.0 MeV. The distance of closest approach for such an particle incident on a gold foil
(Z = 79) is equal to

(2)(79)
d(Au) = (1.44 eV · nm) = 2.8 × 10−14 m (6.11)
(8.0 × 106 eV)

The radius of the gold nucleus is therefore less than 3.0 × 10−14 m, well under the radius
of the atom (10−10 m) as a whole.

Note that the distance of closest approach can be less than the nuclear radius if we
increase the kinetic energy of the projectile, or decrease the electrostatic repulsion by
using a target nucleus with low Z. When this happens, the projectile no longer feels the
full nuclear charge, and the Rutherford scattering formula no longer holds.
Chapter 6 Structure of the Atom 89

6.3 Line Spectra

The radiation from atoms can be classified into continuous spectra as well as discrete
or line spectra. In a continuous spectrum, all wavelengths from the minimum to the
maximum are emitted. The radiation from a hot glowing object is an example of this
category. On the other hand, if an atomic gas or vapor is suitably “excited”, usually by
forcing an electric discharge in a tube containing a small amount of the gas or vapor,
light is emitted at a few discrete wavelengths only. Such line spectrum is known as the
emission line spectrum. An example of this kind of spectra is shown in Figure 6.12.

Figure 6.12: Emission line spectra of hydrogen, helium and mercury in the visible region.

Another possible experiment is to pass a beam of white light (containing all wave-
lengths) through a sample of gas. The gas would absorb light of certain wavelengths seen
in the emission spectrum. Therefore, we will see some dark lines on a bright background
in the resulting spectrum. Such line spectrum is called the absorption line spectrum
(Fig. 6.13).

In general, the interpretation of line spectra is very difficult in complex atoms, and
so we now deal with the spectra of the simplest atom, hydrogen. Regularities appear in
both the emission and absorption spectra of hydrogen atom, as shown in Figure 6.14.
Chapter 6 Structure of the Atom 90

Figure 6.13: The dark lines in the absorption line spectrum of an element corresponds to
bright lines in its emission spectrum.

Figure 6.14: Emission and absorption spectral series of hydrogen. The lines get closer
together as the limit of each series (dashed line) is approached. Only the Lyman series
appears in the absorption spectrum; all series are present in the emission spectrum.
Chapter 6 Structure of the Atom 91

In the absence of a theory that explains the data, Johannes Balmer (1825-1898) noticed
that the wavelengths of the group of emission lines of hydrogen in the visible region can
be very accurately calculated from the empirical formula:

n2
λ = 364.5 (6.12)
n2 − 4
where λ is in units of nm and n can take on integer values beginning from 3. This formula
is now known as the Balmer formula and the series of lines that it fits is called the
Balmer series. The wavelength 364.5 nm, corresponding to n → ∞, is called the series
limit. It was soon discovered that all of the groupings of lines can be fit with a similar
formula of the form
n2
λ = λlimit (6.13)
n2 − n20
where λlimit is the wavelength of the appropriate series limit and where n takes integer
value beginning from n0 + 1. For the Balmer series, n0 = 2. The other series are today
known as Lyman (n0 = 1), Paschen (n0 = 3), Brackett (n0 = 4), and Pfund (n0 = 5).

Any successful model of the hydrogen atom must be able to explain the occurrence of
these interesting arithmetic regularities in the emission spectra.

6.4 The Bohr Model

Following Rutherford’s proposal that the mass and positive charge are concentrated in
a very small region at the center of the atom, the Danish physicist Niels Bohr in 1913
suggested that the atom was in fact like a miniature planetary system, with the elec-
trons circulating about the nucleus like planets circulating about the Sun. The atom
thus doesn’t collapse under the influence of the electrostatic Coulomb attraction between
nucleus and electrons. It’s the attractive force providing the centripetal acceleration nec-
essary to maintain the orbital motion.

For simplicity, we consider the hydrogen atom, with a single electron circulating about
a nucleus with a single positive charge, as shown in Figure 6.14. An electron of mass m
moves with constant tangential speed v in a circular orbit of radius r. The attractive
Coulomb force provides the centripetal acceleration. So:

1 |q1 q2 | 1 e2 mv 2
F = = = (6.14)
4π0 r 2 4π0 r 2 r
Chapter 6 Structure of the Atom 92

Figure 6.15: The Bohr model of the atom (Z = 1 for hydrogen).

Thus the kinetic energy of the electron is


1 1 e2
K = mv 2 = (6.15)
2 8π0 r
The potential energy of the electron-nucleus system is the Coulomb potential energy:
1 e2
U =− (6.16)
4π0 r
Assuming the nucleus to be at rest, the total energy of the electron is given by
1 e2
E =K +U =− (6.17)
8π0 r

We have ignored one serious difficulty with this model so far. Classical physics requires
that an accelerated electric charge must continuously radiate electromagnetic energy. As
it radiates this energy, its total energy decreases. Thus an orbiting electron would spiral
toward the nucleus leading to collapse of the atom. According to classical physics, an
electron cannot revolve around the nucleus in a stable orbit.

To overcome this difficulty, Bohr proposed that there are certain special state of mo-
tion, called stationary states, in which the electron may exist without radiating elec-
tromagnetic energy. According to Bohr, the angular momentum L of the electron in
stationary states takes values that are integer multiples of ~, i. e. L = ~, 2~, 3~, . . .. This
is called the quantization of angular momentum.

For an electron in a circular orbit, its angular momentum has magnitude L = rp =


mvr. Thus Bohr’s postulate is
mvr = n~ (6.18)
Chapter 6 Structure of the Atom 93

where n is any positive integer. Using this expression with equation (6.15), we obtain
 2
1 1 n~ 1 e2
K = mv 2 = m = (6.19)
2 2 mr 8π0 r

Then we find a series of allowed values of the orbit radius r:


4π0 ~2 2
rn = n = a 0 n2 (6.20)
me2
where the Bohr radius a0 is defined as

4π0 ~2
a0 = = 0.0529 nm (6.21)
me2
This result is very different from what we would expect in classical physics. For example,
a satellite may revolve around the Earth at any desired radius by proper operation. But
this is not true for an electron’s orbit — only certain radii are allowed by the Bohr model.

Combining our expression of r with equation (6.17), we can find the electron’s energy
in the permitted orbit of radius rn inside a hydrogen atom:

me4 1 −13.6 eV
En = − 2 2 2 2
= (6.22)
32π 0 ~ n n2

Figure 6.16: The energy levels of atomic hydrogen.

The energies specified by the above equation are called the energy levels of the hydrogen
atom, which are indicated schematically in Figure 6.16. The electron’s energy is quantized
— only certain energy values are possible. In its lowest energy level, with n = 1, the
Chapter 6 Structure of the Atom 94

electron has energy E1 = −13.6 eV and orbits with a radius of r1 = 0.0529 nm. This
state is the ground state. The higher energy states are the excited states.

An electron is nearly free when it is separated from the nucleus by an infinite distance,
corresponding to n = ∞ and E ≈ 0. To remove an electron in an excited state n from
the atom, we need to supply an energy |En | to the electron. This energy is known as the
binding energy of the state n. If we supply more energy than |En | to the electron, the
excess energy will appear as kinetic energy of the “escaped” electron. On the other hand,
we can “excite” a ground state electron to jump to the excited state n by supplying the
so-called excitation energy to the electron. The excitation energy of the state n is just
the energy above the ground state, En − E1 .

Figure 6.17: An electron emits a photon as it jumps from a higher energy state to a lower
energy state.

To test the validity of the Bohr model, let us examine whether this model can explain
the features of the atomic line spectra. Bohr postulated that, even though the electron
doesn’t radiate when it remains in any particular stationary state, it can emit radiation
when it moves to a lower energy level. In this lower level, the electron has less energy
than before, and the energy difference appears as a quantum of radiation whose energy
hf is equal to the energy difference between these levels. That is, if the electron jumps
from n = n1 to n = n2 (see Figure 6.17), a photon would appear with energy
 
me4 1 1
hf = En1 − En2 ⇒ f = − (6.23)
64π 3 20 ~3 n22 n21
The wavelength of the emitted radiation is
 2 2   2 2 
c 64π 3 20 ~3 c n1 n2 1 n1 n2
λ= = 2 2
= (6.24)
f me 4 n1 − n 2 R∞ n1 − n22
2
Chapter 6 Structure of the Atom 95

The parameter R∞ is called the Rydberg constant which is defined by:

me4
R∞ = (6.25)
64π 3 20 ~3 c

and its presently accepted numerical value is 1.097 × 107 m−1 .

These wavelengths are remarkably close to the values of the two longest wavelengths
of the Balmer series. In fact, Equation (6.24) gives
 
n21
λ = (364.5 nm)
n21 − 4

for the wavelength of transitions from any state n1 to n2 = 2. This is identical with
Equation (6.12) for the Balmer series. Thus we see that the radiations identified as the
Balmer series correspond to transitions from higher levels to the n = 2 level. Similar
identifications can be made for other series of radiations, as shown in Figure 6.18.

Figure 6.18: Energies and spectral lines of hydrogen atom.


Chapter 6 Structure of the Atom 96

The Bohr theory for hydrogen atom can be extended to describe any single-electron
atom, even though the nuclear charge Z may be greater than 1. For example, we can
calculate the energy levels of singly ionized helium, doubly ionized lithium, etc. Note
that the nuclear charge enters the Bohr theory in only one place — in the expression for
the electrostatic force between nucleus and electron. We can show that the orbits in the
higher-Z atoms are closer to the nucleus and have larger (negative) energies. That is to
say, the electron is more tightly bound to the nucleus in these atoms.

Deficiencies of the Bohr Model


The Bohr model gives us a picture of how electrons move around the nucleus, and
many of our attempts to explain the behavior of atoms refer to this picture, even though
it is not strictly correct.

Actually, the good agreement between our calculated values of the Balmer series wave-
lengths and the experimental values is just by chance, because we made two errors. The
first error results from our neglect of the motion of the nucleus which has finite mass. The
electron does not orbit around the nucleus, but instead the electron and nucleus both orbit
about their common center of mass. The second mistake we made was in converting the
frequencies into wavelengths. To find the wavelength, we use the expression c = f λ, which
is correct only in vacuum. However, the experiment is usually performed in laboratories
in air. It can be shown that our two mistakes offset one another to some extent.

In spite of its successes, the Bohr model is at best an incomplete model. It is useful
only for atoms that contain one electron since we have considered only the force between
electron and nucleus, not the force between electrons themselves. Furthermore, if we
look very carefully at the emission spectrum, we find that many lines are in fact not
single lines, but very closely spaced combinations of two or more lines; the Bohr model
is unable to account for these doublets of spectral lines. The model is also limited in
its usefulness as a basis from which to calculate other properties of the atom. Although
we can accurately calculate the energies of the spectral lines, we cannot calculate their
intensities. A complete theory should provide a way to calculate this property.

A serious defect in the Bohr model is that it gives incorrect predictions for the angular
momentum of the electron. For the ground state of hydrogen (n = 1), the Bohr theory
gives L = ~ while experiment clearly shows L = 0.

A more serious deficiency of the model is that it completely violates the uncertainty
Chapter 6 Structure of the Atom 97

principle. The uncertainty relationship ∆x∆px ≥ ~ is valid for any direction in space. If
we choose the radial direction, then ∆r∆pr ≥ ~. For an electron moving in a circular orbit,
we know the value of r and pr exactly, so ∆r = 0 and also ∆pr = 0. This simultaneous
exact knowledge of r and pr violates the uncertainty principle.

However, we do not wish to discard the Bohr model completely. The Bohr model gives
us a mental picture for the structure of atom that is sometimes useful. Indeed, there are
many atomic properties can be understood on the basis of Bohr orbits. Moreover, we
will find that the energy levels of hydrogen atom calculated by solving the Schrödinger
equation are in fact identical with those predicted by the Bohr model.

6.5 The Frank-Hertz Experiment

Let us imagine the following experiment performed with the apparatus shown schemati-
cally in Figure 6.19. The electrons emitted from the cathode C are accelerated toward the
grid by the potential difference V . Electrons pass through the grid and reach the plate if
V exceeds V0 , a small retarding potential between the plate and the grid. The current of
electrons reaching the plate is measured using the ammeter A.

Figure 6.19: Frank-Hertz apparatus.

Now suppose the tube is filled with atomic hydrogen gas at a low pressure. As the
voltage V is increased from zero, more and more electrons reach the plate, and the current
rises accordingly. The electrons inside the tube may collide with hydrogen atoms, but lose
no energy in these collisions — the collisions are perfectly elastic. The only way the
Chapter 6 Structure of the Atom 98

electron can give up energy in a collision is if the electron has enough energy to cause
the hydrogen atom to make a transition to an excited state. Thus, when the energy of
the electrons reaches and barely exceeds 10.2 eV (= E2 − E1 ), the electrons can make
inelastic collisions, leaving 10.2 eV of energy with the atom (now in the n = 2 level),
and the original electron moves off with very little energy. When it reaches the grid, the
electron may not have sufficient energy to overcome the small retarding potential and
reach the plate. Thus when V = 10.2 V, a drop in the current is observed. As V is
increased further, we begin to see the effects of multiple collisions. Thus, if a drop in the
current is observed at V , a similar drops are observed at 2V , 3V , . . ..

Figure 6.20: Results of Frank-Hertz experiment using mercury vapor.

This experiment should thus give rather direct evidence for the existence of discrete
energy levels in atoms. Unfortunately, it is not easy to do this experiment with hydrogen,
because hydrogen occurs naturally in the molecular form of H2 , rather than in atomic
form. The molecules can absorb energy in a variety of ways, which would confuse the
interpretation of the experiment. A similar experiment was done in 1914 by James Franck
(1882-1964) and Gustav L. Hertz (1887-1975), using a tube filled with mercury vapor.
Their results are shown in Figure 6.20. It shows current drops at voltages of 4.9 V, 9.8 V
and 14.7 V, which is a clear evidence for an excited state at 4.9 eV. Coincidentally, the
emission spectrum of mercury indicates transition of electrons between the same 4.9 eV
excited state and the ground state. The Frank-Hertz experiment showed that an electron
must have a certain minimum energy, which is the energy of an excited state of an atom,
to make an inelastic collision with an atom. For their direct experimental confirmation of
Bohr’s postulate of discrete energy levels in atoms, Franck and Hertz were awarded the
1925 Nobel prize in physics.
Chapter 7

The Hydrogen Atom

In this chapter, we study the solutions of the Schrödinger equation for the hydrogen atom.
These solutions differ from the Bohr model by allowing for the uncertainty in localizing
electron. But they lead to the same energy levels as those predicted by the Bohr model.

Other deficiencies of the Bohr model are not so easily eliminated by solving the
Schrödinger equation. First, the so-called “fine structure” splitting of the spectral lines,
which appears when we examine the lines very carefully, cannot be explained by our so-
lutions. Second, solving the Schrödinger equation for atoms with two or more electrons
is difficult, so we restrict our discussion in this chapter to one-electron atoms. We will
discuss the structure of many-electron atoms in the next chapter.

7.1 The Schrödinger Equation for the Hydrogen Atom

A hydrogen atom consists of a proton, a particle of charge +e, and an electron, a particle
of charge −e. The atomic electron is bound to the nucleus due to the attractive Coulomb
force between opposite charges. Because an electron is 1836 times lighter than a proton,
we take the approximation that the proton is stationary, with the electron moving around
but prevented from escaping by the electric field of the nucleus. Thus the problem is
simplified from one involving an electron and a proton moving around each other to one
involving only an electron moving under the influence of the Coulomb electrostatic force.
As in the Bohr theory, the correction for the motion of the proton is simply a matter of
replacing the electron mass m by the reduced mass µ = mM/(m + M ) where M is the
mass of a proton.

99
Chapter 7 The Hydrogen Atom 100

For convenience, we set up the coordinate system with the proton at the origin. So
we consider an electron moving under the influence of the Coulomb potential:

1 e2 e2
U (x, y, z) = − p =− (7.1)
4π0 x2 + y 2 + z 2 4π0 r
p
where x, y, z are the coordinates of the electron, and r = x2 + y 2 + z 2 is the distance
of the electron from the proton. Because the potential energy U depends only on r,
the force is necessarily along the radial direction and is thus referred to as central force.
We again restrict our attention to stationary states in which the energy is well-defined
and time-independent. The stationary-state wave functions have the form Ψ(x, y, z, t) =
ψ(x, y, z)φ(t). The spatial part ψ(x, y, z) is a solution to the time-independent Schrödinger
equation:
~
− ∇2 ψ(x, y, z) + U (x, y, z) ψ(x, y, z) = Eψ(x, y, z) (7.2)
2m
∂2 ∂2 ∂2
where the Laplacian ∇2 = 2 + 2 + 2 .
∂x ∂y ∂z

Figure 7.1: Spherical polar coordinates r, θ, φ.

Solving (7.2) in Cartesian coordinates is very difficult due to the dependence of the
potential energy U on r. To take advantage of this radial symmetry, we transform the
Cartesian coordinates (x, y, z) to the spherical polar coordinates (r, θ, φ) (see Figure 7.1).
Chapter 7 The Hydrogen Atom 101

In spherical coordinates, the Laplacian is:


     
2 1 ∂ 2 ∂ ∂ ∂ 2 ∂2
∇ = 2 r + csc θ sin θ + csc θ 2 ,
r ∂r ∂r ∂θ ∂θ ∂φ
Putting this expression into the Schrödinger equation, we obtain
     
~2 1 ∂ 2 ∂ ∂ ∂ 2 ∂2
− r + csc θ sin θ + csc θ 2 ψ(r, θ, ψ)
2m r 2 ∂r ∂r ∂θ ∂θ ∂φ
+ U (r)ψ(r, θ, ψ) = Eψ(r, θ, ψ) (7.3)

e2
where U = − .
4π0 r
Note that (7.3) is the time-independent Schrödinger equation for any time-dependent
potential energy U that depends only on r, i.e. for any conservative central force. Besides,
it is a separable partial differential equation which can be solved by the separation of
variables. We will discuss this in details in the next section.

7.2 Solving the TISE for the Hydrogen Atom

Here, we solve the TISE for the hydrogen atom (Eqn. (7.3)) by the separation of variables.
We consider a solution which is a product of three functions, each depending only on one
of the coordinates r, θ, φ. In other words, we consider a solution of the form:

ψ(r, θ, φ) = R(r)Θ(θ)Φ(φ) (7.4)

Substituting this solution into Eq. (7.3) yields three ordinary differential equations6 :
   
1 d 2 dR(r) 2m ~2 C
r + 2 E − U (r) + R(r) = 0 (7.5)
r 2 dr dr ~ 2mr 2
 
d dΘ(θ)
sin θ sin θ − (C sin2 θ + D)Θ(θ) = 0 (7.6)
dθ dθ
d2 Φ(φ)
= −DΦ(φ) (7.7)
dφ2
where C and D are the separation constants.

Let us study these ordinary differential equations one by one. Equation (7.7) is known
as the azimuthal equation. Its solution Φ(φ) is sinusoidal if D > 0 or exponential if
D < 0. For a physically meaningful Φ(φ), it must meets itself smoothly as the azimuthal
6
Read appendix C for the derivation of these ordinary differential equations
Chapter 7 The Hydrogen Atom 102

angle φ goes from a given value through 2π about the z-axis, i. e. Φ(φ) = Φ(φ + 2π) for
any φ. It can be shown that exponential solution cannot satisfy this restriction. So we
only consider the case that Φ(φ) is sinusoidal which implies D is positive. For such case,
a solution of (7.7) is

Φ(φ) = ei Dφ


If this function meets itself smoothly when φ changes by 2π, then D must be an integer.
We choose the symbol m` for this integer and so we have

Φ(φ) = eim` φ m` = 0, ±1, ±2, ±3, . . . (7.8)

where the integer m` is called the magnetic quantum number. Note that it is the
smoothness condition that gives us the magnetic quantum number m` associated with
the φ coordinate.

Next we consider equation (7.6) which governs polar motion and is called the polar
equation. Putting in D = m2` , we rewrite the equation in terms of z = cos θ as
   
d 2 dΘ m2`
(1 − z ) − C+ Θ=0
dz dz 1 − z2
It is the well-known associated Legendre equation. The solutions Θ(θ) are polynomi-
als in cos θ known as the associated Legendre polynomials. To obtain a physically
acceptable solution, Θ(θ) must be smooth and not diverging at the limits θ = 0 or θ = π
or both. Such conditions are met only if

C = −`(` + 1) ` = 0, 1, 2, . . .
(7.9)
m` = 0, ±1, ±2, . . . , ±` ≤ `

where the integer ` is called the orbital quantum number. A few of the associated
Legendre polynomials satisfying the physical requirements are listed in Table 7.1. Notice
that the products Θ(θ)Φ(φ) specifies the full angular dependence of the wave function
ψ(r, θ, ψ), and are known as spherical harmonics denoted as Y`m` (θ, φ).

Equation (7.5) is the radial equation which determines the radial part R(r) of the
wave function ψ. Inserting C = −`(` + 1) and putting ρ = αr, the radial equation can be
rewritten as:
   
1 d 2 dR(ρ) 2m `(` + 1)
ρ + [E − U (ρ)] − R(ρ) = 0
ρ2 dρ dρ 2
~ α 2 ρ2
To find R(ρ), we need to solve another well-known differential equation known as the
associated Laguerre equation. It’s too tedious to solve this equation here. Instead,
Chapter 7 The Hydrogen Atom 103

Table 7.1: Some Associated Legendre Polynomials P`m` (cos θ).


P00 = 1 P22 = sin2 θ
P10 = 2 cos θ P30 = 24(5 cos3 2θ − 3 cos θ)
P11 = sin θ P31 = 6 sin θ(5 cos2 θ − 1)
P20 = 4(3 cos2 θ − 1) P32 = 6 sin2 θ cos θ
P21 = 4 sin θ cos θ P33 = sin3 θ

we shall only summarize the key results. The solution of the radial wave equation R(r)
requires the introduction of another quantum number called the principal quantum
number n. For the solution to be physically acceptable, n must be a positive integer and
the orbital quantum number ` must obey the condition:

` = 0, 1, 2, . . . n − 1 (7.10)

In summary, we found that three quantum numbers (n, `, m` ) are need to spceify the
solutions of the Schrödinger equation for the hydrogen atom. Their allowed values are:

principal quantum number n 1, 2, 3, . . .


orbital quantum number ` 0, 1, 2, . . . , n − 1
magnetic quantum number m` −`, −(` − 1), . . . , −1, 0, 1, . . . , (` − 1), `

Example 7-1

What are the possible sets of quantum numbers with n = 4 in a hydrogen atom?

Solution:
The possible values of ` are ` = 0, 1, 2, 3 because `max = n − 1. For each value of `, m`
goes from −` to +`.

n ` m`
4 0 0
4 1 −1, 0, 1
4 2 −2, −1, 0, 1, 2
4 3 −3, −2, −1, 0, 1, 2, 3

Each set of quantum numbers (n, `, m` ) corresponds to a very different wave function,
and therefore represents a very different state of motion of the electron in the hydrogen
Chapter 7 The Hydrogen Atom 104

atom. To show the dependence of R, Θ and Φ upon the quantum numbers (n, `, m` ), we
write the separated solution of the time-independent Schrödinger equation (Eq. (7.3)) as
ψn,`,m` (r, θ, φ) = Rn,` (r)Θ`,m` (θ)Φm` (φ) (7.11)
Table 7.2 lists some of these wave functions expressed in terms of the Bohr radius:
4π0 ~2
a0 = = 0.0529 nm
me2

Table 7.2: Normalized Wave functions of the Hydrogen Atom for n = 1, 2, and 3.

n ` m` R(r) Θ(θ) Φ(φ)


2 1 1
1 0 0 3/2
e−r/a0 √ √
a0 2 2π
 
1 r 1 1
2 0 0 √ 3/2
2− e−r/2a0 √ √
2 2 a0 a0 2 2π

1 r −r/2a0 6 1
2 1 0 √ 3/2
e cos θ √
2 6 a 0 a0 √
2 2π
1 r 3 1
2 1 ±1 √ 3/2 e−r/2a0 sin θ √ e±iφ
2 6 a 0 a0 2 2π
 
2 r r2 1 1
3 0 0 √ 3/2 27 − 18 + 2 2 e−r/3a0 √ √
81 3 a0  a0 a0 2 2π
 √
4 r r −r/3a0 6 1
3 1 0 √ 3/2 6 − e cos θ √
81 6 a0  a0 a0 2 2π
 √
4 r r −r/3a0 3 1
3 1 ±1 √ 3/2 6 − e sin θ √ e±iφ
81 6 a0 a0 a0 2 2π

4 r 2 −r/3a0 10 1
3 2 0 √ 3/2 2
e (3 cos2 θ − 1) √
81 30 a0 a0 √
4 2π
4 r 2 −r/3a0 15 1
3 2 ±1 √ 3/2 2
e sin θ cos θ √ e±iφ
81 30 a0 a0 √
2 2π
4 r 2 −r/3a0 15 2 1
3 2 ±2 √ 3/2 2
e sin θ √ e±2iφ
81 30 a0 a0 4 2π

7.3 Quantization of Energy and Angular Momentum

For each quantum number that arises, a physical property is quantized. For the hydrogen
atom, the boundary conditions imposed on the wave function lead to three quantum
numbers and thus three quantized quantities.
Chapter 7 The Hydrogen Atom 105

The principal quantum number n results from the solution of the radial wave function
R(r). Because the radial equation includes the potential energy U (r), it is not surprising
that the boundary condition on R(r) quantizes the energy E. The result for this quantized
energy is
me4 1 13.6 eV
En = − 2 2 2 2
=− n = 1, 2, 3, . . . (7.12)
32π 0 ~ n n2
which is exactly the same formula found by the Bohr’s theory. The negative value of the
energy E indicates that the electron and proton are bound together. Besides, the energy
levels of the hydrogen atom depend on the principal quantum number n only. (It is true
only for atoms with single electron.) So two states with different sets of quantum numbers
(n, `, m` ) and (n0 , `0 , m0` ) would have the same energy if n = n0 . These energy levels are
thus said to be degenerate. For example, the energy for the n = 2 level is the same for
all possible values of ` (` = 0, 1) and m` (m` = 0, ±1, . . . , ±`). We would say the n = 2
level has a degeneracy of 4.

Figure 7.2: Classical orbits with the same energy but different angular momentum L.

The orbital quantum number ` is associated with the functions R(r) and Θ(θ) in the
wave function ψ(r, θ, φ). It determines the orbital angular momentum of the orbiting
electron. The electron-proton system has orbital angular momentum as the particles pass
~ = ~r × p~ or
around each other. Classically, this orbital angular momentum is L

~ = mvorbital r
|L|

~ is perpendicular to the plane of the orbit and


where vorbital is the orbital velocity. L
~ In Figure 7.2, we show several classical orbits
the orbit of the particle depends on |L|.
corresponding to the same orbital energy. These orbits differ in their angular momentum
L, which is the largest for the nearly circular orbit (like the Earth) and the smallest for
the highly elongated elliptical orbit (like the comets). Because there is no torque applied
Chapter 7 The Hydrogen Atom 106

on the hydrogen atom in the absence of external fields, the classical angular momentum
~ is a constant of motion and thus is conserved.
L
~ is related with
In quantum physics, the magnitude of the orbital angular momentum L
the orbital quantum number ` as
p
~ =
|L| `(` + 1)~ ` = 0, 1, 2, . . . , n − 1 (7.13)

The boundary condition on ψ(r, θ, φ) specifies that ` must be a non-negative integer


smaller than n, and therefore the magnitude of L~ is quantized. Note that the orbital
quantum number ` only determines the magnitude of the angular momentum L, ~ but it
~ In addition, we should not
does not contain any information about the direction of L.
take the classical elliptical orbit literally for electrons; only probability functions consistent
with the uncertainty principle can describe the position of the electron.
~ on ` (|L|
The curious dependence of |L| ~ > `~) is a wave phenomenon — it results from
the application of the boundary conditions on ψ(r, θ, φ). We will give the justification for
it later in this section. In addition, the quantum result disagrees with that proposed by
the Bohr model of the hydrogen atom, where |L| ~ = n~. Apparently, we have to discard
Bohr’s semi-classical “planetary” model in which electrons orbit around a nucleus.

The magnetic quantum number m` arises from the solution of the azimuthal equation.
Physical requirements on ψ(r, θ, φ) restrict m` to be an integer with |m` | ≤ `. It can be
shown that m` is related to the z component of the angular momentum L ~ as:

Lz = m ` ~ m` = 0, ±1, ±2, . . . , ±` (7.14)

~ together with Lz in terms of the vector


It is convenient to visualize the quantization of |L|
model depicted in Fig. 7.3. From the figure, we see that the angle θ that L ~ makes with
the z-axis is also quantized. Its possible values are specified by the relation:
Lz m`
cos θ = =p (7.15)
~
|L| `(` + 1)

~ can point in any


Classically, θ can have any value, i. e. the angular momentum vector L
direction in space. However, quantum physics predicts that the possible orientations of
~ are only those consistent with Equation (7.15). The number of these orientations is
L
equal to (2` + 1) and the magnitudes of their successive z components always differ by
~. This phenomenon is called space quantization because only certain orientations of
~ are allowed in space.
L
Chapter 7 The Hydrogen Atom 107

Figure 7.3: (Left) Angular momentum quantization for ` = 1 and ` = 2. (Right) Possible
angles between L~ and z-axis for ` = 1 and ` = 3.

Have we established a preferred direction in space by choosing the z axis? In fact, the
choice of the z axis is completely arbitrary unless there is an external magnetic field to
define a preferred direction in space. It is customary to choose the z axis to be along the
magnetic field B~ if it exists. That’s why m` is called the magnetic quantum number.

Will the angular momentum be quantized along the x and y axes as well? The an-
~ to be quantized along only one direction in space.
swer is that quantum theory allows L
~ the knowledge of a second component would imply
Because we know the magnitude of L,
a knowledge of the third component as well because of the relation L2 = L2x + L2y + L2z .
~
But this would violate the Heisenberg uncertainty principle. If all three components of L
~ would also be known. In this case, we would have a
were known, then the direction of L
precise knowledge of one component of the electron’s position in space, because the elec-
~ But confinement of the
tron’s orbital motion is confined to a plane perpendicular to L.
electron to that plane means that the electron’s momentum component along L ~ is exactly
zero. This simultaneous knowledge of the same component of position and momentum is
forbidden by the uncertainty principle.
~ and Lz may be specified simultaneously. The values of Lx and
Only the magnitude L
Ly should be consistent with the relation L2 = L2x + L2y + L2z but cannot be specified
Chapter 7 The Hydrogen Atom 108

individually. Physicists refer L and Lz as “sharp” because of their known values, and Lx
and Ly as “fuzzy” because of their unknown values. The angular momentum vector L ~
p
~ = `(` + 1)~ > |Lz |max = `~.
never points in the z direction (see Figure 7.3) because |L|

The space quantization just mentioned is an experimental fact. The values of Lz /~


range from −` to +` in steps of 1, for a total of 2` + 1 allowed values. Because there is
nothing special about the three directions x, y, and z, we expect the average of the angular
momentum components squared in the three directions to be the same, i. e. hL2x i = hL2y i =
hL2z i. The average value of hL2 i is equal to three times the average value of the square
of any one components. We choose the z components so that hL2 i = 3hL2z i. To find the
average value of L2z , we just have to sum up L2z for all possible quantum numbers and
then divide by the total number, 2` + 1.
m ` =`
2 3 X
hL i = 3hL2z i = m2 ~2 = `(` + 1)~2 (7.16)
2` + 1 m =−` `
`

This argument explains the `(` + 1) dependence for the expectation value of L2 .

Example 7-2

What is the degeneracy of the n = 3 level of the hydrogen atom? That is, how many
different quantum states are contained in the energy level n = 3?

Solution:
The energy eigenvalues for the hydrogen atom depends only on the principal quantum
number n. For each value of n, there can be n different values of ` (` = 0, 1, . . . , n−1). For
each value of `, there are (2`+1) different values of m` (m` = −`, −(`−1), . . . , 0, 1, . . . , +`).
Therefore, to find the degeneracy for n = 3, we have to sum up all these possibilities.

n ` m` (2` + 1)
3 0 0 1
3 1 −1, 0, 1 3
3 2 −2, −1, 0, 1, 2 5
total = 9

So the degeneracy of the n = 3 level of the hydrogen atom is 9. You may notice that, in
general, the degeneracy is n2 .
Chapter 7 The Hydrogen Atom 109

7.4 Probability Densities

In three dimensions, the squared modulus of the wave function |Ψ(~r, t)|2 gives the prob-
ability per unit volume of finding the particle in the vicinity of the position ~r. Just like
in one dimension, |Ψ(~r, t)|2 = |ψ(~r)|2 for stationary state Ψ(~r, t) = ψ(~r)φ(t). For the case
of a hydrogen atom, the probability density at (r, θ, φ) is given by:

|ψn,`,m` (r, θ, φ)|2 = |Rn,` (r)|2 |Θ`,m` (θ)|2 |Φm` (φ)|2 (7.17)

To compute the actual probability of finding the electron, we multiply the probability
density |ψ|2 by the volume element dV :

dV = r 2 sin θ dr dθ dφ (7.18)

and the result is

|ψn,`,m` (r, θ, φ)|2 dV = |Rn,` (r)|2 |Θ`,m` (θ)|2 |Φm` (φ)|2 r 2 sin θ dr dθ dφ (7.19)

From the probability density |ψ|2 , we can obtain many information about the electron’s
spatial distribution. For example, we can find the radial probability P (r)dr for locating
the electron somewhere between r and r + dr no matter what the values of θ and φ are.
It is equal to the probability of the electron being within the volume of a thin spherical
shell of radius r and thickness dr. So the radial probability is given by:
Z π Z 2π
2 2 2
P (r)dr = |Rn,` (r)| r dr |Θ`,m` l (θ)| sin θdθ |Φm` (φ)|2 dφ (7.20)
0 0
The θ and φ integrals are each equal to unity since the functions R, Θ and Φ are normalized
individually. Thus the radial probability density is

P (r) = r 2 |Rn,` (r)|2 (7.21)

Figure 7.4 shows this function for several lowest energy levels of the hydrogen atom. Note
that all these wave functions extend into the classically forbidden region.

The radial probability density allows us to calculate the expectation value of the
distance of the electron from the nucleus. Just like in the Bohr model, this average
distance increases approximately as the square of principal quantum number n. So an
n = 2 electron is on the average about four times farther from the nucleus than an n = 1
electron, and so forth. Another measure of the electron’s location is its most probable
location where P (r) attains the maximum value. For each n, P (r) for the state with
` = n − 1 has only a single maximum, which occurs at the location of the Bohr orbit,
r = n 2 a0 .
Chapter 7 The Hydrogen Atom 110

Figure 7.4: Radial probability densities P (r) and energies E in the hydrogen atom for
n = 1, 2, and 3.
Chapter 7 The Hydrogen Atom 111

Example 7-3

An electron is in the n = 1, ` = 0 state. What is the probability of finding the electron


closer to the nucleus than the Bohr radius?

Solution:
We are again interested in the radial probability
4r 2 −2r/a0
P (r)dr = r 2 |R1,0 (r)|2 dr = e dr
a30
The total probability of finding the electron between r = 0 and r = a0 is
Z a0 Z
4 a0 2 −2r/a0
P = P (r)dr = 3 r e dr
0 a0 0
By letting x = 2r/a0 , we rewrite the integral as
Z
1 2 2 −x
P = x e dx = 0.323
2 0
That is, 32.3 % of the time the electron is closer than one Bohr radius to the nucleus.

Example 7-4

What is the most likely distance from the origin for an electron in the n = 2, ` = 1 state?

Solution:
In the n = 2, ` = 1 level, the radial probability density is
r 4 −r/a0
P (r) = r 2 |R2,1 (r)|2 = e
24a50
We wish to find where this function has its maximum. In the usual fashion, we take the
first derivative of P (r) and set it equal to zero:
   
dP (r) 1 d 4 −r/a0 1 3 −r/a0 4 1 −r/a0
= (r e )= 4r e +r − e =0
dr 24a50 dr 24a50 a0
 
1 −r/a0 3 r4
⇒ e 4r − =0
24a50 a0
The only solution that yields the maximum is r = 4a0 .

On the other hand, the angular position of the electron is described by the angular
probability density P (θ, φ) which is the angular part of |ψ|2 . It can be shown that

P (θ, φ) = |Θ`,m` (θ)Φ(φ)m` |2 = Θ`,m` (θ)2 (7.22)


Chapter 7 The Hydrogen Atom 112

Figure 7.5: Angular probability densities in the hydrogen atom for ` = 1, 2, and 3.

Figure 7.5 shows the polar plot of the angular probability densities for ` = 1, 2, 3. We
can see that all the probability densities are cylindrically symmetric since they do not
depend on the azimuthal angle φ. Moreover, the probability density for ` = 0 is also
spherically symmetric, i. e. it is independent of direction. For the ` = 1 case, the angular
probability densities have two distinct shapes. For m` = 0, the electron is found primarily
in two regions of maximum probability along the positive and negative z axis, while for
m` = ±1, the electron is found primarily near the xy plane. When m` = 0, the electron’s
angular momentum vector lies in the xy plane. Classically, the angular momentum vector
is perpendicular to the orbital plane and so the electrons should spend most of its time
in a direction perpendicular to the xy plane, i. e. along the z axis. However, the angular
Chapter 7 The Hydrogen Atom 113

momentum vector has its maximum projection along the z axis when m` = ±1. Classical
physics predicts that the electron will spend most of its time near the xy plane.

Figure 7.6: Electron probability densities in the hydrogen atom for n = 1, 2, and 3.
Chapter 7 The Hydrogen Atom 114

Putting together the radial and angular probability densities, we can obtain repre-
sentations of the complete electron probability density |ψ|2 , as shown in Figure 7.6. We
can regard these illustrations as representing the “smeared out” distribution of electronic
charge in the atom, which results from the uncertainty in the electron’s location. They
also represent the statistical outcomes of a large number of measurements of the position
of the electron in the atom. These spatial distributions have important consequences for
the structure of atoms with many electrons, and also for joining atoms into molecules.

7.5 Spectroscopic Notation and Selection Rule

It is customary to specify electron angular momentum by a letter, with s corresponding


to ` = 0, p to ` = 1, and so on, according to the following scheme:

Value of ` 0 1 2 3 4 5 6
Letter s p d f g h i

Thus an s state is the state with no angular momentum, a p state has the angular

momentum 2~, and so forth.

The combination of the principal quantum number n with the letter that represents
the orbital angular momentum provides a convenient and widely used notation for atomic
electronic state, known as spectroscopic notation. In this notation, a state with n = 2,
` = 0 is labeled as 2s state. Fig. 7.7 illustrates the labeling of the hydrogen energy levels
in the spectroscopic notation. Note that the degeneracy of the energy levels with respect
to ` (and also m` ) would be removed in the presence of a magnetic field.

Also shown in Figure 7.7 are lines representing some different photons that can be
emitted when the atom makes a transition from one state to a lower state. We can use
the wave function obtained from solving the Schrödinger equation to calculate the tran-
sition probabilities for the electron to change from one state to another. The results
of such calculations show that electrons absorbing or emitting photons are most likely to
change states when ∆` = ±1. Such transitions are called allowed transitions. Other
transitions, with ∆` 6= ±1, are theoretically possible but occur with much smaller prob-
abilities and are called forbidden transitions. There is no selection rule restricting the
change of the principal quantum number ∆n. The selection rule for the magnetic quantum
Chapter 7 The Hydrogen Atom 115

Figure 7.7: A partial energy level diagram of hydrogen, showing the spectroscopic notation
of the levels and the allowed photon transitions between some levels.

number is ∆m` = 0, ±1. We summarize the selection rules for allowed transitions:

∆n = anything (7.23)
∆` = ±1 (7.24)
∆m` = 0 or ± 1 (7.25)

For example, the transitions of 3p → 2p and 3d → 2s are not allowed transitions because
these transitions violate the ∆` = ±1 selection rule.

7.6 Orbital Magnetism and the Normal Zeeman Ef-


fect

One way to observe the space quantization is to place the atom in an external magnetic
field. From the interaction between the magnetic field and the magnetic dipole moment
~ and also to determine
of the atom, it is possible to observe the separate components of L
` by counting the number of separated components.
Chapter 7 The Hydrogen Atom 116

~ the torque τ acting on a magnetic dipole7


In an uniform external magnetic field B,
with magnetic dipole moment µ
~ is:

~ = µB sin θ
τ = |~µ × B| (7.26)

where θ is the angle between µ ~ For convenience, we set the potential energy U = 0
~ and B.
when θ = π/2, i. e. when µ ~ The potential energy at any other
~ is perpendicular to B.
orientation of µ
~ is equal to the external work must be done to rotate the dipole from
θ0 = π/2 to the angle θ corresponding to that orientation. Hence
Z θ
U= τ dθ = −µB cos θ (7.27)
π/2

When µ ~ U = −µB — its minimum value. This follows


~ points in the same direction as B,
from the fact that a magnetic dipole tends to align itself with an external magnetic field.

Figure 7.8: Magnetic dipole moment µ ~


~ of an orbiting electron of angular momentum L.

The magnetic dipole moment of an orbiting electron in a hydrogen atom depends on


~ We can use the Bohr model with a circular orbit to obtain the
its angular momentum L.
~ and µ
relation between L ~ , which turns out to be the same as the quantum mechanical
results. The magnetic dipole moment of a current loop is

µ = IA

where I is the current and A is the area it encloses. An electron moving with speed
v = p/m around a circular orbit of radius r is equivalent to a current I = dq/dt = q/T
7
Read appendix D for a brief review of magnetic dipole
Chapter 7 The Hydrogen Atom 117

where q is the charge of an electron and T = 2πr/v = 2πrm/p is the period of the orbiting
motion of the electron. So the magnetic dipole moment of the electron is
q 2 q q ~
µ = IA = πr = rp = |L|
T 2m 2m
~ = rp. Putting in the electronic charge q = −e, we obtain
since |L|
e ~
~L = −
µ L (7.28)
2m
~ and µ
The negative sign indicates that the vectors L ~ L point in opposite directions (see
Fig. 7.8). The subscript L on µ
~ L reminds us that this magnetic dipole moment arises
from the orbital angular momentum L ~ of the electron.

~ is perpendicular to L
~L × B
Figure 7.9: The torque µ ~ and thus causes L
~ to precess.

~ that points along the z


Suppose a hydrogen atom is in an external magnetic field B
axis. From (7.28), the atom would experience a torque:
~ = − e (L
~L × B
~τ = µ ~ × B)
~
2m
Chapter 7 The Hydrogen Atom 118

This causes the angular momentum to precess about the magnetic field line, as shown in
Fig. 7.10. From the figure, the rate of precession is given by:

~ = − e L
τ = |~µL × B| ~ ×B ~
2m
(L sin θ)dφ e
⇒ = LB sin θ
dt 2m
dφ eB
⇒ = (7.29)
dt 2m
It implies that while the z-component of µL is constant, its x and y components change
continuously at a rate proportional to the magnetic field strength. Therefore, we cannot
~ along the x and y axes.
observe the quantization of L

According to (7.28), the magnetic potential energy of a hydrogen atom in a magnetic


field is:  e 
U= ~ cos θ
|L|B (7.30)
2m
which depends on both B and θ. Since B~ is assumed to point in the z direction, the angle
~ and the z axis. Therfore, the angle θ can only have
θ is equal to the angle between L
values specified by:
m`
cos θ = p
`(` + 1)
p
~ are given by |L|
since the permitted values of |L| ~ = `(` + 1)~. By putting the above
result into Equation (7.30), the magnetic potential energy of an atom with magnetic
quantum number m` in a magnetic field B~ is found to be:
 
e~
U = m` B (7.31)
2m

The quantity e~/2m is called the Bohr magneton:


e~
µB = = 9.274 × 10−24 J/T
2m
The Bohr magneton is a convenient unit for expressing atomic magnetic dipole moments,
which typically have values of the order of µB .

In a magnetic field, the energy of a particular atomic state of hydrogen not only
depends on the value of n but also on the value of m` since the magnetic potential energy
U ∝ m` . Thus the state of principal quantum number n breaks up into several substates
when the atom is in a magnetic field, and their energies are slightly more or slightly less
than that of the state in the absence of the field. This phenomenon leads to “splitting” of
Chapter 7 The Hydrogen Atom 119

Figure 7.10: The normal Zeeman effect.

individual spectral lines into separate lines when the hydrogen atom radiate in a magnetic
field. The spacing of the lines depends on the magnitude of the magnetic field.

The splitting of spectral lines by a magnetic field is called the Zeeman effect after the
Dutch physicist Pieter Zeeman (1865-1943), who first observed it in 1896. The Zeeman
effect is a vivid confirmation of space quantization. Because m` can have 2` + 1 values
from +` through 0 to −`, a state of given orbital quantum number ` is split into 2` + 1
substates that differ in energy by µB B when the atom is in a magnetic field. However,
because changes in m` are restricted to ∆m` = 0 or ±1, we expect a spectral line from
a transition between two states of different ` to be split into only three components, as
shown in Figure 7.10. The normal Zeeman effect consists of splitting of a spectral line
Chapter 7 The Hydrogen Atom 120

of frequency ν0 into three components whose frequencies are


B e
ν1 = ν 0 − µ B = ν0 − B (7.32)
2π~ 4πm
ν2 = ν 0 (7.33)
B e
ν3 = ν 0 + µ B = ν0 + B (7.34)
2π~ 4πm
In the next chapter, we will see that this is not the whole story of the Zeeman effect.
Chapter 8

Many-Electron Atoms

In atoms with more than one electron, each electron is not only attracted by the nucleus
but also repelled by other electrons. Such problem involving the mutual interactions of
three or more objects is an example of the so-called many-body problem. It is very difficult
to find exact, closed form solutions of the Schrödinger equation for such type of problems.
In this chapter, we consider an approximate set of energy levels for multi-electron atoms,
and we try to understand some of the properties of these atoms based on those energy
levels. We will begin from the study of the electron spin, which plays an important role
in determining the structure of multi-electron atoms.

8.1 Electron Spin

The theory developed in the previous chapter cannot account for a number of well-known
experimental observations. One is the fact that many spectral lines actually consist of
two separate lines that are very close together, which are known as fine structure.
Another failure occurs in the Zeeman effect. In a magnetic field, the spectral lines of an
atom should be split into the three components specified by Eqs. (7.30)-(7.33). While
the normal Zeeman effect is observed in the spectra of a few elements under certain
circumstances, more often it is not. Four, six, or even more components may appear, and
even when three components are present their spacing may not agree with Eqs. (7.30)-
(7.33). Such observations is called the anomalous Zeeman effect (see Figure 8.1).

In order to account for these experimental results, Samuel Goudsmit (1902-1978) and
George E. Uhlenbeck (1900-1988) proposed in 1925 that:

121
Chapter 8 Many-Electron Atoms 122

Figure 8.1: The normal and anomalous Zeeman effects in various spectral lines.

Every electron has an intrinsic angular momentum, called spin, whose magnitude is the
same for all electrons. Associated with this angular momentum is a magnetic moment.

What Goudsmit and Uhlenbeck had in mind was a classical picture of an electron as
a charged sphere spinning on its axis. The rotation involves angular momentum, and
because the electron is negatively charged, it has an intrinsic magnetic dipole moment µ
~s
opposite in direction to its intrinsic angular momentum vector S.~ The notion of electron
spin proved to be successful in explaining not only the fine structure and the anomalous
Zeeman effect but a wide variety of atomic effects as well. However, the classical picture
of an electron as a spinning charged object is not technically correct although it is a useful
concept to visualize electron spin. The fundamental nature of electron spin was confirmed
in 1929 by Paul Dirac’s development of relativistic quantum mechanics.
p
~ = `(` + 1)~, the magnitude of a particle’s intrinsic
Analogous to the relationship |L|
angular momentum |S| ~ is quantized according to the value of the spin quantum number
s8 :
p
~ =
|S| s(s + 1)~ (8.1)

The orbital angular momentum can only point in those directions in space such that its
projection on the z-axis equal Lz = m` ~ for m` = −`, −` + 1, . . . , ` − 1, `. By analogy,
the z-component of the intrinsic angular momentum Sz is also quantized, taking on only
8
In fact, all “elementary particles” possess an intrinsic angular momentum and the value of the spin
quantum number s depends on the kind of particle.
Chapter 8 Many-Electron Atoms 123

2s + 1 possible values specified by:

Sz = m s ~ ms = −s, −s + 1, . . . , s − 1, s (8.2)

where ms is the spin magnetic quantum number.

For the electron, we must have s = 21 , which follows from Dirac’s theory as well as
spectral data. So the magnitude of the electron spin is given by:
s   √
~ = 1 1 3
|S| +1 ~= ~ (8.3)
2 2 2
1
With s = 2
, there are only 2s + 1 = 2 possible values of Sz specified by ms = + 21
(“spin-up”) and ms = − 21 (“spin-down”), i. e.

1
Sz = ± ~ (8.4)
2
Figure 8.2 illustrates the quantization of the electron’s intrinsic angular momentum and
intrinsic magnetic dipole moment.

Figure 8.2: The two possible spin states of an electron.

A particle’s intrinsic magnetic dipole moment is related to its intrinsic angular mo-
mentum by:
q ~
µ
~s = g S (8.5)
2m
where q is the particle’s charge, m is its mass and g is the gyromagnetic ratio. Accurate
measurement have shown that the gyromagnetic ratio g is very nearly equal to 2 for the
electron. From (8.5), the electron’s intrinsic magnetic dipole moment is given by:
e ~
~s = −
µ S (8.6)
m
Chapter 8 Many-Electron Atoms 124

The possible components of µ


~ s along the z axis are therefore limited to
e e 1 1
µs,z = − S z = − ms ~ ms = − , + (8.7)
m m 2 2

Figure 8.3: Stern-Gerlach apparatus: An atom with magnetic dipole moment passing
through a non-uniform magnetic field.

The existence of the intrinsic angular momentum of the electron was first demonstrated
in 1921 in an experiment by Otto Stern (1888-1969) and Walter Gerlach (1889-1979) to
measure the magnetic dipole moment of atoms. In their experiment, a beam of silver
atoms is sent through the gap between the poles of a magnet shaped to produce a non-
uniform magnetic field which has non-zero gradient only along the z-axis (see Fig. 8.3). A
non-uniform field exerts a net force on any magnetic dipole. Thus each atom is deflected
in the gap by the force:

~ = µz ∂Bz ẑ
F~ = −∇U = −∇(−~µ · B) (8.8)
∂z
where µz is the z-component of its magnetic dipole moment. If the magnetic dipole
moment of an electron is only contributed by its orbital angular momentum as in (7.28),
we can rewrite (8.8) as:
 e  ∂B
z
F~ = − Lz ẑ
2m ∂z
Thus the force should be quantized since Lz is quantized due to the space quantization
~ Using the quantization condition Lz = m` ~, we obtain
of L.
e ∂Bz
F~ = − m` ~ ẑ m` = −`, −` + 1, . . . , ` − 1, ` (8.9)
2m ∂z
Chapter 8 Many-Electron Atoms 125

Therefore, the atomic beam should split into (2` + 1) discrete components, one for each
distinct orientation of its orbital angular momentum with respect to the magnetic field.

The Stern-Gerlach experiment produced a surprising result: The silver atomic beam
was clearly split — but into only two components, not the odd number (2` + 1) expected
from the space quantization of L!~ Silver atoms in their ground state have no orbital
angular momentum (` = 0), so there is no splitting due to the orbital magnetic dipole
moment for such case. However, there is still a force due to the intrinsic magnetic dipole
moment µ
~ S of the electron. Inserting (8.7) in (8.8), we obtain the force acting on a silver
atom:  e  ∂B
z e ∂Bz 1 1
F~ = − Sz ẑ = − ms ~ ẑ ms = − , + (8.10)
m ∂z m ∂z 2 2
This gives rise to the splitting of the silver beam into two (= 2s + 1) components. Stern
and Gerlach has made the first direct observation of electron spin and space quantization.

We have previously described all the possible electronic states in hydrogen by three
quantum numbers (n, `, m` ). The introduction of electron spin into the atomic theory
means that a total of four quantum numbers (n, `, m` , ms ) are needed to describe each
possible state of an atomic electron, as shown in Table 8.1. (Note that we don’t need to
specify the spin quantum number s since it is always equal to 1/2.) For example, the
ground state of hydrogen atom was previously labeled as (n, `, m` ) = (1, 0, 0). With the
addition of ms , this would become either (1, 0, 0, + 21 ) or (1, 0, 0, − 12 ). The degeneracy of
the ground state is now 2. Since there are two possible labels for each previous single
label, the degeneracy of each level is 2n2 instead of n2 .

Table 8.1: Quantum Numbers of an Atomic Electron.


Name Symbol Possible Values Quantity Determined
Principal n 1, 2, 3, . . . Electron energy
Orbital ` 0, 1, 2, . . . , n − 1 Orbital angular momentum magnitude
Magnetic m` −`, . . . , 0, . . . , +` Orbital angular momentum direction
Spin Magnetic ms − 12 , + 21 Electron spin direction

We may wonder how such an inherently quantized, non-position-dependent property


can be mated to the other aspect of a particle’s state that we have thus far represented
with a wave function. We will find it sufficient simply to keep these two aspects grouped
but separate. For an electron in an atom, the quantum numbers n, `, and m` are said
Chapter 8 Many-Electron Atoms 126

to specify the electron’s spatial state, while the quantum number ms specifies its spin
state. Its state may be represented as:

ψn,`,m` ,ms (r, θ, φ) = ψn,`,m` (r, θ, φ) ms (8.11)

Because it can take on only two values, related to two opposite components, we often find
it useful to use an arrow for ms so that:

ψn,`,m` ,+ 1 (r, θ, φ) = ψn,`,m` (r, θ, φ) ↑ Spin up (8.12)


2

ψn,`,m` ,− 1 (r, θ, φ) = ψn,`,m` (r, θ, φ) ↓ Spin down (8.13)


2

Neither ms nor the arrow should be regarded as multiplying ψn,`,m` (r, θ, φ) the usual sense,
because spatial and spin-states are of entirely different character. But spin is a necessary
part of the overall description of the state, so it is fair to place the symbol alongside the
wave function.

8.2 Identical Particles

As we know, quantum mechanics limits the knowledge we may have of a one-particle


system. It places further limitations on our knowledge of multiparticle systems. In fact,
these limitations are inextricably linked to spin, but their foundation can be understood
without it. To simplify things, we put spin aside for the time being.

Suppose that there are two indistinguishable particles, i. e. two identical particles
sharing the same space. By “share the same space”, we mean that the particles are not
physically isolated. There is some region of space in which they “overlap”, i. e. a region
where either particle may be found.

To proceed further, we need a two-particle Schrödinger equation. To specify kinetic


and potential energies in the two-particle case, it is reasonable that we would need x1 and
x2 . Still, we seek a true two-particle equation in which the particles are not treated as
completely independent but are described by a single wave function ψ(x1 , x2 ). To avoid
confusion, the state of the whole system ψ(x1 , x2 ) is usually called multiparticle state,
while the term individual-particle state is often used for the state occupied by a single
particle ψ(x). It turns out that the two-particle Schrödinger equation is:
 
~2 ∂ 2 ~2 ∂ 2
− − ψ(x1 , x2 ) + U (x1 , x2 )ψ(x1 , x2 ) = Eψ(x1 , x2 ) (8.14)
2m ∂x21 2m ∂x22
Chapter 8 Many-Electron Atoms 127

where E is the energy of the entire two-particle system.

As a simple context in which to confront the central point in multiparticle systems, let
us consider an infinite well in which the particles exert no forces on each other. For this
case, equation(8.14) is satisfied by a simple product of particle-in-a-box wave functions
for each particle, each with its own quantum number.
r  nπx 
2
ψ(x1 , x2 ) = ψn (x1 )ψn0 (x2 ) where ψn (x) = sin (8.15)
L L
Let us examine (8.15) for the case n = 4 and n0 = 3. The system’s multiparticle wave
function would be:
   
2 4πx1 3πx2
ψ(x1 , x2 ) = ψ4 (x1 )ψ3 (x2 ) = sin sin
L L L
Since this wave function is real, we would expect its square to give us the probability
density P (x1 , x2 ) for this two-particle system:
   
4 4πx1 3πx2
2
P (x1 , x2 ) = ψ (x1 , x2 ) = 2 sin2 sin 2
(8.16)
L L L
As a logical generalization of its meaning in the one-particle case, this would be the
probability of simultaneously finding particle 1 within a unit distance at x1 and particle 2
within a unit distance at x2 , i. e. probability/(distance1 · distance2 ). If we integrate (8.16)
over both x1 and x2 between limits 0 and L, we obtain a sensible result: the probability
of finding both particles somewhere within the box is one.

4
 
Figure 8.4: The probability density P (x1 , x2 ) = L2
sin2 4πx1
L
sin2 3πx2
L
.

Fig. 8.4 shows the total probability density (8.16) as a function of the two variables
x1 and x2 . Note how the behaviors with respect to x1 and x2 differ. In particular, we see
Chapter 8 Many-Electron Atoms 128

that the probability of finding particle 1 at the center of the box is 0, independent of the
value of x2 . In contrast, the probability of finding the particle 2 at the center is in general
non-zero, depending on the value of x1 .

But this cannot be true. Particles 1 and 2 are identical. They bear no distinctive
label, they cannot be watched to keep track of them separately, and their wave functions
share the same space. If an experiment is performed and a particle found, we simply
cannot know where the particle it is. Because it is impossible to verify any asymmetry
in the behaviors of the two particles, we demand that the probability density obey the
following restriction:
Probability density must be unchanged
if the labels of indistinguishable particles are switched.
In other words, the probability density must be symmetric.

How might we alter P (x1 , x2 ) so that switching particle labels 1 and 2 leaves it un-
changed? To produce a symmetric probability density, we must start with the wave
function. There are two ways of modifying the product solution in equation (8.15) to
meet the requirement:

ψS (x1 , x2 ) = ψn (x1 )ψn0 (x2 ) + ψn0 (x1 )ψn (x2 ) Symmetric (8.17)
ψA (x1 , x2 ) = ψn (x1 )ψn0 (x2 ) − ψn0 (x1 )ψn (x2 ) Antisymmetric (8.18)

Because the Schrödinger equation is a linear differential equation, both of these are
as valid mathematically as (8.15). But in solutions (8.17) and (8.18), the notion that one
particle occupies state n and the other state n0 is abandoned. Both are combinations of
particle 1 in state n, particle 2 in state n0 and of particle 1 in state n0 , particle 2 in state
n. Moreover, wave functions (8.17) and (8.18) lead to a symmetric probability density.
The probability densities corresponding to wave functions (8.17) and (8.18) are:

PS (x1 , x2 ) = ψS2 (x1 , x2 ) = ψn2 (x1 )ψn2 0 (x2 ) + ψn2 0 (x1 )ψn2 (x2 ) + 2ψn (x1 )ψn0 (x2 )ψn0 (x1 )ψn (x2 )
PA (x1 , x2 ) = ψA2 (x1 , x2 ) = ψn2 (x1 )ψn2 0 (x2 ) + ψn2 0 (x1 )ψn2 (x2 ) − 2ψn (x1 )ψn0 (x2 )ψn0 (x1 )ψn (x2 )

Interchanging particle labels 1 and 2 leaves both expressions unchanged. For convenience,
we refer to these probability densities as “symmetric” or “antisymmetric” — i. e. according
to the corresponding wave function — but both probability densities are symmetric.

Both the symmetric and antisymmetric probability densities preserve certain feature
of the unsymmetrized probability density (8.16). When properly normalized, they yield
Chapter 8 Many-Electron Atoms 129

one when integrated over all values of x1 and x2 , and they may be interpreted as the
probability of finding particle 1 within a unit distance at x1 and particle 2 within a unit
distance at x2 . Now, however, the probability is the same as if 1 and 2 are switched.

Fig. 8.5 shows PS (x1 , x2 ) and PA (x1 , x2 ) for our special case n = 4 and n0 = 3. Like
the unsymmetrized probability density, they have 12 bumps, but in both PS and PA , they
are divided symmetrically between x1 and x2 . Nevertheless, there is a crucial distinction
between PS and PA . The largest values of PS occur where the particles are close together,
near the line x1 = x2 . However, PA is identically zero when x1 = x2 . On average, the
particles are farther apart in an antisymmetric spatial state than in a symmetric one.

Figure 8.5: Symmetric and antisymmetric probability densities when n = 4 and n0 = 3


states are occupied.

Example 8-1

Two particles in a box occupy the n = 4 and n0 = 3 states. Equations (8.17) and (8.18)
say that the symmetric and antisymmetric wave functions are of the form:
√         
2 4πx1 3πx2 3πx1 4πx2
ψS (x1 , x2 ) = sin sin + sin sin
L L L L L
√         
2 4πx1 3πx2 3πx1 4πx2
ψA (x1 , x2 ) = sin sin − sin sin
L L L L L

The normalization constant 2/L ensures that the total probability of finding the box
somewhere in the box is one. For both ψS and ψA , calculate the probability that both
particles would be found in the left half of the box, i. e. between 0 and L/2.
Chapter 8 Many-Electron Atoms 130

(Hints:
Use the integral formula
Z L/2 

kπx
2 L
sin dx =
0 L 4
Z L/2      
kπx jπx L sin[π(k − j)/2] sin[π(k + j)/2]
sin sin dx = −
0 L L 2π k−j k+j
where k and j are integers. )

Solution:
We integrate the square of the wave function only over values of x1 and x2 between 0 and
L/2. Using a ± sign to treat both cases together, we have:

probability
Z L/2 Z L/2 ( √         )2
2 4πx1 3πx2 3πx1 4πx2
= sin sin ± sin sin dx1 dx2
0 0 L L L L L
Z L/2   Z L/2  
2 2 4πx1 2 3πx2
= 2 sin dx1 sin dx2
L 0 L 0 L
Z L/2   Z L/2  
2 2 3πx1 2 4πx2
+ 2 sin dx1 sin dx2
L 0 L 0 L
Z L/2     Z L/2    
4 4πx1 3πx1 3πx2 4πx2
± 2 sin sin dx1 sin sin dx2
L 0 L L 0 L L
The first four integrals evaluate to L/4, and the last two evaluate to 4L/(7π). Thus
        
2 L L 2 L L 4 4L 4L
probability = 2 + 2 ± 2
L 4 4 L 4 4 L 7π 7π
(
1 0.382 for symmetric wave function
= ± 0.132 =
4 0.118 for antisymmetric wave function

Ignoring symmetrization, the probability of either particle individually being found in


the box’s left half would be 12 , so the probability of both being there would be 41 . The
symmetrization requirement yields two states: the symmetric, in which the probability is
greater than 14 , and the antisymmetric, in which it is less.

Note that ψS and ψA have equal energy, the same as that of the unsymmeterized wave
function, for the two particle-in-a box case. Accordingly, it would seem equally likely to
find the system in the antisymmetric state as in the symmetric. But this wouldn’t be the
case if the particle exerted forces on one another. Were the two particles electrons, the
Chapter 8 Many-Electron Atoms 131

antisymmetric state would be of lower energy. It’s because the particles in such a state
are less likely to be found close together — the repulsive energy would be reduced. This
argument holds even though the interparticle potential energy renders the Schrödinger
equation unsolvable except by numerical methods.

To ensure a symmetric probability density, it is not sufficient to consider just the spatial
state which refers to particles being found at given locations. We must ensure that the
probability of finding them at given locations and in given spin states is symmetric under
exchange of labels. Equations (8.17) and (8.18) may be generalized as follows:

ψS (x1 , x2 ) = ψn (1)ψn0 (2) + ψn0 (1)ψn (2) Symmetric (8.19)


ψA (x1 , x2 ) = ψn (1)ψn0 (2) − ψn0 (1)ψn (2) Antisymmetric (8.20)

Here, n represents all quantum numbers needed to specify the individual-particle state,
and the arguments of the functions are the particles’ labels. For example, to represent an
electron with spin up in a hydrogen’s ground state, n is actually the set {n, `, m` , ms } =
{1, 0, 0, + 12 }, and ψn (2) would mean ψ1,0,0 (r2 , θ2 , φ2 ) ↑ in accord with equations (8.12) and
(8.13).

We would use the term exchange symmetry to refer to the character of a state
when particle labels are exchanged. If it doesn’t change sign when labels are switched, its
exchange symmetry is symmetric; if it changes sign, its exchange symmetry is antisym-
metric.

8.3 The Pauli Exclusion Principle

Why is spin so crucial to the behavior of multiparticle systems? It is fundamental to


nature that a system of multiple indistinguishable particles will be in a multiparticle state
of definite exchange symmetry, and whether it is symmetric or antisymmetric depends on
the spin of the system’s particles. All fundamental particles have spin s that is either
integral or half-integral, and this divides them into two categories. Theory prediction as
well as experiment verification indicates that:

1. Bosons: Particles for which s = 0, 1, 2, . . . manifest a symmetric multiparticle state

2. Fermions: Particles for which s = 21 , 23 , 52 , . . . manifest a antisymmetric multiparticle


state
Chapter 8 Many-Electron Atoms 132

Table 8.2 gives several examples of bosons and fermions. It also reveals that all the
familiar building blocks of nature — electrons, protons, and neutrons — are spin- 12 . Being
fermions, they always assume antisymmetric multiparticle states. This fact may seem
harmless, but the consequences are resounding. Let us investigate this in details.

Table 8.2: Spins of some selected particles.


Fermions (Half-integral spin) Bosons (Integral spin)
Particle s Particle s
1
Electron, e− 2
Pion, π 0 0
1
Proton, p 2
Alpha particle, α 0
1
Neutron, n 2
Photon, γ 1
1
Neutrino, ν 2
Deuteron, d 1
3
Omega, Ω− 2
Graviton 2

Consider a system of two fermions occupying individual-particle states n and n0 . If n


and n0 are equal, the antisymmetric two-particle state is:

ψn (1)ψn (2) − ψn (1)ψn (2) = 0

We conclude that two fermions cannot have the same quantum numbers – their spatial
state or their spin state or both must differ. This is known as the Pauli exclusion
principle which is as follows:

No two indistinguishable fermions may occupy the same individual-particle state.

To specify a particle’s full set of quantum numbers is to specify its state, so an equivalent
way of expressing the exclusion principle is that no two indistinguishable fermions may
have the same set of quantum numbers. Although we have considered only the simple
case of two fermions, the exclusion principle holds for any number. It was discovered by
Wolfgang Pauli in 1924, and the achievement won him the 1945 Nobel Prize.

The exclusion principle applies only to indistinguishable fermions which means that
they must (1) be of the same kind and (2) share the same space. We don’t demand
exchange symmetry for particles that can be distinguished, so the exclusion principle
does not apply to a system of different kinds of fermions, such as hydrogen’s proton and
electron. The second condition is a more delicate issue. We might argue that everything in
Chapter 8 Many-Electron Atoms 133

the universe shares the same space that the wave functions of all things overlap. However,
for all practical purposes, electrons far apart, whose wave functions overlap only negligible,
do not share the same space, so the requirement of exchange symmetry can be ignored.

The importance of the exclusion principle cannot be overstated. As we shall soon


see, were there no exclusion principle, nature would be profoundly different. It is the
foundation of chemistry.

The exclusion principle arises directly from the nature of an antisymmetric multipar-
ticle state, specifically the minus sign that causes (8.20) to be 0 if n and n0 are equal.
However, bosons manifest the symmetric state (8.19), which is not 0, so occupation of the
same individual-particle state is perfectly acceptable. The exclusion principle does not
apply to bosons.

Fundamental particles are either fermions or bosons, but composite particles can also
be one or the other. In fact, most of the particles listed in Table 8.2 are composites.
Integers must always add to integers, so when bosons are “glued” together into a composite
unit — they need not even be identical — that unit is a boson; and half-integers can add
to integers or half-integers, so an even number of fermions forms a boson, and an odd
number a fermion. For example, the proton comprising three spin- 21 quarks is a fermion.

8.4 Electronic States in Many-Electron Atoms

As in the case of the hydrogen atom, orbits with the same value of n all lie at about the
same average distance from the nucleus in multi-electron atoms. The set of orbits with
the same value of n, with about the same average distance from the nucleus, is known as
the atomic shell. The atomic shells are designated by letters as follows:

n 1 2 3 4 5
Shell K L M N O

The levels with a certain value of n and ` (e. g. 2s or 3d) are known as subshells. The
number of electrons that can be placed in each subshell is 2(2` + 1). According to this
scheme, the 1s subshell has a capacity of 2 electrons and the 3d subshell has a capacity
of 10 electrons.

Although the multielectron atom cannot be solved exactly, approximation techniques


can be used to find the energy ordering of the subshells in many-electron atoms. Table 8.3
Chapter 8 Many-Electron Atoms 134

shows the order of filling as well as the number of electrons capacity of each subshell. The
1s subshell remains the lowest energy state. In addition, the 2s and 2p subshells remain
fairly close in energy with the 2s subshell always less energetic. Electrons in the 2s and
2p subshells would be shielded, or “screened”, by the two 1s electrons from “seeing” the
full nuclear charge +Ze. Moreover, the 2s electron has less average distance from the
nucleus than the 2p electron. So it is more likely for the 2s electron to be closer to the
nucleus than the Bohr radius, where it feels the attraction of the entire nuclear charge.
The 2s electron would on average feel a slightly greater attractive force from the nucleus
than the 2p electron. Thus the 2s electron is more tightly bound and has less energy.

Table 8.3: Order of filling and capacity of atomic subshells.


Subshell 1s 2s 2p 3s 3p 4s 3d 4p 5s 4d 5p 6s 4f 5d 6p
Capacity
2 2 6 2 6 2 10 6 2 10 6 2 14 10 6
2(2` + 1)

A more extreme example of the tighter binding of the penetrating orbits occurs
for the n = 3 subshells. The 3s electron penetrates the inner orbits, and the 3p electron
penetrates almost as much. The 3d electron has negligible penetration of the inner orbits.
As a result, the 3s and 3p subshells are more tightly bound and therefore lower in energy
than the 3d subshell. For a given n, energy is lower for lower-` orbits.

Figure 8.6: The energy ordering of atomic subshells varies with the atomic number Z.

Table 8.3 gives the order of filling of the energy subshells, and so only the “outer” or
valence electrons are considered. However, as the outer subshells is filled, the complex
Chapter 8 Many-Electron Atoms 135

electron interactions cause the ordering of inner subshells to change. A lower n is a lower
energy, whereas a lower ` is still a lower energy within a shell. Figure 8.6 schematically
depicts this progressive change in subshell ordering.

8.5 The Periodic Table

Figure 8.7: Periodic table of the elements.

Figure 8.7 shows the periodic table in which the chemical elements are listed in order
of increasing atomic number Z in series of rows so that elements with similar properties
form vertical columns.

In the periodic table, the vertical columns are known as groups which is composed of
elements with similar chemical and physical properties. For example, group VII consists
of the halogens — volatile nonmetals that form diatomic molecules in the gaseous state.
Like the alkali metals, the halogens are chemically active, but as oxidizing agents rather
than reducing agents. On the other hand, the horizontal rows in the periodic table are
called periods corresponding to the filling of the subshells. The first three periods are
broken in order to keep their members aligned with the most closely related elements of
the long periods below. Most of the elements are metal (Fig. 8.8). Across each period
is a more or less steady transition from an active metal through less active metals and
Chapter 8 Many-Electron Atoms 136

then weakly active non-metals to highly active nonmetals and finally to an inert gas
(Fig. 8.9). Within each column, there are also regular changes in properties, but they are
far less obvious than those in each period. For example, increasing atomic number in the
alkali metals is accompanied by greater chemical activity, while the reverse is true in the
halogens.

Figure 8.8: The majority of the elements are metals.

Figure 8.9: The variation of chemical activity in the periodic table.

The observed ordering of the periodic table is related with the order of filling of the
atomic subshells. The filling of electrons into the subshells must follow two rules:

(i) The capacity of each subshell is 2(2` + 1).


(ii) The electrons occupy the lowest energy states available.

To indicate the electron configuration of each element, we use a notation in which


the identity of the subshell and the number of electrons in it are listed. The identity of
the subshell is indicated in the usual way, and the number of electrons in that subshell
is indicated by a superscript. Thus hydrogen has the configuration 1s1 for one electron
in the 1s subshell, and helium has the configuration 1s2 . Table 8.4 lists the electron
configuration of some of the elements.
Chapter 8 Many-Electron Atoms 137

Table 8.4: Electron configuration of some elements.


H 1s1 Ar [Ne] 3s2 3p6 Kr [Ar] 4s2 3d10 4p6
He 1s2 K [Ar] 4s1 Rb [Kr] 5s1
Li 1s2 2s1 Sc [Ar] 4s2 3d1 Y [Kr] 5s2 4d1
Be 1s2 2s2 Cr [Ar] 4s1 3d5 Mo [Kr] 5s1 4d5
B 1s2 2s2 2p1 Mn [Ar] 4s2 3d5 Ag [Kr] 5s1 4d10
Ne 1s2 2s2 2p6 Cu [Ar] 4s1 3d10 In [Kr] 5s2 4d10 5p1
Na [Ne] 3s1 Zn [Ar] 4s2 3d10 Xe [Kr] 5s2 4d10 5p6
Al [Ne] 3s2 3p1 Ga [Ar] 4s2 3d10 4p1 Cs [Xe] 6s1

What is most remarkable about this scheme is that the arrangement of the periodic
table was known well before the introduction of atomic theory. The elements were orga-
nized into groups and periods based on their physical and chemical properties by Dmitri
Mendeleev (1834-1907) in 1859. Understanding this organization in terms of atomic levels
is a great triumph for the atomic theory. What remains is to interpret the chemical and
physical properties based on the atomic levels.

8.6 Properties of the Elements

In this section, we briefly study how our knowledge of atomic structure helps us to un-
derstand the physical and chemical properties of the elements. Our discussion is based
on the following two principles:

(i) Filled subshells are normally very stable configurations. An atom with one electron
beyond a closed shell will readily give up that electron to another atom to form a
chemical bond. Similarly, an atom lacking one electron from a closed shell will readily
accept an additional electron from another atom in forming a chemical bond.
(ii) Filled subshells do not normally contribute to the chemical or physical properties of
an atom. Only the electrons in the unfilled subshells need to be considered. (X-ray
energies are an exception to this rule.)

We consider a number of different physical properties of the elements, and try to


understand those properties based on atomic theory.
Chapter 8 Many-Electron Atoms 138

1. Atomic Radii. We have already learned that the radius of an atom is not a precisely
defined quantity since the electron probability density determines the “size” of an
atom. In fact, different kinds of experiments may give different values for the radius.
One way of defining the radius is by means of the spacing between the atoms in a
crystal containing the element. Figure 8.10 shows how such typical atomic radii vary
with atomic number Z.

Figure 8.10: Atomic radii, determined from ionic crystal atomic separations.

2. Electrical Resistivity. For a bulk material of uniform density with length L and
cross-sectional area A, the resistance is R = ρL/A where ρ is the resistivity. A good
electrical conductor has a small resistivity (e. g. ρ = 1.7 × 10−8 Ω · m for copper) while
a poor conductor has a large resistivity (e. g. ρ = 2 × 1015 Ω · m for sulphur). From an
atomic viewpoint, electrical current depends on the movement of relatively loosely
bound electrons, which can be removed from their atoms by the applied potential
difference. It also depends on the ability of the electron to travel from one atom to
another. Thus elements with s electrons, which are the least tightly bound and also
travel farthest away from the nucleus, are expected to have small resistivities. Figure
8.11 shows the variation of electrical resistivity with atomic number Z.
3. Ionization Energy. The minimum energy needed to remove an electron from an
atom is known as its ionization energy. For example, hydrogen has an ionization
energy of 13.6 eV. Helium has an ionization energy of 24.6 eV for the first electron
and 54.4 eV for the second electron. Figure 8.12 shows the variation of the ionization
energy with atomic number Z.
Chapter 8 Many-Electron Atoms 139

Figure 8.11: Electrical resistivities of the elements.

By examining Figures 8.10 to 8.12, you can see the remarkable regularities in the properties
of the elements. In particular, notice how similar the properties of the different sequences
of elements are, e. g. the electrical resistivity of the d-subshell elements. We now look at
how the atomic structure is responsible for these properties.

Inert Gases
The inert gases occupy the last column of the periodic table. Since they have only
filled subshells, the inert gases do not generally combine with other elements to form
compounds. These elements are very reluctant to give up or accept an electron. They are
monoatomic gases at room temperature. Since their atoms don’t like to join with each
other, the boiling point are very low (typically −200◦ C). Their ionization energies are
much larger than those of neighboring elements, because of the extra energy needed to
break open a filled subshell.

p-Subshells Elements
The elements of the column next to the inert gases are the halogens. These atoms lack
one electron from a closed shell and have the configuration np5 . Since a closed p subshell
is a very stable configuration, these elements readily form compounds with other atoms
that can provide an extra electron to close the p subshell. The halogens are therefore
extremely reactive.
Chapter 8 Many-Electron Atoms 140

Figure 8.12: Ionization energies of neutral atoms of the elements.

As we move across the series of six elements in which the p subshell is being filled, the
atomic radius decreases. This “shrinking” occurs because the nuclear charge is increasing
and pulling all of the orbits closer to the nucleus. Notice from Fig. 8.8 that the halogens
have the smallest radii within each p subshell series. Besides, the p electrons also become
more tightly bound as we increase the nuclear charge. From Figure 8.12, we can see how
the ionization energy increases systematically as the p subshell is filled.

s-Subshells Elements
The elements of the first two columns (groups) are known as alkalis (configuration ns1 )
and alkaline earths (ns2 ). The single s electron makes the alkalis quite reactive; the
alkaline earths are similarly reactive, in spite of the filled s subshell. This occurs because
the s shell electron wave function can extend rather far from the nucleus, where the
electron is screened (by other electrons) from the nuclear charge and therefore not tightly
bound. For this reason, the ns1 and ns2 elements have the largest atomic radii but the
smallest ionization energies. Moreover, they are relatively good electrical conductors.

Transition Metals
The three rows of elements in which the d subshell is filling (Sc to Zn, Y to Cd, Lu to Hg)
are known as the transition metals. Many of their chemical properties are determined
by the “outer” electrons — those whose wave functions extend furthest from the nucleus.
For the transition metals, these are always s electrons, which have a larger mean radius
than the d electrons. As the atomic number increases across the transition metal series,
Chapter 8 Many-Electron Atoms 141

we add one d electron and one unit of nuclear charge; the net effect on the s electron
is very small, since the additional d electron screens the s electron from the additional
nuclear charge. The chemical properties of the transition metals are very similar, as the
small variation in radius and ionization energy shows.

The electrical resistivity of the transition metals shows two interesting features: a
sharp rise at the center of sequence, and a sharp drop near the end. The sharp drop near
the end of the sequence indicates the small resistivity of copper, silver, and gold. If we
fill the d subshell in the expected sequence, copper would have the configuration 4s2 3d9 ;
however the filled d subshell is more stable than a filled s subshell, and so one of the
s electrons is transferred to the d subshell, resulting in the configuration 4s1 3d10 . This
relatively free, single s electron makes copper an excellent conductor. Silver (5s 1 4d10 ) and
gold (6s1 5d10 ) behave similarly.

At the center of the sequence of the transition metals there is sharp rise in the resis-
tivity. Apparently a half-filled shell is also a stable configuration, and so Mn (3d5 ) and Re
(5d5 ) have larger resistivities than their neighbors. (The elements Tc, with configuration
4d5 , is radioactive and is not found in nature; its resistivity is unknown.) A similar rise
in resistivity is seen at the center of the 4f sequence.

Lanthanides (Rare Earths)


The rare earths are rather similar to the transition metals in that an “inner” subshell
(the 4f ) is being filled after an “outer” subshell (the 6s). For the same reasons discussed
above, the chemical properties of the rare earths should be rather similar, since these are
mainly determined by the 6s electron; the variation in the radii and ionization energies
show that this is true.

Actinides
The actinide is the row of elements in which the 5f subshell is filling. They should
have chemical and physical properties similar to those of the rare earths. Unfortunately,
most of the actinide elements (those beyond uranium) are radioactive and do not occur
in nature. They are artificially produced elements and are available only in microscopic
quantities. We are thus unable to determine their bulk properties.
Chapter 8 Many-Electron Atoms 142

8.7 X-ray Spectra

X-rays are electromagnetic radiations with wavelengths approximately from 0.01 to 10 nm.
In chapter 3, we have discussed the continuous X-ray spectrum emitted by accelerated
electrons in the bremsstrahlung process. Here we are concerned with the discrete X-ray
line spectra emitted by atoms.

X-rays are emitted in transitions between normally filled lower energy levels of an
atom. The inner electrons are so tightly bound that the energy spacing of these levels
is about right for the emission of photons in the X-ray range of wavelengths. The outer
electrons are relatively weakly bound, and the spacing between these outer levels is only a
few electron-volts. Transition between these levels gives photon of energies of a few eVs,
corresponding to the visible region of the spectrum.

Since all the inner shells of an atom are filled, X-ray transitions do not occur between
these levels under normal circumstances. For example, a 2p electron cannot make a
transition to the 1s subshell, because all atoms beyond hydrogen have filled 1s subshells.
In order to observe such a transition, we must remove an electron from the 1s subshell.
We can do so by bombarding the atoms with electrons (or other particles) accelerated to
an energy sufficient to knock loose a 1s electron following a collision.

Figure 8.13: Characteristic X-rays are produced when an inner-shell hole is made.

Once we have removed an electron from the 1s subshell, an electron from a higher
subshell will rapidly make a transition to fill that vacancy, emitting an X-ray photon in
Chapter 8 Many-Electron Atoms 143

the process (see Fig. 8.13). The energy of the photon is equal to the energy difference of
the initial and final atomic levels of the electron that makes the transition.

When we remove a 1s electron, we are creating a vacancy in the K shell. The X-rays
that are emitted in the process of filling this vacancy are known as K-shell X rays, or
simply K X rays. The K X ray that originates from the n = 2 shell (L shell) is known as
Kα X ray while the K X rays originating from higher shells are known as Kβ , Kγ , and so
forth. Figure 8.14 illustrates these transitions.

Figure 8.14: X-ray series.

It is also possible that the bombarding electrons can knock loose an electron from the
L shell, and electrons from higher levels will drop down to fill this vacancy. The photons
emitted in these transitions are known as L X rays. The lowest-energy X ray of the L
series is known as Lα , and the other L X rays are labeled in order of increasing energy as
shown in Figure 8.14. In a similar manner, we label the other X-ray series by M , N , nd
so on. Figure 8.15 shows a sample X-ray spectrum emitted by silver.

A vacancy in the K shell can be filled by a transition from the L shell, with the
emission of the Kα X ray. However, the electron that made the jump from the L shell
left a vacancy there, which can be filled by an electron from a higher shell, with the
accompanying emission of a L X ray. Thus it is possible to have a L X ray emitted
directly following the Kα X ray.
Chapter 8 Many-Electron Atoms 144

Figure 8.15: Characteristic X-ray spectrum of silver.

We did not consider the energy differences of the subshells within the major shells.
For example, the Lα X ray could originate from one of the n = 3 subshells (3s, 3p, 3d) and
end in one of the n = 2 subshells (2s, 2p). The energies of these different transitions will
be slightly different, so that there will be many Lα X rays, but their energy differences
are very small, e. g. compared with the energy difference between the Lα and Lβ X rays.
In fact, we don’t even notice this small energy splitting in many applications.

The regularity of the X ray spectra was first observed by the British physicist Henry
Moseley (1887 - 1915). He also interpreted the X ray spectra on the basis of the Bohr
model. He argued that an electron in the L shell is screened by the two 1s electrons, and
so it sees an effective nuclear charge of Zeff ' Z − 2. When one of the 1s electrons is
removed in the creation of a K-shell vacancy, only the remaining 1s electron shields the L
shell, and so Zeff ' Z − 1. Thus Moseley concluded that Kα X ray can be considered as a
transition from the n = 2 to the n = 1 level in a one-electron atom with Zeff = Z −1. Using
the energy equation from the Bohr model (Equation (6.22)), he find that the frequency
of the Kα transition in an atom of atomic number Z is given by

3me4 3cR∞
ν= 3 2 3
(Z − 1)2 = (Z − 1)2 (8.21)
256π 0 ~ 4

by letting n1 = 2, n2 = 1 and replacing e4 by (Z − 1)2 e4 . So the plot of ν as a function
p
of Z should be a straight line with slope 3cR∞ /4. Such expectation was found to be in
good agreement with experimental results as shown in Figure 8.16.

This method gives us a powerful and yet simple way to determine the atomic number
Z of an atom, and was first demonstrated by Moseley. He was also the first to demonstrate
the type of linear relationship shown in Fig. 8.16 and so such graphs are now known as
Chapter 8 Many-Electron Atoms 145

Moseley plots. His discovery provided the first direct means of measuring the atomic
numbers of the elements. Previously, the elements has been ordered in the periodic table
according to their increasing masses. Moseley found certain elements were listed in wrong
order, in which the element of higher atomic number Z happened to have the smaller
mass. He also found gaps corresponding to elements which were not yet discovered.

Figure 8.16: Square root of frequency versus atomic number for Kα X rays. The slope of
p
the line is in good agreement with the expected value 3cR∞ /4 = 4.966 × 107 s−1/2 .

Moseley’s work was of great importance in the development of atomic physics. Moseley
not only provided a direct confirmation of the Bohr model, but he also made a strong link
of atomic theory with the periodic table.
Chapter 9

Statistical Mechanics

Statistical mechanics is not modern physics in the same sense that special relativity
and quantum mechanics are. Rather, it is a distinct area of physics that applies to many
other areas, classical as well as quantum. In statistical mechanics, our concern is making
predictions for the properties and behaviors in systems consisting of huge number of
particles that a complete knowledge of the system is impossible. Faced with incomplete
knowledge, we make statistical predictions based on the applicable laws of mechanics —
classical or quantum. Necessarily, our prediction of what will be observed is the average of
all the things that might possibly occur. For instance, air pressure is perfectly explained
as an average of forces exerted in isolated microscopic collisions.

Statistical mechanics is able to make successful predictions when the number of parti-
cles is large. And averages over very large numbers are much more precise than those over
small numbers. A system in which the number of particles is large enough that statistical
predictions become very precise is known as a thermodynamic system. It is only in
such macroscopic systems that the average properties of pressure, temperature, and den-
sity have real meaning. The molecules of air in a room, the electrons in a conducting wire,
the photons within a bed of glowing coals — each constitutes a thermodynamic system.

In this chapter, we will focus on understanding energy distributions. An energy dis-


tribution specifies the fraction of a system’s particles that will be in a given energy state.
Generally speaking, the higher the energy of a state, the smaller will be the number of
particles in that state. To show how large numbers lead to reliable predictions, let us
begin with another kind of distribution in one of the simplest thermodynamic system
imaginable.

146
Chapter 9 Statistical Mechanics 147

9.1 A Simple Thermodynamic System

Consider a “two-sided room” which is a box divided in half by an imaginary line and
containing N point particles that are free to move back and forth. The distribution we
consider here is not an energy distribution, but a simpler one: the number of particles
on a given side. Obviously, on average, half the particles should be on each side. What’s
important here is that we can easily show that deviations from the expected distribution
are negligible when N is large.

Being statistical predictions, distributions rest upon probabilities, and probabilities


rest upon numbers of ways of doing something. For our two-sided room, the number of
ways WNNR of arranging air molecules so that NR are on the right side is given by the
binomial coefficient9 : !
N N!
WNNR = =
NR NR !(N − NR )!

For example, if the N were 4, WN4 R gives the number of ways of distributing four particles
on two sides, as illustrated in Fig. 9.1.

Figure 9.1: Number of ways of distributing four particles on two sides of a room.

In the above calculation, we have assumed a point fundamental to statistical mechan-


ics: All ways of arranging specific particles on given side are equally probable. But a state
!
9 n
Note that the binomial coefficient can be regarded as the number of ways that k objects can
k
be chosen from among n objects, regardless of order.
Chapter 9 Statistical Mechanics 148

in which the distribution of particles is uniform, regardless of which specific particles are
on which side, is the most probable, simply because there is more number of ways to
obtain it.

To see how large numbers lead to a precise average, consider cases of increasing N
as shown in Fig. 9.2. For the case N = 4, the total number of ways of distributing the
particles is 16 and is approximately the area under the curve. When N = 40, the curve
is peaked more sharply at NR = N/2 than in the N = 4 case. The trend continues for
N = 400. For the macroscopic case of 1023 , the total number of ways occupies a very
small region around NR = N/2. So our conclusion is: Each individual particle has a
probability of 1/2 of being in either of its two possible “states” (left or right), and the
distribution of particles will indeed have half in each state if it is a truly thermodynamic
system.

Figure 9.2: Number of ways of distributing particles on two sides of a room — variation
as total number of particles increases from 4 to 1023 .

There are some important terminology that helps us keep in mind the crucial distinc-
tion between the microscopic and the macroscopic. The microstate is the state of the
system, given complete microscopic knowledge of the states of the individual particles.
In a two-sided room, the individual-particle states are simply “left” and “right”. Each
Chapter 9 Statistical Mechanics 149

“way” of obtaining the distribution for a given NR is a different microstate.

Realistically, however, such microscopic information is unknowable. We assume that it


is plausible to know at least the number of molecules on the two sides, but this constitutes
knowledge of only the macrostate. To know the macrostate of a system is to know the
properties that do not depend on the exact microscopic states of every individual particle:
the overall properties of number, energy, and volume and the average local properties
of pressure, temperature, and particle concentration. In a two-sided room, only one
microstate corresponds to the macrostate NR = N . On the other hand, the NR = N/2
macrostate is the one obtainable in the greatest number of microscopic ways. Because
it corresponds to the greatest number of microstates, all of which are assumed equally
probable, it is the most probable macrostate. The most probable macrostate is what we
know as the equilibrium state.

If left alone, thermodynamics systems eventually reach equilibrium, a macrostate


whose macroscopic properties of pressure, temperature, and particle concentration do not
change with time. Begin in a nonequilibrium macrostate, one for which the number of mi-
crostates is relatively small, a thermodynamic system will pass through other macrostates,
its macroscopic properties varying in space and time, while inevitability approaching its
unchanging equilibrium state. When this equilibrium point is reached, the microstate still
changes but the macrostate does not. And if the system’s macrostate is the same for-
ever, so are its macroscopic properties. This unchanging, overwhelmingly most probable
macrostate defines the equilibrium state.

In a more general system, where many different energy states are available to indi-
vidual particles, we might initially distribute the total energy in a strange way. But the
system would eventually assume its most probable state, with a predictable fraction of
the particles in each individual-particle state. It is such equilibrium energy distributions
that we are going to study in next section.

9.2 Boltzmann Distribution

One of the most important and useful predictions of statistical mechanics is how energy
will be distributed among a large number of identical particles that are able to exchange
energy and thus reach a common equilibrium state. In the following sections, we study
several such equilibrium energy distributions and their applications. Each gives the
Chapter 9 Statistical Mechanics 150

number of particles in a state of a given energy. It is one of the wonders of physics that
knowing only whether the particles are bosons, fermions, or “classically distinguishable”
particles, we can say how many will be in each state. Once again, the simple reason is
that average behaviors become quite predictable when N approaches Avogadro’s number.

We begin with the Boltzmann distribution, a special case that applies when par-
ticles are so spread out that the indistinguishability of identical bosons or fermions may
be ignored. A good example is a gas under ordinary conditions.

Let us first consider a simple special case for the Boltzmann distribution. Suppose
there is a system of N identical but separately identifiable harmonic oscillators, exchanging
energy in some unspecified way — one oscillator jumps to a higher level as another drops
to a lower. We simplify things further by shifting the potential energy by − 21 ~ω0 . The
energy of the ith oscillator in its ni th energy level is thus

Eni = ni ~ω0 (ni = 0, 1, 2, . . .) (9.1)

and the total energy of the N oscillators is


N
X N
X
E= ni ~ω0 = M ~ω0 where M = ni
i=1 i=1

Note that the integer M , the sum of the quantum numbers of all oscillator, is directly
proportional to the total energy.

To find the probability of a given oscillator in a given energy state, we must use some
elementary probability and statistics. An axiom of statistics is that, given numerous
equally likely possibilities, the probability that a particular subset will occur is the number
of ways the subset might occur divided by the total number of ways of obtaining all
possibilities. Applying this to our case, the probability that particle i will possess energy
Eni — that its quantum number will be ni — is:
number of ways energy can be distributed with ni fixed
Pni = (9.2)
total number of ways energy can be distributed
Now, the number of ways N integers — the particles’ quantum numbers — can be added
to give the integer M is
!
(M + N − 1)! M +N −1
number of ways N integers can add to M = =
M !(N − 1)! M
This would be the total number of ways the energy could be distributed among the N
quantum numbers. The number of ways energy can be distributed with ni fixed is the
Chapter 9 Statistical Mechanics 151

number of ways the other N − 1 quantum numbers can be added to give M − ni , which
is proportional to the remainding energy. Thus, the ratio in (9.2) becomes
! !
(M − ni ) + (N − 1) − 1 M +N −1
Pni = / (9.3)
M − ni M

The probability given by (9.3) for all values of ni are shown in Fig. 9.3. The curve drops
sharply as ni increases, which leads to an important conclusion.:

Varying the energy of just one particle causes a sharp change in the number of ways of
distributing the remaining energy among the other particles. The greatest freedom to
distribute the remaining energy occurs when that one particle has the least energy.
Therefore, the most probable state for a given particle. the state in which the number of
ways of distributing the energy among all particles is greatest, is one of lower energy.

Figure 9.3: Probabilities of a given oscillator being in its ni state and Boltzmann proba-
bility.

Equation (9.3) is cumbersome and limited to the special case of harmonic oscillators.
However, it may be shown that in the limit of a large systems of distinguishable particles,
all cases converge to the Boltzmann probability:

P (En ) = Ae−En /kB T (9.4)

This is the probability that in a large system at temperature T , an individual particle will
be in a state of energy En , where n stands for the set of quantum numbers necessary to
Chapter 9 Statistical Mechanics 152

specify the individual-particle state, e. g. (n, `, m` , ms ). Probability drops exponentially


with energy.

It is important to note that the Boltzmann probability and all the distributions yet to
come can be applied to subsystems of large systems. In small subsystems, there might be
significant fluctuations, but the prediction still holds on average, becoming more precise
as the subsystem grows. The larger remainder of the overall system is what we call
reservoir. The temperature is a well-defined property of the overall macroscopic system.

Let’s make a quantitative comparison of (9.3) and (9.4). Doing so require a diversion,
but a very important one, for it spotlights one of the most basic tasks in statistical
mechanics: finding an average. We begin by noting the probability summed over all
individual-particle states n must be 1.
X X
P (En ) = Ae−En /kB T = 1
n
1
⇒ A= X
e−En /kB T
n

This step enable us to eliminate the multiplicative constant A and write the probability
as:
e−En /kB T
P (En ) = X (9.5)
e−En /kB T
n

As always, the average of a quantity is the sum of the quantity’s possible values times the
probabilities for each. So the average energy of a particle is:
X
En e−En /kB T
X
E= En P (En ) = nX (9.6)
e−En /kB T
n

Now returning to the comparison of equations (9.3) and (9.4). Inserting (9.1) into
(9.6), we have

X
n~ω0 e−n~ω0 /kB T
n=0 ~ω0
E= ∞ = (9.7)
X e~ω0 /kB T − 1
e−n~ω0 /kB T
n=0

This result provides the link between temperature found in the Boltzmann probability
(9.4), and the integers M and N of the exact probability (9.3). We merely ensure that
Chapter 9 Statistical Mechanics 153

the average energies agree. The average energy in the exact approach is E = E/N =
M ~ω0 /N . Solving this by substituting E in (9.7) gives us the link:
~ω0
kB T =
ln(1 + N/M )
Inserting this expression and (9.1) back into (9.5) then gives the Boltzmann probability
in terms of M and N :
N
P (En ) = e−n ln(1+N/M ) (9.8)
M +N
As shown in Fig. 9.3, the exponential Boltzmann probability (9.9) agrees quite well with
exact probabilities (9.3). The slight deviation is due to the fact that the N we have
considered is so small. In a realistic macroscopic limit, the messy factorials and smooth
Boltzmann function coincide precisely.

Fig. 9.4 shows the Boltzmann probability versus energy for different temperatures T .
The curves represents cases in which the average particle energy is less than, comparable
to, and greater than the first excited-state energy. Sensibly, at the low temperature, we
see that it is quite unlikely for a given particle to be in any energy level higher than the
ground state. However, at the high temperature, occupation of the ground state is not
much more likely than occupation of the second, the third, and so forth.

Figure 9.4: Variation of Boltzmann probability from kB T < ~ω0 to kB T > ~ω0 .

Often we are interested not so much in the probability of occupying a given state as
in the number of particles expected to occupy that state, referred to as the occupation
Chapter 9 Statistical Mechanics 154

number, N . In fact, the number expected in a given state is just the probability of occu-
pying that state times the total number of particles N . Thus, the Boltzmann distribution
follows directly from (9.4):

N (En )Boltz = N Ae−En /kB T (9.9)

Logically, summing probability (9.4) gives 1, summing (9.9) must give N .

Before going any further with the Boltzmann distribution, we call attention to a
distinction often overlooked but crucial to applying all energy distributions. From their
derivations, we find that they give a number of particles in a given state of energy E, not
the number of a given energy E. If there are multiple state of the same energy, then

(number of particles of energy E)


= (N , number of particles in a given state of energy E) (9.10)
× (number of states of energy E)

Example 9-1

Consider a sample of hydrogen atoms at a temperature of 300 K.

(a) What is the ratio of the number of atoms in the n = 2 energy levels to those in n = 1?
1
(b) At what temperature would the ratio be 10
?

Solution:

(a) Hydrogen’s degeneracy is 2n2 — the n2 due to different ` and m` values and the 2
due to the electron’s two allowed values of spin, ms . Thus the n = 2 level has eight
different states with the same energy. Equation (9.9) says that the number of atoms
in, for instance, the state (n, `, m` , ms ) = (2, 1, 1, 21 ), is:
!
E2,1,1, 1
N (E2,1,1, 1 ) = N A exp − 2
2 kB T

All other n = 2 states have the same energy E2 . Similarly, there are two n = 1
states, (1, 0, 0, 12 ) and (1, 0, 0, − 21 ), of the same energy E1 . So the ratio of the number
of atoms in n = 2 states to those in n = 1 states is:
number with energy E2 N (E2 ) × 8 4 N A e−E2 /kB T
= = = 4e−(E2 −E1 )/kB T
number with energy E1 N (E1 ) × 2 N A e−E1 /kB T
Chapter 9 Statistical Mechanics 155

The hydrogen atom energies are −13.6/n2 eV. So


   
13.6 13.6
E2 − E1 = − 2 eV − − 2 eV = 10.2 eV
2 1
and at 300 K,

kB T = (1.38 × 10−23 J/K)(300 K) = 4.14 × 10−21 J = 0.0259 eV

Therefore,
number with energy E2
= 4e−10.2/0.0259 ∼
= 10−171
number with energy E1
At room temperature, the probability of finding even a single hydrogen atom in an
excited state is practically zero.
1
(b) Using the result in part (a), if the ratio is 10
,

0.1 = 4e−10.2 eV/kB T ⇒ T ∼


= 32000 K

The Sun’s surface is about 6000 K, giving a ratio of about 10−8 . We conclude that
except at extremely high temperatures, the overwhelming majority of hydrogen atoms
are in their ground states.

For the sake of consistency with things to come, let us reexpress the average in (9.6)
in terms of the occupation number (9.9) and leave off the subscript on N to allow for
other distributions later. Thus,
X
En N (En )
n
E= X (9.11)
N (En )
n

Often in statistical mechanics, the spacing of the quantum levels is much smaller than
typical particle energies. If quantum levels are indeed closely spaced, a sum over states
may be replaced by an integral.

However, a quantum number is rarely the most convenient integration variable (i. e. ∆n
→ dn). Energy is a logical alternative, but here we must be careful. Suppose we wish to
calculate an average energy in a collection of hydrogen atoms via equation (9.11). In the
numerator, the sum can be written out as:
X X
En N (En ) = En N A e−En /kB T
n n

= E1 N A e−E1 /kB T × 2 + E2 N A e−E2 /kB T × 8 + . . . (9.12)


Chapter 9 Statistical Mechanics 156

Since the energy actually depends only on thr quantum number n, this sum over states
can be written as a sum over energies as follows:

E1 N A e−E1 /kB T × 2 + E2 N A e−E2 /kB T × 8 + . . .


X
= En N Ae−En /kB T × (number of states of energy En )
En
X
= En [N (En ) × (number of states of energy En )] (9.13)
En

To proceed to an integral, we would replace the discrete En by a continuous variable E,


but we still need to include the number of states at the given energy. In fact, this just
restates the principle in (9.10).

The final step in going from the sum in (9.11) to an integral must somehow give us a
dE in the integrand. It arises naturally. If E indeed becomes a continuous variable, the
number of states at a particular energy is differentially small, and we naturally account
for this through the number of states per range of energy. That is to say,

differential number of states within range dE of E


differential number of states within range dE of E
= dE (9.14)
dE
The left side of (9.14) is what appears in brackets in (9.13), but we still instead insert
the right side. In this way, we avoid choosing a new symbol for the left side, which could
possibly be confusing. More important, the noninfinitesimal quotient multiplying the dE
on the right side contains the truly important information, as we will see. It is given the
symbol D(E) and the descriptive name density of states. That is to say,

differential number of states within range dE of E


D(E) ≡ (9.15)
dE
Note that density of states has dimensions “one over energy”. Thus, applying (9.12)
through (9.15), average (9.11) in the limit of closely spaced energy levels becomes:
R
E N (E)D(E)dE
E= R (9.16)
N (E)D(E)dE

Actually determining densities of states can be complicated and depends on the system
under study. As a first example, consider the simplest case, a system of oscillators with
potential energy shifted by − 21 ~ω0 , ignoring spin. For this system, the quantization
Chapter 9 Statistical Mechanics 157

condition is E = n~ω0 . Therefore, the number of states in an energy range dE is


 
E 1
dn = d = dE
~ω0 ~ω0
differential number of states 1
⇒ D(E) = = (9.17)
dE ~ω0

Next we consider a slightly different case: a collection of particles in a box. In this case,
E would again be proportional to a single quantum number n, but squared. Contrary to
the oscillator’s equally spaced energies, the levels get farther apart as n increases. By the
same procedure as above, it can be shown that if E ∝ n2 , then D(E) ∝ E −1/2 . It does
decrease as energy increases.

We close the section with a basic question: What would be the average energy in a
system of oscillators if the temperature were high enough that the quantum levels could
be considered “closely spaced”? Inserting (9.9) and (9.17) into (9.16), we obtain:
R∞
EN A e−E/kB T (1/~ω0 )dE
E = R0 ∞ = kB T (9.18)
0
N A e−E/kB T (1/~ω0 )dE

Is our switch to integration valid? In the limit of closely spaced energy levels, ~ω0  kB T ,
the average energy obtained via summing (9.7) becomes:

~ω0
E= lim = kB T
~ω0 /kB T →0 e 0 B T
~ω /k −1

It agrees with the prediction obtained by the integral approach.

9.3 Quantum Distributions

In the previous section, when we count the number of ways of distributing energy for a
number of particles, we implicitly assumed that switching the labels of any two particles
yielded a different way of distributing energy. However, if the particles are indistinguish-
able, then these are not different ways. And if the number of ways are different quantum
mechanically than classically, then probabilities will also be different. From these ideas,
perhaps seeming of little practical interest, arise many startling and far-reaching conse-
quence.

Recall from Chapter 8 that there are two different types of indistinguishable particles:
bosons (integral spin) and fermions (half-integral spins). We now investigate how and why
Chapter 9 Statistical Mechanics 158

the distributions of energy in these two should differ from the Boltzmann distribution and
from each other. To illustrate this idea, we consider a simple case: four particles, a, b,
c, and d, bound by a harmonic oscillator potential energy. It is convenient to define the
ground-state energy as 0 and the spacing between levels as δE.

Ei = ni ~ω0 = ni δE (ni = 0, 1, 2, . . .)

Now suppose that the system has total energy 2δE, that is,

na + n b + n c + n d = 2

Table 9.1 shows all possible ways for distinguishable particles under this constraint. We
see that the smallest quantum number is most common; the largest, least. As always, the
probable number of particles in a given state decreases with increasing energy.

Table 9.1: Ways of distributing energy 2δE among distinguishable particles a, b, c, d.

Number of Probability Probable number


times n P of particles
n Ways appears (#) (#/40) (P × 4)
2 a b c d 4 0.1 0.4
1 ab ac ad bc bd cd 12 0.3 1.2
0 bcd acd abd abc cd bd bc ad ac ab 24 0.6 2.4
Totals 40 1.0 4.0

If the particles are indistinguishable bosons, however, there are no labels, and truly
different states are distinguished only by having different numbers of particles at different
levels. Table 9.2 shows how the energy distributed in this case. Again, the probable
number of particles drops with increasing energy, but in a different way. It would seem
that bosons “like” being together in the ground state more than classical particles do. Yet
1
another way is exhibited by indistinguishable s = 2
fermions, for which any distribution
with more than two particles in the same n is strictly forbidden by the exclusion principle.

Now let us move on from four-particle case to truly thermodynamic systems. For
distinguishable particles, detailed, case-specific probabilities converge to the Boltzmann
distribution. For quantum-mechanically indistinguishable particles, they converge to two
different distributions: the Bose-Einstein for bosons and the Fermi-Dirac for fermions.
Chapter 9 Statistical Mechanics 159

Table 9.2: Ways of distributing energy 2δE among four indistinguishable bosons and
spin- 21 fermions.

Bosons
Number of times Probability P Probable number of
n Ways n appears (#) (#/40) particles (P × 4)
2 X 1 0.125 0.5
1 XX 2 0.250 1.0
0 XXX XX 5 0.625 2.5
Totals 8 1.000 4.0
Fermions
Number of times Probability P Probable number of
n Ways n appears (#) (#/40) particles (P × 4)
2 0 0.0 0
1 XX 2 0.5 2
0 XX 2 0.5 2
Totals 4 1.0 4

Although introducing no new principles, the derivations of these quantum distributions are
rather involved. Here we merely present the distributions and study how they compare.


 1 Boltzmann

 E/k T


 Be B (distinguishable)


Case-specific large N 1 Bose-Einstein
−→ N (E) = (9.19)
probabilities E/k
 Be B − 1
 T (bosons)


 1


 Fermi-Dirac

Be E/k B T +1 (fermions)
Each is a probable number of particles occupying a given state. All are decreasing function
of E — the higher the state’s energy, the fewer will be the particles occupying it. And in
all cases, B is related to the total number of particles N . The most conspicuous difference
is whether −1, 0, or +1 is added to the exponential in the denominator. The consequence
can be drastic.

Demonstrating this quantitatively involves a determination of B, for although it is in-


dependent of E, B does in general depend on T , sometimes in a complicated way. Fig. 9.5
shows the three distributions in a simple system of one-dimensional oscillators. Whether
Chapter 9 Statistical Mechanics 160

the temperature qualifies as high or low is measured relative to a constant E defined as


the topmost energy level that would be occupied if particles filled all individual-particle
states, one particle per state, “upward” from the ground state. Of course, only fermions
obey an exclusion principle, but E nonetheless serves as a good indicator of how spread
out among energies we expect particles to be. If kB T is much greater than E, we would
expect particles to be spread out so diffusely among energy levels that there would be
little chance of any two occupying the same state. So it shouldn’t matter whether the
particles are bosons, fermions, or classically distinguishable. And the quantum distribu-
tions thus agree with the classical. If kB T is much less than E, suggesting that particles
might tend to “pile up”, bosons should show their enhanced tendency to crowd together
into the ground state, while fermions should scrupulously avoid occupying the same state.
Fig. 9.5 bears out these expectations perfectly.

Figure 9.5: The three distributions for oscillators at low, intermediate, and high temper-
ature.
Chapter 9 Statistical Mechanics 161

Plots of average particle energy, shown in Fig. 9.6, are also illuminating. At high
temperatures, kB T  E, both boson and fermion cases are essentially plots of a linear
relationship E = kB T , the same as the classical result (9.18). The particles are spread
so diffusely among states that indistinguishability is unimportant. Even so, the energy is
somewhat lower for bosons, which tend to congregate at lower levels more than would be
expected classically, and is somewhat higher for fermions, which avoid already-occupied
lower energy states. At lower temperatures, the average energy for bosons approaches 0,
while that for fermions approaches 21 E, the logical average when equally spaced levels fill
uniformly from 0 to a maximum of E.

Figure 9.6: Average oscillator energy versus temperature for the three types of particles.

The smooth curves in Fig. 9.6 rely on the assumption that integration is valid —
that kB T is comparable to E, which is a fairly high quantum level. But at very low
temperatures, where kB T is comparable to the spacing ~ω0 between individual levels,
summation would be required. The boson case is particularly sensitive; the minus sign of
N (E)BE in the denominator of (9.19) causes a pronounced increase in N (E) near E = 0.
One result is that the vast majority of particles in a boson gas may drop into the ground
state, leading to “superfluid” behavior. But let us first take a look at an aspect of the
Fermi-Dirac distribution that has nothing to do with the specific system.

As Fig. 9.5 shows, at low temperature, the Fermi-Dirac occupation number is nearly
1 out to a certain energy, where it quickly drops to nearly 0. This energy is known as the
Fermi energy EF . Strictly speaking, the Fermi energy is defined as the energy at which
the occupation number is 12 . The convenience of this definition is that the constant B of
N (E)FD in (9.19) may be written in terms of EF , no matter what the system might be.
Chapter 9 Statistical Mechanics 162

1
Given that N (E)FD = 2
at E = EF , we have
1 1
N (EF )FD = = ⇒ B = e−EF /kB T
Be+EF /kB T +1 2
Thus,
1
N (E)FD =
e(E−EF )/kB T
+1
In this form, it is easy to see that in the limit T → 0, the Fermi-Dirac distribution is a
“step function”: 1 for E < EF and 0 for E > EF .

Since it is really just another way of writing B, which is, in general, temperature
dependent, the Fermi energy does vary with temperature, but its value at T = 0 is most
important. Fig. 9.7 for the oscillator case shows why. AT T = 0, the occupation number
drops to zero abruptly at EF 0 . When kB T is not 0 but still small compared with EF 0 ,
1
the plot passes through 2
at nearly the same energy — the Fermi energy is essentially
constant. Moreover, it deviates from a true step function nearly symmetrically about
the Fermi energy. Only when kB T becomes comparable to EF 0 is the bulge significantly
1
depleted and the plot passes through 2
at significantly lower energy. When we speak of a
Fermi energy, we are usually referring to very low temperature, so it is justifiable to refer
to the Fermi energy as if it were constant.

Figure 9.7: The Fermi-Dirac distribution at T = 0 and above. For low temperatures, the
Fermi energy EF differs little from its T = 0 value.

We can derive the expression of EF as a function of T on which Fig. 9.7 is based and
also verifies that its T = 0 limit, EF 0 , is what it should be: the highest occupied energy in
one-per-state filling. Although our observations here have been based mostly on a system
of simple oscillators, the qualitative features are common to all systems of fermions.
Chapter 9 Statistical Mechanics 163

9.4 The Quantum Gas

We now apply the quantum distributions to a more realistic situation. The system we
consider here is one in which N massive particles are free to move in three dimensions,
as are molecules in a classical gas. However, we are not considering ordinary gases, but
systems in which particle concentration is high enough that quantum indistinguishability
must be taken into account. It may be a gas of bosons or a gas of fermions. The particles
are assumed to be moving freely, yet they are still bound to some region so that they
occupy discrete quantum states. The logical model is the infinite well, in which the
particles experience no force except at the confining walls. Of course, to allow for motion
in three dimensions, we must use the 3D well.

Before looking at the quantitative side, a crude yet useful picture summarizes what
we might expect to find. Figure 9.8(a) depicts a relatively high temperature, with speedy,
short-wavelength particles bouncing about their well and enjoying their distinct individ-
ually. Whether they are bosons or fermions matters little, for they are spread out many,
many states. When things get cold, however, wavelengths get longer, freedom is restricted,
and the line between boson and fermion becomes clear. Each fermion claims its own state,
while bosons indulge their preference for piling together in the same lowest state.

Figure 9.8: Classical distinguishable particles at high T , and two different low-T behaviors.
(Adapted from Nobel lecture by Wolfgang Ketterle, Rev. Mod. Phys. 74, October 2002.)

So far as calculations go, we will mostly consider circumstances where quantum states
are closely spaced and integration is valid, so we will need the density of states for our
Chapter 9 Statistical Mechanics 164

system. In chapter 5, the allowed energies in a 3D infinite well are shown to be:

π 2 ~2
Enx ,ny ,nz = (n2x + n2y + n2z ) (9.20)
2mL2
How do we determine the number of quantum states in an energy interval dE? This is
one of those junctures in physics where we learn that the best resort is a trick: Picture the
set (nx , ny , nz ) as coordinates in three-dimensional “quantum number space”, depicted in
Fig. 9.9, with each point representing a quantum state and a unit volume thus enclosing
one state. The radial “distance” n from the origin in this space would be given by:
q
n = n2x + n2y + n2z

Therefore, r
π 2 ~2 2 2mL2 E
E= n ⇒ n = (9.21)
2mL2 π 2 ~2
Sensibly, the “radius” increases with energy. Now, if the energy increases by dE, the
radius increases by a corresponding dn, so an energy interval of dE encloses a “volume”
1
in quantum-number space of 8
4πn2 dn. But a volume in quantum-number space is a
number of quantum states, so we may write
1
differential number of states in energy range dE = 4πn2 dn
8

Figure 9.9: Allowed states in a 3D infinite well. The shell encloses all states in range dn
at “radius” n — in this case, n = 5.
Chapter 9 Statistical Mechanics 165

Thus by definition,
differential number of states in energy range dE 1 dn
D(E) = = 4πn2
dE 8 dE
Using relationship (9.21) between n and E, n may be eliminated, leaving

m3/2 L3 1/2
D(E) = √ E
π 2 ~3 2
We must multiply D(E) by 2s + 1 to account for the multiple allowed spin states. There-
fore, in terms of V = L3 , we have

(2s + 1)m3/2 V 1/2


D(E) = √ E (9.22)
π 2 ~3 2

From (9.22), we can calculate the familiar average particle energy. Unfortunately, the
parameter B in the quantum distributions in (9.19) depends on the system, the complexity
of these formulas and D(E) prevents a closed-form calculation of B and therefore, of E(T )
for the quantum gas. However, even an approximate solution is revealing. The result is:
 3 3
  
3 π ~ N
E∼ = kB T 1 ∓ +... (9.23)
2 (2s + 1)(2πmkB T )3/2 V

In contrast to our oscillator system, in which volume was not even a consideration, the
average energy in a quantum gas depends on both temperature and particle density N/V .
Squeeze the particles together – increase N/V — and the average fermion (bottom sign)
energy increases, while for bosons it decreases. Note that this approximation is good only
when T is not too low and the particle density is not too high.

Let’s turn to study another result from (9.22) that has many applications. It is based
on how diffusely the particles are spread out among the energy levels. To begin, we
calculate the Fermi energy at T = 0. Note that now we view the Fermi energy as a one-
particle-per state standard. If the energy of a system (no matter what kind of particle),
is much greater than this standard, it is unlikely that multiple particles inhabit any state,
and classical behavior prevail. The starting point is an expression for the total number
of particles: Z ∞
N= N (E)FD D(E)dE
0
A great simplification is that we consider T = 0, where N is a step function: 1 for E < EF
and 0 for E > EF . Thus, we simply integrate the number of states per energy D(E) from
Chapter 9 Statistical Mechanics 166

0 to EF and solve for EF .


Z EF  
(2s + 1)m3/2 V 1/2 (2s + 1)m3/2 V 2 3/2
N= √ E dE = √ E
0 π 2 ~3 2 π 2 ~3 2 3 F
 2/3
π 2 ~2 3 N
⇒ EF = √ (9.24)
m (2s + 1)π 2 V
This result says that even at zero temperature, the greater the density of fermions, the
higher will be their energies. We can show how it matches with (9.23). If we rearrange
(9.24) a little bit, we have
~3 N 3/2
∝ EF
(2s + 1)m3/2 V
Thus we see that the second term in (9.23) is proportional to (EF /kB T )3/2 . If kB T  EF ,
only the classical first term survives. The particles are spread so diffusely among states
that quantum indistinguishability can be ignored.

Summarizing, a gas will behave classically so long as the “quantum term” in (9.23) is
small, which can be expressed as:
 3
λ N ~3
 1 or 1 Classical limit (9.25)
d V (mkB T )3/2
where d (∝ (V /N )1/3 ) is the linear distance between particles. When quantum effects
begin to rise, the average energy in a boson gas is lower than in a classical one. Being
related to average energy, the pressure is also lower. Conversely, the average energy and
pressure in a fermion gas is higher than in a classical gas.

Example 9-2

Show that the nitrogen molecules in the atmosphere need not to be treated as indistin-
guishable particles.

Solution:
Equation (9.25) should provide the answer. If air does indeed behave classically, the ideal
gas law allows us to replace N/V by P/kB T . So the L.H.S. of (9.25) becomes:
P ~3 P ~3
=
kB T (mkB T )3/2 m3/2 (kB T )5/2
Now inserting 80% — nitrogen’s share — of atmospheric pressure, a plausible room tem-
perature at 300 K, and the molecular mass of diatomic nitrogen, we arrive at:
P ~3 (0.8 × 1.01 × 105 Pa)(1.055 × 10−34 J · s)3
= = 8.59 × 10−9
m3/2 (kB T )5/2 (28 × 1.66 × 10−27 kg)3/2 [(1.38 × 10−23 J/K)(300 K)]5/2
Chapter 9 Statistical Mechanics 167

The ratio is much less than 1, as it would clearly also be for oxygen and for any other
naturally occurring temperatures, so the criterion for classical behavior is met.

From the above example, we see that our atmosphere is “classical”. But in many
important bound systems where quantum indistinguishability cannot be ignored, our 3D
infinite well model yields excellent predictions. One application arises in cosmology. By
treating them as systems of gravitational bound fermion and using result of (9.24), we may
predict the radii of so-called white dwarf stars and neutron stars. Another very common
application is that of the electrostatically bound conduction electrons in a metal.

In metals, some of each atom’s valence electrons are shared among all atoms, moving
essentially freely in three dimensions through the solid. They are known as conduction
electrons and are good example of a fermion gas. This “gas” is confined in a potential
well whose origin is the net effect of the potential energies of the positive ions spaced
regularly throughout the solid. In a crystalline solid, the density of positive ions is so
great that a finite well becomes an excellent approximation. Often the even simpler
infinite well model we have developed is valid. Accepting this, let us investigate whether
the electron gas in a typical metal must indeed be treated as a quantum gas.

Example 9-3

Assume that the conduction electrons in a piece of silver behave as a fermion gas, with
each atom contributing one electron. Calculate the Fermi energy, then compare it with
the energy that might be expected classically at room temperature. (The density of silver
is 10.5 × 103 kg/m3 .)

Solution:
Equation (9.24) gives the Fermi energy in terms of properties easily determined. We know
the mass and spin of electrons, but we must also determine the number of conduction
electrons per volume. For one conduction electron per atom, this is equivalently the
number of silver atoms per unit volume. So we have
N mass/volume 10.5 × 103 kg/m3
= = −27
= 5.86 × 1028 m−3
V mass/atom (107.9 u)(1.66 × 10 kg/u)
Thus,
 2/3
π 2 ~2 3 28 −3
EF = √ (5.86 × 10 m ) = 8.82 × 10−19 J = 5.51 eV
9.11 × 10−31 kg 2π 2
Chapter 9 Statistical Mechanics 168

Were we apply classical physics to the conduction electron gas, the average energy
would be 23 kB T . At a room temperature of 300 K, this is only 0.04 eV, and we have shown
that even at zero temperature, electrons in silver must fill states up to 5.51 eV! Here we
have an excellent example of kB T  EF , where (9.23) would be an awful approximation.
1
Because kB T at room temperature is only about E ,
100 F
the occupation number distri-
bution would resemble the extreme non-classical T = 0 plot of Fig. 9.7. To electrons in
a metal, room temperature is very cold. Temperatures hundreds of times higher would
be needed before they could spread out among higher energy levels, and because such
temperatures would vaporize most metals, we conclude that conduction electron behavior
is a decidedly quantum-mechanical affair.

That the exclusion principle forces conduction electron to fill up states unusually high
by classical standards has a famous consequence, one that puzzled physicists terribly
before quantum mechanics matured. The electrons’ freedom to move about in a metal
was obvious, and if behaving as a classical gas, that freedom should increase the heat
capacity of metals relative to nonmetals. But the electrons’ contribution is in fact very
small. The reason is that most of them, sandwiched between states already occupied by
others, simply cannot gain energy from modest increase in temperature. Only those few
at the top near the Fermi energy, with some “room” to spread out, can gain energy. Thus,
the majority of electrons contribute nothing to the energy storage.

Figure 9.10: Electron energies in a “cold” metal.


Chapter 9 Statistical Mechanics 169

Other properties of metals also come into clearer focus. Work function, the minimum
energy required to remove an electron from a metal, is simply the energy difference be-
tween the Fermi energy and the top of the finite well in which the conduction electrons
are bound, pictured in Fig. 9.10. In other words, the work function

φ = U0 − EF

9.5 Massless Bosons: The Photon Gas

Photons, being spin-1, are bosons. The quantum density of states (??) isn’t applicable
for photons since it is based on energy quantization condition (9.20) for massive particles.
For a photon gas, the correct density of states is:
differential number of states in range df 8πV
= 3 f2
df c
Using E = hf , we convert this to the usual number per unit energy:
 2
differential number of states in range dE 8πV E
= 3
dE/h c h
8πV
⇒ D(E) = 3 3 E 2 (9.26)
hc

The photon application differs in another important way. Whereas situation abound
in which a fixed number of massive objects is confined to some volume, this isn’t the case
for photons. All objects are constantly emitting and absorbing photons. So the number of
particle in a photon gas is determined by whatever conditions lead to equilibrium between
the photons and the surrounding matter. But this would seem to lead to difficulty: How
do we use the Bose-Einstein distribution if the number N is not fixed, for we have said that
the distribution’s parameter B is related to the total number of particles? A resolution
is setting B = 1. Distribution in hand, we are ready to use the tools we have developed.

Example 9-4

Internal energy in the form of electromagnetic radiation permeates all matter, since all
matter contains oscillating charged particles. How do the energies of the radiation and
matter compare?

(a) First calculate the total photon internal energy in a volume V , calling it Uphoton .
Chapter 9 Statistical Mechanics 170

Z ∞
x3 π4
(Hints: Use the integral formula dx = .)
0 ex − 1 15
(b) Assume that an ideal monatomic gas at STP (273 K, 1 atm) is in equilibrium with
electromagnetic radiation. Find the ratio of the photon internal energy to the internal
kinetic energy Umatter of the gas.

Solution:
(a) As noted earlier, the numerator in (9.15) is the total energy. To calculate it, we need
N (E) and D(E) — equations (9.19) and (9.26) respectively, with B = 1.
Z  
1 8πV 2
Uphoton = E E dE (9.27)
e B − 1 h3 c 3
E/k T

We allow for all possible energies from 0 to infinity, so we may use the integral
provided if we make the definition: E = kB T x. Thus
4 4 Z ∞
 4
8πV kB T x3 8πV kB4 4
T π 8π 5 V kB4 4
T
Uphoton = 3 3 x
dx = 3 3
= 3 3
(9.28)
hc 0 e −1 hc 15 15h c
We see that the energy in a photon gas grows as temperature to the fourth power. As
a rule, we expect average particle energies to be proportional to T to the first power,
and this is indeed the case for the photon gas. The total grows so quickly because
the number of photons grows as the third power of T .
(b) For electromagnetic radiation,
8π 5 V (1.38 × 10−23 J/K)4 (300 K)4
Uphoton = = 6.09 × 10−6 J/m3 × V
15(6.63 × 10−34 J · s)3 (3 × 108 m/s)3
The kinetic energy of the gas is the number of atoms N times the average energy of
3
k T.
2 B
Using the ideal gas law,
3 3 3
Umatter = N × kB T = P V = (1.01 × 105 Pa) × V
2 2 2
The ratio is:
Uphoton 6.09 × 10−6
= 3 5)
= 4.02 × 10−11
Umatter 2
(1.01 × 10
We conclude that under ordinary conditions, the fraction of a thermodynamic sys-
tem’s total energy in the form of electromagnetic radiation is very small.

Quantum statistical mechanics provides the crucial link between temperature and
electromagnetic radiation. Previously, the nature of the electromagnetic radiation com-
ing from an object of a given temperature was mostly a matter of observation. For
Chapter 9 Statistical Mechanics 171

example, the Stefan-Boltzmann law is a very important empirical law because it governs
heat transfer via electromagnetic radiation, yet it lacked theoretical justification. Accord-
ingly, equation (9.28), from which the law follows, was one of statistical mechanics’ more
celebrated early achievements.

To gain further insight, let us examine the integrand in (9.27). It is the contribution
to the total electromagnetic energy of those photons whose energies lie in a small range
dE around E.  
1 8πV 2
dUphoton = E E dE
eE/kB T −1 h3 c 3
Using E = hf , we may reexpress this in terms of f , giving the contribution due to the
photons in the frequency range df around f :
 
hf 3 8πV
dUphoton = hf /k T
df (9.29)
e B −1 c3

This is Planck’s spectral energy density. Clearly, the total photon energy has a maximum
at the frequency which is directly proportional to the temperature. This is the essence of
Wien’s law which was originally put forth without theoretical justification.

It is worth noting that our derivation of equation (9.29) here treats electromagnetic
radiation as indistinguishable boson particles. It differs considerably from the alternative
approach in which the radiation is treated as a standing waves, which are distinguishable,
justifying use of the Boltzmann distribution.

9.6 The Laser

The term laser stands for Light Amplification by Stimulated Emission of Radiation. It
is a device that produces a light beam with some remarkable properties:

1. The light is coherent — it travels in only one direction, is of a single wavelength,


and is exactly in phase (see Figure 9.11). An interference pattern can be obtained
not only by placing a double slit in a laser beam but also by using beams from two
separate laser.
2. A laser beam diverges hardly at all. A laser beam sent from the earth to a mirror left
on the moon by the Apollo 11 expedition remained narrow enough to be detected on
its return to the earth.
Chapter 9 Statistical Mechanics 172

Figure 9.11: Coherent versus incoherent light.

3. The beam is extremely intense, more intense by far than the light from any other
source. To achieve an energy density equal to that in some laser beams, a hot object
would have to be at temperature of 1030 K.

Figure 9.12: Three different photon-atom interactions.

There are three means by which radiation can interact with the energy levels of atoms
(depicted in Figure 9.12). If the atom is initially in the lower state E0 , it can be raised
Chapter 9 Statistical Mechanics 173

to the state E1 by absorbing a photon of energy E1 − E0 = hf . This process is called


absorption. If the atom is initially in the upper state E1 , it can drop to the state E0
by emitting a photon of energy hf . This is spontaneous emission. Einstein was the
first to point out a third possibility, stimulated emission. During this process, an
incident photon of energy hf causes a transition from E1 to E0 . In stimulated emission,
the radiated light waves are exactly in phase with the incident one. They also have the
same energy and move in the same direction as the incident radiation. So the result is an
enhanced beam of coherent light.

To understand how we exploit stimulated emission to produce a coherent light source


of many photons, we need to consider the exchange and balance of energy between mat-
ter and electromagnetic radiation with which it interacts. As we do so, an intriguing
conclusion arises.

Let us assume that the matter is a gas of atoms and restrict our attention to transition
between just two of their energy states, as in Fig. 9.12. The lower state we refer to as level
1 and the upper as level 2. It is reasonable that the rate of spontaneous emission, Rspon ,
should be proportional to the number of atoms poised to drop down from the upper level,
N2 . Denoting the proportionality constant as Aspon , we have

Rspon = Aspon N2

Similarly, the rate of absorption, Rabs , should be proportional to the number of particles
in the lower level that are poised to be promoted, N1 . But it must also depend on the
number of nearby photons that are of just the right energy to promote them. This energy
would naturally be E2 − E1 = ∆E. For the number of photons of this energy, let us
temporarily use the simple symbol Y (∆E). Therefore,

Rabs = Babs N1 Y (∆E)

Here we use Babs for the proportionality constant. It was Einstein who first charted
the course we follow, the coefficients are known universally as the Einstein A and B
coefficients. Finally, the rate of stimulated emission, Rstim , should depend on the number
at the upper level available for emission, N2 , and again on the number of photons of just
the right energy, Y (∆E). That is to say,

Rstim = Bstim N2 Y (∆E)

where Bstim is the proportionality constant.


Chapter 9 Statistical Mechanics 174

Now assuming equilibrium between atoms and photons, the rate at which atoms make
upward transitions from level 1 to level 2 must equal the downward transition rate. This
is known as the principle of detailed balance. Simply put, the two emission rates
must add to the absorption rate:

Aspon N2 + Bstim N2 Y (∆E) = Babs N1 Y (∆E)

Solving for Y (∆E),


Aspon /Babs
Y (∆E) =
N1 /N2 − Bstim /Babs
As the gas atoms are diffuse, they obey the Boltzmann distribution. So the ratio N1 /N2
is equal to e−E1 /kB T /e−E2 /kB T = e∆E/kB T . Thus,
Aspon /Babs
Y (∆E) = ∆E/k
e BT − B
stim /Babs

Here is what “intriguing conclusion” arises. We defined the quantity Y (∆E) as the
number of photons of energy ∆E, but we know what it must be! It is the photon gas
spectral energy density (9.29), with ∆E replacing hf in that formula. The e∆E/kB T is in
place where it needs to be, and the only way the rest of the denominator can be correct is
if Bstim = Babs . Of course, we don’t want the rates of absorption and stimulated emission
equal in a laser. We want stimulated emission to predominate, but how?

In fact, simply making making N2  N1 would do it — Rstim would be much larger


than Rabs – but such a situation is certainly not the way things are in equilibrium, where
number drops with energy. And such an overpopulation of higher levels is crucial to
laser operation and is known as population inversion. Being unnatural, it must be
established by some external means. But so long as it exists, one photon of the proper
frequency, perhaps the result of spontaneous emission, may become two, which become
four, and so on, very quickly resulting in a large number of coherent photons. This is the
basis mechanism of the operation of laser.

The first laser was constructed by T. H. Maiman in 1960. It is based on a three-


level atom as shown in Figure 9.13. The laser medium is a solid ruby rod, in which
the chromium atoms are responsible for the action of the laser. The atoms, originally
in the ground state, are “pumped” into the excited state by an external energy source.
The excited state decays very rapidly to a lower excited state, which is a metastable
state. The atom remains in a metastable state for a relatively long times, perhaps 10 −3 s,
compared with 10−8 s for the short-lived states. As a result, N2 is greater than N1 , and
the situation is ripe for a cascade of stimulated emissions between E2 and E1 .
Chapter 9 Statistical Mechanics 175

Figure 9.13: Energy transitions in the three-level laser.

If the pumping action is successful, there are more atoms in the metastable state
than in the ground state, and a population inversion is achieved. However, as the lasing
transition occurs, the population of the ground state is increased, thereby upsetting the
population inversion. This excess of population in the ground state allows absorption of
the lasing transition which removes photons that might contribute to the lasing action.

Figure 9.14: Energy transitions in the four-level laser.

The four-level laser illustrated in Fig. 9.14 relieves this difficulty. The ground state
is pumped to an excited state that decays rapidly to a metastable state, just like the
three-level laser. The lasing transition proceeds from the metastable state to yet another
Chapter 9 Statistical Mechanics 176

excited state, which in turn decays rapidly to the ground state. The atom in its ground
state thus cannot absorb the energy of the lasing transition, and we have an efficient laser.

Figure 9.15: Schematic diagram of a He-Ne laser.

The helium-neon laser is an example of a four-level laser. A mixture of helium and


neon gas (about 90 % helium) is contained in a narrow tube, as shown in Figure 9.15.
An electrical current in the gas “pumps” the helium from its ground state to the excited
state at an energy of about 20.6 eV. Mirrors are carefully aligned at the ends of the tube
to help the formation of the coherent wave, as it bounces back and forth between the
two ends of the tube, causing additional stimulated emission. One of the mirrors is only
partially silvered, allowing a portion of the beam to escape through one end.
Appendix A

X-ray Diffraction

X-ray wavelength is about 10−10 m which is of the same order of magnitude as the atomic
spacing in a crystal. So a beam of X-rays sees the regular spacing of the atoms in a
crystal as a sort of three-dimensional diffraction grating. In view of this, Max von Laue
(1879-1960), William H. Bragg (1862-1942) and William L. Bragg (1890-1971) suggested
using single crystal such as calcite as the diffraction grating for the diffraction of X-ray.

Figure A.1: Bragg scattering of X-ray photons from multiple atomic planes.

The diffraction of X-rays from multiple atomic planes can be easily understood using
the method proposed by W. L. Bragg in 1912. Consider a beam of X-ray directed at an
angle θ with respect to a surface atomic plane as shown in Fig. A.1. Note that the waves

177
Appendix A X-ray Diffraction 178

scattering from any atoms in the same plane, e. g. rays 1 and 4, always travel the same
distance to the detector since the incident beam and detector are at the same angle. Thus
we only view the interference between the waves scattering from successive planes. If the
path length difference of two rays scattering from successive planes, e. g. rays 1 and 2,
equals integral multiple of X-ray wavelength, they will interfere constructively. From the
diagram, we can see that constructive interference will occur when

nλ = 2d sin θ (A.1)

where λ is the X-ray wavelength, n is the order of the intensity maximum and d is the
spacing between the atomic planes. For fixed d and λ, there are several maxima at different
angles corresponding to n = 1, 2, 3, . . .. Equation (A.1) is known as the Bragg equation
which was used with great success by the Braggs to determine the atomic positions in
crystals. A diagram of a Bragg X-ray spectrometer is shown in Fig. A.2. The crystal
is slowly rotated until a strong reflection is observed, which means that Equation (A.1)
holds. If λ is known, d can be calculated, and the crystal structure may be determined
from the series of values found.

Figure A.2: A Bragg crystal X-ray spectrometer. The crystal is rotated about an axis
through P .
Appendix B

Differential Equations

The Schrödinger equation is a differential equation — equation containing derivatives or


differentials of one or more variables. Differential equations can be classified into two
types:

• Ordinary differential equation (ODE)


It is a differential equation consist of derivatives of one or more dependent variables
with respect to a single independent variable.

• Partial differential equation (PDE)


It is a differential equation containing at least one partial derivative of some depen-
dent variables.

Obviously, the time-independent Schrödinger equation is an ordinary differential equa-


tion while the time-dependent Schrödinger equation is a partial differential equation.

Example B-1

Which of the following differential equations are ODE? Which are PDE?
dy
(a) = ex + sin x
dx
(b) y 00 − 2y 0 + y = cos x
∂2u ∂2u ∂u
(c) 2
+ 2 =
∂x ∂y ∂t

179
Appendix B Differential Equations 180

Partial Derivatives

Suppose we have a function f (x, y) of two variables, x and y,


and we want to know how f varies with only one of them, say x.
To find out, we differentiate f with respect to x while treating
the other variable y as a constant. The result is the partial
derivative of f with respect to x, which is written as:
 
∂f df
=
∂x dx y=constant

For example, if f = x2 y,
∂f
= 2xy
∂x
∂2f ∂f
To find ∂x2
, we first calculate ∂x
and then differentiate with
respect to x again, still keeping y constant. For example, if
f = x2 y, then
∂2f ∂
2
= (2xy) = 2y
∂x ∂x
Similarly,
∂2f ∂ 2
= (x ) = 0
∂y 2 ∂y

The order of differential equations is defined as the order of the highest-order deriva-
tive appeared in the equation. For example, the equation

dy 3 dy 2 dy
3
+ 4x 2
+ 5 − x2 y = cos 3x
dx dx dx
is of third order.

A differential equation is linear if and only if each term involving the dependent
variables and their derivatives is of the first degree in those variables and their derivatives.
An nth-order linear ordinary differential equation has the form

an (x)y (n) + an−1 (x)y (n−1) + . . . + a1 (x)y 0 + a0 (x)y = b(x)

If b(x) = 0, the equation is said to be homogeneous. Otherwise, the equation is said to


be inhomogeneous. If a differential equation cannot be written in this form, it is called
a non-linear differential equation. For example,
Appendix B Differential Equations 181

Differential Equation Linearity and order


dy
(a) (x2 − 1) + 6y = (x − 2)2 1st-order linear ODE
dx
2
dy dy
(b) 2
− 2y + xy = 2 2nd-order non-linear ODE
dx dx
∂2u ∂2v ∂u
(c) 2
+ 4 2
−7 + 2v = eu 2nd-order non-linear PDE
∂x ∂t ∂x

The one-dimensional time-independent Schrödinger equation is a homogeneous second


order linear ordinary differential equation. In general, solving differential equations like
this one is complicate. For a few simple equations, there are some techniques which can
solve them directly. But in most cases, differential equations are solved by inspection and
trial-and-error.

An nth-order homogeneous linear differential equation always has n linearly inde-


pendent solutions. If y1 (x) and y2 (x) are the two linearly independent solutions of an
2nd-order homogeneous linear DE, the general solution of this equation is:

y(x) = c1 y1 (x) + c2 y2 (x)

where c1 and c2 are some arbitrary constants. Any value for these constant will satisfy the
differential equation. However, every problem must have additional information or con-
straints in order to specify the value of these two constants. If the additional constraints
are specified at the initial point, generally time point, they are called initial conditions.
But if the constraints are specified at the boundary points, generally space points, they
are called boundary conditions. The general solution, along with two additional initial
or boundary conditions (for the equation has two arbitrary constants since it is a second
order equation) constitutes the solution of the Schrödinger equation for a given system.

Linearly Independent

A set of functions y1 (x), y2 (x), . . . , yn (x) is said to be linearly


dependent if there exist constants c1 , c2 , . . . , cn , not all zero, such
that
c1 y1 (x) + c2 y2 (x) + . . . + cn yn (x) = 0

If this is not the case, the functions are said to be linearly inde-
pendent.
Appendix C

Separation of Variables of the


Time-independent Schrödinger
Equation for the Hydrogen Atom

Assuming the nucleus is stationary, the time-independent Schrödinger equation for a


hydrogen atom in spherical coordinates is
     
~2 1 ∂ 2 ∂ ∂ ∂ 2 ∂2
− r + csc θ sin θ + csc θ 2 ψ(r, θ, ψ)
2m r 2 ∂r ∂r ∂θ ∂θ ∂φ
+ U (r)ψ(r, θ, ψ) = Eψ(r, θ, ψ) (C.1)

where U = −(1/4π0 )(e2 /r).

We consider a separable solution of the above equation with the form:

ψ(r, θ, φ) = R(r)Θ(θ)Φ(φ)

Putting this solution back into the equation, we have


     
~2 1 d 2 dR d dΘ 2
2 d Φ
− ΘΦ r + RΦ csc θ sin θ + RΘ csc θ 2
2m r 2 dr dr dθ dθ dφ
+U (r)RΘΦ = ERΘΦ
   
1 d 2 dR 2mr 2 1 d dΘ 2
2 1 d Φ
⇒ − r − [E − U (r)] = csc θ sin θ + csc θ
R dr dr ~2 Θ dθ dθ Φ dφ2
| {z } | {z }
depends only on r depends only on θ and φ

Because this equation holds for any (r, θ, φ), both sides must equal a constant.

182
Appendix C Separation of Variables of the Time-independent Schrödinger Equation . . . 183

Suppose the right side of (C.1) is equal to C, we have


 
1 d 2 dR 2mr 2
− r − [E − U (r)] = C
R dr dr ~2
   
1 d 2 dR(r) 2m ~2 C
⇒ 2 r + 2 E − U (r) + R(r) = 0
r dr dr ~ 2mr 2

Setting the left side of (C.1) to (C) gives:


  2
1 d dΘ 2 1 d Φ
csc θ sin θ + csc θ =C
Θ dθ dθ Φ dφ2
 
1 d dΘ 1 d2 Φ
⇒ sin θ sin θ − C sin2 θ = −
Θ dθ dθ Φ dφ2
| {z } | {z }
depends only on θ depends only on φ

As the above equation holds for any (θ, φ), both sides must equal a constant, too, and we
choose the symbol D. Setting each side to D, we obtain

d2 Φ(φ)
= −DΦ(φ)
dφ2
and
 
1 d dΘ
sin θ sin θ − C sin2 θ = D
Θ dθ dθ
 
d dΘ(θ)
⇒ sin θ sin θ − (C sin2 θ + D)Θ(θ) = 0
dθ dθ
Appendix D

Magnetic Dipole

A charged particle moving in a small closed orbit constitutes a magnetic dipole. A


simple example of magnetic dipole is a current loop. A bar magnet can also be considered
to be a magnetic dipole since it produces the same pattern of magnetic field lines as the
current loop (see Fig. D.1).

Figure D.1: Both current loop and bar magnet are examples of magnetic dipole.

We can characterize a magnetic dipole by a vector quantity called the magnetic


dipole moment µ
~ . For a current loop with N turns and area A, the magnetic dipole
moment is equal to
µ
~ = N iA~n

where i is the current of the loop and ~n is a unit vector perpendicular to the loop in a
direction determined by the right-hand rule. The magnetic field created by a current loop

184
Appendix D Magnetic Dipole 185

can be expressed in terms of its magnetic dipole moment:

~ = µ0 µ
B
~
2πz 3

~ equal but opposite


When a magnetic dipole is placed in a uniform magnetic field B,
forces exert on each side of the dipole. As a results, the dipole experiences a torque ~τ
given by:
~
~ ×B
~τ = µ

The resulting torque will tend to align the dipole with the applied field. Suppose a dipole
µ
~ rotates an angle dθ, the work done on the dipole is

dW = −dU = −τ dθ

where dU is the change in the potential energy of the dipole. Therefore, when the angle
between µ
~ and B~ is θ, the potential energy of the dipole is
Z
U = τ dθ = −~µ · B ~

~
~ ⊥ B.
by setting the potential energy U = 0 for µ

Вам также может понравиться