Quantum Mechanics PDF

TIGHT BINDING BOOK
OU_160147>m g< ^
CO
CONTENTS
Removal of degeneracy. The Stark effect. Splitting of spectral The Lorentz classical theory of dispersion. Index of
terms.
of refraction. Polarization. Radiation damping. Quantum theory u of dispersion. Oscillator strengths. Ramon effect. Stokes* and "anti-Stokes" lines. P^nergy correction in second order perturbation
theory.
A.i harmonic
.
oscillator.
Harmonic oscillator.
3 Matrix elements of x
(Perturbation theory for the continuous
spectrum.
Ingoing
and
outgoing
Phase shift. Connection between men sional formulations )

.
wave boundary conditions. partial wave and three di-
v^
PART
II.
RELVriVISTIC QUANTUM MKCII \NICS

The Klein-Gordon Scalar RelativiMic Wa\e Equation
Relativistic invanancc of the de Broglie relations. Relativistic energy-momentum relation of a free particle. The Klein-Gordon
257
15.
259
equation. Charge and current density. Nonrclativistic limit. The initial data problem. Indefiniteness of the sign of charge. Inter-
with an external electromagnetic field. Relativistic energy levels of a spinless particle in a Coulomb field. Fine structure constant. The case Za > /2. Charge and current density
action
l
in the presence of an electromagnetic field.
16.
Motion of an Electron
in a
Magnetic Eield. Electron Spin

effect. Interaction energy of a
268
Classical theory of the
Zeeman
dipole. Larmor precession. Magnetic moment of a moving electron. Zeeman effect in nonrelativistic Schrodinger theory. Orbital magnetic moment. Bohr magneton. Normal and anomalous Zeeman splitting. Emstein-de Haas experiment. Land g factor. Stern-Gerlach experiment. Uhlenbeck-Goudsmit hypothesis of intrinsic angular momentum. Half-integral quantum numbers for angular momentum. El ecJjQn^spin The Pauli equation.
magnetic
Two-component wave functions. The operator for intrinsic magnetic moment. Pauli matricies. Coupled Schro'dinger equations. Matrix elements. Spin operators. Commutation relations for spin
operators. Vedtorial character of spin operators. Separation of' spin and space variables in a homogeneous magnetic field.
Eigenvalues of the spin operator along an arbitrary direction.

Probability distribution of spin directions.
17.
The Uirac \\ave Equation

Linearization of the energy operator. Dirac matricies and their relation to Pauli matricies. The Diracêquation. Charge and
current density. External electromagnetic field. Velocity operator. Statistics in
285
second quantization. Transformation proper-
ties of the spinor
wave function under Lorentz transformations
and spatial rotations.
CONTENTS
18.
The Dirac Theory of the Motion of an Electron

'
in
a Central
'
Field of Force
spin and total angular momenta. Conservation laws. Properties of the total angular momentum operators. Quantization of total angular momentum. Clebsch-Gordan coefficients. Spherical spinors. The vector model of the addition of angular momenta. Motion in a central field including spin effects. Theory of the rotator. Selection rules. Parity of a state. Conservation of parity. Solution of the Dirac equation for a free particle. Negative energy states. Nonrelativistic limit. Four-vector transOrbital,
293
formation law of the energy-momentum operators under Lorentz transformations. Relativistic invariance of the scalar wsrve
equation. Vector model. Charge conjugation.
19.
Approximate Form Two component Pauli form. "Small" and "large" components. 2 Relativistic increase of mass. Correction terms to order (i>/c) Interaction of the intrinsic magnetic moment. Spin-orbit interaction. Contact interaction. The velocity operator and Ehrenfest's theorem in the Dirac theory.
The Dirac Equation
in
308
20.
The Fine
Structure of the Spectra of Hydrogen-like
Atoms
314
Advantages of the approximate method. Relativistic and spin effects. Contact interaction. Stable motion for Z < 137. Fine structure in the Dirac theory. Experimental verification of the
fine structure theory. Lamb-Rutherford experiment. Anomalous Zeeman effect. Weak magnetic field. Lande' g factor. Strong
magnetic fields. Paschen-Back effect. "Breaking" of spin-orbit coupling. Paramagnetism and diamagnetism. Anomalous Zeeman effect in the vector model. (Stark effect. Quenching of metastable states. Intermediate field Paschen-Back effect.)
21.
The Effect of Nuclear Structure on Atomic Spectra Reduced mass. Effect of finite nuclear size. Mesic atoms. Approximate harmonic oscillator potential for large Z. Spin of the muon. Application of the Dirac equation to the neutron and proton. Anomalous (Pauli) magnetic moment. Experimental determination of the magnetic moments of the neutron and proton. Limitations on the measurement of angular momentum. Experiments of Bloch and Alvarez and of Rabi. Nuclear magneton. Hyperfine structure of the hydrogen spectrum.
334
22. 'The Electron -Positron
Vacuum and
the Electromagnetic
Vacuum
347
A. Dirac theory of "holes." Negative energy states. Discovery of the positron. Pair creation and annihilation. Antiparticles. Rigorous validity of conservation laws. Positronium. Interconvertibility of particles. B. The Lamb shift of energy levels of atomic electrons. Fluctuations of the electromagnetic vacuum.
CONTENTS
Virtual particles. "Smearing out" of a point electron. C. Elec-
Xl
tron-Positron
netic
vacuum. Vacuum polarization. Anomalous magmoments of electron, proton, and neutron. D. Renormalization. Quantum electrodynamics. Quantum theory of fields. Cherenkov radiation.
Theory of the Helium Atom Neglecting Spin States
358
23.
Basic principles of the theory of multielectron atoms. Indistinguishability of electrons. J Exchange forces. Perturbation theory solution of the helium atom. Permutation of electrons.
and antiJSxchange degeneracy^ Exchange energy. Symmetric symmetric wave functions. XUoulomb interaction between electrons, "lonization energy. 'The variational method. Derivation of the SchrOdinger equation by the variational method. Hartree-
F]pck method__of self-con si ^tent fields. Investigation of the "exchange energy. Exchange time.
24.
Elementary Theory of Multielectron Atoms Including Spin States

Symmetric and antisymmetric states. Permutation operator. Fermi-Dirac and Bo se-Ein stein statistics. Tt]^ Pauli cntplnsinn principle. Fermi on s. Bosons. Determinental wave function. Addition of angular momentum. Russell-Saunders coupling. Clebsch-Gordan coefficients. LS coupling, jj coupling. Wave function of the helium atom including spins. Triplet and singlet states. Parahelium and orthohelium. Energy spectrum of the helium atom. Variational wave function for a Yukawa potential.
378
Diamagnetic susceptibility of parahelium.

25. Optical Spectra of Alkali
Metals
397
The
The Thomas-Fermi statistical Boundary conditions for neutral and ionized atoms. Solution of the Thomas-Fermi problem by the Ritz variational method. Total ionization energies. Charge distribution in argon. Energy levels of alkali atoms. Atomic core. "Penetrating"
structure of complex atoms.
method.
orbits.
Polarization of the atomic core. "Effective principle quantum number." Smearing of the atomic core. Fundamental series. Multiplet structure of spectral lines. Spectral terms of sodium. Sharp, principle and diffuse series.
26.
Mendeleyev's Periodic System of Elements

X-ray spectra of atoms. Continuous spectra. Bremsstrahlung. Characteristic spectra of atoms and the structure of their inner shells. Moseley's law. Multiplet structure of x-ray spectra.
Relativistic and spin effects. Regular and irregular doublets. Filling of the electron shells. Application of the Thomas-Fermi method. Peri-
420
The discovery of Mendeleyev's periodic law.

odicity properties of the elements.
Xll
CONTENTS
27.
The Theory of Simple Molecules

Chemical bond. Heteropolar molecules. Affinity. Valence. KosMo 1 ecular h.vdiQgeg_ign J^change foj^s^-Jâluajjoa _of se
1 .
.
137
Spin and symmetry. Orthohydrogen and parahydrogen. The valence theory. Spin valence. Mascrs and lasers.
.
PART
III.
SOME APPLICATIONS TO NL1CLFAR PHYSICS

465
rule.
28. Elastic Scattering of Particles
'Time-dependent perturbation theory. Golden

tude.
Cross section
for elastic scattering. Uncertainty of energy. Scattering ampli-
force.
tral
Born approximation. Scattering by a Yukawa center of Range of nuclear force. Fast-electron scattering by neuatoms. Validity of Horn approximation. Partial-wave cross
sections.
Phase
shift.
Scattering from a spherical barrier and
spherical well.
of final states.)
29.
Resonant scattering. (Golden Rule #2, Density
Second Quant t/.ut ion
480
Second quantization of the Schrodingcr equation. The Ileisenberg equation of motion, q numbers and c numbers. Commutation relations for Boson field amplitudes. Creation and destruction
operators.
An ti commutation relations describing particles obeystatistics. Fermi Quantization of Maxwell's field equations. ing Spontaneous emission. Dipole approximation. Beta decay. Pauli's hypothesis of the neutrino. The Fermi theory. Weak
and
strong
interactions.
Fermi and Gamow-Teller selection
Feynman and Gell-Mann theory. /3-decay spectrum. Nonconservation of parity in weak interactions. Lee and Yang. Helicity of the neutrino* Pion decay.
rules.
APPENDIX V
Hilbert Space and Transformation
Theory
497
APPENDIX
I).
The
Statistical Assertions of
Quantum Mechanics
505
511
PROBLEMS
Preface
lectures to students at the Mos(1945 to 1948) and Moscow In writing this book we set ourselves the difficult task of treating in a single volume the fundamentals of atomic theory, that is, Schrodinger's nonrelativistic theory, Dirac's relativistic theory, the theory of multielectron atonis, and the basic applications of quantum mechanics to solid state
This textbook is based on
my
cow Regional Pedagogical University from 1945 on.
Institute
physics. Our aim was to combine the exposition of general theoretical principles with examples of the application of quantum mechanics to specific problems connected with atomic structure. To avoid overloading this book, we have abridged the treatment of certain specialized topics, but in such cases we have endeavored to supply references to standard works on the subject. In most textbooks the solution of specific problems with the
help of Schrodinger's equation is handled in fairly elegant form. The basic mathematical tools required for this purpose are a knowledge of second-order differential equations and various special functions (including the Hermite, Legendre and Laguerre polynomials). However, applications of Dirac's theory to specific problems (such as the hydrogen atom) are on the whole handled less satisfactorily. In some cases the calculations are so long
and cumbersome that

ing of the solutions.
it
is difficult to
perceive the physical mean-
In others there is no actual derivation of the results or only a rough proof is given. In an attempt to avoid these pitfalls, we have used an approximate form of Dirac's equation for our treatment of the hydrogen atom (Chapter 19). This approximation still enables us to obtain the formula for the fine structure of the energy levels and the selection rules (Chapter 18
and 20). Our analysis of the Lamb shift due to the electronpositron vacuum is also somewhat simplified (Chapter 22). Several good problem books in quantum mechanics are available, and therefore we shall consider only a few problems chosen with the aim of elucidating and supplementing the general discussion.
The first part of this book was written jointly by me and Yu. M. Loskutov, and the second part jointly by me and I. M. Ternov. Great assistance was rendered by M. M. Kolesnikova in condensing notes based on my lectures on quantum mechanics and in
preparing the manuscript for the press. Chapter 25 was carefully
XlV
PREFACE
read by N. N. Kolesnikov, who made a number of valuable comments. I would like to mention the great pains taken by S. I. Larin in editing the whole manuscript.
A. A. Sokolov
Introduction
Quantum mechanics dates only from the 1920's. This important branch of theoretical physics deals with the fundamental problem of the behavior of micropar tides (for instance, the behavior of electrons in an atom). As a theo-.y, quantum mechanics represents an extension of classical mechanics, electrodynamics (including the theory of the electron and the theory of relativity) the kinetic theory of matter, and other branches of theoretical physics. Historically, the development of every branch of theoretical physics involves two main stages. First comes the accumulation of experimental facts, the discovery of semiempirical laws, and the development of preliminary hypotheses and theories. This is followed by the discovery of general laws, which provide a basis for interpreting a large number of phenomena. For example, the first or pre-Newtonian stage of mechanics consisted of the discovery of a number of seemingly unrelated laws: the law of inertia, the law of free fall under the action of a gravitational field and Kepler's laws of planetary motion. Most of these laws were discovered only after years of painstaking work by many scientists. Thus, many astronomical observations preceded the discovery of Kepler's
,
We may recall the great efforts of Copernicus, Bruno, Galand others to establish that the Sun is the center of our planetary system and that the Earth is only a planet like Mars, Venus, or Jupiter. It was only after working for fifteen years on Tycho Brahe's extremely valuable observational data that Kepler found the semiempirical laws describing planetary motion. After these preliminary, seemingly independent laws had been established, Newton was able to show that they all rested on the same theoretical foundation. Newton's three laws of motion and the law of universal gravitation opened a new stage in the development of theoretical mechanics. One of the great triumphs of Newtonian mechanics was Leverrier's prediction of the existence of a new planet, Neptune, from perturbations in the motion of Jupiter. In a similar fashion, Maxwell's formulation of the laws of electrodynamics was preceded by the discovery of empirical laws describing various electric and magnetic phenomena. Coulomb's law of interaction between electric charges and magnetic poles 1 and the Biot-Savart law of interaction between an electric current and a magnetic pole were found by analogy with Newton's law of
laws.
ileo,
1
verified
As magnetic monopoles do not exist by means of magnetic dipoles.
in nature,
Coulomb's law
in
magnetostatics is
XVl
INTRODUCTION
gravitation. All of these phenomena were explained on the basis of the principle of "action at a distance," according to which one charge acts directly on another through the intervening space. After Newton, and independently of investigations of electric and magnetic phenomena, considerable attention was devoted to optics. At a relatively early stage, it was established that light consists of transverse waves propagating with a finite velocity
c^3-10 cm/ sec. mained unknown.

of
l
The nature
of these
waves, however, re-
All of these preliminary studies belonged to the first stage of development of electrodynamics: they prepared the ground for Maxwell's theory, which had approximately the same unifying role in electrodynamics as Newton's laws in mechanics. Maxwell's equations provided a powerful tool for the investigation of electric, magnetic and optical phenomena. Maxwell's theory predicted the existence of electromagnetic fields, which carry the interaction continuously from point to point, and of electromagnetic waves, which were later discovered by Hertz. The theory of propagation of electromagnetic waves underlies all of modern radio engineering. Another important result of Maxwell's theory was aproof of the
wave nature of light. The view that matter and electricity have an atomic structure was of considerable importance in connection with the appearance
quantum mechanics. This view had very ancient roots, but remained without scientific foundation until the discovery of the fundamental law of chemistry the law of exact proportions. The kinetic theory of matter and, in particular, the kinetic theory of gasesbased on the classical Maxwell- Boltzmann statistics were important steps in the development of atomic theory. It is worth noting that the classical Maxwell- Boltzmann statistics, which rests on probability theory, cannot be completely explained in terms of Newtonian mechanics and contains certain features that are characteristic only of large collections of particles (for example, the irreversibility of certain processes). Statistical methods made it possible to explain a number of macroscopic properties of matter, such as temperature and specific heat; this provided indirect
of
evidence of the atomic structure of matter. One of the decisive proofs of the atomic theory of matter was the discovery of fluctuations, that is, statistical fluctuations in the behavior of individual molecules. Brownian motion was particularly important in this connection, as it provided the evidence of molecular movement in a liquid. Even more suggestive proofs of the atomic structure of matter were provided by Laue's observation of the diffraction of x-rays in crystals and Aston's mass-spectrographic measurements of the atomic weights of individual isotopes of various elements. From an analysis of Faraday's laws of electrolysis, Helmholtz showed that there must be a fundamental quantity of electricity,
INTRODUCTION
XVII
10 4.8- 10 esu, such that any charge, positive or negative, an integral multiple of this charge. Studies of anode rays indicated that positive charges always appear as ions; that is, a positive charge is always associated with the basic mass of an atom. The lightest positive ion is that of a hydrogen atom. It is known as a proton and it's mass is nearly the same as the mass of a neutral hydrogen atom. The carrier of a negative charge can take the form of a negative ion or of a much lighter particle known as an electron. From measurements of the deflection of cathode rays (a beam of electrons) in electric and magnetic fields, it was found that the mass of the electron was about 1/1836 of the mass of the proton. 2 These discoveries led to Lorentz's electron theory, which represents an interesting synthesis of Maxwell's electrodynamics for a vacuum and the atomic view that matter consists of positive and negative charges. In Lorentz's theory the magnetic permeability, dielectric constant, and conductivity of a medium were obtained by averaging Maxwell's equations for a vacuum over charges and currents of particles of the medium. A conductor was treated as a medium filled with free electrons or, in other words, an "electron gas." It followed from Lorentz's theory that the dielectric constant depends on the frequency of electromagnetic waves, whereas in Maxwell's theory it had been assumed that this quantity is a constant. Lorentz's theory provided an explanation of the dispersion of light. The appearance of this theory was accompanied by the extension of electrodynamics to frames of reference traveling with constant relative velocities; this culminated in the special theory of relativity. It is well known that all laws of motion whether they be Newton' slaws or Maxwell's equations for the motion of an electromagnetic field must be associated with a frame of reference. Newton believed that his laws were related to an absolute frame of reference. Even in his writings, however, this notion remained purely metaphysical, and Newton himself discovered the principle of relativity in mechanics according to which it is impossible to detect a uniform rectilinear motion of a body (or a frame of reference) relative to this absolute system, because all frames of reference moving linearly with constant relative velocities with respect to each other are completely equivalent. Consider the Galilean- Newtonian transformation from one inertial system to another, moving along the x axis with relative
equal to
is
velocity v
In 1932, a particle with a positive charge and mass equal to that of an electron, known as the positron, was discovered Positrons are formed in small quantities when cosmic rays pass through matter. Under ordinary conditions a positron cannot exist for any significant length of time because it combines with an electron and the two particles are converted into gamma-ray photons (see Chapters 3 and 22).
XVili
INTRODUCTION
where the primed coordinates refer to the moving system, and the unprimed coordinates to the stationary system. We find that accelerations and forces are identical in the two frames of reference, and therefore the equations of mechanics (in which the
velocity does not appear) are invariant under this transformation. If the Galilean- Newtonian transformation is applied to the MaxwellLorentz equations, they assume different forms in different inertial systems, because the equations contain the velocity of propagation of electromagnetic waves which, added vector ially, has different
values in different inertial systems. The original Michelson-Morley and other numerous experiments showed, however, that the speed of light is the same in any direction in all inertial coordinate systems. As a result, Einstein generalized the Newtonian principle of relativity in a way that led directly to the so-called Lorentz transformations
*=
*=!!
<
B.
P=4.
The classical laws of electrodynamics are invariant under this transformation. Since the equations of Newtonian mechanics, however, are not invariant under the Lorentz transformations,
they
had
to
mass m of a moving its rest mass m by the
be replaced by relativistic equations in which the particle was related to its velocity v and
relationship
At low velocities, where P2 ^0, the relativistic equations reduce to the Newtonian formulations. The Maxwell- Lorentz equations for an electromagnetic field and the relativistic equations of motion of an electron constituted the culminating stage of the classical electron theory. According to this theory, light consists of electromagnetic waves and an electron is a particle whose motion is described by relativistic mechanics. The success of the Maxwell- Lorentz theory in accounting for certain microscopic phenomena (the propagation and dispersion of light, the motion of an electron in electric and magnetic fields, and so forth) was accompanied by the discovery of experimental facts that could not be explained with classical concepts. These experiments will be described in Chapters 13, and therefore we shall mention them here only very briefly. In the first place, it was found that black-body radiation, the photoelectric effect, and the Compton effect could be explained only
INTRODUCTION
xlx
on the basis of corpuscular properties of light. This was the implicit assumption of the Planck- Einstein photon theory, in which the discrete structure of light was described in terms of Planck's constant h = 6.62* 10"" 27 erg- sec. The photon theory was also successfully used by Bohr in constructing the first quantum theory of the atom, based on the planetary model suggested by Rutherford. In the second place, a number of experimental facts, including the electron diffraction, indicated that in addition to its corpuscular properties, an electron has wave properties. De Broglie's definition of the wavelength of an electron also included Planck's constant h. This led eventually to the development of a new scienceelectron optics which provides a theoretical basis for electron microscopy. The SchrBdinger wave equation (1926) was the first general
theoretical treatment that explained both of these classes of pheunified the preliminary theories of Planck, Einstein, Bohr, and de Broglie. This equation made it possible to discover the laws of behavior of electrons and other elementary particles and to construct a relatively systematic theory of radiation that took into account the quantum nature of light. For atomic physicists, the Schr'ddinger equation was one of the most powerful tools. Many phenomena associated with the behavior of an electron in an atom and with the absorption and emission of light by an atom were provided with a theoretical explanation (see Chapters 4-14). The later development of quantum theory showed that the Schrbdinger equation did not describe all the properties of atoms. In particular, it could not explain correctly the interaction of an atom with a magnetic field (for instance, the anomalous Zeeman effect) and it could not be used to construct a theory of multielectron atoms. One of the main reasons for this was that the Schrbdinger theory did not take into account the electron spin.
nomena and
Dirac's relativistic theory (see Chapters 15-17) was an extension of the Schrbdinger theory that considered relativistic and spin effects of moving electrons (see Chapters 18-20). It turned out that the quantitative corrections due to relativistic effects were relatively small, but that spin effects were of fundamental importance in connection with the fine structure of multielectron atoms (see Chapters 23 and 24). These effects explained the filling of electron shells in an atom and gave a theoretical basis to Mendeleyev's periodic table of elements (see Chapters 25 and 26). Although the fundamental problems related to the structure of the atom were basically solved with the appearance of Dirac's equation, we are constantly adding further details to our knowledge. At present a great deal of attention is being devoted to the influerê of the electron- positron vacuum and magnetic moments on the energy levels of atoms (see Chapters 21 and 22). Quantum mechanics has also been applied to simple molecules (Chapter 27), solid state physics (Chapter 6), and the atomic nucleus.
Part
Nonrelativistic
Quantum Mechanics
Chapter
The Quantum Theory of Light

supremacy of classical physics was challenged by quantum mechanics in the beginning of this century, particle motion was sharply distinguished from wave motion. According to the classical picture, the world consisted of particles (for example, electrons and ions) and fields (for example, light). This picture was completed by Maxwell's theory (1873), which appeared to have definitely established that fields had wave-like
Before the the advent of
properties.
Towards the end of the nineteenth century and in the first years of this century, this state of affairs was disturbed by the discovery of experimental facts that did not fit into the classical conceptual framework. On the one hand, there were certain phenomena, such as the radiation spectrum of an ideal black body, the photoelectric effect, and the Compton effect, which could be understood only in terms of particle-like properties of light. On the other hand, electrons were observed to have wave-like properties, such as diffraction, which later served as a basis for the development of electron
optics.
A.
PRINCIPLES OF THE ELECTRON THEORY
The behavior of the electromagnetic field produced by a given distribution of charge and current is described by the well-known Maxwell-Lorentz equations
dt
= 4wp,
field intensities, re-
where E and //are the electric and magnetic

spectively,
p
is the
charge density (for example, of the electron),
and v
is its velocity.
NONRELATIVISTIC QUANTUM MECHANICS
To start with, an electromagnetic field transmits interactions between the charges. The interaction between stationary charges e is transmitted by an electrostatic field which satisfies Coulomb's
law (v =?*-],
whereas the interaction between moving charges
is transmitted by a system of electric and magnetic fields, since a moving charge can be regarded as an electric current and it is well known that a current interacts with a magnetic field. Electromagnetic fields are always associated with sources of the
appropriate type (for example, charges). Secondly, electromagnetic fields may be regarded as electromagnetic waves, which propagate with the velocity of light c (radio or light waves). As a particular example of a source of light waves, we can take an accelerated charge. The latter emits radiant energy; per unit time, this energy is
W w
2 e* w*
"3~~7~
(I 21 (**/
where e is the charge, and w its acceleration. Once electromagnetic waves have been produced, they can exist independently of their
sources.
The equation describing the propagation of a light wave is obtained from Eqs. (1.1) by setting the charge density p equal to zero. We can then eliminate the vector H from the second Maxwell equation by taking the curl of this equation and substituting into it the first equation. Since V --=(), we can use the vector relation
VxVx=: V(V-)
to obtain the following nents of both vectors E
V2
=
compo-
wave
and
//:
equation, which holds for the
where /
is
A more
any component of the vectors

detailed analysis
of
or//. the Maxwell- Lorentz equations
shows that electromagnetic waves are transverse. This means that the electric field intensity (E) and magnetic field intensity (//) are mutually orthogonal, and also orthogonal to the wave vector k, which points in the direction of propagation of the electromagnetic wave. The vectors form a triad such that when a right-handed screw is turned from E to //, it moves along the direction of k
:
//=*x,
where
(1.4)
k/k
is a unit vector.
THE QUANTUM THEORY OF LIGHT
A charge (say, an electron) moving in externally applied electric and magnetic fields experiences a force
l*x#),
(1.5)
which is called the Lorentz force. Taking into account the relativistic variation of mass, the equation of motion of an electron in an external field has the form
where
We can select a Lagrangian function the variational principle
X
0,
in such a
manner
that
(1.7)
or, in
more
explicit
form,
Jffsf-CjH
('='.*>
3),
d.8)
will yield the equation of motion of an electron (1.6). Here x l 9 z are the spatial coordinates, and the jf denote the x* s=y* x* corresponding velocities. To obtain (1.6), we must set
=x
*->t,
(1.9)
where 4 and o are the vector and scalar potentials of the electromagnetic field. These potentials are related by the Lorentz condition
v.^+j|5=o.
(i.io)
The electric and magnetic field intensities can be expressed terms of A and O by means of the relations
dt
,
in
//=vxA
We
find the following expression for the electron
(1.11)
momentum:

for the generalized force acting on the electron,
Similarly, obtain
we
a-- -+*(
Substituting (1.12) and (1.13) into (1.8) and taking into account (1.10) and (1.11), we obtain Eq. (1.6) for the motion of an electron. Thus, our choice of the Lagrangian is justified. Since we know X, we may also determine the Hamiltonian //:
= PA-*
It is well known that the Hamiltonian should not be expressed in terms of the velocity c$ but in terms of the generalized momentum
t
p=p
equation
A, which,
according to
(1.12),
is
related to
cp
by the
Therefore, the relativistic form of the Hamiltonian is

//== /c9 />a
+ mfc + *$.
4
(1.14)
We
note that
if
the potentials are time- independent , the Hamiltonian
is equal to the total
energy ( E
=H
).
In the nonrelativistic approximation written in the form
(P<m c),
Eq. (1.14) can be
where
H' is approximately equal to the nonrelativistic part of the
Hamiltonian
<W5)
this it can be seen that the relativistic equation for the Hamiltonian (1.14) also includes the rest-mass energy m c9 of the electron. It is very important to take this rest-mass energy into account in studying transformations of elementary particles.
From

B.
THE CLASSICAL THEORY OF BLACK-BODY RADIATION
Among all the phenomena associated with electromagnetic fields, special importance can be attached to the properties of cavity radiation. This can be described as thejradiation inside a_cavi.ty.PQjaawMla^which areieated to a certain
alternatively, as radtetipnl]^ ^ A small hole madejn the wall
In otherwoiggj essentiallyjUl rays entejj^ Hole-Iarela6sorbed. or. moi^_^^cisely^ the reemerga^from the_hole is negligibly small. Consequently*. the cavity radiation escaplnglfrom the hole_may..Jaft regarded ftff the radiation which would be obtainedJzomjui ideal black^body, jtnd it is generally referred to^sj5Iack-body radiation. The analysis of black-body radiation played a particularly important role in the foundation of quantum theory. Although a more or less reasonable classical explanation could be found for many other phenomena, every single theory of black-body radiation constructed on the basis of classical concepts failed to agree with the experimental facts. A systematic theory of black-body radiation
was developed only

introduced
the
in the beginning of this century, when Planck concept of a quantum of energy. This concept was later to play an important role in the development of the first quantum theory of the atom, and afterwards in the development of
quantum mechanics. We shall now consider the theory
of black-body radiation, confining ourselves for the time being to classical concepts. Let the radiation be characterized by its spectral density Po,, 1 which is related to the ordinary electromagnetic energy density
<
1 - 16 >
by the relation
P-
= ZT.
used
(1-17)
In
function
the literature, the function p v is sometimes p^ is related to p v by the equation

1
for the spectral density.
The
277
since
o>
= 2/m

in the
where du is the energy density of the radiation range from o> to o> + d(o. Obviously,
frequency
(1.18)
the basis of the second law of thermodynamics, Kirchhoff that the density p w is determined only by the temperature of the walls of the closed cavity and is entirely independent of the material of which the walls are made; that is, p =/(o), T\
On
showed
(1)
Consequently, the walls of the cavity may be considered as a set of oscillators. The average energy of these oscillators is completely determined by the spectral density of the black-body radiation. We shall show this starling with the equation of motion for an oscillator and taking into account Planck's radiation damping
=
Here
IJ
<
U9
>
e and m are the charge and rest mass of the oscillator, o> is its natural frequency of oscillations, and Ex is the x component of the electric field intensity of the blackbody radiation. Representing Ex in the form of a Fourier series
=
where
2
<*
(1.20)
Exn
is the
amplitude of an individual oscillation of the field with frequency
no>o,
(Ul)
(1.19):
we
obtain the following equation for x(t)
from Eq.
99X
The average energy of the oscillator, which, according to the virial theorem, is twice the average kinetic energy, is given by
(1.23)
where the bar denotes averaging over time. Since
* *
n, n' ***
gtoot (n-n')
oo

where f_n =/f and since, moreover,
,
*>
=!
*
f J
2
where
27i/o)
Eq. (1.23) can be reduced to the form
hence the
amplitude
This equation has a very sharp maximum in the neighborhood of frequency o>,and total energy of the oscillator will actually depend only on those terms of the s u. series for which mo Consequently, in the above equation, the square modulus of the
|
Exn
2
|
can be replaced
by|
ExriQ
|,
where
/*
=
,
and at the same time the
sum
can be changed
to
an integral. According to (1.2 1), 2

d<* n
<o
drt
= =
<o
US)
since dn
1.
Therefore, we obtain
n-
fj.0% "OP
I i
y.
JT/IQ
ing
Replacing the frequency w rt by the variable of integration

oo,
<>
everywhere except = w and extending

rt
<o,
in the difference o> n w, introducthe limits of integration to
we
find
_37t/Z
Ca
"2 ~a>
'
ô
'
'
(U7)
On
the other hand, the energy density u, which is related to the electromagnetic field
(JL 16),
of the radiation by Eq.

tion is isotropic
can also be expressed in terms of
ExriQ
|*.
Since the radia-
we have, on
the basis of Eq. (1.16),

1
71
The reader should not

quantity O) = 277/r, for the relationship O)Q <3C
it
is
a)
uneasy about our equating the differential dco n to the finite always possible to make the period T sufficiently large so that will be satisfied. Mathematically, this corresponds to a transifeel
tion from Fourier series to Fourier integral.
10

1 9
Using the expansion (1.20) and the rule (1.24) for averaging over
we
obtain
Hence, taking into account Eq. (1.18) together with the relationship
for,t
o>
(n
/2
)f
we obtain
Comparing Eqs.
the average energy density of the radiation p^
(1.31) andjl.27), of
we
=;%*
find the relationship between the oscillator and the spectral
d.32)
which forms the basis of the theory of black-body radiation. In classical statistical physics, the energy distribution of particles is given by the function
N(E)
-
= Ae-**.
1
(1.33)
where a= 1/T; k= 1.38 10~ 16 erg.deg' is Boltzmann's constant, and T is the temperature of the medium. The average energy of the
particles is
(1.34)
Substituting this value of
E into
Eq.
(1.32),
we
obtain the Rayleigh-
Jeans formula
9=-4r-kT.
(1.35)

This equation satisfies the Wien's thermodynamic law
(1.36)
which was based on various results in thermodynamics and the electromagnetic theory of light. In the region of long wavelengths (low frequencies), the Ray leigh- Jeans formula is in good agreement with experimental data. At short wavelengths, however, it completely fails to agree with experiment (see Fig. 1.1).
Fig.
1.1.
Radiation spectrum of an ideal black body.
Pa ~po
The heavy dotted line indicates the Ray leigh- Jeans 2 curve pa ~ pQX and the solid line the Planck curve * ~ e x /( 1) which is the same as the experi,
mental curve.
Here p$
0)
(&T) /n h = kT/H.
0)
&)QX,
and
In exactly the same way, the use of the Ray leigh- Jeans formula for calculation of the radiation energy density [see Eq. (1.18)] results in a divergent integral, that is, we obtain the obviously absurd
relationship
1
bT(*
Pd=-JiJT
>
oo.
(1.37)
This was called the "ultraviolet catastrophe" by Ehrenfest. Thus, the classical theory was completely unable to give a satisfactory description of black-body radiation.
C.
PLANCK'S EQUATION
moved
In 1900, Planck put forward an important hypothesis which rethe ultraviolet catastrophe and radically changed a number
of fundamental principles of classical physics.
According to this
12
hypothesis, the energy of microscopic systems (atoms, molecules, and so forth) does not vary continuously and assumes only certain specific discrete values. In particular, the energy of a harmonic oscillator must be a multiple of a certain minimum value e:
=
where
In
AZ
/ie,
(1.38)
O,
1,
....
order to determine the average value of the energy, we must replace the integral (1.34) by the sum
d.39)
Substituting this value of density of the radiation
into Eq. (1.32),
we
obtain the spectral
co
"rf?"
~
e
k1
"
(1.40)
\
We can bring this equation into agreement with Wien's thermodynamic law by letting be proportional to u>
:
= *o.
(1.41)
We
then obtain Planck's equation

fiu*
Pu)==
S
n<j(**
r
1)
'
(1.42)
which was a brilliant achievement of quantum theory. The quantity ft 1.05 10~* 7 erg.sec, which has the dimensions of
action, is called
Planck's constant. 3
At low frequencies
f^<
l),
the exponential e* to/kT may be expanded
in the form of a power series in ft>/W. Restricting ouselves to the linear terms of the expansion, we have
In
6.6249
the literature, Planck's constant is more often taken as the quantity h 27 10 to the frequency l/: sec, which relates the energy erg
(
= hv
13
Thus, Planck's formula (1.35).
equation
(1.42)
reduces to the Rayleigh-Jeans
In the case of high frequencies (jy!>

the
1),
we may
neglect the 1 in
denominator of Eq. (1.42) and write
pw
in the
form
/I 1
V
P.
CD
fl<Q/hT
rcV
-^"/
/I
Q\
Planck's equation (1.42), which describes the dependence of the spectral density p^ of thermal radiation on the frequency u> t is in excellent agreement with experiment (see Fig. 1.1). From Eqs. (1.42) and (1.18), we can find the total radiation
density
OO
00
= J^-nM-3rrIntroducing the variable

l
<
M4)
fi<o/kT
and considering that
we
obtain the well-known Stefan-Boltzmann law 4
where
= 7.56- 10-"
erg.cm-
.deg-
4
.
(1.46)
From Eq. (1.42), it can be seen that the spectral density of black-body radiation has a maximum at some value of co and that the position of this maximum changes with temperature. The equation governing the position of this ma^pum is called Wien's displacement law. More often, Wien's displacement law is expressed
Usually it is not the density u which is measured, but the energy oT which is radiated per second per square centimeter of the black body's surface within a solid angle of 2lT
fl erg
277).
In this case, the

2
Stcfan-Boltzmann constant
is
a-
277
a =
877
a = 5.67
4
10~
cm
deg
sec
14
in
X
terms of the spectral distribution with respect to the wavelengths To determine p A we can use the expression for u
,
00
..;,
Since
X
2^c/a),
transforming
to the spectral distribution
over the
frequencies, we have
from which we
find
Px
2itc
_
X max
\K*cU
2^5
(1.48)
1)
X*(*n
To determine the wavelength
at which
px
has
its
maximum,
we
set $CM
= 0:
-5-h-
<
Setting 2KcH/kTX max
y,
we
obtain the equation

y
= 5(l-rv)
<r
8
)
whose solution can be given with good accuracy

i/^5(l
in the
form
= 4.965.
Thus
max is related^ the temperature 7 by the equation
=3i = =
*
0.29
cm-deg,
(1.49)
which expresses Wien's displacement law, and where b is the Wien's constant. According to this law, as the temperature of an ideal black body increases, the maximum of the radiation intensity is shifted towards shorter wavelengths (see Fig. 1.2).
15
Equations (1.46) and (1.49) relate Planck's constant H and Boltzmann's constant k to the constants a and b. Knowing the numerical values of a and b we can determine ft and k. This is the way in which a numerical value was first
,
obtained for h and a better value found for
k.
1.44
192
Fig. 1.2. Wien's displacement law Curves of the spectral energy distribution as
a function of temperature for an ideal black
body: Amax T
29 cm .deg
Recapitulating, it follows from Planck's hypothesis that processes such as emission and absorption involve discrete quanta. In other words, the eivergy change of particles involved in these processes is discontinuous and not smooth as would follow from the laws of classical physics.
D.
EINSTEIN'S
PHOTON THEORY
In deriving his equation, Planck assumed that the energy of the oscillators is quantized. The original version of the theory, however, does not provide any physical justification for this property. Indeed, Planck himself chose to attribute the "special properties" to the heated body rather than to the electromagnetic radiation. The second important step towards the jgêlopment of quantum theory consisted of Einstein's hypothesis mat oscillators absorb and emit radiation in discrete amounts because electromagnetic radiation itself consists of discrete particles, called photons, which carry an amount of energy ftu>. In effect, Einstein interpreted Planck's equation as a description of the corpuscular properties
of light.
We
photons.
shall
now attempt
to
develop an elementary theory of
16
According
to classical theory, the
energy of a
2
light
wave
is
= -^
(' + //
<Px
= I- J E
<Px,
(1.50)
where
all
d*x is
an element of volume, and the integration extends over
space. The electromagnetic classical theory is
momentum
ji
of the light
wave
in
According
to Eq. (1.4), this
can also be written in the form
Comparing
(l.Sla)
at
momentum
and (1.50), we find the relation between the and the energy e
:
=*
E
.
(1.52)
In the theory of relativity, a similar relation
between energy
t
holds for particles with zero rest mass and it can easily be obtained from Eq. (1.14) by substituting m n -Q <P and 4 0. From these considerations, Einstein concluded that an electromagnetic field can be considered as a set of particles called photons, with zero rest mass and the energy
and
momentum
w.
(1.53)
For
the photon
momentum
n
the following equation is obtained:
= #*2 = #*- = nk,

is the
(1.54)
where
2rcft
and
=-
wave vector
= --
is
the
wave
number).
On the basis of these concepts, in 1905, Einstein constructed a quantitative theory of the photoelectric effect, which had been discovered by Hertz in 1887. What is observed in the photoelectric effect is the following: the potential difference required for a spark to jump between two small charged spheres is reduced if the

illuminated. To explain this the simple equation postulated
17
cathode
is
phenomenon, Einstein
Mo#
3
~-a.
iw
.
/I
c;c;\
j^to
U/.
Â.tJcJ/
This
is essentially the
law of conservation of energy and indicates

!~ of the ejected electron
is
that the kinetic
energy
equal to the
work function
difference between the energy of the absorbed photon Hw and the W of an electron in the metal. It is obvious that if cannot be ejected from the metal. Only if the electrons tiux^W, can electrons leave the energy of the incident photons exceeds metal. The experimental verification of Einstein's theory of the photoelectric effect provided striking confirmation of his basic conclusion that the energy of the ejected electrons depends only on the frequency of the incident light and not on its intensity, and that the
emission
light
a)
of photoelectrons begins only when the frequency of exceeds a certain limit (so-called threshold frequency)
The_implications of the photon theory were brought out and verified in 1923 by experiments on the scattering of x-rays by free electrons (ite Compton effect). jrEe Compton effect was, particularly "interesting "because it .confirmed not pnly the law of c observation of energy (which was already verified by the photoelectric effect) but also the law 3 Scat rm g of ight b y ' free Flg of conservation of momentum. } ; ff electron (the r Compton effect) It is well known that, in classical theory, the frequency of light does not w). By contrast, change when it is scattered by a free electron (u/ #u> is transferred in quantum theory, part of the photon's energy e to the electron (see Fig. 1.3). Consequently, the energy and frequency of a scattered photon should generally be somewhat smaller (e'<e, u/O) . To find the dependence of frequency on the scattering angle, let us write the laws of conservation of energy and momentum, treating the photons as particles:
"
" -
= =
fito
ftu/
c (tn
Wo)
Here m and w m /)/l p* represent the mass of the electron before and after collision; v is its velocity; p w/c; Hk n^/c and
18
represent the momentum of the photon before and after scattering* We rewrite Eq. (1.56) in the form
tik'
fno'/c
u)
u/
2
-
j-
(m
.
(1.56a)
k R
K K
Taking the square of these equations equation from the second, we obtain
too/ (1
and subtracting the first
cos
0)
= !?-
(ceo
ceo').
(1.57)
X' 2irc//, and dividing (1.57) by <DO/, we find 2rcc/u), Substituting X an expression for the increase in wavelength of the scattered light
AX
X'
= 2X
sin
!-,
(1.58)
where
is the
Compton wavelength
of the electron
cm.
therefore see that, according to quantum concepts, the waveof the scattered light X' must be greater than the initial wavelength length X(A/>X) since w'O. This difference increases with the scattering angle 0. Since the Compton wavelength X is relatively
small, lengths
We
Compton scattering
( 8
is
observed
rays
).
x-rays and
gamma
at relatively short waveIndeed, for visible light
(X~10
cm)
whereas for x-rays (X~
10
10~
cm)
in the
Therefore, the Compton shift can be observed experimentally only second case.
In his experiments, Compton studied the scattering of radiation from an x-ray tube by graphite and other substances (lithium,
beryllium, sodium, potassium, iron, nickel, copper, and so on) at different angles ft. The spectral distribution of the intensity of the scattered radiation at different scattering angles was measured by means of an ionization chamber.
19
Figure 1.4 shows the spectral distribution of incident and scattered waves. If the incident wave (upper curve) has one maximum, the scattered wave (lower curve) will have, in addition to this maximum, a second maximum at a longer wavelength. The
Incident
wave
X
Wavelength
Scattered
wave
Wavelength
Kig. the
1 1-.
Compton
Spectral distribution of x-rays in effect before (upper curve)
and after (lower curve) scattering
maxima must correbecause the distance increases with the scattering angle, and, in addition, it does not depend on the type of scattering material [both these facts are in accord with Eq. (1.58) 5 ]. The unshifted maximum corresponds to scattering by electrons which are strongly bound to the nucleus (or more precisely, electrons whose binding energy is greater than the energy of the x-ray quanta). The shifted maximum corresponds to scattering by electrons which are so weakly bound to the nucleus that, in practice, they can be regarded as free. Thus the results of Compton' s experiments completely confirm the quantum nature of light (that is, the photon theory).
distance between the wavelengths of the two
spond to the Compton
shift; this is
Only the intensity of the maxima depends on the type of scattering substance. As the atomic weight of the scattering substance increases, the intensity of the unshifted maximum increases, and that of the shifted maximum decreases.
Chapter 2
The Bohr Quantum Theory

BASIC INFORMATION ON PROPERTIES OF
A.
ATOMS
A theory of the atom was developed only after reliable experimental data had been obtained from studies of the effects described below* 1) Emission of light by atoms. From careful studies of the radiation of atoms, it was established that they have bright-line spectra and that the lines are arranged in certain definite series. For example, all the lines of hydrogen are described by Balmer's formula
1
where R
n'
Rydberg constant, and n'and n are integers. Setting we obtain the Lyman series, which lies in 4, 2 and n = 3, 4, 5, ..., the ultraviolet part of the spectrum. For n' we have the Balmer series, which is located in the visible part of the spectrum and is, therefore, easiest to study. Formula (2.1) can also be written in the form of a difference between two quantities
is the
and
2, 3,
In
spcctroscopy,
it is
customary
to write
Balmcr's formula in the form
where the Rydberg constant

constant in Eq
for
hydrogen
is
(2 1) is related to
K hp by
29
K qp 109,677. h cm" the equation

10
15
1
.
The value
KS
l
of the
Rydberg
2nvtf sp
2/7- 3
sec
20 66
10
sec
Equation (2.1) is not too convenient for spectroscopial use, since it is usually the wavelengths which are experimentally determined. In developing a theory, we do not need to take this reservation into account, since we can always transform from the frequency oj to the wavelength A
-
J?
277 c
THE BOHR QUANTUM THEORY
21
which are called spectral terms. For the hydrogen atom, these terms are given by
Tn
~
n2
as a This possibility of representing the radiation frequencies difference between two terms is a consequence of the Ritz combination principle, which has important spectroscopic applications in regard to the hydrogen atom, as well as more complex atoms. For
example, hydrogen was initially found to have two series, corre2 (the Balmer sponding to /i'=l (the Lyman series) and to n' 2 series). On the basis of the Ritz combination principle, a third 3 This series was /i' and was /i series 4, 5, 6,,... predicted with later discovered by Paschen in the infrared region of the spectrum. 2) The behavior of an atom in external electric and magnetic fields and, in particular, the interaction of the atoms of a substance with fast particles passing through it. The most important experiments in this area were conducted by Rutherford, who succeeded in finding the distribution of positive charges inside the atom from
the analysis of fast-a -particle scattering. 3) Finally, investigation of various properties of molecules provided important data pertaining to the properties of atoms. For example, the formation of simple homopolar molecules and the
valence theory found their explanation only on the basis of the modern quantum theory of the atom.
B.
THE CLASSICAL MODEL OF THE ATOM
Once it had been established that an atom consists of a positively charged part associated with most of the mass, and of light, negatively charged electrons, attempts were made to construct a static model. The reason this approach to the problem was adopted is that, in classical electrodynamics, an accelerated electron emits radiation, the amount of energy emitted per unit time being
*
different
=".
(2 - 2)
The Ritz combination principle was first formulated a? follows if there are two frequencies belonging to the same series, the difference between these freis also a
quencies
series.
The concept
frequency which can also be emitted by the atom, but belongs to another of "terms", permits a relatively simple explanation of this. Indeed
Hence
and thus the Ritz combination principle leads directly
to
Eq
(2 la)
22
NONRELAT1VISTIC QUANTUM MECHANICS

e
where
is
the
ay
elementary charge),
is the
4,80 10" i0 esu is the electron charge (e acceleration of the electron, and c is
Fig. 2.1. Thomson's model of the hydrogen atom (7, - 1)
The positive charge Zr Q is distributed over uniformly

the
volume of a sphere of
radius
The electron (with
the velocity of light in vacuum. The minus sign in front of dE/dt shows that the energy of the electron decreases as a result of the emission of radiant energy. Since an atom does not radiate in the ground state, it follows from the classical theory that the charges in the atom should be at rest. The most interesting classical model was that of Thomson, according to which the positive charge uniformly filled the entire atomic volume, and the electronic, that is, negative point charges were located inside the atom. For example, in the hydrogen atom, the positive charge was supposed to fill uni(see Fig. formly a sphere of radius /? 2.1). The charge density inside the sphere
charge -t' ) is located at a distance x from thf center of

the atom
was
(for
Z=
1)
3*0
In the ground state, the electron was supposed to be located at the center of the sphere, where the electric field is zero. At a distance r .v<]? from the center, the electric field E is directed along the radius and its intensity can be found from Gauss law:
Hence
*
'
(2.3)
e Q and mass w is Therefore, if an electron with charge e placed at a distance x from the center of the atom, it experiences a quasi-elastic attractive force towards the center
With this force, the differential equation describing the motion of

the electron is
j?-f.a)5A:
0,
(2.4)
The solution
of this equation is
23
where
Substituting for to the fundamental frequency observed in the Balmer series, we obtain a very reasonable value for the radius of the
atom, namely, RQ ~IQ-* cm. This value is than the classical radius of the electron
many times greater
The Thomson model agreed completely with the classical Lorentz theory, according to which atoms can be represented as harmonic oscillators. Unfortunately, the Thomson model could not
explain the regularities of the line spectra of atoms and, in particular, the spectral series of hydrogen that are described by Eq. (2.1). Indeed, from the standpoint of classical theory, the Thomson model could emit radiant energy only at the fundamental frequency u) or,
at best, at its
harmonics
ci)
/I
/iaj tt ,
(2.5)
to the Thomson model was dealt by the experiments of Rutherford, who showed that the positive charge is not distributed throughout the entire volume of the atom, but is concentrated virtually at one point. Nevertheless, the Thomson 3 potential inside a nucleus of finite dimensions, with a charge Ze distributed the is volume, uniformly through
where /i=l, 2, 3,... The decisive blow
3
If
the charge of the nucleus is
Ze Q the electric
,
field intensity inside the nucleus is
= ~
__
'
RQ
from which, using the boundary condition that at
point charge
dr
r
~ R the potential Q
is the
same as
for a
we
obtain Eq. (2
6).
24
does play an important role, especially when corrections for the of the nucleus must be made. Moreover, when mesic atoms with large Z are formed (in a mesic atom, an electron is replaced meson, whose mass is 207 times larger than the by a negative mass of the electron), there may be states in which the negative u meson is always inside the nucleus. In this case, the motion of the M meson will be determined mainly by the potential (2.6). The appropriate equations of motion will, however, be quantum-mechanical, rather than classical (see Chapter 21).
volume
\JL
C.
RUTHERFORD'S EXPERIMENTS AND CONCEPTS OF ATOMIC STRUCTURE
Our present model of the atom is based on the famous experiments conducted by Rutherford in 1911 on the passage of alpha particles through matter. It is well known that alpha particles, which are products of nuclear disintegration, possess a sufficient
energy to penetrate into an atom. At the time when Rutherford conducted his experiments, there was no other source which could produce sufficiently heavy particles (that is, particles with a mass comparable to the nucleus) with an energy great enough to penetrate the atom. particles through thin sheets of metal (foil), By passing alpha 4 Rutherford showed that most of the alpha particles which pass through the foil are scattered through relatively small angles (2-3) from their initial direction of motion. Within the framework of the Thomson model, these small deflections could be explained in terms of the statistical theory of random processes, because of the relatively weak interaction between the atoms and the alpha particles. Rutherford and his co-workers, however, also detected individual deflections of alpha particles through very large angles of up to 180. Although the number of these deflections was very small (for example, when a beam of 8000 primary alpha particles from RaC is passed through platinum foil, at most one particle is deflected through an angle greater than 90), it was, nevertheless, much larger than the number which could be predicted on the basis of superposition of a large number of small random deflections. Large scattering angles were also observed when alpha particles were passed through a gas. These could be easily seen in photographs taken in a Wilson cloud chamber. From a general analysis of his experiments, Rutherford established, first, that atoms are fairly transparent to alpha particles
In earlier experiments, Rutherford and his co-workers had established that alpha particles have the same mass as the helium atom, and a positive charge which is twice the magnitude of the electron charge is It now known that alpha particles are the nuclei of
helium atoms
25
(that is, their structure is relatively "open"); and, second, that large deflections can take place only if a very strong electric field exists inside the atom. This electric field must be produced by a positive charge which is associated with a large mass and concentrated in a very small volume. We note in this connection that, according to Eq. (2.6), the largest field produced by a nucleus of radius R Q is
To model
explain these
of the
atom
in
results, Rutherford proposed a planetary which the structure of the atom resembles
a planetary system. A positively charged nucleus constituting almost the entire mass of the atom is concentrated at the center in a very small volume of radius /? 10" 13 10"", and charged
electrons
about this nucleus in closed orbits like planets around the Sun. We note that the potential energy of the Newtonian attraction between a planet of mass m and the Sun (of mass M)
move
Newt
__
X/llAf
~>">
where 7. is the gravitational constant and has the same form as the potential energy of the Coulomb attraction between an electron and a nucleus
From this model, Rutherford developed a quantitative theory of scattering. His calculations were based on the assumption of a Coulomb interaction between the alpha particles and the nucleus. The influence of the atomic electrons was neglected in the first approximation, since their energy is considerably lower than the energy of the bombarding particles.
particle
Let us find, following Rutherford, the trajectory of an alpha s moving in the field of an infinitely heavy point nucleus having a charge Ze . Our calculations will be carried out in a
If the fimteness of the nuclear mass ,\/ nut is taken into account, the nucleus has a certain recoil (like that of the alpha particle) as a result of the interaction In this case, all the calculations must be performed in the center-of-mass the results system and, obtained for the case M nuc -- *, it is necessary to replace the mass of the alpha particle M by the reduced mass
Af
M red
(see Chapter 12, Section
M nuc
i
M nut
for a
discussion of the reduced mass)
26
coordinate system whose origin coincides with the nucleus (see Fig. 2.2). Since the field produced by the nucleus is centrally symmetric, in determining the trajectory of the alpha particles we can use both the law of conservation of energy
= const,
momentum
x
(2.7)
and the law of conservation of angular
L==M
where
and v
(r
v)= const,
(2.8)
is
the
mass
of the alpha particle, r is its coordinate,
is its velocity.
Fig. 2.2. Diagram for the derivation of Rutherford's formula for the cross section of elastic
scattering of alpha particles by nuclei.
Let us introduce the polar coordinates the particle is given by
and
?.
The velocity
of
where v\\=f and vt

and perpendicular
and
$
= r$
are the components of velocity parallel

vector
r,
to the radius
respectively,
andr=~
=t>-
We
then obtain, instead of Eqs. (2.7) and
(2.8),
(2.10)
Lz
=M
(r x v\
= Mtf*$ = const
(2.11)
In the absence of interaction, the alpha particle would pass the nucleus at a distance b (this distance b is called the impact parameter). Setting the initial velocity equal to v (that is, the velocity
{}

r
27
2.2),
_^_oo and
(2.11)
f-*rc, as follows from Fig. can be reduced to the form
then (2.10) and
(2.12)
(2.13)
where the
initial
energy E
is related to the initial velocity VQ
by
the equation
^=
Introducing the
..
(2.14)
new variable
K
=l
(2.13),
(2.15)
and noting that then, according to
|t|
= *=ttf,
(2.16)
and
where
u'
= -~, we transform Eq.
(2.12) to
Differentiating this equation with respect to
cp,
we
obtain
Hence
n
= ^cos ? + Bsin<p --jjgg,.

coefficients
/4
(2.20)
The unknown constant from the initial conditions
and 5 can be determined
and
28
=--0 in Eq. (2.20), Setting <p^=* and u
we
obtain
(2.23)
and, consequently, applying the condition (2.22) to Eq. (2.20),
we
have
B = y.
Thus,
<
2 - 24 )
we
finally obtain
(2.25)
This equation gives the relationship between the absolute value of the radius vector r and the polar angle ?, and thus describes the trajectory of the alpha particle in the Coulomb field of the nucleus. It is an equation for a hyperbolic trajectory in polar coordinates.
cp
definition, the scattering angle & is equal to the angle 7U at which the length of the radius vector r becomes ) (<p^'?o
By
infinite
(that is,
from Fig.
2.2,
= =
-i
'
).
Therefore, from (2.25),
we
find
cot C0t
i-z ~
bEg
> '/.el
(2.26)
It
follows
that
attaining 180
the for 6
scattering
angle increases as
decreases,
(see Fig. 2.3).
Q>*0)
Fig.
Ie
of
2 3
I/
Dependence
the
scattering
h,
angle
on the impact parameter
where
Equation (2.26) can be checked experimentally by photographing

the tracks of alpha particles in a Wilson cloud chamber. maximum scattering angle, it is possible to compute the
From the minimum

value of
b
,
29
which turns out

of
The actual form
to be of the order of the nuclear radius. interaction between an alpha particle and a
nucleus can, however, be determined
more
accurately from an
investigation of the cross section for the scattering of alpha particles by nuclei. For this purpose, Rutherford calculated the relative number of particles scattered at an angle or, to be precise, the number of scattered particles which we would expect to find within the solid angle
di3
= 2*sinOd.
(2.27)
Suppose N particles impinge per unit time on a unit surface placed perpendicularly to the original velocity of the particles. From Eq. (2.26), it follows that the scattering angle depends only on the impact parameter b. For a particle to be scattered through an angle &, it must strike a ring formed by two circles with radii b &ndb db. The area of this ring is 2*bdb. Therefore, the number of particles which hit this area and then, as a result of scattering, are found within the solid angle d& is
dN = N.2\bdb\.
(2.28)
From Eq. (2.26), we obtain for the relative number of particles scattered through an angle d
(2.29)
The ratio dNjN has the dimensions of area and is called the differential cross section. It is usually denoted by do. Taking the derivative of cot 2
in Eq. (2.29),
we
obtain the Rutherford formula for
elastic scattering of alpha particles by a
Coulomb center:
This formula no longer depends on parameter b. If all the quantities in this equation are kept constant, with the exception of 0, we would expect the following equation to hold:
do sin 4
=
\
-Z<>
/
d2
= const.
o
(2.31)
sin
4
For
be
*7c
it
is found,
however, that the quantity
y do
ceases
to
constant and begins to decrease somewhat. This fact was explained by Blackett, one of Rutherford's students, who studied
30
the limits of applicability of the
Coulomb law. Blackett took a large
number of photographs chamber and calculated
of tracks of particles in a Wilson cloud the frequencies with which the various
scattering angles occur. From analysis of the experimental data, he established that the number of particles observed at large scattering angles [according to (2.26), large angles correspond to small values of the parameter b ] is markedly smaller than the number yielded by the formula (2.30). On this basis, Blackett concluded that, in air, for example, Coulomb's law is valid down to distances of the order of 3-10 u cm. At smaller distances, there is a deviation from Coulomb's law. Indeed, at b<^^lfr * cm, the interaction between the alpha particle and the nucleus appears to take the form of a strong mutual attraction. Further experimental investigations have confirmed the existence of a characteristic attractive force at distances less than 10 12 cm. This attraction drops off rapidly with increasing distance from the nucleus. The Rutherford formula (2.30) can be used to find the number Z from an experimental determination of dN and N. This was undertaken by Chadwick, another of Rutherford's students. Chadwick showed that the value of Z is very close to the atomic number, which gives the element's position in Mendeleyev's periodic table. The existence of this phenomenon was rigorously proved at a later
l
date.
Thus the experiments of Rutherford and his colleagues definitely established the planetary model of the atom. These experiments proved that the positive charge of the electron is concentrated in 12 a nucleus with dimensions of 10~ 13 10~ cm, and that, inside the atom, Coulomb forces keep the electrons moving in orbits with a radius of the order of 10~ 8 cm.
D.
THE BOHR THEORY
model
First of all, let us attempt to develop Rutherford's planetary of the atom on the basis of classical theory. We shall re-
strict ourselves to the case of a hydrogen-like atom,
namely, an
atom with a nucleus of charge Ze with a single electron (of e ) moving around it. Particular examples of hydrogencharge e like atoms include hydrogen (Z=\), ionized helium (Z = 2), and
so forth.
we
Introducing the polar coordinates rand cp(# rcoscp, y rsir\y) 9 obtain the following equations for the kinetic energy and the Coulomb potential energy:
31
Then
the Lagrangian is
X=
where m
is the
*L (,
+
we
y)
+ Z^
(2.32)
mass
of the electron.6
this Lagrangian, the motion of the electron:
From
obtain the following equations for
ft
Here
and
pr
= ^jr = mj
(2.34)
represent the generalized momenta associated with the <p and r coordinates, respectively. Since <p does not occur explicitly in words, it is a cyclic coordinate), it follows that (in other
-^-==0. Therefore, the corresponding generalized

cty
momentum
is
constant of the motion
pv
= m r^ = const.
Q
(2.35)
This
is the
law of conservation of angular momentum, which
is well
known from classical mechanics. The second conservation law,

namely, the law of conservation of energy
E
is
= T+V = const
t
(2.36)
obtained from the condition that the time
does not occur ex-
plicitly in the Lagrangian. shall consider only the simplest case, that of circular orbits,
We
for which
-0.
Accordingly, pr
=m
Qf
vanishes, and from (2.33)
we
have
=
6
0.
(2.37)
In the following chapters, the
mass
of the electron in nonrelativistic theory shall be
denoted by
since
will
be used
for the
magnetic quantum number.
32
NONRELATIV1STIC QUANTUM MECHANICS
Hence
*'-l"
Therefore, we obtain for the energy of the electron
<
2 - 38 >
E = -|-Vy
-^-^--l-^^l V.
(2.39)
Let us now express the basic parameters of the atom in terms

of the adiabatic invariants, which were introduced by Ehrenfest. According to Ehrenfest' s definition, in the case of periodic motion,
the quantities
It
(2.40)
(pi is a generalized momentum, and x, a generalized coordinate) remain constant during slow (adiabatic) changes of the parameters of the system (for instance, the charge). The In our case, there is only one degree of freedom (x <p). conditions (2.35) and (2.40) yield the equation
4
P,
= "&** = -&,
*=-S5'
(2.41)
or
<
2 - 42 )
From
<?
Eqs. (2.38) and
(2.42),
we can
obtain expressions for
and
in
terms
of the adiabatic invariant:

(2 ' 43)
^
According
to Eq. (2.39), the
'=43i35r M ,= =!'i.
is then
(2.44)
energy of the electron
= _-2**-^l.
variant:
(2.45)
It follows that the frequency of mechanical oscillation v is given by the derivative of the energy with respect to the adiabatic in-
33
This relation holds not only for the case under consideration, but also for any periodic or quasi- periodic motion.7 Moreover, a system performing any periodic motion can generally emit radiation not only at the first harmonic 1, but also at the harmonics k = 2, 3, 4, . . . . The following expression is, therefore, obtained for the classical frequency of radiation:
k=
= Av
==A4j~.
(2.46a)
The classical theory of the planetary model of the atom led to number of difficulties when it was used to explain the radiation of atoms. Since this model is dynamic, it follows from classical
a
electrodynamics that the centripetal acceleration of the electron

(w
v =~
z
9
where
v is its velocity,
and a
is the
radius of its orbit)
will cause the electron to lose energy at the rate
dE __ ~~
dt
e*w*
c
8
it falls into the nucleus. This is not, however, what actually takes place, and atoms can exist in anon- radiating state for an arbitrarily long time. Moreover, according to the classical theory, the frequency of the radiation should be the same as the frequency
until
of mechanical oscillation (the fundamental frequency CD u> 2irv or, alternatively, an integral multiple of this frequency (one of the m where harmonics 2, 3,4,. . .). Once again, this prediction does not explain Balmer's experimentally established formula (2.1) for the lines in the spectrum of radiation.
)
>
= =
n=
We define
after a certain period of time.
periodic motion as a motion in which a point returns to its initial position As examples of this, we can take harmonic motion
x
-
a cos fM
or motion about an ellipse
a cos
tilt
and
b sin (Ot
in
a special case of which is motion about a circle (a = 6). Quasi-periodic motion is motion which a point does not return, as a rule, to its initial position, but each individual coordinate reassumes its initial value after a certain period of time (which is different
for
each of the coordinates). As an example of quasi-periodic motion, we can take

x
-
a cos
co it o) 2 *
y - b cos
where the frequency
(o\ is
incommensurable with
a> 2
34
NONRELATIV1ST1C QUANTUM MECHANICS
The solution to these difficulties was found in 1913 by Niels Bohr, who added two postulates to the classical laws of motion. First, Bohr assumed that each atom has a series of discrete stationary states in which the electron does not emit radiation, even though its motion is accelerated (the postulate of stationary states). According to Bohr's theory, these stationary states can be determined by quantizing the adiabatic invariants:
,dq { =inh,
(2.47)
where the quantum number n can assume only integral values,

is, n
that
1, 2, 3,
could assume Second, Bohr hypothesized that, when an electron passes from a stationary state with energy E n (the initial state) to a state with an energy *'< (the final state), the atom radiates a quantum of
... (in classical mechanics, the adiabatic invariant/ any constant value).
energy fa^ha*
o>
(the
frequency postulate), whose angular frequency
is
a,
= *-"*'.
(2.48)
This equation can be written in a form similar to the classical expression (2.46a) for the frequency of radiation:
*
= (rt_ n ^. = *j^. =
(2.48a)
Here the quantum number k n n' can be interpreted as the corresponding harmonic, and, moreover, the derivative of E with respect to /, which occurs in the classical expression, is replaced by the ratio of the finite increments. Bohr's postulate that the stable energy states of atoms form a discrete spectrum was confirmed in experiments conducted in 1919 by Franck and Hertz. Passing an electronbeam (a current) through mercury vapor, they showed that, for electron energies less then 4.9 ev, the collisions of electrons with Hg atoms do not effect the magnitude of the current. When the electron energy attains 4.9 ev (see Fig. 2.4), the current suddenly drops. With further increase of the electron energy, periodic sharp dips of the current are observed (approximately every 4.9 ev). This phenomenon can be very simply interpreted from the point of view of the Bohr theory. Let us suppose that the energy of an unexcited Hg atom (that is, the atom before collision) is and, in conformity with the first Bohr postulate, let us take the next possible energy value to be + 4.9 ev. It then follows that a beam of electrons of energy 1== 4.9 ev is not sufficient to raise the atoms to an excited state;
35
therefore, the collisions are elastic and the current does not change. If however, ^> 4.9 ev, the electrons in the beam may give up a part of their energy (namely, 4.9 ev) to the atoms, and therefore, the current will change. If the electrons' energy lies in the range 14.7 ev 9.8 ev, the transfer of energy to the atoms can occur twice; 4.9 ev is given up in the first collision and 4.9 ev in the second.
>>
4.9
9.8
/4.7
Electron energy, ev
Fig.
2.4.
The dependence
in
of current on
!
electron
energy experiment (1 ev
the Franck-Hertz 2 10 /300 ^
erg).
We
shall
now use
the first and second
Bohr postulates [Eqs.
(2.47) and (2.48)] to construct a theory of the hydrogen-like atom. In the equation for the radius of the orbit (2.43) and the equation for the energy (2.45), let us substitute for / its quantized value, which
from Eq.
(2.47) is
We
thus have
(2.49)
(2.50)
For n=\ atom
we
obtain the energy of the lowest (ground) state of the
r M
//ipZ
gfl
/o
'
2ft
l^.o-l-J
t;
i \
and the corresponding radius

ft
(2.52)
36
where
a
= -?L-^ 0.529- ID'
cm
(2.53)
is the
radius of the first Bohr orbit. The second Bohr postulate (2.48), combined with Eq.
a> rtn
,
(2.50) for
the frequency of radiation
gives
__ aw _
Kn
En __
.
a-
Z^ /I - -jar^T n
OT O
-
\
-
1, we obtain Balmer's formula (2.1). The Bohr Accordingly, for 2 theory also relates the Rydberg constant, established prior to this on a purely empirical basis, with the Planck constant ft:
= -*.
(2-55)
This derivation of Balmer's formula, with a value for the Rydberg constant that agreed with experiment, was one of the greatest achievements of the Bohr theory. In spite of this success, however, the Bohr theory has a number of inherent defects, which
became increasingly important in its further development. In the first place, the Bohr theory was obviously semiclassical in nature. In addition, the Bohr theory could be used to compute only the frequency of the spectral lines, and not their intensity. To find the intensity, it was necessary to resort to classical electro3 dynamics, on the basis of the so-called correspondence principle. a could not be to satisthe Bohr used construct postulates Finally, factory theory of multielectron atoms, including that of helium, which possesses only two electrons.
According
theory (here,
theory).
i
to the correspondence principle, in the limit, all the results of a previous lassical electrodynamics) should follow from a new theory (here, the Bohr
For instam e, for // (or at the limit ot large quantum numbers), tho results of quanIn exactly the same way, when tum mechanics should appioach the classical results 0, the results of the lelativistic theory should approach the nonrelativistic results, ft and so on. Thus, the correspondence principle enables us to check the extension of a theory by requiring it to reduce to the classical picture at the limit.
*
>
In the initial stages ot a new theory, when it cannot yet be used to investigate certain phenomena, the correspondence principle may be used for reasonable extensions of the Thus, for example, the Bohr theory could predictions of the old theory to the new theory be used to calculate the frequency oi the radiation, but not its intensity Bohr was able
changes
determine the intensity from the correspondence principle, according to which the only - n ot the quantum number n that were allowed (the selection rules _\ n') were those that coincided with the classically allowed harmonics of the radiation. The quantum
to
theory of radiation of light
was completed only with
the development of quantum electro-
dynamic s.
37
Accordingly, we shall not discuss in greater detail the subsequent history of the Bohr theory. This theory represented only an intermediate stage in the development of quantum mechanics, a theory which can be used to determine both the frequency and intensity of the radiation of atoms. However, we thought that at least a brief discussion of the basic principles of the Bohr theory was advisable, for this theory still retains considerable heuristic significance. In particular, the Bohr theory often serves as the starting point in the analysis of many results related to quantization. The conclusions that follow from Eq. (2.54) will be discussed in Chapter 13, where the problem of the hydrogen-like atom will be solved quantum-mechanically.
Using the Bohr theory, quantize a hydrogen-like atom for the case of is, find the spectrum of the energy levels). Show that, in the nonrelativistic approximation, the coordinates r and y have the same frequency of variation; that is, the motion is periodic. Show that the formula for the frequency of the radiation of an atom remains the same as in the case of circular orbits* What new feature is introduced in the theory by considering elliptic orbits? Solution, Using condition (2.40) and noting that, in this problem, the generalized momenta corresponding to the <f and r coordinates are
//J
r 2<
j>
= const
and p r =5
//i fl
I/ 2//io+
we
find
r
max
()
'nun
There the second integral was calculated from the formula
min
Hence the energy
E is
By direct differentiation of this equation with respect to the variables lr and /f , easily shown that the frequencies with which the r and y coordinates change
it
is
dE
dll
coincide with one another. Using the Bohr quantization rule (2.47), invariants /r and / by nr h and n^h, respectively, obtaining f
we replace
the adiabatic
38
where
n~n r ~\-riy,
nr
= 0,
1, 2, 3,
. . .
and n^
1, 2, 3,.
. .
This expression
is
completely
identical with Eq. (2.50), which was derived for the case of circular orbits; therefore, the radiation frequencies remain the same as before. States with a given n r and different n^ have the same energy, but differ in eccentricity
c:
In
particular,
elliptic trajectory
when n^ n , we have a circular trajectory. As n^ decreases, becomes more elongated.
the
The new feature to appear in the case of elliptic orbits is the selection rules A/^r^ -h 1, A Ht 1, 2, 3,... . These account for the appearance of series that can also be 1 found from the correspondence principle. In the case of circular orbits, Aw bn^ (see also Chapter 13).
Find the classical equation of motion and the trajectory of an electron according to relauvistic theory. Show that, in this case, the r and <p coordinates change with different frequencies; that is, the motion is quasi-periodic. Determine the angle through which the perihelion of the electron is shifted during "one" revolution. Obtain a formula for the energy levels and find their splitting. Compare the results with the corresponding nonrelativistic problem 2.1.
I
Solution. Using the reiativistic Lagrangian function
(2.57)
f
where
we
obtain the equation of motion
d
dt
ni
v
r.
From
Since
/;,.
Eq.
(2.57)
we determine
r
(
the
generalized
momenta /i
=
(p
/)
V
and^,=
*^
--.
T -.
= ~J- r\
where
^=s ~-
it
follows, in accordance with the law of conservation
of energy, that
E==C
\/
m * C*
^P*^
"""
>s
~f"
~ COIlbt
which implies that
Hence we obtain the equation
of the trajectory
r==
1+icos-re
(2 ' 58)

where
39
-~Ze\E*
WfiC \
4
/,/, .+4^
It is
the perihelion,
apparent from Eq. (2.58) that the motion is quasi-periodic. For the shift Ay of we have from (2.58)
With the help of (2.40) and (2.56) we get
where
Then, for the energy
/:,
we
C
obtain the expression
=/fI
+ -.--
Z*e*
(2.59)
4*'
From
this
it
is evident that the
frequencies
(2.47),
to^
^.
y/(p
and ay
= y/
-.
will be different.
r
Using the Bohr quantization rule
we transform Eq.
(2.59) to
- - îf
where
^=
-i
e*
J
G^
.^
is the fine-structure constant.
2 Expanding the formula (2.60) into a series in a and restricting ourselves to quantities of the order of a* f we have
Since M varies from 1 to w t it follows from Eq. (2.61) that the energy levels, which V /i r -j- /i n closely are determined by the principal quantum number /i 9t are split into ed sublevels (this close spacing is a consequence of the small ness of or). spaced
4O

The fine-structure
splitting is
US ^n
=F
l.
-E
***. "
*****
n\ (n 9 '-{)
The splitting, or fine structure, of the levels, is a characteristic result of relativistic effects and essentially distinguishes the predictions of relativistic theory from those of nonrelativistic theory (see Problem 2. IX
roblem
2.3.
JAn electron
is located in a central field
Determine the values of s at which stable states of the system are possible. Answer, s <C 2. Hint. Use the expression for the effective potential energy
~~
2/rtS*
r*
if l/ e ff
and make use of the fact that stable motion
is
possible
has a minimum.
fcoblem 2.4
particles.
Find the scattering cross section for nonrelativistic electrons by nuclei
(Coulomo point charges). Compare the result with the scattering cross section for alpha
Answer.
_roblem 2.5J Show that in the relativistic scattering, unlike the nonrelativistic case, electroWCftfl be captured by nuclei. Obtain the total capture cross section. Solution. According to nonrelativistic theory, the trajectory of electrons in the Coulomb field of a nucleus is given by the equation
COS
-f-
cp
where
O f this equation describes At / hyperbolic motion, and therefore the capture of electrons is impossible. In relativistic mechanics, the equation of the trajectory has the form of Eq. (2.58). %* e If in this equation, 7 becomes imaginary. Ofor 
>
the particle falls to the center, which means it is captured. Since the angular momentum p^ of the electron is related to the initial and the impact parameter b by the expression /> the condition for
9
momentum p
=/^,
capture can be
written in the form b~ ^: -^
'-"
Hence
the total capture cross section is
cap
Chapter 3
Wave
Properties of Particles
A.
DE BROGLIE WAVES
As we mentioned in Chapter 1, the development of modern quantum theory began with the discovery that light has particle properties in addition to wave properties (characterized by the wavelength X and the frequency o>). The energy e and momentum JT of a quantum of light (photon) were established by Einstein as
e
= =
ft,
Av,
rt=ftft=J
(3.1)
Analyzing these equations, the French physicist de Broglie suggested that they could be generalized to apply to ordinary particles and, in particular, electrons. Generally speaking, de Broglie assumed that the wave-particle duality is not an exclusive property of light, but is also a characteristic of electrons and all other particles. Accordingly, a beam of free electrons, whose relative energy E and momentum p are related to the velocity v by the equation 2
1
E = -^=-, ^ p=
-W/I-?'
/l-p
(3.2)
should also exhibit wave-like properties. The corresponding frequency and wave number were defined by equations similar to
Einstein's:
= Aw and = **.
/>
(3.3)
This hypothesis was made by de Broglie with a twofold purpose: first, to provide a physical basis for the Bohr quantization; second, to explain the first experiments on electron diffraction (see below).
From now on, we shall leave it to the reader to distinguish between the relativistic energy (which includes the rest mass energy) and the nonrelativistic energy. Only in cases in which the two energies appear in the same equation will they be distinguished by some kind of index, for example,
\m
42

is
Consequently, the de Broglie wavelength of the moving particles
=A X=^k
P
(3.4)
The de Broglie relations (3.3) thus generalized Einstein equations (3.1), derived from the photon theory. These now became equally applicable to the analysis of light in terms of its particles and of moving electrons in terms of their wave properties. It is worth
if
noting that the dual Planck's constant
-ft
character of particles and light disappears is allowed to go to zero (the correspondence
principle).
we may describe
Taking de Broglie's equations (3.3) as the basis of discussion, the motion of free particles (along, say, the x axis) by the so-called wave function, which for this particular case is analogous to that of light and represents a plane wave:
^(x,t)
= Ae' ^'^ = Ae"^-^ = Ae'^

2rt
(l!t
''
p
^.
(3.5)
the standpoint of Eq. (3.5), we can attempt to explain Bohr's postulate of stationary states [see Eq. (2.47)]. The physical interpretation which we can offer is as follows: the only allowed circular orbits are those which are divisible by an integral number of de Broglie wavelengths, that is,
2
From
T=
(3.6)
Indeed, the
is satisfied.
wave function
is
Furthermore, since
single- valued only when this condition in the nonrelativistic case
\= m
'
Eq. (3.6) yields the Bohr condition for stationary states:

/in.
(3.7)
B.
EXPERIMENTAL OBSERVATION OF THE WAVE PROPERTIES OF PARTICLES
To investigate the wave properties of electrons and to prove that electrons have a specific wavelength X, it is first of all necessary to obtain a monochromatic electron beam (a beam of electrons
moving with the same
such a beam
is
velocity). One possible way of producing " by means of an electron gun" (see Fig. 3.1),
WAVE PROPERTIES OF PARTICLES
43
which emits electrons with a certain definite velocity v. In the nonrelativistic case (u<c), this velocity is given by the equation
(3.8)
300'
is the accelerating potential (in volts) of grid A relative cathode C. Thus, after passage through grid A (at which point no further acceleration is imparted), the electrons will have a de Broglie wavelength of
<D
where
to
the
=
~~
= ~-
--~
(3.9)
To exhibit the wave properties in the most pronounced form, one must impart to them the longest possible wavelength ^. This can be done by decreasing <1>. Since, however, a certain amount of energy, known as the work function, is expended in the ejection of electrons from metal (this energy is of the order of several ev and gives rise to a certain spread in the electron velocities), the smallest potential $ at which the beam will be relatively monochromatic is 15-20 volts. At these conditions, the de Broglie 8 wavelength of the electrons is approximately the same (\~ 10~ cm) as the wavelength of soft x-rays. Before we consider the physical nature of the wave -like character of an electron beam, let us discuss several experiments that have led to the direct detection of de Broglie waves. To begin
with, we shall take the electron diffraction experiments of Davisson and Germer who were
,
the first to
make experimental
observations of electron waves. Since the de Broglie wavelength of an electron is of the order
cm, they proceeded in the same way as for soft x-rays,

10
of
Fig. 3.
An electron gun that produces a 1 monochromatic beam of electrons. C is the cathode, and A is the anode.
using a crystal with a lattice constant of the order of 10 8 cm for the diffraction grating. The set-up of the Davisson-Germer experiments is depicted in Fig. 3.2. After coming out of the electron gun, the beam impinges perpendicularly on the surface of the crystal, where the electrons are scattered at various angles by the surface lattice. Since the penetration of the electrons into the crystal can be neglected, the diffraction grating can be regarded as two-dimensional. Accordingly, the position of the diffraction maxima can be determined from the condition that the path difference s dsin& (d is the
44
NONRELATIVIST1C QUANTUM MECHANICS

t>
is the scattering angle) two-dimensional lattice constant and equal to an integral number of wavelengths (see Fig. 3.3):
is
sintt,,
//A
=n
--,
(3.10)
where the integer n is the order of the given diffraction maximum. The diffraction maxima were detected with a galvanometer, which
registered the intensity of the different angles (see Fig. 3.2).
Electron
beam
of scattered
electrons at
gun
Galvanometer
II
L
,\
\
\
i
j_
Nickel crystal
3 2 Diagram of the Davisson-Genwr experiments on electron diffraction
Fig.
A second prediction of the de Broglie theory was also verified: increases, the angle namely, the prediction that as the potential cor responding to the nth diffraction maximum, will decrease in accordance with Eq. (3.10). Thus the correctness of de Broglie' s equations was completely confirmed by
(
l>
ft,,,
these investigations. It is well known that x-ray diffraction patterns are produced not only by single crystals, but also by polycrystalline formations. This was shown, for example, by Debye and Scherrer. Tartakovskiy and Thomson obtained diffraction patterns similiar to these x-ray
Tip. by a
\\.\\
Electron scattering
tion grating
diffraction
patterns by extending this
two-dimens tonal diffrac-
technique to electron
an electron
beam through
waves and passing a foil.- The

1
following theoretical explanation can be given for the diffraction pattern of electrons. The electron beam, with an energy of several thousands or even tens of thousands of electron volts, impinges upon a polycrystalline foil, where it
*
10
To reduce the absorption cm) was used
of electrons, a relatively thin foil (thickness of the order of
45
encounters single crystals and is reflected from them. The path difference s of two rays (see Fig. 3.4) is related to the lattice constant dot the three-dimensional crystal by the relation
(3.11)
where
is the angle between the ray and the lattice plane. Since <p the single crystals of the foil are oriented at random, a ray can leave the foil at any angle with respect to the original direction (see Fig. 3.5). Among the crystals, there will be some that are oriented at just the right angle to satisfy
Bragg's law
2d
sin
<p
= n\ = n
~-
(3.12)
g.
where
is an integer. Whenever n this is the case, a diffraction maxi-
3 I Reflection from a threedimensional crystal lattice
mum
occurs and a bright spot Q is found on the screen. Since the experimental apparatus is cylindrically symmetric, the bright spots form diffraction rings, whose radiuses R n can be found from the
relation (see Fig. 3.5)

tan 2 f ==
f ,
(3.13)
where L
is the distance from the screen to the polycrystalline foil. Since the angle <p is very small in these experiments (<p^sin< tan ?), Eqs. (3.12) and (3.13) yield
L 12-
10-'
(3.14)
that is, at constant L, d and n,

(3.15)
ments.
These relations were completely confirmed by Thomson's experiElectron diffraction patterns and Debye-Scherrer x-ray
are now widely used in studies of crystal
diffraction patterns structure.
It is worth noting that de Broglie's formula does not only apply to electrons and other elementary particles such as protons and neutrons, but also to complex nuclei, multielectron atoms, and even to molecules. True, their de Broglie wavelength is very small because of their relatively large mass. Nevertheless, Stern and Esterman have succeeded in observing the diffraction of beams of
46

in reflections
helium atoms and hydrogen molecules

crystals.
Polycrystalline
foil
from LiF
Screen
Pencil of electrons
Fig. 3.5. Diffraction of electron waves in a polycrystalline substance (the Tartakovskiy-
Thomson experiments)
A method involving neutron diffraction has been found extremely useful in analyzing the structure of substances. Neutrons have no electric charge and, therefore, pass freely through matter even at low energies (thermal neutrons), when their de Broglie wavelength is relatively large. All these facts provide convincing evidence that wave properties are displayed in some degree or other by all particles and that the de Broglie formula for X is of universal
validity.
The analogy between light waves and electron waves has led to the development of a new branch of physics, electron optics, which is devoted to the study of wavelike processes in electron beams. With the help of this new science, it has been possible to design
and construct electron microscopes, which have found wide application in modern techniques. In ordinary optical microscopes, the upper limit of the resolving power (and, consequently, the magnification) is of the order of the wavelength of the light which is utilized. To achieve the highest possible magnification, it is necessary to reduce the wave length of the light. The wavelength, however, cannot be made arbitrarily small: it is impossible, for instance, to construct an x-ray microscope, since no appropriate lens exists for x-rays. On the other hand, satisfactory electric and magnetic lenses can be produced for electron waves. Thus, the electric lens consists of a capacitor which has an aperture in the middle of the plate, while the magnetic lens consists of ordinary magnetic
coils.
4
Modern
thousand.
optical
microscopes
give
magnification
An electron microscope can give
a magnification of
the electron microscope, the proton microscope is solving power can be made even greater than that of the electron microscope.
approximately one or two more than 100,000. Besides also widely used at present. Its reof

C.
47
WAVE PACKETS. GROUP AND PHASE VELOCITIES
From de Broglie's hypothesis it follows that the motion of a free material particle having an energy E mc* and a momentum p mv can be described by means of the plane wave (3.5). The velocity u of. the de Broglie wave can be found as the time rate of displacement of a constant phase
Et~px = const;
that is, the phase velocity is 5
(3.16)
_**_ M=-JT= dt
E __ c2
p
=v
/Q 17\ f I U.I
I
'
According to the theory of relativity, the velocity of the particle v cannot exceed that of light in vacuum (c). However, the calculated phase velocity of the wave appears larger than c. This would indicate that it is theoretically impossible for a monochromatic wave to transport a particle or carry energy, since all particles and energy must travel with a velocity smaller than that of light, in accordance with the principle of relativity.
This equation can be obtained from the following simple considerations.
function depends only on the phase
this
The wave
(Et
px) -, therefore, at the time
h
phase moves
to the point x^
^ =
+ Af,
t f
Ax, which can be found from the equation

Et
t
Et 1 - px l
px = constant.
TC
Thus
EA - p A
=
is also the velocity of the
Hence the rate of propagation of the constant phase (which wave as a whole, as can be seen from Fig. 3.6) is
u -
Ax
A*
E
.
Position of the
&*
wave
at
time
the
wave
at
time tf-t+At
Fig. 3.6. The propagation of a monochromatic wave. During the time A(, the wave as a whole moves a dis-
tance Ax.
The phase velocity of
the
wave
is
At
48
An escape from this dilemma was found in the early stages of the development of quantum mechanics. This solution retained the wave properties of particles, which had received such striking experimental confirmation. Thus, each particle was associated with a group of waves of nearly equal frequencies, rather than with a single monochromatic wave. This was further suggested by the fact that the diffraction lines of electron waves were always observed to have a certain definite width. Thus, it seemed that several waves with very nearly equal frequencies were diffracted, rather than an individual wave. Yet another basis for this solution was provided by spectroscopic studies, which showed that all spectral lines are also characterized by a certain definite width. One advantage which results from using a set of waves with nearly equal frequencies rather than an individual monochromatic wave is that it is always possible to construct a wave packet whose resultant amplitude is appreciably different from zero only in a certain small region of space. This small region of space can be associated with the position of the particle. On the basis of these considerations, let us attempt to describe the motion of a particle by constructing a wave packet out of a continuous set of waves, assuming that the momentum //, which is connected with the wave number U by the de Broglie relationship
k'
p'/ft 9
ranges from
(//)
/;
to p Jr
(Ap
to
^p).
We
shall take the
amplitudes A
of the individual
waves
be
0.18)
0,
The resultant wave function
ty
is
(3-19)
In this equation, let us
change from the variable p'to the variable

,
;/-;/
;)'
~fj-'^p"
^^
p'
dp'
p:
= dp'\
and expand
'
in
series
about the central point
(3.20)
49
Restricting ourselves for present purposes to the first-order terms and integrating Eq. (3.19) with A/;) (terms proportional to p" obtain to we dp", respect
where the amplitude of the wave packet
is
dE
(3.22)
From this equation, it follows that B does not remain constant either in space or in time. To determine the velocity of the wave packet as a whole, that is, the group velocity corresponding to the motion of some specific amplitude, we shall use the same procedure as for the phase velocity. Let us take a certain constant value of the amplitude such that
//"
(3.23)
We
then obtain the following equation for
u:
Since E
= =M dp
<*x
dt
(3.24)
= c Y P* + mfc
fc> r
a free particle, and

dE ft
c-p =*--
(3-24a)
we
find
(3.24b)
which shows that the group velocity

is
exactly equal to the velocity
it of the wave packet as a whole of the particle itself.
From
the expression for the amplitude B(^)

/
= A-~-
of the
wave
see
packet at time
(when the amplitude corresponds to
,-
*;
Fig. 3.7), it is readily seen that the maximum value of this A lies at the point * 0. At all other points coramplitude B(Q) to the relative maxima the amplitude is smaller. In responding we have particular, considering different values of the argument
50
At the point
at
t
TC,
the amplitude vanishes ((rc)
= 0).
the
wave packet may be regarded as localized
Consequently, in the region
of the first
maximum,
that is, in the region A?
=^ A*~it.
Hence,
we
obtain the relation
^h,
which
is
(3.25)
known as
the uncertainty principle.
Group velocity
-1
Fig. 3.7.
Form
of the
wave packet
at
0.
Since the center of mass of the Wave packet (that is, its principal travels with the velocity of the particle (u v), the wave packet describes the localization of the particle. In particular, it follows from other hypotheses that have been made that the position of the particle is characterized by the square modulus of the amplitude of the x wave, namely,
maximum)
Consequently, the quantities A* and Apmay be regarded as a measure of the accuracy with which it is possible to compute the momentum and position of a particle in space by means of the
wave theory. It was also necessary
to determine whether waves could be identified with the structure of particles, or whether these waves characterize only their motion. The first interpretation of the
<]>
relationship between a particle and its associated wave was proposed by Schrodinger. In terms of his hypothesis, a particle is a wavelike formation and the density of its "smearing out" over space is given by
<!>*<)>.

In theory, a group of
51
waves can always be used
to
form a wave
packet whose size
of the order of the radius of any given particle (for example, an electron). This representation of the particle is, however, unstable. Indeed, as follows from Eq. (3.17), the phase velocity of each of the monochromatic waves contained in the wave packet depends on the corresponding wave number, or momentum p. Accordingly, each of these monochromatic waves propagates with its own phase 6
is
As
a result, the wave packet gradually spreads out with time. The "spreading" time is determined by the time interval in which the in (3.21) becomes cominitially disregarded part of the phase of mensurate with *. According to (3.20), the neglected part of the phase is proportional top'* ~(Ap)*. Thus, from (3. 20) we can obtain a measure of the time A/ which elapses between the initial formation of the wave packet and its distortion:
<|>
Or, the time interval starting with the formation of the wave packet and ending when the distortion of the latter can no longer be neglected is
A*
--**
{3-26)
Let us replace Ap by its value from (3.25). Then, in the nonrelativistic case (p<m c), Eq. (3.24a) yields
dp
=m
jL
m
'
dp*
Thus we obtain for the "spreading" time of the wave packet

(3.27)
It can be seen from this that a stable wave packet can be formed only for a particle which has zero rest mass (m = 0) and which is moving in a vacuum (as an example of such a particle, we may take the photon). It is only in this case that the phase velocities of all the components of the wave package are the same regardless of the wave number.
52
For a particle with w,,^lg and A*~o.l cm, the "spreading" time is A/~10 28 sec, and thus the wave packet does not actually 27 spread. In the case of electrons, however, m ^lCT g and 1J Ax~io~ cm. Therefore, the wave packet of an electron begins 8 to spread out after kt ^10 sec, that is, almost instantaneously. Thus, the electron is not a stable formation in the Schrbdinger theory of the "smeared" electron. This is in obvious contradiction with the experimental facts. Moreover, it becomes impossible to explain the phenomenon of diffraction if a monochromatiq wave, which provides an appropriate description of the motion of several electrons, is replaced by a set of several wave packets. At present, another approach has been adopted, namely, Max Horn's statistical interpretation of the wave function (the quantity as the probability density, or probability of finding the
*
';>
'!>*<);)
electron at various points in space. The statistical interpretation is not concerned with the structure of the electron, and the electron can remain a point charge (or, to be more precise, a charge whose radius rQ does not exceed 10 13 cm). Only the probability of finding the electron at different points in space changes as the wave function changes with time, but the structure of the electron remains
completely unaffected.
standpoint of Horn's statistical interpretation, the quantum-mechanical treatment of problems involving many electrons in identical states presents certain methodological similarities with the treatment of various problems in the kinetic theory of gases on the basis of Maxwell's distribution function (it should be noted, however, that the quantum-mechanical distribution function is devoid of the temperature term). For example, in Horn's / '],*<<) interpretation, the diffraction pattern of an electron beam may be
the
From
explained in the following manner. The bright spots correspond to the maxima of the function f= ty and thus the greatest numbers of electrons travel towards bright spots. On the contrary, the probability is smallest for electron motion in the direction of dark
2
,
regions.
One attempt to explain a certain "freedom of behavior" of an individual electron is based upon the complementarity principle, a solution adopted by Bohr, Heisenberg, and others (see also Chapter 7). The complementarity principle asserts the theoretical impossibility of extending our knowledge of the microscopic world beyond a certain finite, even though small, limit of accuracy,
since our measuring apparatus necessarily must exert certain indeterminable effects on the experimental object (an electron, for example). In particular, for canonically conjugate quantities, such as the position and momentum, the degree of accuracy with which they can be measured is given by the uncertainty
principle.
Heisenberg attempted to provide a more rigorous foundation for the complementarity principle by means of the following hypothetical
53
experiment. Suppose we wish to determine the position of an electron with the aid of an "ultramicroscope," that is, an instrument that is designed for precisely this task and that utilizes a light beam of appropriate wavelength X (it is not actually possible to construct such a microscope). We shall assume that the electron is located at the vertex of a cone of revolution with angle 2?, (the objective being the base), where <f>c=.the angle between the incident (wavelength X) and scattered light beams. Any light that enters the objective after being scattered by the electron must have traveled within this cone of directions. From the laws of optics, follows that the uncertainty in the determination of any of it the electron coordinates in the plane parallel to the plane of the
objective
is
(3.28)
Moreover, since the
light
has a
momentum
(the
= ~i*
part of this
effect).
momentum
is
imparted to the electron
Compton
Ac-
cordingly, the
momentum
only be determined
of the electron in the given direction can with an uncertainty of
(3.29)
The product
relation.
of these
it
two uncertainties yields the uncertainty
seen that to determine the position of the degree of accuracy, it is necessary to use light of the smallest possible wavelength X. Equation (3.29) shows, however, that the smaller the wavelength, the greater is the momentum imparted to the electron (the position and momentum are canonically conjugate quantities). Thus, according to the complementarity principle, there would have to be two classes of experimental aparatus. One of these would be designed to measure the spatial coordinates with any desired degree of accuracy (in the case of "ultramicroscopes," instruments with X->0), the other, to
(3.28),
is
From
electron with the
maximum
measure the corresponding momenta (instruments with X-*^). Then the observer using an instrument of the first class imparts all of the indeterminable effects to the momentum, whereas the
instrument of the second class imparts all of the indeterminable effects to the position. Accordingly, in the opinion of the adherents of the Copenhagen school, we cannot at any given time ascertain both the position and the momentum of the object, although each of
these variables can be measured separately with any desired degree
of accuracy.
54

3.1
L Problem
J Show
that there is an integral
number
of
de Broglie wavelengths in an
eSfpSFetecB&lrorblt. Solution, According to Bohr, the set of stationary the two conditions
0)
elliptic orbits
may
be found from
pr dr
=n h
r
and
(p
where
Since
27*=^ to,
it
that is,
T
2 f
<f
r^
^
ft
where
n=>n ? +
we
obtain the required result
/i
ri
and
2Tdt
O)
y = n.
Problem 3.2/ Show

i
that the
3.2^
wave
L
2
is not
'
k monochromatic for finite values of /. Find the range of wave numbers A& over which the amplitudes of the individual harmonics may be regarded as nonzero. Solution. Let us represent <p (x) in the form of a Fourier integral
00
kQ
<p
(x) =
A (k)
e ittx
dk
where the amplitude s A
(k)
are given by the formula

4-00
4(
=i
(x),
- 00
i(x)r**dx.
Substituting the expression for
<p
we
find
sin
(k
k )t
Hence
it follows that the largest amplitude k is A (* ) corresponding to k //2. Although all the other amplitudes are not equal to zero, in practice only those amplitudes which are of the same order as A (k 9 ) need be regarded as nonzero. These amplitudes
lie within the
range A*
~ -y
The quantity
Aft
2// may be regarded as
the

"spread"
and
/
55
of the
wave vector due
to the finite
width of the wave packet. Setting bk
= -?71
~ Ax, we again obtain the uncertainty relation: ^

A/? AJC
h.
Problem
3. 3.
\Find the
mean
velocity and the "spreading*' time of the

00
wave packet
<p
(x
\ oo
A (k) <?-
'
M-*
v)
dk
(3.30)
if the
amplitude
A (k) is
in the
form
of a
Gaussian curve:
Solution.
f or *
Substituting the expression for 4(fc) into Eq. (3.30) and integrating,
we
find
2
It
follows that the particle is initially localized in the region A.\r~
.This in effect is
equivalent to the uncertainty relation (3.25). To obtain the shape of the wave packet at any other instant of time, the equation
we take
into account
In the nonrelativistic
approximation
(k
<m
c/ft) t
we
then have
Substituting this expression into Eq. (3.30),
we
obtain, after integration
where
Hence we have for
the probability density

? a (x
where t>
it is
is the velocity of the
center of
mass
of the
wave packet. From
this
formula
with the
/Wo
seen that the
maximum
of the
wave packet,
that is, the point*
=v
t,
moves
velocity Vo of the particle.
56
The

effective width of the
wave packet
at
time
is
found to be
If
we
substitute q
into this formula, the "spreading" time of the
wave packet
is
expressed by a relationship which coincides with Eq. (3.27).
Problem 3.4.|Show
-"-"^
that a
damped
oscillatory motion
x (t)
= e-V cos
u>
results in broadening of the spectral line. Solution. The damped oscillation can be represented in the
form
of a Fourier integral
x (t)
where
CO
a (w) cos
<
rfo>,
r
\
(ro)
.v
7t
(t)
cos
(At
dt
i 7t
J
f J
TTT
I
"
V**
"*0/
For the largest amplitude, we obtain
The amplitudes with the same order of magnitude correspond to frequencies lying in the range
It is this equation which represents the line broadening since, in effect, only the amplitudes corresponding to these frequencies differ from zero* Since the damping coefficient 7 is connected with the mean life of the damped oscillations by the relation A 1/7, the above equation yields the familiar optical relationship connecting the broadening of the spectral lines and the mean life of the atom:
Hence, since Aw
A//i,we obtain what
is
known as the fourth uncertainty relation

h.
Chapter 4
The Time-Independent Schrodinger
Wave Equation
Planck's quantum theory, Bohr's postulates, and de Broglie's hypothesis represented very important steps in the development of the theoretical foundations of atomic physics. However, they were overshadowed by the discovery of a fundamental differential equation describing the electron and accounting for its wave properties, and the construction of a theory accounting for the quantum nature of radiation. The crucial move in this connection was made by Schrodinger in 1926, when he proposed a partial differential equation that turned out to be generally applicable to the motion of charged particles in the nonrelativistic case (v^c). This equation
represented
equation to
generalization of the classical Hamilton-Jacobi in which the de Broglie wavelength differs from zero. The Schrodinger equation stands in approximately the same relation to the Hamilton-Jacobi equation as does wave optics to
a
cases
geometrical optics.
A.
DERIVATION OF THE TIME-INDEPENDENT SCHRODINGER EQUATION

show how the Schrodinger equation can be obtained We must insist that there can be no question of a
We
shall
most simply.
rigorous or general derivation of this equation, since it is not, in general, possible to set up a new theory entirely on the basis of old postulates. We shall adopt a mode of presentation which consists essentially of a reasonable generalization of the wave equation from classical electrodynamics or optics
to the
case of de Broglie waves. Here ? is a wave disturbance propagating with velocity u.

chromatic, a solution to Eq.
(4.1)
may
function describing a If the wave is monobe sought in the form

(4.2)
58
the
where u> 2o is the angular frequency, and the spatial part wave function satisfies the equation
<l>(r)of
In this equation,
we can use a
u,
single parameter in place of the two
parameters
u>
and
namely, the wavelength
Xr=
?-
(4.4)
We
then have
^T*(r)-0.
(4.5)
From
ing the
this general equation, we can obtain a wave equation describwave motion of electrons by substituting the de Broglie
wavelength
X
= A == ?55.
' (4.6) x
Using the law of conservation of energy

fm
~^~
aas
^ !SB const
we have
-V(r)].
(4.7)
Substituting this expression into (4.5), (or stationary) Schrodinger equation
we obtain the time -independent
V^(/-)+^[-K(r)]H/-) = 0.
<|>
(4.8)
tion
Once we have found the space -dependent part (r) of the wave funcfrom (4.8), we can use Eq. (4.2), which is valid for all monochromatic waves, to obtain the complete wave function, which
>
depends on both (spatial and time) coordinates. Substituting
-^-,
we have
From now
on,
we
shall write the
time in the form l^(0 be denoted as (//.
wh 1 * 6
t^ e
wave functions which depend on both position and wave functions whose only argument is the position will
THE TIME-INDEPENDENT SCHRODINGER WAVE EQUATION
59
For the complex conjugate

part also satisfies Eq.
(4.8),
function,
whose space-dependent
we have
/
(4 .9a)
B.ÊSTRICTIONS ON THE WAVE FUNCTIONS. EIGENVALUES AND EIG EN FUNCTIONS

The functions (t) which describe the behavior of a particle may be statistically interpreted by means of the Schrodinger theory. In particular, the quantity fy*(t)ty(t) ty*ty, which plays the role of a distribution function, represents the probability density, or probability of finding the particle at any particular region in space. If the probability density differs from zero only in some arbitrarily large, but finite, region of space Q it is accurate enough to say that the particle is located somewhere in this region. In other is words, the probability of detecting the particle in the region unity. Mathematically, this can be expressed in the form
<|>
(4.10)
In
quantum mechanics, relationship
(4.10) is called the
normaliza-
tion condition.
It is important to note that the region of nonzero probability density is not always finite. There are cases where <]>*<)> does not go to zero over all of space (the simplest of these cases is that of the motion of a free particle, which we shall consider below). When this happens, $ty*tyd*x diverges, and a somewhat different formulation of the normalization condition must be given. We shall now give a general analysis of the Schrodinger wave equation. The Schrodinger equation (4.8) is a second-order partial differential equation. Its solution resembles the solution of certain classical problems of mathematical physics, such as the problem of a vibrating string. We note, first, that certain conditions since it is a solution must be imposed on the wave function satisfying a second-order Sturm-Liouville equation. It must be continuous and have a continuous derivative. Moreover, it must be single valued and finite over all of space. Finally, it must satisfy certain boundary conditions. In general, solutions wjnch^satisfjf these requirements do not exist for all values of the parameters, but only for certain specific values, which are known as eigenvalues. In the case of the Schrodinger wave equation, the energy E is a parameter of this sort, its eigenvalues being
<|>
1,
Ea
....
60

the
The solutions of
values
wave equation corresponding

*i,
to these eigen-
fc,
*s,
...
(4.12)
are said to be eigenfunctions. The possible values of the energy form the energy spectrum. We shall see below that, if the motion of a particle is not bounded, its energy spectrum is continuous. If, however, the position of the particle in space is bounded, the energy spectrum is discrete. Let us show that the eigenfunction fy n satisfy the orthonormality condition
lWn<Px = *
where <W
to
tt
n;
(4.13)
is the Kronecker-Weierstrass symbol, which is equal unity for n'~n (the normalization condition), and to zero for n (the orthogonality condition). To prove Eq. (4,13), we write ti the Schrodinger equations for ty n and <|)S:
0,
(4.8a)
(4.
+ -%? (E^ - V) $ =
0.
8b)
If we multiply the first of these equations by and the second by (~~tn)> and then add the two resulting equations, we obtain
<|*J' f
^V'tn
Since
-*
nV
V + ?-
(E n
- E ^n =
n .)
0.
(4.
14)
$'V*y*
W^*=V-B,
^vi*.
where
fl=Wv*
integration of Eq. (4.14) over all of space yields

J
V * dJC
+^
- E^)
J W'tnd'Jf
=
obtain 2
(4.15)
Since the
<|>
function tends to zero at infinity,
we
The surface S - 4nr 2 tends to infinity at r Therefore, the integral goes to zero at the wave function l/J tends to zero more rapidly than r" This condition is always satisfied for a discrete spectrum since the wave function, as a rule, approaches exponentially zero at infinity. The case of the continuous spectrum will be considered separately.
2
>
>.
->
no if

Therefore, instead of Eq. (4.15), we have
d*x
61
*,ty n
0.
(4.16)
We
shall
now assume
that
En
-/-E n
'
(that is,
n'-n). According
to
Eq. (4.16), the following equation then holds (orthogonality condition):

i
([^'tjjflf/'.vszsO
(4.17)
n (or E n E n >), this integral does not go to zero If, however, n' and we may impose the requirement that it be equal to unity (the normalization condition):
v=l.
<J>i,
'[>
(4.18)
Thus the eigenfunctions which correspond to the eigenH ty* and values E lt E andEa, do indeed possess the property of orthonormality (4.13). This is one of the most important characteristics of eigen,
functions.
C.
A PARTICLE
IN
A POTENTIAL
WELL
As an example of the calculation of eigenvalues and eigenfuncwe shall consider the motion of a particle in a potential well. Since the chief interest of this problem is simply that it provides an illustration of methods used in the solution of this example, we may assume a very simple dependence of the potential energy on
tions,
distance (see Fig. 4.1):

Q
for
for
:o<\v<0
(region
I)
V IV
for
< x < co (region

=
<></
(region
II)
(4.19)
III)
In the potential well (region
II),
where E>l/=0,the Schrodinger
equation takes the form

6n
2
|-
<fn
0,
(4.20)
where
and
(4.20a)
62
We
has no physical meaning in this problem. note that the case Since the general solution of Eq. (4. 20) is oscillatory, we have
<]>
<0
n
= BU cos kx + A
ii
sin kx.
(4. 21)
In regions
and
III,
the Schrodinger equation has the
form
(4.22)
Here two cases must be distinguished. In the first case V,,), the solution for these regions is also oscillatory in character (an equation of the elliptic type). It is given by Eq. (4.21), the value of k being
(>
restrictions need to be imposed on the wave functions at infinity. Therefore, the energy E can assume any value in a continuous spectrum of energies. It is better, however, not to investigate the case of a continuous spectrum on the basis of this example, but rather on the basis of the motion of a free particle (see below). The potential well only adds to the mathematical difficulties of the problem, without changing the general character of the solution.
No
Vfjc)
Fig. 4.1.
The motion
of a particle in a potential well.
the second case, namely, the case of a potential barrier ), the solution of Eq. (4.22) is exponential in character (an equation of the hyperbolic type). The general solution can be written in the form
In
(<V
<K
=4
1.
me-*
(4.23)
where
(4.24)
63
If the energy can assume any value without restriction, the wave function inside the potential barrier ) will contain both
(0<<l/
an exponentially increasing part and an exponentially decreasing part (see Fig. 4.2). Therefore, we must choose only those values of E for which exponentially increasing solutions do not exist inside
Exponentially increasing
solution
Exponentially decreasing
solution
I
value of E.
Fig. 4.2.
Wave function
is
for a given
The energy level
taken to be the abscissa of
the
wave function
the potential barrier.

#i
in region
We
Accordingly, we require that the coefficient I(#<0),andthecoefficient j4iu=0in region III
then have
(4.25)
where, for the sake of simplicity, we have made
B in
=Be.
4 0), joining the solutions at the boundary of regions I and II (x and also at the boundary of regions Hand III (# /), and making use of the requirement that the exponentially increasing solution vanish, we obtain the equation for the eigenvalues of the energy E. We shall now further simplify our problem by requiring that
By
V Q9 together with
(4.25) that
*,
<j>i
Eq. for the solution (4.21) inside the potential well (region
<p in
go =
to infinity (see Fig. 4.3).
==
0,
It is apparent from and therefore the boundary conditions

II)
take the
form
for x
and
for
jc
= =
(4.26)
(4.27)
/.
3 the number of unknown coefficients in the wave function is smaller When E < V than the number of imposed conditions. Accordingly, solutions are possible only for certain values of E and a discrete spectrum is obtained.
,
4
first
In joining the solutions, we must match the actual derivatives) at the appropriate point.
wave functions (and also
their
64
Applying Eq. (4.26) and (4.27) to the general solution (4.21) in 0, and the eigenvalue are described by region II, we find that Bu the equation
sin/e/
(4.28)
from which
(4.29)
n from further 1, 2, 3, 4, .... We exclude the value n considerations, since the wave function in this case is identically equal to zero. It is not necessary to consider separately the negative values of //, since the wave functions for negative n are equal
where
^=00
Fig. 4.3. Particle in a potential well with infinitely high walls.
to the
wave functions
l
for positive n, taken with the opposite sign.
Since k
= -j~ E,
we
obtain the following equation for the energy
spectrum
(the eigenvalues):
(4.30)
The wave functions corresponding to these values of energy (eigenfunctions) are

(4.31)
The
coefficient
A n can be found from

/
/
the normalization condition
^dx= An
sin'
vn ~ dx
-^
/!?,=
!,
65
which gives
Substituting the expression for
An
into Eq. (4.31),
we
finally obtain
(4.32)
the general theorem of eigenfunctions [see Eq. Schrodinger equation satisfy the orthogonality condition
According
to
(4.17)] the eigenfunctions (4.32) of the
for
n'^n,
(4.33)
as can be readily seen by performing the direct integration after substituting Eq. (4.32) for ty n We shall now write down a few specific eigenvalues E n and
.
eigenfunctions
<!>,
shown
r L
i
in Fig. 4.3:
-i
. .
,
= ~IiiF> ti=]/7 T' = 4 fc=|/~4sin2*-y, = 9 fc= 3*.

sin
lf
lf
(4.34)
(4.35)
(4 ' 36)
sin
These solutions are very similar

solutions
for
to the familiar standing-wave a vibrating string with fixed ends. The case n 1 2 [see (4.34)] corresponds to the fundamental mode, the case n [see (4.35)], to the first harmonic, etc.
= =
D. THE MOTION OF FREE PARTICLES. NORMALIZATION OF WAVE FUNCTIONS IN THE CASE OF A CONTINUOUS SPECTRUM
We shall consider the motion of a free particle (taking the onedimensional case first), when the Schrodinger equation (4.8) has the same form in all of space ( OD<.V<[OO):
(4.37)
66
where
k*
= ^=.
*'**
(4.38)
The solution of
this equation is
<[
=4
+ &?
ikx
.
(4.39)
To determine the physical meaning of each term in Eq. (4.39), write the complete, time-dependent function
y(t)
It
we
e' mt
y=Ae-'
'-** }
+ Be-*
(m '+**>.
(4.40)
describes the motion of the particle in the positive direction of the # axis, and the second term /fe-ii'+**) t h e motion in the negative direction. If we restrict ourselves to the traveling wave which is propagated in the direction of
is
f
seen that the first term Ae~ i(wt
^ hx)
positive
x,
we have
*t
= Ae
ik
(4.41)
It
is
readily seen that the integral
Y
\
ty*tydx
diverges. Con-
00
we must revise the method of normalization [see Eq. There are two basic methods of normalizing ^functions of this sort. We shall devote most of our attention to one of them, which was proposed by Born. Thus, rather than imposing a boundary condition, we shall subject the wave function to a periodicity
sequently,
(4.10)].
condition
(*)
= * (* +
*-).
(4-42)
where
the length L is said to be the period. It can be made arbitrarily large. As a rule, L does not appear in the final result. Rewriting (4.42) in the form
we
obtain
e ikL
1 ,
which implies
*=?.
where /i=0,
db 1, levels are energy
it
2,
(4.43)
3 .....
From
Eqs. (4.38) and (4.43), the

Since the function condition becomes
<]>
67
is
periodic in the interval L, the normalization

1/2
f -Z/2 -Zja
ty*tydx=
I.
(4.45)
Substituting the expression (4.41) for
<[>,
we have
<
= 4=.
4 46 )
-
Therefore, the normalized solutions are
(4.47)
The
direct
integration shows
that the
functions
(4.47)
are
orthonormal. As can be readily shown by direct integration:

1/2 L/2
-1/2
-T -1/2
\
[l
"'</* =
for
for
n'Âi,
'
4.48)
n.
Therefore, introducing the artificially defined period, we can transform the continuous energy spectrum into a discrete one. If, however, the length L is allowed to go to infinity, the discrete
spectrum again becomes continuous. Since k=^the energy difference between two neighboring levels
A
A r*
=~
r
and A/i=l
is
n"K
2.TI
T = v ~r*
2,TtTl
A c\\
(4.49)
It
immediately follows that A
if
oo; that
is
the energy
spectrum becomes continuous. We shall now generalize the free particle problem to the threedimensional case. The Schrodinger equation can be written in the form
Just as
one-dimensional case, the quantity k 2 is given by shall solve Eq. (4.50) by the method of separation Eq. (4.20a). of variables; that is, we assume a solution of the form:
in
the
We
68
Substituting this expression into (4.50) and multiplying by

1
we
find
<;,"(*)
(.
)'
__ + TOO + r fW + k - n
,
<!/"<
(2)
'
(4 ' 52)
where the primes denote the derivatives of the corresponding^

function with respect to
its
if
this type is satisfied only
argument. We note that an equation of each term (fraction) is independent of
the coordinates and is equal to a certain constant. We thus obtain the following equations for the functions <[>(#) ty(y) and ^(2):
<!/'
(x)
+ wW=
o.
r (y) + W (y) = u> ô
<
4 53 )
-
where
(4.54)
Taking a traveling wave propagating in some specific direction as a solution for each of Eqs. (4.53), we get
^(x)
= Ae ^ x
if
ty(y)
= Be
lll
*v
t
lh ^( z )=Ce ^.
(4.55)
The unknown coefficients A B and C are determined from the normalization condition, assuming that the functions (A;), (y) and
>
<!>
ty
<p(z)
are periodic in the interval

i
L.
We
thus obtain
i
(4.56)
where
and
n "i
n n h* **j
u,
-4--1-
*>
"I-
9 ^
-4- 1 -2w,
...
<
In this case, the particle
has energies of
/HO/.
a
/fioZ,
'
(4,58)
where
(4.59)

Substituting the functions (4.56) into Eq. (4.51),
69
we have
^ = L~2 e i.r
3^
(4.60)
These wave functions
will satisfy the orthonormality condition
in the
The complete, or time-dependent, wave function can now be written form

-r>,
(4.62)
where
p = ftk, E = J
(4.63)
note that the energy spectrum is again continuous, just as case of one-dimensional motion. This can readily be shown in the same way as before by finding the energy difference between two adjacent levels and then letting L go to infinity.
in the
We
E. FUNDAMENTAL PROPERTIES OF 6 FUNCTIONS. 6-FUNCTION NORMALIZATION OF THE WAVE FUNCTION IN THE CASE OF A CONTINUOUS SPECTRUM
ol the fundamental properties of elgenfunctions is orthonormality [see Eq. (4,13)]. An arbitrary function /-' (.v) with no "special" properties can thus be expanded in a series of orthonormal functions belonging to a
As we have already mentionpdfone

set:
>i
their
complete
<M*)
Multiplying obtain
this
(4.64)
equation by
fy*, (.v)
dx and
integrating
over the entire range of X9 we
ft.
(.v)
F (x) dx =
"
Cn
'
4,*. (
X)
<,
(A-)
dx.
Because of the orthonormality condition, we have
^
(4 .6S)
(.v)/-'(-v)^.
order to bring ty n (x) under the integral sign, we change here the variable of integration .v to x' and substitute the expression for Cn into the expression (4.64):
In
x'F (.V) * (*)
}-
(.v).
(4>66)
70
This expansion is a generalized Fourier series. We can obtain the ordinary Fourier series from it by substituting the harmonic functions (4.47) for fy n (x) . We note that the sum over n in Eq. (4.66) is a & function:
(4.67)
transforms the function F(x') into F(x\ resembles the Kronecker-Weierstrass possesses a similar property with respect to the subscripts n and /i
it
since
In this respect, this & function
&
symbol, which
/,fn*nn' =/n.
x9 Just It follows that the & function differs from zero only in a very narrow region x' as the 8 symbol differs from zero only for n I in (4.66), we obtain n. By setting F(JC') one of the fundamental properties of the & function:
1
&
(jt
*')/*'
1.
(4.68)
is interesting to note that the relations (4.66) and (4.68) are completely independent of the particular set of orthonormal functions used to construct the function. In the simplest the & function can be constructed from the case, f orthonormal functions (4.47), which are used in (X.} ( 0(X-Xt expansion into a Fourier series:
It
ft
(4.69)
Introducing a new variable
= -- and
2it/i
making
(since
use
of
the
relation
Aft
An
(4.69) to
Aft*'*
(*
An
Fig. 4.4.
1),
we can transform
j
/
Graph of the 'smeared* 5 function.

let
~\ s= _J_
2rc
*'!
If
we now
L go
to infinity (A&
0),
the last
sum
is
changed into an integral:
&
(x
x')
= -A
1 f dke *
(x
(4.70)
co to The range of integration over k should be taken from oo. We note that the & function belongs to the class of functions that are known as "improper functions." This is manifested in the impossibility of direct evaluation of the integral (4.70), since it diverges at x jc'. Therefore, if we wish to obtain a representation of the & function, we have to "smear out" somewhat the integrand of (4.70). For instance, the 5 function may be "smeared out" as follows:
KS.
(4.71)
A:')'
A graph of the smeared 5 function must have the following properties:
is
given in Fig. 4.4. At the limit a-*
0,
the function

8
71
(x
x') x')
*(x
= co for = for
x = x\
x^x\
ft
(4.72)
As we approach
unchanged:
the limit a
*>
0,
the area between the
function and the
axis remains
(,-^-lta I J d*'
6
ai
+ *_ x)
(
=1-
(4.73)
The three-dimensional
function can be similarly defined as
(r)
6 (x) & (y) 8 (*)
J-
d'*e'*
r
.
(4.74)
It
satisfies the following conditions:
8(r)
if
r^-0,
=l.
(4.75)
The integrals (4.74) and (4.75) are three-dimensional. The former extends over the entire wave-number space (k\ and the latter, over the entire coordinate space (r). It is very convenient to express the density of a point charge with the help of a threedimensional & function:
P(r)
= *5(r).
*
(4.76)
Substituting this expression into Poisson's equation
V*
& = - 4^5 (r) =
2
*
f dke
ikr
,
we can determine
the potential of a point charge
Thus, the three-dimensional

r -*
6 function enables us to describe the singularity at the point of the Laplacian operator applied to 1/r:
In determining the potential $, function by the Laplacian operator:
we made use
of the rule for dividing
an exponential
*r
_ f__
a lkr
The
validity of this rule is immediately obvious, since
v 2 (v 2 rv*'
v2
tkr
72

We
use
shall not discuss the properties of the & function any further and shall proceed to to normalize the continuous spectrum for one-dimensional free motion. From (4.41), the wave functions for this case are of the form
it
<|,* (/>')
/*<?
(4.79)
The normalization
of these functions is related to the 6 function in the following way:
'x
= A*
+?
\
dxe
5
('-4) " = vn
(p
p').
GO
= A^nhb (p
From
this,
/>')
(4.80)
we obtain
___
1
Therefore,
Let us compare the ordinary normalization of wave functions with the 8 function normalization. We can write the orthonormality condition for functions normalized by the ordinary method as follows:
"a
^
\ 7^
'
n"-
//=
J 1 * if
if V), v
/:
/i
n, lles inside the range rii lies outside the range /li-îi
<
Similarly, replacing the sum in Eq. (4.82) by an integral, we can obtain the following generalized form for wave functions which are normalized with the help of a & function:
' J
I*
dp* ^
(
J
O. if
/>
inside the
lies outside the
J
Pi
range p L <zp t
Analyzing the solutions that we have obtained for a particle in a potential well and for a free particle, we arrive to the following conclusions. If the condition V ^> K is satisfied at all points of space at infinity, the energy spectrum is discrete. If, on the other hand, there are regions at infinity in which V <. /.', the energy spectrum is continuous. While this conclusion was obtained on the basis of solutions for a rectangular potential well, it is quite general as long as the potential energy is a continuous function of .v, y and z.
Chapter 5
The Time-Dependent Schrodinger
Wave Equation
It has been shown that the solution of the time- independent or stationary Schrbdinger wave equation (4.8) amounts to a determination of eigenvalues E n (the spectrum of energy levels) and eigenfunctions. The time-dependent wave function ty n (t) of a given state E was found by multiplying ty n by e~ (ilh} n' The function ^,,(0 obtained
9
way describes only strictly monochromatic processes (that is, with only one value of energy). There is, however, a more general form of the Schrodinger equation, which depends explicitly on time and may be used in a much larger class of problems. This equation is known as the time-dependent Schrodinger wave equation.
in this
A.^TRANSITION TO THE TIME-DEPENDENT SCHRODINGER
WAVE EQUATION
To
s*
obtain the time-dependent Schrodinger wave equation, it is necessary to eliminate the energy E from the time- independent equation, where it appears as a constant parameter. The timeindependent Schrodinger wave equation (4.8) can be written in the
form
The relation
that is used to eliminate
E from
this equation is
(5.2)
is
_-fiM=HO.
Accordingly, the time-dependent Schrbdinger wave equation
() -
<
5 - 3>
This equation is more general than the previous one. In particular, it can be used for the description of processes in which the potential energy V is a function of both position and time.
74
If the potential energy V does not depend on time, it is necessary to solve only the time- independent Schrodinger equation by finding
all
the possible energy eigenvalues E n and the corresponding The wave function which satisfies Eq. (5.3) is eigenfunctions n related to these partial solutions by the linear equation
<j>
/-^V^.
(5 .4)
To prove this, we substitute (5.4) into (5.3), remembering that C n are constant coefficients and that ty n satisfies the equation
It
is
then readily seen that <KO
is
a general solution of (5.3), since
(56)
The case
of a monochromatic wave is a special case of the general solution (5.4). The appropriate wave function can be obtained from (5.4) by setting
rto
and Ca
(if
).
it, transition from the time- independent time -dependent equation (5.3) is essentially equation (5.1) equivalent to a simple replacement of the energy E by the expres-
As we have presented
to the
sion
ih -5-.
In
quantum mechanics,
1
this
expression
is
known as the
energy operator
= M jp
(5.7)
The effect of this operator on any function amounts to an ordinary differentiation of the function with respect to time. Thus, E is a linear differential operator. In the case of a monochromatic wave, for which
we have
Operators will be denoted by roman characters.
THE TIME-DEPENDENT SCHRODINGER WAVE EQUATION
75
Hence, it is seen that the energy E n is an eigenvalue of the energy operator E. There are other operators besides the energy operator in quantum mechanics. The most important of these is the momentum operator
its name from the fact that in the case of a free particle, eigenvalue is identical to classical momentum. Indeed, operating with p on the wave function of a free particle [see Eq. (4.62)], we have
which takes
its
Thus,
in this particular case, the
eigenvalue of the operator p
is
the classical
momentum/?.
With this operator notation, which helps to bring out more clearly the relationship between the quantum-mechanical and classical laws of motion, the Schrodinger equation (5.3) has the form
(5 ' 9)
to
Thus, to carry out the formal transition from the classical theory quantum mechanics, it is necessary to replace the energy E and momentum p in the classical equation for the law of conservation
of energy 2
by the corresponding operators and to operate with them on the
wave
function.
T,
The operator J^-
is
known as the
kinetic -energy
operator
andT-f-K as the Hamiltonian operator H. For the sake

shall call the latter the Hamiltonian.
of brevity,
we
If
the electron is in both an electric field and a magnetic field, which is characterized
.by the vector potential A, then, using the classical expression (1.15) for the Hamiltonian, the time-independent Schrodinger equation can be written in the form
E
f
(5.9a)
76
Using this operator notation, Eq.
(5.9)
can be rewritten in the
form
(E-H)'MO = 0.
For the time- independent potential obtained:
V,
(5.10)
the following relation
is
and, therefore, the time-independent Schrbdinger equation reduces

to the
form
(/-
1). !>
().
(5.12)
is apparent from this that, It stationary problems, the eigenvalues of the Hamiltonian are equal to the eigenvalues of the energy, just as, in the classical case, a Hamiltonian function which does not depend explicitly on time is equal to the energy of the system.
B.
CHARGE DENSITY AND CURRENT DENSITY. QUANTUM ENSEMBLES

is
In classical electrodynamics, an important role the equation of continuity
played by
J.
+ V.y=0,
(5.13)
which involves the charge density
p and the cur rent density j. This equation basically represents a general form of the law of conservation of charge. To show this, we multiply (5.13) by d\v and
integrate over all space
-0.
(5.14)
Reversing the order of differentiation and integration in the first integral (this is allowed since time here is only a parameter), and changing the second from a volume to a surface integral, we obtain
-ijV*+fMS=0.
If
(5.15)
there are no charges or currents at infinity, the surface integral vanishes, and we obtain the law of conservation of charge
:
= const.
(5.16)
77
We shall now use the wave theory to find an expression for the charge and current densities. For this purpose, we take the Schrodinger equation (5.3), writing it in the form
a
Similarly, for the complex conjugate equation,
j
(5.17)
we have
(5.18)
or 't-^vV(0-' I
(0=0.
Multiplying Eq. (5.17) by ty*(t), and Eq. (5.18) by ^ (0, and adding the two equations, we obtain
^.'JOiiO
_|_
(/)
vr
(0
(/)
,,,
(/)
_
j
o.
(5>19)
(5.19) with the equation of continuity (5. 13), and considering that the charge density is equal to the charge c of one particle multiplied by the number of particles per unit volume (that is, in this case the probability density), we have
Comparing
(5-20)
From
(5.19)
and
(5.20),
we
find the current density
(5.21)
It
should be noted that for a monochromatic wave
both the charge density
P^r*
and the current density
(5.22)
^=4S-<* V **--** V
*>
(5 ' 23)
are independent of time. In the case of real wave functions (]>* ^), the current density is always identically equal to zero. For instance, the current density of an electron in an infinitely deep, one-dimensional potential
78
(see Fig. 4.3) is zero (/=0). This is quite natural, since oscillations described by real wave functions are actually standing waves, and standing waves cannot give rise to a particle flux. The case of motion of a free particle is somewhat different. According to (4.60), the wave function for this case describes a
well
traveling wave:

(5.24)
into (5.22) and (5.23), we obtain Substituting this expression for the following expressions for the charge density and the current density, respectively:
Hence, when the charge is distributed with uniform probability over the entire volume, its density is equal to the charge divided by the volume, as was to be expected. Moreover, the relationship between the current density and the charge density remains the same as in classical electrodynamics. Let us now explain the physical meaning of the coefficients Cn which appear in the solution^ (Oof the time-dependent wave equation [see (5.4)]. For this purpose, we substitute the expression for ty(t) into Eq. (5.16), which serves to define the conservation of total charge. Then, using (5.20), we obtain
Since the wave functions must satisfy the orthonormality condition

J
^W *==
3
ll
8 n'.
we have
c*i' 2c;c =2i n n
=
(5>25)
We can now give a physical interpretation to the coefficients C n as quantities characterizing the probability that a particle exists in the quantum state n. Indeed, if we know with certainty that the E no ), all of the coefficients C n except particle is in the state /? (E t nn<>. If C,, , will have to be equal to zero; that is, we must set C n the particle has a nonzero probability of occurring in two or more states, then, accordingly, two or more coefficients will differ from
79
zero. The actual probability that a particular state is occupied is 2 given by JCJ , while the probability density of the distribution of the state over entire space is given by ty n \*. When there is a large number N of particles in a region of space, we have, instead of (5.25),
I
Here the coefficients Cn characterize the distribution of the total number of particles among the different quantum states. In this connection, we shall introduce the concept of quantum ensembles: namely, a collection of identical quantum states that are described by the same wave function fy. A quantum ensemble
can be used to describe an electron beam, photon flux, and so on. 3 Two different cases are possible. f N 1) All of the particles are in the same quantum state Ho.'lC/iJ' and iC w for n / n 2) The particles have a definite distribution among the various
2
{}
quantum
states:
1^,1*
= ^,
\C nj
\*
=N
tt2 ,
\CHt
\*
= Nn
,...
(5.26)
where
= N.
In the second case, the motions of particles, say, in the states// /*, and n n 29 cannot be considered independently of one another, because whenever two states ni and'f n are both possible, the total wave function is a superposition of the individual states, namely,
ty
This is important in determining the total probability of a state, which is proportional to the product ^* (O'MO- m addition to terms such as W,(0^.(0 this product will contain mixed terms of the form W, (O i^j(0. In other words, each particle will possess a definite probability of being in both quantum states. The mixed terms differ from zero for coherent y waves (a pure ensemble), and this leads to interference of the de Broglie waves, which does not happen with incoherent waves (a mixed ensemble). Thus, in a pure ensemble, waves are added, whereas in a mixed ensemble, intensities are added. Quantum ensembles are useful with regard
f
See D. 1964
Blokhmtscv, Principles of Quantum Mechcuiics (trans
),
Allyn and Bacon,
80
to the statistical interpretation of the results of
wave theory. For a large number of particles or a large number of states occupied by single particles, it follows from the law of large numbers that the probability of some particular process, when calculated on the quantum-mechanical basis, should agree with the distribution that
can be observed
in experiment (see Chapter 7). Similarly, in classical statistical physics, the probability distribution agrees with experiment independently of the nature of the 'hidden' parameters.
* '
C V CONNECTION BETWEEN THE SCHRODINGER THEORY />AND THE CLASSICAL HA MILTON- JACOBI EQUATION
The Hamilton-Jacob! equation, which is used in classical mechanics to describe the motion of a material particle in a field of forcf is a first-.ozder nonlinear differential equation. We shall briefly Yecapitulate its derivation, sta>tirig~from the classical law of conservation of energy
,
E=
f--\-V 2Mo
'
= T-\r V.
'
' (5.27) *
Let us introduce the action function

t
t
S(t)= where
\Xdt=
\(2T
E)dt
=S
Et
(5.28)
S=$2Tdt.
X
(5.29)
t in (5.29) is only imand therefore we have written tj(t), z(t)) plicit, / it without including as an argument. Thus, we can distinguish between the stationary action function S and the time-dependent action function S (/) in the same way as we distinguish between the stationary wave function and the time-dependent wave function. To show that 5 is not an explicit function of / in (5.29), we take
As we
shall show, the dependence of S and
namely, S
= S(v(/),
the total differential of 5:
dS
= 2Tdt.
d,,+
From
the equations
d,.
-j-
(5-30)
2Tdt = m
(.' -1-
,/*
+ P) dt = pjx + iyly
pt dz,

'
81
wefindthat
and thus 5 does not depend explicitly on
t.
Moreover,
(5.31)
p=VS.
From Eqs. (5.31) and (5.27), we obtain the stationary HamiltonJacobi equation
v
'--
<
5 32 >
-
We
can introduce the time-dependent action function
into this equation
by making use of the relation p
and
eliminate
.
the
parameter
with
the
vS VS(/) help of the relation
We
then obtain the time-dependent Hamilton- Jacobi
equation, which can be used in nonstationary problems:
0.
(5.33)
The stationary and time-dependent Hamilton- Jacobi equations correspond

to the stationary
and time-dependent Schrodinger equations,
particle (K =
respectively. It can be
0,
readily shown that in the case of motion of a free p const, /;= const), the action function is
S(t)
= -Et+pr.
(5.34)
To see
S(0into
sufficient to substitute the expression (5.34) for Let us take special note of the fact that the function for the motion of a free particle is [see (4.60)
this,
it
is
(5.33).
(5.35)
The relationship between the wave function and the action funcwill hold in general form whenever we make the transition from the Schrodinger equation to the Hamilton- Jacobi equation. We
tion
shall consider further only the stationary case, for which
(5.36)
82

start with the Schrodinger equation in operator
We
form
5 37 >
-
o.
<
Making use of the momentum operator p

(5.36)
ih\j,
we
obtain from
(5.38)
'VS)*.
(5.39)
We note from Eq. (5.38) that Eq. (5.36) leads to the same relation between the momentum operator p and the action function S as in the classical theory if we replace the momentum operator by the
classical
momentum p.
Substituting (5.39) into (5.37),
we
obtain
/!_
I
2/Wo
(5.40)
which
is
To obtain last term in
simply the transformed Schrodinger equation. the Hamilton-Jacobi equation, we must neglect the
(5.40), that is, set #~0. It is a well known fact that the quantum-mechanical equations transform exactly into the classical equations when # (). if, on the other hand, h yt 0, but the condition
is satisfied, the quantum-mechanical terms provide only small corrections to the classical equations. The approximation corresponding to this case is known as the quasi-classical approximation. Since /? VS, the above condition can be rewritten as
In particular, for the
one-dimensional case, we have
dx
*.!_
i
*X
dx
(5.41)
Thus, the quasi-classical approximation turns out to be sufficiently accurate in cases where the de Broglie wavelength is constant or changes very little over distances of the order of the wavelength.
Since
83
we can rewrite
the condition (5.41) in the
form
P*
<1,
(5.42)
where F
this,
=
=
^-
is the
classical force acting on the particle.
From
it follows that the quasi-classical approximation becomes inapplicable at small values of the momentum and, in particular, at the points where a particle would come to a stop in the classical Such a state of affairs is obtained, for V, p Q). theory (E example, in a potential well at the points where the direction of a particle is reversed as a result of reflection from the potential barrier (the turning points). A simple explanation can be given for this conclusion: namely, the de Broglie wavelength tends to infinity as and, when this happens, the wave-like properties of a particle become too important for the particle to be treated quasi-
p+
classically.
D.
THE WENTZELis
.APPROXIMATION
As mentioned above, Eq. (5.40) Schrodinger equation. Therefore,
completely equivalent to the would be possible to take Eq. (5.40) as the basis of the wave theory by treating the term which is proportional to h and which does not appear in the classical equation as a new quantum-mechanical potential energy
it
yquan
= _^^^
(5.43)
to be added to the Hamilton- Jacobi equation. The general solution of the nonlinear Schrodinger equation (5.40) is, however, much more complicated than the solution of the linear Schrodinger equation, and therefore the many attempts to develop the quantum theory by means of further analysis of Eq. (5.40) met with failure. Fortunately, Wentzel, Kramers and Brillouin succeeded in finding an approximate solution of the Schrodinger equation (5. 40) by taking only the terms of the order of h. This solution was found to be applicable to a number of problems in quantum mechanics. It is referred to as the approximation. For the sake of simplicity, we shall consider only the onedimensional case, assuming that the potential energy is a relatively smooth function of x (see Fig. 5.1). For particles with energy E, the range of variation of x can be divided into two regions. In the first region (A'<*O ), the energy E is greater than the potential
which has
WKB
energy
(>V),
and in the second region (*>#
E<^V*
It is
obvious
84
that
E=l/(je )at the boundary (x x^) of the two regions. For the one-dimensional case, the original equation (5.40) becomes
ihS"
= 2m (EV) =
(5.44)
X=XQ
Fig. 5.1. Potential V(or) vs
A
X
method
by the
WKB
shall find the solution of this equation for region I can be interpreted as the square the quantity /> of the classical momentum. The solution will besought in the form of a series
2
First, we (>l/), where
>0
S Q -{-Si
-f-Sa-f- ...,
(5.45)
where
is
the quantity S is independent of //, S is proportional to f? S> proportional to // , etc. Substituting the series (5.45) into (5.44) and neglecting quantities proportional toft and to the higher powers of //, we obtain
i}
{
-|
s,;
2$;>s;
/fts;/
/>*.
(5.46)
Taking both sides of the equation, we equate terms that are independent of // and, similarly, terms that are directly proportional to // (here it is necessary to bear in mind that the quantity S is proportional to h). We thus obtain
L
(5.47)
Hence,
it
follows that
(5.48)
This approximate method of solution is basically the same as the perturbation method, which is also successfully used in the solution of the Schrodinger equation (see Chapter
14 below).
THE TIME-DEPENDENT SCHROD1NGER WAVE EQUATION

Therefore, retaining only terms of the order of
AO
85
#,
we have
(5.49)
=S
-\-Si
=
(j
pdx
A
+ to \nl
f
~p.
Substituting (5.49) into (5.36), we obtain the following expression for the wave function in the first region (A'<\Y O ):
to
^^ = -^(A
Q
AO
/;</*
cos| J
fl sin
/x/A
),
(5.50)
In exactly the
same way,
for the second region (x^>x
()
in
which
r <^0, we obtain
where
The wave functions (5.50) and (5.51) are the desired approximate solutions. From these equations it is seen that, when I!^>V, the wave function is cosinusoidal or sinusoidal, as in the case of a potential well [see Eq. (4.21)] or a free particle [sec Eq. (4.47)], whereas, whenV>>Z:, it changes exponentially, as in the case of transmission through a potential barrier [see Eq. (4.23)]. const with the solutions for the Comparing the solutions for case where the potential energy is a function of v, we see that the transition from one case to the other is simply equivalent to replacing the area of the rectangular barrier contained between the x
V=
axis and an axis indicating the constant quantity
_ = V^^V^J^ n
\JP\
by the area enclosed between the x axis and the curve for V -= V (x). Schematically this transition can be represented in the following manner:
similar transition
is
made
in the
case of a potential well.
Thus the specific form of the dependence of the potential energy

on x does not alter the character of the solution. Indeed the solution is determined only by the sign of the difference between E and V, as we indicated at the end of the preceding chapter. The solutions (5.50) and (5.51) give a good approximation only for regions that are relatively far from the special point x (the
{]
86
classical turning point) where the quantity p\ is relatively large. Near the special point (*-*) the quantity p ->0; therefore, the denominator in Eqs. (5.50) and (5.51) vanishes and the actual solution diverges. If we could express the coefficients C and terms of A and B, the foregoing approximation would be entirely JC is comadequate for many problems, since the region \x the relation between these coefficients narrow. However, paratively can be found only if we can connect the functions across the boundary x ) of the regions (by connection of the solutions we mean (x matching the wave functions and also their first derivatives at the boundary x ->* ). The approximate expression for <p must, therel fore, be represented in such a form that at large p Eq. (5.50) holds, while for *=*< when
2
Dm
p*
(x
the approximate solution satisfies the equation

(5.53)
At large
in
z,
terms of an nth order Bessel
the cosine function can be expressed asymptotically function:
'-(*>
<
5 54 >
-
Therefore,
if
we
set
*0
z=^pdx, X
then
(5.55)
<(W
^J/f.Uz).
(5.56)
For large z (and any n), because of the asymptotic formula (5.54), the solution (5.56) transforms into the solution (5.50) found by the method. Let us attempt to choose the order n of the Bessel function in such a way that the solution (5.56) satisfies the Schrodinger equation not only for large z, when x --,Jt but also near the turning point x *x that is, when
WKB
(5.57)
In this
case (X->XQ
0),
the asymptotic expression for ^
becomes
(5.58)

Substituting
(5.58)
87
into
(5.53),
we
find
that
J n must satisfy the
equation
Introducing the new variable
z= ~^-(x
xY^
(the
argument of the
into this equation, and denoting by primes the Bessel function) derivatives with respect to this argument,
.
0.
(5.59)
If the Bessel function is to satisfy Eq. (5.59) and if, at the same time, the wave function is to obey the asymptotic equation (5.53),
we must
(5.50),
put
/j
= ;~.
Thus, instead of the approximate solution
we have
<
5 - 60 >
Similarly, in place of (5.51),
we
obtain for the second region (E <^V)
*=y
(| $\ P
.V<>
\d*)
+
(5.61)
)},
Vfl
where
ment.
/i/ 3
and /_
/8
are the Bessel functions of an imaginary argu-
In order to connect the two solutions, we must find the asymptotic (5.60) and (5.61) for the region*--**o. The appropriate values of z andp are determined from (5.57). For the Bessel function, it is enough to take only the first term of its expansion:
forms of Eqs.
The solutions
(5.60)
and (5.61) become, respectively,
88
NONREUVTIVISTIC QUANTUM MECHANICS
Connecting the solutions at the point x
=x
(}
we
find
Considering the asymptotic forms of the ordinary Bessel function 5 [see (5.54)] and of the Bessel function of imaginary argument
cos
64>
The Bessel function

kind by the equation
of imaginary argument is related to the
Bessel function of the
first
where
The asymptotic form

f
of the Bessel function with
Jt
|
rk
ran DP calculated either
for
J
or for
cos / x
V
1
,
1/7*
7^(^,1/2)) /
fo r
^(/)'
2
-77,
2
(5 64a)
</>
cosf.x
\ 2
77(/j
V'2
)) /
tor
- 2
--
0^
(S 64b)
Morse, and H Feshbach, ^f^'//JO(/s of Theoretical /'/ns/rs, New York McGraw-Hill, 1953, Vol I, p 622) Therefore, the asymptotic form of f n for real ? has a Indeed, using (5 t>4a) and (5 64b), we find for the discontinuity (Stokes phenomenon) two cases
(see, for example, P.
for
0,
(5 64c)
V*
Unfortunately,
II,
Tr/
"
'
Z
.-
/ for
- -
(5 64d)
authors, including Morse and Feshbach (see Methods of Theoretical 1097), use (5 64c) in analyzing the passage of particles through a potential barrier and thus obtain a complex asymptotic expression for the real function f n (^)-
many
P/ivsirs. Vol
The correct procedure

sum
at
</>
(where the function has a discontinuity)

2
is to
take the half-
expected,
of (5 64c) and (5 64d) in order to obtain the asymptotic expression (5 64), which, as is real
89
find that for large
and also taking account of the relation between the coefficients, we z, the formulas (5.60) and (5.61) take the form
(5.65)
dX
Setting
B=A
in the last equality,
we
find the first pair of
connected solutions
(5.66)
for which the exponentially decreasing solution (5.67) in the region x>jc is the analytic extension of the sinusoidal solution (5.66) for the region x <\Y O To determine the analytic extension of the exponentially increasWe then obtain the ing solution (#>#<)), we must set B second pair of connected solutions
ft
= Ab.
= -r^- cos (i J pdx +
-;-)
(5.68)
(5 ' 69)
in According to Eqs. (5.66) and (5.68), the expression for region I (X<^XQ ) is of the form oi standing waves. It can also be written in the form of traveling waves. Indeed, setting
<|>
90
we have
*
(5.70)
The appearance of the factor p in the denominator of (5.70) means that the probability of finding a particie in a unit volume (that is, a 2 quantity proportional to |'^| ) is smaller, the greater the velocity.
Inside the barrier the exponental solution that takes the form
is
connected to (5.70)
p \dx
(5.71)
turning point,
On
cases where the potential barrier is located to the left of the we must interchange the limits of integration, so that the lower limit will always be less than the upper limit.
In
the basis of these results, we are able to quantize the problem of the potential well approximation. Let us assume energy levels of the particles) in the that we have a potential well of arbitrary shape, as illustrated in Fig. 5.2. Obviously,
(that is, find the
WKB
zr
~T
a
x =x
WKB
method
Fig. 5.2. Quantization in to the
potential well according
the process of quantization in the WKB method will consist in finding the conditions under x which the exponentially increasing solutions on both sides of the potential barrier ( A* and .v :> .Vj.) vanish. According to (5.66), the wave function for this case has the following form in that xt region of the potential well which is adjacent to the boundary of the barrier at A* (there being only an exponentially decreasing solution inside the barrier):
<
(5.72)
91
Because of the requirement of an exponentially decreasing solution inside the potential .v lt the following solution is obtained for the potential well in the barrier in the region v region .v:>.Vii
X
(5.73)
solutions must be identical at any arbitrary point x in the potential well (v, <.v as long as we do not take points too close to the boundary of the potential barrier* Joining the solutions (5.72) and (5.73) at some point t (that i s> matching the two wave functions and also their derivatives), we have
Two
*'
sin
X
X*
a'
*+ T)=>
Xi
cos
(l
+ i) + J,rf*
cos(I
',
If
this
determinant
system of homogeneous equations is to have a nontrivial solution for a and of the system must vanish. We then obtain the relation
the
Tlie integral
? p dx cannot be
\
negative, since
=y
2m
I/)
0.
Hence
(5.74)
Thus the quantization rules obtained from the WKB approximation accuracy up to the terms of the order of ft) have the form
(that is t with
an
p dx
These quantization rules
nonzero term -^
differ
2n1l
--
(5.75)
from the Bohr quantization rules by the presence

0).
of a
= '2xh, which corresponds to the lowest state (n
A more
exact solu-
tion of a similar problem in wave theory (for example, the harmonic oscillator problem) shows that a zero-point energy necessarily exists, although it does not affect the radiation
spectrum. Let us now find the normalization coefficient of the wave function for the case of a potential well. In normalizing the function, we can restrict ourselves to integration over x A' a the interval x\ decreases exponentially everywhere (potential well), since outside this region. The normalization coefficient can be then found from the equation
^ ^
<|>
1.
(5.76)
Since the sinusoidal function oscillates rapidly, its square can be replaced by its average value, 1/2, without significantly affecting the accuracy. We then have
(5.77)
92
The

oscillation period t
= 2r
'(<*>
is the angular frequency) is

A
j
-Vj
25
=2
J
I*
dx
where
MQ obtain the expression
is the velocity of the particle.
Hence, for the normalization coefficient, we
Consequently, the eigenfunction (5.73) in the

-
WKB
1
approximation assumes the form

"
9 /'O
nv
^\ IV MV + -V). (n ]
/
f
i
/<^ 7ft\ >
5 78
Proêm 5-Ll Determine the eigenfunctions and eigenvalues of the energy of a particle iiTa^mree-aimWsional potential well bounded by an infinitely high potential barrier:
)
=
oo
outside the potential well.
Find the conditions under which different wave functions correspond to the same energy value /:, that is, the conditions under which the energy levels are degenerate.
Answer.
E 'n nn t
L
_.***'! '
;i /
rt
i '
'Jl
l\
a-
2m Q
'
\ l{
\r
/j
/8
--(/i lf
/i s
,
sin^
- sin -
sin
/!,=
!,
2,
3,...).
In
particular, when
fy n
/a
fy n
=
,
/i,
:
we have degeneracy for
the case of the two different
wave
functions
ntn
and
n n
Problem
5.2.
\Find two classes
of solutions
(symmetric and antisymmetric) for a
pall'flcieTl^S^BWfw^lmensional
symmetric
I/o
potential well:
for
for
x
/
<:
(region
/
I)
l/o
for
.v
< >
A-
<:
/
(region
II)
(5.79)
(region
III)
always have
that, when l/o can have any arbitrary value, only the symmetric solution must at least one energy levei. Find the condition under which an antisymmetric solution is possible. Show that the antisymmetric solutions corresponding to also
Show
,v>0
represent a complete solution of the system in the case where

the function
V (x)
is
described by
oo for
for
<.
0,
/,
<x <
x>l.
(5.80)
VQ for
Theso problems
are
based on the material contained
in
both Chapters 4 and 5

Plot
93
graph
of
the
potential
energy and of the symmetric and antisymmetric wave

in regions
I, II
functions. Solution.
The wave functions
and
III
have the form
==.
Ai
sin
kx
+ 83
cos
kx
<x <
/),
where
The exponentially increasing
solutions have been discarded because they diverge (indeed,
this is the factor that is ultimately responsible for the discrete values of the energy).
From
*
!
the
==.
boundary
i"
conditions
this,
fy(
/4 a
at
* = T/,
it
it
follows
that
ni
ln (,r)
<|/n(*)
and
From
fy(x)
we
A),
find that
dx
solution,
dx
is possible to
have either a symmetric

fy
= -<(/(
for which
x).
or an antisymmetric solution, for which

8
(x)
For a symmetric solution the equation
=0, C
= Di
and the energy levels are found from
tan.=.y"^.,
where
(5.81)
Since ]/p 8
a s /a
>
0,
the
minimum
value of the angle a must lie in the first quandrant
and can be found from the condition
This equation will always have one root for any value of
easily shown graphically from the fact that For the antisymmetric solution [^ (x) x)] t '} ( The energy levels are determined from the equation
in the region
<o<
and
as is
= const >
0.
we have Bt
=Q
Ct =
/?i.
tan a
--
(5.82)
must now
Since the right-hand side of (5.82) has a negative sign t the minimum value of the angle a lie in the second quadrant and can be found from the condition
Sin a
-- tan _
^rrr-^
a
.
/l+tan'a
Introducing the notation
where
the angle 7 lies in the first quandrant ,
we
find
94
is

evident that this equation has at least one root only
It
when
P>-rr.
l.e.
for
Since the wave function of the antisymmetric solution vanishes at A' = t the antisymmetric solution for A:>0 is also the solution for a particle In a potential field described by
(5.80).
.mi
*
the transmission coefficient D of a particle through a potential shape (see Fig. 5.3), if the particle energy /; is less than the height
bamUl
.Problem,5.3.\ Find
U! IMliaiiûlar
of the potential barrier
Vffl
Transmitted
wave
Fig 5.3 Transmission of a particle through a potential

barrier
Solution,
The solution
of the
Schrodmger equation for the various regions has

1
the form:
^rr-./V**
<h,--=/\;,f
-I-
flie1
**
for
for
tkl
A-
**
He*x
< (region < <a (region

I)
A*
II)
*- t
for
(region
III)
Here
k*^~
m"
',
x8
-'"
2
(K
--/:),
/l-j?
/\^
I/ft
and
/?
characterize the incident and
characterizes the transmitted wave; and waves, respectively; /V '*V-<D^ characterizes the reflected wave coming from infinity. Since we have no reflected wave from infinity in our case, we must set Z/ 8 0, 'lo determine the transmission coefficient we shall use the boundary conditions at A* fl and .v We first express /1 3 and B* in terms of /1 8 , making use of the fact that 0. xu ; :
reflected
rt)
I/M v ~
and then express AI In terms of A s
The transmission
(diffusion) coefficient
is
then found to be

where
95
Neglecting the second term in the exponential for D(this
is
possible because the quantity
-y-p
^-is only
slightly different
from
unity),
we
finally obtain
When ft
r~
j
= 0, we get the classical result D = 0.

'
Problem
"*\ A 5.4.1
particle is in a potential field of the

.v
0,
form
<
0,
I)
V0t /<*</! Ion, /i<* 0,

a) b)
0^.\;<zl (region
(region
II)
5- 83 )
(region
III)
Show Show
levels,
that the spectrum is continuous. ra 1 ^~K ) there must exist quasiand / (/i that for */ /) (0 <C that is, states such that there is only a transmitted wave in region III (see
>
>
Problem 5.3X
c) Construct a graph of the wave function corresponding to a quasi-level. d) Find the squared modulus of the wave function inside the potential well and explain why it decays exponentially with time. Solution. In the various regions, the wave function has the following form (see the notation of Problem 5.3):
^s=sA\
sin
kx>
From
the boundary conditions,
we
find
of E, and therefore the spectrum is continuous. obtain quasi-levels by requiring that there should be only a transmitted wave in the external region (region III), that is to say, only a wave moving in the positive A direction. Setting /) 8 0, we obtain the following equation for the energy of the quasi-levels:
These relations hold for any value
We
-2
x
It
+ /F^? _ h
fa
Yp
a*
( \
tana
W*
V
**}
(5.84)
should be noted that the amplitude C* in this case is
much smaller
than the amplitude
96
For <J oo, Eq, (5.84) becomes exactly the same as Eq. (5.82) which gives the energy levels of the potential well (5.80X We let t: Q denote the value of the energy of a particle in the potential well which we obtain from (5.82) and, for present purposes, we disregard all real corrections to , as they make no essential difference to the problem. We then obtain the following expression for the energy:
E^
where
~MA,
(5.85)
2/
and where
under the condition */ 3* \. The presence of an imaginary part in the expression (5.85) for the energy indicates that the wave function decays exponentially with time. The transmission coefficient through the potential barrier for this case can also be found from (5.85). Indeed, the squared modulus of the wave function inside the well is
and thus X, which
is called the
decay constant, characterizes the decrease of the prob-
ability of finding the particle inside the potential well. The quantity X is related to O, the transmission coefficient for a single collision of the particle with the potential barrier, by the equation
where y
,
is the
number
of collisions with the barrier
per unit time. Hence we obtain an
expression forD:
lhe same expression for was obtained in another way in Problem 5.3. In Chapter 6 we shall obtain it for the general case in which the potential energy is an arbitrary function of the coordinates.
The constant
3.4).
also determines the duration of radiation
Therefore, the quantity A
rad
(see also
Problem
Jl\
should characterize the width of an energy level.
Chapter 6
Basic Principles of the
Quantum Theory
of Conductivity
A.
TRANSMISSION OF A PARTICLE THROUGH A POTENTIAL BARRIER (TUNNEL EFFECT)
at
its
According to the classical theory a particle can be located only those points in space where the potential energy V is less than total energy This follows because the kinetic energy of a
.
particle
--=
2tn
' (6.1) *
E (a potential positive. In a region where V has value the momentum an and, classically, barrier), imaginary the particle cannot exist in such a region. Therefore, if two regions of space in which E^>V are separated from one another by a the classical theory does /:, potential barrier inside which V not allow a particle to penetrate from one region into the other. In^the^ wave theory, however, an imaginary value of the TQpmen-
must always remain
>
>
typ^(see
the"
"fib
WKfe Approximation* metFTodT* CHlgleTj?^ simply
arT exponential <3ep0TidelTce of ffiejvave functtonJJrj tKe^co?TrdInafes/"Since the WaveTufifctioh does not'vahisK inside the potential barrier, it is quite possible for a particle to leak through it. This phenomenon is observed in the case of microparticles. The penetration through a potential barrier is called the tunnel
corresponds
It is a specifically quantum-mechanical effect and has no analog in classical mechanics. With the WKB method, it is a relatively simple matter to determine the probability of penetration of a particle through a potential barrier of an arbitrary but sufficiently smooth shape (see Fig. 6.1). Let us assume that a particle is moving in the direction of positive x in the region I ( co<#<\Vi) where E^>V(x). It encounters a potential barrier (*i<*<*j), where E<^V(x), at the *!, and then falls into the region III (*>*J, where again point * V (x). The beginning and end points of the potential barrier can be found from the condition
effect.
>
I'
(*)
(6.2)
90
The de Broglie waves corresponding to the motion of this particle from the potential barrier and partially transmitted through it. The transmitted waves will then propagate
will be partially reflected
in the region III (A:^>JCJ. To determine the probability of penetration of a particle through the potential barrier, let us begin by
Incident
wave
Transmitted wave
Kig
(>
Schematic diagram of
a potential barrier
of an Arbitrary but sufficiently smooth shape The im ident and transmitted waves are repre-
sented by the solid curve, and the reflected wave by the dashed curve
analyzing the wave in region III, where the solution has the simplest form, since in this region there can be no wave moving in the direction of negative A'. According to Eq. (5.70), the solution in region III has the form
....
where
(6.4)
We define the transmission coefficient as the absolute value of the ratio of the flux density of the particles transmitted through the barrier to the flux density of the incident particles: D _|
To determine
A,
Jmc
(6.4a)
the particle flux, we shall make use of Eq. (5.23). Setting the constants e and m equal to unity (which we are allowed to do because we are interested only in the ratio of the particle fluxes), we obtain
THE QUANTUM THEORY OF CONDUCTIVITY
99
Substituting (6.3) into (6.5),

i
we
find
= \Bm\*
=-1*111
4
1
for
''m^
'
/in
for
8m=
<
6 7>
-
It is seen from the above that the amplitude g in characterizes the wave propagating in the direction of positive x, and the amplitude hm the wave propagating in the direction of negative x. Since, as has been already mentioned, there is no wave propagating in the negative x direction in region III, we must set //m = 0. Then
and the corresponding joined solution inside the potential barrier g\\\ in (region II) can be found from Eq. (5.71). Setting /j~(J and g
Eq. (5.71), we find
V*3
*J
I
If
'
Jl
-i
I
i i
_L
snr
,j
jj-k tC O\ (o.yi ' l
J,
where
|
= Y2tfh(V~Ê).
(6.10)
Using the equation
where
Jf 2
.Vj
rf*
==
-J-
lf2Jtfy=~E~)~<tx
(6.12)
the solution (6.9) can be put into a form such that it can be connected to the solution in region I by means of Eq. (5.70). Thus we have
'
dx
100
f
J
'
NONRELATltffSTIC QUANTUM* MECHANICS!

(5.70), the solution in region
I
According to Eq.
(x<^x
has the
form
Substituting this solution into (6.5), the following expressions for the incident and reflected waves are obtained:
/ref=
lft|
(6.16)
Equations (5.70) and (5.71) can be used to relate the coefficients

with
ffi
+ A = Tft
i
""
T
'
fi
6 1? )
-
ft-Ai
= -2/gm^.
(6.18)
From
this,
we
find
(6.19)
In
accordance with the definition of the transmission coefficient (6.4a), we have
>^M'tr
l_
gill
_ *s
I
(6.21)
In exactly the
same way, we
find the reflection coefficient
(6.22)
From
these formulas, it follows that the sum of the transmission and reflection coefficients is equal to unity:
=\.
(6.23)
101
For cases where the quantity 7 is much larger than unity (these are the only ones of practical interest), the transmission coefficient (6.21) is given by the expression
1
D ^ <r
2T
-E) dx
e
(6.24)
In the classical limit (h Q), it is evident that the transmission thus the coefficient becomes zero, as we would expect, and penetration of particles through the potential barrier is impossible.
B.
THE TUNNEL EFFECT AS A MANIFESTATION OF WAVE PROPERTIES
typical
The penetration of particles through a potential barrier is a manifestation of their wave properties. Therefore, an of this effect must occur in analog of wave theory. every type
In
known
optics, this analog is the wellphenomenon of total internal
Glass
reflection, which occurs
when
light is
reflected from an optically less dense medium. Let us assume that a ray of light propagating in glass strikes a glass-air interface (air being the optically less dense medium). (See Fig. Then the wave field, which can 6.2.) be characterized by the electric or
Air
magnetic
in glass
n^>
1)
field strength, is described (where the index of refraction and in air (n 1), by the follow-
Fig.
6.2.
a
from
Propagation of light more dense to a less dense medium.
ing equations, respectively:
(6.25)
'f -~(.vsm0 n
-|-j'c
(6.25a)
In
tial
barrier (see
solving the similar problem of penetration of a particle through a rectangular potenProblem 5 3), we obtained the same exponent as in (6 24), but in front of
2
the exponential there

In the
was
a
16rc
a factor
<1
In
\
,
--
*/
,,V
VV -RJ
which was of the order of

is
unity
case of
ous at
all points), this factor
smooth barrier (that is, a barrier exerting a force which becomes exactly unity
continu-
102

2 Equating these functions at the interface (at the plane y obtain the familiar laws of refraction:
we
ft
sin
Oj;
Substituting these
values
into
(6.25a)
we
V\
find for the refracted
wave
\
---
(vrt
sm
From
this
it
is
seen
propagated
in region
II.
that, if In the
n^mB <^\ an ordinary wave will be case where smOi> 1, we have total
{
internal reflection, the physical analysis of which can not be given on the basis of the laws of geometrical optics. From the standpoint of physical optics, which accounts for the wave properties of light, the electric and magnetic fields are exponentially decaying:
(1)1
.
._
y/i a sma
/U)[
1
Oj
uojf
-|-
xn
sin Oj
If in the case of total internal reflection the refracted wave encounters a second glass surface (region III) (that is, we have two pieces of glass separated by an air layer), then in region III the wave is again propagated according to Eq. (6.25). Its amplitude, however, will be an exponentially decreasing function of the width of the air layer /
.
(6.26)
A diagram
optics is
Wave
of the tunnel effect in

in Fig. 6.3.
shown
Before concluding the physical ^transmitted analysis of the penetration of parthrough ticles through a potential barrier, barrier we should also consider the socalled "tunnel effect paradox." This 6.3. Schematic of the diagram Fig. tunnel effect in optics (total interparadox lies in the fact that, at first nal reflection). glance, it seems as though the real classical particles inside the potential barrier are in a peculiar state characteristized by an imaginary momentum. However, it is important to remember that in this purely quantum-mechanical phenomenon, it is only the probability of a particle being somewhere inside the barrier which decreases exponentially as we recede from the boundary into the potential barrier. Inside the barrier, the momentum and position of a particle are real, and are both given within the framework of the ordinary
We
shall not write here the expression for the reflected wave.
103
uncertainty relation. In order to show this, let us consider a potential barrier with constant V. In the first approximation, the wave function inside the barrier changes according to the equation
?II== ^-
W
,
(6.27)
where
E)
= const.
(6.28)
The right-hand side

Fourier integral
of (6.27) can be represented in the
form
of a
e-**= \f(k)coskxdk,
ff
(6.29)
where
W>=S7?TF7'
(6 ' 29a)
which means that it can be represented as a set of wave functions having real momenta. Obviously, the amplitudes / (k) will effectively to x. differ from zero only when k varies in the range from Thus the uncertainty in momentum is
Ap ^ftx.
According to wave mechanics, the position of a particle inside the potential barrier can be determined only to an accuracy within the order of the width of the barrier:
Multiplying Ap and A*
we
obtain
the accuracy Since our equations are valid only for the case y/ > in determining the momentum of the particle and its position inside the potential barrier will not contradict the uncertainty relation.
1 ,
C.
MOTION OF ELECTRONS
IN
A METAL 3
The theory of the tunnel effect has a number of very important applications both in the theory of metals and in nuclear physics. On the basis of this theory it is now possible to explain a number of phenomena which could not be accounted for in classical physics,
3
See H. Bethe and A. Sommerfeld, Elektrontheorie der Metalle, Handbuch der Physik,
Berlin: Springer, 1933, Vol. 24, part 2.
104
such as cold emission (the emission of electrons from a metal under the action of an electric field), contact potentials, etc. Before discussing these phenomena, we shall say a few words about the theory of an electron gas, which underlies the electron theory of
the conductivity of metals. The high conductivity of metals indicates that electrons are able to move relatively freely inside the entire crystal lattice of a metal. Their excape from the metal into vacuum is, however, hindered because this requires the expenditure of a certain energy, the socalled "work function." This suggests that as a first approximation, we may simply consider the metal as a potential well, inside which the potential energy of an electron can be taken as equal to zero (I/ 0), while outside the metal (that is, in a vacuum)
This simplified model enables us to explain several phenomena occurring in metals. Some of its fundamental results, obtained for the case of free electrons, can be extended (with the help of quantum mechanics) to include the periodic field of a crystal (see below the simplest one-dimensional Kronig- Penney model, which correctly describes at least the qualitative aspects of many phenomena). The electron gas model of a metal was first considered in classical theory (the theories of Drude, Lor entz and others). In this version of the model, the classical Maxwell-Boltzmann statistics, which had successfully explained many phenomena in the kinetic theory of gases, was now applied to electrons. However, the electron gas model encountered great difficulties in developing a theory of specific heat. In accordance with the theorem of equipartition of energy, well-known from classical statistical mechanics, each degree of freedom must have, on the average, an energy 4
l
av
=J
kT,
(6.30)
k is Boltzmann's constant. From this, it is evident that the contribution of each tree electron to the total specific heat will be the same as that of a free atom:
where
This contradicts the experimental facts which indicate that the specific heat of a monatomic metal is the same as that of the lattice atoms; that is, in the first approximation, free electrons make no contribution to the specific heat of a metal.
The specific heats of monatomic substances will be considered Chapter 12 [see formula (12.66)].
in greater detail in
105
This contradiction was resolved by Sommerfeld, who showed

that electrons in a metal do not obey the classical distribution
Instead, the distribution is distribution function
characterized by the Fermi-Dirac
/F-D=-,
_!.
-!.
,*/+,
The Fermi-Dirac quantum statistics is based on the Pauli exclusion principle, according to which each energy level can be occupied by at most two electrons (two quantum states which differ only by the direction of spin). If we are given a three-dimensional potential well of a cubic shape, with side length equal to L, then, according to Eq. (4.57), the components of momentum p fik will be related to the integers HI, t\i and rtj characterizing the energy level by the expressions
*"/."'
We
/T>
note that a unit interval of quantum
numbers
toii\nAn A
is
h >cl*p.
(6.30a)
fore,
associated with only one level, occupied by two electrons. Thereif there are electrons per unit volume, the maximum p momentum of an electron at absolute zero (T 0) can be determined
from the relation

p max
or
f
Pmax=ft(3^p
) /i.
(6.32)
The corresponding maximum
kinetic energy is
(3* P .)V..
(6.33)
This energy
is
called the
Fermi energy.
106
silver.
As an example, The density
let us compute the value of this energy for of silver is 10.5, and its atomis weight is 107.9. Assuming that the number of free electrons is equal to the number of atoms per unit volume, we have
Po
-6.02- 10
= 5.8- 10
22
where we have used the Avogadro number (the number of atoms in one 21 gram-atom, equal to 6.02 10 )*
Hence, Eq.
Fig. 6
4.
(6.33) gives
IA iu
i> "
Model of a potential well

a metal.
.
for
ro
i
Tn a Y
max
= o,D
o e
QTQ *"*&
c o == O,O ev.
K max
is the
function for silver ev, the depth of the potential well in silver is found to be equal to 9 ev. A schematic diagram of the filling of energy levels in a metal is given in Fig. 6.4. The average energy of an electron in a metal is given by the
levels at
T-
upper limit of the filled o (the Fermi energy)
Since the
is
W = 3J
work
equation
770 ^ av
2
"p"o
f
.1
p'_
2/Wo
d*p_
8*
W __
3 5
pn ^ max
'
(6.33a)
agreement with experiment, it follows that at relatively low temperatures the electron gas makes no contribution to the specific
In
heat, since
If
the temperature differs

into higher
from zero, some of the electrons
will
energy levels. The distribution of electrons in the higher levels will not be characterized by a Maxwellian distribution, but by the Fermi distribution function
jump
f E/kT
(6.34)
1
At T
case considered above), this function equals unity if if >max For TV the average energy can be obtained from the expression
(the
'<ro ax and zero

,

Hence, the contribution of each electron to the specific heat
'
107
is
'
(6 ' 35)
which vanishes as T->0. At high temperatures, when the quantity

e E/kT
becomes much greater than
unity, the
Fermi-Dirac distri-
bution function (6.34) approaches the classical Maxwell-Boltzmann distribution

f
= Ae- E
/ kT
,
(6.36)
which, as we know, implies the following expression for the average energy of a free electron:
J7 e l
/fl
*'
~?c
AT 1 K
/C Q7\ D. O
I
expression with the average energy at low temperatures [see Eq. (6.33a)] and using the condition
By comparing
this
^>E V
it
(6.38)
is
possible to define the degeneracy temperature
rdeg
~^(3~
po)
v '.
(6.39)
At temperatures higher than TdeK (T> Tdeg ), we can use classical statistics to describe the behavior of electrons in the metal. If we substitute the value of p 0f say, for silver, we obtain rdeg ^- 1020
Thus, at all temperatures at which a metal exists in the solid state, the electron gas has a certain degree of degeneracy. This means that in discussing the properties of electrons, we must use only Fermi-Dirac statistics, and, moreover, the principal term in the expression for the kinetic energy of free electrons is independent of temperature. This large value of the degeneracy temperature is associated with the small mass of the electron m . Ions and molecules have a mass thousands of times greater than the mass of an electron, and therefore classical statistics applies to them at ordinary
temperatures.
thousand degrees.
D.
REMOVAL OF ELECTRONS FROM A METAL. COLD EMISSION
From
Fig.
the potential-well model of electrons in a metal (see 6.4) we can see that to remove an electron from a metal
108
it
is
necessary
to
impart to
\\7
it
an amount of energy no smaller than

17
the
work function
w
K
o
V ^max'
I CL 4Q\ vo.tuj
As we know, in the external photoelectric effect an electron receives an energy fan from the absorbed photon. Thus, an electron can leave the metal with a kinetic energy
*
ft(0
(6.41)
(Einstein's equation). It follows that the work function represents the minimum amount of energy that must be added to the electron in order to make its energy greater than the height of the potential barrier.
Potential energy in the absence of a field
Metal
Vacuum
Potential energy in the presence of a field
Fig
in
6.5.
the
absence and
line
Potential energy of an electron in a metal in the presence of an external

electric field.
The dashed
energy
shows the behavior of the potential curve when the image force is taken into
account
metal at T some of the electrons occupy energy lying above the Fermi level. If we increase the kinetic energy of the electron gas by heating the metal, a certain fraction of electrons may acquire an energy exceeding the height of the potential barrier, and thus a current will flow from the metal. This phenomenon, which is known as thermionic emission, is used to obtain an electron beam in electron tubes. Under the action of an external electrostatic field, this current may also arise even at lower temperatures. Let us consider the influence of an external electric field g applied to the surface of a conductor in the negative x direction. The potential energy for this case is
In
levels
(x)
=V
ej&x,
(6.42)
where
= ~e
(}
is
the electron charge and
is the
electric field
intensity (see Fig. 6.5).
109
In addition to the external electric field, the electron experiences an electric force called the image force. This force arises because " e induces an an electron with charge image" charge eQ at the metal of the surface (see Fig. 6.6).
Fig. 6 6. Image forces: an electron outside the metal experiences the attractive force
of the induced charge.
Thus the
total force acting

/?
on the electron
<?
is
(6.43)
The effective potential energy, taking is of the form

l/ eff
into account the
image force,
=Vt-e&x-.
at the point
A:
:
(6.44)
The quantity V eff has a maximum
<
6 - 45)
The maximum value of
Veff is less
than
1/
because
(6.46)
V Q -V3&.
Thus, taking into account the electric image force shows that when an external field is applied, the work function decreases and becomes equal to
(6.47)
The electric image forces, however, do not explain cold emission.

In fact, an estimate of the maximum current tungsten, for instance, gives the value
(with W"
0)
for
=^
eo
^2.10 v/cm,
8
(6.48)
whereas experimentally rather strong current

field as
is
obtained with a
low as
g^4
10
v/cm
(Millikan).
110
Thus, within the framework of the classical theory it is impossible to explain quantitatively the cold emission of electrons.
In the quantum theory of this phenomenon (essentially the transmission of electrons through a potential barrier), we limit ourselves to Eq. (6.42) for the potential energy and neglect the electric image force since it does not significantly affect the final result. It can be seen from the graph of potential energy (see Fig. 6.5) that the external electric field produces a potential barrier of finite width. Because of the tunnel effect an electron can penetrate this barrier, the transmission coefficient being given by
D=e
V2mQ
V~V (x)
-E
dx
(6.49)
The integral in the exponent must be taken over the entire width of x given by the condition the barrier from x to the point x
* &*i
that is,
*!=
~E
.
(6.50)
Then
*"!
W(d=Edx=\
= V~*A
V*i-xdx = l
Finally, we obtain the following expression for the transmission coefficient D:
D=e~
MS
=r"
(6.52)
where the quantity 8 depends on the nature of the metal and the energy of the free electrons inside the metal. The cold emission
current
is
proportional to the transmission coefficient
y=/ D==y r
o
s~.
(6.53)
It
follows from the last equation that cold emission should be observed for an electron field of g~10 8 v/cm. This result is in good agreement with experimental data.

E.
III
CONTACT POTENTIALS
Contact potentials, which were discovered by Volta, can also be explained on the basis of the tunnel effect. Let us consider two different metals with different work functions and different Fermi energies (see Fig. 6.7). If these two metals are brought into contact, they will still be separated by a potential barrier of finite width.
Vacuum
Metal
I
Vacuum
Metal
II
Vacuum
Fig. 6.7.
Two metals
before they are placed in contact with each

other.
Wj and
W/2
are work functions, EQJ and &Q2 are e u PP er limits of the filled levels (the Fermi energies)
Since a certain number of filled energy levels in metal I lie above the highest filled level of metal II, electrons can move from metal I into the empty levels of metal II by the tunnel effect. From Fig. 6.7, it is seen that no flow in the opposite direction is possible, since electrons of metal II would then have energies corresponding to filled levels of metal I. It is obvious that the electric current from metal I to II ceases only when the uppermost filled levels of both metals are of the same energy.
Fig.
6.8.
Two
contact.
metals after they have been brought into Formation of the contact potential.
As a result of the tunnel effect, metal II acquires an excess of electrons and is charged negatively, whereas metal I is charged positively. Thus, the energy levels of metal II are shifted upwards relative to those of metal I (see Fig. 6.8). After the Fermi levels
.i*
in both metals are equalized, the electric current ceases, but then there arises a potential difference proportional to the difference
between the work functions of the rnetals:

A(D
^&i J
<&o= a
5-H
f,
(6.54) * *
This quantity
is
called the contact potential.
F.
THE MOTION OF ELECTRONS

KRONIG-PENNEY MODEL)
IN
PERIODIC ELECTRIC FIELD (THE ONE-DIMENSIONAL

As has been previously mentioned, the representation of the motion of an electron in a metal in terms of the potential well model is an approximation in which we average out the periodic
potential of the lattice. A number of characteristic features of the motion of electrons in a crystal appear only when the periodic variation of the potential is taken into account. In the general case, the solution of the problem is very complicated. In order to determine some of the qualitative features of this motion, we may consider, however, a simplified model of a crystal. In the one-dimensional Kronig-Penney model, the periodic electric field produced by the positive ions of the crystal is approximated by a periodic square-well potential of the form shown in Fig. 6.9. The width of each well is denoted by a, and the width of the barrier between two successive wells by b. Thus, the period of the potential (the equivalent of the lattice constant) is c a-\-b. The barrier height is set equal to K
/?=/
n=3
Fig
6.9.
One-dimensional Kronig-Penney model of a

crystal.
The solution of Schrodinger's equation for the nth section of the periodic potential has the following form: for the potential well:
=A
<[>;
sin kxn
+B
cos kxn
(6.55)
for the potential barrier:
= An
sinh
x (xfl
c)
+ B'
cosh
* (*,t
c).
(6.56)
113
Here
is
= j-- =
,
=*
*/i
the coordinate *n
measured from
Similarly, for the
(rc-f- l)st
the origin of the /zth section (that is, the nth well). section, we can write
*n+i
,isinfaf/1+1
fl /1+1
cosft^
(6.57)
We
is,
first join solutions (6.55)
&txn
= a), obtaining
and (6.56)
at the point x
= en
>"
-+-
a (that
A n Biuka-\-B n coaka
^sinhxft
+ Bicoshxft,
(6.57a)
Next, joining the solutions (6.56) and (6.57) at the point x and noting that xn c and ^,=0, we obtain
c (n
-f- 1)
Let us substitute (6.58) into (6.57a) and simplify the problems by considering the limiting Kronig- Penney case in which the width of the barrier between two wells tends to zero (b->0), the height K u tends to infinity, and the width of the well remains constant:
-
Then, since cosh
t.b->
and sinh xb->x&, we have

sin
An
An
cos ka
ka
+B
n cos
ka
=B
lf
Bn
sin /ca
59
,
/1,M1
5 -^ /Ml
Equations (6.59) are linear difference equations; their solutions should be sought in the form
Aa
= dr*
Bn
= C/\
(6.60)
n
Substituting (6.60) into (6.59) and dividing both equations by r we obtain an equation from which we can determine the quantity r and the relationship between the coefficients C and C>:
,
{
Q sin ka = C
C, (?
r sin
(r
cos ka),
(r
ka}
= C,
cos ka).
<
6 61 )
"
114
Multiplying these equations together and dividing both sides of the resulting equation by C^C^ we obtain an equation for r:
r*
2r cos k'a
+ =
1
0,
(6. 62)
where cos
k'a is
given by the equation

cos k'a
=~
sin
ka
+ cos
ka.
(6. 63)
As we
shall see below, (6.63) is the fundamental equation for the energy levels in the periodic field of a crystal. The solution of (6.62) has the form
r
= cos
k'a
sin k'a
ik
'
a
.
(6. 64)
We
note that
if
//will be imaginary,
the right-hand side of (6.63) is greater than unity, and in this case we get an exponential solution
r
lk>\
a==e
*'a
(6.65)
Let us examine in greater detail the solution for the case of real values of k'
r
e ik a .
'
From
Eqs. (6.61) and (6.60), we have

C. 01
** c/smka
008
Substituting the values of ,4,, and A into (6.65) and using the fact that xn an when b->Q, we find an expression for the wave function in the crystal:
~x
y^C^'Uv
where U n
lattice
is
(6.66)
a function with the ~
same
~
e
periodicity as the crystal
*'**'*
sin
^-
w ** sin *
<*
""
a)]
<
6 - 67)
no barrier), we find from case the function / n becomes unity. It follows from (6.66) that an electron can move freely in the crystal if K is a real quantity, that is, if the right-hand side of
In particular, if
(6.63)
that'
(that is, there is
/e;
in this
(6.63) is less than unity.

G.
115
BASIC PRINCIPLES OF THE
ELECTRON THEORY
OF CONDUCTIVITY OF CRYSTALS
The quantum theory of electron motion in a crystal lattice provides a key to distinguishing between conductors, dielectrics or insulators, and semiconductors (which in a sense form an intermediate class of solids). We do not intend to treat this subject quantitatively and shall confine ourselves to a few qualitative remarks based on the onedimensional Kronig- Penney model. As the starting point of our analysis, we shall use Eq. (6.63) to determine the possible values of the electron energy in a crystal lattice. This equation is
cos k'a
p
-r
sin
ka
+ co? ka,
(6.63)
where a
is the lattice constant,
and the quantity k
-~
gives the electron energy.
must be of the electron energy can be found from the condition that real. This condition means that the right-hand side of Eq. (6.63) must be less than unity. oo, we obtain the energy spectrum of isolated atoms (in this case the atoms Setting are separated from one another by an impenetrable barrier). The energy spectrum will then consist of a separate set of levels for each well:
The eigenvalues
P=
L> n
_ n*W
(6.68)
1 , 2, 3 .... We shall not consider the negative values of //, since they give exactly the same values of energy, but correspond to the motion of electrons in the 1, 2) negative x direction. The first two levels (n
where n
'=?>
of isolated
E*=* E
< -
6 69 >
atoms are shown in Pig. 6.10.
Energy levels of isolated atoms

Energy levels
of the
crystal lattice
Fig. 6. 10.
The formation
of energy bands in a crystal lattice.
If P is finite [see Eq. (6.63)], it is easiest to determine the energy levels graphically (see Fig. 6.1 IX The allowed values of k (and therefore of the energy) correspond to the values of the right-hand side of Eq. (6.63) lying between land -\- 1. In Fig. 6.11, these allowed values are denoted by a heavy line. Thus, in a crystal lattice containing atoms each energy level of an isolated atom is split into levels. Each of these groups of levels is called a band (see Fig. 6. 10).
116

The electrons
in the crystal tend to occupy the lowest energy levels and, according to exclusion principle, each level is occupied by at most two electrons with
the
Pauli
opposite spins.
there are
For example, the crystal lattice of an alkali metal contains N valence electrons if N atoms. The electrons occupy only half of the lower band (since there can be at most two electrons in each level). In the ground state, half of the electrons
cos ka
in one direction, and the other half in the opposite direction; consequently, the average current is equal to zero. When an electric field is applied, more than half of the electrons move in some preferred direction, thus producing an electric current and moving up into higher energy levels. Therefore, in a metal, either the first band must contain a sufficient number of unfilled levels, or it must come into
move
Fig. 6.11. Determination of possible energy levels in a crystal lattice.
contact with a second empty band, called the conduction blind* On the contrary, in a perfect insulabands). tor all levels in the first or valence band are occupied and the levels of the second (conduction) band are all empty. The energy spacing between these bands is usually several electron volts. It is easy to show that, when a field is applied, the electrons cannot acquire a preferred direction of motion. The direction of motion of an electron can be reversed only if the electron goes into a different energy state. But since all the energy levels are occupied, this can only happen if the electron occupying that other energy state makes the opposite transition to the state originally occupied by the first electron. Therefore, on the average, there can be no preferred direction of motion of the electrons even when a field is applied. A diagram of the energy levels of a conductor and a dielectric is given in Fig. 6. 12.
cate
the
The heavy lines on the abscissa axis

allowed
values
of
indi-
ka
(allowed
Metal
Perfect dielectric
Fig. 6.12. Energy level diagram of a metal and a dielectric
Nevertheless, every dielectric possesses some (very small) conductivity. The conductivity is much greater in semiconductors, where the forbidden band is considerably smaller than in dielectrics; it is of the order of 1 ev, and sometimes even less (for example, in germanium, the width of the allowed band is 0.66 ev at 7' 300'KX At absolute zero, semiconductors behave like dielectrics; however, their conductivity increases with rising temperature, particularly in the presence of impurities. At room temperature, the resistivity of semiconductors is found to be of the order of 10~ 2 10 2 ohn>cm. At the same time, the resistivity of dielectrics lies within the range of 10 3 10 18 ohnrcm, and that of metals in the range of 10~ B 10~ ohnrcm. The conductivity of dielectrics and semiconductor scan be of two types. Let us assume that under the influence of thermal excitation or the internal photoelectric effect (absorbtion of light by electrons), some of the electrons jump from the valence into the conduction band, leaving vacant states ("holes") in the valence band. The electrons that have jumped
fl
117
into the conduction band become carriers of electric charge, thus producing a current. On the other hand, as soon as several states become empty in the valence band, an electric current can also be produced due to the motion of electrons in the valence band itself. It can be shown that the motion of a system of electrons in an almost completely filled valence band can be treated in terms of the motion of a set of vacant states or "holes." Obviously, the holes move in a direction opposite to the direction of electron motion, so that they behave like positively charged particles. Thus, there are two ways in which charge can be transported and, therefore, two types of electric current in solids: n type ( due to the motion of electrons) and p type (due to the motion of holes). With the help of the band theory, we can easily give a qualitative explanation of a number of interesting phenomena. For example, the conductivity of a metal increases with decreasing temperature because its resistance is due to the interaction between free electrons and the vibrations of the lattice. Since the vibrations of the lattice diminish as the temperature decreases, the resistance will also decrease. On the contrary, the conductivity of semiconductors decreases as the temperature is lowered, because the number of electrons in the conduction band becomes increasingly smaller. The existence of P type conductivity in semiconductors has been demonstrated in investigations of the Hall effect and thermo-
electric emf. The sign of the potentials appearing in these phenomena is determined by the sign of the current carriers. Experiment shows that in some semiconductors the sign of the potentials corresponds to electron carriers, whereas in other semiconductors the sign of the potentials is reversed; that is, the current carriers act as if they were positively charged particles. A natural explanation for this is provided by the concept of "holes." The above conclusions in regard to the conductivity due to transitions of electrons from the valence band into the conduction band (the long arrow on the left-hand side of Fig. 6.13) and due to the motion of holes in the valence band are based on the assumption of a perfect crystal. This type of conductivity is known as intrinsic conductivity. Another type of conductivity, known as impurity conductivity, also plays a significant role in semiconductors. It is caused by the presence of foreign impurities or other structural defects in the lattice. Such disruptions of the perfect periodicity of the lattice lead to local deviations of the field from a perfectly periodic one. As a result, discrete levels may Conduction appear in the forbidden band of the energy specband trum of the electrons (see Fig. 6.13). In practice, the wave functions associated with these levels Donor level differ from zero only in a certain region near the
given defect. Thus, discrete levels are sometimes Forbidden called impurity levels. They do not themselves band contribute to the current (the electrons occupying Accepter them are not free). However, they may affect the level number of electrons contained in both the conduction and the valence bands. Valence Impurity levels (denoted by short arrows on the band of side be can divided into two 6. right-hand Fig. 13) donor A and levels donor types: acceptor levels. is capable of supplying electrons to the conduction Kig- 6.13. Diagram of energy bands so that free m a semiconductor, electrons will appear in the band, conduction band. In contrast to a donor, an acceptor absorbs electrons; as a result a "hole" is formed in the filled band so that electrons in this band are able to Jump into higher energy states (ptype conductivity X If the impurity levels are located sufficiently close to the edge of the corresponding bands, lonization (or other effects) can arise fairly easily due to the energy of thermal motion of the lattice. Therefore, an appreciable number of free electrons (or holes) can exist in bands even under conditions when direct excitation of electrons from the valence band into the conduction band is highly improbable. In the case of impurity conduction, the magnitude and type of the conductivity are determined mainly by the nature and concentration of the impurities. By varying the impurities, it is possible to control within a wide range the magnitude of the conductivity as well as its type ( n type or p type). 5 This fact is widely utilized in semiconductor
,
-i-
may give
Pure germanium has a high resistance but the addition of a small amount of impurities it either n or p type conductivity
118
electronics. It constitutes the basis of operation of modern high-quality crystal diodes and transistor devices which are capable of rectifying AC currents, as well as of amplifying and generating electrical oscillations.
conclusion, we would like to draw attention to two important approximations which implicitly assumed in the electron theory of solids. First, in developing the theory of metals and semiconductors, we were actually concerned only with the problem of single electrons. Accordingly, this theory is called the single-electron theory. The collective properties of electrons in solids were taken into account only in connection with the filling of energy levels by free electrons, when we calculated certain statistical quantities (the Fermi energy, specific heat, etc.). Even in those cases we completely neglected interactions between electrons, although they do in fact experience mutual Coulomb repulsions. It is easy to show that, for the electron densities which we observed in metals, the average energy of this repulsion is not necessarily small in comparison with the Fermi energy. There naturally arises the question of the extent to which this crude single-electron model represents the properties of a real metal. The answer to this question is given in the many-electron theory of solids, which we cannot consider here since this recently developed theory is a highly specialized subject. We may note that the fundamental notion of the electron gas as a free carrier obeying Fermi-Dirac statistics is retained in the multielectron theory. Accordingly, the laws of purely statistical character obtained in the single-electron theory still hold true, such as, for example, the linear dependence of the electronic specific heat on temperature. At most, certain numerical coefficients will have to be slightly changed because of the specific form of the wave functions for a system of electrons. These refinements, however, involve exceptionally large computational difficulties. On the other hand, in dynamical problems, where it is essential to take into account the interactions between electrons, the simplied approach used above (the single-electron theory) is inadequate. One case of this type is the problem of the strength of a metal. The second approximation which is implicit in the single-electron theory concerns the crystal lattice, which is regarded simply as the source of a certain statistical field. The situation is actually more complicated because the lattice ions execute a vibrational motion, which, as we know, persists even at absolute zero (zero-point vibrations). In order to take this fact into account, it is necessary to consider the crystal lattice as a quantum-mechanical system rather than simply as the source of a field (see Chapter 12, D). The essential point in this connection is that the lattice is not isolated from the system of electrons, but coupled to it since the electrons interact with the lattice ions. The energy of this interaction can be uniquely represented by the sum of two terms. One of them represents the potential energy of interaction between the electrons and the stationary lattice. The motion of electrons in a field of this type was considered above in the simple case of the Kronig- Penney model. The second term is connected with the deviations of the ions from their equilibrium positions and represents the interaction energy between the electrons and the lattice vibrations. The average value of this interaction energy is extremely small in comparison with the energy of the electrons at the Fermi level, because the amplitudes of the vibrations of the lattice ions are small (except at temperatures near the melting point). The vibrations of the lattice, however, may play an important role at low temperatures, when the interaction of each electron with the vibrations of the lattice gives rise to an additional interaction in a pair of electrons. If these two electrons have opposite spins, the interaction results in an attraction (in contrast to electrostatic repulsion). As a result, the pair of electrons begins to move with a certain degree of correlation. Such a system is found to have the property of superconductivity. 6 The smallness of the interaction which causes the correlation explains why superconductivity is observed only at
In
we have
extremely low temperatures. 7
A theory of superconductivity based on these concepts was developed independently by the American physicists Bardeen, Cooper, and Schrieffer, and by the Russian physicist Bogoiyubov [See N Bogolyubov, V Tolmachev, D. Shirkov, A New Method in the Theory of Superconductivity (trans ), New York: Consultants Bureau, 1959 1
7
ley,
See R Peierls, Quantum Theory of Solids, Oxford: Clarendon Press, 1955; Electrons and Holes in Semiconductors, New York: Van Nostrand, 1950
Shock-
Chapt er
Statistical
Interpretation of
Quantum
Mechanics
s~
A/ELEMENTS OF THE THEORY OF LINEAR

OPERATORS
In our general investigation of the Schrodinger wave equation (Chapter 5), we saw that in quantum mechanics the momentum operator p i#y is associated with the classical momentum p of
the particle, the energy operator E
= ih-^-
with the energy E, the
Hamiltonian operator or simply Hamiltonian H with Hamilton's function H, etc. Before analyzing the physical meaning of the quantities represented by operators in quantum theory, let us consider certain general aspects of the theory of operators. In the same way as a function relates a number x with another
number
= f(x)
an operator
associates one function/^) with
another function
*(*)
= M/M,
(7-1)
according to some given rule. In order to satisfy the principle of superposition, only linear operators with the two following fundamental properties are used in quantum mechanics:
=A
(
'
where C is an arbitrary constant. The linear operators most commonly encountered are the differentiation sign (for example, the
momentum
operator
M = y -5-
and
the LaplacianM=v*) and the integration sign.

In
Poissonf s equation
V7 (r)
= p(r)
(7.3)
120
the operator is the Laplacian which converts the function / (r) into another function p(/-). Conversely, we can solve Poisson's equation and find
=- J
form
(7.3a)
Here the operator v~* n as the of the operator being
of a definite integral, the kernel
>
'
~4*
|
r f
4*
/(*_*')
+ (y-yj + (-*')
tion by unity
The operators y* a nc V~* which have the same effect as multiplicawhen they are applied in succession, are called inverse
*
operators.
The Hamiltonian operator
consists
of the
sum
of the kinetic energy operator
T==
-~
v%
which
and the potential is of the coordinates. function which a V simply energy We can regard the action of the potential energyj/ (x^ on the ^Ûonji^he action of'AJJJ!^ are satisfieHT THerSelore, besides differential and integral operators, the linear operators that may be used in quantum mechanics include any function of coordinates whose action on the wave function is simply to multiply it. For example, the coordinate r is just as entitled to be considered an operator (the position operator) as the momentum operator p ih\^ which is a differential operator. It is worth noting in this connection that in quantum mechanics r does not represent the position of a particle, but is an argument of the wave function and determines its value in the coordinate space. The quantity which is equivalent to the position can be found from the operator r and the function ^(r) in the same way as the momentum of a particle is found from the momentum operator, that is, by averaging (see the following).
is
directly porportional to the Laplacian,
This will become particularly clear when we write the Schrodinger equation in momentum space (see below), where the wave function depends on p In this case, the momentum operator p will correspond to multiplication by an ordinary function, and the position operator x to differentiation with respect to p.
STATISTICAL INTERPRETATION
B.
121
ELEMENTS OF REPRESENTATION THEORY
x, y and z obviously commute with each since operating with them is equivalent to multiplying by and soon. The zx ordinary numbers. Accordingly, xy yx, xz
The position operators
other,
operators
pv
ih
()
pv
ih
^-
and
p^,
ih
^-
also
is
commute
with each other, since the result of differentiation

of the order in
independent
which
d*
it is
performed
2
-=
oxdy
d ~= dydx
d*
'
d*
=r=
and so on.
,
dydz
0:dy
Similarly, it can readily be seen that the pairs of operators p v and and x and so on, also commute with each other: //, Py
9
that is,
P.V!/=</PJC-
An example
by x and
p^.
of a pair of
noncommuting operators
is
provided
Indeed,
*p^ = -/torjj.-,
whereas
P
(7.4)
^t = -
in
^- =-in(\ + x
$.
(7.5)
Consequently,
(Px^-^pJ * = -**,
that is,
p xx
(7-6)
xp x
ifi.
(7.7)
In a similar fashion,
it
is
readily shown that
The noncommutativity
(7.7)
of these operators, as
and
(7.8), is a significant feature of
expressed by Eqs. quantum mechanics.
The
specific
*
choice of the form of position and
operators x
and(p^ =
momentum
corre-
"""TV
ih\ ~\ satisfying
Eqs. (7.7) and
(7.8)
sponds to the so-UkllUd Coordinate representation,
in
which the
wave function deoends
onIVTSffiH??3T59ri9rflTO
122
The Schrodinger equation can also be written in the momentum representation. For the sake of simplicity, let us take the onedimensional case. Using the (A-) into a Fourier integral
<[>
functions [see (4.70)],
we expand
<|*
(x)
(x
x')
dx'
=~
this equation in the
Since k
=~
wherep = /? A we can rewrite

,
form
V
(x)
,
(7.9)
=^
The Fourier transform
<p(/>)
*t(xyr $*'d*.
of the function
is
ty
namely, the function
momentum
tions.
wave function in the representation. Equations (7.9) and (7.10) relate the wave functions in the coordinate and the momentum representawhich depends on momentum,
called the
Let us find what operator in the momentum representation corresponds to the position coordinate x. We note that in this representation we need not write p as an operator (that is, in
roman
type).
If
we
substitute
fy(x)
for xfy(x) in Eq. (7.10), then

*
X T (p)
= A_
J
-'
x<!f
(x) e
dx
= t*&$-.
x
(7.11)
and thus
in the
form
momentum ~ - -
representation the operator
__-<*
has the
It
is
momentum
easily verified that Eq. (7.7) continues to hold true in the representation, since
satisfied
In exactly the same way, the fundamental relation (7.7) will be if x and p are replaced by certain appropriate matrices.
it
This form of representation is called matrix representation; was introduced by Heisenberg somewhat before the discovery
of
Schrb'dinger's equation. The remainder of our discussion of operators will be conducted in the coordinate representation.
C.
123
AVERAGE VALUES OF OPERATORS
In classical mechanics the motion of an individual point is exactly specified by a function relating its position to time. This dependence can be uniquely determined from the fundamental differential equation of motion
mjr=
V(r).
Once we have determined ras a function of time, we can also determine the momentum and energy of the particle. The situation is somewhat different if there are many particles
involved, as, for example, in the kinetic theory of gases. In this case statistical laws characteristic of a large collection of particles must be used. It turns out that the particles of such a collection obey certain distribution laws, which, generally speaking, apply to both coordinate space and momentum space (that is, a distribution describing both the velocities and the energies). The function / that characterizes this distribution is called the distribution function. Thus, when we deal with a large collection of particles, we can only consider a probability that a particle possesses particular coordinate and momentum values. From the distribution function we can finA that average values of the position and momentum.
xffxfp.
px
= J pjd*xd
p,
and the
mean square
of these quantities
Afj
x*fd*xd*p
and so on,
which in accordance with the law of large numbers should agree with the corresponding experimental values. We must mention one characteristic feature of these statistical laws. In classical physics they are a result of averaging over the so-called "hidden" parameters, which determine the motion of each particle in accordance with Newton's equations. These hidden
in the final results. In principle, however, classical theory enables us to explain why, at any instant of time, the coordinates and momenta of individual particles differ from the average values, even though this explanation may be very complicated mathematically. In quantum mechanics the behavior of particles is described by the wave function ty(r, /), which is a probability function even when the system it describes consists only of a single particle. Thus, quantum mechanics allows us to determine only the average values
parameters do not appear
124
of
whether there is a large must be emphasized that in quantum theory it is in principle impossible to explain the deviations of observed variables from the average values. 2 The method of calculating averages in quantum mechanics is similar to that used in statistical mechanics. The basic formula used for this purpose is
dynamic
number
variables regardless of of microparticles or only one.
It
(7.12)
M is an arbitrary operator (as a special case, it may be a number), and the quantity ty* (O'HO plays the role of the distribution function / provided that the wave functions ty(t) are normalized:
where
The average values of the position and momentum, as has already been mentioned, are in fact computed in basically the same way:
(7.13)
Ic is the coordinate of the center of mass of the wave packet associated with the function (/), and p K is the momentum of this center of mass. Since the outcome of a physical measurement is a real quantity, the average values must be represented by real numbers. There-
Here
<|>
fore, the following equation
must
(A/)*
hold:
Al.
(7.14)
When
satisfied, the corresponding operators requirement are said to be self-conjugate (or Hermitian). In particular, we shall show that the operator p A satisfies the
is
this
condition (7.14), even though it appears to be purely imaginary. As a preliminary, we must first prove an important theorem for "transferring" a derivative. This theorem is as follows. Suppose
we have an
integral
<x
"
uv (n 'dx
(7.15)
that hidden parameters cannot be the basis for the quantum mechanics Von Neumann's proof, howeVer, is valid only within the limits of the actual framework of quantum mechanics itself, and if quantum mechanics is not taken as an ultimate theory, Von Neumann's theorem cannot be regarded as
statistical laws of
Von Neumann has demonstrated
generally valid.
125
where
<n]
= ~^-
Then,
if all
terms of the type

oo oo
'V-2 '],,..,
r<-"w|
(7.16)
vanish, the result of integration of G is not altered if we transfer the /ith derivative of the function v to the function u in (7.15) and place the factor (1)" in front of the integral:
ltt]
vdx.
(7.17)
we carry out an /7-fold integration by parts in (7.15) and that all terms in (7. 16), vanish, we obtain the relationship (7.17). In the case of a discrete spectrum, the conditions (7.16) are always satisfied because the wave function decreases exponentially at infinity. In the case of free motion (that is, a continuous spectrum), these expressions vanish as a consequence of the periodicity condition. Physically, the fulfillment of condition (7.16) means that no particles or currents exist at infinity. Returning to the proof of the self-conjugateness of the operator we substitute , v p
Indeed,
if
assume
u == <p
into
(/),
ihty (t)
and n
Eq. (7.17).
From
this
it
immediately follows that
and thus the self-conjugateness condition (7.14)
is
satisfied for
p^,.
We
is
note that unlike the operator p

its
Ar
ih
real operator-^ ^ v ,the
average value has no physical meaning. has only one eigenvalue (and one eigenfunction <^), it is readily seen that this eigenvalue is identical with the average value of the operator. Indeed, using the general rule (7.12) for determining the average value of an operator and substituting the equation
not self-conjugate and
If
an operator
>.
we
obtain for
M
==*
f <t*(t)tf(t)<Px
\.
(7.19)
126
On the other hand, suppose that the operator M [Eq. (7.18)] has Xn several eigenvalues X t> X.2l corresponding to the functions or this may be the case for example, (* M0--<MO.
.
MO"-
the
energy operator E
<[>(/)
= ih~
-,
for which
X rt
).
Then since the
general solution
can be written in the form

7 ' 20 >
(0.
<
we
find that the
average value of the energy operator E
Ih
is
(7.21)
Here each |Cn 2 is the probability that the particle is in the corresponding quantum state. If all Cn except one,C no are equal to zero, then E E nn and thus the average energy corresponds to the eigenvalue n . Consequently, the eigenvalue no corresponds to the experimentally observed energy. In the case where several coefficients C n differ from zero: C rt|f C n ,, we can obtain Cni any of these values of energy in experimental measurements. If the experiment is repeated many times, the number of measurements which yield an energy E ni should be proportional to the corresponding theoretical probability Cn |*.
|
Chapter 8
Average Values of Operators.
Change
of
Dynamic Variables with Time

A.
DERIVATION OF THE UNCERTAINTY PRINCIPLE
As indicated in the preceding chapter, the observable dynamic quantities associated with the operators must be regarded as average values, given by Eq. (7.12), This is true regardless of whether these operators commute with the Hamiltonian, that is, regardless of whether the corresponding physical quantities are constants of the motion. We shall now show that if t35^gjmamic quantities correspond
to noncommuting operators, they do not have simultaneous definite values in quantum mechanics. Of greatest importance in this respect is the calculation of the deviations from the average values of two canonically conjugate quantities the position* and momentum p x . Our discussion will be carried out in the coordinate representation and we shall restrict ourselves to the case in which the wave function is independent of time (the stationary case). Then the average values of the position and momentum can be found from
the relations
*'
(8.1)
gd'x
(8.2)
First of all, we note that even though the average error or average deviation from the mean, which is given by
==X--Z==Q
(8.3)
is equal to zero, it in no way follows that the particle cannot occupy positions other than x. The reason for this is that the deviations have different signs relative to the mean x, and consequently they cancel out on the average. Accordingly, the deviation from the average value of the operator should be characterized by the variance (mean-square deviation), which is positive for all deviations
128

x.
from
The variance
of the position
can be calculated from the
formula
(Axp
=j
<[>*
(x
x)^d^x
= ^~2
(x)*
+ (xY = tf
2
(x)\
(8.4)
We
0, it follows note, incidentally, that if the variance is (A*y that the probability of the electron occupying a position in space differs from zero only at x x. In this case the average value of the position is equal to the exact value; that is, the corresponding probability of the particle's position can be described by a function similar to the function. Similarly, the variance of the momentum is given by
(8.5)
order to establish the relationship between (Kxf and (Apj* us take a coordinate system whose origin lies at the center of mass of the wave packet (,v 0), and which moves with the same 0). (The use of this coordinate velocity as the center of mass (p* system does not involve any restriction on the generality of the
In
f
let
discussion.)
In this
case
(8.6)
'*
Let us consider the integral
=
where
(8.7)
a is
A-,
(8.7)
some arbitrary
in the
real quantity independent of x. Equation
can be written
form
I(*)
= A**
B*
+C
(8.7a)
where
A= f <]>*x*tyd*x=
8 - 8)
<

Since the integrand in (8.7) is essentially a positive quantity
129
/WÔ,
condition
efficients
9
(8.9)
(8.9) imposes a certain definite restriction on the coA B and C. Indeed, if this relation is satisfied for a a
corresponding to the minimum of the function / (a), it will also be satisfied for any arbitrary a. The value of a can be found from the
condition
/'(,)
= 2A** -fl =
0,
that is, *
'
=^,
and
Hence the minimum value
of
(a) will
B*
be equal to
From this it follows that inequality (8.9) will hold for all real values of a t provided the following condition is satisfied:
B**^4AC.
(8.10)
we
Substituting the values for A, B and_Cfrom_(8.8) and using (8.6), obtain the relationship between (ApJ 8 and (Ax)*
fA)MW^.
=
(8-11)
This inequality represents a rigorous formulation of the uncertainty principle. ih [see Eq. (7.7)], we can By using the relation p xx xp x rewrite (8.11) in the form
xp x
\\
(8.12)
Generalizing (8.12), we can say that whenever two operators M t and M do not commute with each other, they satisfy the uncertainty
2
relation
(AAf ,)'
^j
MjM,
MJVIj
(8.13)
where
= J r (M, -
AT,)'
d^
(/
=1,2).
(8.14)
130
As we have already mentioned,

tions.
the uncertainty principle is a
consequence of the wave-particle duality that underlies quantum mechanics and is in no way connected with the experimental limita-
Experiments may only prove the results which follow the uncertainty principle. The basic meaning of the uncertainty principle consists in the following fact: the probability distributions of variables whose operators do not commute cannot simultaneously take the form of a 8 function (see Fig. 8.1). Moreover, if the probability distribution of one variable approaches a S function, the probability distribution of the other variable will spread out. In the limit, when, for instance, the probability function for x (that is, |<K*)|*) takes the form of a I function [(A*)* 0], the probability function for the momentum p x (that is^ [<p(A*)|') becomes such that it is constant for all values of p x (A/? jr ) 2 oo.
from
Fig.
8.1.
The probability
(b)
distribution function in (a) coordinate

2
If
= H/2. ] the distribution in coordinate space (a) contracts, the distribuspace and
momentum space:
[(Arc)
(Ap)
l/J
tion in
momentum space
(b)
spreads.
The necessary condition for simultaneous measurement of two dynamic quantities is the condition of commutativity of their
corresponding operators.
B.
POISSON BRACKETS IN CLASSICAL
AND QUANTUM
THEORY
The state of a system in classical mechanics is defined by its dynamic variables. The quantities appearing in the canonical, or Hamilton's, equations of motion in classical mechanics depend on
the coordinates x
(
momenta
p and time
t
t,
that is,
(8.15)
/=/(p
|f
t).
131
For example,
in
one-dimensional time-independent problem,
the Hamiltonian depends only on x and p x :
With the help of the canonical (Hamilton's) equations of motion
we
obtain
*-".or
Md
'-.
2,...,
/*),
<
8 - 17 >
If there are n degrees of freedom (/=!, take the form
Eqs. (8.17)
* *'
dH = ^'
H_
.
dH
dxS
f
/ft ' 17a 17a) (8
Hence the time rate of change of the quantity by the equation
[see (8.15)]
is
given
Using the canonical equations (8.17a), we obtain

d
i= i~\
IH.
f],
(8.18)
where
the expression
is called the classical

If /
Poisson bracket.
does not depend explicitly on/, then

/ is
?f=
and consequently
the variation of
completely determined by the Poisson bracket:
g=l//,
If the Poisson bracket vanishes ([//, depend on time, or it is conserved.
/I.
(8.20)
/]
<)),
the quantity
does not
= const.
132

if
the energy does not depend explicitly on time, then it follows that Hamilton's [//, //] f <5///a/ function (the energy in this case) is a constant (// const). Furthermore, substituting the coordinate x l9 and then the momentum p for into (8.20), we obtain the relations (8.17a), that is, Hamilton's / equations of motion. We shall now generalize, the classical Poisson brackets, which
For example,
0.
Since obviously
can be used to fincnne time variati_gn_gijin^^ *he quantumjcaseJ

i'irst
"""""
to
'"
~"
oF"all,
we
recall
that
on).
in
It
quantum mechanics physical
average values of operators is the time rate of change of these average values that we must determine. The average value of any operator f is given in quantum mechanics by Eq. (7.12), in which the time / occurs as a parameter. From this equation, we can find
to the
meaning can be attached only

(position,
momentum, and so
the total derivative of
with respect to time:
= f /(') It
J
at
(/)
**
d
\ J
\" ot
ft
(/)
<px
+
~
Substituting for
~-
and
the expressions
(8.21) to the
^ H^*) and(-
respectively,
we can reduce
form
+{
where
[(H t* (0)
(f
t (0)
- ** (0
(H *
(/))]
d*x,
(8.22)
Using the theorem for transferring a derivative [Eq. (7.17)] and keeping in mind that the potential energy is an ordinary function of coordinates, we readily obtain
Consequently, the change of 7 with time will be given by the equation
(8.23)
133
The expression
JH, f}=4(Hf~fHJ|
(8.24)
mecanical case and
is the generalization of the Poisson bracket (8.19) to the quantumis called the quantum Poisson bracket.
Obviously, in the case where
57
(as a rule
an operator
does
not contain the time explicitly), Eq. (8.23)
becomes
It
follows that in this case the time change of/ is completely deter-
mined by the quantum Poisson bracket. Furthermore, if the operator f commutes with the Hamiltonian operator H, the physical
/ corresponding to this operator is conserved, as can be seen from (8.25). With the help of (8.25) it is easy to prove that the energy of a particle moving in a time- independent potential field V (r) is con-
quantity
served.
The expression
case and therefore
H}=-^-(HH from (8.25) we have
{H,
HH) vanishes
in
this
W = const.
On
(8.26)
the other hand, l\ty n E n ty n according to the time- independent and therefore, when equation, (0 we have [see Schrbdinger
<1>
'])
Eq. (7.20)]
H=
that is, Eq. (8.26) is nothing but the law of conservation of energy const) for a particle moving in a time-independent field of (E force.
C.
EHRENFEST'S THEOREM
We
of
let
f
shall
now
motion
(8.17).
find the quantum analog of the classical equations For this purpose we shall use the quantum Poisson
brackets.
Noting that x and p* do not contain the time explicitly, us use Eq. (8. 25) to determined and ~p X9 substituting into it either x or f as the case may be. In the case of f =x, we find p
je ,
jfc={H,
x}=
(H*
134
where
.
(8.28)
Since x and V
(x)
commute, Eq.
(8.27)
can be reduced
to the
form
Adding the quantity equation, we have
(vx xp x
p x xp x ) to the right-hand side of this
xp x )p x ).
(8.30)
Then, using Eq.
(7.7),
we
obtain
i=
In
m*'
(8 - 31)
we must
order to determine the time rate of change of the momentum substitute the momentum operator p* for the operator f
Then, since
x p x p"
in (8.25).
p!p*
= 0, we find for p
7)
= -?.
(8.32)
Hence, using
(8.31),
we
obtain
F(J).
(8.33)
Equations (8.31)-(8.33) constitute Ehrenfest's theorem, according which the fundamental equations of classical mechanics can be generalized to quantum mechanics by replacing the classical variables by the average values of the corresponding operators.
to
D.
TRANSITION FROM QUANTUM TO CLASSICAL
EQUATIONS OF MOTION
Let us compare the classical equation of motion

(8.34)
with the corresponding quantum-mechanical form (8.33). As was previously stated, x is the quantity which corresponds to the classical position coordinate in quantum theory. Accordingly, we
135
could assume the quantum-mechanical equation to be identical with the classical equation if we had
(8.35)
instead of
position.
of
(8.33).
This would be equivalent to replacing * by
its
average value x
motion
in the classical equation relating the force and the Ehrenfest's theorem asserts, however, that the equation for the quantum case contains the average value of the
actual force, that is, F (x). Therefore, in order to make a transition from quantum equations of motion to classical equations, we must first establish the relationship between F (x) andF(Jf). Let us represent the force operator F (x) in the form
F(x)
= F(x + bx),
in a
(8.36)
where
point x
bx
x.
=x
x,
and expand F(x)
Taylor series about the
Then we obtain
F(x)
= F(X)+(*x)F'(x) + ^-F'(x)+.... =
(jc
(8.37)
Taking the average of this expression in accordance with Eq. (7.12) and considering that "(A*)
x)
0,
we
obtain
-(*)+
the
...
(8.38)
The quantum-mechanical equation form
of motion (8.33) therefore, takes
m
Here the expression
--
= F(x) +
F"
(x) is
^F(X).
(8.39)
*--
the quantum-mechanical correction
to Newton's classical equation. Clearly, the criterion which must be satisfied in transition from quantum equations of motion to classical equations is the inequality
F(x)
(8.40)
It should be noted, however, that mere satisfaction of this inequality is still not sufficient to allow us to apply all classical concepts to the description of the motion of a particle. Indeed, in quantum
mechanics the average value of the kinetic energy f
is
defined as
136
NONREUATIVISTIC QUANTUM MECHANICS
whereas the classical analog of the quantum-mechanical kinetic energy should actually be taken as
T(Px
= J^-
(8.42)
Let us now express the quantum-mechanical definition of the kinetic energy 7'(p.v) in terms of its classical analog T(px ). For this purpose, we shall use the equation
where bp x
p x Removing the parentheses in (8.43) and conpx sidering that after averaging
.
we have
T)
- T (AV) + -- (A/a
2
.
<
8 - 44 )
From
transition
obtain the condition under which we can make a the quantum-mechanical expression for the kinetic energy (8.41) to the classical expression
this
we
from
(^)X pi = 2m T
(/>,).
(8.45)
Multiplying (8.45) by (8,40), we obtain the general condition for the validity of the classical approximation in the microscopic world:
5
2
(A*)
(Ap,)
< 4m r (p x -^L
o
)
(8.46)
If
we
take into account the uncertainty relation
rrT TA~T ^
2
2
condition (8.46)
becomes
r-i /
*
'
16
(8.47)
Let us apply this condition to the hydrogen atom, when

Substituting these values into (8.47),
137
we
obtain the inequality

8 - 48 >
'>-rirSince
r
<
^
fllQPQ
,
=a
where a
is the
radius of the first Bohr orbit, and
2
/z
we
obtain instead of (8.48)
(8.49)
of
and therefore, in the limit of large quantum numbers, the results quantum theory approach the classical results.
p
roblem B &*jjJ Determine the wave function of a freely moving electron in the represe'nt'a'tlonr Write the normalization condition in the p representation. Find the average values of the operators for the momentum and energy of a particle. Solve the problem in the one-dimensional case, and then generalize it to the three-
dimensional case. Solution. Let us choose the
axis along the direction of motion of the electron.
The
po\
wave function
will have the
of a free electron in the
x representation, normalized
in
terms
of
fc
(p'
form [see
(4.81)]
To transform
to the
p representation, we use
Eq.
(7. 10) in
the
*'
form
(Po,
P)
'
-~^
*(/>
x')
<f
dx',
(8.50)
obtaining
T (Po, P)
= *(P
Po)'
The normalization condition has
the
same form
in both
and
p representations:
=
The average value
Po of the
V p*(pi,
f >
operators should be calculated from the equations

Po
,
+ AP
09 x) Mi]/ (p dp'fy* (p'
+ AP
j _ Ap
dpi
cp*
M=
J
Po
x)
dx
=
p
(p' w
p)

In the three-dimensional case,
_ .^
1
(Po,
P)
(P
Po).
a ^Problem 8.2.jDeterrmne the probability of the various values of the momentum of parfrllu in Un ground state, the particle being in a one-dimensional square well with infinitely high walls. Verify the normalization in the p presentation. Solution. Taking the value of the wave function from Chapter 4
and using Eq.
(8.50),
we have
~~
~7T
x
dx.
xe
Evaluating the integral and squaring

distribution
its
modulus, we obtain the required probability
which satisfies the normalization condition
oo
In evaluating the last integral
we may use
the relation
+ 00
ap dp __ n
sin
\a\b
'
b*-p* Jcos
oo
'b
which should then be differentiated with respect to the parameter
b.
^Jroblemj^jj^lnvestigate the motion of a charged particle in a constant and uniform

Solution. This problem is solved most simply in the momentum representation. Since, according to Eq. (7. 1 1), the potential energy in momentum space can be represented in
the
form
the corresponding Schrbdinger equation in
momentum space becomes
The
solution of this equation is

( EP
cp
(E, p)
Y^np
L=
"
where, because of the continuity of the spectrum, the normalization coefficient was found
from
the condition for 8-function normalization:
('-/:).

The wave function
in the position
139
space can be determined with the help of Eq. (7.9)
where
oo
[~
is
5)ef
the
Airy
5),
function, which
is
proportional to the Bessel function of order 1/3 (see

*
.
Chapter
function
and
= x+
f
f-*|~)
Examining the asymptotic behavior
of the
Airy
'
1
for
<
0,
-sin
l
/4
it is
sents
readily shown that the region of large negative values of a potential barrier, whereas the region where E
>
jc,
where
Fx> E,
repre-
Fx
is quasi-classical,
because
1
= -i-
/?</A"
=~
2^
f /2/H
+ ^) dx + const =
electron moves in a constant and uniform magnetic field. Find the time 'SSrWatwe of the average value of the position and momentum of the electron (in other words, generalize the Ehrenfest theorem for the case of motion in a magnetic field). Solution, According to Eq. (5. 9 a), the Hamiltonian of an electron in a magnetic field is
Choosing the direction of the uniform magnetic field to be along the z axis (Hz 0), the field can be specified by the vector potential 7/^
^ O,//^
In order to determine the time rate of change of the electron's position, quantum-mechanical equation of motion
let us
use the
= l{Hr-rH}.
Substituting H,
we
readily find
v==
For
the time derivative of the
J_(p W \
_.!)>
C
j
MQ
operator,
(8.51)
x component of
the
momentum
we
obtain
- vy" ~ n (KP* -P ^-i/KP A \H F* H)H fo py --A "* -~v M] dt m.c y)

\
s'
140
Similarly,
it
NONRELAT1V1STIC QUANTUM MECHANICS

can be shown that
dt
*'
dt
Combining these equations, we have
^=L
i
[y//].
(8.52)
Equations (8.51) and (8.52) constitute the required result.
8.5J As we know, the behavior of an electron in a metal (x 0) can be described wifn a sufficient degree of accuracy by the following potential energy function (see Chapter 6):
<
Determine the coefficient of reflection from the surface of the metal for electrons located inside the metal (x VQ Show that, even 0) in the following cases: (a) E 0), ultimately they return back into the metal. Construct a graph of the change of the potential energy and of the wave function of the moving electrons.
<
>
Answer.
a)
ForE < V0l /?==!, even though
ty
(jc>
0)
0.
Hint. In choosing the solution for x (outside the metal), only the exponentially decreasing solution should be retained in case (a), and only the solution corresponding to a wave traveling along the x axis in case (b).
>
Chapter 9
Elementary Theory of Radiation

A.
SPONTANEOUS AND INDUCED TRANSITIONS
According to classical electrodynamics, an accelerated charge a source of electromagnetic radiation. The amount of energy radiated per unit time is given by the well-known equation
is
1
where r
If
is the acceleration of the particle. the source of radiation is a one-dimensional harmonic oscil-
=w
lator
acosu>t
(9.2)
the frequency of the emitted radiation is the same as the mechanical frequency of vibration of the oscillator, and its intensity is propor2 tional to a
.
In the case
where
the
motion of a charge
is
governed by a more
T
complicated periodic function x
f(t)
with a period
we can
expand the function
f(t)
in a
Fourier series:
=ya
u
cos coW,
(9.2a)
and treat the radiation as if it were generated by a set of oscillators with frequencies (/, Radiation will be where k=\ 2, 3, ... ku>, emitted both at the fundamental frequency (k= 1) and at harmonics feo of the fundamental frequency. The intensity corresponding to the kth harmonic will be proportional to aj. Thus, according to classical theory, the radiation of a system is completely determined by its mechanical properties. Indeed, the
In this chapter the quantum-mechanical averages will be distinguished from time averages by writing the latter with the subscript "av * In accordance with the previous notation, quantum-mechanical averages will be indicated by a bar.
142
frequency of the radiation is either equal to or is a multiple of the mechanical frequency of oscillation of the system, and the intensity of the corresponding harmonic is proportional to the square of the amplitude. In quantum mechanics, the problem of radiation must be approached in a somewhat different manner. According to quantum theory, radiation is emitted only when a particle (or a system) makes a transition from one energy state to a lower energy state
(so-called
"downward"
transition).
The
first
proposed in B (now called the Eins tein coefficients) to characterize the spontaneous transitions and the induced transitions (that is, transitions due to some external effects) of a system from one energy level to another; Einstein also obtained an equation relating these two
coefficients.
quantum treatment of the problem of radiation was 1917 by Einstein. He introduced the coefficients A and
of the quantum theory of radiation are the Suppose one of the electrons of an arbitrary atomic system is in the excited state n with an energy . Then there is a definite probability A nfl per unit time of a spontaneous transition of this electron into a lower energy state n' with an energy E n >. The transition is accompanied by the emission of a photon with an E n E n >. If the number of excited atoms is equal toN n energy ftu the energy radiated per unit time during spontaneous transitions only can be written as
The basic elements
following.
'
iv/spon
"em
KJ "nAnn'nu.
4.
/q \V.3)
ov
When the atoms are subjected to the influence of external electromagnetic radiation, the latter will cause both upward and downward
f
Ann'
induced transitions. The up ward transitions course, be associated with the absorption of photons. Adopting the notation introduced by Einstein, we designate the probabilities of an induced transition from level n to n' by B nn and from level n' to n by Since the number of induced transitions should be
will, of
>
'
Fig. 9. 1. Downward transitions (spontaneous and
induced)
and
upward
proportional to the spectral energy density Pa of the external radiation, we obtain the following equations for the energy radiated and absorbed per unit time in induced
transitions:
vind
(9.4)
transitions (induced)
em
(9.5)
where
Nn
>
is
the
number
of
atoms
in state
n'.
ELEMENTARY THEORY OF RADIATION

Let us consider the case
in
143
which the number of upward and

(see Fig. 9.1):
nn >
n
downward transitions
is
the
same
n
NnA nn'
+ N ^B = N ^B
n ' nt
(9.6)
that is, when a state of the rmodynamic equilibrium exist between the heated atoms and the light radiated by them (black-body radiation), which in turn interacts with the atoms. In this state, the atoms and the radiated light form a closed system. Since, in this case, the energy distribution of the electrons is given by the Maxwell distribution
we obtain
Ann'CKn'* T
+ ^nn^ En" = ?B
T
n ne
.
~E
n'^
(9.7)
E Dividing by the factor e- n /kl and noting that E n
En >~ft(*> we
t
obtain
Since the spectral energy distribution of black-body radiation is completely independent of the specific structure of the atoms or molecules involved, Eq. (9.8) is essentially the same as Planck's formula [see (1.42)]
pw
= _*l".L_^_.
find
n n
>
(9 .9)
omparing
(9.8)
with (9.9)
we
.
B an
It is
= B = -^-
Ann'.
(9. 10)
seen from Eq. (9.10) that the probability coefficients of upward and downward induced transitions are equal to each other and proportional to the coefficient of spontaneous trans ition/4 nn Therefore, to describe the radiation of atoms or molecules, it is sufficient to determine only one of these coefficients.
'.
B.
CALCULATION OF PROBABILITIES OF SPONTANEOUS AND INDUCED TRANSITIONS
of an
quantum mechanics induced transitions are explained in terms interaction between the electrons of an atom and external electromagnetic radiation. The problem of determining the causes of spontaneous transitions was left unexplained by the Schrodinger
In
theory.
144
The answer was obtained only

of radiation
in
after the development of a theory which quantization of the electromagnetic field
(second quantization) was used. The general features of the theory are outlined below. Electrons interact not only with real photons, but also with virtual photons (photons which are in an unobservable state) or, as they are called, vacuum fluctuations of the electromagnetic field (for further details on vacuum fluctuations, see Chapter 22). This interaction causes spontaneous transitions. The classical analog of the interaction between the electrons and the field of virtual photons is the effect of Planck's radiation damping on a moving electron
which represents the self-interaction of the electron with its own electromagnetic field. Under certain conditions this electromagnetic field may detach itself from the electron in the form of electromagnetic radiation. In the language of quantum electrodynamics this amounts to a transition of photons from a virtual state into a
real state. The exact expression for the coefficients A and B can be found on the basis of quantum electrodynamics and, therefore, problems 2 of radiation can be completely solved. In the present discussion we shall obtain coefficient A by means of an appropriate generalization of the results of classical radiation theory to the quantum case. It should be emphasized that this generalization leads to the same results as the rigorous method of
second quantization. In our derivation we shall use the correspondence principle to generalize the classical expression for the radiated energy [Eq.
(9.1)] to
the quantum case. First,
we replace the
classical variable
r by the quantum-mechanical quantity
(9,11)
In addition,
we use an expression
for the radiated energy which is
consistent with the quantum theory:
where the coefficients gn and gn characterize the occupancy of

f
states n and
2
n'
by electrons, since according to the Pauli exclusion

is obtained
See also Chapter 29, where the coefficient A
by the methods of quantum
electrodynamics.

principle
it
145
is
quantum state
Chapter
24).
(for
impossible for two electrons to be in the same more on the Pauli exclusion principle, see
(9.12)
Combining Eqs.
and
(9.1)
and substituting (9.11) we obtain

(9.13)
Let us note that Eq. (9.13) contains two averages. One is the quantum -mechanical average, denoted by a bar, and the other is the time average, denoted by the subscript "av." We shall now assume that the electron has only two possible states with energies E n and E n Then the wave function can be
>.
written as
y(t)=C n e~
The average (over
radius vector
r
is
V <M
C n>e~
**"''
<[v.
(9.14)
the quantum* mechanical states) value of the
C n !'/
Cn
|*
rn
n.
+ CJC^-'/w + C^C e^r ^

n
(9. 15)
where
The matrix elements
form a certain
infinite
matrix
(9.17)
From (9.16), it follows that this matrix changes into its complex conjugate when the rows are replaced by columns, and columns by rows
Matrices satisfying this condition are called Hermitian or selfadjoint matrices. Let us also emphasize that the matrix elements (9.16) are independent of time, and therefore substitution of (9.15)
into (9.13) yields
gngn>fKAnn>
Cn
2
|
Cn
|*
rn 'n
|*.
(9-18)
146

the
fact that the
Here we have used

function is zero, since
time average of a periodic
For further analys is of Eq. (9. 18) we must introduce an additional assumption, which can be rigorously justified only on the basis of quantum electrodynamics. As we already know, quantum mechanics deals with stationary processes and, therefore, there is no am,
l const as the probability biguity in interpreting the quantity \C n \ of finding an electron in the state n. When the emission of radiation is present, the coefficients C n change dis continuously and their physical meaning cannot be simply explained within the usual formalism cf quantum mechanics. We shall, therefore, base our
conclusions on simple physical considerations, which are rigorously proved only in quantum mechanics. Let us substitute the initial values for the coefficients CJ into Eq. (9.18), bearing in mind that the Pauli exclusion principle allows transition only in the case when the quantum state n is initially occupied, while the quantum state n' is empty. Then setting
8*8n'
we
find
that
for
= \Cn\*\C = \C' d = and = the

n .\*
n \*(\
n \C*
\*),
(9.19)
1.
C,i'
product gngn*=
Hence
=B = r 'Wm = HA = y '-Ban
*
.
n .n
n .n \\
(9.21)
\\
nn >
/-
(9. 22)
In these equations
\rn >nl*
= \Xn'n\*+\yn>n\* +
\Zn'n\\
(9.23)
where
and so forth. Thus, the energy eigenvalues can be used to find the frequency of the radiation and the eigenfunctions to find its intensity. Thus, all basic classical radiation properties can be completely generalized to the quantum case by means of the Schrodinger equation. From the last equation above it is evident that the intensity of radiation will be different from zero only for those transitions for

which
zero.
at least
147
>
one of the matrix elements x nn

in
>
ij nn
and
z nn
>
is
non-
These transitions are called

in
quantum mechanics the
allowed transitions. It should be noted that
very many quantum-mechanical
sufficient to calculate the matrix elements alone and thus to set up selection rules, thai is, to find the changes in quantum numbers that correspond to allowed transitions. From a knowledge of the selection rules, one can answer the question of possible frequencies of radiation. In the language of classical electrodynamics, the selection rules correspond to a specification of the harmonics at which radiation can be emitted by a given system. If the matrix elements for a given change (difference) in quantum numbers are equal to zero, there will be no radiation at the corresponding frequencies, and these transitions are said to be forbidden. Here, in speaking oi forbidden transitions, we are restricting the use of this term to electric dipole transitions. By electric dipole transitions we simply mean transitions whose probability depends
problems
it is
on matrix elements
are also cases of quadof higher orders, and transitions transitions, rupole multipole magnetic dipole transitions. The intensity of these transitions turns out to be much smaller than that of the allowed dipole transitions. As an example, if the intensity of an electric dipole transition is of the order
In addition to the dipole transitions, there
w w where
dipole
=yp ~
2
2
<*<
73-
(
\
V2 ea )
a is the linear dimension of the atom, then the intensity of 3 electric quadrupole radiation is of the order
Vad
For an atom
8
~ ^dipole
(f )*
<
9 ' 24 >
5 cmand, therefore, the intensity of dipole radiation is 10 times greater than that of quadrupole radiation. Nevertheless, quadrupole radiation plays a very important part in a number of phenomena. Indeed, if the electric dipole transition is forbidden, it is still possible that a weak quadrupole radiation will be emitted, which can be detected with a very sensitive spectroscope. We note that no dipole radiation occurs in a system consisting of particles having the same charge to mass ratio. The
a~ 10 cm and X~ 10
This subject is treated more fully [see Problem (10.4)].
for the
case of
harmonic oscillator
in
Chapter 10
148
electric dipole moment of such a system is proportional to the r2 ) coordinate of the center of mass /^ -m- [p e(r mj, and therefore the derivative of the dipole moment with respect to time vanishes. This is the situation that should hold for gravitational radiation, since the gravitational charge, or rather the gravitational mass, is proportional to the inert mass m u . Therefore, if gravitational radiation does exist at all, it can only be of quadrupole character. Quadrupole radiation is also of importance in nuclear physics since the charged particles of the nucleus (protons) have the same charge and mass. 4
l
+ = 2^
Solution.
In
Find the probability of quadrupole radiation in the quantum case as a classical formula by applying the corresponding principle. the classical case the intensity of quadrupole radiation is given by the
equation
where
the quadrupole
moment
is
D ab = e
generalize (9.25) to the
(Zx a x b
r^ ab )
(a,
1,2,3).
To quantum case, it is necessary to consider that, in quantum theory, radiation occurs as a result of a transition of the system from one quantum state n to another, /t'. Following the procedure similar to the derivation of Eqs. (9.13)-(9.18), we first replace the classical expression for the quadrupole moment ab by the matrix
element
(Dn'n) a b=
Next, using Eq. (9.3), which relates the intensity of radiation probability A n nt we obtain
,
Wn n
>
to the
emission
A nn
where the frequency
9. 2. i
>
=
o>
^"^ (/Waft
by Eq.
Wn'Jab,
(9.26)
of radiation
is given
(9. 15a).
s~~~~*\ Find the selection rules for dipole and quadrupole radiation for a parade ^Problem
anmniffWtytreep potential well. Answer. For dipole radiation A/I must be an odd number, and for quadrupole radiation A n is an even number. Hint. Using the wave function (4.32), it can be shown that the average value of the x
coordinate is
in
Therefore, the matrix elements corresponding to dipole radiation should be calculated from the equation
(/=
1)
and quadrupole
(j
= 2)
See
Hcitler,
The Quantum Theory
of Radiation, 3rd Ed.,
Now
York: Oxford Univer-
sity Press, 1954
Chapter 10
The Linear Harmonic Oscillatoi

The problem of the linear harmonic oscillator is one of the most important problems of theoretical physics. It arises in connection with the construction of a simple thoery of vibrations in many different branches of physics (mechanics, classical electrodynamics, radio physics, optics, atomic physics, and so on). Most of the new theories developed in atomic physics have been "tested" in a number of simple problems, including, in particular, the problem
harmonic oscillator. The motion of a complex system can often be analyzed in terms of the normal modes of vibrations that are equivalent to the vibrations of harmonic oscillators. For us the harmonic oscillator presents additional interest for methodological reasons because this problem has an exact solution and thus provides a simple illustration of an application of the Schrodinger equation. The harmonic oscillator problem was also very useful in the development of quantum field theory (second quantization) and in the analysis of
of the
tions). In
the zero-point energy of the electromagnetic field (vacuum fluctuaquantum mechanics, the harmonic oscillator problem has direct applications in regard to the black-body radiation, molecular spectra, and the specific heats of diatomic molecules.
A.
THE OSCILLATOR
IN
THE BOHR THEORY
We shall first consider the classical theory of the harmonic oscillator. Let us assume that a certain particle of mass rnQ is acted upon by an elastic force
1
F=-kx,
where k is the elastic constant. Then harmonic oscillator can be written in
(10.1)
the equation of motion for the the form

t
m,x=kx
*In this section
(10.2)
which describes an ordinary oscillatory process.

simplicity,
lator."
we
we shall be concerned with the case of one-dimensional motion and, for shall use the term "harmonic oscillator* rather than "linear harmonic oscil-
150
The solution
of this equation has the
form
(10.3)
where
(o
-*I
= l/
f
ttlQ
is the
angular frequency and a is the ampliit
tude of oscillation. the acceleration
From
Eq. (10.3),
2
follows, in particular, that
w = x=
ao/ cos
co/
(10.4)
differs from zero and, consequently, that the oscillation of a charged particle will be accompanied by radiation, the intensity of which (that is, the radiant energy) is given by the following equation in accordance with Eqs. (10.4) and (2.2):
W
In deriving (10.5), the equation
w/
cl
1e-3-3
Way = -3-T2
s
,
...
fl
o>
/-i
[l n -5 > v
we
calculated the average value of

T
cosV from
Y.
(10.6)
We
shall
total
energy E
known
the intensity of radiation c\ in terms of the of the harmonic oscillator. From the wellequations for the potential energy
now express
= T-{-V
(10.7)
and kinetic energy
T = -? 2
of a
'"" "* a *-
sin
(10.8)
harmonic oscillator, we find

(10.9)
With this equation, we can eliminate the quantity a from

1
(10.5),
obtaining
Thus, on the basis of classical theory one can determine both the intensity and frequency of the radiation; it is also found that this
THE LINEAR HARMONIC OSCILLATOR frequency

the
is the
151
the frequency of mechanical vibrations of of the harmonic oscillator, according to the classical theory, can have any value in a continuous
same as
harmonic oscillator. The energy
range from zero to infinity. Several new features were introduced in the problem of the harmonic oscillator by the Bohr quantum theory. For example, according to Bohr's theory, the energy levels had to. be discrete and could be found from the quantization rule
jx = 2xfin,
where
pA
(10.11)
=g
r
=Wu
e.
(10.12)
Let us substitute p^dx
= m^ -~
En
dt
()
coV
siii*>M/ into
Eq. (10.11).
Then, taking into account Eq. (10.9) and integrating over a complete period, we find
nhv>,
(10.13)
.
where the quantum number = 0, 1, 2, 3, ... We showed above that, according to Bohr's theory, the energy of a harmonic oscillator can take only discrete values, and radiation will be emitted only when the oscillator makes a transition from
one energy level to another. The discovery of a discrete spectrum of energy levels of a harmonic oscillator played an important part in the theory of black-body radiation. Planck's law was first obtained under the assumption that the harmonic oscillator could radiate and absorb light only in the form of discrete quanta of energy /h.
B.
EIGEN FUNCTIONS AND EIGENVALUES OF THE ENERGY
In order to determine the behavior of the wave function in the harmonic oscillator problem, let us first give a graphical representation (Fig. 10.1) of the dependence of the potential energy V on x
i/ --.
the graph, it is seen that inside the potential well, where the energy E of the harmonic oscillator is greater than V (E V), the solutions for will take the form of harmonic functions. Inside the potential barrier (<V) the solutions will contain two parts,
total
From
>
<f
152
one exponentially decreasing and the other exponentially increasing (see Fig. 10.1). It is clear that the solution of the problem reduces to finding the conditions under which there is no exponentially increasing solution. Just as in the case of a rectangular potential well with infinitely high walls (Chapter 4), such levels exist only at certain discrete values of the energy, which we must determine.
Fig. 10.1.
Wave function
of
the
harmonic
oscillator
for
an
arbitrary value of the energy.
Since the potential energy V of a harmonic oscillator depends only on the x coordinate, the Schrbdinger equation can be written as
(10.14)
Setting
B
,\'n
and introducing a new variable

(10.15)
we
obtain
(10.16)
where
(10.17)
at
First, let us find the asymptotic behavior of the wave function co, that is, when the constant X is negligible in comparison
2 .
with
Then
k
%=
0.
(10.18)
THE LINEAR HARMONIC OSCILLATOR
153
We
shall seek a solution of this equation in the

*w
form
(
= ^'.
2e) e**
10 - 19 )
Since
^=
we
find
(4e*P
^ 4e EV'
2
9
,
(10.20)
and, consequently,
Ci
Since the wave function must remain finite at vr*:oo, coefficient must be set equal to zero. Coefficient C4 can be taken to be equal to unity, since the wave function has not yet been normalized. Thus, the asymptotic behavior of the wave function is described by the function
<|i
y 00
= c- '&.
i
(10.21a)
in the
We
shall seek a solution of the

ty
wave function
l
general form
(10.22)
= =
ty 00
= e- '*'u,
+ (P
:
which already takes in account the behavior (10.22) into (10.16) and considering that
l
at infinity. Substituting
(e-
i&
it)"
[//'
2;w'
u]
e~
'**,
we
obtain the following equations for u
K"-2te'
+ (X_l)tt =
0.
(10.23)
Let us look for a solution of this equation in the form of a series
Substituting
terms with the same power
this expression for u into Eq. of 6 , we find
(10.23)
and collecting
Equating
the
coefficients
of
bk
5*
to
zero,
we obtain
a recursion
formula for the coefficients
This formula relates the coefficients b k to 6 A+9 and, therefore, the series (10.24) will consist of even powers (if the minimum subscript k is even) or odd powers (if the minimum subscript k is odd).
,
154

If
the series (10,24) does not terminate at a certain
maximum
power, then beginning with /z>-^-, every term
is positive and,
consequently, the series diverges for large values of ; . This leads to the second asymptotic solution ê l/i^ at ; co which we <tasym 2 it diverges. Therefore, in order for disregarded earlier because Oat the boundary conditions to be satisfied (<); <.o), we must terminate the series (10.24) at a certain km&K ==n. We thus require
,
*W = 0.
From
(10.26) and (10.25),
(10.26)
we
find
= 2n+l,
+ -J
10 - 27 >
and, consequently,
n== /ho(/z
(10.28)
where n can assume any positive integral value, including zero. These are the only energy values for which the wave function
vanishes at
theory [see
infinity.
Comparing
this expression with the one obtained from the Bohr (10.13)], we note the appearance of a term called the
zero -point energy

E,
=~
A<D.
(10.29)
Later, we shall show that the existence of the zero- point energy related to the uncertainty principle and thus to the wave properties of particles. The zero-point energy does not affect the frequency of the radiation, however, since it cancels out in the expression
is
for the frequency
o> nn
/l
n .
--^
Let us now find the wave function of the harmonic The recursion formula (10.25) for the coefficients b k takes the form
-
oscillator.
fe/2 is the
This follows from the fact that at large k the ratio of the coefficients same as for the series expansion of the function e
Therefore,
155
where
/e</i
3
Setting the coefficient of the highest

bn
power
Ar
max =--=/
(10.30)
equal to
= 2*
we
obtain
The power series with a finite number of terms obtained for the function u is called the Hermite polynomial
u
i
-H
/l
n (5)
)
(2
Q
-
- ^fl>
.
(25)-*
+
n
.."(^~ .H
yl
~ 2 )( ~ 3
/2V-4
1-
-f
for odd
& for
(10.32)
even n
In particular,
ffo
(*)=!,
Wi(5)
// 3 (c)
= =
25,
3
(?)
-2,
(10.33)
8E
125
The Hermite polynomials H n (l) can be written
in closed
form
//,)=-i^.
This coefficient can always be chosen
the
arbitrarily,
do.34)
since the normalization factor of

*
,
wave function i// is still undetermined To show this, we introduce the function
v
\-
which satisfies the equation
2fv
--
Differentiating this equation
n +
times, and using the Leibnitz formula
we
obtain
Making the substitution
we
find that the function
satisfies Eq. (10 35), and thus
it
is proportional to the
Hermite
polynomial
The
result,
it
proportionality factor
is
found that
An -
(-l)
A n can be found by equating n from which we obtain Eq

,
the coefficients of (10 34).
2n
As
156
From
vided
A
= 2/i-f-l
(10.32)
it
is
clear that //(;) satisfies Eq. (10.23) proQ
(10.35)
According
to (10.22)
and (10.32), the solution of the Schrbdinger
equation for a harmonic oscillator is

ty n
=C
ne
l/
**H n (l)>
(10.36)
where
ficient
is related to the
Cn
coordinate x by Eq. (10.15). The coefcan be determined from the normalization condition
-Ho
^
00
;i
(/v
-|^cn
.v
Cl
ij
e-^H
tt
(;) //
(?) </;
1.
(10.37)
-03
//(;),
Substituting the we obtain
closed form
(10.34)
for one of the polynomials
rf
(-lyjCoCS f 7/n
00
(5)
dE
1.
(10.38)
Using the rule for transferring the derivative of one function to another [see (7.17)] (that is, we integrate by parts n times), we
obtain
"
f
"
r~ 4 *
dn
%n<k=
1-
(10.39)
Noting that from (10.32)
= 2"/z! ~//n(6) *i
i
(10.40)
and
H^co
^
\\,
(10.41)
00
ê-^=
we
find
2 n /l!
|/'
V
that is,
A'
ao.42)
157
In a similar manner, we can easily prove the orthogonality condition for the wave functions. To do this, it is enough to represent one of the Hermite polynomials, specifically, the one with larger n, in closed form (10.34). The orthogonality condition also follows from the general investigation of the Schrbdinger equation; it can be proved that the eigenfunctions corresponding to the different eigen-
values are orthogonal.
Fig. 10.2. The energy eigenvalues and the behavior of the corresponding eigenfunctions of the harmonic oscillator for small quantum
numbers
classical
(n
0,
1,
2)
For comparison, the
probability distribution functions p n are indicated by the dotted lines
In the
case of a harmonic oscillator, the orthonormality condi-
tion is
+00
\
tyn'tyndx
=
0,
&/m'.
(10.43)
oo
For small quantum numbers
/i
1,
2,
..
when
-,
(10.44)
158

distribution functions
\ty n \*
the probability
(see Fig. 10.2) differ
considerably from the corresponding classical probability functions.

In the classical case the probability of a particle being at a certain point is proportional to the amount of time the particle spends there, and, consequently, is inversely proportional to the particle velocity; therefore, the classical probability
is
proportional to
(a
r )2
^,
where
a is the
maximum
displacement of
a classical oscillator
from the equi-
"
librium position. As we would expect, it is only for large quantum numbers that there is a relatively close agreement, on the average, between the quantum and classical
probabilities (see Fig. 10.3).
C.
Fig. 10.3. Comparison of quantum and classical results for the oscillator in the region of large
ZERO- POINT ENERGY OF THE HARMONIC OSCILLATOR AND THE UNCERTAINTY PRINCIPLE
We have seen that in quantum mechanics the minimum energy of the harmonic oscillator is given by Eq. (10.29) and cannot go to zero, whereas in the classical theory or the Bohr theory, the minimum energy is equal to zero. We shall now show that the existence of the zero- point energy (10.29) in the Schrjjdinêr Jtheory j[gâs mentioned above, very to thq unQprtflfnty principle (8.11). For the "case of a harmonic oscillator the uncertainty principle becomes
numbers (here n 10).
*
quantum
(10.45)
This can be easily shown on the basis of the following simple qualitative considerThe probability of finding a particle at a particular point can be roughly characterized by the absolute value of the reciprocal of its velocity, since the time a particle spends in a region will be greater in regions where the velocity is smaller than in regions where it is greater. Consequently, the probability of finding a particle in a region with larger velocities will be smaller than that for a region with smaller velocities. In the case of a harmonic oscillator, we have from Eq. (10.3) x/a - cos cot and t/eaa - sin a>t.
ations.
'
Taking the square of both equations and adding them, we get

1
and, therefore,

2
159
Here we have replaced (AT) by 'x* and (A/7y by ~p\ This is justified by the fact that the wave functions are real and are either even or
2
odd. Indeed, since the expression
fy*xty
xfy*
is odd,
we have
x==
f
<|>
**<!*/*
0.
Hence
Similarly, using the boundary conditions at infinity,

n
we
find
*"=<>, 00
that is,
Substituting the value of the total energy
/?
from
(10.45) into the equation for
= tf=Jl +
we
obtain
(10.46)
^T^ + -^From
this
<
10 ' 47 >
it is seen that the energy E cannot vanish at_any value r Q, the Indeed, although the second term vanishes for (x ) first term becomes infinite. Conversely, when (**)== oo, the first term vanishes and the second becomes infinite. Thus, the fact that mln differs from zero is directly connected with the uncertainty relation (10.45) or, in other words, with the fact that it is impossible to calculate exactly the position and
of (* 4 ).
momentum
mum.
at which Eq. (10.47) has a mini) the derivative of this function with respect to (P) equal to zero, we obtain
(x
simultaneously. Let us find that value of

Setting
or
160
Substituting this into (10.47),
we have
.
E^~-\-~ =
Hence,
min
(10.48)
=Y
which
is
exactly the
same as the value
for
"0
found
from the wave theory [see (10.29)]. The existence of a finite zero-point energy of the harmonic oscillator is one of the most characteristic manifestations of the wave properties of particles. Thus, the experimental verification 3f the zero-point vibrations was of great significance for quantum was first observed experimechanics. The zero-point energy
mentally in the scattering of x-rays by crystals at low temperatures. If there were no lattice vibrations at low temperatures ( 0), as the for Bohr no interthere would be example, by predicted, theory, action between the x-rays and the crystal lattice, and consequently no scattering would occur. If, on the other hand, the minimum energy were different from zero ( ~f- 0) for T-^0, the scattering cross section at low temperatures should approach a finite limit.
Experiments have confirmed

Schrbdinger wave theory are
that the
second situation corresponds

conclusions of the
to the true state of affairs and, therefore, the
justified.
D.
SELECTION RULES. INTENSITY OF RADIATION
Let us consider the problem of radiation from the harmonic oscillator on the basis of wave mechanics. For this purpose, as was indicated in Chapter 9 [see (9.22)], we must calculate the
matrix elements
Xn n
,
= J ^JC-MJC,
(10.49)
where <îs given by Eq.
(10.36).
we shall derive a recurrence relation for the Hermite polynomials, which will be necessary in our further discussion. From the definition (10.32) of the Hermite polynomials H n (), we find
As
a preliminary step,
H n (S) =2n
from which
it
follows that
,().
(10.50)
161
Next, substituting these equations for the derivatives into Eq. (10.35) and replacing n-+n -\- 1, we obtain a recurrence relation for the Hermite polynomials
r
Wn,(\)=n'H n ,-^) +
Hn>+\ ().
(10.51)
By means
to the
of this formula, the
matrix element (10.49) can be reduced

-foo
form
= xlC Cn>{
n
<x>
^//
we
r
In
terms
of the
wave functions
<|>,
obtain
V+l*ndX+n
^l
Since the functions
<l>
are orthonormal, we have

(10.53)
this expression it follows that the only nonvanishing matrix elements are those for which n' or n' n-}-\\ therefore, the selection rules for the quantum number n are expressed by the
From
= n\
f
equation
A/i
/i
/i
= -l,
(10.54)
which indicates that only transitions between neighboring levels are

possible. For the
wave function as given by Eq.
(10.42),
we
obtain
from
Eq. (10.53)
xn-\,n~ xb Y
V
.
(10.55)
where
theory:
(}
is
determined from Eq.

obtain exactly the
(10.15).
For the frequency
radiation,
we
same expression as in
of the the classical
'
n.n-l
(10.56)
162
NONRELAT1V1STIC QUANTUM MECHANICS
The energy levels and allowed transitions are shown

10.4.
in Fig.
Since spontaneous emission is possible only when transitions occur from higher to lower energy lev els (E n ^>E n .i) t it follows from
(9.22) that the intensity of radiation = nnl of the harmonic oscillator
W W
is
2
'
co
""
,t
-,..,
^_ ^
(1(h5?)
E /h
l''ig.
(10.10),
10 4.
Comparing this equation with Eq. which was obtained from the
theory,
Allowed transitions of the
classical
W6
S66 that for

f
harmonic oscillator
large quantum numbers (n *>\) when both equations yield practicEn

9
<
ally the same result. Transitions to higher energy levels n-+n+\ are possible in the case of induced transitions. The occurrence of spontaneous upward transitions is also possible under the condition that the energy loss in the harmonic oscillator is compensated by the simultaneous liberation of a large amount of energy, as for example, in transitions of atomic electrons (see Chapter 12, spectra of diatomic molecules).
KLjf Find the eigenvalues of the harmonic oscillator using the WKB method. wave function for the harmonic oscillator is
lution. According to (10.14), the
where
According
to Eq. (5.75), the eigenvalues
are determined from the equation
Evaluaung this integral, we find that the energy eigenvalues are the that is, the zero-point energy is also present.
same
as in (10.28);
^jr oblem 10. 2,1 Construct the theory of the harmonic oscillator in the p representation (forlne one-dimensional case). Find the equation of motion, the eigenvalues and the
eigenf unctions.
Solution.
Since x 2
n 2 --- in the p representation, we can write the Schrodlnger
(]2
equation as

that is, transforming the wave equation for the harmonic oscillator tion to the p representation and introducing the new parameters
163
from the
A"
representa-
where
Po
Ymolitot
we
find that the
wave equation changes

*"
identically into itself

9
)
+ (*- 1
=
>)).
(the prime indicates the derivative with respect to (10.42), we can write in the p representation
Using the solutions (10.28) and
and
P --( \ n -V
]
""\Po
It is
easily verified that this wave function satisfies the normalization condition
\f n (p)\
dp^l.
\Problem 10.3jFind the eigenfunctions and energy spectrum of an electron (e = e<> 0) uniform magnetic field. Show that, according to the quantum theory, the "electron gas" must be diamagnetic. Solution. Let the magnetic field be directed along the z axis (A/ v ffy= $Jl =z&W). We can then write for the components of the vector potential A v ~X(f,Av A 2 0. The motion of an electron is described by the Schrodinger equation (see also Problem 8.4)
<
LOvtTl^ irweOflsiant,
2ifT Q
Since the coordinates for a solution in the form
y and
z do not appear explicitly in this equation,
we
shall look
L
For
the function/^),
we
obtain the equation
where
Wl
It
cfik,
is
easily seen that this equation has the
same form as
Eq. (10.14) for the harmonic
oscillator.
Consequently, we can use solutions (10.27) and (10.36) to determine the eigenfunctions and eigenvalues. We thus find
164
where
Hn (;)
o
is the
Hermite polynomial,
-is the frequency of
Cn =
/y
I
Q
!
.v
a >an nnt j (n\)
is the
normalization
<
efficient,
Larmor precession, and
For the eigenvalues, we have
The last term in this equation is simply the kinetic energy of a free electron moving along the z axis, and is of no special interest.
The
first
term
n
= fV^T (2/1+1),
(10.59)
is the Bohr magneton, corresponds to the additional energy acquired by an fi electron in the magnetic field. This additional term represents the energy of electron motion in the xy plane, which is perpendicular to the magnetic field. This conclusion is in agreement with classical theory, according to which an electron placed in a magnetic field precesses with the Larmor frequency o in a plane perpendicular to the magnetic field. In the classical theory, however, the energy of an electron in a magnetic field is determined entirely by its unquantized kinetic energy. Therefore, according to the classical theory, an electron gas generally exhibits no diamagnetic properties. In the quantum theory the energy (10.59) can be interpreted in terms of the appearance of an additional magnetic moment ji of an electron, which makes the following contribution to the energy:
where
10.60)
Comparing (10.60) with
(10.59),
we
find
f^
M2/I+1).
1,
Since the
tional
number In + assumes only positive values (n =0, moment of an electron will be directed along the negative
I
2,
3,
.),
the addi-
z axis.
This naturally
to the diamagnetism of free electrons in a metal. should be noted that in quantum mechanics, solution (10.58) corresponds to harmonic vibrations along the z axis along, whereas in classical theory the circular trajectory means that there are harmonic vibrations along both x andy axes, with a phase difference
It
leads
The reason for

of-^r-.
this is that the
energy
is
independent of the
momentum fik 2 Con.
sequently, degeneracy occurs and the solution for a given energy has the form
(10.61)
where
the coefficients
Ck
are arbitrary amplitudes satisfying the normalization condition
In classical theory there is also an indeterminacy, since the center of the circular trajectory is not uniquely specified. The general solution (10.61) corresponds to a set of circular trajectories having different centers located along the y axis.
165
harmonic oscillations along both x and y axes. This can be seen by examining the expression for the energy [Eq. (10.59)] which "represents a sum of the energies of two harmonic
oscillators (note that
Concluding the above discussion, let us note that the solution (10.61; includes the
ilem 1QA Show that the matrix element of the product of two operators M (A) and which fife"' independent of quantum numbers, is equal to the sum of the products of the matrix elements of these operators, that is,
(A),
(10 .62)
k
Solution. Writing
)^ =
<[
(.v)
M (x) I (x - A') N (*')
ty n (A-')
dx dx'
and using the relation (see Chapter 4)
together with the fact that the operators bers, we readily prove Eq. (10.62).
(,v)
and N
(A-')
are independent of quantum
num-
feroblem^ 10.5J Find the selection rules for quadrupole radiation emitted by the harmOnlc" oscillator. Find the intensity of spontaneous quadrupole radiation and compare it with the intensity of dipole radiation. Obtain Eq. (9.24), which relates the intensity of quadrupole radiation to that of dipole radiation, In Problem 9.1 we found that the quadrupole radiation is proportional Solution. to the matrix element, which according the the preceding problem can be written as
elements
Substituting the values of x n k from Eq. (10.55), of the quadrupole radiation:

,
we
find the following nonvanishing
matrix
(10.63)
That
is, the selection rules for the
quadrupole radiation of the oscillator are
An
0,
it
2.
(9.26).
The probability of spontaneous emission (n' n 2, An For large quantum numbers when E == nh<* we have
t
=
L
2
2) is
calculated from Eq.
16
^
6
15
"'
mfc
According to (10.57), the intensity of dipole radiation
is
166
Using the last two equations, we obtain
where a 9
in
2E
//I
<o
3 a
is the
square of the classical amplitude of oscillations. This relation is
agreement with Eq.
(9.24).
^Problem 10.6J Show that the center of a wave packet composed of the solutions for the nUllllUUft! UUcTllator moves according to the laws of classical mechanics. Show that this wave packet does not spread with time. Obtain the transition to the v ;> l where 2v Us the number of waves in the wave packet). quasi-classical case (n Solution. Let us assume, for simplicity, that the wave packet is composed of 2v -f- 1 eigenf unctions of equal amplitudes
>
where
the eigenfunctions of the harmonic oscillator, and w ty n+ j are frequency of vibration. The coordinate x of the center of the wave packet is given by
is the
mechanical
with
2v
where a
It
= T/
f
^( 2/l
Wo
+ O.i s the classical amplitude of oscillations.

4
follows from the last two equations that for the harmonic oscillator
A:
obeys the classical equation of motion
Evaluating x* in a similar manner, deviation
we
obtain (A*) 2
=x
8
Jt
for the
mean-square
(Kx)*
C2
-~W( B*
}cos2c<
where
'
2"
27+~
Consequently, (A A:)* oscillates about a certain with time.
mean
value and, therefore, does not spread

In the quasi-classical case,
167
we have
1
2v+i;>
and thus
xc
a cos
8
,
(Ax)
c^s
=s.
const.
2v-j-
From this it follows that the larger the of the packet, and finally, for (2v -f 1) 1
number
f
of
>
waves^the smaller
will be the width
the width tends to zero.
Chapter
General Theory of Motion of

a
Particle
in
Centrally Symmetric Field
The problem of the motion of a particle in a central field of force (a field in which the potential depends on the distance alone, and not on the angles) is one of the standard problems of quantum mechanics. This problem provides a basis for the quantum theory of the rotator, which is of considerable importance in connection with the spectra of diatomic molecules, the theory of the hydrogen atom, the nonrelativistic theory of the deuteron, and so on. It is worth noting that in a central field of force the dependence of the wave function on the angles U and <p is completely unrelated to the
specific form of the potential energy. Accordingly, the spherical harmonics are of general validity; they are applicable to any centrally symmetric field. The classical analog of the quantummechanical investigation of the angular parts of the wave function
is the
in
derivation of the law of conservation of angular momentum central field of force. This law is also independent of the
specific
form
of the potential energy.
A.
SCHRODINGER'S EQUATION
IN
SPHERICAL
COORDINATES
The problem
force
of
the motion of a particle in a central field of
F=F(r)$,
is
(H.l)
and cp, which are usually solved in the spherical coordinates r, related to the Cartesian coordinates (see Fig. ll.l)by the equations
Jt
= pcos<p,
r/=psincp, z
= rcosfr,
p=rsinO.
(H.2)
We
shall
now write the Schrodinger equation
in spherical co-
ordinates.
First, using the general definition of the potential energy V as a quantity whose negative gradient is equal to the force /% we have
dV
(F-dr).
(11.3)
MOTION OF A PARTICLE
IN
A CENTRALLY SYMMETRIC FIELD
169
For the case
of central forces (11,1),
we
obtain
dV
and hence
-~(xdx
+ ydy + zdz) =
r
Fdr,
(ll.Sa)
V(r)
=J
F(r)dr,
(11.4)
where the lower limit of integration is chosen the convention that V (r) vanishes at infinity.
In particular,
if
in
accordance with
the central forces are due to
Coulomb
inter-
action
r=-"-r,
? is the charge of an electron where Z<? is the nuclear charge (e we obtain for the potential energy around the nucleus), moving
r
7/>a
-?4dr=-^-. J7/s
expression 2 Laplacian y in spherical coordinates. Using the identity
for the
(ll.Sa)
Now
let us find the
=vV
we
shall first find the
of a vector
(11.6)
components
(11-7)
in spherical coordinates. Bearing in mind that a gradient expresses the spatial rate of a change of scalar field in a certain
direction
Vf fy*= '^r)t obtain, in accordance with Fig. 11.1,
(/=
we
(11.8)
^.
11.1.
Spherical coordinates.
The volume
r sin
element in spherical coordinates.
Let us use the definition of divergence
(BdS)
V B=
lim
170
where d*x is
the

volume element in spherical coordinates
d*x
=r
sin
ft
dr d* dy
(11. 10)
ft and and dS t stands for elementary areas perpendic<p), (XL stands for the coordinates r, ular to the directions dr, rd$ and pflfy respectively:
t
dSy
= rdr db.
(11.11)
With the help of Eq. (11.8), we obtain
'
(g
<1U2)
from which we readily
ordinates
find the
expression for the Laplace operator in spherical co-
Setting in Eq. (11.13)

1
r8
and
we have
so that the Schrodinger equation (4.8) takes the form
where
*t
W = -^[fi-V
(r)]
(11-18)
is,
according
to
Eq. (11.4), a function of the radius
only.
B.
SEPARATION OF VARIABLES. EIGENFUNCTIONS
We shall solve Eq. (11.17) by the method of separation of variables. Let us represent the desired function as a product of the radial and angular parts
cp).
(11.19)
IN

/
171
Multiplying the original equation by l^pU
r2
we
obtain
_:*.
(n.20)
Since the left-hand side depends only on r and the right-hand side and <p, this equation can be satisfied only if only on the angles both the left- and right-hand sides are separately equal to a constant X, called the separation constant. We, therefore, obtain the following equations for the radial and
angular parts, respectively:
0.
(11.22)
The important point to note is that the angular part of the wave function does not contain the variable r and is independent of the specific form of the potential energy V . Consequently f as we mentioned at the beginning of this chapter, the angular solution will be valid for any central force. Using the method of separation of variables for the angular
part alone,
we
set
Y = &(*)$(<?),
(11.23)
and thus obtain the following equations for the functions & and CD:
(11.25)
Here m2
notation:
is the
separation constant and
we have used
the following
<
n 26
-
>
*}
(11-27)
where partial derivatives are replaced by total derivatives, since each of the functions 9 and O depends only on a single variable.
Thus, we have obtained three equations (11.21), (11.24) and and the corresponding eigen(11.25) for the energy eigenvalues functions <f, The last equation contains only a single parameter ni\ whereas the first and second contain two parameters each. Since the solution of one equation yields the eigenvalues for only one parameter, we must begin the solution of the entire problem by
/
172
NONRELATIV1STIC QUA fUM MECHANICS
solving (11.25); then, knowing

finally (11.21).
m w proceed
2
,
to solve (11.24)
and
To
find the normalization constant,

00
we use
sin ta
the relation
2lt
ty*ycPx=
^
b
R+Rr*dr
f
b
9*0
f
b
$*<M<p,
which shows that each of the functions can be normalized separately:

00
J/?*/?rdr=l,
f)
(11.28)
=
l.
l,
(11.29)
(11.30)
(11.25)]
The particular solution for the azimuthal function [see Eq. can be written in two ways:
Q = Ce im *
or

(11.31)
= A cos
(mcp
<p ).
(11.32)
Each of these solutions has a different physical interpretation. The solution (11.31) represents a wave traveling around the circumference of a circle and corresponds, for example, to uniform
circular motion of an electron. On the other hand, the solution (11.32) is associated with standing waves and corresponds, for example, to oscillations of an electron along an arc. In order for the function to describe the motion of an electron around the nucleus, it must have the form of traveling waves (11.31). Moreim * can be obtained from over, since a solution proportional to e the first solution by replacing (m) by (- m), we can take, without any loss of generality,
([>z=Ce tmf ,
(11.33)
where the quantity m assumes both positive and negative values. Since the wave function must be unique (see Chapter 4, Section B), the function $(?) must be periodic with a period
.
(11.34)
It
follows that
IN
173
and therefore the quantity m, which is called the magnetic quantum number, assumes only integral values
m = 0, From
is readily
drl, =t2, dt3
(11.35)
the normalization condition (11.30),
we
find
C=--=.
It
shown by direct calculation

1
that the functions
-*" f
(11.36)
satisfy the condition of orthonormality

2ie
Since we now know the eigenvalues associated with the azimuthal angle <p Eq. (11.24). Introducing the new variable
x
m and the wave we can proceed
function to solve
= cosO
to x
(11.37)
and denoting derivatives with respect
by primes, Eq. (11.24)
becomes
=0.
It
(11.38)
can be seen that (11.38) has singular points at * -!: 1, that is, points at which one of the coefficients of B becomes infinite. To eliminate this divergence, we shall look for a solution 9 in the form
=
Substituting
2
(!
--*)*/2 M
(11.39)
all
(11.39)
into
(11.38)
and dividing
the terms by
(1
r>)'/
we
obtain
u
Q.
(11.40)
We
eliminate the singularity in the last term by setting
Since the fundamental equation for H depends only on w the solutions corresponding to these two values of s both satisfy the
2
,
174
equation, and, consequently, there relationship between them:
same
must be a simple linear

(11.41)
9(m)
It
= 49(
m).
is,
therefore, sufficient to solve Eq. (11.40) for

s
= m^0.
(11.42)
With the help of Eq. (11.41), the solution can be automatically extended to the negative values of in. Under the condition (11.42), Eq. (11.40) becomes
(1
*)w"
2x(m+l)u' + (\
m(m+l))u = 0.
its
(11.43)
Since this equation has no singularities, represented as a polynomial
solution
may be
Substitution of this polynomial into Eq. (11.43) gives
Collecting the terms with the
same powers
of
x,
we
obtain
This yields a recurrence relation

(11.45)
which gives the relationship between the coefficients of the series (11.44). Since the coefficients a k ^ are expressed in terms of a k and thus only alternate terms are related, the function u will be either even or odd depending on whether the main term is even or odd. We require that the series (11.44) terminate at some maximum
power
q,
so that
Then from
(11.45)
we
obtain
+l),
(11.46)
(11.47)
where
(7
0,
1,
2, 3,...
(that is, q is
IN

at
175
Introducing
equal to the the orbital
power
angular
I
which the series is terminated). momentum quantum number I

(11.48)
q
=q+m
we find assume
that,
only
just like the numbers positive integral values

J
and m, this number can

zero),
that is,
(11.49)
(including
...
,
0,
1,
2,
3,
and from (11.48)
it
follows that
l^m.
According
to (11.48)
(11.50)
and
(11.46),
X
we have
/(/
l),
= +
f
(11.51)
and, therefore, Eq. (11.40) can be reduced to the

(1
form
I)]
i
^)a"
2jt(m+l)a
+ [/(/+!) m(m+
0,
(11.52)
where
.
(11.53)
Instead of determining the relationship between the coefficients a q and a qYl by means of the recurrence relation (11.45), let us represent the solution (11.53) in a closed form. For this, we introduce the
function
i;
= (**-!)',
*V + 2*fo =
0,
(11.54)
satisfying the equation

(1
(11.55)
which
is easily obtained by taking the first derivative of v with respect to x. Differentiating Eq. (11.55) with the help of Leibnitz's rule [see (10.34a)] and setting
we
obtain the following equation for the function u^
-m) ^=0.
(11.57)
176
We
note that this equation is exactly the same as the differential equation (11.52) for the function u. Consequently, functions u and other i/i must be proportional to each
w=const
been determined, we can set
to
2//I
MI.
(11.58)
Since the normalization coefficient of the function 6 has not yet this proportionality constant equal
in
order
to
make
the solution (11.58) f or m
identical with
the Legendre polynomial
We
thus obtain
from which, with
the help of Eq. expression for the function ft:
(11.39),
we
find the following
Here P?
equation
is
an associated Legendre polynomial defined by the
andC/ is the normalization coefficient. Although (11.61) was obtained for positive values of m, it can also be extended to include the negative values of m by using
the well-known relation
1
W.
<
n 62
-
>
*To prove Eq. (11

2
62),
we put
it
in the following form with the help of (11.61):

'
'"'
21
dx
+ ]ml
dx'~ {ml
(1163)
Since P/ and P/" must be linearly related to each other [see (11 41)1, it is sufficient for us to show that the coefficients of the leading power of x on both sides of Eq. (11.63) are equal to each other, that is,
"
This
is
easily
shown shown since
"TiT dX
__
("-*)'
fof
(11 64)
for
> n
IN
177
From
(11.61)
and
the quantum
number m
(11.62), is
it
follows that the range of variation of
m = 0,
since for
1,
dr2,...,/;
the solution P m vanishes. 2 m The coefficient C/ in (11.60)canbefoundfrom the normalization condition
|m|>
er
m sin
,
=
J
er
(*) e, (*)
d*
Substituting the solution (11.60) and using (11.62),
we
obtain
Transferring the derivative from the second factor in the integrand /-j-m times (that is, expanding the integral by parts /-{-m times), we obtain
to the first factor
Using the equation [see also
(11.64)]
and
x f \-*'dx= J
we
find
Because
of the linear relationship
between P/
m and
m
P/
,
many authors present the
solution for the function
in the form
m e - cpj '<*).
We
shall not use this form, since in this case the recurrence relation between the associated Legendre polynomials is more complicated than for the solution (11 60) (the recurrence relation is important in connection with the selection rules and the solution of the Dirac equation).
176
Then
ni 66) ; {ii.oo
For the spherical harmonic KF(d, 9), which satisfies Eq. (1.22), the relations (11.23), (11.36) and (11.66) yield
the
The orthonormality condition for the spherical harmonics takes form 3

?'Y YTdQ
mm ,
(11.68)
After having obtained the eigenvalues of the parameters m and to the solution of Eq. (11.21) for the radial part in which there remains only one unknown parameter. The radial solution, however, can be obtained only if the form of the potential energy V (r)is specified, and, therefore, we shall leave this question aside until we come to consider various specific forms of V (r) in the following chapters.
X,
we may proceed
C.
PHYSICAL MEANING OF THE QUANTUM NUMBERS

/
AND
m.
ANGULAR MOMENTUM
/
We
have found that the quantum number

X
eigenvalue /(/-f- l)of the operator -yi,? which is a part of the Hamiltonian
characterizes the
[see (11. 22) and (11.51)],
Comparing
this
Hamiltonian with the classical Hamiltonian function
V (r),
(11.70)
To prove the orthonormality condition (11 68), we substitute the expression (11.67) for the spherical harmonics into (11 68). Integrating over the angle <p, we can readily show that
2TT
Integrating the Legendre polynomials over the angled, loss of generality, we can take /' The case I'- I /.
we can set m' m. Then without was considered above in connecit
tion with the determination of the normalization coefficient. In similar fashion,
can be
readily shown that for l'<l the integral (11 68) vanishes as a result of transferring the derivative from the function with subscript / to the function with subscript /
.
IN
179
m Qr^ we see that the operator ( and L the corresponds square of the angular momentum t a in the classical case, and the operator ( AV) to the square of the radial
where pr
to
= mj,
p?.
momentum
Let us investigate these analogs in more detail. As we know from classical mechanics, the angular momentum L is defined as
L= rxp.
If
(11.71)
external forces
is
exert a torque
M=
r x F, the time rate of
change of L
given by
In the case of central forces (F\\r) 9 no torque
is
exerted and,
consequently,
= const.
law of conservaKepler's theory of planetary motion as the law of conservation of areal velocities (that is,
the law of areas).
In classical mechanics this result is known as the tion of angular momentum; it appears in
To generalize the classical expression for the angular momentum to the quantum case, we replace the classical momentum p in
(11.71)
by the momentum operator

L
=y
y-
We
then obtain
= rxp=yrxv
L x =yp z
zpy
,
(11.73)
or
L^zpt
We
xp,,
&
,
(11.74)
commute
note that the angular momentum operators L x L and L z do not y with each other. For example, by direct calculation of the commutation relation between LândL we find y
,
L x Ly
Ly L x
= (yp,
zpy ) (zp x
xp z )
(zp x
xp g ) (yp z
zpy ).
Using the commutation relation between the momenta and the corresponding coordinates [see (7.7) and (7.8)], we find
L x Ly
Similarly,
it
- L,L, = M (yp x - xpy = ML,.

)
(11.75)
can be shown that
180
To express
the square of the angular
momentum
operator
(11.77)
in spherical coordinates, we must first determine the components L x L and L z in these coordinates. Using the relations (11.2) between
,
Cartesian and spherical coordinates, we have

dty
"W
_ dx
cp
<ty
dr
,
dty
"dS ~i~ ~dy

-f/-
W
(?|
dy
dfy
dz
~dz
W~~
_
^-
r cos
cos
0|
cos d sin
.yz
cp
_ ~r~dx~r~Y~ty ~~*~fc dz _ _ dx_\<fy_ dy d^ ~x d

xz d^ j^
dfy
'
r sin
=
(11.78)
d<\>
"
dty_
<
~d<
'
<
^g-.
Multiplying Eq. (11.78) by
(11.79)
- and
Eq. (11.79)
p*
products, and remembering that
= x*-}-y*
by(^),
we
adding these
obtain the relation
(11.80)
Now
let
us multiply Eq. (11.78) by
and Eq. (11.79) by
r.
Then, proceeding in the same way as before,
we
obtain
Using Eqs. (11.79) and Eqs.
(11.74),
we
find
(11.82)
(11.83)
cos d (which should not be confused with Introducing the variable jc the Cartesian coordinate *), Eqs. (11.82) and (11.83) can be written in the form
dl.85)
IN
181
To determine the effect of these operators on the spherical harmonics, let us take advantage of the fact that a particular spherical harmonic can be represented either as (11.67) or as
Operating directly with L 2 on the spherical harmonic,
we
find
(11.87)
this it follows that the quantum number m characterizes the component of the angular momentum. To determine the effect of the operator L x -\-iLy on the spherical harmonic \f we use the expression (11.67), and for the effect of the operator L x iL v we use the equivalent expression (11.86). Then
,
From
from
the equation
it
follows that
(L, it iL y )
YT
H V(l
m) (/ ip m) YT
l
-
(11.88)
Equations (11.87) and (11.88) yield
X (L, +
The
last equation
Ly)
+ Lj] YT = i) yj*.
ft'yl. f
*T
=
(11.88a)
--^/(/.j-
shows that YT is an eigenfunction of the operators This follows from the fact that the operators L, and L a commute not only with each other, but also with the Hamiltonian H. Since the operators L x and L v do not commute with L, it is impossible to find a wave function that would be a simultaneous eigenfunction of the operator L z and the operators L x or Ly. This does not mean, however, that the direction of the z axis is a preferred
L 2 and LA
f
it can be chosen arbitrarily. The spherical harmonic can be written in such a manner that it will be an eigenfunction of the operators L^. and L 4 In this case,
direction, since
it
will no longer be an eigenfunction of the operator L z (see

12.2).
Prob-
lem
182

D.
ANALYSIS OF THE RESULTS
Quantum-mechanical results are generally analyzed either by finding their classical analogs or by comparison with the results of Bohr's semiclassical theory, which has a simple physical interpretation. To apply Bohr's theory to the motion of a central field, we start from the classical law of conservation of angular momentum; we then conclude that the motion takes place in a single plane and that the angular momentum vector (which is perpendicular to this plane) has the magnitude
L
= p =g = m /^ = const.
9
(11.89)
Applying the quantization rules, we find the discrete values that the angular momentum can assume
Hence
L^=n*H\
where
/i
(11.90)
=l,
2,
3 .....
(11.91)
If
the z
the
Bohr theory allows us

Then
momentum
tion.
axis is not perpendicular to the plane of the orbit, then to quantize the projection of the angular vector on the z axis. Tnis is known as space quantiza-
where
n^
It
n vt
/if+l,
....
.....
/i
1,
nr
(11.93)
follows from this that the angle a between the direction of the angular momentum L and the z axis is given by the equation
cosa
that is,
= ^;
nv
(11.94)
it can assume only certain discrete values. Space quantization is illustrated in Fig. 11.2, which shows that ^ /l 9 corresponds to the case when L is parallel to the z axis (Fig. 11.2a), where n^ corresponds to the case of antithe vectors are mutually parallel L (Fig. 11.2b). Finally, for /i^
\=
IN
183
perpendicular (Fig. 11. 2c). It is obvious that space quantization has a meaning only when there is some preferred direction in space, for 4 If example, the direction of the magnetic field intensity vector. there is no preferred direction, the orientation of the z axis may be taken as perpendicular to the plane of the orbit.
Fig. 11.2.
Space quantization according theory (for L - 1; in units of h)
to
the
Bohr
of
let us compare the quantum-mechanical results with those Bohr theory for the square and the components of the angular momenta:
Now
the
-z
/i= l,
2,
3,
4, ...
It
is
seen that
L'
qm
is
zero when
Q has no classical zero. This means t.h^ thp. gfato ^yjth / analog. It follows* that the angular 'momentum atom in tke lowest ot'*ftn state is "equal, to zero, contrary to me results ol: me soiir theory. T'Ee fiSfpBtimentai aata from atomic spectroscopy fully confirm
this
= 0, whereas =
Lfa
can never be
quantum- mechanical result.
It
is,
of course, understood that in the presence of a magnetic field the central

is disturbed.
symmetry
184
In the Bohr theory the direction of the z axis can be taken to coincide with that of the orbital angular momentum. In this case n ==/z v and, therefore,
In the
wave theory
this
case corresponds
to
m = when
/,
<
^max=
whereas
fr2/2 '
n 96
-
>
The appearance of the additional orbital angular momentum tiH is related to the noncommutativity of the angular momentum operators L xt Ly , and L z , as a result of which the angular momentum components cannot have simultaneous definite values. Therefore, when *==&zmax=^ the components L x and L y do not vanish but have certain minimum values satisfying the relation
^m= Vmax+ (^X )min+ (^)min.

The minimum value
of
(AZ^)*
(11.98)
and (AL V
2
)'
may be
obtained with the
help of the uncertainty principle [see (8.13)]:
(ALXi^)Ur= 1
we may
2
Lx Ly
- Ly Lx = 1 tfl* max = } *V.

\*
(11.99)
Since the problem is symmetric with respect to the x and y axes, 3 set (AL x ) min (A^) min . Hence, we obtain
(^)Jnin=( A ^fein=* T'
(11.100)
and the sum of (AL^)^ in and (AL v ) imln is exactly equal to the additional angular momentum H*l. As a result we arrive at Eq. (11.97). Thus the nature of this additional term is the same as that of the
zero-point energy of the harmonic oscillator. Both are related to the uncertainty principle. For large values of the orbital angular momentum quantum number/, we can neglect the term H*l in (11.96) in comparison with /W% so that in fact we have the Bohr semiclassical solution.
Chapter 12
The Rotator
Spherical harmonics, which are the eigenfunctions of the square the angular momentum, have their main application in the quantum theory of the rotator, that is, in the quantum-mechanical description of the free motion of a point over a sphere. The results of the theory of the rotator can be used directly in connection with the spectra of diatomic molecules. Since, however, the angular part of the wave function in a central field is also described by spherical harmonics, many predictions from the theory of the rotator (for instance, the angular dependence of the wave function and the selection rules for the quantum numbers / and m) remain unchanged in the theory of a particle in a central field (for example, a particle in a Coulomb field in the problem of the hydrogen atom).
of
<1>
A.
EIGENFUNCTIONS OF THE ROTATOR
We
rotator
shall first write the basic results of the quantization of the according to Bohr's theory. These will be used as a
starting point in our further discussion. a Suppose a point is moving over a sphere of radius r const. Let the origin of the coordinate system be at the center of the
= =
sphere. The potential energy V
(r)
is then
in
Since the reference level of the potential energy can be specified any desired manner by defining its value at some point as zero,
set
we
The
total
energy of the rotator
is then equal to its kinetic
energy
(12.2)
E =T = ^f-.
The generalized momentum p^ which here has the meaning angular momentum, is found to be
of the
186
Using Bohr's quantization rule, we obtain

Pf
and, consequently,
/!,*,
(12.4)
where J m a* is the moment of inertia. The quantum- mechanical theory of the rotator is a special case of the problem of motion of a point in a central field of
()
force. Consequently, radial function R(r):
we
shall use
Eq.
(1.21)
to
determine the
0.
(12.6)
We
^
= /(/-f-l)
t
have here set the potential energy equal to zero and substituted a const for in accordance with Eq. (11.51). Since r
rotator,
is
the
function
/?(r)
/?(a)
= const,
that is,
v?#( a )
= = =
energy E
now
easily found
from Eq.
(11.51)
*'*(/+!)
27
7) '
Comparing this equation with Eq. (12.5), we see that in the Bohr /(/+!) As has theory E'^n*, whereas in quantum mechanics already been mentioned in Chapter 11 (Section D), this difference is due to the noncommutativity of the components of L xt Ly and L z of the angular momentum operator; it is one of the characteristic features of quantum mechanics. Both equations become identical only for large quantum numbers, that is, when /*>/. According to Eq. (12.7), the energy of the rotator depends only on the orbital angular momentum quantum number /. The magnetic quantum number m 9 which characterizes the projection of the angular momentum L on the z axis (and, consequently, the orientation of the angular momentum in space), does not appear in
the expression for The eigenfunctions Yf corresponding to the ,. eigenvalue E [see (11.67)] do, however, depend on m. Since m can / to +/ [see (11.50)], each energy eigenvalue E t will vary from have 2/+ 1 corresponding mutually orthogonal eigenfunctions; they describe the state of the rotator and differ only in the orientation of their angular momentum L relative to the z axis. In this case the energy level E 7 is said to be (2/+ l)-fold degenerate. In general, a state of a system (or a given level) is said to be
t
N- fold degenerate if Af linearly independent eigenfunctions correspond to the given energy eigenvalue,
THE ROTATOR
187
Ph v*l6ftl emanation for thfi ^.tlp mer fg levels rotator is tKat ine rotator forms a centrally symmetric system, and consequently all directions passing theough the origin of coordinates are equivalent. From these considerations it follows that degeneracy will occur in any centrally symmetric
faSW*^
of the
system.
there exists a preferred direction, for example, one deterthe direction of a magnetic field, the central symmetry is disturbed and the possible directions of the angular momentum L are no longer equivalent. As a result, the degree of degeneracy is either reduced, or the degeneracy can be completely removed. In spectroscopic notation the energy levels are called terms; for example, the level corresponding to / is called the s term and the level corresponding to 1= , the p term. For the d term, / =2; for the / term, 1 3; for the g term, 1 4; and so on. Correspondingly, the rotator is said to be in the s state when / in the p state when / and so on. 1 Let us consider in more detail the 5 and p states of the rotator. m Q, in the 5 state, it follows from Eq. (11.67) that the Since l is eigenfunction 7J corresponding to the energy eigenvalue
If
mined by
= =
and the probability density
KJ|* is
given by
In the p state the three values
=
1,
1,
and the quantum number m can have any of and -pi. Consequently, the energy eigen-
value
-j-
is
associated with the three eigenfunctions:
=Y\ = y~
f
1"
on
shift,
12 - 10 >
cos*.
(12.11)
(12.12)
The corresponding probability densities are

(12.13)
~cos'&.
(12.14)
188
The probability distribution functions (12.9), (12.13) and (12.14) are plotted in Fig. 12.1; they are shown only in the zy plane because V does not depend on the angle ft. To obtain a complete picture, it is necessary to rotate the graph about the z axis.
|
Fig. 12.1.
Probability density distribution functions for the

rotator.
(12.9) and Fig. 12.1a that, for a rotator state, the angle <p , which gives the direction of the angular momentum L relative to the z axis, is arbitrary. This was to be
It
can be seen from Eq.

s
in the
m=l m=\
expected since the angular momentum L* #*/(/-{- 1) is equal to zero in this case. A material particle at rest has an equal probability of being at any point on the spherical surface of radius a. In other words, all positions of the rotator are possible. There is no classical analog of this state. From Eq. (12.13) and Fig. 12. Ib it follows that in the p state, when /= and m dL the most probable of all the possible orbits of the rotator is the one located in the xy plane. The states with
1
and 1 will have different directions of rotation: for the rotator will rotate clockwise (the angular momentum L is 1 it will rotate counterclockwise parallel to the z axis); for /n (the angular momentum L is antiparallel to the z axis). When 1 1 and m 0, the most probable orbit of the rotator lies in a plane passing through the z axis [see Eq. (12.14) and Fig. 12.1c]. In this case, the orientation of the angular momentum is perpendicular to the z axis. It is worth mentioning that a similar analysis of the angular part of the wave function applies to all systems characterized by central symmetry.
m=
THE ROTATOR
B.
189
SELECTION RULES
As has already been shown, the selection rules indicating the changes of the quantum numbers that correspond to allowed transitions can be expressed in terms of the matrix elements
(r$'=
(12.15)
j',(V7')*rV7^.
If the matrix element vanishes for some particular chapgff oj: foe the corresponding transitioiTis f^rb^dên (there quantum.numbers, ~ wiin>e no radiation). 6nce~ we know the selection rules, we can immediately find both the frequency and the intensity of the radia-
tion [see Eq. (9.22)], Let us replace the coordinates x, y and Eq. (12.15) by the following new variables:
(that is,
replace r) in
(12.16)
(12.17)
(12.18)
From the physical point of view, this is equivalent to resolving the motion of the rotator into three parts an oscillation along the z axis described by the z component; a clockwise rotation in the xy plane (the component) and a counterclockwise rotation in the xy plane (the TI component). In combination, these three components completely describe the motion of a point over the surface of a sphere. In terms of the new variables the determination of the selection rules reduces to a calculation of the matrix elements:
;
;*'
=
(
(Yf)*
sin
M* Y? dQ,
"
(12.20)
($' =
(Y*
)* sin Oe
Yf dQ
(12.21)
where, for the sake of simplicity, we have set
a=
1.
190

1 Using the recurrence relations for spherical harmonics
(12.22) (12.23)
together with the orthonormality condition (11.68),

(z)/m
(tfim
'
we
find
(12.24)
(*i)/m
= const m m = const m m +i*r /1 = const m m

'
,.
',
(12.25)
(12.26)
&
i&/i,
i.
Therefore, we obtain the following selection rules: (a) for vibration along the z axis
Am = m
(b)
m'
0,
A/
= /~~/' =
zt: 1;
(12.27)
for clockwise rotation
Am = -l,
(c)
A/
=
= it
l;
(12.28)
for counterclockwise rotation
Am = +
A/
(12.29)
We have just shown that the only allowed transitions are those for which the changes of the magnetic quantum number tn and the
orbital
quantum number
are
(12.30) (12.31)
The coefficients A and B can be found expansion (11 67) into Eq. (12 22), setting
in
a fairly
simple
way
We
substitute the
Then, dividing
and the coefficients of X
72 and equating the coefficients of x and right-hand sides (nothing further is obtai by equating the coefficients of the remaining powers of x), we find
all the
2 terms by e tm<f (I-* )
l
~m ~
on the
left-
Similarly,
we
find
?
'"'
(2/+l)(2/ + 3)
(12 23a)
"
(/,m)
THE ROTATOR
191
note that these selection rules for the quantum numbers m and hold for any centrally symmetric system including, in particular, the hydrogen atom. From the selection rules, we can find the possible emission (or absorption) frequencies of the rotator:
We
/
will
[see Eq. (12.7)] Substituting the expression for the energy and considering that the moment of inertia of the rotator does not change in this case, we can reduce Eq. (12.32) to the form
,
<%<
= -A
-
[/
(/
+ _
l
/'
(''
)]
(12.33)
From Eqs.
(12.31) and (12.33),
we
obtain
<
12 34 >
tt|i
,+
=_.*(/ -I-
)f
(12.35)
where the frequency
co, ,_j corresponds to a transition from a higher energy level to a lower one (a downward transition) and w/,/+i to an
upward
transition.
C.
SPECTRA OF DIATOMIC MOLECULES
of radiation
There are three main types of spectra the continuous spectrum emitted by a heated body (for example, black-body
Planck's
radiation, with a spectral distribution described by formula); line spectra (or atomic spectra), caused by transitions of atomic electrons between energy levels (for example, the Balmer series for the hydrogen atom); and finally, band spectra character-
izing molecular radiation. A band spectrum consists of bright bands with a sharp edge on the low-freDiagram of a diatomic quency side and a diffuse boundary Fig. 12.2. molecule, on the high-frequency side. Only a high- re solution spectrograph can show that the band actually consists of a series of individual lines. As we shall see below, band spectra are directly related to the rotational motion of molecules. Let us consider a molecule consisting of two atoms with masses mi and m 2 separated by a constant distance r (see Fig. 12.2). An
192
of such a molecule is provided, in a first approximation, the HC1 molecule. It is well known that in the case of diatomic by two or more particles, the center of mass moves as a single particle whose mass is equal to the sum of the masses of all the particles:
example
um=I>.
(12.36)
mass
of the particles is characterized by the reduced the reciprocal of which is equal to the sum of the ///red* reciprocals of the masses of all the particles:
The relative motion
m red
(12 - 37)
To prove this, let us write the Lagrangian of a system consisting, for example, of two mutually interacting particles with masses mi
and
tn.i
X=
where
-^ +
=
i-
-V
(x,
*,),
(12.38)
*, and *2 are the coordinates of the first and second particles, xt is the distance between them. Introducing respectively, and x^ *2 the relative coordinate x Xj and the coordinate of the center
of
mass
c.m
we
obtain
m sum
frc.ny
i
m recj ?
fl
..
from which, using the Lagrange equation
A
dt
we
find that the motion of the center of the total mass
mass
is
characterized by
m sum-*c.m.= const,
and the relative motion of a particle by the reduced mass
(12.40)
as-
THE ROTATOR
the center of mass is at rest (^.mf=0). the coordinates are related to the relative coordinate x by the equations
If
*,
193
and
MI
'
-f- /Wa
//*i
"f"
/#3*
Accordingly, the
moment
J
of inertia of a diatomic molecule is
=m
x\ -f-
4 *S
flVpri**
(12.42)
which
similar to the expression for the moment of inertia of a material particle whose mass is equal to the reduced mass single and whose coordinate is equal to the relative coordinate. Therefore, in all the results obtained for the rotator, we must substitute Eq.
is
u. (12.42) for the moment of inertia, setting * If the radiation is caused only by rotational follows from Eq. (12.34) that its frequency is
<o
transitions,
it
li/
_ 1
= 2/
(12.43)
where
fl
~ *,
(12.44)
From this it is seen that rotational spectra (molecular spectra that result from transitions between rotational levels) consist of sets of equally spaced lines (see 1*4 Fig. 12.3). The rotational spectrum,
however,
lies
in
the
far
infrared
region (radiation wavelengths of the order of 100-300 /i) and its investigation is rather difficult. Absorption lines of this type have been discovered, for example, in the spectra of
2
/
HC1 molecules. A measurement of the spacing between the lines enables

us to determinethe moment of inertia of the molecule. In addition to pure rotational spectra, there are vibrationalrotational spectra, which result from the internal Vibrations Of a molecule in
d)lo
v21
uj32
w43
rotator.
Fig. 12.3.
Spectrum of a
conjunction with its rotation. These spectra are not situated as far in the infrared as the rotational spectra and are more easily studied. Let us consider in general form the theory of a diatomic molecule with varying interatomic distance. This diatomic molecule represents an oscillating rotator. Without considering in detail the atomic interactions in the molecule, and using only simple
194
qualitative arguments, let us find the general
form
of the potential
energy curve V
(r)
First, we must set K(r-*0)-*oo, since the atoms cannot be arbitrarily close to one another. Second, the interaction between the atoms must become negligibly small as r->oo, and hence y ( r _^ co) -> o. Moreover, since a molecule is a stable system, the potential energy V must be negative at a certain finite interatomic distance r a, where it has a minimum. Otherwise, if K^=0 the molecule would rapidly dissociate (see Chapter 27 for a more
detailed account of the bonding energy of atoms in molecules). The general form of the dependence of the potential energy of atoms in a molecule on the interatomic distance is illustrated in Fig. 12.4.
Fig. 12.4. Potential energy diagram for a diatomic
molecule
r a of the molecule from the equilibrium the departures x are position (r a) relatively small (x<a), the potential energy V (r) can be expanded in a series in the neighborhood of the equilibrium point r a
If
= V(r) = V (a
-|-
x)
=V
(a)
4- xV
(a)
+fV
(a)
(12.45)
that the function
Keeping only the first three terms of this expansion and noting V has a minimum at the point r a [V (a) and ], Eq. (12.45) can be reduced to the form
(12.46)
THE ROTATOR
195
D are the elastic constant and K(a) Here and the dissociation energy of the molecule, respectively, 2 To find the energy levels of the molecule (and thus its energy spectrum), we take the Schrbdinger equation (11.21) for the radial part of the wave function, since in our approximation the potential energy (12.46) possesses a spherical symmetry. Since we are interested only in the relative motion of the atoms, we replace the mass by m red in Eq. (11.21). The resulting equa-
V^^m^^
tion is
=o
Noting that
(12.47)
v*=S + }f = }^'
and introducing the function
rR
<
12 - 48 >
u,
(12.49)
we
obtain after substituting (12.46) into (12.47)
Since x<â,
0.
<12.49a)
we may assume
Then
in this equation that ~ r
setting
= (a+ ^ ~ a
.
*>
+D
where B
#/(/ -f
',
(12.50)
=^
and ^==wred
2
,
we reduce Eq.
(12.49a) to the
form
_
This equation is exactly the
oscillator, and, therefore,
te
same as Eq.
(10.14) for the
harmonic
(12.52)
t
where
0,
1,
2, 3,...
The dissociation energy (-))
into
is defined as the work required to split the molecule atoms (neglecting the vibration al energy). This energy is usually of the order of
several electron volts.
196
Thus, when we consider both the vibrational and the rotational motion, the energy of the molecule is
D + fl*/ (/+!)+ Aw
(ft
+ */)
(12.53)
The first term here represents the dissociation energy, while the second and third terms describe the rotation and vibration of the molecule, respectively. We note that the molecule has only a finite number of discrete energy levels, since it is dissociated when
Qualitatively, the dissociation of a molecule at large quantum 1, the amplitude of will in effect stop interacting over this distance and the molecule will cease to exist as a bound system. Moreover, if the orbital angular momentum numbers / which characterize the rotational energy are too large, centrifugal forces can also split the molecule. We shall now proceed to investigate the vibrationalrotational shall assume that the spectrum is determined spectrum. the vibrational energy of the oscillations because its primarily by order of magnitude is larger than the rotational energy (A vib lOjj, and Xrot~ 100 p.). We must bear in mind that spontaneous transitions can occur only from higher to lower energy levels, that is, with a change from k to k [in accordance with the selection rules, the 1) or higher quantum number / may change to either lower (/ ->/ !) values]. Then for the frequency of the radiation (/->/
numbers can be explained as follows. When k vibration may become so large that the atoms
-->
We
tu
,^/:(V)-/:(fe
n
-1, /j:l)
we
obtain
from
(12,53)
u/
ft)-|-o)/r .
(12.54)
Here, in accordance with (12.34) and (12.35), !, and
we have
u>,
= 2Bl,
Thus, two branches are obtained (see Fig. 12.5):

w*
= w vib + 2Bi and or = w vib
25 (/
(12.55)
THE ROTATOR
197
Such vibrationalrotational spectra are observed, for instance, CO molecules, and the investigation of vibrationalrotational spectra is of great importance for the study of molecular structure. From these spectra it is possible to determine various properties of molecules, such as their moments of inertia and the isotopic composition (since the moments of inertia of two molecules consisting of different isotopes of the same element will be somein
HC1 and
what different).
Negative branch
Positive branch
l-f
Fig. 12.5.
Vibrational-rotational
spectrum of a
diatomic molecule
To conclude this section, let us consider the spectrum of a molecule when one of the atoms is in an excited state. In this case the vibrationalrotational radiation is accompanied by the transition of one of the atomic electrons from an energy level n to a lower level n'. The energy of such a molecule can be written in the form
M
-(-, 4- E /f
(12.56)
where E n is the energy of the excited atom. For hydrogen, E n is given by the Balmer formula (see Chapter 2)
(12.57)
The energies
tively,
of the vibrational
and rotational motions are, respec-
(12.58)
and
,
/(/+!).
(12.59)
As a
result of the transition, the energy EM* of the molecule

EM'
changes and becomes equal to
=E
n -f- Eh' -f- EI>

>
(12.56a)
198
Since the main part of the radiant energy is now due to the electronic transition n -> n' in the atom, the quantum numbers k and / may either increase or decrease
K = k\,
/'
= /iLl.
(12.60)
Regardless of the changes in k and /, the overall change in energy is a decrease because of the emission of radiation due to the
electronic transition.
This case is characterized by a further important feature, namely, the strong dependence of the energy bonding the atoms in a molecule on the number of the particular shell in which the electron is located. Therefore, a transition causes a change in the bonding energy, which in turn leads to a change in the interatomic distance a. In transitions from an excited to the ground stated the
distance generally increases, together with the moment of inertia 2 ~ decreases. Owing to the red a , whereas the quantity B
y=m
change B-+B', there
is
an additional slight change
in the rotational
part of the energy, which now becomes

r
B'ft/'
(/'+!),
(12.59a)
where
B'<B
Taking
transitions,
in the present case. into account all possible
vibrational and rotational

UJ ;M
we
find that the frequency of radiation
is
(12.61)
where
M = fl/(/ +
.
l)
BT (/'+!).
we can write Eq.
-f-a> /fl
i.
(12.62)
With the notation
o)
*~
v
>,
(12.61) as
u>
o>
(12.61a)
Finally,
we
obtain three frequency branches for the molecular band
spectra
(12.63)
(12.64)
(12.65)
In these formulas the first or positive branch (the R branch) corresponds to downward transitions between the rotational levels,
THE ROTATOR
199
and the second or negative branch (the P branch) to upward transiThe third branch (called the zero branch or Q branch) appears when there are no transitions between rotational levels; it is due entirely to changes in the moment of inertia caused by transitions within the atom. Using Eq. (12.62) we can represent u>+, ur and u> in the form
tions.
u>+
u>
-f (B
B'
(12.63a)
(12.64a)
(12.65a)
plotted
These branches are depicted in Fig. 12.6, where the frequency is along the abscissa and the orbital angular momentum quantum number / along the ordinate. It can be seen from this
Fig. 12.6. Molecular band spectra: *>+ - the positive branch; oT - the negative branch; f*P the zero branch.
diagram that the superposition of the rotational lines o> /t on the electronic-vibrational line o) gives rise to a whole band of lines with a sharp edge on the left and a diffuse edge on the right. This is in complete agreement with the experimental facts.
,
200
^Problem
the oTBitai

12 iy Find the explicit form of the spherical harmonics for the cases when 2 and / 3. Verify their orthoangular-momentum quantum number is /
ff
normality by direct calculation. Plot a graph of
8
I
Xj
for
in
0,
1,
2.
Answer.
'
Kf-
sin
cos
',
--
J~
sin
9 (5 cos*
is 2ttlU
Problem ^2.^1 Find the eigenfunction of the operator ftliJ the orbital angular momentum quantum number
/
Solution. Let us look for a solution of the equation for
L.v , is
1
/=
given that
1.
its
eigenvalue
and
in the
form
Using (11.88), with
/=
and
^Ci m=+
f
1,
1,
0,
and applying the normalization condition
<p*t|;sin to/Orf?
=1,
we
find
Investigate the general form of the motion of a free particle in spherical coordfnalfes. juetermine the normalized functions for / 0(s states). Solution, The Schrodinger equation for a free particle in spherical coordinates is
written as
where
"The problems
in this
chapter apply also
to
Chapter
11.
THE ROTATOR
Introducing the
201
new
function
X=
1
yr /?, we
transform Eq. (12.79) to

1
\*\
The solution of this equation is a Bessel function of half-integral order. Since the wave function must remain finite as r 0, we retain only the Bessel function of the first kind. Then
and for a free particle with a given energy the general solution of the wave equation in spherical coordinates can be written as
nn
/
1 '
"~*
2
(12.67)
Om-/
we have
In particular, for /
(s state),
C
Since the spectrum is continuous, this expression must be normalized by a
oo
8
(12 68^
function
f
j
o
/?*
(fc' f
r)
R (*,
r)
r*dr
(k
A').
Hence, we
find;
C= ]/ ~~
Problem 12,4^ Find the selection rules for quadrupole radiation from a rotator. iolTOon. Ttis necessary to find the transitions for which the matrix element of the quadupole moment
V
<?
=
J
*Pm' ( 3 ^s^'
rfQ
- * ^')
2
= '(3
does not vanish. Here
V
s'
f 4fm't/m
-V
^
**
s,
=
^r
1,
2,
3;
*8 = z = a cos
sin
ft,
,v lf3
ly
=a
=
{
-~ .
Since the matrix element of a bilinear combination can be expressed as the products of linear matrix elements (see Problem 10.4),
sum
of the
m", /"
and, moreover, it follows from Eqs. (12.24), (12.25) and (12.26) that the only nonvanishing linear matrix elements are of the form
e/
*/,
1.
4-
'
,/
'/,
1,
,' "/,
we
obtain the following selection rules for the quadrupole radiation from a rotator:
Am = 0,
1,
2;
A/
0,
2.
202
As we shall see later, the parity of a spherical harmonic is determined only by the value of / and Is independent of m (see Chapter 17). Consequently, for dipole radiation, 1 ), the only allowed transitions are from an odd state to an even state or vice versa (A/ whereas for quadrupole radiation, the transitions can occur only between even states or between odd states.
Problem
12. 5.)
Expand the plane wave

t|/
= e'*" =
~y
*'*
rcos *
(12.69)
in
terms of spherical waves.
form
and cos Solution. Introducing the notation kr of an expansion in Legendre polynomials
r,
let
us look for a solution in the
Using the orthonormality condition for Legendre polynomials [see (11.68)], we find
i
Substituting the expression for the Lengendre polynomials as given by Eq. (11.59) and i x transferring the derivative from the function ^ x * __ \ y to the function e ^ 1 times (that Is, integrating by parts / times), we obtain
1 i
Bi
=
Z'/
\
'
llyl I "f")
(1
*W
+
dX
'
4 Using the well-known relation from the theory of Bessel functions
(i
-*.)V^= yu/i Ly
(
/?/.
"VJ+l/3 (y).
of a plane
we determine
the coefficients
The desired expansion
wave
in
terms of the
spherical waves is now written as
P,(cos).
(12.70)
Let us emphasize that the plane wave

particle
e ikz
satisfies the Schrodinger equation for a free
and, therefore, the right-hand side of Eq. (12.83) also represents a linear combination of particular solutions of the above equation written in spherical coordinates. This explains the expected proportionality between the coefficients Bi and the Bessel functions
See,
I,
for
part
p. 572,
example, P. M. Morse and H. Feshbach, Method? of Theoretical Physics McGraw-Hill Book Co.
Chapter 13
The Theory of the Hydrogen-like Atom

(Kepler's Problem)
The Bohr theory of the hydrogen-like atom (see Chapter 3) is semiclassical in nature and can only provide a very incomplete explanation of some of the basic properties of the atom. It does not enable us to calculate the intensity of radiation emitted by an atom or to construct a theory of an atom with more than one electron. The wave- mechanical theory of the atom is able to deal with these problems without any fundamental difficulties. The problem of the hydrogen-like atom presents in addition a certain methodological interest since it can be solved exactly, like the harmonic oscillator and rotator problems. Mathematically this problem can be regarded as a generalization of the classical problem of the motion of a planet around the sun (Kepler's problem).
A.
ENERGY EIGENFUNCTIONS AND EIGENVALUES

of interaction
The energy
between an electron and a nucleus
= -Q
(13.1)
depends only on the distance r between them. The problem of the hydrogen-like atom a single electron moving about a nucleus is, therefore, a typical example of motion in a central field of force. If the origin of the coordinate system is at the center of the nucleus, we may regard the angular part Y? of the wave function as known [see Eq. (11.67)] and find the energy levels and the radial part from Eq. (11.21), which for the present case is written /?(/) as
204

1
Let us introduce the effective potential energy of the electron

e(f
31
(13.3)
2/rtor
where the
first term represents the Coulomb interaction and the second the centrifugal forces.
E<0
Fig. 13.1.
The
effective
potential energy function of distance
(solid
line)
as
The dashed
line
shows
the behavior of the
wave function
A graph of V e n is given in Fig. 13.1. This graph the total energy of an electron is negative (<0), it
l
shows
that
if
moves within
We
shall attempt to interpret Eq. (13.3) from the standpoint of the classical theory,
using the classical relationship
A. 2m
o
2m O r 2
(13 3a)
Since p
- constant
for central forces,
we may
write
To generalize
this expression for the
quantum case, we must replace p
by
its
quantum-
mechanical value p
in Eq. (13 2) as
tl
l(l
1)
In the
same way we can regard

in the
the expressio n
1~[ 2m
corresponding to the term pj?/2m
classical theory
THE THEORY OF THE HYDROGEN-LI KE ATOM
205
a region bounded on both sides by potential barriers (the case of elliptic orbits in the classical analog), so that the energy spectrum will be discrete. For E 0, there is no barrier on the right (r -- co) and the position of an electron can range to infinitely large R (the case of hyperbolic orbits in the classical analog). Since the electron's position in the atom must be bounded by a certain r max (elliptic orbits), it is necessary to assume that <0 in order to develop a theory of the atom. Accordingly, Eq. (13.2) becomes
>
where
**/A
Introducing the
=B>
and
^j-
== A
>
0.
(1 3.5
new variable
P
= 2M"r,
(13.6)
we
obtain the equation
~~"
' '
'
'
tfA
'
where
R'
= (dR,'dp)
the graph of V eff, we can see the general nature of the solution. Clearly, inside the well (>'min <C /'<C'"max) the solution will be oscillatory, whereas outside the well (/--*() and r oo) there are two solutions (on each side) one that tends to zero and a second that tends to infinity. To make the solution an acceptable wave function, we must impose on it restrictions that will eliminate the solution that tends to infinity. Just as in the problem of the harmonic oscillator, it turns out that this requirement can be satisfied only for certain discrete values of the energy of the
:
From
electron.
Since the potential well is not symmetric, we shall look for asymptotic solutions for p *0 and for P--CO separately. The asymptotic solution for p-*oo may be found, according to Eq. (13.7),
from
/JilRaÔ,
which gives
P.
(13.8)
(13.9)
206
To eliminate the exponentially increasing solution, we must 0. Coefficient Q can be included in the normalization Cj coefficient of the wave equation and can, therefore, be set equal to
set
unity.
We
then have
/?
= *-'/*.
p
(13.10)
The asymptotic solution for

(13.7),
-0
can be obtained from Eq.
which
in this
case reduces
to the
form
#o
0,
(13.11)
Setting
ft
/?o
=
i
p*.
we
find
q(g
+
'
!)
/(/
+ =
1)
.
0,
and thus
</i
/,
(/
-f
).
Consequently,
/?.
=C +C
lP
-1
a p-
(13.12)
infinity at
Setting
p
C*
and Cj
(to
l f
we
exclude the obtain

/?O
solution increasing to
P'-
(13.13)
Equation (13.7) can be rewritten as
and we assume
its
general solution to be of the form

.
(13.14)
In this
case
p/?
= V/ a = vu,
<+
ap
(13.15 )
and for an unknown function u we obtain the equation
According to (13.15),
lno
=-
and, therefore,
-^ + '-ii.
THE THEORY OF THE HYDROGEN- LIKE ATOM
207
Furthermore, we have
<,=_<+!+(_ j +
I)V
and
i_i_!l+^Jl.
to
Using these formulas, we transform (13.16)
=
If
0.
(13.17)
origin
the behavior of the solution for R is to be determined at the and at infinity, by the asymptotic formulas (13.10) and (13.13), we must find the conditions under which the function u will be a finite polynomial of degree k without negative powers:
Substituting
(13.18)
into
(13.17)
and collecting powers of
we
have
(13.19)
Since aft+J
and
a*
^ 0, we obtain
tj
I.
I
VA
/j-j-/-)-.
=H.
*i
/I
(iG.Û)
Q OflV
Here the quantum number n, which is equal orbital angular momentum quantum number
/==0,
1,
to the
sum
of the
2, 3,...
and the radial quantum number

*
0,
1,
2, 3,.,.
(13.21)
It
plus unity, the values
is
called the principal quantum number n.

n
may assume
(13.22)
1,2,3
To determine the unknown coefficients a v of the series (13.18), we derive a recurrence relation with the a id of Eqs. (13.19) and (13.20):
(13.23)
208
and Setting the coefficient of the highest power in (13.18) a k 1)" ( the with of coefficients all the help (13.23), remaining calculating we obtain the following expression for the function u:
2
,
y=o
The series (13.24) is an associated Laguerre of order k and may also be represented in closed polynomial Q(p)
1
where s=2/ +
form
= Ql =
(p)
e?p~
(e-V**).
(13.25)
Thus, for the radial function R nl (r) we finally obtain

i
(P)
= C^-'/V
we
find
<
+,'_, (P),
(13.26)
where
value of
= 2|A/ir.
fl
Recalling that
(13.5),
n,
and substituting the
from Eq.
where
= ~r~T
WO^Q
is the
radius of the first Bohr orbit. Calculating

(see below),
we
the coefficient obtain
Cnl from the normalization condition
(13.28)
We
shall
0, as can be easily verified by taking the first derivative of v with respect to p Differentiating this equation k t 1 times according to Leibnitz's rule, we reduce it to the form
Eq. (13 17)
show here that the function u, written in the closed form (13 25), The function v -= e~ pp k * s satisfies the equation pu' + (p k - s)v -
satisfies
Introducing the
new
function
&
4
v
(s
\>Pp~*
f
we
obtain the differential equation

-,
pw
0)u;' f fcu>
which is identical with Eq (13 17) for the function u (since (/J''Â) /- 1 = k]. Since can be readily shown that the coefficient of the leading term (p ) of the function
)
it
is identical
with the corresponding coefficient of (13 24),
we have demonstrated
that the
relation (13.25) is correct

Therefore, the radial wave function is
I
209
v.
rrr-
" 6
I
(Jn
~~~
I
\
I
/-I
\m*o
(lo.Aoaj
o OOn
As we know,
the normalization condition for the radial part of the
wave function
is
dr
1.
Substituting the expression for ance with (13.27), we obtain
Rn
from (13,26) and replacing r by 9 ~
pf
in accord-
Now
let us represent one of the polynomials as a series ( 13.24 )t leaving the other polynomial in the closed form (13.25JL The normalization condition then becomes
Q'
Applying the theorem for the transfer of a derivative (see Chapter 7, Section C) to this
equation,
we
find
It is obvious that the remaining terms in the series representing the function QjJ vanish, since the order of the successive derivatives is higher than the corresponding powers
of
p.
Using the well-known integral

00
(13.29)
we
obtain the expression (13.28) for Cnl v average value of (r~ ) 1, 2, <*, 4>, which
In
p=
will be useful in the later

00
a similar fashion, we can determine the development:
On
the basis of the above equations
we can rewrite
CO
*
this as
(7-*)=CL nl ("%}* \2Z

v
("*'}
\2Z
{ P~
v+1
( i
/ l)*{p*~ if *(* v
.1)1.
210

theorem for
the Setting v 1, 2, 3 and 4 in the last expression, and using once again the transfer of a derivative, we obtain after a few simple operations
(13 ' 29a)
-/(/+D
In calculating r~ t we__retain only the leading term p* in the polynomial @J. On the 8 we retain only the last term p. For r~ 8 , we retain the last contrary, in calculating r~ two terms, and so on. The expressions for r~ a and r~* are obtained under the assumption that^ / ^_0_ For the s states (/ 0), interactions which are usually proportional to either r~ a or /-~ 4 are generally replaced by a contact interaction (see below, Chapters 19
!
,
and
20).
To supplement this general treatment, we shall calculate the OT normalization coefficient of the wave function ^ n/OT =/? n /K/ for the ground (or lowest) energy state, which is characterized by the
quantum numbers
According
for
/
Eq. (12.9), the spherical harmonic Y", is a constant as is the polynomial Q^ for k m Consequently, from (13.26) we have for ^ 10 o
= m = 0,
to
'
<hoo
= GT
j>
a
>
(13.30)
and, therefore, the normalization condition takes the

2
form
(13.31)
j *:..ti*d'A-= J
Substituting the explicit
in
..*i^
î2 -
1.
form
00
of the
wave function
d
^ 100
and bearing
mind
its
independence of the angles

" fl
and
9,
we
obtain
">dr = e J_2Zr
o
l.
(13.31a)
and using (13.29), Introducing the variable .v=--Û
2/^r
we
find
The same value for C can also be obtained from the general equation (13.28) by setting n / 1 and 0. Therefore, the wave function for the ground state is
THE THEORY OF THE HYDROGEN-LIKE ATOM

'
211
The energy spectrum of the hydrogen- like atom Eqs. (13.20) and (13.5):
is
found from
where
/?
is the
Rydberg constant
This expression for the energy, which we note is in complete agreement with the corresponding expression of the Bohr theory [see Eq. (2.50)], depends only on the principal quantum number /_]-,& _|_ (that is, only on the sum of the orbital and radial quantum numbers / and k) and is independent of the magnetic
/j
i
quantum number m. At the same time, the wave function n im Rn iYf depends on all three quantum numbers //, / and m individually. Consequently, it follows from the Schrbdinger wave theory that the energy levels are degenerate. Since each value of / can vary from to ;i / to and m can vary from the degree -\~l [see (13.20)], of degeneracy is
ty
1
n-l
n-1
Chapter 12 that the degeneracy with respect to and is related to the fact that there is no preferred direction passing through the origin. The degeneracy with respect to the orbital angular momentum quantum number / appears, however, only in the case of pure Coulomb interaction. In most other centrally symmetric systems there is no degeneracy; that is, the energy level for a n of is value given split into n sublevels corresponding to different 3 If the system is placed in an external field (for example, a /. magnetic field) which removes the central symmetry, the degeneracy with respect to m also disappears. In this case, the nth energy level is split into ri* distinct sublevels.
It
was shown
in
is characteristic of all central fields of force
In particular, as we shall see later, the degeneracy with respect to / vanishes even in the case of the hydrogen atom if we take into account the relativistic effects, the nuclear
volume, and the so-called vacuum corrections Similarly, in the spectrum of alkali metals, which have one valence electron in the outer shell, the influence of the electrons in the
inner shell removes the degeneracy with respect to
/
212

B. SEMICLASSICAL INTERPRETATION OF THE PRINCIPAL RESULTS OF THE QUANTUM-MECHANICAL
THEORY OF THE HYDROGEN-LIKE ATOM
In classical theory the quantity
(13 . 34)
possible only when the radius rmax ) which can be found by r lies within certain limits ('mm^ r of side the equal to zero. Using the (13.34) right-hand setting quantum-mechanical expression for the energy (13.33), we find
|
[see (E
(13.3a)]
must be greater than zero.

it
For
elliptic orbits
E <0)
|
is
seen that this
is
<13 - 35)
The equation of an ellipse
in polar coordinates is
r==
7+7^1-
<
13 - 36 >
where the parameter P

the
e
b2
~^
is defined
as the ratio of the square of

a,
semiminor axis
ellipse
b to the
semimajor axis
and the eccentricity

s
characterizes the elongation of the ellipse (for
0,
the
becomes
a circle).
:
From
(13.36)
we
readily obtain
equations for rmax and rmin
=a(l +e), -n(\ ^ E 'mm "V

max
r
1
<
13 ' 37 >
;-
Comparing Eqs.
(13.37) with (13.35),
we
find
(13.38)
We
see that the classical analog of the quantity
turns out to
is
be the semimajor axis of an ellipse whose eccentricity by Eq. (13.39).
given

Substituting
into
213
l l
(13.39) the
Bohr value
and then the quantum- mechanical value of
= H nl = H of = we obtain pj **/(/+
pj,
(l-^ i)
1),
(13.40)
Vn=V
From
this
it
>-' U ,J-'
<
13 - 41 >
only in the
clear that the eccentricity becomes exactly zero n 1 In wave mechanics, for l n (l 1) the eccentricity has a minimum nonzero value given by
is
Bohr theory
111111
quan
^!/"!. \ n
(13.42) ^
This shows that in quantum mechanics, we can speak of the classical analog of states with circular orbits only when the quantum number / has the value n Furthermore, it should be noted that state (/ in the classical approximation, an 1, 0) gives which corresponds to a parabolic orbit. This case, however, cannot be associated with a parabolic orbit from the quantum-mechanical point of view. As we can see from Eq. (13.33), in quantum mechanics the energy is negative for all values of n (we recall that in the classical case the total energy is zero for parabolic orbits); conis limited only by its maximum sequently, the radius for / value
1
?>
w~
This lack of agreement between the quantum- mechanical and classical solutions for the state simply means that the case / has no classical analog. In general, we are, of course, entitled to speak only of the probability of particular events in the context of wave mechanics. Therefore, all the results obtained from the Schrbdinger theory must be interpreted in terms of probability considerations. Let us
*>
show, for instance, that the Bohr radius a

for the case of circular orbits (/ mechanics to the most probable value
= -*~
1)
[see Eq. (13.38)]

in
/*
corresponds
wave
of the electron, r a. According to the normalization condition
of the position
coordinate
the distribution of radial probability density D(r) is

r Rli.
!
(13.43)
214
In the
case of circular orbits, when
=n
and
0,
Eq. (13.43)
gives
/-*/?,__,,
(13.43a)
According
to
Eqs. (13.25) and (13.26),
Rn
and, therefore,
13.2):
irt _,
= const
p"'
we
find the following expression for D(r) (see Fig.

~
D (r) = const f n i
Determining the value of
r
(13.44)
at
which this function has a maximum:
fdD(r)\
=0j
we
obtain
rn
= a= "'a
9u
(13.45)
It
is
radius
1, the interesting to note that if we set Z r is the radius a of the first Bohr orbit.
most probable
Fig. 13.2. Radial probability density distribution function in the case of circular orbits.
the radial quantum number k ^ 0, the orbits can be said to be elliptical and the probability distribution D(r) assumes the form
If
Zr
(13.46)
The equation for the extremals
of this function is
= const f"e~*{ +
(21
2)
QfH
(13.47)

1
215
th degree, Eq. (13.47) has k Since xQj^ is a polynomial of the and r roots (not counting points r oo) and fe+ maxima. This case is very similar to the probability distribution for the motion of a free particle inside a potential well (sinusoidal variation of the wave function, see Chapter 4) or for the motion of an oscillator
(see Chapter 10, Fig. 10.2).
C.
SELECTION RULES, EMISSION SPECTRA OF
HYDROGEN- LIKE ATOMS

To determine
ments
the
it
(Kepler's problem),
selection rules for hydrogen-like atoms is necessary to calculate the matrix ele-
Wm' = J
Substituting here
ty
'<
W*00
(13.48)
nlm
~Y?R we
nlt
obtain
r f Yf
As we know from Eqs.
\ J
Rn>v r*R nl dr
(13.48a)
(12.24), (12.25) and (12.26), integration over the angles & and gives the selection rules for the orbital /' / :1) and the angular momentum quantum numbers (A/ m m' 0, rl). Using these magnetic quantum numbers (Am results we can write Eq. (13.48a) as
Rmdr.
(13.48b)
Evaluating the integral
r'R n t .Rnt dr
,
,**'
<? t
This integral can be evaluated by Introducing the new variable p =

Q^/ is found that only the integral
oo
~ a
O
-f
\n
-~\ Then, n
J
expressing
term, It
Q k and
in the form of polynomials!
we can
perform the integration term by
vanishes because of the orthogonality condition.
216
it
is easily shown that it does not vanish for any value of n'\ that is, for all allowed transitions the principal quantum number can
change arbitrarily.
Fig. 13.3.
Spectral
series
of the hydrogen
atom
The wavelengths corresponding to the indicated transitions are expressed in Angstrom

units
Having obtained the selection rules for the hydrogen-like atom, us investigate its emission spectrum. We shall first introduce certain conventional symbols designating the energy levels of an atom. The spectral terms ( E nl /h) which depend, in the general case, on both n and / are denoted by the symbol (nl), that is,
let
(13.49)
1, 2, 3, ..., and I is replaced by one of the letter symbols as 2, 4, 3, 5, ..., 0, 1, corresponding to / g h indicated in Chapter 12. Since the quantum number the only possible terms are Is; 2s, 2p, 3s, 3,0, M\ 4s, 4p, 4d, 4/; 5s, 5p, 5d 5f, 5g; and so on. There cannot, for instance, be a \p term, since in this case we would have n= and /= 1, nor can there be a3/ term, since this would give us n / 3. The radiation frequencies expressed in term symbols (nl) have the form
s,
where n=
p, d,
/,
l^n\
= =
217
(13.50)
where
it
is
necessary
(13.33),
orbital angular
momentum quantum number
Using Eq.
bear in mind the selection rules for the /, namely, V l_\. the term (nl) can also be represented in the
to
form
(13.51)
where R
is
the Rydberg constant
R = -f-
We
:
thus obtain the
following equation for the radiation frequency
ow
(13.52)
In the case of the hydrogen atom(Z l), the Lyman series (see Fig. 13.3), which corresponds to a transition to the lowest energy level n'= (the Is level), is given by
\
(13.53)
where n 2, 3, 4, .... For the Balmer series (see Fig. 13.4), which corresponds to a transition to the level n' 2 from the levels there of are three types //^>2, ^ allowed frequencies:
(13.54)
Since the energy states of the hydrogen atom are degenerate with respect to the orbital quantum number, these three lines merge intO One (See
Fig. 13.3), and, consequently,
Fig. 13.4. Balmer series.
The
the
wavelengths
visible
corresponding
to
lines
Ha
Hfi
H-y and
HS
(13 - 55)
are given in S Angstroms (A). H^ gives the theoretical position of the limit of the series.
similar result is obtained for the Paschen series, whose frequencies (see Fig. 13.3) are given by
en^^V^"
where n
"it"
(13.56
4,
5,
218
D.
IN
MOTION OF A PARTICLE IN A COULOMB FIELD THE CASE OF A CONTINUOUS SPECTRUM
Although the wave function for a continuous spectrum can be expressed in terms of a confluent hypergeometric function, the usual procedure adopted in the investigation of the hyperbolic solution is to consider the asymptotic behavior of the hypergeometric function for large values of r These asymptotic solutions can be obtained directly with the help of the quasi-classical WKB method. To investigate the hyperbolic orbits ( 0), let us first use this approximation to obtain certain general results applicable to the case of central forces. According to Eqs. (11.21) and (11.51), the equation for the radial part of the wave function in the case of central forces is
>
=
where
u
In the
Q,
(13.57)
= Rr.
(13.58)
more
required to decrease as r->0. Furthermore, we are entitled to choose the potential energy in such a way that
slowly
than
WKB method, the potential energy V (r) is

r
1
Equation (13,57) has a singularity at the point /--> associated with the term 1(1+ l)r\ The effective potential energy here forms a potential barrier of infinite height. Consequently, the asymptotic forms of the wave function which we obtained earlier for connecting the WKB solutions across a slowly varying potential barrier do not provide a good approximation in the present case. Therefore, in order to use the WKB method in the case of central forces, we must obtain either a different asymptotic expression in place of the Bessel functions of order V or we must remove the 3
oo by introducing Q to the point x== singularity from the point r the new variable x \nr. We shall use the second of these methods and introduce a new wave function
(13.59)
Equation (13.57) then takes the form

d V
t
_9.tr/2//Io
i-i
2/flo
tr/.v\
/l
QAM
..
/I
Q CQ\
to
which the WKB method is applicable. For the argument z which determines the asymptotic behavior of ^ [see Eq. (5.55)], we obtain
219
2m a V(ex )
dx-.
-ijlpl*.
It
(13.61)
should be noted that here
\p\
no longer represents the momentum.

r>
Transforming back
obtain
to the original independent variable
we
dr.
(13.62)
it follows, in particular, that in using the quasi-classical expression for the one- dimensional radial equation, we must make the following change in the orbital angular momentum:
From
this
(13.63)
Let us now apply Eq. (13.61) to the motion of a particle in a
Coulomb
field
for the case of hyperbolic orbits (>0). In the quasi-classical treatment of the problem, it is necessary to use the connected asymptotic solutions (5.66) (5.69). This means that in our case
[see (13.62)]
we must
evaluate the integral
(13.64)
where
(13.65)
and the value of rn is found from the condition /(r ) 0. First let us determine the asymptotic solution of the wave
function for
r
>
e c
->
0.
*
Substituting
\*
~ +
I -4(<
b &
2m --
Ze
.
2m E -^
into Eq. (13.64) and introducing the notation
(13.66)
220

obtain
we
On
the basis of Eq. (5.51), the asymptotic solution for this case
has the form
where
in
accordance with
the radial
wave function R
(13.61). Since, according to (13.58) and (13.59), is related to x by the equation

<
we
*=^x.
find
13 - 67 >
In
set
order for the solution to remain finite at the origin, we must D =0; that is, we choose the solution in the form
/?
Cr<,
(13.68)
which
is in complete agreement with the asymptotic solution obtained by another method [see (13.13)]. For the other limiting case r-*oo, we have the following asymptotic solution [see (5.67) and (5.66)]:
f).
where, according
to (13.61),
(13.69)
is
equal to
(13.70)
= rkk.
for r->Q,
Equation (13.69) is an analytic extension of the asymptotic solution To determine the quantity z from (13.62), we use the integral (13.64) with the following substitutions:
THE THEORY OF
THE.
HYDROGEN-LIKE ATOM
221
(13.71)
Then, since
YJ ^fcr + T- we
obtain at r->oo
&
We
shall also
(13.72)
make use
of the relations
- arc
ilSLjr arc tan
L-V
(13.73a)
Here,
that
in writing the
imaginary part, we took account of the fact

and therefore the angle determining the
7>0
and
f/-f
"^)>
imaginary part of the logarithm in Eq. (13.73) lies in the first quadrant, and the angle of the logarithm in Eq. (13.73a) in the fourth quadrant. The expression for the argument z may now be transformed to
(13.74)
where the phase

f
&/
is
= -/ +
arc
tan^Lj
r!nl +
+T* +7-
(13.74a)
From a more accurate calculation (a calculation of the asymptotic expression for the argument z from an expansion of the confluent hyper geometric function), the following value is obtained for the phase ty (see P M Morse and H. Feshbach, Methods of Theoretical Physics, Part
II,
New
York: McGraw-Hill, 1953):

1
+ 17)
If
\l
+ iy\ !>
1,
then using Stirling's formula

*
f
ru
we
17)
|r
1,
we may write
as an approxima-
Hence, taking into account Eqs. (13.69), (13.67) and (13.70), we

find
kr
.
(13.75)
is
Here C
is the
normalization coefficient, and the total phase

2/fer.
(13.76)
Setting f~0 in (13.76), we obtain the asymptotic function for the case of free motion;
form
of the radial
(13.77)
Thus
the potential energy in a central field of force is taken into account by means of the phase shift
/A
/"O
-^ V
(r)
- + r* 1)'
2
dr.
(13.78)
(/
where
root of the first integrand. case of a Coulomb field, this phase is given by Eq. (13.76). The presence of the logarithmic term depending on r in the phase shift is the result of the long-range character of electrostatic forces, which can influence the particle even at a
rj
is the
In the special
very large r. The phase shifts

(that is,
fy
are functions not only of
/,
but also of
It*
of the
energy E = ^-~).
They are an essential character-
istic of the eigenfunctions of the continuous spectrum. The value of the phase shift cannot be found in a general form. The quantity
has to be determined separately for each specific problem, &, usually by approximation methods. The phase shifts are of particular importance in scattering problems, where they are used to express the effective cross sections (see Chapter 29). The wave functions of a continuous spectrum are normalized in terms of a 8 function. The expression (13.75) for the wave function is valid practically in all space; there is only a small region near the center where this expression takes a somewhat different form.
223
In evaluating the coefficient C we may, therefore, assume that Eq. (13.75) gives a correct expression for the wave function; that is, valid in all space. In this case, the wave function is normalized by
the relation
We
substitute the expression (13.75) for /?,(&) and neglect the logarithmic term in the expression for the phase shift 8, because it increases slowly compared with r. Consequently, we find
~
and the value of the normalization coefficient in the asymptotic
solution for the central field of force is
(13.79)
An energy level diagram for the hydrogen atom, showing both discrete levels and the continuous spectrum, is given in Fig. 13.3. This diagram clearly shows the degeneracy with respect to /, which results in a merging of all levels with the same n into a
single level.
Besides the ordinary transitions between discrete levels, two other processes can take place ionization and capture. Basically, each of these processes is the reverse of the other. In ionization, an electron jumps from a discrete level (E<0), such as, for example, the ground state, to the region of positive energies (>>0) which forms a continuous spectrum (hyperbolic orbits). This process involves the absorption of energy. Conversely, in the process of electron capture, a free electron jumps into one of the possible discrete levels, at the same time liberating a corresponding amount
of energy.
A
from
the ground state given by (see Fig. 13.3)
certain amount of energy is required to transfer an electron (n=\) to the region >0. This energy is
where T
= ~-
is
the kinetic energy of an electron which is no
lon longer bound to the nucleus. The energy represents the ionization energy of the atom. It is at its minimum when T 0; this corresponds to the minimum energy (E 0) transition of an electron
224
from
atom
the level n l to the continuous spectrum. As a result of this transition, the electron can leave the atom. For the hydrogen
-= 13.59
ev.
E.
CALCULATION OF THE EFFECTS OF THE MOTION OF THE NUCLEUS

we have
until
our theory is rigorous only for the case of infinitely large nuclear mass. In general, this is a relatively rough approximation, particularly in the case of light elements such as hydrogen and helium. By taking into account the motion of the nucleus, it is possible to explain a number of important experimental facts. We shall allow for the effects of the nuclear mass M in the same way as we did in our discussion of the spectra of diatomic molecules (Chapter 12), replacing everywhere the electron mass m by the reduced mass
(13.80)
now ignored
In developing the theory of hydrogen- like atom; the motion of the nucleus. Accordingly,
M
The Rydberg constant then becomes
As a consequence
shifted:
of this
change, the term values are slightly
(/!/)=
-jjj^l
'-^j.
(13.82)
The radiation frequency
is therefore
given by the relationship

(13.83)
which differs from the previous one [see Eq.
(13.52)]
by the factor
Since the frequency of the radiation depends on the nuclear mass M, atomic weights can also be determined by spectroscopic methods, as well as by conventional chemical methods. One successful outcome of the application of the spectroscopic method was the
225
proof of the existence of heavy hydrogen and ionized hellium. Previously, the average atomic weight of hydrogen (relative to oxygen) was found by chemical means. The mass spectrograph made it possible to measure the atomic weight of each atom. These measurements gave somewhat different values for the atomic mass:
Me:
- Af tn sp
MOO
^chem
% ^0.0 145%
(13.84)
This led Birge and Menzel to predict the existence of a hydrogen isotope, called deuterium or heavy hydrogen, with anatomic weight twice that of hydrogen (D= H*). The presence of deuterium in natural hydrogen explains the greater atomic weight obtained in chemical measurements. The mass spectrograph measures the atomic weight of H along, wince the spectral lines of iH* atoms fall at a different place on the scale. Hydrogen (H*) Deuterium, like hydrogen, can enter into reactions forming, for example, heavy water D*O, discovered by Urey and Osborn in 1932. Deuterium is usually obtained via the electrolytic decomposition of water. The rate of evolution of ordinary hydrogen, at the cathode, greatly exceeds the rate of evolution of deuterium; thus, the concenFig. 13.5. Diagram of the relative tration of deuterium in the residual position of the spectral lines of electrolyte increases. (It is almost hydrogen and its isotopes. deuterium in to detect impossible natural water because of its low concentration.) The presence of deuterium was confirmed by spectroscopic studies which showed that not only does the Balmer series (n' 2) consist of the lines
t
Balmer
(13.85)
but that each of these lines is associated with a second line situated somewhat to the right. This second series of lines (see Fig. 6 13.5) is described by the equation
Balmer
'
(13.86)
According
to the latest
experimental data,
R m = 2nc
109737,
KH =
27rf
109678,
RQ =
2frc
109707,
where the numbers represent the values of
R 9p
(see footnote, page 20).
226
which can be readily obtained from (13.83) by setting the mass M equal to twice the mass of the hydrogen nucleus and substituting It is worth noting that the large relative difference between Z the masses of the deuterium and hydrogen atoms causes far greater differences in their physical and chemical properties than is usual with isotopes of other elements. Thus, although heavy water outwardly resembles ordinary water, its melting and boiling points are 3.81 C and 101.4C, respectively, its viscosity is greater and it is a poorer solvent for salts. With the development of nuclear physics, heavy water has become particularly important because it is a good moderator for fast neutrons, and can also be used as a source of deuterium. At present, we know of a third hydrogen isotope, namely tritrium whose nucleus consists of two neutrons and one proton. It (T jH-*), forms a compound with oxygen similar to water. The ratio of tritium atoms to iH atoms in natural water is approximately 10~ 18 whereas the ratio of iD' atoms to ill atoms is 1/6800. In a mixture with deuterium, tritium is a very important substance for the production of thermonuclear reactions: the reaction between ,D* and 4 iT nuclei leads to the formation of 2 He and one neutron. Each such reaction releases more than 17 Mev of energy. Tritium is also a beta emitter (with a half- life of 12 years) and consequently is widely used as a radioactive indicator in chemical and biological investigations. The positions of spectral lines of tritium are slightly displaced relative to the hydrogen and deuterium lines (see Fig. 13.5) and are given by the equation
"aimer
"
'
M/l
(13.87)
Another very important consequence of accounting for the motion of the nucleus was the discovery of ionized helium, first detected in spectroscopic studies of the sun. The solar spectrum was found to contain a series of lines, with positions described by
the equation
(13.88)
released In the fusion of deuterium and tritium into helium, just as in the 239 or Pu under neutron excitation. Fusion, however, is possible only if the potential barrier of the Coulomb repulsion between the D and T nuclei is overcome, 8 High temperatures (~ 10 degrees) are therefore required if this reaction is to be self-sustaining, whereas in the case of fission even low-energy neutrons can easily penetrate the nucleus, and that even at low temperatures. To obtain a thermonuclear reaction, a mixture of deuterium and tritium must first be heated to a temperature of tens of millions of degrees. Such temperatures may be created in an atomic explosion.
Energy
of
is
fission
U 23S
227
where
n^
takes the values
"i
= f-
'
T-
'
I'""
<
13 ' 89)
This series is, in effect, the Balmer hydrogen series (/n 3, 4, 5,...) with a number of intermediate lines, which form the so-called Pickering series, characterized by the half-integral quantum numbers n^= 5/2, 7,2, 9/2, .... At first, the Pickering series was explained by assuming that hydrogen was in a special state in the Sun, so that the quantum number n could assume halfintegral values. The spectral lines, however, were later found to be located further to the right than is indicated by Eq. (13.89), and consequently this assumption had to be abandoned. The second hypothesis assumed that the observed spectrum arises from singly whose nuclear mass is Al ionized helium (, He ) 4 and 7360/?2 whose charge is Z 2. According to (13.83), its radiation frequencies then are
4
(13.90)
Setting
;i'
4,
we reduce
(13.90) to the
form
of whether the Pickering series was hydrogen atoms (under the assumption that the quantum numbers may assume half-integral values) or to the radiation of ionized helium atoms (with the usual integral values of the quantum numbers), it was necessary to find an experimental value of the Rydberg constant. In the case of hydrogen,
where = 5, 6, 7, 8, To answer the question

/i
. . . .
due
to the radiation of
(13.92)
whereas for helium

(13.93)
Careful spectroscopic studies confirmed that the Rydberg constant has the value (13.93), and thus it was shown that the Pickering series represents the spectrum of ionized helium.
228
tl on*S
Problem 13.J^ Starting with Eqs. (13.26) and (13.28), show that the radial wave funcand 3 are as follows: I, 2 "Wn/Mr TflSTSrincipal quantum numbers n
'" (6
\1
>
-6p+
P ),
,j^r 1
where
_2Zr
On
the
basis
of these specific
examples, show that the functions
Rnl are
orthonormal,
that ts f
Hint. In proving orthonormality, make use of the fact that at different n and n\ the corresponding values for p also differ from one another.
r*^
Hint.
?o l
1
LProblem 13.ZJ Show that the average electrostatic potential produced by a hydrogen-* ground state is
+ aa~
iyrj
(,=-!,). m ^/ \
<J>
Find the average electrostatic potential produced by the electron

(r
)
.
(r)
(r')
.v'
and add
it
to
the potential of the nucleus. Integrating
over the solid angle
Q'>
use the identity
bution"
r*R* in the states the nucleus (circular orbits). (see Fig. 13.6)7
Snow
that at
^=
If
^ e maximum
of the probability density distri, rt
Is, 2/?
and 3d occurs at distances of a 4o and q</ from Why do the 2s, 3s and fy states have several maxima
THE THEORY OF THE HYDROGEN- LI KE ATOM

3s
229
r--9aa
Fig. 13.6.
Graph
of the radial probability density distribution for various states.
D = r*R*
equa-
problem 13.4J Using

lons
!T>r'""tt=i=
the functions given in
Problem
13.1, verify the following
T,~~i
and
3:
and
r* r 8 of the radial deviations for these states. Find the spread (Ar) 3 On the basis of Fig. 13.6 and the uncertainty principle, explain why (Ar) a does not /anish for circular orbits in the quantum theory.
_________
Cgj-phignq ly^Ry means of the quasi-classical WKB method, find the discrete energy spectrum of the hydrogen-like atom. Solution. According to (5.75) and (13.62), the eigenvalues of the discrete spectrum
[E
< 0)
may be
found from the equation
where
r\
and r are roots of the integrand and
Since the value of this integral Is 8
ve obtain exactly the same expression for the energy of the hydrogen-like atom as via he Schrodinger theory.
This integral can be easily evaluated from (13 64) by setting a ~ A, b - 2B and 2 The logarithm and regarding r l and r 2 *> rj as roots of the equation /"(r) 4. 1/2) f the complex quantity [assuming that a tends to zero from the direction of positive num>ers (a = 0)1 may be taken as:
8
__(/
?0
/3
fl
^
,
in
<Q
2Tt/3>0
230

that
Problem 13.6J Determine

aTUlii
MMUi
li>
Ttue to the orbital
magnetic field intensity at the center of the hydrogen motion of an electron. Find its numerical value for the
1p state.
Answer.
Hx
Q,
//,=-/
For
the 2p state (m
1)
gauss.
Hint,
lake
the
classical
expression
//
= --?-
r x
for the intensity of the
magnetiv field produced by a moving charge. Transform to a quantum treatment of the problem by replacing r x p by the angular momentum operator L. Then, using Eq. (13.29a), calculate the average value of//.
Chapter 14
Time-Independent Perturbation Theory
^
(A
BASIC PRINCIPLES OF THE PERTURBATION
TREATMENT OF PROBLEMS
A relatively large number of problems in quantum mechanics cannot be solved exactly with present-day mathematical methods. Thus, various approximate calculations have to be used. One of the most widely used approximations is the perturbation theory method, which was first developed to handle problems in celestial mechanics. It is well known that in Newtonian mechanics it is possible to solve exactly only the two-body problem (for example, the Earth-Sun or the Moon-Earth problems). We cannot, however, neglect interplanetary forces and consider only the attraction of the planets by the Sun since many delicate phenomena are associated with these additional interactions (it is worth recalling in this connection that Lever rier predicted the existence of Neptune on the basis of the orbital deviations of Jupiter, after which the planet was discovered by astronomers). It thus became necessary to consider the many-body problem, which has no exact solution in classical mechanics. In celestial mechanics it was found that the perturbation problems could be handled by means of an approximation based on the fact that the forces between the planets are much smaller than the force of attraction to the Sun. In this method, one starts
by solving the two-body problem (the zero-order approximation), then takes the "perturbation" into account and finds the correction to the solution (the first-order approximation). In other words, the 'perturbation method" involves taking the principal forces acting on a body, finding the rigorous solution for these forces, and then taking the "perturbing" forces into account. Similarly, in the quantum-mechanical treatment of the motion of several electrons in an atom, it is necessary to consider first the principal forces, such as, for example, the force between the nucleus and an electron. In this case, the perturbing forces may be taken to be the Coulomb forces of mutual repulsion between the electrons. In the problem of an atom subjected to an external electric or magnetic field whose strength is small relative to the electric field of the nucleus, the perturbation may be taken to be the energy of the electron in the external field.
4
232

}
FUNDAMENTAL EQUATIONS OF PERTURBATION THEORY
We shall now develop the perturbation theory in a form suitable for stationary problems, that is, problems in which the Hamiltonian of the system does not depend on time. Suppose the Hamiltonian has the form
H = T + l/ = T + l/
I/',
(14.1)
where the perturbation energy K, and the main part of the potential energy l/ is chosen in such a manner that the Schrbdinger equation of the system
(E
H), when the perturhas an exact solution, characterized by bation V is neglected. Setting T H (the zero-order approxV* imation) and using Eq. (14.1), we can write (14.2) as
+ =
1/'H
(_H
.
<|>
0-
(14.2a)
The basic problem in perturbation theory is to calculate from this equation the energy values E n and the corresponding eigenfunctions The solutions are sought in the form of the series n
E"-|-...
(14.4)
where
to
<!>;
<]/
,
and
<]/'
E' are terms of the first order of smallness relative and E" are terms of the second order of smallness;
and so on.
As a rule, the perturbation energy V can be represented in the form of a potential energy of the same order as V Q multiplied by some small parameter X(X<1). The solutions (14.3) and (14.4) should then appear as expansions in terms of X. Thus, and will be independent of this parameter, E' and f will be propor,

tional to
let
2 and and so on. In the expressions for X, E" and <]/' to X us restrict ourselves to terms of the first order of smallE, ness (that is, we shall retain only terms which are independent of X or directly proportional to X). Substituting (14.3) and (14.4) into ': (14. 2a), we obtain the following equation for <j/ and
,
<|*
(
f
'
ft
V) (f
+f =
)
1
0.
(14. 2b)
Collecting terms of the

(
same order, we have

(
- H) f -f [(' - V") f +
- H) f + ('- V) f=0.
(14.2c)
TIME-INDEPENDENT PERTURBATION THEORY
233
Equation (14.2c) may be regarded as an exact equation, since we have not neglected any terms in it, and since <|/ and may be taken to represent the sums of all terms of different orders of smallness
'
To obtain the first-f ...,'->' ...). order approximation of perturbation theory, we neglect the secondorder term (E'V) <]/ in (14.2c) and use the equation for the zero(that
is .
''
f ^ f -f-
<|/'
+ +
order approximation
(
H)f = 0.
p
(14.5)
This equation yields the zero-order eigenvalues

170
1>
ît
*"3
"tit
po
and eigenfunctions
which are connected by the relationship

();>- HH;V
0.
(14.6)
Keeping all this in mind, we shall now investigate the equation for the first-order approximation of perturbation theory
(<
- H) f = - (' - V)
,;
<!/>.
(14.7)
We assume
state
tion,
(14.7)
n'
=n
that
at
the
Since
'==
and<j*'
and
and
'j>
beginning, the system is in the quantum W in the zero-order approximain the first-order approximation, Eq.
|/'
becomes
(S - H
*;
= -(;- V)
ft.
(14.7a)
We
recall that an arbitrary function can be expanded in a series of orthonormal functions forming a complete set and satisfying the same boundary conditions as the original function; therefore, we to be of the form may assume the solution for
,^
(14.8)
Our problem, therefore, reduces to determination of the unknown coefficients C n of a generalized Fourier series. Substituting (14.8) into (14.7a), we have
>
234
or, taking into account (14.6),

9
we
find
n>
(E n
- M
ft*
= -&- V)
(14.9a)
NONDEGENERATE CASE
Let us suppose that the system is nondegenerate. Therefore, each energy eigenvalue En corresponds to one and only one eigenfunction <[;. Then, multiplying Eq. (14.9a) on the left by <JJ* and integrating over all space, we have
n
n>
(E n
- EM
*nn>
= - E' +
H
*1* V^S
#*
(14.10)
In obtaining this equation, of the eigenfunctions

<|>JJ:
we have made use
of the orthonormality
Since the left-hand side of Eq. (14.10) is equal to zero (J for rt'==tt, and &/m' for n n n), the energy correction E'
f
J.
is
^==V;m
where
(14.11)
Thus, the energy correction Ei of the system quite naturally turns out to be equal to the average value of the perturbation energy V. It is worth noting that the expression (14.11) for the energy correction E' obtained by setting the left-hand side of Eq. n was to zero after it had been multiplied by <[J|* and in[14. 7a) equal tegrated over all space. Since y* is a solution of the homogeneous aquation (14.6), it follows that the right-hand side of the inhomogeneous equation
M't
/
(14.12)
is orthogonal to the solution of the corresponding homogeneous 0, that is, equation Mf
3 */d *
0.
(14.13)
235
the
Let us now proceed to determine the wave functions (that is, of the Schrodinger equation (14.7a) in firstcoefficients C n order perturbation theory. We write Eq. (14. 9a) as
>)
Multiplying on the left by ^y(n'^n) using the orthonormality condition, and integrating over all space, we obtain
9
where
(14.15)
V;*=Jt?y'ld"*.
Thus for
,
n ty'
we have
T-i
= CnW + 2' C"*'

summation symbol indicates
n'
(14. 16)
that the sum Finally, the as yet undetermined coefficient C n of the zero-order wave function can be found from the normalization condition
where
is
the
prime on
all n'
the
taken over
except
= n.
^=l
for the total
(14.17)
wave function
t
= W + # = Ci W + 2' C
n'
""W'>
(14. 18)
where
=!+<;.
Substituting (14.18) into (14.17) and keeping only the first order, we have
-
(14.19)
terms up
to
'JCS*^ J
**!' <P*
+ CJ'CJ J ttft d'x} =
(14.20)
Making use
factor, which
of the orthonormality condition and ignoring the is of no interest to us, we obtain
phase
CS=1.
and therefore Cn
(14.21)
=Q
<]
Consequently, for the wave function perturbation theory, we finally obtain

'
in
the
first-order
-f
E^
Eqs. (14.22) and (14.11), it can be seen that both <K and are proportional to the first power of the perturbation energy (that is, to the parameter A). If we were to compute the corrections to the energy and the wave function in second-order perturbation would turn out to be proportional to the and n theory, both E" second power of V"(that is, to X 2 ).
E' n
<]>J
From
D.
DEGENERATE CASE
We shall now develop the perturbation theory in a form applicable to degenerate systems, in which, in the absence of any perQ turbation, a given energy eigenvalue E n has associated with it ;
eigenfunctions
iJ-.'^
*^-
>-.^
^
**'(*
(*
\*>\
It
is
obvious that any linear combination of these functions

/
(14 ' 23)
is
itself a solution of the
wave equation
in the
zero-order approxi-
mation
If there is a perturbation V, this arbitrariness is removed and the coefficients may become connected with one another by certain definite relationships. Let us show this, proving first that, just as in the case of nondegenerate states, any particular solution of the homogeneous equation (14.6) is orthogonal to the righthand side of the inhomogeneous equation (14. 7a) of first-order

perturbation theory* For this purpose, left by tyn* and integrate over all space.
f
)
237
(14. 7a)
we multiply
on the
We
(E n
then obtain
W (En
*
H)
<!>'
d'x
f
J
<|J* *
V) K d*x.
(14.24)
Using the theorem for the transfer of a derivative [see Eq.
(7.17)],
we have
j
(Ei
- H") W;
d'jc
=- J
T JJ
(; - F)
#x.
(14.25)
Since ^Ji* is a solution of the Schrodinger equation (En we arrive at the equation
,,
H")'^!
0,
^* (;
V) 2] C!^S
rfjc
|f
(14.26)
Without any restriction on the generality, we may assume that all 1 the eigenfunctions <1>H,, are orthonormal. Then, since
we obtain instead
of (14.26)
^' C Q (El ~V',,)=2

i
*>'
Vli '>
14 - 2? )
iSi
where
<Px y'y mi
t
(14.28) (14.29)
;v'wx*,
and the prime on the summation sign indicates that the sum extends over all /"s, except i' Since the subscript / in (14.27) can take any value from 1 to /, we have a system of j homogeneous equations from which we can determine the energy E'n and the coefficients C]
im
- c; v - c;i/; - + c; (; -^ y =
,;
(14.30)
o.
;,)
If
the functions
i/; fl
are not orthonormal,
by
means of
linear transformations,
new
it is always possible to construct from them functions which possess this property.
238

<11
Recalling that the wave function

condition
*tS
must
satisfy the normalization
<***=!,
(14.17a)
that the correction E n to the energy of the unperturbed state En of the system is uniquely determined, as are the coefficients Ci (and hence alsot^). Since, in particular, the system of equations (14.30) can have a nontrival solution only if its determinant is equal to zero, we have the following equation for E' ni
we see
0,
(14.31)
This equation is called the secular equation, a term taken from - ^ celestial mechanics. If the secular equation has several roots (not necessarily / different roots), each one of them will correspond to a completely determined set of coefficients CJ, which can be found from (14.17a) and (14.30) by substituting the given root E'flk for E'n . Consequently, the different corrections E'n to the energy lead to different zeroorder wave functions. Thus, if a system is /-fold degenerate in the absence of any perturbation, a possible effect of a perturbation is to reduce or completely remove the degeneracy [this will occur in the case when Eq* (14.31) has / different roots].
E.
THE STARK EFFECT
If an atom is placed in an electric field, its spectral lines are generally split into components. This phenomenon was discovered by Stark in 1913. Experiments have shown that the effect of an electric field on a hydrogen atom is different than that on other atoms. In a weak field, the splitting of the energy levels of hydrogen (for example, the Balmer series) is proportional to the first power of the field intensity (the linear Stark effect) whereas for all other atoms the splitting is proportional to the second power of the field intensity (the quadratic Stark effect). In a stronger field 5 (of the order of 10 volts/cm), there is an additional splitting ef,
power
fect, proportional to higher powers of the field intensity (second in the case of hydrogen). In very strong electric fields, the
spectral lines disappear completely. The Stark effect could not be explained before the development of quantum mechanics. In terms of classical concepts, the motion of
239
an electron in an atom can always be resolved into three mutually orthogonal oscillations. Let us consider the oscillation along the which can be taken to coincide with the direction of the z axis E V 0, E z The equation describconstant electric field E(E X ) 2 e ) ing the motion of the electron along the z axis is (for e
,
= =
+M
=~
m
where w
is
Q u>lz
=
<
<>&,
(14.32)
the electronic
It
oscillations.
is
is the angular frequency of mass and readily seen that the solution of Eq. (14.32) is
= - ~!r + A cos
(u) '
+ *>
<
14 - 33)
It
is clear that in classical theory the only effect of the constant force ( e g) is to change the position of the point of equilibrium of the system. The frequency of oscillations remains completely unaffected. Consequently, classical concepts imply that the frequency of the radiation emitted by an atom remains the same whether or not the atom is placed in an electric field, since the frequency of the radiation is the same as the mechanical frequency of oscillations of the atomic electrons. Accordingly, an electric field cannot produce any shift of the spectral lines in classical
theory.
We shall now consider the Stark effect in terms of quantum concepts. As we have just mentioned, there are two basic forms of the Stark effect: namely, linear and nonlinear. The linear effect is characteristic only of hydrogen-like atoms because they are degenerate not only with respect to the magnetic quantum number tn, but also with respect to the orbital angular momentum quantum number / (see Chapter 13), with which the linear Stark effect is associated. In all other atoms, there is no degeneracy with respect to /, and therefore the linear Stark effect is not observed. Let us discuss in greater detail the theory of the linear Stark effect for the hydrogen atom. We shall confine our treatment to 3 the second quantum level (/i The external electric field 2). l 10 10* volts/ cm in experiments) is much (which is of the order of weaker than the intraatomic field produced by the nucleus, which is
nucl
10 9
fljj
volts/cm,
orbit.
(in the
where a
{}
is
the radius of the first
Bohr
use the results of perturbation theory
Therefore, we may form developed for
The electric field intensity will have no effect on the oscillations along the x and y axes, which are perpendicular to E.
The
first
quantum level
(n - 1) is
nondegenerate and, therefore, is not
split.
240
degenerate problems). For the perturbation V", we must take the potential energy of the electron in the external electric field
V'
&z.
is
(14.34)
In
the unperturbed
state,
the energy of the electron
[see Eq.
(13.33)]
which has four wave functions associated with
it:
= Ko.o = R
(r)
Y:
= R n (r),
(14.35)
$3
= Ki.o == #21
(r)
Y\
=y
/-
RH
(r)
cos
ft,
(14.36)
= <hu = #a (0 = "[/"^ ^?n (0

>'l
tl
e 9
'
^F=
>
(14.37)
R (r)e-'.
(14.38)
When
and
are replaced by Cartesian coordinates, these wave
functions take the form:
W=/iW,
(14.35a)
f =/
2
(I>1
(/)*,
(14.36a)
YJ
=r /2V/ / / r"\
,
'-
^t
'-^
^~
/1J.
î^t.o/a;
Q7o\
<K
/j
(0
7 ,^
(14.38a)
where
(14.39)
The general wave function of an electron

4
will have the
form
(14.40)
we have the Since the degree of degeneracy is four (/ 4), of four unknown for the equations determining following system coefficients E\ and the corrections C! to the energy E\ of the unperturbed state:
where
ci'i/;,
- v; - avM = - cs v; - civ* + Q (/; - i

3)
o, o,
\/; 4 )
*2*!rjc.
(14.42)
When we
t'n.
integrate over the volume, the matrix elements

'
va 2 v;3l
,
v,
v;3t v;4
v;,,
\/2 ;,
v; lf v;2
v/;4
v;
t ,
v;2
,,
n <i
v;3
vanish, since for each of them the integrand must be an odd function with respect to at least one of the coordinates x, y and z. Only the matrix elements
V'u
and
all
V' n
= V'
U9
which are even functions of

r
three coordinates, differ from zero

1
V=V =
ii
.8
J/i
(Of. (O^
^.
f { (r)
(14.43)
Let us substitute the values of Eq. (14.39) for note that, according to Eq. (13.28a),
and
f 2 (r)
and
and
242
Then, integrating (14.43) over the angles $ and ? (remembering

that z
= r cos
0),
we
obtain
(14.43a)
Next, taking into account the equality

00
we
obtain
(14.44)
in
Using the obtained values of the matrix elements Vi< we find, accordance with (14.31), that the secular equation for the correct ,
tions
E't
is
.00
E\ o
;
o E\
0,
(14.45)
which can be rewritten
in the
form
(14.45a)
This equation has four roots
(14.46)
each one of which, according to Eq. (14.41), is associated with a quite specific set of coefficients:
C2> J
(14.47)
of the coefficients O/ C\ indicates the solution (or root) of Eq. (14.45a) with which it is associated. Thus, it follows from (14,40) and (14.47) that the energy level
1
Here the superscript
(14.48)

is
243
associated in the zero-order approximation with the wave func-
tion
Because of the normalization condition
this
wave function becomes

t
(1)
=y^(^.o + *i.M).
(14.49)
In a similar fashion,
it
can be shown that the state with energy

(14.50)
is
described by the zero-order wave function

(i)
To describe
the states with energy
which remain unperturbed by the electric

approximation, we may
field in the first-order
equally well take the function
or
or, alternatively,
we may
take a linear combination of these func-
even in the tions, since the system remains degenerate for m presence of an electric field. Thus, when the z component of the 1 in units of angular momentum is not equal to zero (m #), that is, the electron is moving largely in the xy plane (see Chapter 12,
1
=+
Section A), there will be no splitting of the energy levels (and, therefore, of spectral lines) in an electric field. If, on the other hand, the orientation of the angular momentum is such that its 2 component is equal to zero m Q and, consequently, the electron
244
is
moving in a plane which includes the z axis (again, see Chapter 12, Section A), then an electric field does lead to splitting of the spectral lines (see Fig. 14.1).
'
Rh 4
14.
Splitting of the second spectral term of hydrogen an electric field (the linear Stark effect):
a) enorgy level in the b) energy level in the
absence of
a field
~-0),
presence of a field
2 may be interpreted as Qaulitatively, the Stark effect for n Since the wave function describing the motion of the follows. 2 (see Fig. 12.1) is not centrally symmetric, the electron for n atom has a certain electric moment p. Consequently, when the
atom
is
placed in an electric field
it
acquires an additional energy

V"
(p) =
pgcos
Tf
(14.52)
where
is the angle between the direction of the electric dipole of the atom and the z axis. Comparing this expression with (14.46), we see that the electric dipole moment of the atom 3a e Q . The solution ty" u) corresponds to the case 7 is p and the solution <p ola) to the case f rc. For the third and fourth solu7
moment
l}
tions,
it is
necessary
to set f
:li^.
We
note that, in the last case,
the electric dipole moment is oriented perpendicularly to the electric field and, consequently, no additional energy appears. To con-
clude this discussion, the linear Stark effect arises because of the intrinsic electric dipole moment of the hydrogen atom at n 2. Predictions obtained on the basis of quantum mechanics are in good agreement with experimental data only in the case of weak fields (10* volts/cm). At higher field intensities (^10 ' volts /cm), there is an additional splitting (the quadratic Stark effect) due to the removal of the degeneracy with respect to the magnetic quantum number m. Finally, at field intensities greater than 10* volts/ cm, the Stark effect completely disappears. This is the result of autoionization of the atoms, that is, removal of electrons from the excited levels.

F.
245
PRINCIPLES OF THE CLASSICAL THEORY OF DISPERSION
Perturbation theory has
many important
applications in studies
of the interaction between light and matter. The predictions obtained in this way differ from classical results and receive excellent confirmation from experiment. This section is concerned with the classical theory of dispersion (that is, the scattering of light) in a dielectric medium. According to classical notions, a dielectric is characterized by the index of refraction
where
(A
is
is the dielectric permittivity (the magnetic permeability taken to be equal to unity). If the index of refraction n becomes
is,
larger as the frequency of light increases (that

is
^>0),the
said to be normal. A typical example of normal dispersion is the dispersion spectral resolution of white light by a glass or quartz prism (the deflection of violet rays from their initial direction is larger than that of red rays).
If,
however,
~ <0in a
certain
range of frequencies, the dispersion in this region is said to be anomalous. As a rule, anomalous dispersion occurs at frequencies at which light is absorbed by the medium. To determine the index of refraction (one of the most important problems in the theory of dispersion), we use the equation relating the electric field intensity E, the displacement vector Z>, and the 4 polarization vector P
D=e
Since
e
= E + 4*P.
(14.53)
/^,
we have
P=Ê.
(14.53a)
Thus, to determine n we are required to find the relationship between P and E on the basis of the microscopic picture of the structure of matter. We shall be better able to appreciate the contribution of quantum theory in this connection after we have completed our review of the basic principles of the classical theory
of dispersion.
According to the Lorentz classical theory, atoms may be regarded as harmonic oscillators in which, in the simplest case, all the electrons oscillate with the same angular frequency u>. If
The polarization P
is
defined as the total dipole moment of the atoms per unit volume
246
we
take the z axis to be parallel to the direction of propagation of the electromagnetic wave, then, since the wave is transverse, we 0), and , may direct vector E along the x axis (, $, ^ vector along the y axis. If we neglect the force exerted on the atomic electrons by the magnetic field (since the magnitude of this
= =
only a fraction (~\ of the force exerted by the electric field), the oscillation of the electrons can be described by the equation
force
is
o-
(14. 54)
We
shall
assume
that the frequency of the incident light
wave
is
u>.
Then
a
cos
w - -*-
(14.55)
If the energy transported by the wave is much smaller than the bonding energy of the electrons in the atom, it follows that the
ratio
- may be neglected because
it
is
small compared to the
atomic dimensions (x~ a~ 10~ 8 cm, while the wavelength A is of the order of 10~ 5 cm). Thus, under the assumptions we have made, the electric field of the wave may be considered to be quasistationary inside the atom. As a result of this simplification, Eq. (14.54)
becomes
-=
^g
N
cos
cot
(14.54a)
where Multiplying (14.54a) by Q N), per unit volume, and substituting

e^x
(e
is
the
number
of
atoms
=p
xt
and
P x = Np x
form
= &,
Py
=P
=:O
we reduce
this equation to the
ffi
-{-
>;$*
So cos w/,
(14.54b)
from which we obtain
Comparing Eqs.
(14.56) and (14.53a), we obtain an equation for the index of refraction which should be familiar from optics:
Wo
247
We note that if the atom is different eigenfrequencies
assumed
to contain electrons with
a more general equation
is
obtained instead of (14.57):
/*
(14.57a)
is the number of electrons per unit volume oscillating with the frequency v k from (14.57) that at radio frequencies (o><a> )the It follows index of refraction may be assumed to be constant, without significantly affecting the accuracy. Its value is given by the relation
where A^
Ne*
(14.58)
On
the other hand, for frequencies <o>o> refraction is given by the relation
Ne*
471
the value of the index of
(14.59)
The index of refraction is, therefore, a constant greater than unity for o><<to , whereas for w^vu> it is less than unity, approaching it as o> CQ. At frequencies close to <D O the magnitude of the index of refraco> it has a discontinuity tion increases without limit, and at o> for this reason beThe 14.2). (see Fig. havior of the function is that Eq. (14.54) does not include the radiation damping of
,
the electron /-"damp,
_r_
_!,
#, which arises
from the interaction between the moving electron and its own field. When /âmp is included, the dispersion curve has the form
indicated by the dotted line in Fig. 14.2. Classical Fig. 14.2. Consequently, the dispersion is anomalous near the resonance frequency u> . Since the Ncg P ersion curve region of anomalous dispersion coincides with the region of the eigenfrequencies of oscillation of electrons in the atom, it follows that anomalous
dispersion
is
accompanied by strong absorption.
348
F.
QUANTUM THEORY OF DISPERSION
Let us now develop a quantum theory of dispersion. By analogy with the classical case, we shall assume that all the electrons in the atoms are in the same quantum state kf We shall use the perturbation method to solve our problem, since the energy of the interaction with the external field is generally small compared with the bonding energy of the electrons. In the nonrelativistic case (when we may neglect the "magnetic" force), the external force which acts on an electron can be obtained from Eq. (14.55) 6
:
The perturbation energy
is
given by
V'
xg
cosa>/.
(14.60)
is
Consequently, the Schrodinger equation for the electron

O.
(14.61)
Let us suppose that Eq. (14.61) has an exact solution for
l/'
= 0:
(14.62)
<pi(0
= tfc- V
(l/n)
==
^-H.^
where
<J$
and*
satisfy the equation
(E,-H)<j*
0.
(14.63)
In accordance with perturbation theory, in the form
we
shall look for a solution
0-
(14.64)
Since (14.62) and (14.63) yield
(-7 'a--"'
This assumption is analogous to the assumption made in the classical treatment, according to which all the electrons have the same eigenfrequencies of oscillation.
Just as in the classical case, we have assumed here that the electric field is quasistationary over distances of the order of the dimensions of the atom.
TIME- INDEPENDENT PERTURBATION THEORY

the equation for
<]>i
249
(0 and the first-order correction to the energy
Ek
is
Substituting
V from
(14.60),
we have
^
To eliminate
tion
(|i
tfc
"*"*
+^
ft(- *
+"
}'
14 65a >
-
the time
from
this equation, let us look for a solu-
(/)
in the
form
^
\Ve then
(/)
= ue - u <,-)
_(_
or-
(+*>
(14.66)
u
have the following equations for the functions

H
and v
(^ __ W ) _
(o) A
01)
H'} u ==
()
1 c*8.,
cojcfctl.
(14.67)
(14.68)
{^
+ -H
=1
note that these two equations have exactly the same form. Consequently, it is only necessary to solve Eq. (14.67) for , since the solution for v of Eq. (14,68) can be obtained from u by substiu> for w. Since the time does not appear explicitly in Eq. tuting we can find u by the perturbation method in the form ap(14.67), plicable to stationary problems. Thus we shall look for a solution in the form of an expansion in eigenfunctions of the unperturbed
We
problem [see
(14.8)]:
**'
where the
<|4"
(14.69)
satisfy the equation
(^-H'')<p2''==0.
(14.70)
Accordingly,
we may reduce
(14.57) to the
form
(14.67a)
where the frequency of radiation
is
^,.=^1^1.
(14.71)
250
Let us multiply (14.67a) on the left by <K* and integrate over all space, taking into account the orthonormality of the eigenfunctions
We
coefficients
then obtain the following equations for the

:
CV and the function u
(14.72)
~
where the matrix element xvk
is
**'.
(14.73)
equal to
:.
(14.74)
Substituting function v:
u>
for
u>
in (14.73),
we obtain an expression
for the
<
14 ' 75)
From
total
(14.64),
(14.66),
ty k
(14.73)
is
and (14.75),
it
follows that the
wave function
(t)
3-
^ J*^! N'*
^
(f )
cos
ico
sin
orf]|
(14.76)
From
the
wave function
field, we can readily obtain the polarization <^. In the classical theory we had
of the electron in the external vector of the medium
To generalize
p by
its
this expression to the quantum case, we must replace average value. Then
(14.77)
Substituting in g we have
<|> fc
(f)
from
(14.76) and retaining only first-order
terms
TIME- INDEPENDENT PERTURBATION THEORY

In deriving this expression,
we used the
relation
which follows from the fact that the integrand is an odd function of x. Comparing Eqs. (14.78) and (14.53a), we obtain the dispersion formula
(14.79)
4w
By introducing
the
new variable
(14,80)
which
is
called the oscillator force,
we transform
Eq. (14.79) to
Nel
(14.81)
Here let us make an observation similar to the one made in regard to the classical treatment: namely, if we had included the radiation damping in the quantum-mechanical treatment, we should have obtained a finite value of ri* for frequencies o> in the neighborhood of u k 'k (see Fig. 14.3a f the dotted line).
/7*-f
n*-1
UJ
a
Fig. 14.3.
b
Dispersion curves
a) positive dispersion C^/c" ^/c'/c)'
b) negative dispersion (a>ft~&>fcJk')
tion
Equation (14.81) has a structure similar to the classical equa(14.57). In actual fact, however, the quantum results are fundamentally different from the classical results. From quantum theory it follows that anomalous dispersion occurs in the neighborhood of frequencies corresponding to allowed transitions, and not,
252
as in classical theory, in the neighborhood of the eigenfrequencies of oscillation of the electrons. This particular conclusion can be seen to be directly related to the role of the oscillator force fvu
in (14.81), which is specified by the matrix element xvk [see (14.80)] which characterizes the selection rules (and thus the allowed transitions). This prediction of quantum theory was experimentally verified by D. S. Rozhdestvenskiy. A second, very important difference from the classical results is that quantum theory leads to negative dispersion (see Fig. 14.3b)a phenomenon which has no classical analog. This can be understood by simply noting that when light is scattered by excited atoms, it is necessary to take into account the states with Ek^>Ek for which
For these states the dispersion formula
(14.81)
becomes
*=
47i
V _!4*L = - *i w ufo
7^
k'
co*
'
(1 ll
and the dispersion curve is represented by the dotted line in Fig. 14.3b. The experimental discovery of negative dispersion was made by Ladenburg; thus, this prediction of quantum theory was also confirmed. Let us now find the value of the oscillator force /v* and the dispersion formula for a harmonic oscillator. The only nonvanishing matrix elements in this case are [see (10.55)] and **_,..- /
,
-,
"By chance,"
it
of radiation are tion
turns out that the quantum-mechanical frequencies identical with the eigenfrequencies of oscilla-
îi,
and
<k-l, *
w o.
We
thus obtain
/*!.*
Consequently
be written as
(sinceY
\
*'
f*. k
= (*+ = IK the
/
1),
/*-!.*
= -*.
(14 82)
dispersion formula (14,81) can
253
mQ
(14.83)
cog
can see that in this particular problem the quantum and classical theories yield the same value for the index of refraction n. The phenomenon of negative dispersion is not observed. The reason for this is that the regions of positive and negative dispersion coincide since |>*,i,ft| i = =|* i, J", so that the stronger effect of positive dispersion masks the negative dispersion.
:
We
G.
RAMAN EFFECT
Let us consider the phenomenon of dispersion from the standpoint of energy diagrams. Suppose that a photon with energy
6
*(o
(14.84)
Ev impinges on an atom with only three energy levels E k " (see Fig. 14.4). In general, the scattering of this photon (that is, dispersion) will be a second-order effect. The first form which this process can take is absorption of the photon. This is ac/ companied by excitation of an
electron from level k to some intermediate state (which may even be a forbidden state 7 see
;
<*<
hw
ficu'
ID
Fig. 14.4,1) and, subsequently, by emission of a photon. If, as a result, the electron returns to its initial state, it follows from the law of conservation of energy that the frequency <>' of the scattered photon is the same as the frequency of the inci>
hw
n
Fig.
1
1.4.
Energy
diagram
for
photon
scattering
h(o is the
dent photon. Alternatively,
energy of the incident photon,
the order of
tl)
I
the process may be reversed: the atom first emits a photon
the energy of the scattered photon, and II represent elastic scattering

,
(hu>ha) k i k and hw/-fla) kk "), III and IV represent induced transitions (hat^fia}^ k
(see Fig. 14.4,11) and then absorbs the incident photon. As in the preceding case, the frequency a/ of the scattered photon will be equal to the frequency o> of the incident photon if the atom re-
turns to
its initial state.
More precisely, the law of conservation of energy may be violated

states
It is
in intermediate
required to hold only in the final result
254
Finally,
resonance occurs when w^wwv*. In this case, both scattering and absorption of the photons take place (see Fig, 14.4, III); as a result of the last process the electrons in the atom undergo induced transitions. The probability of these transitions is given by the Einstein processes
coefficient B k * [see (9.21)]. An external field increases the number of downward transitions (see Fig. 14.4, IV), which results in some additional radiation proportional to the coefficient B**.
So far, we have been concerned

Fig. 14.5.
The Raman
of
effect.
with cases in which atoms return to
their initial state after scattering. It may happen, however, that after the atom has absorbed the incident photon, the electron does not return scattered photon corresponding to from the intermediate state to the "anti- Stokes" lines but instead makes a transilevel tion to the level k' or k" (see Fig. 14.5). In this case, the frequency of the scattered light (/ or /') is not equal to the frequency of the incident light. This type of scattering is called the Raman effect, after the Indian physicist who first discovered this phenomenon in liquids. In solids the Raman effect was discovered by the Soviet physicists L. I. MandePshtam and G. S. Landsberg (1928). From Fig. 14.5, it can be seen that the frequency of the scattered photon may be either lower or higher than the frequency of the incident photon. In the former case, the lines
is
hw
the
energy
the
incident
photon, ftea' the energy of the scattered photon corresponding to "Stokes" lines; and hu", the energy of the
t
= (D
correspond to excitation of the atom, since the atom ends up in a higher energy state. These lines are known as "Stokes" lines (the levels are shifted towards the red part of the spectrum). The second case corresponds to "anti-Stokes" lines (shifting towards the violet part of the spectrum):
u>"
0)
-{-
co;
these lines appear only when the light is scattered by excited atoms. It is obvious that at low temperatures only Stokes lines can be observed. As the temperature increases and the atoms of the substance begin to undergo transitions to excited states, antiStokes lines appear.
The
Raman
effect provides
much important
In
studies of molecular structure.
information in Chapter 12, Section C, we saw
255
that the rotational and vibrational levels (and also the vibrationalrotational levels), which provide data on molecular structure, are all located in the far infrared region of the spectrum and are very difficult to observe In studies of the Raman effect, a it is possible to use visible light in j>
,
,
determining molecular spectra, since these spectra are superposed on the

lines
light.
G>"
-w
'
w"
\
in
the
spectrum of the incident

Flg I4>6 superposition of the molecular frequencies on the frequency of incident light:
.
The experimental values
(See
of a/ and Fig. 14.6) yield the molecular
frequencies
u> k , k
a)
=
.
a)
(o'
and
-,
a)**''
spectral line
a)
in
the ab-
<o"
(0,
from which the selection rules can be

derived.
^Problemu*w 14.1.J Find
'\>
/.
,,
sence of molecular oscillations, and 6) shift of the spectral line due to molecular oscillations
the energy correction for a system in second-order perturbation
theory.
the the
Solution. Including the
terms up to and including the second order in the expansions of into wave function (14.3) and the energy E (14.4), and substituting these expansions Schrodmger equation (14.2a), we obtain the equation
(/:
- H) ^ = - (E'n - K')K - <K-
Since the solution <|* of the homogeneous equation^ H) 4*^=0 is orthogonal to the we have right-hand side and since we can substitute the expression (14.22) for f
The value
of
Vn n
,
is
given by
14. 15),
and we have used the relationship

v' v nn
,
v'*, n*
which holds for Herrmn an operators. We note that the second-order correction (14.85) to the energy of the ground state is always negative, since all the other levels En' are higher than the level j, that is,
Using the results of the perturbation theory, find the energy of the terms up to the order of fi a ; take the Hamiltonian of
Erern__l4.
arih"aTrn6nIc""o2ci ator including the the system to be equal to
where
V = ax* +
Solution.
P-^ (&e constants a and P are classical quantities),, The energy of a harmonic oscillator (V 0) is
= *(/! +>/.).
Taking
as the perturbation energy, the first-order approximation gives
En =Vnn^* (x
256
It
can easily be shown that
*U=Jl*.
since the integrand is an odd function. To calculate the matrix elements
P (x*) nn
we us e
ê multiplication rule for matrix
elements (see Problem
10.4), obtaining
Substituting the value of (x*) nk from Problem 10.5, for the first approximation of the perturbation energy
we
En".
obtain the following expression
Our problem, however,
is
not yet fully solved, since in the second-order approximation

;-
there is a contribution proportional to

the
~#
which arises from the first term of

this contribution into account.
A'
14
perturbation
energy
a,v 3 ,
and
we must
1
take
The
second-order contribution from the termjU'

it
is proportional
and accordingly to-^"-~ H\
may
the first
be neglected in our approximation. The second-order correction arising from term of the perturbation energy can be calculated from Eq. (14.85):
._
"'
\(
~
x *)nn'( x *)n'n
"'
"'
The only nonvanishing matrix elements are [see Eq.
(10.55) and
Problem
10.5]
n -^ ^
(n
__ 1)
(n
2) -;
where
A* O
Hence
Part
Relativistic
II
Quantum Mechanics
Chapter 15
The Klein-Gordon Scalar Relativistic
Wave Equation
RELATIVISTIC MECHANICS AND THE KLEIN-GORDON EQUATION. RELATIVISTIC INVARIANCE
The Schrodinger wave equation is nonrelativistic: it is suitable only for particles whose velocity v is much smaller than the velocity of light c. Itîs not invariant with respect to the Lorentz ;ransformations of tlie special theory of relativity since~there IS in asymmetry between the time and space coordinates^ (the/ equation contains a first derivative with respect to time, and second derivatives with respect to the space coordinates). According to the special theory of relativity, it is necessary for the time and space coordinates to be treated on the same basis. It is interesting to note that the de Broglie relations
p = Uk> E = Uu
are
(15.1)
In the Lorentz transformation, they relatively invariant. behave like a four-vector p^ with components
Pli
,,,=/>,
ft
=.
(15.2)
This indicates that

to
the
it is possible to generalize quantum mechanics case of particles traveling with a velocity of the order of
the velocity of light. A method of extending the nonrelativistic wave equation in a way consistent with the special theory of relativity was proposed
by Klein and Gordon in 1926. (This method was also put forward by Schrodinger and by Fok.) The simplest way of obtaining the Klein-Gordon equation consists in taking the relativistic relationmomentum p, and rest mass m of a ship between the energy
,
free particle
*
<?V
mjc
=0,
(15.3)
260
RELATIVISTIC QUANTUM MECHANICS
substituting the energy and
momentum
operators
iftV,
= tt-^., p =
<|>(r, t).
1
(15.4)
which act on the wave function 2 2 dividing by h c' we obtain

,
Replacing
by Uk^c and
=
Here
*,:=:*,
(15.5)
x*~y,
x<
z,
xi
ict
(15.6)
(a double occurrence of the subscript p, in a term indicates that it should be summed from 1 to 4). Since our initial equation was the
relativistic relationship (15.3), Eq. (15.6) is relativistically invariant and, therefore, it is symmetric with respect to the space and time coordinates. We shall not attempt to prove the invariance of the Klein-Gordon Equation more rigorously, and shall now pro-
ceed to examine
its
properties.
B*yTHE CHARGE AND CURRENT DENSITY

As in the nonrelativistic theory, equations for the charge and current density can be obtained on the basis of the equation of
continuity
We
t
.
of this
multiply (15.5) on the left by ]*, and the complex conjugate equation [that is, Eq. (15.5) with <1>* substituted for ] by Then we subtract the second of the resulting equations from
<!>
the first
(15.8)
After
some simple transformations, Eq.
(15.8)
becomes
=0.
(15.9)
Here and in the subsequent chapters, we shall not write ^ as a function of t, as was done m the Schodmger theory. In the wave equation for monochromatic (K ~ const) waves, for which only the time-independent part of the function must be considered, we shall use the energy eigenvalue instead of the operator.
THE KLEIN-GORDON SCALAR RELATIVISTIC WAVE EQUATION
261
Defining the charge density and the current density as
note that these expressions satisfy the equation of continuity 7). Moreover, they define a four-dimensional vector
ere
Xi
ict,
]\icp.
(15.13)
current density (15.11) is identical with the nonrelativistic jression (5.21), and the charge density (15.10) reduces to the
irelativistic expression (5.20)
when
i/*<c.
Substituting ih d
e (15.4)] into (15.10),
we obtain
p r
= ^V^<]),
7M C
J
(15.14) *
ich
becomes
)roximation
E~m $*
(
the usual expression

m
e^*ij>
in the nonrelativistic
ich the relativistic
Thus we have selected a normalization in values of p and / reduce to the corresponding
irelativistic expressions
It
when
^-J
<;
1 .
is worth noting that the definition of the particle density distinguished from the charge density)
es
rise to
some
difficulties
is
d^
dt
sin-Gordon wave equation

1,
in the relativistic theory. The a second-order differential equation
therefore, both
<|>
and
can be arbitrarily defined at some
-en
time
*ined,
t. Consequently, the density p n (15.15) is not positively unlike the nonrelativistic probability density
p.
= **t-
(15.16)
262
particle
Accordingly, the expression (15.15) cannot be interpreted as the density (that is, the "number of particles per unit volume"). The underlying reason for this is that the same relativistic equation describes particles with either positive or nega-
charge (and, indeed, * mesons, to which the Klein-Gordon equation is applicable, may be either positive or negative in charge). The quantity p therefore, can have both signs.
tive
,
C.
RELATIVISTIC THEORY OF THE HYDROGEN
ATOM
(NEGLECTING THE ELECTRON SPIN)

treat the interaction of a particle with an electromagnetic (defined as usual by a vector potential A and a scalar potential <t>), we introduce the same operators as in the nonrelafield
tivistic
To
case
(15.17)
From
(15.3),
we can
obtain the Klein-Gordon equation
We
shall use this equation to study the
like atom. Setting
4=0.
and
V=e
spectrum of the hydrogen~* in (15.19), we have

(15.20)
^ + ^v {(E-lO'-mJcnÔ.
substitution
Since the potential energy does not depend on time in this equation, we can transform to the time-independent case by means of the
In this equation, we have not included the rest mass energy the particle in the energy E. As a result, Eq. (15.20) becomes
m^
of
(15.21)
This equation can also be obtained from the relativistic Hamiltonian for a particle in an electromagnetic field
"'O^
is only
(15 18)
It
p and A by
necessary to transfer <?$ to the left-hand side, square both sides, and replace their quantum-mechanical operators.
THE KLEIN-GORDON SCAU\R RELATIVISTIC WAVE EQUATION

Just as in the Schrodinger theory, this equation in the form
263
we
shall look for a solution to
,?).
(15.22)
The equation for the radial part
is
=
Here
a
o.
(15.23)
= -t|-^y~
is
a dimensionless quantity, called the fine
structure constant, and
When c*---oo, the expressions (15.24) reduce exactly to their nonrelativistic counterparts (see Chapter 13). The somewhat improved values that we have obtained for A and B by taking into account the relativistic effect do not change in any way the general character of the solution that was obtained
in
nonrelativistic theory.
Formally, the additional term -^-
Z*oc
in
Eq* (15.23) can be treated as a relativistic, attractive potential energy, which obeys an inverse square law and which affects the solution under certain conditions. A detailed analysis of the role of this term will be given below. First of all, let us consider the asymptotic solution /? as /-+0. In this case, Eq. (15.23) reduces to
--
~dr*
--s
7*
p ~~ Q
We
shall look for a solution of this equation in the
form
We
then obtain an equation for
s(s+
the solution of which is
!)
/(/+
1)
+ ZV =
0,
(15.26)
- -~
(15.27)
Consequently,
(15.28)
264
If
both roots s, and s, will be real for all values of/ (I = 0, 1, 2, . . .). We retain only the solution for r/? that does not diverge at r = 0; that is, we set C a 0. Similarly, only the exponentially decreasing oo should be kept when solution for the wave function as r The asymptotically decreasing solutions for the two <0 (A >0). r of values yield the same equation for the energy spectrum limiting as in the Schrbdinger theory, as can be seen from Eq. (13.20) by
substituting
Sj
for
/.
Thus, for the eigenvalues we have the equation
V.
(15.29)
Substituting the relativistic values (15.24) of the constants A and B, we have (for n = k + I + 1 )
(15.30)
expression into a series of powers of Z 2 a 2 and first two nonvanishing terms, we obtain an retaining energy spectrum which includes relativistic effects:
Expanding
this
only the
n
1
--
3 \1
'
T-
T]
/J
(15.31)
identical to the nonrelativistic expression. The is proportional to the square of the fine structure constant a ^- 1/137, gives the relativistic correction. The relativistic correction for the hydrogen atom (Z is ) interesting because it removes the degeneracy with regard to /. The level for a given n is split into n closely spaced sublevels 2 (the close spacing is a consequence ofthesmallness of a ) because the orbital quantum number / can assume n different values = 0, 1, 2, . . . , n 1 In order to compare these results with (/ ). experiment, let us compute the distance between the doublet states of the Balmer series (rc 2). We find
first
is
The
term
second
term,
which
=
A, A
21
rt
20
~~~
*.&*.
3
16
/l^ *^ iio.twj
states
Experimental data show that the actual distance between doublet in the Balmer series is only one third of the distance
THE KLEIN-GORDON SCALAR RELATIV1STIC WAVE EQUATION
265
given by Eq. (15.32). This discrepancy arises because the fine structure of the hydrogen levels cannot be explained entirely in
terms of the relativistic relationship between mass and velocity. As we shall see in Chapter 19, it is also necessary to consider the
electron spin (that is, the intrinsic angular momentum of the electron, which gives rise to a magnetic moment). At first it was thought that the Klein-Gordon equation could be used to describe a relativistic electron. As a result of the discrepancies between its predictions and experiment, however, it was established that it describes particles with spin of 0, whereas the electron has a spin of 1/2. Consequently, the Klein-Gordon equation can be used for TU mesons, which have a spin of 0. Finally, let us consider the case in which
Za>-~
in
(15.33)
Eq. (15.27). In this case, the solution does not consist of a correction added to the nonrelativistic solution, but is fundamentally both roots s and s4 are complex. Therefore, new. Indeed, for 1 the asymptotic solution (15.28) is
flo
= y=f (C,'T' + Ctf- n

l
(15.34)
where
Ci = 0)
7= I/ 2V
on
--.
We
cannot impose the condition C 2
(or
our problem because both solutions have the same as r-+Q. Since the solution remains unrestricted by a singularity potential barrier at small r, even when E<0 the energy spectrum is continuous. Consequently, the particle will "fall" to for
J
the center.
The question of the stability of the motion of the particle is very important in studies of the central forces. The above results can be used to analyze the solution of the Schrodinger equation in the general case of an attractive potential
On obtaining the asymptotic solution for r->Q [15.25], we see that the solution will vanish at the origin only for a maximum value of q equal to 2 and that the particle will not fall to the center if
It is
of
3 interesting to note in this connect ion that the r dependence potential energy occurs fairly frequently in the theory of
266
elementary particles since the potential energy of this form characterizes the inter action of two elementary magnet icdipoles. Actually, two cases have to be distinguished. In the first case, V ^r* only at relatively large distances, while at small distances V varies as a r" . This behavior of V is observed for the spin-orbit interaction in the Dirac theory and it does not give rise to any difficulties. Moreover, it is found that stable motion corresponds to a value of Z greater than any in the present periodic system of elements 137) because the spin effects reduce the influence of the (Z relativistic effects (whereas in the Klein-Gordon theory Z is confined to relatively small values Z y 137). 3 In the second case, V continues to vary as r~ even at small distances (r->0), and particles cannot be combined into atomlike systems. This case can be observed in the meson theory of nuclear forces, where quasi-magnetic interactions are of considerable importance. The formation of an atomlike system becomes possible only if the potential is cut off at small distances
<
<
from the
origin.
""
Problem 15.1.1 Fi nd an expression for p and j if the scalar relativistic equation conTBlllt'lffllug from the presence of an electromagnetic field.
Solution.
Let us substitute p
A,
E -* E
e<b
ln the Klein-Gordon equation,

field.
which now describes the motion of a particle in an electromagnetic
Repeating the calculations that lead to Eq. (15.15), we obtain the generalization
where the four-dimensional
potential
A^ has
the
components
i^Problem 15.2A Show that in the case of time-independent potentials A and $, the space* n liiiLiiini 'TfCordinates in the Klein-Gordon equation can be separated and the wave function can be written as
1
1
Find the wave function of a free particle described by the KleinGordon equation using for normalization the expression for the density p . Show that in
the relativistic case, p is the charge density rather than the particle density. Solution. Suppose the momentum of the particle is directed along the z axis and that the particle travels in a segment of length L (the one-dimensional case). The solution of the Klein-Gordon equation can be written in the form
--=
THE KLEIN-GORDON SCALAR RELATIV1ST1C WAVE EQUATION

where
267
by the
E=
cMC
is the
energy of the particle. Since the charge density

teH
is given
expression
f=
the total charge is [see (15.14)]
-1/2
It follows that p is the charge density rather than the particle density, since this relation can be interpreted only if it Is assumed that particles described by an amplitude B (negative energies) have a charge of opposite sign from particles described by an
amplitude
(E >0).
Chapter 16
Motion of an Electron
in
Magnetic
Field.
Electron Spin
In 1896, Zeeman found that when atoms are placed in a strong magnetic field, their spectral lines are split into several components. This phenomenon is known as the Zeeman effect. The Zeeman effect played an important role in the investigation of the structure and magnetic properties of the atom. It led in particular to the discovery of the spin (intrinsic mechanical moment) and magnetic moment of the electron. Accordingly, it is worth elaborat-
ing the theory of this effect in
some
detail.
A.
THE CLASSICAL THEORY OF THE ZEEMAN EFFECT

in Lorentz* selection
The simplest model of the radiating atom

theory is
based on the assumption that the electron moves under the influence of an elastic force
F=-kr.
The
elastic
(16.1)
constant
ft
is
angular frequency of oscillation
related to the electron o> by the expression
mass and
the
ft=:w
o)J.
(16.2)
The equation for the oscillations

constant magnetic field
//,
of an electron in a
homogeneous,
therefore,
becomes
(16.3)
m /M-/W==-~7 fxH,
where
e
is
the electron charge.
(16.3) along the coordinate tion of the field //, so that
H x = Hy = 0,
axes (the
axis
Taking the components of is chosen in the direcH, <JtT), we find
(16.4)
MOTION OF AN ELECTRON
Multiplying the first, we obtain
IN
A MAGNETIC FIELD
i
269
it to
second equation by
(i= K-T)
and adding
the
(16.5)
where
is the
jLtn^c
Larmor frequency of precession and

of (16.5) is of the
= *-{-
iy.
For o the solution
form
(16.6)
and
it
follows from (16.4) that the
component
is
z=:CeM.
(16.7)
From the above expressions it can be seen that the frequency of oscillation of the electron (a three-dimensional oscillator) changes under the influence of a magnetic field. An atom placed in a magnetic field should emit radiation at three frequencies:
0)
0,
OJ
0l
(t)
-j~0.
(16.8)
According to the classical theory, an oscillator does not emit energy in the direction of oscillation. Therefore, when we observe the light emitted by an atom in the ? direction (the direction of the magnetic lines of force), we are able to detect only two lines (there will be no component <> due to oscillations along the z axis). In other directions, we are able to observe all three components (the normal Zeeman effect). Equations (16.6) and (16.7) indicate that the oscillations are resolved in a longitudinal component in the direction of the magnetic field (the z axis) and two transverse components corresponding to two directions of rotation (a righthanded rotation and a left-handed rotation). Thus, the magnetic field has no effect on the longitudinal oscillations and acts only on the circular rotations in the plane perpendicular to the magnetic
quantum theory a change in the frequency of oscillation is always associated with a change of energy. At first glance, it may seem strange that the magnetic field changes the energy of the
electron, since the Lorentz force
the velocity, and therefore the
field. In
F=
vxHis
perpendicular to
the the
work done by any

Even
in
centripetal force,
work done by this force, just as must be equal to zero. On
order of 10
the case of very strong fields (&%* '^ 10 gauss) the quantity o is of the sec whereas the frequency of oscillation of an electron in an atom (the
,
optical spectrum) is always satisfied
10
sec"
Therefore, the inequality
u>
Q is practically
270
RELAT1VISTIC QUANTUM MECHANICS
other hand, it is well known that an electron rotating in a circle (that is, a current loop) forms a dipole. The energy of this dipole in a magnetic field is equal to
'
ym*g
/f.
(16.9)
These two conflicting conclusions may be explained in the following manner. As the magnetic field changes from zero to a certain
constant value
the electron experiences a force directed 9 along one of the components of the electric field g. This force
H g = <W
imparts an additional energy to the electron. The magnitude of this component of the electric field can be found from Faraday's law of induction (second Maxwell equation):
Assuming
orbit,
that ^T and g depend only on time and that the switching on of the magnetic field does not alter the radius of the stationary
we
find
2c~dt
The additional velocity imparted to the electron electric field can be found from the equation
dt
(e=
by this
2w
dt
which gives
0mag_ J?JL 2/w f

As we can
see, f ma e
is
is
o^T
independent of the rate of change of the switched on. Since the magnetic field is directed along the z axis and the induced electric field (and therefore also u ma s) is perpendicular to it and to the radius of the orbit, we may write in vector form
magnetic field when
it
From
this
it
is
rotation
of
the
clear that the magnetic field produces an additional electron (Larmor precession) with an angular
velocity
0=3^.
now determine the unknown additional energy acquired by a rotating electron when the magnetic field is switched on. Since the energy of an electron placed in a magnetic field is determined entirely by its kinetic energy, we have
We may
IN
A MAGNETIC FIELD
271
is the velocity of the electron before the magnetic field is switched on. Retaining only terms that are linear with respect to v ma s, we obtain
where v
V mag
Comparing
of
= ^^Hxr)=: ^ff^ rXv)f

we see
is
(16.10)
(16.10) with (16.9),
an electron moving in a circle

p
that the magnetic moment given by the expression
= =m
g r*v.
is
(16.11)
Recalling that
its
angular
momentum
,
equal to
x #,
we
find a simple relation
between these two quantities
>=-&*
(16 - 12)
It is worth noting that the magnitude of the magnetic moment can also be found from other considerations. As we know, the magnetic moment of a current loop is
-?*
where
In the
n* is a unit vector normal to the plane of the current loop. above relationship the current strength is equal to
where T
=^
r
-
is
is
the period of rotation, while the area enclosed
by the current
Combining these last relations we again obtain the expression (16.11)

for the quantity ^ .
B.
THE ZEEMAN EFFECT IN THE NONRELATIVISTIC SCHRODINGER THEORY
order to obtain the Schrodinger equation for an electron moving in a magnetic field, we shall use the general rule for transformation of the classical Hamiltonian to the quantum case (see, for
In
272
example, Chapters). To do this, we substitute the momentum operator p into the classical expression for the energy of an electron in the presence of electrostatic and magnetic fields. The Schrodinger equation for the central forces in the presence of a magnetic field characterized by the vector potential A then takes the form
(~HS)^ =
where the Hamiltonian
of the Schrodinger equation is
(16.13)
niQ
(16.14)
and the operator P
is
called the generalized
momentum
operator. For the case of a constant, homogeneous magnetic field Hy Q H Z <F) 9 we may write directed along the z axis (H x
= =
Using the fact that
where
and neglecting the terms proportional

field strength
*#',
to the
square of the magnetic
we
find
=
Remembering
that
0.
(16.15)
where
we reduce
the Schrodinger equation to the
form
IN
A MAGNETIC FIELD
273
It is easy to show that this equation is satisfied by the usual wave function for a centrally symmetric field:
t== /?( r )r(ft
cp).
(16.18)
Substituting this solution into (16.17) and recalling that
we
obtain the equation
which also includes the effect of the magnetic This equation may be written as
field on the
atom.
C16.19)
where we take the charge
of the electron as
The last term in the Hamiltonian may be attributed to the presence of the orbital magnetic moment of the atom, which gives rise to an additional energy
j!{;
(16.20)
Therefore, the orbital magnetic the Schrodinger theory is
moment
obtained on the basis of
Recalling the expression for the
component of the angular momen-
tum
Lg
= hm,
the
we
obtain the
angular
same relation between the magnetic moment and momentum as in the classical theory [see (16.12)]:
=It
<
16 - 22 >
moment are
follows, therefore, that the components of the orbital magnetic multiples of a certain unit magnetic moment
274
^=5fi
which
= 9 273
-
'
10
~ 21
erg.gauss-1,
(16 . 2 3)
is one of the most can be seen from It an atom. important magnetic properties electron placed of an orbital additional that the energy Eq. (16.20) in a magnetic field is given by the expression
is called the Bohr magneton. The orbital magnetic moment of an electron
of
(16.24)
since V mas is a constant, where o is the Larmor frequency. Because of the selection rules for the magnetic quantum number (A/n 0, :+- 1), the additional radiation frequencies due to the Zeeman splitting are the same as in the classical theory [see (16.8)], namely,
A/ii
0,
+o.
(16.25)
The ilormal Zeeman

and doublets)
is
splitting
of the spectral lines
(triplets
encountered only in the case of a strong field (the Paschen-Back effect) or in the case when the total spin of the electrons in the atom is equal to zero (for example, in parahelium, whose outer shell consists of two electrons with oppositely directed spins). In cases in which the spectral lines are split into more than three components, the Zeeman effect is said to be anomalous. The anomalous Zeeman effect is connected with the spin properties of electrons, and an explanation of this effect can be constructed only on the basis of Dirac's theory, which takes into account the
spin effects (see Chapter 20).
C.
THE EXPERIMENTAL DISCOVERY OF ELECTRON SPIN
It was shown in the last section that is able ^theJfohrQdingar theory e^plaii\_only the orbital angular momentum and magnetic mofffent The basic equations character! zing these prop'er ties are Eq. (16.22) for the ratio of the orbital magnetic moment and
^
Let us note that the reasons for the use of the terms "normal Zeeman effect" and "anomalous Zeeman effect" are purely historical Before the discovery of electron spin,
only
the classical theory of triplet
splitting (normal
Zeeman
effect)
was known When
more complicated splitting was discovered it was called the anomalous Zeeman effect because no theoretical explanation could be given for it until the development of the theory
of electron spin
IN
A MAGNETIC FIELD
275
the orbital angular momentum and Eq. (16.24), which indicates that the number of possible orientations of the magnetic moment relative to the z axis is necessarily odd, since the number of states with different quantum
is equal to 2/ + l.TheSchrodinger theory, however, does not adequately account for all the experimental data, the analysis of which led to the discovery of the spin properties of electron. Let us briefly discuss these experimental results. 1. First of all let us consider the Einsteinde Haas experiment (1915), which was carried out in order to verify Eq. (16.22):
numbers m
carrying coil, 3) ferand an angular momentum whose magnitude ro-magnetic rod. can be determined from the angular rotation of the quartz fiber. If an alternating current is passed through the coil, an alternating torque will arise, causing torsional vibrations in the ferromagnetic sample. In addition, resonance can be used to enhance the rotational effect. Experimental measurements of the gyromagnetic ratio (16.22) show that the sign of this ratio is negative, so that it can be definitely concluded that the magnetization of the ferromagnetic sample is due to the motion of electrons. The value of the Lande g factor, however, turned out to be equal to two ( rather than to the unity that was required by the 2), classical or Schrodinger theories. This g value was not explained until the development of the theory of electron spin (see below).
where g , the Lande factor, should be equal to unity for orbital moments. In this experiment, a ferromagnetic rod is suspended on a quartz fiber and magnetized by passing a current through a coil (see Fig. 16.1). As a result, the rod acquires a magnetic moment
Fig 16 1. Diagram of the Einstein-do Haas experiment for the determination

of
1)
the
Lande g
fiber,
factor.
2)
quartz
cur-
rent
<?mag H(z)
II
n
Fig. 16.2. Diagram of the Stern-Gerlach experiment for the determination of the magnetic moment of monovalent atoms
276
2.
atoms
Stern and Gerlach (1921) studied the behavior of a beam of in an inhomogeneous magnetic field in order to check the
theoretical result (16.23)
which describes the spatial quantization. In their classical experiments a beam of monovalent atoms (hydrogen, lithium, silver), traveling along the x axis, crossed a magnetic field directed along Hy Q H z e?f). This magnetic field was very the 2 axis (H x that it had large gradients. Then a magnetic so inhomogeneous,
= =
=
J*
dipole of
moment
= *mag'.
" and
/
(16.26)
is the length of the
z
where
m ag
is
the
"magnetic charge
dipole, will experience a force directed along the

1
axis 3
- ^ - COS = = p,^ = -to d-!Jfm.

(*
*)}
(16.27)
Let us calculate in a simplified fashion the displacement experienced by a particle under the action of the force Fz during the time t. If the particle moves with a velocity v perpendicular to the magnetic field (that is, to the z axis) and travels a distance L vt the displacement along the direction of the z axis will
equal
(16 ' 28)
In this
case the acceleration

and M
is
w= ^
where the force Ft
is
given
by (16.27),
particles
is the mass of the atom. Consequently, a beam of will be split into possessing a magnetic moment components as it passes through an inhomogeneous magnetic field. The number of components is determined by the possible number of projections of the magnetic moment ^ on the direction of the
|u,
field.
of
and Gerlach studied the splitting s state. In this state, the angular momentum and consequently the magnetic moment of an atom are m 0), and therefore there shouldbe no splitting. equal to zero (/ If the electron is in the p state 1), then triple splitting should
In their experiments, Stern a beam of atoms in the
= =
(=
it
We note
dipole,
it
that in order to determine the motion of the center of
mass
of the magnetic
is
quite immaterial whether
we regard
as a rigid dipole or a current loop
IN
A MAGNETIC FIELD
277
be observed because of the three possible values of the magnetic quantum number
Experiments on hydrogen, lithium, silver, and other atoms show, however, that the beam is split into only two components. This proved the existence of a magnetic moment for atoms in the s state. The projection of this magnetic moment on the z axis can assume only two values. The measurements of the quantity p. showed that it is equal to one Bohr magneton
<
16 - 29)
In order to reconcile the results of these two classical experiments, Uhlenbeck and Goudsmit introduced the hypothesis that an electron posses an intrinsic angular momentum in addition to its orbital angular momentum. At first it was believed that this intrinsic angular momentum could be treated by analogy with a top spinning about an axis, and therefore it was called the electron spin. It must, however, be emphasized that no rigorous classical theory of spin exists. According to the hypothesis of Uhlenbeck and Goudsmit, the intrinsic angular momentum of an
electron is equal to
sz
~h;
its
\
(16.30)
that is, the

z
quantum number characterizing
projection on the
.
axis takes on half- integral values
ms =
The
important
distinction between the integral (orbital magnetic quantum number m) and the half- integral (spin quantum number /;/, ) values of quantum numbers lies in the number of possible states. Integral quantum numbers always give us an odd number of states (for / we have one state m Q\ for l \ there are three states On the other hand, half- integral 1; and so on). m=0, -|-l f quantum numbers give us an even number of states (for example,
for
1
-
there are two
states
m s = -\-^ ~~7
t
113
*
r 5==
lf
^ ere
are four states; and so on).
The assumption
of the half-integral
quantum numbers was
introduced even before Uhlenbeck and Goudsmit in order to explain the double splitting of terms for the monovalent atoms. The SternGerlach experiment showed that there are two possible electron states in a monovalent atom; that is, the electron spin must be
278
RELATIV1STIC QUANTUM MECHANICS
described by the half- integral quantum numbers corresponding to two opposite orientations. Recalling that the Einstein-de Haas experiment showed that the Lande g factor in Eq. (16.29) is equal to two (g~2) and the intrinsic angular momentum is given by Eq. (16.30), we find the following expression for the z component of an intrinsic magnetic moment:
^ = -c*. =
which
=FI*o.
(16.31)
is simply one Bohr magneton. The introduction of the electron spin also made it possible to explain the multiple splitting of the spectral lines of atoms, as well as their magnetic properties.
D.
PAULI EQUATION
that
nonrelativistic
wave equation
includes
the intrinsic
magnetic moment of the electron was first proposed by Pauli. For this purpose he took the ordinary Hamiltonian of the Schrodinger equation and added to it a term represent ing the interaction between the magnetic moment of the electron M- and the external magnetic
field
l/
\L*H.
(16.32)
Then the time- independent Schrodinger equation takes the form

{
HS -\p. H}ty
II s
Q,
(16.33)
where the Hamiltonian
is
HS
Next,
the
it
=2
to find suitable quantities to describe of the electron. It is well known spin is related to introduction of a
was necessary
intrinsic
magnetic moment
that introduction of the fourth quantum number, characterizing the internal properties of an electron. On the other hand, the wave function ^ of a particle depends only on three quantum numbers, corresponding toquantization of the three spatial coordinates. Jn_order to describe spin, Pauli introduced two wave functions Wj and U^JTifplace of the single Jwavejunctlon <J. One of the wave functions describes a state 'With one spin orientation and the other wave function describes a state with the opposite spin orientation. The actual wave equation represents a system of two equations. It is possible to represent a system of two or more equations, such as
IN
A MAGNETIC FIELD
279
by a single equation in matrix notation

'
<
16 ' 35a >
where the multiplication

the
is carried out according to the rule for of a matrices (c) (a)(b): namely, an element matrix product is obtained by multiplying each element in the appropriate row of the first matrix by the corresponding elements of the appropriate column of the second matrix and taking the sum
multiplication of
of these products, that is,
*i*
= 2fliA*n
*F
(16.36)
Pauli suggested selecting the wave function
in the
form of a one-
column matrix W
/ur \ l
{
\^3'
and setting the intrinsic magnetic
moment
of an electron equal to
ji,
^0',
a'
(16.37)
where
|A O
is the
Bohr magneton and
stands for the three
2x2
Pauli matrices
These matrices are denoted by the letter a with a prime (the same without a prime will be used to denote the 4 x 4 Dirac matrices). These matrices characterize the components of the
letter
spin vector along the coordinate axes. Using the rule (16.36) for matrix multiplication, it can be readily shown that the square of each Pauli matrix is equal to unity
aj'ââ.; ^!,
(16,39)
where
I'
denotes a 2 x 2 unit matrix L

j
It
can also be shown
that different
matrices antic ommute with one another:
(16.40)
280
In
terms
of the
above matrix expressions, the nonrelativistic
Pauli equation has the form
->.]}(;:;)=.
equations,
This matrix equation is equivalent to a system of two ordinary each of which corresponds to one of the rows of the
matrix
(_HS
(E _.
-ptHJVt-friHs-iHJV^Q.
.f
|JLo
HS
//j V,
(A()
(//,
+ tHy
\V,
o.
Let us consider the case of an electron moving in a magnetic field H V =(\ //,=- JT). Using the Hamildirected along thez axis (H X tonian (16.19), which includes the effect of a magnetic field, we obtain two equations of motion for the electron
E + e.<S> -
wX'm + \^' ~
2J
'
1r i
Uf a
= =
0,
0,
where iv^'/// is the energy of interaction between the magnetic field G>V"and the orbital magnetic moment, and driv^' is the energy of interaction between the magnetic field and the spin magnetic
moment.
to zero,
In the s
state the magnetic
quantum number m
is
equal
so that the Pauli equation takes the form
(16.44)
describes a state in which the intrinsic of the electron is parallel to the z axis, and the wave function ^3, a state in which the magnetic moment is antiparallel to the z axis. These are the orientations of the intrinsic magnetic moment which were observed in the Stern-Gerlach
that is, the
wave function
magnetic moment
experiment.
of
the function *F h Pauli suggested taking the Hermitian adjoint + W, that is, the matrix \F =(V^J) whose elements are obtained by taking the complex conjugates of the elements of UT
t
As
and transposing them (interchanging rows and columns). Thus,
if
a column matrix, density will be given by
is
IN
A MAGNETIC FIELD
281
\F
will
be a row matrix. The probability
^^-pjnqr^
(16.45)
which includes the possibility of two spin orientations. The other matrix elements are formed in a similar manner. For example,
_ ^<*'~~^
UT*IF
urnir
"
(16.46)
that is, ur?^, and UTJ^ represent the probability densities of states in which the electron has a spin orientation parallel and antiparallel
to the z axis, respectively. Using the expression for the intrinsic magnetic moment in the Pauli theory
and the Einstein-de Haas relation between the intrinsic magnetic
moment and
the angular
momentum
e
r v.=--
s,
/fluf
we
find that
(16.47)
Thus, in agreement with the other experimental facts, the z component of the spin angular momentum is equal to iL V, . Since the spin operator is expressed in terms of the matrices a', the spin components do not commute. In this they resemble the components of the orbital angular momentum, which are operators depending on derivatives [see (11.75) and (11. 76)]. The commutation relations satisfied by the spin operators can be easily established from (16.40) and (16.47):
(16.48)
282
Concluding our discussion, we note that the Klein-Gordon theory, which includes the relativistic effects but neglects the spin effects, and the Pauli theory, which, on the contrary, neglects the relativistic effects but includes the spin effects, were predecessors of the more rigorous theory of the electron developed by Dirac, which predicts all the elementary properties of the electron?) It should be noted in this connection that the absolute value of tTre intrinsic magnetic moment was introduced in the Pauli theory from purely empirical
considerations.
that in nonrelativistic quantum mechanics, just as in the classical Zeernan effect is due to the precessional motion of the orbit in a magnetic field, the motion having the Larmor frequency. Solution. From the Hamiltonian of the Schrodinger equation for the case of an electron moving in a magnetic field directed along the z axis
^ P^fthlfilD,jLfiji"* Show
theory, the
we can
find the time derivatives of the angular
momentum
o is the Larmor frequency. It follows that the component of the angular momentum in the field direction (z axis) is a constant of the motion. The components along the x and y axes, however, precess around the z axis with the frequency o .
where
>lem 16,2Ji
1 inear
Show
that the spin operator S is vectorial; that is, if
we construct
the
comTSmatton
S;.
= a,S + PiSj+TiS-,
v
S'
= a SA +
a
PsSj,
+ T&,
where
a,
(J,
are the directional cosines, then

S;S'V
S S; V
= is;
and so on.
roblern 16.3J Show that in a homogeneous magnetic field which is a function of time only, BUT Wave function of the Pauli equation can be resolved into a product of coordinates and spin functions. What form does this solution take if the field is time independent? Solution. Let us look for a solution of the Pauli equation in the form
It is readily shown that the coordinate part of the wave function ^ ordinary Schrodinger equation without the spin
(r,
t)
satisfies the
while the spin part of the wave function
may
be obtained from the equation
The spin part
of the
IN
A MAGNETIC FIELD
283
wave function
is
normalized as follows
In the case of a stationary magnetic field part in the above equations. We simply set
it
is
easy to determine the time-dependent
'
(E-E)t
Then the time-independent parts
of the
wave function are determined from
Find the eigenvalues of the operator of a spin component along the the spherical angles d and <p. Investigate the particular cases in which this direction is the x 9 y or z axis. Solution. Consider first the case in which the spin is directed along the z axis. Then the initial equation takes the form
16.4.h
direction
SpWlH&a by
where
s
_ft
/i
o'
This matrix equation
is
equivalent to two homogeneous algebraic equations
-- C.
+ XC, =
0.
The normalized
solutions of these equations have the
form
first evidently corresponds to the case in which the spin is directed along the z axis; the second to the case in which the spin is directed along the -z axis. The operator for the component of spin along the direction defined by the spherical and y with respect to the coordinate axes is equal to angles
The
sin
ft
cos
<p
S*
sin
ft
sin
<?
S y
cos & S^,
where
284
RELATIVIST1C QUANTUM MECHANICS
Hence from the equation

*
(?;}='()
is parallel to this
we
find two solutions:

(a) the solution
corresponding to the case in which the spin
direction
}}
sin
l e *
(b) the solution corresponding direction
to
the
case in which the spin
is antiparallel to this
cos
\
<?P ^
J
Setting
ft
0,
<p
we obtain
the
same
directed along the x or y axes
may
solution as above. Hie cases in which the spin is be obtained by setting, respectively,
x^ Problem Ib-ST^The electron spin is parallel to the 2 axis. Find the probability that the component of the spin (a) in a direction parallel to the i axis, and (b) in a direction
making an angle
D with the t axis, will
have the values y h and

-,
hm
Hint. Take the wave function describing the state in which the spin is parallel to the axis and then expand it in terms of the functions corresponding to the cases in which the spin is parallel and antiparallel to the direction forming an angle & with the z axis. Both these functions are given in Problem 16.4. Without loss of generality we may set the angle cp 0. 'ITien the squared modulus of the expansion coefficients gives the probabilities
z
of the
components
of the spin along the corresponding directions; these are equal to
In
order
to find the
x component
of the spin,
we must
set d ==
in the last equations.
Chapter
The Dirac Wave Equation 1

A.
LINEARIZATION OF THE ENERGY OPERATOR. DIRAC MATRICES AND THEIR RELATION TO PAULI MATRICES
As indicated in Section 15, relativistic quantum mechanics is based on the well-known relativistic relation between the energy E, momentum p and rest mass m
9
:
= cV7* +
//#*.
(17.1)
To obtain the wave equation describing a free particle, the appropriate operators into this equation
E
we
substitute
= 0lj
= -f*v.
It is
(17.2)
and act with these operators on the wave function. however, to make a direct transition to operators
^
impossible,
in (17.1)
because
the radical sign|frj * s therefore necessary to get rid of the square root"Tn^(tT.l). Cm? way of doing this is to take the square of Eq. (17.1). This gives the relativistic Klein-Gordon wave equation with a one-component wave function. 2 As already noted, this equation describes the motion of spinless particles and is not applicable to electrons, whose spin is equal to 1/2 (in units of ft). A different method of obtaining a linear relativistic wave equation was adopted by Dirac (1928). This method gave a firstorder wave equation and consisted in linearizing the relation (17.1). It led to the discovery of the relativistic wave equation for the electron. This equation plays a fundamental role in relativistic quantum mechanics and quantum field theory since it provides a
êe P A M
Press, 1958
Dirac, Principles of
Quantum Mechanic?, New York
Oxford University
More exactly, we
in fact
cot since a second derivative with respect to time appears in the fundamental equation One degree of freedom corresponds to particles with positive energy, the other to particles with negative energy It was shown by Pauli and Weisskopf that the negative energy states can be eliminated by carrying out a second quantization of the scalar equation and introducing
have a function with two components 0j ~
/>
and
\f/2
~~5r
spinless particles with charges of opposite signs
286
suitable
description of the motion of particles of spin 1/2. The discovery of this equation was the most important advance in the theory of the electron since the Maxwell- Lorentz equations of classical electrodynamics. Bohr's semiclassical theory and nonrelativistic quantum mechanics served only as transitional theories. The relativistic relation between energy^ and momentum is
inearpEjC&y, .êxtr^^ normal with .the aid of matrices or m~

1
"f
>
For
this
purpose we represent
where
p
()
=m
()
pi
=p
xt
Pi
=P
PA
PZ-
(17.4)
We
note that
To determine what conditions the quantities a must satisfy, we square both sides of (17.3). Then, if the operators p^ and /V commute, we
,
jJL
P*'W
1
=Y
JJ.
|JL
[J.
p.'
Equation (17.6) coincides with (17.5) only
if
VV + VV = 2S
that is, all four quantities a
a
(L
;i
iM^'
(17. 7)
anticommute with one another

ji^n,'
V + VV =
is
(17.8)
and the square of each of them
equal to unity
(17.9)
aj=l
We
recall that the
2x2
Pauli matrices also possess analogous
properties
if there is no electromagnetic field ThereDirac proposed that one should first extract the square root of the operator for a free particle, and then generalize the resulting equation to the case when fields are present
These operators commute with each other
fore,
THE DIRAC WAVE EQUATION
287
since they anticommute [see (16.40] and the square of each is equal To extract the square root of the four-term polynomial, however, it is necessary to have four relations (17.7) = 0, 1, 2, 3), instead of three (fj, [Eqs. (16.39) and (16.40)] that are satisfied by the Pauli matrices. Accordingly, Dirac proposed that we take a system of 4 x 4 matrices n and p that are related to the 2 x 2 matrices by the
to unity [see (16.39)].
expressions
(i7.il)
(17.12)
where
a' n
are the Pauli matrices
Hence we
find, for
example,
/O
1
0\
aj== (o o o ? \0 1 O/
andsoon
The properties of these matrices are similar to those of the Pauli matrices, as may be easily checked by direct multiplication. In particular, it turns out that their squares are equal to unity
oi
= =
P
(17.13)
or,
more
exactly, are equal to the

/I
4x4
Ov
unit matrix
I==
(o
\0
1
Or
1
(17.14)
As
in the
case of Pauli matrices, we have

i
<3
= = On(v
32<*i
3 .i.
Pip2
P^Pi
ipa
and so on.
_v
rv0(i
(.'!,
2i3 >
(17 ' 15)
From
this
it
one another
matrices):
(a
follows that the different matrices a anticommute with similar conclusion is also true for the system of p
The matrices
an
and
Pn'f
however, commute with each other.
288

that
Dirac proposed chosen as follows:
the
matrices
o^
[see
Eq.
(17.3)]
be
an
p,o n==
:(0;^
l\
(===!. 2.3),
r
(17.17)
o"\
(17.18)
In conventional notation these
4x4
matrices have the form

/O
i\
/
ai
/O I/O ""
1
1\
1
0\
'
/O
-'"~l
v,
0\
'
~t
V.oooV
oo
o/
g)
v/
Multiplying the above matrices by one another, that they satisfy the relations (17.7).
B.
is
it
easy to show
THE DIRAC EQUATION. CHARGE DENSITY AND CURRENT DENSITY
Let us substitute the corresponding operators into the linearized relativistic relation (17.3) between the energy and momentum. We obtain the Dirac equation for a free particle 4
(EH)t = 0,
where the operators E andp are, as usual, equal
E
to
(17.20)
= ift~, p=
i/>,
Because
of the four
components of the wave function
each state can have either
positive or negative energies (see below) and two directions of the spin (see Chapter 18) In the classical case the relation (17 1) between the energy and momentum can be
represented in a form similar to (17 20)
This equation
is easily verified if
it
is
remembered that
for a free particle
Consequently the matrix
> v c must play the role of the velocity, while p 3 ->\/l~ a scalar that characterizes the Lorentz contraction
fi
289
and the Hamiltonian
is
given by
c
2
.
H = f(a.p) + Po m
(17,21)
<|
must have Since a and p 3 are 4x4 matrices,- the wave function four components, which we combine to form a single- column matrix
(7-22)
The complex conjugate of

Hermitian adjoint, that
is,
this
function
is
understood to be the
17 - 23 >
the
row matrix
(
**=(*;*?*:)
The Dirac wave equation system of four equations
(E
(E (E
-4-
(17.20) is therefore equivalent to a
m c <h m^) <h

2
c (p x
ip y ) ^4
-j-
c (p v
ipy )
4> 3
m$*) ^ 3
2
c(p x
ipy ) <h
(E -f mac
<|>
c (p, -j- *p v ) ti
= = + cp^ cp^, = + cp A =
q>*4>3
4
0,
0,
(17.24)
0,
0.
In the case of motion of an electron in the electromagnetic field specified by the given vector and scalar potentials A and a>, we can still use Eqs. (17.20) and (17. 24), but the energy and momentum operators have to be generalized in accordance with the general
laws of quantum mechanics:
/
<&,
F = iH
P=
iflV
Â
(17.25)
The complex conjugate of the wave equation may also be represented in the form of a single matrix equation
f-(F
c(o-P)
p3
mc
a
0,
(17.26)
where the action of the operators
iti -*.
and
iflV
on the
wave
function which is on their left should be taken to be the in Eq. (17.20) but with opposite sign, that is,
same as
_ îh V = iHVp;
For a free particle, Eqs.
'*
yih dT
ih
<!>+.
(17. 27)
(17.20) and (17.26)

icH (a
'
now become
'
W +
*
V) *
P ;jm
^* =
(17.28)
'*
^+ W"
icn
'
+ vVP^ =
/
0-
(17.29)
290

<>+
Multiplying (17.28) on the left by and adding them, we obtain
and (17.29) on the right by
<|>
V- fa* 7<J *> +
= 0,
(17.30)
which may be interpreted as the equation of continuity for the probability density p and the current density j\
=0,
where
s
(17.31)
ef<|>,
j==aâ'l.
(17.32)
If we write the last equation in terms of components of the functions, rather than in terms of matrices, we obtain
wave
(17 ' 33)
that is,
PO
)
is
matrix consisting of a single element
(it is
just a
number
In
exactly the
same way
O
it
is
readily shown
that
J J J J i o o oy
(17.33a)
note that, contrary to the Klein-Gordon theory, the density a positively defined quantity. This does not mean, however, that in the Dirac theory p,, can be considered the particle density. Just as in the Klein-Gordon theory, there will be particles with a sign opposite to that of the electrons (positrons). From Eq. (17.32) it can again be concluded that ca should be regarded as the velocity operator.
p
We
is
Similar relations will also hold in the rase
when
a field is present
In second quantisation, the definition of PQ as a positive quantity means only that Fermi statistics should be applied to the particles (for example, in the case of the Dirac equation), if p Q may take cither positive or negative values (for example, in the case of the Klein-Gordon equation), then Bose statistics should be applied to the particles

C.
291
TRANSFORMATION PROPERTIES OF THE WAVE FUNCTION UNDER LORENTZ TRANSFORMATIONS AND SPATIAL ROTATIONS
to
general principles of the special theory of must be independent of the choice of the laws relativity, physical Lorentz frame of reference. Therefore, the Maxwell equations, the Klein-Gordon equation, and the Dirac equation must all be invariant under the Lorentz transformations. Let us investigate the transformation properties of the Dirac wave function. The Lorentz transformations can be written as
According
the
c/'~ctfcoshY
Arsinlv,-,
x?
= Xcosh'(
-BJ
^ sinh
-1^1
^
Y, if
y,
z'~z,
(17.34)
where
cosh
T
=--J_
V\
sinhY
'
$=--.
f
3-'
This transformation must be satisfied by all four-dimensional vectors, including, in particular, the charge and current densities
cp'
= cpcosh
sinhY,
/;
/,.
cosh
cpsinhy,
j'
vs
j VtZ
(17.35)
The
gives
definition of these quantities, according to the Dirac theory,
a,
<|/*-
sinh
r)
'|>
a ,<|/
ty+ (a,
cosh
sinh T )
ty
= = yw"^,
<j/e~T*n|>,
(17.36)
Here we have used the

sinh
fact that
{
e~ ^*
y it
= cosh
Y?I
sinh YI = cosh
since a-f" = 1, v] n = Y a, Y order to satisfy the above relations,
where // we must set
is
an integer. In
<!/=(cosh V
3,sinh
=^ J^ 2j
Tai
^,
f
(17.37)
Then, since
a^
it
-JU, 2
=^=(?
_J.
2
ff
__L 71
2
c/j,
JL
7,2 <?
=e
72
7 OQ\ î/.ooj
/I
is
(17.37)
easy to show that the relations (17.36) are correct. From it can be seen that the wave functions do not transform as
292
(whole angle 7) or as tensors (double angles 7), but as spinors, the transformation of which is characterized by the angles 7/2. jpinorg_frjfi alfigugalled tensors of rank 1/2. In a similar a nner",""" "If"IHay"" Be' 'aBowST'tflftTthe transformation law of a spinor under an ordinary spatial rotation (for example, around the z axis by the angle <p ) is as follows:
vectors
'
tb
./ -
I3s
"
~~ * 8
~2~
r\.
4.
T*
rz=
'li
/1
7 QQ\
^JLi.u*//
The above relations follow from the transformations

rent vector:
/'*
for the cur-
]' y
= =
/*
iy
cos
<p
/y
sin
<p,
cos 9
j\ sin
<p,
(17.40)
these transformations are represented in the Dirac theory as

l
f
By
that
it
l
'f
i[/
a,r|/
a cos f
+
<!)
+a
sm <p) (j,
(17.41)
a3
t|)
and so on.
substituting the values for
r
<j>
from
(17.39),
and using the fact
~-
+
,
ra 3 sin
=
a,
cos
ro.,
sin
at
=
we can
-*,!
a,,
obtain the relations (17.40).
Chapter
The Dirac Theory of the Motion of an Electron

in
Central Field of Force
A.
ORBITAL, SPIN AND TOTAL ANGULAR MOMENTA.
CONSERVATION LAWS
We shall determine the angular momentum of an electron from the conservation laws characterizing the motion of an electron in a central field of force:
V
(for
e<b(r)
(18.1)
field of a nucleus
example, an electron moving in the Coulomb

Ze*\
t
V _ --^
it
was shown
in the nonrelativistic
is
Schrodinger theory
that the orbital angular
momentum
conserved in a central field

(18.2)
L=rxp.
In the Dirac theory, however, which takes into account the electron spin, the component of the orbital angular momentum does not
commute with
the Hamiltonian
V(r)
(18.3)
and therefore
it is
not a constant of the motion. Indeed
a*,)
* 0.
(18.4)
In
order
to generalize the
mentum
to particles having spin,
law of conservation of angular we shall use the relation
mo-
-lPy).
It
(18.5)
follows
from
this that the operator

~ftcr
=L+S
(18.6)
294
commutes with the Hamiltonian operator H and thus is a constant of the motion. This result may be interpreted in the following manner. The electron has an intrinsic angular momentum (spin). We have just found that only the total angular momentum is conserved (the orbital angular momentum plus the spin). The orbital angular momentum in the s state is equal to zero, and therefore we have here the law of conservation of spin angular momentum. For the square of the spin we obtain
J
S2
=4^(c>-Kl
J)
s( S
-H)ft'=Aft;
(18.7)
that is, units of
the
ft)
electron spin takes half- integral values
5=
1/2
(in
B. PROPERTIES OF THE TOTAL ANGULAR MOMENTUM OPERATORS. QUANTIZATION OF THE TOTAL ANGULAR
MOMENTUM. VECTOR MODELS
total angular
that the operators for the components of the in the Dirac theory satisfy the same commutation relations as the operators for the components of the orbital angular momentum in the nonrelativistic quantum theory (see Chapter 11). It can be seen that the operators L and S commute with each other, because they act on different variables.
We
shall
now show
momentum
Therefore,
and
J
(18.9)
The last two relations are obtained from the first by cyclic permutation of the coordinates
>
z ->x.
mentum
Since only the total quantity (18 6) is conserved, the separation of the angular mointo spin and orbital parts is not rigorous in the general case This separation is found to be possible only in certain special cases (see Chapter 20)
THE DIRAC THEORY OF THE MOTION OF AN ELECTRON
295
is
The operator
to be
of the square of the total angular

a
momentum
seen
P = (L + S)
=L +
3
S'
-l-2(L t S;e
+ L + LA).
)
,S v
(18.10)
which commutes not only with the Hamiltonian but also with any of its components, for example, with the z component:
J*J,
J J
it
(18.11)
By analogy to the orbital and spin angular moments, we conclude that the square of the total angular momentum and one of its components (for example, J z ) may have simultaneous eigenvalues.
The quantization rules of the total angular momentum may be found from the quantization rules of its orbital component (for a spinless particle):
I*
^/r7 (/+!),
(/
0,1,..
),
= Hm,
S,=
(m=
0)
I,
....
-|-/)
(18.12)
and
its
spin component (for example, for t-S*
= aMs+l),
(s-=V*).
ft/Hi,
K=
-U/i)
(18.13)
This problem can be solved exactly in general form in the Dirac theory. For the sake of simplicity we shall, however, solve it in the Pauli approximation, that is taking into account the spin of the particle and assuming that the particle itself is nonrelativistic. If the particle moves in a central field of force, the components of the wave function
,
means
which obeys the Pauli equation (see Chapter 16), can be related by of the law of conservation of angular momentum
where
of the
r x p is the orbital angular are the two-component Pauli matrices.
L=
momentum operator and We shall look for a solution

or'
system of equations
(18.15) in the
form 2
V,==C
2
V r~
l ,
il'^C.rr,
(18.16)
tum
In this particular choice of a solution, only the square of the orbital angular is conserved, not its projection on the z axis.
momen-
296
where
since
Y? are the spherical harmonics (see Chapter
11).
Then,
Eqs. (18.15), (18.12) and (18.13) give
or
where
q
i (j
4-
(/
4.
_1
(18. 18a)
Using the relations (11.87) and (11.88), we have

(18.19)
<L A
+ *L
V)
YT = - ft
]/</
--
C-H m) Yf--
(18.20)
Then, taking into account

(</
(18.16),
we may write Eq.

1
(18.18) as
o,
- +
,
c,
+ V(7+ - //oF+^j c, =
From
to zero, we find types of solution
the requirement that the determinant of the system be equal two values of q corresponding to the two possible
^^
1). /
(18.22)
= /-%. ^
3
2 ,
coefficients C, and C which determine the relationship between the spherical harmonics in the sum of the orbital and spin angular momenta, are called the Clebsch-Gordan coefficients.
The
We note that this relationship between the spherical harmonics is obtained only in the case of spin-orbit interaction, which we have taken into account with the ai*d of the relations (18 IS) If there is no interaction, the two solutions will be completely independent
THE D1RACT THEORY OF THE MOTION OF AN ELECTRON
297
Using also the normalization condition C;

the first type of solution
+ Q= 1> we may write

1,
.
when/=/
-I,
= 0,
i
in the fol-
lowing form:
F
2/1-1
_ _/
/_i_l/.^
(18.24)
For /==/
/4
'=
If
2,
...
(the
second type of solution), the wave
function has the
form
1
/T
'yw'+T
ym
'
1
I
IJKJJT K
îV'
"",
(18.25)
27+7
where Vl m are the so-called spherical spinors. The orthonormality

;)
condition for the spherical spinors can be written as

Y\ m
= Ww*mm',
in
(18.26)
where
/2
corresponds to the case
which the spin and orbital
/ Vato the case when they angular momenta are parallel, and / are antiparallel. This condition follows immediately from the fact that the spherical spinor Y\ ^ is a single-row matrix and from the orthonormality condition for spherical harmonics. The spherical spinors (18.24) and (18. 25) are spinor generalizations of the ordinary spherical harmonics (see Chapter 11) and represent the angular part
J)
of the solution for all problems involving the motion of a particle of half- integral spin in a central field of force. Substituting these solutions for the function W into (18.15), we find that the component J z of the total angular momentum takes the
value J z
= hm n where the quantum number m is equal to = m ~. can b6 seen from (18.24) For the first type of solution = + ^)> = = to m = that ranges from (m = ^ ~
y
/w
(/'
it
it
/)
-|-
-|-
-5-
since the coefficient at the function Y? \ which does exist for the last value of m, vanishes. In exactly the same way the number m
in
the
second type of solution
(/
from ^) ranges
/+
Thus our results can be summarized as follows. of the total angular momentum has the eigenvalues
The square
T'
'
(18.26a)
'
298
that is,
it

is
quantized similarly to the orbital angular momentum, the quantum number /, which is called the total except 4 momentum quantum number, takes half-integral values. angular The eigenvalues of the component of the angular momentum along any axis are also characterized by half- integral quantum numbers
that
(18.27)
the relations (18.6) and (18.7) and the quantization rules and (18.26) (18.27), it is easy to obtain quantization rules for the scalar products L S and J S, which are important in spectroscopy
},
From
(18.28)
and by analogy
S
(18.29)
We shall consider here the vector model of the addition of angular moments. In spite of the lack of mathematical rigor of this model, it enables us to resolve a number of complicated questions and often gives accurate results. As we know, the orbital angular momentum does not have a specific direction in space in quantum mechanics. The absolute value (square) of the angular momentum and one of its components, for example, along the z axis, have, however, simultaneous definite values. These facts can be
represented geometrically by an angular momentum vector describes a cone about the 2 axis. The projection of the angular momentum on the z axis will then have a welldefined value, whereas the projections on the x and y axes remain indeterminate. These arguments apply with equal validity to the spin angular momentum since it has the same commutation properties as the orbital angular momentum. The spin and orbital angular momentum vectors are oriented in space in such a way that their sum forms a vector / that is constant in magnitude. Thus, the vectors L and S do not have arbitrary directions; they precess about J like two coupled gyroscopes. The dimensionless quantities y*, /*, and s* are drawn in Fig. 18.1. Each of the vectors /* and s* is defined on the surface of a cone. They "precess" around J* like a
that
FIR. 18.1. Addition of the spin and orbital angular
system. We note that, according to Eq, (18.26), addition of the vector /* (/ 0, 1, 2, . . .) and the vector s * (S^H:'/*) leads to e total angular-momentum vector J* with half-integral values of the total angular mocoupled
the
&
l
moments.
mentum quantum number j =
f*.
The number / is also called the internal quantum number. This number was introduced by spectroscopists before the discovery of spin on a purely empirical basis. It expressed certain internal properties ot particles that were still unclear at that stage.
The lack From
the
vector j is equal to
of mathematical rigor lies, for example, in the fact that the square of the 2 }(j 1) rather than to y
\
standpoint of the classical theory, this coupling of the orbital motion by the magnetic field.
coupling can be interpreted as a

From
can find
the vector model the quantization
299
we can
rule
of the
quickly find a number of quantities. For example, we angle between the vectors /* and s*. From the
oblique triangle
we
obtain
cos (t<s*)
= -^J
f7*
/"
s* 2 },
(18.30)
that is,
l)s(sfl)
C.
MOTION
we wish
in
IN
SPIN EFFECTS.
If
A CENTRAL FIELD OF FORCE INCLUDING THEORY OF THE ROTATOR
field
the
to investigate the motion of a particle in a central nonrelativistic approximation with the inclusion of
(
spin effects, we must use spherical spinors Y m characterizing the states in which the total angular momentum (orbital plus spin) is conserved, instead of the spherical harmonics YT describing the states where only the orbital angular momentum is conserved. Since the spherical spinors (in the nonrelativistic approximation) are
J}
i
composed of spherical harmonics having the same quantum number /, we obtain the same radial equation as fora nonrelativistic spinless
particle, that is,
=a! o.
(18.32)
For the wave functions of an electron

n W\*
in a central field
we
obtain
= RY\ m
}]
(18.33)
where the spherical spinor

(18.24) and (18.25).
Y\ m
is
defined by the expressions
In particular, for the rotator we may set r const and the radial part of the wave function /?=!. It is seen that the spin effects in the nonrelativistic approximation do not give any additional terms for the rotator energy, which will be given by the same expression (12.7) as for a spinless particle, that is,
= a=
therefore bers /
,
The wave function will be given by the spherical spinor Y} m we must find the selection rules for the quantum numand y These selection rules hold not only for the nij rotator, but also for any problem of motion of a particle in a
s]
\
central field of force (for example, the electron in the hydrogen atom).
300
In
place of Eq. (12.19), from which the selection rules for spinless particles were established, we now have
(q)'
where
g
=
jl
]
1
(V {,'V)
qY\' m
dQ
(18.35)
may have
q
three values
= = cos
z
ft,
<7
= #Jr*# =
sinflc!
'*
(18.36)
a~l).
(for simplicity let us take the radius of the rotator equal to unity: If in place of the spherical spinors we substitute their
values from
(18.24)
and (18.25), the matrix element (18.35) be-
comes
~
)*
'
D</;>
f (YF'-
qY?
d'J
+ C"'
<f
(Y? ')* </>?
dS2
(18.37)
From this it is seen that the two integrals in (18.37) agree exactly with the integrals in Eqs. (12.19)-(12.22). The selection rules for the quantum numbers / and m will therefore be the same as for a spinless rotator, that is,
We
nij
shall
/
now
Since
find the selection rules for the
quantum numbers
and
is
related to
by the relation
=w
for
same
both types of solutions, the selection rules of as for m, that is,
will be the
the selection rules for / let us consider first the case in which the transitions occur between states characterized -> ^ /' ~> y / or /' by the same type of solutions (/' '/* li -|~ /a / It follows from (18.24) and (18.25) that" the coefficients /a). / D</v> and QM are always positive and therefore such transitions are always allowed. In this case the possible change of; must be the same as the change of the orbital quantum number /, that is, =tl A/ A/ If the transitions occur between states characterized by different /' types of solution (/ Vi->y /~V or / /'- /i^/ / /*). then by taking into account A/=itl, we obtain three possible values for A/ 0, +2, -2. Here, however, we must consider the fact that the coefficients D ^) and CW have different signs. For -:2 the two terms cancel each other, so that this transition A/
-
To determine
= =
= +
= +
3Ol
the difference between the two terms is is forbidden. For A/ not zero, but owing to the fact that the two terms occur with different signs, the intensity of the radiation will be weaker than in the case of transitions between states characterized by ^t 1. This can be shown with the same type of solutions, when A/ the help of a specific example. Let us suppose that the initial / /' state is and the final state is / / V* %. We shall m w 0. Then, using (12.22), we reconsider the case Aw duce the appropriate matrix element (18.37) to the form
= +
/,
m)],
where
/'
/'-{-
Y*
and /=/.
Ya-
Substituting the expressions
DO' - <'
i-
vs. y
= - 1/
/
8)
= --^-^J
,.,_.
and
(18.24) and (18.25), and the expressions for A and B from (12.22a), we find that the coefficient of &/./ + vanishes; that is, the transition A/ 2 is forbidden. At the same time, the coefficient of 8/'. /_i does not vanish, that is, the transition A/ is allowed, but the intensity of the lines is weak in comparison with A/ In a similar manner it is easy to show that the nh 1. transition is forbidden not only for q~z, but also for 2 A/
from
(Aw=;t:l). In accordance with the above discussion, the selection rules for the quantum numbers which characterize the state of a particle in a central field of force, when the spin is taken into account, have the form
(/
= x.iy
(18.39)
1
(normal intensity), (weaker intensity).
D.
PARITY OF A STATE
law of conservation of for Dirac particles we shall now define more clearly the meaning of the quantum numbers / and / in the Dirac theory of a particle in a central field. In nonrelativistic quantum
In connection with the formulation of the
angular
momentum
302
theory, the orbital quantum number / is associated with the square of the angular momentum L" #*/(/ 1), which is a constant of the motion; therefore in both Schrbdinger's and Pauli's theory the quantum number / represents a quantity that is constant in time. In the Dirac theory the orbital angular momentum does not commute with the Hamiltonian,and therefore it is not a constant of the motion. Consequently the quantum number / has only an approximate meaning when used in connection with the law of
conservation of angular momentum. It turns out, however, that /characterizes an additional property of a particle in quantum theory, namely, the parity of a state. By the parity of a state we mean the behavior of the wave function with respect to space inversion:
*
= -.*',
y==-y',
= -~z'.
7
(18.40)
The parity operator
is
defined as follows
I*(r)
= K-r);
(18.41)
that is,
it
values
twice:
reverses the signs of the space coordinates. The eigenof this operator may be found by applying this operator
a
I
i[
X 8 <|>.
(18.42)
This
double application of the parity operator leaves the coordinates unaltered. From (18.41) and (18.42) it follows that
X
= rtl;
X=
(18.43)
that is, either the wave functions remain unchanged with respect to space inversion (even functions, 1), or they reverse their sign (odd functions, X 1). We shall now find the quantities which determine the parity of a wave function in a central field. In the spherical coordinate system r D, <p, space inversion affects only the angular part
(18.44)
as can easily be seen from the fact that the sign of the coordinates
A*
= rsin frcos
<p,
r/
= rsinftsin
cp,
= rcosO
(18.45)
This operator converts a right-handed system of coordinates into a left-handed system, and vice versa In the Dirac theory I0(r) - /0 3 0( r).
303
changes. The radial part of the wave function remains unchanged with respect to space inversion, whereas the angular part changes in accordance with the relationship
m
IK/ (a,
<p)
= l7(ie
ft,
+ = constP/m
<p)
-cos d)*' m <* + *>==
(18.46)
because
Thus the parity of a state in a spherically symmetric field is determined by the parity of the number / Furthermore, it can be seen that the Hamiltonian in a central field remains unchanged under space inversion; therefore, the inversion operator I and the Hamiltonian operator H commute with each other. It follows that the parity of a state is a constant of the motion, since
Î=^(HI
The law
the
IH)
0.
(18.47)
of conservation of parity has no classical analog, unlike conservation laws (energy, momentum, angular momentum). Consequently, in nonrelativistic wave mechanics the number / characterizes two constants of the motion: the square of the angular momentum and the parity of a state. In the Dirac theory, the number / does not have the significance of the square of the angular momentum, but the relationship between this number and the parity of a state is preserved. We shall see later that parity plays a particularly important role in the physics of elementary particles. All wave functions whether for one or more particles can be classified as functions of odd or even parity. Dipole transitions can occur between states with different parities. In the case of two or more particles, the parity depends on the total spin of the system and also on the type of statistics obeyed by the particles. These concepts will be analyzed in some more detail in the treatment of specific examples.
other
E.
SOLUTION OF THE DIRAC EQUATION FOR A FREE PARTICLE
stant
Let us consider the mot ion of a free particle of spin 1/2 and conmomentum such as, for example, an electron. Without loss of
generality
we may take
momentum;
the z axis to lie along the direction of that is, in Eq. (17.20) we set
P*
= P, =
0,
p,*0.
(18.48)
304
The Dirac equation then takes the form

(ill
-I-
tk
&
- PV
iltKt
=0.
form
(18.49)
We
shall look for a solution of this equation in the

-17;
be~
+ '**,
(18.50)
where L3
the
is
same value as
the normalization volume, and the wave number k has in the nonrelativistic Schrbdinger theory,
namely,
*
2
(/i
0,
1,
2,
..
),
and the 4 x 4 matrix

bl
satisfies the normalization condition
4=1
To determine
system
the quantity
e
(18.51)
b^
and the coefficients
we use
the
of equations (17.24), setting
We
then obtain
-=
e=
Hence we find two values for the quantity e: 1 (the energy of the electron is positive) and s 1 (the energy of the electron is negative); while we obtain four values for the matrix 6, the components of which satisfy the normalization condition (18.51). Two of the values of the matrix b refer to states with e l:
305
and two of them refer to states with
'
(18.54)
that
is, the states (18.53) differ from the states (18.54) in the of the energy. sign To determine the physical significance of the states 6 (/) with different f= 1, 2, 3, 4, let us find the projection of the spin on the direction of motion, that is, on the z axis. First of all, we
L z xp v note that, since when a particle is moving //p t along the z axis, the projection of spin on the z axis must be conserved. This follows directly from the fact that the matrix a, commutes with the Hamiltonian in (18.49). We can find the eigenvalues of this operator o 3 by applying (l} it to the spin functions b We have then
.
000
0-10
(1
0010
001
_ I__
?
I
= "
I '
'
&">;
<-
(18.55)
*,_
that is, for this solution the eigenvalue of the operator In exactly the same way it is easy to show that Vi
/2
a,
equals
Io ^>==:--l^,
3
-J.o 3
&<
>
=!&<>
and
lo^=-lft
(4)
.
Thus, the four possible states correspond to the four possible combinations of the sign of the energy and the spin direction. The solution b n} corresponds to positive energy (e= 1) and the projection of this spin along the positive z direction (s= I). In a similar way we have 1 for the solution //*'and we have e 1, 1, .$= 1 and 1 f or the solutions 6 (:t> andb (i), respectively, where 1, s s is double the projection of the spin in the direction of the mo-
e=
s~
mentum.
In the nonrelativistic limit
(v
**>
c),
the
wave functions
<|>
and^ 4
will
be
of the
order of
-^~~
times the wave functions
^ and^
(t|) 3
{/,)
for positive energy states
(s=
1).
For states with nega-
tive
energies [see (18.54)], on the contrary,
<h~
fy.
306
We have considered the special case of motion of a particle along the z axis. This does not restrict, however, the generality of the investigation of the general motion of a particle. Whenever the direction of momentum is characterized by spherical angles & and 9, it is always possible to choose a primed coordinate system in such a manner that the z axis is directed along the momentum. Then, by carrying out two rotations, one rotation through an angle d around the y' axis directed perpendicularly to the 22' plane, the other, second rotation through an angle <f around the z axis, we may transform
1
from the solution
in the primed coordinate system (momentum along the z' axis) to the general case (direction of momentum characterized by the spherical angles &, <p X Using the fact that under a rotation of the coordinate system the wave function changes in accordance with (17.39), we may write the solution for this general case
-I-
-a)
(18.56)
which
is a
generalization of Eqs. (18.50), (18.53) and (18.54).
ill and p now that the energy and momentum operators E Jgrgblem 18. 1. respectively ,Transform like a four- vector under the Lorentz transformations:
= ^
in V,
J2
c
=
= ~,
-7!^
t
P^=P;-
P.-
where
?
E'î/fcgry
and so on.
Hint. Use the Lorentz coordinate transformation [Eq. variables in the process of differentiation.
(17.34)]
and change
to
new
Problem
particle. Hint.
18.2.
"""ili
prove
the
relativistic invariance
of
the scalar equation for a free
First, let us prove the invariance of the operator relation

8
E'
c 2 p'
=E
cV
by using the results of the preceding problem.
- i *"-V
1
i
igblem
JAJ Show
that
in
,
by the angle
the case of spatial rotation of the coordinate system the wave function transforms according to the relation
Hint.
Use
the
method which leads
to the relation (17.39).
lft^ With the aid of the vector model of addition of angular moments, angles between J* and s* and between ;* and /* taking into account the geometric vector addition in quantum mechanics; that is, find
fin<r~the
J^
W em
cos (;*s*);
cos (;*/*)
These problems also
refer to the material in Chapter 17
THE D1RAC THEORY OF THE MOTION OF AN ELECTRON

Hint.
307
Use a method similar
to the one that led to Eqs. (18.30)
and (18.31X
*~*
/a a p8, and fy*is =C<|>*, where C 18.5} Show that the wave function njugate (but not the Hermiaan adjoint, that is, nott^) of the Dirac wave function for an electron with negative energy satisfies the Dirac equation with positive energy and opposite (positive) sign of the charge, that is, describes the motion of a
*^N roblem
positron (a charge conjugate transformationX Solution. The Dirac equation is
for the
a;
=o
complex conjugate of
a*
a.,,
lt
aj
=a
this equation,
pj
<f
we may write
(taking into account the fact that
p,):
We
note that the complex conjugate ^* differs from

k*i
<],+
(the
Hermitian conjugate): namely,
I
whereas
Let
then
us
substitute
that
<f
== /?'!'*
into
the
complex conjugate of the Dirac equation. We

if the
find
satisfies
the
Dirac equation
charge e
Is
replaced by -
e.
Since the state
fy
(r 9
t)
=e
jh
<|>
(r) is
treated as a state with positive energy (E
)t
and
(
the
1
state
|
fy* (r t t)
=e
<j>*
(r)
),
we must interpret
the
is treated as a state with negative energy sign of the energy in "fy differently than in ty\
Chapter 19
The Dirac Equation

In
in
Approximate Form
the Dirac theory,
fl
many problems which are solved by

Therefore we
we
(V
final results.
in the
may immediately
write the Dirac equa/

(
tion in an
It
approximate form, retaining quantities of the order of
a
.
will be
shown below
that the role of both the relativistic
and spin
terms
displayed in this approximation. Let us consider the motion of an electron with positive energy "">() in an electromagnetic field which does not depend on time. Ih this case we may replace the energy operator by its eigenvalue, separating out the rest mass energy m c':
is clearly
to
->w
The wave equation
(E
c'
+ E.
;
(19.1)
(17.20) then becomes

,
e-I.)
;=c (a'-P)
stands for
(2/v*
+ E - *)
=c K-P)
(19.2)
where
p
a'
2x2
Pauli matrices [see (16.18)], and P
A.
Equations (19,2) are simply a different form of the exact

fy,
Dirac equation. For the components of

(E
we
obtain
from
(19.2)
fid>)
i,
-C
(P.v
- iPJ ^ - cP& -
and so on. This equation can also be obtained from the first of Eqs. (17.24), if we substitute into it both (19.1) and (17.25).
limit the
As was mentioned in components
ty A
the preceding section, in the nonrelativistic and ^ 4 are "small" for positive energy
states, since they are of the order of

1
y
V
times the "large functions

l
and
<|>
a.
in eliminating the
The transition to the approximate Dirac equation consists from Eqs. (19.2) "small" components -^ and
'
and retaining terms of the order of

for the "large"
/
(
\*
in the
j
remaining equations
(19.2)
components
and
fy.
Thus from
we
obtain
THE DIRAC EQUATION

First of
î\
IN
APPROXIMATE FORM
309
all
we change from
the four- component
wave functions
to the
two-component functions
\jr*
by setting
<
19 * 4 >
where N is the normalization coefficient. This coefficient determined from the " renormalization" relationship
may
be
;)
(19.5)
Since the "small" wave functions

(19.5),
"!
and
ti
occur as squares
in
we may
set
that is, in (19.3) we neglect second-order terms and replace P by p. This change of operators is permissible in calculating the normalization coefficient since they differ from one another by a first-order term (inversely proportional to the velocity of light). When applied to second-order terms (squares of the "small" wave functions), this term of P gives only third-order terms, which we discard. Then, substituting Eqs. (19.4) and (19.6) into the left-hand side of (19.15) and using the equation
1
(or'-
a)(a'. b)
= (a
b)+ i'
(xft),
(19.7)
which holds both for the Pauli and Dirac matrices, we find
(19.8)
From
this
we
obtain
In order to prove this equality, we may, without loss of generality, assume that vector a is directed along the z axis (a x - a - 0, a z a), and that vector b is located in the zx y - 0) Then we obtain plane (b y
(a'
-
a)(a'
b)
\a
(b
<y')l
i[o'
(a x 6)]
31O
This
(19.4),
approximation gives us, in accordance with (19.3) and
8mc"
Substituting (19.9) into the first of Eqs. (19.2) and neglecting 3 we find of the order of ( v / c )
,
terms
Using the relation (19.7), we have
(0'.
P)(o'.P)
=P
J
-f t[0-(P *
1
P)]=
=P
-(^l(px>)-f(^xp)])=P
-P
where //is the magnetic
-T.'-^)-
field since the operator v acts only on the vector potential A, and not on <J>. In exactly the same way, with the aid of the relations
',
p)
(E
- eto)
(a'-
p)
= (E -
*<D) p*
~ iAe
(a'-
(^ p) ==
and
-,P*
^p^(-^) = (E^^D) P
(
2
-l-
f-
-p
-|-/^V^.
(19.13)
where
v ^ is the electric field intensity, we may reduce the Dirac equation (19.10) to the following approximate form:
(19.14)
The left-hand side of Eq. (19.14) describes the motion of a particle with a nonrelativistic velocity in a stationary electromagnetic field. The right-hand side of (19.14) contains an additional interaction energy that describes the relativistic and spin corrections.
THE DIRAC EQUATION
IN
APPROXIMATE FORM
of (19.14)
311
The
first
term on the right-hand side
<
19 ' 15)
takes into account the correction due to the relativistic velocity of the particle. A similar additional energy must also appear in the relativistic Klein-Gordon equation. The classical analog of this term will be obtained if the relativistic expression for the Hamiltonian is expanded in a series, retaining terms of the order of
The second term on the right-hand side

as
of (19.14)
may
be written
K mag
ji-//.
(19.16)
From
this
it
is
clear that the quantity
JLI
= ~?
n
<r'
may
be treated
as the Dirac magnetic moment of an electron, which appears exin the nonrelativistic approximation only through this transition. This interaction energy turns out to be of the order of From the intrinsic angular momentum [Eq. (18.6)] of the v/c. electron
plicitly
= -o',
(19.17)
we find the relationship between S and \i that is required by experiment and follows automatically from the Dirac theory
The next term of the expansion characterizes the so-called spinorbit interaction
*'
-(x
P)l,
(19.19)
which describes the interaction of a moving magnetic dipole with an electric field.
This interaction may also be interpreted from the classical point of view in the followway: a magnetic dipole moving with a velocity i> (the spatial component of a tensor
ing
312

an additional electric moment (space-time component of the same
quantity) acquires tensor quantity)
lel
=7
l=
px|i
,,!oc
'
(19.20)
which interacts with the electric

action is
cl
field of the nucleus.
This additional energy of inter-
=-
*. |lel
-..
-^
|a'.(xp)]
(19.21)
This classical expression for the interaction energy is twice as large as the corresponding quantum expression [see (19.19)]. Even before the advent of the Dirac theory, an attempt was made to explain the fine structure by the semiclassical introduction of spinorbit interaction. To obtain an agreement with experiment. Thomas and Frenkel suggested that we substitute the coefficient 1/2 into the classical expression for the interaction energy (19.21). This interaction, which follows automatically from the Dlrac
theory, is called the Thomas-Frenkel correction.
In particular, for the
Coulomb
field of a nucleus
(19.22)
The
interaction
between the moving magnetic dipole
and the
nucleus
according to (19.19) becomes
"-*= --*&-where
S ' L'
(19 - 23)
S=
n o'/2 is the spin,
and
L~rxp
is the orbital
angular
momentum.
We note that there is no spin-orbit interaction for an atom in the s state since the orbital angular momentum in this state vanishes. Finally, the last term of the interaction, which in the 3 case of the Coulomb field is equal to
<
19 - 24 >
is
called the
it
contact
interaction.
The
additional energy corre-
sponding to
=
2
(19.25)
Oxford University Press, 1938, Vol. 2, J. I. Frenkel, Wave Mechanics, New York see also (16.11), where the coefficient 1/2 appears in the magnetic moment produced by a
moving charge.
In the derivation of (19 24),
we have used
=
the fact that, according to (4 78),

(x) S (v) 8 (2)
V 1
2
- - 4778 -47r8(r)
(19 24a)
THE DIRAC EQUATION

is
2
I
IN
APPROXIMATE FORM
313
and it will differ from zero only for the 5 proportional to ^ (0) to state since, according (13.28a), only in this case |*F(0)|* ^ 0. For all other states (' ^ 0) this square of the wave function vanishes when r Q. In this sense the contact term may be regarded as the spin-orbit interaction for the s state. We can see now that the last two terms in the interaction energy (19.14) characterize the spin properties of an electron.
|
r~ Problem
v
19,
1J Show Pfi^""PS!rtlcle a
*#v)
.
that the
is
not
matrix ca is the velocity operator, and that In the case a constant of the motion, unlike the momentum operator
Explain this difference. Determine in what case the average velocity
=
will be related to the average
yd
and so on,
momentum
by the classical relation
v=c*^.
Hint.
(19.26)
From
the
H am il Ionian
(17.21)
it is
possible to obtain the velocity operator
It can also be shown that the velocity is not a constant of the . Conmotion, since o sequently, if for a given k we take a linear combination of positive and negative energy states [see Eqs. (18.53), (18.54)] there will exist interference terms that will fluctuate with time (~* 2<c/rt ). As a result, Ehrenfest's theorems will hold only on the average in the Dirac theory. The interference terms will disappear in the calculation of the average value of the momentum operator. Equation (19.26) holds only if the states with positive energies are retained (e= 1).
Chapter 20
The Fine Structure of the Spectra of

Hydrogen-like Atoms
A.
STATEMENT OF THE PROBLEM.
of the motion of an electron in a hydrogen-like atom (Kepler's problem) is rightfully considered as the touchstone of all forms of quantum theory. There are two main reasons for this. First, it has great physical significance, since the problem of motion in a Coulomb field can be solved exactly. Second, the results may be compared with experiment to a high degree of
The problem
for example, the emission spectra of atoms can be observed by optical and microwave spectres copy. The solution of the problem of motion of an electron in a Coulomb field of a nucleus (hydrogen atom) on the basis of the Schrodinger equation, obtained in Chapter 13, gives an expression for the energy
accuracy;
= _*Jf,
(20.1)
which is in good agreement with experimental data. This expression for the energy may be taken as the zero-order approximation. A more detailed study of atomic spectra shows, however, that the spectral lines have a fine structure which, of course, must
be associated with the detailed structure of the energy levels. The Schrodinger theory does not give an adequate description of the regularities frequently occurring in spectra, since it neglects at least two important facts: the relativistic dependence of mass upon velocity and the spin properties of the electron. Both these facts, as we already know, are accounted for by the Dirac theory, and therefore application of the Dirac equation to the Kepler problem gives results that accurately describe the multiplet structure of energy levels. As was pointed out, the Kepler problem can be solved exactly in the Dirac theory. The solution, however, requires many tedious calculations (much more complicated than in the Schr'odinger theory, because in this case we have not one but four equations). Moreover, in the course of these calculations one does not always
THE FINE STRUCTURE OF SPECTRA
315
perceive the physical meaning of the results, the analysis of which is of primary importance to us. We shall therefore use a more elementary method, based on the approximate equations of the preceding section. This method not only enables us to obtain formulas characterizing the fine structure up to terms of the
\2
(V
electron.
'
but
of
also to interpret the

relativistic
individual
terms as
of the
manifestations
or
spin properties
B.
RELATIVISTIC AND SPIN EFFECTS
As follows from Chapter 18 [see (18.24) and (18.25)], the wave function of a particle obeying the Dirac equation, taking into account the spin properties, is
W = RYp
Here Rnl
is the
(20.2)
radial part of the wave function and Y\^ is a spher/-|- /*the spin is parallel to the orbital angular y momentum and for / / V* it is antiparallel. Although terms of the order of (v 'cY are not formally accounted for in Eq. (20.2), the relationship between the spherical harmonies in the spherical spinor that determines the zero-order approximation of the wave function is established by the spin-orbit interaction, which is of the order of (vjcY* The spherical spinor can therefore be used only when the atom is not subject to external perturbing forces of magnitude greater than those involved in the spin-orbit interaction. If that is not the case, the spin-orbit coupling will be disrupted and a new set of premises must be set forth in order to establish a relationship between the spherical functions. Spherical spinors, just as spherical harmonics, satisfy the
ical spinor: for
equation
vi.^ja
= -'(' + )^
(20.3)
therefore, taking into account (11.17), the radial function in (20.2) satisfies the same equation that was derived in the nonrelativistic
Schrbdinger theory:
There are several cases
enables us
to find a relation
approximation
quantum mechanics in which a small interaction energy between the coefficients of the functions in the zero-order We have already encountered a similar situation in the treatment of the
in
Stark effect (see Chapter 14).
316
The wave function (20.2) completely determines the selection rules for all quantum numbers. The selection rules for the quantum numbers /, /, and m; are given by formula (18.39), while the selection rules for the principal quantum number n will evidently be the same as in Schrbdinger's theory [see (13, 48c)], since the radial function remains unchanged. Considering all this, we obtain the following selection rules for a theory of the hydrogen-like atom which takes into account spin effects:
A/
l,
A/
0,
itl,
Am =0,
y
(AM
is
an integer). (20.4)
for the expression for the energy, we cannot restrict ourselves problem to its nonrelativistic value (20.1), since the latter does not determine the fine structure of the energy levels. Knowing the zero-order approximation of the wave function (20,2), and also the additional perturbation energy describing the relativistic [see (19.15)] and spin [see (19.23) and (19.24)] effects, we may find the energy levels characterizing the multiplet structure of the spectrum. According to formula (19.15), the relativistic correction to the energy levels is
As
in
this
= __ J
Since in the present case
ww<Px.
(tfwj'-Jj--.
(20.5)
(20.6)
we see
angles
tt,
that this additional energy will be independent of the solid that is, integrating over the solid angle we get <p;
(20.7)
Then the additional energy characterizing the

is
relativistic effects
\prel aC
9,
__ IW
*
f//7o
r*
3\
(20.8)
where
~ ='He
gzl
Vn?
is
the fine structure constant.

In the derivation of (20.8)
317
we have used
(13.29a)
Equation (20.8) agrees exactly with the formula for the relativistic energy, which was calculated in identical approximation by means of the relativistic Klein-Gordon equation [see (15.31)].
In a similar manner, with the aid of (19.23), tional energy due to the spin-orbit interaction
we
find the addi-
-=
^
L)
(S
L)
(/-').
(20.9)
Using expression (13.29a) for (r*)
and expression (18.28) for(S
=
o
2
for /-O,
for
o,
we
obtain the following value

s
for the energy (20.9):
^ --^-^n^w^r
In these equations
20 - 10 >
At first glance it may seem that the spin-orbit interaction, which is inversely proportional to the third power of the distance, cannot give a stable state. This, however, is not so. At small distances the spin-orbit interaction behaves just like the relativistic interaction, that is, it is inversely proportional to the square of the distance. Indirect proof of 8 rel this is the fact that Aft differs from Ah' only by a numerical factor of the order of
unity
318
and the quantity
Finally, the energy cor responding to the contact interaction, according to (19.24), is given by
where
+
|V(0)p=/?i(o)y}ft
yjft.
(20.13)
Furthermore, considering the expression for
[see
= V2. we /
(13.28a)], find
and using the fact that |K^| 2
when/
and
that is,
Acont
= /?ft^!8
/0
.3
(20.15)
From this we obtain the following expression for the additional energy which accounts for the relativistic effects, the spin-orbit and contact interactions:
A
=A
rel
{-A
-*-f-
Acont
<i_n*/p)
__
V.)(7+l)
-j /0 1
J'
Incidentally, Eq. (20.10) for the contact interaction may be obtained when the expression for the spin-orbit interaction (Eq. 20.10) is allowed to go to the limit as / ~> 0, if we discard the factor 6/0 in 20.10. Therefore, many authors use this procedure and neglect
the contact interaction in deriving the fine-structure formula. However, the agreement between the two formulas is accidental since for the a states the numerator of Eq. (20.10) is always zero, while the denominator vanishes only in the nonrelativistic approximation. In
a number of other problems, for example, an atom containing several electrons, the energy associated with contact interaction is no longer a limit of the expression for the spin-orbit
interaction.

Substituting here the value of q
319
from
(20.11),
we have 4
(20.16)
Therefore, summing both results [(20.1) and (20.16)], we obtain 5 the fine structure formula for the spectrum of a hydrogen-like atom
(20.17)
From this it is seen that the splitting of the levels is proportional to the square of the fine structure constant.
C.
THE FINE STRUCTURE
IN
THE DIRAC THEORY
When we
take the fine structure into account the position of the
energy levels in the hydrogen atom is found to depend also on the total angular momentum quantum number /. Therefore the terms will be denoted in the following manner:
<
20 ' 18 >
this formula it is seen that the fine structure, according Dirac theory, depends only on the principal quantum number n and the total angular momentum quantum number y. In contrast to the Klein-Gordon theory, it is independent of the orbital angular momentum quantum number / (up to terms of the order of a *).
to the
This problem can be solved exactly in the Dirac theory. We then obtain a closed formula for the energy levels, in this formula, the first expansion term (which is independ2 ent of a ) gives the nonrelativistic formula (20.1). The second term, which is proportional to a gives the additional energy (20.16). The third expansion term, which is proportional m this approximation, can be neglected, since it is smaller than the so-called to a vacuum corrections, which are proportional to a (see Chapter 22).
,
From
An exact solution of the Dirac equation gives the following generalization of Eq (15 30), which takes into account the relativistic effects in the case where spin is also
present:
fl
r
1
1 .
zv ~-~
(
>* -m
1
_,-y^VcMy$)
-zV)
(20 17a)
Equation (20 17) may be obtained from (20 17a), if the latter is expanded in a series and we restrict ourselves to the first two terms Since the minimum value of j is equal to 1/2, we find that stable motion in the Coulomb field of a point nucleus, according to the Dirac theory, will extend to Z cr - 137, whereas ui the Klein-Gordon theory it was limited by Z cr - 137 [see (15 33)1 Such an increase of
Z cr is, as we have already mentioned, due to the slight compensation of the relativistic effects by the spin effects
_
32O
gram
RELATJVISTIC QUANTUM MECHANICS
in Fig. 20.1 shows that all terms are doubly to since each value of / there correspond two values of /; split, for example, instead of a single term 2/> (/=-!) we now have two
The diagram given
terms
E/h--0
The exceptions and 2/j. terms (/ are the 0), for which j Thus can have only one value (j ';). the relativistic and spin effects somewhat reduce but do not split the s terms
I
2pi
_,.
,s
(see Fig. 20.1).

'
The degree of degeneracy also changes owing to splitting of the energy
2pi/2
2s1/?t
2pr/2
js
Energy hwoi dia FIR. 20. i. of the hydrogen atom
We know that the principal levels. quantum number may take the following values: ;j~l, 2, 3, 4,.... The orbital angular momentum quantum number / varies from (s state) to /=//!. The total angular momentum quantum
/
number / takes
and
the values/
/ r*
/2 (/ -/.
0)
ji/^l
.
netic
quantum number m
- f
j,
-}
/,
U) and, finally, the magthat is, for a given / there
half- integral values of m r The degree of degeneracy, which is characteristic for any central field of force and is related to the equivalence of the various directions in space, is therefore for particles with a spin of 1/2 (we remember that equal to 2/-|for spinless particles it was equal to 21 \- 1). in contrast to the relativistic spinless theory, the degeneracy with respect to / is still present when we take into account terms of the order of a 2 and even the following expansion terms proportional to a 4 When the finite size of the nucleus is taken into account, the degeneracy with respect to / is removed. We note incidentally that even greater splitting with respect to / is due to vacuum fluctuations (see Chapter 22). The magnitude of the splitting of spectral lines can be determined from the selection rules (20.4). For the Lyman series we then have two lines (instead of a single one):
are
2/+1
(weak
line, since
,)
A/=)
,).
(20.19)
U)U=(l$i
(/!/?<
The Balmer series
lines are split as follows:
w>=i2si
(i><*>
)-(/i/i,, J
),
(2*i
J
/J
(np*
,),
=(2^
=(2/>.
)-.(ws l3 ),
(nsi
co>=(2/?i; J )~
io<*'
J,
(20.20)
,)
(/;</, .),
321
and, finally, the transition 2pi , ->nd*>, is forbidden, because in this If the degeneracy with respect to / is not removed, case Ay 2. l4) the lines w (1) and w coincide, since the initial and final levels have the same values of the principal quantum number /* and the total angular momentum quantum number j. In a similar manner we may determine the splitting of all other lines. The lowest split Let us consider in greater 2. energy level corresponds to n detail the splitting of this level in the case of the hydrogen atom which is the one most carefully investigated experil), (Z 2 level would be split into three submentally. In general, the n levels, and, according to our theory, two of these sublevels would
combine:
(20 . 21)
The transition frequency between these

Dirac theory,
Au)
levels
is,
according to the
(2/)i /a )
(2p v , )
=R
ft
(20.22)
about 1.095- 10 4 Mc. If only the relativistic effects are taken into account (Klein-Gordon equation) the corresponding splitting is [see (15.32)]
which
is
A W K-C
(2s)
(2/>)
={^
(20.23)
that is, the frequency is almost three times greater than the one found from the Dirac theory. Consequently, the spin properties of particles somewhat reduce the influence of the relativistic
effects.
The conclusions
of Dirac' s theory have been accurately con-
firmed by experiment.
It is interesting to note that the fine structure of the spectrum of the hydrogen atom was first theoretically calculated by Sommerfeld who applied a relativistic Hamiltonian to the steady states of the Bohr classical theory. Sommerfeld obtained [see (2.61)]
1 Me - 10 that is, the angular frequency sec with the frequency v, expressed in Me, by the relation
,
(j),
expressed
in
sec
is
connected
322
the following expression for the relativistic theory (20.22) without taking into account the spin effects:
Aa)
Somm
-=(2 S)-(2p)
=%'.
(20.24)
the Sommerfeld result with the conclusion of Dirac's theory was, however, only accidental. Sommerfeld's theory did not take into account the spin effects, and therefore it was unable 2 level into three sublevels, the to predict the splitting of the rc presence of which was later confirmed experimentally.
Agreement of
D.
EXPERIMENTAL VERIFICATION OF THE FINE STRUCTURE THEORY
The major accomplishment of Dirac's theory was its treatment of the fine structure. The theory was in good agreement with the experimental facts and was able to explain this structure as a manifestation of the relativistic and spin effects caused by the motion of the electrons within the atom. However, further and more detailed studies showed divergencies between the theory and fact. Thus, special attention was given to the 2si/ 2 and 2pi/ 2 levels
which, according to the Dirac theory [see (20.21)], should coincide in a hydrogen atom. Among spectroscopists, doubts about the validity of this conclusion were expressed as early as 1934. However, the techniques of the time did not allow greater experimental accuracy, and the discrepancy between the theory and optical observations (that is, the splitting of the levels) being small, no great attention was paid to it. Better experimental data on this splitting were obtained consider ably later, when microwave spectroscopic techniques were used. The microwave spectroscopic method was invented and rapidly developed in the postwar years as a result of technical progress in microwave engineering. 7 Microwave spectroscopy, which has now developed into a special branch of physics, gives valuable results when used in the investigation of nuclei, atoms and molecules. Microwave spectroscopic methods are also applied to the physics of solids and liquids. In 1947 Lamb and Rutherford employed this method to studies of the 2si/ a and 2/?i levels, making use of a special property characteristic of the 2si 7j state. This state is
/d ,
By microwave ultrahigh- frequency radio emission we mean the region of the electromagnetic spectrum located in the wavelength range from millimeters to tens of centimeters - 10 Me). Successful application of microwave spectroscopy to the investigation of (10 atomic spectra is due to the fact that the distances between the components of the levels split by the relativistic, spin and vacuum effects are of the same order of magnitude as the
wavelengths in the microwave region.
323
metastable, since a dipole transition from the 2$ 1/a state to the lower lsi/ a state is forbidden by the selection rules A/ [see Eq. (20.4)].* Transition from the metastable state may be associated either with the emission of two photons (the probability of such a transition is 10 8 lower than that for the allowed transition), or with a preliminary transition to the 2p level. Lamb and Rutherford
investigated the latter type. Let us describe the general features of their experiment (see Fig. 20.2). A beam of hydrogen atoms in the unexcited Isi/, state is obtained as a result of dissociation of molecular hydrogen
at high temperatures (tungsten furnace). A bombarding beam of electrons then excites some fraction of the atoms in the hot beam (approximately one out of 10 8 ) to the metastable state 2si/ 2 . The metastable atoms, unlike the unexcited atoms, readily give up their energy of excitation upon striking a metallic target. In so doing they remove electrons from the metal. The resulting current is measured by a sensitive galvanometer.
Fig. 20.2. Diagram of the Lamb- Rutherford experiments on the detection of the splitting of the 2s and 2py, levels: 1) tungsten furnace emitting a beam of hydrogen atoms; 2) beam of electrons exciting the hydrogen atoms, 3) radio
t/,
frequency field; 4)target,
5)
galvanometer
If the beam of metastable atoms is subjected in transit to a perturbation capable of causing a 2s->2;? transition, then the atoms will almost instantaneously pass to the lsi/ state (prior to reaching the target). As a result, the current reading on the galvanometer is lower.
s
In
the
Lamb-Rutherford experiment such transitions were
induced by microwave radiation (the probability of the corre4 sponding spontaneous transition, proportional to w , is vanishingly small as a consequence of the smallness of w); a strong damping
This
transition
is
between these states
correct for a dipole transition, but calculation shows that the quadrupole is also forbidden
324
action, resulting in a decrease of target current,
was observed
at
This o> was assumed to be the resonance transitions 2si ->2pi,, or 2si/, -+2p>/, with causes which frequency a subsequent practically instantaneous transition to the lsi/ 2 level; the energy difference between these corresponds to fiw. Thus, one can very precisely measure the relative positions of the
some frequency
cw.
/Jt
levels
2si /a 2/;i /a
and 2;;^
9
.
These measurements showed that the level 2$i/ is shifted upwards relative to the level 2/>./. by approximately one tenth of the distance between the doublet levels 2/>i/, 2/h/,, which is equal
2
to (n in
y^
/?.
The arrangement
of the levels of a hydrogen-like
atom
derived from the Lamb-Rutherford experiment is given Fig. 20.3. The disposition of these levels according to the Dirac theory is given for comparison. According to the latest data, the shift of the 2si/, level is approximately 1057.77 Me or, in wavelengths, ~28 cm.
= 2)
OKI
?*> 7/2
Fig 20.3. Splitting of energy levels in the hydrogen atom rj) experimental data, /)) according to the Dirac theory (neglecting vacuum effects) The frequencies of the corresponding transitions and the distances are given in Me.
This apparently negligible discrepancy between theory and experiment led to remarkable progress in theoretical physics and, in particular, in quantum electrodynamics. This subject will be considered in greater detail in Chapter 22.
Lamb and Rutherford, the frequency of the microwave radiation fixed and the resonance condition, corresponding to the difference in the Zeeman components between the states 2M/2 and 2pi/2 or 2p3/, was obtained by adjusting the magnetic
In the experiments of
was
field
shift.
&%* Then,
extrapolating the results to the case
o%" =
0,
the authors found the level

E.
325
ANOMALOUS ZEEMAN EFFECT
The complete theory of the Zeeman effect (both normal and anomalous) must be based on the Dirac theory, because the latter takes into account both the relativistic and the spin corrections. Since the anomalous Zeeman effect is due to the spin effects in the atom, neither classical theory nor Schrbdinger s wave mechanics was able to give a satisfactory explanation of the Zeeman effect, and for obvious reasons. As a starting point of the theory let us take the approximate Dirac equation (19.14), in which these effects are taken into account
T
up to the terms (y
that is, H X according to (16.16),
z
Let the magnetic field be directed along the

-II V
axis,
Q,
..
>'%'.
Then, using the fact that,
'
(20.25)
we reduce Eq. (19.14), describing the motion Coulomb field of a nucleus, to the form
;
of an electron in the
;H^^
<
20 26)
-
where K rel
(19*24),
K s -- and V cont are given by Eqs. (19.15), respectively. Upon averaging of these terms
,
(19.23)
and
x
10
(20.27)
we
obtain the fine structure formula (20.16), that
is,
(20.28)
When a magnetic field is present, side of Eq. (20.26) the interaction
we
obtain on the right-hand
(20.29)
10 is of fundamental importance in Generally speaking, the spin-orbit interaction V this case. Since, however, the relativistic terms are of the same order as the spin-orbit
interaction,
""
we may
set
326
which gives the following value for the additional energy of the atom:
A/I
= lv5r J (*;;) (-
|- -f
d 1*.
a.;) (
J;)
(20.30)
The appearance of either anomalous (case of a weak magnetic field) or normal (case of a strong magnetic field) Zeeman effects depends on the relative proportion between the additional energies on
the right-hand side of (20,26). Let us assume that we have a comparatively weak magnetic field, whose interaction with the atomic electrons is smaller than the relativistic or spin-orbit interaction. Then the zero-order approximation will be expressed by the wave functions (20.2) that are obtained when the spin- orbit coupling is retained. Substituting these functions into (20.30), the additional energy
becomes
A
mag
= ^&T J
Rnf
|i
r dr
,|
d Q (r ,;> r t
,;)
Y <.
r is
(20.31)
In
(20.31)
we should
note that the integral over
equal to
unity
/?n i
,|V <//=!.
(20.32)
Substituting in place
(18.24)
of the spherical spinors their values from and (18.25), and using the orthogonality condition for the
spherical harmonics
we
/
= +
'
find
'/.:
the following expression for the additional energy
when
A mag
[(t _)_
m)
In exactly the
same way when
/==/
'/,
we
obtain

Recalling that m, as a single formula
327
=m
A
V
maS
the last two expressions
may
be written
=
*
(20.33)
where
W = -~
is the
Larmor frequency, and
the
Lande
g-
factor is
(20.34)
Thus, in the case of the anomalous Zeeman effect, the expression for the additional energy contains the Lande g factor, which in the case of the normal Zeeman effect [see (16.23)] is equal to unity.. The additional energy (20.33) does not lead to the usual triplet splitting (normal Zeeman effect), but to a more complex splitting
pattern (anomalous Zeeman effect). In view of the fact that mj can assume 2/-f-l different values,
each level
into
in the case of the anomalous Zeeman effect is split separate sublevels; that is, the external magnetic field completely removes the degeneracy, which is present even in the relativistic theory of the hydrogen atom.
2/4
-1/2
-1/2
3-2
.SL.
Fig. 20.4. Zeeman effect: a) position of the energy levels in the absence of a field, h) anomalous Zeeman effect,
<) normal
Zeeman
effect
To obtain the splitting pattern, it is necessary to take into 2 for the si/,, states, account the value of the Lande g factor (g 4 and so on) and for the for the states, ;>i/., states, g P)/ */8 g also the selection rules for the magnetic quantum number w7 . In particular, when Aw; =0, the emitted components are polarized parallel to the z axis (that is, parallel to the magnetic field), and when Am; the components are polarized perpendicular to the 1
/;,
magnetic
field.
328
Equation (20.33) gives us the following value for the frequency

of the radiation:
u>
o) ()
-f
(gm;
gm,),
(20.35)
netic field
the frequency of the radiation in the absence of a magM and J/" are the Lande g factors of the initial 0); # ( and final states; the magnetic quantum number m ; of the final state may take three values: w, -= w;, m ; 1. Figure 20. 4b shows the splitting of the spectral levels 1'si and
where
<
is
'
2'pi
in a
weak magnetic
field, the
Lannor frequency being taken
as the unit of the splitting. From Fig. 20.4b it is seen that in this case there are four, and not three (as in the normal Zeeman effect) shifted lines. The magnitude of the displacement is given by
(20.35). In the
case of a weak field [according to (20.34)] we find
Hence
.K-Jr,.
=---;!"
36 )
atom and
Equation (20.34) for the I ande ^factor is applicable to the hydrogen to atoms having a single valence electron. In the general case, the Lande g factor becomes
where L, S, and / are the orbital, spin and total angular of the atoms and
J
In particular, for
momenta
= \LS\.
s
Eqs.
L elements of the first group (/ l, 5 /, /) and the are The Lande g same. (20.37) identically (20.34)
!
factor attains
its
maximum
value for
states
(1
= = = = = s=
l
0,
/t):
gs
=2
(20.38)
For atoms with two electrons in the outer shell (for example, helium atoms), single lines (5 0, J L) are possible along with the triplet state S=l. For the single lines we have g= 1, and, therefore, in this case spin effects should be of no importance! only the normal Zeeman effect (that is, triplet splitting) should be observed in either a weak or strong field.

F.
329
STRONG MAGNETIC FIELDS. PASCHEN-BACK EFFECT
It
in the
has been indicated that the anomalous Zeeman effect appears case of weak fields, when the external magnetic field cannot
disrupt the spin-orbit coupling,
ma Mathematically this means that AE g [see (20.33)] is much smaller than the natural splitting of the lines AE S *' given by Eq.
(20.28)
AE s -<v -AEâe.
(20.39)
In the latter case we first solved the problem by taking into account the spin-orbit interaction; this establishes a relation between the spherical harmonics that form the spherical spinor (18.24) or (18.25); then we found an additional energy that leads to the anomalous Zeeman effect, since the Lande j factor does not equal
unity.
In the
magnetic
case of strong fields, when the splitting due to the external field is greater than that due to the spin-orbit interaction
AEroaJV-AE
the magnetic field
8-,
(20.39a)
"breaks" the spin-orbit coupling and the zeroorder approximation solutions, expressed in terms of spherical spinors [see (18.24 and 18.25)], are no longer true. Vs and In this case we may neglect the interactions V /rel
*
,
/cont
in (20.26),
which, when (20.29)
is
taken into account, becomes
Using the fact that the functions

the
Fj
and
i
W must be
2
i
proportional to
i
spherical harmonics
Yj"
with
^-
yy
= mYy
>
we
find
two
independent equations for these wave functions:
_
from which
it
(20.41)
is
evident that the additional energy equals

(20.42)
and
A (&E ) m ms
*=p^ (Am
330
that is, the
wave function î corresponds

is
electron spin wave function
to the case in which the directed along the magnetic field (m s */t), and the l F a corresponds to the case in which the spin direc-
tion is opposite to that of the magnetic field. such that the same energy value is obtained If we choose m m 1 for the functional , for both functions, then we must set m 1
and
^=m-\-l for the function Wj. In this case the wave functions
m,1
(20.43)
will be mutually orthogonal, so that
dr
d<2
(Y?-
<
Yf
HI
=0.
(20.44)
Since the interaction between the atom and the external magnetic field (20.29) contains only the matrix a',, which does not couple the wave functions l and W, transitions from the state with m,= Va to induced by this interaction, will be forthe state with m V* bidden in this case and hence A/?z Taking into account this circumstance, and also the selection rules for the quantum number m (A/7/ 0, :.!), we find from (20.42) an expression for the Zeeman splitting of the spectral lines
l
=
f
<\ui
= oAm = O
-o
(20.45)
which agrees with the result of Schrodinger's theory, which explains

the
normal Zeeman
Thus,
in
strong fields
effect (triplet splitting of the spectral lines). ma ^> A s --), the anomalous effect (A
becomes
the normal effect, which is in agreement with experimental data (Paschen-Back effect). It is interesting to note that the passage from the anomalous Zeeman effect to the normal effect can be illustrated by Fig. 20.4, if the Lande g factor is set equal to unity (see case c). Then the splitting will be
that is, we obtain three components of the split line instead of four. St nia8 for one In special cases, when AE energy level and *<A mft * AE f or i eve i f or when the energies e er conversely <>> of both levels are of the same order of magnitude, the Zeeman
^ ^
splitting becomes complex. Since these are all special cases, shall not elaborate them here.
we
331
roblem 20. IJ Investigate the dlamagnetism and paramagnetism of atoms by placing homogeneous magnetic field (H*=Hy *Q H ; &%"); contrary to the investigation of the Zeeman effect, keep all terms containing g%âs well as terms proportional to <2%^ in the Hamiltonian [see (20.25)]. Indicate the atoms in which diamagnetism may be observed. Solution, When terms portional to e^âre taken into account, we have, instead of
t
(20.25)
Consequently, on the right-hand side of (20.26) we have another term
which in conjunction with (20.33) gives the following expression for the additional energy of an electron in a magnetic field:
Here when calculating the perturbation energy proportional to i 2 -l-.V 3 we have used the spherical symmetry, which must occur in the zero-order approximation -=0), and
,
i
(,.
have set
Hence
the magnetic
atom
1
=--=---- =
dA/i
moment
of
an atom in a magnetic field
is
mag
elWr*
param
diam
,
The latter relationship is a generalization of a familiar equation relating A/, ma S a in the case where A/^mag i s a nonlinear function of jtf\ The diamagnetism of atoms is characterized by the second term of Eq. (20.47), which is proportional to ^j/T
,
The magnetic susceptibility per gram-atom
is
where A^ is Avogadro's number. The quantity /.diam j g never zero and must always be negative (r a j>0). 'Iherefore, the diamagnetic effect must occur in all atoms. As for the first term on the right-hand side of Eq. (20.47), which is proportional to
my,
it
may
take either positive or negative values, since
my ~
.j* ,
~~
,
...
j.
the negative values for trip which give a smaller value for the energy, will be preferred. On the average we therefore obtain a positive value for the paramagnetic susceptibility
In a state of
thermodynamic equilibrium, however,
>
param_ ~~[t^ _ /(;+ 3 kT
1)
Expression (20.47) is obtained for a weak magnetic field, when the anomalous Zeeman effect occurs. It can, however, be easily extended to the case of a strong magnetic field. To do this we must set ff m in (20.47) [see (20.42)]. 1, m ;
U See
Becker, Electron Theory.
332
Since the paramagnetic susceptibility is considerably larger than the diamagnetic / diam) t the atoms exhibit paramagnetic properties when j -+. 0. For hydroger-like atoms ; differs from zero (the minimum value of / equals l /s) and, Only for atoms with an even number of therefore, they are always paramagnetic. electrons can the quantum number / vanish (for example, parahelium in the ground state, an atom be will see Chapter 24). Such diamagnetic.
aram susceptibility (XP
>
eeman
Adding geometrically the orbital and spin angular momenta, show that effect is associated with the fact that the total magnetic moment not parallel to the total angular |i is momentum J. With the aid of the geometric model, also explain the Paschen/ -. Back effect. >V Solution. First let us find geometriw\ cally the angular and the magnetic moment vectors WH
In the geometric representation (see Fig. 20.5) 'we may choose the scale so
that u
'
^~ 2rn
(it
is
immaterial
at
the moment whether the vectors parallel or antiparallel).
are
Then
Fig 20 5 Geometric interpretation of the anomalous Zeernan efiect
|i,=-L, and
|i
= 2S
that is, the total magnetic moment ji will undergo two rotations in a magnetic field: one with an angular velocity <> around the direction of the total angular momentum (this angular velocity corresponds to the frequency associated with the transition between components of the spin-orbit splitting of the spectral lines <o - J\<*~), and the other corresponding to the Larmor frequency of precession around the direction of the magnetic
field
H (//, = // = 0,
(D
II s
= ,jn-
When
>
o,
the additional energy should be calculated
from the relation
J3
Since the magnetic
moment
is
directed on the average along
J,
we have
where
f = -"
W
,
and the Lande g factor
is
equal to
-j.
cos (L
y)
+2
S
-j
cos (5-
Substituting the values of the cosines of the angles
2JL
and remembering that J 3 WJ (J 1), and so on, we obtain the expression (20. 37) for Lande' g factor. Let us pay attention to the fact that if the vectors |i and J were parallel, the Lande' g factor would be unity,
the

In strong fields o
333
>
<o
we must consider independently

z axis.
the spin
moments
ji
around the
Then
the rotation of the orbital and the additional energy becomes
= - ji.^ =^|
which leads directly
to the
(U
+ 2SJ =
ofc
(m
1),
normal Zeeman
splitting
(Paschen-Back
effect).
/Problem 20.3^ By means

tion,
of the relativisuc scalar
wave equation and the Dirac equa-
Show
and
3 states. MIR! 'HIS 'frequency of the allowed transitions between the n - 2 and that, according to the Dirac theory t there are seven lines, five of which are distinct,
that,
according to the scalar theory, there are only three distinct lines.
Chapter 21
The Effect of Nuclear Structure

on Atomic Spectra
A.
INTRODUCTORY REMARKS
As has been mentioned in Chapter 13 t the position of the spectral lines is shifted when the finiteness of the nuclear mass is taken into account. The Rydberg constant /? in the expression for the energy of a hydrogen-like atom
* > (21.1)
\
n
is
somewhat reduced and becomes equal
to
*=
where
*(! -$)
PI.2)
the Rydberg constant corresponding to infinite nuclear mass. Consequently, the Rydberg constant will have somewhat different values for hydrogen, deuterium, and
is
tritium. With the great accuracy attainable in modern spectroscopic techniques, this effect can be used to detect the presence of different isotopes (see Chapter 13). In a similar fashion, the finite size and the magnetic moment of the atomic nucleus have certain effects on atomic spectra,
B.
EFFECT OF THE FINITE SIZE OF THE NUCLEUS
When the motion of an electron in the field of a nucleus is treated as a problem in classical theory, it is quite immaterial whether the nucleus is regarded as a point or as a particle with finite dimensions. All that matters is that the electron should at all times be outside the nucleus and that the nuclear charge should be spherically symmetric, since the potential outside a spherically symmetric charge distribution is the same as the potential of a point charge. In quantum mechanics the situation is somewhat different. The wave function must differ from zero inside the nucleus; therefore, there is a certain probability (however small) that the electron will be located inside the nucleus. Consequently, the charge distribution inside the nucleus must in some manner influence the energy levels of the electrons in the atom. To estimate the effect of the finite nuclear size on the energy spectrum of a hydrogenlike atom, we shall assume that the nucleus can be represented by a sphere of radius with charge distributed uniformly throughout the volume. The potential energy will y?jV
be given by (see Fig. 2U1)
THE EFFECT OF NUCLEAR STRUCTURE ON ATOMIC SPECTRA
335
=-
for
(2U)
The shift of energy levels due to the finite size of the nucleus can be calculated with the help of perturbation theory. We shall assume that the perturbation energy consists of the difference between the potential energy of a point nucleus and the potential energy of a nucleus with charge uniformly distributed over the nuclear volume
71-iH
with
1
(21.5)
for
for
< R Nt
(21.6)
.
r>/? Ar
The perturbing force therefore
differs
from zero only inside
the nucleus.
Potential energy with the dimensions of the nucleus taken into account The dashed curve shows
Fig. 21.1
finite
the variation of the potential energy
which would be obtained inside the nucleus, if the potential were described by the Coulomb law both inside and outside the nucleus
In first-order perturbation
theory the shift of the levels is given by
AEvol
s
| I
<|,*vol tyd* Xt
Since does not change appreciably in the region r ty /?, this integral can be readily evaluated by substituting for ^ j" its value at the origin.
|
(2J.8)
336

3
|
Substituting the expression (20. 14) for
fy
(0)
we
obtain
that is, in nonrelativistic theory, the shift of the
for s states
It
(/
energy levels
is different
from zero only

IRj\r\* order - I
0;.
can be shown that for p levels this

a
shift will contain a factor of the

/A*
4
(where a rt
a n/
),
and for d states a factor of the order of
^j
and so on.
Con-
sequently, the shift in the energy levels for p and d states can be neglected in the first approximation. For hydrogen, the first-order shift in the energy levels is about 1 Me; this is much too small to account for the Lamb shift, which is equal to approximately 1,057 Me (see
Chapter 20X The volume of the nucleus is important in connection with the isotope shift, that is, the shift in the energy levels of atoms with the same atomic number Z and different mass numbers A. The chief factors that give rise to the isotope shift are the different masses (the mass effect) and different volumes (the volume effect) of the isotopes. The mass effect is manifested in a shift of the spectrum lines towards the ultraviolet as the the highest frequencies are found mass number A increases. For example, for Z in tritium with A 3. then deuterium with A 2", and, finally, ordinary hydrogen with A [see (21.2) and' also Chapter 13]. On the other hand, the volume eliect is manifested in a shift of the spectrum lines towards the infrared as A increases. For instance, it can be seen from Eqs. (21.1), (21.9) and (21.13) [see below], that the energy levels of a hydrogen-like atom will be given by the following expression when the shift due to the volume effect is taken into account:
1
RZ*hl
Z-/?8
Vi
n*~V~~5
1/T"
Experiment shows that an isotope shift towards ultraviolet is observed in elements whose atomic number Z is less than 40-50. For elements with a larger value of / an Isotope shift in the opposite direction is observed, that is, towards the infrared. This indicates that, for relatively light elements, the isotope shift is caused mainly by the mass effect, whereas for heavier elements it is caused by the volume effect. This, however, is a rather simplified picture of the isotope effect, and other features associated with the structures of the atom (for example, nuclear spin and polarization of nuclei by electrons) also have to be taken into account.
C.
The
MESIC ATOMS
finite
energy levels in a meslc atom

a nucleus. The
size of the nucleus has a particularly important effect on the position of an atomic system consisting of a u meson revolving about u meson is a particle that has the same spin as the electron (that is, spin
To calculate
pansion of
the shift of the p levels
we must
substitute the second term
the ex-
|0|(r) into (21 7),
namely,
2
dr
since the main term
2
'
|i^(0)|
vanishes for p states
337
207/H ), so that the }i meson is 1/2 in units of ft) and a mass 207 times greater (m |X basically a "heavy" electron. Me sic atoms can be produced by passing negative JLI mesons through matter. After losing its energy and slowing down, a p, meson may be captured In an orbit about a nucleus, forming in this way a ji-mesic atom. Mesic atoms have been obtained for almost all elements of the periodic system, from hydrogen up to the heavy elements (uranium, neptunium, and so on).2 The motion of a fi meson about the nucleus is determined mainly by the Coulomb attraction, just like the motion of an electron. A u meson, however, also has nonelectromagnetic interactions with the electron neutrino and nuclear fields; these interactions may result in spontaneous decay of the p, meson into an electron, neutrino, and antineutrino 2.2 x 10" sec). The p. meson has, In addition, (the lifetime of a jz meson at rest is T a definite probability of being captured by a nucleus. Ihus the lifetime of a u-mesic atom is determined by two competing processes: natural decay of a JLI meson into an electron, neutrino, and antineutrino and nuclear capture of the ji meson. In light mesic atoms ( Z <: 10) the probability of the first process is greater than the probability of the second; that is^ the lifetime of a mesic atom is determined by the lifetime of a meson at rest (T^10~ secX For Z>10, nuclear capture begins to predominate ji 3 and the lifetime slowly decreases to 7 10~ 8 sec (for Z 81>). In the theory of }i -mesic atoms, ordinary electrostatic interaction plays a fundamental role. In the first approximation, we can regard the nucleus as a point charge and calculate the energy of the mesic atom and the radius of the orbit using the equations derived for an ordinary hydrogen-like atom, replacing the electronic mass by the mass of a p, meson. Then the energy and radius of the orbit will be given by [see (13.33),
(13.45)]
207/// . It can be seen that the energy of the u meson in a mesic atom is tn^ 207 times greater than the corresponding energy of an electron in the atom, and that the radius of the orbit on the contrary is reduced by the same factor. If electrons remain in the atom along with the p. mesons, they will move about the nucleus in considerably larger orbits than the p. meson and therefore cannot exert a significant Influence on the meson rotating around the nucleus. A mesic atom, therefore, may be regarded as a p, hydrogen-like atom that can have both large and small values of Z. Since the radius of the "Bohr" orbit of the meson is 207 times smaller than that of the electron orbit, the probability that the meson will be located in the nucleus is considerably greater than for an electron in a hydrogen-like atom. The main correction to the energy levels of a mesic atom will therefore come from the volume effect. The equation for the energy of s states can be obtained from (21.9) and (21.10) by replacing the Bohr radius GO by the corresponding radius in a mesic atom
where
= ~>>
(2UD
obtaining, therefore,
These subjects
For
are treated in
more detail
in a
paper by
D D Ivanenko and G E Pus-
tovalova, llspekhi fiztcheskikh nauk, 61, 27 (1957)

77 mesons, which strongly interact with nuclei and are responsible for nuclear the lifetime with respect to decay into a t meson and neutrino is equal to 2 6 X 8 10~ sec, whereas the lifetime with respect to capture by a nucleus is many times smaller
forces,
In particular, for Tr-mesic
hydrogen the capture time of a negative pion from the orbit is of
338
Using the fact that the nuclear radius is related to the mass number expression
A^ 2Z
by the
(21. 13)
where
const,
we obtain
in the first
approximation
i-o,
yi
J
v/
'
follows that t for the s levels, the energy correction due to the nuclear volume will be proportional to Z*'* and therefore attains very large values for heavy elements. In heavy elements, the orbit of a ji meson may even be inside the nucleus, at large Z and small n. lhe Bohr radius of the mesic atom becomes equal to the nuclear radius
It
r
forZ
Z cr -45
Z cr ) 9 the main part of the potential will no longer For orbits inside the nucleus (Z be determined by the Coulomb law (21.4) but instead by formula (21.3) which corresponds to the potential of a three-dimensional harmonic oscillator (on the assumption of a simplified model of the nucleus in which the charge is uniformly distributed over the volume). Thus the energy has to be determined from the following equation instead of
(13.4):
if'/?
,
>
dR
dr
dr*
^
1
?/*,.
ft-
X
In Eq. (21.15) let us
r + 2/?7
3^jj
Ze^
7^7
1(1
'2
tm-r*-)^
\-U1P\n
(21.15)
-
change the variable by setting
R=Y
r ^'. r
P"-
Since
it follows that the equation for the energy of a three-dimensional harmonic oscillator formally identical with the equation for the hydrogen-like atom
is
but has different values of the constants, namely t
4ft
= /'C' +
l),
(21.17)
To determine
the eigenvalues of the energy
we may use Hq.
(13.20), according to
which
Substituting for B t A and /'their values from (21*17) and using the fact that k we can find the energy of the meson in an orbit inside the nucleus:
- /I,
(21.18)
/+
8
s ),

where
the frequency of mechanical vibrations of the three-dimensional oscillator is
339
The quantity
it
is three
8 / 2 ft(o represents the zero-point energy of the three-dimensional oscillator; times greater than the corresponding zero-point energy of a one-dimensional
oscillator.
The quantity V
=v 2
>>
- is the greatest depth of the oscillator potential well.
Equation (21,18) is correct on the assumption that the potential energy varies in accordance with (21.3) from zero to infinity. If the finite size of the nucleus is taken into account, an additional energy is obtained that represents the difference between the potential energy of the particle in the oscillator well and its potential energy in the Coulomb field (this difference being averaged over the space outside the nucleus X The equation for the additional energy is
AE
vol
=J
(21.19)
Since the wave function of a spherical oscillator is similar to the wave function for a field (It is determined from the same wave equation with r replaced by P r 9), it decreases exponentially as r increases. The __ correction (21.19) is therefore significant only $, when the radius rn of the me sic atoms is close
Coulomb
______________
to
let us consider the ?p ~* Is transition in lead (Z 82). If the mesic atom of lead is assumed to have a point nucleus, the energy released in this transition can be found
RN
As an example,
..
fQ
'*'/*
from
(21.10)
'
E3
The
the
/; 4
r=
s f* 1m Z -1--
=M
^'^
^'^
Ene
"gy
levels in a mesic
Mev.
(21.20)
atom
relativistic and spin effects in the
amount
A
?/J 1/t2
* ls
t/i J
transition increase this energy by
(A/;)
^2
Mev.
(21.21)
Incidentally, such a significant role of the relativistic and spin effects is due to the fact that the energy is expanded in terms of (Zot) s . This quantity is comparatively large for
82, 1/137). Comparison with experiment shows, however, that in this transition an energy of 6 Mev is liberated instead of the predicted 16 Mev. This discrepancy between theory and experiment arises because the 1 s state for lead lies inside the nucleus. In the >p state the orbit lies outside the nucleus and the volume effect of the nucleus is small. If we take the energy of the Is level from (21.18), and the energy of the 2p level from (21.10) for a point nucleus, the energy of the 2p Is transition will be 3.6 Mev. If we add to this the correction (21.19) for the energy of the Is level, the energy of the transition will come to about 5 Mev, which is relatively close to the experimental value. A study of the multiplet structure of the '}p level in mesic atoms enables us to determine the spin of a ji meson. For a particle with integral spin, the level splits into an odd number of components (for spin no splitting occurs, for spin 1 three lines are observed, etc.). Since the 2p level splits into two components (2/?i and 2p*/ a ), it /a was established that the spin of a p, meson is 1/2. On the assumption of a point nucleus, the theoretical splitting calculated from Eq. (20.17) should amount to about 0.55 Mev, with the 'Ipif and 2si/ 3 levels coinciding. When the finite size of the nucleus is taken into account in a mesic lead atom, the splitting of the 2/?/ 3 levels is reduced 2pi/^ to 0.2 Mev and the level is raised above the 2/? level. This is illustrated in 2s, x
{
lead
(Z=
a=
8/
340
Fig. 21.2. From the above data It follows that heavy me sic atoms will emit gamma quanta having energies of several Mev. Lighter mesic atoms emit x-rays. Because of the significant influence of the size of the nucleus on the spectra of heavy mesic atoms, the charge distribution inside the nucleus can be determined from an analysis of the spectra. It has been found that the value that should be substituted for /Vo in the formula for the electromagnetic radius of a nucleus with mass number A:
KN *=R A
is
1, 2
'*
(21.22)
x 10 -13
cm
rather than
1.4 x
10-13
cm
(the
value assumed for nuclear inter-
actions),
determine the magnetic moment of the
Similarly, the multiplet structure of the spectral lines of mesic atoms can be used to its value is close to the muon magneton ji meson;
1. where the Lande g factor is g The theory of -mesic atoms is based mainly on electromagnetic interactions. By contrast, in the theory of TT -mesic atoms, a great part is played by the nuclear interactions, the theory of which is far from complete. Further experimental study of TT -mesic atoms and an explanation of the semiempirical laws that describe their behavior will have important bearings on future work in the theory of nuclear forces. These topics, however, lie beyond the scope of this textbook.
JLI
D.
APPLICATION OF THE DIRAC EQUATION TO THE NEUTRON AND THE PROTON
The Dirac equation describes the motion of particles with spin 1/2. It applies to electrons as well as to protons and neutrons. In the presence of an electromagnetic field it is necessary to take into account the charge of the proton, as well as the so-called anomalous magnetic moments of the proton and the neutron. We recall that the energy of interaction between a charged Dirac particle and an electromagnetic field is
V e = e<l>e **A
(21.24)
In the nonrelativistic approximation this expression contains a the intrinsic (spin) angular momentum (h '2o)
magnetic moment due to
This quantity is known as the kinematic or Dirac magnetic moment. In passing to the relativlstic equation, we must replace the mass ///,,in Eq. (21.25) by Its relativistic value
/W
-
3 ,
and, therefore, the Uirac magnetic
moment vanishes as
the velocity approaches
the velocity of light (v c). In addition to the Dirac magnetic
moment, which appears only in the nonrelativistic approximation and which depends on the charge, a particle may hfve an anomalous magnetic moment that does not vanish even in the relativlstic case and is independent of the particle's charge. We shall now find the energy of interaction due to the anomalous magnetic moment. The energy of interaction (21.24) of an electron with an electromagnetic field is a scalar quantity, since in four-dimensional space il ^l AÂi, Vy /lj, A? <4 8. In the same wav the unit matrix I is the fourth component of the velocity matrix a ^ 4 The interaction energy (21,24) may, therefore, be represented as a (that is, a 4 /l). scalar quantity in four-dimensional notation
1 ,
4 More precisely, the quantity }^ = en/iâî// [see (17 32)1, where transform as a four-vector
a-
^1,2, 3,^
'!
Wlil

The electromagnetic
field
341
forms an antisymmetric tensor of second rank
"-- ^V
dA
// v =//,i,
<
21- 27 >
where
It
follows that
//,=//,
The interaction energy
is, therefore,
//,= /
*
of the anomalous magnetic
moment
with the electromagnetic field
given by
i
KOT
where
=n
JX,
>]
V-^l
//
[JlV
(21.29)
a is a second rank tensor composed of the Dirac matrices. s [lv Using the rules for transformation of a wave function under the Lorentz [see (17.38)] and spatial rotations [see (17.39)], we can show that the quantities
are the matrix elements forming a second rank tensor. The energy of interaction between the anomalous magnetic moment and the electromagnetic field takes the form
l'm==Mp 3 '-//-kp a *-l
(21.31)
electron has a charge, a spin, and also a Dirac magnetic moment. Its anomalous magnetic moment is relatively small (see below). A neutron has no charge, but it does have an anomalous magnetic moment; this magnetic moment determines the interaction between the neutron and the electromagnetic field. As for the proton, it has both a charge and a spin, and hence a Dirac magnetic moment; in addition, it has an anomalous magnetic moment. It should be noted that nuclear interactions are of great importance in the theory of nucleons.
An
E.
EXPERIMENTAL DETERMINATION OF THE MAGNETIC MOMENT OF THE NEUTRON AND THE PROTON
The procedure for determining the magnetic moment of the neutron, proton, and complex nuclei is basically the same as for the magnetic moment of the electron (the Stern-Gerlach experimentX The basic principle consists in applying a magnetic field perpendicular to the direction of motion of the particle. The particle will react differently depending on whether its magnetic moment is oriented parallel or antiparallel
to the field.
Let us first consider the possibility of determining the Dirac magnetic moment and the anomalous magnetic moment of a free particle. Suppose a free particle moves perpendicularly to the z axis. The Hamiltonian describing its motion has die form
H = câ^ + cpja^ +
The component
motion
of the intrinsic angular
p 3 /;/ c
2
,
(21.32)
to the direction of this
momentum
perpendicular
S,= 4More precisely, the quantity
fl
'
(21.33)
^^ny^v
*s a
second rank tensor
342
does not commute with this Hamiltonian. The component of the total angular along the z axis does commute with the Hamilton! an
n<3*>
momentum
(21.34)
and can therefore be determined exactly together with the energy. Let us evaluate the error in the determination of the orbital angular
momentum by
means
of the uncertainty relation. If the origin of the coordinate system Is taken to be at the center of the wave packet, the error in the orbital angular momentum will be
In
accordance with the uncertainty relation, we have
v *y A** A/?
~~
Since the
A.V*
errors
may
be either positive or negative, we find

.
| I
Ax
|
ti
(21.35)
to the Ay |. motion of the particle is of the order of the spin, and therefore the perpendicular components of the intrinsic angular momentum and the Dirac magnetic moment cannot be determined simultaneously.
will
The error
bL z
be
minimum when
AA*
Thus the error bL z due
translational
Transmitted
Fig. 21
Experiments
moment
for determining of the neutron
the
magnetic
We recall that the Stern-Gerlach experiment allowed for the determination of a magnetic moment of a bound electron; since, however, the orbital angular momentum in the s state is zero, it was the spin (or Dirac) magnetic moment that was actually measured. In accordance with (21.31), the interaction energy associated with the perpendicular component of the anomalous magnetic moment is
I/
vm
'
===
S
.
'
I^PJ^S <2sl
{1*30)
/O
7A\
This component commutes with the Hamiltonian (21.32) and can therefore be measured exactly. Consequently, it is possible to measure the magnetic moment of a free neutron when the magnetic field is perpendicular to its motion, as was done by Bloch and Alvarez b In their experiments when a beam of neutrons was passed through a piece of (1940). 7 magnetized iron, the most pronounced scattering was observed for those neutrons whose magnetic moment was parallel to the magnetic induction vector inside the iron. The
emerging beam consisted, therefore, mostly

antlparallel
to
of neutrons whose magnetic moment was the magnetic Induction vector. If the neutron beam passes now through two magnetized iron plates in succession, the experiment is completely analogous to the transmission of light through two Nicol prisms; that is, the first iron plate acts as
Similarly,
it
may be shown
that only
the longitudinal
it
with the Hamiltonian
In principle, therefore,
component of spin commutes also can be measured experimentally

Actually,
their only interactions
Neutrons have no electric charge and they pass quite freely through matter occur on collisions with the nuclei.
343
a polarizer and the second as an analyzer. This phenomenon was used to determine the magnetic moment of the neutron. A schematic diagram of the Bloch- Alvarez experiments is given in Fig. 21.3. Unpolarized neutrons moving along the direction of the x axis pass through the polarizer (first iron plate, with a magnetic induction vector directed upwards). The emerging beam consists mostly of neutrons whose magnetic moment is directed downwards. These polarized neutrons will pass freely through the analyzer if its magnetic induction vector is directed upwards, like that of the polarizer. On the contrary f they will be transmitted much more weakly if the iron plates are oppositely magnetized. Between the polarizer and analyzer there is a device that reorients the neutron spin. This device is similar in principle to the one used by Rabl in his nuclear magnetic resonance experiments to determine the magnetic moment of the proton and of heavier nuclei. The basic principle of the instrument is as follows. In the space between the polarizer and the analyzer, a relatively strong, constant magnetic field is applied
parallel
to
their
moment
is antiparallel to this
magnetization vectors (see Fig. 21.3). A neutron whose magnetic magnetic field acquires the additional energy
!*<#".
If,
(21.37)
however, the magnetic moment
is parallel to the field, the
neutron loses an energy

(21.38)
//
P ar
rrr-jj^ST.
In addition, there is a relatively weak oscillatory magnetic field larly to the field
applied perpendicu-
Hy = A
cos
f.
This oscillatory field will reorient the spin of the neutron particularly strongly when the frequency w is close to the resonance frequency
(2U9)
At this resonance frequency, the number of neutrons undergoing a reorientation of the magnetic moment reaches its maximum value. To find when this happens, it is necessary to determine when the number of neutrons passing through the analyzer is minimum in the case of parallel magnetic induction vectors, or maximum in the case of antiparallel magnetic vectors. Once the frequency <o has been determined (it is, in effect, twice as great as the Larmor frequency of precession), the magnetic moment of the neutron can be found, 8 According to recent data it is equal to
where the
unit for
measuring magnetic moments
is the
nuclear magneton
where
is the mass of a proton and JA O is the Bohr magneton. the resonance frequency of the oscillatory field, we can determine the magnitude of the magnetic moment but not its sign. If, however, we replace the oscillatory magnetic field by a rotating magnetic field, we can also determine the sign of the moment, since for resonance it is necessary that the vector equation o> 2o should hold, where
From
mp
<jw
is the
Larmor frequency
of precession of the neutron spin.
The minus sign
shows
that
the
magnetic moment of the neutron, just as for the electron, is directed
opposite to the spin.

8
Subacquent improvements on the Bloch and Alvarez experiments are described in

I,
E. Segre, Experimental Nuclear Physics, Vol.
1953.
344
The resonance frequencies at which the magnetic moment of a neutron, proton, or other nucleus is reoriented in a field Q%"'~ 10* oersted are in the microwave region. From (21.39) these resonance frequencies will be of the order of 10 2 Me (this corresponds to a wavelength of several meters). The magnetic moment of the proton can be found only in a bound state since the proton is a charged particle and therefore has a Dlrac magnetic moment in addition to the anomalous magnetic moment. The experiments cannot be carried out with individual atoms, but only with molecules in which the electron spins cancel out. For instance, the magnetic moment of the proton can be found by applying the method of nuclear magnetic resonance to the H^ molecule. To obtain molecules with polarized protons, we do not pass them through magnetized iron (H 2 molecules do not pass through iron), but instead split the beam by passing it through an inhomogeneous magnetic field in the same way as in the experiments of Stern and Gerlach. This method of measuring the magnetic moment of a proton was first used by Rabi (1934), even before the experiments of Bloch and Alvarez. The magnetic moment of the proton has a positive sign; that is, it is parallel to the It is equal to (in units of a nuclear magneton). It contains two -f- 2.7928 spin. p 9 components: one (fx = I) corresponds to the Dirac magnetic moment and the other
\*.
= 1.7928) to the anomalous magnetic moment. The anomalous magnetic moments of the proton and neutron are approximately equal and have opposite signs. The magnetic moments of the deuteron (f*]-)= 0.8573) and other nuclei have been determined by similar methods. One of the main results of the experimental determination of the magnetic moments of nuclei was the fact that they are of the order of the nuclear magneton, whereas the magnetic moments of atoms are of the order of the Bohr magneton, which is (m p //w a ) times larger than the nuclear magneton. This difference served initially as a strong argument in favor of the proton-neutron model of the atomic nucleus, according to which electrons (which have a large magnetic moment) are not present in nuclei.
(j,anom
F.
HYPERFINE STRUCTURE OF THE
HYDROGEN SPECTRUM
The hyperfine structure
netic
the
field
moment
hydrogen
of spectral lines is due to the interaction between the magof the nucleus and the magnetic moment of the electron. If the nucleus of
atom
acting
Paul! matrices
a magnetic moment n p where ap are the (Z 1) has ppVp on the wave function of the nucleus, it gives rise to a magnetic
4=Vx 5VX tf=Vx

the the
A.
This magnetic field acts on the magnetic moment of the electron ji o (a acts on fA wave function of the electron), which results in an additional interaction between nucleus and the electrons and leads to the hyperfine structure of the energy levels
r 1
(21.40)
In the first approximation we may assume that there are no preferred directions and, hence, making use of the equations
(W)(vV)==^-(a'-ap )V
and
(21.41)
This should properly read:
ir
At?

we
obtain
345
<
21- 42 >
Consequently, In the first approximation, the interaction of magnetic moments just like in (21.42) the contact interaction influences only the s state. The expression (o'-a^) can be found from the following simple considerations. The spin matrices of the proton v' and the electron a must satisfy the relation
1
IW
where S
is the
(a'
a' )* p
= ft'S (S +
1),
(21.43)
absolute value of the total spin, which is equal to either zero (antiparal-
lell spins)
or
to unity (parallel spins).
Then
Using the fact that
12
+ o^ =
2
6,
we
obtain
(a'.ap
= 2S(S+l)-3.
)==
(21.44)
Since integration when the
function is present gives
<MO)
we
obtain
the
following expression for the
shift
of s levels (hyperfine structure) of
hydrogen:
A 4
,
= |- mp, -L, [2S (S + 1)
3]
(21.45)
where a
from Eq.
fi*
MOÔ
is the
radius of the first Bohr orbit, and the value of
8
|

proton and the electron are parallel (S
1);
then
,-.
(2U7)
The difference between these levels represents splitting of the s term due to the interaction between the electron and the magnetic moment of the nucleus
(21.48)
If we use (21,48) to calculate the s-term splitting for the case the value of ILP obtained In Rabi's experiments and setting fi neton, we find
1, then substituting equal to the Bohr mag-
A(o
theor
=1417 Me.
On
the
microwave spectroscopic methods has given
other hand, a careful experimental verification of the splitting of this level by the value
Aw e *P == 1420 Me.
346
The

relativistlc corrections and the corrections for the finiteness of the nuclear
mass
160 *" ex to the required value Ao> do not raise the frequency Ato* P. The proton's magnetic moment has also been measured very accurately. To explain this anomaly, therefore, it remained to assume that the magnetic moment of the electron is not exactly equal to the Bohr magneton, but is instead somewhat larger, Kusch and Foley showed that to obtain agreement with experiment the magnetic moment of the electron must be taken to be
(2U9)
where, according
to recent data,
6
= 0,00116.
have a very small anomalous magto
These considerations show

netic
(
f
that an electron will
moment
u).
anom =
/JL
fi
&
in
addition
the
Dirac or kinematic magnetic
moment
We
shall
discuss the nature of the anomalous magnetic
moments
in the next
chapter,
Concluding this section, it is necessary to point out that the hyperfine structure cannor explain the Lamb shift of the 2si/ a level (1.058 Me relative to the 2/7 1/^ level). First of all, it follows from (21.48) that the splitting of the 2si term is of the order of 200 /0 Me, and, in addition, the center of mass of the s terms is not shifted. Suppose, the 1 (parallel spins of the electron and the proton) is raised by a certain level with S amount [see (21.47)]; then the level with S (antiparallel spins of the electron and the proton) is lowered by an amount three times as large [see (21.46)]. Since a state with S 1 is three times more probable than a state with S (when S == 1 the spin may be directed along the z axis, opposite to the z axis and perpendicular to the z axis), the center of mass of the s states remains unaltered, so that it occurs in the same position as when the magnetic moment of the nucleus is neglected. The hyperfine structure cannot therefore account for the Lamb shift. The theory of this effect will be considered in the next chapter.
shift of the Is levels in the light p -me sic atoms as a result in&QElHJir of the nuclear structure, taking into account the nuclear motion and also the variation of the wave function within a distance comparable to the size of the nucleus. Hint. The shift of energy levels should be calculated by the perturbation method. We
roblem 21.Q) Find the

.....
of
iliS
use Eqs. (21.5) and (21.7) where the wave functions are those of hydrogen-like atoms with the meson mass substituted for the electron mass. Because of the finiteness of the nuclear mass, the motion of the nucleus has much greater influence on the position of the energy levels of a meslc atom than on the levels of an ordinary atom. It is therefore necessary to use the reduced meson mass in more accurate calculations. The
wave functions
a
|
of a meson change appreciably inside the nucleus; therefore, the quantity can be replaced by its value at the origin only in a comparatively crude qualitative estimate, as in the derivation of (21,14). Answer. In the general case
]/
|
~Tb + '2~[tP + & + {j

1
'>
*i
~i
'
\
y>\
where
M
For small b and
Al
-*>
co, this expression
may
be obtained from Eq.
Chapter 22
The Electron-Positron Vacuum and the

Electromagnetic Vacuum
iIRAC
THEORY OF "HOLES." DISCOVERY OF THE POSITRON

effects and relativistic effects,
The Dirac theory, which includes spin

account
for
the
fine
was able
to
structure of the spectral lines of hydrogen-like atoms and the anomalous and normal Zeeman effects. The Dirac theory, however, also gave rise to a number of major difficulties in connection with the interpretation of negative energy states. These difficulties were not overcome for some time, but they eventually led to fundamental new discoveries In relativistic quantum mechanics. In our treatment of the motion of a free particle (Chapter 17), we mentioned that the Dirac equation allowed solutions corresponding to both positive and negative values of energy. It is worth noting in this connection that solutions with negative energy are not characteristic of the Dirac theory alone, but appear in any relativistic theory. In relativistic mechanics, the energy of a free particle is connected with the momentum and rest mass by the well-known expression
which has two roots
regions of positive and negative energies are separated by an interval equal to (see Fig. 22,1). At first glance, states with negative energy do not appear to have a real physical meaning, since the region of negative energies extends to infinity co) and, therefore, there is no lowest ( state. This would imply that no ordinary state is stable, since a spontaneous transition to a lower energy state would always be possible, Furthermore, a particle with negative mass (negative energy) would have a number of strange properties: for example, it would repel a particle with positive mass. In classical physics, states with negative energy do not cause any difficulties, because the energy of a moving particle can only change
The
2m
continuously; therefore, transitions from the states with positive energy to the states with negative energy cannot take place, since the
energy would have to change dlscontinuously by the amount A^2//ioC 2, Defining the energy to be positive from the start, we may, therefore,
neglect the negative energy states.
Fig. 22.1. Allowed energy levels of a free Dirac particle
The situation is quite different in quantum theory, where transitions can take place between the states in a discrete spectrum, as well as in a continuous spectrum. States with negative energy cannot be excluded simply by defining the energy to be positive
348
at the initial time, because the probability of the transition between the states with W e 2 is not equal to zero. energy -f m c* and In order to avoid transitions of electrons to negative states, Dirac suggested (1931) that we regard all negative energy Jevels as occupied by electrons (see Fig. 22.2), so that an electron with a positive energy cannot negative energy level under ordinary Jump into a " 1 The state in which all negative _. ._" ". conditions. energy levels and no positive energy levels are _ occupied is called the "electron vacuum," Q c* Let us now assume that a 7-ray photon with 2 2 2m c energy *> 2/HoC excites an electron from the electron vacuum into a positive energy state. In 2 Q this case, the absorbed 'Y-ray photon will be "_ replaced by an electron with positive energy, and *' a "hole" will appear In one of the negative energy ;
'
"
'
"
~
'
.;
a_
Fig 22 2
K ram
of
Zero-point energy diathe electron-positron
varuum um
the "zero-point energy" E v ac (the the energies of the electrons in the
states (see Fig. 22.3). The decisive factor that led to the success of Dirac's hypothesis was that he interpreted the "hole " a * * particle (the "positron") with positive mass ecl ua^ to t*ie ma ss of the electron, but with the opposite charge. 2 Let us suppose that there are no particles at the initial time. Then energy of the electron vacuum) is equal to the sum of negative energy states /*_
The "zero-point charge"
is
equal to
Thus, from the standpoint of the hole theory, the absence of any real particles means that all positive energy states are empty, and all negative energy states are occupied. This case corresponds to the electron vacuum (see Fig. 22.2). When an electron Jumps from a negative energy state //_ to a positive energy state w + , the total energy change of the system is
(22.2)
This change represents the sum of the positive energies of two nascent particles. 3 Similar arguments with regard to charge show that the charge of the nascent particle corresponding to the "hole" is opposite to the electron charge
(22.3)
In accordance with the Pauli exclusion principle (see Chapter can occupy each state
2
6),
only one electron
A similar conclusion was reached in the treatment of the hole theory of semiconductors (see the discussion of the band theory of conduction in Chapter 6)
The prime attached
all states r'_
to the
except the state n'_
summation sign (^ - H_
means
that the
summation extends over
ELECTRON-POSITRON AND ELECTROMAGNETIC VACUUMS
349
4 This particle, which was predicted by energy. Dirac, was called a "positron" and was discovered by Anderson in cosmic radiation (1932). Once it is interpreted in this way, the Dirac theory describes in a natural manner both the electron and the positron. The positron is an antiparticle; its wave function satisfies the Dirac equation with positive energy and positive charge
Thus, the transition of an electron from a negative energy state to a positive energy state (as a result of the absorption of a 7-ray quantum with energy greater than 2/ f M ) leads to the creation of a pair of particles. The unoccupied negative energy state ("hole") e and positive may be regarded as a state occupied by a particle with positive charge
(see Problem 18.5). In the Dirac theory,
pair annihilation
reverse of pair creation is also allowed. This process takes place when an electron with positive energy jumps into a hole. In this case the electron and positron are conFig. 22.3. Formation of an electronverted into 7*-rays. positron pair In these transformations, the laws of conservation of energy and momentum are rigorously obeyed. As already mentioned, pair creation due to the absorption of a 7-ray photon can occur only in the presence of a third particle (for example, a nucleus), that takes up the excess momentum of the photon
is the
process which
Ze
(22.4)
Similarly, the conversion of an electron-positron pair into T-rays takes place in accordance with the laws of conservation of energy and momentum; as a result of pair annihilation, at least two 'Y-ray photons are created
*,+*__
27.
(22.5)
In order to show this, we may choose, without the loss of generality, a coordinate in which the electron and positron move with opposite momenta, so that k _
system
fc
~k
(the center-of-mass system). Then, according to the law of conservation of momentum, the total momentum of the two photons which are formed as a result of annihilation must also equal zero
(ki
+ k) =-
0.
(22. 5a)
Using the law of conservation of energy

cH
(ki
4- k 2 )
~2
J/OTiJic
-f
c-'
W,
is
(22.5b)
we find that & t ==
j,
and the energy of each of the photons

e
equal to
s=
cfiki
cfc/Ja
=
.
fmlc*
c*1l*k*.
The lowest value of the photon energy is obtained when k (that is, when the electron a and positron are at rest). Then t mln ; c These two 7-ray photons move apart with the same energy and oppositely directed momenta. It is easy to see that the electron0), since the laws of positron pair cannot be converted into a single 7-ray photon (& conservation of energy and momentum cannot be satisfied simultaneously with only one y -ray photon.
With the help of quantum field theory, we can construct a theory of the electronpositron vacuum that is symmetric with respect to charge. However, even with the theory described above, which is asymmetric with respect to electrons and positrons (an electron
is a particle, whereas a positron is a "hole") we can give phenomena involving the transformation of particles
a clear explanation of
many
350
The law of conservation of total angular momentum (orbital plus spin) is also very Important in the annihilation processes. If an electron moves with a nonrelativistic velocity, this law (as shown in Chapter 18) can be resolved into a law of conservation of orbital angular momentum and a law of conservation of spin. The law of conservation of spin can be observed particularly clearly in the annihilation of positromum (a hydrogen atom in which the proton is replaced by a positron, or more exactly, a system in which an electron and a positron rotate about their common center of massX In this atom, the nucleus (that is, the positron) and the electron have the
same mass, and

ium atom
therefore the reduced
mass
is
-y-.
The energy
of levels in the positron-
will, therefore,
be one half as large as In the hydrogen atom
Ln ~~
RK
2/i 5 "'
while the radius of the orbit will be twice as large. The velocities of the electron and positron may be regarded as nonrelativistic, Just as in the hydrogen atom. If the spins of the electron and positron are antiparallel (parapositronium), positronium can decay into two %-ray photons (the corresponding mean life Is I 25 10~ 10 sec). The total spin of parapositronium is equal to zero, and therefore the two photons move apart with opposite directions of spin (that is, their total spin is zeroX If, however, the spins of the electron and positron are parallel (orthopositronium) the 7 system must decay into three 7-ray photons (the corresponding mean life is 1.4 10~ sec). The spin of orthopositronium is equal to unity. Orthopositronium cannot decay into one >-ray photon with spin 1, because in this case the law of conservation of momentum would be violated. It cannot decay into two 7 -ray photons because then the law of conservation of spin would be violated (the total spin of two 7-ray photons Is either two or zero). Only if orthopositronium decays into three 7-ray photons will the law of conservation of momentum and the law of conservation of spin be satisfied. The discovery of the positron opened a new stage in the study of elementary particles. This discovery showed that particle shad a new fundamental property interconvertibility and confirmed the existence of antlparticles. We can regard the creation of a positron as the conversion of a 7-ray photon into an electron-positron pair, and the annihilation of an electron-positron pair as the conversion of an electron-positron pair into 7-ray photons.
-
B.
THE LAMB SHIFT OF ENERGY LEVELS OF ATOMIC ELECTRONS
in an atom, it interacts with the electromagnetic vacuum, as with the atomic nucleus and the electron-positron vacuum. The interaction between the electron and the electromagnetic vacuum exerts a particularly strong influence on the motion of the electron in the atom, and it explains the shift of the 2s
When an electron moves

well
as
1/a
level
to the 2/; 1/4> level (in the hydrogen atom). theory of this phenomenon can be constructed only by means of quantum electrodynamics, which is based on the theory of second quantization. But even without referring to this theory we can still obtain the appropriate equations with accuracy up to coefficients of the order of unity, while using comparatively simple physical arguments. One of the basic ideas of the quantum field theory is that each wave or field can be associated with a particle, 'thus, for example, the Dlrac ty waves correspond to electrons and positrons, and light waves correspond to photons. It is very well known, however, that Maxwell's equations describe not only light waves, but also electrostatic and magnetic fields, which depend on charges and their velocities (an accelerating charge produces electromagnetic or light wavesX The electrostatic and magnetic fields can be associated with "pseudo-photons," which have observable effects only in the presence of charges. An electrostatic field can be expanded in a Fourier series, that is, schematically represented as a set of oscillators with different frequencies. An analogous expansion holds
upwards relative
A complete
for the
vacuum
field of
"pseudo-photons"
P.P.
=2E
(0>)
C S
w ''

Since the rest mass of a "pseudo-photon" is zero, we frequency w and the wave number k as follows:
351
its
may
write the relation between
/**+*; +*t
The components k x k v k z are related Chapter 6) by the expressions
,
(22 - 7)
to the Integers n h n 3
n t and the period L (see
kx
= -j
and so forth,
where
= it
1,
2,
3 ..... Hence
so that
or
<o
dw
-?r-*
If
8rc 2 __ TT-.
the
system
is
spherically symmetric, this relation
may
be written as
(22.8)
With the aid of Eq. (22.6), we find that the energy of the electrostatic field inside a region
of
volume
L* is
p.p.
=i
()2 *'*
=
8
(to))2 -
(22 - 9)
In deriving this expression,
we have used
cos
tot
the relation
cos
"'Oav
TT
z
&<w
Just as in the theory of the harmonic oscillator, the energy of the field in the lowest energy state is not equal to zero, but instead is equal to the sum of the zero-point energies of the harmonic oscillator (see Chapter 10)
The coefficient 2 in front of the summation sign takes into account the fact that each harmonic corresponds to two different states of polarization of the "pseudo-photons." In the state where there are no real photons the total energy of an electromagnetic field (the vacuum), must be equal to this zero-point energy (e p^ = o). Hence, taking
(22.8) into account,
we
find
E(W=*H=**.-y.
(22.K)a)
In the absence of real particles and external fields, the vacuum (including the electromagnetic one containing photons and pseudo-photons) does not, as a rule, have any because it is isotropic. On the other hand, when real particles observable effects
352
and external fields are present, the Isotropy is disrupted and virtual particles ("pseudophotons" or electron-positron pairs) are created and subsequently annihilated (vacuum
fluctuations).
A simple physical picture of a few basic notions involved in the theory of the vacuum was obtained somewhat unrlgorously by Welton by means of a semiclassical, nonrelativistic treatment of the motion of an electron, taking into account its interaction with the vacuum fluctuations. As an illustration of Welton's calculations, let us attempt to give a more concrete meaning to the zero-point fluctuations (22. 10JL In a rough approximation we
can use the ordinary classical equation
morÊp.p.
to
find
(22.11)
field of
how an electron
p
will be affected by the
vacuum
"pseudo-photons." Ex-
panding
- in a Fourier series
p.p.
2
U)
(<l>)coso> '
(22- 12)
and integrating Eq. (22.11), we find the displacement of the electron positron due
to the
vacuum
field:
^The mean-square deviation

of the position is given by
(SR)i
(22 - 13 >
= (7F-
(r)'
- >!
since
and therefore r
Substituting (22.10a),
we obtain a divergent
integral
==
l ,/_
n
tic
_}Y'<" m c I J u>
* / \ (22.15)
from which a finite (observable) part can be separated out if the range of variation of the frequency is cut off from above at a frequency corresponding to the rest energy of the
electron
and from below at a frequency corresponding to the an atom 5

4
minimum energy
of the electron in
]/:j
"*in=V
m * = aB?'
n
<
22- 17 >
Substitution of (22.17) and (22.16) into (22.15) gives
^.
2
( 22.18)
The
limits of the range of variation of
<y
are specified more accurately in renormali-
zation theory
ELECTRON- POSITRON AND ELECTROMAGNETIC VACUUMS
353
Thus it can be seen that the vacuum field of "pseudo-photons" will cause the electron perform a motion somewhat resembling Brownian motion with a definite value of the mean square displacement. It is well known that the Brownian motion of a particle is due to its collisions with randomly moving molecules of the surrounding medium. In similar fashion, an electron undergoes "collisions" (of a special kind) with the assembly of virtual particles forming the vacuum. As was shown by N. Bogolyubov and S. Tyablikov, the vacuum fluctuations cause a certain "smearing out" of a point electron. As a result, the electronic radius turns out to be the geometric mean of the classical radius and the Compton wavelength
to
TT^ m
a
y~JL m c
'
(22.19)
The existence of this effective radius should have several consequences. In particular, the interaction between the electron and the nuclear charge should be changed, and this, in turn, should lead to an additional coupling energy and thus to a shift of the energy levels. The usual expression for the potential energy of an electron in the field of a
nucleus
(22.20)
is
replaced by
V = - efl(r+ Ar) = ~ e
[1
+ (Ary) + \ (^v)
+...]* (r).
(22.21)
Changing to the average values and using the relations
V,
we
obtain
(22.22)
+
The
...
}*(r).
(22.23)
additional energy of interaction between the electron and the nucleus is given by
'<'>
< 22-
24 >
since the Coulomb potential of the hydrogen nucleus satisfies the Pols son equation
(22.25)
obtain an expression for the shift of energy levels in the hydrogen atom, to average the additional interaction energy over the corresponding state
To
it is
necessary
Wp. p.=
(V
- V)
(r)
I*
d*x
c*a
(0)
In
(22.26)
ITiis shift occurs only for s states, since the quantity 1 (0) a in the approximation under consideration vanishes for other states (/= 1. 2, 3 t ...), whereas for the s state
1
|
(22.27)
2 where a Q ft /MO?? is the radius of the first Bohr orbit of the hydrogen atom, and n is the principal quantum number. If we now substitute this value into (22.26), we obtain a formula for the s -level shift
Ap.p
=3 ^'f in?.
(22.28)
This formula was first derived by Bethe.
354

substitution of numerical values for the 2s state (n
=
.
2)
gives
(22.29)
A
p-
= p
^
7,8/? = 1040 Me
1057.77 Me) for This is in fairly good agreement with recent experimental data (A the Lamb shift (see Chapter 20 X A complete study of the shift of energy levels of atomic electrons with the help of relatlvistic quantum field theory gives considerably better agreement between theoretical and experimental results than the semiclasslcal formula (22.28), The discrepancy is reduced to less than 1 Me. For the sake of brevity, we shall not discuss the modern theory of the vacuum in greater detail, and we shall confine ourselves to a mere enumeration of the main results
that this theory has yielded.
C.
ELECTRON-POSITRON VACUUM
Lamb
shift
Equation (22.28) for the
was obtained from a calculation
of the interaction
between electrons and the electromagnetic vacuum. In addition to the electromagnetic vacuum, there exists an electron-positron vacuum and also vacuum states of other particles. The method of second quantization (which is in some measure applicable to all fields) can be used to calculate the influence of the electron- positron vacuum. In modern quantum field theory, the study of the properties of the vacuum states of different particles plays a particularly important role. The vacuum gives rise to interactions between particles. In particular, the electromagnetic interaction of two electrons (Coulomb's law) may be regarded as an interaction which takes place through the electromagnetic vacuum, with one electron emitting a virtual photon and the other electron
absorbing
it.
On
the
other hand, the
vacuum represents
a sort of "reservoir,"
from which real
particles are "drawn" when they are created, and to which they "return" when they are annihilated. We have already come across the electron-positron vacuum as a "collection" of electrons in negative energy states. Unfortunately, it has no classical analog and therefore we cannot use a semiclassical analysis, as was possible in the case of the electromagnetic vacuum. The Coulomb field of a nucleus can polarize the electronpositron vacuum (so that an electron will behave as if it were in a dielectric), giving rise to an additional interaction
A
t
r).
(22.30)
Comparing (22.30) with (22.24), we see that the level shift due to interaction with the electron-positron vacuum is about 1/40 times as large as the level shift due to the fluctuations of the electromagnetic field, and is opposite in sign. The electron-positron vacuum exerts a particularly strong influence on the magnetic properties of the electron. It was shown by Schwinger that the magnetic moment of the electron becomes somewhat larger than the Bohr magnetron
(22.31)
The change
into account,
in the magnetic
moment
of the electron, with the secondp-order
term taken
AH e- _ p ==
is in
0.328
(g;
ji
0.001 1596
(&
(22.32)
good agreement with experimental data obtained with the aid of microwave spectroscoplc methods (see Chapter 21X Qualitatively, the appearance of the additional magnetic moment of the electron and the sign of the moment can be explained as follows. The initial electron A (see Fig. 22.4), whose spin is directed upwards (we specify in this manner a preferred direction), creates a virtual pair electron A and positron 5' which generally
}
355
have opposite spin directions. Since we have a preferred direction (determined by the can be directed in two ways: spin of electron A) the spins of electron A' and positron either the spin of A will be antiparallel to the spin of A' and parallel to the spin of B and antiparallel (case 1 in Fig. 22.4), or the spin of A will be parallel to the spin of
'
to the spin of (case 2 in Fig. 22.4). In determining the additional magnetic moment, it is necessary to consider the following possibilities of pair creation and annihilation: B' is created and then annihilated. (a) Pair This is possible in case 1 and case 2. No correction to the magnetic moment should arise, because the probability of the creation process is the same in both cases, and the magnetic moments have opposite signs in the two cases. (b) Since .4 and B* have opposite spins in case 2, positron B may be annihilated together with A\ as well as with the initial electron A. In case 1, however, this process is less probable, since the spins of pair A and /?' are
'
if.
Antiparallel spins of electrons
2 Parallel
parallel to one another. This process gives a preferred state in which the additional magnetic moment is parallel to the initial magnetic moment of electron A, since, as can be easily seen, the magnetic moment of all three particles (A, A' and B ) is directed downwards. Consequently, there arises an additional magnetic moment of the electron, equal to
l
A and
/I
spins of electrons A and A

1
Fig. 22.4 Virtual creation of an electron-positron pair by an electron The spins of the real and virtual particles
are depicted by solid and
dashed
ar-
t==
AH-e.-p.
2fVf,
rows, respectively
a numerical factor which determines the probability that the initial electron A will be annihilated together with positron B". These simple considerations give the correct sign of this anomalous magnetic moment. More rigorous calculations led to Eq. (22.32). In addition to the Dirac magnetic moment, ji=j~--fi 3, an electron will have an
f
is
where
anomalous magnetic moment
ji
,>-!
whlch arises owing
to the interaction
between the electron and the electron-positron vacuum. The anomalous magnetic moment of nucleons (neutrons and protons) can be explained in meson theory. The accuracy of the results, however, is much less striking than in calculations of the anomalous magnetic moment of the electron. This is due to the fact that meson theory is still in a much less satisfactory state than quantum electrodynamics. According to the meson theory, protons and neutrons interact with the ir-meson field. Since the proton can dissociate into a neutron and a positively charged TT meson, and the neutron into a proton and a negatively charged IT meson, the anomalous magnetic moments of protons and neutrons due to the ir-meson field must be approximately equal in magnitude and opposite in sign.
The appearance of the additional magnetic moment of a neutron is explained as follows. The neutron has a definite probability of dissociating into a proton and a ir meson. Since the spin of a TT meson equals 0, it does not possess an intrinsic magnetic moment. Therefore, there will be a contribution to the magnetic moment of the neutron from a pion which is, for example, in a p state (we recall that the orbital angular momentum In the
s
state is also equal to 0).
The magnetic moment
of a
v meson
is
times greater
than the nuclear magneton, and therefore the ir meson produces the main contribution to the anomalous magnetic moment of the neutron. For the virtual process to fulfill the law of conservation of angular momentum, it is necessary that the direction of the orbital angular momentum (equal to 1) of the virtual Trmeson should coincide with the direction of spin of the neutron, while the spin of the virtual proton (equal to Va, Just like the spin of the neutron) should be directed opposite to the spin of the neutron so that
the w meson has a negative charge and its s s/t s 3 1- Since Sfl Sn pt p angular momentum is parallel to the spin of the neutron, it gives rise to the negative magnetic moment of the neutron.
=,
356
RELjmVISTIC QUANTUM MECHANICS
In order to explain the magnitude of the anomalous magnetic moment of the neutron, one must assume that the neutron spends / of its time in its dissociated state. This estimate is perfectly reasonable and gives the correct sign and order of magnitude of the n 2p. nuc ). Similarly, it can be shown that the dismagnetic moment of the neutron (|x
!
sociation of the proton into a neutron and a n + meson gives rise to a positive anomalous n -}- 2u nuc ). Moreover, the proton has a Dirac magmagnetic moment of the proton (\>* data on the magnitude of the magnetic moment More accurate netic moment (M-jp l^nucX
of the proton and neutron are given in the preceding chapter. Thus we have a basis for regarding the anomalous magnetic moment of Dirac particles as a secondary effect which can be explained on the basis of the field theory. This moment does not appear in the initial equations, but arises as a result of the interaction either between the electric charge and the electron-positron field (in the case of electrons),
or between the nuclear charge and the 7r-meson field
(in the
case of protons and neutrons).
D.
RENORMALIZATION
One of the most important sections of modern quantum field theory is concerned with the question of renormalization, This subject is not yet in a mathematically satisfactory state, but a number of important results have been obtained. The basic idea involved In renormalization is the separation of finite, observable terms from the divergent terms describing the interaction between an electron and the
electromagnetic or electron-positron vacuum. In effect, the question of renormalization first arose in classical electrodynamics for example, in the theory of the electromagnetic mass of the electron. If we assume that the electron charge is distributed inside a sphere of radius r its classical electromagnetic mass will be equal to
,
m field
where
= 7 -^ r
C
(22.33)
7 is a factor of the order of unity which depends on the charge distribution inside sphere. The attempts to construct a classical electrodynamics in which a finite radius r 10 n cm would give a reasonable value for wfield (the Lorentz theory, the nonlinear Bom-lnfeld theory, the Boppe-Podolsky theory with higher derivatives) did not give any satisfactory results. Moreover, all these theories gave rise to fundamental difficulties in connection with quantization. On the other hand, the theory of a point electron leads to an Infinite value of the mass as r both in the classical and quantum forms of the theory. It was therefore a major achievement of renormalization in modern quantum field theory that It was able to separate, from the infinite interaction energy, finite terms associated with the Lamb
the
atomic levels and the additional magnetic moment of the electron. Modern renormalization theory extends to the problem of the self-mass and charge of the electron. It has been found that when the electron-positron vacuum and the electromagnetic vacuum are taken into account, the electromagnetic mass (22.33) drops out completely. Therefore, the main mass of the electron (the so-called "bare" mass) will not be associated with the electromagnetic field. The interaction with the vacuum (in this case the main contribution is obtained when the electromagnetic vacuum and the electronpositron vacuum are taken into account simultaneously) yields an additional mass which
shift of the
diverges only logarithmically
Renormalization is faced with the important question of finding the value c max at which the logarithmically divergent expression should be cut off. This problem is a theoretically important one, although in practice the logarithmically divergent terms may be regarded to be of the order of unity, and therefore the quantity Am will remain of the order m bare /1 37 over a comparatively wide range of values max.
ELECTRON- POSITRON AND ELECTROMAGNETIC VACUUMS

In exactly
357
vacuum
the same way, the interaction of the electron with the electron-positron (rather than with the electromagnetic vacuum) leads to a decrease of its "bare
charge" by the amount
bare
c8
' v (22.35)
field corrections to the mass and charge of the electron have not yet been experi6 mentally separated, and this important subject requires further investigation.
The
the emission of photons take place for the free motion of an with index of refraction n (the Cherenkov effect)? Why does this emission become impossible in a vacuum (n-=z 1)? Solution. In order for the emission to be possible, it is necessary that the laws of
When can
conservation of energy
and
momentum
be satisfied, where
the photon
p and p are momentum.

1
the initial and final
moments
of the electron
and nkis
Squaring these relations and then subtracting one from the other, ing expression for the cosine of the angle of photon emission:
we
obtain the follow-
Using the relation for the index of refraction n of the medium
==== cjfe = ?*? n

*
is the
phase velocity of light in the medium), we obtain
1 Radiation can take place when p/i (that is, when the electron velocity v remains less than the phase velocity of light in vacuum, but becomes greater than the velocity of light in the medium c ^>v ^> c ). In the classical case (ft 0), the angle at which the photon is emitted satisfies the relation
f
1
>
The emission
cos
of
radiation
in
vacuum
is
impossible because for n
=
c.
we have
and the velocity v cannot become greater than the velocity of light
Quantum Theory of Fidd*, New York- Interscience, 1949, A I. Element s of Quantum Electrodynamics (trans ), London: Oldboume Press, 1962, N N Bogolyubov and D V Shirkov, Introduction to the Theory of
See
Wentzel,
Akhiyezer and
B. Berestetskiy,
Quantized Fields (trans
),
New
York: Interscience, 1959
Chapter 23
Theory of the Helium Atom Neglecting

Spin States
A, BASIC PRINCIPLES
>
OF THE THEORY OF MULTIELECTRON ATOMS

is the
simplest multielectron atom; it conmoving about a nucleus of charge Z 2. In spite of its simplicity, this system exhibits several important features characteristic of the many-body problem in quantum
sists of two electrons
The helium atom
mechanics.
classical theory two electrons can always be identified by subscripts 1 and 2, and the motion of each electron can be followed separately. According to quantum theory two electrons can, in practice, be distinguished from each other only when the distance between them is large. If electrons 1 and 2 are so close to each other that there are points in space where their wave functions are both simultaneously different from zero, then, since electrons are identical particles, we shall be unable to distinguish whether an electron occurring at the point is electron 1
In
or
2.
This indistinguishability or identity of particles is a special feature of quantum theory. It gives rise to the so-called exchange forces, which have no classical analog. In multielectron atoms, spin properties are very important; these properties are neglected in the classical and Bohr theories and are taken into account only in quantum mechanics. In fact, only in the case of a one-electron atom can the spin corrections be neglected in the first approximation. This explains why the predictions of the Bohr theory are applicable only to hydrogen-like
atoms. The Bohr theory could not be extended to atoms with two or more electrons, since it could not account for either the exchange forces or the spin states. In order to investigate the basic features of the quantum theory of the many-body problem, we shall consider in some detail the problem of helium- like atoms (for example, neutral helium, singly ionized Li*, doubly ionized Be ++ and so on).
,
THEORY OF THE HELIUM ATOM NEGLECTING SPIN STATES

B.-
359
SOLUTION OF THE PROBLEM OF THE HELIUM ATOM

VIA
METHODS OF PERTURBATION THEORY
start with, let us determine the physical nature of the exchange forces which are related to the indistinguishability of electrons. We shall neglect the spin properties of the particles. We shall assume that the position of the first and second electrons is given by the position vectors r\ and r4 respectively (their origin coincides with the stationary nucleus; see Fig. 23 V 1). For the sake of brevity, states with quantum numbers (HI, l lt MI) and (n* / 2 w ) will
1 ,
f
,
To
be denoted by, respectively, n^ and m (in this notation, therefore, n stands for the whole set of quantum numbers). If there were no interaction between the electrons, we could determine the motion of each electron separately from the Schrodinger
equation
Fig
2.3
The helium atom
(23.1)
where
(23.2)
and the subscript / takes two values: /=! in the case of the first in the case of the second electron* From these electron, and / equations we can obtain the energy values E nj (see Chapter 13)
(23.3)
will be the same as the wave functions of a nj hydrogen-like atom; they will satisfy the orthonormality condition
The eigenfunctions
=^
If
(23.4)
we now take
into account the interaction of the
two electrons
V'=We
(23.5)
in the form of a product of
are allowed to do this because in this approximation the problem has a solution two functions, depending on the spatial and spin coordinates,
respectively (see Chapter 24).
360
their
motion can no longer be regarded as independent. The Hamiltonian of the complete system will be
H=H
in the
+ H + V =H + K'
f
(23.6)
and to describe this system we must take the Schrbdinger equation
form
r,)
0,
(23.7)
where E is the total energy, and <]>(/*!, ra ) is the total wave function, which depends on the coordinates of both electrons. The quantity r i) characterizes the probability of finding the first 'I^to. ^)'M r
i
electron at position r\ and the second electron at position r2 . Therefore, the normalization condition for tyfa, r*) takes the form
t*(rlf
r,Wi. r,)d^d^=l
(23.8)
where the integration extends over the coordinates of both particles.

Since the exact solution of Eq. (23.7) entails insurmountable we shall use the perturbation theory 3 developed in Chapter 14, assuming that the mutual interaction of the electrons causes only a small change in their individual motions in the Coulomb field of the nucleus (the justification for this approximation will be examined in greater detail later). We shall first consider the zeroth approximation, in which the perturbation energy V can be neglected. The Schrbdinger equation
difficulties,
(23.7)
becomes
(&-W)p(r
Since the
l9
r,)
0.
(23.9)
sum
zero-order Hamiltonian H can be resolved into a H 4 each of which depends only on H a single variable (either r or ra ), it is obvious that in the zeroth approximation the wave function may be written in the form #)
of two Hamiltonians
t
(i|)
r4 ).
Substituting (23.10) into (23.9) and using (23.1),
(23.10)
we
get
- 11) K = {* - (Hi + H
4 )}
^
The problem of the helium-like atom
is a
<k (rO 4m, (/) *i (/-i)Hrf (r4 )}

a
= =
(r,)
f^. (r
9 )}
three-body problem and cannot be solved
exactly, even in the classical approximation.

Hence, the energy
in the
36 1
zeroth approximation
=E +
Wl
(23.11)
where E ni and
* 2 are the energies of the two electrons on the asno of mutual interaction. This result can be explained as sumption follows. In the absence of the perturbation V\ the motion of the electrons is determined by their interaction with the nucleus ZeQ ; that is, their motion is completely described by the Schrbdinger
equation (23.1) which has the eigenvalues E nj [see (23.3)] and eigenfunctions 4> ; . Since one of the electrons is in a state n\ and the second in a state /z*, the total energy of the system is Eni -\- E n ^
when
V'
= Q.
Since the two electrons move independently of one another, the total wave function will be the product of the corresponding two independent one-electron wave functions obtained in the oneelectron problem. By direct substitution into Eq. (23.9) it is, however, easy to show that, in addition to the first solution (23.10), there will be a second solution (fy* v) corresponding to the same
energy eigenvalue
(23.11):
^^(r,)*,,, to).
(23.12)
This solution differs from u by a permutation of the electrons. The first electron is now in state n* and the second electron is in
state HI.
which
If
This state of the system has, therefore, an additional degeneracy is due entirely to the indistinguishability of the electrons;
this is
known as exchange degeneracy. both electrons are in identical states (wi=/z a ), the wave functions u and v are identical and there will be no exchange degeneracy since
u
=v=
<(>,
(r,)
^(r
t ).
(23.12a)
In the case MI ^ /i* , however, functions u and v are different, and therefore the following linear combination should be taken as the general solution fy of the Schrodinger equation (23.9):
f = C u + CiV
t
(23.13)
where C
and C2 are arbitrary constant coefficients, which are v related only by the normalization condition
In order to find the values of the coefficients Q and C 2 and the energy E of the perturbed system (the system with V taken into
362
account), according to the perturbation theory, in the form: the solutions of E and
'!>
we
shall look for
(23.14)
We use a and write
first
it
approximation of the Schrodinger equation
(23.7),
in the
form
H)
<[/
= -.('
I/')
(C!
+C
2 iF).
(23.15)
From Chapter 14, we recall that the solution of the homogeneous equation for the unperturbed problem must be orthogonal to the right-hand side of the inhomogeneous equation for the first-order approximation of the wave function [see (14.13)]. Since the functions u and v are solutions of the unperturbed problem, we have
u
*
('
V) (Citi
+ C&) d
il
*i
d^t
= =
0,
(23. 16)
V* (E'
V)
(C,
+C
3
2 u) rf
3 rf A:a
0.
(23.17)
[see
In (23.17), let us substitute r* for r, and /*i for r l9 The function (23.12)] then becomes u [see (23.10)], and the function v
//
,
becomes
since
|r,
while the perturbation energy V remains unchanged ra r \. Thus Eq. (23.17) takes the form |r,,
|
>
('
K')
(C 4
+C
c/^ 2
c/U'i
=-
0.
(23. 17a)
is, therefore, sufficient to consider only Eq. (23.16) since the results can be extended to Eq. (23.17a) by substituting C t -> C 2 and
It
C 4 ~>C
t .
us substitute into Eq. (23.16) the explicit expressions (23.10) and (23.12) for u and v, and introduce the notation
(23.18)
(23.19)
Let
(23.20)
(23.21)
Here Pnfo) and characterize the probability density distribution of electrons in the states 'h and n*, respectively, whereas
PW

p lt
363
(rt )
and
pî (/*)
exchange) states, when each electron is partially instate partially in state n 8 .
characterize the so-called density of mixed (or rc, and
The orthonormality condition gives

u* u
J
d*Xt d*x*
=J
pn
and
*y
d 3 *! d3 x2
=j
Pli
(rO d *!
8 M (rO d *
0,
and therefore we can reduce (23.16) to the form
EC,
- [Cfl
Jil^M(pL
^^+
(23 - 22)
The
first integral in Eq. (23.22) is
simply the Coulomb interaction
of two
"smeared-out" electrons
u =e A
The second
/,s
f \
Pn (ri) pas (ra) " jt v " Y ^2^1 .y _^ri
/OQ O o\ (Zo.ZoJ
integral gives the so-called exchange energy

t
A = el
"7, -'
d '^
***
(23.24)
corresponding to the interaction of two electrons when each is in mixed state n\ and n a . Unlike the Coulomb energy K, the exchange energy A has no classical analog; it is essentially a quantum- mechanical concept. Using Eqs. (23.23) and (23.24), we obtain instead of Eq. (23.22)
a the following expression:
C,(E'
K)
C /1=0.
2
(23.25)
From Eq. (23.25), we obtain the equation corresponding to (23.17a) by substituting C t -> C and C\ -* C2
t
:
C ('
2
/<)
dd
0.
(23.26)
Those densities have no classical analog
364
RELjVTIVISTIC
QUANTUM MECHANICS
Equations (23.25) and (23.26) give

1)
E'
= K + A,
C,
=C
=
2>
(23.27)
and
2)
e=K
A, Ci
2.
(23.28)
Accordingly, we have two solutions for the wave function [see (23.13)] and for the total energy
1)
symmetric
^ = C,(-{-iO,
>
(23.29)
(23.30)
= + /(-}- 4,
and
4
2)
antisymmetric
V EA
= Ci(u =F+K
l
<]>
v),
(23.31) (23.32)
t
A.
vVe can determine the coefficient C from the normalizatior condition of the wave functions s and y1
:
Hence, 2Cj
or
Ci
= ---
Thus for
s
<]>
and
we
finally
have
t=~(tt +
l
0),
(23.29a)
*'
=^ (-*)-
(23.31a)
When
the functions
both electrons are in the same quantum state (MI /;), // and v are identical. In this case Eqs. (23.16) and
(23.17) reduce to the
same equation
0.
(23.33)
From
this
it is
readily seen that
F = ff,
s
(23.34)
We recall that under a permutation of coordinates (that is, when r^ and r 2 are inter* does not changed) the function u and u transform into one another The wave function change sign as a result of this operation (symmetric function), whereas the function ^ reverses its sign (antisymmetric function)

and thus the exchange energy disappears in this case. For the function we obtain a unique symmetric solution
' i),
i
365
wave
(23.35)
with the corresponding energy of the system

(23.36)
this section, we can make the folof the statement: perturbation method to the probApplication lowing lem of the helium atom 5 leads to two types of solutions a symmetric or antisymmetric solution. This is in complete agreement with the general theory of systems of identical particles (see below).
Summarizing the results of
C.
-COULOMB INTERACTION BETWEEN ELECTRONS
Let us find an expression for the Coulomb energy of two n 1 =l). In this case the electrons in the lowest energy state (n energy and the wave function of each electron are given by
i
<
23 - 3
"
where o j^^r is the radius of the first Bohr orbit. The Coulomb energy of interaction between the two electrons
is
*=
Here
I
*!
(rt )
(r,)
^-i
,
tPxid'x*
(23.38)
vectors
r\
rl r2 and & is the angle between the Yr\ 2/y a cos and r. To integrate (23.38), let us choose the direction
1
of the z axis along the direction of the vector r\ . Substituting the expression (23.37) for the wave functions into Eq. (23.38) and in6 tegrating over the angles, we find
This is due to the fact that the perturbation removes the degeneracy, and therefore the coefficients that were indeterminate in the êroth approximation can now take specific
values (See also Chapter
In integrating
14, the Stark effect
)
over the angle $ (x = cos d),
we used
the relation
(continued)
366
RELATIV15TIC QUANTUM MECHANICS
drt
*
(23.38a)
Next, Integrating over
r,
and
/,,
we
finally obtain
K = 4.M. o a
Since the zero-order energy is
E*
(23.39)
=2 =
1
^
fl
(23.40)
is
the total energy of two electrons in the
ground state
i
.
equal to
B ==
+ = -^- + |z^
/C
(23.41)
Let us now find the ionization energy of the helium atom, that the energy that must be expended in order to remove one electron from the first orbit. For a singly ionized helium atom (a hydrogen-like atom) the bonding energy between the electron and the nucleus is simply E [see (23.37)]. The ionization energy of a helium- like atom is therefore
is,
l
Jon._
-E = ^ (l l
1-
Z)
(23.42)
so that for helium (Z
= 2)
we have
ior
=0.75
r! -.
(23.43)
"o
According
is
to
experimental data, the ionization energy of helium
=0.9 E^S r
The discrepancy between
OQ
^ = 24.48
ev.
(23.43a)
the theoretical value and the experimental
data is a result of the fact that the perturbation energy

is not
|
K=
very small as compared with the zero-order energy E|

ratio
is
=~?
(their
~l/3).
Perturbation theory, therefore, gives us

l r
T\
Considering that tho integrand / i(ri) i//f(/ 2 ) is symmetric with respect to the variable and r 2 we replace (for rj ""* r 2 ) the radius r t by r 2 and vice versa, obtaining
,
for
/j
>
r2
This expression was used
in
evaluating the integral (23 38)
367
only qualitative aspects of the problem. The accuracy of the method have the same order of magnitude. is not very great because K and
D.
THE VARIATIONAL METHOD
The variational method developed by Ritz, Hylleraas and others was first successfully used to find the ground -state energy of
atomic systems, and in recent years ithasbeen applied in collision
theory.
As we know, the average energy of anatomic system
is
given by
(23.44)
If
the
wave function
is
represented as
*
<!*,
(23.44a)
where the coefficients Cn give the probability

in
state n , the average relation [see (7.21)]
of an electron being value of the energy will be given by the
!*
(23.45)
Replacing each eigenvalue E n in the summation by the lowest min and eigenvalue using the fact that for normalized functions
we
find that
in
<
f <i*H']d x;
(23.46)
\
that is, the lowest value of the integral
ty*W>(Px
can be used to
determine the upper limit of the ground- state energy of the system. It was found that the variational method gives very good results when the perturbation energy E' is of the same order as the energy E of the zero-order approximation. The variational method can therefore be used in cases where perturbation theory gives poor results (for example, in calculation of the ground- state energy of the helium atom). When a problem is solved by the variational method, both the additional interaction V and the main interaction are treated equally In the Hamiltonian II of Eq. (23.7). A test function then depends on several parameters and is selected in such a way that the integral can be calculated exactly. The will then be a function of these parameters and it is energy obvious that the minimum value of this function will be close to the true value if the test function resembles the true function.
<j
368
The most difficult part of the method lies in the choice of the best test function. All the available information on the properties of the system must be used in making this choice. It is impossible in the general case to indicate the form of the test function, and therefore it is frequently necessary to rely on physical and mathematical intuition. Very often, a test function is similar in form to the solution of the unperturbed equation. We shall now use the variational method to calculate the groundstate energy of the helium atom. Our procedure will be based on that of Hylleraas (1927). At the end of the discussion we shall compare both perturbation and variational methods. For the test function, Hylleraas chose the ground- state function (23.27) of the hydrogen atom, replacing the charge Z by a certain effective charge 2'. The quantity Z is the unknown parameter which has to be determined from the variational principle. The test function
<
23 - 47 >
normalized to unity, just like (23.37), since its normalization is independent of the value of Z'. The Hamiltonian H in (23.46) must include both the zero-order approximation Hamiltonian H and the perturbation potential energy
will obviously be
V We
.
thus obtain
H = T, + V, + T
where T; and Vj(j=l
t
+ V, + V,
+
(23.48)
2)
are given by (23.2) and the perturbation
potential energy Vis given by (23.5). Since the wave functions are normalized and since T! Vi does not depend on ra and T a -f- V a is independent of the coordinate ri9 the average value of the Hamiltonian is
(23.49)
where
= ST. J
*i
to)
(T
v
')
**
to)
**
<
23 - 5
(ri)2*d*x lt
(23.50a)
V'
=J
ft
(n)
fa)
|n .g ra|
fxiffix*
(23.50b)
369
The
integral (23.50b) agrees exactly with integral (23.38); therefore, Z', we obtain in accordance with (23.39) setting Z
7'
= |.^!L.
(23.51)
The quantity f in Eq. (23.49) represents the average value of the kinetic energy of a hydrogen- like atom with atomic number Z' when the electron is in the lowest state. This value is related to the total energy of the hydrogen- like atom by the well-known exi
pression
T,
1
'
.
<
23 - 52 >
In exactly the same way, replacing Z'in (23.50a) by Z, we can obtain the average value of the potential energy of a hydrogen-like atom, since tt is well known that the potential energy is twice the
total
energy (V
2j ).
Consequently,
V
F
= |j2B = Z
1
/ ^3. a
(23.53)
It follows from (23.49) that the average value of the energy given by the expression
is
E (Z') = 4 (Z'
Ug
2ZZ' -f 4
O Z'),
(23. 54)
which is a function of the parameter Z'. Let us now find the value of the parameter Z' corresponding to the minimum energy of the system. Differentiating E (Z') with respect to Z' and setting the derivative equal to zero, we find
The minimum energy
of electrons in the helium
atom
is
therefore
For the ionization energy we have
In particular, for
helium (Z
2)
we
obtain
5-.
(23.56)
370
This result is considerably closer to the experimental value [see (23.43a)] than (23.43) obtained from the perturbation theory. Hylleraas later improved the agreement with the experiment by mm using several variational parameters. The result (23.55) for has a simple physical interpretation, namely: the interaction between the electrons results in a screening of the positive nuclear charge. The variational method can also be used to find the upper energy limit of one or several excited states. In this case the test function has to be chosen in such a way that it is orthogonal to the wave functions of all the lower states. When the energy levels are it can 1, E* . . .), arranged in order of increasing magnitudes (E is orthogonal to functions <N <N* easily be shown that since .are all equal the corresponding expansion coefficients^, C lt C 2> . to zero. Therefore, in accordance with (23.45), the energy is
, <[
o,
where
the
ft
is
the quantum
a
number
of the given excited state. Using
relation
|C n
=l, we
find that the
minimum
value of this
energy corresponds to the unknown energy E no of the excited state. The application of this method to the calculation of the energies of the highest excited states is rather difficult, because of the necessity of introducing a large number of additional conditions to ensure that the wave function of a given state will be orthogonal to the wave functions of the lower states.
E.
DERIVATION OF THE SCHRODINGER EQUATION BY THE VARIATIONAL METHOD

1
shall consider one of the most general forms of the variational problem, when choice of the test wave function 4 that is used to find the average value of the Hamilton! an
We
the
=
dition
(23.58)
describing the motion of a single particle is restricted only by the normalization con-
ty*fy
d*x
(23.59)
Upon varying E with respect

tain
to
fy
and using the hermiticity of the operator
H f we ob-
8E
(&<|;*H<|;
+ UH*<|/*) d*x =
0.
(23.60)

Here
371
the variations 8^ and ty* cannot be regarded as independent since they are connected by the normalization condition (23.59). These variations can be made independent by varying the condition (23.59)
C ty*<\>d 3
x+
f a<|4* (/*
Multiplying this equation by a constant Lagrangian multiplier X f which is chosen in such a way that the variations are now independent, we add the resulting equation to (23.60). Since the variations ty and H* are now arbitrary, the variational principle gives automatically the Schrodinger equation for fy and ty*:
(H
)<]>
0,
(H*
A
)^*
0,
(23.61)
it is
and the physical meaning energy E (X E).
of the
parameter
becomes clear:
the negative of the
Consequently, the variational principle and the normalization condition together lead to the Schrbdinger equation. It is apparent from the above results that the eigenvalues of the Schrodinger equation (23.61) give the extremals of the variational integral. A more detailed analysis shows that these extremals are minima and that the energy of the ground state corresponds to the absolute minimum the smallest possible energy In order to calculate excited states, it is necessary to impose orthogonality value. conditions on the wave functions of the lower energy state (as mentioned above), whereas orthogonality follows automatically in the Schrbdinger theory.
F.
HARTREE-FOCK METHOD OF SELF-CONSISTENT

FIELDS
We have considered two extreme cases of the variational method. In one case (the Ritz-Hylleraas method) the variation of the wave function amounted to a determination of the "best values" of parameters in a specially chosen wave function. In the other case,
the choice j)f the \yave functions was restricted
onlxbjj^
izattoiTccM^ Schrodinger equation. ~ThereT is also an intermediate case in which the wave function is not specified, but is assumed to be a product of one-electron functions depending on the coordinates only. The specific form of these functions is found by using the method of successive approximations to solve an equation derived on the basis of the variational
principle.
One such method was proposed by Hartree (1928). This method, whose interpretation from the point of view of the variational principle was given later by Fock, may be described as follows.
Let us start with the variational principle for two particles in 7 general form
= f (r
l9
ra )
H'|>
(rlf
r)
d^x, d*x*.
(23,62)
As an additional condition, we require
that the total
wave function
(23.63)
should be a product of one-electron functions

).
This integral can be generalized
in
a similar fashion to the
case of three or more
particles
372
It is

also necessary to allow for the normalization condition
'*i^=l,
which can be written for each particle separately
23 - 64 )
Substituting the test function and 9f we obtain

<|
(23.63) into (23.62) and varying
<|>,
(23.65)
where
Hy
= pJ-j- + V
(ry )
is
the
Hamiltonian
and
-
describing
the
motion of a single electron (/=
1, 2),
represents the inobtain a relation
teraction energy between the electrons. From the normalization condition (23.64) between the variations
we
WtlM* + Wftifc + t?Wi<h + t?W*i
W dM =
8
*.
o.
and Multiplying this equation by the Lagrange multiplier A adding it to (23.65), we can select A so that all the variations 8 ti and so on, will be independent. Therefore, we obtain the
Wi
Hartree equations
l Ii
+
-f
( '
fc'H^d^
$f I
c
I,
(l
+ W -$- h<Px* - fik = r = tf U'h A'i + -^- *td^, -*,

(
'
0,
ria
p*
0.
functions. over the
Analogous equations can also be obtained for the complex conjugate We multiply the first equation by ty* and integrate it whole coordinate space of the second particle; in a similar fashion, we multiply the second Hartree equation by ty* and integrate it over the coordinate space of the first particle. Adding the two resulting equations and dividing the total sum by 2 we obtain the expression for the energy
'
^J
*'**'*'
~V
>>''>'
(23.66a)
373
where
in the case of two particles /, /=!, 2. This equation can also be used successfully for a larger number of particles. If we neglect the interaction energy (that is, if we set the terms
containing -^- equal to zero) and use the relations
and
Hartree' s equation can be resolved into a system of two independent equations

(H, -,)<[,,
describing the motion of each particle individually. In problems treated by the Hartree method, the field in which each electron moves is composed of an external field (for example, the nuclear field) and the field produced by all the other electrons. This method is, therefore, known as the self-consistent field method. Fock (1930) generalized the Hartree method by taking into account the exchange effects. For this purpose it is necessary that the test function in the initial equation (23.62) be consistent with the Paul! exclusion principle. The choice of test functions is, therefore, further restricted by the requirement of antisymmetry (antisymmetric functions satisfying the exclusion principle will be considered in greater detail in the next chapter). Hartree' s system of equations (for example, for an electron shell in an atom) is solved by the method of successive approximations. First of all, the wave functions in the zeroth approximation (that is, neglecting mutual interaction between the electrons) are determined. These wave functions together with the interaction potential between the electrons are then used to obtain the first approximation equations. The first approximation wave function is substituted back into the Hartree- Fock equations to obtain the next approximation and so on. This process is continued until the solutions obtained in successive approximations are identical (within the desired accuracy), that is, until the solutions are selfconsistent. Hartree' s system of equations can be solved only by means of numerical methods of integration. With the help of modern computers it is now possible to use Hartree's method to determine the energies and the wave functions for both light and heavy atoms. Another approximation which can be used in treating heavy
atoms is the Thomas-Fermi statistical method. Although this method is not as accurate as the self-consistent field method,
374
it
comparatively simple and predicts many properties of multiThis method will be used in our subsequent inand we shall consider it in Chapters 25 and 26 in convestigations, nection with the theory of the periodic system of elements.
is
electron atoms.
G.
INVESTIGATION OF THE EXCHANGE ENERGY
Let us consider in somewhat greater detail the physical meaning of the exchange energy (23.24), which represents the average value of the Coulomb interaction between two electrons that are both partly in state n\ and partly in state tii. In accordance with Eqs. (23.30) and (23.32), the total energy of the system is related to the Coulomb energy K and the exchange energy A by the
expression
= F + KA,
to
s
<j>
(23.67)
and the minus sign to a . In order to analyze the exchange energy in greater detail, we shall examine the behavior of the system in time, taking into account the exchange energy. The wave functions of the symmetric and antisymmetric states may be written as
where the plus sign corresponds
<]>
f (0 = f e"
Introducing the notation
and
a
<l
(0
= *Y~
8,
**'.
(23.68)
*=. 4 =
Eq. (23.68)
(23 ' 69)
may
be represented as
<,'(/)='(
(t )
V U -v)e->'+'. = --(
<|
)-'"'-"',
(23.70)
Let us consider a state of the system described by a supers a position of the solutions (0 and (/)
<j
r2
It
= ^(0 = CY<0 + CV(0-
(23.71)
is
solution
easy to show that the function W (t) represents the general of the Schrodinger equation (23.7) in the first-order
shall
perturbation theory.
We
Then
now assume
is
one of the electrons

the function
\V (0)
in state
that at the initial instant of time (/ 0) n l9 and the other is instate n%.
= -- {(C + C
s
a
)
+ (C
Ca) v}
(23.72)

is
375
simply equal to the function
u.
It
follows that
-=(C
that is,
+ <?)=!, and
C s -C a =:0,
CS
/~a
/OQ 7Q\
~777T-
1^0. /OJ
According to Eqs. time t, the function

)
(23.70),
*F
(t)
(23.71)
and
(23.73), at a certain later
becomes
cos
S/
<?-"" {u
iv sin 3/}
==<r ttu ' {Cu u
+C
vv\,
(23.74)
where
(23.75)
is It obvious that the amplitudes Cu and normalization condition
C V9 which satisfy the
| ]
/" C
1
|
1/^1?
tf
-|-
|C.y|*=J,
/OO r7fi\ (2o.7o)
characterize the probability of a system being in states described by u and v, respectively. we have C v the system is initially Since at / and C n in a state represented by the function u. However, after a time
t=
(so that
8/
(23.77)
= y),
it
can be seen from (23.75) that the coefficients
Cu and C v
will have the values
At the time / T,the state of the system will therefore no longer be described by the function u, but instead by the function v. This indicates that whereas at time f one of the electrons of the t, system is in state n\ and the other in state 2 after a time A/ the first electron will be in state /i 2 and the second electron in
state
/ij.
The time
(23.77)
electronic states occurs is nected with the exchange energy A by the simple relationship
" during which the exchange" of called the "exchange time." It is con-
<
23 - 78 >
376
It
T

readily seen that
if
is
there
is
no exchange energy
(A=Q) then
00.
To conclude this section, when the first electron is

2
us evaluate the exchange energy state nt=ls, and the second 2s. The wave function tym is given electron is in the state n tyi can be found from (13.26) by (23.37) and the wave function t 1'
let
in
the
and (13.28)
For the exchange energy, instead
of (23.38),
we
obtain
=J
fc(r a )
|r|
^ ra|
ti
*i (ri)
(23.80)
Carrying out the integration in the same way as was done for the evaluation of energy, we obtain, instead of (23.38a),
Evaluating these integrals (which
is
a relatively simple process)
we
find that
(23.81)
In
Is
accordance with (23.78), the exchange time of electrons and 2s states in the helium atom (Z 2) has the value
in the
The exchange time

the
Is
in the case where one of the electrons is in state and the other in the 10s state is of the order of several
years, and there is practically no exchange. Consequently, it can be seen that the exchange energy plays a significant part only when the probability densities It/iJ* of different states mutually overlap to a considerable extent. When the overlapping of the
is insignificant, the exchange This situation somewhat resembles the from one coupled pendulum to another. It is oscillating pendulum transfers its energy to
wave functions
small.
energy A
is
very
transfer of energy well known that an another pendulum,
See also Problem 13.1.
377
initially at rest, and its amplitude of oscillations after a certain time interval becomes equal to zero. The exchange time of the oscillatory energy depends to a great extent on the relationship between the natural frequencies of oscillation of the two pendulums; the time of exchange attains a maximum when the two frequencies coincide (the case of resonance). It must be emphasized that this analogy is purely formal and is possible only because of the wave properties manifested in both phenomena.
Chapter 24
Elementary Theory of Multielectron Atoms

Including Spin States
A.
SYMMETRIC AND ANTISYMMETRIC STATES
As we pointed out at the beginning of Chapter 23, the quantummechanical theory of assemblies of identical particles contains several features that have no classical analog. The most important of these features arise from the fact that the state of a system is unaffected by interchange of identical particles. This is
the principle of indistinguishability. shall consider here the general properties of a wave function that describes a system composed of two identical particles. The state of the system is characterized by the position vector r, three spatial quantum numbers ti , / and m and the spin quantum number s. For the sake of brevity, we let n denote all three of the quantum numbers /*, / and m 9 The wave function of the system has the form
known as
We
^(/ZiVi; n&r>),
(24.1)
where subscripts
1 and 2 refer to the first and second particles, respectively. Let us introduce an operator p which permutes the position coordinates r\ and r^ or the quantum numbers n lt si and *, &,
P I ("iVY,
v r
'Wi^^faiVa; n^r^
(24.2)
It is
easy to find the eigenvalues of this operator
(24.3)
It
to the initial
follows from (24.2) that two successive applications of wave function
lead
(24.4)
THEORY OF MULTI ELECTRON ATOMS INCLUDING SPIN STATES
379
On
the other hand,
it
follows from (24.3) that
*F
(/i^ri; n 2 s 2 ra )
= W (n&r^
/z 2 s a
r2 ).
(24.5)
The eigenvalues
of the permutation operator are therefore

X
= dbl.
=
(24.6)
This result means that interchanging the particles leaves the function unchanged (symmetric wave functions, with A 1 )
wave
Ws
(riiw,
/z*Si/-a )
iF s
(flis^;
n L s^)
its
(24.7)
or causes the wave function to reverse wave functions with X n
sign (antisymmetric
â
OWt; /iV) =
Vi
(/i 1 s 1
ri
HIS/-,).
(24.8)
According to a principle of quantum mechanics, a set of identical particles can only exist in states with a definite type of symmetry, A state is either symmetric (described by a symmetric
wave function) or antisymmetric (described by an antisymmetric wave function). Quantum transitions between symmetric and antisymmetric states are impossible; states with different types of symmetry do not mix. This suggests that there are two kinds of one kind described by symmetric wave functions, particles: and the other by antisymmetric wave functions.
B.
FERMI-DIRAC AND EINSTEIN-BOSE STATISTICS THE PAULI EXCLUSION PRINCIPLE
of
Experimental and theoretical investigations of the properties systems of identical particles have shown that there are two kinds of particles with fundamentally different statistical propParticles
erties, and that this difference is essentially related to the spin.
with half- integral spin
/
(s
3
t
-^> -^
...
in
units
of the
Planck constant /y obey Fermi-Dirac statistics. These particles are known as fermions and include electrons, protons, neutrons, H mesons and hyperons (all with spin Va). On the other hand, particles with integral spin (s 0, 1, ...) obey Einstein-Bose statistics. These particles are known as bosons and include w mesons, K mesons (both with spin 0), and photons (spin 1). We may note here, without going into a detailed analysis of the statistical properties of aggregates of particles, that in EinsteinBose statistics an unliminted number of particles can occupy each
state. On the contrary, in Fermi-Dirac statistics only a single particle can occupy each state defined by four quantum numbers.
380
This characteristic feature of fermions was established empirically by Pauli (1923) before the discovery of quantum mechanics and quantum statistics; it is known as the Pauli exclusion principle. In order to establish the relationship between the type of symmetry of a state and the kind of statistics that is appropriate for this state, let us consider a system of two particles which have a negligible mutual interaction. The wave function for the system can be written in the form of the product
W (/i^; n&rj =
where
cles.
tyn iSl
(rj
$& (r
% ),
(24.9)
are the wave functions of the individual parti(24.9) corresponds to a state in which the particles with positions r\ and ra are in states characterized by the quantum numbers /tiSi and fl a Sj, respectively. Because of the identity of the particles, the state of the system is not changed when the particles are interchanged. The following solution is and
<ta* s
The
solution
therefore just as valid as (24.9):

;
n lSi r*)
^ ^
(r,)
(/)
(24.10)
in the states
(that is, particles
with coordinates
r^
and r are
respectively). wave functions (24.9) and (24.10) describe the same physical state of the system. We note that any linear combination of these functions will also describe the same physical state. One of the possible linear combinations is the sum of the solutions
,
and
NISI
Both
(24.9)
and (24.10), namely,
y-
*,., to)
W, (r,) +
^ ^
(/-,)
(r,)
(24.11)
(we have assumed that the functions ty*^ and orthogonal and normalized to unity, and therefore
^^
are mutually
introis
we have
duced the normalization factor
y^.
This solution
symmetric
with respect to interchange of the spatial and spin coordinates

(24.12)
Another possible linear combination (24.9) and (24.10)

wa
is
the
difference between
yf
*i*i to) tyn^ (r,)
151
(r,)
3 ,3
(r,)
},
(24. 13)
which gives an antisymmetric solution since

(24.14)
381
the form of *F S and ^ d it follows directly that particles Fermi-Dirac statistics can be described only by antiobeying The Pauli exclusion principle (Fermi stasolutions. symmetric tistics) is satisfied only by a solution of the form (24.13) since when both particles are in the same state (that is, n n^^=n and
From
st
= =
Si
s)
we have
W* (nsr^
nsr*)
0,
but
lrs
(nsrv
nsrj
^ 0.
(24. 15)
On
the other hand,

statistics,
^ a does
Bose
in
not satisfy the requirement of Einsteinwhich an arbitrary number of particles can
occupy the same state. It is easy to generalize the antisymmetric and symmetric wave functions to the case of a system composed of N noninteracting particles. The antisymmetric solution may be represented by the determinant
_1
YN\
(24.16)
The antisymmetric nature of this solution is obvious since interchanging any two columns changes the sign of the determinant. Moreover, it is obvious that the function (24.16) satisfies the Pauli principle because if any of the sets of quantum numbers are equal
(for
example, n
/i 2 ,
Si
s2)
The symmetric wave function can be written
the determinant vanishes. in the
form
of a
sum
of products
WS
= TT^f Wi to) *.'. to

1
'
**^ r^ +
(
"}
C.
(24.17)
ADDITION OF ANGULAR MOMENTA. RUSSELL-SAUNDERS COUPLING, CLEBSCH-GORDAN COEFFICIENTS

Let us consider more fully the antisymmetric wave function system composed of two particles
for a
V*
= -^r
*i' to) fa* to
- *,
(r*)
a ,2
(r,)
}.
(24. 18)
In the many- body problem, we are faced with the question of how to add the angular momenta. In the two-particle case, each
382
particle
is
/
t
RELAT1VIST1C QUANTUM MECHANICS
numbers
characterized by orbital angular and / respectively, for which

2 ,
momentum quantum
L!
tf/t
(/,+
!)
(24.19)
and spins
s4
and
s2 , for
which
.
(24.20)
in which these four quantities can be added. One of these involves adding the orbital and spin angular momenta separately and then finding their total sum
There are two different ways
L
This
= L!-fL
2,
= +S
S,
J=L + S
(24.21)
type of coupling of the particles in an atom is called LS coupling or Russell-Saunders coupling. It corresponds to the case in which the total orbital angular momentum and the total spin angular momentum are conserved separately. This situation is encountered most often in light elements. The second way of combining the angular momenta involves the addition of the orbital and spin angular momenta of each electron individually and then finding the total angular momentum of all the
particles
J^LL + SL, J^Li + S,;
J^ +
J,.
(24.22)
This type of coupling is called// coupling and is encountered mainly in heavy elements. It corresponds to the case in which only the total angular momentum is conserved, and therefore it is important when there is a strong spin-orbit interaction. We shall now compare the order of magnitude of the Coulomb and spin-orbit interactions in the helium atom. The Coulomb energy [see (23.23)] has the form
êl J (
(24.23)
lies approximately between 0.1 and 1). The existence of LS (T coupling depends on the Coulomb energy since the orbital and spin angular momenta are conserved separately in this interaction. The spin-orbit interaction for the helium atom
L s
'
V ~ RMW
(24.24)
(a VIM) is considerably smaller than the Coulomb interaction, and therefore the coupling is of the LS or Russell-Saunders type. It can
THE MULTI ELECTRON ATOMS INCLUDING SPIN STATES
383
be seen from (24.24) that the order of magnitude of the spin-orbit interaction depends strongly on the nuclear charge z (^Z 4 ), so that for large Z (heavy elements) the value of E s may be quite large. In this case the coupling will be of the // type.
-
Let us consider the problem of the addition of the orbital angular momenta of two particles and find the function YLM (1,2) which is an eigenfunction of the operator for the z component of the total angular momentum L L3 L!
= +
L*
YLM (W)
<
L '*
+ L*) YW 0.2) = *MYLM (W).

total
24 - 25 >
and of the operator for the square of the
angular
momentum
lx
ly )
9je
)
L*yLM
(1,2)
= {Lf + LI + (L lx
iLtj,)
+ 2^,1^}
As we
products
YLM
(1,2)
+ il + (L + tL (L - iL,y + (L = L(L+\) = W\YLM (24>26)

tjf
zy )
(1,2);
shall show, the eigenfunction
YLM
(1,2)
can be written as a linear combination of
.?)
and
*"i
>
!'
W V W.
>
(24.27)
where
the
wave functions X
(1)
Y%*
(2)
refer to the first and second particles,
respectively, and
L t,
1
>U
(O-^^
+
>T
(1);
Lfc
y?! 2 (2)-^m a K^
(2)
(2);
M Y" 0) = ^ 2 î
(/i
(1);
LI
/ 3 (/j
1)
(2)
(24.28)
The numerical coefficients ^ m OTi in the linear superposition (24. 27) arc called the vector summation coefficients of Clebsch-Gordan coefficients. For simplicity, we shall restrict ourselves to the case /, /, 1. Substituting (24.27)
= =
into Eq. (24.25)
we
find that
M = Wi + /"sand
yU (W
^n,/n a
= ^m,*
J7
1
Without loss of generality we may set 0;that is, we make the z axis perpendicular to the total angular momentum. Then it follows from (24.27) that
M=
= *-.n
(1)
(2)
.Kr' (1) Y[ (2)
+ ftoKJ (1) yj
(11.88)]
(2).
(24.30)
Taking
into account the equation for spherical

(Li;c
harmonics [see
fLjj)
^
-
(1)
=- *
V(/i
Wi)(/i
mi)
^^'(l),
(24.31)
and using Eq. (24.26), we obtain
L'KZO (1,2)
2
/i
1
{*_, [2K{
l
Kr
}
+ ô [4X? KJ + 2KJ Kf + 2K;
Ki]
(^ /j K, +
ft,
Kf KJ
+&
X{ K?),
(24.32)
This equation is equivalent to a system of three homogeneous equations for the unknown Clebsch-Gordan coefficients
A)*i
=0
H-(2
X)ft. 1=
=0
(24.33)
384

if its
This system has a nonv finishing solution

2
determinant
is equal to
zero
0.
(24.34)
4-X
Calculating this determinant and subjecting the wave function condition
YLJA
to the normalization
J
that is,
fl>W LM
*o
we
obtain three possible solutions:
From the conditions X L (L the case in which the vectors
+
fa
1)
we
fa
find that
and
L 0; that are antiparallel
is, this solution
corresponds to
fa
2)
-,
momentum
vectors at an angle
This case corresponds to the addition of the angular of 60
3)
This solution corresponds
to parallel
fa
and
/,
fa
f A.
quantum numbers, can be oriented
conclude, therefore, that two vectors fa and fa, whose absolute values are integral at angles such that the vector representing their sum is also characterized by integers. For fa these values are /
We
fa
(24.35)
(a total of 2/a
values).
This method of adding the angular momenta is based on a vector model of the atom. We can obtain our previous results by setting /1 =/2 =1 andAf 0. It is interesting to note that the same values of the Clebsch-Gordan coefficients would also be obtained in adding the orbital angular momentum to a spin angular momentum equal to unity (for example, for photons X In this case, however, one would have to use the normalized spin

function instead of the spherical harmonics Y%*. It is easy to coefficients will satisfy the following conditions:
385
show
that the Clebsch-
Gordan
In
M number w 3 second parade is
general the Clebsch-Gordan coefficients bm will depend not only on the quantum m it but also on L and Ai. If the orbital angular momentum of the moments , we obtain /a 1, then adding both orbital angular
for which the Clebsch-Gordan coefficients are given in Table 24.1.
D.
WAVE FUNCTION OF THE HELIUM ATOM

INCLUDING SPIN
We shall now consider more fully the wave function of the helium atom, for which, as we know, the interaction between the spins and the orbital angular momenta of the electrons is of the Russell-Saunders type. Since, therefore, the orbital and spin angular momenta are added independently of one another, the wave function can be written as a product of two parts, one of which depends on the spins of the particles and the other on their coordinates. The wave function must be an antisymmetric function with four quantum numbers
= -C(s
fa
/*)
lf
C(s 2
s,)*,^/-,, ra )
rO.
=
(24.36)
SfcH^ta,
There are two possible cases: either the function is symmetric with respect to the spins and antisymmetric with respect to the coordinates, or the reverse case is true. We have therefore the following two types of solution:
</
r,),
24 - 37 >
(n, /).
(24,38)
We
been obtained (see Chapter

s
'P/iirt.i T/II"J.\
recall that the spatial part of the wave function has already 23). For n^ ^=n^
t
(f\ it
/*a) /
7= (U
|^2
~T~ &0,
/9A ^Q\ ^*.otf^
386
V
s
CO
s
.2
o o
^
pQ
42 "U
Cj
?
O
u
<4H
+ ?3
CD
3
H
aj
THE MULTIELECTRON ATOMS INCLUDING SPIN STATES
387
where
(n).
(24.41)
Let us now investigate the spin part of the wave function. It has already been mentioned that in Russell-Saunders coupling the spin angular momenta are added independently of the orbital angular momenta. The law of conservation of spin, therefore, provides a basis for the construction of the spin wave function for the twoelectron system. We shall take the spin function of each electron to be an eigenfunction of the operator for the z component of the
spin
,
(24.42)
and also of the operator for the square of the spin
(24.43)
Here we write primes
2x2
Pauli matrices
<y
(see Chapter 16) without
Hi-?)Thus
the spin function C=( C M
of a single particle satisifies
two
equations
(24 ' 44)
(24.45)
3 a* and so on, we obtain from Eq. (24.45) X, . The /4 matrix equation (24.44) for Xj is equivalent to a system of two
1 ,
2
Since
homogeneous algebraic equations
0.
(24.46)
from which it follows that there are two solutions corresponding to the two possible orientations of the spin with respect to the z axis
1)
*!
'/
*!=!,
C2
0.
(24.47)
388

is
Here the spin

sponding
parallel to the
l
axis.
The wave function corre-
to the eigenvalues
/2
is
(24.48)
X 1=
2)
'/
d = Q,
ca
=l.
(24.49)
In this
case the spin

is
is antiparallel to the z axis.
The correspond-
ing wave function
/m
(24.50)
In Eqs. (24.48) and (24.50) the value of the z component of the spin is indicated in the parentheses following the amplitude C. It is easily shown that both spin parts of the wave function satisfy the orthonormality condition. If, in accordance with the definition, we take the Hermitian conjugate spin function to be a single-row
matrix
then from (24.48) and (24.50)
we see
that
and
The
effect of Pauli
(24.50), (24.48)
matrices on the spin function, according to Eqs. and (24.44), is
(24.51)
We
shall look for the spin function
C of
the two electrons of the
helium atom in the form of a superposition

'
\
j
r C
I
2
'
ajC, ^-j -f
'
^J
r2 I C
(
\
j
(2 4.52)
THE MULTl ELECTRON ATOMS INCLUDING SPIN STATES
389
where
C, [ db -~\
and C 2
^:
^-J
are the spin functions of the first and
second electrons, respectively, while a {9 a,, a* and ?* are the Clebsch-Gordan coefficients that have to be determined. Let us now attempt to select the coefficients a^ of the function (24.52) in such a way that the latter will be an eigenfunction of the operator for the z component of the total spin
s,
= s; + s; = \ w +
~
"2
a3 ),
(24.53)
and also of the operator for the square of the total spin
-f 2 *'
<J"}
Here the primes and the double primes on the Pauli matrices
that these matrices act on the spin functions of the and second electrons, respectively. We thus have two matrix equations
indicate
first
-W + aJ)C =
ftX 1
(24.55)
~ a'-*" C == /t%C-
(24.56)
Substituting the total spin function (24.52) and taking into account the effect of the Pauli matrices on the spin functions of the individual electrons [see (24.51)], we find from (24.55)
Equating the coefficients at the same spin functions on both sides of the equation, we obtain the equations for the parameter Xj and the coefficients^
M*
X,)
0,
a2 X 1
a3X 1
0,
a4 ( 1
-}- Xj)
(24.57)
390
Similarly, using the relation

(a'-
a")
= aA
1)
C,
(1)
- a,
[C,
j)
Cs
(- 1)
(24.58)
which can be obtained with the aid of (24.51), we obtain from (24.56)
2
j
X 8 a,
aJ
As usual we
total
X2 where 5 is the eigenvalue of the t note that the spin. systems of equations (24.57) and (24.59) must be satisfied simultaneously. From these equations we find that the possible nonvanishing solutions have eigenvalues
set
= -l)a = S(,S+l)
(X a
li
= a=
2
4
2
X 8 a4,
(X 2
~l)a
3.
24 - 59 )
We
X,
2,
with
X1
X1
X2
^0,with
z
= + 1,0, =
1.
(triplet),
(singlet).
Consider now four cases. Case

is
Total spin is equal to unity and
directed along the
allel.
The spins of both particles are parThe corresponding solution, given by (24.52), is symmetric:
axis.
(
24 - 6
Xj
=+
a,
=a =a =
2
0.
Case
axis.
2.
Total
spin is equal to unity and is antiparallel to the
The spins of both particles are parallel. The corresponding symmetric solution is
L
2
-
Si
(
Us
(24 - 61)
>)
a4
=l,
Case 3. Total spin is directed perpendicular to the z axis and is equal to unity. The spins of both particles are parallel. The corresponding symmetric solution is
c'
=T
c
i
'
f)
c*
(-
+c
'
THE MULT ELECTRON ATOMS INCLUDING SPIN STATES

I
391
4. Total spin is equal to zero. The spins of the particles antiparallel. The solution is antisymmetric:
Case
are
note that the choice of the nonvanishing Clebsch-Gordan cowas such that all four solutions are normalized to unity. Returning once again to the spatial part of the wave function for the helium atom and using the fact that the total solution must be antisymmetric we have, in accordance with (24.37) and (24.39),
efficients
We
UT
=0
'
(SL SjH- (ri,
ra )
(24.64)
(three states) and
Ca
(s,,
52
)f
(r,,
r.)
(24.65)
(one state). The spin functions are given by Eqs. (24.60)-(24.63), and the spatial part of the wave function for n ^ th is
{
In the case where both electrons are in the same state there is only one solution with a symmetrical spatial part:
(24.67)
E.
PARAHELIUM AND ORTHOHELIUM
We have obtained the wave functions that describe two states. One state (parahelium) is characterized by antiparallel spins of the electrons [the wave function (24.65), which is symmetric with respect to interchange of the coordinates]. The other state (orthohelium) corresponds to parallel spins [the wave function (24.64), which is antisymmetric with respect to interchange of the coordinates; see Fig. 24.1]. A very interesting property of these states is that both types of helium are stable; in other words, parahelium and orthohelium are not converted into one another. We can convince ourselves that both systems are closed by direct
392
calculation. The matrix element of the dipole moment corresponding to a dipole transition from orthohelium to parahelium
rs-a
= J f (/VaKrj =J (O + = - J f (/VriXn + r
s
*' (/-2,r,)
s
/*)
<!>
(24.68)
turns out to be equal to zero, since

(24. 69)
[In (24.68) we have changed the variables of integration and used the symmetry properties of the wave functions.] Dipole trans it ions o 'arahelium Orthohelium from one state to another are therefore forbidden. Mutual conversion of these Fig 24 1. Electron spin orientation in the helium atom states may, however, take place under the action of other particles. For example, the bombardment of orthohelium by electrons may lead to the replacement of an ejected electron by another electron with opposite spin, and thus orthohelium can be transformed into parahelium, and vice versa.
F.
It
ENERGY SPECTRUM OF THE HELIUM ATOM
has already been noted that the total orbital angular momentum is a result of the addition of the orbital angular momenta / and /.2 of two electrons, assumes integral values (RussellSaunders coupling). In the particular case where t l =l (both electrons are in the p state), the total orbital momentum angular takes the values L In terms of the vector model, these 0. 2. 1, values correspond to the following situations: 1. L 2. The angular momenta are parallel;
L,
which
t
2.
L=\. The angular momenta are
at an angle of
60;

3.
393
The angular momenta are
antiparallel;
At
In the general case integral values
/,
i=/
,
-./ 8 =o.
all
where /i^/ 2 the number L takes
possible
= +
/,
/!
+ /,-!,
/i
-M, -2,
...
/,-/,.
(24.70)
In denoting the
energy levels for multielectron atoms, we follow rules as for hydrogen-like atoms, except that the states with a definite value of the orbital angular momentum L are denoted
the
same
by capital letters
=Q L=2 L= 3
L
S state
L=l
P
F
state
state,
D state
and so on.
Let us enumerate the energy levels of the helium atom. The lowest level corresponds to the total orbital angular momentum L Q with both electrons in the s state. This singlet term of parahelium is denoted by
and corresponds to an antiparallel spin orientation. The superscript at the upper left-hand side of the letter characterizing a particular term indicates the number of states (the multiplicity of states). The next term corresponds to the case when one of the electrons is in the Is state and the other is in the 2s state. In this case, both
parahelium
(ls2s)'S
and orthohelium
momentum and
field.
are possible. Orthohelium, unlike parahelium, has a spin angular exhibits anomalous Zeeman splitting in a magnetic
note that the (ls26') 3 S, level of orthohelium is metastable, because a transition to the lower energy state belonging to parahelium is forbidden by the selection rules. The levels of parahelium are singlets (spin 0), and those of orthohelium are triplets
(spin
1).
We
394
The splitting can be easily explained in terms of the vector model of the orbital and spin angular momenta. According to the vector model, the sum of the two vectors L and S (that is, the
total angular
ev
momentum vector)
can take the values
...
J
,|L-S|.
(24.71)
\1*2S)
J;
For >S, the total these values is
number
of
2S+1,
whereas for
it is
(24.72)
2L+1.
It
(24.73)
follows that for the 5 term 0), the total number of possible states for both ortho(L =
helium and parahelium

Fig 24.2 Energy level diagram for the helium atom (the splitting of the P
level is not in scale)
is
equal
to one. Inexactly the
same way,
is
equal to two (that
is,
s term of the hydrogen atom (one valence electron) is a single, and the number of states the levels are doublets) only for p d terms
the
(for
which />s).
total spin
S of helium assumes two values: 5 0,1. For of states (2S-J- 1) is equal to unity, and for S 1 it is equal to three (25 -f- 1=3); that is, the state is a triplet. A general diagram of the energy levels of the helium atom is given in Fig. 24.2.
The
the
number
~-\ the states group (s^^-andS are quartets or doublets. Thus the number of valence electrons (see Chapter 25) completely determines the splitting of the spectral
lines.
For elements
of the third
The theory
of multielectron
atoms
is
treated in several books.
particle is In the
Yukawa
potential well
êe, for example, D. and Bacon, 1964.

2
I.
Blokhmlsev, Principles of Quantum Mechanics

Chapters 23 and 24
(trans.),
Allyn
These problems
refer to
395
A i e~î r and determine the upper limit of the Choose a test function in the form ^ l values of x at which at least one discrete level will exist. Using the obtained solution, consider the special case of a Coulomb potential (g* ?? * 0). Show that discrete levels always exist for the Coulomb potential. Determine the lowest energy level and the corresponding wave function. Solution. The normalized wave function is of the form [see (23.47)]:
1=
J/7C
The average values
of the kinetic and potential energies are [see (23.52), (23.53)]
~
2/w
'
I'î
+ *)*
The appearance
of the first discrete level is possible
under the conditions
E7=f+K =
from which we
find
0,
-||i=0,
max =0, and

is
_ Wo
2
'
Discrete levels exist for7:<0. This happens when*<y max . For a Coulomb field therefore this condition is always satisfied. The average energy in the Coulomb
case
fc-S-w
Using the condition that Ei
must be minimum, we
find
where o
^ris
the radius of the first
Bohr
orbit.
The expressions for
the
wave func-
tion and the average energy will be exactly the same as those yielded equation for the Is level of the hydrogen atom (see Chapter 13).
by the Schrodinger
roblem 24.J) Calculate the diamagnetic susceptibility of the ground state of one par ahelium , using the expression (23.47) obtained for the wave function by the variational method. Solution* In the ground state of parahelium, the orbital and spin angular momenta are equal to zero(/=0), and therefore parahelium must be diamagnetic (see Problem 20. IX In accordance with (20.48), the diamagnetic susceptibility per gram-atom is given by
g ram- atom"
'
the expression
'(S^^
where
AT is
(24.74)
Avogadro's number and
Ff
+r =
j~
(n
+ r\)
2
|
<!/(/-!,
r,)
rf\v,
(P r 2
(24.75)
From
(23.47)
we obtain an approximate expression for ^
(r lf r.)
396
where for helium
5 ^~~ -r-7
ig
16"
Evaluating the integral (24.75) with the above expression for
<p t
we
obtain
The diamagnetic
susceptibility is therefore
y.theor
= __ =
67
10-,
which
is in fairly
good agreement with the experimental value

1.90
10-.
Chapter 25
Optical Spectra of Alkali Metals

A.
GENERAL DISCUSSION OF THE STRUCTURE OF COMPLEX ATOMS
It has already been pointed out that the position of an atom in the periodic table of elements is determined by the atomic number Z. This number characterizes the nuclear charge in units of e that is, it is an integer that is equal to the number of protons in the nucleus and also to the number of electrons in the neutral atom. Thus, for example, sodium has 11 electrons (Z=ll) and uranium has 92 (Z 92). The most important questions in the theory of a complex atom concern the electron density distribution and the energy spectrum. In studying these questions it is necessary to consider the mutual interaction between all the electrons, in addition to the attraction between the electrons and the nucleus. Owing to the interelectronic interaction, the energy values of the electrons will be smaller (in absolute value) than those given by the well-known express ion
;
Just as in the hydrogen atom, each electron in a complex atom characterized by four quantum numbers. For Russell-Saunders coupling (the case where the spins and orbital angular momenta of the individual electrons in a given shell are added separately) these quantum numbers are as follows:
is
,
the principal quantum number n , 1, 2, 3, 4, . . (2) the orbital angular momentum quantum number / =0,1, 2, .... (/i-l). . , /, and 1, 0, (3) the magnetic quantum number m 1/2 (characterizing the (4) the spin quantum number m projection of the spin on the z axis). In the case of // coupling, the four quantum numbers are as follows: (1) the principal quantum number n , (2) the orbital angular momentum quantum number / , f / , and ( 3) the total angular momentum quantum number / /
(1)
.
398
(4)
1, / (characterizing the component along the z axis). As we have already said, Russell-Saunders coupling (or LS coupling) is characteristic of light elements, while // coupling is characteristic of heavy elements. Both types of coupling give the
my
/,
/-j-1
of the total angular
momentum
same number of levels. A group of closely spaced energy
levels, which is separated
from other energy levels by an appreciable gap, forms a shell
made up of several subshells. Just as for the hydrogen atom, the classification of the shells is based, as a rule, on the principal quantum number n. The relationship between the principal quantum number n and the letters which are used to denote the shells in x-ray spectroscopy is as follows:
n
1
Shell
symbol
K
L
2 3
4 5 6 1
M
N
P Q
Within the shells, electrons with values of the orbital quantum subnumber / equal to 0, 1, 2, 3, form the s, p, d, /, shells. To determine how the shells and subshells are filled, one should take into account the Pauli exclusion principle, according to which there can be only one electron in each quantum state characterized by four quantum numbers. In a state with fixed values of n, /, m, there can be at most two electrons, differing in spin direction. Since the quantum number m, which varies from -/to +/, can take 21 + 1 values, we obtain the following expression
. .
for the
maximum number
of electrons in a given subshell:
(25.1)
The same value

numbers
n,
I,
for
N/
is
found for
jj
coupling For given values of the three quantum

jt
..
,
the fourth quantum number nij may take the values corresponding to 2j + 1 states Hence we obtain the same value for N;
j,
~- l /i,
/a,
J,
where
'
OPTICAL SPECTRA OF ALKALI METALS

It
399
follows that the
s(i
0),
p(/
maximum number
l),
c/(/
2)
and
f(l
3)
of electrons in the subshells is as follows:
Subshells with a larger value of / are not encountered in unexcited atoms, and therefore we shall not discuss them here. We can now find the maximum number of electrons which can occupy a given shell
n-l
rt
=y
1=0
= 2(l + 3 + +
..
.
(2/i-
l))
2/i
"*"
^~"
=2ft a
(25.2)
there can be at most two electrons in the K shell, eight electrons in the L shell, eighteen electrons in the shell, 32 electrons in the N shell, and so on. It should be mentioned that these relations between the quantum number and the number of states in the various shells are true, generally speaking, only for
Thus
hydrogen- like atoms.

establish the order in which the shells and subshells are complex atoms, it is necessary to consider the mutual interaction of the electrons. As we have emphasized several
filled
To
in
times, this problem cannot be solved exactly. The application of quantum mechanics to multielectron systems, however, has led to the discovery of various important properties of such systems, including, in particular, the exchange interaction. Moreover, the development of approximation methods has made it possible to obtain highly satisfactory quantitative results for complex atoms. As indicated in Chapter 23, the simplest approximation methods are the variational methods developed by Ritz, Hylleraas, and others; these give comparatively good results for the light atoms (up to potassium). A more complete analysis of the structure of an atom can be made by means of the self-consistent field method developed by Hartree and Fock. This method has been used to determine the energy and the electron distribution in heavy and light atoms (see Chapter 23). This method also gives the shell 2 structure of complex atoms. Unfortunately, the use of the self-consistent field involves some very complicated computations, which can be carried out only by computers. As a result, the eigenfunctions characterizing the electron distribution are obtained in the form of numerical tables rather than analytical expressions.
For more details see Hartree, John Wiley & Sons, Inc., 1957.
D R The
Calculation of Atomic Structures,
New
York:
400
B.
THE THOMAS- FERMI STATISTICAL METHOD
Besides the approximation methods, there are statistical methods that have been introduced on the basis of principles developed by Thomas and Fermi. These methods apply mainly to heavy atoms. In the statistical treatment, the electrons of an atom are re0, just as in the theory of garded as a degenerate gas with T metals (see Chapter 6). The Thomas-Fermi statistical method is, of course, less accurate than the Hartree-Fock self -consistent field method, because many features of the behavior of the in-
dividual electrons are neglected. In spite of this general shortcoming, the Thomas-Fermi method provides a fairly simple explanation of many important properties which are characteristic for the average behavior of electrons in an atom. The Thomas-Fermi method does not enable us to find the shell structure of an atom, but it does explain several important features of the filling of the electron shells. This method can be used for calculating the total bonding energy of electrons in an atom and the radii of atoms and ions. It can also be used to determine the influence of screening electron shells on the scattering of fast electrons, bremsstrahlung, and the creation of electronpositron pairs due to absorption of x-ray photons in the nuclear
field.
Presently statistical methods are being successfully used to construct a theory of heavy nuclei and a theory of matter at high
pressures
(for
example, inside stars).
In heavy atoms the positively charged nucleus is surrounded by a cloud of negatively charged electrons, which partially screen the electric charge of the nucleus. The potential of an ionized atom at distances greater than its radius is given in the first approximation by the following expression:
tto^V-W",
where
7.
(25.3)
is the atomic number, and is the number of electrons. For a neutral atom and, therefore, <J>oo 0; that is, the electrons completely screen the charge of the nucleus. In constructing a statistical theory, three forms of energy should be taken into account. electrostatic energy of attraction between the electrons and the nucleus. 1. The This energy is related to the electron density po (the number of electrons per unit volume) by the expression
Af=Z
(25.4)
where
is the
charge of the electron, and $ n
is the potential
produced by
the nucleus.

2.
40
The electrostatic energy
of repulsion
between the electrons
ê-e
=-^y
a ? Po*e ^ ^,
(2Ma)
where
3.
The kinetic energy

3
absolute electron
(6.33a)
:
of the electrons in the atom. Just as in the theory of solids at zero, the average kinetic energy of an individual electron is related to the density P O by the following expression, in accordance with Eqs. (6.33) and
rav =
where
*
xp
/,
25.4b)
= ra
'
< 37t2)V
= TO e a
*
(3 * )S/
'
< 25- s >
Hence
the kinetic energy of the electrons is
(25.6)
The total energy of an electron gas in the field of a nucleus is equal to the sum of the potential energy, which consists of two parts [see (25.4) and (25.4a)] f and the kinetic energy [see (25.6)]. Thus the total energy is
"
-.
j
P0
v8
^_! j
o
I>n
l+V: j Saji^-w.
Po </'*
(25.7)
The density
of the electron gas
must
satisfy the condition
= #,
(25.8)
where
is the number of electrons in the atom, With the auxiliary condition (25.8), the variational principle can be formulated as
follows:
a{
+ *o<Mf}=0.
between the
total potential
(25.9)
Using and the density of electrons
this principle,
we can
p
find a relationship
$=
<
+ $e
Po
= -r(2/Wof
(*
Va
<T'o
.
(25. 10)
402

fl
where the Lagrangian multiplier . which plays the role of a constant potential, must be found from the boundary conditions. In the derivation of (25, 10), we have used the
relations
Q
(25.11)
fi^^
Substituting Eq. (25.10) for the electron density into Poisson's equation for a spherically symmetric electron distribution
v a<fr
J.
r$
= 4ne
O po
(25. 12)
and recalling that $ const, we obtain the basis of the statistical model of the atom
Thomas-Fermi equation which forms
the
7 S3 r <*
- *) =
-gF
<
2m
<'>
'
(*
- *>'''
(25.13)
When a specific problem is investigated, the solution of Eq. (25.13) must satisfy certaip definite boundary conditions. For an atom the boundary conditions can be given in the form
and
forr-^0 $-$0 = ?^ r
(25.14)
for r
r,.
Here
zero
r
at
/-
=r
is
determined from the condition that the electron density should be equal to that Is, p (r )= 0. Hence, in accordance with (25.10), we find >
(2
With the help of Poisson's equation (25.12) and (25. 13), the condition (25.8) written as
may
be
(2M7)
$
It
follows
from
(25.17)
we
(25.16) that get, therefore,
and ro
= oo for a neutral atom
(N=Z).
Instead of
This condition will hold
if
the second boundary condition is satisfied
Hm r$ = 0.
(25.18)
403
We
note that the
Thomas-Fermi equation
(25, 13)
has one exact solution
as can be easily shown by substituting (25.19) into (25.13). This solution for a neutral atom ($ 0) will satisfy one of the boundary conditions for r -co [see (25.18)]. The second boundary condition for r-*0 [see (25. 14)] f however, will not be satisfied. Unfortunately, solutions of the Thomas-Fermi equation which satisfy both boundary conditions cannot be expressed in analytic form. 4 Substituting (25. 14) into (25. 10) f we find the behavior of the density po as r - 0,
Po
= const r"-
8/a
(25.20)
4>
The solution (25.19) for a neutral atom gives too high a value for The more exact Hartree-Fock method shows that the electron density
asr*co.
decrease
will
oo . exponentially as r Since we are Interested only in the basic principles of the statistical method, we shall construct an approximate statistical theory of the atom with the help of the variational method. This will enable us to formulate a solution of the problem in analytic form with only small quantitative deviations which are of no importance to us.
C.
SOLUTION OF THE THOMAS- FERMI PROBLEM BY THE RITZ VARIATIONAL METHOD
Solving a problem by the Ritz variational method, one can propose any number of test functions which depend on the different values of the variational parameter X. We shall choose the test function on the basis of the following considerations. The function should agree roughly with the solution of the Thomas-Fermi equation for
We may note that the numerical integration of this equation presents two advantages over the numerical integration of the Hartree-Fock equations First, the Thomas-Fermi equation is much simpler than the Hartree-Fock equations Second, this equation and the for a neutral atom with Z - N) can be transformed boundary conditions (for example, $ = into a universal form which is independent of Z To do this, we must replace $(r) by a
new
function
where
* = a
'
= C"
V128Z,
With the function f(x), Eq. (25.13) becomes
V* ~f
~-
/2
>
(25 13a)
and the boundary conditions (25 14) and (25 18) become
/"(*)
for
x -
f(x)
for
x-00.
(25 14a)
These equations
are of a universal character, that is, they do not necessarily depend on the quantity Z. Therefore, after numerically integrating the Thomas-Fermi equation, we can change the variable (which depends on Z) and use the equation to investigate any
heavy atom. This cannot be done with the Hartree-Fock equation.
404
r(this region is the most important with regard to the solution of the problem as a whole), and it should have a comparatively simple form, so that it can be exactly integrated in the calculation of the total energy. One test function which satisfied these
requirements
is
16*r a/ *
This function
is
already normalized to the total number of electrons
Po
d.v= -5L-1
IJ
yf e- V\r dr = N,
(25.21a)
and therefore the auxiliary condition (25.8) will be satisfied automatically. For r-* 0, the test function (25.21) changes in the same way (p ~r~ 8/2 ) as the solution of the Thomas-Fermi equation [see (25.20)]. As we shall see later, tids explains the good quantitative agreement between results obtained by means of the test function (25.21) and those obtained by means of the potential satisfying the Thomas-
Fermi equation. The potential due
to the electrons in the
atom
r
is
e
- ~ -** - e~ ^ (
1
/x7 lA7 e~
).
(25.22)
This can be easily shown by substituting Eqs. (25.21) and (25.22) for
equation
and $ e into the
Using also the expression
$n
boundary condition (25.15) for
= we = r ro-*oo,
,
can see that the
total potential satisfies the
when
the charge density vanishes together
with the exponential term e~ We can also find an expression for the kinetic energy parameter X. From Eqs. (25.6) and (25.21) we have
in
terms of the variational
We have the following expressions for the potential energy due to the interaction between the nucleus and the electrons [see (25.4)] and for the potential due to the mutual
interaction of the electrons [see (25B 4a)]:
(25. 24)
-6--
(25.25)

Adding Eqs. (25.23)-(25.25), we find that the
total
405
Is
energy (25.7) of the electron cloud
=
where
/tt
-A,
(25.26)
of the
The varlational parameter X , corresponding to atom, can be found from the condition for
-\
the reciprocal of the effective radius the minimum of the total energy /: of
rp
the atom, that is,
^r
0.
Hence we
find
__
/3ny/s _/V_j
(25.28)
T~~100\2~J
(25.29)
\37c
For a neutral atom (N
=Z
),
we have
(25.30)
a? i? (_2 y/9 64 \3ic/
^ 7/
758
<
<* o
It is worth noting that numerical integration of the Thomas-Fermi equation leads to a value for the energy of the atom which is very close to that given by (25.30)
E T- F
0,709
. . .
Z 7/s
20.94
Z ?/3
ev.
(25.30a)
The expression (25.30a) (with the sign reversed) represents the total bonding energy of a neutral atom, that is, the energy required to remove all (or ionization energy) the electrons from the atom. The theoretical values obtained for IF from (25.30a) are quite reasonable even in the case of the hydrogen atom, but they are somewhat higher than the corresponding experimental values. The relative error decreases with increasing Z (see Table 25. IX
Table 25.1
Theoretical and experimental values of the total ionization
energy
of
atoms
(in units
of
e Q *la
406
Concluding this section, let us compare the curves which are obtained for the charge on the basis of the Thomas-Fermi statistical distribution in neutral argon theory and on the basis of Hartree's method of the self-consistent field (see Fig. 25.1). It can be seen from the graph that the curve of ? computed according to the Hartree method (Curve B) has characteristic maxima and minima corresponding to the electron shells, whereas the curve calculated according to the Thomas-Fermi statistical theory (Curve A) describes only the average behavior of the electron density, and therefore has no relative maxima. For large values of r there is also a marked lack of agreement between the two curves: the Hartree-Fock method gives functions which decrease exponentially with increasing r , whereas in the statistical theory the decrease is pro-
(Z18)
4 portional to r' [see (25.19)].s
40
20
s
X;
3 r/a
Fig 25.1. Comparison of the electron density distri= 18) obtained on the basis of the Thomas-Fermi method (Curve A) and the Hartree-Fock method (Curve B) A pseudologanthmic scale is used for the ordmate axis (we have plotted the quantity Da ), where D - Arrr 2 p ) Therefore, the graph &n(\ will be linear for small r>a Q and logarithmic for large D(IQ The choice of this scale enables us to follow the variation of /QQ at large values of r, as well as for r < a
butions in argon (Z
\
,
D.
ENERGY LEVELS OF ALKALI METAL ATOMS
In studying spectral lines in complex atoms, it is necessary to distinguish between outer and inner shells. In the hydrogen atom there is only an outer shell (the K shell), which contains one electron. In helium (Z 2), the /<" shell is completely filled (a noble or inert gas). In litnium (an alkali metal in Group I of the periodic table, Z 3), the inner shell (K shell) is completely filled, but the outer L shell has only one electron. The filling of the L shell is completed in Ne(Z 10). In sodium (an alkali metal, (Z 11) , the inner K and L shells are completely
detailed exposition of the statistical theory of an atom is given
The one des Atoms und
ihre
m P. Gombas, Die Anwendungen, Vienna* Springer- Verlag, 1949.
407
shell. The filling of filled, but there is one electron in the outer the shells in these atoms is illustrated in Fig. 25.2. It should be noted that the bonding energy of an electron in an inner shell is much greater that that of an electron in an outer shell. An indication of this is provided by the ionization energy, which is more than 20 ev for inert gases, whereas for alkali metals it is only slightly more than 5 ev. The removal of the first valence electron in lithium requires the expenditure of 5.39 ev, but removal of the second and third electrons from the inner shells requires the expenditure of 76 and 122 ev, respectively.
Fig. 25.2. Diagram of the filling of electron shells in different atoms. On the right, atoms
with completely filled shells (inert gases) The dark spots represent electrons and the light spots (with a plus sign in the middle) represent
nuclei
The atoms in Group I of the periodic table (Li, Na, K, Rb, Cs, etc.) are known as alkali metals; each has an outer shell containing
one electron, just like the hydrogen atom. Their optical and chemical properties, therefore, should be essentially the same as those of the hydrogen atom (for example, all of these elements are monovalent and they all exhibit doublet splitting of the spectral terms). The optical spectrum is caused by the transition of a valence electron (that is, an electron in the outer shell) from an excited state to a lower state. The excitation of electrons in the inner shells generally requires considerably more energy than does that
of electrons in the outer shells therefore, downward transitions from excited states to the ground states of the inner shell are accompanied
;
408
by the emission of x-rays (see Chapter 26). The atomic nucleus and the electrons in the inner shells together form what is known as the atomic core. Thus the charge of the atomic core is equal to Z n =:Z N, where N is the number of electrons in the inner and thus the shells. For alkali metals (Li, Na, etc.) A/ the is atomic core to of equal 1). Therefore, unity (Z a charge the main part of the potential energy which retains the outer electron in the metal is the same as for hydrogen, namely,
= Z1, =
=-
= -7'
(25 31)
'
The analysis
of the spectra of alkali metals can thus be based on the corresponding expression for the energy which was obtained in Chapter 13
= -^.
25 32 )
-
Similarly,
we can use the hydrogen wave functions as the zeroorder approximation of the wave functions
*
=*/
(25.33)
metals, however, we cannot confine ourselves to the in treating the interaction between the valence electron and the atomic core, and we must also take into account the polarization forces and the "smearing* of the atomic core over a finite volume. This yields various corrections and removes
In
alkali
Coulomb energy
'
the
degeneracy which occurs in hydrogen. Bohr's semiclassical theory, the orbits of valence electrons were rigorously divided into orbits which penetrate the atomic core, and those which do not. In the case of "nonpenetrating" orbits (orbits with nearly circular trajectories) we need to con/
In
sider only the polarization forces, since the potential outside the atomic core (that is, outside the inner shells) is completely independent of the radial distribution of a spherically symmetric
charge. The radial distribution
is very important only for 'penetrating" orbits (elongated ellipses) (see Fig. 25.3). In quantum theory the concept of a trajectory is no longer meaningful. In order to classify orbits as "nonpenetrating" or
"penetrating,"
the
we must introduce a new convention; namely, if wave function describing the behavior of the valence electron
be set equal to zero inside the atomic core, the orbit is "nonpenetrating," and if on the contrary the wave function cannot be set equal to zero inside the atomic core, the orbit is
*
can
'penetrating."

It
409
should be noted in this connection that the s orbit of a complex is always penetrating, since its wave function differs from zero inside the atomic core and, indeed, even inside the nucleus
atom
(25.34)
We shall now calculate the polarization forces that arise between an outer electron and the atomic core. The outer electron will obviously repel electrons in the inner shells and attract the nucleus. As a consequence, the atomic core is polarized, and a polarization force arises between it and the outer electron
l
2*5(2
l).v
(25.35)
The
quantity atomic core.

If
()
(Z
represents the polarization of the

Nonpenetrating
orbit
the atomic core as an elastic dipole, we can set

p
we regard
= p,
(25.36)
where
is
the polarizability of the
atom, and
&==';
(25.37)
Fig. 25 3. *Nonpenetrating" and "penetrating" orbits in alkali metal atoms
is the electric field produced by the outer electron at the center of the atomic core. Thus we can obtain the following expression for the 6 potential energy associated with the polarization
:
(25.38)
Since the polarization can be regarded as a perturbation in this problem, we have the following expression for the polarization energy:
Ep
=J
19<Na
+
),
= - ^ (1).
089(K+),
1.50(Rb
+
),
(25.39)
The coefficient of polarization P> is usually determined from semiempmcal formulas numerical values for alkali metal ions (cores) are as follows (in units of 10"" cm 3 ):
003(Li+),
260(Cs
+
)
410

Since, according to (13.29a),
1
~~
/(/+0
3
:
2i*
relation (25.39)
may
be reduced to
A
= "~"î?'
(25.40)
where
4ai
/_
__
1)
(/
'
1)
(/
+ -|)
(25.41)
We
upon
can, therefore, find the total energy, which in this case depends both / and n (for the time being, we are neglecting spin
corrections)
Substituting expression (25.40) for A
and using the relation
we
obtain
since
1
/,
M'
28
we
Introducing the "effective principal quantum get
number"
n eff
411
It should be noted that Eq. (25.41) cannot be used for s states, since the coefficient Sj becomes infinite when / 0. This happens because polarization forces can be introduced meaningfully only in the case where the outer electron is at a sufficiently great distance from the atomic core. For s terms the wave function does not vanish even when r [see (25.34)]. The influence of the inner electrons on the s orbits (penetrating) is due mainly to the "smearing" of the electronic cloud of the atomic core. In general, the additional energy due to the "smearing" of the electrons over the volume of the atomic core is given by
(25.44)
where Vvo i
is the difference between the potential energy produced by the electrons of the atomic core, taking into account their distribution over a finite volume, and the potential energy produced by an equivalent charge concentrated at the center. To estimate the order of magnitude of the correction S for s electrons of the inner shells terms, let us assume that the Z fill uniformly a volume of radius R. We then have
1
Replacing the wave function for the

[see (25.34)], additional energy
we
s states by its value at the origin find the following approximate expression for the 7 of the s terms
:
<>
where
= --5
a*
'
This expression no longer diverges.

a rule, the correction S for penetrating orbits is taken into account in the following It is assumed that the orbit consists of two parts: an outer part and an inner one J Z a ( aâli In the outer part of the orbit the electron is acted upon by a point charge c metals Z = 1), in the inner orbit, the electron experiences the effect of point charge
7
As
manner
which should be greater than

reduced.
It is
difficult to
CQ^, because the screening effect of the inner electrons is determine the charge Z it theoretically, and it is usually regarded
eQZa
&
l
as an empirical parameter. We thus obtain the following equation for

s,
-z,-
<
2S43 >
where a
is the
maximum distance
/.
(that is, the radius of the atomic core)
from the nucleus to the electron in the inner ellipsoid It is easily shown that the quantity 8j decreases
6^
rapidly with increasing
particularly good value of
is
obtained for s terms.
412
We may note here that according to the the radius of an atom [see (25.28)] is
#==^r,
where the
coefficient
T
>
Thomas- Fermi model,
(25.48)
charge inside the atom,

"penetrating"
s
which characterizes the distribution of the order of unity. Consequently, for the total energy of an electron in the case of
is of the
orbits,
we again obtain an equation
of the
same
form as
(25.42) (25.49)
"eff
8 but now 8 is given by (25.47). order to analyze the difference between the corrections & for "penetrating" and 'nonpenetrating" orbits, we shall take as an example the Li atom. In this atom the p orbit (I l)isnonpenetrating. Equation (25.41) '2). According gives the value 8^0.04 for the lowest state (n to (25.47), the corresponding expression for 8^ for penetrating s orbits must be one order of magnitude greater. It should be noted that for increasing n and 1= const, the eccentricity approaches unity; that is, the elliptical "orbits" become more and more elongated:
where
In
/i
eff
(25.50)
heavy nuclei, therefore, there will also be penetrating orbits as well as those with larger values of/. This is re/ = flected in the corrections 8 for the alkali metal spectra, whose values are given in Table 25.2.
In
with
Table 25.2
Corrections
&
for the spectra of alkali metals
The values
of
for penetrating orbits are indicated by
an asterisk
413
It follows from Eq. (25.49) and also from the Table 25.2 that, for a given n , the largest downward shift (that is, in the direction of decreasing energy) associated with the "smearing" of the inner electrons over a finite volume occurs in states with the lowest /. Q In other words, the largest shift occurs for s terms. The hydrogen atom alone has no penetrating orbits. For the Li atom (Z 3; that is, the next element after hydrogen in the first group of the periodic table), the only penetrating orbit is the outer s orbit. In the next alkali metal, namely, the Na atom (Z=ll), the s and p orbits are penetrating.
E.
FUNDAMENTAL SERIES
The energy levels of the hydrogen atom without relativistic corrections are given by the well-known relation
~^ =
From
this
(25.51)
we can
find the values of the spectral
terms
(IS)
=/?,
(3s)
= (3p) = (3d) = =
-
It follows that the states of hydrogen are degenerate with respect A schematic diagram of the energy levels in the to both / and m hydrogen atom is given in Fig. 25.4. In the Li atom, the energy levels of the K shell (n== 1) are filled (see Fig. 25.2), and therefore it is the L shell which is the outer one. Table 25.2 shows that the K shell exerts the strongest influence on s terms, and for the corresponding terms we have
.
__
(
+ o 588)*"
This shift is so large that it was difficult to determine experimentally whether it belonged to n states or n--\ states. In order
In Chapter 21, when the finite size of the nucleus was taken into account, it was also found that the largest shift occurs for s terms. In that case, however, it was the positive charge that was spread over a finite volume, and therefore the terms were shifted upwards and not downwards
414
make the term notation resemble that of hydrogen, spectros10 \ state copists have originally attributed the shift to the
to
n~
(25.52a)
where n*
/i
and
s=
S5
= 0.588
in
for the original notation (n*s)
We shall use an asterisk order to distinguish it from the

.
correct one
(ns)
Fig. 25-4. Energy-level diagram of monovalent atoms The potential is usually measured (in fu) from the lowest level upwards Here, however, we wish to compare the energy levels of different atoms and, therefore, we have taken the potential at infinity as the
zero level
The shift of the other terms of lithium (/=!, corresponding terms of hydrogen is negligible
2) relative to the
R
(
<
number n in lithium assumes the values n - 2, 3, 4 (the term occupied by two electrons and forms an inner shell), than the quantum number n* takes the values n* - l 2, 3.
If
10
the principal quantum
1 is
415
where
p=-ap =-o.04i, = - 0.002. d=

a rf
The shell to which a shift belonged could be, therefore, uniquely determined and in the old notation the p, d, and other terms were placed in shells which were later shown to be theoretically correct
(Z= 11), the inner (filled) shells are the K and L shells (see Fig. 25.2). As can be seen from Table 25.2, the inner shells of Na have a pronounced influence on both the s terms and the p terms. The original notation which was used for the s terms was
(ns)
see Fig* 25.4). ; (that is, n* In the next alkali metal, Na
=n
= (n*s) = _ n
(/l
p
-fi*-+ 07627)*
li373)
(25.53)
where n*=n S5 0.627. Thus, spectroscopists 2 2, and s= had originally assumed values of the principal quantum number for the s terms of sodium which were low by two units. 11 The original notation that was used for the p terms was
(25.54)
+ =
1 0.117. Thus, the principal quantum 1, 8^ for the p terms was reduced by one unit. The corrections for the states d f, were negligible, and these terms had been . . to shells that were later obtained from the theory. the assigned
where
n*=n
p=
.
number
The energy-level diagram of Na is given in Fig. 25.4. The spectral series of alkali metals are as follows. 1. The principal series. The variable term is the p (principal) term. The spectral frequencies in this series are given by
which means
for H:(ls)-(/zp),
for
Li: (25)
(35)
(np),
(np).
(25.55)
for Na
At present, spectroscopists have also adopted a notation which follows from theoretical calculations This is the notation used in Fig. 25 4
11
416

is the s (sharp)
2. The sharp series. The variable term The spectral frequencies are given by
term.
>
= (2*p) -("*).
which means
for
H:(2/?)
(/is), (/is),
(/is).
for Li:(2p) for Na:(3p)
(25.56)
3. The diffuse series. The variable term is the The spectral frequencies are given by
G>
d (diffuse)
term.
= (2*p)
co
(n*d).
(25.57)
4.
The fundamental series.
= (3*d)
(/iV).
(25.58)
The variable term is the / (fundamental) term. These series take into account the selection rule
names of the multiplet structure.

The
series partly reflect the character of their
F.
MULTIPLET STRUCTURE OF THE SPECTRAL LINES
Just as in the hydrogen atom, the multiplet structure of the spectral lines of alkali metals is due to the spin and relativistic effects. To find the splitting of the terms, let us use a relation including both the relativistic and spin-orbit corrections for a
hydrogen-like atom [see (20.18)]

LEntJ ~~ _RZ**l
*
^ \7F7T")'
n
3\
(2559) S Sy ^
'
>
where e*Jfic= 1/137 is the fine- structure constant. To account for the effect of electrons from the inner shells of
a.
al-
kali metals,
we simply replace Z by a certain
effective value Z eff
417
For "nonpenetrating"
the Z
1
'
orbits,
we may
set Z e ff
= 1, because all
electrons will screen the positive charge of the nucleus. For 'penetrating' orbits the best value of Z e a is chosen from a comparison with experiment.
'
du>4
iT
dWjUI
2s
Principal series, converging doublets
Sharp series, equldistant doublets
Fig 25.5. Spectra of alkali metals
Since the total angular the values
momentum quantum number
assumes
for
/=o
and
for
that all spectral terms of alkali metals should be doublets, except for the s term, where there is no splitting. In order to find the magnitude of the splitting, let us calculate the value of the spectral terms for two cases? first, when the spin and the orbital angular momentum are parallel
we may conclude
(25.61)
and, second,
when they are

A n
antiparallel
yi-Vj_ " ~~
^Z
eff
In
\i
3\
4
(25.62)
)
/i*
418

splitting of the
12
The
terms
is
equal to the difference between (25.62)
and (25.61)
1)'
(25.63)
shall use this equation to explain the doublet splitting of the principal series, that is, transitions originating from the p levels. 1 we find Setting /
We
(25.64)
now that the decrease in splitting will be inversely proto the cube of the principal quantum number n (see Fig. portional 25.5); that is, the spectral of the mamseries are narrowing doublets. In the sharp series, the initial level is a doublet and the final level
It is
evident
(with variable s) is a singlet (see Fig. 25.5). The distance between the doublets in the sharp series is constant (equidistant doublets)
= Aw = =
3
.
.
Act),,
~~~
16
In particular,
for
2 (the and n* Li we have Z eff 1, p orbit is "nonpenetrating"), so that this constant splitting is equal to
Fig. 25.6. Diagram of the spectral terms of the sodium atom including the fine structure
the
spectrum of
A(i) z
= /â
-yg-
A diagram
of the multiplet structure of the principal and sharp series of sodium (Z~ 11) is given in Fig. 25.6. The diffuse series does not obey such an explicit law for the splitting of spectral lines. Each line will be split into three rather than four levels (see Fig. 25.7), since in this case in addition to it is also necessary to take into the selection rules A/ itl
Several other formulas have also been proposed for the doublet splitting For example, on the basis of the quasi-classical picture of penetrating orbits, Lande proposed a formula ~ n - S and Z* where 7, a is the total charge of replacing in (25 63) n by n eff ff by Z\ zj the ion and Z t is the effective charge of the nucleus in the inner region [see Eq. (25.43)] into which the orbit penetrates. For further details we refer the reader to E. Condon and G. Shortley, Theory of Atomic Spectra, New York: Cambridge University Press,1958.
,
12

account the selection rule for the total angular
419
momentum quantum
number
The spectral lines of the fundamental series are also split into three components.
Fig. 25.7.
Splitting
of
the
diffuse
series
splitting of the spectral lines of monovalent atoms be explained only by taking into account the spin properties of electrons. We have already and repeatedly stated that only the half- integral quantum numbers (which characterize the spin) can lead to doublet splitting of the terms (as in the Stern-Gerlach
The multiplet
may
experiments).
Chapter 26
Mendeleyev's Periodic System of Elements

A.
X-RAY SPECTRA OF ATOMS
X-ray spectra provide important information on the structure of the inner shells of atoms and are therefore useful in studying the sequence in which the shells are filled by electrons. We recall that x-rays are emitted when a beam of fast electrons strikes the plate of a cathode-ray tube (see Fig. 26.1). An analysis of the emerging x-rays shows that they consist of two different types of spectra a continuous and a line spectrum. The continuous x-ray spectrum arises as a result of the deceleration of electrons when
they strike the target. The continuous spectrum is therefore a type of bremsstrahlung. If the deceleration of electrons is equal to w(w<^0), then according to classical electrodynamics, the energy radiated by the electrons per unit time is given by the relation
~~
N~ 7T~c r
A characteristic feature of the continuous spectrum is that it cut off at short wavelengths. The wavelength A min at which this occurs decreases as the potential difference between the cathode and the anode increases; the cutoff wavelength X min can be determined from the law of conservation of energy
is
m O&
where V==e^l)
target,
<t>
is
ftu>
cathode,
is the electron energy before colliding with the the potential difference between the anode and the is the energy of the emitted bremsstrahlung photon,
and
1
-
is
the kinetic
energy of the electron after colliding with
the target.
Introducing the wavelength x=s=
instead of the frequency
o>,
we
find that
x=
ch
MENDELEYEV'S PERIODIC SYSTEM
421
From
in the
the last equation
it is evident that X can vary from / = oo an when electron does not lose any energy during case
(as
the
collision with the target (y
mtfJ*
V)
to a certain
minimum
value
min
*-
i/
(26.1)
as in the case when an electron loses all
its
energy in the collision
Fig. 26.1. Schematic diagram of an x-ray tube. C - cathode, T - target
of the radiation intensity on ^ is plotted in of the energy of the primary electrons (25 values for two 26.2 Fig. and 50 kev). It is clearly seen that an increase in <t> corresponds to a decrease of the cutoff wave50 kev length X mm ,in accordance with Eq. w n (26.1). This equation was successfully used in the determination of a more | 9 accurate value of Planck's constant j> h than was given by the Wien and the g
,
The dependence
Stefan-Boltzmann laws (see Chapter

!)
U
g
target
0* 08 06 f,Q material, a line spectrum, Wavelength, A which is known as the characteristic spectrum, is superposed on the conFig. 26.2. Short wavelength limtinuous spectrum. The line spectrum of the continuous x-ray its characterizes material of which the spectrum. target is made (or rather, the structure of the inner shells of the target material) in the same way as the optical spectrum characterizes the structure of the outer shells of the atom. For example, in the case of a rhodium target, a line
l
When the energy of the electrons incident on the target exceeds a certain critical value, determined by the
flj
422
spectrum begins to be superposed on the continuous spectrum at an energy of 31.8 kev. The properties of the line spectrum are identical for all chemical compounds of a given element. The characteristic x-ray spectrum and the optical spectrum are different in this respect, since the latter depends on whether the substance occurs in atomic or molecular state (for example, the optical spectra of atomic oxygen, the Og molecule, and the HsO molecule are completely different). This can be readily understood since only electrons of the outer orbits participate in chemical bonding. The spectral lines of the characteristic x-ray spectrum form regular sequences or series, just like the optical lines of atoms. These series are designated by the capital letters K L, M, N,
9
the
series has the shortest wavelength, followed by the L series, and so on.
B.
CHARACTERISTIC SPECTRA OF ATOMS AND THE STRUCTURE OF THEIR INNER SHELLS
The mechanism which is responsible for the characteristic x-ray spectrum of an element was first explained by Kossel (1914). Let us suppose that an electron which is incident on a target removes an electron from, for example, the K shell of the target atom, and thus leaves a vacant site in the /( shell (see Fig. 26.3). An electron may jump from the L, M, A/, ... shells to this vacant site, giving rise to x-ray lines (denoted by /< a Kp, K >...) The characteristic x-ray spectrum is thus formed
,
r
as a result of transitions of electrons
from one inner shell to another. Since

the bonding energy of electrons moving in the inner orbits is much greater than that of the outer electrons. Electrons of much greater r e (several tens of kev) are required for production of the characteristic x-ray spectrum than for the excitation of optical spectra (where several tens of ev are sufficient).
26.3. schemata diagram of origm of a characteristic x-ray spectrum according to KOSsei: -electrons. The dashed hne
F, fr the
represents the ejection of electron from the K shell
an
methodg
can be uged
lft
constructing a theory of the complex atom, while taking into account the interaction of atomic electrons. In the first method the main potential is taken to be the potential of the nucleus when it is completely screened by the inner electrons. We used this method in constructing the theory of the optical spectra of alkali metals. The potential was determined

by the nuclear charge
inner orbitals
[
423
(Ze
(Z
I)e
].
and the charge of the electrons in the The total potential was equal to
$ = ?i(Z
As
(Z
!))
(26.2)
the perturbation potential we selected an additional potential which took into account the polarization and the space distribution of the electron cloud. This method is particularly suitable for outer electrons, as, for example, in atoms of alkali metals. In studying the motion of electrons in the inner shells, it is convenient to use the potential of the nucleus
=^
(26.3)
as the main term in the expression for the potential, and to regard the additional potential produced by the electron shell as a correction. In this case, the presence of electron shells will result in a screening (an effective decrease) of the nuclear charge ZeQ by the amount S e and the total potential will be
rt
,
z-s
t)*Q.
(26.4)
For example, in the investigation of helium-like atoms, it was shown that the interaction of the electrons in the /< shell reduces the effective nuclear charge [see (25.35)], which then becomes
Z'
->
~ Thus
.
the quantity s n is equal to
tt
/ 16
in this case.
/ .
The screening constant S n must be a function of both n and becomes larger as n increases since there is a greater number of electrons screening the nucleus. It also becomes larger (but more slowly) as increases, since the orbits become less and less penetrating and the effective charge decreases somewhat on the average. In the first approximation it may be assumed that the
It
/
screening constant
is
independent of
/.
The potential (26.4) gives the same formula for the spectral terms as was obtained for a hydrogen-like atom, except that the Sn quantity Z is replaced by Z
:
E.
= -=3p.
(26.5)
From
(26.5)
we can obtain an expression
for the frequency of the
K, line
(26.6)
424
It
follows that the frequency of the x-ray spectral lines increases monotonically as a function of the atomic number Z. This was
first deduced by Mosley (1914), from an analysis of empirical data; he wrote this relationship in a somewhat different form
(26.7)
This formula can be obtained from (26.6) if it is assumed that the S1 S. screening constant for K, and L shells is the same, i.e., S We know, however, that this is not quite correct, and, therefore, in studying x-ray spectra, just as for optical spectra, we should express the frequencies as differences of spectral terms. In accordance with (26.5), the spectral terms may be represented in
l
= =
the
form
(26.8)
This relation
graphically.
R terms
is called Moseley's law; it is usually analyzed Ascribing different values to the principal quantum
number
for
L terms
(n = 2)
n,
we get
(/i
K terms
(see 1)
Fig. 26.4):
ff
35
(26.9a)
30
for L
25
terms
(n
2)
20
15
M terms
(n - 3)
R
for
(26.9b)
(/z
terms
=
(26.9c)
10
5
fO 2030405060708090 Z
Investigations
of
the
experimental
curves
Fig. 26.4. Moseley diagran
I/
-^/(Z)
1
have given the
following average values for the It has also been constants: 5, 3.5, S 3 =10.5. Sj^l, screening established the x-ray spectra change monotonically with increasing Z and that no periodic regularities are observed. This represents a further difference between x-ray spectra and optical spectra, in which periodicity exists (see Section C).
l See A. Somnu'rfeld, Chapters 4 and 5.
Atombau nnd Spcctrallinien
lr
Vieweg.,
Braunschweig, 1951,
425
Thus, periodic properties appear only in valence electrons, and not in inner electrons. The study of x-ray spectra made it possible to prove definitely that the atomic number Z, introduced by Mendeleyev, is determined by the charge of the nucleus, and not by its mass. It turned out that Mendeleyev had correctly arranged the elements Co Ni, Ar K, Te I in a sequence which was not the same as the order in which their atomic weight increased. There were also doubts concerning the correct sequence of the rare earths (elements from Z = 58 to 71), whose chemical properties are very similar. Moseley's law made it possible to verify their arrangement in the periodic system. In addition, the study of x-ray spectra made it possible to determine the filling of the inner shells of the ferromagnetic metals and lanthanides where the Moseley curve is discontinuous.
,
C.
MULTIPLET STRUCTURE OF X-RAY SPECTRA

t
X-ray spectral terms are primarily determined by the quantum and /, from which the quantum numbers n electron has been removed, leaving a "hole." As a rule, 7.5 coupling occurs for outer electrons, whereas //
state characterized by
9
coupling occurs for inner electrons (in sufficiently heavy atoms). atoms ;/ coupling begins to play an important role for outer electrons as well. Since the inner electrons are relatively close to the nucleus, they are mainly under the influence of the field of the nucleus , and therefore, their energy states are close to those of a hydrogen-like atom. We may, therefore, take the following formula as a starting
In the heaviest
point:
+T
where the
fine structure constant is
we must make
account for the screening of neighboring electrons, the substitution Z-+Z S nl in (26.10). We assume in this approximation that the screening constant depends not only on n, but also on /. Then for the x-ray spectra we obtain
In
to
J!
order
.i
__ 1
4
+2
426
Taking the approximate square root we find a generalization of Moseley's law to the case that includes relativistic and spin
effects
"
term
This formula shows that in Moseley's curves there appears a Z A in addition to the term which is proportional to Z . The influence of this additional term becomes marked only at large values of Z. This conclusion is in agreement with experimental
data. In addition, Eq. (26.11) explains the multiplet structure of x-ray terms. We note, first of all, that there is no splitting of the K 0, y term, since only one state (lsi/ a ) is possible (/i=l, / */*)
For L terms, we have three components: Li~2si/

/='/*).
(n
= =
2,
0,
To make Eq. (26.11) agree with the experimental data for screening constants, we must set
We
obtain the following equations for the corresponding terms:

'
/ *7
o ^/
r T1O (Z
(Z
o\ 3)
*i
--r
" /r 4 (LI
v
:
terms)
"T^ + iV
'
"~
4)3
T
j
Ln terms ) ;
26 - 12 )
-i/
72
~^r~
"/,
_ ^ 4| r
is
'
TK 10
(Z x
4) '
-r-
v 4 (Lni
terms) '
These relations are represented graphically in Fig. 26.5. In exactly the same way, it is easily shown that the contain five components
terms
Af,v(3d Vl ),
where the screening constants are S 13. 10, S 8,5, 5,^ There are seven components for the N terms. The parallel doublet L\ and L\\ [see (26.12)], which is due to the screening of the nucleus by the electrons is known as an irregular doublet, whereas the diverging doublet L\ and L\\\ is known as a regular doublet. The reasons for the adoption of this terminology
;iJ
arf
MENDELEYEV'S PERIODIC SYSTEM OF ELEMENTS
427
go back to the first stage of the theory of multiplet splitting of x-ray spectra. It may be recalled that a theory of the fine structure was first constructed by Sommerfeld, starting with a relativistic generalization of the Bohr theory. Sommerfeld' s formula gave a correct value for the splitting of the spectral terms (see Chapter 2), but neglected the spin properties. For x-ray spectra this formula is
where
n^
-f-
Applying this equa-
tion to the analysis of the L terms, it can be shown that these terms split into only two components which correspond to the diverging doublet L\ and in our notation. Thus, only this L\\\
doublet was
feld' s
relativistic
it
explained in theory.
the
L\
SommerAccordregular
L\\
,
ingly,
doublet.
was called The doublet
and
for
which no theoretical explanation was

given for a long time, was called the 55 60 65 70 75 80 85 90 2 irregular doublet. With the advent of the Dirac theory, the irregular doublet Fig. 26.5. Multiplet structure of also found an explanation. Therefore, the x-ray L terms the reasons for the adoption of this 5 " terminology are entirely historical (just as for the normal and "anomalous" Zeeman effect). The correct theory of the multiplet structure of x-ray spectra (and of the anomalous Zeeman effect) was developed much later, when the spin properties of electrons were taken into account.
'
D.
THE DISCOVERY OF MENDELEYEV'S PERIODIC LAW

in
order of increasing various chemical properties tend to recur quasi-periodically. For example, the chemical properties of sodium, potassium and other alkali metals are similar to those of lithium, and the chemical properties of chlorine, bromine, iodine and so on, are similar to those of
atomic
Upon arranging the known elements weight, Mendeleyev discovered
that
fluorine.
Mendeleyev ascribed to each element a number Z (the atomic number) giving its position in the periodic system. Although the increase of Z is for the most part parallel to the increase of the
mass number
instance, with the
( 27
CO
of the elements, there are several exceptions for where the element 28 Ni), B3 I) ( 18 Ar 19 K),
UTe
smaller atomic number has a larger atomic weight.
4E8
REUATIVISTIC QUANTUM MECHANICS
that
Moreover, as we now know, there exists a large number of isotopes, is, atoms having the same Z, but different masses (for example,
1
H',
,D\ J')-
in connection with recent discoveries concerning atomic and nuclear structure. In particular, the study of x-ray spectra and experiments on the scattering of a particles by atoms have established that the atomic number Z characterizes the charge of the nucleus (and the number of electrons in the neutral atom).
The periodic law has acquired particular significance
Sixty-three elements were known when Mendeleyev discovered law (1869). He predicted the existence of ten more and even described the basic chemical and physical elements, properties of three elements which were subsequently discovered,
the periodic
namely, scandium (>i$c), gallium dGa) and germanium ( Ge). The inert (noble) gases were discovered at the end of the 19th century. In Mendeleyev's time, only three elements from the rare earth group (lanthanides) were known: cerium, didymium (a mixture of praseodymium and neodymium) and erbium. At the present time the properties of all fourteen rare-earth elements have been
l)2
investigated. By 1937, ninety-two elements were known, but four of these had not yet been observed. It was later found that these four elements were radioactive and virtually nonexistent in nature. They were produced in the laboratory as a result of nuclear reactions. In 1937, E. Segre produced an element with Z^=43, called technetium, by neutron bombardment of molybdenum. The half-life of <J7 x 10 years. its most stable isotope, 3 Tc , was found to equal 2.6 In 1938, it was first reported that an isotope of the last rareearth element with Z 6l had been produced as a result of deuteron bombardment of neodymium. The half-life of the most stable " is about 20 isotope of this element, G iPm years. In 1940, E. Segre discovered an element with Z-^85 by irradiating bismuth with a particles. He called this element astatine. The half-life of its most stable isotope, ssAt* 10 , is 8.3 hours. A short-lived element with Z~<S7, called francium, was disfl
covered in 1939 by a Frenchwoman, Mile. Perey. This element is 7 produced in the decay of 8 .jAc^ The half-life of its most stable
.
isotope,
And with the development of nuclear physics, it has become possible to produce transuranium elements. These range from neptunium (Z=^93) to lawrencium 2 Thus, the periodic system now consists of 103 elements (Z 103). without any intervening gaps.
Fr 2W f is equal to 22 minutes. finally, we must mention that

7
The largest number of transuranium elements has been discovered by G Seaborg and his students The discovery of lawrencium (Lw) with Z 103 was recently announced.

E.
In
429
FILLING OF THE ELECTRON SHELLS
filled in
quantum mechanics, the levels of the electron shells are accordance with the following rules. (a) There can be at most one electron in each quantum state (Pauli's exclusion principle), and therefore the maximum number of electrons with a given / is equal to 2(21+1) (see Chapter 25). Thus the s, p, d and / subshells can contain no more than 2, 6, 10
and 14 electrons, respectively. (b) Electrons tend to occupy the lowest energy levels. Therewill be filled first, then those with n 2, fore, the shells withn H 3,and so on. The shells would be filled in this way in an ideal scheme in which the wave function of an atomic electron could be calculated on the basis of the assumption that the charge of the nucleus and f electrons are all located at the center. In the charges of the Z this case the energy levels of the remaining electron would be the same as in the hydrogen atom, and therefore they would be degenerate with respect to /. As was shown in the investigation of alkali metals, however, the distribution of the electrons over a finite volume removes the / degeneracy, so that terms with a fixed value
of the principal
quantum number
n (that is, the
terms
in a specific
shell) are arranged in order of increasing /. The s term is therefore filled first, then the p term, and finally, the d term. Moreover, the 4s subshell is located below the 3d subshell (and 5s is below 4rf), while the 6s subshell is below both the 5c/ subshell and the 4/ subshell (similarly, 7s is below 5/). It turns out that the outer shell (in an unexcited atom) can consist only of s and /; subshells. The d and / subshells can be filled when they lie in the first or second inner shell, respectively (the first inner shell is taken to be the shell directly adjacent to the outer shell)? We shall make an attempt to substantiate this by an investigation of the ground-state configuration of the electrons in individual atoms
(see Fig. 26.6).
The
order in which the electron subshells are formed can be remembered most simply
with the help of the following empirical rule: the levels are filled in the order in which the sum of the principal and orbital quantum numbers, n \ /, increases, and levels with the same value of this sum are filled in order of increasing n Since I takes the values
0,
1,
2,
1,
we can
find the rule for filling the terms in
any shell
For example, the
fourth period (see below) will be filled in the order
4s
(rat/-
4),
3d(nt/=5),
4p(n +
5)
and the sixth period
in the order
6s (n
6),
4f(n +
6p(rc
--
7),
Sd(n + l-=7),
7).
430
RELATIVI5T1C QUANTUM MECHANICS
Within the first and second periods of Mendeleyev's system, the order in which the levels are filled conforms to the sequence of levels in the hydrogen atom (the ideal scheme). If this ideal
112-118*
104*- If 2*
m No.103
(32)
-ft/
-7s
-M
-4f(ft)
7 La
Fig. 26.6. Diagram of the filling of energy levels in the periodic system of elements. Only s and p subshells can be present in the
outer shell of an atom.

tt
A d
subshell can be
Ba
-6s
beginning with the first inner shell. An f subshell can be filled only beginning with the second inner shell. The filling of the 3d subshell gives the ferromagnetic elements (Fe, Co, Ni). The filling of the 4/" subshell gives the lanthanides or rare earths (5 8 Ce 7jLu). The
filled
only
filling of the 5f
(
90 Th
103).
The asterisks denote
subshell gives the actinides the atom-
ic
been
numbers of elements which have not yet discovered. The maximum number of
-,,Ar
electrons in a given shell or subshell is indicated in parentheses. Emanation (Em) (the element with Z 86) is also called radon (Rn).
1}K(2)
scheme were applicable
to complex atoms, the 3d subshell would start to be filled beginning with potassium (Z=19). According to the table given in Chapter 25, however 5^ 2,23, for 0,140, and 8 potassium, and therefore the energy of the electrons in the 3d and 4s states will be
>
(3
0.146)*
2,854*'
'
(26.13)
T.77 r
'
It can be seen that E 9d ^>E is and hence the 4s level will be filled before the 3d level. Consequently, the third period will contain only
,
seven elements ( n Na
i8
Ar), just like the
second period.
431
After the 4s subsheli is filled in Ca (Z 20), one might expect that the filling of the subsheli would begin with scandium (Z 21). 28 Ni the Spectroscopic data show, however, that in elements 9 iSc 3d subsheli is filled first. This subsheli becomes filled at the ex-
pense of electrons from 2n Cu to 30 Zn of the 4s subsheli, which then must be refilled after all available states in the 3d subsheli are occupied. Only after that can the refilling of the 4p subsheli start. Thus the fourth period contains 18 elements and consists of the 4s, 3d and 4p subshells (see Fig. 26.6). The fifth period repeats the fourth period (the 5s, 4d, 5p subshells are filled), and thus it also contains 18 elements GvRb 84 Xe). The sixth period contains 32 elements ( 89 Cs 86 Rn), because besides the outer shell, consisting of 6s and 6p states, the first inner subsheli (ten electrons) and the second inner subsheli 4/ (the 14 electrons of the lanthanide or rare- earth elements) will be filled. In exactly the same way the seventh period should repeat the sixth period; that is, it should contain 32 elements (the 7s, 5/, 6d, Ip subshells). So far, however, only 17 elements of this period have been discovered. The so-called actinides, in which the second inner subsheli 5/ is filled ( 90 TH element 103) should have properties similar to those of the lanthanides. The first period therefore contains two elements, the second and third periods eight elements each, the fourth and fifth periods 18 elements each, and the sixth and seventh period 32 elements each (except that the seventh period is incomplete). The order in which the states are filled is illustrated in Fig. 26.6.
F.
APPLICATION OF THE THOMAS- FERMI METHOD TO THE THEORY OF THE PERIODIC SYSTEM OF ELEMENTS
We
shall
now attempt
to treat the
ground-state configuration of the elements
more
rigorously.
In a paper devoted to the statistical theory of the atom ( 1928), Fermi proposed a method, now known as the Thomas-Fermi method (see Chapter 25), to explain the periodic system of elements. With this method, he obtained the minimum values of Z for which s, p, d and / states are filled in atoms. He obtained these values by starting from the
following quasi-classical ideas. In classical theory the angular
momentum
L
of a particle
is related to the
momentum
p by
the expression
r x p.
Consequently,
rt^Ts.
component perpendicular to the position vector r. Obviously, the square p^ of this component of momentum cannot be greater than the square of the maximum momentum, which we shall denote by P. Therefore for a given P and r the possible values of the angular momentum L must satisfy the inequality
is the
where pn
of
momentum
/*>-.
(26.14)
432

of the
It was shown in Chapter 13 that, in the quasi-classical treatment square of the angular momentum must be [see (13.62)]
atom, the
(26.15)
a This relation represents a compromise between the Bohr relation I* and ) (/ b the quantum mechanical relation L* W[ (I 1) . The maximum momentum P is related to the density of the electron gas (electrons in the atom) p by the expression (6.32)
W +
/=a*(3*p
a/
)
(26.16)
The electron density po may be found from the Thorn as- Fermi equation (see Chapter 25), which, as we have already indicated, can be solved only by approximate or numerical methods. A good approximation for p , as follows from a solution of the Thomas- Fermi equation, is provided by the expression [see (25.21)]
where
the coefficient I is found bv the Ritz variational method. Substituting these values for P*and L* into the inequality (26.14),
we
get
(26. 18)
Introducing the
new variable
hr
= x, we find
_
(26.19)
'
x
where
''evident that the right-hand side of (26.19) becomes (r 0) and as x GO. The electrons in the atom / if x lies in the range x\ <.x<. xa for which the inequality (26. 19) is satisfied. Here Xi and x a are roots of the equation
the inequality (26.19)
it
From
is
greater than the left-hand side as A: therefore can have a given value of
>
~
e
I'^Ô x
(26.21)
'
The condition for the appearance of states with a given value of both roots
is the equality of
In this case we should equate not only the two functions themselves, but also their derivatives. Then, in addition to Eq. (26.21), we will have
'
*
(26.22)

These two relations
will be satisfied for
433
that is,
when
Substituting the value for
D from
(26.20),
we
find the value of
at
which electrons with a
given
will first
appear
z = -~r (2/-hD 3
where
0.158.
e
= -r(2/+i)
8
,
(26.23)
= 2.718 ... is
the base of the natural logarithms and the coefficient 7 is equal to

7
numerical solution of the Thomas-Fermi equation gives a very similar value for
This again is a convincing demonstration that the density (26,17) represents a good approximation to the density which is given by a numerical solution of the Thomas-
Fermi
equation.
Equation (26.23) enables us to calculate the Z values at which the s, p9 d m and /states begin to be filled. The results of this calculation are given in Table 26. 1. The first row gives fractional values of Z computed from formula (26.23) with 7T-F =S:0 15 5. The values of Z, calculated with 7 _ 0.158, for which s,/?, d, and /states first appear
are practically identical with those calculated with 7 1 _p = 0.155. The second row gives the nearest greater integral value of Z. The last row of the table gives the empirical values of Z at which the states first appear, and also the symbol of the corresponding elements.
Table 26.1
T F=
Atomic numbers
at
which the s,p,d, and /states
first
appear for a given
From this table it can be seen that this approximate theory is in good agreement with the experimental data. We may note that complete agreement is obtained if the coefficient 7 is taken to be 0. 169 instead of 0. 155. are filled. The It is well known that in light elements (Z 1, 2, 3, 4) only s states 5); this is in complete agreement with filling of the p states begins with boron (Z theoretical data. Table 26.1 shows (in spite of the crudeness of the statistical model) that the filling of the 3d subshell does not begin, as might be expected, in potassium 21); that is to say, it does not begin until the 4s subshell 19), but in scandium (Z (Z
completely filled. Similarly, the Thomas-Fermi model explains the "delay" in the 47). According to filling of the 4/ subshell, which might be expected to begin in Ag (Z the theory, however, the filling of the 4/ subshell should be shifted, and should begin only in cerium (Z 4) would 58). It follows from (26.3) that the filling of the 5g subshell (/
is
24 begin in the element with Z The Thorn as- Fermi model accounts for a very important feature of the ground-state configuration of atoms and explains the departure of the filling of the levels from the ideal scheme (the hydrogen scheme) in terms of the ''smearing" of the electron cloud.
1
434
G.
REUATIVISTIC QUANTUM MECHANICS
PERIODICITY IN THE PROPERTIES
OF ELEMENTS
in the properties of the elements can be explained quite naturally in quantum mechanics. It is connected with the periodic nature of the filling of the outer shell, which contains
The periodicity
most eight electrons ($ and p states) and which determines the chemical and optical properties of atoms. All elements, therefore, can be divided into eight groups, depending on the number of elecat
trons in the outer shell.
The elements in Group I (hydrogen and alkali metals) have an outer shell containing a single electron. As a result, the optical terms (except the s term) are doublets and the elements are monovalent, as will be shown below. The elements of Group II the alkaline earth metals (beryllium, magnesium, calcium, etc.)-have two valence electrons; their spectral terms must therefore be singlets and triplets, and the valence is equal to two. The elements of Group III have an outer shell containing three electrons, and therefore the maximum splitting of their optical terms must be equal to four (quartets); their maximum valence is three. On the contrary, the elements of Group VII the halogens (fluorine, chlorine, etc.) lack just one electron to fill the outer shell completely. Therefore, in addition to the maximum (positive) 1 (the valence of seven, they may have a negative valence of number of electrons required to obtain a stable configuration). They exhibit this valence in the so-called ionic compounds (see
Chapter
27).
Finally, in the inert gas group (neon, argon, krypton, etc.) the outer shell is completely filled. We may even say that there is no outer shell in these elements, because the energy bonding these electrons in the outer shell is larger than that bonding all other shells in the molecule and thus it would be more correct to ascribe it to the inner shell. Thus, these elements may be assigned to the "zeroth" group. The elements of the zeroth group do not as a rule enter into any chemical reactions, and thus these elements
are said to be chemically inactive. There are, however, a number of exceptions to the rule that there are eight elements in each period. The first exception is constituted by hydrogen (Z= 1) and helium (Z 2) which form the first period. In this period there are only two elements, and not eight. This is due to the fact that the /< shell does not include p states. Consequently, the properties of these elements are of a dual nature. Because there is only one electron in the outer shell, hydrogen should have the same chemical and optical properties as the alkali metals. Indeed, just like in these elements, the maximum splitting of the spectral terms of hydrogen is two, and its valence is one. Hydrogen, however, also resembles the halogen group in that it lacks just one electron for a complete outer shell, and it can
MENDELEYEV S PERIODIC SYSTEM OF ELEMENTS

r
435
therefore acquire a second electron forming, in the same way as the halogens, a negatively charged ion. Helium resembles the alkaline earth metals of the second group in that it also has two electrons in the outer shell. The spectral terms of both helium and the alkaline earth elements are either
singlets
(spin
0),
properties, helium because its outer
or triplets (spin 1). However, in its chemical is a typical representative of the inert gases, K shell is completely filled; hence it does not
participate in normal chemical reactions.

the
The maximum valence of elements is determined, as a rule, by number of electrons in the outer shell; that is, the valence of atoms varies from unity (for atoms of the first group) to seven (halogen group). There are, however, certain exceptions among the
elements in which the inner shells are filled after the outer shells. It can be seen from the periodic table that there will be two electrons in the outer shell of all elements from scandium (Z 21) to nickel (Z 28), with the exception of chromium, where there is
states to 4p Owing to transitions of electrons from states, however, the maximum valence of scandium (Z 21) is equal to three, and that of manganese (Z 25) is equal to seven. Consequently, it was necessary to place these elements in groups corresponding to their maximum valences (this was correctly done by 27) and Mendeleyev). Mendeleyev placed iron (Z 26) cobalt (Z nickel (Z 28) in a special group (Group VIII). The introduction of this group is justified from the point of view of modern quantum mechanics, since at most eight electrons can occupy the outer shell. In general, however, iron behaves either as a bivalent or trivalent element. All of these elements have similar properties. In particular, they have distinctive ferromagnetic properties caused by uncompensated spins of the 3d electrons in the inner shell. The presence of these states is due to the fact that from the energy standpoint, the 3d state is more favorable during formation of the crystal lattice than the other states in which the spins of the elec-
only one.
M =
trons can be compensated. In the elements following the ferromagnetic, the 3d subs hell is the first to be completely filled; the filling of the levels then continues in the 4s and then the 4/j> subshells. Krypton completes the structure of the M shell (n therefore its optical and chemical 4) properties will be characteristic of the inert gases. As we have already mentioned, the fifth period, which extends from the alkali metal rubidium (Z 37) to the inert gas xenon (Z 54), is a repetition of the fourth period and exhibits no new features.
436
Quantum theory also explains the characteristic properties of the elements in the lanthanide series (the rare-earth elements), which comes immediately after lanthanum and extends from cerium to lutetium (Z The elements in this series are 58) (7 71). formed by consecutive addition of electrons to the deeper 4/ subshell (the second inner N shell), even though the first inner shell (0) and the outer shell (P) are still incompletely filled. Since the chemical properties are determined mainly by the electrons of the
outer shells,
all 14 rare-earth elements are much closer with regard to chemical properties than are the elements in which the
first inner d subshell is filled.
For a long time hafnium (Z 72) was also included in the A theoretical analysis performed by Bohr lanthanide series. showed, however, that there can be at most fourteen elements in this group (the possible number of /-states) and that, therefore, hafnium must be a chemical analog of zirconium. Careful experiments have confirmed this theoretical conclusion. The actinide series in the seventh and last period is analogous to the lanthanide group. Beginning with thorium (Z 90) , the elements of this series are formed by the consecutive addition of electrons to the deep- lying 5/ subshell of the shell, while the 6s, 7s and filled and subshells the 6d subshell is remain 6;; completely partially filled. The actinides include protactinium (Z 91), uranium 92) and also the following artificially produced transuranium (Z
elements: neptunium (Z 93), plutonium (Zr^94) americium (Z=95), californium 97), 98),, einsteinium (Z= 99), fermium (Z (Z (Z=100), mendelevium(Z=101), nobelium (Z=102) and the recently discovered element 103 (lawrencium)? The question of how many elements can be produced by artificial means and experimentally detected, and the question of where the periodic system ends, have not yet been finally answered. It is clear, however, that the periodic system ends because of he instability of nuclei (due mainly to their spontaneous fission).
curium
The discovery and properties

TtiL>
of the new elements are discussed Transuranium Elements, Reading, Mass., Addison- Wesley, 1958
in
G. T. Seaborg,
Chapter 27
The Theory of Simple Molecules

A.
BASIC FORMS OF THE CHEMICAL BOND
The chemical properties and the optical spectrum of an element are determined mainly by the outer electrons of the atom. Therefore, the regularities in the structure of the outer shell, which account for the optical periodicities, also provide
a basis for the construction of a theory of the periodically recurrent chemical properties of the elements. It should be noted that the chemical properties, unlike the optical properties, are not exhibited by isolated atoms, but appear only in the presence of other atoms, with which the atom forms chemical compounds. The inner electrons have almost no influence on chemical processes, since they are much more strongly bound to the nucleus than the outer electrons. Chemical reactions therefore liberate much less bonding than the energy of the inner electrons. In discussing the chemical properties of an atom, we must distinguish between two main types of chemical bonds: ionic (or heteropolar) and atomic (or homopolar). We shall consider both of these types in greater detail.
B.
HETEROPOLAR MOLECULES
Inorganic salts consist of positive and negative ions held together by an electrostatic (Coulomb) attraction to form a molecule. Compounds of this type are called ionic, and their molecules are said to be heteropolar. The ions may be either positive or negative. The sign of the charge on the ion depends, on the one hand, upon the ionization potential, that is, the energy that must be expended in order to remove an electron from the outer shell; and, on the other hand, on the electron affinity, that is, the energy which the atom must acquire to hold an additional electron in the outer shell. Let us assume that a neutral atom with atomic number Z contains N electrons in the inner orbitals and Z a Z N electrons in the outer orbitals. The electrons in the inner orbitals
438
will
completely screen a corresponding fraction of the nuclear charge, but will do so only in the region outside this inner shell (starting at the outer shell). Thus, the Coulomb potential energy holding the electrons in the outer shell is
za e *
However, inside the atomic core the charge will be Z,>Z that the screening of the nuclear charge will not be complete (see Chapter 25). In exactly the same way, the outer electrons will completely screen the remaining part of the nuclear charge (Z e ) only in the region outside the outer shell (that is, only in the case of excited states). While in this case there appear polarization forces proportional to r~ B they are not able to hold an additional electron. In the outer shell itself the charge will be incompletely compensated; for this reason (but provided there are unfilled states in the outer shell), the incompletely screened part of the nuclear charge will hold additional electrons in this shell, thus forming a negative ion. The rule is that the less electrons there are in the outer shell of a neutral atom, the larger the total screening of the nuclear charge in this shell. Therefore, an alkali metal will lose the one electron in its outer shell more readily
rt
;
is,
rt
than
will acquire additional electrons. the dependence of the ionization potential on Z is plotted in Fig. 27.1. It shows a minimum for alkali metals and a maximum for inert gases. This curve reproduces rather faithit
A curve showing
the periodicity exhibited by the number of electrons in the outer shell. In inert gases the ionization potential reaches its largest value; the removal of an electron from the outer shell and its transfer to another atom require a very large expenditure of energy. In addition, no further electrons can be held in the outer shell, which is completely filled. Therefore, inert gases do not participate in ordinary heteropolar compounds (we shall also see that they do not form homopolar bonds), and hence, as a rule, they exist as unassociated atoms. Atoms of alkali and alkaline earth metals readily give up their valence electrons to another atom (the ionization potential is at
fully
For example, in sodium ('/, 11), the ten electrons in the inner shell completely screen ten units of nuclear charge, which leaves only the eleventh unit of nuclear charge to be (partially) screened by the outer electron. In chlorine (Z -= 17) the ten inner electrons completely screen only ten units of nuclear charge, so that the seven electrons of the outer shell must screen the remaining charge, which they can accomplish only partially.
Therefore, a chlorine atom is able to hold an additional electron more easily than sodium, and is thus converted into the negative ion Cl . On the other hand, a sodium atom gives
up
its
outer electron more readily, and
in this
way forms
a positive ion
Na+
THE THEORY OF SIMPLE MOLECULES

its
439
minimum
here), and thus convert to positive ions (for example,
a Na+ ion).
Em
10
20
30
40
50
60
70
SO
90 Z
Fig. 27.1. The dependence of the lomzation potential of a neutral atom on the atomic number
the contrary, atoms in Group VI (including oxygen) and in VII Group (halogens), and also hydrogen (which resembles Group VII with regard to the number of missing electrons), have a higher electron affinity than the other elements (see Table 27.1).
On
The electron affinity of that of inert gases.
sodium
is practically
equal to zero, like
Table 27.1
Electron Affinity
The first successful attempt to construct a theory of the ionic bond was due to Kossel (1916), who made use of the Bohr theory of the atom. Kossel's theory was based on the fact that the eight- electron shells of the inert gases are closed, so that these atoms have zero valence. Positive valence (or valence with respect to hydrogen) is determined by the number of electrons in the outer shell in atoms in which these electrons are readily lost (atoms of Groups I or II). Negative valence (valence with respect to fluorine or twice the valence with respect to oxygen) is determined by the number of electrons which the atom can acquire, that is, the number of vacant states in the outer shell (see Chapter 26).
440
Negative valence is particularly pronounced in elements of Groups VI and VII, although both types of valence may be exhibited by a given element. For example, in the typical heteropolar compound
Fig. 27.2.
Two
Cl).
(Na and
neutral independent atoms The black dots indicate
electrons, the light dot indicates a state which can be occupied by an electron
owing
to the electron affinity of the atom.
chlorine has a valence of -l although other compounds are possible in which chlorine has a valence of +7. An example of the latter is C1 2 O 7 . We do not intend here to develop a complete theory of the chemical bond, and shall restrict ourselves to an examination of one typical ionic molecule, namely, the hetero2 The energy bonding an outer electron in polar molecule Nad. atomic sodium is 5.1 ev. 3 When the valence electron of sodium is transferred to the outer shell of the chlorine [that is, during formation of Na+ and Cl" ions (see Fig. 27.2)], it carries somet
f
HCl
what less bonding since

is
its
affinity for chlorine
(W&
=
(
3.7 ev)
somewhat lower than the ionization energy
of
sodium
W^
However, in the formation of the molecule, this deficiency compensated by the Coulomb energy of attraction between the Na+ and Cl~ ions (see Fig. 27.3). The total energy bonding these atoms in the NaCl molecule is given by the expression
5.1 ev).
is
M^Na
rylOn
This energy has been very carefully determined from experiment
and found to be equal
to:
UP$ dCI
4.2 ev -
Hence for the Coulomb
See
Herzberg, Atomic Spectra and Atomic Structure,
New
York, 1944
The bonding energy of an electron in an atom (or molecule) is equal to the energy that must be expended in order to remove an electron It is, therefore, equal to the negative of the energy holding the electron in the compound (W - - V), that is, it will be a positive
quantity
THEORY OF SIMPLE MOLECULES

bonding energy between the ions we find
n
441
7{f
_|_
u/N a -f lP?Xa
= 5.6 ev:
Coul = Since V
U? Coul =
-~^,
we
obtain a perfectly reasonable value
for the interatomic distance in the
NaCl molecule:
R=
2.5
10~ 8
cm.
(3.7 ev)
Fig. 27 3 The formation of an NaCl molecule from Na^ and Cl~ ions The lonization potential of sodium (5 1 ev) and the affinity of the chlorine atom for an electron (3 7 ev) are indi-
cated
in
energy
The Coulomb bonding parentheses between the ions in the molecule is

5 6 ev.
It should be noted that we have not considered here all the interactions that occur in a heteropolar molecule. In addition to the Coulomb forces of attraction, there will also be repulsive forces; these exceed the Coulomb forces at small distances and prevent the two atoms from approaching closer than the distance R. In any case, this elementary discussion explains the principal physical processes involved in the formation of heteropolar molecules; it also explains, however qualitatively, the ionic structure of their crystal lattice and the dissociation of these molecules into individual ions, a process which occurs in solutions.
C.
THE MOLECULAR HYDROGEN ION
directly
Aside from ionic compounds, there exist molecules formed from the neutral atoms, rather than from ions. The simplest representatives of these molecules are H 2 O 2 and N s These are called homopolar molecules. The formation of homopolar molecules cannot be explained on the basis of classical theory or Bohr's semiclassical theory. These theories are useful only for compounds held together by electrostatic forces such as, for example, ionic compounds. Before discussing the formation of homopolar molecules, let us
,
442
consider a very simple case, namely, the molecular hydrogen ion Ha which consists of two hydrogen nuclei and a single electron. This analysis is important for methodological reasons, because it enables us to express in comparatively simple mathematical form the features of the bond that arises between two hydrogen nuclei owing to the exchange of an electron (exchange forces). The same forces also appear in the homopolar hydrogen molecule H 2 . Let us denote the distance between the two hydrogen nuclei a and d by R and assume that R changes adiabatically; that is, R changes so slowly that it can be regarded as a constant in solving the Schrodinger equation. In a more exact treatment, it is necessary to take into account the vibrations of the nuclei about the
,
equilibrium position (the vibrational spectrum), and the rotation of the nuclei about the center of mass (the rotational spectrum), These questions have been treated in detail in Chapter 12. Suppose r and S represent the distances between the electron and the nuclei a and d respectively. The Schrodinger equation for the ionized hydrogen molecule can then be written as
,
H)p
0.
(27.1)
Here the Hamiltonian
is
H-T--f i
*?+-5[,
(27.2)
and the kinetic energy operator has the form

/
ft
V2
(27.3)
where
v.v
= vi
(V,
= -~'
vi
= -r)
slnce
(27.4)
R may be regarded as a constant in the problem under consideration. We shall restrict ourselves to an investigation of the ground state, and carry out our calculations with the help of perturbation theory. The spin effects can be neglected in the case of a single electron. In the zeroth approximation we assume that the electron is under the influence of either nucleus a or nucleus a' (see Fig. 27.4).
and
The Schrodinger equations describing these two possible unperturbed states are
(27.5)
443
to
Both eigenvalues and eigenfunctions are identical and correspond the Is state of the hydrogen atom. Since one of the wave functions is associated with nucleus a, and the other with nucleus
a',
we may write
(/i=l,
=m=
(?)
0)
=
fy
(r)~~
(2? * 6)
the
where n im fym is the wave function of ground state of the hydrogen atom
[see (13.32)], namely,
Electron 1 is near nucleus Q
(27.7)
*â
R
Electron 1 is near nucleus a'
The
total energy of the system and the zero-order eigenfunction are as follows:
r-\
r**
Fig. 27.4. Diagram of the interof particles in the H 2 molecular ion The solid arrows
action
(27.8)
particles
The uncertainty
that the
is due to the fact in 6 presence of two nuclei leads to a degenerate state of the system. In solving
represent the interaction of the the zeroth approximation The arrows depicted by
dashed
line represent the in-
teraction
corresponding to the
perturbation
(27.1)by perturbation theory,
we must
set
(27.9)
Substituting (27.9) into (27.1) and restricting ourselves to the firstorder quantities, we find
(27.10)
For the solution
$ a (the
electron near nucleus

ii the
*"
a),
--^ represents the
main
interaction,
<|>
and -
perturbation.
On
the contrary, for

p2
the solution
(the electron
is .
near
a'),
the
main interaction is-yl,
and the perturbation
From Eq. (27*10) we can find the additional energy ', and also a relationship between the coefficients C t and C2 in the wave function ^. This can be done because the perturbation energy removes the degeneracy (just as in the case of the helium atom).
444
To solve this problem, we can make use of the theorem which states that the solution of the homogeneous equation [that is, one of the solutions of Eqs. (27.5)] must be orthogonal to the right-hand side of the same equation, that is, Eq. (27.10). Assuming that the electron is near the nucleus a, we can neglect the perturbation
energy
the the
?
on the left-hand side of the equation, since
it
gives us
a second-order term
when multiplied by <]/. Then the solution of homogeneous equation will be the function a and according to theorem we have just stated, it must be orthogonal to the rightfy
hand side of Eq. (27.10)
-|-?;-^)t
|f
d'v
(27.11)
of the ground Here we have used the fact that the wave function state of the hydrogen atom is real. In exactly the same way, assuming that the electron is near the nucleus a', we may neglect the perturbation energy on the left<j/ (I
hand side of Eq. (27.10)
(in this
case the perturbation energy
is
-*
).
Since the solution of the homogeneous equation will now be the function fy a we obtain a second equation for the unknown quantities
,
(27.12)
replaced by /,^, becomes a whereas in the reverse In both cases, the volume element becomes 6 a d*x remains unchanged (d*x d*tf). Therefore, we can reduce (27.12) to a form that is in agreement with (27.11), but with the coefficients C\ and C 2 interchanged
r is
,
When
<|>
substitution
<J>
rt
'
-]-
C,
[,
{'
-j-
t ,d'x
0.
(27.13)
In further transforming Eq. (27.11), function is normalized to unity

<|
we must
not forget that the
1,
(27.14)
but is not orthogonal to the function
t|
(27.15)
445
This follows from the fact that, although the expression S(R) vanishes for R -> oo (there will be no points where the two functions and ty a both differ from zero), it becomes equal to unity for
<l>a
= 0(^ = ^
)-
Let us introduce the notation
and
where K is the Coulomb energy of interaction between the hydrogen atom and the (atomic) hydrogen ion, and A is the exchange energy, which has no classical analog. This energy arises because the
a simultaneously (that is, a and exchange occurs between states a and d ). Formally, the existence of the exchange energy is reflected in the fact that the expression for contains <jv, as well as a . As we shall see in what follows, it is this exchange energy which gives rise to an attractive force between the nuclei. At certain values of the internuclear separation, this force exceeds the force of repulsion, and as a result the Ha molecular hydrogen ion is formed. It is worth noting that this mixed state cannot arise in the Bohr theory, and therefore the existence of an ionized hydrogen molecule can be explained only in terms of quantum mechanics. With the help of (27.14)- (27. 16), we can reduce Eqs. (27.11) and (27.13) to the form
electron can be in both states
ty
ty
t|>
<|>
G!
(E'Sl
obtaining two solutions: a symmetric one (C ==C2 ) and an antisymmetric one (Ci Q). The symmetric solution is
(27.18)
K^
and the antisymmetric solution
is
(ft y<l
-v
\
'
(27.19)
446
The factor
here that
f * &) V (i the symmetric and antisymmetric solutions are already

.
~-
is
the normalization coefficient.
We may
note
mutually orthogonal
as an additional interaction V (/?), the H+ ion into a stable hydrogen molecule. In order to find the specific form of this interaction as a function of R, we must calculate the values of ,S, K and A. For gktf this purpose we use the well-known expansion of the function
/:'
The quantity
which
binds
the
may be regarded H atom and
into a triple
Fourier integral
~
=W
i
r*
ikr
d*k
'
J F+lfcjf
(27.20)
After differentiating this expression with respect to the parameter k; , we can easily write the wave function for the ground state of the hydrogen atom in the form of a Fourier integral
}
where
In
/e
= -.
(27.15), the
accordance with
expression for 5 becomes

(r
=J
tioo (r) tioo
- K) d*x
(27.22)
M (r) and ^ W o (/" /?) by their expressions in replace Fourier integral (27.21). Integrating with respect to volume, we obtain
<|>i
We may
terms
of the
(27.23)
and thus we find the following value for 5:
TI
(V
-f A*)
"7
(27.24)
447
integral in this equation can be calculated with the aid of Eq. (27.20), which should be differentiated three times with respect to the parameter /?; . In order to calculate the additional Coulomb interaction, we shall use the relation
The
= 4*->". =&
1
b\
J jp
where we must set
^=
->
Then, putting &
in (27.20),
we obtain
<
27 - 25 >
form
same way, we may represent the quantity \/R in the integral (27.25). Substituting these quantities into formula (27.16) and integrating over the volume dAx with aid of (27.23), we find
In exactly the
of the
K=
^^(^ + ^_).
=
9
,
(27.26)
Hence, using (27.20) and setting &

for the additional
we obtain an egression
Coulomb energy
off
(27.27)
In calculating the
exchange interaction
=^ Sfj,,
el
J i^-
m (r-
/?)
pjc
(27.28)
we may use
(27.21) for
M (/*) and
*:/
(27.20) for
^L
tioo(r)_ """
where
ô
= ~.
Hence
pi
==~S
4/?V 8
(*
d*k
j fjrrrjfj).
448
Substituting the values found for 5, K and A into the expression for the interaction energy [see (27.18) and (27.19)], we obtain the k): following expression (R 'a
i
1t
..2
L,-S
(27.30)
Here the upper signs

while the lower signs
(-)
refer to the antisymmetric solution V a (+) refer to the symmetric solution. For small values of R (R - a ) t we have
,
=V =
s
a, t
(27.31)
and thus the energy, as we would expect, determined by the Coulomb energy of repulsion between the two nuclei. For large distances (/?>a ) we have
is
(27.32)
Fig. 27.5. Curves of the interaction energy m the molecular hydrogen ion as a function of the distance H between the nuclei (in units of OQ)
for
the antisymmetric solution (-1-) repulsion, whereas the symgives metric solution (-) gives an attraction. The general nature of the variation of V and l/ s as a function of R is depicted in Fig. 27.5 which also shows the experimental data on the interthat
is,
cl
the
symmetric (V) and a symmetric (V ) states
anti-
action
energy.
It
can be seen from
this figure that only the state is realized in practice.
symmetric
Theoretical values obtained from the graph give the equilibrium distance as /v>,^ 2.50a 1.32 A, and therefore the ionization
energy is
D ---
(/?)
= -f 0.0040 $ =
.76 ev.
(27.33)
The corresponding experimental values are

^l.OiiA,
Z)
exP
=2 79
ev
(27.34)
in the
(the zero-point energy of oscillations is not included theoretical and experimental values given here).
449
The discrepancy between the theoretical and the experimental

due to the fact that here, just as for the helium atom, the perturbation energy is commensurable with the energy of the zeroth approximation. Solving this problem by the variational method, using a test function of the form
data
is
where
the effective charge of the nucleus, is taken as the Z', variational parameter, we can obtain values for /? and D which are in considerably better agreement with experiment
/?=,
If
t0 6
A,
D var =2.25
ev.
several parameters are introduced, the variational method gives that are in practically complete agreement with the experimental data. It can be seen that the formation of the molecular hydrogen ion is essentially due to the quantum- mechanical exchange forces, which in the symmetric state give rise to a stable molecule. From the physical point of view, this can be explained as follows. The probability of the electron being in the symmetric state is
results
(27.36)
whereas the probability for the antisymmetric state
is
(27 - 37)
we plot the curves of constant probability density of the electron (see Fig. 27.6), we see that the electron tends to be located at the midpoint of the line joining the two nuclei in the case of the symmetric solution, whereas in the case of the antisymmetric solution the position probability Danishes at this point. Since the electron binds the two nuclei most strongly when it is halfway between them, it is natural to expect that the first solution, and not the second, will lead to the formation of a molecule. Moreover, in the case of the symmetric solution, the curves showing the electron distribution about the nuclei tend to merge when the nuclei approach one another; 4 this provides a graphical characterization of the homopolar bond.
If
For more details see P. Gombas, Theorie und Losungsmethodcn des Mehrte lichenproblems der Wellenmechantk, Basel Birkhauser-Verlag, 1950.
450
nucleus, can linked by other particles besides electrons. We may mention in this connection the u- me sic molecule (HD) , in which the bonding between the hydrogen and deuterium nuclei is brought about by a negative n meson. Alvarez produced a mesic molecule of this type bypassingnegative^ mesons through a bubble chamber. The radius of such a molecule, as calculated from the equation
also
Two hydrogen nuclei, or a hydrogen and a deuterium

be
1
/?
i=2.5
-'
--p,
will be 1/200 the radius of the
molecular hydrogen
Symmetric
state
Antisymmetric
state
Fig. 27.6.
Electron density distribution in the molecular
hydrogen ion.
ion, since the mass of the u, meson is approximately 1/200 as large as the electron mass. Thus, when the nucleus of the hydrogen atom approaches the nucleus of the deuterium atom, they form a common nucleus, namely that of the >IIe* molecule
As a result, an energy of 5.4 Mev is released and carried away by the n meson. Thus the n meson acts almost as a catalyst of the nuclear reaction.
D.
HOMOPOLAR ATOMIC MOLECULES
the
The first successful attempt to give a theoretical explanation of homopolar molecule was made by Heitler and London (1927) with the help of quantum mechanics. In a homopolar molecule,
exchange forces play a fundamental role. In their treatment, Heitler and London used perturbation theory, which does not give completely accurate quantitative results. Although more accurate quantitative results can be obtained by means of the variational method, the Heitler-London theory enables us to bring out in a very simple way the physical features of the homopolar bond. Let r\ and r denote the position vectors of the first and second electrons relative to nucleus a, and r\ and r^ the position vectors
L

of
451
the
electrons
ra
r\~ri~ /?,
=r
wave functions wave functions
relative to nucleus a' (see Fig. 27.2). Then R. In the zeroth approximation we obtain two which are products of the ground- state hydrogen
2
*aa>
*a'
= = ta
,
(î)
'
*'
W) =
(r;) * a (r,)
<t>100
(r,)
<)
1M (r,
/?),
first solution aa corresponds to the case when the first electron near the nucleus a (and the second electron near the nucleus a'), while the second solution ty a a corresponds to the case when the first electron is near the nucleus a', and the second electron is near the nucleus a. Both these possibilities are depicted in Fig. 27.7, where the solid arrows show the atomic bonds, and the dashed lines show the molecular bonds. When the distance between the nuclei tends to infinity (/?->oo) all molecular bonds vanish.
<!>
The
is
a1
Fig. 27.7. Diagram of the interactions H2 molecule The solid lines join the particles whose interaction is
in the
taken into account in the zeroth apThe dashed lines denote interactions which are regarded as perturbations, a and a' are the nuclei of the hydrogen atoms, 1 and 2 are electrons
proximation
Just as in the problem of the molecular hydrogen ion, the main solutions (27.38) that give rise to an additional degeneracy of the system will not be orthogonal 5
,.
^dîî={ J
ôo (r)
4,0,,
(r
/?)
fx
}
=-.
',
(27.39)
bar over a symbol denotes quantities referring to the neutral molecule.
452
where S for the HJ ion is given by the expression (27.34). The Coulomb energy of interaction of the two atoms is given by
= j tf, {f +
where the
-"# -3}**. d^,
(27.40)
first and second parenthetic terms in the integrand correspond to the potential energy of repulsion between the two nuclei and the two electrons, and the third and fourth terms correspond to the potential energy of attraction between the first electron and the nucleus a' and between the second electron and the nucleus a. In exactly the same way, we obtain the exchange energy
-J
<
27 - 41 >
The expressions for K and A can be computed aproximately by the same method as in the theory of the H| ion. For K we obtain a comparatively simple result, and for A a more complicated result, since A is expressed in terms of an integral logarithm (as shown by Sigura). The general character of the solution, however, remains the same as in the theory of the 1 1] ion. In particular, the main forces which hold the two neutral atoms in the molecule are the exchange forces. These forces have a minus sign at comparatively large interatomic distances and correspond to the mutual attraction of the atoms. Just as in the case of the molecular hydrogen ion, we have two solutions. The first solution is symmetric
2(l| -
5)
(27.42)
and the second
is
antisymmetric
*
a
=i7 -
-(*
S)
1/2(1-
(27.43)
curves of V s for a neutral hydrogen molecule is approximately the same as for the H^ molecular ion; therefore, only symmetric solutions will give stable molecules. For the radius corresponding to the equilibrium position [that is, the minimum of the potential energy V s (R) for the symmetric solution], we obtain /? 1.51a 0.80 A. The corresponding value for the dissociation energy is
The general form
of the
a"
= 0.115=3.2 ev.
(27.44)
453
The experimental values of these

fl
quantities are
c *P
= 0.74A,
D e *P
r=4.73ev.
(27.44a)
We
have omitted here the zero-point energy of oscillations, 0.27 ev (see Chapter 12), from both the theoretical and the experimental
values.
E.
SPIN
AND THE SYMMETRY OF STATES
In the HS ion and the H atom, there is only one electron, and its spin leads only to insignificant spin-orbit interactions. On the other hand, there are two electrons in the H configuration, and the spin plays an important role in the theory of this molecule, even though the spin-orbit and the spin-spin interactions give only small corrections. In the hydrogen molecule, just as in the helium atom, the mutual orientation of the spins of the two electrons determines the type of symmetry of the spatial part of the wave function; this is of primary significance in connection with the stability of a molecule. We shall therefore consider more fully the question of the relation of the spin to the symmetry properties of the molecule. The total wave function ^ must contain a spin part in addition
2
the spatial part. When the potential energy of the spin-orbit interaction can be neglected, then, just as in the case of RussellSaunders coupling, the total wave function can be represented by a product of a spatial part and a spin part. For electrons (which obey Fermi statistics) the total wave function must change sign when coordinates and spins are interchanged (that is, the solution must be antisymmetric). We therefore have two possibilities
to
W 1==Ca
(5.,
s,)<l>s (/
r*)>
(27.45)
= Cs(s
lt
s a )<M/-i,
ra ).
(27.46)
It has been shown in Chapter 24 that the antisymmetric spin function C a describes two electrons with antiparallel spins, and therefore function s which is symmetric in the position coordinates, corresponds to a state with total spin 0. In exactly the same way the symmetric spin function C s , as well as the antisymmetric spatial function, describe a state with total spin 1 (the spins of both electrons are parallel). In the case of the hydrogen molecule, the only solutions which results in attraction is that corresponding to s thus, a stable molecule is obtained only in the case in which the electron spins are antiparallel. We shall now proceed to a general analysis of the states of a molecule, using the symmetry properties. In this connection, we
<]>
<f
454
note that in diatomic molecules the field of force possesses axial symmetry with respect to the line passing through the nuclei (the symmetry axis of the molecule). The absolute value of the component of the total orbital angular momentum along this axis
of
symmetry (which, incidentally, must be conserved) is denoted States corresponding to different A are denoted by the by A
.
following letters:
(A
= 0);
II
(A
= 1);
A (A
2); etc.
In addition, each electronic state must be characterized by the total spin 5 of all the electrons in the molecule. For a given value v 2S 1 states are possible. The quantity v, as in the of 5,
[
case of an atom, determines the multiplicity of the energy level. In we have the case where the total spin is equal to zero (S 0) v= 1. For states with S=l, the multiplicity vr=3, etc. The total spin of electrons in a molecule can therefore be characterized by the multiplicity v, and the corresponding term can be denoted
by
the
!
A.
In this notation, the
symmetric solution for

f
the spatial part of
corresponds to the term v= 1), while the antisymmetric solution 0, S S (that is, 8 a 1, v= 3). It is obvious 0, S corresponds to the term v (A that the term corresponds to three states: in two of the states the spin is directed along the symmetry axis of the molecule (in a parallel or antiparallel direction), while in the third state it is perpendicular to the symmetry axis. It should be noted that symmetry plays a very important role in the theory of molecules (particularly in the case of complex molecules). If, for example, we reflect the wave function in a plane passing through the symmetry axis of the molecule (which we take as the z axis), the energy of the molecule must remain 6 unaltered. At the same time, if the component of the orbital angular momentum or of the spin along the symmetry axis differs from zero (A ;A or 5,^0), the rotations which are associated with these angular momenta will be reversed as a result of this
wave function
A=
(that is, the solution ^s)
(t|)
reflection.
For simplicity, we restrict our treatment to the states in which the orbital angular momentum is zero, that is, A (V terms). In the case where the total spin of the electrons also vanishes, that is, S 0, no change of states will occur in the
mirror
reflection.
In the
case of reflection
in the
r* plane, mirror reflection
amounts
to a
replacement of
v by
As
is
well known, the angular
momentum L
xp
is
an axial vector whose direction
system of coordinates, and the opposite direction in a left-handed svstem) The direction of the contour bounding the area and constructed from the vectors r and p remains, however, unaltered in both the right- and left-handed coordinate systems
(it
is a matter of
convention
has one direction
in a right-handed
455
If, however, the spins of both electrons are parallel (5 1), the following cases are possible. (a) The component of spin along the symmetry axis is equal to zero (S;~0). In this case, the rotation characterizing the spin remains unaltered as a result of mirror reflection (see Fig. 27.8, where the initial and reflected spins are characterized by a rotation denoted by // and IT). The corresponding terms are designated by the symbol %?. (b) The component of spin along the axis of symmetry z differs from zero (S s -" 1). In this case, the rotation which we associate with the spin is reversed as a result of mirror reflection (see Fig. 27.8, where the initial spin is characterized by the rotation / and the reflected spin by the rotation /').
n-n.
Fig. 27.8.
in the
Change of the angular momentum on reflection AA'B'B plane, which passes through the axis of symIf
metry z
momentum occurs
(see
/')
the initial rotation characterizing the angular in a plane perpendicular to A.\ B H (see /), the direction of this rotation will be reversed after reflection
If,
reflection,
however, the rotation takes place in the plane of it will be unaltered by reflection (//'-=//)
The terms whose spin changes on reflection are designated by

the symbol ]~.
Therefore, the following terms of the ground state of the hydro-
gen molecule are possible:

+
v;
2'(A=0, 5 = 0), (A = 0, 5=1, S, = 0),
'v-(A
(27.47)
0,
S=l,
5,.=
1),
where the
If
last
term
is obviously twofold degenerate.
the molecule consists of two identical atoms, there will be an additional symmetry property. A diatomic molecule with identical nuclei must have a center of symmetry, in addition to a plane of symmetry. This center of symmetry is the mid-point on the line joining the nuclei. In Fig. 27.8, it is located at the origin of the coordinate system, that is, at the point 2=0. In this symmetry transformation we must change the sign of the coordinates of all
456
the electrons. In particular, under this symmetry transformation the positions of electron 1 and electron 2 will be interchanged in the hydrogen molecule (the coordinates of the nuclei are left uns changed). The symmetric wave function 6 will remain unaltered; that is, it is even (this is denoted by the subscript g) m The antisymmetric function y changes its sign; that is, it is odd (this is denoted by the subscript u). The main possible states of the hydrogen molecule, taking into account both symmetry properties, can be therefore denoted as follows:
IV 3V" 2^
1
<J
liv
JL
V~
and so forth. The importance of symmetry with regard to the formation of a molecule follows also from the fact that the ground state of most diatomic molecules is a state in which the wave function is invariant is the main term under all symmetry transformations. 8 Thus, of the hydrogen molecule. The question of molecular symmetry, however, lies outside the scope of this book. It should be noted that in stable states of the hydrogen molecule the spins of the two electrons are always oppositely directed. At the same time, there are two types of hydrogen molecules parahydrogen and orthohydrogen. These names refer to the orientation of the nuclear spins, and not to the orientation of electron spins. In parahydrogen the spins of the nuclei are antiparallel, while in orthohydrogen they are parallel. Since the number of possible states for two particles with parallel spins is three times larger than in the case of particles with antiparallel spins, ordinary hydrogen at room temperature will consist of an equilibrium mixture of 25% parahydrogen and 75% orthohydrogen. As the temperature is lowered in the presence of a catalyst (for example, charcoal), the percentage of parahydrogen in the equilibrium mixture increases, and is practically 100% at 0K. Parahydrogen produced at low temperatures is extremely stable and can be preserved in such an equilibrium system for a period of several weeks at room temperature. Orthohydrogen has not yet been obtained in pure form. The difference in the thermal conductivities at low temperatures (the thermal conductivity of parahydrogen is larger) is used for determining the composition of the mixture. Similarly, parahydrogen and orthohydrogen have somewhat different dissociation energies and optical properties.
F.
THE VALENCE THEORY
of
We shall now explain the concept of chemical valence in terms quantum mechanics. By chemical valence we mean the ability
As an exception
to this rule, let
us mention, for example, the C>2 molecule, for which
2L
is the
mam
term

of
457
an atom to combine with a specific number of other atoms. As already mentioned, the first success of quantum theory in connection with the chemical properties of atoms was the explanation of heteropolar chemical compounds (Kossers theory); these compounds are formed as a result of the redistribution of electrons in the outer shells of the participating atoms. According to this theory, the numerical value of the valence is determined by the number of electrons which an atom gives up to another atom (positive ionic valence) or acquires from another atom (negative ionic valence). In the formation of a molecule, the electrons in the outer shells of atoms are redistributed so that the valences of the atoms are saturated. Further progress in the investigation of the formation of a molecule was made with the He itler- London theory. This theory succeeded in explaining the formation of the simplest homopolar molecule H 3 which serves as the basis of our present concept of the covalent bond. According to the He itler- London theory, the spins of the valence electrons are mutually compensated in the homopolar hydrogen molecule. Generalizing these results, it is possible to conclude that the formation of homopolar molecules occurs under the condition of mutual compensation of the spins of the valence electrons. Accordingly, this type of valence is
,
also sometimes called the spin valence. Since the saturation of valence bonds amounts to a compensation of the spins of the valence electrons, the chemical valence of atoms is given by the number of electrons with an uncompensated spin present in the outer shell. To illustrate these general principles, let us consider some specific examples. Figure 27. 9 gives the ground-state configurations of several elements of the periodic system. The electron states are shown as boxes, while the electrons are denoted by arrows whose directions correspond to their spin orientations. It is clear from this figure that the configuration of the outer shell of the 2 hydrogen atom (ls') 5 corresponds to a single valence bond. The valence of hydrogen, which is equal to one, is smaller by a factor of one than the multiplicity of its terms, which is two (the multiplicity is designated by the superscript on the left-hand side of the term symbol S). Similarly, the ground state of the helium atom has the con2 figuration (2s' ) It is evident that the multiplicity equals one ('S), while the valence is equal to zero. The boron atom (Z 5) has the ground state (ls'2^2/? ), corre2 sponding to the doublet ( P), and consequently the valence equals The excited state (Is-'2s 2^) corresponding to the quartet one. 4 is also possible; in this state the valence of boron is equal to ( P) 3. Thus the presence of several different valences in the elements of various groups in the periodic system can be explained in
comparatively simple form (see Table 27.2).
458
Although according to experiment the elements of oxygen and halogen groups can have several different valences, the O and F atoms themselves show only the principal valence. This is due to the fact that their multiplicity can be increased only if an electron is transferred to a shell with a larger value of the principal quantum number. This process is unfavorable from the energy standpoint (the d subshell is absent in O and F). On the contrary, for other elements of these groups there is a possibility of transition between states of the same shell, but having different values of /
is
2s
2p
H
He
F]
(1s* 2V?!
Fig 27 9. Diagram showing the filling of the electron shells of several atoms with the spin taken into account Homopolar valence of atoms is denoted by a dot, and ionic valence is denoted by
a
\
(positive) or
(negative) sign
Table 27.2
Multiplicity and
homopolar valence
The bold-face type indicates
the principal valence.
459
According to the configuration given in Fig. 27.9, nitrogen in the ground state (Is 2s*2p') is trivalent (the three electrons in the 2p shell have parallel spins). However, it can also be univalent (antiparallel spins of the two electrons in the 2p subshell) and even pentavalent (Is*2s l 2p 3 ) (the four spin valences, associated with the parallel spins of the electrons in the 2s and 2p subshells, are augmented by a fifth ionic valence associated with the removal of a second electron from the 2s subshell). In this connection we note that the ionic valence of oxygen and fluorine is the same as the spin valence (covalence).
2
It should be emphasized that, in general, it is impossible to divide rigorously the chemical bonds into homopolar and heteropolar ones. The two types of bonds correspond to the limiting cases of the electron density distribution in the incomplete shells. The limiting case of asymmetry in the distribution of the electron density between the atoms corresponds to a heteropolar molecule. Such a molecule has a dipole moment and can be regarded as an ionic structure. The case of a homopolar bond corresponds to identical electron density distributions in the atoms of the molecule. A homopolar molecule has no dipole moment, and it can be considered as a structure formed from two neutral atoms. Quantum theory provides a general method for the explanation of valence forces and deals with both types of bonds (homopolar and heteropolar) in a single scheme. One of the chief merits of the Heitler- London quantum- mechanical theory of the II 2 molecule is that it explains the saturation of homopolar bonds in terms of the saturation of the spins of the electron shells when electrons combine into pairs with antiparallel spins. When a hydrogen atom approaches a H 2 molecule no additional pairs with compensated spins are formed, and hence there is no gain in energy. The H.,
molecule therefore cannot exist. It must be emphasized, however, that the Heitler- London theory was developed only for molecules consisting of atoms in the s state, and therefore an extension of its conclusions to more complex atoms must be of a somewhat qualitative nature. Further development of the theory of homopolar bonds has shown that in the case of complex atoms quantum laws alone are not sufficient. In this case we must also include the influence of the specific properties characteristic of the chemical compounds.
G.
MASERS AND LASERS
The electromagnetic waves radiated by conventional radio transmitters have a comparatively wide frequency band. Everyone who has used a radio receiver knows that transmitting stations which have nearly equal frequencies overlap one another. This is due to the fact that conventional transmitters have insufficient stability and often "drift" into another frequency band.
460
Thus, the stability of even the best existing quartz oscillators is inadequate in a of cases (for example, high-stability oscillators are necessary for accurate determination of distances by means of radar). For this reason, one of the great achieveQ ments of recent times was the development of masers by Townes and coworkers, and independently by Basov and Prokhorov. In this device, quantum transitions between discrete energy levels in a molecule are used as microwave generators. We note first of all that the process of spontaneous emission, which depends on the Einstein coefficient A, has no essential significance in the case of molecular emission, since its intensity is proportional to w 4 and is extremely small in the radiofrequency region (that is, in the range of frequencies that are low compared with light frequencies). As far as induced transitions, which are proportional to the Einstein coefficient B, are concerned, the probability of upward transitions (resonance absorption) and the probability of downward transitions (induced emission) are identical (see Chapter 9); they are proportional to the energy of the field. Therefore, a system (molecule) with two levels will undergo a transition from one level to another under the influence of a sufficiently strong external field containing the resonant frequency, the transitions being accompanied by the emission and absorption of quanta.
number
Kig. 27.10. Structure of the ammonia molecule showing two mirror-symmetric states a and b of the same energy.
external radiation is passed through a substance, it will interact with its molecules resonant absorption as well as induced emission. In accordance with the Boltzmann distribution (the number of particles in an equilibrium system with energy E ~ is proportional to e K/kT) there must be fewer particles in the higher energy states than in the lower states; therefore, absorption dominates induced emission in the thermal equilibrium. The excess of absorbed energy is completely converted into the energy of thermal motion of the molecules, raising the temperature of the gas. In order for the system to amplify and not absorb the radiation incident on It, it is necessary to disturb this thermal equilibrium in such a manner as to produce greater 10 When this is done, occupancy of the higher energy levels than of the lower ones. such a system will generate electromagnetic waves with extremely small line width under the action of resonant radiation. The first masers were produced with ammonia (NH 3 ) molecules. The ammonia molecule consists of one nitrogen atom and three hydrogen atoms and forms a right pyramid. However, from the laws of symmetry, the stable state of the ammonia molecule corresponds not only to the case in which the nitrogen atom is located above the triangle composed of the hydrogen atoms (Fig. 27.10a), but also to the case In which the nitrogen atom is located at the same distance under the triangle (Fig. 27.10bX The stable states a and have equal energies and correspond to the
If
and cause
The term "masor"

Such
a state
is an
acronym formed from the
first letters of
microwave amplifi-
cation by stimulated emission of radiation.
assumed
that the temperature T,
can also be described by the Boltzmann distribution function, if it is which in this case plays the role of a parameter, assumes
negative values

minima
of the potential
461
energy; they are thus separated by a potential barrier, so that not allow transitions from one to the other. In quantum theory, however, the probability of penetration across the potential barrier is different from zero. Consequently, the nitrogen atom, which is above the triangle, may reposition itself below the triangle in the absence of any external action, and may then return to its
classical
theory does
original position. In molecular physics this phenomenon is called inversion. The process of inversion can be explained in terms of the existence of two perfectly identical potential wells, separated by a potential barrier of finite width. It is well known that this leads to a splitting of the spectral lines, in spite of the fact that the energy states in the two potential wells are completely identical to one another. For the ammonia molecule this splitting, expressed in wavelengths, is 1.27 cm for the most intense line; this corresponds to a radiofrequency wave. Another remarkable property of the ammonia molecule is that, with the help of an electric field, it is fairly easy to separate the molecules in the upper and lower energy levels that have been formed as a result of inversion. It is found that when an ammonia molecule is placed in an electric field, the two levels are shifted in different directions, the upper inversion level being shifted upwards (the energy increases) and the lower inversion level downwards. Since any system tends to a state in which its potential energy is minimum, we first pass a beam of ammonia molecules through a carefully evacuated vessel containing an electrostatic field which is produced by a quadrupole capacitor; this field decreases towards the symmetry axis (see Fig. 27.1 IX In this separating system the molecules in the upper level will tend to move to a region where this field is minimum (since the field increases their energy); that is, they will be focused about the axis of the capacitor. Ilie molecules that are in the lower level, however, tend to move to the region of maximum field (that is, to the periphery, where their energy will be minimum), and thus they will be ejected from the beam. After passing through the above separating system, the beam of molecules, which contains a predominance of excited molecules, enters a cavity resonator which, among its rather wide range of radio frequencies (produced by conventional radio frequency methods) also contains the resonant frequency of the molecular transitions. This external radiation causes molecular transitions which are associated with a line of small width (^ 10 sec-i) ata frequency of 2.4 x 10 10 secî (that is, with a relative error no greater than 10- Q ), corresponding to the wavelength of 1.27 cm. This wavelength is determined mainly by the time of traverse of the molecules through the cavity resonator. A molecular generator produced in such a way exhibits extremely stable frequency. The frequency stability is so high that molecular clocks constructed on this principle have an accuracy of approximately 1 sec per 300 years of continuous operation. * l On the basis of this principle, several types of radio receivers working in the micro-
wave region have been constructed. They all use paramagnetic crystals cooled to very low temperatures. These devices, known as quantum paramagnetic amplifiers, greatly increased the sensitivity of radio astronomical and radar equipment. Recently, atoms and molecules have been used as generators and receivers in the visible spectrum. In this case, we speak of lasers (light amplification by stimulated emission of radiation). What is the difference between an ordinary source of light and a laser? In an ordinary source of radiation, for example, the Sun, the spectrum consists of a broad frequency band and the individual incoherent quanta have arbitrary phases and directions. Optical lasers enable us to obtain a monochromatic beam of light of high intensity (when
it is
focused we obtain a radiation density a thousand times greater than that obtainable
The stability of the frequency in masers made it possible to construct two standard These clocks were used to measure the independently operating, synchronized clocks velocity of light in one direction under terrestrial conditions and therefore to check the fundamental conclusions of the special theory of relatn'ty regarding first-order effects
(proportional to the velocity t> /i). We recall that all interference experiments similar to the Michclson experiment involve, in effect, only one standard clock and are used to measure the velocity of light over a
i
closed path Thus, only second-order effects (proportional to these experiments
2
/3
)
could be detected in
462
by focusing

sunlight), great
coherence and extreme sharpness (with telescopic apparatus reaching the Moon would have a diameter of the order of 3 km). The first laser produced by Maiman was a three-level ruby laser. Chromium atoms, which are present in ruby (aluminum oxide) as a slight impurity, absorb light from a power klystron over a wide band in the green and yellow regions. Initially, these atoms give up (without emitting radiation) part of their energy to the crystal lattice and make a transition to a metastable state from which dipole transitions are forbidden, and hence they can be maintained in that state for a comparatively long time (of the order of several milliseconds).
a
beam
of
light
// / -' /
'
'
i
/ \
\Y V
^\ x
Fig 27.11 capacitor
Electrostatic
in the
field
of the
separating system
At a "negative" temperature (when the number of metastable atoms is greater than number of unexcited atoms and tinder the action of stimulated emission, such a system is capable of generating almost instantaneously monochromatic waves with a wavelength of 6943 A (red regionX 'ITiese transitions are induced by the first emitted quanta, which are retained in the ruby (the ruby has the shape of a circular rod, bounded by silvered parallel ends, one of which is semitransparentjL Photons moving parallel to the axis induce the emission of photons having the same frequency and moving in the same direction. Ihis chain process intensifies until a coherent ray finally passes through the semitransparent mirror (after multiple reflection from the ends). The prospects for practical application of lasers are enormous. These devices can be used for the simultaneous transmission of various types of information, the establishment of cosmic communication, the control of chemical reactions induced by thermal excitation and so forth.
the
Part
III
Some Applications
to
Nuclear Physics
Chapter 28
Elastic Scattering of Particles

A.
In the
TIME-DEPENDENT PERTURBATION THEORY
case when the Hamiltonian is an explicit function of time, as a rule, possible to obtain an exact solution of the Schr'odinger equation. If the time-dependent part of the Hamiltonian V (/) is small in comparison with the time- independent part of the Hamiltonian H, time-dependent perturbation theory can be used
it
is
not,
to solve the
It
problem.
obvious that time-dependent perturbation theory can also be applied when the perturbation energy is independent of time. In this case we obtain the same solutions as in the stationary perturbation theory of Chapter 14, Including the perturbation V'(t), we write the Schr'odinger equais
tion as
-f-T- =
The solution
is
+ V 'W]*(0.
(28.1)
of this equation in the zeroth approximation
(V (t)
Q)
a-'V/^
where the wave function independent wave equation
|f ,
(28.2)
ty n
is
a solution of the unperturbed time-
E^ = H^
n9
(28.3)
and satisfies the orthonormality conditions
'M^W
We
(28.2),
<
28 - 4)
shall look for a solution of Eq. (28,1) in the form of Eq. assuming that the coefficients Cn depend on time: C n -+Cn (t). This approximate method of solution was proposed by Dirac and Is known as the method of variation of constants.
466
FUNDAMENTS OF NUCLEAR PHYSICS
Substituting Eq. (28.2) into Eq. (28.1) and using Eq. (28.3), obtain the following equation for the unknown coefficients Cn
:
we
-ie*
(28.5)
expression over
Cn'
integrating the resulting (28.5) by space, and using the orthonormality condition (28.4), we obtain a system of equations for the unknown coefficients
-E.t '"
Multiplying
Eq.
tyl*e*
all
-T C >=2Cne>n>n>'VM>
n
(t),
(28.6)
n"
where
^..
and the matrix element
=E jq^..
(28.7)
Dirac's system of equations (28.6), taken for all values of n' is completely equivalent to the original wave equation. The approximation used in perturbation theory consists in expanding the solution in the form
9
c.=ci.+c;>+ Q+....
where the zero-order coefficients
coefficients
(28.8)
CJ* are independent of V. The the first and higher orders will be proportional to V, V'\ and so forth. Substituting Eq. (28.8) into Eq. (28.6) and retaining only the first-order terms, we find the following system of equations for the coefficients Cn >:
for
Cn
=
n n"
'
(zeroth approximation) ,
(28.9)
*" 6U
v *'*"(*)
first approximation)
and so on.
The
first of the equations
(28.9)
shows
that the
unknown zerois,
order coefficients must be independent of time, that
CJ.= const.
(28.10)
ELASTIC SCATTERING OF PARTICLES
467
Their values are given by the
initial conditions
and characterize
the state of the electron before the perturbation is applied. / the electron is in the state n. Let us assume that at we may set
Then
C,
=.
(28.11)
This expression specifies the initial conditions of the problem. Substituting Eq. (28.11) into Eq. (28,9), we find
.(').
(28.12)
In
quantum
mechanics one generally calculates the transition
unit time. Since the probability of finding a probability particle in the state n' is given by Cn |* we obtain the following expression for the transition probability:
*
|
w per
(28.13)
of
many quantum-mechanical problems
Equations (28.12) and (28.13) are the basis for the investigation in the first-order time-
dependent perturbation theory.
B.
THE CROSS SECTION FOR ELASTIC SCATTERING
We shall now apply the results obtained above to the study of the elastic scattering of an electron. We assume that at the initial time /o the particle is free; that is, it moves uniformly with a momentum p Uk and an energy
*_*_
o?\
At the instant of time / 0, the particle comes within the range of interactions; that is,
moves in a potential V (/). The particle now has a finite probability of making a tranit
sition
to
the
p'
momentum
E=
the
cHK'
(/C'
= nk = ^);
r
state
with the
and energy
that
is,
Fig. 28.1. Scattering of a particle by a cenof force- nk is the momentum of the incident particle, fik is the momentum of
ter
particle is scattered as result of interaction (Fig.
the
scattered particle,
28.1).
scattering,
& is the angle of denotes the scattering center
468
FUNDAMENTALS OF NUCLEAR PHYSICS
The wave functions of the initial and final states, describing the free motion (in the zeroth approximation), are [see (4.62)]
v *
'
where
period, and the momentum components k and are related to the integers n and n'i by means of k't (*=1, 2, 3) the expressions
is
the
*,
'.
= ?.
we
(28.15)
Substituting the wave functions (28.14) into Eq. (28.12), the coefficient CV
find for
where the matrix element of the perturbation energy

V*
is [(V"
=V
(r)]
= ^ eV
(r) d' x,
=k
#.
Hence, the transition probability
is
We
note that the function

f
/0
/9R 1R (28.18)
has
/C'
sharp maximum for sufficiently large values of t and This means that in practice we can restrict ourselves to only those changes of K' for which the following condition is
a
/C->0.
satisfied:
Since the quantity

initial
tQ
bt is
instant,
and the quantity cft(K
spread resulting from scattering, we

these quantities
A.
the time elapsed from the is the energy find a relationship between
f
~K)=&E
(28.19)
469
This relation can be considered as a fourth uncertainty relation; it is usually obtained from the theory of transition processes. The uncertainty in the energy is characteristic for any wave process; its optical analog is the familiar expression for the broadening of spectral lines, resulting from the finite duration of the emission. For sufficiently large values of t(t->oo) the uncertainty in energy tends to zero and Eq. (28.18) becomes a statement of the law of conservation of energy
This explains why this type of scattering is said to be "elastic." Mathematically, this follows from the fact that the function (28.18) is a 8 function at t oo> Integration of the & function leads to the K of To show this, let us consider the integral /(. replacement by
_ J_ J^K^-KL F (K') dK'. * ~K

f
71
(28. 20)
a'
Introducing a
new variable
rf(/C
-/()=?,
we
get
-ctK
If
the function
F has no
singularities,
we
obtain as /->oo
+ 00
On
that
the other hand,
from the
definition of the
function
it
follows
&(/'
K)F(K )dK' = F(K).

r
(28.21)
in
An example of inelastic scattering is provided by bremsstrahlung a scattering process which an electron emits a photon, so that K ^ K.
470

it
is clear that expression (28.18) becomes a 8 function as and, consequently, Eq. (28.17) for the transition probability can be written as
Hence
t->oo,
).
(28.22)
Replacing the summation (28.22) by integration we use, in accordance with Eq. (28.15), the following relation:
8
|
tion,
k'"
dVdQ
= kk'dK'dQ.
(28.23)
The scattering process

which
is
is
usually characterized by a cross sec-
particles N to the incident beam. Obviously the particles that strike this surface per unit time are those located at a distance not exceeding the velocity of the particles v 9 that is, the particles contained in the volume vS v. The number A/ is equal to the number of parti3 cles per unit volume p multiplied by a volume which is to the numerically equal velocity of the particle
equal to the ratio of the probability w to the number of incident per unit time on a unit surface S perpendicular
"==
<
28 - 24 >
With the aid of Eqs. (28.22) - (28.24), we find the following expression for the scattering cross section:
(28.25)
The integrand characterizing the number

incident per solid angle d$(d spherical scattering angles),
tion, is equal to
'
= smbdbdy
known
of scattered particles where & and 9 are the as the differential cross sect
<
28 - 26 >
In particular,
when
the scattering center is spherically symmetric,
we have
where
in Eq.
dQ' is the solid angle associated with the vector r, whereas (28.25) dQ is the solid angle associated with the vector k'.

Integrating the
find
l/ x
471
last
expression over the solid angle dQ' we

9
4*
rs[n*rV(r)dr.
From this it is clear that the elastic scattering is equal to
differential cross section of the
(28.27)
where
x=
and the quantity
|A
r
\
= 2ksin~
(28.27a)
(28.28)
called the scattering amplitude. Equation (28.27), describing the elastic scattering of the particles by a center of force V (r) in first-order perturbation theory, was originally developed by Born; it is therefore called the Born
is
approximation.
We
note that this
problem can be also solved
in the
time-
independent perturbation theory, since the potential energy of interaction is time- independent. To obtain the scattering cross section we have used, however, the time-dependent perturbation theory, the mathematical apparatus of which is comparatively simple but is much more general. In particular, it can be used to solve many problems in modern quantum electrodynamics, taking account of the interaction of electrons with the doubly quantized electromagnetic field (see Chapter 29). The expression for o(0), obtained from the perturbation theory, is applicable only within some definite limits. In the case of shortrange forces (nuclear forces, neutral atom, impenetrable spheres, and so forth), which can be neglected for r /?, where R is a certain effective radius, the magnitude of the cross section must be either less or of the same order as their geometric cross section 2 rc/? (even if these forces create an absolutely impenetrable barrier). For short-range forces, therefore, the range of applicability of the perturbation method is given by
>
o<>/?
Equation
if
2
.
(28.29)
(28. 27) is not applicable to long-range forces (Coulomb the forces) scattering angle is small. This question requires a more detailed analysis (see below).
472
C.
SCATTERING BY THE YUKAWA CENTER OF FORCE

is
The form of the potential energy for the Yukawa Interaction

as follows:
V=-A
where A
is
e ~**
r
.
(28.30)
a constant related to the charge and
is /?==]-KO
the effec-
tive range of these forces. The interaction (28.30) has some important applications in the theory of nuclear forces and, in particular, in meson theory. For nuclear forces, the quantity A is equal to g* , where g determines the strength of the potential. The range of nuclear forces is equal to the Compton wavelength of a pion
l
/?
= -*
<
10
13
cm.
' (28.31) v
In the case of fast-electron scattering (or a -particle scattering) by neutral atoms, the interaction potential given by the ThomasFermi model can be approximated by the expression (28. 30).2 In this case the quantity A Zel, where Z is the atomic number, and the effective radius R of the atom in the Thomas- Fermi model is
equal to
R = -fh*
where
f is
(28.32)
a numerical factor of the order of unity. R *oo, we obtain the Coulomb potential of the nuclear field which, consequently, can be considered as a limiting
Finally, setting
case of Eq, (28.30). Substituting Eq. (28.30) into (28.27) and using the relation
we obtain
the following expression for the differential cross section of elastic scattering:
(28.33)
As has been mentioned in Chapters 25 and 27, a more accurate approximation of the Thomas-Fermi potential is given by the expression (25 22) The results of the two approximations, however, do not differ greatly from one another (which is a consequence of the short-range character of the forces). The approximation (28 30) is more convenient in calculations of scattering.
473
Here, according to Eq. (28.27a),
x*
= 4#
2 sin --
=4
n*
^,-
where p
1.
analysis of Eq. (28.33). of slow when */?<*! for all particles, relatively Scattering scattering angles. In this case Eq. (28.33) shows that a (ft) is independent of the angle ft and becomes equal to
is the momentum of the particle. Two cases should be distinguished in the
(28.34)
of the scattering cross section of the angle ft a characteristic feature of scattering of relatively low- energy particles by a center of short-range forces. 2. In the case of relatively fast-particle scattering, the differential cross section becomes independent of R (the effective range of force) for all scattering angles satisfying the condition x/? 1.
(isotropy)
is
The independence
Equation
(28.33) in this
case reduces to
(28.35)
evident that for angles satisfying this condition the scattering the Yukawa potential is the same as the scattering by a Coulomb by center. Therefore, in the case of fast electrons or *- particle scattering by a neutral atom through comparatively large angles, the influence of atomic electrons is not important and the scattering is determined only by the potential of the nucleus.
It
is
Setting
A=Ze]
and
= -^-sin-|
x
in
Eq. (28.33),
we obtain the
familiar Rutherford equation
Z*eM
' (28.36) x
which was obtained by the classical method in Chapter 2. Equation (28.36) shows that for long-range forces there is a strong dependence of the scattering cross section on the scattering
angle
ft .
However, for any large values of the wave vectors k

can always find small angles
is true:
ft
we
such that the following inequality
-sin<l.
(28.37)
474
In particular, as ft 0, Rutherford's formula gives a divergent value for <*(&); in this case we must take into account the shortrange character of the forces resulting from the screening action of the electron shell. The condition (28.37) now determines the
region of inapplicability of the Rutherford formula.
For
0,
that is, for
forward scattering (x==0), we find from
Eqs. (28.32) and (28.33) the following expression for the differential cross section:
a (0)
d-t-O
= 4 X^
T
s/
flJZ'/i.
(28.38)
The
total
cross section can be obtained from Eq. (28.33)
sinfo/fc
W
ft*
[1+2W(1
4^8D8
j.
~
costyf
, 2839 \4OoOiJ) v
J
L_
(28.29),
Finally, with the aid of Eq. applicability of the perturbation
we can
method for our problem
find the range of in the two
limiting cases considered above
<
28 * 40)
Represent the cross section for scattering of particles by a spherically as a sum of partial cross sections (the sum of the cross sections for waves with a well-defined value of the orbital quantum number / ). Obtain the scattering cross section for the general case and for the Born approximation, nk and velocity Solution. Suppose the incident particle has a momentum p v=zftk/m , directed along the 2 axis. The wave function of the incident particle is of the form of a plane wave
s y mtliu 11 ic
|JU Umti al
A plane wave can be expanded in spherical harmonics (see the asymptotic expression for the Bessel function as r *oo
sm kr
( \
Problem
21.5).
Then, using
'
/-|-
~ = .W-l/l
/~o7U
the incident
wave can be represented as
'= 7 ."'-i-n
-
*(*--?) L_ !/_ P/(cos8)
(28.41)
475
In the presence of the potential energy I/ (r), the asymptotic expression for the wave function of the particle in a centrally symmetric field, in accordance with Eqs. (13.75) and (13.78), should be chosen in the form
oo
n
C/P/(cos8)
where the phase shift & z can be determined from the asymptotic solution of the Schrodinger equation for the radial function in the presence of the potential V (r)
Clearly the scattered wave
is
X -<
01
The unknown coefficients C/ can be determined from the condition that the function fy s must be a diverging spherical wave. Thus the coefficient of the converging wave"
2 '
must be equal
to zero. ITien
d;
Tscat
=
,
The function /(&)
is the scattering
amplitude [see (28.28)], which, according to the
exact theory, is equal to
~
00
(28.43)
The
angle
time
differential cross section characterizing the scattering of particles through an is equal to the ratio of the probability of scattered particles passing per unit through an element of the spherical surface dS
to the number of particles incident per unit time on a unit surface perpendicular to their velocity, that is, perpendicular to the z axis
From
this
we
find the differential cross section
sinfo/d.
mc
Here,
assuming the scattering
field
to
(28.44)
be axially symmetric,
we
set the solid angle
equal to
dQ
27c sin
9dd.
476
Substituting the value obtained for the scattering amplitude and using the orthonormahty property of Legendre polynomials, after integrating over the angles we get
/McostyP^cosOJsini
which gives us the following expression for the
oo
total
cross section:
(2M5)
Z=0
This expression for a is the desired sum of partial cross sections. Let us compare the expression for the scattering amplitude found in the Born approximation [see (28.28)] with the exact expression (28.43JL The comparison shows that the exact expression gives the same result as the Born approximation for small values of the scattering phase angles 8/. Indeed, whenfy 1, expression (28.43) becomes
<
00
< (
cos 8)
'
(28- 45a)
Solving
Eq.
(28.42) for a given
(for a partial
wave) by perturbation methods, we can
show
that
T
o
kr ) dr
(28.46)
Next, using the expansion
where
7
= 2k
sin
the
scattering amplitude (28.45a) can be reduced to the
form
(28.28) found in the
Born
approximation.
^Problem 2 8f ^Determine the cross section for scattering and radius a symmetric potential barrier of height 1/
>
of particles by a spherically
-fVo V v ~~
I
for for
when
the radius a is much less than the de Broghe wavelength of the scattered particles, that is, when ka-' 1. Show that in this case the s wave (wave with / 0) is the main contributor to the
scattering process. Compare the exact solutions with those obtained by perturbation methods. Solution. It may be seen from Eq. (28.46) that at ka 0) is the main 1, the s wave (/ contributor. Solving the problem by perturbation methods (Born approximation), that is, using Eq. (28.28), we find the scattering amplitude
<

Consequently, the cross section in the Born approximation
is
477
equal to
(28.47)
Let us determine the scattering phase shift fy in this simple problem. We restrict and low energies E when we can ourselves to a determination of the phase shift for /
set
/?#< 1. When /
= 0, the wave equations have the form = forr>a = for r<a, XJ

t
*'*/.
where
X.
*.!>,
k*
= Ê,
(0)
1
i'
^^. -/) = /-*.

8 )
Using the boundary conditions X

as
0,
the solution of Eq. (28.48) can be represented
_ ~
sinhx'r
\ sin (kr
for r for
<
a,
Equating the wave functions and their derivatives at the boundary of the region r find in our case of small E (E <^ 1/ )
we
= arc tan \ J- tanh

where
*' fl
-ka^ka
]
(^^
(28-49)
Hence, according
to Eq. (28.45),
we have
= 4n
In the
a'(^-.)
*0
(28.49a)
case
<
(28.50)
we can
set
Then Eq. (28.49) gives an expression for a corresponding to the Born approximation [see (28.47)]. -* oo) the cross section reaches its maximum value Whenxa 1 (that is, whenl/
>
(28.51)
We
note that the condition (28.50), which in this problem can also be written as
2 B < no
is
equivalent to the condition (28.29) for the applicability of perturbation method
478
The
last expression is four times greater than the corresponding cross section for elastic scattering by an impenetrable sphere, calculated according to classical mechanics when i is determined simply by the geometric cross section of the sphere *a\ This discrepancy is due to the appearance of wave properties (diffraction) of the scattered
particles.
'*'
Problem 28.D Determine
*X
b^TSphCPWaTTysymmetric
the cross section for scattering of slow particles potential well:

[
:
(â<
1)
"
V n
for for
r<a,
Indicate what distinguishes the cross section for scattering of particles by a potential well from the cross section for scattering by a potential barrier. Solution. In the exact solution of the problem, Eq. (28.48) should be replaced by
= =
where
for for
r<a,
The phase
shift &
is given
by the following expression [instead of Eq. (28.49)]:

R
= arc tan A tan k'a\

(
ka
(28.54)
V , we again obtain Eq. (28.47), which was derived in firstWhen '0<! 1 and order perturbation theory. In this case the cross sections for scattering by a potential well and by a potential barrier are identical.
The difference appears when
the quantity x<2 fx 2
<
=?
/;
l/
j
becomes comparable
-f
to
or
greater than unity. Thus, in the case of a potential barrier, the quantity
ically
monoton-
approaches
zero
as
V9
increases, whereas in the case of a potential well the
corresponding
I/o
quantity
vanes periodically over
the
range from
to
oo as
increases.
In particular, if the quantity k'a
approaches
we
find the following expression for
the cross section in the region
tan k'a
< ka ^
Because of the smallness

greater than the potential barrier
of the quantity
k*a*=--
a 3, the expression for a is
much
maximum
(^
max
value for the cross section in the case of scattering by a 2 4*fl when Vo co [see (28.51)]). Since the relation k'a
-j of the first level in the spherically
]/"x
a3
^ 8fl8
for ka
<
is actually
equivalent to the condition of

potential well,xa
appearance
symmetric
rt
f>
^^
the cross section
(28.55) corresponds to the case of "resonant" scattering. Subsequent resonance
maxima
of the cross section occur
when *o
--.
,
and so forth.
Note. The scattering phase shift, together with the expression for the cross section, can be accurately determined for a very limited class of problems (scattering by a spherically symmetric potential barrier or potential well, scattering by a potential inversely proportional to the square of the distance, and so forth). However, In the general
case we must use approximate methods. For intermediate values of the potential energy V (for instance, in the case of scattering of charged particles by a Coulomb field the constant

characterizing the potential energy must satisfy the condition -=
479
Z*
nv <^
1)
perturbation
theory (that is, the Born approximation) gives good results. In cases when the Born approximation is no longer applicable, the phase can be determined by other approximate methods, for example, the WKB method [see Eq. (13.78)], the variational method and so forth. In view of the special nature of these methods we shall not discuss them in this 4 It should be noted that many qualitative features, which are chargeneral treatment. acteristic of the scattering of particles by various potentials, are very well illustrated in the scattering by a potential barrier or a potential well.
4 The theory of scattering is given a more extended treatment in N. Mott and G. Massey, The Theory of Atomic Collis ions, New York. Oxford, 1949, L. Schiff, Quantum Mechanics,
New
York- McGraw-Hill, 1955.
Chapter 29
Second Quantization
A.
SECOND QUANTIZATION OF THE SCHRODINGER EQUATION
As an example of second quantization, we shall consider the nonrelativistic Schrodinger equation, and then generalize the results to the case of the Maxwell and Dirac relativistic equations. In Chapter 5, it was shown that the solution of the time-dependent Schrodinger equation in the general case can be represented in the
form
-3"
(29.1)
where the coefficient Cn (t) , which characterizes the probability that a particle is in the state represented by<J> rt includes the timedependent part of the wave function
,
C n (t)
=C
ne
n
.
(29.2)
Here the quantities E n are the energy eigenvalues, and the eigenfunctions ty n of the time- independent Schrodinger equation satisfy the orthonormality condition
M *==
3
8 n'
(29.3)
In transferring from classical quantities to ones, the position .v, the momentum p v , and the
quantum-mechanical remaining quantities
are replaced by their average values
(29.4)
SECOND QUANTIZATION
481
The time variation of these average values cannot be determined from the classical Poisson brackets, which in this case coincide
with Hamilton's canonical equations
_ dH
dt
dPcl
'
and must be calculated from the commutator relation

i
= ^(HF=n?lT).
(29.5)
Here the Hamiltonian
is
equal to
"HSr+^w-
(29.6)
The quantum equation of motion [Eq. (29.5)] can be considered as the fundamental equation describing the quantization of the classical equations of motion. This process is called first quantization. In order to connect p and x by the usual relation
i==
7<
29 ' 7 >
we must assume that in Eq. (29.5) the operators p and x do not commute with each other and obey the commutation relation
p.v-A'p=4'
Thus
the
<
29 ' 8)
relation (29.8), which is basic to the can be regarded as a consequence of the Schrbdinger theory, quantum equation of motion (29.5), and the transition from the classical equation of motion to the Schrcdinger equation is equivalent to a transition from corpuscular concepts to the wave concepts.
In order to include in the theory the corpuscular properties of the de Broglie waves, it is necessary to introduce a number of additional hypotheses (for example, a hypothesis concerning the probabilistic nature of the waves and the meaning of the Cn * as the probability of an electron being in state coefficients
<[>
]
\
commutation
and so forth). The process of second quantization enables us to take into account both the corpuscular and wave properties of the particles.
n
,
The
that in this case
name "second quantization" originated from the fact we quantize the equation which has already been
quantized as a result of the first quantization.
We
note that as a
482
result of second quantization, the coefficients Cn become operators (q numbers), whereas in the Schrbdinger theory they remained ordinary constants (c numbers). In order to carry out the second quantization, it is necessary to find the average value of the energy operator
(29.9)
Substituting (29.1) in place of find
ty(t)
and recalling that
Hty n
rt
<|
we
Taking into account the orthonormality condition Eq.

obtain
(29.3),
we
t).
(29.10)
is
We note that the last expression apparent from the equations
is
independent of
t,
since
it
that the
time factor in the product C*(t)-Cn (t)

C*(t)C n
(t)
is
simply equal
to
unity
=C C
n
Expression (29.10) is not an operator in the Schrbdinger theory, since the coefficients Cn are ordinary numbers (c numbers). In the theory of second quantization, however, these quantities and the Hamiltonian (29.10) should be regarded as operators, that is, as
q
numbers.
,
To find the commutation relation for the coefficients Cn we must substitute C n (t) and C(t) in place of x and pin the quantum equation of motion_ (29.5), and the operator H must be replaced by
its
average value
//.
We
get
=
According
to Eq. (29.11),
(C
n (<)
- C (t)H).
n
(29.12)
SECOND QUANTIZATION
483
is
and therefore from (29.12) we find the following relation, which a fundamental postulate of the second quantization:
- LvCV = IlC n - C
,
fl
.//.
(29. 13)
Substituting
from Eq,
Bn'Cn'
(29.10),
we
obtain
'
=Ê
n
n C n Cn (C*
C n 'CnC n ).
^9. 14)
The
statistics
last equation has two solutions, corresponding to the Bose and Fermi statistics, respectively. The first solution is 1 Cn'Cn
c c* ^n'^rt
,
C n Cn '=Q c* c _* Q nn'* Wi^rt'

t
q (29.15)
This can be easily verified by substituting the expressions
C>
= C>C
'
(29. ID)
into the first and second terms, respectively, on the right-hand side of Eq. (29.14). We then obtain the identity
From Eq. (29.15), operators. Setting
it
follows that the coefficients C n and CJ are
C;C n
Af,
(29.17)
where N
is
the
number
of particles in the state n
we
find
from
Eqs. (29.15)
Cn CH=\+N.
.''
(29.18)
It follows, ^herefore, that these operators do not vanish even in the case where there are no particles present. Indeed, even though o when N we still have in that case CBC* 1. This CiCn Q nonzero value for the combination of coefficients C gives rise to a relationship between the vacuum (the field of virtual particles) and the real particles.
lt
follows from
i/j*
Eq
(29 15) that the
wave functions
=
l/f(r,
t)
and
0(r',
commute, while
and
do not commute
V>(r, t)
0"(r'f
0-
<AV,
00
S0n *(r')
^rn (r)
- S(/-
r)
484
should It is clear from the above equations that the operator C be regarded as the "creation" operator for particles, and C n as the "destruction" operator. If at the initial time there are no
\ particles present, the condition C n Cn signifies that a particle can first appear (owing to the action of operator CJ), and then imn Cn disappear (owing to the action of Cn ). The condition C* plies that in the absence of particles the inverse process, in which a particle is first absorbed and then emitted, cannot occur. In the solutions (29.17) and (29.18), the quantity N must be a is not restricted to any maximum value. It positive integer. Therefore, these commutation relations correspond to the Bose
statistics, which allows any number of particles to be present in one state. The second solution of Eq. (29.14) can be represented as
which
is
easily verified by the direct substitution of Eq. (29.19)
into Eq. (29.14). Setting
C* n Cn
=N
(29.20)
we
find
Cn CJ=l
JV.
(29.21)
Noting that C*Cn and Cn C* cannot be negative, we find that the number of particles in the state n can assume only two values: Af and N=l. Consequently, the Pauli exclusion principle is already included into this solution and the particles obey the Fermi statistics.
In particular, if there are no particles present initially just as in the case of the Bose statistics we have
(/V
= 0), then
(29.22)
C*Cn
0,
Cn C;=L
The equation obtained
in
second quantization describes, there-
fore, a state with a variable (integral) number of particles. Consequently, electrons will be similar to photons not only because their motion is described by a wave equation, but also because electrons, just like photons, can be created and destroyed. Since the creation and annihilation of electrons requires an energy more than twice as great as the rest-mass energy of an electron (since an electron is always created together with a positron), the second quantization of thenonrelativisticSchrodinger equation is of purely methodological significance. In order to consider real wave processes associated with the creation or annihilation of particles, one should extend the method of second
SECOND QUANTIZATION
485
uantization to relativistic equations namely, the wave equation DP photons (Maxwell's equations) and the relativistic wave equation 3r electrons and positrons (Dirac's equation).
B.
QUANTIZATION OF MAXWELL'S EQUATIONS
It is well known that the photon field (electromagnetic field) can e described by a vector potential which satisfies the d'Alemberts quation
0,
(29.23)
nd
is
subjected to the following condition:

V
A = Q.
is a
(29.24)
Since the vector potential he solution of (29.23) as
real quantity,
we can represent
t,
^^
X
-'*')
'
(29.25)
yhere
*
=
,
|*|,
and
**
= j/h
(n t
0,
drl,
:2,
=b3...).
The condition (29.24) o vector x that is,
means
that the vector
is
perpendicular
x-a =0.
(29.26)
The normalization
coefficient
^ I/
-~~
is
determined from
he condition that the energy of the electromagnetic field
(29 - 27)
nust be equal to the
mergies
chf.
sum of the products obtained by multiplying the of the individual particles by the corresponding squared
.mplitudes. Since the Hamiltonian is proportional to the combinations of CC* (the amplitudes of a in this case coefficients of the formC*C )lay the role of the coefficients C), the quantum equation of motion
486
(29.13) for the field of photons permits only a solution corresponding 2 to the Bose statistics. Furthermore, taking into account the condition (29.24), we obtain the following commutation relations:
(29.28)
2, 3 and the amplitudes a and a"V refer to the same which plays the role of the quantum number n [see Eqs. The coefficient Aw expresses the transverse character of
t
where
i.
i'
= l,
vector x
(29.15)].
the photon field
v4,
In
= xM,, =
0.
(29.29)
particular,
if
instant, only the following bilinear
there are no photons present at the initial combination of amplitudes will
be nonvanishing:
a ai
t
= u ,-^.
*
(29.30)
C.
SPONTANEOUS EMISSION
In Chapter 9 we found the Einstein coefficients A for the spontaneous emission by using the correspondence principle. In such treatment, the reasons for transitions of electrons from higher to lower levels remained unexplained (see Chapter 9). Quantum electrodynamics (the name given to the theory that includes second quantization of the electromagnetic field) explains the transitions in terms of the interaction of an electron with the field of vacuum (virtual) photons. As can be seen from Eq. (29.30), there are quadratic combinations of the quantized amplitudes of the electromagnetic field which differ from zero even in the absence of real photons. The time- dependent Schrodinger equation, taking into account the doubly quantized field of photons, can be represented as
-o.
<
29 31 >
-
Neglecting the second-order terms proportional to A*, and taking into account the transverse character of the electromagnetic waves
(29.32)
CC* (this If these amplitudes occur in the Hamiltonian in the combination C*C happens, for example, for Dirac particles), then only the Fermi statistics would hold for the corresponding amplitudes This is easily verified by substituting into the Hamiltonian the different solutions for the Bose statistics [see Eq. (29.16)] and for Fermi statistics [see Eq. (29.19)].
SECOND QUANTIZATION
since
487
V"4
Eq. (29.31) can be reduced to
0,
=
Here
the
0.
(29.33)
unperturbed
Hamiltonian
H =l/4-y7-P*
is
is
time-
independent, and the operator of the perturbation energy
Let us assume that the electron is initially in the state n under the influence of virtual photons 3
Then
"'.
the electron can
In
(29 - 34)
jump
into the state n
accordance with Eq. (28*12), we obtain the following expression for the coefficient Cn *(t), characterizing this transition:
'.,.,.,
co
,
A (29.35)
fl/l
where
the
matrix element
n'n
=J
tS'
(29.36)
and a>=rc*
is
the frequency of the emitted photons.

is
The quantity x r ~ j
light is
small since the wavelength of the emitted
of
\^ 10 r^ 10 cm.
-"
cm, and the dimensions of the atom are of the order Therefore, in a first approximation, the exponential
er l **
f
factor in Eq. (29.36)
=\
/x-r-f.
..
should be set equal to unity. Such transitions are called the dipole
transitions.
3 4
Taking
into account the
vector potential
in the
commutation relations in (29 30) in the expression for the case when there are no photons, we should retain only amplitudes
operators for the creation of particles
proportional to a*, that

4
is,
If we include the next term in the expansion we obtain quadrupole radiation. Quadrupole radiation is (r/A) times weaker than dipole radiation and is of importance only when dipole transitions are forbidden.
488
The probability of a spontaneous transition level n to the level n' is equal to

'
from the energy
=
.
,,
(29.37)
Here instead of two sums over the wave numbers
of amplitudes a
we
left
only one, since the only combinations of amplitudes a and a*
different
from zero are those which refer to the same momentum Using the last commutation relations, we find
(a-rf>.)(a*.
x.
/>) =
v.
[xJ/v,]
|.
Let us replace the

also Eq. (28.23)]
sum over
by an integral in Eq. (29.37) [see
<
29 ' 38 >
and include the fact that for sufficiently large values of time have, in accordance with (28*20),
we
= *(-*
n ,).
(29.39)
Then in the case of spontaneous transitions Eq. (29.39) reduces to the law of conservation of energy
W=
.L,
(29.40)
since from Eq. (29.40) it follows that the energy of the emitted E n ) lost by the atom as a photon hu is equal to the energy (E n result of transition. For the probability of spontaneous emission,
>
we obtain
we
Evaluating this integral and remember ing that Pn'n*= obtain the final expression for the probability of spontaneous
4 e "* nn
,
' \
emission
'
,4
r nn
.
|*
I
(29.41) \AV.-J
SECOND QUANTIZATION
489
which was already derived in Chapter 9 with the aid of the semi classical correspondence principle [see Eq. (9.20)].
D.
BETA DECAY
of the application of second quantization us consider the theory of )3 decay. This phenomenon consists of the emission of an electron (positron) by a nucleus, leading to an increase (decrease) of the nuclear charge by unity. The theory of ]8 decay, which resembles in some respects the emission of photons by atoms, was constructed on the basis of second quantization. There are no photons in an atom. A photon is created from the vacuum when an atom makes a transition from one energy state to another. Similarly, a nucleus does not contain electrons; they are created only in the process of /3 decay. As a result of experimental investigation of j3 decay it has been established that these electrons have a continuous spectrum bounded by a certain maximum energy equal to the difference in the energy of the nucleus before and after the decay. It was also established that in )3 decay the angular momentum of the nucleus changes by a multiple of h whereas the angular momentum carried away by the electron equals ( j^h The apparent violation of the laws of conservation of energy and angular momentum in /} decay was resolved in the hypothesis of Pauli, who assumed that the emission of an electron is accompanied by the emission of another particlea neutrino possessing halfintegral spin and a rest mass close to zero. According to the Fermi theory, constructed on the basis of this hypothesis, j3 decay should be considered as the transformation of one of the nuclear neutrons (n) into a proton (p), an electron (e~) and an antineutrino (v)
let
,
l
m
As another example
decay should be conSimilarly, the emission of positrons in sidered as the transformation of a nuclear proton 5 into a neutron, 6 a positron and a neutrino
p
>/i-|-e -f- v
+
-
We note that since the rest mass of a neutron is greater than the total rest mass of the proton, electron, and antineutrino, it follows that the decay of a free neutron should also be observed. The decay of a free proton appears impossible from the energy standpoint and
therefore, positron decay can be observed only can be taken up from the nucleus
in a
bound proton, when the required energy
We shall explain the difference between the neutrino and antineutrino at the end of the present chapter in the discussion of the nonconservation of parity
490
Furthermore, the capture of a bound electron is also possible; as a result of electron capture a proton is changed into a neutron and e~ ~* n emits a neutrino (p v) . As a rule, an electron from the K shell is absorbed in this process, and therefore this phenomenon is called K capture. K capture is similar in nature to positron j3 decay, since in both cases the charge of the nucleus is reduced by unity. We shall not consider here the details of j3 decay; our task will be to describe in general terms the creation of an electron and antineutrino following the Fermi theory. The energy of interaction of a neutron with the electronantineutrino field can be written as
V* =ft?tf
where f is a of / is very
(29.42)
coupling constant introduced by Fermi. The magnitude small (/^ 1.4 x 10~ 49 erg x cm 3 ) so that this interaction is called a weak interaction. The spontaneous decay of particles is caused mainly by weak interactions; therefore, the lifetime of elementary particles or nuclei is comparatively large and varies from a fraction of a second to billions of years. Nuclear processes, on the other hand, are caused by strong interactions, which sometimes are a thousand times larger than the electromagnetic interactions. The duration of the processes caused by such interactions is very short (of the order of 10~ 23 sec). Neglecting spin effects, the wave functions i|*J and <|~ can be represented as
l
<>
_ =
-*/**icKt-ik
r
r
-a
HI-X
'
(29.43)
.
there are no particles present at the initial instant, the following relations hold for the amplitudes a* and b*: a*a~b*b Q, aa*
Then, according to Eq. (28.12), we have the following expression

for the coefficient C(t):
t
C (0
=
Jr-3
V pna*b*
dterWr-Kp-K-*)
(29.44)
Here E n
= chK
n is
energy. The matrix element
the neutron energy and E p of the interaction
= cHK
Vpn
is p is
the proton
*.
(29.45)
wave function of the neutron, XP is the wave function and 7 is the Dirac matrix which determines the nature of the interaction. Since the interaction energy must be a scalar quantity, an analogous matrix should also relate the spinor
X n is the of the
where
proton,
SECOND QUANTIZATION
491
amplitudes of the wave functions of the electron and neutrino in Eq. However, the influence of the specific choice of the Dirac matrices manifests itself only in the spin effects, which are neglected in the present treatment.
(29.42).
Just as in the investigation of dipole radiation, the quantity in the exponent in the matrix element (29.45) is much less than unity, that is, inside the nucleus
where R is tue nuclear radius. The exponential, therefore, may be ^ expanded in a series
e -K*f x)r
= _
i
(k -fx).
r+...
If
the matrix element (29.45) does not vanish upon replacement of the exponential by unity, the corresponding transitions are said to be allowed /? transitions* We note that the allowed transitions, which correspond to the dipole transitions in the theory of photons, are associated with definite selection rules. 7
7
In the
case of vector interaction, the matrix element
(the fourth component of the four-dimensional velocity, the first three components for the nucleon at rest vanish) will differ from zero (allowed transitions) if the spin of the nucleus (that is, the total angular momentum) remains unchanged in the /? decay (AJ - 0, Fermi transitions) This Fermi selection rule is satisfied for the majority of nuclei. There are, however, some cases of allowed /3 decay, for example,
He
->
3 Li
+ e~ + v
in
is
which the spin of the nucleus changes by unity (AJ - 1) The spin of the 2 He nucleus equal to zero (one alpha particle plus two neutrons with antiparallel spins), while the spin of the 3 Li nucleus is equal to unity (one alpha particle plus one proton and one
neutron with parallel spins, just as in deuterium). In order to explain these selection rules, Gamow and Teller pointed out that when we form the interaction energy operator, which is a scalar (V- V pn V e y), one may take a
product of two vectors, as well as other relativistically invariant combinations of the four wave functions For example, we may take a product of two pseudovectors, which is also a
relativistic invariant
The pseudovcctor
interaction
A J - 0, "tl It should be emphasized that for a nucleon at components will be different from zero -> In view of the fact that the pseudovector interaction forbids transitions, which were nevertheless observed experimentally, as well as for several other reasons, presentday theory uses a combination of vector and pseudovector interactions (as in the Feynman and Gell-Mann version of the theory of fi decay). This version enables us to explain the
leads to the selection rules
rest, the first three
basic experimental data obtained
in the investigation of
/?
decay.
492
In the subsequent discussion, we shall restrict ourselves to the treatment of allowed transitions; in this case it is sufficient to make the exponential term in the matrix element (29.35) equal to unity. We then obtain the following expression for the coefficient C:
'
JL
n
(29.46)
where E a ==chKnp
light
= ch(K
Kp
particles
(electron,
is the energy carried away by the ) neutrino or antineutrino) in ]S decay.
Energy distribution of electrons in (according to the Fermi theory). KQ is the maximum energy of the /i spectrum The origin of the coordinate system corresponds
Fig.
29.1
decay
to the energy
Using Eq. (29.39) and changing the summation over the of electron and neutrino to an integration [see (29.38)], an expression for the probability of /3 decay
momenta we obtain
= i I c c = TfifeT- J
*
k,
Integrating this expression over all possible angles at which the electron and antineutrino emerge from the nucleus and also over the energy of the antineutrino, we obtain an equation for the energy distribution (E cHK) of the electrons (the /? spectrum)
w=
f
J
w(E)dE
(29.48)
where
w (E) =
(29.49)
SECOND QUANTIZATION
493
The energy-distribution curve of the j8 electrons, obtained in the Fermi theory, is plotted in Fig. 29.1. 8 From this curve it is clear
the energy of the emitted electrons lies between mm m^ E n Ep and E max =Eo to the Pauli- Fermi theory, there is no violation of According the law of conservation of energy, since the total energy of the emitted antineutrino and electron must always be equal to the total 9 energy lost by the nucleus during ft decay.
that
E.
NONCONSERVATION OF PARITY OF PARTICLES
IN
THE DECAY
One of the fundamental discoveries in the theory of weak interactions was the discovery of the nonconservation of parity by Lee and Yang ( 1956), This phenomenon gives rise to a spatial asymmetry in the spontaneous decay of elementary particles and, in
particular, in nuclear
decay.
The nonconservation of parity can be observed experimentally in the following two phenomena. 1. The asymmetry of the angular distribution of electrons in the 8 decay of nuclei with an oriented spin (the number of electrons emerging along the direction of nuclear spin does not equal the number of electrons emerging in the opposite directionX 2. The existence of circular polarization (hehcity) in the particles formed during decay (for example, electrons formed in decay or /j mesons formed in the decay of mesons), even in the case when the decaying system has zero spin. The phenomenon of nonconservation of parity in decay or in the decay of a TT meson was explained with the help of the theory that assigned a definite circular polarization (helicity) to the neutrino. At one time it was thought that the neutrino (which is formed in positron decay) and the antineutrino (which is formed in electron decay) were identical particles (Majorana's hypothesis X It was suggested that this hypothesis could
it
be tested in double decay. If the neutrino and antineutrino were identical, one would expect comparatively large values for the probability of double decay without the emission of a neutrino (one neutron of the nucleus emits an electron and a neutrino, while another neutron emits an electron and absorbs this neutrino), that is, with the emission of only two electrons. If, however, double decay consisted simply of two successive identical decays with the emission of two electrons and two antineutrino s, the probability of decay should be much less. Experiment has confirmed the correctness of the second hypothesis and clearly demonstrated that the neutrino must be different from the antineutrino. We note, incidentally, that both particles are neutral and have a spin 1/2. Physicists, however, were able to establish a difference between the particles from phenomena associated with nonconservation of parity. It has been demonstrated experimentally that the asymmetry observed during positron B decay of a nucleus with an oriented spin, when a neutrino is emitted together with a positron, is the reverse of the asymmetry that is observed during electron decay, in which an antineutrino is emitted together with an electron. It was assumed, therefore, that the neutrino differs from the antineutrino by the type of circular polarization. In order to explain the experimental data, it was necessary to postulate that a neutrino resembles a photon with left-hand circular polarization, while an antineutrino resembles a photon with right-hand polarization. The
should be noted that the maximum of this curve
is slightly shifted
It
towards small
energies. This asymmetry is due to the fact that the antineutrino mass is equal to zero, while the electron mass is different from zero If the m a ss of the antineutrino were equal
to the electron mass, -this curve (neglecting the Coulomb attraction of the electron and the nucleus) would be symmetrical, that is, the maximum would occur at the point E/2
For more details, see

John Wiley
&
Sons, Inc
Bethe and 1958
Morrison, Elementary Nuclear Theory,
New
York:
494

1 (in
only difference is that the spin of the photon is the neutrino is 1/2.
units of H),
whereas the spin of
$(left)
Neutrino
29 2
Antineutrmo
Fig. 29.3 Hehcity of a lefthanded neutrino in right-hand and left-hand coordinate
Hehcity of the neuThe and antmeutrmo neutrino has a left-hand circular polarization and the antineutnno has a right-hand circular polariFig.
trino
systems
zation
The circular (longitudinal) polarization of the neutrino is generally called the helicity. The neutrino has left-hand circular polarization, or negative helicity. This means that if a left-handed screw rotates along the direction of polarization, it moves in the direction of the momentum. The antineutrino, however, has right-hand circular polarization, or positive helicity (see Fig. 29.2). In order to conserve helicity when changing from one Lorentz frame of reference to another, so that it can be adopted as a characteristic of the neutrino, to zero. 10
it
is
necessary that the rest mass
of the neutrino
be exactly equal
Several authors describe polarization with the aid of an axial vector, which is perpendicular to the plane of rotation and has a different direction in the right-hand and left-hand coordinate systems (see Fig. 29.3). It should be noted, however, that, although the axial vector s and the polar vector p have different mutual orientations in the righthand and left-hand coordinate systems, the helicity is nevertheless conserved; that is, the helicity of the neutrino is still negative, and only the method of description has 1 1 changed. Starting with the polarization properties of the neutrino, we shall make an attempt to give a qualitative explanation of the nonconservation of parity during the spontaneous
decay of particles.
Let us consider, for example, the decay of nuclei with oriented spin. We note that the spin of the nucleus (its longitudinal polarization) is more naturally described by a rotation, since the direction of the axial spin vector is arbitrary. The spatial asymmetry which should be observed in the phenomena characterized by the nonconservation of parity is associated with the fact that in electron decay, right-handed antmeutrmos are emitted upwards and downwards with spins oriented, respectively, parallel and antiparallel to the spin of the nucleus. The electrons will be formed predominantly with a helicity opposite to that of an antineutrino (that is, with a negative helicity). This produces a spatial asymmetry due to which the number of electrons emitted in the direction of nuclear spin does not equal the number of electrons emitted in the opposite direction (see Fig. 29.4). In general, we obtain the following equation for the number of emitted electrons as a function of the angle between the direction of electron momentum and the upward direction, which two directions form a right-handed system with the polarization of a nucleus: ^ (29>50) J8) ([
=^ _
^^
where a
is positive
and
is
equal to approximately 0,4.

differs from zero, the parallel spin and momentum vectors transition from one Lorentz frame of reference to another
If the rest mass of a particle may be directed at an angle after
U Smce
mentum p
is different in the right-hand
the direction of the (axial) spin vector * relative to the polar vector of the moand left-hand coordinate systems, Lee, Yang, Lan-
dau [Zhurnal Ekspenmental'noy i Teoretichebkoy Fiziki, 32, 405 (1957)], and others assume, on the contrary, that in this case the negative helicity of a neutrino changes to a positive helicity. In other words, they assume that the method of geometric description of the particle can change its internal properties (helicity).
SECOND QUANTIZATION
495
Asymmetry was detected in the electron 3 decay of Co 60 nuclei with oriented spins and also in the angular distribution of electrons in the 3 decay of free, polarized neutrons (for which 0.1). Since a neutrino with a negative helicity emerges during positron decay, the asymmetry pattern will be opposite to the one described above. The number of emitted posi-
a^
trons is related to the angle & by the equation
w e+ ()
= w ^ + a cos
(1
&);
(29.51)
that is, the positrons that are formed have mainly a positive helicity and are emitted primarily upwards. An asymmetry opposite to that of electron decay was observed ex-
58 nuclei with oriented perimentally in positron 3 decay of Co spins.
ck-
4i
Electron 3 decay
Positron 3 decay
Fig. 29.4 Schematic diagram of the ft decay of nuclei with oriented spins The direction
of rotation characterizes the direction of the
spin
of
the
particles,
is
the
particle
momentum
Longitudinal polarization was observed particularly clearly in the spontaneous decay of pions into muons and a neutrino. Let us choose a coordinate system in which the pion is at rest and consider the negative pion which decays into a negative muon and an antineutrino. Since the antineutrino has a positive helicity and the momenta of the muon and the antineutrino must be equal and opposite, we find that the muon also must have a positive helicity. Indeed, only in this case will the total spin of the muon-antineutrino system be equal to the initial spin, that is, zero. In the decay of positive pions into a neutrino and positive muons, the muons, however, will obviously have a negative helicity (see
Fig. 29.5X
Meson
/
TT
Neutrino \Neut
Meson
Fig. 29.5.
Decay of
a pion at rest into a
muon and
a neutrino
496
The

total longitudinal polarization of the
created muons has also been confirmed experi-
mentally.
A more
of this
detailed discussion of the nonconservation of parity lies outside the scope 12 book, and we must refer the reader to the special literature on this subject.
See the papers in T D Lee and C. N. Yang, Nobel Lectures, Physics, Elsevier Publishing Company, 1964
New
York:
Appendix A
Hilbert Space and Transformation Theory
There is an elaborate formal structure of quantum mechanics that is important for several reasons: the wave mechanics and the matrix mechanics can be unified in one coherent scheme. The conceptual structure has a deep intuitive appeal that gives the
theory a sense of completeness and solidity c The physical content can be embedded in an extensive and rigorous mathematical framework. And, most important for the validity of the physical theory, it is a powerful and flexible phenomenological tool that embraces a wide variety of empirical knowledge. The conceptual structure and statistical foundations of quantum mechanics are formulated most fully in the framework of "abstract Hilbert space." It is, however, not necessary to go into the mathematical technicalities to achieve an accurate physical grasp of the theory. For a complete treatment of the mathematical and statistical
one should consult the original literature. In this appendix we shall outline some definitions and notational apparatus around three broad topics: vector spaces, operators and the inner product.
foundations
Vector Spaces
An abstract (complex) vector space
is a set of
abstract ele-
ments called vectors (or points in the space) which together with complex numbers obey the following axioms: (1) If V a and V b are vectors, V a * V h = V b * - a f V + * c (2) V a x If and y are complex numbers and V is a vector, then xV (3)
f-
fc
is a
vector and
(x
x(y*V) = (xy)*P.
x*V
(4)
(5)
O^P ^
+ y)<P -
yW and
x(
fl
HV
-
= x
fl
*Wb*
f
There
is a null
vector V n uii
such that
V nu n
and
V null for all*.
^.A.M. Dirac, The Principles of Quantum Me chart icv, fourth edition, Clarendon Press, Oxford, 1958. W. Heisenberg, The Physical Principles of Quantum Theory, translated by C. Eckart and C. Hoyt, University of Chicago Press, Chicago, 1930. (Dover Publications,
New
York, 1949). J. von Neumann, Mathematical Foundations of Quantum Mechanics, translated by R. T. Beyer, Princeton University Press, Princeton, 1955.
498
APPENDIX
The structure of a vector space is appropriate for introducing the notion of linear independence: the set of vectors Vi, 2* Vk are linearly independent if the only set of complex numbers zi, + zk^k = ^nuii *k satisfying the relation 21^1 + 22^2 + 22. - z k - 0. If it is possible to specify is the trivial set z\ = 2 2 * .
-
linearly independent vectors but notn + 1, the space is said to be ^-dimensional. If there is no limit to the number of linearly independent vectors that may be specified, then the space is "infinite dimensional." In an n-dimensional space a set of n linearly indeforms a basis that spans the space pendent vectors î, *P 2 in the sense that any element y a of the space can be written as a w, linear combination of the basis vectors V,, - 1, 2,
n
,
.
n are complex numbers that charThe coefficients a 1, 2, acterize the vector V a in the basis ^n. If a and &it 2 = 1, 2, n are the components of the vectors *P a - 2a V, and /
t ,
a, + b, are the components of the vector V c 16,^, then c Sc,V, = ^ a f V^ and za are the components of the vector 2W a where z is a complex number. The geometric concept which is the fe-dimensional generaliza-
Vb
--
=-
tion of the notion of "lines and planes passing through the origin," can be developed with the following definition: A subset 9W of a vector space is called a linear manifold if it contains all the linear combinations y\^\ v ^2^2 + + yk^k along with any k(k - 1,2,. of its elements Vi, W 2> W*. Alternatively, if 8 is an arbitrary set of vectors containing the distinct then the set of all linear combinations vectors Wi, ^2, ^fe,
.
.
*iVi + X2^2 + ^cfe^fe (with arbitrary complex numbers x\, X2, '* linear manifold xfe) is a linear manifold 9W. SK is called the If the vectors V are SI." spanned by linearly indeVi, 2 pendent, the manifold is fc-dimensional. The utility of the concept of a linear manifold is that it is the domain of definition of a linear operator. In general an operator is a mapping or correspondence from a domain consisting of certain points in the vector space into a range consisting of certain other points, denoted symbolically by V + R(). In quantum mechanics we are concerned with linear operators for which by defini*. . . ,
>
tion
R(V
That
+ *) = B(V) + R(*)
and
B(aV) = atf ()
is, the domain of definition of the operator R is a linear manifold. Note that a linear operator need not be defined for all vectors in the space. For example, if the space is the "function space" consisting of square integrable functions on the interval
H1LBERT SPACE AND TRANSFORMATION THEORY
499
for which xtf/ or then there are vectors (functions) not are square integrable. d*///dx There is a "projection property" associated with a linear manifold: any vector V may be resolved into the sum of two vec= + where V|| is in the manifold 2 (i.e., there are tors j - c c& such that V|| i*Pi 4- 02^2 f complex numbers cj, C2, V c-fcfc f where it V* span the SK) and^j is entirely outside 2 2H. As a notational device we may define the linear operator P^ for all The domain of definition of P^ is as follows: . the entire space, while the range consists of 2ft and the null vector. If *! lies outside 2K, P î = Vnu ii. If V 2 lies wholy within 9, PW V 2 = m If ^ is an arbitrary vector, P^W lies entirely within 2tt and 2. -= p p (pa W) - ps v - Since this relation holds for all W, we may g assert the operator equation
< x < ~,
if/
||
_,
--
||
PW
m^
An (Hermitian) operator
or a projection operator.
satisfying this relation is called idempotent

It
has eigenvalues
and
1.
Operators
stand for a physical observable which upon sharp measof a sequence of values 8{, 82, which are characteristic of the observable. The set of values Hi, 82, is called the spectrum of 8. An algebra of observables is set up as follows: let ^ be a complex number, then z$l is an observable with characteristic values ztti, z%2, and if a measurement of 8 2 yields 8jfe, the measurement of z 8 yields z8iU The observable B = 2 88 yields the measured value (8fe) when 8 yields Si. If 8 and 8 are simultaneously measurable then the sum 8 t 8 is defined to be the observable that yields 8^ f S3? when 8 yields 8^ and SB yields may be defined by the artifice Finally, the product 8
Let
81
urement yields any one
'/.
8
If
2
--J
(8
^(8 4
S)
8 and
SB
are simultaneously measurable. These definitions pro-
vide a "physical" construction of polynomials of several simultaneously measurable observables 8, 83, etc. In quantum mechanics the mathematical idealization of an observable is a linear operator on a Hilbert space. Let us first outline the purely algebraic aspects of operators which can be defined without reference to the space on which they act. An abstract operator algebra is a collection of elements called operators which together with complex numbers are endowed with the structure of a vector space (in a technical sense, not to be confused with the Hilbert space) and in addition an operator product. Explicitly,
500
(1)
(2)
If
APPENDIX
M and
+
(IV
/V
are operators, then M + N = N + Mis also.

- <M +
IV)
f 0)
0.
Multiplication by a complex an operator and x(yM) - (xy)M.

N
(3)
(4)
(x
number
is
allowed;
x/V.
i.e.,
xM
is
+ y)M is
yM and x(M
(5)
There
a null operator
IV) - xM such that
M and OM
for all M.
Furthermore, there is a composition law (the "operator product") which assigns to each ordered pair of operators (M,/V) another operator which is denoted by M/V. The composition law obeys the following axioms (we now denote operators by TI, T 2 T 3 ): (a) T\T% is a operator if TI andT 2 are; i.e., the operator algebra
,
is
closed with respect to the product. - <TiT 2 )T 3 (associative rule). (b) Ti(T 2 T 3 bT and (aT 2 + 3 = aTiT 2 + bTiTa (c) Ti(aT 2
) )
(d)
There
is a unit
operator T unit
such that IT
- TI -
T for
operators T. Once a concrete identification of the operators is made and a labeling scheme adapted a "multiplication table" can be setup. The multiplication table can be formulated with the help of the notion of linearly independent operators. Assuming there are only a finite number n of linearly independent operators TI, T 2 ..... T fl we may express any operator as a linear combination of these. In particular, as the operator algebra is closed,
all
,
where
the
complex coefficients C
k
t
are the structure constants of
the algebra.
The algebraic properties defined by operator axioms (l)-(5)and can be realized concretely in terms of linear transformations on a vector space. The operator TI is identified with the transformation which maps V into TiW. The transformation is
(a)-(d)
linear
Sum and difference TI T 2 is identified with (T! T 2 )*P - TI*P while zTi corresponds to the transformation (zTtfV - z(TiV), these relations being valid for all for which TI and T 2 *P are defined. If the range of T 2 includes the domain of TI, the operator product TiT 2 is defined by (TiT 2 )V - Ti<T 2 ) for all V in the domain of T 2 . A transformation is nonsingular if it transforms distinct vectors into distinct vectors; that is, if / 2 , then Tî -/ T^ 2 . An alternative definition: T is nonsingular if to each vector $ x - T^! there is a unique solution VL In particular, if TVi - Vnuiu then
HILBERT SPACE AND TRANSFORMATION THEORY
501
The unique solution î may be expressed in terms of an operator S defined by the relation î = S$i. It then fol4>i by lows that î - STî. If this latter equation holds for all *P lf we may assert the operator equation ST - 1; S is called the left inverse of T. If TS - 1, then S is the right inverse of T. If ST - 1 - TS, then S is the inverse of T and is given the notation T" 1 also, T - S' 1 . 1 - B~ l A~ l If A and B have inverses, then the inverse of AB is (AB)' if A B not vectors and do the even if commute. Finally, Vi, ^2* are linearly independent and span a manifold 2)1 and if V Wjfe is a nonsingular transformation, thenVVi, Wk are linearly independent and span the transformed manifold V2W.
.
V^
Inner Product
An Hermitian inner product is a mapping that assigns to each ordered pair of vectors W, $ a complex number designated by (, 4>). The number (V,<t>) is called the inner product of W and<J>. An inner product has the following properties:
(V, $)
<<&,
)*
(Hermitian symmetry)
(,a<D
(V,V)
60) =
a(V,)
6(,Q)
implies V
and(y,)
O
i,fe
The norm or "length"

tance between V and
orthogonal. basis if
the
of a vector
is
||
is
If
. .
||
W||
V<,)
-
V
.
- 0||.
(W,)
.
and the disthe vectors are
A
+
set
-
of vectors
- 1,2,
(&)
^1,^2,
.
forms an orthonormal
and the triangle in-
5^,
_i
Some important theorems are

||$||
Schwartz inequality
||W|| ||$||
|(V,<W| <||V||
2R there is an orthonormal basis which spans 9M. The projection operator P^ of a manifold 3M spanned by the vectors Wi,*P2 ^k can D ^ written in terms of the inner product as
equality
||W + $||.
For each linear manifold
>
>
W
Z
E ^(VnW
i=
i
,
The projection operator onto the ray or one dimensional manifold defined by y, is P W If V!, V 2 ... is an orthonormal set the projection operators are related to the vectors by the equations
,
<).
t
i
Pt^k - 5,jk^. The orthonormality property is P,P - d^Pk and the completeness relation is SP = 1, the unit operator.
fe
^1,^2* ... be the eigenfunctions of an operator M corresponding to the eigenvalues Mi,M2, .... For simplicity, let us assume the eigenvalue saredistinct, Mj ^ Mj^for/ ^ k\ then (*,,^) and the vectors may be normalized so they form an orthonormal
Let
502
basis.
APPENDIX
A
of this orthoP,M -
The relation of the projection operators P, normal basis to the operator M is the following: MP, The spectral resolution of M
trary function of a real or tion F(M) is defined by
is
MjP/.
EM,'P;
;=
1
Let F(A)be an arbi-
complex variable A. The operator func00
F(M) =
i=i
E F(M;>P
In the mathematical realization of an observable SI in terms of an operator A, the eigenvalues Ai,A2 of the operator A are identified with the characteristic values 8li, 82 of the observable. The construction of polynomials of observables is identified with the construction of functions of an operator given just above. There is a particularly important notation, the Dirac notation, > is placed around the symbol *P to denote a vector. in which a ket V > is called a ket vector. The inner product between vectors V and & is denoted by < W a W 6 >. The expression < V a is regarded as a vector in its own right; it is called a bra vector or simply a bra. The primary distinction between bra and ket vectors is that P transforms as ^ -> UV under a unitary if the abstract vector transformation U, the bra |>-[7|>= !(/>, while the ket <V\ + 1 transforms (contragrediently) as <V\ <UV\ = <V\U = <|U" . Under these transformation properties the inner product < W| remains invariant. Matrix elements of an operator M are written as <V|M|$> = (T,MO). In the Dirac notation, the defining equation for a projection operator reads
\ |
fl
-=
> is arbitrary it is a permisfor an arbitrary ket W >. Since sible notational device to omit writing the >. Then the expression for a projection operator in the new notation is
| |
P<Dt
IV.xVJ
î,^
. .
.
where
the
the
sum runs over
the vectors
V n which span
manifold 9W.
terms of its normal basis $i,$2> ... by the notation
An arbitrary operator A may be expressed in matrix elements A^ - (0,, AO&) in a complete ortho-
There is a notational simplification possible when working with a particular basis <Di,<&2, in that the ubiquious <D is a redundancy
.
HILBERT SPACE AND TRANSFORMATION THEORY

fc
503
1
vector $ > can be abbreviated to read i >. The Hilbert space for a specific physical problem is built on the cononical coordinates Q V Q 2 Qk f r a system of fe-degrees of freedom. The abstract vectors have no numerical significance, rather the q'a serve to label what will be considered a complete orthonormal set, yq v q 2 q k . The ket will be denoted by \q l9 In the (improper) 5- function normalization scheme q2 q k >.
in the notation: the ket
|
the orthonormality relation is
<q v
q2
qk
î'^'
a
qk >
=
i
in the Vq basis:
An arbitrary vector
=
may be expanded
, . . .
*(t
a (q v q 2
where the expansion coefficient ^ a (q v q 2

.
.
.
q fe ) is the probability
j
amplitude for finding particle 1 at position q l$ etc.; i.e., ^ a (q v q 2 q k ) is the Schrodinger wave function in the coordinate representation. Taking the inner product of To > with the bra <q v .q k and using the orthonormality relation, one may verify that
,
.
In quantum mechanics a particle may have a discrete internal degree of freedom, spin, for example, where the discrete variable o takes on either of two values, t for spin up or i for spin down. For a particle with spin and one spatial degree of freedom x an arbitrary state may be expanded as
where O a (x)
is the probability
amplitude for finding the particle at

with no spin, by first ex-
x with spin a (see
Chapter
16).
&
we write
Treating now a particle on the interval -> < * < the inner product of two vectors V a and panding W in terms of the set V,
fr
then taking the inner product using the distributive rule under the integral. The result is
>oo
(
<k =
*r<X)
504
APPENDIX
A
The projection operator
the ordinary tunction space inner product. P a b onto the interval a < x 
-
is
(V x ,MVy )
expressed in terms of by the formula
its
x- space
>CO
/OO */-00
.0
<x|M|y> \V x
(ket) state
xV
\dxdy
> is written
the action of an operator
M on a
v*CO
vector
*P
|
as
/-00 /CO /
JO
|V
JC
:xx|M|yxy|M'>
to
dxdy
where
it
is
now appropriate
M.
integral operator thing, the expectation value of
The average value M
<-x|_M|y> the kernel of the or what is the same f in the state <t>, is
call
--=
operator V is called local if its matrix elements have the special form <x|M|y > - 3(x - y)V(x), where 3(x - y) is the Dirac S-function and V(x) is an ordinary function. The expectation value of a local operator is
An
r
V
/
/-co
The kinetic energy operator T has the matrix element < x T

\
y > =
/2 ~
tin
^x8(x -
y),
which
is also
regarded as a local operator. With
the help of Integra tion-by- parts the expectation value of T can be written
/*
2m / -no
where
Appendix B
The Statistical Assertions of Quantum
Mechanics
consider a cononical system of k degrees of freedom, emq k to specify its configuration and ploying the coordinates g the cononically conjugate momenta to specify its condition of motion. In the wave mechanical mode of description, in the coordinate representation, everything that can be said about the "state of the system at one time" (its configuration, condition of motion and the values of all its physical quantities) must be derived from the wavefunction <f>(q^ q k ). The functions admitted as wave functions are those that are square integrable (normalizable, 0|| finite) and furthermore normalized to unity,
.
We
||
y00
y00
*^
...
00
Z
*^
\<t.(q l ,--CD
q k )\
dq l
.^dq h
= 1,
although in applications to continuous spectra this requirement may be relaxed. There are three primary statistical assertions: (1) the probability of finding the system within a volume V of configuration space is
energy of the system has the operator H with eigenvalues and eigenfunctions e probability that tlien the system has the energy value E 7 in the state is
(2)
if
the
Ei,E 2
^^^
-co
'
' '
-co
./-co
/
/-co
^M\
and
(3)
the average value of a physical observable

is
81
to
which the
operator A corresponds
/_oo
*/-oo
0<g t
q k )*A
fo
.
,
506
A
APPENDIX B
in the quantum state <. The interrelation of these three assertions will now be examined. In order to present assertion (1) in a general form, we introassociated with duce the projection operators Py (/ y ), ; = < < The the intervalue l defined operator is
!,,*,
by
q]
QJ
qj.
projection
defined by
The projection operator associated with the (rectangular) volume V is Pi(/i)P2(/2> P*C*) and the integral specifying the probin V may be reduced of the particle described by finding ability
to the
expression
For assertions (2), let Psn denote the projection operator associated with the eigenvalue E n "of H, that is,
PE n
~
1/fm
^n
= HJ
^d
for
j4
tl
The probability of finding the value E n of the energy of a system in the state $
upon a sharp measurement can then be put in the form
The probability E'< E <:E"is
of finding
the
system within an energy interval
where
P(/E) is the projection operator associated with the interval
PE>
Pz n
Assertions (1) and (2) may now be unified in what most general probability assertion possible":
is
called "the
Statistical Postulate: The probability that in the state the physical quantities with the operators Ai,A2, ,A m take
-
on values from
the^
respective intervals
/j,/2
'
is
where Pi,P2,
to the
P m are
.
. .
the projection operators belonging
operators AI
Am .
THE STATISTICAL ASSERTIONS OF QUANTUM MECHANICS Assertion assertion

(1)
-
507
has m = fe, with A\ - q\, A% - q^ A^ = <?&; while = to that this In insure m with order has H. AI 1, (2) postulate is a coherent statistical statement the following properties must be varified: (a) Since the order of the operators is arbitrary in formulating the statistical question, the order of the projection operators in the statistical assertion must be immaterial. This implies that the projection operators must commute for arbitrary intervals Jm and this in turn implies that the operators Aj,A2 /b/2 ,A m must commute among themselves. (b) Vacuous propositions may be inserted at will, vis. if the interval /' is contained within the interval /, P(/)P(/') - P(/'), or the projection operator for obtaining any value whatever is the
-
identity.
(c) If the interval 1 3 lies outside the spectrum of A/, for some ;', the corresponding projection operator is the null operator and the probability is zero. (d) Probabilities are additive; i.e., if an interval / is resolved
into
P(/")
two disjoint subintervals / - /' + /", with operators corresponding to /' and l" respectively, then
t
P</')
and
2
||P(/)<D||
||
PC/')
(D
||
||P(n<D||
< W < 1 The total probability W ranges over the values for normalized <1>. Assertion (3) at the beginning of this appendix, can be expressed in terms of the inner product. If we introduce the notationExp{9,i
(e)
to
stand for the statement "the expectation value of the physical observable SI in the state <V' assertion (3) may be stated
Exp!3(,<J>j
=,
(<D,A<D),
where A
is the operator corresponding to 81. Let F(A)be any function of the real parameter A. There is a theorem to the effect that F(S) if the observable 9 has the operator A, then the observable has the operator F(A). Assertion (3) may then be generalized to
read
ExpiF($l),4>! = F(A)
--
(O,F(A)0)
(3')
Assertion (3') can be derived from the statistical postulate as follows: subdivide the interval (-<, ~) into a sequence of subinter< A- n < < A-i < AO < AI < > A < vals / = \\ n Xn +i .where + 00. Let \n be some number in the interval A; < A; < A; +i. The average value F(A) in this mesh is
,
I
FXA) /= -n
A;)
508
APPENDIX B
If we introduce the monotonic increasing projection operation E(A) defined by
A;
-
<A
<*>
and the E(A; +i) - E(A,)) and let the n -> (conversely, P(/ 7 ) size vanish, the sum approaches the Stieltjes integral
mesh
which by definition of a function of an operator is (<J>,F(A)<J>). So far we have indicated that assertions (3) and (3') follow from the Statistical Postulate. What is remarkable is that the converse is
also true: the Statistical Postulate follows from assertion (3). The proof of this statement is obtained by a technical application of the ,A TO be a set of mutually comfollowing theorem: Let Ai,A 2 muting operators; there exists an operator R and functions Fj( A),
,
such that AI -- Fi<fl),A 2 - Fo(tf) ..... F m - F m (R). preceding discussion we have assumed that the state Is a pure state, that is, it is described by single vector <J>* In general, in the physical preparation of a state some of the variables are left uncontrolled and the state Is not completely specified. This situation can be formulated in terms of a classical probability distribution that is superimposed on top of the quantum mechanical uncertainty. Let us suppose that the system is in one of the states V ,a = 1,2, but we don't know precisely which one. Let Wa be
F 2 (A) ..... F m (A)

In the
classical probability that the system is in the state W a . The basic statistical postulate can now be reformulated as
the
^
and assertion
(3)
E W PlUl)P
a
2 </2>
'
>
Pm^m) Vail
read
can be generalized
Expl
to
Kô.VJ
!><1'a,AVa) a
of
is
mutually orthogonal.
S\V a - 1.
The states V a are an arbitrary set For the Wa 's it
states; they
required that W a >
need not be and
The statistical assertions can also be formulated in the trace. The trace of an operator A is
trA i
terms of
(^.AV,)
where the sum runs over a complete orthonormal basis Vi,V 2 .... The trace is independent of which basis is used to define it. The
,
THE STATISTICAL ASSERTIONS OF QUANTUM MECHANICS

statistical assertions are
509
W
and
trpPi</i)P 2 </2>'
'
'
Pm</m>
Expl
a^Vj
* a.
trpA
where
p is the density n P
matrix (operator)
V* / * IV "a P
a
V /
a.
*P
Ya
'
^W<WI "a ^ T
a.[
The density matrix is useful for making a general statement about the time evolution of the system. Expectation values change with time according to the general rule
dt
ExplH;tfa
tripA + P A
In the Schrodinger picture the burden of change is put on /?; namely, A and + ih p ^ Up - pll, where H is the Hamiltonian. In the and ~i~K A - HA - AH, (see Problem 8.4). Heisenberg picture p = The Hamiltonian controls the evolution of an isolated system. When the system is subjected to a measurement of one of its observables ft and the eigenvalue R' obtained the uncontrollable disturbance of the measuring process forces the system discontinuously into the state described by the eigenvector W W corresponding to the eigenvalue R\ If the measurement of R is repeated (before the evolution generated by the Hamiltonian moves the system into another state) the system is already in an eigenstate^'of # and
'
thus yields the value R' with certainty.
Problems
Chapter
1.1
Express the space and time dependence of E, H and A in terms of amplitudes and phase angles for a plane wave moving in the positive x direction. What are the conditions for plane or circularly polarized light? What is the initial data at t = for solution of Maxwell's equations or the wave equation for A? Express the solution in terms of a power in terms of a series in time. Express the solution for t <
solution for
t
>
with appropriate initial data.
1.2
Determine the magnitudes of E and H for unpolarized sunlight characterized by the solar constant (Poynting vector)
S x = 2 cal^cm^mln" 1. What is the ratio of electric to magnetic force on an electron moving with the speed of an electron in the first Bohr orbit? Compare the wave and photon descriptions of normal reflection of sunlight from a mirror. What is the energy density, pressure against the mirror and the density of photons? Take X = 55 00 A, What is the electric field intensity of a photon
1.3
1.4
1.5
absorbed in a 1 cc detector? Find an expression for the angular momentum density of an electromagnetic field. Show that a photon carries an angular momentum of magnitude H. Calculate the torque on a quarter wave plate by a normally incident left- circularly (rightscrew) polarized light. Determine the spectral density of radiation under the assumption that the oscillator representing the behavior of the walls for angular frequency o> can assume the energy value E = or any one of a continuum of values E > o. Show that the gap in the energy spectrum is related to the behavior of 7to>. P^ in the quantum region kT
Answer. The partition function Z - 2e" aEn
e'^/m*.
Chapter 2
2.1
Consider the scattering of 4 MeV alpha particles on gold atoms (Z = 79). Show that the distance of closest approach is
r
mm
(ZZ'e$/2E^
(l
csc|0)
and evaluate
it
for scattering
512
PROBLEMS
angles 5, 20 and 80. At what angles will the scattering deviate from the Rutherford formula? Consider the scattering of alpha particles from gold foil. A radium source yields 4.8 MeV a-particles at a rate of 3.7 x 10 10 particles per second per gram of Ra. The target has 22 atoms-cm" 3 a thickness of 4 xlO~ 5 cm f anda 5.9 x 10 2 The target is situated 1 cm cross sectional area of 8 . 2 area a Consider from the source. detecting screen of 2
,
2.2
mm
mm
the target. What amount of radium is required for a counting rate of 30 per minute at scattering angles 10 and
5
cm from
120?
2.3
Derive (2.46) for a general one- dimensional potential by

evaluating dl/dE directly. Reduce the validity of the Bohr quantization rule for large n by equating the classical expression for the frequency of the expression in terms of the energy level spacing. (This application of the correspondence principle shows that (2.46) and (2.47) give a consistent classical limit if the same constant ft appears in both equations.)
2.4
2.5
Examine the classical and quantum descriptions of the emission of light from a harmonic oscillator. On the basis of the correspondence principle, argue that the n appearing in the Bohr quantization rule is the same Planck's constant characterizing the corpuscular nature of light. Deduce the energy level spacing and infer the levels between which electric dipole transitions are possible (selection rules). Use the correspondence principle to infer the energy levels of a rigid rotator consisting of a mass m held at distance b from a fixed axis of rotation. Note that the angular momentum
is
8<6
Show
of H.
independent of the mechanical parameters. that for circular motion in a general centrally symmetric field, the angular momentum is quantized in steps
2.7
Find the energy levels of circular orbits in the Bohr planetary model of the hydrogen atom using the correspondence principle directly. Note that as usual n is left undetermined up to an additive constant of integration. Infer the selection rules for electric dipole transitions between circular orbits
2.8
and find the accompanying angular momentum change. Estimate the lifetime of a stationary Bohr orbit by computing the time required for the corresponding classical motion to radiate away an amount of energy equal to the quantum energy level spacing. Compute the cumulative time for several successive transitions between circular orbits and compare it with the time taken to radiate the same amount of energy
classically. Use the Bohr quantization conditions to determine the energy levels of an isotropic three-dimensional harmonic oscillator.
2.9
PROBLEMS
Let the potential energy be
2.10
i ma> 2 r 2
513
and use polar coordinates
in the plane of the classical motion. Find the energy levels of the relativistic linear
harmonic
oscillator according to the Bohr-Somerfeld quantization rule. Evaluate the lowest order correction to the nonrelativistic energies. The potential may be introduced via the scalar
potential
e$
kx
2
.
~kx 2
or by replacing the
mass
invariant by
c2 +
l
*
Chapter 3
3.1
the deBroglie wave lengths of the following pareach with a kinetic energy 500 keV: photon, electron, proton, and alpha particles. Also of thermal oxygen atoms at
Calculate
ticles
300K.
3.2
a>(k) if the group velocity is dispersion law a> inversely proportional to the phase velocity? How is this case realized physically? Compare the reflection of a particle and a wave from a moving surface. Show that AE/Ao - Ap/A/e, where AE(Ap) is the change in the energy (momentum) of the particle and Ao>(Afe) is the change in frequency (wave number) of the wave. Show this independently of the dispersion law of the wave and the energy- momentum relation of the particle. How can this result be generalized and what is its significance? Calculate the deflection of a charged particle by a thin slab of magnetic field, expressing the change of momentum in terms of the vector potential. How is the difflection explained in terms of the wave picture? What is the relation between momentum and wave vector in the presence of a magnetic field?
What
is the
3.3
3.4
Answer.
3.5
k =
across the slab and
p + Ae/c
--
Tfk.
3.6
Consider a particle bouncing back and forth in a rigid box of linear dimension L. What is the minimum measurable kinetic energy of the particle? How much energy is required to constrain an electron to remain within a volume of nuclear size? Determine the maximum time a free particle will remain within a volume of radius R by considering the limitations on the specification of the initial data in the classical description of motion.
3.7
Find an uncertainty relation connecting angular momentum and angular orientation. Consider a rigid rotator. What is
the
minimum
uncertainty in L?
3.8
ball bounces on another with a center- to-center height of ten times the radius. What is the optimal horizontal
One billiard
514
localization to
PROBLEMS
maximize
maximum number
Answer,
3.9
n
the number of bounces? of bounces?
What
is the
50.
What is the optimal localization of an ideal pendulum to maximize the time it will remain balanced in an inverted
position and what is the
maximum time?
Answer. About six times the period for small oscillations.
Chapter 4
4.1
Under what circumstances can a narrow potential be approximated by a ^-function? What is the effect of a 5-function
potential on the continuity property of a solution of Schrodinger's equation? Show that the potential V(x) = ~gd(x 2 - a 2 ) has a discrete 2 eigenfunction for g > H /m and not one for anitsymmetric z < H /m. 8 -V S(jc) Consider the potential V(x} = for \x\ > L and V(x) for x < L. Show that for VQ > there is one eigenvalue E < and that it remains discrete as L approaches infinity. Find a complete set of orthonormal eigenfunctions for the potential V - -| VQ |5(x) on the interval -<*> < x < <*>. Check the completeness relation (4.67). (Sturm- Liouville theorem) Let \l/\ and 2 be solutions with energies E\ and E 2 of two Schrbdinger equations with potentials V\ and V 2 respectively. Show that
,
4.2
4.3
--=
4.4
4.5
t//
0* [<E 2 -
4.6
for any interval a < x < 6. Show that if y>i and ift 2 &re solutions for the same potential and if E 2 > EI, then i// 2 has at least one node between each pair of consecutive nodes of ^ t . Over the entire interval, then, ^ 2 has at least one more node than \l/ lm (The eigenfunctions are ordered in energy according to the number of nodes. A set of eigenfunctions is complete if there is one for each integral number of nodes.) (Comparison potential for discrete states) Consider the in< x < M with the boundary condition (7(0) = 0. Find terval the potential for which U = % n is a zero energy solution of the Schrodinger equation. Show that the potential obtained has no
discrete eigenfunctions for n real.

i'/3,
If
n is
complex,
n =
I
2
show
discrete)
that the potential has arbitrarily many states. The solution may be taken to
bound (i.e., be real U =
Vx cos
3 In*.
PROBLEMS
4.7
515
Let VQ be the comparison potential obtained in problem 4.6.
Show
(a) If
the following: V > VQ for x > XQ, there is no discrete eigenfunctions

,
(b) If
V with a node located beyond XQ. there is a discrete eigenfunction V < VQ for x > x with arbitrarily many nodes beyond XQ. There are inof
finitely many discrete states of V. < x < XQ, there is no solution for the poV > VQ for < x < XQ. tential V with E < that has a node in the interval < -<* V < if V ), x more than If V for x (i.e., rapidly VQ (d)
(c) If
then there are eigenfunctions with arbitrarily

in the interval.
many nodes
4.8
Scale the function in problem 4.6 to find some eigenfunctions of the potential V - ~ for x < a and x > 6, V ^ -V (l/x 2 < a < x < 6. Is there a restriction l/b 2 )a 2 b*/(b 2 - a 2 ) for among the parameters VQ, a, and b for this method to work? Can one find a complete set this way?
Chapter 5
5.1
Let a particle be prepared in the state = 0, where a > t explipx f a(x - XQ)! at
i(l + signx).
t
^r(x,0)
0,
XQ <
= V2a0(xo - x) and 0(x) =
Assume
/>(x,0)
that no forces act

in
on the particle for
> 0.
Expand
terms
t
of energy- momentum eigen-
functions and show that for

,i
*\
> 0,
V8a&
expG'px -
u/\Xyt)
i8p
+ /y 2 /4/3)
(2a/3 + ly)
C J
e~
1
L dL
iz
L2
2
.
where p
terpret
5.2
i/r
^/2m, y =
x - x
2j9p,
and
z - (1 +
iy/2apf
In-
for ^ - oo. Reconsider the initial

the
wave function specified

is
Assume
Expand
particle
^(x,0) in
terms
in problem 5.1. in a potential field V = -|Vo|S(x). of the eigenfunctions of problem 4.4
5.3
and follow the motion in the dispersion-free approximation (3.20). Compute the probability of the particle being reflected by the potential. What is the probability of finding the particle in the ground state? The initial value problem of the time- dependent Schrodinger equation can be solved in the form of an integral operator
with the Green's function kernel G t (x,x') =
n
)
^n
tl*
(*')* e~
lt* ni
where
is a complete set of orthonormal energy eigen\l/ n functions. (a) Find the differential equation and boundary conditions for G t in the case of a general one-dimensional potential.
516
(b)
PROBLEMS
that the force- free motion of a non-relativistic parin an s- dimensional Euclidian space is given by s/2 ' 2 -' - (m/2mW s G t (iM-') exp|m(r can G t serve as an approx(cf. Problem 3.3 in the text. How
Show
ticle
5.4
imation to the S-function?) Find the angular dependence and total current of charged particles flowing from a source at the origin. The wave lkr function far from the source is cos6e /r in spherical coordinates.
5.5
A particle is in a uniform gravitational field in a region bounded from below by a perfectly reflecting surface. Adapt the WKB method to the case of an impenitrable wall and show 3/4) 2/3 for large n. Show that that the energy levels go as(rc = ft for large n, where w is the classical angular & n _i
\-
frequency.
Chapter 6
6.1
6.2
6.3
6.4
6.5
Find the probability for a particle of charge +//CQ and speed v to penetrate through the Coulomb barrier to a nucleus of charge /ô. (Gamow factor) Derive (6.35), using integration by parts to pick out terms of order T 2 in Ejv . Find the pressure exerted by the walls to contain a particle in a very deep potential well of volume V at absolute zero temperatures. Find a relation between the volume and pressure of an electron gas at 0K. Show that (6.66) holds for an arbitrary periodic potential (Block's theorem).
Chapter 7
7.1
7.2
7.3
7.4
Express 1/r as an integral operator in the momentum representation. ______ Find the kernels for the expression of p 2 and Vp 2 f m 2 c 2 as integral operators in the coordinate representation. Evaluate the commutator p x V(r) ~ V(r)p x . Let ft(r,p) be a polynomial in the components of r and p. Show that
-
dp x
1i
7.5
and indicate explicitly the meaning of the partial derivative. l in powers (a) Find an expression for the operator (A - zB)~ of 2, where A and B are non-commuting operators.
PROBLEMS
(b)
517
Let A and B be IV x IV matrices and let zi be the root of which is smallest in absolute the equation det(A - 28) value. Prove that the expansion of part (a) converges whenever z < z\ and diverges whenever z > z\ |.
|
7.6
7.7
that in general any quantum mechanical operator can be expressed as an integral operator. Find an expression for the translation operator T a , defined by T a ^(x) - i//(x + a).
Show
7.8
7.9
The solution of the initial value problem of the time- dependent Schrodinger equation can be expressed in terms of an operinto ator U(t) that transforms the wave function at time the wave function at time t\ that is, i//(t) - U(t) i/y(0). (a) Upon what properties of the Schrodinger equation does this depend? (b) Express 11 (t) as an integral operator in terms of a complete set of energy eigenfunctions. (c) Evaluate the kernel of the integral operator V(t) for the motion of a free particle. The concept of the quantum mechanical average value has its natural generalization in the notion of the "inner product" in the representation space (see Appendix). The average value M - (MMVW is the inner product of V with the vector Mf. Demonstrate the following: /- I..., be a complete set of basis vectors (a) Let which is orthonormal with respect to the inner product, fi Show that the action of an operator M is (i, Wfr) completely determined by its matrix elements M^ - (V,,
t =.
z
fe.
(b)
(c)
(d)
An operator U is Hermitian if (^//O) - (OT,<t>) for all vectors V and <. Show that the eigenvalues of /] are real, the eigenvectors corresponding to distinct eigenvalues are orthogonal, and there exists a basis in which H is diagonal. Find the conditions on the matrix elements H k such that 11 is Hermitian. If (VJQV) for all vectors V and if (0 is Hermitian, then and $. One may then assert the oper(0, OT) = o for all ator equation (0 - 0. If (% A$) = (OT,) for all ^ and 4>, then B is said to be the Hermitian conjugate of A and is given the special notation B - A + . Find the relation between the matrix elements of A and A + Show that A is the Hermitian conjugate of A 4 and that a Hermitian matrix is self-conjugate, H = H+.
t
Express the Hermitian conjugate AC in terms A* and C\

(e)
of the operator product
Show
(<J),<f>)
that
>
if the inner product is positive definite, i.e., for all $ but the null vector, then the eigenvalues
of AA* are non-negative.

(f)
An operator
(/
is
unitary
if
<W,U<1>)
-=
(^,<J))
for all W and $.
518
If
PROBLEMS
= !..., is an orthonormal basis, then ^J = U^\ is the conditions on the matrix elements U k such Find also. that U is unitary. Show that U is nonsingular. (g) Show that the operators defined in problems 7.7 and 7.8 are unitary. 7.10 If A and B are N x N matrices and if AB = I, show that BA = N x N unit matrix. Construct a counter example / being the result is not true in general for "infinite to show this matrices, "i.e.,/V
%,
/,
>.
Chapter 8
8.1
Show
that the commutation relations between the operators for the velocity components of a charged particle in a mag-
netic field are v x u y - v y v x 8.2
H z and cyclic permutations.
Show that if V, and ^^ are energy eigenfunctions and if M does not depend explicitly on time, then the time dependence of offdiagonal matrix elements is givenby
at
where fau^
8.3
E J9
8.4
Let U(t) be the operator that generates the solution WU) of the time- dependent Schrodinger equation (with Hamiltonian H) from arbitrary initial data V(0) in the form (*) - U(t) See problem 7.8. (a) Show that U(t) satisfies the operator equation HU(t), with initial data U<0) = 1. (b) Deduce from the differential equation that U(t) is unitary. (c) Specify under what conditions and in what sense U(t) may be written as U(t) = exp(-i////^). In the Schrodinger picture observables are represented by operators that do not depend explicitly on time, the change of dynamic variables with time being described by the change of state vector V(t). In the Heisenberg picture the time dependence is transferred to the operators themselves by means of the unitary operator in problem 8.2. The procedure is M(t) = = (0) is the state vector ) where V (VU) M0)) - (W ,h(f) at / = and where ft() - U(tr l MU(t). - tfft, where M = U(tr l HU(t). If H is (a) Show thatiTztfu independent of time # = H. This is the Heisenberg equation of motion. - BA = /C, then (IB - I3Q = iC. (b) If AB Derive an expression for the time derivative of the op(c) erator product Q(f)@(t). (d) Find the position operator in the Heinsenberg picture for the motion of a free particle. Express the result in the coordinate representation.
f
PROBLEMS
(e)
519
Using the position operator in part (d), find the time dependence of the variance of the position of a freely moving
8.5
particle. that the mean value of the kinetic energy in a state belonging to the discrete spectrum is related to the mean value of the potential energy by the relation 2T = (r VV). If is a a homogeneous function of degree a, V(Ar) = \ V(r) f then 2T = aV
Show
8.6
(virial theorem). Verify the following
"sum
rules."
2
(b)
Z]
*
6>
lfc
r,*
("F-sum rule
2
2m
fe
mo
e ) 2- <^^
T tk
diverges
ljfe
= E, - E*. Note that <y > Here 7* if k designates the ground state. Sum rules are important because they provide a means of testing the physical content of the theory even though the mathematical problem cannot be solved comljfe
pletely.
Chapter 9
9.1
Formulate the set of coupled differential equations governing the population of a large number of quantum states. Show that the total number of levels occupied does not with time.
change
Identify the conditions under which the occupation numbers are positive definite. Derive the relations among the Einstein
coefficients
which specify the approach
to
thermodynamic
equilibrium.
9.2
Compute
the
current density of the non-stationary state
which describes the transition from the 2p to the Is state of hydrogen and calculate the radiation field from this source. Identify the angular distribution and state of polarization for each value of the magnetic quantum number m. See Chapter 13.
Chapter 10
10.1
At
oscillator
what quantum level can a wave packet in a harmonic be localized to 1% of the total excursion? Use the
10.2
uncertainty principle. Find the stationary states of a linear harmonic oscillator in a uniform electric field.
520
10.3 Suppose
-
PROBLEMS
that
is
at
= 0,
a particle in the potential field V

^(x^))
22 2 mo> x
2
described by the wave function
= const.
the
exp
iikQX -
-a 2 (x
- *o) 2
where
a2 =
mo>/X
Calculate
probability amplitude for the particle being in each of the energy eigenstates. Find the wave function for / > 0. Discuss the spreading of the wave packet. 00
10.4
Show
that e
- s **
2st
n= o
-" n
1
H n (0
is the
generating function for

it
Tchebycheff- Her mite polynomials. Use appropriate orthonormality relations.

x00
to establish the
10.5 10.6
Show
that
ll
n (x)e-**e"'*d*
- \/n
m m e -P 2/4 . p
10.7
10.8
10.9
Infer the matrix elements of the position operator of a harmonic oscillator from the spectrum, selection rules and the F-sum rule. Solve the Heisenberg equations of motion to find the timedependent Heisenberg operators for the position and momentum of a harmonic oscillator. Find the time-dependent Green's function for the harmonic oscillator. Normalize by comparing it to the free particle Green's function at a time ait -' 1. Study the harmonic oscillator using the operator properties of the variables
a V<
\fCh/2mcj) d/dx
x -
Prove the following statements:

(a)
The Hamiltonian operator
is
- a* a +
where the energy
(b)
is measured in units of 7?w. The commutation relations
aa* - a + a
Ha - aH ^ -a
Ha + - a f H - a +
hold as operator equations. The Hermitian conjugate (or adjoint)
(c)
(d)
H is Hermitian (selfadjoint). If U r is an energy eigenfunction with eigenvalue c, then is an eigenfunction with energy e - 1 and (a*) n U e is aUe an eigenfunction with eigenvalue t + n. This gives a series of equally spaced energy levels.
of a
is
PROBLEMS
(e)
(f)
521
There is a finite lowest eigenvalue. The series must terminate at the lower end by arriving at a "ground state" UQ that satisifes the equation alJQ ^
(= the null vector). Solve this differential
equation for UQ
(g)
and normalize. The "excited states" are

(n + ~\
(;
- (n\y l/z (a*Y Uo

l
with energies
Tzw.This gives a complete set of orthonormal eigen-
functions.
(h)
Determine the matrix elements of a, a + x and p by algebraic means. Show that there are no finite dimensional matrices which satisfy the specified multiplication prop,
erties.
Chapter 11
11.1
Prove the orthogonality and normalization condition of Legendre polynomials using the generating function (1 - 2rx +
r
)-*
n-
rPn M.
11.2
Derive the commutation relations between the components of angular momentum and the components of (a) the position operator, and (b) the momentum operator. 11.3 Let A x A y and A z satisfy the commutation relations L x A y A y L x - iHA Z9 L V A X - A x L y - -iHAz, L r A T - A x L x - 0, andcyclic permutations, with respect to the angular momentum operators. Such operators are called "vector operators"; examples are r and p. Prove the following relations: 2 - A2 L = K (a) L X A L) - (A L)L X (b) L X (A 2 - iH(A 2 - A,L V - L V A,) v Lt + L z A y (c) L A X - A X L
, . --
(d)L'
/^v
2 2 2 (l*A x - A.L ) - (L A X - A X L 2 )L 2 2 4L X (A.L) 2(L A X + A X L )

'
-=
i/i
f
I\/AN"*^
>
'
'
(e)/(/
D(^
/m
(L
jr
A) nl (L) lm
A \ '*
i '
'
/ r
"t
.
11.4
11.5
Show
mean
that in a state v> with a sharp value of L^, L e values of L x and L y are zero.
<//
TO^,
the
Suppose a system can be resolved into two weakly interacting subsystems 1 and 2 so that the total angular momentum L is LI + L.2. If the subsystems are in states characterized by definite values of the quantum numbers /i, l\ z and /2 h* re2 spectively, what _are the possible values of L and what is the 2 of ? L average value 11.6 Express the spherical harmonics for 1 = 0, 1 and 2 as polynomials in x, y and 2. 11.7 Find the transformation rule of the spherical functions Y n> YIQ and Y 1-1 for a rotation of the coordinate system through Eulerian angles a, ft and y.
522
11.8
PROBLEMS
Let L - L x
(a) (L
iLy.
Verify the following relations:
LT
(b) (c) (d)
(e)
(f)
L2 = L2 L L+LL2 Lt L2 L 2
L2 + <L+L- + L-L+) - L L2 HL - L-L+ - 2fiL 3 - L L2 f - LsL 2 =
11.9
Study the
way in which the multiplication properties in problem 11.8 can be represented in terms of matrices, or what is the same thing, linear transformations on a finite dimensional
Show
the following:
.
vector space. Let V\^ be a finite set of degenerate eigenfunctions of L 2 with eigenvalue A. Here n labels the elements
of the set.
(a)
(b)
(c)
(d)
(e)
(f)
(g)
(h)
This implies that A has the form /(/ + l)7z 2 where / may have the values 0, 1/2, 1, 3/2, .., and that m runs over range -/,-/ + 1, ...,/- 1, /. Find the matrices of order 2/ + 1 that represent the action of the operators L z and L on the vector space spanned by m - ' ôr a ^ xec^ value of /. For example, ^/m' *"' = m ') V ( P ,L 2 /m /m ^mm'' Select a phase convention ('Vmm' compatiable with (11.88). Write out the explicit matrices for / = 1/2 and verify the commutation relations by direct matrix multiplication.
,
l
the condition that the norm of L m is nonnegative, 2 that A > m(m l)^ the follows (respectively for ) the if vector is null. equality holding The condition for the existence of finite multipletes is n terminates at both ends. that the series generated by (L )
it
,
an eigenfunction of L 2 is some linear combination of the Vs which is an eigenfunction of L^. Denote the eigenvalue by mU and the particular linear combination by ^\ m . 2 L+V\ m is an eigenfunction of L with eigenvalue A. L t W Xm is an eigenfunction of L z with eigenvalue (m l)7z.
L z VXM There
is
From
Chapter 12
12.1
Find the energy levels and eigenfunctions of an isotropic three-dimensional harmonic oscillator. Find energy eigenfunctions that are also eigenfunctions of L 2 and L 2 . What is the degeneracy of each level? What part of the degeneracy
of the potential? out the explicit / = I, m = 1, O - 1, eigenfunctions of lowest energy of a three dimensional spherically symmetric harmonic oscillator. Express these eigenfunctions as linear combinations of the solution obtained by using spearation of variables in Cartesian coordinates. If each triplet of wave
t
stems from isotropy
12.2 Write
PROBLEMS
523
12.3
functions is orthonormal, show that the matrix of coefficients of the transformation is a unitary matrix. For the isotropic potential V(r) = -g8(r - a), determine the
range of g for which there

12.4
= 1 discrete state. In the case of a continuous
/
is
an
discrete state but no
from
spectrum, the wave function far ' the scattering center is characterized by the 'phase
shift" 8i(k) defined as follows: rR k j(r)

for
kr
/
c sin (kr - !z +
S/(fe))
range" Chapter
in
= 0), where Hk = \/2mE k . For a "short potential, S/ is independent of r (see page 222 and phase shift for the potential 28B). Calculate the / (kr
1
for
problem
12.3.
t
3 ~J Ji + i&(p) j (p) occurs frequently and is (\Vi called the "spherical Bessel function." The second solution,
irregular at the origin p =
0,
is n/(p> =
-W)'
(
)
" ^/.^^P^ con

(J p co8rrp
sistent with the definition of Neumann functions, N p = J.p)/sin77p. Verify the following properties: (a) differential equation,
/, 2
\
"
(b)
explicit form,
sin
~P
sin p
cos p
cos p
sin
(c)
recurrence relations,
-
/-I
21 + 1
.,
,
+
p
p
(d)
J;
Jt
asymptotic form
/ .
(p
/,
1),
;/ (p)
) w/ ( p
L ^ ___ COS
/
^p
/77
^ J
(e)
behavior at P
= 0,
2'Z' A(*
7
524
PROBLEMS
(f)
integral representation (see page 202),

+i
1
2 -/-i
(g)
orthonormality,
2
j(/er)r
dr
8(k'k')
12.6
Find the conditions under which the spherical potential well " r for r -> a, can support IV V for a, and V = _| VQ
_-_
--
s-wave bound states, but not

12.7
In
Nil.
zero kinetic energy, an s-wave phase shift behaves as, 8 (k) - -ka, where a is the "scattering length." Find an approximate relation between scattering length and the energy of a loosely bound state, KH 0.
the
limit of
Chapter 13
Find the momentum distribution of an electron in the Is, 2s and 2p states of hydrogen. 13.2 Calculate the lifetime of a hydrogen atom in the 2p state. 13.3 Give an expression for the transition rates between consecutive circular orbits in a hydrogen-like atom.
13.1
13.4
Show
that
(1
-1
exp|\f/(l - 01 k=
-1
,
<fr)
Lk
Mt k
is the
gen-
erating function for Leguerre polynomials L^(x) - Q^(*). At what quantum level is a muonic atom the same size as a normal hydrogen atom in its ground state? 13.6 What is the probability of finding an electron with quantum numbers n, I inside a nucleus of radius R n ? 13.7 Show that if the energy is regarded as a complex variable, the Coulomb "scattering amplitude'* for a definite value of /,
13.5
has a simple pole singularity at the energy \)/2tk, value corresponding to each level of the discrete spectrum.
\c
l<t
Chapter 14
14.1
Evaluate the shift in energy levels of a harmonic oscillator produced by a perturbing 5-function potential that is centrally located. State the limit of validity of the approximation. 14.2 Calculate the lowest order effect on the spectrum of a linear harmonic oscillator due to the relativistic increase in mass
PROBLEMS
of the particle. In
525
what circumstance is the relativistic effect pronounced? When is the perturbation approximation valid? Cf. problem 2.10. 14.3 Treat the Stark effect of the n -= 3 level of hydrogen. Determine the pattern of splitting and residual multiplicity by symmetry arguments. How does the secular equation factor in the representation in which L z is diagonal? What linear combinations of unperturbed eigenfunctions are energy eigenf unctions in the presence of the field?
14.4
14.5
electronic polar izability of a hystate. For the continuous part of a spectrum, the states may be labeled by the eigenfunctions of the free particle Hamiltonian
Evaluate the static

its
(o>
0)
drogen atom in
ground
H and an (exact) integral equation for the eigenfunctions W k r of the Hamiltonian f/o f V set up along the pattern of perturbation theory. The integral form of the Schrodinger
(
equation
is
W k (r)
where GR
exp/k-r
is the
energy Green's function

4n\r ,
r
|
characteristic of the free particle Hamiltonian. The Green's function can be defined in terms of a complete set of energy eigenfunctions n (r) of the (any) Hamiltonian HO as follows:
(a)
(b)
Formulate a differential equation and boundary conditions for the Green's function. Evaluate the explicit form of the free particle Green's function from its definition in terms of a complete set of
energy eigenstates. Find the relation between the energy Green's function and the time Green's function G t (See problem 5.3.) The effect of a "short range" potential on the wave func.
(c)
(d)
tion at large distances from the scattering center is described by the "scattering amplitude f(0) 9 " defined by the
asymptotic form (for
e lk
'
f(0)
(See Chapter 28, Section B) Derive an exact expression for in terms of the exact solution of the Schrodinger k
equation.
526
(e)
PROBLEMS
14.6
Formulate a perturbation approximation for * k and f(0)on the basis of the above results. For a definite value of the angular momentum the integral from of the Schrodinger equation for a spherically symmetric potential takes the form
u ki
(r)
= sinkr +
n*
/ JQ
gEj(iy') V(r')uki(r')dr'
where
SEQ(r s'} =
\sink K L
\
r'
|
- sink
r
\
r'
\
for
(a)
(b)
if
= 0.
/
Evaluate gEi(rs') for
> 0.
between the wave functions and between the Green's functions of the three-dimensional (problem 14.5) and the partial wave formulations. It is necessary to distinguish ingoing and outgoing wave boundlkr /r. The addition formula for spherical ary conditions e harmonics is helpful,
What
is the relation
21 f 1
m=-/
where
(c)
cos
= cos0 cos0' + sintf
An exact expression
for the s-wave phase shift is
T
/
/I
00
si sinfer
/Q
Verify this relation and generalize it for I > 0. What normalization condition is implied for u k ? See problem 12.4.
Chapter 15
15.1
Identify the conditions under which the Klein-Gordan equation reduces to the Schrodinger. How does the initial data problem for the second order differential equation reduce to that of the first order equation? Show that the function- space inner product
15.2
**" '/[*
is
I**
if
f -IT
and
are solutions of the Kleinof the inner product in mo-
independent of time
Gordan equation. Find the form
mentum
space. Is the inner product positive definite?
PROBLEMS
15.3
527
A general solution of the Klein-Gordan equation is a superposition of a positive-frequency and a negative-frequency ta)t and e* "', respectively. part, with time dependence e~ Show that under a Lorentz transformation that does not reverse the sense of time, a positive-frequency solution is transformed into a positive- frequency solution, i.e., that the decomposition is Lorentz invariant. 15.4 Find a complete set of positive- frequency solutions of the Klein-Gordan equation that are orthonormal with respect to the inner product in problem 15.2. Show that the completeness relation (for unequal times) is
1
where
<y
- <u(k) - cVfc 2 +
m2c2/^2
and where the Minkowski

t'}
space inner product
is k(x -x') = <^(t -
- k
(\
- \).
Chapter 16
16.1
What
is the
value of the g factor of a particle described by

)
the Hamiltonian (l/2w
[V
(p
- (e/c)
A)]
16.2
particle of spin 1/2 and magnetic precessing magnetic field
moment
Hz
=
p,
moves
in a
Hx
H smO
cosa)t,
Hy
sintfsinojf,
H cos0.
At time t - 0, the spin is parallel to the z axis. What is the probability of the spin being antiparallel to the z axis at some later time? 16.3 Find the (time- dependent) position operators in the Heisenberg representation of a particle of spin 1/2 and magnetic moment u moving in a nonhomogeneous magnetic field
Hx
16.4
= 0,
Hy
= -fey,
H2
= HO + kz.
particle
Determine the energy spectrum and wave functions of a charged moving in uniform electric and magnetic fields that
are perpendicular to one another.
Chapter 17
that the four matrices (/',o') are Hermitian, linearly independent and form a complete basis for 2x2 matrices. 17.2 How many matrices are required to form a complete set of Hermitian, linearly independent NxN matrices? How many
17.1
Show
528
PROBLEMS
mutually commuting, Hermitian, linearly independent matrices are there? 17.3 In the Dirac theory, a and p are mutually commuting linear operators. Develop a notation for the linear vector space on which the operators act. Find an expression for the inner product in the composite (direct product) space. Show that the Dirac Hamiltonian is Hermitian. 17.4 Use the Heisenberg equation of motion to show
(a)
--(x
i
iKp 3 a/2mc)
2
zfipg/2/nc)
-=
p 3 (p
- (e/c)\)/m
(b) ---(/
p 3 (U
- e<D)/mc 2
17.5
What transformation to restore the form

gauge
of the spinor wave function must be made of the Dirac equation after a change of
17.6
Show
in the electromagnetic potentials? that the form of the Dirac equation
remains unchanged
by Lorentz transformation or spatial rotation.
Chapter 18
18.1
Suppose a system consists of two weakly interacting subsystems each of spin 1/2. The total spin is S = si + 82. What are the possible eigenvalues of S 2 and S 2 ? Compute the value s 2 in the triplet (spins parallel) and singlet (spins of si antiparallel) states of the composite system. Find eigenfunctions of S 2 and S^ as linear combinations of products of eigenfunctions of the subsystems,
18.2
An electron moving
in a central field of force is in a state
specified by the quantum numbers ljm jm What are the possible values of the z components of orbital and spin angular momentum and what is the average value of each? 18.3 Is the parity operator / a linear operator? Is it Hermitian? What are the commutation properties of / with the operators
r,
L, S
and
,1?
18.4
axial vector transforms as a vector under proper rotations (i.e., rotations without space inversion) but does not change sign under inversion. Classify the following as being either vector or axial vector: 1% H, A, S, L, p and u x v
An
18.5
where u and v are vectors. Show that if a system is in a state characterized by a sharp value mj for the 2 component of the total angular momentum,
the
z'
mean
making an angle
value of the total angular with the ^ axis
momentum
is
about an axis
m}
cos
0.
Evaluate the particle flux of positive energy and negative energy plane wave solutions of the Dirac equation. Also calculate the flux of the corresponding charge conjugate solutions. 18.7 Can the charge conjugation transformation be represented by a linear operator?
18.6
PROBLEMS
529
Chapter 19
19.1
19.2
j for a Dirac particle in an electromagnetic the Lorentz force is given correctly. that Verify Calculate the spin-orbit and contact potentials for an electron outside a closed atomic core consisting of a hydrogen-
Evaluate
and
field.
like
atom.
Chapter 20
20.1
weak
Calculate the energy level splitting of a hydrogen atom in a electric field. Neglect the Lamb shift and assume the Stark effect is small in comparison with the fine structure. Account for the latter by using eigenfunctions of J 2 J z andL 2
,
for the unperturbed states.
20.2
Evaluate the Stark effect for the n

for the case
2, j
1 level of hydrogen
where the Stark effect is comparable with the Represent the Lamb shift by a phenomenological perturbation matrix element that touches only the s- state. 2 in the 20.3 Study the hydrogen Stark effect for the level n transition region where the Stark effect and the fine structure are of the same order of magnitude. Plot the energy levels
Lamb
shift.
--
20.4
20.5
Compute
as a function of P. the lifetimes of the 2py and 2p^ states of hydrogen.
Find an expression for the lifetime of the metastable 2s^ state of hydrogen in a very weak electric field. At what field 3 strength will the lifetime by 10" sec? Is the Stark level shift this at field appreciable strength? Hint: Evaluate the matrix
element of
using a perturbed 2s^ wave function that constate and take advantage of the
tains an admixture of the 2p
20.6
smallness of the Lamb shift. Find the mean value of the operator u - L + g s S in a state characterized by the quantum numbers J, J 2 L, S. (This gives a generalization of the Lande formula in the case where the g factor of the electron is not exactly equal to two. See page 346.) Hint: Use problem 11.3(e).
,
Chapter 21
21.1
Determine the form of the contact interaction when the

nuclear size
the
It
finite
is
taken into account. Calculate the splitting of
2s-*--2pl
levels of hydrogen
0.1
stemming from
this effect.
is
about
Me.
530
PROBLEMS
21.2 Calculate the hyperfine splitting of the ls^ state of hydrogen

i
using a classical model in which the nucleus is represented by a uniformly magnetized sphere of radius %. Give a classical explanation of the sign of the splitting. 21.3 Find the magnetic field at the nucleus of an s state electron. 21.4 Consider a hydrogen atom in the ground state in a uniform magnetic field. Find the appropriate linear combinations of electron- proton spin wave functions that are energy eigenfunctions in the case where the interaction with the external field is the same order of magnitude as the electron- proton dipole-dipole interaction. 21.5 Calculate the hyperfine splitting of a hydrogen-like atom in a state of non-zero orbital angular momentum. Hint: Use
problem
Chapter 22
11.3(e).
Specify a complete set of commuting constants of the motion for positronium. 22.2 Calculate the fine structure of positronium. Obtain the
22.1
Hamiltonian by semiclassical arguments. Account especially ' for "hyperfine' splitting, knowing that the magnetic moment of the positron is equal in magnitude and opposite in sign to that of the electron. Present the results in an energy level diagram for n = 1,2. 22.3 Evaluate the Lamb shift for positronium. 22.4 Construct a theory for positronium using a Dirac Hamiltonian for each particle and the Coulomb interaction between them. Separate the motion of the center of mass from the relative motion in the approximation of retaining only the lowest order relativistic corrections for the relative motion. Does the "hyperfine" interaction emerge automatically? Hint: The action of spin operators on a product wave function can be expressed as follows:
Chapter 23
Construct a complete set of orthonormal two-particle eigenfunctions from a complete set of one-particle eigenfunctions. 23.2 Derive the Hartree-Fock equations for determining the best single-particle functions to give an antisymmetric (or symmetric) two-particle wave function of lowest energy. 23.3 Calculate the n - 1 and 2 levels of a hydrogen-like atom, including fine structure and the Lamb shift, using a variational
23.1
method.
PROBLEMS
23.4
531
23.5
= method with trial functions if the a A's are 2A e~ can be reduced to matrix procedure regarded as the variational parameters. The a's may be selected by intuition or by an itteration scheme. Note how this gives a finite set of orthonormal vectors. Evaluate the ground state energy of helium using the following "self-consistent" variational method: assume the first electron is in a hydrogenic Is state with Z = Z' and calculate the screened field seen by the second electron. Describe the second electron by a hydrogenic Is function with Z = Z". Vary Z" to find the lowest energy for a given Z'. Then impose
Show how
a| r
'
the
variational
i//
the
symmetry by
setting Z' - Z".
Chapter 24
24.1 Give the possible values of the total angular momentum for 3 2 the following states (terms): 1 S, 2 S, 3 S, 2 P. P, D and 4 D. 24.2 Which terms are possible for the following two-electron configurations:
=
'?
(a)
nsn's,
(b)
nsn'p,
Which terms are consistent with
(c) nsn'd, and (d) np, n'p? the exclusion principle if
24.3 Couple three unit angular momenta /i = /2 = /a = 1 to yield a 2 of L , where L = li + \2 + la. resultant eigenstate L = 1, L* = How many independent states of this sort are there? 24.4 Estimate the low-lying excited state energies of helium in the approximation of neglecting exchange effects. Do this assuming that one electron is described by the Is function found in the ground state calculation and carrying out the variational procedure for a hydrogenic 2s (and independently, 2p) wave function in the screened Coulomb field. The 2s state must be taken orthogonal to some appropriate Is state. The root of the variational equation may be found by a rapidly converging iteration method. 24.5 Calculate the low- lying excited state energies of helium taking
account exchange and spin effects. Use appropriately symmetrized product eigenfunctions for the configurations (Is, 2s) and (Is, 2p), where the Is function is that found in the ground state calculation. The required matrix elements are given in problem 24.6 for reference. 24.6 Consider the hydrogenic wave functions
into
6(1 -
where
a - Zi s /ao, /3 - Z2 S /2ao, and y - Z2 p /2ao and where 6, B and c are determined by the orthogonality and normalization
532
PROBLEMS
conditions. Verify the following one- and two-particle matrix
elements:
T 2p
<
.
?L
I/
Vgp
f'
>7 e 2Z Y O
a 2 y 2 <a 2 + 5ay + lOy 2 ) 6 (a +

28e
3 (a
2 8_ 5
y)
2 a2
I-
y
5
y)
a 2 -aft +
a2 " \
<u + ^) 2
a^
2
/3
,
2(a
3
/3)
(a
a/6
2
j3
)
(a +
5
/S)
(a
aft + ^8)
24.7
Calculate the ground state of lithium by
means of
a variational
method based on an an ti symmetrized product wave function for the configuration (Ls 2 2s). The appropriate matrix elements can be reduced to those given in problem 24.6.
Answer: The best variational wave function is given by Zis = 2.694 and Z 2s = 1.534. The energy is
200.8 eV compared with the perturbation theory value 190.8 eV and the observed value- 202.54 eV.
24.8
(Is 2p) excited state of lithium by the variational method, (a) neglecting exchange effects and assuming a helium core, (b) neglecting exchange effects and using the 2 Ls core found in problem 24.7 and (c) taking exchange effects into account using the antisymmetrized wave function for the
Calculate the
configuration
(Ls
2p).
Chapter 25
25.1
Show
in the
that the virial
theorem 2T Ve - e + Vn - c Thomas- Fermi model. Does it also hold

\
is fulfilled in the varia-
tional solution?
PROBLEMS
533
25.2 Using the virial theorem, show that in the Thomas- Fermi model of a neutral atom the energy of electrostatic repulsion between the electrons is 1/7 the magnitude of the electrostatic attraction between the electrons and the nucleons. 25.3 Estimate the order of magnitude of the following quantities in a neutral atom according to the Thomas- Fermi model: (a) the size of the atom, (b) the average electrostatic repulsion between two electrons, (c) the average kinetic energy of one electron, (d) the average speed of an electron, (e) the average angular momentum of an electron, (f) the mean radial quantum number. 25.4 Show that the mean perturbation of all states of a given term is zero for the spin- orbit interaction. 3 25.5 Calculate the L S splitting of the P term of helium. S L of a s Evaluate the 25.6 ingle np electron in a spherisplitting the result in terms of an potential. Express cally symmetric arbitrary radial matrix element. Estimate the radial matrix element for the doublet splitting of sodium. 2 25.7 Two electrons move in an (np) configuration in a spherically symmetrical potential. Regarding the electrostatic repulsion between the electrons as a perturbation, evaluate the splitting of terms in first order approximation. Neglect the spinorbit interaction. Use qualitation considerations to infer
ordering of the terms. Hint: To facilitate diagonalization of the secular equation, use a representation in which M^ and MS are diagonal and note that the sum of the roots of a secular equation is equal to the sum of the diagonal matrix elements.
Chapter 26
Find the possible atomic terms (a term is characterized by L and S) in a configuration of two equivalent d- electrons. Give the total number of states and the number of states in each term. What values of J are possible to each term? 26.2 In an atomic configuration the term that has the lowest energy can be determined by Hund's semi-empirical rules: (1) The ground state will have the largest value of S consistent with the Paul! principle, (2) L will have the largest value consistent with the value of
26.1
(3)
determined in rule (1), The total angular momentum

S
of the
ground state
is J
|L - S| if half full;
full.
the
it
unfilled
is J -
subshell is half full or less than S if the subshell is more than half
Give a qualitative physical justification for each of these rules.
534
PROBLEMS
26.3 Using Hund's rules (see problem 26.2), find the ground state of the configuration np* for x = 1, 2, . . . , 6, For each value of x, state an element for which this case is realized physically.
26.4
Do problem 26.3 for the configuration
nd x for x =
1, 2,
. .
10.
Chapter 27
Estimate the relative frequencies and separation of energy levels for the electronic, vibrational and rotational motions of a diatomic molecule. 27.2 Derive the Schrodinger equation describing the motion of the nuclei of a diatomic molecule in the approximation that the nuclei move much more slowly than the electrons and thus experience only an interaction with the electrons that is averaged over many electron revolutions. This procedure provides a separation of variables between the electronic and nuclear motions. The approximation is called the adiabatic
27.1
27.3
or Born-Oppenheimer approximation. What are the possible symmetry states of the diatomic molecules D2, N 2 LiH formed from the bonding of the two atoms
,
ground states? 27.4 What spin symmetries are possible for the rotational states of the deuterium molecule D 2 in the electronic ground state? The deuterium nucleus has spin 1. 27.5 Calculate the energyôf a rigid electric dipole in a uniform electric field. Use second order perturbation theory. 27.6 Show that the force between two hydrogen atoms in their 7 ground states varies as l/R if the atoms are separated by a
in their
large distance R.
Chapter 28
28.1
The rate for making transitions from an

final state
f is
initial state
to the
Wlf=
T |W
2/7
lir
'
where
p f is the number of final states per unit energy interval ("Golden Rule No. 2"). in first order perturbation (a) Find the expression for U' lf theory. What normalization convention is implied for the
wave function?
(b)
Derive an expression for

theory.
M^
in
second order perturbation
PROBLEMS
(c)
535
28.2
Evaluate the density of states p f for a final state consisting of two free particles of definite total energy and momentum. (d) Evaluate p f for a final state consisting of three equal mass (free) particles. Express the results in the center of mass system. How is the partition of energy among the three particles accounted for by a probability distribution? (e) Calculate the dependence of pf on energy near the energy threshold for an N-particle final state. Calculate the rate for induced transitions from state m to state n of an atomic system in an electric field with spectral
density 8
the
(o>)
at transition
frequency
<
1 (E m
n
- E n ). Identify
Einstein coefficient B and thereby infer the rate for spontaneous emission. 28.3 Consider a particle of mass m bound in a three-dimensional
harmonic oscillator potential ~kr 2 9 The particle z

with a mild pulse
t
is irradiated
<
oo.
< for the time interval Determine which transitions are possible and calculate
&' - ex 2 e~ (t/r ^
their probabilities. Identify the limiting cases of sudden and adiabatic perturbations. State the limits of validity of the
28.4
perturbation calculation. At t = O a hydrogen atom in its ground state is irradiated with a uniform periodic electric field. Determine the minif
mum frequency of the field necessary to ionize the atom and compute the ionization probability per unit time. As an approximation, the electron in the final state may be regarded
28.5
as free. State the limits of validity of the approximation. Show that the scattering of slow particles in a short range 2l+l . Find the proportionpotential is characterized by d t (k) ~ k ality constant in Born approximation and state the conditions
28.7
under which the approximation is valid. Taking into account the symmetry of the wave function, give the differential cross section for elastic Coulomb scattering of an electron on an electron and of an alpha particle on an alpha particle. Distinguish spin states and also give the formula for the scattering of unpolarized electrons. Identify quantum effects and show how they disappear in the classical limit. For reference, the exact scattering amplitude for a
fixed
Coulomb
fc (0)
potential is
2
= _
. "
Fd
iy)
2sm2|
t
HI-
^ "
e -2i-ytn
sm
0/2
iy)
28.8
where y = e$/Hv. Cf. Chapter 13, Section D. Compute in Born approximation the differential scattering cross section for the scattering of fast neutrons by a Coulomb
field.
536
28.9
PROBLEMS
Set up coupled Schrodinger equations to describe the "two channel" reaction and scattering processes
(11) a + b a + b
c
}
>
a + b
(12)
(22)
>
c + d, c
+
rf,
(21)
c + d
a + 6.
Assume the particles are spinless and of unequal mass. Separate out the motion of the center of mass. Vi 1,^12, ^21,^22 be related in (a) How must the potentials order to describe a system that is invarient with respect to reversal of the sense of time? How are the cross sections for reactions (12) and (21) related? (detailed balance) (b) Give expressions for the effective cross sections for these reactions in Born approximation. (c) Formulate an expression for the reaction and scattering amplitudes in terms of partial wave amplitudes of definite angular momentum. What relation among fi i J\ 2, ( 21^22 (f r a definite value of /) is implied by conservation of probability?
Note:
matrix
The results are presented most simply in terms of a notation. The many-channel generalization of the 2
'
quantity
5/
e>
is
called the S- matrix.
Chapter 29
29.1
Verify the antic ommutation relations

a <r,f)0< P ;) +
i-
^(K/)<k<r,f)
tt/
,S<r-f)
y^(r,0i/^(r,'*)
^(^t)^^)
--
satisfied by the second quantized Dirac wave function. The a is the ft indicate components of the Dirac spinor and / a/? unit and + matrix. The positive-frequency parts of describe the creation of positrons and electrons .respectively, while the negative- frequency parts describe the annihilation of
and
4x4
<//
i//
electrons and positrons, respectively. Electrons and positrons are described by independent sets of mutually anticommuting creation and annihilation operators
,
29.2
Evaluate the commutation relations between components of current and charge density of a quantized Dirac wave field, i.e., evaluate j^{r,0jv (i#) - 7 v (r;0; (r,0 at equal times.
the
M
PROBLEMS
29.3 Calculate
field
0(r,) =

537
for the
the
Born approximation matrix element
production of an electron- positron pair in an external electric
COS kt COS a)t
Express the energy and momentum operators of a quantized Dirac wave field in terms creation and distruction operators. 29.5 Let M(x) be a massive pseudoscalar meson field satisfying the Klein-Gordan equation. Calculate the force between two stationary Dirac "nucleons" in second order perturbation theory, assuming the interaction energy between the "nucleon" and meson fields is
29.4
H' - g /
i/

Quantum Mechanics PDF

Uploaded by

Copyright:

Available Formats

Quantum Mechanics PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Quantum Mechanics PDF

Uploaded by

Copyright:

Available Formats

TIGHT BINDING BOOK

(Perturbation theory for the continuous

Phase shift. Connection between men sional formulations )

wave boundary conditions. partial wave and three di-

RELVriVISTIC QUANTUM MKCII \NICS

in the presence of an electromagnetic field.

Magnetic Eield. Electron Spin

Classical theory of the

Eigenvalues of the spin operator along an arbitrary direction.

The Uirac \\ave Equation

second quantization. Transformation proper-

ties of the spinor

wave function under Lorentz transformations

and spatial rotations.

The Dirac Theory of the Motion of an Electron

The Dirac Equation

Structure of the Spectra of Hydrogen-like

22. 'The Electron -Positron

Elementary Theory of Multielectron Atoms Including Spin States

Diamagnetic susceptibility of parahelium.

Mendeleyev's Periodic System of Elements

The discovery of Mendeleyev's periodic law.

The Theory of Simple Molecules

SOME APPLICATIONS TO NL1CLFAR PHYSICS

28. Elastic Scattering of Particles

'Time-dependent perturbation theory. Golden

for elastic scattering. Uncertainty of energy. Scattering ampli-

Scattering from a spherical barrier and

Resonant scattering. (Golden Rule #2, Density

Second Quant t/.ut ion

Fermi and Gamow-Teller selection

Hilbert Space and Transformation

This textbook is based on

cow Regional Pedagogical University from 1945 on.

and cumbersome that

perceive the physical mean-

As magnetic monopoles do not exist by means of magnetic dipoles.

c^3-10 cm/ sec. mained unknown.

waves, however, re-

mass m of a moving its rest mass m by the

The Quantum Theory of Light

PRINCIPLES OF THE ELECTRON THEORY

where E and //are the electric and magnetic

charge density (for example, of the electron),

NONRELATIVISTIC QUANTUM MECHANICS

whereas the interaction between moving charges

equation, which holds for the

any component of the vectors

or//. the Maxwell- Lorentz equations

THE QUANTUM THEORY OF LIGHT

We can select a Lagrangian function the variational principle

NONRELATIVISTIC QUANTUM MECHANICS

Therefore, the relativistic form of the Hamiltonian is

the potentials are time- independent , the Hamiltonian

is equal to the total

In the nonrelativistic approximation written in the form

Eq. (1.14) can be

H' is approximately equal to the nonrelativistic part of the

THE QUANTUM THEORY OF LIGHT

= #2 = #- = nk,