QUANTUM MECHANICS SECOND EDITION EUGEN MERZBACHERWiley International Edition QUANTUM MECHANICS SECOND EDITION EUGEN MERZBACHER University of North Carolina by, Se 3 o 30 John Wiley & Sons, Inc., New York « London + Sydney + Toronto Toppan Company, Ltd., Tokyo, JapanAuthorized reprint of the edition published by John Wiley & Sons, Inc., New York and London. Copyright © 1961, 1970, by John Wiley & Sons, Inc. All rights reserved. No part of this book may be reproduced by any means, nor transmitted, nor translated into a machine language without the written permission of the publisher. Wiley International Edition This book is not to be sold outside the country to which it is consigned by the publisher. Library of Congress Catalogue Card Number: 74-88316 Printed by Toppan Printing Company, Limited, Tokyo, Japan.To my Father xerMing a Oo Y “eye, % o}werMing, a 2 Oo Y “eye,Preface In a time of rapidly increasing specialization, physics appears to have suffered less damage from the fragmentation of knowledge than many other scientific disciplines. Such cohesion may be accounted for by the central role of quantum mechanics as a unifying principle in contemporary physics. This function of quantum mechanics is reflected in the training of physicists who take a common course, whether they expect to specialize in atomic, molecular, or nuclear physics, in solid state physics, or in the physics of elementary particles. This book—a revised and expanded version of a book first published almost ten years ago—was written for such a course. It is intended to be a comprehensive introduction to the principles of quantum mechanics and to their application in a variety of fields to which physicists turn. The material contained in this new edition is appropriate for three semesters (or four quarters). The first eighteen chapters make up the content of a two- semester (or three quarter) course on nonrelativistic quantum mechanics. Chapters 19-24 discuss the elements of formal scattering theory, second quantization, relativistic quantum theory, and other subjects, sometimes classified as “Advanced Quantum Mechanics,” which provide the back- ground for an understanding of quantum electrodynamics, theories of elementary particles, and many-body theories. The book could also be divided by drawing a line between the first eleven (or thirteen) chapters, containing the elementary theory which is often taught in one semester to advanced undergraduates, and the remainder, starting with Chapter 12 or 14 and covering about two semesters (or three quarters) of graduate-level quantum mechanics, with standard wave mechanics (and perhaps simple spin theory) as a prerequisite. Although the text has been revised in many places—even merging two chapters (old 10 and 11) into one (new 10) and dividing a long chapter (old 13) into two (new 12 and 13)—the first fourteen chapters of the two editions remain in close correspondence, with one exception. To Chapter 4 has been added a section on the variational principle which leads to the SchrédingerPreface equation and to first-order perturbation theory of discrete stationary states. This device has made it possible to move many problems and illustrative applications to earlier chapters. A number of new problems have been added in the second edition. As before, the book begins with a glance back into history and a preview of the ideas to be encountered and develops ordinary wave mechanics inductively with the aid of wave packets moving like classical particles. The Schrédinger equation is established, various operators are identified, and the probability doctrine emerges. Chapters 5, 6, and 7 provide practice in the use of fundamental concepts, including the construction of an elementary form of the scattering matrix from symmetry principles. Two-by-two matrices are employed in the calculations for transmission through a barrier by the WKB approximation. Spin mechanics with matrices is developed from experimental facts in Chapter 12 as a companion to wave mechanics, and the general dynamics of two-level systems is presented in Chapter 13. The discussion of the spin, as a system with only two basis states, is more elaborate than usual. The inherent simplicity of the mathematical formalism permits attention to be focused on the essential features of quantum mechanics, the superposition principle and the physical meaning of a state. The connection between SU(2) and the rotation group is explicitly established at this elementary level, preparatory to a more general discussion of angular momentum in Chapter 16. The general principles of quantum mechanics are the subject of Chapters 14 and 15. As a synthesis of wave mechanics and spin mechanics—both concrete examples of the abstract theory—these chapters contain the for- mulation of quantum dynamics in Hilbert space and the elements of quantal transformation theory. The essential mathematical background is provided, and the dual bra and ket spaces are used whenever convenient. The general formulation of quantum mechanics in terms of state vectors and linear operators is then used throughout the remainder of the book. As an example, the forced linear harmonic oscillator is treated in the Heisenberg picture, affording an introduction to the notion of “in” and “out” states and to the theory of coherent states. From Chapter 15 on, the new edition of this book differs substantially from the first. With emphasis on rotations and angular momentum, Chapter 16 is devoted to symmetries, including reflection and time reversal, and to the rudiments or the theory of group representations. Bound state pertur- bation theory, with its applications to atoms, is covered in Chapter 17, and time-dependent perturbation theory is now consolidated in Chapter 18. Chapter 19 returns to the study of scattering, considered as a transition between free particle states, and is intended as a bridge from elementaryPreface ix quantum mechanics to modern collision theory. Those who do not intend to cross this bridge may omit Chapter 19. Chapters 20-24 are entirely new. They are based on the belief that, for the quantum mechanics of identical particles, the second quantization formalism is conceptually simpler than either wave mechanics in configuration space or the canonical quantization of classical field equations. Chapter 20 is con- structive in nature and attempts to spell out a set of simple assumptions from which the existence of only two kinds of statistics, with their characteristic commutation relations for creation and annihilation operators, can be deduced. In Chapter 21, the second quantization formalism is applied to a broad range of problems: angular momentum for systems of identical particles, including Schwinger’s method of treating angular momentum by coupling spin 1/2 bosons in the stretched configuration; quantized spin waves in solids (magnons); the theory of atomic structure in outline, and a derivation of the Hartree-Fock equations; a basic account of the BCS theory of superconductivity, expressed in terms of the quasiparticles introduced by Bogoliubov; and a brief derivation of the distribution law for an ideal gas of bosons or fermions in thermal equilibrium. Continuing in an ahistorial vein, Chapter 22 starts with photons as fundamental entities which compose the electromagnetic field with its local dynamical properties like energy and momentum. The most elementary consequences (emission and absorption) of the interaction between the quantized matter and radiation fields are worked out by perturbation theory in the nonrelativistic approximations, leaving all higher-order processes to more advanced textbooks on quantumelectrodynamics. However, the emission and absorption of photons from a given current distribution, which can be described classically, is treated rigorously on the basis of the forced linear harmonic oscillator. Similarly, in Chapter 23, positrons and electrons are taken as the elements of the relativistic theory of leptons and the Dirac equation is derived as the quantized field equation for charged spin 1/2 fermions moving in an externally prescribed electromagnetic field which may be approximated by a classical description. The relativistic covariance of the theory is shown, and the discrete symmetry operations, especially charge conjugation, are discussed. Finally, by a reversal of the usual arguments about hole theory, the one- particle Dirac theory of the electron (or positron) is obtained as an approx- imation to the complete many-electron-positron theory of Chapter 23. The magnetic moment of the Dirac electron and the fine structure of the hydrogen spectrum are derived in Chapter 24. The reader of this book is assumed to know the basic facts of atomic and nuclear physics. No previous contact with quantum mechanics is explicitly required, but some earlier exposure to elementary quantum mechanics atx Preface the undergraduate level will be found advantageous. A good knowledge of classical mechanics and some familiarity with the elements of electromagnetic theory are also assumed. There are almost four-hundred exercises and problems, which form an integral part of the book. The exercises supplement the text and are woven into it. The problems, which appear at the end of each chapter, are usually more independent extensions and applications of the text and require more work. One small departure from the first edition must still be mentioned. Since recent textbooks on electromagnetic theory give preference to Gaussian units, these are used exclusively in the new edition. The quoted references are not intended to be exhaustive, but the footnotes should serve as a guide to further reading and bear witness that many sources have contributed to this book. Comments from many quarters were useful in the preparation of the new edition, and it is impossible to acknowledge the help of every student and colleague who has furnished constructive criticism. The author owes particular thanks to the following physicists: S. Borowitz, W. A. Bowers, B. Bransden, B. Chern, C. DeWitt, S. T. Epstein, N. and P.O. Fréman, B. Hoffmann, H. P. Kennedy, L. Parker, R. E. Peierls, C. P. Poole, and D. D. Sharma. Most of the work on the new edition of this book was done at the University of North Carolina and, during a leave of absence, at the University of Washington. The material in the last four chapters was first developed in lectures given at the University of Colorado. Some sections were written with support from the Atomic Energy Commission, Chapel Hill, Boulder, and Seattle provided excellent environments for writing, and the encourage- ment of colleagues and friends in these places is gratefully acknowledged. Eugen MerzbacherContents nN . Introduction to Quantum Mechanics . Wave Packets and Free Particle Motion . The Wave Equation . The Wave Function and the Schrodinger Equation . The Linear Harmonic Oscillator . Piecewise Constant Potentials in One Dimension . The WKB Approximation . The Principles of Wave Mechanics ). Central Forces and Angular Momentum ). Piecewise Constant Potentials in Three Dimensions and the Hydrogen Atom . Scattering . The Spin . Dynamics of Two-Level Systems . Linear Vector Spaces in Quantum Mechanics . Quantum Dynamics . Rotations and Other Symmetry Operations . Bound State Perturbation Theory . Time-Dependent Perturbation Theory . The Formal Theory of Scattering ). Identical Particles + Applications of Second Quantization . Photons and the Electromagnetic Field . Relativistic Electron Theory . One-Electron Dirac Theory Index 13 28 35 53 80 116 139 172 191 215 251 276 294 334 371 413 450 488 508 525 553 557 SOL 609werMing, a 2 Oo Y “eye,QUANTUM MECHANICS KS a Oo Y Gene, KMing © 2werMing, a 2 Oo Y “eye,CHAPTER | Introduction to Quantum Mechanics Quantum mechanics is the theoretical framework within which it has been found possible to describe, correlate, and predict the behavior of a vast range of physical systems, from elementary particles, through nuclei, atoms, and radiation, to molecules and solids. This introductory chapter aims at setting the stage with a brief review of the historical background and a preliminary discussion of some of the essential concepts which we shall encounter. 1. Quantum Theory and the Wave Nature of Matter. Matter at the atomic and nuclear level reveals the existence of a variety of particles which are identifiable by their distinct properties, such as mass, charge, spin, and magnetic moment. All of these seem to be of a quantum nature in the sense that they take on only certain discrete values. This discreteness of physical properties persists when elementary particles combine to form atoms and nuclei. For instance, the notion that atoms and nuclei possess discrete energy levels is one of the basic facts of quantum physics. The experimental evidence for this fact is overwhelming and familiar. It comes most directly from observations on inelastic collisions (Franck-Hertz experiment) and selective absorption of radiation, and somewhat indirectly from the interpretation of spectral lines. Consider as familiar an object as the hydrogen atom. The evidence that such an atom consists of a nucleus and an electron, bound to the former by ~ forces of electrostatic attraction, is too well known to need recapitulation. The electron can be removed from the atom and identified by its charge, mass, and spin. It is equally well known that the hydrogen atom can be excited by absorbing certain discrete amounts of energy and that it can return the excitation energy by emitting light of discrete frequencies. These are empirical facts. Niels Bohr discovered that any understanding of the observed discreteness requires, above all, the introduction of Planck's constant, h = 6.6255 x 10-*7 erg sec. In old-fashioned language this constant is often called the quantum2 Chapter 1 Introduction to Quantum Mechanics of action. By the simple relation AE = hy (ut) it links the observed spectral frequency » to the jump AE between discrete energy levels. And, divided by 47, the same constant appears also as the unit of angular momentum, the discrete numbers n(h/47)(n = 0, 1, 2, . . .) being the only values which a component of the angular momentum of a system can assume. More specifically, Bohr was able to calculate discrete energy levels of an atom by formulating a set of quantum conditions to which the canonical variables q, and p, of classical mechanics were to be subjected. For our purposes it is sufficient to remember that in this “old quantum theory” the classical phase (or action) integrals for a conditionally periodic motion were required to be quantized according to br, dq, = n;h (1.2) where the quantum numbers n, are integers, and each integral is taken over the full period of the generalized coordinate g,. The quantum conditions (1.2) give good results in calculating the energy levels of simple systems but fail when applied to such systems as the helium atom. Exercise 1.1. Calculate the quantized energy levels of a linear harmonic oscillator of angular frequency w in the old quantum theory. It is well known that (1.1) played an important role even in the earliest forms of quantum theory. Einstein used it to explain the photoelectric effect by inferring that light, which through the nineteenth century had been so well established as a wave phenomenon, can exhibit a particle-like nature and is emitted or absorbed only in quanta of energy. Thus, the concept of the photon asa particle with energy E = hv emerged. The constant h connects the wave (») and particle (E) aspects of light. De Broglie proposed that the wave-particle duality may not be a monopoly of light but a universal characteristic of nature which becomes evident when the magnitude of A cannot be neglected. He thus brought out a second funda- mental fact, usually referred to as the wave nature of matter. This means that in certain experiments beams of massive particles give rise to interference and diffraction phenomena and exhibit a behavior very similar to that of light.? 4 This is true for systems which are composed of several particles (atoms, nuclei) as well as for the elementary particles themselves. Of course, these particles may be no more “elementary” than atoms and nuclei. The ease with which they can be created or destroyed and converted into each other suggests that they may be composite structures. 2 If we sometimes speak of matter waves, this term is not intended to convey the impres- sion that the particles themselves are oscillating in space.§1 Quantum Theory and the Wave Nature of Matter 3 Although such effects were first produced with electron beams, they are now commonly observed with slow neutrons from a reactor. When incident on a crystal these behave very much like X rays. From suclr experiments on the diffraction of free particles we infer the very simple law that the waves associated with the motion of a particle of momentum p propagate in the direction of motion and that their (de Broglie) wavelength is given by h “5 (1.3) This relation establishes contact between the wave and the particle pictures. The finiteness of Planck’s constant is the basic point here. For if A were zero, then no matter what momentum a particle had the associated wave would always correspond to A = 0 and would follow the laws of classical mechanics, which can be regarded as the short wavelength limit of wave mechanics in the same way as geometrical optics is the short wavelength limit of wave optics. Indeed, a free particle would then not be diffracted but go on a straight rectilinear path, just as we expect classically. Let us formulate this a bit more precisely. If x is a characteristic length invoived in describing the motion of a body of momentum p, such as the linear dimension of an obstacle in its path, the wave aspect of matter will be hidden from our sight, if a_i za ap! (1.4) ive., if the quantum of action A is negligible compared with xp. Macroscopic bodies, to which classical mechanics is applicable, satisfy the condition xp >> h extremely well. To give a numerical example, we note that even as light a body as an atom moving with a kinetic energy corresponding to a temperature of T= 10-°°K still has a wavelength no greater than about 10-*cm! We thus expect that classical mechanics is contained in quantum mechanics as a limiting form (h —» 0). Indeed, the gradual transition which we can make conceptually as well as practically from the atomic level with its quantum laws to the macroscopic level at which the classical laws of physics are valid suggests that quantum mechanics must not.only be consistent with classical physics but should yield the classical laws in a suitable approximation. This requirement, which is a guide in discovering the correct quantum laws, is called the correspondence principle. We shall see later that the limiting process which establishes the connection between quantum and classical mechanics can be exploited to give a useful approximation of quantum mechanical problems (see WKB approximation, Chapter 7). We may read (1.3) the other way around and infer that any wave phen- omenon also has associated with it a particle, or quantum, of momentum4 Chapter | Introduction to Quantum Mechanics p = hia. Hence, if a macroscopic wave is to carry an appreciable amount of momentum, as a classical electromagnetic or an elastic wave may, there must be associated with the wave an enormous number of quanta, each contribut- ing a very small momentum. A classically observable wave will result only if the elementary wavelets representing the individual quanta add coherently. For example, the waves of the electromagnetic field are accompanied by quanta (photons) for which the relation E = hv holds. Since photons have no mass, their energy and momentum are related by E = cp. It follows that (1.3) is valid for photons as well as for material particles. At macroscopic wave- lengths, corresponding to radio frequency, a very large number of photons is required to build up a field of macroscopically discernible intensity. Yet, such a field can be described in classical terms only if the photons can act coherently. This requirement, which will be discussed in detail in Chapter 22, leads to the peculiar conclusion that a state of exactly n photons cannot represent a classical field, even if n is arbitrarily large. Exercise 1.2. To what velocity would an electron (neutron) have to be slowed down, if its wavelength is to be 1 meter? Are matter waves of macroscopic dimensions a real possibility? 2. Complementarity. As we have seen, facing us at the outset is the fact that matter, say an electron, exhibits both particle and wave aspects.* This dupli- city of character was described in deliberately vague language by saying that the de Broglie relation (1.3) “associates” a wavelength with a particle momentum. The vagueness reflects the fact that particle and wave aspects, when they show up in the same thing such as the electron, are incompatible with each other, unless we modify our traditional concepts to a certain extent. Particle means to us traditionally an object with a definite position in space. ‘Wave means a periodically repeated pattern in space with no particular emphasis on any one crest or valley, and it is characteristic of a wave that it does not define a location or position sharply. How a synthesis of these two concepts might be accomplished can, for a start, perhaps be understood if we recall that the quantum theory must give an account of the discreteness of certain physical properties, ¢.g., energy levels in an atom or nucleus. Yet discreteness is not new in physics. In classical macroscopic physics discrete, “quantized,” physical quantities appear naturally as the frequencies of vibrating bodies of finite extension, such as 2 The massless quanta corresponding to elastic (e.g., sound) waves are called phonons and behave similarly to photons, except that c is now the speed of sound. 4 It will be convenient to use the term electron frequently when we wish to place equal ‘emphasis on the particle and wave aspects of a constituent of matter. The electron has been chosen only for definiteness of expression. Quantum mechanics applies equally to protons, neutrons, mesons, etc.§2 Complementarity 5 strings, membranes, air columns. We speak typically of the natural modes of such systems. These phenomena have found their simple explanation in terms of interference between incident and reflected waves. Mathematically the discrete behavior is enforced by boundary conditions: the fixed ends of the string, the clamping of the membrane rim, etc. It is tempting likewise to see in the discrete properties of atoms the manifestations of bounded wave motion and to connect the discrete energy levels with standing waves. In such a picture the bounded wave must somehow be related to the confinement of the particle to its “orbit,” but it is obvious that the conventional concept of an orbit as a trajectory covered with definite speed cannot be maintained. A wave is generally described by its velocity of propagation, wavelength, and amplitude. Since in a standing wave it is the wavelength which assumes discrete values, it is evident that, if our analogy is meaningful at all, there must be a correspondence between the energy of the particle and the wave- length of the associated wave. For a free particle, one that is not bound in an atom, the de Broglie formula (1.3) has already given us a relationship connect- ing wavelength with energy (or momentum). The equation linking wavelength and momentum is likely to be much more complicated for an electron bound to a nucleus as in the hydrogen atom, and it is one of our tasks to uncover the appropriate connection. For very loosely bound electrons, moving in circular orbits far from the nucleus, (1.3) should be a good approximation. Under these circumstances the wavelength is short compared with the distance between electron and nucleus. Hence, it should be possible to describe this type of almost-free motion either as a particle moving in a well-defined classical orbit around a center of force or as a wave which can be thought of as following this orbit. From Figure 1.1 it is clear that such a situation can be described by a wave only if the orbit accommodates exactly an integral number of wavelengths (Figure 1.1a), Otherwise (Figure 1.16) the wave is out of phase with itself, and destructive interference will prevent a stable oscillation around the nucleus. Hence, the condition for having a standing wave is ds =n (5) a the integral to be taken over the path of the trajectory. In the classical limit nis a very large integer. Using (1.3), we obtain $e ds = nh (1.6) which for the special case of a circular orbit (ds = r dg) is easily seen to be identical with a quantum condition of the form (1.2), if the aximuth @ is used as the generalized coordinate, p, = L, = rp being the canonically conjugate momentum.6 Chapter 1 Introduction to Quantum Mechanics Figure 1.1. (a) Constructive interference of de Broglie waves in an atom distinguishes the allowed stable Bohr orbits. (6) Destructive interference of de Broglie waves in an atom disallows any orbit which fails to satisfy the quantum conditions. Of course, conditions (1.2) and (1.6) are really useful and interesting only if the quantum numbers 7 are small integers, i.e., outside the domain of validity of the classical approximation. The bold extrapolation from large to small quantum numbers is in accordance with Bohr’s demand, formulated in the correspondence principle, that a quantum equation which, like (1.6), has been established to be valid in the classical limit (n — 00) should be tentatively applied also in the unclassical quantum domain (small n). The fact that for n = 1, 2, .. . (1.6) yields the correct energy levels of the Bohr atom shows the usefulness of the correspondence principle and, more important for us, indicates that the wave-particle dualism contains the ingredients for a satisfactory description of atomic systems. Exercise 1.3. Assuming that the electron moves in a circular orbit in a Coulomb field, derive the Balmer formula from the quantum condition (1.6). It might be well to emphasize that the quantum conditions (1.2) or (1.6), although they give astonishingly good values for the energy levels of some quantum systems, are in general not accurate for small values of n. It should not be surprising that a naive extrapolation from large to small quantum numbers can lead to incorrect results. The correspondence principle merely requires that quantum theory must be consistent with classical physics in the limit of large quantum numbers but does not specify uniquely the form of the§2 Complementarity 1 new theory. In Chapter 7 it will be shown how the old quantum conditions are obtained from quantum mechanics in a semiclassical approximation. We owe mainly to Bohr the discovery of the principles that allow us to understand how the two seemingly contradictory concepts of wave and particle, describing the same thing, may be reconciled. We shall see in later chapters that this reconciliation requires modification of some classical attributes with which we commonly endow waves and particles. When these concepts have been appropriately broadened, one of the two pictures alone will prove inadequate for the description of physical happenings. But their new broadened definitions will—owing to a departure from the strictly classical characterizations—carry within themselves enough freedom for the two aspects to coexist. Bohr, who was guided by the empirical fact of the dual nature of matter, has elevated this program to a principle of complemen- tarity. According to this principle, wave and particle nature are considered complementary aspects of matter, both equally essential for a full description of the phenomena. It may be argued that the reasoning which has led us so far utilizes classical and conventional notions, appropriate to macroscopic phenomena but patently inapplicable in the atomic domain. After all, Bohr’s quantization tules, successful as they are, entail a whole set of assumptions which are in conflict with classical physics. Chief among these is the interdiction against the emission of radiation while the electron circulates about the nucleus. This insures the stability of the Bohr orbits, but it violates classical principles concerning the radiation from accelerated charges, and it fails to help us understand the occurrence, albeit infrequently, of transitions between energy levels which give rise to the observed spectral lines. Might we not be better off if we shed all pretext of making pictures of the quantum phenomena in terms of particles and waves and the like? Why not simply establish suitable mathematical laws for the description of the observations, as Newton urged for a branch of physics reaching maturity? Such a point of view ignores the process by which new physical theories are arrived at; new ideas always depend on understanding the novel in terms of the familiar. Of course, as the recently “new” ideas become more familiar and we become accustomed to them, the need for the older points of reference diminishes. It is for this reason, for instance, that a complete review of the elaborate calculations of the “old” quantum theory is no longer regarded as particularly helpful in the study of quantum mechanics. Nevertheless, when all this is recognized, there still remains the pos- sibility—according to Bohr and his Copenhagen school the necessity—that all our experience of the quantum level of nature may have to be perceived and interpreted in classical, macroscopic terms. In support of this view it is pointed out that all physical quantities such as position, momentum,8 Chapter 1 Introduction to Quantum Mechanics energy, angular momentum, etc., which are used in quantum mechanics as much as in classical mechanics, can be defined only by stating how they are measured. All such measurements, ultimately designed to convey informa- tion to our senses, must contain amplification mechanisms by which microscopic effects are translated into macroscopic effects accessible to our understanding. In the final analysis, atomic, nuclear, and particle experi- ments are all described in classical terms, since they invariably result in the reading of scales, the motion of large pointers, the activation of computer elements, etc., all of which are subject to the laws of classical physics. According to this view, there is a subtle and profound relationship between quantum mechanics and classical physics. We have already encountered one facet of this relationship; it is the correspondence principle, which demands that quantum mechanics must be consistent with classical mechanics. Bohr went far beyond this obvious requirement and contended that quantum mechanics presupposes classical physics in a logical sense for its very formulation. Although it is not clear that this epistemological view of the relationship between classical and quantal physics and the associated doctrine of complementarity will remain central to the interpretation of quantum mechanics, an appreciation of Bohr’s attitude is important because, through stimulation and provocation, it has greatly contributed to the development of the subject. 3. The Wave Function and Its Interpretation. So far we have considered only the length of matter waves, and we have found how the wavelength is connected with particle momentum. There are, of course, other important properties of these waves to be discussed. In particular, their amplitude may or may not have directional (i.e., polarization) properties. We shall see in Chapter 12 that the spin of the particles corresponds to the polarizability of the waves. However, for many purposes the dynamical effects of the spin are negligible in first approximation, especially if the particles move with non- relativistic velocities.> We shall neglect the spin for the time being, much as in a simple theory of wave optical phenomena (interference, diffraction, geometrical limit, etc.) the transverse nature of light waves is neglected. Hence, we attempt to build up quantum mechanics by the use of scalar waves first. For particles with zero spin, e.g., pi mesons, this gives an appropriate description. For particles with nonzero spin, ¢.g., electrons, nucleons, mu mesons, suitable corrections must be made later (see Chapters 12 and 17). Mathematically these waves are represented by a function y(z, y, 2, t) which in colorless terminology is called the wave function. Upon its introduc- tion we immediately ask such questions as these: Is y a measurable quantity, and what precisely does it describe? In particular, what feature of the particle aspect is related to the wave function? 5 This statement requires important corrections when we deal with systems of many identical particles (Chapter 20).§3 The Wave Function and Its Interpretation 9 We cannot expect entirely satisfactory answers to these questions before we have become familiar with the properties of these waves and with the way in which y is used in calculations. As we progress through quantum mechanics, we shall become accustomed to y as an important addition to our arsenal of physical concepts. If its physical significance must as yet remain somewhat obscure to us, one thing seems certain: y must in some sense be a measure of the presence of a particle. Thus we do not expect to find a particle in those regions of space where y = 0. Conversely, in the regions of space where a particle may be found, p must be different from zero. But the function v(x, y, 2, t) itself cannot be a direct measure of the likelihood of finding a particle at position x, y, z at time ¢. For if it were that, it would have to be a positive number everywhere. Yet it is impossible that y be positive (or zero) everywhere, if destructive interference of the y waves is to account for the observed dark interference fringes and for the instability of any but the distinguished Bohr orbits in a hydrogen atom (Figure 1.16). In physical optics interference patterns are produced by the superposition of waves of E and H, but the intensity of the fringes is measured by E? and H®. In analogy to this situation we assume that the positive quantity lp(x, y, 2, t)|? measures the probability of finding a particle at x, y, z. (The absolute value has been taken because it will turn out that y can have complex values.) The full meaning of this interpretation of y and its internal consistency will be discussed in detail in Chapters 4 and 8. Here we merely want to advance some general qualitative arguments for this so-called probability interpretation of the quantum wave function y. This interpretation was introduced into quantum mechanics by Born. The discussion in the preceding section of the role that standing waves play in the hydrogen atom leads us to associate the wave p with a single particle as a representative of a statistical ensemble. The alternative would be to regard the wave as describing the behavior of a beam of particles, so that, for instance, the observed diffraction patterns would be the result of some collective phenomena. Owing to their characteristic properties, such as charge and mass, particles can, fortunately, be identified singly in the familiar detection devices of modern experimental physics: counters, bubble chambers, photographic plates, etc. With the aid of these tools it has been established abundantly that the interference fringes shown schematically in Figure 1.2 are the statistical result of the effect of a very large number of independent particles hitting the screen. The interference pattern appears only after many particles have passed through the slits, but each particle retains its discrete individuality and each goes its own way, being finally deposited somewhere on the screen. Note that the appearance of the inter- ference effects does not require that a whole beam of particles go through the slits. In fact, particles can actually be accelerated and observed singly, and the interference pattern can be produced over a length of time, a particle hitting10 Chapter 1 Introduction to Quantum Mechanics Figure 1.2. Double-slit interferences produced in analogy with Young’s optical experiment. C is the location of a bright fringe on the screen if AC — BC equals an integral multiple of the wavelength. the screen now here, now there in seemingly random fashion. When many particles have come through, a regular interference pattern will be seen to have formed. The conclusion is almost inevitable that y describes the behavior of single particles, but that it has an intrinsic probabilistic meaning. The quantity |y|? would appear to measure the chance of finding the particle at a certain place. In a sense, this conclusion was already implicit in our earlier discussion regarding a plane wave as representative of a particle with definite momentum (wavelength) but completely indefinite position. At least if p is so interpreted, the observations can be correlated effortlessly with the mathematical formalism. The opposition to the probability interpretation of p originates in the discomfort felt by some (including very eminent) physicists with the lack of determinism which the probabilistic view is said to ascribe to nature. The idealized experiment shown in Figure 1.2 illustrates the argument. Single particles are subject to wave interference effects, and some are found deposited on the screen at locations which they could not reach if they moved along classical paths through either slit. The appearance of the interference fringes depends on the passage of the wave through both slits at once, just as a light wave goes through both slits in the analogous optical Young interference experiment, If the wave describes the behavior of a single particle, it follows that we cannot decide through which one of the two slits the§3 The Wave Function and Its Interpretation 11 particle has gone. If we try to avoid this consequence by determining experi- mentally with some clever monitoring device through which slit the particle has passed, we shall by the very action of the monitoring mechanism change the wave drastically and destroy the interference pattern. A single particle now goes definitely through one slit or the other, and the accumulation of a large number of particles on the screen will result in two well-separated traces. Exactly the same traces are obtained if we close one slit at a time, thereby predetermining the path of a particle. We conclude that the conditions under which the interference pattern is produced forbid a determination of the slit through which the particle passes. More quantitative statements about this peculiar circumstance (Heisenberg uncertainty principle) will be made later, but the basic feature should be clear from the present example: The simultaneous appearance of wave and particle aspects compels us to be resigned to an inevitable indetermination in some of the conditions of an experiment. Wave aspect and particle aspect in one and the same thing are compatible only if we forgo asking certain questions of nature which no longer have any meaning (such as: “Let us see the inter- ference fringes produced by particles whose paths through an arrangement of slits we have followed!””). Accepting this situation as inescapable, it is then not paradoxical to say that the interference fringes appear if the particle goes through both slits at once. The probability doctrine of quantum mechanics asserts that the indeter- mination, of which we have just given an example, is a property inherent in nature and not merely a profession of our temporary ignorance from which we expect to be relieved by a future better and more complete theory. The conventional interpretation thus denies the possibility of an ideal theory which would encompass the present quantum mechanics but would be free of its supposed defects, the most notorious “imperfection” of quantum mechanics being the abandonment of strict classical determinism. Loose talk about these problems may even lead to the impression that quantum mechanics forsakes the classical goal of a wholly rational description of physical processes. Nothing could be further from the truth, as this book hopes to demonstrate. * Many references to the vast literature on quantum mechanics are found in the footnotes of this book. Other books that will contribute to an understanding of quantum mechanics are: H. A. Bethe and R. W. Jackiw, Intermediate Quantum Mechanics, 2nd ed., W. A. Benjamin, New York, 1968. D. Bohm, Quantum Theory, Prentice-Hall, New York, 1951. P. A. M. Dirac, The Principles of Quantum Mechanics, 4th ed., Clarendon Press, Oxford, 1958; Ist ed.: 1930; 2nd ed.: 1935; 3rd ed.: 1947. S. Gasiorowicz, Elementary Particle Physics, John Wiley and Sons, New York, 1967. H. S. Green, Matrix Mechanics, P. Noordhoff Ltd., Groningen, 1965.12 Chapter | Introduction to Quantum Mechanics Footnote continued J. M. Jauch, Foundations of Quantum Mechanics, Addison-Wesley Publishing Company, Reading, 1968. L. D. Landau and E. M. Lifshitz, Quantum Mechanics, translated by J. B. Sykes and J. S. Bell, Addison-Wesley Publishing Company, Reading, 1958. G. Ludwig, Die Grundlagen der Quantenmechanik, Springer, Berlin, 1954. A. Messiah, Quantum Mechanics. Volume 1 translated by G. M. Temmer, Volume II translated by J. Potter, North-Holland Publishing Company, Amsterdam, 1961 and 1962. W. Pauli, Die allgemeinen Prinzipien der Wellenmechanik, in S. Fliigge, ed., Encyclopedia of Physics, vol 5/1, pp. 1-168, Springer Verlag, Berlin, 1958. P. Roman, Advanced Quantum Theory, Addison-Wesley Publishing Company, Reading, 1965. S. 8, Schweber, An Introduction to Relativistic Quantum Field Theory, Row, Peterson and Company, Evanston, 1961. A.J. W. Sommerfeld, Atombau und Spektrallinien, Volume IL, F. Vieweg, Braunschweig, 1939. S. Tomonaga, Quantum Mechanics, translated by Koshiba, Volume I, North-Holland Publishing Company, 1962; Volume II, John Wiley and Sons, New York, 1966. B. L. van der Waerden, Sources of Quantum Mechanics, North-Holland Publishing Company, Amsterdam, 1967,CHAPTER 2 Wave Packets and Free Particle Motion 1. The Principle of Superposition. We have learned that it is reasonable to suppose that a free particle of momentum p is associated with a harmonic plane wave. Defining a vector k which points in the direction of wave propagation and has the magnitude 2: katt (2.1) we may write the fundamental de Broglie relation as p= ik (2.2) fi is a common symbol in quantum physics and denotes the frequently recurring constant he = = 1.0545 x 107” erg sec = 6.5819 x 10-2 MeV sec 7 It is true that the diffraction experiment does not give us any direct information about the detailed dependence on space and time of the periodic disturbance which produces the alternatingly “bright” and “dark” fringes, but all the evidence points to the correctness of the simple inferences embodied in (2.1) and (2.2). The comparison with optical interference suggests that the fringes come about by linear superposition of two waves, a point of view which has already been stressed in Section 1.3 in the discussion of the simple double-slit interference experiment, Figure 1.2. Mathematically these ideas are formulated in the following fundamental assumption about the wave function y(x,y,2,t): if yy(x,y,2,t) and po(x, y, z,t) describe two waves, their sum y(z, y,z, ft) = y + pe also describes a possible physical situation. This assumption is known as the principle of superposition and is illustrated by the interference experiment of Figure 1.2. The intensity produced on the screen by opening only one slit at a time is |y,/? or |pal2. When both slits are open, the intensity is determined by14 Chapter 2 Wave Packets and Free Particle Motion lv: + val? This differs from the sum of the two intensities, |y,|? + |yal? by the interference terms yyy2* + y:*y2. (An asterisk will denote complex conjugation throughout this book.) A careful analysis of the interference experiment would require detailed consideration of the boundary conditions at the slits,! but there is no need here for such a thorough treatment, because our purpose in describing the idealized double-slit experiment was merely to show how the principle of superposition accounts for some typical interference and diffraction phen- omena. Such phenomena, when actually observed as in diffraction of particles through crystals, are impressive direct manifestations on a macroscopic scale of the wave nature of matter. We therefore adopt the principle of superposition to guide us in developing quantum mechanics. The simplest type of wave motion to which it will be applied is an infinite harmonic plane wave propagating in the positive 2- direction with wavelength A = 2n/k and frequency w. Such a wave is associated with the motion of a free particle moving in the x-direction with momentum p = hik. If we followed our first impulse, we would represent this plane wave by a function of the form (2, t) oc cos (ke — wt + a) (2.3a) and a plane wave moving in the negative x-direction would be written as y(z, t) x cos (ke + wt + B) (2.36) The coordinates y and z can obviously be omitted in describing these one-dimensional processes. In order to decide if (2.3a) and (2.3b) offer satisfactory means of descril plane waves in quantum mechanics, it is necessary to give some thought to the time dependence of the wave function. Classical mechanics permits the prediction of what happens at time #, if the initial conditions (coordinates and velocities) of the system at ¢ = 0 are known. The correspondence prin- ciple requires that quantum mechanics must be equally deterministic and that it should be possible to calculate the condition of a system at an arbitrary time ¢ from knowledge about it at time ¢ = 0. But we have no a priori knowledge about the degree of information needed at t = 0, if a complete prediction of the evolution of the system is to be made, and we do not know whether the initial wave function p(, y, z, 0) determines y(z, y, 2, ¢) at all times, or whether additional information about the state at ¢ = 0 is necessary. Judging from experience with other types of wave motion, it might at first sight seem questionable that knowledge of (x,y, 2,0) could suffice to 1 This is similar to the situation found in wave optics, where the uncritical use of Huygens’ principle must be justified by recourse to the wave equation and to Kirchhoft’s approximation.§1 The Principle of Superposition 15 determine y(x, y, z, t). For example, if y were an elastic displacement of a vibrating string, dp/@t would have to be specified at t = 0 in addition to yp. We need only to remember that a snapshot taken of a vibrating string at t= 0 tells us nothing about its subsequent behavior. We must also know what it looks like a little later in order to decide whether the wave is propa- gating to the right, or to the left, or whether it is a standing wave. This is clearly the reason why the differential equation governing the behavior of a vibrating string must be of the second order in the time derivative. As an illustration we note that the initial wave form (x, 0) = cos ke (2.4) is obtained from either (2.3a) or (2.36) by setting « = 6 = 0 and t=0. The initial condition (2.4) does not tell us how much of the wave travels to the right, (2.3a), and how much travels to the left, (2.35). In fact, the wave function y(e, t) = acos (kx — wt) + (1 — a) cos (kx + wt) (2.5) coincides with p(x, 0) = coskx at ¢=0 for any value of a. However, knowledge of the first time derivative of (2.5) at t= 0 Otho determines a and hence the form of (2.5) unambiguously. We conclude that we are confronted by the following choice of working hypotheses: (2) = (2a — 1)w sin ke I, Equations (2.3a) and (2.36) describe plane waves, as proposed, but lz, y, z, 0) does not determine the wave function (x, y, z, t) at other times. Il. p(x, y, z, 0) determines (x,y,z, 1), but the expressions (2.34) and (2.3) do not describe simple harmonic plane yp waves. Since, unlike elastic displacements or electric field vectors, the y waves of quantum mechanics cannot be observed directly, it is tempting to construct the theory by the use of working hypothesis [I and to infer from the formalism the appropriate y for plane waves, The only way in which we can generalize the expression (2.3a) for a harmonic plane wave propagating in the positive x-direction is to adjoin to the cosine a sine term, thus forming yilz, t) cos (kx — wt) + 6, sin (ke — wt) (2.6) Similarly, a wave traveling in the opposite direction is pala, t) oc cos (kx + wt) + dy sin (kx + wt) (2.7) The constants 6, and 6, can be determined by the requirement that the two oppositely moving plane waves , and y, should be linearly independent at16 Chapter 2. Wave Packets and Free Particle Motion all times,? and that an arbitrary displacement of z or ¢ should not change the physical character of the wave. The linear independence insures that any harmonic plane wave can be written at all times as a linear combination of y, and yz. It is then eminently reasonable to demand that the motion of a free particle moving in the positive z-direction should at all times be represented by a pure plane wave of the type y without any admixture of ys, which moves in the opposite direction. Hence, it must be required that y, at time ¢ should be a simple multiple of y, evaluated at f = 0. Mathematically we must have cos (kx + ) + 4, sin (kx + &) = a,(e)(cos kx + 6, sin kz) for all values of x and e. Comparing coefficients of cos kx and of sin kx, this last equation leads to cose +6,sine=a, and 6, cos e — sine = a,6, These equations are compatible for all ¢ only if é%=-1 of b= Hi Similarly, from (2.7) we obtain 6,2 = —1. We choose 6, = i for the wave (2.6), and 6, = —i for (2.7), and we are thus led to the conclusion that y waves describing free particle motion must be complex quantities if working hypothesis II is to succeed. Summarizing our conclusions, we see that in this scheme a harmonic plane wave propagating toward increasing « is pr(a, t) = Aalto? (2.8) and a wave propagating toward decreasing = is val, #) = Beteto? (2.9) The initial values of these waves are ,(x, 0) = Ae and a(x, 0) = Be respectively. w is the frequency of oscillation which so far we have not brought into connection with any physically observable phenomenon. Generally, it will be a function of k (see Section 2.3). We shall refer to wave functions like (2.8) and (2.9) briefly as plane waves. A plane wave propagating in an arbitrary direction has the form yx, y, 2, f) = AeeOM — Agilkertiywtht-od (2.10) Equations (2.8) and (2.9) are special cases of this with k, =k, = 0 and k, = +k. Of course, it is always possible to replace a complex wave function by two equivalent real functions (e.g., the real and imaginary parts of y), but to do 2 The waves (2.3a) and (2.36) violate this requirement. For t = (a — B)/2u they are equal, hence linearly dependent.§2 Wave Packets and the Uncertainty Relation 17 so would violate the spirit of working hypothesis II. The essence of this hypothesis is that a single function y determines the future state of the system. We now recognize that a price must be paid for this assumption: Complex functions y must be admitted. There can be at this point hardly any objection to the use of such functions. In particular, since the diffraction pattern presumably measures only the amplitude of y, we have no physical reason for rejecting complex-valued wave functions. However, it should be stressed that with the acceptance of complex values for y we are by no means excluding wave functions which are real. For example, wave function (2.4) can be written as the sum of two exponential functions: p(x, 0) = cos kx = fel + forte (2.11) According to the principle of superposition, this must be an acceptable wave function. How does it develop in time? The principle of superposition, in conjunction with working hypothesis II, suggests (but does not logically require) a simple answer to this important question: Each of the two waves, into which y can be decomposed at t = 0, develops independently, as if the other component were not present. The wave function, which at t = 0 satisfies the initial condition p(x, 0) = cos kx, thus becomes for arbitrary t: Ya, 1) = pelo) 4 yertlerron It is seen that our rule, to which we shall adhere and which we shall gener- alize, assures that p(x, 0) determines the future behavior of the wave uniquely. If this formulation is correct, we shall expect that the complex y waves may be described by a differential equation which is of the first order in time.* 2. Wave Packets and the Uncertainty Relation. The foregoing discussion points to the possibility that by allowing the wave function to be complex we might be able to describe the state of motion of a particle at time ¢ completely by y(z, t). The real test of this assumption is of course its success in account- ing for experimental observations. However, strong support for it can be gained by demonstrating that the correspondence with classical mechanics can be established within the framework of this theory. To this end we must find a way of making (zx, t) describe, at least approxi- mately, the classical motion of a particle which has both reasonably definite position and reasonably definite momentum. The plane waves (2.8) and (2.9) correspond to particle motions with momentum which is precisely defined by 3 Maxwell's equations are of the first order even though the functions are real, but this is accomplished by using both E and B in coupled equations, instead of describing the field by just one Function. The heat flow equation is of the first order in time and still describes the behavior of a real quantity, the temperature; but none of its solutions Tepresent waves propagating without damping in one direction like (2.10).18 Chapter 2. Wave Packets and Free Particle Motion Figure 2.1. Example of a one-dimensional momentum distribution. The function (k,) = 7/42! cosh (kz — 3)}-* represents a wave packet moving with mean wave number kz = 3. (kz) is normalized to unity. Figure 2,2. Normalized wave packet corresponding to the momentum distrib of Figure 2.1, The absolute value |y(z, 0)| = =/4[2 cosh (n/*z/2)]" is plotted.§2 Wave Packets and the Uncertainty Relation 19 (2.2); but, having amplitudes |y| = const. for all x and ¢, these plane waves leave the position of the particle entirely unspecified. However, by super- position of several different plane waves a certain degree of localization can be achieved, as the fringes on the screen in the diffraction experiment attest. The mathematical tools for such a synthesis of localized compact wave packets by the superposition of plane waves of different wave number are available in the form of Fourier analysis. Maintaining for simplicity the fiction that only one spatial coordinate, x, need be considered, we can write 4.00 v(x, 0) = ez f Ake dky (2.12) at Joo and the inverse formula 1 ft : ok.) = e f v(x, Oe dex (2.13) under very general conditions to be satisfied by y(x, 0). Formulas (2.12) and (2.13) show that y(k,) determines the initial wave function (x, 0), and vice versa. We now apply the rule, established in the last section, that each component plane wave contained in (2.12) propagates independently of all the others according to the prescription of (2.10). The wave function at time ¢ is thus: 1 rte y(a, t) = ie E Ok, det dk, (2.14) Formula (2.14) is, of course, incomplete until we determine the dependence of » on k, This determination will be made on physical grounds in Section 2.3. For simplicity we now assume that g is a real positive function and that it has a shape similar to Figure 2.1, i.c., an approximately symmetric distribu- tion of k, about a mean value k,. Making the simple change of variables unk. -E we may write = —L pte (** tue 21. WO) = Te | ou + Ryde du (2.15) This is a wave packet whose absolute value is shown in Figure 2.2. The particle is most likely to be found at a position where y is appreciable. It is easy to see for any number of simple examples that the widths of the amplitude ¢ as a function of k, and of the wave packet p stand in a reciprocal telationship: Az Ak, = 1 (2.16) A proof of this assertion will be supplied at the end of this chapter.20 Chapter 2. Wave Packets and Free Particle Motion Exercise 2.1. Assume o(k,) = V2n for ky — 8