The Conceptual Framework of Quantum Field Theory Anthony Duncan OXFORD UNIVERSITY PRESSOXFORD UNIVERSITY PRESS Great Clarendon Street, Oxford, OX2 GDP, United Kingdom Oxford University Press is a department of the University of Oxford. If furthers the University’s objective of excellence in research, scholarship, and education by publishing wo-ldwide. Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries © Amhony Duncan 2012 ‘The moral rights of the author have been asserted First Edicion published in 2012 Impression: 1 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by licence or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above ‘You must not circulate this work in any other form ‘and you must impose this same condition on any acquirer British Library Cataloguing in Publication Data Data available Library of Congress Cataloging in Publication Data Data available BN 978-0-19-957326-4 Printed and bound by CPI Group (UK) Ltd, Croydon, CRO 4YY. Links to third party websites are provided by Oxford in good faith and for information only. Oxford disclaims any responsibility for the materials contained in any third party website referenced in this work.Preface In the roughly six decades since modern quantum field theory came of age with the introduction in the late 1940s of covariant field theory, supplemented by renormaliza- tion ideas, there has been a steady stream of expository texts aimed at introducing each new generation of physicists to the concepts and techniques of this central area of modern theoretical physics. Each decade has produced one or more “classics”, attuned to the background, needs, and interests of students wishing to acquire a proficiency in the subject adequate for the beginning researcher at the time. In the 1950s the seminal text of Jauch and Rohrlich, Theory of Photons and Electrons, provided the first systematic textbook treatment of the Feynman diagram technique for quantum electrodynamics, while more or less simultaneously the first field-theoretic attacks on the strong interactions were presented in the two-volume Mesons and Fields of Bethe, de Hoffmann, and Schweber. The 1960s saw the appearance of the massive treatise by Schweber, Introduction to Relativistic Quantum Field Theory, which addressed in much greater detail formal aspects of the theory, including the LSZ asymptotic formalism and the Wightman axiomatic approach. The dominant text of the late 1960s was undoubtedly the two-volume text of Bjorken and Drell, Relativistic Quantum Mechanics and Relativistic Quantum Fields, which combined a thorough introduction to Feynman graph technology (in volume 1) with a more formal introduction to Lagrangian field theory (in volume 2). In the 1970s the emergence of non-abelian gauge theories as the overwhelmingly favored candidates for a successful field-theoretic description of weak and strong interactions coincided with the emergence of functional (path-integral) methods as the appropriate technical tool for quantization of gauge theories. In due course, these methods received full treatment with the appearance in 1980 of Itzhykson and Zuber’s encyclopedic Quantum Field Theory. In a similar way, the surge to prominence of supersymmetric field theories throughout the 1980s necessitated a full account of supersymmetry, which is the sole subject of the third volume of Weinberg’s comprehensive three-volume The Quantum Theory of Fields, the first edition of which appeared in 1995. With such a selection of classic expository treatises (not to mention many other fine texts not listed above—with apologies to authors of same!) one may well doubt the need for yet another introductory treatment of quantum field theory. Nevertheless, in the course of teaching the subject to graduate students (typically, second year) over the last 25 years, I have been struck by the number of occasions on which important conceptual issues are raised by questions in the classroom which require a careful explanation not to be found in any of the readily available textbooks on quantum field theory. To give just a small sample of the sort of questions one encounters in the classroom setting: “Of the plethora of quantum fields introduced to describe Nature at subatomic scales, why do so few (basically, only electromagnetism and gravity) have classical macroscopic correlates?”; “If there are many possible quantum fields availableiv Preface to ‘represent’ a given particle, can, or in what sense does, quantum field theory prescribe a unique all-time dynamics?”; “If the interaction picture does not exist, as implied by Haag’s theorem, why (or in what sense) are the formulas derived in this picture for the S-matrix still valid?”; “Are there non-perturbative phenomena amenable to treatment using perturbative (i.e., graph-theoretical) methods?”; and so on. None of these questions require an answer if one’s attitude in learning quantum field theory amounts to a purely pragmatic desire to “start with a Lagrangian and compute a process to two loops”. However, if the aim is to arrive at a truly deep and satisfying comprehension of the most powerful, beautiful, and effective theoretical edifice ever constructed in the physical sciences, the pedagogical approach taken by the instructor has to be quite a bit different from that adopted in the “classics” enumerated above. In the present work, an attempt is made to provide an introduction to quantum field theory emphasizing conceptual issues frequently neglected in more “utilitarian” treatments of the subject. The book is divided into four parts, entitled respec- tively, “Origins”, “Dynamics”, “Symmetries”, and “Scales”. Although the emphasis is conceptual-the aim is to build the theory up systematically from some clearly stated foundational concepts—and therefore to a large extent anti-historical, I have included two historical chapters in the “Origins” section which trace the evolution of the modern theory from the earliest “penumbra” of quantum-field-theoretical phenomena detected by Planck and Einstein in the early years of the twentieth century to the emergence, in the late 1940s, of the recognizable structure of modern quantum field theory, in the form of quantum electrodynamics. The reader anxious to proceed with the business of logically developing the framework of modern field theory is at liberty to skim, or even entirely omit, this historical introduction. ‘The three remaining sections of the book follow a step-by-step reconstruction of this framework beginning with just a few basic assumptions: relativistic invariance, the basic principles of quantum mechanics, and the prohibition of physical action at a distance embodied in the clustering principle. The way in which these physical ingre- dients combine to engender some of the most dramatic results of relativistic quantum. field theory is outlined qualitatively in Chapter 3, which also contains a summary of the topics treated in later chapters. Subsequent chapters in the “Dynamics” section of the book lay out the basic structure of quantum field theory arising from the sequential insertion of quantum-mechanical, relativistic, and locality constraints. The rather extended treatment of free fields allows us to discuss important conceptual issues (e.g., the classical limit of field theory) in greater depth than usually found in the standard texts. Some applications of perturbation theory to some simple theories and processes are discussed in Chapter 7, after the construction of covariant fields for general spin has been explained. A deeper discussion of interacting field theories is initiated in Chapters 9, 10, and 11, where we treat first general features shared by all interacting theories (Chapter 9) and then aspects amenable to formal perturbation expansions (Chapter 10). The “Dynamics” section concludes with a discussion of “non-perturbative” aspects of field theory—a rather imprecise methodological term encompassing a wide variety of very different physical processes. In Chapter 11 we attempt to clarify the extent to which certain features of field theory are “intrinsically”Preface v non-perturbative, requiring methods complementary to the graphical expansions made famous by Feynman. In the “Symmetries” section we explore the many important ways in which symmetry principles influence both our understanding and our use of quantum field theory. Of course, at the heart of relativistic quantum field theory lies an inescapable symmetry of critical importance: Lorentz-invariance, which, together with translational symmetry in space and time, makes up the larger symmetry of the Poincaré group. The centrality of this symmetry explains the dominance of Lagrangian methods in field theory, even though from a physical standpoint the Hamiltonian would appear (as is typically the case in non-relativistic quantum theory) to hold pride of place. The role played by Lorentz-invariance in restricting the dynamics of a field theory is the main topic of Chapter 12, which also includes an introduction to the extension of the Poincaré algebra to the graded superalgebra of supersymmetric field theory. Discrete spacetime symmetries, and the famous twin theorems of axiomatic field theory—the Spin-Statistics and TCP theorems—are the subject of Chapter 13. The discussion of global symmetries, exact and approximate, in Chapter 14 leads naturally into the very important topics of spontaneous symmetry-breaking and the Goldstone theorem. The “Symmetries” section of the book closes with a treatment of local gauge symmetries in Chapter 15, which imply remarkable new features not present in theories where the only symmetries are global (i.e., involve a finite-dimensional algebra of spacetime-independent transformations). With the final section of the book, entitled “Scales”, we come to perhaps the most characteristic conceptual feature of quantum field theory: the scale separation property exhibited by theories defined by an effective local Lagrangian. Given that essentially all of the information obtained from scattering experiments at accelerators concerns asymptotic transitions (i.e., the infinite time evolution of an appropriately prepared quantum state, terminated by a detection measurement) it is critically important for theoretical progress that the probabilities of such transitions not depend in a sensitive way on interaction details at much smaller distances than those presently accessible in accelerator experiments (roughly, the inverse of the center-of-mass energy of the collision process). The insensitivity of field theory amplitudes to our inescapable ignorance of the nature of the interactions at very short distances (or equivalently, high momentum) is therefore of central importance if we are to infer reliably an underlying microdynamics from the limited phenomenology available at any given time. Remarkably, in this respect quantum field theories are far kinder to us than their classical (particle or field) counterparts, where non-linearities almost always introduce chaotic behavior which effectively precludes the possibility of accurate predictions of state evolution over long time periods. The technical foundations needed for examining these issues are taken up in Chapter 16, which contains an account of regularization, power-counting, effective Lagrangians, and the renormalization group. Applications to the proof of perturbative renormalizability, and a discussion of the “triviality” phenomenon (the absence of a non-trivial continuum limit), follow in Chapter 17. Chapters 18 and 19 then explore important features of the behavior of quantum field theories at short distance (e.g., the operator product expansion and factorization) and long distance (in particular, the complications in defining the correct physical state space in unbroken abelian and non-abelian gauge field theories).vi Preface To the beginning student, quantum field theory all too often takes on the appear- ance of a multi-headed Hydra, with many intertwined parts, the understanding of any ‘one of which seems to require a prior understanding of the rest of the frightening anatomy of the whole beast. The motivation for the present work was the author's desire to provide an introduction to modern quantum field theory in which this rich and complex structure is seen to arise naturally from a few basic conceptual inputs, in contrast to the more typical approach in which Lagrangian field theory is presented as a theoretical fait accompli and then subsequently shown to have the desired physical features. Much (perhaps most) of the attitude towards quantum field theory expressed in this book is the result of innumerable conversations, over four decades, with colleagues and students. For laying the foundations of my knowledge of field theory I wish especially to thank my predoctoral and post-doctoral mentors (Steven Weinberg and Al Mueller, respectively). In the case of the present work I am extremely grateful to Estia Eichten, Michel Janssen, Adam Leibovich, Max Niedermaier, Sergio Pernice, and Ralph Roskies for reading extensive parts of the manuscript, and for many useful comments and suggestions. Any remaining solecisms of style or content are, of course, entirely the responsibility of the author.Contents Origins I: From the arrow of time to the first quantum field 1.1 Quantum prehistory: crises in classical physics 1.2 Early work on cavity radiation 1.3. Planck’s route to the quantization of energy 1.4 First inklings of field quantization: Einstein and energy fluctuations 1.5 The first true quantum field: Jordan and energy fluctuations 2 Origins II: Gestation and birth of interacting field theory: from Dirac to Shelter Island 2.1 Introducing interactions: Dirac and the beginnings of quantum electrodynamics 2.2 Completing the formalism for free fields: Jordan, Klein, Wigner, Pauli, and Heisenberg 2.3. Problems with interacting fields: infinite seas, divergent integrals, and renormalization 3 Dynamics I: The physical ingredients of quantum field theory: dynamics, symmetries, scales 4 Dynamics 11: Quantum mechanical preliminaries 4.1 The canonical (operator) framework 4.2. The functional (path-integral) framework 4.3 Scattering theory 4.4 Problems 5 Dynamics III: Relativistic quantum mechanics 5.1 The Lorentz and Poincaré groups 5.2 Relativistic multi-particle states (without spin) 5.3 Relativistic multi-particle states (general spin) 5.4 How not to construct a relativistic quantum theory 5.5. A simple condition for Lorentz-invariant scattering 5.6 Problems 6 Dynamics IV: Aspects of locality: clustering, microcausality, and analyticity 6.1 Clustering and the smoothness of scattering amplitudes 6.2 Hamiltonians leading to clustering theories 6.3 Constructing clustering Hamiltonians: second quantization 6.4 Constructing a relativistic, clustering theory 6.5 Local fields, non-localizable particles! Rowe 30 31 40 46 57 69 70 86 96 106 108 108 111 114 121 125, 130 132 133, 138 144 159viii 7 8 9 10 1 Contents 6.6 From microcausality to analyticity 6.7. Problems Dynamics V: Construction of local covariant fields 7.1 Constructing local, Lorentz-invariant Hamiltonians 7.2. Finite-dimensional representations of the homogeneous Lorentz group 7.3 Local covariant fields for massive particles of any spin: the Spin-Statisties theorem 7.4 Local covariant fields for spin-} (spinor fields) 7.5 Local covariant fields for spin-I (vector fields) 7.6 Some simple theories and processes 7.7 Problems Dynamics VI: The classical limit of quantum fields 8.1 Complementarity issues for quantum fields 8.2 When is a quantum field “classical”? 8.3. Coherent states of a quantum field 8.4 Signs, stability, symmetry-breaking 8.5 Problems Dynamics VII: Interacting fields: general aspects 9.1 Field theory in Heisenberg representation: heuristics 9.2. Field theory in Heisenberg representation: axiomatics 9.3 Asymptotic formalism I: the Haag-Ruelle scattering theory 9.4 Asymptotic formalism II: the Lehmann-Symanzik-Zimmermann (LSZ) theory 9.5 Spectral properties of field theory 9.6 General aspects of the particle-ficld connection 9.7 Problems Dynamics VIII: Interacting fields: perturbative aspects 10.1 Perturbation theory in interaction picture and Wick’s theorem 10.2. Feynman graphs and Feynman rules 10.3 Path-integral formulation of field theory 10.4 Graphical concepts: N-particle irreducibility 10.5 How to stop worrying about Haag’s theorem 10.6 Problems Dynamics IX: Interacting fields: non-perturbative aspects 11.1 On the (non-)convergence of perturbation theory 11.2 “Perturbatively non-perturbative” processes: threshhold bound states 11.3. “Essentially non-perturbative” processes: nou-Borel-summability in field theory 11.4 Problems 164 169 171 171 173 sues 184 198 202 215 219 219 223 228 234 238 240 241 253, 268 281 289 297 304 307 309 314 325, 341 359 371 374, 376 386 400 41112 13 14 15 16 Contents Symmetries I: Continuous spacetime symmetry: why we need Lagrangians in field theory 12.1. The problem with derivatively coupled theories: seagulls, Schwinger terms, and T* products 12.2 Canonical formalism in quantum field theory 12.3 General condition for Lorentz-invariant field theory 12.4 Noether’s theorem, the stress-energy tensor, and all that. stuff 12.5 Applications of Noether’s theorem 12.6 Beyond Poincaré: supersymmetry and superfields 12.7 Problems Symmetries II: Discrete spacetime symmetries 13.1 Parity properties of a general local covariant field 13.2 Charge-conjugation properties of a general local covariant field 13.3 Time-reversal properties of a general local covariant field 13.4 The TCP and Spin-Statistics theorems 13.5 Problems Symmetries III: Global symmetries in field theory 14.1 Exact global symmetries are rare! 14.2 Spontaneous breaking of global symmetries: 14.3. Spontaneous breaking of global symmetrie: 14.4 Problems the Goldstone theorem dynamical aspects Symmetries IV: Local symmetries in field theory 15.1 Gauge symmetry: an example in particle mechanics 15.2 Constrained Hamiltonian systems 15.3 Abelian gauge theory as a constrained Hamiltonian system 15.4 Non-abelian gauge theory: construction and functional integral formulation 15.5 Explicit quantum-breaking of global symmetries: anomalies 15.6 Spontaneous symmetry-breaking in theories with a local gauge symmetry 15.7. Problems Scales I: Scale sensitivity of field theory amplitudes and effective field theories 16.1 Scale separation as a precondition for theoretical science 16.2 General structure of local effective Lagrangians 16.3 Scaling properties of effective Lagrangians: relevant, marginal, and irrelevant operators 16.4 The renormalization group 16.5 Regularization methods in field theory 16.6 Effective field theories: a compendium 16.7 Problems ix 414 414 421 426 431 443 464 469 470 ATA ATT 478 485 487 489 492 495 507 509 509 512 519 529 544 552 565 569 570 571 574 581 588 595 608x Contents 17 18 19 Scales II: Perturbatively renormalizable field theories 17.1 Weinberg’s power-counting theorem and the divergence structure of Feynman integrals 17.2. Counterterms, subtractions, and perturbative renormalizability 17.3. Renormalization and symmetry 17.4 Renormalization group approach to renormalizability 17.5 Problems Scales III: Short-distance structure of quantum field theory 18.1 Local composite operators in field theory 18.2 Factorizable structure of field theory amplitudes: the operator product expansion 18.3 Renormalization group equations for renormalized amplitudes 18.4 Problems Scales IV: Long-distance structure of quantum field theory 19.1 The infrared catastrophe in unbroken abelian gauge theory 19.2. The Bloch—Nordsieck resolution 19.3 Unbroken non-abelian gauge theory: confinement 19.4 How confinement works: three-dimensional gauge theory 19.5 Problems Appendix A The functional calculus Appendix B Rates and cross-sections Appendix C Majorana spinor algebra References Index 610 613 645 652 660 662 664 679 698 708 aaa 7131 Origins I: From the arrow of time to the first quantum field 1.1 Quantum prehistory: crises in classical physics The first indications of serious inadequacies in the framework of classical physics— deficiencies which eventually could only be resolved by the introduction of quantum- theoretical concepts and methods—can already be found in Maxwell’s discussion of the anomalously low specific heat of gases in the mid-1870s. Nevertheless, the birth of quantum theory as such, and in particular the clear identification of a new fundamental constant of Nature characteristic of quantum phenomena, is usually located in the year 1900 with Planck’s invention (Planck, 19000) (“derivation” would perhaps be too charitable a term) of a novel formula for the distribution of energy over frequencies in thermal radiation (namely, electromagnetic radiation in the interior of a sealed enclosure which has been allowed to come to thermal equilibrium with the walls of the enclosure, themselves maintained at a fixed temperature T). ‘The problem of thermal cavity (or “blackbody” radiation would in modern terms be regarded as one of thermal quantum field theory, so we have the strange situation that the first historical impetus to the discovery of quantum principles actually lay in a problem of quantum field theory, which is generally supposed to be a much later invention arising from a fusion of quantum, relativistic, and locality principles. In fact, Planck’s papers of 1900 make no reference to quantization of the electromagnetic field-energy (a concept which Planck would continue to resist strenuously until the mid-1920s): the energy quantization principle for Planck is strictly a statement about the distribution of energy among the idealized material oscillators constituting the walls of the enclosure, and general principles of thermodynamics are then brought to bear to fix the electromagnetic energy distribution which must necessarily obtain once equilibrium between these oscillators and the interior radiation has been achieved. Only five years later, in his remarkable paper entitled “A heuristic point of view concerning the creation and conversion of light” (Einstein, 19052), Einstein was to extend boldly and explicitly the idea of energy quantization to the electromagnetic field itself, introducing the idea of “light quanta” (in modern language, photons). From a conceptual (if not technical) point of view, this paper therefore marks the true birth of quantum field theory. In order to understand the origins of quantum theory in the problem of black- body radiation, which to physicists of Planck’s generation must have appeared quintessentially classical (merging as it did well-established principles of Maxwellian electromagnetic theory and thermodynamics), we need to push our time horizon back a quarter century or so and survey the overall situation in classical physics at the start2 Origins |: From the arrow of time to the first quantum field of the final quarter-century of the 1800s. The three great: edifices of classical physics Newtonian mechanics (amplified and deepened, of course, by the contributions of Laplace, Lagrange, Hamilton, and many others), electromagnetic theory, only recently completed by Maxwell (in Philosophical Transactions, vol. 155, 1865), and, somewhat later, put in a form recognizable to the modern student of the subject by Hertz, and thermodynamics, which reached essential conceptual completeness at the hands of Clausius, also around 1865—stood as precise descriptions of natural phenomena, each apparently unassailable in its natural domain of applicability. In a sense, further progress keeping strictly within the limits of each of these disciplines had become difficult or impossible. However, precisely at this time, natural phenomena requiring the simultaneous application of more than one of these formal structures began to demand the attention of physicists. It is possible to trace the origins of the core disciplines of twentieth-century physics—quantum theory, statistical mechanics, and relativity—to developments at the interfaces of the three basic classical frameworks. We can summarize these developments very briefly as follows: 1. Developments at the interface of thermodynamics and electromag- netic theory. The discovery and classification of solar spectral lines by Fraunhofer, and the development of spectroscopy as an analytical tool by Bunsen and Kirchhoff in the 1850s, led naturally to an investigation of the radiation emitted by hot bodies, and thus to the study of the relation between thermodynamics and electromagnetic phenomena. It was immediately recognized by Kirchhoff that the intensity and frequency of the radiation emitted by a perfectly absorbing body had a fundamental significance. Various arguments—some of a rigorous thermodynamic nature, others of an heuristic character—led, by 1896, to a widely accepted form for this “blackbody distribution” (the Wien Law). The experimental failure of this “law” was the final stimulus which led Planck to (reluctantly) advance the quantum hypothesis in his seminal papers (Planck, 1900) in Annalen der Physik, 1900. » . Developments at the interface of mechanics and thermodynamics. Attempts to reconcile classical mechanics, regarded as the underlying dynamical description of all phenomena, with the formal principles of thermodynamics led to the development of kinetic theory by Clausius, Maxwell, and Boltzmann. The more general framework developed by the last of these has since come to be called statistical mechanics. Boltzmann was the first to understand clearly the role of statistical and probabilistic considerations in reconciling mechanics with heat theory (as thermodynamics was called at the time). However, his views were highly controversial at the time, especially with regard to his claim of a statistical origin for the phenomenon of irrever 'y in thermal physics. Although he would prove to be absolutely right on this point, Boltzmann’s methods proved incapable of explaining the observed specific heats of gases: this difficulty, early (before 1875) recognized by Maxwell (Maxwell, 1875) as a serious anomaly in classical theory, would only finally be removed by the application of quantum ideas in the 1920s, fully fifty years after the problem was first recognized.Early work on cavity radiation 3 3. Developments at the interface of electromagnetic theory and mechanics. ‘The historical and conceptual preeminence of classical mechanics implied a spe- cial status which led naturally to an attempt to interpret all natural phenomena in mechanical terms. In particular, the attempt to weld clectromagnetic theory with mechanics (begun by Maxwell himself) by main force led by the second half of the nineteenth century to the development of a profusion of aether theories of increasing complexity and artificiality. The failure to produce any direct evidence for the existence of an aether in optical experiments (by measuring relative motion of the aether and the optical apparatus of choice) grew from a mere annoyance into an outright crisis with the null result of the experiments of Michelson (1881) and Michelson and Morley (1887). The entire class of complicated and messy dynamical aether theories concocted to surmount this impasse were demolished with one stroke by Einstein (Einstein, 19054) in 1905, by accepting the kine- matical structure natural to electromagnetic phenomena as generally valid in the mechanical sphere also. Tn our outline of the conceptual origins of quantum mechanics and quantum field theory, the first item above holds pride of place: firstly, because by common consent quantum theory begins with Planck’s introduction of a new universal constant of Nature in his blackbody distribution formula of 1900, and secondly, because, as we shall see later, the first explicitly quantum field-theoretic calculation, Jordan’s derivation of the mean-square energy fluctuations in a subvolume of a cavity containing (a one- dimensional version of) electromagnetic radiation in the final section of the Drei- Ménner-Arbeit (1925) of Born, Heisenberg, and Jordan (Born et al., 1926), came directly out of an attempt to reproduce a remarkable result of Einstein dating from 1909 (Binstein, 1909b,a) in which the apparent paradox of simultaneous wave and particle behavior of light was first exposed with full clarity. The essence of quantum field theory is to provide a unified dynamical framework in which these apparently dis- parate behaviors can coerist in a conceptually consistent fashion. The electromagnetic radiation contained in a cavity at thermal equilibrium therefore plays a central role in the conceptual origins both of quantum theory generally and quantum field theory in particular. For this reason we shall retell in this chapter the history of thermal radiation in some detail, paying particular attention to those aspects important for understanding the conceptual origins of quantum field theory. The story will lead us continuously from the early arguments surrounding the role of the Second Law of Thermodynamics (and the “arrow of time” it implies) in blackbody radiation, to the appearance of the first truly quantum-field-theoretical analysis of electromagnetic radiation. We begin in the next section by describing some important milestones on the way to the understanding of blackbody radiation as it stood in 1900 when Planck took the first steps along the road to modern quantum theory. 1.2 Early work on cavity radiation The fact that heated bodies glow with a color and intensity varying with their temper- ature must surely have been apparent in prehistoric times (at least since the discovery4 Origins |: From the arrow of time to the first quantum field of fire!). The precise nature of thermal radiation became the subject of intense study in the nineteenth century, and as we shall see, led directly to the discovery of the quantum principle. Wedgewood, the porcelain manufacturer, observed in the 1790s that heated bodies all became red at the same temperature. The Scottish physicist Balfour Stewart noted (in 1858) that a block of rock salt at 100° C strongly absorbs the radiation emitted by a similar block at the same temperature, and suggested the rule of equality of radiating and absorbing power of bodies for rays of any given type (ie., wavelength). About a year later (independently of Stewart) Kirchhoff put all of this phenomenology into a comprehensible framework by the use of very general thermodynamic arguments (Kirchhoff, 1859, 1860). The arguments given by Kirchhoff established the universal character of blackbody, or “cavity”, radiation—the radiation filling the interior of a hollow material cavity, the walls of which are maintained at a fixed temperature 7’ It is important to understand at the outset the role of the cavity in these arguments. After heating the cavity to the desired temperature 7, a fixed amount of radiant energy fills the interior. To examine the nature of this radiation we are at liberty to drill a very small hole in the walls, as the small amount of radiant energy emerging will not sensibly disturb the established equilibrium. Note that from the point of view of the external world, the punctured cavity is essentially “black”, in the sense that any radiation entering through the pinhole will have to scatter around in the interior of the cavity for a very long time before having an opportunity to escape. Such a cavity is therefore (effectively) a perfect absorber, or “black body”. The essence of the problem is that radiation in the cavity is forced to interact with the walls (or contents, if any) of the cavity until thermal equilibrium is reached. Kirchhoff showed that the energy (per unit volume, and per wavelength interval) of the cavity radiation was uniform and isotropic throughout the interior. The arguments he gave were subsequently simplified by Pringsheim. The latter observed that by inserting a reflecting plane surface at some point inside the cavity, the equality of radiation in opposite directions follows, as otherwise the unequal radiation pressure exerted on the two faces would allow the spontaneous conversion of heat to work, thereby violating the Second Law of Thermodynamics. The existence of a pressure exerted by radiation reflected from a surface is crucial to these arguments: the precise relation (to be discussed further below) between this pressure and the energy density of the radiation had been derived earlier by Maxwell (Maxwell, 1873). Similar arguments employing two mirrors can be used to prove the isotropy and homogeneity of the radiation in the cavity. The existence of filters selectively absorbing and transmitting radiation of different wavelengths means that these statements hold separately within each interval of wavelength. Thus, we can define a function ¢(A, 7) such that (A, T)AA is the radiant energy per unit volume between wavelengths ,\ + AA anywhere in the cavity. "The use of the term “blackbody”, though historically predominant, frequently confuses the beginning student, who wonders quite naturally how a truly black body can radiate! In the forthcoming discussion we prefer the use of the term “cavity (or thermal) radiation”. The German terminology, “Normal Spektrum", is not particularly illuminating either.Early work on cavity radiation 5 Next, one easily sees that (A, 7) is the same (universal!) function for any cavity, irrespective of size, shape, or material constitution. This can be established by connecting two cavities at the same temperature by a thin tube allowing radiation to pass in either direction. Unequal radiation densities would result in unequal fluxes down the tube, and hence in a spontaneous flow of heat between two systems at the same temperature. Again, by placing filters which pass only a limited range of wavelengths in the tube, this equality is found to hold in each wavelength interval a The universal function ¢(A, 7) appears in the description of radiation emitted from any hot surface in the following way. At thermal equilibrium the radiation impinging on the surface must be balanced by the total radiation leaving, as otherwise the surface would grow progressively cooler or hotter. Suppose the surface (say, the interior wall of our cavity discussed above) absorbs a fraction Ay of the incident radiation flux ¢, (at wavelength A). It is clear that the radiation emitted (as opposed to reflected), or the “emissive power” of the surface By must just equal Aye), or equivalently, the ratio E)/Ay of emissive power to absorption coefficient is a universal function (essentially our old friend @) of wavelength and temperature. This result, of course, lies at the core of Stewart’s observations mentioned above. It also explains why thermos flasks are silvered (making A, small) in order to get them to radiate less. The total energy density of cavity radiation (at all wavelengths) is evidently just the integral of Kirchhoff’s function ¢: aT) = footer (a) In 1879 Josef Stefan proposed on the basis of some preliminary experiments the form (7) = aT* (1.2) with a a universal constant. In 1884 Boltzmann derived this formula thermodynam- ically, essentially by the following argument. Consider a spherical cavity of radius r. Maxwell had shown that radiation of energy density p exerts a radiation pressure 4p. Thus the internal energy is U = $rp and the heat absorbed in an infinitesimal reversible expansion is given by the First Law of Thermodynamics as? dQ = dU +4W = ag ar®p) + dar? xen (1.3) with the corresponding entropy change 4 dy 4 dp dT a8 = 4/7 = 409 @ 4 Cant = Sui OE 4 Cart a pig epg arta ep OA) 2 We remind the reader that U and S are state functions, unlike work W and heat Q: only the changes in the latter are meaningful, whence the difference in notation in the associated differentials W, dQ.6 Origins |: From the arrow of time to the first quantum field Equating 498 to 238 we find dp _ 4, ar" T° (18) which immediately implies the T* law stated above, and now known as the Stefan— Boltzmann Law. Serious attempts to measure experimentally the intensity and spectral composition of blackbody radiation began with Langley, the American astronomer, who invented the bolometer, a device for measuring the intensity of radiation in the infrared, using the principle of temperature-dependent resistance of a thin filament placed in the path of the rays after they were refracted through a rock-salt prism. These measurements (1886) extended up to about wavelengths of 5. He found “a real though slight progression of the point of maximum heat towards the shorter wave-lengths as the temperature rises” and the asymmetric form of the maximum of (A,7), steeper on the shorter wavelength side. By 1895 the greatly improved measurements of Paschen established the rule that the wavelength Am of maximum intensity was inversely proportional to the temperature T. Paschen’s measurements led him to propose, in 1896, the form 60,7) = BY exp(-4) (1.6) with the constant C’ somewhere in the range of 5-6. In 1893 Wien derived, using purely thermodynamic arguments, an important constraint on the Kirchhoff function $(A,T). Consider once again the spherical cavity used above in the derivation of the Stefan—Boltzmann Law. Imagine that the sphere undergoes slow, adiabatic compression, where the radius contracts steadily at speed v (v << ¢). At every reflection from this inwardly contracting sphere light of wavelength 2 suffers a Doppler shift to wavelength \(1—2v/c). During a contraction by Ar, occurring in time Ar/v, the light undergoes cAr/2ur reflections across a diameter of the sphere (it can be shown that light not incident perpendicular to the walls suffers a smaller Doppler shift cach time, but is reflected correspondingly more frequently: the net result is the same). The result is a total blue shift to wavelength 20, gar Ar a-)#a~a- ya so the wavelength is shifted by 44 = —4* in this adiabatic compression. As there is no heat transfer dS = 0 and (see Eq. (1.4) above) Tor ago aoa aa ala aia, T dra Te 37 7 1 sora a. r The essence of the thermodynamic argument given by Wien lies in the observation that the adiabatic process described here gives at every stage cavity radiation inEarly work on cavity radiation 7 equilibrium at the new temperature T' inversely proportional to r. If the result were otherwise, producing more or less radiation in some wavelength band than appropriate for blackbody radiation at the new temperature, introduction of filters absorbing in this range immediately allows the generation of temperature differences and a consequent violation of the Second Law. In the time interval At of the adiabatic compression, the radiant energy originally in the wavelength band (A, + 4), namely grr", r)AX (17) is shifted down to the interval (A(1 — 244), (A + AA)(1 — 24¢)), and increased by the amount of adiabatic work done against the radiation pressure, which is (4er)(0At)(40,7)A) (18) Thus 4 a(n — vAt)SO(A(L - , r—vasara — “At 3 7 = gros) - (Grr? wane, r)AX (1.9) Expanding to first order in At, one finds the equation or a. a5 gy + rg, )O%A) =0 so that, $= fA/r) = FAT) Wien phrased this result slightly differently. Since 1 o> orslOP) it follows that if iT, = A2T», then + sen) = 70027) ? which Wien called the “T®” law, but is now commonly called the Wien Displacement Law. If we know the radiation function (1,71) at temperature 7), the law allows us to “displace” it into the appropriate curve for any other temperature Tp, as (A, Ta) = Fol ap, qT)8 Origins |: From the arrow of time to the first cuantum field It follows immediately from the Wien Displacement Law that the total energy density or) = [ ax0.7) = { ans fear) oe ie f de, : x (x) =aT* (1.10) satisfies the Stefan-Boltzmann Law (provided, of course, that the integral converges: a condition by no means to be taken for granted, as we shall see). Another imme- diate corollary is the result later verified experimentally by Paschen (but suggested previously by several workers in this field, notably H. F. Weber) that the maximum in wavelength displaces inversely with the temperature. Finally, the Wien Displacement Law immediately fixes the value of the constant C in the form proposed by Paschen (see (1.6)) to be 5 exactly. Independently of Paschen’s work, Wien in 1896 arrived at the form (1.6) on the basis of an ad hoc assumption concerning the emission of radiation by molecules distributed according to a Maxwellian velocity distribution. This form, which is now a complete specification of the Kirchhoff function ¢, was called the Wien Distribution Law, and was to play, with its “corrected” version, the Planck Distribution Law, a critical role in the evolution of attempts to understand quantization of the electromagnetic field. 1.3. Planck’s route to the quantization of energy Max Planck, the father of quantum theory, was born in Kiel, Germany, in 1858, and attended the Gymnasium (high school) and University in Munich before going to Berlin for his doctoral degree, where he had classes from Helmholtz and Kirchhoff. His doctoral thesis concerned the application of thermodynamics (2 la Clausius) to problems of “Evaporation, Melting, and Sublimation”. Planck was fascinated by the extraordinary scope, power and (apparent) infallibility of the energy conservation principle—or First Law of Thermodynamics—and accorded to the Second Law of Thermodynamics (with its concomitant “arrow of time”) an equal degree of validity. He was therefore convinced that the Second Law could not rest on the purely mechanical foundations of Boltzmannian gas theory, in which the reversibility objections of Loschmidt and Zermelo (the latter a Planck assistant in Berlin) would necessarily lead to spontaneous processes (albeit rare) where the entropy decreased.* The consideration of the paradox of Maxwell’s demon also led Planck to the rather peculiar conclusion that the Second Law could never be valid in @ system comprised of discrete particles Instead, Planck began to investigate (in 1897) the possibility that irreversible thermal phenomena could somehow be traced back to irreversible processes in a continuous medium—in particular, the electromagnetic field. The archetypal process considered by Planck was the apparently irreversible conversion of plane radiation 5 For a beautiful retelling of this remarkable period in the development of statistical heat theory, see the biography by Martin Klein of Paul Ehrenfest (Klein, 1970).Planck's route to the quantization of energy 9 incident on a charged oscillator into outgoing spherical waves. This subject was explored with great thoroughness in a series of five papers in the Berliner Berichte (1897-99) entitled “ber irreversible Strahlungsvorginge” (“On irreversible radiation processes”) (Planck, 1900a). The subject of absorption and re-emission of electromag- netic radiation from an oscillator led Planck naturally into the subject of thermal cavity radiation. Here the oscillators constitute the material of the walls of the cavity, absorbing and re-emitting the radiation in the interior. The universal character of the thermal radiation discussed above allowed Planck the freedom of making a very simple model of the constituent particles of the walls (essentially charged simple harmonic oscillators), as the spectral distribution of the cavity radiation would have to be independent of the specific material constitution of the cavity once equilibrium is reached. The irreversibility that Planck relies upon in his radiation studies can be seen clearly in the damped oscillator equation that he derived as a prelude to his studies of the coupled field-oscillator problem: 2. 2 gs nit + ke — a = eF cos(2nvt) (1.11) The third (“radiation damping”) term has three time-derivatives and evidently changes sign under time-reversal. It arises because the damping force times the velocity must give the power lost to radiation, which is proportional to the acceleration of the charged particle squared. The average of the third term above times the velocity 42 over a cycle of the periodic system is easily seen to be the same as the average power radiated, by a single integration by parts. Planck was particularly impressed by the fact that the irreversibility in this system arises without any recourse to non- conservative processes, in which ordered energy is lost (as in friction or air resistance) to disordered heat. Instead, the energy appears to flow irreversibly from an ordered source (an incoming plane wave incident on the oscillator) to an equally ordered form: outgoing spherical radiation. In 1898 Boltzmann succeeded in convincing Planck (in a paper entitled “On the supposedly(!) irreversible radiation processes” (Boltzmann, 1898)) that the hope of deriving irreversible phenomena. from electromagnetic theory without additional statistical assumptions was bound to fail, as Maxwell’s equations are just as invariant under time-reversal as those of classical mechanics. In fact (Boltzmann claimed), in the course of a careful derivation of radiation damping one is forced to apply boundary conditions to the fields which amount to a field analog of the assumption of molecular disorder implicit in the Boltzmann approach to gas theory.* Planck admitted this promptly and abandoned the attempt at a “microscopic” explanation of irreversibility based on electrodynamics. In the fifth of his papers on irreversible radiation processes (Planck, 1899), Planck derived a crucial formula relating the distribution function for cavity radiation to the average energy of his fictional oscillators (at equilibrium). Before stating this formula, 4See (Klein, 1970), (Kubn, 1978) for masterful expositions of the remarkable developments at the interface of mechanics and heat theory summarized all too briefly above.10 Origins I: From the arrow of time to the first quantum field aslight change in notation will be convenient. Let p(v,T)dv be the energy/unit volume of cavity radiation in the frequency interval (v,1 + dv), where vA = ¢, |dv| = dA, so that pv, T)dv = ant) gaa = (A, T)dd 2 ce p.t) = ~9(a,7) = So(2,7) (1.12) In terms of p(v,T), the T® law takes the form eT) = 4(Z) (1.13) and the Wien Distribution Law is p(v,T) = av’ exp(—Bv/T) (1.14) where the new constants a, @ are related to those appearing in the Paschen result (1.6) by a= 4,8 = 4 (recall that the constant C was fixed previously by purely thermodynamic reasoning to be 5). The equation derived by Planck, obtained by equating at equilibrium the energy absorbed and emitted by the oscillator, stated simply E(vo,T) (Yo, T) (1.15) ~ Be? It relates the average energy of an oscillator of natural frequency 1, the blackbody distribution function. That such a relation must exist is physically clear. Planck showed that if the left-hand side exceeded the right, energy would flow from the oscillators to the electromagnetic field, while if the intensity of radiation at vy became large enough that the right-hand side exceeded the left the oscillators would tend to absorb energy from the field. The importance of this equation (a full derivation of which we must unfortunately forego, in the interests of brevity) in Planck’s intellectual journey can scarcely be overemphasized: it allowed him to restrict the application of energy quantization to the material oscillators alone (left-hand side of (1.15)), while relying on the equilibrium condition to transfer the resultant average distribution of energy by main force, as it were, to the continuous electromagnetic radiation (right-hand side) in the interior of the cavity. Planck would continue to insist on the continous, purely classical character of electromagnetic radiation for the next 25 years. In his final paper on irreversible radiation processes (see (Planck, 1900a)), Planck gave a “derivation” of the Wien Law based on purely thermodynamic arguments together with the crucial formula (1.15) above. This was done by making a plausible assumption for the form of the entropy J of the oscillator as a function of energy E, using for the inverse temperature T~! = 35, and solving for E as a function of T. Planck showed that his assumption for S(E) implied that the entropy of the whole system (oscillators plus radiation) would necessarily increase in time, in agreement with the Second Law of Thermodynamics. He was also under the (as it later turnedPlanck's route to the quantization of energy 11 out, erroneous) impression that this was the only possible choice for $(Z) consistent with the Second Law. Consequently, at this point Planck was quite convinced that he had finally managed a complete derivation of the blackbody spectrum from pure thermodynamics (even if he had now to agree with Boltzmann that the Second Law had a statistical rather than absolute significance, even in radiation phenomena) On the afternoon of Sunday, 7 October 1900, Planck was visited at home by an experimental colleague from the Physikalische-Technische Reichsanstalt (the Physical- ‘Technical Imperial Institute, or PTR), H. Rubens. He learnt from Rubens that recent experiments at the PTR had established incontrovertible deviations from the Wien Distribution Law on the infrared (low-frequency) side. In particular, the intensity was roughly proportional to temperature in this regime, instead of the saturation at high temperatures implied by the Wien Law (1.14). Planck realized that a more general form for the oscillator entropy S(E) would in turn allow the derivation of a modified distribution law . 1 = BI OT) =o I (1.16) which clearly reproduces the Wien Law at higher frequencies, but behaves like | ~ ST LA? 2 (117) in the infrared (small v), showing the desired linear behavior with T. This interpolating formula, which Planck appears to have constructed in the few hours following the visit of Rubens, was checked within the next week and a half and found to match exactly the experimental data. Planck was perfectly aware that his interpolating formula was nothing more than an enlightened guess at this stage, and he began right away to search for a proper understanding of the formula (1.16). His strategy was preciscly the inverse of the one he had followed heretofore. He used (1.15) to obtain the average oscillator energy, assuming the validity of the Planck distribution (1.16): hv BON sp Buf (1.18) where h = 2%. He then reconstructed the corresponding expression for oscillator entropy as a function of energy, using the thermodynamic relation TdS = dE valid for a reversible transformation involving transfer of heat but no external work. Here, oscillators of a fixed natural frequency v (called vp above) are considered. Solving (1.18) for 1/T’ as a function of E: (1.19)12 Origins |: From the arrow of time to the first quantum field and integrating, one obtains, — i = es [oe + hv) —In(Z))dE h E E E = gilt pin + iw! apart from an irrelevant, integration constant. The problem now shifted to finding a “fundamental” explanation for this last expression. By this point in late 1900, Planck had been converted to Boltzmann’s statistical approach, and he now adopted the teckniques used by the latter for gas theory in an attempt to establish (1.20) by microstatistical reasoning. Thus, the entropy was to be determined by taking the logarithm of the number of available microscopic states consistent with the stated macroscopic parameters, S = kIn(W) (here k is Boltzmann's constant). Like Boltzmann, Planck introduced a finite-energy unit € to facilitate the counting. The total energy Ey shared by N oscillators was a (large!) integer P number of these units, Ey = Pe. Planck then “counted” W by simply computing the number of ways in which P units of energy could be distributed among the N oscillators. The combinatorial formula needed for this can be derived rapidly using a characteristically elegant trick due to Ehrenfest. Write out a string of P energy units €, with dividers to indicate how many units belong to the first, second, etc., oscillator: (1.20) eelecelelee...lee There are P of the « symbols and N — 1 dividers. First assume that all these symbols are distinguishable. There are then (P-+N —1)! ways of ordering them. As the dividers and energy units are (separately) indistinguishable, we have overcounted by a factor (N — 1)!P!. Thus the desired result (using Stirling’s approximation to evaluate the factorials of large numbers) is (P+N-1)! $= kw Sy yr ) ~ k(N + P)In(N + P) — Pin(P) — Nin(N)) The average entropy of each oscillator S = oscillator is E = Ey = Fe. Consequently 4Sw while the average energy of a single S=HQ+ =)t (1.21) This is exactly the relation (1.20), provided we identify = hv, 4 = k. In other words, the derivation of the new distribution formula forced Planck to keep the energy units ¢ finite, even at the end of the calculation. Setting ¢ to zero here, as Boltzmann had done at the end of his gas theory calculations, would be equivalent to setting the constant h to zero, which would lead to an incorrect distribution law (the Rayleigh-Jeans Law,Planck's route to the quantization of energy 13 to be discussed further below). Apparently, the oscillators in the walls of a cavity were only allowed to have energies in integer multiples of the basic energy “quantum” ¢ = hv! The arguments outlined above were presented in Planck’s paper in Annalen der Physik 4 (1901),p. 553, “Uber das Gesetz der Energicverteilung im Normalspectrum” (“On the law of energy distribution for the normal [i.e., blackbody] spectrum”). The famous Planck's constant h appears here for the first time, In modern notation, the blackbody distribution thus takes the form 8x hy “S exp(hv/kT) —1 (1.22) pv, T) = From the experimental fits, Planck determined h = 6.55 x 107?” erg/sec, and k (Boltz- mann’s constant)= 1.346 x 10-1 ergs/degree. The latter value allowed Planck to obtain the first decently accurate value for Avogadro's number N = R/k (where R is the gas constant). It is a strange historical irony that Planck’s modification of the Wien’s Law, motivated by the pressure of the Kurlbaum-Rubens experimental results, was actually a move towards a “more classical” result: as Einstein was to emphasize in his epochal 1905 paper (Einstein, 19052), in which the revolutionary idea of field quantization was introduced, the Wien Law is in a sense an extreme manifestation of the quantal properties of light. The deviations observed from this law in the infrared by Kurlbaum and Rubens are harbingers of the reappearance of the classical wave-like aspects of electromagnetic phenomena. To understand this we must realize that despite Planck’s heroic efforts to obtain a rigorous and unique classical result for the distribution func- tion of cavity radiation throughout the 1890s, leading up to the quantum-theoretically correct Planck distribution (1.22), the first derivation of the blackbody distribution based on a consistent and full application of classical principles is actually due to Lord Rayleigh. In a short (two-page) paper published in 1900 (Rayleigh, 1900) Rayleigh derived the correct classical form of the distribution function from the classical equipartition theorem applied directly to the electromagnetic modes in the cavity. Consider a cubical LxLxL box containing electromagnetic radiation in the form of standing waves. A typical standing wave mode takes the form mymE, . meTy . MsTz ) sin( re sin( 2H) sin( ST) where the associated frequency is v = 3571 and 71 is the vector with (positive) integer Cartesian components (n1,n2,n3)- The number of such modes in the shell ((7i), || + djit|) (octant of positive components only!~—an error of Rayleigh’s later corrected by Jeans, see below) is evidently 1 Bs gitinPdin| = 4a V7dv and each of these modes receives a total of 2kT at equilibrium by the equipartition principle (namely, 447 each into electric and magnetic field energy, and each of two polarization modes). Thus the energy per unit volume in the field in the frequency interval (¥,v + dv) is14 Origins |: From the arrow of time to the first quantum field pv, T)dv : RT (4 a Pdi = 8 ekT dy (1.23 Ae ee eae ) ‘a result which has since become known as the Rayleigh-Jeans Law. (The error mentioned above of an overall factor of eight made by Rayleigh in his original paper was subsequently corrected by Jeans. As Pais points out in his biography of Einstein (Pais, 1982), the correction was made also in Einstein's 1905 paper on the light quantum, so the result should perhaps more properly be called the Rayleigh-Jeans-Einstein Law.) Rayleigh was perfectly aware that this result could not be correct: the total energy contained in the cavity radiation, when integrated over all frequencies, would then be infinite! Instead, he assumed that it was correct only for the “graver modes” (i.e., lower frequencies) and that: the distribution wes modified for some as yet unknown reason. at higher frequencies (Rayleigh simply inserted an exponential suppression factor at high frequencies, and the resultant formula was in fact his final result). In any event the simple linear dependence on temperature in the Rayleigh-Jeans Law flies in the face of experience: a bar of steel at room temperature (300 K, say) does not emit radiation at one-tenth the blinding intensity of a bar at 3000 K ! The infinite amount of energy present in the classical radiatior. field under equipartition would later (1911) be referred to by Ehrenfest (Ehrenfest, 1911) as the “ultraviolet catastrophe” Of course, if Planck had finished his Boltzmannian calculation of the average oscillator energy by taking the energy units € to zero, as Boltzmann had done previously in his discussion of gas theory, 1e would have arrived precisely at Rayleigh’s result (though he does not seem to have been aware of Rayleigh’s work during the critical period leading up to the 1901 paper), as the Rayleigh-Jeans Law is simply the h — 0 limit of the Planck distribution. That he did not do so is probably due to a combination of reasons: 1. He does not seem to have regarded equipartition as a fundamental guiding principle to the same extent as other physicists of a more “mechanist” bent. Planck attacked the problem from the point of view of the behavior of the oscillators at thermal equilibrium, rather than by directly considering the modes of the electromagnetic field itself, which would have led much more quickly to the (wrong!) classical result. The result obtained by setting the energy units to zero would not have agreed with the Wien Law, with which Planck had started and which he knew to be empirically correct at higher frequencies. n bad 1.4 First inklings of field quantization: Einstein and energy fluctuations Although Planck succeeded in obtaining an absolutely correct expression for the equilibrium thermal frequency distribution of electromagnetic radiation in a cavity, there is absolutely no indication that he supposed any sort of energy quantization to hold for the electromagnetic field itself. Instead, the (at this point frankly magical) effect of the energy quantization imposed on the material oscillators receiving from and transferring energy to the radiation in the interior was forcibly transferred to the electromagnetic field via the equilibrium formula (1.15). The field itself, PlanckFirst inklings of field quantization: Einstein and energy fluctuations 15 was to insist for almost another full quarter century, was a continuous, fully classical entity regulated by Maxwell’s equations. The situation was to change dramatically with Einstein’s remarkable 1905 paper, “On a heuristic point of view concerning the creation and conversion of light” (Einstein, 1905a). Although this paper is now commonly referred to as the “photoelectric paper”, Einstein spends much more time in it on an analysis of the volume-dependence of blackbody radiation (pp. 92-102) than on the brief discussion (pp. 104-105) of the photoelectric effect. ‘After pointing out that a strictly classical analysis must necessarily lead to the Rayleigh-Jeans result (1.23), with its inescapable concomitant ultraviolet catastrophe, Einstein goes on to analyse cavity radiation in the high-frequency domain, drawing some extraordinarily non-classical conclusions from the quintessentially “classical” (at least from an historical point of view) Wien Law. Einstein’s approach in this paper is radically different from Planck’s. He focusses first and foremost on the thermodynamic and statistical properties of the electromagnetic radiation in the interior of the cavity. Taking a cavity of volume Vo and considering only the electromagnetic radiation in the frequency interval (v,v + dv), the energy E of such radiation in the high-frequency domain where Wien's Law (1.14) holds is given by 8rhv? o EB Voe" BF dv (1.24) Solving this equation for 4 and repeating the integration procedure of (1.20) to obtain ‘an expression for the entropy S of the electromagnetic radiation in this frequency interval, one finds (the 0 subscript indicates that the radiation in the entire cavity of volume Vo is being considered—we shall shortly consider radiation in a subcavity) KE E So =~, (ln (Geaenagae) ~ 13 (1.25) ‘The same amount of radiation confined to a smaller volume V would lead to an entropy $ with exactly the same form as (1.25) but with Vo replaced with V. Accordingly, the difference in entropy for the two situations is S-—S =—b(— = kin( eM (1.26) Vo ‘The fundamental Boltzmannian association of entropy with the probability W of the associated microstates of the system, S = kln W, then leads to the conclusion that _ Wa Elke (1.27) i.e., that the probability of an energy fluctuation leading to a concentration of all the electromagnetic radiation in the frequency interval (v,v + dv) in the subvolume V of the full cavity Vo takes exactly the form which we would expect if that radiation consisted of = “mutually independent energy quanta” (cach of energy hv) moving freely throughout the cavity, in complete analogy to the behavior of molecules in a gas. This is as far from the classical picture of electromagnetic radiation as extended waves subject to mutual (destructive and constructive) interference as it is possible to16 Origins J: From the arrow of time to the first quantum field get. The result (1.26)—extraordinarily simple, but profoundly baffling, from a classical point of view—clearly had a deep impact on Einstein’s thinking. He was to hold firmly to the concept of energy (and later momentum) quantization of the electromagnetic field over the next 20 years—a period of time in which the majority of physicists were firmly on Planck’s side and resistant to any notion of quantization of the sacred classical Maxwellian fields. The centrality of blackbody radiation to Einstein’s thinking about the nature of the electromagnetic field is clear once one reflects on the number of occasions on which he would return to the subject: to take the most prominent cases, in 1909 in two papers (Einstein, 19096,a) (one entitled “On the present status of the radiation problem”, the other “On the development of our conceptions on the nature and constitution of radiation”) in which energy fluctuations were once more used as a diagnostic for exposing the underlying properties of radiation, and in 1917, in the famous “A-B coefficients” paper (Einstein, 1916, 1917), of critical importance in the later development of dispersion theory by Kramers, and thereafter in the 1925 development of matrix mechanics at the hands of Heisenberg, Born, and Jordan.5 Here we briefly review Einstein's results of 1909, which proved to be a critical inspiration for Jordan’s introduction in 1925, in the last section of the “hree-Man” paper of Born, Heisenberg, and Jordan (Born et al., 1926), of the first true quantum field. In returning to the problem of energy fluctuations in cavity radiation, Einstein decided to relax the simplifying assumption of high-frequency (or low-density) radia- tion described by the Wien Law, and to enquire into the implications of the full Planck distribution (1.16), valid at all densities and frequencies, for the fluctuation properties of thermal radiation. In this case, instead of considering the highly non-Gaussian process whereby a fluctuation would concentrate 100% of the radiation energy in a given interval (v,v+dv) in a subvolume V (giving the result (1.27), later to be called “Einstein’s first fluctuation theorem” by Jordan) Einstein decided to calculate the mean-square energy fluctuation of the energy in this interval in the subvolume V. The formula for such mean-square fluctuations is a standard result of statistical mechani 2 2 dE) : (Any) = eet (1.28) where T is the temperature and (£2) the mean energy, which in this case is clearly just Vplv,T)dv. We can distinguish three interesting choices for the energy distribution p(v,T) and corresponding mean-square energy fluctuation. We shall distinguish the results obtained for the mean-square energy fluctuation in each case by a subscript indicating the assumed form for the universal Kirchhoff distribution function p(v,T): “RJ” for the completely classical Rayleigh—Jeans form, for the Wien Law, and “P” for the final result of Planck. In the case of the Rayleigh-Jeans Law valid at low frequencies, 5 For a thorough study of the role played by dispersion theory in the birth of modern quantum mechanics, see the two-part paper by M. Janssen and the present author (Duncan and Janssen, 20074,6)