Qc Sponsoring Etor: Peer Rene Project Eto: Naney ight Manuseript Eto: Rath Ves Designers: Gary A. Head and Shaten I, Sith Production Coontinatr: Frank Mickell ‘Mlastaton Coordinator: Batyoh Fanos Arts: Flin Coopee Conposior:SyisxIteratcoal DAA SS OK S2 ADSo c oby 4 Library of Congress Cataloging in Puiston Data Kitt, Charis, “Theta psc, Bibliography: Includes cx Hsrben 1928" oincauthor aL Tei QEIII.KS2 isxd 7 qp.tcen ISBN 67167-10889 (Copyright® 980 by W. H. Freinan and Company No pata ths book may be reprontcs by any ‘weshanical, photographie, oF eetraie process, off 1s form oF phonograph vecoeding. nar may be sored in retrieval ern aramid, cr hers ‘opie foe public or privat use without writen petmisson fromthe publics Printed inthe United Stat of Ameen ‘Twenty-fiest printing, 2000 Staite thermodyeunis, 1. Krome, About the Authors Charles Kittel has aught solid state physics at the University of California at Berkeley since 1951, having previously been at the Bell Laboratories, His undergraduate work in physics was done at MALT. and at the Cavendich Laboratory of Cambridge University. His Ph.D. research was in theoretical ‘nuclear physics with Professor Gregory Breit at the University of Wisconsin, He has been awarded three Guggenheim fellowships, the Olivet Buckley Prive ‘or Solid State Physics, and, for contributions to teaching, the Oersted Medal of the American Association of Physics Teachers. He is a member of "ie National Academy of Science and of the Ametican Academy of Arts a ‘Selnces. His research has been in magnetism, magnetic resonance, semi: ‘ductors, and the statistical mechanics of solids Hebert Krositer is Professor of Elzetrical Engincering at the University of Californie at Santa Varbara, His background ani training are in solid stave Dhysies, He receive a Ph.D. in physics in 1982 from the University of Cttingen in Germany with a thesis on hot electron effets in the then new keatsister From 1952 through 1968 he worked ia several semiconductor research lab foris in Germany and the United States. In 1968 he became Professor of [lectical Enginecring at the University of Colorado; he eame to UCSH fa 1976, tis research hus been in the physics and technology of semiconductors and semiconductor devices, including high-frequency transistors, negative: ‘nats effects in semiconductor, injection lasers, the Gunn effect, electron-hole «drops, and semiconductor heterojunctions.Preface ‘This book gives an elementary account of thermal physics. The subject is simple, the methods are powerful, and the results have broad applica tions, Probably no other physical theory is used more widely throughout science and engineering. ‘We have written for undergraduate students of physics and astronomy, and for electrical engineering students generally, ‘These fields for ove Purposes have strong common bonds, most notably a concem with Ferri ftses, whether in semiconductors, metals, slurs, oF nuclei, We develop methods (not original, Lut aut easily accessible elsewhere) that are well suited to these fields. We wrote the book in the fist place because we ‘were delighted by the clarity of the "new" methods as compared to those ‘we were tought when we were students ourselves. We have not emphasized several traditional wopies, some because they ‘are no longer useful and some because their eeliance on classical satisti- cal mechanics would make the course more difficult than we believe a first course should be, Also, we have avoided the use of combinatorial ‘methods where they are unnecessary. Notation aad units: We generally use the SI and CGS systems in parallel, We do not use the calorie. The kelvia teriperature T is related to the fundamental temperature 7 by += Fy7, aud the conventional entropy. 5 is telated to the fundamental entropy o by $ kya The symbol log, will denote natural logarithm throughout, simply because In is less ex Pressive when set in type, The notation (18) refers to Equation (18) of the current chapter, but (3.18) refers t0 Equation (18) of Chapter 3 The book is the successor fo course notes developed with the assist. ance of grinis by the University of California, Edveard M. Purcell con tributed many ideas to the first edition. We benefited from review of th secand edition by Seymour Geller, Paul L, Richards, and Nicholas Whecter. Help was given by Ibrahim adawi, Bemard Black, G. Domo- kos, Margaret Geller, Cameron Hayne, K. A. Jackson, S. Justi, Peter Kittel, Richard Kistler, Manin J, Klein, Ellen Leverene, Brice H. J. fcKellar, F. E. O'Meara, Norman E.Phillips, B. Roswell Russell, T. M.treface Sanders, B. Stozchly, John Verhoogen, Wichmann. We thenk Carol Tua Wilde for her help with the index. An clemeniary treatment of the greenhouse effect in the sphere was added in 1994 on page 115, following an at by Professor Richard Muller. A page on atomic gas experiments on the Bose-Einstein condensation was added 10 page 223 in 2000, For instructors who have adopted the course for classroom use, @ solutions manual is available via the freeman web site (hitpsshfrecman, ‘comu/hermalphysics). Joh Wheatley, and Eyvind wg for the typed manuscript and Sari aeth's skmo- rguiment suggested Berkeley and Santa Barbara Charles Kitel Herbert Kroemer Note to the Student mum coverage of the concepts presented in each chapter, the authors ‘ecommend the following exercises. Chapter 2: 1, 2,3; Chapter 3: 1.23.4, &, 14; Chapter 4:1, 2,4, 5, 6,8; Chapter 5: 1,3,4, 6, 8 Chapter 6: 14, 15; Chapter 7:2, 3.5, 6,7, 11; Chaptec 8: 1, 2,3, 5,6, 7; Chapter 9: i, 2,35 ‘Chapter 10: 1,2, 3; Chapter 11: 1,2, 3; Chapter 12: 3, 4 5; Chapter 13: 1,2, 3.7.8, 10; Chapter 14: 1,3, 4, 5; Chaptor 15: 2, 3,4, 6Contents Guide to Fundamental Definitions xiii General References xv Introduetion 1 Chapter 1 States ofa Model System 5 Chapter Enttopy und Temperature 27 Chapter Raltzmana Distribution end Hctuholiz Free Energy 55 Chapter 4 Thermal Radiation and Planck Distribution 87 Chapter 5 Chemical Potential and Gibbs Disuibution 117 Chapter6 Weal Gas 151 Chapter? Fermi and Bose Gases 181 Chapter8 Heat ahd Work 225 Chapter 9 Gibbs Free Energy and Chemical Reactions 261 Chapter 10 Phase Transformations 275, Chapter 11 Binary Mixtures 309 Chapter 12 Cryogenics 333 Chapter 13 Semiconductor Statistics 353 Chapter 14 Kinetie Theory 389 Chapter 15 Propagation 423 Appendiv. A Some Integrals Containing Appendix B Temperature Scales 445 xponentials 439 xAppendix © Appendix D Appendix Inder 465 Poisson Distribution 433 Pressure 459 Negative Temperature 460 Guide to Fundamental Definitions Absolute activity, 2 expt) 139 Accessible state 29 Boltzmann constant, by 25 Boltzmann factor, expl—e/s) 6 Boson 183, Chemical potential, 119 Classical regime, ce my 74 Ensemble of systems 31 Enthalpy, HH = Ut p¥ 246 Entropy,e 40 Fermion 183, Gibbs factor, expf(Nu — ele] 138 Gibbs free energy, G —rot py 262 Gibts or gfand sum, 3138 Heat capacity, C63 Heat,Q 68,227 Helmholtz free energy, F = U~ 10 68 Landau free energy function, F, 298 Multiplicity, 7 Orbital 9 Partition function, Z 61 ‘aiGuide to Fundamental Defaitons Quantum concentration, ng = (Ma/2xh%? 73 Reversible process 64 Temperature, 41 ‘Thermal average 62 Thermal equilibrium — 39 Work, WW 227 General References ‘Thermodynamics A.B, Pippard, Elements of classical shermodnanics, Cambridge Univenity Press 196 M.W. Zemansky and R. H. Dittman, Heat and thermodynamics: an inermediare ‘exibook, 6 ed, MEGraw-Hil, 198) Suisical Mechanics BLK, Agaoval and M. Eisner, Staiteu! mechanics, Wiley, 1988 TL Mil Statsical mechanis: prncptes and selected applications, Dover Pubtica- inn, 987, 1956, © Kitt, Brocton staistcal pysce, Wiley, 1938, Pans 2 and 3 ea applications to noise and 10 elementary transport theory. Part thas been expanded into the R. Kubo, Statice! mechanic, North Hollnd, 1990, 1965, RR. Kubo, M. Tada, N. Hashitsime, Staniticalphysis 1! (Noneguilibrtu, Springer, 1983, 1. Landauand E, M.Liphit,Staristca physics, 3d ed by EM. LiShitgand LP Pitaersi, Pergamon, 1980, pat 1 Shang-Keog Ma, Staiic mackanics, Wosld Sientifc, 198. M, Toda, 8 Kubo, N. Suto, Statistical physice £ (Equilibrium), Springer, 1983 Mathematical tables HB. Deight, Tables of integrals and other mathematical data, th ed, Macmvittaa, 1961, A widely useful small collection, Applications Asteophysiss RJ. Taylon The star: their siictare and evohaton, Spingse, 1972, S. Weinbus, The fist thrce minuess a madern view of the ogi ofthe taverse, new fa, Bautsin Books, 1984 ‘Biophysics and macromolecules T.L Hil, Cooperatvityheory in bischenistry: steady state and equiltbeian systems, Springer, 1985General References CCoyopenies end low temperature physi. G. K, White, Experimental tetniques in low-temperature physics, 3d ed, Oxford Univesity Press, 1947, 1979 «pa, 4. Wika and D, 8. Beas, a intron tof guid helium, 2nd ed, Oxford Univesity Pres, 1987 Irreversible thermodynamics 2A McLennaa, Introductian 1 nom equ statistical mechanics Peete Mad, 1539, 1. Prgorine and 1, Stenger, Onder out of chaos: man's new dialog Random House, 1984, ee Kinetic theory aad transpon phenomena §.G.Bossh, The kind of motion we cll heat, North-Holland, 1986, e191. HL Smith and HHL Jensen, Transpar phenamena, Onfocd Univey Press, 1989, Plasma phases 1 Spitee I Physical proceses inthe intratllar miedvan, Wiley, 1978. Phase transitions PP. Peay and G. Toulouse, Snraduaion othe renormalization group end to teat phenomena, Wiles, 1977 HE Stanley, Jetceton to phase trsiions an eritical phenomena, Onferd Usk veaity Press, 1987. ‘Metals 2nd atlogs P. Hasen, Ppsical metllvgy, 20d el, Cambridge University Press, 1986, Superb, Bousdary value problems HS. Cardaw and J.C. Jaeger, Conduction of heat in solids, 2nd ed, Oxford Univer: sity Pres, 1986, ©1989, Semiconductor devices R, Dalen, Fzoducton co applied slid stare physics, Plenum, 1990, K Seeger, Semiconductor phyticr: on inraduction, Sth ed, Springer, 1594 SUM. Sze, Physics of semiconductor devices, 208 td, Wiley, (981. state physics C Kittel, Intracuction solid state physics, 6th ed, Wiley, 1986, Reeredtoas ISSP.Thermal PhysicsIntroduction Ourapproach tothermal physics diters fram the tradition follosed in beginning physics courses. Therefore we provide this introduction to Set out what we are ‘oing to do in the chapters that follow. We show the main lies ofthe logical structure: in thi subject all the physis comes from the loge, In ordee of their appearance, the leading characters in aur story are the entropy, the lemperature, the Boltzmann factor, the chemical potential, the Gibbs factor, and the eistribus sion fonctions ‘The entropy measures the number af quantum states accessible toa ystor A close system might be in any of these quantum states and (we assume) with ‘equal probability. The fundamentt statistic element, the fundamental logical assumption, i that quantum states are either accessible or inaccessible to the system, sd the system is equally ikely to be In any one accessbte state a8 any other accessible state. Given g accessible states, the entropy is defined as 2 = logg. The entropy thus defined will be a function of the energy U, the number of particles N, and the volume 1 of the system, because these param cers emter the determination of g2 other parameters may enter as well. The use of the logarithm is a mathematical eonvenionce: itis easier to write 102° than exp({0), and it is more natural for wo systems to speak of ay + oy than of gis ‘When two systems, each of specified energy, are brought into thermal coatact ‘hey may wansfer energy; their total energy remains constant, but the constraints, on their individual energies are lifted. A transfer of energy in one direction, or peshaps inthe other, may increase the product gg, that measures the number of accessible states of the combined systems. The fundamental assumption biases the outcome ia favor ofthat allocation ofthe total energy that maximizes the number of accessible states: more is better, and moe likely, This statement is the kernel ofthe law of increase of entropy, whict isthe general expression of the second law of theemody namics, ‘We have brought 1wo systems into thermal contact so that they may transfor energy. What isthe most probable outcome of the encounter? One system will, ‘gain energy at the expense of the other, and meanwhile the total entropy ofthe two systems will increase. Eventually the entropy will reach @ maximum for ‘the given total energy. Is not dificult to show (Chapter 2) that the maximum, 1Invocton raitained when the value of @6/€U}.y for one system is equal to the value of the same quantity forthe sec "ond sjstem, Thisequality property fortwo systems aera contact isjentthe propery weexpect ofthe emperatue, Accordisly, We define the fundarncutal tesnpecature + by the telition 0 ‘The use ofl atsures that energy wil Now fom high +10 low r;n0 more come Pleated eelaton is eccded. It will follow thatthe Kelvin temperatore sisoty proportional to, with r = AyT, where ky isthe bolo he conventional enteony 8 is given by § = ine Now consider a very sinyle exsnupe ‘Chapter 3. et sasall syste with say eon the Melizmsnun factor treated jin ies, une a enerny @ a naray ¢ be placed in thermal contact with a arge system that we eat ihe fessrvoit. The total energy of the combined systems is Uss when the smell ‘yatem is in the state of energy 0, the reservoir has energy‘ Us and wil hneg atl ohstates accesible voit, When the small scm isin the state of enersy nha ‘Metvoie wil hse energy Up ~ eand will have gle ~ «states accesible to i By the fanamental assumption, the rato of the probability of finding the Small sytem with emery « 19 the probably of nding it wih energy Oe MO We ~ 9 ofelt—o] : Fo) oO” ~ expat ; ‘The ceservoirenicopy a nigy be expanded in & Taylot series: SIU, - = al) ~ Alef) = olUy) ~ sie, % Dy the defiition (1) of the temperature, Higher order ray be dropped. Cancellation of the term exglaiU, ‘numerator and denorninator of (2) after the sul terms i the expansion a} which occurs in the stitution of (3), leaves us with Peay expel. “ ‘This is Boltemann’s result, To show is us ‘erny (e) ofthe twa state system in ther perature ¢ , weealeulate the theemal average mal contact with a reservoir at tem- © = YaPe) = 0-700) + ew) w 6 Iuadecton Where we have imposed the normalization eondit abilities: ® on the sum of the probe PO) + Phe) = 1 0 ‘The argument ean be generalized immediatly to find the average eneray of a harmonic oscillator at temperature s, and we da this in Chapter 4 as the frst step in the derivation of the Planck radiation law. ‘The most important extension of the theory i to systems that can transfer particles a8 well as enerey with the reservoir, Fortwo systems in difusive and thermal contaet, the entropy will be a maximum with respect to the transior of particles as well a8 to the transfer of energy. Not only must (26/CU.» be ‘equ for the to systems, but (76/Nyy- must also be equ, where A fers 19 the womsbe of parties ofa given specks He new eulity eto i the son for the iatvenbuction fa nee qustity, the bal pte y o For two systems in thermal and difusive contact, ry = ry and ly = py. The sign in (7) cuosen to ensure that the direction of particle fow as equitbrium 's approached is front high chemical potential to low chemical potent “The Gibbs factor of Chapter 5 isan extension of the Boltzmann factor and allows «to treat systems that ean zransfer particles. The simplest example is System with two states, one withO particlesand 0 energy, and one with 1 particle and energy «. The system isin contact with a reervolr at temperature and shiemical potential . We extend (3) forthe reservoic entropy: {Ug ~ €:Ny = 1) = a(UaiNe) ~ elea!dv = 1 @efanig = a(UaiNe) ~ ut + af. 8 By analogy with (4), we have PU.ey POM) = expllu ~ ae), ° for the ratio of the probability the system is occupied by t putticle at energy & to the probability the system is unoccupied, with eneray 0. The result () after normalization is readily expressed as t Pe Ses aT 0)‘This particutae resus is known a6 the Fermi-Dirac distsibution function and is used partioularly in the theory of metals to describe the electron gas at low temperature and high concentration (Chapter 7}, “The classical distribution function used in the derivation ofthe ideal gas tas js just te Tit of 10) when the occupancy P(e) is much less than PQs) = expfln — av on “The properties ofthe ideal gas are developed from this result in Chapter 6 ‘The Helmholtz free energy Fm U — te appears as an important computa ‘ional function, because the relation (2F/ét)sy = —< offers the easiest method for finding tke entropy, once we ave found out how to calculate F from the energy eigenvalues (Chapter 3} Othor powerful tools for the calevation of thcrimodynsmic functions are developed in the text, Most ofthe remainder of the text concerns applications that ae usafil in their own right and that illum nate the meaning and utility ofthe principal thermodynainic anctions. ‘Thermal physies connects the world of everyday objects, of astro objects, and of chemical and biologie provesses withthe world af molecu tom, und eleetconieayatema. I ites the two pasts war wut ta misty scopie and the macroscopic. Chapter 1 States of a Model System BINARY MODEL SYSTEMS Enumeration of States and the Multigliity Function Binary Alloy System ‘Sharpness of the Multiplicity Function AVERAGE VALUES : Enorgy of the Binary Magnetic Systema Example: Multiplicity Funotion for Hatmonie Oscillators SUMMARY. » 6 6 18 2 2 2 261 Gibbs A theory ts the mre in more diferent kinds spplicabiliy. Therefore the deep ‘pan me. Iris the only physical theory of uniter vill necer be averdvown, within she, concepts. pressive the greater she sinplicty ofits premises, the of things it relaes, and the more extended ts area of inpression tha classical thermodynanies made content which Ian convinced Sronework of applicability of us basic A. Finstcin haper ts es of a Modet Sytem Thermal physics is the treit ofthe union of statistical and mechanical principles Mechanics tells us the meaning of work; thermal phsics ell us the meaning of heat, There are three ne¥ quantities in thermal physics that do aot appeat tn ordinary mechanies: entropy, temperature, and free energy. We shall motivate their definitions in the frst three chapters and deduce their consequences theceater Our point of departure fr the development f thermal physics is the concept of the stationary quantum states of sjstem of particles. When ve ean count the quantum states accessible to a system, we know the entiopy ofthe system, for the entropy is defined as the logirithm of the numberof states (Chapter 2) ‘The dependence ofthe entropy an the energy ofthe system defines the tempera. tore. From the entropy. the temparatucondihe heecctey we NG the pressure, the chemical potential, and all ther thermodynamic propeties of the systerm For system ina stationary quantum state al ebserable physical proserticg such as the energy and the nutnber of panicles are independent ofthe tine For bievity we usually omit the word stationary; the quantum states that we teet ars stationary except when we discuss transport processes in Chapters 4.15 The systems we discuss may be composed ofa single particle of, more often, ‘of many particles. The theory is developed to handle general systems of inten, acting patticles, but powerful simplifications ean be made in special problems for which the interactions may be neglected Each quantum state has a definite enesgy, States with identical energies ace said to belong to the same energy level. The multiplicity or degeneracy of an nerey level isthe number of quantum states with very nearly the same energy, {tis the umber of quantum states that is important in thermal physics, not the number ofenergy levels, Wes frequently deal with sums overall quantum states, Two states atthe same enerey must always be counted 38 tivo states, os one fevel Let ustook a the ysantun systems The simplest is hydrogen, with one ecciean stl one proton. The lowly ing ‘energy levels of hyrogen are shown in Figuce 1. The zero of energy in the figure is taken at the state of toxest energy. The number ef quantum states belonging to the ame energy level esa energy lvstvofseser ten isin parentheses. ta the figure we overlook that the proton has a spin of Yh and has two indspendentofientations, paralelCompre Ve States of Sade Sper up 7 t | ew si a 2b ——u soe —w 5 4 + ——a tes ‘3 & 4 4 = — a J 3 2 y tof | eee Hydrogen ——a of —_w Lithium Boron Figure Lt Lowstying energy feels of atomishydrogts thin, snd boron. The ergs se given in electron vols with LeV = 1402 » 10°"? erp The numbers fo parentheses pve the numberof quactum sates having the Same cacy, with vo account laken ofthe spin of the nedeus. The eco a enerey inthe pure is tae for convenience at she lowest energy state of exch atom, rantiparallel to the direction ofan arbitrary external axis, such asthe dicetion, of a magnetic field. To take account ofthe wo orientations we should double the values ofthe multiplicities shown for atomic hydrogen, ‘An atom of lithiuen has three electcons which move about the nucleus. Each ‘lectron interacts with the nucleus, and each electron also interacts with all the leper 22 Stes of Mat Sten o mr Mubiptcity of quantuan nunbets —— « Soboiyaas -—— 6 sa ask -— 3 422 —— 332 t—— « s21 2s] —— 3 2430 f——— 3 ang h—— 3 322 os gad 1 222 -— > 314 3 224 3 zat Figure 1.2 Energy evel, clipes, and quantum numbers sty ofa parti coined 03 cube other electrons. The energies ofthe levels of lithium shown ia the figure are the collective energies ofthe entiesystem-The energy levels shown for boron, which thas five electrons, are also the energies of the entice system, “Theeneegy of systemisthe t nergy ofall particles kinetic plus potential, th account taken of interactions betwicen paticls. A quantum state ofthe system is a state of all particles. Quantum states of a one-particle system are called orbitals, The low-lying energy levels ofa single particle of mass M con- fined to a cube of side L are shoven in Figure 1.2. We shal find in Chapter 3wo % Particle can be charactsied by thee postive integral ‘The enesgy is quantum sumbersn, WP fx\? eB (Jot eat ens ® Tr eis ofthe letls ars incited inthe Rue. Te thee orbitals Tehetanteaunt ob) (ht) and (i s)alhaven? + m2 tas 8; the coresponding energy evel has the multipisiy 3, cate the statistical properties ofa system of panei extemal Sosened ee convenient arbitrary way, but two dieeat sate should ane Ce sssigned the same index 118 800d ides to start our program by studying the properties of simple Roel aatem or which the energies) can be cleulted exactly Wechoneg sound ate sinble binaty system Because the general satnicel peop Fuad fr the model system are believed ta apply equly well to aig see Ebsfal sstem. This assumption leads to predictions tat sluaye sere te seve e ata Eenera statistical properties are afconcer wil become ca 2s we goalong, BINARY MODEL SYSTEMS Pambicd OF snoscupied, z10 oF one, plus one or minus one. The ties ee phoniatat siles with dierent numbers are supposed nat to overlay ta Bisset soe: Youmight even thnk of hess as mumbuted prkitgseas ec Sat batking lot asin Figuce 14 Each pking ypocehas tostaten eoenng ‘occupied by one et Matever the nuture af our objects, we may designate the we sta cro ate only point straight up or stesight dn Ifthe magnet ping magne et he magnetic moment is +m. ithe magnet ins dove ihe ‘magnetic mement ism, Binary Model Systems 7 8 9 w te Figwe 13 Model system composed of 10 elementary magnets afte sites ona in, each having magnetic ‘moment +m. The aumbers shown ae attached io the sitess ch site has its own magnel. We assune tere ae ne Ieteractons among ths mapncs and theres no excceat ‘aeti eld. Each magnetic moment may te olen a {07a w75, up OF down, so that there ase 2" dine rangement of he 10 magoetic momseats shown inthe ‘gure. the arrangements are selected ina Fado process {he probability of ding the parila arrangement showy mine, | Number of the sit 2 Fine 14 State ofa parkieg lot with 10 numbered parking sp3ei. The ©'s denote spaces ocaied bya ans the O's oe vacant spaces This partis state equate to hat shown in Figure 13, Now consisr W dierent sts, each of which beats a moment that may issue the values tin. Each moment may be oriented ia two ways with & Probsbility independ of the orientation of all other moatents, The totsl nanber of arrangements of the N moments is 2x 2x 2x02 a2 4 stale ofthe systems is pacified hy giving the oricutatfon ofthe sito: thete are 2* states We any slate of the system of W sites lomeat on each ¥ vse the following simple notation for a single tnunitt @(Chapter Is States of Mosel System He L3 The four diferent states of meee || || || | I ‘The sites themselves are assumed to be arranged in a definite order. We may ‘namberthem in sequence fom left to right, as we di in Figure 13 According to this convention the state (2)also ean be weitten a8 Ifalalalstetatatotso eo Both sis of symbols (2) and (3) denote the same state of the system, the state in which the magnetic moment on sitet i +m on site 2, the moment is +m ‘onsite 3, the moment is ~ ry; and go forth, kis not hard to convince yourself that every distinet state ofthe system fs sontained ina symubullc product of factors (he + Wl + LMS Fe ta bed 6 ‘The multiplication rue is defined by (hth +) sit tlt hte + hb. 6 ‘The function (4) on multiplication generates 3 sum of 2" terms, one for each of ‘the 2Y possible states. Bach term isa product of N individual magnetic moment symbols, with one symbol for each elementary magnet on the line. Each term denotes an independent state of the system auld is @ simple product ofthe form Tefal "Ty, for example. Fora system of two elementary magnets, we multiply (Ty + L,)by (ts + La) to obtain the four possible states of Figure 1.5 (a FAs +1) Tila tele + Lata # dele © ‘The sumis nota statebut isa way of isting the four possiblestates of thesystern. ‘The product on the let-tand side ofthe equation iscalled a generating function: it generates the states of the system, Binary Moet ‘The genecoting function for the tates ofa system of theee magnets is (hy # Lita Lala + La ‘This expression on moltiplication generates 2” = 8 aiferent states ‘Three magnets up ht Twomsgnetsup: fytals Tilaly —nfat Onemagnetup: Tylaly Lataly —falats None up: Alas. ‘The total magnetic moment of ovr chedel system of N magnets each of ‘magnetic moment m willbe denoted by Af, whieh we will relate to the energy ina magneticield, The value of M varies from Nin to ~ Niu. The set of possible values is given by A= Nom, (N= 2pm, 49m, (N= Oe, Km ‘The set of possible values of Mis obtained if we start with the tate for which ail ‘mogaets are up(M = Nin)and reverse one ata time. Wemay reverse N magnets to obiain the uhimate state for which all magnets are down (M = — Nn). ‘There are N + 1 possible values of the totat moment, whereas there ate 2° states. When N21, we have 2° > N+ 1. There are many mote states than values ofthe total moment, IFN = 10, there are 2°° = 1024 staes dsieibuted Among 11 different values ofthe total magnetic moment. For lerge N’ many different states of the system may have the same valve of the total moment A. We will elculae in the next section how many states have s given value of M. Only one state ofa system has the moment Af s: Nor: that state i ttt tit 6 re N ways to form a state with one magnet down: Mites ttt ° is one such state; anothoe fs titan (10)“ Chapter: Stores "Model Sytem And the other states with one magnet down are form ed from (8) by reversing any single ignet The states (9) and (10) have total moment Af = N, Enumeration of States and the Multiplicity Funct ‘Nese the word spin s a shorthand fr elementaty magaet tis convenient to assume that N isan even number. We necd a mathematical expression for the {aber of states with Ny ~ AN 4 s magnets up and N= Ws magnets Govsn, where an integer. When we turn one magnet fom the upto the cows orientation, WY + s goes to UN +5— 1 and $N ~s goes oN — sat Tie dilfecence (number up ~ number dovn) changes om 2s 102s 3. The dierence Ny-M,= 2s ay jected the spaecess, The spin exces ofthe 4 states in Figuee 1.362.040, ~2, {am eto sit. The factor of nH) appcirs to bea nuisance a this sage Dut twill prove to he convenient. ‘The product jn (4) may be written symbolically as be ‘We may drop the ste tabels (the subicrpts) from (4) when we are interested only in how many of the magnets in a state ate up oF down, and not in which brticular sites have magnets up or down. {t we drop the labels and neglect ‘he order in sich the arrows appear in a given produet, then (5) becomes (HM Tr ate ty further, es tits se att + Uy We find (F + )* forarbitcary N by the binomial expansion aa AN NSIT & AN — Doty? goog ye (2) Enumeration of Sates and he Biutipity Function We may write the exponents of x and yin a lightly different, but equivalent form by replacing ¢ with 4 — 3 dnl FS (ety airy (3) With this result the symbolic expression (f+ 1)* becomes Pee nt ses pave = Ser arcgel ro) ‘The coefficient of the term in) {s the number of states having x, iN + § magnots up and Ny = 4 — s magnets down. This class of states has spin excess Ny ~ WY, = 25 and net magnetic moment Jom Let us tlenote the numberof states in this class by g(Ns), for a system of magnets a5) ‘Thus (14 s written as ae ANS) AY 9) ‘We al cat (the multiplicity function; tis the number of states having the same value of s. The reason for our definition emerges when a magnetic field is applied tothe spin system: in a magnetic fel, states of diferent values of S have diferent values of the enexgy, so that our gis equal to the multiplicity ofan energy level in a magnetic field. Until we introduce a magnet field, alt states ofthe model system have the same energy, which may be taken as zero, Note feom (16) thatthe total numberof states is given by wy XY aitts) = + 1 = 2% an 9 Examples related to g(N.s) for = 10 are given in Figures 1.6 and £2. Fora coin, “heads” could stand for “magnet up” and “xails" could stand for “magnet down” B16 Chapter I: Stoter ofa Model Sytem 210} [210 Fee 16 Number of distinct aerangemenss 2f5 4 sspins upand 3 ~ sins dwn, Volos of gia) efor N= TOs where is the spineauess NT WL The total aber of| i” 120 E ati0.. ‘The values ofthe sare taken froma table oF the bizomis coe eins io | | L apoRae aw Spin excess 26 Binary Alloy System To illustrate that the exact nature ofthe Wyo states on each sites irrelevant to the result, we consider an alternate ssstem—an alloy crystal with WN distinct sites, numbered from 1 through 12 in Figure L8, Each sites occupied by ether ‘an atom of chemical species A or an atom of ches sion for vacant sites, In brass, A could be single state ofthe alloy system can be ical species B, with no provi- ‘copper and B zine. In analogy to (3), ADBIBSADBSAGDBeDyA yA Ars > = 08) Binary Ay Sytem ” | ven namber of heads occurred fh CTisasereow Number of heads igure 17m expzroen was Gone a wien 18 pennies wee thrown 100 times The number ofteads in eh Shrove was econ, ©0080 © OO) @ Serene ‘ccupy dintinet numtere sterates of e Moule Sytem Every distinet state of a binary alloy system on NV sites is contained in the symbolic product of W factors: (Qc # BIA + BAL + BY y 4B), (9) in analogy to (4). The average composition of a binary alloy is specified con= ventionally by the chemical formula A, B,, which means that out ofa total of N atoms, the number of A atoms is Ny = (1 —sN and the number of B atoms is Ng = AN. Here x lies between D nd | ‘The symbolic expression wt (A + By = J ar 2" 8 Le ia {is analogous to the result (12). The coefficient of the term in AY“! BY gives the hhumber 20%, of possible arcangements of states of N— atoms A and 1 atonts B on N sites ae Wom" yy en which s identical tothe result (15) for the pin model system, except for notation ‘Sharpness of the Meltiplicity Funetion We know fiom common experience that systems held at constant temperature ‘usually have well-defined properties; this stability of physical properties is a ‘major prediction of thermal physics. The stability follows a8 a consequence of Ubeexceedingly sharp peak in the multiplicity function and of the steep variation ‘of that function away from the peak, We can show explicitly that for a very large system, the function g(¥,s) defined by (15) is peaked very sharply about the value s = 9, We look for an approsimation that allows us to examine the form of g{N.s) versus s when N>> 1 and {sfc N. We cannot fook up these Values in tables: common tables of fetorials do not go above N = 100, sid we ‘may be interested in = 10° ofthe ander of the number of atoms in a solid specimen big enaush to be seen and fel. An approximation is clearly needed, andl a good one is avaiable, 1k is convenient 10 work with logg. Becept where atherwine spot afl Fogatitans are understood to be log base, writen heres log. The international standacd usage is Infor fog base e, but itis clearer ta write fog when there ig no ambiguity whatever. When you confront a very, very large number such is Sharpness ofthe Mattiplisiy Function 10° isa simplification tolook atthe Logarithm ofthe number. se the logarithm of both sides of (15) 0 obtain og aN. fog! — log(h + 3) ~ tog = 91, (22) by virwe ofthe characteristic property ofthe logarithm ofa product. logay = logs + logy; tog(vs) = tose ~ logy, (23) With he notation Nees My en for the number of magnets up end down, (22) appears as legets) BN! — log Ny! ~ log Nut e9 We evaluate the logarithm of WH in 25) by use of the according to which log anpensinnation, resto sel) st 2aNENM exp Nb Y2N) + 1 6) for N > 1. This result is derived in Appendix A, For sufficiently large N, the terms 1,121) + +> in the argument may be neglacted in comparison with N. We take the logarithm of bath sides of (26) 10 obtain log! & Hlog2e + (N + S)logN — N. en Similarly logy! & Hog2n + (Ny + NlogM, — Ns (8; logit = Hogle + (Mi + Dloga ~ Me. (2) Aller rearcangement of (27), log! & Hosea Eth +L N+ logy = (Ny ND, wesliave uss for (25) BN fesubteact ta (29) from GO}tw obtain ogg & Hog(l/2nN) - ON, + HlogiNy/A) = OW + HlogN uN OH »0 Copter): States ofa Model System This may be simplified because FoulNUN) = log {Cl + 291N) = —log? + foglt + 2s!) log? + (2578) ~ s4/N4) (32) by virtue of the expansion logll + x) ex fst valid for x & 1 Sinularly, lop(N./N) = log ~ 25/8) = log? — 251M) - GS/NR, 03) (On subsiution in (31) we obtain ogg & {log(/xN) + Ntog2 - 224)v, 04) We write this esutt as [ a AN Dhexpl—29)N} , 03) where (0.0) = (fey, 26) Such 8 distribution of values of sis called a Gaussian distribution. The integeal™ (0 (35) over the runge ~ co to +00 for s gives the correct valve 2Mor the total umber of states. Several useful integeas are treated in Appendix A, ‘Thc exact value of g(N,0) is given by (15) with s w Tm OW ot 1B asco sm by a8 neat seh oC. fl Mavs doesnt nade ‘ict erre Foreampla thecal nt eh x 44M) to Meds = $n Inequaliot 4 UM, which approaches 1a. appecactes 0, Sharpness ofthe Matiplsy: Faection » Figore 19. The Gaussian approximation to tne binomial coeFicents lM} pated on a linea sale On tis stale not possible ssingush onthe drawing the approsimation from the exact vals evs the range os plotted The entire range of sis rom ~ SO10 “+50. The dashed line redrawn fom the points a je of he simon alu of (100, For N = 50, the value of (500) fs 1.264 > 10, from (37), The approximate value from (30)is 1.270 x 10" The distribution plotted in Figure 9 is centered ina maximum at s = 0. When s* = N, the value of gis reduced toe"! of the ‘maximum value. That is, when ai = (RN)? , 29) the valve ofg is e°* of (NO) The quantity (1/20) is thus a reasonable mea sure of the fcactional widlh of the distbution. For N= 10%, the fractional ‘width is ofthe order of 10°"! When Nis very lage, the distribution is exceed ingly sharply defined in a relative sens. Ik is this sharp peak and the continued sharp variation of the multiplicity function fr from the peak that will ead to 2 prediction that the physical properties of systems in thermal equilibrium ate well defined. We now consider one such property the mean value of 2,a (Chapter I: States ofa Model System AVERAGE VALUES ‘Tis rege value or mean value, ofa fiction fs) taken aver a probabitity ‘stibution function P(s) is defined as P= Esorre) , 69) ‘Provided that the distribution function is normalized to unity LP 1 40) ‘The binomial distribution (15) has the property (17) that Lava) = 2, (4) nd is not normalized to unity. Hall states ae e GIN.32, nd we have SPs) willbe "ally probable, thea P(s} = = 1. The average of 3) aver this distribution D =F M9 PW, 2) ‘Consider the function ts) = 3. in the approximation that led to (38) and (4, we replace in (2) the sum Y, over s by an integral [+ ds between = and +, Then 1 _ CUA? 2° fide exp 25409 a fasten : BreAy (AIM F deers Pa CM yay, whence GP =A «aH 483) The quantity <7) is the mean square spn exces. The root mean square : spin excess is QsPy'? = WN, (44) Energy of he Binary Magnctic Sytem 45) ‘The larger N is, the smaller isthe fractional actuation. This meens that the ental peak of the distribution faction becomes relatively mose sharply defined as the size of the system increases, the size being measured by the Dumber of sites W. For 10° particles, # = 10-4, which is very small ‘Bnergy of the Binary Magnetic System ‘The thermal properties of the model sysem become phy the elementary magnets are placed in a magnetic field for then the cneteie of the diferent states are no longer il equal. Ftheensrgy ef the system is specie then ouly the states having this energy way eacuts The encryy af interstion ‘of single magnetic moment m with a fixed external magnetic eld Bi Cs =m, 40) hiss the potential energy of inthe itd For the model system of 2° ceseatary magnets, each with two allowed orientations ina uniform magnetic fc 8, the total poteatial energy U is De 30 = 0: Fm, 2) a, the expression M for the total magnetic moment 2am In this example the spectrum of values of the energy U is discrete. We shall see later that 2 con- ‘inuous or quasi-continuous specicum wil create no dificulty. Furthermore, the spacing between adjacent snergy levels of this model is constant, as in Figute 110. Constant spacing is a special fextute of the particulur move, bu this feature will not restrict the generality of the ar tive following sections, ‘The ealue of the energy for momeets What interact only with the external magnetic ticid i comeletcly determined by the value of s. This funetional Sependence ws wcicatia by writing UUs) Reversing a single moment lowers 2s by ~2, lowers the total magnatie moment by nd rises the enerey by 2mB. The energy dillerence between adjacent levels is donoted by de, where iment that i developed in Ae = Uy) Uls $1) = 2m «sy 2Chapter 1a States ofa Blake Systems : Lorna 56) tog sts) “St o wa “a tsa tty E535 o——— 9 om 553 FE s35, 9 was st Ht ao Ht ° Figure 1.10 Energy hve of the mode system of 10 ‘netic momeatsm ina magnetie eld The eels sv abso Uy st sale, whete 251s the spn teen “SAN + 95 + 515 he numberof up spi The ‘encebes U() and lips fate shown Fo his problem the engy leveled egy we Separation Az = Ill betwosnsjesot ene Example: Muliticy funtion for barmoic ocilaters, sysen isthe simplest problem for which an tacts “nowa, Another exactly solvable preblemis the hao ‘The problam ofthe binary mode ion for the multiplicity function is icosilator, for which the souton where the quanturs number sisa the exilatr. The umbe of Bato = ho (0) Expy ofthe Binary Megnatc Syters ‘an be distributed among the esilatrs, That ig, we wan the mliplisiy function la) foc the ocintocs. The ossitator mulpiity function not the Samet the spin lis plist asction found esti, We begin the aoaysisby ging back tothe makipiity fncion fro single ent forshich gl) = [forall atuetof the quantum numbers bereidenical ton Totove he rable of (3) below, we teed a fancion to repretcut or fscrae the sees othane = Fe co) AI ra fom Oto oo. Here ris just & temporary tol tha sil help us Gnd the cess (53) but does not appear in the final eesslt The answers (32) Provided we assume |< 1. For the problem of N oscitatrs, the geetating fonction is the tesa the ber of ways en am aa it the HA pos is pely sur of etal ways ia eich de integer mean be faces the smb N nomena inegers ‘Weiner that Wy 8 oe sin = tin (EY ava 1 (dy yx = ton (f) 9 = EM EO OO 6 “That for the system of ssilators, Nan witW =O! (59) (08) = "Ths res willbe needed in solving a problem in he next hapler Fy26 (Chapter 1: States of « Model Syne SUMMARY 1, The moltiplieity function for asystem of N magnets with spin excess 25 = Ny Mis i “NINE waa Ji the limit <1, with A'2» 1, we have the Gaussian approximation ats) = 7 N}AD exp ~25/N), 2 tall states of th i #€ modet spin system are equally likely, the verage value of 4s Lesa] f* asain =, in the Gaussian approximation 3 The factional Huctuation of st is defined as C425! pnt 4. Theeneray oft IN and is equal 10 vodel spin system in a state of spin excess 2sis UG) =, sn, ‘here mis the magnetic moment of one spin and B isthe magnetic fed Chapter 2 Entropy and Temperature FUNDAMENTAL ASSUMPTION » PROBABILITY a Example: Construction of an Ensemble 2 Most Probable Configuration a Example: Two Spin Systems in Thermal Contact 2 ‘THERMAL EQUILIBRIUM » ‘TEMPERATURE ENTROFY Example: Entropy tncrease on Heat Flow 4 Law of Increase of Enttopy 4“ LAWS OF THERMODYNAMICS, 4“ Enteopy as a Logarithm 0 Example: Perpetual Motion of the Second Kind 50 SUMMARY st PROBLEMS 1. Entropy and Temperature 2. Poramagnetisin 2 3. Quantum Harmonie Oscilator 2 4. The Meaning of "Never" 3 5. Aduitivity ofthe Entropy for Two Spin Systems 3 6. Integrated Deviation St by Probons 3 edo aotempPs Chapter 2: Eniopy and Temperature ne shoul nor tmagine the 0 gees i a 0.1 Her container, il is, then agin after a few days separa, : te contary, one fins... 19° years will hereby a recom int mice, then mis again on forth. On hot not unitate ens ng compared ts ny dceable nny af the gosee. One my that this is practically equivalent to never. : 7 1 Botcmae Ye wish 0 find in rational mechanics an a of thermodynamics, cntropy. prior foundation forthe principles Ne must Seek mechanical definitions of temperature and A.W. Gibbs The aeneral conection horween enorgy and temperature nuiy nly be esrabtshed by probability considerations. [Two systems} are hn seats! equilorium when ‘transfer of energy does not increase the prababliny M, Planck Fundamental aasumgtion We stun this chapter with a definiion of probability that enables us 1 defiae the average value of a physical property of a system, We then consider systems in thermal equilibrium, the definition of entropy, and the definition of temperature. The second iaw of thermodynamics will appear as the law of increase of entropy. This chapter is perhaps the most abstract in the book. The chapters that follow ssl apply the eoneepis to physical problems. FUNDAMENTAL ASSUMPTION ‘The fundamental assumption of thermal physis is that a closed system is equally likely to be in any of the quantum states accesible to it, Al accessible quartunt states re assumed to be equally prudable—sheve & Hu HeaseR 4O prefer some accessible states over other accessibe states, A closed sy hhave constant energy, 2 constant number of particles, constant volume, and constant values of all external parameters that may influence the system, including gravitational, electric, and magnetic feds. quantum state is accessible if ils properties are compatible with the plysical specification ofthe system: the energy of the state must be in the range within which the energy ofthe system is specified and the number of particles must be jn the range within which the numberof particles is specified, With lage systems ‘we can never know either of these exactly, but it will sufice to have BUJU « 1 and BN/N 1 Unusual properties of a system may sometimes make it impossible for cxttain sates tobe accessible during the time the tysier is under observation. For example, the states of the erystaline form of SiO, are inaccessible at low temperatures in any observation that starts with the glassy oF amorphous {orm: fused silica will not convert to quartzin our lifetime ina low-temperature caperiment. You will recognize many exclusions of this type by common sense, We treat all quantum states as accessible unless they are excluded by the specification ofthe system (Figure 2.1) and the time scale of the measurement Process, States that are not accessible ae std to have zero probability. Of cours, i is possible to specify the configuration of a closed system to a point that its statistical properties as such are of no interest. Ifwespecify that the »2 (Chapter 2s Entropy and Temperate Wig 2A pu oma ceieseis an csc quant atc ofa ed pe, The {urdanestalssngtion of sutra pei etn ‘tem eal iy tote imanpotine eae ne Accs ti, The emay cts epee os ey Ina ace not acess bcs the poperon oo roioee shespeciction of hess oie ns ingame std spo system i exactly ina stationary qua the problem, ania PROBABILITY Suppose we have a closed system that we know is equally likely to be in ofthe g accessible quantum states, Let she general sate late fan wot one ‘the spin excess The probability (9) of finding the system inthis se any hall 7 Mo w ithe states is acocesible and P() = 0 atherwse consent with the fun damental assumption. We shall be concered fle with syste that ae ane closed, for which the energy U and particle number N may tay. For thee ‘stems PO)will no bea constant asin} but willhavesfunetonal dependence on U and on N. Probubity The sum Ps) ofthe probability ove all sates is aways equal to unly, because the total probability thatthe system isin some state iy Leet rH The probabilities defined by (1} lead to the definition of the average value of any physical property. Suppose that the physical property X has the valve X(} when the system isin the state s. Here X might denote magnetic moment, ‘nergy, square of the energy charge density near a point r,o¢ any property that ‘san be observed when the system isin a quanturn state, Then the average of the observations of the quantity X taken over a system described by the proba. bilities Pu is mao CD = Exits, | a ‘This equation defines the average vatue of X. Mere P(x is the probability hat the system is in the state s. The angular brackets (+) are used to denote average value, Fora closod system, the average value of X'is baer ot = = EXC. co) because now all g accessible states are equally likely, with PC) = Vg. The ‘average in (4) ig an elementary example of what aay be called an ensenbte average: ke imagine g similar systems, one in each accessible quantum state uch a group of systems constructed alike is called an ensemble of systems. The average of any property over the group is called the ensemble average of that propery. = ve0'$ ‘Aa mile of spstems I composed of many systems, all constructed alike. Each system in the ensemble is replica of the actual system in one of the santum states accessible 10 the system If there are g accessible states, then {here will be g systems inthe ensemble, one system for each state. Each ssn in the ensembic is equivalent for all practical purposes to the actwal system, Each system satisfies all external requirements placed on the original system ‘and in this sense is “just as good” as the actual system, Every quantum state aChapter 2s Eneopy ond Temperate At ft Vigwe 22 Th of 10 spins ‘The shitty gt ‘rer in which the various systems in the embe ee lise aso sgniiance Accessible fo the acta syster is represented in the weensemble by one system in a Stationary quantum st as in Figure 22. We assume that the ensemble ‘epresens the rea system —thissimplied inthe fundamental assumption, Example: Consracion of n ensenSle. We consruct in Figuse 23 an ensemble to ‘sheen closed sytem of fie spins. ach system wth spin encess 2 = I, The eeray of chin mugneic itd is ~mB. (Do not coatoe the use of in spin exces eaten reswen us fata tate ins or abel) Each ple repesets en ofthe muipls of \ Mose Probable Cofguration Figwe23 Theensemble repesetsa system wth N = S spins and sin exces 2s « | 2 opal stuanon ‘uso states a thisenetgs, The number such tess given bythe multi function cs a= 9%, m0, SN = yyy 0 The 1Osystsms shown in Fgute 23 wake wp the cnsemb ahs ener nthe magnetic fal wee sul hate \hcensebke ts in Figure 24 romp heblaticn gS eauahand the new ensembe oat reeset 2 ysl of wih $ systems ave 25 35 10 systems have 2e= 1 10 sjstcas have fave 26 “3; and Lyeona has 2s 3 hale Most Probable Configuration et two systems $1 and $3 be brought into contact so that energy can be wansfereed freely (fom one tothe othce. This is called thermal contact (Figure 2) The two systems in contact form a larger closed system $= 3, + 3 with constant energy U = U, + Uy. What determines whether there will bea ‘net flow ofencray from onesystem to another? The answer leads to the concept of temperature. The direction of energy fo is not simply a matter of whether the encray of one system is greater than the energy of the other, because theas Chaper 23 Entropy and Temperaare Two dosed [Uy tem aot |], incomact |]. gy Tnslation The systems sre in Therast contact lesulstiag The#mal conductor allows ‘vchangs of energy Fine? Estbtsbment of her contat betwcin to systems 8, ant ‘ystems can be diferent in siz and constitution. A constant total energy can be shared in many ways between two systems, The most probable division of the tal en system has the maxiniin aumabs accessible states af two models ergy sthat for Which the combined eof ccesibe states. We shall enumerate the ystems an then siady what characteries the Systema when in dermal contact, We fist sole in detail the prablein of thermal contact between vo spinsystes, land 2,ina magnetic ld which is introduced in order to define the ener. The numbers of pins, Np may bs diferent, and the values of the spin excess 2,, 25, may be diferent forthe two systems, All spins have magnetic moment m, The aemateschange of energy might take place ‘ia Some weak (residual) coupling between the spins near the interfce between the two systems, We assume that the quaniutn states of the total system $ ean be represented accurately by a combination of any tate of 8, with any state of ‘Sr We keep iY, Ny constant, but the valves ofthe spin excess are allowed to change. The spin excess of sate ofthe combinsd system wll be denoted by 25, wheres = 5, + sy. The eneigy of theeombined system is directly proportional to the total spin excess \ U0) = Usted + Us) = ~ Bs, + 83 Bs, 6 ‘The total number of pacicles is N= Ny 4 Ny ‘Most Probable Configuration ‘Wenssume that the energy splittings between adjuceit energy levels are equal 0 2m in both systems, so that the magnetic energy given up by system 1 when one spin is reverse can be taken up by the reversil af one spin ofsysiem 2 in the opposite sense. Any large physical system will have enough diverse modes of nergy storage so that encigy exchange with another system is always possible, ‘The value of s = 5, 4+ 5, is constant because the total energy is constae. but when the two systems are brought into thermal contact a redistribution is permitted in the separate values of s,, 5, and thus inthe energies Uy, Us. ‘The multiplicity function g(W3} of the combined system $ is related 19 the product of the multiplicity functions of the individual systems 4 and Sy by the relation ats) = DadM sills — 50. 6) ‘where the multiplicity functions 9,, 9, are given by expressions of the form of (1.13), The range of 5 in the summation is from ~4N, 10 40. iC Ny < Ny ‘To soe how (6) comes about, consider fest that config system for which the Fics system has spin excess 2s, and the socond system has spin excess 255. configuration is defined asthe st ofall states with specified values of and 52. The fist system has 9 (V5) accessible states, each of which tnay occur together with any of the gy(N gy) accessible states of the second sslem, Ths total number of states in one configuration of the combined system is given by the product g\(N,,5,)9,(N,s1) of the multiplicity functions of 8, and S,, Because s, = s ~ s, the product of the g's may be written as bf the combined gala slgtNas = 5p. o This product forms one term of the sum (6) Different configurations of the combined system are characterized by different ‘aloes of. Wesum over alt possible values of, to obtain the total number of ates ofall the configurations with Bxed ¢ or fred energy. We thus absain (6), Where g{N.s) is the number of accessible states of the combined sysein in the sum we hold 3, Ny, and 1, constant, as part of the specification of sheseyal The result (6) isa sum of products ofthe form (7) Suc a product will be a ‘maximum for some value of, say 8), 10 be read as “sy hat” of “3, caret” ‘The configuration for which gq, 1 a maximum it ealled the most probable ‘configuration; the mumber of states in itis 9lVas)9IN 3 ~ 54) @ eagle f? k e hove ob a