Physics 224 – Winter 2009 F. H. Shu Lectures 2 and 3 Reaction Equilibria and Einstein Coefficients Law of Mass Action The law of mass action is fundamental to any reaction (chemical or nuclear) that occur in LTE. Consider a reaction where integer numbers, say, 3 and 2, of particles A and B combine to form integer numbers, say, 2 and 1, of particles C and D: 3 A + 2 B = 2C + D. In thermodynamic equilibrium, the forward and backward reactions are in balance, which is why we write the reaction with an equal sign rather than with an arrow. A more general notation moves everything to the left-hand side and reads !# a " a = 0, a where ! a is the particle species of type a, and ! a is an integer, positive or negative depending on whether we think of the species as a reactant or as a product. In the previous example, if we label A, B, C, D as types 1, 2, 3, 4, respectively, then " 1 = 3, " 2 = 2, " 3 = !2, " 4 = !1. Notice, however, that a flip in sign of what we consider as reactants and products does not affect the final law of mass-action, since the above equation can be multiplied by minus one without changing its validity. In equilibrium, the entropy S of a closed system attains a maximum value if the system is kept at constant volume V and energy E. In keeping with the usual conventions of thermodynamics (see, e.g., the book by Callen), we have changed the notation here so that E is the total energy, not the energy per unit volume as in Lecture 6. We consider the closed system to have a fixed total mass. When dealing with LTE rather than TE, the closest analog to what we are doing here is therefore the energy per unit mass that we denoted in Lecture 6 as E . In any case, it turns out that the maximum entropy principle implies the minimum energy principle: the energy E of a closed system attains a minimum value if the syestem is kept at constant volume V and entropy S. The proof proceeds as follows. Suppose the energy E were not a minimum. Then it would be possible to extract energy from the system at constant entropy S by doing PdV work. The work can be converted to heat TdS and added back to the system to bring it to the original level of energy E and then allowing the system adiabatically to expand to its original volume V. The system now has a higher level of entropy S for the same E and V, which contradicts the assumption that the original system had a maximum value of the entropy S. Thus, it must not be possible to lower the energy E for given S and V by doing PdV work, i.e., the original system had the minimum possible value of E. In many experimental and natural situations, what are kept constant are not the entropy S and volume V, but the pressure P and temperature T (for example, a small volume inside a star in mechanical and local thermal equilibrium with its surroundings). In such a case, the minimum energy principle applies not to the internal energy E but to the Gibbs energy G, defined by the Legendre transformation: G = E ! TS + PV . The internal energy E satisfies the fundamental equation of thermodynamics, dE = TdS ! PdV + µdN , where the last term accounts for the change in E if the number N of particles changes, with the coefficient µ , called the “chemical potential,” being defined as an intensive variable complementary to the extensive variable N by analogy to the corresponding intensive coefficients T and P that are complementary to the extensive S and V. For homogeneous single-component systems, the particle number N does not change, so the above equation reduces to the simpler form used in our prior discussion. The third term µdN acquires importance only when we have reactions in multi-component systems. In any case, when it is T and P which are kept constant rather than S and V, what is minimized at equilibrium is not E, but G. The demonstration of the minimum G principle is analogous to the proof the minimum E principle, so we forego it here. The important point for us is that taking the differential of the expression for G and substitution of the expression for dE yields dG = ! SdT + VdP + µdN , i.e., the natural variables for G are T, P, and N, which is the original motivation for the Legendre transformation defining G. When the system consists of multiple components that we donote by a subscript a, with partial entropies and pressures that we denote as S a and Pa , and with individual chemical potentials per particle and particle numbers that we denote by µ a and N a , we may write the above relationship for G as ' $ ' $ dG = (% ! S a "dT + Vd % ! Pa " + ! µ a dN a . & a # & a # a At fixed T and P " ! Pa , chemical (or nuclear) equilibrium requires G to be at a a minimum, dG = ! µ a dN a = 0; a i.e., G is to be minimized subject to variations of the individual particle numbers that may result when chemical reactions produce some particles types at the expense of other particle types. The individual variations dN a are not independent because the appearance or disappearance of individual particles must satisfy Dalton’s integer proportions as defined by the original reaction equation; i.e., dN1 : dN 2 : dN 3 : ... = ! 1 : ! 2 : ! 3 : ... Another way to express the above is dN a = " a d! where d! is a common scale factor for each a and is the only independent variable of the problem. Substitution of the above expression into the requirement for the minimization of G for arbitrary d! now yields the result that at chemical equilibrium, !" a µ a = 0, a which is the law of mass action. The law of mass action states that at chemical equilibrium, the chemical potentials of the individual species, when multiplied by its Dalton integer in the fundamental reaction equation, must sum to zero. Commit this result to memory. The chemical potential µ is generally a function of the number density of the species and the ambient temperature T. The law of mass action therefore provides a constraint on the relative number densities (or concentrations) of the reactants and products of a reaction in equilibirum at temperature T. The calculation of the chemical potentials of different species as a function of number density and temperature is a relatively simple task in the gaseous state when we deal with well-separated molecules, atoms, or nuclei with welldefined internal and external degrees of freedom. Relativity makes the principle even simpler since the energy of a particle is 2 ! = mc , where m is the particle mass, whether the particle is an elementary particle or a compound particle. (Do not confuse our use here of ! and m with the nuclear energy generation rate per unit mass and the mean particle mass in a cosmic mixture – there are only so many letters in the Latin and Greek alphabets.) In other words, the binding energy between 2 hydrogen atoms plus an oxygen atom and a water molecule is simply c2 times the mass difference between 2H+O and H2O. The chemical binding energy simply represents the “mass deficit” of the water molecule in comparison with its separated atoms moved to infinity. We usually think in these terms only for nuclear energy, but, of course, relativity theory is universal. If one includes general relativity, what one calls the binding energy of the Earth to the Sun, which is the sum of the gravitational potential energy and the kinetic energy in Newton’s way of thinking, is in Einstein;s universe just c2 times the mass difference of the Earth and Sun imagined infinitely separated compared to the mass that an observer at infinity would ascribe to the actual Earth-Sun system. In relativity, there is no such thing as gravitational potential energy, kinetic energy, chemical binding energy, etc., only ! = mc 2 . The reader my be surprised by the claim that there is no such thing as kinetic energy, since the conventions of modern physics are to define kinetic energy as the difference between the actual total energy of a particle and its rest energy. This is okay for truly elementary particles, but it runs into conceptual problems for compound particles. Consider the proton. Does it have a rest mass? All textbooks say so, but a proton is made of moving quarks. What are the relative contributions to the “rest mass” of the proton of the “kinetic energies” of the moving quarks, their rest masses, and their QCD interactions? It becomes frightfully complex quickly, doesn’t it? Yet, Einstein solves the practical problem for us in a clean stroke. What is the mass of the proton? Measure its gravitational mass. And if it’s moving or in an excited internal state, it will have a higher gravitational mass than if it is stationary. Such is the power and liberating effect of Einstein’s formula, ! = mc 2 . Relativistic Approach to Notion of Chemical Potential The occupation number of fermions (upper sign) or bosons (lower sign) with energy ! and chemical potential µ (not to be confused with the reduced mass of the reactants) at temperature T is !(" ) = 1 e ( " # µ )/ kT ±1 . Notice that ! is a function of an individual particle whereas µ is a function of the distribution of particles; i.e., the latter is a function of the density and temperature while the former is not. In the interiors of normal stars, the densities are not high enough that we need to worry about the ± 1, i.e., the distinction between fermions and bosons; nor are the temperatures high enough that we need worry about relativistic effects. Thus, the relationship among particle energy, momentum, and mass is given by the usual nonrelativistic formula, except for an additive rest energy: ! = m0 c 2 + p2 . 2m0 The chemical potential µ is now obtained by requiring that we get the number density n when we make an appropriate integral of the occupation number: & n = gh !3 ' "(# )4$ p 2 dp = g%T !3e( µ ! m 0 c 0 2 )/ kT , where !T is the thermal de Broglie wavelength associated with a particle of momentum (2!m0 kT )1 / 2 : "T # h , (2!m0 kT )1 / 2 and g is the number of equivalent internal states with the same rest mass m0 . (For example, if we are talking about the ground state of a free proton, g would equal 2 for the two spin states of an s = 1/2 particle. For more complex nuclei, we need more information on nuclear shell models than we have learned in this book in order to calculate the degeneracy factor g, a detail that we shall forego since the g’s are usual factors of only a few.) Thus, the chemical potential of a gas of particles with rest mass m0 at density n and temperature T is given by [ 3 ] µ =m 0 c 2 + kT ln n!T / g . Boltzmann’s Law We first use these formulae to derive Boltzmann’s law (which is the analog of what Goldstein said in another context akin to “using a sledge hammer to crack a peanut.”) Consider the collisional excitation and de-excitation equilibrium: (i,1) + e ! " (i, n) + e ! . The relativistically-correct rest-energy of an ion in state i and excited electronic level n with energy En (i) above the ground state n = 1 is m0n (i)c 2 = m10 (i)c 2 + En (i) ! E1 (i). Because the free electron appears on both left and right of the excitation/de-excitation equilibrium, it does not enter into the law of mass action (in other words, it doesn’t matter what is colliding with ion i, as long as it has a thermal distribution; it could be blackbody photons instead, for example): &g n( 3# µ1 ' µ n = 0 = '( E n ' E1 ) + kT ln$$ n 1 T 13 !!, % g1 nn (Tn " which holds for either i or i+1. Because the electronic energies are much smaller than 3 3 rest energies of the ion, we can set !T 1 / !Tn = 1, and derive nn g n !( En ! E1 ) / kT , = e n1 g1 which is Boltzmann’s law. If we sum over all n, we can also express the above as nn (i ) = ni g n e ![ En (i ) ! E (i )1 ] / kT Zi " where Z i $ ! g n e #[ En (i ) # E1 (i )] / kT , n =1 nn (i + 1) = ni +1 g n e ![ En (i +1) ! E (i +1)] / kT Z i +1 " where Z i +1 $ ! g n e #[ En (i +1) # E1 (i +1)] / kT , n =1 where ni and ni+1 are the densities of the total number of ions in stage i and i+1. Einstein Coefficients for Radiative Transitions Between Discrete Levels Consider now the condition of detailed balance when we have atoms or molecules in thermodynamic equilibrium with a blackbody radiation field B! (T ) : nu [Aul + Bul B! (T )]= nl Blu B! (T ). In this notation, Boltzmann’s law, with " u # " l = h! , reads nu g u " h! / kT = e . nl gl Division by nl in the equation of detailed balance and the substitution of the formula for the Planck function B! (T ) yield, after multiplication through by the terms containing exponentials: ) & 2h. 3 # & 2 h. 3 # gu , - h. / kT - h. / kT + $$ 2 !! Bul e * Aul 1 - e ' = $$ 2 !! Blu . gl + % c " ( % c " ( ) Being atomic or molecular constants, Aul , Bul , and Blu cannot depend on T; thus, the terms with and without the exponential factor e " h! / kT must be independently set equal to each other. This requirement secures the identifications: Aul = g 2h' 3 Bul = l 2 gu c & 2h' 3 # $$ 2 !! Blu . % c " The above relations were established by Einstein in 1916. Note especially the relationship Bul / Blu = gl / gu that we cited without proof in Lecture 5 when we corrected true absorption for stimulated emission. The genius in Einstein’s derivation reproduced above lay in anticipating that the form of Planck’s law, with eh! / kT " 1 in the denominator of the photon occupation number, instead of the classical inverse Boltzmann factor by itself eh! / kT , implies that the radiiative emission and absorption processes contains something extra compared to superelastic and inelastic collisions between classical material particles. The extra !1 in the denominator must be associated with a process of stimulated emission that exists as an extra process on top of spontaneous emission and true absorption. Moreover, whatever were the quantum laws governing radiative transitions, they have to be consistent with the two equations above relating the three coefficients Aul, Bul and Blu . If one knows one coefficient, one would know all three. It would take another decade before the quantum mechanics of Heisenberg and Schrödinger for the quantized levels of atoms, and the time-dependent perturbation theory of Dirac for radiative transitions between different discrete levels would verify by direct computation the Einstein relations. Moreover, of the two methods, Einstein’s argument from the principle of detailed balance is the more general since it holds to all perturbation orders. We warn the reader that the relationship between the Einstein coefficients hold in the form s written above only when they are defined in terms of I! . If the absorbed and stimulated rates are written using instead I ! , then a factor of c /" 2 = !2 / c needs to be introduced in the relationship between the corresponding Einstein B’s to compensate for the difference between I! and I ! . We now derive the relationship between Einstein’s coefficient for true absorption and the radiative cross-section that an atom or molecule has for absorbing a photon of the appropriate frequency. In terms of the angle-independent cross-sections that we introduced in Lecture 5, the rate at which true absorption is taking place per unit time per unit volume equals !! d#! " 0 I$ nl % abs ($ )d$ , h$ (7.8) where the division by the energy of the photon h! is necessary to convert I! from a monochromatic specific intensity for energy into one for photon number, and where the integration over d! accounts for photons of the appropriate frequency propagating in all directions past the atoms. We suppose now that " abs (! ) has the form given by the classical model for electronic transitions in atoms (see Chapter 14 of Shu, Radiation), except that we include a correction factor, called the f-value, f lu , which corrects for quantum effects: # abs (! ) = "e 2 f lu L(! ), me c where L(! ) is a sharply peaked function about " = " ul $ (! u # ! l )/ h that integrates to unity when we perform the indicated integration over d! . Often L(! ) is taken to be the Lorentzian profile (see eq. 23.18 of Shu, Radiation): % ( " ul / 4# L(! ) = 4 ' ,, 2 2 * & (! $ ! ul ) + (" ul / 4# ) ) where ! ul / 4" is called the width of the transition and is given by quantum calculations for isolated atoms as the sum of the Einstein A’s out of the levels u and l that, by the uncertainty principle, make the levels “fuzzy.” ! ul = " Aun + " Aln . n <u n <l In general, the lifetimes of atoms against spontaneous radiative decay, even for the strongest processes, are long compared to the inverse frequency of the radiation that they emit, i.e., Aul ! ! ul , so that the typical atom emits photons of many wave periods before the energy difference of the upper and lower level is carried away, yielding a line profile that is well-defined (i.e., very narrow) about the line frequency ! ul . When the Lorentzian is a sharply peaked function of ! , L(! ) behaves essentially as a Diract delta function, and the absorption rate (7.8) becomes " 4! e2 % " flu % nl $ ** I(ul d). # me c '& $# h( lu '& ! The corresponding rate written in Einstein’s notation is nl Blu 1 4! ! $$ I " ul d# , where we have introduced a factor of 1 / 4! because historically the volumetric rate associated with Einstein’s B coefficient was defined in terms of the mean specific intensity. Cancelling out the common factors, we obtain upon equating the two expressions " 4! 2 e2 % ful 1 Blu = $ = ' # me c & h( ul h( ul +) abs (( )d( . (7.9) *( The first identification for Blu holds only for electronic transitions in atoms or ions; the second identification, where the integration occurs over the line-width holds more generally, for example, for rotational-vibrational transitions in molecules as well as their electronic transitions, as long as ! abs (" ) is an accurately measured or calculated crosssection for true absorption. From Einstein’s general relations, equation (7.9) yields an expression for the rate coefficient of spontaneous emission of electronic transitions in atoms or ions: g Aul = l gu & 8( 2 e 2' ul 2 $ $ m c3 e % # ! f ul . ! " The total power radiated on average by an atom undergoing spontaneous emission is the Einstein A-value times the energy of the emitted photon h" ul = !! ul : 2 g 2e 2! ul P= l f ul !! ul . g u me c 3 We compare this expression with Larmor’s formula for the radiative power of an electric dipole freely oscillating at natural radian frequency ! 0 (see eq. 14.13 of Shu, Radiation): P= 2 e 2 !x!2 3c 3 2e 2! 0 = 3c 3 4 x2 Except for a dimensionless factor ( g l / g u )3 f ul , if we identify ! 0 with ! ul , we see that there is a correspondence between the quantum and classical expressions if we identify the expectation value for the oscillator energy (half in kinetic and half in potential energy), when first excited, to be totally radiated away by the atom or ion as a photon of energy !! ul : 2 me! 0 x 2 ~ !! ul . This correspondence suggests correctly that for normal (strong resonance) lines, the quantum-mechanical factor ( g l / g u )3 f ul is of order unity. However, in interstellarmedium astrophysics, one often deals with much weaker lines, particularly so-called “forbidden lines,” where the f-values are much less than unity. It is conventional to express hydrogen-atom f-values for transitions from a lower level ! = n! to an upper level u = n in terms of semi-classical values computed by Kramers times a quantum-mechanical correction g nn! called the Gaunt factor: f nn! = 2 5 ' n! 3 n $ g . % 2 2 3 " nn! 3) 3 & (n ( n! ) # Carrying out the indicated multiplications, together with Rydberg’s formula for hydrogen-line frequencies, we obtain Ann! = 1.57 # 1010 [n 3 n!(n 2 " n! 2 )]"1 g nn! s -1. For atomic hydrogen, the Gaunt factors are of order unity for large n and n! since quantum mechanics becomes classical mechanics in the limit of large quantum numbers. Even for the transition 1 ! 2 , we have g 21 = 0.717, which is not very different from unity. Numerical values for some common transitions are Lyman alpha: A21 = 4.68 " 108 s !1 ; Lyman beta: A31 = 5.54 " 10 7 s !1 ; Balmer alpha (H ! ) : A32 = 4.39 " 10 7 s !1 ; Balmer beta (H ! ) : A42 = 8.37 " 10 6 s !1 . In fact, the Einstein A values can be computed analytically for all the transitions of atomic hydrogen (see Prob. Set 5 of Shu, Radiation), so an approach via f-values is not needed for such a simple quantum system. It is much more useful in application to complex atoms and ions whose spectra are measured in the laboratory. Application to Ionization Equilibrium – Saha’s Equation Let us now consider the recombination-ionization equilibrium out of the ground electronic states of i and i+1: (i + 1,1) + e " ! (i,1). The law of mass action applied to this equilibrium reaction reads & g1 (i )ne (Te 3 n1 (i + 1) # µ1 (i + 1) + µ e ' µ1 (i ) = 0 = I i + kT ln $ !, %$ g e g1 (i + 1)n1 (i ) "! where ge = 2 for the two spin states of the free electron, and where we have taken the 3 3 liberty of setting !T 1 (i + 1) / !T 1 (i ) = 1 on the approximation that the electron rest energy is also small in comparison with the ion rest energies. Thus, we have the equation ' n1 (i + 1) $ ' g1 (i ) $ !3 ! I i ; kT . % "% " ne = g e (Te e & g1 (i + 1) # & n1 (i ) # (7.10) Applying Boltzmann’s law to n = 1 (the principle of detailed balance permits us to use any n or any sum of them), we may express the ground-state fractions as n1 (i + 1) ni +1 = , g1 (i + 1) Z i +1 g1 (i ) Z i = , ni (i ) ni which now allows us to write equation (7.8) in the usual form stated for Saha’s equation: ni +1 ne Z i +1 g e !3 = "Te e ! I i / kT . ni Zi (7.11) It is usual to adopt the convention of defining the ground electronic-energy levels as zero, E1 (i ) ! 0 and E1 (i + 1) = 0, in the electronic partition functions. Equation (7.11) has the following breakdown. The combination ni +1 ne / ni on the left-hand side is what we expect – product of densities of reactants over product of densities of products – from the so-called “law of mass action” that we learn from chemistry for an equilibrium reaction of dilute gases of the form given by equation (i + 1) + e " ! i . The right-hand side of equation (7.11) is a reaction “constant” that depends only on the temperature T. The term e ! I i / kT is the usual Boltzmann factor that favors the less ionized state i in comparison with the more-ionized state i+1 because of the need to supply ionization energy Ii to convert the former into the latter. The partition functions Zi and Zi+1 are simply the generalization of the usual degeneracy factors to account for the number of ionic states that have the same energies, but now apportioned according to !3 their Boltzmann distributions across excited electronic states. The factor g e "Te represents the analogous quantity for the electrons, accounting for both its internal (spin only) degrees of freedom as well as its translational degrees of freedom due to its thermal motions. The left- and right-hand sides must have the same dimension of inverse !3 volume, which is a second justification for the extra factor "Te on the right-hand side of equation (7.11). The thermal de Broglie wavelength !Te characterizes the minimum scale over which thermal electrons will not experience the quantum presence of other electrons because of the Pauli exclusion principle. In other words, in order for the derivation that 3 led to equation (7.11) to be valid, we require ne !Te << 1, so that the electrons are not degenerate in physical space when they have typical momenta ~ (2!me kT )1 / 2 . The 3 inequality ne !Te << 1 is well satisfied for the relatively low densities that prevail in stellar atmospheres. Assuming that the partition functions in equation (7,11) yield only factors of order unity (see below), we see that ionization to the next stage from i becomes relatively important when e " I i / kT ne !Te 3 ~ 1 , i.e., when T ~ Ii / k 3 " ln(ne !Te ) . (7.12) Let us apply this equation to the partial ionization of hydrogen, where I i / k = 158,000 K. 3 In a typical stellar atmosphere, ln(ne "Te ) ~ !15 ; thus, the partial ionization of hydrogen is strongest when T ~ 11,000 K, which is the effective temperature at the top of the range for A stars (Lecture 3). Although we naively expect the partial ionization of hydrogen to occur only when the temperature approaches 158,000 K because of the binding-energy advantage of the neutral atom in its ground electronic configuration compared to the same in the ion, the actual temperature required to produce appreciable ionization in a stellar photosphere is more than an order of magnitude lower than the naive expectation because of the advantage gained statistically by an electron if it is freed to explore phase space. In some sense, it is easier to ionize an atom than it is to excite it. A tension exists between a preference for greater binding energy at low temperatures and a preference for more freedom to explore phase space at high temperatures. Einstein-Mile Coefficients for Continuum Processes While the derivation of Saha’s equation is fresh in our memory, we use the result to derive the so-called Einstein-Milne coefficients, A(n) and B(n), associated with spontaneous and induced recombination by the radiative emission of the energy difference between an electron in the continuum and a bound state n of the hydrogen atom. The relevant equation of detailed balance reads [ ] n(H[n]) p! (n) B! (T ) = [vf e ( v)]4"v 2 dv ne np [$(n) + #(n) B! (T )]. In the above, p! (n) is related to the photo-ionization cross-section " ! (n) of a hydrogen atom in level n, which we denote with the symbol H[n], for a photon of energy h! , by the equation p! (n) = 4# " ! ( n) . h! The function f e ( v) is the Maxwell-Boltzmann distribution at temperature T of random velocities v of electrons e relative to protons p, whose number densities are, respectively, ne and np, which are recombining into atomic hydrogen in level n, with number density n(H[n]), spontaneously and by induced radiation with Einstein-Milne rate-coefficients A and B, respectively. By energy conservation, we require h" = ! n + me v 2 / 2, where ! n is the ionization potential from level H[n], and where we have not bothered to distinguish between the mass of the electron me and its reduced value in the electronproton system. Note that the last equation requires the following relationship between differentials: hd! = me vdv. If we substitute into the equation of detailed balance, the appropriate expressions for the Maxwell-Boltzmann distribution of the electron, the Planck function, and the Saha equation for the electron-proton-H[n] equilibrium, we get after some cancellation of common factors: 2 & # , 2 h0 3 ) g e g p , m e ) , 2h0 3 ) p0 ** 2 '' = * ' 41v 2 $ /(n) 1 . e . h0 / kT + -(n)** 2 ''e . h0 / kT ! . gn + h ( + c ( + c ( % " ( ) Since A, B,and p! are atomic constants independent of T, we require "( n) = 2h# 3 !(n), c2 - gnh .( n ) = + +g g m , e p e *& # 10 ($ , with g n / g e g p = n 2 . (7.13) ( 0 (h0 ' / ) ! n " )% As usual, we are able to express any elementary rate coefficient for a radiative process in terms of the absorption cross-section for the process, Apparent Divergence of Partition Functions Naively, one might think that the partition functions Z i and Z i +1 containing an infinite number of terms might not sum to an order-unity quantity. Consider, for example, the case when i is the neutral hydrogen atom. In this case, the degeneracy factor g n = 2n 2 and the energy level E n = I i (1 ! n !2 ); thus, " " n =1 n =1 Z i = ! g n e # En / kT = 2e # I i / kT ! n 2 e I i / n kT , 2 which is a divergent expression since even the sum of n2 from 1 to infinity diverges, and we have inside the sum something larger than n2. The formal difficulty is that the number of bound Rydberg states increases dramatically as n " ! , so at any T, the freedom to explore the large amount of phase space available to a bound electron at large n would seemingly win out over any advantage for more binding energy at small n. If this formal result corresponded to reality, then any container of thermal hydrogen atoms would have most of the atoms occupying very high Rydberg states rather than having a large fraction existing, say, in the ground level n = 1. This is a patently false conclusion, so there must be a problem in allowing a sum over an infinite number of bound states. To see that only a finite number of Rydberg states is a realistic approximation, note that the effective size of the neutral hydrogen atom in level n scales as rn ~ n 2 a 0 where a0 is the Bohr radius ~0. 5 ! 10 "8 cm. Thus, high-level Rydberg atoms are very large, and perturbations from nearby material may truncate the existence of such states. For example, in the solar atmosphere, the neutral hydrogen density might be 1018 cm !3 , so that hydrogen atoms are spaced by 10 !6 cm apart on average. If we set the largest rn at 10 !6 cm, we get a maximum value of n ~14. (There are more sophisticated ways to find nmax, but they do not affect our general conclusion.) If the upper limit ! in the definition of Zi is replaced by 14, then the sum is dominated by the first term for values of T ~ 10 4 K. In this case, Zi is only slightly larger than 2, and most neutral hydrogen atoms are in the ground state n = 1. Those of you who do radio astronomy will know that a famous radiorecombination line in emission from H II regions is H109α, arising in a radiative transition (recombination cascade) from n = 110 to n =109 in the hydrogen atom. A hydrogen atom in n = 110 has a radius rn ~ 6 ! 10 "5 cm, i.e., the interstellar medium contains atoms the size of a large bacterium! Atoms are not always microscopically small because interstellar space is very empty.