Peter Atkins Julio de Paula Ron Friedman Physical Chemistry Quanta (0511-0561)

488 10 Magnetic resonance
of the molecules and the field. In solution, when the molecule

Self-test 50.1 At what magnetic field would the methyl radi-
cal come into resonance in a spectrometer operating at is tumbling rapidly, only the average value of the g-value is
34.000 GHz (radiation belonging to the Q band of the micro- observed. Therefore, anisotropy of the g-value is observed only
wave region)? for radicals trapped in solids.
Answer: 1.213 T
50.2 Hyperfine structure

The g-value is related to the ease with which the applied field
can stir up currents through the molecular framework, and the The most important feature of an EPR spectrum is its hyper-
strength of the magnetic field the currents generate. Therefore, fine structure, the splitting of individual resonance lines into
the g-value gives some information about electronic structure components. In general in spectroscopy, the term ‘hyperfine
and plays a similar role in EPR to that played by shielding con- structure’ means the structure of a spectrum that can be traced
stants in NMR. to interactions of the electrons with nuclei other than as a result
Two factors are responsible for the difference of the g-value of the latter’s point electric charge. The source of the hyperfine
from ge. Electrons migrate through the molecular framework structure in EPR is the magnetic interaction between the elec-
by making use of excited states (Fig. 50.1). This circulation gives tron spin and the magnetic dipole moments of the nuclei pre-
rise to a local magnetic field that adds to the applied field. The sent in the radical which give rise to local magnetic fields.
extent to which these currents are induced is inversely propor-
tional to the separation of energy levels, ΔE, in the radical or
complex. Secondly, the strength of the field experienced by the
(a) The effects of nuclear spin
electron spin as a result of these electronic currents is propor- Consider the effect on the EPR spectrum of a single H nucleus
tional to the spin–orbit coupling constant, ξ (Topic 21). We can located somewhere in a radical. The proton spin is a source of
conclude that the g-value differs from ge by an amount that is magnetic field and, depending on the orientation of the nuclear
proportional to ξ/ΔE. This proportionality is widely observed. spin, the field it generates adds to or subtracts from the applied
Many organic radicals, for which ΔE is large and ξ (for carbon) field. The total local field is therefore
is small, have g-values close to 2.0027, not far removed from
ge itself. Inorganic radicals, which commonly are built from Bloc = B0 + amI mI = ± 12 (50.3)
heavier atoms and therefore have larger spin–orbit coupling
constants, have g-values typically in the range 1.9 to 2.1. The where a is the hyperfine coupling constant. Half the radicals in
g-values of paramagnetic d-metal complexes often differ con- a sample have mI = + 12 , so half resonate when the applied field
siderably from ge, varying from 0 to 6, because in them ΔE is satisfies the condition
small on account of the small splitting of d orbitals brought
h␯ 1
about by interactions with ligands (Topic 45). h␯ = g μB (B0 + 12 a ) , or B0 = − a (50.4a)
The g-value is anisotropic: that is, its magnitude depends on g μB 2
the orientation of the radical with respect to the applied field.
The anisotropy arises from the fact that the extent to which an The other half (which have mI = − 12 ) resonate when
applied field induces currents in the molecule, and therefore the
h␯ 1
h␯ = g μB (B0 − 12 a ) , or
magnitude of the local field, depends on the relative orientation B0 = + a (50.4b)
g μB 2
Therefore, instead of a single line, the spectrum shows two lines

of half the original intensity separated by a and centred on the
field determined by g (Fig. 50.2).
If the radical contains an 14N atom (I = 1), its EPR spectrum
consists of three lines of equal intensity, because the 14N nucleus
has three possible spin orientations, and each spin orientation
is possessed by one-third of all the radicals in the sample. In
general, a spin-I nucleus splits the spectrum into 2I +1 hyper-
fine lines of equal intensity.
Figure 50.1 An applied magnetic field can induce circulation When there are several magnetic nuclei present in the radi-
of electrons that makes use of excited-state orbitals (shown cal, each one contributes to the hyperfine structure. In the case
with a white line). of equivalent protons (for example, the two CH2 protons in the
www.ebook3000.com
Atkins09819.indb 488 9/11/2013 12:37:07 PM

50 Electron paramagnetic resonance 489
No hyperfine splitting αN
Example 50.1 Predicting the hyperfine structure
Hyperfine splitting βN
due to one proton of an EPR spectrum
A radical contains one 14N nucleus (I = 1) with hyperfine con-
stant 1.61 mT and two equivalent protons (I = 12 ) with hyper-
hν fine constant 0.35 mT. Predict the form of the EPR spectrum.
α
β
hν Method Consider the hyperfine structure that arises from
each type of nucleus or group of equivalent nuclei in succes-
sion. So, split a line with one nucleus, then each of those lines
βN
is split by a second nucleus (or group of nuclei), and so on.
It is best to start with the nucleus with the largest hyperfine
αN
splitting; however, any choice could be made, and the order
Figure 50.2 The hyperfine interaction between an electron in which nuclei are considered does not affect the conclusion.
and a spin- 21 nucleus results in four energy levels in place of the Answer The 14 N nucleus gives three hyperfine lines of equal
original two. As a result, the spectrum consists of two lines (of intensity separated by 1.61 mT. Each line is split into doublets
equal intensity) instead of one. The intensity distribution can of spacing 0.35 mT by the first proton, and each line of these
be summarized by a simple stick diagram. The diagonal lines doublets is split into doublets with the same 0.35 mT splitting
show the energies of the states as the applied field is increased, (Fig. 50.4). The central lines of each split doublet coincide, so
and resonance occurs when the separation of states matches the proton splitting gives 1:2:1 triplets of internal splitting
the fixed energy of the microwave photon. 0.35 mT. Therefore, the spectrum consists of three equivalent
1:2:1 triplets.
radical CH3CH2) some of the hyperfine lines are coincident. If
the radical contains N equivalent protons, then there are N + 1
hyperfine lines with an intensity distribution given by Pascal’s
1.61 mT
triangle (Topic 48, reproduced here as 1). The spectrum of the 0.35 mT
benzene radical anion in Fig. 50.3, which has seven lines with
intensity ratio 1:6:15:20:15:6:1, is consistent with a radical con-
taining six equivalent protons. More generally, if the radical
contains N equivalent nuclei with spin quantum number I, then
there are 2NI + 1 hyperfine lines with an intensity distribution
based on a modified version of Pascal’s triangle, as shown in
Example 50.1. 1:2:1 1:2:1
1 Figure 50.4 The analysis of the hyperfine structure

1 1
1 2 1 of radicals containing one 14N nucleus (I = 1) and two
1 3 3 1
1 4 6 4 1
equivalent protons.
1 5 10 10 5 1
Self-test 50.2 Predict the form of the EPR spectrum of a radi-
1 cal containing three equivalent 14N nuclei.
Answer: See Fig. 50.5
Field strength
1 3 6 7 6 3 1
Figure 50.3 The EPR spectrum of the benzene radical anion,
C6H6− , in fluid solution, with a the hyperfine splitting of the Figure 50.5 The analysis of the hyperfine structure of radicals
spectrum. The centre of the spectrum is determined by the containing three equivalent 14N nuclei.
g-value of the radical.
Atkins09819.indb 489 9/11/2013 12:37:11 PM

(b) The McConnell equation (c) The origin of the hyperfine interaction
The hyperfine structure of an EPR spectrum is a kind of fin- The hyperfine interaction is an interaction between the mag-
gerprint that helps to identify the radicals present in a sample. netic moments of the unpaired electron and the nuclei. There
Moreover, because the magnitude of the splitting depends on are two contributions to the interaction.
the distribution of the unpaired electron in the vicinity of the An electron in a p orbital centred on a nucleus does not
magnetic nuclei, the spectrum can be used to map the molecu- approach the nucleus very closely, so it experiences a field that
lar orbital occupied by the unpaired electron. For example, appears to arise from a point magnetic dipole. The resulting
because the hyperfine splitting in C 6 H6− is 0.375 mT, and one interaction is called the dipole–dipole interaction. The contri-
proton is close to a C atom that has one-sixth the unpaired bution of a magnetic nucleus to the local field experienced by
electron spin density (because the electron is spread uniformly the unpaired electron is given by an expression like that in eqn
around the ring), the hyperfine splitting caused by a proton in 48.10 (a dependence proportional to (1 – 3 cos2θ)/r3). A char-
the electron spin entirely confined to a single adjacent C atom acteristic of this type of interaction is that it is anisotropic and
should be 6 × 0.375 mT = 2.25 mT. If in another aromatic radical averages to zero when the radical is free to tumble. Therefore,
we find a hyperfine splitting constant a, then the spin density, hyperfine structure due to the dipole–dipole interaction is
ρ, the probability that an unpaired electron is on the atom, can observed only for radicals trapped in solids.
be calculated from the McConnell equation: An s electron is spherically distributed around a nucleus and
so has zero average dipole–dipole interaction with the nucleus
a = Qρ McConnell equation (50.5) even in a solid sample. However, because it has a nonzero prob-
ability of being at the nucleus (Topic 17), it is incorrect to treat
with Q = 2.25 mT. In this equation, ρ is the spin density on a C the interaction as one between two point dipoles. As explained
atom and a is the hyperfine splitting observed for the H atom to in Topic 48, an s electron has a Fermi contact interaction with
which it is attached. the nucleus, a magnetic interaction that occurs when the point
dipole approximation fails. The contact interaction is isotropic
(that is, independent of the radical’s orientation), and conse-
Brief illustration 50.2 The McConnell equation quently is shown even by rapidly tumbling molecules in fluids
The hyperfine structure of the EPR spectrum of the radical anion (provided the spin density has some s character).
C10H8− , the naphthalene radical anion, can be interpreted as aris- The dipole–dipole interactions of p electrons and the Fermi
ing from two groups of four equivalent protons. Those at the 1, 4, contact interaction of s electrons can be quite large. For exam-
5, and 8 positions in the ring have a = 0.490 mT and those in the ple, a 2p electron in a nitrogen atom experiences an average
2, 3, 6, and 7 positions have a = 0.183 mT. The densities obtained field of about 3.4 mT from the 14N nucleus. A 1s electron in a
by using the McConnell equation are, respectively (2), hydrogen atom experiences a field of about 50 mT as a result
of its Fermi contact interaction with the central proton. More
a
0.490 mT 0.183 mT values are listed in Table 50.1. The magnitudes of the contact
ρ= = 0.218 and ρ = = 0.0813 interactions in radicals can be interpreted in terms of the s
2.25 mT 2.25 mT

orbital character of the molecular orbital occupied by the
Q
unpaired electron, and the dipole–dipole interaction can be
0.22 – interpreted in terms of the p character. The analysis of hyper-
0.08 fine structure therefore gives information about the compo-
sition of the orbital, and especially the hybridization of the
2 atomic orbitals.
− the anthracene radi-
Self-test 50.3 The spin density in C14 H10 ,
cal anion, is shown in 3. Predict the form of its EPR spectrum.
0.097 0.193
Table 50.1* Hyperfine coupling constants for atoms, a/mT
–
0.048 Nuclide Isotropic coupling Anisotropic coupling
1H 50.8 (1s)
3
2H 7.8 (1s)
Answer: A 1:2:1 triplet of splitting 0.43 mT split into a 1:4:6:4:1 14N 55.2 (2s) 4.8 (2p)
quintet of splitting 0.22 mT, split into a 1:4:6:4:1 quintet
of splitting 0.11 mT, 3 × 5 × 5 = 75 lines in all 19F 1720 (2s) 108.4 (2p)
*More values are given in the Resource section.
www.ebook3000.com
Atkins09819.indb 490 9/11/2013 12:37:15 PM

50 Electron paramagnetic resonance 491
Brief illustration 50.3The composition of a molecular Hund

Pauli Fermi
orbital from analysis of the hyperfine structure
From Table 50.1, the hyperfine interaction between a 2s elec- C H
tron and the nucleus of a nitrogen atom is 55.2 mT. The EPR
spectrum of NO2 shows an isotropic hyperfine interaction of
Low energy High energy
5.7 mT. The s character of the molecular orbital occupied by
the unpaired electron is the ratio 5.7/55.2 = 0.10. For a continu- (a) (b)
ation of this story, see Problem 50.6.
Figure 50.6 The polarization mechanism for the hyperfine
Self-test 50.4 In NO2 the anisotropic part of the hyperfine interaction in π-electron radicals. The arrangement in (a) is
coupling is 1.3 mT. What is the p character of the molecular lower in energy than that in (b), so there is an effective coupling
orbital occupied by the unpaired electron? between the unpaired electron and the proton.
Answer: 0.38
between a proton and the α electrons (ms = + 12 ) which results

We still have to explain the source of the hyperfine struc- in one of the electrons tending to be found with a greater prob-
ture of the C 6 H6− anion and other aromatic radical anions. The ability nearby (Fig. 50.6). The electron with opposite spin is
sample is fluid, and as the radicals are tumbling the hyper- therefore more likely to be close to the C atom at the other end
fine structure cannot be due to the dipole–dipole interaction. of the bond. The unpaired electron on the C atom has a lower
Moreover, the protons lie in the nodal plane of the π orbital energy if it is parallel to that electron (Hund’s rule favours
occupied by the unpaired electron, so the structure cannot be parallel electrons on atoms), so the unpaired electron can
due to a Fermi contact interaction. The explanation lies in a detect the spin of the proton indirectly. Calculation using this
polarization mechanism similar to the one responsible for model leads to a hyperfine interaction in agreement with the
spin–spin coupling in NMR. There is a magnetic interaction observed value of 2.25 mT.
Checklist of concepts
☐ 1. The EPR resonance condition is written in terms of the lines, with an intensity distribution given by a modified
g-value of the radical. version of Pascal’s triangle.
☐ 2. The value of g depends on the ability of the applied field ☐ 5. Hyperfine structure can be explained by dipole–dipole
to induce local electron currents in the radical. interactions, Fermi contact interactions, and the
☐ 3. The hyperfine structure of an EPR spectrum is its split- polarization mechanism.
ting of individual resonance lines into components ☐ 6. The spin density is the probability that an unpaired
by the magnetic interaction between the electron and electron is on the atom.
nuclei with spin.
☐ 4. If a radical contains N equivalent nuclei with spin
quantum number I, then there are 2NI + 1 hyperfine
Checklist of equations
Property Equation Comment Equation number
EPR resonance condition h␯ = g μBB0 No hyperfine interaction 50.2
h␯ = g μB (B0 ± 12 a) Hyperfine interaction between an electron and a proton 50.4
McConnell equation a = Qρ Q = 2.25 mT 50.5
Atkins09819.indb 491 9/11/2013 12:37:19 PM

Focus 10 on Magnetic resonance
Topic 47 General principles

Discussion questions
47.1 To determine the structures of macromolecules by NMR spectroscopy, 47.2 Compare the effects of magnetic fields on the energies of nuclei and the
chemists use spectrometers that operate at the highest available fields and energies of electrons.
frequencies. Justify this choice.
47.3 What is the Larmor frequency?
Exercises
47.1(a) Given that g is a dimensionless number, what are the units of γN 47.7(a) Calculate the magnetic field needed to satisfy the resonance condition
expressed in tesla and hertz? for unshielded 14N nuclei in a 50.0 MHz radiofrequency field.
47.1(b) Given that g is a dimensionless number, what are the units of γN 47.7(b) Calculate the magnetic field needed to satisfy the resonance condition
expressed in SI base units? for unshielded protons in a 400.0 MHz radiofrequency field.
47.2(a) For a proton, what are the magnitude of the spin angular momentum 47.8(a) Use the information in Table 47.2 to predict the magnetic fields
and its allowed components along the z-axis? What are the possible at which (a) 1H, (b) 2H, (c) 13C come into resonance at (i) 500 MHz,
orientations of the angular momentum in terms of the angle it makes with the (ii) 800 MHz.
z-axis? 47.8(b) Use the information in Table 47.2 to predict the magnetic fields at
47.2(b) For a 14N nucleus, what are the magnitude of the spin angular which (a) 14N, (b) 19F, and (c) 31P come into resonance at (i) 400 MHz,
momentum and its allowed components along the z-axis? What are the (ii) 750 MHz.
possible orientations of the angular momentum in terms of the angle it makes
47.9(a) Calculate the relative population differences (δN/N, where δN denotes
with the z-axis?
a small difference Nα − Nβ) for protons in fields of (a) 0.30 T, (b) 1.5 T, and
47.3(a) What is the resonance frequency of a proton in a magnetic field of (c) 10 T at 25 °C.
13.5 T? 47.9(b) Calculate the relative population differences (δN/N, where δN denotes
47.3(b) What is the resonance frequency of a 19F nucleus in a magnetic field a small difference Nα − Nβ) for 13C nuclei in fields of (a) 0.50 T, (b) 2.5 T, and
of 17.1 T? (c) 15.5 T at 25 °C.
47.4(a) 33S has a nuclear spin of 23 and a nuclear g-factor of 0.4289. Calculate 47.10(a) The first generally available NMR spectrometers operated at a
the energies of the nuclear spin states in a magnetic field of 6.800 T. frequency of 60 MHz; today it is not uncommon to use a spectrometer that
47.4(b) 14N has a nuclear spin of 1 and a nuclear g-factor of 0.404. Calculate operates at 800 MHz. What are the relative population differences of 13C spin
the energies of the nuclear spin states in a magnetic field of 10.50 T. states in these two spectrometers at 25 °C?
47.10(b) What are the relative population differences of 19F spin states in
47.5(a) Calculate the frequency separation of the nuclear spin levels of a 13C
spectrometers operating at 60 MHz and 450 MHz at 25 °C?
nucleus in a magnetic field of 15.4 T given that the magnetogyric ratio is
6.73 × 10−7 T−1 s−1. 47.11(a) What magnetic field would be required in order to use an EPR
47.5(b) Calculate the frequency separation of the nuclear spin levels of a 14N X-band spectrometer (9 GHz) to observe 1H-NMR and a 300 MHz
nucleus in a magnetic field of 14.4 T given that the magnetogyric ratio is spectrometer to observe EPR?
1.93 × 10−7 T−1 s−1. 47.11(b) Some commercial EPR spectrometers use 8 mm microwave radiation
(the Q band). What magnetic field is needed to satisfy the resonance
47.6(a) In which of the following systems is the energy level separation larger?
condition?
(a) A proton in a 600 MHz NMR spectrometer, (b) a deuteron in the same
spectrometer.
47.6(b) In which of the following systems is the energy level separation larger?
(a) A 14N nucleus in (for protons) a 600 MHz NMR spectrometer, (b) an
electron in a radical in a field of 0.300 T.
Problems
47.1‡ The relative sensitivity of NMR lines for equal numbers of different 47.2 With special techniques, known collectively as magnetic resonance
nuclei at constant temperature for a given frequency is Rν ∝ (I + 1)μ3, whereas imaging (MRI), it is possible to obtain NMR spectra of entire organisms.
for a given field it is RB ∝ {(I + 1)/I2}μ3. (a) From the data in Table 47.2, A key to MRI is the application of a magnetic field that varies linearly
calculate these sensitivities for the deuteron, 13C, 14N, 19F, and 31P relative to
the proton. (b) Derive the equation for RB from the equation for Rν. ‡ These problems were supplied by Charles Trapp and Carmen Giunta.
www.ebook3000.com
Atkins09819.indb 492 9/11/2013 12:37:20 PM

Exercises and problems 493
across the specimen. Consider a flask of water held in a field that varies in (Similar equations may be written for gradients along the x- and y-directions.)
the z-direction according to B0 + Gzz, where Gz is the field gradient Application of a 90° radiofrequency pulse with ν = νL(z) will result in a signal
along the z-direction. Then the water protons will be resonant at the with an intensity that is proportional to the number of protons at the position
frequencies z. Now suppose a uniform disk-shaped organ is in a linear field gradient,
and that the MRI signal is proportional to the number of protons in a slice of
γN width δz at each horizontal distance z from the centre of the disk. Sketch the
␯ L (z ) = (B +G z ) shape of the absorption intensity for the MRI image of the disk before any
2π 0 z
computer manipulation has been carried out.
Topic 48 Features of NMR spectra

48.1 Discuss in detail the origins of the local, neighbouring group, and solvent 48.2 Discuss how the Fermi contact interaction and the polarization
contributions to the shielding constant. mechanism contribute to spin–spin couplings in NMR.
Exercises
48.1(a) What are the relative values of the chemical shifts observed for nuclei 48.4(a) Sketch the form of the 19F-NMR spectra of a natural sample of 10 BF4−
in the spectrometers mentioned in Exercise 47.10(a) in terms of (a) δ values, and 11 BF4− .
(b) frequencies? 48.4(b) Sketch the form of the 31P-NMR spectra of a sample of 31 PF6− .
48.1(b) What are the relative values of the chemical shifts observed for nuclei
48.5(a) From the data in Table 47.2, predict the frequency needed for 19F-NMR
in the spectrometers mentioned in Exercise 47.10(b) in terms of (a) δ values,
in an NMR spectrometer designed to observe proton resonance at 800 MHz.
(b) frequencies?
Sketch the proton and 19F resonances in the NMR spectrum of FH2+ .
48.2(a) The chemical shift of the CH3 protons in acetaldehyde (ethanal) is 48.5(b) From the data in Table 47.2, predict the frequency needed for 31P-NMR
δ = 2.20 and that of the CHO proton is 9.80. What is the difference in local in an NMR spectrometer designed to observe proton resonance at 500 MHz.
magnetic field between the two regions of the molecule when the applied field Sketch the proton and 31P resonances in the NMR spectrum of PH +4 .
is (a) 1.5 T, (b) 15 T?
48.6(a) Sketch the form of an A3M2X4 spectrum, where A, M, and X are
48.2(b) The chemical shift of the CH3 protons in diethyl ether is δ = 1.16 and
protons with distinctly different chemical shifts and JAM > JAX > JMX.
that of the CH2 protons is 3.36. What is the difference in local magnetic field
48.6(b) Sketch the form of an A2M2X5 spectrum, where A, M, and X are
between the two regions of the molecule when the applied field is (a) 1.9 T,
protons with distinctly different chemical shifts and JAM > JAX > JMX.
(b) 16.5 T?
48.7(a) A proton jumps between two sites with δ = 2.7 and δ = 4.8. At what
48.3(a) Sketch the appearance of the 1H-NMR spectrum of acetaldehyde
rate of interconversion will the two signals collapse to a single line in a
(ethanal) using J = 2.90 Hz and the data in Exercise 48.2a in a spectrometer
spectrometer operating at 550 MHz?
operating at (a) 250 MHz, (b) 800 MHz.
48.7(b) A proton jumps between two sites with δ = 4.2 and δ = 5.5. At what
48.3(b) Sketch the appearance of the 1H-NMR spectrum of diethyl ether using
rate of interconversion will the two signals collapse to a single line in a
J = 6.97 Hz and the data in Exercise 48.2b in a spectrometer operating at
spectrometer operating at 350 MHz?
(a) 400 MHz, (b) 650 MHz.
Problems
48.1 You are designing an MRI spectrometer (see Problem 47.2). What field 48.3‡ Various versions of the Karplus equation (eqn 48.14) have been used
gradient (in microtesla per metre, μT m−1) is required to produce a separation to correlate data on vicinal proton coupling constants in systems of the type
of 100 Hz between two protons separated by the long diameter of a human R1R2CHCHR3R4. The original version (M. Karplus, J. Am. Chem. Soc. 85,
kidney (taken as 8 cm), given that they are in environments with δ = 3.4? The 2870 (1963)) is 3JHH = A cos2 φHH + B. When R3 = R4 = H, 3JHH = 7.3 Hz; when
radiofrequency field of the spectrometer is at 400 MHz and the applied field R3 = CH3 and R4 = H, 3JHH = 8.0 Hz; when R3 = R4 = CH3, 3JHH = 11.2 Hz.
is 9.4 T. Assume that only staggered conformations are important and determine
which version of the Karplus equation fits the data better.
48.2LG Refer to Fig. 48.14 and use mathematical software, a spreadsheet,
or the Living graphs (labelled LG) on the website of this book to draw a 48.4‡ It might be unexpected that the Karplus equation, which was first derived
family of curves showing the variation of 3JHH with φ for which A = +7.0 Hz, for 3JHH coupling constants, should also apply to vicinal coupling between the
B = −1.0 Hz, and C varies slightly from a typical value of +5.0 Hz. What is the nuclei of metals such as tin. T.N. Mitchell and B. Kowall (Magn. Reson. Chem.
effect of changing the value of the parameter C on the shape of the curve? In 33, 325 (1995)) studied the relation between 3JHH and 3JSnSn in compounds of
a similar fashion, explore the effect of the values of A and B on the shape of the type Me3SnCH2CHRSnMe3 and found that 3JSnSn = 78.86 3JHH + 27.84 Hz.
the curve. (a) Does this result support a Karplus-type equation for tin? Explain your
Atkins09819.indb 493 9/11/2013 12:37:23 PM

reasoning. (b) Obtain the Karplus equation for 3JSnSn and plot it as a function of 48.6 In a liquid, the dipolar magnetic field averages to zero. Show this result by
the dihedral angle. (c) Draw the preferred conformation. evaluating the average of the field given in eqn 48.15. Hint: The surface area
element is sin θ dθdφ in polar coordinates.
48.5 Show that the coupling constant as expressed by the Karplus equation
passes through a minimum when cos φ = B/4C.
Topic 49 Pulse techniques in NMR

49.1 Discuss in detail the effects of a 90° pulse and of a 180° pulse on a system 49.4 Discuss the origin of the nuclear Overhauser effect and how it can be
of spin- 12 nuclei in a static magnetic field. used to measure distances between protons in a biopolymer.
49.2 Suggest a reason why the relaxation times of 13C nuclei are typically much 49.5 Discuss the origins of diagonal and cross-peaks in the COSY spectrum of
longer than those of 1H nuclei. an AX system.
49.3 Suggest a reason why the spin–lattice relaxation time of a small molecule
(like benzene) in a mobile, deuterated hydrocarbon solvent increases, whereas
that of a large molecule (like a polymer) decreases.
Exercises
49.1(a) The duration of a 90° or 180° pulse depends on the strength of the B1 1
field. If a 180° pulse requires 12.5 μs, what is the strength of the B1 field? How NO2CH2CH2CH3
long would the corresponding 90° pulse require?
49.1(b) The duration of a 90° or 180° pulse depends on the strength of the B1 2
NO2CH2CH2CH3
field. If a 90° pulse requires 5 μs, what is the strength of the B1 field? How long δ
would the corresponding 180° pulse require?
3
49.2(a) Figure F10.1 shows the proton COSY spectrum of 1-nitropropane.
Account for the appearance of off-diagonal peaks in the spectrum.
49.2(b) Account for the observation that the COSY spectrum of alanine has 4
only two off-diagonal peaks between Δ = 1.00 and 8.50.
NO2CH2CH2CH3
5
5 4 3 δ 2 1
Figure F10.1 The COSY spectrum of 1-nitropropane (NO2CH2CH2CH3). The

circles show enhanced views of the spectral features. (Spectrum provided
by Prof. G. Morris.)
Problems
49.1‡ Suppose that the FID in Fig. 49.5 was recorded in a 400 MHz (b) Explore how the shape of the FID curve changes with changes in the
spectrometer, and that the interval between maxima in the oscillations in the Larmor frequency and the spin–spin relaxation time. (c) Use mathematical
FID is 0.12 s. What is the Larmor frequency of the nuclei and the spin–spin software to calculate and plot the Fourier transforms of the FID curves you
relaxation time? calculated in parts (a) and (b). How do spectral linewidths vary with the value
of T2? Hint: This operation can be performed with the ‘fast Fourier transform’
49.2 To gain some appreciation for the numerical work done by computers
routine available in most mathematical software packages. Please consult the
interfaced to NMR spectrometers, perform the following calculations.
package's user manual for details.
(a) The total FID F(t) of a signal containing many frequencies, each
49.3 (a) In many instances it is possible to approximate the NMR lineshape by
corresponding to a different nucleus, is given by
using a Lorentzian function of the form
F (t ) =
∑S
j
0j cos(2π␯Ljt )e −t /T2 j
I Lorenztian (ω ) =
S0T2
1+ T22 (ω −ω 0 )2
where, for each nucleus j, S0j is the maximum intensity of the signal, νLj is the where I(ω) is the intensity as a function of the angular frequency ω = 2πν, ω0 is
Larmor frequency, and T2j is the spin–spin relaxation time. Plot the FID for the resonance frequency, S0 is a constant, and T2 is the spin–spin relaxation time.
the case Confirm that for this lineshape the width at half-height is 1/πT2. (b) Under certain
circumstances, NMR lines are Gaussian functions of the frequency, given by
S01 = 1.0 νL1 = 50 MHz T21 = 0.50 μs

IGaussian (ω ) = S0T2 e −T2 (ω −ω 0 )
2 2
S02 = 3.0 νL2 = 10 MHz T22 = 1.0 μs
www.ebook3000.com
Atkins09819.indb 494 9/11/2013 12:37:27 PM

Confirm that for the Gaussian lineshape the width at half-height is equal to 49.6 The z-component of the magnetic field at a distance R from a magnetic
2(ln 2)1/2/T2. (c) Compare and contrast the shapes of Lorentzian and Gaussian moment parallel to the z-axis is given by eqn 49.10. In a solid, a proton at a
lines by plotting two lines with the same values of S0, T2, and ω0. distance R from another can experience such a field and the measurement of
the splitting it causes in the spectrum can be used to calculate R. In gypsum,
49.4 The shape of a spectral line, I(ω), is related to the free-induction decay
for instance, the splitting in the H2O resonance can be interpreted in terms
signal G(t) by
of a magnetic field of 0.715 mT generated by one proton and experienced by
the other. What is the separation of the protons in the H2O molecule?
∞
I (ω ) = a Re
∫
0
G(t )eiωt dt 49.7 In a liquid crystal a molecule might not rotate freely in all directions
and the dipolar interaction might not average to zero. Suppose a molecule
is trapped so that, although the vector separating two protons may rotate
where a is a constant and ‘Re’ means take the real part of what follows.
freely around the z-axis, the colatitude may vary only between 0 and θ ′. Use
Calculate the lineshape corresponding to an oscillating, decaying function
mathematical software to average the dipolar field over this restricted range
G(t) = cos ωt e−t/τ.
of orientation and confirm that the average vanishes when θ ′ is equal to π
49.5 In the language of Problem 49.4, show that, if G(t) = (a cos ωt + b cos ωt)e−t/τ, (corresponding to free rotation over a sphere). What is the average value of
then the spectrum consists of two lines with intensities proportional to a and b the local dipolar field for the H2O molecule in Problem 49.6 if it is dissolved
and located at ω = ω1 and ω2, respectively. in a liquid crystal that enables it to rotate up to θ ′ = 30°?
Topic 50 Electron paramagnetic resonance

50.1 Discuss how the Fermi contact interaction and the polarization 50.2 Explain how the EPR spectrum of an organic radical can be used to
mechanism contribute to hyperfine interactions in EPR. identify and map the molecular orbital occupied by the unpaired electron.
Exercises
50.1(a) The centre of the EPR spectrum of atomic hydrogen lies at 329.12 mT 50.4(a) Predict the intensity distribution in the hyperfine lines of the EPR
in a spectrometer operating at 9.2231 GHz. What is the g-value of the electron spectra of (a) ·CH3, (b) ·CD3.
in the atom? 50.4(b) Predict the intensity distribution in the hyperfine lines of the EPR
50.1(b) The centre of the EPR spectrum of atomic deuterium lies at 330.02 mT spectra of (a) ·CH2CH3, (b) ·CD2CD3.
in a spectrometer operating at 9.2482 GHz. What is the g-value of the electron
50.5(a) The benzene radical anion has g = 2.0025. At what field should
in the atom?
you search for resonance in a spectrometer operating at (a) 9.313 GHz,
50.2(a) A radical containing two equivalent protons shows a three-line (b) 33.80 GHz?
spectrum with an intensity distribution 1:2:1. The lines occur at 330.2 mT, 50.5(b) The naphthalene radical anion has g = 2.0024. At what field should
332.5 mT, and 334.8 mT. What is the hyperfine coupling constant for each you search for resonance in a spectrometer operating at (a) 9.501 GHz,
proton? What is the g-value of the radical given that the spectrometer is (b) 34.77 GHz?
operating at 9.319 GHz?
50.6(a) The EPR spectrum of a radical with a single magnetic nucleus is split
50.2(b) A radical containing three equivalent protons shows a four-line
into four lines of equal intensity. What is the nuclear spin of the nucleus?
spectrum with an intensity distribution 1:3:3:1. The lines occur at 331.4 mT,
50.6(b) The EPR spectrum of a radical with two equivalent nuclei of a
333.6 mT, 335.8 mT, and 338.0 mT. What is the hyperfine coupling constant
particular kind is split into five lines of intensity ratio 1:2:3:2:1. What is the
for each proton? What is the g-value of the radical given that the spectrometer
spin of the nuclei?
is operating at 9.332 GHz?
50.7(a) Sketch the form of the hyperfine structures of radicals XH2 and XD2,
50.3(a) A radical containing two inequivalent protons with hyperfine constants
where the nucleus X has I = 25 .
2.0 mT and 2.6 mT gives a spectrum centred on 332.5 mT. At what fields do
50.7(b) Sketch the form of the hyperfine structures of radicals XH3 and XD3,
the hyperfine lines occur and what are their relative intensities?
where the nucleus X has I = 23 .
50.3(b) A radical containing three inequivalent protons with hyperfine constants
2.11 mT, 2.87 mT, and 2.89 mT gives a spectrum centred on 332.8 mT. At what
fields do the hyperfine lines occur and what are their relative intensities?
Problems
50.1 It is possible to produce very high magnetic fields over small volumes by 50.2 The angular NO2 molecule has a single unpaired electron and can be
special techniques. What would be the resonance frequency of an electron spin trapped in a solid matrix or prepared inside a nitrite crystal by radiation
in an organic radical in a field of 1.0 kT? How does this frequency compare to damage of NO2− ions. When the applied field is parallel to the OO direction
typical molecular rotational, vibrational, and electronic energy level separations? the centre of the spectrum lies at 333.64 mT in a spectrometer operating
Atkins09819.indb 495 9/11/2013 12:37:30 PM

at 9.302 GHz. When the field lies along the bisector of the ONO angle, the 50.6 When an electron occupies a 2s orbital on an N atom it has a hyperfine
resonance lies at 331.94 mT. What are the g-values in the two orientations? interaction of 55.2 mT with the nucleus. The spectrum of NO2 shows an
isotropic hyperfine interaction of 5.7 mT. For what proportion of its time is
50.3 The hyperfine coupling constant in ·CH3 is 2.3 mT. Use the information in
the unpaired electron of NO2 occupying a 2s orbital? The hyperfine coupling
Table 50.1 to predict the splitting between the hyperfine lines of the spectrum
constant for an electron in a 2p orbital of an N atom is 3.4 mT. In NO2 the
of ·CD3. What are the overall widths of the hyperfine spectra in each case?
anisotropic part of the hyperfine coupling is 1.3 mT. What proportion of its
50.4 The p-dinitrobenzene radical anion can be prepared by reduction of time does the unpaired electron spend in the 2p orbital of the N atom in NO2?
p-dinitrobenzene. The radical anion has two equivalent N nuclei (I = 1) What is the total probability that the electron will be found on (a) the N atom,
and four equivalent protons. Predict the form of the EPR spectrum using (b) the O atoms? What is the hybridization ratio of the N atom? Does the
a(N) = 0.148 mT and a(H) = 0.112 mT. hybridization support the view that NO2 is angular?
50.5 The hyperfine coupling constants observed in the radical anions 1, 2, and 50.7 Sketch the EPR spectra of the di-tert-butyl nitroxide radical (4) at
3 are shown (in millitesla, mT). Use the value for the benzene radical anion to 292 K in the limits of very low concentration (at which electron exchange is
map the probability of finding the unpaired electron in the π orbital on each negligible), moderate concentration (at which electron exchange effects begin
C atom. to be observed), and high concentration (at which electron exchange effects
predominate).
NO2 NO2
NO2
0.011 NO2 0.112 0.112
0.450 0.272 –
– – N
0.112 0.112
0.0172 0.011 0.108 NO2 O
0.0172 0.450 NO2
1 2 3 4 di-tert-Butyl nitroxide
Integrated activities
F10.1 In Problem F6.15 you are asked to use molecular electronic structure During the course of these electron transfer reactions, a tyrosine radical
methods to investigate the hypothesis that the magnitude of the 13C chemical forms, with spin density delocalized over the side chain of the amino acid. (a)
shift correlates with the net charge on a 13C atom by calculating the net charge The phenoxy radical shown in 5 is a suitable model of the tyrosine radical.
at the C atom para to the substituents in the following series of molecules: Using molecular modelling software and the computational method of your
methylbenzene, benzene, trifluoromethylbenzene, benzonitrile, nitrobenzene. choice (semi-empirical or ab initio method), calculate the spin densities at
The 13C chemical shifts of the para C atoms in each of the molecules that you the O atom and at all of the C atoms in 5. (b) Predict the form of the EPR
examined are given below: spectrum of 5.
Substituent CH3 H CF3 CN NO2

O
δ 128.4 128.5 128.9 129.1 129.4
(a) Is there a linear correlation between net charge and 13C chemical shift of
5 Phenoxy radical
the para C atom in this series of molecules?
F10.4 NMR spectroscopy may be used to determine the equilibrium constant
(b) If you did find a correlation in part (a), use the concepts developed in
for dissociation of a complex between a small molecule, such as an enzyme
Topic 48 to explain the physical origins of the correlation.
inhibitor I, and a protein, such as an enzyme E:
F10.2 EPR spectra are commonly discussed in terms of the parameters that
occur in the spin-hamiltonian, a hamiltonian operator that incorporates EI E + I K I = [E][I]/[EI]
various effects involving spatial operators (like the orbital angular
In the limit of slow chemical exchange, the NMR spectrum of a proton in
momentum) into operators that depend on the spin alone. Show that if you
I would consist of two resonances: one at νI for free I and another at νEI for
use H = − g eγ eB0 sz − γ eB0lz as the true hamiltonian, then from second-order
bound I. When chemical exchange is fast, the NMR spectrum of the same
perturbation theory, the eigenvalues of the spin are the same as those of
proton in I consists of a single peak with a resonance frequency ν given
the spin-hamiltonian H = − g γ eB0 sz (note the g in place of ge) and find an
by ν = fIνI + fEIνEI, where fI = [I]/([I] + [EI]) and fEI = [EI]/([I] + [EI]) are,
expression for g.
respectively, the fractions of free I and bound I. For the purposes of analysing
F10.3 The computational techniques described in Topics 28–30 have shown the data, it is also useful to define the frequency differences δν = ν – νI and
that the amino acid tyrosine participates in a number of biological electron Δν = νEI – νI. Show that when the initial concentration of I, [I]0, is much
transfer reactions, including the processes of water oxidation to O2 in plant greater than the initial concentration of E, [E]0, a plot of [I]0 against δν−1 is a
photosystem II and of O2 reduction to water in cytochrome c oxidase. straight line with slope [E]0Δν and y-intercept –KI.
www.ebook3000.com
Atkins09819.indb 496 9/11/2013 12:37:35 PM

FOCUS 11 ON Statistical thermodynamics
Topic 51 Topic 52 Topic 53 Topic 54

The Molecular
Molecular The canonical
Boltzmann partition
energies ensemble
distribution functions
Independent
molecules Interacting
Focus 3 Focus 9 Focus 12–15
molecules
The quantum
Molecular Thermo-
mechanics
spectroscopy dynamics
of motion
One of the most important equations linking the properties of molecules to the properties of the
bulk matter to which the molecules contribute is the ‘Boltzmann distribution’, which is established
and developed in this group of Topics. The derivation of the distribution (Topic 51) is based on the
supposition that molecules are distributed over the available energy levels in the most probable way
subject to the total energy and the total number of particles of the system having particular values. It
can then be used to predict the population of each level for a system at a given temperature. Indeed,
the Boltzmann distribution provides insight into the nature of temperature as the single parameter
that determines the population of levels of a collection of molecules. As is demonstrated in other
Topics, the Boltzmann distribution lies at the heart of molecular interpretations of thermodynamics.
The Boltzmann distribution introduces the concept of ‘partition function’ (Topic 52), which is a
kind of ‘thermodynamic wavefunction’ of the system in the sense that it can be used to calculate
all its thermodynamic properties, as explained in the groups of Topics on The First Law of thermo-
dynamics, The Second and Third Laws of thermodynamics, Physical equilibria, and Chemical equilibria.
A ‘molecular partition function’, which is appropriate when molecules can be treated as not interact-
ing with one another, can be calculated from molecular properties, such as the mass (for the trans-
lational contribution), the moment of inertia (for the rotational contribution), and from the force
constant (for the vibrational contribution). Because parameters such as these can be determined
from Molecular spectroscopy, the partition function provides a very important link between spec-
troscopy and thermodynamics. When molecular interactions cannot be ignored, it is necessary to
make use of the concept and techniques related to a ‘canonical ensemble’ and the ‘canonical parti-
tion function’, as explained in Topic 54.
The most straightforward application of the molecular partition function is to the calculation of
the contribution of the various modes of motion (translation, rotation, and vibration as introduced
in The quantum mechanics of motion) to the mean energy of a collection of independent molecules
(Topic 53). That is a basis for the calculation and interpretation of the internal energy of a system, as
used in the First Law of thermodynamics.
Atkins09819.indb 497 9/11/2013 12:37:35 PM

TOPIC 51
The Boltzmann distribution

The crucial step in going from the quantum mechanical prop-
Contents erties of individual molecules to the properties of bulk samples,
51.1 Configurations and weights 499 the province of statistical thermodynamics, is to recognize
(a) Instantaneous configurations 499 that the latter deals with the average behaviour of large num-
Brief illustration 51.1: The weight of a configuration 499 bers of molecules. For example, the pressure of a gas depends
(b) The most probable distribution 500 on the average force exerted by its molecules and there is no
Brief illustration 51.2: The Boltzmann need to specify which molecules happen to be striking the wall
distribution 501 at any instant. Nor is it necessary to consider the fluctuations in
(c) The relative population of states 501 the pressure as different numbers of molecules collide with the
Example 51.1: Calculating the relative populations wall at different moments. The fluctuations in pressure are very
of rotational states 501 small compared with the steady pressure: it is highly improb-
51.2 The derivation of the Boltzmann distribution 502 able that there will be a sudden lull in the number of collisions,
(a) The role of constraints 502 or a sudden surge. Fluctuations in other bulk properties also
Example 51.2 : Using the method of Lagrange occur, but for large numbers of particles they are very much
multipliers 503 smaller than the mean values.
(b) Solving for the most probable distribution 503 We consider a system composed of N molecules, with N a
(c) The final step 504 large number (of the order of 1023 for systems typical of chem-
Checklist of concepts 504 istry). Although the total energy is constant at E, it is not pos-
Checklist of equations 505
sible to be definite about how that energy is shared between the
molecules. Collisions result in the ceaseless redistribution of
energy not only between the molecules but also between the
quantum states that each molecule occupies. The closest we can
➤ Why do you need to know this material? come to a description of the distribution of energy is to report
The Boltzmann distribution is the key to understanding the population of a state, the average number of molecules that
a great deal of chemistry. All thermodynamic properties occupy it, and to say that on average there are Ni molecules in
can be interpreted in its terms, as can the temperature a state of energy εi. The populations of the states remain almost
dependence of equilibrium constants and the rates of constant, but the precise identities of the molecules in each
chemical reactions. It also illuminates the meaning of state may change at every collision.
‘temperature’. There is, perhaps, no more important The problem we address in this Topic is the calculation of
unifying concept in chemistry. the populations of states for any type of molecule in any mode
of motion at any temperature. The only restriction is that the
➤ What is the key idea? molecules should be independent, in the sense that the total
The most probable distribution of molecules over the energy of the system is a sum of their individual energies. We
available energy levels subject to certain constraints are discounting (at this stage) the possibility that in a real sys-
depends on a single parameter, the temperature. tem a contribution to the total energy may arise from interac-
tions between molecules. We also adopt the principle of equal
➤ What do you need to know already? a priori probabilities, the assumption that all possibilities for
You need to be aware that molecules can exist only in the distribution of energy are equally probable. ‘A priori’ in this
certain discrete energy levels (Topics 9 −14) and that in context loosely means ‘as far as one knows’. We have no reason
some cases more than one state has the same energy. to presume otherwise than that for a collection of molecules at
thermal equilibrium, a vibrational state of a certain energy, for
www.ebook3000.com
Atkins09819.indb 498 9/11/2013 12:37:36 PM

51 The Boltzmann distribution 499
instance, is as likely to be populated as a rotational state of the

same energy.
One very important conclusion that will emerge from the
following analysis is that the overwhelmingly most probable
populations of the available states depend on a single param-
eter, the ‘temperature’. That is, the work we do here provides
a molecular justification for the concept of temperature and
some insight into this crucially important quantity. Figure 51.1 Whereas a configuration {5,0,0,…} can be achieved
in only one way, a configuration {3,2,0,…} can be achieved in
the ten different ways shown here, where the tinted circles
represent different molecules.
51.1 Configurations and weights
of the configuration. The weight of the configuration {N0,N1,…}
Any individual molecule may exist in states with energies ε0, is given by the expression
ε1, …. For reasons that will become clear, we shall always take
N!
the lowest available state as the zero of energy (that is, we set W= Weight of a configuration (51.1)
N 0 ! N1 ! N 2 !
ε0 = 0), and measure all other energies relative to that state. To
obtain the actual energy of the system we may have to add a with x! = x(x − 1)…1 and by definition 0! = 1. Equation 51.1 is a
constant to the energy calculated on this basis. For example, if generalization of the formula W = 12 N (N −1) , and reduces to it
we are considering the vibrational contribution to the energy, for the configuration {N − 2,2,0,…}.
then we must add the total zero-point energy of any oscillators
in the system.
Brief illustration 51.1 The weight of a configuration
(a) Instantaneous configurations To calculate the number of ways of distributing 20 identi-
cal objects with the arrangement 1, 0, 3, 5, 10, 1, we note that
At any instant there will be N0 molecules in the state 0 the configuration is {1,0,3,5,10,1} with N = 20; therefore the
with energy ε0, N1 in the state 1 with ε1, and so on, with weight is
N0 + N1 + … = N, the total number of molecules in the system.
Initially we shall suppose that all the states have exactly the same 20 !
W= = 9.31 × 108
energy. The specification of the set of populations N0, N1, … 1!0 !3!5!10 !1!
in the form {N0, N1, …} is a statement of the instantaneous con-
Self-test 51.1 Calculate the weight of the configuration in
figuration of the system. The instantaneous configuration fluc- which 20 objects are distributed in the arrangement 0, 1, 5, 0,
tuates with time because the populations change, perhaps as a 8, 0, 3, 2, 0, 1.
result of collisions. At this stage the energies of all the configura- Answer: 4.19 × 1010
tions are identical so there is no restriction on how many of the
N molecules are in each state.
We can picture a large number of different instantaneous
configurations. One, for example, might be {N,0,0,…}, corre-
Justification 51.1
sponding to every molecule being in state 0. Another might The weight of a configuration
be {N − 2,2,0,0,…}, in which two molecules are in state 1. The
First, consider the weight of the configuration {N − 2,2,0,0,…},
latter configuration is intrinsically more likely to be found
which is prepared from the configuration {N,0,0,0, …} by the
than the former because it can be achieved in more ways: migration of two molecules from state 0 into state 2. One can-
{N,0,0,…} can be achieved in only one way, but {N − 2,2,0,…} didate for migration to state 1 can be selected in N ways. There
can be achieved in 12 N (N −1) different ways (Fig. 51.1; see the are N − 1 candidates for the second choice, so the total number
following Justification). If, as a result of collisions, the system of choices is N(N − 1). However, we should not distinguish the
were to fluctuate between the configurations {N,0,0,…} and choice (Jack, Jill) from the choice (Jill, Jack) because they lead
{N − 2,2,0,…}, it would almost always be found in the second, to the same configurations. Therefore, only half the choices
more likely configuration (especially if N were large). In other lead to distinguishable configurations, and the total number
words, a system free to switch between the two configurations of distinguishable choices is 12 N (N −1) .
would show properties characteristic almost exclusively of the Now we generalize this remark. Consider the number of
second configuration. A general configuration {N0,N1,…} can ways of distributing N balls into bins. The first ball can be
be achieved in W different ways, where W is called the weight selected in N different ways, the next ball in N − 1 different
Atkins09819.indb 499 9/11/2013 12:37:39 PM

500 11 Statistical thermodynamics
ways for the balls remaining, and so on. Therefore, there are is adequate. Then the approximate expression for the weight is
N(N − 1)…1 = N! ways of selecting the balls for distribution
over the bins. However, if there are N0 balls in the bin labelled lnW = {N ln N − N } − ∑{N ln N − N }i i i
0, there would be N0! different ways in which the same balls i
could have been chosen (Fig. 51.2). Similarly, there are N1! = N ln N − N − ∑N ln N + N [because∑N = N ]
i i i
ways in which the N1 balls in the bin labelled 1 can be chosen, i i
and so on. Therefore, the total number of distinguishable ways = N ln N − ∑N ln N

i i (51.3)
of distributing the balls so that there are N 0 in bin 0, N1 in i
bin 1, etc. regardless of the order in which the balls were cho-
sen is N!/N0!N1!…, which is the content of eqn 51.1.
(b) The most probable distribution
N = 18
We have seen that the configuration {N − 2,2,0,…} dominates
{N,0,0,…}, and it should be easy to believe that there may be
other configurations that have a much greater weight than
both. We shall see, in fact, that there is a configuration with so
great a weight that it overwhelms all the rest in importance to
such an extent that the system will almost always be found in
it. The properties of the system will therefore be characteristic
of that particular dominating configuration. This dominating
3! 6! 5! 4! configuration can be found by looking for the values of Ni that
lead to a maximum value of W. Because W is a function of all
Figure 51.2 The 18 molecules shown here can be
the Ni, we can do this search by varying the Ni and looking
distributed into four receptacles (distinguished by the
for the values that correspond to dW = 0 (just as in the search
three vertical lines) in 18! different ways. However, 3! of the
for the maximum of any function), or equivalently a maxi-
selections that put three molecules in the first receptacle
mum value of ln W. However, there are two difficulties with
are equivalent, 6! that put six molecules into the second
receptacle are equivalent, and so on. Hence the number of this procedure.
distinguishable arrangements is 18!/3!6!5!4!. At this point we allow the states to have different energies.
The first difficulty that results from this change is the need to
take into account the fact that the only permitted configura-
It will turn out to be more convenient to deal with the natu- tions are those corresponding to the specified, constant, total
ral logarithm of the weight, ln W, rather than with the weight energy of the system. This requirement rules out many con-
itself. We shall therefore need the expression figurations; {N,0,0,…} and {N − 2,2,0,…}, for instance, have
different energies (unless the states 0 and 1 happen to be
N!
lnW = ln = ln N ! − ln N 0! N1! N 2 ! degenerate), so both cannot occur in the same isolated system.
N 0! N1! N 2!
It follows that in looking for the configuration with the great-
= ln N ! − ln N 0! − ln N1! − ln N 2! − est weight, we must ensure that the configuration also satisfies
= ln N ! − ∑ ln N !i the condition
∑N ε = E
i
Energy Constant
i i constraint total energy (51.4)
where in the first line we have used ln(x/y) = ln x − ln y and in i
the second ln xy = ln x + ln y. One reason for introducing ln W where E is the total energy of the system.
is that it is easier to make approximations. In particular, we can The second constraint is that, because the total number of
simplify the factorials by using Stirling’s approximation molecules present is also fixed (at N), we cannot arbitrarily vary
all the populations simultaneously. Thus, increasing the popu-
ln x ! ≈ ( x + 12 ) ln x − x + 12 ln 2π x >> 1 Stirling’s
approximation
(51.2a) lation of one state by 1 demands that the population of another
state must be reduced by 1. Therefore, the search for the maxi-
This approximation is in error by less than 1 per cent when x is mum value of W is also subject to the condition
greater than about 10. We deal with far larger values of x, and
∑N = N
Constant total Number
the simplified version i number of (51.5)
constraint
i molecules
ln x !≈ x ln x − x x >> 1 Stirling’s approximation (51.2b) We show in the next section that the populations in the con-
(for large x)
figuration of greatest weight, subject to the two constraints in
www.ebook3000.com
Atkins09819.indb 500 9/11/2013 12:37:45 PM

eqns 51.4 and 51.5, depend on the energy of the state according (c) The relative population of states
to the Boltzmann distribution:
If we are interested only in the relative populations of states,
Ni e − βε i
the sum in the denominator of the Boltzmann distribution
= Boltzmann distribution (51.6a)
N ∑ i e − βε i need not be evaluated, because it cancels when the ratio is
taken:
The denominator in eqn 51.6a is denoted q and called the parti-
tion function: N i e − βε i
Thermal Boltzmann
= = e − β (ε −ε )
i j
equilibrium population (51.8a)
N j e − βε j
ratio
q= ∑e − βε i
Definition Partition function (51.6b)
i
That β ∝ 1/T is plausible is demonstrated by noting from eqn
In this Topic, the partition function is no more than a con- 51.8a that for a given energy separation the ratio of popula-
venient abbreviation for the sum; but in Topic 52 we see that tions N1/N0 decreases as β increases, which is what is expected
it is central to the statistical interpretation of thermodynamic as the temperature decreases. At T = 0 (β = ∞) all the popula-
properties. tion is in the ground state and the ratio is zero. Equation 51.8a
Equation 51.6a is the justification of the remark that a single is enormously important for understanding a wide range of
parameter, here denoted β, determines the most probable pop- chemical phenomena and is the form in which the Boltzmann
ulations of the states of the system. We shall confirm in Topic distribution is commonly employed (for instance, in the dis-
54 and anticipate throughout this Topic that cussion of the intensities of spectral transitions, Topic 16). It
tells us that the relative population of two states falls off expo-
1
β= (51.7) nentially with their difference in energy.
kT
A very important point to note is that the Boltzmann distri-
where T is the thermodynamic temperature and k is Boltzmann’s bution gives the relative populations of states, not energy lev-
constant. In other words: els. Several states might correspond to the same energy, and
each state has a population given by eqn 51.6. If we want to
The temperature is the unique parameter that governs the
consider the relative populations of energy levels rather than
most probable populations of states of a system at thermal
states, then we need to take into account this degeneracy. Thus,
equilibrium.
if the level of energy εi is gi-fold degenerate (in the sense that
there are gi states with that energy), and the level of energy εj
is gj-fold degenerate, then the relative total populations of the
Brief illustration 51.2 The Boltzmann distribution levels are given by
Suppose that two conformations of an organic molecule differ
in energy by 5.0 kJ mol−1 (corresponding to 8.3 × 10 −21 J for a Ni g i e − βε gi Thermal Boltzmann
single molecule), so conformation A lies at energy 0 and con- = = i e − β (ε −ε )
i j
equilibrium, population (51.8b)
N j g j e − βε g j
j
degeneracies ratio
formation B lies at ε = 8.3 × 10−21 J. At 20 °C (293 K) the denom-
inator in eqn 51.6a is
∑e
−21
J )/(1.381 ×10−23 J ) × (293 K )
− βε i
= 1 + e − ε /kT = 1 + e −(8.3 ×10 = 1.113
i Example 51.1 Calculating the relative populations
of rotational states
The proportion of molecules in conformation B at this tem-
perature is therefore Calculate the relative populations of the J = 1 and J = 0 rota-
−21
tional states of HCl at 25 °C.
J )/(1.381 ×10−23 J ) × (293 K )
N B e −(8.3 ×10
= = 0.11 Method Although the ground state is non-degenerate, the
N 1.13
level with J = 1 is triply degenerate (MJ = 0, ±1); see Topic 41.
or 11 per cent of the molecules. The remaining 89 per cent of From that Topic, the energy of state with quantum number J
the molecules are in conformation A. ( J + 1) . Use B = 10.591 cm −1. A useful relation is
is ε J = hcBJ
kT/hc = 207.22 cm−1 at 298.15 K.
Self-Test 51.2 Suppose that there is a third conformation a fur-
ther 0.50 kJ mol−1 above B. What proportion of molecules will Answer The energy separation of states with J = 1 and J = 0 is
now be in conformation B?
ε1 − ε 0 = 2hcB
Answer: 0.10, 10 per cent
Atkins09819.indb 501 9/11/2013 12:37:55 PM

The first constraint recognizes that the total energy must not
The ratio of the population of a state with J = 1 and any one of
change, and the second recognizes that the total number of
its three states MJ to the population of the single state with J = 0
is therefore
molecules must not change. These two constraints prevent us
from solving d ln W = 0 simply by setting all (∂ ln W/∂Ni) = 0
N J ,MJ because the dNi are not all independent.
= e −2hcBβ
N0 The way to take constraints into account was devised by the
French mathematician Lagrange, and is called the method of
The relative populations of the levels, taking into account the undetermined multipliers (The chemist’s toolkit 51.1). All we
threefold degeneracy of the upper state, is need here is the rule that:
NJ r a constraint should be multiplied by a constant and then
= 3e −2hcBβ
N0 added to the main variation equation.
Inser t ion of hcB β = hcB /kT = (10.591cm −1 )/(207.22 cm −1 ) = The variables are then treated as though they were all inde-
0.05111 then gives pendent, and the constants are evaluated at the end of the
NJ calculation.
= 3e −2 × 0.05111 = 2.708
N0
We see that because the J = 1 level is triply degenerate, it has The chemist’s toolkit 51.1 The method of undetermined
a higher population than the level with J = 0, despite being of multipliers
higher energy. As the example illustrates, it is very important
Suppose we need to find the maximum (or minimum) value of
to take note of whether you are asked for the relative popula-
some function f that depends on several variables x1, x 2, …, xn.
tions of individual states or of a (possibly degenerate) energy
When the variables undergo a small change from xi to xi + δxi
level.
the function changes from f to f + δf, where
Self-test 51.3 What is the ratio of the populations of the levels n
⎛ ∂f ⎞
with J = 2 and J = 1 at the same temperature?
Answer: 1.359
δf = ∑ ⎜⎝ ∂x ⎟⎠ δx
i =1
i
i
At a minimum or maximum, δf = 0, so then

n
⎛ ∂f ⎞
51.2The derivation of the Boltzmann ∑ ⎜⎝ ∂x ⎟⎠ δx = 0
i
i
distribution
i =1
If the x i were all independent, all the δx i would be arbitrary,

We remarked earlier that ln W is easier to handle than W. and this equation could be solved by setting each (∂f/∂x i) = 0
Therefore, to find the form of the Boltzmann distribution, we individually. When the xi are not all independent, the δxi are
look for the condition for ln W being a maximum rather than not all independent, and the simple solution is no longer valid.
dealing directly with W. Because ln W depends on all the Ni, We proceed as follows.
when a configuration changes and the Ni change to Ni + dNi, the Let the constraint connecting the variables be an equa-
tion of the form g = 0. The constraint g = 0 is always valid, so g
function ln W changes to ln W + d ln W, where
remains unchanged when the xi are varied:
⎛ ∂ lnW ⎞
d lnW = ∑ ⎜⎝
i
dN
∂N i ⎟⎠ i δg =
n
⎛ ∂g ⎞
∑ ⎜⎝ ∂x ⎟⎠ δx = 0 i
i
i =1
(a) The role of constraints Because δg is zero, we can multiply it by a parameter, λ, and
add it to the preceding equation:
All the last expression states is that a change in ln W is the sum
n
⎧⎛ ∂f ⎞ ⎛ ∂g ⎞ ⎫
∑ ⎨⎩⎜⎝ ∂x ⎟⎠ + λ ⎜⎝ ∂x ⎟⎠ ⎬⎭δx = 0
of contributions arising from changes in each value of Ni. At a
maximum, d ln W = 0. However, when the Ni change, they do so i =1
i i
i
subject to the two constraints

This equation can be solved for one of the δx, δxn for instance,
∑ε dN = 0
i
i i ∑dN = 0
i
i Constraints (51.9) in terms of all the other δx i. All those other δx i (i = 1, 2, …,
www.ebook3000.com
Atkins09819.indb 502 9/11/2013 12:38:07 PM

n − 1) are independent, because there is only one constraint the constraint in the form g = 0. Introduce the Lagrange mul-
on the system. But λ is arbitrary; therefore we can choose it tiplier λ for this constraint and construct and solve the corre-
so that sponding Lagrange variation equations.
⎛ ∂f ⎞ ⎛ ∂g ⎞ Answer Let the sides of the box be a, b, and c; then its vol-
⎜⎝ ∂x ⎟⎠ + λ ⎜⎝ ∂x ⎟⎠ = 0 (A)
ume is V = abc. The total area of the surface of the box is
n n
2(ab + bc + ca), so the constraint on the construction is
Then ab + bc + ca = A/2. In the notation used in The chemist’s toolkit
n−1
51.1, g = ab + bc + ca − A/2 = 0. On introducing the Lagrange
⎧⎛ ∂f ⎞ ⎛ ∂g ⎞ ⎫
∑ ⎨⎩⎜⎝ ∂x ⎟⎠ + λ ⎜⎝ ∂x ⎟⎠ ⎬⎭δx = 0
i =1
i i
i
multiplier as λ, the three equations we have to solve to find the
maximum volume subject to the constraint are
∂V ∂g
Now the n −1 variations δxi are independent, so the solution of −λ = 0 x = a, b, c
this equation is ∂x ∂x
⎛ ∂f ⎞ ⎛ ∂g ⎞ That is,
⎜⎝ ∂x ⎟⎠ + λ ⎜⎝ ∂x ⎟⎠ = 0 i = 1, 2,…, n −1
i i ∂V ∂g
−λ = bc − λ (b + c) = 0
∂a ∂a
However, eqn A has exactly the same form as this equation, so ∂V ∂g
−λ = ac − λ (a + c) = 0
the maximum or minimum of f can be found by solving ∂b ∂b
∂V ∂g
⎛ ∂f ⎞ ⎛ ∂g ⎞ −λ = ab − λ (a + b) = 0
⎜⎝ ∂x ⎟⎠ + λ ⎜⎝ ∂x ⎟⎠ = 0 i = 1, 2, …, n ∂c ∂c
i i
Multiplication of the first by a and the second by b and then

If there is more than one constraint, g1 = 0, g2 = 0, …, and this subtracting the second from the first gives
final result generalizes to
λ (ab + bc) − λ (ab + ac) = 0, or bc − ac = 0
⎛ ∂f ⎞ ⎛ ∂g 1 ⎞ ⎛ ∂g 2 ⎞
⎜⎝ ∂x ⎟⎠ + λ1 ⎜⎝ ∂x ⎟⎠ + λ2 ⎜⎝ ∂x ⎟⎠ + = 0, i = 1, 2, …, n or a = b. Likewise, the other solutions are a = c and b = c.
i i i
We conclude that the box of greatest volume is a cube
with a corresponding multiplier, λ1, λ2, … for each constraint. with sides a = b = c and volume V = a 3. From the constraint,
g = 3a 2 − A/2 = 0, so a = (A/6)1/2. In this instance, with A = 1.0 m2,
a = 0.41 m, so V = 0.069 m3.
We employ the technique as follows. The two constraints in Self-test 51.4 Suppose that you require the height of the box,
eqn 51.9 are multiplied by the constants −β and α, respectively a, to be fixed at the value h = 0.40 m. What is the maximum
(the minus sign in −β has been included for future conveni- volume of the box given the same sheet of metal?
ence), and then added to the expression for d ln W: Answer: V = 0.067 m3
⎛ ∂ lnW ⎞
d lnW = ∑ ⎜⎝
i
dN + α
∂N i ⎟⎠ i ∑dN − β ∑ε dN
i
i
i
i i
⎧⎛ ∂ lnW ⎞ ⎫ (b) Solving for the most probable

= ∑ ⎨⎩⎜⎝
i
⎟
∂N i ⎠
+ α − βε i ⎬ dN i
⎭ distribution
All the dNi are now treated as independent. Hence the only way
of satisfying d ln W = 0 is to require that, for each i,
Example 51.2 Using the method of Lagrange
multipliers ⎛ ∂ lnW ⎞
⎜⎝ ∂N ⎟⎠ + α − βε i = 0 (51.10)
Suppose you have a sheet of metal of area A and want to build i
a box of the greatest volume possible from that sheet. What is

the maximum volume when A = 1.0 m2? when the Ni have their most probable values.
Equation 51.3 for W is
Method Identify the volume in terms of the lengths of the
sides (which are mutually perpendicular in a box) and identify lnW = N ln N − ∑N ln N
i
i i
Atkins09819.indb 503 9/11/2013 12:38:18 PM

There is a small housekeeping step to take before differentiating ∂Nj lnN j ⎧⎪⎛ ∂Nj ⎞ ⎛ ∂ ln N j ⎞ ⎫⎪
ln W with respect to Ni: this equation is identical to ∑j
∂N i
= ∑ ⎨⎩⎪⎜⎝ ∂N ⎟⎠ ln N + N ⎜⎝
j i ∂N i ⎟⎠ ⎬⎪
j
⎭
j
lnW = N ln N − ∑ Nj ln N j ⎧⎪⎛ ∂N j ⎞ ⎫
⎛ ∂N j ⎞ ⎪
j = ∑j
⎨⎜ ∂N ⎟ ln N j + ⎜ ∂N ⎟ ⎬
⎩⎪⎝ i ⎠ ⎝ i ⎠ ⎭⎪
because all we have done is to change the ‘name’ of the states
⎛ ∂N j ⎞
from i to j. This step makes sure that we do not confuse the i in
the differentiation variable (Ni) with the i in the summation.
= ∑ ⎜⎝ ∂N ⎟⎠ (ln N +1)
j i
j
Now differentiation of this expression gives

= ∑δ (ln N + 1) = ln N +1
j
ij j i
∂ lnW ∂N ln N ∂N j ln N j
∂N i
=
∂N i
− ∑ j
∂N i
(51.11)
and therefore
The derivative of the first term on the right is obtained as
∂ lnW N
follows: = ln N + 1− (ln N i + 1) = − ln i (51.16)
∂N i N
∂N ln N ⎛ ∂N ⎞ ⎛ ∂ ln N ⎞ It follows from eqn 51.10 that
=⎜ ln N + N ⎜
∂N i ⎝ ∂N i ⎟⎠ ⎝ ∂N i ⎟⎠
(51.12) Ni
∂N − ln + α − βε i = 0
= ln N + = ln N + 1 N
∂N i
and therefore that
The ln N in the first term on the right in the second line arises
N i α −βε (51.17)
because N = N1 + N2 + … and so the derivative of N with respect to =e i
N
any of the Ni is 1: that is, ∂N/∂Ni = 1. The second term on the right
in the second line arises because ∂(ln N)/∂Ni = (1/N)∂N/∂Ni.
(c) The final step
The final 1 is then obtained in the same way as in the preceding
remark, by using ∂N/∂Ni = 1. At this stage we note that
For the derivative of the second term we first note that
∂ ln N j 1 ⎛ ∂Nj ⎞
N= ∑N = ∑Ne
i
i
i
α − βε i
= Neα ∑e
i
− βε i
= (51.13)
∂N i Nj ⎜⎝ ∂N i ⎟⎠ Because the N cancels on each side of this equality, it follows that
Moreover, if i ≠ j, Nj is independent of Ni, so ∂Nj/∂Ni = 0.
1
However, if i = j, eα = (51.18)
∑ e − βε i
∂N j ∂N i i
= =1 (51.14)
∂N i ∂N i and
Therefore,
N i α −βε e − βε i
∂N j =e = eα e − βε =
i i
Boltzmann distribution (51.19)
=δ ij (51.15) N ∑ e − βε i
∂N i i
with δij the Kronecker delta (δij = 1 if i = j, δij = 0 otherwise). Then which is eqn 51.6a.
☐ 1. The instantaneous configuration of a system of N mol- ☐ 3. The principle of equal a priori probabilities assumes
ecules is the specification of the set of populations N0, that all possibilities for the distribution of energy are
N1, … of the energy levels ε0, ε1, … . equally probable.
☐ 2. The Boltzmann distribution gives the numbers of mol- ☐ 4. The relative populations of energy levels, as opposed
ecules in each state of a system at any temperature. to states, must take into account the degeneracies of the
energy levels.
www.ebook3000.com
Atkins09819.indb 504 9/11/2013 12:38:30 PM

Boltzmann distribution N i / N = e − βεi /q β = 1/kT 51.6a
Partition function q=
∑ei
− βε i See Topic 52 51.6b
Boltzmann ratio N i /N j = ( g i / g j )e − β (εi −ε j ) gi, gj are degeneracies 51.8
Atkins09819.indb 505 9/11/2013 12:38:32 PM

TOPIC 52
Molecular partition functions
Contents ➤ What do you need to know already?

52.1 The significance of the partition function 506 You need to know that the Boltzmann distribution
Brief illustration 52.1: A partition function 507 expresses the most probable distribution of molecules
52.2 Contributions to the partition function 508 over the available energy levels (Topic 51). In that Topic
(a) The translational contribution 509 we introduce the concept of partition function, which is
Brief illustration 52.2: The translational developed further here.
partition function 510
(b) The rotational contribution 510
Example 52.1: Evaluating the rotational
partition function explicitly 510 The partition function q = Σ i e − βε is introduced in Topic 51 sim-
i
Brief illustration 52.3: The rotational contribution 511 ply as a symbol to denote the sum over states that occurs in the
Brief illustration 52.4: The symmetry number 513 denominator of the Boltzmann distribution (eqn 51.6a). But it
(c) The vibrational contribution 514 is far more important than that might suggest. For instance, it
Brief illustration 52.5: The vibrational partition contains all the information needed to calculate the bulk prop-
function 514 erties of a system of independent particles. In this respect q
Example 52.2: Calculating a vibrational plays a role for bulk matter very similar to that played by the
partition function 514 wavefunction in quantum mechanics for individual molecules:
(d) The electronic contribution 515 q is a kind of thermal wavefunction. This Topic shows how the
Brief illustration 52.6: The electronic partition partition function is calculated in a variety of important cases
function 515 in preparation for seeing how thermodynamic information is
Checklist of concepts 516
extracted.
The significance of the

52.1
➤ Why do you need to know this material? partition function
Statistical thermodynamics provides the link between
molecular properties that have been calculated or The molecular partition function is
derived from spectroscopy and thermodynamic properties,
including equilibrium concepts. The connection is the
partition function. Therefore, this material is an essential q= ∑e
states i
− βε i
Definition Molecular partition function (52.1a)
foundation for understanding physical and chemical
properties of bulk matter in terms of the properties of the
constituent molecules. where β = 1/kT. As emphasized in Topic 51, the sum is over
states, not energy levels. If gi states have the same energy εi (so
➤ What is the key idea? the level is gi-fold degenerate), we write
The partition function is calculated by drawing on calculated
or spectroscopically derived structural information about
molecules. q= ∑g e
levels i
i
− βε i Alternative
definition
Molecular partition
function (52.1b)
www.ebook3000.com
Atkins09819.indb 506 9/11/2013 12:38:37 PM

52 Molecular partition functions 507
where the sum is now over energy levels (sets of states with 1.4 2
the same energy), not individual states. Also as emphasized in
Topic 51, we always take the lowest available state as the zero of
energy and set ε0 = 0.
Partition function, q
Brief illustration 52.1 A partition function 1.2 1.5
Suppose a molecule is confined to the following non-degen-

erate energy levels: 0, ε , 2 ε , … (Fig. 52.1; later in this Topic
we shall see that this array of levels is used when considering
molecular vibration). Then the molecular partition function is
1 1
0 0.5 1 0 5 10
( )
2
q = 1+ e − βε
+e −2 βε
+…= 1 + e − βε
+ e − βε
+ Temperature, kT/ε Temperature, kT/ε
Figure 52.3 The partition function for a two-level system as

The sum of the geometrical series 1 + x + x 2 + … is 1/(1 − x), so in a function of temperature. The two graphs differ in the scale
this case of the temperature axis to show the approach to 1 as T → 0
1 and the slow approach to 2 as T → ∞.
q=
1 − e − βε
This function is plotted in Fig. 52.2. We have derived the following important expression for the
partition function for a uniform ladder of states of spacing ε:
1
q= Uniform ladder Partition function (52.2a)
1− e − βε
ε
We can use this expression to interpret the physical signifi-
...
3ε cance of a partition function. To do so, we first note that the

2ε Boltzmann distribution for this arrangement of energy lev-
ε els gives the fraction, pi = Ni/N, of molecules in the state with
0 energy εi as
Figure 52.1 The equally spaced infinite array of energy e − βε i
pi =
levels used in the calculation of the partition function. q Uniform ladder Population (52.2b)
= (1− e )e
A harmonic oscillator has the same spectrum of levels. − βε − βε i
10
Figure 52.4 shows how pi varies with temperature. At very low

temperatures (large β), where q is close to 1, only the lowest
Partition function, q
state is significantly populated. As the temperature is raised, the

population breaks out of the lowest state, and the upper states
5
become progressively more highly populated. At the same
time, the partition function rises from 1 towards 2 and then
higher, so we see that its value gives an indication of the range
of states populated at any given temperature. The name ‘parti-
0 tion function’ reflects the sense in which q measures how the
0 5 10
Temperature, kT/ε
total number of molecules is distributed—partitioned—over
the available states.
The corresponding expressions for a two-level system,
Figure 52.2 The partition function for the system shown in
Fig. 52.1 (a harmonic oscillator) as a function of temperature.
derived in Self-test 52.1, are
Self-test 52.1 Suppose the molecule can exist in only two q = 1+ e − βε Two-level system Partition function (52.3a)
states, with energies 0 and ε. Derive and plot the expression
for the partition function. e − βε i
e − βε i
Answer: q = 1 + e−βε; see Fig 52.3. pi = = (52.3b)

q 1+ e − βε Two-level system Population
Atkins09819.indb 507 9/11/2013 12:38:43 PM

Low High When T is close to zero, the parameter β = 1/kT is close to infin-
temperature temperature ity. Then every term except one in the sum defining q is zero
because each one has the form e−x with x → ∞. The exception is
the term with ε0 ≡ 0 (or the g0 terms at zero energy if the ground
state is g0-fold degenerate), because then ε0/kT ≡ 0 whatever
the temperature, including zero. As there is only one surviving
term when T = 0, and its value is g0, it follows that
lim q = g 0
T →0
βε: 3.0 1.0 0.7 0.3
q: 1.05 1.58 1.99 3.86 That is, at T = 0, the partition function is equal to the degen-
eracy of the ground state (commonly, but not necessarily, 1).
Figure 52.4 The populations of the energy levels of the
Now consider the case when T is so high that for each term
system shown in Fig. 52.1 at different temperatures, and the
in the sum εj/kT ≈ 0. Because e−x = 1 when x = 0, each term in the
corresponding values of the partition function as calculated
sum now contributes 1. It follows that the sum is equal to the
from eqn 52.2a. Note that β = 1/kT.
number of molecular states, which in general is infinite:
In this case, because ε0 = 0 and ε1 = ε, lim q = ∞

T→∞
1 e − βε In some idealized cases, the molecule may have only a finite

p0 = p1 = (52.4)
1+ e − βε 1+ e − βε number of states; then the upper limit of q is equal to the num-
ber of states, as we saw for the two-level system.
These functions are plotted in Fig. 52.5. Notice how the pop- In summary:
ulations are p0 = 1 and p1 = 0 and the partition function is q = 1
The molecular partition function gives an indication of
(one state occupied) at T = 0. However, the populations tend
the number of states that are thermally accessible to a
towards equality ( p 0 = 12 , p1 = 12 ) and q = 2 (two states occupied)
molecule at the temperature of the system.
as T → ∞. Incidentally, a common error is to suppose that when
T = ∞ all the molecules in the system will be found in the upper At T = 0, only the ground level is accessible and q = g0. At very
energy state; however, we see from eqn 52.4 that as T → ∞, high temperatures, virtually all states are accessible, and q is
the populations of states become equal. The same conclusion correspondingly large.
is true of multi-level systems too: as T → ∞, all states become
equally populated.
We can now generalize the conclusion that the partition
function indicates the number of thermally accessible states.
Contributions to the partition
52.2
function
1 1
p0 The energy of a molecule is the sum of contributions from its
different modes of motion:
ε i = ε iT + ε iR + ε iV + ε iE
Population, p
p0 (52.5)
0.5 0.5
p1 where T denotes translation, R rotation, V vibration, and E the
electronic contribution. The separation of terms in eqn 52.5 is
p1 only approximate (except for translation) because the modes
are not completely independent, but in most cases it is satisfac-
0
tory. The separation of the electronic and vibrational motions
0
0 0.5 1 0 5 10 is justified provided only the ground electronic state is occu-
Temperature, kT/ε Temperature, kT/ε
pied (for otherwise the vibrational characteristics depend on
Figure 52.5 The fraction of populations of the two states of a the electronic state) and, for the electronic ground state, that
two-level system as a function of temperature (eqn 52.4). Note the Born–Oppenheimer approximation is valid (Topic 22). The
that as the temperature approaches infinity, the populations separation of the vibrational and rotational modes is justified to
of the two states become equal (and the fractions both the extent that the rotational constant (Topic 41) is independ-
approach 0.5). ent of the vibrational state.
www.ebook3000.com
Atkins09819.indb 508 9/11/2013 12:38:57 PM

Given that the energy is a sum of independent contributions, function also increases with increasing temperature (decreas-
the partition function factorizes into a product of contributions: ing β), because more states become accessible.
Justification 52.1
q= ∑e i
− βε i
= ∑e
i ( all states )
− βε iT − βε iR − βε iV − βε iE The partition function for a particle in
a one-dimensional box
= ∑ ∑ ∑ ∑ e − βε i − βε i − βε i − βε i
T R V E
The energy levels of a molecule of mass m in a container of
i ( translational ) i ( rotational ) i ( vibrational ) i ( electronic ) length X are given by eqn 9.8b with L = X:
⎛ ⎞⎛ ⎞
=⎜ ∑ e − βε ⎟ ⎜ ∑
e − βε ⎟
T R
i i n2 h 2
En =
⎝ i ( translational ) ⎠ ⎝ i (rotational ) ⎠ 8mX 2
⎛ ⎞⎛ ⎞
×⎜ ∑ e − βε ⎟ ⎜ ∑
e − βε ⎟ The lowest level (n = 1) has energy h 2/8mX 2 , so the energies
V E
i i
⎝ i ( vibrational ) ⎠ ⎝ i (electronic ) ⎠ relative to that level are
ε n = (n2 −1)ε ε = h2 /8mX 2

That is,
The sum to evaluate is therefore
q = q Tq R q V q E Factorization of the partition function (52.6)
∞
This factorization means that we can investigate each con-

q XT = ∑e
n=1
− (n2 −1)βε
tribution separately. In general, exact analytical expressions

for partition functions cannot be obtained. However, closed The translational energy levels are very close together in a
approximate expressions can often be found and prove to be container the size of a typical laboratory vessel, in the sense
very important for understanding chemical phenomena; they h2/8mkTX 2 1 (for instance, for H2 in a vessel of length 1 cm
are derived in the following sections and collected at the end of at 298 K, h 2/8mkTX 2 ≈ 4 × 10 −17); therefore, the sum can be
the Topic. approximated by an integral:
∞ ∞
q XT = ∫ e −(n −1) βε dn ≈ ∫ e −n βε dn
2 2
(a) The translational contribution 1 0
The translational partition function for a particle of mass m The extension of the lower limit to n = 0 and the replacement
free to move in a one-dimensional container of length X can of n2 − 1 by n2 introduces negligible error but turns the integral
be evaluated by making use of the fact that the separation of into standard form (Integral G.1 of the Resource section). We
energy levels is very small and that large numbers of states are make the substitution x2 = n2βε, implying dn = dx/(βε)1/2, and
accessible at normal temperatures. As shown in the following therefore that
Justification, in this case
π/2

1/2
1/2 1/2 1/2 1/2

⎛ 2πm ⎞ ⎛ 1 ⎞ ∞
⎛ 1 ⎞ π1/2 ⎛ 2πm ⎞
∫
One- Translational
q XT = ⎜ 2 ⎟ q XT = ⎜ ⎟ e − x dx = ⎜ ⎟ =
2
X X
2 ⎜⎝ h2 β ⎠⎟
(52.7a)
⎝hβ⎠ dimensional partition function ⎝ βε ⎠ 0 ⎝ βε ⎠
It will prove convenient to anticipate once again that β = 1/kT

and to write this expression as q XT = X /Λ, with The total energy of a molecule free to move in three dimen-
sions is the sum of its translational energies in all three directions:
h
Λ= Definition Thermal wavelength (52.7b)
(2πmkT )1/2 ε n n n = ε n( X ) + ε n(Y ) + ε n( Z )
1 2 3 1 2 3
(52.8)
The quantity Λ (uppercase lambda) has the dimensions of where n1, n2, and n3 are the quantum numbers for motion in
length and is called the thermal wavelength (sometimes the the x-, y-, and z-directions, respectively. Therefore, because
‘thermal de Broglie wavelength’) of the molecule. The thermal ea+b+c = eaebec, the partition function factorizes as follows:
wavelength decreases with increasing mass and temperature. (X)
∑e ∑e
(X)
− βε n1 − βε n( 2Y ) − βε n( 3Z ) − βε n1 − βε n( 2Y ) − βε n(3Z )
This expression shows that the partition function for transla- qT = = e e
n1 ,n2 ,n3 n1 ,n2 ,n3
tional motion increases with the length of the box and the mass
of the particle, for in each case the separation of the energy lev- ⎛ ⎞⎛ ⎞⎛ (Z ) ⎞
∑e ∑e ∑e
(X) (Y )
− βε n1 − βε n2 − βε n3
els becomes smaller and more levels become thermally acces- =⎜ ⎟⎜ ⎟⎜ ⎟
⎜⎝ n1
⎟⎠ ⎜⎝ n2
⎟⎠ ⎜⎝ n3
⎟⎠
sible. For a given mass and length of the box, the partition
Atkins09819.indb 509 9/11/2013 12:39:11 PM

That is, The validity of the approximations that led to eqn 52.10 can
be expressed in terms of the average separation, d, of the parti-
q T = q XT q YTq ZT (52.9) cles in the container. Because q is the total number of accessible
states, the average number of states per molecule is qT/N. For
Equation 52.7a gives the partition function for translational this quantity to be large, we require V/NΛ3 1. However, V/N
motion in the x-direction. The only change for the other two is the volume occupied by a single particle, and therefore the
directions is to replace the length X by the lengths Y or Z. average separation of the particles is d = (V/N)1/3. The condi-
Hence the partition function for motion in three dimensions is tion for there being many states available per molecule is there-
fore d3/Λ3 ≫ 1, and therefore d ≫ Λ. That is, for eqn 52.10 to
3/2
⎛ 2πm ⎞ be valid, the average separation of the particles must be much
qT =⎜ 2 ⎟ XYZ (52.10a)
⎝hβ⎠ greater than their thermal wavelength. For H2 molecules at 1 bar
and 298 K, the average separation is 3 nm, which is signifi-
The product of lengths XYZ is the volume, V, of a rectangular cantly larger than their thermal wavelength (71.2 pm).
container, so we can write The validity of eqn 52.10 can be expressed in a different
way by noting that the approximations that led to it are valid if
V Three- Translational many states are occupied, which requires V/Λ3 to be large. That
qT = (52.10b)
Λ3 dimensional partition function will be so if Λ is small compared with the linear dimensions of
the container. For H2 at 25 °C, Λ = 71 pm, which is far smaller
with Λ as defined in eqn 52.7b. As in the one-dimensional case, than any conventional container is likely to be (but comparable
the partition function increases with the mass of the particle (as to pores in zeolites or cavities in clathrates). For O2, a heavier
m3/2) and the volume of the container (as V); for a given mass molecule, Λ = 18 pm.
and volume, the partition function increases with temperature
(as T3/2). As in one dimension, qT → ∞ as T →∞ because an infi-
nite number of states becomes accessible as the temperature is
(b) The rotational contribution
raised. Even at room temperature, qT ≈2 × 1028 for an O2 mol- ( J + 1) with J = 0,
The energy levels of a linear rotor are ε J = hcBJ
ecule in a vessel of volume 100 cm3. 1, 2, … (Topic 41). The state of lowest energy has zero energy,
so no adjustment need be made to the energies given by this
Brief illustration 52.2 expression. Each level consists of 2J + 1 degenerate states.
The translational partition function
Therefore, the partition function of a non-symmetrical (AB)
To calculate the translational partition function of an H2 mol- linear rotor is
ecule confined to a 100 cm3 vessel at 25 °C we use m = 2.016mu;
then ε

J
g
J
( J + 1)
Λ=
6.626 ×10 −34
Js qR = ∑ (2 J + 1) e − β hcBJ
(52.11)
( )
1/2
⎧⎪2π × 2.016 ×1.6605 ×10−27 kg ⎫⎪ J
⎨ ⎬
(
⎪⎩× 1.381 ×10 JK × 298 K
−23 −1
) ( ) ⎪⎭ The direct method of calculating q R is to substitute the experi-
mental values of the rotational energy levels into this expres-
= 7.12 ×10−11 m
sion and to sum the series numerically.
where we have used 1 J = 1 kg m2 s−2. Therefore,
1.00 × 10−4 m3
qT = = 2.77 × 1026 Example 52.1
(7.12 ×10 )
3
−11
m Evaluating the rotational partition
function explicitly
About 1026 quantum states are thermally accessible, even at Evaluate the rotational partition function of 1H35Cl at 25 °C,
room temperature and for this light molecule. Many states given that B = 10.591cm −1.
are occupied if the thermal wavelength (which in this case is
71.2 pm) is small compared with the linear dimensions of the Method We need to evaluate eqn 52.11 term by term. We use
container. kT/hc = 207.224 cm−1 at 298.15 K. The sum is readily evaluated
by using mathematical software.
Self-test 52.2 Calculate the translational partition function
for a D2 molecule under the same conditions. Answer To show how successive terms contribute, we draw up
Answer: qT = 7.8 × 1026, 23/2 times larger the following table by using hcB /kT = 0.05111 (Fig. 52.6):
www.ebook3000.com
Atkins09819.indb 510 9/11/2013 12:39:21 PM

Justification 52.2 The rotational contribution for

J 0 1 2 3 4 … 10
(2J + 1)e−0.05111J(J+1) 1 2.71 3.68 3.79 3.24 … 0.08
linear molecules
When many rotational states are occupied and kT is much
The sum required by eqn 52.11 (the sum of the numbers in the
larger than the separation between neighbouring states, the
second row of the table) is 19.9, hence q R = 19.9 at this tempera-
sum in the partition function can be approximated by an inte-
ture. Taking J up to 50 gives q R = 19.902. Notice that about ten
gral, much as we did for translational motion:
J-levels are significantly populated but the number of popu-
lated states is larger on account of the (2J + 1)-fold degeneracy ∞
∫

of each level. qR = (2 J + 1)e − βhcBJ ( J +1)dJ
0
4
This integral can be evaluated without much effort by making
the substitution x = βhcBJ ( J + 1), so that dx /dJ = βhcB (2 J + 1)
and therefore (2 J + 1) dJ = dx /βhcB . Then
Contribution to q

1

∞
1 1
∫
2
qR = e − x dx =
βhcB 0 βhcB
1
which (because β = 1/kT) is eqn 52.12a.
0
0 1 2 3 4 5 6 7 8 9 10
Quantum number, J
Brief illustration 52.3 The rotational contribution
Figure 52.6 The contributions to the rotational partition
function of an HCl molecule at 25 °C. The vertical axis is For 1 H 35 Cl at 298.15 K we use kT/hc = 207.224 cm−1 and

the value of (2 J +1)e − βhcBJ ( J +1) . Successive terms (which are B =10.591 cm −1 . Then
proportional to the populations of the levels) pass through
a maximum because the population of individual states kT 207.224cm −1
qR = = = 19.59
decreases exponentially, but the degeneracy of the levels hcB 10.591cm −1
increases with J.
The value is in good agreement with the exact value (19.02,
Self-test 52.3 Evaluate the rotational partition function for Example 52.1) and with much less work.
1H35Cl at 0 °C. Self-test 52.4 Evaluate the rotational contribution to the parti-
Answer: 18.26
tion function for 1H35Cl at 0 °C.
Answer: 17.93; exact value 18.26, Self-test 52.3
At room temperature, kT/hc≈ 200 cm−1. The rotational con-

stants of many molecules are close to 1 cm−1 (Table 43.1) and Justification 52.3
The rotational contribution for
often smaller (though the very light H2 molecule, for which nonlinear molecules
B = 60.9 cm −1 , is one exception). It follows that many rota-
The energies of a symmetric rotor (Topic 41) are
tional levels are populated at normal temperatures. When this
is the case, we show in the following two Justifications that the ( J + 1) + hc( A − B )K 2
E J ,K , M J = hcBJ
partition function may be approximated by
with J = 0, 1, 2, …; K = J, J − 1, …, −J; and M J = J, J − 1, …, −J.
kT
qR = (52.12a) Instead of considering these ranges, the same values can be
hcB
Linear rotor
covered by allowing K to range from −∞ to ∞, with J confined
3/2 1/2 to |K|, |K| + 1, …, ∞ for each value of K (Fig. 52.7). Because the
⎛ kT ⎞ ⎛ π ⎞ energy is independent of MJ, and there are 2J + 1 values of MJ
qR =⎜ ⎟ ⎜⎝ ⎟⎠ Nonlinear rotor (52.12b)
⎝ hc ⎠ ABC for each value of J, each value of J is (2J + 1)–fold degenerate. It
follows that the partition function
where A , B , and C are the rotational constants of the mole- ∞ J J
cule expressed as wavenumbers. However, before using these
expressions, read on (to eqns 52.13 and 52.14).
q= ∑∑ ∑ e
J =0 K =− J M J =− J
− βE J ,K , M J
Atkins09819.indb 511 9/11/2013 12:39:48 PM

K = +1, J = 1,2,3,...
K = +2, J = 2,3,4,...
K = –2, J = 2,3,4,...
K = –1, J = 1,2,3,...
A useful way of expressing the temperature above which the
K = 0, J = 0,1,2,...
K = –J K = +J J = |K| J = |K| rotational approximation is valid is to introduce the character-
K
–5 –4 –3 –2 –1 0 +1+2 +3 +4+5 istic rotational temperature, θ R = hcB /k. Then ‘high tempera-
J ture’ means T ≫ θR and under these conditions the rotational
5 partition function of a linear molecule is simply T/θ R. Some
4 typical values of θR are shown in Table 52.1. The value for H2
3 is abnormally high and we must be careful with the approxima-
2
tion for this molecule.
1
0
The general conclusion at this stage is that molecules with
(a) (b) large moments of inertia (and hence small rotational constants
and low characteristic rotational temperatures) have large
Figure 52.7 (a) The sum over J = 0, 1, 2, … and K = J, J − 1, …, −J rotational partition functions. The large value of qR reflects the
(depicted by the circles) can be covered (b) by allowing K to range closeness in energy (compared with kT) of the rotational levels
from −∞ to ∞, with J confined to |K|, |K| + 1, …, ∞ for each value of K. in large, heavy molecules, and the large number of rotational
states that are accessible at normal temperatures.
can be written equivalently as We must take care, however, not to include too many rota-
∞ J ∞ ∞
tional states in the sum. For a homonuclear diatomic molecule or
a symmetrical linear molecule (such as CO2 or HC ≡ CH), a rota-
q= ∑∑ (2 J +1)e
J =0 K =− J
− βE J ,K , M J
= ∑∑(2 J +1)e
K =−∞ J =|K |
− βE J ,K , M J
tion through 180° results in an indistinguishable state of the mol-

∞ ∞ ecule. Hence, the number of thermally accessible states is only
( )
∑e − hc β A − B K 2
∑(2 J +1)e
( J +1)
− hc βBJ
= half the number that can be occupied by a heteronuclear dia-
K =−∞ J =|K | tomic molecule, where rotation through 180° does result in a dis-
tinguishable state. Therefore, for a symmetrical linear molecule,
As in Justification 52.2 we assume that the temperature is so
high that numerous states are occupied and that the sums may kT T Symmetrical Rotational partition
qR = = R (52.13a)
be approximated by integrals. Then
2hcB 2θ linear rotor function
∞ ∞
( ) ( 2 J + 1) e − hc βBJ ( J +1)dJdK
∫ ∫
− hc β A − B K 2
q= e The equations for symmetrical and non-symmetrical molecules
−∞ |K |
can be combined into a single expression by introducing the
symmetry number, σ, which is the number of indistinguish-
As before, the integral over J can be recognized as the integral
of the derivative of a function, which is the function itself, so, able orientations of the molecule. Then
as you should verify, T
qR = Linear rotor Rotational partition function (52.13b)
σθ R
∞
∫ |K |
(2 J + 1) e−hcβBJ ( J +1)dJ = hc1βB e−hcβBK 2
For a heteronuclear diatomic molecule, σ = 1; for a homo-

nuclear diatomic molecule or a symmetrical linear molecule,
We have also supposed that |K| ≫ 1 for most contributions and σ = 2.
replaced |K| (|K| + 1) by K 2. Now we can write The same care must be exercised for other types of symmet-
rical molecule, and for a nonlinear molecule we write
∞ ∞
1 ( ) e 1
∫ ∫
− hc β A − B K 2 − hc βBK
2
q= dK = e − hc βΑK dK
2
e
hc βB hc βB
3/2 1/2
−∞ −∞ 1 ⎛ kT ⎞ ⎛ π ⎞ Rotational
qR = Nonlinear
σ ⎜⎝ hc ⎟⎠ ⎜⎝ ABC
⎟⎠ (52.14)
1/2 partition
rotor
1 ⎛ π ⎞ function
=
hc βB ⎜⎝ hc βΑ ⎟⎠
Table 52.1* Rotational and vibrational temperatures
For the final integration, we have used Integral G.1 of the
Resource section. We conclude that Molecule Mode θV/K θR/K
H2 6330 88
1/2 3/2 1/2
⎛ π ⎞
1 ⎛ kT ⎞ ⎛ π ⎞
q= 3/2 ⎜ 2 ⎟
=⎜ ⎟ ⎜⎝ AB
2 ⎟⎠
HCl 4300 15
(hcβ ) ⎝ AB ⎠ ⎝ hc ⎠ I2 39 0.053
CO2 ν1 1997 0.561
For an asymmetric rotor, one of the B is replaced by C,
to ν2 3380
ν3 960
give eqn 52.12b.
* More value are given in the Resource section; use hc/k = 1.439 K cm.
www.ebook3000.com
Atkins09819.indb 512 9/11/2013 12:40:03 PM

Justification 52.4 nuclear spin of the bosons is 0, then only even-J states are
The origin of the symmetry number
admissible. Because only half the rotational states are occu-
The quantum mechanical origin of the symmetry number is pied, the rotational partition function is only half the value of
the Pauli principle, which forbids the occupation of certain the sum obtained by allowing all values of J to contribute (Fig.
states. We saw in Topic 42, for example, that H 2 may occupy 52.9).
rotational states with even J only if its nuclear spins are paired
(para-hydrogen), and odd J states only if its nuclear spins are
parallel (ortho-hydrogen). There are three states of ortho-H2 to
each value of J (because there are three parallel spin states of
the two nuclei).
Population
To set up the rotational partition function we note that
‘ordinary’ molecular hydrogen is a mixture of one part para-
H2 (with only its even-J rotational states occupied) and three
parts ortho-H2 (with only its odd-J rotational states occupied).
Therefore, the average partition function per molecule is
∑ (2 J +1)e ∑(2 J +1)e

1 ( J +1) 3 ( J +1)
qR = − βhcBJ
+ − βhcBJ
0 1 Rotational quantum number J
4 4
even J odd J
Figure 52.9 The relative populations of the rotational
The odd-J states are more heavily weighted than the even-J energy levels of CO2. Only states with even J values are
states (Fig. 52.8). From the illustration we see that we would occupied. The full line shows the smoothed, averaged
obtain approximately the same answer for the partition func- population of levels.
tion (the sum of all the populations) if each J term contributed
half its normal value to the sum. That is, the last equation can
be approximated as
Some typical values of the symmetry numbers are given in
∑
1
qR = (2 J + 1)e − βhcBJ ( J +1) Table 52.2. For the way that group theory is used to identify the
2
J value of the symmetry number, see Problem 52.9.
and this approximation is very good when many terms con-
tribute (at high temperatures, T 88 K). Table 52.2* Symmetry numbers
The same type of argument may be used for linear symmet-
Molecule σ
rical molecules in which identical bosons are interchanged
by rotation (such as CO2). As pointed out in Topic 42, if the H2O 2
NH3 3
CH4 12
C6H6 12
ortho-H2
* More values are given in the Resource section.
Population
Brief illustration 52.4 The symmetry number

para-H2
The value σ (H 2O) = 2 ref lects the fact that a 180° rotation
about the bisector of the HOH angle interchanges two indis-
tinguishable atoms. In NH 3, there are three indistinguish-
0 1 Rotational quantum number J able orientations around the axis, as shown in 1. For CH4 ,
any of three 120° rotations about any of its four CeH bonds
Figure 52.8 The values of the individual terms leaves the molecule in an indistinguishable state (2), so the

(2 J +1)e − βhcBJ ( J +1) contributing to the mean partition symmetry number is 3 × 4 = 12. For benzene, any of six ori-
function of a 3:1 mixture of ortho- and para-H2. The partition entations around the axis perpendicular to the plane of the
function is the sum of all these terms. At high temperatures, molecule leaves it apparently unchanged (Fig. 52.10), as does
the sum is approximately equal to the sum of the terms over a rotation of 180° around any of six axes in the plane of the
all values of J, each with a weight of 21 . This is the sum of the molecule (three of which pass along each CeH bond with
contributions indicated by the curve.
Atkins09819.indb 513 9/11/2013 12:40:10 PM

10
A F E D
F B E A D F C E
Partition function, qV
E C D B C A B F
C D B C B B C A
B D A C C A D B
5
A E F D D F E A
D F E E F E A F
E C F D A E B F
1
F B A C B D C E 0
A B C D 0 5 10
Temperature, kT/hcν~
Figure 52.10 The 12 equivalent orientations of a benzene
Figure 52.11 The vibrational partition function of a molecule
molecule that can be reached by pure rotations, and give
in the harmonic approximation. Note that the partition
rise to a symmetry number of 12. The six pale colours are the
function is linearly proportional to the temperature when the
underside of the hexagon after that face has been rotated
temperature is high (T ≫ θ V, with θ V = hcν~/k).
into view.
Brief illustration 52.5 The vibrational partition function
the remaining three passing through each Ce C bond in the
plane of the molecule). To calculate the partition function of I2 molecules at 298.15 K
C3 we note from Table 43.1 that their vibrational wavenumber is
C3 D 214.6 cm−1. Then, because, at 298.15 K, kT/hc = 207.224 cm−1, we
have
A B hc␯ 214.6 cm −1
A βε = = = 1.036
B kT 207.244 cm −1
C C
1 2
Then it follows from eqn 52.15 that
1
qV = = 1.55
Self-test 52.5 What is the symmetry number for a naphtha- 1 − e−1.036
lene molecule?
Answer: 3 We can infer that only the ground and first excited states are
significantly populated.
Self-test 52.6 Evaluate the populations of the first three vibra-
tional states.
(c) The vibrational contribution Answer: p 0 = 0.645, p1 = 0.229, p2 = 0.081
The vibrational partition function of a molecule is calculated

by substituting the measured vibrational energy levels into
the exponentials appearing in the definition of qV, and sum- In a polyatomic molecule each normal mode (Topic 44) has
ming them numerically. However, provided it is permissible its own partition function (provided the anharmonicities are so
to assume that the vibrations are harmonic, there is a much small that the modes are independent). The overall vibrational
simpler way. In that case, the vibrational energy levels form a partition function is the product of the individual partition
uniform ladder of separation hc␯ (Topic 43), which is exactly functions, and we can write qV= qV(1)qV(2)…, where qV(K) is
the problem treated in Brief illustration 52.1 and summarized the partition function for the Kth normal mode and is calcu-
in eqn 52.2a. Therefore we can use that result with ε = hc␯ and lated by direct summation of the observed spectroscopic levels.
conclude immediately that
Example 52.2 Calculating a vibrational partition
1 Harmonic Vibrational
q = V
(52.15) function
1− e − βhc␯ approximation partition function
The wavenumbers of the three normal modes of H 2O are
This function is plotted in Fig. 52.11 (which is essentially the 3656.7 cm−1, 1594.8 cm−1, and 3755.8 cm−1. Evaluate the vibra-
same as Fig. 52.1). Similarly, the population of each state is tional partition function at 1500 K.
given by eqn 52.2b.
www.ebook3000.com
Atkins09819.indb 514 9/11/2013 12:40:21 PM

Method Use eqn 52.15 for each mode, and then form the prod- kT High-temperature Vibrational
qV ≈ (52.16)
uct of the three contributions. At 1500 K, kT/hc = 1042.6 cm−1. hc␯ approximation partition function
Answer We draw up the following table displaying the contri- The temperatures for which eqn 52.16 is valid can be
butions of each mode: expressed in terms of the characteristic vibrational tempera-
ture, θ V = hc␯ /k (Table 52.1). The value for H2 is abnormally
Mode: 1 2 3
high because the atoms are so light and the vibrational fre-
␯ /cm −1 3656.7 1594.8 3755.8 quency is correspondingly high. In terms of the vibrational
hc␯ /kT 3.507 1.530 3.602 temperature, ‘high temperature’ means T ≫ θV, and when
qV 1.031 1.276 1.028 this condition is satisfied, qV = T/θV (the analogue of the rota-
tional expression).
The overall vibrational partition function is therefore
q V = 1.031 × 1.276 × 1.028 = 1.352 (d) The electronic contribution

Electronic energy separations from the ground state are usu-
The three normal modes of H 2O are at such high wavenum-
ally very large, so for most cases q E = 1 because only the ground
bers that even at 1500 K most of the molecules are in their
state is occupied. An important exception arises in the case of
vibrational ground state. However, there may be so many nor-
atoms and molecules having electronically degenerate ground
mal modes in a large molecule that their excitation may be
significant even though each mode is not appreciably excited. states, in which case q E = g E, where g E is the degeneracy of the
For example, a nonlinear molecule containing 10 atoms has electronic ground state. Alkali metal atoms, for example, have
3N − 6 = 24 normal modes (Topic 44). If we assume a value of doubly degenerate ground states (corresponding to the two ori-
about 1.1 for the vibrational partition function of one nor- entations of their electron spin), so q E = 2.
mal mode, the overall vibrational partition function is about
qV ≈ (1.1)24 = 9.8, which indicates significant vibrational excita-
tion relative to a smaller molecule, such as H2O.
Brief illustration 52.6 The electronic partition function
Self-test 52.7 Repeat the calculation for CO2, where the vibra-
tional wavenumbers are 1388 cm−1, 667.4 cm−1, and 2349 cm−1, Some atoms and molecules have low-lying electronically
the second being the doubly degenerate bending mode. excited states. An example is NO, which has a configuration
Answer: 6.79 of the form …π1 (Topic 25). The energy of the two degener-
ate states in which the orbital and spin momenta are paral-
lel (giving the 2Π3/2 term, Fig. 52.12) is slightly greater than
In many molecules the vibrational wavenumbers are so great that of the two degenerate states in which they are antipar-
allel (giving the 2Π 1/2 term). The separation, which arises
that βhc␯ >1. For example, the lowest vibrational wavenumber
from spin–orbit coupling, is only 121 cm−1. If we denote the
of CH4 is 1306 cm−1, so βhc␯ = 6.3 at room temperature. Most
energies of the two levels as E1/2 = 0 and E3/2 = ε, the partition
CeH stretches normally lie in the range 2850 to 2960 cm−1, so
function is
for them βhc␯ ≈14 . In these cases, e − βhc␯ in the denominator
of qV is very close to zero (for example, e−6.3 = 0.002), and the
vibrational partition function for a single mode is very close
qE = ∑g e
levels i
i
− βε i
= 2 + 2e − βε
to 1 (qV = 1.002 when βhc␯ = 6.3 ), implying that only the zero-
point level is significantly occupied. This function is plotted in Fig. 52.13. At T = 0, q E = 2, because
Now consider the case of bonds with such low vibrational only the doubly degenerate ground state is accessible. At high
frequencies that βhc␯ 1. When this condition is satisfied, temperatures, q E approaches 4 because all four states are
the partition function may be approximated by expanding the accessible. At 25 °C, q E = 3.1.
exponential (ex = 1 + x + …): Self-test 52.8 A certain atom has a fourfold degenerate ground
state and a sixfold degenerate excited state at 400 cm−1 above
1 1
qV = = the ground state. Calculate its electronic partition function at
1− e − βhc␯
1 − (1 − βhc␯ +) 25 °C.
Answer: 4.87
That is, for weak bonds at high temperatures,
Atkins09819.indb 515 9/11/2013 12:40:32 PM

S
4
2
Π3/2
L
Partition function, q E
S
121.1 cm–1
S 3
L L
S
2
Π1/2 L
2
5 10
0 Temperature, kT/ε
Figure 52.12 The doubly degenerate ground electronic Figure 52.13 The variation with temperature of the
level of NO (with the spin and orbital angular momentum electronic partition function of an NO molecule. Note
around the axis in opposite directions) and the doubly that the curve resembles that for a two-level system (Fig.
degenerate first excited level (with the spin and orbital 52.3), but rises from 2 (the degeneracy of the lower level)
momenta parallel). The upper level is thermally accessible at and approaches 4 (the total number of states) at high
room temperature. temperatures.
☐ 1. The molecular partition function is an indication of ☐ 4. The vibrational partition function of a molecule may
the number of thermally accessible states at the tem- be approximated by that of a harmonic oscillator.
perature of interest. ☐ 5. Because electronic energy separations from the ground
☐ 2. If the energy of a molecule is given by the sum of con- state are usually very big, in most cases the electronic
tributions, then the molecular partition function is a partition function is equal to the degeneracy of the
product of contributions from the different modes. electronic ground state.
☐ 3. The symmetry number takes into account the number of
indistinguishable orientations of a symmetrical molecule.
Molecular partition function

q=
∑e
states i
− βε i Definition, independent molecules 52.1a
q=
∑g e
levels i
i
− βε i Definition, independent molecules 52.1b
Uniform ladder q = 1/(1 − e−βε) 52.2a

Two-level system q = 1 + e−βε 52.3a
Thermal wavelength Λ = h/(2πmkT)1/2 52.7b
Translation qT = V/Λ3 52.10b
Rotation q R = kT /σhcB T ≫ θR, linear rotor, θ R = hcB /k 52.13
q R = (1/σ ) ( kT / hc ) (π / ABC )
1/2
3/2 T ≫ θR, nonlinear rotor 52.14
Vibration q V = 1/(1− e − βhc␯ ) Harmonic approximation 52.15
www.ebook3000.com
Atkins09819.indb 516 9/11/2013 12:40:43 PM

TOPIC 53
Molecular energies
This Topic sets up the basic equations that show how to use the
Contents molecular partition function to calculate the mean energy of a col-
lection of independent molecules. In Topic 58 we see how those
53.1 The basic equations 517
mean energies are used to calculate the thermodynamic property
Brief illustration 53.1: Mean energy of a two-level of ‘internal energy’, and in Topic 60 we extend the discussion to
system 518
the calculation of the other central thermodynamic property, the
53.2 The translational contribution 518 ‘entropy’. The equations for collections of interacting molecules are
Brief illustration 53.2: The equipartition theorem 519 very similar (Topic 54), but much more difficult to implement.
53.3 The rotational contribution 519
Brief illustration 53.3: Mean rotational energy 520
53.4 The vibrational contribution 520 53.1 The basic equations
Brief illustration 53.4: The mean vibrational energy 521
53.5 The electronic contribution 521 We begin by considering a collection of N molecules that do
Example 53.1: Calculating the electronic contribution not interact with one another. Any member of the collection
to the energy 521 can exist in a state i of energy εi measured from the lowest
Brief illustration 53.5: The spin contribution to the energy state of the molecule. The mean energy of a molecule,
energy 522 〈ε〉, relative to its energy in its ground state, is the total energy
Checklist of concepts 522 of the collection, E, divided by the total number of molecules:
∑N ε
E 1
〈ε 〉 = = i i (53.1)
N N
i
In Topic 51 it is shown that the overwhelmingly most probable

population of a state in a collection at a temperature T is given
➤ Why do you need to know this material? by the Boltzmann distribution (eqn 51.6, N i /N = (1/ q )e − βε ), so ι
we can write
The partition function contains thermodynamic information,
but it needs to be extracted. Here we show how to extract
∑ε e
1
one particular property: the average energy of molecules, 〈ε 〉 = i
− βε i
(53.2)
q
which plays a central role in thermodynamics. i
➤ What is the key idea? with β = 1/kT. To manipulate this expression into a form involving
only q we note that
The average energy of a molecule in a collection of
independent molecules can be calculated from the
d − βε
molecular partition function. ε i e − βε = −
i
e i
dβ
➤ What do you need to know already?
It follows that1
You need know how to calculate the molecular partition
function from calculated or spectroscopic data (Topic
∑ dβ e ∑e
1 d 1 d 1 dq
52) and its significance as a measure of the number of 〈ε 〉 = − − βε i
=− − βε i
=−
q q dβ q dβ (53.3)
accessible states. The Topic also draws on expressions i i
for the rotational and vibrational energies of molecules 1 To reinforce the analogy between statistical thermodynamics and quan-
(Topics 41 and 43). tum mechanics, note the resemblance of eqn 53.3 written as 〈ε 〉q = − dq /dβ
to the time-dependent Schrödinger equation written as HΨ = −dΨ /d(it / ).
Atkins09819.indb 517 9/11/2013 12:40:53 PM

There are several points that need to be made in relation to

0.6
eqn 53.3. Because ε0 = 0 (remember that we measure all ener- 0.4
gies from the lowest available level), 〈ε〉 should be interpreted 0.5
as the value of the mean energy relative to its ground-state
Mean energy, 〈ε〉/ε

0.4
(zero-point) energy. If the lowest energy of the molecule is in
fact εgs rather than 0, then the true mean energy is εgs + 〈ε〉. For
0.2
instance, for a harmonic oscillator, we would set εgs equal to
the zero-point energy, 12 hc␯ . Secondly, because the partition 0.2
function may depend on variables other than the tempera-
ture (for example, the volume), the derivative with respect to
β in eqn 53.3 is actually a partial derivative with these other
0 0
variables held constant. The complete expression relating the 0 0.5 1 0 5 10
Temperature, kT/ε Temperature, kT/ε
molecular partition function to the mean energy of a molecule
is therefore Figure 53.1 The mean energy of a two-level system
(expressed as a multiple of ε) as a function of
1 ⎛ ∂q ⎞ temperature, on two temperature scales. The graph on
〈ε 〉 = ε gs − ⎜ ⎟ Mean molecular energy (53.4a) the left shows the slow rise away from zero energy at
q ⎝ ∂β ⎠ V
low temperatures; the slope of the graph at T = 0 is 0. The
graph on the right shows the slow rise to 0.5 as T → ∞ as
An equivalent form is obtained by noting that dx/x = d ln x: both states become equally populated.
⎛ ∂ lnq ⎞
〈ε 〉 = ε gs − ⎜
⎝ ∂β ⎟⎠ V
Mean molecular energy (53.4b) 53.2 The translational contribution
For a one-dimensional container of length X, for which
These two equations confirm that we need know only the par- qT = X/Λ with Λ = h(β/2πm)1/2 (Topic 52), we note that Λ is a
tition function (as a function of temperature) to calculate the constant multiplied by β1/2, and obtain
mean energy.
1 ⎛ ∂q T ⎞ Λ⎛ ∂ X⎞
〈ε T 〉 = − =− ⎜
q T ⎜⎝ ∂β ⎟⎠ V X ⎝ ∂β Λ ⎟⎠ V
constant × β 1/2 d ⎛ 1 ⎞
=− ×X×
Brief illustration 53.1 Mean energy of a two-level system X dβ ⎜⎝ constant × β 1/2 ⎟⎠
1
If a molecule has only two available energy levels, one at 0 and − β −3/2

2
the other at an energy ε, its partition function is d 1 1
= −β 1/2 =
dβ β 1/2 2 β
q = 1 + e − βε
Therefore, the mean energy of a collection of these molecules That is,

at a temperature T is
〈ε T 〉 = 12 kT One dimension Mean translational energy (53.5a)
1 d(1 + e − βε
) εe − βε
ε
〈ε 〉 = − = =
1 + e − βε dβ 1 + e − βε e βε + 1
For a molecule free to move in three dimensions, the analogous
calculation leads to
This function is plotted in Fig. 53.1. Notice how the mean Mean
Three
energy is zero at T = 0, when only the lower state (at the zero 〈ε T 〉 = 23 kT dimensions
translational (53.5b)
of energy) is occupied, and rises to 12 ε as T → ∞ , when the two energy
levels become equally populated.
The equipartition theorem of classical mechanics is consistent
Self-test 53.1 Deduce an expression for the mean energy when with this result and provides a useful shortcut. First, we need to
each molecule can exist in states with energies 0, ε, and 2ε. know that a ‘quadratic contribution’ to the energy means a con-
Answer: 〈ε〉 = ε (1 + 2x)x/(1 + x + x 2), x = e−βε tribution that can be expressed as the square of a variable, such
as the position or the velocity. For example, the kinetic energy
www.ebook3000.com
Atkins09819.indb 518 9/11/2013 12:41:07 PM

53 Molecular energies 519
of an atom of mass m as it moves through three-dimensional sums can be replaced by integrals. The average value of any
space is term that depends on x 2 is therefore
Ek = 12 mvx2 + 12 mv 2y + 12 mvz2
∞ ∞
e − βε ( x , p)
〈x2 〉 = ∫ ∫
−∞ −∞
x2
q
dxdp
and there are three quadratic contributions to its energy. In its
simplest form, and as developed in the following Justification,
The term in blue is the Boltzmann factor and the partition
the equipartition theorem then states that:
function is
For a collection of particles at thermal equilibrium
at a temperature T, the average value of each quadratic ∞ ∞
contribution to the energy is the same and equal to

1
q= ∫ ∫
−∞ −∞
e − βε ( x , p)dxdp
2 kT.
where k is Boltzmann’s constant. (For the molar energy, mul- If the energy has the form ε(x, p) = ap2 + bx 2 (for instance, a free
tiply by Avogadro’s constant and use NAk = R.) The equiparti- particle has a = 1/2m and b = 0), then
tion theorem is a conclusion from classical mechanics and 1
( π/b3 β 3 )1/2
is applicable only when the effects of quantization can be
2

ignored. In practice, it can be used for molecular translation ∞ ∞ ∞
and rotation for all but the lightest of molecules, but not for 〈x 〉 =
2 ∫ xe−∞
2 − bβx 2
dx ∫ e
−∞
− aβp2
dp
=
∫ x e dx =
−∞
2 − bβx 2
1 ⎛ 1 ⎞
⎜⎝ bβ ⎟⎠
∞ ∞ ∞ 2
∫ e dx ∫ e ∫ed x
vibration. − bβx 2
− aβp 2
dp − bβx 2
−∞ −∞ −∞
A note on good practice You will commonly see the equipar- ( π/bβ )1/2
tition theorem expressed in terms of the ‘degrees of freedom’
rather than quadratic contributions. That can be misleading
and is best avoided, for a single vibrational degree of freedom where we have used Integrals G.1 and G.3 from the Resource
has two quadratic contributions (the kinetic energy and the section. It follows that the average value of a contribution
potential energy). to the energy of the form bx 2 is 1/2β = 12 kT A similar argu-
ment applies for the average value of p2 and its analogues (for
instance, angular momentum).
Brief illustration 53.2 The equipartition theorem
According to the equipartition theorem, the average energy
of each term in the expression above for Ek is 12 kT. Therefore,
the mean energy of the atoms is 23 kT and their molar energy
53.3 The rotational contribution
is 23 RT. At 25 °C, 23 RT = 3.7 kJmol−1 , so translational motion
contributes about 4 kJ mol−1 to the molar internal energy of The mean rotational energy of a linear molecule is obtained
a gaseous sample of atoms or molecules (the remaining con- from the rotational partition function (eqn 52.11):
tribution arises from the internal structure of the atoms and
molecules). qR = ∑(2J +1)e
J
( J +1)
− βhcBJ
Self-test 53.2 Estimate the mean molar rotational energy of a

linear molecule. When the temperature is low (in the sense T < θ R), the series
Answer: RT; see Brief illustration 53.3
must be summed term by term, which for a heteronuclear
diatomic molecule or other non-symmetrical linear molecule
gives

Justification 53.1 q R = 1+ 3e −2 βhcB + 5e −6 βhcB +
The equipartition theorem
We consider a one-dimensional system; the approach is eas- Hence, because
ily generalized to more dimensions. In general, the energy
depends on the position of a particle, x, and the linear momen-
dq R
= −hcB (6e −2 βhcB + 30e −6 βhcB +)

tum, p, both of which can range from −∞ to +∞. Both quanti-
ties are continuously variable (this is a classical system), so any dβ
Atkins09819.indb 519 9/11/2013 12:41:22 PM

(q R is independent of V, so the partial derivative has been Brief illustration 53.3 Mean rotational energy
replaced by a complete derivative), we find
To estimate the mean energy of a nonlinear molecule we rec-
1 dq R ognize that its rotational kinetic energy (the only contribu-
〈ε R 〉 = − R
q dβ tion to its rotational energy) is Ek = 12 I aω a2 + 12 Ibω b2 + 12 I cω c2 .
hcB (6e −2 βhcB + 30e −6 βhcB +)
As there are three quadratic contributions, its mean rotational
= energy is 23 kT. The molar contribution is 23 RT. At 25 °C, this
1+ 3e −2 βhcB + 5e −6 βhcB +
contribution is 3.7 kJ mol−1, the same as the translational con-
Unsymmetrical Mean tribution, for a total of 7.4 kJ mol−1. A monatomic gas has no
linear molecule rotational (53.6a)
energy rotational contribution.
Self-test 53.3 How much energy does it take to raise the tem-
This ungainly function is plotted in Fig. 53.2. At high tempera- perature of 1.0 mol H 2O(g) from 100 °C to 200 °C? Consider
tures (T ≫ θR), q R is given by eqn 52.13 in the form only translational and rotational contributions to the mean
energy.
1 Answer: 2.5 kJ
qR =
σβhcB
where σ = 1 for a heteronuclear diatomic molecule. It then fol-

lows that

−1/β 2 53.4 The vibrational contribution
1 dq d R
1 d 1
〈ε R 〉 = − R = −σβhcB = −β The vibrational partition function in the harmonic approxima-
q dβ dβ σβhcB dβ β
tion is given in eqn 52.15:
and therefore that

1
qV =
1 Linear molecule, Mean 1− e − βhc␯
〈ε R 〉 = = kT high temperature rotational (53.6b)
β (T ≫ θR) energy
Because qV is independent of the volume, it follows that
The high-temperature result, which is valid when many rota-
tional states are occupied, is also in agreement with the equi-
dq V d 1 hc␯ e − βhc␯
partition theorem, because the classical expression for the = =− (53.7)
dβ dβ 1 − e − βhc␯
(1− e − βhc␯ )2
energy of a linear rotor is Ek = 12 I ⊥ω a2 + 12 I ⊥ω b2 and therefore
has two quadratic contributions. (There is no rotation around
the line of atoms.) It follows from the equipartition theorem and hence from
that the mean rotational energy is 2 × 12 kT = kT.
1 dq V hc␯ e − βhc␯
〈ε V 〉 = − = (1− e − βhc␯ )
1.5 q dβ
V
(1− e − βhc␯ )2
~
hc␯ e − βhc␯
Mean energy, 〈ε R〉/hcB

=
1− e − βhc␯
1
that
0.5
hc␯ Mean
〈ε V 〉 = βhc␯
Harmonic vibrational (53.8)
0 e −1 approximation energy
0 1 2
Temperature, T/θ R
Figure 53.2 The mean rotational energy of a non-symmetrical The zero-point energy, 12 hc␯ , can be added to the right-hand
linear rotor as a function of temperature. At high temperatures side if the mean energy is to be measured from 0 rather than
(T ≫θ R), the energy is linearly proportional to the temperature, the lowest attainable level (the zero-point level). The variation
in accord with the equipartition theorem. of the mean energy with temperature is illustrated in Fig. 53.3.
www.ebook3000.com
69_Atkins_Ch53.indd 520 9/11/2013 2:04:45 PM

10
The addition of the zero-point energy (corresponding to
1
2
× 214.6 cm−1 ) increases this value to 225.3 cm−1. The equipar-
Mean energy, 〈ε V〉/hc ~
ν
tition result is 207.224 cm−1 the discrepancy reflecting the fact

that in this case it is not true that T ≫θV and (as deduced in
Brief illustration 52.5) only the ground and first excited states
5
are significantly populated.
Self-test 53.4 What must the temperature be before the energy
estimated from the equipartition theorem is within 2 per cent
of the energy given by eqn 53.8?
0 Answer: 625 K; use a spreadsheet
5 10
0 Temperature, T/θ V
Figure 53.3 The mean vibrational energy of a molecule in When there are several normal modes that can be treated as
the harmonic approximation as a function of temperature.
harmonic, the overall vibrational partition function is the prod-
At high temperatures (T ≫θV ), the energy is linearly
uct of each individual partition function, and the total mean
proportional to the temperature, in accord with the
vibrational energy is the sum of the mean energies of each mode.
equipartition theorem.
53.5 The electronic contribution

At high temperatures, when T θV, or βhc␯ 1 (from Topic
52, θ V = hc␯ /k ), the exponential functions can be expanded We shall consider two types of electronic contribution: one
(ex = 1 + x + …) and all but the leading terms discarded. This arising from the electronically excited states of a molecule and
approximation leads to one from the spin contribution.
In most cases of interest, the electronic states of atoms and
molecules are so widely separated that only the electronic
hc␯ High-
Mean
〈ε V 〉 = temperature ground state is occupied. As we are adopting the convention
(1+ βhc␯ +) −1 approximation
vibrational (53.9)
that all energies are measured from the ground state of each
energy
1 (T ≫θ V)
≈ = kT mode, we can write
β
〈ε E 〉 = 0 Mean electronic energy (53.10)
This result is in agreement with the value predicted by the In certain cases, there are thermally accessible states at the
classical equipartition theorem, because the energy of a one- temperature of interest. In that case the partition function and
dimensional oscillator is E = 12 mvx2 + 12 kf x 2 and the mean energy hence the mean electronic energy are best calculated by direct
of each quadratic term is is 12 kT. Bear in mind, however, that summation over the available states. Care must be taken to take
the condition T θV is rarely satisfied. any degeneracies into account, as we illustrate in Example 53.1.
Example 53.1 Calculating the electronic contribution to

the energy
Brief illustration 53.4 The mean vibrational energy
A certain atom has a doubly degenerate electronic ground
To calculate the mean vibrational energy of I2 molecules at state and a fourfold degenerate excited state at 600 cm−1 above
298.15 K we note from Table 43.1 that their vibrational wavenum- the ground state. What is its mean electronic energy at 25 °C
ber is 214.6 cm−1. Then, because, at 298.15 K, kT/hc = 207.224 cm−1, expressed as a wavenumber?
from eqn 53.8 with
Method Write down the expression for the partition func-
hc␯ 214.6 cm −1 tion at a general temperature T (in terms of β ) and then
βε = = = 1.036
kT 207.244 cm −1 derive the mean energy by differentiating with respect to β.
Finally, substitute the data. Use ε = hc␯ , 〈ε E 〉 = hc 〈 ␯ E 〉, and
it follows that kT/hc = 207.226 cm−1 at 25 °C.
Answer The partition function is
214.6 cm −1
〈ε V 〉/hc = = 118.0 cm −1
e1.036 −1 q E = 2 + 4e − βε
Atkins09819.indb 521 9/11/2013 12:41:54 PM

The mean energy of the spin is therefore

The mean energy is therefore
−4
εe
− βε
−2 βμB B
−2 μ Be
1 dq E
1 d
B

〈ε E 〉 = − =− (2 + 4e − βε )
q E dβ 2 + 4e − βε dβ 1 dq S 1 d
〈ε S 〉 = − =− (1+ e −2 βμ B
) B
4εe − βε ε q S dβ 1+ e −2 βμ BB dβ
= = 1 βε
2 + 4e − βε
e +1 2 μ Be −2 βμ B B
2 = B −2 μ B
1+ e B
and expressed as a wavenumber

That is,
␯
〈 ␯ E 〉 = 1
2
ehc␯ /kT + 1
2 μ BB
From the data, 〈ε S 〉 = (53.13)
e2 βμ B + 1
B Mean spin energy
600cm −1
〈 ␯ E 〉 = = 59.7cm −1
1
2 e600/207.226 + 1 This function is essentially the same as that plotted in Fig. 53.1.
Self-test 53.5 Repeat the problem for an atom that has a

threefold degenerate ground state and a sevenfold degenerate
excited state 400 cm−1 above.
Answer: 101 cm−1 Brief illustration 53.5 The spin contribution to the
energy
Suppose a collection of radicals is exposed to a magnetic field
An electron spin in a magnetic field B has two possible of 2.5 T (T denotes tesla). With μ B = 9.274 ×10 −24 J T−1 and a
energy states that depend on its orientation as expressed by the temperature of 25 °C,
magnetic quantum number ms and which are given by
2μBB = 2 × (9.274 × 10−24 JT−1 ) × 2.5T = 4.6…× 10−23 J
2 × (9.274 ×10−24 JT−1 ) (2.5 T)
Em = 2μBBms Electron spin energies (53.11) 2 βμBB = = 0.011…
s
(1.381 ×10−23 JK −1 ) × (298 K )
where μB is the Bohr magneton (see inside the front cover). The mean energy is therefore
These energies are discussed in more detail in Topic 50, where
4.6…× 10−23 J
we see that the integer 2 needs to be replaced by a number very 〈ε S 〉 = = 2.3 × 10−23 J
e0.011… + 1
close to 2. The lower state has ms = − 12 , so the two energy lev-
els available to the electron lie (according to our convention) This energy is equivalent to 14 J mol −1 (note joules, not
at ε−1/2 = 0 and at ε+1/2 = 2μBB. The spin partition function is kilojoules).
therefore
Self-test 53.6 Repeat the calculation for a species with S = 1 in
the same magnetic field.
qS = ∑e
ms
− βε ms
= 1+ e −2 βμ B
B Spin partition function (53.12) Answer: 0.0046 zJ, 28 J mol−1
☐ 1. The mean molecular energy can be calculated from the ☐ 2. The (classical) equipartition theorem states that, for a
molecular partition function (see the following list of collection of particles at thermal equilibrium at a tem-
equations). perature T, the average value of each quadratic contribu-
tion to the energy is the same and equal to 12 kT.
www.ebook3000.com
Atkins09819.indb 522 9/11/2013 12:42:02 PM

Mean energy 〈ε 〉 = ε gs − (1/q )(∂q /∂β )V 53.4a
〈ε 〉 = ε gs − (∂ ln q /∂β )V Alternative version 53.4b
Translation 〈ε T 〉 = d2 kT In d dimensions 53.5
Rotation 〈ε R 〉 = kT Linear molecule, T θR 53.6b
〈ε R 〉 = 23 kT Nonlinear molecule, T θR; see Brief illustration 53.3

Vibration 〈ε V 〉 = hc␯ /(e βhc␯ −1) Harmonic approximation 53.8
〈ε V 〉 = kT T θV 53.9
Atkins09819.indb 523 9/11/2013 12:42:08 PM

TOPIC 54
The canonical ensemble

Topic 52 deals with a system composed of independent mole-
Contents cules. Here we consider the formalism appropriate to systems in
54.1 The concept of ensemble 524
which the molecules interact with one another, as in real gases
(a) Dominating configurations 525 and liquids. The crucial concept we need when treating systems
Brief illustration 54.1: The canonical distribution 525
of interacting particles is the ‘ensemble’. Like so many scientific
(b) Fluctuations from the most probable distribution 526 terms, the term has basically its normal meaning of ‘collection’,
Brief illustration 54.2: The role of the density
but it has been sharpened and refined into a precise significance.
of states 526 The expressions presented in this Topic are the analogues of
54.2 The mean energy of a system 526
those derived in Topics 52 and 53, and details of the calcula-
tions are found there. The implementation of the equations,
Brief illustration 54.3: The expression for the energy 527
however, is much more difficult, and except in certain cases is
54.3 Independent molecules revisited 527
best left to specialized texts. This Topic is included principally
Brief illustration 54.4: Indistinguishability 527
for the sake of completeness and as the endpoint of our journey
54.4 The variation of energy with volume 528 laying the foundations of statistical thermodynamics.
Brief illustration 54.5: A configuration integral 528
Checklist of concepts 529
54.1 The concept of ensemble
To set up an ensemble, we take a closed system of specified
volume, composition, and temperature and think of it as rep-
➤ Why do you need to know this material? licated N times (Fig. 54.1). All the identical closed systems
Whereas the derivation of the molecular partition function are regarded as being in thermal contact with one another, so
(Topic 52) deals with independent molecules, in practice, they can exchange energy. The total energy of all the systems
molecules do interact. So, this material is essential for is E and, because they are in thermal equilibrium with one
constructing models of real gases, liquids, and solids and another, they all have the same temperature, T. The volume
of any system in which intermolecular interactions cannot of each member of the ensemble is the same, so the energy
be neglected. levels available to the molecules are the same in each system,
and each member contains the same number of molecules,
➤ What is the key idea? so there is a fixed number of molecules to distribute within
A system composed of interacting molecules is described each system. This imaginary collection of replications of
in terms of a canonical partition function, from which its the actual system with a common temperature is called the
thermodynamic properties may be deduced. canonical ensemble. The word ‘canonical’ means ‘according
to a rule’.
➤ What do you need to know already? There are two other important ensembles. In the micro-
This material draws on the calculations in Topic 52: the canonical ensemble the condition of constant temperature is
calculations here are analogous to those, and are not replaced by the requirement that all the systems should have
repeated in detail. This Topic also draws on the calculation exactly the same energy: each system is individually isolated.
of energies from partition functions (Topic 53); here too the In the grand canonical ensemble the volume and temperature
calculations are analogous to those presented in that Topic. of each system is the same, but they are open, which means that
matter can be imagined as able to pass between the systems; the
www.ebook3000.com
Atkins09819.indb 524 9/11/2013 12:42:09 PM

54 The canonical ensemble 525
limit of N → ∞, this dominating configuration is overwhelm-

ingly the most probable, and it dominates the properties of the
system virtually completely.
N, The quantitative discussion follows the argument in Topic 51
8 9 10
with the modification that N and Ni are replaced by N and N i .
V, 6 7
3 4 5
T 1 2
Ener
gy The weight W of a configuration {N 0 , N 1 ,…} is
N, 19 20 N !
17 18 W = Weight (54.1)
N1 ! N 2 !…
V, 15 16
T 12 13 14
11
The configuration of greatest weight, subject to the constraints
Figure 54.1 A representation of the canonical ensemble, in this that the total energy of the ensemble is constant at E and that
case for N = 20. The individual replications of the actual system the total number of members is fixed at N, is given by the
all have the same composition and volume. They are all in canonical distribution:
mutual thermal contact, and so all have the same temperature.
Energy may be transferred between them as heat, and so N i e − βE i
they do not all have the same energy. The total E of all 20 = Canonical distribution (54.2a)
N Q
replications is a constant because the ensemble is isolated
overall. where
composition of each one may fluctuate, but now the property Q = ∑e − βEi
Canonical partition function (54.2b)
known as the chemical potential, μ (which is described in Topic i
69), is the same in each system: in which the sum is over all members of the ensemble, each one
having an energy Ei. The quantity Q , which is a function of the
Microcanonical ensemble: N, V, E common temperature, is called the canonical partition function. Like
Canonical ensemble: N, V, T common the molecular partition function, the canonical partition func-
Grand canonical ensemble: μ, V, T common tion contains all the thermodynamic information about a sys-
tem, but in this case allowing for the possibility of interactions
The microcanonical ensemble is the basis of the discussion in between the constituent molecules.
Topic 51 (but not so called there); we shall not consider the
grand canonical ensemble explicitly. Brief illustration 54.1 The canonical distribution
The important point about an ensemble is that it is a collec-
tion of imaginary replications of the system, so we are free to let Suppose that we are considering a sample of a monatomic
the number of members be as large as we like; when appropri- real gas that contains 1.00 mol atoms. Then at 298 K its total
ate, we can let N become infinite. The number of members of energy is close to 23 nRT = 23 (1.00 mol) × 8.3145 J K −1mol −1 ) ×
the ensemble in a state with energy Ei is denoted N i , and we can (298K ) = 3.72 kJ. Suppose that for an instant the molecules are
speak of the configuration of the ensemble (by analogy with the present at separations where the total energy is 3.72 kJ and an
configuration of the system used in Topic 51) and its weight, instant later are present at separations where the total energy is
Note that N is unrelated to N, the number of molecules in
W. lower than 3.72 kJ by 0.000 000 01 per cent (that is, by 3.72 × 10−7 J).
the actual system; N is the number of imaginary replications of To predict the ratio of numbers of members of the ensemble
with these two energies we use eqn 54.2 in the form
that system.
N (lower ) −( −3.70 ×10−7 J)/(1.381×10−23 JK −1 ) × (298 K )
=e
(a) Dominating configurations N (higher )
= e3.33×10
7
Just as in Topic 51, some of the configurations of the canoni-

cal ensemble will be very much more probable than others. For At first sight, the number of members that have the lower

instance, it is very unlikely that the whole of the total energy, E, energy vastly outweighs the number with the higher energy.
Why that is not necessarily the case is explained below.
will accumulate in one system. By analogy with the discussion
in Topic 51, we can anticipate that there will be a dominating Self-test 54.1 Repeat the calculation for members of the same
configuration, and that we can evaluate the thermodynamic ensemble with energies that differ by 1.0 ×10−20 per cent.
properties by taking the average over the ensemble using that Answer: N (lower )/N (higher ) = e90 ≈ 1×1039
single, most probable, configuration. In the thermodynamic
Atkins09819.indb 525 9/11/2013 12:42:26 PM

(b) Fluctuations from the most probable Brief illustration 54.2 The role of the density of states
distribution
A function that increases rapidly is x N, with N a large value.
The canonical distribution in eqn 54.2 is only apparently an A function that decreases rapidly is e−Nx, once again with N a
exponentially decreasing function of the energy of the system. large value. The product of these two functions, normalized so
We must appreciate that the equation gives the probability that their maxima coincide,
of occurrence of members in a single state i of the entire sys-
f (x )= e N x N e − Nx
tem of energy Ei. There may in fact be numerous states with
almost identical energies. For example, in a gas the identities is plotted for three values of N in Fig. 54.4. We see that the
of the molecules moving slowly or quickly can change without width of the product does indeed decrease as N increases.
necessarily affecting the total energy. The density of states, the
number of states in an energy range divided by the width of the 1
range (Fig. 54.2), is a very sharply increasing function of energy.
It follows that the probability of a member of an ensemble hav- 0.8
ing a specified energy (as distinct from being in a specified 10
state) is given by eqn 54.2, a sharply decreasing function, multi- 0.6
20
plied by a sharply increasing function (Fig. 54.3). Therefore, the f(x)
50
overall distribution is a sharply peaked function. We conclude 0.4
that most members of the ensemble have an energy very close
to the mean value. 0.2
0
0 0.5 1 1.5 x 2 2.5 3
Figure 54.4 The product of the two functions discussed in

Energy
Width of
range
Brief illustration 54.2, for three different values of N.
Self-test 54.2 Show that the product of the functions x 2N and

Number of e−Nx, suitably normalized, behaves similarly.
states
Answer: Plot f(x) = (1/2)2N e2N x 2N e − Nx for 0 ≤ x ≤ 4
Figure 54.2 The energy density of states is the number of

states in an energy range divided by the width of the range.
54.2 The mean energy of a system
Probability Number of Just as the molecular partition function can be used to calcu-
of state states
late the mean value of a molecular property, so the canonical
Probability partition function can be used to calculate the mean energy of
of energy an entire system composed of molecules (that might or might
not be interacting with one another). Thus, Q is more general
than q because it does not assume that the molecules are inde-
pendent. We can therefore use Q to discuss the properties of
condensed phases and real gases where molecular interactions
are important.
Energy and there are N mem-
If the total energy of the ensemble is E,
Figure 54.3 To construct the form of the distribution of bers, the average energy of a member is 〈 E 〉 = E / N . Because the
members of the canonical ensemble in terms of their energies, fraction, pi , of members of the ensemble in a state i with energy
we multiply the probability that any one is in a state of given Ei is given by the analogue of eqn 51.6 (N i /N = e − βε / ∑ e − βε i i
i
energy, eqn 54.2, by the number of states corresponding to written as p i = e − βε /q ) as
i
that energy (a steeply rising function). The product is a sharply

peaked function at the mean energy, which shows that almost e − βE
i
pi = (54.3)
all the members of the ensemble have that energy. Q
www.ebook3000.com
Atkins09819.indb 526 9/11/2013 12:42:38 PM

it follows that Q= ∑e i
− βε i (1) − βε i (2 ) − − βε i ( N )
∑ ∑
1
〈E〉 = p i Ei = Ei e − βE
i
(54.4)
Q The sum over the states of the system can be reproduced by let-
i i
ting each molecule enter all its own individual states (although
By the same argument that led to eqn 53.4, we meet an important proviso shortly). Therefore, instead of
summing over the states i of the system, we can sum over all
1 ⎛ ∂Q ⎞ ⎛ ∂ lnQ ⎞ the individual states j of molecule 1, all the states j of molecule
〈E〉 = − = −⎜
Q ⎜⎝ ∂β ⎟⎠ V ⎝ ∂β ⎟⎠ V
Mean energy of a system (54.5) 2, and so on. This rewriting of the original expression leads to
⎛ ⎞⎛ ⎞ ⎛ ⎞
As in the case of the mean molecular energy, we must add to
this expression the ground-state energy of the entire system if
Q =⎜
⎜⎝ ∑ j
e − βε j ⎟ ⎜
⎟⎠ ⎜⎝ ∑j
e − βε j ⎟ ⎜
⎟⎠ ⎜⎝ ∑
j
e − βε j ⎟ = q N
⎟⎠
that is not zero.
Brief illustration 54.3 The expression for the energy If all the molecules are identical and free to move through
space, we cannot distinguish them and the relation Q = q N is not
If the canonical partition function is a product of the molecu-
valid. Suppose that molecule 1 is in some state a, molecule 2 is
lar partition function of each molecule (which we see below
in b, and molecule 3 is in c, then one member of the ensemble
is the case when the N molecules of the system are independ-
has an energy E = εa + εb + εc. This member, however, is indis-
ent), then we can write Q = q N, and infer that the energy of the
tinguishable from one formed by putting molecule 1 in state b,
system is
molecule 2 in state c, and molecule 3 in state a, or some other
1 ⎛ ∂q N ⎞ Nq N −1 ⎛ ∂q ⎞ N ⎛ ∂q ⎞ permutation. There are six such permutations in all, and N! in
〈E〉 = − =− N ⎜ =− ⎜ = N 〈ε 〉
N ⎜ ∂β ⎟
q ⎝ ⎠V q ⎝ ∂β ⎟⎠ V q ⎝ ∂β ⎟⎠ V general. In the case of indistinguishable molecules, it follows
that we have counted too many states in going from the sum
That is, the mean energy of the system is N times the mean over system states to the sum over molecular states, so writing
energy of a single molecule. Q = q N overestimates the value of Q. The detailed argument is
Self-test 54.3 Confirm that the same expression is obtained if
quite involved, but at all except very low temperatures it turns
Q = qN/N!, which is another case described below. out that the correction factor is 1/N!. Therefore:
For distinguishable independent molecules :Q = q N (54.7a)
54.3 Independent molecules revisited For indistinguishable independent molecules : Q = q N /N !

(54.7b)
We shall now see how to recover the molecular partition func-
tion from the more general canonical partition function when At low temperatures, where only a few states are thermally acces-
the molecules are independent. When the molecules are inde- sible, account must be taken of the role of the Pauli principle
pendent and distinguishable (in the sense to be described), the in governing the occupation of states. This is the role of Bose–
relation between Q and q is Einstein statistics (for bosons) and Fermi–Dirac statistics (for
fermions), both of which are beyond the scope of this discussion.
Q =q N (54.6)
Brief illustration 54.4 Indistinguishability

Justification 54.1
The relation between Q and q For molecules to be indistinguishable, they must be of the
same kind: an Ar atom is never indistinguishable from a Ne
The total energy of a collection of N independent molecules
atom. Their identity, however, is not the only criterion. Each
is the sum of the energies of the molecules. Therefore, we can
identical molecule in a crystal lattice, for instance, can be
write the total energy of a state i of the system as
‘named’ with a set of coordinates. Identical molecules in a lat-
Ei = ε i (1) + ε i (2) +… + ε i (N ) tice can therefore be treated as distinguishable because their
sites are distinguishable, and we use eqn 54.7b. On the other
In this expression, εi(1) is the energy of molecule 1 when the hand, identical molecules in a gas are free to move to different
system is in the state i, εi(2) the energy of molecule 2 when the locations, and there is no way of keeping track of the identity
system is in the same state i, and so on. The canonical parti- of a given molecule; we therefore use eqn 54.7b.
tion function is then
Atkins09819.indb 527 9/11/2013 12:42:46 PM

In the second line, we have used the relation (∂ 2f/∂x∂y) =

Self-test 54.4 A re ident ic a l mole c u le s i n a l iqu id
indistinguishable? (∂ 2f/∂y∂x) and then noted that Λ is independent of volume, so its
Answer: Yes
derivative with respect to volume is zero.
For a real gas of atoms (for which the intermolecular inter-
actions are isotropic), Z is related to the total potential energy
EP of interaction of all the particles, which depends on all their
relative locations, by
The variation of energy
54.4
with volume Z=
1
N! ∫
e − βE dτ 1dτ 2 …dτ N
p Configuration integral (54.11)
When there are interactions between molecules, the energy of where dτi is the volume element for atom i. The physical ori-
a collection depends on the average distance between them, gin of this term is that the probability of occurrence of each
and therefore on the volume that a fixed number occupy. This arrangement of molecules possible in the sample is given by a
dependence on volume is particularly important for the discus- Boltzmann distribution in which the exponent is given by the
sion of real gases (Topic 36), but we have enough information potential energy corresponding to that arrangement.
to establish the dependence here.
We need to evaluate (∂〈E〉/∂V〉T, the variation in energy of Brief illustration 54.5 A configuration integral
a system with volume at constant pressure. (In Topic 36, this
quantity is identified with the ‘internal pressure’ of a gas and Equation 54.11 is very difficult to manipulate in practice, even
denoted πT.) To proceed, we substitute eqn 54.5 and obtain for quite simple intermolecular potentials, except for a perfect
gas for which E P = 0. In that case, the exponential function
⎛ ∂E ⎞ ⎛ ∂ ⎛ ∂ lnQ ⎞ ⎞ becomes 1 and
⎜⎝ ∂V ⎟⎠ = − ⎜⎝ ∂V ⎜⎝ ∂β ⎟⎠ ⎟⎠ (54.8)
(∫ )
N
1 1 VN
∫
dτ 1dτ 2 …dτ N =
V T
T
Z= dτ =
N! N! N!
We need to consider the translational contribution to Q since
translational energy levels depend on volume, but to develop just as it should be for a perfect gas.
eqn 54.8 we also need to find a way to build an intermolecular
Self-test 54.5 Go on to show that, for a perfect gas,
potential energy into the expression for Q.
The total kinetic energy of a gas is the sum of the kinetic
(∂〈 E 〉/∂V )T = 0.
Answer: Z in this case is independent of temperature
energies of the individual molecules. Therefore, even in a
real gas the canonical partition function factorizes into a
part arising from the kinetic energy, which for the perfect If the potential has the form of a central hard sphere sur-
gas is Q = VN/Λ3NN!, where Λ is the thermal wavelength, eqn rounded by a shallow attractive well (Fig. 54.5), then detailed
52.7b, and a factor called the configuration integral, Z, which calculation, which is too involved to reproduce here, leads to
depends on the intermolecular potentials (don't confuse this Z
⎛ ∂〈 E 〉 ⎞ an2
with the compression factor in Topic 36). We therefore write ⎜⎝ ∂V ⎟⎠ = V 2 Attractive potential (54.12)
T
Z
Q = (54.9)
Λ3N
with Z replacing VN/N!, and expect Z to equal VN/N! for a per-
Potential energy, V
fect gas (see Brief illustration 54.5). It then follows that Determines b
⎛ ∂E ⎞ ⎛ ∂ ⎛ ∂ ln (Z / Λ3 N ) ⎞ ⎞
⎜⎝ ∂V ⎟⎠ = − ⎜ ⎜ ⎟ ⎟
⎜⎝ ∂V ⎝ ∂β ⎠ ⎟
T V ⎠T Intermolecular separation
0
⎛ 0 ⎞
⎛ ∂ ⎛ ∂ ln Z ⎞ ⎞ ⎜ ∂ ⎛ ∂ ln (1/Λ3 N ) ⎞ ⎟ Determines a
= −⎜ ⎜ ⎟ ⎟ −⎜ ⎜ ⎟ ⎟ (54.10)
⎝ ∂V ⎝ ∂β ⎠ V ⎠ T ⎜ ∂β ⎝ ∂V ⎠T ⎟
⎜⎝ ⎟⎠ Figure 54.5 The van der Waals equation of state (Topic 36)
V
⎛ ∂ ⎛ ∂ lnZ ⎞ ⎞ ⎛ ∂ 1 ⎛ ∂Z ⎞ ⎞ can be derived on the basis that the intermolecular potential

= −⎜
∂ ⎜⎝ ∂β ⎟⎠ ⎟ = − ⎜ ∂V Z ⎜⎝ ∂β ⎟⎠ ⎟ energy has a hard core of range b surrounded by a long-range,
⎝ V V ⎠T ⎝ V ⎠T shallow attractive well with an area proportional to a.
www.ebook3000.com
Atkins09819.indb 528 9/11/2013 12:42:54 PM

where n is the amount of molecules present in the volume V increases as it expands isothermally (because (∂〈 E 〉 / ∂V )T > 0,
and a is a constant that is proportional to the area under the and the slope of 〈E〉 with respect to V is positive). The energy
attractive part of the potential. In Topic 36 we see that exactly rises because, at greater average separations, the molecules
the same expression is implied by the van der Waals equation spend less time in regions where they interact favourably.
of state. At this point we can conclude that if there are attrac-
tive interactions between molecules in a gas, then its energy
☐ 1. The canonical ensemble is an imaginary collection ☐ 3. The mean energy of the members of the ensemble can
of replications of the actual system with a common be calculated from the canonical partition function.
temperature.
☐ 2. The canonical distribution gives the number of mem-
bers of the ensemble with a specified total energy.
Canonical distribution N i /N = e −βEi /Q 54.2a
Canonical partition function Q=

∑ei
− βEi Definition 54.2b
Mean energy 〈 E 〉 = −(1/Q )(∂Q /∂β )V = −(∂ lnQ /∂β )V 54.5
Configuration integral Q = Z /Λ3N 54.9
∫
Z = (1/N !) e − βEp dτ 1dτ 2 …dτ N Isotropic interaction 54.11
Variation of mean energy with volume (∂〈 E 〉/∂V )T = an2 /V 2 van der Waals gas; 0 for a perfect gas 54.12
Atkins09819.indb 529 9/11/2013 12:43:03 PM

Focus 11 on Statistical thermodynamics

Assume that all gases are perfect and that data refer to 298 K unless otherwise stated.
Topic 51 The Boltzmann distribution

51.1 Discuss the relationship between ‘population’, ‘configuration’, and ‘weight’. 51.3 What is temperature?
What is the significance of the most probable configuration?
51.4 Summarize the role of the Boltzmann distribution in chemistry.
51.2 What is the significance and importance of the principle of equal a priori
probabilities?
Exercises
51.1(a) Calculate the weight of the configuration in which 16 objects are 51.4(b) What is the temperature of a two-level system of energy separation
distributed in the arrangement 0, 1, 2, 3, 8, 0, 0, 0, 0, 2. equivalent to 300 cm−1 when the population of the upper state is one-half that
51.1(b) Calculate the weight of the configuration in which 21 objects are of the lower state?
distributed in the arrangement 6, 0, 5, 0, 4, 0, 3, 0, 2, 0, 0, 1.
51.5(a) Calculate the relative populations of a linear rotor in the levels with
51.2(a) Evaluate 8! by using (a) the exact formula; (b) Stirling’s approximation, J = 0 and J = 5, given that B = 2.71cm −1 and a temperature of 298 K.
eqn 51.2b; (c) the more accurate version of Stirling's approximation, eqn 51.5(b) Calculate the relative populations of a spherical rotor in the levels with
51.2a. J = 0 and J = 5, given that B = 2.71cm −1 and a temperature of 298 K.
51.2(b) Evaluate 10! by using (a) the exact formula; (b) Stirling’s
51.6(a) A certain molecule has a non-degenerate excited state lying at 540 cm−1
approximation, eqn 51.2b; (c) the more accurate version of Stirling’s
above the non-degenerate ground state. At what temperature will 10 per cent
approximation, eqn 51.2a.
of the molecules be in the upper state?
51.3(a) What are the relative populations of the states of a two-level system 51.6(b) A certain molecule has a doubly degenerate excited state lying at
when the temperature is infinite? 360 cm−1 above the non-degenerate ground state. At what temperature will
51.3(b) What are the relative populations of the states of a two-level system as 15 per cent of the molecules be in the upper state?
the temperature approaches zero?
51.4(a) What is the temperature of a two-level system of energy separation
equivalent to 400 cm−1 when the population of the upper state is one-third
that of the lower state?
Problems
51.1 A sample consisting of five molecules has a total energy 5ε. Each 51.4 A certain atom has a doubly degenerate ground level pair and an upper
molecule is able to occupy states of energy jε, with j = 0, 1, 2, …. (a) Calculate level of four degenerate states at 450 cm−1 above the ground level. In an
the weight of the configuration in which the molecules are distributed atomic-beam study of the atoms it was observed that 30 per cent of the atoms
evenly over the available states. (b) Draw up a table with columns headed by were in the upper level, and the translational temperature of the beam was
the energy of the states and write beneath them all configurations that are 300 K. Are the electronic states of the atoms in thermal equilibrium with the
consistent with the total energy. Calculate the weights of each configuration translational states?
and identify the most probable configurations.
51.5 Explore the consequences of using the full version of Stirling’s
51.2 A sample of nine molecules is numerically tractable but on the verge of approximation (eqn 51.2a), x ! ≈ (2π)1/2 x x +1/2 e − x , in the development of the
being thermodynamically significant. Draw up a table of configurations for expression for the configuration of greatest weight. Does the more accurate
N = 9, total energy 9ε, in a system with energy levels jε (as in Problem 51.1). approximation have a significant effect on the form of the Boltzmann
Before evaluating the weights of the configurations, guess (by looking for the distribution?
most ‘exponential’ distribution of populations) which of the configurations
51.6 The most probable configuration is characterised by a parameter we
will turn out to be the most probable. Go on to calculate the weights and
know as the ‘temperature’. The temperatures of the system specified in
identify the most probable configuration.
Problems 51.1 and 51.2 must be such as to give a mean value of ε for the
51.3 Use mathematical software to evaluate W for N = 20 for a series of energy of each molecule and a total energy Nε for the system. (a) Show
distributions over a uniform ladder of energy levels, ensuring that the total that the temperature can be obtained by plotting pj against j, where pj is the
energy is constant. Identify the configuration of greatest weight and compare (most probable) fraction of molecules in the state with energy jε. Apply the
it to the distribution predicted by the Boltzmann expression. Explore what procedure to the system in Problem 51.2. What is the temperature of the
happens as the value of the total energy is changed. system when ε corresponds to 50 cm−1? (b) Choose configurations other than
www.ebook3000.com
Atkins09819.indb 530 9/11/2013 12:43:06 PM

the most probable, and show that the same procedure gives a worse straight the radius of the planet, temperature, atmospheric composition, and other
line, indicating that a temperature is not well defined for them. factors. Prove that the atmosphere of planets cannot be in an equilibrium state
by demonstrating that the Boltzmann distribution leads to a uniform finite
51.7‡ The variation of the atmospheric pressure p with altitude h is predicted
number density as r → ∞. Hint: Recall that in a gravitational field the potential
by the barometric formula to be p = p0 e −h/H , where p0 is the pressure at
energy is V(r) = –GMm/r, where G is the gravitational constant, M is the mass
sea level and H = RT/Mg with M the average molar mass of air and T the
of the planet, and m is the mass of the particle.
average temperature. Obtain the barometric formula from the Boltzmann
distribution. Recall that the potential energy of a particle at height h above the 51.9 Consider a protein P with four distinct sites, with each site capable of
surface of the Earth is mgh. Convert the barometric formula from pressure to binding one ligand L. Show that the possible varieties (‘configurations’) of
number density, N. Compare the relative number densities, N(h)/N(0), for O2 the species PLi (with PL0 denoting P) are given by the binomial coefficients
and H2O at h = 8.0 km, a typical cruising altitude for commercial aircraft. (for binomial coefficients, see Mathematical background 7).
51.8‡ Planets lose their atmospheres over time unless they are replenished. A
complete analysis of the overall process is very complicated and depends upon
Topic 52 Partition functions

52.1 Describe the physical significance of the partition function. 52.3 What is the difference between a ‘state’ and an ‘energy level’? Why is it
important to make this distinction?
52.2 Describe how the mean energy of a system composed of two levels varies
with temperature. 52.4 Why and when is it necessary to include a symmetry number in the
calculation of a partition function?
Exercises
52.1(a) Calculate (a) the thermal wavelength, (b) the translational partition 52.6(a) The rotational constant of CH4 is 5.241 cm−1. Evaluate the rotational
function at (i) 300 K and (ii) 3000 K of a molecule of molar mass 150 g mol−1 partition function explicitly (without approximation but ignoring the role
in a container of volume 1.00 cm3. of nuclear statistics) and plot its value as a function of temperature. At what
52.1(b) Calculate (a) the thermal wavelength, (b) the translational partition temperature is the value within 5 per cent of the value calculated from the
function of a Ne atom in a cubic box of side 1.00 cm at (i) 300 K and approximate formula?
(ii) 3000 K. 52.6(b) The rotational constant of CCl4 is 0.0572 cm−1. Evaluate the rotational
partition function explicitly (without approximation but ignoring the role
52.2(a) Calculate the ratio of the translational partition functions of H2 and He
of nuclear statistics) and plot its value as a function of temperature. At what
at the same temperature and volume.
temperature is the value within 5 per cent of the value calculated from the
52.2(b) Calculate the ratio of the translational partition functions of Ar and Ne
approximate formula?
at the same temperature and volume.
= 5.097 cm −1 and
52.7(a) The rotational constants of CH3Cl are A
52.3(a) The bond length of O2 is 120.75 pm. Use the high-temperature
B = 0.443 cm −1 . Evaluate the rotational partition function explicitly (without
approximation to calculate the rotational partition function of the molecule
approximation but ignoring the role of nuclear statistics) and plot its value as
at 300 K.
a function of temperature. At what temperature is the value within 5 per cent
52.3(b) The bond length of N2 is 109.75 pm. Use the high-temperature
of the value calculated from the approximate formula?
approximation to calculate the rotational partition function of the molecule = 6.196cm −1 and
52.7(b) The rotational constants of NH3 are A
at 300 K.
B = 9.444 cm −1 . Evaluate the rotational partition function explicitly (without
52.4(a) The NOF molecule is an asymmetric rotor with rotational constants approximation but ignoring the role of nuclear statistics) and plot its value as
3.1752 cm−1, 0.3951 cm−1, and 0.3505 cm−1. Calculate the rotational partition a function of temperature. At what temperature is the value within 5 per cent
function of the molecule at (a) 25 °C, (b) 100 °C. of the value calculated from the approximate formula?
52.4(b) The H2O molecule is an asymmetric rotor with rotational constants
52.8(a) Give the symmetry number for each of the following molecules:
27.877 cm−1, 14.512 cm−1, and 9.285 cm−1. Calculate the rotational partition
(a) CO, (b) O2, (c) H2S, (d) SiH4, and (e) CHCl3.
function of the molecule at (a) 25 °C, (b) 100 °C.
52.8(b) Give the symmetry number for each of the following molecules:
52.5(a) The rotational constant of CO is 1.931 cm−1. Evaluate the rotational (a) CO2, (b) O3, (c) SO3, (d) SF6, and (e) Al2Cl6.
partition function explicitly (without approximation) and plot its value as a
52.9(a) Estimate the rotational partition function of ethene at 25 °C given that
function of temperature. At what temperature is the value within 5 per cent of
A = 4.828 cm −1 , B = 1.0012 cm −1 , and C = 0.8282 cm −1 . Take the symmetry
the value calculated from the approximate formula?
number into account.
52.5(b) The rotational constant of HI is 6.511 cm−1. Evaluate the rotational
52.9(b) Evaluate the rotational partition function of pyridine, C5H5N, at
partition function explicitly (without approximation) and plot its value as a
25 °C given that A = 0.2014 cm −1 , B = 0.1936cm −1 , C = 0.0987 cm −1 . Take the
function of temperature. At what temperature is the value within 5 per cent of
symmetry number into account.
the value calculated from the approximate formula?
‡ These problems were supplied by Charles Trapp and Carmen Giunta.
Atkins09819.indb 531 9/11/2013 12:43:13 PM

52.10(a) The vibrational wavenumber of Br2 is 323.2 cm−1. Evaluate the 52.12(a) Calculate the vibrational partition function of CCl4 at 500 K given the
vibrational partition function explicitly (without approximation) and plot its wavenumbers 459 cm−1 (symmetric stretch, A), 217 cm−1 (deformation, E),
value as a function of temperature. At what temperature is the value within 776 cm−1 (deformation, T), 314 cm−1 (deformation, T).
5 per cent of the value calculated from the approximate formula? 52.12(b) Calculate the vibrational partition function of CI4 at 500 K given the
52.10(b) The vibrational wavenumber of I2 is 214.5 cm−1. Evaluate the wavenumbers 178 cm−1 (symmetric stretch, A), 90 cm−1 (deformation, E),
vibrational partition function explicitly (without approximation) and plot its 555 cm−1 (deformation, T), 125 cm−1 (deformation, T).
value as a function of temperature. At what temperature is the value within
52.13(a) A certain atom has a fourfold degenerate ground level, a non-
5 per cent of the value calculated from the approximate formula?
degenerate electronically excited level at 2500 cm−1, and a twofold degenerate
52.11(a) Calculate the vibrational partition function of CS2 at 500 K given the level at 3500 cm−1. Calculate the partition function of these electronic states at
wavenumbers 658 cm−1 (symmetric stretch), 397 cm−1 (bend; two modes), 1900 K. What is the relative population of each level at 1900 K?
1535 cm−1 (asymmetric stretch). 52.13(b) A certain atom has a triply degenerate ground level, a non-degenerate
52.11(b) Calculate the vibrational partition function of HCN at 900 K given the electronically excited level at 850 cm−1, and a fivefold degenerate level at
wavenumbers 3311 cm−1 (symmetric stretch), 712 cm−1 (bend; two modes), 1100 cm−1. Calculate the partition function of these electronic states at 2000 K.
2097 cm−1 (asymmetric stretch). What is the relative population of each level at 2000 K?
Problems
52.1 This problem is best done using mathematical software. Equation 52.15 is Calculate the electronic partition function at 298 K and 1000 K by direct
the partition function for a harmonic oscillator. Consider a Morse oscillator summation. Hint: The degeneracy of a level J is 2J + 1.
(Topic 43) in which the energy levels are given by eqn 43.15 in the form
52.6 The pure rotational microwave spectrum of HCl has absorption lines
at the following wavenumbers (in cm−1): 21.19, 42.37, 63.56, 84.75, 105.93,
2
⎛ 1⎞ ⎛ 1⎞ 127.12, 148.31, 169.49, 190.68, 211.87, 233.06, 254.24, 275.43, 296.62, 317.80,
Ev = ⎜ v + ⎟ hc␯ − ⎜ v + ⎟ hcx e ␯
⎝ 2⎠ ⎝ 2⎠ 338.99, 360.18, 381.36, 402.55, 423.74, 444.92, 466.11, 487.30, 508.48.
Calculate the rotational partition function at 25 °C by direct summation.
Evaluate the partition function for this oscillator, remembering to measure 52.7 Calculate, by explicit summation, the vibrational partition function
energies from the lowest level and to note that there is only a finite number and the vibrational contribution to the energy of I2 molecules at (a) 100 K,
of bound-state levels. Plot the partition function against temperature for a (b) 298 K given that its vibrational energy levels lie at the following
variety of values of xe, and—on the same graph—compare your results with wavenumbers above the zero-point energy level: 0, 213.30, 425.39, 636.27,
that for a harmonic oscillator. 845.93 cm−1. What proportion of I2 molecules are in the ground and first two
52.2 Explore the conditions under which the ‘integral’ approximation for the excited levels at the two temperatures?
translational partition function is not valid by considering the translational 52.8‡ Consider the electronic partition function of a perfect atomic hydrogen
partition function of an H atom in a one-dimensional box of side comparable gas at a density of 1.99 × 10−4 kg m−3 and 5780 K. These are the mean
to that of a typical nanoparticle, 100 nm. Estimate the temperature at which, conditions within the Sun's photosphere, the surface layer of the Sun that
according to the integral approximation, q = 10 and evaluate the exact is about 190 km thick. (a) Show that this partition function, which involves
partition function at that temperature. a sum over an infinite number of quantum states that are solutions to the
52.3 (a) Calculate the electronic partition function of a tellurium atom at (i) Schrödinger equation for an isolated atomic hydrogen atom, is infinite.
298 K, (ii) 5000 K by direct summation using the following data: (b) Develop a theoretical argument for truncating the sum and estimate the
maximum number of quantum states that contribute to the sum. (c) Calculate
Term Degeneracy Wavenumber/cm−1 the equilibrium probability that an atomic hydrogen electron is in each
quantum state. Are there any general implications concerning electronic states
Ground 5 0
that will be observed for other atoms and molecules? Is it wise to apply these
1 1 4707 calculations in the study of the Sun’s photosphere?
2 3 4751 52.9 A formal way of arriving at the value of the symmetry number is
3 5 10 559 to note that σ is the order (the number of elements) of the rotational
subgroup of the molecule, the point group of the molecule with all but
(b) What proportion of the Te atoms are in the ground term and in the term the identity and the rotations removed. The rotational subgroup of
labelled 2 at the two temperatures? H2O is {E,C2}, so σ = 2. The rotational subgroup of NH3 is {E,2C3}, so
σ = 3. This recipe makes it easy to find the symmetry numbers for more
52.4 The four lowest electronic levels of a Ti atom are: 3F2, 3F3, 3F4, and 5F1, at complicated molecules. The rotational subgroup of CH4 is obtained
0, 170, 387, and 6557 cm−1, respectively. There are many other electronic states from the T character table as {E,8C3,3C2}, so σ = 12. For benzene, the
at higher energies. The boiling point of titanium is 3287 °C. What are the rotational subgroup of D6 h is {E,2C6,2C3,C2,3C2′,3C2″}, so σ = 12. (a)
relative populations of these levels at the boiling point? Hint: The degeneracies Estimate the rotational partition function of ethene at 25 °C given that
of the levels are 2J + 1. A = 4.828cm −1 , B = 1.0012cm −1 , and C = 0.8282cm −1 . (b) Evaluate the
52.5‡ J. Sugar and A. Musgrove (J. Phys. Chem. Ref. Data 22, 1213 (1993)) have rotational partition function of pyridine, C5H5N, at room temperature
published tables of energy levels for germanium atoms and cations from Ge+ (A = 0.2014 cm −1 , B = 0.1936cm −1 , C = 0.0987 cm −1 ).
to Ge+31. The lowest-lying energy levels in neutral Ge are as follows:
3P 3P 3P 1D 1S
0 1 2 2 0
(E/hc)/cm−1 0 557.1 1410.0 7125.3 16 367.3
www.ebook3000.com
Atkins09819.indb 532 9/11/2013 12:43:17 PM

Topic 53 Molecular energies

53.1 Discuss the conditions under which energies predicted from the 53.2 Describe the behaviour of the mean energy of a two-level system in the
equipartition theorem coincide with energies computed by using partition limit of (i) very low and (ii) very high temperature.
functions.
Exercises
53.1(a) Compute the mean energy at 298 K of a two-level system of energy 53.5(b) Evaluate, by explicit summation, the mean vibrational energy of I2
separation equivalent to 500 cm−1. and plot its value as a function of temperature. At what temperature is the
53.1(b) Compute the mean energy at 400 K of a two-level system of energy equipartition value within 5 per cent of the accurate value? Use ␯ = 214.5cm −1 .
separation equivalent to 600 cm−1.
53.6(a) Evaluate, by explicit summation, the mean vibrational energy of
53.2(a) Evaluate, by explicit summation, the mean rotational energy of CS2 and plot its value as a function of temperature. At what temperature
CO and plot its value as a function of temperature. At what temperature is the equipartition value within 5 per cent of the accurate value? Use the
is the equipartition value within 5 per cent of the accurate value? wavenumbers 658 cm−1 (symmetric stretch), 397 cm−1 (bend; two modes),
CO) = 1.931cm −1 .
B( 1535 cm−1 (asymmetric stretch).
53.2(b) Evaluate, by explicit summation, the mean rotational energy of 53.6(b) Evaluate, by explicit summation, the mean vibrational energy of
HI and plot its value as a function of temperature. At what temperature HCN and plot its value as a function of temperature. At what temperature
is the equipartition value within 5 per cent of the accurate value? is the equipartition value within 5 per cent of the accurate value? Use the
HI) = 6.511cm −1 .
B( wavenumbers 3311 cm−1 (symmetric stretch), 712 cm−1 (bend; two modes),
2097 cm−1 (asymmetric stretch).
53.3(a) Evaluate, by explicit summation, the mean rotational energy of
CH4 and plot its value as a function of temperature. At what temperature 53.7(a) Evaluate, by explicit summation, the mean vibrational energy of
is the equipartition value within 5 per cent of the accurate value? CCl4 and plot its value as a function of temperature. At what temperature
CH ) = 5.241cm −1 .
B( is the equipartition value within 5 per cent of the accurate value? Use the
4
53.3(b) Evaluate, by explicit summation, the mean rotational energy of wavenumbers 459 cm−1 (symmetric stretch, A), 217 cm−1 (deformation,
CCl4 and plot its value as a function of temperature. At what temperature E), 776 cm−1 (deformation, T), 314 cm−1 (deformation, T). A modes are
is the equipartition value within 5 per cent of the accurate value? non-degenerate, E modes are doubly degenerate, and T modes are triply
CCl ) = 0.0572cm −1 .
B( degenerate.
4
53.7(b) Evaluate, by explicit summation, the mean vibrational energy of
53.4(a) Evaluate, by explicit summation, the mean rotational energy of CH3Cl
CI4 and plot its value as a function of temperature. At what temperature
and plot its value as a function of temperature. At what temperature is the
is the equipartition value within 5 per cent of the accurate value? Use the
equipartition value within 5 per cent of the accurate value? A = 5.097 cm −1
wavenumbers 178 cm−1 (symmetric stretch, A), 90 cm−1 (deformation,
and B = 0.443cm −1 .
E), 555 cm−1 (deformation, T), 125 cm−1 (deformation, T). A modes are
53.4(b) Evaluate, by explicit summation, the mean rotational energy of NH3
non-degenerate, E modes are doubly degenerate, and T modes are triply
and plot its value as a function of temperature. At what temperature is the
degenerate.
equipartition value within 5 per cent of the accurate value? A = 6.196cm −1
and B = 9.444 cm −1 . 53.8(a) Calculate the mean contribution to the electronic energy at 1900 K for
a sample composed of the atoms specified in Exercise 52.13(a).
53.5(a) Evaluate, by explicit summation, the mean vibrational energy of
53.8(b) Calculate the mean contribution to the electronic energy at 2000 K for
Br2 and plot its value as a function of temperature. At what temperature
a sample composed of the atoms specified in Exercise 52.13(b).
is the equipartition value within 5 per cent of the accurate value? Use
␯ = 323.2cm −1 .
Problems
53.1 An electron trapped in an infinitely deep spherical well of radius R, such only these six energy levels need be considered. Hint: Remember to measure
as may be encountered in the investigation of nanoparticles, has energies energies from the lowest level.
given by the expression Enl = 2 Xnl
2 /2m R2 , with X the value obtained by
e nl
53.2 The NO molecule has a doubly degenerate excited electronic level
searching for the zeroes of the spherical Bessel functions. The first six values
121.1 cm−1 above the doubly degenerate electronic ground term. Calculate
(with a degeneracy of the corresponding energy level equal to 2l + 1) are as
and plot the electronic partition function of NO from T = 0 to 1000 K.
follows:
Evaluate (a) the term populations and (b) the mean electronic energy at 300 K.
Evaluate the partition function and mean energy of an electron as a function
53.3 Consider a system with energy levels εj = jε and N molecules. (a) Show
of temperature. Choose the temperature range and radius to be so low that
that if the mean energy per molecule is aε, then the temperature is given by
n 1 1 1 2 1 2
l 0 1 2 0 3 1 1 ⎛ 1⎞
β= ln 1+
ε ⎝⎜ a ⎠⎟
Xnl 3.142 4.493 5.763 6.283 6.988 7.725
Atkins09819.indb 533 9/11/2013 12:43:27 PM

Evaluate the temperature for a system in which the mean energy is ε, taking ε 53.4 Deduce an expression for the root mean square energy, 〈ε2〉1/2, in terms
equivalent to 50 cm−1. (b) Calculate the molecular partition function q for the of the partition function and hence an expression for the root mean square
system when its mean energy is aε. deviation from the mean, Δε = (〈ε2〉 – 〈ε〉2)1/2. Evaluate the resulting expression
for a harmonic oscillator.
Topic 54 The canonical ensemble

54.1 Why is the concept of a canonical ensemble required? 54.3 Under what circumstances may identical particles be regarded as
distinguishable?
54.2 Explain what is meant by an ensemble and why it is useful in statistical
thermodynamics. 54.4 What is meant by the ‘thermodynamic limit’?
Exercises
54.1(a) Identify the systems for which it is essential to include a factor of 54.1(b) Identify the systems for which it is essential to include a factor of 1/N!
1/N! on going from Q to q : (a) a sample of helium gas, (b) a sample of carbon on going from Q to q : (a) a sample of carbon dioxide gas, (b) a sample of
monoxide gas, (c) a solid sample of carbon monoxide, (d) water vapour. graphite, (c) a sample of diamond, (d) ice.
Problem
54.1‡ For a perfect gas, the canonical partition function, Q, is related to the p = kT(∂ ln Q/∂V)T. Use the expression for q to derive the perfect gas law
molecular partition function q by Q = q N/N!. In Topic 66 it is established that pV = nRT.
Integrated activities
F11.1 An electron spin can adopt either of two orientations in a magnetic (d)LG Consider a three-level system with levels 0, ε, and 2ε. Plot the functions
field, and its energies are ±μBB, where μB is the Bohr magneton. Deduce an p0, p1, and p2 against kT/ε.
expression for the partition function and mean energy of the electron and (e)LG Plot the temperature dependence of the vibrational contribution to the
sketch the variation of the functions with B. Calculate the relative populations molecular partition function for several values of the vibrational wavenumber.
of the spin states at (a) 4.0 K, (b) 298 K when B = 1.0 T. Estimate from your plots the temperature above which the harmonic
oscillator is in the ‘high-temperature’ limit.
F11.2 A nitrogen nuclear spin can adopt any of three orientations in a
(f)LG Plot the temperature dependence of the electronic partition function
magnetic field, and its energies are 0, ±γNB, where γN is the magnetogyric
for several values of the energy separation ε between two doubly degenerate
ratio of the nucleus. Deduce an expression for the partition function and
levels. From your plots, estimate the temperature at which the population of
mean energy of the nucleus and sketch the variation of the functions with
the excited level begins to increase sharply.
B. Calculate the relative populations of the spin states at (a) 1.0 K, (b) 298 K
(g)LG Draw graphs similar to those in Fig. 53.1 for a three-level system with
when B = 20.0 T.
levels 0, ε, and 2ε.
F11.3 Use mathematical software, a spreadsheet, or the Living graphs (h)LG Plot the temperature dependence of the mean rotational energy
(labelledLG) on the website of this book for the following: for several values of the rotational constant (for reasonable values of the
(a)LG Plot the partition function of a harmonic oscillator against temperature rotational constant, see the Resource section). From your plots, estimate the
for several values of the energy separation ε. How does q vary with temperature at which the mean rotational energy begins to increase sharply.
temperature when T is high, in the sense that kT ε (or βε 1)? (i)LG Plot the temperature dependence of the mean vibrational energy for
(b)LG Consider a three-level system with levels 0, ε, and 2ε. Plot the partition several values of the vibrational wavenumber (for reasonable values of the
function against kT/ε. vibrational wavenumber, see the Resource section). From your plots, estimate the
(c)LG To visualize the content of Fig. 52.4 in a different way, plot the functions temperature at which the mean vibrational energy begins to increase sharply.
p0, p1, p2, and p3 against kT/ε.
www.ebook3000.com
Atkins09819.indb 534 9/11/2013 12:43:27 PM

Probability theory 535
Mathematical background 7 Probability theory

Probability theory deals with quantities and events that are dis- However, if the order in which ‘heads’ come up is unimpor-
tributed randomly and shows how to calculate average values tant, there are 12!/5!7! ways of achieving 5 ‘heads’ in 12 tosses,
of various kinds. We shall consider variables that take discrete and in general N!/n!(N – n)! ways of achieving n ‘heads’ in N
values (as in a one-dimensional random walk with a fixed step trials. The probability of getting exactly n ‘heads’ is therefore
length) and continuous values (as in the diffusion of a particle the product of pnqN−n and the number of ways of distributing n
through a fluid). ‘heads’ over N trials:
⎛ N⎞ ⎛ N⎞ N!
MB7.1 Discrete distributions P (n) = ⎜ ⎟ pnq N −n where ⎜ ⎟ =
⎝ n⎠ ⎝ n ⎠ n!(N − n)!
We denote a variable x and the discrete values that it may take
xi, i = 1, 2, ..., N. If the probability that xi occurs is pi, then the Binomial
distribution (MB7.3)
mean value (or expectation value) of x is ⎛ N⎞
The symbol ⎜ ⎟ is the binomial coefficient, as it occurs in the
N ⎝ n⎠
〈x〉 = ∑x p
i =1
i i
Mean value (MB7.1a) binomial expansion:
N
⎛ N⎞
The mean values of higher powers of x may be computed
similarly:
( x + y )N = ∑ ⎜⎝ n ⎟⎠ x y
n=0
n N −n Binomial
expansion (MB7.4)
N
The expression P(n) is accordingly called the binomial distri-
〈xn 〉 = ∑x p
i =1
n
i i
Mean nth power (MB7.1b)
bution (Fig. MB7.1).
Manipulations of the binomial distribution are often facili-
Although the mean is a useful measure, it is important to tated by the following summation procedure:
know the width in the scatter of outcomes around the mean. 1. Introduce a dummy variable a.
There are two related measures: one is the variance, V(x), and
2. Express the sum as a derivative with respect to a, using
the other is the standard deviation, σ(x), the square root of
nan = adan/da to eliminate the factor n; if a power of n
the variance:
occurs, apply d/da the appropriate number of times.
V ( x ) = 〈 x 2 〉 − 〈 x 〉2 Variance (MB7.2a) 3. Evaluate the resulting sum: note the binomial expansion
(eqn MB7.4).
4. Evaluate the derivative.
σ (x ) = V (x )1/2 = {〈 x 2 〉 − 〈 x 〉2 }
1/2
Standard
deviation (MB7.2b) 5. Finally, set a = 1.
In certain cases, the probabilities can be expressed in a simple

0.25
way, depending on the nature of the events being considered. 10
0.2
(a) The binomial distribution
0.15
In a Bernoulli trial, the outcome of an observation is one of a
P(n)
mutually exclusive pair (such as ‘heads’ or ‘tails’ in a coin toss) 50
0.1
and successive trials are independent (so that getting ‘heads’ 100
on one toss does not influence the following toss). Suppose the 150
0.05
probability of outcome 1 is p and that of the alternative out-
come 2 is q, with p + q = 1. For a fair coin, p = q = 12 . Then one
series of N = 12 trials might be 0
0 20 40 60 80 100
n
thhtththttht Probability of occurrence = p5q 7 , Figure MB7.1 The binomial distribution for different values of
and in general pnq N −n N and p = q = ½.
Atkins09819.indb 535 9/11/2013 12:43:41 PM

536 Mathematical background 7
Brief illustration MB7.1 1

The binomial distribution 0.4
We can use the binomial distribution to determine the mean

number of times that ‘heads’ will be obtained in a series of N 0.3
trials (this is the average value of n, denoted 〈n〉): P(n)
2
0.2 4
N N N
⎛ N⎞ ⎛ N⎞
∑ ∑ ∑
6
〈n〉 = nP (n) = n ⎜ ⎟ pnq N −n = n ⎜ ⎟ pn (1 − p)N −n
n=0 n=0
⎝ n⎠ n=0
⎝ n⎠
0.1
Application of the summation procedure to this expression

0
gives: 0 1 2 3 4 5 6 7 8 9
n
Step1 N Step2 Figure MB7.2 The Poisson distribution for different values of Δt/τ.

⎛ N⎞
N
⎛ N⎞ n n
∑ ∑
d
〈n〉 = n ⎜ ⎟ an pn (1 − p)N −n = a ⎜ n ⎟ a p (1 − p)
N −n
⎝ ⎠
n d a ⎝ ⎠
n=0 n=0 just the probability of getting n ‘heads’ in a total interval that
Step3 Step 4

d
spans N of the tiny intervals:
= a (ap +[1 − p])N = Nap(ap +[1 − p])N −1
da
Step5

⎛ N⎞ ⎛ N ⎞ ⎛ Δt ⎞ n ⎛ Δt ⎞ N −n
= Np P (n) = ⎜ ⎟ pn (1− p)N −n = ⎜ ⎟ ⎜ ⎟ ⎜ 1− ⎟
⎝ n⎠ ⎝ n ⎠ ⎝ Nτ ⎠ ⎝ Nτ ⎠
For example, the mean number of times ‘heads’ is obtained for If then we suppose that N is very large and make use of the
a fair coin ( p = 12 ) in 10 trials is 5. relation
lim(1+ x )1/x = e (MB7.3)

x→0
Note that although the width of the distribution increases (as then in a slightly involved but straightforward calculation
N1/2), its value relative to the mean decreases (as N1/2/N = 1/N1/2). (which we do not reproduce here) we arrive at the Poisson
For tosses of a fair coin, when p = 12 , 〈n〉 = 12 N (half the tosses distribution:
turn up ‘heads’) and σ (n) = 12 N 1/2 .
(Δt /τ )n −Δt /τ
P (n) = e Poisson distribution (MB7.4)
n!
(b) The Poisson distribution
In another important type of trial, an event either takes place for the probability that n events will occur in the interval Δt
or does not, such as an excited molecule dissociating into (Fig. MB7.2). As shown in the following Brief illustration, the
fragments or (more mundanely) a bus arriving. At first sight mean number of events in the interval Δt is just Δt/τ, so τ can
it appears that we cannot assign a meaning to the number of be interpreted as the average time between events.
times an event does not occur (how many times did the bus not
arrive in an interval?). However, we can still assign a probability Brief illustration MB6.2 The Poisson distribution
that an event occurs by imagining an interval of time Δt that
is divided into N regions, each of duration Δt/N (or, similarly, To calculate the average number of events that occur in an
regions of space Δx/N) that are so small that the probability interval Δt (and in the limit N → ∞), we need to evaluate
that two or more events occurs in it is negligible and we have N N N
(Δt / τ )n −Δt /τ −Δt /τ (Δt /τ )n
a ‘heads’ for the event occurring in that brief interval or tiny 〈n〉 = ∑n=1
nP (n) = ∑ n=1
n
n!
e =e ∑ (n −1)!
n=1
region and a ‘tails’ if it does not.
N N
(Δt /τ )
n−1
(Δt / τ )n
∑ ∑
If the events occur at random, the probability of an event
= (Δt / τ )e −Δt /τ = (Δt /τ )e −Δt /τ
occurring within this tiny interval is proportional to the length (n −1)! n!
n=1 n=0
of the interval, and we can write p = Δt/Nτ, where 1/τ is a con- = (Δt / τ )e −Δt /τ e Δt /τ = Δt / τ
stant of proportionality (we give it a physical meaning later). It
then follows that in a set of Bernouilli trials (where ‘heads’ now where we have used the Taylor series expansion (eqn MB1.7b)
corresponds to ‘did occur’ and ‘tails’ to ‘did not occur’) the total of ex.
probability that n events occur in the interval Δt = N(Δt/N) is
www.ebook3000.com
Atkins09819.indb 536 9/11/2013 12:43:59 PM

Probability theory 537
0.14
(c) The Gaussian distribution (a) Even
0.12
Suppose that a variable can take positive and negative integer Mean
values centred on zero and that to reach a certain value n the 0.1
P(n) (b) Odd
system jumps to the left or right by taking steps of length λ at 0.08
random for a total of N steps. The number of ways of taking NR
0.06
steps to the right and NL to the left (with N = NR + NL) in any one
such trial of N steps is 0.04
0.02
N!
W= (MB7.7)
N L !N R ! 0
–10 –9 –8 –7 –6 –5 –4 –3 –2 –1 0 1 2 3 4 5 6 7 8 9 10
n
Then, because there are 2N possible choices of direction in the
Figure MB7.3 The Gaussian distribution for N even (40) and
course of N steps, the probability of being n steps from the ori-
N odd (41), and their mean. The bell-shaped mean curve is
gin, with n = NR – NL in a trial of N steps, is the average of the two distributions, and corresponds to the
Gaussian distribution function for a continuous variable.
W N! N!
P (n) = = N = (MB7.8)
2 N
2 N L !N R ! ⎛ N +n⎞ ⎛ N −n⎞ When N is very large and the xi values are so closely spaced
2N ⎜ ! !
⎝ 2 ⎟⎠ ⎜⎝ 2 ⎟⎠ that x can be regarded as varying continuously, it is useful to
express the probability that an outcome will lie between x and
A point of some subtlety and which we draw on later is that if N x + dx as ρ(x)dx, where the function ρ(x) is the probability
is even, then n must be even (you cannot end up an odd num- density. The mean value of x continuously varying between –∞
ber of steps from the origin if you take an even number of steps; and ∞ is then given by
think about N = 4); similarly, if N is odd, then n must be odd too
(you cannot end up an even number of steps from the origin if ∞
you take an odd number of steps; think about N = 5). Because 〈x〉 = ∫ −∞
x ρ(x )dx (MB7.11)
it then follows that N + n and N – n are both even numbers, the
factorials we have to evaluate in P(n) are of whole numbers. with analogous expressions for the variance and standard
To develop this expression in the case of large numbers of deviation.
steps, we take logarithms and use Stirling’s approximation: To derive the Gaussian version of the probability density we
use the random-walk model in Section MB7.1 and write x = nλ,
⎛ 1⎞ Stirling’s allowing λ to be very small and n to be very large and effec-
ln x ! ≈ ln(2π )1/2 + ⎜ x + ⎟ ln x − x approximation (MB7.9)
⎝ 2⎠ tively continuous. If dx spreads over a sufficiently wide range
of points, then instead of dealing with a distribution like that
This approximation leads, after a fair amount of algebra, to the in Fig. MB7.3a or like that in Fig. MB7.3b, we can deal with the
Gaussian distribution average of the two, as shown by the curves superimposed on
the distribution. That is, for a continuous distribution we use
2
P (n) = e −n /2 N
2
Gaussian distribution (MB7.5)
(2N π ) 1/2
P (n) =
1
e − n /2 N
2
(MB7.12)
(2N π )1/2
with, remember, n even if N is even and n odd if N is odd. This
bell-shaped curve is illustrated in Fig. MB7.3 for N even and The total probability of being in the range dx = λdn at x = nλ is
N odd. The Gaussian distribution is commonly used to dis- therefore
cuss a continuous function, as we demonstrate in the following
section. 1 1 dx
ρ(x )dx = e −n /2 N dn = e − x /2 N λ
2 2 2
(MB7.13)
(2N π )1/2 (2N π )1/2 λ
MB7.2 Continuous distributions It follows that
A continuous distribution is a distribution in which the vari-
able can take on a continuum of values. One of the most impor-
tant examples can be developed from the Gaussian distribution 1
ρ(x ) = e − x /2 N λ
2 2
(MB7.14)
(2 N π λ )
1/2
2
of the preceding section, as follows.
Atkins09819.indb 537 9/11/2013 12:44:22 PM

538 Mathematical background 7
This expression is commonly expressed as the Gaussian distri-

To evaluate the mean square value of x we write
bution function (or normal distribution function):
∞ 1/2 ∞
⎛ 1 ⎞
Gaussian 〈x2 〉 = ∫ x 2 ρ(x )dx = ⎜ ∫ x 2e −( x − 〈 x 〉) /2σ dx
1/2 2 2
⎛ 1 ⎞ − x /2σ ⎝ 2πσ 2 ⎟⎠
ρ(x ) = ⎜
2 2
e distribution (MB7.15) −∞ −∞
⎝ 2πσ 2 ⎟⎠ function
where σ = N1/2λ turns out to be the standard deviation of the Now ma ke t he subst itut ion z = (x – 〈x〉)/(2σ 2 )1/2 , so
distribution (see below). dx = (2σ2)1/2dz and x 2 = 2σ2 z2 + 2 (2σ2)1/2z〈x〉 + 〈x〉2, and obtain
1/2
⎛ 1 ⎞
〈x2 〉 = ⎜ (2σ 2 )1/2 ×
⎝ 2πσ 2 ⎟⎠
Brief illustration MB7.3 Gaussian distribution function
⎧
0 ⎫
⎪⎪ ∞ ∞ ∞ ⎪⎪
∫ ∫ ∫
If the bell-shaped curve of the Gaussian distribution function 2σ 2 z 2e − z dz + 2(2σ 2 )1/2 〈 x 〉
2
z e − z dz + 〈 x 〉 2
2
e− z
2
⎨ dz ⎬
is centred on 〈x〉, the distribution becomes ⎪ −∞ −∞ −∞
⎪
⎪⎩ ⎪⎭
1/2
⎛ 1 ⎞ =
1
{π1/2 σ 2 + 〈 x 〉2 π1/2 } = σ 2 + 〈 x 〉2
ρ(x ) = ⎜ e −( x − 〈 x 〉) /2σ
2 2
⎝ 2πσ 2 ⎟⎠ π1/2
www.ebook3000.com
Atkins09819.indb 538 9/11/2013 12:44:27 PM

Peter Atkins Julio de Paula Ron Friedman Physical Chemistry Quanta (0511-0561)

Uploaded by

Copyright:

Available Formats

Peter Atkins Julio de Paula Ron Friedman Physical Chemistry Quanta (0511-0561)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Peter Atkins Julio de Paula Ron Friedman Physical Chemistry Quanta (0511-0561)

Uploaded by

Copyright:

Available Formats

488 10 Magnetic resonance

of the molecules and the field. In solution, when the molecule

50.2 Hyperfine structure

Therefore, instead of a single line, the spectrum shows two lines

Atkins09819.indb 488 9/11/2013 12:37:07 PM

1 Figure 50.4 The analysis of the hyperfine structure

Atkins09819.indb 489 9/11/2013 12:37:11 PM

Atkins09819.indb 490 9/11/2013 12:37:15 PM

Brief illustration 50.3The composition of a molecular Hund

between a proton and the α electrons (ms = + 12 ) which results

EPR resonance condition h␯ = g μBB0 No hyperfine interaction 50.2

h␯ = g μB (B0 ± 12 a) Hyperfine interaction between an electron and a proton 50.4

McConnell equation a = Qρ Q = 2.25 mT 50.5

Atkins09819.indb 491 9/11/2013 12:37:19 PM

Focus 10 on Magnetic resonance

Topic 47 General principles

Atkins09819.indb 492 9/11/2013 12:37:20 PM

Topic 48 Features of NMR spectra

Atkins09819.indb 493 9/11/2013 12:37:23 PM

Topic 49 Pulse techniques in NMR

Figure F10.1 The COSY spectrum of 1-nitropropane (NO2CH2CH2CH3). The

S01 = 1.0 νL1 = 50 MHz T21 = 0.50 μs

S02 = 3.0 νL2 = 10 MHz T22 = 1.0 μs

Atkins09819.indb 494 9/11/2013 12:37:27 PM

Topic 50 Electron paramagnetic resonance

Atkins09819.indb 495 9/11/2013 12:37:30 PM

Substituent CH3 H CF3 CN NO2

Atkins09819.indb 496 9/11/2013 12:37:35 PM

Topic 51 Topic 52 Topic 53 Topic 54

Atkins09819.indb 497 9/11/2013 12:37:35 PM

The Boltzmann distribution

Atkins09819.indb 498 9/11/2013 12:37:36 PM

instance, is as likely to be populated as a rotational state of the

Atkins09819.indb 499 9/11/2013 12:37:39 PM

and so on. Therefore, the total number of distinguishable ways = N ln N − ∑N ln N

Atkins09819.indb 500 9/11/2013 12:37:45 PM

Atkins09819.indb 501 9/11/2013 12:37:55 PM

At a minimum or maximum, δf = 0, so then

If the x i were all independent, all the δx i would be arbitrary,

subject to the two constraints

Atkins09819.indb 502 9/11/2013 12:38:07 PM

Multiplication of the first by a and the second by b and then

⎧⎛ ∂ lnW ⎞ ⎫ (b) Solving for the most probable

a box of the greatest volume possible from that sheet. What is

Atkins09819.indb 503 9/11/2013 12:38:18 PM

Now differentiation of this expression gives

Atkins09819.indb 504 9/11/2013 12:38:30 PM

Boltzmann distribution N i / N = e − βεi /q β = 1/kT 51.6a

Boltzmann ratio N i /N j = ( g i / g j )e − β (εi −ε j ) gi, gj are degeneracies 51.8

Atkins09819.indb 505 9/11/2013 12:38:32 PM

Molecular partition functions

Contents ➤ What do you need to know already?

The significance of the

Atkins09819.indb 506 9/11/2013 12:38:37 PM

Suppose a molecule is confined to the following non-degen-

Figure 52.3 The partition function for a two-level system as

3ε cance of a partition function. To do so, we first note that the

Figure 52.4 shows how pi varies with temperature. At very low

state is significantly populated. As the temperature is raised, the

〈ε R 〉 = 23 kT Nonlinear molecule, T θR; see Brief illustration 53.3