284 7 Molecular symmetry

symmetry operations. However, as explained in Topic 31, we

Criterion 4 is fulfilled, as it is immaterial how the operations
distinguish ‘symmetry operations’ from ‘symmetry elements’, are grouped together.
the axes, planes, and so on with respect to which the operation
is carried out. Finally, there is a third use of the word ‘element’, Self-test 32.1 Confirm that the operations of the group C3v,
to denote the number lying in a particular location in a matrix. which are illustrated in Fig. 32.2, form a group.
Be very careful to distinguish element (of a group), symmetry
element, and matrix element.

Example 32.1 Showing that symmetry operations form C3+ C3–

a group σv′
σv″
Show that the symmetry operations of the group C2v fulfil the
criteria for being a group in the mathematical sense. σv
Method We need to show that combinations of the operations
match the criteria set out above. The operations are set out in
Figure 32.2 The symmetry operations of the group C3v.
Topic 31.
Answer: Criteria are fulfilled
Answer Criterion 1 is fulfilled because the collection of sym-
metry operations includes the identity E. Criterion 2 is fulfilled
because in each case the inverse of an operation is the opera-
tion itself. Thus two twofold rotations are equivalent to the
identity: C2C2 = E and likewise for the two reflections and the
Symmetry operations fall into the same class if they are of
identity itself. Criterion 3 is fulfilled, because in each case one the same type (for example, rotations) and can be transformed
operation followed by another is the same as one of the four into one another by a symmetry operation of the group. The
symmetry operations. For instance, a twofold rotation C2 fol- two threefold rotations in C3v (namely C3+ and C3− ) belong to
lowed by the reflection σ v′ is the same as the single reflection the same class because one can be converted into the other by a
σv (Fig. 32.1). Thus: σ v′ C2 = σ v . The following group multipli- reflection (see Fig. 32.2); the three reflections all belong to the
cation table for the point group can be constructed similarly, same class because each can be rotated into another by a three-
where the entries are the product symmetry operations RR′: fold rotation. The formal definition of a class is that two opera-
tions R and R′ belong to the same class if there is a member S of
R↓ R′ → E C2 σv σ′v the group such that
E E C2 σv σ′v
R′ = S −1 RS Membership of a class (32.1)
C2 C2 E σ′v σv

σv σv σ′v E C2 where S−1 is the inverse of S.

σ′v σ′v σv C2 E

Brief illustration 32.1 Classes

C2
To show that C3+ and C3− belong to the same class in C 3v
(which intuitively we know to be the case), take S = σv. The
reciprocal of a reflection is the reflection itself, so σ v−1 = σ v .
It follows that

σ v−1C3+σ v = σ v C3+σ v = σ vσ v′ = C3−

σv σv′ Therefore, C3+ and C3− are related by an equation of the form
of eqn 32.1 and hence belong to the same class.

Figure 32.1 A twofold rotation C2 followed by the reflection Self-test 32.2 Show that the two reflections of the group C2v
σ v′ is the same as the single reflection σv. fall into different classes.

www.ebook3000.com

Atkins09819.indb 284 9/11/2013 11:41:03 AM

 32 Group theory 285

32.2 Matrix representations –

S +
Group theory takes on great power when the notional ideas –
presented so far are expressed in terms of collections of num- – +
bers in the form of matrices.
– B
A +
(a) Representatives of operations
Consider the set of three p orbitals shown on the C2v SO2 mol-
ecule in Fig. 32.3. Under the reflection operation σv, the change Figure 32.3 The three px orbitals that are used to illustrate
(pS,pB,pA) ← (pS,pA,pB) takes place. We can express this trans- the construction of a matrix representation in a C2v
formation by using matrix multiplication (Mathematical back- molecule (SO2).
ground 5):
D (σ )
v

⎛ 1 0 0⎞ (b) The representation of a group
(pS , pB , p A ) = (pS , p A , pB ) ⎜ 0 0 1⎟ = (pS , p A , pB )D(σ v ) The set of matrices that represents all the operations of
⎜ ⎟ (32.2)
⎜⎝ 0 1 0⎟⎠ the group is called a matrix representation, Γ (uppercase
gamma), of the group for the basis that has been chosen.
The matrix D(σv) is called a representative of the operation σv. The dimensionality of a representation is the number of
Representatives take different forms according to the basis, the rows (or columns) of each of its matrix representatives. An
set of orbitals that has been adopted. In this case, the basis is n-dimensional representation is denoted Γ(n); The matrices
(pS,pA,pB). of a representation multiply together in the same way as the
operations they represent. Thus, if for any two operations R
Brief illustration 32.2 Representatives and R′ we know that RR′ = R″, then D(R)D(R′) = D(R″) for a
given basis.
We use the same technique to find matrices that reproduce
the other symmetry operations. For instance, C2 has the effect
(–pS,–pB,–pA) ← (pS,pA,pB), and its representative is Brief illustration 32.3 Matrix representations
⎛ −1 0 0 ⎞ In the group C2v, a twofold rotation followed by a reflection in
D(C2 ) = ⎜⎜ 0 0 −1⎟⎟ a mirror plane is equivalent to a reflection in the second mir-
⎜⎝ 0 −1 0 ⎟⎠ ror plane: specifically, σ v′ C2 = σ v . When we use the represent-
atives specified in Brief illustration 32.2, we find
The effect of σ v′ is (–pS ,–pA,–pB) ← (pS ,pA,pB), and its repre-
⎛ −1 0 0 ⎞ ⎛ −1 0 0 ⎞
sentative is
D (σ v′ ) D (C2 ) = ⎜⎜ 0 −1 0 ⎟⎟ ⎜⎜ 0 0 −1⎟⎟
⎛ −1 0 0 ⎞ ⎜⎝ 0 0 −1⎟⎠ ⎜⎝ 0 −1 0 ⎟⎠
D (σ v′ ) = ⎜⎜ 0 −1 0 ⎟⎟
⎛ 1 0 0⎞
⎜⎝ 0 0 −1⎟⎠
= ⎜⎜ 0 0 1⎟⎟ = D(σ v )
⎜⎝ 0 1 0⎟⎠
The identity operation has no effect on the basis, so its repre-
sentative is the 3 × 3 unit matrix:
This multiplication reproduces the group multiplication. The
⎛ 1 0 0⎞ same is true of all pairs of representative multiplications, so
D(E ) = ⎜⎜ 0 1 0⎟⎟ the four matrices form a representation of the group.
⎜⎝ 0 0 1⎟⎠ Self-test 32.4 Confirm the result that σ vσ v′ = C2 by using the
matrix representatives developed here.
Self-test 32.3 Find the representative of the C2 operation of an
H2O molecule for the basis (H1sA,H1sB).
⎛ 0 1⎞ The discovery of a matrix representation of the group means
Answer: D(C2 ) = ⎜ ⎟
⎝ 1 0⎠ that we have found a link between symbolic manipulations of
operations and algebraic manipulations of numbers.

Atkins09819.indb 285 9/11/2013 11:41:15 AM

 286 7 Molecular symmetry

(c) Irreducible representations

Inspection of the representatives of the group C2v shows that
they are all of block-diagonal form:
⎛■ ⎞
⎜
D=⎜ ■ ■⎟⎟ Block-diagonal form
⎜⎝ ■ ■⎟⎠
The block-diagonal form of the representatives shows us that
the symmetry operations of C2v never mix pS with the other two
functions. Consequently, the basis can be cut into two parts,
one consisting of pS alone and the other of (pA,pB). It is readily
verified that the pS orbital itself is a basis for the one-dimen-
sional representation
D(E ) = 1 D(C2) = −1 D(σ v) = 1 D(σ v′ ) = −1
– B +
B
A + A –
– –
+ +
(32.3)
Figure 32.4 Two linear combinations of the basis orbitals
shown in Fig. 32.3. The two combinations each span a one-
dimensional irreducible representation and their symmetry
species are different.
Answer Under σv, pB + pA ← pA + pB and pB – pA ← pA – pB ;
therefore, under σv, (p1, − p2) ← (p1,p2). Similar analysis for the
other symmetry operations results in the following represent-
atives in the new basis (p1,p2):

⎛ 1 0⎞ ⎛ −1 0⎞
which we shall call Γ(1). The remaining two basis functions are a D( E ) = ⎜ ⎟ D(C2) = ⎜ ⎟
⎝ 0 1⎠ ⎝ 0 1⎠
basis for the two-dimensional representation Γ(2):
⎛1 0 ⎞ ⎛ −1 0 ⎞
D(σ v ) = ⎜ ⎟ D(σ v′ ) = ⎜ ⎟
⎛ 1 0⎞ ⎛ 0 −1⎞ ⎝ 0 −1⎠ ⎝ 0 −1⎠
D( E ) = ⎜ ⎟ D(C2 ) = ⎜ ⎟
⎝ 0 1⎠ ⎝ −1 0 ⎠ The new representatives are all in block-diagonal form, and the
⎛ 0 1⎞ ⎛ −1 0 ⎞ two combinations are not mixed with each other by any opera-
D(σ v ) = ⎜ ⎟ D(σ v′ ) = ⎜ ⎟ tion of the group. The reduction of Γ (2) has been reduced to the
⎝ 1 0⎠ ⎝ 0 −1⎠
sum of two one-dimensional representations. Thus, p1 spans
These matrices are the same as those of the original three- D(E ) = 1 D(C2) = −1 D(σ v ) = 1 D(σ v′ ) = −1
dimensional representation, except for the loss of the first row
and column. We say that the original three-dimensional repre- which is the same one-dimensional representation as that
sentation has been reduced to the ‘direct sum’ of a one-dimen- spanned by pS, and p2 spans
sional representation ‘spanned’ by pS, and a two-dimensional
D(E )=1 D(C2) = 1 D(σ v ) = −1 D(σ v′ ) = −1
representation spanned by (pA,pB). This reduction is consist-
ent with the common-sense view that the central orbital plays
which is a different one-dimensional representation; we
a role different from the other two. We denote the reduction
denote these two representations Γ (1) and Γ (1) ′, respectively.
symbolically by writing At this stage we have reduced the original representation as
Γ (3) = Γ (1) + Γ (2) Direct sum (32.4) follows:

The one-dimensional representation Γ(1) cannot be reduced
any further, and is called an irreducible representation of the
group (an ‘irrep’).
Example 32.2 Reducing a representation
Demonstrate that the two-dimensional representation Γ (2)
is reducible for the basis (pA,pB) in the group C2v by switching
attention to the linear combinations p1 = pA + pB and p2 = pA – pB.
These combinations are sketched in Fig. 32.4.
Method The representatives in the new basis can be con-
structed from the old by noting, for example, that under σv,
(pB,pA) ← (pA,pB).
Γ (3) = Γ (1) + Γ (1) + Γ (1) ′
Self-test 32.5 Consider the H 2O molecule and focus on the
two H1s orbitals. Is the two-dimensional representation on
the basis (H1sA,H1sB) reducible?
Answer: Yes; adopt H1sA + H1sB, H1sA – H1sB.
(d) Characters and symmetry species
The character, χ (chi), of an operation in a particular matrix
representation is the sum of the diagonal elements of the rep-
resentative of that operation. Thus, in the original basis we are
using, the characters of the representatives are

www.ebook3000.com

Atkins09819.indb 286 9/11/2013 11:41:24 AM

 32 Group theory 287

R E C2 σv σ′v In C2v, for instance, there are four classes (four columns in
the table of characters), so there are only four species of
D(R)
⎛ 1 0 0⎞ ⎛ −1 0 0 ⎞ ⎛ 1 0 0⎞ ⎛ −1 0 0 ⎞ irreducible representation. The character table in Table 32.1
⎜ ⎟ ⎜0 0 ⎟ ⎜ ⎟ ⎜0 ⎟
⎜ 0 1 0⎟ ⎜ −1⎟ ⎜ 0 0 1⎟ ⎜ −1 0 ⎟ therefore shows the characters of all the irreducible repre-
⎜⎝ 0 0 1⎟⎠ ⎜⎝ 0 −1 0 ⎟⎠ ⎜⎝ 0 1 0⎟⎠ ⎜⎝ 0 0 −1⎟⎠ sentations of this group. Another powerful result relates the
sum of the dimensions, di, of all the symmetry species Γ(i) to
χ(R) 3 −1 1 −3
the order of the group, the total number of symmetry opera-
tions, h:
The characters of one-dimensional representatives are just the
representatives themselves. The sum of the characters of the ∑d
Species i
2
i =h Dimensionality and order (32.6)
reduced representation is unchanged by the reduction:

R E C2 σv σ′v Brief illustration 32.4 Symmetry species

χ(R) for Γ (1) 1 −1 1 −1 There are three classes of operation in the group C3v (E,C3,σv:
χ(R) for Γ (1) 1 −1 1 −1 we don’t need to know the number in each class), so there are
three symmetry species (they turn out to be A1, A 2 , and E).
χ(R) for Γ (1)′ 1 1 −1 −1
The order of the group is 6 (we now need to know that the
Sum: 3 −1 1 −3 elements of the group (that is, the symmetry operations) are
(E,2C3,3σv), for h = 6), so if we already knew that two of the
At this point we have found two irreducible representations symmetry species are one-dimensional, we could infer that
of the group C2v. Although the notation Γ(n) can be used for the remaining irreducible representation is two-dimensional
general representations, it is common in chemical applications (E) from 12 + 12 + d2 = 6.
of group theory to use the labels A, B, E, and T to denote the Self-test 32.6 How many symmetry species are there for the
symmetry species of the representation: group Td , with elements (E,8C3,3C2 ,6σ d ,6S 4)? Can you infer
their dimensionalities?
A: one-dimensional representation, character +1 under
Answer: five species; 2A + E + 2T, 2 × (1)2 + 1 × (2)2 + 2 × (3)2 = 24 = h
the principal rotation
B: one-dimensional representation, character –1 under
the principal rotation
E: two-dimensional irreducible representation 32.3 Character tables
T: three-dimensional irreducible representation
The tables we have been constructing are called character
Subscripts are used to distinguish the irreducible representa- tables and from now on move to the centre of the discussion.
tions if there is more than one of the same type: A1 is reserved The columns of a character table are labelled with the symmetry
for the representation with character +1 for all operations. All operations of the group. For instance, for the group C3v the col-
the irreducible representations of C2v are one-dimensional, and umns are headed E, C3, and σv (Table 32.2). The numbers mul-
the table above is labelled as follows: tiplying each operation are the numbers of members of each
class. The rows under the labels for the operations summarize
Symmetry species E C2 σv σ′v the symmetry properties of the orbitals. They are labelled with
the symmetry species.
B1 1 –1 1 –1
B1 1 –1 1 –1
A2 1 1 –1 –1 Table 32.1* The C2v character table

C2v, 2mm E C2 σv σ′v h=4

Are these the only irreducible representations of the group C2v?
There is in fact only one more species of irreducible representa- A1 1 1 1 1 z z2, y2, x2
tions of this group, for a surprising theorem of group theory
A2 1 1 –1 –1 xy
states that
B1 1 –1 1 –1 x zx
Number of symmetry species B2 1 –1 –1 1 y yz
Number of species (32.5)
= number of classes
* More character tables are given in the Resource section.

Atkins09819.indb 287 9/11/2013 11:41:29 AM

 288 7 Molecular symmetry

Table 32.2* The C3v character table

Self-test 32.7 A buckminsterfullerene molecule, C 60 (19 of
C3v, 3m E 2C3 3σv h=6 Topic 31), belongs to the icosahedral point group. What is the
maximum possible degree of degeneracy of its orbitals?
A1 1 1 1 z z2, x2 + y2
Answer: 5
A2 1 1 –1
E 2 –1 0 (x, y) (xy, x2 – y2), (yz, zx)
* More character tables are given in the Resource section.
(b)The symmetry species of
atomic orbitals
(a) Character tables and orbital degeneracy The characters in the rows labelled A and B and in the col-
The character of the identity operation E tells us the degen- umns headed by symmetry operations other than the identity
eracy of the orbitals. Thus, in a C3v molecule, any orbital with E indicate the behaviour of an orbital under the corresponding
a symmetry label A1 or A2 is non-degenerate. Any doubly operations: a +1 indicates that an orbital is unchanged, and a –1
degenerate pair of orbitals in C3v must be labelled E because, indicates that it changes sign. It follows that we can identify the
in this group, only E symmetry species have characters symmetry label of the orbital by comparing the changes that
greater than 1. (Take care to distinguish the identity opera- occur to an orbital under each operation, and then comparing
tion E (italic, a column heading) from the symmetry label E the resulting +1 or –1 with the entries in a row of the character
(roman, a row label).) table for the point group concerned. By convention, irreduc-
Because there are no characters greater than 2 in the column ible representations are labelled with uppercase Roman let-
headed E in C3v, we know that there can be no triply degenerate ters (such as A1 and E) and the orbitals to which they apply are
orbitals in a C3v molecule. This last point is a powerful result of labelled with the lowercase equivalents (so an orbital of sym-
group theory, for it means that with a glance at the character metry species A1 is called an a1 orbital). Examples of each type
table of a molecule, we can state the maximum possible degen- of orbital are shown in Fig. 32.5.
eracy of its orbitals.

Example 32.3 Using a character table to judge sN

degeneracy
Can a trigonal planar molecule such as BF 3 have triply
a1 a2
degenerate orbitals? What is the minimum number of atoms
from which a molecule can be built that does display triple
degeneracy?
Method First identify the point group, and then refer to the
corresponding character table in the Resource section. The
maximum number in the column headed by the identity E e
is the maximum orbital degeneracy possible in a molecule
Figure 32.5 Typical linear combinations of orbitals in a C3v
of that point group. For the second part, consider the shapes
molecule.
that can be built from two, three, etc. atoms, and decide which
number can be used to form a molecule that can have orbitals
of symmetry species T. Brief illustration 32.5 Symmetry species of atomic orbitals
Answer Trigonal planar molecules belong to the point group Consider the O2px orbital in H2O (the x-axis is perpendicular to
D3h. Reference to the character table for this group shows that the molecular plane; the y-axis is parallel to the H–H direction;
the maximum degeneracy is 2, as no character exceeds 2 in the z-axis bisects the HOH angle). Because H2O belongs to the
the column headed E. Therefore, the orbitals cannot be triply point group C2v, we know by referring to the C2v character table
degenerate. A tetrahedral molecule (symmetry group T) has (Table 32.1) that the labels available for the orbitals are a1, a2, b1,
an irreducible representation with a T symmetry species. The and b2. We can decide the appropriate label for O2px by noting
minimum number of atoms needed to build such a molecule is that under a 180° rotation (C2) the orbital changes sign (Fig.
four (as in P4, for instance). 32.6), so it must be either B1 or B2, as only these two symmetry

www.ebook3000.com

Atkins09819.indb 288 9/11/2013 11:41:30 AM


C2
+ symmetry species of x, y, and z, which appear on the right-hand
– side of the character table. Thus, the position of z in Table 32.2
σv′
σv
Figure 32.6 A px orbital on the central atom of a C2v
molecule and the symmetry elements of the group.
types have character –1 under C2. The O2px orbital also changes
sign under the reflection σ v′ ; which identifies it as B1. As we
shall see, any molecular orbital built from this atomic orbital will
also be a b1 orbital. Similarly, O2py changes sign under C2 but
not under σ v′ ; therefore, it can contribute to b2 orbitals.
Self-test 32.8 Identify the symmetry species of d orbitals on
the central atom of a square-planar (D4h) complex.
Answer: A1g + B1g + B2g + Eg
For the rows labelled E or T (which refer to the behaviour
of sets of doubly and triply degenerate orbitals, respectively),
the characters in a row of the table are the sums of the charac-
ters summarizing the behaviour of the individual orbitals in the
basis. Thus, if one member of a doubly degenerate pair remains
unchanged under a symmetry operation but the other changes
sign (Fig. 32.7), then the entry is reported as χ = 1 – 1 = 0. Care
must be exercised with these characters because the transfor-
mations of orbitals can be quite complicated; nevertheless, the
sums of the individual characters are integers.
+
+ sA
–
–1 +1
Figure 32.7 The two orbitals shown here have different
properties under reflection through the mirror plane: one
changes sign (character –1), the other does not (character +1).
Atkins09819.indb 289
32 Group theory 289
The behaviour of s, p, and d orbitals on a central atom under
the symmetry operations of the molecule is so important that
the symmetry species of these orbitals are generally indicated
in a character table. To make these allocations, we look at the
shows that pz (which is proportional to zf(r)) has symmetry
species A1 in C3v, whereas px and py (which are proportional to
xf(r) and yf(r), respectively) are jointly of E symmetry. In tech-
nical terms, we say that px and py jointly span an irreducible
representation of symmetry species E. An s orbital on the cen-
tral atom always spans the fully symmetrical irreducible repre-
sentation (typically labelled A1 but sometimes A1′ ) of a group
as it is unchanged under all symmetry operations.
The five d orbitals of a shell are represented by xy for dxy, etc.,
and are also listed on the right of the character table. We can see
at a glance that in C3v, dxy and d x − y on a central atom jointly
2 2
belong to E and hence form a doubly degenerate pair.
The symmetry species of linear
(c)
combinations of orbitals
So far, we have dealt with the symmetry classification of indi-
vidual orbitals. The same technique may be applied to linear
combinations of orbitals on atoms that are related by sym-
metry transformations of the molecule, such as the linear
combinations p1 and p2 of Example 32.2 and the combination
ψ1 = ψA + ψB + ψC of the three H1s orbitals in the C3v molecule
NH3 (Fig. 32.8). This latter combination remains unchanged
under a C3 rotation and under any of the three vertical reflec-
tions of the group, so its characters are
χ ( E ) = 1 χ (C3 ) = 1 χ (σ v ) = 1
Comparison with the C3v character table shows that ψ1 is of
symmetry species A1, and therefore that it contributes to a1
molecular orbitals in NH3.
sC sB
Figure 32.8 The three H1s orbitals used to construct linear
combinations in a C3v molecule such as NH3.
9/11/2013 11:41:39 AM
 290 7 Molecular symmetry

Example 32.4 Identifying the symmetry species N

+
of orbitals –
O
Identify the symmetry species of the orbital ψ = ψ A – ψ B in a –
O +
C2v NO2 molecule, where ψ A is an O2px orbital on one O atom
and ψ B that on the other O atom.
Method The negative sign in ψ indicates that the sign of ψ B is Figure 32.9 One linear combination of O2px orbitals in the
opposite to that of ψ A. We need to consider how the combina- C2v NO2 molecule.
tion changes under each operation of the group, and then write
the character as +1, –1, or 0 as specified above. Then we compare Self-test 32.9 Consider PtCl −4 , in which the Cl ligands form a
the resulting characters with each row in the character table for square planar array of point group D4h (1). Identify the sym-
the point group, and hence identify the symmetry species. metry type of the combination ψ A – ψ B + ψC – ψ D.
Answer The combination is shown in Fig. 32.9. Under C 2 ,
ψ changes into itself, implying a character of +1. Under the A B
reflection σv, both orbitals change sign, so ψ → –ψ, implying
a character of –1. Under σ v′ , ψ → –ψ, so the character for this D C
operation is also –1. The characters are therefore 1

χ (E ) = 1 χ (C2)=1 χ (σ v ) = −1 χ (σ v′ ) = −1 Answer: B2g

These values match the characters of the A 2 symmetry species,

so ψ can contribute to an a 2 orbital.

Checklist of concepts
☐ 1. A group in mathematics is a collection of transforma-
tions that satisfy the four criteria set out at the start of
the Topic.
☐ 2. A matrix representative is a matrix that represents the
effect of an operation on a basis.
☐ 3. The character is the sum of the diagonal elements of a
matrix representative of an operation.
☐ 4. A matrix representation is the collection of matrix rep-
resentatives for the operations in the group.
☐ 5. A character table consists of entries showing the char-
acters of all the irreducible representations of a group.
☐ 6. A symmetry species is a label for an irreducible repre-
sentation of a group.
☐ 7. The character of the identity operation E is the degen-
eracy of the orbitals that form a basis for an irreducible
representation of a group.

Checklist of equations
Property Equation Comment Equation number

Class membership R′ = S−1RS All elements members of the group; 32.1

R and R′ in same class

Number of species rule Number of symmetry species = number of classes 32.5

∑ d =h
Character and order h is the order of the group 32.6
2
i
Species i

www.ebook3000.com

Atkins09819.indb 290 9/11/2013 11:41:47 AM

 TOPIC 33

Applications of symmetry
Group theory shows its power when brought to bear on a vari-
Contents ety of problems in chemistry, among them the construction of
33.1 Vanishing integrals 291 molecular orbitals and the formulation of spectroscopic selec-
(a) Integrals over the product of two functions 292 tion rules. This Topic describes these two applications after
Example 33.1: Deciding if an integral must be zero 1 292 establishing a general result relating to integrals. In Topics 6
(b) Decomposition of a direct product 293 and 7 it is explained how integrals (‘matrix elements’) are cen-
Brief illustration 33.1: Decomposition of a tral to the formulation of quantum mechanics, and knowing
direct product 293 with very little calculation that various integrals are necessarily
(c) Integrals over products of three functions 293 zero can save a great deal of calculational effort as well as add-
Example 33.2: Deciding if an integral must be zero 2 293 ing to insight about the origin of properties.
33.2 Applications to orbitals 294
(a) Orbital overlap 294
Example 33.3: Determining which orbitals can
contribute to bonding 294 33.1 Vanishing integrals
(b) Symmetry-adapted linear combinations 294
Example 33.4: Constructing symmetry- An integral, which we shall denote I, in one dimension is equal
adapted orbitals 295 to the area beneath the curve. In higher dimensions, it is equal
33.3 Selection rules 295 to volume and various generalizations of volume. The key point
Example 33.5: Deducing a selection rule 295 is that the value of the area, volume, etc., is independent of the
Checklist of concepts 296 orientation of the axes used to express the function being inte-
Checklist of equations 296 grated, the ‘integrand’ (Fig. 33.1). In group theory we express
this point by saying that I is invariant under any symmetry
operation, and that each symmetry operation brings about the
trivial transformation I → I.
➤ Why do you need to know this material?
This Topic explains how the concepts introduced in Topics
31 and 32 are put to use. The arguments here are essential y y
for understanding how molecular orbitals are constructed
and underlie the whole of spectroscopy.
x x
➤ What is the key idea?
An integral is invariant under symmetry transformations
of a molecule.

➤ What do you need to know already? (a) (b)

This Topic develops the material that began in Topic
31, where the symmetry classification of molecules is
introduced on the basis of their symmetry elements,
and draws heavily on the properties of characters and
character tables described in Topic 32.
Figure 33.1 The value of an integral I (for example, an area) is
independent of the coordinate system used to evaluate it: the
dark shaded regions have the same area in (a) and (b). That is,
I is a basis of a representation of symmetry species A1 (or its
equivalent).

Atkins09819.indb 291 9/11/2013 11:41:49 AM

 292 7 Molecular symmetry

(a) Integrals over the product of two functions Example 33.1 Deciding if an integral must be zero 1
Suppose we had to evaluate the integral May the integral of the function f = xy be nonzero when evalu-
ated over a region the shape of an equilateral triangle centred
∫
I = f1 f 2 dτ (33.1) on the origin (Fig. 33.2)?

where f1 and f2 are functions and the integration is over all space.
For example, f1 might be an atomic orbital A on one atom and f2
an atomic orbital B on another atom, in which case I would be
their overlap integral (denoted S). If we knew that the integral
is zero, we could say at once that a molecular orbital does not
result from (A,B) overlap in that molecule. We shall now see
that the character tables introduced in Topic 32 provide a quick
way of judging whether an integral is necessarily zero.
The volume element dτ is invariant under any symmetry
operation. It follows that the integral is nonzero only if the inte-
grand itself, the product f1 f2, is unchanged by any symmetry
operation of the molecular point group. If the integrand changed
sign under a symmetry operation, the integral would be the sum
of equal and opposite contributions, and hence would be zero. It
follows that the only contribution to a nonzero integral comes
from functions for which under any symmetry operation of the
molecular point group f1 f2→ f1 f2, and hence for which the char-
acters of the operations are all equal to +1. Therefore, for I not to
be zero, the integrand f1 f2 must have symmetry species A1 (or its
equivalent in the specific molecular point group).
The following procedure is used to deduce the symmetry
species spanned by the product f1 f2 and hence to see whether it
does indeed span A1:
r
Identify the symmetry species of the individual
functions f1 and f2 by reference to the character table for
the molecular point group in question and write their
characters in two rows in the same order as in the table.
– +
y
x
+ –
Figure 33.2 The integral of the function f = xy over the
tinted region is zero. In this case, the result is obvious by
inspection, but group theory can be used to establish
similar results in less obvious cases. The insert shows the
shape of the function in three dimensions.
Method First, note that an integral over a single function f is
included in the previous discussion if we take f1 = f and f2 = 1 in
eqn 33.1. Therefore, we need to judge whether f alone belongs
to the symmetry species A1 (or its equivalent) in the point
group of the system. To decide, we identify the point group
and then examine the character table to see whether f belongs
to A1 (or its equivalent).
Answer An equilateral triangle has the point-group symme-
try D3h. If we refer to the character table of the group, we see
that xy is a member of a basis that spans the irreducible repre-
sentation E′. Therefore, its integral must be zero, because the
integrand has no component that spans A1′.

r
Multiply the two numbers in each column, writing the
results in the same order.
r
Inspect the row so produced, and see if it can be expressed
as a sum of characters from each column of the group.
The integral must be zero if this sum does not use A1.
Self-test 33.1 Can the function x 2 + y2 have a nonzero integral
when integrated over a regular pentagon centred on the origin?
Answer: Yes (Fig. 33.3)

A shortcut that works when f1 and f2 are bases for irreducible

representations of a group is to note their symmetry species;
y
if they are different (B1 and A2, for instance), then the integral x
of their product must vanish; if they are the same (both B1, for
instance), then the integral may be nonzero.
It is important to note that group theory is specific about
when an integral must be zero, but integrals that it allows to
be nonzero may be zero for reasons unrelated to symmetry.
For example, the NeH distance in ammonia may be so great Figure 33.3 The integration of a function over a pentagonal
that the (s1,sN) overlap integral, where s1 is the combination region. The insert shows the shape of the function in three
sA + sB + sC of the three H1s atomic orbitals, is zero simply dimensions.
because the orbitals are so far apart.

www.ebook3000.com

Atkins09819.indb 292 9/11/2013 11:41:53 AM

 33 Applications of symmetry  293

(b) Decomposition of a direct product Brief illustration 33.1 Decomposition of a direct product
In many cases, the product of functions f1 and f2 spans a sum To find whether A1 does indeed occur in the product with
of irreducible representations. For instance, in C2v we may find characters 8, −2, −6,4 in C2v, we draw up the following table:
the characters 2,0,0,−2 when we multiply the characters of f1
E C2v σv σ ′v h = 4 (the order of the group)
and f2 together. In this case, we note that these characters are
the sum of the characters for A2 and B1: f1f2 8 −2 −6 4 (the characters of the product)
A1 1 1 1 1 (the symmetry species we are interested in)
8 −2 −6 4 (the product of the two sets of characters)
E C2v σv σ v′
A2 1 1 −1 −1 The sum of the numbers in the last line is 4; when that
B1 1 −1 1 −1 number is divided by the order of the group, we get 1, so A1
A2 + B1 2 0 0 −2 occurs once in the decomposition. When the procedure is
repeated for all four symmetry species, we find that f1 f2 spans
A1 + 2A 2 + 5B2.
To summarize this result we write the symbolic expression Self-test 33.2 Does A 2 occur among the symmetry species
A2 × B1 = A2 + B1, which is called the decomposition of a direct of the irreducible representations spanned by a product with
product. This expression is symbolic. The × and + signs in characters 7,−3,−1,5 in the group C2v?
this expression are not ordinary multiplication and addition Answer: No
signs: formally, they denote technical procedures with matri-
ces called a ‘direct product’ and a ‘direct sum’. Because the sum
on the right does not include a component that is a basis for
an irreducible representation of symmetry species A1, we can
(c) Integrals over products of three functions
conclude that the integral of f1 f2 over all space is zero in a C2v Integrals of the form
molecule.
Whereas the decomposition of the characters 2,0,0,−2 can be
done by inspection in this simple case, in other cases and more
∫
I = f1 f 2 f 3 dτ (33.3)

complex groups the decomposition is often far from obvious.
For example, if we found the characters 8,−2,−6,4, it might not
be obvious that the sum contains A1. Group theory, however,
provides a systematic way of using the characters of the repre-
sentation spanned by a product to find the symmetry species of
the irreducible representations. The formal recipe is
are also common in quantum mechanics for they include
matrix elements of operators (Topic 7), and it is important to
know when they are necessarily zero. As for integrals over two
functions, for I to be nonzero, the product f1 f2 f3 must span A1
(or its equivalent) or contain a component that spans A1. To test
whether this is so, the characters of all three functions are multi-
plied together in the same way as in the rules set out above.

∑χ
1
n(Γ ) = (Γ )
(R) χ (R) Decomposition of direct product (33.2)
h R
Example 33.2 Deciding if an integral must be zero 2
We implement this expression as follows:
Does the integral ∫(3d z 2 )x(3d xy )dτ vanish in a C2v molecule?
r
Write down a table with columns headed by the
Method We must refer to the C2v character table (Table 32.1)
symmetry operations, R, of the group. Include a column
and the characters of the irreducible representations spanned by
for every operation, not just the classes.
3z2 – r2 (the form of the d z 2 orbital), x, and xy; then we can use the
r
In the first row write down the characters of the procedure set out above (with one more row of multiplication).
representation we want to analyse; these are the χ(R).
Answer We draw up the following table:
r
In the second row, write down the characters of the
irreducible representation Γ we are interested in; these E C2 σv σ ′v
are the χ(Γ)(R). f3 = dxy 1 1 −1 −1 A2
r
Multiply the two rows together, add the products f2 = x 1 −1 1 −1 B1
together, and divide by the order of the group, h. f1 = d z 2 1 1 1 1 A1

The resulting number, n(Γ), is the number of times Γ occurs in f1 f2 f3 1 −1 −1 1

the decomposition.

Atkins09819.indb 293 9/11/2013 11:42:00 AM

 294 7 Molecular symmetry
The characters are those of B2. Therefore, the integral is neces-
sarily zero.
Self-test 33.3 Does the integral ∫(2p x )(2p y )(2pz )dτ necessar-
ily vanish in an octahedral environment?
Answer: No
33.2 Applications to orbitals
The rules we have outlined let us decide which atomic orbitals
may have nonzero overlap in a molecule. It is also very useful
to have a set of procedures to construct linear combinations of
atomic orbitals to have a certain symmetry, and thus to know
in advance whether or not they will have nonzero overlap with
other orbitals.
(a) Orbital overlap
An overlap integral, S, between two sets of atomic orbitals ψ1
and ψ2 is
∫
S = ψ 2*ψ 1dτ Overlap integral (33.4)
and clearly has the same form as eqn 33.1. It follows from that
discussion that only orbitals of the same symmetry species may
have nonzero overlap (S ≠ 0), so only orbitals of the same sym-
metry species form bonding and antibonding combinations. It
is explained in Topic 23 that the selection of atomic orbitals that
had mutual nonzero overlap is the central and initial step in the
construction of molecular orbitals by the LCAO procedure. We
are therefore at the point of contact between group theory and
the material introduced in that Topic.
Example 33.3
Determining which orbitals can
contribute to bonding
The four H1s orbitals of methane span A1 + T 2 . With which
of the C atomic orbitals can they overlap? What bond-
ing pattern would be possible if the C atom had d orbitals
available?
Method Refer to the Td character table (in the Resource sec-
tion) and look for s, p, and d orbitals spanning A1 or T2.
Answer An s orbital spans A1, so it may have nonzero over-
lap with the A1 combination of H1s orbitals. The C2p orbit-
als span T 2 , so they may have nonzero overlap with the T 2
combination. The d xy, dyz , and d zx orbitals span T 2 , so they
may overlap the same combination. Neither of the other two
d orbitals span A1 (they span E), so they remain nonbonding
www.ebook3000.com
Atkins09819.indb 294
orbitals. It follows that in methane there are (C2s,H1s)-
overlap a1 orbitals and (C2p,H1s)-overlap t 2 orbitals. The
C3d orbitals might contribute to the latter. The lowest energy
configuration is probably a12 t 62 , with all bonding orbitals
occupied.
Self-test 33.4 Consider the octahedral SF6 molecule, with the
bonding arising from overlap of S orbitals and a 2p orbital on
each F directed towards the central S atom. The latter spans
A1g + Eg + T1u. What s orbitals have nonzero overlap? Suggest
what the ground-state configuration is likely to be.
2 6 4
Answer: 3s(A1g), 3p(T1u), 3d(Eg); a1g t1u eg
(b) Symmetry-adapted linear combinations
In the discussion of the molecular orbitals of NH3
(Topic 32) we encounter molecular orbitals of the form
ψ = c1sN + c2(sA + sB + sC), where sN is an N2s atomic orbital
and sA, sB, and sC are H1s orbitals. The sN orbital has nonzero
overlap with the combination of H1s orbitals as the latter has
matching symmetry. The combination of H1s orbitals is an
example of a symmetry-adapted linear combination (SALC),
which are orbitals constructed from equivalent atoms and
having a specified symmetry. Group theory also provides
machinery that takes an arbitrary basis, or set of atomic orbit-
als (sA, etc.), as input and generates combinations of the speci-
fied symmetry. As illustrated by the example of NH3, SALCs
are the building blocks of LCAO molecular orbitals and their
construction is the first step in any molecular orbital treat-
ment of molecules.
The technique for building SALCs is derived by using the
full power of group theory and involves the use of a projection
operator, P(Γ), an operator that takes one of the basis orbitals and
generates from it—projects from it—a SALC of the symmetry
species Γ:
∑χ
1
P (Γ ) = (Γ )
(R)R Projection operator (33.5)
h R
To implement this rule, do the following:
r
Write each basis orbital at the head of a column
and in successive rows show the effect of each
operation R on each orbital. Treat each operation
individually.
r
Multiply each member of the column by the character,
χ(Γ)(R), of the corresponding operation.
r
Add together all the orbitals in each column with the
factors as determined in the second step.
r
Divide the sum by the order of the group, h.
9/11/2013 11:42:03 AM
 33 Applications of symmetry  295

We now form the overall molecular orbital by forming a linear The other columns give
combination of all the SALCs of the specified symmetry species.
In this case, therefore, the a1 molecular orbital is ψ = cNsN + c1s1,
1
6
(2s A − s B − sC ) 1
6
(2s B − s A − sC ) 1
6
(2sC − s B − s A )
as specified above. This is as far as group theory can take us. The
coefficients are found by solving the Schrödinger equation; they However, any one of these three expressions can be expressed as
do not come directly from the symmetry of the system. a sum of the other two (they are not ‘linearly independent’). The
difference of the second and third gives 12 (s B − sC ) , and this com-
Example 33.4 Constructing symmetry-adapted orbitals bination and the first, 16 (2s A − s B − sC ) are the two (now linearly
independent) SALCs we have used in the discussion of e orbitals.
Construct symmetry-adapted linear combinations of H1s
orbitals for NH3.
Method Identify the point group of the molecule and have 33.3 Selection rules
available its character table. Then apply the projection-oper-
ator technique. It is explained in Topic 16 and developed further in Topic 45 that
Answer From the (sN,sA,sB,sC) basis in NH3 we form the fol- the intensity of a spectral line arising from a molecular transi-
lowing table, with each row showing the effect of the operator tion between some initial state with wavefunction ψi and a final
shown on the left: state with wavefunction ψf depends on the (electric) transition
dipole moment, μfi. The z-component of this vector is defined
sN sA sB sC through

∫
E sN sA sB sC
μz , fi = − e ψ f* zψ i dτ Transition dipole moment (33.6)
C3+ sN sB sC sA
C3− sN sC sA sB
where −e is the charge of the electron. The transition moment
σv sN sA sC sB has the form of the integral in eqn 33.3; so, once we know the
σ v′ sN sB sA sC symmetry species of the states, we can use group theory to for-
σ v′′ sN sC sB sA mulate the selection rules for the transitions.

To generate the A1 combination, we take the characters Example 33.5 Deducing a selection rule
for A1 (1,1,1,1,1,1); then the second and third rules lead to
ψ ∝ sN + sN + …= 6sN. The order of the group (the number of Is px → py an allowed transition in a tetrahedral environment?
elements) is 6, so the combination of A1 symmetry that can be Method We must decide whether the product p yqp x , with
generated from sN is sN itself. Applying the same technique to q = x, y, or z, spans A1 by using the Td character table.
the column under sA gives
Answer The procedure works out as follows:
ψ= 1
(s
6 A
+ s B + sC + s A + s B + sC ) = 1
(s
3 A
+ s B + sC )
E 8C3 3C2 6σd 6S4

The same combination is built from the other two columns, f3(py) 3 0 −1 1 −1 T2
so they give no further information. The combination we have f2(q) 3 0 −1 1 −1 T2
just formed is the one we see above and denote s1 (apart from f1(px) 3 0 −1 1 −1 T2
the numerical factor). f1f2f3 27 0 −1 1 −1

Self-test 33.5 Generate the symmetry-adapted linear combi-

nations of the H1s orbitals in H2O.
Answer: H1sA + H1sB, H1sA – H1sB
We run into a problem when we try to generate an SALC of
symmetry species E, because, for representations of dimension
2 or more, the rules generate sums of SALCs. This problem can
be illustrated as follows. In C3v, the E characters are 2, −1, −1, 0,
0, 0, so the column under sN gives
ψ = 16 (2s N − s N − s N + 0 + 0 + 0) = 0
We now use the decomposition procedure summarized by eqn
33.2 to deduce that A1 occurs (once) in this set of characters, so
px → py is allowed. A more detailed analysis (using the matrix
representatives rather than the characters) shows that only q = z
gives an integral that may be nonzero, so the transition is z-polar-
ized. That is, the electromagnetic radiation involved in the tran-
sition has a component of its electric vector in the z-direction.
Self-test 33.6 What are the allowed transitions, and their
polarizations, of an electron in a b1 orbital in a C 4v molecule?
Answer: b1 → b1(z); b1 → e(x,y)

Atkins09819.indb 295 9/11/2013 11:42:11 AM

 296 7 Molecular symmetry

Checklist of concepts
☐ 1. For an integral not to be zero, the integrand must have ☐ 3. A symmetry-adapted linear combination (SALC) is a
symmetry species A1 (or its equivalent in the specific linear combination of atomic orbitals constructed from
molecular point group). equivalent atoms and having a specified symmetry.
☐ 2. Only orbitals of the same symmetry species may have
nonzero overlap (S ≠ 0).

Checklist of equations
Property Equation Comment Equation number

Decomposition of direct product n(Γ ) = (1/ h) ∑χ

R
( Γ ) ( R) χ ( R)
Real characters* 33.2

Overlap integral S = ∫ ψ 2*ψ 1dτ Definition 33.4

Projection operator P (Γ ) = (1/ h) ∑χR

(Γ ) (R)R
33.5

Transition dipole moment μz , fi = −e ∫ ψ f* zψ i dτ z-component 33.6

* In general, characters may have complex values; throughout this text we encounter only real values.

www.ebook3000.com

Atkins09819.indb 296 9/11/2013 11:42:15 AM

 Exercises and problems  297

Focus 7 on Molecular symmetry

Topic 31 The analysis of molecular shape

Discussion questions
31.1 Explain how a molecule is assigned to a point group. 31.3 State and explain the symmetry criteria that allow a molecule to be polar.

31.2 List the symmetry operations and the corresponding symmetry elements 31.4 State the symmetry criteria that allow a molecule to be optically active.
of the point groups.

Exercises
31.1(a) The CH3Cl molecule belongs to the point group C3v. List the symmetry 31.5(a) Assign (a) cis-dichloroethene and (b) trans-dichloroethene to point
elements of the group and locate them in a drawing of the molecule. groups.
31.1(b) The CCl4 molecule belongs to the point group Td. List the symmetry 31.5(b) Assign the following molecules to point groups: (a) HF, (b) IF7
elements of the group and locate them in a drawing of the molecule. (pentagonal bipyramid), (c) XeO2F2 (see-saw), (d) Fe2(CO)9 (3), (e) cubane,
C8H8, (f) tetrafluorocubane, C8H4F4 (4).
31.2(a) Identify the group to which the naphthalene molecule belongs and
locate the symmetry elements in a drawing of the molecule.
31.2(b) Identify the group to which the anthracene molecule belongs and H
locate the symmetry elements in a drawing of the molecule. CO
Fe
31.3(a) Identify the point groups to which the following objects belong: CO
(a) a sphere, (b) an isosceles triangle, (c) an equilateral triangle, (d) an
unsharpened cylindrical pencil.
31.3(b) Identify the point groups to which the following objects belong: (a) a F
sharpened cylindrical pencil, (b) a three-bladed propeller, (c) a four-legged
table, (d) yourself (approximately). 3 4

31.4(a) List the symmetry elements of the following molecules and name 31.6(a) Which of the following molecules may be polar? (a) pyridine,
the point groups to which they belong: (a) NO2, (b) N2O, (c) CHCl3, (b) nitroethane, (c) gas-phase HgBr2, (d) B3N3H6.
(d) CH2 = CH2. 31.6(b) Which of the following molecules may be polar? (a) CH3Cl,
31.4(b) List the symmetry elements of the following molecules and name (b) HW2(CO)10, D4h (c) SnCl4.
the point groups to which they belong: (a) furan (1), (b) γ-pyran (2),
31.7(a) Identify the point groups to which all isomers of dichloronaphthalene
(c) 1,2,5-trichlorobenzene.
belong.
31.7(b) Identify the point groups to which all isomers of dichloroanthracene
belong.
31.8(a) Can molecules belonging to the point groups D2h or C3h be chiral?
O O Explain your answer.
1 Furan 2 γ-Pyran 31.8(b) Can molecules belonging to the point groups Th or Td be chiral?
Explain your answer.

Problems
31.1 List the symmetry elements of the following molecules and name the complex if each CF3 group has a CF bond in that plane (so the CF3 groups
point groups to which they belong: (a) staggered CH3CH3, (b) chair and do not point to either CN group preferentially) and the CF3 groups are
boat cyclohexane, (c) B2H6, (d) [Co(en)3]3+, where en is ethylenediamine (i) staggered, (ii) eclipsed.
(1,2-diaminoethane; ignore its detailed structure), (e) crown-shaped S8.
Which of these molecules can be (i) polar, (ii) chiral?
CF3
31.2a‡ In the square-planar complex anion [trans-Ag(CF3)2(CN)2]−, the Ag–
CN groups are collinear. (a) Assume free rotation of the CF3 groups (that is, Ag
disregarding the AgCF and AgCH angles) and name the point group of this
complex ion. (b) Now suppose the CF3 groups cannot rotate freely (because
CN
the ion was in a solid, for example). Structure (5) shows a plane which bisects CN
the NC–Ag–CN axis and is perpendicular to it. Name the point group of the
CF3
‡ These problems were supplied by Charles Trapp and Carmen Giunta. 5

Atkins09819.indb 297 9/11/2013 11:42:23 AM

 298 7 Molecular symmetry

31.3‡ B.A. Bovenzi and G.A. Pearse, Jr. (J. Chem. Soc. Dalton Trans., 2763 HO OH
(1997)) synthesized coordination compounds of the tridentate ligand N N
pyridine-2,6-diamidoxime (C7H9N5O2, 6). Reaction with NiSO4 produced N
H2N NH2
a complex in which two of the essentially planar ligands are bonded at
right angles to a single Ni atom. Name the point group and the symmetry
operations of the resulting [Ni(C7H9N5O2)2]2+ complex cation.
6

Topic 32 Group theory

Discussion questions
32.1 Explain what is meant by a ‘group’. 32.4 Explain what is meant by the reduction of a representation to a direct
sum of representations.
32.2 Explain what is meant by (a) a representative and (b) a representation in
the context of group theory. 32.5 Discuss the significance of the letters and subscripts used to denote the
symmetry species of a representation.
32.3 Explain the construction and content of a character table.

Exercises
32.1(a) Use as a basis the valence pz orbitals on each atom in BF3 to find the 32.3(a) Show that all three C2 operations in the group D3h belong to the same
representative of the operation σh. Take z as perpendicular to the molecular class.
plane. 32.3(b) Show that all three σv operations in the group D3h belong to the same
32.1(b) Use as a basis the valence pz orbitals on each atom in BF3 to find the class.
representative of the operation C3. Take z as perpendicular to the molecular
32.4(a) What is the maximum degeneracy of a particle confined to the interior
plane.
of an octahedral hole in a crystal?
32.2(a) Use the matrix representatives of the operations σh and C3 in a basis 32.4(b) What is the maximum degeneracy of a particle confined to the interior
of valence pz orbitals on each atom in BF3 to find the operation and its of an icosahedral nanoparticle?
representative resulting from σhC3. Take z as perpendicular to the molecular
32.5(a) What is the maximum possible degree of degeneracy of the orbitals in
plane.
benzene?
32.2(b) Use the matrix representatives of the operations σh and C3 in a basis
32.5(b) What is the maximum possible degree of degeneracy of the orbitals in
of valence pz orbitals on each atom in BF3 to find the operation and its
1,4-dichlorobenzene?
representative resulting from C3σh. Take z as perpendicular to the molecular
plane.

Problems
32.1 The group C2 h consists of the elements E, C2, σh, i. Construct the group
multiplication table and find an example of a molecule that belongs to the
group.
32.2 The group D2h has a C2 axis perpendicular to the principal axis and a
horizontal mirror plane. Show that the group must therefore have a centre of
inversion.
32.3 Consider the H2O molecule, which belongs to the group C2v. Take as a
basis the two H1s orbitals and the four valence orbitals of the O atom and set
up the 6 × 6 matrices that represent the group in this basis. Confirm by explicit
matrix multiplication that the group multiplications (a) C2σ v = σ v′ and
(b) σ v σ v′ = C2 . Confirm, by calculating the traces of the matrices, (a) that
symmetry elements in the same class have the same character,
(b) that the representation is reducible, and (c) that the basis spans
3A1 + B1 + 2B2.
32.4 Confirm that the z-component of orbital angular momentum is a basis
for an irreducible representation of A2 symmetry in C3v.
32.5 Find the representatives of the operations of the group Td in a basis of
four H1s orbitals, one at each apex of a regular tetrahedron (as in CH4).
32.6 Confirm that the representatives constructed in Problem 32.5 reproduce
the group multiplications C3+C3− = E , S4C3 = S4′ , and S4C3 = σd.
32.7 The (one-dimensional) matrices D(C3) = 1 and D(C2) = 1, and D(C3) = 1
and D(C2) = –1 both represent the group multiplication C3C2 = C6 in the group
C6v with D(C6) = +1 and –1, respectively. Use the character table to confirm
these remarks. What are the representatives of σv and σd in each case?
32.8 Construct the multiplication table of the Pauli spin matrices, σ, and the
2 × 2 unit matrix:
⎛ 0 1⎞ ⎛ 0 −i ⎞ ⎛1 0 ⎞ ⎛ 1 0⎞
σx =⎜ ⎟ σ y = ⎜ i 0 ⎟ σ z = ⎜ 0 −1⎟ σ 0 = ⎜ 0 1⎟
⎝ 1 0⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠
Do the four matrices form a group under multiplication?
32.9 The algebraic forms of the f orbitals are a radial function multiplied by
one of the factors (a) z(5z2 – 3r2), (b) y(5y2 – 3r2), (c) x(5x2 – 3r2), (d) z(x2 –
y2), (e) y(x2 – z2), (f) x(z2 – y2), (g) xyz. Identify the irreducible representations
spanned by these orbitals in (a) C2v, (b) C3v, (c) Td, (d) Oh. Consider a

www.ebook3000.com

Atkins09819.indb 298 9/11/2013 11:42:28 AM

 Exercises and problems  299

lanthanoid ion at the centre of (a) a tetrahedral complex, (b) an octahedral

complex. What sets of orbitals do the seven f orbitals split into?
32.10‡ A computational study by C.J. Marsden (Chem. Phys. Lett. 245, 475
(1995)) of AMx compounds, where A is in Group 14 of the periodic table
and M is an alkali metal, shows several deviations from the most symmetric
structures for each formula. For example, most of the AM4 structures were not
tetrahedral but had two distinct values for MAM bond angles. They could be
derived from a tetrahedron by a distortion shown in (7). (a) What is the point
group of the distorted tetrahedron? (b) What is the symmetry species of the
7 8
distortion considered as a vibration in the new, less symmetric group? Some
AM6 structures are not octahedral, but could be derived from an octahedron 32.11‡ The H3+ molecular ion, which plays an important role in chemical
by translating a C–M–C axis as in (8). (c) What is the point group of the reactions occurring in interstellar clouds, is known to be equilateral
distorted octahedron? (d) What is the symmetry species of the distortion triangular. (a) Identify the symmetry elements and determine the point group
considered as a vibration in the new, less symmetric group? of this molecule. (b) Take as a basis for a representation of this molecule the
three H1s orbitals and set up the matrices for this basis. (c) Obtain the group
multiplication table by explicit multiplication of the matrices. (d) Determine
if the representation is reducible and, if so, give the irreducible representations
obtained.

Topic 33 Applications of symmetry

Discussion question
33.1 Identify and list four applications of character tables.

Exercises
33.1(a) Use symmetry properties to determine whether or not the integral
∫pxzpzdτ is necessarily zero in a molecule with symmetry C2v.
33.1(b) Use symmetry properties to determine whether or not the integral
∫pxzpzdτ is necessarily zero in a molecule with symmetry D3h.
33.2(a) Is the transition A1 → A2 forbidden for electric dipole transitions in a
C3v molecule?
33.2(b) Is the transition A1g → E2u forbidden for electric dipole transitions in a
D6h molecule?
33.3(a) Show that the function xy has symmetry species B2 in the group C4v.
33.3(b) Show that the function xyz has symmetry species A1 in the group D2.
character table in the Resource section). What irreducible representations does
it span?
33.6(b) A set of basis functions is found to span a reducible representation
of the group D2 with characters 6,–2,0,0 (in the order of operations in the
character table in the Resource section). What irreducible representations does
it span?
33.7(a) What states of (a) benzene, (b) naphthalene may be reached by electric
dipole transitions from their (totally symmetrical) ground states?
33.7(b) What states of (a) anthracene, (b) coronene (9) may be reached by
electric dipole transitions from their (totally symmetrical) ground states?

33.4(a) Consider the C2v molecule NO2. The combination px(A) − px(B) of the
two O atoms (with x perpendicular to the plane) spans A2. Is there any orbital
of the central N atom that can have a nonzero overlap with that combination
of O orbitals? What would be the case in SO2, where 3d orbitals might be
available?
33.4(b) Consider the D3h ion NO3− . Is there any orbital of the central N atom
that can have a nonzero overlap with the combination 2pz(A) – pz(B) – pz(C)
of the three O atoms (with z perpendicular to the plane)? What would be the 9 Coronene
case in SO3, where 3d orbitals might be available?
33.5(a) The ground state of NO2 is A1 in the group C2v. To what excited states 33.8(a) Write f1 = sin θ and f2 = cos θ, and show by symmetry arguments using
may it be excited by electric dipole transitions, and what polarization of light the group Cs that the integral of their product over a symmetrical range
is it necessary to use? around θ = 0 is zero.
33.5(b) The ClO2 molecule (which belongs to the group C2v) was trapped in a 33.8(b) Write f1 = x and f2 = 3x2 – 1, and show by symmetry arguments using
solid. Its ground state is known to be B1. Light polarized parallel to the y-axis the group Cs that the integral of their product over a symmetrical range
(parallel to the OO separation) excited the molecule to an upper state. What is around x = 0 is zero.
the symmetry species of that state?
33.6(a) A set of basis functions is found to span a reducible representation
of the group C4v with characters 4,1,1,3,1 (in the order of operations in the

Atkins09819.indb 299 9/11/2013 11:42:34 AM

 300 7 Molecular symmetry

Problems
33.1 What irreducible representations do the four H1s orbitals of CH4 span?
Are there s and p orbitals of the central C atom that may form molecular
orbitals with them? Could d orbitals, even if they were present on the C atom,
play a role in orbital formation in CH4?
33.2 Suppose that a methane molecule became distorted to (a) C3v symmetry
by the lengthening of one bond, (b) C2v symmetry by a kind of scissors action
in which one bond angle opened and another closed slightly. Would more d
orbitals become available for bonding?
33.3 Does the product 3x2 – 1 necessarily vanish when integrated over (a) a
cube, (b) a tetrahedron, (c) a hexagonal prism, each centred on the origin?
33.4‡ In a spectroscopic study of C60, Negri, et al. (J. Phys. Chem. 100, 10849
(1996)) assigned peaks in the fluorescence spectrum. The molecule has
icosahedral symmetry (Ih). The ground electronic state is A1g, and the lowest-
lying excited states are T1 g and Gg. (a) Are photon-induced transitions allowed
from the ground state to either of these excited states? Explain your answer.
(b) What if the molecule is distorted slightly so as to remove its centre of
inversion?
33.5 In the square planar XeF4 molecule, consider the symmetry-adapted
linear combination p1= pA − pB + pC − pD where pA, pB, pC, and pD are 2pz
atomic orbitals on the fluorine atoms (clockwise labelling of the F atoms).
Using the reduced point group D4 rather than the full symmetry point group
of the molecule, determine which of the various s, p, and d atomic orbitals on
the central Xe atom can form molecular orbitals with p1.
33.6 The chlorophylls that participate in photosynthesis and the haem groups
of cytochromes are derived from the porphine dianion group (10), which
belongs to the D4h point group. The ground electronic state is A1g and the
lowest-lying excited state is Eu. Is a photon-induced transition allowed from
the ground state to the excited state? Explain your answer.
N
N– N–
33.7 The NO2 molecule belongs to the group C2v, with the C2 axis bisecting
the ONO angle. Taking as a basis the N2s, N2p, and O2p orbitals, identify the
irreducible representations they span, and construct the symmetry-adapted
linear combinations.
33.8 Construct the symmetry-adapted linear combinations of C2pz orbitals
for benzene, and use them to calculate the Hückel secular determinant. This
procedure leads to equations that are much easier to solve than using the
original orbitals, and show that the Hückel orbitals are those specified in
Topic 26.
33.9 The phenanthrene molecule (11) belongs to the group C2v with
the C2 axis in the plane of the molecule. (a) Classify the irreducible
representations spanned by the carbon 2pz orbitals and find their symmetry-
adapted linear combinations. (b) Use your results from part (a) to calculate
the Hückel secular determinant. (c) What states of phenanthrene may be
reached by electric dipole transitions from its (totally symmetrical)
ground state?
11 Phenanthrene
33.10 Some linear polyenes, of which β-carotene is an example, are important
biological cofactors that participate in processes as diverse as the absorption
of solar energy in photosynthesis and protection against harmful biological
oxidations. Use as a model of β-carotene a linear polyene containing 22
conjugated C atoms. (a) To what point group does this model of β-carotene
belong? (b) Classify the irreducible representations spanned by the carbon 2pz
orbitals and find their symmetry-adapted linear combinations. (c) Use your
results from part (b) to calculate the Hückel secular determinant. (d) What
states of this model of β-carotene may be reached by electric dipole transitions
from its (totally symmetrical) ground state?

10

www.ebook3000.com

Atkins09819.indb 300 9/11/2013 11:42:36 AM

 FOCUS 8 ON Interactions

Topic 34 Topic 35 Topic 36 Topic 37 Topic 38

Electric Interactions Bonding
Real Crystal
properties between in
gases structure
of molecules molecules solids

Topic 39
Focus 6 Focus 12 Focus 9 Electrical,
optical, and
The First Law magnetic
Molecular Molecular
of thermo- properties
structure spectroscopy
dynamics of solids

From a knowledge of electrostatics (Foundations, Topic 2) and Molecular structure it is possible to con-
struct models for the interactions between atoms or molecules. The result is a better understanding
of the factors that govern the properties of gases at high pressure and the structures and properties
of liquids and solids.
We begin with an account of the electric properties of molecules, such as ‘electric dipole moments’
and ‘polarizabilities’ (Topic 34). All these properties reflect the degree to which the nuclei of atoms
exert control over the electrons in a molecule. The description of the basic theory of interactions
then focuses on ‘van der Waals interactions’ between closed-shell molecules and ‘hydrogen bond-
ing’ (Topic 35). All liquids and solids are bound together by one or more of these cohesive interac-
tions. Deviations from perfect gas behaviour and the thermodynamic properties of ‘real gases’ are
also explained in terms of these interactions (Topic 36).
The solid state includes most of the materials that make modern technology possible. To under-
stand solids, it is necessary to understand the regular arrangement of atoms in crystals and the sym-
metry of their arrangement. The basic principles of ‘X-ray diffraction’ are central to the determination
of structures and we explain how the diffraction pattern obtained in this technique is interpreted in
terms of the distribution of electron density in a ‘unit cell’ (Topic 37). X-ray diffraction studies lead to
important information about the structures of metallic, ionic, and molecular solids (Topic 38). The
energetics of ionic solids can be understood using concepts introduced in The First Law of thermody-
namics. Equipped with a knowledge of structural features, and aided by the principles of Molecular
spectroscopy, we show how the electrical, optical, and magnetic properties of solids stem from the
arrangement and properties of the constituent atoms (Topic 39).

What is the impact of this material?

Interactions between atoms or molecules play important roles in biochemistry, biomedicine, and
technology. Biological polymers can be studied by X-ray diffraction (Impact 8.1), which reveals the
importance of molecular interactions in establishing the three-dimensional structures and biochem-
ical functions of proteins and nucleic acids (Impact 8.2), and the ways in which drugs bind to receptor

Atkins09819.indb 301 9/11/2013 11:42:37 AM

 sites in biopolymers, leading to the inhibition of the progress of disease (Impact 8.3). The manipu-
lation of molecular interactions could have significant technological consequences. One example
is the design of assemblies that can store and deliver hydrogen gas efficiently, thereby making it a
viable fuel for commercial development of a host of devices (Impact 8.4). Another is the synthesis of
‘nanowires’, nanometre-sized atomic assemblies that conduct electricity, which is a major step in the
fabrication of a new generation of electronic devices (Impact 8.5).

To read more about the impact of this material, scan the QR code or go to
http://bcs.whfreeman.com/webpub/chemistry/qmc2e/impact/qchem_
impact8.html.

www.ebook3000.com

Atkins09819.indb 302 9/11/2013 11:42:37 AM

 TOPIC 34
Electric properties of molecules
Contents
34.1 Electric dipole moments
Brief illustration 34.1: Symmetry and the polarity
of molecules
Brief illustration 34.2: Molecular dipole moments
Example 34.1: Calculating a molecular dipole
moment
34.2 Polarizabilities
Brief illustration 34.3: The induced dipole moment 306
Checklist of concepts
Checklist of equations
➤ Why do you need to know this material?
Because the molecular interactions responsible for the
formation of condensed phases and large molecular
assemblies (and which are treated in Topic 35) arise from
the electric properties of molecules, you need to know
how the electronic structures of molecules lead to these
properties.
➤ What is the key idea?
The nuclei of atoms exert control over the electrons in
a molecule and can cause electrons to accumulate in
particular regions, or permit them to respond more or less
strongly to the effects of external fields.
➤ What do you need to know already?
You need to be familiar with the Coulomb law (Founda-
tions, Topic 2), molecular geometry (from introductory
chemistry), and molecular orbital theory, especially the
relevance of the energy gap between a HOMO and LUMO
(Topic 26).
Atkins09819.indb 303
The electric properties of molecules are responsible for many
of the properties of bulk matter. The small imbalances of
charge distributions in molecules allow them to interact with
303
one another and with externally applied fields. One result
of this interaction is the weak cohesion of molecules to form
304
the bulk phases of matter. Molecular interactions are also
304
responsible for the shapes adopted by biological and synthetic
305
macromolecules.
306
307
34.1 Electric dipole moments
308
An electric dipole consists of two electric charges +Q and −Q
with a separation R. A point electric dipole is an electric dipole
in which R is very small compared with its distance from the
observer. The electric dipole moment is a vector μ (1) that
points from the negative charge to the positive charge and has a
magnitude given by
μ = QR Definition Magnitude of the electric dipole moment (34.1)
–Q μ +Q
R
1 Electric dipole
Although the SI unit of dipole moment is coulomb metre (C m),
it is still commonly reported in the non-SI unit debye, D, named
after Peter Debye, a pioneer in the study of dipole moments of
molecules:
1D = 3.335 64 × 10− 30 C m (34.2)
The magnitude of the dipole moment of a pair of charges +e
and −e separated by 100 pm is 1.6 × 10−29 C m, corresponding
to 4.8 D. The magnitudes of dipole moments of small molecules
are typically about 1 D.1
A polar molecule is a molecule with a permanent electric
dipole moment. The permanent dipole moment stems from
1 The conversion factor in eqn 34.2 stems from the original definition of
the debye in terms of c.g.s. units: 1 D is the dipole moment of two equal and
opposite charges of magnitude 1 e.s.u. separated by 1 Å.
9/11/2013 11:42:40 AM
 304 8 Interactions

the partial charges on the atoms in the molecule that arise from resultant of two chlorobenzene dipole moments arranged at 60°
differences in electronegativity or other features of bonding to each other. This technique of ‘vector addition’ can be applied
(Topics 25 and 26). Nonpolar molecules acquire an induced with fair success to other series of related molecules, and the
dipole moment in an electric field on account of the distortion magnitude of the resultant, μres, of two dipole moments, μ1 and
the field causes in their electronic distributions and nuclear μ2, that make an angle θ to each other (4) is approximately (see
positions; however, this induced moment is only temporary, Mathematical background 4)
and disappears as soon as the perturbing field is removed. Polar
molecules also have their existing dipole moments temporarily μres ≈ ( μ12 + μ22 + 2 μ1 μ2 cosθ )1/2 (34.3a)
modified by an applied field.
All heteronuclear diatomic molecules are polar, and typical
μ1 μ
values of μ include 1.08 D for HCl and 0.42 D for HI (Table
34.1). Molecular symmetry is of the greatest importance in
deciding whether a polyatomic molecule is polar or not (see θ
also Topics 31 and 32). Indeed, molecular symmetry is more μ2
important than the question of whether or not the atoms in the 4 Addition of dipole moments
molecule belong to the same element. For this reason, and as
we see in Brief illustration 34.1, homonuclear polyatomic mol- When the two dipole moments have the same magnitude (as in
ecules may be polar if they have low symmetry and the atoms the dichlorobenzenes), this equation simplifies to
are in non-equivalent positions.
1+ cosθ =2 cos2 12 θ

μres ≈ {2 μ12 (1+ cosθ )}1/2 = 2 μ1 cos 12 θ (34.3b)

Brief illustration 34.1 Symmetry and the polarity
of molecules
Brief illustration 34.2 Molecular dipole moments
The angular molecule ozone (2) is homonuclear. However, it
is polar because the central O atom is different from the outer Consider ortho (1,2-) and meta (1,3-) disubstituted benzenes,
two (it is bonded to two atoms, which are each bonded only for which θortho = 60° and θmeta = 120°. It follows from eqn 34.3
to one). Moreover, the dipole moments associated with each that the ratio of the magnitudes of the electric dipole moments
bond make an angle to each other and do not cancel. The is
heteronuclear linear triatomic molecule CO2 (3) is nonpolar
because, although there are partial charges on all three atoms, = =
(
μres , ortho cos 12 θ ortho cos 12 × 60°
=
(3)1/2 /2 )
= (3)1/2 ≈ 1.7
the dipole moment associated with the OC bond points in the μres ,meta cos 2 θ meta cos 1 × 120°
1
2 ( 1/2 )
opposite direction to the dipole moment associated with the
CO bond, and the two cancel. Self-test 34.2 Calculate the resultant magnitude of two dipole
moments of magnitude 1.5 D and 0.80 D that make an angle of
μ
109.5° to each other.
δ+ δ+ δ+ δ+
δ– δ– δ– δ– Answer: 1.4 D

2 Ozone, O3 3 Carbon dioxide, CO2 A more reliable approach to the calculation of dipole
moments is to take into account the locations and magnitudes
Self-test 34.1 Is SO2 polar? of the partial charges on all the atoms. These partial charges are
Answer: Yes included in the output of many molecular structure software
packages. To calculate the x-component, for instance, we need
to know the partial charge on each atom and the atom’s x-coor-
dinate relative to a point in the molecule and form the sum
To a good first approximation, the dipole moment of a polya-
tomic molecule can be resolved into contributions from vari-
ous groups of atoms in the molecule and their relative locations
μx = ∑Q x
J
J J (34.4a)

(Fig. 34.1). Thus, 1,4-dichlorobenzene is nonpolar by symmetry

on account of the cancellation of two equal but opposing C − Cl Here QJ is the partial charge of atom J, xJ is the x-coordinate
moments (exactly as in carbon dioxide). 1,2-Dichlorobenzene, of atom J, and the sum is over all the atoms in the molecule.
however, has a dipole moment which is approximately the Analogous expressions are used for the y- and z-components.

www.ebook3000.com

Atkins09819.indb 304 9/11/2013 11:43:06 AM

 34 Electric properties of molecules  305

C2v D2h corresponding to μ x = +0.42 D. The expression for μy is

μ y = (−0.36e) × (0 pm) + (0.45e) × (0 pm) + (0.18e) × (−87 pm)

+ (−0.38e) × (107 pm)
= −56e pm
(a) μobs = 1.57 D (b) μobs = 0, μcalc = 0 = −19.0 ×10−30 C m

C2v C2v
It follows that μy= –2.7 D. The amide group is planar, so μz = 0
and

μ = {(0.42 D)2 + (−2.7 D)2 }1/2 = 2.7 D

(c) μobs = 2.25 D, μcalc = 2.7 D (d) μobs = 1.48 D, μcalc = 1.6 D
Figure 34.1 The resultant dipole moments (red) of
the dichlorobenzene isomers (b to d) can be obtained
approximately by vectorial addition of two chlorobenzene
dipole moments (with magnitude 1.57 D). (The point groups
of the molecules are also indicated.)
We can find the orientation of the dipole moment by arrang-
ing an arrow of length 2.7 units of length to have x, y, and z
components of 0.42, –2.7, and 0 units, respectively; the orien-
tation is superimposed on 5.
Self-test 34.3 Calculate the electric dipole moment of formal-
dehyde by using the information in 6.
–0.38 (0,118,0)

For an electrically neutral molecule, the origin of the coor- +0.45 (0,0,0)
dinates is arbitrary, so it is best chosen to simplify the meas- +0.18 +0.18
urements. In common with all vectors, the magnitude of μ is (–94,–61,0) (94,–61,0)

related to the three components μx, μy, and μz by 6

Answer: 2.3 D
μ = ( μ x2 + μ 2y + μz2 )1/2
(34.4b)

Example 34.1 Calculating a molecular dipole moment Molecules may have higher multipoles, or arrays of point
Estimate the electric dipole moment of the amide group charges (Fig. 34.2). Specifically, an n-pole is an array of point
shown in (5) by using the partial charges (as multiples of charges with an n-pole moment but no lower moment. Thus, a
e) and the locations of the atoms shown, with distances in monopole (n = 1) is a point charge, and the monopole moment
picometres. is what we normally call the overall charge. A dipole (n = 2),
(182,–87,0)
as we have seen, is an array of charges that has no monopole
+0.18 H moment (no net charge). A quadrupole (n = 3) consists of an
+0.45
(132,0,0) N C (0,0,0) array of point charges that has neither net charge nor dipole
–0.36 moment (as for CO2 molecules, 3). An octupole (n = 4) consists
O (–62,107,0)
–0.38
5 Amide (peptide) link
Monopole
Method We use eqn 34.4a to calculate each of the components
of the dipole moment and then eqn 34.4b to assemble the
Dipole Octupole
three components into the magnitude of the dipole moment.
Note that the partial charges are multiples of the fundamental Quadrupole
charge, e = 1.609 × 10−19 C.
Answer The expression for μ x is
Quadrupole Octupole
μ x = (−0.36e) × (132 pm) + (0.45e) × (0 pm) + (0.18e) × (182 pm)
+ (−0.38e) × (−62.0 pm)
= 8.8e pm Figure 34.2 Typical charge arrays corresponding to electric
multipoles. The field arising from an arbitrary finite charge
= 8.8 × (1.602 ×10 −19
C) × (10 −12
m) = 1.4 ×10 −30
Cm
distribution can be expressed as the superposition of the fields
arising from a superposition of multipoles.

Atkins09819.indb 305 9/11/2013 11:43:46 AM

 306 8 Interactions

of an array of point charges that sum to zero and which has nei- metre (C2 J−1 m−1), it follows that α ′ has the dimensions of
ther a dipole moment nor a quadrupole moment (as for CH4 volume (hence its name). Polarizability volumes are similar in
molecules, 7). magnitude to actual molecular volumes (of the order of 10−30
m3, 10−3 nm3, 1 Å3).
δ+

δ–
δ + δ– Brief illustration 34.3 The induced dipole moment
δ–
δ+
δ– The polarizability volume of H2O is 1.48 × 10 −30 m3. It follows
from eqns 34.5a and 34.6 that μ* = 4πε 0α ′E and the dipole
δ+ moment of the molecule (in addition to the permanent dipole
7 Methane, CH4 moment) induced by an applied electric field of strength
1.0 × 105 V m−1 is

μ* = 4π × (8.854 × 10−12 J −1 C 2 m −1 ) × (1.48 × 10−30 m3 )

34.2 Polarizabilities
The failure of nuclear charges to control the surrounding elec-
trons totally means that those electrons can respond to external
fields. Therefore, an applied electric field can distort a molecule
as well as align its permanent electric dipole moment. When
the applied field is weak, the magnitude of the induced dipole
moment, μ*, is proportional to the field strength, E, and we write
× (1.0 × 105 JC −1 m −1 )
= 31.6 × 10−35 C m = 4.9 × 10−6 D = 4.9 μD
where we have used 1 V = 1 J C −1.
Self-test 34.4 What strength of electric field is required to
induce an electric dipole moment of magnitude 1.0 μD in a
molecule of polarizability volume 2.6 × 10−30 m3 (like CO2)?
Answer: 11 kV m−1

μ * = αE Definition Polarizability (34.5a)

The constant of proportionality α is the polarizability of
the molecule. The greater the polarizability, the larger is the
induced dipole moment for a given applied field. In a formal
treatment, we should use vector quantities and allow for the
possibility that the induced dipole moment might not lie paral-
lel to the applied field, but for simplicity we discuss polarizabili-
ties in terms of (scalar) magnitudes.
When the applied field is very strong (as in tightly focused
laser beams), the induced dipole moment is not strictly linear
in the strength of the field, and we write
μ * = αE + 12 βE 2 +… Definition Hyperpolarizability
The experimental polarizability volumes of some mol-
ecules are given in Table 34.1. As shown in the follow-
ing Justification, polarizability volumes correlate with the
HOMO–LUMO separations in atoms and molecules (Topic
26). The electron distribution can be distorted readily if the
LUMO lies close to the HOMO in energy, so the polarizabil-
ity is then large. If the LUMO lies high above the HOMO,
an applied field cannot perturb the electron distribution
significantly, and the polarizability is low. Molecules with
small HOMO–LUMO gaps are typically large, with numer-
ous electrons.
(34.5b)

The coefficient β is called the hyperpolarizability of the

Table 34.1* Magnitudes of dipole moments (μ) and
molecule.
polarizability volumes (α')
Polarizability has the units (coulomb metre)2 per joule
(C2 m2 J−1). That collection of units is awkward, so α is often μ/D α′/(10−30 m3)
expressed as a polarizability volume, α ′, by using the relation
CCl4 0 10.3
H2 0 0.819
α
α′= Definition Polarizability volume (34.6) H2O 1.85 1.48
4π ε 0
HCl 1.08 2.63
HI 0.42 5.45
where ε0 is the vacuum permittivity (Foundations, Topic 2). * More values are given in the Resource section.
Because the units of 4πε0 are coulomb-squared per joule per

www.ebook3000.com

Atkins09819.indb 306 9/11/2013 11:44:00 AM

 34 Electric properties of molecules  307

Justification 34.1 density from the distribution ψ 0 to the distribution ψn and ψi,
Polarizabilities and molecular and E(0)
i are the wavefunctions and energies, respectively, in
structures the absence of the electric field. By comparing the two expres-
The energy E of a molecule in an electric field of magnitude E sions for the energy, we conclude that the polarizability of the
is related to the molecular dipole moment by molecule in the z-direction is

E = −μE 2
μ z , 0n
It follows that when the electric field is increased by dE, the α =2 ∑E
n≠0 n − E0
(0 ) (0 ) (34.7)
energy changes by −μdE and, if the molecule is polarizable,
we interpret μ as the induced dipole moment μ* (eqn 34.5). The content of eqn 34.7 can be appreciated by approximat-
Therefore, the change in energy when the field is increased ing the excitation energies by a mean value ΔE (an indication
from 0 to E is of the HOMO–LUMO separation) and supposing that the
E E most important transition dipole moment is approximately
1
ΔE = − ∫0
μ *dE = − ∫ 0
αEdE = − αE 2
2 equal to the charge of an electron multiplied by the molecular
radius, R, of the molecule. then
The contribution to the hamiltonian when a dipole moment is
exposed to an electric field E in the z-direction is 2e 2 R2
α≈
ΔE
 (1) = −μ E
H z
This expression shows that α increases with the size of the
Comparison of these two expressions suggests that we should molecule and with the ease with which it can be excited (the
use second-order perturbation theory to calculate the energy smaller the value of ΔE).
of the system in the presence of the field, because then we shall If the excitation energy is approximated by the energy
obtain an expression proportional to E2 . According to eqn needed to remove an electron to infinity from a distance R
15.6 of Topic 15, the second-order contribution to the ground- from a single positive charge, we can write ΔE ≈ e2/(4πε 0R).
state energy is When this expression is substituted into the equation above,
both sides are divided by 4πε 0, and the factor of 2 is ignored
2
in this approximation, we obtain α ′ ≈ R3, which is of the same
∫ψ H*
ψ 0 dτ
(1)
∫ψ n* μ zψ 0dτ
2

∑ ∑
n
E (2)
= =E 2 order of magnitude as the molecular volume.
n≠0
E0(0) − En(0) n≠0
E0(0) − En(0)
2
μ z , 0n
=E2 ∑En≠0 0 − En
(0) (0)
For most molecules, the polarizability is anisotropic, by

∫
where μz ,0n = ψ n* μ zψ 0dτ is the transition electric dipole
moment in the z-direction. Transition dipole moments are
introduced in Topic 16 and discussed further in Topic 45:
for our purposes here they can be interpreted as the electric
dipole moment associated with the migration of electron
which is meant that its value depends on the orientation of
the molecule relative to the field. The polarizability volume of
benzene when the field is applied perpendicular to the ring is
0.0067 nm3 and it is 0.0123 nm3 when the field is applied in the
plane of the ring.

Checklist of concepts
☐ 1. An electric dipole consists of two electric charges +Q
and –Q separated by a distance R.
☐ 2. The electric dipole moment μ is a vector that points from
the negative charge to the positive charge of a dipole.
☐ 3. A polar molecule is a molecule with a permanent elec-
tric dipole moment.
☐ 4. Molecules may have higher electric multipoles: an
n-pole is an array of point charges with an n-pole
moment but no lower moment.
☐ 5. The polarizability is a measure of the ability of an elec-
tric field to induce a dipole moment in a molecule.
☐ 6. Polarizabilities (and polarizability volumes) correlate
with the HOMO–LUMO separations in atoms and
molecules.
☐ 7. For most molecules, the polarizability is anisotropic.

Atkins09819.indb 307 9/11/2013 11:44:14 AM

 308 8 Interactions

Checklist of equations
Property Equation Comment Equation number

Magnitude of the electric dipole moment μ = QR Definition 34.1

Magnitude of the resultant of two dipole μres ≈ ( μ12 + μ22 + 2 μ1 μ2 cos θ )1/2 34.3a
moments
Magnitude of the induced dipole μ* = αE Linear approximation; α is the 34.5a
moment polarizability

μ * = αE + 12 βE 2 Quadratic approximation; β is the 34.5b

hyperpolarizability

Polarizability volume α ′ = α/4πε0 Definition 34.6

www.ebook3000.com

Atkins09819.indb 308 9/11/2013 11:44:15 AM

 TOPIC 35

Interactions between molecules

Contents ➤ What is the key idea?

35.1 Interactions between partial charges 309
Attractive interactions result in cohesion but repulsive
Brief illustration 35.1: The interaction energy
interactions prevent the complete collapse of matter to
of two partial charges 310
nuclear densities.
35.2 The interactions of dipoles 310
(a) Charge–dipole interactions 310 ➤ What do you need to know already?
Brief illustration 35.2: The energy of interaction
You need to be familiar with electrostatics, specifically
of a point charge and a point dipole 311
the Coulomb interaction (Foundations, Topic 2), and with
(b) Dipole–dipole interactions 311
the relationships between the structure and electric
Brief illustration 35.3: The dipolar interaction 312
properties of a molecule, specifically its dipole moment
Brief illustration 35.4: The Keesom interaction 313
and polarizability (Topic 34).
(c) Dipole–induced dipole interactions 314
Brief illustration 35.5: The dipole–induced dipole
interaction 314
(d) Induced dipole–induced dipole interactions 315 We begin by examining the interactions between the partial
Brief illustration 35.6: The London interaction 315
charges of polar molecules. Then we discuss van der Waals
interactions: attractive interactions between closed-shell mol-
35.3 Hydrogen bonding 315
ecules that depend on the separation of the molecules as the
Brief illustration 35.7: The hydrogen bond 316
inverse sixth power (V ∝ 1/r6), although this precise criterion
35.4 The total interaction 317 is often relaxed to include all nonbonding interactions. Finally,
Example 35.1: Calculating an intermolecular we see that repulsive interactions arise from Coulomb forces
force from the Lennard-Jones potential energy 318
and, indirectly, from the Pauli principle (Topic 19) and the
Checklist of concepts 319
exclusion of electrons from regions of space where the orbitals
Checklist of equations 319
of neighbouring species overlap.

Interactions between partial

35.1
charges
➤ Why do you need to know this material?
You need to understand the many types of molecular In general, atoms in molecules have partial charges arising from
interactions responsible for the formation of condensed the spatial variation in electron density in the ground state. If
phases and large molecular assemblies. The molecular these charges were separated by a vacuum, they would attract
interactions described here are of prime importance for or repel each other in accord with Coulomb’s law (Foundations,
solving one of the great problems of molecular biology: Topic 2), and we would write
how complex molecules, like proteins and nucleic acids,
fold into their three-dimensional structures. Q1Q2
V= Vacuum Coulomb potential energy (35.1a)
4πε 0 r

Atkins09819.indb 309 9/11/2013 11:44:18 AM

 310 8 Interactions

where Q1 and Q2 are the partial charges, r is their separation, 0

Potential energy, V/(Q1Q2/4πε0r0)

and ε0 is the vacuum permittivity. However, we should take into 3
account the possibility that other parts of the molecule, or other –2 1

molecules, lie between the charges and decrease the strength of

the interaction. We therefore write –4

Q1Q2 –6
V= Any medium Coulomb potential energy (35.1b)
4πεr
–8

where ε is the permittivity of the medium lying between the

–10
charges. The permittivity is usually expressed as a multiple of 0 1 2 3
Distance, r/r0
the vacuum permittivity by writing ε = εrε0, where εr is the rela-
tive permittivity (formerly known as the dielectric constant). Figure 35.1 The Coulomb potential for two (opposite) charges
The effect of the medium can be very large: for water εr = 78, and its dependence on their separation. The two curves
so the potential energy of two charges separated by bulk water correspond to different relative permittivities (εr = 1 for a
is reduced by nearly two orders of magnitude compared to the vacuum, εr = 3 for a fluid; r0 is a scaling factor).
value it would have if the charges were separated by a vacuum
(Fig. 35.1).
(a) Charge–dipole interactions
A point dipole is a dipole in which the separation l between the
Brief illustration 35.1 The interaction energy of two charges is much smaller than the distance r at which the dipole
partial charges is being observed (l << r). We show in the following Justification
The energy of interaction between a partial charge of –0.36 that the potential energy of interaction between a point dipole
(that is, Q1 = −0.36e) on the N atom of an amide group and the with a dipole moment of magnitude μ1 = Q1l and the point
partial charge of +0.45 (Q2 = +0.45e) on the carbonyl C atom charge Q2 in the arrangement shown in 1 is
at a distance of 3.0 nm, on the assumption that the medium
between them is a vacuum, is μ1Q2 Energy of interaction between a
V =− point dipole and a point charge (35.2)
(0.36e) × (0.45e) 4π ε 0 r 2
V =−
4 π ε 0 × (3.0 nm)
0.36 × 0.45 × (1.602 × 10−19 C)2 r
=− μ1
4 π × (8.854 × 10−2 J −1C −2m −1 ) × (3.0 × 10−9 m)
= −1.2 × 10−20 J +Q1 –Q1 Q2
l

This energy (after multiplication by Avogadro’s constant) cor- 1

responds to −7.5 kJ mol−1. However, if the medium has a ‘typi-
cal’ relative permittivity of 3.5, then the interaction energy is
reduced by that factor to −2.1 kJ mol−1.
Self-test 35.1 Repeat the calculation for bulk water as the
medium.
Answer: −0.96 kJ mol−1
With μ in coulomb metres, Q2 in coulombs, and r in metres,
V is obtained in joules. The potential energy rises towards zero
(the value at infinite separation of the charge and the dipole)
more rapidly (as 1/r2) than that between two point charges
(which varies as 1/r) because, from the viewpoint of the point
charge, the partial charges of the dipole seem to merge and can-
cel as the distance r increases (Fig. 35.2).

35.2 The interactions of dipoles

Justification 35.1
The interaction between a point
Most of the discussion in this and the following sections is based
charge and a point dipole
on the Coulombic potential energy of interaction between two
charges (eqn 35.1a). This expression can be adapted to find the The sum of the potential energies of repulsion between like
potential energy of a point charge and a dipole and extend it to charges and attraction between opposite charges in the orien-
the interaction between two dipoles. tation shown in 1 is

www.ebook3000.com

Atkins09819.indb 310 9/11/2013 11:44:25 AM

 35 Interactions between molecules  311

⎛ ⎞ Self-test 35.2 Consider the arrangement in 1 and calculate the

1 ⎜ Q1Q2 Q1Q2 ⎟ Q1Q2 ⎛ 1 1 ⎞
V= − + = − + molar energy required to reverse the direction of the water
4π ε0 ⎜ 1 1 ⎟ 4 π ε 0r ⎜⎝ 1 − x 1 + x ⎟⎠
⎜⎝ r − l r + l ⎟⎠ molecule when it is at 300 pm from the Li+ ion.
2 2
Answer: 119 kJ mol−1
where x = l/2r. Because l << r for a point dipole, this expres-
sion can be simplified by expanding the terms in x by using
(Mathematical background 1)
1 1
(b) Dipole–dipole interactions
= 1 − x + x 2 − = 1 + x + x 2 +
1+ x 1− x We show in the following Justification that the preceding dis-
cussion can be extended to the interaction of two dipoles
and retaining only the leading surviving term:
arranged as in 2. The result is
2 xQ1Q2 2Q Q l μ1 μ2
V =− = − 1 22 V =−
Arrangement Energy of interaction
4 πε 0r 4 πε 0r as in 2 (35.3)
2π ε 0 r 3 between two dipoles

With μ1 = Q1l, this expression becomes eqn 35.2. The equation

should be multiplied by cos θ when the point charge lies at an r
μ1 μ2
angle θ to the axis of the dipole.
+Q1 l –Q1 +Q2 l –Q2

This interaction energy approaches zero more rapidly (as 1/r3)

than for the previous case: now both interacting entities appear
neutral to each other at large separations.

Justification 35.2 The interaction energy of two dipoles

To calculate the potential energy of interaction of two dipoles
Figure 35.2 There are two contributions to the diminishing separated by r in the arrangement shown in 2 we proceed
field of an electric dipole with distance (here seen from the in exactly the same way as in Justification 35.1, but now the
side). The potentials of the charges decrease (shown here by a total interaction energy is the sum of four pairwise terms, two
fading intensity) and the two charges appear to merge, so their attractions between opposite charges, which contribute nega-
combined effect approaches zero more rapidly than by the tive terms to the potential energy, and two repulsions between
distance effect alone. like charges, which contribute positive terms.
The sum of the four contributions is
Brief illustration 35.2 The energy of interaction of a
1 ⎛ Q1Q2 Q1Q2 Q1Q2 Q1Q2 ⎞
point charge and a point dipole V= − + + −
4 π ε 0 ⎜⎝ r + l r r r − l ⎟⎠
Consider a Li+ and a water molecule (μ = 1.85 D) separated by Q1Q2 ⎛ 1 1 ⎞
=− −2+
1.0 nm, with the point charge on the ion and the dipole of the 4 π ε 0r ⎜⎝ 1 + x 1 − x ⎟⎠
molecule arranged as in 1. The energy of interaction is given
by eqn 35.2 as with x = l/r. As before, provided l  r we can expand the two
QLi+ μH2O terms in x and retain only the first surviving term, which is
 
(1.602 × 10−19 C) × (1.85 × 3.336 × 10−30 C m) equal to 2x 2. This step results in the expression
V =− −2
4 π × (8.854 × 10−12 J −1C −1m −1 ) × (1.0 × 10−9 m) 2 x 2Q1Q2
   V =−
ε0 r 4 πε 0r
= −8.9 × 10−21 J
Therefore, because μ1 = Q1l and μ2 = Q2l, the potential energy of
This energy corresponds to −5.4 kJ mol−1. interaction in the alignment shown in 2 is given by eqn 35.3.

Atkins09819.indb 311 9/11/2013 11:44:36 AM

 312 8 Interactions

Justification 35.2 represents only one possible orientation of
two dipoles. More generally, the potential energy of interac-
tion between two polar molecules is a complicated function of
their relative orientation. When the two dipoles are parallel and
arranged as in 3, the potential energy is simply
the repulsion of the former is cancelled by the attraction of the
latter. Mathematically, this result arises from the fact that, as we
show in the following Justification, the average (or mean value)
of the function 1 – 3 cos2 θ is zero.

Energy of
μ1 μ2 f (θ ) interaction Justification 35.3
V= f (θ ) = 1− 3cos2θ (35.4) The dipolar interaction between two
4π ε 0r 3 between two
freely rotating molecules
fixed parallel
dipoles
Consider the unit sphere shown in Fig. 35.3. The average value
(or mean value) of f(θ) = 1 − 3 cos2 θ is the sum of its values in
μ1 l each of the infinitesimal regions on the surface of the sphere
+Q1 –Q1 (that is, the integral of the function over the surface) divided
by the surface area of the sphere (which is equal to 4π). With
r the area element in spherical polar coordinates as sin θ dθ dφ,
θ θ ranging from 0 to π, and φ ranging from 0 to 2π, the average
μ2
value 〈f(θ)〉 of f(θ) is
+Q2 l –Q2
1 2π π
3
〈 f (θ )〉 =
4π 0 0 ∫ ∫ (1 − 3 cos2θ )sinθ dθ dφ
1 2π π
=
4π 0 ∫ ∫
dφ (1 − 3 cos2θ )sinθ dθ
0
Brief illustration 35.3 1 π
∫
The dipolar interaction = (1 − 3 cos2θ )sinθ dθ
2 0
We can use eqn 35.4 to calculate the molar potential energy of
the dipolar interaction between two amide groups. Supposing
that the groups are separated by 3.0 nm with θ = 180° (so that z
cos θ = −1 and 1 − 3 cos2 θ = −2), we take μ1 = μ2 = 2.7 D, corre-
sponding to 9.1 × 10−30 C m, and find θ
sin θ dθ
dφ
μ 1 μ2
 1−
3 cos2θ

(9.1 × 10−30 C m)2 × (−2)

V=
4 π × (8.854 × 10−12 J −1 C 2 m −1 ) × (3.0 × 10−9 m)3 y
 
ε0 r3 x
(9.1 × 10−30 )2 × (−2) C 2 m2
=
4 π × (8.854 × 10−12 ) × (3.0 × 10−9 )3 J −1 C 2 m −1 m3
= −5.5 × 10−23 J Figure 35.3 A unit sphere showing the area element
sin θ dθ dφ.
where we have used 1 V C = 1 J. This value corresponds to
−33 J mol−1. Note that this energy is considerably less than that
between two partial charges at the same separation (see Brief The integral is calculated as follows:
illustration 35.1). 
− d cos

θ
π π π

Self-test 35.3 Repeat the calculation for an amide group and a ∫ 0

(1 − 3 cos θ )sinθ dθ =
2
∫
0 ∫
sinθ dθ − 3
0
cos θ sinθ dθ
2

water molecule separated by 3.5 nm with θ = 90°, in a medium

Integrals T.1 and T.10

⎛ 1 π⎞
π
with relative permittivity of 3.5. = − cosθ 0 − 3 ⎜ − cos3θ ⎟
Answer: −2.1 J mol−1 ⎝ 3 0⎠


+2   −2
π π
= − cosθ 0 + cos3θ 0 = 0
Equation 35.4 applies to polar molecules in a fixed, parallel
orientation in a solid. In a fluid of freely rotating molecules, where we have used the standard forms listed in the Resource
the interaction between dipoles averages to zero because f(θ) section. It follows that 〈 f(θ)〉 = 0, and, from eqn 35.4, that the
changes sign as the orientation changes, and its average value dipolar interaction between two freely rotating molecules
is zero. Physically, the like partial charges of two freely rotating vanishes.
molecules are as close together as the two opposite charges, and

www.ebook3000.com

Atkins09819.indb 312 9/11/2013 11:44:46 AM

 35 Interactions between molecules  313

The interaction energy of two freely rotating dipoles is zero. It follows that
However, because their mutual potential energy depends on 1 π
1 π
their relative orientation, the molecules do not in fact rotate 〈 f (θ )〉 = ∫ f (θ )dθ − ∫ f (θ ) (V / kT ) dθ +
π 0 π 0
completely freely, even in a gas. In fact, the lower-energy ori- 1 π
1 π
μ1μ2
entations are marginally favoured, so there is a nonzero average =
π ∫0
f (θ )dθ −
π ∫
0 4 π ε 0kTr 3
f (θ )2 dθ +
interaction between polar molecules. We show in the following 〈 f (θ ) 〉0
2
 〈 f (θ )0 〉
    
Justification that the average potential energy of two rotating
1 π μ1μ2 ⎛ π 1 ⎞
molecules that are separated by a distance r is =
π 0 ∫ f (θ )dθ − 3 ⎜
4 π ε 0kTr ⎝ 0 ∫π
f (θ )2 dθ ⎟ +
⎠
μ1μ2
Average potential = 〈 f (θ )〉0 − 〈 f (θ )2 〉0 +
C 2 μ12 μ12 4 π ε 0kTr 3
〈V 〉 = − C= energy of two
(35.5)
r6 3(4 π ε 0 )2 kT rotating polar
molecules
where 〈〉 0 denotes an unweighted spherical average. The
This expression describes the Keesom interaction, and is the spherical average of f(θ) is zero (as in Justification 35.3), so
first of the contributions to the van der Waals interaction (when the first term in the expression for 〈 f(θ)〉 vanishes. However,
that is taken to be a 1/r6 interaction). the average value of f(θ)2 is nonzero because f(θ)2 is positive at
all orientations, so we can write

μ12 μ12 〈 f (θ )2 〉0
Justification 35.4
〈V 〉 = −
The Keesom interaction (4π ε 0 )2 kTr 6
The detailed calculation of the Keesom interaction energy
The average value 〈 f(θ)2〉0 turns out to be 2/3 when the calcula-
is quite complicated, but the form of the final answer can be
tion is carried through in detail. The final result is that quoted
constructed quite simply. First, we note that the average inter-
in eqn 35.5.
action energy of two polar molecules rotating at a fixed sepa-
ration r is given by

μ1μ2 〈 f (θ )〉 The important features of eqn 35.5 are:

〈V 〉 =
4π ε 0r 3
r
The negative sign shows that the average interaction is
where 〈f(θ)〉 now includes a weighting factor in the averaging attractive.
that is equal to the probability that a particular orientation
will be adopted. This probability is given by the Boltzmann r
The dependence of the average interaction energy on the
distribution, p ∝ e −E/kT, with E interpreted as the potential inverse sixth power of the separation identifies it as a

Physical interpretation
energy of interaction of the two dipoles in that orientation. van der Waals interaction.
That is, r
The inverse dependence on the temperature reflects
the way that the greater thermal motion overcomes
μ1μ2 f (θ ) the mutual orientating effects of the dipoles at higher
p ∝ e −V /kT V=
4π ε 0r 3 temperatures.
r
The inverse sixth power arises from the inverse third
When the potential energy of interaction of the two dipoles
power of the interaction potential energy that is
is very small compared with the energy of thermal motion,
weighted by the energy in the Boltzmann term, which is
we can use V << kT, expand the exponential function in p, and
retain only the first two terms: also proportional to the inverse third power of the
separation.
p ∝1− V /kT +

We now write the weighted average of f(θ) as

Brief illustration 35.4 The Keesom interaction
π

∫ f (θ ) pdθ
1 π
1 π
Suppose a water molecule (μ1 = 1.85 D) can rotate 1.0 nm from
〈 f (θ )〉 = 0
π
=
π ∫ f (θ ) pdθ =
π ∫ f (θ )(1−V /kT )dθ + an amide group (μ2 = 2.7 D). The average energy of their inter-
∫
0 0
dθ action at 25 °C (298 K) is
0

Atkins09819.indb 313 9/11/2013 11:44:55 AM

 314 8 Interactions

The reason for the even steeper decrease with distance is the same
μ1 μ2
      as before: the array of charges appears to blend together into neu-
2 × (1.85 × 3.336 × 10−30 C m)2 × (2.7 × 3.336 × 10−30 C m)2 trality more rapidly with distance the higher the number of indi-
〈V 〉 = −
3 × (1.710 × 10−43 J −1 C −2 m −2 K −1 ) × (298 K) × (1.0 × 10−9 m)6 vidual charges that contribute to the multipole. Note that a given
   
( 4 π ε 0 )2 k T r molecule may have a charge distribution that corresponds to a
superposition of several different multipoles, and in such cases
This interaction energy corresponds (after multiplication by
the energy of interaction is the sum of terms given by eqn 35.6.
Avogadro’s constant) to −24 J mol−1, and it is much smaller
than the energies involved in the making and breaking of
chemical bonds. (c) Dipole–induced dipole interactions
A note on good practice Note how the units are included A polar molecule with dipole moment μ1 can induce a dipole in
in the calculation and cancel to give the result in joules. a neighbouring polarizable molecule (Fig. 35.4). The induced
It is far better to include the units at each stage of the dipole interacts with the permanent dipole of the first mole-
calculation and treat them as algebraic quantities that cule, and the two are attracted together. The average interaction
can be multiplied and cancelled than to guess the units energy when the separation of the molecules is r is
at the end of the calculation.
Potential energy of
C μ12α 2′
Self-test 35.4 Calculate the average interaction energy for V =− C= a polar molecule
(35.7)
pairs of molecules in the gas phase with μ = 1 D when the sepa- r6 4π ε 0 and a polarizable
molecule
ration is 0.5 nm at 298 K. Compare this energy with the aver-
age molar kinetic energy of the molecules. where α ′2 is the polarizability volume (Topic 34) of molecule
Answer: 〈V 〉 = −0.07 kJ mol −1 << 3
RT = 3.7 kJ mol −1
2 and μ1 is the magnitude of the permanent dipole moment of
2
molecule 1. Note that the C in this expression is different from
the C in eqn 35.5 and other expressions below: we are using the
same symbol in C/r6 to emphasize the similarity of form of each
Table 35.1 summarizes the various expressions for the expression.
interaction of charges and dipoles. It is quite easy to extend The dipole–induced dipole interaction energy is independ-
the formulas given there to obtain expressions for the energy ent of the temperature because thermal motion has no effect
of interaction of higher multipoles (electric multipoles are on the averaging process. Moreover, like the dipole–dipole
described in Topic 34). The feature to remember is that the interaction, the potential energy depends on 1/r6: this dis-
interaction energy falls off more rapidly the higher the order of tance dependence stems from the 1/r3 dependence of the field
the multipole. For the interaction of an n-pole with an m-pole, (and hence the magnitude of the induced dipole) and the 1/r3
the potential energy varies with distance as dependence of the potential energy of interaction between the
permanent and induced dipoles.
1
V∝ Energy of interaction between multipoles (35.6)
r n+m+1
Brief illustration 35.5 The dipole–induced dipole
interaction
Table 35.1 Interaction potential energies For a molecule with μ = 1.0 D (3.3 × 10 −30 C m, such as HCl)
separated by 0.30 nm from a molecule of polarizability volume
Interaction Distance Typical Comment
type dependence of energy α ′ = 10 × 10 −30 m 3 (such as benzene, Table 34.1), the average
potential energy (kJ mol−1) interaction energy is
Ion–ion 1/r 250 Only between ions (3.3 × 10−30 C m)2 × (10 × 10−30 m3 )
V =−
Hydrogen 20 Occurs in X–HY, 4 π ×(8.854 × 10−12 J −1 C 2 m −1 ) × (3.0 ×10−10 m)6
bond where X, Y = N,
O, or F = −1.4 ×10−21 J
Ion–dipole 1/r2 15   which, upon multiplication by Avogadro’s constant, corre-
Dipole–dipole 1/r3 2 Between stationary sponds to –0.83 kJ mol−1.
polar molecules
1/r6 0.3 Between rotating
Self-test 35.5 Calculate the average interaction energy, in
polar molecules units of joules per mole (J mol−1), between a water molecule
London 1/r6 2 Between all types and a benzene molecule separated by 1.0 nm.
(dispersion) of molecules Answer: −2.1 J mol−1
and ions

www.ebook3000.com

Atkins09819.indb 314 9/11/2013 11:45:03 AM


(a) (b)
Figure 35.4 (a) A polar molecule (dark arrow) can induce
a dipole (light arrow) in a nonpolar molecule, and (b) the
orientation of the latter follows that of the former, so the
interaction does not average to zero.
(d)Induced dipole–induced dipole
interactions
Nonpolar molecules (including closed-shell atoms, such as Ar)
attract one another even though neither has a permanent dipole
moment. The abundant evidence for the existence of interac-
tions between them is the formation of condensed phases of
nonpolar substances, such as the condensation of hydrogen or
argon to a liquid at low temperatures and the fact that benzene
is a liquid at normal temperatures.
The interaction between nonpolar molecules arises from the
transient dipoles that all molecules possess as a result of fluc-
tuations in the instantaneous positions of electrons. To appreci-
ate the origin of the interaction, suppose that the electrons in
one molecule flicker into an arrangement that gives the mol-
ecule an instantaneous dipole moment of magnitude μ1* . This
dipole generates an electric field that polarizes the other mol-
ecule, and induces in that molecule an instantaneous dipole
moment of magnitude μ2. The two dipoles attract each other
and the potential energy of the pair is lowered. Although the
first molecule will go on to change the size and direction of its
instantaneous dipole, the electron distribution of the second
molecule will follow; that is, the two dipoles are correlated in
direction (Fig. 35.5). Because of this correlation, the attraction
between the two instantaneous dipoles does not average to zero,
and gives rise to an induced dipole–induced dipole interaction.
This interaction is called either the dispersion interaction or the
London interaction (for Fritz London, who first described it).
(a) (b)
Figure 35.5 (a) In the dispersion interaction, an instantaneous
dipole on one molecule induces a dipole on another molecule,
and the two dipoles then interact to lower the energy. (b)
The two instantaneous dipoles are correlated, and although
they occur in different orientations at different instants, the
interaction does not average to zero.
Atkins09819.indb 315
35 Interactions between molecules  315
The strength of the dispersion interaction depends on the
polarizability of the first molecule because the instantaneous
dipole moment of magnitude μ1* depends on the looseness
of the control that the nuclear charge exercises over the outer
electrons. The strength of the interaction also depends on the
polarizability of the second molecule, for that polarizability
determines how readily a dipole can be induced by another
molecule. The actual calculation of the dispersion interaction is
quite involved, but a reasonable approximation to the interac-
tion energy is given by the London formula:
V =−
C 3 II
C = α ′2 α ′2 1 2 London formula (35.8)
r6 2 I1 + I 2
where I1 and I2 are the ionization energies of the two molecules.
This interaction energy is also proportional to the inverse sixth
power of the separation of the molecules, which identifies it
as a third contribution to the van der Waals interaction. The
dispersion interaction generally dominates all the interactions
between molecules other than hydrogen bonds.
Brief illustration 35.6 The London interaction
For two CH4 molecules separated by 0.30 nm, we can use eqn
35.8 with α ′ = 2.6 × 10−30 m3 and I ≈ 700 kJ mol−1 and obtain
3 ⎧⎪ (7.00 ×105 Jmol −1 )2 ⎫⎪
× (2.6 ×10−30 m3 )2 × ⎨ −1 ⎬
⎪⎩ 2 × (7.00 ×10 Jmol ) ⎪⎭
2 5
V =−
(0.30 × 10−9 m)6
= −4.9 kJmol −1
A very approximate check on this figure is the enthalpy of
vaporization of methane, which is 8.2 kJ mol−1. However, this
comparison is insecure, partly because the enthalpy of vapori-
zation is a many-body quantity and partly because the long-
distance assumption breaks down.
Self-test 35.6 Estimate the energy of the London interaction
for two He atoms separated by 1.0 nm.
Answer: −0.071 J mol−1
35.3 Hydrogen bonding
The interactions described so far are universal in the sense that
they are possessed by all molecules independent of their spe-
cific identity. However, there is a type of interaction possessed
by molecules that have a particular constitution. A hydrogen
bond is an attractive interaction between two species that arises
from a link of the form A–H…B, where A and B are highly elec-
tronegative elements and B possesses a lone pair of electrons.
Hydrogen bonding is conventionally regarded as being limited
9/11/2013 11:45:13 AM
 316 8 Interactions
to N, O, and F but, if B is an anionic species (such as Cl−), it may
also participate in hydrogen bonding. There is no strict cut-off
for an ability to participate in hydrogen bonding, but N, O, and
F participate most effectively.
The formation of a hydrogen bond can be regarded either
as the approach between a partial positive charge of H and a
partial negative charge of B or as a particular example of delo-
calized molecular orbital formation in which A, H, and B each
supply one atomic orbital from which three molecular orbit-
als are constructed (Fig. 35.6). Experimental evidence and
theoretical arguments have been presented in favour of both
views and the matter has not yet been resolved. The electro-
static interaction model can be understood readily in terms of
the discussion in Section 35.1. Here we develop the molecular
orbital model.
Thus, if the A−H bond is regarded as formed from the over-
lap of an orbital on A, χA, and a hydrogen 1s orbital, χH, and
the lone pair on B occupies an orbital on B, χB, then, when the
two molecules are close together, we can build three molecular
orbitals from the three basis orbitals:
ψ = c1 χ A + c2 χ H + c3 χ B
One of the molecular orbitals is bonding, one almost non-
bonding, and the third antibonding. These three orbitals need
to accommodate four electrons (two from the original A−H
bond and two from the lone pair of B), so two enter the bond-
ing orbital and two enter the nonbonding orbital. Because the
antibonding orbital remains empty, the net effect—depending
on the precise energy of the almost nonbonding orbital—may
be a lowering of energy.
A H B
Energy
Figure 35.6 The molecular orbital interpretation of the
formation of an A–HB hydrogen bond. From the three A, H,
and B orbitals, three molecular orbitals can be formed (their
relative contributions are represented by the sizes of the
spheres). Only the two lower-energy orbitals are occupied (four
electrons total, two from the original A–H bond and two from
the B lone pair), and there may therefore be a net lowering of
energy compared with the separate AH and B species.
www.ebook3000.com
Atkins09819.indb 316
In practice, the strength of the bond is found to be about
20 kJ mol−1. Because the bonding depends on orbital overlap,
it is virtually a contact-like interaction that is turned on when
AH touches B and is zero as soon as the contact is broken. If
hydrogen bonding is present, it dominates the other intermo-
lecular interactions. The properties of liquid and solid water,
for example, are dominated by the hydrogen bonding between
H2O molecules. The structure of DNA and hence the trans-
mission of genetic information is crucially dependent on the
strength of hydrogen bonds between base pairs. The struc-
tural evidence for hydrogen bonding comes from noting that
the internuclear distance between formally nonbonded atoms
is less than expected on the basis of their van der Waals radii,
which suggests that a dominating attractive interaction is pre-
sent. For example, the O−O distance in O−H…O is expected to
be 280 pm on the basis of van der Waals radii, but is found to be
270 pm in typical compounds. Moreover, the H…O distance is
expected to be 260 pm but is found to be only 170 pm.
Hydrogen bonds may be either symmetric or unsymmet-
ric. In a symmetric hydrogen bond, the H atom lies midway
between the two other atoms. This arrangement is rare, but
occurs in F–H…F−, where both bond lengths are 120 pm. More
common is the unsymmetrical arrangement, where the A−H
bond is shorter than the H…B bond. Simple electrostatic argu-
ments, treating A−H…B as an array of point charges (partial
negative charges on A and B, partial positive on H) suggest that
the lowest energy is achieved when the bond is linear, because
then the two partial negative charges are furthest apart. The
experimental evidence from structural studies supports a linear
or near-linear arrangement.
Brief illustration 35.7 The hydrogen bond
A common hydrogen bond is that formed between OKH
groups and O atoms, as in liquid water and ice. In Problem
35.4, you are invited to use the electrostatic model to calcu-
late the dependence of the potential energy of interaction on
the OOH angle, denoted θ in 4, and the results are plotted
in Fig. 35.7. We see that at θ = 0 when the OHO atoms lie in a
straight line; the molar potential energy is −19 kJ mol−1.
O
R
θ
O H
r
4
Self-test 35.7 Use Fig. 35.7 to explore the dependence of the
interaction energy on angle: at what angle does the interaction
energy become negative?
9/11/2013 11:45:21 AM
 35 Interactions between molecules  317

300
that the energy of interaction of three (or more) molecules is
the sum of the pairwise interaction energies alone. The total
Potential energy, V/(kJ mol–1)

dispersion energy of three closed-shell atoms, for instance, is

200 given approximately by the Axilrod–Teller formula:

100 C6 C C C′
V =− − 6 − 6 + Axilrod−Teller
(35.10a)
rAB6 rBC 6 rCA 6 (rAB rBC rCA )3 formula

0 where

C ′ = a(3 cos θ A cosθ B cosθ C + 1) (35.10b)

–100
–180 –90 0 90 180
Angle, θ/°
The parameter a is approximately equal to 34 α ′C6 ; the angles θ
Figure 35.7 The variation of the energy of interaction are the internal angles of the triangle formed by the three atoms
(according to the electrostatic model) of a hydrogen bond (5). The term in C′ (which represents the non-additivity of the
as the angle between the O–H and :O groups is changed. pairwise interactions) is negative for a linear arrangement of
atoms (so that arrangement is stabilized) and positive for an
Answer: Only ±12°, so that the energy is negative (and the interaction is equilateral triangular cluster (so that arrangement is destabi-
attractive) only when the atoms are close to a linear arrangement. lized). It is found that the three-body term contributes about
10 per cent of the total interaction energy in liquid argon.
B

35.4 The total interaction rAB

rBC
θB
Here we consider molecules that are unable to participate in A θA
θC
hydrogen bond formation. The total attractive interaction rCA
energy between rotating molecules is then the sum of the C
5
dipole−dipole, dipole−induced dipole, and dispersion interac-
tions. Only the dispersion interaction contributes if both mol- When molecules are squeezed together, the nuclear and elec-
ecules are nonpolar. In a fluid phase, all three contributions to tronic repulsions begin to dominate the attractive forces. The
the potential energy vary as the inverse sixth power of the sepa- repulsions increase steeply with decreasing separation in a way
ration of the molecules, so we may write that can be deduced only by very extensive, complicated molec-
ular structure calculations of the kind described in Topics
C6
V =− (35.9) 28–30 (Fig. 35.8).
r6
where C6 is a coefficient that depends on the identity of the
molecules.
Although attractive interactions between molecules are often
Potential energy, V(r)

expressed as in eqn 35.9, we must remember that this equation

has only limited validity. First, we have taken into account only Repulsion
dipolar interactions of various kinds, for they have the longest
range and are dominant if the average separation of the mole-
cules is large. However, in a complete treatment we should also Total
consider quadrupolar and higher-order multipole interactions, 0
particularly if the molecules do not have permanent dipole
Attraction
moments. Secondly, the expressions have been derived by
Separation, r
assuming that the molecules can rotate reasonably freely. That
is not the case in most solids, and in rigid media the dipole– Figure 35.8 The general form of an intermolecular potential
dipole interaction is proportional to 1/r3 (as in Justification energy curve (the graph of the potential energy of two closed-
35.2) because the Boltzmann averaging procedure is irrelevant shell species as the distance between them is changed). The
when the molecules are trapped into a fixed orientation. attractive (negative) contribution has a long range, but the
A different kind of limitation is that eqn 35.9 relates to the repulsive (positive) interaction increases more sharply once the
interactions of pairs of molecules. There is no reason to suppose molecules come into contact.

Atkins09819.indb 317 9/11/2013 11:45:31 AM

 318 8 Interactions

In many cases, however, progress can be made by using a Table 35.2* Lennard-Jones parameters for the (12,6) potential
greatly simplified representation of the potential energy, where
ε/(kJ mol−1) r0/pm
the details are ignored and the general features expressed by
a few adjustable parameters. One such approximation is the Ar 128 362
hard-sphere potential energy, in which it is assumed that the Br2 536 427
potential energy rises abruptly to infinity as soon as the parti- C6H6 454 6177
cles come within a separation d: Cl2 368 448.5

Hard-sphere H2 34 297
V = ∞ for r ≤ d V = 0 for r > d potential energy
(35.11)
He 11 258
Xe 236 426
This very simple expression for the potential energy is sur-
prisingly useful for assessing a number of properties. Another * More values are given in the Resource section.

widely used approximation is the Mie potential energy:

a very poor representation of the repulsive potential energy, and
C C that an exponential form, e- r /r , is greatly superior. An exponen-
0

V = nn − mm Mie potential energy (35.12)

r r
with n > m. The first term represents repulsions and the sec-
ond term attractions. The Lennard-Jones potential energy is
a special case of the Mie potential energy with n = 12 and m = 6
(Fig. 35.9); it is often written in the form
⎧⎪⎛ r ⎞ 12 ⎛ r ⎞ 6 ⎫⎪
V = 4ε ⎨⎜ 0 ⎟ − ⎜ 0 ⎟ ⎬ Lennard-Jones potential energy
⎪⎩⎝ r ⎠ ⎝ r ⎠ ⎪⎭
The two parameters are ε, the depth of the well (not to be con-
fused with the symbol of the permittivity of a medium), and r0,
the separation at which V = 0 (Table 35.2).
Although the Lennard-Jones potential energy has been used in
many calculations, there is plenty of evidence to show that 1/r12 is
tial function is more faithful to the exponential decay of atomic
wavefunctions at large distances, and hence to the overlap that is
responsible for repulsion. The potential energy with an exponen-
tial repulsive term and a 1/r6 attractive term is known as an exp-6
potential energy. These expressions for the potential energy can
be used to calculate the virial coefficients of gases, as explained in
Topic 36, and through them various properties of real gases, such
as the Joule–Thomson coefficient (Topic 56). They are also used
(35.13) to model the structures of condensed fluids.
With the advent of atomic force microscopy (AFM), in
which the force between a molecular sized probe and a surface
is monitored (Topic 95), it has become possible to measure
directly the forces acting between molecules. The force, F, is the
negative slope of the potential energy, so for the Lennard-Jones
potential energy between individual molecules we write

dV 24ε ⎧⎪ ⎛ r0 ⎞ ⎛ r0 ⎞ ⎫⎪
13 7
12
F =− = −
r0 ⎨⎪ ⎜⎝ r ⎟⎠ ⎜⎝ r ⎟⎠ ⎬⎪
2 (35.14)
dr
Repulsion, ⎩ ⎭
Potential energy, V(r)/4ε

8
1/r12

4 Total
Example 35.1 Calculating an intermolecular force from
21/6r0
0 the Lennard-Jones potential energy
Use the expression for the Lennard-Jones potential energy
–4 –ε
Attraction, to estimate the greatest net attractive force between two N2
–1/r6 molecules.
–8
0 0.5 1 1.5 Method The magnitude of the force is greatest at the distance
Separation, r/r0
r at which dF/dr = 0. Therefore differentiate eqn 35.14 with
respect to r, set the resulting expression to zero, and solve for
Figure 35.9 The Lennard-Jones potential energy is another
r. Finally, use the value of r in eqn 35.14 to calculate the cor-
approximation to the true intermolecular potential energy
responding value of F.
curves. It models the attractive component by a contribution
that is proportional to 1/r6 and the repulsive component by Answer Because dx n = nx n−1, the derivative of F with respect
a contribution that is proportional to 1/r12. Specifically, these to r is
choices result in the Lennard-Jones (12,6) potential. Although
there are good theoretical reasons for the former, there dF 24ε ⎧ ⎛ 1 ⎞ ⎛ 1 ⎞⎫ ⎧ 7 26r 6 ⎫
= ⎨2 ⎜ −13r013 14 ⎟ − ⎜ −7r07 8 ⎟ ⎬ = 24εr06 ⎨ 8 − 140 ⎬
is plenty of evidence to show that 1/r12 is only a very poor dr r0 ⎩ ⎝ r ⎠ ⎝ r ⎠⎭ ⎩r r ⎭
approximation to the repulsive part of the curve.

www.ebook3000.com

Atkins09819.indb 318 9/11/2013 11:45:40 AM

 35 Interactions between molecules  319

It follows that dF/dr = 0 when From Table 35.2, ε = 1.268 × 10 −21 J and r 0 = 3.919 × 10 −10 m. It
follows that
7 26r06
− = 0 or 7r 6 − 26r06 = 0
r 8 r 14 2.396 × (1.268 ×10−21 J)
F =− = −7.752 ×10−12 N
3.919 ×10−10 m
or
where we have used 1 N = 1 J m−1. That is, the magnitude of the
1/6
⎛ 26 ⎞ force is about 8 pN.
r = ⎜ ⎟ r0 = 1.244r0
⎝ 7⎠
Self-test 35.8 At what separation re does the minimum of the
potential energy curve occur for a Lennard-Jones potential?
At this separation the force is
Answer: re = 21/6r 0

24ε ⎧⎪ ⎛ r ⎞ 13 ⎛ r ⎞ 7 ⎫⎪
F= ⎨2 ⎜ 1.244r ⎟ − ⎜ 1.244r ⎟ ⎬ = −2.396ε /r0
0 0
r0 ⎪⎩ ⎝ 0⎠ ⎝ 0⎠ ⎪
⎭

Checklist of concepts
☐ 1. A van der Waals interaction between closed-shell mol- ☐ 3. A hydrogen bond is an interaction of the form X–HY,
ecules is inversely proportional to the sixth power of where X and Y are typically N, O, or F.
their separation. ☐ 4. The Lennard-Jones potential energy function is a
☐ 2. The following molecular interactions are important: model of the total intermolecular potential energy.
charge–charge, charge–dipole, dipole–dipole, dipole–
induced dipole, dispersion (London), hydrogen bonding.

Checklist of equations

Property Equation Comment Equation number

Potential energy of interaction between V = Q1Q2/4πεr The relative permittivity of 35.1b

two point charges in a medium the medium is εr = ε/ε0

Energy of interaction between a point V = −μ1Q2/4πε0r2 35.2

dipole and a point charge

Energy of interaction between two fixed V = μ1μ2f(θ)/4πε0r3, f(θ) = 1−3cos2 θ Parallel dipoles 35.4
dipoles

Energy of interaction between two V = −2μ12 μ22 / 3(4 π ε 0 )2 kTr 6 35.5

rotating dipoles

Energy of interaction between a polar V = −μ12α 2′ /4 π ε 0r 6 35.7

molecule and a polarizable molecule

London formula V = − 23 α1′ α 2′ [ I1I2 /(I1 + I2 )]/ r 6 35.8

Axilrod–Teller formula 6 − C / r 6 − C / r 6 + C ′ /(r r r )3

V = −C6 / rAB 6 BC 6 CA AB BC CA Applies to closed-shell 35.10
atoms

Lennard-Jones potential energy V = 4ε{(r0/r)12 − (r0/r)6} 35.13

Atkins09819.indb 319 9/11/2013 11:45:49 AM

 TOPIC 36

Real gases

Contents ➤ What do you need to know already?

36.1 Molecular interactions in gases 321 For the first three sections you need to be familiar with
Brief illustration 36.1: Interactions in gases 321 the perfect gas law (Foundations, Topic 1) and the origin
36.2 The virial equation of state 321 of the attractive and repulsive molecular interactions, as
Brief illustration 36.2: The virial equation of state 322 expressed by the Lennard-Jones potential energy function
36.3 The van der Waals equation 323 (Topic 35). You should be familiar with the qualitative
(a) The development of the equation 323 features of the kinetic theory of gases (which is treated in
Example 36.1: Using the van der Waals equation depth in Topic 78). For the more advanced final section, you
to estimate a molar volume 324 need to be aware of the concept of internal pressure (Topic
(b) The reliability of the equation 325 58) and the development of thermodynamic equations of
Brief illustration 36.3: Criteria for perfect state (Topic 66).
gas behaviour 325
(c) Critical behaviour 326
Brief illustration 36.4: The critical temperature 326 The state of a pure gas is specified by giving its volume, amount
36.4 Thermodynamic considerations 327 of substance, pressure, and temperature. However, it has
(a) The internal pressure 327 been established experimentally that it is sufficient to specify
Example 36.2: Writing an expression for the only three of these variables, for then the fourth variable is
internal pressure of a real gas 327 fixed. That is, it is an experimental fact that each substance is
(b) The statistical origins of equations of state 328 described by an equation of state, an equation that expresses
Brief illustration 36.5: The configuration integral 328 any one of these four variables in terms of the other three. Thus,
Checklist of concepts 329 the general form of an equation of state is
Checklist of equations 329
p = f (T ,V , n) Equation of state (36.1)

where p is the pressure, V the volume, n the amount of mol-

➤ Why do you need to know this material?
To understand the atmospheres of this and other planets,
you need to understand gases. Also, many industrial
processes involve gases, and both the outcome of the
reaction and the design of the reaction vessels depend on
a knowledge of their properties.
➤ What is the key idea?
In a real gas weak attractions and repulsions can cause
deviations from the perfect gas law.
ecules, and T the (absolute) temperature. This equation tells us
that if we know the values of n, T, and V for a particular sub-
stance, then the pressure has a fixed value.
Each substance is described by its own equation of state, but
the explicit form of the equation is known in only a few special
cases. As discussed in Topics 1 and 66, the equation of state of
a perfect gas, which is the limiting form of the equation of state
for any gas as p → 0, is
pV = nRT Perfect gas law (36.2)
where R is a universal constant, the same for all gases
(Foundations, Topic 1). It is established in Foundations, Topic
2, that R = NAk, where NA is Avogadro’s constant and k is
Boltzmann’s constant. Our task is to discover how this equation
is modified when intermolecular interactions are important.

www.ebook3000.com

Atkins09819.indb 320 9/11/2013 11:45:53 AM


At the microscopic level, a perfect gas is characterized by
the lack of intermolecular interactions and a completely ran-
dom distribution of molecules in ceaseless motion (Topic 78).
However, in a real gas there are weak attractions and repul-
sions which have minimal effect on the relative locations of
the molecules but which cause deviations from the perfect gas
law. Deviations from the law are particularly important at high
pressures and low temperatures, especially when a gas is on the
point of condensing to liquid at low temperatures. At that point,
the molecules of a gas have insufficient kinetic energy to escape
from each other’s attraction and they stick together. Although
molecules attract each other when they are a few diameters
apart, they repel each other as soon as they come into contact.
This repulsion is responsible for the fact that liquids and solids
have a definite bulk and do not collapse to an infinitesimal point.
36.1 Molecular interactions in gases
Repulsive forces between molecules assist expansion and attrac-
tive forces assist compression. Repulsive forces are significant
only when molecules are almost in contact: they are short-range
interactions, even on a scale measured in molecular diameters
(Fig. 36.1). Because they are short-range interactions, repulsions
can be expected to be important only when the average separa-
tion of the molecules is small. This is the case at high pressure,
when many molecules occupy a small volume. On the other
hand, attractive intermolecular forces have a relatively long range
and are effective over several molecular diameters. They are
important when the molecules are close together but not nec-
essarily touching (at the intermediate separations in Fig. 36.1,
Potential energy, Ep
Repulsion dominant
0 Vm
Attraction
dominant
Separation
Figure 36.1 The variation of the potential energy of two
molecules (at very small separations) indicates that the
interactions between them are strongly repulsive at these
distances. At intermediate separations, where the potential
energy is negative, the attractive interactions dominate. At
large separations (on the right) the potential energy is zero and
there is no interaction between the molecules.
Atkins09819.indb 321
36 Real gases  321
within a few molecular diameters of one another). Attractive
forces are ineffective when the molecules are far apart (well to the
right in Fig. 36.1). Intermolecular forces are also important when
the temperature is so low that the molecules travel with such low
mean speeds that they can be captured by one another.
Brief illustration 36.1 Interactions in gases
To gain some insight into distances that can make interac-
tions between particles in the gas phase important, consider
two Ar atoms. A model of the intermolecular potential energy
of two atoms is the Lennard-Jones potential energy intro-
duced in Topic 35, V = 4ε{(r 0/r)12 – (r 0/r)6}, which is a mini-
mum (most negative) at r = 21/6r 0 and passes through zero at
r = r 0. Because r 0 = 362 pm for argon atoms (Table 35.2), the
potential energy of interaction between two Ar atoms passes
through zero at r = 362 pm. This is the distance below which
repulsions become dominant. Because the diameter of an Ar
atom is 142 pm, this distance corresponds to about 2.5 atomic
diameters. The potential energy of interaction between two Ar
atoms is a minimum (that is, their mutual attraction is great-
est) at r = 21/6 × 362 pm = 406 pm, which corresponds to 2.9
atomic diameters.
Self-test 36.1 Estimate the molar energy of the dispersion
interaction (use the London formula, eqn 35.8) for two Ar
atoms separated by (i) 407 pm (about 3 atomic diameters) and
(ii) 1.0 nm (about 7 atomic diameters).
Answer: (i) −691 J mol−1; (ii) −3.1 J mol−1
36.2 The virial equation of state
The general form of the equation of state of a real gas can be
inferred from measurements of the pressure for various val-
ues of the temperature, volume, and amount of molecules in a
sample. To do so, it is convenient to begin by defining the com-
pression factor, Z, of a gas as the ratio of its measured molar
volume, Vm = V/n, to the molar volume of a perfect gas, Vm° , at
the same pressure and temperature:
Z= Compression factor of a gas (36.3a)
V m°
For a perfect gas Vm = V °m , so Z = 1 under all conditions; there-
fore, deviation of Z from 1 is a measure of departure from per-
fect behaviour. Furthermore, because Vm° can be replaced by
RT/p, an alternative version of eqn 36.3a is
pVm
Z= , implying pVm = ZRT for all gases (36.3b)
RT
Some experimental values of Z are plotted in Fig. 36.2. At very
low pressures, all the gases shown have Z ≈ 1 and behave nearly
9/11/2013 11:45:58 AM
 322 8 Interactions

2
C2H4
Table 36.1* Second virial coefficients, B/(cm3 mol−1)

CH4 Temperature
Compression factor, Z

H2 273 K 600 K
Ar −21.7 11.9
Perfect
1 CO2 −142 −12.4
H2
1
p/atm
N2 −10.5 21.7
Z
10
0.98 CH4 Xe −153.7 −19.6
*More values are given in the Resource section.
NH3 0.96 C2H4
NH3
0
0 200 400 600 800
first approximation (in this case pVm = RT) is treated as the first
p/atm term in a series in powers of a variable (in this case p). A more
convenient expansion for many applications is
Figure 36.2 The variation of the compression factor, Z, with
pressure for several gases at 0 °C. A perfect gas has Z = 1 at all ⎛ B C ⎞
pressures. Notice that although the curves approach 1 as p → 0, pVm = RT ⎜ 1 + + +…⎟ Virial equation of state (36.4b)
⎝ Vm Vm2 ⎠
they do so with different slopes.
These two expressions are two versions of the virial equation of
perfectly. At high pressures, all the gases have Z > 1, signify- state (the name virial comes from the Latin word for force). By
ing that they have a larger molar volume than a perfect gas at comparing the expression with eqn 36.3 we see that the term in
the same temperature. Repulsive forces are now dominant. At parentheses can be identified with the compression factor, Z:
intermediate pressures, most gases have Z < 1, indicating that
the attractive forces are reducing the molar volume relative to B C
Z = 1+ + 2 +… Compression factor in terms of
(36.5)
that of a perfect gas. Vm Vm the virial coefficients
Figure 36.3 shows the experimental isotherms, plots of data
(in this case pressure and volume data) obtained at constant The coefficients B, C, … (which are sometimes denoted B2,
temperature, for carbon dioxide. At large molar volumes and B3, …) depend on the temperature and are the second, third,
high temperatures the real-gas isotherms do not differ greatly … virial coefficients (Table 36.1); the first virial coefficient is 1.
from perfect-gas isotherms. The small differences suggest that The third virial coefficient, C, is usually less important than the
the perfect gas law is in fact the first term in an expression of second coefficient, B, in the sense that at typical molar volumes
the form C /Vm2 << B /Vm.

pVm = RT (1+ B ′ p + C ′ p2 + …) Virial equation of state (36.4a)

Brief illustration 36.2 The virial equation of state
This expression is an example of a common procedure in physi-
To use eqn 36.4b (up to the B term), to calculate the pressure
cal chemistry, in which a simple law that is known to be a good
exerted at 100 K by 0.104 mol O2(g) in a vessel of volume 0.225
dm3, we begin by calculating the molar volume:
140
V 0.225 dm3
120 Vm = = = 2.16 dm3 mol −1 = 2.16 ×10−3 m3 mol −1
50 °C nCO2 0.104 mol
100
40 °C Then, by using the value of B found in Table 36.1 of the
80 F 30.04 °C (Tc)
p/atm

Resource section,
20 °C
60
0 °C RT ⎛ B ⎞
E D p= 1+
Vm ⎜⎝ Vm ⎟⎠
C
40 B

A (8.3145 J mol −1 K −1 ) × (100 K) ⎛ 1.975 × 10−4 m3 mol −1 ⎞

20 = ⎜⎝ 1 − 2.16 × 10−3 m3 mol −1 ⎟⎠
2.16 × 10−3 m3 mol −1
0
0 0.2
Vm/(dm3 mol–1)
0.4 0.6 = 3.50 × 105 Pa , or 350 kPa

Figure 36.3 Experimental isotherms of carbon dioxide at where we have used 1 Pa = 1 J m−3. The perfect gas equation of
several temperatures. The ‘critical isotherm’, the isotherm at the state, eqn 36.1, would give the calculated pressure as 385 kPa,
critical temperature, is at 30.04 °C.

www.ebook3000.com

Atkins09819.indb 322 9/11/2013 11:46:08 AM

 36 Real gases  323

Table 36.2* Boyle temperatures of gases

or 10 per cent higher than the value calculated by using the
virial equation of state. The deviation is significant because TB/K
under these conditions |B/Vm| ≈ 0.1, which is not negligible Ar 411.5
relative to 1.
CO2 714.8
Self-test 36.2 What pressure would 4.56 g of nitrogen gas in a He 22.64
vessel of volume 2.25 dm3 exert at 273 K if it obeyed the virial O2 405.9
equation of state? * More values are given in the Resource section.
Answer: 104 kPa

this temperature, which is called the Boyle temperature, TB,

The virial equation can be used to demonstrate the impor-
tant point that although the equation of state of a real gas may
coincide with the perfect gas law as p → 0, not all its proper-
ties necessarily coincide with those of a perfect gas in that limit.
Consider, for example, the value of dZ/dp, the slope of the
graph of compression factor against pressure. For a perfect gas
dZ/dp = 0 (because Z = 1 at all pressures), but for a real gas from
eqn 36.3 and 36.4a we obtain
the properties of the real gas do coincide with those of a per-
fect gas as p → 0. According to eqn 36.6b, Z has zero slope as
p → 0 if B = 0, so we can conclude that B = 0 at the Boyle tem-
perature. It then follows from eqn 36.4 that pVm ≈ RTB over a
more extended range of pressures than at other temperatures
because the first term after 1 (that is, B/Vm) in the virial equa-
tion is zero and C /Vm2 and higher terms are negligibly small.
For helium TB = 22.64 K; for air TB = 346.8 K; more values are
given in Table 36.2.

dZ
= B ′ + 2 pC ′ + …→ B ′ as p → 0
dp
However, B′ is not necessarily zero, so the slope of Z with
respect to p does not necessarily approach 0 as p → 0, as can be
seen in Fig. 36.2. By a similar argument,
dZ
→ B as Vm → ∞, corresponding to p → 0 (36.6b)
d(1/Vm )
Because the virial coefficients depend on the tempera-
ture, there may be a temperature at which Z → 1 with zero
slope at low pressure or high molar volume (Fig. 36.4). At
(36.6a)
36.3 The van der Waals equation
Conclusions from the virial equations of state can be drawn
only by inserting specific values of the coefficients. It is often
useful to have a broader, if less precise, view of all gases.
Therefore, we introduce the approximate equation of state
suggested by J.D. van der Waals in 1873. This equation is an
excellent example of an expression that can be obtained by
thinking scientifically about a mathematically complicated
but physically simple problem; that is, it is a good example of
‘model building’.

Higher (a) The development of the equation

Compression factor, Z

Boyle temperature
temperature
The van der Waals equation is
Perfect 2
gas nRT ⎛ n⎞
p= −a
V − nb ⎜⎝ V ⎟⎠
1 van der Waals equation (36.7a)

Lower and a derivation is given in the following Justification. The

temperature equation is often written in terms of the molar volume
Vm = V/n as
Pressure, p

Figure 36.4 The compression factor, Z, approaches 1 at low RT a

p= − Alternative form van der Waals equation (36.7b)
pressures, but does so with different slopes. For a perfect gas, the Vm − b Vm2
slope is zero, but real gases may have either positive or negative
slopes, and the slope may vary with temperature. At the Boyle The (positive) constants a and b are called the van der Waals
temperature, the slope is zero and the gas behaves perfectly over parameters. They are characteristic of each gas but independ-
a wider range of conditions than at other temperatures. ent of the temperature (Table 36.3).

Atkins09819.indb 323 9/11/2013 11:46:14 AM

 324 8 Interactions
Table 36.3* van der Waals parameters
a/(atm dm6 a/(Pa m6 b/(10−2
mol−2) mol–2) mol−1)
Ar 1.337 0.1355 3.20
CO2 3.610 0.3658 4.29
He 0.0341 0.00346 2.38
Xe 4.137 0.4192 5.16
* More values are given in the Resource section.
Justification 36.1 The van der Waals equation of state
The repulsive interactions between molecules are taken into
account by supposing that they cause the molecules to behave
as small but impenetrable spheres. The nonzero volume of the
molecules implies that instead of moving in a volume V they
are restricted to a smaller volume V – nb, where nb is approxi-
mately the total volume taken up by the molecules themselves.
This argument suggests that the perfect gas law p = nRT/V
should be replaced by
nRT
p=
V − nb
when repulsions are significant. The closest distance of two
hard-sphere molecules of radius r, and volume Vmolecule = 43 πr 3 ,
is 2r, so the volume excluded is 43 π( 2r)3 , or 8Vmolecule . The
volume excluded per molecule is one-half this volume, or
4Vmolecule, so b ≈ 4VmoleculeNA.
The pressure depends on both the frequency of collisions
with the walls and the force of each collision. Both the fre-
quency of the collisions and their force are reduced by the
attractive forces, which act with a strength proportional to
the molar concentration, n/V, of molecules in the sample.
Therefore, because both the frequency and the force of the
collisions are reduced by the attractive forces, the pressure is
reduced in proportion to the square of this concentration. If
the reduction of pressure is written as −a(n/V)2 , where a is a
positive constant characteristic of each gas, the combined
effect of the repulsive and attractive forces is the van der Waals
equation of state as expressed in eqn 36.7.
In this Justification we have built the van der Waals equation
using plausible arguments about the volumes of molecules
and the effects of forces. The equation can be derived in other
ways, but the present method has the advantage that it shows
how to derive the form of an equation out of general ideas. The
derivation also has the advantage of keeping imprecise the
significance of the parameters a and b: they are much better
regarded as empirical parameters than as precisely defined
molecular properties (but see eqn 36.11 below and Topic 66 for
a precise thermodynamic interpretation of a).
Atkins09819.indb 324
Example 36.1 Using the van der Waals equation to
dm3 estimate a molar volume
Estimate the molar volume of CO2 at 500 K and 100 atm by
treating it as a van der Waals gas.
Method To express eqn 36.7b as an equation for the molar vol-
ume, we multiply both sides by (Vm − b)Vm2 , to obtain
(Vm − b)Vm2 p = RTVm2 − (Vm − b)a
and, after division by p, collect powers of Vm to obtain
⎛ RT ⎞ 2 ⎛ a ⎞ ab
Vm3 − ⎜ b + V + V − =0
⎝ p ⎟⎠ m ⎜⎝ p ⎟⎠ m p
Although closed expressions for the roots of a cubic equa-
tion can be given, they are very complicated. Unless analyti-
cal solutions are essential, it is usually more expedient to solve
such equations by using mathematical software.
Answer According to Table 36.3, a = 3.610 dm 6 atm mol −2
and b = 4.29 × 10 −2 dm 3 mol−1. Under the stated conditions,
RT/p = 0.410 dm3 mol−1. The coefficients in the equation for Vm
are therefore
b + RT/p = 0.453 dm3mol −1
a/p = 3.61 × 10− 2 (dm3mol −1 )2
ab/p = 1.55 × 10− 3 (dm3mol −1 )3
Therefore, on writing x = Vm/(dm3 mol−1), the equation to solve
is
x 3 − 0.453x 2 + (3.61 × 10− 2 )x − (1.55 × 10− 3 ) = 0
The acceptable root is x = 0.366, which implies that Vm = 0.366
dm3 mol−1. The molar volume of a perfect gas under these con-
ditions is 0.410 dm3 mol−1.
Self-test 36.3 Calculate the molar volume of argon at 100 °C
and 100 atm on the assumption that it is a van der Waals gas.
Answer: 0.298 dm3 mol−1
It is too optimistic to expect a single, simple expression to be
the true equation of state of all substances, and accurate work
on gases must resort to the virial equation, use tabulated val-
ues of the coefficients at various temperatures, and analyse the
systems numerically. The advantage of the van der Waals equa-
tion, however, is that it is analytical (that is, expressed symboli-
cally) and allows us to draw some general conclusions about
real gases. When the equation fails we must use one of the other
equations of state that have been proposed (some are listed in
Table 36.4), invent a new one, or go back to the virial equation.
www.ebook3000.com
9/11/2013 11:46:23 AM
 36 Real gases  325

Table 36.4 Selected equations of state

Equation Critical constants 1.5

pc Vc Tc

Pressure, p
1.0
RT
Perfect gas p=
Vm 0.8

RT a a 8a
van der Waals p= − 3b
Vm − b Vm2 27b2 27 Rb
,T
re
atu
RT a 1 ⎛ 2aR ⎞
1/2
2 ⎛ 2a ⎞
1/2 Volume, V p er
Berthelot p= − 3b m
Vm − b TVm2 12 ⎝⎜ 3b3 ⎠⎟ 3 ⎝⎜ 3bR ⎠⎟ Te

Figure 36.5 The surface of possible states allowed by the van

RTe −a/RTVm a a
Dieterici p= 2b der Waals equation. The curves are labelled with the ‘reduced
Vm − b 4e2b2 4bR
temperature’, Tr = T/Tc.
RT ⎛ B(T ) C(T ) ⎞
Virial p= ⎜ 1+ V + V 2 +…⎟
Vm ⎝ m m ⎠ 1.5
1.5

Reduced pressure, p/pc

(b) The reliability of the equation 1 1

We can begin to judge reliability of the equation by compar-

ing the isotherms it predicts with the experimental isotherms
in Fig. 36.3. Some calculated isotherms are shown in Fig. 36.5
and Fig. 36.6. Apart from the oscillations they do resemble
experimental isotherms quite well. The oscillations, the van der
Waals’ loops, are unrealistic because they suggest that under
some conditions an increase of pressure results in an increase of
volume. Therefore they are replaced by horizontal lines drawn
so the loops define equal areas above and below the lines: this
procedure is called the Maxwell construction (1). The van der
Waals parameters, such as those in Table 36.3, are found by fit-
ting the calculated curves to the experimental curves.
Equal areas
0.5
0.8
0
0.1 1 10
Reduced volume, Vm/Vc
Figure 36.6 Isotherms calculated by using the van der Waals
equation of state. The axes are labelled with the ‘reduced
pressure’, p/pc, and ‘reduced volume’, Vm/Vc, where pc = a/27b2
and Vc = 3b. The individual isotherms are labelled with the
‘reduced temperature’, T/Tc, where Tc = 8a/27Rb. The van der
Waals’ loops are normally replaced by horizontal straight lines.

Brief illustration 36.3 Criteria for perfect gas behaviour

1 Vm = RT/p = 33 dm3 mol−1, so the criterion Vm ≫ b for perfect
An important feature of the van der Waals equation is that
perfect gas isotherms are obtained at high temperatures and large
molar volumes. When the temperature is high, RT may be so
large that the first term in eqn 36.7b greatly exceeds the second.
Furthermore, if the molar volume is large, in the sense Vm  b,
then the denominator Vm − b ≈ Vm. Under these conditions, the
equation reduces to p = RT/Vm, the perfect gas equation.
For benzene a = 18.57 atm dm 6 mol −2 (1.882 Pa m 6 mol −2)
and b = 0.1193 dm 3 mol−1 (1.193 × 10 −4 m 3 mol−1); its normal
boiling point is 353 K. Treated as a perfect gas at T = 400 K
and p = 1.0 atm, benzene vapour has a molar volume of
gas behaviour is satisfied. It follows that a/Vm 2 ≈ 0.017 atm,
which is 1.7 per cent of 1.0 atm. Therefore, we can expect ben-
zene vapour to deviate only slightly from perfect gas behav-
iour at this temperature and pressure.
Self-test 36.4 Can argon gas be treated as a perfect gas at 400 K
and 3.0 atm?
Answer: Yes

Atkins09819.indb 325 9/11/2013 11:46:45 AM

 326 8 Interactions

(c) Critical behaviour Brief illustration 36.4 The critical temperature

Consider what happens when we compress (reduce the volume The critical temperature of oxygen signifies that it is impos-
of) a sample of gas initially in the state marked A in Fig. 36.3 at sible to produce liquid oxygen by compression alone if its tem-
constant temperature by pushing in a piston. Near A, the pres- perature is greater than 155 K. To liquefy oxygen—to obtain a
sure of the gas rises in approximate agreement with the perfect fluid phase that does not occupy the entire volume—the tem-
gas law. Serious deviations from that law begin to appear when perature must first be lowered to below 155 K, and then the gas
the volume has been reduced to B. compressed isothermally.
At C (which corresponds to about 60 atm for carbon diox- Self-test 36.5 Under which conditions can liquid nitrogen be
ide), all similarity to perfect behaviour is lost, for suddenly the formed by the application of pressure?
piston slides in without any further rise in pressure: this stage Answer: At T < 126 K
is represented by the horizontal line CDE. Examination of the
contents of the vessel shows that just to the left of C a liquid
appears, and there are two phases separated by a sharply defined
surface. As the volume is decreased from C through D to E, the We show in the following Justification that the critical con-
amount of liquid increases. There is no additional resistance stants are related to the van der Waals parameters as follows:
to the piston because the gas can respond by condensing. The
Critical constants
pressure corresponding to the line CDE, when both liquid and 8a a
Vc = 3b Tc = pc = in terms of van der (36.8)
vapour are present in equilibrium, is called the vapour pres- 27 Rb 27b2 Waals parameters
sure of the liquid at the temperature of the experiment.
At E, the sample is entirely liquid and the piston rests on its The first of these relations shows that the critical volume is
surface. Any further reduction of volume requires the exertion about three times the volume occupied by the molecules them-
of considerable pressure, as is indicated by the sharply rising selves. Table 36.4 shows expressions for the critical constants
line to the left of E. Even a small reduction of volume from E to derived from several equations of state.
F requires a great increase in pressure.
The isotherm at the temperature Tc (304.19 K, or 31.04 °C
Justification 36.2
Relating the critical constants to the
for CO2) plays a special role in the theory of the states of mat-
ter. An isotherm slightly below Tc behaves as we have already van der Waals parameters
described: at a certain pressure, a liquid condenses from the gas We see from Fig. 36.6 that, for T < Tc, the calculated isotherms
and is distinguishable from it by the presence of a visible sur- oscillate, and each one passes through a minimum followed by
face. If, however, the compression takes place at Tc itself, then a a maximum. These extrema converge as T → Tc and coincide
surface separating two phases does not appear and the volumes at T = Tc; at the critical point the curve has a flat inflexion (2).
at each end of the horizontal part of the isotherm have merged
to a single point, the critical point of the gas. The tempera-
ture, pressure, and molar volume at the critical point are called
the critical temperature, Tc, critical pressure, pc, and critical
molar volume, Vc, of the substance. Collectively, pc, Vc, and Tc
are the critical constants of a substance (Table 36.5).
At and above Tc, the sample has a single phase which occupies
the entire volume of the container. Such a phase is, by definition, 2
a gas. Hence, the liquid phase of a substance does not form above
the critical temperature. The single phase that fills the entire vol- From the properties of curves, we know that an inflexion of
ume when T > Tc may be much denser that we normally consider this type occurs when both the first and second derivatives are
typical of gases, and the name supercritical fluid is preferred. zero. Hence, we can find the critical temperature by calculat-
ing these derivatives and setting them equal to zero. It follows
Table 36.5* Critical constants of gases from eqn 36.7b that the first and second derivatives of p with
pc/atm Vc/(cm3 mol–1) Tc/K respect to Vm are, respectively,
Ar 48.0 75.3 150.7
dp RT 2a
CO2 72.9 94.0 304.2 =− +
dVm (Vm − b)2 Vm3
He 2.26 57.8 5.2
d2 p 2RT 6a
O2 50.14 78.0 154.8 = −
* More values are given in the Resource section. dVm2 (Vm − b)3 Vm4

www.ebook3000.com

Atkins09819.indb 326 9/11/2013 11:46:50 AM

 36 Real gases  327

We now have another way of interpreting the role of the van der
At the critical point T = Tc , Vm = Vc , and both derivatives are
Waals parameter a: it gives the contribution of molecular inter-
equal to zero:
actions to the internal energy of the gas. Moreover, because the
RTc 2a internal energy is positive for a van der Waals gas, we can infer
− + 3 =0 that the internal energy increases as the gas expands at constant
(Vc − b) Vc
2

temperature. That is plausible: the molecules become further

2RTc 6a
− =0 apart on average and thus interact favourably less strongly.
(Vc − b)3 Vc4

Solving this pair of equations gives (as you should verify) the
expressions for Vc and Tc in eqn 36.8. When they are inserted Example 36.2 Writing an expression for the internal
in the van der Waals equation itself, we find the expression for
pressure of a real gas
pc given there too.
Write an expression for the internal pressure of a gas that
obeys the virial equation of state.

36.4 Thermodynamic considerations Method Begin by rearranging eqn 36.4b to

RT ⎛ B ⎞
The empirical results discussed above can be understood in terms p= 1+ +…⎟
Vm ⎜⎝ Vm ⎠
of the principles of chemical and statistical thermodynamics
developed throughout the text, and especially in Topics 58 and 66.
and then use eqn 36.9, noting that the virial coefficients
The result is deeper insight into the properties of real gases.
depend on temperature.
Answer Because the virial coefficients depend on tempera-
(a) The internal pressure ture, we write
The virial equation and the van der Waals equation are special
⎡
f   g
⎤

cases of the thermodynamic equation of state introduced in ⎢
⎛ ∂p ⎞ ∂ RT ⎛ B ⎞⎥
Topic 66: ⎜⎝ ∂T ⎟⎠ = ⎢ ∂T V ⎜⎝ 1 + V +…⎟⎠ ⎥
V ⎢ m m ⎥
⎢⎣ ⎥⎦
⎛ ∂p ⎞ V
πT =T ⎜ −p (36.9)
⎝ ∂T ⎟⎠ V
Thermodynamic equation of state product rule:
d ( fg)= fdg + gdf

RT ⎡ ∂ ⎛ B ⎞⎤ ⎛ B ⎞ ⎡ ∂ ⎛ RT ⎞ ⎤
= ⎢ ∂T ⎜⎝ 1 + V +…⎟⎠ ⎥ + ⎜⎝ 1 + V +…⎟⎠ ⎢ ∂T ⎜⎝ V ⎟⎠ ⎥
where πT is the internal pressure, the dependence of the inter- Vm ⎣ m ⎦V m ⎣ m ⎦
V
nal energy on the volume at constant temperature:
After evaluating the derivatives and using eqn 36.4b again to
⎛ ∂U ⎞
πT = ⎜ (36.10)
⎝ ∂V ⎟⎠ T
Internal pressure simplify the resulting expression, we obtain

⎛ ∂p ⎞ ⎧⎪ RT ⎛ ∂B ⎞ ⎫⎪ R ⎛ B ⎞
with πT = 0 for a perfect gas, because in the absence of molecu- ⎜⎝ ∂T ⎟⎠ = ⎨ V 2 ⎜⎝ ∂T ⎟⎠ +…⎬ + V ⎜⎝ 1 + V +…⎟⎠
lar interactions, a change in volume—that is, a change in aver- V ⎪⎩ m V ⎪⎭ m m

p/T ( eqn 36.4 b )
age intermolecular separation—has no effect on the internal
energy of a perfect gas at constant temperature. For a van der ⎧⎪ RT ⎛ ∂B ⎞ ⎫⎪ p
=⎨ 2 ⎜ ⎟ +…⎬ +
Waals gas, ⎪⎩ Vm ⎝ ∂T ⎠ V ⎪⎭ T

⎛ ∂p ⎞ nR
⎜⎝ ∂T ⎟⎠ = V − nb It follows from eqn 36.9 that
V

and it follows from eqn 36.9 that ⎡ ⎧⎪ RT ⎛ ∂B ⎞ ⎫⎪ p ⎤

π T = T ⎢⎨ 2 ⎜ ⎟ +…⎬ + ⎥ − p
⎢⎣ ⎪⎩ Vm ⎝ ∂ T ⎠V ⎪⎭ T ⎥⎦
nRT ⎧ nRT n2 ⎫
πT = −⎨ −a 2 ⎬ van der Internal (36.11) ⎧⎪ RT 2 ⎛ ∂B ⎞ ⎫⎪ RT 2 ⎛ ∂B ⎞
V − nb ⎩ V − nb V ⎭ Waals gas pressure =⎨ 2 ⎜ +…⎬ + p − p = 2 ⎜ +…
⎪⎩ Vm ⎝ ∂ T ⎟
⎠V ⎪⎭ Vm ⎝ ∂T ⎟⎠ V
n2
=a
V2

Atkins09819.indb 327 9/11/2013 11:47:03 AM

 328 8 Interactions

For a real gas of atoms (for which the intermolecular interac-

Self-test 36.6 Estimate π T for argon at 275 K given that
tions are isotropic), Z is related to the total potential energy Ep
B(250 K) = −28.0 cm 3 mol−1 and B(300 K) = −15.6 cm 3 mol−1 at
1.0 atm. Hints: (a) Rewrite the expression for πT from Example of interaction of all the particles by
36.2 in terms of the ratio of differences ΔB/ΔT, and then take 1 − βE Real
T as the mean value of the two given temperature values; Z=
N!∫e dτ 1dτ 2 …dτ N
p
gas of
atoms
Configuration
integral
(36.15)
(b) because deviation from perfect gas behaviour is expected
to be small, use the perfect gas law to estimate Vm. where dτi is the volume element for atom i. The physical ori-
Answer: 0.30 kPa gin of this term is that the probability of occurrence of each
arrangement of molecules possible in the sample is given by a
Boltzmann distribution in which the exponent is given by the
potential energy corresponding to that arrangement.
(b)The statistical origins of equations
of state
Brief illustration 36.5 The configuration integral
To establish contact between the empirical equations of state Equation 36.15 is very difficult to manipulate in practice, even
and the statistical thermodynamic discussion in Topics 51–54, for quite simple intermolecular potential energy functions.
we need to use the canonical partition function Q to calculate However, the case of a perfect gas is manageable because when
the internal pressure because only that formulation allows the molecules do not interact with one another, Ep = 0 and
us to include the effects of intermolecular interactions. From hence e − βΕ p =1. Then
the  expression for U in terms of Q (〈E〉 = −(∂ ln Q/∂β)V and
 
V  
N

U = U(0) + 〈E〉), we can write 1

1
1 VN
Z=
N!
e − βEp
∫dτ 1dτ 2 …dτ N = ∫ ∫
dτ 1 dτ 2 … dτ N =
N!  N! ∫
⎛ ∂ ⎛ ∂ lnQ ⎞ ⎞ Internal pressure
V V V
πT = −⎜ ⎜ ⎟ ⎟ in terms of Q (36.12)
⎝ ∂V ⎝ ∂β ⎠ V ⎠ T
where ∫ dτi = V, with V as the volume of the container.

where β = 1/kT. To develop this expression, we need to find
a way to build an intermolecular potential energy into the
expression for Q. The total kinetic energy of a gas is the sum
of the kinetic energies of the individual molecules. Therefore,
even in a real gas the canonical partition function factorizes
into a part arising from the kinetic energy and a factor called
the configuration integral, Z, which depends on the contribu-
tions to the intermolecular potential energy. We write
Z
Q = Q in terms of the configuration integral
Λ3N
where Λ is the thermal wavelength (see eqn 52.7b), Λ = h/
(2πmkT)1/2. It then follows that
Self-test 36.7 Use the result from Brief illustration 36.5 and
eqn 36.12 to calculate the internal pressure of a perfect gas.
Answer: πT = 0; the same result is obtained in Topics 58 and 66
When the potential energy function has the form of a
central hard sphere surrounded by a shallow attractive well
(Fig. 36.7), then detailed calculation, which is too involved
to reproduce here, leads to πT = an2/V2, where a is a constant
(36.13) that is proportional to the area under the attractive part of
Potential energy, Ep

Determines b
⎛ ∂ ⎛ ∂ ln(Z / Λ 3 N ) ⎞ ⎞
πT = −⎜ ⎜ ⎟⎠ ⎟
⎝ ∂V ⎝ ∂β V⎠ T

⎛ ∂ ⎛ ∂ ln Z ⎞ ⎞ ⎛ ∂ ⎛ ∂ ln(1/ Λ 3 N ) ⎞ ⎞ (36.14)
= −⎜ ⎜ ⎟ ⎟ −⎜ ⎟⎠ ⎟
⎝ ∂V ⎝ ∂β ⎠ V ⎠ T ⎝ ∂V ⎜⎝
Intermolecular separation
∂β V ⎠T
0

⎛ ∂ ⎛ ∂ ln Z ⎞ ⎞ ⎛ ∂ ⎛ 1 ∂Z ⎞ ⎞ Determines a
= −⎜ ⎜⎝ ∂β ⎟⎠ ⎟ = − ⎜ ∂V ⎜⎝ Z ∂β ⎟⎠ ⎟
⎝ ∂V V ⎠T ⎝ V ⎠T

Figure 36.7 The van der Waals equation of state can be

In the second line, although the derivative of Λ with respect to derived on the basis that the intermolecular potential
temperature and hence β is nonzero, Λ is independent of vol- energy has a hard core surrounded by a long-range, shallow
ume, so the derivative with respect to volume is zero. attractive well.

www.ebook3000.com

Atkins09819.indb 328 9/11/2013 11:47:13 AM

 36 Real gases  329

the potential. Of course, this is also eqn 36.11 for the inter- isothermally (because πT > 0, and the slope of U with respect
nal pressure of a van der Waals gas. At this point we can to V is positive). The energy rises because, at greater average
conclude that if there are attractive interactions between mol- separations, the molecules spend less time in regions where
ecules in a gas, then its internal energy increases as it expands they interact favourably.

Checklist of concepts
☐ 1. In real gases, molecular interactions affect the equation
of state.
☐ 2. The true equation of state of a gas is expressed in terms
of virial coefficients.
☐ 3. The van der Waals equation of state is an approxima-
tion to the true equation of state in which attractions
are represented by a parameter a and repulsions are
represented by a parameter b.
☐ 4. At or above the critical temperature a single form of
matter, a supercritical fluid, fills a container at every
value of the pressure, and there is no separation of a liq-
uid from the gas.
☐ 5. The internal pressure of a perfect gas is zero; the internal
pressure of a real gas can be written in terms of experimen-
tal parameters (such as the virial coefficients or the van der
Waals parameters) or the canonical partition function.

Checklist of equations
Property Equation Comment Equation number

Compression factor Z = Vm /V m
° Definition 36.3a

Virial equation of state pVm = RT (1+ B /Vm + C /Vm2 +…) 36.4b

van der Waals equation p = RT/(Vm − b) − a /Vm2 a: effect of attractions; b: effect of repulsions 36.7b

Critical constants in terms of the van Vc = 3b 36.8

der Waals parameters Tc = 8a/27Rb
pc = a/27b2

Internal pressure πT = −an2/V2 van der Waals gas 36.11

πT = (∂(∂ ln Q/∂β)V/ ∂V)T, Q is the canonical partition function; Z is the 36.12–36.13
Q = Z/Λ3N configuration integral; β = 1/kT

Atkins09819.indb 329 9/11/2013 11:47:16 AM

 TOPIC 37

Crystal structure

Contents ➤ What is the key idea?

The details of the regular arrangement of atoms in periodic
37.1 Periodic crystal lattices 330
crystals can be determined by X-ray diffraction techniques.
Brief illustration 37.1: Bravais lattices 333
37.2 The identification of lattice planes 333 ➤ What do you need to know already?
(a) The Miller indices 333
You need to be familiar with the wave description of
Brief illustration 37.2: Miller indices 334
electromagnetic radiation (Foundations, Topic 3) and
(b) The separation of planes 334
the significance of Fourier transforms (Mathematical
Example 37.1: Using the Miller indices 334
background 6). Light use is made of the de Broglie relation
37.3 X-ray crystallography 335 (Topic 4) and the equipartition theorem (Foundations,
(a) Bragg’s law 336
Topic 2).
Brief illustration 37.3: Bragg’s law 1 336
Brief illustration 37.4: Bragg’s law 2 336
(b) The electron density 337 A crucial aspect of the link between the structure and proper-
Example 37.2: Calculating a structure factor 338 ties of a solid is the pattern in which the atoms (and molecules)
Example 37.3: Calculating an electron density are stacked together, so here we examine how the structures
by Fourier synthesis 339
of solids are described and determined. First, we see how to
(c) Determination of the structure 339
describe the regular arrangement of atoms in solids. Then we
Brief illustration 37.5: The Patterson synthesis 340
consider the basic principles of X-ray diffraction and see how
37.4 Neutron and electron diffraction 340 the diffraction pattern can be interpreted in terms of the distri-
Example 37.4: Calculating the typical wavelength bution of electron density in a crystal.
of thermal neutrons 341
Checklist of concepts 342
Checklist of equations 342

37.1 Periodic crystal lattices

➤ Why do you need to know this material?
It is necessary to understand the details of the structures of
metallic, ionic, and molecular solids if you want to account
for their properties. To be prepared for the study of the
mechanical, electrical, optical, and magnetic properties
of solids that can form the basis of new materials and
technologies you need to know how the crystal structures
of solids are determined.
A periodic crystal is built up from regularly repeating ‘struc-
tural motifs’, which may be atoms, molecules, or groups
of atoms, molecules, or ions. A space lattice is the pattern
formed by points representing the locations of these motifs
(Fig. 37.1). The space lattice is, in effect, an abstract scaffold-
ing for the crystal structure. More formally, a space lattice is
a three-dimensional, infinite array of points, each of which is
surrounded in an identical way by its neighbours, and which
defines the basic structure of the crystal. In some cases there
may be a structural motif centred on each lattice point, but
that is not necessary. The crystal structure itself is obtained by
associating with each lattice point an identical structural motif.

www.ebook3000.com

Atkins09819.indb 330 9/11/2013 11:47:17 AM


Lattice point
Structural motif
Figure 37.1 Each lattice point specifies the location of a
structural motif (for example, a molecule or a group of
molecules). The crystal lattice is the array of lattice points; the
crystal structure is the collection of structural motifs arranged
according to the lattice.
The solids known as quasicrystals are ‘aperiodic’, in the sense
that the space lattice, though still filling space, does not have
translational symmetry. Our discussion will focus on periodic
crystals only and, to simplify the language, we refer to these
structures simply as ‘crystals’.
The unit cell is an imaginary parallelepiped (parallel-sided
figure) that contains one unit of the translationally repeating
pattern (Fig. 37.2). A unit cell can be thought of as the funda-
mental region from which the entire crystal may be constructed
by purely translational displacements (like bricks in a wall). A
unit cell is commonly formed by joining neighbouring lattice
points by straight lines (Fig. 37.3). Such unit cells are called
primitive. It is sometimes more convenient to draw larger non-
primitive unit cells that also have lattice points at their centres
or on pairs of opposite faces. An infinite number of different
unit cells can describe the same lattice, but the one with sides
that have the shortest lengths and that are most nearly per-
pendicular to one another is normally chosen. The lengths of
the sides of a unit cell are denoted a, b, and c, and the angles
between them are denoted α, β, and γ (Fig. 37.4).
Figure 37.2 A unit cell is a parallel-sided (but not necessarily
rectangular) figure from which the entire periodic crystal
structure can be constructed by using only translations
(not reflections, rotations, or inversions).
Atkins09819.indb 331
37 Crystal structure  331
Figure 37.3 A unit cell can be chosen in a variety of ways, as
shown here. It is conventional to choose the cell that represents
the full symmetry of the lattice. In this rectangular lattice, the
rectangular unit cell would normally be adopted.
a
b c β
α
γ a a
c b c
b γ β
α
Figure 37.4 The notation for the sides and angles of a unit cell.
Note that the angle α lies in the plane (b,c) and perpendicular
to the axis a.
Unit cells are classified into seven crystal systems by noting
the rotational symmetry elements they possess. A symmetry
operation is an action (such as a rotation, reflection, or inver-
sion) that leaves an object looking the same after it has been
carried out. There is a corresponding symmetry element for
each symmetry operation, which is the point, line, or plane
with respect to which the symmetry operation is performed.
For instance, an n-fold rotation (the symmetry operation)
about an n-fold axis of symmetry (the corresponding symme-
try element) is a rotation through 360°/n. (See Topics 31–33 for
a more detailed discussion of symmetry.)
The following are examples of unit cells:
r
A cubic unit cell has four threefold axes in a tetrahedral
array (Fig. 37.5).
r
A monoclinic unit cell has one twofold axis; the unique
axis is by convention the b axis (Fig. 37.6).
r
A triclinic unit cell has no rotational symmetry, and
typically all three sides and angles are different (Fig. 37.7).
Table 37.1 lists the essential symmetries, the elements that
must be present for the unit cell to belong to a particular crystal
system.
9/11/2013 11:47:22 AM
 332 8 Interactions

C3 C3 a a

C3 C3
a

Cubic P Cubic I Cubic F

a b
a a
β
c c

Figure 37.5 A unit cell belonging to the cubic system has four Tetragonal P Tetragonal I Monoclinic P Monoclinic C
threefold axes, denoted C3, arranged tetrahedrally. The insert a b
shows the threefold symmetry.
c

Orthorhombic P Orthorhombic C Orthorhombic I Orthorhombic F

C2 b a a a a
a
γ 120° a 120°
c α β c

Triclinic Hexagonal Trigonal R

Figure 37.6 A unit cell belonging to the monoclinic system has
a twofold axis (denoted C2 and shown in more detail
in the insert).
Figure 37.8 The 14 Bravais lattices. The points are lattice
points, and are not necessarily occupied by atoms. P denotes
a primitive unit cell (R is used for a trigonal lattice), I a body-
centred unit cell, F a face-centred unit cell, and C (or A or B) a
cell with lattice points on two opposite faces. Trigonal lattices
may belong to the rhombohedral or hexagonal systems
(Table  37.1).

There are only 14 distinct space lattices in three dimensions.

These Bravais lattices are illustrated in Fig. 37.8. It is conven-
tional to portray these lattices by primitive unit cells in some
cases and by non-primitive unit cells in others. The following
notation is used:
r
A primitive unit cell (with lattice points only at the
Figure 37.7 A triclinic unit cell has no axes of rotational
corners) is denoted P.
symmetry.
r
A body-centred unit cell (I) also has a lattice point at its
Table 37.1 The seven crystal systems centre.
r
A face-centred unit cell (F) has lattice points at its
System Essential symmetries
corners and also at the centres of its six faces.
Triclinic None r
A side-centred unit cell (A, B, or C) has lattice points at
Monoclinic One C2 axis its corners and at the centres of two opposite faces.
Orthorhombic Three perpendicular C2 axes
Rhombohedral One C3 axis
For simple structures, it is often convenient to choose an atom
belonging to the structural motif, or the centre of a molecule, as
Tetragonal One C4 axis
the location of a lattice point or the vertex of a unit cell, but that
Hexagonal One C6 axis
is not a necessary requirement. Equivalent lattice points within
Cubic Four C3 axes in a tetrahedral arrangement the unit cell of a Bravais lattice have identical surroundings.

www.ebook3000.com

Atkins09819.indb 332 9/11/2013 11:47:26 AM


Brief illustration 37.1 Bravais lattices
Consider a body-centred cubic unit cell of sides a and one of its
corners with coordinates x = 0, y = 0, z = 0 (Fig. 37.9). Starting
from this corner, the centre of the edge that runs along the
y-axis has coordinates x = 0, y = 12 a, z = 0. It follows that the
centres of each edge are equivalent to this point with coordi-
nates x = 0, y = 12 a, z = 0.
z a
0 a/2 a y
a
x
Figure 37.9 The body-centred cubic unit cell used in Brief
illustration 37.1. The arrows show some of the ways in which
the initial (black) point is related by symmetry operations to
the remaining points halfway along each edge.
Self-test 37.1 What points within a face-centred cubic unit cell
are equivalent to the point x = 12 a, y = 0, z = 12 a ?
Answer: The centres of each face
The identification of lattice
37.2
planes
There are many different sets of lattice planes in a crystal
(Fig. 37.10), and we need to be able to identify them. Two-
dimensional lattices are easier to visualize than three-dimen-
sional lattices, so we shall introduce the concepts involved by
referring to two dimensions initially, and then extend the con-
clusions by analogy to three dimensions.
(a) The Miller indices
Consider a two-dimensional rectangular lattice formed from a
unit cell of sides a, b (as in Fig. 37.10). Each plane in the illus-
tration (except the plane passing through the origin) can be
distinguished by the distances at which it intersects the a and b
axes. One way to label a plane would therefore be to quote the
smallest intersection distances. For example, we could denote a
representative plane of each type in Fig. 37.10 as (a) (1a,1b), (b)
( 12 a, 13 b), (c) (−1a,1b), and (d) (∞a,1b), where ∞ is used to show
that the planes intersect an axis at infinity. However, if we agree
to quote distances along the axes as multiples of the lengths of
the unit cell, then we can label the planes more simply as (1,1),
Atkins09819.indb 333
37 Crystal structure  333
b
a
(a) (b)
(c) (d)
Figure 37.10 Some of the planes that can be drawn
through the points of a rectangular space lattice and their
corresponding Miller indices (hkl): (a) {110}, (b) {230}, (c) { 110},
and (d) {010}.
( 12 , 13 ), (−1,1), and (∞,1), respectively. If the lattice in Fig. 37.10
is the top view of a three-dimensional orthorhombic lattice in
which the unit cell has a length c in the z-direction, all four sets
of planes intersect the z-axis at infinity. Therefore, the full labels
are (1,1,∞), ( 12 , 13 , ∞), (−1,1,∞), and (∞,1,∞).
The presence of fractions and infinity in the labels is incon-
venient. They can be eliminated by taking the reciprocals of the
labels. As we shall see, taking reciprocals turns out to have fur-
ther advantages. The Miller indices, (hkl), are the reciprocals of
intersection distances. To simplify the notation while providing
a great deal of information, the following rules apply:
r
Negative indices are written with a bar over the number,
as in ( 110) .
r
If taking the reciprocal results in a fraction, then the
fraction can be cleared by multiplying through by an
appropriate factor.
For example, a ( 13 , 12 , 0) plane is denoted (2,3,0) after multipli-
cation of all three indices by 6.
r
The notation (hkl) denotes an individual plane. To
specify a set of parallel planes we use the notation {hkl}.
Thus, we speak of the (110) plane in a lattice, and the set of all
{110} planes that lie parallel to the (110) plane.
A helpful feature to remember is that the smaller the abso-
lute value of h in {hkl}, the more nearly parallel the set of planes
is to the a axis (the {h00} planes are an exception). The same
is true of k and the b axis and l and the c axis. When h = 0, the
planes intersect the a axis at infinity, so the {0kl} planes are par-
allel to the a axis. Similarly, the {h0l} planes are parallel to b and
the {hk0} planes are parallel to c.
9/11/2013 11:47:35 AM
 334 8 Interactions

Brief illustration 37.2 Miller indices Justification 37.1 The separation of lattice planes
The {1,1,∞} planes in Fig. 37.10a are the {110} planes in the Consider the {hk0} planes of a square lattice built from a unit
Miller notation. Similarly, the { 12 , 13 , ∞} planes are denoted cell with sides of length a (Fig. 37.12).
{230}. Figure 37.10c shows the {110} planes. The Miller indi-
ces for the four types of plane in Fig. 37.10 are therefore {110}, a
{230}, {110}, and {010}. Figure 37.11 shows a three-dimensional
(hkl)
representation of a selection of planes, including one in a lat-
tice with non-orthogonal axes.

b
a/k φ dhkl
a
c φ
a
a/h
(110) (111)
a Figure 37.12 The dimensions of a unit cell and their relation
to the plane passing through the lattice points.
b
We can write the following trigonometric expressions for the
angle φ shown in the illustration:
c
(100) (111) d hd d kd
sin φ = = hk 0 cos φ = = hk 0
(a / h) a (a / k ) a
Figure 37.11 Some representative planes in three
dimensions and their Miller indices. Note that a 0 indicates Because the lattice planes intersect the horizontal axis h times
that a plane is parallel to the corresponding axis, and that and the vertical axis k times, the length of each hypotenuse
the indexing may also be used for unit cells with non- is calculated by dividing a by h and a by k. Then, because
orthogonal axes. sin2 φ + cos2 φ = 1, it follows that
2 2
Self-test 37.2 Find the Miller indices of the planes that inter- ⎛ hdhk 0 ⎞ ⎛ kdhk 0 ⎞
sect the crystallographic axes at the distances (3a, 2b, c) and ⎜⎝ a ⎟⎠ + ⎜⎝ a ⎟⎠ = 1
(2a, ∞b, ∞c)
2
Answer: {236} and {100} which we can rearrange by dividing both sides by dhk0 into

1 h2 k 2 h2 + k 2
2
= + =
dhk a2 a2 a2
(b) The separation of planes 0

which is eqn 37.1a.

The Miller indices are very useful for expressing the separation
of planes. It is shown in the following Justification that the sepa-
ration of the {hk0} planes in the square lattice in Fig. 37.12 is
given by The corresponding expression for a general orthorhombic lat-
tice (one in which the axes are mutually perpendicular) is the
1 h2 + k 2
2
= or generalization of this expression:
dhk 0 a2
Square lattice Separation of planes (37.1a)
a 1 h2 k 2 l 2 Orthorhombic Separation
dhk 0 = 2 2 1/2 = 2 + 2 + 2 (37.1c)
(h + k ) 2
dhkl a b c lattice of planes

By extension to three dimensions, the separation of the {hkl}

planes of a cubic lattice is given by
Example 37.1 Using the Miller indices
1 h2 + k 2 + l 2
2
= or Calculate the separation of (a) the {123} planes and (b) the
dhkl a2
Separation
a Cubic lattice
of planes (37.1b) {246} planes of an orthorhombic unit cell with a = 0.82 nm,
dhkl = 2 b = 0.94 nm, and c = 0.75 nm.
(h + k 2 + l 2 )1/2

www.ebook3000.com

Atkins09819.indb 334 9/11/2013 11:47:49 AM