You are on page 1of 19

32 Polymer Elasticity

& Collapse

Polymeric Materials Are Often Elastic


Rubber is elastic, and gooey liquids such as raw eggs are viscoelastic, because
polymer molecules have many conformations of nearly equal energy. Stretch-
ing the polymer chains lowers their conformational entropy. The chain retract
to regain entropy. The entropy is also lowered when polymers become com-
pact, as when proteins fold or when DNA becomes encapsulated within virus
heads. The simplest molecular description of polymer elasticity is the random-
flight model.

Polymer Chains Can Be Modelled as Random Flights

We consider a polymer chain to be a connected sequence of N rigid vectors,


each of length b. For now, each vector represents a chemical bond, but starting
on page 612 we consider other situations in which each vector can represent a
virtual bond, a collection of more than one chemical bond.
There are different ways to characterize the size of a polymer chain. The
contour length L is the total stretched-out length of the chain,

L = Nb (32.1)

(see Figure 32.1(a)). The contour length has a fixed value, no matter what the
chain conformation.

609
(a) Contour Length L (b) End-to-end Length r
b b
b b N
r
b 2
b
1
3 4
L = Nb

Figure 32.1 Two measures of the size of a polymer molecule. If there are N bonds
and each bond has length b, the contour length (a) is the sum of the lengths of all
N
bonds, L = Nb. The end-to-end length (b) is the vector sum, r = i=1  i , where  i is
the vector representing bond i.

Another measure of polymer chain size is the end-to-end length, the length
of the vector pointing through space from one end of the chain to the other,
when the chain is in a given conformation (see Figure 32.1(b)). The end-to-end
length varies from one chain conformation to the next. The end-to-end vector
r is the vector sum over the bonds  i for bonds i = 1, 2, 3, . . . , N:

N
r= i. (32.2)
i=1

The end-to-end length cannot be greater than the contour length. The end-
to-end length changes when a polymer is subjected to applied forces. It is
related to the radius of the molecule, and can be used to interpret the vis-
cosities of polymer solutions, the scattering of light, and some of the dynamic
properties of polymer solutions.
For any vector, you can always consider the x, y, and z components indi-
vidually. The x component of the end-to-end vector is

N
rx = xi , (32.3)
i=1

where xi = b cos θi is the projection on the x-axis of the ith vector and θi is
the angle of bond i relative to the x-axis. If all the bond vectors have the same
length b, you get

N
rx = b cos θi . (32.4)
i=1

In solution, there are many molecules, each with a different conformation.


The physical properties that can be measured by experiments are averages
over all the possible conformations. To get these averages, you can use a sim-
ple model, called the freely jointed chain, or random-flight model (‘flight’ in
three dimensions, or ‘walk’ in two dimensions). In this model the angle of
each bond is independent of every other, including the nearest neighboring
bonds. Assuming that all conformations have equal probabilities, the average
x-component of the end-to-end vector is

610 Chapter 32. Polymer Elasticity & Collapse


3 4

N
rx  = b cos θi =b cos θi  = 0, (32.5)
i=1 i=1

because the average over randomly oriented vectors is cos θi  = 0 (see Exam-
ple 1.23). This means that monomer N of a random flight chain is in about
the same location as monomer 1, because there are as many positive steps as
negative steps, on average. Because the x, y, and z components are indepen-
dent of each other, the y and z components can be treated identically and
ry  = rz  = 0.
Because the mean value of the end-to-end vector is zero (rx  = 0), rx 
doesn’t contain useful information about the size of the molecule. The mean
square end-to-end length r 2  or rx2  is a more useful measure of molecular
size. It is related to the radius of gyration Rg by Rg2 = r 2 /6 for a linear chain
[1]. We won’t prove that relationship here. The radius of gyration, which is
always positive, is a measure of the radial dispersion of the monomers. We use
r 2  instead of Rg because the math is simpler.
The square of the magnitude of the end-to-end vector can be expressed as
a matrix of terms:

11 + 12 + 13 + · · · + 1N +


 2

N21 + 22 + 23 + · · · + 2N +


r·r= i = .. .. (32.6)
i=1 . .
N 1 + N 2 + N 3 + · · · + N N .

This sum involves two kinds of terms, ‘self terms’  i  i (along the main
diagonal) and ‘cross terms’  i  j , i = j. The self terms give  i ·  i  = b2 ,
because the product of a vector with itself is the square of its length. The cross
terms give  i ·  j  = b2 cos θ = 0 because there is no correlation between
the angles of any two bond vectors in the random-flight model. Therefore the
only terms that contribute to the sum are the self terms along the diagonal,
and there are N of those, so

r 2  = Nb2 . (32.7)

The root-mean-square (rms) end-to-end distance is r 2 1/2 = N 1/2 b. Because


the molecular weight of a polymer is proportional to the number of monomers
it contains, one important prediction of the random-flight theory is that the
mean radius (and the rms end-to-end length) of a polymer chain increases in
proportion to the square root of the molecular weight. This result is the cen-
terpiece of much of polymer theory.
How do real chains differ from random flights? An important difference is
chain stiffness.

Chain Stiffness

Contrast the mathematical description of a random flight with that of a rod,


for which all vectors point in the same direction. For a rod, all the terms in
Equation (32.6) contribute b2 to the sum, and there are N 2 terms in the matrix,
so the end-to-end length equals the contour length, and r 2  = (Nb)2 = L2 .

Polymeric Materials Are Often Elastic 611


Because r 2  ∼ N when bonds are uncorrelated, and r 2  ∼ N 2 when
bonds are perfectly correlated, you might expect that partial correlation be-
b tween bonds, which is a more realistic model for polymers, would lead to a
ψ dependence of r 2  on N with some exponent between 1 and 2. But this is not
the case.
bK To illustrate, consider a model in which bonds have a weak angular corre-
lation between first neighbors,  i ·  i+1  = b2 γ, where γ < 1 is a positive
Figure 32.2 The Kuhn constant. There are no correlations beyond first neighbors. If N is large, Equa-
model of a polymer. b is tion (32.6) contains only terms along the main diagonal and the two adjacent
the length of the chemical diagonals. So for large N, you have r 2  ≈ Nb2 + 2Nb2 γ = Nb2 (1 + 2γ).
bond, bK is the length of a Angular correlations between neighboring bonds do not change the scaling,
virtual, or Kuhn, bond r  ∼ N. Neighbor correlations only change a multiplicative constant. This
2

directed along the chain conclusion also holds if you include second neighbors and third neighbors,
axis, and ψ is the angle etc., provided only that the correlations decay to zero over a number of bonds
between the chain axis and |i − j|  N that is much smaller than the chain length. So the random-flight
the chemical bond. model is useful even for real chains, for which bonds are not fully independent
of their neighbors. Chain stiffness is a term sometimes used to describe the
degree of angular correlation between neighboring bonds.
In general, the effects of bond correlations can be described by

r 2  = CN Nb2 , (32.8)

where the constant CN is called the characteristic ratio. The subscript indicates
that CN can depend on the chain length N. The characteristic ratios of poly-
mers are typically greater than one, indicating that there are correlations of
orientations between near-neighbor bonds along the chain.
These arguments suggest a strategy for treating real chains. The idea is
to represent a real chain by an equivalent freely jointed chain. The equivalent
chain has virtual bonds, or Kuhn segments, each of which represents more than
one real chemical bond. The number of virtual bonds NK and the length of each
bond bK are determined by two requirements. First, the Kuhn model chain
must have the same value of r 2  as the real chain, but it is freely jointed so
its characteristic ratio equals one,
2
r 2  = CN Nb2 = NK bK . (32.9)

Second, the Kuhn chain has the same contour length as the real chain,

L = Nb cos ψ = NK bK , (32.10)

where ψ is the angle between the real bond and the long axis of the extended
real chain (see Figure 32.2), since that axis defines the direction of the virtual
bonds. Dividing Equation (32.9) by (32.10) gives
bK CN
= (32.11)
b cos ψ
so

NK cos2 ψ
= . (32.12)
N CN
Example 32.1 gives the Kuhn length for polyethylene.

612 Chapter 32. Polymer Elasticity & Collapse


EXAMPLE 32.1 The Kuhn model of polyethylene. Polyethylene has a
measured characteristic ratio of 6.7. For tetrahedral valence angles, ψ =
70.5◦ /2 = 35.25◦ [1]. Equations (32.11) and (32.12) give the Kuhn length
as bK /b = 6.7/ cos 35.25◦ ≈ 8 times the chemical bond length, and there are
N/NK = 6.7/ cos2 (35.25◦ ) ≈ 10 chemical bonds per virtual bond.

Another model is the wormlike chain model [2]. Its measure of stiffness is
the persistence length, which is half the Kuhn length. The persistence and Kuhn
lengths characterize the number of bonds over which orientational correlations Figure 32.3 Example of a
decay to zero along the chain. These are simplified models of bond correlations. random-walk chain
A better treatment, which we won’t explore here, is the rotational isomeric state conformation. Source: LRG
model [1, 3], which accounts not only for the angles and lengths of chemical Treloar, Physics of Rubber
bonds, but also for the different statistical weights of the various possible bond Elasticity, 2nd edition,
conformations. Oxford University Press,
To understand why polymeric materials are elastic, you need to know more New York, 1958.
than just the first and second moments, r  = 0 and r 2  = Nb2 ; you need the
full distribution function of the end-to-end lengths.

Random-flight Chain Conformations Are Described


by the Gaussian Distribution Function
To get the distribution of end-to-end lengths of a polymer chain, we use the
random-flight model that predicted the diffusion of a particle in Chapter 4
(pages 57 to 59). In diffusion, a particle moves a distance b in a random direc-
tion at each time step. For a polymer, imagine laying down a bond at a time,
from bond i = 1 to N, so the growing end of a random-flight chain moves a
distance b in a random direction as each bond vector is added to the chain.
Adding each randomly oriented bond vector to the growing chain corresponds
to a particle moving one time step in the diffusion model. Figure 32.3 shows
how the conformation of a random-flight polymer chain looks like the path of
a diffusing particle.
To be more quantitative, we first consider a one-dimensional random walk,
then we generalize to three dimensions. Each bond vector, because it is ori-
ented randomly, can have a different projection onto the x-axis. We first com-
pute the distribution of the number of forward and reverse steps, then we
compute the average x-axis distance travelled per step.
We want the probability, P (m, N) that the chain takes m steps in the +x
direction out of N total steps, giving N − m steps in the −x direction. Just like
the probability of getting m heads out of N coin flips, this probability is given
approximately by the Gaussian distribution function, Equation (4.34),
∗ )2 /N
P (m, N) = P ∗ e−2(m−m ,

where m∗ is the most probable number of steps in the +x direction.


Now we convert from the number of steps m to coordinate position x. The
net forward progress x is a product of two factors: (1) the number of forward
minus reverse steps, which equals m−(N −m) = 2m−N; (2) the average x-axis
distance travelled per step. The average x-axis projection of the bond vectors
is rx  = 0 (see Equation (32.5)) because of the symmetry between forward and

Random-flight Chain 613


reverse steps. However, because we want the average step length, without √its
sign, we use the rms x-axis projection: b2 cos2 θ1/2 = bcos2 θ1/2 = b/ 3
(see Equation (1.45)). √
The product of these two factors is x = (2m − N)b/ 3. Since√the average
number of forward steps is m∗ = N/2, you have x = 2(m−m∗ )b/ 3. Squaring
both sides and rearranging gives

(m − m∗ )2 = 3x 2 /4b2 .

Substituting this expression into Equation (4.34) gives

P (x, N) = P ∗ e−3x
2 /2Nb 2
. (32.13)

You can find the normalization constant P ∗ by integrating:



P ∗ e−3x /2Nb dx = 1.
2 2

−∞

This is solved by using the integral (Appendix D, Equation (D.1))


∞ 5
−βx 2 π
e dx = , where β = 3/(2Nb2 ). (32.14)
−∞ β

Equation (32.13) becomes


 1/2  1/2
β −βx 2 3
e−3x /2Nb .
2 2
P (x, N) = e = (32.15)
π 2π Nb2
Equation (32.15) gives you the relative numbers of all N-mer chain conforma-
tions that begin at the origin of the x-axis and end at x. The Gaussian distribu-
tion has a peak at x = 0, implying that most of the chains have about as many
+ steps as − steps, as we noted in Chapter 4.
Relatively few chains are highly stretched. If you want to convert from
the fractions of chain conformations to total numbers of chain conformations,
you can multiply by the approximate total number of chain conformations zN ,
where z is the number of rotational isomers per bond, to get zN P (x, N). Now
we show that the Gaussian model predicts that rubber and other polymeric
materials are highly deformable and elastic.

Polymer Elasticity Follows Hooke’s Law


When a polymer is stretched and then released, it retracts like a Hooke’s law
spring. The retractive force is proportional to the extension. The retractive
force is entropic.
Suppose you pull the two ends of a polymer chain apart along the x-direction.
Because all the conformations of random-flight chains have the same energy,
the free energy is purely entropic, F = −T S. Substituting Equation (32.15)
into the expression S/k = ln P (x, N) gives, for the entropy and free energy of
stretching,

F S 3x 2
= − = − ln P (x, N) = βx 2 + constant = + constant. (32.16)
kT k 2Nb2

614 Chapter 32. Polymer Elasticity & Collapse


Force (pN)

Force (pN)

20
0.4

0.2

10

0
0 10 20 26 30
Extension (µm)

0
0 10 20 30
Extension (µm)

Figure 32.4 Stretching a single molecule of DNA leads to a retractive force that is
linear in the extension at low extensions, but steeper at higher extensions. The inset
is an enlargement of the y-axis, for small deformations. Source: SB Smith, L Finzi and
C Bustamante, Science 258, 1122–1126 (1992).

The retractive force felastic is defined as the derivative of the free energy,
dF 3kT x
felastic = − = −2kT βx = − . (32.17)
dx Nb2
Equation (32.17) shows that the retractive force felastic is linear in the displace-
ment x, like a Hooke’s law spring. Polymeric materials can be stretched to many
times their undeformed size. (Metal can be stretched too, but only to a much
smaller extent, and by a different, energetic, mechanism.) Fully stretched, the
size of a chain is determined by its contour length, L = Nb. In its undeformed
state, its average size√is r 2 1/2 = N 1/2 b, so a polymer chain can be stretched
by nearly a factor of N. f
N 1
Figure 32.4 shows how the retractive force depends on the stretched lengths
of single molecules of DNA. You can see the linearity between force and exten-
Figure 32.5 Chain
sion for small stretching. At large deformations, the retractive forces become
conformational elasticity is a
much stronger and the Gaussian distribution no longer applies.
vector force tending to
Now we generalize to three dimensions to describe elasticity as a vector
cause the chain ends
force that drives the chain ends together (see Figure 32.5).
(labeled N and 1) to be near
each other, on average.
Elasticity in Two and Three Dimensions

Let r represent the vector from the origin, where the chain begins, to (x, y, z),
where the chain ends (see Figure 32.6(a)). The length of the end-to-end vector

Polymer Elasticity Follows Hooke’s Law 615


(a) (b)
P ( r, N ) = P ( x, y, z, N ) P ( r, N ) = 4 πr P ( r, N )

z
z
dx dy dz

r
x

r
x y
dr
y

r r

Figure 32.6 Two different probability densities for the termini of chains that begin
at the origin (0, 0, 0). (a) The probability density P (r, N) = P (x, y, z, N) that the chain
end is between (x, y, z) and (x + dx, y + dy, z + dz). The most probable
termination point is the origin. (b) The probability density P (r , N) = 4π r 2 P (r, N) that
the chain end is anywhere in a radial shell between r and r + dr . The peak of this
distribution is predicted by Equation (32.23). Source: CR Cantor and PR Schimmel,
Biophysical Chemistry, Part II, WH Freeman, San Francisco, 1980.

is r , where r 2 = x 2 + y 2 + z2 . Because each bond is oriented randomly, the x,


y, and z projections of a random-flight chain are independent of each other.
So the probability density of finding the chain end in a small volume element
between (x, y, z) and (x + dx, y + dy, z + dz) is the product of independent
factors:

P (x, y, z, N) dx dy dz = P (x, N)P (y, N)P (z, N) dx dy dz


 3/2
β
e−β(x +y +z ) dx dy dz.
2 2 2
= (32.18)
π
In terms of the vector r, you have
 3/2
β
e−βr .
2
P (r, N) = P (x, y, z, N) = (32.19)
π
Equations (32.18) and (32.19) give the probability that the chain terminus is
located at a specific location r between (x, y, z) and (x + dx, y + dy, z + dz),
if the chain originates at (0, 0, 0) (see Figure 32.6(a)). However, sometimes
this is not exactly the quantity you want. Instead you may want to know the
probability that the chain terminates anywhere in space at a distance r from
the origin (see Figure 32.6(b)). We denote this quantity P (r , N), without the
bold r notation. Because the number of elements having a volume dx dy dz
grows with distance r from the origin as 4π r 2 , the probability of finding the

616 Chapter 32. Polymer Elasticity & Collapse


Number of Molecules Figure 32.7 Distribution of end-to-end
distances r of T3 DNA adsorbed on
200
cytochrome c film (in microns, µm).
The line represents the theoretical
distribution for two-dimensional
100 random walks. Source: D Lang,
H Bujard, B Wolff, and D Russell, J Mol
Biol 23, 163–181 (1967).
0
0 2 4 6
r (µm)

two ends separated by a distance r in any direction is P (r , N) = 4π r 2 P (r, N).


The normalization is given by
∞ ∞ ∞ ∞
P (x, y, z, N) dx dy dz = 4π r 2 P (r, N) dr = 1. (32.20)
−∞ −∞ −∞ 0

The function P (r , N) has a peak because it is a product of two quantities: the


probability P (r, N) diminishes monotonically with distance from the origin, but
the number of volume elements, 4π r 2 , increases with distance from the origin.
Figure 32.7 shows that the two-dimensional end-to-end distance distribution of
T3 DNA molecules adsorbed on surfaces and counted in electron micrographs
is well predicted by two-dimensional random-walk theory.
Example 32.2 shows how the Gaussian distribution function is used for pre-
dicting polymer cyclization equilibria and kinetics.
Figure 32.8 For polymer
cyclization, the two chain
EXAMPLE 32.2 Polymer cyclization (Jacobson–Stockmayer theory). What
ends must be close
is the probability that the two ends of a polymer chain come close enough
together.
together for them to react with each other? This probability is useful for cal-
culating the rate and equilibrium constants for cyclization processes in which
linear chains form circles (see Figure 32.8). Here we follow the random-flight
model of H Jacobson and WH Stockmayer [4].
Suppose that the polymer has N bonds. To determine the probability that
the chain end is within a bond distance b of the chain beginning, integrate
P (r , N), the probability of finding the ends a distance r apart, from 0 to b:
b
Pcyclization = P (r , N) dr
0
 3/2 b
3
e−3r
2 /2Nb 2
= 4π r 2 dr . (32.21)
2π Nb2 0

When the ends are together, r = b is small (r 2 /Nb2  1), so e−3r


2 /2Nb 2
≈ 1.
b
Then the integral in Equation (32.21) is 0 4π r 2 dr = 4π b3 /3, and
 3/2    1/2
3 4 6
Pcyclization = π b3 = N −3/2 .
2π Nb2 3 π
The main prediction of the Jacobson–Stockmayer theory is that the cyclization
probability diminishes with chain length as N −3/2 . The longer the chain, the
smaller is the probability that the two ends are close enough to react. This

Polymer Elasticity Follows Hooke’s Law 617


(a) Concentration of (b) Higher Resolution
Cyclized Long Chains (M) Cyclization Concentrations (M)
10− 7
10−7

10−8 10− 8

10−9
102 103 104 235 245 255
DNA Length (base pairs) DNA Length (base pairs)

Figure 32.9 (a) For long chains the DNA cyclization probability diminishes as
predicted by the Jacobson–Stockmayer theory. Shortening the chain stiffens it,
reducing the cyclization probability below the Jacobson–Stockmayer value. (b) At
higher resolution the cyclization probability depends on periodicities in the polymer,
neglected in the random-flight treatment. Source: D Shore and RL Baldwin, J Mol Biol
170, 957–981 (1983).

Figure 32.10 Experimental molar N


Log K 20 30 40 50 100
cyclization equilibrium constants K (in
mol dm−3 ) for cyclic
−1
[O(CH2 )10 OCO(CH2 )4 CO]n in
poly(decamethylene adipate) melts at
423 K ( ) versus chain length n are
−2
compared with values calculated ( )
from the Jacobson–Stockmayer theory.
Source: JA Semylen, Cyclic Polymers,
−3
Elsevier, London, 1986.
1.2 1.6 2.0
Log N

prediction is confirmed by experiments shown in Figures 32.9 and 32.10. Fig-


ure 32.9 shows that the probability of DNA cyclization diminishes with chain
length for long chains, as predicted by the theory, but you can also see the
effect of chain stiffness, which is not treated by the theory. If chains are too
short, their stiffness prevents them from bending enough to cyclize, so their
cyclization probability is smaller than is predicted by the random-flight model.
Also, chemical details can matter for short chains: the ends must be oriented
correctly with respect to each other to cyclize.

Example 32.3 shows how to find the most probable radius R0 .

EXAMPLE 32.3 Most probable radius R0 . What is the most probable end-to-
end distance R = R0 of a Gaussian chain? Because this is related to the size
of the molecule, this quantity is often called the most probable radius. Begin
with the entropy, S(r ) = k ln P (r , N). Because F = −T S, using Equation (32.19)
gives

618 Chapter 32. Polymer Elasticity & Collapse


Figure 32.11 An affine deformation of a material from
dimensions (x0 , y0 , z0 ) to (λx x0 , λy y0 , λz z0 ).

λ yy 0 λ xx 0

λ zz 0

z0

y0 x0

F S 
= − = − ln 4π r 2 P (r, N) = βr 2 − 2 ln r + constant. (32.22)
kT k
To get the equilibrium value r = R0 , find the minimum free energy:
 
d F 2
= 2βR0 − =0 (32.23)
dr kT r =R0 R0
1 2Nb2
⇒ R02 = = .
β 3

The quantity R02 = 2Nb2 /3 = (2/3)r 2  is the square of the most probable ra-
dius of the Gaussian chain. We use it starting on page 621 for treating polymer
collapse and folding processes.

Now we use the theory of polymer chains to describe elastic materials.

The Elasticity of Rubbery Materials


Results from the Sum of Chain Entropies
Rubber and other polymeric materials are elastic. Polymeric elastomers are co-
valently cross-linked networks of polymer chains. Here we describe one of the
simplest and earliest models for the retractive forces of polymeric materials,
the affine network model.
Suppose you have m chains, all of the same length N. The bond length
is b. The chain ends are covalently cross-linked at junction points. Assume
there are no intermolecular interactions, and that the total elastic free energy
of the material is the sum of the elastic free energies of each of the chains.
For the undeformed material, indicated by subscript 0, the end of a chain is at
(x0 , y0 , z0 ) and

r02 = x02 + y02 + z02 = Nb2 . (32.24)

If the material is deformed by a factor λx in the x-direction, λy in the y-


direction, and λz in the z-direction (see Figure 32.11), so that x = λx x0 , y =
λy y0 , and z = λz z0 , the chain end is moved to (x, y, z),

r 2 = x 2 + y 2 + z2 = λ2x x02 + λ2y y02 + λ2z z02 . (32.25)

Elasticity of Rubbery Materials 619


Using F = −kT ln P (x, y, z, N), β = 3/(2Nb2 ), and Equation (32.19), you have
the free energy for deforming a single chain, F1 :
P (x, y, z, N)
∆F1 = Fdeformed − Fundeformed = −kT ln = kT β(r 2 − r02 ).
P (x0 , y0 , z0 , N)
Summing over m independent chains gives the total free energy Fm :

∆Fm = kT β (r 2 − r02 ). (32.26)


m

Assuming that all chains are equivalent, you have m (r
2
− r02 ) = m[r 2  −
r02 ], and Equation (32.26) becomes

∆Fm = kT βm(r 2  − r02 )


 
= kT βm (λ2x − 1)x02  + (λ2y − 1)y02  + (λ2z − 1)z02  . (32.27)

If the undeformed chains are isotropic (all the directions are equivalent), then
x02  = y02  = z02  = (Nb2 )/3, and Equation (32.27) becomes
∆Fm m 2
= (λ + λ2y + λ2z − 3). (32.28)
kT 2 x
In a macroscopic material, applying a force in one direction can cause forces
and deformations in other directions. The forces per unit area in various di-
rections are called stresses and the relative displacements are called strains.
Stresses are derivatives of the free energy with respect to strains. If you deform
a material by a factor α, where the x-axis, y- axis, and z-axis deformations de-
pend on α, the free energy is a function F (λx (α), λy (α), λz (α)). The derivative
can be written in terms of λ2x , λ2y , and λ2z as
       
∂F ∂F dλ2x ∂F dλ2y ∂F dλ2z
= + + . (32.29)
∂α ∂λ2x dα ∂λ2y dα ∂λ2z dα

Suppose that you stretch an elastomer along the x-direction by a factor


α = x/x0 . Then the x-direction force is
     
∂∆Fm 1 ∂∆Fm α ∂∆Fm
fx = − = =− . (32.30)
∂x x0 ∂α x ∂α
The stress τ acting in the opposite direction equals the force given by Equation
(32.30) divided by the deformed cross-sectional area yz,
 
fx α ∂∆Fm
τ=− = , (32.31)
yz V ∂α
where V = xyz is the final volume of the material [5, 6]. Examples 32.4 and
32.5 are applications of this elasticity model.

EXAMPLE 32.4 Stretch a rubber band along the x-axis. If you stretch rub-
ber, its volume remains approximately constant. Therefore, stretching along
the x-direction by λx = α leads to
1
λy = λz = √ . (32.32)
α

620 Chapter 32. Polymer Elasticity & Collapse


Then Equation (32.28) becomes τ (kg cm−2)
 
∆Fm m 2 2
= α + −3 . 60
kT 2 α
Because ∂∆Fm /∂λ2x = ∂∆Fm /∂λ2y = ∂∆Fm /∂λ2z = mkT /2, Equation (32.28)
gives
  40
∂Fm 1
= mkT α − 2 . (32.33)
∂α α
Substituting Equation (32.33) into Equation (32.31) gives
  20
mkT 1
τ= α2 − . (32.34)
V α
Figure 32.12 shows that the model predicts experimental data adequately at
extensions below about α = 3–5, but that the chains become harder to stretch 0
than the model predicts at higher extensions. 2 4 6 8
α

Figure 32.12 Stretching a


EXAMPLE 32.5 Stretch a rubber sheet biaxially. Stretch a rubber sheet rubber string gives stress τ
along the x-axis by an amount λx = α1 and along the y-axis by an amount versus elongation α, ( )
λy = α2 , where α1 is independent of α2 . If the volume is constant, λz = experimental data; ( )
1/(α1 α2 ). Then the stress along the x-axis is curve predicted by
    Equation (32.34). Source:
mkT dλ2x dλ2z mkT 2 1
τx = α1 + = α1 − 2 2 . (32.35) P Munk, Introduction to
2V dα1 dα1 V α1 α2
Macromolecular Science,
Wiley, New York, 1989. Data
The retractive stress in elastomers depends not only on the deformation but are from LRG Treloar, Trans
also on the cross-link density, through m/V , the density of chains. In a unit Faraday Soc 40, 59 (1944).
volume there are 2m total chain ends. If each junction is an intersection of j
chain ends, then there will be (1 junction/j chain ends) × (2m chain ends) =
(2m/j) junctions, so the number of junctions is proportional to m. Therefore
the retractive force increases linearly with the cross-link density of the network.
Bowling balls are made of a type of rubber that has a much higher cross-link
density than rubber bands.
The main advance embodied in theories of polymer elasticity was the recog-
nition that the origin of the force is due mainly to the conformational freedom
of the chains, and is mainly entropic, not energetic. The conformational en-
tropies of polymers are important not only for stretching processes. They also
oppose the contraction of a polymer chain to a radius smaller than its equilib-
rium value.

Polymers Expand in Good Solvents,


Are Random Flights in Theta Solvents,
and Collapse In Poor Solvents
According to the random-flight theory, the ‘size’ of a molecule increases in
proportion to N 1/2 , where size means either the average end-to-end distance
r 2  = Nb2 or the most probable radius R02 = 2Nb2 /3. We showed on pages

Polymers Expand 621


611–613 that near-neighbor bond angle correlations (called local interactions)
do not change this scaling relationship. But now we show that solvent (non-
local) interactions can change this relationship because different solvents or
temperatures cause the chain to swell or contract.
Random-flight behavior, which is dominated by local interactions, applies
only to a limited class of solvent and temperature conditions, called θ-condi-
tions, θ-solvents, or θ-temperatures. In contrast, in a good solvent, monomer–
solvent interactions are more favorable than are monomer–monomer interac-
tions. Chains expand in good solvents. The radius of a polymer molecule grows
more steeply with N than is predicted by the random-flight model, mainly be-
cause of the self-avoidance of the segments of the chain. In a third class of con-
ditions, called poor solvents, the chain monomers are attracted to each other
more strongly than they are attracted to the solvent. Poor solvents cause an
isolated chain to collapse into a compact globule, with radius ∼ N 1/3 . Poor sol-
vents can also cause multiple chains to aggregate with each other. For example,
oil chains such as polymethylene phase-separate from water because water is
a poor solvent. Here is the simplest model of the expansion and collapse of a
single chain, due to PJ Flory [7, 8].
How does the conformational free energy depend on the radius? Consider
a chain with N monomers. As in Equation (32.23), our strategy is to find the
radius r = R that maximizes the entropy, or minimizes the free energy. How-
ever, now we include an additional contribution to the free energy. In addition
to the elastic free energy Felastic , we now also account for the solvation free
energy Fsolvation , using the Flory–Huggins theory of Chapter 31,
d
(Felastic (r ) + Fsolvation (r ))r =R = 0. (32.36)
dr
The solvation free energy depends on the radius through the mean chain
segment density, ρ = N/M. M is the number of sites of a lattice that contains
the chain. We define M in terms of the chain radius below. Low density, ρ → 0,
means the chain is expanded, while high density, ρ → 1, means the chain is
compact.
Chain collapse is opposed by the conformational entropy but is driven by the
gain in monomer–monomer contacts, which are favorable under poor-solvent
conditions. To get the conformational entropy for Fsolvation , you can begin with
Equation (31.4), which gives ν1 , the number of conformations of a single chain.
But we make two changes. First, we leave out the factor (z − 1)N−1 because it is
a constant that doesn’t change with the density ρ, and leaving it out simplifies
the math. (The independence of the conformational entropy from the factor
(z − 1)N−1 in the Flory model has the important implication that chain stiffness
and local interactions do not contribute to entropies of collapse or expansion.)
Second, our focus on a single chain means that its center-of-mass position in
space is irrelevant. This is equivalent to neglecting the placement of the first
monomer, so we replace the factor M −(N−1) with M −N and the conformational
entropy of the chain is
S M!
= ln ν1 = ln . (32.37)
k (M − N)!M N
Multipy the numerator and denominator of Equation (32.37) by N! to put part

622 Chapter 32. Polymer Elasticity & Collapse


Figure 32.13 The contact free energy ∆g describes
desolvating two polymer chain segments P1 and P2 and
P1 + P2 P1 P2 bringing them into contact. The interaction parameter χ
is defined for the process with the opposite sign (see
Figure 15.6).

of this expression into a more familiar form,


   
S M! N!
= ln + ln
k (M − N)!N! MN
     
N N N
= −N ln − (M − N) ln 1 − + N ln − N ln e
M M M
 
N
= −(M − N) ln 1 − − N, (32.38)
M
by using Stirling’s approximation N!/M N ≈ (N/eM)N = (N/M)N e−N . Now
divide by N to get the entropy per monomer, and express the result in terms
of the average density ρ = N/M,
 
S 1−ρ
=− ln(1 − ρ) − 1. (32.39)
Nk ρ

You can check that S → 0 as ρ → 0 and S → − Nk as ρ → 1.

EXAMPLE 32.6 Polymer collapse entropy. When a protein folds, it col-


lapses to a nearly maximally compact state. What is the entropic component
of the free energy opposing collapse at T = 300 K? If the chain length is N = 100
monomers, the equations below Equation (32.39) give

∆Fcollapse = −T (Scompact − Sopen ) = NkT


≈ (100)(1.987 cal mol−1 K−1 )(300 K) ≈ 60 kcal mol−1

Next, we determine how the contact energy depends on the chain radius,
or density. Let wpp , wss , and wsp represent the energies of a contact between
two polymer segments, between two solvent molecules, and between a sol-
vent molecule and a polymer segment, respectively. Figure 32.13 and Equa-
tion (15.11) indicate that the free energy ∆g for desolvating two polymer seg-
ments and forming a contact is
 
wss + wpp 2
∆g = −2 wsp − = − χkT . (32.40)
2 z
The minus sign indicates that the process defining ∆g is the reverse of the
process defining χ (see Figure 15.6). The mean-field approximation gives an
estimate of the number of contacts among chain monomers. The probability
that a site adjacent to a polymer segment contains another polymer segment is
ρ = N/M, and there are z sites that are neighbors of each of the N monomers,
so the contact energy U relative to the fully solvated chain is
Nρz
U= ∆g = −NkT ρχ. (32.41)
2

Polymers Expand 623


The factor of 1/2 corrects for the double counting of interactions. Combining
the entropy Equation (32.39) with the energy Equation (32.41) gives the confor-
mational free energy of the chain as a function of its compactness,
 
Fsolvation U S 1−ρ
= − = ln(1 − ρ) + 1 − ρχ. (32.42)
NkT NkT Nk ρ

When the chain is relatively open and solvated, the density is small, ρ  1.
Then you can use the approximation ln(1 − ρ) ≈ −ρ − (1/2)ρ 2 − . . . to get
    
1−ρ 1−ρ ρ2
ln(1 − ρ) + 1 ≈ −ρ − − ... + 1
ρ ρ 2
1
≈ ρ + ρ2 . (32.43)
2
Substituting Equation (32.43) into Equation (32.42), and keeping only the first-
order approximation, gives the solvation free energy as a function of the aver-
age segment density,
 
Fsolvation 1
= Nρ −χ . (32.44)
kT 2
In Flory theory, a simple approximation relates the density ρ to the size r
N Nv
ρ= = 3 , (32.45)
M r
where v is the volume per chain segment. This relationship defines the value
of M, the number of sites on the lattice that contains the polymer chain. Com-
bining Equation (32.45) with Equations (32.22) and (32.44) gives
 
Felastic Fsolvation N 2v 1
+ = βr 2 − 2 ln r + − χ + constant. (32.46)
kT kT r3 2
Taking the derivative of Equation (32.46) and finding the value r = R that
causes the derivative to be zero (the most probable value of r ; Equation (32.36))
gives
 
2 3N 2 v 1
2βR − − − χ = 0. (32.47)
R R4 2
You can express this in terms of R02 = 2Nb2 /3 = β−1 , the most probable ra-
dius of the unperturbed chain (see Equation (32.23)). Multiplying both sides of
Equation (32.47) by R 4 /(2R03 ), and rearranging gives
 5  3   
R R 3N 2 v 1
− = − χ
R0 R0 2R03 2
 5/2  √
3 v 1
= −χ N. (32.48)
2 b3 2

Theta Solvents Give Random-flight Behavior

For solvent and temperature conditions that cause χ = 0.5, the right-hand side
of Equation (32.48) equals zero. In that case, multiplying Equation (32.48) by

624 Chapter 32. Polymer Elasticity & Collapse


Expansion Factor Figure 32.14 Homopolymers collapse when the
1.8 temperature T is less than the temperature θ at which
χ = 0.5. Chains expand at higher temperatures (χ < 0.5,
good solvents). The transition steepens with increasing
chain lengths. (( ) molecular weight M = 2.9 × 103 ,
( ) M = 1 × 105 , ( ) M = 2.6 × 107 .) Source: ST Sun,
I Hishio, G Swislow and T Tanaka, J Chem Phys 73,
1.0
5971–5975 (1980).

0.2

0.90 1.00 1.10


T/θ

(R0 /R)3 gives the random-flight prediction,


 2
R 2Nb2
=1 ⇒ R 2 = R02 = , (32.49)
R0 3
given by Equation (32.23). That is, when the solvation free energy is zero (the
monomer–monomer attraction just balances the excluded volume), then the
most probable radius of the chain is given simply by the elastic free energy
alone.

Good Solvents Expand Polymers

For solvent and temperature conditions that cause χ < 0.5 (called good sol-
vents), the right-hand side of Equation (32.48) is positive. In this regime, ex-
cluded volume causes chain expansion. For large N, the fifth-power term is
much larger than the third-power term, so
 5  5/2  √
R 3 v 1
≈ − χ N ⇒ R 5 ∝ R05 N 1/2 ∝ N 3
R0 2 b3 2
⇒ R ∝ N 0.6 . (32.50)

In good solvents, the chain radius grows more steeply with chain length (R ∝
N 0.6 ) than in θ-solvents (R ∝ N 0.5 ).

Polymers Collapse in Poor Solvents

Solvent and temperature conditions that cause χ > 0.5 are called poor solvents.
For a poor solvent, the right-hand side of Equation (32.48) is negative, so R/R0 <
1, and the polymer collapses into a compact conformation. For poor solvent
conditions, Equation (32.48) is no longer sufficient because of the low-density
approximation we used to derive it. A collapsed chain has a high segment
density. If you were to keep the next higher term in the density expansion
for ln(1 − ρ), however, you would find that the model predicts R ∝ N 1/3 , as
expected for a compact chain.

Polymers Expand 625


Polymers can undergo very sharp transitions from the coil to compact states
as the solvent and temperature are changed. These are called coil-to-globule
transitions. The reason they are so sharp is found in Equation (32.48). For large
N, the right-hand side of Equation (32.48) can change abruptly from being very
positive to being very negative with only a small change in χ at about χ = 0.5.
Figure 32.14 shows the collapse process in homopolymers.
Two natural collapse processes are the folding of proteins into their com-
pact native states in water, and the compaction of DNA molecules for insertion
into virus heads and cell nuclei. While this homopolymer collapse model illus-
trates the principle of coil-to-globule transitions, neither protein folding nor
DNA collapse follow it exactly, because both polymers also have electrostatic
interactions and specific monomer sequences.

Summary
Polymers have many conformations of nearly equal energy, so they have broad
distributions of conformations. Stretching or squeezing or otherwise perturb-
ing polymers away from their equilibrium conformations leads to entropic
forces that oppose the perturbations. This is the basis for rubber elasticity.
One of the most important models is the random-flight theory, in which the
distribution of chain end distances is Gaussian. The rms distance between
the two ends of the chain increases as the square root of the chain length N,
r 2 1/2 = (Nb2 )1/2 . The random-flight theory applies when the chain is in a
θ-solvent. In that case, the steric tendency to expand is just balanced by the
self-attraction energy causing the chain to contract. When the solvent is poor,
the self-attractions between the chain monomers dominate and chains collapse
to compact configurations. When the solvent is good, the self-attractions are
weak and chains expand more than would be predicted by the random-flight
theory.

626 Chapter 32. Polymer Elasticity & Collapse


Problems [2] JE Mark ed. Physical Properties of Polymers Hand-
book, Chapter 5, American Institute of Physics,
1. Stretching a rubber sheet. What is the free energy Woodbury, 1996.
for stretching an elastomeric material uniformly along the
x and y directions, at constant volume? [3] WL Mattice and UW Suter. Conformational Theory of
Large Molecules: the Rotational Isomeric State Model
2. Stretching DNA. Figure 32.4 shows that it takes
in Macromolecular Systems, Wiley, New York, 1994.
about 0.1 pN of force to stretch a DNA molecule to an
extension of 20 µm. Use the chain elasticity theory to es- [4] H Jacobson and WH Stockmayer. J Chem Phys 18,
timate the undeformed size of of the molecule, r 2 1/2 . 1600 (1950).
3. Stretched polymers have negative thermal expansion
[5] B Erman and JE Mark. Ann Rev Phys Chem 40, 351–
coefficients. The thermal expansion coefficient of a ma-
374 (1989).
terial is α = (1/V )(∂V /∂T )p . Consider the corresponding
one-dimensional quantity αp = (1/x)(∂x/∂T )f for a sin- [6] JE Mark and B Erman. Rubberlike Elasticity: a Molec-
gle polymer molecule stretched to an end-to-end length x ular Primer, Wiley, New York, 1988.
by a stretching force f .
(a) Compute αp for the polymer chain. [7] PJ Flory. Principles of Polymer Chemistry, Series ti-
(b) What are the similarities and differences between a tle: George Fisher Baker Nonresident Lectureship in
polymer and an ideal gas? Chemistry at Cornell University, Cornell University
Press, Ithaca, 1953.
4. An ‘ideal’ solvent expands a polymer chain. If a
polymer chain is composed of the same monomer units [8] HS Chan and KA Dill. Ann Rev Biophys and Biophys
as the solvent around it, the system will be ideal in the Chem 20, 447–490 (1991).
sense that the polymer–polymer interactions will be iden-
tical to polymer–solvent interactions, so χ = 0.
(a) Write an expression for the most probable radius R Suggested Reading
for a chain in an ideal solvent.
(b) Show that such a chain is expanded relative to a The classic texts on rotational isomeric state and random-
random-flight chain. flight models:
(c) Describe the difference between an ideal solvent and CR Cantor and PR Schimmel, Biophysical Chemistry, Vol
a θ-solvent. III, WH Freeman, San Francisco, 1980.
5. Computing conformational averages. Using the ex- M Doi, Introduction to Polymer Physics, trans by H See, Ox-
pression for the distribution P (r , N) for the end-to-end ford University Press, New York, 1996.
separation of a polymer chain of length N, compute r 2  PJ Flory, Principles of Polymer Chemistry, series title:
and r 4 . George Fisher Baker Nonresident Lectureship in Chem-
istry at Cornell University. Cornell University Press,
6. Contour length of DNA. The double-stranded DNA
Ithaca, 1953.
from bacteriophage λ has a contour length L = 17 ×
10−6 m. Each base pair has bond length b = 3.5 Å. WL Mattice and UW Suter, Conformational Theory of
(a) Compute the number of base pairs in the molecule. Large Molecules: the Rotational Isomeric State Model
(b) Compute the molecular weight of the DNA. in Macromolecular Systems, Wiley, New York, 1994.

7. Using elasticity to compute chain concentrations.


Excellent summaries of rubber elasticity:
Figure 32.12 shows the stress–strain properties of a rub-
ber band. Use the figure and chain elasticity theory to: B Erman and JE Mark, Ann Rev Phys Chem 40, 351–374
(a) Estimate the number of polymer chains in a cubic (1989).
volume 100 Å on each side.
JE Mark and B Erman, Rubberlike Elasticity: a Molecular
(b) If each monomer occupies 100 Å3 , what is the length Primer, Wiley, New York, 1988.
of each chain between junction points?

One of the first models of a polymer collapse, applied to


References DNA:

CB Post and BH Zimm, Biopolymers 18, 1487–1501


[1] PJ Flory. Statistical Mechanics of Chain Molecules, Wi-
(1979).
ley, New York, 1969.

Problems 627

You might also like