Introduction to Polymer Physics.

1 1
The references I have consulted are
the 2012 lectures of Polymer physics
Pankaj Mehta from the 2012 Boulder Summer School.
Notes and videos available here https:
March 8, 2021 //[Link]/2012/
boulder-school-2012-lecture-notes.
We have also used Chapter 1 of De
In these notes, we will introduce the basic ideas of polymer physics – Gennes book, Scaling concepts in Poylmer
with an emphasis on scaling theories and perhaps some hints at RG. Physics as well as Chapter 1 of M. Doi
Introduction to Polymer Physics, this
nice review on Flory theory https:
//[Link]/abs/1308.2414, as well
Introduction these notes from Levitov available at
[Link]
Polymers are just long floppy organic molecules. They are the basic notes/polymers_notes.pdf.

building block of biological organisms and also have important in-

dustrial applications. They occur in many forms and are an intense
field of study, not only in physics but also in chemistry and chemical
lectures. In these lectures, we will just touch on these ideas with an
aim of understanding the basics of protein folding. A distinguishing
feature of polymers is that because they are long and floppy, that
entropic effects play a central role in the physics of polymers.
A polymer molecule is a chain consisting of many elementary
units called monomers. These monomers are attached to each other
by covalent bonds. Generally, there are N monomers in a polymer,
with N  1. This means that polymers behave like thermodynamic
objects (see Figure 1). It will be helpful to understand some basic
scales for the problem of polymers.

• First, entropic effects will be important and we will often ask

about exerting forces on the polymers. For this reason it is helpful
to keep in mind that at room temperature 1pNnm ≈ 4.1k B T.

• To break covalent bonds between monomers, we need 1000K. So

they are essentially never broken by thermal fluctuations

• However, “bending” and non-covalent interactions (electrostatics)

compete with k B T.

• Monomers are typically of order 1 Angstrom or 1nm.

• Polymers are typically composed of N ∼ 10 − 109 monomers with

lengths of 10nm − 1m.

It’s also helpful to look at some examples. Fig. 2 shows polymers of

various kinds that occur in biological systems.
We see that there is a lot of diversity in polymers. What are the
key things we have to pay attention to. Well there are number of
things that will be important. In particular, the things that we will
care about are:
 introduction to polymer physics. 2

Figure 1: A polymer is composed

of many monomers. Figure from
Grossberg Polymer lectures in Boulder
Summer School 2012.

Polymer molecule is a chain:

Examples: polyethylene (a), polysterene
• Polymeric from Greek poly- (b), polyvinyl chloride (c)…
mer s having many parts;
First Known Use: 1866
(Merriam-Webster);
• Polymer molecule consists
of many elementary units,
called monomers;
• Monomers – structural
units connected by covalent
bonds to form polymer;
• N number of monomers in a … and DNA
polymer, degree of
polymerization;
• M=N*mmonomer molecular
mass.

• The first important property is whether the polymer is a homopoly-

mer – composed of a single kind of monomer – or a hetropolymer –
composed of many kind of monomers. Most of the interesting bi-
ological polymers are hetropolymers (DNA with 4 bases, proteins
with 20 amino acids, etc.)

• The second major thing that will be important is how flexible

the polymer is. The more flexible, the more entropic configura-
tions that are available. All polymer bend but the question is how
much?

• Another thing that will be important is if the polymer is charged.

This is because electrostatic interaction compete with entropic
interactions.

• Finally, the basic topology is important. We can have a single

chain, or branching chains, or complicated topologies. We are
focusing almost entirely on single chains here.

The final important thing about physics is that the properties of

polymers are often strongly effected by the solvent in which they are
dissolved. The reason for this is again because the solvent changes
the free energy of the polymers by changing the balance between en-
tropic and energetic effects. In fact, we can characterize a polymer by
its radius of gyration R defined the typical distance between the two
 introduction to polymer physics. 3

Figure 2: Polymers in living systems.

Figure from Grossberg Polymer lectures
in Boulder Summer School 2012.

Polymers in living nature

DNA RNA Proteins Lipids Polysaccha
rides

N Up to 1010 10 to 1000 20 to 1000 5 to 100 gigantic

Nice Bioinformatics, Secondary Proteomics, Bilayers, ??? Someone
physics elastic rod, structure, random/designed liposomes, has to start
models charged rod, annealed heteropolymer,HP, membranes
helix-coil branched, funnels, ratchets,
folding active brushes
Uses

Molecule

ends. In general, this is much smaller than a fully stretched polymer

since polymers like to bend. In fact, we will see that typically this
radius goes like the number of polymers to some power

R ∼ Nν (1)

The solvent, elasticity, and electrostatic interactions can change this

power dramatically from ν = 1 for repulsive polymers to ν = 1/3 in
poor solvents. This is summarized in Fig. 3.
We can also directly measure this in experiment using X-ray crys-
tallography in the small k limit.

Neutral Flexible Polymers

We will start with “ideal” polymers. These are neural, flexible poly-
mers that serve as an important starting point for understanding
polymer physics. Ideal polymers ignore all interactions between
monomers, except between neighboring monomers. Conceptually,
they play the same role as an ideal gas for understanding the statisti-
cal mechanics of gases.
 introduction to polymer physics. 4

Figure 3: (Top) Polymer sizes in dif-

Polymer Size ferent solvents. (Bottom) Analogy to
Monomer size b~0.1nm; Number of monomers N~102-1010; understand dramatic change in sizes.
Contour length L~10nm – 1m;
Depending on how much polymer is bent, Figure from Grossberg Polymer lectures
its overall size R varies widely and depends on solvent quality
in Boulder Summer School 2012.
Long-range repulsion Good solvent -solvent Poor solvent

R~1m R~100mm R~10mm R~100nm

Astronomical Variations of Polymer Size

Increase monomer size by a factor of 108: b ~ 1cm; let N=1010.

Poor solvent -solvent

Good solvent
Long-range repulsion

Figure 4: In Freely-Jointed Chain, a

polymer can be viewed as a random
walk where monomers are connected
by bonds whose orientation is uncor-
related with the orientation of other
bonds. Notice that we have assumed
that there are no other interactions
(electrostatic, excluded volume, etc).
Picture from Wikipedia.

Freely Jointed Chain

We can start by considering polymers as composed of monomers

joined by “bonds” between monomers. As a first approximation, we
assume that the bonds are of a fixed length b but the orientation of
every bond can vary and is uncorrelated with all other bonds (see
Figure 4). We will see that this very simplistic picture captures many
of the essential features of polymers— especially entropic effects.
In fact, we will be mostly concerned with long polymers where the
length L is much larger than b so that the total number of monomers
N = L/b  1. In this case we expect entropy to dominate energetic
effects. The FJC model model does this by treating polymers as ran-
dom walks. In this way, we can assign a probability to each allowed
configuration. In doing this, we have neglected things like energetic
interactions and excluded volume.
Let us now analyze the FJC in greater detail. Let us call the posi-
 introduction to polymer physics. 5

tion of the j-th monomer ~r j . Let us also define the bond vector for
the j-th bond by
~τj = ~r j −~r j−1 (2)
By definition, we know that the bond vectors satisfy the relationships

|~τj | = b (3)
h~τj i = 0 (4)
2
h~τi~τj i = b δij . (5)

The first of these just fixes the length of the bond, the second that the
bond is equally likely to be oriented in all directions, while the final
equation is simply the statement that the bonds are uncorrelated.
Let us start with first calculating the mean end-to-end displace-
ment of the polymer ~R. We know that

h~Ri = h∑ ~τj >= ∑h~τj i = 0. (6)

j j

This is simply the statement that polymer is equally likely to be

pointed in all directions just like a random walk. However, we can
also look at the root-mean square displacement R defined by

R2 = h~R · ~Ri = h∑ ~τj~τk >= ∑h~τj~τk i = b2 N. (7)

j,k j,k

This is the more accurate measure of the size of the polymer that we
discussed earlier. We see that this argument gives us a simple scaling
relation
R ∼ bN 0.5 , (8)

and a scaling exponent ν = 0.5 (defined in Eq. 1).

At this point, it is worth better understanding what this exponent
ν means. Notice that if we have

R ∼ bN ν , (9)

then we can invert this relationship to get

 1
R ν
N∼ (10)
b

This implies that the fraction of the polymer contained in a radius R0

df
is just R1/ν
0 ≡ R0 , where this equation defines the fractal dimension
1
d f = ν . This is the usual way we define dimension since for d =
1, 2, 3 we would expect number of things contained to go like R0 , R20 ,
and R30 respectively. This is an interesting line of reasoning that tells
us something about the geometry of polymers.
 introduction to polymer physics. 6

From FJC to Gaussian Chain

It will also be helpful to derive a general probability distribution for

this chain. To do so, we will make use of the general relationship
between random walks and the diffusion equation (Fokker-Planck)
equation. It will be helpful now to consider a more concrete setting
of a polymer in d-dimensions. Let us label the three components of ~τ
by τα with α = 1, . . . , d labels the different directions. We known that

h~τ i = ∑hτα i = 0. (11)

α

From symmetry, we conclude that in fact we must have that each

of these individual directions is zero. More tricky, is to consider the
correlation function
hτiα τjβ i. (12)
To calculate this, we rewrite

h~τi~τi i = b2 δij (13)

in component form to get

∑hτiα τjα i = b2 δij . (14)

α

Once again, by symmetry we know that all directions are equivalent

so that we conclude
b2
hτiα τjα i = δij (15)
d
Finally, since different components are uncorrelated, we can write

b2
hτiα τjβ i = δ δ (16)
d ij αβ

To proceed, we will write a recursive equation for the probability

P(~R, N ) that a polymer with N monomers has end-to-end displace-
ment ~R. In particular, using Bayes theorem we can write
Z
P(~R, N ) = d~τ p(~τ ) P(~R − ~τ , N − 1), (17)

where p(~τ ) is just the probability of having an orientation ~τ for the

last bond. In the limit where N  1 and ~R  b, we can perform a
Taylor expansion of the right hand side. This yields (in component
notation)
Z
!
∂P(~R, N ) ∂P(~R, N ) 1 ∂2 P(~R, N )
P(~R, N ) = d~τ p(~τ ) P(~R, N ) − − ~τ + ∑ τα τβ
∂N ∂~R 2 αβ ∂Rα ∂R β
(18)
 introduction to polymer physics. 7

This yields using expectation values above the d-dimensional effec-

tive diffusion equation

∂P(~R, N ) b2 ∂2 P(~R, N )
= , (19)
∂N 2d ∂~R2
with N playing the role of time and effective diffusion constant
De f f = b2 /2d. We already know the solution to this equation is a
Gaussian distribution of the form
 d  ~2 
~ d 2 − dR
P( R, N ) = e 2Nb2 (20)
2πNb2
In other words, the polymer behaves like a Gaussian chain. This sug-
gests that we should be able to replace the more complicated FJC by
a Gaussian model and still capture the long-distance physics of the
problem. In fact, the reason for this is that the chain is essentially
composed on N random steps each with variance b2 /d. Since vari-
ances of independent processes add, this tells us that We will return
to this universality in a little bit.
This same argument also essentially tells us about the probability
distribution describing the difference between ~Rn − ~Rm . In particular,
we know that this will be a sum of n − m terms each with variance
b2 /d. For this reason, we know that
  d  d(~Rn −~Rm )2 
d 2 −
~n , ~Rm ) = 2| n − m | b2
P( R e (21)
2π |n − m|b2

Polymers as springs

Before proceeding, this also gives us some idea about how entropic
forces work. In the absence of external forces, polymers of course
like to contract. We can ask, how much force f is needed to fully
extend the polymer to distance R f . We will now treat this as a one-
dimensional problem in the direction of the force. In other other
words, how much do you have to pull the polymer in order to . Well
we know that we can also thing of this as a partition function
− F ( R,N )
P(~R, N ) ∝ e kB T
, (22)

where F ( R, N ) is the effective free energy which we can identify as

k B TR2
F ( R, N ) ∼ . (23)
2Nb2
Notice this means that a polymer essentially behaves like a spring
with effective spring constant that is proportional to the temperature:

kB T
ke f f ∼ , (24)
2Nb2
 introduction to polymer physics. 8

since partition function for spring is just

−k e f f x2
Pspring ( x ) ∼ e kB T
(25)

Now, we know if we exert a force f that the free-energy will be modi-

fied to yield
F ( R f ,N )− f R f
−
P(~R, N ) ∝ e kB T
, (26)
This combined free energy must be minimized at the force needed to
stretch polymer implying

∂F ( Re f f , N ) k B TR f
= ∼ f (27)
∂R Nb2
This basic idea that entropy can give rise to forces is an interesting
one – and one that periodically gets revived in fundamental physics
as a possible origin of quantum gravity (most recently by Verlinde
[Link]

Beyond Gaussian Chains

So far we have ignored everything except for Gaussian effects. How

can we incorporate these non-thermodynamic interactions. In gen-
eral, this will be really hard. However, surprisingly mean-field theory
does an incredibly good job of capturing the essential physics.

Accounting for excluded volume/short range repulsive interactions

Let us start with the simplest version of mean-field theory. Let us

try to take into account excluded volume. In particular, let us write
the volume of one segment as vc . Then the probability that a given
monomers does not overlap a second monomer is just 1 minus the
fraction of volume occupied by the second segment in d-dimensions
is just
q ∼ (1 − vc /Rd ). (28)
In general, for a polymer composed on N monomers, there are N 2
such potential overlaps. The probability that none of the segments
overlap is given by

2 − N 2 vc
w( R) = q( N ( N − 1)/2) ≈ (1 − vc /Rd ) N ≈ R3 >>vc e Rd (29)

where in writing this, like in all mean-field models we have ignored

the correlations between monomers.
Now we make the further assumption, that the we can write the
probability of having a polymer of length R with excluded volume
 introduction to polymer physics. 9

interactions, p f lory ( R) is just the probability of having a Gaussian

chain of length R given by Eq. 20 times the probability that no
monomers occupy the same volume

p f lory ( R) = p( R)w( R)
 
− d~R2 − N 2 vc
∝e 2Nb2 e Rd (30)
2 ~2
− N dvc − dR 2
∝e R 2Nb (31)

This allows us to identify an effective scaling free energy

N 2 vc d~R2
F∼ + . (32)
R d 2Nb2
The equilibrium R∗ will minimize this energy. Let us now differenti-
ate this equation to get

dN 2 vc dR∗
− + =0 (33)
Rd∗+1 Nb2

which yields the scaling relation

3
R ∗ ∼ N ν = N d +2 . (34)

Thus, we see that the repulsive interaction have modified our expo-
nent ν from the ideal model where ν = 1/2 to ν = 3/5 in d = 3
dimensions and ν = 3/4 in two dimensions. Surprisingly, this is in
good agreement with experiments!

Basic Phase Diagram of Polymers

We thus far considered only repulsive interactions. One can also

think about attractive interactions. Obviously, attractive interac-
tions will make the polymer more compact. With attractive inter-
actions and hard core repulsion due to steric occlusion, the polymer
monomer would like to stay as close as possible. In particular, we
expect the density of monomer to be O(1). Thus, we expect that
N/Rd ∼ 1, so that we have

R ∼ N 1/d . (35)

These basic considerations allow us to think of a stylized phase

diagram for polymers. Since the temperature controls the relative
effects of entropic versus energetics. Thus, at low temperatures we
would expect that attractive interactions dominate and the polymer
is “collapsed” with R ∼ N 1/d . At some transition temperature often
called T = Tθ , we expect a phase transition to an extended phase
 introduction to polymer physics. 10

where R ∼ N v . These two regions are separated by a critical regions

around T = Tθ where R ∼ N 1/2 . This is summarized in Figure 5.
7
We will not have time to work through this in detail. However, rest
2. Collapse
assured this can be made more precise using field theoretic methods
The case of attractive interaction may also be mentioned here. With attraction, and hard-core repulsion, the
and methods
monomers wouldfrom RG.
like to stay as closeThese
as possible. are quite
This gives a more interesting
or less compact packingand involved
of spheres so that the
monomer density inside a sphere enclosing the polymer is O(1) in N . Note that the density for the repulsive case
d 1−dν
N/R ∼ N → 0, for large N . A compact phase, also called a globule, would then have
calculations. 1
R ∼ N 1/d , i.e., ν = . (compact) (2.13)
d
The collapsed state is not a unique state and the polymeric nature is important in determining its overall property.
One expects a generic phase diagram, as schematically depicted in Fig.2, with a theta point at T = Tθ , a high Figure 5: Phase diagram of polymers
temperature (T > Tθ ) swollen or coiled phase and a low temperature (T < Tθ ) compact phase. This will be discussed
in detail in Section V. from "Flory Theory of Polymers"
(arxiv:1308:2414).

1/2
R~ N
1/d
ν
R ~N R ~N

T <Tθ T =Tθ T >Tθ

Collapsed Ideal Extended

FIG. 2: Schematic phase diagram of an isolated homopolymer. At high temperature T > Tθ , the polymer is in a swollen phase
(right), whereas one expects a compact globule at sufficiently low temperatures T < Tθ (left). These two regimes are separated
by a transition regime at T = Tθ (center) where the polymer behaves more or less as a Gaussian chain, at least in d > 3.

III. THE EDWARDS CONTINUUM MODEL

A. Discrete Gaussian model

The central limit theorem, as explained in Appendix A, allows us to describe a polymer by the distribution
W (r0 , . . . , rN ) of N bonds, τ1 = r1 − r0 ,. . ., τN = rN − rN −1 , each having a Gaussian distribution, as
N N
"# $d/2 % &'
! ! 1 1 τj2
W (r0 , . . . , rN ) = p (τj ) = 2
exp − 2 , (3.1a)
2πb 2 b
Understanding Universality and Self-Similarity: From(3.1b)
= Z exp [−βH ],
j=1
−1
j=1
WLC back
G
G

to FJC
where we have introduced the Gaussian Hamiltonian
N N
1 ( 2 1 (
βHG = τ = 2 (rj − rj−1 )2 , (3.1c)
2b2 j=1 j 2b j=1

One ofwiththe most

the partition powerful
function Z = (2πb ) and
G . interesting ideas to come out of poly-
2 N d/2

The Gaussian Hamiltonian is another representation of a polymer where the monomers are connected by harmonic
mer physics is Atthe
springs (Fig. 1c). ideatemperature,
any nonzero of scaling 2 . These
the equipartition ideas
theorem gives $τ %/b have
= d, whichtheir
allows theorigin
2
j
2
bonds in 2
Here we will follow the discussion
2 2
to have a nonzero rms length. The size of the polymer is given by $R % = db N .
the work of Michael Fisher and collaborators on phase transitions, of Doi as wall as Bhattacharjee et al in
arxiv:1308:2414
but polymers are arguably the place where they were most widely
and successfully used outside this original setting. The basic idea
of the scaling theory is that at long distances, the physics should be
independent of the scale at which I view the problem at short dis-
tances.
One of the defining properties of random walks is that they are
self-similar. This essentially means that as I view a random walk
at different scales, it essentially looks the same 3 . The under lying 3
A great demonstration of this can
reason for this is that we we ask how the radius of the random walk be found from Wolfram applet found
[Link]
scales with the number of steps, this takes the form of a powerlaw: SelfSimilarityInRandomWalk/

R ∼ Nν. (36)

To see why power laws are special, let us consider two kinds of
functions
A
f l (x) = α (37)
x
 introduction to polymer physics. 11

and
−x
Be ξ
f s (x) = . (38)
xα
Notice that the second of these has a natural length scale associated
with it ξ whereas there is no length scale with the first function.
Imagine now, we measured x on a different length scale x 0 = λx,
the we we that by setting A0 = Aλ−α the functional form of f l ( x )
remains invariant. Thus, no matter the scale λ at which one looks
the basic power law behavior remains the same. This is the idea of
scale invariance. We can contrast this with the function f s . If λx  ξ
we cannot simply rescale the parameter B and then obtain the same
functional form.
In fact, one can go a little further. We have shown that any power
law function is scale invariant. In fact, one can also show the opposite
must also hold: a scale invariant function (defined below) must take
the form a power law. Assume we have a scale-invariant function so
that:
f (λx ) ∼ λ− p f ( x ) (39)

for all choices of λ (this is known as continuous scale invariance). In this

case, we see that by setting λ = 1/x, we get that

f (1/x ) ∼ f (1) x p . (40)

Substituting, y = 1/x we come to the conclusion

f (y) ∼ y− p . (41)

Thus, we see that continuous scale invariance implies power law

scalings and vice versa.
In the context of polymers, let us revisit the FJC model, and in
particular Fig. 4. Notice that the scale b0 we chose for the length of
the links and the number of corresponding monomers N = L/b
was somewhat arbitrary. We could have just as easily chosen a dif-
ferent scale b0 and different number of monomers N 0 . However, we
expect that the long distance physics should be independent of these
choices. In particular, for long polymers L  1, we expect the physics
to be independent of all these choices.
Consider the probability P(~R, L) that a polymer of length L has
end to end displacement ~R. In principle, this probability depends on
all the exact details of our microscopic model. However, we know
that for long distances this should only depend on L. Let us now
imagine changing the macroscopic scale of the problem so that

L → λL, (42)
 introduction to polymer physics. 12

while keeping the microscopic scale b fixed. This is called a scale

transformation. Notice in Flory theory, we will be keeping the micro-
scopic scale fixed. In particular, since we have

R ∼ b1−ν ( Nb)ν ∼ Lν (43)

Thus, under a scale transformation L → λL and R → λν R. The

probability will show scaling if

P(~R, L) = λ x P(λν ~R, λL), (44)

for all λ. Let us now choose a λ = 1/L. This yields

1 ~R 1 ~R
P(~R, L) = x P( v , 1) ≡ x P ( v ), (45)
L L L L
where in the last line we have defined the scaling function P .
Now, we know that
Z
dr d P(~R, L) = 1. (46)

This implies that

Z ~R
1
1∼ dr d
x
P( x )
L L
Z ~
R
L x ∼ dr d P ( x )
L
Substituting r = r̃Lv gives
Z
L x ∼ Ldν dr̃ d P (~R̃) (47)

This implies that x = dν and our scaling function takes the form

1 ~R
P ( ) (48)
Ldν Lv
We can compare this with our expression in Eq. 20 using L = Nb
and ν = 1/2. In particular, we have that
 d  
d 2 − d~R2
P(~R, L) = e 2bL
(49)
2πLb

and indeed this scales as ∼ L−dv .

More scaling analysis of polymers in solutions

Thus far we have basically considered an isolated polymer. However,

polymers often occurs in solution. We are interested in understand-
ing polymers in a solution. After all the cell, is a very crowded place
as this famous picture from David Goodsell shows (see Fig. ??. We
will largely confine ourselves to some simple scaling arguments.
 introduction to polymer physics. 13

Figure 6: This illustration shows a

cross-section of a small portion of an
Escherichia coli cell. The cell wall, with
two concentric membranes studded
with transmembrane proteins, is shown
in green. A large flagellar motor crosses
the entire wall, turning the flagellum
that extends upwards from the surface.
The cytoplasmic area is colored blue
and purple. The large purple molecules
are ribosomes and the small, L-shaped
maroon molecules are tRNA, and the
white strands are mRNA. Enzymes are
shown in blue. The nucleoid region
is shown in yellow and orange, with
the long DNA circle shown in yellow,
wrapped around HU protein (bacterial
nucleosomes). In the center of the
nucleoid region shown here, you might
find a replication fork, with DNA
polymerase (in red-orange) replicating
new DNA. copyrigh David S. Goodsell
Critical Concentrations 1999.

Consider polymers in solution at some concentration c of monomers

in solution. We know that if this concentration is very small, we can
think about the polymers as isolated and non-interacting. However,
at large concentrations where the distance between polymers ap-
proaches the radius of the polymers themselves R g , the properties of
the polymer solution will differ from the isolated polymers. This is
depicted in Fig. ?? from Doi’s book.
Let us denote the concentration that separates these regimes by
c∗ . We can actually recover this from simple scaling arguments. We
know that
c∗ ∼ N/R3g (50)
However, we know that R g ∼ N ν so that we have

c∗ ∼ N 1−3ν (51)

Recalling that ν = 3/(d + 2), we see that for d = 3 that we have that
c∗ ∼ N −0.8 so that if N is very large that there is strong interactions
even at very small concentrations. For example, polyesterene with
molecular weight ∼ 106 starts to interact at 0.5% weight concentra-
tions.

Correlations length of polymers

Since the polymers are interacting in solution, the polymer solution

is no longer scale free but instead defined by a correlation length ξ.
Let us now try to use dimensional analysis and scaling arguments to
try to figure out the scaling properties of the correlation length. By
 introduction to polymer physics. 14

Figure 7: Figure 2.1 from Doi showing

meaning of critical concentration.

dimensional analysis, we know that we we expect that the properties

of the polymer will take the form

ξ ∼ R g f (cb3 , N ), (52)

where f is some arbitrary function of the dimensionless numbers cb3

and N.
We will ask what happens as we “coarse-grain” the chain so that
we group together some segments into a smaller number of seg-
ments: N → λ−1 N where λ > 1. In particular, we expect that for
large N the fundamental physics should not change under such a
coarse graining. Under such a transformation, we have that the con-
centration will decrease like N but the length scales will scale as λν :

N → λ−1 N, b → λν b, c → λ−1 c (53)

Since the correlation lengths and the radius of gyrations are physical
quantities, they must be invariant under such a transformation so that

ξ → R g f (λ3ν−1 cb3 , λ−1 N ) (54)

 introduction to polymer physics. 15

Let us choose λ = N, to get that

cb3
ξ = Rg f ( , 1)
N 1−3ν
 1−3ν −1
3 N
= R g f (cb , 1)
b3
c
= R g f ( , 1)
c∗
c
≡ Rg F ( ) (55)
c∗
This shows us the correlation function is a scaling function of c/c∗ .
To proceed, we will make some more scaling assumptions. In
particular, for c > c∗ , we know the correlations do not depend of the
details of the polymer involved. In fact, the correlation length must
be independent of N. Since we know that the powers of N come from
R g ∼ N ν and c∗ ∼ N 1−3ν . This means that
 x
c c
F( )∼ ∼ N − x(1−3ν) . (56)
c∗ c∗

This implies that since ξ is independent of N that

x = ν/(1 − 3ν) (57)

so that we get
 ν/(1−3ν)
c
ξ = Rg , (58)
c∗
which for d = 3 implies ξ ∝ c−3/4 which has been confirmed experi-
mentally.