Theory of Friction between Neutral Polymer

Brushes
J. B. Sokoloff
Department of Physics and
Center for Interdisciplinary Research on Complex Systems
Northeastern University, Boston, Massachusetts 02115, U.S.A.,
E-mail address: [email protected]
April 25, 2007

Abstract

A type of static friction between two polymer brush coated sur-

faces, resulting from fluctuations from mean field theory is found, but

with creep-like motion for forces below the force of static friction,

which is much more rapid than the usual creep between solid surfaces

in contact. At sufficiently light loads, it is shown that polymer brush

coated surfaces can slide, with the load supported entirely by osmotic

pressure, and thus exhibit no static friction and only extremely weak

viscous kinetic friction.

1
1 Introduction

A polymer brush consists of a surface with a fairly concentrated coating of

polymer chains, each one of which has one, and only one, of its two ends

tightly bound to the surface. The mean spacing of the points of attachment

to the surface is much smaller than the radius of gyration of an isolated

polymer. Aside from the end that is attached to the surface, the remainder

of each polymer interacts very weakly with the surface. Polymer brushes

serve as extremely effective lubricants, producing friction coefficients as low

as 0.001 or less[1]. They must be immersed in a liquid which is a good solvent

for the polymers making up the brush in order to function as a lubricant.

It will be shown that a reasonably light load can be supported entirely by

osmotic pressure due to the solvent, resulting in extremely low viscous fric-

tion. Since joints in the human body are known to be immersed in a fluid,

known as the synovial fluid, it is possible that they are lubricated by some

type of polymer brush consisting of polymers known as lubricin which are

attached to the cartilage which coats the bones making up the joints[2]. The

density profile of a polymer brush (i.e., the density of monomers as a func-

2
tion of distance from the surface to which the polymers are attached) is well

established[3, 4]. What is not understood is how the interaction of polymer

brush coated surfaces in contact with each other is able to account for the

observed friction. For example, molecular dynamics studies generally do not

predict static friction[5], whereas surface force apparatus measurements due

to Tadmor, et. al.[6], report a force of static friction. What is observed is

that the shear force between the surfaces of the surface force apparatus in-

creases quite rapidly as a function of degree of sliding, reaches a maximum

value, and then decreases as the sliding continues. Presumably, if the force

that produces the sliding were below this maximum value, the sliding would

cease, although a small amount of motion, due to the elasticity of the poly-

mer brushes and the apparatus, would continue for a short time. This fits

the usual definition of static friction. The static friction that they report

increases logarithmically with the time that the polymer coated surfaces are

in contact (see Fig. 1 in Ref.[6]). The logarithmic increase of static friction

with resting time is qualitatively similar to a similar effect for solid-solid

contact, even though the mechanisms for it are much different[7].

3
 Pincus, Witten, Milner, etc. [3, 4], present the following picture: over all

except the very outer edge of a brush, the monomer density at a distance z

from the solid surface has the form A − Bz 2, where A and B are constants,

but there is a low density tail for z > hmax = (A/B)1/2, where hmax is the

height of an uncompressed polymer brush. When two brushes are put in

contact they are compressed and the resulting elastic force (which is really

of entropic origin) supports the load.

Kinetic friction is discussed in a paper by Joanny[8], assuming that it

occurs by reptation of polymers from one brush that get entangled in the

second brush. He finds only a viscous friction, but if the polymers near

the interface get entangled enough, there must be something that resembles

static friction if the times scale associated with the reptation motion is very

long compared to the time for the surfaces to move some very small distance

in creep motion as they attempt to slide. Even for solid-solid friction, there

can be creep motion[9]. We shall see, however, that creep for the polymer

brushes happens over a much shorter time scale than for bare solid-solid

contact.

4
 In this article, it is shown that, as a result of fluctuations from mean

field theory, polymer brushes in contact will exhibit a type of static friction,

resulting from interpenetration of polymers belonging to one brush into the

second. This is not the usual static friction between solid surfaces in con-

tact, for which when a force below it is applied, the surfaces are stationary

for extremely long times. For polymer brush coated surfaces, in contrast,

there is relatively rapid creep motion, which can be a few millimeters per

hour or faster. The mechanism for friction discussed in this article is to be

distinguished from the mechanism for friction discussed in Ref. 1b. The

mechanism in Ref. 1b is a mechanism for viscous kinetic friction in a region

in which the monomer density is in the dilute limit (as might occur in a re-

gion between but outside of the main part of the monomer density profile of

the two brushes. The friction in this mechanism results from viscous drag as

blobs from a polymer are dragged through such a region. It is also shown in

the present work that at lighter loads, the polymer brushes can support the

load without the mean field theory monomer profiles of the brushes being in

actual contact. In this case, the load is entirely supported by osmotic pres-

5
sure due to the monomers that fluctuate into the interface region between the

two brushes. The osmotic pressure in this case is not of the usual textbook

example in which a solute is trapped in an impenetrable membrane. Here

what plays the role of the membrane are the forces that hold the monomers

together in each polymer chain. Since each polymer is anchored to a one of

the two surfaces, the monomers cannot escape from the region between the

two surfaces. Solvent is pulled into the region between the plates to dilute

the monomer density, resulting in osmotic pressure.

2 Calculation of Friction between Polymer

This section deals with a type of static friction expected to occur for two

polymer brush coated surfaces in contact. Although the creep rate estimated

in the present work is considerably faster than that between typical bare

surfaces in contact and under shear stress, it can still be sufficiently slow to

allow one to define a force of static friction. In order to calculate the static

friction it is necessary to determine the number of monomers belonging to

polymers of one brush that penetrate into the second brush. This can be

6
accomplished using an approach similar to that used in Refs. [3, 4]. Consider

two polymer brushes which are pressed together. The mean field potential

for the hard core interaction of a monomer belonging to a polymer under

consideration with other monomers is given by

w[φ(z) + φ(D − z)], (1)

where D is the separation between the surfaces, to which the two polymer

brushes that we are considering are attached, z is the distance from one

of the two surfaces, which we will consider to be the lower surface, φ(z) is

the monomer density of one of the polymer brushes and w is the standard

monomer excluded volume parameter[10]. In equilibrium, φ has the standard

parabolic density[3]

wφ(z) = A − Bz 2, (2)

for z < h and zero for z > h, where h is the actual height of the brush

and B = π 2 kB T /(8a2 N 2 ), the value that B must have in order for a polymer

attached on one end to the surface at z=0 to be a solution to the equations of

equilibrium, as will be shown explicitly under Eq. (4) below. The quantity

A can be written as A = Bh2max[(2/3)(hmax /h) + (1/3)(h/hmax )2 ], which

7
represents the free energy of the brush and is minimum for h = hmax , which

denotes the equilibrium height when the brushes are not in contact with

each other. (In Ref. [3], it is shown that hmax = (12/π 2 )1/3(a2σw/kB T )1/3N ,

where the factors of kB T and a, which were set equal to one in Ref. [3], have

been inserted here. Here, a is the intermonomer distance, kB is Boltzmann’s

constant, T is the absolute temperature and σ is the number of polymers

belonging to the polymer brush attached to the surface per unit area.

When the brushes are in contact and the two brushes are pushed together

under a load, h < hmax , which means that the brushes get ”flattened out” so

that φ(z) no longer goes to zero at z=h. Within mean field theory, which is

expected to be accurate for sufficiently large N, almost all of the monomers

belonging to the brush coating the lower surface remain within that brush,

and the monomers belonging to the brush coating the upper surface remain

in that brush. For all finite values of N, however, there exist fluctuations

from mean field theory in which polymers from the lower brush penetrate

into the upper brush and polymers belonging to the upper brush penetrate

into the lower one. In order to calculate the degree of penetration, let us

8
consider the free energy needed to add a polymer to one of the brushes [3],

which represents the single polymer free energy,

Z N drn 2
F (k) − F (k − 1) = dn[(1/2)(kB T /a2)| | +
0 dn

w(φ(zn ) + φ(D − zn ))], (3)

where F(k) is the free energy for a two brush system containing k polymers

and rn is the location of the nth monomer in the polymer that was added to

the lower brush[3]. Here, zN denotes the z-component of the location of the

monomer at the end of the polymer which is attached to the surface and z0

is the location of the opposite end of the polymer under consideration, which

can have any value less than h. The minimum of the single polymer free

energy given in Eq. (3) is a local equilibrium configuration for the polymer

under consideration. We can find a local minimum value of Eq. (3) by setting

the variation of this expression with respect to rn equal to zero, which gives

d2 zn dφ(zn ) dφ(D − z)
(kB T /a2) 2
= w[ + ], (4)
dn dzn dzn

which looks like Newton’s second law in classical mechanics with the param-

eter n playing the role of time, and with dxn /dn = dyn /dn are constant. The

9
solutions of Eq. (4) are subject to the condition that the n=N monomer lies

on the surface to which the polymer under consideration is attached. This

means that the solution for zn for a polymer attached to the surface at z=0

satisfies the condition zN = 0.

For z < h, Eq. (4) is

d2 z dφ(z)
(kB T /a2) 2
=w = −2Bz, (5a)
dn dz

which has the form of Newton’s second law for a simple harmonic oscillator.

The solution to this equation for z0 < h, subject to the condition that zN = 0

is

zn = ccos(ωn), (5b)

where c is a constant, the choice of ”frequency” ω = (2Ba2 /kB T )1/2 =

π/(2N ) guarantees that zN = 0 for all polymers in this brush, and c=z0 < h.

These are the mean field theory equilibrium solutions for a polymer belong-

ing to the brush attached to the surface at z=0. By a similar argument, the

solution for zn for a polymer attached to the surface at z=D is

zn = D + acos(ωn), (5c)

10
with ω given above and a = −z0 > −h. These are the equilibrium solutions

within mean field theory. In mean field theory, for any finite value of N,

there is always penetration, resulting from thermal fluctuations, which can

be estimated to lowest order in 1/N by solving Eq. (4) for a polymer which

is attached to the surface at z=0 and penetrates into the region for which

z > h. For zn > h for two brushes in contact (i.e., for D < 2hmax ), Eq. (4)

reduces to
d2 z dφ(D − z)
(kB T /a2 ) 2
=w = −2B(z − D). (6a)
dn dz

The solution of Eq. (6a) is

z = D + acos(ωn) + bsin(ωn) (6b)

If we assume that for n > k1 , the value of n for the monomer belonging

to a polymer connected to the lower surface, which first penetrates into the

second brush, we must demand that the values of zn given by Eq.’s (5b) and

(6b) and their derivatives with respect to n be equal at n = k1 , since these

are second order differential equations. This condition gives

a = −h[cos(ωk1 ) − sin(ωk1 )tan(ωk1)], (7a)

11
 b = −2hsin(ωk1 ), (7b)

h
c= . (7c)
cos(ωk1 )

Since we expect for large N and low temperature that k1 << N , we find from

Eq. (6b), using Eq. (7) that

z0 − h ≈ (3/2)h(ωk1 )2 (8a)

and
dzn
|n=0 ≈ −2hω 2 k1 . (8b)
dn

Following the discussion in Ref. [3], we find that the increase in the free

energy of a polymer belonging to the lower brush when k1 of its monomers

penetrate into the upper brush is equal to the amount of work by the ten-

sion that would have to be applied to the n=0 monomer (which is equal

to (kB T /a2 )(dzn /dn|n=0 ) to pull it out of the lower brush by an amount

y = z0 − h, which is given by

Z h+y dzn
∆S = (kB T /a2 ) |n=0 dz0 . (9)
h dn

If we change variable of integration to k1 using Eq.’s (8), we obtain

∆S = 2(kB T /a2 )ω 4 h2k13 . (10)

12
Using Eq. (10), we find that the probability of k1 monomers of a polymer

from the lower brush penetrating into the upper brush is given by the Boltz-

mann factor

e−∆S/KB T = exp[−2ω 4(h/a)2 k13 ]. (11a)

Thus, from Eq. (11a) we conclude that the mean number of monomers of

one polymer belonging to one brush that gets entangled in the second brush

< k1 > is of order [(1/2)(a/h)2 ω −4 ]1/3, which can be written as

< k1 > /N = 2/π 4/3(N 1/2a/hmax )2/3u−2/3, (11b)

where u = h/hmax . The quantity N 1/2a is recognized as the radius of gyration

for a non-swollen polymer[10], which as we see gives a length scale for h. If h is

comparable or less than N 1/2 a, the fraction of monomers that fluctuate out of

the classical mean field monomer desnsity profile of Ref. [3] is large signifying

that this mean field theory must break down at that point. In fact, once

h < N 1/2a, it is certainly less than N 3/5a, the Flory radius, signifying that

we no longer have a polymer brush. For the case for the experiments reported

in Ref. [6], as N ≈ 1300, hmax = 480Ao and a ≈ 5Ao . From Eq. (11b) one

concludes that even for h = hmax , < k1 > is about 22 percent of N. One must

13
keep in mind, however, that the theoretical treatment employed here is only

expected to give rough orders of magnitude for and scaling relations between

the quantities of interest for this problem. In fact, numerical calculations

based on mean field theory[11] and molecular Dynamics simulations[5], show

that the actual value of the ratio < k1 > /N is actually noticeably smaller

than this value, which is not totally unexpected. Nevertheless, it is clear

from the present results that for values of h well below N 1/2a, the mean field

treatment of polymer brushes[3, 4, 10] is not expected to be applicable to

this system. Therefore, the discussion in this article will be restricted to

sufficiently large values of N and/or sufficiently dense packing of polymers

on the surfaces so that the condition for applicability of mean field theory,

< k1 > /N << 1 is satisfied. Understanding the predictions of mean field

theory, which is a manageable theoretical problem, is a first step towards

understanding the mechanisms for friction between polymer brushes. As will

be seen, this approach shows tendencies towards the observations reported

in Ref. [6], as the limit of highly compressed brushes studied in that work is

approached. A study of the concentration limit needed to obtain a complete

14
understanding of the work of Ref. [6] will be considered in future work using

other approaches.

If the < k1 > segments belonging to one brush, which penetrate into the

second are sufficiently long to form ”blobs” which are close packed in the

brush that it has penetrated, one must apply a force to disentangle these

blobs, which we identify with the force of friction. This is expected to occur

when the polymers belonging to the two brushes are in the semi-dilute regime

[10]. The formation of blobs for a polymer that penetrates a semi-dilute solu-

tion of polymers (as exists for a polymer brush into which it has penetrated)

is illustrated in Fig. 1. The condition for most parts of the polymer brushes

to be in the semidilute regime is cR3F /N >> 1. Since c ≈ (N/s2 hmax ), this

inequality becomes (N/s2 hmax )(R3F /N) >> 1. We may write RF in terms of

the interaction parameter w as RF = (w/kB T )1/5a2/5N 3/5 = R1 N 3/5, where

R1 = (a2 b3)1/5 and where b = (w/kB T )1/3. Substituting the expressions for

RF and hmax given above, we find that the left hand side of the last inequality

is equal to (hmax /s)4/5, implying that the brushes will be in the semidilute

regime as long as hmax >> s. The mean value of the monomer density in

15
the region of the brush into which the polymers from the other brush have

penetrated is approximately equal to

c = φ(z = h) = (N/s2 hmax )(α/u) (12)

where s is the distance between polymers at a surface, defined by σ = s−2

and α = (1 − u3). Then, from Ref. [10], the mesh size in the brush is ap-

proximately ξ = RF (cR3F /N)−3/4 . (The parameter ξ actually represents a

correlation length, as opposed to the dimension of a static mesh.) Substitut-

ing the relationship R3F /(s2 hmax ) = (hmax /s)4/5 in the above expression for

ξ, we obtain ξ = R1(Ns/hmax )3/5(u/α)3/4. The mean number of monomers

that get entangled in a brush to which they do not belong < k1 > can form

blobs of size ξ[10], provided that < k1 > is greater than g, the number of

monomers in a blob of linear size of the order of ξ. The quantity g is given

by g = (ξ/R1 )5/3. Substituting the above expression for ξ in this expression,

we obtain

g = (Ns/hmax )(u/α)5/4 (13)

. If < k1 > is greater than g, there is a force of static friction f per polymer

that penetrates, which is the force needed to pull a ”blob” through the mesh,

16
(illustrated in Fig. 1) which is (from section I.4.1 in Ref. [10]), given by

kB T
f= . (14)
ξ

If < k1 > is not greater than g, there will be no static friction. As will

be discussed later, this is not true static friction of the usual kind, because

even when a force per polymer less than f is applied, there will be observable

motion on time scales considerably shorter than the time scales usually asso-

ciated with creep motion under applied forces below the force of static fric-

tion. From the above condition for the occurrence of static friction, namely

< k1 > greater than g, we find from Eqs. (11b) and (13) that the relative

brush height u below which there will be static friction satisfies

u23/12
( 5/4
) < (2/π 4/3)(N 1/2a/hmax )2/3(hmax /s). (15)
α

There are two regimes of solutions to Eq. (15). When the right hand side of

the inequality is greater than one, there will be static friction for all values

of u, below a value which is very close to one (when α−1 , defined under Eq.

(12), is large). The reason for this is that since u is always less than or

equal to 1, the left hand side of the inequality will be less than one, except

17
for values of u for which u is sufficiently close to 1 so that α is sufficiently

small to make the left hand side greater than one. This cross-over occurs for

u ≈ 0.7. That is, there will be static friction for almost all values of u. When

the right hand side is less than one, there can be values of u above which

static friction disappears (i.e., when is Eq. (15) is not satisfied). From Eq.

(15), we see that the condition that the right hand side of this inequality be

less than one, namely (2/π 4/3)(N 1/2a/hmax )2/3 < (s/hmax ), is very difficult

to satisfy, since in general when the condition for the existence of a polymer

brush is satisfied i.e., the radius of gyration of an isolated polymer is less than

hmax and less than the spacing of polymers attached to the surface (i. e.,

s), (N 1/2a/hmax > s/hmax . This implies that for polymer brushes in contact,

there will virtually always be friction due to entanglement of the polymers.

The static friction per polymer is the force needed to dislodge a blob from

the mesh, which is given by

kB T kB T α
f= =( )(hmax /Ns)3/5 ( )3/4. (16a)
ξ R1 u

For u close to 1, but still small enough so that α is not too much smaller

18
Figure 1: The mechanism for friction discussed here is illustrated for just
two blobs for clarity. Dark lines represent polymers belonging to a polymer
brush. The lighter lines (red if viewed in color) represent a polymer that has
penetrated into the brush represented by the dark lines. The mesh spacing in
which these blobs reside are of size ξ, which is actually a correlation length,
rather than a static mesh spacing[10]. Its size is exagerated here. In reality,
it is only about a factor of 2 larger than the intermonomer spacing within an
interpenetrating blob.

than 1,

f σ = (6.7 × 103 N/m2 )(α/u)3/4, (16b)

using a = 5Ao and b = 2.42Ao , the value of b which gives hmax = 480Ao , the

uncompressed height of the polymer brushes of Ref. [6].

When a force below the maximum force of static friction is applied to the

19
surfaces, we anticipate that the entangled polymer blobs will get pulled out

of the brush in which they are entangled in a time of order

ηs ξ 3 3
τ ≈[ ]N , (17a)
kB T 0

where N0 =< k1 > /g, ηs is the viscosity of the solvent and ξ is the blob size,

from Eq. (VIII.16) in Ref.[10]. Combining Eq. (17a) and our expressions for

< k1 >, ξ and g, we get

τ = (8/π 4 )(ηs R31 /kB T )(a/s)6/5(Na/hmax )4/5N 2 α3/2 /u7/2. (17b)

Using the values for a and b given under Eq. (16) we obtain (8/π 4 )((ηs R31 /kB T ) =

0.698×10−12 s, at room temperature for η = 0.01poise, the viscosity of water.

For more viscous solvents, τ will be larger than the values quoted here. On

the average, an entangled polymer will work itself free in a time τ , resulting

in the center of masses of the two surfaces sliding a distance s/Ne relative to

each other where Ne is the number of entangled polymers. Multiplying this

by the rate at which the entangled polymers disentangle by reptation, which

is given by τ −1Ne , we get a mean creep velocity of s/(τ ). Although for u

comparable to 1, this quantity is of the order of a cm/s, since τ ∝ h−7/2 , as

20
the polymer brushes are compressed more and more, τ increases rapidly. For

sufficiently long chains which are sufficiently compressed, creep motion will

more closely resemble that for applied forces below the force of static friction

for bare surfaces in contact. This shows a trend towards the static friction

observed in Ref. [6]. For example, for the above values for a and b and for

N=1300, s/τ ≈ 10−2 cm/s for u=0.2. If N=10,000, it becomes for this value

of u, approximately 10−4 cm/s, or about a mm per hour.

The normal force per polymer that can be supported by two brushes, each

with a parabolic monomer distribution is the derivative of the free energy per

polymer of one of the brushes given in Eq. (31) of Ref. [3], which can be

written as
N 2 σw 1 u2 u5
F =( )[ + − ], (18)
hmax 2u 2 10

where u = h/hmax . The resulting normal force fN per unit area, which is

equal to minus the derivative of F with respect to h multiplied by σ, is given

by
N 2 σ 2w 1 u4
fn σ = ( )[ − u + ], (19a)
h2max 2u2 2

21
which can be written as

1 u4
fn σ = (kB T /a3 )(Na/hmax )2(b3 a/s4 )[ − u + ]. (19b)
2u2 2

The quantity fn σ is zero for u=1 and increases as u decreases, increasing as

u−2 for sufficiently small u. Since the force of friction per unit area from

Eq. (16b) increases only as u−3/4 as u decreases, the friction coefficient

decreases as u5/4 for small u. For a = 5Ao , b = 2.42Ao , s = 84A0 and

N=1300, the quantity multiplying the square bracketed expression is equal

to 8.33 × 103 N/m2 . From Eq. (16b) we see that the friction coefficient only

becomes small when the brushes are fairly strongly compressed (i.e., for small

u).

The kinetic friction in the slow speed sliding limit (but still much greater

than the creep rate) which must be comparable to the static friction can be

found from the Tomlinson model [12], as follows: We may crudely treat an

entangled polymer as a spring, one end of which is moving in a potential

possessing several minima, taken to represent the potential due to polymers

in the brush in which it is entangled. The opposite end of the spring is

assumed to move with a sliding velocity v. Then the total potential of this

22
polymer is given by

V = (1/2)α(x − vt)2 + u(x), (20)

where α is the force constant of the spring, u(x) is a potential which pos-

sesses several minima (which represents the energy needed to pull a polymer

d2 u(x)
out of a brush in which it is entangled, and t is the time). If α < dx2 max
| ,

d2 u(x)
this potential possesses multiple minima[12]. Here dx2 max
| is the maximum

force constant of the bottom of the entanglement potential well. Following

the treatment of a semi-dilute solution of polymers in Ref. [10], the entangle-

ment potential represents the energy that must be expended to pull a ”blob”

through the mesh created by the other polymers among which the polymer

under consideration is entangled. In order to pull the polymer through the

mesh, each ”blob” must be stretched so that it will become small enough to

fit through. It follows that the force constant of the mesh potential must be

equal to the force constant for a single ”blob.” Since the chain as a whole

consists of a bunch of ”blobs” connected in series, the effective force constant

of the chain must be equal to the force constant of a single ”blob” divided by

the number of ”blobs” of the polymer chain under consideration which are

23
entangled in the other brush. Then the above condition for multistability will

be satisfied, if the stretching of the part of the polymer that is still residing in

the brush to which it belongs is not inhibited significantly by the surrounding

polymers in the brush. As the chain is pulled through the potential a poten-

tial minimum containing the end of the chain periodically becomes unstable,

allowing the end of the chain to drop into a potential well of lower energy.

When this occurs, the resulting drop in potential energy gets converted into

kinetic energy, which is assumed to get rapidly dissipated into excitations of

the system. Setting the rate at which such energy dissipation occurs equal

to the product of the kinetic friction and v, we determine that there must

be nonzero kinetic friction in the low v limit. Clearly there can only be mul-

tistability of the type described above (and hence ”stick-slip” motion) if the

time it takes to pull a blob out of a region in which it is entangled is small

compared to τ , where τ is defined in Eq. (17) (i.e., (s/v) << τ ).

3 Floating Polymer Brushes

Let us now consider the possibility that at sufficiently light load the bulk of

the two polymer brushes might not be in contact, meaning that the load could

24
be entirely supported by osmotic pressure due to those polymers that extend

out of each brush into a thin interface region of thickness D − 2h. (Because

of the osmotic pressure in the interface region, h will be compressed below

hmax , even though the brushes are not in contact.) If this interface region is

sufficiently thick so that the polymers from the two brushes do not entangle

to any significant degree, there can be negligibly small static friction. Then,

the end of the polymer z0 is located in the interface region. For the part of

the polymer in the interface region (0 < n < k1 ), dzn /dn = −v1, a constant,

and hence zn = z0 − v1n. For n > k1 , zn is still given by Eq. (5b) and

(7c). Again requiring continuity of zn and dzn /dn at n = k1 and assuming

that ωk1 << 1, we obtain v1 = hωk1 , and hence, the free energy to pull k1
R k1
monomers belonging to a polymer out of a brush ∆S = (kB T /a2 ) 0 v1dz0 =

(2/3)(kB T /a2)h2 ω 4 k13 and hence the probability that k1 monomers of the

chain penetrate into the interface region is given by

P = exp[−(2/3)(hω 2 /a)2 k13 ] (21a)

Hence < k1 > /N ≈ (3/2)1/3 (a/hω 2 )2/3 = (24/π 4 )1/3(N 1/2a/hmax )2/3u−2/3.

As was pointed out in the last section, this estimate for < k1 > /N is only an

25
order of magnitude estimate, and numerical mean field theory calculations

generally give a smaller value[11]. In order for there to be no significant

penetration of polymers belonging to one brush into the second (which might

lead to static friction), we require that

D − 2h > z0 − h ≈ hω 2 < k1 >2 . (21b)

In order to estimate z0 − h, the amount that a polymer extends out of the

mean field polymer brush density profile, we must first find h by setting

the osmotic pressure equal to the load per unit area that can be supported

by a polymer brush. The mean monomer density in the interface region is

c ≈ [σ/(D − 2h)] < k1 >. The maximum value of the monomer concentra-

tion that can occur without significant penetration of polymers belonging to

one brush into the second brush is obtained by substituting for D-2h using

Eq, (21b), which gives for the maximum possible monomer concentration

(2/π)2 (N/s2 h)(N/ < k1 >). As this is greater than the monomer density in-

side a brush, given by Eq. (12), since < k1 > /N, 1, we conclude that c in the

interface region can be made as large as we wish if D-2h is made sufficiently

small. Even though this expression for the maximum possible value of c in

26
the interface is can be greater than the monomer density inside a brush, such

a situation will never occur. The reason for this is that before this happens,

the brushes will be subject to very high normal force, which will compress

them to comparable densities. Of course, at this point the brushes can for

all purposes be considered to be in contact.

If D-2h is sufficiently large, we can be in the dilute regime, in which the

osmotic pressure is given by[10]

ca3
Posm = (1/2)(kB T /a3)( ). (22)
< k1 >

The equilibrium value for h is obtained by setting this expression for the os-

motic pressure equal to the load per unit area given by Eq. (19b) and solving

the resulting expression for h. For a = 5Ao , the quantity (1/2)(kB T /a3) =

1.6 × 107 N/m3 . Then for ca3 ≈ 0.002 (a value below which one is likely

to be in the dilute regime), the osmotic pressure can support a load of or-

der 103 N/m2 . The dilute regime for the monomers that fluctuate out of

a polymer brush can be determined by introducing a Flory radius R0F for

that segment of the polymer which extends into the interface region, in

an analogous manner to which it is defined for a whole polymer, namely

27
R0F = R1 < k1 >3/5. The dilute regime is defined by cRF03 / < k1 ><< 1.

When this condition is satisfied, the two polymer brush coated surfaces will

only exhibit viscous friction (i.e., friction that vanishes as the sliding velocity

approaches zero. Using the fact that c =< k1 > /(s2 (D − 2h), this con-

5/3
dition becomes < k1 ><< (s10/9/R1 )(D − 2h)5/9 . Blobs formed from a

segment of a polymer that fluctuates into the region between the two mean

field theory parabolic profiles of the two polymer brushes will not get en-

tangled in polymer segments that fluctuate into this region from the other

brush (which if it did occur might lead to static friction) if g >< k1 >. The

number of monomers in a single blob g defined by g = (ξ 0 /R1 )5/3, where ξ 0 ,

the mesh size for the fluctuating monomers from the second brush is given by

ξ 0 = R0F (cR03
F / < k1 >)
−3/4
. Then from the above condition for being in the

dilute region, namely cRF03 / < k1 ><< 1, we find that ξ 0 >> R0F , which from

the above definitions of g and R0F we find that the latter condition implies

that g >>< k1 >.

As D-2h becomes smaller, the interface region will clearly switch into

the semidilute regime, resulting in the occurrence of a type of static fric-

28
tion resulting from monomers belonging to one brush getting entangled in

monomers belonging to the second brush, likely to be of the same order of

magnitude and have a similar creep rate as the static friction found for the as-

sumed case of brushes which touch each other, considered in the last section.

This shows a trend towards the behavior found in Ref. [6], i.e., the friction

only becomes large when the brushes are pushed together with sufficiently

large load.

4 Conclusions

We conclude that entanglement of polymers belonging to one polymer brush

in a second brush, in which it is in contact, shows a tendency towards the

occurrence of the static friction observed in Ref. [6] for two polymer brushes

in contact, as the number of monomers N decreases from values for which

mean field theory is valid towards the value of N for the polymer studied

in Ref. [6], although strictly speaking, the mean field theory used here is

only marginally applicable for the polymers studied in that work. Although

it was estimated that there is a ”creep velocity,” which is much larger than

the creep velocities found for bare solids in contact, it can be sufficiently

29
small so that on short time scales one can consider there to be static friction.

Osmotic pressure alone was found to be able to support a load of the order of

103 P a while keeping the polymer brushes sufficiently far apart to prevent the

entanglement that results in static friction. When the polymers are chared,

counterions can provide about 105 P a of osmotic pressure[13].

Acknowledgment

I wish to thank the Department of Energy (Grant DE-FG02-96ER45585). I

would also like to thank R. Tadmor for many discussions of his work.

References

[1] Klein, J.; Annual Reviews of Material Science 1996, 26, 581-612;

Taunton, H.; Toprakcioglu, C.; Fetters, L. J.; and Klein, J.; Macro-

molecules 1990, 23,571-580; Klein, J.; Kumacheva, E.; Mahalu, D.; Per-

ahia, D.;Fetters, L. J.; Nature 1994, 370, 634; Klein, J.; Perahia, D.;

Warburg, S.; Fetters, L. J.; Nature 1991, 352, 143;Klein, J.; Colloids

Surf. A 1994 86, 63; Klein, J.; Kumacheva, E.; Perahia, D.; Mahalu, D.;

Warburg, S.; Faraday Discussions 1994, 98, 173; Granick, S.; Demirel,

30
 A. L.; Cai, L. L.; Peanasky, J.; Israel Journal of Chemistry 1995, 35, 75;

Raviv, U.; Glasson, S.; Kampf, N.; Gohy, J. F.; Jerome, R.; Klein, J.;

Nature 2003, 425, 163.

[2] Klein, J.; Proceedings of the Institute of Mechanical Engineers: part J,

Journal of Engineering Tribology, special Issue on Biolubrication 2006,

220, 691-710.

[3] Milner, S. T.; Witten, T. A.; and Cates, M. E.; Macromolecules 1988,

21, 2610-2619; Zulina, E. B.; Priamitsyn, V. A., Borisov, O. V.; Vysoko-

mol Soedin 1989, Ser. A 30, 1615; Semenov, Sov. Phys. JETP 1985,

61, 733; Milner, S. T. Polymer Brushes. Science 1991, 251, 905-914;

de Gennes, P. G.; Advances in Colloid and Interface Science 1987, 27,

189-209; Alexander, S.; J. Physique 1977, 38, 983; Klein, J.; J. Phys.:

Condens. Matter 2000, 12, A19 - A27; Milner, S. T.; Polymer Brushes.

Science 1991, 251, 905-914; Wijmans, C. M.; Zhulina, E. B.; and Fleer,

G. J.; Macromolecules 1994, 27, 3238-3248; Klein, J., and Pincus, P.;

Macromolecules 1982, 15(4), 1129-1135.

31
[4] Witten, T. A.; Leibler, L.; and Pincus, P.A.; Macromolecules 1990, 23,

824-829.

[5] Grest, G. S.; Advances in Polymer Science 1999, 138, 149; Murat, M.;

Grest, G. S.; Macromolecules 1989, 22, 4054-4059; Grest, G. S.; J. Chem.

Phys. 1996, 105,5532; Grest, G. S.; Current Opinion Colloid Interface

Science 1997, 2, 271; Grest, G. S.; in ”Dynamics in Small Confined

Systems III (Materials Research Society, Pittsburg, 1997), ed. Drake, J.

M.; Klafter, J.; Kopelman, R.; vol. 464, 71; Grest, G. S.; Phys. Rev. Lett.

1996, 76, 4979.

[6] Tadmor, R.; Janik, J.; Klein, J.; and Fetters, L. J.; Physical Review

letters 2003, 91, 115503; Tadmor, R., private communication.

[7] Baumberger, T.; Solid State Communications 1997, 102, 175.

[8] Joanny, J.-F.; Langmuir 1992, 8 989.

[9] E. Rabinowicz, ”Friction and Wear of Materials” (John Wiley and Sons,

New York, 1966), ch.4; Heslot, F; Baumberger, T.; Perrin, B.; Carloli,

B.; Caroli, C.; Phys. Rev. E 1994, 49, 4973.

32
[10] de Gennes, P.-G.; ”Scaling Concepts in Polymer Physics,” Cornell Uni-

versity Press, Ithaca 1979).

[11] Kim, J. U.; Matsen, M. W.; unpublished; Matsen, M. W.; J. Chem.

Phys. 2004, 121, 1939-1948; Matsen, M. W.; J. Chem. Phys. 2002, 117,

2351; Milner, S. T.; Witten, T. A.; Macromolecules 1992, 25, 5495-5503.

[12] Caroli, C.; and Nozieres, Ph.; European Physical Journal 1998, B4, 233;

Physics of Sliding Friction, edited by Persson, B. N. J.; and Tosatti, E.;

NATO ASI Series E: Applied Sciences(1996), Vol. 311 (Kluwer Academic

Publishing, Dordrecht); Brillouin, M.; Notice sur les Travaux Scientifiques

(Gautier-Villars, Paris 1909); Tomlinson, G. A.; Philosophical Magazine

1929, series 7, 905.

[13] J. B. Sokoloff, unpublished.

33