Document [original]

Adhesion of surfaces mediated by adsorbed particles: Monte Carlo

simulations and a general relationship between adsorption isotherms and

effective adhesion energies

Tillmann Stieger,

Martin Schoen

and Thomas R. Weikl*

Received 3rd July 2012, Accepted 10th September 2012

DOI: 10.1039/c2sm26544c

In colloidal and biological systems, interactions between surfaces are often mediated by adsorbed

particles or molecules that interconnect the surfaces. In this article, we present a general relationship

between the adsorption isotherms of the particles and the effective, particle-mediated adhesion energies

of the surfaces. Our relationship is based on the analysis and modeling of detailed data from Monte

Carlo simulations. As general properties that should hold for a wide class of adsorption scenarios, we

find (i) that the particle-mediated adhesion energies of surfaces are maximal at intermediate bulk

concentrations of the particles, and (ii) that the particle coverage in the bound state of the surfaces is

twice the coverage in the unbound state at these bulk concentrations.

1. Introduction

Adhesion and adsorption are important phenomena in both

colloidal and biological systems. Characteristic aspects of these

systems are that the constituent molecules or particles typically

differ in size, and that the interactions between these constituents

are often dominated by surface interactions. Adsorption refers to

the binding of molecules or particles to the surfaces of larger

constituents and is typically characterized by adsorption

isotherms, i.e. by the surface concentrations of adsorbed mole-

cules or particles as a function of their bulk concentration or

chemical potential. Adhesion refers to the binding of two

surfaces that are typically large compared to molecular dimen-

sions and is characterized by adhesion energies per area.

Adsorption can lead to adhesion if molecules or particles bind

to two apposing surfaces, e.g. to the surfaces of two larger

particles or objects. The adhesion and aggregation of nano-

particles or microparticles, for example, can be mediated by

adsorbed proteins

1–5

or polymers.

6–10

Nanoparticles can affect

the adhesion of microparticles.

The adhesion of lipid

membranes can be caused by adsorbed proteins

12,13

or multiva-

lent ions

that crosslink the membranes. Membrane adhesion

may also be mediated by soluble proteins that interconnect

receptor and ligand proteins anchored in apposing

membranes.

15,16

In this article, we consider an ensemble of particles between

two parallel surfaces in Monte Carlo simulations. The two

surfaces can be seen as surface segments in the contact zone of

two constituents in colloidal or biological systems that are

significantly larger than the particles. The particles adsorb on the

surfaces and mediate adhesion if the separation of the surfaces is

close to the diameter of the particles. In our Monte Carlo

simulations, we determine the pressure that the particles exert on

the surfaces and the area concentrations of the adsorbed particles

at different surface separations. The effective particle-mediated

adhesion energy of the surfaces is then obtained by integrating

the pressure. Interestingly, the effective adhesion energy is

maximal at intermediate bulk concentrations of the particles.

Our analysis of the Monte Carlo results indicates that the

surface concentrations of the adsorbed particles depend in good

approximation on a single parameter, the sum of the chemical

potential and the binding energy of the particles, at least for

binding energies that are significantly larger than the thermal

energy kT where kis Boltzmann’s constant and Tdenotes the

temperature. Integration of these surface concentrations, or

adsorption isotherms, leads to free energies of adsorption in the

bound and unbound state of the surfaces. These free energies of

adsorption provide the basis for a simple model to calculate

effective, particle-mediated adhesion energies of surfaces that can

be generalized to a wide class of adsorption isotherms. The

simple model is in good agreement with the effective adhesion

energies obtained directly from the pressure measured in our

Monte Carlo simulations. In addition, the model explains why

the particle-mediated adhesion energies of surfaces are maximal

at intermediate bulk concentrations of the particles, and why the

particle coverage in the bound state of the surfaces is twice the

coverage in the unbound state at these bulk concentrations. Our

Technische Universit€

at Berlin, Stranski-Laboratorium f€

ur Physikalische

und Theoretische Chemie, Straße des 17. Juni 115, 10623 Berlin, Germany

North Carolina State University, Department of Chemical and

Biomolecular Engineering, 911 Partners Way, Raleigh, NC 27695, USA

Max Planck Institute of Colloids and Interfaces, Department of Theory

and Bio-Systems, Science Park Golm, 14424 Potsdam, Germany

This journal is ªThe Royal Society of Chemistry 2012 Soft Matter, 2012, 8, 11737–11745 | 11737

Dynamic Article LinksC

Soft Matter

Cite this: Soft Matter, 2012, 8, 11737

www.rsc.org/softmatter PAPER

Published on 01 October 2012. Downloaded by TU Berlin - Universitaetsbibl on 30/03/2016 13:32:14.

View Article Online

/ Journal Homepage

/ Table of Contents for this issue

model generalizes and helps to understand previous results

obtained in the special case of Langmuir adsorption.

17,18

2. The simulation model

We consider spherical particles between two parallel surfaces (see

Fig. 1). The particles repel each other, but are attracted by the

surfaces. In our model, the interaction potential V

between the

particles and the surfaces is short-ranged and decays to zero at

separations zof the particles from the surfaces close to the

particle diameter d(see Fig. 2). The interaction potential V

attains its minimum value Uat the separation of z¼d/2 at

which the particles are in close contact with the surfaces. The

depth Uof the potential minimum corresponds to the binding

energy of the particles at the surfaces. The soft, pairwise repul-

sion of the particles has the form V

¼4kT(d/r)

where ris the

distance between two particle centers.

We assume that the two parallel surfaces are segments of

colloidal objects that are large compared to the particles and

surrounded by the particle solution. The number of particles

between the parallel surfaces then varies because these particles

can exchange with the surrounding bulk of particles. We further

assume that the bulk particles constitute a large particle reser-

voir, with a bulk concentration X

of particles that is determined

by the chemical potential mof the particles (see Fig. 3). The

ensemble of particles between the two parallel surfaces consid-

ered here then corresponds to a grand-canonical ensemble with

chemical potential m.

3. Excess pressure and effective adhesion potential

In this section, we determine the effective, particle-mediated

adhesion potential of the surfaces from the pressure that the

particles exert on the surfaces. This pressure depends on the

separation Lof the surfaces and can be obtained from Monte

Carlo simulations (see Appendix for details). We consider here

the excess pressure to be

Dp(L)¼p(L)p(L¼N) (1)

since we assume that the two surfaces are surface segments of

larger objects that are fully surrounded by the particles. There-

fore at large separations, the overall forces exerted by the

particles are zero.

The effective, particle-mediated adhesion potential V

of the

surfaces is obtained by integration over the excess pressure

Dp(L):

Fig. 1 Monte Carlo snapshot for the surface separation L¼10dwhere dis the diameter of the particles. The particles repel each other, but are attracted

by the surfaces. In this snapshot, the chemical potential of the particles is m¼12.32kT, which corresponds to a bulk concentration of X

¼0.01/d

the particles away from the surfaces. The binding energy U¼10kT of the particles at the surfaces leads to an area concentration of X

¼0.42/d

of the

particles in the adsorption layers.

Fig. 2 The particle–surface interaction potential V

depends on the

distance zof the particle center from the surface. The interaction

potential attains its minimum value Uat the separation z¼0.5dat

which the spherical particle is in contact with the surface. The depth U>

0 of the minimum is the binding energy of the particle with the surface.

The potential is composed of a soft repulsive and a Yukawa-like

attractive term (see eqn (12) and (13) in the Appendix section for details).

Fig. 3 Bulk concentration X

versus chemical potential mof the particles

in our model (data points). At small values of X

and m, the two quantities

are related via ln(d

)x7.8kT +m(dashed line).

11738 | Soft Matter, 2012, 8, 11737–11745 This journal is ªThe Royal Society of Chemistry 2012

Published on 01 October 2012. Downloaded by TU Berlin - Universitaetsbibl on 30/03/2016 13:32:14.

View Article Online

Vef ðLÞ¼ð

DpL0dL0(2)

In Fig. 4, the excess pressure Dp(L) and the effective adhesion

potential V

(L) are shown for the binding energy U¼10kT and

chemical potential m¼12.32kT, which corresponds to a bulk

concentration X

¼0.01/d

of the particles. The effective adhe-

sion potential exhibits a minimum value U

at surface sepa-

rations Lclose to the diameter dof the particles since the particles

can bind to both surfaces at this separation. The depth U

of this

minimum corresponds to the effective, particle-mediated adhe-

sion energy of the surfaces. At surface separations Laround 1.6d,

the effective adhesion potential V

has a local maximum because

particles can no longer bind to both surfaces, and because

particles that bind to one of the surfaces sterically obstruct the

binding of particles to the other surface (see Fig. 5 and 6). This

maximum of height U

constitutes a barrier for adhesion. At

larger surface separations LT3d, the effective potential V

decays to zero because the particles adsorb independently on the

two surfaces.

A characteristic feature of the effective, particle-mediated

adhesion energy U

is that it exhibits a maximum at an inter-

mediate value m¼m* of the chemical potential (see Fig. 7). With

increasing binding energy Uof the particles, the location of this

maximum is shifted to smaller values of the chemical potential

(see Fig. 8) and, thus, to smaller bulk concentrations X

of the

particles. In the following, we will show that the maximum of the

function U

(m) can be understood from the adsorption isotherms

and adsorption free energies of the particles. Our starting point is

the surface concentration of adsorbed particles considered in the

next section.

4. Surface concentrations of particles

The surface concentrations of the particles in the adsorption

layers can be calculated from the concentration profiles X(z)of

the particles between the surfaces (see Fig. 6). Here, zis the

Cartesian coordinate perpendicular to the two surfaces, which

are located at z¼0 and z¼L. For large surface separations LT

3d, the particle concentration X(z) has two pronounced peaks

Fig. 4 (a) Excess pressure Dpexerted by the particles and (b) effective,

particle-mediated adhesion potential V

of the surfaces as functions of

the surface separation Lin units of the particle diameter d. In this

example, the binding energy of the particles is U¼10kT and the chemical

potential is m¼12.32kT, which corresponds to a bulk particle

concentration of X

¼0.01/d

. The dots in subfigure (a) represent the

Monte Carlo data, and the line results from interpolation. The effective

potential V

in subfigure (b) is obtained from the excess pressure Dp via

integration. The effective potential exhibits a minimum at a surface

separation close to the particle diameter at which the particles are firmly

bound to both surfaces. The depth U

of the potential minimum is the

effective binding of the surfaces. Because of the entropy of the confined

particles, the minimum is located at a surface separation slightly larger

than the separation L¼dwhere the total binding energy to both surfaces

is minimal for a particle. In this example, the minimum is located at Lx

1.01d, and the effective binding energy is U

x4.20kT/d

. The effective

potential exhibits a barrier of height U

at intermediate separations at

which particles that bind to one of the surfaces obstruct the binding of

particles to the other surface. In this example, the barrier is located at

Lx1.60dand has the height U

x0.83kT/d

Fig. 5 Monte Carlo snapshots at the surface separations L¼3d, 1.6d,

and dfor the same parameters as in Fig. 4.

This journal is ªThe Royal Society of Chemistry 2012 Soft Matter, 2012, 8, 11737–11745 | 11739

Published on 01 October 2012. Downloaded by TU Berlin - Universitaetsbibl on 30/03/2016 13:32:14.

View Article Online

with maxima at z-values close to 0.5dand L0.5dwhere the

particle–surface interaction potential V

is minimal. These two

peaks correspond to the single layers of adsorbed particles at the

two surfaces (see Fig. 5 and 6). At intermediate z-values in the

range d<z<Ldbetween the two peaks, the particle

concentration Xtends towards the bulk concentration X

because the particle–surface potential V

is practically 0 for

these z-values, and because packing effects of the particles

between the surfaces are negligible for the bulk concentrations X

< 0.1d

considered here. For surface separations Lclose to the

binding separation d, the concentration profile X

has a single

peak that corresponds to a single layer of particles bound to both

surfaces. The surface concentration X

of particles in the

adsorption layers is obtained by integration over the peaks in the

concentration profiles X(z):

Xs¼ð

XðzÞdz(3)

For large surface separations, the surface concentration X

defined in eqn (3) is the area concentration of the single layer of

particles adsorbed to one of the surfaces. For the surface

Fig. 6 Concentration profiles X(z) of the particles between the surfaces

for the same surface separations Land parameters as in Fig. 5. The two

peaks in the profiles for L¼3dand L¼1.6dcorrespond to the single

layers of particles adsorbed at the two surfaces. The peaks at the sepa-

ration L¼1.6dare lower in height than the peaks at L¼3dbecause

particles that bind to one of the surfaces sterically obstruct the binding of

particles to the other surface at this separation (see the snapshot in Fig. 5

for L¼1.6d). The single peak in the concentration profile at the sepa-

ration L¼dcorresponds to a layer of particles bound to both surfaces.

Fig. 7 Effective, particle-mediated binding energy U

of the surfaces as

a function of the chemical potential mfor the binding energies U¼8, 10,

and 12kT of the particles. The effective binding energy U

exhibits a

maximum at intermediate values of the chemical potential. The points

represent the Monte Carlo data, and the lines the simple model based on

eqn (5).

Fig. 8 Values m* of the chemical potential at which the effective binding

energy U

of the surfaces is maximal versus binding energy Uof the

particles. The data points result from an interpolation of Monte Carlo

data for U

as a function of m(see e.g. data points in Fig. 7). The line

results from eqn (7) of the simple model with the fit function for the

adsorption isotherm X

(m+U

) indicated as a dashed line in Fig. 9(b).

The simple model is in good agreement with the Monte Carlo results for

particle binding energies U$7kT, but deviates at smaller binding

energies.

11740 | Soft Matter, 2012, 8, 11737–11745 This journal is ªThe Royal Society of Chemistry 2012

Published on 01 October 2012. Downloaded by TU Berlin - Universitaetsbibl on 30/03/2016 13:32:14.

View Article Online

separation L¼d, the surface concentration X

is the area

concentration of the particles that are bound to both surfaces.

In Fig. 9(a), the surface concentration X

is shown as a func-

tion of the chemical potential mat the binding separation L¼dof

the surfaces and at the large surface separation L¼10d, for the

three binding energies U¼8, 10, and 12kT of the particles. The

area concentration X

increases with mand with the binding

energy Uof the particles, and is significantly larger at the surface

separation L¼dbecause the particles bind to both surfaces.

When plotted as a function of m+U

with U

¼Ufor large L

and U

¼2Ufor L¼d, the six curves of Fig. 9(a) fall onto a

single curve (see Fig. 9(b)), which indicates (i) that the surface

concentration X

depends on the sum of the chemical potential

and binding energy of the particles, and (ii) that the binding

energy at the surface separation L¼dis approximately twice the

binding energy at large separations, which is plausible since the

particles bind to both surfaces at this separation. The small

deviations between the curves in Fig. 9(b) presumably result from

small differences in the entropies of bound particles at L¼dand

at large L, which appear to be negligible compared to the binding

energies, at least for the binding energies Umuch larger than the

thermal energy kT considered here. Because of the soft repulsive

interactions of the particles, the surface concentration X

does

not saturate at large values of m+U

. A scaling argument

indicates that X

increases proportional to (m+U

)

1/6

for large

values of m+U

at which the adsorbed particles are arranged in a

hexagonal lattice (see Appendix).

The surface concentration X

of the particles at large surface

separation determines the height U

of the barrier in the effec-

tive potential V

. In Fig. 10, the barrier height U

of the

effective potential is shown as a function of this surface

concentration for the three binding energies U¼8, 10, and 12kT

of the particles. The values of X

here correspond to the values in

Fig. 9(a) at the large surface separation L¼10d. For a given

binding energy U, different values of X

in Fig. 10 result from

different values of the chemical potential mof the particles. The

three curves shown in Fig. 10 fall onto a single curve since the

steric interactions between the two adsorbed layers of particles

that lead to the potential barrier only depend on the concentra-

tions of the particles in these layers.

5. Adsorption free energies

In the grand-canonical ensemble, particle concentrations can be

expressed as derivatives of the grand-canonical potential, or ‘‘free

energy’’ with respect to the chemical potential m. The concen-

tration profile X(z) thus can be related to a z-dependent grand-

canonical potential f(z)via X(z)¼vf(z)/vm, and the surface

concentrations X

defined in eqn (3) can be associated with a

surface potential, or ‘‘free energy’’ of adsorption f

. In the

previous section, we have shown that the surface concentrations

at the surface separation L¼dand at large separations

depend in good approximation on a single parameter, the

rescaled chemical potential m+U

with U

¼2Ufor L¼dand

Fig. 9 (a) Surface concentration X

of particles in the adsorption layers

as a function of the chemical potential mat the large surface separation

L¼10d(three bottom lines) and at the binding separation L¼dat which

the particles strongly bind to both surfaces. At both separations, the

surface concentration X

increases with the chemical potential mand with

the binding energy Uof the particles. (b) Same surface concentrations X

as a function of the rescaled chemical potential m+U

with U

¼Ufor

L¼10dand U

¼2Ufor L¼d. In this rescaled plot, the six curves of

subfigure (a) fall onto a single curve. The dashed line represents a 9

order polynomial fit to the Monte Carlo data (see Appendix).

Fig. 10 Height U

of the barrier in the effective potential V

as a

function of the surface concentration X

of adsorbed particles at the large

surface separation L¼10dfor the binding energies U¼8, 10, and 12kT

of the particles.

This journal is ªThe Royal Society of Chemistry 2012 Soft Matter, 2012, 8, 11737–11745 | 11741

Published on 01 October 2012. Downloaded by TU Berlin - Universitaetsbibl on 30/03/2016 13:32:14.

View Article Online

¼Ufor large L(see Fig. 9). Therefore, we consider here a free

energy of adsorption f

(m+U

) that depends on m+U

and is

defined via X

(m+U

)¼vf

(m+U

)/vm, or alternatively via

fsðmþULÞ¼ð

XsðmþULÞdm(4)

up to an integration constant.

The free energy of adsorption f

(m+U

) is related to the

effective, particle-mediated binding energy U

via

(m,U)x(f

(m+2U)2f

(m+U)) (5)

because U

can be understood as the difference between the

adsorption free energies at the binding separation L¼dand at

large separations Lof the surfaces. The factor 2 in the second

term on the right-hand side of eqn (5) results from the fact that

we have two adsorption layers of particles at large surface

separations L. In Fig. 7, the simple model based on eqn (5) is

compared to the Monte Carlo data. The function f

(m+U

) here

has been obtained by integrating the dashed fitting function of

Fig. 9(b), with an integration constant determined from a fit to

the Monte Carlo data (see Appendix for details).

According to eqn (4), the chemical potential m* at which the

effective binding energy U

is maximal follows from the equation

vUef

vm xXsðmþ2UÞ2XsðmþUÞ¼0 (6)

An interesting consequence of eqn (4) thus is that we have

(m*+2U)x2X

(m*+U) (7)

at m¼m*, i.e. the surface concentration X

of particles at the

binding separation L¼dis twice the surface concentration X

large separations L. Within the numerical accuracy, this is indeed

the case for our MC results at the binding energies U¼8, 10, and

12kT (see Table 1). The location m¼m* of the maximum of the

effective binding energy U

(m) obtained from eqn (7) is in a good

agreement with Monte Carlo results for particle binding energies

U$7kT (see Fig. 8). The deviation at the smaller binding energy

U¼6kT presumably results from contributions of the binding

entropies of the particles, which are neglected in the simple model

based on eqn (5). For binding energies U#5kT of the particles,

the effective binding energy U

determined from Monte Carlo

simulations does not exhibit a maximum at an intermediate value

m* of the chemical potential.

6. Generalization to other adsorption isotherms

Our arguments in the previous section can be generalized to

adsorption scenarios with particle–surface interactions V

and

particle–particle interactions V

different from our simulation

model, provided the particles adsorb in single layers in these

scenarios, with binding energies Uthat are significantly larger than

the thermal energy kT. Adsorption scenarios are typically char-

acterized by adsorption isotherms, i.e. by the surface concentration

of adsorbed particles as a function of the bulk concentration X

of the particles, or alternatively, as a function of the chemical

potential mof the particles. For binding energies U[kT of the

particles, it seems plausible that X

is a function of the rescaled

chemical potential m+U

with U

¼Ufor single surfaces and U

2Ufor two surfaces with ‘‘binding separation’’ Lclose to the

particle diameter, as in our simulation model. In general, the

adsorption isotherms X

(m+U

) are monotonously increasing

functions, with a more or less pronounced S-shape as shown in

Fig. 9(b). For such isotherms, it seems likely that there are values m*

of the chemical potential that satisfy eqn (7) for given binding

energies U, which implies that the effective, particle-mediated

adhesion energy U

defined in eqn (5) is maximal at these valuesm*.

In the Langmuir adsorption scenario, for example, the parti-

cles are assumed to bind independently to ‘‘adsorption sites’’ at

the surfaces, which leads to the surface concentration:

17,18

XsðmþULÞx1

qeðmþULÞ=kT

1þqeðmþULÞ=kT (8)

with a numerical factor qand the area d

per binding site for

binding energies Umuch larger than the thermal energy kT.

Here, d

simply is the probability that a binding site is occupied

by a particle. The surface concentration X

in the Langmuir

model ‘saturates’ for large values of m+U

,i.e. it tends towards

the limiting value 1/d

, in contrast to the surface concentration X

of the soft particles in the model considered here, which increases

proportional to (m+U

)

1/6

for large values of m+U

according

to a scaling argument (see Appendix). From eqn (4) and (5), we

obtain the Langmuir free energy of adsorption

fsxkT

d2ln1þqeðmþULÞ=kT (9)

and the effective, particle-mediated adhesion energy

Uef xkT

d2ln 1þqeðmþ2UÞ=kT

ð1þqeðmþUÞ=kT Þ2(10)

which is maximal at the value

m*xUkTln q(11)

of the chemical potential.

7. Discussion and conclusions

In this article, we have derived a general relationship between the

surface concentration, or adsorption isotherm, X

, of adsorbed

particles and the effective, particle-mediated adhesion energy U

of two surfaces that are bound together by the adsorbed parti-

cles. The derivation of this relationship is based on a detailed

analysis of Monte Carlo results. Our main results are:

(1) The surface concentration X

of the adsorbed particles

depends in good approximation on the single parameter m+U

with U

¼2Uin the bound state of the surfaces and U

¼Uin

the unbound state, for binding energies Uthat are large

compared to the thermal energy U(see Fig. 9). An integration of

Table 1 Surface concentrations X

at the values m* of the chemical

potential that maximize the effective binding energy U

(see Fig. 7)

U[kT]m*[kT]X

(L¼d)[1/d

]2X

(L¼10d)[1/d

]

812.24 0.03 0.62 0.01 0.63 0.01

10 13.76 0.03 0.67 0.01 0.68 0.01

12 15.44 0.03 0.71 0.01 0.71 0.01

11742 | Soft Matter, 2012, 8, 11737–11745 This journal is ªThe Royal Society of Chemistry 2012

Published on 01 October 2012. Downloaded by TU Berlin - Universitaetsbibl on 30/03/2016 13:32:14.

View Article Online

the adsorption isotherms X

(m+U

) leads to free energies of

adsorption f

(m+U

) (see eqn (4)).

(2) The effective, particle-mediated adhesion energy U

of the

surfaces can be calculated as a difference of adsorption free

energies f

(m+2U) and 2f

(m+U) in the bound and unbound

state of the surfaces (see eqn (5)). This calculation is in good

agreement with values for the effective adhesion energy deter-

mined from the pressure on the surfaces measured in our Monte

Carlo simulations (see Fig. 7).

(3) The effective adhesion energy U

is maximal at an inter-

mediate value m* of the chemical potential. This intermediate

value follows from eqn (7) for the adsorption isotherms X

(m+

2U) and X

(m+U) in the bound and unbound state of the

surfaces (see Fig. 8).

(4) At the optimal chemical potential m* for adhesion, the

surface concentration in the bound state of the surfaces is twice

the surface concentration in the unbound state. This is a direct

consequence of eqn (7).

In our model, the general relationship between the adsorption

isotherm X

of the particles and the effective adhesion energy U

of the surfaces described by eqn (4) and (5) holds for binding

energies U$7kT of the particles (see Fig. 8). For these binding

energies, the differences between the binding entropies of the

particles in the bound and unbound state of the surfaces

apparently can be neglected. These differences arise since parti-

cles bound to both surfaces experience the superposition V

(z)+

(Lz) of particle–surface interaction potentials, which has a

different shape than the potential V

experienced by a particle

bound to one of the surfaces. The threshold for the particle

binding energies Ubeyond which eqn (4) and (5) hold may be

different for other particle–surface interaction potentials V

which in general will lead to other binding entropies.

At the optimal chemical potential m¼m* for adhesion, the

binding of the surfaces requires only a local rearrangement of

particles since the surface concentration X

(m*+2U) in the bound

state of the surfaces is equal to the sum 2X

(m*+U) of the surface

concentrations in the unbound state. At larger or smaller values of

m, in contrast, the equilibration of the surface concentrations

during binding requires a global transport of particles in or out of

the contact zone. For large contact zones, the shear strain in the

bound particle layer

19,20

may impede such a transport and, thus,

may lead to even larger differences between the effective adhesion

energies at m¼m* and at values of msmaller or larger than m*.

In experiments, maxima in adhesion strength have been

observed for intermediate concentrations of proteins that inter-

connect receptors and ligands in apposing membranes,

16,21

and

for intermediate concentrations of nanoparticles that affect the

adhesion of microparticles.

In principle, the effective surface

interactions induced by adsorbed particles can be measured

directly, e.g. via the surface-force apparatus,

12,22

or can be

inferred from the phase behavior of colloidal systems.

23,24

colloidal systems, changes in the particle concentrations may

lead to reentrant transitions in which surfaces or colloidal objects

first bind with increasing concentration of adhesive particles, and

unbind again when the concentration is further increased beyond

the optimum concentration at which the effective adhesion

energy is maximal.

In this article, we have focused on particles that exhibit purely

repulsive pair interactions V

and a short-ranged attraction V

to the surfaces (see Fig. 2). However, as argued in Section 6, our

general relationship between the adsorption isotherm of the

particles and the effective adhesion energy of the surfaces should

also hold for other particle–particle interactions V

or particle–

surface interactions V

at least as long as these interactions do

not lead to adsorption in multilayers. This is the case for weakly

attractive particle–particle interactions V

and other short-

ranged particle–surface interactions V

. For more strongly

attractive particle–particle interactions V

or long-ranged

particle–surface interactions V

that lead to multilayers of

adsorbed particles, the effective adhesion potential V

will

exhibit several minima that correspond to one, two, or more

layers of particles between the surfaces.

Layers of particles can also arise if the bulk of particles in

contact with the surfaces is quite dense, or ‘liquid-like’, not dilute

as assumed here. Such a ‘layering’ has been known from

computer simulation studies of ‘simple’ fluids composed of

spherically symmetric molecules or particles that have just three

translational degrees of freedom.

Layering manifests itself as

periodic oscillations of the particle concentration X(z) along the

normal of the surfaces, with a spacing between neighboring

peaks that approximately matches the diameter of the spherical

particles. The oscillations are damped as one moves away from

the surfaces because the particle–surface interaction potential

decays to zero with increasing distance from the substrate.

Experimentally, layering near solid surfaces can be detected as

oscillations in the force profile measured with the surface forces

apparatus. In this apparatus, one brings a thin film composed of,

e.g. nearly spherical octamethylcyclotetrasiloxane (OMCTS)

molecules between the surfaces of a pair of macroscopic cylinders

coated with a thin mica sheet.

22,26

The cylinders are arranged

such that their axes form a right angle. By varying the distance h

between the cylinders, one can measure the pressure p(h) exerted

by the confined film on the cylinders with molecular resolution.

Like the particle concentration X(z), p(h) also exhibits damped

oscillations with a wavelength that is equal to the bulk correla-

tion length.

If the confined fluid is composed of particles or molecules that

also possess rotational degrees of freedom, interesting orienta-

tional effects may arise. In the case of confined liquid crystals, for

example, prewetting phenomena arise at a solid surface that are

driven by the precise anchoring of individual particles at the

surface.

Here, ‘anchoring’ refers to an energetic preference of a

molecule’s orientation with respect to the plane of the solid

surface. In addition to these static effects, diffusion of liquid

crystals in nanoconfinement is also quite unique.

We have focused here on planar surfaces. An interesting aspect

of flexible surfaces such as lipid membranes is that they can wrap

around adhesive particles.

30–33

A partial wrapping can lead to

effective, surface-mediated interactions between the adsorbed

particles

34–36

and, thus, to different adsorption isotherms of the

particles, compared to planar surfaces.

Appendix

Interactions

In our model, the spherical particles are confined between two

planar and parallel surfaces separated by a distance Lalong the

This journal is ªThe Royal Society of Chemistry 2012 Soft Matter, 2012, 8, 11737–11745 | 11743

Published on 01 October 2012. Downloaded by TU Berlin - Universitaetsbibl on 30/03/2016 13:32:14.

View Article Online

z-axis. The interaction potential of the particles and the surfaces

is the sum of a soft repulsive and a Yukawa-like attractive term:

Vps ¼U"a1 d

z!10

a2

exph^

z



z#(12)

here, ^

z¼z+d/2, and zis the distance of the particle center from a

surface. The parameters

a1¼hdþ1

hd9and a2¼

10dehd

hd9(13)

are chosen such that the minimum of the potential V

is located

at z¼d/2, with minimum value Uwhere Uis the binding energy

of the particle. The interaction range of the potential depends on

the parameter h. We have chosen the value h¼7dfor which the

interaction potential decays to zero at separations zof the

particles from the surfaces close to the particle diameter d(see

Fig. 2).

The particle–particle interaction is purely repulsive:

Vpp ¼4kTd

r12

(14)

here, ris the distance between the two particle centers.

Pressure calculations

In the grand-canonical ensemble, equilibrium states correspond

to minima of the grand potential whose exact differential may be

given as

dF¼SdTNdmP

LdAP

AdL(15)

where Sdenotes entropy, Tis temperature, mis the chemical

potential of the particles, Ais the area of the surface, and P

½(P

) and P

are diagonal components of the pressure

tensor P. Because the particle–surface interaction depends only

on distances from the surfaces in the z-direction, properties of

our model are translationally invariant in the x- and y-directions.

To make contact with a microscopic level of description we

introduce the expression:

F¼kT ln X(16)

where kis Boltzmann’s constant and

XðT;m;A;LÞ¼X

expðbmNÞQðT;N;A;LÞ(17)

is the grand-canonical partition function with b¼1/kT.In

eqn (17)

QhZ

N!L3N(18)

is the partition function of the canonical ensemble for a system

with 3Ntranslational degrees of freedom, Lhﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

bh2=2pm

pis the

thermal de Broglie wavelength of a particle of mass m,his

Planck’s constant, and

Z¼ðdRexp½bVðRÞ (19)

is the configuration integral with the configurational energy

VðRÞ¼1

i¼1X

jsi

j¼1

VpprijþX

k¼1X

i¼1

Vpsðzi;kÞ(20)

Here, Rh{r

,.,r

} is a short-hand notation for the

configuration of the Nparticles, r

h|r

r

| is the distance

between the particles iand j, and z

i,1

¼z

and z

i,2

¼Lz

are the

distances of particle ifrom the two walls located at z¼0 and z¼

L, respectively.

A key quantity in this article is the pressure tensor component

. From eqn (15) and (16), it is easy to verify that

Pzz ¼kT

vln X

¼P

N!L3NvZ

vLT;m;A

¼Pid þPp

(21)

where the ideal-gas contribution is P

¼hNikT/Vwith V¼AL

and h.idenotes an average in the grand-canonical ensemble.

The contribution from particle interactions is given by

zz ¼ 1

2VX

i¼1X

jsi

j¼1DV0

pprijrij^

rij$^

ez2E1

2VX

k¼1

X

i¼1V0

psðzi;kÞzi;k(22)

where the first and second terms on the right side arise because of

particle–particle and particle–surface interactions, respectively,

0¼dV

/dr

0¼dV

/dz

¼r

, and ^

is a unit vector

pointing along the z-axis. In the limit L/N, we have P

where the bulk pressure is given by

Pb¼Pid 1

6VX

i¼1X

jsi

j¼1DV0

pprij rij E(23)

The pressure pin eqn (1) that the particle exerts on the surfaces

is identical to the pressure tensor component P

and calculated

from eqn (21) and (22) in our Monte Carlo simulations.

Monte Carlo simulations

In our Monte Carlo (MC) simulations, we numerically realize a

Markov process with a limiting distribution in configuration

space proportional to exp{b[U(R)mN]ln N!3Nln L}.

To achieve this we employ an algorithm originally proposed by

Adams for a simple Lennard-Jones fluid.

It proceeds in a

sequence of pairs of steps where particles are displaced and

created or destroyed.

We refer to a MC cycle as a sequence of Nattempts to displace

a molecule and Nattempted creations of new or removals of

already existing molecules where Nis the actual number of

molecules present at the beginning of a new cycle. To avoid

biasing the generation of configurations, displacements and

rotations as well as creation and removal are attempted with

equal probability. Our simulations are based upon 6 10

cycles

for equilibration followed by 10

cycles to compute ensemble

averages. To save computing time we employ a combination of a

11744 | Soft Matter, 2012, 8, 11737–11745 This journal is ªThe Royal Society of Chemistry 2012

Published on 01 October 2012. Downloaded by TU Berlin - Universitaetsbibl on 30/03/2016 13:32:14.

View Article Online

conventional Verlet with a link-cell neighborlist as described in

Allen and Tildesley’s book.

A particle is considered as a

neighbor of a reference particle if it is located within a sphere of

radius r

¼3.5d. In addition, fluid–fluid interactions are cut off

beyond an intermolecular separation of r

¼3dwhich we use

throughout this work; no such cutoff is applied to fluid–substrate

interactions.

Curve fitting

The dashed line in Fig. 9(b) represents the 9

-order polynomial

fit XsðmþULÞxð1=d2ÞP9

n¼0cnxnwith x¼(m+U

)/kT and fit

parameters c

¼0.513, c

¼0.0335, c

¼0.00212, c

¼0.000311,

¼3.03 10

5

¼2.914 10

6

¼6.42 10

7

1.13 10

8

¼2.63 10

9

, and c

¼1.13 10

10

Integration of X

(m+U

) leads to the adsorption free energy

fsðmþULÞxðkT=d2Þðcint P9

n¼0cnxnþ1=ðnþ1ÞÞ (see eqn (4)).

We have determined the value c

int

¼2.69 for the integration

constant from a fit of eqn (5) to the Monte Carlo data for the

effective binding energy U

shown in Fig. 7.

Asymptotic limit of large surface concentration

At large surface concentrations, the particles in the adsorption

layers are packed in a hexagonal lattice. To determine the

scaling form of the surface concentration X

in this limit, we

consider a surface area Awith Nadsorbed particles arranged in

a hexagonal lattice. The adsorption energy of the particles is

¼N(m+U

), and the sum of the repulsive interactions of

the particles is E

rep

¼3NV

(r(N)) ¼12N(d/r(N))

with the N-

dependent distance rðNÞ¼ ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

2A=N

p=31=4between neighboring

particles. Minimization of the total energy E

rep

with

respect to the particle number Nleads to a particle density of

x(0.55/d

)(m+U

)

1/6

Acknowledgements

We would like to thank Bartosz R

zycki and Marco G. Mazza

for valuable comments and fruitful discussions. Financial

support from the Deutsche Forschungsgemeinschaft (DFG) via

the International Research Training Group 1524 ‘‘Self-Assem-

bled Soft Matter Nano-Structures at Interfaces’’ is gratefully

acknowledged.

References

1 S. Srivastava, A. Verma, B. L. Frankamp and V. M. Rotello, Adv.

Mater., 2005, 17, 617–621.

2 D. Zhang, O. Neumann, H. Wang, V. M. Yuwono, A. Barhoumi,

M. Perham, J. D. Hartgerink, P. Wittung-Stafshede and

N. J. Halas, Nano Lett., 2009, 9, 666–671.

3 J. C. Ang, J.-M. Lin, P. N. Yaron and J. W. White, Soft Matter, 2010,

6, 383–390.

4 B. Bharti, J. Meissner and G. H. Findenegg, Langmuir, 2011, 27,

9823–9833.

5 M. Kendall, P. Ding and K. Kendall, Nanotoxicology, 2011, 5, 55–65.

6 K. Wong, P. Lixon, F. Lafuma, P. Lindner, O. Charriol and

B. Cabane, J. Colloid Interface Sci., 1992, 153, 55–72.

7 R. Shenhar, T. B. Norsten and V. M. Rotello, Adv. Mater., 2005, 17,

657–669.

8 R. Sczech and H. Riegler, J. Colloid Interface Sci., 2006, 301, 376–

385.

9 Y. Ofir, B. Samanta and V. M. Rotello, Chem. Soc. Rev., 2008, 37,

1814–1825.

10 D. Babayan, C. Chassenieux, F. Lafuma, L. Ventelon and

J. Hernandez, Langmuir, 2010, 26, 2279–2287.

11 C. Hodges, Y. Ding and S. Biggs, Adv. Powder Technol., 2010, 21, 13–

18.

12 Y. Hu, I. Doudevski, D. Wood, M. Moscarello, C. Husted,

C. Genain, J. A. Zasadzinski and J. Israelachvili, Proc. Natl. Acad.

Sci. U. S. A., 2004, 101, 13466–13471.

13 Y. Min, K. Kristiansen, J. M. Boggs, C. Husted, J. A. Zasadzinski

and J. Israelachvili, Proc. Natl. Acad. Sci. U. S. A., 2009, 106,

3154–3159.

14 T. Franke, R. Lipowsky and W. Helfrich, Europhys. Lett., 2006, 76,

339–345.

15 A. Saterbak and D. A. Lauffenburger, Biotechnol. Prog., 1996, 12,

682–699.

16 M. L. Dustin, T. Starr, D. Coombs, G. R. Majeau, W. Meier,

P. S. Hochman, A. Douglass, R. Vale, B. Goldstein and A. Whitty,

J. Biol. Chem., 2007, 282, 34748–34757.

17 B. R

zycki, R. Lipowsky and T. R. Weikl, Europhys. Lett., 2008, 84,

26004.

18 B. R

zycki, R. Lipowsky and T. R. Weikl, J. Stat. Mech.: Theory

Exp., 2009, P11006.

19 P. Bordarier, M. Schoen and A. Fuchs, Phys. Rev. E: Stat. Phys.,

Plasmas, Fluids, Relat. Interdiscip. Top., 1998, 57, 1621–1635.

20 M. Cates, J. Wittmer, J. Bouchaud and P. Claudin, Phys. Rev. Lett.,

1998, 81, 1841–1844.

21 R. N. Gutenkunst, D. Coombs, T. Starr, M. L. Dustin and

B. Goldstein, PLoS One, 2011, 6, e19701.

22 J. N. Israelachvili, Intermolecular and Surface Forces, Academic Press,

2nd edn, 1992.

23 M. M. Baksh, M. Jaros and J. T. Groves, Nature, 2004, 427, 139–

141.

24 E. M. Winter and J. T. Groves, Anal. Chem., 2006, 78, 174–180.

25 M. Schoen and S. H. L. Klapp, Nanoconfined Fluids: Soft Matter

Between two and three Dimensions, Wiley-VCH, Hoboken, 2007.

26 M. L. Gee, P. M. McGuiggan, J. N. Israelachvili and A. M. Homola,

J. Chem. Phys., 1990, 93, 1895–1906.

27 S. H. L. Klapp, Y. Zeng, D. Qu and R. von Klitzing, Phys. Rev. Lett.,

2008, 100, 118303.

28 M. Greschek and M. Schoen, J. Chem. Phys., 2011, 135, 204702.

29 M. G. Mazza, M. Greschek, R. Valiullin, J. Kaerger and M. Schoen,

Phys. Rev. Lett., 2010, 105, 227802.

30 R. Lipowsky and H. G. D€

obereiner, Europhys. Lett., 1998, 43, 219–

225.

31 M. Deserno, Phys. Rev. E: Stat., Nonlinear, Soft Matter Phys., 2004,

69, 031903.

32 C. Fleck and R. Netz, Europhys. Lett., 2004, 67, 314–320.

33 S. Cao, G. Wei and J. Z. Y. Chen, Phys. Rev. E: Stat., Nonlinear, Soft

Matter Phys., 2011, 84, 050901.

34 T. R. Weikl, Eur. Phys. J. E: Soft Matter Biol. Phys., 2003, 12, 265–

273.

35 M. M. M€

uller, M. Deserno and J. Guven, Europhys. Lett., 2005, 69,

482–488.

36 B. J. Reynwar, G. Illya, V. A. Harmandaris, M. M. M€

uller,

K. Kremer and M. Deserno, Nature, 2007, 447, 461–464.

37 D. J. Adams, Mol. Phys., 1975, 29, 307–311.

38 M. P. Allen and D. J. Tildesley, Computer Simulation of Liquids,

Clarendon Press, Oxford, 1987.

This journal is ªThe Royal Society of Chemistry 2012 Soft Matter, 2012, 8, 11737–11745 | 11745

Published on 01 October 2012. Downloaded by TU Berlin - Universitaetsbibl on 30/03/2016 13:32:14.

View Article Online