You are on page 1of 45

Theoretical and Experimental Chemistry, Vol. 43, No.

3, 2007

COMPETITIVE ADSORPTION

V. M. Gun’ko UDC 541.183

A review is presented on the multicomponent competitive adsorption, static equilibrium adsorption, and
dynamic nonequilibrium adsorption from gaseous and liquid media onto the surface of porous and disperse
solids. The state of the theory of multicomponent adsorption is analyzed. Not only traditional methods were
used for investigations in this field but also NMR spectroscopy (difference in the spectra of adsorbed and free
compounds), FTIR spectroscopy (difference in the spectra of surface groups and bonds in adsorbates),
calorimetry, thermally stimulated depolarization, photon correlation spectroscopy, electrophoresis, and
cryoporometry.

Key words: multicomponent competitive adsorption, equilibrium adsorption, dynamic adsorption, surface
reactions, multicomponent adsorption theory.

INTRODUCTION

Several different adsorbates are almost always present in real systems on solid–gas, solid–liquid, or liquid–gas phase
separation boundaries. These different adsorbates compete for adsorption sites characterized by maximum heat of adsorption
and minimum free energy of adsorption. The adsorption of different substances such as polar and nonpolar compounds on a
heterogeneous surface may occur on different types of sites. In this case, the competition may be insignificant and the
predominant adsorption of one of the adsorbates on certain sites causes selective adsorption. Studies of competitive adsorption
hold great importance since the selectivity of adsorbents, efficiency of concentrating and separating compounds, filtration,
heterogeneous catalysis, biocompatibility of artificial biomaterials, and may other methods, effects, and phenomena depend in
some way specifically on competitive adsorption on phase separation boundaries [1-6]. Many studies, monographs and reviews
[1-22], dissertations [23, 24], special reports [25], and detailed in-depth studies [26-47] have been devoted to competitive,
multicomponent and multiphase adsorption and have provided a basic understanding of both multicomponent adsorption and
the adsorption of individual compounds. With this in mind, the present author has examined mainly recent publications on
competitive adsorption and some trends in this field.
We should note that high purity of the adsorbates in the adsorption of individual compounds such as nitrogen, argon,
benzene, and CO2 has been a necessary condition for precise measurements and accurate calculations and, hence, competitive
adsorption in this case is extremely undesirable. However, data on the adsorption of individual compounds are often used in
analyzing suitable multicomponent mixtures [1, 2]. Competitive or multicomponent adsorption, similar to individual
adsorption, leads to a decrease in the free energy of the system. In this case, the entropy contribution (TDS) is always
destabilizing [3]. Both the enthalpy and entropy factors of adsorption depend on the molecular size of the adsorbates, chemical
structure of the molecules, characteristics of the surface and medium, and type of adsorption complexes. The multifactorial
nature of competitive adsorption (balance of thermodynamic and dynamic factors, textural effects of the adsorbents, and effect
of the medium) in static and dynamic adsorption may lead to very different results, for example, for monomers and their
___________________________________________________________________________________________________
A. A. Chuiko Institute of Surface Chemistry, National Academy of Sciences of Ukraine, Vul. Generala Naumova, 17,
Kyiv 03164, Ukraine. E-mail: gun@voliacable.com. Translated from Teoreticheskaya i Éksperimental’naya Khimiya,
Vol. 43, No. 3, pp. 133-169, May-June, 2007. Original article submitted June 4, 2007.

0040-5760/07/4303-0139 ©2007 Springer Science+Business Media, Inc. 139


corresponding polymers. We should note that the effect of the conditions of both equilibrium (static) and nonequilibrium
(dynamic) adsorption is greater in competitive adsorption since the adsorbates may differ considerably in their rate of diffusion,
kinetics and free energy of adsorption, nature of interaction both with the surface (monodentate, bidentate, and polydentate
adsorption) and the medium, and in their concentration. On the whole, competitive adsorption leads to a reduced rate of
reaching equilibrium. Preadsorption or a higher diffusion rate gives a momentary advantage to the first or more mobile
adsorbate, which is most evident in dynamic adsorption and less so in equilibrium adsorption. The time to reach equilibrium in
the case of preadsorption or concurrent competitive adsorption may be different but it is greater than in the individual
adsorption of the same compounds.
There are two approaches to the description of competitive adsorption involving additive and nonadditive models. In
the former case, the adsorption of the i-th adsorbate is seen as the adsorption of an individual compound taking account of its
partial pressure or concentration, in which the parameters of the adsorbent such as adsorption capacity are redetermined taking
account of the adsorption of other adsorbates. We use fitting or additional parameters such as activity coefficients in these
models [11]. In the case of nonadditive models, adsorption is described by a system of related equations, determining the
corresponding parameters using different variants for uncoupling equations and iterative self-consistency of their solutions.
Special program packages have been developed for analyzing the data for multicomponent mixtures. Thus, for example, the
TMVOC package [25] (similar to the TOUGH2 or T2VOC packages) was developed at the Berkeley National Laboratory in
the United States for modeling systems involving a nonisothermal water flow, soil gas, and a mixture of volatile organic
compounds in a heterogeneous porous medium. The 3DHYDROGEOCHEM package was developed by the Environmental
Protection Agency in the United States for analogous purposes. These packages are applicable both for “natural”
environmental conditions and engineering systems.
On the whole, there have been many studies of multicomponent adsorption but nevertheless less than for the adsorption
of individual compounds and many of these works have been more practical than theoretical although many aspects of the
theory of multicomponent adsorption require further development. Thus, in the present work, we analyzed some aspects of
concurrent adsorption and their theoretical description and determined features of the corresponding phenomena on phase
separation boundaries. Furthermore, we demonstrated the usefulness of the regularization method, which permits us to solve
the integral equations describing both static and dynamic multicomponent adsorption. It is impossible to encompass all the
aspects of multicomponent adsorption in a short review and, thus, some areas in this field are only briefly mentioned to show
the wide variety of phenomena, in which concurrent adsorption is manifest.

1. STRUCTURAL AND ADSORPTION CHARACTERISTICS OF


ADSORBENTS AND COMPETITIVE ADSORPTION

The results of both dynamic nonequilibrium and equilibrium static multicomponent competitive adsorption depend
significantly on the structural and adsorption characteristics of the adsorbents. Various classifications exist for adsorbents
relative to shape, pore type and size, and the corresponding adsorption isotherms such as the IUPAC nomenclature based on the
classification of Dubinin [3, 4, 19-22]. These classifications encompass from three to seven morphological types of pores such
as intraparticle, interparticle, interaggregate, and intercluster pores, which differ also in their shape: cylindrical, slit, conical,
spherical, bottle-like, voids between spherical, cylindrical, and other particles, pores with disordered structure, etc. For the
purposes of this work and the features of competitive adsorption, adsorbents may be divided into three major morphological
groups.

Adsorbents Consisting of Particles with


Internal Micropores and Mesopores

Materials with the above porosity include activated microporous charcoals (AC), mesoporous silica gels, mesoporous
sieves, zeolites, and porous polymers [3, 4, 19-22]. We should note several features of these materials such as high specific
surface (up to 3000 m2/g), narrow pores with diameter d < 2 nm with slit, cylindrical, or other forms, high adsorption potential
even for nonpolar compounds (for example, the change in the Gibbs free energy in the adsorption of nitrogen in these pores

140
Fig. 1. Pore size distribution functions for materials of various
morphology. First group: 1) microporous activated charcoal (AC),
2) MCM-41 silica, 3) Si-40 silica gel, 4) copolymer of styrene and
divinylbenzene LiChrolut. Second group: 5) pyrogenous silica
A-500, 6) graphitized carbon black (PSD´10). Third group: 7) AC,
8) mordenite (PSD´10) and Amberlite XAD-16 polystyrene. The
calculations were carried out using the density functional theory for
models of slit pores (AC), cylindrical pores (MCM-41, Si-40,
polymers, and mordenite), and voids between spherical particles
(A-500 and carbon black).

–DG > 8-10 kJ/mol and –DG < 1 kJ/mol for capillary condensation in mesopores), the existence of mesopores with d = 4-50 nm
(especially, narrow mesopores with d < 10 nm), the absence or very minor contribution of macropores (d > 50 nm) (Fig. 1),
significant adsorption volume (0.3-3 cm3/g), the existence of functional groups, primarily oxygen-containing groups, as
adsorption sites, relatively high mechanical particle strength, chemical inertness (silica) or adjustable catalytic activity
(aluminum oxides, titanium oxides, and mixed oxides), apparent density from 0.01 (aerogel) to 0.5 g/cm3 or higher, and
adjustable particle morphology [1-6, 19-22]. The time for the establishment of equilibrium in the multicomponent adsorption
of complex compounds such as polymers on such adsorbents may be several days since the redistribution of adsorbates with
different molecular weights in narrow pores (Fig. 1), which are initially occupied by the most mobile adsorbate, is hindered at
high concentrations. The establishment of equilibrium in the multicomponent adsorption of low-molecular-weight compounds
proceeds more rapidly and requires minutes or tenths of minutes depending on the pressure in a gas mixture or concentration in
solution.

Highly Disperse Materials Consisting of


Nonporous Primary Nanoparticles

Characteristic representatives of this morphological group include pyrogenous nanooxides, blacks, aerosols, and
polymer powders. Most of these materials are the product of high-temperature processes, which accounts for the lack of pores
in the primary particles. For example, individual (SiO2, TiO2, Al2O3, etc.) and mixed nanooxides (Al2O3/SiO2, SiO2/TiO2,
Al2O3/SiO2/TiO2, etc.) are obtained by the high-temperature hydrolysis of chlorides of the corresponding elements (or other
precursors such as organometallic compounds) in a hydrogen–oxygen flame. The high synthesis temperature (³1300 K) results
in very similar morphology of the nonporous pyrogenous oxide nanoparticles with different chemical structure. The structural
and adsorption characteristics of nanooxides depend on the particle size distribution of the primary particles j(d)

141
(diameter d = 3-100 nm), which determines the specific surface (S » 50-500 m2/g), aggregate structure (50-1000 nm) of the
primary particles and agglomerates (>1 µm), which determines the textural porosity parameters (Fig. 1), significant mass
fractal index (2.2-2.6), low bulk density (0.04-0.2 g/cm3), structure of surface of the primary particles, distribution of the phases
on the surface in the case of mixed oxides, their hydrophilicity due to various OH groups [26], existence of Brönsted and Lewis
acid sites, Lewis base sites, as well as linked functional groups, and the amorphous nature or crystallinity of the primary
particles or phases in mixed oxides [48]. The first two characteristics are virtually independent of the chemical composition of
the pyrogenous oxides and are mainly a function of the synthesis conditions, primarily, the synthesis temperature. With
increasing size of the primary particles, a trend is noted toward broadening of the j(d) and a reduction in the aggregation of
nanooxides, which affects the adsorption volume, behavior of the particles in the liquid medium, and nature of their interaction
with macromolecules and supramolecular structures and concurrent adsorption on their surface. Further control of the
adsorption properties of nanooxides is possible through creation of a rough surface of the nanoparticles (i.e., fine pores) during
the synthesis, which increases the adsorption potential and adsorption volume relative to low-molecular-weight compounds
(for example, the adsorption of water from the atmosphere is increased by a factor of 3-4). Chemical modification of the surface
by inorganic, heterorganic, and organic compounds permits us to control the structural, adsorption, and other properties of
nanomaterials by means of controlled functionalization [3, 5, 48, 49]. The time for establishing equilibrium in multicomponent
adsorption on disperse materials is much less than in the case of adsorbents of the first morphological group since the primary
particles are nonporous, the “pores” or cavities between the primary particles in their aggregates (Fig. 1) are short, and the ratio
of the radius of the pores to their length (R/h) is several orders of magnitude greater than for microporous adsorbents. This
circumstance leads to rapid adsorption and rather rapid equilibration of the adsorption phase both in the case of
single-component adsorption and in the case of multicomponent adsorption.

Materials with Internal and Textural Porosity

Materials of the above type include natural clays and zeolites, micro/wide-porous carbon adsorbents, nanotubes,
specially synthesized silicas, hybrid carbon–mineral adsorbents, and polymers. The major feature of such materials is true
porosity of the primary particles and the existence of significant cavities in the secondary particles. These materials may have
all the properties of the materials of the first two morphological groups. For example, they may have a broad pore size
distribution (PSD), including peaks characteristic for adsorbents of the first two groups (Fig. 1). In many cases, these materials
are the most efficient adsorbents superior in their parameters to representatives of the first two groups since the existence of
micropores provides for high adsorption potential for low-molecular-weight compounds, while the existence of textural
porosity provides for the adsorption of high-molecular-weight compounds and the rapid transport of low-molecular-weight
compounds in the micropores [1-6, 19-22]. The latter property is especially significant in the dynamic adsorption of toxic
substances from air or water [49]. The time to establish equilibrium may be somewhat less than for materials of the first group
but significantly greater than for the second group as a consequence of the existence of significant microporosity (Fig. 1).
In light of the structural features of the adsorbents, layered filters are often used consisting of morphologically and
chemically different materials such as activated charcoal and zeolites. This approach permits us to minimize the size of the
filters and enhance the efficiency of the adsorbents, for example, in the cyclic separation of mixtures such as H2/CO2/CH4/CO
[50] and in the purification of water. We should note that a discontinuity is observed on the boundary between two adsorbents
in the rate of propagation of an adsorbate “wave” and these changes may lead either to an increase or decrease. This, in turn,
leads to the appearance of transverse waves in the layered filters, which may produce deleterious results. Models developed
taking account of the differences in the interaction of the mixture components with various adsorbents permits the design of an
optimal layered structure corresponding to the flow composition and rate [50]. Goj et al. [51] carried a theoretical study of the
effect of size and shape of zeolite pores on the adsorption of CO2 and N2. Lu et al. [52] offered a model for the adsorption and
separation of ternary and quaternary mixtures of short linear alkanes on zeolites. Experimental and theoretical methods were
used to study the adsorption and diffusion of mixtures of hexane and 2-methylpentane [53]. The diffusion of mixtures of
adsorbates in the case of concentration fluctuations was studied in nanoporous adsorbents [54]. Chen and Sholl [55]
investigated the reliability of the ideal adsorbed solution theory (IAST) using a Monte Carlo model of multicomponent
adsorption. Significant attention has been given to the adsorption, separation, concentration, and accumulation of light

142
Fig. 2. Incremental pore size distribution (DFT calculation). The hatched region is for pores
with d £ 2 nm (a); binary adsorption isotherm for CO2 (2) – N2 (1) at 298 K and 1.05 bar on
activated charcoal, 3) total adsorption (b) (Fig. 2b is adapted from Elsayed et al. [66]).

hydrocarbons, H2, CO2, CO, NOx, and SO2 mainly in carbon adsorbents [56-60] and zeolites [61-64] since the efficient
adsorption of these compounds requires significant microporosity. The adsorption of gases such as N2, O2, H2, CO, CO2, and
CH4 proceeds efficiently under normal conditions or even at elevated pressures only in narrow pores with half-width not
exceeding two diameters of the adsorbate molecules. Thus, considering the size of these molecules, the range of dimensions of
efficiently adsorbing pores corresponds only to micropores. As noted above, significant microporosity is found for activated
charcoals, some porous polymers, and zeolites (Fig. 1) [1-6, 19-22]. Wu et al. [56] studied the adsorption equilibrium of a
methane/nitrogen/hydrogen mixture on an activated charcoal adsorbent. Choi et al. [57] studied the adsorption of a more
complex methane/ethane/ethylene/nitrogen/hydrogen mixture on activated charcoal, while Nam et al. [61] investigated the
adsorption of the same mixture on zeolite 5A. We should note the role of the adsorption of water vapor in such processes [58].
Multicomponent adsorption has been studied not only by adsorption methods but also by desorption methods (TPD MS and
thermogravimetry) [59]. Important information is obtained using kinetic methods and by studying thermal effects [1, 2, 60] and
spectral methods [65].
In the case of mixtures such as N2/CO2, competitive adsorption may proceed almost at the same sites in activated
charcoal [66]. However, as a result of differences in the electronic structures of molecules of N2 and CO2, their polarity, and
polarizability, CO2 molecules may interact more efficiently with the polar surface groups such as COH, and COOH. The N2
molecule is smaller than the CO2 molecule. Thus, some of the narrow pores in AC are available only for N2 molecules. These
circumstances lead to a larger total adsorption of N2 and CO2 (Fig. 2), which is greatest at about 75% CO2 in the mixture, than
for the individual compounds [66]. A molecular sieve adsorbent has been developed with pore diameter of ~0.4 nm for the
selective adsorption of CO from mixtures without noticeable adsorption of other gases such as halogens [63]. The competitive
adsorption of CO2 and water from the air was studied on a layered adsorbent consisting of alumina and zeolite [64]. A method
was used in the adsorption of CO2 from a gas flow (with 10% CO2 and higher) with periodic alteration of the pressure and
temperature for the preliminary elimination of moisture prior to adsorption [67]. Complex filters consisting of different
adsorbents are used for the efficient adsorption of multicomponent systems [68]. A method has been developed for the thermal
regeneration of zeolite CaX and SrX adsorbents in the adsorption of compounds, which may react with the adsorbent surface at
the regeneration temperature. A hot nitrogen gas flow is used so that the adsorption process is rapid, permitting reduction of the
passage time to less than 1 s [69] and in the probability of deleterious reactions. A method for the separation and extraction of
carboxylic acids from a gaseous mixture containing water vapor is based on oscillating adsorption and separation on a complex
alumina–zeolite adsorbent featuring adsorption, washing, desorption, and introduction of the gas flow at elevated pressure
[70]. A study was carried out on the adsorption of multicomponent aqueous solutions of organic compounds on activated
charcoals and the selective adsorption of organic homologs from dilute aqueous solutions on heterogeneous adsorbents [6].
Various approximations were made in analyzing the data such as a solvophobic interaction and the theory of congruity and the
effect of the adsorbent structure on multicomponent adsorption was studied [6].

143
The theoretical description of multicomponent and multiphase adsorption is much more complicated than in the case of
the adsorption of individual compounds due to the heterogeneity of the mixtures, competitive adsorption, nonadditive effects,
heterogeneity of the adsorbent surface, broad PSD, and complications due to lateral interactions. Let us examine some of the
theoretical approaches to the description of competitive adsorption in greater detail.

2. THEORY OF COMPETITIVE ADSORPTION

The Langmuir additive model is often used to describe the adsorption of compounds A and B under isothermal
conditions [1-3]. The rate of adsorption of molecules of A (rA,a) is proportional to the partial pressure pA and number of
unoccupied adsorption sites (1 – qA – qB)

rA,a µ (1– qA – qB)pA. (1)

Since the rate of desorption rA,d µ qA at equilibrium is equal to the rate of adsorption

(1– qA – qB)bApA = qA (2)

where bA is a parameter dependent on the free energy of adsorption of A and analogously for B. After straightforward
transformations of the equations for the two adsorbates, we obtain for adsorbate A

bA p A
θA = . (3)
1 + bA p A + bB p B

The analogous equation describes the adsorption of B. Let us generalize Eq. (3) for N adsorbates:

bi p i
θi = .
N (4)
1+ ∑bj pj
j =1

Parameters bi are functions of the structure–adsorption characteristics of the adsorbent. Assuming that the pore size distribution
f(R) is a known function (for example, a function calculated using the low-temperature adsorption of nitrogen or argon), we
give the adsorption of the i-th compound averaged over pore size as:

R max
bi ( R ) p i
θi = ∫ N
f ( R )dR.
(5)
R min 1+ ∑ b j (R ) p j
j =1

Furthermore, parameters bi are functions of the free energy of adsorption (DG), which depends on the size of the pores, in which
adsorption occurs, and lateral interactions (z), which depend on the partial pressure of the adsorbates (pi). For simplicity, we
assume that there is only one type of adsorption sites. If this assumption is rejected, a function of the distribution of adsorption
free energy ji(DG) for each adsorbate dependent of the PSD and then,

 − ( ∆Gi ( R ) + z i ( R )) 
R max ε max
p i exp 
 Rg T 
θi = ∫ ∫ N  − ( ∆G j ( R ) + z j ( R )) 
ϕ i ( ∆G ) f ( R )d ( ∆G )dR
(6)
1 + ∑ p j exp
R min ε min

j =1  Rg T 

144
where Rg is the gas constant. Equation (5) may be given in another form in order of increasing molecular size of the adsorbates
in some series [39]:

a i +1 ai + 2
(i +1)
θi = ∫ θ i ( p1 , p 2 ,K , p i ; R ) f ( R )dR + ∫ θi ( p1 , p 2 ,K , p i +1 ; R ) f ( R )dR + K +
(i)

ai a i +1

a max
+ ∫ θi ( p1 , p 2 ,K , p i ,K , p N ; R ) f ( R )dR
(i)
(7)
aN

where i = 1, 2,…, N and ai is the minimum half-width of the pores, into which the i-th adsorbate may penetrate, and

bi ( R ) p i
θ iM ( p1 , p 2 ,K , p M ; R ) = M (8)
1+ ∑ b j (R ) p j
j =1

where M £ N, M = k, k+1,…, N. The first term in Eq. (7) corresponds to the contribution of the i-th compound to adsorption in
the pores available for compounds from 1 to i-th and the last term corresponds to adsorption in the pores accessible to all
compounds.
Equations (5)-(7) correspond to a nonadditive model for multicomponent adsorption and are poorly suited to the
problem since slight fluctuations in the experimental data caused, for example, by white noise, lead to considerable variations
in the distribution functions. Thus, direct minimization procedures or transformations are not applicable for such equations,
which are solved using various regularization procedures including the method of self-consistent regularization, developed for
solving integral equations involving several distribution functions [48, 49, 71]. In calculations with such equations, we should
use the maximum entropy method (MEM) [71]. In this case, an additional regulator (VAR + VARMEM) may be given as a vector

r r
S ( p( f )) 
2
VAR + α  1−  ® min, (9)
 S max 

where a is a regularization parameter and S is the entropy,

r r r r r
p 0 ( f ) = f , p1i ( f ) = f i +1 − f i + ( f max − f min ), i = 1, …, N – 1,

r r
p i2 ( f ) = f i +1 − 2 f i + f i −1 + 2( f max − f min ), i = 2, …, N – 1,

r N fk  1
S ( f ) = − ∑ sk ln( sk ), sk = , S max = − ln   .
N N (10)
k =1
∑ fk
k =1

r r
Vector p j ( f ) corresponds to MEM of the j-th order. This approach was used for modifying the CONTIN/MEM-j algorithm
[71].
In a model for multicomponent adsorption taking account of heterogeneity of the adsorbed phases (multispatial
adsorption model, MSAM), we used the isotherms of the individual adsorbates and parameters describing the adsorbent taking
account of the temperature dependence [72]. We should note the elegant work of Tvardovski et al. [40], in which the authors

145
used a phenomenological thermodynamic model and equation of state, in which pressure was given as a function of the
isosteric heat of adsorption taking account of the compressibility factor and a correction for unoccupied volume, and proposed
an equation for multicomponent adsorption for the adsorbed phase including the Langmuir, Fowler–Guggenheim, Temkin, and
BET equations, generalized for mixtures of adsorbates, as special cases. However, some difficulties may arise in the practical
application of this approach in determining the values of the parameters entering the equations.
Prediction of the multicomponent adsorption equilibrium using the data for single-component adsorption is one of the
outstanding problems in adsorption. The lack of precise and consistent experimental data in broad temperature and adsorption
ranges is one of the major hindrances to further developments in this field and verification of the corresponding theories [73].
Detailed adsorption calorimetry studies of binary and ternary systems including CO2, C2H4, and C2H6 on zeolite adsorbents led
to the calculation of the activity coefficients, change in free energy, enthalpy, and entropy of adsorption using a three-parameter
equation and taking account of the nonideal nature of the systems [41, 73]. A good correlation was established between the
thermodynamic functions for multicomponent systems and the parameters of individual adsorption.
The excess free energy for binary adsorption is given by the sum [41, 73]

NC NC
ge = ∑ ∑ ( A ij + B ij T )xi x j (1− exp( −C ij Ψ ) (11)
i =1 j =1

where j > i, Aij, Bij, and Cij are equation parameters, and xi and xj are the mole fractions of the adsorbates;

p
 ∂ ln f 
a
Φ
Ψ=− = ∫ ad ln f = ∫   da, (12)
Rg T 0  ∂ ln a 
0 T

T is the constant temperature, F is the adsorption potential, f is the volatility, and a is the amount of adsorption. The differential
enthalpy of desorption (isosteric heat) for the i-th component in the case of multicomponent adsorption of a real gas and
nonideal adsorbed solution is given as [73]

 2  ∂a
−e  
 ∑ x j G j a j ( ∆h j − ∆h j ) + R g T  
0 0 0 0

 ∂ ln γ i   1  ∂ ln γ i   j  ∂T  Ψ, x 
∆hi = ∆hi0 + R g T 2   +  0 +    (13)
 ∂T  Ψ, x  a i  ∂Ψ  T , x    ∂a − e  
 ∑ x j G j −  ∂Ψ 
0

 j T,x 

where ∆h 0j is the differential enthalpy for individual adsorbates, gi is the activity parameter [73],

 ∂ ln γ i  A
  =− (1− exp( −CΨ ))x 2j for i ¹ j; (14)
 ∂T  Ψ, x Rg T 2

1  ∂ ln( a 0j ) − e 
G 0j ≡ +   . (15)
( a 0j ) 2  ∂ ln f j0 
T

For a binary mixture [73]

1  x1G1 a1 ( ∆h1 − ∆h1 ) + x 2G2 a 2 ( ∆h2 − ∆h2 ) 


0 0 0 0 0 0 0 0
∆h1 = ∆h10 +  . (16)
a10  x1G10 + x 2G20 

146
Fig. 3. Comparison of the experimental data for equilibrium adsorption and values calculated
using the IAST (dotted lines), HIAST (solid lines), and HEL models (dashed lines) for binary
ethane (2)–propane (1) mixtures adsorbed on Norit activated charcoal (303 K) (a) and
ethylene (1)–CO2 (2) mixtures adsorbed on Nuxit activated charcoal (293 K) (b); Y1 is the
mole fraction of compound 1 in the gas phase (adapted from the work of Wang et al. [74]).

Wang et al. [74] compared the results of calculations for multicomponent adsorption in the framework of three
different approaches: IAST, heterogeneous ideal adsorbed solution theory (HIAST), and the heterogeneous expanded
Langmuir model (HEL). HIAST was shown not always to give the better results than IAST (Fig. 3), which depends on the
adsorption energy distribution used. The observed concentration of the adsorbed phase in the HIAST and HEL models
corresponds to the integral of the local isotherm


C µ ( k ) = ∫ C µ [ k , E ( k )]F [ k , E ( k )]dE ( k ) (17)
0

1
where F[k, E(k)] = for Emin < E < Emax and F = 0 for other values, E(k) is the energy of local interactions of
E max ( k ) − E min ( k )
the k-th compound, and Cµ[k, E(k)] is the local concentration of the adsorbed phase on sites with a given energy calculated
using IAST for an individual component (Langmuir isotherm) (model 1) or using an expanded multicomponent Langmuir
equation (model 2). The local Langmuir equation for the individual k-th component has the form

 E (k )
b0 ( k )C ( k )exp 
 Rg T 
Cm[k, E(k)] = C µs ( k ) (18)
 E (k )
1+ b0 ( k )C ( k )exp 
 Rg T 

where b0 is the affinity constant at the zero energy level and C is the concentration of the gas phase.
The modified Monte Carlo (MC) method was used to model the isotherms of two-component adsorption of molecular
hydrogen isotopes on carbon nanotubes (10,10) [75]. Individual adsorption obeys Henry’s law at low and medium
occupancies. The selectivity of the adsorption of T2 relative to H2 increases with increasing pressures up to saturation of the
nanotubes. The results of the MC modeling and IAST calculations were in good accord at low and medium occupancies. At
higher occupancies, the MC calculations gave almost constant selectivity, while IAST predicts a constant increase in selectivity
[75].
The dynamics of adsorption of a gas mixture, from which only one gas is adsorbed efficiently on the porous adsorbent,
is accompanied by a spontaneous temporary increase in pressure in the pores (Fig. 4) [76]. Construction of a model for the

147
Fig. 4. Experimental dependence of the relative pressure on
elapsed time in the transport of a ternary system H2 ®
(C3H8 + N2) (the arrow indicates that hydrogen displaces the
two other components, which causes desorption of propane)
in the pores of a commercial hydrogenation catalyst ICI 52/1
for C3H8/N2 mole ratio = 0 (1), 0.5 (2), and 1 (3) using a
cylindrical pellet of ICI 52/1 (CuO+ZnO/Al2O3) with
height/diameter = 3.81/5.50 mm. The mean pore diameter
was 16.4 nm. The specific surface S = 73 m2/g. The pore
volume was 0.377 cm3/g and the porosity was 0.604
(adapted from the work of Hejtmanek et al. [76]).

changes in pressure using the basic Maxwell–Stephan equation shows that, upon equilibrium adsorption, the predicted pressure
exceeds the temporary dynamic pressure within the pores. The dusty gas model (DGM) and mean transport pore model
(MTPM) based on the Maxwell–Stephan theory include contributions of the bulk and Knudsen diffusion and a penetrating flow
(taking account of viscous and Knudsen flows). The vector form for the relationship between the molar densities of the flow
N = {N1, N2,…, Nn}T and molar concentration gradients c = {c1, c2,…, cn}T is the same for both models:


H(c)·N + c=0 (19)
∂x

1 ciα i n cl
where H(c) is a square matrix (n´n) with elements hij dependent on the concentration hii = + +∑ ,
Dik Dik l =1 cDil
ciα i ci
hij = − for i ¹ j; Dij = YDij is the effective binary bulk diffusion coefficient, Dij is the binary bulk diffusion
D kj cDij

2 8R g T
coefficient, Dik = rΨ is the effective Knudsen diffusion coefficient, Y is a geometric model parameter (effective
3 πM i
porosity of the transport pores), c is the total molar concentration of the gas mixture, ci is the molar concentration of the i-th
component, ai is a parameter dependent on diffusion, concentration, and flow in the pores of the i-th component, n is the
number of mixture components, x is the coordinate along the axis of the adsorbent granule, and r is the mean radius of the
transport pores. The matrix elements hij depend on the transport properties of the individual gases and their mixture, pore
structure characterized by parameters Y, r Ψ (effective radius of the transport pores), and r 2 Ψ (effective viscous flow
parameter) [76].
The mass balance for an n-component gas mixture for m (m £ n) adsorbed gases in the pore space is described by the
equation

148
r r r
∂c ( t , x ) ∂a ( t , x ) ∂N ( x, t )
ε +ρp =− (20)
∂t ∂t ∂x

r
where e is the total granule porosity, rp is the apparent granule density, a is a vector describing the adsorption of all
components, and t is the time. In the case of adsorption equilibrium, which is described by the HEL model, we have a kinetic
equation

 m  
a max K i   1+ ∑ K j c j  δ ij − K j c i 
∂a i m ∂a ∂c m   j =1   ∂c j
=∑ i =∑
j
(21)
∂t j =1 ∂c j ∂t j =1  m 
2 ∂t
 1+ ∑ K j c j 
 
 j =1 

where i = 1, …, m, dij are Kronecker symbols, and Ki is the adsorption equilibrium constant of the i-th component. This equation
may be given as

∂a i   m a 
= k i  c i  a max − ∑ a j − i   (22)
∂t   Ki 
j =1 

where ki is the adsorption rate constant of the i-th component.


Let us examine some characteristic examples of multicomponent adsorption and examples of the application of theory
to some real and model systems.

3. ADSORPTION FROM A GASEOUS MEDIUM

The experimental differential enthalpies (isosteric heats) of the binary adsorption of CO2 and C2H6 on zeolite NaX
exceed the ideal values calculated using the IAST model (Fig. 5), which is in accord with a negative deviation from ideal
behavior for the selectivity parameter (Fig. 6) [73]. The IAST error increases with increasing adsorption. Despite systematic
errors (~2 kJ/mol) of the calculations according to the ABC expression (Eq. (11)), the agreement between the calculated and
experimental data is quite suitable considering that only one fit parameter is used for the binary mixture. The mean absolute
error in calculations of the selectivity (Fig. 6) was 12%, while the mean error in the enthalpy calculations was ~2 kJ/mol (Fig. 7)
[73]. On the whole, the multicomponent equilibrium may be predicted with accuracy of only 10% using the data for the
corresponding binary systems, which, however, does not guarantee the absence of large errors in the case of strong
intermolecular interactions in complex systems [41, 73]. Analysis of the excess functions (Fig. 8) is useful for evaluating the
deviation of complex systems from ideal behavior. The destabilizing entropy contribution (Tse = he – ge) is relatively small
(Fig. 7). In the case of a constant adsorption potential, the excess function curves should be symmetrical. However, we find a
displacement of the minimum toward the component with greater adsorption (greater adsorption potential). The relative
magnitudes of the deviations of the adsorbed phase from ideal behavior (Fig. 9) show that the adsorption enthalpies for systems
CO2–C3H8 and CO2–C2H6 on NaX are similar despite considerable differences in the free energy of adsorption. This
discrepancy may be attributed to the circumstance that the energy effects are similar upon the adsorption of a mixture of polar
and nonpolar adsorbates on a heterogeneous adsorbent, while the entropy effects differ significantly due to differences in the
molecular structure of these adsorbates. According to Myers [41, 73], multicomponent adsorption can be predicted using
single-component isotherms and adsorption enthalpies as well as potential parameters obtained in single-component
adsorption (Fig. 10) and molecular modeling.

149
Fig. 5. Experimental and calculated values of the differential enthalpies of
adsorption (isosteric heats) for CO2 (1-3) and C2H6 (4-6) adsorbed on zeolite NaX:
1), 4) experimental data, 2), 5) calculation using Eq. (13), 3), 6) calculation using
Eq. (16) and IAST (adapted from the work of Siperstein [73]).

Fig. 6. Comparison of the experimental selectivity (points) for the ternary


CO2 (1)–C2H4 (2)–C2H6 (3) system adsorbed on zeolite NaX with selectivity
calculated using Eq. (11) (solid lines), sij = (xiyj)/(xjyi); x and y are the mole
fractions of the components in the adsorbed state and in the gas phase (adapted
from the work of Siperstein [73]).

A model for multicomponent adsorption based on the vacant solution theory and the Dubinin–Polanyi theory has been
used to described the adsorption of CH4/CO2 and CH4/C2H6 mixtures on a charcoal adsorbent [77]. The predictive power of the
model may be enhanced by using equations with various activity coefficients. The model developed permits us to predict the
results of multicomponent adsorption at different temperatures using data on the adsorption of the individual compounds
obtained at a single temperature.
The equilibrium and dynamic adsorption of the CO2/CH4 mixture on a carbon fiber adsorbent (ACF) was compared
with the adsorption of individual CO2 [78]. The electrooscillatory adsorption/desorption at 30 V was studied at atmospheric
pressure. The individual adsorption of CO2 (~40 mg/g) was approximately double the adsorption of CH4. However, in the case
of two-component adsorption, this ratio rose to 5.2. The application of a voltage caused heating of ACF to 200 °C over 30 s,
which provided for the efficient desorption of CO2 since desorption occurs efficiently even at about 60 °C (i.e., at a lower
voltage). This method of desorption proved 20% more efficient than the vacuum method. Wang [23] carried out a detailed
study of the thermal regeneration of multicomponent adsorbents. Modeling the cycles of nonisothermal multicomponent
adsorption permits a deeper understanding of the mechanism of separation of the gaseous components [24, 79, 80].

150
Fig. 7. Comparison of experimental differential enthalpies of adsorption for the
ternary CO2 (1)–C2H4 (2)–C2H6 (3) system upon preadsorption of CO2 (a) or
C2H4 (b) on zeolite NaX with theoretically calculated values as functions of the total
adsorption (adapted from the work of Siperstein [73]).

Fig. 8. Isothermal (295 K) and isobaric (13.3 kPa) values of excess enthalpy (2, 4) and
free energy (1, 3) in the adsorption of a) CO2–C3H8 (1, 2) and CO2–C2H6 (3, 4) on NaX,
b) C2H4–C2H6 (1, 2) on NaX and SF6–CH4 (3, 4) on silicalite, xi is the mole fraction of
the first component in the adsorbed phase (adapted from the work of Siperstein [73]).

Fig. 9. Dimensionless value of the excess free energy as a function of the occupancy q
for an equimolar mixture (xi = 0.5): A) CO2–C3H8 on NaX, B) CO2–C2H6 on NaX,
C) C2H4–C2H6 on NaX, D) SF6–CH4 on silicalite, E) CO2–C2H4 on NaX (adapted
from the work of Siperstein [73]).

151
Fig. 10. Correlation between the constant A0 = A + BT and
∆h 0 − ∆h20 ∆h2s V c 2
the term z = 1 s
where Dhs is the molar
Rg T ∆h1 V c1
integral enthalpy of adsorption and Vc is the critical gas
volume in adsorption on zeolites NaX (1), NaA (3), and
silicalite (2) (adapted from the work of Siperstein [73]).

Fig. 11. Adsorption isotherms (303 K) from a mixture of


30% ethane (2) and 70% propane (1) on a carbon
adsorbent (adapted from the work of Wang [39]).

The adsorption of light hydrocarbons is more efficient on microporous adsorbents since, as noted above, if the pore
diameter exceeds the molecular dimensions of the adsorbate by more than four times, adsorption in such pores is low. The ratio
of the molecular dimensions in a homologous alkane series determines the ratio of values of the equilibrium adsorption on any
adsorbents since the disperse interaction making the major contribution to the energy of interaction of nonpolar compounds
with any surface increases with increasing dimensions. Thus, for example, when the propane/ethane ratio in the mixture is
approximately 2, the propane/ethane adsorption ratio is approximately 10 (Fig. 11) [39].
CO2 is selectively adsorbed on cyanide–palladium–cobalt gels synthesized from aqueous solutions of Na2PdCl4 and
K3[Co(CN)6] [81]. Cyanide bridges link palladium and cobalt in the lattice. Hence, these adsorbents were termed cyanogels.
The thermodynamics and kinetics of adsorption and desorption of CO2 showed that cyanoaerogel has about twice the
adsorption capacity as the cyanoxerogel. The adsorption of CO2 is completely reversible on both adsorbents. The experimental
results showed that the adsorption proceeds molecularly and obeys a first-order adsorption equation. Under the conditions
studied, SF6 and N2O are adsorbed on cyanoaerogel but gases such as N2, Ar, O2, and CH4 are virtually not adsorbed on both
adsorbents. An increase in the CO2 pressure from 1 to 3 atm leads to a decrease in the number of vacant sites on the xerogel from
10-15% to 5-7% over 3 h adsorption [81].

152
Fig. 12. Experimental (points) and calculated dependence (solid lines) of the pressure
on elapsed time for binary systems in the adsorption (1, 2) C2H4 ® N2 (a) and
C3H8 ® N2 (b) and desorption (3, 4) N2 ® C2H6 (a) and N2 ® C3H8 (b) on
CuO+ZnO/Al2O3 catalyst (adapted from the work of Hejtmanek [76]).

Fig. 13 Fig. 14

Fig. 13. Experimental (points) and calculated (solid lines) dependence of the pressure on
elapsed time in the adsorption of the mixture (C3H8+N2) ® H2 (C3H8/N2 = 0.25/0.75) taking
account of the adsorption kinetics of propane (the adsorption rate constant k = 10.9 cm3/mol·s)
on the CuO+ZnO/Al2O3 catalyst (adapted from the work of Hejtmanek [76]).

Fig. 14. Experimental (points) and calculated (solid lines) relative volumes of components CH4
(1, 2), N2 (3, 4), and CO2 (5, 6) in a flow at the exit of an activated charcoal adsorbent (mean
granule diameter about 0.25 mm, porosity 0.37, the adsorbent was packed in a 25-cm-long
cylinder with diameter 4.25 cm in the displacement of adsorbed methane by a 46/54 CO2/N2
mixture dependent on the injection volume of this mixture (pressure 4.1 MPa, T » 303 K) with
constant inlet rate 0.5 cm3/min (adapted from the work of Tang [83]).

There are very considerable expenses in the chemical industry for the separation of desired products, side products, and
unreacted starting materials [82]. Several variants have been developed for the extraction at equilibrium of vapor/liquid,
liquid/liquid, solid/liquid, and vapor/solid phases. Thermodynamic models have been developed such as the ge model,
equations of state, and group contribution method for multicomponent systems, in which only binary experimental data are
used. The group contribution method is based on an approximation, in which the system is seen as consisting not of molecules
but of functional groups. This method is convenient since the number of functional groups in various organic adsorbates is

153
much less than the number of compounds [82]. Corresponding databases have been developed, which are readily used to
analyze the enormous number of compounds on the basis of parameters for several dozen groups.
The agreement between the calculated and experimental data for multicomponent adsorption is often unsatisfactory.
This has been seen, for example, in the data on the dependence of the pressure on elapsed time for binary systems in the
adsorption of C2H6 ® N2 and C3H8 ® N2 on the ICI 52/1 hydrogenation catalyst (CuO+ZnO/Al2O3) and the desorption
N2 ® C2H6 and N2 ® C3H8 (Fig. 12) although the shapes of the theoretical and experimental curves are similar [76]. A possible
reason for the observed deviation is capillary condensation of the organic compounds in mesopores. Deviation in the case of
inert gases is less. The inclusion of surface diffusion of the adsorbed component specifies prediction of a more significant
change in pressure than observed experimentally. The most reliable predictions were obtained upon inclusion of the adsorption
kinetics into consideration (Fig. 13) [76].
A comparison of the experimental data and calculations of transport dynamics and the adsorption of multicomponent
mixtures on carbon adsorbents was carried out by analyzing the accumulation and extraction of methane [83]. The agreement
of the calculated and experimental values is qualitative in nature (Fig. 14). We should note that the introduction of a correction
taking account of the adsorption behavior of nitrogen permits a significant improvement in the agreement between the
theoretical and experimental data both relative to methane and nitrogen.

4. ADSORPTION UNDER CRITICAL AND


SUPERCRITICAL CONDITIONS

The accumulation of gases such as H2 and CH4 in microporous adsorbents is carried out under high pressure [27,
84-86]. Description of the adsorption of multicomponent mixtures under critical or supercritical pressures requires special
approaches since systems under these conditions become highly nonideal, which limits the use of models developed to describe
adsorption under low or normal pressures.
The dependence of the density on the pressure for binary systems in the critical or supercritical states (i.e., equation of
state) is described well by the generalized Bender equation [27]

p = ρT [ R g + Bρ + Cρ 2 + Dρ 3 + Eρ 4 + Fρ 5 + (G + Hρ 2 )ρ 2 exp( − a 20ρ 2 )] (23)

where B = a1 – a2/T – a3/T2 – a4/T3 – a5/T4; C = a6 + a7/T + a8/T2; D = a9 + a10/T; E = a11 + a12/T; F = a13/T;
G = a14/T3 + a15/T4 + a16/T5; H = a17/T3 + a18/T4 + a19/T5; a20 = ρ −2
c ; ai are constants. The densities of the adsorbate in the
gaseous (subscript g), liquid (subscript l), and fluid as a multilaminar adsorbate (m) may be related through the equation for
volatility f

ρ
f (T , ρ ) p (T , ρ ) 1 dρ
R g T ∫0
ln = − 1+ [ p(T , ρ ) − R g Tρ] 2 (24)
R g Tρ R g Tρ ρ

and equation

E 
f l, m = f g exp l, m  (25)
 RT 

where E is the energy of interaction an adsorbate molecule with the pore wall and adjacent molecules, which is calculated using
the Lennard–Jones potentials for various types of pores [1, 3]. In the generalized variant of the equation of state, the constants in
Eq. (23) are redetermined using six parameters for each compound: critical values of the temperature Tc, pressure pc, and
density rc as well as the molecular mass M, acentricity factor w and polarity factor c. These parameters are given for a series of
compounds by Platzer and Maurer [27]. In the case of mixtures, the system is characterized by pseudocritical parameters Tcm

154
and rcm and pseudopure factors wm and cm. One of the most important parameters for describing mixtures is the volatility
coefficient jk of the k-th mixture component. This parameter is calculated from the volatility parameter for the mixture jm,
which is found from the generalized expression for compressibility of the mixture (Z)


dV
ln ϕ m = ∫ ( Z − 1) − ln Z + Z − 1, (26)
V
V

 ∂ ln ϕ m  N  ∂ ln ϕ m 
ln ϕ k = ln ϕ m +   − ∑ xi   , (27)
 ∂x k  p, T , x i =1  ∂x k  p, T , x
i≠ k i≠ k

Z = 1+ B *δ + C *δ 2 + D*δ 3 + E *δ 4 + F *δ 5 + (G* + H *δ 2 )δ 2 exp( −δ 2 ), (28)

where t = T/Tc; d = r/rc; B* = e1 – e2/T – e3/t2 – e4/t3 – e5/t4; C* = e6 + e7/t + e8/t2; D* = e9 + e10/t; E* = e11 + e12/t; F* = e13/t;
G* = e14/t3 + e15/t4 + e16/t5; H* = e17/t3 + e18/t4 + e19/t5 and ei(w, c) = g1iw + g2ic + g3iwc + g4i + g5ic2.
An additional parameter dependent on temperature is usually introduced to describe the equilibrium in a binary
mixture. However, a greater number of correlation parameters for the transition from describing individual systems to
describing their mixture is required to describe the dependence of properties on temperature, especially outside the range of
experimental temperatures. According to Platzer [27], the set of these parameters is given from the equations

1
∑ ∑ xk x j ν c
ηm
Tcm = Tckj ,
ν ηc m k j
kj
m

Tckj ν cηm = (Tck Tc j )( ν ck ν c j ) ηm k kj ,


kj

ν cm = ∑ ∑ x k x j ν c ,
kj
k j

1 1/ 3
ν ckj = ( ν ck + ν1c/j 3 ) 3 ξ kj ,
8

ν m = ∑ xk ω k ,
k

χ m = ∑ xk χ k ,
k
1
ηm = ∑
1/ 2N ( N − 1) k
∑ ( xk + x j )η kj ,
j >k

R g Tcm Z cm
p cm = ,
ν cm

Z cm = ∑ x j Z c j ,
j
pcj ν cj
Zc j = (29)
R g Tc j

155
where xi and yi are the mole content of the i-th component in the mixture in the liquid and gaseous state. The binary parameters
hkj, kkj, and xkj are determined from the condition of minimization of the sum of the relative discrepancies

 p − p  2  y 2
1 exp − y1 calc  
N
S = ∑  exp calc
 +   (30)

i =1  p exp   y1 exp  
 

in comparing the calculated parameters with the experimental values. We should note that the mean errors in the calculations
using the generalized Bender equation relative to pressure and content of the liquid phase for a large set of binary systems are
much less than when using the Lee–Kessler–Plucker approach [27].
Redetermination of the parameters of Eqs. (24) and (25) is required in the case of adsorption of a mixture in the critical
or supercritical state. The mole content of the i-th condensed phase will depend both on the porous structure of the adsorbent,
adsorption potential, and on the competitive adsorption of the different phases. The content of the liquid phases in the
multicomponent mixture will clearly increase with increasing microporosity of the adsorbents such as activated charcoal both
relative to the bulk state of the fluids and upon their adsorption on wide-pore adsorbents.

5. COMPETITIVE ADSORPTION FROM THE LIQUID MEDIUM

Multicomponent competitive adsorption from the liquid medium has great importance since the elimination of
impurities from water and the extraction, separation, and concentration of many compounds is accomplished precisely through
a competitive adsorption mechanism. Many studies have been published on this question [7-10, 28] and many models have
been developed to describe competitive adsorption from liquid media, which has been described in detail in the monographs of
Lyklema [10], Lipatov [7, 8], and Parfitt and Rochester [9], especially relative to chromatographic methods (some of which
will be discussed below) and the adsorption of polymers and metal ions. Thermodynamic, diffusion, adsorption, static,
quantum chemical, molecular mechanical, and dynamic models have been proposed for adsorption from solution. We should
note the works of Tolmachev et al. [87, 88] on the Ohno–Kondo, Ohno–Kondo–Aranovich, and
Ohno–Kondo–Aranovich–Tolmachev lattice models with different numbers of model parameters, the studies of Rusanov [29,
89, 90] on equations of state based on the concept of the exclusion factor and approximation hierarchy, the work of Schay [30]
on multicomponent excess adsorption from different phase separation boundaries, and other studies [42, 91, 92]. As in
multicomponent adsorption from the gas phase, various Langmuir models are often used (the concentration of the solutes is
included instead of pressure in equations similar to Eqs. (4)-(7)), mass balance equations similar to Eqs. (19)-(22),
thermodynamic models similar to Eqs. (11)-(16), and others [7-10, 28-30, 42, 87-92]. Let us examine the application of these
approaches to some concrete systems.
Wurster et al. [42] studied the adsorption of barbiturates such as phenobarbital (luminal), mephobarbital, and
primidone, which differ only in the side-groups on the heterocycle (see Fig. 15), on activated charcoal from a liquid model of
stomach fluid using several models of multicomponent adsorption: competitive Langmuir model (CLM) (the pressure in
Eq. (4) is replaced by equilibrium concentration), modified CLM, and the Le Van–Vermeulen model. The two latter models
give very good agreement with the experimental data, while CLM gives less satisfactory results since the adsorption capacity
for adsorbates is assumed to be the same in this model, while the two other models take account of the difference in the
adsorption capacity for different mixture components. The modified CLM has two additional parameters, namely, the
maximum adsorption (adsorption capacity) of the adsorbates (Xmi). The equations for the amount of adsorbed components
(g/m2) for a binary system when Xm1 > Xm2 appear as follows:

( X m1 − X m 2 ) AC eq 1 X m 2 AC eq 1
X1 = + , (31)
1+ AC eq 1 1+ AC eq 1 + BC eq 2

156
a b c

Fig. 15. Structure of barbiturates: a) phenobarbital, b) mefobarbital, c) primidone.

X m 2 BC eq 2
X2 = (32)
1+ AC eq 1 + BC eq 2

where A and B are the affinity constants for the first and second adsorbates found from the data for individual adsorption. The
first term in the right-hand side of Eq. (31) corresponds to the Langmuir equation describing the adsorption of compound 1
without competition, while the second term takes account of competition with compound 2. In the case of the adsorption of
three compounds, the CLM equations become [92]

( X m1 − X m 2 ) AC eq 1 ( X m 2 − X m 3 ) AC eq 1 X m 3 AC eq 1
X1 = + + , (33)
1+ AC eq 1 1+ AC eq 1 + BC eq 2 1+ AC eq 1 + BC eq 2 + CC eq 3

( X m 2 − X m 3 )BC eq 2 X m 3 BC eq 2
X2 = + , (34)
1+ AC eq 1 + BC eq 2 1+ AC eq 1 + BC eq 2 + CC eq 3

X m 3CC eq 3
X3 = (35)
1+ AC eq 1 + BC eq 2 + CC eq 3

where C is the affinity constant of the third adsorbate.


IAST relative to solutions is based on the thermodynamic equivalence of the spreading pressure of each adsorbate. The
spreading pressure of the i-th (pi) is defined as the difference between the boundary tensions of the pure solvent and of the
solvent with adsorbate on the boundary with the adsorbent

C *eq i
Rg T Xi
πi =
S ∫ C eq i
dC eq i (36)
0

C *eq i is the equilibrium concentration of the i-th adsorbate in individual adsorption and S is the specific surface. We may assume
that Ceq i = C *eq i x is , where x is is the molar adsorption fraction of the i-th adsorbate. In the Le Van–Vermeulen model [42, 92]

x m AC eq 1
X1 = + ∆ L 2 (1+ ∆ L 3 ) (37)
1+ AC eq 1 + BC eq 2

where x m is the normalized monolayer capacity

157
Fig. 16. Adsorption of barbiturates from individual (open
circles) and ternary solutions (equal initial component
concentrations, closed circles) on SuperChar G812R
activated charcoal (Gulf Bio-Systems): 1, 2) phenobarbital,
3, 4) mefobarbital, and 5, 6) primidone (adapted from the
work of Wurster [42]).

X m1 AC eq 1 + X m 2 BC eq 2
xm = +
AC eq 1 + BC eq 2

( X m1 − X m 2 ) 2 AC eq 1 BC eq 2  1 1 
+2   +  ln (1+ AC eq 1 + BC eq 2 ) − 1, (38)
X m1 + X m 2 ( AC eq 1 + BC eq 2 )   AC eq 1 + BC eq 2 2 
2


Xm 1 and Xm 2 are the adsorption capacity for components 1 and 2, respectively,

AC eq 1 BC eq 2
∆ L 2 = ( X m1 − X m 2 ) ln(1+ AC eq 1 + BC eq 2 ),
( AC eq 1 + BC eq 2 ) 2

X m1 − X m 2 1
∆L3 = ×
X m1 + X m 2 AC eq 1 + BC eq 2

 ( BC eq 2 ) 2 + 2BC eq 2 − 4 AC eq 1 − ( AC eq 1 ) 2
× ln (1+ AC eq 1 + BC eq 2 ) +
 AC eq 1 + BC eq 2

3( AC eq 1 ) 2 + 4 AC eq 1 + AC eq 1 BC eq 2 − 2BC eq 2 − 2( BC eq 2 ) 2 
+ .
1+ AC eq 1 + BC eq 2 

In comparing the calculated and experimental adsorption for two-component barbiturate systems, the correlation
coefficients are equal to 0.946-0.956 (CLM), 0.945-0.977 (modified CLM), and 0.958-0.975 (Le Van–Vermeulen model) [42].
The correlation coefficients for a three-component system (Fig. 16) are equal to 0.946-0.976 (modified CLM) and
0.920-0.941 (CLM) [42, 92], i.e., on the whole, all the models give rather reliable results. The difference in the adsorption of
barbiturates is a function of differences in the free energy of their solvation, geometry, polarity, and polarizability of their

158
Fig. 17. Adsorption isotherms for nitrobenzene (a, b) and para-nitrophenol (a, c)
individually and in the presence of the other adsorbate in different concentration ratios
(from aqueous solution) on activated charcoal (adapted from Koganovskii [11]).

Fig. 18. Adsorption free energy distribution function for nitrobenzene (a) and
para-nitrophenol (b) adsorbed individually and in the presence of the other adsorbate at
different concentration ratios (from aqueous solution).

molecules. We should note that the heterocyclic ring is most planar for phenobarbital due to the interaction of the p electrons of
the C=O bonds and unshared electron pairs of the nitrogen atoms. The replacement of C=O by CH2 in primidone leads to a
breakdown in this conjugation and the ring takes a chair-like conformation. In the case of mephobarbital, the distortions in the
almost planar ring structure are the same as in phenobarbital. These structural features of barbiturates give the following
solvation free energies: DGs = –28 (phenobarbital), –15 (mephobarbital), and –56 kJ/mol (primidone) (the DGs values were
calculated using the IEFPCM/B3LYP/6-31G(d,p) method [93, 94] for molecular geometry calculated using the SM5.42/PM3
method [95]). An increase in hydrophobic character upon replacing >N—H by >N—CH3 enhances the free energy of
solvation. However, there is no greater hydrophobic character of primidone relative to phenobarbital since the electron-donor
properties of the nitrogen atom in the case of the chair-like heterocycle are greatly enhanced, i.e., the capacity of these atoms to

159
form strong hydrogen bonds with water molecules is increased. This structural and electronic rearrangement of the heterocycle
probably also leads to greater adsorption for primidone relative to the other barbiturates (Fig. 16). The steric factor is greater in
the adsorption of mephobarbital than in the adsorption of phenobarbital. Thus, the adsorption of phenobarbital is somewhat
greater both in the case of individual systems and in adsorption from mixtures.
Molecules of nitrobenzene (NB) and para-nitrophenol (NP) have rather similar dimensions but NP molecules are more
hydrophilic. Thus, the solvation free energy calculated by the IEFPCM/B3LYP/6-31G(d,p) method DGs = –22.7 kJ/mol (NB)
and –35.4 kJ/mol (NP). This leads to less adsorption of NP from aqueous solution (Fig. 17) [11] since the desolvation energy
upon adsorption is greater. The total adsorption of NB+NP from an equimolar solution is approximately 20 and 50% greater
than the individual adsorption of NB and NP, respectively, although the adsorption of the individual compounds is less than the
total for their mixture. Thus, the adsorption sites of activated charcoal may divided into two types. The competitive adsorption
of NB and NP occurs on type-1 sites, for example, in micropores (the adsorption isotherms are similar at low concentrations).
There is virtually no competitive adsorption on type-2 sites. Evidence for competitive adsorption is seen both in the adsorption
isotherms (Fig. 17) and changes in the adsorption free energy distribution functions f(DG) (Fig. 18) calculated in this work. A
shift in the second maximum of f(DG) occurs with increasing concentration of the second component in the direction
corresponding to weaker adsorption.
The adsorbate in the multicomponent potential theory (MCPT) is seen as a separated mixture in an external potential
field created by the adsorbent [96]. This permits us to use the equations of state describing the thermodynamic properties of the
separate and bulk phases. The advantage of MCPT, as in the case of IAST, is the possibility of using the data for the adsorption
of individual compounds to predict the results of adsorption of the corresponding mixtures [96]. In the model of
multicomponent adsorption based on the Polanyi potential theory, the mixture is seen as a heterogeneous substance separated
in an external field created by the adsorbent and described by the standard equation of state for separated and bulk phases. The
model entails several parameters required for correlation of the adsorption isotherms of the individual components [97]. The
model for the separation of food products (liquid oils) by adsorption was studied in the case of a mixture of benzene and
cyclohexane in their adsorption on silica gel. However, the use of a bulk liquid and ideal adsorbed phase gives unsatisfactory
results [98]. Analysis of the different variants of multicomponent equilibrium adsorption using the Langmuir model, IAST,
Freundlich isotherms taking account of a constant separation factor showed that IAST should be used with caution when one or
several components are poorly adsorbed [12].
All chromatographic methods are based on competitive multicomponent and multiphase adsorption [13-15, 31]. In
contrast to analytical applications of chromatography, in which small volume samples contain dilute solutions, large volumes
and/or high concentrations of compounds in the probe are used in preparative chromatography. In this case, the distribution
equilibrium of the compounds between phases is not linear and, furthermore, the separation results depend strongly on
competition between the adsorbed compounds, heterogeneity of the adsorbent surface, and the PSD of the adsorbent. The most
significant feature of separation under column overload conditions is the strong dependence of the adsorbate holding time on
the amount of probe introduced [13-15, 43]. The stationary layer consisting of well-packed fine particles (with diameter of
fractions of a millimeter) of an adsorbent such as silica gel may be adequately characterized in the equilibrium disperse model.
All the effects in this model leading to broadening of the chromatographic zone are reflected by the apparent diffusion
coefficient Da, which correlates with the number of theoretical plates [43]. The corresponding material balance equation for the
i-th component of the mixture consisting of k compounds is given by the following expression:

∂c i 1− ε ∂q i ( c1 , c 2 ,K , c k ) ∂c i ∂ 2ci
+ +u = Dai (39)
∂t ε ∂t ∂z ∂z 2

for i = 1, 2,…, k
uL
Dai = , (40)
2N i

where ci is the concentration of the i-th component, z is the longitudinal coordinate, e is the column porosity, u is the linear
velocity of the mobile phase, L is the column length, and function qi describes the relationship between the concentration of

160
component i in the stationary phase and the concentration of all the components of the mixture c1, c2,…, ck in the mobile phase.
In order to use this model to describe the separation process, we must know the usually nonlinear relationship qi = qi(c1, c2,…,
ck). We must also know the boundary and initial conditions of the concrete system. In elution chromatography, it is usually
assumed that prior to introduction of the probe, the column is completely regenerated and the probe is introduced onto the
column as a rectangular pulse. The system of mass balance Eqs. (39) and (40) may be solved numerically by the finite
difference method after suitable discretization. The Langmuir model for adsorption from a multicomponent mixture is
commonly used to describe the equilibrium between the mobile and stationary phases due to its simplicity and the good
agreement with experimental data [13-15, 43].
Chromatography provides for the efficient separation of multicomponent mixtures due to differences in the adsorption
affinity of different compounds. There are many types of gaseous and liquid chromatography and the scope of these methods is
expanding. For example, a new method has been developed for continuous chromatographic separation described in detail by
Dunnebier [99], who proposed several methods for obtaining models of separation processes including a nonlinear
thermodynamic model, equilibrium theory, and theory for the propagation of nonlinear waves taking account of the nonideal
nature of the system. The extraction of trace amounts of organic compounds from different solutions holds practical importance
in analytical chemistry. The efficiency of adsorption, concentration, and subsequent desorption are functions of the
competitive adsorption of the organic compounds and the solvents. Adsorption is carried out in some solvents (the minimum
free energy, i.e. |DGs|, increases upon adsorption), while desorption is carried out in other solvents (|DGs| increases upon
desorption) [100-102]. These processes may be described in the framework of competitive adsorption, including the solvent in
the examination.
We should note the lattice Boltzmann method (LBM), which was developed to describe the behavior of
multicomponent systems in the complex space of pores taking account of surface tension and adsorption [103]. In describing
the configuration of the separation boundaries in the case of partial pore occupancy, LBM successfully imitates the
Young–Laplace equation and the adsorption of a liquid film. Construction of a model of the behavior of liquids in pores with
simple geometry using the lattice Boltzmann method taking account of capillary effects and adsorption is in good accord with
the analytical solution, which confirms the reliability of this method ands its feasibility for describing the behavior of liquids in
pores with complex geometry, migrations on separation boundaries, and transport effects in partially occupied pores [103].
Many studies have been published on the multicomponent adsorption of low- and high-molecular-weight organic
compounds from liquid media. Thus, the competitive adsorption of polymethyl methacrylate (PMMA) and polystyrene (PS) on
cobalt nanoparticles (mean diameter 27 nm) was the subject of a thermogravimetric study [104]. In the case of a good solvent
for both polymers, the adsorption layer contains only PMMA but if PS has low solubility in the solvent, the adsorption layer
will contain a mixture of the polymers. A theoretical analysis of the separation boundaries between an aqueous surfactant
solution and the air or oil showed that kinetics for adsorption of individual compounds in the case of kinetically-limited
adsorption differs considerably from the adsorption kinetics for mixtures due to a strong interaction between the compounds
[105]. Multicomponent competitive adsorption was examined for nonionogenic surfactants and polymers on silica gel [106],
acid dyes on activated charcoal [107], sulfates in complex systems serving as soil models [108], C5-C10 alkanes on silicalite
[109], acetone and toluene on activated charcoal [110], anions of a series of antibiotics on ion-exchange polymers [111], and
humic and fulvic acids on copper-containing acid soils [112].
Many studies have been carried out on the competitive multicomponent adsorption of metal ions since these effects
have great practical importance both in industry and in environmental protection [32, 33, 44, 46, 113-116]. Prediction of the
behavior of active compounds in porous materials requires accurate adsorption models. Various single-site and multiple-site
models of cation exchange have been studied for the example of the adsorption of Ca(II) and Na(I) ions and Vulava et al. [113]
showed that the model of single-site cation exchange obtained using the well-known Gaines–Thomas equation [114] gives
satisfactory results for predicting cation transport although there were several deviations from the experimental data. The most
reliable predictions were obtained using a multiple-site model for cation exchange with a distribution of different ion-exchange
sites [113].
The multicomponent model for ion exchange (MMIE) [34], which gives better results than such simplified approaches
such as linear distribution coefficients, Langmuir or Freundlich isotherms, describes the competitive adsorption of ions using a
mass action expression, total exchange capacity, and activity coefficients in the multiple-site adsorption model [6, 32-36, 44,
46, 47, 113-119]. In the case of a binary system for cations Au+ and Bv+, we have MMIE equations in differential form

161
ud ln f B + d[(1− β B ) ln K v ] = − ln K v dβ B − uvn w d ln a w , (41)

vd ln f A + d (β B ln K v ) = − ln K v dβ B − uvn w d ln a w (42)

v{BX v } v{BX v } fv
where bB is the equivalent fraction of B cations in the ion-exchange resin β Β = = ; Kv = K eq Au is
u{AX u } + v{BX v } C oe fB
AX u BX v
the selectivity coefficient, Keq is the equilibrium constant, fA = and fB = are the mole fractions of the cations, NA =
NA NB
{AX u } {BX v }
and NB = are the mole fractions of the ion-exchange sites.
{AX u } + {BX v } {AX u } + {BX v }
The standard state in studying adsorption from solution corresponds to the lack of interaction of the solutes at infinite
dilution in the solution and in the adsorption phase. Activity coefficients are introduced to correct for deviation of the solution
and the adsorption phase from the standard state. For example, three parameters are used for a one-component solution: Ks is a
parameter reflecting the effect of the energy heterogeneity of the adsorbent surface, Ka = fa2/fa1 is a parameter taking account of
the change in the ratio of the activity coefficients of the molecules or ions of the solute and of the solvent upon increasing
adsorption, and Kv = f2/f1 is a parameter accounting for the deviation of the solution in the bulk from the standard. These
parameters and the corresponding equation are readily generalized to the case of multicomponent adsorption [11].
Considerable attention has been given to the multicomponent adsorption of organic compounds, especially in the
presence of metal ions and microorganisms, from aqueous solution on carbon adsorbents, zeolites, and soil elements [6,
120-129]. The equilibrium parameters for the glucose–sucrose–sorbitol–water system adsorbed on zeolite 13X were calculated
using chromatographic data for binary systems, IAST, and potential theory [130].
Bacteria may control transport and competitive adsorption of pollutants in the surface layer, i.e., in the region of
supramembrane and membrane structures, due to their capacity to adsorb a significant amount of metal cations and organic
compounds [121]. Bacteria may affect the dissolution and precipitation of mineral compounds and the decomposition of
organic pollutants as well as the valence state of redox-sensitive metals by means of metabolic and/or enzyme processes. The
competitive adsorption of humic acid on the surface of Bacillus subtilis bacteria was studied in the presence of Cd(II). The
presence of Cd(II) has virtually no effect on the adsorption of humic acid and the adsorption of acid has no effect on the
adsorption of Cd(II) since the formation of their complexes is possible [121].
4 ) and molybdate anions (MoO 4 ) (at 6.7·10 and 6.7·10 M) on the adsorption of
–7 –6
The effect of phosphate (PO2− 2−

As(V) (6.7·10–7 M) on mineral adsorbents such as kaolinite, montmorillonite, and yllite (concentration 2.5 g/L) was studied
relative to pH at constant ionic strength (0.1 M NaCl) [126]. Competitive adsorption is observed for the As(V)+PO2−
4 system
and noncompetitive adsorption is observed on a different type of sites for the As(V)+MoO2−
4 system. The model of constant
capacity upon surface complexation gives a satisfactory description of the competitive adsorption of ions. The Freundlich-type
Scheindorf–Rebhun–Sheintuch equation based on a quasithermodynamic balance taking account of specific soil types (three
soil types were studied) has been used to calculate the competitive adsorption constants (CAC) of arsenate, phosphate, and
molybdate anions [127]. The isosteric heats of adsorption (qiso), which drop in the series H2PO −4 > H2AsO −4 > MoO2− 4 , were
determined in the individual adsorption of the ions using the Clausius–Clapeyron equation. The CAC values determined for
binary solutions are in accord with the distribution of heats of adsorption of the individual compounds, i.e., large CAC values
correspond to greater values of qiso [127].
Let us examine some characteristic examples of the competitive adsorption of polymers. Competitive adsorption was
studied for polyoxyethylene (POE) and a nonionic surfactant such as POE-alkyl ether on the silica gel surface from aqueous
solution [128]. Complexes of the polymer and surfactant were not observed in the solution. Since both compounds are bound to
the surface through the same mechanism, slight differences in the structure of the layers in the adsorption phase or other
specific parameters determine whether one or two components will be adsorbed. The surfactant molecules may replace the
POE molecules up to some critical molecular mass. Surfactants with a high aggregation number (in the solution and on the
surface) may replace high-molecular-weight POE but the a longer time for replacement will be required in this case. The

162
Fig. 19. Adsorption of polyvinyl alcohol (PVA) on an alumina surface (IEP = 9.2) with
and without polyacrylic acid (PAA) (negative effect) (a) and adsorption of PAA with
and without PVA (no effect) (b) (adapted from the work of Santhiya [129]).

Fig. 20. Adsorption of polyvinyl alcohol (PVA) on an alumina surface with and without
polyacrylic acid (PAA) (negative effect) (a) and adsorption of PAA with and without
PVA (no effect) as a function of pH (b) (adapted from the work of Santhiya [129]).

existence of a critical molecular weight and corresponding change in the adsorption kinetics are attributed to a change in the
critical surface association [128].
The competitive adsorption of polyacrylic acid (PAA) and polyvinyl alcohol (PVA) on an alumina surface (a-Al2O3,
S » 10 m2/g, isoelectric point 9.2) was studied relative to the component concentrations and solution pH (Fig. 19) [129]. The
adsorption of PAA drops with increasing pH since the electrostatic interaction of the oppositely-charged PAA molecules
(polyions) and Al2O3 surface decreases and, at pH > 9.2 when there is the same charge of the molecules and surface, the
adsorption of PAA becomes negligible (Fig. 20). In the case of PVA, adsorption decreases with increasing pH since PVA is a
nonionic polymer, which is adsorbed mainly due to hydrogen bonding. The presence of PVA in binary systems does not affect
the adsorption of PAA but the amount of PVA adsorption decreases in the presence of PAA and the free energy of adsorption is
low (Fig. 21) (the calculations were carried out in this work using a regularization method). The adsorption isotherm for PAA in
the acidic region shows Langmuir character with high affinity in contrast to PVA, which has low affinity (Fig. 19a) [129].
The adsorption kinetics of individual polystyrenes (molecular weight 96, 126, and 355 kDa) and their binary mixtures
on silica gel MB-800 was studied relative to concentration [130]. The time to establish adsorption equilibrium (10-35 h)
depends on the molecular weight of the polystyrene. In the case of a mixture, the shorter polymer is initially adsorbed, but the
larger polymer predominates in the adsorption layer with time and equilibrium is achieved in approximately 35 h [130].

163
Fig. 21. Distribution function of the free energy of adsorption of
individual PAA (1), PVA (2), and PVA in the presence of PAA (3).

Fig. 22. Particle size distribution (PSD) relative to light scattering


intensity (PSDI) (a), particle volume (PSDV) (b), and number of
particles (PSDN) (c) in aqueous suspensions of A-300, PVA/A-300,
and lecithin(II)/PVA(I)/A-300 [132].

164
Fig. 23. Size distribution function for structures of nonfrozen water (a, b)
and benzene (b) in the pores of mesoporous silica gel MCM-41 at different
hydration h = 0.186 (V H 2O << Vp) and 5.2 g (V H 2O >>Vp) water per g silica gel
(a, 1, 2) and at h = 0.37 g/g (b, 1) and benzene content 0.4 g/g (b, 2) at VC6 H 6 +
V H O » Vp; the pore size distribution (PSD) is shown for MCM-41 (b, 3)
2

calculated using the nitrogen adsorption isotherm [139].

Fig. 24. Size distribution function for structures of nonfrozen benzene (1),
water (2, h = 1.33 g/g), and water (3, h = 0.67 g/g) in the presence of
chloroform (0.8 g/g) and DFT pore size distribution (4) using the nitrogen
adsorption isotherm in adsorption on activated charcoal [140].

Similar effects were observed in the adsorption of many polymers and proteins from solution. Adsorption methods,
NMR, TSD, IR, photon correlation spectroscopy, diffusion, and dynamic microweighing on a quartz detector (QCM) were
used to analyze the kinetics and dynamics of the adsorption of macromolecules and the established equilibrium [37, 38,
131-137]. A special mathematical apparatus has been developed to analyze the experimental data obtained by these methods.
Some of these systems were studied under adsorption equilibrium conditions, which, however, give significantly different
results than dynamic nonequilibrium adsorption [37, 38, 131-137]. Figure 22 shows the dependence of the particle size
distribution (PSD) on elapsed time in aqueous suspensions of pyrogenous silica gel A-300 (S = 232 m2/g) in the adsorption of
polyvinyl alcohol (PVA, 43 kDa) and then lecithin (0.77 kDa) [132]. Since the molecular weight of lecithin is much less than
for PVA, the probability of the replacement of PVA molecules by lecithin molecules during competitive adsorption is low.
However, particle aggregation is enhanced after the introduction of lecithin, which has surfactant properties, and
lecithin/PVA/A-300 aggregates appear with diameter greater than 1 µm.
In addition to various adsorption methods, NMR spectroscopy is also used to analyze multicomponent adsorption.
Thus, Evanas et al. [138] used 19F NMR spectroscopy to study the adsorption of fluorine-containing surfactants on alumina.

165
NMR spectroscopy was also used to evaluate the selectivity of adsorption from multicomponent mixtures. However, in this
case, the specific surface should be not less than 10 m2/g, i.e., this method is applicable for porous and highly disperse
adsorbents. A number of workers have analyzed the individual and competitive adsorption of water and various organic
compounds such as benzene, acetone, DMSO, acetonitrile, chloroform, trinitrotoluene, dinitrochlorobenzene, floroglucine,
polyvinylpyrrolidone, and fibrinogen on carbon and oxide adsorbents using 1H NMR spectroscopy with freezing out of the
liquid phase and cryoporimetry. Many of these systems were also studied by thermostimulated depolarization (TSD) [150].
The use of NMR and TSD methods for studying the adsorption of multicomponent systems based on change in the properties of
the adsorbates by the action of the adsorbent surface such as distortion of the structure of the hydrogen bond lattice
characteristic for the bulk liquid phase and change in the mobility of the molecules and their functional groups relative to the
nature and structure of the surface and the presence of other adsorbates. The 1H NMR spectra of water structurized in the pores
are sensitive to the appearance of other adsorbates. Thus, the removal of water into large-diameter pores occurs in the
adsorption of water and benzene on MCM-41 (Fig. 23) and of water and chloroform on activated charcoal (Fig. 24). This effect
is attributed to a reduction in the free energy of the system upon change in the contact area of the organic compounds occupying
the pores with water, which is eliminated.
The model for the dynamics of wetting of a self-organized monolayer of hydrocarbons (C18H37) by a water/DMSO
mixture showed that wetting of the hydrophobic hydrocarbon layer is achieved by DMSO molecules in a broad concentration
range [151]. In this case, DMSO molecules are oriented with their methyl groups toward the hydrophobic layer and the S=O
bonds are directed toward the water layer, i.e., a water layer covers the DMSO layer.

6. ADSORPTION OF PROTEINS

Considerable attention has been given to the multicomponent adsorption of proteins since this problem has great
practical significance [16, 17, 48, 152-159]. Progress in the study of proteins and other biomacromolecules is largely a function
of developments in the retention and identification in biomaterials in a broad concentration range (from 10–3 M to individual
molecules) [48]. The most promising areas of study are nanotechnological physicochemical and biospecific procedures for the
separation of multicomponent protein mixtures with subsequent concentration on a solid base and the identification and
binding of individual molecules using molecular detectors [48].
Theoretical calculations of the interaction of proteins with complex membranes within the framework of the mean field
theory taking account of local rearrangements of lipids and lateral interactions of the adsorbed proteins show that the lateral
repulsive interactions of the protein molecules affect the modulation profile of the membrane and, at high occupancies, lead to a

Fig. 25. Concentration of HSA on a silica gel surface (particles with


diameter 36 µm, S = 0.076 m2/g) as a function of time elapsed after initial
adsorption of labeled HSA from a 0.05% solution a t = 0-3 h, injection of
an 0.05% solution of unlabeled immunoglobulin (IgG, 150 kDa) at t = 3 h,
washing and injection of a 0.01% solution of polyclonal anti-IgG (IgY,
170 kDa) at t = 6 h (adapted from the work of Lutante et al. [158]).

166
Fig. 26. Concentration of immunoglobulin (IgG, 150 kDa)
adsorbed on a silica surface as a function of the time elapsed
after washing and adding an 0.01% solution of unlabeled
polyclonal anti-IgG (IgY, 170 kDa) in two experiments:
1) IgG was mixed with 0.05% solution of HSA and adsorbed
at t = 0-3 h and 2) IgG was added in a solution after the
adsorption of HSA on silica gel over 3 h (adapted from the
work of Lutante et al. [158]).

Fig. 27. Concentration of proteins adsorbed on untreated,


APTES-modified, dextran-modified (D, acidified for
0.5-24 h), and Na hyaluronate-modified silica surfaces
(adapted from the work of Ombelli et al. [159]).

significant reduction of the interaction energy [154]. It is important to consider the mobility of lipids and the protein–protein
interactions, while the formation of lipid–protein domains is enhanced due to the electrostatic adsorption of proteins.
Hindrance was found for diffusion of protein molecules (monoclonal antibodies) within porous particles as the result of
adsorption [155]. Netz et al. [156] used the mean field model for free energy calculations and their study of the supramolecular
interaction of proteins and lipids forming a monolayer on the air–water boundary showed that localization or delocalization of
protein molecules in the lipid layer or outside this layer depends on the nature of their interaction with lipids, concentration, and
temperature. Proteins compete for space on the liquid–vapor or liquid–solid boundaries in accord with their molecular mass
and concentration [157]. At equilibrium, the ratios of concentrations on the liquid–vapor boundary and in the bulk are identical.
The absolute concentration of each component on the separation boundary is less than in the case of adsorption from an
individual solution. Smaller proteins having greater diffusion velocity are concentrated more rapidly on separation boundaries
but then, larger, slower proteins replace a fraction of the lighter proteins (according to the Vroman effect) and equilibrium is
settled [157]. The Vroman effect is observed also on the liquid–solid boundary but the concentration ratio in the bulk and on the

167
Fig. 28. a) Adsorption isotherms for gelatin (maintenance time 2 h) from water on
silica gel A-300 (1 mass %) with preadsorbed polyethylene glycol (PEG, 20 kDa)
0 (1), 20 (2), 50 (3), and 200 mg (4) per g silica gel, b) corresponding distribution
functions for change in Gibbs free energy upon the adsorption of gelatin [132].

Fig. 29. Adsorption of gelatin (a, b) or ovalbumin (b) on A-300 1 mass % (a) and 3
or 4 mass % (b) as a function of the concentration of preadsorbed polyethylene
glycol (PEG, 20 kDa) from water and of polyvinylpyrrolidone (PVP) (water and
phosphate buffer) (a) and PVP (b) [132].

surface may differ significantly due to the specific interaction of proteins with the surface in contrast to the liquid–vapor
separation boundary, where such an interaction is lacking, and differences in the structure of the macromolecules [16, 17, 157].
The affinity of polyclonal anti-IgG and human immunoglobulin (IgG) was studied by several methods in competitive
adsorption on silica gel [158]. Comparison of the activity of IgG adsorbed directly or by exchange with preadsorbed albumin
shows that, in the former case, the reaction between anti-IgG and the adsorbed protein proceeds in a 1 : 1 ratio, while, in the
latter case, no reaction is observed [158].
The initial response of blood to exposure to an artificial surface corresponds to the competitive adsorption of blood
proteins and then bioreactions such as inflammation and blood coagulation take place [159]. On the whole, the competitive
adsorption of proteins plays a key role in the biocompatibility of biomaterials. Polysaccharides, which are the basic
components of the endothelial cell glycocalyx, are capable of decreasing the nonspecific adsorption of proteins and cell
adhesion. Thus, the deposition of these substances on a biomaterial surface enhances their biocompatibility. A comparative
study was carried out on the competitive adsorption of bovine serum albumin (BSA) and fibrinogen (FG) on the silica surface
both untreated and modified with g-aminopropyltriethoxysilane (APTES), dextran, and sodium hyalurate (Fig. 27). The
polysaccharide-modified silica surface adsorbed a much smaller amount of proteins than the untreated or APTES-modified
silica surface. The concentration of adsorbed BSA in this case ranges from 69 to 93% (in the case of the same initial protein
concentration) [159].

168
Fig. 30. Change in the frequency of a quartz crystal (QCM) in
the adsorption of FG–HSA and HSA–FG on Au (1, 2),
DLC (3, 4), Ti (5, 6), and TiN (7, 8); 1), 3), 5), 7) adsorption of
FG, then HSA, 2), 4), 6), 8) adsorption of HSA, then FG [131].

TABLE 1. Thickness of Protein Layer (h)


Adsorbed on Different Surfaces in Individual and
Consecutive Adsorption of HSA (14´4 nm) and
FG (45´9´6 nm) [131]

h, nm
Surface
FG HSA FG–HSA HSA–FG

Au 12.3 5.2 8.9 13.5


DLC 34.8 21.0 31.0 35.3
Ti/TiOx 35.6 10.1 17.8 37.8
TiN 28.9 22.5 30.9 30.0

The preadsorption of polymers such as polyethylene glycol (PEG, 35 kDa) and polyvinylpyrrolidone (PVP, 12.6 kDa)
with a lower molecular weight than proteins such as gelatin and ovalbumin inhibits the strong adsorption of proteins on
pyrogenous silica (Figs. 28 and 29). In this case, the concentration of the adsorbed protein decreases almost linearly with
increasing concentration of the adsorbed polymer (Fig. 29). The adsorption of the protein is close to zero in the case of
monolayer or multiplayer adsorption of the polymer. The replacement of the polymer molecules by protein molecules is not
observed, probably due to insufficient maintenance time (2 h) since the equilibrium in this system may be achieved only after a
much longer period.
In a study of the competitive adsorption of human serum albumin (HSA) and fibrinogen on the surface of gold, DLC
(diamond-like carbon), Ti (TiOx), and TiNx [37, 131, 133], the ratio of their concentrations corresponds to this ratio in human
blood. Even though CHSA/CFG = 8, the amount of adsorbed HSA by quartz crystal microweighing (QCM) (Fig. 30) assuming G
~ –Df(t), is always less than for FG: GFG/GHSA = 1.3¸3.4. This ratio depends on the surface type. The maximum value of

169
GFG/GHSA is found for TiOx, while the minimal value is found for TiNx. On the other hand, the ratio of the adsorption of FG and
HSA is less than the ratio of their molecular weights: GFG/GHSA < MFG/MHSA » 5.1. Thus, the relative packing density of
adsorbed FG molecules is less than for HSA (Table 1) [131]. We should note the considerable differences between the
three-dimensional structures of HSA and FG; the FG molecule is more extended and consists of several domains. HSA may be
adsorbed in smaller surface pores than FG. These findings and measurement of the adsorbed layer thickness (Table 1, h)
indicate that the structure of this layer for FG and HSA on the surfaces studied differs considerably due to the protein molecular
structure, conformational changes upon adsorption, and differences in adhesion to different surfaces. These effects give rise to
differences in coatings and in hematocompatibility.
The surfaces of Au, DLC, Ti (TiOx), and TiNx in the study are not smooth on the nanometer level or even the micron
level [131]. The “bumpy” topography of the surfaces remains also after protein adsorption since a rather thin protein monolayer
is adsorbed (Table 1, h), whose thickness is less than the surface roughness. For FG adsorbed on all surfaces, dmin < h < dmax,
where dmin and dmax are the minimum and maximum molecular diameters. A similar relationship is also found for HSA
adsorbed on the surface of Au and Ti but h > dmax in the adsorption of HSA on DLC and TiNx. Desorption of fibrinogen from
the surfaces studied in the QCM measurements is not observed (Fig. 30) in contrast to HSA adsorbed on gold. The adsorption of
HSA on Ti is less than for FG, which provides for the marked adsorption of FG (as a second protein) after the adsorption of
HSA on Ti in comparison with the other surfaces. The DLC surface is the least polar of the coatings studied [131]. Thus,
electrostatic forces in protein adsorption give less contribution to the interaction than disperse forces in comparison with the
other more polar and hydrophilic surfaces. On the other hand, the amount of bound water on the hydrophobic surface/protein
separation boundary is less than in adsorption on a hydrophilic surface [17]. The adsorption of FG on the most hydrophobic
coatings of DLC and TiNx is greater than the adsorption of HSA. A significant amount of FG is adsorbed on these surfaces with
preadsorbed albumin.
The differences in the slopes of the Df(t) curves after completion of individual and binary adsorption may be due to
changes in the structure of the adsorbed protein layer, for example, its compaction due to folding and oligomerization of the
macromolecules and slow penetration of molecules into the pores (slow in comparison with adsorption) [17, 134, 160].
Furthermore, ions from the buffer solution may be adsorbed on the protein molecules, lowering the vibrational frequencies of
the charged quartz crystal. These processes depend on the structure of the surface and protein type. Thus, the rate of compaction
of the adsorbed layer after adsorption is greatest in the adsorption of HSA on DLC and least for Au. According to the QCM
data, it is difficult to differentiate the processes of adsorption and folding of the macromolecules. According to literature data
[17], an increase in the FG concentration leads to a decrease in the effects of molecular unfolding on the surface and surface
occupancy due to unfolding of much less coating as a result of adsorption Qu < Qa or for constants ku << kaCFG when
CFG > 3·10–4 g/cm3. In our experiments [131], the value for CFG was greater by an order of magnitude. Thus, upon formation of
a monolayer, we may neglect the coating due to unfolding of the macromolecules since Qa/(Qa + Qu) ® 1. Correspondingly,
the slower processes of molecular folding and unfolding may play a role only after the completion of adsorption. FG molecules,
which are larger and have greater adhesion to the surfaces studied, may replace HSA molecules from the surface in accord with
the Vroman effect [17]. The results obtained indicate that the Vroman effect is greatest for surfaces with greatest roughness
[131]. This may explain the heterogeneity of the adsorbed HSA layer on the bumpy surface, which permits FG more readily to
find available surface segments and then replace a portion of the HSA molecules. Adsorption of the smaller HSA molecules
occurs more rapidly than the larger FG molecules. Thus, DDf12 > 0 and also Df1/Df2 < 1. This occurs as the result of the more
rapid diffusion of the smaller albumin molecules since protein adsorption is diffusion-controlled [17]:

0. 5
dΓ  D
= C0   (43)
dt  πt 

where D is the diffusion coefficient and C0 is the protein concentration. Furthermore, the HSA concentration is higher. Since
FG molecules are larger and have greater affinity toward the surfaces studied, over time Df1/Df2 >1 and DDf12 < 0. The kinetics
of the individual and competitive adsorption may be described by the following equation [131]:

170
Fig. 31. Adsorption rate constants calculated using Eqs. (44)-(46), FG (1, 3, 5, 7), HSA (2,
4, 6, 8) (a) and HSA after FG (1, 3, 5, 7) and FG after HSA (2, 4, 6, 8) (second shift Df) (b)
on the surface of Au (1, 2), DLC (3, 4), Ti (5, 6), and TiNx (7, 8) [131].

∆f
ln = − kt (44)
A

where k is the pseudo-first-order rate constant. This constant decreases over time due to an increase in the relative occupancy of
the surface Q. Equation (44) may be written as [131]

  γ (t − α )ν  
∆f = A exp −  − 1 (45)
  1+ β ( t − α )  

where a is a constant corresponding to the time between injection of the protein into the flow and the onset of its adsorption, b,
g, and n are equation parameters dependent on the concentration of the first and second proteins, characteristics of the QCM
sensor and liquid, and temperature, exp [ ] < 1, n >1, and A » –Dfm is the maximum change in frequency as the result of
adsorption. Parameters b, g, and n are determined by minimization of the functional

2
t max    γ (t − α )ν   
Φ(α , β , γ , ν ) = ∫  spline ( ∆f exp ( t )) − A  exp −  − 1
 1+ β ( t − α )  
 dt → min, (46)
t min   

i.e., when ¶F/¶xi = 0 (xi = b, g, and n), where spline(Dfexp(t)) is the cubic spline for the experimental data, tmin and tmax
correspond to the time boundaries from the injection of the protein until complete removal of the unbound protein by the liquid
flow. In order to describe the multicomponent adsorption of proteins over a prolonged period, the change in the vibrational
frequencies of the crystal may be described by the following equation [131]:

 


∆f = A  ( t − α ) δ exp − γ ( t − α ) ν  b ( t − α ) + c( t − α ) λ 

(47)

where A, b, c, a, d, g, l and n are equation parameters dependent on the parameters of the protein mixture and order of their
adsorption. Equation (47) gives better agreement with the experimental data than Eq. (45) at longer observation times but in the
case of a sharp drop in frequency upon adsorption in a short time period, the two equations give similar results.

171
Fig. 32. Rate constant distribution function in individual and binary adsorption j(k) of proteins:
FG (1), HSA after FG (2), HSA (3) and FG after HSA (4) on the surface of Au (a) and Ti (b) [131].

Fig. 33. PSD relative to their volume in aqueous suspensions of A-300 a) PEG/A-300,
BSA(2)/PEG(1)/A-300 and b) PVA/A-300, BSA(2)/PVA(1)/A-300; 1 and 2 are the first and
second adsorbates [132].

Kinetic equation (44) may be written taking account of the decrease in the adsorption rate for the i-th protein dG/dt
(assuming that dG/dt ~ d(–Dfexp(t))/dt ® 0 as Q ® 1 on the basis of experimental observations) upon an increase in the relative
surface occupancy Q

dG/dt ~ d(–Dfexp(t))/dt ~ 1– Q = 1 – (t – a )/c (48)

as a type-I Fredholm integral equation for the i-th protein [37, 131]

k max
νi
∆f exp i ( t ) = A i ∫ exp[ k ( t − α i ) ] [1− ( t − α i ) / χ i ]ϕ i ( k )dk (49)
0

where ni and ci are equation parameters and ji(k) is the distribution function of adsorption rate constants k for the i-th protein.
Equation (49) was solved by regularization with ji(k) > 0 for any k. The calculations for the adsorption rates using the two
approaches (minimization and regularization) and Eqs. (45), (46), and (49) give similar results. The condition kFG(t) > kHSA(t)
(Fig. 31) although we have the opposite relationship at the onset. This effect corresponds to a displacement of the adsorption
rate constant distribution function j(k) for HSA in the region of large k relative to FG (Fig. 32). For the second protein,

172
Fig. 34. Diffusion coefficient distribution function of molecules of
FG (1) and BSA (2-4) through an Integra artificial skin membrane
without preadsorption (1, 2) and after the preadsorption of BSA (3)
and FG (4) on the membrane [133]. The arrows indicate the value
of D in the diffusion of the proteins in a dilute individual solution.

kFG-HSA < kHSA-FG (Fig. 31b) and the corresponding peak on j(k) for FG (as the second protein in the HSA–FG pair) is shifted
toward greater values of k relative to HSA for FG–HSA pair (Fig. 32). The peaks on j(k) for FG and HSA adsorbed on Ti are
shifted toward larger values of k (Fig. 32b) in comparison with adsorption on Au (Fig. 32a). This result is in accord with the
position of the corresponding k(t) curves (Fig. 31a). The surface of Au coated with HSA is more homogeneous (less
microroughness) for subsequently adsorbed FG. The j(k) peak is narrower than in the adsorption on Ti (Fig. 32a,b, curves 4). A
similar effect was observed for the FG–HSA pair (curves 2). We should note that photon correlation spectroscopy, similar to
QCM but in contrast to adsorption methods (with centrifugation), permits us to observe the interaction of the protein molecules
with the pyrogenous silica surface with a preadsorbed polymer monolayer (Fig. 33). Depending on the nature of the polymer
(whether it can form hydrogen bonds as a proton donor), we find enhanced aggregation (BSA is adsorbed on PVA/A-300 since
PVA forms hydrogen bonds as a proton and electron donor) or reduced aggregation (BSA/PEG/A-300, since PEG is only an
electron donor) [132].
The effect of competitive adsorption of proteins on biocompatibility is also seen when using Integra artificial skin
[161]. The preadsorption of BSA has less effect on the diffusion of BSA molecules through the Integra membrane than the
preadsorption of fibrinogen (Fig. 34). The diffusion coefficient distribution function f(D) was calculated by solving the integral
equation [133]

c0
Dmax
1  x2 
c( x, t ) =
2 π
∫ Dt
exp  −  f ( D )dD
 4Dt 
(50)
Dmin

where c0 is the initial protein concentration in the first cell, c is the protein concentration in the second cell separated from the
first cell by a membrane with thickness x, t is the elapsed time, Dmin and Dmax are the minimum and maximum diffusion
coefficients in the integration (with the condition f(D) ³ 0 for any D). The positions for the maxima on f(D) (the right-hand
maximum for BSA) in diffusion without preadsorption correspond to the diffusion coefficients of the proteins in the bulk since
Integra is macroporous [161]. However, on the whole, we have a rather broad distribution of f(D), while it is bimodal in the case
of BSA and the membrane markedly inhibits the diffusion of the macromolecules. In the case of the preadsorption of proteins, a
shift in the distributions occurs toward reduced mobilities. However, a slight shift of the right-hand peak toward larger rates is
observed in the case of BSA/BSA, which can be attributed to desorption of a portion of the proteins in the second cell.

173
7. DYNAMIC ADSORPTION FROM DIFFERENT MEDIA

The dynamic adsorption of mixtures of gases, for example, toxic compounds from moist air on carbon or combined
filters holds great practical importance. The adsorption of tert-butylbenzene, cyclohexane, and dimethyl methylphosphate as
model toxicants from dry and moist air flows was studied on various adsorbents, i.e., in competitive adsorption with water [137,
162, 163]. The total adsorption of an organic compound on a filter over time t is described by the following equation [162]

− ∆Gmax t
 − ∆G   v L ( t− t b ) 
C Σ (t ) =A ∫ ∫ f ( ∆G )exp
 RT 
 exp
 αLV p 
 dtd ( − ∆G ) (51)
0 0

where A and a are constants, f(DG) is the adsorption free energy distribution function DG, vL is the adsorbent pore volume, L is
the filter thickness, Vp is the adsorbent pore volume, and tb is the passage time of the compound at the minimum critical
concentration. The total adsorption is described by an equation taking account of the adsorption rate constant distribution
function f(be) [162]

− ∆Gmax t
 − ∆G   v L ( t− t b ) 
C Σ (t ) =A ∫ ∫ f ( ∆G )exp
 RT 
 exp
 αLV p 
 dtd ( − ∆G )
(52)
0 0

where be is the adsorption rate constant and c0 and c are the initial concentration of the organic compound and the current
concentration of this compound after the filter. The constants and distribution functions in Eqs. (51) and (52) depend both on
the preadsorption of water on the filter and the humidity of the air (Figs. 35-37), i.e., the competitive adsorption of the toxicants
and water leads to poorer functioning of the filter. The most negative results were found in the case of water preadsorption from
moist air even in the case of only slight water preadsorption [137, 162, 163] since most of the water adsorption sites on carbon
adsorbents are located on external graphene plane faces, while adsorbed water as clusters and nanodrops [164] blocks access to
the micropores, in which most of the toxicants are adsorbed. Analysis of the data obtained indicates that a certain balance is
required for optimal functioning of the filter (maximum values of be and tb with decrease in free energy of adsorption) between
the transport pores (wide mesopores and macropores) to facilitate diffusion with the granules and micropores for efficient
adsorption of the toxicants. Furthermore, a decrease in adsorbent granule size improves functioning of the filter [162, 163].
Further optimization of the filtration system is necessary for the separation of gas mixtures in a pulse mode [165].

Fig. 35. Effect of the air flow humidity and water preadsorption on the passage of
tert-butylbenzene through a carbon filter, initial activated charcoal (a, b, 1),
oxidized (a – 2, 3; b – 2) and reduced adsorbent (a – 4, 5; b – 3) from dry (a – 1, 2, 4;
b – 1), and moist air (a – 3, 5) and with preadsorbed water (b – 2, 3), granule
fraction 0.5-0.7 mm (b) and 0.7-1 mm (a) [162].

174
Fig. 36. Dynamic adsorption rate constant distribution
functions for tert-butylbenzene on a carbon filter (initial (1,
2), oxidized (3, 4), and reduced adsorbent (5)) from dry (1,
3, 5) and moist air with 80% humidity (2, 4) [162].

Fig. 37. Gibbs free energy distribution function for the


adsorption of tert-butylbenzene on a carbon filter (initial (1,
4, 7), oxidized (2, 6, 8) or reduced adsorbent (3, 5)) from dry
(1, 4-6) or moist air (7, 8) with preadsorbed water (2, 3),
granule fraction 0.5-0.7 mm (1-3) and 0.7-1 mm (4-8) [162].

Many works have been published on the diffusion and dynamics of various multicomponent systems. Diffusion as the
rate-limiting step in multicomponent dynamic adsorption has been described within the framework of the orthogonal
collocation method based on a generalized approximation of the driving forces independent of the mass transport model
relative to the corresponding parabolic differential equations in partial derivatives describing diffusion within a porous particle
using mass balance equations [166]. This approach yields approximate solutions, which were obtained previously by analytical
solution of a uniform diffusion equation.
A model for the diffusional transport in microporous complex matrix membranes (clinoptilolite flakes dispersed in a
matrix of microporous silica) was developed using the Maxwell–Stephan theory [167]. The behavior of the two-component
N2/CH4 system in its interaction with a complex adsorbent may be predicted using Henry’s law, the expanded Langmuir model,
and IAST. The partition parameters for the ortho and para isomers of xylene on a silica–silicate matrix were found using
Henry’s law. The selective penetration of a membrane is a function of the pressure and obeys the expanded Langmuir model
and IAST with efficient exchange and weak spatial limitation taking account of the dependence of the Maxwell–Stephan

175
diffusion coefficient on the load (pressure). This pressure dependence was analyzed in terms of adsorption selectivity and
penetration balanced between the silicate flakes and silica matrix [167].

8. COMPETITIVE ADSORPTION AND CHEMICAL REACTIONS ON


PHASE SEPARATION BOUNDARIES

Surface reactions as in heterogeneous catalysis or in functionalization of a solid surface feature both competition
between reagent molecules for adsorption sites and competition between active sites of the heterogeneous surface for reagents.
The first type of competitive reactions include processes occurring without change in the structure of the surface sites (for
example, heterogeneous catalysis without poisoning of the catalytic sites), while the second type includes processes such as
functionalization and poisoning of catalytic sites, which lead to change in the surface structure. Thus, two major reactions are
possible in the thermal decomposition of polydimethylsiloxane (PDMS) on the surface of untreated pyrogenous silica gel and
silica gel with a supported zirconium dioxide phase: oxidation leading to SiO2, CO2, and H2O and depolymerization leading to
the formation mainly of three- and four-membered cyclic siloxanes. The appearance of a catalytically active ZrO2 phase leads
to a reduction in the temperature for depolymerization (type-I reaction) and decrease in the contribution of oxidation (type-II
reaction) [168].
The structure of the carbon species formed differs significantly in the carbonization of the same precursors at the same
concentration on a surface of mesoporous silica gels and nanosilicas [39, 169-174]. This is a result of different availability and a
different number of sites for the formation of nuclei of the carbon phase in the silica gel pores and on the surface of nonporous
primary nanosilica particles as well as different kinetics of the processes on surfaces of porous and nonporous particles.
Inhibition to diffusion occurs in narrow pores, especially upon blockage of access by pyrocarbon particles. Thus, the carbon
content in the pores, especially narrow pores, as in silica gel Si-40, proves less than on the external particle surface. The
external surface of mesoporous silica gel particles (textural microporosity) is minimal. Thus, the number of carbon phase nuclei
also proves to be insignificant. However, the carbon phase nuclei are more active in the addition of subsequent portions of
pyrocarbon than the silica surface. Thus, a corresponding increase is observed on the kinetic curve for increasing pyrocarbon
content after a minimum due to reactions on the carbon species [171]. The sizes of the carbon particles on silica gels, which
depend in the case of identical content also on the PSD of the silica gels, and nanosilicas differ by more than an order of
magnitude.
Somewhat different behavior is observed in the carbonization of organic precursors on mixed oxides, which have
catalytic activity in acid–base and redox reactions [39, 169-174]. These processes are analogous to coking of heterogeneous
catalysts for organic reactions. Adsorption sites on the surface of mixed oxides, titanium oxide, and other catalysts are equally
or more active relative to sites of the carbon phase formed. Thus, competitive processes on these sites cause a different
morphology of the carbon particles: 1) spherical carbon particles (20-2000 nm), which grow from carbon nuclei on the external
surface of the particle surface, 2) graphenes (1-2 nm) formed in pores and on the active sites of the matrix, and 3) film structures
(when there is a large amount of active surface sites or after formation of a new, catalytically active noncarbon phase). In the
latter case, competitive reactions become more complex when organometallic compounds such as metal acetylacetonates or
alcoholates are used since a new, often catalytically active metal oxide or reduced metal phase is formed, leading to a more
complex morphology and pyrocarbons [169].
The reaction temperature plays an important role relative to adsorption in the functionalization of a surface in the case
of binary or multicomponent adsorption when one of the adsorbates is a catalyst such as triethylamine since the lifetime of the
adsorbed complexes decreases with increasing temperature and the catalytic effect disappears at some critical value [175-177].
Similar effects are characteristic for the heterogeneous catalysis of organic synthesis. Thus, multicomponent adsorption may be
not only competitive but also a promoter of surface reactions.
Partial blockage of the surface is used to obtain a regular mosaic coating of the adsorbent surface by functional groups
or clusters of another phase. In previous work [178], we studied the effect of preadsorbed polyvinylpyrrolidone (PVP,
12.6 kDa, monolayer with 0.175 g per g silica) on the reaction of nanosilica with hexamethyldisilazane (HMDS) (2 h at 423 K
and then 8 h at 363 K). The monolayer PVP coating was found not to inhibit the reaction of nanosilica with HMDS. The
reaction of Ti[OCH(CH3)2]4 with the nanosilica surface in the presence of adsorbed PVP (10.7 kDa) in 2-propanol also takes

176
place very efficiently, i.e., the adsorption concentration of PVP relative to low-molecular-weight reagents is relatively
insignificant [179]. Thus, lighter but more reactive compounds may replace heavier but less reactive components of the
mixture. We should note that preliminary partial silylation of the surface by pyrogenous silica inhibits the formation of TiO2
particles to a greater extent than adsorbed PVP [179], i.e., the weak negative effect of PVP on these reactions is caused by
mobility of the polymer in the adsorption phase. The efficient migration of PVP molecules between particles of coated
(monolayer) and noncoated nanosilica was observed upon stirring and slight warming (to 353 K) by the IR photon correlation
method [180].
The rates of multicomponent organic reactions are higher in water [181] in the presence of various catalysts and solid
plates [18, 182-188]. In this case, the reaction selectivity depends on the selectivity of multicomponent adsorption and the
selective action of the catalysts. The photocatalytic oxidation of individual herbicides (isoproturon, simazine, and propazine) in
a suspension of TiO2 at an initial herbicide concentration of from 3 to 70 mg/L, which corresponds to typical concentrations in
ground and surface waters, is described satisfactorily by the Langmuir–Hinshelwood (LH) reaction rate equation [189]. Kinetic
parameters found for individual systems are used in a variant of the LH equation extended to describe time-dependent
degradation in a multicomponent system. Comparison of the binding constants for the herbicides upon dark adsorption and in
the case of photocatalytic oxidation suggests that the degradation of a mixture of these compounds occurs as a consequence of
surface reactions or reactions near the surface occurring by a competitive LH mechanism. This corresponds to the condition of
a monomolecular coating of TiO2 (the herbicide concentration did not exceed 1 mg/L). At higher concentrations, departure is
seen from the LH mechanism due to multilayer adsorption, leading to more rapid degradation [189].
The effect of the multicomponent adsorption of normal C6-C9 alkanes on zeolite catalysts Pt/H-Y (Si/Al = 2.7) and
Pt/USY (Si/Al = 13 and 30) was analyzed for a series of cracking, addition, and isomerization reactions [190]. The reaction
rates were found to be virtually independent of the type of catalyst studied. Thus, Denayer et al. [190] assumed that these
reactions proceed more rapidly by a classical mechanism through formation of alkylcarbenium ions than by a mechanism with
transition states not completely separated from the surface sites.
A study was carried out on the adsorption of NO2, NO (O2), and the catalytic reduction of NO by methane on the
CoH-ZSM-5 catalyst (SiO2/Al2O3 = 25, Co/Al = 0.132-0.312) [191]. The adsorption of NO and —NOy, formed in the
adsorption of NO2, is unstable since efficient desorption occurs at temperatures below the reaction temperature (523 K).
Although NO is weakly adsorbed on CoH-ZSM-5, the competitive adsorption of NO and —NOy leads to a reduction in the
adsorption of —NOy (i.e., the adsorption of NO2). If the presence of NO2 is significant for the formation of adsorbed —NOy
species, the amount of these species upon the adsorption of NO + O2 should be, at least, between the adsorption of NO2,
NO + O2, and NO + NO2 due to minimum concentration of NO2 and maximum concentration of NO. Indeed, the amount of
adsorbed —NOy corresponded to the average between the two other adsorption processes, indicating that the origin of the
adsorbed —NOy compounds may be related only to the adsorption of NO2 [191].
The adsorption of individual CO and CO2 was studied in comparison to the adsorption of mixtures H2/CO2, CO/CO2,
and H2/CO/CO2 on zeolite BaZSM-5 at 323-473 K and atmospheric pressure [192]. The adsorption of CO2 is much higher than
for CO and higher from the H2/CO2 mixture than upon individual adsorption (according to TCD and IR photon correlation
data). This result is attributed to a chemical reaction on the catalyst involving CO2 and H2. The adsorption of CO2 decreases
only slightly with increasing partial pressure of CO (at constant total pressure). On the other hand, the adsorption of CO
decreases significantly with increasing CO2 partial pressure. The expanded Langmuir model satisfactorily describes
multicomponent adsorption on zeolite BaZSM-5 for these systems [192].
Feasibility has been demonstrated plasmocatalytic methods (TiO2 catalysts, Aerolyst® 7706 TiO2, CuOMnO2/TiO2)
for the elimination of pollutants such as NOx, O3, and toluene from the air [193]. The competitive adsorption of water and
toxicant molecules leads to a reduction in the efficiency of catalytic air purification. For example, the reduction of NOx is lower
for air with 50% humidity in comparison with processes with dry air, which is attributed to the formation of HNO3 in the plasma
discharge with subsequent adsorption on the catalyst and its poisoning [193]. H2S gives a negative effect in the catalytic
conversion (hydrodeoxygenation) of benzofuran on reduced or sulfidized Ni-Mo/Al2O3 catalysts [194]. The major effect was
shown to be caused by the competitive adsorption of H2S and benzofuran and the formation of SH groups largely involving
molybdenum upon raising the reaction temperature [194].

177
CONCLUSION

Thus, the results of multicomponent, multiphase, competitive equilibrium static and nonequilibrium dynamic
adsorption depend on many factors such as the mixture composition, differences in the molecular weights and chemical nature
of the components, gas phase parameters (temperature, pressure, flow rate), liquid phase parameters (concentration, pH, ionic
strength, solvent characteristics), chemical structure and heterogeneity of the adsorbent surface (group types, catalytic
activity), and textural characteristics and morphology of the particles. The isotherms in the case of equilibrium multicomponent
adsorption with a monolayer coating (or less) are described satisfactorily by different modified variants of the multicomponent
Langmuir equation, especially if corrections are introduced for adsorbent heterogeneity and the difference in adsorption
capacity for different adsorbates and the pore size distribution is taken into consideration. Important information is derived
from thermodynamic approaches, which permit us to find the change in the free energy, entropy, and isosteric heat upon
multicomponent adsorption. We should note that the solution of the most general integral equations of multicomponent
adsorption requires the use of special mathematical procedures such as self-consistent regularization, which permits us to
calculate several independent distribution functions for similar or different parameters.
Some common features were observed in static and dynamic multicomponent adsorption and multicomponent
chemical reactions on the surface of porous or disperse materials. These common features are attributed to the nature of the
interaction of adsorbates or reagents with active surface sites and to effects due to the topology and chemistry of the solid
surface as well as the medium since there is an intermolecular interaction in all cases between the adsorbate/reagent molecules
and the active surface sites. The subsequent existence of these complexes, their decomposition upon desorption or removal by
another component or their transformation in reactions depend primarily on temperature, since subsequent reaction steps
(following adsorption) require an activation energy, and on the medium. When only one of the components is actively adsorbed
or reacts, while the other components weakly interact with the surface, the adsorption of the active compound individually and
from a mixture may be virtually identical. The removal of low-molecular-weight compounds by high-molecular-weight
compounds occurs rather rapidly if the surface is available for the latter, that is, the contribution of micropores is negligible. On
the other hand, adsorbed polymers cannot hinder the interaction of highly active low-molecular-weight reagents with a surface.
The competitive adsorption of high-molecular-weight compounds with different molecular weights and different affinity
toward the surface may be very hindered on porous adsorbents and the time to establish equilibrium is much greater than for the
adsorption of the individual compounds, especially in the case of preadsorption of one of the components, even with a lower
molecular weight.

REFERENCES

1. D. D. Do, Adsorption Analysis: Equilibria and Kinetics, Imperial College Press, London (1998).
2. L. P. Ding, Multicomponent Adsorption in Heterogeneous Microporous Solids, St. Lucia, Queensland, Australia
(2002).
3. A. W. Adamson and A. P. Gast, Physical Chemistry of Surface, John Wiley & Sons, New York (1997).
4. S. J. Gregg and K. S. W. Sing, Adsorption, Surface Area and Porosity, Academic Press, London (1982).
5. J. A. Schwarz and C. I. Contescu, Surfaces of Nanoparticles and Porous Materials, Marcel Dekker, New York (1999).
6. M. J. McGuire and I. H. Suffet (eds.), Treatment of Water by Granular Activated Carbon, Oxford University Press,
Oxford, England (1983).
7. Yu. S. Lipatov and L. M. Sergeeva, Polymer Adsorption [in Russian], Naukova Dumka, Kiev (1992).
8. Yu. S. Lipatov, Phase Transfer Phenomena in Polymers [in Russian], Naukova Dumka, Kiev (1980).
9. G. Parfitt and C. Rochester (eds.), Adsorption from Solution on Solid Surfaces [in Russian], Mir, Moscow (1986).
10. J. Lyklema, Fundamentals of Interface and Colloid Science, Academic Press, London, Vol. 1 (1991), Vol. 2 (1995).
11. A. M. Koganovskii, T. M. Levchenko, and V. A. Kirichenko, Adsorption of Solutes [in Russian], Naukova Dumka,
Kiev (1977).
12. R. A. Matusko and R. A. Lamontagne, Theoretical Aspects of Multicomponent Adsorption Equilibria, Pentagon Rep.,
A228523 (1997).

178
13. G. Guiochon, S. Golshan-Shirazi, and A. M. Katti, Fundamentals of Preparative and Nonlinear Chromatography,
Academic Press, Boston (1994).
14. G. Guiochon and B. Lin, Modeling for Preparative Chromatography, Academic Press, Amsterdam (2003).
15. C. Horvath (ed.), High Performance Liquid Chromatography – Advances and Perspectives, Academic Press,
New York (1986).
16. M. Malmsten (ed.), Biopolymers at Interfaces, Marcel Dekker, New York (1998).
17. J. L. Brash and T. A. Horbett (eds.), Proteins at Interfaces: Physicochemical and Biochemical Studies, American
Chemical Society, Washington, D. C. (1987).
18. J. Zhu and H. Bienyamé (eds.), Multicomponent Reactions, Wiley-VCH Verlag GmbH & Co., Weinheim, Germany
(2005).
19. V. B. Fenelonov, Porous Carbon [in Russian], Institute of Catalysis, Novosibirsk, Russia (1995).
20. M. M. Dubinin, Adsorption and Porosity [in Russian], Voen. Akad. Khim. Zashchity, Moscow (1972).
21. M. M. Dubinin, Pure Appl. Chem., 48, 407-414 (1976).
22. A. P. Karnaukhov, Adsorption. Texture of Disperse and Porous Materials [in Russian], Nauka, Novosibirsk, Russia
(1999).
23. C. C. Huang, Thermal Regeneration of Fixed Multicomponent Adsorption Beds, Ph.D. Dissertation, University of
Texas, Austin (1987).
24. C. Chu, Modeling and Simulation of Multicomponent, Nonisothermal Adsorption Cycles for Gas Separations, Ph.D.
Dissertation, University of Texas, Austin (1991).
25. K. Pruess and A. Battistelli, TMVOC, Simulator for Multiple Volatile Organic Chemicals,
http://repositories.cdlib.org/lbnl/LBLN-52400.
26. G. D. Parfitt, Pure Appl. Chem., 48, 415-418 (1976).
27. B. Platzer and G. Maurer, Fluid Phase Equilib., 84, 79-110 (1993).
28. A. Vrij, Pure Appl. Chem., 48, 471-483 (1976).
29. A. I. Rusanov, Usp. Khim., 74, No. 2, 126-137 (2005).
30. G. Schay, Pure Appl. Chem., 48, 393-400 (1976).
31. G. Guiochon, J. Chromatogr. A, 965, 129-162 (2002).
32. C. A. J. Appelo, J. A. Hendriks, and M. van Veldhuizen, J. Hydrol., 146, 89-113 (1993).
33. P. L. Bjerg, H. C. Ammentrop, and T. H. Christensen, J. Contam. Hydrol., 12, 291-311 (1993).
34. C. A. J. Appelo, Reactive Transport in Porous Media, Reviews in Mineralogy, P. C. Lichtner, C. I. Steefel, and E. H.
Oelkers (eds.), Vol. 34 (1996), pp. 193-227.
35. C. I. Steefel, S. A. Carroll, P. Zhao, and S. Roberts, J. Contam. Hydrol., 67, 219-246 (2003).
36. A. Voegelin, V. M. Vulava, F. Kuhnen, and R. Kretschmar, J. Contam. Hydrol., 46, 319-338 (2000).
37. S. V. Mikhalovsky, V. M. Gun’ko, K. D. Pavey, et al., Surfaces and Interfaces for Biomaterials, P. Vadgama (ed.),
CRC Press and Woodhead Publ., Cambridge (2005), pp. 322-385.
38. V. M. Gun’ko, A. V. Klyueva, Yu. N. Levchuk, and R. Leboda, Adv. Colloid Interface Sci., 105, 201-328 (2003).
39. K. Wang and D. D. Do, Adsorption, 5, 25-37 (1999).
40. A. Tvardovski, D. Tondeur, and E. Favre, J. Colloid Interface Sci., 265, No. 2, 239-244 (2003).
41. A. L Myers and F. Siperstein, Colloids Surfaces A, 187/188, 73-81 (2001).
42. D. E. Wurster, K. A. Alkhamis, and L. E. Matheson, AAPS PharmSciTech, 1, No. 3, article 25, 1-15 (2000).
43. A. Zaidel’-Morgenshtern, Ros. Khim. Zh., 47, No. 1, 80-89 (2003).
44. M. Cernik, K. Barmettler, D. Grolimund, et al., J. Contam. Hydrol., 16, 319-337 (1994).
45. M. Cernik, B. Borkovec, and J. C. Westall, Langmuir, 25, 6127-6137 (1996).
46. M. Cernik, K. Barmettler, D. Grolimund, et al., J. Contam. Hydrol., 16, 319-337 (1994).
47. Y. D. Ivanov, V. M. Govorun, V. A. Bykov, and A. I. Archakov, Proteomics, 6, 1399-1414 (2006).
48. V. M. Gun’ko, Teor. Éksp. Khim., 36, No. 6, 349-353 (2000) [Theor. Experim. Chem., 36, No. 6, 319-324 (2000)].
49. V. M. Gun’ko, V. V. Turov, and R. Leboda, Teor. Éksp. Khim., 38, No. 4, 119-225 (2002) [Theor. Experim. Chem., 38,
No. 4, 199-228 (2002)].
50. J.-H. Park, J.-N. Kim, S.-H. Cho, et al., Chem. Eng. Sci., 53, No. 23, 3951-3963 (1998).

179
51. A. Goj, D. S. Sholl, E. D. Atken, and D. Kohen, J. Phys. Chem. B, 106, No. 33, 8367-8375 (2002).
52. L. Lu, Q. Wang, and Y. Liu, Langmuir, 19, No. 25, 10617-10623 (2003).
53. D. Schuring, A. O. Koriabkina, A. M. de Jong, et al., J. Phys. Chem. B, 105, No. 32, 7690-7698 (2001).
54. Y. Wang and M. D. Le Van, Ind. Eng. Chem. Res., 46, No. 7, 2141-2154 (2007).
55. H. Chen and D. S. Sholl, Langmuir, 23, No. 11, 6431-6437 (2007).
56. Q. Wu, L. Zhou, J. Wu, and Y. Zhou, J. Chem. Eng. Data, 50, No. 2, 635-642 (2005).
57. B.-U. Choi, D.-K. Choi, Y.-W. Lee, et al., J. Chem. Eng. Data, 48, No. 3, 603-607 (2003).
58. M. P. Cal, M. J. Rood, and S. M. Larson, Energy Fuels, 11, Vol. 2, 311-315 (1997).
59. A. J. Fletcher, M. J. Benham, and K. M. Thomas, J. Phys. Chem. B, 106, No. 30, 7474-7482 (2002).
60. S. Sircar, Ind. Eng. Chem. Res., 46, No. 10, 2917-2927 (2007).
61. G.-M. Nam, B.-M. Jeong, S.-H. Kang, et al., J. Chem. Eng. Data, 50, No. 1, 72-76 (2005).
62. G. Calleja, J. Pau, and J. A. Callas, J. Chem. Eng. Data, 43, 944-1003 (1998); M. Sukuth, S. Sander, and J. Gmehling, J.
Chem. and Eng. Data, 44, No. 6, 1427-1428 (1999).
63. K. O. Christe, “Method for the selective separation of gases,” US Patent No. 4,695,295, Publ. Sept. 22, 1987.
64. R. Kumar, “Removal of water and carbon dioxide from atmospheric air,” US Patent No. 4,711,645, Publ. Dec. 8, 1987.
65. M. Mizukami and K. Kurihara, Chem. Lett., 29, No. 3, 256 (2000).
66. M. Elsayed, P. Hall, and M. Heslop, Ninth International Conference on Fundamentals of Adsorption,
May 20-25, 2007, Giardini Naxos, Sicily, Italy.
67. K. Hashimoto, T. Kono, M. Kurimoto, and M. Uno, “Process for adsorbing and separating carbon dioxide from gas
mixtures,” US Patent No. 4,812,147, Publ. Febr. 23, 1988.
68. W. E. BeVier, “Multicomponent adsorption process,” US Patent No. 4,812,147, Publ. May 5, 1989.
69. S. Sircar, R. R. Conrad, and W. R. Koch, “Closed-loop thermal regeneration of adsorbents containing reactive
adsorbates,” US Patent No. 4,971,606, Publ. Nov. 20, 1990.
70. M. Uno, K. Ueda, M. Inoue, and S. Kaji, “Process for separating and recovering carbonic acid gas from gas mixture
by adsorption,” US Patent No. 4,986,835, Publ. Jan. 22, 1991.
71. V. M. Gun’ko, V. V. Turov, R. Leboda, et al., Langmuir, 23, No. 6, 3184-3192 (2007).
72. C. R. C. Jensen, N. A. Seaton, V. Gusev, and J. A. O’Brien, Langmuir, 13, No. 5, 1205-1210 (1997).
73. F. R. Siperstein and A. L. Myers, AIChE Journal, 47, No. 5, 1141-1159 (2001).
74. K. Wang, S. Qiao, and X. Hu, Separ. Purif. Technol., 20, 243-249 (2000).
75. S. R. Challa, D. S. Sholl, and J. K. Johnson, J. Chem. Phys., 116, No. 2, 814-824 (2002).
76. V. Heftmánek, F. „apek, O. Šolcová, P. Schneider, Chem. Eng. J., 74, 171-179 (1999).
77. C. R. Clarkson, SPE Journal, 8, No. 3, 236-251 (2003).
78. S.-H. Moon, and J.-W. Shim, J. Colloid Interface Sci., 298, No. 2, 523-528 (2006).
79. P. Jacob and D. Tondeur, Chem. Eng. J., 26, 143 (1983).
80. H. Rhee, E. D. Heerdt, and N. R. Amundson, AIChE Journal, 28, No. 3, 423 (1982).
81. R. S. Deshpande, S. L. Sharp-Goldman, and A. B. Bocarsly, Langmuir, 18, No. 20, 7694-7698 (2002).
82. J. Gmehling, Pure Appl. Chem., 71, No. 6, 939-949 (1999).
83. G.-Q. Tang, K. Jessen, and A. R. Kovscek, SPE 95947, SPE Ann. Tech. Conf. and Exhibition, Dallas, Texas,
October 8-12, 2005.
84. O. Di Giovanni, W. Dörfler, M. Mazzotti, and M. Morbidelli, Langmuir, 17, 4316-4321 (2001).
85. S. Himeno, T. Komatsu, and S. Fujita, J. Chem. Eng. Data, 50, No. 2, 369-376 (2005).
86. C. Nguyen and D. D. Do, Langmuir, 17, No. 5, 1552-1557 (2001).
87. A. M. Tolmachev, M. I. Godovikova, and T. S. Egorova, Zh. Fiz. Khim., 79, No. 1, 110-115 (2005).
88. A. M. Tolmachev, M. V. Borodulina, A. B. Arzamastseva, et al., Vestn. Moskovsk. Univ. Khim., 42, No. 4, 244-247
(2001).
89. A. I. Rusanov, Zh. Fiz. Khim., 77, No. 10, 1736-1741 (2003).
90. A. I. Rusanov, Zh. Fiz. Khim., 77, No. 10, 1764-1771 (2003).
91. N. S. Polyakov, M. L. Gubkina, A. V. Larin, and G. A. Petukhova, Sorpts. Khromatograf. Prots., 1, No. 6, 936-939
(2001).

180
92. K. A. Alkhamis and D. E. Wurster, AAPS PharmSciTech, 3, No. 3, article 23, 1-8 (2002).
93. M. W. Schmidt, K. K. Baldridge, J. A. Boatz, et al., J. Comput. Chem., 14, 1347-1363 (1993) (GAMESS version 7.3).
94. A. A. Granovsky, PC GAMESS version 7.0.6, http://classic.chemmsu.su/gran/gamess/index.html.
95. J. D. Xidos, J. Li, T. Zhu, et al., GAMESOL – version 3.1, University of Minnesota, Minneapolis, 2002, based on the
GAMESS.
96. M. A. Monsalvo and A. A. Shapiro, Fluid Phase Equilib., 254, Nos. 1/2, 91-100 (2007).
97. A. Shapiro, J. Colloid Interface Sci., 201, 146-157 (1998).
98. O. Aoyi, K. O. Asiedu, and O. Osembo, J. Food Technol. Africa, 6, No. 2, 63-67 (2001).
99. G. Dünnebier and K.-U. Klatt, Chem. Eng. Sci., 55, 373-380 (2000).
100. R. Leboda, V. M. Gun’ko, W. Tomaszewski, and B. J. Trznadel, J. Colloid Interface Sci., 238, No. 2, 489-500 (2001).
101. W. Tomaszewski, V. M. Gun’ko, R. Leboda, and J. Skubiszewska-Zi“ba, J. Colloid Interface Sci., 266, 388-402
(2003).
102. W. Tomaszewski, V. M. Gun’ko, R. Leboda, and J. Skubiszewska-Zi“ba, J. Colloid Interface Sci., 282, No. 2, 261-269
(2005).
103. M. C. Sukop and D. Or, Water Resourc. Res., 40, N W01509, 1-11 (2004).
104. K. David, N. Dan, and R. Tannenbaum, Surface Sci., 601, No. 8, 1781-1788 (2007).
105. G. Ariel, H. Diamant, and D. Andelman, Langmuir, 15, No. 10, 3574-3581 (1999).
106. B. R. Postmus, F. A. M. Leermakers, L. K. Koopal, and M. A. CohenStuart, Langmuir, 23, No. 10, 5532-5540 (2007).
107. K. K. H. Choy, J. F. Porter, and G. McKay, J. Chem. Eng. Data, 45, No. 4, 575-584 (2000).
108. J. C. L. Meeussen, J. Kleikemper, A. M. Scheidegger, et al., Environ. Sci. Technol., 33, No. 19, 3443-3450 (1999).
109. M. S. Sun, O. Talu, and D. B. Shah, J. Phys. Chem., 100, No. 43, 17276-17280 (1996).
110. C. Li and W. M. Moe, Environ. Sci, Technol., 39, No. 7, 2349-2356 (2005).
111. L. A. M. van der Wielen, M. J. A. Lankveld, and K. C. A. M. Luyben, J. Chem. Eng. Data, 41, No. 2, 239-243 (1996).
112. L. Weng, E. P. M. J. Fest, and J. Fillius, et al., Environ. Sci. Technol., 36, No. 8, 1699-1704 (2002).
113. V. M. Vulava, R. Kretzschmar, K. Barmettler, and A. Voegelin, Water Resourc. Res., 38, No. 5, 1-7 (2002).
114. G. L. Gaines and H. C. Thomas, J. Chem. Phys., 21, 714-718 (1953).
115. V. M. Vulava, R. Kretzschmar, U. Rusch, et al., Environ. Sci. Technol., 34, 2149-2155 (2000).
116. J. C. L. Meeussen, A. Scheidegger, T. Hiemstra, et al., Environ. Sci. Technol., 30, 481-488 (1996).
117. E. Brouwer, B. Baeyens, A. Maes, and A. Cremers, J. Phys. Chem., 87, 1213-1219 (1983).
118. J. Griffioen, Water Resourc. Res., 29, 3005-3019 (1993).
119. P. C. Lichtner, S. Yabusaki, K. Pruess, and C. I. Steefel, Vadose Zone J., 3, 203-219 (2004).
120. P. K. Muralidharan and C. B. Ching, Ind. Eng. Chem. Res., 36, No. 2, 407-413 (1997).
121. P. G. Wrightman and J. B. Fein, Chem. Geology, 180, 55-65 (2001).
122. J. M. Zachara, C. T. Resch, and S. C. Smith, Geochim. et Cosmochim. Acta, 58, 553-566 (1994).
123. M. F. Benedetti, C. J. Milne, D. G. Kinniburgh, et al., Environ. Sci. Technol., 29, 446-457 (1995).
124. D. A. Fowle and J. B. Fein, Geochim. et Cosmochim. Acta, 63, 3059-3067 (1999).
125. L. K. Koopal, W. H. van Riemsdijk, and J. C. M. de Wit, J. Colloid Interface Sci., 166, 51-60 (1994).
126. B. A. Manning, and S. Goldberg, Clays Clay Mater., 44, No. 5, 609-623 (1996).
127. W. R. Roy, J. J. Hassett, and R. A. Griffin, Eur. J. Soil Sci., 40, No. 1, 9-15 (1989).
128. B. R. Postmus, F. A. M. Leermakers, L. K. Koopal, and M. A. StuartCohen, Langmuir, ASAP
Article 10.1021/1a063525z S0743-7463(06)03525-6.
129. D. Santhiya, S. Subramanian, K. A. Natarajan, and S. G. Malghan, J. Colloid Interface Sci., 216, 143-153 (1999).
130. M. Kawaguchi, Y. Sakata, S. Anada, et al., Langmuir, 10, 538-541 (1994).
131. V. M. Gun’ko, S. V. Mikhalovskii, L. I. Mikhalovskaya, et al., Physical Chemistry of Nanomaterials and
Supramolecular Structures, A. P. Shpak and P. P. Gorbik (eds.) Vol. 2, Naukova Dumka, Kiev, Ukraine (2007).
132. V. M. Gun’ko, V. I. Zarko, E. F. Voronin, et al., J. Colloid Interface Sci., 300, No. 1, 20-32 (2006).
133. S. V. Mikhalovsky, L. I. Mikhalovska, S. L. James, et al., Combined and Hybrid Adsorbents, Fundamentals and
Applications (NATO Security Science Series C), J. M. Loureiro and M. T. Kartel (eds.), Springer, Dordrecht, Germany
(2006), pp. 309-320 (2006).

181
134. M. Melillo, V. M. Gun’ko, L. I. Mikhalovska, et al., Langmuir, 20, 2837-2851 (2004).
135. V. M. Gun’ko, S. V. Mikhalovskii, M. Melilo, et al., Teor. Éksp. Khim., 40, No. 3, 133-138 (2004) [Theor. Experim.
Chem., 40, No. 3, 137-143 (2004)].
136. V. M. Gun’ko, N. N. Vlasova, L. P. Golovkova, et al., Colloids Surfaces Sci. A, 167, No. 3, 229-243 (2000).
137. J. P. Blitz and V. M. Gun’ko (eds.), Surface and Environmental Science (NATO Science Series II), Vol. 228, Springer,
Dordrecht, Germany (2006).
138. L. Evenäs, I. Furó, P. Stilbs, and R. Valiullin, Langmuir, 18, 8096-8101 (2002).
139. V. M. Gun’ko, V. V. Turov, A. V. Turov, et al., Central Eur. J. Chem., 5, No. 2, 420-454 (2007).
140. V. M. Gun’ko, O. P. Kozynchenko, V. V. Turov, et al., Carbon, in press (2007).
141. V. M. Gun’ko, V. I. Zarko, V. V. Turov, et al., Teor. Éksp. Khim., 37, No. 2, 73-77 (2001) [Theor. Experim. Chem., 37,
No. 2, 75-79 (2001).
142. V. V. Turov, V. M. Gun’ko, R. Leboda, et al., J. Colloid Interface Sci., 253, No. 1, 23-34 (2002).
143. V. V. Turov, V. M. Gun’ko, M. D. Tsapko, et al., Appl. Surface Sci., 229, 197-213 (2004).
144. T. A. Alexeeva, N. I. Lebovka, V. M. Gun’ko, et al., J. Colloid Interface Sci., 278, 333-341 (2004).
145. V. M. Gun’ko, V. V. Turov, J. Skubiszewska-Zi“ba, et al., Appl. Surface Sci., 214, Nos. 1-4, 178-189 (2003).
146. V. M. Gun’ko, V. V. Turov, V. N. Barvinchenko, et al., Colloids Surfaces A, 278, Nos. 1-3, 106-122 (2006).
147. A. A. Rugal, V. M. Gun’ko, V. N. Barvinchenko, et al., Central Eur. J. Chem., 5, No. 1, 32-54 (2007).
148. R. Leboda, V. V. Turov, W. Tomaszewski, et al., Carbon, 40, 389-396 (2002).
149. V. M. Gun’ko, V. V. Turov, V. M. Bogatyrev, et al., Adv. Colloid Interface Sci., 118, 125-172 (2005).
150. V. M. Gun’ko, V. I. Zarko, E. V. Goncharuk, et al., Adv. Colloid Interface Sci., 131, Nos. 1/2, 1-89 (2007).
151. J. Vieceli, and I. Benjamin, Langmuir, 19, No. 13, 5383-5388 (2003).
152. G. M. S. Finette, B. S. Baharin, Q.-M. Mao, and M. T. W. Hearn, Biotechnol. Prog., 13, No. 3, 265-275 (1997).
153. S. Damodaran, K. Anand, and L. Razumovsky, J. Agr. Food Chem., 46, No. 3, 872-876 (1998).
154. S. May, D. Harries, and A. Ben-Shaul, Biophys. J., 79, 1747-1760 (2000).
155. A. Susanto, T. Herrmann, E. J. von Lieres, and J. Hubbuch, J. Chromatogr. A, 1149, No. 2, 178-188 (2007).
156. R. R. Netz, D. Andelman, and H. Orland, J. Phys. II France, 6, 1023-1047 (1996).
157. A. Krishnan, C. A. Siedlecki, and E. A. Vogler, Langmuir, 20, No. 12, 5071-5078 (2004).
158. E. Lutante, J. C. Voegel, P. Schaff, et al., Proc. Nat. Acad. USA Biochem., 89, 9890-9894 (1992).
159. M. Ombelli, L. B. Costello, Q. C. Meng, et al., Mater. Res. Soc. Symp. Proc., 845, AA8.6.1-AA8.6.6 (2005).
160. V. M. Gun’ko, W. R. Betz, S. Patel, et al., Carbon, 44, No. 7, 1258-1262 (2006).
161. L. I. Mikhalovska, V. M. Gun’ko, V. V. Turov, et al., Biomaterials, 27, No. 19, 3599-3607 (2006).
162. V. M. Gun’ko, D. Palijczuk, R. Leboda, et al., J. Colloid Interface Sci., 294, No. 1, 53-68 (2006).
163. D. Palijczuk, V. M. Gun’ko, R. Leboda, et al., J. Colloid Interface Sci., 250, 5-17 (2002).
164. J. K. Brennan, T. J. Bandosz, K. T. Thomson, and K. E. Gubbins, Colloids Surfaces A, 187/188, 539-568 (2001).
165. L. Jiang, L. Biegler, and G. Fox, AIChE Journal, 49, 1140-1157 (2003).
166. P. Cruz, F. D. Magalhaes, and A. Mendes, Chem. Eng. Sci., 61, No. 11, 3519-3531 (2006).
167. J. A. Sheffel and M. Tsapatsis, J. Membrane Sci., 295, Nos. 1/2 , 50-70 (2007).
168. V. M. Gun’ko, M. V. Borysenko, P. Pissis, et al., Appl. Surface Sci., 253, No. 17, 7143-7156 (2007).
169. V. M. Gun’ko, R. Leboda, J. Skubiszewska-Zi“ba, and J. Rynkowski, J. Colloid Interface Sci., 231, No. 1, 13-25
(2000).
170. V. M. Gun’ko, J. Skubiszewska-Zieba, R. Leboda, and V. I. Zarko, Langmuir, 16, No. 2, 374-382 (2000).
171. V. M. Gun’ko, R. M. Leboda, M. Marciniak, et al., Langmuir, 16, No. 7, 3227-3243 (2000).
172. V. M. Gun’ko, V. I. Zarko, R. Leboda, et al., J. Colloid Interface Sci., 230, 396-409 (2000).
173. V. M. Gun’ko, R. Leboda, V. A. Pokrovskiy, et al., J. Anal. Appl. Pyrol., 60, No. 2, 233-247 (2001).
174. V. M. Gun’ko and R. Leboda, Encyclopedia of Surface and Colloid Science, A. T. Hubbard (ed.), Marcel Dekker,
New York (2002), pp. 864-878.
175. V. A. Tertykh and L. A. Belyakova, Chemical Reactions Involving the Silica Surface [in Russian], Naukova Dumka,
Kiev, Ukraine (1991).
176. V. M. Gun’ko, E. F. Voronin, E. M. Pakhlov, and A. A. Chuiko, Langmuir, 9, 716-722 (1993).

182
177. V. M. Gun’ko and A. A. Chuiko, Colloidal Silica: Fundamentals and Applications, H. E. Bergna (ed.),
Taylor & Francis, Salisbury (2005), pp. 465-497.
178. V. M. Gun’ko, E. F. Voronin, L. V. Nosach, et al., Appl. Surface Sci., 253, 2801-2811 (2006).
179. V. M. Gun’ko, V. M. Bogatyrev, V. V. Turov, et al., Powder Technol., 164, 153-167 (2006).
180. V. M. Gun’ko, E. F. Voronin, L. V. Nosach, et al., Adsorp. Sci. Technol., 24, No. 2, 143-157 (2006).
181. M. C. Pirrung and K. D. Sarma, J. Am. Chem. Soc., 126, 444-445 (2004).
182. P. A. Tempest, Curr. Opinion Drug. Discov. Developm., 8, No. 6, 776-788 (2005).
183. S. E. Denmark and Y. Fan, J. Org. Chem., 70, No. 24, 9667-9676 (2005).
184. K. M. Short and A. M. M. Mjalli, Tetrahedron Lett., 38, 359-362 (1997).
185. I. Ugi, Pure Appl. Chem., 73, No. 1, 181-191 (2001).
186. J. F. Denayer, G. V. Baron, W. Souverijns, et al., Pure Appl. Chem., 36, No. 8, 3242-3247 (1997).
187. G. G. Condorelli, M. Favazza, C. Bedoya, et al., Chem. Mater., 18, No. 4, 1016-1022 (2006).
188. C. A. Grande and A. E. Rodrigues, Ind. Eng. Chem. Res., 40, No. 7, 1686-1693 (2001).
189. A. Gora, B. Toepfer, V. Puddu, and G. Li Puma, Appl. Catal. B, 65, Nos. 1/2, 1-10 (2006).
190. J. F. Denayer, G. V. Baron, G. Vanbutsele, et al., J. Catal., 190, No. 2, 469-473 (2000).
191. J. Zhang, Y. Liu, W. Fan, et al., Fuel, 86, Nos. 10/11, 1577-1586 (2007).
192. S. K. Wirawan and D. Creaser, Separat. Purific. Technol., 52, No. 2, 224-231 (2006).
193. J. Van Durme, J. Dewulf, W. Sysmans, et al., Appl. Catal. B, 74, Nos. 1/2, 161-169 (2007).
194. A. Y. Bunch, X. Wang, and U. S. Ozkan, J. Mol. Catal. A, 270, Nos. 1/2, 264-272 (2007).

183

You might also like